Low expression of EXOSC2 protects against clinical COVID-19 and impedes SARS-CoV-2 replication

Association of lung gene expression with risk of COVID-19

Association testing was performed using FUSION (Gusev et al, 2016) using pre-computed weights for lung gene expression from GTEx (v7) (Lonsdale et al, 2013). All eQTLs were used regardless of significance. We used the ANA7 phenotype from DF2 which included 1,294 individuals with clinical evidence of COVID-19 and 26,969 asymptomatic controls.

Affinity purification of Nsp8-associated complexes and mass spectrometry analysis

HEK293T cells were transfected with Strep-Nsp8 (Gordon et al, 2020) and Strep-Nsp8 together with untagged Nsp7 using PEI for 24 h. Mock-transfected cells were used for control purifications. Cell pellets were collected, proteins lysates were prepared, and affinity purification was performed using four biological replicates of each group as previously described (Gordon et al, 2020). Elution of affinity-purified proteins was performed by incubation of MagStrep “type 3” beads with 50 μl of elution buffer (5% SDS, 50 mM Tris, pH 7.4) at 70°C for 15 min. Protein reduction was performed by adding TCEP to a final concentration of 5 mM and incubation at 70°C for 15 min and alkylation by adding iodoacetamide to a final concentration of 10 mM and incubation at 37°C for 30 min. Sample clean-up was performed using suspension trapping (S-Trap) according to the manufacturer’s instructions (ProtiFi). Tryptic digestion was performed by adding 1 μg of trypsin (sequencing grade; Pierce) and incubating at 47°C for 60 min. Eluted peptides were dried to completion in a vacuum concentrator (Eppendorf).

Samples were re-suspended in 40 μl of 0.5% formic acid, and 18 μl was analysed by nanoflow LC-MS/MS using an Orbitrap Elite (Thermo Fisher Scientific) hybrid mass spectrometer equipped with a nanospray source, coupled to an Ultimate RSLCnano LC System (Dionex) at the biOMICS facility at the University of Sheffield. The system was controlled by Xcalibur 3.0.63 (Thermo Fisher Scientific) and DCMSLink (Dionex). Peptides were desalted on-line using an Acclaim PepMap 100 C18 nano/capillary BioLC, 100A nanoViper 20 mm × 75 μm I.D., and particle size 3 μm (Thermo Fisher Scientific) at a flow rate of 5 μl/min and then separated using a 125-min gradient from 5% to 35% buffer B (0.5% formic acid in 80% acetonitrile) on an EASY-Spray column, 50 cm × 50 μm ID, PepMap C18, 2 μm particles, and 100 Å pore size (Thermo Fisher Scientific) at a flow rate of 0.25 μl/min. The Orbitrap Elite was operated with a cycle of one MS (in the Orbitrap) acquired at a resolution of 60,000 at m/z 400, with the top 20 most abundant multiply charged (2+ and higher) ions in a given chromatographic window subjected to MS/MS fragmentation in the linear ion trap. An FTMS target value of 1 × 10⁶ and an ion trap MSn target value of 1 × 10⁴ were used with the lock mass (445.120025) enabled. Maximum FTMS scan accumulation time of 100 ms and maximum ion trap MSn scan accumulation time of 50 ms were used. Dynamic exclusion was enabled with a repeat duration of 45 s with an exclusion list of 500 and an exclusion duration of 30 s.

Mass-spectrometry data analysis

Raw data files were processed using MaxQuant (Version 1.6.10.43) (Tyanova et al, 2016). Data were searched against a combined human and SARS-CoV-2 UniProt sequence database (Dec 2019) using the following search parameters: digestion set to trypsin/P with a maximum of two missed cleavages, oxidation (M), N-terminal protein acetylation as variable modifications, cysteine carbamido-methylation as a fixed modification, match between runs enabled with a match time window of 0.7 and a 20-min alignment time window, label-free quantification was enabled with a minimum ratio count of 2, minimum number of neighbours of 3, and an average number of neighbours of 6. A first search precursor tolerance of 20 ppm and a main search precursor tolerance of 4.5 ppm were used for FTMS scans and a 0.5-D tolerance for ITMS scans. A protein FDR of 0.01 and a peptide FDR of 0.01 were used for identification-level cut-offs.

