physiological insights gained from gene expression ... of biochemistry, ... important discoveries...

NU30CH06-Attie ARI 12 April 2010 23:23

RE V I E W

S

IN

AD V A

NC

E

Physiological InsightsGained from GeneExpression Analysisin Obesity and DiabetesMark P. Keller and Alan D. AttieDepartment of Biochemistry, University of Wisconsin-Madison, Madison,Wisconsin 53706-1544; email: [email protected]

Annu. Rev. Nutr. 2010. 30:6.1–6.24

The Annual Review of Nutrition is online atnutr.annualreviews.org

This article’s doi:10.1146/annurev.nutr.012809.104747

Copyright c© 2010 by Annual Reviews.All rights reserved

0199-9885/10/0821-0001$20.00

Key Words

diabetes, obesity, insulin resistance, insulin signaling, β-cells, muscle

Abstract

Microarray technology permits the interrogation of nearly all expressedgenes under a wide range of conditions. Patterns of gene expression inresponse to obesity and diabetes have yielded important insights into thepathogenesis of diabetes and its relationship to obesity. In muscle, mi-croarray studies have motivated research into mitochondrial function.In adipose tissue, clues have pointed to the importance of inflammationin obesity. New adipocyte-derived hormones involved in insulin resis-tance have been found; a notable example is retinol binding protein4. In liver, genes responsive to master regulators of lipid metabolismhave been identified. In β-cells, genes involved in cell survival, cellproliferation, and insulin secretion have been identified. These studieshave greatly expanded our understanding of mechanisms underlying thepathogenesis of obesity-induced diabetes. When combined with geneticinformation, microarray data can be used to construct causal networkmodels linking gene expression with disease.

6.1

Review in Advance first posted online on April 20, 2010. (Changes may still occur before final publication online and in print.)

Changes may still occur before final publication online and in print.

Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


DE: differentialexpression

Contents

INTRODUCTORY COMMENTSAND WORDS OF CAUTION . . . . 6.2

HIGHLIGHTS OFMICROARRAY-DERIVEDRESULTS IN THE CONTEXTOF OBESITY AND DIABETES . . 6.2

MUSCLE . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2ADIPOSE . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6LIVER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.8β-CELLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.10CORRELATION VERSUS

CAUSATION . . . . . . . . . . . . . . . . . . . . . 6.12GENETICAL GENOMICS:

EXPRESSION QUANTITATIVETRAIT LOCI . . . . . . . . . . . . . . . . . . . . . 6.12

IDENTIFICATION ANDVALIDATION OF CAUSATIVEGENES FOR FAT MASS INMICE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.13Genetic Architecture of Hepatic

Gene Expression in Humans andIts Use for Identification ofDisease-Causing Genes . . . . . . . . . 6.14

Variation in Genomic CopyNumber Contributes toeQTLs . . . . . . . . . . . . . . . . . . . . . . . . . 6.15

Concluding Remarks . . . . . . . . . . . . . . . 6.16

INTRODUCTORY COMMENTSAND WORDS OF CAUTION

The advent of microarray technology, ten yearsago, has had a profound impact on all of biologyand on the way in which biology experimentsare conducted. Within a short period of time,research organizations worldwide installed thenecessary equipment and acquired the exper-tise to conduct microarray experiments. Whenconsidered relative to the amount of data ob-tained, the experiments, if properly conducted,can be very informative and cost-effective (seesidebar, Guidelines for Planning a MicroarrayExperiment).

Numerous spectacular discoveries havebeen made through microarray experiments.In many cases, the discoveries were entirelyserendipitous; i.e., they were not anticipatedas part of an a priori hypothesis. Since manygenes whose expression is assessed in microar-ray experiments are unannotated, the context inwhich differential expression (DE) is detectedcan help to define their biological functions.

One undesirable consequence of these suc-cess stories is that it has motivated many inves-tigators to undertake microarray studies whenthey feel like they have exhausted all of theirideas. Essentially, they are employing the ex-periment to produce the hypothesis. Althoughimportant discoveries can still be made, this of-ten results in a large amount of data with manyhundreds of DE genes and a failure to producethe desired epiphany.

HIGHLIGHTS OFMICROARRAY-DERIVEDRESULTS IN THE CONTEXTOF OBESITY AND DIABETES

This review provides a selective sampling ofimportant discoveries made through microar-ray analysis of muscle, adipose tissue, liver, andpancreatic islets in the context of obesity and di-abetes. We apologize to the many outstandinginvestigators whose work is not described here.Our goal is to highlight a small number of keyobservations and wherever relevant, point outsome of the strengths of the experimental de-sign that were of special importance.

MUSCLE

Muscle is responsible for >80% of insulin-stimulated glucose clearance from the blood-stream (161). Consequently, many gene expres-sion studies aimed at identifying key genes thatmight be responsible for insulin resistance andtype 2 diabetes focused on muscle.

Early microarray studies on muscle from in-dividuals with insulin resistance and/or type2 diabetes were disappointing. When cor-rected for the large multiple-testing penalty,

6.2 Keller · Attie

R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


essentially no genes emerged as differentiallyexpressed between the experimental groups(139, 170). Mootha et al. approached this prob-lem with a dimension-reduction approach tothe data analysis (98). They organized the genesof the microarray into 146 lists according topathways. They profiled skeletal muscle fromthree groups of subjects; those with normal glu-cose tolerance, those with impaired glucose tol-erance, and those with frank type 2 diabetes.They ranked the DE lists according to the ex-tent of DE. They tested the hypothesis that theordering of the genes in this fashion would co-incide with a particular category of genes; i.e.,that there would be enrichment of the DE genesaccording to function, a method termed geneset enrichment analysis (GSEA) (145). This in-creased their statistical power to detect differ-ential expression by about 100-fold and also al-lowed them to place modestly DE genes into abroader context.

Application of the GSEA method identi-fied just one enriched pathway; genes whoseproducts play a role in oxidative phosphory-lation were modestly downregulated, but thekey finding was that the expression of thesegenes is coregulated. This coregulated expres-sion correlated with a modest (∼20%) re-duction in the expression of a transcriptionalcoactivator involved in mitochondrial biogene-sis, peroxisome proliferator-activated receptorγ-1α (PGC1α). A parallel study in MexicanAmericans also demonstrated reduced expres-sion of NRF-regulated oxidative phosphoryla-tion genes, accompanied by a ≈50% reductionin expression of PGC1α and β in both insulin-resistant and diabetic subjects (111). It might beimportant that in both of these studies, the nor-mal glucose tolerance groups had lower bodymass index (BMI) than the other two groups.Another potentially important difference be-tween the two studies was that the former studysurveyed muscle under hyperinsulinemic clampconditions, whereas the Mexican American sub-jects in the latter study were under fastingconditions.

The mitochondrial content of skeletalmuscle is responsive to environmental factors

GUIDELINES FOR PLANNINGA MICROARRAY EXPERIMENT

If you’re planning to do a microarray study, be clear about whyyou’re doing it, what you expect to obtain, what will you do withthe data, and ask yourself if this is really the best next experimentto do.

1. Learn as much as you can about the physiology and bio-chemistry of your system before doing a microarray study.This is critical in order to decide what tissue will be pro-filed and under what conditions. For obesity phenotypes,the decision is not easy because many tissues can influencebody weight (brain, intestine, liver, endocrine organs, andof course, adipose tissue). For example, a food intake mea-surement would help to understand if an obesity phenotypemight be due to a change in satiety control; this would mo-tivate the sampling of gene expression in the brain, perhapsthe hypothalamus.

2. The ability to discern distinct DE patterns is related to theeffect size of the observed phenomena and the biologicalexperimental variance. Manipulation of the experimentalconditions to maximize the effect size can greatly increasethe power to detect meaningful DE patterns. Pilot exper-iments to test various experimental conditions can help todetermine the optimal design for the microarray experi-ment.

3. Try to eliminate or minimize extraneous sources of ex-perimental variance. If this is not possible, systematicallyrecord factors that might introduce variation; they can beused as covariates in the analysis of the data. An impor-tant factor that affects the expression of perhaps as many as40% of all genes is sex (157). Many physiological phenom-ena show sexual dimorphism. For example, in most rodentmodels of diabetes, males are more severely affected thanfemales. A variable that is often overlooked is time of day. Alarge number of genes show circadian rhythms (154). Thus,whenever possible, all animals in an experiment should besampled at the same time of day (unless, of course, time ofday is an experimental variable).

4. The greatest difficulty in the statistical analysis of mi-croarray experiments is that they suffer from a largemultiple testing problem (143). This issue and the manysources of experimental variation demand that there bea sufficient number of biological replicates or a well-designed pooling scheme (74, 76). Ideally, one shoulddo a pilot study and measure the expression of several genes,

www.annualreviews.org • Gene Expression in Obesity and Diabetes 6.3

R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


covering a wide range in mRNA abundance, in order to beable to estimate the extent of variance within the experi-mental groups. This enables a power calculation to deter-mine then umber of replicates needed to detect a particularlevel of DE at a particular frequency. In our experience,fewer than five replicates is underpowered.

5. The dynamic range of microarrays (60 to 100-fold) is farless than that of real-time PCR or RNA-seq. The data areless reliable for transcripts of very low or very high abun-dance. Results obtained by microarray analysis should al-ways be verified by real-time PCR, and whenever possibleand relevant, protein abundance should also be evaluated.The overall correlation between mRNA and protein abun-dance is only about 0.4 (53). Discrepancies between mRNAand protein abundance can provide useful clues to investi-gate post-transcriptional control mechanisms.

6. Tissues have a mixture of cell types. It is important to bemindful of the mix of vascular tissue, blood-borne cells,tissue macrophages, and “minor” cell types when inter-preting tissue-derived gene expression data. Differentialexpression changes might reflect changes in cell number(e.g., increased vascularization) rather than true changesin gene expression. It might be desirable to perform mi-crodissection or cell type enrichment prior to extractingmRNA. This is especially important in the endocrine pan-creas, where islet cell mass comprises only approximately1% of total organ mass.

7. Of special relevance to those studying nutritional factors, itis critical to take special care in choosing the right controldiet. Often overlooked is the fact that high-fat diets havemore synthetic ingredients than ordinary chow and thusmight not be matched for fiber content, phytoestrogens,and simple carbohydrates (160).

GSEA: gene setenrichment analysis

such as exercise, aging, and caloric restriction(51). Nuclear magnetic resonance studies inhuman subjects showed a substantial declinein mitochondrial respiration associated withaging (112). Muscle from patients with type2 diabetes has a reduction in mitochondrialdensity (59, 73). Offspring of type 2 diabeticsalso show a decline in mitochondrial function,along with insulin resistance and an increasein the accumulation of myocellular lipid (12,99, 113). Shulman and coworkers (100) havesuggested that reduced mitochondrial respira-

tory capacity results in diminished fatty acidβ-oxidation, resulting in an accumulation oflipid signaling molecules, specifically diacyl-glycerols, which activate several atypical pro-tein kinase C isoforms that blunt insulin sig-naling through the serine phosphorylation ofIRS1.

Holloszy (60) has pointed out that skeletalmuscle has an oxidative capacity far in excessof what is normally required even during stren-uous exercise, enabling a 150-fold increase inoxygen consumption. He goes on to argue thatinsulin-resistant individuals have fewer mito-chondria, but their reduction in oxidative ca-pacity is not sufficient to present a bottleneckfor substrate oxidation. However, this does notaddress the potential link between mitochon-drial function/capacity and insulin sensitivity.

Rats and mice fed high-fat diets havelong been used as a model of human insulinresistance. However, recent studies show thatalong with the insulin resistance, these animalsexperience an increase in number of musclemitochondria and oxidative function (57, 150),especially fatty acid oxidation (165) (Figure 1a,see color insert). This occurs with an increasein PGC1α protein, but not mRNA (57), em-phasizing poor correlation between the tran-scriptional and post-transcriptional regulationof PGC1α. This result has not been uniformlyreproduced; Sparks et al. (138) observed adecrease in several mitochondrial oxidativephosphorylation (ox/phos) genes and PGC1α

protein in humans and mice on a high-fat diet.Even within the same laboratory, the effects ofhigh-fat diets on muscle mitochondrial geneexpression and PGC1α protein can be highlyvariable (D. Muoio, personal communication).

