Comprehensive copy number profiles of breast cancer cell model genomes

bcr1370 BCJ Research article Comprehensive copy number profiles of breast cancer cell model genomes Shadeo Ashleen ashadeo@bccrc.ca Lam L Wan wanlam@bccrc.ca

Cancer Genetics and Developmental Biology, British Columbia Cancer Research Centre, Vancouver BC, V5Z 1L3, Canada

Breast Cancer Research 1465-5411 2006 8 1 R9 http://breast-cancer-research.com/content/8/1/R9 16417655 10.1186/bcr1370 12 8 2005 9 9 2005 11 10 2005 18 11 2005 3 1 2006 2006 Shadeo and Lam; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Introduction

Breast cancer is the most commonly diagnosed cancer in women worldwide and consequently has been extensively investigated in terms of histopathology, immunochemistry and familial history. Advances in genome-wide approaches have contributed to molecular classification with respect to genomic changes and their subsequent effects on gene expression. Cell lines have provided a renewable resource that is readily used as model systems for breast cancer cell biology. A thorough characterization of their genomes to identify regions of segmental DNA loss (potential tumor-suppressor-containing loci) and gain (potential oncogenic loci) would greatly facilitate the interpretation of biological data derived from such cells. In this study we characterized the genomes of seven of the most commonly used breast cancer model cell lines at unprecedented resolution using a newly developed whole-genome tiling path genomic DNA array.

Methods

Breast cancer model cell lines MCF-7, BT-474, MDA-MB-231, T47D, SK-BR-3, UACC-893 and ZR-75-30 were investigated for genomic alterations with the submegabase-resolution tiling array (SMRT) array comparative genomic hybridization (CGH) platform. SMRT array CGH provides tiling coverage of the human genome permitting break-point detection at about 80 kilobases resolution. Two novel discrete alterations identified by array CGH were verified by fluorescence in situ hybridization.

Results

Whole-genome tiling path array CGH analysis identified novel high-level alterations and fine-mapped previously reported regions yielding candidate genes. In brief, 75 high-level gains and 48 losses were observed and their respective boundaries were documented. Complex alterations involving multiple levels of change were observed on chromosome arms 1p, 8q, 9p, 11q, 15q, 17q and 20q. Furthermore, alignment of whole-genome profiles enabled simultaneous assessment of copy number status of multiple components of the same biological pathway. Investigation of about 60 loci containing genes associated with the epidermal growth factor family (epidermal growth factor receptor, HER2, HER3 and HER4) revealed that all seven cell lines harbor copy number changes to multiple genes in these pathways.

Conclusion

The intrinsic genetic differences between these cell lines will influence their biologic and pharmacologic response as an experimental model. Knowledge of segmental changes in these genomes deduced from our study will facilitate the interpretation of biological data derived from such cells.

Introduction

Breast cancer is the most prevalent cancer worldwide and is the second leading cause of cancer-related deaths in women in North America 12. It is a complex disease in which multiple genetic factors can combine to drive pathogenesis 345. Changes in copy numbers of genes such as ERBB2 and c-MYC have been extensively documented in breast cancer and are present in model cell lines 6789. Amplified (and overexpressed) genes are prime therapeutic targets as for example, the use of the drug trastuzumab against ERBB2 has been shown to improve breast cancer survival rates alone or in combination with other treatments 101112.

Strategies to detect gene copy number alterations will facilitate the identification of novel molecular targets. Previous studies with 10-megabase (Mb) resolution conventional metaphase comparative genomic hybridization (CGH) have identified gross regions of recurrent chromosomal aberrations in multiple breast cancer cell lines including loci within chromosomes 1q, 8q, 11q13, 17q and 20q13. Many of these alterations proved to be relevant because they were also present in primary tumors investigated 131415. Recent advances in array CGH have greatly improved the resolution of this technology, enabling the detection of segmental copy losses and gains 1617. Regional genomic arrays, providing contiguous or tiling coverage of a locus of interest, have been constructed for the fine mapping of commonly altered regions in breast cancer (such as 20q13) 181920. Whole chromosome arrays have been used to provide information at 500 kb intervals. For example, a chromosome 17 array was used to identify 13 regions of change present in breast cancer cell line models and primary breast cancers 21. Similarly, a genome-wide array containing nearly 2,500 bacterial artificial chromosome (BAC) clones with a resolution at about 1.4 Mb was used to illustrate the detection of copy number alterations (CNAs) in various breast cancer cell lines 22. Recently, a separate study using an array of 422 genomic loci detected frequent alterations at 1, 6, 7p, 9, 11q, 12q, 17, 20q and 22q in archival breast cancer specimens 23. cDNA arrays have also detected DNA copy changes of amplicons containing ERRB2 on 17q 24252627. More recently, a cDNA array containing 6,691 mapped human genes was used to explore the relationship between copy number alteration and gene expression changes in breast tumors and cell lines 28. While large-insert clone megabase-interval CGH arrays and cDNA arrays provide a robust platform for the rapid survey of tumor genomes, valuable information could be overlooked as a result of their limited resolution. It is clear that a more detailed description of breast tumor genomes would require re-examination with a higher-resolution array platform.

