INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The epidermal growth factor (EGF) receptor (EGF-R) is a protein tyrosine kinase receptor that is the target for high-affinity ligands, such as EGF and transforming growth factor alpha. EGF-R shares extensive homology with the erbB oncogene product of the avian erythroblastosis virus (12, 17, 46, 64), and overproduction and/or amplification of EGF-R has been detected in several different types of human cancers (12, 39, 40, 46, 47, 63), including breast cancer (20, 21, 30, 53). Extensive studies have shown that breast tumors which overexpress EGF-R are most often estrogen receptor (ER) negative and have a poor prognosis (57, 59). Despite the clinical importance of the EGF-R in human breast cancer, the molecular mechanisms governing the control of EGF-R gene expression in these cells remains to be fully elucidated.

MDA-MB-468 and BT-20 estrogen-unresponsive breast cancer cell lines have similar phenotypes and overexpress EGF-R to similar degrees (1 × 10⁶ to 2 × 10⁶ binding sites per cell) (29, 32). Treatment of MDA-MB-468 cells with EGF increases EGF-R mRNA and protein levels (3, 32), due in part to a moderate transcriptional upregulation (19). However, the vast proportion of the EGF-induced increase in EGF-R mRNA levels in MDA-MB-468 cells cannot be accounted for by transcriptional enhancement. This suggests that posttranscriptional events must play a significant role in the regulation of EGF-R mRNA levels in these cells. Little is known of the regulation of EGF-R expression in BT-20 cells.

The regulation of mRNA decay is a central mechanism in the control of gene expression (51). Specific cis-acting structural RNA motifs can confer instability to mRNAs under appropriate conditions. A major class of these regulatory cis elements comprises adenosine-uridine pentamers (AUUUA) which are termed AU-rich elements (AREs). Shaw and Kamen initially observed that an ARE in the 3' untranslated region (3'-UTR) of granulocyte-macrophage colony-stimulating factor (GM-CSF) mRNA could stimulate the degradation of the normally stable -globin mRNA, reducing the half-life from many hours to less than 30 min (56). Similarly, the 3'-UTR of c-fos, which contains a 69-nucleotide (nt) ARE, reduced the stability of -globin mRNA (8, 9). AREs that function as RNA-destabilizing elements and target the mRNA for rapid degradation in the cytoplasm have been found in numerous mRNAs, including c-myc, junB, nur77, beta interferon, various interleukin (IL-1, IL-2, and IL-3), and tumor necrosis factor (TNF) mRNAs (8, 9, 10, 13, 24, 31, 51, 56). In vivo, mice lacking AREs in their TNF gene have defective destabilization and translational regulation of TNF mRNA and also develop specific phenotypes, suggesting a potential etiopathological role for the ARE (31).

Typically the ARE contains an AUUUA pentamer, repeated once or several times within the 3'-UTR. It is often found within a U-rich region of the mRNA. Recent work has suggested an ARE containing the nonamer UUAUUUA (U/A) (U/A) is more indicative of rapid destabilization (34, 68). Variations of the nonamer, such as extended pentamers (AUUUUA and AUUUUUA), are present in many complex AREs and have been identified as putative binding targets by a random RNA selection procedure (37). However, little is known of the role of isolated extended AU pentamers in the regulation of specific mRNAs in vivo. This is particularly relevant, as it is the combination of functionally and structurally distinct sequence motifs, such as AU pentamers, nonamers, extended AU pentamers, and U-rich stretches, that determines the ultimate destabilizing ability of each particular ARE.

A family of proteins that bind to AU-rich, and sometimes U-rich, RNAs with high affinity has been characterized by RNA electrophoretic mobility shift assay (REMSA). In many cases, binding of these AU-rich region-binding proteins (AUBPs; 30 to 45 kDa) regulates the turnover of ARE-containing mRNAs (15, 27). One of the best-characterized AUBPs is HuR, a 36-kDa member of the elav family of RRM (RNA recognition motif)-containing RNA-binding proteins (27, 37). HuR binds with high affinity to ARE sequences (27, 37) and plays an active role in the stabilization of specific mRNAs containing AREs, such as GLUT1, c-fos, GM-CSF, plasminogen activator inhibitor type 2, and p21^WAF1 mRNAs (18, 22, 26, 27, 38, 44). AUF1 is another well-characterized RNA-binding protein that binds to AREs, in particular that of c-myc (6, 16, 61, 67). AUF1 has been implicated in the regulation of many cytokine and G protein-coupled receptor mRNAs (61) and plays a major role in development (33). Interestingly, the binding of some AUBPs to AREs is regulated by activators of protein kinase C, such as phorbol esters (phorbol 12-myristate 13-acetate [PMA]) (4).

The EGF-R mRNA contains four separate AU-rich sequences in the 3'-UTR. Interestingly, A431 epidermoid cancer cells express both the full-length EGF-R mRNA and a truncated EGF-R mRNA which lacks the 3'-UTR. The truncated EGF-R mRNA transcript is more stable than the full-length transcript (28), suggesting that one or more of the AU-rich sequences in the EGF-R mRNA 3'-UTR may contribute to basal and possibly regulated changes in EGF-R mRNA stability. We used breast cancer cells (MDA-MB-468 and BT-20) to determine the contribution of mRNA stability to the regulation of EGF-R expression. In both cell lines, EGF stabilized EGF-R mRNA >2-fold. Transfection experiments identified a novel 260-nucleotide (nt) AU-rich (66%) cis element in the 3'-UTR of EGF-R mRNA that contained two ~75-nt AU-rich destabilizing sequences (EGF-R1A and EGF-R2A). trans-acting protein factors (55 to 80 kDa) that targeted the cis element and whose binding affinity was regulated by EGF and phorbol esters were identified. Mutational analysis demonstrated an important role for the extended AU pentamer at the 3' end of the cis element in regulating EGF-R mRNA stability. These findings demonstrate a central role for mRNA turnover in the regulation of EGF-R expression in breast cancer cells and illustrate that complex cis element RNA-protein interactions contribute to basal and regulated EGF-R mRNA decay.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and culture conditions. Cell lines were obtained from American Type Culture Collection (ATCC; Manassas, Va.). MDA-MB-468 (HTB 132) and HT1080 (CCL 121) cells were routinely cultured in Dulbecco's modified Eagle's medium-F12 medium supplemented with 10% fetal calf serum (FCS; Gibco-BRL, Melbourne, Australia). BT-20 (HTB 19), LNCaP (CRL 1740), SK-BR-3 (HTB 30), and MDA-MB-231 (HTB 26) cell lines were grown in RPMI 1640 supplemented with 10% FCS. MCF7 (HTB 22) cells were cultured in RPMI 1640 supplemented with 5% FCS. All cell lines were cultured in the presence of penicillin (100 U/ml) and streptomycin (100 mg/ml). Cells were grown to 70 to 80% confluence and within eight passages of the original stock received from ATCC for all experiments.

