Gene Expression Profiling to Identify Oncogenic Determinants of Autocrine Human Growth Hormone in Human Mammary Carcinoma*

We have exploited a discrepancy in the oncogenic potential of autocrine and exogenous human growth hormone (hGH) in an attempt to identify molecules that could potentially be involved in oncogenic transformation of the human mammary epithelial cell. Microarray analysis of 19,000 human genes identified a subset of 305 genes in a human mammary carcinoma cell line that were remarkably different in their response to autocrine and exogenous hGH. Autocrine and exogenous hGH also regulated 167 common genes. Semiquantitative reverse transcription-PCR confirmed differential regulation of genes by either autocrine or exogenous hGH. Functional analysis of one of the identified autocrine hGH-regulated genes, TFF3, determined that its expression is sufficient to support anchorage-independent growth of human mammary carcinoma cells. Small interfering RNA-mediated knockdown of TFF3 concordantly abrogated anchorage-independent growth of mammary carcinoma cells and abrogated the ability of autocrine hGH to stimulate oncogenic transformation of immortalized human mammary epithelial cells. Further functional characterization of the identified subset of specifically autocrine hGH regulated genes will delineate additional novel oncogenes for the human mammary epithelial cell.

We have exploited a discrepancy in the oncogenic potential of autocrine and exogenous human growth hormone (hGH) in an attempt to identify molecules that could potentially be involved in oncogenic transformation of the human mammary epithelial cell. Microarray analysis of 19,000 human genes identified a subset of 305 genes in a human mammary carcinoma cell line that were remarkably different in their response to autocrine and exogenous hGH. Autocrine and exogenous hGH also regulated 167 common genes. Semiquantitative reverse transcription-PCR confirmed differential regulation of genes by either autocrine or exogenous hGH. Functional analysis of one of the identified autocrine hGH-regulated genes, TFF3, determined that its expression is sufficient to support anchorage-independent growth of human mammary carcinoma cells. Small interfering RNA-mediated knockdown of TFF3 concordantly abrogated anchorage-independent growth of mammary carcinoma cells and abrogated the ability of autocrine hGH to stimulate oncogenic transformation of immortalized human mammary epithelial cells. Further functional characterization of the identified subset of specifically autocrine hGH regulated genes will delineate additional novel oncogenes for the human mammary epithelial cell.
The mammary gland is one of very few organs that are subjected to substantial postnatal development; cycles of growth, differentiation, apoptosis, regression, and remodeling are maintained almost constantly during the lifetime of the organism (1). Growth hormone (GH) 1 is obligatory for normal pubertal mammary gland development by acting on both the mammary stromal and epithelial components to result in ductal elongation and the differentiation of ductal epithelia into terminal end buds (2,3). Expression of the human GH (hGH) transgene in mice results in precocious development of the mammary gland (4,5) and the development of neoplasia (5). Conversely, spontaneous or experimentally engineered functional deficiency of GH (6 -9) results in severely impaired mammary gland development and virtual resistance to the spontaneous development of hyperplastic alveolar nodules (6) and chemically induced mammary carcinogenesis (7,9). Similarly, in a primate model, hGH administration results in marked hyperplasia of the mammary gland with an increased epithelial proliferation index (10).
The hGH gene is also expressed in epithelial cells of the normal human mammary gland (11). Increased epithelial expression of the hGH gene is associated with the acquisition of pathological proliferation, and the highest level of hGH gene expression is observed in metastatic mammary carcinoma cells (11). hGH receptor gene expression per mammary epithelial cell remains constant throughout the process of neoplastic progression (12), and therefore, alterations in the local concentration of ligand are likely to be pivotal in determining the effects of hGH on mammary epithelial cell behavior. We have recently generated a model system to study the role of autocrine-produced hGH in mammary carcinoma by stable transfection of either the hGH gene or a translation-deficient hGH gene into mammary carcinoma cells (13)(14)(15)(16)(17). We have further demonstrated that autocrine production of hGH in immortalized human mammary epithelial cells concomitantly enhances proliferation and offers protection from apoptosis, forming the basis for abnormal mammary acinar morphogenesis, oncogenic transformation, and tumor formation in vivo. Thus, simple forced expression of the hGH gene is sufficient for oncogenic transformation of the immortalized human mammary epithelial cell (18). Furthermore, autocrine production of hGH, in mammary carcinoma cells with epithelial morphology, promotes mesenchymal cellular morphology, increased cell migration, and increased metalloprotease activity with subsequent acquisition of invasive behavior both in vitro and in vivo (19). In stark contrast to the oncogenic and metastatic potential of autocrine hGH, exogenous hGH supports neither tumor formation nor invasion by human mammary epithelial cells.
