Hypoxia-inducible Factor 1α–dependent Induction of Alveolar miR-147b and Direct Targeting of Viral RNA in SARS-CoV-2–associated Acute Respiratory Distress Syndrome

Abstract

Background:

Acute respiratory distress syndrome (ARDS) causes significant morbidity and mortality during viral pneumonia, including SARS-CoV-2 infections. Nevertheless, most patients with SARS-CoV-2 infections recover seamlessly without developing ARDS, suggesting the existence of endogenous pathways to protect the lungs. Since microRNAs (miRNAs) can regulate endogenous molecular pathways involved in lung protection, the authors hypothesized that alveolar miRNAs could function to dampen SARS-CoV-2–associated lung injury.

Methods:

Screening studies in human alveolar epithelial cells and SARS-CoV-2–infected mice were performed to identify miRNAs induced during infection. Candidate miRNAs were confirmed via reverse-transcription quantitative polymerase chain reaction. The role of hypoxia-inducible factor 1α in regulating miRNA expression was examined in molecular studies. Loss- and gain-of-function approaches in a murine SARS-CoV-2-associated ARDS model were used to assess the physiologic relevance of miRNA. Viral sequence analyses and site-directed mutagenesis were used to determine direct miRNA–viral RNA interactions.

Results:

Screening studies identified miR-147b (hsa-miR-147b-3p or mmu-miR-147-3p) as the leading candidate during infection of human alveolar epithelia or mice with SARS-CoV-2. Functional and molecular studies implicate hypoxia-inducible factor 1α in miR-147b induction during alveolar injury or SARS-CoV-2 infection. Studies in mice with induced deletion of miR-147b in alveolar epithelia (miR147^fl/fl Sftpc-CreER^T2(SPC-CreER) mice) or nanoparticle-mediated miR-147b overexpression revealed a protective role of alveolar-expressed miR-147b during murine SARS-CoV-2–associated ARDS. Moreover, the authors identified the Open Reading Frame 8 (ORF8) region within the SARS-CoV-2 template strand as a direct target of miR-147b, with an ORF8 silent mutation of the SARS-CoV-2 miR-147b–binding site abolishing the observed protection in vitro and in vivo.

Conclusions:

Alveolar epithelial cell–derived miR-147b serves as an endogenous lung protective miRNA against SARS-CoV-2–associated ARDS by directly targeting virus-encoded RNA, revealing a previously unrecognized antiviral and lung-protective mechanism.

miR-147b was identified as an endogenous lung-protective microRNA in SARS-CoV-2–infected human alveolar epithelial cells as well as in mice with SARS-CoV-2–induced lung injury, where it attenuated lung inflammation by reducing viral load. Additionally, functional and molecular studies identified hypoxia-inducible factor 1α as an inducer of miR-147b.

Editor’s Perspective

What We Already Know about This Topic

• Patients with SARS-CoV-2 infection frequently develop acute lung injury, but only a subset progress to acute respiratory distress syndrome. The underlying protective mechanisms, however, remain poorly understood.

• MicroRNAs are known to be involved in lung-protective pathways.

What This Article Tells Us That Is New

• miR-147b was identified as an endogenous lung-protective microRNA in SARS-CoV-2–infected human alveolar epithelial cells as well as in mice with SARS-CoV-2–induced lung injury, where it attenuated lung inflammation by reducing viral load.

• Additionally, functional and molecular studies identified hypoxia-inducible factor 1α as an inducer of miR-147b.

Acute respiratory distress syndrome (ARDS) is a leading cause of morbidity and mortality in patients experiencing critical illness, accounting for more than 500,000 deaths in the United States annually.¹ ARDS is a life-threatening clinical condition of hypoxemic respiratory failure resulting from dysregulated inflammation in response to an initial insult such as infection or trauma.^2,3 Despite advances in supportive care, ARDS outcomes remain ominous, with a 40% mortality, and effective pharmacologic therapies have been elusive.^4–6 Besides the classical Berlin definition, the 2024 New Global Definition of ARDS highlighted a new group, “nonintubated ARDS.”³ Critically, these patients requiring substantial oxygen support have not yet progressed to “intubated ARDS,” the most severe form requiring intubation and mechanical ventilation.³ The most recent definition of ARDS, based on the four-round Delphi study, encompasses both milder and severe forms of ARDS and highlights the heterogeneity of clinical presentations and underlying etiologies.⁷ ARDS is frequently triggered by viral pneumonia, contributing to more than 7 million deaths worldwide during the COVID-19 pandemic caused by SARS-CoV-2 infections.⁸ In this study, we seek to identify molecular mechanisms that influence the severity of lung injury.

Previous studies have shown that microRNAs (miRNAs) are a group of functional RNA molecules strongly implicated in the molecular control of infections.⁹ They constitute a family of short, noncoding RNA molecules, 20 to 25 nucleotides in length, that regulate gene expression posttranscriptionally.¹⁰ Numerous studies suggest that miRNAs can operate in endogenous feedback loops, orchestrating protective responses during organ injury, mucosal inflammation, or viral infection.⁹ For instance, hepatocyte-derived miR-122 is increased during liver ischemia–reperfusion injury and provides liver protection by promoting the stabilization of hypoxia-inducible factor (HIF) 1α.¹¹ Other studies suggest that miR-223 is upregulated during ARDS or mucosal inflammation, providing endogenous mucosal protection.^12,13 The functional role of miRNAs in SARS-CoV-2 is currently under intense investigation. For example, circulating miRNAs have been explored as predictive biomarkers of SARS-CoV-2 infection outcomes,¹⁴ and miR-150-5p has been shown to control viral infection in vitro.¹⁵ However, the molecular function of alveolar epithelial-derived miRNAs during SARS-CoV-2 infection in vivo is poorly characterized.

While some patients develop severe viral pneumonia–associated ARDS, most infected with SARS-CoV-2 develop only mild symptoms and recover seamlessly without significant complications, implying that they can mount an appropriate immune response to limit viral replication and promote pathogen clearance. These findings suggest the existence of molecular pathways critical for endogenous control of viral replication and for alleviating ARDS. Since many miRNAs play important roles in endogenous regulatory pathways, we hypothesized that they function in endogenous lung protection during SARS-CoV-2–associated ARDS.

Materials and Methods

Sex as a Biological Variable

Sex-based differences in responses to murine SARS-CoV-2 infection have been extensively reported in previous publications.^16,17 Our study examined male and female animals, and similar findings for the functional role of miR-147b during SARS-CoV-2 infection were reported for both sexes in the main figures.

Human Subject Research

Human subject research of this study is approved by the Committee for the Protection of Human Subjects at the University of Texas Health Science Center at Houston (UTHealth, Houston, Texas; No. HSC-MS-18-0113). COVID-19 ARDS patients were recruited to the ARDS Biobank (No. HSC-MS-18-0113) for tracheal aspirate samples. De-identified COVID-19–negative tracheal aspirate samples were obtained from Discovery Life Sciences (DSL, USA). The sex and age of the human research participants are included in Supplemental Digital Content table 1 (https://links.lww.com/ALN/E668).

Animal Study

Animal care was performed according to the guide for the care and use of laboratory animals of the National Institutes of Health (Bethesda, Maryland). All experimental procedures were approved by the UTHealth Institutional Animal Care and Use Committee (protocol Nos. AWC-20-0120 and AWC-23-0083). All mice utilized in the study were from breeding colonies within UTHealth. Breeding pairs of C57BL/6J and K18-Human Angiotensin-Converting Enzyme 2 (hACE) mice were initially purchased from The Jackson Laboratory (USA). The miR147^fl/fl mice¹⁸ bred to SPC-CreER (B6.129S-Sftpctm1^{(cre/ERT2)Blh}/J) mice to achieve alveolar type II cell–specific deletion of miR-147b. The SPC-CreER mice were characterized previously for tissue specificity.¹⁹ For miR-147b tissue–specific knockout in the alveolar epithelium, miR-147^fl/fl SPC-CreER mice or SPC-CreER control animals were injected with 50 mg/kg tamoxifen (Cayman Chemical, USA) in corn oil daily for 5 days intraperitoneally as described.¹⁹ Our study examined male and female animals, and similar findings are reported for both sexes.

