Glucose-Regulated Protein 78 Is a Potential Serum and Imaging Marker for Early Detection of Ovarian Cancer.

Normal ovarian and fimbrial ( n = 10) tissues and BRCA+ fimbrial tissues ( n = 10) were obtained from subjects who underwent surgery for hysterectomy due to non-ovarian reason. Ovarian benign tumors ( n = 10, including serous cystadenomas = 3 and cystadenofibromas = 3) or high-grade serous carcinoma (HGSC) at early ( n = 8) and late stages ( n = 16) were obtained from the Department of Pathology, Rush University Medical Center (RUMC) following surgery. Staging and histological types of tumors were obtained during surgery and from final pathology reports. In addition, ovaries with endometriosis ( n = 3) were also used as an additional control with benign condition. Blood samples from subjects and patients were collected from Pathology following surgery. Serum was separated and stored at −80 °C for use in immunoassay. Specimens were examined by routine staining, immunohistochemistry (paraffin sections, 5 µm thick), proteomic and molecular biological studies (mRNA expression). White Leghorn laying hens were reared under standard poultry husbandry practices. Three to four-year-old hens ( n = 190) were scanned with traditional TVUS as reported earlier [ 17 ]. The number of hens for control (healthy) and OVCA (experimental) groups were determined to yield a true between-group difference in their tissue and/or serum level of GRP78. Accordingly, healthy hens with normal ovaries ( n = 10) or hens with ovarian tumors at early ( n = 10, serous = 4, endometrioid = 3, mucinous = 3) or late stages ( n = 10, serous = 4, endometrioid = 4, mucinous = 2) were used. Blood was collected, hens were euthanized and examined for presence of solid masses, and the extent of dissemination of OVCA was recorded. Normal and tumor tissues were processed for immunohistochemistry, protein and gene expression studies. Serum was separated and stored at −80 °C for later use. Three to four-year-old laying hens ( n = 50) with or without tumors were selected by traditional TVUS imaging as mentioned earlier [ 17 ]. GRP78-targeted imaging agents were developed using anti-GRP78 antibodies (Abcam, Cambridge, MA, USA) as reported earlier, with little modification [ 18 ]. Briefly, anti-GRP78 antibodies were biotinylated using biotinylation kit (Abcam, Cambridge, MA, USA) and then conjugated with microbubbles containing streptavidin (Targeson, Inc., San Diego, CA, USA). Hens were scanned with TVUS imaging before and 5 min after injection with 10 µL of GRP78-targeted imaging agents per kilogram body weight of hen, as reported earlier [ 18 ]. TVUS images were archived and analyzed to determine the signal intensities of the tissues before and after injection of targeted-imaging agent and expressed as mean ± standard error of the mean (SEM) in 20 mm 2 area of tissue. Hens were euthanized following targeted imaging, examined for the presence of OVCA, and tissues were processed. Total tissue protein, nuclear matrix protein (NMP), and total RNA were collected from all samples as reported previously [ 19 , 20 , 21 , 22 ]. Additionally, lysates of normal human ovarian surface epithelial cells (HOSE) and ovarian malignant cell lines including OVCAR3, SKOV3 and Caov3 (ATCC, Manassas, VA, USA) were collected and examined for GRP78 expression by immunoblotting. Immunohistochemical detection of GRP78 expression was performed using anti-GRP78 antibodies, mentioned above, as reported previously [ 23 ]. The intensities of GRP78 expression by normal and malignant cells were determined as reported earlier [ 23 ], and expressed as the mean intensity ± standard error of the mean (SEM) in 20 mm 2 area of the tissue. Intensities of immunohistochemical staining were reported as arbitrary values as defined by the computer-assisted imaging software (MicroSuite TM version 5, Olympus American, Inc., Center Valley, PA, USA). Serum levels of GRP78 in representative normal and OVCA patients and hens were determined by immunoassay using commercial GRP78 ELISA kits for human and chicken, respectively, as per the manufacturer’s instruction (MyBioSource, Inc., San Diego, CA, USA). A standard curve was generated, and serum GRP78 levels were determined with reference to the standard curve as per manufacturer’s recommendation using a software program (Gen5, version 2.00, Biotek Instruments, Inc., Winooski, VT, USA). Values are presented as mean concentration ± standard error of the mean (SEM). Immunohistochemical expression of GRP78 was confirmed by 1-& 2-D-WB using the same antibodies mentioned above and similarly reported