A suppression subtractive cDNA library representing mRNAs expressed at a higher level in the malignant human breast cancer cell line, MCF-7, relative to a benign breast tumor-derived cell line, Huma 123, contained a cDNA, M36, which was expressed in estrogen receptor α (ERα)–positive breast carcinoma cell lines but not in cell lines from normal/benign/ERα-negative malignant breast lesions. M36 cDNA had an identical coding sequence to anterior gradient 2 (AGR2), the human homologue of the cement gland–specific gene (Xenopus laevis). Screening of breast tumor specimens using reverse transcription-PCR and immunocytochemistry with affinity-purified anti-AGR2 antibodies showed that the presence of AGR2 mRNA and protein were both statistically significantly associated with ERα-positive carcinomas (P = 0.007, Fisher's exact test) and with malignancy (P ≤ 0.025). When an expression vector for AGR2 cDNA was introduced into benign nonmetastatic rat mammary tumor cells, and three separate clones and two pools of cells were transferred to the mammary glands of syngeneic hosts, there were no consistent differences in the mean latent periods of tumor formation. However, metastases occurred in the lungs of animals receiving the AGR2 transfectants in 77% to 92% of animals with primary tumors (P = 0.0001) compared with no metastases in the control groups. The AGR2 transfectants exhibited enhanced rates of adhesion to a plastic substratum and extracellular AGR2 enhanced the rate of attachment of AGR2-negative but not AGR2-positive cells. These experiments are the first to link mechanistically the developmental gene product, AGR2, with metastasis in vivo.

Altered levels of gene products are thought to be associated with the changed behavior of cancer cells (1). Modern array techniques are now being used to relate patterns of gene expression to both pathologic appearance and malignant behavior of breast cancer (1, 2) and other cancers (35). However, it is also important to identify the key changes in gene expression that cause the malignant properties of cancer cells. Such an approach has led to the discovery of the metastasis-inducing proteins S100A4 (6, 7) and osteopontin (8) that have been shown to have such a dramatic effect on patient survival in a group of breast cancer patients (9, 10).

PCR-selected suppression subtractive hybridization (11) has been used to identify cDNAs representing mRNAs differentially expressed (1214) between an estrogen receptor α (ERα)–negative benign human mammary epithelial cell line, Huma 123 (15), and the ERα-positive malignant human mammary epithelial cell line, MCF-7 (16). The resulting subtracted libraries contained well-characterized, differentially expressed cDNAs that have been associated previously with tumor progression (12). In this article, we report one novel cloned cDNA, M36, which matches identically the coding sequence of human anterior gradient 2 (AGR2) cDNA, the human homologue (previously hAG-2) of the Xenopus laevis cement gland–specific gene, XAG-2 (17). The XAG-2 gene product has developmental significance in Xenopus embryos (18). Here, we show that the human homologue of this developmentally associated protein is differentially expressed between benign and malignant human breast carcinoma specimens. Furthermore, its cDNA, when introduced into a benign, nonmetastatic, rat mammary cell line, confers a metastatic phenotype on benign nonmetastatic cells.

Cell lines and cell culture. The normal human mammary epithelial cell line, Huma 7, was subcloned from primary cultures of reduction mammoplasty specimens of normal breast tissue immortalized with SV40 (19). The benign human mammary epithelial cell line, Huma 123, and the derivative benign human mammary myoepithelial-like cell line, Huma 109 (15), were derived from HMT-3522, itself obtained from a primary cell culture of human benign breast disease displaying prominent epithelial hyperplasia (20). These cell lines, and the malignant human mammary epithelial cell lines derived from pleural effusions of breast cancer patients, MCF-7, T47D, ZR-75, and MDA-MB-231 (21), were cultured as described previously (12, 15, 19). The benign rat mammary epithelial cell line, Rama 37, was cultured as described previously (22). The transfected derivative cell lines were grown in medium containing 1 mg/mL geneticin. All cells were passaged on reaching 70% confluency. Culture in medium depleted of steroid hormones was carried out as described previously (23, 24).

Subtractive hybridization and Northern hybridization screening. A suppression subtractive (11) library consisting of PCR products representing mRNAs expressed at a higher level in the malignant breast epithelial cell line, MCF-7, relative to a benign human breast-derived cell line, Huma 123, was constructed using a PCR-Select cDNA Subtraction kit (Clontech, Palo Alto, CA) as described previously (14). Reverse Northern screening of the subtracted cDNA library was carried out as described previously (12). Total cellular RNA was prepared using the guanidinium-isothiocyanate-cesium chloride method (2527). Poly(A)-containing RNA was isolated from total RNA using the Fast Track mRNA isolation kit (Invitrogen, Groningen, the Netherlands). Northern hybridization procedures were done as described previously (12). cDNA probes were radioactively labeled to 1 × 109 dpm/μg DNA by random-primed DNA synthesis (28) using a labeling kit (Roche Molecular Biochemicals, Mannheim, Germany). The constitutive probe, 36B4, a cDNA to human acidic ribosomal phosphoprotein PO mRNA (29), was used to normalize RNA loading on the gel.

Coupled transcription and translation assay in vitro. Transcription and translation assays in vitro were carried out using a TNT T7/T3-coupled reticulocyte lysate system (Promega, Madison, WI) to produce a protein product labeled with [35S]methionine in vitro. DNA template (2 μg) was transcribed and translated in 50 μL containing 40 units RNAsin RNase inhibitor, 25 μL TNT rabbit reticulocyte lysate, 20 μmol/L amino acid mixture without methionine, 20 units (T3 or T7) RNA polymerase, 20 μCi [35S]methionine (>1,000 Ci/mmol at 10 mCi/mL), and 2 μL TNT reaction buffer. The mixture was incubated at 30°C for 90 minutes. The resulting 35S-labeled proteins and nonradioactive standards were fractionated on urea-containing SDS, 15% (w/v) polyacrylamide gels (SDS-PAGE) with 6% (w/v) polyacrylamide stacking gels (30). The gels were stained with Coomassie blue, destained with 40% (v/v) methanol, 7% (v/v) acetic acid, dried under vacuum, and autoradiographed with Kodak X-Omat film (Eastman Kodak, Rochester, NY) at −70°C for 3 to 10 days.

Production and purification of recombinant protein anterior gradient 2 and its antiserum. The full-length M36 cDNA or one with the M36 signal sequence deleted was cloned into the expression vector, pET-16b (Novagen, Madison, WI), downstream of the His tag, to yield a recombinant cDNA construct designated pET-M36, which was first verified by automated DNA sequencing and then transformed into Escherichia coli BL21DE3 cells. Induction of recombinant protein was carried out by adding isopropyl-l-thio-β-d-galactopyranoside (1 mmol/L) to the culture medium (A600 = 0.5) for 2 hours. Purification of recombinant AGR2 protein to a single band on SDS-PAGE gel was carried out using His-Bind resin (Novagen). The amino acid sequence of the purified recombinant AGR2 protein was confirmed by an automated sequencer. The production of rabbit anti-AGR2 serum was conducted by Eurogentec (Seraing, Belgium). The anti-AGR2 antibodies were affinity purified by their binding to antigen immobilized on a polyvinylidene difluoride (PVDF) membrane. Briefly, recombinant AGR2 (1 mg) was subjected to SDS-PAGE and electrophoretically transferred to a PVDF membrane and the part of the membrane containing the immobilized antigen was incubated with serum from a rabbit immunized with recombinant AGR2. Bound antibody was eluted with a 100 mmol/L glycine buffer (pH 2.5) followed by neutralization with 1 mol/L Tris buffer (pH 8.0).

