| Nucleic Acids Research | Pages |
Identification of differentially expressed genes associated with HER-2/neu overexpression in human breast cancer cells
Introduction
Materials And Methods
Cell culture
RNA preparation
Construction of the cDNA library
Differential hybridization
Probe generation for northern hybridization
Northern hybridization
DNA sequencing and computer analysis
In vitro transcription/translation
Extraction of total RNA from breast tumor samples
Results
Isolation of differentially expressed genes associated with HER-2/neu overexpression
DNA sequencing and identification of the clones
Confirmation of differential expression in ovarian cancer cell counterparts
Differential expression of two of the identified cDNAs in primary human breast cancer samples
Discussion
Acknowledgements
References
Identification of differentially expressed genes associated with HER-2/neu overexpression in human breast cancer cells
Received July 7, 1999; Revised and Accepted August 19, 1999
ABSTRACT Amplification and resulting overexpression of the HER-2/neu proto-oncogene is found in ~30% of human breast and 20% of human ovarian cancers. To better understand the molecular events associated with overexpression of this gene in human breast cancer cells, differential hybridization was used to identify genes whose expression levels are altered in cells overexpressing this receptor. Of 16 000 clones screened from an overexpression cell cDNA library, a total of 19 non-redundant clones were isolated including seven whose expression decreases (C clones) and 12 which increase (H clones) in association with HER-2/neu overexpression. Of these, five C clones and 11 H clones have been confirmed to be differentially expressed by northern blot analysis. This group includes nine genes of known function, three previously sequenced genes of relatively uncharacterized function and four novel genes without a match in GenBank. Examination of the previously characterized genes indicates that they represent sequences known to be frequently associated with the malignant phenotype, suggesting that the subtraction cloning strategy used identified appropriate target genes. In addition, differential expression of 12 of 16 (75%) cDNAs identified in the breast cancer cell lines are also seen in HER-2/neu-overexpressing ovarian cancer cells, indicating that they represent generic associations with HER-2/neu overexpression. Finally, up-regulation of two of the identified cDNAs, one novel and one identified but as yet uncharacterized gene, was confirmed in human breast cancer specimens in association with HER-2/neu overexpression. Further characterization of these genes may yield insight into the fundamental biology and pathogenetic effects of HER-2/neu overexpression in human breast and ovarian cancer cells.
INTRODUCTION
The human HER-2/neu (c-erbB-2) proto-oncogene encodes a transmembrane receptor tyrosine kinase with extensive sequence homology to the epidermal growth factor receptor (EGFR) (1). Amplification and/or overexpression of HER-2/neu has been found in one-third of human breast and one-fifth of ovarian cancers (2,3). In addition, the HER-2/neu alteration is associated with a poor clinical outcome in that women whose tumors contain it experience earlier disease relapse and shorter overall survival (3-6). Two hypotheses potentially account for this prognostic association. First, HER-2/neu overexpression may be an epiphenomenon serving merely as a marker of aggressive breast cancers. Conversely, the alteration may be causal for the aggressive phenotype. Considerable circumstantial evidence now supports the latter possibility, with data suggesting that overexpression plays a direct causal role in the pathogenesis of the malignancies in which it occurs (7-9).
In order to more directly evaluate the potential biological role of HER-2/neu overexpression in the pathogenesis of breast cancer, we developed an experimental model in which the effects of overexpression in human breast cancer cells can be studied (10-13). In these experiments, human MCF-7 breast cancer cells which express normal levels of the receptor were transfected with a retroviral expression vector containing a full-length cDNA encoding the human HER-2 gene (14). Multiple rounds of infection and sorting of the top 5% of HER-2 overexpressing, pooled transfectants generated a stably transfected cell line, MCF-7/HER-2, in which HER-2 expression levels were comparable to those observed in human HER-2 overexpressing breast cancer specimens (12). MCF-7 cells were similarly transfected using an empty neomycin resistance vector including multiple infections, producing the MCF-7/control cell line. The amount of HER-2/neu protein expressed, as determined by quantitative western blot analysis, was ~1.62 pg/cell for MCF-7/HER-2 cells as compared to 0.36 pg/cell for MCF-7/control cells (5). In vitro and in vivo studies of these engineered cells demonstrated that the growth characteristics of the MCF-7/HER-2 human breast cancer cell line are significantly altered by overexpression of HER-2/neu (10-14). Increased cell proliferation was seen in the HER-2 overexpressing cell line as assessed by [3H]thymidine incorporation and in vitro cell proliferation assays. In addition, HER-2 overexpression markedly improved soft agar cloning efficiency, and the cells exhibited increased tumorigenicity in nude mice (13,14). Together, the data confirmed that overexpression of the HER-2 receptor tyrosine kinase plays a role in altering the biological behavior of human breast cancer cells. The exact molecular mechanism(s) by which this overexpression promotes a more aggressive phenotype of these cells, however, remains unknown. There are multiple potential mechanisms by which the observed phenotypic changes may occur. Increased amounts and/or activation of this cell surface receptor may affect either the expression or function of other molecules involved in regulation of cell proliferation. Direct effects of HER-2 overexpression on other cellular proteins can be accomplished by changes in: (i) expression at the mRNA transcript level; (ii) protein production at the translational level; or (iii) protein activation/modification at the post-translational level. The cellular changes associated with HER-2/neu overexpression are likely to be induced by most or all of these mechanisms. To identify those changes associated with differential expression of genes at the transcript level, we undertook a differential screening analysis.
