Acute respiratory distress syndrome (ARDS) is widely recognized in patients with hypoxemic respiratory failure, and bilateral pulmonary infiltrates, not attributable to left heart failure3. Viral pneumonia is one of the major causes of ARDS, which showed significant morbidity and mortality during the COVID-19 pandemic4-6, and attribute to more than 6 million deaths worldwide, including 1.07 million deaths in the US alone and rising. SARS-CoV-2 docks onto alveolar-expressed angiotensin-converting-enzyme 2 (ACE2), causing direct epithelial injury in COVID-19-associated ARDS7. Therefore, understanding the host response to SARS-CoV-2 infection in alveolar epithelial cells would be of the highest significance.
Interestingly, despite the severity of COVID-19-associated ARDS, most of the population infected with SARS-CoV-2 recovered without hospitalization. According to the COVID-19 hospital data available from the Centers for Disease Control and Prevention, the percentage of hospitalized COVID-19 patients to develop ARDS is less than 10% (https://www.cdc.gov/nchs/covid19/nhcs/other-respiratory-illnesses.htm, accessed on 9/23/22). Thus, it is conceivable that although viruses usually develop sophisticated mechanisms to evade the immune system, most patients successfully launch the proper immunologic response to restrict viral replication and promote viral clearance. Therefore, we hypothesize the existence of endogenous host-protective pathways that are important to combat viral-mediated immune evasion and prevent the development of severe COVID-19.
MicroRNAs (miRNAs) constitute a family of short non-coding RNA. molecules of 20 to 25 nucleotides in length that regulate gene expression at the posttranscriptional level8. Many studies indicated that miRNAs could serve endogenous tissue protective pathways during organ injury or mucosal inflammation9-12. For instance, hepatocyte-derived miR-122 is increased during liver ischemia-reperfusion injury and offers liver protection via enhancing of hypoxia signaling pathways10. Other studies implicate miR-223 to be upregulated during ARDS or intestinal inflammation while providing endogenous mucosal protection11,12. Similarly, host-derived miRNAs have been involved in regulating antiviral immunity and restricting viral replication. For example, miR-485 targets viral mRNAs during influenza virus infection to control viral replication13. However, how host miRNAs interact with SARS-CoV-2 remains poorly understood.
To identify miRNA-based endogenous pathways in alveolar epithelial cells during COVID-19-associated ARDS, we performed an initial screening of human alveolar epithelial cells infected with SARS-CoV-2. We identified miR-147 as the highest-induced miRNA, which was validated using a mouse model of SARS-CoV-2 infection. A combination of genetic and pharmacologic treatment studies indicates that miR-147 is critical for promoting proper antiviral responses in airway epithelia. miR-147 induction in alveolar epithelial cells represses viral ORF8 and represents an endogenous adaptive pathway to launch the appropriate antiviral immune responses during SARS-CoV-2 infection.
