Insulin resistance is associated with altered amino acid metabolism and adipose tissue dysfunction in normoglycemic women

1 Scien i ic RepoR s | 6:24540 | DOI: 10.1038/s ep24540 www.na u e.com/scien i ic epo s Insulin esis ance is associa ed wi h al e ed amino acid me abolism and adipose issue dys unc ion in no moglycemic women Pe i Wiklund1,2, Xiaobo Zhang1, Sa u Pekkala2, Reija Au io3, Lingjia Kong4, Yi an Yang1, Si kka Keinänen-Kiukaanniemi5, Ma kku Alen5,6 & Sulin Cheng1,2 Insulin esis ance is associa ed adiposi y, bu he mechanisms a e no ully unde s ood. In his s udy, we aimed o iden i y ea ly me abolic al e a ions associa ed wi h insulin esis ance in no moglycemic women wi h a ying deg ee o adiposi y. One-hund ed and en young and middle-aged women we e di ided in o low and high IR g oups based on hei median HOMA-IR (0.9 ± 0.4 s. 2.8 ± 1.2). Body composi ion was assessed using DXA, skele al muscle and li e a by p o on magne ic esonance spec oscopy, se um me aboli es by nuclea magne ic esonance spec oscopy and adipose issue and skele al muscle gene exp ession by mic oa ays. High HOMA-IR subjec s had highe se um b anched-chain amino acid concen a ions (BCAA) (p < 0.05 o bo h). Gene exp ession analysis o subcu aneous adipose issue e ealed signi ican down- egula ion o genes ela ed o BCAA ca abolism and mi ochond ial ene gy me abolism and up- egula ion o se e al in lamma ion- ela ed pa hways in high HOMA-IR subjec s (p < 0.05 o all), bu no di e en ially exp essed genes in skele al muscle we e ound. In conclusion, in no moglycemic women insulin esis ance was associa ed wi h inc eased se um BCAA concen a ions, down- egula ion o mi ochond ial ene gy me abolism and inc eased exp ession o in lamma ion- ela ed genes in he adipose issue. One o he ea lies de ec able de ec s in he me abolic con inuum leading o ype 2 diabe es is insulin esis ance1. Impai ed glucose homeos asis is associa ed wi h obesi y2, bu he means by which excessi e adiposi y induces insulin esis ance and glucose in ole ance emain con o e sial. Indeed, only abou a qua e o he a iance o insulin esis ance is explained by BMI in he gene al popula ion3. S udies ha e shown ha inc eased isk o hea disease is independen o cen al obesi y in indi iduals wi h me abolic synd ome4, and lean indi iduals may be as insulin esis an as hose wi h ype 2 diabe es5. On he o he hand, no all o e weigh and obese indi iduals de elop insulin esis ance o o he me abolic diso de s6, sugges ing ha he quan i a i e impac o obesi y pe se on insulin sensi i i y may no be as la ge as p e iously hough 7. Obesi y-associa ed inc ease in plasma ee a y acids and accumula ion o ec opic lipids a e linked wi h he onse o pe iphe al and hepa ic insulin esis ance8. Dys egula ed amino acid me abolism is also associa ed wi h obesi y- ela ed impai ed insulin sensi i i y9 and inc eased isk o u u e diabe es10. We ha e ecen ly used high- h ough pu me aboli e quan i ica ion o iden i y me abolic di e ences be ween seden a y obese indi iduals wi h and wi hou me abolic synd ome. We ound ha b anched-chain and a oma ic amino acids we e s ongly associa ed wi h insulin esis ance and o he me abolic isk ac o s, independen o a mass and wais ci cum e - ence11. This inding sugges s ha excess a mass alone does no explain he associa ions o hese amino acids wi h insulin esis ance. Thus, he sou ce o inc eased b anched-chain and a oma ic amino acids, and he mechanisms by which hey migh con ibu e o impai ed insulin sensi i i y emain incomple ely unde s ood. Recen e idence sugges s ha obesi y is associa ed wi h al e ed adipose issue me abolism, which in u n a ec s sys emic glucose homeos asis and induces insulin esis ance in skele al muscle12,13. The e is li le 1Shanghai Jiao Tong Uni e si y, Shanghai, China. 2Depa men o Heal h Sciences, Uni e si y o Jy äskylä, Jy äskylä, Finland. 3School o Heal h Sciences, Uni e si y o Tampe e, Tampe e, Finland. 4Depa men o Signal P ocessing, Tampe e Uni e si y o Technology, Tampe e, Finland. 5Cen e o Li e Cou se Heal h Resea ch, Uni e si y o Oulu, Oulu, Finland. 6Depa men o Medical Rehabili a ion, Oulu Uni e si y Hospi al, Oulu, Finland. Co espondence and eques s o ma e ials should be add essed o S.C. (email: [email p o ec ed]) Recei ed: 26 No embe 2015 accep ed: 31 Ma ch 2016 Published: 15 Ap il 2016 OPEN www.na u e.com/scien i ic epo s/ 2 Scien i ic RepoR s | 6:24540 | DOI: 10.1038/s ep24540 knowledge on he al e a ions in sys emic, adipose issue and skele al muscle me abolism in ela ion o insulin esis ance in no moglycemic indi iduals wi h a ying deg ee o adiposi y. The e o e, we aimed o in es iga e he sys emic me aboli e and gene exp ession p o iles o subcu aneous adipose issue and skele al muscle in ( as ing) no moglycemic women bu di e ing in insulin esis ance. Resul s Clinical cha ac e is ics o he low and high HOMA-IR g oups. The high HOMA-IR g oup had highe body weigh and BMI han he low HOMA-IR g oup (p < 0.01 o bo h, Table1). High HOMA-IR subjec s also had highe pe cen body a , o al and isce al a mass (p < 0.01 o bo h). No signi ican di e ence in hepa ic o in a-muscula a was ound. Fas ing glucose concen a ions we e wi hin he no moglycemic ange in he majo - i y o he subjec s (92%). HOMA-IR indices we e almos h ee imes highe in indi iduals in he high HOMA-IR g oup