Leucine-rich alpha-2-glycoprotein-1 is upregulated in sera and tumors of ovarian cancer patients
© Andersen et al; licensee BioMed Central Ltd. 2010
Received: 10 May 2010
Accepted: 10 September 2010
Published: 10 September 2010
New biomarkers that replace or are used in conjunction with the current ovarian cancer diagnostic antigen, CA125, are needed for detection of ovarian cancer in the presurgical setting, as well as for detection of disease recurrence. We previously demonstrated the upregulation of leucine-rich alpha-2-glycoprotein-1 (LRG1) in the sera of ovarian cancer patients compared to healthy women using quantitative mass spectrometry.
LRG1 was quantified by ELISA in serum from two relatively large cohorts of women with ovarian cancer and benign gynecological disease. The expression of LRG1 in ovarian cancer tissues and cell lines was examined by gene microarray, reverse-transcriptase polymerase chain reaction (RT-PCR), Western blot, immunocytochemistry and mass spectrometry.
Mean serum LRG1 was higher in 58 ovarian cancer patients than in 56 healthy women (89.33 ± 77.90 vs. 42.99 ± 9.88 ug/ml; p = 0.0008) and was highest among stage III/IV patients. In a separate set of 193 pre-surgical samples, LRG1 was higher in patients with serous or clear cell ovarian cancer (145.82 ± 65.99 ug/ml) compared to patients with benign gynecological diseases (82.53 ± 76.67 ug/ml, p < 0.0001). CA125 and LRG1 levels were moderately correlated (r = 0.47, p < 0.0001). LRG1 mRNA levels were higher in ovarian cancer tissues and cell lines compared to their normal counterparts when analyzed by gene microarray and RT-PCR. LRG1 protein was detected in ovarian cancer tissue samples and cell lines by immunocytochemistry and Western blotting. Multiple iosforms of LRG1 were observed by Western blot and were shown to represent different glycosylation states by digestion with glycosidase. LRG1 protein was also detected in the conditioned media of ovarian cancer cell culture by ELISA, Western blotting, and mass spectrometry.
Serum LRG1 was significantly elevated in women with ovarian cancer compared to healthy women and women with benign gynecological disease, and was only moderately correlated with CA125. Ovarian cancer cells secrete LRG1 and may contribute directly to the elevated levels of LRG1 observed in the serum of ovarian cancer patients. Future studies will determine whether LRG1 may serve as a biomarker for presurgical diagnosis, disease recurrence, and/or as a target for therapy.
Ovarian cancer is the most lethal gynecologic malignancy ; about 22,000 women are diagnosed annually in the U.S. and ~16,000 patients succumb to the disease . New biomarkers that either replace or are used in conjunction with the current ovarian cancer serum biomarker, CA125, are needed to improve diagnosis and treatment [1–4]. Biomarkers that distinguish between malignant and benign abdominal masses prior to surgery could identify those patients who should be referred to a gynecologic oncologist . Initial cytoreductive surgery by a gynecologic oncology surgeon has been shown to result in improved outcomes for advanced ovarian cancer patients . In addition, a biomarker that could be used to monitor the efficacy of therapy would be ideal to detect disease recurrence.
To date, serum biomarker discovery has been impeded by an abundance of twelve proteins that comprise ~95% of the serum proteome, and can mask lower abundance proteins . We have previously reported the use of immunoaffinity depletion columns coupled with complementary mass spectrometry-based proteomic technologies to identify several differentially expressed proteins in the pooled sera of serous ovarian cancer patients compared to healthy women [8, 9]. One such differentially expressed protein, leucine-rich α-2-glycoprotein-1 (LRG1), is ~3-fold more abundant in ovarian cancer serum compared to non-cancer control serum, and represents a potential serum biomarker for ovarian cancer.
Human LRG1 is a serum glycoprotein of 312 amino acids in length with a predicted unmodified molecular weight of 34 to 36 kD . LRG1 has five potential glycosylation sites; 2 D SDS-PAGE results show LRG1 molecular weight ranges from 44 to 55 kD with isoelectric points ranging from 4.52 to 4.72 , suggesting that modifications occur. LRG1 has a normal plasma concentration of 21-50 μg/ml [12, 13].
The function of LRG1 remains unknown, although reports have predicted its role in cell adhesion [14, 15] due to its leucine-rich repeats, granulocytic differentiation due to its expression in neutrophil lineage experiments , and cell migration due to its overexpression in high-endothelial venules and tendency to bind extracellular matrix proteins . LRG1 has been implicated as a protein involved upstream of the TGF-βR II pathway [18, 19], suggesting a role in signalling. Serum LRG1 binding to cytochrome c has been recently demonstrated  and is proposed to play a role in cell survival and apoptosis [13, 21].
In this study, we have validated our proteomic discovery experiments using sera, tissue, and cell lines from ovarian cancer patients and non-cancer controls.
Serum from patients with serous ovarian carcinoma (n = 58) and healthy female controls (n = 56) were obtained from the Gynecologic Oncology Group (GOG) Tissue Bank. The majority of ovarian cancer patients had stage III or IV serous tumors (n = 51), the others had stage I and II tumors (n = 7). The median age of the ovarian cancer patients was 52 years (range: 35-85 years) compared to 46 years (range: 19-58 years) for the non-cancer controls.
Additional sera were obtained from the University of Minnesota Tissue Procurement Facility (Minneapolis, MN). These samples were obtained immediately prior to surgery from women with suspected ovarian cancer. All patients were consented in accordance with the University of Minnesota Institutional Review Board (IRB) guidelines. Definitive diagnoses were determined by pathologists. A total of 193 samples were selected from patients with the following pathology: 10 benign mucinous ovarian cystadenomas, 10 fibromas, 19 cases of endometriosis, 16 cystadenomas, 30 other benign ovarian masses, 21 ovarian tumors of low malignant potential, 8 clear cell (5stage I or II; 3 stage III), and 79 serous ovarian cancers (11 stage I or II; 63 stage III or IV; 5 not staged). Collection, processing, and storage of all blood samples was strictly standardized as follows. Blood samples were collected in a vacutainer tube, allowed to clot at room temperature (RT) for 30 min, and centrifuged at ~2500 × g for 10 min at RT. The serum was removed and immediately divided into 100 μl and 1 ml aliquots, and stored at-80°C.