The subtraction cloning technique termed differential hybridization, also known as plus/minus screening (15), was used to isolate genes which are differentially expressed in MCF-7/HER-2 cells as compared to MCF-7/control cells. This approach has the advantage of comparing two human breast cancer cell lines which are identical except for HER-2/neu overexpression, allowing for a direct comparison of cDNAs derived from the two cell populations. In the current study, we identified a series of genes, either previously characterized or entirely novel, whose expression levels are altered in association with HER-2/neu overexpression. It is possible that some of these genes might be mediators of the HER-2 overexpressing phenotype since we have confirmed their differential expression not only in human ovarian cancer cell lines which overexpress HER-2 but also primary breast cancer specimens containing this alteration.
MATERIALS AND METHODS
Cell culture
Cells were grown in RPMI medium 1640, supplemented with 10% fetal bovine serum, 2 mM glutamine and 1% penicillin G/streptomycin/fungizone solution. Cells were harvested at 80% confluency for total RNA extraction.
RNA preparation
Total cellular RNA was purified by guanidinium/cesium chloride ultracentrifugation (16). mRNA was isolated by two passages through an oligo(dT) cellulose column (T3; Collaborative Research) (17). The quality and mRNA composition of the resulting RNA population were confirmed by northern blot analysis by probing with [beta]-actin and HER-2 cDNAs. Both MCF-7/control and MCF-7/HER2 mRNA pools contain equivalent, basal expression of the endogenous HER-2 transcript whereas the transcript representing transfected HER-2 cDNA was present only in the MCF-7/HER2 cells.
Construction of the cDNA library
Five micrograms of MCF-7/HER2 poly(A)+ RNA was constructed into ZAP ExpressTM vector (Stratagene) according to the manufacturer's protocol using materials provided in the cDNA synthesis kit. The recombinant phage were packaged in Gigapack II Gold packaging extract (Stratagene). The packaged library was amplified for one round through passage on XL1-Blue MRF[prime] host cells (6 × 105 p.f.u./µl titer). As determined by the X-gal/IPTG color assay, the background (non-recombinant) phage level was <0.1%.
Differential hybridization
The MCF-7/HER2 cDNA library was plated on XL1-Blue MRF[prime] host cells at a density of 2000 p.f.u. in each of eight 150 mm Petri dishes. After plating, actin and HER-2 clones purified from the same library were each loaded onto four designated spots within the individual plates to be used as hybridization controls. The nitrocellulose filters (Millipore) were placed on the agar plates 1.5 min for the first filter, 3 min for the second, and 7 min for the third. The phage DNA was denatured for 3 min in a solution containing 0.5 M NaOH, 1.5 M NaCl, neutralized for 3 min in a solution containing 3 M NaCl, 0.5 M Tris, pH 7.5, and rinsed in 2× SSC. The treated filters were air dried and baked at 80°C for 1 h.
The radiolabeled cDNA probes (MCF7/control and MCF-7/HER2) were prepared as follows. Poly(A)+ RNA was randomly labeled, using both random hexamers and oligo(dT) primers, in 20 µl solution containing 1.0 µg poly(A)+ RNA, 1× MMLV buffer, 1 mM each dATP, dGTP and dTTP, 0.045 mM dCTP, 100 µCi [[alpha]-32P]dCTP, 0.5 µg oligo(dT)15, 0.2 µg random primer, 20 U RNase inhibitor, and 200 U Moloney murine leukemia virus (MMLV) reverse transcriptase. The reaction was incubated at room temperature first for 10 min to allow primer annealing and further incubated at 37°C for 1 h. Upon addition of 4.6 µl of 0.5 M NaOH, the reaction mix was incubated at 70°C for 20 min. Incorporated counts were eluted from a spin column (Chromaspin 100; Clontech) in 1× TEN buffer (0.1 M NaCl, 10 mM Tris, pH 8.0, 1 mM EDTA).
Approximately 5.9 × 107 d.p.m. of MCF-7/control probes were added to a first hybridization containing 26 ml hybridization solution and the first set of eight filters obtained from each plate. Equal counts of MCF-7/HER2 probes were added to the second set of filters. The third set of filters were hybridized with radioactive HER-2 cDNA in order to avoid selecting the HER-2-containing clones. Hybridization solution contained 50% formamide, 25× Denhardt's, sonicated salmon sperm DNA, NaPO4, pH 6.8, sodium pyrophosphate, and riboATP. Prehybridization was performed at 42°C for 4 h and hybridization at 42°C for 4 days (overnight hybridization for the third set of filters). Filters were washed at room temperature for 5 min (three times) in 0.2× SSC, 0.1% SDS, and at 60°C for 15 min (seven times) in the same solution. The washed filters were exposed with an intensifying screen to Kodak-XAR5 films at -70°C for various time periods. Autoradiograms were analyzed to compare differences in focal signal intensity between the replica filters.
