Disabled-2 (Dab2), a candidate tumor suppressor of ovarian carcinoma, frequently (around 80%) loses its expression in ovarian tumors. Expression of exogenous Dab2 in tumor cell lines negatively regulates growth and suppresses the downstream signal of the Ras/mitogen activated protein kinase mitogenic pathway. The inactivation of Dab2 is believed to be an early event in ovarian tumorigenicity.
The authors analyzed the correlation among expression of Dab2, presence of basement membrane (collagen IV and laminin), morphologic alteration of the surface epithelial cells, and expression of the mitotic index marker Mib-1 in 50 archived ovarian tumors by an immunohistochemical approach. The stainings of Dab2, Mib-1, collagen IV, and laminin in premalignant lesions bordering both normal and neoplastic ovarian surface epithelium from adjacent slides were analyzed in 50 ovarian tumors.
In these 50 ovarian tumors, the percentage of Mib-1 positive tumor cells distributed in a wide range, from 1% to 75%, and there has no strong correlation with the expression of Dab2. However, in the premalignant regions bordering tumor and normal ovarian surface epithelium, the loss of Dab2 expression closely correlated with the dysplastic morphologic transition and Mib-1 expression of the ovarian surface epithelial cells. In 20 foci in 12 out of the 50 tumors, a transition from normal to neoplastic morphology within a contiguous epithelium was observed, and in all cases the morphologic change correlated with the loss of Dab2 staining. In addition, the collagen and laminin staining of the basement membrane were absent or weakened in pre-malignant epithelium prior to loss of Dab2 expression in all these 20 cases. Nevertheless, collagen IV and laminin were detectable in established adenomas on the same tumor slides.
The majority of ovarian carcinomas arise from the epithelial cells covering the surfaces of the ovaries.1–3 The ovarian surface consists of a single layer of flat cuboidal or simple squamous epithelial cells organized by a well-defined basement membrane, in which collagen IV and laminin are major components.4, 5 Derived from mesothelial cells during embryonic development, the surface epithelial cells are important for ovulation in the adult. The release of ova during each ovulatory cycle ruptures the surface epithelium layer, and the wound is healed as a result of proliferation and migration of the epithelial cells located around the site of follicular rupture.6, 7 These cells also deposit a basement membrane as part of the repair process. Increased cell proliferation as the result of repeated ovulation is thought to lead to genetic and epigenetic alterations in the epithelial cells, and the accumulation of genetic changes ultimately allows the development of ovarian carcinoma. This incessant ovulation hypothesis1, 6–8 is supported by epidemiologic data that the accumulated number of menstrual cycles correlates with the risk of ovarian carcinoma.7 Laboratory studies have shown that continuous culturing of rat ovarian surface epithelial cells leads to spontaneous cell transformation,9, 10 further supporting this hypothesis.
