| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
FJ
EXPRESS SUMMARY ARTICLE The Full-length version of this article is also available, published online June 27, 2001 as doi:10.1096/fj.00-0750fje. |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Vascular Biology, The Scripps Research Institute, La Jolla, California 92037, USA
4Correspondence: The Scripps Research Institute, Department of Vascular Biology/VB-3, 10550 North Torrey Pines Rd., La Jolla, CA 92037, USA. E-mail: loskutof{at}scripps.edu
SPECIFIC AIM
Elevated levels of plasma and adipose tissue plasminogen activator inhibitor 1 (PAI-1) are frequently observed in obese humans and rodents and correlate strongly with visceral fat mass and the degree of insulinemia. In the present study, genetically obese and diabetic (ob/ob; C57BL/6J) mice lacking the PAI-1 gene (PAI-1-/-) were generated and used to test the hypothesis that elevated PAI-1 contributes to the hyperglycemia, hyperinsulinemia, and insulin resistance associated with the obese and diabetic phenotype.
PRINCIPAL FINDINGS
1. Lack of PAI-1 is associated with a significant reduction in
adiposity
In general, PAI-1-deficient ob/ob
(PAI-1-/- ob/ob) mice appeared to be smaller
than age-matched ob/ob mice containing the PAI-1 gene (i.e., WT for
PAI-1 or WT ob/ob). In fact, ob/ob mice lacking PAI-1 weighed 
25%
less than WT ob/ob mice whereas heterozygous
(PAI-1-/+) ob/ob mice were intermediate in size.
The lower weight of the PAI-1-/- ob/ob mice was
related to the rate of fat accumulation since the mean weight of the
epididymal fat pads from PAI-1-/- ob/ob mice
was significantly less than those from WT ob/ob mice.
2. Lack of PAI-1 improves the hyperglycemia associated with obesity
Nonfasting serum glucose levels were determined in the lean and
ob/ob mice. The mean serum glucose level of the WT ob/ob mice was
363 ± 44.6 mg/dl at the age of 2–4 months vs. 209 ± 14.8
mg/dl in age-matched lean controls. Unexpectedly, the serum glucose
levels in the PAI-1-/+ and
PAI-1-/- ob/ob mice were significantly lower.
In fact, PAI-1-/- ob/ob mice had normal mean
serum glucose levels at all ages tested. WT ob/ob mice also were
hyperglycemic in the fasting state (379±40.2 mg/dl), and again the
mean serum glucose levels of PAI-1-/- ob/ob
mice were considerably lower (274±4.0 mg/dl, P<0.05). Lean
mice of all PAI-1 genotypes responded to the administration of
intraperitoneal (i.p.) glucose with elevated serum glucose levels at 15
and 30 min; as expected, the effects were transient and serum glucose
concentrations decreased to the normal range by 60 min. In contrast,
ob/ob mice of all PAI-1 genotypes were unable to restore normal serum
glucose levels by 60 min after the exogenous glucose administration.
These results indicate that WT and PAI-1-/-
ob/ob mice have impaired glucose tolerance and that the absence of
PAI-1 does not restore normal glucose tolerance in ob/ob mice.
3. Lack of PAI-1 improves the hyperinsulinemia associated with
obesity
Hyperinsulinemia was a common finding in most of the WT ob/ob
mice. For example, in mice aged 2–4 months, nonfasting serum insulin
levels were substantially increased compared with age-matched lean
controls (mean values, 28.9±6.6 ng/ml vs. 2.0±0.6 ng/ml,
P<0.0001). Although serum insulin levels in the
PAI-1-/- ob/ob mice also were elevated over the
lean controls, their mean insulinemia (13.3±3.5 ng/ml) was
significantly lower than that of their WT ob/ob counterparts
(P<0.05). Mean serum insulin levels in the heterozygote
(PAI-1-/+) ob/ob mice were intermediate between
those of the WT and PAI-1-/- ob/ob mice.
Determination of fasting serum insulin levels immediately before
glucose administration again revealed that WT ob/ob mice had
significantly elevated insulin concentrations compared with lean
controls. Again, the fasting insulin levels of the
PAI-1-/- ob/ob mice were significantly lower
than those of WT ob/ob mice (P<0.05). Although i.p. glucose
administration resulted in a dramatic increase in serum insulin levels
in the WT ob/ob mice, the increase in the
PAI-1-/- ob/ob mice was relatively small. In
lean mice, serum insulin levels increased only slightly after the
glucose injection. Finally, both WT and
PAI-1-/- ob/ob mice were relatively resistant
to insulin as monitored by changes in serum glucose levels in response
to insulin administration. For example, serum glucose levels were
unchanged or even slightly higher in ob/ob mice of both PAI-1 genotypes
after insulin treatment. In contrast, serum glucose levels in the lean
mice decreased in response to the insulin administration, reaching
levels significantly lower than their ob/ob counterparts. Thus, the
insulin tolerance test confirmed the insulin-resistant state of WT and
PAI-1-/- ob/ob mice.
