key: cord-266520-n439dwcx
authors: Levanova, Alesia A.; Kalke, Kiira M.; Lund, Liisa M.; Sipari, Nina; Sadeghi, Mohammadreza; Nyman, Marie C.; Paavilainen, Henrik; Hukkanen, Veijo; Poranen, Minna M.
title: Enzymatically synthesized 2'-fluoro-modified Dicer-substrate siRNA swarms against herpes simplex virus demonstrate enhanced antiviral efficacy and low cytotoxicity
date: 2020-08-13
journal: Antiviral Res
DOI: 10.1016/j.antiviral.2020.104916
sha: 
doc_id: 266520
cord_uid: n439dwcx

Chemical modifications of small interfering (si)RNAs are used to enhance their stability and potency, and to reduce possible off-target effects, including immunogenicity. We have earlier introduced highly effective antiviral siRNA swarms against herpes simplex virus (HSV), targeting 653 bp of the essential UL29 viral gene. Here, we report a method for enzymatic production and antiviral use of 2’-fluoro-modified siRNA swarms. Utilizing the RNA-dependent RNA polymerase from bacteriophage phi6, we produced 2’-F-siRNA swarms containing either all or a fraction of modified adenosine, cytidine or uridine residues in the antisense strand of the UL29 target. The siRNA containing modified pyrimidines demonstrated high resistance to RNase A and the antiviral potency of all the UL29-specific 2’-F-siRNA swarms was 100-fold in comparison with the unmodified counterpart, without additional cytotoxicity. Modest stimulation of innate immunity signaling, including induced expression of both type I and type III interferons, as well as interferon-stimulated gene 54, by 2’-F-cytidine and 2’-F-uridine modified siRNA swarms occurred at early time points after transfection while the 2’-F-adenosine-containing siRNA was similar to the unmodified antiviral siRNA swarm in this respect. The antiviral efficacy of the 2’-F-siRNA swarms and the elicited cellular innate responses did not correlate suggesting that innate immunity pathways do not significantly contribute to the observed enhanced antiviral activity of the modified siRNAs. The results support further applications of enzymatically produced siRNA molecules with incorporated adenosine nucleotides, carrying fluoro-modification on ribose C2’ position, for further antiviral studies in vitro and in vivo.

RNA interference (RNAi) is a natural antiviral defense mechanism in plants, fungi, invertebrates, and, under specific conditions, in mammals Schuster et al., 2019) . Antiviral

RNAi comes about via a sequence-specific binding of small interfering RNAs (siRNAs) to viral mRNAs or genomic RNAs, resulting in their degradation or translational repression. In general, siRNA is prone to fast degradation with RNases, and the delivery to the target cells might be difficult. A number of chemical modifications introduced to different positions in the siRNA (sense, antisense, or both strands) have been tested to improve siRNA stability and reduce offtarget activities (Braasch et al., 2003; Choung et al., 2006; Czauderna et al., 2003; Fedorov et al., 2006; Jackson et al., 2006; Watts et al., 2008; Hassler et al., 2018) .

Instead of chemical synthesis and post-synthesis hybridization of RNA oligonucleotides, we produce high-quality molecules for RNAi using viral RNA polymerases and recombinant

Giardia Dicer enzyme (Levanova and Poranen, 2018) . This results in a swarm of 25-bp-long Dicer-substrate siRNAs (DsiRNAs) (MacRae et al., 2006; Romanovskaya et al., 2012) , which can cover several kbp sequence of the target (Nygardas et al., 2009; Niehl et al., 2017; Jiang et al., 2019) . When introduced into a cell, such siRNAs are processed by the endogenous Dicer, which enhances the RNAi potency and efficacy (Kim et al., 2005) . In the enzymatically synthesized siRNA swarms, each individual siRNA species is present only in low concentrations, diluting out the potential off-target effects (Levanova and Poranen, 2018) . The siRNA swarm contains predominantly 5'-monophosphates as only the single siRNAs originating from the very ends of the produced dsRNA molecules contain the 5'-triphosphate moiety, which could induce innate immunity responses. The swarm, comprising of numerous different siRNAs against the J o u r n a l P r e -p r o o f target, also widens the antiviral spectrum (Paavilainen et al., 2016; Jiang et al., 2019) and minimizes the potential of emergence of viral variants resistant to a single siRNA.

