Flow cytometry-based quantification of targeted knock-in events in human cell lines using a GPI-anchor biosynthesis gene PIGP

Abstract Targeted knock-in supported by the CRISPR/Cas systems enables the insertion, deletion, and substitution of genome sequences exactly as designed. Although this technology is considered to have wide range of applications in life sciences, one of its prerequisites for practical use is to improve the efficiency, precision, and specificity achieved. To improve the efficiency of targeted knock-in, there first needs to be a reporter system that permits simple and accurate monitoring of targeted knock-in events. In the present study, we created such a system using the PIGP gene, an autosomal gene essential for GPI-anchor biosynthesis, as a reporter gene. We first deleted a PIGP allele using Cas9 nucleases and then incorporated a truncating mutation into the other PIGP allele in two near-diploid human cell lines. The resulting cell clones were used to monitor the correction of the PIGP mutations by detecting GPI anchors distributed over the cell membrane via flow cytometry. We confirmed the utility of these reporter clones by performing targeted knock-in in these clones via a Cas9 nickase-based strategy known as tandem paired nicking, as well as a common process using Cas9 nucleases, and evaluating the efficiencies of the achieved targeted knock-in. We also leveraged these reporter clones to test a modified procedure for tandem paired nicking and demonstrated a slight increase in the efficiency of targeted knock-in by the new procedure. These data provide evidence for the utility of our PIGP-based assay system to quantify the efficiency of targeted knock-in and thereby help improve the technology of targeted knock-in.


Introduction
Targeted knock-in is a methodology of genome engineering with which a specified DNA element or a base substitution is introduced into a predetermined position of the genome. Because targeted knock-in allows for modifications of the genome exactly as designed, this methodology is considered promising for future applications in various fields of life sciences, including clinical medicine and agriculture [1][2][3][4]. Recently, the efficiency of targeted knock-in was dramatically improved by the application of genome editing technologies such as the CRISPR/Cas systems [5] to enhance homology-directed DNA recombination [6,7]. However, efforts are still being made to improve the efficiency, precision, and specificity of CRISPR/Cas-assisted targeted knock-in [8][9][10][11][12].
To modify the procedure of targeted knock-in and improve the efficiency achieved, we first need a molecular system that allows for the sensitive and quantitative detection of targeted knock-in events. We previously developed such a molecular system in which an inactive mutant PIGA gene in a human cell line is corrected to a wild-type gene by targeted knock-in [13]. The PIGA gene, located on the X chromosome, is essential to biosynthesize glycosylphosphatidylinositol (GPI) anchors which, after being synthesized, are distributed over the cell membrane [14,15]. Thus, cells carrying wild-type but not mutant PIGA are sensitive to aerolysin, a cytolytic toxin that binds to GPI-anchored proteins and forms pores on cell membranes. PIGA-disrupted cells are therefore selectable with a precursor of aerolysin, namely proaerolysin [16,17]. In addition, cells carrying inactive and reverted PIGA can be counted by flow cytometry (FCM) detecting fluorescently stained GPI anchors on the cell surface [18,19]. This latter assay, named PIGA correction assay, allowed us to sensitively measure the efficiency of targeted knock-in occurring within an endogenous human gene with a simple, high-throughput, and relatively inexpensive procedure [13].
Despite the advantages of the FCM-based quantification of targeted knock-in events elicited within an endogenous gene, to the best of our knowledge, the PIGA correction assay has been the only established monitoring system achieving all the above-mentioned benefits. To eliminate possible gene locus-specific effects and experimental artifacts unique to the PIGA correction assay, a second system measuring the efficiency of targeted knock-in based on a different endogenous platform gene is critical. Accordingly, in the present study, we developed a distinct FCM-based molecular system that allows us to quantitatively monitor the correction of an inactivating mutation in two near-diploid human cell lines, using PIGP as a platform gene. PIGP is another GPI-anchor biosynthesis gene whose product forms a complex with PIGA [20]. We also employed the resultant PIGP-based assay (hereafter referred to as PIGP correction assay) in an attempt to improve the procedure of tandem paired nicking (TPN) [13], a recently developed nick-based strategy for precision targeted knock-in.

