Identifying Extrinsic versus Intrinsic Drivers of Variation in Cell Behavior in Human iPSC Lines from Healthy Donors

摘要

class="head no_bottom_margin" id="sec1title">IntroductionNow that the applications of human induced pluripotent stem cells (hiPSCs) for disease modeling and drug discovery are well established, attention is turning to the creation of large cohorts of hiPSCs from healthy donors. These offer a unique opportunity to examine common genetic variants and their effects on gene expression and cellular phenotypes (, , , , ). Genome-wide association studies (GWASs) and quantitative trait locus (QTL) studies can be used to correlate SNPs and other genetic variants with quantitative phenotypes (). As a contribution to this effort, we recently described the generation and characterization of over 700 open access hiPSC lines derived from 301 healthy donors through the Human Induced Pluripotent Stem Cell Initiative (HipSci) (; ). In addition to creating a comprehensive reference map of common regulatory variants affecting the transcriptome of hiPSCs, we performed quantitative assays of cell morphology and demonstrated a donor contribution in the range of 8%–23% to the observed variation (). In the present study, we set out to identify genetic drivers of cell behavior.Previous attempts using lymphoblastoid cell lines to link genetics to in vitro phenotypes have had limited success (, ). In that context, confounding effects included Epstein Barr virus (EBV) viral transformation, the small number of lines analyzed, variable cell culture conditions, and line-to-line variation in proliferation rate. These factors decrease the power to detect true relationships between DNA variation and cellular traits (). In contrast, we have access to a large number of hiPSC lines derived using standard protocols from healthy volunteers, including multiple lines from the same donor. In addition, HipSci lines present a substantially lower number of genetic aberrations than reported for previous collections (, ). Cells are examined over a limited number of passages, and cell properties are evaluated at single-cell resolution during a short time frame, using high-throughput quantitative readouts of cell behavior.Stem cell behavior reflects both the intrinsic state of the cell (, href="#bib22" rid="bib22" class=" bibr popnode">Kyttälä et al., 2016) and the extrinsic signals it receives from its local microenvironment, or niche (href="#bib23" rid="bib23" class=" bibr popnode">Lane et al., 2014, href="#bib36" rid="bib36" class=" bibr popnode">Reimer et al., 2016). We hypothesized that subjecting cells to different environmental stimuli increases the likelihood of uncovering links between genotype and cell behavior. For that reason, we seeded cells on different concentrations of the extracellular matrix (ECM) protein fibronectin that support cell spreading to differing extents and assayed the behavior of single cells and cells in contact with their neighbors. We took a “cell observatory” approach, using high-throughput, high-content imaging to gather data from millions of cells 24 h after seeding. We then applied a multidimensional reduction method, Probabilistic Estimation of Expression Residuals (PEER) (href="#bib42" rid="bib42" class=" bibr popnode">Stegle et al., 2012), to reveal the underlying structure in the dataset and correlated cell behavior with the expression of a subset of genes and the presence of rare deleterious non-synonymous single nucleotide variants (nsSNVs). The strategy we have developed bridges the gap between genetic and transcript variation on the one hand and cell phenotype on the other, and should be of widespread utility in exploring the genetic basis of inter-individual variability in cell behavior.