The progression of primary tumors to an invasive phenotype requires dynamic changes in multiple cellular and local tumor microenvironment markers. In this study, we report a genomic approach to assess gene transcriptional changes upon overexpression of ErbB receptors, in vitro and in vivo, focusing on markers involved in the regulation of the tumor microenvironment. ErbB receptors (ErbB-1/epidermal growth factor receptor, ErbB-2, ErbB-3, and ErbB-4) were stably overexpressed in a polyclonal cell population as single or paired combinations using murine and human breast cell models. The overall numbers of known genes that are up- or down-regulated was significantly higher in cells and tumors overexpressing paired combinations of receptors compared with cells and tumors overexpressing single ErbB receptors. Genes encoding components of cell-cell structures, extracellular matrix, coagulation factors, and angiogenesis were predominantly affected by the most active ErbB receptor combinations and were predictive of the aggressive in vivo tumorigenicity, a feature that was not always seen in vitro. Among ErbB-regulated tumor microenvironment markers detected by the genomic analysis, thrombospondin 1, an endogenous inhibitor of angiogenesis, was additionally validated in relation to tumor growth phenotype. Thrombospondin 1 mRNA and protein were down-regulated by specific ErbB receptors, in vitro and in both rodent and human ErbB-induced tumors, consistent with the extent of tumor growth and tumor vascularization associated with specific ErbB receptors. In summary, our genomic results highlight the broad diversity of ErbB-regulated cancer-associated genes and revealed several novel targets that may have potential therapeutic applications for targeting tumor progression involving aberrations of ErbB receptors.

The impact of the ErbB3 tyrosine kinase receptor family, which include ErbB-1 (epidermal growth factor receptor/HER-1), ErbB-2 (HER-2), ErbB-3 (HER-3), and ErbB-4 (HER-4), on biochemical pathways has been extensively investigated in two-dimensional in vitro cell culture systems. This has provided evidence of the oncogenic properties of these receptors and the diversity of their signal transduction pathways attributed primarily to their propensity to form many homodimeric and heterodimeric receptor combinations. Furthermore, these in vitro studies have established that the biological activity of the ErbB receptors is primarily associated with ErbB heterodimers compared with less active homodimers (reviewed in Refs. 1, 2).

Under normal conditions, ErbB receptors are expressed at very low levels in a variety of tissues and cells of epithelial, mesothelial, endothelial, and neuronal origin where they elicit broad biological functions, including organ development, maintenance of cell proliferative and differentiation states, and the regulation of tissue homeostasis (2). This is highlighted by the finding that inactivation of erbB-1, erbB-2, erbB-3, or erbB-4 genes in mice results in embryonic or early lethality because of aberrant development of several organs, predominantly the heart, brain, and gastrointestinal lining (1, 2). Additional evidence has increasingly supported that abnormal expression of members of the ErbB receptors in cell lines and transgenic mice is among the earliest changes that lead to the imbalance between proliferation and differentiation, and tumorigenesis. For instance, transgenic mice overexpressing the ErbB-1 ligand transforming growth factor α or the ErbB-3/4 ligand heregulin develop various types of hyperplasia and neoplasia (3, 4, 5, 6) and mice heterozygous for heregulin develop hyperplasia in epithelial tissues, including the mammary gland, pancreas, and intestine (7). In humans, overexpression and/or amplification of specific members of the ErbB receptor family is a common feature of many solid tumors, including carcinomas of the breast, lung, prostate, ovary, bladder, head and neck, and the aerodigestive tract (reviewed in Ref. 1). Clinical investigations have correlated high ErbB-2 expression to shorter survival, invasive disease, and in many cases to poor response to specific chemotherapy drugs, hormonotherapy, and/or radiotherapy (reviewed in Ref. 8). Similar clinical observations have reported the association between overexpression of ErbB-1 and poor clinical prognosis (reviewed in Ref. 9), whereas very little is known of ErbB-3 and ErbB-4 overexpression and therapeutic outcome. These prognostic correlative studies are not without exception as there is some evidence that low levels of ErbB-2 expression, for example, may also predict a poor prognosis compared with tumors with intermediate levels of ErbB-2 expression (10, 11, 12, 13). It is likely that the association between ErbB overexpression and clinical outcome of disease may depend on the tumor type where multiple markers can cooperate or operate simultaneously (14, 15, 16, 17). Although most clinical studies on ErbB and prognosis have focused on a single receptor, mostly ErbB-1 or ErbB-2, co-overexpression of more than one member of the ErbB family and non-ErbB receptors occurs frequently within the same tumor and can contribute to a distinct phenotype when compared with those tumors overexpressing a single receptor. For instance, co-overexpression of ErbB-1 and ErbB-2 is common in many solid tumors (18, 19, 20, 21). Co-overexpression of ErbB-2 and ErbB-4 is common in childhood medulloblastomas (22) and ovarian carcinomas (23). Tumors overexpressing ErbB-2, including breast carcinomas, also can display elevated ErbB-3 phosphorylation (24). This is consistent with the observation that mammary tumors from neu transgenic mice exhibit elevated expression of ErbB-3 (25). In addition, there are clinical studies where ErbB-3 overexpression was correlated with breast cancer progression (26, 27). Changes in ErbB-4 expression, either low or overexpression, have also been seen in granulosa cell tumors, adenocarcinomas, and squamous cell carcinomas (28, 29). The clinical impact of the ErbB receptors on disease phenotype under conditions where concomitant co-overexpression of more than one receptor occurs in the same tumor/cell remains to be established. This is important as in vitro studies show differential signaling and biological functions of specific ErbB heterodimers or homodimers (1, 2).

In this study, we conducted a comprehensive screen for transcriptional changes associated with ErbB receptors, expressed as single or paired combinations, in cells maintained in a two-dimensional cell culture system. Because results on cell lines often represent a distorted and incomplete picture of the in situ physiopathology of cancer where host and the tumor microenvironment play critical roles in tumor growth and progression, e.g., cross signals between epithelial, stromal, and neighboring endothelial and mesothelial cells, we expanded our screening to matched primary tumors using xenograft models. A correlation between transcriptional changes in cellular factors, tumor microenvironment, and tumorigenesis is established using genomic analysis. We report that several components of the ECM and angiogenic pathways are affected by the most active ErbB receptor combinations and are predictive of the in vivo tumorigenicity associated with these receptors. We have identified several previously undescribed markers associated with ErbB overexpression that may account for the resultant aggressive properties associated with specific ErbB receptor combination in vivo. These results expand current knowledge of ErbB-regulated genes and revealed several novel targets for potential therapeutic intervention in tumors displaying aberrant ErbB receptor expression.

Cell Culture.

The NIH-3T3 cell clone used in this study was established from NIH-3T3 mouse fibroblasts (American Type Culture Collection, Manassas, VA; ATCC CRL1658). The human cancer cell lines MCF-7, MDA-435, and BT-20 were from the American Type Culture Collection. Cells were maintained in the DMEM or RPMI medium (Mediatech, Washington, DC) supplemented with 10% FCS (NIH-3T3) or 10% fetal bovine serum (MCF-7, MDA-435, and BT-20 cells) and penicillin-streptomycin antibiotics. Cells were maintained in an atmosphere of 5% CO2. The 293GPG-packaging cells were kindly provided by Dr. Jacques Galipeau (McGill University, Montreal, Quebec, Canada) and were maintained in DMEM (Mediatech), 10% heat-inactivated FBS supplemented with 300 μg/ml G418 (Mediatech), and 2 μg/ml puromycin (Sigma Chemical Co., Oakville, Ontario, Canada), 1 μg/ml tetracycline (Fisher Scientific, Nepean, Ontario, Canada), and 50 units/ml of penicillin-streptomycin antibiotics.

Stable Expression of ErbB Receptors.

ErbB receptors were expressed, as single or paired combinations, using bicistronic retrovectors that coexpress each ErbB receptor member with the enhanced GFP as marker for proviral transfer in the target cells. Detailed methodology for vector construction, selection of packaging cells, and transduction of target cells were described in detail in Yen et al.(33). Briefly, recombinant retroparticles were generated by stable transfection of the 293GPG packaging cell line with subsequent production of high titer retrovirus. Target cells were transduced with single or paired combinations of AP2-ErbB retroviral particles. Stably transduced cells were expanded, and flow cytometry analysis was performed with an Epics XL/MCL Coulter analyzer to verify gene transfer efficiency as measured by GFP fluorescence.

Immunofluorescence Analysis.

For immunofluorescence study, cells overexpressing ErbB receptors, either singly or in paired combinations, were seeded on coverslips for 2 days at a density of 50,000 cells/35-mm dish. The cells were rinsed with PBS and fixed with 3% (w/v) PFA in PBS for 5 min, followed by an incubation in precooled methanol (−80°C) for 15 min at −20°C. The cells were then rinsed with PBS and blocked with 2% BSA, 2% normal goat serum, and 0.2% gelatin in PBS. The cells were then incubated with the following primary antibodies: monoclonal anti-ErbB-1 (Ab-1; Calbiochem); monoclonal anti-ErbB-2 (Ab-3; Oncogene Science, Cambridge, MA); polyclonal anti-ErbB-3 (C-17; Santa Cruz Biotechnology); or polyclonal anti-ErbB-4 (clone H4.77.16; NeoMarkers) for 1 h at room temperature. All antibody dilutions were made in blocking solution. Cells were washed three times with 0.2% BSA in PBS and then incubated with appropriate secondary antibodies conjugated to Texas Red (Jackson ImmunoResearch Laboratories, West Grove, PA) for 30 min. The coverslips were then washed with PBS and mounted with Gelvatol (Air Products and Chemicals, Inc., Allentown, PA). Confocal analysis was performed using a Zeiss LSM 410 confocal microscope.

Growth of ErbB Tumor Xenografts in Nude Mice.

In vivo studies were approved by the McGill Animal Care Committee (Protocol no. 4101) and were conducted in accordance with institutional and Canadian Federal Guidelines. Subconfluent cells were suspended in PBS and injected s.c. into the flanks (106/0.1 ml) of BALB/c Nu/Nu mice. Tumor volumes were measured every second or third day by external measurement using a caliper and the equation: volume = π/6 (length × width2). At the end of experiments, tumors were isolated under sterile conditions, snap frozen in liquid nitrogen and stored at −80°C, or fixed in formalin for subsequent immunohistochemical studies as indicated in the results. Comparison of growth rates between cells expressing the various receptors was determined by variance analysis between specific groups. Statistical significance between various ErbB-expressing tumors was defined at the 0.05 level using a two-tailed t test.

Sample Preparation and Oligonucleotide Array Hybridization.

