Evolution from adherent to suspension: systems biology of HEK293 cell line development

Abstract

The need for new safe and efficacious therapies has led to an increased focus on biologics produced in mammalian cells. The human cell line HEK293 has bio-synthetic potential for human-like production attributes and is currently used for manufacturing of several therapeutic proteins and viral vectors. Despite the increased popularity of this strain we still have limited knowledge on the genetic composition of its derivatives. Here we present a genomic, transcriptomic and metabolic gene analysis of six of the most widely used HEK293 cell lines. Changes in gene copy and expression between industrial progeny cell lines and the original HEK293 were associated with cellular component organization, cell motility and cell adhesion. Changes in gene expression between adherent and suspension derivatives highlighted switching in cholesterol biosynthesis and expression of five key genes (RARG, ID1, ZIC1, LOX and DHRS3), a pattern validated in 63 human adherent or suspension cell lines of other origin.

Figure 1

Comparisons of genomic and transcriptomic profiles of HEK293 cells showed taxonomic divergence between parental HEK293 and progeny cell lines. (a) A schematic overview of the lineage relationship of the six HEK293 cell lines used in this study. Blue dots represent adherent cells whereas grey dots represent suspension cell lines. (b) Genomic comparison between HEK293 cell lines based on Spearman correlation coefficients of read counts. Darker blue color indicates higher correlation. (c) Sample-to-sample comparison between transcriptomes illustrated by a heatmap and hierarchical clustering of taxonomical divergence between samples. Darker blue color indicates shorter Euclidean distance between samples and more similarity. (d) PCA plot showing the separation in expression pattern between samples. (e) RNA expression levels (in DESeq2 median of ratios) and standard deviations (n = 3) of stably integrated viral genes (EBNA-1, Large T, E1A and E1B) in HEK293 cell lineages determined by RNA sequencing.

Figure 2

Copy number variation analysis of HEK293 progeny cells compared to the parental HEK293 revealed conserved patterns of copy number gain and loss. (a) Genomic copy number gain (red) or loss (blue) of chromosomes 1, 8, 9, 13 and 18 of progeny HEK293 cell lines compared to parental HEK293 cells. The black line indicates the centromere position of each chromosome. (b) Genomic copy number gain or loss (log2 fold change) compared to HEK293 for each cell line of the FH, KMO, TLE4 and ADAM3A genes.

Figure 3

Differential expression analysis emphasized processes and genes with common changes in all progeny cell lines compared to the parental HEK293. (a) Consensus heatmap of GO biological processes with a different expression pattern between progeny cell lines compared to the parental HEK293. Low consensus scores, represented by a dark blue color, indicate more significant differences. (b) Common differentially expressed (DE) genes in pairwise comparisons of all HEK293 cells. Blue bars show number of DE genes in each pairwise comparison. Green bar shows 329 common DE genes in pairwise comparisons of progeny with HEK293 parental cells. Red bar shows common 38 DE genes in the comparison of suspension cells against adherent cells. (c) Top ten significant GO cellular components of the 329 common DE genes in pairwise comparisons between progeny cells and HEK293.

Figure 4

Gene set analysis identified biological process and metabolic changes between suspension and adherent HEK293 progeny cell lineages. (a) Heatmap of gene set analysis using GO biological processes and Piano consensus scores showed a different expression pattern between suspension cells (293-H and 293-F) and adherent cells (293E and 293T). Low consensus scores, represented by a dark blue color, indicate more significant differences. (b) Metabolic genes set analysis for comparing metabolic differences between suspension and adherent HEK293 progeny cell lines. The size of each node corresponds to the number of genes in each of these pathways, thickness of connections between nodes corresponds to the number shared genes between pathways and the colors of the nodes shows the p-value for the given metabolic process.

Figure 5

Evaluation of common DE genes between adherent and suspension HEK293 cells identified cholesterol biosynthesis as the main enriched pathway. (a) Heat map with TPM values for each DE gene common to all adherent to suspension comparisons. (b) The top ten most enriched biological GO terms of the 38 common DE genes between adherent and suspension cells based on gene enrichment analysis. Length and color of bars both show significance of adjusted p-value for the hypergeometric test. Also, genes mentioned in each bar are the genes that belong to enriched GO term and present in the list of 38 DE genes.

Figure 6

Gene expression validation of the 38 previously identified differentially expressed genes in 63 human cell lines, identified nine significantly differentially expressed genes between suspension and adherent cell lines. (a) PCA transcriptomic data of 63 human cell-lines from the Human Protein Atlas shows a clear separation of suspension and adherent cell lines from different tissues. (b) Range of normalized counts in HPA cell lines for each of the previously identified 38 genes, differentially expressed between all adherent and suspension HEK293 cell lines. The black line in each box shows median of normalized counts for the gene. (c) Genes that are differentially expressed between adherent and suspension cells using a Mann–Whitney U-test, with p-values < 0.01, are highlighted in purple, where length of bars shows logarithmic fold change of median between two groups and the color of bars denotes degree of significance of p-value. Non-significant genes have gray bars.