Immortalization of primary marmoset skin fibroblasts by CRISPR-Cas9-mediated gene targeting

ABSTRACT Immortalized cell lines can be used for diverse in vitro experiments, providing invaluable data before conducting in vivo studies Callithrix jacchus, the common marmoset, is a non-human primate model utilized for studying various human diseases. However, only a few immortalized marmoset cell lines are currently available. In the present study, we reveal that CRISPR-Cas9-mediated targeting of the p53 gene or CDKN2A locus is an effective means for immortalizing primary marmoset skin fibroblasts. In addition to frameshift mutations that result in premature stop codons, in-frame mutations potentially destroying the DNA-binding motif of p53 are frequently detected in immortalized cells. Like Cdkn2a-deficient mouse cells, CDKN2A-deficient marmoset cells express wild-type p53 proteins normally respond to genotoxic stresses, including adriamycin and etoposide. Taken together, these findings indicate that Cas9- mediated gene targeting of the p53 gene or CDKN2A locus is an effective tool for establishing immortalized marmoset cell lines with defined genetic alterations.


Introduction
The common marmoset, Callithrix jacchus, is an emerging non-human primate model used in neuroscience, immunology, reproductive science, and toxicology studies (Kishi et al. 2014). Despite the increasing demand for these animals, their supply is limited, as common marmoset is an endangered species (Malukiewicz et al. 2015). The supply of immortalized cell lines is unlimited, and their use for in vitro experiments before animal studies is an essential tool to minimize the number of in vivo studies sacrificing animals (Dumenco et al. 1995). However, only a limited number of immortalized marmoset cell lines are currently available (Guo et al. 2018;Petkov et al. 2018). This necessitates development of immortalized marmoset cell lines from specific tissues to be studied.
Because cell immortalization is a critical process for carcinogenesis, numerous immortalized cell lines have been isolated from various human cancers. Different from mouse embryonic stem (ES) cells , normal diploid human cells eventually enter replicative senescence after a finite number of cell divisions (Hayflick 1965) and thus, to become immortalized, cells must escape replicative senescence, acquiring the capability of infinite cell division. Replicative senescence is attributable to shortening of telomeres, the ends of chromosomes, with each cell division (Blackburn 2000). Because normal somatic cells lack the telomerase activity necessary for telomere elongation, telomeres are gradually shortened by successive cell divisions, and finally, critically short telomeres trigger replicative senescence (Shay et al. 2001). Consistently, exogenous telomerase overexpression causes elongation of telomeres in primary diploid cells, which then actively proliferate, accompanied by suppressed expression of senescence biomarkers (Bodnar et al. 1998). Similarly, primary marmoset skin fibroblasts can be immortalized using the human telomerase catalytic subunit (hTERT) gene (Petkov et al. 2018).
p53 is the most frequently mutated gene in human cancers, and approximately 90% of p53 mutations are missense mutations in the DNA-binding domain (Baugh et al. 2018). p53 mutations are common in spontaneously immortalized primary mouse embryonic fibroblasts (MEFs) (Harvey and Levine 1991), and overexpression of the mutant p53 gene (del329) immortalizes normal human mammary epithelial cells (MECs) (Gao et al. 1996). Similarly, cellular senescence does not occur in p53-deficient primary MEFs (Harvey et al. 1993).
The CDKN2A locus, also known as INK4A/ARF locus alternatively encodes two tumor suppressor proteins, p16 INK4A and p14 ARF (p19 Arf in mice), and is one of the most frequently mutated genes in human cancers. Most mutations cause functional loss of both p16 INK4A and p14 ARF (Sharpless and DePinho 1999). In mice, genetic ablation of both p16 Ink4a and p19 Arf allows primary MEFs to consistently grow with no detectable senescent phase (Serrano et al. 1996). p19 Arf -deficient MEFs with retention of p16 Ink4a gene expression are also immortal (Kamijo et al. 1997), but immortalization of p16 Ink4a -null MEFs is significantly lower than that of MEFs null for the Cdkn2a locus or p19 Arf gene (Sharpless et al. 2001), indicating a predominant role for the p19 Arf gene in cellular senescence.
In the present study, we determined if deficiencies in the p53 gene or CDKN2A locus could immortalize primary common marmoset skin fibroblasts. To induce null mutations in these genes, a CRISPR-Cas9 system from Streptococcus pyogenes (Cho et al. 2013;Cong et al. 2013;Mali et al. 2013) was adapted and provided a simple and robust means for efficient immortalization of primary skin cells from common marmosets. This platform can potentially be used to establish immortalized cell lines that are essential for in vitro studies prior to in vivo experiments using common marmoset monkeys, which are available in very limited numbers due to their status as an endangered species.

