PERSIST platform provides programmable RNA regulation using CRISPR endoRNases

Abstract

Regulated transgene expression is an integral component of gene therapies, cell therapies and biomanufacturing. However, transcription factor-based regulation, upon which most applications are based, suffers from complications such as epigenetic silencing that limit expression longevity and reliability. Constitutive transgene transcription paired with post-transcriptional gene regulation could combat silencing, but few such RNA- or protein-level platforms exist. Here we develop an RNA-regulation platform we call "PERSIST" which consists of nine CRISPR-specific endoRNases as RNA-level activators and repressors as well as modular OFF- and ON-switch regulatory motifs. We show that PERSIST-regulated transgenes exhibit strong OFF and ON responses, resist silencing for at least two months, and can be readily layered to construct cascades, logic functions, switches and other sophisticated circuit topologies. The orthogonal, modular and composable nature of this platform as well as the ease in constructing robust and predictable gene circuits promises myriad applications in gene and cell therapies.

Fig. 1. Engineering PERSIST-OFF- and ON- motifs for regulated gene expression.

a Schematic of RNA-based OFF-switch and ON-switch motif designs that are regulated by RNA cleavage. b Gene expression reduction due to appending increasing numbers of degradation motifs. Fold change was calculated as described in Supplementary Fig. 3. n = 3 biologically independent samples where each sample represents the evaluation of >1000 transfected cells (HEK293FT). Data are presented as mean ± s.d. c Triplex structure rescues gene expression only after removal of degradation motifs by RNase P. Data were analyzed from the total number of samples indicated in the bars where each sample represents the evaluation of >1000 transfected cells (HEK293FT). Data are presented as mean ± s.d. d PERSIST shows resistance to epigenetic silencing in clonal populations. Three different constructs (left) each expressing mKO2 (constitutive, Tet-On switch, or PERSIST-ON switch) were genomically integrated and single-cell sorted to evaluate clonal silencing effects (see Supplementary Fig. 5). The table (middle) shows the maintenance and induction conditions of each cell line. 22 and 55 days after sorting 12 clonal lines for each cell type were induced with either 4 μM Dox or transfected with Csy4 both in the presence and absence of TSA and evaluated by flow cytometry two days later. The ratio of ±TSA for mKO2 expression was calculated for each cell line in each sample type where a ratio >1 indicates rescue of response by TSA. The Tet-On system shows increased mKO2 expression in the presence of TSA, which indicates epigenetic silencing effects. The PERSIST switch resists silencing similar to the constitutively expressed reporter. Data were analyzed from the total number of clones indicated in the bars where each sample represents the evaluation of ≥6 CHO-K1 cells. Data are presented as mean ± s.d. e CRISPR endoRNases activate the PERSIST-ON motif and repress the PERSIST-OFF motif. Fold change is normalized to the effect of adding endoRNase to a control reporter without an endoRNase recognition site as described in Supplementary Fig. 4. Constructs and endoRNases were optimized (see Supplementary Figs. 6–8; the best variants are shown here). n = 2 biologically independent samples where each sample represents the evaluation of >1000 transfected cells (HEK293FT). Data bars are presented as mean.

Fig. 2. PERSIST endoRNase platform is orthogonal and composable.

a Evaluation of Cas endoRNase pairwise orthogonality. The ability of each Cas endoRNase to repress reporters containing each endoRNase recognition site was evaluated by poly-transfection. Fold repression values are indicated by color bar. While some pairs should be avoided, most endoRNases cleave orthogonally (see Supplementary Figs. 11–12). b Cas endoRNase two-stage repression cascade. c Cas endoRNase two-stage activation/repression cascade. b, c For plasmid ratio calculations, CasE plasmid was maintained at 75 ng. CasE plasmid transfection efficiency was tracked by a plasmid encoding constitutive mKO2 (x axis) while Csy4 and EYFP reporter plasmid transfection efficiencies were tracked by a plasmid encoding constitutive tagBFP. Summary values were calculated for tagBFP-positive cells. d Cas endoRNase auto-positive feedback. We compared a CasE construct tagged with EYFP alone (gray line), a construct containing the PERSIST-ON motif but no endoRNase target site (blue line), and a construct containing a CasE target site in the PERSIST motif (green line). Construct transfection efficiencies were tracked by a plasmid encoding constitutive tagBFP. b–d n = 2 biologically independent samples as shown by separate curves where each sample represents the evaluation of >1000 transfected cells (HEK293FT). The shaded region is the maximum interquartile range of all cells in each bin.