Protein group output files generated by MaxQuant were loaded into Perseus version 1.6.10.50. The matrix was filtered to remove all proteins that were potential contaminants, only identified by site and reverse sequences. The LFQ intensities were then transformed by log2(x), normalised by subtraction of the median value, and individual intensity columns were grouped by experiment. Proteins were filtered to keep only those with a minimum of three valid values in at least one group. The distribution of intensities was checked to ensure standard distribution for each replicate. Missing values were randomly imputed with a width of 0.3 and downshift of 1.8 from the SD. To identify significant differences between groups, two-sided t test were performed with a permutation-based FDR of 0.05.

Calu-3 cell culture

Calu-3 cells were cultured in DMEM/F12 (1:1) GlutaMAX (Thermo Fisher Scientific) supplemented with 10% FBS, 1% NEAA, 1% sodium pyruvate, 1% penicillin-streptomycin, and maintained at 37°C, 5% CO₂, and passaged with TrypLE Express 1X (Thermo Fisher Scientific) when ∼80% confluent. All experimental work was performed on cells within the range of 20–30 passages.

CRISPR-Cas9 editing

An sgRNA targeting exon 1 of EXOSC2 (5′-GAUACAAUCACUACGGACAC-3′) was designed using the CRISPOR tool (http://crispor.tefor.net/) (Concordet & Haeussler, 2018). Design was guided by available protospacer adjacent motifs and predicted on- and off-target efficiencies. A validated, commercially available sgRNA targeting HPRT (IDT) was used as a CRISPR control. sgRNA duplexes were assembled from tracrRNA and crRNA in a thermocycler according to manufacturer’s instructions under RNase-free conditions. Cells were cultured to ensure 70–90% confluency on the day of transfection. 24-well plates containing 500 μl of antibiotic-free (DMEM)/F12 (1:1) GlutaMAX (Thermo Fisher Scientific) were incubated at 37°C. CRISPR/Cas9 ribonucleoproteins were formed by complexing 240 ng sgRNA duplex with 1,250 ng Alt-R V3 Cas9 Protein (IDT) in 10 μl buffer R (Neon Transfection System 10 μl Kit; Thermo Fisher Scientific)—a 1:1 M ratio—for 10 min at RT. 100,000 viable cells were aliquoted per transfection and centrifuged at 400g for 5 min at RT. Cells were washed in calcium- and magnesium-free Dulbecco’s Phosphate Buffered Saline (Sigma-Aldrich) and centrifuged at 400g for 5 min at RT. Cell pellets were resuspended in 10 μl buffer R containing Cas9 protein and sgRNA duplexes. 2 μl of 10.8 μM electroporation enhancer (IDT) was added, and the solution mixed thoroughly to ensure a suspension of single cells. 10 μl of this mixture was loaded into a Neon transfection system (Thermo Fisher Scientific) and electroporated according to manufacturer’s instructions (1,400 V, 2 pulse, 20 s pulse width). Cells were then transferred to pre-warmed media in 24-well plates.

To assess editing efficiency, genomic DNA was isolated from CRISPR-edited and control cells using a GenElute Mammalian DNA Miniprep Kit (Sigma-Aldrich) according to manufacturer’s instructions. A ∼400-bp region around the expected Cas9 cut site in exon 1 of EXOSC2 was amplified by PCR using PfuTurbo DNA polymerase (Agilent), according to the manufacturer’s instructions using primers: AGGCCGTGAGTTCTCATTGG (fwd) and GGGTTTCAGGGAGCTGAGAC (rvs) (Sigma-Aldrich). Expected amplification was confirmed using gel electrophoresis, and the products were Sanger-sequenced (Source BioScience). Sequencing trace files were uploaded to TIDE (http://shinyapps.datacurators.nl/tide/) and ICE (https://ice.synthego.com), and an indel efficiency calculated.

