AIB1 gene amplification and the instability of polyQ encoding sequence in breast cancer cell lines

Primary breast tumor specimens with matching normal breast tissue samples were obtained from Fu Jen Catholic University, and Cardinal Tien Hospital, Taiwan, after surgical removal of the tumor according to the IRB approved protocol. The ER positive breast cancer cell lines were obtained from Georgetown University Lombardi Comprehensive Cancer Center Tissue Culture shared resource and American Type Cell Culture. A total of 25 cancer and 4 non-cancer breast tissue cell lines were studied. MCF-7 variants include different passages, MCF-7 p19, MCF-7 p72 and MCF-7 derivatives: LCC1 (selected for growth in vitro without estrogens) [27], LCC2 (selected from LCC1 by treatment with non-steroid antiestrogen 4-OH TAM) [28], LCC9 (selected from LCC1 by treatment with steroid antiestrogen ICI 182,780) [29], LY-2 (resistant to 4-OH TAM, KEO and LY 117018) and R27 (resistant to 4-OH TAM). AK-47 is derived from parental ER positive cell line ZR-75-1 with the loss of expression of ER. LCC6 is a more aggressive form of MDA-MB-435 [30]. A1N4 is a normal breast cell line that is ER negative. DNA from tissues, blood and cell cultures was extracted by salting out method [31].

A region of 439 bp from exon 5 of AIB1 gene was amplified with the forward primer; 5'-CAAGCGATCAAATGAGGGTAG-3' and the reverse primer; 5'-CATTGTTTCATATCTCTGGCG-3'. A fragment of 85 bp from 3' untranslated region of β ₂ -microglobulin gene (β ₂ -M) was amplified with the forward primer; 5'-TGCTGTCTCCATGTTTGATGTATCT-3' and the reverse primer; 5'-TCTCTGCTCCCCACCTCTAAGT-3' [32‐34]. These PCR products were cloned into the pCR 2.1-TOPO vector (Invitrogen). The plasmid DNA was isolated and quantified using the DU640 Spectrophotometer (Beckman, Fullerton, CA, USA). The copy numbers were calculated from absorbance at 260 nm and based on the molecular weight of the resulting plasmid. The plasmid DNAs were serially diluted over four logs to establish the standard curve giving a range from 400,000 to 40 copies/μl. In additional set of experiments the standard curve was constructed using genomic DNA prepared as a pool of equal amounts of blood DNA from 7 control individuals with normal AIB1 copy number. 'Normal' genomic DNA (100 ng/μl) was diluted in water over four logs. Since 1 ng of genomic DNA contains approximately 330 copies of a single copy gene, five standards used range from 33000 to 3.3 copies/μl.

In real-time Q-PCR analysis, the primers used were 5'-GAGTTTCCTGGACAAATGAG-3' (forward) and 5'-CATTGTTTCATATCTCTGGCG-3' (reverse) for AIB1 gene (Exon 5), and the same primers as used for standard DNA preparation for β ₂ -M gene, yielding 134 bp and 85 bp PCR products, respectively. The TaqMan probes were FAM-5' GCCGTATGTTGATGAAAACACCACA 3'-TAMRA and VIC-5' TTGCTCCACAGGTAGCTCTAGGAGG 3'-TAMRA, for AIB1 and β ₂ -M gene respectively, each labeled with FAM or VIC (reporter dye) at the 5' end and TAMRA (quencher dye) at the 3' end. Each 10 μl real time Q-PCR reaction mixture contained 1 × TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA), 10 ng of genomic DNA, 0.3 μM of each primer, and 0.1 μM probe. The actin gene was also used as a reference. However, since the actin gene has multiple homologous copies, the data presented here were referenced to β ₂ -M gene. The amplification was carried out according to the conditions suggested by the manufacturer (initial denaturation at 95°C for 10 min and 40 cycles of 95°C for 15 s and 60°C for 1 min) using an ABI Prism 7700 Sequence Detection System (Applied Biosystems, Foster City, CA). Each measurement was performed in triplicate and the threshold cycle numbers (C_T) were measured. The copy number was generated from the C_T value and standard curve according to previously described procedures [32‐34].

The poly Q containing fragment was amplified by the forward primer F: 5' GTCTTATACCTGGTGTATTG 3' and the reverse primer R: 5' CTGGGGGAAGCAGTCACATTAG 3', yielding a PCR product of 314 bp. The high fidelity amplification was carried out in a 30 μl reaction mixture containing 10 ng of genomic DNA, 0.2 μM of each primer, 1 × HF 2 PCR buffer, dNTPs, and Advantage-HF 2 polymerase according to the manufacturer's recommendation (Clontech Laboratories, Palo Alto, CA). After 1 min of initial denaturation at 94°C, the DNA was amplified by 30 cycles of 45 s at 95°C, 45 s at 55°C and 45 s at 72°C, followed by a final extension at 72°C for 5 min. The PCR products were purified and cloned into pCR2.1-TOPO (Invitrogen) vector according to the manufacturer's protocol. At least 8 clones from each sample were picked for sequencing using BigDye sequencing kit and analyzed on an ABI 377 DNA Sequencer (Applied Biosystems, Foster City, CA). Two primers, F and F2 (5' AGCAGGGTTTTCTTAATGCTC 3') were used for sequencing and loading of reactions onto alternate lanes for easy tracking. The sequence results were analyzed using sequence analysis software version 3.4.

