SARS-CoV-2-Mediated Lung Edema and Replication Are Diminished by Cystic Fibrosis Transmembrane Conductance Regulator Modulators

ABSTRACT Coronaviruses (CoVs) of genera α, β, γ, and δ encode proteins that have a PDZ-binding motif (PBM) consisting of the last four residues of the envelope (E) protein (PBM core). PBMs may bind over 400 cellular proteins containing PDZ domains (an acronym formed by the combination of the first letter of the names of the three first proteins where this domain was identified), making them relevant for the control of cell function. Three highly pathogenic human CoVs have been identified to date: severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East respiratory syndrome coronavirus (MERS-CoV), and SARS-CoV-2. The PBMs of the three CoVs were virulence factors. SARS-CoV mutants in which the E protein PBM core was replaced by the E protein PBM core from virulent or attenuated CoVs were constructed. These mutants showed a gradient of virulence, depending on whether the alternative PBM core introduced was derived from a virulent or an attenuated CoV. Gene expression patterns in the lungs of mice infected with SARS-CoVs encoding each of the different PBMs were analyzed by RNA sequencing of infected lung tissues. E protein PBM of SARS-CoV and SARS-CoV-2 dysregulated gene expression related to ion transport and cell homeostasis. Decreased expression of cystic fibrosis transmembrane conductance regulator (CFTR) mRNA, essential for alveolar edema resolution, was shown. Reduced CFTR mRNA levels were associated with edema accumulation in the alveoli of mice infected with SARS-CoV and SARS-CoV-2. Compounds that increased CFTR expression and activity, significantly reduced SARS-CoV-2 growth in cultured cells and protected against mouse infection, suggesting that E protein virulence is mediated by a decreased CFTR expression.

found that the PBMs of these three hCoVs are virulence factors required for optimal replication and the generation of alveolar edema. Furthermore, we constructed a collection of SARS-CoV mutants in which the E protein PBM was replaced with one derived from virulent or attenuated hCoVs and studied their virulence. A gradient of virulence was observed, depending on whether the E protein was derived from an attenuated or virulent hCoV. Following infection of mice with SARS-CoV, we performed transcript profiling in lung tissues using RNA sequencing (RNA-seq). These studies showed that the E protein PBMs of SARS-CoV or SARS-CoV-2 altered the expression of genes involved in ion transport and cell homeostasis. Of note, we observed decreased expression of cystic fibrosis transmembrane conductance regulator (CFTR) mRNA, encoding a protein involved in liquid absorption in the alveolus. The reduction in CFTR transcript abundance was associated with increased alveolar edema in the lungs of mice infected with SARS-CoV or SARS-CoV-2. We further investigated the effects of small molecule CFTR modulators on the replication of SARS-CoV-2 in cultured cells and observed that these compounds significantly reduced SARS-CoV-2 replication and protected against severe disease in a mouse model. These results demonstrate the importance of the E protein PBMs in CoV replication and virulence and identify novel cellular targets for the selection of antivirals.

RESULTS
Relevance of E protein PBM from virulent hCoVs on viral replication in vitro and in vivo. Using an infectious cDNA clone engineered in our laboratory, we constructed a collection of mutant viruses by replacing the 4 residues of the PBM core with glycines (Fig. 1A). The growth kinetics of mouse-adapted SARS-CoV and SARS-CoV-2 wild-type and deletion mutants were assessed in Vero E6 cells. Similar studies of modified MERS-CoV were performed in Huh-7 cells. While the growth of rSARS-CoV with or without E PBM was similar, the E PBM deletion mutants of SARS-CoV-2 and MERS-CoV exhibited between 10 and 1,000-fold lower viral titers, respectively, than the parental (wt) viruses (Fig. 1B). This suggests that SARS-CoV-2 and MERS-CoV required an E PBM for optimal virus replication in cultured cells.
