Modeling human cancer predisposition syndromes using CRISPR/Cas9 in human cell line models

The advancement of CRISPR mediated gene engineering provides an opportunity to improve upon preclinical human cell line models of cancer predisposing syndromes. This review focuses on using CRISPR/Cas9 genome editing tools to model various human cancer predisposition syndromes. We examine the genetic mutations associated with neurofibromatosis type 1, Li‐Fraumeni syndrome, Gorlin syndrome, BRCA mutant breast and ovarian cancers, and APC mutant cancers. Furthermore, we discuss the possibilities of using next‐generation CRISPR‐derived precision gene editing tools to introduce a variety of genetic lesions into human cell lines. The goal is to improve the quality of preclinical models surrounding these cancer predisposition syndromes through dissecting the effects of these mutations on the development of cancer and to provide new insights into the underlying mechanisms of these cancer predisposition syndromes. These studies demonstrate the continued utility and improvement of CRISPR/Cas9‐induced human cell line models in studying the genetic basis of cancer.

myelinating microglia cell of the peripheral nervous system. 1 These NF1 À/À tumors are benign but can be extremely disfiguring and have the ability to transform into malignant peripheral nerve sheath tumors (MPNST), a distinctly heterogenous soft tissue sarcoma derived from Schwann cells lineage which currently has no effective treatment. 2 Modeling the transformation of wild-type primary human Schwann cells (SC) to MPNST remains a challenge. Problems with long term culturing of hSC using current tissue culture conditions is a significant roadblock in modeling NF1 related tumors. The generation of immortalized hSC lines through transduction of murine Cdk4 and hTERT has allowed for continual propagation in vitro. 3 These immortalized SCs show similar phenotypic characteristics as primary hSCs providing a quality model of disease mechanisms, drug screening, 4 and different CRISPR screens for frailties. 5 MPNST cells lines have also been generated from patient biopsies to capture common mutations in their development. Frequent mutations have been found in either the polycomb repressor complex 2 (PRC2) or TP53. 6 While these cell lines are important, how well they recapitulate the primary tumor is an important unknown detail. Advancements in current model technologies can be applied to MPNST to better recapitulate these heterogenous tumors associated with NF1 loss.
Modeling all types of NF1-associated peripheral nerve tumors is critical to understanding their development from SCs to MPNST.
Using induced pluripotent stem cell (iPSC) reprogramming methods, Carrio et al. successfully generated iPSCs from several NF1À/À plexiform neurofibromas (PNF). 7 These tumor derived iPSCs showed faithful differentiation potential toward neural crest cell (NCC) and hSC lineages through detection of lineage markers in 2D and 3D culture. 7,8 When injected into the sciatic nerves of nude mice, these NF1 À/À PNF-derived SCs generated PNF like-tumors. 8 Other models of neurofibromas and MPNSTs have been created with iPSC derived Schwann cell lineage populations. The introduction of patient-specific loss of function mutations in NF1 allowed the increased expansion of an iPSC derived Schwann cell precursor (SCP) population, and elicited abnormal SC lineage differentiation. 9 Implantation these NF1 À/À SC lineage cells into the sciatic nerve of nude mice generated humanized neurofibromas. 9 In clinical NF1-associated MPNSTs, p53 and PTEN loss is sometimes observed. 10 Mo et al. modeled this by introducing CRISPR-Cas9 knockout of TP53 and NF1 in human iPSCs. This inhibited differentiation to SCPs and formed MPNST-like tumors when injected orthotopically into the sciatic nerve of nude mice. 9 These experiments show the promise and utility of iPSC-based modeling of NF1-associated MPNST.
The polycomb repressive complex (PRC2) complex plays a role in the development of NF1-associated MPNSTs through loss one of two PRC2 components SUZ12 or EED. 11 Generating iPSC derived SCs and SCPs lacking NF1 and SUZ12 could offer a novel preclinical model in which to test a range of known therapies, as well as discover novel drivers of MPNST-genesis.
In addition to peripheral nerve sheath tumors, patients with germline mutations in NF1 are at increased risk for neurological problems such as autism and attention deficient disorder. 1 The ease and accuracy of CRISPR-mediated gene editing allows for a larger number of these variants to be studied further and any potential genotype-phenotype relationships to be observed. Some prominent genotype-phenotype relationships in different NF1 mutations that have been uncovered and dictate a different prognosis. 12 Neurofibromin mediated activation of PKCζ leads to GPCR desensitization of its inhibitor GRK2 and subsequent cAMP elevation from adenylyl cyclase. 13 Initial work from Anastasaki et al. has combined iPSC and CRISPR technology to create 3-D organoid models that recapitulate this inherent variability caused by various NF1 mutations at both cellular and tissue levels. 14 Neurofibromin is a large gene spanning 350 kilobases and containing 60 exons, yet no defined "hotspot" NF1 mutations have been described.
Further investigation into potential correlation between specific variants and clinical relevance is currently underway. Recently, a large-scale analysis of 4610 germline NF1 mutations uncovered 783 benign and 938 pathogenetic variants. 15 Additional iPSC-derived human models of these specific mutations could be generated with next generation CRISPR techniques. For example, phenotypic correlation of targeted missense mutations at specific pathogenic domains could be generated with CRISPR-pass technology. 16 Other common pathogenic variants include frameshift mutations, splice-site alterations, nonsense mutations, and few synonymous mutations. 15 Once characterized, select candidate mutations could be generated in 2D or 3D cultures using human cell line models to create a more effective and predictive preclinical model.

