key: cord-0330851-m2zw1019
authors: Pawlica, Paulina; Yario, Therese A.; White, Sylvia; Wang, Jianhui; Moss, Walter N.; Hui, Pei; Vinetz, Joseph M.; Steitz, Joan A.
title: SARS-CoV-2 expresses a microRNA-like small RNA able to selectively repress host genes
date: 2021-09-08
journal: bioRxiv
DOI: 10.1101/2021.09.08.459464
sha: 3e724ee2ff7949127ae7a4ed9d97f87ca1577d88
doc_id: 330851
cord_uid: m2zw1019

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent of coronavirus disease (COVID-19), continues to be a pressing health concern. In this study, we investigated the impact of SARS-CoV-2 infection on host microRNA (miRNA) populations in three human lung-derived cell lines, as well as in nasopharyngeal swabs from SARS-CoV-2 infected individuals. We did not detect any major and consistent differences in host miRNA levels after SARS-CoV-2 infection. However, we unexpectedly discovered a viral miRNA-like small RNA, named vmiR-5p (for viral miRNA), derived from the SARS-CoV-2 ORF7a transcript. Its abundance ranges from low to moderate as compared to host miRNAs. vmiR-5p functionally associates with Argonaute proteins — core components of the RNA interference pathway — leading to downregulation of host transcripts. One such host messenger RNA encodes Basic Leucine Zipper ATF-Like Transcription Factor 2 (BATF2), which is linked to interferon signaling. We demonstrate that vmiR-5p production relies on cellular machinery, yet is independent of Drosha protein, and is enhanced by the presence of a strong and evolutionarily conserved hairpin formed within the ORF7a sequence. Significance statement We discovered that severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) expresses a small viral non-coding RNA, named vmiR-5p (for viral miRNA), derived from the ORF7a transcript. vmiR-5p associates with the cellular RNA interference machinery to regulate host transcripts likely via target silencing. The production of vmiR-5p relies on cellular machinery and the formation of a strong hairpin within ORF7a sequences. This newly-described vmiR-5p may contribute to SARS-CoV-2 pathogenesis and could become a target for therapeutic intervention.

Coronaviruses are large single-stranded positive-sense RNA viruses that infect various animals; 2 in humans coronaviral infection results in mild to severe respiratory disease. -coronaviruses, 3 such as severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East 4 respiratory syndrome coronavirus (MERS-CoV), cause very severe disease in humans, while 5 the recently emerged severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) results in 6 coronavirus disease characterized by wide array of symptoms from mild to serious 7

illness. SARS-CoV-2 is the cause of the COVID-19 pandemic due to its high transmissibility and 8 emergence of novel variants (1). It is crucial to understand all aspects of SARS-CoV-2 9 pathogenesis in order to develop multiple tools to combat the constantly-adapting virus. 10 microRNAs (miRNAs) are ~22 nucleotide (nt) small non-coding RNAs (ncRNAs) involved 11 in post-transcriptional regulation of gene expression in animals and plants (2). During canonical 12 miRNA biogenesis, RNA polymerase II transcribes primary miRNAs (pri-miRNAs) characterized 13 by formation of strong hairpins that are recognized and cleaved by the endonuclease Drosha, 14 resulting in ~70 nt-long precursor miRNAs (pre-miRNAs). Pre-miRNAs are exported to the 15 cytoplasm, where they are subjected to yet another cleavage by Dicer, giving rise to ~22 nt-long 16 miRNA duplexes, which are then loaded onto Argonaute (Ago) proteins. The passenger strand 17 is removed, and the mature miRNA guides Ago proteins to partially complementary target 18 sequences, usually located in the 3′ untranslated region (UTR) of messenger RNAs (mRNAs). 19 miRNA binding leads to translation inhibition and/or mRNA decay by decapping and 20 deadenylation. In the special case where the miRNA is nearly perfectly complementary to its 21 target mRNA, it directs Ago2 (one of four Ago proteins in humans) to cleave the mRNA target; a 22 function typically associated with small interfering RNAs (siRNAs) (2). These modes of target 23 repression belong to RNA interference (RNAi) pathway, but Ago2-mediated cleavage is more 24 potent in target silencing and requires lower intracellular copy numbers than the canonical 25 miRNA action mode (3) . Importantly, miRNA profiles differ between cell types and developmental stages, and a single miRNA can downregulate multiple transcripts, precisely 1 regulating gene expression dependent on the cell's needs. In humans, most mRNAs are 2 regulated by miRNAs, and aberrant miRNA levels are linked to disease (2). 3