miR-147 (hsa-miR-147b or mmu-miR-147-3p) induction in alveolar epithelial cells during SARS-CoV-2 infection
SARS-COV-2 infections commonly resolve spontaneously but the mechanisms involved in self-limiting infection remain unknown. We examined whether miRNAs could protect the host from SARS-CoV-2-associated ARDS. Since SARS-CoV-2 directly targets the alveolar mucosa by docking to surface-expressed ACE2, we examined miRNA expressional pattern using a qPCR-based high-capacity array in human pulmonary epithelial cells overexpressing ACE2 (A549-hACE2)14-16. We found increased spike protein inside the cells 24 hours after the initial infection with 0.1 MOI SARS-CoV-2 (WA117) (Fig. 1a, Extended Data Fig. 1a). While several miRNAs were differentially regulated (Supplementary table 1), we found miR-147 (hsa-miR-147b) had the highest induction in response to SARS-CoV-2 infection (Fig. 1b, c). Several models of murine SARS-CoV-2 infection have been developed in the past two years18-20. Thus, we next determined expressional alterations of miRNAs in the lungs using a murine-adapted SARS-CoV-2 (MA1020). C57BL6 mice were infected with 2.25x105 PFU of SARS-CoV-2 (MA10). MA10 infection was associated with severe lung injury and concomitant increases in inflammation (Extended Data Fig. 1b-e). Consistently, we found that miR-147 (mmu-miR147-3p) was the highest induced miRNA (Fig. 1d-f, and Supplementary table 2). Thus, we identified miR-147 as top induced miRNA in infections of alveolar epithelial cells, and in murine models of SARS-CoV-2 infection. Next, we pursued further confirmation of the miRNA arrays. Firstly, we confirmed robust induction of miR-147 in A549-hACE2 cells after SARS-CoV-2 infection with increases in both mature miR-147 and primary miR-147 (pri-hsa-miR-147b) (Fig. 1g, h). Real-time RT-PCR also indicated mature miR-147 induction in murine-adapted SARS-CoV-2 infection (Fig. 1i). K18-hACE mice that show severe ARDS upon SARS-CoV-2 infection21 (Extended Data Fig. 1f-h) and significantly increased miR-147 levels in their lungs (Fig. 1j) further validating our findings. Finally, alveolar epithelial cells isolated from C57BL6 mice infected with MA10 or mock condition showed increased 6-fold miR147 transcript levels when compared to mock control (Fig. 1k). Thus, we identified miR-147 as the highest-induced miRNA in alveolar epithelia following both in vitro and in vivo infections with SARS-CoV-2.
Induction of miR-147 during alveolar injury or SARS-CoV-2 infection involves hypoxia-inducible factor HIF-1A
We next examined the transcriptional mechanism of miR-147 induction in SARS-CoV-2 infection. We used cyclic mechanical stretch22, as a screening tool to identify transcriptional responses to alveolar injury. Firstly, we confirmed the induction of miR-147 during cyclic mechanical stretch in primary human alveolar epithelial cells. Notably, we found increased miR-147 expression as early as 2 hours after the initiation of mechanical stretch in primary alveolar epithelial cells (Fig. 2a). Additional studies in Calu3 cells also confirmed robust induction of miR-147 following mechanical stretch (Extended Data Fig. 2a). For further molecular studies, Calu3 was chosen as the cell line due to the high transfection efficiency. To screen for potential transcription factor candidates, we carried out a Signosis Promoter-Binding TF Profiling Array23. The promoter binding TF array pointed us toward HIF as the top transcriptional inducer with the highest luciferase activity (Fig. 2b). HIF is an essential transcription factor for tissue adaptation to hypoxia as well as other physiological conditions24-27. Since previous studies provided evidence for the functional stabilization of HIF during ARDS22,28, we further pursued its role in miR-147 induction during alveolar injury or SARS-CoV-2 infection.
The sequence of human pri-miR-147b is located within exon 4 of the normal mucosa of esophagus-specific gene 1 (NMES1), and is downstream of the NMES1-coding sequence29. Furthermore, transcription factor binding analysis (TFSearch, http://www.cbrc.jp/research/db/TFSEARCH.html) indicated consensus sites for multiple transcription factors, including three binding sites for hypoxia-inducible factor (HIF1A; hypoxia response element, HRE) and for NF-kB, which had both been previously implicated in the transcriptional regulation of gene expression during ARDS30. We demonstrated robust induction of the full-length promoter (1809 bp). However, all shorter fragments (882, 470, 367 bp) lost stretch inducibility, indicating a “hot” (or critical) area within the promoter located between 1809 and 882 bps (Fig. 2c, d). To further investigate the role of HIF in the induction of miR-147, we exposed pulmonary epithelial cell lines to ambient hypoxia (1% oxygen; Extended Data Fig. 2b, d) or treatment with the HIF activator dimethyl-oxaloylglycine (DMOG) (Extended Data Fig. 2c, e), and observed induction of hsa-miR-147b.
