Secretion of annexin A3 from ovarian cancer cells and its association with platinum resistance in ovarian cancer patients

Abstract Early detection of resistance to platinum-based therapy is critical for improving the treatment of ovarian cancers. We have previously found that increased expression of annexin A3 is a mechanism for platinum resistance in ovarian cancer cells. Here we demonstrate that annexin A3 can be detected in the culture medium of ovarian cancer cells, particularly these cells that express high levels of annexin A3. Levels of annexin A3 were then determined in sera from ovarian cancer patients using an enzyme-linked immunosorbent assay. Compared with those from normal donors, sera from ovarian cancer patients contain significantly higher levels of annexin A3. Furthermore, serum levels of annexin A3 were significantly higher in platinum-resistant patients than in platinum-sensitive patients. To gain insight into the mechanism of secretion, the ovarian cancer cell lines were examined using both transmission electron microscopy and immunoelectron microscopy. Compared with parent cells, there are significantly more vesicles in the cytoplasm of ovarian cancer cells that express high levels of annexin A3, and at least some vesicles are annexin A3-positive. Moreover, some vesicles appear to be fused with the cell membrane, suggesting that annexin A3 secretion may be associated with exocytosis and the release of exosomes. This is supported by our observation that ovarian cancer cells expressing higher levels of annexin A3 released increased numbers of exosomes. Furthermore, annexin A3 can be detected in exosomes released from cisplatin-resistant cells (SKOV3/Cis) by immunoblotting and immunoelectron microscopy.


Introduction
Ovarian cancer is the leading cause of gynaecological cancer mortality. Its treatment usually consists of optimal cytoreductive surgery and platinum-based combination chemotherapies. Although the platinum compounds cisplatin and carboplatin are among the most effective anticancer drugs, development of resistance to the platinum-based therapies has emerged as a major obstacle in the treatment of ovarian cancers. As a result, the 5-year survival rate for advanced ovarian cancer (stages III and IV) is only about 20-30% [1]. It has become apparent that early detection of drug resistance and prompt adjustment of the chemotherapy regimen are critically important for improving the outcome of ovarian cancer treatment.
There have been extensive studies on the mechanisms of platinum resistance in ovarian cancers. A number of genes, including GST-pi, LRP, MDR1, XIAP, HER2/neu, hMLH2 and hMSH1 have been found associated with platinum resistance in ovarian cancer cells [2][3][4][5][6]. However, it is still not well understood whether and how these genes can confer specific resistance to the platinum drugs. It also remains to be determined whether any of the associated gene products can be used as biomarkers for the early detection of platinum resistance in ovarian cancer patients. Through proteomic analysis, we have found that the expression of a number of proteins, including annexin A3, destrin, cofilin 1, glutathione-S-transferase omega 1 and cytosolic NADP ϩ -dependent isocitrate dehydrogenase, are significantly altered in platinumresistant ovarian cancer cells [7]. Further studies demonstrated that the expression of annexin A3 is also significantly increased in tumours from platinum-resistant ovarian cancer patients [8]. Moreover, enforced expression of annexin A3 specifically conferred platinum resistance to ovarian cancer cells both in culture and in an animal model, whereas down-regulation of annexin A3 in the platinum-resistant ovarian cancer cells made them more sensitive to cisplatin [8]. These results indicated that increased annexin A3 plays a critical role in platinum resistance and raised the possibility that it could be a biomarker for resistance to platinum-based therapies in ovarian cancer patients.
Annexins are a family of Ca 2ϩ and phospholipid-binding proteins that are ubiquitously expressed in various cells. The 12 human and vertebrate annexins are catalogued into group A of the family and designed A1-A13 [9]. They all contain a conserved 70 amino acid residue annexin repeat that is responsible for Ca 2ϩ and phospholipids binding. A number of studies have shown that annexins are able to induce the generation of an inward vesicle, a formation of lipid domains and intermembrane contact, which participates cellular processes such as membrane-cytoskeleton organization, membrane trafficking, endocytosis, exocytosis and cytokinesis [10]. These activities appear to be responsible for the diversified functions of annexins, which include mediating the anti-inflammation action of glucocorticoids, participating in the regulation of blood coagulation and modulating ion channel transportation [11]. In addition, the ability of annexin A5 to bind phosphatidylserine also makes it a valuable tool to detect apoptotic cells [12]. However, there are relatively few studies on the functions of annexin A3, the 36-kD protein expressed in the cytoplasm of many cells [13].
In this study, we demonstrate that annexin A3 is secreted from ovarian cancer cells. The amount of secretion is associated with the level of intracellular annexin A3 and platinum resistance. Both electron microscopy studies and biochemical analysis indicated that the secretion is associated with exocytosis and the release of exosomes. We also found that the level of annexin A3 in sera from platinum-resistant ovarian cancer patients is significantly higher than that of normal donors as well as that of platinum-sensitive ovarian cancer patients, indicating that annexin A3 may serve as a potential biomarker for ovarian cancer and platinum resistance.