For the secondary screenings, the primary screening procedure was repeated except that each clone was separately plated onto a 100 mm Petri dish at a low density of 25-50 plaques/plate.
Probe generation for northern hybridization
The pBK-CMV phagemid was in vivo excised from the [lambda] phage vector according to the manufacturer's instructions (Stratagene). The cDNA inserts were isolated from the plasmid either by restriction enzyme digestion or by PCR amplification using T3 and T7 sequences as primers.
Northern hybridization
Either 2 µg of poly(A)+ RNA or 20 µg of total RNA was loaded onto a 1% formaldehyde-agarose gel and electrophoresed at 70 V for 4 h. The RNA was transferred to a nylon membrane in 10× SSC. The purified cDNA inserts were random labeled in a 50 µl reaction mix which contained 50 ng template, [[alpha]-32P]dCTP, 20 µg BSA, and 6 U Klenow. Incorporated counts were eluted from a G-50 Sephadex spin column (Pharmacia). Approximately 3 × 106 d.p.m./ml hybridization solution were used. The hybridization was carried out in 50% formamide, 2× SSC, 0.1% SDS, 10 mg/ml salmon sperm DNA, and 10% dextran sulfate, at 42°C for 16 h. Membranes were washed in 2× SSC, 0.1% SDS at 25°C for 10 min (three times), and in the same solution at 65°C for 5 min (twice). The washed membranes were exposed with an intensifying screen to Kodak-XAR film at -70°C.
DNA sequencing and computer analysis
Minipreparations of pBK-CMV plasmid vector (Qiawell 8 Ultra; Qiagen) were sequenced with T3/T7 promoter primers and internal primers using an automatic DNA sequencer (Applied Biosystems Model 373A). The sequence similarity search was performed using the GenBank and EMBL DNA databases.
In vitro transcription/translation
The cDNA inserts weretranslated into polypeptides in a TNT coupled reticulocyte system (Promega) according to the manufacturer's protocol. Approximately 1 µg of purified plasmid template was transcribed and translated in the 50 µl reaction containing T3 RNA polymerase, rabbit reticulocyte lysate, [35S]methionine, etc. A sample of 5 µl of the end product was aliquoted to estimate the molecular size of the in vitro translated protein using 10% SDS-PAGE and prestained protein size markers (Bio-Rad).
Extraction of total RNA from breast tumor samples
Breast tumors were obtained from patients at the time of surgery as part of a core tissue procurement resource sponsored by the DOD breast cancer program. All tumor samples were snap frozen in liquid nitrogen and kept at -70°C before extraction of RNA. Frozen tissues were pulverized in liquid nitrogen prior to homogenization in cold 4 M guanidine thiocyanate buffer (7.5 ml/g tissue). The homogenates were centrifuged for 10 min at 4°C at 8000 g in order to remove cell debris. RNA was sedimented through a cesium chloride gradient (5.7-2.4 M CsCl2) via ultracentrifugation (18 h at 36 000 r.p.m., 20°C). The separated RNA phases were extracted with phenol/chloroform prior to a wash with 100% ethanol. The RNA pellet was precipitated by adding 2 ml of 0.4 M sodium acetate and 2.5 vol of 100% ethanol, and storing overnight at -20°C. After centrifugation (20 min at 10 000 g), the pellets were dried and dissolved in DEPC-treated water.
RESULTS
Isolation of differentially expressed genes associated with HER-2/neu overexpression
In our first round of differential screening, 16 000 clones from the MCF-7/HER-2 library were analyzed. Clones showing a stronger signal intensity hybridized with the MCF-7/control cell cDNA probes were labeled C clones (C1, C2, C3, etc.), whereas those demonstrating a stronger signal intensity hybridized with HER-2-overexpressing cell cDNA probes were labeled H clones (H1, H2, H3, etc.). From this primary screening, a total of 127 differentially expressed clones were isolated including 77 C clones and 50 H clones representing genes whose expression levels are decreased (C clones) or increased (H clones), respectively, in association with HER-2 overexpression. Each clone was ranked according to degree of differential hybridization based on signal intensity ranging from more than a 5-fold to less than a 2-fold change based on visualization.
Forty-three C clones and 36 H clones which demonstrated the greatest differences in signal intensity in the primary screening were taken through secondary screening to ensure consistent differential expression and to isolate pure colonies. Subsequent to this isolation, the clones were cross-hybridized to determine redundancy. This resulted in a total of seven non-redundant C clones and 12 non-redundant H clones. Finally, to confirm our screening technique, differential expression patterns of the selected clones were evaluated by northern blot analysis of RNA from MCF-7/control and HER-2 cells. A total of five C clones and 11 H clones showed expression patterns consistent with expectations from the differential hybridization approach (Fig. 1), while two C clones and one H clone failed to demonstrate the anticipated pattern.
Figure 1. Northern blot analysis of candidate gene expression. Differential expression patterns were confirmed by northern blot analyses for five C clones and 11 H clones. For clones H35 and H45, 20 µg of total RNA was loaded in each lane. For the remaining clones, 2 µg of poly(A)+ RNA was loaded (C, MCF-7/control mRNA; H, MCF-7/HER-2 mRNA). Ethidium bromide staining of RNA gel is shown below autoradiograms to illustrate equal loading and quality of RNA. The size of the differentially expressed transcript is indicated on the left.