The array analysis was carried out at the McGill Genome Centre (McGill University, Montreal, Quebec, Canada). Sample preparations were based on the Affymetrix Expression Analysis Sample Prep Protocol (Affymetrix, Huntsville, AL). Total RNA was isolated from either cells or frozen tumor tissue using RNAzol B (Tel-Test, Inc., Friendwood, TX). First strand cDNA was synthesized from 20 μg of total RNA with a special oligo (dT)24 primer containing a T7 RNA polymerase promoter at its 5′ end in 20 μl of first strand reaction mix at 42°C for 1 h. The second strand was synthesized in a second strand reaction mix for 2 h at 16°C. The double-stranded cDNA was then phenol/chloroform extracted and ethanol precipitated. Biotin-labeled cRNA was generated from the cDNA by in vitro transcription using a BioArray High Yield RNA Transcript Labeling Kit (Enzo Diagnostics, Farmingdale, NY). The labeled cRNA was purified using a RNeasy Mini Kit (Qiagen, Valencia, CA) and then fragmented in fragmentation buffer [40 mm Tris-acetate (pH 8.1), 100 mm potassium acetate, and 30 mm magnesium acetate] for 35 min at 94°C. As specified in the Affymetrix protocol, the fragmented cRNA was hybridized to the murine U74A (Affymetrix) chip for 16 h at 45°C in a GeneChip Hybridization Oven 640 (Affymetrix). Immediately after the hybridization, the hybridized probe was washed and stained in an automated protocol by a GeneChip Fluidics Station 400 (Affymetrix). Detailed methodology for screening of gene oligonucleotide microarray has been described earlier (32).

Array Data Analysis.

The arrays were scanned, and the digitized image data were processed with GeneChip software available from Affymetrix. The mouse chip we used contains 12,000 genes where each gene is represented by 20 pairs of oligonucleotide probes. Each probe pair is a 25-bp in length and is either a perfect match or a single central mismatch. Hybridization signals fluctuate between different probe pairs and, presumably, reflect nonspecific hybridization by background RNAs and differences in hybridization kinetics between different sequences. To correct for this, hybridization intensities are expressed as average differences (subtracting the intensity of the mismatch from the intensity of the perfect match). Also, average difference values for probe pairs emitting hybridization signals much stronger than their neighbors (>3 SD away from mean) are filtered out to prevent significant contribution to the reading of average level of intensity of the gene. Average difference values from the murine GeneChip analysis were exported into Excel (Microsoft Corp.) and then converted to ratios of fold expression over background (NIH-3T3 cells transduced with control AP2 virus). Each array analysis was done in duplicate or triplicate experiments, and the average ratios are reported. Compiled data were screened by a nonparametric ANOVA using a P of <0.05. Fold expression ratios were sorted using Excel and only genes with expression ratios > 2.5-fold, which corresponded to a minimum magnitude change of 200 fluorescence units, were considered significant. Fold expression levels < 2.5 are considered to be unchanged unless statistically significant over many experiments (Affymetrix). In vitro ErbB heterogeneity was analyzed by grouping significantly changed genes from the relevant pairwise comparisons (e.g., ErbB-1/2 versus AP2, ErbB-1 versus AP2, and ErbB-2 versus AP2). These groupings were additionally organized by Venn diagrams where genes common to some or all pairwise comparisons are visually represented by discrete and intersecting sections of the diagram. Our in vivo study of heterodimer heterogeneity was performed slightly differently because it was not possible to compare expression levels of tumors to control AP2-transduced NIH-3T3 cells because these are not tumorigenic. Because an ErbB heterodimer consists of two receptor family members, we have attempted to identify genes that presumably contribute to the overexpressed heterodimer phenotype by pairwise comparisons with the relevant homodimers (e.g., ErbB-1/2 versus ErbB-1 and ErbB-1/2 versus ErbB-2). The intersection of these groupings, as represented in the in vivo Venn diagrams, identified differentially expressed candidates.

Semiquantitative RT-PCR for erbB Gene Expression.

Total RNA was isolated using RNAzol B according to manufacturer’s indications (Tel-Test, Inc.). Reverse transcription and subsequent amplification were carried out in a single tube using the One-Step RT-PCR kit (Qiagen), using primers specific for each erbB gene member (Table 1). Primers specific for amplifying gapdh gene were used to control for the amounts of cDNA generated from each sample (Table 1). Amplification was carried out using the supplied enzyme mix of Omniscript reverse transcriptase, Sensiscript reverse transcriptase, HotStarTaq DNA polymerase (Qiagen), and buffer and primer concentrations of 0.6 μm in a final volume of 25 μl. Reverse transcription proceeded during a 30 min incubation at 50°C, followed by an activation step of 15 min at 95°C, and the reactions were cycled 20–35 times at 94°C for 30 s, 60–65°C for 30 s, and 72°C for 1 min, followed by a last extension for 10 min at 72°C. PCR cycling was optimized to ensure that product intensity was within the linear range of amplification. PCR products were visualized on 1% agarose gels and stained with ethidium bromide.

Northern Blot Analysis.

Cells at 70% confluence were lysed, and RNA was extracted using RNAzol B reagent (Tel-Test, Inc.). Twenty μg of total RNA were resolved by electrophoresis through a 1% denaturing formaldehyde gel and then transferred to Zeta-probe membrane (Bio-Rad). Equal loading of RNA in each lane was evaluated by ethidium bromide staining before transfer. Tsp1 probe was prepared as follows: a cDNA from a mouse mammary library was subjected to PCR amplification using a 5′ primer (5′-CCTGCCAATCCCTGCTTT-3′) and a 3′ primer (5′-GTCTTCCTGCCCCGAGTT-3′). Amplified 500-bp tsp1 product was agarose gel-purified, then 32P-labeled (Oligolabeling kit; Amersham-Pharmacia). Membranes were hybridized overnight with TSP1 probe at 42°C as described previously (30). Membranes were stripped and then reprobed for gapdh, used as internal control.

Immunohistochemistry.

Tumors were either snap frozen in liquid nitrogen (TSP1 and CD31) or fixed in 10% neutral-buffered formalin and embedded in paraffin wax (ErbB2). Sections were immunostained using a mouse monoclonal anti-TSP1 (Ab-4; NeoMarkers), a rat monoclonal antimouse CD31 (Mec 13.3; PharMingen), or a mouse monoclonal against ErbB-2 (Ab-3; Novacastra). Immunostaining was performed on 5-μm thick sections as described previously (31, 33). Sections were counterstained with Harris’ hematoxylin and mounted. All sections were analyzed by conventional light microscopy and digital photography (Leitz Aristoplan).

In Vitro and in Vivo Characterization of the ErbB Cell Model.

To address the effect of ErbB signaling on the gene transcription profile, we transduced polyclonal cell populations derived from a NIH-3T3 clone with bicistronic retrovirions that express each member of the ErbB family and the enhanced GFP. Control cells were transduced with retroviral particles without the coding sequence for an ErbB receptor (AP2). As shown in Fig. 1,A, no detectable levels of ErbB-1, ErbB-2, ErbB-3, and ErbB-4 were seen in cells transduced with control retroviral particles (AP2) by immunofluorescence (Fig. 1,A) or Western blot analysis (33). Also exposure of serum-starved control AP2-transduced cells to 20 ng/ml epidermal growth factor, the ligand for ErbB-1, or heregulin β1, the ligand for ErbB-3 and ErbB-4 receptors, failed to induce cell proliferation (data not shown). A very high efficiency of ErbB receptor expression was seen in cells expressing either single (Fig. 1,B) or paired receptor combinations (Fig. 1 C; both immunofluorescence and fluorescence-activated cell sorting analysis revealed that >95% of transduced cells expressed the receptors). The level of expression of each ErbB receptor was confirmed by Western blot analysis, and ligand-specific tyrosine phosphorylation of the activated receptors was reported earlier (33).

To determine the impact of ErbB receptor overexpression on in vivo tumorigenicity, control and ErbB-transduced cells were transplanted s.c. into Balb/C nude mice. Overexpression of the various ErbB receptor combinations on the ErbB-expressing tumors was confirmed on tumor sections collected at the time of sacrifice using RT-PCR and immunohistochemistry (described below and Ref. 33). As expected, none of the control AP2-derived cells were able to form tumors (Fig. 2). Overexpression of ErbB-3, ErbB-4, or ErbB-1/3 resulted in the development of small localized nodules that become palpable only after ∼35 days but remained stable or regressed thereafter. Overexpression of ErbB-1 or ErbB-2 alone resulted in tumor formation with ErbB-2 being the most potent in terms of early tumor formation [day 23 compared with day 31 for ErbB-1 (P = 0.01)]. Interestingly, cells overexpressing ErbB-1/2, ErbB-1/4, and ErbB-2/4 receptor combinations were the most aggressive, i.e., tumors appeared by day 4 for ErbB-1/2 tumors and within 7–11 days for ErbB-2/4 and ErbB-1/4 tumors (P < 0.005). Within ∼1 week, these same tumors reached an approximate size of 1 cm3. In contrast, tumors induced by cells overexpressing ErbB-2/3 appeared only on day 23 but showed very aggressive growth thereafter compared with the other receptor combinations (P < 0.001). Tumors induced by cells overexpressing ErbB-3/4 appeared on day 27 and displayed a slow growth rate. This pattern of tumor growth was confirmed in three independent in vivo experiments using different cell stocks and with n ≥ 8 mice/condition for each experiment. Immunohistochemistry on tumor sections revealed a lack of correlation between PCNA staining and tumor growth rates (data not shown).

In Vitro Comparisons of Gene Expression Profiles.

The transcriptional profile associated with the overexpression of ErbB receptors was carried out on NIH-3T3 serum-stimulated cells to mimic the in vivo situation where multiple growth stimuli operate. The ErbB receptor overexpression in the transduced cells was confirmed by RT-PCR on duplicate samples (Fig. 3,A). We used an Affymetrix murine U74A microarray chip containing 12,000 genetic elements. mRNA from cells transduced with control retroviral particles (AP2) was used as baseline. To analyze the specificity of gene expression profiles associated with specific members of the ErbB receptors, we compiled gene expression profiles using Venn diagram distributions. Fig. 4 summarizes the distribution of genes whereby each circle represents the whole group of differentially expressed genes, including ESTs, as revealed when making comparison of gene expression between cells overexpressing a single ErbB receptor versus the corresponding paired receptor combinations. Overlapping intersections in the Venn diagram represent concomitant gene expression within these groupings and highlight the presence of both distinct and overlapping genes among cells expressing each ErbB paired combination versus matched cells expressing the respective single receptors.

The overall number of cancer-associated genes affected by ErbB receptors was higher in cells overexpressing ErbB-paired combinations compared with cells overexpressing single receptors (Tables 2 and 3). Also, many genes that are associated with the overexpression of specific ErbB receptor combinations are distinct from those seen in cells overexpressing either of the respective single receptors (Tables 2 and 3). For example, 7–9-fold down-regulation of the stromal cell-derived factor 1, a chemokine involved in hematopoiesis and tumor invasion, was observed in cells overexpressing all ErbB single or paired combinations. Genes such as TSP1, an endogenous inhibitor of angiogenesis, was down-regulated by specific single and paired receptor combinations, but fold changes were more pronounced in cells overexpressing paired receptor combinations compared with matched cells overexpressing single receptors. More important, a variety of genes such as TSP2, tenascin C, vinculin, plasminogen activator, proteoglycans, procollagens, and several kinases were selectively affected by distinct receptor pairs compared with a single receptor (Tables 2 and 3). This points to a synergy that is specific for overexpression of paired ErbB receptors compared with single receptors. Overall comparisons revealed that most genes that are affected by ErbB in serum-stimulated conditions have been reported to play key roles in cell signaling, cell cycle, and cell-cell interaction. In addition to known genes, many ESTs were differentially affected by ErbB receptors (data not shown but the full list is available online4).

In Vivo Comparisons of Gene Expression Profiles.

