| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Obstetrics and Gynaecology, British Columbia Womens Hospital, University of British Columbia, Vancouver, British Columbia, V6H 3V5, Canada
| Abstract |
|---|
|
|
|---|
I. Introduction
II. Embryonic Development
III. OSE in the Adult
A. Structure
B. Functions
C. Differentiation
IV. Neoplastic Progression of OSE
A. Epidemiology and etiology of the epithelial ovarian carcinomas
B. OSE in women with histories of familial ovarian cancer
C. Epithelial ovarian carcinomas
V. OSE in Culture
A. Culture methods
B. Properties
C. Three-dimensional culture systems
D. Extension of the life span of surface epithelial cells
E. Variation in OSE characteristics among species
F. Culture of OSE from women with family histories of ovarian cancer
VI. Regulation by Hormones, Growth Factors, and Cytokines
A. OSE
B. Ovarian carcinomas
VII. Concluding Remarks
| I. Introduction |
|---|
|
|
|---|
| II. Embryonic Development |
|---|
|
|
|---|
|
| III. OSE in the Adult |
|---|
|
|
|---|
|
The OSE is separated from the ovarian stroma by a basement membrane
and, underneath, by a dense collagenous connective tissue layer, the
tunica albuginea, which is responsible for the whitish color of the
ovary. It is thinner and less resilient than the tunica albuginea in
the testis, but likely provides a partial barrier to the diffusion of
bioactive agents between the ovarian stroma and OSE. The OSE differs
from all other epithelia by its tenuous attachment to its basement
membrane, from which it is easily detached by mechanical means. Until
recently, the resulting loss of OSE in surgical specimens was
responsible for the widely held opinion that OSE is frequently absent
in ovaries of older women. Whether this loose attachment has any
physiological consequences is not known. With age, the human ovary
assumes increasingly irregular contours and forms OSE-lined surface
invaginations (clefts) and epithelial inclusion cysts in the ovarian
cortex. It has been suggested that the squamous and cuboidal forms of
OSE cells on the ovarian surface represent cell groups that,
respectively, have or have not undergone postovulatory proliferation
(46). In addition, OSE cells tend to assume columnar shapes, especially
within clefts and inclusion cysts. Whether these shape changes are the
result of crowding or whether they reflect genetically determined
metaplastic changes is not always clear, but they may be derived by
either process. The importance of surface invaginations and inclusion
cysts lies in the propensity of the OSE in these regions to undergo
metaplastic changes, i.e., to take on phenotypic
characteristics of Mullerian (usually tubal) epithelium, which include
columnar cell shapes and several markers found in ovarian neoplasms,
including CA125 and E-cadherin (6, 31, 40, 47, 48, 49). Furthermore,
OSE-lined clefts and inclusion cysts, rather than surface OSE, are not
only common sites of benign metaplasia but also of early neoplastic
progression (50, 51, 52). It has been suggested that the inclusion cysts
form from OSE fragments that are trapped in or near ruptured follicles
at the time of ovulation (53, 54). However, Scully (52) reported that
inclusion cysts are more numerous in ovaries of multiparous women than
in nulliparous women who ovulate more frequently, and the cysts are
particularly numerous in women with polycystic ovarian disease, a
condition that is characterized by anovulation or infrequent ovulation.
He proposed as an alternative that inclusion cysts arise through
inflammatory adhesions of surface OSE which becomes apposed at sites of
surface invaginations, combined with localized stromal proliferation.
There is currently no definitive explanation for the predilection of
inclusion cysts as preferred sites of neoplastic progression of OSE,
but these preferential locations strongly suggest the presence of
specific tumor-promoting microenvironmental factors in these sites. Two
different scenarios, which are not mutually exclusive, can be
envisaged: 1) OSE within inclusion cysts is not separated from
underlying stroma by the tunica albuginea. Therefore this OSE likely
has more access to stromally derived growth factors and cytokines as
well as to blood-borne bioactive agents that may promote neoplastic
progression. This hypothesis is supported by the observation that, in
inclusion cysts located near the ovarian surface, metaplastic and
dysplastic changes tend to be more pronounced on the side near the
stroma than on the side adjacent to the tunica albuginea (51, 52). 2)
Neoplastic progression in OSE-lined cysts and clefts may be promoted by
autocrine mechanisms through OSE-derived cytokines and hormones, since
these agents may accumulate to bioactive levels in such confined sites
but not on the ovarian surface where they diffuse into the pelvic
cavity. The hypothesis that these factors participate in autocrine
loops is supported by the capacity of normal OSE to secrete bioactive
cytokines including interleukin (IL)-1 and IL-6 (55) and by reports
that IL-1 and IL-6 enhance the proliferation of ovarian carcinomas
(56), that IL-1 causes changes in gene expression including the
induction of tumor necrosis factor (TNF)-
, which is a mitogen for
OSE (57, 58), and that human CG (hCG) is produced by normal and
neoplastic OSE (47) and is also mitogenic for rabbit OSE cells (59).