We have previously generated three antiviral siRNA swarms against herpes simplex virus type 1 (HSV-1) mRNAs encoding essential viral proteins, including glycoprotein B (UL27), the infected cell protein 8 (ICP8; UL29), and ICP27 (UL54) (Romanovskaya et al., 2012; Paavilainen et al., 2016) . The siRNA swarm targeting mRNA of HSV single-stranded DNA binding protein ICP8, encoded by the UL29 gene, harbors a most pronounced antiviral effect (Paavilainen et al., 2016) and induces only minimal non-specific cellular responses (Romanovskaya et al., 2012) . The HSV UL29-derived siRNA swarm elicits modest IFN-λ1 (IL-29), IFN-λ2/3, ISG54, and TLR3 expression . However, the observed slight activation of innate immune responses did not cause a decrease in cell viability (Romanovskaya et al., 2012; Paavilainen et al., 2015) . The anti-HSV-UL29 siRNA swarm also controlled HSV infection in mice and inhibited local virus replication in corneal epithelia in vivo (Paavilainen et al., 2017) .

Here, we set up an enzymatic production method for modified DsiRNAs using T7 DNAdependent RNA polymerase (DdRp), phi6 RNA-dependent RNA polymerase (RdRp) and

Giardia Dicer. Using this method, we produced UL29-specific siRNA swarms harboring 2'fluoro-modifications in the ribose moieties of uridine, adenosine or cytidine residues. The incorporated fluoro-modifications generally improved siRNA stability. The 2'-F-siRNA swarms were well tolerated by cells and had enhanced HSV-specific antiviral activity compared to unmodified siRNA swarms.

J o u r n a l P r e -p r o o f

The 653 bp sequence of the UL29 gene from HSV-1 prototype strain (17+) (GenBank JN555585.1, nucleotides 59,302 to 59,954) was PCR amplified to produce a DNA template for ssRNA production (Romanovskaya et al., 2012) Dicer (Romanovskaya et al., 2012; Paavilainen et al., 2017) . The siRNA molecules were purified by anion-exchange chromatography on a CIMac QA column (BIA Separations, Slovenia, 110.5113-1.3) using an ÄKTApurifier FPLC system (GE Healthcare) (Romanovskaya et al., 2013) at the Instruct-HiLIFE Biocomplex unit (University of Helsinki). The same protocol was used to produce non-specific unmodified control siRNA swarms from a 413 bp long sequence of Escherichia coli lacI gene in pET32b vector using primers: Fwd_pET32b_1581: 5'-TAATACGACTCACTATAGGGCTGCCTGCACTAATGTTCC-3' and Rev_pET32b_1983: 5'-GGAAAAAAATCGTCGTATCCCACTACC-3'. As confirmed with BLASTn, the selected J o u r n a l P r e -p r o o f control sequence has no significant similarities with human, herpesviral or mouse transcriptomes. The 88 bp dsRNA was produced as described previously (Jiang et al., 2011) . ChemiDoc Touch Imaging System (Bio-Rad) and the band intensities measured using Fiji (Schindelin et al., 2012) . The relative amount of undegraded siRNA was calculated according to the formula: a%=(S RNaseA ×100)/S untreated , where S RNaseA is a density of RNaseA-treated siRNA and S untreated is a density of the equivalent amount of untreated sample.

An astrocytoma-glioma cell line (U373MG) obtained originally from ATCC (Manassas, VA) was used. The line has been subsequently reclassified as U251 human glioblastoma, but referred here as U373MG to ensure continuity with our previous studies (Paavilainen et al., , 2016 Romanovskaya et al., 2012) . U373MG cells were maintained in DMEM with 10% heat J o u r n a l P r e -p r o o f inactivated fetal bovine serum (FBS) and 2 mM L-glutamine at 37˚C, 5% CO 2 . Vero cells (ATCC), used for plaque titration, were maintained in M199 medium with 5% FBS.