han in hose in he low HOMA-IR g oup. This di e ence was la gely due o he signi ican ly highe as ing insulin le els in he high HOMA-IR g oup (p < 0.001). Se um iglyce ide and lep in concen a ions we e highe and physical ac i i y and ae obic i ness was lowe in he high HOMA-IR han low HOMA-IR g oup (p < 0.05 o all), bu no di e ence in adiponec in was ound. Die a y in ake did no di e be ween he g oups, no was he e a signi ican di e ence in plasma ee a y acids. No g oup by gene a ion in e ac ion was ound on any o he a iables. low HOMA-IR (n = 55) high HOMA-IR (n = 55) p- alue G oup by gene a ionMean 95% CI Mean 95% CI An h opome y Age (yea s) 35.1 (35.9, 36.3) 36.0 (35.0, 36.9) 0.269 0.096 Heigh (cm) 165.7 (163.9, 167.7) 166.1 (164.5, 167.6 ) 0.837 0.562 Weigh (kg) 61.3 (58.4, 64.1) 67.3 (65.1, 69.5) 0.001 0.604 BMI (weigh (kg)/heigh (m)2) 22.3 (21.4, 23.2) 24.4 (23.7, 25.1) 0.002 0.497 Body composi ion Pe cen body a 29.5 (27.3, 31.6) 35.0 (32.9, 37.1) 0.001 0.209 FFM (kg) 40.6 (39.3, 42.0) 40.7 (39.6, 41.8) 0.944 0.417 FM (kg) 17.8 (15.5, 20.1) 23.7 (21.9, 25.5) 0.001 0.266 VAT (kg) 0.5 (0.47, 0.62) 0.7 (0.59, 0.70) 0.044 0.164 Li e a (%) 2.5 (0.9, 4.1 ) 2.7 (1.5, 4.0) 0.787 0.577 IMCL (%) 0.2 (0.16, 0.25) 0.2 (0.17, 0.23) 0.796 0.261 EMCL (%) 0.2 (0.15, 0.31) 0.3 (0.27, 0.39) 0.068 0.925 Me abolic bioma ke s s-glucose (mmol/l) 5.2 (5.0, 5.4) 5.5 (5.3, 5.6) 0.042 0.982 s-insulin (μU/ml) 4.1 (3.0, 5.1) 9.9 (9.0, 10.8) < 0.001 0.110 HOMA-IR 0.9 (0.6, 1.2) 2.4 (2.2, 2.7) < 0.001 0.112 Lipids FFA (mmol/l) 3.6 (3.0, 4.2) 4.0 (3.5, 4.5) 0.308 0.968 T iglyce ides (mmol/l) 0.9 (0.7, 1.0) 1.1 (1.0, 1.2) 0.013 0.250 Adipokines Lep in (ng/ml) 14.6 (9.0, 20.2) 27.4 (23.1, 31.7) 0.001 0.490 Adiponec in (μg/ml) 10.4 (18.2, 20.6) 10.1 (8.4, 11.8) 0.830 0.375 Die Ene gy (kcal) 1840.0 (1690, 1990) 1780.0 (1650, 1900) 0.523 0.583 P o ein (E%) 17.7 (16.3, 18.8) 18.3 (17.3, 19.3) 0.324 0.545 Fa (E%) 34.5 (31.4, 36.0) 32.3 (30.1, 34.0) 0.283 0.564 Ca bohyd a es (E%) 47.8 (44.8, 49.5) 49.4 (46.4, 50.3) 0.414 0.868 Physical ac i i y and i ness LTPA (hou s/week) 4.5 (3.8, 5.2) 3.6 (3.0, 4.1) 0.049 0.466 VO2max (ml/kg/min) 41.1 (36.8, 45.5) 35.1 (31.3, 38.9) 0.042 0.540 Table 1. Gene al cha ac e is ics in he low and high HOMA-IR g oups (MIXED model es ima ed ma ginal means wi h 95% con idence in e als a e gi en aking in o accoun gene ic simila i y and sha ed en i onmen (daugh e and mo he ) and con as es ima es’ p- alues we e used o localize he signi ican di e ences be ween he wo g oups and g oup by gene a ion in e ac ion). FFM = a - ee mass; FM = a mass; SAT = subcu aneous adipose issue; VAT = isce al adipose issue; IMCL = in a-myocellula lipids; EMCL = ex a-myocellula lipids; HOMA-IR = homeos a ic model assessmen o insulin esis ance; OGTT = o al glucose ole ance es ; Ma suda index = insulin sensi i i y index; FFA = ee a y acids; E% = pe cen age o o al ene gy in ake; LTPA = leisu e- ime physical ac i i y; VO2max = maximum oxygen up ake. www.na u e.com/scien i ic epo s/ 3 Scien i ic RepoR s | 6:24540 | DOI: 10.1038/s ep24540 Se um me aboli es. Se um me aboli e p o ile analysis e ealed many simila i ies bu also signi ican di - e ences be ween he high and low HOMA-IR g oups (Table2). The e we e no signi ican di e ences in he a y acid and phospholipid (Table2) and lipop o ein subclass concen a ions (Supplemen a y Table S1) be ween he g oups. Howe e , concen a ions o b anched-chain amino acids (BCAA = isoleucine, leucine and aline), a o- ma ic amino acids (AAA = phenylalanine and y osine), glyce ol and o osomucoid we e signi ican ly highe in he high HOMA-IR g oup han he low HOMA-IR g oup (p < 0.05 o all). Only phenylalanine showed g oup by gene a ion in e ac ion. The di e ences be ween he g oups emained signi ican a e adjus ing o age, o al o isce al a mass, physical ac i i y and ae obic i ness. To con i m ha he ci cula ing me aboli e le els we e no con ounded by di e ence in adiposi y, we in es- iga ed whe he he di e ences in me aboli es be ween he HOMA-IR g oups we e also consis en ly p esen low HOMA-IR (n = 55) high HOMA-IR (n = 55) pBH G oup by gene a ionMean 95% CI Mean 95% CI Low-molecula weigh me aboli es be ahyd oxybu y a e 0.049 (0.033, 0.066 0.061 (0.046, 0.077) 0.357 0.253 ace a e 0.039 (0.036, 0.042) 0.043 (0.040, 0.046) 0.081 0.785 ace oace a e 0.038 (0.032, 0.044) 0.034 (0.029, 0.040) 0.961 0.793 alanine 0.374 (0.353, 0.394) 0.399 (0.380, 0.418) 0.261 0.305 ci a e 0.098 (0.092, 0.104) 0.102 (0.096, 0.108) 0.406 0.923 c ea inine 0.049 (0.046, 0.053) 0.051 (0.048, 0.055) 0.197 0.311 glu amine 0.511 (0.494, 0.529) 0.533 (0.517, 0.550) 0.685 0.174 glyce ol 0.056 (0.047, 0.064) 0.074 (0.066, 0.082) 0.003 0.134 glycine 0.257 (0.239, 0.276) 0.276 (0.259, 0.292) 0.685 0.211 o osomucoid 1.225 (1.168, 1.281) 1.344 (1.291, 1.397) 0.045 0.443 his idine 0.050 (0.047, 0.053) 0.056 (0.053, 0.058) 0.362 0.329 isoleucine 0.038 (0.035, 0.041) 0.043 (0.040, 0.046) 0.040 0.959 leucine 0.061 (0.057, 0.064) 0.069 (0.066, 0.073) 0.005 0.813 aline 0.154 (0.144, 0.163) 0.174 (0.165, 0.183) 0.014 0.351 BCAAsum 0.253 (0.237, 0.268) 0.288 (0.273, 0.302) 0.009 0.666 phenylalanine 0.061 (0.058, 0.064) 0.066 (0.063, 0.069) 0.035 0.045 y osine 0.041 (0.037, 0.044) 0.047 (0.044, 0.050) 0.030 0.089 py u a e 0.068 (0.061, 0.075) 0.077 (0.070, 0.083) 