Western blot analysis. Cells were grown to 70% to 80% confluence, washed twice with ice-cold PBS buffer, and lysed in lysis buffer [50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1% (v/v) NP40, 0.5% (w/v) sodium deoxycholate, 1 mmol/L EDTA, 0.1% (w/v) SDS], and a protease inhibitor cocktail tablet (Roche Molecular Biochemicals) was added. The cleared lysates were collected by centrifugation at 12,000 × g for 20 minutes at 4°C. The protein concentration in the lysate was measured by Bio-Rad protein assay (Bio-Rad Laboratories, Hemel Hempstead, Herts, United Kingdom). Lysates containing equal amounts of total proteins were resolved by SDS-PAGE. The proteins were electrotransferred onto PVDF membranes using a Bio-Rad semidry transfer apparatus. The membranes were incubated with the affinity-purified, in-house rabbit polyclonal anti-AGR2 antibody. After washing and incubating with anti-rabbit horseradish peroxidase–conjugated IgG, the membranes were washed and detected by the Supersignal West Pico Chemiluminescent Substrate (Pierce Biotechnology, Inc., Perbio Science, Cramlington, Northumberland, United Kingdom) according to the manufacturer's instructions. The membranes were reprobed with mouse monoclonal anti-actin antibody (Sigma, Poole, Dorset, United Kingdom) to ensure equal protein loading.

Reverse transcription-PCR. Total RNA (2 μg) was reverse transcribed in 10 μL with 200 units SuperScript RNase H reverse transcriptase (Invitrogen Ltd., Paisley, United Kingdom). Subsequently, the first-strand cDNA reaction mixture (1 μL) was amplified by PCR with Taq DNA polymerase (Invitrogen). For M36 cDNA, the forward primer (5′ position at nucleotide 87, Genbank accession no. NM_006408.2) was 5′-GCTCCTTGTGGCCCTCTCCTACAC-3′ and the reverse primer (5′ position at nucleotide 440, Genbank accession no. NM_006408.2) was 5′-ATCCTGGGGACATACTGGCCATCAG-3′. For the human glyceraldehyde-3-phosphate dehydrogenase cDNA, the forward primer 5′-ACCACAGTCCATGCCATCAC-3′ and the reverse primer 5′-TCCACCACCCTGTTGCTGTA-3′ were used to provide a normalization control. PCR was done as follows: 94°C for 3 minutes followed by 25 cycles at 94°C for 30 seconds, 60°C for 30 seconds, and 72°C for 1.5 minutes. PCR products were visualized with ethidium bromide following agarose gel electrophoresis (31).

Transfection of plasmid DNA into rat cells. Exponentially growing benign rat mammary epithelial Rama 37 cells were harvested, seeded at a density of 0.5 to 0.7 × 106 cells per 9 cm diameter culture dish in routine medium, and incubated for 24 hours at 37°C. pcDNA-M36 construct (5 μg) containing full-length M36 cDNA, which had been verified by DNA sequencing, was transfected into Rama 37 cells using calcium phosphate as described previously (32). Empty pcDNA vector (5 μg) was transfected into Rama 37 cells as a negative control. Colonies were visible after ∼7 days following selection in 1 mg/mL geneticin. Three single colonies and two pools of cells were picked, expanded, and subsequently transferred several times before being frozen for storage.

Tumorigenesis and metastasis. Cultured cells were harvested by trypsinisation and centrifugation and washed twice before being resuspended in cold PBS (4°C) at a concentration of 107 viable cells/mL. Female Ludwig Furth-Wistar rats (5-6 weeks old), maintained on expanded rat and mouse diet no.1 (B.P. Ltd., Essex, United Kingdom) and tap water ad libitum, were injected s.c. with 2 × 106 viable cells at the site of the left or right inguinal mammary fat pad. All rats were observed at 3- to 4-day intervals up to 5 months. Tumor-bearing rats were necropsied when their primary tumors reached 10% of body mass or earlier if the tumors ulcerated or caused serious morbidity. The lungs, liver, axillary lymph nodes, spleen, kidney, and heart were examined for gross metastases. Samples of the primary tumors and lungs with abnormal appearance were fixed in Methacarn [methanol/trichloroethane/acetic acid 6:3:1 (v/v)] and processed for histology (7). Animal experiments were conducted according to the United Kingdom Coordinating Committee for Cancer Research guidelines and the New York Academy of Sciences Committee on Animal Research under Home Office Project Licenses PPL 40/1515 and 40/2395 to Prof. Philip S. Rudland.

Human breast specimens. Human breast specimens, normal specimens from reduction mammoplasties, benign fibroadenomas, and invasive ductal carcinoma of no special type were obtained from the Cancer Tissue Bank Research Centre (Liverpool, United Kingdom) with full and informed patient consent and with ethical approval. The carcinomas were subdivided into two groups based on immunocytochemical staining for ERα, a cutoff at 5% of the carcinoma cells stained by antibodies to ERα divided the negative from the positive group.

Histology and immunocytochemistry. The histology of 4 μm tissue sections was determined after staining with H&E. Immunocytochemical staining for vimentin, skeletal muscle actin, myoglobin, and ERα was carried out as described previously (33, 34). Immunocytochemical staining for AGR2 was done with affinity-purified AGR2 antibodies with or without prior incubation with 0.1 mg/mL recombinant AGR2 protein either for sections of human specimens [1:500 dilution in PBS buffer containing 2% (w/v) bovine serum albumin incubated at room temperature for 2 hours] or for sections of rat specimens [1:200 dilution in PBS buffer containing 0.5% (w/v) bovine serum albumin incubated at room temperature overnight]. The bound antibodies were detected using biotinylated donkey anti-rabbit serum followed by ABC complex/horseradish peroxidase kit (DAKO Ltd., Cambridgeshire, United Kingdom). The sections were visualized as a brown stain by incubating with 3,3′-diaminobenzidine (Sigma, Dorset, United Kingdom) and 0.075% (v/v) H2O2, counterstained with Mayer's hemalum, and mounted in dibutyl polystyrene xylene (Merck, Dorset, United Kingdom). All staining results were examined by three independent observers and scored as plus/minus using 5% of carcinoma cells staining as a cutoff. Photography was carried out as described previously (33).

Cell adhesion assays. The cells were grown to 70% to 80% confluence, washed twice with PBS, trypsinized, and counted using a Coulter counter (Beckman-Coulter UK Ltd., High Wycombe, Buckinghamshire, United Kingdom). The cells were resuspended at 2 × 105 cells/mL and counted again to check the concentration before adding 1 mL cells to each well of a 24-well plate. After incubation for 30 minutes at 37°C, the cells were washed thrice with PBS buffer to remove any cells in suspension, and cells adhering to the wells were trypsinized and counted. The number of adherent cells was calculated as the percentage of the total number of cells that had adhered after 30 minutes. In some experiments, the cell culture plates were coated with either 2 or 20 μg of recombinant AGR2 lacking the signal sequence before carrying out cell attachment assays as described above. Three independent experiments were carried out using triplicate wells and error bars represent the SDs of the means of the three separate experiments.

Statistical analysis. Statistical analyses were done by the two-tailed Fisher's exact test or Mann-Whitney U-test using Arcus Pro-Stat Dos version 3.28 software (Medical Computing, Aughton, United Kingdom).

