Addressing the Complexity of Engineered Cell Lines

By: Gary L. Davis, Ph.D., Manager of Cell Line Engineering, Biotechnology R&D, Biowire Spring 2012, 4–5

Biowire Spring 2012 — Live Cell Imaging of Signaling Pathways

Once Upon a Time...

More than a few years ago, I recruited a post-doctoral student from the prestigious lab of Dr. Pierre Chambon to join my research team in pharmaceutical discovery. To understand tissue development, this young scientist had been disrupting human somatic cell genes in culture using a technique called homologous recombination (HR)1. Using HR, he was able to completely ablate the production of his particular proteins and then study the effect of their absence. Replacing or removing genes at will in cultured cells was at the time revolutionary and held the promise of rapidly propelling our understanding of what certain cellular proteins did and, for us, which were candidate drug targets. Well, upon arrival and hearing the actual project goals, he enlightened us on the challenge — the near impossibility, as it turns out — of doing this at the scale and within the timelines that discovery-for-profit demanded. He went on to an impressive career in cancer research while the field of eukaryotic gene manipulation plodded along. Fortunately, since then, new technologies have surfaced that make it practical to expeditiously “engineer” many types of user-defined changes into living cells2.

One such technological advent is the zinc finger nuclease (ZFN) system3. Like the old methods, ZFN technology also relies on the HR biology, as well as another cellular DNA repair mechanism called nonhomologous end joining (NHEJ)4; together or separately, these natural cellular activities can be used to deliberately sculpt the genetics of the target genome and make designer cell lines.

The ZFN system is commercially distributed by Sigma-Aldrich® as CompoZr®, and is now being used by my current team of cell biologists to create genetically engineered cell line products. These ZFN-altered model cell lines are available as both Sigma-Aldrich catalog items and as customer-defined custom products (Cell Design Studio™). In this issue’s companion article, Wemhoff, Holroyd, and Keller give an example of the custom cell line engineering process and how this can be effectively applied to a once perplexing problem in cancer biology.

Complexity Redux

In my earlier Biowire review on cancer2, I discussed the application of ZFN-derived cell-based models to study complex diseases. Complexity enters the cell scientist’s world in two places: once at the level of the biological problem being investigated (bio-complexity), and again in the biology of the cell-based system being used as a tool to study the problem (bio-technical complexity). Both are not trivial, both are important, and both are obviously interrelated.


Most complex biological phenomena arise from large arrays of multi-molecular interactions. This complexity is compounded by the natural variation present in the “activities” (kinetics) of a population’s interactants. And it is this variation that is due to the unique genetic constellation present in the individual — whether that is an individual person or an individual cell. Often, the complexity becomes so complicated, that cause-and-effect generalizations about a complex phenotype are woefully insufficient, if not misleading (cf.2,5). The challenge of “personalized medicine”6, as discussed here in the context of cell line engineering by Wemhoff, Holroyd, and Keller, is a direct result of the necessity to examine the unique genetic complement of an individual’s “interactome”7.

The solution to the Bio-Complexity problem, pertinent to the most important biomedical questions, can be addressed through the deliberate fusion of: 1) broad, system-wide analyses (systems biology) and 2) hypothesis-driven investigations of critical mechanisms2,8. Philosophically speaking, it is the co-application of the Baconian datadriven (“omics”) approaches with the hypothesis-driven methods of Galileo9 to the problem. Because engineered cell lines permit the systematic investigation of many independent variables while keeping others controlled, they are superior tools for Galilean-type experiments. This is partly true because ZFN-altered cell lines can be tested in parallel against both other “isogenic” engineered lines and the unmodified “isogenic” parental control cells. Using genetically matched sets of cells is now routine10,11.

Bio-Technical Complexity
Complexity also enters the experimental cell biologist’s realm when they begin to consider the experimental system itself. When using ZFNs to make genetic changes in a cell line, a host of biological situations have an impact on the technology’s facility.

Chromosomal Clutter: Ploidy
As Wemhoff, Holroyd, and Keller point out, increases in the number of chromosomal targets per cell (ploidy) will reduce the frequency of engineered mutations detected. In many transformed cell lines, the genetic instability and resultant aneuploidy, which permits the cell’s limitless proliferation, can greatly increase the numbers of individual chromosomes. Theoretically, this increases the number of cells to screen in an engineering experiment; practically, however, it has never prevented us from eventually isolating a full-knockout clonal cell line.

Problematic Prolificacy: Gene Amplification
What could prevent success, is if all alleles in a cell needed to be modified and the target sequences had been extensively amplified. For example, in the breast cancer cell line MCF-7, karyotype analysis would suggest that the “q” arm of chromosome 8, at base 116M, is tetraploid; however, a fine copy number variation (CNV) analysis shows that TRPS1, which is located at base 116M, is actually at 10 copies per MCF-7 cell12.

Marred Mending: Homologous Recombination Ability
In the case of knockin cell lines, where HR is required to insert a donor DNA fragment at the ZFN cut site, it is critical that the cell line be at least moderately proficient in this activity. The human cell line HT29 is an example of a cell without one repair activity; in our studies, while having NHEJ activity equivalent to other lines, it has virtually undetectable HR activity (unpublished).

Genetic Reshuffling: Mitotic Recombination and LOH
The goal of most cell line engineering projects is to deliver complete changes in all alleles present (homozygous). In some situations, however, it is not possible to modify all alleles (because complete loss — or gain — affects proliferation or viability); and, in others, the heterozygous state is desired because it is the most biologically relevant (cf.2). Regardless, if within a cell there exists both modified and unmodified alleles, the potential for inter-chromosomal recombination is present and can eventually result in both wildtype and homozygous-mutant cell lineages being present in the cell population. Termed loss of heterozygosity (LOH )13, if a growth advantage results from these genetic rearrangements, then that sub-population will begin to predominate in the culture.