To determine the gene transcription changes associated with in vivo tumorigenesis induced by overexpression of single or paired combinations of ErbB receptors, we carried out a similar array study on tumors induced by the same cells used for the in vitro study. The ErbB receptor expression was confirmed by RT-PCR on duplicate samples taken from tumor tissues at the time of sacrifice (Fig. 3,B). Because control cells transduced with empty viral particles do not form tumors in vivo, we were not able to make pairwise comparisons with these cells and, therefore, not able to reveal genes differentially expressed in association with ErbB homodimer overexpression. Instead, we have attempted to indirectly dissect the effects of ErbB heterodimer overexpression by pairwise comparison with its respective receptor components (Fig. 5). For example, if analysis of ErbB-1/2 with ErbB-1 as background represents genes that are differentially expressed in association with ErbB-2 overexpression and analysis of ErbB-1/2 using ErbB-2 as background represents differentially expressed genes in association with ErbB-1 overexpression, then the intersection of these two groups of genes are presumably associated indirectly with ErbB-1/2 overexpression. Similar analyses were performed for all other heterodimers. We included homodimers containing ErbB-3, ErbB-4, and ErbB-1/3 in our comparisons, although these cells formed only very small nodules after a delayed period. However, these nodules were confirmed to overexpress the appropriate ErbB receptors (Fig. 3 B).

Overexpression of ErbB heterodimers, in general, was associated with the most aggressively growing tumors, e.g., tumors overexpressing ErbB-1/2, ErbB-2/3, ErB-1/4, and ErbB-2/4 are more powerful in bringing about such an aggressive growth phenotype compared with ErbB-1/3 and ErbB-3/4. It is therefore clear that there is a cooperative effect when two receptors are overexpressed, resulting in a unique gene expression profile for each specific heterodimer. As expected, very different expression profiles was observed when we compare the composition of genes residing in the intersections of the in vivo Venn diagrams (Fig. 5 and Table 4). Noticeable changes in genes that are involved in the regulation of the tumor microenvironment appear a shared characteristic for the most aggressive receptor combinations in vivo. The genes that are responsible for the resultant phenotype seem specifically associated with distinct ErbB heterodimer. For example, the insulin-like growth factor 2 and insulin-like growth factor-binding protein 5 encoding genes were highly elevated in aggressive tumors overexpressing ErbB-1/4 and ErbB-2/3, whereas genes such as granzyme D-G and gap junction membrane channel protein β 3 were elevated in the weakly aggressive ErbB-3/4 tumors (Table 4). Complete details of genes and ESTs that are affected in vivo are available online.4

Target Validation of Selected Markers Involved in the Regulation of Tissue Microenvironment.

Comparison of the gene expression patterns between ErbB-expressing NIH-3T3 cells maintained in vitro and their matched tumors revealed that the predominant changes associated with overexpression of the most aggressive ErbB combinations occur in markers associated with the regulation of the local tumor microenvironment (Tables 2,3,4). We have chosen the gene tsp1 as a candidate marker for additional validation because it is a potent endogenous inhibitor of angiogenesis, a process that we have shown to be regulated by specific ErbB receptors (30, 33). Fig. 6,A shows the expression profile of tsp1 in NIH-3T3 cells examined by Northern blot analysis. Compared with control cells or cells overexpressing a single ErbB receptor, tsp1 mRNA was reduced in cells expressing ErbB-2 and, to a lesser extent, ErbB-1 receptor. However, tsp1 was strongly down-regulated in paired combinations of ErbB receptors with the exception of cells overexpressing ErbB-3/4. The most pronounced effect was observed in cells overexpressing ErbB-1/2, ErbB-1/4, and ErbB-2/4 receptors. In vivo, TSP1 expression was examined by immunohistochemistry on tumor tissue sections. The extensive cell heterogeneity attributable to tumor infiltration by mesothelial cells, endothelial cells, inflammatory cells, and components of the ECM made Northern blot analysis nonreliable. As shown in Fig. 6,B, TSP1 immunostaining was reduced in tumors overexpressing ErbB-2, ErbB-1/3, ErbB-2/3, and ErbB-2/4 in comparison to tumors overexpressing other ErbB combinations. In some tumors, e.g., ErbB-1 and ErbB-4, a large heterogeneity in TSP1 immunostaining was observed within the same tumor section, likely a consequence of cell heterogeneity. Also, we noted that tumors overexpressing some ErbB combinations such as ErbB2/3 have low TSP1 expression and were highly vascularized as revealed by CD31 staining (Fig. 6 B).

To validate the relevance of the results on NIH-3T3 model to human cancer, we compared the expression of TSP1 in human tumor cells and xenografts induced by MCF-7, BT-20, and MDA-345 breast cancer cells overexpressing ErbB-2. These cells also express various levels of ErbB-1, ErbB-3, and ErbB-4, and therefore, it is likely that any biological effect can be contributed by heterodimeric interactions between the four ErbB receptor members. As indicated in Fig. 7,A, tsp1 mRNA is significantly down-regulated in cultured cells overexpressing ErbB-2 as well as in matched tumors (Fig. 7 B), which is consistent with the array analysis results and the enhanced vascularity of the ErbB-induced tumors (33).

In this study, we sought to: (a) identify the differential transcriptional status of genes regulated by various members of the oncogenic ErbB receptor family in vitro and in vivo; (b) interface the in vitro and in vivo situations to understand the relative contribution of each member of the ErbB family to the regulation of tumor-microenvironment; and (c) identify ErbB-associated tumor markers that may represent potential therapeutic targets.

We focused on gene expression profiles using a murine model consisting of nontransformed and ErbB-transformed polyclonal NIH-3T3 cells overexpressing single or paired combinations of ErbB receptors. These cells were chosen because they do not express detectable amounts of ErbB receptors unlike cells of epithelial origin. The results were additionally validated in human cancer cells. The parental NIH-3T3 cells transduced with control retroviral particles were not transformed and nontumorigenic when injected in immunocompromised nude mice. Cells overexpressing specific combinations of ErbB receptors, however, induced tumor formation in vivo. The tumor growth phenotypes varied depending on the type of ErbB receptors expressed: from none or very low tumorigenicity for cells overexpressing ErbB-3, ErbB-4, and ErbB-1/3, in agreement with previous studies (34, 35), to very aggressive tumorigenicity for cells overexpressing ErbB-1/2, ErbB-1/4, and ErbB-2/4. Intermediate tumorigenicity was observed for tumors overexpressing ErbB-2 and ErbB-2/3, whereas cells overexpressing ErbB-1 and ErbB-3/4 showed delayed tumorigenicity. These in vivo results revealed some discrepancies with a previous study by Cohen et al. using a similar cell model. This study reported that cells overexpressing any single ErbB receptors were unable to induce tumors, ErbB-1/4 was not tumorigenic, and coexpression of ErbB-1 with ErbB-3 or ErbB-4 reduced tumorigenesis, whereas ErbB-2/4 and ErbB-3/4 were not reported. Unlike our study using a polyclonal population, Cohen et al.(36) used single selected NIH-3T3 cell clones, and their in vivo results were determined at 4 weeks after injection, which may exclude detection of tumors that appear at a later stage, e.g., ErbB-1/3.

Analysis of gene expression profiles revealed distinct patterns depending on the expression of single or paired combinations of ErbB receptors. The numbers of genes that are affected are significantly higher in cells overexpressing paired receptor combinations compared with cells expressing a single receptor, supporting earlier studies on the potency of ErbB heterodimers in eliciting biological functions (reviewed in Ref. 1). The number of genes affected by the kinase-deficient ErbB-3 receptor was unexpectedly high because this receptor is known to be kinase deficient. ErbB-3, however, has been shown to affect gene expression via protein-protein interaction with cellular factors. Yoo et al.(34) have reported that the cytoplasmic domain of ErbB-3 interacts with p23, the human homologue of the mouse transplantation antigen p198; this interaction is suggested to play a role in the regulation of cell proliferation. Moreover, ErbB-3 has been shown to interact with the regulatory subunit of phosphoinositide 3-kinase (37). Our model does not exclude, however, the possibility of ErbB-3 interaction with the reported very low levels of ErbB-1 that may be expressed endogenously in the NIH-3T3 cell model (35). The in vitro array analysis revealed several points of potential interaction between ErbB receptors and loss of suppression of cell proliferation, including up-regulation of cyclins D1 and D2 (observed in cells overexpressing ErbB-1, ErbB-2, ErbB-2/4, and ErbB-2/3) and down-regulation of p21 (observed in cells overexpressing ErbB-1/3). Cyclin Ds are important downstream target for ErbB receptors (38), and an array study by Sweeney et al.(39) reported cyclin D1, cyclin D3, and E2F to be stimulated by ErbB ligands in mammary breast cancer cells.

In contrast to cells maintained in vitro, our array analysis comparing tumor tissues overexpressing the various ErbB receptors argues for the implication of additional host mechanisms that operate in vivo as revealed by the wide divergence of gene expression profiles between in vitro and in vivo conditions. As noted in the results, cells expressing control retroviral particles are not tumorigenic, thus we focused our in vivo array comparisons between tumors overexpressing paired combinations of receptors and tumors overexpressing a single ErbB receptor and related it to tumor growth phenotype. Also, tumor growth rates in vivo were dependent on the ErbB receptor types that were overexpressed. This growth pattern created, however, several variables that should account for comparative variation in gene expression between the in vitro and in vivo conditions. Firstly, tissue specimens used for the array analysis were taken when tumors reached ∼0.8 cm3 (with the exception of ErbB-3, ErbB-4, and ErbB-1/3 tumors which formed smaller tumors or nodules), and hence the time delay, because of different growth rates, can account for differences in gene regulation among the various receptors. Nevertheless, this is a biological phenomenon that has clinical relevance when considering matched cells that differ only in ErbB expression status. Secondly, tumor cell heterogeneity is another variable. In vivo, a variety of cell types populate the stroma compartment, including inflammatory cells, fibroblasts, myofibroblasts, and endothelial cells. This cell heterogeneity is clinically relevant as it occurs in all solid tumors, and it is a hallmark problem in clinical oncology as it creates various types of cell subpopulations that can affect tumor progression as well as response to therapeutics (40). We confirmed the expression changes for selected targets, e.g., TSP1, in multiple tissue samples taken from the same tumor and from tumors isolated from independent animals. The results were highly reproducible, making it likely that the contaminating host cells are randomly represented in the samples analyzed. Also, immunohistochemical analysis was performed on duplicate specimens used for the array, and the expression of the various ErbB receptors were observed in >80% of tissue sections, thus supporting homogeneity between tissue specimens in regard to receptor overexpression during growth in vivo. We conclude that markers we identified by the array analysis are most likely representative of molecular features associated with ErbB expression as also revealed for TSP1 using human cancer cell models overexpressing ErbB receptors.

Overall, our in vivo array analysis highlights the tissue microenvironment as a key player in tumor aggressiveness based on the many genes identified that are involved in the regulation of the local tumor microenvironment. These include several markers that converge, directly or indirectly, on the regulation of the ECM, cell-cell structure, angiogenesis, and inflammatory cytokines (e.g., integrins, TSPs, TIMPs, thrombin receptor, tenascins, prostaglandins, collagens, proteoglycans, and several multiadhesive matrix proteins such as laminins); most of these markers are contributed primarily by the host (41, 42, 43, 44, 45), and aberrant expression of some was reported in human tumor subclasses overexpressing the ErbB-2 receptor (46, 47).