Viruses often hijack the miRNA pathway; either by depleting host miRNAs or by 4 producing their own miRNAs (reviewed in (4, 5) ). Some examples of host miRNA regulation 5 include: selective host miRNA decay mediated by certain herpesviral transcripts in a process 6 called target-directed miRNA degradation (TDMD; reviewed in (6, 7)), and widespread miRNA 7 polyadenylation by poxvirus poly(A) polymerase, which results in miRNA decay (8). Viruses can 8 produce their own miRNAs, often via non-canonical biogenesis pathways; especially 9 cytoplasmic viruses have devised multiple strategies to bypass the requirement for nuclear 10 Drosha (4, 9) . It has been proposed that coronaviruses, SARS-CoV and SARS-CoV-2, produce 11 small viral RNAs (svRNAs) involved in viral pathogenesis independently of RNAi pathway (10, 12 11). One recent study suggested the existence of miRNA-like RNAs produced by SARS-CoV-2, 13 which might contribute to inflammation and type I interferon (IFN) signaling (12). 14 In this study, by using a small RNA-seq library preparation protocol that selectively 15 enriches for miRNAs, we unexpectedly discovered a miRNA-like viral ncRNA, named vmiR-5p 16 (for viral miRNA), expressed by SARS-CoV-2. We show that it associates with Ago proteins and 17 has the potential to regulate host transcripts likely via target silencing. We demonstrate that 18 vmiR-5p production relies on cellular machinery, but is independent of Drosha proteins and is 19 enhanced by the presence of a strong hairpin formed within ORF7a sequences. This newly-20 described vmiR-5p may contribute to SARS-CoV-2 pathogenesis and could become a target for 21 therapeutic intervention. 22

Results 1

To explore how SARS-CoV-2 affects host miRNA populations, we infected three lung-derived 3 human cell lines either at high multiplicity of infection (MOI of 5) to detect the possible direct 4 impact of the incoming virions on host miRNAs, or at low MOI (0.05) to assess the effect of viral 5

replication on these small RNA species. We used non-small-cell lung cancer cells Calu-3, lung 6 adenocarcinoma cells A549 transduced with the human viral receptor, angiotensin-converting 7 enzyme 2 (ACE2), and lung adenocarcinoma cells PC-9, all of which we found to support 8 SARS-CoV-2 replication (red, yellow and blue dots, respectively, in Fig. 1A ). SARS-CoV-2 9 replicated better in Calu-3 and PC-9 cell lines than in the A549-hACE2 cell line, possibly 10 because the viral receptor is greatly overexpressed in the last cell line (Fig. S1A ), which could 11 sequester budding virions at the cell surface. 12

Small RNA sequencing was performed at both 6 h and 24 h post infection (hpi) at high 13 MOI and 48 hpi at low MOI, as well as for the uninfected controls (at the 24-h time point). The 14 vast majority (84% ± 5%) of obtained reads mapped to host miRNAs (green bars in Fig. 1B) with 15 an average read length of 22-23 nts (Fig. S1B) . Overall, only a small fraction of small RNA 16 reads mapped to the viral genome (0.07±0.01%, 0.21±0.04% and 1.63±0.67% at high MOIs 24 17 hpi for Calu-3, PC-9 and A549-hACE2, respectively; blue bars in Fig. 1B ). This number is in 18 agreement with similar studies of svRNAs in other RNA viruses (10-12). 19 SARS-CoV-2 infection resulted in minimal alteration of some host miRNA levels, but no 20 miRNA displayed a uniformly significant change across all three cell lines (red and blue dots in 21