Genetic, biochemical and pharmacologic studies of breast cancer have been greatly dependent on several commonly used model breast cancer cell lines: MCF-7, BT-474, SK-BR-3, T-47D, UACC-893, MDA-MB-231 and ZR-75-30. That is, a summation of studies involving at least one of these seven cell lines produces over 13,500 hits on Medline. These cells are known to harbor gross chromosomal aberrations; measuring the precise segmental copy number status across their entire genome may uncover novel discrete changes. In the current study we expanded the use of array CGH to survey the genomes of these breast cancer cells at unprecedented detail with a recently developed whole-genome tiling path array that covered the genome with 32,433 overlapping BAC clones 29. Analysis at this resolution has led to the identification of novel features in these genomes and to the delineation of segmental genetic alterations that have escaped detection by conventional molecular cytogenetic techniques and previous marker-based or interval array CGH analysis.

Materials and methods

Cell line DNA

A panel of seven breast cancer-derived cell line DNA was obtained from the American Type Culture Collection: MCF-7, T-47D, Sk-Br-3, MDA-MB-231, BT-474, UACC-893 and ZR-75-30. Pooled normal female DNA was used as reference for all array CGH experiments (Novagen, Mississauga, ON, Canada). DNA was quantified with a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).

Array CGH

The seven cell lines were assayed for genetic alterations with a whole-genome tiling path BAC array in comparative genomic hybridization experiments. The submegabase-resolution tiling set (SMRT) array contains 32,433 overlapping BAC-derived DNA segments that provide tiling coverage over the human physical genome map. All clones were spotted in triplicate, resulting in 97,299 elements over two sides 293031. A detailed protocol is provided in Additional file 1.

Additional File 1

A Microsoft Word file containing the detailed SMRT array CGH protocol.

Click here for file

Imaging and analysis

Hybridizations were scanned with an imaging system based on a charge-coupled device (Arrayworx eAuto; Applied Precision, Issaquah, WA, USA) and analyzed with SoftWoRx Tracker Spot Analysis software. Stringent criteria were applied to filter spot intensity data. A standard deviation greater than 0.075 between triplicate spots was deemed unreliable and such spots were therefore excluded from our analysis 29. Only data points with a ratio of signal intensity to background intensity noise exceeding 15 were used in the analysis.

Custom software (SeeGH) was used to visualize log₂ratios of clones with respect to location in the genome 3233. Because of the complexity of the genomes of these cell lines with respect to ploidy, we have set thresholds for high-level gains and losses to +0.8 and -0.7, respectively, to limit the number of regions for discussion. This threshold encompasses high-level or multi-copy changes previously reported while excluding the abundant number of low-level or single-copy changes common to these cell lines. The complete data set has been made publicly available for further inquiry. In addition, only those loci containing two or more altered overlapping clones were included in the analysis to reduce false positives, and breakpoints were confirmed with the publicly available aCGH-smooth software 3435.

Fluorescence in situ hybridization

For fluorescence in situ hybridization (FISH) probe synthesis, DNA samples from BAC clones RP11-118L18, RP11-419H8, RP11-813P3 and RP11-790I13 were amplified with a modified ligation-mediated polymerase chain reaction protocol as described previously 31. Imaging and analysis were performed as described previously 36.

Results and discussion

Whole-genome tiling path analysis of segmental alterations

SMRT array CGH technology provides a tool for assessing genomic aberrations comprehensively in great detail. Comprehensive genomic profiles of segmental gains and losses for seven commonly used breast cancer model cell lines were revealed with this technology. Because of the large amount of data generated, we present the complete genomic profiles and frequency analysis in Additional files 2, 3, 4, 5, 67 and 8 (Figure 1). The raw data of the signal intensity ratios of the 97,299 spots for each array CGH experiment have been made publicly available 33 and also deposited at the gene expression omnibus (GEO) database at NCBI, series accession number GSE3106.