Among the selected tumor microenvironment markers we validated, we noticed clear differences between cells and tissues overexpressing specific members of the ErbB family, e.g., TSP1, tenascin XB, troponin T3, TIMP-3, vinculin, and meltrin-β (data not shown). We then focused our validation study on TSP1, an endogenous inhibitor of endothelial cell proliferation and angiogenesis (48). Generally, TSP1 was down-regulated in tumors overexpressing ErbB combinations compared with cells expressing single receptors. Interestingly, TSP1 was also reduced in tumor tissues from NIH-3T3 as well as human xenografts expressing ErbB combinations. The differential profile of TSP1 inhibition between in vitro and in vivo would suggest that TSP1 is regulated by other types of interactions operative in vivo. As noted in Tables 2,3,4, many regulators of coagulation and inflammation were affected by ErbB expression. In addition to their role in homeostasis and thrombosis, deregulation of coagulation factors such as overexpression of tissue factor, the principle initiator of coagulation in vivo, has been shown to inhibit the transcription of TSP and induce mitogenic activity of endothelial cells (49), supporting the presence of multiple cross-talks between hematopoiesis factors and angiogenesis. TSP1 is a member of a protein family widely present in the ECM and plays a key role in the organization of tissue microenvironment (50) and angiogenesis (51, 52). TSP1 has a multimodular structure that includes three epidermal growth factor-like type II repeats and has been shown to synergize with epidermal growth factor to promote cell proliferation (53). In relation to ErbB expression in the clinical setting, several studies have reported that diminished TSP1 expression or alteration of its spatial distribution correlates with aggressive tumor growth and poor survival (51). However, in our ErbB tumor models, we cannot attribute aggressive tumor growth or high tumor vascularization solely to changes in TSP1 expression because many other important regulators of ECM and angiogenesis were also affected by ErbBs. It is therefore challenging to dissect the molecular mechanisms by which ErbB signaling affects TSP1 regulation, as well as the contribution of the other markers we identified by the genomic study to tumorigenesis.

In summary, the important aspect of the comprehensive genomic study presented here is to dissect ErbB signaling diversity associated with all possible ErbB receptor dimer combinations, with a specific focus on markers of the tumor microenvironment. We conclude that the multiplicity of autocrine and paracrine loops involving ErbB receptors have a great impact on host tumor tissue microenvironment, which, in turn, can play a key role in determining the in vivo growth and invasive phenotypes. It is intriguing that most genes involved in host-tumor interaction are contributed by the host and presents the tumor ecosystem as an alternative target for cancer therapeutics. The coordinated expression of the candidate genes involved in the regulation of the tumor microenvironment suggests cofunctionality, although how ErbB signaling regulates these markers and how they cooperate in tumor progression remains to be tested at the molecular level.

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.

1

This work was supported by the Canadian Breast Cancer Research Initiative of the National Cancer Institute of Canada and, in part, by a strategic grant from the Cancer Research Society, Inc. M. A. A-J. is supported by a Chercheur National Award from the Fonds de la Recherché en Santé du Québec.

3

The abbreviations used are: ErbB, tyrosine kinase receptors of the ErbB family; GFP, green fluorescent protein; CD31, Mr 130,000 integral membrane protein also known as platelet endothelial adhesion molecule (PECAM1); EST, expressed sequence tag; GAPDH, glyceraldehydes-3-phosphate dehydrogenase; ECM, extracellular matrix; RT-PCR, reverse transcription-PCR; TIMP, tissue inhibitor of metalloproteinase; TSP, thrombospondin.

4

Internet address: http://www.medicine.mcgill.ca/pharma/alaouijamalilab.

Fig. 1.

Characterization of ErbB receptor expression in NIH-3T3 cells overexpressing ErbB receptor combinations. Immunofluorescence analysis was performed on subconfluent NIH-3T3 cells using antibodies for the respective receptors and Texas Red secondary antibodies as described in “Materials and Methods.” Immunofluorescence of cells transduced with control (GFP) retrovirus (AP2; A), single ErbB receptors (B), or paired ErbB receptor combinations (C).

Fig. 1.

Characterization of ErbB receptor expression in NIH-3T3 cells overexpressing ErbB receptor combinations. Immunofluorescence analysis was performed on subconfluent NIH-3T3 cells using antibodies for the respective receptors and Texas Red secondary antibodies as described in “Materials and Methods.” Immunofluorescence of cells transduced with control (GFP) retrovirus (AP2; A), single ErbB receptors (B), or paired ErbB receptor combinations (C).

Close modal
Fig. 2.

Effect of ErbB receptor expression on tumor formation and tumor growth. Subconfluent cells were transplanted s.c. in the flanks of nude mice. Tumor growth was monitored daily, and tumor size was calculated as described in “Materials and Methods.” Each time point corresponds to the average of 12 (AP2) and 8 (ErbB) animals ± SE.

Fig. 2.

Effect of ErbB receptor expression on tumor formation and tumor growth. Subconfluent cells were transplanted s.c. in the flanks of nude mice. Tumor growth was monitored daily, and tumor size was calculated as described in “Materials and Methods.” Each time point corresponds to the average of 12 (AP2) and 8 (ErbB) animals ± SE.

Close modal
Fig. 3.

RT-PCR validation of ErbB receptor expression. Total RNA from cells and tumor tissues was amplified using gene specific primers, and RT-PCR was carried out as described in “Materials and Methods.” gapdh expression was used to optimize PCR cycles and RNA concentrations. Products were run on 1% agarose gels and visualized with ethidium bromide staining. Each photograph is a representative RT-PCR experiment from at least four independent experiments carried out on independent cells and tumors; each lane corresponding to single or paired combinations of ErbB receptors overexpressed in cells maintained in vitro (A) or in vivo (B).

Fig. 3.

RT-PCR validation of ErbB receptor expression. Total RNA from cells and tumor tissues was amplified using gene specific primers, and RT-PCR was carried out as described in “Materials and Methods.” gapdh expression was used to optimize PCR cycles and RNA concentrations. Products were run on 1% agarose gels and visualized with ethidium bromide staining. Each photograph is a representative RT-PCR experiment from at least four independent experiments carried out on independent cells and tumors; each lane corresponding to single or paired combinations of ErbB receptors overexpressed in cells maintained in vitro (A) or in vivo (B).

Close modal
Fig. 4.

In vitro transcriptional gene profile analysis using microarray. Cells at 80% confluence were harvested, total RNA was extracted, reverse transcribed, and hybridized to a microarray as described in “Materials and Methods.” Comparisons of cell lines overexpressing single or paired combinations of ErbB receptors to control cells transduced with the control retroviral particles (AP2) were analyzed for genes that were differentially expressed. Known genes in which the fold expression ratios that were ≥2.5-fold were organized into Venn distributions such that the independent effects of heterodimer expression could be identified from homodimeric forms. Circles represent all genes of pairwise comparisons, whereas intersecting areas represent genes common to another pairwise comparison. Genes of interest are described in Tables 2 and 3.

Fig. 4.

In vitro transcriptional gene profile analysis using microarray. Cells at 80% confluence were harvested, total RNA was extracted, reverse transcribed, and hybridized to a microarray as described in “Materials and Methods.” Comparisons of cell lines overexpressing single or paired combinations of ErbB receptors to control cells transduced with the control retroviral particles (AP2) were analyzed for genes that were differentially expressed. Known genes in which the fold expression ratios that were ≥2.5-fold were organized into Venn distributions such that the independent effects of heterodimer expression could be identified from homodimeric forms. Circles represent all genes of pairwise comparisons, whereas intersecting areas represent genes common to another pairwise comparison. Genes of interest are described in Tables 2 and 3.

Close modal
Fig. 5.

In vivo transcriptional gene profile analysis using microarray. Tumors with approximate size of 0.8 cm3 (with the exception of the ErbB-3, ErbB-4, and ErbB-1/3 tumors, which were taken at a later time points) were removed, snap frozen, and total RNA was extracted, reverse transcribed, and hybridized to a microarray as described in “Materials and Methods.” Overexpression of paired combinations of ErbB receptors was dissected by pairwise comparison with their respective receptor components to reveal genes that are putatively responsible for the phenotype in vivo. Only known genes in which the fold expression ratios were ≥2.5 were included. Genes of interest are described in Table 4.

Fig. 5.

In vivo transcriptional gene profile analysis using microarray. Tumors with approximate size of 0.8 cm3 (with the exception of the ErbB-3, ErbB-4, and ErbB-1/3 tumors, which were taken at a later time points) were removed, snap frozen, and total RNA was extracted, reverse transcribed, and hybridized to a microarray as described in “Materials and Methods.” Overexpression of paired combinations of ErbB receptors was dissected by pairwise comparison with their respective receptor components to reveal genes that are putatively responsible for the phenotype in vivo. Only known genes in which the fold expression ratios were ≥2.5 were included. Genes of interest are described in Table 4.

Close modal
Fig. 6.

A, analysis of tsp1 mRNA expression in the NIH-3T3 cells overexpressing ErbB receptors. Total RNA from subconfluent cells was prepared for Northern blot analysis as described in “Materials and Methods.” A representative autoradiogram is shown where each lane corresponds to cells transduced with single or paired combinations of ErbB receptors. gapdh mRNA was used as an internal control for loading. The graph represents the average ratios of tsp1/gapdh mRNA (±SE) from two independent experiments. B, immunohistochemical analysis of TSP1 expression in tumors induced by specific ErbB-overexpressing cells. NIH-3T3-ErbB tumors were induced s.c., snap frozen, and sections were stained with H&E, or immunolabeled for TSP1 or CD31. For negative controls (−), the primary antibodies were omitted. Scale bar: 400 μm.

Fig. 6.

A, analysis of tsp1 mRNA expression in the NIH-3T3 cells overexpressing ErbB receptors. Total RNA from subconfluent cells was prepared for Northern blot analysis as described in “Materials and Methods.” A representative autoradiogram is shown where each lane corresponds to cells transduced with single or paired combinations of ErbB receptors. gapdh mRNA was used as an internal control for loading. The graph represents the average ratios of tsp1/gapdh mRNA (±SE) from two independent experiments. B, immunohistochemical analysis of TSP1 expression in tumors induced by specific ErbB-overexpressing cells. NIH-3T3-ErbB tumors were induced s.c., snap frozen, and sections were stained with H&E, or immunolabeled for TSP1 or CD31. For negative controls (−), the primary antibodies were omitted. Scale bar: 400 μm.

Close modal
Fig. 7.

A, analysis of tsp1 mRNA expression in human breast cancer cells. The human breast cell lines MCF-7, MDA-435, and BT-20 were transduced with control retroviral particles (AP2) or viral particles containing the erbB2 gene. After cell characterization, total RNA from subconfluent cells was prepared for Northern blot analysis as described in “Materials and Methods.” A representative autoradiogram is shown where each lane corresponds to cells transduced with single or paired combinations of ErbB receptors. gapdh mRNA was used as a control for loading. The graph represents the average ratios of tsp1/gapdh mRNA (±SE) from two independent experiments. B, immunohistochemical analysis of TSP1 expression in human tumor xenografts. MCF7 and BT20 cells were implanted into the mammary fat pad of nude mice. Tumors reached a size of 0.8–1 cm3, tumors were collected at ∼0.8–1 cm3 in size, and sections were stained with H&E or immunolabeled for ErbB-2 or TSP1. For negative controls (−), the primary antibodies were omitted. Scale bar: 400 μm.

Fig. 7.