Materials and methods
Animal, biopsy, and primary skin cell culture A 3-year-old female common marmoset was purchased from laboratory animal supplier, CLEA (Kawasaki, Japan). Animal procedures were approved by the Institutional Animal Care and Use Committee of the Biomedical Research Institute at the Seoul National University Hospital, an accredited research institute of the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC; SNUH-IACUC No: 20-0161). An abdominal skin sample was collected during autopsy. Primary skin fibroblasts were isolated as described previously (Seluanov et al. 2010). Briefly, 7 days after digestion of skin tissues using Liberase TM (Roche) in DMEM/F12 (Gibco), cells were transferred to EMEM (ATCC) containing 10% (v/v) FBS (Sigma-Adrich) and 10 unit/ml Penicillin/Streptomycin. Upon confluence, cells were detached using 0.05% Trypsin-EDTA (Invitrogen) and were then cultivated with 10% (v/v) FBS/DMEM media supplemented with Penicillin/ Streptomycin.
For Sanger sequencing, PCR products were subcloned using a T-blunt TA cloning kit (SolGent Co., Ltd., Seoul, Korea), and miniprep DNA samples were then prepared using a DNA-spin™ Plasmid DNA Purification Kit (iNtRON Biotechnology, Inc., Seongnam, Korea) and were sequenced at Macrogen (Seoul, Korea).
In vitro growth and senescence-associated βgalactosidase (SAβ-gal) assays To measure cumulative population doublings, cells were plated at 3 × 10 5 cells/60 mm dish or 1 × 10 6 cells/ 100 mm dish. Confluent cultures were trypsinized, viable cells were counted, and cells were replated at the same density. For viable cell counting, aliquots of trypsinized cells were stained with Trypan Blue staining solution (Invitrogen) and were counted using a Countess II cell counter (Invitrogen) equipped with a Countess TM cell counting chamber slide (Invitrogen). If tissue cultures were not confluent, cells were grown until confluence after adding fresh media. To count senescent cells, SAβ-gal assays were conducted using a Senescence β-Galactosidase Staining Kit (Cell Signaling) according to the manufacturer's instructions ). Photographs were taken using a Nikon eclipse Ts2 microscope (Nikon, Tokyo, Japan) equipped with an HK3.1 CMOS digital camera and HKBASIC Software (KOPTIC).

Western blot analysis
Cells were treated with adriamycin (Sigma-Aldrich) or etoposide (Sigma-Aldrich), and SDS protein samples were then prepared by directly lysing the cells in SDS sample buffer (Elpisbiotech. Inc., Daejeon, Korea) after washing with ice-cold phosphate-buffered saline (PBS).