Additional File 2

A PDF file containing a BT474 Karyogram.

Click here for file

Additional File 3

A PDF file containing a MCF7 Karyogram.

Click here for file

Additional File 4

A PDF file containing a T47D Karyogram.

Click here for file

Additional File 5

A PDF file containing a SKBR3 Karyogram.

Click here for file

Additional File 6

A PDF file containing a MDA MB 231 Karyogram.

Click here for file

Additional File 7

A PDF file containing a ZR 75 30 Karyogram.

Click here for file

Additional File 8

A PDF file containing a frequency plot of 7 cell lines.

Click here for file

Figure 1

Comprehensive submegabase-resolution tiling set (SMRT) array comparative genomic hybridization profile of cell line UACC-893

Comprehensive submegabase-resolution tiling set (SMRT) array comparative genomic hybridization profile of cell line UACC-893. Whole-genome SeeGH karyogram UACC-893. Individual data points denote log₂ratios plotted to corresponding chromosomal location. Log₂± 0.5 scale bars are included for reference. Displacement of data points to the right and left of the centre line represents gain and loss, respectively. The inset shows a magnified view of complex alteration on chromosome 17q.

Figure 1 demonstrates the details of a tiling path SeeGH karyogram, summarizing SMRT array CGH results for cell line UACC-893. Whole chromosomal arm gains can be seen at 1q, 5p, 7p, 8q and 10p, whereas arm losses are evident at 3p, 4p, 5q, 8p, 13q, 17p, 19p, 20p and Xp. Smaller segmental changes such as the telomeric gained region of 6p or loss at 10q are readily detected. Complex alterations indicating multiple levels of change are denoted by higher-level peaks embedded within a region of change, for example the central region of the 2p arm. The magnified display of 17q demonstrates the identification of a discrete CNA. Beginning at the centromere, we can see two regions of segmental loss separated by a high-copy-number amplicon containing the ERBB2 gene. The centromeric breakpoint of this amplicon is located between the overlapping regions of clones RP11-25P3 and RP11-592L16, whereas the telomeric breakpoint is located between clones RP11-686E5 and RP11-259G21. The second region of segmental loss at 17q21.1-q21.31 is followed by a large segmental gain and a second discrete multiple copy amplification at 17q25.1.

To establish detection sensitivity, we first examined previously reported regions of CNA. Our data indicated high-level gains at the c-MYC locus in SK-BR-3 and MCF-7 (+2.84 and +1.19 log₂ratios, respectively) corresponding to previously reported change in copy number 73738. Similarly, BT-474, ZR-75-30, UACC 893 and SK-BR-3 are known to harbor a high-level amplification of the ERBB2 locus. SMRT array CGH, in addition to detecting the ERBB2 locus, revealed several additional discrete changes on the 17q arm in these cell lines. In another example, a previously reported homozygous deletion at 3q13.31, detected by a 10K single nucleotide polymorphism (SNP) array in MCF-7, yielded a log₂ratio of -1.2 in our SMRT array CGH analysis 39. Further comparison of SNP data and SMRT array CGH for cell line BT-474 showed that many of the alterations detected by SMRT array CGH were not clearly delineated or were not detected by the SNP platform (Additional file 9). Although SNP arrays offer the advantage of genotype data, they are only suited to the detection of large-scale changes in copy number. However, the two technologies are clearly complementary because each is designed to address a different question.

Additional File 9

A PDF file containing an SNP SMRT aCGH comparison of 4q and 17q in BT474.

Click here for file

Six of the seven cell lines (not MDA-MB-231) were previously profiled for genomic alterations with the use of a 6,691-gene cDNA microarray 28. Pollack and colleagues showed numerous genomic alterations, both gains and losses, which were correlated with expression patterns on the same array platform. All the CNAs reported were detected by SMRT array CGH, along with the discovery of numerous novel alterations when re-evaluated at tiling path resolution. Known and novel CNAs for the seven genomes are summarized in Table 1. Interestingly, not all CNAs contain annotated genes, which is consistent with the fact that the annotation of coding and non-coding transcripts within the human genome sequence is a continuing process.

