Systematic analysis of RNAi-accessible SARS-CoV-2 replication steps identi�es ORF1 as promising target

A promising approach to tackle the Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) could be small interfering (si)RNAs. The proof of concept that SARS-CoV-2 can be inhibited with siRNAs, however, is missing. Here, we report that siRNAs can target genomic RNA (gRNA) of SARS-CoV-2 after cell entry, terminating replication before start of transcription and preventing cytopathic effects. Coronaviruses replicate via negative sense intermediate transcripts using a unique discontinuous transcription process. As a result, each viral RNA contains identical sequences at the 5’ and 3’ end. Surprisingly, siRNAs were not active against intermediate negative sense transcripts. Targeting sequences shared by different viral transcripts allowed simultaneous suppression of gRNA and subgenomic (sg)RNAs by a single siRNA. The most effective suppression of viral replication and spread, however, was achieved by siRNAs targeting open reading frame 1 (ORF1) which is solely part of gRNA. We propose two independent mechanisms for this: An increased accessibility of translational-active ORF1 before the start of transcription, as well as highly abundant sgRNAs out-competing siRNAs that target common sequences of transcripts. Our work encourages the development of siRNA-based therapies for COVID-19 and suggests that an early therapy start, together with targeting ORF1, might be key for high antiviral e�cacy.


Introduction
Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) is causing a pandemic with so far unforeseeable consequences on global health, politics and economy.SARS-CoV-2, like other coronaviruses affecting humans, is mainly transmitted via respiratory secretions (1) and replicates primarily in respiratory epithelial cells (2).Due to its lytic cell cycle (3), it causes severe endothelial injury and widespread microangiopathy (4), which can result in respiratory failure and death.While some progress has been made by repurposing the RNA polymerase inhibitor Remdesivir (5) or by ameliorating SARS-CoV-2 induced lung injury using Dexamethason (6), lethality of COVID-19 remains high (7).A promising alternative approach could be to deliver small interfering (si)RNAs (8-11) to the respiratory tract by inhalation (12) and induce degradation of viral RNAs by the RNA-interference (RNAi) machinery.Studies performed with Severe Acute Respiratory Syndrome Coronavirus-1 (SARS-CoV-1) or Middle East Respiratory Syndrome Coronavirus (MERS-CoV), showed that siRNAs can silence viral RNA and relieve symptoms caused by related coronaviruses (13)(14)(15)(16)(17).However, while several publications have proposed siRNAs as potential therapeutic also for COVID- 19 (8-11), until today there is no proof that SARS-CoV-2 can be inhibited by siRNAs, not to mention a detailed systematic analysis, which replication steps are accessible for RNAi.SARS-CoV-2, as other coronaviruses, has a positive sense, single-stranded RNA genome with a length of approximately 30,000 nucleotides.Following binding to the cellular receptor angiotensin-converting enzyme 2 (18), the virus is taken up via endocytosis (19).After fusion with the endosomal membrane (20), the ribonucleocapsid is released into the cytoplasm.Here, the viral genome serves as template for initial translation of the polyprotein 1ab (pp1b) from Open Reading Frame 1 (ORF1) by the cellular ribosomal machinery.Pp1ab is cleaved into 16 non-structural proteins (NSPs) of which several assemble around the viral genome to form the replication/transcription complex (RTC) (21).As for other positive sense RNA viruses, transcription does not take place in the cytosol, but exclusively within doublemembrane vesicles (22).Therefore, the viral RTC associates with endoplasmic reticulum membranes to form so-called viral replication organelles (ROs).Here, the viral genome serves as template for transcription of full-length progenitor genomic (g)RNA as well as subgenomic (sg)RNAs encoding for structural (S, envelope protein [E], membrane protein [M], Nucleocapsid [N]) as well as accessory proteins (3a, 6, 7a, 7b, 8 and 10) (23).As a results of the discontinuous transcription (24,25), each coronaviral RNA contains an identical 3´ (the ∼70 nucleotides long leader sequence [L]) as well as 5' end (N ORF and 3ú ntranslated region [3´UTR]) (23).Next, sgRNAs are released from ROs (26), translated into the corresponding protein and gRNA packaged by the structural proteins to assemble progeny virions.