To investigate the role of E protein PBM in virus replication in vivo, BALB/c, K18-hDPP4, or K18-hACE2 mice were infected with either wild-type (wt) SARS-CoV, MERS-CoV, or SARS-CoV-2, respectively, or their corresponding mutants lacking E PBMs. Lung tissue titers were determined at days 2 and 4 postinfection (dpi) (Fig. 1C). The MERS-CoV and SARS-CoV-2 E protein deletion mutants grew to 1,000-to 10,000-fold lower titers that the corresponding parental viruses, indicating that E protein PBMs contributed to efficient virus replication in vitro and in vivo. We then monitored weight loss and survival through 10 dpi. All mice infected with parental viruses lost weight and exhibited mortality ranging from 80% to 100%, as expected (Fig. 1D). In contrast, all mice infected with virus lacking E protein PBM expression survived, further supporting the conclusion that the E protein PBM is a CoV virulence factor.
Human CoVs cause mild to severe respiratory disease. To assess whether disease severity is influenced by the source of the PBM, we engineered a set of recombinant SARS-CoVs expressing the E protein PBM sequences of virulent or attenuated b-hCoV, including SARS-CoV-2, MERS-CoV, HCoV-OC43, and HCoV-HKU1 ( Fig. 2A). As a control, a rSARS-CoV without a PBM core sequence was generated (rSARS-PBM 2 ). We inoculated BALB/c mice intranasally with the recombinant viruses and monitored weight loss (Fig. 2B) and survival (Fig. 2C) for 10 dpi. While all mock-infected mice and mice infected with SARS-CoV lacking E protein PBM survived, mice infected with recombinant viruses containing the parental virus PBM from SARS-CoV-2 (PBM-SARS2) or MERS-CoV (PBM-MERS) lost weight and died between 4 and 6 dpi, respectively. Interestingly, we observed a gradient of lethality among these highly virulent viruses. All mice infected with PBM-SARS2 died with less than 20% weight loss. In contrast, all mice infected with PBM-MERS exhibited more than 25% weight loss and required euthanasia. Mice infected with viruses containing the E protein PBM core sequences from the attenuated hCoVs HKU1 and OC43 demonstrated intermediate survival phenotypes   S1 in the supplemental material). E protein PBM conditions b-hCoV pathogenesis. At 2 dpi, SARS-CoV-2-infected K18-hACE2 mice displayed mild alveolar septal thickening and mild perivascular and peribronchiolar mononuclear infiltrates. These lesions increased by 4 dpi, and mice also developed alveolar edema ( Fig. 3A and B) with a histopathology score of 14.0. However, mice infected with the mutant SARS-CoV-2-PBM 2 exhibited no significant inflammatory or vascular changes having a score of 4.7. K18-hDPP4 mice infected with MERS-CoV showed moderate lung inflammatory lesions at 2 dpi that increased in severity by day 4 pi with a score of 11.4, including the presence of diffuse alveolar septal thickening, and moderate perivascular and peribronchiolar mononuclear infiltrates. In FIG 1 Legend (Continued) lysis plaque formation method. (C) Groups of six 16-week-old BALB/c, K18-hDPP4, or K18-hACE2 mice were intranasally inoculated with 10,000 PFU of the WT viruses (red columns) or each of the mutants lacking E protein PBM (blue columns). Three mice from each group were euthanized at 2 and 4 dpi to analyze virus production in lung. Vertical bars represent the standard error of the mean. Statistically significant data are represented according to the P value obtained in Student's t test analysis: *, P , 0.05; ***, P , 0.001. (D) Groups of five 16-week-old BALB/c, K18-hDPP4, or K18-hACE2 mice were mock-infected (gray lines) or intranasally inoculated with 10,000 PFU with each of the parental (WT) virus (red lines), or with their corresponding mutants without PBM of E protein (PBM 2 ) (blue lines). Weight loss (top) and survival (bottom) of mice were monitored for 10 days postinfection. Vertical bars represent the standard error of the mouse weight mean. contrast, mice infected with the mutant MERS-CoV-PBM 2 