Cell Culture and SARS-CoV-2 Infection

Calu-3 human airway epithelial cells, HEK-293 cells or human pulmonary epithelial A549 cells were purchased from American Type Culture Collection (USA; HTB-55, RRID:CVCL_0609; CRL-1573; CCL-185). Calu-3 cells were grown in Advanced Minimum Essential Medium (GIBCO, USA), HEK-293 cells were grown in Eagle’s Minimum Essential Medium and A549 cells were grown in Dulbecco’s Modified Eagle Medium (DMEM)–F12 (GIBCO), all cell lines with 10% fetal bovine serum (FBS), and 0.02% l-glutamine. The following reagent was obtained through Biodefense and Emerging Infections Research Resources Repository (BEI Resources), National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH): human lung carcinoma cells (A549) expressing human angiotensin-converting enzyme 2 (A549-hACE2, NR-53821, RRID:CVCL_C0Q5). A549-hACE2 cells were cultured according to BEI Resources’s suggested protocol. Cell lines are free of mycoplasma contamination, as they are initially tested by the vendor for mycoplasma contamination before shipment and then tested annually using polymerase chain reaction (PCR). Human pulmonary alveolar epithelial cells (catalog No. 3200, ScienCell Research Laboratories, USA) were cultured according to the supplier’s instructions. SARS-CoV-2 (Washington-1 [WA1]) was propagated and characterized according to previous publications and recommended protocol from BEI Resources (NR-53821). Infection of SARS-CoV-2 in A549-hACE2 cells was performed as previously described.²⁰ Cells were washed with phosphate buffered saline (PBS) and harvested for RNA/protein isolation or immunofluorescent staining at various time points after the infection. The following reagent was obtained through BEI Resources, NIAID, NIH: SARS-related coronavirus 2, mouse-adapted (MA) 10 variant (in isolate USA-WA1/2020 backbone), infectious clone (ic2019-nCoV MA10) in Calu-3 cells, NR-55329. The following reagent was deposited by the Centers for Disease Control and Prevention (Atlanta, Georgia) and obtained through BEI Resources, NIAID, NIH: SARS-related coronavirus 2, isolate USA-WA1/2020, NR-52281.

BaseScope Staining of Human Lung Tissue

BaseScope staining of pre–miR-147b was designed by ACDBio (Advanced Cell Diagnostics, USA). The signal was detected using the BaseScope Reagent Kit V2–Red according to the supplier’s recommended protocol with the following modification: in step 7, the hybridization time is 45 min instead of the recommended 30 min.

Quantification of SARS-CoV-2 Viral Titer in Lung Tissue

Viral load quantification was performed on Vero E6 cells following the previously described protocol.²¹ Briefly, 6 × 10⁵ cells per well were seeded in six-well plates to achieve 100% confluence overnight in a complete medium (DMEM, 200 mM glutamine, 10% FBS). Samples were prepared by homogenizing equal amounts of lung tissue (milligrams) and PBS (microliters, w/v). Homogenates were centrifuged at 200g, 4°C, for 5 min, and then 10-fold serial dilutions of the virus sample (10⁻¹ to 10⁻⁶) were prepared in infection medium (DMEM, 200 mM glutamine, and 2% FBS). Vero E6 cell monolayers were washed with Dulbecco’s Phosphate-Buffered Saline, and 100 μl of each dilution was added to the wells. Virus adsorption was achieved for 1-h incubation at 37°C, rocking every 15 min. A semisolid overlay medium (containing 1.5% carboxymethyl cellulose) was added to each well, and plates were incubated for 3 days at 37°C, 5% CO₂. Cells were fixed with 10% buffered formalin for 1 h and stained with crystal violet (0.5% in 20% ethanol) for 15 min to visualize plaques. Clear circular plaques on the stained monolayer were counted. The viral titer in plaque-forming units (PFU)/ml was calculated applying the following formula, typically utilizing the dilution that yields 5 to 100 plaques per well: PFU/ml equals number of plaques/volume of dilution used multiplied by dilution factor. SARS-CoV-2 total viral RNA and genomic RNA were measured as described.²²

In Vitro Stretch Model

To study injurious mechanical ventilation in vitro, we performed cyclic mechanical stretch utilizing a previously described procedure.²³ All cells were cultured in six-well plates with 4 ml media per well: Calu-3 cells were grown in Advanced Minimum Essential Medium (GIBCO), and human pulmonary alveolar epithelial cells were grown in Alveolar Epithelial Cell Medium (ScienCell Research Laboratories). Cells were stretched at 30% maximum with 5-s on and off cycles.

Chromatin Immunoprecipitation

Chromatin immunoprecipitation was performed using the chromatin immunoprecipitation assay kit according to the manufacturer’s instructions (Active Motif, USA). Primers used for PCR were as follows: miR-147b sense, CCCAAAGTGTTGGGATTACAGT; antisense, CTTGTGAGGAGTTTGGAGGTTT. PCR products were visualized by running on 1% agarose gels and stained with ethidium bromide.

Small Interfering RNA Transfection in Pulmonary Epithelial Cells

A549-hACE2 cells were transfected with Lipofectamine 2000 (Thermo Fisher Scientific, USA) using the supplier-recommended protocol. Briefly, ON-TARGETplus Human HIF1A (No. 3091) small interfering (siRNA) or ON-TARGETplus Non-targeting Pool (Horizon Discovery, United Kingdom) were transfected in A549-hACE2 cells (30 to 50% confluency) for 48 h. Then cells were infected with 0.2 multiplicity of infection SARS-CoV-2 (WA1) for 24 h before harvesting for RNA isolation and reverse-transcription quantitative PCR (RT-qPCR).

Transcriptional Analysis

Total RNA was isolated from human epithelial cells/cell lines, murine lung tissue, or alveolar epithelial cells using the miRNeasy Micro Kit (Qiagen, Germany) according to the supplier’s recommended protocol. RNA quantification was accomplished by Take3 assay from Cytation 5 Cell Imaging Multimode Reader (Agilent Technologies, USA). RNA concentration was adjusted to 100 ng/µl before proceeding to RT-qPCR. For mRNA gene expression assay, cDNA was synthesized using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, USA). Quantitative PCR (qPCR) was performed using the Taqman probe system (Thermo Fisher Scientific) or SYBR system (Gendepot, USA). miRNA profiling of A549-hACE2 infected with SARS-CoV-2 was performed using Human Panel I, V5, miRCURY LNA miRNA miRNome PCR Panel (Qiagen, No. YAHS-301YE). miRNA profiling of C57BL6 mice infected with SARS-CoV-2 was performed using Mouse Immunopathology Focus, miRCURY LNA miRNA Focus PCR Panel (Qiagen, No. YAMM-204ZD). Measurement of hsa-miR-147b was performed using miRCURY LNA RT Kit and miRCURY LNA miRNA PCR Assay. Measurement of mmu-miR-147-3p was performed using Taqman microRNA assay (Thermo Fisher Scientific).

Hsa-miR-147b Promoter Studies

The promoter assay is performed in A549 cells using a previously established protocol.²⁴ In brief, A549 cells were transfected with luciferase reporter plasmid containing the promoter region of Hsa-miR-147b or control plasmid (pGL3), together with Renilla reporter plasmid. Luciferase signal after mechanical stretch is measured by using Dual Luciferase Reporter Assay (Promega, USA).

Binding of miR-147b to ORF8

HEK293 cells were transfected with ORF8 containing luciferase-expressing plasmid (wild-type [WT] or silent mutation [siMUT], customized by Genecopoeia, USA) and control or miR-147b mimic using the siRNA Plasmid Co-Transfection Protocol with Lipofectamine 2000 (Thermo Fisher Scientific) in a 12-well plate–based format. Forty-eight hours after the transfection, cells were harvested for dual-luciferase assay using Dual Luciferase Assay Kit (Promega) according to the manufacturer’s instructions using a Biotek Synergy 2 Multimode Microplate Reader.

Immunoblotting Experiments

To measure HIF1A, HIF2A, or SARS-CoV-2 ORF8 protein content, Western blotting experiments were performed according to a previously described protocol.²⁵ In brief, cells were harvested in Radio-Immunoprecipitation Assay lysis buffer with protease and phosphatase inhibitor (Roche). The membranes were incubated in primary antibody (HIF1A, Cell Signaling Technology [CST], USA, catalog No. 14179S; HIF2A, Novus, USA, catalog No. NB100-122; LaminB1, CST, catalog No. 12586; SARS-CoV-2 ORF8, R&D Systems, USA catalog No. MAB10820; β-actin, Cell Signaling Technology, catalog No. 5125) at concentrations given for 1 h at room temperature. After three 10-min washes in Tween-Tris Buffered Saline, the membranes were incubated in a secondary antibody (Peroxidase AffiniPure Goat Anti-Mouse IgG, Jackson ImmunoResearch Laboratories Inc., USA, catalog No. 115-035-174; Peroxidase AffiniPure Goat Anti-Rabbit IgG, Jackson ImmunoResearch Laboratories Inc., catalog No. 111-035-144).

mRNA Sequencing Analysis

RNA was isolated from lung tissues in miR-147^fl/fl SPC-CreER and SPC-CreER mice infected with SARS-CoV-2 (MA10) using miRNeasy kit from Qiagen according to the recommended protocol. mRNA sequencing was performed by Cancer Genomic Center at UTHealth. RNA samples were submitted to Cancer Genomics Center at the UTHealth (Cancer Prevention and Research Institute of Texas No. RP180734). Total RNA was quality-checked as previously described.²⁶ RNA with an integrity number greater than 7 was used for library preparation. Libraries were prepared as previously described.²⁶ The quality of the final libraries was examined and then sequenced according to a previously established protocol. The mRNA sequencing data that support the findings of this study are available in the Gene Expression Omnibus (National Center for Biotechnology Information, Bethesda, Maryland) with the identifier GSE217897.

