- Research article
- Open Access
- Open Peer Review
This article has Open Peer Review reports available.
Data integration from two microarray platforms identifies bi-allelic genetic inactivation of RIC8Ain a breast cancer cell line
- Aslaug Aamodt Muggerud†1, 2,
- Henrik Edgren†3, 4,
- Maija Wolf3, 4,
- Kristine Kleivi1, 4,
- Emelyne Dejeux5,
- Jörg Tost5,
- Therese Sørlie†1, 6Email author and
- Olli Kallioniemi†3, 4Email author
© Muggerud et al; licensee BioMed Central Ltd. 2009
Received: 03 December 2008
Accepted: 11 May 2009
Published: 11 May 2009
Using array comparative genomic hybridization (aCGH), a large number of deleted genomic regions have been identified in human cancers. However, subsequent efforts to identify target genes selected for inactivation in these regions have often been challenging.
We integrated here genome-wide copy number data with gene expression data and non-sense mediated mRNA decay rates in breast cancer cell lines to prioritize gene candidates that are likely to be tumour suppressor genes inactivated by bi-allelic genetic events. The candidates were sequenced to identify potential mutations.
This integrated genomic approach led to the identification of RIC8A at 11p15 as a putative candidate target gene for the genomic deletion in the ZR-75-1 breast cancer cell line. We identified a truncating mutation in this cell line, leading to loss of expression and rapid decay of the transcript. We screened 127 breast cancers for RIC8A mutations, but did not find any pathogenic mutations. No promoter hypermethylation in these tumours was detected either. However, analysis of gene expression data from breast tumours identified a small group of aggressive tumours that displayed low levels of RIC8A transcripts. qRT-PCR analysis of 38 breast tumours showed a strong association between low RIC8A expression and the presence of TP53 mutations (P = 0.006).
We demonstrate a data integration strategy leading to the identification of RIC8A as a gene undergoing a classical double-hit genetic inactivation in a breast cancer cell line, as well as in vivo evidence of loss of RIC8A expression in a subgroup of aggressive TP53 mutant breast cancers.
Cell culture and emetine treatment
We used six commonly studied breast cancer cell lines, representing different types of breast cancer; MDA-MB-468 (ER-, PR-, TP53 mutation), MDA-MB-231 (ER-, PR-, TP53 mutation), ZR-75-1 (ER+, PR+), MCF7 (ER+, PR+), BT-474 (ER+, PR+, TP53 mutation) and T-47D (ER+, PR+, TP53 mutation) and three non-malignant cell lines; HMECs (non-malignant human mammary epithelial cells), IMR90 (normal lung fibroblasts) and WS1 (normal skin fibroblasts). All cell lines were obtained from American Type Culture Collection and grown in accordance with the distributor's instructions. Both malignant and non-malignant cell lines were treated with the translation inhibitor emetine dihydrochloride hydrate (Sigma-Aldrich, St Louis, MO). For each cell line, parallel cell cultures were grown in 175 cm2 flasks until 70–80% confluence. Half of the subconfluent cultures were treated with 100 μg ml-1 of emetine dihydrochloride hydrate while the other half were left as untreated controls. Both emetine-treated and untreated cells were incubated at 37°C for 10 h. After drug exposure, cells were trypsinated, cell pellets snap-frozen and total RNA was extracted using RNeasy mini kit (Qiagen, GmbH, Hilden Germany) as described by the manufacturer. RNA quality was evaluated using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, California, USA). Genomic DNA was extracted from untreated cells using the NucleoSpin Tissue kit with the protocol for cultured cells (Macherey-Nagel, Düren, Germany).
We used Affymetrix Human Genome U133 plus 2.0 GeneChip oligonucleotide arrays to compare emetine-treated and untreated cells. 3 μg of total RNA per sample was prepared according to the manufacturer's protocols. For hybridization, fragmented cRNA (15 μg from each sample) was used. The arrays were stained with phycoerythrin-streptavidin and the signal intensity was amplified by treatment with a biotin-conjugated anti-streptavidin antibody, followed by a second staining using phycoerythrin-streptavidin. Finally, the stained arrays were scanned using the GeneChip Scanner 3000 (Affymetrix). Data were analyzed using GeneChip Operating Software 1.1 (Affymetrix).