Next, we extended the above studies of stretch-induced alveolar epithelial injury to SARS-CoV-2 infection. Here, we examined the contribution of HIF in miR-147 induction during SARS-CoV-2 infection. Western blot analysis demonstrated a robust induction of HIF1A, but not HIF2A (Fig. 2e-g). Functionally, the inducibility of hsa-miR-147b was completely blocked following siRNA-mediated repression of HIF1A (Fig. 2h, Extended Data Fig. 2f, g). Together, these findings identify HIF1A as a novel transcriptional regulator of miR-147 during alveolar injury or SARS-CoV-2 infection.
Deletion of miR-147 in alveolar epithelia increases susceptibility to ARDS in SARS-CoV-2 infection
To investigate whether alveolar epithelial miR-147 induction functions as an endogenous feedback mechanism protecting the lungs during SARS-CoV-2 infection, we generated transgenic mice with a floxed miR-147 locus31 to create a novel mouse line with alveolar-specific deletion of miR-147 (miR147fl/fl SPC-CreER mice). To induce miR-147 deletion, we treated miR147fl/fl SPC-CreER mice or SPC-CreER control animals matched for age, gender and weight with daily tamoxifen injections for 5 days. The efficiency of miR-147 deletion in alveolar epithelial cells was confirmed by RT-qPCR (Extended Data Fig. 3a). After infection of SARS-CoV-2 (MA10, 2.25x105), miR147fl/fl SPC-CreER developed increased lung injury at day 3 post-infection marked by increased ALI score and total protein in the BAL (Fig 3a- c, and Extended Data Fig. 3b), suggesting that alveolar epithelial miR-147 provides lung protection. To investigate the role of miR-147 on viral replication in alveoli we measured viral spike protein expression in this model. We found significant increases in viral load as shown by increased spike RNA level in miR147fl/fl SPC-CreER (Fig. 3d). To further investigate the functional mechanism of miR-147-elicited lung protection we next sequenced mRNA from the lung tissues of infected mice. We identified 40 genes that were induced and 46 genes that were repressed upon MA10-SARS-CoV-infection of miR147fl/fl SPC-CreER mice (Supplementary table 3). Pathway enrichment indicated several biological/immunological gene clusters were highly differentially regulated (Fig. 3e, Extended Data Fig. 3c). Most notably, several antiviral pathways were significantly attenuated in miR147fl/fl SPC-CreER mice indicating acute immune suppression. We validated the 5 most differentially regulated leading edge genes of the top five pathway by RT-qPCRs (Fig. 3f, Supplementary table 4), which revealed a remarkable decrease of MHCII molecules in miR147fl/fl SPC-CreER mice, implicating an immune-compromised phenotype. These studies indicate a protective role for alveolar miR-147 by promoting antiviral responses during SARS-CoV-2 infection in vivo.
Identification of SARS-CoV-2 ORF8 as a key mRNA target of miR-147.
Because we found that miR-147 governs proper anti-viral responses during SARS-CoV-2 infection, we next sought to identify specific miR-147 mRNA targets32. Therefore, we cross-referenced induced genes from the mRNA seq in miR147fl/fl SPC-CreER mice with predicted miR147 targets using different in silico prediction platforms (Targetscan, miRDB, supplementary table 5). Notably, we did not find any match from the target prediction scans with differentially regulated genes in miR147fl/fl SPC-CreER mice (Extended Data Fig. 4). Prior studies have shown that miRNAs can target the viral genome33,34, therefore, we next examined whether miR-147 can directly target SARS-CoV-2 RNA. A manual search for potential miRNA targets in the SARS-CoV-2 genome (GenBank: ON311289.1) identified three potential miR-147 binding sites that expressed a complementary sequence to the miR-147 seed sequence, including ORF1ab, envelop protein (E), and ORF8 (Fig. 4a, Extended Data Fig 4). Among those targets, only ORF8 had been previously implicated as a critical mediator for immune evasion of SARS-CoV-235,36. Therefore, we focused on miR-147-ORF8 targeting and measured ORF8 levels in miR147fl/fl SPC-CreER mice after SARS-CoV-2 infection. ORF8 showed increased levels in mice with alveolar epithelial deletion of miR-147 (Fig. 4b). We next used two plasmids containing containing WT37 (Addgene # 141278) or mutated38 (Addgene # 141390) miR-147 binding site (Fig. 4c) to confirm the targeting in vitro. Following co-transfection of WT or mutated ORF8 together with control or miR-147 mimic into HEK293 cells, successful overexpression of ORF8 and miR-147 was confirmed (Extended Data Fig. 4). Over-expression of hsa-miR-147b was associated with decreased ORF8 mRNA expression (Fig. 4d). However, this response was abolished following mutation of ORF8 in complementary sequence to seed sequence of miR-147, which accounts for hsa-miR-147b-binding to its target genes (Fig. 4d). These findings identify SARS-CoV-2 ORF8 as a previously unidentified miR-147 target. In conjunction with the above results showing dramatic increases in viral load following the alveolar deletion of miR-147 (Fig. 3d), those findings implicate ORF8 targeting as a mechanism for miR-147-mediated immunity during SARS-CoV-2 infection.
