The Insulin-Related Ovarian Regulatory System in Health and Disease

The Insulin-Related Ovarian Regulatory System in Health and Disease LEONID PORETSKY, NICHOLAS A. CATALDO, ZEV ROSENWAKS, AND LINDA C. GIUDICE Division of Endocrinology, Department of Medicine (L.P.) and Division of Reproductive Endocrinology,
Department of Obstetrics and Gynecology (Z.R.), New York Presbyterian Hospital and Weill Medical
College of Cornell University, New York, New York 10021; and Division of Reproductive Endocrinology
and Infertility, Department of Obstetrics and Gynecology, Stanford University Medical Center (N.A.C.,
L.C.G.), Stanford, California 94305 I. Introduction II. Insulin and Insulin Receptor A. Structures of insulin and insulin receptor B. Presence of insulin and insulin receptor in the ovary C. Insulin action and the ovary D. Summary III. IGFs and Their Receptors A. IGF peptides and receptors B. Expression of IGFs and IGF receptors in the ovary C. Role of IGFs in ovulatory function and steroido- genesis D. Summary IV. IGF-Binding Proteins (IGFBPs) and Proteases A. Structural relationships among IGFBPs B. IGFBP expression in the ovary C. IGFBP proteases in the ovary D. IGFBP actions in the ovary E. Role of IGFBPs in follicular development and atresia F. Summary V. Polycystic Ovary Syndrome (PCOS) A. Clinical features B. Theories of pathogenesis C. Insulin resistance in PCOS D. Alterations of IGFs and IGFBPs in PCOS E. Summary VI. The Insulin-Related Ovarian Regulatory System: Implications for Therapy
A. Treatment of PCOS B. Therapeutic use of IGF-I and IGF-II C. Use of GH in ovulation induction VII. Summary and Conclusions I. Introduction I NSULIN, a pancreatic peptide hormone produced in the -cells of the islets of Langerhans, plays a major role in the regulation of carbohydrate, fat, and protein metabolism (1). The classical target organs for insulin action are muscle,
adipose tissue, and liver (2). Until approximately a decade
ago, insulin was not thought to play a significant role in the
regulation of ovarian function, despite suggestions of the
“gonadotropic” function of insulin (3) in observations of
abnormal ovarian function in young women with type 1
diabetes mellitus by Joslin et al. (4), which predated the
discovery of insulin more than 75 years ago (5). A resurgence
of interest in the ovarian effects of insulin was stimulated by
observations of severe ovarian hyperandrogenism in women
with syndromes of extreme insulin resistance (6, 7), which
led to the hypothesis that high levels of circulating insulin
may cause excessive androgen production in these patients
(8, 9). The demonstration of insulin’s ability to stimulate
steroidogenesis in ovarian cells in vitro (10) and the demon-
stration of insulin receptors in both stromal and follicular
compartments of the human ovary (11, 12) established the
ovary as another important target organ for insulin action. This field was further expanded by studies of the ovarian production and ovarian effects of the insulin-like growth
factors, IGF-I and IGF-II, by the discovery of ovarian type I
and type II IGF receptors, and by the discovery of the ovarian
production of binding proteins [IGF-binding proteins
(IGFBPs)] for these two growth factors (13–15). Thus, in ad-
dition to insulin, a role for the structurally related IGFs in
ovarian function has gained recognition. Over the last de-
cade, a significant amount of information has accumulated
about the role of insulin and IGFs in the ovary at the mo-
lecular, cellular, and clinical levels in a variety of normal and
pathological conditions. Therefore, a need has arisen for a
comprehensive review of what we term the insulin-related
ovarian regulatory system. This system consists of the fol-
lowing components (Table 1): insulin; IGF-I and IGF-II; in-
sulin receptor; type I and type II IGF receptors; IGFBPs 1– 6;
and IGFBP proteases. While the pituitary ovarian regulators, LH and FSH, are of paramount importance to ovarian function (16, 17), the in-
sulin-related ovarian regulatory system likewise participates
in normal follicle development (3, 14, 18 –23). Its alterations
may be important in the ovarian dysfunctions observed in a
number of disorders, including diabetes mellitus, obesity,
polycystic ovary syndrome (PCOS), and syndromes of ex-
treme insulin resistance (9, 24 –28). The physiological and Address reprint requests to: Leonid Poretsky, MD, New York Pres- byterian Hospital and Weill Medical College of Cornell University, 525
East 68th Street, New York, New York 10021 USA. * This work was supported in part by NIH Grants M01 RR-00047 (L.P.), NICHD R035618 – 01A1 (L.P. and Z.R.), K08 HD-01141 (N.A.C.)
and R01 HD-31579 (L.C.G.). 0163-769X/99/$03.00/0
Endocrine Reviews 20(4): 535–582
Copyright © 1999 by The Endocrine Society
Printed in U.S.A. 535 by on November 4, 2008 edrv.endojournals.org Downloaded from clinical significance of this regulatory system is underscored
by recent observations which demonstrate that pharmaco-
logical agents capable of manipulating the components of
this system may be useful in the therapy of some of these
disorders (29 –38). This article reviews the role of each component of the insulin-related ovarian regulatory system in both normal
ovarian physiology and in relevant pathological states, the
interactions among the components of this system, and the
therapeutic implications of this system for women with ab-
normal ovarian function. II. Insulin and Insulin Receptor A. Structures of insulin and insulin receptor Detailed reviews of the structures of insulin and its re- ceptor are available (1, 2, 39 – 42), and thus only a brief over-
view will be presented here. Insulin is a 5900 mol wt polypeptide secreted by the -cells of the pancreatic islets of Langerhans. The human insulin
gene is located on chromosome 11 (39) and encodes pre-
proinsulin, a 110-amino acid single-chain polypeptide that is
the precursor of insulin (1). Pre-proinsulin is proteolytically
converted to proinsulin, which consists of the A chain, B
chain, and C peptide. Proinsulin is homologous with IGF-I
and -II and can bind to the insulin receptor with approxi-
mately 10% of the affinity of insulin. Insulin is produced after
the C-peptide is cleaved from proinsulin by endopeptidases
active in the Golgi apparatus and in secretory granules. The
endopeptidases preferentially cleave either at the C pep-
tide/B chain junction, between Arg31 and Arg32 (endopep-
tidase type I), or at the C peptide/A chain junction, between
Lys64 and Arg65 (endopeptidase type II). The resulting in-
sulin molecule consists of an A chain (21 amino acids) and
a B chain (30 amino acids), with three disulfide bridges: two
between the A and the B chains (A7-B7 and A20-B12) and one
within the A chain (A6-A11). The insulin receptor is a heterotetramer consisting of two - (135 kDa molecular mass) and two - (95 kDa molecular mass) subunits (2). The gene for the insulin receptor is located
on the short arm of chromosome 19 (43– 45), contains 22
exons, is more than 150 kb in length, and encodes the pro-
receptor, a single-chain polypeptide with a molecular mass
of 190 kDa that contains one and one -subunit. The mature 2 2 heterotetrameric form of the receptor results from dimerization and several posttranslational processing steps,
including proteolytic cleavage. An isoform of the receptor
lacking 12 amino acids encoded by exon 11 results from
alternative mRNA splicing. Insulin receptors lacking exon 11 may have biological properties somewhat different from
those containing exon 11 (46), although no significant dif-
ferences in insulin binding and insulin receptor kinase ac-
tivity between these two variants were observed (47). Insulin receptor -subunits are extracellular structures possessing cysteine-rich domains that serve as insulin-bind-
ing sites. Insulin receptor -subunits have extracellular, transmembrane, and intracellular domains, the latter con-
taining an ATP-binding site and several tyrosine autophos-
phorylation sites. After insulin binds to the -subunits, the -subunits become phosphorylated on tyrosine residues and acquire kinase activity, initiating a cascade of intracellular
protein phosphorylation (48, 49). The most important intra-
cellular proteins phosphorylated under the influence of the
insulin-receptor tyrosine kinase are the insulin receptor sub-
strates (IRS), several of which have been described (50 –58).
IRS-1, the first of these to be discovered (2, 59), has a mo-
lecular mass of 131 kDa and possesses 14 potential tyrosine
phosphorylation sites. IRS-1 appears to be important in in-
sulin receptor function and its variant forms are sometimes
associated with diabetes (60, 61). Mice deficient in IRS-2
develop a syndrome resembling type 2 diabetes (62). Some
IRS-1 mutations are associated with insulin resistance and
hyperinsulinemia (63), and codon 972 polymorphism of the
IRS-1 gene is associated with impaired glucose tolerance,
PCOS (64), and late onset of type 2 diabetes mellitus (65).
IRS-1 binds phosphatidylinositol-3-kinase (PI-3 kinase), a src
homology-2 (SH2) domain-containing enzyme, activation of
which is necessary for the initiation of glucose transport (2,
59, 66 – 69). In addition to PI-3 kinase activation, mitogen-
activated protein kinase (MAPK) is also phosphorylated after
insulin receptor binding (2, 49, 59, 70). MAPK activation is
thought to be responsible for the growth-promoting effects
of insulin (2). MAPK can be activated not only by the insulin
receptor, but also by other tyrosine kinase receptors, such as
the type I IGF receptor, and receptors for epidermal growth
factor (EGF) and platelet-derived growth factor (PDGF), as
well as G protein-linked receptors (2, 71, 72). The molecular
link between the MAPK cascade and the insulin receptor
may be p21 Ras, a highly conserved protein involved in cell
growth that may be a critical element in growth factor re-
ceptor and insulin receptor tyrosine kinase action (2, 49, 59). Tyrosine kinase activation is believed to be the main sig- naling mechanism of the insulin receptor (48); it appears to
be the earliest postbinding event and is necessary for many,
although not all, of insulin’s effects, including transmem-
brane glucose transport (73, 74). Overexpression of tyrosine
kinase-deficient insulin receptors in muscle causes insulin
resistance in transgenic animals (75). Tyrosine kinase activity
is required in vivo for phosphorylation of IRS-1 and for PI-3
kinase activation (76). An alternative signaling pathway for the insulin receptor has also been described. It involves generation of inositol-
glycan second messengers at the cell membrane after insulin
binding to receptor -subunits but independently of -sub-
unit tyrosine kinase activation (77). This alternative pathway
for receptor signaling may mediate some of insulin’s effects,
including stimulation of ovarian steroidogenesis (78 – 80)
(Fig. 1), but the role of this system in propagating the insulin T ABLE 1. Components of the insulin-related ovarian regulatory system Insulin
IGF-I
IGF-II
Insulin receptor
Type I IGF receptor
Type II IGF receptor
IGFBPs 1–5
IGFBP proteases 536 PORETSKY ET AL. Vol. 20, No. 4 by on November 4, 2008 edrv.endojournals.org Downloaded from signal for glucose transport and other insulin effects has not
been fully established. Insulin binding to its receptor results in a plethora of metabolic effects, including stimulation of DNA and protein
synthesis, lipogenesis, transmembrane electrolyte transport,
and a variety of effects on carbohydrate metabolism, the most
important of which is stimulation of transmembrane glucose
transport (2). This transport is carried out by a family of
glucose transporter proteins (GLUTs) (81) which, in their
resting phase, reside in intracellular vesicles. After insulin
binds to its receptor, these vesicles are translocated to and
fuse with the plasma membrane. The GLUTs are then in-
serted into the plasma membrane and become functional.
Once glucose transport is completed, GLUTs are recycled to
intracellular vesicles. Insulin signaling for glucose trans-
porter activation is mediated by PI-3 kinase. Insulin receptor-like proteins are present in lower organ- isms that do not produce insulin. For example, in certain
species of worms, daf-2, a gene similar to that of the insulin
receptor, regulates glucose metabolism and longevity (82).
Mutation of the insulin receptor in Drosophila leads to small
ovaries lacking oocytes, and thus sterility (83). Insulin re-
ceptor-like molecules are present in mosquito ovaries (84).