DNA sequencing and identification of the clones
Individual clones whose differential expression was confirmed by northern blot analysis were subsequently analyzed by DNA sequencing, and these data demonstrate that full-length cDNAs were obtained for most of the clones. The differentially expressed genes were grouped into three different classes based on computer searches against the GenBank and EMBL databases: (i) known genes with previously characterized function; (ii) previously identified genes with relatively uncharacterized function; (iii) novel sequences (summarized in Table 1). In addition, each clone was grouped into three different categories based on significance of their relative difference in expression (Table 1). Even genes whose differential expression is small (~2-fold) were included if this difference was consistently reproducible in multiple analyses.
Table 1. Identity of differentially expressed clones
| Clonea | Sizeb (bp) | Clone identity | mRNA sizec (kb) | Relative difference in expressiond | Accession no. |
| Genes with previously characterized function | |||||
| C29 | 1777 | Keratin 8 | 1.8 | [darr][darr] | X74929 |
| C31 | 2053 | Cathepsin D | 2.1 | [darr][darr] | M11233 |
| C49 | 1423 | Keratin 18 | 1.4 | [darr][darr] | M26326 |
| C72 | 1940 | [gamma]-Actin | 1.9 | [darr] | X04098 |
| H16 | 933 | Ribosomal protein L8 | 0.9 | [uarr] | Z28407 |
| H18 | 2530 | 90 kDa heat shock protein | 1.2, 2.5 | [uarr][uarr][uarr] | M16660 |
| H31 | 1284 | Glyceraldehyde 3-phosphate dehydrogenase | 1.3 | [uarr][uarr] | M33197 |
| H35 | 948 | LLRep3 | 0.9 | [uarr][uarr][uarr] | X17206 |
| H45 | 2317 | Succinyl CoA:3-oxoacid CoA transferase | 1.5, 3.3, 5.3 | [uarr][uarr] | U62961 |
| Recently identified genes with relatively uncharacterized function | |||||
| H13 | 1027 | DNA fragmentation factor (DFF) | 1.0, 1.4, 3.4, 6.4 | [uarr][uarr] | AF103799e |
| H14 | 2214 | Density-regulated protein-1 (DRP-1) | 0.96, 2.2, 6.5 | [uarr][uarr][uarr] | AF103800e |
| H37 | 3091 | RNA binding motif protein 5 (RBMS) | 1.9, 3.1, 6.5 | [uarr][uarr] | AF103802e |
| Novel sequences | |||||
| C40 | 1750 | Not previously identified | 1.8, 2.6, 4.9 | [darr] | AF103798e |
| H17 | 1981 | Not previously identified | 2.0, 4.4 | [uarr][uarr][uarr] | AF103801e |
| H41 | 3346 | Not previously identified | 1.9, 2.7, 3.3, 4.0 | [uarr][uarr][uarr] | AF103803e |
| H63 | 2068 | Not previously identified | 1.9, 4.5 | [uarr] | AF103804e |
bThe size of cDNAs cloned in the differential screening and used as probes for northern blot analysis.
cThesizes of differentially expressed transcripts on northern autoradiogram.
dDirections of arrows indicate expression level increase ([uarr]) or decrease ([darr]), respectively, in HER-2 overexpressing cells. The number of arrows indicates relative difference in expression level change by visualization, i.e. [uarr], ~2-fold; [uarr][uarr], 3-5-fold; [uarr][uarr][uarr], >5-fold.
eThe nucleotide sequences reported in this study have been submitted to the GenBank/EMBL database and assigned these accession nos.
As stated above, the MCF-7/HER-2 cells behave significantly differently than their isogenic control parental counterparts; with increases in DNA synthesis, cell growth in vitro, soft agar cloning efficiency and tumorigenicity. Nine of the differentially expressed genes identified in this study are known to be associated with the malignant phenotype. For example, cytokeratin 8 (C29) (18), cytokeratin 18 (C49) (19) and [gamma]-actin (C72) (20) are cytoskeletal proteins essential for maintaining both cell shape and motility of normal cells and their expression differs from levels seen in the aberrant cytoskeleton of cancer cells (21). Similarly, GAPDH (H31) (22) and succinyl CoA transferase (H45) (23) are involved in the metabolic pathway of more rapidly growing cells, including cancer cells. The increased expression of ribosomal proteins L8 (H16) (24) and LLrep3 (H35) (25) is consistent with the increased rate of protein translation required for cancer cell growth. Two other genes identified in our study whose cellular functions have been previously characterized are cathepsin D (C31) (26), an acidic lysosomal protease, and the 90 kDa heat shock protein (H18) (27), which is a chaperone protein associated with steroid hormone receptor genes. Both of these genes are known to be differentially expressed in malignant cells (28,29).