Plasmid clones, cDNA probes, and RNA transcripts. The EGF-R plasmid cDNA contained the entire coding region of EGF-R and the first 131 nt of the 3'-UTR in pBluescript II KS (pBlue) (14). For Northern blots, a 1-kb BglII/HindIII fragment from the EGF-R coding region was random-prime labeled using [³²P]dCTP (approximately 3,000 Ci/mmol; Amersham, Sydney, Australia). A 1.1-kb 18S rRNA probe was used as a loading control. BglII, HindIII, and BamHI fragments of EGF-R cDNA were subcloned into the HpaI site of the Rous sarcoma virus-luciferase (RSV-Luc) expression vector (Fig. 1B). Additional sequences in the 3'-UTR were amplified from MDA-MB-468 cell total RNA using reverse transcriptase (RT)-mediated PCR (RT-PCR) and cloned into RSV-Luc (Fig. 1B). The c-fos AU-rich element which generates an unstable mRNA was also cloned into RSV-Luc and used in transfections (68). A plasmid encoding RSV--galactosidase (RSV--Gal; ATCC) was cotransfected as an internal control. The authenticity of all constructs used for transfections was verified by sequencing.

View larger version (55K): [in this window] [in a new window]   FIG. 1.   Schematic of EGF-R cDNA, RSV-Luc/EGF-R chimeric plasmid constructs, pBlue EGF-R constructs, and DNA oligomers. (A) Schematic representation of the human EGF-R cDNA indicating restriction enzyme sites and AU-rich 3'-UTR sequences. (B) RSV-Luc/EGF-R constructs generated for transfections by subcloning various portions of the EGF-R into the HpaI site in the RSV-Luc expression vector. The length of each EGF-R insert is indicated in nucleotides. EGF-R1A, nt 4017 to 4089; EGF-R2A, nt 4116 to 4190; EGF-R22, nt 4116 to 4153; EGF-R23, nt 4154 to 4190. (C) pBlue EGF-R constructs containing specific sequences of the 3'-UTR of the EGF-R cDNA. The AU-rich elements within the 3'-UTR are underlined. EGF-R1A, EGF-R2A, EGF-R22, and EGF-R23 are clones derived from the same portions of the 3'-UTR as for the RSV-Luc/EGF-R constructs in panel B. EGF-R22G and EGF-R23G are mutant clones in which three thymidine nucleotides in the AU-extended pentamers have been replaced by guanine (underlined). EGF-R9T contains a consensus nonamer AU-rich sequence. (D) Schematic of EGF-R2A showing location and orientation of competitor oligonucleotides. EGF-R22a.s and EGF-R22a.as represent oligonucleotides spanning the 5' 20 nt of EGF-R22 in the sense and antisense orientations, respectively; EGF-R22b.s and EGF-R22b.as represent oligonucleotides to the remaining 14 nt of EGF-R22; EGF-R22c.s and EGF-R22c.as represent the first 12 nt of EGF-R22 (similarly for EGF-R23a.s, EGF-R23a.as, EGF-R23b.s, and EGF-R23b.as).

RNA isolation and northern analysis. Cells were solubilized in 4 M guanidinium isothiocyanate, and total RNA was isolated by the method of Chomczynski and Sacchi (11). RNA (10 to 15 µg per sample) was size fractionated on a 1% agarose-formaldehyde gel and transferred to a Hybond-N+ membrane (Amersham). RNA was UV cross-linked to the membrane, which was prehybridized for 4 h at 42°C in a buffer containing 50% formamide, 0.75 M NaCl, 0.075 M sodium citrate (pH 7.0), 5× Denhardt's solution, 1% sodium dodecyl sulfate (SDS), and 200 mg of salmon sperm DNA/ml and then hybridized in the same buffer overnight at 42°C with ³²P-labeled EGF-R cDNA probe at 10⁶ cpm/ml. The membrane was washed with 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) containing 0.1% SDS and then with 0.2× SSC containing 0.1% SDS at 22°C. Membranes were analyzed by autoradiography using Kodak EM-1 film, imaged with a PhosphorImager (Molecular Dynamics, Sunnyvale, Calif.), and quantified using ImageQuant software. In all experiments, an 18S rRNA probe was used for normalization.

mRNA turnover studies. Cells were grown to 70 to 80% confluence and not treated or treated with EGF (25 ng/ml, 4 nM; Promega, Madison, Wis.) for 8 h, followed by addition of the transcription inhibitor actinomycin D (Act D; 7.5 mg/ml; Sigma, St. Louis, Mo.). Total RNA was isolated from the cells at 0-, 2-, 4-, 8-, and 12-h time intervals after addition of Act D and subjected to Northern analysis as described earlier. EGF-R mRNA half-life was determined by linear regression analysis.