MTT assay

Cell viability was assessed using the MTT Assay Kit (Abcam) according to the manufacturer’s instructions. Viable cells but not dead cells can metabolise MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) to an insoluble formazan product; when solubilised, the compound can be read at OD590 nm. The measured absorbance is proportional to the number of viable cells. 60,000 Calu-3 cells were seeded in each well of a 96-well plate. After 48 h, the culture medium was discarded, and 50 μl of fresh serum-free medium and 50 μl MTT reagent were added to each well, including to a background (no cells) control. The plates were returned to the 37°C incubator for 3 h before the MTT/media solution was discarded and 150 μl of MTT solvent was added to each well. The plates were incubated for a further 15 min in foil on a hula mixer, and the absorbance recorded at 590 nm. The replicate values were corrected for background and averaged.

Cloning and viral transduction

The human EXOSC2 open-reading frame was amplified from HEK293 cDNA using oligonucleotides: gctagcATGGCGATGGAGATGAGGC (fwd) and ctcgagTCCCTCCTGTTCCAAAAGCCT (rvs) and cloned as a NheI/XhoI PCR fragment into the NheI/XhoI restriction sites of a lentiviral self-inactivating transfer vector (SIN) containing a woodchuck hepatitis virus post regulatory element (W) to overexpress EXOSC2 under a PGK promoter (pLV_SIN-W-PGK-EXOSC2). All plasmids were validated by Sanger sequencing, and sequences are available upon request.

HEK293T cells were used for lentiviral production, plated at a density of 3 × 10⁶ per 10-cm dish. Cells were transfected using a calcium chloride transfection containing 0.5 M calcium chloride (Sigma-Aldrich), 2× HEPES Buffered Saline (Sigma-Aldrich), and four lentiviral component plasmids; pCMV delta 8.2 (13 μg), pRSV-Rev (3 μg), pMD.G (3.75 μg) (Addgene), and pLV_SIN-W-PGK-EXOSC2 (13 μg) (Déglon et al, 2000). Transfection mix added dropwise to each plate and left overnight, with a full media change carried out the following morning. Cells were incubated for a further 48 h before all media was collected and filtered using a 0.45-μm filter (Sigma-Aldrich). Equally loaded tubes (Beckman Coulter) were then spun at 19,000 rpm/90 min/4°C using an ultracentrifuge and a SW28 hanging rota (Beckman Coulter). All supernatant was removed, and each viral pellet was resuspended with 300 μl of 1% bovine serum albumin (Tocris Bioscience) in Phosphate Buffer Solution (Sigma-Aldrich). Each tube was incubated on ice for 1 h and then combined into one homogeneous solution before being aliquoted and stored at −80°C.

Viral titres were measured through qPCR against a virus of known biological titre (FACS titration). Genomic DNA was isolated from cells (described previously) which had been transduced with a serial dilution of virus. Viral genomic integration was measured using WPRE primers: CCCGTACGGCTTTCGTTTTC (fwd) and CAAACACAGAGCACACCACG (rvs).

SARS-CoV-2 production

The SARS-CoV-2 strain Victoria was produced by infecting Vero E6 cells at an MOI of 0.01 in DMEM supplemented with 2% FCS for 72 h, until cytopathic effects were visible. The supernatant containing viral particles was harvested, aliquoted, and stored at −80°C. The titre of the SARS-CoV-2 stock was determined by plaque assay using Vero E6 cells.