A, analysis of tsp1 mRNA expression in human breast cancer cells. The human breast cell lines MCF-7, MDA-435, and BT-20 were transduced with control retroviral particles (AP2) or viral particles containing the erbB2 gene. After cell characterization, total RNA from subconfluent cells was prepared for Northern blot analysis as described in “Materials and Methods.” A representative autoradiogram is shown where each lane corresponds to cells transduced with single or paired combinations of ErbB receptors. gapdh mRNA was used as a control for loading. The graph represents the average ratios of tsp1/gapdh mRNA (±SE) from two independent experiments. B, immunohistochemical analysis of TSP1 expression in human tumor xenografts. MCF7 and BT20 cells were implanted into the mammary fat pad of nude mice. Tumors reached a size of 0.8–1 cm3, tumors were collected at ∼0.8–1 cm3 in size, and sections were stained with H&E or immunolabeled for ErbB-2 or TSP1. For negative controls (−), the primary antibodies were omitted. Scale bar: 400 μm.

Close modal
Table 1

Primer sequence used for RT-PCR

GenePrimer sequencesProduct’s size (bp)
ErbB-1 AGAGGAGAACTGCCAGAAACTG Sense 550 
 AGGAGGAGTATGTGTGAAGGAGTC Antisense  
ErbB-2 GAGATCTTGAAAGGAGGGGTCT Sense 484 
 CGTCCGTAGAAAGGTAGTTGTAGG Antisense  
ErbB-3 GTGGATGGCCCTTGAGAGTA Sense 499 
 ACGTGGCCGATTAAGTGTTC Antisense  
ErbB-4 GGGAACTGATGACCTTTGGA Sense 500 
 CACTCCTTGTTCAGCAGCAA Antisense  
GAPDH GAAGGCCATGCCAGTGAGCT Sense 125 
 CCGGGAAACTGTGGCGTGAT Antisense  
GenePrimer sequencesProduct’s size (bp)
ErbB-1 AGAGGAGAACTGCCAGAAACTG Sense 550 
 AGGAGGAGTATGTGTGAAGGAGTC Antisense  
ErbB-2 GAGATCTTGAAAGGAGGGGTCT Sense 484 
 CGTCCGTAGAAAGGTAGTTGTAGG Antisense  
ErbB-3 GTGGATGGCCCTTGAGAGTA Sense 499 
 ACGTGGCCGATTAAGTGTTC Antisense  
ErbB-4 GGGAACTGATGACCTTTGGA Sense 500 
 CACTCCTTGTTCAGCAGCAA Antisense  
GAPDH GAAGGCCATGCCAGTGAGCT Sense 125 
 CCGGGAAACTGTGGCGTGAT Antisense  
Table 2

List of common genes differentially expressed in cells overexpressing single ErbB receptors and maintained in vitro in two-dimensional cell culture condition

Gene descriptionAccession no.ErbB-1ErbB-2ErbB-3ErbB-4
Receptors/signal transduction      
 IGF binding protein 4 X76066 −3.5 −5.1 nsa −17.6 
 R-ras M21019 −2.6 ns −4.3 ns 
 Platelet derived growth factor receptor M57683 −3.4 ns −2.8 ns 
 Suppressor of cytokine signaling-3 U88328 −5.1 ns −5.1 −3.1 
 Phosphatase and tensin homologue U92437 ns ns −2.8 ns 
 Oncostatin receptor AB015978 −3.2 ns ns ns 
Transcription/nuclear factors      
 Max-interacting protein 1 L38822 −3.4 −4.2 −3.1 −2.9 
 Early growth response 1 M28845 ns ns −6.1 −5.8 
 High mobility group protein 1-C X99915 5.5 8.6 3.4 2.8 
 DNA damage-inducible transcript X67083 −3.9 ns ns −3.0 
 Mutated in multiple advanced cancers U92437 −3.1 −3.0 −2.8 ns 
Cell cycle      
 Cyclin D1 AI849928 3.3 3.9 2.6 ns 
 Cyclin G2 U95826 −2.6 ns −3.5 ns 
Cytoskeleton      
 Actin, β M12481 −5.8 −4.2 −4.0 ns 
 Actin, α 2 X13297 −5.0 ns −4.5 −7.6 
ECM and associated proteins/angiogenesis      
 TSP1 M62470 −3.3 −6.7 −3.6 ns 
 Matrilin 2 U69262 −3.2 ns −3.9 ns 
 Microfibrillar-associated protein 5 AW121179 ns −4.5 ns −3.5 
 Hyaluronan synthase 2 U52524 ns −5.1 −5.1 ns 
Protein degradation regulation      
 Serine protease inhibitor 2-1 M64085 −9.6 −6.2 −7.9 ns 
 Serine protease inhibitor 2-2 M64086 −6.0 −4.7 −9.3 −3.1 
 Serine protease inhibitor, Kazal type 4 Y11505 ns −2.8 −16.9 −32.1 
 Protease-nexin 1 X70296 −3.1 ns ns −6.2 
Hematopoiesis/inflammation      
 Stromal cell-derived factor 1 L12030 −7.4 −7.5 −7.5 −9.3 
 Haptoglobin M96827 −5.6 −11.9 −3.7 ns 
 Coagulation factor III M26071 −4.5 ns −4.2 ns 
ESTs and genes with unknown function      
 Total number of genes up- or down-regulated  27 23 40 50 
Gene descriptionAccession no.ErbB-1ErbB-2ErbB-3ErbB-4
Receptors/signal transduction      
 IGF binding protein 4 X76066 −3.5 −5.1 nsa −17.6 
 R-ras M21019 −2.6 ns −4.3 ns 
 Platelet derived growth factor receptor M57683 −3.4 ns −2.8 ns 
 Suppressor of cytokine signaling-3 U88328 −5.1 ns −5.1 −3.1 
 Phosphatase and tensin homologue U92437 ns ns −2.8 ns 
 Oncostatin receptor AB015978 −3.2 ns ns ns 
Transcription/nuclear factors      
 Max-interacting protein 1 L38822 −3.4 −4.2 −3.1 −2.9 
 Early growth response 1 M28845 ns ns −6.1 −5.8 
 High mobility group protein 1-C X99915 5.5 8.6 3.4 2.8 
 DNA damage-inducible transcript X67083 −3.9 ns ns −3.0 
 Mutated in multiple advanced cancers U92437 −3.1 −3.0 −2.8 ns 
Cell cycle      
 Cyclin D1 AI849928 3.3 3.9 2.6 ns 
 Cyclin G2 U95826 −2.6 ns −3.5 ns 
Cytoskeleton      
 Actin, β M12481 −5.8 −4.2 −4.0 ns 
 Actin, α 2 X13297 −5.0 ns −4.5 −7.6 
ECM and associated proteins/angiogenesis      
 TSP1 M62470 −3.3 −6.7 −3.6 ns 
 Matrilin 2 U69262 −3.2 ns −3.9 ns 
 Microfibrillar-associated protein 5 AW121179 ns −4.5 ns −3.5 
 Hyaluronan synthase 2 U52524 ns −5.1 −5.1 ns 
Protein degradation regulation      
 Serine protease inhibitor 2-1 M64085 −9.6 −6.2 −7.9 ns 
 Serine protease inhibitor 2-2 M64086 −6.0 −4.7 −9.3 −3.1 
 Serine protease inhibitor, Kazal type 4 Y11505 ns −2.8 −16.9 −32.1 
 Protease-nexin 1 X70296 −3.1 ns ns −6.2 
Hematopoiesis/inflammation      
 Stromal cell-derived factor 1 L12030 −7.4 −7.5 −7.5 −9.3 
 Haptoglobin M96827 −5.6 −11.9 −3.7 ns 
 Coagulation factor III M26071 −4.5 ns −4.2 ns 
ESTs and genes with unknown function      
 Total number of genes up- or down-regulated  27 23 40 50 
a

ns, not significant.

Table 3

List of unique genes differentially expressed in cells overexpressing specific paired combinations of ErbB receptors and maintained in vitro in two-dimensional cell culture condition

