Optimizing a CRISPR-Cas13d Gene Circuit for Tunable Target RNA Downregulation with Minimal Collateral RNA Cutting

Abstract

The invention of RNA-guided DNA cutting systems has revolutionized biotechnology. More recently, RNA-guided RNA cutting by Cas13d entered the scene as a highly promising alternative to RNA interference to engineer cellular transcriptomes for biotechnological and therapeutic purposes. Unfortunately, "collateral damage" by indiscriminate off-target cutting tampered enthusiasm for these systems. Yet, how collateral activity, or even RNA target reduction depends on Cas13d and guide RNA abundance has remained unclear due to the lack of expression-tuning studies to address this question. Here we use precise expression-tuning gene circuits to show that both nonspecific and specific, on-target RNA reduction depend on Cas13d and guide RNA levels, and that nonspecific RNA cutting from trans cleavage might contribute to on-target RNA reduction. Using RNA-level control techniques, we develop new Multi-Level Optimized Negative-Autoregulated Cas13d and crRNA Hybrid (MONARCH) gene circuits that achieve a high dynamic range with low basal on-target RNA reduction while minimizing collateral activity in human kidney cells and green monkey cells most frequently used in human virology. MONARCH should bring RNA-guided RNA cutting systems to the forefront, as easily applicable, programmable tools for transcriptome engineering in biotechnological and medical applications.

Keywords: CRISPR; Cas13d; RNA-targeting; collateral activity; gene downregulation; synthetic gene circuit.

Figure 1

Characterization of the mNF-Cas13d gene circuit LP-integrated into the AAVS1 safe harbor site. (A) Illustration of the need to understand and optimize multiple aspects of RNA reduction by various CRISPR/Cas13d RNase activities. This could be fulfilled by fine-tuning both Cas13d and crRNA expression levels using stably integrated synthetic gene circuits. Arrows indicate a desired optimization of each aspect, where the blue arrow corresponds to an increase, the red arrow corresponds to a decrease and the gray arrow corresponds to no change. (B) Diagram of the synthetic mNF-Cas13d gene circuit for Dox-controlled, joint tuning of tetracycline repressor (TetR), mCherry reporter and Cas13d protein levels after AAVS1 site-specific LP-integration. TetO: Tetracycline Operator; P2A: self-cleaving peptide. (C) Representative dose–responses of red fluorescence intensity histograms for the stably integrated mNF-Cas13d gene circuit, measured at 0, 0.01, 0.02, 0.05, 0.1, 0.2, 0.5, 1, 2, 5, 10, 100 ng/mL Dox levels, respectively. (D) Dose–responses of mean fluorescence intensity of the mCherry reporter for the stably LP-integrated mNF-Cas13d gene circuit in HEK 293 cells (n = 3). (E) Dose–responses of the coefficient of variation (CV) of mCherry reporter for the stably LP-integrated mNF-Cas13d gene circuit in HEK 293 cells (n = 3).

Figure 2

Negative-feedback transcriptional regulation of both Cas13d protein and crRNA improves dose–response characteristics. (A) Diagram of single-vector transfection to test Cas13d’s Dox dose-dependent on-target activity. The plasmid encoding the BFP reporter and the BFP-targeting crRNA or nontargeting crRNA was transfected into HEK 293 cells with mNF-Cas13d stably LP-integrated. Cells were Dox-induced and incubated for 72 h before flow cytometry measurements. (B) Dox dose–responses of mean mCherry fluorescence intensity in HEK 293 cells with stably LP-integrated mNF-Cas13d gene circuit 72 h post-transfection. Unpaired two-tailed t test, n = 3, **P < 0.01, ***P < 0.001. (C) Dox dose-dependent reduction of transiently transfected BFP reporter by Cas13d targeting. All BFP expression levels were normalized to uninduced cells with nontargeting guide, n = 3. (D) Diagram of single-vector transfection to test Cas13d’s Dox dose-dependent, on-target activity with regulated crRNA expression. BFP-targeting crRNA was driven by a modified human U6 (hU6) promoter containing two Tetracycline Operator (TetO2) sites flanking the TATA-box. This hU6-2O promoter can then be repressed by hTetR expressed from the mNF-Cas13d gene circuit. (E) Comparison of mCherry dose–responses from the mNF-Cas13d gene circuit 72 h post-transfection with all three transfection constructs. Unpaired two-tailed t test, n = 3, **P < 0.01, ***P < 0.001. (F) Comparison of Dox dose-dependent BFP reduction by Cas13d, 72 h post-transfection with all three transfection constructs. All BFP expression levels were normalized to uninduced samples with nontargeting guide, n = 3. (G) Diagram of all-in-one constructs to test Cas13d’s on-target activity on stably genome-integrated targets. The GFP reporter-targeting crRNA is expressed from the TetR-regulated hU6-2O promoter, while the nontargeting guide is freely expressed from the normal hU6 promoter. The whole construct was FLP-RMCE integrated using the same HEK 293 LP parental cells. (H) Dose–responses of mean mCherry fluorescence intensity for stably integrated all-in-one constructs in engineered HEK 293 cells 72 h postinduction. Unpaired two-tailed t test, n = 3, **P < 0.01, ***P < 0.001. (I) Dose-dependent reduction of stably integrated GFP reporter expression by Cas13d. All GFP expression levels were normalized to uninduced samples with nontargeting guide, n = 3.

