High Throughput Screening in Primary Mammalian Cells

As mentioned above, many screens were previously done in immortalized cell lines for a variety of reasons, including cost, the ease of culture and expansion, as well as the relative ease of transfection, but these conveniences were achieved to a certain extent by using cells that are not normal. Compared to primary cells, cell lines often have major differences in ploidy, chromosome number, and even gene expression and receptor mutations, and sometimes these differences are seen even in the same culture (Mol Cancer Res: 14(1): 3–13 (2016); A. Goodspeed, et al.; “Tumor-Derived Cell Lines as Molecular Models of Cancer Pharmacogenomics”) making them a bad choice in the field of functional genomics. Even in the realm of cancer biology, there is concern as to whether cancer cell lines truly resemble tumors, especially in the tumor microenvironment, due in part to cell culture models that are sometimes not consistently reflecting clinical gene expression patterns. An additional factor would be that prolonged cell culture is likely to induce the occurrence of secondary genomic changes, such as copy number variations or transcriptomic drifts. (J Natl Cancer Inst. 105(7): 452–458. (2013), Jean-Pierre Gillet, et al.; “The Clinical Relevance of Cancer Cell Lines”). So although useful, there are potentially serious issues with the reliability of data generated with cancer lines (and actually, with all cell lines), and questions concerning the translatability of this data into the clinical or diagnostic arena, or into in vivo models. Within certain limits, cell lines can be a useful model system for cancer, but increasingly, primary cells are being looked at as a better option, especially when evaluating disease-related noncancerous cell behavior and responses.

The trend in screening is being driven towards using more relevant models that can more accurately depict the in vivo situation. Cell lines are increasingly being shown to be somewhat deficient in depicting the true in vivo situation, and alternatives are being developed. Whether using special incubators with custom settings for oxygen, humidity, and pressure, or through elaborate co-culturing of cells, or the use of 3D cell culture modeling to mimic the in vivo environment, once it has been demonstrated that cells respond differently to stimuli under these different and more relevant scenarios, it raises the unpleasant question as to whether these differences would have an impact in the analysis of candidate drugs and their in vivo efficacy, or otherwise impact the translation of in vitro results to in vivo applications (J. Lim and F. Aswad; https://www.ddw-online.com/therapeutics/p315005-ex-vivo-vs-in-vivo-the-challenge-of-generating-meaningful-results-with-traditional-cell-culture.html; “Ex Vivo vs In Vivo - The challenge of generating meaningful results with traditional cell culture”).

This brings us to the growing use of primary cells in high throughput screens for functional genomics. Primary cells have the advantage of being normal cells, expressing all of the receptors and signal transduction pathways, . The hypothesis is that these primary cells will respond in a more realistic manner to stimuli, and that the results are more likely to translate accurately into an in vivo setting. There are a few disadvantages of the use of primary cells in high-throughput library screening, some of which might be 1) the limited number of population doublings available for primary cells, possibly limiting the cell numbers available for large screens, 2) the difficulties in transfecting or transducing primary cells effectively, and 3) the relatively higher costs of using primary cells. These issues will be discussed in greater detail below.

Terminally-differentiated cells are obtained in limited numbers from donors. These primary cells also have a limited number of population doublings available to them before they begin to change and adapt to in vitro cultivation, and possibly lose some of their natural phenotype. In order to take advantage of the desirable qualities of primary cells, they must be used within this defined window of opportunity. This necessarily means that there may not be adequate time to expand the cells sufficiently from a single donor to the numbers needed for a large scale CRISPR library screen, and that researchers will either be forced to do smaller, more targeted screens with cells from a single donor, or will combine the cells from multiple donors and do pooled cell screens. These results from pooled screens should later be confirmed on more targeted screens with cells from single donors.