General molecular biology techniques
Extraction of genomic DNA (gDNA) was conducted using the PureLink Genomic DNA Mini Kit (Thermo Fisher Scientific, Waltham, MA, U.S.A.). Polymerase chain reaction (PCR) solution was prepared using KAPA HiFi HotStart ReadyMix (Roche, Basel, Switzerland) as per the manufacturer's protocol, and amplification was performed in a Veriti thermal cycler (Thermo Fisher Scientific). AE-6932GXES Printgraph (ATTO, Tokyo, Japan) was used for the digital imaging of DNA fragments fractionated on agarose gels. Sequencing samples were prepared using the BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific) and a Veriti thermal cycler (Thermo Fisher Scientific) and analyzed by a 3500 genetic analyzer (Thermo Fisher Scientific).
To create dual single guide RNA (sgRNA) vectors, sgRNA expression cassettes on PX462-based vectors (strings of a U6 promoter, an inserted spacer sequence, and a sgRNA scaffold with a transcription stop signal) were PCR-amplified using a pair of primers shown in Supplementary Table S2. The resultant PCR products were digested with XbaI and inserted into the XbaI site of different PX462-based vectors in such a way that two sgRNA expression cassettes are placed in the same orientation.
Donor plasmids bearing wild-type PIGP homologous regions (Donor-PIGPex3 and Donor-PIGPex2) were constructed by PCR amplifications of the PIGP genomic region and the insertion of the PCR products into pBluescript II KS(+) (Agilent Technologies, Santa Clara, CA, U.S.A.), using restriction enzymes whose recognition sites were situated within PCR primers (Supplementary Table S2). The PIGP mutations shown in Supplementary Figures S1 and 2 were incorporated into the above-described wild-type donor plasmids (Donor-PIGPex3 and Donor-PIGPex2, respectively) to create donor plasmids bearing mutant PIGP homologous regions.

Creation of PIGP reporter clones
Supplementary Table S1 indicates the sequences of Cas9 nuclease target sites employed to create the initial heterozygous PIGP deletion in both cell lines (PIGP-nuclease-A and PIGP-nuclease-B), as well as those used to extend the deletion in HCT116 (TTC3-nuclease-C and Junction-nuclease-D).
Analytical PCRs probing the initial heterozygous PIGP deletion were performed with the following primer pairs: PIGPint2-F1 and PIGPint2-R2, PIGPdownstream-F1 and PIGPdownstream-R1, and PIGPint2-F1 and PIGPdownstream-R1. Analytical PCRs probing the creation of the extended heterozygous PIGP deletion in HCT116 were performed with the following primer pairs: TTC3int1-F and TTC3int1-R, PIGPint2-F1 and PIGPint2-R2, PIGPdownstream-F1 and PIGPdownstream-R1, and TTC3int1-R and downstream-R1. Two reverse primers were used to amplify the joined deletion ends because PIGP and TTC3 are situated in head-to-head configuration. The primer sequences are shown in Supplementary Table S2.
Southern blot analysis to confirm the heterozygous deletions was performed as previously described [13,22,23], except that the BglII restriction enzyme was used to digest gDNA. A probe was produced using a primer pair amplifying the region downstream from the last PIGP exon (Figure 1 and Supplementary Table S2).
To create the PIGP reporter clones, an inactivating mutation was introduced into the remaining PIGP allele in the PIGP-heterozygous clones by TPN-based targeted knock-in using the following Cas9 nickases: PIGP-nickase-1 and 3 for the DLD-1-derived clone, and PIGP-nickase-6 and 7 for the HCT116-derived clone ( Figure 2 and Supplementary  Table S1).

Figure 2. A flow chart depicting the construction of reporter clones
The flow chart is shown together with explanatory schemes for respective cell engineering steps. Heterozygous deletions were created exploiting the fact that two adjacent DSBs introduced by Cas9 nucleases on the same chromosome sometimes elicit a genomic deletion between DSBs and sometimes create two indels at the cleaved sites. In HCT116, a 10.7 kb-long heterozygous deletion was introduced by 2-round sequential deletions of 6.3-and 4.4-kb neighboring regions. Truncating mutations were incorporated into the genome by the TPN strategy using Cas9 nickases. Panels (A-E) noted in the schemes represent analytical PCRs whose results are shown in Figure 3A-E. Asterisks indicate PCR amplifications encompassing the knock-in sites, and the nucleotide sequences amplified by these PCRs are shown in Figure 4B.

PIGP correction assay
The PIGP correction assay (reversion of mutant PIGP allele to a wild-type allele) was initiated by transfecting PIGP reporter clones (1 × 10 5 cells) with a donor plasmid bearing a wild-type PIGP sequence along with a Cas9 nuclease or tandem paired Cas9 nickases shown in Supplementary Table S1. Donor-PIGPex3 and Donor-PIGPex2 were employed as donor plasmids for transfecting DLD-1-derived and HCT116-derived PIGP reporter clones, respectively. Transfected cells were incubated for three days to allow the production of functional PIGP protein from the corrected PIGP allele and subsequent GPI-anchor synthesis. Cells were then stained with FLAER and analyzed by FCM to determine the percentage of FLAER-positive cells.
To obtain FLAER-positive ratios, the percentages of FLAER-positive cells in the respective samples were divided by the transfection efficiency determined by transfecting the PIGP reporter clones with PX458.