Nuclear run-on transcription assay. Exponentially growing MDA-MB-468 and BT-20 cells were treated with EGF (25 ng/ml, 4 nM) for 6 to 8 h. Nuclei were isolated as described previously (3), rapidly frozen, and stored at 85°C. The transcription assay was performed as described previously (65). Briefly, for the transcription assay, the nuclei were thawed on ice, resuspended in 100 µl of reaction buffer (10 mM Tris-HCl [pH 8.0], 5 mM MgCl₂, 300 mM KCl, 5 mM dithiothreitol, 0.5 mM each ATP, CTP, and GTP, 100 µCi of [³²P]UTP [800 Ci/mmol; NEN-Du Pont]), and incubated at 30°C for 30 min. Labeled RNA was isolated and hybridized to nitrocellulose filters onto which 5 µg of EGF-R and 18S cDNAs had been blotted. Filters were washed, analyzed by autoradiography, and quantified using a PhosphorImager.

Immunoblot assay for EGF-R protein. Cells were treated with EGF (25 ng/ml, 4 nM) for 8 h, harvested, and lysed in ice-cold lysis buffer (1% Triton X-100, 20 mM Tris-HCl [pH 7.4], 1 mM EDTA). After 10 min on ice, the lysate was centrifuged at 750 × g (Eppendorf model 5415C centrifuge) for 10 min at 4°C, after which the supernatant was recovered and stored at 85°C. Total protein concentration of the lysate was determined using the Bio-Rad protein assay; the proteins (5 to 10 µg/lane) were separated by polyacrylamide gel electrophoresis (PAGE) on SDS-6% polyacrylamide gels, and subsequently transferred to nitrocellulose. The membrane was blocked with 5% nonfat dried milk in TBS-T (20 mM Tris-HCl [pH 7.4], 150 mM NaCl, 0.1% Tween 20) at 22°C for 1 h prior to incubation with an EGF-R polyclonal antibody (1:2,000; Upstate Biotechnology, Lake Placid, N.Y.) for 1 to 2 h at 22°C and horseradish peroxidase-conjugated anti-sheep goat immunoglobulin G (1:2,000; Amersham); the EGF-R protein was visualized by ECL (Amersham) and subjected to autoradiography. The blots were also probed with a human polyclonal actin antibody (1:500; Santa Cruz Biotechnology, Santa Cruz, Calif.) as a control for loading. The 170-kDa EGF-R protein band was quantified using a Kodak DCS-420c digital camera and ImageQuant software.

Transfection, Luc, and -Gal assays. MDA-MB-468 cells (70 to 80% confluent) were transfected by electroporation with 10 µg of RSV-Luc or RSV-Luc EGF-R and 6 µg of RSV--gal as a control. After electroporation, cells were cultured in medium in the presence or absence of EGF (25 ng/ml) for 6 h prior to lysate extraction and assays for luciferase (Luc) and -galactosidase (-Gal) activity. Lysates were prepared by harvesting the cells from the plate in phosphate-buffered saline (PBS). The mixture was centrifuged at 450 × g (Jouan model C3-12 centrifuge) for 5 min, the supernatant was removed, and the cell pellet was resuspended in 250 µl of lysis buffer (125 mM Tris [pH 7.6], 0.5% Triton X-100). The solution was centrifuged at 16,500 × g (Eppendorf model 5415C centrifuge) for 10 min at 4°C, and the supernatant was used in the Luc assay. For each Luc assay, 50 µl of the cell lysate was used with 250 µl of assay buffer (25 mM glycylglycine [pH 7.8], 15 mM MgSO₄). Samples in triplicate were analyzed in an automated Berthold luminometer, with 100 µl of luciferin mixture (containing luciferin [50 mg/ml; Promega], 5 mM ATP, and assay buffer) added to each sample. -Gal activity was determined for each extract as described elsewhere (52).

LightCycler PCR assay. MDA-MB-468 cells grown to ~80% confluence were transfected using Fugene (Roche, Indianapolis, Ind.) with RSV-Luc, EGF-R1A, or EGF-R2A plasmid and with -globin as an internal cotransfection control. Total RNA was isolated from the cells, and 5 µg was reverse transcribed using avian myeloblastosis virus RT (Promega) in a mixture containing 5 mM MgCl₂, 1 mM dNTPs, 1 U of RNasin/ml, 10 U of avian myeloblastosis virus RT, and 50 ng of oligo(dT) for 60 min at 42°C. The cDNA was then used in a real-time quantitation assay using a LightCycler (Roche Molecular Dynamics) (62). A total reaction volume of 20 µl contained 2 µl of cDNA from the RT reaction, 2 µl (10 pmol/µl) of Luc antisense and 2 µl (10 pmol/µl) of Luc sense primers, 1 µl of 40 mM dNTPs, 2 µl of 10× PCR buffer, 1 µl of bovine serum albumin (25 mg/ml), 2 µl of 5× Sybr green (Sigma), 1 µl of Taq polymerase (AmpliTaq; 1U/µl; Perkin-Elmer), and 7 µl of water. The Luc primers were 5' TAC TGG GAC GAA GAC GAA CAC 3' (sense) and 5' CAC GCC CGC GTC GAA GAT GTT 3' (antisense); the -globin primers were 5' GAG TCC TTT GGG GAC CTG TCC 3' (sense) and 5' GAA GTT CTC AGG ATC CAC GTG 3' (antisense).