earlier [ 23 , 24 ]. Two-dimensional gel electrophoresis was performed as reported previously [ 25 ]. Proteins were separated and transferred to a nitrocellulose membrane [ 26 ], and immunoreactive GRP78 protein on the membrane was determined as reported earlier [ 19 ]. GRP78 mRNA expression was assessed by qRT-PCR assays using human-specific GRP78 primer, and the 18s SnRNA primer (an endogenous primer) was designed from Qiagen (Foster City, CA, USA). The differences in GRP78 mRNA expression levels were calculated as fold-changes as reported earlier [ 22 , 27 ]. Semi-quantitative RT-PCR was performed for GRP78 mRNA expression, as reported earlier [ 28 ] (data not shown). Differences in the intensities of GRP78 expression among different groups of clinical specimens, as well as hens, were assessed by ANOVA. Differences in the intensity of GRP78 protein expression (in immunoblotting), GRP78 mRNA expression or serum GRP78 levels among normal or tumor ovaries, both in women and hens, were analyzed similarly. Significant differences in the signal intensity of TVUS imaging due to GRP78-targeted imaging agent were examined using two-sample t-tests (pre-targeted vs. post-targeted imaging from OVCA hens). Correlations in tissue GRP78 protein and GRP78 gene (mRNA) expression were examined using the Pearson co-efficient of correlation at 95% confidence interval ( n = 4). All reported p values are 2-sided, and p < 0.05 was considered significant. Statistical analyses were performed using GraphPad Prism (GraphPad software version 6, La Jolla, CA, USA).

Approximately 90% of all ovarian cancers (OVCA) are epithelial, and type-II OVCA is the most lethal form which, in most cases, is detected at stage-III and stage-IV [ 1 , 2 ]. The 5-year survival rate of OVCA patients when detected at late stages is approximately 45%, as opposed to >90% when detected at early stage [ 3 ]. OVCA recurs frequently [ 4 ], so an effective early detection test for OVCA is urgently needed. Serum levels of cancer antigen 125 (CA-125) with or without traditional transvaginal ultrasound (TVUS) imaging, are the currently used methods for OVCA detection. CA-125 is not specific for early-stage OVCA, as its serum level is also elevated in ovarian or non-ovarian benign conditions, and limited resolution of traditional TVUS cannot detect early changes associated with OVCA. Further, CA-125 alone or in combination with TVUS did not improve the rates of early detection of OVCA [ 5 ]. In addition, no target(s) in the ovary indicative of malignant transformation has been established for TVUS imaging. Therefore, effective serum markers, imaging target(s) in the ovary and an imaging agent to detect this target need to be developed to establish an early detection test for OVCA [ 2 ]. Chronic inflammation and oxidative stress have been suggested as risk factors for malignant development [ 6 ]. Ovaries and fimbriae of the fallopian tubes are exposed to various internal and external inflammatory factors, including ovulation and/or infection [ 7 ]. Ovulation is followed by the influx of immune cells and frequent ovulation leads to chronic inflammation and oxidative stress [ 7 , 8 , 9 ]. Long-term exposure to oxidative stress increases production of glucose-regulated protein 78 (GRP78), an endoplasmic reticulum (ER) resident protein and marker of oxidative stress. To withstand and survive from stressful conditions, GRP78 is expressed by the cell surface and secreted into the circulation [ 10 , 11 , 12 ]. Thus, GRP78 has the potential to be used for the detection of OVCA. The goals of this study were to examine systematically: (1) whether expression of GRP78 increases at cellular and molecular levels during OVCA development, and (2) whether GRP78 can be detected by TVUS imaging. As access to patients with early-stage OVCA is difficult, the study on GRP78-targeted TVUS imaging was performed in laying hens. Rodents do not develop OVCA spontaneously, and induced OVCA in rodents does not represent spontaneous OVCA in humans. In contrast, laying hens ( Gallus gallus domesticus ) develop OVCA spontaneously with high incidence rates and remarkable similarities to human OVCA [ 13 , 14 , 15 , 16 ]. Ovarian tumors in hens can also be detected using an ultrasound scanner similar to that used in the clinic [ 17 ]. Three experiments were conducted in this study, including the determination of GRP78 expression in normal or OVCA patients (experiment 1) or hens (experiment 2). In the third experiment, a GRP78-targeted imaging agent was developed, and its feasibility for OVCA detection by GRP78-targeted TVUS imaging was examined.

