Cell lines
Four hundred and forty-seven Epstein-Barr virus (EBV)-transformed LCLs derived from 34 Caucasian Utah CEPH families of northern and western European descent (1331, 1332, 1333, 1334, 1340, 1341, 1344, 1345, 1346, 1347, 1349, 1350, 1356, 1358, 1362, 1375, 1377, 1408, 1413, 1416, 1418, 1420, 1421, 1423, 1424, 1444, 1447, 1454, 1459, 1463, 13291, 13292, 13293, 13294) were used for heritability and linkage analyses. Fifty-seven out of 60 unrelated cell lines (from the same families) from the HapMap Project (GM06985, GM06993, GM06994, GM07000, GM07022, GM07034, GM07055, GM07056, GM07345, GM07357, GM11829, GM11830, GM11831, GM11832, GM11839, GM11840, GM11881, GM11882, GM11992, GM11993, GM11994, GM11995, GM12003, GM12004, GM12005, GM12006, GM12043, GM12044, GM12056, GM12057, GM12144, GM12145, GM12146, GM12154, GM12155, GM12156, GM12234, GM12239, GM12248, GM12249, GM12264, GM12750, GM12751, GM12760, GM12761, GM12762, GM12763, GM12812-GM12815, GM12872, GM12873, GM12874, GM12875, GM12891, GM12892) and 11 unrelated cell lines from the Perlegen Project (GM06990, GM07349, GM10842-10845, GM10848, GM10850, GM10852, GM10853, GM10858) were phenotyped for sensitivity to carboplatin. Cells were purchased from Coriell Institute for Medical Research http://ccr.coriell.org/Sections/Collections/NIGMS?SsID=8. Two HapMap samples (GM12716 and GM12717) were not phenotyped owing to the inability to obtain > 85 per cent viability. Additionally, another sample (GM12236) was not phenotyped because it was not available from Coriell at the time of phenotyping, owing to the cell line not meeting all quality control standards at Coriell (eg slow growth). Cell lines were cultured in RPMI 1640 medium containing 15 per cent heat-inactivated foetal bovine serum (Hyclone, Logan, UT, USA) and 20 mM L-glutamine. Cell lines were diluted three times per week at a concentration of 300,000-350,000 cells/ml and were maintained in a 37°C, 5 per cent CO2 humidified incubator. Medium and components were purchased from Cellgro (Herndon, VA, USA).
Drugs
Cis diamine (1,1-cyclobutanedicarboxylato) platinum (carboplatin) was purchased from Sigma Chemical Co. (St Louis, MO, USA). Figure 1 illustrates the structure of carboplatin. Carboplatin was prepared as a 20 mM stock in water, filter sterilised and diluted in medium immediately prior to addition to cells.
Cell growth inhibition
Cytotoxicity was assessed in the absence (control) and presence of increasing concentrations of carboplatin, utilising a high-throughput alamarBlue™(Biosource International, Camarillo, CA, USA) assay, as previously described [14]. Final concentrations of carboplatin used for the heritability and linkage analyses were 10, 20, 40 and 80 μM, with an exposure time of 72 hours. Cytotoxicity measurements were performed in triplicate for each drug concentration per experiment, with two to three experiments per cell line. Final cytotoxicity values were averaged from at least six replicates taken from two separate experiments. IC50, the concentration required to inhibit 50 per cent of cell growth, was calculated for each cell line by curve fitting each concentration using Microsoft Excel™software. For cytotoxicity studies following small interfering RNA (siRNA) knockdown, cells were plated one hour after transfection and treated with 20, 40, 80, 160, 320 and 640 μM carboplatin for 48 hours to expose cells to drug at the time of greatest CD44 mRNA and protein knockdown. Three independent knockdown experiments were performed in triplicate for each of two cell lines with high CD44 expression (GM06985 and GM11881).
Heritability analysis
Heritability analysis was performed using Sequential Oligogenic Linkage Analysis Routines (SOLAR; http://www.sfbr.org/solar/) computer software to estimate narrow sense heritability at each carboplatin concentration [18]. SOLAR uses likelihood ratio tests to evaluate heritability by comparing a purely polygenic model with a sporadic model in the case of testing heritability. This analysis is used to determine whether there is a significant proportion of cytotoxicity at each dose that can be explained by genetic factors. All phenotype data were transformed using the inverse normalisation of the percentile rank function in Microsoft Excel™software. Covariates such as age, sex and the age-sex interaction were tested as previously described [14].
