key: cord-0980359-wr4zrggo
authors: Zhu, Chi; Lee, Justin Y.; Woo, Jia Z.; Xu, Lei; Nguyenla, Xammy; Yamashiro, Livia H.; Ji, Fei; Biering, Scott B.; Van Dis, Erik; Gonzalez, Federico; Fox, Douglas; Rustagi, Arjun; Pinsky, Benjamin A.; Blish, Catherine A.; Chiu, Charles; Harris, Eva; Sadreyev, Ruslan I.; Stanley, Sarah; Kauppinen, Sakari; Rouskin, Silvi; Näär, Anders M.
title: An intranasal ASO therapeutic targeting SARS-CoV-2
date: 2021-05-18
journal: bioRxiv
DOI: 10.1101/2021.05.17.444397
sha: 397b129303b644a1132bf066f99f54e6f1c495df
doc_id: 980359
cord_uid: wr4zrggo

The COVID-19 pandemic is exacting an increasing toll worldwide, with new SARS-CoV-2 variants emerging that exhibit higher infectivity rates and that may partially evade vaccine and antibody immunity1. Rapid deployment of non-invasive therapeutic avenues capable of preventing infection by all SARS-CoV-2 variants could complement current vaccination efforts and help turn the tide on the COVID-19 pandemic2. Here, we describe a novel therapeutic strategy targeting the SARS-CoV-2 RNA using locked nucleic acid antisense oligonucleotides (LNA ASOs). We identified an LNA ASO binding to the 5’ leader sequence of SARS-CoV-2 ORF1a/b that disrupts a highly conserved stem-loop structure with nanomolar efficacy in preventing viral replication in human cells. Daily intranasal administration of this LNA ASO in the K18-hACE2 humanized COVID-19 mouse model potently (98-99%) suppressed viral replication in the lungs of infected mice, revealing strong prophylactic and treatment effects. We found that the LNA ASO also represses viral infection in golden Syrian hamsters, and is highly efficacious in countering all SARS-CoV-2 “variants of concern” tested in vitro and in vivo, including B.1.427, B.1.1.7, and B.1.351 variants3. Hence, inhaled LNA ASOs targeting SARS-CoV-2 represents a promising therapeutic approach to reduce transmission of variants partially resistant to vaccines and monoclonal antibodies, and could be deployed intranasally for prophylaxis or via lung delivery by nebulizer to decrease severity of COVID-19 in infected individuals. LNA ASOs are chemically stable and can be flexibly modified to target different viral RNA sequences4, and they may have particular impact in areas where vaccine distribution is a challenge, and could be stockpiled for future coronavirus pandemics.

To address these challenges, we developed a novel strategy to inhibit the replication of SARS-80

CoV-2 using antisense oligonucleotides (ASOs) targeting viral RNAs. ASOs, which rely on 81

Watson-Crick base-pairing to target specific complementary RNA sequences, can be quickly 82 designed to target any viral or host RNA sequence, including non-coding structural elements that 83 may be important for viral replication, and may recruit RNase H for cleavage (gapmers) or act 84 through steric hindrance (mixmers) 4 . ASOs are typically well tolerated, and a number of ASO 85 therapeutics have been approved for clinical use 12 . Additionally, ASO manufacturing is well 86 established and can be readily scaled-up. We have employed chemically modified gapmer and 87 mixmer ASOs containing interspersed locked nucleic acid nucleotide bases (LNAs) and DNA 88 nucleotides linked by phosphorothioate (PS) bonds. The introduced chemical modifications confer 89 increased affinity, stability and improved pharmacokinetic/pharmacodynamic properties 13, 14, 15 . 90 91 SARS-CoV-2 is a compact (30 kilobases) positive-sense single-stranded RNA virus, with a 5' 92 untranslated region (UTR), the ORF1a/b RNA encoding non-structural viral proteins, and a 3' 93 segment encoding the structural RNAs, such as the Spike protein that binds to the ACE2 receptor 94 on host cells, and the nucleocapsid N protein involved in virion assembly, and a 3'UTR 16 . The 5' 95 UTR, a non-coding segment consisting of multiple highly conserved stem-loop and more complex 96 secondary structures, is functionally critical for viral translation and replication by affording 97 protection from host cell antiviral defenses and through selective promotion of viral transcript 98 translation over those of the host cell, at least in part through the recruitment of the viral non-99 strctural protein 1 (Nsp1) 17 . The 5' UTR begins with a short 5' leader sequence (nucleotides 1-69), 100

which is added via discontinuous transcription to the 5' end of all sub-genomic RNA transcripts 101

encoding the viral structural proteins, and regulates their translation as well as translation of 102

