SARS-CoV-2-encoded small RNAs are able to repress the host expression of SERINC5 to facilitate viral replication

Abstract

Serine incorporator protein 5 (SERINC5) is a key innate immunity factor that operates in the cell to restrict the infectivity of certain viruses. Different viruses have developed strategies to antagonize SERINC5 function but, how SERINC5 is controlled during viral infection is poorly understood. Here, we report that SERINC5 levels are reduced in COVID-19 patients during the infection by SARS-CoV-2 and, since no viral protein capable of repressing the expression of SERINC5 has been identified, we hypothesized that SARS-CoV-2 non-coding small viral RNAs (svRNAs) could be responsible for this repression. Two newly identified svRNAs with predicted binding sites in the 3'-untranslated region (3'-UTR) of the SERINC5 gene were characterized and we found that the expression of both svRNAs during the infection was not dependent on the miRNA pathway proteins Dicer and Argonaute-2. By using svRNAs mimic oligonucleotides, we demonstrated that both viral svRNAs can bind the 3'UTR of SERINC5 mRNA, reducing SERINC5 expression in vitro. Moreover, we found that an anti-svRNA treatment to Vero E6 cells before SARS-CoV-2 infection recovered the levels of SERINC5 and reduced the levels of N and S viral proteins. Finally, we showed that SERINC5 positively controls the levels of Mitochondrial Antiviral Signalling (MAVS) protein in Vero E6. These results highlight the therapeutic potential of targeting svRNAs based on their action on key proteins of the innate immune response during SARS-CoV-2 viral infection.

Keywords: MAVS; SARS-CoV-2; SERINC5; innate immune response; viral miRNAs.

Figure 1

Analysis of the levels of SERINC5 mRNA in nasopharyngeal and saliva samples of COVID-19 patients. (A) RT-qPCR analysis of the expression of SERINC5 mRNA in nasopharyngeal (swabs) and saliva samples from COVID-19 patients (P) with respect to healthy patients (C). The control value represents the mean of all control samples. The ΔΔCt method was used for relative quantification using RPP30 mRNA as an endogenous control. Data are represented as log2 fold change with respect to control samples. (B) Correlation analysis between log2 fold changes obtained for SERINC5 mRNA and log2 viral titers in swabs (top) and saliva (bottom) samples. Viral titer was expressed as E mRNA copies/mL sample. Differences from control values were found to be statistically significant at *p < 0.05, **p < 0.01, and ***p < 0.001.

Figure 2

In silico prediction of SARS-CoV-2 svRNAs interacting with SERINC5 mRNA. (A) Identification of svRNA 1 using RNA Central Resources. Sequence and genomic location of svRNA 1 and alignment of svRNA 1 sequence with those of pre-miRNA-431 from different species are shown. The mature miRNA sequences are in grey shadow and the conserved nucleotides among species are in bold letters. (*) ORF10 has so far little experimental support as a protein-coding gene (B) Identification of svRNA 2 by reanalysis of GSE148729 dataset. Count coverage of small RNA sequences from SARS-CoV-2-infected Calu3 cells aligning with the SARS-CoV-2 genome is indicated. The top panel shows the region’s count coverage containing the svRNA 2 at 24 hpi. The red shadow indicates the svRNA 2 sequence. (C) Predicted interaction of svRNAs 1 and 2 with the 3’UTR of SERINC5 mRNA according to the Diana MR-microT tool. Grey shadow indicates nucleotides involved in the interaction. (D) Alignment of SARS-CoV-2 genome-encoded svRNA 1 and 2 sequences (dark shadow) with the genomes of other coronaviruses: MERS-CoV, SARS-CoV, HCoV-229E, HCoV-HKU1, HCoV-OC43, and the SARS-like betacoronavirus Bat coronavirus RaTG13. A higher intensity of the blue shade denotes higher conservation.