It is thought that as each ovulation occurs, genetic damage accumulates in epithelial cells with each phase of cell proliferation, and multiple genetic alterations are required for the development of ovarian carcinoma.1–3 Premalignant lesions in ovarian surface epithelium have been observed,11–15 and presumably the precursor cells contain a number of genetic mutations and epigenetic changes. Often located in deep invaginations or inclusion cysts, these lesions probably form when ovarian surface epithelial cells are trapped inside ovaries during the repair of the wound caused by ovulation.16, 17 Upon malignant transformation, the surface epithelial cells hyper-proliferate and either disseminate into the peritoneal cavity or invade and grow inside the ovarian cortex. The malignant tumor cells are no longer bound by a basement membrane, and disorganization is the most prominent hallmark of the malignancy.4
Disabled-2 (Dab2)18 is a candidate ovarian tumor suppressor gene,19–22 and its inactivation is believed to be an early event in ovarian tumorigenicity.21 Dab2 was first cloned from a mouse macrophage library as a 96 kD phosphoprotein responsive to mitogen/growth factor stimulation.18 The role of Dab2 in ovarian carcinoma was first suggested from studies using RNA fingerprinting to identify differentially expressed RNAs in epithelial carcinoma cells.19, 20 A partial cDNA, named DOC-2, was found to be expressed in all normal cells but absent in ovarian epithelial carcinoma cell lines and tissues. The full-length DOC-2 shares 93% homology with the mouse Dab2.23, 24 The expression of Dab2 is lost in the vast majority (around 80%) of ovarian carcinomas,21 and reexpression of Dab2 by cDNA transfection suppresses growth and tumorigenicity of Dab2-negative tumor cells.20, 22 It has been shown that Dab2 is a Grb2 binding protein,25 and Dab2 acts to uncouple Ras/mitogen activated protein kinase (MAPK) activation and c-Fos expression26 and thus terminates mitogenic signaling. Dab2 is thought to function in epithelial cell positioning organization, and inactivation of Dab2 has been proposed as enabling the disorganized growth of tumor cells in ovarian tumorigenicity.22
In the current study, we have analyzed the expression of Dab2 in a panel of 50 human ovarian tumors by immunohistochemical staining. The correlation of Dab2 expression with dysplastic morphologic transformation, loss of basement membrane components, and the proliferative index of pre-malignant lesions, especially ovarian surface epithelium bordering normal and tumor areas, was examined. Based on our observations, we have postulated several common events in the dysplastic transformation of ovarian surface epithelium: loss of collagen IV and laminin in the basement membrane and the inactivation of Dab2 accompanying morphologic transformation and proliferation of ovarian surface epithelial cells in pre-malignant lesions.
MATERIALS AND METHODS
Ovarian Tumors and Tissues
A representative set of 50 archived human ovarian tumor tissues and 5 nontumor human ovaries were selected from the ovarian tumor collection held by the Department of Pathology, Emory University School of Medicine. All these tissues were obtained from surgeries at the Emory University Hospital. Tumor grades and histologic subtypes were obtained from the pathology record without reference to the patients' personal information. The use of these human tissues was examined and approved by the institutional Human Investigation Committee, and safety and ethical guidelines were followed in using the human tumor tissues according to institutional requirements.
The tissues were embedded in paraffin, and grade and histologic subtype were confirmed in sections stained with hematoxylin and eosin. All the tumors were identified as being of ovarian surface epithelial origin. Of the examined cases, there were 2 endometrial carcinomas, 36 serous papillary and cystadenocarcinomas, 5 mucinous adenocarcinomas, and 7 poorly differentiated adenocarcinomas. The age of the patients ranged from 43 to 79 years, with a mean age of 58.5 years.
The tissue paraffin blocks were cut into 5 μm sections and placed on positively charged glass slides. Adjacent sections from the same tissue block were stained with anti-Dab2, anticollagen IV, antilaminin, and anti-Mib1 following steam heat induced antigen retrieval. The procedure used has been described in our previous report.21, 22 The antibodies and dilutions used are listed in Table 1.
Table 1. Antibodies and Dilutions Used
Dako (Carpinteria, CA)
Transduction Lab (Lexington, KY)
Immuno (West Brook, ME)
The appropriate antibody dilution was determined using positive and negative control slides prior to the study. Both positive and negative controls were included during immunostaining of the tested slides.
Mib-1 Positive Quantification
The percentage of Mib-1 positive tumor cells in each tumor was quantified by computer-assisted automated image cytometry as described previously.27 The results were confirmed by further visual inspection.