Transmission electron microscopy (TEM)
Cells were fixed and processed as described [16]. Briefly, after treatment with 1% osmium tetroxide for 1 hr at 4ЊC, samples were dehydrated in ascending grades of ethanol and embedded in spur resin. Following overnight polymerization, ultrathin sections (70-80 nm) were made using an ultramicrotome. The sections were laid on copper grids, stained with uranyl acetate and lead citrate, and examined under the transmission electron microscope (JEOLTEM 1010, JEOL, Japan). Exosomes on grids were fixed in 1% glutaraldehyde and negatively stained with 1% uranyl acetate before examining with TEM.

Exosome purification
Cells (1ϫ10 7 ) were cultured with DMEM without FBS for 24 hrs to avoid potential contamination of exosomes from serum-derived products [19]. For purification of exosomes, 60 ml of culture medium was cleared by centrifugation at 1000 ϫ g for 10 min. and concentrated to ~1. 5

Release of annexin A3 from cultured ovarian cancer cells
Although annexins do not contain a signal sequence for protein secretion [20], some family members, including A1, A2, A3 and A6, have been found outside cells under many circumstances [21][22][23]. Therefore, we asked whether increased expression of annexin A3 in ovarian cancer cells can lead to their secretion to culture medium. Compared with those from parent SKOV3 and A2780 cells, concentrated supernatants from platinum-resistant cells SKOV3/Cis and A2780/Cis contained significantly higher levels of annexin A3 (Fig. 1). Supernatants from SKOV3 and A2780 cells transfected with an annexin A3 expressing plasmid also had elevated levels of annexin A3 (Fig. 1A and B). Furthermore, down-regulation of annexin A3 in SKOV3/Cis and A2780/Cis with antisense annexin A3 significantly decreased the amount of annexin A3 in the medium ( Fig. 1A and B). These results indicate that annexin A3 can be secreted into culture medium and the secretion is significantly increased in cells that express elevated levels of cytoplasmic annexin A3.

Expressions of annexin A3 in sera from ovarian cancer patients
The There is also a significant difference between the levels of annexin A3 in the sera from platinum-sensitive patients and that of normal donors (P ϭ 0.006).
The annexin A3 levels of individual patients are shown in Figure  1C and Table 1 ; Fig. 1D]. These results indicate that serum annexin A3 levels may be a prognostic biomarker for resistance to platinum-based chemotherapy in ovarian cancer patients with reasonable sensitivity (63%) and specificity (80%).