Isolation of Murine Alveolar Epithelial Cells

The protocol to isolate alveolar epithelial cells is adapted from previous publications.²⁷ For alveolar epithelial cell isolation, mice were euthanized. Lungs were lavaged and then perfused with PBS via the right ventricle. Dispase (STEMCELL, Canada, No. 07913) was delivered to the lung via a catheter through tracheotomy. A low-melting-point agarose plug was instilled after dispase. Lungs were then harvested and incubated at room temperature for 1 h. Tissue was dissociated by grinding through a 70-μm filter (Thermo Fisher Scientific) in DMEM supplemented with heat-inactivated FBS and deoxyribonuclease (40 μg/ml). Cells were incubated with antibody cocktails (6.5 µl CD16/32, 20 µl TER119, 15 µl CD45, 15 µl CD90) for 15 min at room temperature. After the incubation, 50 µl EasySep Mouse Streptavidin RapidSpheres beads (STEMCELL) were added and incubated for 15 min. The mixture was attached to EasySep Magnet (STEMCELL) for 5 min, and a negative portion was collected as alveolar epithelial cells.

In Vivo Treatment of miR-147b Mimic

To achieve overexpression of miR-147b in vivo, we utilized systemic delivery of miR-147b mimic via packaging in 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC).^28–30 The overexpression of miR-147b in the lung was confirmed by qPCR measurement of miR-147b levels 24 h after a single dose of 10 µg miRNA-147b mimic (Dharmacon, USA) packaged in DOPC via intraperitoneal injection. To address the function of miR-147b overexpression during SARS-CoV-2 infection, single injection of 10 µg control or miR-147b mimic was injected by intraperitoneal injections every 48 h from day 3 after SARS-CoV-2 infection in mice with angiotensin-converting enzyme 2 (ACE2) receptor driven by the K18 gene promoter (K18-hACE2).

Murine SARS-CoV-2 Infection Model

Murine-adapted SARS-CoV-2 infection (using mice 4 to 6 months old) and infection of K18-hACE2 mice with native SARS-CoV-2 (using mice 2 to 3 months old) were performed according to a protocol from a previous publication³¹ with the following modification. Instead of intranasal instillation, mice were anesthetized by isoflurane and infected via the oropharyngeal route.³² Mock group was anesthetized by isoflurane and received sterile PBS. Mice were monitored and weighed daily and were euthanized in a humane manner according to predetermined criteria in order to reduce distress in accordance with approved protocols. The mouse clinical assessment score variables comprised fur aspect, posture activity level, respiration quality, eyelids, and ear positioning. Clinical symptoms were categorized according to their severity using a scale in which 0 indicated normal, 1 to 4 reflected moderate pain and distress, and 5 or more represented severe pain and distress as a humane endpoint for euthanasia. To obtain bronchial alveolar lavage (BAL) fluid, mice were euthanized by pentobarbital overdose, and tracheostomy was performed to allow the insertion of n 23-gauge blunt-ended catheter. The lungs were lavaged three times with 0.5 ml PBS. All removed fluid was either flash-frozen for total protein measurement or centrifuged to separate the cellular component for the isolation of BAL leukocytes. Total protein in the BAL was measured by the Pierce (USA) Rapid Gold Bicinchoninic Acid Protein Assay Kit according to the manufacturer’s recommended protocol. After BAL collection, the lung was perfused via by instilling 10 ml PBS through the right ventricle. After perfusion, lung tissue was harvested and flash-frozen for RNA isolation or fixed in 4% formalin for histology study. hematoxylin and eosin staining and acute lung injury scoring of the lung tissue were performed blindly as previously described.³³

GenBank-based SARS-CoV-2 Binding Site Analysis

To investigate the association of the miR-147b binding site across different SARS-CoV-2 variants, we analyzed the complete genome sequences of SARS-CoV-2, aligning the motif sequence CGCACA and TGTGCG to identify matches across variants and both strands. We retrieved 3,101,231 genome records from GenBank on February 4, 2025, using the search query (“Severe acute respiratory syndrome coronavirus 2”[Organism] OR SARS-CoV-2[All Fields]) AND “complete genome”[Title] NOT “Homo sapiens”[Primary Organism]. To facilitate the analysis, we first extracted GenBank accession identification numbers and saved them locally via Application Programming Interface, storing them as a dictionary. We then extracted key genomic features, including origin, mature peptide, gene, and coding sequence annotations. The motif-matched sequences were saved in FASTA format. For variant classification, we utilized Nextclade³⁴ to cluster SARS-CoV-2 genomes into clades and World Health Organization (Geneva, Switzerland)–named variants: Alpha, Beta, Delta, Epsilon, Eta, Gamma, Iota, Kappa, Lambda, Mu, Omicron, and Theta. From the downloaded genome data, we identified 12 genes associated with the binding site: E, M, N, S, ORF1a, ORF1b, ORF3a, ORF6, ORF7a, ORF7b, ORF8, and ORF9b. To determine variant-specific gene presence, we compared gene start and end positions with motif sequence coordinates, linking them with GenBank accession identification numbers, gene annotations, and CDS positions. Finally, we quantified the frequency of each gene across the variants.

Generation of Mutated SARS-CoV-2 Virus in miR-147b Binding Site on ORF8

Recombinant ORF8 siMUT SARS-CoV-2 was generated by engineering ORF8 mutations into an infectious cDNA clone of USA-WA1/2020, following our previously established protocols.³⁵ Briefly, ORF8 mutations were introduced into a subcloned plasmid pCC1-F567, which contained 12,308 nucleotides of the 3’ terminus of SARS-CoV-2 genome using a standard overlap PCR with a pair of primers, miR147b-F (5’- AACCAGCCTCATCCACACATAATTCAATTAAAGGTGC-3’) and miR147b-R (5’- GCACCTTTAATTGAATTATGTGTGGATGAGGCTGGTT-3’). ORF8 mutations were underlined in the primer sequences. The resulting pCC1-F567 containing ORF8 mutations was validated through Sanger sequencing. Full-length infectious cDNAs were assembled via in vitro ligation and used as a template for in vitro transcription. The resulting genomic RNA transcripts were electroporated into VeroE6-TMPRSS2 cells. Infectious siMUT viruses were recovered in the supernatants of the transfected cells on day 2 after transfection. The genome sequence of the recovered siMUT virus was confirmed by Sanger sequencing without undesired mutations.

Statistics

Detailed statistical tests are listed in the figure legends. For all statistical analysis except mRNA sequencing, GraphPad Prism (USA) 9.0 software was used. For experiments with numbers less than five, we reported the mean ± SD for parametric data and the median with minimum/maximum for nonparametric data. For experiments with sample sizes of five or greater, a two-tailed P value was shown for analysis using the Student t test (or Welch t test) or the Mann–Whitney test (for nonparametric data); ANOVA-related analysis was corrected for multiple comparisons using the Bonferroni method. The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written. Data analysis of the mRNA sequencing data was performed using Galaxy³⁶ version 22.05.1. Sequence reads were trimmed for adaptor sequence/low-quality sequence using the Galaxy Cutadapt tool (parameter, minimum length, 20; quality cutoff, 20). Trimmed sequence reads were mapped to mm10 using the Galaxy HISAT2 tool. Read count extraction and normalization were performed using the Galaxy featureCounts tool. Differential expression was achieved using the Galaxy limma-voom tool³⁷ (parameter, single count matrix; filter lowly expressed genes, yes; filter on CPM or count values, CPM; minimum CPM, 0.5; minimum samples, 2).

Data Availability

The mRNA sequencing data that support the findings of this study are available in the Gene Expression Omnibus with the identifier GSE217897. All other data are available in the main text or the Supplemental Digital Content (https://links.lww.com/ALN/E668, https://links.lww.com/ALN/E669).