The existence of these homologous proteins in insects sug-
gests that the growth and regulatory functions of the insulin/
IGF receptor family arose before the divergence of insects
and vertebrates more than 600 million years ago (83). Con-
servation of the insulin receptor over this length of time in
a variety of organisms indicates its importance for their sur-
vival. Indeed, mice with a genetic knockout of the insulin
receptor die in the neonatal period (85). B. Presence of insulin and insulin receptor in the ovary Circulating insulin levels in the peripheral blood of normal women are approximately 10 U/ml in the fasting state and
up to 50 U/ml within 1 h after an oral glucose load. In obese
women, these levels are somewhat higher, averaging ap- proximately 15 U/ml in the fasting state and up to 60 U/ml after a glucose load. In insulin-resistant hyperinsu- linemic states such as PCOS or the early stages of type 2
diabetes mellitus, serum insulin levels range from 20 –35 U/ml in the fasting state to 120 –180 U/ml after a glucose load (9, 86). In patients with syndromes of extreme insulin
resistance, circulating insulin levels may be as high as 200 U/ml in the fasting state and up to 1400 –2000 U/ml after a glucose load (9). Ovarian follicular fluid (FF) insulin concentrations range from less than 2 U/ml to 65 U/ml, with a mean value of
approximately 16 U/ml (87). These do not correlate with plasma insulin or FF estradiol (E 2 ) or androstenedione (A) con- centrations, but do correlate directly with those of progesterone
(P) (87). Insulin likely reaches FF from the circulation by tran-
sudation. To our knowledge, intrafollicular concentrations of
insulin have not been reported in women with insulin resis-
tance with or without ovulatory dysfunction. Both in humans and in animal models, insulin receptors are widely distributed throughout all ovarian compartments,
including granulosa, thecal, and stromal tissues (3, 11, 12,
88 –91) (Table 2). Ovarian insulin receptors have the same
heterotetrameric 2 2 structure as insulin receptors in other organs. They possess tyrosine kinase activity (12) and may
stimulate the generation of inositolglycans (79). The regulation of insulin receptor expression in the human ovary has been investigated (92, 93). As in other organs,
insulin itself plays a major role in this process: in vitro, insulin
exposure leads to receptor down-regulation, followed by a
return to normal receptor number approximately 4 h after
insulin exposure ends (92). In vivo, down-regulation of ovar-
ian insulin receptors by insulin has been observed in rats
with experimentally induced hyperinsulinemia (94). In post-
menopausal women, in vivo studies have demonstrated a
positive correlation between insulin receptor number on cir-
culating white cells and in the ovary (93). This relationship
was not found in premenopausal women. Since insulin is the F IG . 1. Insulin receptor, its signaling pathways for glucose transport, and hy-
pothetical mechanisms of stimulation
or inhibition of steroidogenesis. The
main pathways for the propagation of
the insulin signal include the following
events: after insulin binds to the insulin
receptor -subunits, the -subunit ty- rosine kinase is activated; IRS-1 and -2
are phosphorylated; PI-3 kinase is ac-
tivated; GLUT glucose transporters are
translocated to the cell membrane, and
glucose uptake is stimulated. An alter-
native signaling system may involve
generation of inositolglycans at the cell
membrane after insulin binding to its
receptor. This inositolglycan signaling
system may mediate insulin modula-
tion of steroidogenic enzymes (see text
for more details and references). August, 1999 INSULIN-RELATED OVARIAN REGULATORY SYSTEM 537 by on November 4, 2008 edrv.endojournals.org Downloaded from major regulator of receptor number on peripheral leuko-
cytes, these observations suggest, albeit without direct evi-
dence, that insulin is the major regulator of ovarian receptors
in postmenopausal women. In premenopausal women, how-
ever, other circulating factors such as gonadotropins or sex
steroids, or locally produced autocrine regulators such as
IGFs and IGFBPs, may be involved in insulin receptor reg-
ulation. These factors may account for the observation that
in premenopausal women with PCOS and other hyperinsu-
linemic states, ovarian insulin receptor expression is pre-
served (88, 89, 95) and that the insulin receptor may mediate
some of the ovarian effects of insulin despite the presence of
peripheral insulin resistance (9, 79, 96, 97). Insulin-induced hyperandrogenism is unlikely to result from an action of insulin through its own receptor, however,
in disorders in which receptor expression or availability is
significantly compromised, such as the type A syndrome of
insulin resistance and acanthosis nigricans, caused by insulin
receptor mutations, or the type B syndrome, associated with
antiinsulin receptor antibodies (6, 7). In the latter two con-
ditions, insulin receptors likely function as inefficiently in the
ovary as in other organs, and another receptor, such as the
type I IGF receptor, is more likely to mediate the effects of
hyperinsulinemia in the ovary (9). C. Insulin action and the ovary Numerous actions of insulin on the ovary have been dem- onstrated both in vitro (Table 3) and in vivo (Tables 3 and 4),
with no significant differences between humans and other
species (3). 1. Effects on steroidogenesis. a. In vitro studies. In vitro, insulin stimulates ovarian ste- roidogenesis by both granulosa and thecal cells, increasing
production of androgens, estrogens, and progesterone (3, 10, 96 –101). In some studies, the concentration of insulin re-
quired to achieve a stimulatory effect is supraphysiological
(3, 10), suggesting that insulin may be acting through the type
I IGF receptor. Several lines of evidence, however, suggest
that insulin receptors mediate the stimulation of steroido-
genesis by insulin. Willis and Franks (97) demonstrated that
insulin-stimulated steroid production by granulosa cells ob-
tained from both normal women and those with PCOS could
be inhibited by antiinsulin receptor antibodies, but not by
antibodies against the type I IGF receptor. Nestler et al. (79)
recently demonstrated in cultured thecal cells obtained from
women with PCOS that insulin stimulation of testosterone
(T) production could not be inhibited by an antibody against
the type I IGF receptor, suggesting that this effect of insulin
was also mediated by the insulin receptor. Since circulating
levels of insulin rarely are high enough to produce significant T ABLE 3. A summary of insulin effects related to ovarian function Effect Organ Directly stimulates steroidogenesis Ovary Acts synergistically with LH and FSH to stimulate steroidogenesis Ovary Stimulates 17 -hydroxylase Ovary Stimulates or inhibits aromatase Ovary, adipose
tissue Up-regulates LH receptors Ovary Promotes ovarian growth and cyst formation synergistically with LH/hCG Ovary Down-regulates insulin receptors Ovary Up-regulates type I IGF receptors or hybrid insulin/type I IGF receptors Ovary Inhibits IGFBP-1 production Ovary, liver Potentiates the effect of GnRH on LH and FSH Hypothalamus/
pituitary Inhibits SHBG production Liver See text for details and references. T ABLE 2. Expression of IGFs, IGFBPs, IGFBP proteases, type I and type II IGF receptors, and insulin receptors in the human ovary a IGF-I IGF-II IGFBP-1 IGFBP-2 IGFBP-3 IGFBP-4 IGFBP-5 IGFBP-2 protease IGFBP-3 protease IGFBP-4 protease Type I IGF-R Type II IGF-R Insulin receptor Early antral follicles (3–5 mm) O 2 2 G / 2 / 3 /4 / 2 1 3 /3 3 1 T 2 / 2 / 3 /4 2 4 / 2 /2 3 1 S / / / 4 /4 4 /4 1 V / 4 Late antral follicles (7–20 mm) O 2 4 2 G / 4 /4 4 / /4 3 / /4 / 4 cu 4 /4 4 2 T / / 4 /4 3 / 2 /2 2 /2 3 2 S / / / 4 /3 4 /4 1 V /4 /4 /4 Corpus luteum (and granulosa luteal
cells) G 4 /4 4 / / b 4 / b 2–3 T / 3 S / weak V /4 a [Data are from Refs. 88,90,91,344,437,458, and 514.] Since there are discrepancies between the groups using in situ hybridization, these results are reported in the format of Ref. 88/Ref. 344. Data are presented as strongly positive (4 ) to weakly positive (1 ). If no number appears,
the data were not reported. IGF-R, IGF receptor; O, oocyte; G, granulosa; T, theca; S, stroma; V, vascular endothelium; cu, cumulus. b Type I IGF receptor mRNA expression present in granulosa-luteal cells (90). 538 PORETSKY ET AL. Vol. 20, No. 4 by on November 4, 2008 edrv.endojournals.org Downloaded from binding to the type I IGF receptor, the actions of insulin on
the ovary are likely mediated mainly by the insulin receptor. At this time, there is only limited knowledge about the specific effects of insulin on ovarian steroidogenic enzymes.
A stimulatory effect of insulin on aromatase has been sug-
gested by some studies of animal and human ovarian cells
in vitro (102–105), but one study (106) failed to confirm this
finding. 17 -Hydroxylase activity appears to be stimulated
by insulin (29, 107–109), but a recent study of 28 women with
PCOS and 18 normal controls found no correlation between
insulin levels and 17-hydroxyprogesterone (17-OHP) levels
after treatment with GnRH agonist (GnRHa) (110). Insulin
increases P 450 side chain cleavage (scc) enzyme mRNA in porcine granulosa cells (111) and P 450 scc activity in goldfish follicles (112). A similar effect could not be demonstrated,
however, in a human ovarian thecal-like tumor line (101). In
the latter study, insulin had no effect on the enzyme activity
or mRNA concentration of 17 -hydroxylase/17,20-lyase
(P 450 c17) or 3 -hydroxysteroid dehydrogenase (HSD), but forskolin stimulation of 3 -HSD mRNA was enhanced by
insulin. In human luteinized granulosa cells, 3 -HSD ex-
pression was found to be stimulated by insulin (106). b. In vivo studies (Table 4). It has not been consistently demonstrated that insulin stimulates ovarian steroidogene-
sis in vivo (113). Several studies have examined the in vivo
effects of insulin on aromatase. In rats with experimental
hyperinsulinemia, an increased estrone (E 1 ) to A ratio was demonstrated, consistent with a stimulatory effect of insulin
on ovarian or peripheral aromatase (94). In women, an in-
sulin infusion study has suggested a similar effect (114), and
in hyperinsulinemic women with PCOS, an increased E 2 /A ratio was seen after gonadotropin stimulation, compared
with normoinsulinemic women with PCOS (115). Relatively
insulin-deficient women with type 2 diabetes show reduced
aromatase activity (116). The increase in circulating A level
observed during insulin infusions in women (117, 118), on
the other hand, suggests that insulin may inhibit aromatase.
In short, it remains unclear whether or how insulin regulates
aromatase in vivo. The effect of insulin on ovarian androgen production in women has been extensively studied (Tables 3 and 4). In PCOS,
a positive correlation has been reported between insulin and T
or A levels (119–122) in several studies, while more recent
studies (123–127) failed to find such a relationship. In insulin
infusion studies that maintained hyperinsulinemia for several
hours, a stimulatory effect of insulin on ovarian androgen pro-
duction has not been consistently found. Stuart and associates
(117, 118, 128) demonstrated elevation of A and dehydroepi-
androsterone (DHEA) in normal lean and obese women and in
women with insulin resistance and acanthosis nigricans during
a euglycemic, hyperinsulinemic clamp study. Micic et al. (129)
demonstrated an increase of T in patients with PCOS during a
4.5-h insulin infusion. On the contrary, Diamond et al. (130)
could demonstrate no change in total or free T or in A during
either insulin or glucose infusion in normal women. Similarly,
Nestler et al. (131) could not demonstrate a rise in T in normal
women during insulin infusion. Dunaif and Graf (114) exam-
ined gonadotropin and sex hormone levels basally and during
insulin infusion in normal and PCOS women. No effect on
gonadotropins was demonstrated; E 2 levels rose in response to insulin in normal women. In PCOS women, A levels increased,
but T, free T, and dihydrotestosterone (DHT) levels declined. Another group of studies has examined the effects of food intake or oral or intravenous administration of glucose on
circulating androgen concentrations. In normal women,
Parra et al. (132) found an increase in free T and no change
in A after breakfast, but a decline of free T after an oral
glucose load. Elkind-Hirsch et al. (133) failed to demonstrate
a rise of either T or A during a tolbutamide-enhanced in-
travenous glucose tolerance test (IVGTT). Smith et al. (134)
found a positive correlation between insulin responses and
A, T, and DHT levels during oral glucose tolerance testing
(OGTT) in hyperandrogenic and normal women, but Tiitinen
et al. (135) demonstrated no significant change in T or A in
women with PCOS or weight-matched normal controls after
an oral glucose load and Tropeano et al. (136) demonstrated
a decline of T, A, and DHEA during an OGTT. On occasion,
both a stimulatory response and the lack of it have been
observed in the same study. For example, Anttila et al. (137)
reported a tendency to increased serum T levels during
OGTT mainly in a subgroup of PCOS patients with both
hyperinsulinemia and elevated LH levels; most PCOS pa-
tients, however, showed a decline in T. Fox et al. (138) found
that serum androgens declined in PCOS patients during
OGTT, but A rose during a 2-h intravenous insulin infusion
in obese controls. Since a decline of serum T in the course of
a 3- or 4-h OGTT may be attributed to diurnal variations of
T, the lack of an increase of T under these conditions argues
against a significant acute stimulatory or inhibitory effect of
insulin on ovarian androgen production in vivo. While studies that raise circulating insulin concentration have produced variable effects on serum androgen levels,
studies in which insulin levels were reduced have consis-
tently demonstrated a decline in serum androgen levels in
insulin-resistant hyperandrogenic women (139, 140) (see Sec-
tion VI.A). Whether insulin levels are lowered with diazoxide
(30, 141), octreotide (34, 142), metformin (29, 31, 108, 143–
146), troglitazone (35, 36), or through weight loss (147–156),
a decline in serum androgen levels is usually found and
ovulatory function improves (Table 4). In contrast to the
studies in which insulin levels were elevated acutely for
several hours, the effect of the reduction of circulating insulin
can be studied over many weeks. If insulin-induced stimu-
lation of ovarian steroidogenesis requires a prolonged ex-
posure to excess circulating insulin, the latter group of stud-
ies is more likely to be able to demonstrate, albeit indirectly,
a stimulatory effect of insulin on circulating steroids. A con-
founding factor in some of these studies is a decline in cir-
culating LH, which may be responsible, at least in part, for
the reduced androgen secretion (157). In summary, it appears that insulin may have stimulatory or inhibitory effects on ovarian steroidogenic enzymes, but
the responses of specific enzymes may vary with cell type
and possibly among species. Further studies are needed on
the effects of insulin on steroidogenic enzymes in the ovaries
both in vitro and in vivo. 2. Interactions with gonadotropins. Acting at the ovarian level,
insulin appears to potentiate the steroidogenic response to