Figure 3

Analysis of the levels of svRNA 1 and 2 in nasopharyngeal and saliva samples of COVID-19 patients. (A) RT-qPCR analysis of the expression of svRNA 1 and svRNA 2 in nasopharyngeal (swabs) and saliva samples from COVID-19 patients (P) with respect to control samples (C). The control value represents the mean of all control samples. The ΔΔCt method was used for relative quantification with U6 snRNA as an endogenous control. Data are represented as log2 fold change with respect to control samples. (B) Correlation analysis between log2 fold changes obtained for svRNA 1 or svRNA 2 and log 2 viral titers in swabs (top) and saliva (bottom) samples. (C) Correlation analysis of log2 fold changes obtained for svRNA 1 or svRNA 2 and SERINC5 mRNA in swabs (left) and saliva (right) samples. Differences from control values were found to be statistically significant at *p < 0.05, **p < 0.01, and ***p < 0.001.

Figure 4

Analysis of the expression of svRNAs 1 and 2 and SERINC5 in Vero E6 and HEK293T-hACE2 infected cells. (A) RT-qPCR analysis of svRNAs 1 and 2 expression and correlation analysis between log2 fold changes obtained for svRNA 1 and svRNA 2 and log 2 viral titers in SARS-CoV-2-infected (MOI, 1 PFU/cell) Vero E6 (left panels) and HEK293T-hACE2 (right panels) cells at 4, 8, and 16 hpi. The ΔΔCt method was used for relative quantification, with U6 snRNA as an endogenous control. Data are represented as log2 fold change with respect to mock values and are the mean ± SD of at least three independent experiments. Viral titer was expressed as E mRNA copies/mL sample. (B) RT-qPCR analysis of SERINC5 mRNA expression in SARS-CoV-2-infected (MOI, 1 PFU/cell) Vero E6 (left panel) and HEK293T-hACE2 (right panel) cells at 4, 8, and 16 hpi. The ΔΔCt method was used for relative quantification with RPP30 mRNA as an endogenous control. Data are represented in logarithmic scale as fold change with respect to mock values and are the mean ± SD of at least three independent experiments. (C) Western blot analysis of SERINC5 protein in SARS-CoV-2-infected (MOI, 1 PFU/cell) Vero E6 (left panels) and HEK293T-hACE2 (right panels) cells at 8, 16, and 20 hpi. Blots are representative of at least three independent experiments. The scatter plot shows the densitometric analysis of the protein normalized to β-tubulin and represented as fold change relative to mock. Differences from mock values were found to be statistically significant at *p < 0.05, **p < 0.01, and ***p < 0.001.

Figure 5

Analysis of the levels of SERINC5 mRNA in Vero E6 and HEK293T-hACE2 cells transfected with svRNAs 1 and 2 mimics. (A) RT-qPCR analysis of SERINC5 mRNA level in Vero E6 (left panel) and HEK293T-hACE2 (right panel) cells transfected with an oligonucleotide that mimics the precursor of svRNA 1 or 2 (Pre-svRNA 1 and 2, respectively) or an irrelevant precursor svRNA (NC-pre-svRNA) as a negative control. Data are represented as fold change with respect to values from control samples. (B) RT-qPCR analysis of SERINC5 mRNA level in HEK293T-hACE2 cells co-transfected with Pre-svRNA 1 and a GFP-expressing plasmid. The study was performed on these cells after sorting them into three populations, non-expressing (−GFP), intermediate expressing (+GFP), and highly expressing (++GFP) GFP cells. Data are represented as fold change with respect to values from negative control-transfected cells.

Figure 6

Study of the binding capacity of svRNA 1 and 2 to the 3’UTR of SERINC5 mRNA. Vero E6 cells were co-transfected with a negative control oligonucleotide (NC-pre-svRNA), the oligonucleotide mimic of svRNA 1 or 2 (pre-svRNA 1 and 2) together with the reporter constructs containing the Firefly Luciferase gene fused with the SERINC5-3′UTR in the direct (+) (left panel) or reverse (−) direction (right panel), and the Renilla Luciferase control vector. At 48 h post-transfection the luciferase activity was analyzed and the data obtained were normalized to the levels of Renilla Luciferase (Control of transfection). Differences from control values were found to be statistically significant at *p < 0.05, **p < 0.01, and ***p < 0.001.