Fig. 3. PERSIST endoRNase platform is modular.

All 16 two-input Boolean logic functions can be created using Cas endoRNases and the PERSIST-ON- and OFF-switches. Each logic function was constructed using EYFP as output with other constructs included as necessary to confer a designed response to two inputs: Csy4 and CasE. Cse3 endoRNase was used as an intermediate regulator where necessary. Each set of constructs encoding specific logic functions was evaluated without input endoRNases, with CasE only, with Csy4 only, and with both Csy4 and CasE via poly-transfection (Supplementary Fig. 15). The expected Boolean logic output is depicted below each chart. All logic functions can be correctly identified using the same approximately 10-fold separation between ON and OFF values (gray region). n = 2 or n = 3 biologically independent samples as indicated by the points where each sample represents the evaluation of >1000 transfected cells (HEK293FT). Data bars are presented as mean normalized output.

Fig. 4. Dual function of Cas endoRNases in the PERSIST platform.

a A tagBFP-encoding OFF-switch reporter (blue line) and an EYFP-encoding ON-switch reporter (yellow line) can be acted on simultaneously by Csy4 in the same cell. The ON/OFF-switch reporters and Csy4 plasmid were co-transfected with plasmids encoding constitutive iRFP720 and mKO2 respectively in order to track their corresponding transfection efficiencies. Summary values were calculated for iRFP720-positive cells. b, A coherent feed-forward loop with dual function Cas protein improves the response of the PERSIST-ON-switch motif. For plasmid ratio calculations, CasE plasmid was maintained at 75 ng. Transfection efficiency of the EYFP reporter and intermediate Csy4 plasmid was tracked by a plasmid encoding constitutive tagBFP while CasE transfection efficiency was tracked by a plasmid encoding constitutive mKO2. Summary values were calculated for tagBFP-positive cells. Fold change (inset) was calculated by normalizing to the EYFP value at the lowest mKO2 bin (see Supplementary Fig. 16) and evaluated at mKO2 = 10,000. Over 1000-fold dynamic range is achieved for certain Csy:CasE ratios compared to roughly 100-fold dynamic range achieved by the ON-switch alone (gray). a, b n = 2 biologically independent samples where each sample represents the evaluation of >1000 transfected cells (HEK293FT). The shaded region is the maximum interquartile range of all cells in each bin.

Fig. 5. PERSIST dual function enables single-element positive feedback + repression motif.

a Schematic of positive feedback + repression motif. b Csy4 activates itself through its own PERSIST-ON motif and represses another element. Transfection efficiency of the EYFP reporter was tracked by a plasmid encoding constitutive tagBFP while Csy4-containing plasmid transfection efficiency was tracked by a plasmid encoding constitutive mKO2. Summary values were calculated for tagBFP-positive cells. n = 2 biologically independent samples, as shown by separate curves where each sample represents the evaluation of >1000 transfected cells (HEK293FT). The shaded region is the maximum interquartile range of all cells in each bin. c The positive feedback + repression motif can be used to improve the performance of the second stage of a three-stage repression cascade. An EYFP reporter tracks the output for subsequent additions of each stage. A larger percentage of stage-2 cells exhibit “ON" behavior with endoRNase positive feedback (right) compared to when repression alone is used (left), with a trade-off existing of an increase in the third stage response. Each bar represents an evaluation of cells within a large range of transfection levels of each stage (Supplementary Fig. 17). n = 2 biologically independent samples, where each sample represents the evaluation of >1000 transfected cells (HEK293FT). d The positive feedback + repression motif is used to make a genetic bistable switch. mKO2 and EYFP reporter expression was evaluated for iRFP720- and tagBFP-positive cells. Only bistable switch plasmids were used with no additional inputs, yet 97% of cells display expression in only a high-mKO2/low-EYFP (red dots) or high-EYFP/low-mKO2 (yellow dots) state after 48 hours. Few cells expressed high levels of both (blue dots) or none (gray dots). Percentages of cells in each state are shown in the inset and depicted in the bar. n = 1 representative sample (HEK293FT).