Data Robustness and Reproducibility in Gene Editing Applications: Today’s Limits and Tomorrow’s Potential

Sponsored content brought to you by

CRISPR’s use has vast potential in basic research to add to the understanding of cellular activity and in healthcare for precision therapies for today’s untreatable diseases. To date, some 30 clinical trials using CRISPR genome editing technology are registered at ClinicalTrials.gov.

New versions of Cas9 are being continually discovered and used. As an example, the generation of nuclease deficient Cas9 versions enabled the generation of CRISPR interference (CRISPRi) and CRISPR activation (CRISPRa). CRISPRi and CRISPRa allow the repression or overexpression, respectively, of target genes rather than complete elimination to address critical gaps in target ID and validation during development work.

When used together, CRISPRi and CRISPRa can reveal genes that are critical to both inhibition and activation of a pathway of interest and also ‘switch’-like genes that display opposing effects. The approach offers researchers a novel, powerful, and systematic way to gain valuable genetic insights into cellular physiology, explore drug mechanism of action, identify novel biomarkers, and provide compelling targets for the development of novel combination therapies.

Challenges remain in gene editing. Frequently, when beginning to investigate novel gene function and their role in disease, researchers edit readily available, and relatively simple to manipulate immortalized cell lines. However, as DNA repair pathways can be cell type-dependent, reproducing a phenotype when moving to more biologically-relevant primary cells may prove difficult.

Off-target effects also must be taken into consideration. Genome editing commonly depends on a double-strand DNA break made at the intended site by a programmable nuclease. But if the nuclease and guide RNA complex lacks perfect specificity, secondary off-target sites may also be cleaved, leading to unplanned modifications. Off-target edits are especially important in human and other mammalian applications, especially if a deleterious effect results; analyzing for and alleviating these off-target effects are crucial.

GEN asked six outstanding CRISPR researchers about the use of CRISPR for knockout (CRISPR KO), interference, and activation, ramping up to editing primary cells from immortalized cell lines, and how they approach off-target effects. And we asked them to look ahead and share what they think the future holds.

GEN: Where does CRISPRa/CRISPRi fit within your overall CRISPR toolbox? Would using CRISPR knockout in addition to CRISPRi as an orthogonal validation method increase confidence in your hits?

Beeke Wienert: Absolutely! I think showing your knockout results in an orthogonal way is a great way to confirm your results. Recently it has been shown that introducing indels to knockout (KO) a gene can result in other aberrant proteins and partial preservation of protein function. Thus, additional verification using CRISPRi/a can confirm your results. It might also give you an idea how the “dosage” of your gene correlates with phenotype changes, as CRISPRi/a lets you “fine-tune” expression. (Two references are in Nature Methods, www.nature.com/articles/s41592-019-0614-5 and Nature Communications, www.nature.com/articles/s41467-019-12028-5.)

I also think that CRISPRi is an invaluable tool to study essential or haplo-insufficient genes that result in lethality of cells or animals when knocked out. When performing a CRISPR screen, these genes would drop out when using CRISPR KO, while CRISPRi could result in insights of their gene function.

Stanley Qi: We use a lot of CRISPRa/i for our research. CRISPRa allows enhanced gene expression, which is a missing feature when using CRISPR knockouts. CRISPRi allows partial repression, which is useful for studying essential genes. Furthermore, we found CRISPRa/i allow better multi-gene regulation to study genetic interactions. Using CRISPRi as an orthogonal validation of CRISPR knockout is a good idea, as this may avoid off-target effects from the particular guide used for the knockout.

Sanne Klompe: The CRISPR toolbox is as versatile as it is because clever engineering yields tools like CRISPRa and CRISPRi. Being able to up- and down-regulate genes transiently or in a multiplexed fashion is an invaluable asset for studying gene function and pathway mapping. Similarly, CRISPRa and CRISPRi could be used to better mimic natural changes or different cell-states than complete knockouts might. Combining CRISPR knockout and CRISPRi could yield additional information on the temporal aspects of gene essentiality and function.

Jenny Hamilton: CRISPRi is especially powerful for knocking down expression of essential genes, where CRISPR knockout would result in cellular lethality. For non-essential genes, CRISPR knockout is an appropriate alternative method for validating genes identified in a CRISPRi screen.

Mitch O’Connell: We have yet to use these in the lab. Having orthogonal methods to modulate gene expression and validate our CRISPR edits would be useful if we end up needing to do CRISPR screens, where validation becomes a little more difficult/time consuming. The same goes for the use of other CRISPR-Cas tools like Cas13, which we’ve had some success setting up for gene knockdown (KD) experiments.

GEN: Are there reproducibility concerns when moving from immortalized cell lines into primary cells? If so, how are these risks addressed?

Jenny Hamilton: It’s sometimes challenging to move from immortalized cell lines to primary cells because the pathways favored to repair DNA damage can vary by cell type. As gene editing outcomes are dictated by DNA repair pathways, immortalized cell lines may not always accurately predict the outcome of gene editing in a primary cell. This makes it critically important to move gene editing experiments from immortalized cell lines into primary cells as quickly as is feasible, as cell lines may not be appropriate surrogates for studying the on/off-target specificity and efficiency of gene editing.