ORF1a/b from full-length genomic RNA 18 . The ORF1a/b also contains a structured and highly 103 conserved frameshift stimulation element (FSE) near its center that controls a shift in the protein 104 translation reading frame by one nucleotide of ORF1a/b genes 3' to the FSE. The FSE and accurate 105 frame shifting is crucial for the expression of ORF1b, which encodes five non-structural proteins 106

including an RNA-dependent RNA polymerase (RdRP) essential for SARS-CoV-2 genome 107 replication 19 . 108 109 We designed multiple LNA ASOs targeting the 5' leader sequence, downstream sequences in the 110 5' untranslated region (UTR) of ORF1a/b, and the ORF1a/b FSE of SARS-CoV-2 ( Fig. 1a and 111

Extended Data Table 1 ). Additionally, given the importance of the Spike protein in host cell entry 112

for SARS-CoV-2 infection, LNA ASOs targeting the Spike coding sequence were also tested (Fig. 113 1a and Extended Data Table 1 ). To evaluate the viral repressive effect of LNA ASOs, the initial 114 screening was carried out in Huh-7 human hepatoma cells, which exhibit excellent transfection 115 efficiency and are readily infected by SARS-CoV-2. Cells transfected with LNA ASOs were 116 infected with SARS-CoV-2 and both cells and medium were collected at 48h post-infection (hpi) 117

for RNA extraction and infectious viral particle determination, respectively. The viral titer was 118 measured by detecting the expression level/copy number of Nucleocapsid (N) and Spike (S) using 119 reverse transcription (RT)-quantitative PCR (qPCR). The screening results showed that the 120 treatment with certain LNA ASOs lead to a dramatic decrease of N and S expression ( Fig. 1b and  121 1c, Extended Data Fig. 1a and 1b) . 122 123

Interestingly, we found that LNA ASOs targeting the 5' leader region of SARS-CoV-2 were 124 particularly effective in suppressing viral RNA levels in infected cells ( Fig. 1b and Extended Data 125 Fig. 1a ). This is consistent with the fact that the 5' leader sequence is present in all viral RNA 126 transcripts and is required for viral replication. The most potent LNA ASO targeting the 5' leader, 127 5'-ASO#26, was selected for further investigation of its viral repression capability. The repressive 128 effect of 5'-ASO#26 was demonstrated in a dose-dependent manner by measuring the expression 129 level of viral RNAs (Fig. 1d ) and by directly determining the viral titer of infectious particles with 130 the Fifty-percent Tissue Culture Infectious Dose (TCID50) assay in Vero E6 African green monkey 131 kidney epithelial cells (Fig. 1e ). 132 133

Although the 5' UTR nucleotide sequences are somewhat divergent amongst the coronavirus 134 family, the secondary structure of the 5' UTR is highly conserved 20 , and it has been shown that 135 two stem-loop structures, SL1 and SL2, are formed by the 5' leader sequence 21 . Since the 136 complementary sequence of 5'-ASO#26 aligns along the 3' portion of SL1 (marked in pink frame) 137 (Fig. 2c) , we hypothesized that the viral repressive effect of 5'-ASO#26 may be in part due to its 138 ability to disrupt the secondary structure of the 5' leader sequence upon binding to the viral 139 genomic or sub-genomic RNAs, interfering with the formation of the SL1 stem-loop structure. To 140 test if 5'-ASO#26 can disrupt the SL1 structure, we carried out in vitro transcription of the SARS-141

CoV-2 5' UTR sequence, added 5'-ASO#26 and treated samples with dimethyl sulfate (DMS). 142

DMS is able to specifically and rapidly methylate unpaired adenines (A) and cytosines (C) within 143 single-stranded sequences and not those that are complexed as RNA secondary structure or to the 144 LNA ASO, allowing for the unpaired A or C nucleotides to be detected by DMS-MaPseq 22 . Our 145 results strongly indicated that the secondary structure of SL1 was indeed interrupted by 5'-146 ASO#26 in a dose-dependent manner (Fig. 2a) . We also performed DMS-MaPseq with SARS-147