Finally, the proliferative response to cytokines of cervical cells
(which are developmentally related to OSE) changes with immortalization
so that the immortalized cells acquire a selective advantage over
normal cells (60). Within inclusion cysts, such cytokines and hormones
might act as immediate autocrine growth regulators, or they might cause
secondary changes in gene expression that promote neoplasia.
B. Functions
The OSE transports materials to and from the peritoneal cavity and
takes part in the cyclical ovulatory ruptures and repair. Most of these
functions vary with the reproductive cycle and thus are likely to be
hormone dependent (1, 6, 59). It is well established that OSE must
proliferate to repair ovulatory defects in the ovarian surface, and
Osterholzer et al. (59) demonstrated directly that in rabbit
ovaries, this proliferative activity is both localized to the vicinity
of the ovulatory site and peaks at, and immediately after, the time of
ovulation. Several reports, based on electron microscopy and
histochemistry, have suggested that the OSE contains lysosome-like
inclusions and produces proteolytic enzymes, which may contribute to
follicular rupture (61). These reports were supported by direct
observations of protease secretion by cultured OSE (29). However, this
concept has been questioned because of inconsistencies in the timing of
the appearance of the dense lysosome-like granules in the OSE, their
biochemical nature, and the observation that follicles denuded of
overlying OSE can also rupture (reviewed in Ref. 62). Furthermore,
electron microscopy in various species has revealed that OSE cells
degenerate and slough off the follicular surface shortly before
ovulatory rupture. There is evidence that this cyclic, localized loss
of OSE near the time of ovulation is due to apoptosis that is induced
by prostaglandins (63, 64) and perhaps mediated by the Fas antigen (65, 66). It is possible that, as the tunica albuginea in the area of the
stigma thins and ultimately disappears before ovulation, the OSE in
this region is exposed to stromal influences that induce apoptosis.
However, the possibility cannot be ruled out that the OSE alters the
tunica albuginea and underlying stroma in the area of incipient
ovulation just before its disappearance. The proteolytic capacity of
OSE might contribute to the remodeling, as well as the breakdown, of
the ovarian cortex. OSE likely also takes part in the restoration of
the ovarian cortex by the synthesis of both epithelial and connective
tissue-type components of the extracellular matrix (ECM) (27, 29, 67)
and by its contractile activity, which resembles the contractile
capacity exhibited by connective tissue fibroblasts during wound
healing (68). Like fibroblasts, which convert to myofibroblasts when
engaged in tissue repair, OSE cells in culture contain smooth muscle
actin (our unpublished observations). This is in keeping with
their dual epithelio-mesenchymal phenotype and with the proposition
that OSE cells, like many other cell types, acquire a regenerative
rather than stationary phenotype when they are explanted into culture.
Contraction by OSE cells may also play a role in the shrinkage of the
ovaries that occurs with age and results in their typical convoluted
shape and the formation of the OSE-lined clefts and inclusion cysts.
C. Differentiation
Normal OSE covering a nonovulating ovary is a stationary
mesothelium with both epithelial and mesenchymal characteristics. In
contrast to mesothelia elsewhere, OSE retains the capacity to alter its
state of differentiation along pathways leading either to stromal or to
ectopic (aberrant) epithelial phenotypes. In response to stimuli that
initiate a regenerative (repair) response, such as ovulatory rupture
in vivo or explantation into culture, OSE cells assume
phenotypic characteristics of stromal cells. Alternatively, OSE
acquires complex epithelial characteristics of the Mullerian
duct-derived epithelia, i.e., of the oviduct, endometrium,
and endocervix, when it undergoes metaplasia, benign tumor formation,
and neoplastic progression. Together, these characteristics show that
the differentiation of OSE is not as firmly determined as in other
adult epithelia and that OSE is closer to its pleuripotential
mesodermal embryonic precursor form than other celomic epithelial
derivatives.
Normal stationary OSE has no known tissue-specific differentiation markers. In situ, it can be distinguished from extraovarian mesothelium by the lack of CA125 (20) and by the differential expression of mucin, cilia, 17ß-hydroxysteroid dehydrogenase, and several antigenic markers (5, 6, 47, 49, 69, 70). It has classical epithelial features, which include desmosomes, tight junctions, basement membrane, keratin, and apical microvilli, but other aspects of epithelial differentiation are less defined. For example, E-cadherin and CA125 in human OSE are rare while both markers occur in oviductal and endometrial epithelium, and CA125 is also secreted by extraovarian pelvic mesothelium and by abdominal and pleural peritoneum (1, 6, 20, 69, 70). OSE cells also constitutively coexpress keratin with vimentin, which is a mesenchymal intermediate filament, expressed by most epithelial cells only in response to wounding, explantation into culture, or pathological conditions (71, 72, 73). Expression of the connective tissue collagen types I and III has been shown in cultured OSE but not in situ (27).