Results
Targeted inactivation of the p53 gene in primary marmoset skin fibroblasts using CRISPR-Cas9 Because most mutations of the p53 DNA-binding domain cause reduction or complete loss of p53 transcriptional activity (Baugh et al. 2018), we designed a sgRNA specific for exon 7, which encodes critical amino acid residues of the p53 protein DNA-binding domain (Figure 1(A)). We could not find any potential off-target sites with less than three mismatches from the sgRNA on the marmoset genome. When this sgRNA was expressed with Cas9 using a lentiviral vector, indel mutations were robustly induced in primary marmoset skin fibroblasts (Figure 1(B)). As polyclonal marmoset skin fibroblasts actively proliferated and contained diverse indel mutations ( Figure S1), we established multiple clonal cell lines with in-frame and/ or frameshift mutations (Figure 1(C)).
Among the p53-mutated clones, we selected mutant clone #4, which was homozygous for a 1 bp insertion in the target sequence (Figure 1(C)). Because of the frameshift mutation, clone #4 expressed a mutant p53 protein comprised of ∼1-244 amino acid residues of wild-type marmoset p53 protein (395 amino acid residues,  XM_002747948) and 20 aberrant amino acid residues. Accordingly, while Western blot analysis detected a wild-type p53-specific signal in parental primary skin fibroblasts, p53 mutant cells strongly expressed a smaller mutant p53 protein that is approximately 43 kDa, but not wild-type p53 protein (Figure 1(D)).
Targeted mutations of the CDKN2A locus in primary marmoset skin fibroblasts using CRISPR-Cas9 The CDKN2A locus encodes two proteins, p16 INK4A and p14 ARF , with exon 1 sequences, 1α and 1β, respectively, and alternative reading frames of the common exons in humans and mice (Quelle et al. 1995). The CDKN2A locus in the marmoset genome also encodes the p16 INK4A gene (Figure 2(A)). Although marmoset p14 ARF protein and transcript have not yet been annotated, the marmoset genome harbors a putative exon 1β that is upstream of exon 1α (Figure 2(A)). Accordingly, the predicted transcript variant X2 (XM_035306720), which harbors putative exon 1β, alternatively encodes the putative marmoset p14 ARF protein ( Figure S2A), which is highly homologous to human p14 ARF protein ( Figure S2B).
To inactivate both the p16 INK4A and p14 ARF genes simultaneously, we selected an sgRNA specifically targeting the common exon 2 (Figure 2(A)) and confirmed that the sgRNA actively induced indel mutations in primary marmoset skin fibroblasts (Figure 2(B)). The sgRNA has an off-target site with a 1-bp mismatch on the highly homologous CDKN2B gene. Except for this off-target site, there is no potential off-target sites with less than three mismatches from our CDKN2A gene-specific sgRNA on the marmoset genome (Tsai et al. 2015;Kim et al. 2016). The CDKN2A-targeted marmoset skin fibroblasts were used to establish clonal cells deficient for both p16 INK4A and p14 ARF . We detected a sequence variation, XM_035306720 c.286G > A (p.Gly96Ser), which was prevalent in our wild-type marmosets ( Figure S3). However, the 1 bp mismatch between the target sequence and the sgRNA did not compromise the action of CRISPR-Cas9, and thus most of the clones had both insertions and deletions resulting in simultaneous frameshifts in the p16 INK4A and p14 ARF genes (Figure 2(C)). For example, the 23 bp-deleted clone #1 encoded mutant p16 INK4A protein was composed of 76 amino acid residues of wild-type p16 INK4A protein at the N-terminus and 28 amino acid residues of p14 ARF protein at the C-terminus ( Figure S4A), reflecting the alternative reading frame in the CDKN2A locus. For the p14 ARF gene, the frameshift mutation of clone #1 involved the third reading frame that was not used by either the p16 INK4A or p14 ARF genes, and thus the open reading frame encoded a mutant p14 ARF protein containing 97 amino acid residues of wild-type p14 ARF protein at the N-terminus and 103 mutated amino acid residues at the C-terminus ( Figure S4B).
We confirmed expression of p16 INK4A and p14 ARF in CDKN2A-mutant clone #1. When compared with those of parental cells, transcription of the p16 INK4A gene was slightly decreased, while transcription of the p14 ARF gene was unchanged (Figure 2(D)). Due to the 23 bp deletions in CDKN2A-mutant clone #1, the sizes of the p16 INK4A and p14 ARF cDNAs were slightly smaller than those of the parental cells (Figure 2(D)). We also measured expression of p16 INK4A and p14 ARF proteins in the CDKN2A-mutant cells using p53-mutant clone #4 as a positive control (Figure 2(E)). p16 INK4A protein was robustly expressed in p53-mutant cells, but not in CDKN2A-mutant cells (Figure 2(E)). Although a weak non-specific signal was present, p53-mutant cells robustly expressed p14 ARF protein, but CDKN2A-mutant cells did not (Figure 2(E)).
We also monitored the off-target mutation of the CDKN2B gene ( Figure S5). The sequencing result showed the co-existence of wild-type and 9-bp deleted mutant alleles ( Figure S5).