We have utilized this discrepancy in the oncogenicity of autocrine and exogenous hGH in an attempt to identify molecules that could potentially be involved in oncogenic transformation of the human mammary epithelial cell. We were able to extract a subset of 305 genes that were remarkably different in their response to autocrine and exogenous hGH. Functional analysis of one of the identified autocrine hGH regulated genes, TFF3 (trefoil factor-3), determined that its expression is sufficient to support anchorage-independent growth of human mammary carcinoma cells. Further functional characterization of the identified subset of specifically autocrine hGH-regulated genes will delineate additional novel oncogenes for the human mammary epithelial cell.

MATERIALS AND METHODS
Cell Lines-The MCF-7 and MCF-10A cell lines were obtained from the ATCC and stably transfected with an expression plasmid containing the wild-type hGH gene (pMT-hGH) (20) under the control of the metallothionein 1a promoter (designated MCF-hGH) (13). For control purposes, the ATG start site in pMT-hGH was disabled via a mutation to TTG generated by standard techniques (pMT-MUT) (14), and MCF-7 cells stably transfected with this plasmid were designated MCF-MUT (13). MCF-MUT cells therefore transcribe the hGH gene but do not translate the mRNA into protein. A detailed description of the characterization of these cell lines has been published previously (13,14). Neither MCF-7, MCF-10A, nor MCF-MUT cells produce detectable amounts of hGH protein under serum-free conditions, whereas MCF-hGH cells secrete ϳ100 pM hGH into 2 ml of medium over a 24-h period under the culture conditions described here. MCF-7 and MCF-MUT cells behave identically to each other in terms of proliferation, transcriptional activation, and cell spreading (13).
Preparation of Total RNA-Total RNA was isolated from MCF-7, MCF-MUT, and MCF-hGH cells using TRI-REAGENT according to the manufacturer's instructions and resuspended in diethyl pyrocarbonatetreated water. Quantification and purity of the RNA was assessed by A 260 /A 280 absorption, and RNA quality was assessed by agarose gel electrophoresis. Three independently derived total RNA samples from the respective cell lines were pooled before labeling for hybridization. RNA samples with ratios greater than 1.6 were stored at Ϫ80°C for further analysis.
Oligonucleotide Microarray Manufacture-The human 19-K oligonucleotide arrays utilized herein consist of 60-mer oligonucleotide probes (manufactured by Sigma-Genosys), representing 18,861 genes. The probes were resuspended in 3ϫ SSC at 20 M concentration and were spotted onto poly-L-lysine-coated glass slides using GeneMachines OmniGrid spotter at the Genome Institute of Singapore. The printed arrays were postprocessed essentially as described by Eisen et al. (21).
Microarray Analysis-For fluorescence labeling of cDNAs, 30 g of total RNA from serum-free untreated cells and from hGH-treated cells at each time point were reverse transcribed in the presence of Cy3-dUTP and Cy5-dUTP using the Superscript II reverse transcription kit (Invitrogen). Total RNA from MCF-MUT and MCF-hGH cells was also labeled identically. Labeled cDNAs were hybridized overnight (14 -16 h) to the arrays at 42°C in a water bath. Before scanning, slides were washed in 2ϫ SSC with 0.1% SDS for 1 min, and then 1ϫ SSC, 0.2ϫ SSC, and 0.05ϫ SSC, sequentially for 1 min, 30 s, and 10 s. The arrays were scanned on a GenePix 4000S scanner (Axon Instruments), and the images were analyzed by GenePix Pro 3.0 software to measure fluorescence signals and format data for data base deposition. All of the array data were deposited in the GIS microarray data base (available on the World Wide Web at www.gismadb.gis.a-star.edu.sg), where expression ratios (Cy5/Cy3; ratio of the means) were calculated and transformed to log2. Log2 expression ratios were normalized by subtracting the median of the ratios. Agglomerative hierarchical clustering by Pearson distance measurement was accomplished using the cluster and tree view program developed by Eisen and Brown (22).
Selection of hGH-regulated Genes-The time course expression data were initially distilled to the set of array features having intensity of Ն100 with signal over background in both channels for at least six of the seven time points between 1 and 24 h. Within this data set, hGHresponsive genes were identified as those with temporal expression profiles demonstrating the greatest magnitude and consistency of change. This set of outliers was composed of genes showing a Ն1.5-fold change at least two consecutive time points. To eliminate outliers potentially due to experimental technique, a control screen was performed in which two different reference RNA samples (i.e. total RNA from cells without hGH treatment) were used to generate labeled cDNA and hybridized against each other on three separate microarrays. Genes identified in this control screen with expression ratios of Ն2.0 in any one experiment or Ն1.4 in average of three experiments were excluded from the final set of hGH-responsive genes. The autocrine hGH data demonstrating a 1.5-fold or greater change at four replicas within five experiments were selected.