During postovulatory repair and in culture (see Section IV)
OSE cells have the ability to modulate to a fibroblast-like form that
reflects their close developmental relationship to ovarian stromal
cells. The exact mechanisms regulating this conversion have not been
defined. However, as shown later in this review, epidermal growth
factor (EGF), collagen substrata, and ascorbate are all conducive to
epithelio-mesenchymal conversion of OSE in culture. In addition,
transforming growth factor (TGF)-ß, which is an autocrine regulator
of OSE growth (74), causes epithelio-mesenchymal conversion in a number
of epithelial cell types (75). Similar epithelio-mesenchymal
conversions occur in vivo in mesodermally derived cell types
closely related to OSE, such as pleural mesothelial cells responding to
injury (76) and the cells of the developing Mullerian duct during
regression in response to Mullerian inhibiting substance (77). This
capacity of OSE to undergo epithelio-mesenchymal conversion likely
confers advantages during the postovulatory repair of the ovarian
surface: it increases motility, alters proliferative responses and
capacities to modify ECM, and renders the cells contractile (see
below). Epithelio-mesenchymal conversion might also function as a
homeostatic mechanism to accommodate OSE cells that become trapped
within the ovary at ovulation, to allow them to become incorporated
into the ovarian stroma as stromal fibroblasts. As a related
hypothesis, an inability to undergo epithelio-mesenchymal conversion
would preserve the epithelial forms within the ovarian stroma, which
could lead to OSE cell aggregation and subsequent inclusion cyst
formation (Fig. 3
). Factors that have
been shown in culture to enhance epithelio-mesenchymal conversion of
OSE include EGF (16), ascorbate (our unpublished data), and
growth in collagen gels and other three-dimensional matrices (68, 78)
(see Section V). It is important to note that OSE at the
site of ovulatory rupture is exposed to all these influences. In
contrast to epithelio-mesenchymal conversion, which is part of normal
OSE physiology, the differentiation of metaplastic and neoplastic OSE
along the lines of Mullerian duct-derived epithelia is clearly a
pathological process, based on complex epigenetic and genetic changes
that will be discussed briefly in Section IV.
|
| IV. Neoplastic Progression of OSE |
|---|
|
|
|---|
The etiology of the epithelial ovarian carcinomas is poorly understood. Over the years, environmental agents that have been implicated but not proven to play a role include diet, talc, industrial pollutants, smoking, asbestos, and infectious agents (7). Epidemiological studies point to possible racial and geographic, social, and hormonal causative factors (7, 81, 82, 83). There is convincing evidence that nulliparity and, probably, hyperovulation treatment for infertility increase the risk of ovarian cancer, while oral contraceptives and pregnancies are protective. These observations support the hypothesis, first proposed by Fathalla in 1971 (149) and subsequently supported by epidemiological and experimental data (84, 85; reviewed in Ref. 8), that frequent ovulation contributes to increased risk because the repeated rupture and repair of the OSE at the sites of ovulation provide an opportunity for genetic aberrations. Recently, it has been suggested that inflammation may be a contributing factor in ovarian cancer development, because tubal ligation and hysterectomies act as protective factors, perhaps by preventing passage of environmental initiators of inflammation (86). Another major known risk factor is a strong family history of ovarian cancer, which accounts for 510% of cases.
B. OSE in women with histories of familial ovarian cancer
At present, a strong family history of ovarian cancer is the most
important and best-defined risk factor for development of this disease,
and it is associated with 510% of ovarian epithelial carcinomas. The
risk increases from 1.4% in the general population to 5% for women
with one first-degree relative and to 8% for women with two
first-degree relatives affected (first-degree relatives include
parents, siblings, and children, while second-degree relatives include
grandparents, uncles, aunts, cousins, and grandchildren). There is also
a strong association with familial breast cancer, and a lesser
association with familial cancers of the colon and endometrium. Three
hereditary ovarian cancer syndromes with autosomal dominance (reviewed
in Ref. 87) are listed below.
1. Hereditary site-specific ovarian cancer, where a family history of ovarian cancer only is associated with an overall 3.6-fold increase in risk. No specific gene responsible for this syndrome has been identified.
2. Hereditary nonpolyposis colon cancer/ovarian cancer (Lynch Syndrome II or HNPCC), where ovarian cancer occurs in families that also have a high incidence of carcinomas of the colon and endometrium. It is associated with mutations in the DNA mismatch repair genes hMSH1, hMSH2, hPMS1, and hPMS2 (88). In this syndrome, the increase in risk has not been defined.