CoV-2-infected Huh-7 cells in the presence or absence of 5'-ASO#26, and monitored secondary 148 structure changes of SL1 at 6 hpi and 12 hpi. Similar to the findings with the in vitro assay, the 149 result from infected cells confirmed the disruption of the secondary structure of SL1 due to binding 150 of 5'-ASO#26 (Fig. 2b ). 151 152

To evaluate the effects of 5'-ASO#26 in vivo, we employed humanized transgenic K18-hACE2 153 mice, which are expressing human ACE2 allowing SARS-CoV-2 cell entry and infectious spread 23 . 154

K18-hACE2 mice were inoculated with 1×10 4 TCID50 units of the USA-WA1/2020 SARS-CoV-155 2 strain via intranasal administration. No significant weight loss was observed at 4 days post-156 infection (dpi) (Extended Data Fig. 2a) , consistent with previous studies indicating that weight loss 157 starts from 5 dpi 23 . Mice were treated with 5'-ASO#26 (400 µg) once-daily via intranasal 158 administration in saline, from 3 days before infection until 3 dpi ( Fig. 3a) . High levels of infectious 159 SARS-CoV-2 viral particles were detected in the lungs of the saline-treated control group (Saline), 160

whereas a remarkable decrease (>80-fold) in the viral load in lungs was observed in the LNA ASO-161 treated group (Fig. 3b ). As expected, the levels of viral N and S RNA were also potently (98-99%) 162

repressed after LNA ASO treatment, as measured by RT-qPCR (Fig. 3c ). Previous studies have 163

shown that golden Syrian hamsters can be infected with SARS-CoV-2 24 . We inoculated hamsters 164

with 10 TCID50 units of the USA-WA1/2020 SARS-CoV-2 strain and treated them with 5'-165 ASO#26 (600 µg) following the same schedule as for the mouse studies (Fig. 3a) . About 10-fold 166 decreased viral load in lung was observed in the LNA ASO-treated group and with no significant 167 weight change (Extended Data Fig. 2d , 2e). Considering that the dose was approximately 2.5-fold 168 lower per kg, the viral replication was still efficiently repressed in hamsters. To further investigate 169 the physiological effects of LNA ASO in mice, histological analyses were carried out with control 170

and LNA ASO-treated lung tissue after SARS-CoV-2 infection. The analyses revealed clear 171

repression of N expression with LNA ASO treatment (Fig. 3d) . Meanwhile, we did not observe 172 significant histological changes by H&E staining in either Saline or LNA ASO-treated mice, which 173

is consistent with previous studies demonstrating that significant inflammation and severe alveolar 174 wall thickening could only be observed at 7 dpi 23 . However, we did notice that the localized signal 175 of N correlated with local enrichment of CD3 (T cell marker), B220 (B cell marker) by 176 immunohistochemistry, and moderate thickening of the alveolar wall (by H&E staining) in Saline-177 treated mice, but not in LNA ASO-treated mice (Extended Data Fig. 2c) . The viral repressive effect 178 of the LNA ASO was also examined by RNA-seq of lung tissues from mice with saline or LNA 179 ASO treatment. Consistent with results from the TCID50 assay, viral reads were dramatically 180 decreased in LNA ASO-treated mice when compared with that of the Saline control (Extended 181

Data Fig. 3a ). To evaluate the in vivo effect of 5'-ASO#26, previous RNA-seq data from infected 182 K18-ACE2 mice 23 was used to define up-and down-regulated genes in response to SARS-CoV-2 183 infection at 4 dpi (Extended Data Fig. 3e ). As expected, expression profile changes induced by 184 infection were markedly rescued by LNA ASO treatment ( Fig. 3e and Extended Data Fig. 3b ). 185

Gene set enrichment analysis (GSEA) revealed strong enrichment of gene involved in type I and 186

