Programmable multistep CRISPR gene activation via control of RNA polymerase III termination

Abstract

Although genomes encode instructions for mammalian cell differentiation with rich syntactic relationships, existing methods for genetically programming cells have only modest capabilities for stepwise gene regulation. Here, we develop a sequential genetic system that transcriptionally activates endogenous genes in a preprogrammed, stepwise manner. This system uses the removal of an RNA polymerase III termination sequence to trigger both the transcriptional activation and DNA endonuclease activities of a Cas9-VPR protein, driving progression through a cascade of gene activation events. The system's functionality in human cells, including iPSCs, enables the development of a path for cellular programming by controlling the sequential order of gene activation to influence cellular states.

Fig. 1.. The proGuide is a conditional CRISPR sgRNA.

(A) Schematic illustrating the conversion of an inactive proGuide to an active matureGuide. (Top) Plasmid DNA for Cas9-VPR expression and a trigger sgRNA plasmid encode the components, which when transfected into cells, target Cas9 activity to the CTS corresponding to the spacer (red). (Bottom left) In the absence of a trigger, an inactivation moiety embedded in the proGuide prevents activity as a guide RNA. (Bottom right) In the presence of a trigger guide RNA specific to the CTS (red), the inactivation moiety is excised from the proGuide plasmid DNA, resulting in transcription of an active matureGuide RNA. (B) Schematic illustrating progression of a proGuide cascade encoded in plasmid DNA. Color elements for steps 1 and 2 follow the rubric from (A). Steps 3 and 4 add new spacer and CTS for each step, and step 4 includes multiple proGuides with the same CTS (blue) and different 14-nt spacers target three genes for CRISPRa.

Fig. 2.. Inactivation of proGuides by RNA Pol III transcriptional termination sequences.

(A) Schematic representation of a cascade of proGuide plasmids, wherein the spacer sequence functions as an output activity (black oval) of one guide RNA and corresponds to the CTS (white oval) as the sensing input in a downstream proGuide DNA. (B to D) Reduction of GFP-positive HEK293T cells harboring a genomic EGFP expression cassette was determined by flow cytometry 72 hours after transient transfection of guide RNA plasmids and a Cas9 expression plasmid. Control transfections either lacked an EGFP targeting guide RNA (Trfx; white bars, dotted line) or included an sgRNA targeting EGFP (sgRNA; black bars). (B) First-generation proGuides displayed activity in the absence of a trigger RNA (light gray bars) when inactivation sequences were positioned in the tetraloop (T) region. ProGuides exhibited incomplete conversion to an active state when located in the hairpin (H) region. (C) Inactivation with only the ribozyme (Rz), only the 6-nt polyT tract (PolyT) or both (Rz + PolyT) all displayed leakiness. (D) Increasing the length of the polyT tract reduced leakiness of EGFP proGuides below that of the first-generation inactivation sequences (Rz + PolyT dotted line). (E) Schematic of the DNA sequence (top) encoding a dual polyT tract and predicted RNA secondary structure (bottom). (F) Absence of detectable leak from proGuides harboring the dual polyT tract depicted in (E). All data represent mean of biological triplicates ± SD. Statistical significance is denoted as follows: *P < 0.05, **P < 0.01, and ****P < 0.0001; ns indicates P > 0.05 (unpaired t test). SD values smaller than the figure symbol were omitted for clarity.

Fig. 3.. Effects of the orientation of CTS repeats on conversion of proGuides to active guide RNA.

(A) Schematic of the arrangement of different proGuide components in a DNA sequence (top) and predicted RNA sequence structures (bottom) following a perfect NHEJ repair–mediated nested deletion between the two Cas9 cut sites. (B) Transient transfections (as described in Fig. 2) with proGuide expression plasmids depicted in (A). (C) Conversion of IR1 configuration to an active guide RNA state was not significantly changed by the termination sequences (schematic on left). All data represent the mean of biological triplicates ± SD. **** indicated P < 0.0001 (unpaired t test). rRNA, ribosomal RNA.

Fig. 4.. Effects of orientation of CTS repeats on the RNA sequence of converted matureGuides.

(A to E) DNA sequencing results of reverse transcribed guide RNA from HEK293T cells transiently transfected with proGuides containing CTS configurations depicted in Fig. 3A. Sequences were analyzed and stratified on the basis of RNA transcript size and mapped to proGuide reference sequences. The frequency of deletions at each base in the proGuide (left graph) is depicted with the expected Cas9 cleavage sites assuming perfect NHEJ repair (arrows). Shading corresponds to regions of the CTS (purple), nontarget repeat sequence (teal), and polyT inactivation moiety (red). The distribution of deletion sizes for each CTS variant (right graph) is depicted with perfect repair between the Cas9 cut site (pink), near-perfect repair within 3 bp at either end of the Cas9 cleavage site (green), or other deletion sizes (gray). proGuide DNA containing CTSs oriented as IRs (IR1) displays higher frequency of perfect repair than DR sequences. Data represent all sequences with at least two reads in a sample.

