International Journal of Obesity (2000) 24, Suppl 4, S11±S14 ß 2000 Macmillan Publishers Ltd All rights reserved 0307±0565/00 $15.00 www.nature.com/ijo A-ZIP=F-1 mice lacking white fat: a model for understanding lipoatrophic

ML Reitman1* and O Gavrilova1

1Diabetes Branch, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, Maryland, USA

The A-ZIP=F-1 mouse is lacking virtually all white adipose tissue. Like humans with extensive de®ciencies of adipose tissue, the A-ZIP=F-1 mice develop a severe form of resistant diabetes. We have studied the physiology of the A-ZIP=F-1 mice. Their adaptation to fasting is notable for its rapidity and the use of torpor, a hibernation-like state, to minimize energy needs. Transplantation of adipose tissue reversed the metabolic manifestations in the mice, demonstrating that the lack of adipose tissue is the cause of the . Leptin replacement is not very effective in reversing the diabetes of the A-ZIP=F-1 mice, which contrasts with its ef®cacy in the aP2-SREBP-1c mouse. International Journal of Obesity (2000) 24, Suppl 4, S11±S14

Keywords: lipoatrophy=; diabetes; insulin resistance; adipose tissue; fasting; torpor; leptin

Introduction to human identi®ed.7 However, in most patients, the gene iden- tities, mechanisms and even tissues of action remain a lipodystrophy=lipoatrophy mystery. The =lipoatrophies are a heterogeneous group of rare disorders, having in common a diminished amount of adipose tissue.1,2 Patients with these disorders Murine models of fat exhibit a wide range in the distribution of fat loss, from ablation=lipoatrophy generalized to partial to focal. The age of onset ranges from congenital through adulthood, and is often char- We have developed a transgenic mouse, named A- acteristic of the particular subtype of the disease. ZIP=F-1, which has virtually no white adipose tissue Patients with lipoatrophy, particularly those with greater (WAT).8 These mice express, selectively in adipose degrees of fat de®ciency, develop . This is tissue, a dominant negative protein that heterodi- an intriguing contrast to obesity, in which an excess, merizes with and inactivates members of the C=EBP rather than a de®ciency, of adipose tissue is associated and JUN families of B-ZIP transcription factors. The with diabetes. A-ZIP=F-1 phenotype strikingly resembles that of There are multiple causes of lipoatrophic diabetes, humans with severe lipoatrophic diabetes, such as such as genetic, autoimmune and presumed toxic (eg congenital generalized lipodystrophy (Berardinelli ± HIV patients treated with protease inhibitors). Seip syndrome). Transgenic mice with similar, but Mechanisms both intrinsic and extrinsic to the adipo- less severe, phenotypes resulted from adipose expres- cyte have been proposed, including destruction or sion of a modi®ed diphtheria toxin9,10 or a constitu- inadequate production of adipocytes, abnormal non- tively active SREBP-1c transcription factor.11 A adipocyte signals affecting metabolism, and intrinsic number of other mice with diminished amounts of adipocyte abnormalities that alter their metabolic fat have been reported. In some the reduced fat mass response.3 A number of candidate genes have been may be due to a nonspeci®c manifestation of a sick or excluded.4 Recently, causative mutations, in the lamin undernourished animal, and not a primary adipose A=C gene, were identi®ed for the Dunnigan variety of tissue phenotype. Here we will discuss the physiology partial lipodystrophy.5,6 The chromosomal location of the A-ZIP=F-1 mice. (9q34) of the mutation in a subset of patients with congenital generalized lipodystrophy has also been General phenotype of the A-ZIP=F-1 mice (Table 1)8 *Correspondence: ML Reitman, Diabetes Branch, National The A-ZIP=F-1 mice have virtually no visible WAT at Institute of Diabetes and Digestive and Kidney Diseases, NIH Building 10, Room 8N-250, Bethesda, MD 20892-1770, USA. any time during development. However, the mice are E-mail: [email protected] not completely WAT-free, since some adipocytes are Lipoatrophic diabetes ML Reitman and O Gavrilova S12 Table 1 Summary of the A-ZIP=F-1 phenotype