To directly address the role of free fatty acidsin the expression of genes for mitochondrialproteins, Holloszy’s team (44) performed dailyheparin injections of rats maintained on a high-fat diet to increase serum free fatty acid lev-els eightfold (heparin releases lipoprotein lipasefrom the endothelial lining of blood vessels).This resulted in an induction in the expres-sion of skeletal muscle genes encoding TCAcycle enzymes and oxidative phosphorylation

6.4 Keller · Attie

R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


proteins and increased recruitment of the tran-scription factor PPARδ to the gene encodingcarnitine palmitoyl transferase 1 (CPT1) (44).Overexpression of a VP16-PPARδ fusion geneleads to increased mitochondrial gene expres-sion (85, 158), suggesting that PPARδ is likelyto mediate the fatty acid induction of mito-chondrial biogenesis. However, Kleiner et al.(78) recently showed that although treatmentof mice with the PPARδ agonist GW501516induced CPT1, PDK4, and UCP3, some othermitochondrial genes, as well as PGC1α andPGC1β, were not induced, suggesting thatPPARδ induces some mitochondrial genes butdoes not induce mitochondrial biogenesis. Theresponse of CPT1 and PDK4 to the agonist wasreduced in skeletal muscle from PGC1α-nullmice. It therefore appears that PGC1α plays acrucial role in maintaining overall mitochon-drial content, whereas PPARδ acts specificallyon several mitochondrial genes, and in the pres-ence of PGC1α, PPARδ induces expression ofgenes that promote fatty acid oxidation.

More recent studies by Patti and coworkersshow that in obese diabetic, but not in nondia-betic insulin-resistant subjects, there is a down-regulation of mitochondrial oxidative phospho-rylation genes in skeletal muscle (M.E. Patti,personal communication). These results sug-gest that the response tends to be reactive to di-abetes and hyperglycemia rather than causal forinsulin resistance. Ritov et al. (121) showed thatin moderately obese (both diabetic and nondi-abetic) individuals (BMI >30), there is an im-balance between TCA cycle activity and elec-tron transport chain activity in skeletal muscle.This was inferred from the observation that cit-rate synthase activity was normal while NADHoxidase activity was reduced. These changeswere not responsive to caloric restriction orexercise, interventions that normally improveinsulin sensitivity and suggesting that obesitymight affect mitochondrial function indepen-dent of its effects on insulin sensitivity.

The microarray data in human skeletal mus-cle (98, 111) predicted that overexpression ofPGC1α in muscle should increase, whereas aknockdown or deletion of PGC1α should de-

crease, insulin sensitivity. Paradoxically, thesepredictions have not been borne out. Muscle-specific deletion of Pgc1α in mice, although itresulted in decreased mitochondrial respiratorycapacity, resulted in increased glucose uptakeinto muscle under hyperinsulinemic conditions(58). Muscle-specific overexpression led to in-creased mitochondrial density and respiratorycapacity and greater susceptibility to high-fatdiet-induced insulin resistance (25). The high-fat diet effects on mitochondrial fatty acid oxi-dation capacity are not linked to any changes ininsulin signaling, suggesting that these two pro-cesses may not be mechanistically linked (165).

Three other transgenic mouse models inwhich mitochondrial function is disruptedin muscle have increased insulin sensitivity(Figure 1b). Mice with a muscle-specificdeletion of Transcription factor A (Tfam)show improved glucose tolerance and en-hanced insulin-stimulated glucose uptake intoextensor digitorum longus muscle (164).

Apoptosis-inducing factor (AIF), in additionto its role in apoptosis, is essential for mitochon-drial electron transport, especially at Complex1 (67). Mice deficient in the Aif gene havea reduction in mitochondrial respiratory ca-pacity, as expected. They have a reduction infat oxidation but increased anaerobic glucosecatabolism. The mice also have increased in-sulin sensitivity (117).

Mice with a muscle-specific deletion of theThioredoxin interacting protein (Txnip) gene, likethe AIF-knockout mice, have a defect in aerobicglucose and fat oxidation, suggesting that theirmitochondria are dysfunctional (64). The miceare extraordinarily glucose tolerant, and thiscorrelates with an increase in phosphorylatedAkt in muscle in response to insulin. A proposedmechanism is that Txnip normally preventsthe thioredoxin-mediated activation of phos-phatase and tensin homolog (PTEN) throughreduction of its redox-sensitive disulfide.PTEN is a phophatidylinositol 3-phosphatasethat blunts insulin signaling. Parikh et al.(110) identified Txnip as a regulator of glucosemetabolism in humans, initially through a mi-croarray survey. They identified TXNIP as one


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


of just two genes repressed by insulin. The geneis induced by glucose in β-cells (18, 132) andParikh et al. discovered that it is upregulated inthe skeletal muscle of people with impaired glu-cose tolerance or type 2 diabetes (Figure 1c).These studies provide clues that link “redox cir-cuitry” (102) to insulin action and offer a plau-sible link between mitochondrial function andthe regulation of insulin signaling.

To determine whether the reduction in ex-pression of genes encoding mitochondrial res-piratory enzymes is a consequence of reducedinsulin signaling or of diabetes itself, Yechooret al. (173) surveyed gene expression in mus-cle from mice that were insulin-deficient andhyperglycemic due to streptozotocin treatmentversus mice with a muscle-specific knockoutof the insulin receptor. Diabetes was corre-lated with a decreased expression of these geneswhereas loss of receptor signaling was not, sug-gesting that hyperglycemia might suppress ex-pression of mitochondrial respiratory genes;i.e., diabetes might cause the changes in mi-tochondrial respiratory gene expression ratherthan vice versa.

It should be emphasized that although therole of mitochondrial function in insulin signal-ing in muscle is not fully understood, mitochon-drial dysfunction in β-cells can lead to diabetes(89). Mutations in mitochondrial genes havelong been known to cause insulin-dependentdiabetes (87). Stimulation of insulin secretionby glucose depends upon glucose catabolism,and in β-cells, this is primarily aerobic. Thiswas originally thought to be due to a paucityof lactate dehydrogenase in β-cells (129), but areappraisal has shown that this is not the case(130). Thus, anything that compromises β-cellmitochondrial function will interfere with glu-cose oxidation and thus insulin secretion via aneffect on glucose sensing.

ADIPOSE

Two early microarray studies of adipose tissuefrom Leptinob/ob mice showed a coordinateddecrease in the expression of genes encoding li-pogenic enzymes as well as the master regulator

of lipogenesis, the transcription factor Srebf1c(104, 137). In addition, both studies detectedan increase in the expression of genes relevantto inflammatory and acute-phase processes andgenes involved in adipocyte differentiation.These studies provided clues that obesityrepresents an inflammatory condition and thatadipocytes in obese animals have an alterationin their differentiation state. The decreasein lipogenic gene expression has also beenobserved in adipose tissue of obese humansubjects (34, 35, 131). It has been suggestedthat a limited lipogenic and/or adipogeniccapacity in adipocytes might lead to spillover ofexcess lipids to other tissues, where lipotoxicitycan contribute to the pathogenesis of diabetes(29, 103, 155).

The lipid spillover concept was tested byoverexpression of the leptin receptor in adi-pose tissue of db/db (leptin receptor–deficient)mice under the control of the adipocyte fattyacid–binding protein (aP2) promoter (155). Al-though the mice were just as hyperphagic asdb/db mice, they were much leaner. The re-markable result was that they were still insulinresistant, and this was correlated with steato-sis in nonadipose tissues: heart, liver, and islets.Unger and coworkers (156) have suggested thatfor adipose tissue to play its role as a high-capacity lipid storage depot, leptin must, in anautocrine fashion, suppress functions normallyassociated with brown adipose tissue; i.e., re-duce uncoupled lipid oxidation and thermoge-nesis and promote lipogenesis (Figure 2, seecolor insert).

It is now well established that obesity is as-sociated with the infiltration of proinflamma-tory (M1) macrophages in adipose tissue (82,83, 127, 162, 166). Many of the genes whose ex-pression are upregulated in response to obesityand are associated with inflammation are ex-pressed in macrophages rather than adipocytes.This underscores an important caveat in theinterpretation of microarray data from tissuesamples. Differences in mRNA abundance be-tween experimental groups may reflect changesin the cell composition of the tissue ratherthan changes in the expression of the genes

6.6 Keller · Attie

R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


in one particular cell type. Indeed, up to 40%of the cells in adipose tissue from obese an-imals are macrophages and lymphocytes. Inaddition to the altered expression profilederived from macrophage infiltration, themacrophages are able to elicit changes in theexpression of adipocyte genes, as demonstratedin coculture experiments (84).

Adipocytes secrete hormones, termedadipokines, that modulate insulin signaling inan autocrine fashion in adipose tissue or in aendocrine fashion toward other tissues, princi-pally muscle and liver (4). Leptin, adiponectin,and vaspin improve insulin sensitivity, whereasresistin, tumor necrosis factor-α, interleukin-6,plasminogen activator inhibitor-1, fasting-induced adipose factor (aka angiopoietin-like4), and retinol binding protein-4 (RBP4) bluntinsulin signaling.

The discovery of RBP4 and its role in insulinresistance illustrates a judicious application ofmicroarray technology (168). Although adiposetissue does not contribute to insulin-stimulatedglucose clearance nearly as much as muscle,overexpression or deletion of adipose tissueGlut4 results in altered insulin signaling and ac-tion in muscle and liver (2, 133). This does notoccur in isolated muscle tissue, implying that acirculating factor originating in adipose tissuemight be responsible for the changes in insulinresponsiveness in muscle and liver. Based onthis premise, Yang et al. (168) performed mi-croarray analyses of adipose tissue from miceoverexpressing Glut4 or with a deletion of theGlut4 gene specifically in adipocytes. They fil-tered their data, looking for genes that weremodulated in opposite directions under the twoconditions and encoded secreted proteins. Thefilter yielded just five mRNAs, including Rbp4.Follow-up studies showed that the Rbp4 pro-tein, when introduced into the circulation ofmice, is sufficient to blunt insulin signaling inmuscle and liver. RBP4-overexpressing miceare insulin resistant, and RBP4 knockout micehave enhanced insulin sensitivity (168). Thesefindings have been extended to human sub-jects, where serum RBP4 protein levels corre-late with various measures of insulin resistance

and metabolic syndrome even in large epidemi-ological studies (3, 5, 9, 10, 26, 45, 48, 49, 54,119, 172). Furthermore, genetic studies showthat a single nucleotide polymorphism (SNP) inthe RBP4 promoter is associated with increasedserum RBP4 levels and enhanced risk of type 2diabetes (101, 152).

In a follow-up of the original study of theadipocyte Glut4 knockout mice, gene set en-richment analysis was applied to identify anypathway that would be enriched among tran-scripts that are differentially regulated whenGlut4 is selectively overexpressed or knockedout in adipose tissue. The results showed thatpathways regulating metabolism of other majorfuels are enriched, suggesting that the differ-ences in glucose uptake in adipose tissue leadto alterations in genetic programs in fat thatregulate systemic fuel homeostasis (B.B. Kahn,personal communication).

The distribution of the increased adiposityassociated with obesity is highly variable acrossindividuals and populations. Excess adipositycan accumulate predominantly in visceral fator in subcutaneous depots distributed aroundthe body. Deposition in visceral fat is associ-ated with type 2 diabetes, whereas subcutaneousfat deposition appears to be much more be-nign (66). Microarray analysis of mouse sub-cutaneous versus visceral adipose tissue identi-fied 197 genes that are differentially expressedbetween the two depots (46). Twelve of thesegenes are involved in adipocyte development.These differences persisted after the adipocyteswere removed from the animals and incubatedin culture for six days, indicating that they arenot dependent upon extrinsic factors in theintact animal. Tbxl5, Shox2, En1, Sfrp2, andHoxc9 were more highly expressed in subcu-taneous adipose tissue, whereas Nr2f1, Gpc4,Thbd, Hoxa5, and HoxC8 were more highlyexpressed in visceral fat. Many of the resultsfrom the mouse study were confirmed in afollow-up study in 53 lean human subjects,although in some instances the direction ofthe differences was not the same as seen inthe mouse. The role of these genes in devel-opment led Gesta et al. (46) to suggest that


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


adipocytes residing in different fat depots mightbe derived from distinct adipocyte precursorcells.

LIVER

Microarray analysis helped to reliably delin-eate the genes regulated by two transcrip-tion factors that are master regulators of lipidmetabolism gene expression. Sterol regulatoryelement–binding protein (SREBP) is expressedas three isoforms, SREBP-1a, SREBP-1c, andSREBP2. The latter two are the predominantisoforms in intact liver. They are translated as aprotein that resides in the ER, but when ERmembrane cholesterol drops below a criticalthreshold, the SREBP proteins are transportedto the Golgi, where proteolytic cleavage re-leases a cytoplasmic domain, which translocatesto the nucleus, where it regulates gene expres-sion. The SREBP cleavage-activating protein(SCAP) helps to sense cholesterol and escortSREBPs to the Golgi (17).