3. Hereditary breast/ovarian cancer. There is a 50% increase in ovarian cancer risk among women with family histories of breast cancer and a similar increase in breast cancer risk among women with family histories of ovarian cancer. Germline mutations in two genes involved in this syndrome, BRCA1 and BRCA2, appear to be responsible for a high proportion of cancers in women with familial cancer histories. The BRCA1 and BRCA2 proteins regulate DNA damage responses (89) and have been defined as tumor suppressor genes. BRCA1, in particular, plays a major role in ovarian cancer susceptibility (90). Intensive screening for BRCA1 mutations is ongoing but the large size of the gene and the great variety of different mutations that have been found complicate screening and risk predictions (91). The observation that BRCA1 and BRCA2 germline mutations cause increases in cancer incidence predominantly in the breast, ovary, and prostate, although they are present in all tissues, points to interrelationships with hormonal influences. Interactions between BRCA1 and estrogen as well as PRL have indeed been reported in cancer cells (92, 93, 94), but there seems to be no information available on similar interactions in normal OSE. Importantly, not all of the carriers of these predisposing mutations develop ovarian cancer, which suggests a role for interactions with other, as yet unidentified, genetic and epigenetic influences.
There have been several contradictory reports on the occurrence of histological changes in the OSE of overtly normal ovaries that were removed by prophylactic oophorectomy from healthy women with histories of familial ovarian cancer. A nonblind study (95) demonstrated increased papillomatosis and pseudostratification of the OSE, as well as an increase in inclusion cysts and invaginations in ovaries from women with familial ovarian cancer. In another blind study, only nuclear changes were observed in the OSE of such women (96), while in two other reports no significant differences were observed (97, 98). Thus, it is still not clear whether, in situ, overtly normal OSE from women with family histories of ovarian cancer is distinct at the phenotypic level.
C. Epithelial ovarian carcinomas
1. Pathology. Histopathologically and immunocytochemically,
ovarian carcinomas are among the most complex of all human malignancies
(99, 100). One of the most unusual aspects of ovarian carcinogenesis is
the change in differentiation that accompanies neoplastic progression.
As discussed above, OSE is a simple, rather primitive epithelium with
some stromal features, but as it progresses to malignancy it loses its
stromal characteristics and acquires the characteristics of the
Mullerian duct-derived epithelia, i.e., the oviduct,
endometrium, and uterine cervix. This aberrant differentiation occurs
in such a high proportion of ovarian carcinomas that it serves as the
basis for the classification of a high proportion of these cancers as
serous (fallopian tube-like), endometrioid (endometrium-like), and
mucinous (endocervical-like) adenocarcinomas (Fig. 4
). Serous adenocarcinomas comprise
approximately 80% of all epithelial ovarian cancers. Among the less
common forms are clear cell carcinomas that express features resembling
mesonephros. It has also been proposed that at least some endometrioid
carcinomas may arise in endometriotic lesions derived from endometrial
implants (101), and that some mucinous ovarian adenocarcinomas may
actually be metastases of gastrointestinal malignancies because the
mucus in these lesions is of the gastrointestinal rather than the
endocervical variety (102).
|
|
Histopathologically detectable early malignant changes occur more
frequently in OSE-lined clefts and inclusion cysts (Fig. 2
) than on the
ovarian surface that faces the pelvic cavity. The evidence for
inclusion cysts as the preferred sites of ovarian carcinogenesis was
reviewed by Scully (51, 52): 1) Most early carcinomas appear to be
confined within the ovary without involvement of its surface; 2) tubal
metaplasia is 10 times more common in epithelial inclusion cysts than
on the ovarian surface; 3) inclusion cysts are significantly more
numerous and the OSE lining them is 23 times more often metaplastic
in women with contralateral epithelial ovarian tumors than in women
without such cancers (105); 4) several ovarian carcinoma tumor markers
(e.g., CA125, CA199) are significantly more common in the
epithelium of epithelial inclusion cysts than in the surface epithelium
itself (20, 47, 106, 107). The localization of early malignant changes
in crypts and cysts has given rise to speculations that neoplastic
progression may be promoted by the particular microenvironment to which
preneoplastic OSE is exposed within these confined spaces.
2. Genetic changes. The genetic basis of the epithelial ovarian carcinomas is too complex to be reviewed in detail here, but numerous excellent reviews exist on this subject. In brief, amplification, altered expression, and mutations in a number of oncogenes and tumor suppressor genes play a role in the development of ovarian epithelial neoplasms. Oncogenes that are frequently overexpressed or amplified in ovarian carcinomas include cMYC, particularly in serous adenocarcinomas (108); KRAS, especially in mucinous carcinomas that may exhibit enteric mucinous differentiation (109); and ERBB2, EGF-R, and cFMS (the receptor for colony-stimulating factor 1), all of which are associated with a poor prognosis (110, 111, 112). Recently, phosphatidyl inositol 3 kinase (PI3K) and its downstream effector AKT2 were also shown to be amplified in a significant proportion of ovarian carcinomas (113, 114). Among tumor suppressor genes, p53 is mutated in about 50% of late-stage tumors but rarely in low-stage tumors and borderline lesions (115), and the PI3K inhibitor PTEN is mutated in a significant proportion of endometrioid ovarian carcinomas (116). As mentioned in Section IV.B., mutations in the tumor suppressor genes BRCA1 and BRCA2 appear to form the basis for most cases of familial ovarian cancer. The expression of a recently described tumor suppressor gene, NOEY2 (ARHI), is decreased specifically in carcinomas of the ovary and breast (117).