Figs. 1C and S1C). To identify any consistent shifts in host miRNA levels across the three cell 22 lines, we generated a heat map of fold-changes for miRNAs that were significantly altered, by at 23 least two-fold, in at least one condition (Fig. S1D ). Some of these trends were further examined 24 by Northern blotting of RNA from Calu-3 cells infected with SARS-CoV-2 at low MOI. The most 25 reproducibly changed (upregulated) miRNA, although not reaching statistical significance, was although slightly more weakly than most host-cell miRNAs; its expression levels are comparable 1 to low or moderately expressed host miRNAs (Fig. 3A) . To obtain enough material for Northern 2 blotting experiments, we performed an anti-HA RIP from Calu-3 cells transduced with either 3 empty vector or FLAG-HA-tagged human Ago2 (27). Although weak, Ago2 selectively bound 4 vmiR-5p, but not vmiR-3p. In these experiments, we observed slight RNA degradation likely 5 arising during the SARS-CoV-2 inactivation step (30 min incubation at RT). Perhaps this is the 6 reason why vmiR-5p bands in Fig. 3B migrate somewhat faster than those in Fig. 2D . 7

The role of Ago-associated small RNAs is to repress targeted mRNAs via sequence 8 complementarity, we thus computationally searched the repository of human mRNA transcripts 9 (obtained from GENECODE) for putative vmiR-5p targets. Considering the moderate expression 10 levels of vmiR-5p as compared to those of host miRNAs, we hypothesized that vmiR-5p might 11 rely on Ago2's ability to cleave its mRNA target. Since Ago2-mediated target cleavage tolerates 12 a limited number of mismatches (28, 29), we searched for mRNAs with high complementarity to 13 vmiR-5p (Table 1 ; Fig. 3C ), and selected some of these for further validation. 14 Due to the widespread host-shutoff effect after infection with SARS-CoV-2 (16), we 15 were unable to perform meaningful quantification of mRNAs from infected cells. We thus 16 transfected synthetic vmiR-5p (annealed to a passenger strand) into HEK293T cells and 17 assayed transcript levels of predicted target mRNAs (Fig. 3D ). Two mRNAs, Basic Leucine 18

Zipper ATF-Like Transcription Factor 2 (BATF2) and (Heparan Sulfate Proteoglycan 2) HSPG2, 19

were significantly downregulated (Fig. 3D ). BATF2 has been linked to INF-gamma signaling via 20 association with Irf1 (30). HSPG2 is the core protein of a large multidomain proteoglycan that 21 binds and cross-links many extracellular matrix components; it interestingly participates in the 22 attachment of some viruses, including the coronavirus NL63 (31). We hypothesize that vmiR-5p 23 has the potential to repress human mRNAs that support the host IFN response, and perhaps to 24 inhibit viral superinfection. 25

To assess whether vmiR-5p biogenesis can occur in a virus-free system, we transfected in vitro 2 transcribed RNA into HEK293T cells. A 100 nt-long viral RNA sequence (WT), alongside 3 counterparts containing mutations either in the 3′ portion (m3p) or in the 5′ portion (m5p) of the 4 hairpin ( Fig. 4A ) were transfected into cells and RNA samples collected after 6 h were analyzed 5 by Northern blotting (Fig. 4B ). The WT sequence was processed inside cells to produce a 21 6 nt-long band of vmiR-5p, while the m3p did not yield efficient production of an RNA of such size. 7

To investigate whether the produced RNA functionally associates with Ago proteins, we 8 transfected in vitro transcribed RNA together with luciferase reporters containing 3′ UTR 9 sequences complementary to either vmiR-5p or scrambled controls. The luciferase activity of 10 the reporter containing viral sequences was significantly decreased in the presence of the WT 11

ORF7a RNA (Fig. 4C ), but not by the other constructs. Finally, we used Drosha knockout cell 12 lines (32) to show that the production of vmiR-5p occurs independently of this enzyme's activity 13 (Fig. 4D) . These results clearly demonstrate that vmiR-5p is processed from a hairpin 14

independently of viral proteins and of cellular Drosha, and that it can be functionally loaded on 15

Ago. 16 17

Here, we began by investigating the impact of SARS-CoV-2 infection on host miRNA 19 populations using three human lung-derived cell lines (at various MOIs) and found no consistent 20 changes detected in all three cell lines (Figs. 1C and S1C). We also assessed the levels of 21 selected miRNAs in patient samples and again did not detect major differences after SARS-22