SARS-CoV-2 challenge

1 million Calu-3 cells were seeded in each well of a 12-well plate. Cells were left to adhere for 24 h. The cells were infected with the SARS-CoV-2 strain Victoria at MOI 1 in 2% FCS-medium, allowing a total volume of 1 ml per well. The cells were returned to the incubator and harvested at 17 h post-infection when 20% of cells showed cytopathogenic effects. To neutralise SARS-CoV-2, 1 ml of SARS-CoV-2 was incubated for 1 h before infection with 15 μl of the antibody clone EY11A with regular mixing. EY11A inhibits SARS-CoV-2 activity by binding strongly to its spike protein. At the end point, the supernatant was collected and preserved at −80°C for future determination of virus titre, and after two washes with PBS, RNA was extracted from cells using the RNeasy Mini Kit (QIAGEN) following manufacturer’s instructions.

For the RNA sequencing experiment, 2 million cells were seeded in six-well plates, and after 24 h, cells were infected with THE SARS-CoV-2 Victoria strain at MOI 1. At 17 h post infection, cells were washed twice with PBS and harvested for RNA extraction using the QIAGEN RNeasy Mini Kit. The quality of the RNA was assessed by NanoDrop ratios and TapeStation 4200 (Agilent). RNAs with a 260/230 ratio > 2 and RIN > 9.8 were submitted for library preparation and sequencing.

TCID50 titration of virus after challenge

96-well plates were seeded with 15,000 Vero E6 cells per well 24 h before adding virus in 120 μl of 2% FCS medium per well. 40 μl/well of preserved supernatant was added to the monolayer of Vero E6 cells in column 1 of the plate (rows A to H); this provides eight replicates to determine TCID50/ml. The virus was mixed by carefully pipetting six times before changing tips at each column and transferring 40 μl to the next column, performing a serial dilution (1:4) across the plate. The plates were incubated at 37°C for 72 h and stained with crystal violet. TCID50/ml was determined visually by recording cell death at each dilution and deriving the titre using the freely available online software (https://www.klinikum.uni-heidelberg.de/fileadmin/inst_hygiene/molekulare_virologie/Downloads/TCID50_calculator_v2_17-01-20_MB.xlsx).

Quantification of the SARS-CoV-2 RNA copy number

The SARS-CoV-2 RNA copy number was quantified in RNA samples extracted from infected cells using the QuantiTect probe RT-PCR Kit (QIAGEN). We prepared a standard curve using the ORF9 nucleoprotein coding sequence cloned in pcDNA3. 10 μg of plasmid was linearised with Apa1 at 37°C for 2 h. DNA was purified with the PCR cleanup kit from QIAGEN, and 1 μg of DNA was used in a T7 transcription reaction using Promega T7 RiboMAX (P1320) at 37°C for 30 min. The transcribed RNA was cleaned using Zymo RNA clean and concentrator 25 (R1017; Zymo Research). The copy number was derived using an online calculator (https://nebiocalculator.neb.com/ ssRNA mass to moles converter).

The RNA standard curve was prepared from diluting RNA template from 10 × 10⁸ to 10 × 10¹ in RNase-free water. RNA samples were diluted 1:5,000, and 1 μl of standard or sample was dispensed into each well of a 384-well plate to perform the RT-PCR step following manufacturer’s instructions. Quantitation of the copy number was performed using N1 and N2 standard probes for SARS-CoV-2 (IDTDNA).

EXOSC2 reconstitution

5 μg of pLenti-EXOSC2 with 2.5 μg pR8.91 and 2.5 μg of VSV-G plasmids were transfected in a 15-cm dish of Hek293T cells at 80% confluency. The transfection mix was prepared using Fugene 6 (Promega) at 3 μl/μg DNA in 500 μl (total volume) of Opti-MEM (Thermo Fisher Scientific). The transfection mix was incubated following manufacturer’s recommendations and added dropwise to the dish. After 24 h, the medium was changed to 15 ml 2% FCS medium. After a further 24-h incubation, the supernatant was collected, spun at 200g to remove cellular debris, and passed through 0.45-μM syringe filter (Sartorius). The supernatants were preserved at −80°C.

To transduce cells with EXOSC2-lentivirus, a T25 of EXOSC2 CRISPR-edited cells was incubated for 48 h with 2 ml of culture media supplement with 1 ml of viral supernatant and 10 μg of polybrene. The cells were then returned to normal propagation media.