The epidemiology, histopathology, and clinical course of OSE-derived ovarian carcinomas differ profoundly from those of the mesotheliomas, which arise in extraovarian mesothelium, e.g., in response to asbestos exposure, and lack Mullerian phenotypes. This difference reflects, among other factors, the different developmental histories of these two components of the pelvic peritoneum, which may include inductive signals emanating from the ovary and acting on the developing OSE (2, 6, 26).
| V. OSE in Culture |
|---|
|
|
|---|
|
B. Properties
1. Differentiation. Cultured OSE is highly responsive to
environmental influences. Over several passages under standard culture
conditions, freshly explanted OSE cells respond to the culture
environment by modulating from an epithelial to a more mesenchymal
phenotype (Table 1
). Immediately upon
explantation into primary culture they retain mesenchymal markers that
are present in vivo, such as vimentin, and acquire
additional mesenchymal characteristics, such as collagen type III
secretion. They rapidly lose some epithelial differentiation markers,
including villin and desmoplakin, but retain others, e.g.,
keratin, for longer periods. With passages in culture, the cells may
assume a more definitive fibroblast-like phenotype as indicated by a
change to anterior-posterior polarity, reduced intercellular cohesion,
gel contraction, increased secretion of collagen types I and III, and
loss of the epithelial marker keratin (27, 78, 123). Such
epithelio-mesenchymal conversion is more consistent and prominent in
three-dimensional than in two-dimensional culture (29, 78). It is
enhanced by epithelial growth factor (16), collagen substrata (29), and
ascorbate (our unpublished data). It varies widely in frequency
between laboratories and within laboratories with time. The reasons for
this variation and the precise mechanisms underlying the mesenchymal
conversion of OSE have not been defined, but they most likely depend on
as yet undefined serum factors. Similar epithelio-mesenchymal
conversions occur in the culture of other mesodermally derived
epithelia (reviewed in Refs. 87, 124). Generally, cells respond to
explantation into culture as they would to wounding and undergo changes
in phenotype and in gene expression that are similar to those that
occur in regenerative responses. In analogy, the response of OSE cells
to explantation into culture likely mimics their normal response to
ovulatory rupture. Thus, the phenotype observed in culture should
perhaps be compared with that of regenerating OSE rather than to the
phenotype of stationary OSE covering a nonovulating ovary.
|
3. Intercellular adhesion. Similar to its in vivo phenotype (31, 32), intercellular contact of cultured OSE is maintained by N-cadherin, which is expressed constitutively while Ecadherin is expressed only conditionally, when OSE cell shapes approach those of metaplastic epithelium (36). In contrast to the human, cultured rat OSE expresses E-cadherin consistently (128). N-cadherin-mediated adhesion appears to have an antiapoptotic effect in OSE of the rat (129), but whether it has a similar function in the human is not known. In general, expression of N-cadherin alone or of N- and E-cadherin together characterize adhesive mechanisms of mesodermally derived tissues (reviewed in Ref. 36).
C. Three-dimensional culture systems
Pathological changes in OSE, including neoplastic conversion and
endometriosis, often involve three-dimensional formations such as
OSE-lined clefts and cysts. To reproduce OSE growth in such confined
spaces, several three-dimensional culture systems have been
investigated: 1) rat tail tendon-derived collagen gel that is rich in
collagen type I and permits differentiation of many cell types; 2) a
rat OSE-derived matrix plus collagen gel to produce OSE
"organoids"; 3) Matrigel (Collaborative Research, Bedford,
MA), a mouse yolk sac tumor-derived basement membrane substitute
rich in laminin and other basement membrane components (29, 68); and 4)
Spongostan (Health Design Industries, Rochester NY), a pig skin-derived
denatured collagenous sponge that provides a rigid skeleton (78). In
collagen gel cultures, human OSE cells converted to a mesenchymal form,
dispersed in the gel in a manner resembling connective tissue
fibroblasts, and then remained stationary and eventually died (29).
However, if cocultured with endometrial stromal cells in the presence
of 17ß-estradiol, OSE were reported to form structures composed of
monolayered polarized cells surrounding lumina and expressing markers
of endometrial cells. This system may represent an experimental model
for OSE-derived endometriosis (130, 131). When cultured on the rat
OSE-derived matrix plus collagen gel, the OSE cells again converted to
a mesenchymal form and dispersed and then contracted the relatively
loose matrix into smaller, denser structures (68). Such contractile
function is generally considered as characteristic of fibroblasts in
the process of wound healing. On Matrigel, OSE cells aggregated into
solid cell clumps. Depending on the lot of Matrigel, the cells showed a
varying propensity to lyse the matrix and eventually form monolayers on
the underlying plastic (29). This variation may have depended on growth
factor contaminants known to occur in Matrigel. In their ability to
lyse this matrix, these presumably normal cells resembled cancer cells,
which are commonly assumed to be the only cells capable of invading
Matrigel. In Spongostan, cells were grown for several weeks until they
filled the sponges. In contrast to ovarian cancer cells, which form
epithelial linings along the sponge spicules, human OSE cells under
these conditions again underwent mesenchymal conversion: they assumed
morphological and functional characteristics of stromal cells as they
dispersed in intercellular spaces, took on fibroblast-like shapes, and
secreted ECM (78). Thus, in all three-dimensional systems except for
Matrigel, OSE cells converted to mesenchymal phenotypes.
