Recombination phenotypes of the NCI-60 collection of human cancer cells
© Stults et al; licensee BioMed Central Ltd. 2011
Received: 1 January 2011
Accepted: 17 May 2011
Published: 17 May 2011
The NCI-60 is a collection of tumor cell lines derived from a variety of human adult cancer tissue types and is commonly used for genetic analysis and screening of potential chemotherapeutic agents. We wanted to understand the contributions of specific mechanisms of genomic instability to the etiology of cancers represented by the NCI-60.
We screened the NCI-60 for dysregulated homologous recombination by using the gene cluster instability (GCI) assay we pioneered, and for defects in base excision repair by sensitivity to 5-hydroxymethyl-2'-deoxyuridine (hmdUrd). We identified subsets of the NCI-60 lines that either displayed the characteristic molecular signature of GCI or were sensitive to hmdUrd. With the exception of the NCI-H23 lung cancer line, these phenotypes were not found to overlap. None of the lines examined in either subset exhibited significant changes in the frequency of sister chromatid exchanges (SCE), neither did any of the lines in either subset exhibit microsatellite instability (MSI) indicative of defects in DNA mismatch repair.
Gene cluster instability, sensitivity to hmdUrd and sister chromatid exchange are mechanistically distinct phenomena. Genomic instability in the NCI-60 appears to involve only one mechanism of instability for each individual cell line.
Genomic instability is a fundamental characteristic of most solid tumors and adult leukemias. The term encompasses a broad range of defects that arise by a variety of damaging events and/or mechanistic failures of individual DNA repair pathways. Whatever its source, genomic destabilization is believed to begin early in tumor progression, creating heterogeneity within a population of cells, and conferring, in concert with other events, a selective advantage to a given cell which dominates in proliferation [1, 2]. Instability may or may not continue as the tumor progresses. One means of genomic destabilization is defective or dysregulated homologous recombination.
Homologous recombination (HR) is a mechanism for repairing double-strand breaks (DSBs) during S and G2 phase of the cell cycle. In contrast to non-homologous end joining (NHEJ), which results in a loss of genetic material, homologous recombination is considered error-free repair because it uses the available, identical sequence from the sister chromatid to repair the DSB. Although NHEJ is capable of repairing frank DSBs during G2/M, HR is preferred, especially for repairing the DSBs that arise at stalled replication forks, for example from forks that encounter single strand breaks or cross-links . Nevertheless, mitotic HR is a complex, varied, and tightly regulated process, and defects in several of the components of HR have long been associated with cancer (reviewed in [4, 5]). One study shows overexpression of several HR-associated genes in patients with non-small cell lung cancer . Approximately 5% of the human genome is comprised of large repetitive elements called low copy repeats (LCRs), also known as segmental duplications, which possess sufficiently high sequence identity to cause structural genomic instability via non-allelic homologous recombination (NAHR) between regions of identical sequence but differing genomic context, resulting in insertions, deletions, and translocations (, reviewed in ).
The most established means of detecting dysregulated homologous recombination, whether in cells with defective/deficient HR capacity, or in response to damage, is the sister chromatid exchange assay (SCE) that differentially stains sister chromatids, allowing for microscopic detection of the physical exchange of DNA which occurs with crossover HR . With the advent of straightforward techniques, the SCE assay has been in popular use since the 1970s for the purpose of identifying potential "chromosomal mutagenicity" of chemical agents . Chemicals that generate cross-linking of DNA are potent inducers of SCE, since HR is required to repair the resultant blockage during replication . Conditions and drugs which increase the number of single-strand breaks (SSBs) also increase the number of SCEs, presumably by overburdening the base-excision repair (BER) pathway such that unrepaired SSBs remain, become DSBs during replication, and must be repaired by homologous recombination . Accordingly, HeLa cells with downregulated XRCC1, a key component in the base excision repair pathway, show a 1.7-fold increase in SCEs, and an almost 2-fold increase when methyl methansulfonate (MMS), a DNA methylating agent, is added . Likewise, the thymidine analog 5-hydroxymethyl-2'-deoxyuridine (hmdUrd) at a 1 μM dose induces sister chromatid exchanges resulting in a 6-fold increase over background in Chinese hamster ovary (CHO) cells , again presumably through either saturation of BER activity, or through DNA replication across nicked BER intermediates. Inhibition or deficiency of poly(ADP-ribose) polymerase (PARP-1) also increases levels of sister chromatid exchange . Essentially, the HR pathway compensates, at least partially, for the defects or inadequacy of the BER response. PARP inhibitors can induce synthetic lethality in cells with mutations in BRCA-1 or BRCA-2, which are components of the HR pathway [15, 16]. Similarly, exposure to hmdUrd (or the related compound 5-chloro-2'-deoxyuridine) is synthetically lethal with loss of key BER components such as XRCC1 .