Tissue and ascites samples
Tissue and ascites samples were obtained from the University of Minnesota Tissue Procurement Facility, as previously described [22, 23]. All tissues were snap frozen in liquid nitrogen within 30 min of resection and stored in the vapor phase of liquid nitrogen. Tissue sections were made from each sample, stained with hematoxylin and eosin (H&E), and examined by a pathologist by light microscopy to confirm the pathological state of each sample; a second pathologist confirmed the diagnosis of each sample, documented the percent tumor (typically 100%), and documented any necrosis (typically none). The following tissues were analyzed in this study: 21 cases of serous ovarian cancer, 22 cases of serous ovarian cancer metastatic to the omentum, 24 cases of serous ovarian cancer widely metastatic to other regions (including peritoneal surfaces, bowel serosa, lymph nodes, liver, uterus, and the mesentery of the small bowel), 17 benign ovary tumors, 8 cases of ovarian tumors of low malignant potential, and 57 normal ovaries were analyzed for global gene expression. An additional 7 ovarian cancer and 13 normal ovary samples were used for RT-PCR and/or Western blot experiments. Ascites was obtained from 29 women undergoing surgery for the removal of serous ovarian cancer, as soon as it was released by pathology (typically within 1 hr of removal from the patient). Ascites was centrifuged at 600 × g for 10 min at RT, and the supernatant was immediately divided into small aliquots and frozen at -80°C.
Ovarian cancer cell lines SKOV3, ES-2, NIH:OVCAR3, HEY, C13, OV2008, OVCA429, OVCA433, A2780-S, and A2780-CP, provided by Dr. Barbara Vanderhyden (University of Ottawa, Canada); NIH:OVCAR5, provided by Dr. Judah Folkman (Harvard Medical School, Boston, MA); MA148 provided by Dr. Sundaram Ramakrishnan (University of Minnesota, Minneapolis, MN); CAOV3 provided by Dr. Robert Bast Jr. (University of Texas, Houston, TX) were maintained as previously described [24–26]. Immortalized normal ovarian surface epithelial (NOSE) cell lines 1816-575, 1816-686, HIO117, IMCC3, IMCC5, HIO3173-11, and HIO135, provided by Dr. Patricia Kruk (University of South Florida, Tampa, FL), were cultured as previously described [27, 28]. All cells were maintained in a humidified chamber at 37°C with 5% CO2 and were routinely subcultured with trypsin/EDTA.
Mouse IgG monoclonal antibody (mAb) 2F5.A2 against human sera LRG1 was used in the ELISAs . Mouse IgG mAb 2E3 against recombinant human LRG1 (Abnova Corporation, Taipei, Taiwan) was used for Western blots and immunocytochemistry. Normal mouse IgG (Equitech-Bio, Inc. Kerville, TX) was used as a negative control for all experiments. Mouse mAb AC-74 against β-actin (Sigma Aldrich, St. Louis, MO) was used on Western blots as a loading control.
The ELISA for LRG1, which employs cytochrome c as the capture ligand, was conducted as described previously . All samples were tested at least two times in triplicate. Concentrations of LRG1 were calculated from a purified standard . The ELISA samples were compared as follows: for the GOG samples, mean LRG1 concentrations were compared across patients with ovarian cancer and control samples using general linear model for repeated measures, adjusted for age. For pre-surgical samples, mean LRG1 concentrations were compared across patients of the eight diagnoses using a general linear model for repeated measures and the least squared means are reported. T-tests were used to make comparisons between groups; all reported p-values are adjusted for multiple comparisons using a Bonferroni correction. CA125 levels were provided from the medical records. The CA125 levels were highly skewed and the log transformation was used for all analyses. Pearson's correlation was used to determine the association between CA125 and LRG1. The diagnostic value of LRG1, when used in addition to CA125, was considered using receiver operating characteristic (ROC) curves. ROC curves were constructed by plotting sensitivity versus 1-specificity and the areas under the curve (AUC) were calculated. Patients with a benign mass, mucinous ovarian tumors, fibroma, endometriosis and cystadenomas were defined as having benign pathology, patients with clear cell and serous ovarian cancer were defined as having cancer and patients diagnosed as having low malignant potential disease were excluded from the ROC analysis.
All values reported are means ± standard deviation (SD) unless otherwise noted. Statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, NC).
Gene Expression Analysis
Ovarian tissues from 149 patients were obtained from the Tissue Procurement Facility as described above; tissue samples were provided to Gene Logic Inc. (Gaithersburg, MD) for microarray analysis. On receipt of the tissue samples at Gene Logic Inc., H&E-stained slides were examined by a pathologist to verify the diagnosis and percentage of tumor tissue present, and the absence of necrosis. All tissue samples underwent stringent quality control measures to verify the integrity of the RNA before use in gene array experiments [22, 23]. Total RNA was isolated and gene expression was assayed via the Affymetrix U133 Set gene array at Gene Logic Inc. Data was analyzed with the Gene Logic Genesis Enterprise System® Software, using the Gene Logic normalization algorithm, as previously described [22, 23]. The mean expression of LRG1 for each tissue type was calculated using the normalized expression values for Affymetrix probeset 228648_at, which is the only probe targeting LRG1 on this platform.
Reverse Transcriptase PCR
Total RNA was isolated from cell lines and tissues as previously described . The following oligos (Invitrogen, Carlsbad, CA) were used: LRG1 (forward, 5'CCATCTCCTGTCAACCACCT); reverse, 5'GTTTCGGGTTAGATCCAGCA) and β-actin (forward, 5'GGCCACGGCTGCTTC; reverse, 5'GTTGGCGTACAGGTCTTTGC). Select LRG1 cDNA amplicons were extracted, gel-purified, and sequenced with both LRG1 forward and reverse primers; sequences matched solely to LRG1 mRNA and genomic DNA sequences. As LRG1 is produced in the liver , we used liver mRNA as a positive control. β-actin served as a loading control.