Array-based CGH was carried out using DNA from the untreated breast cancer cell lines. Agilent Human Genome CGH microarray 44 k was utilized according to the protocol provided by the manufacturer (Agilent Technologies, Palo Alto, California, USA). Genomic DNA pooled from healthy female donors was used as reference in all hybridizations. Briefly, six micrograms of digested and purified tumour and reference DNA was labelled with Cy5-dUTP and Cy3-dUTP (Perkin-Elmer, Wellesley, Massachusetts, USA), respectively, in a random priming reaction using BioPrime DNA Labelling System (Invitrogen, Carlsbad, California, USA). Labelled tumour and reference samples were pooled and hybridized onto the arrays according to the protocol. After hybridization arrays were washed and scanned with a laser scanner (G2565 Scanner, Agilent Technologies). Feature Extraction software version 7.5.1 (Agilent Technologies) was used to extract the signal intensities. CGH Analytics software version 3.4.27 (Agilent Technologies) was used for data visualization.
Microarray data analysis
Data from the Affymetrix gene expression arrays were normalized using the dChip method . Gene expression and aCGH data were combined by calculating, for each probe set, the median of all aCGH oligos located between the start and end base pair positions of the gene the probe set mapped to. Probe set to gene mappings were retrieved from NetAffx (April 2005) and all base pair positions are based on the May 2004 human genome build. The data have been deposited in NCBIs Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/projects/geo/) and are accessible through GEO Series record GSE15477.
Selection and prioritization of genes from the NMD microarray data
The most promising candidate genes for harbouring truncating mutations based on the NMD data in each breast cancer cell line were selected using the following criteria: I) the gene was induced by emetine at least 1.5 fold (log2 scale) in the cancer cell line and not in any of the other cell lines. II) Expression in the clean (no emetine treatment) sample was lower than in any of the clean samples from all other cell lines. III) The gene was located in a heterozygously deleted region. IV) Expression after emetine treatment was over 50 units (to exclude genes expressed at levels which are not measured reliably).
A previously described cohort of early stage breast cancer (I/II) [7, 12] was used for validation. From this, DNA from 127 tumours was used to sequence the RIC8A gene. Gene expression data from 115 tumours of the same cohort were used for in silico validation of expression levels and 38 of these were validated by qRT-PCR. Methylation analysis was performed on 86 tumours from a cohort of early lesions of breast cancer, 27 pure DCIS, 27 pure invasive (< 15 mm) and 32 mixed lesions (invasive with in situ component)  and in 75 DNA tumour samples from patients with locally advanced breast cancer (T3/T4) . Microarray data of RIC8A were further in silico validated in 251 primary breast tumours .
PCR primers with M13 tails were designed to amplify the exons plus 50 bp of intronic sequence at both ends of each exon (for primer sequences see Additional file 1). PCR products were purified with PCR cleanup filter plates (Millipore) and sequenced using M13 primers and BigDye v.1.1 on a 3730 DNA analyzer (Applied Biosystems).
cDNA synthesis and real-time PCR analysis
cDNA was generated in a total volume of 20 μl with 50 ng total RNA using GeneAmp® Gold RNA PCR Core kit (Applied Biosystems). Real-time PCR experiments were performed using TaqMan Gene Expression Assays (Applied Biosystems) on an ABI Prism 7900 HT sequence detector system (Applied Biosystems). The primers included were designed to overlap exon-exon boundaries to avoid genomic DNA contamination. Universal human reference RNA (Stratagene) was used to generate standard curves. Each sample was run in triplicate and the relative gene expression levels were determined with the standard curve method and normalized using PGK1 as reference gene.
DNA methylation analysis by pyrosequencing
1 μg of DNA was bisulphite converted using the MethylEasy™ HT Kit for Centrifuge (Human Genetic Signatures, North Ryde, Australia) according to the manufacturer's instructions. Quantitative DNA methylation analysis of the bisulphite treated DNA was performed by pyrosequencing . 98 CpGs spanning 2 kb of genomic sequence were quantitatively analyzed at single nucleotide resolution. The investigated regions comprised the entire promoter region as well as other regions of the large CpG island (nucleotides -943 to +1338 relative to the transcription start site). Oligonucleotides for PCR amplification and pyrosequencing were synthesized by Biotez (Buch, Germany) and sequences are given in Additional file 2.