D. Extension of the life span of surface epithelial cells
One of the problems in human OSE research is the small number and
short life span of cells obtained at surgery. To alleviate this
problem, "immortalizing" genes such as SV40 large T antigen (Tag)
(132) and the HPV genes P6 and P7 (123, 133, 134) have been introduced
into OSE. Expression of these genes does not truly immortalize human
OSE cell lines in that their population-doubling capacity is greatly
extended but not infinite; however, the lines provide sufficiently
large cell numbers for molecular studies. One advantage of these lines
is that they tend to retain some, although not all, of the
tissue-specific properties of the cells from which they are derived.
For example, many of these lines retain keratin, and most, if not all
of them, continue to express N-cadherin and lack E-cadherin (in common
with normal, and in contrast to neoplastic OSE). Although such lines
are nontumorigenic in SCID mice (18), their growth controls are
profoundly disturbed, which confer on them properties of neoplastic
cells such as genetic instability, increased saturation density
reduced serum requirements, and variable degrees of anchorage
independence. Tag and E6/E7 inactivate the tumor suppressor genes p53
and p105RB (135, 136). Importantly, 3080% of epithelial ovarian
carcinomas have p53 mutations that disrupt controls of the cell cycle,
DNA repair, and apoptosis (137). Sometimes, a few cells of
such "immortalized" OSE cultures survive crisis and become truly
immortal, continuous lines. Recently, we introduced constitutively
expressed E-cadherin into an SV40 Tag-immortalized line derived
from normal OSE. The resulting phenotype closely resembled neoplastic
OSE, and the cells formed adenocarcinomas in SCID mice (17, 18). These
adenocarcinomas resembled Mullerian duct-derived epithelia in that they
formed papillae and cysts and expressed CA125 and E- cadherin. The
line, IOSE-29EC, became not only tumorigenic but also acquired an
indefinite, truly immortal growth potential. While the exact
relationships between the introduction of T-antigen and E-cadherin to
tumorigenicity need to be examined in additional lines, this is the
first experimental transformation of normal human OSE to ovarian
adenocarcinoma cells and the first direct confirmation that OSE is
capable of such a transformation. The results support the hypothesis
that E-cadherin may act as an inducer of the Mullerian epithelial
differentiation that accompanies neoplastic conversion of OSE (36).
E. Variation in OSE characteristics among species
Important issues that are frequently overlooked in the
interpretation of data derived from studies of OSE are the structural
and physiological differences among OSE from different species. For
extrapolations of results to human OSE, one of the best tissue culture
models appears to be bovine OSE because of the relative similarity
between the reproductive systems of these two species (138). One
example of differences between species, discussed in Section
III.A, is the constitutive expression of E-cadherin by OSE of
rodents and pigs but not humans. Other differences include the
dependence of human but not rat OSE on high calcium levels in culture
media for growth (122) and the propensity of rat OSE but not human OSE
to undergo spontaneous transformation to immortal cell lines in culture
(11). Studies of rabbit OSE have provided some of the earliest and most
detailed information on hormonal regulation of OSE. In this species,
the responses to hormonal stimulation are associated with morphological
changes that differ significantly from those of the human (2, 59), The differences between OSE from different sources are likely
related to variations in the reproductive biology of different species
and might provide clues for the striking interspecies variation in
their propensity to develop epithelial ovarian cancers. Therefore, in
order to avoid reporting confusing and irreproducible results, it is
mandatory to specify species in discussions of OSE.
F. Culture of OSE from women with family histories of ovarian
cancer
One of the pressing problems in ovarian cancer management is the
lack of markers for the detection of preneoplastic or early neoplastic
changes in the OSE. Our laboratory and others have investigated this
problem by studying the properties of overtly normal OSE from women
with histories of familial ovarian cancer and, in particular, women
with proven predisposing mutations. As stated in Section
IV.B, the evidence for phenotypic changes in OSE in
situ of women with these predisposing mutations is controversial.
However, it appears that such OSE expresses an altered phenotype in
culture that might reveal early changes and, perhaps, be a source of
predictive markers for ovarian carcinogenesis (78, 139, 140).
