Insulin Action and Its Disturbances in Disease - part 5 pptx

232 RELATIONSHIP BETWEEN FAT DISTRIBUTION AND INSULIN RESISTANCE
130. Tomlinson, J. W., Sinha, B., Bujalska, I., Hewison, M. and Stewart, P. M. (2002)
Expression of 11beta-hydroxysteroid dehydrogenase type 1 in adipose tissue is not
increased in human obesity. J Clin Endocrinol Metab 87 (12), 5630–5635.
131. Moore, J. S., Monson, J. P., Kaltsas, G., Putignano, P., Wood, P. J., Sheppard, M. C.,
Besser, G. M., Taylor, N. F. and Stewart, P. M. (1999) Modulation of 11beta-
hydroxysteroid dehydrogenase isozymes by growth hormone and insulin-like growth
factor: in vivo and in vitro studies. J Clin Endocrinol Metab 84 (11), 4172–4177.
132. Tomlinson, J. W., Moore, J., Cooper, M. S., Bujalska, I., Shahmanesh, M., Burt, C.,
Strain, A., Hewison, M. and Stewart, P. M. (2001) Regulation of expression of 11beta-
hydroxysteroid dehydrogenase type 1 in adipose tissue: tissue-specific induction by
cytokines. Endocrinology 142 (5), 1982–1989.
133. Bujalska, I. J., Kumar, S., Hewison, M. and Stewart, P. M. (1999) Differentiation of
adipose stromal cells: The roles of glucocorticoids and 11β hydroxysteroid dehydroge-
nase. Endocrinology 140 (7), 3188–3196.
134. Palermo, M., Shackleton, C. H. L., Mantero, F. and Stewart, P. M. (1996) Urinary free
cortisone and the assessment of 11β hydroxysteroid dehydrogenase activity in man. Clin
Endocrinol (Oxf) 45, 605–611.
135. Stewart, P. M. and Krozowski, Z. S. (1999) 11 beta-hydroxysteroid dehydrogenase. Vit
Horm 57, 249–324.
136. Rask, E., Olsson, T., Soderberg, S., Andrew, R., Livingstone, D. E., Johnson, O. and
Walker, B. R. (2001) Tissue-specific dysregulation of cortisol metabolism in human
obesity. J Clin Endocrinol Metab 86 (3), 1418–1421.
137. Rask, E., Walker, B. R., Soderberg, S., Livingstone, D. E., Eliasson, M., Johnson, O.,
Andrew, R. and Olsson, T. (2002) Tissue-specific changes in peripheral cortisol meta-
bolism in obese women: increased adipose 11beta-hydroxysteroid dehydrogenase type
1 activity. J Clin Endocrinol Metab 87 (7), 3330–3336.
138. Weaver, J. U., Taylor, N. F., Monson, J. P., Wood, P. J. and Kelly, W. F. (1998) Sex-
ual dimorphism in 11 beta hydroxysteroid dehydrogenase activity and its relation to fat
distribution and insulin sensitivity; a study in hypopituitary subjects. Clin Endocrinol
(Oxf) 49 (1), 13–20.

139. Livingstone, D. E., Jones, G. C., Smith, K., Jamieson, P. M., Andrew, R., Kenyon, C. J.
and Walker, B. R. (2000) Understanding the role of glucocorticoids in obesity: tissue-
specific alterations of corticosterone metabolism in obese Zucker rats. Endocrinology
141 (2), 560–563.
140. Kotelevtsev, Y., Brown, R. W., Fleming, S., Kenyon, C., Edwards, C. R., Seckl, J. R.
and Mullins, J. J. (1999) Hypertension in mice lacking 11beta-hydroxysteroid dehydro-
genase type 2. J Clin Invest 103 (5), 683–689.
141. Masuzaki, H., Paterson, J., Shinyama, H., Morton, N. M., Mullins, J. J., Seckl, J. R.
and Flier, J. S. (2001) A transgenic model of visceral obesity and the metabolic syn-
drome. Science 294 (5549), 2166–2170.
142. Horton, R. and Tait, J. F. (1966) Androstenedione production and interconversion rates
measured in peripheral blood and studies on the possible site of its conversion to
testosterone. J Clin Invest 45, 301–313.
143. Bruch, H. R., Wolf, L., Budde, R., Romalo, G. and Schweikert, H. U. (1992) Andro-
stenedione metabolism in cultured human osteoblast-like cells. J Clin Endocrinol Metab
75 (1), 101–105.
144. Jakob, F., Hormann, D., Seufert, J., Schneider, D. and Kohrle, J. (1995) Expression and
regulation of aromatase cytochrome P450 in THP-1 human myeloid leukemia cells. Mol
Cell Endocrinol 110, 27–33.
REFERENCES 233
145. Baynard, F., Clamens, S., Delsol, G., Blaes, N., Maret, A. and Faye, J. C. (1995) Oes-
trogen biosynthesis, oestrogen metabolism and functional oestrogen receptors in bovine
aortic endothelial cells. Ciba Foundation Symp 191, 122–132.
146. Labrie, F. (1991) Intracrinology. Mol Cell Endocrinol 78, 113–118.
147. Labrie, F., Belanger, A., Cusan, L. and Candas, B. (1997) Physiological changes in
dehydroepiandrosterone are not reflected by serum levels of active androgens and estro-
gens but their metabolites: intracrinology. J Clin Endocrinol Metab 82, 2403–2409.
148. Deslypere, J. P., Verdonck, L. and Vermeulen, A. (1985) Fat tissue: a steroid reservoir
and site of steroid metabolism. J Clin Endocrinol Metab 61 (3), 564–570.
149. Simpson, E. R., Merrill, J. C., Hollub, A. J., Lorence, S. G. and Mendelson, C. R.

(1989) Regulation of oestrogen biosynthesis by human adipose cells. Endo Rev 10,
136–148.
150. Simpson, E. R., Mahendroo, M. S., Means, G. D., Kilgore, M. W., Hinselwood, M. M.,
Graham-Lorence, S., Amarneh, B., Ito, Y., Fisher, C. R., Micheal, M. D., Mendelson,
C. R. and Bulun, S. E. (1994) Aromatase cytochrome P450, the enzyme responsible for
estrogen biosynthesis. Endo Rev 15, 352–354.
151. Bulun, S. E. and Simpson, E. R. (1994) Competitive reverse transcription-polymerase
chain reaction analysis indicates that levels of aromatase cytochrome P450 transcripts
in adipose tissue of buttocks, thighs, and abdomen of women increase with advancing
age. J Clin Endocrinol Metab 78 (2), 428–432.
152. Zhao, Y., Nichols, J. E., Valdez, R., Mendelson, C. R. and Simpson, E. R. (1996) Tumor
necrosis factor-alpha stimulates aromatase gene expression in human adipose stromal
cells through use of an activating protein-1 binding site upstream of promoter 1.4. Mol
Endocrinol 10 (11), 1350–1357.
153. Simpson, E. R., Zhao, Y., Agarwal, V. R., Michael, M. D., Bulun, S. E., Hinshelwood,
M. M., Graham-Lorence, S., Sun, T., Fisher, C. R., Qin, K. and Mendelson, C. R.
(1997) Aromatase expression in health and disease. Recent Prog Horm Res 52,
185–213.
154. Jones, M. E. E., Thornburn, A. W., Britt, K. L., Hewitt, K. N., Wreford, N. G., Proi-
etto, J., Oz, O. K., Leury, B. J., Robertson, K. M., Yao, S. and Simpson, E. R. (2000)
Aromatase-deficient (ArKO) mice have a phenotype of increased adiposity. Proc Natl
Acad Sci 97, 12 735–12 740.
155. Poehlman, E. T. and Tchernof, A. (1998) Traversing the menopause: changes in energy
expenditure and body composition. Coron Artery Dis 9 (12), 799–803.
156. Hope, P. J., Turnbull, H., Breed, W., Morley, J. E., Horowitz, M. and Wittert, G. A.
(2000) The effect of ovarian steroids and photoperiod on body fat stores and uncou-
pling protein 2 in the marsupial Sminthopsis crassicaudata. Physiol Behav 69 (4/5),
463–470.
157. Misso, M. L., Murata, Y., Boon, W. C., Jones, M. E., Britt, K. L. and Simpson, E. R.
(2003) Cellular and molecular characterization of the adipose phenotype of the