Results

miR-147b (hsa-miR-147b-3p or mmu-miR-147-3p) Induction in Alveolar Epithelial Cells during Respiratory Viral Infection

SARS-CoV-2 infections commonly resolve spontaneously, but the molecular mechanism involved in self-limiting viral replication remains unknown. Therefore, we set up studies to examine whether miRNAs could protect the host from SARS-CoV-2–associated ARDS. Since SARS-CoV-2 directly enters the alveolar mucosa by docking to surface-expressed ACE2, we investigated miRNA expression pattern using a qPCR-based high-capacity array in human pulmonary epithelial cells overexpressing ACE2 (A549-hACE2).²⁰ We found increased spike protein inside the cells 24 h after the initial infection with 0.1 multiplicity of infection SARS-CoV-2 (WA1³⁸; fig. 1A; Supplemental Digital Content fig. 1A, https://links.lww.com/ALN/E669). While several miRNAs were differentially regulated (Supplemental Digital Content table 2, https://links.lww.com/ALN/E668), we found that miR-147b (hsa-miR-147b-3p) had the highest induction in response to SARS-CoV-2 infection (fig. 1, B and C). As a second step, we explored those findings in an in vivo murine model of viral pneumonia. Several models of murine SARS-CoV-2 infection have been previously described.³⁹ Since we were interested in using genetically modified mice on a C57/BL6 background in subsequent functional studies, we determined miRNA expression alterations in the lungs using a murine-adapted SARS-CoV-2 (MA10³⁹), which allows studies in animals under a C57/BL6 background. For this purpose, we infected C57BL6 mice 4 to 6 months old with 2.25 × 10⁵ PFU SARS-CoV-2 (MA10). Murine MA10 infection was associated with severe weight loss (Supplemental Digital Content fig. 1, B and C), increased lung edema (Supplemental Digital Content fig. 1D, mock, 0.1695 ± 0.244 mg/ml; infected, 1.911 ± 0.4927 mg/ml) and acute lung injury score (Supplemental Digital Content fig. 1, E and F). Consistent with our data on human cells, we found that mmu-miR-147-3p (murine homolog of hsa-miR-147b-3p) was the highest induced miRNA (fig. 1, D to F; Supplemental Digital Content table 3, https://links.lww.com/ALN/E668). Together, those studies identified miR-147b as the most highly induced miRNA in infections of human alveolar epithelial cells or in the lungs of mice infected with murine-adapted SARS-CoV-2.

Fig. 1.

Identification of miR-147b as the leading microRNA (miRNA) in SARS-CoV-2 infection. (A) A549-hACE2 cells infected with SARS-CoV-2 (WA1, 0.1 MOI) and immunofluorescent staining of spike and ACE2. Scale bar, 25 µm. (B and C) High-capacity miRNA array of A549-hACE2 cells infected with SARS-CoV-2 (0.1 MOI, 24 h). (B) Volcano plot with red line representing adjusted P value of 0.05 and (C) heat map of the array representing top five induced miRNAs. The red circle represents hsa-miR-147b. n = 3 each group. (D to F) C57BL6 mice 4 to 6 months old were exposed to SARS-CoV-2 infection (MA10, 2.25 × 10⁵ PFU), and mock-infected animals were used as controls. (D) Experimental design, (E) volcano plot grid line representing adjusted P value of 0.05, and (F) heat map of the array representing top five induced miRNAs. Red circle representing mmu-miR-147-3p. n = 3 each group. (G) RT-qPCR confirmation of hsa-miR-147b (n = 6 each group) and (H) primary hsa-miR-147b (n = 3 each group) induction in A549-hACE2 cells infected with 0.1 MOI of SARS-CoV-2 (WA1). Mean ± SD is shown; **P < 0.01; ****P < 0.0001; one-way ANOVA. (I) RT-qPCR analysis of mmu-miR-147-3p in C57BL6 mice 4 to 6 months old infected with 2.25 × 10⁵ PFU SARS-CoV-2 (MA10; n = 4 to 10 each group) and (J) in K18-hACE2 mice 2 to 3 months old infected with 1.4 × 10⁴ PFU SARS-CoV-2 (WA1; n = 4 to 8 each group). Mean ± SD is shown. (K) RT-qPCR analysis of mmu-miR-147-3p (n = 4 or 5 each group) induction in lung alveolar epithelial cells isolated from 4 to 6 months old C57BL6 mice infected with 2.25 × 10⁵ PFU of SARS-CoV-2 (MA10). Mean ± SD is shown. (L) RT-qPCR of hsa-miR-147b in trachea aspirate samples from control or SARS-CoV-2 ARDS patients (n = 10 for control; n = 9 for SARS-CoV-2 ARDS). Median ± minimum/maximum are shown; *P < 0.05; Mann–Whitney test. (M and N) BaseScope staining and quantification of hsa-miR-147b precursor on lung tissues from control or SARS-CoV-2 ARDS patients (n = 5 for control; n = 13 for SARS-CoV-2 ARDS). Median ± minimum/maximum are shown; *P < 0.05; Mann–Whitney test. Scale bar, 26 µm. A549-hACE2, human lung carcinoma cells (A549) expressing human angiotensin-converting enzyme 2; ACE2, angiotensin-converting enzyme 2; MA, mouse-adapted; MOI, multiplicity of infection; PFU, plaque-forming units; RT-qPCR, reverse-transcription quantitative polymerase chain reaction; WA1, Washington-1.

As the next step, we pursued confirmation of the miRNA array findings. First, we observed a robust induction of miR-147b in A549-hACE2 cells after SARS-CoV-2 infection, with increases in both mature and primary miR-147b (pri-hsa-miR-147b; fig. 1, G and H). Real-time RT-qPCR also indicated mature miR-147b induction in murine-adapted SARS-CoV-2 infection (fig. 1I). As a second model of murine SARS-CoV-2 infection, we examined mice with overexpression of hACE2 in lung epithelial cells (K18-hACE2 mice).⁴⁰ Consistent with our previous studies using MA10 infection, we observed significantly increased miR-147b levels in the lungs of K18-hACE2 mice infected with the WA1 strain of SARS-CoV-2 (fig. 1J; fold change compared to mock, day 2, 0.5982 ± 0.4165; day 4, 21.4 ± 36.77; day 7, 1.657 ± 1.76), further validating our initial screening studies. Finally, alveolar epithelial cells isolated from C57BL6 mice infected with MA10 showed up to sixfold increases in miR-147b transcript levels when compared to mock control (fig. 1K; fold change compared to mock, day 2, 2.537 ± 0.5503; day 4, 6.176 ± 1.869). Collectively, these studies identified miR-147b as the highest-induced miRNA in alveolar epithelia after both in vitro and in vivo infections with SARS-CoV-2.

Next, we performed proof-of-principle studies in humans to expand the laboratory findings on miR-147b in endogenous host protection toward patients with SARS-CoV-2 infections. For this purpose, we included patients with a positive SARS-CoV-2 PCR test who were hospitalized for COVID-19 treatment (patient demographics are displayed in Supplemental Digital Content table 1, https://links.lww.com/ALN/E668). As the first step, RT-qPCR indicated increased levels of miR-147b in tracheal aspirates of intubated COVID-19 ARDS patients compared to COVID-19–negative/culture-negative intubated controls (fig. 1L). Furthermore, we used BaseScope⁴¹ to display the expression of miR-147b directly in the lungs of deceased SARS-CoV-2-associated ARDS patients. Indeed, we found that miR-147b is enriched in the lungs of SARS-CoV-2-associatedARDS patients compared to SARS-CoV-2–negative control lungs obtained during thoracic surgery (fig. 1, M and N). These studies suggest that miR-147b is increased in human SARS-CoV-2 ARDS.

Induction of miR-147b during Alveolar Inflammation Involves HIF1A

Next, we examined the transcriptional mechanism governing miR-147b induction during alveolar inflammation. Initially, we used cyclic mechanical stretch, which models injurious mechanical ventilation,²³ as a screening tool to identify transcriptional responses during alveolar injury. First, we investigated the induction of miR-147b during injurious cyclic mechanical stretch in primary human alveolar epithelial cells. Notably, we found increased miR-147b expression as early as 2 h after the initiation of mechanical stretch in primary alveolar epithelial cells (fig. 2A; fold change compared to 0 h, 2 h, 5.003 ± 1.834; 4 h, 2.392 ± 1.974; 8 h, 1.750 ± 0.4751; 24 h, 2.890 ± 2.369). Additional studies in Calu-3 cells also demonstrated robust induction of miR-147b after mechanical stretch (Supplemental Digital Content fig. 2A, https://links.lww.com/ALN/E669). Due to their high transfection efficiency, Calu-3 cells were chosen as the cell line for further molecular studies. As the next step, we used the Signosis (USA) Promoter-Binding Transcription Factor Profiling Array⁴² to screen for potential transcription factor candidates. The promoter binding transcription factor 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 or other inflammatory conditions.^43,44 Since previous studies provided evidence for the functional stabilization of HIF during ARDS,^23,25,33 we further investigated its role in miR-147b induction during alveolar injury or SARS-CoV-2 infection.

Fig. 2.