The microarray study examined three setsof mouse livers; those from SCAP−/− miceand those from transgenic mice overexpress-ing either SREBP-1a or SREBP-2 (61). Thestudy confirmed that SREBP-1a primarily tar-gets genes involved in de novo fatty acid synthe-sis, fatty acid desaturation, and the generationof the necessary reducing equivalents to supportde novo lipogenesis. SREBP-2 primarily targetsenzymes that take acetate all the way throughthe biosynthetic pathway leading to cholesterol,as well as the LDL receptor. There were 33genes that were upregulated by both SREBPisoforms and downregulated in the SCAP−/−

mice. All but three of these genes encode en-zymes in the sterol or fatty acid biosyntheticpathway or proteins related to sterol transportor regulation of sterol metabolism (LDL recep-tor, Pcsk9, Insig-1).

The Leptinob/ob mouse is the most widelyused animal model for the study of morbidobesity. These mice have a global upregulationof expression of lipogenic enzymes and theirmaster regulator, Srebf-1c, in the liver (39, 70,72, 81). The upregulation of hepatic Srebf-1c

also occurs in lipodystrophic mice, which,like Leptinob/ob mice, are leptin deficient (135).Like leptin deficiency, leptin resistance alsoinduces the expression of Srebf-1c (147). Theseconsequences of leptin deficiency and leptinresistance are opposite to what occurs in theadipose tissue, suggesting from the transcriptprofile that the lipogenic burden appears to beshifted from adipose tissue to the liver (103).The upregulation of lipogenic enzymes asso-ciated with leptin deficiency in the Leptinob/ob

mouse, lipodystrophic mice (135), and inleptin-resistant rodent models (11, 70) is ac-companied by a failure of the increased insulinlevels to suppress hepatic gluconeogenesis.Thus, there is insulin insensitivity of the path-way by which insulin normally suppresses theexpression of gluconeogenic genes, togetherwith a normal lipogenic response of lipogenicgenes, chiefly through the induction of Srebf-1c.

Studies to evaluate the reversal of the lipidand diabetic phenotypes of leptin deficiencyshowed that intracerebroventricular (ICV) ad-ministration of leptin is much more effectiveat regulating hepatic gene expression than isperipheral leptin administration (6). At lowerdoses, ICV leptin reversed hyperglycemia andhyperinsulinemia in lipodystrophic mice with-out ameliorating the hepatic lipid abnormali-ties, suggesting that the hepatic lipid abnormal-ities are not causal for diabetes.

In an effort to define the downstreameffectors of leptin that mediate the lipogeniceffects of leptin deficiency, Cohen et al. (28)identified genes that were upregulated by leptindeficiency but not by high-fat-diet-inducedobesity and are specifically repressed by leptin.Stearoyl-CoA desaturase-1 (Scd1) emerged asthe gene that best fulfilled these criteria. Theinhibitory effect of leptin on Scd1 expressionoccurs in animals with a targeted deletion ofthe hepatic insulin receptor, indicating thatit does not depend on insulin signaling (13).These results predicted that deletion of theScd1 gene might rescue some of the phenotypesassociated with leptin deficiency. This predic-tion was tested in a naturally occurring animalmodel of Scd1 deficiency, the asebia mouse

6.8 Keller · Attie

R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


(28, 174), and in a mouse with a targeted disrup-tion of the Scd1 gene (108). When the Leptinob/ob

mutation was combined with either modelof Scd1 deficiency, the mice had a reversal oftheir hepatic steatosis, reduced body weight,and a striking increase in O2 consumption (28,108). Interestingly, despite these reversals inphenotypes normally associated with increasedsusceptibility to diabetes, the Leptinob/ob Scd−/−

mice actually have an increased susceptibility todiabetes (41). However, loss of Scd1 improvesmeasures of glucose tolerance and insulinsensitivity in animals that are leptin resistantrather than leptin deficient (97). On thebasis of these studies, several pharmaceuticalcompanies have chosen Scd1 as a target for thetreatment of obesity and diabetes (31).

A sharp reduction in the ability to synthesizemonounsaturated fatty acids de novo appearsto make animals more dependent on a criticalsupply of polyunsaturated fatty acids. Flowerset al. (42) showed that Scd1−/− mice on a very-low-fat diet developed cholestasis, hypercholes-terolemia, and severe liver disease. Microarrayanalysis revealed that the liver had an activationof the unfolded protein response and inflamma-tory pathways (43). These studies identify po-tential markers that should be monitored in thedevelopment of drugs targeted to Scd.

The foregoing studies all employed micewith a loss of Scd1 in all tissues. Several of thestudies strongly implied that the tissue sites re-sponsible for the Scd1-mediated effects on hep-atic steatosis, metabolic rate, body weight, andinsulin sensitivity would be the liver and/or adi-pose tissue. Yet, tissue-specific knockout of Scd1in the liver (96) or adipose tissue ( J.M. Ntambi,personal communication) did not reproduce thephenotype of the whole-body Scd1 knockoutmouse. Rather, a knockout of the gene in theskin reproduces most of the whole-body knock-out phenotype (123). This result points to oneof the trickiest aspects of microarray and geneexpression studies: inferring causality from cor-relative data. We address this issue later in thisarticle.

Glucose has a direct effect on the expres-sion of several genes involved in converting

carbohydrate into fatty acids (149). Early stud-ies focused on pyruvate kinase and identi-fied a site in its promoter that is responsiveto glucose (91, 151). The glucose effect isnot dependent on insulin. Rather, it is medi-ated by a pentose shunt metabolite of glucose,xylulose-5-phosphate (69, 106). This metabo-lite activates a phosphatase that dephosphory-lates and thus activates the transcription factorcarbohydrate-responsive element-binding pro-tein (ChREBP), which has two isoforms, knownas Mondo1 and Mondo 2. Both ChREBP iso-forms form a heterodimer with another pro-tein, Max-like factor X (Mlx) (140). Using adominant negative form of Mlx, Ma et al.(86) identified genes whose expression was sup-pressed by the inactivation of ChREBP. Theyfound that about half of those genes were in-duced by glucose. Many of the genes overlapwith those induced by Srebp-1c—Acc1, Fasn,G6pdh, and Me—but also include Pyruvate ki-nase, Malate dehydrogenase, Transketolase, Fruc-tokinase, Glycerol phosphate dehydrogenase, andMicrosomal triglyceride transfer protein, genes notresponsive to SREBP-1c. It is interesting thatfibroblast growth factor-21, a recently identi-fied hepatic hormone that affects whole-bodyglucose homeostasis, is also induced by glucoseand repressed by dominant-negative Mlx.

In contrast to muscle, liver expression ofoxphos genes seems to be responsive to in-sulin sensitivity. Patti’s group (114) showedthat in morbidly obese individuals (BMI >50),the expression of several mitochondrial oxphosgenes was downregulated; these were genes thatare normally induced by thyroid hormone inmuscle.

A recent study by Cheng et al. (24) sup-ports a causal relationship between insulin re-sistance and mitochondrial function in the liver.In mice lacking both Irs1 and Irs2, there was areduction in oxygen consumption in the pres-ence versus the absence of adinosine diphos-phate (respiratory control ratio). These physio-logical outcomes were associated with increasedexpression of genes regulated by the transcrip-tion factor Foxo1, a protein normally negativelyregulated by insulin-induced phosphorylation.


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


One of the Foxo1 target genes is Hmox1, whichencodes a protein that decreases the concentra-tion of mitochondrial heme, a key componentof electron transport complexes II and IV.

β-CELLS

The high content of ribonucleases in adult pan-creatic acinar cells poses an extra challenge inthe study of gene expression in pancreatic tis-sue. It makes it almost impossible to carry outin situ hybridization studies. Special care has tobe exercised in obtaining islets for microarrayexperiments (92, 163).

During the early stages of the onset oftype 2 diabetes, β-cells secrete a higher-than-normal level of insulin, but this is still notsufficient to compensate for insulin resistanceand titrate blood glucose. Over time, β-cellfunction and/or β-cell mass diminishes, leadingto more severe hyperglycemia; a feed-forwardcycle of disease pathogenesis (116). Proposedmechanisms to explain the link between hy-perglycemia and β-cell dysfunction fall intotwo categories: (a) consequences of glucosemetabolism and (b) glucose effects on gene ex-pression. The deleterious effects of excess glu-cose metabolism have been proposed to involvethe glucosamine pathway, formation of sor-bitol, and formation of reactive oxygen species(122, 146). It is interesting that reactive oxygenspecies have also been implicated in insulin re-sistance (63). In addition to glucose, fatty acidsmay play an important role in β-cell dysfunc-tion and death, a process termed lipotoxicity(79, 118).

The earliest human islet microarray exper-iment identified genes that were upregulatedin response to 16.7 versus 1.67 mM glucose.The most responsive gene was Thioredoxin in-teracting protein (Txnip; aka Vitamin D3 upregu-lated gene 1) (132), a gene discussed above in thecontext of muscle insulin signaling. Txnip wasoriginally identified as a gene whose abundanceis increased by 1,25-dihydroxy vitamin D3 inHL-60 leukemia cells (22), and consequentlyTxnip was found to act as a proapoptotic tumorsuppressor gene (56). Txnip is ubiquitously ex-

pressed and plays a role in controlling the redoxstate in a variety of cell systems (68, 107).

Consistent with the initial findings inhuman islets, incubation of primary mouseislets or INS-1 rat insulinoma cells with 25 mMglucose for 24 hours led to an induction of Txnipas well as of cleaved caspase 3 protein, resultingin an increase in apoptosis as judged by ter-minal deoxynucleotidyl transferase-mediateddUTP-biotin nick end labeling staining (94).In fact, Txnip overexpression was sufficient torender INS-1 cells more susceptible to apopto-sis (94), and microarray analysis revealed geneexpression changes consistent with increasedapoptosis in the Txnip-overexpressing β-cells(95). Isolated primary islets from the HcB-19mouse, a naturally occurring model of Txnipdeficiency (14, 65, 134), were completelyprotected from glucose toxicity–induced β-cellapoptosis (21). Interestingly, the β-cell protec-tive effects of the antidiabetic drug Exenatideseem to be mediated in part by a reduction ofTxnip expression (19).

The HcB-19 mice have also been found tohave mild hypoglycemia (20, 65, 134), and theirβ-cell mass is increased twofold, as assessedby morphometry and measurement of whole-pancreas insulin content (20). In addition,their Txnip deficiency also protected themfrom β-cell death induced by streptozotocin,a model of type 1 diabetes (20). Crossing theBTBR Leptinob/ob mouse as a model of obesity-induced diabetes with the Txnip-deficientHcB-19 mouse (double-mutant congenicBTBR lepob/obtxniphcb/hcb) further revealed thatlack of Txnip was also sufficient to protectagainst this severe model of type 2 diabetes,again primarily by reducing β-cell apoptosis(>98%) and preserving β-cell mass (20). Fi-nally, a β-cell-specific Txnip knockout mousegenerated by the Cre-LoxP system was alsocompletely protected against diabetes (20).

Glucose-induced Txnip expression occursat the transcriptional level, but surpris-ingly does not necessarily require glucosemetabolism (93). Promoter analysis of thehuman Txnip promoter revealed a unique car-bohydrate response element consisting of two

6.10 Keller · Attie

R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


nonpalindromic E-boxes (94). More recentwork, including chromatin immunoprecip-itation studies, demonstrated that glucoseenhances binding of ChREBP to the Txnippromoter in INS-1 cells as well as in humanislets (93). ChREBP, forming a heterodimerwith Mlx, in turn recruits the coactivator p300,which also acts as a histone acetyltransferaseand specifically leads to increased acetylationof histone H4 and recruitment of RNApolymerase II (93). In β-cells, ChREBP is nec-essary and sufficient for glucose-induced Txnipexpression (93), whereas in skeletal muscleand renal epithelial cells, the ChREBP paralogMondoA is responsible for this function (142).

The pathogenesis of type 2 diabetes followsa feed-forward pattern whereby hyperglycemiaexerts deleterious effects on β-cells and exac-erbates the syndrome. A microarray study ofhuman islets from nondiabetic versus diabetichuman subjects identified the genes most al-tered in response to diabetes (52). A key ob-servation was an 82% decline in the expres-sion of ARNT, a transcription factor that playsa role in embryonic development. In additionto ARNT, one of the MODY genes, HNF4α

was also sharply reduced. Genes participatingin the insulin-signaling pathway, the insulin re-ceptor, IRS2, AKT2, showed reduced expres-sion, while a phosphatase that blunts insulinsignaling, SHIP2, was upregulated, consistentwith a picture of insulin resistance in the β-cells.Knockout of the Arnt gene in β-cells in mice re-sulted in animals with glucose intolerance, withgreater severity in females than in males. Theglucose intolerance was due to reduced glucose-stimulated insulin secretion. This was associ-ated with reductions in enzymes responsible forglucose oxidation, suggesting that glucose sens-ing was impaired. Hnf4α expression was also re-duced in the knockout islets, as were the insulinreceptor and Akt2, drawing a clear parallel withthe observations in islets from human diabeticsubjects. Together, the effects of hyperglycemiaon TXNIP and ARNT expression offer a modelto explain increased β-cell apoptosis and de-creased insulin secretion as a consequence ofdiabetes (Figure 3, see color insert).