In order to correct for gene-specific dye bias, we also performed arrays with reciprocal dye labeling. The above selected genes that were not reproducible in the reversed hybridization were further eliminated.
The complete list of hGH-responsive genes, GenBank TM accession numbers, and corresponding expression ratios are available on the World Wide Web at www.gis.a-star.edu.sg/homepage/toolssup.jsp. Note that all gene names are Unigene cluster ID names from Unigene build 164.
RT-PCR and Primers-The genes HOXB9 (homeobox B9), PAX4 (paired box gene 4), CRX (cone rod homeobox), TFF3, SEPP1 (seleno-FIG. 1. A, expression patterns of exogenous hGH-responsive genes in MCF-7 cells. A, temporal expression profiles of 587 arrayed features exhibiting change over time in response to exogenous hGH were hierarchically clustered and visualized using Eisen software (see "Materials and Methods"). Microarray experiments for each time point are shown in columns; individual genes are in rows. Red, increasing mRNA levels; green, decreasing mRNA levels. The degree of color saturation reflects the magnitude of the ratio (see color key at the bottom). B, cluster profiles of exogenous hGH-regulated genes. Gene expression profiles were algorithmically subdivided into eight clusters (shown to the right of the clustergram by colored bars). The average expression profile of each gene cluster is shown in a line graph and is colored according to the matching bars. N, number of array features within each cluster.
protein P-1), SNK (Ste20-related serine/threonine kinase), and TC10 (Ras-like protein), identified to be regulated from the array, were analyzed using semiquantitative RT-PCR. RT-PCR was performed in a final volume of 50 l containing 0.2 g of mRNA template, 0.6 M primers, 2 ml of enzyme mix, 400 M each dNTP, 10 l reaction buffer, and 10 l Q-Solution by use of the Qiagen One-step RT-PCR kit. Briefly, RNA template was reverse-transcribed into cDNA for 30 min at 50°C; Hotstart TaqDNA polymerase was activated by heating for 15 min at 95°C; and the denatured cDNA templates were amplified by the following cycles: 94°C for 30 s, 55°C for 30 s, and 72°C for 60 s. A final extension was performed for 10 min at 72°C. In order to compare the PCR products semiquantitatively, 20 -35 cycles of PCR (annealing temperature 55°C) were performed to determine the linearity of the PCR amplification, and the amplified ␤-actin cDNA served as an internal control for cDNA quantity and quality. All RNA samples were treated with DNase I to avoid genomic DNA contaminations. Primers used were as follows: HOXB9 (sense), 5Ј-CGACTGCAAAGCCAGTGCTG-3Ј; Amplified PCR products were visualized on a 1% agarose gel. Am-plification yielded the predicted size of the respective amplified fragments.
Confocal Laser-scanning Microscopy-MCF7-hGH and MCF-MUT cells were grown in cavity slides, fixed in 4% paraformaldehyde/PBS, pH 7.4, washed, and blocked with BBX (PBS 0.1% Triton, 0.1% bovine serum albumin, 250 mM NaCl). The cells were incubated with rabbit TFF3 antiserum (kindly provided by Dr. Andy Giraud) (23) overnight at 4°C. They were again washed and blocked with BBX and incubated with the anti-rabbit-TRITC secondary antibody in BBX and mounted. Labeled cells were visualized with a Carl Zeiss (Jena, Germany) Axioplan microscope equipped with epifluorescence optics and a Bio-Rad MRC1024 confocal laser system. Images were converted to the tagged information file format and processed with the Adobe Photoshop program.
Luciferase Reporter Assay for TFF3 Promoter-MCF-hGH or MCF-MUT cells were grown in 6-well plates to 40 -60% confluence and were transiently transfected using Effectene with the TFF3 promoter-LUC construct (kindly provided by Prof. Nikolaus Blin) (24). Briefly, 0.2 g of the promoter construct was transfected per well in RPMI medium, and after 6 h, cells were transferred to serum-free medium. After the respective treatment (see below), the cells were finally washed in phosphate-buffered saline and lysed with 300 ml of 1ϫ lysis buffer (25 mM Tris phosphate, pH 7.8, 2 mM EDTA, 2 mM dithiothreitol, 10% glycerol, 1% Triton X-100) by a freeze-thaw cycle, and lysate was collected by centrifugation at 14,000 rpm for 15 min. The supernatant was used for the assay of luciferase and ␤-galactosidase activity. The luciferase activities were normalized on the basis of protein content as well as the ␤-galactosidase activity of the pCMVB vector. The ␤-galactosidase assay was performed with 20 ml of precleared cell lysate according to a   Cloning of Human TFF3-To clone human TFF3 cDNA, total RNA was extracted from MCF-7 cells using TRI-REAGENT, and the cDNA was amplified using the primers TFF3 cDNA Top (5Ј-CTC TGC ATG CTG GGG CTG GTC-3Ј) and Bot (5Ј-GGA GGT GCC TCA GAA GGT GCA TTC-3Ј) and cloned in to PCR-script TM AmpSK ϩ vector. TFF3 cDNA inserts were reamplified with primers TFF3 cDNA-myc Top (5Ј-GCG AAG CTT ATG CTG GGG CTG GTC) and TFF3 cDNA-myc Bot (5Ј-GGA GGT CCG CGG GAA GGT GCA TTC-3Ј) and cloned in frame using the enzyme sites HindIII and SacII inserted in the primers in pcDNA3.1/Myc-His-B vector, and the sequence was verified. Expression of TFF3 was verified by RT-PCR and Western blot analysis (data not shown).