Researchers shouldn’t see this as an intimidating jump—while immortalized cell lines sometimes show increased editing efficiencies over primary cells, this is not always the case! In our hands, we’ve observed more favorable editing efficiencies in primary human T cells compared to Jurkats, an immortalized T cell line.

Ben Kleinstiver: In our experience, the performance of most CRISPR-Cas enzymes translate generally well between different cell types when comparable effective doses of the editor can be delivered. That said, the edit profiles of certain types of CRISPR technologies might change, since they are subject to the expression profiles of endogenous repair factors that vary in different cell types.

Stanley Qi: There are higher variations when working with primary cells. The source of primary cell is one major factor—cells from different people are different. Factors including gender, age, health and culture conditions are all confounding factors that impact the outcome when working with these cells. We have to admit that we know so little about primary cells. It is a false expectation to assume they can work similarly to immortalized cell lines. More efforts are needed to truly understand their biology and develop new protocols and standards to work with them in a more reproducible manner.

GEN: What methods do you use to analyze and limit off-target editing? What current limitations restrict the ability to analyze  off-target editing?

Ben Kleinstiver: The manner in which we consider off-targets depends on the type of experiment that we’re performing. Generally, we start with in silico analyses using software including Cas-OFFinder, which comprehensively annotates all closely matched putative off-targets, and CHOP-CHOP (or other similar software), which aggregates on- and off-target predictions.

To empirically determine the specificities of different CRISPR-Cas enzymes in human cells, we use an unbiased genome-wide assay called GUIDE-seq. We follow this with targeted sequencing of GUIDE-seq-identified off-target sites and also closely matched sites nominated by Cas-OFFinder. One potential limitation of existing off-target analytical methods is their relative inability to effectively identify long deletions or larger structural changes induced by Cas enzymes. New assays that remove these and other blind spots will be instructive for the field.

Mitch O’Connell: All of the genome editing we do in my biochemistry lab is to study gene function in cell culture (no therapeutic goals at this point), so the amount of concern for off-targeting editing depends on the application and the limitations also vary (it is much easier to sequence a cell line than a whole organism to get a sense of off-target events). We take steps to ensure we can minimize any potential off-targets and do appropriate controls to make sure any phenotype we go on to study is a result of the desired edit.

On the front end, we use several guide RNA design tools to judiciously select several gRNAs with the lowest off-target potential, with the aim of making several distinct clonal lines with ‘different’ edits in the same gene, to ensure any phenotype we see is recapitulated across different edits. We use RNP transfections at the lowest concentration possible to an effective editing rate to minimize CRISPR-Cas9 dose/exposure to the cell—the dose makes the poison! Once we’ve detected sufficient edits in our initial polyclonal population (using amplicon sequencing and ICE analysis), we generate monoclonal populations and repeat the amplicon sequencing by subcloning out the edited region into plasmids and sequencing 10-20 clones to make homogenous populations at the DNA level.

We then conduct KO analysis both at the RNA (qPCR/RNA-seq) and protein level (western blot/MS) to ensure we are seeing efficient nonsense-mediated decay (NMD) and loss of protein expression. And finally, as we have some molecular phenotypes we can measure upon KO, we transfect our cell lines with a wild-type copy of our gene and make sure we see rescue of this phenotype.

As for current limitations, we may miss rare alleles with our approach and I do think we should carry out NGS or even whole-genome sequencing (WGS) more routinely (although there are limitations with both). Once we start tackling genes where a molecular phenotype may be harder to identify initially, we will definitely need to be more careful in this respect. We also need to be careful about not detecting larger on-target deletions that may occur in our population, which could be achieved by longer-read sequencing, which is rapidly coming of age. Finally, as we work in transformed cell lines, aneuploidy/genome instability is also a perpetual concern to us (and how off-target genome editing can contribute to this). We attempt to minimize its contribution by having the appropriate controls in our experiments.

Sanne Klompe: Different types of editing will require different approaches to analyze off-target editing events. As the activity of CRISPR-transposons results in the integration of a specific donor rather than in small sequence changes, detecting off-target events is pretty straightforward. Instead of doing WGS, we can fragment the genomic DNA and use the donor sequence to enrich for integration events, whether through PCR amplification or sequence-specific pull-down methods.

All editing events by our CRISPR-transposons (on- or off-target) are then revealed by sequencing any donor-encoding genomic fragments. Analyses of off-target edits and ChIP-seq data elucidate more detailed spacer-protospacer base-pairing requirements and are used to inform gRNA design to prevent off-target events. Understanding how off-targeting is facilitated is crucial in limiting off-target events; structures of off-target-bound editors can explain the molecular basis of these events with increasing clarity.

Beeke Wienert: Avoiding off-targets usually starts with rational design of your editing experiment, which means picking gRNAs that have a low off-target profile by in silico prediction methods (such as CRISPOR). The level of depth you have to get into to chase off-targets varies depending on the application.