Normal ovaries from women at the perimenopausal stage showed ovarian surface epithelium (OSE) with occasional presence of stromal follicles. In contrast, ovaries in normal postmenopausal women contained OSE and stromal fibroblasts with no follicles ( Figure 1 A). Malignant ovarian tumors, including early and late stages used in this study, were serous carcinoma ( Figure 1 B,C). Expression of GRP78 was detected in OSE cells in normal ovaries and in tumor cells of benign and malignant ovaries ( Figure 1 D–F). Similarly, fimbrial surface epithelial (FSE) cells of normal or in BRCA1+ fallopian tube, one of the sites of origin of ovarian HGSC, were examined for GRP78 expression ( Supplementary Figure S1 ). Normal FSE showed weak expression for GRP78 while an intense expression for GRP78 was observed in the FSE of BRCA+ subjects (high risk for OVCA development). In addition to cell surface expression, malignant cells also showed cytoplasmic and nuclear staining for GRP78. Compared with normal ovaries and ovaries with benign tumors, the staining for GRP78 was stronger in HGSC. Furthermore, ovarian endometriotic lesions showed a weak expression for GRP78 ( Supplementary Figure S2 ). Compared with normal (19.20 × 10 4 ± 0.155 × 10 4 in 20 mm 2 area), the intensity of GRP78 expression was higher in benign tumors (24.03 × 10 4 ± 0.157 × 10 4 in 20 mm 2 ). The intensity of GRP78 staining was significantly ( p < 0.0001) higher in malignant tumors at early (44.94 × 10 4 ± 0.48 × 10 4 in 20 mm 2 area) and late stages (219.2 × 10 4 ± 18.0 × 10 4 in 20 mm 2 area) when compared to normal and benign tumors ( Figure 2 ). Tissue expression of GRP78 was confirmed by 1-D- & 2-D-WB of normal and malignant ovaries. 1-D-WB showed multiple bands with one at approximately 78 kDa for GRP78. 2-D-WB confirmed the intense expression of GRP78 at 78 kDa in malignant ovaries ( Figure 3 A,B). In addition, 1-D-WB of cell lysates showed stronger bands for GRP78 in OVCAR3, SKOV3 and Caov3 cancer cells than normal cells (HOSE) ( Figure 3 C). Immunoblot showed similar patterns of signal intensities of GRP78 protein expression in cancer cells to that of immunohistochemical expression of GRP78 ( Figure 3 E,F). Compared with normal ovaries, quantitative assays showed significantly higher expression of GRP78 gene in malignant ovaries ( p < 0.01). Fold changes in GRP78 gene expression ( Figure 3 D) were positively correlated with the intensity of GRP78 protein expression (in immunoblotting, Figure 3 E) (R 2 = 0.96, p < 0.05, n = 4). Thus, protein and gene expression data for GRP78 support immunohistochemical observations of an increase in GRP78 expression during OVCA development and progression. Normal ovaries in older (3–4 year old) hens had fewer preovulatory follicles (2–3 large follicles) ( Figure 4 A). In hens that had ceased laying, the ovaries became atrophied with no large follicle ( Figure 4 B). In the early stage of OVCA, solid masses were limited to the ovary accompanied with or without little to moderate ascites ( Figure 4 C). Tumors at late stage of OVCA metastasized to distal organs and were accompanied