Construction of Small Interfering RNA to Human TFF3-We generated a human TFF3 RNAi construct in pRNA-U6.1/Hygro vector from Genscript targeting the sequence ACTAGGAAGACAGAATGCA. 5 ϫ 10 4 MCF-7 cells were seeded into 6-well plates and were cultured as above. Cells were transiently transfected with 1 g of the RNAi constructs or the pRNA-U6.1/Hygro vector and grown for additional 18 h.
Soft Agar Colony Formation-To determine the effect of altered TFF3 expression (5 ϫ 10 4 ) MCF-7 or MCF-10A cells were seeded into 6-well plates and were cultured as above in RPMI medium. Cells were transiently transfected with either 1 g of the TFF3 cDNA construct or the pcDNA3.1/Myc-His-B vector, with either TFF3 RNAi or the vector pRNA-U6.1/Hygro, and grown for an additional 18 h before soft agar colony formation. For the soft agar colony formation assay, 6-well plates were first covered with an agar layer (RPMI 1640 with 0.5% agar and 10% fetal bovine serum). The upper layer contained 5 ϫ 10 3 cells in RPMI 1640 with 0.35% agar and 10% fetal bovine serum. Cells transfected with TFF3 cDNA or the pcDNA3.1/Myc-His-B and TFF3 RNAi or the vector pRNA-U6.1/Hygro were trypsinized and seeded as above. To determine whether TFF3 mediates autocrine hGH-induced oncogenic transformation, MCF10A cells were alternatively transfected with vector, hGH, or hGHϩ TFF3 RNAi constructs, and the number of colonies in soft agar was assessed as below. Medium was added to prevent drying. The plates were incubated for 9 days (for MCF-7 cells) or 14 days (for MCF-10A cells), after which the cultures were inspected and photographed. The colonies in the plates were counted after incubating them at 37°C for 9 days (MCF-7 cells) and 14 days (MCF-10A cells), and the relative number of colonies was calculated.

Temporal Analysis of Gene Expression in Response to
Exogenous hGH-We utilized the described microarray platform to characterize and compare the gene expression pattern of human mammary carcinoma cells (MCF-7) in response to stimulation with either autocrine hGH or exogenous hGH. We first profiled temporal gene expression in human mammary carcinoma cells treated with exogenous hGH to provide a framework for interpretation of the gene expression patterns stimulated by autocrine hGH. Using rigorous selection criteria (see "Materials and Methods"), we identified 587 distinct genes demonstrating time-dependent changes in mRNA level in response to exogenous hGH (Supplemental data set 1; available on the World Wide Web at www.gis.a-star.edu.sg/homepage/ toolssup.jsp). A hierarchical clustering algorithm based on Pearson correlation coefficients was applied to group genes on the basis of expression pattern similarity over the examined time points. The expression level of each gene was represented by pseudocolor in matrix format, with red representing increasing mRNA levels and green denoting decreasing mRNA levels and color intensity representing the magnitude of the expression ratio. As shown in Fig. 1A, 440 genes displayed a greater than 1.5-fold increase in expression level at two consecutive time points in response to exogenous hGH, 128 displayed a greater than 1.5-fold decrease in expression, and a small portion of the total regulated genes (19 genes) displayed a dynamic up-and down-regulated pattern.
Genes were further segregated according to the timing of their peak expression level. We grouped the gene outliers into early (peak at Յ4 h), intermediate (peak at 6 or 8 h), late (peak at Ն12 h), and sustained (ratios Ն1.5-fold at four time points but with less than 20% variance) temporal expression groups.
Among 440 up-regulated genes, 47 genes (11%) responded rapidly with peak response at 4 h or earlier, and 103 genes (23%) were up-regulated in a sustained manner by hGH. Most genes (66%) responded after 6 h, 179 genes achieved the peak response at 6 or 8 h after hGH stimulation, and 111 genes achieved the peak response after 8 h of hGH stimulation. For the 128 down-regulated genes, 11 and 10 genes responded early before 2 and 8 h, respectively; the majority of genes (107 genes, 84%) achieved the peak response at or after 8 h.