Coronaviruses protect their RNA well.Besides the envelope and capsid protecting virions, nucleocapsid proteins bind directly to the viral genome.Thus, even between uncoating and incorporation into doublemembraned ROs, the genome is not present as naked RNA (27).Furthermore, while sgRNAs are exported from ROs for translation, this does not seem to be the case for gRNA which remains associated with double-membraned vesicles (26).Currently, it is not clear whether the different viral RNA species produced during replication are accessible for an RNAi-based therapy.Furthermore, certain viral components might be essential for replication, whereas the suppression of others might be tolerated by the virus.Thus, suppression of reporter constructs as often performed during siRNA development does not accurately predict the effect of siRNAs on viral replication and spread.To shed light upon these questions, we systematically analyzed which steps of the SARS-CoV-2 life cycle can be targeted by siRNA and how this would affect viral replication.

siRNA design and synthesis
The siRNAs used in the study were designed using siDirect (http://sidirect2.rnai.jp)and siSPOTR (https://sispotr.icts.uiowa.edu/sispotr/index.html)online tools (28-30), structured upon guidelines delineated from previous ndings (31)(32)(33)(34).We designed siRNAs against four regions of the SARS-CoV-2 genome: Leader sequence, ORF1, Nucleocapsid (N) and 3' untranslated region (3'UTR), based on the fulllength reference sequence available on RefSeq database (NCBI Accession number: NC_045512.2).The siRNAs were designed with an asymmetric design (28) and occasional G:U wobbles (35) for improved speci city (Supplementary Table S1).Two additional siRNAs targeting GFP and Fire y Luciferase were designed as negative controls.The luciferase-speci c siRNA served as control for infection experiments and the GFP-speci c siRNA for experiments in which Luciferase reporters were used.The siRNAs were purchased in desalted form (Microsynth AG, Balgach, Switzerland), resuspended and maintained in RNAse free water upon arrival.Activity of siRNAs was tested using luciferase reporters (see below) before proceeding to the SARS-CoV-2 infection models.

Cloning of luciferase reporters
Initial siRNA screenings, testing of siRNA strand-speci c activities and the competition assay (shown in Figure 4D) were performed using the dual luciferase expressing psiCHECK TM -2 vector (Promega GmbH; Walldorf, Germany) with siRNA target sites cloned into a multiple cloning site present downstream of the Renilla luciferase translational stop codon.For this, the full-length SARS-CoV-2 leader sequence, both negative and positive sense N sequence, as well as the individual siRNA binding sites of siRNAs targeting ORF1 and the 3'UTR were inserted by cloning via XhoI/NotI digestion (FastDigest TM , Thermo Fisher Scienti c; Dreieich, Germany).The individual siRNA binding sites and the whole Leader sequence were purchased as single-stranded DNA oligonucleotides, designed to form overhangs mimicking digested oligonucleotide fragments after annealing.For this, equal amounts of complementary oligonucleotides were mixed and heated at 95 °C for ve minutes followed by gradual cooling for 2h at 30 0 C to allow forming of oligonucleotide duplexes.These were directly used in a ligation reaction with the digested psiCHECK-2 TM vector.The positive sense N coding sequence was PCR ampli ed using primers E-N Fw BamHI and E-N Rev EcoRI from cDNA of SARS-CoV-2 infected VeroE6 cells and cloned into the pcDNA1/Amp plasmid vector.In a next step, the N coding sequence was PCR-ampli ed using primers N CDS Fw XhoI and N CDS Rev NotI and cloned into the luciferase reporter.The full-length negative sense N gene was purchased as desalted, pre-annealed double-stranded DNA oligonucleotide and used directly for the annealing reaction with digested psiCHECK TM -2 vector.A list of used oligonucleotides is given in Supplementary Table S2.