0.064 0.915 lac a e 0.891 (0.799, 0.984) 0.967 (0.877, 1.057) 0.253 0.101 u ea 0.052 (0.045, 0.058) 0.049 (0.043, 0.055) 0.731 0.131 Lipid ex ac cons i uen s es e i ied choles e ol 2.691 (2.528, 2.855) 2.831 (2.675, 2.987) 0.147 0.422 ee choles e ol 1.020 (0.952, 0.1087) 1.097 (1.032, 1.161) 0.127 0.386 omega3 a y acids 0.361 (0.316, 0.406) 0.369 (0.326, 0.411) 0.699 0.551 omega6 a y acids 2.789 (2.626, 2.952) 2.920 (2.824, 3.136) 0.328 0.683 omega7and 9 a y acids 4.918 (4.510, 5.325) 5.422 (5.032, 5.812) 0.253 0.446 o al a y acids 8.068 (7.505, 8.630) 8.770 (8.233, 9.308) 0.259 0.699 linoleic acid 2.303 (2.157, 2.449) 2.501 (2.362, 2.641) 0.328 0.738 polyunsa u a ed a y acids 1.647 (1.493, 1.801) 1.690 (1.543, 1.837) 0.792 0.747 docosahexanoic acid 0.143 (0.122, 0.164) 0.145 (0.125, 0.165) 0.716 0.872 monounsa u a ed a y acids 2.183 (1.976, 2.389) 2.429 (2.232, 2.626) 0.253 0.370 o al phosphoglyce ides 0.620 (0.567, 0.673) 0.671 (0.620, 0.721) 0.685 0.763 phosphocholines 1.526 (1.413, 1.639) 1.622 (1.514, 1730) 0.685 0.700 sphingomyelines 0.202 (0.187, 0.216) 0.207 (0.193, 0.220) 0.685 0.520 omega3/ o al a y acid a io 4.453 (4.066, 4.840) 4.165 (3.795, 4.536) 0.458 0.173 omega6/ o al a y acid a io 34.87 (33.80, 35.95) 34.40 (33.36, 35.42) 0.676 0.206 omega7and9/ o al a y acid a io 60.67 (59.61, 61.73) 61.45 (60.43, 62.46) 0.458 0.076 a y acid leng h 17.92 (17.84, 18.00) 17.94 (17.87, 18.01) 0.857 0.264 Table 2. Se um low-molecula weigh me aboli es and lipid ex ac cons i uen s in he low and high HOMA-IR g oups(MIXED model es ima ed ma ginal means wi h 95% con idence in e als a e gi en aking in o accoun gene ic simila i y and sha ed en i onmen (daugh e and mo he ) and con as es ima es’ p- alues we e used o localize he signi ican di e ences be ween he wo g oups and g oup by gene a ion in e ac ion). P- alues a e adjus ed o mul iple compa isons using Benjamin-Hochbe g co ec ion. All me aboli es a e in mmol/l. www.na u e.com/scien i ic epo s/ 4 Scien i ic RepoR s | 6:24540 | DOI: 10.1038/s ep24540 in no mal weigh indi iduals. We ound ha se um BCAAs and o osomucoid we e signi ican ly highe in he high HOMA-IR g oup han he low HOMA-IR g oup (p < 0.05 o all) (Supplemen a y Table S2). No di e ence be ween he g oups in he o he me aboli es was ound. To al se um BCAA co ela ed wi h s insulin ( = 0.388), e en a e adjus ing o age, pe cen a mass and isce al adipose issue. The concen a ion o plasma ee a y acids was no associa ed wi h s insulin ( = 0.108), HOMA-IR ( = 0.168), a mass ( = 0.093) o isce al adipose issue ( = 0.153) (p > 0.05 o all). Adipose issue gene exp ession. To elucida e he me abolic pa hways cha ac e izing o con ibu ing o insulin esis ance, we s udied global ansc ip p o iles o adipose issue and skele al muscle. Mic oa ay anal- ysis e ealed 1093 di e en ially exp essed genes (688 up- egula ed and 405 down- egula ed) in he adipose issue o he high HOMA-IR g oup (Supplemen a y Table S4). Kyo o Encyclopedia o Genes and Genomes (KEGG) en ichmen analysis o he di e en ially exp essed genes (p < 0.05) iden i ied 9 down- egula ed and 15 up- egula ed pa hways. The up- egula ed pa hways in he adipose issue e ealed inc eased ac i a ion o immune esponse in ol - ing bo h inna e and adap i e immune sys ems (Table3). Lysosome was he mos up- egula ed pa hway (p = 5.6 × 10−13), wi h con ibu ions by genes in ol ing all aspec s o lysosome biogenesis and unc ion, includ- ing s uc u al genes (LAMP1, LAMP2 and LAPTM5), se e al lysosomal acid hyd olases and anspo p o eins o lipids and choles e ol (MCOLN1 and NPC2) as well as p o eins equi ed o lysosome acidi ica ion (ATP6V0D2 and ATP6AP1). Se e al chemokines (e.g., CCL2, CCL3, CCL4 and CCL5), which a e p oduced by inna e immune cells, and also by p e-adipocy es and ma u e adipocy es14 we e also p esen in he up- egula ed pa hways. Consis en ly, leukocy e ans-endo helial mig a ion, Fc gamma R-media ed phagocy osis, oll-like ecep o sign- aling, and he complemen sys em, (complemen C1 [C1S, C1QA, C1QB, C1QC]) we e up- egula ed in he high HOMA-IR subjec s. Up- egula ion o he B cell ecep o signaling pa hway was also obse ed, p o iding u he e idence o he in ol emen o he adap i e immune sys em and in il a ion o in lamma o y cells o he adipose issue in insulin esis ance. The mos down- egula ed pa hway in he adipose issue was aline, leucine and isoleucine deg ada ion (p = 1.1 × 10−7, Table4). The genes ha mapped o his pa hway encode he cy osolic and mi ochond ial com- ponen s o he pa hway and included bo h genes common o he deg ada ion o all BCAAs, namely isoleucine, leucine and aline (BCAT1, BCKDHB) and hose speci ic o he deg ada ion o leucine (MCC2, AUH), isoleucine (PCCA, PCCB) and aline (HIBADH, ALDH6A1). Fu he , we ound ha sys emic insulin esis ance was also associa ed wi h signi ican down- egula ion o he genes encoding p o eins ha play impo an oles in cellula ene gy homeos asis. These included a mas e egula o o mi ochond ial biogenesis and unc ion, pe oxisome p oli e a o -ac i a ed ecep o -gamma coac i a o -1α (PPARGC1A) and mi ochond ial ace yl-coenzyme A ca - boxylase be a (ACACB), which is an impo an egula o o a y acid oxida ion and syn hesis. Fu he , pe oxi- some p oli e a o -ac i a ed ecep o alpha (PPARA), which p omo es he up ake, u iliza ion, and ca abolism o a y acids, was