Identification of Genes Regulated by the Autocrine Production of hGH in Human Mammary Carcinoma Cells-To identify genes regulated by autocrine production of hGH in mammary carcinoma cells, RNA from the MCF-7 cell line stably transfected with the hGH gene under the control of the metallothionein 1a promoter (MCF-hGH) was hybridized to that from a translation-deficient hGH control (MCF-MUT). Microarray analysis identified 473 genes that exhibited a 1.5-fold or greater change. Among these, 321 genes were up-regulated, and 152 were down-regulated (supplementary data set 2, available on the World Wide Web at www.gis.a-star.edu.sg/ homepage/toolssup.jsp).
Clustering Analysis Reveals Expression Patterns That Distinguish Autocrine or Exogenous hGH-Analysis by comparison of expression profile between exogenous hGH and autocrine hGH demonstrated a different pattern of genetic regulation (Fig. 2). Tables I and II list genes either exclusively regulated by exogenous hGH (Table I, clusters E and F) or by autocrine hGH (Table II, clusters A and B). Only genes with known or predicted functions are presented; the complete list can be obtained from supplemental data sets 1 and 2.
As shown in Fig. 2, we were able to extract a subset of 305 genes that were remarkably different in their response to autocrine and exogenous hGH. The expression of these genes is either increased (cluster A, 162 genes) or decreased (cluster B, 143 genes) in response to autocrine hGH and exhibits no response or the opposite response to exogenous hGH. This subset of genes represent candidate genes potentially involved in the oncogenic transforming effects of autocrine hGH. Interestingly, within the subset of the genes regulated by both autocrine hGH and exogenous hGH (167 genes; 158 up-regulated and 9 downregulated), ϳ80% are genes whose gene functions are unknown or not clear.
Functional Classification of the Differential Responsive Genes to Autocrine or Exogenous hGH-To assess the potential functionality of genes in response to autocrine hGH or exogenous hGH, the gene products were grouped into Gene Ontology (available on the World Wide Web at www.geneontology.org) categories with the assistance of Standford SOURCE data base (38). Of the 400 exclusively exogenous hGH-responsive genes, 189 had known biological roles; of 305 exclusively autocrine hGH-responsive genes, 179 had known biological roles. These genes were classified by broad biological function. The encoded proteins are involved in a broad range of cellular functions and metabolic pathways. The functions of these genes include regulation of apoptosis, cell cycle, development, metabolism, biosynthesis, cell adhesion, cellular transport, stress response, and signal transduction. Regulatory molecules involved in sig-    nal transduction and metabolism comprised the bulk of the gene lists (Fig. 3). Of note, this analysis did not elucidate any functional categories of genes capable of distinguishing autocrine from exogenous hGH regulation. RT-PCR Confirms the Microarray Results-In order to verify the fidelity of data generated from the microarray, we analyzed by semiquantitative RT-PCR a number of genes identified to be regulated by exogenous and autocrine hGH, respectively. For exogenous hGH-regulated genes, we chose three genes, CRX (peak expression ratio 1.62), HOXB9 (peak expression ratio 2.00), and PAX4 (peak expression ratio 2.12). ␤-Actin served as a gene not regulated by exogenous hGH for control purposes.
As observed in Fig. 4 and consistent with the temporal microarray data, exogenous hGH stimulated an increase in the mRNA level of the three examined genes in a time-dependent manner. Furthermore, the extent of increase in expression upon hGH stimulation correlated with the peak expression ratios obtained from microarray analysis. We subsequently examined by RT-PCR a number of genes identified by microarray to be regulated by autocrine hGH in mammary carcinoma cells. We examined four genes (SNK, TC10, TFF3, and SEPP1), two of which (SNK (peak expression ratio 0.65) and TC10 (peak expression ratio 0.58)) were down-regulated by autocrine hGH. The other two genes (TFF3 (peak expression ratio 2.61) and SEPP1 (peak Genes were grouped into Gene Ontology (available on the World Wide Web at www.geneontology.org) categories with the assistance of Stanford SOURCE data base (32). expression ratio 2.05)) were up-regulated in response to autocrine hGH in human mammary carcinoma cells. RT-PCR analysis concordantly demonstrated down-regulation of the mRNA for SNK and TC-10 and up-regulation of mRNA for TFF3 and SEPP1 in response to autocrine hGH (Fig. 4, A and B). The level of ␤-actin mRNA did not alter with autocrine hGH, as described previously. To test the capacity of the microarray platform and the selection criteria to identify genes regulated exclusively by autocrine hGH and not by exogenous hGH, we therefore examined the effect of exogenous hGH on the four genes (SNK, TC10, TFF3, and SEPP1) by RT-PCR at a variety of times after stimulation. As observed in Fig. 4, C and D, there was no significant effect of exogenous hGH on the examined genes. We are therefore able to utilize the microarray data to identify specific genes exclusively regulated by autocrine hGH and not exogenous hGH. TFF3 Is a Transcriptional Target of Autocrine hGH Stimulation in Human Mammary Carcinoma Cells-The rationale for our approach was to identify specifically autocrine hGHregulated genes that are potentially involved in the ability of autocrine hGH to stimulate oncogenic transformation. One such potential gene was TFF3. Members of trefoil factor family, TFF1 and TFF3, are frequently expressed at high levels in breast cancer and in primary tumors of other tissues (39). We therefore proceeded to examine the interaction of autocrine hGH with TFF3.