HIF1A-dependent induction of miR-147b during alveolar injury and SARS-CoV-2 infection. (A) Expression level of hsa-miR-147b in human primary alveolar epithelial cells after cyclic mechanical stretch (30%). n = 3. Mean ± SD is shown. (B) Transcription factor binding array of miR-147b promoter in Calu-3 cells 2 h after cyclic mechanical stretch (30%). (C) Schematic of promoter construct of miR-147 demonstrating HRE and NF-κB responsive elements. (D) Plasmid with the constructs shown in c was transfected into Calu-3 cells for 24 h. Dual-luciferase activity was measured after 18 h of mechanical stretch (Firefly/Renilla). n = 3. Mean ± SD is shown. (E) Chromatin immunoprecipitation (ChIP) assay to examine for direct binding of HIF1A to the HRE regions within the hsa-miR-147b promoter regions during mechanical stretch (8 h) in Calu-3 cells (gels represent two individual experiments). (F) Calu-3 cells were exposed to ambient hypoxia (1% oxygen) and expression level of hsa-miR-147b was measured by RT-qPCR. n = 6. Mean ± SD is shown. *P < 0.05. One-way ANOVA. (G) A549 cells were exposed to ambient hypoxia (1% oxygen), and expression level of hsa-miR-147b was measured by RT-qPCR. n = 5. Mean ± SD is shown. *P < 0.05; **P < 0.01. One-way ANOVA. (H) Calu-3 cells were treated with dimethyloxaloglycine (DMOG; 1 µm, 24 h), and expression level of hsa-miR-147b was measured by RT-qPCR. n = 6. Mean ± SD is shown. **P < 0.01. Mann–Whitney test. (I) A549 cells were treated with DMOG (1 µm, 24 h), and expression level of hsa-miR-147b was measured by RT-qPCR. n = 5. Mean ± SD is shown. **P < 0.01. Mann–Whitney test. (J) Protein levels of HIF1A or HIF2A in A549-hACE2 cells after SARS-CoV-2 infection (0.1 MOI) by Western blot. (K and L) Quantification of HIF1A and HIF2A in A54-hACE2 cells after SARS-CoV-2 infection (0.1 MOI). n = 9. Mean ± SD is shown. ns P > 0.05; **P < 0.01; **** P < 0.0001. One-way ANOVA. (M) A549-hACE2 cells with siRNA-mediated repression of HIF1A were exposed to SARS-CoV-2 infection (WA1; 0.2 MOI) or mock-infected. Hsa-miR-147b transcript levels were examined by real-time RT-PCR. n = 3 or 4. Mean ± SD is shown. hACE2,human angiotensin-converting enzyme 2; HIF, hypoxia-inducible factor; HRE, hypoxia-response element; MOI, multiplicity of infection; NF, nuclear factor; ns, not significant; RT-PCR, reverse-transcription polymerase chain reaction; RT-qPCR, reverse-transcription quantitative polymerase chain reaction; siRNA, small interfering RNA; WA1, Washington-1.

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 sequence.⁴⁵ Transcription factor binding analysis (TFSearch, https://www.cbrc.jp/research/db/TFSEARCH.html) identified consensus binding sites for multiple transcription factors, including three HIF (hypoxia response element, HRE) binding sites and one binding site for nuclear factor (NF)–κB, both of which had been previously implicated in transcriptional regulation of gene expression during ARDS.^46–48 We observed a robust induction of the full-length promoter (1809 bp) during cyclic mechanical stretch exposure of Calu-3 cells transfected with promoter constructs. 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 bp (fig. 2, C and D; full-length promoter, 1809 bp, control 2.02 ± 0.219; stretch 4.463 ± 2.112; truncated promoter, 882 bp, control 1.067 ± 0.552; stretch 1.7 ± 1.187). Given that this region contained a hypoxia response element, and HIF1A is known to be induced during cyclic mechanical stretch,⁴⁹ we next performed chromatin immunoprecipitation assay and confirmed direct binding of HIF1A to the hsa-mir-147b promoter in the HRE located between 1809 and 882 bp (fig. 2E). To further investigate the role of HIF in the induction of miR-147b, we exposed pulmonary epithelial cell lines to ambient hypoxia (1% oxygen; fig. 2, F and G) or HIF activator dimethyloxaloylglycine (fig. 2, H and I) and observed induction of hsa-miR-147b. Our results have identified a previously unrecognized molecular regulatory pathway governing the induction of hsa-miR-147b, involving direct interaction between HIF1A and the HRE located between 1809 and 882 bp in the promoter region of hsa-miR-147b.

Next, we extended the studies of stretch-induced alveolar epithelial injury to SARS-CoV-2 infection. Here, we examined the contribution of HIF in miR-147b induction during SARS-CoV-2 infection. Western blot analysis demonstrated a robust induction of HIF1A but not HIF2A (fig. 2, J to L). The inducibility of hsa-miR-147b was completely blocked after siRNA-mediated repression of HIF1A (fig. 2M; fold change compared to mock of respective siRNA treatment, control siRNA, 60.291 ± 44.015; HIF1A siRNA, 1.555 ± 0.244; Supplemental Digital Content fig. 2, B and C, https://links.lww.com/ALN/E669, fold change compared to mock of respective siRNA treatment, control siRNA, 41.288 ± 13.94; HIF1A siRNA, 18.14 ± 4.784). Together, these findings identify HIF1A as a novel transcriptional regulator of miR-147b during alveolar injury or SARS-CoV-2 infection.

Deletion of miR-147b in Alveolar Epithelia Increases Susceptibility to SARS-CoV-2 Infection

To investigate whether alveolar epithelial miR-147b induction functions as an endogenous feedback mechanism protecting the lungs during SARS-CoV-2 infection, we generated transgenic mice with a floxed miR-147b locus¹⁸ to create a novel mouse line with inducible alveolar type II–specific deletion of miR-147b (miR147^fl/fl SPC-CreER mice) under a C57BL/6J background. To induce miR-147b deletion, we treated 4- to 6-month-old miR147^fl/fl SPC-CreER mice and SPC-CreER control animals matched for age, sex, and weight with daily tamoxifen injections for 5 days, then rested the animals for 10 days. The efficiency of miR-147b deletion in alveolar epithelial cells was confirmed by RT-qPCR (fig. 3, A and B, SPCCreER, 1 ± 0.5764; miR147^fl/fl SPC-CreER, 0.1004 ± 0.05487). To study the functional role of alveolar epithelial-derived miR-147b during murine SARS-CoV-2 infection, we used a murine-adapted strain of virus (MA10) that induces lung pathology in 4- to 6-month-old animals on a C57BL/6J background.³⁹ After infection of SARS-CoV-2 (MA10, 2.25 × 10⁵ PFU), miR147^fl/fl SPC-CreER mice developed increased lung edema at day 3 after infection compared to SPC-CreER mice marked by elevated total protein in the BAL (fig. 3, C and D). Furthermore, we measured the levels of cytokines and chemokines in the BAL fluid using MILLIPLEX (USA) Mouse Cytokine/Chemokine Magnetic Bead Panel Immunology Multiplex Assay with the following analytes: granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), interferon-gamma (IFN-γ), interleukin (IL)-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-9, IL-10, IL-12p40, IL-12p70, IL-13, IL-15, IL-17, interferon gamma-induced protein 10 (IP-10/CXCL10), keratinocyte-derived chemokine (KC/CXCL1), leukemia inhibitory factor (LIF), lipopolysaccharide-induced CXC chemokine (LIX/CXCL5), monocyte chemoattractant protein-1 (MCP-1/CCL2), macrophage colony-stimulating factor (M-CSF), monokine induced by gamma interferon (MIG/CXCL9), macrophage inflammatory protein-1α (MIP-1α/CCL3), macrophage inflammatory protein-1β (MIP-1β/CCL4), macrophage inflammatory protein-2 (MIP-2/CXCL2), regulated upon activation, normal T cell expressed and secreted (RANTES/CCL5), tumor necrosis factor-alpha (TNF-α), vascular endothelial growth factor (VEGF), and eotaxin (CCL11). Surprisingly, we found that the majority of the cytokine and chemokines were unaltered comparing miR147^fl/fl SPC-CreER and SPC-CreER control mice, except for the significantly increased KC and reduced IL-4 levels in the miR147^fl/fl SPC-CreER (Supplemental Digital Content fig. 3A, https://links.lww.com/ALN/E669). Blinded lung pathologic scoring indicated that lung injury was more severe in miR147^fl/fl SPC-CreER mice (fig. 3, E and F; Supplemental Digital Content fig. 3B). To investigate the role of miR-147b in viral replication and clearance, we measured the titer of live virus in this model. We found significant increases in viral titer in the lungs of SARS-CoV-2 (MA10)–infected mice, measured by plaque assay, in miR147^fl/fl SPC-CreER (fig. 3G), implicating the importance of miR-147b in controlling viral load in the lung. To determine whether inducible deletion of miR-147b in alveolar epithelial cells results in baseline lung injury, we further characterized lung injury in mock-infected mice. We observed no significant changes in pulmonary edema (Supplemental Digital Content fig. 4, A and B, SPCCreER, 99.94 ± 16.13 μg/ml; miR147^fl/fl SPC-CreER, 87.4 ± 10.72 μg/ml) or lung injury scores (Supplemental Digital Content fig. 4C) between miR147^fl/fl SPC-CreER mice and SPC-CreER mice, demonstrating that the exaggerated lung injury seen in miR147^fl/fl SPC-CreER mice during SARS-CoV-2 infection is not due to baseline changes in the lung.

Fig. 3.