Figure 3

Collateral activity due to Cas13d hyperactivation depends on Cas13d and crRNA expression levels. (A) Diagram of experimental setup to assess collateral activity from hyperactivated Cas13d with different guide control scenarios. The SV40 promoter-driven GFP target was cotransfected with either TetR-regulated GFP-targeting crRNA, constitutively expressed GFP-targeting crRNA or nontargeting crRNA. All three crRNA plasmids contain the same BFP gene driven by the EF1α promoter. Cells were transfected under 72 h induction before flow cytometry measurements. (B) Comparison of mCherry dose–responses from the mNF-Cas13d gene circuit 72 h post-transfection for all three cotransfection scenarios. Unpaired two-tailed t test, n = 3, **P < 0.01. (C) Comparison of relative GFP levels indicating dose–responses of on-target activity from the mNF-Cas13d gene circuit 72 h post-transfection with all three cotransfection scenarios. All GFP expression levels were normalized to the uninduced sample with nontargeting guide, n = 3. (D) Comparison of relative BFP levels indicating dose–responses of off-target activity from the mNF-Cas13d gene circuit 72 h post-transfection with all three cotransfection setups. All the BFP expression levels were normalized to the uninduced sample with nontargeting guide, n = 3. (E) Diagram of all-in-one constructs to test Cas13d’s on-target and off-target activities on stably genome-integrated targets. GFP reporter-targeting crRNA is expressed from the hTetR-regulated hU6-2O promoter, while the nontargeting guide is constitutively expressed from the normal hU6 promoter. The whole construct is FLP-RMCE-integrated using the same HEK 293 LP parental cells. (F) Dose–responses of mCherry reporter indicating Cas13d expression levels for stably integrated all-in-one constructs in HEK 293 cells 72 h postinduction. Unpaired two-tailed t test, n = 3, *P < 0.05, **P < 0.01. (G) Dose–responses of relative GFP reporter levels indicating on-target activity for stably integrated all-in-one constructs in HEK 293 cells 72 h postinduction. All GFP expression levels were normalized to uninduced samples with nontargeting guide, n = 3. (H) Dose–responses of relative BFP reporter levels indicating off-target activity for stably integrated all-in-one constructs in HEK 293 cells 72 h postinduction. All BFP expression levels were normalized to uninduced samples with nontargeting guide, n = 3.