also down- egula ed in he high HOMA-IR g oup. Acco dingly, down- egula ion o he a y acid deg ada ion pa hway and ica boxylic acid cycle (TCA cycle) was obse ed. O he down- egula ed pa hways in ol ed a oma ic amino acid (phenylalanine and yp ophan) and sho -chain a y acid (p opanoa e) me abo- lism, and lysine biosyn hesis. To ensu e ou obse a ions we e no biased, we alida ed ou esul s wi h wo o he independen expe imen s in obese insulin- esis an subjec s by using gene exp ession omnibus (GEO) and he GSE2663715 and GSE2095016 and da a se s, espec i ely. The up- egula ed in lamma ion- ela ed genes in he adipose issue we e simila in ou s udy han in he s udy o So onen e al.15 and Ha dy e al.16 (Fig.1). In addi ion, he in lamma ion and ene gy me abolism- ela ed pa hways in he adipose issue we e simila ly up- and down- egula ed han in obese insulin esis an women in he s udy o So onen e al.15. We hen assessed he associa ions be ween adipose issue gene exp essions and clinical ai s in non-obese indi iduals. Cha ac e is ics o he s udy pa icipan s a e p esen ed in Supplemen a y Table S3. The mean cen oid o he BCAA ca abolism pa hway in he adipose issue was associa ed wi h insulin sensi i i y (Ma suda index) and s-insulin (p < 0.05 o all) (Fig.2). These associa ions emained signi ican a e adjus ing o age and pe cen a mass o isce al a mass. The BCAA ca abolism pa hway co ela ed closely wi h mi ochond ial espi a ion and biogenesis, i.e., wi h he TCA cycle and PPARGC1A (p < 0.001) (Fig.3). Maximum oxygen up ake (VO2max) co ela ed wi h he BCAA ca abolism ( = 0.543) and he TCA cycle ( = 0.522) (p < 0.05 o all). In lamma ion pa hways we e coo dina ely up- egula ed wi h insulin esis ance and adipokines, e.g., he chemokine signaling pa hway co ela ed wi h insulin sensi i i y (Ma suda index) ( = − 0.807), s-insulin ( = 0.858), s-adiponec in ( = − 0.598) and lep in ( = 0.428) (p < 0.001 o all). The chemokine signaling pa hway displayed signi ican associa ions wi h he TCA cycle and BCAA ca abolism ( = − 0.812 and = − 0.788, espec i ely, p < 0.001 o bo h). Skele al muscle gene exp ession and signaling p o ein phospho yla ion. Unexpec edly, when he ansc ip omic da a o skele al muscle we e s udied, no di e en ially exp essed genes we e ound be ween he low and high HOMA-IR samples. Howe e , since skele al muscle is he p ima y si e o insulin-s imula ed glucose disposal, we s udied whe he he e we e di e ences in he phospho yla ion le els o se e al p o eins ela ed o glucose up ake, insulin signaling and mi ochond ial ene gy me abolism. No di e ences in he phos- pho yla ion le els o insulin ecep o β o i s downs eam a ge Ak we e ound (Supplemen a y Figu e S1). The le el o phospho yla ed AS160, which p omo es ansloca ion o glucose anspo e s o he cell memb ane, was also simila be ween he g oups. In addi ion, no di e ences in he exp ession o mi ochond ial espi a o y chain complex subuni s, namely ATP5A, UQCRC2, MTCO1, SDHB and NDUFF88 be ween he g oups we e ound (Supplemen a y Figu e S1). www.na u e.com/scien i ic epo s/ 5 Scien i ic RepoR s | 6:24540 | DOI: 10.1038/s ep24540 Discussion In his s udy wi h young and middle-aged no moglycemic women, we ound ha insulin esis ance was asso- cia ed wi h inc eased se um BCAA le els, independen o obesi y. Consis en wi h his, we ound a signi ican down- egula ion o genes ela ed o BCAA ca abolism and mi ochond ial ene gy me abolism, concu en ly wi h inc eased exp ession o in lamma ion- ela ed genes in he adipose issue. Plasma ee a y acids a e commonly ele a ed in obese indi iduals due o inc eased adipose issue lipolysis17. Ele a ed ci cula ing ee a y acids may accumula e in o he insulin- esponsi e issues, such as skele al muscle and li e (whe e hey in e e e wi h insulin signaling and cause insulin esis ance)18. Howe e , we ound ha in no moglycemic women wi h a ying deg ee o adiposi y, insulin esis ance was no associa ed wi h inc eased plasma ee a y acids, bu ins ead wi h inc eased se um amino acid concen a ions. Ou obse a ion pa ly ag ees wi h a ecen s udy in no mal weigh subjec s disco dan o insulin sensi i i y19, and a me a-analysis by Ka pe e al.20. I is emp ing, he e o e, o sugges ha he con ibu ion o ci cula ing ee a y acids o insulin esis ance may be ela i ely small. Inc eased se um BCAA concen a ions ha e been associa ed wi h obesi y- ela ed insulin esis ance in ea lie s udies21–23. Ou esul s sugges ha pe u ba ions in sys emic BCAA homeos asis a e ela ed o insulin esis - ance a he han o obesi y pe se, since signi ican di e ence in hese amino acids was obse ed be ween he high and low HOMA-IR g oups also in no mal weigh indi iduals (Table S2). The a e age di e ence in o al BCAA be ween low and high HOMA-IR g oups was ~14% in whole s udy popula ion and ~10% in no mal weigh indi iduals. Whe he such di e ence is physiologically meaning ul is no clea . Howe e , Sunny e al.24 ecen ly demons a ed ha insulin-s imula ed inc eases (10–20%) in plasma BCAA co ela ed signi ican ly wi h insulin esis ance indices in humans. They concluded ha such small bu ch onic inc ease in ci cula ing BCAA wi h insulin esis ance may be su icien o dis up signaling e en s in he mi ochond ia o he muscle and li e he eby con ibu ing o mi ochond ial dys unc ion. Howe e , ou s udy canno show empo al ela ionships, al