Lymphoblastoid cell lines (LCLs) are widely used in pharmacogenomic studies and the applicability of LCLs for various clinical phenotypes is emerging. Early studies have yielded promising results for LCLs as a proxy for genetic variants in treatment outcome for a number of cancers, as well as toxicity in varying tissue types including taxane-induced neuropathy. Although LCLs have demonstrated utility in the elucidation of functional mechanisms for results of clinical genotype–drug phenotype studies, there are more relevant cell-based models developing.
Utility of LCLs as a resource in pharmacogenomic discovery
The most relevant system to determine how an individual‘s genetic profile influences their likelihood of responding to a drug or developing toxicity would be humans; however, performing appropriate pharmacogenomic trials requires a homogenous population of patients treated with the same dosage regimen with minimal confounding variables. Depending on the effect size, an adequately powered study could require thousands of patients. Furthermore, a replication trial is often necessary to increase confidence in the findings, particularly if the genotypes are being considered for clinical implementation. In the field of oncology, this is difficult – if not impossible – because the standard of care tends to change as new therapies emerge and for most cancers, patients receive a multidrug regimen. However, effective pharmacogenomic prediction in oncology would potentially have great utility because nonresponse and adverse events can be life threatening.
An approach for the identification of predictive markers that has emerged to overcome the difficulty of performing large clinical trials is the use of cell-based models. Epstein–Barr virus-transformed LCLs are a widely used model including those from the International HapMap Project with publicly available genetic data Citation[1,2]. Studies using LCLs from 11 different world populations to address interindividual and interpopulation comparisons have included a variety of cellular phenotypes that go beyond chemotherapeutic-induced cytotoxicity and apoptosis, including LDLR levels in response to simvastatin treatment Citation[3], Salmonella-induced pyroptosis Citation[4], radiation-induced expression changes Citation[5] and other phenotypes (reviewed in Wheeler and Dolan Citation[1]). In addition to genetic (genotype and sequencing) data in the public domain, there are also baseline expression data using various array platforms Citation[6–8], baseline miRNA data Citation[9] and methylation data Citation[10]. Genetic variants identified in genome-wide association studies (GWAS) can be identified as potentially functional through their association with baseline mRNA expression, miRNA expression or modified cytosines. The cells provide a cost-effective testing system in which environmental factors, such as concomitant medications, can be controlled Citation[1].
Although there are a number of advantages of HapMap LCLs, the limitations include confounding variables, such as cellular proliferation rate, Epstein–Barr virus copy number and lack of expression of CYP450 enzymes; also, the fact that they represent one cell type has been a concern (reviewed previously Citation[1,2]). The fundamental concern of whether these LCLs recapitulate clinical outcomes is being addressed through various studies comparing LCL results with clinical outcomes.
Applicability of LCLs to clinically relevant phenotypes
Since LCLs arise from B lymphocytes, one might assume they better represent blood-based toxicities, such as neutropenia, than toxicities in major organs, such as nephrotoxicity or cardiotoxicity. Different expression patterns in tissues may account for differences in the degree to which they are damaged by outside stimuli Citation[11]. To no surprise, there is even greater skepticism that LCLs can identify relevant genetic variants predictive of tumor response because of the many somatic changes and heterogeneity within tumors. However, targets discovered through pharmacogenomic studies in LCLs have been replicated in clinical trials, arguably the ultimate measure of utility Citation[12,13]. Top SNPs from LCLs associated with antileukemics were associated with clinical response in acute myeloid leukemia Citation[14]. Those associated with platinum susceptibility were also associated with outcomes in head and neck cancer patients Citation[15]; a top carboplatin SNP from LCLs was found to be significantly associated with greater cellular sensitivity to carboplatin, and when tested in ovarian cancer patients, was associated with decreased progression-free survival and poorer overall survival Citation[16]. A recent study evaluating temozolomide response identified a SNP in the LCLs as associated with differential gene expression of MGMT (p < 10-26) that was also associated with differential methylation in glioblastoma samples Citation[17].
In addition, functional validation of clinical GWAS results in LCLs is also showing great promise. In cardiovascular pharmacogenomics, a haplotype associated in patients with poor simvastatin response was functionally evaluated in LCLs to discover a potential mechanism through alternative splicing in which the associated haplotype affected patient response Citation[18]. In asthma pharmacogenomics, discovery studies in large patient population identified a genetic variant located in the promoter of GLCCI1 that was functionally evaluated in LCLs. Utilizing LCLs, the variant was shown to influence expression of GLCCI1 and dexamethasone was shown to induce expression of GLCCI1 Citation[19].