It has been shown in a variety of cells that annexins can affect vesicle trafficking and exocytosis by promoting membrane
fusion [24]. Therefore, we examined the ultrastructure of the ovarian cancer cells using TEM. The cell nucleus, many organelles in cytoplasm, and the cell membrane could be clearly identified under TEM (10,000ϫ). Compared with that of platinum-sensitive SKOV3 and A2780 cells, the cytoplasm of platinum-resistant cells SKOV3/Cis and A2780/Cis contains significantly increased numbers of membrane-bound round or elliptical vesicles, ranging from 0.1 to 1 m in size. Some of these vesicles contain high-density particles, suggesting that they may be related to phagocytosis or endosomes ( Fig. 2A). Under higher magnification (50,000ϫ), it is evident that some of the vesicles look like multivesicular bodies (MVBs) with contents of various densities. Some vesicles can be found fusing with cell membrane (Fig. 2B), indicating that they participate in exocytosis. Interestingly, SKOV3/Ann and A2780/Ann, the ovarian cancer cells that were transfected with an annexin A3-expressing vector, also possess increased numbers of vesicles (Fig. 2C). Furthermore, SKOV3/Cis/R and A2780/Cis/R, the platinumresistant cell lines that had been transfected with a vector expressing an antisense annexin A3 and had reduced levels of annexin A3, also had fewer vesicles (Fig. 2C, lower)

. These results demonstrate that increased expression of annexin A3 in the ovarian cancer cells results in the formation of increased numbers of MVBs in the cytoplasm.
To further explore the association of annexin A3 secretion and the cytoplasmic vesicle, cells were examined by IEM using an anti-annexin A3 antibody and a colloidal gold-labelled secondary antibody. To avoid its potential effect on antibody-recognized epitopes of annexin A3, osmic acid was not used in the staining of cells for electron microscopy. Therefore, the pictures from these studies are lighter and have less contrast compared to regular TEM pictures. Nevertheless, the increased numbers of the vesicles could be identified in SKOV3/Cis cells under this condition

ovarian cancer cell lines SKOV3 and SKOV3/Cis and A2780 and A2780/Cis. The arrows indicate the increased numbers of vesicles in platinum-resistant SKOV3/Cis and A2780/Cis cells. (B) Higher magnification (50,000ϫ) photographs of SKOV3/Cis cells. Arrows indicate the MVB-like vesicle (upper) and exocytosis of a vesicle (lower). (C) Representative TEM (10,000ϫ) photographs of ovarian cancer cell lines. Arrows in the upper panel indicated increased vesicles in annexin A3-expressing A2780/Ann and SKOV3/Ann cells. The lower panels are pictures of A2780/Cis/R and SKOV3/Cis/R cells. The vesicles observed in A2780/Cis and SKOV3/Cis (A)
have largely disappeared. (Fig. 3). As shown in Figure 3B, the high-density gold particles that are indicative of annexin A3 were present in the cytoplasmic region of the cells. Particularly, annexin A3 appeared to be highly expressed both around and inside the membrane of the vesicles. Pictures taken under higher magnification showed that some annexin A3-containing vesicles appear to be fusing with cell membranes (Fig. 3C and 3D), indicating that annexin A3 might be secreted through exocytosis.  demonstrated that annexin A3 is associated with these exosomes (Fig. 4B and C). However, because of the size of the gold particle (12 nm), it is difficult to determine whether annexin A3 is associated with the exosome membrane. The association of annexin A3 with these exosomes was further examined by anti-annexin A3 immunoblotting. As shown in Figure 4D, exosomes from SKOV3 cells contain both annexin A3 and the characteristic hsp70. These results indicate that at least some annexin A3 is secreted through exocytosis inside the exosomes. The amount of exosomes released from various ovarian cancer cells was further assessed by measuring total protein in exosome preparations. As shown in Figure 5A, enforced expression of annexin A3 significantly increased the amount of exosome released, whereas down-regulation of annexin A3 in platinum-resistant cells reduced exosome release significantly. The same phenomenon was observed when exosome preparations were analysed by immunoblotting with anti-hsp70 and annexin A3 antibodies (Fig. 5B).