Characterization of Dab2, Collagen IV, Laminin, and Mib-1 in the Ovarian Epithelium
In the ovary, the surface epithelial cells grow as a single layer in contact with the basement membrane,28 which acts as both a mechanical barrier and an organizer of the tissue structure. The epithelial cells are laterally interconnected through specialized adhesive areas, including adhesive junctions, desmosomes, and tight junctions.29 As shown in Figure 1, the ovarian surface epithelium was stained positive (arrow) for Dab2, collagen IV, and laminin. Dab2 staining was within the surface epithelial cells, which were well-organized and in contact with the basement membrane. Collagen IV staining was found only in a well-defined layer (indicated by an arrow), the basement membrane, between the surface epithelial cells and the stroma. Laminin staining was seen in the basement membrane layer as well as in a defuse trace around the surface epithelial cells. In disease-free ovaries, there was no significant collagen IV and laminin staining in the stroma. Dab2 staining was found mostly in the surface epithelial cells (arrow), though some macrophages and fibroblasts scattered in the stroma were found to be Dab2-positive. The ovarian surface epithelial cells were normally quiescent, and Mib-1 positive cells were rarely observed in the epithelium.
Correlation of Dab2 Expression with Tumor Grade and Proliferative Index of Ovarian Tumors
Previously, Dab2 expression has been analyzed by immunostaining of 27 ovarian tumors.21 To confirm and extend the current study, we analyzed the expression of Dab2 in an additional set of 50 ovarian tumors as well as the proliferative index of the tumors in adjacent slides. As shown in Table 2, Dab2 expression was lost in the majority (74%) of tumors; however, the percentage of tumors in which Dab2 expression was lost did not progressively increase from low grade to high grade tumors. These results confirm our previous conclusion that the inactivation of Dab2 is an early event in ovarian tumorigenicity.
Table 2. Correlation Between Dab2 Expression and Tumor Grades
A set of 50 representative human ovarian carcinomas (50 tumors) was selected and immunostained with anti-Dab2. The Dab2 stainings, negative (−), light positive (+), positive (++), and strong positive (+++), are listed correlating with tumor grades. No Dab2 + (light positive) tumors were denoted in this tumor set. Dab2: Disabled-2.
In immunostaining and histological analysis of tumor sections, Mib-1 (also known as Ki67), a nuclear protein expressed in proliferating cells with as yet unclear cellular function, is often used as an indication of the tumor proliferative index.30–32 The percentage of Mib-1 positive tumor cells in each tumor was quantified by computer-assisted automated image cytometry and confirmed by visual inspection. The percentage of Mib-1 positive cells in this set of tumors showed a very wide range, from 1 to 75% (Fig. 2). There was no striking correlation between Mib-1 status and tumor grade (Fig. 2). Nevertheless, the five tumors with greater than 50% Mib-1 positive cells were all Grade III cancer. Dab2-positive tumors were found only in Grade II and III, but not in Grade I, tumors (Table 2, Fig. 2A). Although the average number of Mib-1 positive cells was greater in Dab2-negative tumors (Fig. 2B), two of the five tumors having the highest percent of Mib-1 positive cells were also strongly Dab2 positive (Fig. 2A). Thus, no significant correlation was found among Mib-1 index, Dab2 expression status, and tumor grade in this set of 50 ovarian tumors.
Loss of Dab2 in Premalignant Ovarian Surface Epithelial Cells
In sections of ovarian tumors, it is observed that both morphologically normal and malignant cells are present on the same continuous surface epithelium. It is reasonable to consider that the morphologically normal epithelial cells and the physically immediate neighboring transformed cells have similar histories and probably derived relatively recent from the same precursor cells. Thus, the differences between these cells are little, probably one or a few critical genetic or epigenetic changes which cause cell transformation. The morphologically normal neighboring epithelial cells are thus likely precursors of cancer and known as preneoplastic or premalignant lesions.
We focused our attention on the premalignant lesions found bordering the tumor and nontumor tissues. Out of the 50 archived tumor tissue blocks, 12 tumors were found to contain a nontumor region with normal ovarian surface epithelium morphology. As controls, we first characterized the nontumor surface epithelium and inclusion cysts from the tumor specimens by immunohistochemistry. A typical example is shown in Figure 3 of an inclusion cyst (or deep invagination of the surface epithelium, indicated by an arrow) in an ovarian section stained with antibodies against Dab2, Mib-1, collagen IV, and laminin (Fig. 3). The ovarian surface epithelium (Fig. 3) in this noncancerous part of the tumor tissue (S92-12477-4Y, serous cystadenocarcinoma, Grade III) resembles normal ovarian surface epithelium (Fig. 1): it is Dab2-positive, Mib-1 negative, and the epithelial cells are organized by a well-defined basement membrane that is collagen IV- and laminin-positive (arrows).