II IFN signaling in infected mice treated with Saline ( Fig. 3f and Extended Data Fig. 3c 195 196 To further evaluate the effect of 5'-ASO#26, we first confirmed that 5'-ASO#26 repressed viral 197

replication in vivo in a dose-dependent manner (Extended Data Fig. 2b ). To explore the optimal 198 treatment time course, we assessed the efficacy of 5'-ASO#26 viral repression in pre-infection and 199

post-infection treatment regimens (Fig. 4a ). We found that the strongest viral repressive effect was 200 observed in the Prophylactic #2 group (Fig. 4a) , in which the mice were treated with LNA ASO 201 for four days, followed by infection 24 hrs later. Viral repression was also validated by staining of 202 N in mouse lung tissues (Extended Data Fig. 4a ). Notably, a prophylactic effect was still observed 203 one week after LNA ASO treatment (Prophylactic #1, Fig. 4a ). Starting LNA ASO treatment at 6 204 hpi also revealed a moderate repressive effect, whereas beginning treatment from 1 dpi showed no 205

repressive effect in mice (Treatment #3 and #4, Fig. 4a ). These results show that 5'-ASO#26 206 exhibits potent effects as a prophylactic therapeutic. The diminished ability of late post-infection 207 treatment of 5'-ASO#26 to inhibit viral replication may be a result of rapidly accumulating sub-208 genomic viral RNAs saturating the amount of administered LNA ASO in the lung due to the very 209 large dose of virus used to inoculate the mice. Considering that we directly administered the naked 210 5'-ASO#26 intranasally in saline, we believe that additional modifications of the LNA ASO (such 211 as lipid-conjugation 28,29,30 ) may further promote cellular uptake of LNA ASO in lung, improving 212 the effect of post-infection LNA ASO treatment. Direct delivery of the LNA ASO to the lung via 213 nebulizer may also improve post-infection therapeutic effect. 214 215

Because the 5' leader sequence of SARS-CoV-2 is highly conserved and as 5'-ASO#26 does not 216 target Spike, we predicted that 5'-ASO#26 should also be able to repress the replication of SARS-217

CoV-2 variant strains. Therefore, we tested several reported variants with 5'-ASO#26 in cell-based 218 assays. Our results showed that regardless of the mutations, 5'-ASO#26 exhibits potent repressive 219

activity on viral replication of multiple SARS-CoV-2 variants, including B. however the viral titers of both variant strains were repressed to a similar degree by LNA ASO 227 treatment ( Fig. 4c and 4e) . Of note, similar to animals infected with the WA1 strain, the B.1.351 228 strain did not induce weight loss at 4 dpi (Fig. 4d) . However, we noticed that there was a significant 229 weight loss induced by B.1.427 strain infection from 3 dpi in the saline-treated group (Fig. 4f) . In 230 contrast, the LNA ASO-treated mice did not exhibit significant weight loss after B.1.427 strain 231 infection (Fig. 4f) GGGATTAAAGGTTTATACCTTCCC-3' and Rv 5'-TCGTTGAAACCAGGGACAAG-3'). 372

The PCR product was purified using E-Gel TM SizeSelect TM II 2% agarose gel (Invitrogen Immunohistochemistry was performed on a Bond Rx autostainer (Leica Biosystems) with enzyme 382 treatment (1:1000) using standard protocols. Antibodies used were rabbit monoclonal CD3 383 primary antibody (Abcam, ab16669, 1:100), rabbit monoclonal B220 primary antibody (Novus, 384 NB100-77420, 1:10000), rabbit monoclonal SARS-CoV-2 (COVID-19) nucleocapsid primary 385 antibody (GeneTex, GTX635686, 1:8000) and rabbit anti-rat secondary (Vector, 1:100). Bond 386

Polymer Refine Detection (Leica Biosystems) was used according to the manufacturer's protocol. 387

After staining, sections were dehydrated and film coverslipped using a TissueTek-Prisma and 388

Coverslipper (Sakura). Whole slide scanning (40x) was performed on an Aperio AT2 (Leica 389

Biosystems). 390 391

Statistical analysis. Data are presented as mean values, and error bars represent SD. Data analysis 392 was performed using GraphPad Prism 8. Data were analyzed using unpaired t-test; one-way or 393 two-way ANOVA followed by Turkey or Dunnett test as indicated. P value < 0.05 was considered 394 as statistically significant. 395 396

Data availability. All data are available in the manuscript and associated files. Source data is 397 provided with this paper. strain-infected Huh-7 cells treated with LNA ASO (100 nM) and cell culture medium were 525 collected at 48 hpi. Viral RNA levels were analyzed by RT-qPCR for LNA ASO screening. Each 526 LNA ASO was tested in duplicate. d) Dose-dependent effects of 5'-ASO#26 were evaluated in 527

infected Huh-7 cells with increasing doses of the LNA ASO (as indicated) by RT-qPCR. e) The 528

infectious virus was measured by TCID50 assay. The infected Huh-7 cells with different doses of 529 LNA ASO treatment were collected at 48 hpi. For d) and e), one-way ANOVA with Dunnett's test 530 was used to determine significance (** P < 0.01, *** P < 0.001, **** P < 0.0001, ns, not 531 significant 

