Age, Sex, and Genetic Architecture of Human Gene Expression in EBV Transformed Cell Lines

Manuel A. Rivas, Mark J. Daly, Itsik Pe’erarXiv 2009 · DOI · arXiv

Abstract

Individual expression profiles from EBV transformed cell lines are an emerging resource for genomic investigation. In this study we characterize the effects of age, sex, and genetic variation on gene expression by surveying public datasets of such profiles. We establish that the expression space of cell lines maintains genetic as well as non-germline information, in an individual-specific and cross-tissue manner. Age of donor is associated with the expression of 949 genes in the derived cell line. Age-associated genes include over-representation of immune-related genes, specifically MHC Class I genes, a phenomenon that replicates across tissues and organisms. Sex associated genes in these cell lines include likely candidates, such as genes that escape X-inactivation, testis specific expressed genes, androgen and estrogen specific genes, but also gene families previously unknown to be sex associated such as common microRNA targets (MIR-490, V_ARP1_01, MIR-489). Finally, we report 494 transcripts whose expression levels are associated with a genetic variant in cis, overlapping and validating previous reports. Incorporating age in analysis of association facilitates additional discovery of trans-acting regulatory genetic variants. Our findings promote expression profiling of transformed cell lines as a vehicle for understanding cellular systems beyond the specific lines.

Introduction

In modern genomics, Epstein-Barr virus (EBV) transformed lymphoblastoid cell lines (LCLs) are frequently used as a proxy for human biology due to their accessibility and ease of maintenance. However, a persistent concern in the field is the extent to which the "immortalization" process—the transformation of primary cells into a self-perpetuating laboratory resource—might obscure the original biological signatures of the donor. If individual traits like age and sex are erased during transformation, LCLs lose their utility as models for aging or sex-specific disease manifestation.

This study investigates whether these individual signatures survive the transformation process and how accounting for them can improve the detection of expression quantitative trait loci (eQTLs). By analyzing public datasets, the researchers demonstrate that LCLs retain a significant "biological memory" of their donors, which, if properly modeled, can reveal a more accurate picture of the genetic architecture of gene expression.

Methodology and Data Integration

To isolate the signals of age, sex, and genetics from the noise of the transformation process, the researchers utilized a multi-faceted computational approach:

  • Linear Regression and Genomic Control: Expression levels were modeled as functions of age, sex, and family structure. To account for the relatedness within pedigrees (such as the CEPH families), genomic control methods were applied to prevent false-positive associations.
  • Meta-Analysis and Imputation: The study combined data across four HapMap populations (CEPH, YRI, JPT, and CHB) using Fisher’s exact test. To increase the resolution of genetic mapping, the team integrated microsatellite and SNP data to infer genotypes for 3 million markers.
  • Cross-Tissue Validation: To ensure that the findings were not artifacts of the cell culture environment, the researchers compared LCL expression patterns with independent human kidney expression data.

The Persistence of Age and Sex Signatures

Biological Memory of Aging

The study identified 949 genes whose expression levels in LCLs were significantly associated with the donor's chronological age. Most notably, there was a consistent increase in the expression of MHC Class I genes with age. This signature was not unique to LCLs; it was also observed in human kidney tissue and has been documented in other organisms like mice. This suggests that certain aging mechanisms are deeply "hard-wired" into the cellular machinery and persist even after the cells are removed from the donor and immortalized.

Sex-Specific Regulation

Beyond the expected expression of genes on the sex chromosomes or those that escape X-inactivation, the researchers discovered sex-associated patterns in autosomal genes. Interestingly, these patterns included targets of specific microRNAs, such as MIR-490 and MIR-489. This finding suggests a layer of sex-based gene regulation in LCLs that was previously unrecognized, potentially mediated by microRNA activity.

Genetic Architecture and eQTL Discovery

By treating age and sex as covariates rather than noise, the researchers were able to refine the search for genetic variants that regulate gene expression. They identified 494 transcripts associated with a genetic variant in cis (nearby).

Furthermore, incorporating age into the association analysis facilitated the discovery of trans-acting (long-distance) regulatory variants. A key example is the discovery of a variant (rs10928136) in the LRP1B gene that interacts with age to influence the expression of APOE, a gene critical to Alzheimer’s disease and human longevity. This highlights how biological variables like age can act as modifiers of genetic effects, and ignoring them may mask important regulatory relationships.

Discussion and Limitations

While this study confirms that LCLs are a powerful vehicle for understanding human cellular systems, several areas remain for future exploration. The authors note that while some signatures are cross-tissue, many are likely specific to the lymphocyte lineage. Additionally, while the study hypothesizes that epigenetic mechanisms like DNA methylation might be the medium for this "biological memory," this was not directly tested.

Future research with larger sample sizes will be necessary to detect more subtle trans-acting effects and to explore how other environmental factors—such as diet, stress, or disease history—might also be recorded within the expression profiles of immortalized cell lines.

Conclusion

This research establishes that EBV-transformed cell lines are not biological blank slates. They retain significant information about the donor's age and sex, reflecting broader physiological processes. By accounting for these factors, genomic studies can achieve higher precision, uncovering the complex interplay between our environment, our age, and our genetic blueprint.

Summary updated

Abstract

Individual expression profiles from EBV transformed cell lines are an emerging resource for genomic investigation. In this study we characterize the effects of age, sex, and genetic variation on gene expression by surveying public datasets of such profiles. We establish that the expression space of cell lines maintains genetic as well as non-germline information, in an individual-specific and cross-tissue manner. Age of donor is associated with the expression of 949 genes in the derived cell line. Age-associated genes include over-representation of immune-related genes, specifically MHC Class I genes, a phenomenon that replicates across tissues and organisms. Sex associated genes in these cell lines include likely candidates, such as genes that escape X-inactivation,testis specific expressed genes, androgen and estrogen specific genes, but also gene families previously unknown to be sex associated such as common microRNA targets (MIR-490, V_ARP1_01, MIR-489). Finally, we report 494 transcripts whose expression levels are associated with a genetic variant in cis, overlapping and validating previous reports. Incorporating age in analysis of association facilitates additional discovery of trans-acting regulatory genetic variants. Our findings promote expression profiling of transformed cell lines as a vehicle for understanding cellular systems beyond the specific lines.
BibTeX
@misc{rivas2009age,
  author = {Manuel A. Rivas and Mark J. Daly and Itsik Pe’er},
  title = {{Age, Sex, and Genetic Architecture of Human Gene Expression in EBV Transformed Cell Lines}},
  year = {2009},
  eprint = {0911.0215},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/0911.0215},
  howpublished = {arXiv},
}
Figure