Figure 7

Expression levels of svRNAs 1 and 2, viral proteins and cellular SERINC5 in Vero E6 cells treated with mimic or anti-sense molecules for svRNA 1 and 2 and infected with SARS-CoV-2. Vero E6 cells were transfected with specific oligonucleotides mimic for svRNA 1 or svRNA 2 (Pre-svRNA 1 and 2), or specific antisense oligonucleotides for svRNA 1 or svRNA 2 (Anti-svRNA 1 and 2), and at 36 h post-transfection the cells were infected with SARS-CoV-2 (1 PFU/cell). The levels of svRNAs, N, S and SERINC5 proteins were analyzed at 20 hpi and the virus production in the cell supernatants at 24 hpi. (A) RT-qPCR analysis of svRNAs 1 and 2 levels in cells transfected with Pre-svRNA 1 and 2 (left panel), or Anti-svRNA 1 and 2 (right panel). Data are represented as fold change respect to the negative control (NC-pre-svRNA or NC-anti-svRNA)-transfected and SARS-CoV-2-infected cells (control samples). The control value represents the mean of all control samples. (B) Western blot analysis of viral N and S proteins. The scatter plot shows the densitometric analysis of N and S proteins normalized to β-tubulin and represented as fold change relative to negative control-transfected cells. The control value represents the mean of all control samples. (C) Virus production. The viral titers in the cell supernatants were determined at 24 hpi by plaque assay. Data are represented as fold change respect to the negative control (NC-pre-svRNA or NC-anti-svRNA)-transfected and SARS-CoV-2-infected cells (control samples). (D) Western blot analysis of SERINC5. The scatter plot shows the densitometric analysis of SERINC5 protein normalized to β-tubulin and represented as fold change relative to SARS-CoV-2-infected, negative control-transfected cells. The control value represents the mean of all control samples. Differences from negative control values were found to be statistically significant at *p < 0.05, **p < 0.01 and ***p < 0.001.

Figure 8

Analysis of MAVS expression during SARS-CoV-2 infection of Vero E6 and HEK293T-hACE2 cells treated or not with anti-sense molecules for svRNA 1 and 2. (A) Western blot analysis of SERINC5 and MAVS in Vero E6 (left) or HEK293T-hACE2 cells (right) mock-infected or infected with SARS-CoV-2 (MOI of 1 PFU/cell) at 8, 16, and 20 hpi. The scatter plot shows the densitometric analysis of MAVS protein normalized to β-tubulin and represented as fold change relative to non-infected (mock) cells. (B) Western blot analysis of MAVS in Vero E6 cells mock-transfected or transfected with anti-svRNA 1, anti-svRNA 2, or NC-anti-svRNA for 36 h and then infected with SARS-CoV-2 with a MOI of 1PFU/cell for 20 h. The scatter plot shows the densitometric analysis of MAVS and SERINC5 proteins normalized to β-tubulin, represented as fold change relative to non-transfected mock-infected (mock) cells. Differences from control values were found to be statistically significant at *p < 0.05, **p < 0.01, and ***p < 0.001.

Figure 9

Expression levels of IFNβ, ISG20, and CCL20 mRNAs in Vero E6 cells treated or not with anti-sense molecules for svRNA 1. Vero E6 cells were mock transfected or transfected with either a specific antisense oligonucleotide for svRNA 1 (Anti-svRNA1) or an irrelevant oligonucleotide (NC-anti-svRNA) as a negative control. At 36 h post-transfection the cells were infected with SARS-CoV-2 (1 PFU/cell) and the levels of IFNβ, ISG20 and CCL20 mRNAs were analyzed at 18 hpi by RT-qPCR. The ΔΔCt method was used for relative quantification with RPP30 mRNA as an endogenous control. Data are represented as fold change relative to non-transfected mock-infected (mock) cells and are the mean ± SD of at least three independent experiments. Differences from control values were found to be statistically significant at *p < 0.05 and **p < 0.01.