Despite the striking visible result upon staining, sister chromatid exchange is genetically silent. It represents a very large scale physical relocation of genetic material which is the consequence of a crossover recombination event; but there is no gain or loss of genetic information between two identical sisters. Presumably, these crossovers happen at the submicroscopic level as well. Our lab has developed an assay which measures non-silent, NAHR-mediated molecular level changes to genomic architecture by monitoring the stability of the length of gene clusters or tandemly repeated segmental duplications . For this gene cluster instability (GCI) assay, we use the gene clusters that produce the 45S precursor transcript to the 18S, 5.8S and 28S ribosomal RNA molecules. These clusters of tandemly repeated genes are located on the small arms of the five acrocentric chromosome pairs, representing a total of approximately 600 copies of the 43kb unit gene . We characterized the lengths of these rDNA gene clusters from healthy blood donors and found complete heterozygosity on each of the five chromosomes, and between the parental pairs of homologs. We also detected abundant evidence of both human meiotic  and mitotic rearrangement . We recently used the GCI assay to compare matched normal tissue to tumor tissue in patients with lung or colorectal cancer and found that approximately 50% of the tumors show changes in the sizes of the clusters compared to the normal tissue, as well as evidence of ongoing instability and heterogeneity within the tumor population indicating that HR has at some point become dysregulated within the tumor cells . Notably, loss or knockdown of the RecQ homolog defective in Bloom syndrome (BLM) causes a remarkable 100x increase in rDNA gene cluster instability rates along with the well-characterized 10-fold elevation in rates of sister chromatid exchange in these cells , suggesting elevated HR with crossing-over as the most likely mechanism  for this destabilization. We also demonstrated that loss of the ataxia-telangiectasia-mutated (ATM) protein causes a 10x elevation in rDNA gene cluster instability, even though loss of ATM in the absence of exogenous DNA damaging agents does not increase levels of sister chromatid exchange [24–26].
We are interested in the manner by which elevated and/or dysregulated recombination may be involved in the etiology of cancer and the development of chemotherapeutic resistance. We reasoned that elevated recombination could be caused either by an increase in recombination initiating lesions as the result of BER deficiency as seen in XRCC1 mutants, or by alterations in the downstream biochemistry of recombination causing an increase crossover vs. non-crossover recombination as seen in BLM mutants. Accordingly, we screened the NCI-60 panel of human cancer cell lines for defective BER by sensitivity to hmdUrd, and for altered recombination outcomes by the gene cluster instability assay. Lines exhibiting either phenotype were subsequently characterized by sister chromatid exchange in order to cross-compare three potential mitotic recombination indicators.
Results and Discussion
Gene Cluster Instability Survey
Lines with ongoing GCI: TK-10, K562, T-47D
In our previous work with ATM and BLM deficient cells [Killen, 2009 #1052], we estimated GCI rates in terms of observed new minor intensity bands per subclone. To a first approximation, wild-type cells had a GCI rate of 0.1 minor bands per subclone compared to rates of 1.0 and 10 minor bands per subclone in ATM and BLM deficient cells respectively. By this measure, the rates of instability in TK-10, K562, T-47D cells are more qualitatively similar to the 10x elevated rate in ATM-deficient cells over wild-type rates, rather than the more extreme 100x elevated rate in BLM-null cells.