Functional role of miR-147b in SARS-CoV-2 infection. (A and B) miR-147^fl/fl SPC-CreER or age/sex-matched SPC-CreER control mice were treated with intraperitoneal (i.p.) injection of tamoxifen (50 mg ·per kg per day) for 5 days and rested for 10 days. (B) Alveolar epithelial cells were isolated, and mmu-miR-147 level was measured by RT-qPCR. n = 4 or 5. Mean ± SD is shown. (C) Alveolar type 2 (ATII) cell–specific deletion of miR-147b was achieved by tamoxifen injection. SPC-CreER and miR-147^fl/fl SPC-CreER mice 4 to 6 months old matched in age, sex, and weight were injected with tamoxifen (50 mg/kg) for 5 days, rested for 10 days, and exposed to SARS-CoV-2 infection (MA10, 2.25 × 10⁵ PFU). (D) The total protein of the BAL was measured by Rapid Gold BCA assay. n = 10 and 13. Mean ± SD is shown. **P < 0.01. Unpaired Student t test. (E) Representative histology image of the lung. Scale bar, 150 µm. (F) Blinded acute lung injury scoring of the lung tissue. n = 11 and 12. Median ± minimum/maximum are shown. *P < 0.05. Student t test. (G) Titer of live virus in the lung was measured by plaque assay. n = 8 and 14. Median ± minimum/maximum are shown. *P < 0.05. Mann–Whitney test. (H) C57BL6 mice 2 to 3 months old were injected with 10 µg intraperitoneal control miR-DOPC or miR-147b–DOPC, and miR-147b levels in the lung were measured by RT-qPCR at 48 h after injection. n = 4. Mean ± SD is shown. (I) K18-hACE2 mice 2 to 3 months old were infected with SARS-CoV-2 (WA1, 280 PFU) and treated with control miR-DOPC or miR-147b-DOPC every 48 h starting at day 3 after infection and (J) probability of survival. n = 14 each group. P value is from log-rank (Mantel–Cox) test. (K) K18-hACE2 mice 2 to 3 months old were infected with SARS-CoV-2 (WA1, 280 PFU) and treated with control miR-DOPC or miR-147b–DOPC every 48 h starting at day 3 after infection, and samples were harvested at day 5. Representative of hematoxylin and eosin staining in lung tissues from control or miR-147b mimic treated mice. Scale bar, 100 µm. (L) Blinded acute lung injury scoring of the lung tissue. n = 8 and 7. Median ± minimum/maximum are shown. *P < 0.05. Mann–Whitney test. (M) Titer of live virus in the lung was measured by plaque assay. n = 7. Mean ± SD is shown. *P < 0.05. Unpaired Welch t test. BCA, bicinchoninic acid; Ctrl, control; DOPC, 1,2-dioleoyl-sn-glycero-3-phosphocholine; K18-hACE2, ACE2 receptor driven by the K18 gene promoter; MA, mouse-adapted; MOI, multiplicity of infection; PFU, plaque-forming units; RT-qPCR, reverse-transcription quantitative polymerase chain reaction; SPC-CreER, Sftpc-CreER^T2; WA1, Washington-1.

In addition to alveolar epithelial cells, immune cells are activated and recruited to the lung during respiratory viral infections. To investigate the contribution of infiltrating leukocytes on the induction and functional roles of miR-147b, we isolated BAL leukocytes and measured expression level of miR-147b at days 2 and 4 after SARS-CoV-2 infection (MA10) in C57BL6 mice (Supplemental Digital Content fig. 5A, https://links.lww.com/ALN/E669). In contrast to alveolar epithelial cells (fig. 1), we did not observe any induction of miR-147b in BAL leukocytes (Supplemental Digital Content fig. 5B, fold change compared to mock, day 2, 0.019 ± 0.007; day 4, 0.036 ± 0.032). Furthermore, miR-147b levels in the BAL are reduced in miR147^fl/fl SPC-CreER mice (Supplemental Digital Content fig. 5, C and D, fold change compared to SPCCreER, 1 ± 0.4201; miR147^fl/fl SPC-CreER, 0.2982 ± 0.1273) but remain unaltered in miR147^fl/fl lysozyme M-Cre recombinase (LysMCre) mice compared to Cre plus control mice during SARS-CoV-2 infection (Supplemental Digital Content fig. 5, E and F, fold change compared to LysMCre, 1 ± 1.878; miR147^fl/fl LysMCre, 0.7071 ± 0.3915), suggesting that alveolar epithelial cells instead of infiltrating immune cells can release miR-147b in the BAL. Finally, myeloid-specific deletion of miR-147b was not associated with any significant increases in lung injury susceptibility, as miR147^fl/fl LysMCre mice showed similar levels of pulmonary edema (Supplemental Digital Content fig. 5G) and histological lung injury compared to LysMCre mice (Supplemental Digital Content fig. 5H). Taken together, these results suggest that the deletion of miR-147b in alveolar epithelial cells is associated with increased susceptibility to lung injury during SARS-CoV-2 infection in vivo.

Nanoparticle-mediated Overexpression of miR-147b Protects against SARS-CoV-2 Infections

After demonstrating that alveolar deletion of miR-147b is associated with more severe lung injury during SARS-CoV-2–associated ARDS, we next pursued studies targeting miR-147b therapeutically. We used DOPC²⁸ nanoliposomes to achieve pulmonary overexpression of miR-147b and directly assess its therapeutic potential during SARS-CoV-2 infection in vivo. Previous studies suggested that the intraperitoneal administration of DOPC is an effective means of systemic miRNA delivery, likely through portal circulation.²⁹ Thus, we first treated mice with an intraperitoneal injection of DOPC-packaged miR-147b mimic or control mimic and demonstrated that miR-147b mimic treatment significantly increased the endogenous level of miR-147b in the lung (fig. 3H, fold change compared to control mimic, control, 1 ± 0.3364; miR-147b mimic, 2.443 ± 0.5112). After establishing successful overexpression of miR-147b, we treated K18-hACE2 mice at day 3 after SARS-CoV-2 (WA1) infection with DOPC-packaged miR-147b mimic or control mimic every 48 h, and then tracked clinical outcomes (fig. 3I). Posttreatment with miR-147b mimic resulted in a significant improvement in clinical score at day 5 after infection (Supplemental Digital Content fig. 6, A and B, https://links.lww.com/ALN/E669). Strikingly, we observed significantly improved survival in mice treated with miR-147b mimic (fig. 3J). Blinded histologic injury scores of the lungs indicated that miR-147b mimic treatment is associated with attenuated histologic lung injury on day 5 during the infection, before the mortality starting on day 7 (fig. 3, K and L; Supplemental Digital Content fig. 6C). Furthermore, we observed an overall reduction of viral load in the lung in mice treated with miR-147b mimic compared to control mimic, as marked by reduced live virus titer by plaque assay on day 5 after infection (fig. 3M). These results suggested that miR-147b overexpression in vivo improves outcomes from SARS-CoV-2 infection by higher survival, reduced lung injury, and decreased viral load in the lungs. These studies further support the lung protective role of miR-147b during SARS-CoV-2 infection in vivo.

miR-147b-mediated Lung Protection Involves Direct Targeting of Viral Genome

To further examine the mechanism of miR-147b–dependent protection and the associated reduction in viral load, we performed mRNA sequencing in lung tissues from miR147^fl/fl SPC-CreER and SPC-CreER mice at 3 days after infection (fig. 4A). To search for potential host target genes for miR-147b, we cross-referenced induced genes from the mRNA sequencing in miR147^fl/fl SPC-CreER mice with predicted miR-147b targets using different in silico prediction platforms (Targetscan, microRNA DataBase, Supplemental Digital Content tables 4 and 5, https://links.lww.com/ALN/E668). Surprisingly, we found no overlap from the target prediction scans with differentially regulated genes in miR147^fl/fl SPC-CreER mice (fig. 4B). We further investigated whether miR-147b could significantly alter host target gene expression in SARS-CoV-2 infection by comparing expression changes in miR-147b target genes to the distribution of expression changes in all other host genes. For these two subsets, the distributions of logFC values between control and miR147^fl/fl SPCCre mice were not significantly different by the Kolmogorov–Smirnov test (fig. 4C). Overall, these results indicate that the lung protection mediated by miR-147b is unlikely to be caused by a direct targeting of host target genes.

Fig. 4.

Identification of SARS-CoV-2 ORF8 as miR-147b target. (A and B) mRNA sequencing of RNA isolated from lung tissue in SPC-CreER or miR-147^fl/fl SPC-CreER mice and overlap between upregulated genes from mRNA sequencing in miR-147^fl/fl SPC-CreER mice 4 to 6 months old compared to SPC-CreER control infected with SARS-CoV-2 (MA10, 2.25 × 10⁵ PFU) with predicted mmu-miR-147b target genes. (C) Cumulative frequency distribution of miR-147b–predicted targets and nontarget host genes. (D) Heatmap of SARS-CoV-2 genes overlapping miR-147b binding motifs across variants showing the frequency of 12 SARS-CoV-2 genes containing the miR-147b–associated binding motifs (CGCACA and TGTGCG) across 16 major World Health Organization–classified SARS-CoV-2 variants and clades. MA, mouse-adapted; ORF8, open reading frame 8; PFU, plaque-forming units; SPC-CreER, Sftpc-CreER^T2.