Student's t-test and Mann-Whitney test were performed using SPSS (version 15.0. SPSS Inc, Chicago, Illinois, USA) for association with clinical parameters. In the microarray datasets used for validation, samples with RIC8A signal intensity < 650 were considered as low expressed and RIC8A signal intensity > 950 as highly expressed. With this range, about 15% of the tumours showed low expression, a frequency similar to what was observed in other published data sets [16, 17].
Identification of genes subject to NMD-degradation by microarray analysis
Sequencing of RIC8Ain clinical breast cancer samples
To further determine the relevance of the mutation in clinical specimens, we sequenced all ten exons of RIC8A in 127 tumour samples obtained from patients with early-stage breast cancer. We could not find nonsense mutations or other amino acid changing mutations in any of these tumours. Notably, we have previously demonstrated the ability to detect mutations in only 10–15% of cells in a sample with this sequencing technology.
To evaluate if RIC8A could be silenced by CpG hypermethylation in the promoter regions, we pyrosequenced 98 CpGs spanning 1 kb upstream of the transcription start site as well as the 3' end of the large CpG island downstream of the transcription start site in pre-invasive (DCIS) (n = 27) and invasive tumours with DCIS components (n = 32), in early invasive breast cancer (n = 27) as well as a cohort of 75 locally advanced breast cancers. All CpG positions upstream of the transcription start were found to be unmethylated in all samples, while some methylation was observed at CpGs downstream of the transcription start as well as at the 3' end of the CpG island where methylation levels reached 40–80%. However, DNA from normal breast tissue samples displayed identical methylation patterns indicating that RIC8A is not a target for epigenetic silencing (data not shown).
Bioinformatics and qRT-PCR analysis of RIC8Ain clinical breast cancer cohorts
We describe here an integrated genomic approach for the genome-wide-screening for putative tumour suppressor genes in cancer. Given the multitude of potential target genes in frequently deleted regions of the cancer genome, strategies to narrow down the candidates for detailed study would lead to more efficient identification of cancer genes. Here, we describe how a data integration approach combining three microarray technologies (aCGH, gene expression arrays and NMD microarrays) made it possible to narrow the search for potential mutated genes to three tumour suppressor candidates, one of which, RIC8A, was confirmed to show a truncating mutation, deletion and loss of expression in a breast cancer cell line. Furthermore, while the clinical breast cancers had no clear pathogenetic mutations, a subset of clinical breast cancers associated with aggressive features showed low levels of RIC8A expression indicating that inactivation of this gene may play a role in the pathogenesis of a subset of breast cancers.
This represents the second example of NMD-based discovery identifying a novel truncating mutation in cancer. Our previous studies of prostate cancer cell lines identified EPHB2 as a target for similar deletion-truncation events in the DU-145 cells . Therefore, it is clear that some, but not all mutations are highlighted by the NMD-based strategy, particularly in combination with the aCGH data. Treatment with translation inhibitors, such as emetine, extends the half-lives of numerous transcripts . In addition, nonsense-mediated mRNA decay (NMD) is known to regulate a significant number of genes under physiological conditions , as well as to prevent incorrectly spliced genes from being translated . To avoid selection of stress-response genes or those normally regulated by NMD, we applied stringent filtering criteria to the NMD microarray data from six breast cancer cell lines. These included first, presence of the candidate gene in a deleted region of the genome in that cell line, corresponding with the two-hit Knudson hypothesis of tumour suppressor gene inactivation, and second, a low relative baseline mRNA expression. Using this integrated genomic approach we limited the list of candidates for sequencing to 51. The three top candidate genes with the clearest increase in normalized NMD ratios were sequenced in all cell lines. Of these, RIC8A showed a truncating mutation.