by profuse ascites ( Figure 4 D). Routine histological examination confirmed the presence of tumors in hens ( Figure 5 ). Microscopically, normal ovaries in hens contained surface epithelial cells, embedded stromal follicles, stromal atretic follicles, and a few scars and remnants of regressing postovulatory follicular tissues ( Figure 5 A). Histological examination of tumors showed the tumors were serous, endometrioid and mucinous ( Figure 5 B–D). In normal ovaries, GRP78 was detected in few OSE cells (red arrow) while other OSE cells did not stain for GRP78 (black arrow, Figure 6 A). Malignant cells showed intense staining for GRP78 ( Figure 6 B,C). Compared with normal ovaries (19.96 × 10 4 ± 3.5 × 10 4 ), the intensity of GRP78 expression was significantly ( p < 0.03) higher in malignant tumors at early stages, including serous (34.70 × 10 4 ± 2.58 × 10 4 ), endometrioid (32.15 × 10 4 ± 2.91 × 10 4 ) and mucinous (34.60 × 10 4 ± 4.0 × 10 4 ) in 20 mm 2 area of tissue ( Figure 6 D). Significant differences were not observed in GRP78 expression among different histological types of malignant tumors. Compared with normal levels (9007.76 ± 816.54 pg/mL), serum levels of GRP78 were higher ( p < 0.05) in patients with early stage (12,730.59 ± 817.35 pg/mL) and late-stage OVCA (13,930.12 ± 202.35 pg/mL) ( p < 0.01) ( Figure 7 A). Compared with normal (222.62 ± 181.69 pg/mL), the levels of serum GRP78 were significantly ( p < 0.005) higher in hens with early-stage OVCA (590.19 ± 198.18 pg/mL), and increased further in hens with late-stage OVCA (1261.38 ± 372.85) ( p < 0.003) ( Figure 7 B). Thus, these results suggest that serum GRP78 levels increase in association with OVCA development and progression in patients and hens. Compared with normal hens, 1-D-WB and 2-D-WB showed strong expression of GRP78 in tumor homogenates (total protein) from hens with OVCA at early and late stages ( Figure 8 A–F), and by tumor NMP ( Figure 8 G). A similar pattern of staining was shown by the cancer cells from early and late-stage OVCA in hens ( Figure 8 B,C) as was observed in patients ( Figure 3 ). Thus, protein and gene expression data for GRP78 support immunohistochemical observations. Further, 1-D-WB of tumor NMP supports the nuclear staining of GRP78 observed in immunohistochemical studies. Compared with pre-targeted (43.31 × 10 3 ± 10.7 × 10 3 ), the signal intensities of GRP78-targeted TVUS imaging from ovarian tumors (145.00 × 10 3 ± 15.4 × 10 3 ) were significantly higher ( p < 0.0001) ( Figure 9 ). Gross examination of hens after targeted imaging confirmed the presence of tumor-associated mass in the ovary. Similar patterns of increase in signal intensities due to GRP78-targeted imaging were observed among different histological types of ovarian tumors. Immunohistochemical, Western blotting and gene expression studies for GRP78 expression supported the observations of GRP78-targeted-TVUS imaging. Therefore, targeted imaging agents bonded with their targets (GRP78) expressed on the surface of the malignant cells and enhanced the signal intensities of TVUS imaging for ovarian tumors.

