letter

nature genetics • volume 24 • april 2000 377

In vivo modulation of Hmgic reduces obesity

Ashim Anand & Kiran Chada

Department of Biochemistry, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, Piscataway, New Jersey, USA.Correspondence should be addressed to K.C. (e-mail: Chada@umdnj.edu).

The HMGI family of proteins consists of three members1,2,HMGIC, HMGI and HMGI(Y), that function as architecturalfactors3–5 and are essential components of the enhancesome6,7.HMGIC is predominantly expressed in proliferating, undifferenti-ated mesenchymal cells and is not detected in adult tissues8,9. Itis disrupted and misexpressed in a number of mesenchymaltumour cell types10–12, including fat-cell tumours12 (lipomas). Inaddition Hmgic–/– mice have a deficiency in fat tissue13. To studyits role in adipogenesis and obesity, we examined Hmgic expres-sion in the adipose tissue of adult, obese mice. Mice with a par-tial or complete deficiency of Hmgic resisted diet-inducedobesity. Disruption of Hmgic caused a reduction in the obesityinduced by leptin deficiency (Lepob/Lepob) in a gene-dose–depen-dent manner. Our studies implicate a role for HMGIC in fat-cellproliferation, indicating that it may be an adipose-specific targetfor the treatment of obesity.We initially examined gene expression in adipose tissue isolatedfrom wild-type mice fed either a standard or a high-fat diet, as thelatter induces obesity14. RT–PCR analysis of RNA isolated fromthe individual fat depots (Fig. 1a) and variousadult tissues (data not shown) from mice fed astandard diet confirmed observations15 in humanthat Hmgic is not expressed. In contrast, Hmgicexpression was detected in RNA isolated from indi-vidual fat depots of wild-type mice after one weekon a high-fat diet (Fig. 1a). Similarly, Hmgicexpression was observed in the individual fatdepots of two genetically obese mouse models16,17,Lepob/Lepob and Leprdb/Leprdb (Fig. 1b), but not inother tissues (data not shown). This provided thefirst indication for a role of Hmgic in obesity.

We had generated a specific Hmgic-nullmutant8, which allowed us to examine the in vivoeffect of Hmgic in obesity. We induced obesity byfeeding mice a high-fat diet. After 26 weeks, wild-type mice had developed obesity compared withmice on the standard diet (Fig. 2a). There was nota statistically significant difference in the finalweight of Hmgic–/– or Hmgic+/– mice fed either a

standard or a high-fat diet (Fig. 2b,c). The Hmgic mutation isrecessive, and at 30 weeks of age there was not a significant dif-ference between the weights of Hmgic+/– and wild-type mice fedthe standard diet (27.8±1.38 g versus 29.04±1.59 g, P=0.48).Thus, a haploinsufficiency effect was observed between wild-type and Hmgic+/– mice (35.24±1.86 g versus 29.87±2 g,P=0.03), leading to the conclusion that under the stimulus of ahigh-fat diet, there are phenotypic ramifications of the Hmgicmutation in the heterozygous state. The lack of weight gain can-not be attributed to a decrease in food intake, because daily foodconsumption was not statistically different between wild type(4.15±0.36 g) and either Hmgic+/– (4.28±0.34 g, P=0.82) orHmgic–/– (4.02±0.40 g, P=0.70) genotypes. Thus, absence of oneor both Hmgic alleles enables mice to resist the obesity thatresults from a high-fat diet.