aromatase-deficient mouse. Endocrinology 144 (4), 1474–1480.
158. Belanger, C., Luu-The, V., Dupont, P. and Tchernof, A. (2002) Adipose tissue intra-
crinology: potential importance of local androgen/estrogen metabolism in the regulation
of adiposity. Horm Metab Res 34 (11/12), 737–745.
159. Bulun, S. E. (1998) The syndrome of aromatase deficiency. Curr Opin Endo Diab 5,
350–356.
160. Rosenbaum, M. and Leibel, R. (1999) Role of gonadal steroids in the sexual dimor-
phism in body composition and circulating concentrations of leptin. J Clin Endocrinol
Metab 84 (6), 1784–1789.
234 RELATIONSHIP BETWEEN FAT DISTRIBUTION AND INSULIN RESISTANCE
161. Rebuff
´
e-Scrive, M., Enk, L., Crona, N., Lonnroth, P., Abrahamsson, L., Smith, U. and
Bjorntorp, P. (1985) Fat cell metabolism in different regions in women. Effect of men-
strual cycle, pregnancy, and lactation. J Clin Invest 75, 1973–1976.
162. Rebuff
´
e-Scrive, M., Eldh, J., Haafstrom, L. and Bjorntorp, P. (1986) Metabolism of
mammary, abdominal and femoral adipocytes in women before and after the menopause.
Metabolism 35, 792–797.
163. Price, T. M., O’Brien, S. N., Welter, B. H., George, R., Anandjiwala, J. and Kilgore, M.
(1998) Estrogen regulation of adipose tissue lipoprotein lipase – possible mechanism of
body fat distribution. Am J Obstet Gynecol 178, 101–107.
164. Wing, R. R., Matthews, K. A., Kuller, L. H., Meilahn, E. N. and Plantinga, P. L. (1991)
Weight gain at the time of the menopause. Arch Intern Med 151, 97–102.
165. Pasquali, R., Casimirri, F., Labate, A. M., Tortelli, O., Pascal, G., Anconetani, B.,
Gatto, M. R., Flamia, R., Capelli, M. and Barbara, L. (1994) Body weight, fat dis-
tribution and the menopausal status in women. The VMH Collaborative Group. Int J
Obes Relat Metab Disord 18 (9), 614–621.
166. Haarbo, J., Mareslew, U., Gotfredsen, A. and Kritiansen, C. (1991) Postmenopausal

hormone replacement therapy prevents central redistribution of body fat after the meno-
pause. Metabolism 10, 1323–1326.
167. Troisi, R., Wolf, A., Mason, J., Klinger, K. and Colditz, G. (1995) Relationship of body
fat distribution to reproductive factors in pre-and postmenopausal women. Obesity Res
3, 143–151.
168. Krotkiewski, M. and Bjorntorp, P. (1976) The effect of progesterone and of insulin
administration on regional adipose tissue cellularity in the rat. Acta Physiol Scand 96,
122–127.
169. Steingrimsdottir, L., Greenwood, M. R. C. and Brasel, J. A. (1980) Effect of preg-
nancy, lactation and a high fat diet on adipose tissue in Osborne-Mendal rats. JNutr
110, 600–609.
170. Iverius, P H. and Brunzell, J. D. (1988) Relationship between lipoprotein lipase activ-
ity and plasma sex steroid levels in obese women. J Clin Invest 82, 1106–1111.
171. Rebuff
´
e-Scrive, M., Basdevant, A. and Guy-Grand, B. (1983) Effect of local applica-
tion of progesterone on human adipose tissue lipoprotein lipase. Horm Metab Res 15,
566.
172. Mayes, J. S., McCann, J. P., Ownbey, T. C. and Watson, G. H. (1996) Regional dif-
ferences and up-regulation of progesterone receptors in adipose tissues from oestrogen-
treated sheep. J Endocrinol 148, 19–25.
173. Roncari, D. A. K. and Van, L. R. (1978) Promotion of human adipocyte precursor repli-
cationby17β -estradiol in culture. J Clin Invest 62, 503–508.
174. Rubin, G., Jones, M., Clyne, C., Zhao, Y. and Simpson, E. (1997) PPARgamma estro-
gens and aromatase expression in adipose tissue. 80th Annual Meeting of The Endocrine
Society, New Orleans, LA, P1–22.
175. Anderson, L. A., McTernan, P. G., Barnett, A. H. and Kumar, S. (2001) The roles of
androgen and oestrogen on proliferation: effect of site and gender. J Clin Endocrinol
Metab 86, 5045–5051.
176. Casabiell, X., Pineiro, V., Peino, R., Lage, M., Camina, J., Gallego, R., Vallejo, L. G.,

Dieguez, C. and Casanueva, F. F. (1998) Gender differences in both spontaneous and
stimulated leptin secretion by omental adipose tissue ‘in-vitro’: dexamethasone and
estradiol stimulate leptin release in women but not in men. J Clin Endocrinol Metab
83, 2149–2155.
177. Marin, P. and Arvers, S. (1998) Androgens and abdominal obesity. Ballieres Clin Endo-
crinol Metab 12 (3), 441–451 [review].
REFERENCES 235
178. Bhasin, S., Storer, T. W., Berman, N., Callegari, C., Clevenger, B., Phillips, J., Bun-
nell, T. J., Tricker, R., Shirazi, A. and Casaburi, R. (1996) The effects of supraphysi-
ologic doses of testosterone on muscle size and strength in man. N Engl J Med 335,
1–7.
179. Marin, P., Od
´
en, B. and Bjorntorp, P. (1995) Assimilation and mobilisation of triglyc-
erides in subcutaneous abdominal and femoral adipose tissue in-vivo in men. J Clin
Endocrinol Metab 80, 239–243.
180. Xu, X., De Pergola, G. and Bjorntorp, P. (1991) Testosterone increases lipolysis and the
number of beta-adrenoceptors in the rat male adipocyte. Endocrinology 128, 379–382.
181. Xu, X., De Pergola, G., Eriksson, P. S., Fu, L., Carlsson, B., Yang, S., Eden, S. and
Bjorntorp, P. (1993) Postreceptor events involved in the up-regulation of beta-adrenergic
receptor mediated lipolysis by testosterone in rat white adipocytes. Endocrinology 132,
1651–1657.
182. Anderson, L. A., McTernan, P. G., Harte, A. L., Barnett, A. H. and Kumar, S. (2002)
Androgen mediated regulation for altering lipolysis and lipogenesis by dihydrotestos-
terone in human adipose tissue. Diabetes Obesity Metab 4 (3), 209–214.
183. Lovejoy, J. C., Bray, G. A., Bourgeois, M. O., Macchiavelli, R., Rood, J. C., Greeson,
C. and Partington, C. (1996) Exogenous androgens influence body composition and
regional fat distribution in obese postmenopausal women – a clinical research centre
study. J Clin Endocrinol Metab 82, 2198–2203.