Preparation of cytoplasmic extracts. MDA-MB-468 cells were grown to 70 to 80% confluence in 100-mm-diameter culture dishes. Cytoplasmic extracts were prepared as described elsewhere (58). Medium was replenished 12 to 24 h prior to ligand treatment with EGF (25 ng/ml) or PMA (50 ng/ml). Cells were scraped from the culture dishes with chilled PBS and centrifuged for 4 min at 450 × g (Jouan model C3-12 centrifuge) at 4°C; the supernatant was discarded. The cells were washed with cold PBS and centrifuged at 450 × g (Jouan model C3-12 centrifuge) for 4 min. Cell pellets were incubated with cytoplasmic extract buffer (10 mM HEPES, 3 mM MgCl₂, 40 mM KCl, 5% glycerol, 0.2% NP-40, 1 mM dithiothreitol), with freshly added protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 10 µg of leupeptin/ml, 2 µg of aprotinin/ml [Roche]), for 20 min and then centrifuged for 2 min at 12,100 × g (Eppendorf model 5415C centrifuge) and 4°C; the supernatant was snap-frozen in liquid nitrogen. Extracts from MDA-MB-231, MCF7, HT1080, SK-BR-3, and LNCaP cells were all processed in a similar manner. The HT1080 fibrosarcoma line required extra steps for cell lysis: heating at 37°C for 5 min followed by snap-freezing in liquid nitrogen for 2 min, repeated five times. Protein concentrations were determined using a Bio-Rad protein assay kit.

REMSA. Binding reactions were performed as described elsewhere (35, 36, 58) with 5 µg of cytoplasmic extract and 10⁵ cpm of RNA (~10 to 20 pg). Briefly, following incubation at 22°C for 30 min, 0.3 U of RNase T₁ (Roche) was added for 10 min, followed by the addition of heparin (final concentration, 50 mg/ml; Sigma) for 10 min. Samples were subjected to electrophoresis on a 4% native acrylamide gel (acrylamide/bisacrylamide ratio of 36:1), dried, and analyzed with a PhosphorImager followed by autoradiography as described elsewhere (58).

UVXL of RNA-protein complexes. RNA-protein binding reactions were carried out as described above, using 20 to 30 µg of cytoplasmic extract and 1.5 × 10⁵ cpm of RNA (15 to 30 pg) of ³²P-riboprobe (35, 55, 58). Following the addition of heparin, samples were placed on ice in a microtiter tray and UV irradiated for 10 to 15 min, 1 cm below the Stratalinker UV light source (240-nm UV bulb; Stratagene, La Jolla, Calif.). Samples were then incubated with RNase A (100 mg/ml; Roche) at 37°C for 15 min. The samples were boiled for 3 min in SDS sample buffer; RNA-protein complexes were separated by SDS-PAGE on 8 to 10% gels and analyzed by autoradiography. ¹⁴C-labeled Rainbow molecular weight markers (Amersham) were used as standards.

Secondary structure prediction of EGF-R mRNA. Nucleic acid sequence was obtained from GenBank accession number X00588, and stem-loop structure prediction plots were modeled using foldRNA and Squiggles programs (webANGIS) on a Macintosh workstation.

Statistical analysis. Statistical analysis was performed using General Linear Models. Analysis of variance was performed to assess the significance of treatment (EGF) over different time intervals from Act D chase assays. Where appropriate, means comparisons at specific time points were made using Student's t test. A P value less than or equal to 0.05 was considered statistically significant.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

EGF increases EGF-R mRNA and protein levels in MDA-MB-468 and BT-20 cells. In preliminary experiments, we evaluated the biological potency of EGF by examining levels of phosphotyrosine activation of the EGF-R. Incubation of MDA-MB-468 cells for 15 min with EGF at 12.5 or 25 ng/ml (~4 nM) induced substantial EGF-R tyrosine phosphorylation (data not shown), so that in subsequent experiments we used either 12.5 or 25 ng of EGF/ml.

View larger version (57K): [in this window] [in a new window]   FIG. 2.   EGF upregulates EGF-R mRNA and protein levels in MDA-MB-468 and BT-20 cells. (A and B) Northern blot analysis of total RNA extracted from MDA-MB-468 (A) and BT-20 (B) cells after treatment with EGF (25 ng/ml, ~4 nM) as indicated. RNA was fractionated on 1% agarose gels, transferred to a nylon membrane, and hybridized to a 32P-labeled EGF-R-specific cDNA probe. Each blot was normalized using a 32P-labeled 18S rRNA probe. Arrows denote EGF-R mRNA species at 10 and 5.6 kb and 18S rRNA. Blots were quantitated using a PhosphorImager and autoradiography. Each value is shown relative to an arbitrary value of 1 at time zero and is representative of at least three separate experiments performed in duplicate. Error bars represent standard errors of the means. Black and grey columns represent 10- and 5.6-kb EGF-R mRNA species, respectively. (C and D) Immunoblot analysis of EGF-R protein levels in MDA-MB-468 (C) and BT-20 (D) cells following treatment with EGF (25 ng/ml) for 8 h. Cells were lysed, and 5 µg (MDA-MB-468 cells) or 10 µg (BT-20 cells) of protein lysate was electrophoresed on SDS-6% polyacrylamide gels, transferred to nitrocellulose membranes, and analyzed for EGF-R protein expression using a sheep polyclonal anti-human EGF-R antibody and ECL detection (see Materials and Methods). The size of the EGF-R protein is ~170 kDa. The bar graphs depict the quantified changes in EGF-R protein after EGF treatment relative to an actin control (data not shown) and are representative of at least three different experiments performed in duplicate. Error bars represent standard errors of the means. Rel., relative.

EGF stabilizes EGF-R mRNA and increases EGF-R mRNA transcription in MDA-MB-468 and BT-20 cells. The contribution of posttranscriptional events to the regulation of EGF-R mRNA expression in breast cancer cells has not previously been addressed. Act D chase studies were used to determine the half-life of EGF-R mRNA in each cell line (Fig. 3C and D). The basal half-lives of EGF-R mRNA were ~6.5 and 8.5 h in MDA-MB-468 and BT-20 cells, respectively. In both cell lines, EGF stabilized EGF-R mRNA, prolonging the half-life to greater than 14 h (>2-fold). Nuclear run-on assays demonstrated that EGF upregulated transcriptional activity 2-fold in each cell line (Fig. 3A and B). Thus, the increase in EGF-R mRNA (~3-fold) induced by EGF in each cell line resulted from a combined increase in mRNA stability and transcriptional upregulation, with the former appearing predominant.