4. Lack of PAI-1 reduces TNF-
gene expression in the adipose
tissue of obese mice
Since overexpression of the tumor necrosis factor (TNF-)
gene in the adipose tissue is a common finding in obese humans and most
animal models of obesity, and since TNF- may be an important
mediator of insulin resistance, we studied the effect of deletion of
the PAI-1 gene on adipose tissue TNF- mRNA and protein levels. As
expected, a significant increase in TNF- mRNA was detected in the
adipose tissues of WT ob/ob mice, with a maximum increase (15-fold) at
2 to 4 months. The concentration of TNF- mRNA gradually decreased in
the older mice and approached control levels at 10–12 months of age.
Unexpectedly, TNF- mRNA levels in the adipose tissues of
PAI-1-/- ob/ob mice did not undergo this
dramatic increase at 2 to 4 months and were only slightly higher than
those of the lean controls at this age. Although there was a
significant increase in TNF- gene expression in
PAI-1-/- ob/ob mice at 8 months of age, mean
expression levels were still lower than those of WT ob/ob mice. In the
lean controls, TNF- gene expression levels were not different
between WT and PAI-1-/- mice at any age. In
situ hybridization experiments revealed a strong signal for TNF-
mRNA in cells that appeared to be adipocytes. Elevated TNF- mRNA was
detected in 20–25% of tissue sections from the WT ob/ob mice, but
not in tissue sections from PAI-1-/- ob/ob, WT
lean mice, or PAI-1-/- lean mice.
Experiments were performed to determine whether changes in adipose
tissue TNF- mRNA levels were reflected by changes in TNF- protein
levels. Adipose tissues explants obtained from 2-month-old WT ob/ob
mice secreted higher amounts of TNF- protein than those from lean
controls. Moreover, the level of TNF- protein secreted from adipose
tissue explants of PAI-1-/- ob/ob mice was
considerably lower than that from WT ob/ob mice. Immunohistochemical
analysis revealed increased amounts of TNF- antigen in adipose
tissue of WT ob/ob mice compared with the levels in adipose tissues of
WT lean, PAI-1-/- lean and
PAI-1-/- ob/ob mice. The strong immunosignal
was detected predominantly within adipocytes and the distribution of
TNF- antigen was similar to that of TNF- mRNA.
CONCLUSIONS AND SIGNIFICANCE
Obesity is a pathological condition frequently associated with
metabolic disturbances such as insulin resistance and
non-insulin-dependent diabetes mellitus. PAI-1, the primary
physiological inhibitor of fibrinolysis in vivo, is elevated in the
plasma and adipose tissue of obese humans and rodents, and increased
PAI-1 activity is associated with the increase in adiposity as well as
with hyperinsulinemia and development of the metabolic syndrome
(Fig. 1
). A variety of observations suggest that the increase in PAI-1 is also
due to the chronic elevation in TNF- associated with this condition.
Increased plasma PAI-1 levels may contribute to the impaired endogenous
fibrinolysis and the increased cardiovascular risk of patients with
visceral fat accumulation and/or non-insulin-dependent diabetes
mellitus. Despite this well-documented link between PAI-1 levels and
obesity, however, no information is available regarding the possibility
that the elevated PAI-1 itself may play a role in the development of
adiposity or of obesity-associated insulin resistance/hyperinsulinemia
(Fig. 1
, shaded arrows).
|
To study the contribution of PAI-1 to the obese phenotype, we generated
genetically obese mice that lacked the PAI-1 gene and thus were unable
to express the inhibitor. We then examined the effect of PAI-1 gene
deletion on body weight, hyperglycemia, hyperinsulinemia, insulin
resistance, and adipose tissue TNF- mRNA and protein expression. Our
results demonstrate that both wild-type and PAI-1-deficient ob/ob mice
develop substantial adiposity. However, PAI-1-deficient ob/ob mice were
smaller and weighed significantly less than WT ob/ob mice. Moreover,
the epididymal fat pads from PAI-1-/- ob/ob
mice were significantly smaller than those from the WT ob/ob mice.
Thus, deletion of the PAI-1 gene seems to decrease the rate of fat
accumulation in ob/ob mice. Unexpectedly, fasting and nonfasting serum
insulin levels also were significantly lower in
PAI-1-/- ob/ob mice than WT ob/ob mice and
serum glucose levels in these mice were as low as those of lean
controls. Finally, i.p. glucose tolerance tests revealed that
PAI-1-/- ob/ob mice responded to the exogenous
glucose load by elaborating substantially lower insulin levels than WT
ob/ob mice. These results suggest that PAI-1 deficiency also partially
protects against the development of glucose-induced hyperinsulinemia in
obesity. However, ob/ob mice of all PAI-1 genotypes did not respond to
the administration of insulin with a fall in serum glucose levels as
did the lean mice. These latter studies suggest that the
PAI-1-/- ob/ob mice remain insulin resistant.
More specific tests that dissect insulin resistance into defects in
insulin sensitivity or insulin responsiveness need to be performed
before the implication of these observations can be fully appreciated.
In any case, these results indicate that deletion of the PAI-1 gene
leads to a significant reduction in body weight and partially protects
genetically obese mice from obesity-associated hyperglycemia and
hyperinsulinemia.