Characterization of p53-and CDKN2A-mutant marmoset cells
We examined whether p53-and CDKN2A-mutant marmoset cells continued to divide without developing senescence. Cumulative population doubling levels (PDLs) of mutant marmoset cells were measured and compared with that of primary marmoset cells ( Figure  3(A,B)). The proliferation of primary cells gradually slowed and eventually ceased (Figure 3(A,B)). For p53mutant marmoset cells, both polyclonal and monoclonal cells continuously proliferated without decreased growth rates and had similar growth characteristics (Figure 3(A)). p53-mutant cells (clone #4) were thawed at passage 8 and cell growth was measured from passage 10 to passage 33, resulting in 17 cumulative population doublings (PDs). For CDKN2A-mutant cells, the growth rate of polyclonal cells did not decrease (Figure 3(B)). Compared with polyclonal cells, CDKN2Amutant clone #1 exhibited an increased growth rate (Figure 3(B)) with 63 cumulative PD when cell growth was measured from passage 18 to passage 35. CDKN2A-mutant cells retained higher proliferation capacities than those of p53-mutant cells (Figure 3(A,  B)). Most parental cells were enlarged and flattened by passage 18 compared with passage 8, a characteristic feature of senescent cells. In addition, p53 and CDKN2A-mutant cells did not lose their fibroblast characters (Figure 3(C) and Figure S6). Notably, senescence-associated β-galactosidase (SA-βgal)-positive cells were frequently present among primary marmoset cells at passage 18, but not in p53 (passage 23)-and CDKN2A (passage 24)-mutant clones (Figure 3(C)). Taken together, these data support establishment of genetically defined immortalized marmoset cell lines.
Genotoxic stresses activate p53 and would thus be expected to increase expression of p53 target genes in the CDKN2A-mutant cell line with intact p53. We treated CDKN2A-and p53-mutant cell lines with adriamycin and etoposide and measured expression of p53 and its target genes (Figure 4). Adriamycin stabilized p53 protein and increased MDM2 and p21 WAF1 protein levels in the CDKN2A-mutant marmoset cell line, which did not occur in the p53-mutant marmoset cell line (Figure 4(A)). Transcript levels of p53 target genes MDM2 and p21 WAF1 were concomitantly increased in the CDKN2A-mutant cell line, but not in the p53-mutant cell line (Figure 4(B)). Consistently, etoposide treatment also increased p53 protein levels and target gene expressions (Figure 4(C,D)).
Cumulatively, these findings demonstrated that CRISPR-Cas9-mediated knockout of the p53 gene and CDKN2A locus is a simple means to establish immortalized cell lines from marmoset skin biopsies. In turn, the resulting combination of p53-proficient (CDKN2Amutant) and p53-deficient cells is an excellent in vitro model for the study of genotoxic stresses.

Discussion
Immortalized human cell lines are essential for in vitro experiments and have contributed enormously to our understanding of the cellular and biochemical mechanisms underlying biological phenomena and pathological conditions. Common marmosets are a versatile nonhuman primate model suitable for studying a broad range of human diseases and generating gene-edited animal models. Because extensive in vitro experiments should be conducted prior to in vivo experiments using live marmosets due to limited animal availability and ethical considerations, immortalized cell lines derived from marmoset tissues of interest will greatly enhance the efficiency of these in vitro studies.
Recently, several reports have described successful generation of immortalized marmoset cell lines, and our study will provide an alternative and complementary method (Guo et al. 2018;Petkov et al. 2018;Orimoto et al. 2022). As in normal human cells that undergo a finite number of cell divisions (Bodnar et al. 1998), piggy-Bac transposon-mediated expression of the hTERT gene successfully immortalizes marmoset skin fibroblasts (Petkov et al. 2018). By contrast, hTERT expression alone is not sufficient for immortalization of marmoset muscle fibroblasts; a mutant form of cyclin-dependent kinase 4 (CDK4 R24C ) and overexpression of wild-type Cyclin D1 are also required (Orimoto et al. 2022). CDK4 R24C mutation predisposes humans to hereditary melanoma and abolishes the ability of CDK4 protein to bind p16 INK4A protein (Wolfel et al. 1995;Zuo et al. 1996). Therefore, it can be reasonably expected that the p16 INK4A -mediated senescence signal was perturbed by expression of CDK4 R24C in the immortalized marmoset muscle fibroblasts. Based on our results, CDKN2A gene knockout could exert a comparable effect to introducing the CDK4 R24C transgene.
As previously reported in human-induced hepatocytes (hiHeps) (Huang et al. 2014), the simian virus 40 (SV40) large T-antigen also immortalizes marmoset hepatic progenitor cells (Guo et al. 2018). SV40 large Tantigen is effective for cellular immortalization but inactivates the p53 gene (Ahuja et al. 2005). This suggests that SV40 large T-antigen-immortalized cells would not be suitable for study of p53-dependent cellular responses. Therefore, marmoset cells immortalized by CDKN2A gene targeting are preferable, as these cells retain functional p53. In support of this, we demonstrated that genotoxic stress-induced gene expression could be analyzed using CDKN2A-and p53-deficient immortalized cells as a pair.
The present study contributes to the development of novel marmoset cell lines from tissue types of interest. Our strategy can be improved by avoiding the offtarget mutation detected in the highly homologous CDKN2B gene ( Figure S5). It will be accomplished by targeting the exon 1β to generate p14 ARF -deficient but p16 INK4A -proficient cell lines as p19 Arf deficiency alone makes primary MEFs immortal (Kamijo et al. 1997). In addition, vector-free or adeno-associated virus (AAV) vector systems are suitable for minimizing off-target effects induced by the constitutive expression of the CRISPR-Cas9 system. These cell lines will expedite biomedical studies using common marmoset monkeys and minimize the scale of animal experiments by allowing conduct of extensive in vitro experiments prior to initiating in vivo studies.