View larger version (34K): [in this window] [in a new window]   FIG. 3.   EGF increases EGF-R gene transcription and stabilizes EGF-R mRNA in MDA-MB-468 and BT-20 cells. (A and B) Transcriptional analysis of MDA-MB-468 (A) and BT-20 (B) cells after treatment with EGF (25 ng/ml) for 6 h. EGF-R transcription rates were determined in isolated nuclei by run-on transcription assays, analyzed by autoradiography, and quantitated using a PhosphorImager. Results were normalized relative to 18S RNA transcription levels and are representative of two separate experiments performed in duplicate. (C and D) Act D chase studies in MDA-MB-468 (C) and BT-20 (D) cells. Cells were grown to 70 to 80% confluence, treated with EGF (25 ng/ml) for 8 h, and then chased with 7.5 µg of Act D/ml. Total RNA was extracted from the cells at 0, 2, 4, 8, and 12 h after Act D treatment and analyzed by Northern blotting using a 32P-labeled cDNA EGF-R-specific probe as in Fig. 2. The 10-kb species of EGF-R mRNA was normalized against 18S RNA (image not shown) and quantitated using a PhosphorImager. Half-lives were determined by linear regression analysis. The graph at the bottom shows composite data from three separate experiments performed in duplicate. Rel., relative. Means comparisons: (a) P < 0.05; (b) P < 0.02.

Identification of cis-acting elements in EGF-R mRNA 3'-UTR. To localize the cis element(s) contributing to the regulation of EGF-R mRNA stability, we developed a transient transfection system in MDA-MB-468 breast cancer cells. A set of plasmids containing different portions of the EGF-R mRNA inserted downstream of the Luc coding region in RSV-Luc was used (Fig. 1B). Initially we examined the change in reporter activity induced by various portions of the EGF-R mRNA. Interestingly, basal Luc activity was modified with the RSV-Luc/3'-UTR construct to ~60 to 70% of control levels (Fig. 4A). However, no effect was seen with any of the constructs that contained 5'-UTR or coding region sequences of EGF-R mRNA. This included sequences containing the 5' end of the EGF-R mRNA (EGF-R2) and several smaller clones from within that sequence, EGF-R3, -4, -5, and -6 (Fig. 4A). Reporter activity of the RSV-Luc/c-fos ARE plasmid was reduced to ~10% of the basal level. As the c-fos ARE sequence is known to have a short half-life in other systems (25), this observation was consistent with it conferring rapid decay and shortening the mRNA half-life of the Luc reporter. Taken together, these results suggested that the 3'-UTR was the major contributor to the regulation of EGF-R mRNA turnover.

View larger version (39K): [in this window] [in a new window]   FIG. 4.   The 3'-UTR of EGF-R mRNA contains an EGF-regulated 260-nt ARE that regulates Luc activity in MDA-MB-468 cells. (A and B) MDA-MB-468 cells were transfected by electroporation with 10 µg of one of several RSV-Luc constructs, either RSV-Luc or an RSV-Luc/EGF-R hybrid (Fig. 1B), or c-fos ARE and 6 µg of RSV--Gal. Cells were harvested 24 h later, and the lysate was analyzed in Luc and -Gal assays (see Materials and Methods). The luciferase results were normalized to the -Gal reading and plotted. The graphs are representative of at least five separate experiments performed in triplicate. Error bars denote standard deviation. (C and D) Analysis of the AU-rich sequences within the 260-nt EGF-R mRNA cis-acting element. MDA-MB-468 cells were transfected as for panels A and B, EGF (25 ng/ml) was added to the cells 24 h posttransfection, cells were harvested 8 h later, and analysis was performed as described above.

View larger version (22K): [in this window] [in a new window]   FIG. 5.   The EGF-R2A mRNA sequence acts as a cis destabilizing and EGF-regulated element. MDA-MB-468 cells were transfected with pRSV-Luc or pRSV-Luc/EGF-R2A in triplicate. After treatment with EGF (25 ng/ml) for 6 h, the cells were treated with Act D (7.5 µg/ml), and RNA was extracted at 0, 2, 4, and 8 h followed by RT-PCR for Luc mRNA using a LightCycler (see Materials and Methods). Samples were normalized using a cotransfected control of -globin. Half-lives were determined by linear regression analysis. The graph is representative of data from four independent experiments in duplicate. Rel., relative.

MDA-MB-468 cells contain cytoplasmic proteins that bind specifically to the EGF-R mRNA cis element. To investigate whether the 260-nt cis-acting EGF-R mRNA element was a target for RNA-binding proteins, we generated riboprobes to the EGF-R1A and EGF-R2A sequences (Fig. 1C, EGF-R1A and EGF-R2A, respectively) for use in REMSA. Cytoplasmic proteins were identified from MDA-MB-468 cells that bound specifically to the EGF-R1A and EGF-R2A riboprobes (Fig. 6A and B). Addition of ~100-fold excess unlabeled pBlue competitor RNA did not diminish the formation of the RNA-protein complex (Fig. 6A and B, lanes 5). However, addition of excess unlabeled EGF-R1A (Fig. 6A, lane 2) or EGF-R2A (Fig. 6B, lane 2) virtually abolished RNA-protein complex formation, confirming binding specificity for these transcripts. Addition of excess unlabeled EGF-R2A had no significant effect on binding of the EGF-R1A complex (Fig. 6A, lane 3). Similarly, excess unlabeled EGF-R1A had only a marginal effect on the EGF-R2A complex (Fig. 6B, lane 3). No RNA-protein complex was observed with ³²P-pBlue riboprobe (Fig. 6A, lane 6).

