How real are our data? Copy number variation in lymphoblastoid and other cell lines


Human genetic studies require human samples, and obtaining these samples, with sufficient material for experiments over time, is a challenge. Many of us have come to rely on lymphoblastoid cell lines (LCLs). These transformed cells are relatively easy to establish and provide an almost unlimited supply of cells from individuals. Use of these cells prevents researchers from having to continuously go back to obtain biological material from their subjects. But how representative are these cells of the genomic content of the parental cells from which they were derived?

In this issue, Shirley et al. (Hum Mutat 33:1075–1086, 2012) analyze LCLs and their parent blood cells to determine the concordance for chromosomal variation. They used multiple comparisons between similarly derived LCLs and peripheral blood cells. They also studied 3 differentiated cell lines (fibroblasts, keratinocytes and melanocytes) derived from the same foreskin sample. Furthermore, they obtained genotyping data from an available data set that included replicates and used the genotyping data to analyze copy number variation.

Overall 26 of 29 LCLs prepared from the 5 individuals were found to have SNP genotyping results highly similar to those derived from peripheral blood. In the 3 remaining LCL cell lines, there were mosaic differences not seen in peripheral blood. Notably, the frequency of mosaicism was higher in the LCL lines than in the differentiated cell lines or in the genotyped dataset, and mosaicism was found for both regions of homozygosity and aneuploidy. In addition, the alterations in the LCLs did not appear to be random and included loci with immunoglobulin genes. The authors catalogued these LCL-specific alterations as a resource for other investigators.

One can conclude that LCLs are a pretty good proxy for peripheral blood samples, with the caveat that change does happen, and this generally appears in a mosaic form (making it easier to recognize). However, the frequency of mosaic changes is high enough that it is worthwhile to routinely characterize them via SNP array genotyping to monitor any changes, especially for experiments that involve gene expression or pharmacogenomic analysis.