In some cases, ovarian surface layers or inclusion cysts were present, containing both morphologically normal and neoplastic epithelium, suggesting an ongoing transition, versus an established transformed lesion. An example shown in Figure 4A is the inclusion cyst (or deep invagination) found on another area of the same slide of the serous cystadenocarcinoma, also shown in Figure 3. The loss of Dab2 closely correlated with the dysplastic morphologic transition of the epithelial cells (Fig. 4B). In the 12 ovarian tumor specimens containing nontumor tissues, there were at least 20 areas that contained an epithelium in transition from morphologically normal to neoplastic. In all cases, we consistently observed that loss of Dab2 expression closely correlated with epithelial dysplastic transformation. In adjacent tissue sections, Mib-1 positive cells were present in the morphologically transformed, Dab2-negative cells (double arrows), but not in the morphologically normal, Dab2-positive cells (arrow) (Fig. 4B). Thus, we conclude that loss of Dab2 is closely associated with the timing of early neoplastic transformation.
In transformed ovarian surface epithelium containing established adenomas, the loss of Dab2 associated with cell proliferation and morphologic transformation, as illustrated in Figure 5. These portions of an ovarian papillary serous cystadenocarcinoma (Fig. 5A) or an ovarian papillary cystadenocarcinoma (Fig. 5B) contained established adenomas in the transformed surface epithelium with multicell layers and pseudostratified cells (arrow). Dab2 staining was absent in these cells, and Mib-1 was positive in cells scattered within and around the epithelium. A basement membrane was present between the epithelium and the stroma, as defined by collagen IV and laminin staining (arrow, Fig. 5).
Dissolution of the Basement Membrane Prior to Ovarian Surface Epithelium Transformation
Somewhat surprisingly, the stainings for collagen IV and laminin were absent in the basement membrane of the preneoplastic epithelium (arrow, Fig. 4B). Another example of an ovarian surface epithelium in transition from normal to neoplastic morphology is shown in Figure 6. In all 20 cases of the tumors with areas of morphologic transition, the collagen IV and laminin staining of the preneoplastic epithelial basement membrane was very weak or completely absent, as shown in Figures 4B and 6 (the position of the presumed basement membrane is indicated by an arrow). In an adjacent site on the same slide as Figure 4, collagen IV and laminin were strongly positive in the basement membrane of the epithelium of a normal inclusion cyst indicated by an arrow, Fig. 3, which served as a control for the staining. In addition, the neighboring blood vessel walls stained positively for collagen IV and laminin. The tumor and transformed epithelial cells were positive for laminin around the cells, suggesting that laminin was produced but that the basement membrane was not intact or maintained. Therefore, in the morphologically normal, Dab2-positive preneoplastic cells, collagen IV and laminin were already absent (Figs. 6 and 4B, arrow), indicating that the dissolution of the basement membrane occurred prior to the loss of Dab2. Absence of collagen IV and laminin was observed, in all cases, only in the epithelium undergoing transformation, but not in the normal or the established tumors. Collagen IV and laminin staining, however, was often observed in established adenocarcinomas of the transformed epithelium, as in an example shown (arrow) in Figure 5A, photographed from the same set of slides shown in Figure 6. Light collagen IV and laminin staining were observed in the basement membrane of the dysplastic epithelium (double arrows, Fig. 6) located immediately next to the collagen IV and laminin-negative but morphologically normal ovarian surface epithelium (arrow, Fig. 6). Thus, we conclude that the loss of collagen IV and laminin in the basement membrane is a common event prior to the dysplastic morphologic transformation of ovarian surface epithelium (Fig. 7). The loss of collagen IV and laminin appears to be transient, and collagen IV and laminin positive basement membrane is often reestablished in adenocarcinomas soon after dysplastic transformation.