Understanding how well the overall genetic architecture of pharmacologic phenotypes can be captured by the cell-based models is of paramount importance. One approach recently developed that goes beyond evaluating only the strongest associations from LCLs in a patient population employed a permutation resampling analysis. The enrichment approach evaluates all SNPs with nominally significant associations with cytotoxicity in studies in cell-based models against those at least nominally significant in human clinical trials Citation[20]. This was applied to GWAS results associated with paclitaxel-induced cytotoxicity in the LCL model and GWAS results of paclitaxel-induced neuropathy in patients. SNPs associated with patient paclitaxel-induced neuropathy were found to be enriched for SNPs associated with paclitaxel-induced cytotoxicity in HapMap LCLs and vice versaCitation[21]. This significant enrichment confirms that LCLs are a useful model in the study of a subset of shared genes involved in patient toxicity. As more clinical GWAS results become available, the robustness of the LCLs for additional tissue types and toxicities can be evaluated.
Conclusion
The advantage to using cell-based models for pharmacogenomics discovery is the ability to perform experiments outside of humans in a well-controlled environment where in vivo confounders such as concomitant medications are eliminated. International HapMap LCLs also have the added advantage of a wealth of genetic, epigenetic, expression and pharmacologic data available. The full breadth of pharmacologic phenotypes that can be assessed in LCLs is not yet appreciated; however, researchers have evaluated cell growth inhibition, apoptosis, gene-expression modulation following drug treatment and other pharmacologic end points. Evidence comparing genetic predictors of patient outcomes with genetic predictors arising from LCL models has indicated robustness for the cell-based model. Genetic variants identified in LCLs correspond to treatment response in a number of diverse cancers include ovarian, head and neck, and leukemia, and in toxicities such as peripheral neuropathy associated with paclitaxel. The use of enrichment studies between clinical GWAS signals in patient studies and LCL GWAS signals for a particular drug will help identify the utility of the cell-based model.
Future perspective
As more pharmacogenomic clinical trials are completed and compared with LCL results for both overlap among top signals and enrichments among suggestive signals, we will have a better sense of where the LCLs are most useful in pharmacogenomic discovery. In general, clinical pharmacogenomics studies are hindered by the availability of well-phenotyped data in patients. The robustness of LCL predictors generated with the same agent will continue to be assessed clinically. To some degree, the applicability of LCLs in the discovery phase of genetic variants is primarily with oncology agents.
As studies linking genetic variants to mRNA, methylation, miRNA and other functional units continue to grow, LCLs remain an attractive system to functionally evaluate genetic variants that are robustly associated with clinical phenotypes. Follow-up studies in LCLs can link functional annotation to potential mechanism of associated signals for clinical studies. Furthermore, bringing in protein expression to the vast knowledge on these cells is an area of great interest. The application of microwestern arrays, which represent the latest generation of proteomic analysis methods for the large-scale analysis of protein abundance and modification, will add great value to these cell-based models Citation[22]. By using the microwestern array method one might uncover additional loci that may influence the translation or degradation rate of proteins that are key, high-level executors of cell decision-making processes.
Another area of potential growth in pharmacogenomics is the use of well-genotyped induced pluripotent stem cells that can then be converted to relevant tissues including neurons, hepatocytes and cardiomyocytes. These cell types have the potential for both discovery of genotype–phenotype relationships in tissues of toxicity and functional evaluation of potential mechanisms. A large repertoire of cell lines with a known genetic background from different tissues of origin would be extremely beneficial to the field of pharmacogenomics.
Financial & competing interests disclosure
The authors are supported by NIH/NIGMS Pharmacogenomics of Anticancer Agents grant U01GM61393, University of Chicago SPORE grant NCI P50CA125183, R21HG006367, LLSA SCOR grant and NIH CA136765. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
Additional information
Funding
References
- Welsh M , MangraviteL, MedinaMW et al. Pharmacogenomic discovery using cell-based models. Pharmacol. Rev. 61(4) , 413–429 (2009).
- Wheeler HE , DolanME. Lymphoblastoid cell lines in pharmacogenomic discovery and clinical translation. Pharmacogenomics13(1) , 55–70 (2012).