For tissue, ~ 100 mg of snap-frozen tissue was extracted using a PowerGen 125 hand-held homogenizer (Thermo-Fisher Scientific, Waltham, MA) in 2 ml of T-PER™ Tissue Protein Extraction Reagent (Thermo-Fisher Scientific) containing a serine-and cysteine-protease inhibitor cocktail (Roche Applied Science, Basel, Switzerland). Insoluble cellular components were removed by centrifugation at ~20,000 × g for 20 min. For cell lines, cells were grown to >90% confluency under normal conditions, rinsed twice with PBS and harvested with a rubber policeman. Cells were then pelleted at 7300 × g for 2 min and resuspended in 50 mM Tris-HCl, 150 mM NaCl, 0.1% (v/v) NP-40, pH 8.0; with Halt™ protease inhibitor cocktail, EDTA-free (Pierce Biotechnology, Rockford, IL). After a 30 min incubation on ice with intermittent vortexing, cell suspensions were sonicated at 20% duty cycle, output 2 with a Sonifier 450 analog (Branson Ultrasonics, Danbury, CT). Cellular debris was removed by centrifugation at 16000 × g for 20 min. Protein concentration was determined by the BCA method (Thermo-Fisher Scientific).
For deglycosidation, cell extracts or LRG1 purified from human plasma  were denatured and treated with Peptide: N-Glycosidase F (PNGase F) for 2 hr following the manufacturer's instructions (New England BioLabs, Ipswich, MA).
Protein samples in Laemmli buffer (2% SDS (w/w), 50% glycerol, 0.1 M DTT, 50 mM Tris, pH 6.8), were separated on a 4-20% or 10% Tris-HCl Criterion gel (Bio-Rad Laboratories, Hercules, CA), and electroblotted to a polyvinylidene difluoride (PVDF) membrane in 20% methanol, 25 mM Tris base, 192 mM glycine, pH 8.0. The PVDF membranes were blocked with 5% BSA in 20 mM Tris base, pH 7.6, containing 200 mM NaCl, and 0.05% Tween-20, and then incubated with primary antibodies at 1 μg/ml for 1 hr at RT. Membranes were then washed and incubated with a horseradish peroxidase-conjugated secondary antibody (Thermo-Fisher Scientific) and proteins were detected by enhanced chemiluminescence, using SuperSignal West Femto Maximum Sensitivity substrates (Thermo-Fisher Scientific) and exposed to film (Midwest Scientific, Valley Park, MO).
Nineteen of 21 cell lines were examined by immunocytochemistry; the ovarian cancer cell line HEY and NOSE cell line IMCC5 were not analyzed. Cell lines were seeded into Nunclon™ 24 well plates (Nalge Nunc International, Rochester, NY) and grown to confluence. Cells were rinsed twice with PBS and then fixed with 100% methanol overnight at -20°C. Cells were rehydrated with PBS at RT and blocked with 5% v/v goat serum in PBS containing 0.1% Tween-20. Mouse mAb 2E3 (Abnova) against rLRG1 was added at a 1:50 dilution in blocking buffer and incubated overnight at 4°C. Cells were washed and incubated in a 1:50 dilution of fluorescein-labeled secondary antibody (goat polyclonal antibody against mouse heavy and light chains (IgG and IgM), Roche International, Basel, Switzerland) in the dark. Cells were washed, followed by incubation with 4', 6-diamidino-2-phenylindole (DAPI; Roche International) in blocking buffer. Cells were then washed with blocking buffer and stabilized with a SlowFade® Antifade kit (Invitrogen, Carlsbad, CA). Cells were observed with an Olympus IX70 fluorescence microscope with a 20 × objective lens (Olympus, Tokyo, Japan) and a PixCell IIe™ Image Archiving Workstation camera (Molecular Devices, Sunnyvale, CA). Images were digitized using DVC View, v.2.2.8 software (DVC Company, Austin, TX). DAPI fluorescence was observed with a 285-330 nm excitation filter and a 420 nm absorption filter (U-MWU; Olympus). FITC fluorescence was observed with a 470 to 490 nm excitation filter and a 520 nm absorption filter (U-MP; Olympus).
Processing of Serum-Free Conditioned Media
The ovarian cancer cell line NIH:OVCAR5, and the NOSE cell line, 1816-575, were grown to >90% confluency in media with serum [RPMI 1640 supplemented with L-glutamine, 0.2 U/ml bovine pancreas insulin (Sigma Aldrich), 50 U/ml penicillin and 50 μg/ml streptomycin (Mediatech, Inc., Manassas, VA) and 10% heat inactivated fetal bovine serum (FBS, Atlanta Biologicals, Lawrenceville, GA); or a 1:1 mixture of M199: MCDB 105 (Sigma Aldrich) supplemented with 0.1 mg/ml gentamicin (Invitrogen) and 15% FBS], as previously described [24–26]. Media was decanted and cells were rinsed three times with PBS. Cells were cultured for an additional 24 hr in serum-free MCDB 105 media (Sigma Aldrich). The media was collected and cellular debris was pelleted at 50,000 × g at 4°C for 1.5 hr. The media was concentrated using a 4 ml, 5000 MWCO PES membrane concentrator (VivaScience, Hanover, Germany) centrifuged at 5000 × g to a final volume of ~100 μl. Buffer exchange into PBS was accomplished by three reservoir changes with PBS. Protein concentration was determined by the BCA method.
Proteins were subjected to tryptic digestion, dried down in a SpeedVac and rehydrated in water/ACN/FA (95:5:0.1). Mass spectrometry was performed on a linear ion trap (LTQ, Thermo Electron Corp., San Jose, CA). Peptide mixtures were desalted and concentrated on a Paradigm Platinum Peptide Nanotrap (Michrom Bioresources, Inc., Auburn, CA) precolumn (0.15 × 50 mm, 400-μl volume) and subsequently to a microcapillary column, packed with Magic C18AQ reversed-phase material on a flow splitter (Michrom Bioresources, Inc.) at ~250 nl/min. The samples were subjected to a 60 min (10-40% ACN) gradient and eluted into the microcapillary column set to 2.0 kV. The LTQ was operated in the positive-ion mode using data-dependent acquisition with (collision energy of 29%) on the top four ions detected in the survey scan. An inclusion list representing LRG1 (NCBI: gi|4712536) with m/z of +2 and +3 were included in the method.
MS/MS samples were analyzed using SEQUEST (ThermoFinnigan, San Jose, CA) and X! Tandem http://www.thegpm.org. The search was done using an NCBI reference sequence of the Homo sapiens database (Oct, 2007; 33029 entries including known contaminants). The search parameters were carbamidomethyl-cysteine and oxidized methionine with 2 trypsin miscleavages. Scaffold (version Scaffold-01_05_14, Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identification. Protein probabilities were assigned by the Protein Prophet algorithm . Proteins of interest with fewer than three peptides for ID were verified using manual inspection of product ion spectra in relation to candidate peptide sequences. Peptide candidates were judged as correct if a continuous series of a minimum of four b-or y-type product ions were present, if all product ion peaks were at least 3 times the intensity of background and if all experimental fragment ions could be matched to theoretical fragment ions.
Quantification of LRG1 in Serum
LRG1 concentration in sera from serous ovarian cancer patients and healthy female controls
42.99 +/- 9.88
89.33 +/- 77.90
62.52 +/- 36.53
93.01 +/- 81.50
Concentration of LRG1 in sera collected prior to surgery
Benign Ovarian Mass
We found a moderate correlation between CA125 and LRG1 (r = 0.47, p < 0.0001). In order to examine the diagnostic value of LRG1 in distinguishing patients with benign tumors from those with ovarian cancer, we compared receiver operator curves (ROC) for CA125 alone, LRG1 alone and in combination with CA125 (Figure 1C). The ROC of the combined markers was not significantly different from the ROC of CA125 alone; the area under the curve (AUC) for CA125 alone was 0.88 (95% CI: 0.82, 0.94) and the AUC for CA125 and LRG1 was 0.89 (95% CI: 0.84, 0.96; p = 0.2728). There was no significant improvement in sensitivity when adding LRG1.
Ascites fluid from 29 women with serous ovarian cancer was also tested by ELISA for LRG1 protein and was found to be elevated relative to serum levels with a mean value of 142.28 ± 73.56 μg/ml.
Differential Expression of LRG1 mRNA
Using RT-PCR, we also detected increased LRG1 mRNA expression in ovarian tumors compared to normal ovaries (Figure 2B). Eight tissue samples from patients with stage II or higher serous ovarian cancer and seven normal ovaries were tested. Six of the eight ovarian cancers expressed higher levels of LRG1 mRNA than normal ovaries. As LRG1 is an acute-phase protein, primarily produced in the liver , we used liver mRNA as a positive control.
To control for the possible influence of stromal, endothelial, and blood cells present in tissue samples, we examined LRG1 mRNA expression levels in ovarian cancer and NOSE cell lines by RT-PCR. LRG1 mRNA expression was observed in 7 of the 12 ovarian cancer cell lines tested, but no measurable expression was detected in the 4 immortalized NOSE cell lines (Figure 2C).
Differential Expression of LRG1 Protein
Because surface epithelial cells comprise only a minor fraction of the normal ovary, we also examined the expression of LRG1 protein in cell lines derived from ovarian cancer cells and normal ovarian surface epithelia. In Western blot analysis of cell lines, the ~47 and ~51 kD forms of LRG1 protein were present in both ovarian cancer and NOSE cell lines (Figure 3B); the predominant form detected in all cases was 47 kD. Interestingly, four of the five serous ovarian cancer cell lines, OVCA433, OVCAR3, A2780-S, and A2780-CP expressed predominantly the ~47 kD form of LRG1 and little to none of the ~51 kD protein band. Two other serous ovarian cancer cell lines, CAOV3 and MA148, also expressed high levels of the ~47 kD band, but not the ~51 kD band (data not shown). In addition, the cisplatin-resistant cancer line A2780-CP expressed higher levels of the ~47 kD protein band compared to its cisplatin-sensitive counterpart A2780-S (Figure 3B). No LRG1 protein was detected in the NOSE cell line 1816-686.
To establish whether the multiple iosforms of LRG1 observed by Western blot represent different glycosylation states, we treated purified LRG1 protein and cell-free extracts with the enzyme PNGase F to remove carbohydrate residues from the LRG1 protein backbone. As shown in Figure 3C (left panel), LRG1 purified from human plasma has an apparent molecular weight of ~ 47 kD prior to PNGase F treatment. After digestion, the molecular weight of LRG1 is reduced to ~ 34 kD, indicating protein deglycosylation. Similar results were observed in cell-free extracts of the ovarian cancer cell line SKOV3 and the NOSE cell line 1816-575 (Figure 3C, right panel), where multiple higher molecular weight species were reduced to a single lower molecular weight band upon digestion with PNGase F.
Cellular Localization of LRG1
Using immunocytochemistry, LRG1 protein was detected in the cytoplasm of all 19 cell lines tested; representative examples are shown in Figure 3D. LRG1 also localized to the plasma membrane in most of the ovarian cancer cell lines. Three NOSE cell lines (HIO135, HIO117, and IMCC3) also had moderate amounts of LRG1 localized to the plasma membrane. Punctate cytoplasmic localization was observed in NIH:OVCAR5, HEY, C-13, OV2008, ES-2, and OVCA429 ovarian cancer cell lines and all six of the NOSE cell lines. Consistent with the Western blot, the cisplatin-resistant cancer line A2780-CP demonstrated more intense staining compared to its cisplatin-sensitive counterpart A2780-S (data not shown).
Identification of LRG1 in NIH:OVCAR5 Conditioned media
LRG1 peptides identified by mass spectrometry
LRG1 was also detected in the conditioned media of the NIH:OVCAR5 cells by Western blotting. We observed two major LRG1 protein bands of ~47 and ~51 kD, as well as minor protein bands of ~34/36, ~39/40, and ~65 kD in the NIH:OVCAR5 conditioned media (Figure 4C). By comparison, Western blots of the conditioned media from the NOSE cell line 1816-575 detected major LRG1 protein bands of ~47 and ~39/40 kD as well as a minor protein band of <37 kD. Finally, we used the ELISA to detect LRG1 in the NIH:OVCA5 conditioned media (data not shown). Taken together, these results demonstrate that, in addition to being synthesized in the liver, ovarian cancer cells synthesize and secrete LRG1, and may therefore contribute to the elevated LRG1 levels observed in the sera of the ovarian cancer patients.
We recently identified leucine-rich alpha-2-glycoprotein-1 (LRG1) as one of several proteins overexpressed in the serum of patients with ovarian cancer [8, 9]. In this study, we sought to validate this observation and quantitate the levels of LRG1 in a larger cohort of patients' sera. We have also shown that ovarian cancer cells may directly contribute to the elevated levels of LRG1 observed in patients' sera.
The increased serum LRG1 levels in ovarian cancer patients that we had observed by Western blot in pooled samples , were also evident by ELISA in individual samples. When the initial 114 serum samples were tested by ELISA, serum LRG1 was found to be approximately 2-fold greater in serous ovarian cancer patients' sera compared to sera from healthy control women; however, the variance among the ovarian cancer patient samples resulted in unfavorable estimates of sensitivity and specificity. This led us to explore the levels of serum LRG1 among women with different types of benign and malignant ovarian masses. Using a separate set of 193 patient serum samples obtained immediately prior to surgery for a suspicious adnexal mass, LRG1 values were significantly higher (1.7-fold) in patients with serous and clear cell ovarian cancer compared to those with benign gynecological diseases. Although our gene microarray data showed that LRG1 mRNA expression levels were greatest in low malignant potential tumors, the level of serum LRG1 protein in the LMP tumors was significantly lower than for both serous and clear cell ovarian cancer.
Although a biomarker for the early detection of ovarian cancer would have a greater impact, the ability to distinguish malignant from benign disease prior to surgery could be useful in determining which patients would benefit from treatment by a gynecological oncologist. Recently, a panel of biomarkers was approved by the FDA to aid in the diagnosis of ovarian tumors prior to surgery (OVA1; ). This panel includes CA125 as well as beta-2 microglobulin, apolipoprotein A1, transthyretin and transferrin, but not LRG1.
Though the mean concentration of serum LRG1 in serous ovarian cancer patients differed between samples in the two data sets, differences in serum preparation and storage may have affected the quantity of LRG1 detected. For example, Govorukhina et al.,  recently reported that LRG1 levels were decreased in serum with clotting time of longer than 1 hr. We maintained a strict protocol for sample collection and storage for the samples taken from patients at the University of Minnesota, in order to minimize these types of variations (see Methods), and this likely explains the higher LRG1 values in the second dataset compared with the first set of samples obtained from the GOG.
Initially, LRG1 was classified as an "acute-phase protein" involved in the body's response to bacterial and viral infection , but has since been identified as elevated in a variety of disease states, both malignant and benign, including toxic-shock syndrome , and during inflammatory responses of cystic fibrosis . LRG1 is also increased in serum of patients with hepatocellular carcinoma following therapeutic ablation treatment . Differential expression techniques employing affinity depletion of high abundance proteins and 2 D electrophoresis have found serum LRG1 to be upregulated in lung and pancreatic cancer [35–37]. Proteomic research using 2 D SDS-PAGE to analyze body fluids found LRG1 to be upregulated in cerebrospinal fluid and serum of patients with hydrocephalus and silicosis [19, 38].
We conducted a series of experiments examining ovarian cancer tumor cells as a possible source of serum LRG1. Others have identified LRG1 peptides by mass-spectrometry in the secreted or cell surface fractions of CAOV3 and OVCAR3 serous ovarian cancer cell lines, but not in the clear cell ovarian cancer cell line ES-2 . LRG1 peptides have also been identified in ascites fluid and cells from ovarian cancer patients . Recently, elevated levels of LRG1 have been identified in chemoresistant ovarian tumor tissue , and in immunodepleted serum, using ICAT quantitative proteomics . Additionally, LRG1 peptides have been identified in the conditioned media of prostate cancer [43, 44], and breast cancer cell lines  and in the peritoneal fluid of women with uterine leiomyomas . The production and secretion of LRG1 by tumor cells suggests there may be a more direct relationship between tumor burden and serum levels of LRG1 than for other acute phase proteins secreted only by the liver. For example, although haptoglobin levels are increased in the sera of ovarian cancer patients, no haptoglobin RNA or protein were detected by Ye et al.  in seven ovarian cancer cell lines.
In a limited number of cases, we have analyzed sera from patients prior to surgery and following treatment for ovarian cancer. We have found that serum LRG1 levels appear to be more directly related to tumor burden compared to CA125. For example, in three of six patients with sub-optimal debulking surgery, CA125 levels dropped substantially, while LRG1 levels remained elevated. In six cases, serum LRG1 dropped dramatically post chemotherapy. In five cases, LRG1 levels appeared to rise prior to CA125 levels and the onset of recurrent disease. However, given the very low numbers of patients that we have analyzed to date, the use of LRG1 as a marker for disease recurrence, while tantalizing, is purely speculative.
By immunocytochemistry, LRG1 was localized to the cytoplasm of all of the ovarian cell lines tested, both cancer and normal, and was observed on the plasma membrane of most. The serous papillary ovarian cancer cell line, NIH:OVCAR5, had the most intense plasma membrane staining for LRG1. In addition, this cell line expressed high levels of the ~51 kD LRG1 protein band. The LRG1 sequence contains a predicted transmembrane domain  which overlaps the signal sequence; this may allow for the expression of LRG1 at the cell surface. Alternatively, the localization of LRG1 that we observed on the surface of the NIH:OVCAR5 cells may be indicative of cells in the process of secreting LRG1.
Examination of ovarian tumor extracts and cell lines by Western blot revealed increased expression of LRG1 protein in malignant serous tumors and ovarian cancer cell lines compared to their respective controls, as well as the presence of several isoforms of LRG1, though notably the ~47 kD LRG1 band was most intense in each of the malignant ovarian tumor protein extracts. The presence of numerous isoforms for LRG1 has previously been shown by 2 D SDS-PAGE [8, 11, 32, 34, 36–38], suggesting the presence of multiple glycosylated isoforms of LRG1. Indeed, we showed that glycosidase treatment of LRG1, both purified and in extracts of ovarian cancer and NOSE cell lines, reduced the apparent molecular weight of LRG1 indicating the presence of carbohydrate modifications of the protein backbone.
The N-glycosylation of LRG1 produced by the ovarian cancer cells is consistent with its secretion. The ~51 kD band was found at very low levels in the ovarian cancer tumor extracts and was present in the protein extracts of only a few of the ovarian cancer cell lines. It is possible that this ~51 kD glycoform of LRG1 is secreted by the serous ovarian cancer cells, and may contribute to the elevated levels of LRG1 quantitated by ELISA in the sera of these patients. This hypothesis is supported by our Western blot findings that an ~51 kD band was found in the conditioned media of the NIH:OVCAR5 cells but not the NOSE cells, again suggesting that the ~51 kD glycoform of LRG1 may be preferentially secreted, or aberrantly glycosylated in ovarian cancer.
Glycosylation of serum proteins in cancer states is well documented, and serum glycoproteins are being investigated for use as biomarkers in prostate, breast, lung, ovarian and other gynecologic cancers [49–52]. Glycosylation of surface proteins on ovarian carcinoma cells has been reported to mediate adhesion, migration, and invasion through the ECM . Given that murine LRG1 has been shown to bind to several extracellular matrix proteins, and also TGFβ , a possible role for LRG1 in ovarian cancer progression is intriguing.
Alternatively, LRG1 may be playing a role in apoptosis. We have found that MCF-7 breast cancer cells transfected with LRG1 are more resistant to apoptosis induction than non-transfected cells due to cytoplasmic LRG1 binding cytochrome c and inhibition of Apaf-1 activation (Jemmerson and colleagues, manuscript in preparation). In addition, transformed granulocytic cells transfected with LRG1 were reported by Ai et al.  to be more viable than non-transfected cells when transferred between different media. Likewise, LRG1 may be a survival factor for ovarian cancer cells, possibly rendering them more resistant to chemotherapy. It is interesting to note that the cisplatin-resistant A2780-CP cells express higher levels of LRG1 protein than their more sensitive counterparts A2780-S (Figure 3B); however, no difference in LRG1 protein expression was found for the cisplatin resistant cell line C13, compared to the corresponding cisplatin sensitive cell line OV2008.
We have demonstrated the potential for using LRG1 as a serum biomarker for ovarian cancer. Furthermore, we showed the expression of LRG1 mRNA and protein in ovarian cancer tissues and cell lines, signifying that the tumor cells could be contributing to the increased levels of LRG1 in sera of ovarian cancer patients. Though future studies using a larger patient cohort are needed to determine whether LRG1 may serve as a biomarker for presurgical diagnosis of ovarian cancer, for the detection of recurrent disease, and/or as a target for therapeutic treatment, these initial results are encouraging.
We would like to thank Dr. Patricia Kruk (University of South Florida, Tampa, FL), Dr. Barbara Vanderhyden (University of Ottawa, Ottawa, Ontario, Canada), Dr. Judah Folkman (Department of Vascular Biology, Boston Children's Hospital, Boston, MA), Dr. Robert Bast Jr. (University of Texas, Houston, TX), and Dr. Sundaram Ramakrishnan (University of Minnesota, Minneapolis, MN) for the cell lines; Sarah Bowell, Diane Rauch, and Marissa Mackey of the University of Minnesota Tissue Procurement Facility for providing tissue, blood, and ascites samples; the Gynecologic Oncology Group Tissue Bank for the serum samples; Robin Bliss of the Masonic Cancer Center's Biostatistics Core Facility; the Minnesota Supercomputing Institute; and the staff of Gene Logic Inc., Gaithersburg, MD, for performing the gene expression experiments with the human tissue samples. This work was supported by grants from the Minnesota Ovarian Cancer Alliance (APNS), National Institutes of Health/National Cancer Institute R01-CA106878 (APNS), and Cancurables (APNS).
- Ries L, Harkins D, Krapcho M, Mariotto A, Miller BA, Feuer EJ, Clegg L, Eisner MP, Horner MJ, Howlader N, Hayat M, Hankey BF, Edwards BK: SEER Cancer Statistics Review, 1975–2003. Book SEER Cancer Statistics Review, 1975–2003 (Editor ed.^eds.). City 2005.
- Jemal A, Siegel R, Ward E, Hao Y, Xu J, Murray T, Thun MJ: Cancer statistics, 2008. CA Cancer J Clin 2008, 58: 71–96. 10.3322/CA.2007.0010View ArticlePubMed
- Badgwell D, Bast RC Jr: Early detection of ovarian cancer. Dis Markers 2007, 23: 397–410.PubMed CentralView ArticlePubMed
- Yurkovetsky ZR, Linkov FY, D EM, Lokshin AE: Multiple biomarker panels for early detection of ovarian cancer. Future Oncol 2006, 2: 733–741. 10.2217/14796622.214.171.1243View ArticlePubMed
- Fung ET: A recipe for proteomics diagnostic test development: the OVA1 test, from biomarker discovery to FDA clearance. Clin Chem 56: 327–329. 10.1373/clinchem.2009.140855
- Mercado C, Zingmond D, Karlan BY, Sekaris E, Gross J, Maggard-Gibbons M, Tomlinson JS, Ko CY: Quality of care in advanced ovarian cancer: the importance of provider specialty. Gynecol Oncol 117: 18–22. 10.1016/j.ygyno.2009.12.033
- Anderson NL, Anderson NG: The human plasma proteome: history, character, and diagnostic prospects. Mol Cell Proteomics 2002, 1: 845–867. 10.1074/mcp.R200007-MCP200View ArticlePubMed
- Andersen JD, Boylan KLM, Xue FS, Anderson LB, Witthuhn BA, Markowski TW, Higgins L, Skubitz APN: Identification of candidate biomarkers in ovarian cancer serum by depletion of highly abundant proteins and differential in-gel electrophoresis. Electrophoresis 2010., 31: 10.1002/elps.200900441
- Boylan KLM, Andersen JD, Anderson LB, Higgins L, Skubitz APN: Quantitative proteomic analysis by iTRAQ®for the identification of candidate biomarkers in ovarian cancer serum. Proteome Sci 2010, 8: 31. 10.1186/1477-5956-8-31PubMed CentralView ArticlePubMed
- Schwick HG, Haupt H: Purified human plasma proteins of unknown function. Jpn J Med Sci Biol 1981, 34: 299–327.View ArticlePubMed
- Hoogland C, Mostaguir K, Sanchez JC, Hochstrasser DF, Appel RD: SWISS-2DPAGE, ten years later. Proteomics 2004, 4: 2352–2356. 10.1002/pmic.200300830View ArticlePubMed
- Haupt H, Baudner S: [Isolation and characterization of an unknown, leucine-rich 3.1-S-alpha2-glycoprotein from human serum (author's transl)]. Hoppe Seylers Z Physiol Chem 1977, 358: 639–646.View ArticlePubMed
- Weivoda S, Andersen JD, Skogen A, Schlievert PM, Fontana D, Schacker T, Tuite P, Dubinsky JM, Jemmerson R: ELISA for human serum leucine-rich alpha-2-glycoprotein-1 employing cytochrome c as the capturing ligand. J Immunol Methods 2008, 336: 22–29. 10.1016/j.jim.2008.03.004View ArticlePubMed
- Kobe B, Kajava AV: The leucine-rich repeat as a protein recognition motif. Curr Opin Struct Biol 2001, 11: 725–732. 10.1016/S0959-440X(01)00266-4View ArticlePubMed
- Takahashi N, Takahashi Y, Putnam FW: Periodicity of leucine and tandem repetition of a 24-amino acid segment in the primary structure of leucine-rich alpha 2-glycoprotein of human serum. Proc Natl Acad Sci USA 1985, 82: 1906–1910. 10.1073/pnas.82.7.1906PubMed CentralView ArticlePubMed
- O'Donnell LC, Druhan LJ, Avalos BR: Molecular characterization and expression analysis of leucine-rich alpha2-glycoprotein, a novel marker of granulocytic differentiation. J Leukoc Biol 2002, 72: 478–485.PubMed
- Saito K, Tanaka T, Kanda H, Ebisuno Y, Izawa D, Kawamoto S, Okubo K, Miyasaka M: Gene expression profiling of mucosal addressin cell adhesion molecule-1+ high endothelial venule cells (HEV) and identification of a leucine-rich HEV glycoprotein as a HEV marker. J Immunol 2002, 168: 1050–1059.View ArticlePubMed
- Sun D, Kar S, Carr BI: Differentially expressed genes in TGF-beta 1 sensitive and resistant human hepatoma cells. Cancer Lett 1995, 89: 73–79.View ArticlePubMed
- Li X, Miyajima M, Jiang C, Arai H: Expression of TGF-betas and TGF-beta type II receptor in cerebrospinal fluid of patients with idiopathic normal pressure hydrocephalus. Neurosci Lett 2007, 413: 141–144. 10.1016/j.neulet.2006.11.039View ArticlePubMed
- Cummings C, Walder J, Treeful A, Jemmerson R: Serum leucine-rich alpha-2-glycoprotein-1 binds cytochrome c and inhibits antibody detection of this apoptotic marker in enzyme-linked immunosorbent assay. Apoptosis 2006, 11: 1121–1129. 10.1007/s10495-006-8159-3View ArticlePubMed
- Ai J, Druhan LJ, Hunter MG, Loveland MJ, Avalos BR: LRG-accelerated differentiation defines unique G-CSFR signaling pathways downstream of PU.1 and C/EBPepsilon that modulate neutrophil activation. J Leukoc Biol 2008, 83: 1277–1285. 10.1189/jlb.1107751PubMed CentralView ArticlePubMed
- Hibbs K, Skubitz KM, Pambuccian SE, Casey RC, Burleson KM, Oegema TR Jr, Thiele JJ, Grindle SM, Bliss RL, Skubitz AP: Differential gene expression in ovarian carcinoma: identification of potential biomarkers. Am J Pathol 2004, 165: 397–414.PubMed CentralView ArticlePubMed
- Skubitz AP, Pambuccian SE, Argenta PA, Skubitz KM: Differential gene expression identifies subgroups of ovarian carcinoma. Transl Res 2006, 148: 223–248. 10.1016/j.trsl.2006.06.001View ArticlePubMed
- DeRycke MS, Andersen JD, Harrington KM, Pambuccian SE, Kalloger SE, Boylan KL, Argenta PA, Skubitz AP: S100A1 expression in ovarian and endometrial endometrioid carcinomas is a prognostic indicator of relapse-free survival. Am J Clin Pathol 2009, 132: 846–856. 10.1309/AJCPTK87EMMIKPFSPubMed CentralView ArticlePubMed
- Subramanian IV, Bui Nguyen TM, Truskinovsky AM, Tolar J, Blazar BR, Ramakrishnan S: Adeno-associated virus-mediated delivery of a mutant endostatin in combination with carboplatin treatment inhibits orthotopic growth of ovarian cancer and improves long-term survival. Cancer Res 2006, 66: 4319–4328. 10.1158/0008-5472.CAN-05-3297View ArticlePubMed
- Shaw TJ, Senterman MK, Dawson K, Crane CA, Vanderhyden BC: Characterization of intraperitoneal, orthotopic, and metastatic xenograft models of human ovarian cancer. Mol Ther 2004, 10: 1032–1042. 10.1016/j.ymthe.2004.08.013View ArticlePubMed
- Kruk PA, Maines-Bandiera SL, Auersperg N: A simplified method to culture human ovarian surface epithelium. Lab Invest 1990, 63: 132–136.PubMed
- Nicosia SV, Wilbanks GD, Saunders B, Mayer J, Cardosi RJ, Kruk PA, Cheng J, Bai W, Coppola D, Fiorica J: Cytology of human ovarian surface epithelial brushings. Cancer 2004, 102: 1–10. 10.1002/cncr.20001View ArticlePubMed
- Shirai R, Hirano F, Ohkura N, Ikeda K, Inoue S: Up-regulation of the expression of leucine-rich alpha(2)-glycoprotein in hepatocytes by the mediators of acute-phase response. Biochem Biophys Res Commun 2009, 382: 776–779. 10.1016/j.bbrc.2009.03.104View ArticlePubMed
- Keller A, Nesvizhskii AI, Kolker E, Aebersold R: Empirical statistical model to estimate the accuracy of peptide identifications made by MS/MS and database search. Anal Chem 2002, 74: 5383–5392. 10.1021/ac025747hView ArticlePubMed
- Govorukhina NI, de Vries M, Reijmers TH, Horvatovich P, van der Zee AG, Bischoff R: Influence of clotting time on the protein composition of serum samples based on LC-MS data. J Chromatogr B Analyt Technol Biomed Life Sci 2009, 877: 1281–1291. 10.1016/j.jchromb.2008.10.029View ArticlePubMed
- Bini L, Magi B, Marzocchi B, Cellesi C, Berti B, Raggiaschi R, Rossolini A, Pallini V: Two-dimensional electrophoretic patterns of acute-phase human serum proteins in the course of bacterial and viral diseases. Electrophoresis 1996, 17: 612–616. 10.1002/elps.1150170333View ArticlePubMed
- Norkina O, Kaur S, Ziemer D, De Lisle RC: Inflammation of the cystic fibrosis mouse small intestine. Am J Physiol Gastrointest Liver Physiol 2004, 286: G1032–1041. 10.1152/ajpgi.00473.2003View ArticlePubMed
- Kawakami T, Hoshida Y, Kanai F, Tanaka Y, Tateishi K, Ikenoue T, Obi S, Sato S, Teratani T, Shiina S, et al.: Proteomic analysis of sera from hepatocellular carcinoma patients after radiofrequency ablation treatment. Proteomics 2005, 5: 4287–4295. 10.1002/pmic.200401287View ArticlePubMed
- Yu KH, Rustgi AK, Blair IA: Characterization of proteins in human pancreatic cancer serum using differential gel electrophoresis and tandem mass spectrometry. J Proteome Res 2005, 4: 1742–1751. 10.1021/pr050174lView ArticlePubMed
- Kakisaka T, Kondo T, Okano T, Fujii K, Honda K, Endo M, Tsuchida A, Aoki T, Itoi T, Moriyasu F, et al.: Plasma proteomics of pancreatic cancer patients by multi-dimensional liquid chromatography and two-dimensional difference gel electrophoresis (2D-DIGE): up-regulation of leucine-rich alpha-2-glycoprotein in pancreatic cancer. J Chromatogr B Analyt Technol Biomed Life Sci 2007, 852: 257–267. 10.1016/j.jchromb.2007.01.029View ArticlePubMed
- Okano T, Kondo T, Kakisaka T, Fujii K, Yamada M, Kato H, Nishimura T, Gemma A, Kudoh S, Hirohashi S: Plasma proteomics of lung cancer by a linkage of multi-dimensional liquid chromatography and two-dimensional difference gel electrophoresis. Proteomics 2006, 6: 3938–3948. 10.1002/pmic.200500883View ArticlePubMed
- Zeng JW, Wang SX, Zhao XF, Wei MT, Tu ZG: [Utilizing 2-DE and MALDI-TOF MS/MS to screen differentially expressed serum proteins of silicosis]. Zhonghua Lao Dong Wei Sheng Zhi Ye Bing Za Zhi 2007, 25: 136–141.PubMed
- Faca VM, Ventura AP, Fitzgibbon MP, Pereira-Faca SR, Pitteri SJ, Green AE, Ireton RC, Zhang Q, Wang H, O'Briant KC, et al.: Proteomic analysis of ovarian cancer cells reveals dynamic processes of protein secretion and shedding of extra-cellular domains. PLoS ONE 2008, 3: e2425. 10.1371/journal.pone.0002425PubMed CentralView ArticlePubMed
- Gortzak-Uzan L, Ignatchenko A, Evangelou AI, Agochiya M, Brown KA, St Onge P, Kireeva I, Schmitt-Ulms G, Brown TJ, Murphy J, et al.: A proteome resource of ovarian cancer ascites: integrated proteomic and bioinformatic analyses to identify putative biomarkers. J Proteome Res 2008, 7: 339–351. 10.1021/pr0703223View ArticlePubMed
- Pan S, Cheng L, White JT, Lu W, Utleg AG, Yan X, Urban ND, Drescher CW, Hood L, Lin B: Quantitative proteomics analysis integrated with microarray data reveals that extracellular matrix proteins, catenins, and p53 binding protein 1 are important for chemotherapy response in ovarian cancers. Omics 2009, 13: 345–354. 10.1089/omi.2009.0008View ArticlePubMed
- Lin B, White JT, Wu J, Lele S, Old LJ, Hood L, Odunsi K: Deep depletion of abundant serum proteins reveals low-abundant proteins as potential biomarkers for human ovarian cancer. Proteomics Clin Appl 2009, 3: 853–861. 10.1002/prca.200800141PubMed CentralView ArticlePubMed
- Sardana G, Jung K, Stephan C, Diamandis EP: Proteomic analysis of conditioned media from the PC3, LNCaP, and 22Rv1 prostate cancer cell lines: discovery and validation of candidate prostate cancer biomarkers. J Proteome Res 2008, 7: 3329–3338. 10.1021/pr8003216View ArticlePubMed
- Sardana G, Marshall J, Diamandis EP: Discovery of candidate tumor markers for prostate cancer via proteomic analysis of cell culture-conditioned medium. Clin Chem 2007, 53: 429–437. 10.1373/clinchem.2006.077370View ArticlePubMed
- Kulasingam V, Diamandis EP: Tissue culture-based breast cancer biomarker discovery platform. Int J Cancer 2008, 123: 2007–2012. 10.1002/ijc.23844View ArticlePubMed
- Ferrero S, Gillott DJ, Remorgida V, Anserini P, Ragni N, Grudzinskas JG: Increased expression of one isoform of leucine-rich alpha-2-glycoprotein in peritoneal fluid of women with uterine leiomyomas. Arch Gynecol Obstet 2009, 279: 365–371. 10.1007/s00404-008-0741-1View ArticlePubMed
- Ye B, Cramer DW, Skates SJ, Gygi SP, Pratomo V, Fu L, Horick NK, Licklider LJ, Schorge JO, Berkowitz RS, Mok SC: Haptoglobin-alpha subunit as potential serum biomarker in ovarian cancer: identification and characterization using proteomic profiling and mass spectrometry. Clin Cancer Res 2003, 9: 2904–2911.PubMed
- Hirokawa T, Boon-Chieng S, Mitaku S: SOSUI: classification and secondary structure prediction system for membrane proteins. Bioinformatics 1998, 14: 378–379. 10.1093/bioinformatics/14.4.378View ArticlePubMed
- An HJ, Lebrilla CB: A glycomics approach to the discovery of potential cancer biomarkers. Methods Mol Biol 600: 199–213. full_text
- Saldova R, Royle L, Radcliffe CM, Abd Hamid UM, Evans R, Arnold JN, Banks RE, Hutson R, Harvey DJ, Antrobus R, et al.: Ovarian cancer is associated with changes in glycosylation in both acute-phase proteins and IgG. Glycobiology 2007, 17: 1344–1356. 10.1093/glycob/cwm100View ArticlePubMed
- Saldova R, Wormald MR, Dwek RA, Rudd PM: Glycosylation changes on serum glycoproteins in ovarian cancer may contribute to disease pathogenesis. Dis Markers 2008, 25: 219–232.PubMed CentralView ArticlePubMed
- Fujimura T, Shinohara Y, Tissot B, Pang PC, Kurogochi M, Saito S, Arai Y, Sadilek M, Murayama K, Dell A, et al.: Glycosylation status of haptoglobin in sera of patients with prostate cancer vs. benign prostate disease or normal subjects. Int J Cancer 2008, 122: 39–49. 10.1002/ijc.22958View ArticlePubMed
- Casey RC, Oegema TR Jr, Skubitz KM, Pambuccian SE, Grindle SM, Skubitz AP: Cell membrane glycosylation mediates the adhesion, migration, and invasion of ovarian carcinoma cells. Clin Exp Metastasis 2003, 20: 143–152. 10.1023/A:1022670501667View ArticlePubMed
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.