Fig. 5.. Sequence composition of CTS affects efficiency of proGuide conversion.

(A) The EGFP disruption activity of different ratios of proGuide: trigger plasmid DNA was determined by flow cytometry 72 hours after transient transfection (similar to Fig. 2). Three proGuides and their matched trigger plasmids used different CTS (CTS101, CTS102, and CTS103). (B) Activation of the endogenous CXCR4 gene determined by cell surface protein expression was determined by flow cytometry 48 hours after transient transfection of a proGuide (CTS101, CTS102, and CTS103) with a 14-nt spacer targeting the CXCR4 promoter, a trigger plasmid, and a Cas9-VPR expression plasmid. Transfections lacking a guide RNA and with an sgRNA with the 14-nt CXCR4 spacer were used as negative/positive controls. (C) Two proGuide cascades constructed in opposing directions using spacers and CTS sequences corresponding to CTS101, CTS102, and CTS103 (fig. S10). Cascade plasmids were transiently transfected into HEK293T cells with a Cas9-VPR expression plasmid and assessed by flow cytometry for surface protein expression of CXCR4. (D) Activity of proGuides (+/− trigger) harboring 20 engineered CTS sequences determined by EGFP disruption, as described in Fig. 2 and fig. S1. (E) Two proGuide cascades in opposing directions using engineered CTS sequences (D) in the sequential order indicated above the graph. All data represent the mean of biological triplicates ± SD. SD values smaller than the length of the figure symbol were excluded for clarity.

Fig. 6.. Control over kinetics of endogenous gene activation by transient transfection of proGuide plasmids.

(A) Schematic depicting plasmid DNA composition of proGuide cascades used in (B), as described in fig. S11. (B) CXCR4 surface protein expression was measured by flow cytometry at 12-hour intervals after transient transfection of HEK293T cells with Cas9-VPR and plasmid mixes depicted in (A). (C) The order of the CTS used in cascades of proGuides was iterated on by generating new proGuide plasmids for stepwise progression to activation of CXCR4 transcriptional activation by Cas9-VPR. Each graph shows CXCR4 surface protein expression for cascades designed to activate CXCR4 at the indicated step number. All data represent the mean of biological triplicates ± SD. SD values smaller than the length of the figure symbol were excluded for clarity.

Fig. 7.. Syntactic activation of CD105 and CXCR4 genes using cascades of proGuides.

(A) Schematic depicting plasmid DNA composition of proGuide cascades used in (B). (B) CXCR4 and CD105 surface protein expression was measured by flow cytometry at 12-hour intervals after transient transfection of HEK293T cells with Cas9-VPR and plasmid mixes depicted in (A). Graphs show the percentage of cells positive for CXCR4/CD105 relative to the number of positive cells observed in sgRNA-positive control transfections at each time point. Note that since the sgRNA-positive control did not display CD105 positivity at 12 hours, normalized data are not shown for CD105 at that time point. All data represent the mean of biological triplicates ± SD. SD values smaller than the length of the figure symbol were excluded for clarity.

Fig. 8.. Pool of proGuide plasmids developed for use in iPSCs.

Schematic illustrates proGuide plasmid arrangement in three different cascades, each designed to activate different permutations of DLL4, CD4, and CD105 at step 1, 2, 5, or 7. The schematic follows the format described in fig. S11. Note that four guide RNAs are used to target the CRISPRa activity on each endogenous gene with binding of four different spacer sequences arranged on the gene’s promoter region.

Fig. 9.. Syntactic activation of several genes in iPSCs using cascades of proGuides.

(A) The expression of cell surface marker proteins, DLL4, CD4, and CD105, in human iPSCs after nucleofection of one of the proGuide cascades shown in Fig. 8. Protein expression was measured every 12 hours by flow cytometry, and the gating strategy described in fig. S17 was used to determine the percentage of positive cells from the parental population [i.e., those expressing the indicated marker from the pool of cells that expressed the marker(s) upstream in the cascades]. Each graph corresponds to one of the cascades shown in Fig. 8. (B) The rate of activation of each gene was determined from the frequencies in (A) and represented as the percent increase per 12 hours. (C) Relative activation of surface marker proteins, DLL4, CD4, and CD105, in iPSCs programmed at steps 1, 2, 5, and 7. Each graph shows data from cells nucleofected with different proGuide cascades. All data represent the mean of biological triplicates ± SD. SD values smaller than the length of the figure symbol were excluded for clarity.