Green fluorescent protein (GFP)-expressing HSV-1 strain, HSV-1(17+)Lox-P mCMV GFP (abbreviated here as HSV-1-GFP), originally received from prof. Beate Sodeik (MHH Hannover Medical School, Germany), was used for antiviral studies (Snijder et al., 2012; Mattila et al., 2015) and propagated as previously described (Romanovskaya et al., 2012) .

The U373MG cells were transfected on 96-well plates with either modified or unmodified siRNA swarms, 88 bp dsRNA or water (mock transfection) using Lipofectamine RNAiMAX (#13778-150; Invitrogen, Carlsbad, CA) according to the manufacturer's forward transfection protocol (Romanovskaya et al., 2012) . Depending on the experiment, the cells were transfected with 1-10 pmols of RNA per well, which is equivalent to a concentration of 10-100 nM. All experiments were repeated at least twice with three or more biological replicates each.

Cellular viabilities in siRNA treatments were assessed 48 hours post transfection (hpt) with

CellTiter-Glo (#G7571; Promega, Madison, WI) luminescent assay (Romanovskaya et al., 2012; Turunen et al., 2016) . The luminescence was quantified with VICTOR Nivo Multimode Plate Reader (Perkin Elmer, Waltham, MA). Viability of more than 80% compared to mock-and untreated cells was considered acceptable.

J o u r n a l P r e -p r o o f

The total cellular RNA was extracted from siRNA-treated U373MG cells at 8, 24, and 48 hpt with TRI Reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. The RNA was processed to complementary DNA using RevertAid H Reverse Transcriptase (Thermo Fisher Scientific, Waltham, MA) with random hexamer primers (Thermo Fisher Scientific) after treatment with DNAse (Thermo Fisher Scientific) (Romanovskaya et al., 2012; Paavilainen et al., , 2016 .

Quantitative PCR (qPCR) was performed with Rotor-Gene Q, as previously described (Nygardas et al., 2011) . The GAPDH housekeeping mRNA values of the corresponding sample were used for normalization of mRNA expression levels. The mRNA expression of interferon beta (IFN-β) (Peri et al., 2008) , lambda 1 (IL-29; IFN-λ1) (Paavilainen et al., 2016) , interferon stimulated gene 54 (ISG54) (Romanovskaya et al., 2012) , and GAPDH (Nygardas et al., 2009 ) was quantified using previously published primer sequences.

The RNA-or control-treated cells were challenged with a clinically relevant infectious dose of HSV-1-GFP at 1000 plaque forming units ( 

We tested the capability of phi6 RdRp to utilize 2'-F-dNTPs and 2'-OMe-NTPs in the dsRNA synthesis. HSV UL29 gene-specific ssRNA, produced using T7 DdRp, was used as a template for the phi6 RdRp-catalyzed dsRNA synthesis. Initially, the reactions were done by replacing one of the canonical NTPs in the reaction mixture with the corresponding modified NTP.

Although dsRNA synthesis was not detected in the presence of 2'-OMe-modified NTPs (Fig. S1 ) and the quality of the dsRNA product was somewhat compromized in the presence of 2'-F-dGTP ( Fig. S2A) , phi6 RdRp was able to produce full-length dsRNA products using 2'-F-dATP, 2'-F-dUTP, and 2'-F-dCTP substrates (Fig. 1A) , suggesting that the 113 adenosines, 128 uridines, and 185 cytidines in the antisense strand of the HSV UL29 target sequence can potentially be substituted with their fluoro-modified counterparts. Wild-type phi6 RdRp and a mutant RdRp J o u r n a l P r e -p r o o f containing substitution K→A at position 147 (K147A) were initially tested for dsRNA production. The K147A RdRp consistently provided higher dsRNA yields (Fig. S2B) and was therefore used for the subsequent siRNA production. Complete replacement of a canonical rNTP with the corresponding 2'-F-dNTP resulted in slightly decreased yields, which was especially pronounced when all three NTPs (ATP, UTP and CTP) carried the modification (Fig. 1A) .