Immunoblotting for EXOSC2

500,000 Calu-3 cells were seeded in 12-well plates and left to adhere for 24 h. Then the media was discarded, cells were subsequently washed with PBS, and 250 μl of RIPA buffer (10 mM Tris–HCl pH 8, 140 mM NaCl, 1% Triton-X100, 0.1% SDS, 0.1% sodium deoxycholate, 1 mM EDTA, 1 mM EGTA) containing protease inhibitor (04693132001; Roche) was added to the cells. Cells were scraped from wells, dispensed into an Eppendorf tube, and agitated for 20 min at 4°C, then spun at 30,000g for 20 min at 4°C. The supernatant was collected, and 100 μg of total protein (determined by qubit) was supplemented with Laemmli buffer, boiled for 5 min, and resolved on 4–20% TGX mini protean gels (Bio-Rad). The proteins were transferred to PVDF membranes using Bio-Rad Trans-Blot semi-dry transfer. The membrane was blocked for 1 h with constant rotation in 2.5% milk-PBS 0.1% Tween 20 and incubated overnight at 4°C with the EXOSC2 antibody (66099-1-Ig; ProteinTech Group) or GAPDH antibody (14C10; Cell Signalling Technology) diluted in 2.5% milk-PBS-0.1% Tween 20. After three washes of 15 min each in PBS-0.1% Tween 20, the membranes were incubated with secondary antibodies for 1 h, and after three further washes with PBST, the membrane was transferred to a fresh falcon tube and incubated with 1 ml of Western Lightning Plus ECL chemiluminescent reagent (PerkinElmer). Proteins bands were visualised with an iBright instrument (Thermo Fisher Scientific). The intensity of EXOSC2 band against the GAPDH loading control was assessed using ImageJ.

Immunoblotting for the total set of RNA exosome proteins

Calu-3 cells were harvested by centrifugation at 400g for 4 min at RT. Cell pellets were washed with ice-cold PBS and lysed in RIPA buffer (10 mM Tris–HCl, pH 8, 140 mM NaCl, 1% Triton-X100, 0.1% SDS, 0.1% sodium deoxycholate, 1 mM EDTA, 1 mM EGTA) containing protease inhibitor cocktail (Roche) for 15 min on ice. Lysates were centrifuged at 17,000g for 5 min at 4°C, and the supernatant was transferred to fresh Eppendorf tubes. 30 μg total protein lysate was mixed with Laemmli buffer, boiled at 95°C, and fractionated on 12% SDS polyacrylamide gels. Protein was electrophoretically transferred to nitrocellulose membranes using a Bio-Rad semi-dry transfer apparatus. Membranes were blocked in 5% milk + TBST (20 mM Tris, 137 mM NaCl, 0.2% [vol/vol] Tween 20, pH 7.6) for 1 h at RT, then incubated overnight at 4°C in the primary antibody (EXOSC1, 12585-1-AP; Proteintech; EXOSC2, 14805-1-AP; Proteintech; EXOSC3, 15062-1-AP; Proteintech; EXOSC4, 15937-1-AP; Proteintech; EXOSC5, 15627-1-AP; Proteintech; EXOSC7, 25292-1-AP; Proteintech; EXOSC8, 11979-1-AP; Proteintech; EXOSC9, 24470-1-AP; Proteintech; EXOSC10, 11178-1-AP; Proteintech; DIS3, 14689-1-AP; Proteintech; DIS3L, 25746-1-AP; Proteintech; DIS3L2, 25792-1-AP; Proteintech; β-actin, ab8227; Abcam) diluted in 5% milk + TBST. Membranes were washed 3× in TBST then incubated in an HRP-conjugated secondary antibody (diluted in 5% milk + TBST) for 1 h at RT, followed by a further three washes in TBST. Membranes were incubated in ECL Western blotting substrates (Bio-Rad) for 5 min at RT, and protein bands were visualised using an LI-COR Odyssey XF Imaging System. Quantification of band intensity was performed using FIJI (NIH) and normalised to β-actin.