View larger version (43K): [in this window] [in a new window]   FIG. 6.   Proteins from MDA-MB-468 cells bind specifically to EGF-R mRNA. (A) MDA-MB-468 cytoplasmic extract (~5 µg) was incubated with either 32P-labeled EGF-R1A (lanes 1 to 5) or pBlue riboprobe (lane 6), and REMSA was performed as described in Materials and Methods. In lanes 2 to 5, ~100-fold excess unlabeled (cold) competitor RNA was added to the labeled EGF-R1A probe for 10 min at 22°C before addition of extract and REMSA. Arrow denotes RNA-protein complex (RPC). (B) MDA-MB-468 extract (~5 µg) was incubated with either 32P-labeled EGF-R2A (lanes 1 to 5) or EGF-R9T (lanes 6 to 8), and REMSA was performed as above. In lanes 2 to 5, 7, and 8, ~100-fold excess of various unlabeled competitor RNA was added to the labeled probe prior to the extract. Arrows denote RNA-protein complexes (RPC). (C) MDA-MB-468 extract was incubated with labeled EGF-R2A in the absence (lane 1) or presence of increasing concentrations (1, 10, and 50 ng) of unlabeled homopolymers RNA, poly(C) (lanes 2 to 4), poly(A) (lanes 5 to 7), and poly(U) (lanes 8 to 10). Arrow denotes RNA-protein complex (RPC). The data are representative of at least three individual experiments.

View larger version (52K): [in this window] [in a new window]   FIG. 7.   EGF-R mRNA-binding proteins are present in a variety of cancer cells and are regulated by PMA and EGF. (A and B) REMSA using 32P-labeled EGF-R1A (A) or EGF-R2A (B) and extracts from a variety of human cancer cells, including prostate (LNCaP), breast (SK-BR-3, MDA-MB-231, MCF7, and MDA-MB-468), and fibrosarcoma (HT1080) cells. Arrows denote RNA-protein complexes (RPC). (C and D) REMSA using 32P-labeled EGF-R2A and extracts isolated from MDA-MB-468 cells treated with 100 nM PMA (C) or 4 nM EGF (D) at the time intervals shown. The top panel shows the REMSA; the lower graph depicts the level of binding relative to a control at each time point and normalized to an arbitrary value at 100% at time zero. The data are representative of four experiments performed in duplicate. Error bars represent standard errors of the means. Means comparisons: (a) P < 0.01; (b) P < 0.05.

View larger version (30K): [in this window] [in a new window]   FIG. 8.   EGF-R mRNA is a target for multiple RNA-binding proteins. (A) Extract from MDA-MB-468 cells was incubated with either 32P-labeled EGF-R1A (lanes 2 and 3) or EGF-R2A (lanes 4 and 5), followed by addition of RNase T1 and heparin. The mixture was then UV cross-linked for 15 min (see Materials and Methods) before digestion with RNase A, SDS-PAGE (7.5% gel), and analysis by PhosphorImager. In lanes 3 and 5, ~100-fold excess unlabeled (cold) RNA was added (EGF-R1A and EGF-R2A, respectively). 14C-labeled size markers are indicated in lane 1. (B and C) Stem-loop secondary structure plots of EGF-R1A (B) and EGF-R2A (C), using the Genetics Computer Group Squiggles program (see Materials and Methods). The E values for EGF-R1A and EGF-R2A were 6.4 and 9.2 kJ mol1, respectively. Locations of the AU pentamers (B) and AU-extended pentamers (C) are indicated by the thick lines.

1

1

Mutations within EGF-R2A radically alter RNA protein binding. A set of short sense and antisense DNA oligomers (Fig. 1D) was used to identify which half of EGF-R2A was critical for binding. The predominant binding site for EGF-R22 resided in the 5' end of the sequence within which the AU-extended pentamer was located (Fig. 9A, lanes 5 and 6 versus lane 7). Because of the marked effect of antisense DNA oligomer 6 with EGF-R22, we generated oligomers to the 5' 12 nt of EGF-R22 (oligomers 4 and 8), which did not include the AU-rich sequence. We found that sequence antisense to the 12-nt region abrogated binding completely, while its sense counterpart had no effect (Fig. 9A, lanes 8 and 4, respectively). A similar but less striking pattern was observed for EGF-R23: a major reduction in binding was observed when the 3'-end antisense oligomer was used (Fig. 9B, compare lanes 3 and 6).

View larger version (38K): [in this window] [in a new window]   FIG. 9.   Analysis of the EGF-R mRNA binding motif using DNA oligomers and AU-extended pentamer mutants. (A and B) REMSA using MDA-MB-468 extracts and 32P-labeled EGF-R22 (A) or EGF-R23 (B). Before addition of cell extract, a sense or antisense DNA oligomer was incubated with the labeled probe (1:1 molar ratio) for 10 min at 22°C. In panel A, sense oligomers were added to samples in lanes 1 to 4, and antisense oligomers were added to samples in lanes 5 to 8. In panel B, sense oligomers are in lanes 1 to 3, and antisense oligomers are in lanes 4 to 6. The upper portion of each panel illustrates the length and location of each oligomer relative to the labeled probe. The oligomer, probe, and extract mixture was treated as above for REMSA and analyzed by PhosphorImager. The sequence and orientation of each oligonucleotide and probe are shown in Fig. 1D. (C and D) Stem-loop secondary structure plots of EGF-R22 and EGF-R23 and their respective triple-G AU-extended pentamer mutations. The thin line denotes the location of the mutation. The E values for EGF-R22, EGF-R22G, EGF-R23, and EGF-R23G were 3.5, 0.7, 4.0, and 5.7 kJ mol1, respectively. (E and F) MDA-MB-468 extract was incubated with either 32P-labeled EGF-R22 or EGF-R22G (E) or EGF-R23 or EGF-R23G (F). REMSA was performed as for Fig. 6 and analyzed by PhosphorImager. RPC, RNA-protein complex.