We determined whether TFF3 is a transcriptional target of autocrine hGH. We therefore examined the response of a human TFF3 promoter-LUC construct to both autocrine hGH and exogenous hGH. Autocrine hGH production by MCF-hGH cells increased the activity of the TFF3 promoter by 3-4-fold in comparison with MCF-MUT cells. Concordant with the array, exogenous hGH did not stimulate TFF3 promoter activity (Fig.  5A). Surprisingly, however, exogenous hGH significantly abrogated the ability of autocrine hGH to stimulate TFF3 promoter activity. TFF3 is therefore subject to transcriptional up-regulation by autocrine hGH.
The hGH antagonist B2036 combines a single amino acid substitution impairing receptor binding site 2 (G120K) with eight additional amino acid substitutions that improve binding site 1 affinity (40). B2036 does not bind, activate, or antagonize the human PRL receptor and therefore is suitable to determine cellular effects mediated specifically through the hGH receptor (41). We have previously utilized this hGH receptor-specific antagonist in MCF-7 cells to demonstrate that the effects of autocrine hGH on mammary carcinoma cell behavior are mediated via the hGH receptor (15). We therefore utilized B2036 to demonstrate that transcriptional up-regulation of TFF3 by autocrine hGH was a specific effect of autocrine hGH through the hGH receptor and not due to a clonal selection artifact. As observed in Fig. 5B, B2036 completely abrogated the ability of autocrine hGH to stimulate the transcriptional activation of TFF3 in MCF-hGH cells.
To determine whether autocrine hGH up-regulation of TFF3 transcription also resulted in increased TFF3 protein, we examined both MCF-MUT and MCF-hGH cells by confocal laserscanning microscopy. TFF3 expression was detected in MCF- MUT cells (Fig. 6A), and autocrine production of hGH in MCF-hGH cells resulted in a higher level of expression of TFF3. Furthermore, significantly more and larger granular deposits of TFF3 were observed in MCF-hGH cells compared with MCF-MUT (Fig. 6B). Autocrine production of hGH in human mammary carcinoma cells therefore resulted in increased TFF3 protein production.
TFF3 Increases Anchorage-independent Growth in Human Mammary Carcinoma Cells-One characteristic of oncogenically transformed cells is the capacity for anchorage-independent growth. One measure of anchorage-independent growth is the ability of transformed cells to form colonies in soft agar. To determine whether forced expression of TFF3 would regulate anchorage-independent growth of human mammary carcinoma cells, we first cloned the complete human TFF3 cDNA, sequence-verified the clones, and determined expression by RT-PCR and Western blot analysis ( Fig. 7C; data not shown). Transient transfection of human mammary carcinoma cells with TFF3 cDNA increased the capacity for anchorage-independent growth, as indicated by soft agar colony formation compared with that of vector-transfected cells (Fig. 7A). The human mammary carcinoma cell line (MCF-7) used here also produces TFF3 independent of autocrine hGH stimulation. We therefore reasoned that if TFF3 increased the capacity for anchorage-independent growth, then knockdown of TFF3 would concordantly abrogate the ability of mammary carcinoma cells to form colonies in soft agar. We therefore examined anchorage-independent growth of mammary carcinoma cells as indicated by soft agar colony formation after transient transfection of a TFF3 RNAi construct we generated as described under "Materials and Methods" (Fig. 7D). Small interfering RNA to TFF3 reduced the level of TFF3 in MCF-7 concordant with transfection efficiency (ϳ60%; Fig. 7D). Inhibition of TFF3 expression abrogated the number of colonies formed by MCF-7 cells in soft agar by more than half (Fig. 7B). TFF3 therefore regulates the oncogenicity of human mammary carcinoma cells.