This is the first report describing the changes at the cellular and molecular levels of GRP78 (an endoplasmic reticulum resident protein and a marker of cellular stress) during OVCA development in patients and hens, as a preclinical model of spontaneous OVCAS. This is also the first report to show that GRP78-targeted imaging agents enhanced the resolution of traditional TVUS imaging. This study also showed increase in GRP78 levels during OVCA development. Similarities in expression of GRP78 during OVCA development in humans and hens suggest the suitability of laying hens for translational studies to develop an early detection test for OVCA based on GRP78-targeted imaging and its serum levels. Information on the early molecular and cellular changes associated with OVCA development are limited, and understanding such factors involved in malignant changes are critical to establishing an early detection test and/or treatment of OVCA. This study showed that, compared with normal and benign ovarian tumors, the intensity of GRP78 expression was higher in ovarian malignant tumors at early and late stages. Similar findings were also observed in an immunohistochemical study on the expression of GRP78 in OVCA [ 29 ] as well as malignancies of several other organs [ 30 , 31 , 32 , 33 ]. Thus, these results suggest that increase in GRP78 expression may be an indicator of early changes associated with OVCA development. However, the reason(s) for increase in GRP78 expression and its association with OVCA development is (are) unknown. The Cancer Genome Atlas (TCGA) reports expression of GRP78 in several cancers, including ovarian, lung, breast, and skin cancers [ 34 , 35 ]. Further, increase in expression of GRP78 is associated with poor survival in renal carcinoma, as reported by the Human Protein Atlas [ 36 ]. Information on changes in GRP78 expression is critical for designing treatment intervention as well as early detection of HGSC, as observed in this study. Longstanding chronic inflammation and cellular stress are hallmarks of malignant transformation [ 6 , 37 , 38 ]. The ovarian surface and the fimbria of the oviduct are constantly exposed to various bio-molecules associated with ovulation, and frequent ovulation has been suggested as a risk factor for OVCA [ 7 ]. Ovulation is an inflammatory event, and during ovulation the ovarian surface epithelium (site of ovulation) and the fimbrial surface epithelium (site of receiving of ovulated egg) are exposed to various factors including cytokines [ 39 ]. Furthermore, ovulation is followed by the influx of immune cells to these sites, leading to a chronic inflammatory state in the ovary. When the inflammatory condition continues and remains unresolved, further influx of immune cells results in an oxidative burst, leading to a hypoxic state and production of reactive oxygen species [ 7 , 12 ]. Under stressful environments, including hypoxic condition, cells enhance the secretion of GRP78 as a mechanism to withstand stress and survive [ 12 ]. Laying hens are frequent ovulators, and this study showed increased expression of GRP78 in hens with OVCA. Thus, it is possible that stressful conditions, including frequent ovulation, to which OSE cells and the fimbria of the fallopian tube are exposed, may lead to the increased expression of GRP78, which may be involved in malignant transformation in these tissues. However, the mechanism of GRP78-induced malignant transformation of the OSE or fimbria is unknown. Following enhancement in expression under a stressful condition, GRP78 has been reported to have three main fates: one portion moves to the cell surface from the endoplasmic reticulum, another portion is secreted into the circulation, and the third portion translocates to the nucleus [ 40 ]. After translocating to the nucleus, GRP78 has been suggested to inhibit DNA-damage repair mechanisms [ 41 ], facilitating the growth of abnormal cells with altered or mutated DNA sequences. Uncontrolled growth of abnormal cells leads to the development of a malignant condition. In this study, both the cytoplasmic and nuclear staining of GRP78 were observed in ovarian malignant cells in women and hens. Moreover, in addition to the enhanced expression of GRP78 by malignant cells, serum levels of GRP78 also increased in association with OVCA development and progression in both patients and hens. These results suggest that GRP78 may be a potential serum marker for early OVCA detection. Furthermore, the portion of GRP78 expressed by the cell surface offers a potential target for imaging to detect OVCA at early stage by improving or enhancing the resolution of TVUS imaging. In this study, we developed, for the first time, a GRP78-targeted imaging agent, and this agent showed binding with its target and enhanced the signal intensity of traditional TVUS imaging from ovarian tumors in hens.

This study has several translational strengths. Enhanced expression of GRP78 by the cell surface during OVCA development can be detected by GRP78-targeted TVUS imaging. In addition, GRP78 becomes available in serum and is detectable by immunoassay. Thus, GRP78 has the potential to be a serum marker as well as an imaging target for the detection of OVCA at early stage. Moreover, OVCA in hens can be used to study and develop GRP78-targeted therapeutics, and to examine their effectiveness. As GRP78 has more than 90% homology between human and hens [ 42 ], information on GRP78 obtained from hen OVCA can be easily translated to humans. Thus, the laying hen represents a feasible model to study and establish an early detection test for OVCA using serum levels of GRP78 and GRP78-targeted TVUS imaging. Smaller sample size is a limitation of this study. However, taken together, the results of this study showed that enhancement in GRP78 expression at molecular and cellular levels is associated with OVCA development and progression. GRP78 represents a potential target to be detected by targeted-TVUS imaging and a serum marker for early detection of OVCA. This study will serve as the foundation for a larger clinical study to establish the feasibility of GRP78 as a serum marker and imaging target for early detection of OVCA.