View larger version (63K): [in this window] [in a new window]   FIG. 10.   UVXL analysis of EGF-R mRNA mutants. MDA-MB-468 extract was incubated with one of several different 32P-labeled probes (Fig. 1C), UV cross-linked, resolved by SDS-PAGE on an 8.5% gel, and analyzed by PhosphorImager. Lane 1 shows 14C-labeled molecular mass standards. Arrows a, b, and c denote specific RNA-protein complexes discussed in the text.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

These data establish the importance of mRNA stability in the regulation of EGF-R gene expression by EGF in ER-negative breast cancer cell lines that overexpress EGF-R. In both MDA-MB-468 and BT-20 breast cancer cells, the EGF-induced stabilization of EGF-R mRNA (>2-fold) was accompanied by a ~2-fold increase in total EGF-R protein. Taken together with previous studies in which we reported EGF-induced stabilization of EGF-R mRNA in epidermoid KB cancer cells (~2-fold) (45) and in human prostate cancer cells (LNCaP and DU-145; ~2-fold) (39), it is apparent that this effect on EGF-R mRNA stability is shared across many different cancer cell lines and is of primary importance in the regulation of EGF-R gene expression.

Despite increasing evidence for tight control of EGF-R expression at the level of mRNA decay, little is known of the molecular mechanisms involved. In particular, there have been no attempts to locate and characterize a cis-acting element(s) within the EGF-R mRNA or the trans-acting protein(s) that binds these regions. The 3'-UTR of the EGF-R mRNA contains four major AU-rich regions, two within a ~260-nt sequence (EGF-R8) at the 5' end and two smaller regions further toward the 3' end. Our transfection studies indicated that the 5' 260-nt AU-rich (66%) region contains a complex cis-acting element. Interestingly, the two other AU-rich regions in the 3'-UTR do not appear to be involved in the regulation of reporter activity. The location of the cis element within the 3'-UTR is typical of most instability sequences described to date (1, 2, 23). However, the EGF-R mRNA cis-acting element is distinctive in that it is devoid of a consensus nonamer (34, 68); instead it has two classical pentamers (AUUUA) (9, 23, 42, 43) and two extended pentamers (AUUUUUA), residing in separate discrete ~75-nt regions, EGF-R1A and EGF-R2A, respectively. Each of these regions is capable of independently regulating basal and EGF-stimulated Luc mRNA turnover, suggesting that both components of the cis element (EGF-R1A and EGF-R2A) play an important role in the EGF-induced stabilization of EGF-R mRNA. Based on our observations, we presume that EGF-R1A and EGF-R2A function in a cooperative manner to regulate EGF-R mRNA turnover. It is important to note, however, that neither the EGF-R1A nor the EGF-R2A sequence reduced reporter activity to the level of the c-fos ARE sequence, suggesting that these EGF-R AU-rich sequences are less potent than the well-characterized destabilizing nonamers in the c-fos ARE (34, 68). The EGF-R mRNA cis element has intermediate capacity to confer instability to a heterologous reporter, consistent with the intermediate basal endogenous EGF-R mRNA half-life of ~6.5 to 9 h documented for these two cell lines. Interestingly, we found the use of the LightCycler significantly facilitated analysis of Luc mRNA half-life for turnover studies and also verified that in these cells, the Luc reporter protein assay was an excellent surrogate for Luc mRNA, as the two measurements closely followed each other.

Analysis of our mutant constructs in transfections provided insight into the potential contribution of each AU-extended pentamer to the regulation of EGF-R mRNA turnover. Interestingly, mutation of the 5' extended pentamer in EGF-R2A (EGF-R22G) did not alter reporter activity in the presence or absence of EGF. However, mutation of the 3' extended pentamer (EGF-R23G) upregulated basal reporter activity to ~50% of the wild-type level (EGF-R23) while having a similar effect on EGF-induced reporter activity (~2-fold increase). These data suggest that the 5' extended pentamer does not contribute to the regulation of EGF-R mRNA turnover, while the 3' extended pentamer is likely to be involved in regulating basal, but not EGF-induced, changes in EGF-R mRNA turnover.

Our REMSA and UVXL studies identified novel cytoplasmic proteins from ER-negative breast cancer cells that bound specifically to the 260-nt cis-acting element of EGF-R mRNA. These are the first proteins from breast cancer cells to be described that target the EGF-R mRNA. Interestingly, these proteins appear to be widely distributed in a variety of human breast and prostate cancer, as well as fibrosarcoma, cell lines. Taken together with our detection of a similar complex in epidermoid KB cells (45), these data support a central role for these proteins in the regulation of EGF-R gene expression in cancer cells. Interestingly, binding was detected in breast cancer cell lines that were ER positive, such as MCF7 cells, and also in ER-negative cells, such as MDA-MB-231 and MDA-MB-468 cells. Furthermore, the intensity of binding to the cis-acting element appeared unrelated to the cellular level of EGF-R expression.

We found that PMA and EGF both rapidly downregulated binding activity of these proteins to the EGF-R2A probe. The decrease in binding was detectable within 5 min and may be due to a change in the phosphorylation state of one or more of these proteins that modifies binding. The binding activity of several other AUBPs has been shown to be regulated by PMA (4). However, these have been predominantly proteins that target classic AREs comprising nonamers, such as AU-rich binding factor (AUBF) (43). In addition, phosphorylation of the iron regulatory protein by PMA induces rapid changes in binding to the iron-responsive element (54). Given that PMA and EGF both activate the protein kinase C pathway in these cells, this finding provides a ready explanation for the similar effects observed in REMSA with the two ligands. Further studies will be required to determine which isotype of protein kinase C mediates these effects, whether other downstream effectors of EGF signaling regulate binding, whether these proteins are ubiquitously expressed, and when they appear during development.