Three genes (H13, H14 and H37) found to be overexpressed in the HER-2-overexpressing cells matched cDNA sequences which were previously identified by other investigators but not fully characterized. The H13 clone appears to be an alternative splice variant of DNA fragmentation factor (DFF) (GenBank accession no. U91985) (30); the first 261 amino acid sequences contained in both H13 and the DFF open reading frames are identical, but H13 lacks 70 amino acids at the 3[prime]-end and contains seven different amino acids in their place. The predictive amino acid sequence for H13 is highly similar to the ICAD (inhibitor of caspase-activated DNase) S and L proteins (31,32) (73 and 69% sequence identity, respectively). In order to ensure that the cDNAs cloned, either uncharacterized or novel, can be efficiently translated into protein products of expected sizes, we performed in vitro translation experiments. As predicted, the H13 cDNA was translated into a polypeptide of ~30 kDa (Fig. 2). Clone H14 is identical to DRP-1 (density-regulated protein-1, GenBank accession no. AF038554) (33) except that it lacks 285 bp at the 5[prime]-end and has an additional 23 bp at the 3[prime]-end plus a poly(A) tail. Both the H14 and DRP-1 cDNAs are likely to be partial, 3[prime]-end sequences of a larger transcript since no suitable initiating codon was found in either sequence, and the H14 cDNA hybridized with an additional transcript of ~6.5 kb on northern blot analysis. According to the EST (expressed sequence tags) database analysis, the 5[prime]-end of the H14 cDNA sequence can be extended, and the EST covering the extended portion of the gene is designated THC202438 (deposited in the Tentative Human Consensus Effort) (34). The H37 cDNA sequence has previously been deposited in GenBank as RNA-binding motif protein 5 (RBM5) (accession no. AF091263; unpublished results) found within a region reported to be homozygously deleted in lung cancer and believed to contain a major tumor suppressor gene(s) involved in a majority of small cell and non-small cell lung cancers (35). The H37 cDNA contains an open reading frame of 816 amino acids and is translated in vitro into a predicted protein product of ~90 kDa (Fig. 2). Analysis of the putative H37 protein against the PROSITE protein profile database recognized the presence of two RNA-binding domains, located at amino acid residues 140-147 and 274-281, which are perfect matches with the consensus eukaryotic sequence for a putative RNA-binding region RNP-1 (36).
Figure 2. In vitro transcription/translation of the proteins from the identified differentially expressed transcripts. The transcription-coupled translation reaction was performed using T3 RNA polymerase, rabbit reticulocyte lysate and [35S]methionine labeling. The first lane represents the 61 kDa luciferase protein product which was used as a positive control. The C40, H13, H17 and H37 protein products are seen as distinct bands at 55, 30, 50 and 90 kDa, respectively, whereas the H41 cDNA produced two faint bands at 30 kDa and a lower molecular weight. The protein molecular weight marker is shown on the left.
Four clones (C40, H17, H41 and H63) represented as yet unknown genes in the DNA databases, and three of these (C40, H17 and H41) were found to contain probable open reading frames. The 1750 bp long C40 clone contains a 510 amino acid open reading frame (Fig. 3A). The coding region of this gene begins with a start codon at nucleotide position 74 and has an in-frame stop codon at position 1604 (Fig. 3A). The C40 clone was in vitro translated into the predicted major protein product of ~55 kDa (Fig. 2). The 385 bp region (nt 568-952) of this gene is 89% identical to a GenBank transcript (accession no. U56429); however, the putative C40 protein does not share any significant homology with any known proteins in the databases. Examination of the C40 protein sequence using the GCG program Motifs revealed the presence of a leucine zipper motif at amino acid positions 104-125, which is a perfect match with the consensus sequence for the leucine zipper pattern (37; Fig. 3A).
Figure 3. Schematic representation of the three differentially expressed novel genes. The thin line indicates a stretch of nucleotide sequences ending at the poly(A) tail, denoted by (A)N, at the base pair number written next. The filled box illustrates the location of the most probable open reading frame, with the numbers below indicating the base pair positions of the start and stop codons, respectively. (A) Map of the C40 cDNA. The leucine zipper motif is denoted by LZ and shown above is the corresponding amino acid positions and sequences. (B) Map of the H17 cDNA. The asterisks above the poly(A) tail indicate the presence of polyadenylation signals. The hatched box illustrates the amino acid region of shared homology and the sequence alignments are shown below the gene. Identical residues are indicated by shading (Z77667, a C.elegans cDNA of unknown function; AE001086, sarcosine oxidase). Numbers at the right represent corresponding amino acid positions. Gaps introduced for maximal alignment are marked with dashes. (C) Map of the H41 cDNA. NLS, nuclear localization signal; AF005858, one of the `fast evolving' Drosophila genes of unknown function.
The 1981 bp long H17 clone contains a consensus initiation codon (38) at nt 66 followed by a 486 amino acid open reading frame and a 458 bp 3[prime]-untranslated region including a polyadenylation signal (AATAAA) (Fig. 3B). In vitro translation generated the predicted protein product of ~50 kDa (Fig. 2). The putative H17 protein has 39.3% identity with a Caenorhabditis elegans cDNAof unknown function(Z77667) over a 422 amino acid region (amino acids 61-482) and is similar to a metabolic enzyme sarcosine oxidase (AE001086), with 29.2% identity in a 171 amino acid region (amino acids 66-236) (Fig. 3B).