In rodent animal models (15) and perhapsalso in humans, pregnancy evokes an expan-sion of β-cell mass through an induction ofβ-cell proliferation, primarily through the ac-tion of placental lactogen and prolactin (136).Rieck et al. (120) carried out microarray anal-yses of islet gene expression in pregnant miceat the time when β-cell proliferation is at itspeak, at 14.5 days of gestation. They foundthat cyclins A2, B1, B2, D3, E1, F, and Cdk4were all upregulated. In addition, genes in-volved in serotonin and tryptophan biosynthe-sis were induced. Of special interest was thegene Birc5 (aka Survivin), an antiapoptotic gene.This gene is upregulated in the pregnant mice,but also in two other models of β-cell prolif-eration: obesity induced by leptin deficiency inB6 mice (71) and recovery from apoptosis in atransgenic model in which caspase 8 is induced(159).

Mouse strains differ widely in their suscep-tibility to obesity-induced diabetes (27). Onefactor that affects susceptibility to diabetes isthe proliferative response of β-cells to obesity.One model of obesity-induced diabetes is theBTBR strain into which the Leptinob mutationwas been introgressed to produce a congenicstrain, BTBR.ob. This strain develops severediabetes by six weeks of age (71, 141). Kelleret al. carried out a microarray study in isletsand five other tissues and explored three vari-ables; age (four weeks and ten weeks of age),leptin-deficiency induced obesity, and mousestrain (B6 versus BTBR). Using a method toidentify transcripts whose expression is highlycorrelated (47), 105 gene expression “modules”were identified in the six tissues, where therewas a distinct pattern of differential expressionas a function of one or more of the three vari-ables. One particular module was enriched incells encoding genes involved in cell replication(Figure 4, see color insert). In the B6 strain, theexpression of these genes increased with obe-sity but decreased with age. In contrast, in theBTBR strain, the response to obesity was lost;the genes were only responsive to age. To estab-lish that the behavior of this module was indeedpredictive of cell proliferation, the mice were


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


eQTL: expressionquantitative trait locus

given 2H20 two weeks before sacrifice, and theenrichment of 2H in DNA was determined asa measure of cell proliferation as described byNeese et al. (105). The direct measure of cellproliferation agreed with the prediction fromthe behavior of the cell cycle gene-expressionmodule (71).

A therapeutic objective in both type 1 andtype 2 diabetes is to try to boost β-cell masswith agents that prolong β-cell survival or β-cell proliferation. An active area of research andcontroversy is whether or not adult β-cells pro-liferate in vivo or if new β-cells arise from dif-ferentiation of progenitors or transdifferentia-tion of other cell types (32, 167). Microarrayanalysis has been helpful to define some of thegenes that participate in pathways leading toβ-cell proliferation. Transcription factors in-volved in the development of the endocrinepancreas have been tested for their ability topromote β-cell proliferation. One of these tran-scription factors, Nkx6.1, when overexpressedin isolated islets, promotes β-cell proliferation(128). Microarray analysis identified a patternof gene expression whereby the cell cycle genescyclins B1, A2, and E1 are induced by this gene.A follow-up study showed that overexpressionof cyclin E1 in islets is sufficient to stimulateβ-cell proliferation (128).

It is difficult to obtain reliable data at theextremes of gene expression from microarrays.This is because the probes can become satu-rated when mRNAs are abundant, and at thelow end, there is a great amount of variabil-ity in the data. An alternative albeit expensiveapproach is massive parallel signature sequenc-ing (MPSS). This approach involves sequenc-ing the transcripts and counting the number oftimes each transcript is sequenced. The methodis extremely sensitive and has a very broad dy-namic range. Using this approach, Kutlu et al.(80) identified the most enriched genes fromhuman islets and estimated that 13% of all themRNA in the samples was comprised of insulinmRNA. Based on an estimate of 50–55% ofthe islet sample containing insulin-producingcells, this suggests that approximately 30% ofβ-cell mRNA is comprised of insulin mRNA.

The massive parallel signature sequencing dataare available online (http://t1dbase.org).

CORRELATION VERSUSCAUSATION

Microarray analysis delivers large datasets fromwhich one can identify genes or sets of geneswhose expression varies or covaries across agiven set of conditions. For example, as de-scribed above, genes controlling lipogenesis,Srebf-1c, and its targets are upregulated in thelivers of obese animals. The expression of thesegenes is coordinately upregulated, and this canbe easily discerned in a microarray experiment.But the microarray experiment by itself doesnot allow one to infer the line of causalitybeginning with Srebf-1c and leading to genesencoding lipogenic enzymes. The microarraydata simply provide correlation information forwhich there are multiple potential interpreta-tions. Below, we describe the use of geneticinformation together with microarray data toovercome this limitation and enable the devel-opment of causal models. The method involvesmapping the loci that control gene expressionand then using these loci as anchor points todevelop causal models.

GENETICAL GENOMICS:EXPRESSION QUANTITATIVETRAIT LOCI

When employed in the context of an intercrossin model organisms or association mapping inan outbred population, genomic loci can bemapped that regulate the abundance of specificmRNA molecules. When the genotype at aparticular locus (usually an SNP) correlateswith the level of mRNA abundance, the locusis termed an expression quantitative trait locus(eQTL). The coverage provided by a genome-wide microarray, representing all knowngenes, means that thousands of eQTLs can beidentified in these studies, which have beentermed genetical genomics. The developmentof statistical tools required to analyze thesevoluminous data sets has been the subject of


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


intense research by several groups (16, 75, 88,175). Furthermore, integrating eQTL analyseswith the more traditional disease-related QTL(cQTL) studies (e.g., loci governing bodyweight) has led to the development of causalgene-gene regulatory networks as part of asystems biology approach to better understandcomplex disease states (33, 47, 126).

There are two classes of eQTLs: cis andtrans. Cis-eQTLs occur when the genomicposition of the gene associated for a particularmRNA is coincident with its eQTL (Figure 5,see color insert). Conceptually, a cis-eQTLimplies genetic variation at or near the tran-scription start site, whereby the variationresults in a change in the expression of thegene (153). In contrast to cis-mapping eQTLs,trans-eQTLs show linkage to genomic regionsthat are distant to their associated gene. Asimple example would be variation in a geneencoding a transcription factor, leading tovariation in the abundance of the mRNAs ofits target genes. The mRNA abundance traitsof the targets would map to the location of thegene encoding the transcription factor.

A key premise that permits directionallycausal networks to be constructed from QTL-driven expression data sets is that variation inDNA is the origin, or starting point, of thecausative relationship.

DNA → mRNA → clinical trait

This premise derives from the idea thatDNA variation can cause changes in gene ex-pression, but not vice-versa. This one-way re-lationship between DNA variation and changesin gene expression greatly simplifies the pos-sible combinations that relate genomic loci,eQTLs, and cQTLs. In simplified form, threeof the many possible models that interrelatethese network components are (124):

(a) Locus → eQTL → clinical trait (causal)(b) Locus → clinical trait → eQTL (reactive)(c) eQTL ← locus → clinical trait

(independent)

Note that in all models, the locus is theanchor point of the causal model in a unidi-

rectional fashion. Furthermore, if we assumethat the locus is directly affecting the eQTL,we can infer it to be cis-mapping in nature.Importantly, the cQTL illustrated in causalmodel (a) will have an overlapping QTL withthe eQTL directly upstream. Assuming the cis-eQTL depicted in this causal model is con-trolling the cQTL that is downstream (modela above), we would expect the expression andclinical traits associated with each QTL tobe correlated (positively or negatively). Noveldisease-causing genes have been identified ina variety of systems by using this integrativeapproach.

IDENTIFICATION ANDVALIDATION OF CAUSATIVEGENES FOR FAT MASS IN MICE

Schadt and colleagues (124) applied geneti-cal genomics to identify and validate novelgenes that regulate diet-induced changes in fatmass. An F2 intercross was generated betweenC57BL/6J (B6) and DBA/2J (DBA) mice, whichwere maintained on chow diet up to 12 monthsof age, followed by a high-fat diet for the an ad-ditional 4 months, after which mice were sac-rificed, and genome-wide expression profilingwas done in liver.

Whereas the expression of thousands ofgenes was found to significantly vary acrossthe panel of F2 mice, only ∼90 were identifiedas being key regulators of changes in fat massin response to the dietary challenge. Such areduction in number of candidate regulatorsresulted from applying several key filters:(a) determining which eQTLs favored acausal versus reactive or independent model;(b) requiring causal eQTLs to be correlatedwith changes in fat mass; and (c) selectingthose causal and correlated eQTLs to haveoverlapping QTLs for fat mass. Finally, thecandidate regulators were ranked according towhat proportion of fat mass variance could beexplained by the variation in transcript level.

The generation of genetic knockout ortransgenic overexpression animal models isconsidered the gold standard for testing the


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


predictive power of this integrative approachfor the identification of disease-causing genes.Three genes (Zfp90, C3ar1, and Tgfbr2) weretested with genetic manipulation and found tosignificantly affect fat mass accumulation in re-sponse to a high-fat diet (124). Furthermore, ina series of follow-up studies, additional geneswere validated as playing a key role in regulat-ing fat mass, including Lpl and LactB (23, 169)and Gas7, Gpx3, and Me1 (169). Importantly,the mouse models utilized for these validationstudies were whole-body deletions or ubiqui-tous promoters for the knockout and transgenicconstructs, respectively.

Although the expression profiling studiesused to identify the causal nature of these geneswere conducted in liver, it remains unclear iftheir role as regulators of fat mass actually oc-curs in the liver. It is possible that the QTL-based causal associations established betweencis-eQTLs and overlapping clinical phenotypeQTLs (e.g., fat mass) translate across tissuesand, thus, profiling liver can be informativefor these relationships in other tissues. Iden-tification of the tissue or cell type responsiblefor the effect on fat mass would require tissue-specific knockout or transgenic mouse models.It is also possible that while whole liver wasused for gene expression profiling, nonhepato-cyte liver cells (e.g., Kupffer cells) actually con-tributed to the eQTLs. Remarkably, five of thetop ten candidate genes (Tgfbr2, C3ar1, Zfp90,Lpl, and LactB) originally profiled in liver (124)were recently identified in a network highlyenriched for macrophage gene signatures andwere shown to closely associate with obesity andother metabolic syndrome phenotypes (23).

Several recent studies have clearly demon-strated the important role that adipose tissuemacrophage infiltration plays in obesity-induced insulin resistance (30, 38, 62, 127).Very recently, it was shown that C3ar1, a lead-ing candidate for regulating fat mass in mouse(124, 169), was found to be highly expressedin adipocytes as well as macrophages (90).Furthermore, not only were C3ar1−/− miceresistant to diet-induced obesity, but they alsowere protected from insulin resistance, hepatic

steatosis, adipose macrophage infiltration, andincreased circulating levels of proinflammatorycytokines, all of which are associated withobesity. Taken together, these results suggestthat the hepatic QTL study that originallyidentified C3ar1 as an obesity-causing gene(124) may have actually been assessing the roleof C3ar1 in macrophages that had infiltratedinto adipose and potentially other tissues (90),supporting the notion that QTL-driven genediscovery in one organ or tissue type canbe diagnostic about a gene’s role in diseasecausation in multiple tissues.

Genetic Architecture of HepaticGene Expression in Humans andIts Use for Identification ofDisease-Causing Genes

To what extent does QTL-driven discoveryfor disease-causing genes translate from inbredmouse strains to human biology? Schadt andcolleagues (125) addressed this question byperforming genome-wide expression profilingin liver samples from 427 human subjectsthat were genotyped at >780,000 SNPs. Ap-proximately 3,000 cis-eQTLs were identified,classified as having linkage to SNPs within1 Mb of the structural gene, including manygenes already identified as being associatedwith human diseases (BRCA1, CFH) anddrug responsiveness (VKORC1) or previouslyimplicated in genome-wide association studiesto the development of type I diabetes (e.g.,CTLA4, HLA-DRB1, IL2RA, LONRF2, andCHST10). By leveraging the human expressionand genotype data sets with eQTL-drivenexpression networks constructed from threemouse F2 intercrosses, some genes thought tobe associated with human disease were not sup-ported, while novel disease-causing genes wereidentified.