Discussion
We have found in this study that annexin A3 can be secreted from ovarian cancer cells. Up-regulation of annexin A3 in these cells leads to a significant increase in its secretion, whereas down-reg-ulation of annexin A3 results in its reduction in culture medium. This is consistent with the observation that annexin A3 can be detected in urine from prostate cancer patients [22]. Examination of these ovarian cancer cells using TEM and IEM revealed that increased expression of annexin A3 results in the formation of MVBs-like vesicles. At least some of the vesicles contain annexin A3 and appear to fuse with the cell membrane. It was also found that annexin A3 is associated with exosomes released from platinum-resistant ovarian cancer cells. Because MVBs often contain exosomes that can be released through exocytosis [25], these results led us to propose that annexin A3 promotes the formation of MVBs that are secreted with exosomes upon fusion of the MVBs with the cell membrane. It is worth noting that other annexins, such as A1, A2 and A6 have also been found to be secreted through nonclassical pathways [23,[26][27]. Furthermore, it has been shown that annexin A2 is associated with MVBs in enterocytes [26], suggesting that exocytosis of MVBs might be a common mechanism for annexin secretion. This has been further supported by the finding that annexins are components of exosomes from a variety of cells [28].
We also examined the levels of annexin A3 in sera from 30 normal female donors and 50 ovarian cancer patients by ELISA. Compared to that of normal donors, there is a significant increase of annexin A3 in sera from ovarian cancer patients.  Furthermore, the concentration of annexin A3 in sera from platinum-resistant patients is significantly higher than that of platinum-sensitive patients. Ovarian cancer patients with higher levels of serum annexin A3 also have a significantly shorter progress-free time. These results indicate that annexin A3 may be a biomarker for ovarian cancer and a prognostic marker for platinum resistance. Consistent with our observation, increased annexin A3 immunostaining has been found in approximately two-thirds of colorectal cancer patients [29], in lung adenocarcinoma with lymph node metastasis [30] and in prostate tumours [31]. In particularly, platinum drugs are also commonly used in the treatment of patients with lung cancer. It could be very informative to further assess the serum annexin A3 levels in these patients before, during, and after chemotherapies. Interestingly, a number of patients enrolled in our study had particularly high levels of serum annexin A3. It is conceivable that, by assessing the level of annexin A3 in a significantly larger group of patients, we may be able to associate the high serum A3 level with a particular pathological type, stage and size of ovarian cancer. To further examine the prognostic role of annexin A3, we are also planning to compare serum annexin A3 levels in ovarian cancer patients before, during and after cytoreductive surgery and chemotherapy. It is also worth noting that some of the platinum-resistant patients do not have increased serum A3 levels. This may reflect the notion that drug resistance in ovarian cancer can result from abnormal pharmacokinetics and tumour microenvironments [32]. Because we have shown previously that annexin A3 specifically confers resistance to the platinum compounds [8], it is also conceivable that the insensitivity to platinum-based chemotherapy in some patients is because of resistance to other agents such as paclitaxel [33,34].
Two recent studies have found that, when compared with other types of ovarian cancer, annexin A4 is up-regulated in ovarian clear cell carcinomas and is associated with platinum resistance [35,36]. Given the structural and functional similarities among the annexin family members, it is conceivable that annexins A3 and A4 may act through similar mechanisms in ovarian cancer cells. We are also interested in examining the expression of additional annexins in various types of ovarian cancers. It has been reported that annexin A3 in urine detected by immunoblotting is a highly specific marker for early detection of prostate cancer [22]. Given   level, we are interested in examining annexin A3 in urine from ovarian cancer patients in the future. In addition to the 36-kD annexin A3 that we investigated in this study, a 33-kD alternatively spliced form of annexin A3 has been found in primary cell cultures from renal cell carcinoma and normal renal cortex tissue [37]. Intriguingly, although 36-kD isoform was down-regulated in renal cell carcinoma, the 33-kD isoform was up-regulated in these tumour cells [37]. We are interested in further exploring whether the smaller isoform is also associated with other types of tumours. Because the N-terminal region of annexin A3 plays an important role in mediating protein-protein interaction, it would be interesting to determine whether enforced expression of the 33-kD isoform can still increase the formation of vesicles in ovarian cancer cells, which would help to understand the mechanism of annexin A3-promoted vesicle formation.