gonadotropins, both in vitro and in vivo (96, 102, 157–163). In August, 1999 INSULIN-RELATED OVARIAN REGULATORY SYSTEM 539 by on November 4, 2008 edrv.endojournals.org Downloaded from granulosa cells, this effect may be mediated by an increase in
LH receptor number, since insulin in concert with FSH in-
creases ovarian LH-binding capacity (13, 164). In addition,
insulin may act on the pituitary to increase gonadotrope
sensitivity to GnRH. Evidence for this effect comes both from
in vitro studies (165, 166) and indirectly from studies in in-
sulin-resistant patients treated with insulin sensitizers, in
whom circulating LH declined concomitantly with insulin
(29, 31, 35, 108). On the other hand, in rats with experimental
hyperinsulinemia maintained over six 4-day estrous cycles,
the response of gonadotropins to GnRH did not differ from
that of controls (94). In normally cycling women, increasing
body mass index (BMI) did not have an effect on gonado- tropin secretion and in women with PCOS BMI and LH levels
were inversely related (167–169), while gonadotropin re-
sponsiveness to GnRH did not change after insulin infusion
(114). In summary, it remains unclear whether hyperinsu-
linemia significantly enhances gonadotrope responsiveness
to GnRH in vivo, as it does in vitro. 3. Effects on ovarian growth and cyst formation. In a rat model,
a synergistic interaction between LH/hCG and insulin on the
ovary can be demonstrated directly during experimentally
induced hyperinsulinemia, which enhances hCG-induced
ovarian growth and cyst formation (28, 170) (Fig. 2). This
synergistic action of insulin with LH/hCG is seen regardless T ABLE 4. Selected in vivo studies of the effect of insulin on circulating ovarian androgens, SHBG and LH Ref Correlative studies Burghen et al., 1980 119 PCOS and control, obese I positively correlated with T, A Chang et al., 1983 120 PCOS, nonobese I positively correlated with T, A Pasquali et al., 1983 121 PCOS, obese and nonobese I positively correlated with A Elkind-Hirsch et al., 1991 133 PCOS, obese and nonobese; nonobese controls I positively correlated with T Anttila et al., 1991 124 PCOS without acanthosis nigricans, obese and
nonobese I did not correlate with T or A Toscano et al., 1992 123 Hirsute women, with and without PCOS, obese and
non-obese I did not correlate with T Buyalos et al., 1993 125 PCOS, obese and nonobese Basal and integrated I on OGTT did not correlate with T or A Studies in which circulating insulin levels were raised Insulin infusion: Nestler et al., 1987 131 Nonobese normal women; one obese woman with
IR/HA No change or 2T in normals; no change in T in IR/HA; Stuart et al., 1987 118 Normal obese and nonobese women; obese women
with IR/HA; 1A in all groups Micic et al., 1988 129 PCOS, obese 1T Dunaif and Graf, 1989 114 PCOS with IR, most obese; obese controls PCOS: 1A, 2T, 2fT, 2DHT Normals: A,T,fT,DHT unchanged Stuart and Nagamani,
1990 128 Normal women and women postoophorectomy 1A in both groups; no change in T Fox et al., 1993 138 Normal and PCOS women, obese and nonobese 1A in normal obese; T unchanged in all groups Diamond et al., 1991 130 Normal, nonobese women No effects on T, fT, or A IVGTT tolbutamide: Elkind-Hirsch et al., 1991 133 PCOS, obese and nonobese; nonobese controls No change of A or T in either group OGTT: Smith et al., 1987 134 Normal nonobese and HA obese women I positively correlated with A, T, DHT in both groups Tiitinen et al., 1990 135 Obese and nonobese PCOS; nonobese controls No significant effect on A or T in either group Anttila et al., 1993 137 Normal and PCOS, obese and nonobese 2T, 2A in PCOS; 2T in normals Fox et al., 1993 138 Normal and PCOS, obese and nonobese 2T, 2A in all groups except nonobese normal; 1T in nonobese normal Tropeano et al., 1994 136 Normal and PCOS, obese and nonobese 2T, 2A in both groups; no correlation between I and T Parra et al., 1995 132 Normal women 2fT after OGTT, 1fT after breakfast Studies in which circulating insulin levels were lowered Diazoxide: Nestler et al., 1989 30 Obese PCOS 2T, 2fT, 2A/E; A and LH unchanged Krassas et al., 1998 141 PCOS, obese and nonobese 2fT, 2A, 1SHBG; LH unchanged Octreotide: Prelevic et al., 1992 34 PCOS 2T, 2A, 2LH Fulghesu et al., 1995 142 PCOS 2T, 2A, 2LH only if hyperinsulinemic 540 PORETSKY ET AL. Vol. 20, No. 4 by on November 4, 2008 edrv.endojournals.org Downloaded from of cotreatment with a GnRH antagonist, suggesting that the
growth- and cyst-promoting effects of insulin are exerted
directly on the ovary. Indeed, insulin can stimulate prolif-
eration of both human and rat theca-interstitial cells in vitro
(171–173). In humans, the ability of high insulin levels to
stimulate ovarian growth in vivo has been suggested by a case
report of a patient with the type B syndrome of insulin
resistance, whose sonographically determined ovarian vol-
ume doubled during a prolonged insulin infusion (174). Fur-
thermore, in women with PCOS, circulating insulin levels are
correlated with ovarian volume (175, 176), and after gonad-
otropin stimulation, the increase in ovarian dimensions ob-
served in hyperinsulinemic PCOS is greater than in normo-
insulinemic PCOS (115). 4. Effects on sex hormone-binding globulin (SHBG) production.
Closely linked to the steroidogenic effects of insulin is its
inhibitory effect on hepatic SHBG production, which has
been shown both in vitro and in vivo (177–180). In fact, SHBG
levels may be useful for screening individuals for insulin
resistance, since they correlate negatively with circulating
insulin levels (181–184). An increase in circulating SHBG, as
may be seen in women with PCOS given insulin sensitizers
(see Section VI.A.3) (29, 31, 35), may lead to decreased cir-
culating levels of free steroid hormones, including free T.
Suppression of SHBG production may be largely responsible
for hyperandrogenism in some patients with hyperinsuline-
mic insulin-resistant states. 5. Effects on IGFBP-1 production. Another protein under the
regulatory control of insulin is IGFBP-1. Insulin and BMI are
the major determinants of circulating IGFBP-1 levels in both
obesity (185–187) and PCOS (183, 188 –192). Insulin inhibits
IGFBP-1 production in the liver (193–198), thereby reducing circulating IGFBP-1 levels. Insulin also inhibits IGFBP-1 pro-
duction in ovarian granulosa cells (see Section IV.B), acting
through its own receptor (199). A detailed discussion of the
role of IGFBPs in ovarian function and their regulation in the
ovary is presented in Section IV.D. 6. Ovulation in diabetes mellitus and in states of extreme insulin
resistance. Insulin and IGFs have been shown to suppress
apoptosis in ovarian follicles, thus reducing rates of their
atresia (200, 201). A variety of clinical and experimental ob-
servations in patients with type 1 and type 2 diabetes mellitus
and states of extreme insulin resistance suggest that insulin
may be involved, either directly or indirectly, in the process
of ovulation (3, 9, 202). Insulin deficiency in type 1 diabetes has been associated with disordered ovulation (3, 202). In rats, streptozotocin-
induced diabetes is associated with cessation of ovulatory
cycles, which can be restored with insulin treatment (203). In
mice with alloxan-induced diabetes, a similar reduction in
ovulation rate has been reported (204). While the current
availability of insulin therapy does not allow observation of
a similar phenomenon in human type 1 diabetes, in the
preinsulin era, girls who developed diabetes prepubertally
failed to enter puberty (3, 4). It is difficult to determine
whether it was insulin deficiency itself, the state of chronic
diabetic ketoacidosis, the starvation diets used for treatment,
or the dramatic weight loss that caused the failure of pubertal
development in these girls. In patients with type 1 diabetes
treated with insulin, the hypothalamic-pituitary-gonadal
axis appears to be relatively hypoactive, mainly because of
failure of the GnRH pulse generator (205, 206); low serum sex
hormone levels, including low luteal-phase P levels, have
been described (207, 208). Even with insulin treatment, up to T ABLE 4. Continued Studies in which circulating insulin levels were lowered (continued) Weight loss: Kopelman et al., 1981 148 Obese, HA 2T, 2A, 1SHBG Bates and Whitworth, 1982 150 Obese PCOS 2T, 2A Harlass et al., 1984 149 Obese with irregular menses 2T, 2LH, 1SHBG Pasquali et al., 1989 151 Obese, HA 2T, 2LH Kiddy et al., 1992 147 Obese PCOS 2fT, 1SHBG; T unchanged Holte et al., 1995 155 Obese PCOS 2T, 1SHBG; A and LH unchanged Guzick et al., 1994 152 Obese PCOS 2fT, 1SHBG; LH and T unchanged Metformin: Crave et al., 1995 153 Obese, hirsute 2fT, 2A, 1SHBG, T unchanged with weight loss; no additional
effect of metformin Velazquez et al., 1994 144 Obese PCOS 2T, 2fT, 2A, 2LH, 1SHBG Nestler and Jakubowicz, 1996 29 Obese PCOS 2fT, 217-OHP, 2LH, 1SHBG Nestler and Jakubowicz, 1997 108 Nonobese PCOS 2T, 2fT, 2A, 2LH, 1SHBG Diamanti-Kandarakis et al., 1998 145 Obese PCOS 2fT, 2A, 1SHBG; T unchanged Morin-Papunen et al., 1998 146 Obese PCOS 2fT; T, SHBG, LH unchanged Troglitazone: Dunaif et al., 1996 35 Obese PCOS 2fT, 2A, 2LH, 1SHBG Ehrmann et al., 1997 36 Obese PCOS 2T, 2fT, 2A, 1SHBG; LH unchanged I, Insulin; fT, free testosterone; A, androstenedione; DHT, dihydrotestosterone; LH, luteinizing hormone; SHBG, sex hormone binding globulin; IR, insulin resistance; HA, hyperandrogenism (hyperandrogenic). August, 1999 INSULIN-RELATED OVARIAN REGULATORY SYSTEM 541 by on November 4, 2008 edrv.endojournals.org Downloaded from one third of young women with type 1 diabetes may expe-
rience delayed menarche and oligomenorrhea of hypotha-
lamic origin (205). Hyperinsulinemia resulting from exogenous insulin ad- ministration is often present in treated patients with type 1
diabetes. If such patients gain excessive weight, their LH:FSH
ratio increases, SHBG levels decrease, and more than 70%
develop polycystic ovaries (209); the response of 17-OHP to
GnRHa in oligomenorrheic diabetic adolescents is exagger-
ated, resembling the response reported in insulin-resistant
patients with PCOS (29, 108, 210). Some patients with type
2 diabetes have mildly elevated androgen levels or increased
androgen responses to GnRH stimulation (116, 202) as well
as reduced SHBG levels (211), particularly in the early, hy-
perinsulinemic stage of the disease (116, 212). It should be
noted that hyperinsulinemia in patients with diabetes is rel-
atively mild, compared with that seen in patients with syn-
dromes of extreme insulin resistance, and that significant
hyperandrogenism is not characteristic of women with either
type 1 or type 2 diabetes (9). Hyperandrogenism and polycystic ovaries or ovarian hy- perthecosis are commonly found in states of extreme insulin
resistance (9, 140, 213). These conditions are sometimes
caused by mutations of the insulin receptor gene (214 –216)
and include the type A syndrome (6), leprechaunism (9, 217,
218), Rabson-Mendenhall syndrome (9, 215), and syndromes
characterized by defective insulin receptor signaling (74, 219,
220). Premenopausal patients with the type B syndrome (in-
sulin resistance and acanthosis nigricans associated with the
presence of antiinsulin receptor antibodies) also exhibit hy-
perandrogenism (7, 8). Although there is evidence that hyperinsulinemia contrib- utes to the development of hyperandrogenism, not all clinical
conditions associated with hyperinsulinemia lead to ovarian
androgen overproduction. For example, most women with
type 1 diabetes, who are often hyperinsulinemic because of
exogenous insulin administration but usually do not exhibit
significant insulin resistance, do not become hyperandro-
genic, but rather exhibit hypothalamic-pituitary-ovarian axis
hypofunction. It is not clear why hyperinsulinemia devel- oping in the setting of insulin resistance, rather than any form
of hyperinsulinemia, is associated with ovarian hyperandro-
genism, particularly since correction of hyperinsulinemia
without correction of insulin resistance may improve ovarian
function (38, 221–223). Dissecting the effects of hyperinsulinemia from those of insulin resistance is difficult (224, 225). One can postulate,
however, that because the postbinding insulin receptor path-
ways may diverge (2, 9, 226), in conditions characterized by
hyperinsulinemia without primary insulin resistance all in-
sulin receptor-signaling pathways are significantly down-
regulated, whereas when hyperinsulinemia is caused by in-
sulin resistance, only some of these pathways (e.g., glucose
transport) may be deficient, while others may be hyper-
stimulated (9, 227, 228). Thus, if hyperinsulinemia promotes
androgen production by activating insulin-signaling path-
way(s) distinct from those involved in glucose transport,
hyperandrogenism would be more likely to develop in the
setting of insulin resistance and compensatory hyperinsu-
linemia. 7. Interactions of insulin with leptin; leptin-mediated effects on
ovulation. New insights into the relationship between weight
and ovulation and the role that insulin may play in modi-
fying this relationship emerged with the discovery and char-
acterization of leptin. Leptin is a 16-kDa protein produced by
adipose cells (229 –233). Circulating leptin levels are stimu-
lated by estrogen and inhibited by androgens (234 –236) and
are directly proportional to adipose tissue mass (236 –241).
Leptin regulates body weight by binding to specific receptors
in the hypothalamus and thus decreasing food intake (242–
244). Leptin is encoded by the ob gene, which is defective in
genetically obese ob/ob mice (229, 231, 237, 245). These ani-
mals are also insulin resistant and infertile. Replacement of
leptin in ob/ob mice produces weight loss, reverses metabolic
abnormalities, and restores ovulation and fertility (246, 247).
Db/db mice and Zucker fatty rats have a similar phenotype,
which results from a genetic abnormality of the leptin re-
ceptor (237, 245, 248). A human kindred with an ob mutation
has been described, in which two prepubertal cousins with F IG . 2. The effects of 23 days of daily injections of normal saline (control), hCG, insulin, or insulin plus hCG and GnRHant on gross ovarian morphology in rats. Female Sprague-Dawley rats were randomized into the following treatment groups: vehicle; high-fat diet (to control for
the effects of weight gain); insulin; hCG; GnRH antagonist (to control for possible central effects of insulin vs. direct effects on the ovary);
GnRHant and HCG; insulin and GnRHant; insulin and hCG; insulin, hCG, and GnRHant. Ovarian morphology in the group treated with insulin
and hCG (not shown) did not differ from that seen in the group treated with insulin, hCG, and GnRHant (shown above). [Reproduced with
permission from L. Poretsky et al.: Metabolism 41:903–910, 1992 (170). W. B. Saunders Co.] 542 PORETSKY ET AL. Vol. 20, No. 4 by on November 4, 2008 edrv.endojournals.org Downloaded from a frameshift mutation in the ob gene suffer from massive
obesity (249). It is not yet known whether they will develop
reproductive abnormalities. Similarly, a mutation of the hu-
man leptin receptor gene associated with obesity has been
reported (250). A rise in circulating leptin levels is associated with and precedes puberty (251), and higher circulating leptin levels
are associated with a younger age at menarche (252, 253),
possibly because leptin serves as a signal for the initiation of
an early pubertal gonadotropin-secretory pattern (254 –257).