Forced Expression of TFF3 in Immortalized Human Mammary Epithelial Cells Results in Oncogenic Transformation-To indicate whether increased expression of TFF3 would result in de novo oncogenic transformation of human mammary epithelial cells, we utilized the immortalized human mammary epithelial cell line MCF-10A. When grown attached to a plastic substrate, these cells display normal epithelial morphology and do not form colonies in soft agar (42). We first determined the expression of TFF3 by RT-PCR in MCF-10A cells (Fig. 8B). In contrast to MCF-7 cells, in which TFF3 expression was readily detected with 15-20 cycles of PCR, detection of TFF3 expression in MCF-10A cells required 35-40 cycles of PCR. Transient transfection of TFF3 cDNA increased TFF3 expression in MCF-10A cells. Control transfected cells were largely ineffective in colonization of soft agar, whereas a significant number of colonies formed from cells transiently transfected with TFF3 cDNA (Fig. 8A). This is remarkable, given the relative inefficiency of transient transfection and that cells were grown for 14 days in soft agar before determination of colony formation. Thus, increased expression of TFF3 is sufficient to support anchorage-independent growth and potentially to oncogenically transform immortalized human mammary epithelial cells.
TFF3 Mediates Autocrine hGH-stimulated Oncogenic Transformation-We have previously demonstrated that autocrine hGH stimulates oncogenic transformation of immortalized human mammary epithelial cells (43). Since TFF3 expression is regulated by autocrine hGH and TFF3 itself supports anchorage-independent growth, we reasoned that at least part of the autocrine hGH-stimulated oncogenic transformation will be mediated by TFF3. We therefore transiently transfected the immortalized human mammary epithelial cell line MCF-10A with either vector plus TFF3 RNAi or hGH cDNA plus TFF3 RNAi and assessed oncogenic transformation by soft agar colony formation as described. As demonstrated in Fig. 9, transient transfection of hGH cDNA resulted in soft agar colony formation by MCF-10A cells in accord with our previously published data (18). Depletion of TFF3 expression largely abrogated the ability of autocrine hGH to stimulate anchorageindependent growth. Autocrine hGH-stimulated oncogenic transformation of human mammary epithelial cells is therefore dependent on TFF3. DISCUSSION We have utilized a genome-wide approach with a stringent selection algorithm to identify a subset of 305 genes that are specifically regulated by autocrine production of hGH in mammary carcinoma cells and not by exogenously added hGH, which would mimic the effect of endocrine-delivered hGH. We exploited the differential oncogenic capacity of autocrine hGH compared with exogenous hGH (18) to identify, within this subset, one gene with the capacity to oncogenically transform the human mammary epithelial cell. Such an approach based on differential functional activity is therefore valid for the identification of novel human mammary epithelial oncogenes. Based on further functional analysis, we have identified one gene, TFF3, with the potential capacity to result in oncogenic transformation of the human mammary epithelial cell.
Although we have utilized the differential oncogenic capacity of autocrine hGH to identify novel human mammary oncogenes, we cannot exclude the possibility that some of the genes co-regulated or solely regulated by exogenous hGH possess the potential for oncogenic transformation. One explanation is that exogenous hGH may also stimulate or repress other regulatory pathways, which would abrogate the oncogenic effect of the potential oncogene. Indeed, we have recently described a situation whereby exogenous hGH results in the transcriptional up-regulation of a homeodomain-containing gene, PITX1, which subsequently results in the direct transcriptional upregulation of p53. 2 However, exogenous hGH stimulation of mammary carcinoma cells does not result in p53-mediated apoptosis. One mechanism by which a potential apoptotic signal stimulated by exogenous hGH may be avoided is by the regulation of genes that abrogate the effect of p53. One such identified gene also regulated by exogenous hGH is PTTG1, 2 which results in the proteosomal degradation of p53 (44). Adding more complexity is the fact that p53 transcriptionally down-regulates PTTG1 (45) and PITX1 transcriptionally upregulates PTTG1. 3 Thus, regulation of single linear pathways or a single gene is not entirely reliable to predict the response of a cell to a specific stimulus. In any case, we have demonstrated the utility, if not comprehensive, of the approach to identify potentially oncogenic genes based on the differential oncogenic potential of autocrine hGH.