qRT–PCR for EXOSC2

Calu-3 cells were lysed on ice using Tri Reagent (Sigma-Aldrich) for 5 min in RNase-free conditions. Total RNA was extracted using a Direct-zol RNA Miniprep Kit (Zymo) according to the manufacturer’s instructions. RNA concentration was determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific). 2 μg total RNA was converted to cDNA by adding 1 μl 10 mM dNTPs, 1 μl 40 μM random hexamer primer, and DNAse/RNAse–free water to a total reaction volume of 14 μl. The mixture was heated to 70°C followed by a 5-min incubation on ice. 4 μl 5× first strand buffer, 2 μl 0.1 M DTT, and 1 μl M-MLV reverse transcriptase (Thermo Fisher Scientific) were then added, and cDNA conversion was performed in a thermocycler (37°C for 50 min, 70°C for 10 min). cDNA was amplified using RT-PCR with Brilliant III SYBR Green (Agilent) as per the manufacturer’s instructions using primers: AACCTGGAGCCTGTCTCTCTT (fwd) and TGATCTGATGTGGAAGGGATGC (rvs). CT analysis was performed using CFX Maestro software (Bio-Rad).

RNA sequencing

RNA was extracted from Calu-3 cells in the presence and absence of EXOSC2 gene editing at 17 h post–SARS-CoV-2 infection at MOI 1. Extracted RNA was of high quality (RIN∼10); the input mass was 100 ng. Libraries were prepared according to manufacturer’s instructions; the NEBNext rRNA Depletion Kit v2 was applied to remove human ribosomal RNA before strand-specific libraries were constructed using the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (E7760). The libraries were indexed with custom adaptors and barcode tags (including dual indexing [Lamble et al, 2013]) and sequenced on an Illumina NovaSeq6000 v1.5 in 150-bp paired-end mode. A minimum of 80 million reads were obtained per sample.

Human transcriptome analysis

Raw Fastq files were trimmed for the presence of Illumina adaptor sequences using Cutadapt v1.2.1 (Martin, 2011). Reads were aligned to hg19 transcripts (n = 180,253) using Kallisto v0.46.0 (Bray et al, 2016) to produce gene-level TPM estimates by aggregating transcripts per gene. Differential expression analysis was performed using edgeR (Robinson et al, 2010). Read counts were first TMM-normalised (Robinson & Oshlack, 2010) to account for differences in library size. Differentially expressed genes were identified by a genewise exact test (Robinson & Smyth, 2008); only genes which were significant after Benjamini–Hochberg multiple testing correction were reported as differentially expressed.

Viral transcriptome analysis

The raw fastQ files were filtered for viral reads using ReadItAndKeep (Hunt et al, 2022), and the files of viral reads produced were then used for downstream analysis of Subgenomic RNA. Briefly, ReadItAndKeep removes host reads from viral sequencing data by aligning all reads against a target genome using minimap2 and retaining only reads that match at least 50 bp or 50% of the length of the read. This approach has been shown to have 100% sensitivity and 99.894% specificity in distinguishing human from viral reads with Illumina sequencing. ReadItAndKeep version 0.1.0 was downloaded from Bioconda and run using the Wuhu-Hu-1 genome (NCBI Reference Sequence: NC_045512.2, download from www.ncbi.nlm.nih.gov/nuccore/NC_045512) as the target sequence the paired Illumina reads in fastq format. The percentage of viral reads was calculated as the number of reads retained by ReadItAndKeep divided by the total number of reads.

To calculate the proportion of genomic:subgenomic reads, we measured read-depth over SARS-CoV-2 Orf1 which is present in all genomic RNAs and none of the canonical subgenomic RNAs to the read-depth over Orf10 which is present in all canonical subgenomic RNAs (Long, 2021). A limitation of our method is that we do not consider low-frequency non-canonical subgenomic RNAs.