| LI-FRAUMENI SYNDROME (TP53)
Li-Fraumeni Syndrome (LFS) is a hereditary cancer predisposition syndrome that results from a mutation in a single allele of TP53. 17 TP53 an extremely well-known tumor suppressor gene that is aptly named 'the guardian of the genome' due to its central role in responding to multiple forms of DNA damage. The likelihood of developing cancer for a LFS patient carrier is nearly 75% and 100% for males and females, respectively. 17 There is not an organ specific predilection for cancer initiation in LFS patients that carry a germline TP53 mutation as all their somatic cells are heterozygous for TP53. The most common forms of cancer in LFS patients are breast, sarcoma, brain, and bone. 17 Sequencing LFS patient genomes revealed different types of germline mutations throughout the TP53 coding sequence. Missense mutations were most common (74%), followed by nonsense mutations (9%), and splicing mutations (8%). 17 Most of these mutations landed in the DNA binding domain (codons 175, 245, 248, 273, and 282). Other significant mutational areas include a 72Arg polymorphism shown to have an increased affinity toward MDM2, causing higher levels of p53 degradation, and a 16-bp duplication in intron 3 (PIN3). Those patients carrying one minor allele (16 bp duplication) were found to have a first cancer diagnosis an average of 17.1 years later than individuals carrying two major (nonduplicated) alleles. 17 As LFS was one of the first described cancer predisposition syndromes there have been several preclinical genetically engineered mouse and human cell line models developed to search for effective therapeutics. These models are well reviewed 18

| GORLIN SYNDROME (PTCH1)
Gorlin syndrome (GS) is a rare genetic disorder caused by mutations in the PTCH1 gene. PTCH1 encodes a receptor that functions as a negative regulator of the Hedgehog signaling pathway, which is involved in various cellular processes such as cell division, differentiation, and growth.
More than 90% of GS cases are associated with mutations in PTCH1. 20 The most common PTCH1 mutations associated with GS are frameshift and nonsense mutations. Other described PTCH1 mutations include splicing mutations, large deletions, and missense mutations. 20 However, there is no clear relationship between specific mutations and the resulting phenotype, and no clear mutation hotspots have been  22 Further, introduction of cooperating mutations into these PTCH1 mutant NES cells, either in DDX3X or GSE1 led to an acceleration of tumorigenesis. 22 A similar study confirmed that iPSCs derived from four independent GS patients heterozygous for PTCH1 +/À could generate medulloblastoma when implanted into immunodeficient mice while wild-type controls could not. 21 These initial studies provide evidence to support that further investigation into specific PTCH1 mutations associated with GS is warranted using this novel model approach. Introduction of other mutations could be achieved through a variety of CRISPR-Cas9 mediated techniques discussed in this review.