A second method of obesity induction used the genetic mousemodel17 Lepob/Lepob, which results from the deficiency of the neu-roendocrine hormone leptin18. Double-homozygous mice(Hmgic–/–, Lepob/Lepob) were reduced in size compared with

Fig. 1 Hmgic expression in the white adipose tissue (WAT) ofobese mice. a, A high-fat diet induces Hmgic expression inadult mice. RT–PCR was performed on total RNA isolated frominguinal (i), parametrial (p), mesenteric (m) and epididymal(e) fat pads of adult wild-type mice (f, female; m, male) fedeither a standard (sd) or a high-fat (hfd) diet. Four-week-oldmice were weaned (0) and fed the high-fat or the standarddiet for one week (1). RNA samples from 12.5 days post-coitum embryos and Hmgic–/– WAT were used as positive andnegative controls, respectively, for Hmgic expression (top).We used Gapd as a positive control to determine the qualityof total RNA (bottom). b, Hmgic expression in Lepob/Lepob

and Leprdb/Leprdb WAT. A similar RT–PCR analysis was per-formed on total RNA isolated from inguinal, parametrial,mesenteric and epididymal fat pads of adult (8-week-old)Lepob/Lepob, Leprdb/Leprdb and wild-type mice.

a

b

© 2000 Nature America Inc. • http://genetics.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

378 nature genetics • volume 24 • april 2000

Lepob/Lepob mice (Fig. 3). By 30 weeks of age, double-homozygousmice weighed only 27% that of Lepob/Lepob mice (P<0.0001; Figs 3,4a). This is not due to the smaller size of the Hmgic–/– mouse,which weighs 47% that of a wild-type mouse (Fig. 4a). When theLep null mutation was placed on an Hmgic+/– background(Hmgic+/–, Lepob/Lepob), there was less expression of Hmgic (ref. 8)in the adipose tissue (data not shown) and the mice weighed only72% that of Lepob/Lepob mice (Figs 3, 4b). The results obtained withHmgic+/–, Lepob/Lepob mice diminish the possibility that theobserved weight reduction of Hmgic–/–, Lepob/Lepob mouse arisesfrom an absence of Hmgic during embryogenesis, rather than adirect effect of its loss in adult mice.

The overall weight reduction in mice of the relevant genotypescompared with Lepob/Lepob mice was reflected in a decrease in theweights of the discrete fat pads19 (Fig. 5a). The greatest and leastweight reduction occurred in the mesenteric (10.9%, P=0.002)and inguinal (31.4%, P=0.004) Hmgic–/–, Lepob/Lepob fat pads,respectively. We observed a similar weight-reduction pattern forHmgic+/–, Lepob/Lepob fat pads (Fig. 5a) with no significant differ-ence between the inguinal fat pads. These results demonstrate adifferential response among the various fat pads by inactivationof Hmgic on a Lepob/Lepob background.

The increase in weight of adipose tissue in Lepob/Lepob mice isdue to hypertrophy and hyperproliferation of the fat cells20. Theadipocytes of Hmgic–/–, Lepob/Lepob mice appear similar to thosein Lepob/Lepob mice (Fig. 5b,c) and we observed no differences inthe expression levels and regulation of genes (Pparg, Adn andAp2) involved in adipogenesis (data not shown). Instead, thereduction in fat pad weight of Hmgic–/–, Lepob/Lepob mice com-pared with Lepob/Lepob mice is due to a decrease in the numberof cells in the respective adipose depots (Fig. 5d). Similar resultswere obtained in a preliminary study with Hmgic–/– andHmgic+/– mice fed a high-fat diet.

Leptin deficiency is a more extreme stimulus forthe induction of weight gain (200%) than diet-induced obesity (21%). Although there is a reduc-tion in the weight of Lepob/Lepob mice on anHmgic–/– background (from 80.05±2.8 g to22.31±1.1 g), the mice still weigh 59% more thanthe parental Hmgic–/– mice. This is in contrast to anegligible weight gain by Hmgic–/– mice on a high-fat diet and suggests the existence of a minor pro-liferative pathway that is independent of Hmgic.

Although Hmgic is epistatic to Lep in the fat tis-sue, it should be noted that this is not a classicalepistatic21 relationship because other phenotypicfeatures of Lepob/Lepob mice are still observed inHmgic–/–, Lepob/Lepob mice. For example, glyco-

suria, plasma glucose (186.9±9.23, 208.8±23.1 mg dl–1, P=0.50)and insulin (3.99±0.57, 3.48±0.57 ng ml–1, P=0.37) levels werenot significantly different between Lepob/Lepob and Hmgic–/–,Lepob/Lepob genotypes. This suggests that Hmgic and Lep func-tion in independent genetic pathways.