Experiments were performed to investigate the mechanisms whereby
deletion of the PAI-1 gene from ob/ob mice partially restores normal
glucose homeostasis. The primary hypothesis was that TNF- was
involved since overexpression of TNF- in the adipose tissue is an
almost universal finding in both obese humans and mice (Fig. 1)
.
Moreover, adipocytes had earlier been identified as the main cellular
source of adipose tissue TNF- gene expression, and TNF- mRNA
levels in fat strongly correlate with the extent of hyperinsulinemia.
Recent reports suggest that TNF- also influences glucose and lipid
metabolism, as well as adipocyte differentiation and regulation of
adipose tissue gene expression. TNF- may contribute to the
insulin-resistant state as well. For example, TNF- was reported to
block the action of insulin through its ability to inhibit insulin
tyrosine kinase activity and through down-regulation of glucose
transporter molecules such as Glut 4 or the adipocyte fatty acid
binding proteins aP2. Moreover, studies in mice demonstrated that
glucose tolerance improved both after in vivo neutralization of TNF-
and in obese mice lacking the TNF- gene. Neutralization of TNF-
or deletion of both TNF- receptors in ob/ob mice also resulted in
lower plasma insulin and PAI-1 levels and led to a substantial
reduction of adipose tissue PAI-1 mRNA expression. Deletion of the
PAI-1 gene from ob/ob mice leads to a significant reduction in adipose
tissue TNF- gene expression, as demonstrated by RT-PCR and in situ
hybridization. These results thus support the hypothesis that the
effects of PAI-1 on body weight and on insulin and glucose metabolism
may be mediated, at least in part, by TNF- (Fig. 1)
.
There is no information available to explain how the absence of PAI-1
can lower TNF-. It is possible that PAI-1 inhibits an enzyme
necessary for the activity of TNF- or that PAI-1 alters TNF- by a
mechanism that is independent of its ability to function as a protease
inhibitor. PAI-1 was recently shown to inhibit cell adhesion and
migration by a mechanism that does not require its inhibitory activity.
PAI-1 also may alter other cytokines known to contribute to insulin
resistance, such as transforming growth factor ß, or alter
fat-specific proteins that play key roles in glucose metabolism in fat
such as the glucose transporter protein Glut4, adipsin, hexokinase-II,
or the fatty acid binding protein aP2. Whether the adipose tissue
expression of these or other genes is altered in PAI-1-deficient ob/ob
mice remains to be determined. Since food restriction in obese mice or
weight loss in humans leads to a significant decrease in adipose tissue
TNF- gene expression, the reduction of TNF- mRNA synthesis in fat
and the amelioration of hyperinsulinemia observed in the
PAI-1-/- ob/ob mice could (in part) be
secondary to the reduction of body weight and/or changes in feeding
behavior. However, the weight difference between WT and
PAI-1-/- ob/ob mice was at most only 25%, and
it seems unlikely that this difference could account for the complete
restoration of normal glucose levels, the more than 50% reduction in
serum insulin levels, and the absence of TNF- overexpression until
the age of 8 months.
We did not detect elevations in circulating TNF- in plasma from
ob/ob mice. These latter results are consistent with previous studies
of obese humans and various animal models showing significantly
elevated adipose tissue TNF- protein and mRNA without detectable
increases in plasma TNF- levels. These observations suggest an
autocrine-paracrine mode of action of TNF- rather than an endocrine
route.
In summary, our findings demonstrate a previously unknown link between
PAI-1, a potent inhibitor of the fibrinolytic system, and the
development of the obese phenotype in ob/ob mice (Fig. 1)
. Our data
suggest that the increase in PAI-1 levels in ob/ob mice contributes to
the weight gain and metabolic abnormalities of these mice and that the
increase in PAI-1 is not simply a secondary phenomenon of these
disorders. The effects of PAI-1 on obesity may be explained, at least
in part, by its effects on TNF- gene expression since reduction of
adiposity and amelioration of the diabetic phenotype in PAI-1-deficient
ob/ob mice are paralleled by a dramatic reduction of the chronic
overexpression of adipose tissue TNF-. The mechanisms by which PAI-1
interferes with TNF- synthesis are unknown and will be the subject
of future studies. Ultimately, attempts to use PAI-1 gene transfer to
rescue the obese and diabetic phenotype will be needed to unequivocally
imply PAI-1 in the pathogenesis of metabolic disorders associated with
obesity.
FOOTNOTES
1 To read the full text of this article, go to
http://www.fasebj.org/cgi/doi/10.1096/fj.00-0750fje ; to cite this
article, use FASEB J. (June 27, 2001)
10.1096/fj.00-0750fje 
2 Present address: Georg-August University of
Goettingen, Department of Cardiology, Robert Koch Str. 40, D-37075
Goettingen, Germany.
3 Present address: Tokyo Metropolitan Bokutou
General Hospital, Department of Endocrinology 1–16-2–1002 Sugamo,
Toshima-Ku, Tokyo 170–002 Japan.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