Observed phenotype in A-ZIP=F-1 mice Possible explanation Reference

Near-complete lack of white adipose tissue Direct effect of transgene 8 BAT initially normal, but becomes inactive and Direct effect of transgene 8 reduced in amount with aging Elevated serum FFA and triglycerides Lack of WAT to store triglyceride; hepatic production 8 Fatty liver Lack of WAT to store triglyceride; hepatic production; low leptin 8,18 Increased muscle triglyceride levels Lack of WAT to store triglyceride 14 Insulin resistance, Increased muscle and liver triglyceride levels 8,14 Increased food, water intake Trying to keep up with the glucosuria; low leptin 8,18 Somatomegaly, visceral organomegaly IGF receptor stimulation by elevated insulin levels 8 Initially runt ?Unable to handle high fat content of milk 8 Reduced viability ?Hyperosmolar coma leading to death 8 Reduced fertility Normal adaptation to low triglyceride stores; low leptin 8 Increased bone density Low leptin 19 Accelerated response to fasting Normal adaptation to low triglyceride stores 12 Torpor when fasting Normal adaptation to low triglyceride stores 12 Elevated glucocorticoid levels ?Low leptin 12

visible upon microscopic examination. We estimate that body weight. This continued growth may be due to the WAT mass is reduced by  99%. Extrahepatic insulin-like growth factor (IGF) effects of the very triglyceride content is reduced 20-fold. In the A- high insulin levels. Of note, the body weight under- ZIP=F-1 mouse, transgene expression is driven by the estimates the difference in lean body mass between A- aP2=422 adipose fatty acid binding protein promo- ZIP=F-1 and wild-type mice, since the A-ZIP=F-1 ter=enhancer, which turns on during adipose differentia- body weight is essentially a lean body mass, while tion later than markers such as PPARg-2 and C=EBPa. normal adult mice have  15% fat. Thus, one might have expected that some adipocytes We routinely propagate the A-ZIP=F-1 mice by would be observed before the effects of the dominant breeding hemizygous transgenic males with wild- negative protein would ablate the cells. However, we type females. A-ZIP=F-1 mice are born at approxi- rarely observe lipid-containing cells, so the transgene mately the expected frequency, demonstrating no must act before adipocytes can accumulate lipid. prenatal loss. However the mortality between birth At birth, brown adipose tissue (BAT) is found in the and weaning can approach 50%. After weaning, the A-ZIP=F-1 mice in near-wild-type amounts. Histolo- A-ZIP=F-1 mice have a much reduced, but still higher gically, this BAT appears normal and active. However, mortality than wild-type mice. No pups born to as the mice age, the BAT decreases in size and transgenic mothers have survived to weaning, possi- acquires a less eosinophilic staining pattern, with bly due to poor milk production. All experiments to more lipid per cell. Expression of UCP-1, the marker date have been performed using hemizygous mice of active BAT, is also very low. Thus the BAT with a pure FVB=N background; at this time neither undergoes accelerated, premature involution, becom- homozygous mice nor the effects of genetic back- ing reduced in amount and inactive. Since the brown ground have been studied. Many aspects of the phe- adipocytes express the transgene, it is likely that the notype discussed below show a gender dimorphism, BAT inactivation is the direct result of effects of the A- with males being more severely affected. ZIP=F protein. We do not know why transgene expres- In addition to the lack of WAT and the abnormal sion results in a milder phenotype in the BAT, as BAT, the A-ZIP=F-1 mice have other anatomic compared to the WAT. One possible explanation is abnormalities. The liver is greatly enlarged and that the A-ZIP=F transgene is expressed at different engorged with lipid, with a centrilobular pattern to levels in WAT and BAT. Another is that the transgene the steatosis. Quantitatively, the liver triglyceride inactivates a slightly different set of endogenous B- content is up 11-fold. The kidneys, heart and spleen ZIP transcription factors in WAT than in BAT. are enlarged. The mice are severely diabetic, with The growth pattern of the A-ZIP=F-1 mice is elevated serum glucose (3-fold) and insulin (50 ± 400- complicated, with a number of different phases. fold) levels, accompanied by polyuria, polydypsia and Their size at birth is indistinguishable from wild- hyperphagia. The pattern is that of type 2 diabetes. In type littermates. However, the A-ZIP=F-1 mice imme- addition, the mice have elevated serum free fatty acids diately fall behind in growth, and by one week of age, (FFA; 2-fold), and triglycerides (3 ± 5-fold) and they are only half the weight of their littermates. At reduced leptin levels (20-fold). weaning (3 ± 4 weeks of age), the A-ZIP=F-1 mice begin catch-up growth, and by  8 weeks of age they weigh as much as their littermates. We postulate that A-ZIP=F-1 mice respond to fasting by the poor growth while nursing is due to the high lipid entering torpor12 content of milk, which the A-ZIP=F-1 pups cannot handle. The A-ZIP=F-1 mice continue to grow in The total nonprotein energy stores in the A-ZIP=F-1 adulthood, eventually surpassing their littermates in mice are reduced 10-fold, to 8.7 kcal=mouse. This is