Lines with historical GCI, but currently stable: A549, NCI-H23, EKVX
Lung cancer cell line A549 does not appear to demonstrate ongoing rDNA cluster instability (Figure 4). The major banding pattern from the clonal parental line is faithfully transmitted to each of the daughter lines with the exception of a single additional major band in subclone E (Figure 4, open triangle). Lung cancer line NCI-H23 shows similar results, with the disappearance of a single band in subclone D (Figure 4). All nine subclones also appear to have lost two bands that are non-stoichiometric but apparent in the parental line. We attribute this change to the technical details of culturing this specific line: NCI-H23 does not form single-cell-derived colonies easily, cells tend to migrate toward one another, and it may be that the parental population was not purely derived from expansion of a single cell but rather became mixed with an independent clone during the expansion process. We classify the third lung cancer cell line showing cluster length laddering, EKVX, as having low GCI on the basis of subclonal analysis (Figure 4). The parental line shows a single non-stoichiometric band thought to have arisen during expansion by mitotic recombination. Three of the subclonal lines appear to have inherited this band (subclones C, E, and G), while the other five (A, B, D, F, and H) did not.
Lack of overlap between GCI and defective mismatch repair
Nine of the NCI-60 collection of human tumor lines have defects in mismatch repair with resulting microsatellite instability (Figure 2). The lack of overlap between lines with gene cluster instability and mismatch repair defective lines emphasizes the mechanistic differences between these two aspects of genomic instability.
Sensitivity to hydroxymethyl deoxyuridine (hmdUrd): HOP-62, NCI-H23, SNB-19, U251, OVCAR-5, NCI/ADR-RES, 786-0
Sister Chromatid Exchange
Comparison of the two lines showing the greatest difference in median SCE/chromosome values, TK-10 (0.07) and U251 (0.13) indicates that the difference between these two lines is statistically significant (P = 0.017, Bonferroni corrected Mann-Whitney test). Nevertheless, we observed that, although the median number of chromosomes per metaphase in TK-10 is not out of line with values for the other lines, the chromosomes in TK-10 tended to be shorter on average than the other lines, which may limit our ability to score SCEs near the ends of these shorter chromosomes and artifactually reduce the TK-10 SCE values. A statistically significant difference involving pairwise comparisons between the other lines scored is not seen.
Levels of BLM Protein in Lines With Ongoing GCI
We were surprised to find that gene cluster instability in the NCI-60 panel of cells did not correspond with markedly elevated sister chromatid exchange. This observation establishes clear mechanistic differences between these two processes. Likewise, our observation that hmdUrd sensitivity in the NCI-60 panel of human cell lines did not correspond well to elevated SCE activity previously demonstrated in CHO cells may point to differences in physiology and metabolism between humans and rodents.
In summary, of the six lines we identified as definitively demonstrating rDNA GCI instability on initial screen, only three (T47D, TK10, and K562) continue to show evidence of ongoing instability at the present time. The other three appear to have accumulated gene cluster changes in the process of establishing and culturing the cell lines as the result of transient exogenous influences, rather than due to an endogenous mechanistic switch to an elevated gene cluster instability phenotype. These results are in good accord with our previous observation in gene cluster unstable human solid tumors  that half of the cases could be attributed to historical genotoxic damage, while half were more suggestive of a biochemical alteration towards an ongoing predisposition to gene cluster instability.
NCI-60 cell lines were acquired frozen from the National Cancer Institute's Developmental Therapeutics Program (DTP), and grown in RPMI-1640 culture medium supplemented with antibiotics and 5% fetal bovine serum. Cells were maintained in this medium, at 37C, in 5% CO2, in a humidified incubator for the duration of these experiments.
For subclonal GCI analysis, clonal cultures were initially generated from the bulk NCI-60 cell populations as received from the DTP by isolating single cells through limiting dilution followed by unrestricted expansion. For generating multiple subclonal cultures, single cells from the freely expanding clonal cultures were again isolated by limiting dilution followed by unrestricted expansion.
High Molecular Weight DNA Isolation
Single cell suspensions of 1x107 cells/mL in 0.8% low melting point were drawn into a 1-mL syringe, and chilled on ice until solidified. High molecular weight DNA was prepared from the solid-phase agarose cell suspension by digestion with 1% sarkosyl/500 mM EDTA/0.5 mg/mL proteinase K solution at 50C for at least 16 h, after which it was treated with phenylmethylsulfonyl fluoride in 10 mM Tris/1 mM EDTA (TE pH 8.0), extensively rinsed in TE/50% glycerol, and stored at -20C.