hough ou esul s a e compa ible wi h p e ious s udies25,26, which ha e sugges ed ha BCAAs associa e wi h insulin esis ance. I is unclea why se um BCAA is ele a ed in obesi y and insulin esis ance. Di e ences in die 23, p o ein u no e (muscle loss)27 o li e a ( a y li e disease) can a ec ci cula ing le els o amino acids. Howe e , in ou s udy no di e ence in die (p o ein in ake), a - ee mass o li e a con en was obse ed be ween he low and high HOMA-IR g oups. The posi i e co ela ion o BCAA wi h insulin concen a ion in ou s udy is line P- alue Coun Size Pa hway name Gene Names 5.6 × 10−13 36 121 Lysosome ACP5, AP1B1, ARSB, ATP6AP1, ATP6V0B, ATP6V0D2, CD68, CTSA, CTSB, CTSG, CTSH, CTSS, CTSZ, DNASE2B, FUCA1, GAA, GBA, GLA, GLB1, GM2A, GUSB, HEXB, LAMP1, LAPTM5, LGMN, MAN2B1, MCOLN1, NAGA, NPC2, PLA2G15, PPT1, PSAP, LAMP2, SLC11A2, SMPD1, TCIRG1 2.2 × 10−528 156 Phagosome ACTB, ACTG1, ATP6AP1, ATP6V0B, ATP6V0D2, ATP6V1B2, C1R, CD14, CLEC7A, CORO1A, CTSS, CYBA, FCGR2A, FCGR2B, FCGR3A, ITGB2, ITGB5, LAMP1, MARCO, MSR1, NCF2, NCF4, SEC61A1, TCIRG1, TUBA1C, TUBB2A, TUBB2B, VAMP3 2.0 × 10−432 189 Chemokine signaling pa hway ADCY6, ADCY7, ADRBK2, ARRB2, CCL13, CCL18, CCL19, CCL2, CCL22, CCL3, CCL4, CCL5, CCR1, CXCL10, CXCL16, CXCR4, DOCK2, FGR, GNAI1, GNB4, GNG2, GRB2, HCK, PIK3R5, PREX1, PRKCB, PRKX, RAC2, STAT2, STAT3, TIAM1, VAV1 7.4 × 10−421 117 Leukocy e ansendo helial mig a ion ACTB, ACTG1, ACTN1, CXCR4, CYBA, EZR, F11R, GNAI1, ICAM1, ITGAL, ITGB2, MMP9, MSN, MYL9, NCF2, NCF4, PIK3R5, PRKCB, RAC2, THY1, VAV1 2.5 × 10−316 95 Fc gamma R-media ed phagocy osis ARF6, CFL2, DOCK2, FCGR2A, FCGR2B, FCGR3A, HCK, LAT, PIK3R5, PPAP2B, PRKCB, PTPRC, RAC2, SCIN, SYK, VAV1 1.9 × 10−317 102 Toll-like ecep o signaling pa hway CCL3, CCL4, CCL5, CD14, CD86, CXCL10, IRAK1, LBP, LY96, MAP2K3, MAPK10, PIK3R5, SPP1, TLR1, TLR5, TLR7, TLR8 2.8 × 10−310 65 Glycolysis/Gluconeogenesis ADH1B, ALDH2, ALDH3B1, ALDH7A1, ENO1, GALM, HK3, PDHB, PFKP, PGAM2 7.8.0 × 10−35 17 Renin-angio ensin sys em AGTR1, ANPEP, CTSA, CTSG, NLN 3.3 × 10−36 17 O he glycan deg ada ion FUCA1, FUCA2, GBA, GLB1, HEXB, MAN2B1 1.3 × 10−212 75 B cell ecep o signaling pa hway BLNK, DAPP1, FCGR2B, GRB2, NFKBIE, PIK3AP1, PIK3R5, PRKCB, PTPN6, RAC2, SYK, VAV1 1.3 × 10−25 19 Glycosaminoglycan deg ada ion ARSB, GLB1, GUSB, HEXB, HPSE 1.1 × 10−29 48 Amino suga and nucleo ide suga me abolism CHI3L1, CHIT1, GNE, GNPDA1, HEXB, HK3, NAGK, NPL, PMM1 1.7 × 10−211 69 Complemen and coagula ion cascades C1QA, C1QB, C1QC, C1R, C1S, C3AR1, C5AR1, F13A1, PLAU, PLAUR, SERPINE1 3.5 × 10−217 136 Na u al kille cell media ed cy o oxici y BID, CD48, FCER1G, FCGR3A, GRB2, HCST, ICAM1, ITGAL, ITGB2, LAT, PIK3R5, PRKCB, PTPN6, RAC2, SYK, TYROBP, VAV1 1.2 × 10−225 200 Focal adhesion ACTB, ACTG1, ACTN1, BIRC3, CCND2, COL6A1, COL6A2, COL6A6, FLNA, GRB2, ITGB5, LAMB3, MAPK10, MYL9, PDGFA, PIK3R5, PPP1CA, PRKCB, PTEN, RAC2, SPP1, TNC, VAV1, VEGFA, ZYX Table 3. Up- egula ed pa hways in he adipose issue o high HOMA-IR g oup. Coun = Amoun o di e en ially exp essed genes ha map in pa hway. Size = To al amoun o genes in ol ed in pa hway. www.na u e.com/scien i ic epo s/ 6 Scien i ic RepoR s | 6:24540 | DOI: 10.1038/s ep24540 wi h ea lie epo s, which ha e sugges ed ha hese amino acids may s imula e insulin sec e ion om he pan- c eas21,28,29. On he o he hand, ele a ed insulin may inc ease ci cula ing BCAA, possibly by a enua ing BCAA ca abolism in di e en issues as sugges ed by Sunny e al.24. Indeed, obesi y- ela ed inc eases in ci cula ing BCAAs ha e been associa ed wi h dec eased BCAA ca abolism in adipose issue30. He e, we showed signi ican down- egula ion o he BCAA deg ada ion pa hway genes in he adipose issue o no moglycemic subjec s wi h high insulin esis ance. The ac ha he e was no di e ence in a e age BMI o pe cen body a be ween he low and high HOMA-IR g oups sugges s ha down- egula ion o he BCAA ca abolism pa hway was no a ibu able o adiposi y alone. Thus, he dec ease in he BCAA ca abolism can p obably be asc ibed o educed mi ochon- d ial espi a ion and biogenesis (as indica ed by he close co ela ion o he BCAA ca abolism wi h he TCA cycle and PPARGC1A genes). Since physical ac i i y and ae obic i ness a e known o imp o e insulin sensi i i y and ene gy me abolism31, i is possible ha di e ences in physical ac i i y and ae obic i ness may ha e ampli ied he obse ed di e ences in gene exp ession be ween he low and high HOMA-IR g oups. The close co ela ion be ween VO2max and BCAA ca abolism and he TCA cycle u he suppo his no ion. G owing e idence indica es ha obesi y-associa ed low-g ade in lamma ion o adipose issue con ibu es o he de elopmen o insulin esis ance32. Ou esul s complemen his no ion by showing ha up- egula ed in lamma ion- ela ed genes we e closely associa ed wi h insulin esis ance, se um adiponec in and lep in also in no moglycemic indi iduals, e en a e adjus ing o measu es o adiposi y. Ea lie s udies ha e demons a ed ha plasma adiponec in and lep in le els a e associa ed wi h insulin esis ance independen o a mass33–35, and hey may play a key ole in he egula ion o in lamma ion and immuni y36. In line wi h ea lie s udies37,38, we ound ha he inna e in lamma o y componen ela ed o mac ophages was coo dina ely up- egula ed wi h insu- lin esis ance. The physiological ole o mac ophages is p