Previous studies have shown that miRNAs can target the viral genome directly.⁵⁰ Thus, we hypothesized that miR-147b may directly interact with the SARS-CoV-2 viral genome to protect the lungs. To test this hypothesis, we aligned the motif sequence CGCACA (positive strand) or TGTGCG (template strand) against 3,101,231 complete SARS-CoV-2 genome records retrieved from GenBank on February 4, 2025, and quantified the frequency of each motif-containing gene across World Health Organization–named variants⁵¹ (see Materials and Methods for full details). Our findings reveal that ORF8, ORF1ab, and E are the most frequently observed genes across all major SARS-CoV-2 variants, suggesting a potential role in viral evolution and adaptation (fig. 4D). As shown in Supplemental Digital Content figure 7 (https://links.lww.com/ALN/E669), among those targets, the ORF8 region on the template strand of WA1 SARS-CoV-2 had the most nucleotide matches with miR-147b, with a total of 13 matches. Therefore, we focused on the functional role of the interaction between miR-147b and the ORF8 region of the viral genome.

Silent Mutation of the miR-147b Binding Site within SARS-CoV-2 ORF8 Prevents miR-147b–elicited Lung Protection

To further investigate the direct targeting of miR-147b on SARS-CoV-2 ORF8, we designed two luciferase reporter vector plasmids containing WT ORF8 or ORF8 with silently mutated miR-147 binding sites (fig. 5A) to investigate their interaction in vitro. First, the successful overexpression of miR-147b in HEK-293 cells was confirmed by RT-qPCR (Supplemental Digital Content fig. 8, https://links.lww.com/ALN/E669, fold change compared to control mimic, control, 1 ± 0.2868; miR-147b mimic, 57,009 ± 16,583). As second step, WT or silently mutated ORF8 dual-luciferase reporter vectors were cotransfected together with control or miR-147b mimic into HEK-293 cells. Twenty-four hours after cotransfection, luciferase reporter activity was measured using a dual-luciferase assay. Overexpression of hsa-miR-147b was associated with decreased luciferase reporter activity, suggesting repressed expression of ORF8 (fig. 5B). Importantly, miR-147b–mediated repression was abolished after silent mutation of the miR-147b binding site within ORF8, suggesting direct binding of hsa-miR-147b to ORF8 in vitro (fig. 5B).

Fig. 5.

Establishment of SARS-CoV-2 ORF8 as miR-147b target. (A) Sequence of silent mutation (siMUT) of miR-147b binding site on SARS-CoV-2 ORF8 for luciferase assay (B). HEK-293 cells were cotransfected with ORF8 WT or siMUT luciferase vectors and control or miR-147b mimic. Luciferase activity was measured by dual-luciferase reporter assay 48 h after the transfection (Firefly normalized to Renilla). n = 6 to 9. Mean ± SD are shown. *P < 0.05; ns P > 0.05. Two-way ANOVA. (C) Design of the mutated SARS-CoV-2 virus with silent mutation of miR-147b on the WA1 background (ORF8 siMUT). (D) K18-hACE2 mice 2 to 3 months old were infected with SARS-CoV-2 (ORF8 siMUT, 280 PFU) and treated with control miR-DOPC or miR-147b–DOPC every 48 h starting at day 3 after infection. (E) Probability of survival. n = 10 and 13. P value from log-rank (Mantel–Cox) test. (F and G) Blinded acute lung injury scores and representative hematoxylin and eosin staining of the lung. n = 9 or 10. Median ± minimum/maximum are shown. ns P > 0.05. Mann–Whitney test. Scale bar, 100 µm. Ctrl, control; ns, not significant; ORF8, open reading frame 8; RLU, relative luciferase unit; WA1, Washington-1; WT, wild-type.

Since our in vitro studies provided strong evidence supporting the direct targeting of SARS-CoV-2 ORF8 by miR-147b, we next investigated a functional role for the miR-147b–ORF8 link in vivo. As a first step to examine the targeting of ORF8 by miR-147b in vivo, we generated a mutated SARS-CoV-2 using a WA1 backbone with a silent mutation of the miR-147b binding site within the ORF8 gene (ORF8^miR147mut, fig. 5C). This approach allows for disruption of miR-147b binding to the ORF8 without altering the amino acid sequence of the translated ORF8 protein. To determine the impact of the silent mutation of ORF8 on miR-147b–mediated lung protection, K18-hACE2 mice were infected with ORF8^miR147mut SARS-CoV-2 and treated with DOPC packaged miR-147b mimic or control mimic every 48 h on day 3 after infection and tracked the clinical outcomes for 7 days (fig. 5D), the same schedule as figure 3. In support of a functional role for miR-147b in targeting ORF8, miR-147b mimic treatment failed to improve survival or alleviate lung injury after infection with ORF8^miR147mut (fig. 5, E to G; Supplemental Digital Content fig. 9, https://links.lww.com/ALN/E669), suggesting that silent mutation of miR-147b binding site on ORF8 abolished miR-147b mediated lung protection in vivo. Together with our findings above, these studies identify a previously unrecognized role for miR-147b in lung protection during SARS-CoV-2 infection, both in vitro and in vivo.

Discussion

The molecular functions of host-derived miRNAs in endogenous lung protection during viral pneumonia, including SARS-CoV-2–associated lung injury, remain poorly characterized. Combining miRNA screening approaches in human alveolar epithelia and in mice infected with SARS-CoV-2, we identified miR-147b as a leading miRNA during SARS-CoV-2 infection, fundamental to the antiviral response. Studies examining the transcriptional regulation of miR-147b identified a novel role for HIF1A as a critical transcription factor in miR-147b induction during alveolar injury or SARS-CoV-2 infection. Mice with alveolar epithelial deletion of miR-147b showed exaggerated lung injury and inflammation after SARS-CoV-2 infection, associated with attenuated antiviral responses and increased viral burden. Furthermore, pharmacologic enhancement of miR-147b using nanoparticulate-mediated overexpression improved survival during an in vivo SARS-CoV-2 challenge. Mechanistically, miR-147b promoted antiviral responses by interacting with SARS-CoV-2 ORF8. These findings identified a previously unrecognized role for miR-147b as an endogenous feedback mechanism that protects the lung during SARS-CoV-2 infection by attenuating viral replication.

The current study identified HIF1A as a key transcriptional regulator of miR-147b induction during alveolar injury caused by SARS-CoV-2 infection. These findings align with previous studies indicating a protective role for HIF1A in pathogen-associated ARDS. One of the first studies to implicate HIFs in lung protection during pathogen-induced lung injury demonstrated that mice with bacterial ARDS experienced better clinical outcomes when kept at a lower oxygen concentration (50% oxygen) than at higher levels.⁴⁶ The authors attributed their findings to the effects of iatrogenic hyperoxia on the stabilization of HIFs in the lungs.⁵² Other studies have implicated HIF1A as crucial for the bactericidal activity of myeloid cells, as myeloid-selective deletion of HIF1A results in systemic infection spread. At the same time, pharmacologic HIF activators enhance bactericidal capacity, which is associated with HIF-dependent enhancement of macrophage phagocytosis.⁵³ Another molecular mechanism of HIF1A-dependent antimicrobial activity relates to the production of antimicrobial peptides during epithelial defense against pathogens.^54–56 Furthermore, two recent studies have suggested a protective role for HIF stabilization during SARS-CoV-2 infection, as exposure to ambient hypoxia or treatment with the HIF activator roxadustat suppresses viral replication in lung epithelial cells in vitro.⁵⁷ Moreover, the HIF activator roxadustat protects hamsters infected with SARS-CoV-2 in vivo by reducing epithelial damage and increasing the number of ciliated cells.²² Finally, a recently published randomized clinical trial using the Food and Drug Administration (Silver Spring, Maryland)–approved HIF activator vadadustat for the treatment of lung injury in patients with SARS-CoV-2 infection found a very high likelihood of lung protection with vadadustat treatment, particularly in patients admitted with severe hypoxia.²⁵ These findings are consistent with our current studies, suggesting that HIFs may drive the induction of microRNAs to repress mediators potentiating lung injury and inflammation. Notably, the most significant clinical improvement with the HIF stabilizer was observed in patients with severe hypoxia, not in those with lower levels of hypoxia. Thus, future studies are needed to determine whether only the most hypoxic patients with SARS-CoV-2 infection exhibit miR-147b induction. However, several mechanisms beyond microRNA induction may have contributed to clinical improvement in patients, given the complex nature and heterogeneity of human infection.

