54 128
designed gRNAs against a total of nine genomic loci. The loci
included core pluripotency transcription factors (
OCT4
,
NANOG
,
and
SOX2
), kinases (
ROCK1
and
GSK3-
b
), a cardiac mesoderm-
transcription factor (
MESP1
), and cardiac disease-associated
genes (
BAG3
,
MYBPC3
, and
HERG
). Except for
MESP1
(ex-
pressed only transiently in cardiacmesodermcells) and
MYBPC3
(expressed only in cardiomyocytes), all other genes are ex-
pressed in iPSCs at different levels. We generated populations
of CRISPRi iPSCs containing stably integrated gRNA-expression
constructs. We then cultured these stable polyclones or clonal
populations either with or without doxycycline for at least 7 days.
Three to five gRNAs were designed to target near the TSS of
each gene and initially were tested individually in polyclonal
populations. Approximately half of the tested gRNAs were active
in polyclonal populations with a silencing activity of over 70%
(Figure S4A). We did not observe a difference in the knockdown
efficiency between gRNAs targeting either the template or non-
template strands (Figures 3A, S4A, and S4B). The most active
gRNA-containing polyclonal line was further passaged and
subcloned for more detailed knockdown analysis. Using the
most active gRNA, we achieved 90%–99% knockdown of the
gene of interest in a selected population of iPSCs after doxycy-
cline treatment (Figure 3B). As expected, when we subcloned
polyclonal populations via single-cell cloning, we observed a
higher percentage of transcriptional knockdown. With immuno-
fluorescence analysis we found that iPSC clones expressing
gRNAs against
OCT4
,
NANOG
,
SOX2
, and
BAG3
showed com-
plete loss of target protein expression 7 days after doxycycline in-
duction. In cells expressing gRNAs against the core pluripotency
transcription factors
OCT4
,
NANOG
, and
SOX2
, we observed
clear morphological changes and a loss of pluripotency after
doxycycline induction; however, loss of a non-pluripotency
gene (
BAG3
) did not affect pluripotent morphology (Figure 3C).
Using the Gen1 CRISPRi knockin vector, we targeted non-
iPSCs with a different genetic background to determine how
broadly this technology can be applied to other cell types. A T-
lymphocyte (CEM) CRISPRi line was generated, as described in
Experimental Procedures. Similar to the iPSC experiments,
gRNAs were introduced to the stable CEM CRISPRi cell line,
and cells cultured in either the presence or absence of doxycy-
cline for 10 days. Three gRNAs were tested to knock down
CD4
in CEM-CRISPRi cells, and all showed greater than 70%
knockdown efficiency in polyclonal populations (Figure S4B).
The most active gRNA-containing polyclone was subcloned,
and three independent clonal lines were isolated and assayed
for knockdown, where greater than 95% knockdown efficiency
was observed (Figure S4C). These results clearly demonstrate
the doxycycline-inducible CRISPRi vector system is highly versa-
tile and transportable to other cell lines and shows high efficiency
of knockdown across a range of cell types and genetic loci.
CRISPRi Knockdown Is Reversible and Tunable and Can
Be Allele Specific
GCaMP is a calcium-sensitive modified GFP and, thus, can be
used as a fluorescent reporter under steady-state levels of cyto-
plasmic Ca
2+
(Apa´ ti et al., 2013). Using GCaMP (driven off the
strong constitutive promoter, CAG), we monitored the green-
fluorescence signal in iPSCs to determine if we could knock
down GCaMP and then reverse its expression by removing
doxycycline from the culture. We found that adding doxycycline
for 7 days knocked down GCaMP expression by 98%, which
was completely restored after removing doxycycline for
14 days (Figure 4A). Similarly, we targeted the
BAG3
endoge-
nous locus and achieved efficient transcript knockdown post-
doxycycline treatment.
BAG3
expression was fully restored after
doxycycline withdrawal (Figure 4B). These findings indicate that
CRISPRi knockdown is fully reversible in iPSCs.
To determine if we could achieve variable levels of knockdown
with different gRNA sequences, we tested two additional gRNAs
targeting GCaMP (g+24 and g+91) (Figure 4C). These gRNAs
knocked down GCaMP expression by only 30% and 50%,
as measured by flow cytometry (Figures 4D and 4E). Therefore,
by changing the location of the gRNA-binding site, we can
tune the level of knockdown when trying to mimic haploinsuffi-
ciency or reduced protein levels (rather than complete loss of
function). In addition, we tested whether the knockdown level
is tunable by titrating the doxycycline concentration. Careful
titration of the doxycycline concentration enabled homogenous
modulation of GCaMP expression (Figure S5).
We next sought to further test the tunability of knockdown with
CRISPRi. We determined if we could use single-nucleotide poly-
morphisms (SNPs) to specifically target one allele for knockdown
to achieve a heterozygous-like state. In our CRISPRi iPSCs,
there is a SNP near the TSS of
OCT4
. Thus, we designed a
gRNA in which the heterozygous SNP is located in the PAM
sequence (AGG versus AGA). Because an ‘‘NGG’’ sequence is
required for dCas9 to target DNA, we could selectively target
only one
OCT4
allele (Figure 4F). After doxycycline induction,
we found that the iPSC population carrying the SNP-specific
OCT4
gRNA (
OCT4
g 4) remained OCT4 positive ( 97%) by
flow cytometry analysis. However, the median intensity of
OCT4 staining was reduced by 40%after 7 days of doxycycline
treatment, implying that OCT4 expression was homogeneously
reduced in all cells and not just a fraction of them (Figures 4G
and 4H). We confirmed this finding with immunocytochemistry
and TaqMan qPCR (data not shown).
CRISPRi Knockdown Is Highly Specific
To assess the specificity of CRISPRi targeting, we designed a
gRNA that targets the GCaMP transgene, since its silencing
should have few downstream transcriptional and cellular conse-
quences. Indeed, expression of the GCaMP transcript was over
30-fold lower in the doxycycline-treated condition, while few
other endogenous transcripts changed expression level with
the exception of
VIM
as previously discussed (Figure 5A).
CRISPRi to Promote iPSC Differentiation
To show that our system can release iPSCs from their pluripotent
state to promote differentiation, we tested the efficiency of
CRISPRi in knocking down core pluripotency transcription factors
(
OCT4
,
SOX2
, and
NANOG
) without adding small molecules or
cytokines to the mTeSR media. We targeted gRNA against these
genes and performed a time-course analysis of a selected num-
ber of transcripts by TaqMan qPCR (Figure 5B). We found that
knocking down these target transcripts caused cell differentiation,
as indicated by morphological changes and transient expression
of the lineage-specific transcript
T
(mesoderm marker), and
expression of
PAX6
(neuronal progenitor marker). After 3 days
546
Cell Stem Cell
18
, 541–553, April 7, 2016
ª
2016 Elsevier Inc.