Premalignant lesions in the ovarian surface epithelium have been documented,11–15 and it is believed that ovarian carcinoma often arises from the epithelium in the inclusion cysts.16, 17 An accumulation of genetic and epigenetic alterations is associated with the stepwise progression of the epithelial cell transformation; however, there is little understanding of the molecular events associated with the progression. In analyzing transition epithelia found in ovarian tumors bordering both tumor and nontumor tissues, we identified that the loss of the candidate ovarian tumor suppressor Dab2 correlates closely with the dysplastic morphologic transformation of ovarian surface epithelial cells. The current data also indicate that collagen IV and laminin staining of the basement membrane is often absent in morphologically normal epithelium neighboring the premalignant lesion, suggesting that the loss of collagen IV and laminin in the basement membrane is an even earlier event than the loss of Dab2. However, the loss of collagen IV and laminin is a transient event. Based on these observations, we suggest that the following events occur in the initiation of a high proportion of ovarian carcinomas: transient loss of basement membrane components such as collagen IV and laminin, inactivation of Dab2, transformation to atypical morphology of the epithelial cells, and subsequent disorganized proliferation of tumor cells (Fig. 7).
Loss of Dab2 and Proliferation in Ovarian Tumors
By immunohistochemic staining, we have previously found a loss of Dab2 expression in approximately 80% of ovarian21 and breast33 tumors, but the loss of Dab2 expression did not correlate with tumor grade, suggesting that it may occur as an early event in ovarian malignancy.21 This conclusion has been further confirmed in the current study using an additional set of 50 tumors; Dab2 expression was lost in 74% of the tumors (37 out of 50 tumors were Dab2 negative).
The proliferative marker Mib-1 has been shown to correlate with prognosis, but not tumor grade.34–36 The data from the current set of tumors suggest a similar conclusion, although we did find that tumors with greater than 50% of the cells Mib-1 positive were high grade tumors. Although the average percentage of Mib-1 positive cells was slightly lower in Dab2 positive tumors, some Dab2 positive tumors also had a high Mib-1 positive index. Thus, there was no correlation between proliferation index and Dab2 expression. Additionally, all Dab2 positive tumors were Grades II and III, confirming the previous result21 that the loss of Dab2 does not correlate with tumor progression and that the inactivation of Dab2 is an early event. Some tumors (26%) developed without suppressing Dab2 expression, and we speculate that those tumors most likely inactivated another target in the Dab2 signaling pathway.
Thus, although Dab2 transfection can suppress cell growth and induce cell death in tissue culture studies,20, 22, 26 Dab2 expression does not correlate directly with cell proliferation in tumor tissues. This is consistent with the hypothesis that Dab2 functions in maintaining epithelial cell positioning organization22 and that inactivation of Dab2 or its signaling pathway allows cell growth that was otherwise suppressed by the restraint imposed by positioning organization.
The Role of Disabled Family in Cell Positioning
Dab2 is one of two mammalian homologs of Drospholia disabled (dDab), which was identified and cloned as a haploinsufficient dependent on abl mutations gene.37–39 Genetic analysis in Drosophila embryonic development indicates that Drosophila abl kinase and Dab function in controlling cell contact and the final position of neuronal cells.37–39 The other mammalian homolog, Disabled-1 (Dab1), has also been implicated in the intracellular signaling from the extracellular matrix protein reelin to determine the positioning of brain cells in mouse embryonic development.40–43
In contrast to Dab1, which is mainly expressed in the brain, Dab2 is found in various tissues and is strongly expressed in the surface epithelial layer of the ovary.18, 19, 21 We have hypothesized that Dab2 may share a similar function as Dab1 in controlling the position of epithelial cells.22 Previously, we have speculated that Dab2 may function as a tumor suppressor in ovarian carcinoma by negatively regulating Ras-mediated cell growth.25 The C-terminal proline-rich domain of Dab2 binds Grb2, an adapter protein that couples tyrosine kinase receptors to Son-of-Sevenless (Sos), both in vitro and in vivo, and competes with Sos for binding. Thus, Dab2 may downregulate Ras activation by sequestering Grb2 from Sos. Recently, Dab2 was found to uncouple Ras/MAPK activation and c-Fos expression,26, 44, 45 thus regulating the downstream effect of Ras signaling.