Statistical analysis
EZR software [24] version 1.54 was used for the statistical analyses performed in the present study. The efficiencies of targeted knock-in achieved using dual sgRNA vectors and those using PX462-based expression vectors were compared in DLD-1-derived and HCT116-derived PIGP reporter clones, separately. Statistical analyses were conducted applying two-way repeated measures analysis of variance using 'vector' (1 μg each of PX462-based expression vectors, 1 μg of dual sgRNA vector, and 2 μg of dual sgRNA vector) and 'sgRNA pair' (two pairs) as two independent variables. After no interactions between these independent variables were found, a Bonferroni post hoc test was performed to detect significant differences among three 'vector' levels.

Results
Creation of a DLD-1-derived reporter clone with which to monitor the correction of a PIGP mutation GPI-anchor biosynthesis is governed by a molecular system comprised of approximately 20 proteins [25]. Many of these proteins are essential for the synthesis of GPI anchors and could therefore be exploited as platforms to monitor the efficiency of gene editing using a cellular phenotype as a surrogate marker, i.e. abrogation of one of these genes would result in the loss of GPI anchors from the cell surface. However, among the genes known to be required for GPI-anchor biosynthesis, PIGA is the only gene located on the X chromosome [15]. Therefore, to establish a reporter system to monitor gene editing events using a GPI-anchor biosynthesis gene other than PIGA as a platform, we planned to delete an allele of a GPI-anchor biosynthesis gene on an autosome so that the absence of GPI anchors on the cell surface determined by FCM analyses can correctly report the frequency of edits created on the other allele. We chose PIGP (chromosome 21), the product of which forms a complex with PIGA and plays an essential role in GPI-anchor biosynthesis [20], as the substrate of our genetic manipulation ( Figure 1). DLD-1, a near-diploid human cell line, was employed as the first platform cell line for the genetic manipulation.
To introduce a large deletion within a PIGP allele, we initially transfected DLD-1 cells with two plasmids each expressing Cas9 and a sgRNA (a Cas9 nuclease, collectively), which create double-strand breaks (DSBs) within PIGP intron 2 and downstream from the last PIGP exon, respectively ( Figure 2). Puromycin selection was then performed for 48 h to remove untransfected cells. At 14 days after transfection, the selected cells were stained with FLAER and subjected to the FCM-based sorting of FLAER-positive cells to remove cells bearing both PIGP alleles disrupted by Cas9 nucleases. Single-cell clones were next isolated from the sorted cell population and processed for PCR screening using three sets of primer pairs; the three PCRs amplified the joint portion of both deletion ends, a genomic region encompassing the Cas9 target site within PIGP intron 2, and that downstream from the last PIGP exon, respectively (Figures 2 and 3A,C,D). Cell clones indicating positivity in all three PCRs were identified, and PCR products derived from these clones were sequenced. The obtained reads verified the correct joining of deletion ends and the creation of small indels at the Cas9 nuclease target sites that supposedly occurred in the remaining PIGP allele. We also performed southern blot analysis to confirm the introduction of the programmed PIGP-heterozygous deletion in a representative clone ( Figure 4A). This clone is hereafter referred to as a DLD-1-derived PIGP-heterozygous clone.
Next, we sought to create a reporter clone that allows us to monitor the correction of a PIGP-inactivating mutation by introducing a truncating mutation into the remaining PIGP allele in the DLD-1-derived PIGP-heterozygous clone. To this end, we transfected the PIGP-heterozygous clone with two plasmids each expressing Cas9 (D10A) and a sgRNA (a Cas9 nickase, collectively), together with a donor plasmid carrying the PIGP genomic sequence harboring a truncating mutation at exon 3 ( Figure 2). This transfection should elicit targeted knock-in of the PIGP truncating mutation into the PIGP-heterozygous cell clone via TPN, a nick-based strategy for precision targeted knock-in [13]. We thus stained the transfected cells with FLAER and collected FLAER-negative cells by FCM-based cell sorting to obtain a cell population in which both PIGP alleles were inactivated by a deletion and a mutation, respectively. The sorted cells were subjected to single-cell cloning, and the knock-in site at PIGP exon 3 in the single-cell clones was sequenced ( Figure 4B). These procedures led to the identification of cell clones in which the programmed mutation had been incorporated into the intended genomic position. The absence of functional PIGP alleles in the identified clones was confirmed by cell staining with FLAER followed by FCM analysis. The cell clones were thereby established as DLD-1-derived PIGP reporter clones, which were to be employed for detecting targeted knock-in events.