Overexpressed M36 cDNA in the malignant human breast cancer cell lines corresponds to anterior gradient 2 mRNA. A suppression subtractive library was constructed containing cDNAs expressed at a higher level in the malignant mammary cell line, MCF-7, than in the cell line, Huma 123, derived from a benign mammary lesion (Materials and Methods). Four cloned cDNAs (M36, M40, M202, and M234) of 174 cloned cDNAs sequenced each exhibited 100% identity to the coding sequence of human cDNA AGR2 (Genbank accession no. NM_006408). AGR2 is the human homologue of the X. laevis AGR2 (XAG-2) gene, which is expressed by the cement gland of the developing Xenopus embryo (18). The nucleotide sequences for M36 or M202 cDNA contained an open reading frame of 175 amino acids that was identical in amino acid sequence to that of AGR2 protein (Genbank accession no. NP_006399). In a T7/T3 RNA polymerase–coupled transcription/translation system, clones, M36 and M202, but not empty vector, yielded a single 35S-labeled primary translation product of 20 kDa on urea-containing SDS-PAGE (data not shown), a size that corresponds to the 20 kDa derived from the amino acid sequence.

Quantitative reverse Northern hybridizations using as probes double-stranded mixed cDNAs showed that the level of AGR2 mRNA was >15-fold higher in the RNA from MCF-7 cells than in that from the Huma 123 cells (data not shown). Northern hybridization experiments showed that the M36 probe hybridized to a major band of RNA with a molecular size of 0.9 kb (mean of three independent experiments) corresponding to the AGR2 mRNA and to an additional faint band at 1.6 kb in all the positive lanes (Fig. 1). The 1.6-kb band corresponds in size to the recently updated 1.7-kb variant mRNA of AGR2 containing a longer untranslated 3′-end (Genbank accession no. NM-006408). The AGR2 mRNA was present in all three ERα-positive breast cancer cell lines tested, MCF-7, T47D, and ZR-75, but undetectable in the ERα-negative MDA-MB-231 breast cancer cell line and in the SV40-immortalized normal human breast cell line, Huma 7, the benign human breast tumor cell line, Huma 123, and its myoepithelial-like convert, Huma 109 (Fig. 1). The same distribution was found for the AGR2 protein using Western blotting with the AGR2 antibody (see Materials and Methods; Fig. 1), the signal being abolished by prior incubation of the antibody with 0.1 mg/mL recombinant AGR2 (data not shown). These results suggest that the expression of AGR2 mRNA and protein correlates with the presence of ERα at least in these cell lines.

Figure 1.

Northern and Western blots of AGR2 mRNA and protein in human mammary cell lines derived from benign and malignant breast tumors. A, total RNAs isolated from the SV40-immortalized normal human mammary cell line, Huma 7, the benign human mammary derived cell line, Huma 123, a benign myoepithelial-like convertant cell line, Huma 109, derived from Huma 123, the ER-negative, malignant, human mammary epithelial cell line, MDA-MB-231, and the ER-positive, human mammary cell lines MCF-7, T47D, and ZR-75, were subjected to Northern hybridization (Materials and Methods) using a 32P-labeled probe to AGR2 mRNA (A, top) or to the mRNA for ribosomal phosphoprotein cDNA, 36B4 (A, bottom) for normalization purposes. B, protein (10 μg) from Huma 7, Huma 123, MCF-7, T47D, ZR-75, and MDA-MB-231 or recombinant, His-tagged AGR2 (1 μg) was subjected to SDS-PAGE and blotted onto PVDF membranes as described in Materials and Methods. The membranes were incubated with antibodies to actin (B, top) or AGR2 (B, bottom) and visualized by chemiluminescence. nAGR2, natural AGR2; rAGR2, recombinant AGR2. The recombinant AGR2 is larger than the natural AGR2 due to the presence of a histidine tag and a protease factor X cleavage site.

Figure 1.

Northern and Western blots of AGR2 mRNA and protein in human mammary cell lines derived from benign and malignant breast tumors. A, total RNAs isolated from the SV40-immortalized normal human mammary cell line, Huma 7, the benign human mammary derived cell line, Huma 123, a benign myoepithelial-like convertant cell line, Huma 109, derived from Huma 123, the ER-negative, malignant, human mammary epithelial cell line, MDA-MB-231, and the ER-positive, human mammary cell lines MCF-7, T47D, and ZR-75, were subjected to Northern hybridization (Materials and Methods) using a 32P-labeled probe to AGR2 mRNA (A, top) or to the mRNA for ribosomal phosphoprotein cDNA, 36B4 (A, bottom) for normalization purposes. B, protein (10 μg) from Huma 7, Huma 123, MCF-7, T47D, ZR-75, and MDA-MB-231 or recombinant, His-tagged AGR2 (1 μg) was subjected to SDS-PAGE and blotted onto PVDF membranes as described in Materials and Methods. The membranes were incubated with antibodies to actin (B, top) or AGR2 (B, bottom) and visualized by chemiluminescence. nAGR2, natural AGR2; rAGR2, recombinant AGR2. The recombinant AGR2 is larger than the natural AGR2 due to the presence of a histidine tag and a protease factor X cleavage site.

Close modal

AGR2 mRNA was present at a 7.3 ± 0.2-fold (mean ± SD of three independent experiments) higher level in MCF-7 cells grown in the presence of estrogen than in cells grown in estrogen-depleted conditions, whereas the level of a previously described, estrogen-responsive mRNA, that of pS2 (35), was increased only 3.4 ± 0.1-fold. In these quantitative results, mRNA levels were normalized with respect to 36B4 mRNA, a mRNA that is not dependent on the presence of estrogen and its receptor for its production (29).

Identification of anterior gradient 2 mRNA and protein in human breast tumor specimens. The occurrence of AGR2 mRNA in human benign breast lesions and malignant breast carcinomas was examined by reverse transcription-PCR (RT-PCR; Fig. 2). Using 25 cycles of PCR, only 3 of 9 (33%) normal and 13 of 25 (52%) benign samples were positive for AGR2 mRNA (positivity defined as a single PCR band of 354 bp), whereas 44 of 56 (79%) breast carcinoma samples were positive for AGR2 mRNA (Table 1). This proportion was significantly different from the normal and benign specimens (P = 0.0029, Fisher's exact test). Moreover, 31 of 34 (91%) ERα-positive carcinoma specimens yielded a strong PCR product, whereas only 13 of 22 (59%) ERα-negative carcinomas were positive for AGR2 mRNA, values that were also significantly different (P = 0.007, Fisher's exact test). These results show that AGR2 mRNA is dependent on the presence of ERα in the majority of breast carcinoma specimens as well as in the breast carcinoma cell lines.

Figure 2.

Identification of AGR2 mRNA from human breast tumor specimens using RT-PCR. RNA from human breast carcinoma specimens (lanes 1-28) was amplified by RT-PCR using primers specific for AGR2 cDNA (A) or human glyceraldehyde-3-phosphate dehydrogenase cDNA (B) to yield PCR products of 354 and 452 bp, respectively. The ERα status of specimens was recorded (+ or -) as described in Materials and Methods. The resulting RT-PCR products were subjected to agarose gel electrophoresis and stained with ethidium bromide as described in Materials and Methods. Lanes M, DNA molecular weight markers.

Figure 2.