A rapid decline of circulating leptin levels is observed during
caloric restriction (258) or starvation (244, 259, 260). A decline
in leptin may be responsible for the activation of the hypo-
thalamic-pituitary-adrenal axis and the inhibition of the go-
nadotropic axis observed with stress (261, 262), since these
responses can be abolished in animals by leptin administra-
tion (233, 263). Leptin receptors are present in the ovary (264 –266). Their functional capacity and their role in both normal and ab-
normal ovarian function remain to be firmly established
since two leptin receptor isoforms exist, one with a full-
length and another with a truncated intracellular domain
(267). While the action of leptin on gonadotropin secretion is
stimulatory, the direct effects of leptin on ovarian steroido-
genesis may be either inhibitory or stimulatory (264, 266,
268). For example, leptin inhibits insulin-induced P and E 2 production in bovine granulosa cells (264) and reduces syn-
ergism between FSH and IGF-I on E 2 production in rat gran- ulosa cells (268). On the other hand, leptin appears to stim-
ulate ovarian 17 -hydroxylase (265). Insulin stimulates secretion of leptin by adipocytes (269 – 272). In addition, by promoting lipogenesis, insulin may
increase adipose tissue mass, thereby further enhancing lep-
tin production. However, there is no apparent acute effect of
feeding on leptin levels (260, 273, 274) and no correlation
between leptin and insulin sensitivity in vivo (273). Never-
theless, circulating leptin levels rise with acute massive over-
feeding over a 12-h period (275). Leptin inhibits insulin secretion from isolated pancreatic islets in some studies (276, 277), but stimulates insulin se-
cretion in others, either by a direct stimulatory effect on
pancreatic -cells (278) or because of its inhibitory effect on
somatostatin (279). Leptin may affect pancreatic function
through the autonomic nervous system (280) and was shown
to improve insulin sensitivity in normal rats, reducing glu-
cose and insulin levels (281). When administered intracere-
broventricularly, leptin enhanced insulin-stimulated glucose
metabolism (282). Leptin has been shown to possess antidi-
abetic properties in some studies (283, 284), but in other
studies it did not affect glucose-stimulated insulin secretion
and did not have a significant effect on glucose transport or
insulin action in either adipocytes or muscle cells (285, 286).
In some circumstances, as, for example, in the setting of
obesity, leptin may contribute to the development of insulin
resistance and diabetes (287–290). The above observations point to a complex relationship among insulin, leptin, body weight, ovarian steroidogenesis,
and ovulation (Fig. 3). If a certain “threshold” level of leptin
is needed to activate the hypothalamic-pituitary-ovarian
axis, then a certain mass of adipose tissue must be present for ovulation to occur (291). In states characterized by hypoin-
sulinemia, such as starvation, weight loss, or untreated type
1 diabetes mellitus, amenorrhea may develop (292, 293), pos-
sibly because of a decline in circulating leptin (294) and a
resultant deactivation of the hypothalamic-pituitary-ovarian
axis (233, 293, 295). Thus, insulin deficiency may contribute
to abnormalities of ovulatory function either directly, by
affecting gonadotropins or the ovaries, or indirectly, by neg-
atively influencing secretion of leptin. On the other hand,
states characterized by insulin excess may be associated with
higher circulating levels of leptin. Whether such putative
leptin excess would play a role in the development of the
hyperandrogenism or anovulation observed in hyperinsu-
linemic states remains to be determined. 8. Effects of insulin on expression of ovarian type I IGF receptors.
In addition to participating, directly or indirectly, in the
regulation of ovarian steroidogenesis and insulin receptor
number in the ovary, insulin may also affect the expression
of ovarian type I IGF receptors. In vivo studies in rats dem-
onstrated that experimental hyperinsulinemia, while down-
regulating ovarian insulin binding, increased ovarian IGF-I
binding (94) (Fig. 4). That this phenomenon may also occur
in humans is suggested by the observations of Samoto et al.
(95) and Nagamani and Stuart (296), who demonstrated that
in women with hyperthecosis or PCOS, ovarian type I IGF
receptors are up-regulated, while insulin receptors are
down-regulated. Pepper and colleagues (297) have reported
that ovarian [ 125 I]IGF-I binding in a patient with ovarian hyperthecosis was increased over that found in normal con-
trols (12, 298). Interestingly, an increase in type I IGF receptor
expression in PCOS may not be limited to the ovaries: a rise
in erythrocyte type I IGF receptors in these patients has also
been reported (299). Further, hyperinsulinemia may increase
expression of hybrid insulin/type I IGF receptors in a variety
of insulin target tissues (300), although this process has not
yet been described in the ovary. F IG . 3. The relationships among insulin, leptin, pituitary gonado- tropins, and ovarian steroidogenesis. Insulin stimulates leptin secre-
tion, enhances pituitary gonadotropin response to GnRH, and pro-
motes ovarian steroidogenesis. Leptin stimulates the hypothalamic-
pituitary-gonadal axis at the level of the hypothalamus and/or
pituitary; it inhibits ovarian E 2 and P production, but may stimulate androgen production by stimulating 17 -hydroxylase activity or ex-
pression. Leptin and insulin potentiate each other’s secretion, al-
though leptin may inhibit insulin secretion under some circum-
stances. Ovarian sex steroids inhibit FSH production and either
inhibit (E 2 , T, P) or stimulate (E 1 ) LH responsiveness to GnRH. August, 1999 INSULIN-RELATED OVARIAN REGULATORY SYSTEM 543 by on November 4, 2008 edrv.endojournals.org Downloaded from In addition to up-regulating type I IGF receptors in the ovary, insulin may also increase the cellular pool of p21 Ras
(49, 301). Both up-regulation of type I IGF receptors and an
increase in the pool of p21 Ras may amplify the effects of
IGF-I on steroidogenesis and follicle development. Further-
more, up-regulation of type I IGF receptors may also amplify
the effects of IGF-II, the dominant ligand for the type I IGF
receptors in human granulosa cells (see Section III.B). Finally,
up-regulation of type I IGF receptors by insulin may amplify
the effects of insulin itself in states of extreme insulin resis-
tance, in which circulating concentrations of insulin are very
high and insulin receptors are either genetically defective or
blocked by antiinsulin receptor antibodies. Under these cir-
cumstances, as discussed previously, insulin may act mainly
by binding to the type I IGF receptor via the “specificity
spillover” effect (9, 302). Thus, the ability of hyperinsulin-
emia to up-regulate ovarian type I IGF receptors may con-
tribute to the ovarian growth and stimulation of steroido-
genesis by IGF-I, IGF-II, and insulin. D. Summary The role of insulin in the ovary may be summarized as follows: 1) Insulin receptors are widely distributed through-
out all ovarian compartments. Ovarian insulin receptors
have a subunit structure identical to insulin receptors in
other organs, possess tyrosine kinase activity, and are capa-
ble of stimulating the generation of inositolglycan second
messengers. 2) At this time there is no convincing direct in
vivo evidence that hyperinsulinemia acutely stimulates ovar-
ian steroid production, but there is direct in vitro evidence
and indirect in vivo evidence for a stimulatory effect of insulin
on ovarian steroidogenesis. The in vitro evidence suggests
that the stimulatory effect of insulin on steroidogenesis is
mainly mediated by the insulin receptor and may involve the
inositolglycan pathway. The in vivo evidence is largely de-
rived from experiments in which a reduction in circulating
insulin levels produces a decline of circulating androgens
and from clinical observations in women with both insulin
deficiency and insulin excess. 3) The effects of insulin on
ovulation are complex. A threshold level of insulin is likely to be required for the normal function of the hypothalamic-
pituitary-ovarian axis, either because of the direct stimula-
tory effects of insulin on this axis or because of the stimu-
latory effects of insulin on leptin secretion (both direct, with
insulin stimulating adipocyte production of leptin, and in-
direct, because of insulin-stimulated lipogenesis). Leptin, in
turn, participates in the initiation of puberty and activation
of the hypothalamic-pituitary-gonadal axis. On the other
hand, excessive circulating insulin, particularly in the setting
of insulin resistance, may enhance ovarian androgen pro-
duction and thus may contribute to the development of
anovulation. 4) Insulin may amplify its own effects, the ef-
fects of IGFs, and those of gonadotropins by up-regulating
type I IGF receptors and gonadotropin receptors, as well as
by inhibiting production of IGFBP-1, both in the liver and
ovary. In the setting of insulin resistance and hyperinsulin-
emia, therefore, a cycle of events that leads to a self-perpet-
uating amplification of the ovarian effects of insulin and IGFs
can develop (Fig. 5). In reviewing the literature dealing with the effects of in- sulin on ovarian function, it is important to distinguish those
effects that have been mainly demonstrated in vitro or in
animal systems, and therefore may contribute only in a lim-
ited way to our understanding of normal and abnormal
human ovarian physiology, from those that have been clearly
demonstrated in women in vivo. In our opinion, the only
insulin-related effects on ovarian function that have been
consistently observed in women in vivo are insulin-induced
suppression of hepatic SHBG and IGFBP-1 production. The
importance of these effects in both normal and pathological
conditions still needs to be clarified. The importance for
normal and abnormal human ovarian function of the other
insulin effects discussed in this section, such as its direct
effects on ovarian steroidogenesis, growth, and cyst forma-
tion; its effects on the expression of ovarian receptors for
insulin, IGF-I, and LH; and its synergistic action with go-
nadotropins, remains to be established. The reported ovarian
effects of insulin in vitro and in vivo are summarized in Tables
3 and 4. F IG . 4. [ 125 I]IGF-I binding to ovarian homogenates from normal rats (A) and
rats with experimentally induced hy-
perinsulinemia (B). Female Sprague-
Dawley rats were treated with either
vehicle (A) or insulin for 23 days.
[ 125 I]insulin (not shown) and [ 125 I]IGF- I binding to ovarian homogenates was
examined. In rats treated with insulin,
a doubling of [ 125 I]IGF-I binding was observed, suggesting amplification of
the number of type I IGF receptors or
hybrid insulin/type I IGF receptors.
[Reproduced with permission from L.
Poretsky et al.: Endocrinology 122:581–
585, 1988 (94). © The Endocrine Soci-
ety.] 544 PORETSKY ET AL. Vol. 20, No. 4 by on November 4, 2008 edrv.endojournals.org Downloaded from III. IGFs and Their Receptors A. IGF peptides and receptors 1. IGF-I. IGF-I is a 70 amino-acid, single-chain polypeptide
that shares significant sequence homology with IGF-II, pro-
insulin, and relaxin. The human IGF-I gene is located on
chromosome 12. The major source of circulating IGF-I is the
liver, but IGF-I is widely expressed in most tissues, especially
during postnatal development (303). IGF-I was first known
as somatomedin C and identified as a mediator of GH action
(304). GH rapidly activates IGF-I gene transcription and also
regulates changes in chromatin structure within the IGF-I
gene, delineating a target within the chromatin for GH action
(305). In addition to GH, other activators of IGF gene tran-
scription include estradiol, experimental diabetes, and an-
giotensin II (306). Null mutants for IGF-I are severely growth
restricted in utero but are fertile (307, 308). 2. IGF-II. IGF-II is a 7.5-kDa, 67-amino acid, single-chain
polypeptide that is approximately 70% homologous with
IGF-I and 50% homologous with proinsulin (14, 309 –312).
The human IGF-II gene is located on chromosome 11, con-
tiguous with the insulin gene. Pre-pro-IGF-II, the precursor
of IGF-II, is a 22-kDa protein. Inactivation of the IGF-II gene
in animals (308, 313) produces growth-deficient but fertile
and otherwise normal individuals. IGF-II is highly expressed
in fetal tissues and tumors, as well as in normal adult tissues.
IGF-II can bind to type I and type II IGF receptors (see below),
as well as to the insulin receptor (302, 314). 3. Type I IGF receptor. The type I IGF receptor precursor
protein consists of 1367 amino acids, comprising both the -
and -subunits of the receptor. The human type I IGF re-
ceptor gene is located on chromosome 15. The mature type
I IGF receptor protein is a heterotetramer consisting of two - and two -subunits and is highly homologous with the insulin receptor (315, 316). The cysteine-rich regions of the -subunits of the insulin receptor and type I IGF receptor are 64 – 67% homologous, whereas the tyrosine kinase domains
of the -subunits are 84% homologous. In addition to IGF-I,
the type I IGF receptor can also bind IGF-II and insulin,
although with somewhat lower affinity. In addition to bind-
ing IGF-I, IGF-II, and insulin, the type I IGF receptor has also
been reported to interact with IGFBPs (317), but the signif-
icance of this finding remains to be determined. Type I IGF
receptor postbinding events, similar to those of the insulin
receptor, include tyrosine phosphorylation of receptor -subunits and IRS proteins, interactions with PI-3 kinase, and activation of MAPK (69, 315, 318, 319). Type I IGF re-
ceptor knockout mice weigh 45% of normal at birth and die
immediately afterward (320). Patients with a deletion of the
distal arm of chromosome 15 lack one copy of the IGF-I
receptor gene and exhibit both intrauterine and postnatal
growth restriction (321, 322). 4. Hybrid insulin/type I IGF receptors. Hybrid receptors that
combine an / insulin hemireceptor and an / type I IGF
hemireceptor have been reported in a variety of tissues, al-
though not in the ovary (41, 323). These receptors can form
in tissues coexpressing both insulin and type I IGF receptors,
theoretically including the ovary. Hybrid receptors have
properties similar to type I IGF receptors, binding IGF-I with
high affinity and insulin with lower affinity. Interestingly, in
situations that are characterized by insulin receptor down-
regulation, the number of hybrid insulin/type I IGF recep-
tors tends to increase (228). 5. Type II IGF receptor. The type II IGF receptor is identical to
the mannose-6-phosphate (Man-6-P) receptor (309, 324 –326).