9
PPARγ and Glucose
Homeostasis
Robert K. Semple and Stephen O’Rahilly
It has long been known that various xenobiotic compounds, when adminis-
tered to mice, give rise to exuberant proliferation of hepatic peroxisomes, and
ultimately to tumour development. In 1990 the mediator of this response was
cloned and identified as a nuclear hormone receptor subsequently called per-
oxisome proliferator-activated receptor (PPAR).
1
When two homologues were
later cloned in Xenopus
2
and then in all mammalian species studied, the three
receptors were designated PPARα,PPARγ and PPARδ. Independently of these
developments, large scale chemical screening in the 1980s identified thiazo-
lidinediones as potent agents for lowering blood glucose and improving lipid
profiles in animal models of diabetes and obesity.
3
The convergence of these
two lines of investigation with the realization that the molecular target of the
thiazolidinediones was PPARγ
4
placed this receptor right at the centre of the
interplay between lipid and glucose metabolism. This occurred at a time in the
early 1990s when the ‘glucocentric’ view of type 2 diabetes as a disease princi-
pally of glucose metabolism (perhaps, in part, a historical accident)
5
was being
usurped by the resurgent appreciation that it is a complex metabolic disease in

which abnormal lipid and glucose homeostasis are intimately and inextricably
linked. In the decade since then, a wealth of experimental data has confirmed the
importance of PPARγ as a central regulator of the metabolic cross-talk between
insulin-sensitive tissues, and thiazolidinediones have proved beneficial therapeu-
tically as the first new class of insulin-sensitizing agents for several decades.
While PPARγ has afforded investigators a valuable handle on the intractable
pathophysiology of this most prevalent condition, many questions remain about
Insulin Resistance. Edited by Sudhesh Kumar and Stephen O’Rahilly
 2005 John Wiley & Sons, Ltd ISBN: 0-470-85008-6
238 PPARγ AND GLUCOSE HOMEOSTASIS
its biology. Here the progress of investigation to date and the outstanding issues
will briefly be reviewed.
9.1 Evidence from cell and rodent models
PPARγ binds to specific promoter response elements as a heterodimer with
the retinoic acid receptor (RXR). In the presence of ligand it recruits co-
activator molecules, which target chromatin-decondensing complexes to the
promoter region and render it accessible for the initiation of transcription. Con-
versely, in the presence of an antagonist, and perhaps in the unliganded state,
PPARγ recruits co-repressor molecules, which lead to the condensation of chro-
matin and sequestration of promoter elements. In addition, there is an evolving
appreciation that PPARγ may influence gene expression indirectly, and usu-
ally negatively, through competition with other transcription factors for such
accessory molecules. Although thiazolidinediones have been identified as potent
synthetic ligands of PPARγ, it is not clear whether any physiologically relevant,
potent endogenous ligands exist. The most widely studied candidate has been 15
deoxy-
12,14
-prostaglandin J2, identified as a potent activator in in vitro studies,
but a more recent reanalysis has suggested not only that in vivo concentrations
are too low for it to be a relevant ligand, but also that levels fail to correlate

with PPARγ activity.
6
Furthermore, a large number of unsaturated fatty acids,
eicosanoids and prostaglandins have also been shown in vitro to activate the
receptor. The binding affinities of these agents tend to be rather low, leading
to the suggestion that, instead of conforming to the paradigm of receptors with
single, very high affinity ligands, PPARγ functions as a more generic sensor
of fatty acid flux, a property which might help subserve a role as a nutritional
sensor and co-ordinator of metabolic responses. Further complexity is attested
to by the ability of RXR ligands, too, to stimulate the transactivational activity
of the PPARγ–RXR heterodimer, and the further modulation of this activity by
phosphorylation of PPARγ.
7
PPARγ is expressed at the highest levels in brown and white adipose tissue,
where around 30 per cent of its protein expression is accounted for by a splice
variant known as PPARγ
2
.
8, 9
This variant has an additional 28 N-terminal
amino acids, and appears to be specific to white adipose tissue. PPARγ is also
expressed at high levels in large intestine and white blood cells of both the
lymphoid and myeloid lineages, and at lower levels in kidney, liver, skeletal
and smooth muscles, pancreas and small intestine.
9–11
The relative importance
of PPARγ in each of these tissues from the point of view of glucose homeostasis
is incompletely understood, and will form the remainder of this discussion.
White adipose tissue
Although obesity is robustly associated with impaired insulin sensitivity, the

severe insulin resistance of both humans with lipodystrophy and of mice with
EVIDENCE FROM CELL AND RODENT MODELS 239
genetically ablated white adipose tissue bears witness to the importance of
normal amounts of this tissue in glucose homeostasis.
12
Murine models of
complete
13
or near complete
14
lipoatrophy exhibit ectopic fat accumulation
in liver and muscle with severe insulin resistance progressing to diabetes.
15
Importantly, transplantation of white adipose tissue into these mice dramatically
improves insulin sensitivity and related parameters,
16
demonstrating that it is the
absence of fat per se that leads to the abnormal metabolic phenotype. Human
subjects with lipodystrophy exhibit a similar pattern of severe insulin resistance
and dyslipidaemia, and are discussed in detail in Section 17.5.
PPARγ is known to play a pivotal role in preadipocyte differentiation in well
characterized in vitro models of adipogenesis, as detailed in Figure 9.1. Mouse
embryo-derived preadipocyte cell lines such as 3T3-L1 have been key tools in
establishing the transcriptional cascade of adipogenesis. Comparison of this data
C/EBPβ
C/EBPδ
PPARγ
C/EBPα
dexamethasone
cAMP

IGF-1
ERKs
P
Adipocyte genes
Proadipogenic factors
Anti-adipogenic factors
O/E-1, FGF10, FGF2, FBI
GATA-2,3, TGFβ, wnt10b
10% serum
Insulin
Dexamethasone
IBMX (raises cAMP)
10% serum
Insulin
10% serum
(a)
(b)
ERK
C/EBPα
PPARγ
C/EBPδ
C/EBPβ
aP2, Glut4 etc.
48 h
96 h
Figure 9.1 Role of PPARγ in adipogenesis in vitro
240 PPARγ AND GLUCOSE HOMEOSTASIS
with adipose phenotypes of genetically modified animals suggests that they do,
at least in part, model the in vivo situation, although it is also clear that many
more influences on in vivo adipogenesis remain to be discovered. Figure 9.1(a)

shows a typical pattern of expression of some of the key genes implicated in 3T3-
L1 differentiation, showing details of the artificial differentiation medium used.
Figure 9.1(b) shows a simplified model of the transcriptional cascade, showing
a complex series of kindling reactions leading to a robust mutually sustaining
expression of PPARγ and C/EBPα, which then drive the full programme of
adipocyte gene expression.
In view of the physiological importance of adipose tissue, the simplest inter-
pretation of the role of PPARγ in modulating insulin sensitivity is that the
beneficial effects of its activation derive solely from its ability to promote
the expansion of adipose tissue. However, thiazolidinediones are not used in
clinical practice principally as a means of inducing adipogenesis in lipodys-
trophic subjects, but rather are used effectively in patients of normal or more
commonly increased adiposity to enhance insulin sensitivity. Thus, apparently
paradoxically, a pro-adipogenic agent is used to treat a condition that is often
precipitated by the development of excessive adipose tissue. This paradox is at
least partly resolved by consideration of the complex biology of adipose tis-
sue in vivo, which cannot be replicated fully in vitro: far from the historical
perception of adipose tissue as a relatively inert reservoir for excess dietary
fat, it is now understood that white adipose tissue is a complex ‘organ’, which
plays a key role in orchestrating numerous metabolic processes. It is constantly
sensing the nutritional status of the whole organism, is in continuous commu-
nication with other tissues such as liver and muscle and is moreover spatially
heterogeneous, with fat depots at different anatomical sites exhibiting markedly
different patterns of gene expression, presumably reflecting distinct metabolic
functions. Thus, modifying the hypothesis by invoking depot-selective responses
of adipose tissue to PPARγ activation is necessary.
Support for the concept of such depot-selective PPARγ effects is provided
by pharmacological studies in mice: administration of potent and selective thi-
azolidinediones results in a preferential expansion of inguinal fat, analogous to
human subcutaneous adipose tissue, at the expense of retroperitoneal and other

depots.
17
Possibly because this remodelling favours the accretion of lipid in
depots that are less hormonally sensitive, and that do not have direct access
to the portal circulation and hence the liver, insulin sensitivity is enhanced.
However, the increased mass of inguinal fat pads is not simply due to accu-
mulation of more tissue of the same morphology: analysis of the distribution
of adipocyte size reveals that, while the total number of cells does increase,
these cells are of smaller size due to a combination of hyperplasia of precursor
cells and apoptosis of larger, hypertrophic adipocytes.
18–22
Correlational studies
in different genetic and dietary models of obesity have consistently revealed a
positive relationship between adipocyte size and insulin resistance,
23–28
and so
EVIDENCE FROM CELL AND RODENT MODELS 241
PPARγ is instrumental not only in modulating the amount and distribution of
adipose tissue, but also in regulating the function of that mature tissue.
A further experimental approach to the question of PPARγ and glucose sen-
sitivity has been to manipulate mice genetically in order to look at the effects of
altering PPARγ expression. Attempts to generate homozygous knockout animals
foundered due to the embryonic lethality of the deficiency,
22
but study of het-
erozygous knockout mice has been instructive, and has revealed some surprising
results. Two groups have determined independently that PPARγ heterozygote
knockout mice are more insulin sensitive than their wild type counterparts at
baseline,
22, 29