Transfection siRNAs were either transfected 6h before or 3h after SARS-CoV-2 infection using Lipofectamine RNAiMAX (Thermo Fisher Scienti c; Dreieich, Germany).A nal concentration of 50nM siRNA was used per well.For transfections before SARS-CoV-2 infection, a reverse-transfection protocol was used.Whereas, for transfections after virus infection, a conventional forward-transfection protocol was employed -both according to the manufacturer's instructions.All transfection experiments were performed with at least three biological replicates.For the pre-selection of siRNAs, the determination of strand speci c activities of N-speci c siRNAs, and the competition assay, siRNAs were co-transfected together with respective plasmid expressing a luciferase reporter.In brief, 200ng of reporter plasmid and 6pmol of siRNA were mixed with 1µl of transfection reagent (Lipofectamine 2000, Thermo Fisher Scienti c; Dreieich, Germany) diluted with Opti-MEM to a nal volume of 100µl.siRNA-plasmid containing transfection complexes were added on top of con uent cells, resulting in 10nM nal concentration of siRNA per well.For the prescreening of siRNAs and the determination of strand speci c activities of N-speci c siRNAs, constructs were transfected into 85-90% con uent HEK293T cells and for the competition assay into con uent VeroE6 cells.

Dual-Luciferase based reporter assay and competition experiment
As a surrogate model to determine siRNA activity, siRNAs were co-transfected into cells with plasmids expressing dual luciferase reporters.After co-transfecting siRNAs and plasmids (for details see paragraph above), cells were lysed after 48h (siRNA prescreening and strand speci c activities) with 100µl Passive Lysis Buffer (Promega GmbH; Walldorf, Germany), and luciferase activity from 10µl cell lysate measured using the Dual Luciferase® Reporter Assay System (Promega GmbH; Walldorf, Germany) according to instructions using a Tecan In nite 200 PRO Microplate reader (Tecan Group Ltd.; Männedorf, Switzerland).Relative activity of Renilla luciferase (normalized to Fire y luciferase activity as an internal transfection control) was indicated as silencing e ciency of the siRNA and compared with a control siRNA targeting GFP or Luciferase.For the competition experiment (shown in Fig. 4D-E), siRNAs and the respective luciferase reporter plasmid were co-transfected into VeroE6 cells as described previously, which were 6h later infected with wildtype SARS-CoV-2 (MOI 0.1) and 24h later luciferase activity and knockdown e cacy were determined.

SARS-CoV-2 infection
VeroE6 cells were seeded in 24-well format at least 6h before infection to gain approximately 90-95% con uency at time of infection.The SARS-CoV-2 stock was pre-diluted for a multiple of infection (MOI) of 0.1 in 100µl media.At time of infection, old growth media was removed, and the pre-diluted SARS-CoV-2 solution added to cells.After 1h incubation at 37°C, a medium exchange was performed.Different termination time points were performed from 1 to 24h post infection based on the investigated step of the viral replication cycle.The SARS-CoV-2 wildtype virus used in this study was isolated from a patient in March 2020 at Institute of Virology, TU Munich.The full-length sequence was uploaded onto GISAID database (https://www.gisaid.org/)under name hCoV-19/Germany/BAV-PL-virotum-nacq/2020 and accession ID: EPI_ISL_582134.
Real-time monitoring of virus spread using rSARS-CoV-2-GFP and automated uorescence analysis with the IncuCyte® Live-Cell Analysis VeroE6 cells in growth media were seeded at least 6h before infection into 96-well plates to gain approximated 90-95% con uency at time of infection.Cell were then infected with a recombinant SARS-CoV-2, expressing Green Fluorescent Protein (GFP) from a sequence integrated at the ORF7 locus (rSARS-CoV-2-GFP) (36).For this, the rSARS-CoV-2-GFP virus infection solution was pre-diluted for a MOI of 1 in 50µl growth media.After adding 50µl of the infection solution to cells, media was exchanged after 1h and multi-well plates placed into IncuCyte® Live-Cell Analysis machine and phase contrast as well as uorescence pictures of the whole well acquired every 4h for three days.Total cell number was determined via phase contrast microscopy and infected cell population using the GFP channel using the IncuCyte S3 software (Essen Bioscience; version 2019B Rev2).

Determination of cell viability
The cell viability was determined using the CellTiter-Blue Cell Viability Assay kit (Promega GmbH, Walldorf, Germany) according to manufacturer's instructions.In brief, VeroE6 cells were reversely transfected in 96-well plate with siRNAs 6h before SARS-CoV-2 infection (MOI 1), and cell viability assessed at different time points.For this, CellTiter-Blue reagent was diluted 1:5 with culture medium and applied to cells for 60 min at 37 °C.A distinct color change was observed in the untreated controls in comparison to the empty well controls.Conversion from resazurin to resoru n was analysed with uorescence lters 550 / 590 nm at a Tecan In nite F200 (Tecan Group Ltd.; Männedorf, Switzerland).