The 3346 bp H41 clone contains a 323 bp 5[prime]-untranslated region followed by an initiation codon with a satisfactory Kozak context (38) and an extensive, 2249 bp 3[prime]-untranslated region (Fig. 3C). The 258 amino acid residues encoded by its open reading frame was translated in vitro into the predicted protein product of ~30 kDa and an additional protein of lower molecular weight (Fig. 2). The putative H41 protein is related to one of the `fast evolving' Drosophila genes of unknown function (AF005858) (39), with 28.7% identity in a 167 amino acid region (amino acids 9-175) (Fig. 3C). According to a pSORT protein database search, the H41 gene product is predicted to be a nuclear protein based on the presence of a nuclear localization signal, four basic amino acid (Lys) residues, at its N-terminus (94.1% reliability by Reinhardt's method) (40; Fig. 3C). For the above three novel cDNA sequences (C40, H17 and H41), we could not find any ESTs which would extend our sequences further at either the 5[prime]- or 3[prime]-ends.
Lastly, the novel H63 clone is believed to be a partial, 3[prime] sequence of a longer transcript because this sequence did not contain any probable open reading frames, and the 2068 bp DNA hybridized with a transcript of ~4.5 kb on a northern blot. According to the EST database analysis, the 5[prime]-end of the H63 cDNA sequence can be extended and assembled as THC175350.
Confirmation of differential expression in ovarian cancer cell counterparts
To ensure that differential expression of these genes is a phenomenon consistently associated with HER-2/neu overexpression rather than a unique event restricted to a single cell line, we evaluated the expression of these genes in human ovarian cancer cells (CaOv-3) engineered to overexpress HER-2/neu in a fashion identical to the MCF-7/HER-2 cells (10,12). The amount of HER-2/neu protein expressed, as determined by quantitative western blot analysis, was ~1.62 pg/cell for MCF-7/ HER-2 cells and 1.14 pg/cell for the CaOv-3/HER-2 cells, as compared to 0.36 and 0.41 pg/cell for the control transfected cells, respectively (5). In addition, the biological changes induced by HER-2/neu overexpression in the human ovarian cancer cells were similar to those seen and described above in the human breast cancer cells (13). Based on this consistent pattern of biological changes induced by HER-2/neu overexpression for both the MCF-7/HER-2 and CaOv-3/HER-2 cell lines, we would anticipate that at least some of the changes in mRNA expression patterns associated with HER-2 overexpression might be similar between these two cell lines if these genes are relevant to the HER-2/neu-overexpressing phenotype. Northern blot analysis of the ovarian cancer cell line pair demonstrated that 12 of 16 (75%) clones found to be differentially expressed in MCF-7/HER-2 as compared to isogenic control breast cancer cells were also differentially expressed in the CaOv-3 ovarian cancer cells (Fig. 4). In addition, for most of the clones, the degree of differential expression was consistent between these two distinct epithelial cell lines. This phenomenon is seen with clones H17, H18, H35, H37 and H41, which show marked increases in expression levels in association with HER-2/neu overexpression, and clones C49, C72, H16, H45 and H63, which show more subtle differences (Fig. 4). Four of the MCF-7 differentially expressed clones (C31, C40, H14 and H31) did not demonstrate any noticeable difference in expression in CaOv-3/HER-2 versus control cells.
Figure 4. Confirmation of differential expression in CaOv-3 ovarian cancer cells overexpressing HER-2. The differential expression patterns of three C clones and nine H clones identified in MCF-7 breast cancer cells were reproduced in CaOv-3 ovarian cancer cell counterparts on northern blots (C, control; H, HER-2 transfectant). Aliquots of 20 µg of total RNA were loaded in each lane. Ethidium bromide staining of 18S rRNA is shown as a loading control below autoradiograms. The transcript sizes are as shown in Figure 1.
Differential expression of two of the identified cDNAs in primary human breast cancer samples
To further confirm differential expression of the novel (or uncharacterized) cDNAs in actual human malignancies, we examined a panel of primary human breast cancer specimens by northern blot analyses. Among these, we were able to detect clear signals on the tumor northern blots for two of the tested clones, H37 and H41. Less success with the other cDNAs (i.e. C40, H13, H14, H17 and H63) can be best explained by their relatively rare message level. For the H37 cDNA, 15 individual cancer samples were analyzed, and eight of these (nos 3, 4, 5, 7, 8, 12, 14 and 15) overexpress HER-2/neu (Fig. 5A).Seven of these eight tumors (88%) demonstrated overexpression of the H37 transcript, while only one of seven (14%) of the non-HER-2 overexpressors overexpresses this cDNA ([kappa] = 0.732, P < 0.005) (Fig. 5A). For the H41 cDNA, five of seven (71%) of the HER-2 overexpressing malignancies overexpress the gene while two of eight (25%) non-HER-2 overexpressors were found to have increased levels of this novel transcript ([kappa] = 0.464, P < 0.075) (Fig. 5B). Tumor 8 did not express high levels of either H37 nor H41 despite its high HER-2 expression level. Overall, these data suggest that the above two novel genes may contribute in some way to the phenotype associated with HER-2 overexpression. Further studies of this association are currently underway using greater numbers of tumor specimens.