A large genome-wide association study of∼17,000 white European subjects and sevencommon diseases identified a locus on chromo-some 12q13, containing ERBB3 as a candidategene that is associated with type 1 diabetes(1, 148). An additional, more directed


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


association study in a Japanese population hasconfirmed the ERBB3 association, suggestingthe locus on 12q13 contributes to diabetessusceptibility across ethnic subgroups (7).However, there are key differences betweenstudies that associate disease to a particulargenomic region and those that relate changesin the expression of candidate genes to diseasesusceptibility.

Association studies do not offer a measure ofgene expression changes, nor do they identifythe tissue(s) involved in the disease relationship.For example, the expression of ERBB3 was notassociated with DNA variation at the 12q13 lo-cus, whereas the expression of a flanking gene,RPS26, was found to be highly correlated tothe SNP at this locus, giving rise to an eQTLin the human liver cohort (125). However, onceagain, we must ask whether an eQTL in liveris informative for how a gene may function inanother tissue. Importantly, only ∼30% of thecis-eQTLs identified in the human liver sam-ples were preserved across human blood andadipose tissue samples (37, 125), showing thatcis-eQTLs are sometimes, but not always, tissuespecific.

More than 1,000 mice were generated as partof three separate F2 intercrosses between theB6, C3H, and CAST mouse strains (125, 171).Gene expression was profiled in a number oftissues of each cross, including liver, adipose,muscle, and brain, and combined with genotypedata to reconstruct causal gene-gene expressionnetworks. Interestingly, Rps26, but not Erbb3,was identified in a subnetwork that was highlyenriched with other genes (H2-Eb1, H2-Ab1,H2-T22, H2-M3, H2-DMa, H2-Aa, H2-DMb2,and H2-T23) annotated as type 1 diabetes–related genes (125). It should be noted, how-ever, that a recent statistical analysis has chal-lenged the conclusion that genetic variation inthe expression of RPS26 is responsible for theassociation of type 1 diabetes at the 12q13 locus(115).

Applying the lessons learned from the type 1diabetes locus on 12q13 and the differences ob-served between Erbb3 and Rps26 as likely can-didate genes, additional genes (Psrc1, Sort1, and

Celsr2) were identified as candidates for a locuson chromosome 1p13.3 shown to associate withcoronary artery disease and LDL-cholesterollevels (125). Although uncertainties remain re-garding the reproducibility of expression net-works from one tissue to another as well as thedegree to which cis-eQTLs are preserved acrossmultiple tissues, the opportunity for gene dis-covery is enhanced by the approach of integrat-ing the eQTL data from the human liver cohortwith multiple human genome-wide associationstudies and with the genetical-genomic analy-sis of gene expression and clinical phenotypesin segregating mouse populations.

Variation in Genomic Copy NumberContributes to eQTLs

Copy number variants (CNVs) are genomicsegments that have been duplicated or deletedbetween individual mouse strains (50). Theycan range in size between a few kilobases toseveral megabases (with an average ∼0.3 Mb)and may contain known genes (50). CNVs havebeen shown to affect gene expression in humanlymphoblastoid cells collected from unrelatedindividuals from four geographically distinctpopulations (Utah, China, Japan, and Nigeria)(144). It is interesting to note that CFH, a geneassociated with age-related macular degenera-tion in humans (36, 55, 77), is contained withina CNV that is well conserved across species (50)and was identified as a cis-eQTL in a study ofhuman liver samples (125).

To determine the extent to which CNVscan influence eQTL analysis in a segregatingmouse cross, the effect on gene expression of19 CNVs, identified between the B6 and C3Hmouse strains (50), was determined in an F2 in-tercross between the same two strains (109). Alarge number of eQTLs occurred within or verynear CNV regions in adipose, brain, liver, andmuscle tissues. Remarkably, ∼84% of the genesknown to reside within the 19 CNVs were dif-ferentially expressed as a function of genotypeof the nearest marker to each CNV.

In addition to cis-eQTLs, trans-eQTLswere found to overlap several CNVs,


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


suggesting that regulatory elements con-tained within the CNV may have beenmediating their linkage. Both eQTLs andmetabolic QTLs for body weight, plasmacholesterol, triglyceride, glucose, and in-sulin overlapped CNVs on chromosomes1, 4, and 17. Furthermore, cis-eQTLswithin these CNVs tested causal for thecolocalized metabolic QTLs and await val-idation with conditional knockout and/ortransgenic animal models, as discussedabove. Taken together, the results indi-cate that CNVs represent another class ofgenomic polymorphism that can influencegene expression.

Concluding Remarks

The ability to interrogate the entire tran-scriptome in an unbiased fashion, throughmicroarray technology, has had a profoundimpact in virtually every field of biology.This review highlights examples where clearexpression patterns emerged that pointed to

a regulatory pathway or where changes inexpression pointed to an important role of anovel gene in diabetes pathogenesis. Tech-nology enabled these dramatic developments.Now, new technology in mass spectrometry isenabling similarly dramatic developments inproteomics and metabolomics. The ability toaccurately quantitate protein abundance andto assess post-translational modifications (e.g.,phosphorylation, acetylation, methylation,ubiquitination, glycosylation, prenylation, andfatty acylation) promises to reveal importantnew regulatory mechanisms and their relationto cell function and disease. Metabolomicsstudies will likely identify novel metabolites.Targeted metabolomics studies are beginningto show clear patterns of metabolites associatedwith metabolic diseases and cancer (8). Thesestudies might also lead to reliable biomarkersfor designing and monitoring disease treat-ments. Finally, the integration of these varioushigh-volume data sources can yield testablehypotheses connecting metabolites, protein,and/or transcript levels (40).

SUMMARY POINTS

1. Coexpressed genes are often coregulated.

2. Genes encoding mitochondrial respiratory function in muscle are dysregulated in dia-betes, likely a consequence of diabetes.

3. Mutations that disrupt mitochondrial function in muscle are associated with improvedinsulin sensitivity.

4. Obesity is associated with an inflammatory response in adipose tissue. Adipocytes,macrophages, and lymphocytes produce hormones that affect whole-body insulin sig-naling. Inadequate lipid storage in adipose tissue leads to spillover and a pathologicalaccumulation of lipids in other tissues.

5. Genes involved in hepatic lipogenesis are coordinately regulated through transcriptionfactors SREBP and ChREBP. These pathways are dysregulated in obesity and diabetes.

6. β-cell gene expression is responsive to hyperglycemia. Two genes that are dysregulated indiabetes are Txnip and Arnt, leading to β-cell apoptosis and impaired glucose-stimulatedinsulin secretion, respectively.

7. When combined with genetics, gene expression data can be used to construct causalnetwork models linking gene expression with disease phenotypes.


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


DISCLOSURE STATEMENT

The authors are not aware of any affiliations, memberships, funding, or financial holdings thatmight be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS

We are grateful to Dale Abel, Klaus Kaestner, Vamsi Mootha, Debbie Muoio, Mary ElizabethPatti, and Anath Shalev for their comments and insights. Work of the Attie laboratory is fundedby NIDDK grants DK66369 and DK58037, an American Diabetes Association Mentor-BasedFellowship, and a grant from the JDRF.

LITERATURE CITED

1. 2007. Genome-wide association study of 14000 cases of seven common diseases and 3000 shared controls.Nature 447:661–78

2. Abel ED, Peroni O, Kim JK, Kim YB, Boss O, et al. 2001. Adipose-selective targeting of the GLUT4gene impairs insulin action in muscle and liver. Nature 409:729–33

3. Aeberli I, Biebinger R, Lehmann R, L’Allemand D, Spinas GA, Zimmermann MB. 2007. Serum retinol-binding protein 4 concentration and its ratio to serum retinol are associated with obesity and metabolicsyndrome components in children. J. Clin. Endocrinol. Metab. 92:4359–65

4. Ahima RS, Lazar MA. 2008. Adipokines and the peripheral and neural control of energy balance.Mol. Endocrinol. 22:1023–31

5. Aigner E, Bachofner N, Klein K, De Geyter C, Hohla F, et al. 2009. Retinol-binding protein 4 inpolycystic ovary syndrome—association with steroid hormones and response to pioglitazone treatment.J. Clin. Endocrinol. Metab. 94:1229–35

6. Asilmaz E, Cohen P, Miyazaki M, Dobrzyn P, Ueki K, et al. 2004. Site and mechanism of leptin actionin a rodent form of congenital lipodystrophy. J. Clin. Invest. 113:414–24

7. Awata T, Kawasaki E, Tanaka S, Ikegami H, Maruyama T, et al. 2009. Association of type 1 diabeteswith two loci on 12q13 and 16p13 and the influence coexisting thyroid autoimmunity in Japanese.J. Clin. Endocrinol. Metab. 94:231–35

8. Bain JR, Stevens RD, Wenner BR, Ilkayeva O, Muoio DM, Newgard CB. 2009. Metabolomics appliedto diabetes research: moving from information to knowledge. Diabetes 58:2429–43

9. Balagopal P, Graham TE, Kahn BB, Altomare A, Funanage V, George D. 2007. Reduction of elevatedserum retinol binding protein in obese children by lifestyle intervention: association with subclinicalinflammation. J. Clin. Endocrinol. Metab. 92:1971–74

10. Barber TM, Hazell M, Christodoulides C, Golding SJ, Alvey C, et al. 2008. Serum levels of retinol-binding protein 4 and adiponectin in women with polycystic ovary syndrome: associations with visceralfat but no evidence for fat mass–independent effects on pathogenesis in this condition. J. Clin. Endocrinol.Metab. 93:2859–65

11. Becker W, Kluge R, Kantner T, Linnartz K, Korn M, et al. 2004. Differential hepatic gene expres-sion in a polygenic mouse model with insulin resistance and hyperglycemia: evidence for a combinedtranscriptional dysregulation of gluconeogenesis and fatty acid synthesis. J. Mol. Endocrinol. 32:195–208

12. Befroy DE, Petersen KF, Dufour S, Mason GF, de Graaf RA, et al. 2007. Impaired mitochondrialsubstrate oxidation in muscle of insulin-resistant offspring of type 2 diabetic patients. Diabetes 56:1376–81

13. Biddinger SB, Miyazaki M, Boucher J, Ntambi JM, Kahn CR. 2006. Leptin suppresses stearoyl-CoAdesaturase 1 by mechanisms independent of insulin and sterol regulatory element-binding protein-1c.Diabetes 55:2032–41

14. Bodnar JS, Chatterjee A, Castellani LW, Ross DA, Ohmen J, et al. 2002. Positional cloning of thecombined hyperlipidemia gene Hyplip1. Nat. Genet. 30:110–16


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


15. Brelje TC, Parsons JA, Sorenson RL. 1994. Regulation of islet beta-cell proliferation by prolactin in ratislets. Diabetes 43:263–73

16. Broman KW. 2005. Mapping expression in randomized rodent genomes. Nat. Genet. 37:209–1017. Brown MS, Goldstein JL. 2009. Cholesterol feedback: from Schoenheimer’s bottle to Scap’s MELADL.

J. Lipid Res. 50(Suppl.):S15–2718. Cha-Molstad H, Saxena G, Chen J, Shalev A. 2009. Glucose-stimulated expression of Txnip is mediated

by carbohydrate response element-binding protein, p300, and histone H4 acetylation in pancreatic betacells. J. Biol. Chem. 284:16898–905

19. Chen J, Couto FM, Minn AH, Shalev A. 2006. Exenatide inhibits beta-cell apoptosis by decreasingthioredoxin-interacting protein. Biochem. Biophys. Res. Commun. 346:1067–74

20. Chen J, Hui ST, Couto FM, Mungrue IN, Davis DB, et al. 2008. Thioredoxin-interacting proteindeficiency induces Akt/Bcl-xL signaling and pancreatic beta cell mass and protects against diabetes.FASEB J. 22:3581–94

21. Chen J, Saxena G, Mungrue IN, Lusis AJ, Shalev A. 2008. Thioredoxin-interacting protein: a criticallink between glucose toxicity and beta cell apoptosis. Diabetes 57:938–44

22. Chen KS, DeLuca HF. 1994. Isolation and characterization of a novel cDNA from HL-60 cells treatedwith 1,25-dihydroxyvitamin D-3. Biochim. Biophys. Acta 1219:26–32

23. Chen Y, Zhu J, Lum PY, Yang X, Pinto S, et al. 2008. Variations in DNA elucidate molecular networksthat cause disease. Nature 452:429–35

24. Cheng Z, Guo S, Copps K, Dong X, Kollipara R, et al. 2009. Foxo1 integrates insulin signaling withmitochondrial function in the liver. Nat. Med. 15:1307–11

25. Choi CS, Befroy DE, Codella R, Kim S, Reznick RM, et al. 2008. Paradoxical effects of increasedexpression of PGC-1alpha on muscle mitochondrial function and insulin-stimulated muscle glucosemetabolism. Proc. Natl. Acad. Sci. USA 105:19926–31

26. Choi SH, Kwak SH, Youn BS, Lim S, Park YJ, et al. 2008. High plasma retinol binding protein-4 andlow plasma adiponectin concentrations are associated with severity of glucose intolerance in women withprevious gestational diabetes mellitus. J. Clin. Endocrinol. Metab. 93:3142–48

27. Clee SM, Attie AD. 2006. The genetic landscape of type 2 diabetes in mice. Endocrine Rev. 28:48–8328. Cohen P, Miyazaki M, Socci ND, Hagge-Greenberg A, Liedtke W, et al. 2002. Role for stearoyl-CoA

desaturase-1 in leptin-mediated weight loss. Science 297:240–4329. Danforth E Jr. 2000. Failure of adipocyte differentiation causes type II diabetes mellitus? Nat. Genet.