Multiple mRNA-binding proteins that target AREs in early response genes and cytokines have been described in the past few years. The vast majority of these proteins have molecular masses in the range of 30 to 50 kDa. The first AUBP described that bound an ARE was termed AUBF (~44 kDa) (42). Binding of this protein (a three-subunit complex) was upregulated by PMA and calcium ionophores (43). Subsequently, other ARE-binding proteins have been characterized, and some have been cloned. AUF1, a 37-kDa protein that contains two RRMs, was recently cloned (66). Three other proteins, termed AU-A, AU-B, and AU-C, that bound to AREs were described 2 years earlier (5). Several other studies have identified proteins of 82, 71, 66, and 37 kDa (66), as well as 70, 45, 40, 38, and 32.5 kDa (50), that target AREs within the c-fos and GM-CSF mRNAs. Several of these proteins appear to be cross-reactive with AUF1 antibodies (45, 40, and 38 kDa) or hnRNP antibodies (40 and 38 kDa) (50). Most recently, HuR, a member of the elav family of RRM-containing RNA-binding proteins, was cloned (41). HuR is a ubiquitously expressed ~32- to 38-kDa protein that binds with high affinity to AREs present in a number of early response genes. HuR is remarkable in that it shuttles between the nucleus and cytoplasm (18) and has been shown to be an important component of the mRNA destabilization machinery (48). Almost all of the studies involving the above proteins have utilized sequences based on nonamer-pentamer repeats, and none to our knowledge have focused on the characterization of proteins that bind to extended AU pentamers. In this context, several lines of evidence suggest that the proteins we identified are not classical AU-rich nonamer-binding proteins. First, the AU-rich sequences in EGF-R1A and EGF-R2A are pentamers and extended pentamers and do not contain any nonamers. Second, antibodies to two well characterized AUBFs, HuR and AUF1, did not supershift the complex with either probe. Third, competition studies with excess unlabeled nonamer binding probe (EGF-R9T) riboprobe had little effect on the EGF-R1A and EGF-R2A probes. Last, the sizes of the proteins detected (~55 to 80 kDa) were in a range above that of most of the AUBFs described to date (~30 to 50 kDa) (7, 49, 54, 60).

The EGF-R2A sequence was of great interest because of the complex set of proteins binding on UVXL, the presence of two extended pentamers, and its stable stem-loop structure. Interestingly, UVXL showed that the 5' end of EGF-R2A mRNA, EGF-R22, appeared to bind only one of the proteins observed with the EGF-R2A. In contrast, the 3' end of EGF-R2A, EGF-R23, appeared to contribute the majority of binding, as we detected all of the RNA-protein complexes observed with the EGF-R2A probe, as well as additional smaller complexes at ~35 and 45 kDa. To further define the major binding motif in EGF-R2A, we used a combination of DNA oligomers and mutant probes. The oligomer data suggested, somewhat unexpectedly, that the 5' AU-extended pentamer in EGF-R2A was unlikely to be involved in protein binding. Instead, the 12 nt 5' of the extended pentamer was shown to be critical for binding a single protein of ~55 kDa. These findings were corroborated when we showed that the triple-G mutant, designed to disrupt the 5' AU-extended pentamer and significantly modify the predicted stem-loop structure (EGF-R22G), bound with increased affinity to the protein extract in both REMSA and, in particular, UVXL. Thus, introduction of the mutation in the 5' extended pentamer of EGF-R2A induced changes that increased rather than abolished binding. Studies using the EGF-R23 probe were quite different. The DNA oligomer data with portions of the EGF-R23 sequence suggested that the 3' end of the sequence contained the major protein binding site. When we examined the binding profile of EGF-R23G, which was designed to disrupt the 3' extended pentamer, we found that two RNA-protein complexes (~35 and 45 kDa) were abolished, and there was increased binding of the other major protein moieties at ~55 and 80 kDa. This confirmed a direct role for the extended pentamer in binding of at least two of the several proteins binding to EGF-R23. However, it also indicated that a significant proportion of protein binding to the EGF-R23 sequence occurs outside the extended AU pentamer region. Thus, there is differential binding of proteins to the two extended pentamers within the EGF-R2A sequence, and the majority of binding to EGF-R2A is directed to non-AU-rich sequence. These data suggest a model in which the 5' extended AU pentamer is not involved in the regulation of EGF-R mRNA stability or binding of trans-acting factors, while the 3' extended pentamer plays an important role in regulating basal mRNA stability, in part, through binding of specific trans-acting factors.

In summary, these data provide compelling evidence for a central role of mRNA turnover in the regulation of EGF-R gene expression in ER-negative breast cancer cells. The control of EGF-R mRNA stability is a complex process involving regulated interactions between a novel 3'-UTR cis element and multiple trans-acting protein factors. The cis element contains AU pentamers and extended pentamers and is a target for EGF-regulated trans-acting proteins. Interestingly, only one of the extended pentamers in this region is likely to contribute, in part, to the regulation of EGF-R mRNA stability and binding of specific proteins. These EGF-R mRNA-binding proteins are widely distributed in multiple cancer cell types, and their size and binding characteristics suggest that they are not classical ARE-binding proteins. The identification of a novel cis element in EGF-R mRNA will allow detailed assessment of the mechanisms governing mRNA turnover in a variety of relevant cell types in which overexpression of EGF-R is linked with proliferation. Furthermore, the isolation and cloning of proteins that target the cis element should provide valuable insight into the regulation of EGF-R mRNA expression in breast and other cancer cells and may provide the basis for designing strategies to disrupt EGF-R expression and consequently alter rates of cellular proliferation and growth.