Therefore, to produce siRNA swarms for the cellular experiments we used 2'-F-dATP, 2'-F-dCTP, or 2'-F-dUTP, but not their mixture. Furthermore, we set up a dsRNA synthesis reaction in which only a fraction (10%) of a given NTP was fluoro-modified (Fig. S2 ). The mass spectrometry analysis confirmed that the "10% modified" dsRNA products contained 2'-fluoromodified nucleotides (Fig. S3) . Furthermore, the Giardia Dicer could process the produced UL29-2'-F-dsRNA into a swarm of Dicer-substrate 2'-F-siRNAs (Fig. 1B) .

RNase A treatment in low salinity conditions was performed to assess sensitivity of the 2'-F-siRNA swarms to RNase. Under the conditions used, about 50% of the unmodified siRNA was degraded ( Fig. 1C) , whereas fluoro-modifications in the uridine and cytidine residues significantly enhanced the stability of the siRNAs.

Cellular toxicity of the 2'-F-siRNA swarms was assessed using U373MG cells in 96-well plates.

Cell were transfected with each of the HSV-specific UL29-gene derived-2'-F-siRNA swarms, an unmodified UL29-siRNA swarm, a non-specific siRNA swarm derived from E. coli lacI gene, a J o u r n a l P r e -p r o o f cytotoxic 88-bp-long control dsRNA, or water. The viability of the cells was assessed 48 hpt with luminescent assay (Fig. 2) . Transfections with the 2'-F-siRNA swarms resulted in similar levels of cellular viability to that with unmodified UL29-siRNA swarms, with the exception of the fully 2'-F-dCTP modified swarm, which was significantly less tolerated than the unmodified siRNA swarm at 50 and 100 nM. Nevertheless, all the 2'-F-siRNA swarms were well tolerated, with relative cellular viability consistently higher than 80% at doses relevant for antiviral applications. Hence, 2'-F-modifications had no effect on tolerability of the siRNA swarms in the studied concentration range. The UL29-targeting siRNA swarms did not display increased cytotoxicity in comparison to the non-specific siRNA swarm, targeting a non-relevant bacterial sequence. The 88 bp dsRNA used as a toxic control resulted in a clear decrease in cell viability at 10 nM concentration (Fig. 2) . The cytotoxicity profile of the modified siRNA swarms was confirmed with non-cancerous human corneal epithelial cells, in which only the fully 2'-F-dCTP modified swarm was significantly less tolerated than the unmodified siRNA swarm (Fig. S4) .

The antiviral efficacy of the UL29-2'-F-siRNA swarms was quantified at 48 hpt by plaque titration from culture supernatants of HSV-1-GFP-infected U373MG cells. Before infection, cells were treated with HSV-targeted modified or unmodified UL29-siRNA swarms, a nonspecific siRNA swarm, water (mock-treated), or left untreated (Fig. 3) . After treatment with the unmodified UL29 siRNA swarm, the viral titer was reduced at least three orders of magnitude compared to the untreated samples (Fig. 3A) . The antiviral effect of the UL29-2'-F-siRNA swarms was even more pronounced, resulting in a five orders of magnitude reduction in HSV J o u r n a l P r e -p r o o f titer (Fig. 3B ). Compared to treatment with the non-specific siRNA swarm, treatment with the modified and unmodified UL29 siRNA swarms resulted in a significant decrease of titer (p = 0.00000000165 and p = 0.000000178, respectively), confirming the sequence specificity of the antiviral effects (Fig. 3A) . Furthermore, treatment with the UL29-2'-F-siRNA swarms in comparison with the unmodified counterparts resulted in significantly lower titers (p = 0.018) ( Fig. 3A) and an extensive decrease of viral shedding (p = 0.000001) (Fig. 3C) . The difference in antiviral efficacy between the 10% and 100% modified swarms was non-significant (Fig. 3A) .