As discussed earlier in this review, normal OSE cells have a tendency to undergo epithelio-mesenchymal conversion in culture. In contrast, ovarian carcinoma cells are nonresponsive to the environmental signals that induce this conversion and remain epithelial in culture indefinitely. The first indication to suggest that overtly normal OSE from women with family histories of ovarian cancer (FH-OSE) differs from the OSE of women with no family history (NFH-OSE) not only genetically but also phenotypically came in 1995, when CA125 in cultured OSE was found to be expressed in more cells and for longer durations in FH-OSE (141). CA125 is an ovarian tumor marker used to monitor the clinical progress of ovarian cancer patients, but it is also an epithelial differentiation marker that is expressed by normal oviductal and endometrial epithelium. The increased expression of CA125 suggested that FH-OSE cells might have a diminished capacity for epithelio-mesenchymal conversion. This hypothesis was supported by subsequent observations that showed an increased tendency of FH-OSE cells to retain an epithelial cellular morphology and growth patterns in two- and three-dimensional culture and to express the epithelial markers keratin and E-cadherin more frequently and over longer periods in culture than NFH-OSE. At the same time, the capacities for sponge contraction and collagen type III secretion, which are mesenchymal markers, were reduced compared with NFH-OSE cultures (36, 78).
Recently, we showed that the Met receptor for hepatocyte growth factor
(HGF) was down-regulated in prolonged cultures of NFH-OSE but was
stabilized in FH-OSE cultures at all passages, similar to ovarian
carcinoma lines. As Met is characteristically expressed by epithelial
cells, the presence of this receptor represents yet another epithelial
differentiation marker that persists longer in FH-OSE. In view of the
capacity of HGF to induce glandular morphogenesis (142), Met expression
may enhance the susceptibility of the FH-OSE cells to the aberrant
Mullerian differentiation that accompanies ovarian carcinogenesis (139, 143). Our data also revealed concomitant expression of HGF and Met,
suggestive of autocrine regulation by HGF-Met in most cases of FH-OSE
but rarely in NFH-OSE (Fig. 7
).
|
|
An additional difference from NFH-OSE was observed in SV-40 large T antigen-immortalized FH-OSE cultures, which were found to exhibit increased telomeric instability and a reduced growth potential indicative of greater proximity to replicative senescence (140). These observations are particularly relevant to the unexplained earlier age of onset that characterizes ovarian cancer in women with hereditary ovarian cancer syndromes (148).
A possible reason why differences between FH-OSE and NFH-OSE were detected mainly in culture may relate to the particular nature of these changes: most of them involve differences in the stability, rather than type, of phenotypic characteristics in culture. Since the response of cells to explantation into culture is thought to mimic their response to injury, the nature of the changes suggests the interesting possibility that FH-OSE may respond abnormally to regenerative stimuli. This possibility is particularly intriguing in view of the apparent role of ovulation as a predisposing factor in ovarian carcinogenesis (8, 149).
| VI. Regulation by Hormones, Growth Factors, and Cytokines |
|---|
|
|
|---|
1. GnRH and gonadotropins. Recently, we showed that GnRH is an autocrine growth inhibitor for normal OSE. Using RT-PCR and Southern blot analysis, we cloned the GnRH and GnRH receptor in human OSE cells and found that they have sequences identical to those found in the hypothalamus and pituitary, respectively (150). It has been shown that gonadotropins stimulate cell proliferation of normal OSE of several species in vivo and in vitro (59, 151). Human OSE cells also have receptors for FSH (152). The presence of these receptors lends support to the hypothesis that the high FSH levels in peri- and postmenopausal women may play a promoting role in ovarian carcinogenesis, since this is the age of the peak incidence of epithelial ovarian carcinomas (153). Human and rabbit OSE cells express LH receptors since hCG, which is secreted by human OSE (47), stimulates their proliferation (120, 154) and LH also stimulates rabbit OSE growth in culture (59).
2. Steroids. Receptors for estrogen, progesterone, and
androgen were found at the mRNA and/or protein level in rat OSE (12)
and human OSE (104, 155). SV-40 large T-immortalized OSE cells
expressed ER
but not ERß (156). No direct effects of these
steroids on OSE proliferation have been demonstrated (104), but there
is increasing evidence for indirect actions. Expression by OSE of the
GnRH receptor appears to be reduced by estrogen (156A ), and estrogen
also modulates levels of HGF (157) and EGF both of which stimulate OSE
growth (see below). Furthermore, in ovarian carcinoma cells, estrogen
and progesterone markedly influence the steady state levels of mRNA for
the HGF receptor Met (145), and 5
-dihydrotestosterone down-regulates
the expression of mRNA for the TGFß receptors (158), suggesting that
these steroids may also have indirect effects on the growth regulation
of normal OSE. Although there is no evidence for a direct mitogenic
effect of ovarian steroids on OSE, it has been known for a long time
that corticosteroids enhance OSE proliferation in culture and that
combinations of EGF and hydrocortisone are among the most potent
mitogens for cultured OSE (16) (see below). Steroidogenic factor 1, a
transcription factor that regulates the differentiation of granulosa
cells and inhibits their proliferation, is also growth inhibitory in
rat OSE cells (159).