but only one of these groups found these animals to be protected
from high-fat-induced insulin resistance.
22
Further analysis of the mechanism
underlying this showed that, as in thiazolidinedione-treated wild type animals,
the mean size of the adipocytes decreased, though in this case they also declined
in number, so that body weight and fat mass of the heterozygotes was reduced.
However, when these heterozygous knockout mice were treated with antagonists
of PPARγ and/or RXR they did indeed become insulin resistant,
30
consistent
with data from humans harbouring rare loss-of-function mutations in PPARγ,
discussed later. The other group generating PPARγ heterozygote knockout mice
found no difference in adipocyte hypertrophy and insulin resistance between het-
erozygous knockout and wild type animals, but did find the heterozygous animals
to be relatively protected from the age-related decline in insulin sensitivity.
31
A second, and complementary, genetic approach involved generation of mice
with homozygous PPARγ alleles which have a point mutation preventing serine
phosphorylation at position 112.
32
Phosphorylation at this site has been shown
in vitro to reduce PPARγ transactivational activity, and so loss of the poten-
tial for phosphorylation would be expected to result in a more active PPARγ,
at least intermittently. The homozygous mice had no more adipose tissue than
wild type counterparts, and were protected from high-fat-diet-induced insulin
resistance and adipocyte hypertrophy.
Thus the relationship between the level of PPARγ activity and insulin sensi-
tivity is more complex than first imagined, with either stimulation or a moderate
reduction in its action apparently leading to metabolic benefits. These metabolic

benefits are lost when PPARγ activity drops below a certain critical thresh-
old. It appears that the unifying feature of the two situations is a change in
adipocyte morphology, such that the cells are predominantly smaller and less
lipid laden. The possible functional connections between these adipocyte mor-
phological changes and enhanced insulin sensitivity may broadly be classified
into three groups: first, PPARγ may influence glucose tolerance through direct
effects on the insulin sensitivity of the adipocytes, thus augmenting the rate of
glucose disposal in adipose tissue. Second, the trapping of fatty acids in adipose
tissue in the fed state may be rendered more efficient, and finally the change in
adipocyte phenotype may result in an altered profile of secretory factors, which
have remote effects on other insulin-sensitive tissues.
242 PPARγ AND GLUCOSE HOMEOSTASIS
Direct effects on adipose tissue insulin sensitivity
To the extent that preadipocyte differentiation involves the expression of many
genes that confer insulin sensitivity upon adipose tissue, it is to be expected
that thiazolidinediones, via PPARγ, will enhance the insulin-stimulated glu-
cose uptake of adipose tissue. However, besides effects on the total number of
adipocytes and on the population distribution of differentiated and undifferen-
tiated cells, there is also evidence that enhanced glucose uptake in response to
insulin results directly from PPARγ activation in mature adipocytes. Thus, in
the 3T3-L1 murine embryo fibroblast model of adipogenesis, it has been shown
that in cells differentiated for 15 days, and regarded as equivalent to mature
adipocytes, exposure to rosiglitazone markedly induces expression of IRS-2
33
and Glut4.
34
Conversely, a decline in the expression of these genes and a reduc-
tion in cell size and triglyceride content are seen in the presence of a dominant
negative PPARγ.
35

The induction of Glut4 expression and glucose transport in response to insulin
is known to depend in part on activation of phosphatidyl inositol-3-kinase down-
stream from IRS1 and 2. However, a second pathway, involving interaction of
the tyrosine kinase cCbl with the insulin receptor, is also thought to be involved.
This, too, has been implicated directly in the PPARγ-mediated sensitization of
adipocytes to insulin: cCbl interacts with the insulin receptor only via the adap-
tor protein CAP, or cCbl-associated protein, and expression of CAP appears to
be rate limiting for the recruitment of cCbl to the insulin receptor.
36–38
Thia-
zolidinediones have been shown to upregulate CAP expression,
39
and promoter
analysis of the CAP gene has confirmed the presence therein of a functional
PPARγ-binding response element.
40
In the presence of CAP, cCbl is phosphory-
lated by the receptor, and the CAP–cCbl complex then translocates to specific
lipid membrane rafts enriched in caveolin.
38
The presence of the CAP–cCbl
complex in these rafts further recruits the CrkII–C3G complex to the rafts,
which phosphorylates the small GTP-binding protein TC10. This step has been
shown to be necessary for normal translocation of Glut4-containing vesicles to
the cell surface.
36
Thus PPARγ appears to enhance signalling through the two characterized
arms of the signalling network that links insulin binding to Glut4 translocation
and increased glucose uptake in adipose tissue. However, as glucose uptake
into adipose tissue accounts for only a small proportion of whole body glucose

disposal, this is likely to account at most for only part of the enhanced insulin
sensitivity that results from PPARγ activation.
Effects on dietary lipid handling
The central metabolic role played by adipose tissue is in the storage of excess
caloric intake as triglycerides in the postprandial state, the controlled release
of this stored energy under fasting conditions and above all the tight coupling
EVIDENCE FROM CELL AND RODENT MODELS 243
of these processes to the prevailing nutritional conditions. Consonant with this,
expression of PPARγ does appear to be entrained to nutritional status, with sig-
nificantly lower levels of mRNA found in adipose tissue in the fasting state and
in experimental states of insulin deficiency, and higher levels seen after high
fat feeding.
41
Furthermore, it is likely that, if the relevant endogenous ligands
are indeed a constellation of polyunsaturated fatty acids and their derivatives,
the post-prandial state correlates with a period of high PPARγ activity. Hence it
is likely to be of great functional significance that the transcriptional response
elaborated by activated PPARγ encompasses a range of genes that play key
roles in lipid uptake and trapping. Thus, lipoprotein lipase,
42
which acts at the
cell surface to hydrolyse chylomicron triglycerides, CD36 and FATP,
43
which
mediate uptake of the resulting free fatty acids into the adipocyte, fatty acid
binding protein
8
and acyl-CoA synthase
44
are upregulated by PPARγ, while

genes that induce lipolysis and release of fatty acids, such as the β
3
-adrenergic
receptor
45
and leptin,
46, 47
are repressed. In order for free fatty acids to be
stored in the adipocyte as triglyceride, glycerol is also required. This may be
provided by uptake of circulating glycerol, mediated by the adipocyte homo-
logue of aquaporin (also upregulated by PPARγ)
48
or by glyceroneogenesis.
The rate-limiting step in this last process is catalysed by phosphoenolpyruvate
carboxykinase, which, too, is upregulated by PPARγ.
49–52
Finally, activation
of glycerol prior to esterification with fatty acids is undertaken by glycerol
kinase. Until recently, the dogma has been that this enzyme is not present in
adipocytes, preventing a futile cycle between triglyceride hydrolysis and resyn-
thesis being established. However, one report has now shown not only that this
enzyme is present, but also that it is strongly upregulated by PPARγ activation
in both mouse and human adipocytes,
53
although the human findings have sub-
sequently been strongly countered by negative findings from a second group.
54
Thus the sum of these actions of PPARγ strongly favours free fatty acid trapping
in adipose tissue in situations where ligand and receptor are plentiful. This is
schematized in Figure 9.2. As indicated by the asterisks and bold lines, PPARγ

upregulates the transcription of almost all stages of fatty acid trapping, from
fatty acid release from lipoproteins and uptake into adipocytes, to esterifica-
tion to glycerol. The futile glycerol cycle recently proposed to be stimulated by
PPARγ is illustrated by the open circular arrows.
These actions of PPARγ form the basis of the ‘lipid steal hypothesis’, which
provides at present the best supported and most widely held explanation for the
insulin-sensitizing action of thiazolidinediones. According to this hypothesis,
PPARγ, by ensuring that almost all circulating free fatty acid is trapped effi-
ciently in adipose tissue, prevents the exposure of other insulin-sensitive tissues
such as liver and skeletal muscle to these molecules. It is well established that
there is a strong correlation between ectopic accumulation of lipid at these sites
(particularly in the form of intracellular fatty acyl-CoA)
55
and insulin resistance,
and so activation of PPARγ enhances insulin sensitivity and glucose disposal.
244 PPARγ AND GLUCOSE HOMEOSTASIS
aquaporin*
CD36*
TG-rich LP
glycerol
TG
FFA
DHAP
ACS*
acylCoA
pyruvate
PEPCK*
GK*
FFA
FFA