Gene Cluster Instability Analysis
The rRNA gene clusters were analyzed by pulsed-field gel electrophoresis and Southern blotting essentially as described . Briefly, approximately 1 μg of genomic DNA in solid phase was digested with EcoRV (New England Biolabs) overnight to release intact gene clusters from bulk genomic DNA. Digested DNA was placed into wells of a 1% Pulse Field Certified (Bio-Rad) agarose gel constituted in 0.5x TBE (44.5 mM Tris/44.5 mM boric acid/1.0 mM EDTA pH 8.0) and samples sealed into the wells using 0.8% low-melting-point agarose. Gels were run using a CHEF-MAPPER (Bio-Rad) at 14C, with two 6V/cm field vectors at 120° separation, with a switch time of from 3" to 90" over 24 hours using a 0.357 ramp factor. Following electrophoresis, gels were equilibrated to a final concentration of 0.5% glycerol in water and dried at 65C. Dried gels were rehydrated and the DNA denatured using 0.4 N NaOH/0.8 mM NaCl, neutralized in 0.5 M Tris pH 8.0/0.8 mM NaCl, then prehybridized at 65C in a hybridization buffer of 2x SSC (300 mM NaCl/30 mM Na-citrate) with 7% SDS and 0.5% casein. Gels were probed in fresh hybridization buffer with an rDNA-specific, 32P-labeled probe (radiolabeled PCR products amplified from a plasmid containing cloned human rDNA sequence using primers 5'-GGGCTCGAGATTTGGGACGTCAGCTTCTG and 5'-GGGTCTAGAGTGCTCCCTTCCTCTGTGAG) at 65C overnight. Southern probing in rehydrated dried gels maintains quantitative hybridization signal strength while avoiding documented difficulties involved in transferring DNA from pulsed-field gels onto membranes [32, 33]. Following two rinses in 2x SSC/1% SDS solution and two rinses in 0.5x SSC/1% SDS, the gels were imaged using a PhosphorImager (Molecular Dynamics).
Drug Sensitivity Screen
The in vitro growth inhibition screen of NCI-60 screen with a single 10 μM dose of 5-hydroxymethyl-2'-deoxyuridine (hmdUrd) was performed by the National Cancer Institute's Developmental Therapeutics Program according to the methods described online (http://dtp.nci.nih.gov/branches/btb/ivclsp.html). Briefly, cells are replica-plated in 96-well microtiter plates. After growth for 24 hours, time-zero plates are fixed with trichloroacetic acid (TCA) while experimental plates are treated with drug or left as untreated controls. Experimental plates are incubated for growth for an additional 48 hours before TCA fixation. Fixed cells are quantified spectrophotometrically by staining with sulforhodamine B and growth values are calculated as the mean of duplicate experiments.
Sister Chromatid Exchange Analysis
Sister chromatid exchanges were prepared using BrdU and visualized essentially as described . Individual metaphase spreads were photographed using bright-field microscopy and a 60x oil immersion objective.
Protein extracts were prepared using RIPA buffer as described previously . Following SDS-PAGE, gels were blotted onto Hybond-ECL nitrocellulose membrane (Amersham Biosciences). Proteins were detected using rabbit antibodies to BLM (Cell Signaling Technologies) and β-tubulin (NeoMarkers) with a horseradish peroxidase conjugated donkey anti-rabbit secondary antibody (Pierce) and an ECL Plus western blotting detection system (GE Healthcare).
The authors acknowledge the Markey Cancer Center for generous financial support.