Nonetheless, the 2'-F-siRNA swarms harboring modifications in cytidine or adenosine residues, i.e. 2'-F-dCMP-siRNA and 2'-F-dAMP-siRNA swarms respectively, yielded significantly higher antiviral potency compared to those having uridine modifications (2'-F-dUMP-siRNA), when using the unmodified UL29-siRNA swarm as a reference (Fig. 3B ). The 10% modified 2'-F-dUMP-siRNA swarm was the only UL29-2'-F-siRNA swarm that did not show improved antiviral efficacy in comparison with the unmodified UL29-swarm (Fig. 3B ).

The efficacy of the HSV-targeted siRNA swarms against HSV-1-GFP infection on U373MG cells was visualized by living cell fluorescent imaging at 48 hpt (at 44 hpi; Fig. 3D ). The virusderived GFP signal was prominent in infected cells, untreated or control-treated, whereas in cells pretreated with unmodified UL29-siRNAs or UL29-2'-F-siRNAs, the GFP signal could not be detected using standard imaging parameters. 2'-F-dUMP-and 2'-F-dCMP-siRNAs induced higher IFN-β, IL-29 and ISG54 expression compared to unmodified UL29-siRNA swarm, particularly at the earliest time point (Fig. 4) .

Innate immunity signaling induced by 2'-F-dAMP-siRNA swarms and that by the unmodified UL29 siRNA were at equal levels, or even lower. Overall, the induction of interferon signaling by modified or unmodified siRNA swarms was not extensive (only ca. 10-fold compared to mock-and untreated cells).

We have previously reported a novel enzymatic approach to synthetize large pools of siRNA molecules, which we refer to as siRNA swarms. The antiviral siRNA swarms targeting the current target of choice, the UL29 gene of HSV-1, have previously demonstrated their efficacy against circulating pathogenic HSV strains in vitro and corneal infection in vivo, unveiling their potential as antiviral drugs. In the current study, we pursued the modification of the RNAs using nucleotides with 2'-modified ribose, in order to further enhance the antiviral potency and stability of the siRNA preparations. Eventually, we are able to report successful enzymatic incorporation of 2'-F-dNTP nucleotides to biologically active dsRNA molecules. All 2'-F-siRNA swarms were well tolerated and highly effective against HSV-1 in vitro. The observed J o u r n a l P r e -p r o o f enhanced antiviral efficacy of the 2'-F-siRNA swarms was unrelated to the elicited cellular innate immunity responses, proving the sequence specificity of the modified siRNAs.

We produced siRNA swarms, in which either all or a fraction of adenosines, cytidines or uridines were replaced by 2'-F-dNMPs. Replacement of a 2'-OH group of ribose with a fluorine has been shown to stabilize RNA molecules against nucleases (Monia et al., 1993) . SiRNAs are primarily degraded by RNase A (Turner et al., 2007) , which recognizes the 2'-OH of pyrimidine nucleosides for cleavage (Cuchillo et al., 2011) . Therefore, a common strategy is to modify all pyrimidines (Hassler et al., 2018) to increase the siRNA stability. Accordingly, substitution of the canonical CMPs or UMPs by the corresponding 2'-F-dNMPs resulted in a significant increase in siRNA stability in our experiments (Fig. 1C) .

Conventionally, modifications are introduced into RNA molecules using chemical solid phase synthesis. Some wild-type or mutant DdRps of autographiviruses (T7, Syn5) can be used for enzymatic synthesis of single-stranded 2'-F-RNAs (Sousa and Padilla, 1995; Zhu et al., 2015) which may be further hybridized to produce dsRNA. We demonstrate here that bacteriophage phi6 RdRp efficiently utilizes 2'-F-dNTPs allowing enzymatic production of perfectly duplexed dsRNAs without the often error-prone post-synthesis hybridization. Similarly, to other wild-type bacteriophage polymerases (Sousa and Padilla, 1995; Zhu et al., 2015) , phi6 RdRp did not tolerate 2'-OMe modifications (Fig. S2) .

All of the modified siRNA swarms were non-toxic for U373MG cells at the relevant antiviral concentrations. The findings were reproducible in human corneal epithelial cells (Fig. S4) , which represent another potential target tissue for antiviral therapy of diseases caused by HSV-1.