After discovery of a mutation in a cancer cell line, the more challenging task is to ascertain the significance and prevalence of such mutations in clinical samples. The key question here is as to whether the mutation is critical to the pathogenesis of cancer and common across a set of cancers, or possibly a "passenger mutation" selected along with a more important other mutation. Although mutations in the coding region of RIC8A were not found in the cohort of 127 early-stage breast cancers, we cannot rule out that mutations might exist at a low frequency in more advanced cases. Exhaustive pyrosequencing of the entire RIC8A promoter region in early invasive breast cancers as well as in pre-invasive tumours (DCIS) and locally advanced (T3/T4) tumours excluded also promoter hypermethylation as a mechanism for RIC8A gene silencing. Recent estimates of the prevalence of somatic mutations in human cancers are highly variable both within and between classes of cancer . The emerging pattern is that there are relatively few frequently mutated genes, (most of these already known) and a very large number of rare mutations, often highly varying from one case to another . Based on the lack of mutations identified in clinical breast cancer samples as well as the lack of evidence from cancer genome sequencing studies indicating involvement of RIC8A [22, 23], we postulated that RIC8A inactivation could either be a rare genetic event in breast cancer, or possibly, its inactivation in clinical samples could take place more often through other mechanisms, not involving direct truncating mutations or methylation.
RIC8A was originally found by genetic studies in C. elegans and reported to act in vitro as a guanine nucleotide exchange factor (GEF) for G protein alpha subunits . Several studies have since reported that RIC8A is required for G-protein signalling and is involved in centrosome movements during early embryogenesis in C. elegans [24, 25]. In addition, RIC8A was discovered to be required for proper asymmetric division of one-cell-stage in both C. elegans embryos  and Drosophila . However, the physiological role of the mammalian homolog RIC8A in G protein-coupled receptor signalling in intact cells is largely unknown. Yet, a human mitotic phosphorylation motif associated with protein localization to the mitotic apparatus was recently identified in RIC8A, suggesting a role for RIC8A in mitosis . This point to the possibility that RIC8A would play a role in cancer relevant pathways and that its inactivation could be important to the pathogenesis of the disease.
Bioinformatics analysis of microarray data from 251 consecutive breast cancers  showed that low RIC8A expression was significantly associated with hormone receptor negativity (ER and PR) as well as presence of TP53 mutations. A similar analysis of gene expression in 115 tumours from another cohort focussing on early breast cancer indicated a similar, but statistically weaker trend (Figure 3). The association between low RIC8A expression and TP53 mutations was validated by the analysis using qRT-PCR of 38 tumours from the early stage cohort . Furthermore, a previously published analysis of 171 primary breast tumours showed that the chromosome 11p telomeric region is lost at an approximate frequency of 15% . The frequency is highest towards the distal part of the chromosome, where RIC8A is located, and declines towards the centromere. In the same tumours, genes in this region, including RIC8A, were shown to be statistically significantly deleted and underexpressed when combining copy number and gene expression data. Also, this deletion was statistically significantly associated to both recurrence and distant metastases in this patient cohort. There was no significant association with any other clinical parameter. Yet another study showed that 11p15.5-p15.4 copy number loss in primary breast cancers was associated with a higher incidence of recurrent disease in tamoxifen-treated patients . While RIC8A is located in one of the most frequently deleted regions in breast cancer , the deletion obviously involves a large number of genes. We therefore evaluated whether the flanking genes from RIC8A would show similar statistical associations between low expression and ER/PR negativity and TP53 mutation, respectively. For example, this region harbours the SIRT3 gene, which has been associated with cancer [28, 29] and whose expression showed correlation with copy number loss together with RIC8A . However, in our analyses RIC8A showed the strongest association with aggressiveness. Together with the presence of the truncating mutation, this suggests a specific importance of the RIC8A gene in a subset of breast cancers.
We demonstrate here that integrating microarray data from three different methods (aCGH, expression, nonsense-mediated mRNA decay) facilitated the identification of putative tumour suppressor genes undergoing bi-allelic inactivation in cancer. Here, this strategy led to the identification of RIC8A as a gene undergoing deletional and mutational inactivation in a breast cancer cell line, as well as in vivo evidence of loss of its expression in a subgroup of aggressive breast cancers.
This study was supported by the Norwegian Cancer Society, Norwegian Research Council, Marie Curie Ext program, Academy of Finland, Finnish Cancer Society, Sigrid Juselius Foundation, and the French-Norwegian cooperation program Aurora grant 15842WE (AAM and JT). We thank Suet-Feung Chin and Carlos Caldas for help with access to aCGH data for validation.