The control of epithelial positioning necessarily involves cell adhesion molecules and the basement membrane. Changes in cell to basement membrane adhesion that alter cell positioning control can be separated into two categories: those involving adhesion to basement membrane, e.g., a change in laminin, collagen IV, and cell surface glycoproteins, and those involving intracellular signaling, such as the Ras/MAPK signaling pathway. Alteration of the signaling pathway through inactivation of Dab2 and change in the basement membranes, such as the loss of collagen IV and laminin, presumably contribute to the loss of epithelial cell positioning control and cell neoplastic transformation.
Integrity of the Basement Membrane in Pre-neoplastic Ovarian Surface Epithelium
A hallmark of carcinomas of the ovary and other epithelial cells is loss of a continuous basement membrane.4, 46, 47 Benign cystadenomas and tumors without microinvasion often have a continuous basement membrane. However, with increasing tumor grade and metastases, basement membrane distribution is discontinuous and irregular and contains wide areas of complete absence of laminin and type IV collagen.4, 46, 47 In the current study we found that collagen IV and laminin are absent in the pre-neoplastic lesions immediately adjacent to transformed epithelium. However, we also observed in many cases that a continuous collagen IV and laminin-positive basement membrane is present in established adenocarcinomas, as documented previously,4, 46, 47 and as shown in Figure 5. It is possible that the loss of collagen IV and laminin in the basement membrane in preneoplasic epithelium is a reversible and transient event during the initiation of epithelium transformation. Once established, the collagen IV and laminin deposition in the basement membrane could be restored beneath the noninvasive cystadenomas.
The loss of collagen IV, which forms the structural scaffolding of the basement membrane, is believed to be a result of increased degradation by Type IV collagenases.48–50 Previous studies have shown a correlation between expression of Type IV collagenase, the loss of collagen IV, and malignant transformation.48–51 In serous tumors of the ovary, disruptions in collagen IV (as well as laminin) staining were found to correlate with the loss of differentiation of the tumor, with areas of microinvasion and metastases, and with increased expression of Type IV collagenases.48–51 Alternatively, aberrant synthesis of collagen IV has been reported in ovarian carcinoma cells.52 In addition, other basement membrane components have been reported to be lost or differentially expressed in ovarian carcinoma, including secreted protein acidic, rich in cysteine (SPARC), fibulin, and laminin.53–56
It has been suggested that the integrity of the basement membrane surrounding the mammary epithelial duct is regulated physiologically during mammary involution by the end of lactation. This suggests that the basement membrane may regulate the growth and death of mammary epithelial cells.57, 58 In the ovary, the integrity of the surface epithelium undergoes cycles of disruption and repair and is obviously regulated by physiologic factors.59, 60 This physiologic process is likely activated covertly during neoplastic transformation of the ovarian surface epithelium by either suppressing synthesis, enabling assembly, or elevating degradation of collagen IV and laminin. The loss of basement membrane components, including collagen IV and laminin, may be a permissive factor for the initiation of epithelium dysplastic transformation.
By uncovering the molecular events occurring in pre-neoplastic epithelium, such as loss of collagen IV and laminin in the basement membrane and loss of Dab2 in ovarian surface epithelial cells, we can begin to approach an understanding of the initiation of ovarian malignancies at a molecular level.
The authors thank Diane Lawson for technical assistance in immunostaining, Jonathan Boyd of the Fox Chase Cancer Center Imaging Facility for help with photography, and Patricia Bateman for her secretarial assistance.