Furthermore, modified swarms had higher antiviral activity than the unmodified swarm, as demonstrated by the significant reduction of viral titer in plaque assays. Notably, there was no difference between siRNA swarms containing only a fraction of modified dNMPs and those containing all modified dCMPs, dUMPs, or dAMPs in the antisense strand (Fig. 3) . The antiviral effect of siRNAs containing only a fraction of modified dUMPs was not statistically different from the unmodified siRNA swarm. Interestingly, a higher stability to nucleases of siRNA swarms containing 2'-F-dUMPs or 2'-F-dCMPs did not translate into higher antiviral efficacy than those with 2'-F-dAMPs.

The immunostimulatory potential of 2'-F-dAMP-siRNAs resembled that of the unmodified siRNA swarm and was thus minimal. At the early time point (8 hpt 

The authors declare no conflict of interest.

Zhu, B., Hernandez, A., Tan, M., Wollenhaupt, J., Tabor, S., Richardson, C.C., 2015. Synthesis of 2'-Fluoro RNA by Syn5 RNA polymerase. Nucleic Acids Res. 43, e94. http://doi.org/10.1093/nar/gkv367 Since the number of CMPs in the sequence is significantly higher than that for AMPs or UMPs 

RNA interference in mammalian cells by chemically-modified RNA

Chemical modification of siRNAs to improve serum stability without loss of efficacy

Bovine pancreatic ribonuclease: fifty years of the first enzymatic reaction mechanism

Structural variations and stabilising modifications of synthetic siRNAs in mammalian cells

Off-target effects by siRNA can induce toxic phenotype

Small RNA-based antimicrobial immunity

Comparison of partially and fully chemically-modified siRNA in conjugate-mediated delivery in vivo

Positionspecific chemical modification of siRNAs reduces "off-target" transcript silencing

Innate immune responses in human monocyte-derived dendritic cells are highly dependent on the size and the 5' phosphorylation of RNA molecules

Efficient Inhibition of Avian and Seasonal Influenza A Viruses by a Virus-Specific Dicer-Substrate Small Interfering RNA Swarm in Human Monocyte-Derived Macrophages and Dendritic Cells

Synthetic dsRNA Dicer substrates enhance RNAi potency and efficacy

RNA Interference as a Prospective Tool for the Control of Human Viral Infections

Structural basis for double-stranded RNA processing by Dicer

An investigation of herpes simplex virus type 1 latency in a novel mouse dorsal root ganglion model suggests a role for ICP34.5 in reactivation

Evaluation of 2'-modified oligonucleotides containing 2'-deoxy gaps as antisense inhibitors of gene expression

Synthetic biology approach for plant protection using dsRNA

Treatment of experimental autoimmune encephalomyelitis in SJL/J mice with a replicative HSV-1 vector expressing interleukin-5

Inhibition of coxsackievirus B3 and related enteroviruses by antiviral short interfering RNA pools produced using phi6 RNA-dependent RNA polymerase

Inhibition of clinical pathogenic herpes simplex virus 1 strains with enzymatically created siRNA pools

Innate responses to small interfering RNA pools inhibiting herpes simplex virus infection in astrocytoid and epithelial cells

Topical treatment of herpes simplex virus infection with enzymatically created siRNA swarm

Herpes simplex virus type 1 Us3 gene deletion influences toll-like receptor responses in cultured monocytic cells

Enzymatically produced pools of canonical and Dicer-substrate siRNA molecules display comparable gene silencing and antiviral activities against herpes simplex virus

High-throughput purification of double-stranded RNA molecules using convective interaction media monolithic anion exchange columns

Fiji: an open-source platform for biological-image analysis

Antiviral RNAi in Insects and Mammals: Parallels and Differences. Viruses 11

Single-cell analysis of population context advances RNAi screening at multiple levels

A mutant T7 RNA polymerase as a DNA polymerase

MALDI-TOF mass spectral analysis of siRNA degradation in serum confirms an RNAse A-like activity

HSV-1 Infection Modulates the Radioresponse of a HPV16-positive Head and Neck Cancer Cell Line

Chemically modified siRNA: tools and applications

## ### ### ***