Expression Paradoxes: Gene Dosage and Compensation
It is logical to expect, when multiple alleles are present, that inactivating alterations in increasingly larger numbers of alleles would result in a correspondingly decreased expression profile. Often, however, this is not the case. In gene knockout situations where multiple alleles are targeted, knocking out any number of alleles only slightly reduces expression levels; it is only when all alleles are knocked out that expression disappears (cf. BAX/BAK knockouts)2. It seems, therefore, that for many genes there are compensatory mechanisms that maintain a minimal level of expression.

This is even true when a subset of genes in a functionally related gene family is targeted. In the Caco2 cell line, we inactivated all four copies of the MDR1 gene only to see a significant increase in the BCRP transporter expression; it would seem that the increase in BCRP is to compensate for the loss of function of MDR1(unpublished).

Rogue Biology: Synonymous Mutation Suppression and Stop Codon Read-Through
Other less well-understood phenomena may also confound attempts at gene modification. For instance, sometimes during the engineering of a particular gene, synonymous (translationally silent) mutations are created. While it is usually presumed these have no affect on expression, in fact, sometimes silent mutations drastically reduce the levels of gene products observed14,15. This can be true even when the exogenously added synthetic donor DNA sequence is optimized using human preferred codons. As an example, when we used a human codon-optimized green fluorescent protein (GFP) donor DNA to tag an endogenous gene, this completely inhibited expression of the resultant fusion protein; when the optimized codons were replaced with the naturally occurring GFP codons, however, this restored recombinant protein production and fluorescence (unpublished).

Similarly perplexing is a phenomenon sometimes seen when gene knockouts are made using ZFNs to either directly introduce stop codons or create deletions that result in a translational frameshift. Usually, it is presumed that stop codons in the new reading frame will terminate translation; however, once again, it is not always that simple. Sometimes, it is possible that the new theoretical stops are ignored when the DNA-sequence context of those stop codons is suboptimal; this phenomenon, known to be exploited by animal viruses, can also occur through the action of natural suppressor tRNAs16.


If we are indeed only limited by our imaginations, then it is necessary that we appreciate the complexities facing us. For research scientists, who are encouraged to use Occam’s Razor to simplify their conclusions, what they must not do in their experimental design is trivialize the intricacies of the biology or the technology used to understand it. The sophisticated cell line engineer avoids both.



  1. Boylan JF, Lufkin T, Achkar CC, et al. Targeted disruption of retinoic acid receptor alpha (RAR alpha) and RAR gamma results in receptor-specific alterations in retinoic acid-mediated differentiation and retinoic acid metabolism. Mol Cell Biol. 1995 Feb;15(2):843–51.
  2. Davis GL. Cracking the Cancer Code: Decrypting Its Genesis and Proliferation with Cell-Line Models. Biowire. 2011 April, 25, 2011;6–10, 39, 40.
  3. Urnov FD, Rebar EJ, Holmes MC, et al. Genome editing with engineered zinc finger nucleases. Nat Rev Genet. 2010 Sep;11(9):636–46.
  4. Hartlerode AJ, Scully R. Mechanisms of doublestrand break repair in somatic mammalian cells. Biochem J. 2009;423(2):157–68.
  5. Lander AD. The edges of understanding. BMC Biol. 2010;8:40.
  6. Hoehe MR, Timmermann B, Lehrach H. Human interindividual DNA sequence variation in candidate genes, drug targets, the importance of haplotypes and pharmacogenomics. Curr Pharm Biotechnol. 2003 Dec;4(6):351–78.
  7. Lage K, Karlberg EO, Storling ZM, et al. A human phenome-interactome network of protein complexes implicated in genetic disorders. Nat Biotechnol. 2007 Mar;25(3):309–16.
  8. Kitano H. Looking beyond the details: a rise in system-oriented approaches in genetics and molecular biology. Curr Genet. 2002 Apr;41(1):1–10.
  9. Medawar PB. Advice to a Young Scientist. NewYork: Harper & Row; 1979.
  10. Dreyer AK, Cathomen T. Zinc-finger nucleases-based genome engineering to generate isogenic human cell lines. Methods Mol Biol. 2012;813:145–56.
  11. Sur S, Pagliarini R, Bunz F, et al. A panel of isogenic human cancer cells suggests a therapeutic approach for cancers with inactivated p53. Proc Natl Acad Sci U S A. 2009 Mar 10;106(10):3964–69.
  12. Savinainen KJ, Linja MJ, Saramaki OR, et al. Expression and copy number analysis of TRPS1, EIF3S3 and MYC genes in breast and prostate cancer. Br J Cancer. 2004 Mar 8;90(5):1041–46.
  13. Moynahan ME, Jasin M. Mitotic homologous recombination maintains genomic stability and suppresses tumorigenesis. Nat Rev Mol Cell Biol. 2010;11(3):196–207.
  14. Sauna ZE, Kimchi-Sarfaty C, Ambudkar SV, et al. The sounds of silence: synonymous mutations affect function. Pharmacogenomics. 2007 Jun;8(6):527–32.
  15. Sauna ZE, Kimchi-Sarfaty C, Ambudkar SV, et al. Silent polymorphisms speak: how they affect pharmacogenomics and the treatment of cancer. Cancer Res. 2007 Oct 15;67(20):9609–12.
  16. Beier H, Grimm M. Misreading of termination codons in eukaryotes by natural nonsense suppressor tRNAs. Nucleic Acids Res. 2001 Dec 1; 29(23):4767–82.


Related Links