An enhanced CRISPR repressor for targeted mammalian gene regulation

Abstract

The RNA-guided endonuclease Cas9 can be converted into a programmable transcriptional repressor, but inefficiencies in target-gene silencing have limited its utility. Here we describe an improved Cas9 repressor based on the C-terminal fusion of a rationally designed bipartite repressor domain, KRAB-MeCP2, to nuclease-dead Cas9. We demonstrate the system's superiority in silencing coding and noncoding genes, simultaneously repressing a series of target genes, improving the results of single and dual guide RNA library screens, and enabling new architectures of synthetic genetic circuits.

Figure 1

Repression of endogenous genes using dCas9-KRAB-MeCP2. (a) Schematic of dCAS9-KRAB and dCas9-KRAB-MeCP2 repressors. NLS=nuclear localization signal. (b) RNA expression of targeted single gene. n=2 biologically independent samples (cell cultures). (c) RNA expression during three separate multiplex repression studies. In each study, four different genes were targeted simultaneously. Two non-coding (nc) genes, XIST and TERC, were studied. n=2 biologically independent samples (cell cultures).

Figure 2

dCas9-KRAB-MeCP paired with gRNA at various positions. (a) An array of gRNAs was design to target 1-kb upstream to 1-kb downstream of the TSS for CANX. Shown is RNA expression of CANX when different gRNA was used. T=template strand, NT=non-template strand. n=2 biologically independent samples (cell cultures). (b) Shown is RNA expression of SYVN1 when individual or combinations of different gRNAs were used. n=2 biologically independent samples (cell cultures). (c) RNA expression of the indicated target genes using one or two gRNAs. n=2 biologically independent samples (cell cultures).

Figure 3

dCas9-KRAB-MeCP2-mediated repression is highly specific in human cells. RNA-seq analyses of HEK293T cells transfected with a gRNA targeting CXCR4 along with dCas9, dCas9-KRAB or dCas9-KRAB-MeCP2 repressors. Data are normalized and log₂-transformed counts per million (CPM) values are plotted for each repressor (y-axis) versus that of a negative control transfected with gRNA alone (x-axis). Pearson’s and Spearman’s correlation coefficients are provided for each repressor. n=2 biologically independent samples (cell cultures).

Figure 4

dCas9-KRAB-MeCP2 outperforms previous tools in screens of gene essentiality. (a) Shown are log₂ odd ratios of all sgRNA constructs as compared to the HAP1 wild-type cells at days 14. sgRNAs targeting essential genes are marked in orange and sgRNAs targeting non-essential genes are marked in blue. A similar experiment was repeated in (b) SH-SY5Y and (c) HEK293T cells. Shown are log₂ odd ratios of all sgRNA constructs as compared to the respective wild-type cells at days 14.

Figure 5

dCas9-KRAB-MeCP2 improves genetic interaction mapping. (a) A density plot showing negative and positive selection pressure against gRNA pairs over time. (b) Shown is the hierarchical clustered heatmap of genetic interactions for dCas9-KRAB and dCas9-KRAB-MeCP2. Only the screen using dCas9-KRAB-MeCP2 showed a discernible clustering structure.

Figure 6

Superiority of dCas9-Krab-MeCP2 in regulating complex synthetic circuits. (a) When expressed from a doxycycline inducible Pol II promoter and edited by Csy4, gRNA showed improved repression of EYFP when co-expressed with dCas9-KRAB-MeCP2, relative to dCas9 or dCas9-KRAB. n=4 biologically independent samples (cell cultures). (b) Inducible Pol II-expressed gRNA edited by Csy4 and complexed with dCas9-KRAB-MeCP2 showed improved performance in a two-tier repressor cascade. n=4 biologically independent samples (cell cultures). (c) With full circuit, dCas9-KRAB-MeCP2 decreases CXCR4 protein level. In absence of layer 1, dCas9-KRAB-MeCP2 mediates derepression of CXCR4. In absence of layers 1 and 2, the repressor surpasses dCas9 and dCas9-KRAB in repressing CXCR4 levels. n=3 biologically independent samples (cell cultures). For a-c, data are presented as mean ± s.e.m. One-sided Student T-test was performed for all statistical comparison. # p< 0.05 vs. unrepressed or TALER only control, ¥ p< 0.05 v.s. dCas9, and *p < 0.05 v.s. dCas9-KRAB.