From a genome-wide screen of RNAi molecules against SARS-CoV-2 to a validated broad-spectrum and potent prophylaxis

Abstract

Expanding the arsenal of prophylactic approaches against SARS-CoV-2 is of utmost importance, specifically those strategies that are resistant to antigenic drift in Spike. Here, we conducted a screen of over 16,000 RNAi triggers against the SARS-CoV-2 genome, using a massively parallel assay to identify hyper-potent siRNAs. We selected Ten candidates for in vitro validation and found five siRNAs that exhibited hyper-potent activity (IC50 < 20 pM) and strong blockade of infectivity in live-virus experiments. We further enhanced this activity by combinatorial pairing of the siRNA candidates and identified cocktails that were active against multiple types of variants of concern (VOC). We then examined over 2,000 possible mutations in the siRNA target sites by using saturation mutagenesis and confirmed broad protection of the leading cocktail against future variants. Finally, we demonstrated that intranasal administration of this siRNA cocktail effectively attenuates clinical signs and viral measures of disease in the gold-standard Syrian hamster model. Our results pave the way for the development of an additional layer of antiviral prophylaxis that is orthogonal to vaccines and monoclonal antibodies.

Fig. 1. A genome-wide screen of the SARS-CoV-2 genome by Sens.AI.

a Oligo design. Each oligo had a unique RNAi trigger in the context of a miR-30 backbone along with a 50 nucleotide stretch flanking the viral RNAi target site; b Library design. Each plasmid contained a specific RNAi trigger and its matching target site in cis as part of the 3’UTR of a Venus reporter; c–e Scheme of an in vitro screen in human cells: c The RNAi machinery was first turned off. Dicer^−/− 293FT cells were infected with the plasmid library and sorted for Venus^high expression, which formed the T₀ read out; d The RNAi machinery was then turned on. We restored the RNAi machinery by ectopic expression of a destabilised Dicer (ddDicer); e The activity of the RNAi machinery was modulated quantitatively. Two biological replicates were subjected to seven different conditions, designed to titer Dicer expression either downward, by treatment with an anti-Dicer siRNA, or upwards, by treatment with Shield-1. Upon FACS sorting, Venus^low and Venus^dark cells were collected, and their oligo constructs regions were sequenced; f The resulting data matrix of the screen. Each row represents a specific combination of a treatment condition (the far left column) and a FACS gate (denoted as D, for Dark, and L, for Low). These row vectors describe the enrichment of each tested RNAi trigger (blue) compared to negative (green highlighted) and positive (red highlighted) controls. The area under the curve (AUC) column was calculated based on the ability to distinguish the positive from the negative controls; g Selection of the two best conditions. The two-row vectors with the highest AUC were selected. Precision-recall (left) and receiver operating characteristics (right) curves calculated for the intrinsic controls are shown. The screen score of each RNAi trigger was calculated as the average of these two-row vectors.

Fig. 2. Validation of screen results for n = 17,589 shRNAs.

a Screen scores had a high internal consistency. The screen results showed a Pearson correlation of 72.2% between the scores in the two top-performing conditions; b Enrichment of features associated with potent shRNAs. The probability of not finding an adenine in position 20 of RNAi triggers targeting SARS-CoV-2 increased with screen scores, recapitulating previous studies indicating that when adenine occupies position 20 it weakens shRNA maturation; c Screen scores highly correlated with bioinformatic predictions. The screen scores, representing enrichment statistics as calculated by DESeq2, were highly correlated with a published machine learning algorithm predicting the potency of shRNAs; d Screen results and the selected candidates. The graph shows the resultant screen scores of each RNAi trigger that passed the quality threshold, as a function of position along the viral genome. Orange: RNAi triggers that targeted conserved regions between SARS-CoV and SARS-COV-2. Green: The selected ten candidates. Blue: all other RNAi riggers; e Dose-response curves of the selected ten candidates. siRNAs were tested using a reporter assay. The plot represents the standardised median ratio between the expression of mCherry (reporter gene) and GFP (control gene); f Potency scores of each of the 10 selected candidates. The IC₅₀ value of each of the 10 candidates was calculated based on the dose-response curves in (e). Standard errors for IC50 computations were computed using statistical bootstrap over all data points at each concentration level.