International Journal of Obesity Lipoatrophic diabetes ML Reitman and O Gavrilova S13 less than the resting metabolic expenditure for one day reverse the metabolic derangement. We transplanted (  12 kcal), yet the A-ZIP=F-1 mice typically handle wild-type fat into A-ZIP=F-1 mice and indeed fasting for a day without problems. We therefore observed partial or complete reversal of the all aspects studied the response to fasting in detail. The A- of the diabetic phenotype. The mice developed ZIP=F-1 adaptation to fasting can be summarized approximately normal glucose levels, dramatically brie¯y as a normal response to an abnormal situation: lowered insulin levels, and improved liver and by 2 ± 4 h of fasting, the blood glucose is reduced to muscle insulin sensitivity. The hyperphagia, polyuria, control levels, and thereafter is defended at this level. hepatic steatosis and somatomegaly all improved. In contrast, after a 24 h fast, muscle and liver trigly- However, the reduction in triglyceride and FFA ceride and glycogen are all severely depleted, to much levels was modest. Donor fat taken from parame- greater degree than in controls. Serum free (nonester- trial and subcutaneous sites was equally effective in i®ed) fatty acids (FFA) and triglycerides also drop to reversing the phenotype. The bene®cial effects of very low levels, presumably re¯ecting the exhaustion transplantation were dose-dependent and required of the tissue triglyceride stores. Normally, serum b- near-physiologic amounts of transplanted fat. These hydroxybutyrate levels rise with fasting, re¯ecting experiments demonstrate that the lack of WAT is conversion of FFA to ketone bodies, predominantly causing the metabolic abnormalities. Additionally, by liver, in order to keep the brain supplied with fat transplantation is a new technique for studying energy. However, b-hydroxybutyrate levels do not adipose tissue physiology and the mechanisms of rise during fasting in A-ZIP=F-1 mice. This might insulin resistance and type 2 diabetes. be due to the paucity of FFA or to abnormal regulation of the conversion process. During fasting the blood urea nitrogen rises in the A-ZIP=F-1, but not control mice, indicating that the A-ZIP=F-1 mice catabolize Role of leptin in causing the protein in order to defend their glucose levels. In phenotype of the A-ZIP=F-1 mice addition, the A-ZIP=F-1 mice adapt to fasting with a drop in metabolic rate, which is already evident after Leptin's role in increasing insulin sensitivity has been only a few hours of fasting. To summarize, in A- demonstrated in a number of ways. Leptin treatment of ZIP=F-1 mice a 24 h fast exhausts the carbohydrate ob=ob mice reverses their diabetes. Leptin treatment of and triglyceride stores, but the mice adapt by normal rodents also provides more evidence for bene- catabolism of muscle. ®cial effects of leptin on glucose homeostasis.15,16 In the The A-ZIP=F-1 mice also use a remarkable method aP2-SREBP-1c transgenic model of lipoatrophy, leptin to conserve fuel during fasting. During a 24 h fast, A- treatment is remarkably effective at reversing the insulin ZIP=F-1 (but not control) mice enter a hibernation- resistance.17 However, leptin is only marginally effec- like state called torpor. This saves the energy usually tive in reversing the diabetes of the A-ZIP=F-1 mice.18 used to defend body temperature (one-third of resting The difference between the A-ZIP=F-1 and aP2- metabolic rate at room temperature) and also reduces SREBP-1c mice is probably due to the greater degree need much further. During torpor, the A-ZIP=F-1 of fat loss in the A-ZIP=F-1, although transgene-speci®c mice develop a core body temperature of  2C effects or the different genetic backgrounds may con- above ambient. Although not widely known, torpor tribute. The aP2-SREBP-1c mice have more residual is actually a normal response of mice to insuf®cient adipose tissue. In these mice leptin appears to be limiting food in a quiet and suf®ciently cold environment. and its replacement reverses the diabetes. The A-ZIP=F- Leptin-de®cient ob=ob mice readily enter torpor 1 mice must be missing functions provided by adipose with fasting, and leptin infusion prevents this. In tissue, in addition to leptin secretion. One possibility is contrast, neither leptin nor thyroid hormone prevented that adipose tissue's metabolic functions, affecting fatty torpor in A-ZIP=F-1 mice. These data suggest that acid and triglyceride metabolism are needed. Alterna- there are at least two signals for entry into torpor in tively, adipose tissue could contribute via endocrine mice, a low leptin level and another signal that is mechanisms, secreting hormones, either known or yet to independent of leptin and thyroid hormone levels. be discovered. In conclusion, leptin de®ciency contri- butes to the insulin resistance of generalized lipoatro- phy, but is neither the sole nor the major cause of insulin resistance in severe forms of this disease. Reversal of diabetes in the A-ZIP=F-1 mice by transplantation of adipose tissue13,14 Perspective=future directions The A-ZIP=F-1 mice provide a way to rigorously test The full potential of the A-ZIP=F-1 mice is only the role of WAT in the etiology of lipoatrophic beginning to be realized. One major question is the diabetes. If the lack of fat is causing the diabetes, mechanism underlying the diabetes of the A-ZIP=F-1 reconstitution with normal adipose tissue should mice. If tissue triglyceride is causing insulin