Identification of AGR2 mRNA from human breast tumor specimens using RT-PCR. RNA from human breast carcinoma specimens (lanes 1-28) was amplified by RT-PCR using primers specific for AGR2 cDNA (A) or human glyceraldehyde-3-phosphate dehydrogenase cDNA (B) to yield PCR products of 354 and 452 bp, respectively. The ERα status of specimens was recorded (+ or -) as described in Materials and Methods. The resulting RT-PCR products were subjected to agarose gel electrophoresis and stained with ethidium bromide as described in Materials and Methods. Lanes M, DNA molecular weight markers.

Close modal
Table 1.

Identification of AGR2 mRNA and protein in breast specimens using RT-PCR and immunocytochemistry

Clinical specimensRT-PCR screening of specimens
Immunocytochemical screening of specimens
No. positive*No. negative*TotalNo. positiveNo. negativeTotal
Normal 
Benign 13 12 25 15 
Total 16 18 34 11 20 
ERα-positive carcinomas 31 34 26 29 
ERα-negative carcinomas 13 22 7 15 
Total 44§ 12 56 33 11 44 
Clinical specimensRT-PCR screening of specimens
Immunocytochemical screening of specimens
No. positive*No. negative*TotalNo. positiveNo. negativeTotal
Normal 
Benign 13 12 25 15 
Total 16 18 34 11 20 
ERα-positive carcinomas 31 34 26 29 
ERα-negative carcinomas 13 22 7 15 
Total 44§ 12 56 33 11 44 
*

Positive RT-PCR is defined as a single band of molecular weight 354 bp for AGR2 on the gel; negative RT-PCR is defined as no clear band on the gel, all have a band of 452 bp for control GPDH on the gel.

Positive immunocytochemistry is defined as >5% of epithelial cells staining; negative immunocytochemistry is defined as <5% of epithelial cells staining.

P = 0.007, statistically significantly different from ERα positive (Fisher's exact test).

§

P = 0.0029, statistically significantly different from normal and benign specimens (Fisher's exact test).

P = 0.0033, statistically significantly different from ERα positive (Fisher's exact test).

P = 0.0025, statistically significantly different from benign specimens (Fisher's exact test).

The affinity-purified AGR2 antibodies (Materials and Methods) were used to stain immunocytochemically histologic sections of human breast specimens. The epithelial cells of human normal breast tissue and benign lesions either were stained modestly for AGR2 protein or were unstained (Fig. 3A and B), but the epithelial cells of ERα-positive breast carcinomas were stained strongly for AGR2 protein (Fig. 3C). The positive staining for AGR2 protein was completely abolished by prior incubation of the antibodies with recombinant AGR2 protein (Fig. 3D). The immunocytochemical staining for AGR2 protein showed a granular appearance, reminiscent of secretory granules (Fig. 3E). There was little or no staining for AGR2 in >50% of ERα-negative breast carcinoma specimens (Fig. 3F). Overall, AGR2 protein immunocytochemical positivity (defined as >5% of epithelial cells staining) was found in only 2 of 5 (40%) normal specimens and 7 of 15 (47%) benign breast tumor specimens, but 33 of 44 (75%) breast carcinoma specimens were positive for AGR2 protein. This proportion was significantly different from normal and benign specimens (P = 0.025, Fisher's exact test). Twenty-six of 29 (90%) ERα-positive specimens were positively stained, whereas only 7 of 15 (47%) ERα-negative carcinomas were positive for AGR2 protein, significantly different from the ERα-positive carcinomas (P = 0.0033, Fisher's exact test). These experiments showed quantitative results for AGR2 protein that were similar to those obtained for mRNA by RT-PCR.

Figure 3.

Immunocytochemical staining of human breast specimens for AGR2. Histologic sections showing a normal human breast containing ducts and lobules (A), a benign fibroadenoma (B), an ERα-positive (C-E), and an ERα-negative (F) invasive ductal carcinomas stained immunocytochemically by affinity-purified antibodies to recombinant AGR2 protein (A-C, E, and F) or by antibodies preincubated with 0.1 mg/mL recombinant AGR2 protein (D) as described in Materials and Methods. For A-D and F, bar (shown on A), 50 μm; for E, bar, 20 μm.

Figure 3.

Immunocytochemical staining of human breast specimens for AGR2. Histologic sections showing a normal human breast containing ducts and lobules (A), a benign fibroadenoma (B), an ERα-positive (C-E), and an ERα-negative (F) invasive ductal carcinomas stained immunocytochemically by affinity-purified antibodies to recombinant AGR2 protein (A-C, E, and F) or by antibodies preincubated with 0.1 mg/mL recombinant AGR2 protein (D) as described in Materials and Methods. For A-D and F, bar (shown on A), 50 μm; for E, bar, 20 μm.

Close modal

Identification of anterior gradient 2 as a novel metastasis inducer. The full-length M36 (AGR2) cDNA was inserted into the multiple cloning site of the mammalian expression vector pcDNA3, downstream of the cytomegalovirus promoter, to yield a recombinant cDNA construct designated pcDNA-M36. The pcDNA-M36 expression construct, or the same amount of empty pcDNA vector as a negative control (pcDNA), was transfected into the benign rat mammary epithelial cell line, Rama 37 (Materials and Methods). Following selection in geneticin, single colonies from the Rama 37 pcDNA-M36 transfectants were picked and expanded along with two independent pools from the same transfectants. Using PCR or Southern hybridization, RT-PCR, or Northern hybridization, AGR2 sequences were shown to be present in the DNA and RNA from the transfectant pools and clones but absent from the Rama 37 cells and the Rama 37 cells transfected with empty vector summarized in Table 2.

Table 2.

Incidence of tumors and metastases produced by M36 cDNA transfected cells

Cell linesPresence or absence of AGR2 mRNAPrimary tumors
Metastases
Mean ± SD appearance (days after injection of cells)*P (Mann-Whitney U-test)*Mean ± SD growth period (d)P (Mann-Whitney U-test)Mean ± SD weight (g)P (Mann-Whitney U-test)Incidence of primary tumors (%)§P (Fisher's exact test)§Incidence of metastases (%)P (Fisher's exact test)
Rama 37 1st test Negative N/A — N/A — N/A — 22/46 (48) — 0/22 (0) — 
Rama 37 2nd test Negative 31 ± 7 — 15 ± 5 — 5.5 ± 2.1 — 12/18 (67) — 0/12 (0) — 
pcDNA vector Negative 49 ± 11 0.0002 15 ± 2 0.07 N/A — 10/20 (50) 1, 0.5 0/10 (0) — 
Pool 1 Positive 38 ± 18 0.46 37 ± 14 <0.0001 5.4 ± 2.6 0.99 13/23 (57) 0.6, 0.5 12/13 (92) <0.0001 
Pool 2 Positive 54 ± 15 <0.0001 30 ± 18 0.009 4.8 ± 1.7 0.32 14/26 (54) 0.8, 0.2 12/14 (86) <0.0001 
Clone 1 Positive 81 ± 6 <0.0001 18 ± 5 0.49 N/A — 30/60 (50) 0.9, 0.3 23/30 (77) <0.0001 
Clone 2 Positive 37 ± 14 0.39 16 ± 5 0.94 6.8 ± 2.4 0.25 9/14 (64) 0.4, 1 7/9 (78) 0.0003 
Clone 3 Positive 46 ± 22 0.014 42 ± 20 0.0003 5.9 ± 2.7 0.88 11/37 (30) 0.1, 0.02 9/11 (82) <0.0001 
Cell linesPresence or absence of AGR2 mRNAPrimary tumors
Metastases
Mean ± SD appearance (days after injection of cells)*P (Mann-Whitney U-test)*Mean ± SD growth period (d)P (Mann-Whitney U-test)Mean ± SD weight (g)P (Mann-Whitney U-test)Incidence of primary tumors (%)§P (Fisher's exact test)§Incidence of metastases (%)P (Fisher's exact test)
Rama 37 1st test Negative N/A — N/A — N/A — 22/46 (48) — 0/22 (0) — 
Rama 37 2nd test Negative 31 ± 7 — 15 ± 5 — 5.5 ± 2.1 — 12/18 (67) — 0/12 (0) — 
pcDNA vector Negative 49 ± 11 0.0002 15 ± 2 0.07 N/A — 10/20 (50) 1, 0.5 0/10 (0) — 
Pool 1 Positive 38 ± 18 0.46 37 ± 14 <0.0001 5.4 ± 2.6 0.99 13/23 (57) 0.6, 0.5 12/13 (92) <0.0001 
Pool 2 Positive 54 ± 15 <0.0001 30 ± 18 0.009 4.8 ± 1.7 0.32 14/26 (54) 0.8, 0.2 12/14 (86) <0.0001 
Clone 1 Positive 81 ± 6 <0.0001 18 ± 5 0.49 N/A — 30/60 (50) 0.9, 0.3 23/30 (77) <0.0001 
Clone 2 Positive 37 ± 14 0.39 16 ± 5 0.94 6.8 ± 2.4 0.25 9/14 (64) 0.4, 1 7/9 (78) 0.0003 
Clone 3 Positive 46 ± 22 0.014 42 ± 20 0.0003 5.9 ± 2.7 0.88 11/37 (30) 0.1, 0.02 9/11 (82) <0.0001 