CoV-2 infection (Fig. 1D ). Other groups have also ventured into examining miRNA profiles 23 during SARS-CoV-2 infection, but there is little overlap between these prior studies (33) (34) (35) and 24 our results. Differences between the outcomes likely stem from the use of different methods of 25 RNA-seq library preparation (36, 37) . Our libraries were prepared with the NEXTflex v2 kit, which has been shown to be highly selective for miRNAs (36) , and contained ~84% host 1 miRNAs ( Fig. 1B) , while libraries that rely on template switching used by others (33) yield only 2 ~17% miRNAs (37) . The presence of additional reads in the latter libraries could result in 3 nonspecific alignment to miRNA loci. For example, miR-155-3p, which is believed to represent a 4 host miRNA passenger stand, was reported to increase upon SARS-CoV and SARS-CoV-2 5 infection (33), but we were unable to detect this miRNA either by sequencing or by Northern 6 blotting (data not shown). Similarly, miR-4485 was described to be upregulated in another study 7 (34) , but it did not pass the abundance threshold (> 1 CPM) in our small RNA-seq analysis. 8

Also, other studies assessing the impact of various coronaviruses on host miRNA levels do not 9 support the notion that coronaviruses have major impact on host miRNA populations (38-41). 10

We cannot rigorously exclude the possibility that some host miRNAs levels are altered during 11 SARS-CoV-2 infection, but if so, these changes are likely minor. 12

Interestingly, in samples from individuals infected with SARS-CoV-2 all assessed host 13 miRNAs seemed to be stabilized, whilesimilar to what is known about host mRNAs in 14 infected cells (16) -U6B snRNA and U44 snoRNA levels decreased (Fig. S1H ). These results 15

suggest that miRNAs might be resistant to the viral host-shutoff effect. Indeed, miRNAs can be 16 very stable because of their association with Ago proteins, especially when secreted in 17 exosomes (2, 17, 18). It is tempting to speculate that, because of their high stability, small RNAs 18 could be successfully used as therapeutic agents against SARS-CoV-2. Regardless, we caution 19 against examining miRNA profiles using RT-qPCR and normalizing to ncRNAs from different 20 classes, such as snRNAs and snoRNAs. 21

In this study, we discovered that SARS-CoV-2 expresses an miRNA-like small RNA, 22

which we call vmiR-5p (Fig. 2 ). There are multiple examples of svRNAs expressed by RNA 23 viruses, such as influenza (42), enterovirus 71 (43), Hepatitis C (44), hepatitis A (45), Polio (44), 24

Dengue (44, 46) , Vesicular Stomatitis (44), West Nile (44, 47) , coronaviruses (10-12) and retroviruses (48, 49) . It has been suggested that some svRNAs function through the host-cell 1

RNAi pathway (12, 44, 45) , but most proposed roles are independent of this machinery (10, 11, 2 42, 43, 50) . As compared to other studies of SARS-CoV-2, we did not detect svRNAs derived 3 from the N gene (10, 12), which suggests that the functions of those is not related to RNAi 4 pathway. Interestingly, vmiR-5p was also detected by Meng and colleagues, but was not 5 selected for further validation (12). 6 vmiR-5p is derived from the beginning of the coding sequence of the ORF7a transcript 7 ( Fig. 2B) , which is the most abundant subgenomic viral RNA detected during SARS-CoV-2 8 infection (33, 51) . The sequence of vmiR-5p is preserved in all SARS-CoV-2 isolates and 9

variants of concern, and is related to those of two other coronaviruses (Fig. S2 ). Interestingly, 10

ORF7a deletions, which lead to decreased innate immune suppression, occur frequently in 11 SARS-CoV-2, but affect almost exclusively the C-terminus of the protein, while the N-terminus 12 where the vmiR-5p sequence is locatedis preserved (52). These data and the observed 13 conservation of the hairpin containing vmiR-5p (21) support the notion that this sequence is 14

The abundance of vmiR-5p ranges from low to moderate as compared to host miRNAs 16 ( Fig. 3A) , similar to other svRNAs (10, 45, 53) , and is consistent with the idea that its 17 functionality relies on high complementarity to target mRNA, likely inducing mRNA cleavage. 18