Gene descriptionAccession no.ErbB-1/2ErbB-1/3ErbB-1/4ErbB-2/3ErbB-2/4ErbB-3/4
Receptors/signal transduction        
 IGF binding protein 4 X76066 nsa ns −12.5 ns −15.0 −17.6 
 Insulin-like growth factor I receptor AF056187 ns ns ns −2.9 ns ns 
 EGF receptor pathway substrate 15 AW124983 −3.9 ns ns ns ns ns 
 Platelet-derived growth factor receptor M57683 ns −3.2 ns ns ns ns 
 Growth hormone receptor U15012 ns −2.6 ns −4.0 −3.1 ns 
 Fibroblast growth factor 7 Z22703 −6.7 ns ns ns −2.9 ns 
 CD47 AB012693 −2.6 −3.3 ns ns ns ns 
 Suppressor of cytokine signaling-3 U88328 ns −5.1 ns ns ns −6.5 
 Ephrin B2 U30244 ns ns 3.0 ns ns ns 
 Oncostatin receptor AB015978 ns −4.3 ns ns −3.3 ns 
 RAS p21 protein activator 3 U20238 ns ns ns ns −3.0 ns 
 R-ras M21019 ns −2.9 ns ns ns ns 
 MAP kinase kinase 3 X93150 ns ns ns ns 3.3 ns 
 Serine/threonine kinase 2 AA822531 ns ns ns ns ns −3.6 
 Protein-serine/threonine kinase L41495 ns ns ns ns ns 4.6 
 Secreted frizzled-related sequence U88566 ns ns −5.9 ns −5.9 −5.9 
Transcription/nuclear factors        
 Early growth response 1 M28845 −6.3 ns ns ns ns −9.4 
 c-fos-induced growth factor X99572 ns ns ns ns −4.0 ns 
 Nuclear factor I/B Y07686 −5.1 ns ns ns ns ns 
 Topoisomerase I X70956 −10.4 ns ns ns ns ns 
 Mutated in multiple advanced cancers protein U92437 −10.0 −3.1 −3.8 −3.8 −3.6 ns 
 Pre B-cell leukemia transcription factor 1 AW124932 −12.9 ns ns ns ns ns 
 Growth factor-inducible immediate early X61940 ns ns ns ns ns 3.2 
 DNA damage-inducible transcript 3 X67083 ns ns −5.7 ns ns −3.4 
 Matrin 3 AI835367 −5.2 ns ns ns ns −2.9 
 High mobility group protein I-C X99915 4.1 3.4 3.6 4.2 6.8 ns 
 Cell cycle/apoptosis        
 Cyclin D1 AI849928 ns 2.7 ns 2.8 3.0 ns 
 Cyclin D2 M83749 ns ns ns 4.1 ns ns 
 Cyclin G2 U95826 ns −3.1 ns ns ns ns 
 P21 U09507 ns −2.8 ns ns ns ns 
 TIS21 M64292 ns ns ns ns −5.0 ns 
 Growth arrest specific 5 AI849615 ns −3.3 ns ns ns ns 
 GADD45 protein U00937 ns ns −3.1 ns ns ns 
 Max interacting protein 1 L38822 −4.3 −4.2 ns ns ns −2.8 
 Caspase 11 Y13089 ns ns −2.6 ns ns −4.4 
Cytoskeleton organization/cell-cell interaction        
 Actin, β M12481 −2.8 −4.3 ns −3.5 ns ns 
 Actin, α M15501 ns 2.9 ns ns ns ns 
 Actin, α 2 X13297 ns −4.7 −43.4 ns ns −2.9 
 Actin-crosslinking protein 7 AI843799 −2.9 ns ns ns ns − 
 Capping protein α 2 U16741 −3.0 ns ns ns ns ns 
 Laminin, α 4 U69176 ns −2.9 ns ns ns ns 
 Microfibrillar-associated protein 5 AW121179 ns ns ns ns −6.0 ns 
 Cullin 1 AI849838 −3.2 ns ns ns ns ns 
 Cullin 3 AI840051 −7.4 ns ns ns ns ns 
 Protocadherin 13 AI854522 −2.7 ns ns ns ns ns 
 Cadherin-related neuronal receptor 7 AB008182 ns ns ns ns ns 3.8 
 Connexin 43 M63801 ns ns ns ns 3.0 ns 
ECM, cell adhesion, and angiogenesis        
 Vinculin AI462105 −3.6 ns ns ns ns ns 
 Plasminogen activator, tissue J03520 −3.0 ns ns ns ns ns 
 Fibulin 2 X75285 ns ns ns ns −5.1 ns 
 Glypican 1 AI852765 ns ns 2.6 ns 3.2 ns 
 Meltrin γ U41765 ns ns ns ns ns −2.8 
 Stromal antigen 2 AJ002636 ns ns ns ns ns −3.8 
 Talin X56123 ns ns ns ns ns −5.2 
 Connective tissue growth factor-related protein WISP-2 AF100778 ns ns ns ns −5.4 ns 
 TSP1 M62470 −10.2 −9.5 −6.1 −3.3 −26.7 ns 
 TSP2 L07803 −3.1 ns ns ns ns ns 
 Tenascin C X56304 ns ns −3.1 ns −3.1 −3.1 
 Integrin β 4 binding protein Y11460 4.1 2.6 ns ns ns ns 
 Nidogen L17324 ns ns ns −2.9 −3.6 ns 
 Hyaluronan synthase 2 U52524 ns ns ns −5.1 ns ns 
 Chondroitin sulfate proteoglycan D45889 −9.4 ns ns ns ns ns 
 Matrilin 2 U69262 ns −5.1 ns ns ns ns 
 GPI Anchor attachment protein 1 AB002136 ns ns ns ns 3.9 ns 
 Procollagen, type V, α 2 AA032310 −5.3 ns −4.1 ns ns ns 
 Procollagen, type VI, α 2 Z18272 −2.7 ns ns ns ns ns 
 Procollagen, type VI, α 1 AV010209 ns −2.6 ns −2.7 ns ns 
 Procollagen, type III, α 1 AA655199 ns − 6.8 ns 6.9 ns 
Protein degradation regulation        
 TIMP-3 U26437 ns ns ns ns −2.6 ns 
 Serine protease inhibitor, Kazal type 4 Y11505 ns ns ns −6.0 −2.7 −29.2 
 Protease-nexin 1 X70296 ns ns −6.5 ns −13.5 ns 
 Ubiquitin specific protease 14 AW060186 ns ns − 2.6 ns ns 
 Serine protease inhibitor 2-1 M64085 −4.9 −9.6 ns −9.6 ns ns 
 Serine protease inhibitor 2-2 M64086 −4.4 −14.9 ns ns ns −8.0 
Hematopoiesis/inflammation        
 Stromal cell-derived factor 1 L12030 −7.5 −7.5 −9.3 −7.5 −9.3 −9.3 
 Thrombin receptor AW123850 −3.2 ns ns ns ns ns 
 Plasminogen activator, tissue J03520 −3.0 ns ns ns ns ns 
 Coagulation factor III M26071 ns −5.7 −2.9 ns ns ns 
 Heme oxygenase 1 X56824 ns ns ns ns ns −4.4 
 Kallikrein binding protein X61597 ns ns ns ns −5.0 −3.6 
 Prothymosin β 4 U38967 ns ns ns ns 3.6 ns 
 Haptoglobin M96827 −3.4 −15.9 ns ns ns ns 
ESTs and genes with unknown function        
 Number of genes up- or down-regulated  90 37 16 33 44 77 
Gene descriptionAccession no.ErbB-1/2ErbB-1/3ErbB-1/4ErbB-2/3ErbB-2/4ErbB-3/4
Receptors/signal transduction        
 IGF binding protein 4 X76066 nsa ns −12.5 ns −15.0 −17.6 
 Insulin-like growth factor I receptor AF056187 ns ns ns −2.9 ns ns 
 EGF receptor pathway substrate 15 AW124983 −3.9 ns ns ns ns ns 
 Platelet-derived growth factor receptor M57683 ns −3.2 ns ns ns ns 
 Growth hormone receptor U15012 ns −2.6 ns −4.0 −3.1 ns 
 Fibroblast growth factor 7 Z22703 −6.7 ns ns ns −2.9 ns 
 CD47 AB012693 −2.6 −3.3 ns ns ns ns 
 Suppressor of cytokine signaling-3 U88328 ns −5.1 ns ns ns −6.5 
 Ephrin B2 U30244 ns ns 3.0 ns ns ns 
 Oncostatin receptor AB015978 ns −4.3 ns ns −3.3 ns 
 RAS p21 protein activator 3 U20238 ns ns ns ns −3.0 ns 
 R-ras M21019 ns −2.9 ns ns ns ns 
 MAP kinase kinase 3 X93150 ns ns ns ns 3.3 ns 
 Serine/threonine kinase 2 AA822531 ns ns ns ns ns −3.6 
 Protein-serine/threonine kinase L41495 ns ns ns ns ns 4.6 
 Secreted frizzled-related sequence U88566 ns ns −5.9 ns −5.9 −5.9 
Transcription/nuclear factors        
 Early growth response 1 M28845 −6.3 ns ns ns ns −9.4 
 c-fos-induced growth factor X99572 ns ns ns ns −4.0 ns 
 Nuclear factor I/B Y07686 −5.1 ns ns ns ns ns 
 Topoisomerase I X70956 −10.4 ns ns ns ns ns 
 Mutated in multiple advanced cancers protein U92437 −10.0 −3.1 −3.8 −3.8 −3.6 ns 
 Pre B-cell leukemia transcription factor 1 AW124932 −12.9 ns ns ns ns ns 
 Growth factor-inducible immediate early X61940 ns ns ns ns ns 3.2 
 DNA damage-inducible transcript 3 X67083 ns ns −5.7 ns ns −3.4 
 Matrin 3 AI835367 −5.2 ns ns ns ns −2.9 
 High mobility group protein I-C X99915 4.1 3.4 3.6 4.2 6.8 ns 
 Cell cycle/apoptosis        
 Cyclin D1 AI849928 ns 2.7 ns 2.8 3.0 ns 
 Cyclin D2 M83749 ns ns ns 4.1 ns ns 
 Cyclin G2 U95826 ns −3.1 ns ns ns ns 
 P21 U09507 ns −2.8 ns ns ns ns 
 TIS21 M64292 ns ns ns ns −5.0 ns 
 Growth arrest specific 5 AI849615 ns −3.3 ns ns ns ns 
 GADD45 protein U00937 ns ns −3.1 ns ns ns 
 Max interacting protein 1 L38822 −4.3 −4.2 ns ns ns −2.8 
 Caspase 11 Y13089 ns ns −2.6 ns ns −4.4 
Cytoskeleton organization/cell-cell interaction        
 Actin, β M12481 −2.8 −4.3 ns −3.5 ns ns 
 Actin, α M15501 ns 2.9 ns ns ns ns 
 Actin, α 2 X13297 ns −4.7 −43.4 ns ns −2.9 
 Actin-crosslinking protein 7 AI843799 −2.9 ns ns ns ns − 
 Capping protein α 2 U16741 −3.0 ns ns ns ns ns 
 Laminin, α 4 U69176 ns −2.9 ns ns ns ns 
 Microfibrillar-associated protein 5 AW121179 ns ns ns ns −6.0 ns 
 Cullin 1 AI849838 −3.2 ns ns ns ns ns 
 Cullin 3 AI840051 −7.4 ns ns ns ns ns 
 Protocadherin 13 AI854522 −2.7 ns ns ns ns ns 
 Cadherin-related neuronal receptor 7 AB008182 ns ns ns ns ns 3.8 
 Connexin 43 M63801 ns ns ns ns 3.0 ns 
ECM, cell adhesion, and angiogenesis        
 Vinculin AI462105 −3.6 ns ns ns ns ns 
 Plasminogen activator, tissue J03520 −3.0 ns ns ns ns ns 
 Fibulin 2 X75285 ns ns ns ns −5.1 ns 
 Glypican 1 AI852765 ns ns 2.6 ns 3.2 ns 
 Meltrin γ U41765 ns ns ns ns ns −2.8 
 Stromal antigen 2 AJ002636 ns ns ns ns ns −3.8 
 Talin X56123 ns ns ns ns ns −5.2 
 Connective tissue growth factor-related protein WISP-2 AF100778 ns ns ns ns −5.4 ns 
 TSP1 M62470 −10.2 −9.5 −6.1 −3.3 −26.7 ns 
 TSP2 L07803 −3.1 ns ns ns ns ns 
 Tenascin C X56304 ns ns −3.1 ns −3.1 −3.1 
 Integrin β 4 binding protein Y11460 4.1 2.6 ns ns ns ns 
 Nidogen L17324 ns ns ns −2.9 −3.6 ns 
 Hyaluronan synthase 2 U52524 ns ns ns −5.1 ns ns 
 Chondroitin sulfate proteoglycan D45889 −9.4 ns ns ns ns ns 
 Matrilin 2 U69262 ns −5.1 ns ns ns ns 
 GPI Anchor attachment protein 1 AB002136 ns ns ns ns 3.9 ns 
 Procollagen, type V, α 2 AA032310 −5.3 ns −4.1 ns ns ns 
 Procollagen, type VI, α 2 Z18272 −2.7 ns ns ns ns ns 
 Procollagen, type VI, α 1 AV010209 ns −2.6 ns −2.7 ns ns 
 Procollagen, type III, α 1 AA655199 ns − 6.8 ns 6.9 ns 
Protein degradation regulation        
 TIMP-3 U26437 ns ns ns ns −2.6 ns 
 Serine protease inhibitor, Kazal type 4 Y11505 ns ns ns −6.0 −2.7 −29.2 
 Protease-nexin 1 X70296 ns ns −6.5 ns −13.5 ns 
 Ubiquitin specific protease 14 AW060186 ns ns − 2.6 ns ns 
 Serine protease inhibitor 2-1 M64085 −4.9 −9.6 ns −9.6 ns ns 
 Serine protease inhibitor 2-2 M64086 −4.4 −14.9 ns ns ns −8.0 
Hematopoiesis/inflammation        
 Stromal cell-derived factor 1 L12030 −7.5 −7.5 −9.3 −7.5 −9.3 −9.3 
 Thrombin receptor AW123850 −3.2 ns ns ns ns ns 
 Plasminogen activator, tissue J03520 −3.0 ns ns ns ns ns 
 Coagulation factor III M26071 ns −5.7 −2.9 ns ns ns 
 Heme oxygenase 1 X56824 ns ns ns ns ns −4.4 
 Kallikrein binding protein X61597 ns ns ns ns −5.0 −3.6 
 Prothymosin β 4 U38967 ns ns ns ns 3.6 ns 
 Haptoglobin M96827 −3.4 −15.9 ns ns ns ns 
ESTs and genes with unknown function        
 Number of genes up- or down-regulated  90 37 16 33 44 77 
a

ns, not significant.