Figure 5. Up-regulation of the H37 and H41 transcripts correlates with HER-2/neu overexpression in human breast tumors (P < 0.005 and P < 0.075, respectively). (A) Northern blot analysis was performed to compare expression levels of the HER-2 versus H37 cDNAs in 15 individual breast tumor samples. Aliquots of 10 µg of total RNA were loaded in each lane, and the same blot was stripped for rehybridization with the second probe. (B) The expression levels of the HER-2 versus H41 cDNAs were analyzed in a separate northern blot experiment. The same set of breast tumor samples were used as in (A) except that tumor 16 was substituted for 14 due to depletion of the sample. Aliquots of 15 µg of total RNA were loaded in each lane except for tumor 15, for which only 5 µg were used because of lack of material. The blot was stripped as in (A). For both (A) and (B), ethidium bromide staining of 28S rRNA is shown below the autoradiograms for RNA loading control.
DISCUSSION
To evaluate some of the potential downstream events triggered by HER-2/neu overexpression, several genes whose expression levels are altered in association with HER-2/neu overexpression were identified using a subtraction cloning procedure. The rationale for this approach is the likelihood that some of the biological effects of HER-2/neu overexpression are mediated at least in part by changes in expression levels of specific genes. The differential screening approach compared MCF-7 breast cancer cell lines transfected with a human HER-2/neu cDNA (MCF-7/HER-2) or with an identical empty vector (MCF-7/control). The alternative approach of comparing two different non-engineered cell lines which are not isogenic, i.e. MCF-7 and/or MDA-MB-231 compared against SKBR3 and/or BT-474, respectively, is problematical in that the presence of non-HER-2-associated genetic differences unique to cells derived from different individuals would almost certainly complicate interpretation of the results. Such heterogenetic effects would confound identification of those genes which are differentially expressed in direct association with HER-2/neu overexpression. A relatively conventional subtraction cloning method termed differential hybridization has been successfully used by other investigators in the cloning of genes associated with various biological phenomena, including the galactose-inducible genes of yeast (15), human fibroblast interferon (41), a variety of heat shock proteins (42) and the metastasis suppressor gene nm-23 (43). This screening stratagy has the advantage of obtaining a high yield of full-length clones in contrast to more recent techniques, such as differential display or representational difference analysis (RDA), which require an additional procedure of screening a cDNA library using the DNA fragments obtained. One major limitation of this approach, however, is its low sensitivity, making it difficult to detect either rare transcripts (constituting <0.01% of the mRNA population) or those genes whose expression levels are only subtly changed (44).
From our initial screen of 16 000 MCF-7/HER-2 cDNA library clones, we identified five genes with decreased and 11 genes with increased expression levels in association with HER-2/neu overexpression. These clones include nine genes with previously identified cellular functions, three existing sequences of relatively uncharacterized function, and four novel genes without matching sequences in GenBank. A number of the known genes identified in our screening have been previously reported to be associated with several aspects of human breast cancer and/or tumorigenicity in general. Although the differential screening approach does not provide direct evidence that a given gene plays a critical role in the phenotypic changes associated with HER-2 overexpression, a review of the literature regarding some of the genes in our study indicates that they may be candidates. Recent data, for example, indicate that down-regulation of cytokeratin (C29 and C49) gene expression may result in disorganization of the cytoskeleton leading to enhanced invasive properties (21). Similarly, the [gamma]-actin (C72) transcript level is markedly decreased in salivary gland adenocarcinoma cells on acquisition of metastatic ability (45). These observations are consistent with our findings and are of interest given the fact that HER-2 overexpression is associated with increased metastatic potential (46-48). The observation that cathepsin D (C31) transcript level is decreased in HER-2 overexpressing breast cancer cells is consistent with the most recent clinical data (28,49), which contradict the original reports of high cathepsin D concentrations as indicative of a poorer prognosis (50). The 90 kDa heat shock protein (H18) forms highly stable complexes with the estrogen receptor and thus may play a role in mediating estrogen-dependent growth (51,52). Its potential role in regulating estrogen receptor activity in human breast cancer is interesting in the light of the interactions recently described between HER-2 and the estrogen receptor (53,54). Other known genes found in our screening to be overexpressed in association with HER-2 overexpression, ribosomal proteins L8 (H16) and LLrep3(H35), GAPDH (H31) and succinyl CoA transferase (H45), may be merely reflective of higher proliferation in HER-2 overexpressing tumors. Alternatively, differential expression of these genes may be more specifically linked to HER-2 overexpression. An example of this could be LLrep3, which was also identified in differential hybridization screening of a ras-transfected teratocarcinoma cell line compared to an isogenic cell control as increased 25-fold (55), however, this gene is not differentially expressed when comparing non-tumorigenic and tumorigenic NIH 3T3 cells transformed by Ha-ras, N-ras, v-myc, v-mos, v-src and v-abl (55).