showed a score of around 6.0 with no significant inflammatory lesions at either time point ( Fig. 3C and D). In BALB/c mice infected with SARS-CoV, lung tissue pathology was more severe than in BALB/c mice infected with the SARS-CoV-PBM 2 mutant with histopathology scores of 17.7 and 13.5 at 2 dpi, respectively. These differences included severe interstitial pneumonia, perivascular and peribronchiolar mononuclear infiltrates, and alveolar edema ( Fig. 3E and F). We performed additional histologic analysis of lung tissue sections from mice infected with rSARS-CoVs with PBMs from attenuated or virulent CoVs at 2 or 4 dpi to assess immune cell infiltration and alveolar edema. Mice infected with virus lacking the E protein PBM core sequence (PBM 2 ) showed minimal inflammatory cell infiltrates and edema only at 4 dpi (Fig. 4). Mice infected with rSARS-CoV with an E protein PBM from HCoV-HKU1 (PBM-HKU1) or HCoV-OC43 (PBM-OC43) developed significant cellular infiltrates at 2 dpi that persisted at 4 dpi, while mice infected with virulent SARS-CoVs showed progressive infiltrates at the same time points (Fig. 4). Mice infected with viruses encoding the E protein PBM core from SARS-CoV-2 showed moderate interstitial and peribronchiolar cell infiltration at 2 dpi and more alveolar edema at 4 dpi coincident with the greater lethality of this virus. Interestingly, mice infected with recombinant viruses encoding an E protein PBM core from MERS-CoV showed a phenotype similar to that observed for PBM-SARS2 at 2 dpi but with a significant increase in cell infiltration and alveolar and bronchiolar wall thickening at 4 dpi, compared to mice infected with viruses encoding E PBMs from attenuated CoVs (Fig. 4). These data indicated that virus virulence correlated with lung tissue pathology and the E protein PBM core sequence contributed to these findings.
Effect of E protein PBM virus origin on host gene expression. To investigate the virulence mechanisms of rSARS-CoV with PBMs from the different hCoVs, we performed RNA-seq analysis on lung tissue from infected mice. Because the observed association of virulence and lung histopathology scores was greater at 4 dpi, we focused on this time point. A Venn diagram of differentially expressed genes (DEGs) with a fold change (FC) higher than 1.5 and a false discovery rate (FDR) lower than 0.05 (jFCj>1.5; FDR ,0.05, respectively) of mice infected with rSARS-CoVs with PBMs from hCoVs causing a range of pathology compared to the PBM 2 mutant was determined (Fig. 5A). We sought to identify transcripts with expression driven by the presence of an E protein PBM from attenuated or virulent b-hCoVs. The presence of an E protein PBM from attenuated b-hCoVs reduced SARS-CoV mortality ( Fig. 2B and C). A set of 14 transcripts that increased their abundance after infection with all four viruses, including PBM-HKU1, PBM-OC43, PBM-MERS, or PBM-SARS2, was upregulated in all cases, whereas 1 (STK33) was downregulated in comparison with the infection by a SARS-CoV including a nonfunctional PBM that was replaced by four glycines (Fig. 5B). These transcripts were classified using the Gene Ontology analysis (DAVID software) as mainly related to immune and inflammatory responses (Fig. 5B), suggesting that the presence of a functional E protein PBM from either an attenuated or virulent b-hCoV triggered a host innate immune response.
Replacing the SARS-CoV E protein PBM core with the homologous sequence of a virulent hCoV (PBM-MERS or PBM-SARS2) increased lung injury and virus pathogenicity compared to replacement by the PBM core from an attenuated hCoV (PBM-HKU1 and PBM-OC43). We next focused on the 70 and 392 DEGs that were solely changed in expression by viruses encoding the E protein PBM from MERS-CoV or SARS-CoV-2 at 4 dpi, respectively (Fig. 5A). The 70 DEGs enriched in tissues from mice infected with CoVs encoding a PBM-MERS were associated with host immune and inflammatory responses (Fig. 5C).