FFA
glycerol
glycerol
glycerol
FFA
glycerol
GK*
HSL
HSL
ADIPOCYTE
CAPILLARY
ACS = acylCoA synthetase
GK = glycerol kinase
PEPCK = phosphoenolpyruvate carboxykinase
LPL = lipoprotein lipase
HSL = hormone-sensitive lipase
FFA = free fatty acids
TG = triglyceride
DHAP = dihydroxyacetone phosphate
*upregulated by PPARγ
FATP*
Figure 9.2 Actions of PPARγ that promote fatty acid trapping in adipocytes
EVIDENCE FROM CELL AND RODENT MODELS 245
In type 2 diabetes, the habitually tight coupling between fatty acid trapping in
adipose tissue and nutritional state is dysregulated, rendering lipid metabolism
relatively inflexible in the face of fluctuating nutritional states. Because of this,
mean circulating levels of free fatty acids are high. Pharmacological activation
of PPARγ in this context allows these abnormally high fatty acids to be safely
sequestered in adipocytes, ‘stealing’ them from the other insulin-sensitive tissues
such as skeletal muscle. Indeed, the level of improvement of insulin sensitiv-

ity upon PPARγ activation seems to be tightly associated with a diminution in
lipid accumulation in skeletal muscle.
56
More direct support for this model of
PPARγ function comes from a report describing fatty acid kinetics in rodents
treated with PPARγ agonists: this found that thiazolidinediones increase insulin-
stimulated free fatty acid clearance, and also the rate of fasting free fatty acid
appearance.
57
These findings illustrate that determinations of fasting free fatty
acid levels alone are likely to be only crude indicators of subtly dysregulated
coupling of free fatty acid flux to nutritional status, and perhaps explain some
of the inconsistent findings in human studies.
Effects on adipocytokines
Fatty acids, as metabolic substrates themselves, provide an appealing link between
adipose tissue and other insulin-sensitive tissues, and provide perhaps the simplest
explanation of thiazolidinedione action. However, in the last decade it has become
apparent that adipose tissue also has the capacity to elaborate a wide variety of
small molecules with autocrine, paracrine or endocrine activity, and many of these
molecules are subject to regulation by PPARγ. The term ‘adipocytokine’ has been
coined for some of these molecules, and they have been grouped into those that
enhance insulin sensitivity (such as leptin and adiponectin) and those that blunt
insulin sensitivity (such as TNFα, resistin, and nitric oxide). Each of these has
potential to account for some of the beneficial effects of PPARγ.
The prototypic adipocyte-derived hormone is leptin, cloned in 1994.
58
This
cytokine-like peptide hormone is secreted in proportion to total body fat mass,
and is best characterized as a centrally acting suppressor of appetite and food
intake. Acting principally through the autonomic nervous system, it also induces

increased energy expenditure and oxidation of lipid in various tissues.
59
There is
in addition some evidence in rodents that it has direct local actions on skeletal
muscle and liver to enhance fatty acid catabolism.
60, 61
There are also lep-
tin receptors on adipocytes themselves, and on pancreatic β-cells. Evidence
for the importance of leptin in glucose homeostasis is most compelling in fat-
less or lipodystrophic mice, where insulin sensitivity is significantly enhanced
either by infusion of leptin
62
or by its transgenic overexpression,
63
although the
degree of metabolic improvement is contingent upon the particular fatless model
used, genetic background and details of the leptin regimen. Further to these
observations, it has been demonstrated in A-ZIP/F1 fatless mice that the marked
246 PPARγ AND GLUCOSE HOMEOSTASIS
metabolic benefit seen with transplantation of adipose tissue is not observed with
white adipose tissue from ob/ob mice, which does not secrete leptin.
64
Thus, at
least in the context of this model of complete lipodystrophy, leptin seems to be
pre-eminent as the mediator of the beneficial effects of fat. Disappointingly, it
appears that the dramatically beneficial effects of leptin may only be relevant to
complete lipodystrophy, or genetic deficiency of the hormone, as the vast major-
ity of models of obesity feature high leptin levels, often posited as evidence of
‘leptin resistance’. Such differences between the situation of dramatically low
levels of leptin and that of normal or increased fat mass and high leptin perhaps

accounts for the observation that PPARγ, though increasing insulin sensitivity,
suppresses expression of leptin. This is apparent in the decreased circulating
leptin in mice treated with thiazolidinediones, and the higher levels of leptin
seen in heterozygous PPARγ knockout animals.
21
Furthermore, the oxidative
response to administration of leptin appears to be enhanced in mice with only
one functional PPARγ allele.
21
Thus, although leptin appears to be permissive
for normal glucose homeostasis, its beneficial metabolic effects are not propor-
tional to its concentration over the higher part of its range, and in cases of
insulin resistance in animals that have normal or increased adipose stores other
factors override it in determining insulin sensitivity. PPARγ activation, while
decreasing leptin levels, enhances insulin sensitivity.
Unlike leptin, adiponectin, an adipocyte-derived multimeric hormone with
homology to complement factor 1q, circulates at levels that are inversely related
to the amount of white adipose tissue. Also unlike leptin, circulating levels of
adiponectin have been shown to correlate with insulin sensitivity in both genetic
and dietary models of murine insulin resistance and obesity,
65, 66
while infusion
of adiponectin markedly improves hepatic insulin sensitivity.
67
Furthermore, it
has been shown that in a murine model of lipoatrophy (this time heterozygous
PPARγ knockout mice treated with an RXR antagonist) infusion of adiponectin
alone substantially improved insulin sensitivity, while co-administration of leptin
normalized it.
65

Importantly, these salutary effects of adiponectin were also seen
in two genetic models of obesity-related insulin resistance.
68, 69
Further analy-
sis suggested that these effects of adiponectin were mediated by an enhanced
capacity to oxidize fatty acids in muscle and liver, possibly via AMP-activated
protein kinase,
70
with resulting depletion of the ectopic triglyceride accumulated
at these sites. Interestingly, this protein kinase has also been suggested recently
to mediate the insulin-sensitizing action of metformin.
71
In conjunction with the
observation that PPARγ activation increases adiponectin expression and produc-
tion both in vitro and in vivo, these findings render adiponectin a more appealing
candidate than leptin as a mediator of some of PPARγ’s insulin-sensitizing
action. However, the picture has been clouded by the contradictory findings
from different workers who have produced adiponectin knockout mice: while
one report suggested this resulted in moderate insulin resistance without body
weight change, the other failed to find any alterations in insulin sensitivity.
72, 73
EVIDENCE FROM CELL AND RODENT MODELS 247
Also confusingly, the second report suggested increased levels of β-oxidation in
muscle and liver of knockout animals.
73
One unifying feature of the two reports,
however, was the observation of a marked reduction in neointimal proliferation
and vascular stenosis.
72–74
PPARγ also influences the expression of further secreted factors which may

reduce insulin sensitivity: tumour-necrosis factor α (TNFα) is an adipocyte-
derived signalling molecule that reduces insulin-stimulated glucose uptake
75
and
is found in high concentrations in obese and insulin-resistant individuals.
76, 77
TNFα expression is inhibited by PPARγ activation in adipocytes,
78
which could
be relevant to the observed improvement of glycaemic control on PPARγ acti-
vation. Most support for this idea comes from studies in mice lacking TNFα.
Compared with wild type animals, these mice are mildly resistant to obesity
and insulin resistance induced by high fat feeding or gold thioglucose lesion-
ing of the hypothalamus.
79, 80
Studies of mice in which both isoforms of the
TNFα receptor have been knocked out are less clearcut, however. Although
the double-knockout mice were protected from insulin resistance and hypergly-
caemia when put on a ob/ob background, the same mice, exposed to high fat
diets, were equally or even more insulin resistant in comparison to wild type
animals.
79
Thus, further work remains to be done before the role of TNFα in
the insulin-sensitizing action of PPARγ can be defined with confidence.
Another gene downregulated by PPARγ encodes the secreted dimeric protein
resistin, first identified in 3T3-L1 adipocytes by virtue of this downregulation.
82
Plasma resistin levels are approximately in proportion to adipose tissue mass,
and in some reports correlate with insulin resistance in both dietary and genetic
models of obesity.