We present a hypothesis to explain the mechanism and functionof HMGIC in adipogenesis and obesity (Fig. 6). We propose thatthe proliferative expansion of undifferentiated pre-adipocytesrequires HMGIC expression, which may be mediated by the regula-tion of its binding ability during the cell cycle22,23. Previous studieshave linked HMGIC with proliferation8,12, and stromal-vascularcells, similar to Hmgic-null embryonic fibroblasts8, proliferate moreslowly than their wild-type counterparts (data not shown). In wild-type fat tissue, HMGIC expression is undetectable because the pre-adipocyte population forms a minor component of the adiposetissue. Under obesity-inducing conditions, adipocyte numberincreases24,25 due to an expansion of the pre-adipocyte populationand causes Hmgic expression to become detectable in the adiposetissue. In the absence of Hmgic in vivo, pre-adipocyte proliferationis not as extensive as in wild-type mice and leads to mice with fewermature adipocytes and, hence, 87% less fat13 (Fig. 6). Similarly, theexpansion of the pre-adipocytes in Hmgic–/– and Hmgic+/– mice isseverely retarded under obesity-inducing conditions.

A major focus for therapy in obesity has been the aspect ofappetite suppression through interventions focusing on the cen-tral nervous system26. Our results advocate a complementaryapproach for the treatment of obesity focusing on the periphery,focusing on the actual tissue where the obese phenotype arises.Hmgic has a number of advantages as a potential target in obesity.First, its expression is limited to the adipose tissue of the adult inthe pathological state of obesity. Second, the gene-dosage effectof HMGIC in obesity suggests a titratable, pharmacological

Fig. 2 Hmgic–/– and Hmgic+/– mice areresistant to diet-induced obesity.Growth curves of wild-type (a),Hmgic–/– (b) and Hmgic+/– (c) mice fedeither a high-fat or a standard dietare shown. Growth measurementswere initiated at six weeks and atleast six mice (no sex-related differ-ences were observed) were used foreach growth curve. The total weightgain by wild-type mice on the high-fat diet was 21% (35.24±1.86 g versus29.04±1.59 g, P=0.03). *P<0.05. Bodyweight in grams (g) is provided as themean ±s.e.m.

Fig. 3 Representative Hmgic+/–, Lepob/+ intercross progeny at 30 weeks of age.

a b c

© 2000 Nature America Inc. • http://genetics.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

nature genetics • volume 24 • april 2000 379

action for HMGIC inhibitors. Our studies have shown a centralrole for HMGIC in obesity as two independent methods of obe-sity induction are alleviated by a partial or complete absence ofHMGIC, as well as its potential as a candidate target molecule forthis significant health problem27,28.

MethodsRT–PCR. We used total RNA (2 µg) for first-strand cDNA synthesis. Gene-specific primers for Hmgic and Gapd (encoding glyceraldehyde-3-phos-phate dehydrogenase) were used in the same 20-µl reaction, with Super-script II (Gibco BRL) reverse transcriptase. This reaction (2 µl) was used forthe following PCRs: (i) Hmgic (94 °C for 2 min; 40 cycles of 94 °C for 20 s, 65°C for 30 s, 72 °C for 30 s; 72 °C for 7 min), using the primers 5´–ATGGACTCATACACAGCAGCAGGAG–3´, and 5´–AGGGAATATAGAGAGGAGAGAGAGG–3´; (ii) Gapd (95 °C for 1 min, 49 cycles of 94 °C for 20 s,55 °C for 20 s, 72 °C for 30 s; 72 °C for 2 min), using the primers 5´–ATGGTGAAGGTCGGTGTGAA–3´, and 5´–ACCAGTAGACTCCAACGACAT–3´.