Linkage analysis
MERLIN [19] was used to perform non-parametric quantitative trait locus (QTL) linkage analysis, which is robust to non-normally distributed data. The genotypic data were downloaded from the CEPH database http://www.cephb.fr/cephdb/ and the Marshfield map database http://research.marshfieldclinic.org/genetics/GeneticResearch/comp Maps.asp. The CEPH database (version 10) contains genotypic information on over 32,000 genetic markers and 9,000 microsatellite markers typed in the CEPH LCLs. Overall, this database manages over 6 million genotypes, which can be utilised for genetic studies. The Marshfield map database was used to determine microsatellite locations within particular chromosomal regions of interest. The genetic markers were error checked for Mendelian incompatibility, mis-specified relationship and unlikely recombinations, using a platform for linkage analysis as previously described [13, 14]. The SNP data were downloaded from the SNP Consortium http://www.snp.cshl.org. From the combined pool of SNP and microsatellite markers genotyped in the above databases, approximately 7,209 non-redundant markers were selected based on the availability of genotypes in at least 50 per cent of family members. Physical positions of selected microsatellite and SNP markers were found using Build 36 of the UCSC Genome Browser (http://www.genome.ucsc.edu.). Genetic maps were constructed based on microsatellite and SNP positions in the Marshfield map. These highly heterozygous markers, yielding a dense genetic map, were utilised for the analysis. Age and sex were also included as covariates in the analysis.
Association analysis
SNP genotypes were downloaded from the International HapMap Project http://hapmap.org (release 20) and Perlegen http://perlegen.com genotype databases for the 68 unrelated CEPH/UTAH (57 HapMap plus 11 Perlegen) samples. To exclude possible genotyping errors in overlapping samples from the HapMap and Perlegen datasets, 23,189 Perlegen SNPs with Mendelian transmission errors and 13,803 SNPs with inconsistent genotypes were removed. A total of 277,000 SNPs, located 5 kilobases upstream/downstream of genes, in addition to within genes, were used for the analysis.
The single-locus analysis was performed using one-way analysis of variance (ANOVA), with Bartlett's test for equal variances cutoff as p = 0.01, using R-2.3.1. Since multivariate normality of the phenotypic data is a critical assumption of ANOVA, the IC50 phenotype was transformed into normal distributions using log2 transformations. ANOVA was then performed on unrelated samples to assess the non-linear relationship among the genotypes. False discovery rate (FDR) was used to control for multiple testing within the phenotype using the Q Value package of R-2.3.1 http://www.r-project.com. An association p value with an FDR of less than 5 per cent was defined as significant.
Gene expression
Baseline gene expression values for HapMap CEPH samples were determined as previously described [20] using the Affymetrix GeneChip Human Exon 1.0 ST array.
Quantitative real-time polymerase chain reaction (qRT-PCR)
Up to 5 × 106 total number of cells from ten LCLs with the highest carboplatin IC50 values (IC50 range: 91.3-34.7) and ten LCLs with the lowest carboplatin IC50 values (IC50 range: 10.4-13.3) (out of the 68 unrelated HapMap and Perlegen samples) were evaluated for CD44 baseline expression. The LCLs with the lowest IC50 values were: GM10845, GM10853, GM12006, GM12043, GM12145, GM12146, GM12812, GM12813, GM12815 and GM12872. The LCLs with the highest IC50 values were: GM06985, GM11830, GM11881, GM12044, GM12056, GM12751, GM12761 and GM12891. LCLs in exponential growth were pelleted, washed three times with ice cold phosphate-buffered saline, flash frozen in liquid nitrogen, and stored in a freezer at -80°C until required. Total RNA was isolated from the cells using the Qiagen RNeasy Mini kit (Valencia, CA, USA), according to the manufacturer's protocol. Gene expression was quantified with the Roche LightCycler SYBR Green I kit (Indianapolis, IN, USA) and samples were run (one-step PCR) on the 7900 HT Fast Real Time PCR System from Applied Biosciences (Foster City, CA, USA). PCR primers were designed and tested