CRISPR Gene Editing in Stem Cells

displayed a rapid, uniform, and efficient transcriptional knock-

down. This result was also validated across multiple endoge-

nous loci in iPSCs, cardiac progenitors, and iPS-CMs. By

contrast, using CRISPRn, we found that while all cells harbored

the gRNA-expression vector and had continuous expression of

Cas9, they did not all display complete loss-of-function pheno-

types. Indeed, up to one-third of the cells maintained expression

of the target gene. When we sequenced the target alleles, we

found that of the mutated alleles, over one-third had in-frame

INDELs, potentially resulting in a hypomorphic protein encoded

by a gene that is now resistant to further Cas9 cutting using the

target gRNA. Statistically, we expect that one-third of the

INDELs generated by double-strand breaks induced by Cas9

through the non-homologous end-joining pathway would

produce in-frame mutations. This effect could cause partial

loss-of-function or gain-of-function phenotypes. Additionally,

the location and size of the in-frame INDEL might not change

the function of the mutated protein compared with the wild-

type protein (Boettcher and McManus, 2015; Shi et al., 2015;

Sung et al., 2013).

CRISPRi gRNAs were only effective at promoter regions close

to the TSS, which may reduce the likelihood of off-target effects

by transcriptional interference elsewhere in the genome. Indeed,

RNA-seq analysis showed that the knockdown of GCaMP was

highly specific. Furthermore, expression of dCas9-KRAB did

not cause significant off-target transcriptional changes as

compared to Cas9 expression alone. Although CRISPRi is highly

effective, there are cases when other genetic tools such as

CRISPRn, TALENs, and RNAi may have advantages. For

instance, we and others (Gilbert et al., 2014) have shown that

CRISPRi gRNAs are only effective near the TSS, which restricts

the efficiency of transcript for genes that have poorly defined or

multiple TSSs. CRISPRn and TALENs can be effective at any

exon as long as the genomic region is accessible (Doench

et al., 2014; Kim et al., 2013b). Additionally, RNAi can target

any constitutive portion of the mRNA and has already been

approved for human therapy (Davidson and McCray, 2011;

Haussecker, 2012); however, RNAi has been shown to have

many off-target effects (Jackson et al., 2003; Kim et al., 2013b;

Krueger et al., 2007).

We also demonstrated the feasibility of allele-specific interfer-

ence and the tunable nature of CRISPRi-based knockdown,

which can be used to study the dose-dependent effects of a

gene involved in development and disease. The dosage of tran-

scription factors plays a significant role during development and

organogenesis (McFadden et al., 2005; Takeuchi et al., 2011).

In addition, many human diseases result from haploinsufficiency

in which a mutation in a single copy of a gene produces the

disease phenotype (Armanios et al., 2005; Marston et al., 2012;

Minami et al., 2014; Theodoris et al., 2015). Therefore, to study

the dose-dependent effects of transcription factors in develop-

ment and disease, CRISPRi can be used to homogeneously

tune the level of repression in cells by either choosing the rele-

vant gRNA sequences or empirically titrating the levels of doxy-

cycline to achieve the desired knockdown level. Alternatively,

introducing a single point mutation at different positions in the

gRNA sequence (which leads to mismatches between the

RNA-DNA homology sequence) can be used to tune CRISPRi

knockdown activity (Gilbert et al., 2014). Finally, CRISPRi knock-

down was reversible in iPSCs upon doxycycline withdrawal,

which would support studies involving transient knockdown of

transcripts within a specific window during cell differentiation.

Our studies with CRISPRi in iPSCs showed that knocking

down transcripts involved in maintaining pluripotency is highly

efficient and rapidly causes a complete loss of pluripotent

morphology, followed by cell differentiation in all cells expressing

the appropriate gRNA. We also used this approach to knock

down the

HERG

potassium channel to mimic an LQT2-type

phenotype in iPS-CMs. We found that the inducible TetO pro-

moter is partially silenced during the cardiac differentiation

process, which has been reported to be due to methylation at

CpG dinucleotides (Oyer et al., 2009). This silencing is indepen-

dent of integration at the AAVS1 locus, as CAG-driven trans-

genes integrated at the AAVS1 locus remain active after differen-

tiation. To avoid the effects of promoter silencing, we initiated

transcript knockdown in the iPSC state or progenitor cells (day

5 of differentiation), where the vast majority of the cells respond

to doxycycline. This strategy has proved highly effective at trans-

gene knockdown in cardiac progenitors and iPS-CMs. To

circumvent issues with silencing in future studies, we generated

a non-inducible CRISPRi iPSC line (Gen3; in which dCas9-KRAB

is driven off the CAG promoter), and the knockdown can be initi-

ated upon introduction of gRNA. With this cell line, we expect to

achieve highly efficient knockdown in differentiated cell types,

such as iPS-CMs.

Several groups have used the CRISPR/Cas9 system for loss-

of-function genetic screens in human cells (Shalem et al., 2014;

Wang et al., 2014). Furthermore, some groups have used

genome-scale screens with CRISPRi and CRISPR activation

(CRISPRa) to identify known and novel genes that control cell

growth and sensitivity to cholera-diphtheria toxin (Gilbert et al.,

2014). In this study, we present our CRISPRi iPSC lines as

suitable model systems for performing screens to identify novel

transcripts of pluripotency, drug resistance, and cell survival at

the pluripotent stem cell stage. With genome-scale screens,

we can identify factors that improve cell-specific differentiation

into functional cell types that have been traditionally hard to

obtain, and we can more rapidly generate mature functional

cell types that better mimic in vivo cell counterparts. In addition,

with CRISPRi, we can repress putative disease-associated

genes in a medium- to high-throughput manner to unravel

the molecular mechanisms underlying human disease in vitro.

Finally, we can build on the current power of CRISPRi for devel-

opmental screens by using an orthogonal dCas9-effector sys-

tem for gene activation via CRISPRa, which can synergistically

modulate gene knockdown and activation and direct cell fate

toward a particular lineage.

EXPERIMENTAL PROCEDURES

iPSC Culture

WTB and WTC iPSCs and derivative lines were maintained under feeder-

free conditions on growth factor-reduced Matrigel (BD Biosciences) and

fed daily with mTeSR medium (STEMCELL Technologies) (Ludwig et al.,

2006). Accutase (STEMCELL Technologies) was used to enzymatically

dissociate iPSCs into single cells. To promote cell survival during enzymatic

passaging, cells were passaged with the p160-Rho-associated coiled-coil

kinase (ROCK) inhibitor Y-27632 (10

M; Selleckchem) (Watanabe et al.,

2007). iPSCs were frozen in 90% fetal bovine serum (HyClone) and 10%

550

Cell Stem Cell

, 541–553, April 7, 2016

2016 Elsevier Inc.