Indeed, we have shown that vmiR-5p associates with human Ago proteins and that it can 19 repress human targets (Fig. 3 ). Using bioinformatics, we identified two host transcripts that are 20 significantly downregulated in the presence of a synthetic vmiR-5p, BATF2 and HSPG2 (Fig. 3  21 and Table 1 ). In addition to these, there are likely more targets that may be identified by cross-22

linking ligation and sequencing of hybrid (CLASH; (54)) methodology. To address the 23 functionality of vmiR-5p in the context of viral infection, we attempted to use luciferase reporters, 24 but due to the host-shutoff effect there was almost no luciferase signal in infected cells. We 25 avoided using antisense oligonucleotides to block vmiR-5p because they would undoubtedly also suppress production of the ORF7a protein and inhibit viral replication. Upcoming studies 1 will focus on the construction of mutant viruses in which the ORF7a coding sequence and 2 hairpin formation are preserved. 3

Another intriguing possibility is that SARS-CoV-2 uses vmiR-5p to regulate the level of 4 its own negative-sense subgenomic RNAs, or of antigenomic RNAs (55). It is also plausible that 5 the viral hairpin, in addition to being the source of a viral miRNA, could have an additional 6 function independent of the RNAi pathway. RNA viruses have been shown to express regulatory 7 svRNAs. For example, svRNA1 from enterovirus-71 binds to a viral internal ribosome entry site 8 (IRES) to regulate translation of viral proteins (43) . Another example is svRNAs from influenza 9 that associate with viral RdRp, possibly enabling the switch from transcription to replication (50). 10

Finally, we addressed vmiR-5p biogenesis and were able to demonstrate that its 11 processing occurs via cellular machinery and relies on the formation of an RNA hairpin, but is 12 independent of Drosha protein (Fig. 4 ). Yet, it is possible that viral genes enhance this 13 processing pathway. Many viruses bypass Drosha to produce their pre-miRNAs, e.g. by utilizing 14

RNase Z (56, 57), Integrator complex (58) or a viral protein (such as HIV-1 TAT) (59). In some 15

instances, viral miRNAs biogenesis even relies on Ago2 cleavage (4) . We speculate that Dicer, 16 which is responsible for production of the majority of host and viral miRNAs (2, 45, 59, 60), is 17 involved in vmiR-5p biogenesis. Future efforts will aim to uncover which enzymes are 18 responsible for vmiR-5p production. 19

In summary, viruses develop multiple ways to suppress host gene expression, 20 and multiple overlapping mechanisms often evolve. SARS-CoV-2 regulates host mRNA 21 expression on many levels: stability (16), export (16, 60), translation (61-63) and 22 splicing (61). It is not surprising that yet another strategy, which utilizes the RNAi 23 pathway, to selectively target hostand perhaps viraltranscripts has evolved. 24 1 Acknowledgements 2

We thank the members of the Steitz lab for advice and support. We are grateful to Dr. Craig 3

Wilen for the help in setting up BSL3 work. We thank Dr. Charles Rice for Hu7.5 cells and 4

Huh7.5 Drosha knockout cells (32). We thank Dr. Benjamin Israelow for a plasmid expressing 5 hACE2. We acknowledge the hard work of Yale's Environmental Health and Safety program, in 6 particular Benjamin Fontes. This work was funded by the NIH supplement 3P01CA016038- Plaque assays. Vero-E6 cells were seeded at 4×10 5 cells/well in 12-well plates. The next day, 20 medium was removed, and cells were incubated for 1 h at 37°C with 100 μl serially diluted 21 sample; plates were rocked every 15 min. Next, 1 ml of the overlay media (DMEM, 2% FBS, 22 0.6% Avicel) was added to each well. 3 days later, the plates were fixed with 10% formaldehyde 23 for 30 min, stained with crystal violet solution (0.5% crystal violet in 20% methanol) for 30 min, 24