- Cowell JK, Hawthorn L: The application of microarray technology to the analysis of the cancer genome. Curr Mol Med. 2007, 7: 103-120. 10.2174/156652407779940387.View ArticlePubMedGoogle Scholar
- Noensie EN, Dietz HC: A strategy for disease gene identification through nonsense-mediated mRNA decay inhibition. Nat Biotechnol. 2001, 19: 434-439. 10.1038/88099.View ArticlePubMedGoogle Scholar
- Huusko P, Ponciano-Jackson D, Wolf M, Kiefer JA, Azorsa DO, Tuzmen S, et al: Nonsense-mediated decay microarray analysis identifies mutations of EPHB2 in human prostate cancer. Nat Genet. 2004, 36: 979-983. 10.1038/ng1408.View ArticlePubMedGoogle Scholar
- Wolf M, Edgren H, Muggerud A, Kilpinen S, Huusko P, Sorlie T, et al: NMD microarray analysis for rapid genome-wide screen of mutated genes in cancer. Cell Oncol. 2005, 27: 169-173.PubMedPubMed CentralGoogle Scholar
- Losson R, Lacroute F: Interference of nonsense mutations with eukaryotic messenger RNA stability. Proc Natl Acad Sci USA. 1979, 76: 5134-5137. 10.1073/pnas.76.10.5134.View ArticlePubMedPubMed CentralGoogle Scholar
- Miller LD, Smeds J, George J, Vega VB, Vergara L, Ploner A, et al: An expression signature for p53 status in human breast cancer predicts mutation status, transcriptional effects, and patient survival. Proc Natl Acad Sci USA. 2005, 102: 13550-13555. 10.1073/pnas.0506230102.View ArticlePubMedPubMed CentralGoogle Scholar
- Naume B, Zhao X, Synnestvedt M, Borgen E, Russnes HG, Lingjærde OC, et al: Prescence of bone marrow micrometastasis is associated with different recurrence risk within molecular subtypes og breast cancer. Molecular Oncology. 2007, 1: 160-171. 10.1016/j.molonc.2007.03.004.View ArticlePubMedGoogle Scholar
- Miller KG, Emerson MD, McManus JR, Rand JB: RIC-8 (Synembryn): a novel conserved protein that is required for G(q)alpha signaling in the C. elegans nervous system. Neuron. 2000, 27: 289-299. 10.1016/S0896-6273(00)00037-4.View ArticlePubMedPubMed CentralGoogle Scholar
- Wang H, Ng KH, Qian H, Siderovski DP, Chia W, Yu F: Ric-8 controls Drosophila neural progenitor asymmetric division by regulating heterotrimeric G proteins. Nat Cell Biol. 2005, 7: 1091-1098. 10.1038/ncb1317.View ArticlePubMedGoogle Scholar
- Yang F, Camp DG, Gritsenko MA, Luo Q, Kelly RT, Clauss TR, et al: Identification of a novel mitotic phosphorylation motif associated with protein localization to the mitotic apparatus. J Cell Sci. 2007, 120: 4060-4070. 10.1242/jcs.014795.View ArticlePubMedGoogle Scholar
- Li C, Wong WH: Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection. Proc Natl Acad Sci USA. 2001, 98: 31-36. 10.1073/pnas.011404098.View ArticlePubMedGoogle Scholar
- Wiedswang G, Borgen E, Karesen R, Kvalheim G, Nesland JM, Qvist H, et al: Detection of isolated tumor cells in bone marrow is an independent prognostic factor in breast cancer. J Clin Oncol. 2003, 21: 3469-3478. 10.1200/JCO.2003.02.009.View ArticlePubMedGoogle Scholar
- Warnberg F, Nordgren H, Bergkvist L, Holmberg L: Tumour markers in breast carcinoma correlate with grade rather than with invasiveness. Br J Cancer. 2001, 85: 869-874. 10.1054/bjoc.2001.1995.View ArticlePubMedPubMed CentralGoogle Scholar
- Geisler S, Lonning PE, Aas T, Johnsen H, Fluge O, Haugen DF, et al: Influence of TP53 gene alterations and c-erbB-2 expression on the response to treatment with doxorubicin in locally advanced breast cancer. Cancer Res. 2001, 61: 2505-2512.PubMedGoogle Scholar
- Tost J, Gut IG: DNA methylation analysis by pyrosequencing. Nat Protoc. 2007, 2: 2265-2275. 10.1038/nprot.2007.314.View ArticlePubMedGoogle Scholar