82
Infusion of resistin has also been reported to induce marked
hepatic insulin resistance in mice, but had no effect on the insulin sensitivity of
other peripheral tissues in the same report.
83
However, the relevance of resistin
as a mediator of PPARγ activity remains to be clarified, as other reports have
found that PPARγ activators stimulate rather than decrease resistin expression
in vivo in several different models of obesity and insulin resistance.
84
A further possible link between PPARγ action in adipose tissue and insulin
sensitization lies in its effects on nitric oxide (NO) production: in diet-induced
obesity and insulin resistance it is known that NO is overproduced in adipose
tissue and muscles by inducible nitric oxide synthase (iNOS),
85
and NO has
been shown to impair insulin-stimulated glucose uptake in L6 myotubes and
isolated skeletal muscle.
86
It also exerts effects on skeletal muscle triglyceride
metabolism through an action on lipoprotein lipase activity.
87
Although it has
yet to be demonstrated that PPARγ can repress iNOS expression in adipocytes,
this has already been shown in other cell types, and so NO may potentially
play a role in the improvement of glucose homeostasis resulting from PPARγ
activation.
Thus, tantalizing clues from these studies of adipocytokines suggest that they
may well be important mediators of the action of PPARγ on insulin sensitivity,
248 PPARγ AND GLUCOSE HOMEOSTASIS

but the many confusing and contradictory findings from different experimental
protocols and animal models mean that consensus has yet to be reached.
Skeletal muscle
Although investigation of the role of PPARγ in insulin sensitization has focused
on adipose tissue, it is also expressed in other insulin-sensitive tissues. Indeed,
although the level of PPARγ protein is relatively low in skeletal muscle, the total
mass of muscle and its importance as the site of most insulin-mediated glucose
disposal mean that physiologically relevant effects of PPARγ in muscle cannot
be excluded. Discrimination of the relative importance of PPARγ in adipose
tissue and elsewhere has been attempted by administration of thiazolidinediones
to lipoatrophic mice, but once again results have been inconsistent between dif-
ferent models: while rosiglitazone and troglitazone failed to have any effect on
glucose or insulin levels in A-ZIP/F1 fatless mice (although circulating triglyc-
erides were lowered),
88
troglitazone greatly improved insulin and glucose levels
in mice that had had 90 per cent of their adipose tissue ablated by coupling of
the diphtheria toxin gene to the fat-specific aP2 promoter.
20
It is possible that
even the tiny residual amount of adipose tissue in the second model may have
permitted a beneficial effect of troglitazone to have been expressed through its
effects on those remaining adipocytes.
Direct investigation of the part played by PPARγ in skeletal muscle is com-
plicated by the potential for phenomena that are secondary to improved sys-
temic insulin sensitivity during administration of PPARγ agonists. This prob-
lem has been circumvented experimentally in two ways: first, isolated cells
in culture have been examined, and positive effects of thiazolidinediones on
insulin-stimulated glucose uptake have been reported in both rat-derived L6
myotubes and in cultured human skeletal muscle cells, mediated by enhance-

ment of insulin-stimulated PI3-kinase activity and translocation of GLUT4.
89–94
Second, the muscle-specific deletion of PPARγ has recently been reported, and
has permitted experimentation in a whole animal setting.
95
These animals exhibit
modest whole body insulin resistance, but surprisingly the glucose disposal into
muscle is normal, with the effect attributable instead to hepatic and perhaps
adipose insulin resistance. Moreover, the knockout animals accumulate adipose
tissue at a faster rate than wild type controls despite reduced food consumption.
This evidence of metabolic cross-talk between different insulin-sensitive tissues
is a recurring theme of different tissue-specific genetic manipulations, including
liver-specific deletion of PPARγ
96
(see below) and adipose-specific deletion of
Glut4, but the mechanisms are at best only partly understood. In this case no
excess of intracellular lipid was seen in the livers of the knockout animals, and
no increase in whole animal fatty acid oxidation was observed. However, when
knockout animals were treated with thiazolidinediones, their response was as
good as that of wild type controls, reinforcing the view that, while PPARγ may
EVIDENCE FROM CELL AND RODENT MODELS 249
have complex roles in normal physiology, the principal site of the therapeutic
action of potent agonists is adipose tissue. This genetic model failed to pro-
vide support for the previous suggestion, based on gene expression profiling in
PPARγ agonist-treated Zucker fatty rats, that pyruvate dehydrogenase kinase 4
(PDK4) may be a physiologically relevant target of PPARγ in muscle. PDK4
phosphorylates and inhibits pyruvate dehydrogenase, downregulating glucose
oxidation and reciprocally promoting fatty acid oxidation, and was found to be
downregulated eightfold by PPARγ activation.
97

This may have been an example
of indirect effects on muscle mediated by PPARγ activity elsewhere.
Pancreatic β-cells
Pancreatic β-cells have been demonstrated to express PPARγ,
98
and there is
some evidence that its activation can both enhance insulin secretion and protect
β-cells from the apoptosis thought to be triggered by excessive accumulation
of triglyceride in the metabolic milieu of insulin resistance.
98, 99
Although acti-
vation of PPARγ does not acutely improve insulin secretion in isolated human
pancreatic islets, treatment of insulin-resistant animals with thiazolidinediones
has been shown to increase fatty acid β-oxidation, thus blunting the accu-
mulation of intracellular triglyceride in pancreatic β-cells, and delaying β-cell
failure.
98, 99
Similar prolongation of β-cell survival has been seen on adminis-
tration of troglitazone to mice with streptozotocin-induced type I diabetes.
100
Levels of intracellular triglyceride may not only affect cell survival, but are
also likely to interfere with insulin secretion. A second, more direct mode of
enhancement of insulin secretion has also been suggested, based on the finding of
a functional PPARγ response element in the GLUT2 promoter region: PPARγ-
mediated stimulation of GLUT2 expression would increase glucose uptake and
hence lead ultimately to insulin release.
101
Thus PPARγ appears not only to
be an important player in systemic insulin resistance, but also to play a role
in the β-cell response to that increased demand, the other key process in the

pathogenesis of type 2 diabetes.
Liver
The data relating to effects of chronic thiazolidinedione administration on hep-
atic glucose output in humans is conflicting,
102 – 104
and together with the low
levels of expression of PPARγ in the liver (around 10–30 per cent of levels
in adipose tissue) they suggest that the liver is not a physiologically important
site of action of PPARγ. However, in a range of different rodent models of
diabetes and insulin resistance, encompassing both lipoatrophy and hyperphagic
obesity, hepatic expression of PPARγ is markedly elevated.
41, 20, 105 –108
All
these models feature hepatic steatosis, and patterns of gene expression suggest
that the upregulation of PPARγ may contribute to this. Proof that PPARγ may
250 PPARγ AND GLUCOSE HOMEOSTASIS
indeed stimulate lipid accumulation in hepatocytes in vivo has recently been
established using adenovirally mediated hepatic overexpression of PPARγ in
mice.
109
This resulted in hepatic steatosis accompanied by the upregulation in
expression of a wide range of known PPARγ-responsive genes. However, in
fatty liver induced by fasting or choline deficiency in PPARα knockout mice no
evidence of PPARγ upregulation could be found,
109
suggesting that not all hep-
atic steatosis is the same, and further that fat accumulation per se is insufficient
to induce PPARγ expression. The clearest evidence that PPARγ, at least in some
circumstances, has a significant role in the liver comes from two complemen-
tary approaches. First, detailed analysis of tissue-specific insulin sensitivity in