obably o clea adipose deb is h ough he p ocess o phagocy osis and ac i a e he adap i e immune sys em39. Acco dingly, we ound ha in e ms o o e -exp essed genes he mos up- egula ed pa hways in he adipose issue we e lysosome and phagosome pa hways. These indings, ( oge he wi h ac i a ion and in il a ion o lymphocy es, a oll-like ecep o signaling pa hway, and he complemen sys em) sugges a s a e o ch onic low-g ade in lamma ion in he adipose issue. Fu he mo e, he ch onic in lamma ion may also in pa explain he obse ed impai men s in adipose issue ene gy me abolism (as indica ed by he close in e se co ela ion o chemokine signaling genes wi h BCAA ca abolism and he TCA cycle genes). Impai ed insulin-media ed skele al muscle glucose up ake40 and in amyocellula lipid concen a ions41 a e majo con ibu o s o insulin esis ance and ype 2 diabe es. In ou s udy, in amuscula iglyce ides we e no inc eased in subjec s wi h high HOMA-IR. Consis en wi h his, no abe an gene exp ession in indi iduals wi h high HOMA-IR was ound. To con i m hese indings, we u he s udied whe he he e we e di e ences in he phospho yla ion le els o se e al signaling p o eins ela ed o glucose me abolism. No signi ican di e ences in he phospho yla ion le els o insulin ecep o β and i s downs eam a ge Ak , we e ound, no was he e any signi ican di e ence in he le el o phospho yla ed AS160, which p omo es ansloca ion o glucose anspo e s o he cell memb ane. P e ious s udies ha e epo ed educed muscle ansc ip le els ela ed o oxida i e me ab- olism in diabe ic indi iduals compa ed o heal hy con ols42. We ound no di e ence in mi ochond ial espi a o y chain complex subuni s be ween he low and high HOMA-IR g oups. These indings sugges ha in he as ing s a e glucose and mi ochond ial ene gy me abolism is no signi ican ly al e ed in he skele al muscle in ea ly s ages o insulin esis ance. Howe e , since bo h acu e hype insulinemia43 and hype glycemia44 ha e been shown o induce ansc ip ional and ansla ional egula ion o glucose and ene gy me abolism in he skele al muscle, i may be ha signi ican di e ences could exis du ing hype insulinemic-euglycemic clamp, glucose challenge o mixed meal eeding. This s udy has some limi a ions. Fi s , he numbe o pa icipan s in ou s udy was ela i ely small and consis ed solely o women. Despi e his we we e able o iden i y s a is ically signi ican di e ences in se um P- alue Coun Size Pa hway name Genes 1.1 × 10−717 44 Valine, leucine and isoleucine ca abolism ACADM, ACADSB, ALDH2, ALDH6A1, ALDH7A1, AUH, BCAT1, BCKDHB, DLD, HADH, HIBADH, IL4I1, MCCC2, MUT, OXCT1, PCCA, PCCB 3.7 × 10−410 32 P op iona e me abolism ACACB, ACADM, ACSS3, ALDH2, ALDH6A1, ALDH7A1, MUT, PCCA, PCCB, SUCLG2 2.9 × 10−36 17 Phenylalanine me abolism ALDH3B1, AOC2, IL4I1, MAOB, MIF, PRDX6 4.4 × 10−310 43 Fa y acid deg ada ion ACADL, ACADM, ACADSB, ADH1B, ADH1C, ALDH2, ALDH7A1, CPT1A, HADH, PECI 1.2 × 10−29 42 T yp ophan me abolism ALDH2, ALDH7A1, CAT, CYP1B1, HADH, IL4I1, KMO, KYNU, MAOB 1.6 × 10−27 30 Ci a e cycle (TCA cycle) DLD, DLST, IDH3B, PC, PDHB, SDHB, SUCLG2 2.9 × 10−28 41 Ty osine me abolism ADH1B, ADH1C, ALDH3B1, AOC2, COMT, IL4I1, MAOB, MIF 1.7 × 10−22 3 Lysine biosyn hesis AASS, ALDH7A1 4.7 × 10−24 18 Glyoxyla e and dica boxyla e me abolism HYI, MUT, PCCA, PCCB 1.9 × 10−212 68 Adipocy okine signaling pa hway ACACB, ADIPOQ, ADIPOR1, CPT1A, MAPK10, NFKBIE, PPARA, PPARGC1A, PRKAG1, SLC2A4, STAT3, TNFRSF1B Table 4. Down- egula ed pa hways in he adipose issue o high HOMA-IR g oup. Coun = Amoun o di e en ially exp essed genes ha map in pa hway. Size = To al amoun o genes in ol ed in pa hway. www.na u e.com/scien i ic epo s/ 7 Scien i ic RepoR s | 6:24540 | DOI: 10.1038/s ep24540 me aboli es and gene exp essions in adipose issue be ween he g oups. In addi ion, he pa icipan s we e ca e- ully selec ed in o de o minimize con ounding ac o s and gene ic a iabili y. Fu he mo e, gene exp ession in he adipose issue o pa icipan s wi h high HOMA-IR was alida ed by wo independen expe imen s wi h non-obese ype 2 diabe ic and obese insulin- esis an subjec s. Thus, we belie e ha ou esul s a e no biased and ha he gene exp ession da a can be iewed wi h con idence. Finally, he majo i y o he pa icipan s we e wi hin he no mal as ing glycemic ange and exclusion o hose wi h impai ed as ing glucose did no change he esul s. This ga e us he possibili y o iden i y he bioma ke s associa ed wi h sys emic insulin esis ance in i s ea ly s age. Ou da a demons a e ha se um a y acids, in a-myocellula lipids and li e a con en we e no ele a ed in no moglycemic women wi h high HOMA-IR. Ins ead, we show ha impai ed insulin sensi i i y was associ- a ed wi h a signi ican inc ease in se um BCAA concen a ion, up- egula ion o in lamma ion- ela ed genes and down- egula ion o genes ela ed o BCAA ca abolism and mi ochond ial ene gy me abolism in adipose issue. These indings sugges ha adipose issue in lamma ion and mi ochond ial dys unc ion may be ea ly e en s in he de elopmen o sys emic insulin esis ance. Fu he s udies a e needed o de e mine he ini ial ac o (s) ha igge he ansc ip ional changes ha lead o hese me abolic al e a ions. Figu e 1. Up- egula ed in lamma ion- ela ed genes in he adipose issue in insulin esis an compa ed o insulin sensi i e g oup. The y-axis shows he old change o gene exp