| HEREDITARY BREAST AND OVARIAN CANCER SYNDROME (BRCA1/2)
BReast CAncer genes 1 and 2 (BRCA1/BRCA2) are tumor suppressor genes in the DNA damage repair pathway. Interacting at double stranded DNA breakage sites (DSBs), BRCA1 and BRCA2 direct DNA repair through DNA damage checkpoint activation and repair, and mediation of homology directed repair, respectively. As both of these genes have important functions in maintaining genomic integrity, mutations a single allele in either gene results in hereditary breast and ovarian cancer syndrome (HBOC). 23 Inherited mutations in BRCA1/2 have been discovered and analysis of genetic testing correlates high risk of breast and ovarian cancer development with specific variants. 24,25 In cases of pathogenic BRCA1/2 variants inheritance, the cumulative risk of developing breast cancer is 72% for BRCA1 mutations and 69% for BRCA2 mutations, and the risk of developing ovarian cancer is 44% for BRCA1 mutations and 17% for BRCA2 mutations in women. 26 Recently, the creation of phenotypically relevant organoid's gen- Classic FAP patients possess a dramatically increased risk of colorectal cancer (CRC), and are associated in most cases with truncating mutations in central region of the APC gene. 34 Attenuated FAP has reduced penetrance, is characterized by fewer polyps (mean of 30), a reduced risk of CRC compared to classic FAP, and is associated in most cases with mutations in the extreme 5 0 or 3 0 end of the APC coding region. 34 A third syndrome called gastric adenocarcinoma and proximal polyposis of the stomach (GAPPS) is characterized by gastric polyps, fundic gland polyps, and germline pathogenic mutations in the APC gene 1B promoter. 34 As such, these conditions are a good demonstration of the way phenotypes can vary greatly depending on the mutation or genotype present for a given gene.
In theory, APC gene genotype-phenotype correlations can be studied and recapitulated using iPSC-based models. Standardized methods to created colonic organoids, by in vitro differentiation of iPSC to definitive endoderm and then hindgut endoderm, have been developed and improved. 35 These have allowed classic FAP to be modeled using iPSC using cells from FAP patients and by introduction of APC gene mutations using CRISPR or TALENs, into wild type iPSC or human embryonic stem cell lines. 35,36 These studies have been useful in showing that heterozygous APC mutations cause greater canonical WNT/b-catenin signaling, proliferation, reduced differentiation potential, changes in cell polarity and chromosomal instability. [35][36][37] In addition to the benefits of isogenic human iPSC models for studies like these, it should be possible for these models to be used in many other useful ways. Indeed, Crespo et al. described using the FAP colonic organoids for drug studies aimed at reversing the effects of truncating APC gene mutations on WNT/b-catenin signaling and hyperproliferation. 35 This paper also showed that non-sense mutations in APC could be effectively suppressed using geneticin, which reversed FAP-like features in mutant colonic organoids without effects on wild type organoids. In the future, it should be possible to use CRISPR gene editing to model loss of the wild type allele in FAP colonic organoids, a genetic step that occurs during progression to CRC. As described by others, it should also be possible to use iPSC organoids to model diet-microbiome-host interactions known to be important in risk for development of CRC. 38 While attenuated FAP and GAPPS have not been modeled in iPSC using CRISPR, genotypephenotype studies of this sort will also be possible. It should be possible to perform genetic screens using APC mutant iPSC and colonic organoids for factors that enhance transformation or suppress features of FAP or GAPPS. Finally, it is worth noting that classic FAP patients are predisposed to cancers other than CRC, and to structural abnormalities that could be modeled using an iPSC platform. 34

| NEXT-GENERATION CRISPR DERIVED PRECISION GENE EDITING TECHNOLOGIES
The use of CRISPR-Cas has become synonymous with interrogating gene function and interactions in a wide variety of cell types. There have been numerous uses of the CRISPR-Cas system that deviate from the initially described programmable RNA guided DNA endonuclease system which initiates a double stranded DNA break. These original gene editing CRISPR-Cas9 systems and subsequent derivatives such as base editors and prime editors and their potential uses in iPSC derived models of cancer are thoroughly reviewed in detail elsewhere. [39][40][41][42] Here we provide an up-to-date summary of several novel methods of large-scale gene integration, additional improvements to prime editing, and how these gene editing technologies can be applied to improve human cell line models of cancer predisposition syndromes.