26:1330. de Luca C, Olefsky JM. 2008. Inflammation and insulin resistance. FEBS Lett. 582:97–10531. Dobrzyn A, Ntambi JM. 2005. Stearoyl-CoA desaturase as a new drug target for obesity treatment.

Obes. Rev. 6:169–7432. Dor Y, Brown J, Martinez OI, Melton DA. 2004. Adult pancreatic beta-cells are formed by self-

duplication rather than stem-cell differentiation. Nature 429:41–4633. Drake TA, Schadt EE, Lusis AJ. 2006. Integrating genetic and gene expression data: application to

cardiovascular and metabolic traits in mice. Mamm. Genome 17:466–7934. Dubois SG, Heilbronn LK, Smith SR, Albu JB, Kelley DE, Ravussin E. 2006. Decreased expression of

adipogenic genes in obese subjects with type 2 diabetes. Obesity (Silver Spring) 14:1543–5235. Ducluzeau PH, Perretti N, Laville M, Andreelli F, Vega N, et al. 2001. Regulation by insulin of gene

expression in human skeletal muscle and adipose tissue. Evidence for specific defects in type 2 diabetes.Diabetes 50:1134–42

36. Edwards AO, Ritter R 3rd, Abel KJ, Manning A, Panhuysen C, Farrer LA. 2005. Complement factor Hpolymorphism and age-related macular degeneration. Science 308:421–24

37. Emilsson V, Thorleifsson G, Zhang B, Leonardson AS, Zink F, et al. 2008. Genetics of gene expressionand its effect on disease. Nature 452:423–28

38. Ferrante AW Jr. 2007. Obesity-induced inflammation: a metabolic dialogue in the language of inflam-mation. J. Intern. Med. 262:408–14

39. Ferrante AW Jr, Thearle M, Liao T, Leibel RL. 2001. Effects of leptin deficiency and short-termrepletion on hepatic gene expression in genetically obese mice. Diabetes 50:2268–78


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


40. Ferrara CT, Wang P, Neto EC, Stevens RD, Bain JR, et al. 2008. Genetic networks of liver metabolismrevealed by integration of metabolic and transcriptional profiling. PLoS Genet. 4:e1000034

41. Flowers JB, Rabaglia ME, Schueler KL, Flowers MT, Lan H, et al. 2007. Loss of stearoyl-CoA desaturase-1 improves insulin sensitivity in lean mice but worsens diabetes in leptin-deficient obese mice. Diabetes56:1228–39

42. Flowers MT, Groen AK, Oler AT, Keller MP, Choi Y, et al. 2006. Cholestasis and hypercholesterolemiain SCD1-deficient mice fed a low-fat, high-carbohydrate diet. J. Lipid Res. 47:2668–80

43. Flowers MT, Keller MP, Choi Y, Lan H, Kendziorski C, et al. 2008. Liver gene expression analysis revealsendoplasmic reticulum stress and metabolic dysfunction in SCD1-deficient mice fed a very low-fat diet.Physiol. Genomics 33:361–72

44. Garcia-Roves P, Huss JM, Han DH, Hancock CR, Iglesias-Gutierrez E, et al. 2007. Raising plasma fattyacid concentration induces increased biogenesis of mitochondria in skeletal muscle. Proc. Natl. Acad. Sci.USA 104:10709–13

45. Gavi S, Stuart LM, Kelly P, Melendez MM, Mynarcik DC, et al. 2007. Retinol-binding protein 4 isassociated with insulin resistance and body fat distribution in nonobese subjects without type 2 diabetes.J. Clin. Endocrinol. Metab. 92:1886–90

46. Gesta S, Bluher M, Yamamoto Y, Norris AW, Berndt J, et al. 2006. Evidence for a role of developmentalgenes in the origin of obesity and body fat distribution. Proc. Natl. Acad. Sci. USA 103:6676–81

47. Ghazalpour A, Doss S, Zhang B, Wang S, Plaisier C, et al. 2006. Integrating genetic and network analysisto characterize genes related to mouse weight. PLoS Genet. 2:e130

48. Graham TE, Yang Q, Bluher M, Hammarstedt A, Ciaraldi TP, et al. 2006. Retinol-binding protein 4and insulin resistance in lean, obese, and diabetic subjects. N. Engl. J. Med. 354:2552–63

49. Deleted in proof50. Graubert TA, Cahan P, Edwin D, Selzer RR, Richmond TA, et al. 2007. A high-resolution map of

segmental DNA copy number variation in the mouse genome. PLoS Genet. 3:e351. Guarente L. 2008. Mitochondria—a nexus for aging, calorie restriction, and sirtuins? Cell 132:171–7652. Gunton JE, Kulkarni RN, Yim S, Okada T, Hawthorne WJ, et al. 2005. Loss of ARNT/HIF1beta

mediates altered gene expression and pancreatic-islet dysfunction in human type 2 diabetes. Cell 122:337–49

53. Gygi SP, Rist B, Griffin TJ, Eng J, Aebersold R. 2002. Proteome analysis of low-abundance proteinsusing multidimensional chromatography and isotope-coded affinity tags. J. Proteome Res. 1:47–54

54. Haider DG, Schindler K, Prager G, Bohdjalian A, Luger A, et al. 2007. Serum retinol-binding protein4 is reduced after weight loss in morbidly obese subjects. J. Clin. Endocrinol. Metab. 92:1168–71

55. Haines JL, Hauser MA, Schmidt S, Scott WK, Olson LM, et al. 2005. Complement factor H variantincreases the risk of age-related macular degeneration. Science 308:419–21

56. Han SH, Jeon JH, Ju HR, Jung U, Kim KY, et al. 2003. VDUP1 upregulated by TGF-beta1 and 1,25-dihydroxyvitamin D3 inhibits tumor cell growth by blocking cell-cycle progression. Oncogene 22:4035–46

57. Hancock CR, Han DH, Chen M, Terada S, Yasuda T, et al. 2008. High-fat diets cause insulin resistancedespite an increase in muscle mitochondria. Proc. Natl. Acad. Sci. USA 105:7815–20

58. Handschin C, Choi CS, Chin S, Kim S, Kawamori D, et al. 2007. Abnormal glucose homeostasis inskeletal muscle-specific PGC-1alpha knockout mice reveals skeletal muscle-pancreatic beta cell crosstalk.J. Clin. Invest. 117:3463–74

59. He J, Watkins S, Kelley DE. 2001. Skeletal muscle lipid content and oxidative enzyme activity in relationto muscle fiber type in type 2 diabetes and obesity. Diabetes 50:817–23

60. Holloszy JO. 2009. Skeletal muscle “mitochondrial deficiency” does not mediate insulin resistance.Am. J. Clin. Nutr. 89:463–66S

61. Horton JD, Shah NA, Warrington JA, Anderson NN, Park SW, et al. 2003. Combined analysis ofoligonucleotide microarray data from transgenic and knockout mice identifies direct SREBP targetgenes. Proc. Natl. Acad. Sci. USA 100:12027–32

62. Hotamisligil GS. 2006. Inflammation and metabolic disorders. Nature 444:860–6763. Houstis N, Rosen ED, Lander ES. 2006. Reactive oxygen species have a causal role in multiple forms of

insulin resistance. Nature 440:944–48


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


64. Hui ST, Andres AM, Miller AK, Spann NJ, Potter DW, et al. 2008. Txnip balances metabolic andgrowth signaling via PTEN disulfide reduction. Proc. Natl. Acad. Sci. USA 105:3921–26

65. Hui TY, Sheth SS, Diffley JM, Potter DW, Lusis AJ, et al. 2004. Mice lacking thioredoxin-interactingprotein provide evidence linking cellular redox state to appropriate response to nutritional signals.J. Biol. Chem. 279:24387–93

66. Jensen MD. 1997. Health consequences of fat distribution. Horm. Res. 48(Suppl. 5):88–9267. Joza N, Oudit GY, Brown D, Benit P, Kassiri Z, et al. 2005. Muscle-specific loss of apoptosis-

inducing factor leads to mitochondrial dysfunction, skeletal muscle atrophy, and dilated cardiomyopathy.Mol. Cell Biol. 25:10261–72

68. Junn E, Han SH, Im JY, Yang Y, Cho EW, et al. 2000. Vitamin D3 up-regulated protein 1 mediatesoxidative stress via suppressing the thioredoxin function. J. Immunol. 164:6287–95

69. Kabashima T, Kawaguchi T, Wadzinski BE, Uyeda K. 2003. Xylulose 5-phosphate mediates glucose-induced lipogenesis by xylulose 5-phosphate-activated protein phosphatase in rat liver. Proc. Natl. Acad.Sci. USA 100:5107–12

70. Kakuma T, Lee Y, Higa M, Wang Z, Pan W, et al. 2000. Leptin, troglitazone, and the expression of sterolregulatory element binding proteins in liver and pancreatic islets. Proc. Natl. Acad. Sci. USA 97:8536–41

71. Keller MP, Choi Y, Wang P, Belt Davis D, Rabaglia ME, et al. 2008. A gene expression network modelof type 2 diabetes links cell cycle regulation in islets with diabetes susceptibility. Genome Res. 18:706–16

72. Keller MP, Choi Y, Wang P, Davis DB, Rabaglia ME, et al. 2008. A gene expression network model oftype 2 diabetes links cell cycle regulation in islets with diabetes susceptibility. Genome Res. 18:706–16

73. Kelley DE, He J, Menshikova EV, Ritov VB. 2002. Dysfunction of mitochondria in human skeletalmuscle in type 2 diabetes. Diabetes 51:2944–50

74. Kendziorski C, Irizarry RA, Chen KS, Haag JD, Gould MN. 2005. On the utility of pooling biologicalsamples in microarray experiments. Proc. Natl. Acad. Sci. USA 102:4252–57

75. Kendziorski C, Wang P. 2006. A review of statistical methods for expression quantitative trait locimapping. Mamm. Genome 17:509–17

76. Kendziorski CM, Zhang Y, Lan H, Attie AD. 2003. The efficiency of pooling mRNA in microarrayexperiments. Biostatistics 4:465–77

77. Klein RJ, Zeiss C, Chew EY, Tsai JY, Sackler RS, et al. 2005. Complement factor H polymorphism inage-related macular degeneration. Science 308:385–89

78. Kleiner S, Nguyen-Tran V, Bare O, Huang X, Spiegelman BM, Wu Z. 2009. PPARdelta agonismactivates fatty acid oxidation via PGC-1alpha but does not increase mitochondrial gene expression andfunction. J. Biol. Chem. 284:18624–33

79. Kusminski CM, Shetty S, Orci L, Unger RH, Scherer PE. 2009. Diabetes and apoptosis: lipotoxicity.Apoptosis 14:1484–95

80. Kutlu B, Burdick D, Baxter D, Rasschaert J, Flamez D, et al. 2009. Detailed transcriptome atlas of thepancreatic beta cell. BMC Med. Genomics 2:3

81. Lan H, Rabaglia ME, Stoehr JP, Nadler ST, Schueler KL, et al. 2003. Gene expression profiles ofnondiabetic and diabetic obese mice suggest a role of hepatic lipogenic capacity in diabetes susceptibility.Diabetes 52:688–700

82. Lumeng CN, Bodzin JL, Saltiel AR. 2007. Obesity induces a phenotypic switch in adipose tissuemacrophage polarization. J. Clin. Invest. 117:175–84

83. Lumeng CN, Deyoung SM, Bodzin JL, Saltiel AR. 2007. Increased inflammatory properties of adiposetissue macrophages recruited during diet-induced obesity. Diabetes 56:16–23

84. Lumeng CN, Deyoung SM, Saltiel AR. 2007. Macrophages block insulin action in adipocytes by alteringexpression of signaling and glucose transport proteins. Am. J. Physiol. Endocrinol. Metab. 292:E166–74