Nucleic Acid extraction and qPCR RNA from cultured cells was extracted with the NucleoSpin RNA kit (Macherey-Nagel; Düren, Germany), and cDNA synthesized with the Superscript TM III First-Strand Synthesis System (Thermo Fisher Scienti c; Dreieich, Germany) according to manufacturer's instructions.SARS-CoV-2 transcripts were ampli ed in subsequent qPCR using primers speci c to the N region, essentially covering all the viral transcripts or the RNA dependent RNA polymerase (Rdrp) region, as a measure of gRNA.For quanti cation of viral RNAs, a standard curve was constructed using plasmids with integrated Rdrp or N sequences.Amount of sgRNA was calculated by subtracting the result of PCR reaction using Rdrp primers from the one using N primers (as full-length gRNA is also detected by the N primers).18S rRNA was used as a reference gene for relative quanti cation.All quantitative PCRs were performed on a LightCycler® 480 (Roche Holding AG; Basel, Switzerland) using primers and cycling conditions shown in Table 1.

Strand-speci c cDNA synthesis
To individually determine negative or positive sense SARS-CoV-2 RNA, we speci cally transcribed RNA of a certain polarity to cDNA.For this, rst strand synthesis was performed from total RNA extracts using the SuperScript TM IV First-Strand Synthesis System (Thermo Fisher Scienti c; Dreieich, Germany) with primers speci c either for positive sense mRNA (Oligo(dT) 20 primers) or negative sense mRNA (Oligo(dA) 20 primers).To allow transcription of a house keeping gene also in the reaction transcribing negative sense RNA, primers speci c for the 18S rRNA gene (18S cDNA1-3; Table 1) were added to the reaction.A nal concentration of 50 µM for all primers combined were used for rst strand synthesis reaction and viral RNAs quanti ed by qPCR as described above.

Results
Targeting the genome of incoming SARS-CoV-2 terminates replication before start of transcription and prevents cytopathy Following the events in viral replication cycle, rst, we investigated whether siRNAs can directly target the incoming genome of SARS-CoV-2 after cell entry.We chose ORF1 as target region, as it is only contained in full-length genomic, but not sgRNAs (23).We individually transfected three siRNAs which were active in previous luciferase reporter screens (Supplementary Figure S2A) into VeroE6 cells.After 16h, cells were infected with the GFP-expressing rSARS-CoV-2-GFP vector and viral infection and spread monitored by quantifying GFP + cells every 4h over the course of three days (Figure 1A, top).As the ORF1-speci c siRNAs do not target the transcript from which GFP is expressed, a suppression of GFP expression would indicate that siRNAs targeted full-length gRNA (Figure 1A, bottom).Indeed, we found the number of GFP + cells reduced to approximately 50% by each of the tested siRNAs (Figure 1B-D).Importantly, this difference was already present at the earliest time point (12h post infection [p.i.]) with detectable GFP signal (Figure 1B), indicating that genomes of incoming virus were successfully targeted.We con rmed this by repeating the experiment using wildtype SARS-CoV-2 but lysed the cells 24h p.i. and quanti ed SARS-CoV-2 gRNA by reverse-transcriptase quantitative PCR (RT-qPCR).As indicated by our rst experiment, gRNA was reduced in groups pre-treated with the ORF1-speci c siRNAs (Figure 1E).To further con rm that indeed genomes of incoming virus were degraded, we transfected cells with a pool of three ORF1-speci c siRNAs 6h before infection with wildtype SARS-CoV-2 and quanti ed intracellular viral RNAs at different time points.Viral RNAs were further differentiated into full-length gRNA and sgRNAs (see methods for details).We found that gRNA was reduced as early as 1h p.i. (Figure 1F), before sgRNAs were synthesized (Figure 1G).Treatment with ORF1-speci c siRNAs prevented sgRNA expression (Figure 1G), reduced morphologic changes (Figure 1H), and improved metabolic rate of cells (Figure 1I).Taken together, our data proves that siRNAs can target the genome of SARS-CoV-2 and terminate viral replication at an early replication step preventing cytopathy.