DFF (H13) is also overexpressed in the HER-2 overexpressing cells and has recently been identified as a protein which functions downstream of caspase-3 during apoptosis (30). Its exact cellular role in this process, i.e. inhibition or promotion of apoptosis, however, is as yet undefined (56-58). Lastly, DRP-1 (H14), which is also increased in association with HER-2 overexpression, has been found to be preferentially expressed in cells grown at high density compared to cells at low density. Growth arrest by serum starvation or transforming growth factor B treatment does not however induce this gene's expression (33). Its role, if any, in the HER-2 phenotype remains to be determined.
The possibility that the pattern of differential gene expression observed in this study is unique to a given experimental cell line rather than a generic phenomenon associated with HER-2 overexpression was also addressed. To verify differential expression in another cell line, we utilized CaOv-3 ovarian cancer cells engineered to overexpress HER-2/neu. For 75% of the differentially expressed clones, the patterns identified in the breast cancer cells were also found in the human ovarian cancer cell counterparts. This consistent expression pattern, demonstrated across cell lines from two different epithelia (i.e. breast and ovary), suggest that the expression differences observed in our study are related to HER-2/neu overexpression. In addition, we found a correlation between overexpression of HER-2/neu and up-regulation of the H37 and H41 genes in actual human breast cancer specimens. Those genes which did not yield a signal on northern analysis likely due to rare message level are currently being evaluated by a quantitative RT-PCR approach to circumvent this difficulty. Given the problem in assessing northern blot analyses from whole tissue specimens resulting from dilutional artifacts introduced by surrounding normal cells, these correlations are encouraging. It is intriguing that the H37 cDNA, found to be overexpressed in HER-2 overexpressing cells in the current study and demonstrating convincing differential expression in actual tumor samples, is localized to a region of chromosome 3p21.3 alleged to contain a putative lung cancer tumor suppressor gene(s) (35). Further characterization of this gene at the functional and genomic levels should give further insight into this phenomenon.
The current studies indicate that HER-2/neu overexpression induces a pattern of consistent genetic alterations in target human cells. We recognize that there are more sensitive techniques, such as microarray chip technology, now available for evaluating differential gene expression and plan to reanalyze these cell line pairs using these newer appoaches. It is possible that some of the genes identified may in part be biological mediators of the aggressive biological behavior associated with HER-2/neu overexpression. Future elucidation of the role of these genes, in particular those with an as yet unknown function, in mediating malignant phenotype should provide further insights into the fundamental biology and pathogenetic effects of HER-2/ neu overexpression in human breast and ovarian cancer cells and may suggest novel treatment strategies for patients whose tumors contain these alterations.
ACKNOWLEDGEMENTS
We thank Lillian Ramos for her excellent technical assistance. We are also grateful to Drs Frank Calzone and Denny Afar for critical reading of the manuscript. Thanks are due to Ms Judith Mitchelle for secretarial guidance. This study was supported in part byfunds from theRevlon/UCLA Women's Cancer Research Program, DOD grant DAMD 17-94-J-4234, and NIH PO1 CA32737. J.O. is supported by a post-doctoral fellowship from the Susan G. Komen Breast Cancer Foundation.
REFERENCES
*To whom correspondence should be addressed. Tel: +1 310 825 5193; Fax: +1 310 267 2301; Email: dslamon{at}mednet.ucla.edu Present address: David R. Grosshans, University of Colorado School of Medicine, Denver, CO 80262, USA
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  W. B. Carter, G. Niu, M. D. Ward, G. Small, J. E. Hahn, and B. J. Muffly Mechanisms of HER2-Induced Endothelial Cell Retraction Ann. Surg. Oncol., October 1, 2007; 14(10): 2971 - 2978. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Dolled-Filhart, L. Ryden, M. Cregger, K. Jirstrom, M. Harigopal, R. L. Camp, and D. L. Rimm Classification of breast cancer using genetic algorithms and tissue microarrays. Clin. Cancer Res., November 1, 2006; 12(21): 6459 - 6468. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Callebaut and J.-P. Mornon OCRE: a novel domain made of imperfect, aromatic-rich octamer repeats Bioinformatics, March 15, 2005; 21(6): 699 - 702. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Schlee, T. Krug, O. Gires, R. Zeidler, W. Hammerschmidt, R. Mailhammer, G. Laux, G. Sauer, J. Lovric, and G. W. Bornkamm Identification of Epstein-Barr Virus (EBV) Nuclear Antigen 2 (EBNA2) Target Proteins by Proteome Analysis: Activation of EBNA2 in Conditionally Immortalized B Cells Reflects Early Events after Infection of Primary B Cells by EBV J. Virol., April 15, 2004; 78(8): 3941 - 3952. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. S. Wilson, H. Roberts, R. Leek, A. L. Harris, and J. Geradts Differential Gene Expression Patterns in HER2/neu-Positive and -Negative Breast Cancer Cell Lines and Tissues Am. J. Pathol., October 1, 2002; 161(4): 1171 - 1185. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. J. Oh, A. R. West, M. C. Fishbein, and D. J. Slamon A Candidate Tumor Suppressor Gene, H37, from the Human Lung Cancer Tumor Suppressor Locus 3p21.3 Cancer Res., June 1, 2002; 62(11): 3207 - 3213. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||