The gene for the type II IGF receptor is located on the long
arm of chromosome 6. This receptor targets Man-6-P-con-
taining enzymes from the Golgi apparatus to the lysosomes
and also mediates the rapid internalization of IGF-II (309).
The receptor is a single-chain polypeptide of approximately
300 kDa with a large extracellular domain containing IGF-II
binding sites (325, 327). The cytoplasmic domain is very short
and includes tyrosine, threonine, and serine phosphorylation
sites. Type II IGF receptor knockout mice exhibit elevated
IGF-II levels and die in utero (328, 329). Interestingly, if the
IGF-II gene is knocked out at the same time, about 50% of the
fetuses survive to birth (328). Type I/type II IGF receptor
double-knockout mice differ from normal controls only in
their patterns of growth (328). These observations, taken
together, suggest that excessive activation of the type I IGF
receptor by IGF-II may be lethal in utero. The type II IGF receptor can be released from the cell membrane into the circulation. This mechanism may be prin-
cipally responsible for its loss from the cell surface (330 –333).
The circulating form of the IGF-II receptor retains its affinity
for IGF-II (325, 334) and may participate in the local mod-
ulation of organ size in vivo. For example, overexpression of
the soluble IGF-II/Man-6-P receptor in transgenic mice can
significantly decrease the weight of their alimentary canal
(335). Although the type II IGF/Man-6-P receptor is important for IGF-II internalization and degradation, it is unclear
whether this receptor actively mediates IGF-II signaling. Ex- F IG . 5. Hypothetical insulin/IGF self-enhancement mechanisms in the ovary. Hyperinsulinemia, acting through insulin receptors, type
I IGF receptors, or possibly through hybrid insulin/type I IGF recep-
tors increases the number of type I IGF receptors and/or hybrid
insulin/IGF receptors and increases cellular pool of p21 Ras, which
may be responsible for the mitogenic effects of insulin or of IGFs.
Hyperinsulinemia also inhibits IGFBP-1 production, leading to a
further increase in bioavailable IGFs. Thus, hyperinsulinemia may
lead to a self-perpetuating cycle of events resulting in the exagger-
ation of the ovarian effects of both insulin and IGFs, leading to ovarian
enlargement and excessive androgen production (please see the text
for details and references). Solid arrow, action via a receptor; broken
arrow, regulation of a receptor. August, 1999 INSULIN-RELATED OVARIAN REGULATORY SYSTEM 545 by on November 4, 2008 edrv.endojournals.org Downloaded from amples of such signaling have been reported, including stim-
ulation of G-protein activation and of thymidine incorpora-
tion into rat hepatocyte DNA (325, 336 –338). In most
instances, however, the metabolic and growth-promoting
actions of IGF-II appear to be mediated by the type I IGF
receptor (339) or the insulin receptor (314). The type II IGF
receptor, however, may mediate signals involved in angio-
genesis (340) and other processes. Ligands for the type II IGF
receptor, in addition to IGF-II and Man-6-P, include -ga-
lactosidase and other lysosomal enzymes, proliferin, renin,
latent transforming growth factor (TGF)- (329), and leuke-
mia-inhibitory factor (341). In the context of these observa-
tions, the functions of the type II IGF receptor within the
ovary remain to be determined. B. Expression of IGFs and IGF receptors in the ovary 1. Human and nonhuman primate. Distinctive features of IGF
expression in the primate ovary include the predominance of
IGF-II and its pattern of localization (Table 2). Other molecules
that modulate IGF action, including the IGF receptors, IGFBPs,
and IGFBP proteases, are also differentially expressed in the
primate ovary (see below). While the majority of studies that
examined the ovarian expression of IGFs and that of their re-
ceptors were done on human tissue, ovaries from cycling rhesus
monkeys reveal similar expression patterns of IGF-I, IGF-II, and
type I IGF receptor, and there is strong evidence that IGF-II,
aromatase, and IGFBP-4 can be regarded as markers of the
dominant follicle in the rhesus ovary (342). In the human ovary, IGF peptide expression is follicle stage-specific and compartmentalized (Table 2). IGF-I
mRNA is barely detectable in the adult ovary and not in the
granulosa layer at any stage of follicular development (88, 89,
343). IGF-II mRNA is expressed in the theca and perifollic-
ular vessels of all follicles and in the granulosa cells of some
follicles. In small antral follicles, IGF-II mRNA and protein
are detectable in both granulosa and theca (88, 89, 343). In
atretic antral follicles, on the other hand, IGF-II is minimally
expressed by the theca. IGF-II is abundantly expressed and
secreted by granulosa cells of preovulatory follicles as well
as by granulosa-luteal cells harvested during oocyte retrieval
after controlled ovarian hyperstimulation (COH) (88, 90,
344 –347). These findings, plus the observations that granu-
losa cells do not express IGF-II prepubertally, but do so in a
subpopulation of adult follicles, and that gonadotropins reg-
ulate IGF-II mRNA expression and secretion in human gran-
ulosa-luteal cells in vitro (344, 345), suggest that ovarian
IGF-II gene expression is regulated by gonadotropins. Follicular fluid (FF) constituents such as IGF peptides are derived from the circulation as well as from intraovarian
production. In normally cycling women, FF IGF-I levels are
similar in estrogen-dominant and androgen-dominant folli-
cles and do not correlate with follicular size (348). In contrast,
FF IGF-II levels are higher in estrogen- compared with an-
drogen-dominant follicles and correlate positively with fol-
licle size, cycle day, and E 2 and negatively with androgen- estrogen (A:E) ratio (348). In normally cycling women,
simultaneous measurements of IGF-I, IGF-II, and insulin
concentrations in ovarian and peripheral venous blood re-
veal an ovarian gradient only for IGF-II (349), and serum IGF-I and IGF-II levels in normally cycling women do not
vary during the menstrual cycle (348). These data collectively
suggest that FF IGF-I originates from serum by transudation
and that FF IGF-II derives primarily from local production by
the granulosa and possibly by the theca, in addition to some
contribution from the circulation. After COH, FF IGF-II levels
are about 8 times higher than those of IGF-I, and both IGF-I
and IGF-II levels are lower than in serum (350 –353). In con-
trast to spontaneous cycles, these levels in COH do not cor-
relate with follicle size, oocyte maturity, or FF E 2 . FF IGF-I and IGF-II levels were noted to rise with increasing cycle day
3 serum FSH, an index of ovarian reserve (354). Normal circulating levels of IGF-I are not a prerequisite for normal ovarian follicular development in women, as evi-
denced by cases of ovulation and fertility in individuals with
Laron-type dwarfism, which results from GH receptor de-
ficiency (GHRD) (355–358). Furthermore, a normal follicular
response to injected gonadotropins, leading to ovulation and
conception, has been reported in women with GHRD, whose
serum GH was markedly elevated and both serum and FF
IGF-I barely detectable (355, 356). In such subjects, serum
IGF-II levels were about 25% of normal (FF IGF-II was not
measured). These clinical observations support the conclu-
sion that IGF-I does not play an important role in the ovu-
latory process in women. Both type I and type II IGF receptors are found in the human ovary (88, 298, 343, 359). By in situ hybridization, type I IGF
receptor mRNA is predominantly expressed by granulosa cells
and oocytes, with more intense expression in dominant com-
pared with small antral follicles (88, 343). By this technique,
theca and stroma are negative for type I IGF receptors, but
stromal receptors with the specificity of the type I IGF receptor
have been reported in ligand binding studies (298). Type II IGF
receptors are localized to both granulosa and thecal layers, with
more intense expression in the granulosa and in dominant,
compared with smaller, antral follicles (88). By RT-PCR, both
types of receptors were found to be expressed by granulosa,
theca, and stroma and to persist upon culture of both granulosa
and thecal cells (347). 2. Rodent. In the rat, ovarian IGF-I gene expression and pro-
tein production are granulosa specific (360 –362); signifi-
cantly, IGF-I is selectively expressed in the granulosa of only
healthy antral follicles, not in atretic or luteinized follicles or
in theca-interstitial cells (342, 360, 363, 364). IGF-II mRNA
expression is limited to the thecal compartment and blood
vessels (342, 362, 363), but the postnatal decline in ovarian
IGF-II content (365) argues against a significant role for this
peptide in rat ovarian physiology. While type I IGF receptor
mRNA is abundantly expressed in granulosa cells (365), the
corresponding protein is detected not only in the granulosa
but also in the thecal compartment, regardless of the matu-
rational stage or health status of the follicle (363), suggesting
that regulation of the receptor is unlikely to play a major role
in follicular maturation (366). The patterns of IGF-I, IGF-II, and type I IGF receptor ex- pression are essentially the same in rat and mouse ovary (342,
364, 367). IGF-I expression increases at the secondary pre-
antral stage and is abundant in healthy follicles through the
preovulatory stage. Type I IGF receptor is expressed consti- 546 PORETSKY ET AL. Vol. 20, No. 4 by on November 4, 2008 edrv.endojournals.org Downloaded from tutively, regardless of follicular developmental stage or
health (367). These findings lay the groundwork for studies
of ovarian function in transgenic mouse models with dele-
tions of these components (368). 3. Livestock species. Porcine granulosa cells in culture secrete
abundant immunoreactive IGF-I, which is increased by FSH,
cAMP, GH, EGF, and TGF- . IGF-I is abundant in porcine FF,
especially in large follicles. Its levels increase in response to
PMSG and/or GH treatment (369–371). This finding suggests
that gonadotropin and GH action on the granulosa cells of the
developing porcine follicle is mediated in part by local induc-
tion of IGF-I. IGF-II in the porcine ovary is expressed mainly in
the theca and is not under gonadotropin or GH regulation (15,
370, 372). FF IGF-II levels decline in response to GH (370, 372,
373). In the sheep ovary, at least four localization studies of
IGF-I expression have been published, with divergent findings
(374–377). IGF-II is localized to the theca, and its levels in FF are
4-fold greater than those of IGF-I (377, 378). In the cow, IGF-I
is produced by the ovary (379, 380), and its levels in FF increased
with increasing E 2 concentrations and increasing follicle diam- eter in some (379, 381–384), but not all (385–387), studies. IGF-II
is exclusively expressed in the theca, with greater expression in
dominant follicles, compared with subordinate or nonrecruited
ones (388). C. Role of IGFs in ovulatory function and steroidogenesis
(Table 5) 1. Human. Studies of the effects of IGFs on human granulosa
and thecal cells in vitro have primarily employed IGF-I, al-
though as discussed above, the predominant endogenous
locally produced ligand in vivo is IGF-II. IGF actions on the
ovary include augmentation of DNA synthesis and steroi-
dogenesis. IGF-I stimulates DNA synthesis and basal E 2 se- cretion in granulosa and granulosa-luteal cells and inhibits
IGFBP-1 production (199, 389 –396). It also synergizes with
gonadotropins in augmenting E 2 and P production (393, 397– 400). Several studies have been conducted recently of
the effects of IGF-II on human ovarian cellular constituents.
IGF-II stimulates basal P and E 2 secretion by human gran- ulosa-luteal cells (353, 401). It also stimulates aromatization
of androgen precursors (402) and inhibits IGFBP-1 (396) and
IGFBP-2 (403) production by these cells. The effect of IGF-II
on estradiol production is most pronounced if the cells are
preincubated with insulin (402), possibly due to insulin-in-
duced up-regulation of type I IGF receptors, formation of
hybrid insulin/IGF-I receptors, or inhibition of IGFBP-1 pro-
duction. IGF-II also stimulates granulosa-luteal cell DNA
synthesis and proliferation in vitro (401, 404). In granulosa
cells from both unstimulated and gonadotropin-stimulated
preovulatory follicles, IGF-I, both alone and in synergy with
gonadotropins, stimulates P450 aromatase mRNA expres-
sion and activity (405). IGFs also exert actions on human thecal cells and oocytes. In human thecal monolayer cultures, IGF-I enhances DNA
and androgen synthesis (406) and synergizes with LH in A
production (100), although in vivo, a decline of circulating
IGF-I levels after treatment with clomiphene citrate did not
lead to a reduction in hyperandrogenism in PCOS (407).