A-ZIP/F-1 fatless mice treated with rosiglitazone demonstrated improvement of
muscle insulin sensitivity at the expense of increased intracellular triglyceride
and reduced insulin sensitivity in the liver,
110
and second, two groups have
reported liver-specific genetic ablation of PPARγ, in the context of wild type
animals, of the A-ZIP/F1 fatless model, and ob/ob genetic obesity models. Com-
pared to ob/ob animals with intact PPARγ, those with no hepatic PPARγ had
decreased triglyceride accumulation in the liver, with concomitantly enhanced
hepatic insulin sensitivity. However circulating levels of free fatty acids and
triglycerides were increased, and insulin sensitivity in muscle and adipose tissue
was further impaired. Despite the absence of PPARγ in the liver, rosiglitazone
still resulted in marked improvement in these indices, further supporting the view
that the principal site of its action is elsewhere.
111
Correspondingly, in A-ZIP/F-
1 mice, loss of hepatic PPARγ also led to increased circulating triglyceride and
worsened muscle insulin sensitivity, but this time the effect of rosiglitazone on
whole body insulin sensitivity was abolished.
96
Interestingly, even on a wild
type background, deletion of liver PPARγ led to increased adiposity, hyperlip-
idaemia and insulin resistance compared with controls in the face of a high
fat diet.
Thus the evolving picture is that, while hepatic PPARγ may have little signif-
icance in lean animals, in those with obesity and insulin resistance it undergoes
compensatory upregulation to accommodate excess lipid in the liver. Further-
more, considering all the studies of tissue-specific ablation of PPARγ suggests
that there is a hierarchy of triglyceride storage: by far the largest site of storage,
and probably the only one of relevance in lean animals, is the white adipose

tissue. However, when presented with chronic caloric excess, the liver also
has a significant capacity for fat storage. Only when both these depots are
overwhelmed does skeletal muscle experience the adverse effects of lipid accu-
mulation and insulin resistance. In this regard it is worth reflecting that the diet
of large parts of Western industrialized society most closely resembles the types
of high fat diet used in rodent studies.
A summary of the possible roles of PPARγ in glucose homeostasis is shown in
Figure 9.3. As well reducing adipocyte size, improving the profile of secreted
adipokines, and enhancing lipid trapping in adipose tissue, PPARγ may also
INSIGHTS FROM HUMAN STUDIES 251
FFAs
FFAs
Resistin
TNFα
Leptin
NO
Adiponectin
Resistin
TNFα
Leptin
NO
Adiponectin
HIGH
PPARγ
activity
LOW
PPARγ
activity
pancreas
muscle

liver
Figure 9.3 Potential mechanisms of the insulin-sensitizing effect of PPARγ activation
permit excess triglyceride to be stored in the liver, giving further buffering
capacity before skeletal muscle and β-cells are affected by accumulating intra-
cellular lipid. However, loss of PPARγ from any one of these tissues is likely
to have deleterious consequences for the insulin sensitivity (or secretion) of
the others.
9.2 Insights from human studies
Informing and motivating these rodent and cellular studies has been the estab-
lished utility of potent and selective PPARγ agonists in the treatment of insulin
resistance and type 2 diabetes in humans. As ever, caution must be exercised in
extrapolating the results in these model systems to human pathophysiology, but
the evolving evidence suggests that this is, in large part, appropriate. The prin-
cipal sources of in vivo human data are clinical trials of potent PPARγ agonists
and studies of subjects harbouring naturally occurring PPARγ variants. These
will be considered in turn.
Pharmacological studies
Effects on insulin sensitivity
Three potent and selective PPARγ agonists have been used in large scale
clinical practice. The prototype, troglitazone, was unfortunately withdrawn by
252 PPARγ AND GLUCOSE HOMEOSTASIS
the manufacturer due to the occurrence in a small proportion of patients of
serious, and sometimes fatal, hepatotoxicity, but subsequently both rosiglitazone
and pioglitazone have been used with no evidence of similar problems.
Extensive clinical trial data has accumulated for all three agents. Results
have been consistent: used as monotherapy in patients with type 2 diabetes,
troglitazone,
112 – 114
rosiglitazone
115, 116

and pioglitazone
117
all reduce fasting
and postprandial plasma glucose by around 2 mmol/l, and glycosylated
haemoglobin A1 by between 1 and 1.5 per cent. When combined with either a
sulfonylurea,
117 – 120
metformin
104, 121
or subcutaneous insulin,
122, 123
similarly
beneficial results are seen. When analysed in more detail by hyperinsulinaemic
clamp studies, significant improvements in whole body insulin sensitivity have
been seen with all three agents,
104, 113, 114, 118, 117
accounted for mostly by
increased glucose disposal rates, although a minor suppression of hepatic glucose
output has sometimes been found.
102, 103
Interestingly, one report that examined
the effects of metformin and troglitazone in parallel suggested that their effects
were complementary, with metformin principally suppressing hepatic glucose
output, and troglitazone preferentially acting to increase the rate of glucose
disposal.
104
The response to thiazolidinediones is not confined to patients
with diabetes: when used in 18 obese patients with normal glucose tolerance,
troglitazone still reduced insulin levels in the fasting state and after glucose
challenge, concomitant with a significant increase in insulin sensitivity measured

directly by euglycaemic clamp.
124
Effects on adipose tissue
As in rodents, human PPARγ1 and γ2 are highly expressed in adipose tissue,
125
and exposure of cultured primary human preadipocytes to PPARγ agonists
induces their differentiation,
126
while overexpression of a potent dominant-
negative mutant PPARγ has been shown to block this process.
127
Treatment
with thiazolidinediones promotes weight gain in humans and several studies have
shown that the increase in body weight associated with thiazolidinedione treat-
ment is accounted for principally by accumulation of subcutaneous fat (reviewed
in reference
128
), whereas visceral adipose tissue volume is reduced or unchanged.
These observations are in keeping with ex vivo studies in which preadipocytes
isolated from subcutaneous abdominal adipose tissue differentiated more read-
ily in response to thiazolidinediones than cells from visceral depots of the same
subjects.
129, 126
It is not known whether TZD treatment increases subcutaneous
fat mass in all body regions equally, a question that is of interest in light of
the burgeoning evidence of important functional metabolic differences between
upper body (abdominal) and lower body (including femoro-gluteal) subcuta-
neous fat.
130
Beyond gross effects on white adipose tissue distribution, whether TZD treat-

ment in humans induces apoptosis of hypertrophic adipocytes and increased
INSIGHTS FROM HUMAN STUDIES 253
differentiation of preadipocytes, leading to alterations in average adipocyte size
as seen in rodents, remains unresolved. Other aspects of thiazolidinedione action
on adipose tissue appear to correspond rather variably with data from animal
models. Thus, thiazolidinediones have generally been reported to lower free fatty
acid (FFA) levels in clinical trials, consistent with the ‘lipid steal’ hypothesis of
insulin sensitization. It is likely that many of the mechanisms subserving this
are the same as those outlined earlier in mice, although whether the induction
of glycerol kinase expression and activity in human adipocytes is significant
appears doubtful.
54
As in rodents, alterations of the profile of adipocyte-secreted proteins
may play a role in the therapeutic actions of thiazolidinediones. Of those
adipocytokines discussed earlier, adiponectin appears to be the best candidate in
humans: plasma levels correlate with insulin sensitivity,
131 – 133
and are inversely
proportional to fat mass,
134, 135
and thiazolidinediones increase adiponectin
gene expression.
136, 137
Moreover, circulating adiponectin levels were found to
be dramatically lower in three individuals harbouring loss-of-function PPARγ
mutations when compared with healthy controls or subjects with non-PPARγ-
mediated severe insulin resistance, suggesting a direct correlation between
PPARγ activity and adiponectin expression.
138
Further studies should help to

determine the extent to which this contributes to the insulin-sensitizing effects
of PPARγ agonists.
It is not clear whether resistin is significantly expressed in mature human
adipocytes,
139 – 141
and levels were undetectable in subjects carrying dominant
negative PPARγ mutations.
139
Definitive data on the role of nitric oxide and
TNFα in PPARγ-mediated insulin sensitization in humans have also yet to
be presented.
Studies of human genetic variants
Rare mutations
Recently, three groups have independently identified loss-of-function mutations
in the LBD of human PPARγ.
142 – 144
Together these reports describe eight adult
subjects, all of whom exhibit a stereotyped form of partial lipodystrophy and
severe insulin resistance, with the insulin resistance being evident even in early
childhood in affected individuals.
145
Loss of subcutaneous fat from the limbs
and gluteal region, with relative preservation of both the subcutaneous and
visceral abdominal depots was uniformly reported, although some differences
were observed in facial adipose tissue, said either to be reduced, preserved and
increased in different kindreds carrying different point mutations. Although these
findings are broadly in keeping with the role of PPARγ as a key regulator of
adipogenesis, it is difficult to reconcile the pattern of selective partial lipodys-
trophy observed with current knowledge about adipose tissue PPARγ expression
and the adipogenic response to receptor agonists in humans.