Fig. 3. siRNAs repress live SARS-CoV-2 replication in VeroE6 cells.

Blue and orange: qPCR results of the E and RdRP transcripts, respectively. Unless noted otherwise, the 100% viral load was calibrated to viral level after treatment with an anti-GFP siRNA. a Viral load of SARS-CoV-2 (ancestral strain) after treatment with the top siRNA candidates from the Sens.AI screen; b TCID₅₀ levels of SARS-CoV-2 (ancestral strain) after treatment with four of the top siRNA molecules. In each batch of the experiments, an siRNA against eGFP was used as a negative control (left panel n = 3, right panel n = 1, error bars=SEM, significance was calculated using t-test); c Viral load of SARS-CoV-2 (ancestral strain) after treatment with the top siRNA from the bioinformatic pipeline; d The effect of various siRNA cocktails against SARS-CoV-2 (ancestral strain). The results were calibrated to the repression of S3 at the same concentration; e Viral load of SARS-CoV-2 (ancestral strain) after treatment with siRNA cocktails in ALI culture. Viral load was measured in media collected from the media at 48, 72 and 96 hours postinfection (n = 2 independent samples, error bars=SEM). f Viral load post-treatment with siRNA cocktails against SARS-CoV-2 Delta versus the ancestral strain (n = 3 independent samples, error bars=SEM). g Viral load of SARS-CoV-2 Omicron versus the ancestral strain after treatment with the S5/Hel14 cocktail and other types of antivirals; (h) DeSEQ2 analysis of SARS-CoV-2 replicon treatment with the S3/S5 siRNA cocktail. We observed a sharp coverage decrease around the S3 (~23.5kbase) and S5 (~28kbase) cleavage sites along the replicon sequence (FDR values 4 × 10⁻⁸³ and 6 × 10⁻²², respectively). Error bars in panels e-f represent standard diviation, based on three replicates.

Fig. 4. Saturation mutagenesis.

a The design scheme of oligos used in this setting. The library consisted of the S5 as the shRNA trigger with 2,143 mutations, exhaustively depicting every possible single- and double-mismatch in the siRNA target site; b The effect of mismatches stratified by position for n = 66 shRNAs with a single mismatch. The X axis represents positions along the guide strand. Blue: mean values. Yellow: smoothed mean effect; Squares represent the mean, horizontal lines represent the median, box edges represent the 25% and 75% quartiles, and the whiskers represent the furthest data points within up to 50% of the interquartile range. c The distribution of the effect of mutations on the activity of S5. The vertical line at 1.0 (X axis) signifies scores that are identical to the screen score of a target site devoid of any mutation. Black: the distribution of effects of all 2143 single and double mutations. Grey: the distribution of all single mutations. Orange: the distribution of all double transitions. Red: the distribution of all double transversions.

Fig. 5. Prophylactic treatment against SARS-CoV-2 infection in Syrian hamsters.

a Dosing regimen. Syrian hamsters (n = 6, per group) pre-treated with a non-targeting siRNA (negative control, grey), the LY-CoV555 antibody (positive control, yellow) or our lead siRNA cocktail (treatment, red) were infected with 4 × 10³ PFUs of the ancestral SARS-CoV-2 strain; b Weight change post infection. Box plot of the change in weight by treatment group relative to time postinfection; c, d Viral load at day 5 postinfection. All measurements were based on qPCR of the RdRP gene five days postinfection from homogenised lungs c and nares d. In all panels p value is presented as: *<0.05,**<0.001. Ctrl: control; VL: viral load; Neg.: negative; Pos.: positive; tment: treatment; n.s.: not significant. In panels b–d, horizontal lines represent the median, box edges represent the 25% and 75% quartiles, and the whiskers represent the furthest data points within up to 50% of the interquartile range. P-values were computed using parametric bootstrap (Methods).