Several mechanisms could stabilize HIF during injurious mechanical ventilation or SARS-CoV-2 infection. In the context of injurious mechanism ventilation, previous studies from our laboratory have shown that elevation of succinate can function as an endogenous HIF activator.^23,58 During severe SARS-CoV-2 infection, HIF stabilization can occur during hypoxic conditions in the lung as patients with SARS-CoV-2 ARDS experience profound hypoxia. Furthermore, tissue hypoxia can result from changes in metabolic supply and demand,⁵⁹ which could be due to oxygen depletion by inflammatory cell infiltration into the lung during viral infection or inflammation. For example, important studies from Dr. Sean Colgan’s laboratory have shown that activated neutrophils, when undergoing oxidative burst, can deplete the adjacent tissues of oxygen, causing “hypoxia imprinting” on epithelial cells.⁶⁰ Future studies are needed to elucidate the mechanisms underlying HIF stabilization during SARS-CoV-2 infection.

miR-147b (hsa-miR-147b or mmu-miR-147-3p) is located on chromosome 15 in humans and chromosome 2 in mice. miR-147b has not been widely studied compared to other more well-established miRNAs (e.g., miR-155⁶¹ and miR-146a⁶²). Thus, it was surprising that miR-147b was the most induced miRNA after SARS-CoV-2 infection both in vitro and in vivo. Similarly, the substantial functional roles of miR-147b in attenuating SARS-CoV-2–associated ARDS in vivo were unexpected. Previous studies of miR-147b have focused on cancer biology, where it has been shown to mediate a metabolic shift in cancer cells that confers drug tolerance⁶³ and to regulate the cell cycle.⁶⁴ A recent study found that miR-147b alleviates inflammatory responses in lung epithelial cell lines (A549) by dampening p38 Mitogen-Activated Protein Kinase signaling upon lipopolysaccharide exposure.⁶⁵ Additionally, miR-147b–dependent repression of NADH:Ubiquinone Oxidoreductase Subunit A4 promotes antitumor responses in tumor-associated macrophages by remodeling mitochondrial complex IV.⁶⁶ The targeting of NADH:Ubiquinone Oxidoreductase Subunit A4 by miR-147b has also been observed during inflammatory bowel diseases.⁶⁷ Furthermore, previous studies demonstrate that miR-147b overexpression in macrophages dampens Toll-like receptor activation–induced cytokine production in vitro.⁴⁵ While the current findings identifying miR-147b as an attenuator of SARS-CoV-2 viral replication were surprising, additional studies are needed to elucidate miR-147b’s diverse roles in other physiologic or pathologic conditions, including its functions in other tissue compartments.

Our study has a few limitations. First, our study focuses on miR-147b derived from alveolar epithelial cells in the lung. However, since SARS-CoV-2 infects many different cell types across various organs, it would be beneficial to determine whether the specific viral genome-targeting mechanisms exist in other cells in which SARS-CoV-2 replicates. Furthermore, our study highlights lung injury and inflammation as the primary outcomes. It would be interesting to investigate how miR-147b alters the lung immune landscape during SARS-CoV-2 infection or impacts COVID-19–associated fibrosis.⁶⁸ Although our study focuses on the specific interaction between miR-147b and the SARS-CoV-2 ORF8, many other viral genome targets could interact with miR-147b. Thus, further studies on those interactions will offer a more comprehensive view of how miR-147b interacts with SARS-CoV-2. Moreover, additional mechanisms exist to decrease pathogen burden in alveolar epithelial cells. For example, lipid mediator elovanoid inhibits SARS-CoV-2 binding to the alveolar epithelial cells to confer lung protection during infection.⁶⁹ How miR-147b contributes to these pathways requires additional investigation. In addition, some of the studies included in the manuscript are not powered for sex-based analysis, warranting further investigations. Finally, additional studies in human ARDS that include analysis of predisposing lung conditions of patients, and their heterogeneity in terms of genetics and other medical problems, are needed to further investigate the functional role of miR-147b.

The current studies implicate pulmonary miRNAs in an endogenous host protective pathway during SARS-CoV-2 infection. Indeed, induction of miR-147b via a transcriptional mechanism under the control of HIF1A is associated with attenuated lung inflammation and pulmonary edema by reducing viral replication. From a treatment standpoint, this pathway could be enhanced by nanoparticle-mediated delivery of miR-147b to the lungs or by pharmacologic HIF activators, which have shown promising results in a recent randomized clinical trial for SARS-CoV-2 infection.²⁵ Future studies might explore the use of pharmacologic HIF activators as a novel therapeutic strategy for ARDS associated with other non–SARS-CoV-2 pathogens.

Conclusions

To identify endogenous protective pathways involving miRNAs in alveolar epithelial cells during viral infection–associated ARDS, we performed an initial screening in human alveolar epithelial cells infected with SARS-CoV-2. We identified miR-147b (hsa-miR-147b-3p) as the highest-induced miRNA. Subsequent screening studies in mice infected with murine-adapted SARS-CoV-2 virus suggested its murine homolog (mmu-miR-147-3p) as the highest expressed miRNA during lung infection in vivo. Studies using combinations of genetic and pharmacologic approaches suggest that miR-147b induction is dependent on HIF1A, and miR-147b plays a crucial role in controlling viral load in the lungs. Surprisingly, we found that induction of miR-147b in alveolar epithelial cells directly targets the viral genome within the ORF8 region. Upon silent mutation of the viral ORF8 binding site for miR-147b, the protection provided by miR-147b overexpression was abolished. As such, our findings reveal an unexpected role for alveolar epithelial cell–derived miR-147b in the molecular control of viral replication by targeting the microbial genome. Our study only provided preclinical evidence, and the therapeutic efficacy of targeting this pathway will need to be investigated in patient clinical trials.

Acknowledgments

The authors thank the technical support from the Cancer Prevention and Research Institute of Texas (Houston, Texas; No. RP180734). The following reagent was obtained through BEI Resources, National Institute of Allergy and Infectious Diseases, National Institutes of Health (NIH; Bethesda, Maryland): SARS-related coronavirus 2, mouse-adapted, MA10 variant (in isolate USA-WA1/2020 backbone), infectious clone (ic2019-nCoV MA10) in Calu-3 cells, NR-55329, contributed by Ralph S. Baric (Department of Epidemiology, Gillings School of Global Public Health; Department of Microbiology and Immunology, School of Medicine; University of North Carolina at Chapel Hill, Chapel Hill, North Carolina). The following reagent was deposited by the Centers for Disease Control and Prevention (Atlanta, Georgia) and obtained through BEI Resources, NIAID, NIH: SARS-related coronavirus 2, isolate USA-WA1/2020, NR-52281. The following reagent was obtained through BEI Resources, NIAID, NIH: human lung carcinoma cells (A549) expressing human angiotensin-converting enzyme 2, NR-53821.

Research Support

The research is supported by National Institutes of Health (NIH; Bethesda, Maryland) grant No. R01HL154720; American Lung Association (Chicago, Illinois) Catalyst Award No. CA-622265, and an 2022 International Anesthesia Research Society (Walnut Creek, California) Mentored Research Award. Dr. Vidal Melo is supported by the NIH/National Heart, Lung, and Blood Institute (Bethesda, Maryland; grant No. R01 HL121228). Dr. Makarevich is supported by Department of Defense (Arlington, Virginia) grant No. W81XWH2110032 and a Parker B. Francis Fellowship (Kansas City, Missouri). Dr. Sood is supported by Onxeo, Foundation Medicine, Corcept, StarPharma, Kaida, DSMB (Advenchen, Mural Oncology), and a research grant from Pfizer.

Competing Interests

Drs. Eltzschig, Potnuru, and Yuan receive research funding through a contract between Akebia Therapeutics (Cambridge, Massachusetts) and UTHealth (Houston, Texas) to support a clinical trial on the effect of vadadustat in ARDS patients (No. NCT07086755). The other authors declare no competing interests.

Supplemental Digital Content

Supplemental tables, https://links.lww.com/ALN/E668,

Supplemental table 1, Patient demographic information.

Supplemental table 2, Top differentially regulated microRNAs in SARS-CoV-2 infection in vitro.

Supplemental table 3, Top differentially regulated microRNAs in SARS-CoV-2 infection in vivo.

Supplemental table 4, Differentially regulated genes for RNA sequencing of miR-147^fl/fl SPC-CreER mice compared to SPC-CreER mice during SARS-CoV-2 infection.

Supplemental table 5, In silico predicted target genes for mmu-miR-147.

Supplemental figures, https://links.lww.com/ALN/E669,

Supplemental figure 1, Establishment of SARS-CoV-2 infection models in mice.

Supplemental figure 2, Transcriptional induction of miR-147b by HIF1A during alveolar injury or SARS-CoV-2 infection.

Supplemental figure 3, Additional phenotype of miR-147^fl/fl SPC-CreER mice during SARSCoV-2 infection.

Supplemental figure 4, Baseline phenotype in mock-infected miR-147^fl/fl SPC-CreER mice.

Supplemental figure 5, Immune cell-derived miR-147b didn’t contribute significantly to its induction or functional role during SARS-CoV-2 infection.

Supplemental figure 6, miR-147b mimic attenuates SARS-CoV-2 infection sequelae.

Supplemental figure 7, Binding of SARS-CoV-2 viral genome targets for miR-147b.

Supplemental figure 8, Transfection efficiency of miR-147b mimic in HEK293 cells.

Supplemental figure 9, miR-147b mimic does not attenuate ORF8 siMUT SARS-CoV-2 infection sequelae.

Supplementary Material

aln-145-1204-s001.pdf ^{(3.1MB, pdf)}