NOTE: Fifteen percent of injected rats were omitted from the analysis due to the presence of ascites or ulceration. N/A, not applicable.

*

Mean time of appearance of primary tumors (P values < 0.05, significantly different from Rama 37, second test).

Mean growth period from primary tumor appearance to culling (P values <0.05, significantly different from Rama 37, second test).

Mean weight of primary tumor (all AGR2 transfectants not significantly different from Rama 37, second test).

§

Incidence of primary tumors: % (no. rats with tumors/no. rats injected), all transfectants not significantly different from Rama 37, first test, and all but one transfectant not significantly different from Rama 37, second test.

Incidence of metastases: % (no. rats with metastases/no. rats with tumors), all AGR2 transfectants significantly different from Rama 37, first test, Rama 37, second test, and pcDNA vector transfectant.

The incidences of primary tumors and metastases produced by AGR2 transfection are shown in Table 2 along with statistical analyses. A single s.c. injection of cells transfected with the empty vector yielded primary tumors in the mammary glands with a mean latent period of 49 ± 11 days and with a tumor incidence of 50% (Table 2). This incidence was not significantly different from that obtained with Rama 37 cells (Table 2). None of the tumor-bearing rats exhibited lung metastases, and this was confirmed by subsequent histologic examination of selected tissues, including lung and lymph nodes. However, the mean latent period for palpable primary tumors of the pcDNA transfectants was statistically different from Rama 37 (Table 2). Rats injected with the two independent pools or three independent clones of pcDNA-M36-transfected, AGR2-expressing cells yielded tumors with a range of mean latent periods to palpability (Table 2). All of these were longer, some significantly longer than Rama 37 cells (Table 2), as were the growth periods between tumor appearance and culling (Table 2). There was no relationship between latent period and growth period for the individual clones and pools. The weights of tumors at culling for any of the pools and clones tested were not significantly different from Rama 37 cells (Table 2).

The incidences of primary tumors for the AGR2-transfected clones and pools in the mammary glands ranged from 30% to 64% of injected rats (Table 2), none were statistically different from pcDNA vector-transfected cells (Table 2), Rama 37 cells first test (Table 2), and all but one not significantly different from Rama 37 second test (Table 2). However, 92% and 86% of rats injected with pool 1 and pool 2 cells, respectively, and 77% to 82% of rats injected with the three clones of AGR2-transfected cells developed either gross metastases in the lungs, which were visible at necropsy, or micrometastases evident on subsequent histologic examination (Table 2). These values for the incidences of metastases were significantly different from the control group of pcDNA empty vector-transfected Rama 37 cells and two separate groups of animals injected with Rama 37 cells (all P ≤ 0.0003, Fisher's exact test), in which no lung macrometastasis or micrometastasis was found. There was no significant correlation between metastatic potential and the length of time the tumors took to grow following detection, nor any significant relationship when latent period plus tumor growth period was plotted against metastatic potential (least squares regression analysis of a fit of the points to a straight line yielded probabilities in the range 0.12-0.55), ruling out the possibility that metastasis arose due to a longer period of tumor growth. The results show that transfection of the nonmetastatic cells with an expression vector containing AGR2 induces metastasis in two pools and three separate clones of cells.

Histology and immunocytochemistry of tumors produced by transfected cells. Some of the primary tumors from rats injected with the AGR2-expressing Rama 37 cells transfected with pcDNA-M36 were composed of cuboidal cells, many forming cords that were surrounded by neoplastic spindle cells, whereas others consisted predominantly of neoplastic spindle cells. Many tumors showed central necrotic cores. In many primary tumors arising from cells transfected with the pcDNA-M36 construct, extensive numbers of blood vessels were seen. Some tumor cells had breached the surrounding connective tissue capsules and had invaded the adjacent host skeletal muscle (Fig. 4A). In general, the histology of the metastases was the same as that of the primary tumor. Both cannonball metastases and tumor cells penetrating the surrounding lung tissue were evident (Fig. 4B). The primary tumors and metastases were also extensively stained by antibodies to vimentin (Fig. 4B), and in that case, tumor cells in endothelial cell–lined spaces, possibly lymphatics (Fig. 4C) and in blood vessels (data not shown), were also observed. The primary tumor cells and lung metastases also exhibited staining for milk fat globule membrane antigen (data not shown) and by pan-keratin antibodies (Fig. 4D) and by peanut lectin (data not shown). Differentiation of tumor cells to skeletal muscle–like elements was common in both the primary tumor and its metastases, sometimes forming large multinucleate cells. These skeletal muscle–like elements were immunocytochemically stained by antisera to skeletal muscle actin (data not shown) and to myoglobin (Fig. 4E). Skeletal muscle elements were not found in any of the primary tumors arising from cells transfected with the control pcDNA vector.

Figure 4.

Histology and immunocytochemical staining of primary rat tumors and metastases produced by M36 cDNA transfectants. Histologic sections of tissues from animals injected with Rama 37 cells transfected with the pcDNA3-M36 construct were stained with H&E or immunocytochemically stained. Carcinoma cells (T) of a primary tumor in the mammary gland locally invading muscle tissue (M), visualized by H&E staining (A). A cannonball metastasis (M) in the lung stained by antibodies to vimentin (L; B). Metastatic tumor cells in the lungs (large arrow) possibly in a lymphatic space adjacent to a blood vessel (small arrow) stained by antibodies to vimentin (C). Tumor cells in a lung metastasis stained by antibodies to pan-keratin (D). Skeletal muscle elements (arrow) within a lung metastasis stained by antibodies to myoglobin (E). Histologic sections of primary tumors from animals injected with Rama 37 cells transfected with pcDNA3 empty vector (F) or primary tumors (G and I) and metastases in the lungs (H) produced from animals injected with Rama 37 cells transfected with vector containing M36 cDNA (G-I) were stained with affinity-purified antibodies to recombinant AGR2 protein (F-H) or with antibodies preincubated with 0.1 mg/mL recombinant AGR2 protein (I) as described in Materials and Methods. Bar, 50 μm (A), 200 μm (B), and 20 μm (C); (D-I) as (A).