Among the 392 DEGs enriched during infection with viruses containing the E PBM from SARS-CoV or SARS-CoV-2 ( Fig. 6A), 52 were present within biological activity processes Role of Coronavirus E Protein PBM in CFTR Expression and Pathogenesis mBio related to ion transport and homeostasis ( Fig. 6A). At least 34 of these 52 transcripts were identified in a highly interactive network (Fig. 6B). The genes with increased expression were mainly associated with the regulation of cytosolic Ca 21 concentration while those with decreased expression were associated with ion transport and cell homeostasis (Fig. 6C). These transcriptional changes were associated with greater pulmonary edema scores and mortality. Among these genes we noted that CFTR transcript abundance was reduced only in the presence of E protein PBM from SARS-CoV and SARS-CoV-2. The reduction in CFTR mRNA expression in the lungs of mice infected with SARS-CoV, containing a PBM identical to that of SARS-CoV-2, was confirmed by quantitative PCR at 4 dpi (Fig. 6D). SARS-CoV and SARS-CoV-2 infection reduced CFTR mRNA transcript abundance in Calu-3 2b4 cells and lung tissue of infected mice ( Fig. 7A and B). CFTR modulators modify virus replication in infected cells. In the alveolus, CFTR plays an important role in fluid absorption and thus loss of CFTR function could contribute to alveolar edema (20)(21)(22)(23). Consistent with this, mice infected with SARS-CoV-2 or SARS-CoV had more alveolar edema compared to the PBM-variants ( Fig. 7C and D). Recently, small molecule CFTR modulators were discovered that improve the function of some mutant forms of CFTR protein associated with cystic fibrosis. The corrector compound VX809 (lumacaftor) acts as a chemical chaperone and partially rescues CFTR function in cells expressing the misfolded F508del CFTR protein (24,25). The Groups of six 16-week-old mice were mock infected or inoculated with 100,000 PFU of a SARS-CoV including E proteins with a PBM derived either from the SARS/SARS-2-CoV, or from MERS-CoV, hCoV-OC43, hCoV-HKU1, or with the PBM replaced by four glycines (PBM 2 ). Three mice from each group were euthanized, and lungs were collected at 2 and 4 dpi. Lungs were fixed in formalin with zinc, embedded in kerosene, and cut and stained with hematoxylin and eosin. Cellular infiltration (A) and edema (B) scores were determined in lung samples as indicated in Materials and Methods (n = 3/group). The vertical bars represent the mean and the standard deviation of the data obtained in the lung of each of the three mice used for each experimental setting. Unpaired t test: *, P , 0.05; **, P , 0.01; ***, P , 0.001; ****, P , 0.0001. potentiator VX770 (ivacaftor) increases the probability of the CFTR anion channel associated with G551D and related conductance mutations being in an open conformation (26)(27)(28)(29)(30). Of note, these CFTR modulators also enhance the function of wt CFTR (31)(32)(33). We hypothesized that CFTR modulator treatment would partially rescue CFTR function and modify the disease phenotype. To test this hypothesis, CFTR modulators VX809 and VX770 were administered to Calu-3-2B4 cells and to K18-hACE2 mice infected with SARS-CoV-2. SARS-CoV-2 titers were measured at 48 h postinfection in cell culture, the time point at which the highest titer is reached (Fig. S2). CFTR modulator treatment resulted in a 10-and 7-fold reduction in virus titers in vitro and in vivo, respectively, in relation to nontreated cells or mice ( Fig. 8A and B). At these concentrations, the modulators were nontoxic to Calu-3-2B4 cells (Fig. S3). Remarkably, the combined treatment of mice infected with SARS-CoV-2 with the VX809 and VX770 was associated with 80% survival, while mock-treated animals had 100% mortality. Still, a minimal effect on weight loss was observed, which is in line with previous observations that weight loss is a more sensitive measure of virus pathogenesis than survival (reference 15). Mice receiving VX770 or VX809 alone exhibited 40% and 20% survival, respectively ( Fig. 8C and D). Treatment with the combination of both CFTR modulators delayed the appearance of clinical signs due to SARS-CoV-2 infection. SARS-CoV-2-infected and nontreated mice showed the highest clinical score (3.6) at both 7 and 8 dpi. In contrast, infected mice treated with both CFTR modulators showed a clinical score of 2.6 and 0.5 at the same times, respectively (Fig. 8E). Moreover, infected mice treated with CFTR modulators individually showed less edema score (4.9 for both cases) at 6 dpi in comparison with untreated mice (7.9). The combination of both CFTR modulators also reduced the edema score (4.0) (Fig. 8F).

DISCUSSION
Relevance of E protein PBM core motifs as virulence determinants. In this article, the role of the E protein PBMs in human CoV replication and virulence is investigated, leading to the discovery that E protein PBM from SARS-CoV and SARS-CoV-2 reduces the transcript abundance for several genes involved in ion transport and cellular homeostasis (Fig. 6A), including CFTR ( Fig. 6B and C). It is also shown that the combination of two drugs previously used to treat cystic fibrosis increased by 80% the survival of humanized mice after their infection by SARS-CoV-2.