Diet-induced obesity. We obtained mice from the Hmgic+/– colony main-tained at UMDNJ and genotyped them by allele-specific PCR in a singlereaction. Hmgic–/– mice have a polymorphism due to an insertion of theneomycin gene at the Hmgic locus8. Two forward primers, one wild-type-specific (in the deleted portion of Hmgic; 5´–GTGTCCCTTGAAATGTTAGGCGGGG–3´) and one mutant specific (in the neomycin insertion;5´–AGGAGCCAAGCTGCTATTGG–3´), were designed along with a com-mon reverse primer (5´–CCCACTGCTCTGTTCCTTGC–3´) for PCR (94°C for 2 min; 35 cycles of 94 °C for 10 s, 58 °C for 30 s, 72 °C for 30 s; 72 °Cfor 7 min). The Hmgic allele-specific PCR gave 406-bp and 287-bp bandsfor mutant and wild-type alleles, respectively. Mice (4 weeks old) of the var-ious Hmgic genotypes were weaned on either a high-fat diet (45% total calo-ries in the form of fat; D12451, Research Diet) or a standard diet (23% totalcalories in the form of fat; Prolab RMH 2000, PMI Feeds). Mice had accessto food ad libitum and were weighed every two weeks for up to 32 weeks.

Generation of Hmgic–/–, Lepob/Lepob mice. Lepob/+ mice (C57Bl/6J; JacksonLaboratories) and Hmgic+/– mice8 were crossed to produce double-heterozy-

Fig. 4 Growth curves of Hmgic+/–, Lepob/+ intercross progeny. Body weight ingrams (g) is provided as the mean ±s.e.m. At least ten mice (no sex-related dif-ferences were observed) of each genotype were used for the growth curves.a, The morbid obesity of Hmgic+/+, Lepob/Lepob mice is ameliorated on theHmgic–/– background (Hmgic–/–, Lepob/Lepob mice). b, Hmgic has a gene-dosage–dependent effect on the obesity of Lepob/Lepob mice.

Fig. 5 Adiposity of Hmgic–/–,Lepob/Lepob and Hmgic+/–,Lepob/Lepob mice is reducedcompared with that of Lepob/Lepob mice. a, Mesenteric,parametrial (left), epididymal(left) and inguinal fat padswere dissected from 30-week-old mice. When comparedwith the corresponding fatpads of the wild-type, Lepob/Lepob genotype, all fat padsshowed a significant differ-ence (P<0.02) except for theHmgic+/–, Lepob/Lepob ingui-nal fat pad. Weights areshown in grams ±s.e.m. (n>5).b, Histological sections ofwild-type (left) and Hmgic–/–

(right) WAT show unilocular,polygonal cells with one largelipid depot (clear space).c, The Lepob/Lepob (left) andHmgic–/–, Lepob/Lepob (right)adipocytes exhibit the charac-teristic hypertrophy due togreater lipid storage. Scale bar,20 µm. d, Total cellularity ofmesenteric (right), parametrial(centre) and epididymal (left)fat pads is shown in millions ofcells ±s.e.m. (n>6). The cellular-ity of the three Hmgic–/–,Lepob/Lepob fat pads was sig-nificantly reduced (P<0.002)compared with that of theirLepob/Lepob counterparts.

a

b

a b

c d

© 2000 Nature America Inc. • http://genetics.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

380 nature genetics • volume 24 • april 2000

gous mice (Hmgic+/–, Lepob/+), which were then intercrossed to produce dou-ble-homozygous mice (Hmgic–/–, Lepob/Lepob). The mice were weaned at theage of four weeks, and weighed every two weeks thereafter. We first genotypedthe mice for the Hmgic locus, and then for the Lep locus. Lep genotypinginvolved two separate PCRs, one mutant-specific and one wild-type–specific.Lepob/Lepob mice have a C→T point mutation in Lep (ref. 18). Therefore, twospecific forward primers one for the mutant (5´–TGACCTGGAGAATCTCT–3´) and one for the wild-type (5´–GACCTGGAGAATCTCC–3´)Lep and a common reverse primer (5´–GCACATGGCTCTCTTCT–3´)were designed to distinguish between the two alleles. For each sample, we car-ried out two separate PCRs (94 °C for 4 min; 30 cycles of 94 °C for 30 s, 52 °Cfor 30 s, 72 °C for 30 s; 72 °C for 7 min), one wild-type–specific and onemutant-specific, for Lep genotyping. The mutant-specific PCR generated a295-bp product from the mutant Lep allele, but no product from the wild-type allele, and vice versa for the wild-type –specific PCR.