ession o up- egula ed genes in high HOMA-IR subjec s o he p esen s udy (A) and obese insulin esis an subjec s in he s udy o So onen e al. (B)15 and Ha dy e al.16 (C). www.na u e.com/scien i ic epo s/ 8 Scien i ic RepoR s | 6:24540 | DOI: 10.1038/s ep24540 Ma e ials and Me hods S udy subjec s. This a icle is pa o a la ge amily s udy wi h 282 pa icipa ing amilies and has been desc ibed elsewhe e45. A subg oup o amilies (n = 74), comp ising 222 indi iduals (daugh e , mo he and a he ) wi h no ype I/II diabe es o amily his o y ( i s deg ee ela i e) o diabe es, ca diac diseases, au oimmune dis- eases o majo li e (cance , hepa i is) diseases we e con ac ed by le e o an addi ional s udy aimed a iden i y- ing bioma ke s associa ed wi h insulin esis ance and li e a accumula ion. A o al o 184 indi iduals esponded o ou in i a ion, o whom 163 (53 a he s, 53 mo he s and 57 daugh e s) a ended he labo a o y es s. Fo his epo , all he a he s we e excluded in o de o educe he a iabili y in gene ic a chi ec u e, lea ing only he mo he s and daugh e s (mo he s = 53 and daugh e s = 57). Mo he s and daugh e s did no di e in measu es o adiposi y, insulin esis ance, se um iglyce ides, a y acids o amino acids. The e o e, all da a including he me abolome and mic oa ay da a we e pooled be o e he pheno ypic analysis and all esul s we e adjus ed o age and amilia i y. Fu he , o minimize he me abolic al e a ions occu ing a di e en s ages o he mens ual cycle, blood samples we e collec ed om he women wi h egula menses be ween 2 and 5 days a e (mens u- a ion). Twen y- wo pa icipan s we e in ea ly pos menopause bu none we e on ho monal eplacemen he apy. Including o excluding hese pa icipan s did no in luence he esul s. All subjec s we e clinically eu hy eo ic. The s udy p o ocol was app o ed by he e hics commi ee o he Cen al Finland Heal h Ca e Dis ic . A w i en in o med consen was ob ained om all pa icipan s, and all expe imen s we e pe o med in acco dance wi h ele an guidelines and egula ions. Me hods. A de ailed desc ip ion o he backg ound in o ma ion and me hods a e p o ided in supplemen- a y ex S1. In sho , backg ound in o ma ion including medical his o y, cu en heal h s a us and physical ac i i y was collec ed ia sel -adminis e ed ques ionnai es. Food consump ion and in akes o o al ene gy and ene gy-yielding nu ien s we e assessed om h ee-day ood eco ds. All measu emen s we e pe o med in he mo ning a e o e nigh as ing. Venous blood samples we e ob ained o he analyses o glucose, insulin, Figu e 2. Co ela ions be ween he mean cen oids o he BCAA ca abolism pa hway and s-insulin and insulin sensi i i y (Ma suda index). The alues o s-insulin and he Ma suda index we e ans o med o no mal dis ibu ion by na u al loga i hms. Each do ep esen s an indi idual and he line is a linea eg ession i line. Figu e 3. Co ela ion be ween he mean cen oids o he BCAA ca abolism pa hway and he TCA cycle and PPARGC1A. Each do ep esen s an indi idual and he line is a linea eg ession i line. www.na u e.com/scien i ic epo s/ 9 Scien i ic RepoR s | 6:24540 | DOI: 10.1038/s ep24540 non-es e i ied a y acids, lep in, and adiponec in. The HOMA-IR index (homeos a ic model assessmen o insulin esis ance) was calcula ed as ( as ing glucose x as ing insulin/22.5). Acco ding hei HOMA-IR alues (median = 1.57), he subjec s we e di ided in o low (n = 55) and high (n = 55) g oups. Body composi ion was assessed by DXA, subcu aneous and in a-abdominal adipose issue by MRI46, and ec opic a o li e , mus- cle in a-myocellula lipid (IMCL) and ex a-myocellula lipid (EMCL) by 1HMRS47. Se um me aboli es we e assessed by NMR spec oscopy48. Maximum oxygen up ake (VO2max, ml/kg/min) was assessed by a bicycle e gom- e e es . In addi ion, supe icial abdominal subcu aneous adipose issue and skele al muscle ( as us la e alis) biopsies we e ob ained om 24 indi iduals o assess he di e ences in global gene exp ession p o iles and muscle p o ein exp ession be ween he low and high HOMA-IR g oups. Fu he mo e, a 75-g o al glucose ole ance es (OGTT) was pe o med o subjec s wi h issue biopsies o assess whole body insulin sensi i i y49. Mic oa ay measu emen s we e analyzed by using he Robus Mul ia ay A e aging (RMA) algo i hm in he Bioconduc o R package a y50–52. The Limma R package was used o di e en ially exp essed genes (DEGs). Raw p alues we e adjus ed o con ol o he alse disco e y a e (FDR) using he me hod o Benjamini and Hochbe g53 ( o mo e de ailed in o ma ion on mic oa ay and gene en ichmen analysis see supplemen a y ex S1). S a is ical me hods. Be o e each analysis, con inuous da a we e checked o no mali y by Shapi o-Wilk’s es using PASW s a is ics e sion 21 (IBM Co po a ion, USA). I da a we e no no mally dis ibu ed, hei na u al loga i hms we e used. Clinical cha ac e is ics and se um me aboli es we e compa ed using an independen -samples - es . Since he da a we e om a amily s udy, he amilia i y (gene ic and en i onmen al (household) simila i y) was con olled by using linea mixed model o compa e le els o he ou come a iables be ween he low and high HOMA-IR g oups. Con as es s we e used in mixed models o assess he e ec o gene a ion while con olling o dependency among amily membe s wi h andom e ec s. P- alues we e adjus ed o con ol o he alse disco e y a e (FDR) using he me hod o Benjamini and Hochbe g when compa ing me aboli es be ween he low and high HOMA-IR g oups53. Pea son co ela ion analysis was used o de e mine he ela ionship be ween clinical cha ac e is ics, se um me aboli es and adipose issue gene exp ession. S a is ical signi icance was se a p < 0.05. Re e ences 1. DeF onzo, R. A. & T ipa hy, D. Skele al muscle insulin esis ance is he p ima y de ec in ype 2 diabe es. Diabe es ca e 32 Suppl 2, S157–163, doi: 10.2337/dc09-S302 (2009). 