Using reverse genetics, we generated a collection of SARS-CoV mutants by replacing its E protein PBM with homologous PBMs from attenuated or virulent b-hCoVs (SARS-CoV-2, MERS-CoV, hCoV-OC43, and hCoV-HKU1). The engineered mutants showed significant differences in virulence in mice. The PBMs from common cold hCoVs partially attenuated SARS-CoV and showed less cellular infiltrates and alveolar edema, compared to mice infected with the parental virus (Fig. 4). In contrast, rSARS-CoVs with PBM cores from virulent hCoVs caused the death of all infected mice and were associated with diffuse cellular infiltrates and alveolar edema. While only a small number of viruses were analyzed, these results support the notion that the PBMs play a critical role in virus virulence. Analysis of additional CoVs should be performed to confirm (or not) this interesting association. SARS-CoV and SARS-CoV-2 share identical E protein PBMs. Infection of mice with these two viruses caused the greatest alveolar edema and mortality. In contrast, the introduction   (42,43). Rabies virus (RABV) strains with different G protein PBMs showed a spectrum of virulence regulated by the activation of different signaling pathways (44,45).
Effects of E protein PBMs on host gene expression. We identified host mRNA transcripts with expression significantly altered by infection with recombinant SARS-CoVs in which only the E protein PBM was modified. The introduction of the MERS-CoV PBM into SARS-CoV maintained expression of inflammatory response-related genes, particularly late in the infection, unlike the PBMs from other virulent viruses (Fig. 5C). For example, high levels of SLFN8, CCL3, CCR3, and CCRL2 were observed in addition to factors related to T-cell proliferation, leukocyte recruitment, eosinophil accumulation, and recruitment of effector immune cells (46)(47)(48)(49). Increases in these factors could contribute to the cellular infiltrates observed in the lungs of mice infected with SARS-CoV mutants encoding the MERS-CoV PBM core domain.
We have shown that E protein PBM from SARS-CoV and SARS-CoV-2 reduces the transcript abundance for several genes involved in ion transport and cellular homeostasis (Fig. 6A), including CFTR ( Fig. 6B and C).
Influence of CFTR on the lung pathology produced by SARS-CoV and SARS-CoV-2. CFTR dysfunction in alveolar epithelial cells impairs alveolar fluid absorption by reducing Cl-permeability across the alveolar epithelium (23). CFTR is required to establish the osmotic gradient for fluid absorption from the alveolar space via aquaporin channels and paracellular pathways (20). Interactions between the SARS-CoV and SARS-CoV-2 E protein PBMs and CFTR may increase alveolar capillary and endothelium permeability, favoring edema formation. CFTR may function in an anti-inflammatory capacity (50)(51)(52)(53)(54)(55). The membrane protein (M2) of influenza A has intrinsic ion channel activity and also inhibits CFTR during infection (56). Therefore, it is possible that CFTR inhibition could be a mechanism to induce respiratory virus pathogenesis. In fact, previous studies identified an important role for CFTR in ion and fluid transport across the alveolus. Adenovirus-mediated CFTR expression in rat lungs increased fluid absorption (57), whereas CFTR-deficient animals exhibit impaired alveolar fluid absorption in epithelial cells (58). We found that the CFTR modulators ivacaftor and lumacaftor reduced SARS-CoV-2 replication. Furthermore, infection with native SARS-CoV-2 virus reduced CFTR mRNA expression, while a deletion mutant without an E protein PBM did not ( Fig. 7A and B) (59). Similar observations were reported using in vitro and in vivo models of influenza A virus infection (60). Influenza A virus infection reduced CFTR protein abundance and ion channel activity, and these changes were restored by CFTR modulators (60). Ivacaftor also improved alveolar liquid clearance in isolated pig lung lobes ex vivo and reduced edema in a volume overload in vivo pig model of hydrostatic pulmonary edema (61).