85. Luquet S, Lopez-Soriano J, Holst D, Fredenrich A, Melki J, et al. 2003. Peroxisome proliferator-activatedreceptor delta controls muscle development and oxidative capability. FASEB J. 17:2299–301

86. Ma L, Robinson LN, Towle HC. 2006. ChREBP∗Mlx is the principal mediator of glucose-induced geneexpression in the liver. J. Biol. Chem. 281:28721–30

87. Maassen JA, ’t Hart LM, Van Essen E, Heine RJ, Nijpels G, et al. 2004. Mitochondrial diabetes: molecularmechanisms and clinical presentation. Diabetes 53(Suppl. 1):S103–9


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


88. Mackay TF, Stone EA, Ayroles JF. 2009. The genetics of quantitative traits: challenges and prospects.Nat. Rev. Genet. 10:565–77

89. Maechler P, Wollheim CB. 2001. Mitochondrial function in normal and diabetic beta-cells. Nature414:807–12

90. Mamane Y, Chung Chan C, Lavallee G, Morin N, Xu LJ, et al. 2009. The C3a anaphylatoxin receptor isa key mediator of insulin resistance and functions by modulating adipose tissue macrophage infiltrationand activation. Diabetes 58:2006–17

91. Marie S, Diaz-Guerra MJ, Miquerol L, Kahn A, Iynedjian PB. 1993. The pyruvate kinase gene as a modelfor studies of glucose-dependent regulation of gene expression in the endocrine pancreatic beta-cell type.J. Biol. Chem. 268:23881–90

92. Marselli L, Sgroi DC, Bonner-Weir S, Weir GC. 2009. Laser capture microdissection of human pan-creatic beta-cells and RNA preparation for gene expression profiling. Methods Mol. Biol. 560:87–98

93. Minn AH, Couto FM, Shalev A. 2006. Metabolism-independent sugar effects on gene transcription: therole of 3-O-methylglucose. Biochemistry 45:11047–51

94. Minn AH, Hafele C, Shalev A. 2005. Thioredoxin-interacting protein is stimulated by glucose througha carbohydrate response element and induces beta-cell apoptosis. Endocrinology 146:2397–405

95. Minn AH, Pise-Masison CA, Radonovich M, Brady JN, Wang P, et al. 2005. Gene expression profiling inINS-1 cells overexpressing thioredoxin-interacting protein. Biochem. Biophys. Res. Commun. 336:770–78

96. Miyazaki M, Flowers MT, Sampath H, Chu K, Otzelberger C, et al. 2007. Hepatic stearoyl-CoAdesaturase-1 deficiency protects mice from carbohydrate-induced adiposity and hepatic steatosis. CellMetab. 6:484–96

97. Miyazaki M, Sampath H, Liu X, Flowers MT, Chu K, et al. 2009. Stearoyl-CoA desaturase-1 deficiencyattenuates obesity and insulin resistance in leptin-resistant obese mice. Biochem. Biophys. Res. Commun.380:818–22

98. Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, et al. 2003. PGC-1alpha-responsivegenes involved in oxidative phosphorylation are coordinately downregulated in human diabetes.Nat. Genet. 34:267–73

99. Morino K, Petersen KF, Dufour S, Befroy D, Frattini J, et al. 2005. Reduced mitochondrial densityand increased IRS-1 serine phosphorylation in muscle of insulin-resistant offspring of type 2 diabeticparents. J. Clin. Invest. 115:3587–93

100. Morino K, Petersen KF, Shulman GI. 2006. Molecular mechanisms of insulin resistance in humans andtheir potential links with mitochondrial dysfunction. Diabetes 55(Suppl. 2):S9–15

101. Munkhtulga L, Nakayama K, Utsumi N, Yanagisawa Y, Gotoh T, et al. 2007. Identification of a regula-tory SNP in the retinol binding protein 4 gene associated with type 2 diabetes in Mongolia. Hum. Genet.120:879–88

102. Muoio DM. 2007. TXNIP links redox circuitry to glucose control. Cell Metab. 5:412–14103. Nadler ST, Attie AD. 2001. Please pass the chips: genomic insights into obesity and diabetes. J. Nutr.

131:2078–81104. Nadler ST, Stoehr JP, Schueler KL, Tanimoto G, Yandell BS, Attie AD. 2000. The expression of

adipogenic genes is decreased in obesity and diabetes mellitus. Proc. Natl. Acad. Sci. USA 97:11371–76105. Neese RA, Misell LM, Turner S, Chu A, Kim J, et al. 2002. Measurement in vivo of proliferation rates

of slow turnover cells by 2H2O labeling of the deoxyribose moiety of DNA. Proc. Natl. Acad. Sci. USA99:15345–50

106. Nishimura M, Uyeda K. 1995. Purification and characterization of a novel xylulose 5-phosphate-activated protein phosphatase catalyzing dephosphorylation of fructose-6-phosphate,2-kinase:fructose-2,6-bisphosphatase. J. Biol. Chem. 270:26341–46

107. Nishiyama A, Masutani H, Nakamura H, Nishinaka Y, Yodoi J. 2001. Redox regulation by thioredoxinand thioredoxin-binding proteins. IUBMB Life 52:29–33

108. Ntambi JM, Miyazaki M, Stoehr JP, Lan H, Kendziorski CM, et al. 2002. Loss of stearoyl-CoAdesaturase-1 function protects mice against adiposity. Proc. Natl. Acad. Sci. USA 99:11482–86

109. Orozco LD, Cokus SJ, Ghazalpour A, Ingram-Drake L, Wang S, et al. 2009. Copy number variationinfluences gene expression and metabolic traits in mice. Hum. Mol. Genet. 18:4118–29


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


110. Parikh H, Carlsson E, Chutkow WA, Johansson LE, Storgaard H, et al. 2007. TXNIP regulates periph-eral glucose metabolism in humans. PLoS Med. 4:e158

111. Patti ME, Butte AJ, Crunkhorn S, Cusi K, Berria R, et al. 2003. Coordinated reduction of genes ofoxidative metabolism in humans with insulin resistance and diabetes: potential role of PGC1 and NRF1.Proc. Natl. Acad. Sci. USA 100:8466–71

112. Petersen KF, Befroy D, Dufour S, Dziura J, Ariyan C, et al. 2003. Mitochondrial dysfunction in theelderly: possible role in insulin resistance. Science 300:1140–42

113. Petersen KF, Dufour S, Befroy D, Garcia R, Shulman GI. 2004. Impaired mitochondrial activity in theinsulin-resistant offspring of patients with type 2 diabetes. N. Engl. J. Med. 350:664–71

114. Pihlajamaki J, Boes T, Kim EY, Dearie F, Kim BW, et al. 2009. Thyroid hormone-related regulation ofgene expression in human fatty liver. J. Clin. Endocrinol. Metab. 94:3521–29

115. Plagnol V, Smyth DJ, Todd JA, Clayton DG. 2009. Statistical independence of the colocalized associationsignals for type 1 diabetes and RPS26 gene expression on chromosome 12q13. Biostatistics 10:327–34

116. Poitout V, Robertson RP. 2008. Glucolipotoxicity: fuel excess and beta-cell dysfunction. Endocr. Rev.29:351–66

117. Pospisilik JA, Knauf C, Joza N, Benit P, Orthofer M, et al. 2007. Targeted deletion of AIF decreasesmitochondrial oxidative phosphorylation and protects from obesity and diabetes. Cell 131:476–91

118. Prentki M, Nolan CJ. 2006. Islet beta cell failure in type 2 diabetes. J. Clin. Invest. 116:1802–12119. Reinehr T, Stoffel-Wagner B, Roth CL. 2008. Retinol-binding protein 4 and its relation to insulin

resistance in obese children before and after weight loss. J. Clin. Endocrinol. Metab. 93:2287–93120. Rieck S, White P, Schug J, Fox AJ, Smirnova O, et al. 2009. The transcriptional response of the islet to

pregnancy in mice. Mol. Endocrinol. 23:1702–12121. Ritov VB, Menshikova EV, Azuma K, Wood RJ, Toledo FG, et al. 2009. Deficiency of electron transport

chain in human skeletal muscle mitochondria in type 2 diabetes mellitus and obesity. Am. J. Physiol.Endocrinol. Metab. Epub ahead of print

122. Robertson RP. 2004. Chronic oxidative stress as a central mechanism for glucose toxicity in pancreaticislet beta cells in diabetes. J. Biol. Chem. 279:42351–54

123. Sampath H, Flowers MT, Liu X, Paton CM, Sullivan R, et al. 2009. Skin-specific deletion of stearoyl-coadesaturase-1 alters skin lipid composition and protects mice from high-fat diet-induced obesity. J. Biol.Chem. 284:19961–73

124. Schadt EE, Lamb J, Yang X, Zhu J, Edwards S, et al. 2005. An integrative genomics approach to infercausal associations between gene expression and disease. Nat. Genet. 37:710–17

125. Schadt EE, Molony C, Chudin E, Hao K, Yang X, et al. 2008. Mapping the genetic architecture of geneexpression in human liver. PLoS Biol. 6:e107

126. Schadt EE, Zhang B, Zhu J. 2009. Advances in systems biology are enhancing our understanding ofdisease and moving us closer to novel disease treatments. Genetica 136:259–69

127. Schenk S, Saberi M, Olefsky JM. 2008. Insulin sensitivity: modulation by nutrients and inflammation.J. Clin. Invest. 118:2992–3002

128. Schisler JC, Fueger PT, Babu DA, Hohmeier HE, Tessem JS, et al. 2008. Stimulation of human and ratislet beta-cell proliferation with retention of function by the homeodomain transcription factor Nkx6.1.Mol. Cell Biol. 28:3465–76

129. Sekine N, Cirulli V, Regazzi R, Brown LJ, Gine E, et al. 1994. Low lactate dehydrogenase and highmitochondrial glycerol phosphate dehydrogenase in pancreatic beta-cells. Potential role in nutrientsensing. J. Biol. Chem. 269:4895–902

130. Sener A, Mercan D, Malaisse WJ. 2001. Enzymic activities in two populations of purified rat islet beta-cells. Int. J. Mol. Med. 8:285–89

131. Sewter C, Berger D, Considine RV, Medina G, Rochford J, et al. 2002. Human obesity and type 2diabetes are associated with alterations in SREBP1 isoform expression that are reproduced ex vivo bytumor necrosis factor-alpha. Diabetes 51:1035–41

132. Shalev A, Pise-Masison CA, Radonovich M, Hoffmann SC, Hirshberg B, et al. 2002. Oligonucleotidemicroarray analysis of intact human pancreatic islets: identification of glucose-responsive genes and ahighly regulated TGFbeta signaling pathway. Endocrinology 143:3695–98


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


133. Shepherd PR, Gnudi L, Tozzo E, Yang H, Leach F, Kahn BB. 1993. Adipose cell hyperplasia andenhanced glucose disposal in transgenic mice overexpressing GLUT4 selectively in adipose tissue.J. Biol. Chem. 268:22243–46

134. Sheth SS, Castellani LW, Chari S, Wagg C, Thipphavong CK, et al. 2005. Thioredoxin-interactingprotein deficiency disrupts the fasting-feeding metabolic transition. J. Lipid Res. 46:123–34

135. Shimomura I, Matsuda M, Hammer RE, Bashmakov Y, Brown MS, Goldstein JL. 2000. DecreasedIRS-2 and increased SREBP-1c lead to mixed insulin resistance and sensitivity in livers of lipodystrophicand ob/ob mice. Mol. Cell 6:77–86

136. Sorenson RL, Brelje TC. 1997. Adaptation of islets of Langerhans to pregnancy: beta-cell growth,enhanced insulin secretion and the role of lactogenic hormones. Horm. Metab. Res. 29:301–7

137. Soukas A, Cohen P, Socci ND, Friedman JM. 2000. Leptin-specific patterns of gene expression in whiteadipose tissue. Genes Dev. 14:963–80

138. Sparks LM, Xie H, Koza RA, Mynatt R, Hulver MW, et al. 2005. A high-fat diet coordinately downregu-lates genes required for mitochondrial oxidative phosphorylation in skeletal muscle. Diabetes 54:1926–33

139. Sreekumar R, Halvatsiotis P, Schimke JC, Nair KS. 2002. Gene expression profile in skeletal muscle oftype 2 diabetes and the effect of insulin treatment. Diabetes 51:1913–20

140. Stoeckman AK, Ma L, Towle HC. 2004. Mlx is the functional heteromeric partner of the carbohy-drate response element-binding protein in glucose regulation of lipogenic enzyme genes. J. Biol. Chem.279:15662–69