Real-time Q-PCR analysis allows the measurement of actual copy number of AIB1 gene using a single copy β ₂ -microglobulin gene as a reference. From the threshold cycle number and the standard curve, the ratio of the copy number of AIB1 gene to that of β ₂ -M gene can be calculated. This ratio can be used as a measure of the amplification of the AIB1 gene. The average copy number ratio of the AIB1/β ₂ -M in the blood samples from 48 age matched control individuals is determined to be 1.16 ± 0.38. An AIB1/β ₂ -M ratio above 2 SD of the mean (1.16 + 2 × 0.38 = 1.92) is defined as truly amplified. In addition, all measurements were repeated using a pool of normal genomic DNAs for standard curve construction. The results obtained using both methods were practically identical.

We first evaluate the amplification of AIB1 gene in 26 primary breast tumors (13 ER positive and 13 ER negative) and corresponding surrounding normal breast tissue samples. AIB1 gene was found to be amplified in 1 ER positive tumor sample that constitutes 3.8% of total or 7.6% of ER positive tumors. This result is consistent with previous report [3, 7, 13].

As shown in Table 1, 9 out of 29 cell lines had elevated AIB1 at 2SD above the mean. All of them were ER positive. AIB1 gene was not amplified in 7 ER positive cell lines: BT20, ZR-75-1, T47D, BT483, MDA-MB-361, MDA-MB-468 and MDA-MB-330. None of the 13 ER negative cell lines showed significant AIB1 amplification. Some cell lines are derivatives of others. For example, AK-47 is derived from ER positive ZR-75-1 cell line. In AK-47 cells, loss of ER expression did not have any effect on AIB1 copy number. The amplification of AIB1 gene in different passages of ER positive MCF-7 cell lines remains at high levels of 18–23 fold of control. Loss of estrogen dependence in LCC1 is accompanied by moderate decrease in AIB1 gene amplification (13.5 fold in LCC1 versus 22.2 fold in MCF-7). The level of AIB1 gene amplification was reduced to 14.6 fold when the estrogen-independent cells became resistant to antiestrogen 4-OH TAM treatment (LCC2). Similar decrease in amplification (15.6 fold of control) of AIB1 gene was observed in estrogen-independent cell line that has gained antiestrogen resistance to both non-steroid, 4-OH TAM, and steroid, ICI 182,780 antiestrogens (LCC9) (Table 1).

The polymorphic poly Q encoding region of AIB1 contains CAG repeat that is frequently interrupted by CAA's. The poly Q region is part of the histone acetyltransferase domain. It is also where the recruitment and interactions with other components of the transcription activator complex takes place. In order to investigate if qualitative alteration in this region accompanied the quantitative change of AIB1 gene in breast cancer cell lines, we cloned and sequenced the poly Q encoding region of the gene. The cloning/sequencing technique resolved the heterogeneous poly Q encoding sequences into distinct sequences, thus allowing the analysis at the single molecule level. At least 8 clones from each cell line were selected and sequenced. Theoretically, there should be only 2 distinct sequence patterns if the cell line is heterozygous for AIB1 allele and one distinct sequence pattern if it is homozygous. However, 18/25 (72%) (data partially shown in Table 2) of the cell lines contain at least one poly Q encoding sequence pattern that represents less than 20% of the sequenced clones of the cell line. These results suggest that the under-represented sequences probably arise from the parental sequence by somatic mutation. Indeed these rare sequences differ from their parental sequence by one base pair substitution (CAG to CAA) or by insertion or deletion of CAGs. The high degree (72% of the cell lines) of somatic instability is probably characteristic for cancer cell lines since it only occurs in less than 5% (2/43) of the normal controls. We analyzed normal A1N4 cell line at two different times. The first time, ten clones of A1N4 cell line were sequenced. Pattern 2 (Table 2), (CAG)₆CAA(CAG)₉(CAACAG)₃(CAACAGCAG)₂CAA, was found in 4 clones, and pattern 17, (CAG)₄CAA(CAG)₉(CAACAG)₃(CAACAGCAG)₂CAA was found in 6 clones. The second time, 31 clones were sequenced. Sixteen had pattern 2 and 15 had pattern 17. There was no occurrence of "extra" poly Q encoding sequence. These results suggest that the occurrence of rare sequences is not due to cloning/PCR artifact.

Association between poly Q length or its specific encoding sequence with AIB1 amplification was not recognized (Table 2). Somatic instability occurred in all variants of MCF-7 cell line, although they all maintain the parental allele as the predominant coding sequences (pattern 1). Two new sequences arise by insertion of 2 and 3 CAGs in passage 19. Another two new sequences occur in passage 72 by deletion of 1 and 2 CAG repeats. Similar somatic mutations occur in cell lines LCC1, LCC2, LCC9, LY-2 and R27. These mutations seem to occur randomly and independently in each cell line. There is not any single sequence that occurs more frequently than others, except pattern 5, which occurs 3 times by losing 1 CAG directly from the parental sequence.

AK47 was derived from ZR-75-1 by losing its ER activity. During the establishment of the cell line, additional somatic mutations occurred in the polyglutamine region (patterns 11 and 16). The poly Q encoding sequence of AIB1 gene seems to be quite unstable in MDA-MB435 cell line. It has 4 distinct poly Q encoding sequence patterns with pattern 9 as the predominant one. Its variant LCC6 had a total of 7 different sequence patterns. These data are consistent with the genomic instability that is characteristic for cancer cells. Although poly Q encoding sequence patterns do not seem to directly link to AIB1 gene amplification, it is possible that the alteration in poly Q length affects protein-protein interaction, thus, the transactivation activity of AIB1. While most alterations do not change poly Q length significantly, rare sequence pattern in LCC2 with much shorter (only 14 repeats) poly glutamine tract may affect the co-transactivating activity of AIB1 gene.