and then rinsed with water to visualize plaques. 25 Samples for small RNA sequencing. For small RNA sequencing, 5×10 5 of Calu-3, 5×10 5 of 1 PC-9 or 1×10 5 of A549-hACE2 cells were plated per well in 6-well plates. Cells were infected 2 with SARS-CoV-2 at MOI either of 5 or 0.05 for 1 h in 200 µl of inoculum and incubated for 1 h 3 at 37°C. After that, the cells were rinsed with PBS to remove the unbound virus and fresh were 4 media added. Cells were incubated for either 6 h and 24 h (MOI 5), or 48 h (MOI 0.05). In 5 addition, mock-treated controls were collected after 24h. Cells were harvested by removing the 6 media and adding TRIzol at each time point; supernatants were kept to assess viral titer by 7 plague assay. RNA isolation and library preparation was performed under BSL2+ confinement; 8 RNA concentration was measured by Qubit Fluorimeter (ThermoFisher) inside a biosafety 9

cabinet. 1 g of total RNA was used for library preparation using NEXTflex v2 kit (PerkinElmer) μl specimen of nasopharyngeal swab in transportation medium was processed and the miRNA-4 enriched RNA was collected in 50 μl of elution buffer. miRNA RT-qPCR was performed using 5

TaqMan MicroRNA Assay (Applied Biosystems); for miR-16, miR-210-3p, miR-31-3p, miR-193-6 5p, miR-193-3p, U6B and U44 pre-designed assays were ordered (Cat # 4427975) and for 7 vmiR-5p, a custom assay to detect the sequence UUCUUGGCACUGAUAACAC was ordered 8 (Cat # 4398987). RT-qPCR was performed according to the manufacturer's guidelines; briefly, 9 cDNA was prepared separately for each miRNA using the TaqMan MicroRNA Reverse 10 Transcription Kit (Applied Biosystems) and qPCR was performed using the TaqMan Fast 11

Advanced Master Mix (Applied Biosystems). For detection of viral RdRp, cDNA was prepared 12 using random primers and SuperScript III (Invitrogen) and qPCR was performed according to 13 the protocol (71) using oligonucleotides and a probe ordered from IDT (sequences are given in 14 Table 2 ). All work was done in BSL2+ conditions. 15 16 Northern blot analysis. 5×10 6 of Calu-3 cells were infected with SARS-CoV-2 at MOI 0.05 in 17 T75 flasks for 1h, after which the inoculum was replaced by fresh media, and cells were 18 incubated for either 24 h or 48 h (mock-treated samples were collected after 48h). Samples 19

were inactivated for 30 min in 5 ml of TRIzol and RNA was isolated in BSL2+ conditions. 10 -15 20 µg of total RNA was separated by 15% urea-PAGE, electrotransferred to Hybond-NX 21 membrane (Amersham) and crosslinked with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide 22

(EDC) (72). miRNAs were detected using 32 P 5'-radiolabeled DNA probes. Densitometry was 23 performed by using Quantity One. 24 miR-210-3p target and re-analysis of public data. Counts from RNA-seq experiments of lung 1 biopsies were obtained from (14, 15) . The data were divided into miR-210-3p targets (data from 2 miRTarBase (73); excluding "weak" targets) and other transcripts. The data were plotted as 3 cumulative distribution functions and compared to each other using Wilcoxon test. For anti-HA IP, 5×10 6 of Calu-3 cells, transduced with either EV or Ago2, were infected 20 with SARS-CoV-2 at MOI 0.1 for 48h. Cells were detached with trypsin-EDTA, pelleted and 21 resuspended in 2 ml of Pierce IP Lysis Buffer (Thermo Fisher). Cells were incubated at room 22 temperature for 30 min (approved by the Yale University Biosafety Committee SARS-CoV-2 23 inactivation method). After this time, the samples were moved to the BSL2+ laboratory, 24 sonicated using a Diagenode Bioruptor Pico sonication device, and IP was performed using Anti-HA Magnetic Beads (Pierce). Beads were resuspended in TRIzol, RNA was extracted and 1 analyzed by Northern blot as described above. Table 2 for sequences) were 18 annealed by heating equimolar concentrations for 1 min at 90°C in siRNA buffer (Horizon, 60 19 mM KCl, 6 mM HEPES-pH 7.5, 0.2 mM MgCl2) and then incubating for 1 h at 37°C. 5x10 5 20