2. Muoio, D. M. & Newga d, C. B. Mechanisms o disease:Molecula and me abolic mechanisms o insulin esis ance and be a-cell ailu e in ype 2 diabe es. Na u e e iews. Molecula cell biology 9, 193–205, doi: 10.1038/n m2327 (2008). 3. Abbasi, F., B own, B. W., J ., Lamendola, C., McLaughlin, T. & Rea en, G. M. Rela ionship be ween obesi y, insulin esis ance, and co ona y hea disease isk. Jou nal o he Ame ican College o Ca diology 40, 937–943 (2002). 4. Lee, J. e al. Should cen al obesi y be an op ional o essen ial componen o he me abolic synd ome? Ischemic hea disease isk in he Singapo e Ca dio ascula Coho S udy. Diabe es ca e 30, 343–347, doi: 10.2337/dc06-1866 (2007). 5. Hollenbeck, C. & Rea en, G. M. Va ia ions in insulin-s imula ed glucose up ake in heal hy indi iduals wi h no mal glucose ole ance. The Jou nal o clinical endoc inology and me abolism 64, 1169–1173, doi: 10.1210/jcem-64-6-1169 (1987). 6. Ka elis, A. D., S -Pie e, D. H., Conus, F., Rabasa-Lho e , R. & Poehlman, E. T. Me abolic and body composi ion ac o s in subg oups o obesi y: wha do we know? The Jou nal o clinical endoc inology and me abolism 89, 2569–2575, doi: 10.1210/jc.2004-0165 (2004). 7. Fe annini, E. e al. Insulin esis ance and hype sec e ion in obesi y. Eu opean G oup o he S udy o Insulin Resis ance (EGIR). The Jou nal o clinical in es iga ion 100, 1166–1173, doi: 10.1172/JCI119628 (1997). 8. Boden, G. Obesi y, insulin esis ance and ee a y acids. Cu en opinion in endoc inology, diabe es, and obesi y 18, 139–143, doi: 10.1097/MED.0b013e3283444b09 (2011). 9. Lynch, C. J. & Adams, S. H. B anched-chain amino acids in me abolic signalling and insulin esis ance. Na u e e iews. Endoc inology 10, 723–736, doi: 10.1038/n endo.2014.171 (2014). 10. Wang, T. J. e al. Me aboli e p o iles and he isk o de eloping diabe es. Na u e medicine 17, 448–453, doi: 10.1038/nm.2307 (2011). 11. Wiklund, P. K. e al. Se um me abolic p o iles in o e weigh and obese women wi h and wi hou me abolic synd ome. Diabe ology & me abolic synd ome 6, 40, doi: 10.1186/1758-5996-6-40 (2014). 12. Guilhe me, A., Vi basius, J. V., Pu i, V. & Czech, M. P. Adipocy e dys unc ions linking obesi y o insulin esis ance and ype 2 diabe es. Na u e e iews. Molecula cell biology 9, 367–377, doi: 10.1038/n m2391 (2008). 13. Kahn, S. E., Hull, R. L. & U zschneide , K. M. Mechanisms linking obesi y o insulin esis ance and ype 2 diabe es. Na u e 444, 840–846, doi: 10.1038/na u e05482 (2006). 14. Kabi , S. M., Lee, E. S. & Son, D. S. Chemokine ne wo k du ing adipogenesis in 3T3-L1 cells: Di e en ial esponse be ween g ow h and p oin lamma o y ac o in p eadipocy es s. adipocy es. Adipocy e 3, 97–106, doi: 10.4161/adip.28110 (2014). 15. So onen, J. e al. Adipose issue gene exp ession analysis e eals changes in in lamma o y, mi ochond ial espi a o y and lipid me abolic pa hways in obese insulin- esis an subjec s. BMC medical genomics 5, 9, doi: 10.1186/1755-8794-5-9 (2012). 16. Ha dy, O. T. e al. Body mass index-independen in lamma ion in omen al adipose issue associa ed wi h insulin esis ance in mo bid obesi y. Su ge y o obesi y and ela ed diseases: o icial jou nal o he Ame ican Socie y o Ba ia ic Su ge y 7, 60–67, doi: 10.1016/j.soa d.2010.05.013 (2011). 17. Boden, G. & Shulman, G. I. F ee a y acids in obesi y and ype 2 diabe es: de ining hei ole in he de elopmen o insulin esis ance and be a-cell dys unc ion. Eu opean jou nal o clinical in es iga ion 32 Suppl 3, 14–23 (2002). 18. Shulman, G. I. Ec opic a in insulin esis ance, dyslipidemia, and ca diome abolic disease. The New England jou nal o medicine 371, 1131–1141, doi: 10.1056/NEJM a1011035 (2014). 19. A ne , P. & Ryden, M. Fa y Acids, Obesi y and Insulin Resis ance. Obesi y ac s 8, 147–155, doi: 10.1159/000381224 (2015). 20. Ka pe, F., Dickmann, J. R. & F ayn, K. N. Fa y acids, obesi y, and insulin esis ance: ime o a ee alua ion. Diabe es 60, 2441–2449, doi: 10.2337/db11-0425 (2011). 21. Felig, P., Ma liss, E. & Cahill, G. F., J . Plasma amino acid le els and insulin sec e ion in obesi y. The New England jou nal o medicine 281, 811–816, doi: 10.1056/NEJM196910092811503 (1969). 22. McCo mack, S. E. e al. Ci cula ing b anched-chain amino acid concen a ions a e associa ed wi h obesi y and u u e insulin esis ance in child en and adolescen s. Pedia ic obesi y 8, 52–61, doi: 10.1111/j.2047-6310.2012.00087.x (2013). 23. Newga d, C. B. e al. A b anched-chain amino acid- ela ed me abolic signa u e ha di e en ia es obese and lean humans and con ibu es o insulin esis ance. Cell me abolism 9, 311–326, doi: 10.1016/j.cme .2009.02.002 (2009). 24. Sunny, N. E. e al. C oss- alk be ween b anched-chain amino acids and hepa ic mi ochond ia is comp omised in nonalcoholic a y li e disease. Ame ican jou nal o physiology. Endoc inology and me abolism 309, E311–319, doi: 10.1152/ajpendo.00161.2015 (2015).