International Journal of Obesity Lipoatrophic diabetes ML Reitman and O Gavrilova S14 resistance, how does it do so? Increased tissue (parti- 7 Garg A, Wilson R, Barnes R, Arioglu E, Zaidi Z, Gurakan F, cularly muscle) triglyceride has also been proposed to Kocak N, O'Rahilly S, Taylor SI, Patel SB, Bowcock AM. A cause the much more common obesity-associated type gene for congenital generalized lipodystrophy maps to human chromosome 9q34. J Clin Endocrinol Metab 1999; 2 diabetes. Thus mechanisms elucidated with lipoa- 84: 3390 ± 3394. trophic mice will supply ideas to be tested in other 8 Moitra J, Mason MM, Olive M, Krylov D, Gavrilova O, models. Marcus-Samuels B, Feigenbaum L, Lee E, Aoyama T, Reconstitution of adipose tissue into the A-ZIP=F-1 Eckhaus M, Reitman ML, Vinson C. Life without white fat: a transgenic mouse. Genes Devl 1998; 12: 3168 ± 3181. mice by transplantation is a powerful tool. For exam- 9 Ross SR, Graves RA, Spiegelman BM. Targeted expression of ple, the role of particular genes can be studied by a toxin gene to adipose tissue: transgenic mice resistant to using donor fat from genetically modi®ed mice. This obesity. Genes Devl 1993; 7: 1318 ± 1324. approach is a possible alternative to tissue-speci®c 10 Burant CF, Sreenan S, Hirano K, Ta TA, Lohmiller J, Lukens recombination as a way to achieve adipose-speci®c J, Davidson NO, Ross S, Graves RA. Troglitazone action is independent of adipose tissue. knockouts. For example, leptin physiology can be J Clin Invest 1997; 100: 2900 ± 2908. examined by transplantation of WAT from ob=ob 11 Shimomura I, Hammer RE, Richardson JA, Ikemoto S, Bash- and db=db mice into the A-ZIP=F-1 animals. makov Y, Goldstein JL, Brown MS. Insulin resistance and The A-ZIP=F-1 mice are also a wonderful system diabetes mellitus in transgenic mice expressing nuclear for devising and testing treatments for human lipoa- SREBP-1c in adipose tissue: model for congenital generalized lipodystrophy. Genes Devl 1998; 12: 3182 ± 3194. trophic diabetes. Possible studies include drug trials 12 Gavrilova O, Leon LR, Marcus-Samuels B, Mason MM, and combinations of drugs with fat transplantation. Castle AL, Refetoff S, Vinson C, Reitman ML. Torpor in The uses of lipoatrophic mice in elucidating the mice is induced by both leptin-dependent and -independent physiology and pathophysiology of adipose tissue mechanisms. Proc Natl Acad Sci USA 1999; 96: 14623 ± are limited only by our imagination. 14628. 