If you are looking at what the phenotype of a gene knockout is in your model cell line, the easiest might be to use multiple gRNAs to knock out your gene of interest and then compare phenotypes between those cell lines. This eliminates phenotypes observed through off-target effects. However, if you plan to use a gRNA in a clinical study, a much more detailed off-target profile is needed: the more orthogonal methods the better, which means that one should perform in vitro assays such as CIRCLE-Seq or SITE-Seq in addition to cell-based methods such as GUIDE-Seq or DISCOVER-Seq.

In general, one bottleneck is that we assume there is high sequence similarity between on-target and off-targets and that off-target methods use a “mismatch filter” to filter out false positives. However, DNA bulges, small deletions and genome-specific single nucleotide polymorphisms (SNPs) are much harder to predict, as they don’t fall under the classic “mismatch” category and thus could result in unexpected off-targets.

Stanley Qi: We are switching to WGS to fully evaluate the off-target effects on the genome DNA. However, it is costly for both experiments and computational analysis.

GEN: What will be the next big thing in CRISPR? Will it be the driving force for more therapeutic applications?

Mitch O’Connell: I think we will see increasingly emerging technologies like prime editing and base-editing possibly tackle more therapeutic targets in the clinic, particularly as we’re continuing to see improvements in their performance. I’m hoping improvements in the ability to modulate HDR/NHEJ ratios will help with tougher knock-in edits, although this is a tough problem.

There also seems to have been a boom in the number of start-up companies tackling the hard problem of delivery, particularly for therapeutic applications. We may also start to see the use of CRISPR-Cas-mediated transposases for applications in mammalians cells, which will open up new and exciting ways to manipulate the genome.

Personally, I’m excited to see an increasing use of Cas13 as another orthogonal tool for gene expression modulation. It may become easier to use in cases where CRISPR KD or CRISPRi fail, or simply as another way to validate the biology you’re observing. For this to happen, we need better guide-design tools, and I’m looking forward to seeing whether they are truly generalizable. Improvements in delivery protocols and detection of KDs will help Cas13 be easier for folks to use, as well.

Beeke Wienert: Oh, I wish I knew! I think that people are trying to move away from actually making a double-stranded break (DSB) in the genome. In rare events, DSBs have been shown to cause chromosomal rearrangements, big deletions or micronuclei and if possible, one would like to avoid those in therapeutic applications (even if rare). We have already seen the technology move towards “non-DSB” innovations such as base editors and more recently, prime editors. I really think that this is where the future lies and that we will see much more of these types of CRISPR editors in the future and in therapeutic applications.

Ben Kleinstiver: The genome editing community has access to a really rich toolbox of technologies that enable a wide range of edits for research applications. How adaptable and applicable some of the newer technologies are for therapeutics remains to be fully demonstrated, but the preclinical data that has been publicly presented are encouraging.

One of the major obstacles where breakthroughs are needed is technologies for tissue-specific delivery and/or regulation of genome editors. Other areas where additional discoveries will be illuminating in the coming years will be aspects related to immunity and the long-term tolerance and performance of edited cells.

Jenny Hamilton: I think the next big thing in the CRISPR field will be the improvement of delivery vehicles to enable in-body gene editing. There have been many recent successes with the therapeutic use of CRISPR-Cas9 tools. For example, CRISPR Therapeutics has demonstrated exciting results editing BCL11A to restore fetal hemoglobin expression as a treatment for beta thalassemia and sickle-cell disease. This was an incredible milestone in the gene-editing field! However, the process of extracting stem cells, performing the targeted gene editing in the laboratory, expanding the cells, chemically ablating the patient’s remaining bone marrow, and then restoring the edited cells is incredibly complicated and is a major barrier to expanding this type of revolutionary treatment to all those who can benefit.

The gene editing field needs to continue prioritizing the development of delivery vehicles to enable safe and effective gene editing of targeted cells in the body to improve the patient experience and enable the scalability and global use of these life-changing treatments.

Sanne Klompe: New ways of bioinformatically mining sequence data reveal interesting new variants of CRISPR-Cas systems and associated genes. It seems nature, too, exploited CRISPR-Cas components for vastly varying purposes. I’m especially excited to see the function of nuclease deficient CRISPR-Cas systems revealed and look forward to their potential applications.

We now have a vast range of tools for precise nucleic acid targeting, and further advancements will provide more control over the actual editing outcomes. Beautiful examples of controlled editing are point mutations generated by base-editors and small changes created with prime-editing. Now, CRISPR-transposons open up exciting new avenues for large kilobase-scale editing independent of HDR. New editing platforms that do not rely on the host’s repair pathways could additionally expand the range of
amenable species and cell types and would in turn open up new therapeutic targets.

Stanley Qi: The next big thing in CRISPR will be delivery. How to efficiently and specifically deliver the CRISPR components in vivo to the therapeutically relevant cells is one major hurdle to its applications for treatment.

 To learn more visit: horizondiscovery.com