50 128
more efficient transcriptional interference (Gilbert et al., 2013,
2014; Kearns et al., 2014). To further this effort, we aimed to
use CRISPRi technology to efficiently repress genes to study
early differentiation and model disease with human induced
pluripotent stem cells (iPSCs) (Takahashi et al., 2007).
iPSCs are well suited to study early embryonic development
and disease since they can produce different functional cell
types in vitro (Sterneckert et al., 2014). Early embryonic develop-
ment consists of a series of accurately timed events that affect
gene activation and repression (Bolouri and Davidson, 2003).
Therefore, precisely regulating the timing and dosage of
transcription factors critically affects embryonic development
(McFadden et al., 2005; Takeuchi et al., 2011), and dysregulation
in the timing and dosage of transcripts can lead to disease devel-
opment (Theodoris et al., 2015). In this study, we compared
inducible CRISPR systems for gene knockout (using Cas9) or
knockdown (using dCas9-KRAB) to enable temporal control of
loss-of-function phenotypes in iPSCs and differentiated cell
types.
RESULTS
Generation of CRISPRi and CRISPRn iPSC Lines
For loss-of-function studies, we independently derived multiple
stable CRISPRi and CRISPRn human iPSC clones in two genetic
backgrounds: wild-type B (WTB) and wild-type C (WTC)
(Miyaoka et al., 2014). In separate targeting events, the CRISPRi
and CRISPRn constructs (see Supplemental Experimental Pro-
cedures) were integrated into the AAVS1 locus of WTB and
WTC iPSCs using a TALEN-assisted gene-trap approach (Fig-
ures 1A, 1B, and S1). Transgenes integrated at the AAVS1 locus
remain transcriptionally active in both iPSCs and differentiated
cell types (Hockemeyer et al., 2011; Lombardo et al., 2011).
We generated several different versions of the CRISPRi system
that are either inducible or constitutive; the inducible CRISPRi
(Gen1 and Gen2) clones express dCas9-KRAB (KRAB domain
fused at the N terminus) from the inducible TetO promoter, while
the constitutive CRISPRi clones (Gen3) express dCas9-KRAB
under the constitutively active CAG promoter. The CRISPRn
(Gen1) clones express Cas9 under the inducible TetO promoter
(Figure S1).
The average efficiency of forming stable clones was 350 col-
onies per million iPSCs transfected with AAVS1 TALENs and
donor plasmid (data not shown). From each condition, multiple
independent colonies were isolated and expanded. A subset of
the stable colonies from each targeting vector was screened
using junction PCR. Two putative colonies from each targeting
event were further characterized by stably introducing an
OCT4
-specific gRNA and performing knockdown or knockout
assays with immunofluorescence and western blot analysis. All
putative CRISPRi clones containing an
OCT4
-specific gRNA
showed efficient knockdown (>95%) of OCT4 in bulk popula-
tions, while a significant fraction of the CRISPRn cells remained
OCT4 positive ( 30%–40%) in bulk populations containing
OCT4
-specific gRNA (Figure S1). One clone each from CRISPRi
and CRISPRn (Gen1 lines in the WTC genetic background) were
subsequently used as lead clones for further studies.
To enable non-invasive and high-throughput phenotypic anal-
ysis in iPSC-derived cardiomyocytes (iPS-CMs), we performed
a second targeting event that introduced the green fluorescent
calcium-modulated protein 6 fast type (GCaMP) calcium sensor
(Chen et al., 2013) into the other AAVS1 locus of the CRISPRi
cell line. The GCaMP transgene is driven off the strong, constitu-
tive CAG promoter (Figure S1). We found that CRISPRi iPSCs
could differentiate into iPS-CMs, so that we could measure cal-
cium transients based on theGCaMP-fluorescent intensity (Movie
S1) (Huebsch et al., 2015). Lead CRISPRi and CRISPRn iPSCs
were karyotypically normal (Figures S2A and S2B) and expressed
pluripotency markers, as expected (Figures S2C and S2D).
RNA-sequencing (RNA-seq) analysis indicated that expres-
sion of dCas9-KRAB or Cas9 was undetectable in the absence
of doxycycline, and addition of doxycycline without any gRNA
resulted in robust selective induction of dCas9-KRAB or Cas9,
while the rest of the transcriptome remained virtually unchanged
(Figures S2E and S2F). Furthermore, the RNA-seq data suggest
that the addition of the KRAB domain has no detectable off-
target effects when compared to expression of Cas9 alone.
Remarkably the one gene that appeared to be upregulated
upon doxycycline induction (without gRNA) was the same gene
(Vimentin;
VIM
) for both CRISPRi and CRISPRn cells (Figures
S2E and S2F). Since the same gene is upregulated for CRISPRi
and CRISPRn cells, we suspect it may represent an off-target
activity of the doxycycline-induced transactivator. Importantly,
our experiments suggest that the expression of dCas9-KRAB
alone has no additional effects on gene expression.
We also expressed dCas9-KRAB and Cas9 by continuously
culturing CRISPRi and CRISPRn lines with doxycycline for
3 weeks (four passages). With this long-term treatment, we
observed no cytotoxicity, decrease in proliferation, or change
in morphology in these cells (Figures S2G and S2H). Using a
droplet digital PCR (ddPCR)-based copy-number assay, we
measured the number of integration events (Figure S2I). We
further validated on-target integration sites on the lead CRISPRi
and CRISPRn clones with junction PCR (Figure S2J) and verified
their sequences (data not shown).
To further ensure there was no leaky expression of the single
doxycycline-inducible vector, we measured the protein levels
of dCas9-KRAB and Cas9 in iPSCs. With immunostaining, flow
cytometry and western blots did not detect dCas9-KRAB or
Cas9 protein without doxycycline in either CRISPRi or CRISPRn
iPSCs, indicating that the TetO promoter has high fidelity in the
AAVS1 locus. After doxycycline treatment, all cells in the
CRISPRi and CRISPRn lines expressed dCas9-KRAB or Cas9
within 48 hr, respectively (Figures 1C–1H). dCas9-KRAB and
Cas9 were expressed at similar levels after induction, and both
proteins rapidly degraded after removing doxycycline (Figures
1F, 1H, and S2K). These data showed that dCas9-KRAB and
Cas9 expression could be tightly regulated with the TetO pro-
moter, which would support studies that rely on precisely timing
gene knockdown or knockout.
Comparison of Loss of Function between CRISPRi and
CRISPRn
To compare CRISPRi and CRISPRn for loss-of-function studies,
we designed a gRNA that targets the first exon of
NANOG
,
a transcription factor necessary for maintaining the pluripotency
network. We selected
NANOG
as our first target gene because
its deficiency is sufficient to give an immediate readout, as
542
Cell Stem Cell
18
, 541–553, April 7, 2016
ª
2016 Elsevier Inc.