One major mechanism by which GH affects cellular and somatic function is by regulating the level of specific mRNA species (46). Some of these GH-regulated genes code for trophic factors such as insulin-like growth factor-1 (47), which act in an intermediary role to execute the cellular effects of GH. Indeed, GH has been demonstrated to regulate the level of a number of trophic factors in specific tissues including hepatocyte growth factor in liver (48), epidermal growth factor in kidney (49), basic fibroblast growth factor in chondrocytes (50), interleukin-6 in osteoblasts (51), bone morphogenetic proteins 2 and 4 in fibroblasts (52), interleukin-1␣ and interleukin-1␤ in thymus (53), and preadipocyte factor-1 in adipocytes (54) and islet ␤-cells (55). It is apparent that many of the effects of autocrine hGH on mammary carcinoma cell function are also mediated by genetic regulation of specific trophic factors. Indeed, we have previously demonstrated that autocrine hGH also exerts effects on cellular function by transcriptional repression of protein effector molecules, such as PTGF-␤, that promote cell cycle arrest and apoptosis (17). From the results of this microarray analysis, we have identified numerous other soluble peptide factors that could mediate many of the pleiotropic effects of autocrine hGH on mammary carcinoma cell function. These include TFF1, TFF3, BMP7, laminin 3, the postulated interleukin 27, osteomodulin, and TRH among potential others. Autocrine hGH also down-regulates a number of soluble and secreted peptides including thymosin, laminin 5, and thrombospondin-1 (Tsp1). The identification of Tsp1 as an autocrine hGH-regulated gene is of particular interest, given the well established role of repression of Tsp1 expression in tumor progression. Tsp1 was the first naturally occurring inhibitor of angiogenesis to be identified (56). Tsp1 inhibits the activity of metalloprotease 9 (57), an extracellular matrix metalloproteinase that releases vascular endothelial growth factor sequestered in the extracellular matrix (58). In addition, Tsp1 can act directly to inhibit angiogenesis by binding to the CD36 receptor, which is present on endothelial cell surfaces (59). We have previously demonstrated that autocrine hGH production by human mammary carcinoma cells results in increased activity of matrix metalloprotease 9 associated with epitheliomesenchymal transition to an invasive phenotype (19). Since we observed no change in the level of matrix metalloprotease 9 mRNA (19), it is plausible that the increased metalloprotease 9 activity is consequent to autocrine hGH down-regulation of Tsp1. The inhibition of Tsp1 by autocrine hGH is also suggestive that autocrine production of hGH will increase neovascularization of mammary carcinoma, pivotal for clinical progression of the disease.
We have demonstrated herein that autocrine hGH results in an increase in both TFF1 and TFF3 gene expression. In humans, three distinct members of the trefoil peptides have been identified. TFF1 or pS2 was first detected in a mammary cancer cell line as an estrogen-inducible gene (39). In the human stomach, it is predominantly located in the foveolar cells of the gastric mucosa. TFF2 (formerly spasmolytic polypeptide or 2 D. X. Liu and P. E. Lobie, manuscript in preparation. 3 D. X. Liu and P. E. Lobie, unpublished results. SP) was first purified from porcine pancreas and is expressed in mucous neck cells, deep pyloric glands, and Brunner's glands (60). TFF3, or intestinal trefoil factor, was the last to be identified (61) and is claimed to be predominantly expressed in the goblet cells of the small and large intestine. The trefoil peptides are involved in mucosal healing processes and are expressed at abnormal elevated levels in neoplastic diseases (62). A wide range of human carcinomas and gastrointestinal inflammatory malignancies, including peptic ulceration and colitis, Crohn's syndrome, pancreatitis, and biliary disease, aberrantly express trefoil peptides (63)(64)(65)(66)(67). The first characterized trefoil peptide, pS2-TFF1, was isolated from the mammary carcinoma cell line MCF-7, utilized herein, and is expressed in normal and neoplastic lesions of the human mammary gland (39). Several studies have investigated the value of TFF1 expression as a marker of hormonal responsiveness and its correlation with estrogen receptor status, which may have prognostic and therapeutic relevance (68). TFF3 is widely co-expressed with TFF1 in malignancies of the human mammary gland, whereas TFF2 is not expressed in the mammary epithelial cells (69). The trefoil peptides apparently possess divergent function in the mammary gland, with TFF1 functioning as a motogen and TFF2 stimulating branching morphogenesis and cell survival (70). We have demonstrated here that expression of TFF3 is sufficient to support anchorage-independent survival and proliferation of human mammary epithelial cells. We have therefore identified TFF3 as a potential human mammary epithelial oncogene, since anchorage-independent growth is a pivotal characteristic of oncogenic transformation. Further analysis of the function of TFF3 in the human mammary epithelial cell will delineate its full oncogenic potential and the mechanism by which this is achieved.
In conclusion, we have exploited the discrepancy in the oncogenicity of autocrine and exogenous hGH in an attempt to identify molecules that could potentially be involved in oncogenic transformation of the human mammary epithelial cell. We were able to extract a subset of 305 genes that were remarkably different in their response to autocrine and exogenous hGH. Functional analysis of one of the identified autocrine hGH-regulated genes, TFF3, determined that its expression is sufficient to support anchorage-independent growth of human mammary carcinoma cells. Further functional characterization of the identified subset of specifically autocrine hGH-regulated genes will further delineate novel oncogenes for the human mammary epithelial cell.