for gene homology using the UCSC Genome Browser http://www.genome.ucsc.edu and BLAST http://www.ncbi.nlm.nih.gov/BLAST programs. Primers were designed to be approximately 20-25 base pairs in length, have 50-60 per cent GC content and melting temperatures of 55-65°C. CD44 primers were designed to capture all isoforms with the forward primer 5'-GCATTGCAGTCAACAGTCGAAGAAG-3' and the reverse primer 5'-GG CCTCTCCGTTGAGTCCACTT-3'. Lamin A/C primer sequences included forward primer 5'-ATG ATCGCTTGGCGGTCTAC-3' and reverse primer 5'-GCCCTGCGTTCTCCGTTT-3'. Beta-actin was used as an endogenous control with the forward primer 5'-ATTGCCGACAGGATGCAGA-3' and reverse primer 5'-GCTCAGGAGGAGCAATGAG CTT-3'. Standard curves were prepared for all genes, with fivefold dilutions ranging from 2,000 ng/μl to 0.64 ng/μl. Standard curves were only used if they had r2 values > 0.985. CD44 and beta-actin were run under identical conditions, with parameters as follows: Step 1: 55° × 1,800 seconds; Step 2: 95° × 600 seconds; Step 3: 95° × ten seconds, 58° × 15 seconds, 72° × 20 seconds (repeated 45 times). Lamin A/C primers were run under the following conditions: Step 1: 55° × 1,800 seconds; Step 2: 95° × 600 seconds; Step 3: 95° × 15 seconds, 60° × 60 seconds, 72° × six seconds (repeated 45 times). Melt curve analysis was performed for each run. Three separate experiments were run; within each experiment, samples were run in duplicate. Student's t-test was used for comparison of normalised CD44 expression levels between cells with high and low carboplatin IC50 values.
siRNA knockdown of CD44
LCLs (GM06985 and GM11881) were grown to 7 - 9 × 105 cells/ml and transfected 24 hours after diluting with an Amaxa Nucleofector I device, according to the manufacturer's recommendations (Amaxa Biosystems Inc., Gaithersburg, MD, USA). Electrical parameter X-01 and Nucleofector kit V were used for the single cuvette transfection process. Nucleofected cells were immediately transferred into a flask containing 3 ml medium and then placed in a humidified incubator at 37°C, 5 per cent CO2. Transfection efficiency, typically between 70 and 80 per cent, was determined by flow cytometry using fluorescently labelled siGLO siRNA directed against lamin A/C from Dharmacon (Lafayette, CO, USA). Transfection efficiency was also assessed by mRNA knockdown of lamin A/C 24 hours post-transfection using qRT-PCR conditions as described above. CD44 On-Targetplus pool siRNA and control On-Targetplus siCONTROL non-targeting pool siRNA (Dharmacon) were used. All siRNAs were used at equal concentrations (600 pmol). Typical mRNA knockdown of CD44, as compared with non-targeting control, was assessed after 24 hours.
Western blotting
CD44 protein levels were analysed in total cell extracts prepared by lysing the cells in RIPA buffer in the presence of protease inhibitors (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Protein concentrations were analysed at 595 nm using Bio-Rad's (Hercules, CA, USA) Bradford protein assay. Since CD44 is highly expressed in LCLs, 1 μg total protein was separated on a 4-20 per cent tris-glycine gel (Bio-Rad) under non-reducing conditions and transferred to a polyvinylidene difluoride (PVDF) membrane. The membrane was blocked for at least one hour in 3 per cent milk. Primary CD44 (recognising the extracellular N-terminal domain) and β-actin antibodies (Abcam, Cambridge, MA, USA) were added in 1:5,000 dilutions overnight at 4°C. Proteins were visualised using horseradish peroxidase-conjugated secondary antibodies (1:10,000 dilutions) from Millipore (St Charles, MO, USA) and ECL plus Western Blotting Detection Reagent (General Electric Healthcare, Piscataway, NJ, USA). CD44 protein levels were measured 48-96 hours after siRNA transfection.
Statistical analysis
IC50 values, from at least three separate knockdown experiments, were compared using Student's t-test. To improve the statistical power, the cellular survival results across the two cell lines (GM06985 and GM11881) were combined by performing a mixed model ANOVA, with cell line as a random effect and dose and experimental condition (CD44 siRNA versus non-targeting group) as fixed effects. The interaction between dose and experimental condition was also tested.