Table 4

List of unique genes differentially expressed in tumors induced by cells overexpressing specific paired combinations of ErbB receptors compared with tumors overexpressing single receptors

Gene descriptionAccession no.ErbB-1/2ErbB-1/3ErbB-1/4ErbB-2/3ErbB-2/4ErbB-3/4
Receptors/signal transduction        
 Insulin-like growth factor 2 X71922 nsa 16.29s 117.2 88.93 ns ns 
 Insulin-like growth factor binding protein 5 L12447 ns 4.61 117.2 88.93 ns ns 
 Growth factor-inducible protein M59821 ns −2.93 ns ns ns ns 
 Heparin binding epidermal growth factor-like growth factor L07264 ns ns ns ns ns 4.7 
 Polo-like kinase U73170 ns −10.44 ns ns ns ns 
Transcription        
 Early growth response 1 M28845 ns 3.18 ns 3.38 ns ns 
 Cell cycle/apoptosis        
 Bcl2-associated X protein L22472 ns −3.01 ns ns ns ns 
 Bcl2-associated X protein L22472 ns −3.01 ns ns ns ns 
 Granzyme D X56990 ns ns ns ns ns 10.0 
 Granzyme E M36901 ns ns ns 5.79 ns 5.1 
 Granzyme F J03257 ns ns ns ns ns 5.1 
 Granzyme G J02872 ns ns ns ns ns 7.6 
Cytoskeleton/cell adhesion        
 α-Actinin 3 AF093775 ns 3.48 ns ns ns ns 
 Fibrillin 1 L29454 ns 2.58 ns ns ns −5.1 
 Fibulin 2 X75285 ns 2.92 ns 2.60 ns ns 
 Actin, α 1 M12347 ns ns 36.8 5.92 ns ns 
 Actin, α 2 X13297 ns ns 6.5 52.75 ns −6.8 
ECM/ECM-associated proteins/angiogenesis        
 TSP1 M62470 ns −3.27 ns −5.27 ns ns 
 ECM protein 2 U66166 ns ns ns ns −3.7 ns 
 Cathepsin K AJ006033 ns 3.52 ns ns ns ns 
 Dystroglycan 1 U43512 −2.62 ns ns ns ns ns 
 Tenascin C X56304 ns 3.33 2.9 5.96 2.9 ns 
 Tenascin-X AB010266 ns 13.90 ns ns ns ns 
 Troponin T3 L48989 ns 34.9 ns −12 ns ns 
 Clusterin D14077 ns ns ns −5.46 ns ns 
 TIMP-3 gene U26437 ns ns ns −3.86 ns ns 
 MMP-3 X66402 ns 2.54 ns ns ns ns 
 Mast cell protease 5 M68898 ns −3.42 ns ns ns ns 
 Mast cell protease 7 L00653 ns −3.51 ns ns ns ns 
 Secretory leukocyte protease inhibitor AF002719 ns ns −2.6 ns −2.6 5.3 
 Meltrin, β AA726223 ns ns ns −5.52 2.6 ns 
 GPI anchor attachment protein 1 AB002136 ns ns ns ns 3.2 ns 
 α-1 type I procollagen U03419 ns 7.25 ns ns ns ns 
 Gap junction membrane channel protein β 3 X63099 ns ns ns ns ns 5.3 
Hematopoiesis/inflammation/prostaglandin metabolism        
 Stromal cell derived factor 1 L12029 ns 5.87 ns 6.00 ns ns 
 Prostaglandin-endoperoxide synthase 1 M34141 ns −4.73 ns −7.66 ns ns 
 Prostaglandin I2 (prostacyclin) synthase AB001607 ns ns −5.9 ns ns ns 
 Thromboxane A synthase 1 L18868 ns ns −3.3 ns ns ns 
 β-1-globin V00722 2.71 ns ns ns ns ns 
 Hemoglobin, β adult major chain J00413 2.53 ns ns ns ns −7.3 
 Hemoglobin, β adult minor chain V00722 ns ns ns ns ns −17.9 
 Hemoglobin α, adult chain 1 V00714 ns ns ns ns ns −6.3 
 Haptoglobin M96827 ns 9.76 ns ns ns ns 
 Mast cell chymase 2 M68899 ns −3.94 ns ns ns ns 
 Small inducible cytokine A12 U50712 ns ns −5.2 ns ns ns 
Gene descriptionAccession no.ErbB-1/2ErbB-1/3ErbB-1/4ErbB-2/3ErbB-2/4ErbB-3/4
Receptors/signal transduction        
 Insulin-like growth factor 2 X71922 nsa 16.29s 117.2 88.93 ns ns 
 Insulin-like growth factor binding protein 5 L12447 ns 4.61 117.2 88.93 ns ns 
 Growth factor-inducible protein M59821 ns −2.93 ns ns ns ns 
 Heparin binding epidermal growth factor-like growth factor L07264 ns ns ns ns ns 4.7 
 Polo-like kinase U73170 ns −10.44 ns ns ns ns 
Transcription        
 Early growth response 1 M28845 ns 3.18 ns 3.38 ns ns 
 Cell cycle/apoptosis        
 Bcl2-associated X protein L22472 ns −3.01 ns ns ns ns 
 Bcl2-associated X protein L22472 ns −3.01 ns ns ns ns 
 Granzyme D X56990 ns ns ns ns ns 10.0 
 Granzyme E M36901 ns ns ns 5.79 ns 5.1 
 Granzyme F J03257 ns ns ns ns ns 5.1 
 Granzyme G J02872 ns ns ns ns ns 7.6 
Cytoskeleton/cell adhesion        
 α-Actinin 3 AF093775 ns 3.48 ns ns ns ns 
 Fibrillin 1 L29454 ns 2.58 ns ns ns −5.1 
 Fibulin 2 X75285 ns 2.92 ns 2.60 ns ns 
 Actin, α 1 M12347 ns ns 36.8 5.92 ns ns 
 Actin, α 2 X13297 ns ns 6.5 52.75 ns −6.8 
ECM/ECM-associated proteins/angiogenesis        
 TSP1 M62470 ns −3.27 ns −5.27 ns ns 
 ECM protein 2 U66166 ns ns ns ns −3.7 ns 
 Cathepsin K AJ006033 ns 3.52 ns ns ns ns 
 Dystroglycan 1 U43512 −2.62 ns ns ns ns ns 
 Tenascin C X56304 ns 3.33 2.9 5.96 2.9 ns 
 Tenascin-X AB010266 ns 13.90 ns ns ns ns 
 Troponin T3 L48989 ns 34.9 ns −12 ns ns 
 Clusterin D14077 ns ns ns −5.46 ns ns 
 TIMP-3 gene U26437 ns ns ns −3.86 ns ns 
 MMP-3 X66402 ns 2.54 ns ns ns ns 
 Mast cell protease 5 M68898 ns −3.42 ns ns ns ns 
 Mast cell protease 7 L00653 ns −3.51 ns ns ns ns 
 Secretory leukocyte protease inhibitor AF002719 ns ns −2.6 ns −2.6 5.3 
 Meltrin, β AA726223 ns ns ns −5.52 2.6 ns 
 GPI anchor attachment protein 1 AB002136 ns ns ns ns 3.2 ns 
 α-1 type I procollagen U03419 ns 7.25 ns ns ns ns 
 Gap junction membrane channel protein β 3 X63099 ns ns ns ns ns 5.3 
Hematopoiesis/inflammation/prostaglandin metabolism        
 Stromal cell derived factor 1 L12029 ns 5.87 ns 6.00 ns ns 
 Prostaglandin-endoperoxide synthase 1 M34141 ns −4.73 ns −7.66 ns ns 
 Prostaglandin I2 (prostacyclin) synthase AB001607 ns ns −5.9 ns ns ns 
 Thromboxane A synthase 1 L18868 ns ns −3.3 ns ns ns 
 β-1-globin V00722 2.71 ns ns ns ns ns 
 Hemoglobin, β adult major chain J00413 2.53 ns ns ns ns −7.3 
 Hemoglobin, β adult minor chain V00722 ns ns ns ns ns −17.9 
 Hemoglobin α, adult chain 1 V00714 ns ns ns ns ns −6.3 
 Haptoglobin M96827 ns 9.76 ns ns ns ns 
 Mast cell chymase 2 M68899 ns −3.94 ns ns ns ns 
 Small inducible cytokine A12 U50712 ns ns −5.2 ns ns ns 
a

ns, not significant.

We thank Drs. David J. Riese, II (Purdue University, West Lafayette, IN), and David F. Stern (Yale University, New Haven, CT) for the ErbB receptor cDNAs.