13 Gavrilova O, Marcus-Samuels B, Graham D, Kim JK, Shul- man GI, Castle AL, Vinson C, Eckhaus M, Reitman ML. References Surgical implantation of adipose tissue reverses diabetes in 1 Foster DW. The lipodystrophies and other rare disorders of lipoatrophic mice. J Clin Invest 2000; 105: 271 ± 278. adipose tissue. In: Fauci AS, Braunwald E, Isselbacher KJ, 14 Kim JK, Gavrilova O, Chen Y, Reitman ML, Shulman GI. Wilson JD, Martin JB, Kasper DL, Hauser SL, Longo DL Mechanism of insulin resistance in A-ZIP=F-1 fatless mice. (eds). Harrison's Principles of Internal Medicine. McGraw- J Biol Chem 2000; 275: 8456 ± 8460. Hill: New York, 1998, pp 2209 ± 2214. 15 Kamohara S, Burcelin R, Halaas JL, Friedman JM, Charron 2 Seip M, Trygstad O. Generalized lipodystrophy, congenital MJ. Acute stimulation of glucose metabolism in mice by leptin and acquired (lipoatrophy). Acta Paediatr Suppl 1996; 413: treatment. Nature 1997; 389: 374 ± 377. 2 ± 28. 16 Rossetti L, Massillon D, Barzilai N, Vuguin P, Chen W, 3 Rossini AA, Cahill GFJ. Lipoatrophic diabetes. In: DeGroot Hawkins M, Wu J, Wang J. Short term effects of leptin on LJ, Cahill GF Jr, Martini L, Nelson DH, Odell WD, Potts JT, hepatic gluconeogenesis and in vivo insulin action. J Biol Jr, E Steinberger E, Winegrad AI (eds). Endocrinology. Grune Chem 1997; 272: 27758 ± 27763. & Stratton: New York, 1979, pp 1093 ± 1097. 17 Shimomura I, Hammer RE, Ikemoto S, Brown MS, Goldstein 4 Vigouroux C, Khallouf E, Bourut C, Robert JJ, de Kerdanet JL. Leptin reverses insulin resistance and diabetes mellitus M, Tubiana-Ru® N, Faure S, Weissenbach J, Capeau J, Magre in mice with congenital lipodystrophy. Nature 1999; 401: J. Genetic exclusion of 14 candidate genes in lipoatropic 73 ± 76. diabetes using linkage analysis in 10 consanguineous families. 18 Gavrilova O, Marcus-Samuels B, Leon LR, Vinson C, Reit- J Clin Endocrinol Metab 1997; 82: 3438 ± 3444. man ML. Leptin and diabetes in lipoatrophic mice. Nature 5 Cao H, Hegele RA. Nuclear lamin A=C R482Q mutation in 2000; 403: 850. Canadian kindreds with Dunnigan-type familial partial lipo- 19 Ducy P, Amling M, Takeda S, Priemel M, Schilling AF, Beil dystrophy. Hum Mol Genet 2000; 9: 109 ± 112. FT, Shen J, Vinson C, Rueger JM, Karsenty G. Leptin inhibits 6 Shackleton S, Lloyd DJ, Jackson SN, Evans R, Niermeijer bone formation through a hypothalamic relay: a central control MF, Singh BM, Schmidt H, Brabant G, Kumar S, Durrington of bone mass. Cell 2000; 100: 197 ± 207. PN, Gregory S, O'Rahilly S, Trembath RC. LMNA, encoding lamin A=C, is mutated in partial lipodystrophy. Nature Genet 2000; 24: 153 ± 156.

International Journal of Obesity