IGF-II also increases androgen production by human theca
(158). Maturation of immature human oocytes in vitro can be
augmented by IGF-I (408). 2. Rodent. IGF-I actions in rat granulosa and theca have been
extensively reviewed (14, 23, 409, 410). IGF-I acts as a co-
gonadotropin with FSH to stimulate granulosa cells to pro-
duce E 2 and P, and with LH to stimulate thecal androgen production. IGF-I stimulates LH receptor expression in gran-
ulosa and theca (13, 411, 412) and may be required for FSH T ABLE 5. Ovarian actions of IGF-I and IGF-II Species Granulosa (granulosa/luteal) cells Theca cells/explants Follicles Human Promotes: Promotes: Promotes: Aromatase activity and mRNA Androstenedione production ?Oocyte maturation Basal E 2 and P secretion Testosterone production FSH-stimulated E 2 and P secretion DNA synthesis DNA synthesis
Cellular proliferation
IGFBP-4 proteolysis
IGFBP-5 production
?IGFBP-2 proteolysis Inhibits:
IGFBP-1, IGFBP-2 production Rat Promotes: Promotes: Promotes: Adenylate cyclase Androstenedione production ?Ovulatory rupture Aromatase activity P 450 scc mRNA E 2 secretion 17 -Hydroxylase Inhibits: LH receptor synthesis DNA synthesis Apoptosis Progesterone release Cellular proliferation Inhibin secretion
Proteoglycan synthesis
DNA synthesis Inhibits:
IGFBP-5 proteolysis August, 1999 INSULIN-RELATED OVARIAN REGULATORY SYSTEM 547 by on November 4, 2008 edrv.endojournals.org Downloaded from receptor expression in granulosa (368); it also stimulates
granulosa cell production of inhibin -subunit and augments
the stimulation of this response by FSH (413– 415). Stimula-
tion of inhibin- expression in rat granulosa by FSH requires
activation of protein tyrosine kinases by endogenously pro-
duced IGF-I, suggesting that IGF-I signaling is obligatory for
this response (415). IGF-I also stimulates DNA synthesis in
granulosa and theca-interstitial cells (171, 416). In addition to its role in differentiation and proliferation of granulosa and theca, IGF-I also plays an important role in
granulosa survival, since it can inhibit apoptosis (201). Gran-
ulosa cell apoptosis, associated with regular cleavage of nu-
clear DNA by endonuclease, is associated with follicular
atresia (417). In vitro, this process is suppressed by IGF-I and
gonadotropins and enhanced by the presence of IGFBPs
(200). In the human ovary apoptosis is characteristic of an-
drogen- but not estrogen-dominant follicles (418), but reg-
ulation of apoptosis by IGFs has not yet been demonstrated
in human ovarian follicles or cellular components, as it has
in the rat (201). To our knowledge, there are no studies
examining specific effects of IGF-II in rodent ovaries. 3. Livestock species. In the sow, similar effects of IGFs on
granulosa and thecal cell function have been reported as in
humans and rodents (419 – 421). IGF-I stimulates granulosa
cell proliferation and synergizes with FSH in granulosa cell
differentiation (419). IGF-II enhances the delivery of choles-
terol to the P 450 scc enzyme complex and enhances the func- tional activity of this first committed step in P biosynthesis
(421). In sheep, IGF-I stimulates granulosa cells from small
follicles to proliferate and those from larger follicles to pro-
duce P (422), an effect likely mediated through the type I IGF
receptor (423). In the cow, IGF-I stimulates granulosa and
thecal cell proliferation and steroidogenesis (379, 380, 424). D. Summary Although both IGF-I and IGF-II have been shown in vitro to have multiple ovarian effects in various species, IGF-II
appears to be the predominant ovarian IGF in the human.
The IGF-II gene is expressed in the human ovary, and the
effects of IGF-II appear to be similar to those of IGF-I. The
metabolic and growth-related effects of IGF peptides appear
to be mediated under most circumstances by type I IGF
receptors, which are present in all human ovarian compart-
ments. Their numbers appear to be increased under the in-
fluence of insulin, as discussed in Section II.C. Type I IGF
receptors may mediate the effects of insulin in the ovary in
extreme insulin-resistant states with severe hyperinsulin-
emia. Clarification of the presence and the role of hybrid
insulin/type I IGF receptors in the human ovary awaits
further studies. IV. IGF-Binding Proteins (IGFBPs) and Proteases A. Structural relationships among IGFBPs The bioavailability and, therefore, the actions of the IGFs are regulated, in part, by a superfamily of homologous pro-
teins, called IGFBPs, that bind IGFs with high affinity. There
are six IGFBPs, designated IGFBP-1 through IGFBP-6 (425– 427), whose discovery, gene and protein structures, and
mechanisms of actions have recently been reviewed (329,
428, 429). All six IGFBPs have core molecular masses of 23–32 kDa. They are all at least 50% homologous, and for each IGFBP
there is roughly 80% homology among species. The amino
and carboxy termini are most highly homologous among the
different IGFBPs, while the midsequence shows little simi-
larity. The IGFBPs each contain at least 16 conserved cys-
teines, which are important in determining their conforma-
tion. There is also a group of proteins that share limited
sequence homology with the IGFBPs and bind IGFs with low
affinity. Due to their undefined roles as IGFBPs and limited
structural homology to IGFBPs 1– 6, they have been called
IGFBP-related proteins (IGFBP-rPs) (427, 428). The high-af-
finity IGFBPs have dissociation constant (K d ) values for the IGFs in the range of 10 9 to 10 11 mol/liter, compared with 10 6 to 10 7 mol/liter for the IGFBP-rPs (428). The genes for human IGFBP-1 and IGFBP-3 are located on chromosome 7, the IGFBP-2 and IGFBP-5 genes are on chro-
mosome 2, the IGFBP-4 gene is located on chromosome 17,
and the IGFBP-6 gene is on chromosome 12 (329, 430). IGFBP
genes are in close proximity to homeobox (Hox) gene clusters
(Hox A–Hox D), with which they appear to have coevolved.
Hox genes encode DNA-binding proteins that are transcrip-
tionally regulated by retinoic acid, as are some of the IGFBPs
(430). IGFBP-1 and IGFBP-2 both contain the tripeptide motif
Arg-Gly-Asp (RGD), which can bind to integrins, and their
production and function are related to carbohydrate metab-
olism and metabolic homeostasis. In contrast, IGFBP-3, and
likely the highly homologous IGFBP-5, are primarily in-
volved in growth. The IGFBPs have several functions, which include 1) to transport the IGFs in the circulation; 2) to regulate efflux of
IGFs from the vascular space; 3) to prolong the half-life and
metabolic clearance rates of the IGFs; 4) to prevent IGF-
induced hypoglycemia; 5) to directly modulate interactions
of IGFs with their receptors locally within target tissues; and
6) to directly modulate cellular function, independent of their
ability to bind IGFs. All six IGFBPs have been shown to
inhibit IGF action, likely by limiting bioavailable free IGFs
from interacting with their receptors. IGFBP-1 and IGFBP-3
can also be stimulatory to IGF action, presumably by forming
a pool of “slow-release” IGFs. IGFBP-1 and IGFBP-3 addi-
tionally have IGF-independent actions, including alteration
of cellular motility and inhibition of DNA synthesis, respec-
tively. IGFBP-4 and -5 may also have IGF-independent ac-
tions both in the human ovary (431) and in cell lines derived
from other tissues (430, 432). Since the affinities of IGFBPs
1– 6 for the IGFs are equal to or greater than the affinities of
the type I and type II IGF receptors for the peptides,
mechanisms have evolved to decrease IGFBP affinities
and increase IGF bioavailability to the receptors. These
mechanisms include phosphorylation, glycosylation, and
proteolysis (329). This review will focus on IGFBP expression and regulation primarily in the human and rat ovary and underscore the
mechanisms of ovarian IGFBP production and regulation
common to other species. Also discussed are IGFBP prote-
olysis by specific proteases, the regulation of these enzymes, 548 PORETSKY ET AL. Vol. 20, No. 4 by on November 4, 2008 edrv.endojournals.org Downloaded from and their putative functions in normal and pathological ovar-
ian conditions. B. IGFBP expression in the ovary IGFBPs are expressed by granulosa and thecal cells and are present in the FF of every species studied. Significant dif-
ferences exist in the patterns of ovarian expression and reg-
ulation of individual IGFBP species between the human and
animal models. 1. Human (Table 2). The human ovary expresses mRNAs for
IGFBP-1, -2, -3, -4, and -5. In situ hybridization shows dis-
tinctive patterns of mRNA expression for each of these
IGFBPs in antral follicles, with parallel localization of im-
munostainable protein (89). IGFBP-1 is localized only to the
granulosa cells of dominant follicles, not to theca or small
antral follicles. IGFBP-2 is expressed by granulosa cells only
in small, nondominant antral follicles, but by thecal cells in
both dominant and nondominant follicles. IGFBP-3 expres-
sion is found in the theca of all follicles and the granulosa of
only dominant follicles. IGFBP-4 is found in both granulosa
and theca in all follicles, with a slight increase in granulosa
expression in dominant compared with small follicles.
IGFBP-5 has also been localized to both granulosa and theca;
its expression is unaffected by follicular development. No
IGFBP-6 mRNA or protein was localized by in situ hybrid-
ization (89), but expression was detected by RT-PCR (347). A
recent study found IGFBP-4 to be expressed in luteal cells
and in the granulosa and theca layers of only atretic antral,
not healthy or preantral follicles (433). The expression of
IGFBP-2, -4, and -5 by both granulosa and thecal cells has
been confirmed by Northern analysis (347). Expression of
IGFBP-1 has also been found in the corpus luteum (434). The regulation of IGFBP production by the human ovary has been examined in cell culture studies. Two sources of
tissue have been employed: antral follicles from surgically
excised ovaries, and granulosa-luteal cells obtained at oocyte
harvest for in vitro fertilization (IVF) after controlled ovarian
hyperstimulation (COH). Granulosa cells derived from an-
tral follicles in spontaneous cycles release IGFBP-2 and both
core and glycosylated isoforms of IGFBP-4 and express the
corresponding mRNAs (347, 435, 436). Cultures of thecal
tissue derived from these follicles produce IGFBP-2, -3, and
-4; theca from mature healthy follicles also produces pro-
teolytic fragments of IGFBP-3 and -4 (436 – 438). Thecal
IGFBP-3 accumulation, as determined by ligand blotting,
was stimulated markedly by LH/hCG or GH in one study
(438), but these effects were not noted by others (347, 437).
Thecal expression of mRNA for IGFBP-5, but not IGFBP-1, -2,
-3, or -4, is stimulated by LH (347). Because luteinizing granulosa cells from IVF oocyte har- vests are readily available, this model has been extensively
employed to study human IGFBP production. These cells
express mRNAs for IGFBP-1, -2, -3, -4, and -5 in culture and
accumulate all of these proteins except IGFBP-5, as detected
by ligand blotting of conditioned medium (403, 439 – 443). By
metabolic labeling, they synthesize IGFBP-1 and -2 de novo,
but evidence for IGFBP-3 synthesis is conflicting (403, 444,
445). Although IGFBP-5 mRNA is abundantly expressed (442), no immunoprecipitable IGFBP-5 protein has been de-
tected in conditioned medium (443, 446). These findings sug-
gest that human granulosa cells elaborate an IGFBP-5 pro-
tease as has been reported in the rat (447, 448). Production of each IGFBP species by human luteinizing granulosa cells is uniquely regulated. IGFBP-1 production is
inhibited by FSH, insulin, IGF-I, IGF-II, and the somatostatin
analog octreotide, and increased by LH, EGF, PGs, and phor-
bol ester (199, 396, 439, 449 – 454). The inhibition by insulin
is mediated through its cognate receptor, not the type I IGF
receptor (199). Both IGF-I and IGF-II inhibit IGFBP-1 pro-
duction more potently than insulin (199, 449, 455) and ap-
parently act via the type I IGF receptor. In fact, the concen-
trations of IGFs present in human FF completely inhibit in
vitro granulosa cell IGFBP-1 production. This finding may
explain the production of IGFBP-1 in cultured, but not in
freshly obtained, human granulosa cells (347), as well as the
observation that IGFBP-1 mRNA is not expressed in gran-
ulosa cells of small antral follicles (89). IGFBP-2 production
is negatively regulated by LH/hCG through increased
cAMP; this effect can be reversed by activin-A or interferon-
(IFN- ) (403, 443). IGF-II, but not IGF-I, decreases medium
IGFBP-2, possibly through an action at the type II IGF re-
ceptor (403). In two studies, cAMP agonists promoted the
accumulation of IGFBP-3 (403, 456), while a third found that
FSH did not alter accumulation of immunoreactive IGFBP-3
but decreased its level on ligand blots, consistent with the
action of an IGFBP-3 protease (451). In another study,
IGFBP-3 detected by ligand blotting accumulated in condi-
tioned medium during treatment with IGF peptides but not
insulin, possibly reflecting release of IGFBP-3 from the cell
surface upon binding ligand or protection from proteolysis
(403). IGFBP-4 accumulation is inhibited by LH despite mod-
est stimulation of its mRNA, apparently through elaboration
of an IGFBP-4 protease (see Section IV.C below) (435, 436, 443,
457). IGFBP-5 mRNA expression is stimulated by activin-A
(442). IGFBPs found in human FF may either originate from local production or may reach the FF from an extraovarian source,
such as the liver. FF IGFBPs have been measured both in
antral follicles from cycling women and in hyperstimulated
follicles aspirated for IVF, using both immunoassay and li-
gand blot techniques. FF from cycling women contains im-
munoassayable IGFBP-1, -2, and -3. IGFBP-1 levels range
from 5–32 ng/ml, with levels positively correlated with fol-
licular size and greater in dominant than cohort follicles (348,
446, 458). In one report, FF contained 15 ng/ml IGFBP-2, but
the type of follicle studied was not stated (446). Mean im-
munoassayable IGFBP-3 in estrogen-dominant follicles (2995
ng/ml) was greater than in androgen-dominant follicles
(2352 ng/ml); these levels were indistinguishable from
those in hyperstimulated follicles (348). Immunoassays for
IGFBP-4, -5, and -6 in these follicles have not been reported. By ligand blotting, two distinct IGFBP profiles have been consistently observed in FF from cycling women (446, 459,
460). FF from estrogen-dominant, presumably healthy folli-
cles contains low levels, while FF from androgen-dominant,
presumably atretic follicles contains significantly greater lev-
els of IGFBP-2 and both isoforms of IGFBP-4. The lower level
of IGFBP-4 detectable by ligand blotting in FF from estro- August, 1999 INSULIN-RELATED OVARIAN REGULATORY SYSTEM 549 by on November 4, 2008 edrv.endojournals.org Downloaded from genic compared with androgenic follicles results from the
action of a serine metalloprotease found in estrogenic but not
androgenic FF (see below) (435, 436, 457). An IGFBP-2 pro-
tease was also recently reported in estrogenic FF (436), but
negative regulation of IGFBP-2 gene expression by gonad-
otropins (443) probably plays a more significant role in re-
ducing IGFBP-2 levels in the healthy follicle. By contrast,
IGFBP-3 levels are similar in FF from both types of follicles.
