br Results br Discussion Here we demonstrate

Results
Discussion Here, we demonstrate the generation of hESC reporter lines without the use of drug selection for both active and silent genes through the use of the iCRISPR system. Selection-free gene targeting eliminates the need for removal of drug-resistance cassette after identification of correctly targeted clones, and mini-vector donors with short homology arms (∼500–1,000 bp each) are convenient to make. Thus, this method significantly reduces the time and effort required for establishing hESC reporter lines. Additionally, conventional gene-targeting strategies typically use the Cre-loxP strategy to remove drug-resistance cassettes, which leaves behind a 34-bp loxP “scar” in the endogenous locus. Although not an issue in most cases, this residual sequence could interfere with the kinesin spindle protein of the targeted gene in some situations (Meier et al., 2010). In comparison, our strategy eliminates the need for the selection cassette and thereby minimizes the alteration of the endogenous locus. Although we focused on creating promoter-fusion reporters, the same knockin approach can be readily applied to making protein-fusion reporters for visualizing protein subcellular localization, precisely deleting or replacing specific genomic sequences, and introducing or correcting disease-associated mutations. Previously, we and others have failed to target the PDX1 locus using traditional targeting approaches (Z.Z. and D.H., data not shown; E. Stanley, personal communication), yet the absolute targeting efficiencies using the selection-free method were comparable between the PDX1 and OCT4 loci. It is known that gene-targeting efficiencies can vary significantly depending on the target locus, though the exact reason is unclear. Traditional gene targeting relies on drug selection; thus, the relative targeting efficiencies after drug selection depend, at least in part, on the expression of the drug-resistance gene from the targeted locus. Because the expression of drug-resistance gene may differ significantly between expressed and silent loci, the relative targeting efficiency after drug selection for a lineage-specific gene may appear much lower compared with a pluripotency gene. For certain loci, the drug-resistance gene may be expressed at such low levels that hinder the identification of a correctly targeted clone using the drug-selection method (Rostovskaya et al., 2012). Because our targeting approach obviates drug selection, it may overcome such bias and facilitate the generation of reporter alleles for genes that were previously difficult to target. One may further use this ability to measure absolute targeting frequencies to compare HDR efficiencies across different genomic contexts. There have been concerns about potential off-target mutagenesis with the CRISPR/Cas system (Cho et al., 2014; Fu et al., 2013; Hsu et al., 2013; Mali et al., 2013b; Pattanayak et al., 2013). Our analysis so far did not reveal any off-target mutations at sites without perfect complementarity with the CRISPR target sequence. However, we cannot exclude the possibility of off-target mutations elsewhere in the genome, and a thorough analysis may be necessary before the reporter lines are used in future studies. The CRISPR/Cas system is continuously improved with the development of better algorithms for CRISPR design and off-target prediction. It is reassuring that a recent high-coverage whole-genome sequencing study failed to detect significant incidence of off-target mutations in CRISPR-targeted hPSC lines (Veres et al., 2014). On the other hand, we noticed that some correctly targeted clones carried mutations in the non-targeted allele, though the targeting efficiency is sufficiently high that one could simply discard the minority of clones carrying mutations. One may also target intronic regions with low-sequence conservations to further mitigate any concerns associated with Indel mutations in the non-targeted allele.