Overexpression of miR-147 by DOPC nanoliposomes provides lung protection during SARS-CoV-2 infection.
After demonstrating that miR-147 targeting of ORF8 dampens SARS-CoV-2 immune evasion, we next pursued studies targeting miR-147 therapeutically. Here we used nano-particle-mediated miR-147 overexpression as an intervention to dampen SARS-CoV-2-mediated immune evasion in vivo. We utilized 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC)39 nanoliposomes to achieve pulmonary overexpression of miR-147 to directly address the therapeutic potential of miR-147 overexpression during SARS-CoV-2 infection in vivo. Firstly, we treated mice with i.p. injection of DOPC packaged miR-147 mimic or control mimic and demonstrated that miR-147 mimic treatment significantly increased the endogenous level of miR-147 in the lung (Fig. 5a). After confirming successful overexpression of miR-147, we treated K18-hACE2 mice at day 3 post SARS-CoV-2 infection with DOPC packaged miR-147 mimic or control mimic every 48 hours and then tracked for the clinical outcomes for 11 days. MiR-147 mimic resulted in an improvement in survival (Fig. 5b, c). Furthermore, additional animals were harvested at day 7 and lung injury scoring indicated that miR-147 mimic attenuates lung injury (Fig. 5d, e). miR-147 mimic treatment also results in reduction of ORF8 in the lung (Fig. 5f) with increased expression of MHCII molecule H2-Ab1 (Fig. 5g). Taken together, these studies implicate miR-147 overexpression in improving outcomes during SARS-CoV-2 infection in vivo.
miR-147 in human COVID-19 ARDS
As the final step in our studies, we performed proof-of-principle studies to expand the above laboratory studies on miR-147 in endogenous host responses to SARS-CoV-2 toward COVID-19 patients. Patient demographic information is included in Supplementary table 6. Firstly, we confirmed the expression of miR-147 by BaseScope40 (ACDBio) to directly stain the precursor of miR-147 in lung histological slides from COVID-19 ARDS patients. Indeed, miR-147 is enriched in lung tissue from COVID-19 ARDS patients (Fig. 6a) when compared to COVID-19 negative controls. Additional staining are displayed in Extended Data Fig. 5. Furthermore, RT-qPCR indicated increased levels of miR-147 in tracheal aspirates of COVID-19 ARDS patients compared to COVID negative controls (Fig. 6b). To further study the functional role of miR-147 in human COVID-19 ARDS, we correlated the key leading edge genes equivalent from the mRNAseq in miR147fl/fl SPC-CreER mice with the expression level of miR-147 in tracheal aspirate samples. Consistently, higher miR-147 levels correlated with increased expression of HLA-DRB6 (Fig. 6c), suggesting enhanced antiviral responses. These proof-of-principle studies identified elevated levels of miR-147 in patients with COVID-19 ARDS and suggest a functional role in human SARS-CoV-2 infections.