Another possible alternative to large scale pooled screens of primary cells would be to use differentiated iPSC cells to conduct large scale screens. In general, iPSC lines are primary pluripotent stem cells derived and reprogrammed from somatic cells, which are then characterized with respect to karyotype, and the ability to differentiate into the three standard germ layers, but which also demonstrates that iPSCs are essentially primary cells. An iPSC line can be expanded in an essentially unlimited manner, and researchers are getting increasingly sophisticated in differentiating these cells to different types of neurons, immune cells, cardiomyocytes, and the like, which would then be capable of being fed into large scale screens.

Although there are potential issues with using iPSCs in this fashion, such as the telomeric age of the cells or of epigenetic differences between primary and differentiated iPSC lines, they remain a viable screening option, particularly depending on the choice of starting material used to generate the iPSC line. One could use cells from a patient with a particular genetic disease, with the hope that the resulting iPSCs, when differentiated to the appropriate cell type, would tend to mimic, model, or recapitulate the disease, allowing detailed analysis. Alternatively, one could start with normal cells to generate the iPSC line. This line could then be altered with CRISPR in a targeted fashion, to create two isogenic iPSC lines, identical except for the specific CRISPR modification, which could then be expanded, differentiated into the appropriate cell type, screened, and the results from two iPSC lines then compared. This is an intriguing way to perhaps tease apart and determine the impact of certain biochemical or signal transduction pathways on a particular disease or phenotype.

Costs associated with primary cell culture might seem high at a first glance, the value of the conclusions that can be drawn from a more biologically relevant model may finally compensate for this, as it could help to select good drug targets quicker and also reduce animal studies.

Primary cells can be more difficult to transfect or transduce than cell lines, which is somewhat of a deterrent to the use of primary cells in CRISPR library screenings. Lipids can work well with cell lines, and are easy to use in a screen with their limitations in reliable and reproducible transfection efficiencies. Furthermore, lipids often do not work efficiently on all types of primary cells. Standard electroporation also does not often work well on many types of primary cells; in addition, there are also not many options for higher throughput 96- or 384-well standard electroporation, which limits its utility for siRNA or CRISPR library screening applications. This leaves both viruses, such as lenti- or retroviruses, and improved electroporation technologies like the Nucleofector™ Technology comprising multi-well platforms as the feasible options. Viruses can work very efficiently when matched well with a particular cell type, but may not work as well when paired with an unsuitable cell type (vector tropism). Basically, lenti- or retroviruses often need a co-receptor on the cell surface in order to bind and efficiently transduce a cell, such as with CCR5 and HIV (Receptors Channels 8(1):19-31 (2002); Blanpain C , et al.; “CCR5 and HIV infection”). Because of this, a lentiviral library targeted towards one cell type may not work as effectively with another cell type, limiting its versatility. Also, the costs of construction of a high-titer lentiviral library are not trivial, whether made in-house or purchased commercially, which must also be factored into any screening strategy, not to mention the need to maintain and monitor a BSL-2 (or greater) laboratory, if live viruses are being used.

Another alternative to the screening strategy discussion is the Nucleofector™ Technology. The Nucleofector™ Technology is an advanced form of electroporation which has proven to achieve high transfection efficiencies in primary cells, and the platform has been expanded to include robust and versatile high throughput transfection with the 96-well Shuttle™Device and the 384-well Nucleofector™ System, whose capabilities and throughput are not matched by any other form of electroporation. Both these devices have the ability to deliver a different pulse to each well, if desired, maximizing the flexibility of the platform and allowing for easy determination of optimal transfection conditions. In addition, the Nucleofector™ Technology is largely agnostic with respect to the substrates that can be transfected, and can easily transfect plasmid DNA, mRNA, oligonucleotides, siRNAs, miRNAs, peptides, proteins, ribonucleoproteins (RNPs), morpholinos, Q-Dots, small molecules into relevant cell types, or different combinations of these substrates.

Nucleofection™ technology is also excellent at the transfection of the Cas9 RNP, along with the capability for high NHEJ editing efficiencies in relevant cell types (up to 85-95% NHEJ editing efficiencies in primary human T cells and CD34+ hematopoietic progenitor cells, and up to 75% in NK cells), which greatly facilitates the use of CRISPR library screens in primary cells in high- and/or medium-throughput screen format.