Figure 4.

Histology and immunocytochemical staining of primary rat tumors and metastases produced by M36 cDNA transfectants. Histologic sections of tissues from animals injected with Rama 37 cells transfected with the pcDNA3-M36 construct were stained with H&E or immunocytochemically stained. Carcinoma cells (T) of a primary tumor in the mammary gland locally invading muscle tissue (M), visualized by H&E staining (A). A cannonball metastasis (M) in the lung stained by antibodies to vimentin (L; B). Metastatic tumor cells in the lungs (large arrow) possibly in a lymphatic space adjacent to a blood vessel (small arrow) stained by antibodies to vimentin (C). Tumor cells in a lung metastasis stained by antibodies to pan-keratin (D). Skeletal muscle elements (arrow) within a lung metastasis stained by antibodies to myoglobin (E). Histologic sections of primary tumors from animals injected with Rama 37 cells transfected with pcDNA3 empty vector (F) or primary tumors (G and I) and metastases in the lungs (H) produced from animals injected with Rama 37 cells transfected with vector containing M36 cDNA (G-I) were stained with affinity-purified antibodies to recombinant AGR2 protein (F-H) or with antibodies preincubated with 0.1 mg/mL recombinant AGR2 protein (I) as described in Materials and Methods. Bar, 50 μm (A), 200 μm (B), and 20 μm (C); (D-I) as (A).

Close modal

Although there was no immunocytochemical staining for AGR2 protein in the histologic sections of primary tumors arising from animals injected with Rama 37 cells transfected with empty pcDNA plasmid vector (Fig. 4F), both carcinoma cells of primary tumors (Fig. 4G) and metastases in the lungs (Fig. 4H) produced from animals injected with the AGR2-expressing Rama 37 cells transfected with pcDNA-M36 were stained positively by the affinity-purified antibodies directed against recombinant AGR2 protein. The immunocytochemical staining by the antibodies to AGR2 was blocked completely by their prior incubation with recombinant AGR2 protein (Fig. 4I).

Enhanced adhesion is associated with the anterior gradient 2–transfected cells. The AGR2 transfectants did not exhibit any significantly different growth rates in vitro compared with vector-transfected or parental Rama 37 cells (P = 1.0, Student's t test), nor any altered invasive potential as measured by their behavior in an invasion assay using Matrigel-coated Transwell chambers (P = 1.0, Student's t test). In contrast, a statistically significant increase in the rate of cellular adhesion to a plastic substratum (Fig. 5) was shown by all the AGR2 cDNA-transfected pools and transfected clones of cells examined relative to parental Rama 37 cells (P range, 0.009 to <0.0001, Student's t test) and by all pools and two of three clones relative to Rama 37 cells transfected with empty vector (P range, 0.0074-0.001, Student's t test). Although there was a broad correlation between the metastatic and the adhesive potential of the different transfectants, the similar levels of AGR2 in the pcDNA-M36-transfected cells precluded determining any significant relationship between AGR2 expression and adhesive/metastatic potential of the individual pools/clones of cells. To further identify the mechanism by which AGR2 affects adhesion, coating of the tissue culture dishes with either 2 or 20 μg of recombinant AGR2 protein lacking a signal peptide increased the rate of attachment of the AGR2-negative Rama 37 cells or Rama 37 cells transfected with empty vector but had little effect on the rate of attachment of the AGR2-transfected cell lines or pools (Fig. 5).

Figure 5.

Adhesive and metastatic properties of AGR2-transfected and untransfected rat mammary cell lines. A, Rama 37 or Rama 37 cells transfected with empty pcDNA3 vector (Vector control) or pcDNA3-M36 (Pool 1, Pool 2, Clone 1, Clone 2, and Clone 3) were tested in vitro for their rate of adhesion to a plastic substratum (white columns) and for their metastatic ability (black columns) as described in Materials and Methods. White columns, mean of three separate experiments; bars, SD. No black columns are shown for Rama 37 cells and Rama 37 transfected with vector alone because they did not produce metastases. B, Rama 37, Rama 37 cells transfected with empty vector, or Rama 37 cells transfected with expression vector containing AGR2 cDNA (pools and clones)were allowed to attach to the substratum in 24-well culture dishes and the percentage of cells adhering in 30 minutes was recorded. The experiments were carried out in untreated tissue culture wells (white columns) and with wells coated with 2 μg (gray columns) or 20 μg (black columns) recombinant AGR.

Figure 5.

Adhesive and metastatic properties of AGR2-transfected and untransfected rat mammary cell lines. A, Rama 37 or Rama 37 cells transfected with empty pcDNA3 vector (Vector control) or pcDNA3-M36 (Pool 1, Pool 2, Clone 1, Clone 2, and Clone 3) were tested in vitro for their rate of adhesion to a plastic substratum (white columns) and for their metastatic ability (black columns) as described in Materials and Methods. White columns, mean of three separate experiments; bars, SD. No black columns are shown for Rama 37 cells and Rama 37 transfected with vector alone because they did not produce metastases. B, Rama 37, Rama 37 cells transfected with empty vector, or Rama 37 cells transfected with expression vector containing AGR2 cDNA (pools and clones)were allowed to attach to the substratum in 24-well culture dishes and the percentage of cells adhering in 30 minutes was recorded. The experiments were carried out in untreated tissue culture wells (white columns) and with wells coated with 2 μg (gray columns) or 20 μg (black columns) recombinant AGR.

Close modal

A subtracted library of cloned cDNAs expressed at a higher level in the malignant human mammary cell line, MCF-7, relative to a benign cell line, Huma 123, derived from benign breast disease, yielded among others, a cDNA, M36, identical to the coding sequence of human AGR2, the anterior gradient 2 homologue (36), produced by the cement gland of the developing X. laevis embryo (18). Furthermore, the size of the primary translation products of M36 and the AGR2 protein were identical and the proposed product additionally exhibited 91% identity to the first 175 amino acids of the mouse gob-4 gene, which is expressed in intestinal mucin-secreting goblet cells (37). Transfection of M36 cDNA in an expression vector into a benign rat mammary cell line, Rama 37, induced a metastatic phenotype in vivo when the cells were injected into the mammary fat pads of syngeneic rats, whereas similar transfection of empty vector failed to induce a metastatic phenotype. The nonmetastatic Rama 37 cells have been shown previously to be converted to a metastatic phenotype by genes encoding proteins S100A4 (7), osteopontin (8), and cutaneous fatty acid-binding protein (38) but not by oncogenes, such as Ha-ras or DNA virus-transforming genes (39). Elevated levels of immunoreactive S100A4 (9) or osteopontin (10) in the breast cancers of patients have been shown to correlate with markedly reduced patient survival, whereas cutaneous fatty acid-binding protein occurs preferentially in malignant as opposed to benign prostatic lesions (38). Use of the Rama 37 cell line has thus identified previously cloned genes whose products are biologically relevant to the metastatic process in human cancer, and our present results suggest strongly that AGR2 may also be involved in metastasis. Human AGR2 was reported previously to be expressed in the ERα-positive breast cancer cell line, MCF-7, and not in the ERα-negative cell line, MDA-MB-231 (17), but this is the first report of its direct involvement with metastasis and/or malignancy. The apparent inconsistency of the involvement of a strongly ER-dependent gene/gene product being associated with the process of metastasis of breast tumor cells is supported by the observation that in a group of 225 tamoxifen-treated patients with ER-positive breast cancers, those with AGR2 in their breast cancer cells exhibited a statistically significantly poorer survival than those without AGR2 in their cancer cells. In contrast, the similarly treated 126 patients with ER-negative breast cancers showed no such relationship.4

4

Innes et al., in preparation.