254 PPARγ AND GLUCOSE HOMEOSTASIS
As both partial and generalized lipodystrophy have consistently been found to
be associated with insulin resistance in man as in rodents,
146
it is likely that the
dramatic diminution in peripheral limb and gluteal fat contributes to the severe
insulin resistance of these rare patients. Additionally, even the residual adipose
tissue depots in the individuals studied so far is metabolically inflexible, probably
exacerbating the exposure of skeletal muscle and liver to dysregulated fatty acid
fluxes.
145
The findings in mice with tissue-specific ablation of PPARγ expression
with or without lipodystrophy, discussed earlier, permit the speculation that the
presence of a dominant negative PPARγ species in the liver and skeletal muscle
of these subjects further denies them the compensatory FFA-buffering capacity
afforded in lipodystrophy by the liver in particular, and so abnormalities in
these tissues also may contribute significantly to the severe systemic insulin
resistance observed.
Although study of the metabolic cross-talk between insulin-sensitive tissues
in humans is hampered by the lack of tissue-specific genetic deletions, the recent
report of a kindred with a digenic pattern of severe insulin resistance may offer
a rare opportunity for this type of investigation:
147
in the kindred described, the
combination of a frameshift mutation in PPARγ (effectively a null allele) and
a premature stop mutation in the muscle-specific glycogen-targeting subunit of
protein phosphatase 1 (PPP1R3A) cosegregated with severe insulin resistance,
while either mutation alone appeared to have no effect on insulin sensitivity in
the small number of subjects described. As PPARγ is most highly expressed in
adipose tissue, while the PPP1R3A product is specific for cardiac and skeletal

muscle, further investigation of these subjects, and of analogous animal models,
may provide important insights into the factors mediating the metabolic dialogue
between insulin-sensitive tissues, which has been a prominent but ill understood
feature of many of the genetic models already described. This kindred may also
be a first step away from rare monogenic insulin resistance towards the oligo- or
polygenic patterns which account for most of the population burden of insulin
resistance and diabetes.
With loss-of-function mutations resulting in lipodystrophy, and PPARγ ago-
nists promoting adipogenesis, gain-of-function PPARγ mutations might be antici-
pated to increase body fat mass. Indeed, four morbid subjects have been described
who are heterozygous for a proline to glutamine substitution in the N-terminal
domain of PPARγ2.
148
This mutation is adjacent to a phosphorylation site thought
to mediate downregulation of PPARγ transcriptional activity.
7
The mutation inter-
feres with this phosphorylation, resulting in a receptor with constitutive tran-
scriptional function and enhanced adipogenic activity in vitro.
148
However, no
segregation studies were reported, and no more similar mutations were found in
a larger screen of morbidly obese subjects. Furthermore, the observed severely
obese phenotype is at odds with the recently reported mouse harbouring a mutation
at the phosphorylation site itself: even homozygous animals had no more tendency
to weight gain than wild type littermates, and moreover were protected from
INSIGHTS FROM HUMAN STUDIES 255
diet-induced insulin resistance.
32
Although three of the human subjects were also

reported to have low insulin levels relative to their massive obesity, all three also
had type 2 diabetes, making inferences about their insulin sensitivity impossible.
Pro12Ala polymorphism
The receptor mutations described hitherto are rare, and although they have pro-
vided unique insights into the role of PPARγ in human glucose homeostasis, with
profound phenotypic effects in affected individuals, they make a negligible contri-
bution to the risk of insulin resistance or type 2 diabetes in the general population.
In contrast, by far the most prevalent human PPARγ genetic variant reported to
date is a polymorphism, replacing alanine for proline at codon 12 (Pro12Ala)
in the unique PPARγ2 amino-terminal domain, with an allelic frequency that
approaches 15 per cent in some Caucasian populations.
149, 150
Adipose tissue
mRNA levels of PPARγ2 are increased in morbidly obese individuals, whereas
expression of PPARγ1 is unchanged,
125
and isoform-specific knockout studies
suggest that PPARγ2 is the critical isoform mediating adipogenesis.
151
In some
functional assays, the Pro12Ala variant exhibits reduced binding to DNA and
modest impairment in transcriptional activation, and these functional properties
have been correlated with the association of this polymorphism with reduced
body mass index (BMI), although subsequent studies have failed to confirm this
finding.
150, 152
Evidence for an association between Pro12Ala and type 2 diabetes
was first reported in a Finnish population, in whom a lower BMI appeared to
correlate with improved insulin sensitivity in those carrying the Ala allele, while
in a group of second generation Japanese Americans the Pro/Pro genotype was

found to associate with type 2 diabetes.
152
However, this association was initially
poorly reproducible, with only one of five subsequent studies showing statisti-
cally significant linkage with diabetes risk.
153 – 157
Nevertheless, a meta-analysis
of published association studies confirmed a modest (1.25-fold) but significant
(p = 0.002) increase in diabetes risk with the Pro allele,
149
the discrepancy being
accounted for by the underpowering of many of the individual studies. Thus, if an
entire population carried the Ala allele, the global prevalence of type 2 diabetes
would be reduced by 25 per cent, making PPARγ potentially the most important
common ‘diabetogene’ thus far discovered. Further reports have strengthened
this association further,
158, 159
although the caveat that publication bias may have
favoured positive studies is a significant one.
If the association does stand the test of time, it raises the question of how
the Ala genetic variant influences diabetes risk. In the index study carriers of
the Ala polymorphism had a significantly lower BMI, and after correcting for
this there was no difference in insulin sensitivity between genotypes.
152
This,
in conjunction with the lower transcriptional activity of the Ala variant in vitro,
led to the hypothesis that improved insulin sensitivity might be accounted for
entirely by changes in adiposity. Although this would unify the observation that
256 PPARγ AND GLUCOSE HOMEOSTASIS
PPARγ activity

thiazolidinedione treatment
lean, healthy
• heterozygous knockout mice
• ?Pro12Ala
• dominant negative mutations
• heterozygous knockout mice +
PPARγ/RXR antagonist
Whole body insulin sensitivity
Figure 9.4 Integrated model of the relationship between PPARγ activity and insulin
sensitivity
PPARγ2 has a unique regulatory role in adipogenesis,
151
with the knowledge
that body fat mass is a strong determinant of insulin sensitivity, subsequent
studies have failed to yield consistent findings, with some even demonstrating
a modestly greater BMI in carriers of the Ala allele.
160 – 162
Germane to this are
likely to be lessons from rodent studies, where not only the amount of fat, but
also the size and function of individual adipocytes within it, is crucial to the
optimal physiological function of the adipose tissue. Such issues have yet to be
examined in human subjects. Furthermore, gene–environment interactions are
likely to be more complex in humans, as evidenced by a recent study indicating
that variations in dietary polyunsaturated fat versus saturated fat intake can
influence BMI in carriers of the Ala variant.
163
An attempt to simplify and integrate the relationship between PPARγ activity
and insulin sensitivity in humans, based on the above diverse observations, is
represented in Figure 9.4.
References

1. Issemann, I. and Green, S. (1990) Activation of a member of the steroid hormone
receptor superfamily by peroxisome proliferators. Nature 347, 645–650.
2. Dreyer, C., Krey, G., Keller, H., Givel, F., Helftenbein, G. and Wahli, W. (1992) Con-
trol of the peroxisomal beta-oxidation pathway by a novel family of nuclear hormone
receptors. Cell 68, 879–887.
3. Day, C. (1999) Thiazolidinediones: a new class of antidiabetic drugs. Diabet Med 16,
179–192.
4. Lehmann, J. M., Moore, L. B., Smith-Oliver, T. A., Wilkison, W. O., Willson, T. M.
and Kliewer, S. A. (1995) An antidiabetic thiazolidinedione is a high affinity ligand for