Negative sense SARS-CoV-2 RNAs are not accessible for siRNA therapy Currently it is unclear if both, negative and positive sense coronaviral RNA, or only RNA with a certain polarity is accessible for RNAi silencing.This question is particularly interesting when designing therapeutic siRNAs, as potentially both strands of the siRNA could convey antiviral activity.To gain a more detailed understanding on the kinetic of RNA synthesis during SARS-CoV-2 replication, we lysed wildtype SARS-CoV-2-infected VeroE6 cells at different time points.Positive and negative sense viral RNAs were individually quanti ed by strand speci c rst strand synthesis (see Methods).Negative sense gRNA was detected in low quantities already 1h p.i., but strongly increased at 6h p.i. when it was more abundant than positive sense gRNA (Figure 2A).In contrast, sgRNAs started to appear only at 6h p.i. (Figure 2B).Consistent with other coronaviruses, lower amounts of negative sense sgRNAs were detected as compared to their positive sense counterparts (37).We now investigated whether negative sense SARS-CoV-2 RNA is accessible for RNAi-mediated silencing.We developed siRNAs that speci cally targeted either negative or positive sense SARS-CoV-2 RNA.We chose the N ORF as target region, as it is also part of sgRNAs which are -in contrast to gRNA -exported from ROs (26) and should therefore be easily accessible for siRNAs.siRNA strand-speci c activity was validated by co-transfecting siRNAs with reporter plasmids expressing either the positive or negative sense N coding sequence incorporated downstream to Renilla luciferase gene (see scheme in Figure 2C and methods).The majority of siRNAs showed high selectivity against the RNA strand they were designed (Supplementary Figure S1).We chose siRNAs with almost exclusive activity against either the positive or negative sense reporter (Figure 2D).To our surprise, only siRNAs active against positive sense N ORF succeeded in reducing sgRNA of wildtype SARS-CoV-2 (Figure 2E), thereby inhibiting viral spread in the rSARS-CoV-2-GFP model (Figure 2F).In summary, our data proves that negative sense SARS-CoV-2 RNAs are inaccessible for RNAi.
Targeting common regions of SARS-CoV-2 transcripts allows simultaneous suppression of gRNA and sgRNAs, but leads to loss of e cacy We further went on to investigate whether targeting the common regions shared by all SARS-CoV-2 transcripts (L, N ORF and 3´UTR; Figure 3A) would allow simultaneous suppression of gRNA as well as sgRNAs.In contrast to inceptive experimental setup, we infected VeroE6 cells with wildtype SARS-CoV-2 and afterwards transfected siRNAs targeting either the common regions of transcripts (shown in red) or ORF1 (blue) which is contained only in gRNA.For each target region, we used a pool of three siRNAs which previously showed similar activities against luciferase reporters (Supplementary Figure S2A).As expected, ORF1-speci c siRNAs suppressed only gRNA, whereas siRNAs targeting common regions of transcripts suppressed gRNA and sgRNAs (Figure 3B).We next investigated how targeting sgRNAs in addition to gRNA would affect antiviral e cacy.Therefore, we infected VeroE6 cells with rSARS-CoV-2-GFP and 3h later, individually transfected three siRNAs for each target region.Viral spread was evaluated by quantifying GFP + cells every 4h over a course of three days.All tested siRNAs reduced GFP + cells compared to the control groups.siRNAs targeting gRNA and sgRNAs strongly suppressed GFP expression until 40h p.i. (as they also targeted the GFP-transcript; Supplementary Figure 2B, C) but then, the number of GFP + cells increased exponentially (Figure 3C and Supplementary Figure 2D, E).In contrast, groups treated with ORF1-speci c siRNAs inhibited viral spread more e ciently, showing a linear and generally slower increase of GFP + cells (Figure 3C).In summary, our data indeed showed a concurrent suppression of genomic and subgenomic viral RNAs by siRNAs targeting the common regions in the viral transcripts.
On the other hand, the ORF1 speci c siRNAs which solely target SARS-CoV-2 gRNA, subdue viral replication more e ciently as compared to siRNAs that additionally targeted sgRNAs.

