Saturation variant interpretation using CRISPR prime editing

Background

Our understanding of the genes associated with genetic disorders has advanced considerably with the advent of next generation sequencing technologies including gene panels, microarrays, and exome sequencing. Clinical genetic testing has become routine for obtaining precise genetic diagnoses in many clinical disciplines from oncology to cardiology to neurology. The rate of clinical genetic testing is expected to continue to increase as the cost of next generation sequencing continues to decline and the benefits of early genetic testing become more widely appreciated.

An ideal outcome of a clinical genetic test is the identification of a pathogenic variant which explains the individual’s phenotype and has an associated precision medicine treatment, such as combination lumacaftor/ivacaftor for the F508del CFTR variant in cystic fibrosis. Currently, the most likely outcome is the discovery of a variant of uncertain significance (VUS), i.e., a genetic change in a potentially relevant gene is detected, but lacks the sufficient evidence to determine whether the genetic variant is pathogenic or benign. Functional assays, in which a VUS is expressed in a cell model to evaluate its activity, have been used for many years to determine whether a particular VUS significantly impacts gene function. However, these experiments are time-consuming and only test one VUS at a time. Developing a new technology is therefore necessary to scale this experiment to tackle the large number of VUS observed in clinical genetic testing.

Multiplexed assays of variant effects (MAVEs) allow for the simultaneous functional characterization of large libraries of VUS generated with techniques such as saturation mutagenesis or saturation genome editing (SGE). In SGE, a genome editor such as SpCas9 is used in concert with a library of guide RNAs (gRNAs) and oligo donors to make many different mutations in a region of a gene. SGE is typically paired with a functional assay, such as cell growth, and next generation sequencing to identify variants with normal or altered function. One of the first MAVEs used SGE to reclassify VUS in the tumor suppressor BRCA1.¹ However, there are several technical limitations of SGE, especially (1) an over-reliance on the haploid cell line HAP1 which is known to spontaneously revert to a diploid state and (2) the use of homology-directed repair (HDR) with low editing efficiency due to the frequent occurrence of nonhomologous end joining (NHEJ). In this preprint, Erwood S et al. sought to address both of these technical limitations through the development of novel haploidized cell lines and the adaptation of prime editing for high-throughput variant characterization.

Key findings

Locus specific haploidization

Haploid HAP1 cells are typically used in SGE experiments to limit each cell to expressing a single variant, which is essential for determining the function of the variant after sorting or selecting cells based on a cellular phenotype. However, this has restricted SGE experiments to genes endogenously expressed in HAP1 cells. In order to overcome this limitation, the authors hypothesized that making cells haploid at a specific locus (a single gene) would enable SGE experiments in many cell lines. A pair of gRNAs specific to heterozygous single nucleotide polymorphisms (SNPs) native to HEK293T (HEK) cells were designed using publicly available genome sequencing data. The authors co-transfected HEK cells and selected a line with only a single remaining functional copy of the NPC1 gene associated with a lysosomal storage disorder known as Niemann-Pick disease type C1. The cell line was validated by experiments to confirm the haploidized line had reduced NPC1 expression by Western blotting and normal cholesterol homeostasis by filipin staining. In theory, a similar approach could be used in virtually any cell line that endogenously expresses a particular gene of interest.

Saturation Prime Editing (SPE) of NPC1 and BRCA2

The authors developed an assay based on staining with LysoTracker, a cell-permeable fluorescence dye that stains lysosomal compartments. The assay was validated using well-characterized missense (p.I1061T and p.P1007A) and nonsense (p.C909X) NPC1 variants. Qualitative and quantitative increases in LysoTracker staining in variant expressing cells were observed relative to parental haploidized HEK cells. Next, prime editing gRNAs (pegRNA) libraries encoding NPC1 variants were generated; importantly, each library contained a silent substitution to disrupt the PAM site and prevent further editing (Figure 1). The mutant pegRNA libraries were cotransfected into haploidized HEK cells with expression constructs for the SpCas9 prime editor as well as a nicking gRNA to increase editing efficiency. Cells were FACS sorted based on LysoTracker signal, followed by DNA extraction and sequencing of the target region of the NPC1 gene. 251/256 (98%) variants were successfully classified as either functional or non-functional based on whether these variants were enriched in the sorted pool of cells with high LysoTracker signal. 14/14 (100%) of control nonsense variants were classified as non-functional, while 60/61 (98%) of control synonymous variants were classified as functional. Interestingly, the majority (121/180) of NPC1 missense variants were classified as non-functional. Editing efficiency in libraries was high, ranging from about 45-73% of the sequencing reads with programmed edits and few of the sequencing reads (about 5-17%) containing small indels.

Figure 1. Establishment of a functional assay for variants in NPC1.

To show that this strategy is generalizable to other genes, the authors further developed a cell growth assay for SPE in the tumor suppressor BRCA2. Through a similar haploidization method followed by SPE and downstream sequencing at two timepoints, the authors assayed 465 different BRCA2 variants. In this case, most (252/300) missense variants were classified as functional, while all nonsense (59/59) were classified as nonfunctional. These data suggest that pairing locus-specific haploidization with SPE enables highly efficient functional characterization at scale.

Why I selected this preprint

I have performed research for a number of years aimed at functional characterization of individual missense variants in epilepsy-associated genes such as SCN3A or SZT2. I have been closely monitoring the literature on different strategies for functional characterization at scale as I believe this is a critical new technology for genetics to manage the large numbers of VUS coming from clinical genetic testing. This preprint addresses a number of significant technical limitations associated with SGE and it will be interesting to see whether other laboratories adopt and perhaps even improve upon some of the techniques reported in this preprint.

Questions for the authors

Q: Is there a concern that locus-specific haploidization alone will induce a cellular phenotype, such that sorting cells based on variants introduced by SPE is not possible? For example, in cases of haploinsufficiency where a 50% reduction is protein expression is sufficient to cause a genetic disorder. It may be that prime editing of a single allele may not produce significant effects beyond the difference seen between that haploidized cell line relative to its parental control line. Do you have an alternative strategy for these genes which may work?

Q: Are the costs of performing SPE similar to SGE? I would imagine with the more complex nature of pegRNAs relative to sgRNAs, there might be additional costs associated, but these costs may also be recouped later by the relative lack of indels (ie. Wasted cells).

References

1. Findlay, G.M., Daza, R.M., Martin, B., Zhang, M.D., Leith, A.P., Gasperini, M., Janizek, J.D., Huang, X., Starita, L.M., and Shendure, J. (2018). Accurate classification of BRCA1 variants with saturation genome editing. Nature 562, 217-222.

Tags: brca2, crispr, npc1, sge, spe, variant, vus

doi: https://doi.org/10.1242/prelights.30851

Read preprint (3 votes)