- Anderson GR, Stoler DL, Brenner BM: Cancer: the evolved consequence of a destabilized genome. Bioessays. 2001, 23 (11): 1037-1046. 10.1002/bies.1149View ArticlePubMedGoogle Scholar
- Stoler DL, Chen N, Basik M, Kahlenberg MS, Rodriguez-Bigas MA, Petrelli NJ, Anderson GR: The onset and extent of genomic instability in sporadic colorectal tumor progression. Proc Natl Acad Sci USA. 1999, 96 (26): 15121-15126. 10.1073/pnas.96.26.15121PubMed CentralView ArticlePubMedGoogle Scholar
- Rothkamm K, Kruger I, Thompson LH, Lobrich M: Pathways of DNA double-strand break repair during the mammalian cell cycle. Mol Cell Biol. 2003, 23 (16): 5706-5715. 10.1128/MCB.23.16.5706-5715.2003PubMed CentralView ArticlePubMedGoogle Scholar
- Helleday T: Homologous recombination in cancer development, treatment and development of drug resistance. Carcinogenesis. 2010, 31 (6): 955-960. 10.1093/carcin/bgq064View ArticlePubMedGoogle Scholar
- Reliene R, Bishop AJ, Schiestl RH: Involvement of homologous recombination in carcinogenesis. Adv Genet. 2007, 58: 67-87.View ArticlePubMedGoogle Scholar
- Saviozzi S, Ceppi P, Novello S, Ghio P, Lo Iacono M, Borasio P, Cambieri A, Volante M, Papotti M, Calogero RA, Scagliotti GV: Non-small cell lung cancer exhibits transcript overexpression of genes associated with homologous recombination and DNA replication pathways. Cancer Res. 2009, 69 (8): 3390-3396. 10.1158/0008-5472.CAN-08-2981View ArticlePubMedGoogle Scholar
- Consortium IHGS: Finishing the euchromatic sequence of the human genome. Nature. 2004, 431 (7011): 931-945. 10.1038/nature03001View ArticleGoogle Scholar
- Gu W, Zhang F, Lupski JR: Mechanisms for human genomic rearrangements. Pathogenetics. 2008, 1 (1): 4- 10.1186/1755-8417-1-4PubMed CentralView ArticlePubMedGoogle Scholar
- Wilson DM, Thompson LH: Molecular mechanisms of sister-chromatid exchange. Mutat Res. 2007, 616 (1-2): 11-23.View ArticlePubMedGoogle Scholar
- Perry P, Evans HJ: Cytological detection of mutagen-carcinogen exposure by sister chromatid exchange. Nature. 1975, 258 (5531): 121-125. 10.1038/258121a0View ArticlePubMedGoogle Scholar
- Thompson LH: Unraveling the Fanconi anemia-DNA repair connection. Nat Genet. 2005, 37 (9): 921-922. 10.1038/ng0905-921View ArticlePubMedGoogle Scholar
- Fan J, Wilson PF, Wong HK, Urbin SS, Thompson LH, Wilson DM: XRCC1 down-regulation in human cells leads to DNA-damaging agent hypersensitivity, elevated sister chromatid exchange, and reduced survival of BRCA2 mutant cells. Environ Mol Mutagen. 2007, 48 (6): 491-500. 10.1002/em.20312View ArticlePubMedGoogle Scholar
- Kaufman ER: Induction of sister chromatid exchanges by the thymidine analog 5-hydroxymethyl-2'-deoxyuridine. Somat Cell Mol Genet. 1989, 15 (6): 563-568. 10.1007/BF01534917View ArticlePubMedGoogle Scholar
- Schultz N, Lopez E, Saleh-Gohari N, Helleday T: Poly(ADP-ribose) polymerase (PARP-1) has a controlling role in homologous recombination. Nucleic Acids Res. 2003, 31 (17): 4959-4964. 10.1093/nar/gkg703PubMed CentralView ArticlePubMedGoogle Scholar
- Ashworth A: A synthetic lethal therapeutic approach: poly(ADP) ribose polymerase inhibitors for the treatment of cancers deficient in DNA double-strand break repair. J Clin Oncol. 2008, 26 (22): 3785-3790. 10.1200/JCO.2008.16.0812View ArticlePubMedGoogle Scholar
- Kyle S, Thomas HD, Mitchell J, Curtin NJ: Exploiting the Achilles heel of cancer: the therapeutic potential of poly(ADP-ribose) polymerase inhibitors in BRCA2-defective cancer. Br J Radiol. 2008, 81 (Spec No 1): S6-11.View ArticlePubMedGoogle Scholar
- Horton JK, Watson M, Stefanick DF, Shaughnessy DT, Taylor JA, Wilson SH: XRCC1 and DNA polymerase beta in cellular protection against cytotoxic DNA single-strand breaks. Cell Res. 2008, 18 (1): 48-63. 10.1038/cr.2008.7PubMed CentralView ArticlePubMedGoogle Scholar