Table 1

High-level alterations detected by array CGH

Cell line

Locus

Start clone

End clone

Size (kb)

Locus

Start clone

End clone

Size (kb)

MCF7

Amplifications

Deletions

1p13.3

N0451I14

N0228E23

1,290

3q13.31

N0747H24

N0362H11

700

1p13.2

N0099M15

N0795O09

288

4q34.3-35.2

N0442N05

N0746B09

7,100

1p13.2

N0626F04

N0517B05

1,330

6q25.2-27

M2007C03

M2258B24

14,400

3p14.2-14.1

N0669F02

N0589G04

3,180

8p arm

8q21.2-q24.21

N0133G02

N0315E09

43,300

11p15.5-11.2

N0412M16

M2326E01

45,300

15q21.1-21.3

N0416B20

N0664B09

4,930

11p11.2

N0070A09

M2326E01

1,940

17q23.2-24.3

N0716B04

N0203A19

113,200

11q11-q12.1

N0010E21

F0627I09

4,300

Peak:

17q23.2

N0760B22

N0433B24

4,900

11q14.2-23.3

N0282G16

N0004N09

30,000

17q25.1

N0076G04

N0552F03

766

11q23.3-25(tel)

N0196E01

N0715D10

15,400

20q12

N0385G02

N0476P15

1,790

13q14.2-34(tel)

N0155D15

M2323L19

66,900

20q13.12-20q13.33 (tel)

N0272C13

N0476I15

17,500

Peaks:

20q13.12-20q13.13 A

N0702E03

N0730O20

1,790

20q13.13 B

N0711M06

F0592G15

309

20q13.2 C

N0020J08

N0346B03

1,450

20q13.31 D

N0044A06

N0671P16

411

21q22.13-22.3(tel)

N0094J12

N0457P07

8,450

BT474

Amplifications

1q21.2-q25.1

15q11.2 -q12

N0607H20

N0208F21

5,030

Peaks:

1q21.2-q21.3 A

N0035F14

N0714N02

366

Peaks:

15q11.2 A

N0552D03

M2200G17

949

1q22-q23.1 B

N0647N20

N0740J19

510

15q11.2 B

N0484P15

N0710L06

624

1q24.2 C

N0137J06

N0616K15

540

17q12-21.2

N0196P12

N0278E15

4,090

1q31.3

N0662E13

N0141E20

1,661

17q21.32-23.2

N0771D19

M2014K24

13,800

1q32.1

N0783D13

N0617D19

936

Peaks:

17q21.32-q21.33 A

N0071G24

N0607H13

2,500

1q42.12-q42.13

M2185P06

M2016D17

500

17q22-23.2 B

N0515J20

N0473G17

2,180

1q43

N0236L13

N0614N14

449

17q23.2 C

N0399O18

N0767P09

1,170

1q44-q43

N0440F10

N0794A13

865

17q24.1-24.3

N0583F02

N0693H11

5,300

1q44

N0778E23

N0071K05

1,720

Peak:

17q24.1

N0583F02

N0394K10

482

4p16.1-15.33

N0270I03

N0652B07

1,210

19p13.2-13.12

N0295M02

N0441D06

6,310

4q21.1

N0598G02

N0772N01

2,700

20p12.1

N0134G22

N0022E15

800

9p13.3

N0069E18

N0795P12

2,060

20q11.22

N0601G07

N0552G16

1,300

9q33.1-34.13

M2248M11

N0738I14

12,600

20q13.11-13.32

N0809G24

N0261P09

14,800

11q13.1-13.5

N0813P09

N0360N22

19,800

20q13.33

N0648D07

N0694I10

305

Peaks:

11q13.1 A

N0029K11

N0804F01

704

20q13.33-tel

N0305P22

N0134L13

1,380

11q13.4 B

N0093M11

N0598G03

1,030

Deletions

11q22.1-22.2

N0795D03

N0347H03

3,560

5q14.1-14.3

N0129E04

N0291O24

5,390

14q11.2-q21.1

N0597A11

N0254B15

21,230

6q24.1

N0709J21

M22024J17

3,200

14q31.3-32.12

N0771O08

N0325L17

3,080

20q11.22

N0171G22

N0774C15

670

20p11.23-13.11

N0712N14

N0464F07

7,140

ZR 75 30

Amplifications

8q11.21

N0569I08

N0513O13

682

17q22-23.2

N0349F01

N0153J08

5,600

8q13.3

N0367C12

N0634L17

538

Peaks:

17q23.2 A

N0159D12

M2023J07

298

8q21.2

N0319A24

N0317J10

1,350

17q23.2 B

N0381A05

N0634F05

3,560

8q21.3

N0565H20

N0102D03

264

17q23.3-24.1

N0614F05

N0712A10

481

8q22.1

N0381I07

N0103P22

729

Deletions

8q22.2-24.3

N0281D17

N0620H01

45,600

1p36.33-35.3

N0206L10

N0758C04

26,700

Peaks:

8q23.1 A

N0078J05

N0357L07

907

1p35.2-35.1

N0629A12

N0068H10

775

8q24.12 B

N0760H22

N0088J18

1,440

1p34.3

N0452M14

N0020P17

1,510

8q24.22 C

N0100J23

N0422I20

440

1p21.2-13.3

N0293L15

N0813H10

10,800

8q24.3 D

N0662P06

F0530P15

2,740

4q21.1

N0184F15

N0184J11

151

17q11.1-11.2

N0260A09

N0147N18

3,050

11q13.5-25(tel)

N0149B15

M2270L17

57,600

17q12

N0600K04

N0560P04

980

17q11.2-12

N0082D14

N0104J23

6,020

17q12

N0722D15

N0062P03

1,360

17q21.1-21.32

N0278E15

N0046I22

7,700

17q12-21.1

N0689B15

N0032H06

490

17q23.33-22

N0368A16

N0379H09

994

17q21.32-21.33

N0771D19

M2190C10

3,910

17q22

N0466D20

N0515J20

881

21q11.2-q22.11

N0615H23

N0694N16

16,300

UACC 893

Amplifications

Deletions

7p21.1

N0116D07

N0746H13

733

11p15.1

N0118A14

N0614L12

995

11q13.3-q14.3

M2011L13

N0613M07

24,700

16p12.1

N0104F06

N0674B07

1,380

11q22.1

N0743I15

N0659E10

619

17p arm

17q12-21.2

N0600J16

N0686E05

1,490

17q21.33-24.2

N0095M07

N0068K09

17,700

17q25.1

N0076G04

N0449J21

999

SKBR3

Amplifications

3p22.2

N0091E04

N0325M12

432

17q12-21.2

N0062P03

N0606M07

3,390

3q25.1

N0739J07

N0118L18

678

17q25.3

N0305C04

N0781F24

1,550

3q22.3-q23

N0657M13

N0718H02

2,000

20p arm

3q26.2-q26.31

N0190I05

N0590E04

3,300

20q11.22-11.23

N0454F11

N0254N13

4,510

7q31.1-q32.3

M2023N18

N0019B03

17,800

20q12-13.32

N0434N22

N0290D09

16,200

8q13.3-21.13

N0746L20

N0125J17

10,600

Deletions

8q21.2-21.3

N0509F16

N0561A10

5,950

8q21.3-22.1

N0230C03

N0804A07

6,040

Peaks:

8q21.2 A

N0694L21

N0606L16

309

8q22.3-23.1

N0739L19

N0025P11

4,920

8q21.3 B

N0129P07

N0196C06

559

8q24.22

N0015L05

N1147M08

1,570

8q21.3 C

N0627A06

N0529J09

499

8q24.23-24.3

N0467B24

N0613F12

3,810

8q23.2-24.21

N0114O08

N0294P07

17,100

10q22.3-25.2

N0635P19

N0257E05

30,900

Peaks:

8q23.3 A

N0058O03

N0531C21

746

12q23.3-24.11

N0061L24

M2155D19

1,380

8q23.3 B

N0164M09

M2118B16

365

12q24.21-24.31

N0412G23

N0138I16

5,360

8q24.12 C

N0389M07

N0047A23

821

17q11.2-12

N0634A23

N0607B02

6,460

8q24.21 D

N0288B17

N0294P07

716

17q21.2-23.2

N0400F19

N0767P09

18,100

10q21.1-22.3

N0195P24

N0506E07

25,500

18q12.2-21.2

N0645I23

N0664P08

20,400

14q31.3-32.12

N0046B20

N0386D20

7,020

19p13.11

N0715L15

N0723M22

4,710

17q11.1-11.2

N0458L21

N0193M11

3,170

MDA MB 231

Amplifications

Deletions

6p21.31-21.2

N0479F12

N0450J18

3,510

8q24.13

N0346M14

N0391A17

1,020

6p21.2-21.1

N0259H15

N0769C16

3,290

9p24.2-22.2

N0654D08

N0460F23

14,500

7q35

N0703N05

N0340G20

670

Peak: