A CRISPR/Cas9-based system for reprogramming cell lineage specification

Abstract

Gene activation by the CRISPR/Cas9 system has the potential to enable new approaches to science and medicine, but the technology must be enhanced to robustly control cell behavior. We show that the fusion of two transactivation domains to Cas9 dramatically enhances gene activation to a level that is necessary to reprogram cell phenotype. Targeted activation of the endogenous Myod1 gene locus with this system led to stable and sustained reprogramming of mouse embryonic fibroblasts into skeletal myocytes. The levels of myogenic marker expression obtained by the activation of endogenous Myod1 gene were comparable to that achieved by overexpression of lentivirally delivered MYOD1 transcription factor.

Graphical abstract

Figure 1

Engineering an RNA-Guided, Nuclease-Inactive ^VP64dCas9-BFP^VP64—VdC9BV—Fusion Protein to Enable Robust Transactivation of the Endogenous Myod1 Gene (A) Illustration depicting the reprogramming of MEFs to SkMs initiated by the lentiviral transduction of MEFs with the gRNA, reverse tetracycline transactivator (M2rtTA), and a doxycycline-inducible VdC9BV. (B) Illustration of gRNA-guided transactivation by the VdC9BV fusion protein. (C) The lentiviral constructs utilized to activate endogenous genes. (D) Confocal images of live HEK293T cells transfected with VdC9BV or dC9B, H2BmCherry fluorescent protein (nuclear-localized), and GFP (predominantly cytoplasmic) showing that inclusion of an additional NLS in VdC9BV (3 NLS) enables better nuclear localization than the 2 NLS dCas9-BFP (dC9B) parent construct. The scale bar represents 20 μm. (E) Nucleus-to-cytoplasm (N:C) ratio of BFP fluorescence intensity quantified from confocal images of live transfected HEK293T cells. The analysis shows that the 3 NLS form of dCas9 (VdC9BV) localizes better to the nucleus than the original 2 NLS dCas9 (dC9B) form (n = 20 cells, error bars represent SEM). (F) Lentiviral vector enables robust delivery of nuclear-localized VdC9BV. dC9B (2 NLS) transfected into C3H10T1/2 cells shows lack of nuclear localization. VdC9BV (3 NLS) enables better nuclear localization. However, few cells expressed VdC9BV when transfected into the cells. Transduction of C3H10T1/2 with the lentiviral form of VdC9BV enables higher efficiency of delivery of the construct, as evidenced by BFP fluorescence imaging of live cells and immunofluorescence staining of the FLAG epitope in fixed cells. The scale bar represents 100 μm. (G) Lentiviral delivery of VdC9BV is more efficient than delivery by transfection. By flow cytometry analysis, ∼50% of the transduced cells were found to be BFP positive compared with ∼5% by transfection (two-tailed unpaired t test, p ≤ 0.0001, n = 3 biological replicates; error bars represent SEM). (H) qRT-PCR evaluation of relative Myod1 mRNA expression in C3H10T1/2 on day 6 postinduction of VdC9BV expression in the presence of various combinations of three gRNA molecules (p ≤ 0.001, one-way ANOVA, Dunnett’s post hoc comparing all the groups to the [−] gRNA control group [M2rtTA + VdC9BV + Empty FUW], n = 3 biological replicates, error bars represent SEM).

Figure 2

CRISPR/^VP64dCas9-BFP^VP64 Transactivates the Endogenous Myod1 Gene and Reprograms C3H10T1/2 Cells and MEFs into SkMs (A) ^VP64dCas9-BFP^VP64 (VdC9BV), in the presence of three gRNAs targeting the Myod1 locus, results in upregulation of major SkM genes in C3H10T1/2 cells detected by RT-PCR on day 18 posttransduction. (B) Reprogramming of C3H10T1/2 cells by VdC9BV is characterized by significantly higher levels of Myod1 and Myog mRNA detected by qRT-PCR on day 18 posttransduction (two-way ANOVA, Bonferroni post tests, p ≤ 0.001 for both Myod1 and Myog, n = 3 biological replicates, error bars represent SEM). Fold change in expression is relative to the (−) gRNA control group. (C) Immunocytochemistry (ICC) of reprogrammed C3H10T1/2 cells for SkM markers. The multinucleated myotubes stained for nuclear MYOD1 and MYOG proteins and sarcomeric proteins desmin and ACTN1 (α-actinin). Reprogrammed myotubes were absent in the control group (phase contrast [PC] imaging and ICC). The scale bar represents 100 μm. (D) Cell fusion assay. Cell fusion in the reprogramming cells was quantified by detecting the luciferase released after mixing two lentivirally transduced cell populations: one expressing Cre recombinase (LV-Cre) and a second containing a Cre recombinase responsive Floxed-Stuffer Luciferase cassette (LV-Floxed Luc). Fold change in relative light units (RLUs) was calculated relative to a Luc-only group (VdC9BV, M2rtTA, and LV-Floxed Luc). The RLU fold change of the (+) gRNA3 group is significantly higher on day 22 than on day 14 (two-way repeated measures [RM] ANOVA, Bonferroni post hoc test, p ≤ 0.001, n = 3 biological replicates, error bars = SEM). Overall, the (+) gRNA3 group has significantly higher luciferase activity than other groups (two-way RM ANOVA, p ≤ 0.0001). (E) Timeline denoting the main steps of the MEF reprogramming process. (F) Gene expression analysis (RT-PCR) performed on reprogrammed and control MEFs. (G) Imaging of the reprogrammed MEFs shows the presence of multinucleated myotubes containing SkM markers. No myogenesis was observed in the (−) control group. Confocal imaging shows cross-striated organization of ACTN1 and F-actin.

Figure 3

Characteristics of gRNA-Guided ^VP64dCas9-BFP^VP64-Mediated Activation of the Endogenous Myod1 Gene Locus (A) VdC9BV-mediated activation of the endogenous murine Myod1 gene is stable and sustained following doxycycline removal. The red arrowheads indicate the time points on which addition and removal of doxycycline are performed. dCas9 expression kinetics is significantly different from that of transactivated Myod1 (p ≤ 0.0001, two-way ANOVA). Fold change in expression is relative to levels at day 0.5 posttransduction (n = 4 biological replicates, error bars represent SEM). (B) Illustration of dCas9-based activators. (C) The C-terminal-only fusion of VP64 to dCas9, dCas9-BFP^VP64, fails to activate the endogenous murine Myod1 gene in C3H10T1/2 cells even in the presence of multiple gRNAs. VdC9BV significantly activated transcription of the endogenous Myod1 gene locus when compared with all other groups (p ≤ 0.001, one-way ANOVA, Tukey’s post hoc test, n = 3). Control group: M2rtTA + Empty FUW. Fold change in expression is relative to the (−) control group. (D) The N-terminal-only VP64 fusion to dCas9 (^VP64dCas9-BFP) also failed to match the ability of VdC9BV to transactivate significant levels of endogenous Myod1 expression in the presence of a single gRNA (p ≤ 0.001, one-way ANOVA, Dunnett’s post hoc test, n = 3). Control: M2rtTA +Empty FUW. Fold change in expression is relative to the (−) control group. (E) Only the N- and C-terminal VP64 fusion to dCas9 (VdC9BV) significantly activated the human MYOD1 locus of HEK293T cells in the presence of a gRNA (p ≤ 0.01, one-way ANOVA, Dunnett’s post hoc test, n = 3). (F) A gRNA targeting the antisense strand (gRNA4) can significantly activate endogenous Myod1 expression to levels similar to that achieved by the gRNA targeting the sense strand (gRNA3) (p ≤ 0.001, one-way ANOVA, Dunnett’s post hoc comparing all the groups to the [−] gRNA group, n = 3). (−) gRNA group: M2rtTA + VdC9BV + Empty FUW virus. Fold change in expression is relative to the (−) control group. (G) In the absence of the BFP domain, ^VP64dCas9^VP64 activates Myod1 expression to a lesser extent than the VdC9BV group, although dCas9 is expressed at a significantly higher level using the same MOI (C3H10T1/2 cells, unpaired two-tailed t test p = 0.0051 [dCas9], p = 0.0044 [Myod1], n = 3). Fold change in expression is relative to the VdC9BV group. Measurements for (C–G) occurred at posttransduction day 6, and the data from independent biological replicates are represented as mean ± SEM.

Figure 4

Comparison of Skeletal Reprogramming by gRNA-Guided ^VP64dCas9-BFP^VP64-Mediated Transactivation of the Endogenous Myod1 Gene Locus and by Transgenic MYOD1 Overexpression (A) Illustration depicting the experimental setup for the comparison of skeletal reprogramming by Myod1 transactivation and by transgenic MYOD1 overexpression. (B) Expression kinetics of VdC9BV and MYOD1 transgenes under doxycycline induction from days 2 to 10 posttransduction. (C) Expression kinetics of myogenic genes (Myod1, Myog, and Desmin) in C3H10T1/2 cells detected by qRT-PCR during the 8 days of doxycycline induction of both the systems. The endogenous Myod1 expression over the entire duration of induction is significantly higher in the Myod1 transactivation group than the transgenic MYOD1 overexpression group (p < 0.0001). Expression of Myog (p = 0.3740) and Desmin (p = 0.6588) is similar in both the groups (two-way ANOVA, n = 3). Fold change in expression is relative to the (−) gRNA (−) MYOD1 transgene group. (D) The effect of duration of reprogramming system induction by doxycycline on myogenic marker expression dynamics in C3H10T1/2 cells detected by qRT-PCR on day 18 posttransduction. The endogenous Myod1 and Myog expression is significantly higher in the Myod1 transactivation group than the transgenic MYOD1 overexpression group (p < 0.0001). Expression of Desmin (p = 0.5348) is similar in both the groups (two-way ANOVA, Bonferroni post tests, n = 3). Fold change in expression is relative to the (−) gRNA (−) MYOD1 transgene control group. (E) Percentage of 4',6-diamidino-2-phenylindole-stained nuclei that also express MYOD1/ MYOG protein in C3H10T1/2 cells on day 18 posttransduction is higher in the Myod1 transactivation group (mean = ∼34% MYOD1+, 33% MYOG+) than the MYOD1 overexpression group (mean = ∼10% MYOD1+, 11% MYOG+) (p = 0.0004 [MYOD1], <0.0001 [MYOG], two-tailed unpaired t test, n = 3 biological replicates). Doxycycline induction of VdC9BV and MYOD1 transgenes was carried out for 8 days till day 10 posttransduction. Data from independent biological replicates for (B)–(E) are represented as mean ± SEM.