The mechanism of Nucleofection™ Technology is based on diffusion of the substrate into the cell along the lines of a concentration gradient generated by the concentration of the substrate in the transfection reaction and the electrical fiel, and into the cells through pores generated in the cell and nuclear membranes by the electrical Nucleofector™ Device pulses. This enables one to tweak reaction conditions to vary and improve the results of the transfection by varying the amount of substrate used in the transfection reactions. This concept has been put to practical use by researchers transfecting the Cas9 RNP, and in part is responsible for the high editing efficiencies that they have achieved in primary human CD34+ and both mouse and human T cells (Nature. 17;539(7629):384-389 Nov 2016, DP. Dever and MH Porteus, “CRISPR/Cas9 β-globin gene targeting in human haematopoietic stem cells’; Nature 559: 405-409 (2018), Theodore L. Roth, … Alexander Marson, “Reprogramming human T cell function and specificity with non-viral genome targeting”; J. Exp. Med. 2018 (https://doi.org/10.1084/jem.20171626), A Seki and S Rutz, “Optimized RNP transfection for highly efficient CRI SPR/Cas9-mediated gene knockout in primary T cells”).

Not only is the use of the Nucleofection™ System relatively unrestricted with respect to substrate delivery, but it is also very versatile in throughput and volume due to the different platforms available. The concept of the transferability of transfection conditions across these platforms is what makes the technology so powerful in a screening situation. When using the 20 µl transfection volume in the low-throughput benchtop 4D Nucleofector™ System, perhaps for basic research or assay optimization, those same conditions (transfection solution, program, cell number, and performance) will translate into the 20 µl reactions of both the 96-well Shuttle™ Device and the 384-well Nucleofector™ System precisely. This allows for assays revolving around transfection of specific substrates to be perfected at a low throughput level, and then fed into high throughput screens without having to alter or adapt the assay in any way.

This can be quite convenient for the researcher, as the assay conditions can be worked out in detail and more cost-effectively at small scale, and then fed directly into the higher throughput platforms for siRNA or CRISPR library screenings. This is especially relevant, for example, for central core screening facilities in larger academic, biotech, or Pharma settings, where the core facility receives screening requests from their internal customer labs, possibly even globally. The individual assays can be crafted and tested in the respective laboratories at a smaller scale using the 4DNucleofector™ X Unit or 96-well Shuttle™ Device add-On, and when ready, these assays can then be used for CRISPR library screenings in functional genomics fed into the core facility, located at that same site or elsewhere. Because the transfection conditions will transfer exactly, the core facility will not have to do extensive optimizations to incorporate the assays into their higher throughput 96-well Shuttle™ Device or 384-well Nucleofector™ System, but can instead feed these assays directly into their workflow with minimal additional preliminary confirmation tests, potentially saving a considerable amount of time and effort.

Furthermore, both the 96-well Shuttle™ Unit and the 384-well Nucleofector™ System are capable of being partially or fully integrated into robotic liquid handling systems (e.g. from Tecan, Beckman or Hamilton), which can be more important when it comes to managing large scale CRISPR library screens, reducing hands-on time.

So what is different about this latest trend in high-throughput screening, the use of CRISPR-generated knockout screening in primary cells? First, as described above, there is a trend in using primary cells in screens, based on the idea that primary cells will be more effective predictors of the in vivo efficacy of a drug or drug candidates. Secondly, the use of CRISPR technology to specifically target and knock out genes or target different parts of a particular pathway gives the researcher an opportunity to tease apart pathways and responses in a very clean system an approach used for functional genomics applications.

Using CRISPR technology knockouts can be generated by introducing a double stranded DNA break at a particular target site in the genome. Normally, these ends are perfectly rejoined by Non-Homologous End Joining (NHEJ). However, this repair process is known to be error-prone, and can sometimes result in small nucleotide insertions or deletions (indels). When this occurs in the coding region of a gene, this usually results in a frame-shift mutation, and once the reading frame is thrown off, a random stop codon is then encountered fairly frequently, resulting in a premature termination event, and a knock out of the expressed gene function.