- Killen MW, Stults DM, Adachi N, Hanakahi L, Pierce AJ: Loss of Bloom syndrome protein destabilizes human gene cluster architecture. Hum Mol Genet. 2009, 18 (18): 3417-3428. 10.1093/hmg/ddp282View ArticlePubMedGoogle Scholar
- Henderson AS, Warburton D, Atwood KC: Location of ribosomal DNA in the human chromosome complement. Proc Natl Acad Sci USA. 1972, 69 (11): 3394-3398. 10.1073/pnas.69.11.3394PubMed CentralView ArticlePubMedGoogle Scholar
- Stults DM, Killen MW, Pierce HH, Pierce AJ: Genomic architecture and inheritance of human ribosomal RNA gene clusters. Genome Res. 2008, 18 (1): 13-18.PubMed CentralView ArticlePubMedGoogle Scholar
- Stults DM, Killen MW, Williamson EP, Hourigan JS, Vargas HD, Arnold SM, Moscow JA, Pierce AJ: Human ribosomal RNA gene clusters are recombinational hotspots in cancer. Cancer Res. 2009, 69 (23): 9096-9104. 10.1158/0008-5472.CAN-09-2680View ArticlePubMedGoogle Scholar
- Chaganti RS, Schonberg S, German J: A manyfold increase in sister chromatid exchanges in Bloom's syndrome lymphocytes. Proc Natl Acad Sci USA. 1974, 71 (11): 4508-4512. 10.1073/pnas.71.11.4508PubMed CentralView ArticlePubMedGoogle Scholar
- Wu L, Hickson ID: The Bloom's syndrome helicase suppresses crossing over during homologous recombination. Nature. 2003, 426 (6968): 870-874. 10.1038/nature02253View ArticlePubMedGoogle Scholar
- White JS, Choi S, Bakkenist CJ: Transient ATM kinase inhibition disrupts DNA damage-induced sister chromatid exchange. Sci Signal. 2010, 3 (124): ra44- 10.1126/scisignal.2000758PubMed CentralView ArticlePubMedGoogle Scholar
- Bartram CR, Koske-Westphal T, Passarge E: Chromatid exchanges in ataxia telangiectasia, Bloom syndrome, Werner syndrome, and xeroderma pigmentosum. Ann Hum Genet. 1976, 40 (1): 79-86. 10.1111/j.1469-1809.1976.tb00166.xView ArticlePubMedGoogle Scholar
- Galloway SM, Evans HJ: Sister chromatid exchange in human chromosomes from normal individuals and patients with ataxia telangiectasia. Cytogenet Cell Genet. 1975, 15 (1): 17-29. 10.1159/000130495View ArticlePubMedGoogle Scholar
- NEB: Restriction site densities deviating from expectations. 2010, New England BiolabsGoogle Scholar
- CGP: The Cancer Genome Project. 2008, Wellcome Trust Sanger InstituteGoogle Scholar
- Liscovitch M, Ravid D: A case study in misidentification of cancer cell lines: MCF-7/AdrR cells (re-designated NCI/ADR-RES) are derived from OVCAR-8 human ovarian carcinoma cells. Cancer Lett. 2007, 245 (1-2): 350-352. 10.1016/j.canlet.2006.01.013View ArticlePubMedGoogle Scholar
- Batist G, Tulpule A, Sinha BK, Katki AG, Myers CE, Cowan KH: Overexpression of a novel anionic glutathione transferase in multidrug-resistant human breast cancer cells. J Biol Chem. 1986, 261 (33): 15544-15549.PubMedGoogle Scholar
- Boorstein RJ, Teebor GW: Effects of 5-hydroxymethyluracil and 3-aminobenzamide on the repair and toxicity of 5-hydroxymethyl-2'-deoxyuridine in mammalian cells. Cancer Res. 1989, 49 (6): 1509-1514.PubMedGoogle Scholar
- Leach TJ, Glaser RL: Quantitative hybridization to genomic DNA fractionated by pulsed-field gel electrophoresis. Nucleic Acids Res. 1998, 26 (20): 4787-4789. 10.1093/nar/26.20.4787PubMed CentralView ArticlePubMedGoogle Scholar
- Lee H, Birren B, Lai E: Ultraviolet nicking of large DNA molecules from pulsed-field gels for southern transfer and hybridization. Anal Biochem. 1991, 199 (1): 29-34. 10.1016/0003-2697(91)90265-UView ArticlePubMedGoogle Scholar
- Perry P, Wolff S: New Giemsa method for the differential staining of sister chromatids. Nature. 1974, 251 (5471): 156-158. 10.1038/251156a0View ArticlePubMedGoogle Scholar
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.