In one study, IGFBP-3 levels in dominant follicles declined
slightly but significantly with advancing follicle size and
cycle day (446). IGFBP-1 has not been detected on ligand
blots of FF from spontaneously cycling women (459). FF obtained after hyperstimulation with menopausal go- nadotropins followed by hCG contains IGFBP-1, -2, and -3,
identified by immunoprecipitation (352, 434, 461). By im-
munoassay, mean IGFBP-1 levels are 90 –160 ng/ml (434, 456,
462, 463), while mean IGFBP-3 levels are consistently near
2400 ng/ml (462, 464, 465), and IGFBP-6 levels are 170 ng/ml
(466). By ligand blotting, IGFBP-1, -2, and -3 are detectable
in FF from hyperstimulated cycles (352, 467). 2. Rodent. IGFBPs 2– 6 have been detected in the rat ovary in
both localization and cell culture studies (468 – 470). Studies
of the cycling ovary revealed that IGFBP-4 and -5 are the
predominant species expressed in granulosa cells of antral
follicles. Both are preferentially localized to atretic follicles,
with IGFBP-4 mRNA signal intensity increasing with the
degree of atresia, and both IGFBP-4 and IGFBP-5 mRNA
expression becoming more widespread in atretic follicles
after the proestrous gonadotropin surge (468 – 470). In
PMSG/hCG-treated rats, each gonadotropin treatment in-
creased IGFBP-4 mRNA expression in small antral follicles,
but no expression was seen in large follicles (471). Cultured
granulosa cells from immature, diethylstilbestrol (DES)-
treated rats secrete intact IGFBP-4 and IGFBP-5 into the me-
dium (447, 448, 472). These cells respond to saturating doses
of FSH by decreasing accumulation of both IGFBP-4 and
IGFBP-5. These effects result from both decreases in mRNA
expression and increases in elaboration of protease activities
that degrade these IGFBPs into smaller, inactive fragments
(448, 460, 473). Paradoxically, low doses of FSH (1–3 ng/ml)
stimulate IGFBP-4 and -5 release (460). GnRH agonists,
which induce follicular atresia (473) and granulosa cell ap-
optosis (474), stimulate basal IGFBP-4 accumulation without
affecting IGFBP-4 protease activity and block the effect of
FSH on both IGFBP-4 production and protease activity (473).
IGF-I stimulates IGFBP-5 accumulation and decreases
IGFBP-5 protease elaboration, while GnRH agonists can op-
pose the effects of FSH on both IGFBP-5 mRNA and protein
expression and IGFBP-5 protease elaboration (447, 475, 476).
Cytokines and growth factors known to block FSH-induced
estradiol production, including TGF- , tumor necrosis factor
(TNF)- , basic fibroblast growth factor, and interleukin-1 ,
stimulate IGFBP-4 (477), suggesting that their effects on FSH
action are due to the IGF-I-sequestering properties of
IGFBP-4. Activin-A can decrease both IGFBP-4 and IGFBP-5
mRNA expression and IGFBP-5 protein accumulation (478). In contrast to the expression of IGFBP-4 and -5 by gran- ulosa cells, IGFBP-2 mRNA expression and production in
culture are unique to theca-interstitial cells in the rat ovary. IGFBP-3 expression is limited to theca-interstitial cells and
vascular and perivascular elements of corpora lutea, sug-
gesting that it plays a role in the vascular control of luteal
regression (468, 479 – 481). IGFBP-6 expression is limited to
the thecal layer (422), while no IGFBP-1 expression has been
detected (448, 468). IGFBP production has also been examined in the mouse ovary. Notable differences from the rat include expression of
IGFBP-2 by granulosa cells (364, 367), negative correlation of
granulosa IGFBP-5 expression in antral follicles with atresia
(367), and the failure of FSH to inhibit accumulation of
IGFBP-4 and -5 in granulosa cell-conditioned medium (364,
367). In the mouse ovary, expression of IGFBP-4 was in-
creased in granulosa cells of histologically atretic follicles and
was correlated with positive staining for the DNA fragmen-
tation characteristic of apoptosis (367). 3. Livestock species. The pig ovary expresses IGFBP-2, -3, -4,
and -5, with granulosa cell IGFBP-2 localized by in situ hy-
bridization to small follicles and IGFBP-4 to large follicles
(482). IGFBP-2 mRNA and protein levels decline with ad-
vancing follicular development (483). Cultured porcine gran-
ulosa cells elaborate both IGFBP-2 and -3, with production of
IGFBP-3 and IGFBP-2 stimulated by IGF-I and decreased by
FSH (484, 485). Granulosa cells from medium-sized follicles
also accumulate IGFBP-4 and -5. IGF-I stimulates, while FSH
inhibits, IGFBP-5 mRNA and protein production. FSH stim-
ulates elaboration of 22-kDa IGFBP-4 (484, 486). In porcine
FF, follicular growth is accompanied by a slight increase in
IGFBP-3 and a decrease in IGFBP-2 and IGFBP-4, as assessed
by ligand blotting (487– 489). While IGFBP-4 and IGFBP-5 are
undetectable in FF from preovulatory follicles, atresia is as-
sociated with a marked increase of intrafollicular levels of
IGFBP-2 and IGFBP-4 (487, 489, 490). In the sheep, IGFBP-4 and -5 expression in healthy follicles is mainly limited to the theca (491– 493). In atretic follicles,
both IGFBP-2 and -5 are more strongly expressed in the
granulosa layer than in healthy follicles, while both IGFBP-2
and -4 are more strongly expressed by the theca (493). FF
content of IGFBP-2 and -4 declines, while IGFBP-3 slightly
increases, with follicle growth. Atresia is associated with
increased content of IGFBP-2, -4, and -5 (424, 493). In the cow, as in the sheep, IGFBP-2, -3, -4, and -5 have been identified in FF by immunoprecipitation. By ligand blotting
and mRNA expression analysis, IGFBP-2 and -4 are more
abundant in estrogen-poor, atretic follicles than in estrogen-
rich, healthy ones (384, 387, 494 – 497). Within the dominant
follicle, an increase in IGF-I and IGF-II with a concomitant
decrease in IGFBP-2 may promote follicular dominance
(388). In summary, since granulosa cells from the pig, sheep, and cow express IGFBP-2, these three livestock species are better
models for the human ovary than is the rat. The large animal
models also permit the study of FF IGFBP content in relation
to follicular functional status. In every species in which such
studies have been reported, atretic follicles contain higher
levels of IGFBPs -2, -4, and/or -5. Additionally, in cell culture
models, gonadotropins universally decrease accumulation
by granulosa cells of these small IGFBPs. These findings
suggest that in a highly conserved mechanism, IGFBPs -2, -4, 550 PORETSKY ET AL. Vol. 20, No. 4 by on November 4, 2008 edrv.endojournals.org Downloaded from and -5 serve as IGF antagonists in follicles destined to un-
dergo atresia, and that gonadotropins may exert their an-
tiatretic action in part through down-regulation of IGFBP
production. By contrast, IGFBP-3 may reach FF from thecal
production or from the circulation; its level in FF is not
affected by gonadotropins or atresia, but rather increases
modestly with follicular maturation. By contrast to the
smaller IGFBPs, IGFBP-3 appears not to function as an IGF
antagonist within the follicle, possibly because it is saturated
with ligand. C. IGFBP proteases in the ovary (Table 2) IGFBP protease activity was first demonstrated for IGFBP-3 in human pregnancy serum (498, 499). Subsequent
reports of IGFBP-3 protease activity in pregnancy serum of
other species (500, 501) were followed by nearly a decade of
discovery of IGFBP proteases, which exist for most of the
IGFBP species in a variety of biological fluids and are pro-
duced and secreted by a variety of cell types (329, 430, 502).
The IGFBP proteases comprise a superfamily that includes
several classes of proteases, including metal-dependent pro-
teases, matrix metalloproteinases, disintegrin metallopro-
teinases, kallikreins, and cathepsins. These molecules likely
represent enzymes with multiple active sites, multimeric
proteins with subunit-specific active sites, or a cascade of
enzymes with different activities. Several IGFBP proteases
have been characterized with regard to their active sites and
cofactor requirements, and the human pregnancy serum
IGFBP-3 protease has been purified and characterized as a
disintegrin metalloproteinase (503). Most IGFBP proteases
are specific for particular binding-protein substrates.
IGFBP-3 is the most susceptible to proteolysis by a variety of
proteases, whereas IGFBP-1 appears to be the most resistant
(504). Sequence analyses of IGFBP cleavage sites suggests
that most proteolysis occurs in nonconserved regions (505). The proteolysis of IGFBPs is likely to be an essential mech- anism in the complex regulation of IGF action. IGFBP pro-
teases partially proteolyze IGFBPs, resulting in lowered af-
finities of the IGFBP fragments for IGF peptides, thus
increasing IGF binding to their receptors. In support of this
concept, inhibitory effects of IGFBPs on IGF-stimulated DNA
synthesis and mitogenesis are reversed in the presence of
IGFBP protease activity in cultured chick embryo fibroblasts
and prostatic epithelial cells, respectively (506, 507). In se-
rum, proteolysis of IGFBP-3 releases IGFs for transport to the
extravascular space, where they are likely bound to other
IGFBPs, which are subsequently cleaved to promote release
of the IGFs for action within the tissue. IGFBP-3 fragments
may act at the cell membrane to augment the stimulatory
effects of IGFs (508). Spatial and temporal regulation of
IGFBP proteases is essential for controlled IGF actions, as
well as the actions of IGFBP fragments. It is remarkable that IGFBP-4 protease activity has been found in the ovaries of all species examined, including the
pig, cow, and sheep. In these livestock species, the patterns
of expression of low mol wt IGFBPs and their proteases in
atretic and growing follicles are similar to those observed in follicles of other species. Likely this finding reflects a con-
served mechanism that has evolved to regulate IGF bioavail-
ability in the ovarian follicle (509 –511). In the next sections,
we will review the IGFBP protease activities that have im-
plications for ovarian function in human and rat ovaries. 1. Human. a. IGFBP-4 protease. IGFBP-4 exists as a nonglycosylated 25-kDa form and a 32- to 34-kDa glycosylated protein. While
some IGFBPs have inhibitory as well as stimulatory effects on
IGF actions, IGFBP-4 appears to have exclusively inhibitory
actions (429). IGFBP-4 mRNA and protein are abundantly
expressed in small antral (androgen-dominant) follicles of
normal and polycystic human ovaries (89, 343). As noted
above, the apparent absence by ligand blotting of IGFBP-4 in
FF from estrogen-dominant, compared with androgen-dom-
inant, follicles (446, 459, 460, 512) was demonstrated to be due
to an IGFBP-4 protease that decreases the affinity of IGFBP-4
for IGFs (457, 513). This protease is a metal-dependent en-
zyme with a pH optimum between 7 and 9 (436, 457), which
is produced by nonluteinizing granulosa cells before the LH
surge as well as by luteinizing granulosa (436, 443, 457, 513).
The degree of proteolysis of IGFBP-4 is inversely propor-
tional to the A:E ratio within the follicle (513). IGFBP-4 pro-
tease activity is stimulated by gonadotropins, IGF-I and -II,
activin-A, and IFN- (435, 443, 513); FSH and IGF-II syner-
gistically stimulate this activity in nonluteinizing granulosa
cells (435). When unsaturated with IGF peptide, IGFBP-3 inhibits pro- teolysis of IGFBP-4, whereas when saturated, it permits
IGFBP-4 proteolysis (514). The implication of this finding is
that in estrogen-dominant follicles, where IGF levels are high
and IGFBP-3 is presumably saturated, IGFBP-4 proteolysis
can increase IGF bioavailability from the pool of IGFs bound
to this binding protein. In contrast, in androgen-dominant
follicles, where IGFBP-3 is presumably unsaturated due to
low levels of IGF production, any IGFBP-4 protease activity
present is inhibited by the unsaturated IGFBP-3. b. IGFBP-3 and IGFBP-2 proteases. IGFBP-3 protease in es- trogen-dominant FF (FFe) obtained at oocyte harvest from
patients undergoing IVF was first demonstrated by Gar-
gosky et al. (465). Iwashita et al. (515) also demonstrated a
protease in FFe that cleaved radiolabeled IGFBP-3 into
smaller fragments, whose activity in medium conditioned by
luteinizing granulosa cells was stimulated by increasing
doses of FSH. A 29-kDa fragment of IGFBP-3 was found in
FF from dominant, compared with small antral, follicles,
consistent with the presence of an IGFBP-3 protease (436,
465). With regard to IGFBP-2, immunoblotting revealed al-
most exclusively a 23-kDa IGFBP-2 fragment in FF from
dominant follicles, compared with nearly exclusively intact
IGFBP-2 and minimal fragments in FF from small cohort
follicles (436). These observations are consistent with an
IGFBP-2 protease in FFe, although specific IGFBP-2 prote-
olysis has not yet been demonstrated in these follicles. FSH
action on luteinizing granulosa cells increases IGFBP-3 im-
munoreactivity in conditioned medium and apparently also
increases IGFBP-3 proteolysis. These effects were found to be
dose-dependent (515). These observations underscore the August, 1999 INSULIN-RELATED OVARIAN REGULATORY SYSTEM 551 by on November 4, 2008 edrv.endojournals.org Downloaded from complexity of the mechanisms underlying control of IGF
bioavailability within the human follicle. c. Thecal and stromal proteases. Limited information is avail- able regarding IGFBP protease in the thecal or stromal com-
partments of the ovary of humans or other species. In human
thecal cell-conditioned medium, LH decreases IGFBP-2, -3,
and -4 levels, but no increase in low molecular weight forms
consistent with proteolysis was seen. Conditioned medium
contains an IGFBP-3 protease, which was partially inhibited
by metal chelators. No difference was observed in theca from
patients with normal or polycystic ovaries (438, 516). In summary, since IGFs are potent stimulators of steroi- dogenesis and follicular growth in the human ovary, their
regulation by IGFBPs and IGFBP proteases is temporally and
spatially related within ovarian tissues. This is likely to pro-
vide timed promotion and inhibition of growth factor action
during periods of follicular development and of limited fol-
licular growth or steroidogenesis, respectively. 2. Rodent. Cultured rat granulosa cells secrete intact IGFBP-4
and IGFBP-5 into the medium (see above). When rat gran-
ulosa cells are cultured with FSH, there is a dose-dependent
decrease in intact IGFBP-4 and an increase in a 17.5-kDa
IGFBP-4 fragment, suggesting the stimulation of an IGFBP-4
protease by FSH (448, 473, 517). This proteolytic activity has
a neutral pH optimum and is inhibited by EDTA, but not by
other protease inhibitors, suggesting its dependence on a
divalent cation (517). Some studies, however, failed to find
IGFBP-4 protease activity in granulosa cell-conditioned me-
dium, regardless of FSH stimulation (447, 518). FSH, but not
IGF-I, also stimulates proteolysis of IGFBP-5. The granulosa-
derived IGFBP-5 protease appears to be a zinc-dependent
metalloprotease of molecular mass greater than 100 kDa,
which is specific for IGFBP-5. The resulting degradation frag-
ments were estimated at 18 and 14 kDa in one study (447) and
19.5 and 17.5 kDa in another (518). Under cell-free conditions,
IGF-I attenuates IGFBP-5 proteolysis, suggesting that bind-
ing to IGF-I may be protective (447, 518). GnRH, which
increases IGFBP-4 and IGFBP-5, does not induce protease
activity for either of these IGFBPs under basal conditions, but
it completely blocks the ability of FSH to inhibit IGFBP-4 and
IGFBP-5 accumulation and stimulate protease activity (473,
476, 518). Since IGFBP-4 and IGFBP-5 are effective inhibitors
of FSH action in rat granulosa cells, regulated production of
their proteases is likely to be important in FSH-dependent
control of follicle growth and development. In summary, IGFBP proteases are produced by granulosa and theca cells at distinct times of follicle development in
ovaries from a variety of species. This conservation of ex-
pression and their regulation by gonadotropins, IGFs, and
other peptides and cytokines underscore the importance of
IGFBP proteases in regulating IGF bioactivity at unique
stages of follicle development. The striking absence of
IGFBP-4 protease in androgen-dominant follicles and the
presence of this enzymatic activity in estrogen-dominant
follicles argue strongly for an important role for the IGF
peptides as co-gonadotropins and for IGFBPs as antigona-
dotropins during follicular growth, steroidogenesis, and
atresia. D. IGFBP actions in the ovary Studies of IGFBP actions in the ovary have largely em- ployed IGFBPs purified from the FF of large animals or
prepared by recombinant DNA technology, with cultured
granulosa cells from DES-primed, immature rats as the tar-
get. When IGFBP-1, -2, -3, or -4 is added to cultured rat
granulosa cells, each can inhibit FSH-stimulated steroido-
genesis (471, 519, 520), while IGFBP-6 is ineffective (422).