- Chin SF, Teschendorff AE, Marioni JC, Wang Y, Barbosa-Morais NL, Thorne NP, et al: High-resolution aCGH and expression profiling identifies a novel genomic subtype of ER negative breast cancer. Genome Biol. 2007, 8: R215-10.1186/gb-2007-8-10-r215.View ArticlePubMedPubMed CentralGoogle Scholar
- Han W, Han MR, Kang JJ, Bae JY, Lee JH, Bae YJ, et al: Genomic alterations identified by array comparative genomic hybridization as prognostic markers in tamoxifen-treated estrogen receptor-positive breast cancer. BMC Cancer. 2006, 6: 92-10.1186/1471-2407-6-92.View ArticlePubMedPubMed CentralGoogle Scholar
- Ross J: A hypothesis to explain why translation inhibitors stabilize mRNAs in mammalian cells: mRNA stability and mitosis. Bioessays. 1997, 19: 527-529. 10.1002/bies.950190612.View ArticlePubMedGoogle Scholar
- Mendell JT, Sharifi NA, Meyers JL, Martinez-Murillo F, Dietz HC: Nonsense surveillance regulates expression of diverse classes of mammalian transcripts and mutes genomic noise. Nat Genet. 2004, 36: 1073-1078. 10.1038/ng1429.View ArticlePubMedGoogle Scholar
- Jaillon O, Bouhouche K, Gout JF, Aury JM, Noel B, Saudemont B, et al: Translational control of intron splicing in eukaryotes. Nature. 2008, 451: 359-362. 10.1038/nature06495.View ArticlePubMedGoogle Scholar
- Greenman C, Stephens P, Smith R, Dalgliesh GL, Hunter C, Bignell G, et al: Patterns of somatic mutation in human cancer genomes. Nature. 2007, 446: 153-158. 10.1038/nature05610.View ArticlePubMedPubMed CentralGoogle Scholar
- Wood LD, Parsons DW, Jones S, Lin J, Sjoblom T, Leary RJ, et al: The genomic landscapes of human breast and colorectal cancers. Science. 2007, 318: 1108-1113. 10.1126/science.1145720.View ArticlePubMedGoogle Scholar
- Sjoblom T, Jones S, Wood LD, Parsons DW, Lin J, Barber TD, et al: The consensus coding sequences of human breast and colorectal cancers. Science. 2006, 314: 268-274. 10.1126/science.1133427.View ArticlePubMedGoogle Scholar
- Miller KG, Rand JB: A role for RIC-8 (Synembryn) and GOA-1 (G(o)alpha) in regulating a subset of centrosome movements during early embryogenesis in Caenorhabditis elegans. Genetics. 2000, 156: 1649-1660.PubMedPubMed CentralGoogle Scholar
- Tall GG, Krumins AM, Gilman AG: Mammalian Ric-8A (synembryn) is a heterotrimeric Galpha protein guanine nucleotide exchange factor. J Biol Chem. 2003, 278: 8356-8362. 10.1074/jbc.M211862200.View ArticlePubMedGoogle Scholar
- Afshar K, Willard FS, Colombo K, Johnston CA, McCudden CR, Siderovski DP, et al: RIC-8 is required for GPR-1/2-dependent Galpha function during asymmetric division of C. elegans embryos. Cell. 2004, 119: 219-230. 10.1016/j.cell.2004.09.026.View ArticlePubMedGoogle Scholar
- Winqvist R, Hampton GM, Mannermaa A, Blanco G, Alavaikko M, Kiviniemi H, et al: Loss of heterozygosity for chromosome 11 in primary human breast tumors is associated with poor survival after metastasis. Cancer Res. 1995, 55: 2660-2664.PubMedGoogle Scholar
- Allison SJ, Milner J: SIRT3 is pro-apoptotic and participates in distinct basal apoptotic pathways. Cell Cycle. 2007, 6: 2669-2677.View ArticlePubMedGoogle Scholar
- Ashraf N, Zino S, Macintyre A, Kingsmore D, Payne AP, George WD, et al: Altered sirtuin expression is associated with node-positive breast cancer. Br J Cancer. 2006, 95: 1056-1061. 10.1038/sj.bjc.6603384.View ArticlePubMedPubMed CentralGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1755-8794/2/26/prepub
This article is published under license to 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.