The AGR2-induced tumors in the experimental rats differed somewhat from those induced by S100A4. Whereas S100A4 produced primarily cannon ball metastases in the lungs/lymph nodes (7), AGR2 induced both cannon ball metastases and micrometastases in the lungs, similar to those observed previously for osteopontin (8). The presence of AGR2-induced micrometastases in blood, and possibly in lymphatic vessels, suggests that AGR2 may induce metastasis to the lungs by both blood-borne and lymphatic routes.

The Xenopus AGR2 protein is a product of the mucin-producing cement gland, which is the first ectodermal organ to appear in the developing Xenopus embryo (18). AGR2 is up-regulated in experimentally dorsalized embryos and down-regulated in experimentally ventralized embryos (18). Although the precise role of AGR2 is presently unknown, its injection into early cleavage stage embryos results in enhanced cement gland development (18), and ectopically produced AGR2 not only can signal dorsoanterior ectodermal fate but also can induce neural markers in embryo cells, suggesting that extracellular AGR2 is able to alter the differentiation potential of its target cells (18). It is not yet clear whether AGR2 has similar activities in human development or whether such activities are related to its metastasis-inducing properties. However, in the transfection experiments, some of the cells containing the AGR2 (M36) construct exhibited muscle, and not neuronal, patterns of differentiation. The reason for this pattern of differentiation being evident in the Rama 37 cells transfected with AGR2 is not known; however, this pattern of differentiation has been described before in derivative cells of distinct morphologic appearance, representing possible pluripotent intermediates in differentiation of the Rama 37 epithelial cells to a myoepithelial-like phenotype (40). Thus, the capability to form muscle elements might be an intrinsic property of the recipient Rama 37 cell derivatives that is enhanced by the AGR2 (M36) gene/gene product. The mechanism of such enhancement is not known.

Previously, the XAG-2 (AGR2) protein has been shown to be secreted when expressed in Xenopus oocytes (18). In the present experiments, a granular cytoplasmic appearance of immunocytochemical staining for AGR2 observed in some human carcinoma specimens suggests that AGR2 might be secreted by some carcinoma cells. Pilot experiments using an AGR2 COOH-terminal GFP fusion cDNA5

5

D. Liu, P.S. Rudland, R. Barraclough, unpublished data.

have shown that the fusion protein is secreted at least in HeLa cells, strongly suggesting that the AGR2 secretory signal is active.

To identify a possible mechanism for AGR2-induced metastasis, the effect of added recombinant AGR2 protein on Rama 37 cells was tested. AGR2 protein lacking the signal sequence up to a concentration of at least 24 μmol/L6

6

D. Liu, P.S. Rudland, R. Barraclough, unpublished results.

and pcDNA-M36 transfection failed to affect the growth rate of Rama 37 cells in culture. These results support the lack of a consistent effect of AGR2 on the latent period of tumor formation in vivo (Table 2). Furthermore, pcDNA-M36-transfected cell pools and clones did not show any enhanced invasive ability through Matrigel compared with parental and empty vector-transfected Rama 37 cells. However, the pcDNA-M36 transfectants exhibited an increased adhesive potential to a plastic substratum. The fact that AGR2 contains an active secretory signal is consistent with the punctate staining pattern for AGR2 observed in human breast cancers at higher magnification (Fig. 3E) and suggests that AGR2 functions extracellularly. An extracellular mechanism of AGR2 is shown by the observation that extracellularly added AGR2 enhances the rate of attachment of two AGR2-negative cell lines to that observed with five independent AGR2-producing cell clones and pools but had no effect on the AGR2-producing cell clones and pools. Although the mechanistic link between AGR2-induced increased adhesion to plastic and metastasis is not yet known, another well-characterized metastasis-inducing (8, 10) secreted protein, osteopontin (41), has also been shown to increase the adhesion of cells (42), including Rama 37 cells, to plastic substrata (43). Taken together, these results strongly suggest that AGR2 might also cause metastasis by enhancing this adhesive property of the Rama 37 cells.

In summary, this article describes the first demonstration of metastasis-inducing properties of the developmentally important protein, AGR2. The presence of detectable AGR2 mRNA and protein above a threshold in breast carcinoma cells significantly correlates with carcinoma in preference to benign/normal tissue, and ERα-positive in preference to ERα-negative carcinomas, suggesting that the metastasis-inducing properties of AGR2 may contribute, in some way, toward the malignant progression of some ERα-positive breast cancers. Identification of the receptor for AGR2 will provide the means to identify the signaling pathways that link the enhancement of cell attachment to the process of metastasis.

Grant support: Clatterbridge Cancer Research Trust studentship and UK Committee of Vice Chancellors and Principals ORS scholarship award (D. Liu), Cancer and Polio Research Fund, and North West Cancer Research Fund grant CR532.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Joe Carroll, Barry Cotterill and Karen Collard for excellent technical assistance.