Figure 4

Multilevel negative autoregulation of Cas13d and crRNA in MONARCH reduces the basal target downregulation. (A) Design and rationale for multilevel negative autoregulation of both Cas13d and crRNA expression, to avoid their unwanted overexpression, geberating too many activated Cas13d:crRNA complexes with excessive basal effect and collateral activity. Incorporating the crRNA into the same transcript with Cas13d not only brings both under the same tight transcriptional regulation, but also adds RNA-level regulation via crRNA processing by Cas13d. This may reduce the basal effect as well as the collateral activity. (B) Representative dose–responses of fluorescence intensity histograms from the stably integrated MONARCH 1.0 circuit measured at 0, 0.1, 0.2, 0.5, 1, 10 ng/mL Dox levels, respectively. (C) Dose–responses of mean mCherry fluorescence intensity for the MONARCH 1.0 circuit stably integrated into HEK 293 LP cells (n = 3). (D) Dose–responses of the coefficient of variation (CV) of mCherry reporter expression for MONARCH 1.0 stably integrated into HEK 293 LP cells (n = 3). (E) Diagram for testing on-target activity of Cas13d expressed from MONARCH 1.0 on stably genome-integrated targets. The whole construct is FLP-RMCE-integrated using the same HEK 293 LP parental cells. Basal level is determined with integration of only the SV40 promoter-driven GFP target. (F) Dose–responses of mean fluorescence intensity of mCherry reporter for stably integrated MONARCH 1.0_SV40-GFP construct in HEK 293 cells (n = 3). (G) Dose–responses of coefficient of variation (CV) of mCherry reporter for stably integrated MONARCH 1.0_SV40-GFP in HEK 293 cells (n = 3). (H) Dose–responses of relative GFP levels indicating on-target activity for the MONARCH 1.0_SV40-GFP construct stably LP-integrated into HEK 293 cells. All GFP expression levels were normalized to the basal sample, n = 3. (I) Dose–responses of coefficient of variation (CV) of GFP target expression for MONARCH 1.0_SV40-GFP stably LP-integrated into HEK 293 cells (n = 3). (J) Diagram for testing on-target activity of MONARCH 1.0 on an endogenous target. GFP-targeting crRNA serves only as an RNA-level regulator in this scenario. Multiple crRNAs targeting BACH1 are cloned into a single plasmid and transfected together into the cells. Reverse transcription quantitative polymerase chain reaction (RT-qPCR) was performed after 144 h of induction and incubation post-transfection. (K) Dose-dependent reduction of endogenous BACH1 by Cas13d expressed from MONARCH 1.0 stably LP-integrated into HEK 293 cells. One-way ANOVA with Tukey’s multiple comparisons tests between each dose and native control, n = 3.

Figure 5

Optimized multilevel negative autoregulation of Cas13d and crRNA improves the on-target potency and minimizes the basal effect. (A) Design and rationale for optimized multilevel negative-autoregulation for Cas13d and crRNA expression. Compared to MONARCH 1.0, a tertiary RNA structural motif “Triplex” is incorporated between the CDS (coding sequence) and crRNA to stabilize transcripts lacking poly(A) due to crRNA processing, enabling a moderately extended window of protein expression. (B) Representative dose–responses of fluorescence intensity histograms from the stably integrated MONARCH 2.0 circuit measured at 0, 0.1, 0.2, 0.5, 1, 10 ng/mL Dox levels, respectively. (C) Dose–responses of mean mCherry fluorescence intensity for the MONARCH 2.0 circuit stably LP-integrated into HEK 293 cells (n = 3). (D) Dose–responses of coefficient of variation (CV) of the mCherry reporter for MONARCH 2.0 stably LP-integrated into HEK 293 cells (n = 3). (E) Diagram for testing the on-target activity of Cas13d expressed from MONARCH 2.0 on genomically integrated targets. The whole construct is FLP-RMCE-integrated using the same HEK 293 LP parental cells. The basal level is determined through the integration of only SV40 promoter-driven GFP. (F) Dose–responses of mean mCherry fluorescence intensity for the MONARCH 2.0_SV40-GFP construct stably LP-integrated into HEK 293 cells (n = 3). (G) Dose–responses of coefficient of variation (CV) of the mCherry reporter for MONARCH 2.0_SV40-GFP stably LP-integrated into HEK 293 cells (n = 3). (H) Dose–responses of relative GFP levels indicating on-target activity for MONARCH 2.0_SV40-GFP stably LP-integrated into HEK 293 cells. All the GFP expression levels were normalized to the basal sample, n = 3. (I) Dose–responses of coefficient of variation (CV) of the GFP target for stably integrated MONARCH 2.0_SV40-GFP in HEK 293 cells (n = 3). (J) Diagram for testing on-target activity of MONARCH 2.0 on an endogenous target. GFP-targeting crRNA serves only as an RNA-level regulator in this scenario. Multiple crRNAs targeting BACH1 are cloned into a single plasmid and transfected together into the cells. RT-qPCR was performed after 144 h of induction and incubation post-transfection. (K) Dose-dependent reduction of endogenous BACH1 by Cas13d expressed from MONARCH 2.0 stably LP-integrated into HEK 293 cells. One-way ANOVA with Tukey’s multiple comparisons test between each dose and native control, n = 3.