Fat pad weight and cellularity. Mice were killed by cervical dislocation andthe mesenteric, parametrial, epididymal and inguinal fat pads dissectedaccording to various anatomical landmarks20. The fat pads were washed insaline and weighed on a Mettler AE 50 fine balance, then flash frozen in liq-uid nitrogen for the cellularity experiments. We determined the cell num-ber in these fat pads by fluorometric DNA quantitation29.

Histology. WAT was dissected from approximately seven-month-old mice.Tissues were formaldehyde-fixed, paraffin-embedded, transversely sec-tioned (10 µm intervals) and stained with haematoxylin and eosin.

Food consumption. We housed six-month-old mice individually for oneweek before performing any measurements. Daily food consumption wasmeasured for at least one week for each mouse. Weighed food was providedto the mice ad libitum, and food consumed was calculated after 24 h, takingspillage into account.

Statistical analysis. All of the statistical analyses were carried out usingGraph Pad Prism software (Graph Pad Software). We calculated P valuesusing two-tailed, paired t-tests.

AcknowledgementsWe thank J. D’Armiento, HMGene, Inc. and members of the laboratory for acritical reading of the manuscript. K.C. is supported by National Institutes ofHealth grant CA77929 and NJ Commission on Science and Technology.

Received 5 October 1999; accepted 17 January 2000.

1. Grosschedl, R., Giese, K. & Pagel, J. HMG domain proteins: architectural elementsin the assembly of nucleoprotein structures. Trends Genet. 10, 94–100 (1994).

2. Bustin, M., Lehn, D. & Landsman, D. Structural features of the HMG chromosomalproteins and their genes. Biochem. Biophys. Acta 1049, 231–243 (1990).

3. Wolffe, A. Architectural transcription factors. Science 264, 1100–1101 (1994).4. Zhou, X. & Chada, K. HMGI family proteins: architectural transcription factors in

mammalian development and cancer. Keio J. Med. 47, 73–77 (1998).5. Mantovani, F. et al. NF-κB mediated transcriptional activation is enhanced by the

architectural factor HMGI-C. Nucleic Acids Res. 26, 1433–1439 (1998).6. Thanos, D. & Maniatis, T. Virus induction of human IFNβ gene expression requires

the assembly of an enhanceosome. Cell 83, 1091–1100 (1995).7. Carey, M. The enhancesome and transcriptional synergy. Cell 92, 5–8 (1998).8. Zhou, X., Benson, K., Ashar, H. & Chada, K. Mutation responsible for the mouse

pygmy phenotype in the developmentally regulated factor HMGI-C. Nature 376,771–774 (1995).

9. Hirning-Folz, U., Wilda, M., Rippe, V., Bullerdiek, J. & Hameister, H. The expressionpattern of the Hmgic gene during development. Genes Chrom. Cancer 23,350–357 (1998).

10. Schoenmakers, E. et al. Recurrent rearrangements in the high mobility groupprotein gene, HMGI-C, in benign mesenchymal tumors. Nature Genet. 10,436–443 (1995).

11. Tkachenko, A., Ashar, H.R., Meloni, A.M., Sandberg, A.A. & Chada, K.K.Misexpression of disrupted HMGI architectural factors activates alternativepathways of tumorigenesis. Cancer Res. 57, 2276–2280 (1997).

12. Ashar, H.R. et al. Disruption of the architectural factor HMGI-C: DNA-binding A-Thook motifs fused in lipomas to distinct transcriptional regulatory domains. Cell82, 57–65 (1995).