1
Martin K, Kritzman DM, Price LM, et al. Linking gene expression patterns to therapeutic groups in breast cancer.
Cancer Res
2000
;
60
:
2232
–8.
2
Perou CM, Sorlie T, Eisen MB, et al. Molecular portraits of human breast tumours.
Nature
2000
;
406
:
747
–52.
3
Smid-Koopman E, Blok LJ, Chadha-Ajwani S, Helmerhorst TJ, Brinkmann AO, Huikeshoven FJ. Gene expression profiles of human endometrial cancer samples using a cDNA-expression array technique: assessment of an analysis method.
Br J Cancer
2000
;
83
:
246
–51.
4
Wang T, Hopkins D, Schmidt C, et al. Identification of genes differentially over-expressed in lung squamous cell carcinoma using combination of cDNA subtraction and microarray analysis.
Oncogene
2000
;
19
:
1519
–28.
5
Xu JC, Stolk JA, Zhang XQ, et al. Identification of differentially expressed genes in human prostate cancer using subtraction and microarray.
Cancer Res
2000
;
60
:
1677
–82.
6
Barraclough R, Savin J, Dube S, Rudland P. Molecular cloning and sequence of the gene for p9Ka, a cultured myoepithelial cell protein with strong homology to S-100, a calcium-binding protein.
J Mol Biol
1987
;
198
:
13
–20.
7
Davies BR, Davies MPA, Gibbs FEM, Barraclough R, Rudland PS. Induction of the metastatic phenotype by transfection of a benign rat mammary epithelial cell line with the gene for p9Ka, a rat calcium-binding protein but not with the oncogene EJ ras-1.
Oncogene
1993
;
8
:
999
–1008.
8
Oates AJ, Barraclough R, Rudland PS. The identification of osteopontin as a metastasis-related gene product in a rodent mammary tumour model.
Oncogene
1996
;
13
:
97
–104.
9
Rudland PS, Platt-Higgins A, Renshaw C, et al. Prognostic significance of the metastasis-inducing protein S100A4 (p9Ka) in human breast cancer.
Cancer Res
2000
;
60
:
1595
–603.
10
Rudland PS, Platt-Higgins A, El-Tanani M, et al. Prognostic significance of the metastasis-associated protein osteopontin in human breast cancer.
Cancer Res
2002
;
62
:
3417
–27.
11
Diatchenko L, Lau YFC, Campbell A, et al. Suppression subtractive hybridisation: a method for generating differentially-regulated or tissue-specific cDNA probes and libraries.
Proc Natl Acad Sci U S A
1996
;
93
:
6025
–30.
12
Liu D, Rudland PS, Sibson DR, Platt-Higgins A, Barraclough R. Expression of calcium-binding protein S100A2 in breast lesions.
Br J Cancer
2000
;
83
:
1473
–9.
13
Liu D. The identification of genes differentially expressed in human breast lesions. Ph.D. thesis. Liverpool: University of Liverpool; 2001. p. 1–268.
14
Liu D, Rudland P, Sibson D, Barraclough R. Identification of mRNAs differentially-expressed between benign and malignant breast tumour cells.
Br J Cancer
2002
;
87
:
423
–31.
15
Ke Y, Fernig DG, Wilkinson MC, et al. The expression of basic fibroblast growth factor and its receptor in cell lines derived from normal human mammary gland and a benign mammary lesion.
J Cell Sci
1993
;
106
:
135
–43.
16
Soule HD, Vazquez A, Long A, Albert S, Brennan MA. Human cell line from a pleural effusion derived from a breast carcinoma.
J Natl Cancer Inst
1973
;
51
:
1409
–13.
17
Thompson DA, Weigel RJ. hAG-2, the human homologue of the Xenopus laevis cement gland gene XAG-2, is coexpressed with estrogen receptor in breast cancer cell lines.
Biochem Biophys Res Commun
1998
;
251
:
111
–6.
18
Aberger F, Weidinger G, Grunz H, Richter K. Anterior specification of embryonic ectoderm: the role of the Xenopus cement gland-specific gene XAG2.
Mech Dev
1998
;
72
:
115
–30.
19
Rudland PS, Ollerhead G, Barraclough R. Isolation of simian virus 40 transformed human mammary epithelial stem cell line that can differentiate to myoepithelial-like cells in culture and in vivo.
Dev Biol
1989
;
136
:
167
–80.
20
Briand P, Petersen OW, Van Deurs B. A new diploid non-tumorigenic human breast epithelial cell line isolated and propagated in chemically-defined medium.
In Vitro
1987
;
23
:
181
–8.
21
Engel LW, Young NA. Human breast carcinoma cells in continuous culture: a review.
Cancer Res
1978
;
38
:
4327
–39.
22
Dunnington DJ, Monaghan P, Hughes CM, Rudland PS. Phenotypic instability of rat mammary tumor epithelial cells.
J Natl Cancer Inst
1983
;
71
:
1227
–40.
23
El-Tanani M, Green C. Estrogen-induced genes, pLiv-1 and pS2, respond divergently to other steroid-hormones in MCF-7 cells.
Mol Cell Endocrinol
1995
;
111
:
75
–81.
24
Darbre P, Yates J, Curtis S, King R. Effect of estradiol on human breast cancer cells in culture.
Cancer Res
1983
;
43
:
349
–54.
25
Barraclough R, Kimbell R, Rudland PS. Differential control of mRNA levels for Thy-1 antigen and laminin in rat mammary epithelial and myoepithelial-like cells in culture.
J Cell Physiol
1987
;
131
:
393
–401.
26
Chirgwin JM, Przybyla AE, Macdonald RJ, Rutter WJ. Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease.
Biochemistry
1979
;
18
:
5294
–9.
27
Han JH, Stratowa C, Rutter WJ. Isolation of full-length putative rat lysophospholipase cDNA using improved methods for mRNA isolation and cDNA cloning.
Biochemistry
1987
;
26
:
1617
–25.
28
Feinberg AP, Vogelstein B. A technique for radiolabelling DNA restriction endonuclease fragments to high specific activity.
Anal Biochem
1984
;
137
:
266
–7.
29
Laborda J. 36B4 cDNA used as an estradiol-independent mRNA control is the cDNA for human acidic ribosomal phosphoprotein PO.
Nucleic Acids Res
1991
;
19
:
3998
.
30
Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4.
Nature
1970
;
227
:
680
–5.
31
Sambrook J, Fritsch EF, Maniatis T. Molecular cloning: a laboratory manual. 2nd ed. Cold Spring Harbor (NY): Cold Spring Harbor Laboratory Press; 1989.
32
Jamieson S, Barraclough R, Rudland PS. Generation of metastatic variants by transfection of a nonmetastatic rat mammary epithelial cell line with DNA from a metastatic rat mammary cell line.
Pathobiology
1990
;
58
:
329
–42.
33
Rudland PS, Dunnington DJ, Gusterson B, Monaghan P, Hughes CM. Production of skeletal muscle elements by cell lines derived from neoplastic rat mammary epithelial stem cells.
Cancer Res
1984
;
44
:
2089
–101.
34
Anandappa S, Sibson R, Platt-Higgins A, Winstanley J, Rudland P, Barraclough R. Variant estrogen receptor α mRNAs in human breast cancer specimens.
Int J Cancer
2000
;
88
:
209
–16.
35
Masiakowski P, Breathnach R, Bloch J, Gannon F, Krust A, Chambon P. Cloning of cDNA sequences of hormone-regulated genes from the MCF-7 human breast cancer cell line.
Nucleic Acids Res
1982
;
10
:
7895
–903.
36
Kuang WW, Thompson DA, Hoch RV, Weigel RJ. Differential screening and suppression subtractive hybridization identified genes differentially expressed in an estrogen receptor-positive breast carcinoma cell line.
Nucleic Acids Res
1998
;
26
:
1116
–23.
37
Komiya T, Tanigawa Y, Hirohashi S. Cloning of the gene gob-4, which is expressed in intestinal goblet cells in mice.
Biochim Biophys Acta
1999
;
1444
:
434
–8.
38
Jing C, Beesley C, Foster CS, et al. Identification of the messenger RNA for human cutaneous fatty acid-binding protein as a metastasis inducer.
Cancer Res
2000
;
60
:
2390
–8.
39
Jamieson S, Barraclough R, Rudland PS.Transfection of a non-metastatic diploid rat mammary epithelial cell line with the oncogenes for EJ-RAS-1 and polyoma large T antigen.
Int J Cancer
1990
;
46
:
1071
–80.
40
Rudland P, Paterson F, Monaghan P, Twiston-Davies A, Warburton M. Isolation and properties of rat cell lines morphologically intermediate between cultured mammary epithelial and myoepithelial-like cells.
Dev Biol
1986
;
113
:
388
–405.
41
Denhardt D, Guo X. Osteopontin: a protein with diverse function.
FASEB J
1993
;
7
:
1475
–82.
42
Liaw L, Skinner M, Raines E, et al. Adhesive and migratory effects of osteopontin are mediated via distinct cell-surface integrins—role of α(v)β(3) in smooth-muscle cell-migration to osteopontin in-vitro.
J Clin Invest
1995
;
95
:
713
–24.
43
Moye VE, Barraclough R, West C, Rudland PS. Osteopontin expression correlates with adhesive and metastatic potential in metastasis-inducing DNA-transfected rat mammary cell lines.
Br J Cancer
2004
;
90
:
1796
–802.