3. EGF family. Among growth factors, those of the EGF family
were among the first reported to stimulate human and rabbit OSE
proliferation either with or without costimulation by corticosteroids
(16, 56, 160, 161). OSE cells express receptors for EGF and for TGF
, which is a structural homolog of EGF also binds to the EGF
receptor (162). EGF not only stimulates proliferation of human OSE
cells but also profoundly affects their differentiation: within a few
days of EGF treatment, the cells convert from an epithelial to the
spindle-shaped morphology and lose epithelial differentiation markers
such as keratin (16). EGF is not present in large amounts in the plasma
(163) but is released from platelets during the clotting process. In
the ovary, EGF should therefore be present in increased amounts due to
the hemorrhage that occurs during follicular rupture (164). The
resulting localized stimulation of the OSE likely contributes to its
rapid postovulatory proliferation and perhaps also to
epithelio-mesenchymal conversion of OSE cells trapped within the
ruptured follicle. EGF has numerous functions in the ovary, which
include inhibition of FSH induction of LH receptors (165), inhibition
of estrogen production (166) and of theca differentiation (167), and
stimulation of progestin biosynthesis (168). TGF
has been
demonstrated immunohistochemically in human OSE in vivo and
in vitro and found to stimulate thymidine incorporation by
cultured human OSE cells. It was also demonstrated
immunohistochemically in human theca cells, suggesting that it plays a
role in the reproductive functions of the ovary (169). In OSE cells
whose life span has been extended by transfection with SV40 large T
antigen, EGF does not enhance proliferation but promotes survival
(170). Amphiregulin, another EGF homolog, is also a potent mitogen for
OSE cells and appears to control OSE and ovarian cancer cell
proliferation in a complex manner (171, 172).
Of particular interest for ovarian cancer are the heregulins, including the heregulin/neu differentiation factor, which are a family of ligands that cause phosphorylation of the HER2/neu receptor, a 185-kDs transmembrane protein kinase with extensive homology to the EGF receptor (reviewed in Ref. 173). HER1 (synonymous with EGF receptor), HER2, HER3, and HER4 are members of the type I receptor tyrosine kinase family (RTK I) of epithelial growth factor receptors (174). These receptors interact in multiple ways that modify their influence on a variety of cells (reviewed in Ref. 175). Although normal OSE cells express EGF receptors, they express little or no HER-2/neu (110, 172, 176, 177). However, HER2/neu is amplified and overexpressed in 2530% of ovarian and breast cancers, and this overexpression is associated with a poor prognosis (110, 173).
4. Other growth factors. Among other growth factors, basic
fibroblast growth factor (bFGF), a member of the FGF family of growth
factors (178), stimulates the proliferation of rabbit OSE (161) and
maintains viability in cultured rat OSE cells. The latter function
involves alterations in intracellular calcium levels and can be
mimicked by N-cadherin-mediated intercellular adhesion (129, 179).
Platelet-derived growth factor (PDGF) also stimulates proliferation of
OSE cells (180). Finally, TNF
, produced, for example, by
macrophages, induces both proliferation and TNF
expression in OSE
cells (57, 58, 181). It is significant that EGF and PDGF, which
stimulate OSE growth, are released from platelets during the clotting
process that occurs at ovulation. Recently, it was reported that
keratinocyte growth factor (KGF) and its ligand, Kit, represent an
autocrine mitogenic system for bovine OSE and that KGF/Kit may interact
with HGF in the regulation of this system (138).
5. TGFß family of growth-inhibitory factors. Among agents
that inhibit OSE growth are several members of the TGFß family of
growth factors (182), which affect and/or are produced by OSE. TGFß
itself, a widely distributed growth factor with multiple modes of
action, acts as an autocrine growth inhibitor for cultured human OSE
(74) and also counteracts the growth-stimulatory effect of EGF (183).
In contrast to some other inhibitory factors, TGFß does not induce
apoptosis in OSE cells (184). TGFß inhibits growth of rabbit OSE
(161) and regulates Kit ligand expression in immortalized rat OSE
(185). A detailed examination by immunohistochemistry and in
situ hybridization of TGFß subtypes, the related protein
endoglin, TGFß receptors, and TGFß-binding protein demonstrated the
presence of all of these in human OSE and, with the exception of the
binding protein, levels were lower than in ovarian cancers (186).
Interestingly, 5
-dihydrotestosterone down-regulates the expression
of mRNA for the TGFß receptors I and II in ovarian carcinoma lines
(158), suggesting that it might also counteract growth-inhibitory
effects of TGFß in normal OSE. Welt et al. (187)
investigated the TGFß-related factors, activin, inhibin, and
follistatin, in normal and neoplastic ovarian epithelia. OSE,
immediately after removal from the ovary, expressed mRNA for
follistatin 288 and 315, for the activin receptors IA, IB, II, and IIB,
as well as for the
-subunit and (weakly) the ß-subunit of the
ligands. At the protein level, OSE produced inhibin only. After 1 month
in culture, the
-subunit was undetectable while th