Porcine IGFBP-3 and IGFBP-2 inhibit FSH-stimulated E 2 and P release; their lack of efficacy in the presence of IGF-I an-
tiserum or IGF peptide suggests that they act by neutralizing
endogenous IGF-I (471, 519, 521). In this model, IGFBPs also
decrease mitosis and cAMP generation. Human IGFBP-1, -2,
-3, and -4 all similarly decrease FSH-stimulated P output
(471, 522); human IGFBP-6 does not, possibly because of its
lower affinity for IGF-I, the principal IGF produced by rat
granulosa cells, compared with IGF-II (422). The physiolog-
ical relevance of IGFBP actions on the granulosa is strongly
suggested by in vitro studies showing the greater potency of
IGF peptide analogs that do not bind to IGFBPs, compared
with the native peptides, only under conditions of high-
medium IGFBP levels (522). These observations have led to
the conclusion that intrinsic IGF-I is an obligatory mediator
of FSH-induced E 2 and P production by rat granulosa. Ad- ditional in vivo evidence for the biological relevance of IGFBP
action on the ovary comes from studies showing that injec-
tion of IGFBP-3 into the rat ovarian bursa or introduction of
IGFBP-3 into the in vitro perfusate of rabbit ovaries each can
decrease the rate of follicular rupture at ovulation (523, 524),
and from the recent observation that transgenic mice over-
expressing IGFBP-1 have reduced numbers of ovulations per
estrous cycle (525). IGFBP actions on human granulosa cells are similar to those on cells from the rat. In cultured granulosa-luteal cells,
IGFBP-1 and -3 decrease IGF-I-stimulated E 2 production; IGFBP-1 also decreases IGF-I-stimulated mitosis (390, 399,
513, 526, 527). IGFBP-3 fails to inhibit the steroidogenic effect
of des(1–3)IGF-I, an analog that does not bind to IGFBPs. In
granulosa cells obtained from women during unstimulated
cycles, IGFBP-1 and IGFBP-3 inhibit IGF-I-stimulated E 2 and P production (399). Recombinant human (rh) IGFBP-4 inhibits IGF-stimu- lated E 2 production by human granulosa cells (431, 435, 512, 513). Iwashita et al. (513) employed luteinizing gran-
ulosa cells, whereas Chandrasekher et al. (435) and Mason
et al. (512) used nonluteinizing granulosa cells, showing
that rhIGFBP-4 can inhibit both IGF-II- and FSH-stimu-
lated E 2 production. This inhibition exceeded 80%, while in similar experiments IGFBP-2 or IGFBP-3 inhibited gran-
ulosa cell steroidogenesis by only about 20% (512). In
contrast to the inhibitory effects of intact rhIGFBP-4 on E 2 production, addition of proteolyzed IGFBP-4 was without
effect (513). These findings support an important role for
IGFBP-4 and IGFBP-4 protease in the regulation of follic-
ular steroidogenesis in the human ovary. IGFBP-4 inhibits
FSH-stimulated E 2 production in the absence of added IGF peptide or in the presence of type I IGF receptor antibody,
suggesting either IGF-independent action or antagonism
of a locally produced IGF (431, 435, 512). Nevertheless, 552 PORETSKY ET AL. Vol. 20, No. 4 by on November 4, 2008 edrv.endojournals.org Downloaded from IGFBPs consistently display actions on cultured ovarian
tissues opposite to those of IGF peptides and gonadotro-
pins, suggesting that an excess of IGFBPs can be antigo-
nadotropic (409, 528) and result in either follicular arrest
(as in PCOS) or atresia. In addition to regulating follicular differentiation and mat- uration, IGFs and IGFBPs also likely play a role in regulating
apoptosis of granulosa cells, which is associated with follic-
ular atresia (201). In a rat antral follicle culture system, both
gonadotropins and IGF-I can prevent the apoptosis of gran-
ulosa cells that occurs spontaneously in serum-free medium,
and IGFBP-3 reverses the protection from apoptosis afforded
by hCG, FSH, GH, and IGF-I (200, 529). The restriction of
IGFBP-4 expression in the mouse follicle to histochemically
apoptotic granulosa cells (367) also supports a role for
IGFBPs in promoting follicular atresia in vivo. E. Role of IGFBPs in follicular development and atresia In the growing estrogen-dominant follicle, a number of mechanisms have evolved to increase IGF peptide bioavail-
ability and thereby amplify granulosa responsiveness to the
growth-promoting, steroidogenesis-promoting, and anti-
apoptotic actions of FSH (Fig. 6). These include up-regulation
of IGF receptors by gonadotropins and, in the rat, by estro-
gens (90, 530, 531); increase in IGF expression by gonado-
tropins (345, 532); inhibition by IGFs and gonadotropins of
inhibitory IGFBP synthesis (403); and stimulation by gonad-
otropins and IGF-II of IGFBP protease activity (435, 513). The
net result is maximum bioavailability of IGF peptides. In
contrast, in the androgen-dominant follicle that is arrested in
development or destined for atresia, these mechanisms are
reversed (Fig. 6): FSH receptor numbers are low; IGF ex-
pression is almost undetectable; there is abundant expression
of inhibitory IGFBPs (IGFBP-2 and IGFBP-4); and there is
minimal detectable IGFBP protease activity. The net result is
that aromatase is not induced, and thus precursor androgen persists in these follicles, in association with developmental
arrest or atresia. The question remains, however, whether relative IGFBP expression is causally involved in selection and maturation
of the dominant follicle. The study of IGFBPs in PCOS (see
Section V) had been anticipated to shed some light on their
role in follicular maturation in this disorder. Women with
PCOS appear to have a defect in antral follicular maturation,
but the cause of this defect has not been identified. Levels of
IGFBPs in FF and IGFBP mRNA expression in follicular cells
of the PCOS ovary are similar to those in small antral (largely
atretic) follicles in normal women (89, 347, 533, 534). This
appears to exclude a unique defect in IGFBP regulation in the
ovary as a cause of the PCOS follicular maturation defect.
Rather, in both the PCOS and normal ovary, the challenge is
to explain how FSH can be successful in suppressing IGFBP
production in one follicle (destined for dominance) while
failing to do so in others (cohort follicles destined for atresia). F. Summary The high levels of expression of IGFs and low levels of expression of inhibitory IGFBPs in healthy follicles, and the
reverse in atretic follicles, suggest that the level of bioavail-
able IGFs may play a role in regulating follicular growth,
steroidogenesis, and apoptosis. IGFBPs and IGFBP proteases
could thus assume importance in determining follicular des-
tiny, since they can modulate the bioactivity of members of
the IGF family. V. Polycystic Ovary Syndrome (PCOS) A. Clinical features PCOS is a disorder of unknown, probably heterogeneous, etiology, characterized by chronic anovulation, biochemical
and/or clinical evidence of hyperandrogenism, and en-
larged, polycystic ovaries (535, 536). When first described by
Stein and Leventhal (537) in 1935, the syndrome was defined
by ovarian enlargement and multiple small cysts, in associ-
ation with amenorrhea and hirsutism. PCOS affects between
5–10% of women of reproductive age (538, 539), and the onset
of clinical manifestations often occurs at the time of puberty
(191). In recent years, varying definitions of this syndrome
have been used in studies of this disorder, with some inves-
tigators requiring polycystic ovaries on ultrasound for in-
clusion, and others requiring an elevation of serum LH or
LH:FSH ratio (540). A consensus definition of PCOS was
reached in 1990 under NIH auspices, which requires only
hyperandrogenism of ovarian origin and oligomenorrhea or
amenorrhea, with exclusion of other specific disorders such
as steroid 21-hydroxylase deficiency (541). Other endocrine
abnormalities that are inconsistently present in women with
PCOS include obesity, peripheral insulin resistance and hy-
perinsulinemia, and elevations of serum PRL or DHEA-sul-
fate. Phenotypic differences among PCOS study populations
may reflect underlying genetic differences in etiology or
pathophysiology or in peripheral manifestations such as hir-
sutism (542, 543). Differences in diagnostic selection criteria
can make comparison of studies on PCOS difficult. F IG . 6. Model of IGF, IGFBP, and IGFBP protease actions in human ovary. In the estrogen-dominant, healthy growing follicle (shown at
top left), granulosa cell IGF-II production increases, synergizing with
FSH. IGF-II action is amplified by decreased synthesis and increased
proteolysis of IGFBPs. In the androgen-dominant follicle (shown at
top right), both increased IGFBP synthesis and decreased IGFBP
proteolysis contribute to decreased FSH and IGF-II action on the
granulosa, resulting in atresia or developmental arrest. August, 1999 INSULIN-RELATED OVARIAN REGULATORY SYSTEM 553 by on November 4, 2008 edrv.endojournals.org Downloaded from PCOS is perhaps the most common disorder in which the association between insulin resistance and ovarian function
appears to be important. Since several comprehensive re-
views on this subject are available (26, 27, 140, 535), we focus
herein on the controversial issues related to the pathogenesis
of PCOS and the changes in the insulin-related ovarian reg-
ulatory system observed in PCOS. In the following section,
we will review recent studies that have evaluated the use of
inhibitors of insulin secretion and insulin-sensitizing agents
in the therapy of PCOS. B. Theories of pathogenesis Determining the etiology or etiologies of PCOS has proven elusive. It was recognized as early as 1980 by Yen (544) that
in PCOS a number of endocrine abnormalities perpetuate
themselves in what has been described as a “vicious cycle.”
These include abnormal gonadotropin secretion, with excess
circulating LH and low, tonic FSH levels; hypersecretion by
ovarian thecal and stromal compartments of androgens,
which were viewed as both disrupting follicular maturation
and providing substrate for peripheral aromatization to es-
trogens in adipose and other sites; and negative feedback of
this tonic estrogen production on the pituitary to decrease
FSH secretion and thus trophic support of the granulosa cell
(544). The vicious cycle concept was further supported by
studies suggesting that normal ovulatory function can occur
after disruption of this cycle, e.g., by ovarian wedge resection
or cautery or during recovery from GnRHa-induced sup-
pression (545–548). The vicious cycle concept does not, how-
ever, provide an explanation of how the abnormalities be-
come established. A number of endocrine disorders can
produce similar anovulatory, hyperandrogenic states, such
as functional or drug-induced hyperprolactinemia (549, 550)
and adult-onset congenital adrenal hyperplasia resulting
from 21-hydroxylase deficiency (551, 552). The primary ab-
normality in PCOS has been proposed to be of central, ovar-
ian, adrenal, or peripheral metabolic origin. These theories
will be briefly reviewed below. 1. Central hypothesis. Abnormalities in LH-secretory pattern
and its regulation have been observed in PCOS. Women with
PCOS often have both increased LH pulse amplitude and
frequency, compared with ovulatory controls (168, 553–555).
This results in increased or disordered LH secretion and may
lead to an elevated serum LH:FSH ratio. These central al-
terations may be mediated by the altered steroid milieu of
PCOS rather than being primary, since during recovery from
GnRHa suppression no difference was seen between PCOS
and normal women in the recovery

Comments