Targeting ORF1 with siRNAs prevents RNA transcription and out-competition by sgRNAs
To investigate which replication step was more e ciently targeted by ORF1-speci c siRNAs, we compared transfection of siRNAs before and after infection with wildtype SARS-CoV-2 and assessed viral RNA knockdown after 24h.While ORF1-speci c siRNAs showed a trend towards slightly stronger e cacy compared to siRNAs targeting common regions when transfected after infection, the difference was signi cantly enhanced when siRNAs were transfected before infection (Supplementary Figure S3).This indicated that ORF1-speci c siRNAs had an advantage while targeting an early replication step of SARS-CoV-2.To further pinpoint which replication step was primarily affected, we transfected VeroE6 cells 6h before infection with wildtype SARS-CoV-2 and analyzed the effect on viral RNAs at different time points.Indeed, ORF1-speci c siRNAs showed an enhanced knockdown of gRNA already 1h p.i. (Figure 4A) leading to decreased transcription of negative sense gRNA (Figure 4B) and sgRNAs (Figure 4C).Taken together, our data demonstrates that ORF1-speci c siRNAs more e ciently targeted SARS-CoV-2 gRNA, ergo, more e ciently prevented transcription, than siRNAs targeting common regions of transcripts.
We hypothesized, that a second reason for the decreased e cacy of siRNAs targeting gRNA as well as sgRNA could be the high abundance of sgRNAs outcompeted siRNAs and/or RNA-induced Silencing Complexes (RISC).To test this hypothesis, we co-transfected siRNAs targeting the different SARS-CoV-2 regions together with their respective luciferase reporter.After 6h, we infected cells with wildtype SARS-CoV-2 and analyzed how SARS-CoV-2 replication in cells would affect silencing of the luciferase reporter (Figure 4D).Of note, in this experimental setting, both SARS-CoV-2 RNAs and mRNA transcribed from a luciferase reporter plasmid can be targeted by the respective siRNAs.Indeed, we found that silencing of luciferase reporters by siRNAs which targeted gRNA and sgRNAs was signi cantly impaired by SARS-CoV-2 replication.This was not observed for the ORF1-speci c siRNA which suppressed the luciferase reporter with same e cacy in both, infected as well as non-infected cells (Figure 4E).In summary, our data show that two mechanisms, a more e cient targeting of viral genomes by ORF1-speci c siRNAs, as well as an impaired RNAi silencing affecting siRNAs targeting additionally sgRNAs, are responsible for the observed differences of siRNA activities.

Discussion
A promising approach to develop antiviral therapies against SARS-CoV-2 constitute siRNAs, which is pursued by several academic and industry groups (38,39).We provide the rst proof that SARS-CoV-2 replication and spread can be e ciently controlled using siRNAs.We found that siRNAs, when given in a prophylactic setting, can target the genome of SARS-CoV-2 after cell entry and terminate replication before start of transcription.To our surprise, targeting solely gRNA resulted in a stronger antiviral e cacy than a simultaneous targeting of gRNA and sgRNA.We show that two independent mechanisms were responsible for this.First, ORF1-speci c siRNAs more e ciently targeted genomes of SARS-CoV-2 before start of transcription.As ORF1 is the only translational active region before formation of RTC and ROs, the other genomic regions were probably less accessible for the RNAi machinery, due to secondary RNA structures (40) or as they are shielded by nucleocapsid proteins (41).Second, the impaired RNAi silencing affecting siRNAs targeting gRNA and sgRNAs could result from the high abundance of sgRNAs outcompeting siRNAs and/or RISC complexes.This notion appears especially plausible as Kim et al.
showed that roughly 2/3 rd of the transcriptome of infected cells are made up of SARS-CoV-2 RNAs of which almost all contain the targeted sequences (23).Our ndings on a rst look might contradict a previous report which described that targeting the leader sequence of SARS-CoV-1 with siRNAs would be more e cacious than targeting the S ORF (13).Several factors could explain differences found in our study.First of all, SARS-CoV-1, which -while being the closest related virus -has an amino acid sequence homology of only between 40-94% depending on the ORF (42), thus ndings might not be applicable to SARS-CoV-2.Also, Li et al. compared an siRNA targeting the Leader sequence to siRNAs targeting the S gene, but not to ORF1-speci c siRNAs.As we speculate that one possible mechanism for the enhanced e cacy of ORF1 siRNA stems from ORF1 being the only translational active region before start of transcription, our nding might not apply to siRNAs targeting the S ORF.Third, Li et al. compared only a single Leader-speci c siRNA to two S-speci c siRNAs questioning if the nding can be generalized to the target region or if intra-individual differences of siRNA activity were responsible for the observed differences.Regarding the reason why negative sense SARS-CoV-2 RNA was not accessible for siRNAsilencing, the most plausible explanation to us is that negative sense RNA -in contrast to positive sense RNAs -are not exported from ROs.As probably no RISC complexes are present within the ROs, activity of siRNA is restricted to RNAs which have either not yet entered ROs (gRNA of incoming virus) or positive sense sgRNAs which are exported from ROs for translation.This is in line with the notion that negative sense RNAs are not translated to proteins, creating no necessity to export them into cytosol.