Insights into the secondary structural ensembles of the full SARS-471

CoV-2 RNA genome in infected cells

DMS-MaPseq for genome-wide or targeted RNA structure probing in 474 vivo

SARS-CoV-2 infection of human ACE2-transgenic mice causes 476 severe lung inflammation and impaired function

Pathogenesis and transmission of SARS-CoV-2 in golden hamsters

Genetic Screens Identify Host Factors for SARS-CoV-2 and Common 481

Genome-Scale Identification of SARS-CoV-2 and Pan-483 coronavirus Host Factor Networks

Mice were 612 administered with different treatment regimens of 5'-ASO#26 as indicated. Treatment on the day 613 of infection was carried out at 6 hpi. The viral burden in lungs of mice in each group (N =5) was 614 measured by TCID50 assay using lung homogenates. One-way ANOVA with Dunnett's test was 615 used to determine significance (* P < 0.05). b) Dose-dependent effects of 5'-ASO#26 on inhibition 616 of viral replication of SARS-CoV-2 WA1 and B.1.351 strains were evaluated in infected Huh-7 617 cells by RT-qPCR of

c) and e) The 619 viral burden in lungs of mice treated with Saline (N=5) and 5'-ASO#26 (N =5) was measured by 620 TCID50 assay using lung homogenates. Student t-test was used to determine significance (**** P 621 < 0.0001). d) and f) Weight change was monitored (N=5, symbols represent mean ± SD). Two-622 way ANOVA was used to determine significance (** P < 0.01, **** P < 0.0001). For a), c) and 623 e), center line, median; box limits, upper and lower quartiles

Extended Data Figure 1. In vitro screening of LNA ASOs targeting SARS-CoV-2. a) and b) 663

LNA ASO (100 nM) and cell culture media was collected 664 at 48 hpi. Levels of viral Spike (S) RNA were analyzed by RT-qPCR. Each LNA ASO was tested 665 in duplicate. One-way ANOVA with Dunnett's test was used to determine significance

Extended Data Figure 3. Evaluating the in vivo effects of 5'-ASO#26 in K18-hACE2 mice

Reads mapped to the human genome marked in gray and reads mapped to virus 729 marked in red. b) Expression change of SARS-CoV-2 infection-downregulated genes in Saline-730 and LNA ASO-treated group. Infection. c) GSEA of REACTOME gene sets enriched among 731 upregulated genes in lungs of Saline-treated mice. Terms were ranked by the false discovery rate 732 (q value). d) GSEA plot and heatmap of significantly upregulated genes enriched in cholesterol 733 homeostasis pathway in ASO-treated mice. e) Heatmaps of significantly upregulated and 734 downregulated (> 2-fold, FDR<0.01) genes at day 4 of SARS-CoV-2 infection

Columns represent samples and rows represent genes. Colors indicate gene expression levels (log2 736 RPKM) relative to average expression across all samples

repressing the replication of SARS-CoV-2 740 variant strains. a) Representative images of IHC staining of SARS-CoV-2 nucleocapsid protein 741 in Saline (N=5) or Prophylactic#2 (N=5) regimen groups. Scale bar = 2 mm. b) Dose-dependent 742 efficacy of 5'-ASO#26 in repressing replication of SARS-CoV-2 variant strains was evaluated in 743 infected Huh-7 cells with increasing doses of 5'-ASO#26 by RT

One-way ANOVA with Dunnett's test was used to determine significance (*** 745

# no detection). c) Viral RNA levels in mouse lungs were analyzed 746 by RT-qPCR of Nucleocapsid (N) and Spike (S) RNAs. Center line, median; box limits, upper and 747 lower quartiles; plot limits, maximum and minimum in the boxplot

test was used to determine significance (*** P < 0.001) and for a) and e), Two-way ANOVA was 690 used to determine significance .  691  692  693  694  695  696  697  698  699  700  701  702  703  704  705  706  707  708  709  710  711  712  713  714  715  716  717  718  719  720  721  722  723  724  725