HEK293T cells were transfected with 30 M of either vmiR-5p or control siRNA, by using 21

Lipofectamine RNAiMAX Transfection Reagent (Invitrogen). Cells were collected 48 h later, 22

RNA extractions were performed using TRIzol, samples were treated with RQ1 DNase 23 (Promega). cDNA was made using SuperScript III and random primers (Invitrogen), and qPCR was performed using FastStart Essential DNA Green Master (Roche). Primer sequences are 1 given in Table 2 . Results were analyzed using the comparative Ct method (75). 2 3 In vitro RNA transcription and processing. The sequences were PCR amplified from 4 templates containing desired mutations and by using flanking primers; the forward primer 5 contained the T7 promoter (see Table 2 for sequences). The product was purified and used for control) separated by 20 nt were PCR amplified using four primers overlapping at the unique 20 spacer sequences (listed in Table 2 ). The PCR products were cloned downstream of Renilla 21 luciferase of psiCHECK(TM)-2 vector (Promega) using XhoI and NotI sites. 22 5x10 5 cells (HEK293Ts, Huh7.5 or Huh7.5 KO) were transfected with 30 M of in vitro 23 transcribed RNAs, by using Lipofectamine RNAiMAX Transfection Reagent (Invitrogen). Cells 24 were collected 6 h later, RNA extractions were performed using TRIzol and samples were processed for Northern blotting as described above. 2 4h later, 10 ng psiCHECK reporters and 1 2 μg pBlueScript II (Stratagene) were transfected using TransIT-293 Transfection Reagent 2 (Mirus). After additional 24h, Firefly and Renilla luciferase activities were measured by the Dual-3 summarizing the origin of obtained reads across the sequenced conditions. Error bars signify 7 SD. C) Plots summarizing the fold-change in host miRNA levels 48 hpi with SARS-CoV-2, n = 3. 8

Colored dots denote miRNAs that significantly (p < 0.05) changed (at least two-fold). small RNA reads obtained from the three cell lines infected with SARS-CoV-2 map to a single 3 distinct peak within the viral genome (data for MOI 5, 24 hpi are shown). The replicates for each 4 cell line were overlayed on a single track (represented by different colors) and are normalized to 5 10 7 total reads. MOImultiplicity of infection. B) The reads coming from SARS-CoV-2 (~20 nt-6 long) map near the beginning of the ORF7a gene (encoded amino acids are shown above the 7 nucleotide sequence). Data from Calu-3 cells are shown. C) vmiR-5p forms a hairpin with the 8 sequence immediately downstream in the viral genome. Shaded nucleotides indicate the 9 sequences detected by Northern blot probes: pink for 5p and blue for 3p. D) vmiR-5p can be 10 detected by Northern blotting of extracts from Calu-3 cells infected with SARS-CoV-2 at MOI 1 Figure 3 . vmiR-5p associates with Argonaute and has the capacity to repress the levels 2 of host transcripts. A) vmiR-5p associates with Ago proteins and its levels are comparable to 3 those of moderately expressed host miRNAs. Anti-pan-Ago RNA Immunoprecipitation followed 4 by sequencing was performed on extracts of Calu-3 and A549-hACE2 cells infected with SARS-5

CoV-2 at MOI 5 for 24h. Each plot shows the average of three independent experiments. B) 6 vmiR-5p is selectively loaded on Ago2. Representative Northern blots for viRNA5p and 7

viRNA3p showing anti-HA IP from Calu-3 cells transduced with either empty vector (EV) or with 8 FLAG-HA-tagged Ago2 (Ago2). INinput, 10%. The miR-16 lanes provide size markers. C) 9

Predicted interactions between vmiR-5p and sequences from the CDSs of two targeted host 10 mRNAs. D) Synthetic vmiR-5p downregulates host gene expression. HEK293T cells were 11 transfected with synthetic vmiR-5p (annealed to a passenger strand) and levels of seven 

Reduced sensitivity of 6 SARS-CoV-2 variant Delta to antibody neutralization

Control of translation and mRNA 11 degradation by miRNAs and siRNAs

Idiosyncrasies of Viral Noncoding RNAs Provide Insights into Host Cell 15 Biology