Several factors that have to be considered for clinical application were not addressed in our study.During siRNA design, we did not take the conservation of target regions into account which could affect occurrence of resistance mutations under therapy.However, as ORF1 shows a relatively high conservation compared to other regions of the SARS-CoV-2 genome (43), it appears plausible that an ORF1-speci c siRNA drug candidate with a high resistance barrier could be developed.Also, we did neither chemically modify our siRNAs -which might further enhance activity, stability and tolerability of siRNAs -nor did we use advanced delivery methods which would enhance e cacy an allow clinical application (44)(45)(46)(47).These questions should be addressed in further studies to allow development of potent siRNA-based therapies.Especially the identi cation of the optimal carrier and administration route is an important factor.While pulmonary delivery can be achieved by intranasal (i.n.) or inhalation administration, i.n.administration was chosen as the delivery route in Alnylam's early attempts of delivering siRNA against RSV (48).The big advantage of i.n.delivery is the possibility of administering a liquid formulation as nose drops without the need of nebulizing the formulation.This is particularly of impact for liposomal formulations as they do not withstand shear forces and temperature-related stress during nebulization (49).The biggest disadvantage of i.n.administration, however, is the low pulmonary bioavailability of the administered dose, while a large proportion is swallowed and degraded (50).
Inhalation delivery, in contrast, requires aerosol development of a mist or dry powder.For nebulization of macromolecules such as siRNA, vibrating mesh nebulizers are preferred for decreased effects on biomolecule stability (45).Dry powder inhalation offers the advantages of long shelf-lives and enhanced stability of nucleic acid formulations against chemical, physical and microbial factors (46) but faces engineering challenges when nucleic acids nanoformulations need to be transferred into dry powders (51).Such nanoformulations are, however, particularly important for pulmonary delivery where free nucleic acids do not e ciently diffuse through the mucus barrier for subsequent uptake into the epithelium (52).Numerous siRNA nanoformulations exist based on polymers, lipids, peptides and inorganic materials (53), each of which can be improved in e ciency and speci city with different surface modi cations such as targeting ligands or membrane-active substances (47).
To our knowledge, there is so far no equally detailed analysis of RNAi-targetable replication steps and RNA species for any positive sense RNA virus.Thus, our results might also be of relevance beyond SARS-CoV-2.The reduction of cytopathic effects achieved by antiviral siRNAs could play out crucial, as endothelial injury has been proposed to trigger pathology in lethal COVID-19 cases (4).Along this line, early therapy starts, or possibly even prophylactic application of siRNAs appears as major bene t.
Especially in this setting, targeting ORF1 was advantageous over targeting sequences contained also in sgRNAs which could be a valuable information for designing siRNAs and treatment regimens in clinical studies.Taken together, our study con rms that siRNA-based strategies could allow to develop potent antivirals to reduce pathology of COVID-19, encouraging academia and industry to proceed with ongoing efforts. Figures

4 )
Table1.Oligonucleotides and cycling conditions used during polymerase change reaction.A = Adenine; C = Cytosine; G = Guanine; T = Thymine; Rev = reverse; min = minute; s = second; RDRP = RNA-dependent RNA polymerase Statistical Analysis Statistical analysis was performed with GraphPad Prism (version 8.4.3) for Mac.Normally distributed samples were analyzed using student T-test for independent samples when comparing two groups and with One-way Anova with Dunnett´s multiple comparison correction when comparing three or more groups.Statistical differences of non-normally distributed data were calculated for two groups using Mann-Whitney U or Kruskal-Wallis with Dunn´s multiple comparison correction tests when comparing three or more groups.p-values <0.05 were considered signi cant.