- Mangravite LM , MedinaMW, CuiJ et al. Combined influence of LDLR and HMGCR sequence variation on lipid-lowering response to simvastatin. Arterioscler. Thromb. Vasc. Biol. 30(7) , 1485–1492 (2010).
- Ko DC , GamazonER, ShuklaKP et al. Functional genetic screen of human diversity reveals that a methionine salvage enzyme regulates inflammatory cell death. Proc. Natl Acad. Sci. USA 109(35) , e2343–e2352 (2012).
- Smirnov DA , MorleyM, ShinE, SpielmanRS, CheungVG. Genetic analysis of radiation-induced changes in human gene expression. Nature459(7246) , 587–591 (2009).
- Zhang W , DuanS, KistnerEO et al. Evaluation of genetic variation contributing to differences in gene expression between populations. Am. J. Hum. Genet. 82(3) , 631–640 (2008).
- Pickrell JK , MarioniJC, PaiAA et al. Understanding mechanisms underlying human gene expression variation with RNA sequencing. Nature 464(7289) , 768–772 (2010).
- Stranger BE , NicaAC, ForrestMS et al. Population genomics of human gene expression. Nat. Genet. 39(10) , 1217–1224 (2007).
- Huang RS , GamazonER, ZiliakD et al. Population differences in microRNA expression and biological implications. RNA Biol. 8(4) , 692–701 (2011).
- Bell JT , PaiAA, PickrellJK et al. DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biol. 12(1) , R10 (2011).
- Caliskan M , CusanovichDA, OberC, GiladY. The effects of EBV transformation on gene expression levels and methylation profiles. Hum. Mol. Gen.20(8) , 1643–1652 (2011).
- French D , YangW, ChengC et al. Acquired variation outweighs inherited variation in whole genome analysis of methotrexate polyglutamate accumulation in leukemia. Blood 113(19) , 4512–4520 (2009).
- Ellsworth KA , MoonI, EckloffBW et al. FKBP5 genetic variation: association with selective serotonin reuptake inhibitor treatment outcomes in major depressive disorder. Pharmacogen. Genomics23(3) , 156–166 (2013).
- Mitra AK , CrewsK, PoundsS et al. Impact of genetic variation in FKBP5 on clinical response in pediatric acute myeloid leukemia patients: a pilot study. Leukemia 25(8) , 1354–1356 (2011).
- Ziliak D , O‘DonnellPH, ImHK et al. Germline polymorphisms discovered via a cell-based, genome-wide approach predict platinum response in head and neck cancers. Trans. Res. 157(5) , 265–272 (2011).
- Huang RS , JohnattySE, GamazonER et al. Platinum sensitivity-related germline polymorphism discovered via a cell-based approach and analysis of its association with outcome in ovarian cancer patients. Clin. Cancer Res. 17(16) , 5490–5500 (2011).
- Brown CC , HavenerTM, MedinaMW et al. A genome-wide association analysis of temozolomide response using lymphoblastoid cell lines shows a clinically relevant association with MGMT. Pharmacogen. Genomics 22(11) , 796–802 (2012).
- Medina MW , GaoF, RuanW, RotterJI, KraussRM. Alternative splicing of 3-hydroxy-3-methylglutaryl coenzyme A reductase is associated with plasma low-density lipoprotein cholesterol response to simvastatin. Circulation118(4) , 355–362 (2008).
- Tantisira KG , Lasky-SuJ, HaradaM et al. Genomewide association between GLCCI1 and response to glucocorticoid therapy in asthma. N. Engl. J. Med. 365(13) , 1173–1183 (2011).
- Cox NJ , GamazonER, WheelerHE, DolanME. Clinical translation of cell-based pharmacogenomic discovery. Clin. Pharmacol. Ther.92(4) , 425–427 (2012).
- Wheeler HE , GamazonER, WingC et al. Integration of cell line and clinical trial genome-wide analyses supports a polygenic architecture of paclitaxel-induced sensory peripheral neuropathy. Clin. Cancer Res. 19(2) , 491–499 (2013).
- Ciaccio MF , WagnerJP, ChuuCP, LauffenburgerDA, JonesRB. Systems analysis of EGF receptor signaling dynamics with microwestern arrays. Nat. Methods7(2) , 148–155 (2010).