Balance and Stealth: The Role of Noncoding RNAs in the 18 Regulation of Virus Gene Expression

How Complementary Targets 13

DIANA-TarBase v8: a decade-long collection of experimentally supported 3 miRNA-gene interactions

Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19

Temporal 14 and spatial heterogeneity of host response to SARS-CoV-2 pulmonary infection

SARS-CoV-2 uses a multipronged strategy to impede host protein 20 synthesis

Extracellular 23 microRNAs in human circulation are associated with miRISC complexes that are accessible to 24 anti-AGO2 antibody and can bind target mimic oligonucleotides

Disney 39 MD, Moss WN. A map of the SARS-CoV-2 RNA structurome

Comprehensive in vivo secondary structure of the SARS-CoV-2 genome reveals novel 44 regulatory motifs and mechanisms

In vivo structural characterization of the SARS-CoV-2 48 RNA genome identifies host proteins vulnerable to repurposed drugs

Addressing Bias in Small RNA Library Preparation for Sequencing: A New 2 Protocol Recovers MicroRNAs that Evade Capture by

Integrative 11 deep sequencing of the mouse lung transcriptome reveals differential expression of diverse 12 classes of small RNAs in response to respiratory virus infection

The 15 Coronavirus Transmissible Gastroenteritis Virus Evades the Type I Interferon Response through 16 IRE1alpha-Mediated Manipulation of the MicroRNA miR-30a-5p/SOCS1/3 Axis

Human coronavirus OC43 nucleocapsid 20 protein binds microRNA 9 and potentiates NF-kappaB activation

Epub 2013/10/11

MicroRNome analysis unravels the molecular basis of 24 SARS infection in bronchoalveolar stem cells

A cytoplasmic RNA virus generates functional viral small RNAs and regulates viral 33 IRES activity in mammalian cells

Six RNA viruses and forty-39 one hosts: viral small RNAs and modulation of small RNA repertoires in vertebrate and 40 invertebrate systems

Novel microRNA-43 like viral small regulatory RNAs arising during human hepatitis A virus infection

MicroRNA-like viral small RNA from Dengue virus 2 autoregulates 46 its replication in mosquito cells

Nile virus encodes a microRNA-like small RNA in the 3' untranslated region 50 which up-regulates GATA4 mRNA and facilitates virus replication in mosquito cells

Noncanonical microRNA 3 (miRNA) biogenesis gives rise to retroviral mimics of lymphoproliferative and 4 immunosuppressive host miRNAs

RNA virus microRNA that mimics a B-cell oncomiR

10 tenOever BR. Influenza A virus-generated small RNAs regulate the switch from transcription to 11 replication

SARS-CoV-2 genomic and subgenomic RNAs 14 in diagnostic samples are not an indicator of active replication

SARS-CoV-2 genomic surveillance 19 identifies naturally occurring truncation of ORF7a that limits immune suppression

A mammalian 32 herpesvirus uses noncanonical expression and processing mechanisms to generate viral

Epub 2010/02/05

Identification of novel 36 microRNA-like molecules generated from herpesvirus and host tRNA transcripts

SARS-CoV-2 ORF6 Disrupts 47 Bidirectional Nucleocytoplasmic Transport through Interactions with Rae1 and Nup98. mBio

Protein Trafficking to Suppress Host Defenses

SARS-CoV-2 Nsp1 binds the ribosomal mRNA channel to 8 inhibit translation

Structural basis for translational shutdown 13 and immune evasion by the Nsp1 protein of SARS-CoV-2

Cutadapt removes adapter sequences from high-throughput sequencing 17 reads

featureCounts: an efficient general purpose program for 22 assigning sequence reads to genomic features

Pall GS, Hamilton AJ. Improved northern blot method for enhanced detection of small 44 RNA

Wei 48 FX, Huang HD. miRTarBase 2020: updates to the experimentally validated microRNA-target 49 interaction database

Analyzing real-time PCR data by the comparative C(T) 11 method

R-CHIE: a web server and R package for 14 visualizing RNA secondary structures

cell lines, n = 3. Colored dots denote miRNAs that significantly (p < 0.05) changed (at least two-