13. Benson, K. & Chada, K. Minimouse: phenotypic characterization of a transgenicinsertional mutant allelic to pygmy. Genet. Res. 64, 27–33 (1994).

14. West, D., Boozer, C., Moody, D. & Atkinson, R. Dietary obesity in the mouse:interaction of strain with diet composition. Am. J. Physiol. 262, R1025–R1032(1992).

15. Rogalla, P. et al. HMGI-C expression patterns in human tissues. Am. J. Pathol. 149,775–779 (1996).

16. Green, M. in Genetic Variants and Strains of the Laboratory Mouse (eds Lyon, M.& Searle, A.) 265–266 (Oxford University Press, Oxford and New York, 1989).

17. Herberg, L. & Coleman, D. Laboratory animals exhibiting obesity and diabetessyndromes. Metabolism 26, 59–99 (1977).

18. Zhang, Y. et al. Positional cloning of the mouse obese gene and its humanhomologue. Nature 372, 425–431 (1994).

19. Bates, M., Nauss, S., Hagman, N. & Mayer, J. Fat metabolism in three forms ofexperimental obesity. Am. J. Physiol. 180, 301–303 (1955).

20. Johnson, P. & Hirsch, J. Cellularity of adipose depots in six strains of geneticallyobese mice. J. Lipid Res. 13, 2–11 (1972).

21. Frankel, W. & Schork, N. Who’s afraid of epistasis? Nature Genet. 14, 371–373(1996).

22. Reeves, R., Langan, T. & Nissen, M. Phosphorylation of the DNA-binding domainof nonhistone high-mobility group I protein by cdc2 kinase: reduction of bindingaffinity. Proc. Natl Acad. Sci. USA 88, 1671–1675 (1991).

23. Tavtigian, S., Zabludoff, S. & Wold, B. Cloning of mid-G1 serum response genesand identification of a subset regulated by conditional myc expression. Mol. Biol.Cell 5, 375–388 (1994).

24. Klyde, B. & Hirsch, J. Increased cellular proliferation in adipose tissue of adult ratsfed a high-fat diet. J. Lipid Res. 20, 705–715 (1979).

25. Ailhaud, G., Grimaldi, P. & Negrel, R. Cellular and molecular aspects of adiposetissue development. Annu. Rev. Nutr. 12, 207–233 (1992).

26. Flier, J. & Maratos-Flier, E. Obesity and the hypothalamus: novel peptides for newpathways. Cell 92, 437–440 (1998).

27. Pi-Sunyer, F.X. Medical hazards of obesity. Ann. Intern. Med. 119, 655–670 (1993).28. Flegal, K., Carrol, M., Kuczmarski, R. & Johnson, C. Overweight and obesity in the

United States: prevalence and trends, 1960–1994. Int. J. Obes. Relat. Metab.Disord. 22, 39–47 (1998).

29. Cummings, D.E. et al. Genetically lean mice result from targeted disruption of theRIIβ subunit of protein kinase A. Nature 382, 622–626 (1996).

Fig. 6 A model for HMGIC in adipogenesis and obesity.Multipotent stem cells commit themselves to theadipocyte lineage, and post-natally develop into WAT.Adult WAT is a complex cellular milieu that is com-posed of a variety of cells, including proliferative pre-adipocytes (non–lipid-filled cells, shown in smallirregular shape) and terminally differentiated, non-proliferative adipocytes (lipid-filled cells, shown as cir-cles). In response to an obesity-induced stimulus,wild-type adipose tissue (top) increases in mass byadipocyte hyperplasia due to the proliferation of pre-adipocytes and their subsequent differentiation toadipocytes and adipocyte hypertrophy. In contrast, theHmgic–/– mice (bottom) resist obesity because theHmgic–/– pre-adipocytes do not proliferate to the sameextent as their wild-type counterparts, although theHmgic–/– adipocytes are able to become hypertrophic.

© 2000 Nature America Inc. • http://genetics.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://g

enet

ics.

nat

ure

.co

m