1
Yarden Y., Sliwkowski M. X. Untangling the ErbB signalling network.
Nat. Rev. Mol. Cell Biol.
,
2
:
127
-137,  
2001
.
2
Olayioye M. A., Neve R. M., Lane H. A., Hynes N. E. The ErbB signaling network: receptor heterodimerization in development and cancer.
EMBO J.
,
19
:
3159
-3167,  
2000
.
3
Sandgren E. P., Luetteke N. C., Palmiter R. D., Brinster R. L., Lee D. C. Overexpression of TGF-α in transgenic mice: induction of epithelial hyperplasia, pancreatic metaplasia, and carcinoma of the breast.
Cell
,
61
:
1121
-1135,  
1990
.
4
Jhappan C., Stahle C., Harkins R. N., Fausto N., Smith G. H., Merlino G. T. TGF-α overexpression in transgenic mice induces liver neoplasia and abnormal development of the mammary gland and pancreas.
Cell
,
61
:
1137
-1146,  
1990
.
5
Matsui Y., Halter S. A., Holt J. T., Hogan B. L., Coffey R. J. Development of mammary hyperplasia and neoplasia in MMTV-TGF-α transgenic mice.
Cell
,
61
:
1147
-1155,  
1990
.
6
Krane I. M., Leder P. NDF/heregulin induces persistence of terminal end buds and adenocarcinomas in the mammary glands of transgenic mice.
Oncogene
,
12
:
1781
-1788,  
1996
.
7
O’Shea S., Johnson K., Clark R., Sliwkowski M. X., Erickson S. L. Effects of in vivo heregulin β1 treatment in wild-type and ErbB gene-targeted mice depend on receptor levels and pregnancy.
Am. J. Pathol.
,
158
:
1871
-1880,  
2001
.
8
Wang S. C., Zhang L., Hortobagyi G. N., Hung M. C. Targeting HER2: recent developments and future directions for breast cancer patients.
Semin. Oncol.
,
28
:
21
-29,  
2001
.
9
Nicholson R. I., Gee J. M., Harper M. E. EGFR and cancer prognosis.
Eur. J. Cancer
,
37 (Suppl.4)
:
S9
-S15,  
2001
.
10
Dittadi R., Gion M. More about: prognostic importance of low c-erbB2 expression in breast tumors.
J. Natl. Cancer Inst. (Bethesda)
,
92
:
1443
-1444,  
2000
.
11
Ferrero-Pous M., Hacene K., Tubiana-Hulin M., Spyratos F. Re: Prognostic importance of low c-erbB2 expression in breast tumors.
J. Natl. Cancer Inst. (Bethesda)
,
91
:
1584
-1585,  
1999
.
12
Koscielny S., Terrier P., Spielmann M., Delarue J. C. Prognostic importance of low c-erbB2 expression in breast tumors.
J. Natl. Cancer Inst. (Bethesda)
,
90
:
712
1998
.
13
Piffanelli A., Dittadi R., Catozzi L., Gion M., Capitanio G., Gelli M. C., Brazzale A., Malagutti R., Pelizzola D., Menegon A., Giovannini G., Gardini G. Determination of ErbB2 protein in breast cancer tissues by different methods. Relationships with other biological parameters.
Breast Cancer Res. Treat.
,
37
:
267
-276,  
1996
.
14
Bozcuk H., Uslu G., Pestereli E., Samur M., Ozdogan M., Karaveli S., Sargin F., Savas B. Predictors of distant metastasis at presentation in breast cancer: a study also evaluating associations among common biological indicators.
Breast Cancer Res. Treat.
,
68
:
239
-248,  
2001
.
15
Albanell J., Baselga J. Unraveling resistance to trastuzumab (Herceptin): insulin-like growth factor I receptor, a new suspect.
J. Natl. Cancer Inst. (Bethesda)
,
93
:
1830
-1832,  
2001
.
16
Carneiro F., Sobrinho-Simoes M. The prognostic significance of amplification and overexpression of c-met and c-erb B-2 in human gastric carcinomas.
Cancer (Phila.)
,
88
:
238
-240,  
2000
.
17
Mirza A. N., Mirza N. Q., Vlastos G., Singletary S. E. Prognostic factors in node-negative breast cancer: a review of studies with sample size more than 200 and follow-up more than 5 years.
Ann. Surg.
,
235
:
10
-26,  
2002
.
18
Chow N. H., Chan S. H., Tzai T. S., Ho C. L., Liu H. S. Expression profiles of ErbB family receptors and prognosis in primary transitional cell carcinoma of the urinary bladder.
Clin. Cancer Res.
,
7
:
1957
-1962,  
2001
.
19
Porebska I., Harlozinska A., Bojarowski T. Expression of the tyrosine kinase activity growth factor receptors (EGFR, ERB B2, ERB B3) in colorectal adenocarcinomas and adenomas.
Tumour Biol.
,
21
:
105
-115,  
2000
.
20
Suo Z., Risberg B., Kalsson M. G., Willman K., Tierens A., Skovlund E., Nesland J. M. EGFR family expression in breast carcinomas, c-erbB-2 and c-erbB-4 receptors have different effects on survival.
J. Pathol.
,
196
:
17
-25,  
2002
.
21
Xia W., Lau Y. K., Zhang H. Z., Xiao F. Y., Johnston D. A., Liu A. R., Li L., Katz R. L., Hung M. C. Combination of EGFR, HER-2/neu, and HER-3 is a stronger predictor for the outcome of oral squamous cell carcinoma than any individual family members.
Clin. Cancer Res.
,
5
:
4164
-4174,  
1999
.
22
Gilbertson R. J., Perry R. H., Kelly P. J., Pearson A. D., Lunec J. Prognostic significance of HER2 and HER4 coexpression in childhood medulloblastoma.
Cancer Res.
,
57
:
3272
-3280,  
1997
.
23
Gilmour L. M., Macleod K. G., McCaig A., Gullick W. J., Smyth J. F., Langdon S. P. Expression of erbB-4/HER-4 growth factor receptor isoforms in ovarian cancer.
Cancer Res.
,
61
:
2169
-2176,  
2001
.
24
Alimandi M., Romano A., Curia M. C., Muraro R., Fedi P., Aaronson S. A., Di Fiore P. P., Kraus M. H. Cooperative signaling of ErbB3 and ErbB2 in neoplastic transformation and human mammary carcinomas.
Oncogene
,
10
:
1813
-1821,  
1995
.
25
Siegel P. M., Ryan E. D., Cardiff R. D., Muller W. J. Elevated expression of activated forms of Neu/ErbB-2 and ErbB-3 are involved in the induction of mammary tumors in transgenic mice: implications for human breast cancer.
EMBO J.
,
18
:
2149
-2164,  
1999
.
26
Lemoine N. R., Barnes D. M., Hollywood D. P., Hughes C. M., Smith P., Dublin E., Prigent S. A., Gullick W. J., Hurst H. C. Expression of the ERBB3 gene product in breast cancer.
Br. J. Cancer
,
66
:
1116
-1121,  
1992
.
27
Naidu R., Yadav M., Nair S., Kutty M. K. Expression of c-erbB3 protein in primary breast carcinomas.
Br. J. Cancer
,
78
:
1385
-1390,  
1998
.
28
Furger C., Fiddes R. J., Quinn D. I., Bova R. J., Daly R. J., Sutherland R. L. Granulosa cell tumors express erbB4 and are sensitive to the cytotoxic action of heregulin-β2/PE40.
Cancer Res.
,
58
:
1773
-1778,  
1998
.
29
Srinivasan R., Poulsom R., Hurst H. C., Gullick W. J. Expression of the c-erbB-4/HER4 protein and mRNA in normal human fetal and adult tissues and in a survey of nine solid tumour types.
J. Pathol.
,
185
:
236
-245,  
1998
.
30
Yen L., You X. L., Al Moustafa A. E., Batist G., Hynes N. E., Mader S., Meloche S., Alaoui-Jamali M. A. Heregulin selectively up-regulates vascular endothelial growth factor secretion in cancer cells and stimulates angiogenesis.
Oncogene
,
19
:
3460
-3469,  
2000
.
31
Jaitovich-Groisman I., Benlimame N., Slagle B. L., Perez M. H., Alpert L., Song D. J., Fotouhi-Ardakani N., Galipeau J., Alaoui-Jamali M. A. Transcriptional regulation of the TFIIH transcription repair components XPB and XPD by the hepatitis B virus x protein in liver cells and transgenic liver tissue.
J. Biol. Chem.
,
276
:
14124
-14132,  
2001
.
32
Novak J. P., Sladek R., Hudson T. J. Characterization of variability in large-scale gene expression data: implications for study design.
Genomics
,
79
:
104
-113,  
2002
.
33
Yen L., Benlimame N., Nie Z. R., Xiao D., Wang T., Moustafa A. E., Esumi H., Milanini J., Hynes N. E., Pages G., Alaoui-Jamali M. A. Differential regulation of tumor angiogenesis by distinct ErbB homo- and heterodimers.
Mol. Biol. Cell
,
13
:
4029
-4044,  
2002
.
34
Yoo J. Y., Hamburger A. W. Interaction of the p23/p198 protein with ErbB-3.
Gene (Amst.)
,
229
:
215
-221,  
1999
.
35
Penington D. J., Bryant I., Riese D. J. Constitutively active ErbB4 and ErbB2 mutants exhibit distinct biological activities.
Cell Growth Differ.
,
13
:
247
-256,  
2002
.
36
Cohen B. D., Kiener P. A., Green J. M., Foy L., Fell H. P., Zhang K. The relationship between human epidermal growth-like factor receptor expression and cellular transformation in NIH3T3 cells.
J. Biol. Chem.
,
271
:
30897
-30903,  
1996
.
37
Hellyer N. J., Cheng K., Koland J. G. ErbB3 (HER3) interaction with the p85 regulatory subunit of phosphoinositide 3-kinase.
Biochem. J.
,
333(Pt.3)
:
757
-763,  
1998
.
38
Lee R. J., Albanese C., Fu M., D’Amico M., Lin B., Watanabe G., Haines G. K., III, Siegel P. M., Hung M. C., Yarden Y., Horowitz J. M., Muller W. J., Pestell R. G. Cyclin D1 is required for transformation by activated Neu and is induced through an E2F-dependent signaling pathway.
Mol. Cell. Biol.
,
20
:
672
-683,  
2000
.
39
Sweeney C., Fambrough D., Huard C., Diamonti A. J., Lander E. S., Cantley L. C., Carraway K. L., III Growth factor-specific signaling pathway stimulation and gene expression mediated by ErbB receptors.
J. Biol. Chem.
,
276
:
22685
-22698,  
2001
.
40
Heppner G. H., Miller F. R. The cellular basis of tumor progression.
Int. Rev. Cytol.
,
177
:
1
-56,  
1998
.
41
Bowden E. T., Barth M., Thomas D., Glazer R. I., Mueller S. C. An invasion-related complex of cortactin, paxillin and PKCmu associates with invadopodia at sites of extracellular matrix degradation.
Oncogene
,
18
:
4440
-4449,  
1999
.
42
Coussens L. M., Tinkle C. L., Hanahan D., Werb Z. MMP-9 supplied by bone marrow-derived cells contributes to skin carcinogenesis.
Cell
,
103
:
481
-490,  
2000
.
43
Nakahara H., Howard L., Thompson E. W., Sato H., Seiki M., Yeh Y., Chen W. T. Transmembrane/cytoplasmic domain-mediated membrane type 1-matrix metalloprotease docking to invadopodia is required for cell invasion.
Proc. Natl. Acad. Sci. USA
,
94
:
7959
-7964,  
1997
.
44
Sternlicht M. D., Lochter A., Sympson C. J., Huey B., Rougier J. P., Gray J. W., Pinkel D., Bissell M. J., Werb Z. The stromal proteinase MMP3/stromelysin-1 promotes mammary carcinogenesis.
Cell
,
98
:
137
-146,  
1999
.
45
Werb Z. ECM and cell surface proteolysis: regulating cellular ecology.
Cell
,
91
:
439
-442,  
1997
.
46
Kauraniemi P., Barlund M., Monni O., Kallioniemi A. New amplified and highly expressed genes discovered in the ERBB2 amplicon in breast cancer by cDNA microarrays.
Cancer Res.
,
61
:
8235
-8240,  
2001
.
47
Sorlie T., Perou C. M., Tibshirani R., Aas T., Geisler S., Johnsen H., Hastie T., Eisen M. B., van de Rijn M., Jeffrey S. S., Thorsen T., Quist H., Matese J. C., Brown P. O., Botstein D., Eystein L. P., Borresen-Dale A. L. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications.
Proc. Natl. Acad. Sci. USA
,
98
:
10869
-10874,  
2001
.
48
Chen H., Herndon M. E., Lawler J. The cell biology of thrombospondin-1.
Matrix Biol.
,
19
:
597
-614,  
2000
.
49
Zhang Y., Deng Y., Luther T., Muller M., Ziegler R., Waldherr R., Stern D. M., Nawroth P. P. Tissue factor controls the balance of angiogenic and antiangiogenic properties of tumor cells in mice.
J. Clin. Investig.
,
94
:
1320
-1327,  
1994
.
50
Adams J. C. Thrombospondins: multifunctional regulators of cell interactions.
Annu. Rev. Cell Dev. Biol.
,
17
:
25
-51,  
2001
.
51
de Fraipont F., Nicholson A. C., Feige J. J., Van Meir E. G. Thrombospondins and tumor angiogenesis.
Trends Mol. Med.
,
7
:
401
-407,  
2001
.
52
Sargiannidou I., Zhou J., Tuszynski G. P. The role of thrombospondin-1 in tumor progression.
Exp. Biol. Med. (Maywood)
,
226
:
726
-733,  
2001
.
53
Bornstein P. Thrombospondins: structure and regulation of expression.
FASEB J.
,
6
:
3290
-3299,  
1992
.