Creation of a HCT116-derived reporter clone with which to monitor the correction of a PIGP mutation
To reduce biases in measuring the efficiency of targeted knock-in caused by potential cell line-specific and genomic site-specific factors, we sought to engineer another human cell line and create a second series of PIGP reporter clones bearing a mutation at a different portion of the PIGP gene. Given that a mutation on exon 3 was created in the DLD-1-derived PIGP reporter clone, we planned to introduce a distinct truncating mutation on exon 2 into another near-diploid human cell line, HCT116 (Figure 1). We initially performed the same experimental procedures as described above with the HCT116 cell line to introduce the identical heterozygous PIGP deletion with the DLD-1-derived reporter clone (Figure 2). The resulting HCT116-derived clone was next transfected with two Cas9 nucleases targeted to the junction of deletion ends and intron 1 of the TTC3 gene, respectively. TTC3 is a gene located next to PIGP with a head-to-head configuration, and exon 1s of both genes are located very close to each other (approximately 100 bp apart). If a Cas9 nuclease was designed at a genomic site between two exon 1s, the promoter of the remaining PIGP allele would likely be disrupted by indels created at the Cas9-cleaved site, which would compromise the functionality of PIGP reporter clones generated through subsequent engineering steps. We thus chose to place one of the Cas9 nucleases within TTC3 intron 1. This process would extend the original heterozygous deletion in the HCT116-derived clone; now, the deletion should cover the entire PIGP gene in addition to TTC3 exon 1 (Figures 1 and 2). After transfection with the two Cas9 nucleases, cells were selected with puromycin for 48 h, stained with FLAER at 10 days after transfection, and subjected to FCM-based sorting to collect FLAER-positive cells. Single-cell clones were isolated from the collected cells and then analyzed for their PIGP gene statuses by four different analytical PCRs: one encompassing the extended deletion and the others encompassing the Cas9 nuclease target sites located within TTC3 intron 1, within PIGP intron 2, and downstream from the last PIGP exon, respectively (Figures 2 and 3B-E). Cell clones positive for all four analytical PCRs were identified, and the existence of a functional PIGP gene in these clones was confirmed by cell staining with FLAER followed by FCM analysis. Southern blot analysis was performed to confirm the genomic structure at the PIGP locus in one of the clones ( Figure 4A). This clone is hereafter referred to as a HCT116-derived PIGP-heterozygous clone.
We next employed the HCT116-derived PIGP-heterozygous clone to create a reporter clone allowing us to monitor the efficiency of targeted knock-in by introducing a truncating mutation into exon 2 within the remaining PIGP allele. We chose to accomplish this by TPN-based targeted knock-in and transfected the HCT116-derived PIGP-heterozygous clone with two Cas9 nickases targeted to the region surrounding the knock-in site, along with a donor plasmid carrying a PIGP sequence harboring a truncating mutation on exon 2 ( Figure 2). The cells were then stained with FLAER and subjected to FCM-based sorting to collect the FLAER-negative cells. Single-cell clones were isolated from the collected cells and processed for Sanger sequencing to verify the targeted incorporation of the programmed PIGP mutation in these clones ( Figure 4B). After the confirmation of FLAER-negativity by FCM analysis, these cell clones were established as HCT116-derived PIGP reporter clones for monitoring targeted knock-in events.

Measuring the efficiency of targeted knock-in using the PIGP reporter clones
We next compared the efficiency of targeted knock-in achieved via the TPN strategy versus a conventional Cas9 nuclease-based strategy using the PIGP reporter clones as platforms of the assay. For this purpose, we initially transfected the DLD-1-derived and HCT116-derived PIGP reporter clones with donor plasmids carrying the relevant portions of the wild-type PIGP sequence (Donor-PIGPex3 and Donor-PIGPex2, respectively) together with tandem paired Cas9 nickases or a Cas9 nuclease ( Figure 5). Cells were then stained with FLAER and analyzed by FCM. These assays, namely PIGP correction assays, demonstrated that half of the Cas9 nickase pairs used in the TPN strategy served to achieve efficiencies of targeted knock-in largely comparable to that achieved by Cas9 nucleases (Figure 6). These data are consistent with our previous findings obtained via PIGA correction assays [13], suggesting that the PIGP correction assay is useful as a second assay measuring the efficiency of targeted knock-in similar to the PIGA correction assay.