Figure 6

Mathematical modeling indicates the origins of MONARCH’s benefits. (A) Illustration of hypothesized mechanisms to explain experimental observations from both mNF-Cas13d and MONARCH system. Briefly, with mNF-Cas13d, highly expressed Cas13d and crRNA will form excessive amounts of Cas13d:crRNA complexes, which can be hyperactivated by target RNA cutting. Due to the limited target RNA pool, the excessive hyperactive Cas13d:crRNA complexes will start to degrade nontarget RNA, causing collateral activity. With MONARCH, the Cas13d:crRNA complex abundance is limited by multilevel negative autoregulation and remains sufficiently curtailed compared to the target RNA pool. Then the limited amount of hyperactive complex will mostly cut target RNA, resulting in an elevated on-target efficacy while minimize the collateral activity. Red dashed curve, degraded target RNA; gray dash curve, degraded nontarget RNA; “*” indicates nonspecific, trans cleavage activation (complex hyperactivation). (B) Schematic of modeling reactions. Top: reactions to produce the active Cas13d-crRNA complex (A) in the mNF-Cas13d system. Here, the guide RNA (G) is produced from a separate promoter from the Cas13d transcript (R). The Cas13d transcript is translated into Cas13d protein (C), which then associates with the guide RNA to become activated. Middle: reactions to produce the active Cas13d complex in the MONARCH system. Here, the guide RNA is coexpressed with the Cas13d transcript. When the Cas13d protein processes the RNA, it produces an unstable Cas13d transcript (U), which can still be translated into Cas13d protein. Bottom: for both models, once the active Cas13d complex is created, it can then start cutting target RNA (T). When this happens, the Cas13d complex becomes hyperactive (H), allowing it to cleave both target RNA and nonspecific RNA (N). (See Supplemental Results for more details). (C) Plot of simulated nonspecific RNA levels for each system. N0 (circles): no Cas13d control (i.e., only constitutive synthesis and linear degradation). N1 (full line): mNF-Cas13d system. N2 (dashed line): MONARCH system. Bar chart shows the relative basal levels, calculated relative to N0. (D) Plot of simulated target RNA levels for each system. T0 (circles): no Cas13d control (i.e., only constitutive synthesis and linear degradation). T1 (full line): mNF-Cas13d system. T2 (dashed line): MONARCH system. Bar chart shows the average slope, calculated from the dynamic range of each system (See Supplemental Results for more details). (E) Repeat simulation with c_T (cutting rate of target RNA by hyperactive Cas13d) set to zero, indicating that nonspecific RNA cutting contributes to target RNA reduction. Legend is the same as in (D).

Figure 7

MONARCH established in Vero E6 LP cells demonstrates antiviral capability and potential for improving on-target efficiency. (A) Diagram of experimental setup to repurpose MONARCH to target SARS-CoV-2 viral genome fragments in engineered Vero E6 cells. Both MONARCH systems are stably integrated into the AAVS1 ortholog in the Vero E6 genome using the same Landing-Pad and RMCE strategy. Successful by Cas13d targeting of SARS-CoV-2 viral genome fragments cloned into the 3′UTR of the GFP transcript will result in mRNA degradation, which can be measured by GFP fluorescence reduction. (B) Comparison of mCherry dose–responses from MONARCH 1.0 and 2.0 stably LP-integrated into Vero E6 cells 72 h post-transfection. (C) Comparison of coefficient of variation (CV) of mCherry dose–responses from MONARCH 1.0 and 2.0 stably integrated in Vero E6 cells 72 h post-transfection. (D) Comparison of relative GFP levels indicating dose–responses of on-target activity from MONARCH 1.0 and 2.0 stably LP-integrated into Vero E6 cells, 72 h post-transfection. All the GFP expression levels were normalized to native Vero E6 cells transfected with the target plasmid only. Unpaired two-tailed t test for each dosage comparison, n = 3, **P < 0.01, ***P < 0.001. (E) Diagram of experimental setup to assess further improvement of MONARCH on-target activity by providing extra crRNA. Target donor plasmid is cotransfected with either SARS-CoV-2-targeting CRISPR array, nontargeting crRNA or blank control into the engineered Vero E6 cells. Engineered Vero E6 cells are preinduced with 10 ng/mL doxycycline 72 h before transfection. (F) Relative mCherry expression increases from 10 ng/mL doxycycline preinduced MONARCH 1.0 and 2.0 stably integrated into Vero E6 cells, 72 h post-transfection with extra crRNA plasmids. All mCherry expression levels were normalized to the Vero E6 cells transfected with a blank control plasmid. (G) Relative GFP level reductions 72 h post-transfection with extra crRNA plasmids. All GFP expression levels were normalized to the Vero E6 cells transfected with a blank control plasmid.