Through tweaking the transfection conditions, the editing efficiency of introducing indels, and consequently knockouts, can be increased dramatically. Especially with the high editing efficiency achievable in many relevant cell types using CRISPR (ranging from 85-95% indel formation), the background noise from unedited cells can be moved largely into the background level, minimizing the need for any additional steps or analyses.

By knocking out a gene target with CRISPR as opposed to merely reducing the expression of the target gene with siRNAs, any residual activity of expressed protein can be reduced or eliminated by knocking out the target gene, resulting in a cleaner, less complicated analysis. This is especially important in cases where the phenotype of the modified cell is more subtle and harder to analyze; in cases like these, the phenotype might well be masked by presence of residual protein activity if the mRNA expression was merely reduced instead of eliminated.

The CRISPR knockout strategy could conceivably be of interest in evaluating aspects of drug discovery associated with pathway analysis in functional genomics. If a specific biochemical pathway is implicated with a phenotype that results in a particular disease or response, then knocking out different parts of that pathway might not only confirm the importance of that pathway in the disease or response, but might also provide alternative druggable targets for therapeutic intervention.

There are a number of assay options for these kinds of CRISPR screens. The simplest sort of assay might be viability screens, where the targeted knockout induces lethality. These can be quite useful, but one potential problem is that many genes are not lethal when knocked out. This limitation can impact the utility of these sorts of screens when examining other, more subtle forms of phenotypes. Although in many ways these assays can be difficult to design, phenotypic screens can be revealing with respect to cellular functions and the various factors involved in regulating these functions both positively and negatively. Phenotypic screens done in primary cells can yield results that may well be different from comparable screens done with cell lines.

Previous experiments of phenotypic screens were done, for example, using a cDNA library and high content analysis to examine the roles of kinases, esterases, and phosphatases in neurite extension in primary rat neurons (Molecular Systems Biology 6:391, Buchsner WJ et al. (2010), “Kinase/phosphatase overexpression reveals pathways regulating hippocampal neuron morphology”), or with siRNA libraries that tested the role of a number of genes identified in the C. elegans screens on the mouse macrophage immune response (Alper S et al. (2008) PNAS 105(19):7016–7021, “Identification of innate immunity genes and pathways using a comparative genomics approach”). Using CRISPR library screens, there is also now the opportunity to evaluate cells and cellular responses in more detailed ways. In the relatively brief time that CRISPR has been around, there are already several relevant examples of CRISPR phenotypic screens being employed. DNA repair pathways have been evaluated to determine what other cellular pathways may also be implicated in the CRISPR repair processes (CD Richardson, ….. and JE Corn, Nature Genetics 50:1132–1139 (2018), “CRISPR–Cas9 genome editing in human cells occurs via the Fanconi anemia pathway”) where it was shown that the Fanconi Anemia DNA repair pathway may tend to drive the resolution of the CRISPR-derived double-stranded DNA cleavage away from non-homologous end joining (NHEJ) towards homology-directed repair (HDR). This could be of particular interest in the gene therapy field, if the Fanconi Anemia pathway could be harnessed and utilized to preferentially enrich the number of desirable HDR outcomes (i.e. efficient insertion of the DNA repair/replacement template) and reduce the number of unwanteed NHEJ outcomes. In addition, other potential CRISPR screens offer possibilities, for example, to evaluate transmissible-disease pathogens such as Plasmodium falciparum (malaria), for vulnerabilities that might be exploited therapeutically (PLoS One. 2017 May 22;12(5):e0178163. doi: 10.1371/journal.pone.0178163. eCollection 2017, ED Crawford,… JL DeRisi, “Plasmid-free CRISPR/Cas9 genome editing in Plasmodium falciparum confirms mutations conferring resistance to the dihydroisoquinolone clinical candidate SJ733”).