| TARGETED INTEGRATION OF LARGE DNA CARGO
The insertion of large cargo into a defined locus in mammalian cells has been a stochastic and/or inefficient process that may rely on Yet another exciting targeted large payload integration method was recently described by Blanch-Asensio et al. 45 The method, called STRAIGHT-IN, utilizes a nuclease-free approach and achieves high efficiency and accuracy in gene targeting in human iPSCs ( Figure 1C). The authors report successful insertion of DNA sequences up to 100 kb in size, which is significantly larger than what is achievable using traditional methods. By first inserting "landing pads" into the AAVS1 locus containing an attP site, fluorescent reporter, and positive antibiotic selection marker, the authors isolated stable iPSC clones primed for integration.
Transfection of these iPSCs with serine integrase Bxb1 and DNA cargo containing attB sites allowed for integration of extremely large sequences such as a 173 kb bacterial artificial chromosome. 45 Through clever inclusion of unique loxP sites surrounding their landing pad, exogenous administration of Cre-recombinase after integration led to an almost "scarless" integration through removal of residual landing pad sequences. One could utilize this system to generate a barcoded library of specific variants and integrate into the recipient cells, induce differentiation toward a desired cell type and monitor the frequencies of variants in the population overtime.

| ENHANCING THE UTILITY OF PRIME EDITORS
Prime editor has previously been described to contain SpCas9 nickase directly tethered to a MMLV reverse transcriptase. 46 This can limit the packaging of this complex into viral vectors for delivery in vitro or in vivo due to size constraints. One effort to overcome this restriction is to engineer efficient prime editors with small, untethered reverse transcriptase (RT). 47 Using this approach, different RTs can quickly be screened for enhanced activity in many different contexts. The authors found that using a split DNA plasmid prime editor system with a mutant MMLV RT that lacks the RNAse H domain, they could perform edits at the same efficiency as a tethered RT system while saving almost 1500 bp in packaged sequence ( Figure 2A). When packaged into a dual adeno-associated virus (AAV) vector, their untethered prime editor installed targeted transversion mutations in U2OS cells at a frequency of 4%. Although improvement in efficiency is desired, further optimization and removal of unnecessary sequences in the prime editor system could allow for packaging of proteins described to enhance prime editing in other reports. 48 Additional work from the lab of Dr. David Lui continues to improve upon the primary reports of the prime editing system. The use of base editing or prime editing to directly install a given mutation or create a landing pad for serine-integrase mediated delivery of a sequence of interest are alternative solutions to avoid these potential copy number anomalies.
The expansion of available protospacer adjacent motif (PAM) sequences was not described in detail in this review, but we encourage readers to refer to this insightful review for further reading. 51 A recent study of note demonstrates the use of a near-PAMless SpCas9 variant (SpRY) for digesting double stranded DNA with an apparent lack of PAM sequence requirement. 52 Off target effects were see at nearly identical gRNAs sequences, but this was overcome through tar- include creation of autologous co-culture systems consisting of mutant iPSC derived tumors and immune cells. This would allow for investigation into potential immune checkpoint blockade assays to identify mechanisms of resistance and sensitivity similar to what is seen in ex vivo tumor fragment platforms. 53 Recently, a study identified heritable defects in telomere and mitotic function can predispose patients to develop specific sarcomas. 54 Through whole genome sequencing of more than 1600 sarcoma patients they found several variants of unknown significance that were deemed likely pathogenic (>26 000). Beginning to identify and functionally validate these and other variants of unknown signifi-

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.