141. Stoehr JP, Nadler ST, Schueler KL, Rabaglia ME, Yandell BS, et al. 2000. Genetic obesity unmasksnonlinear interactions between murine type 2 diabetes susceptibility loci. Diabetes 49:1946–54

142. Stoltzman CA, Peterson CW, Breen KT, Muoio DM, Billin AN, Ayer DE. 2008. Glucose sensing byMondoA:Mlx complexes: a role for hexokinases and direct regulation of thioredoxin-interacting proteinexpression. Proc. Natl. Acad. Sci. USA 105:6912–17

143. Storey JD, Tibshirani R. 2003. Statistical significance for genomewide studies. Proc. Natl. Acad. Sci. USA100:9440–45

144. Stranger BE, Forrest MS, Dunning M, Ingle CE, Beazley C, et al. 2007. Relative impact of nucleotideand copy number variation on gene expression phenotypes. Science 315:848–53

145. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, et al. 2005. Gene set enrichmentanalysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad.Sci. USA 102:15545–50

146. Tanaka Y, Tran PO, Harmon J, Robertson RP. 2002. A role for glutathione peroxidase in protectingpancreatic beta cells against oxidative stress in a model of glucose toxicity. Proc. Natl. Acad. Sci. USA99:12363–68

147. Tobe K, Suzuki R, Aoyama M, Yamauchi T, Kamon J, et al. 2001. Increased expression of the sterolregulatory element-binding protein-1 gene in insulin receptor substrate-2(-/-) mouse liver. J. Biol. Chem.276:38337–40

148. Todd JA, Walker NM, Cooper JD, Smyth DJ, Downes K, et al. 2007. Robust associations of four newchromosome regions from genome-wide analyses of type 1 diabetes. Nat. Genet. 39:857–64

149. Towle HC. 2005. Glucose as a regulator of eukaryotic gene transcription. Trends Endocrinol. Metab.16:489–94

150. Turner N, Bruce CR, Beale SM, Hoehn KL, So T, et al. 2007. Excess lipid availability increases mito-chondrial fatty acid oxidative capacity in muscle: evidence against a role for reduced fatty acid oxidationin lipid-induced insulin resistance in rodents. Diabetes 56:2085–92

151. Uyeda K, Yamashita H, Kawaguchi T. 2002. Carbohydrate responsive element-binding protein(ChREBP): a key regulator of glucose metabolism and fat storage. Biochem. Pharmacol. 63:2075–80

152. van Hoek M, Dehghan A, Zillikens MC, Hofman A, Witteman JC, Sijbrands EJ. 2008. An RBP4promoter polymorphism increases risk of type 2 diabetes. Diabetologia 51:1423–28

153. Veyrieras JB, Kudaravalli S, Kim SY, Dermitzakis ET, Gilad Y, et al. 2008. High-resolution mappingof expression-QTLs yields insight into human gene regulation. PLoS Genet. 4:e1000214

154. Walker JR, Hogenesch JB. 2005. RNA profiling in circadian biology. Methods Enzymol. 393:366–76155. Wang MY, Grayburn P, Chen S, Ravazzola M, Orci L, Unger RH. 2008. Adipogenic capacity and the

susceptibility to type 2 diabetes and metabolic syndrome. Proc. Natl. Acad. Sci. USA 105:6139–44


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


156. Wang MY, Orci L, Ravazzola M, Unger RH. 2005. Fat storage in adipocytes requires inactivationof leptin’s paracrine activity: implications for treatment of human obesity. Proc. Natl. Acad. Sci. USA102:18011–16

157. Wang S, Yehya N, Schadt EE, Wang H, Drake TA, Lusis AJ. 2006. Genetic and genomic analysis of afat mass trait with complex inheritance reveals marked sex specificity. PLoS Genet. 2:e15

158. Wang YX, Zhang CL, Yu RT, Cho HK, Nelson MC, et al. 2004. Regulation of muscle fiber type andrunning endurance by PPARdelta. PLoS Biol. 2:e294

159. Wang ZV, Mu J, Schraw TD, Gautron L, Elmquist JK, et al. 2008. PANIC-ATTAC: a mouse modelfor inducible and reversible beta-cell ablation. Diabetes 57:2137–48

160. Warden CH, Fisler JS. 2008. Comparisons of diets used in animal models of high-fat feeding. Cell Metab.7:277

161. Wasserman DH. 2009. Four grams of glucose. Am. J. Physiol. Endocrinol. Metab. 296:E11–21162. Weisberg SPD, McCann M, Desai M, Rosenbaum R, Leibel RL, Ferrante AW Jr. 2003. Obesity is

associated with macrophage accumulation in adipose tissue. J. Clin. Invest. 116:1796–808163. White P, Kaestner KH. 2009. Gene expression analysis in diabetes research. Methods Mol. Biol. 560:239–

61164. Wredenberg A, Freyer C, Sandstrom ME, Katz A, Wibom R, et al. 2006. Respiratory chain dysfunction

in skeletal muscle does not cause insulin resistance. Biochem. Biophys. Res. Commun. 350:202–7165. Wright JJ, Kim J, Buchanan J, Boudina S, Sena S, et al. 2009. Mechanisms for increased myocardial fatty

acid utilization following short-term high-fat feeding. Cardiovasc. Res. 82:351–60166. Xu H, Barnes GT, Yang Q, Tan G, Yang D, et al. 2003. Chronic inflammation in fat plays a crucial role

in the development of obesity-related insulin resistance. J. Clin. Invest. 112:1821–30167. Xu X, D’Hoker J, Stange G, Bonne S, De Leu N, et al. 2008. Beta cells can be generated from endogenous

progenitors in injured adult mouse pancreas. Cell 132:197–207168. Yang Q, Graham TE, Mody N, Preitner F, Peroni OD, et al. 2005. Serum retinol binding protein 4

contributes to insulin resistance in obesity and type 2 diabetes. Nature 436:356–62169. Yang X, Deignan JL, Qi H, Zhu J, Qian S, et al. 2009. Validation of candidate causal genes for obesity

that affect shared metabolic pathways and networks. Nat. Genet. 41:415–23170. Yang X, Pratley RE, Tokraks S, Bogardus C, Permana PA. 2002. Microarray profiling of skeletal muscle

tissues from equally obese, nondiabetic insulin-sensitive and insulin-resistant Pima Indians. Diabetologia45:1584–93

171. Yang X, Schadt EE, Wang S, Wang H, Arnold AP, et al. 2006. Tissue-specific expression and regulationof sexually dimorphic genes in mice. Genome Res. 16:995–1004

172. Yao-Borengasser A, Varma V, Bodles AM, Rasouli N, Phanavanh B, et al. 2007. Retinol binding protein4 expression in humans: relationship to insulin resistance, inflammation, and response to pioglitazone.J. Clin. Endocrinol. Metab. 92:2590–97

173. Yechoor VK, Patti ME, Ueki K, Laustsen PG, Saccone R, et al. 2004. Distinct pathways of insulin-regulated versus diabetes-regulated gene expression: an in vivo analysis in MIRKO mice. Proc. Natl.Acad. Sci. USA 101:16525–30

174. Zheng Y, Eilertsen KJ, Ge L, Zhang L, Sundberg JP, et al. 1999. Scd1 is expressed in sebaceous glandsand is disrupted in the asebia mouse. Nat. Genet. 23:268–70

175. Zhu J, Zhang B, Smith EN, Drees B, Brem RB, et al. 2008. Integrating large-scale functional genomicdata to dissect the complexity of yeast regulatory networks. Nat. Genet. 40:854–61


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


Insulin sensitivity

# of mitochondriaor mitochondrial function

Tfam

Aif

Txnip

Pgc 1α

Insulin sensitivity

# of mitochondria

High fat diet

Diabetes

Insulin sensitivity Glucose

Txnip

a

b

c

Figure 1Experimental observations where mitochondrial function or number of mitochondria is negativelycorrelated with insulin sensitivity. (a) In four different knockout mouse models that affect mitochondrialfunction or number of mitochondria in skeletal muscle, there is a marked improvement in insulin sensitivity.(b) High-fat diets increase the number of skeletal muscle mitochondria. The effect on PGC1α iscontroversial. But, these diets are associated with reduced insulin sensitivity. (c) Hyperglycemia itself isassociated with reduced muscle insulin sensitivity. A highly glucose-responsive gene, Txnip, when highlyexpressed, reduces insulin sensitivity and conversely, as illustrated in panel a, when knocked out in skeletalmuscle, is associated with improved insulin sensitivity.

www.annualreviews.org • Gene Expression in Obesity and Diabetes C-1

R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


Obesity

freefatty acids

insulinrersistance

lipotoxicity in β-cells

lipotoxicity &insulin resistance

in muscle

adiposelipogenesis

adiposelipolysis

adiposeinflammation

hepaticlipogenesis

hepaticsteatosis,

leptin

insulin

leptin resistance

Figure 2Obesity causes metabolic and endocrine changes in adipose tissue. Leptin acts in an autocrine fashion on adipocytes to promotelipogenesis and reduce uncoupled oxidative phosphorylation. Obesity promotes leptin resistance, thus reducing lipogenesis in adiposetissue and promoting it in other tissues, leading to accumulation of triglycerides in the liver (hepatic steatosis). In muscle, increasedfatty acid flux leads to insulin resistance. In β-cells, fatty acids can promote lipotoxicity and impair β-cell function and survival.Adipocytes in obese individuals produced adipokines that promote insulin resistance (e.g., RBP4 and resistin). Obesity promotes aninflammatory response in adipose tissue wherein macrophages and lymphocytes are recruited to adipose tissue and promoteinflammation and insulin resistance. The insulin resistance in adipocytes increases the rate of triglyceride lipolysis, increasing the flux offree fatty acids to other tissues.

C-2 Keller · Attie

R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


Hyperglycemia

TxnipGlucosesensing

β-cellapotosis

insulinsecretion

Arnt

Figure 3Diabetes has detrimental effects on β-cells. In the early stages of type 2 diabetes, β-cells produce insulin atan elevated level in an attempt to compensate for insulin resistance. However, hyperglycemia compromisestwo key mechanisms by which β-cells mount a compensatory response to insulin resistance. First, glucoseinduces the expression of Txnip, a proapoptotic gene. Second, hyperglycemia downregulates the expressionof the transcription factor Arnt. This leads to reduced glucose oxidation and reduced glucose sensing,resulting in reduced insulin secretion.


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


4 wk 4 wk10 wk 10 wkB6 BTBR

lean leanlean leanob obob

lean leanob ob

ob

lean leanob ob

log

gene

exp

ress

ion

ratio

n

heat

map

a

b0.3

0.2

0.1

0

-0.1

-0.2

-0.3

30

20

10

0

p<0.01 p<0.001 NS NS

% n

ew c

ells

c

lean leanob ob

B6 BTBE

NSp<0.001

Figure 4Coexpression modules enriched with cell cycle regulation accurately predict diabetes and obesity.(a) Expression heat maps and (b) the PC1 on log10 scale of the cell cycle regulatory modules in islets (217transcripts) and adipose (96 transcripts) are shown. For the heat maps, red shows increased expression, greenshows decreased expression, and black is neutral. Bar plots in (b) show the PC1 for individual mice andcorrespond to an decreased expression for negative values and increased expression for positive values.(c) The percentage of new cells, derived from an in vivo measure of 2H incorporation into newly synthesizedDNA, is shown for islets and adipose tissue. Where significant obesity-dependent differences were observed,P-values are shown. Arrows are used to show influence of obesity. NS, not significant. Figure and legendreprinted from (71).

C-4 Keller · Attie

R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.


acis-eQTL

trans-eQTL

master regulator in trans

b

c

Figure 5Cis- and trans-acting expression quantitative trait loci (eQTLs). (a) When mRNA abundance maps as aphenotype near the physical location of the gene encoding the mRNA, it is termed a cis-eQTL or a proximaleQTL. (b) When the mRNA abundance maps to a genomic location that is distinctly separate from thephysical location of the gene encoding the mRNA (e.g., a different chromosome, as shown), it is described asa trans-eQTL or distal eQTL. (c) When multiple mRNA abundance phenotypes map to the same genomiclocation in trans, then one can hypothesize that there might be one or more regulators encoded at thatlocation that coordinately regulate this set of mRNAs. This result is one step in building network models ofgene regulation.


R

E V I EW

S

IN

AD V A N

CE


Ann

u. R

ev. N

utr.

201

0.30

. Dow

nloa

ded

from

arj

ourn

als.

annu

alre

view

s.or

gby

Uni

vers

ity o

f W

isco

nsin

- M

adis

on o

n 07

/07/

10. F

or p

erso

nal u

se o

nly.

physiological insights gained from gene expression ... of biochemistry, ... important discoveries...

Documents