The use of dual sgRNA vectors enhances the efficiency of targeted knock-in via TPN
We lastly sought to harness the PIGP correction assay to improve the targeted knock-in procedure. We and others have demonstrated the utility of TPN to conduct precise and relatively efficient targeted knock-in [13,26,27]. Although two sgRNAs used for TPN in these studies were transfected into cells using two plasmids each expressing Cas9 (D10A) and a sgRNA, we postulated that the use of all-in-one vectors expressing two sgRNAs and Cas9 (D10A) from a single plasmid would allow the efficient simultaneous expression of all these genome editing tools within recipient cells and would thereby help achieve higher efficiency of targeted knock-in via TPN. We therefore created plasmids harboring three gene cassettes expressing a Cas9 (D10A) and two sgRNAs targeted near the PIGP knock-in site, respectively ( Figure 7A). Similar all-in-one vectors expressing up to seven sgRNAs and a Cas9 nuclease were previously created and employed for multiplex DSB-mediated genome engineering in human cells [28]. The plasmids that we created, named dual sgRNA vectors, were transfected into DLD-1-derived and HCT116-derived reporter clones; the transfected clones were then subjected to PIGP correction assays. As a result, the average efficiencies of the targeted knock-in achieved using dual sgRNA vectors were 1.1-to 1.9-fold higher than those obtained with two PX462-based expression vectors in all comparisons performed using a total of four tandem Cas9 nickase pairs with DLD-1-derived and HCT116-derived reporter clones ( Figure 7B). However, the differences reached statistical significance only in half of the dataset derived from the DLD-1-derived reporter clone. Overall, the data obtained in these assays demonstrated that the use of a dual sgRNA vector slightly enhances the efficiency of targeted knock-in achieved via TPN compared with the use of two PX462-based expression vectors.

Discussion
In the present study, we created reporter clones in which two PIGP alleles were inactivated by a genomic deletion and a truncating mutation, respectively. The reporter clones were employed to perform the PIGP correction assay that monitors the correction of PIGP mutations achieved via targeted knock-in. Although a previously established PIGA reporter clone used in the PIGA correction assay derives from a single cell line (HCT116) [13], the PIGP reporter clones derive from two cell lines (DLD-1 and HCT116), and these clones have truncating mutations on different PIGP exons. Therefore, the use of the PIGP correction assay will help reduce concerns for genomic site-specific and cell line-specific biases that potentially affect the outcomes of the assays. In addition, the joint use of PIGA-and PIGP-based assays will further provide variety in genomic loci employed as assay platforms. Similar to the PIGA correction assay, the PIGP correction assay monitors the efficiency of targeted knock-in harnessing a GPI-anchor biosynthesis gene as a reporter. In both of these assays, GPI anchors protruding from the cell membrane are detected as a surrogate marker for the presence of a functional reporter gene. FCM-based quantification of GPI-anchor-positive cells has been well established as a sensitive, simple, and relatively inexpensive methodology to determine the efficiency of targeted knock-in in the PIGA correction assay, and the PIGP correction assay shares the same benefits.
We initially created large heterozygous deletions within and over the PIGP gene to create the PIGP reporter clones. This initial deletion step enables the use of autosomal genes, such as the PIGP gene, to quantify the efficiency of targeted knock-in. Other GPI-anchor biosynthesis genes, including PIGL [17], will also be useful as platform genes for FCM-based detection of targeted knock-in events, once a heterozygous deletion is initially created. In addition, although the cell lines employed in the present study (DLD-1 and HCT116) are of male origin, it should also be possible to use female cell lines for an assay monitoring gene correction events by first introducing the heterozygous deletion of a reporter gene. Thus, the initial deletion step will be able to further increase the choice of  reporter genes and platform cell lines in monitoring the efficiency of targeted knock-in. In summary, our current study demonstrated that the PIGP correction assay is useful for measuring the efficiency of targeted knock-in and, together with the PIGA correction assay, will help further improve the efficiency of CRISPR/Cas-assisted targeted knock-in.

Conclusion
Collectively, the present study has served to extend the variety of reporter genes and platform cell lines harnessed to measure the efficiency of targeted knock-in and will thereby help minimize the biases in the observed efficiency of targeted knock-in caused by locus-specific and cell line-specific factors.

Data Availability
Data and materials will be made available from the corresponding author upon reasonable request.