Loss of CFTR function also alters Ca 21 homeostasis and contributes to increased inflammatory responses in CF (62). Viruses can also disrupt Ca 21 homeostasis by forming membrane structures used for viral particle formation (63). Ivacaftor and lumacaftor normalize mitochondrial and intracellular Ca 21 levels and reduce oxidative stress and proinflammatory cytokines levels in CF lungs (64)(65)(66). The E protein PBM of SARS-CoV-2 may affect virus replication by altering Ca 21 homeostasis, leading to the reduction in

Role of Coronavirus E Protein PBM in CFTR Expression and Pathogenesis mBio
SARS-CoV-2 titers observed in cells infected and treated with lumacaftor or ivacaftor (Fig. 8) (67)(68)(69). In fact, genes responsible for Ca 21 homeostasis, such as ATP2A3 and ATP2B2, also are underregulated due to the presence of E protein PBM from SARS-CoV-2 (Fig. 6C). Complementary studies have demonstrated that the use of Ca 21 channel blockers may reduce the mortality from SARS-CoV-2 in humans (70,71). In summary, loss of CFTR function in the setting of severe CoV infection may contribute to lung injury through multiple mechanisms including impaired fluid and electrolyte transport, altered Ca 21 homeostasis, and NLRP3 inflammasome activation. Ivacaftor and lumacaftor treatment partially reversed this pathology and improved survival in a mouse model of severe disease. FDA-approved CFTR modulators, now widely used for patients with cystic fibrosis, may also help to reduce morbidity and mortality due to high-pathogenicity human coronaviruses. Viruses. Mouse-adapted SARS-CoV (MA15) (72) parental wild-type (wt) and recombinant viruses were rescued from infectious cDNA clones generated in a bacterial artificial chromosome (BAC). rSARS-CoV-PBM 2 was generated in our lab previously (15).

MATERIALS AND METHODS
Generation of recombinant virus infectious clones. The cDNA of SARS-CoV-MA15 assembled in a pBAC (pBAC-SARS-CoV-MA15) (73) was used to introduce the corresponding mutations affecting the E protein of SARS-CoV. To generate viruses with mutations in the SARS-CoV E protein carboxy terminus, DNA fragments including nucleotides 26017 to 26884 of the SARS-CoV-MA15 genome flanked by BamHI and XcmI restriction sites were generated using the oligonucleotides indicated in Table 1. These fragments carried different mutations located on the E gene generating amino acid changes at the E protein carboxy terminus: PBM-MERS, PBM-HKU1, and PBM-OC43. The fragments were digested with enzymes BamHI and XcmI and introduced into an intermediate plasmid pBAC-BamHI-RsrII-SARS-CoV containing nucleotides 26044 to 29783 of the SARS-CoV infectious clone. Next, these plasmids were digested with enzymes BamHI and RsrII, and the fragments containing the different mutations were inserted into plasmid pBAC-SARS-CoV-MA15 digested with the same enzymes, obtaining the respective infectious clones. The integrity of cDNA was verified by restriction analysis with the enzyme HindIII and by Sanger sequencing.
SARS-CoV-2 cDNA was assembled in a pBAC (pBAC-SARS-CoV-2) following previously described procedures (13,74,75). This infectious cDNA clone was used to introduce the corresponding mutations affecting the PBM of the SARS-CoV-2 E protein. First, a DNA fragment was obtained by chemical synthesis (GeneScript) that included nucleotides 26386 to 26751 of the SARS-CoV-2 genome, flanked by HpaI and AgeI restriction sites. This fragment carried the carboxy terminus mutation eliminating the PBM (DLLV ! GGGG). The fragment digested with HpaI and AgeI was introduced into the corresponding sites of plasmid psL-F6-SARS-CoV-2, which contains nucleotides 25314 to 28610 of the SARS-CoV-2 genome, generating plasmid psL-F6-SARS-CoV-2-E-PBM 2 . Plasmid psL-F6-SARS-CoV-2-E-PBM 2 was then digested with restriction enzymes BamHI and AvrII and the resulting fragment was cloned into an intermediate plasmid pBAC-F6-SARS-CoV-2. Subsequently, this plasmid with was digested with BamHI and RsrII and the fragment containing the E gene mutation was inserted into the unique BamHI and RsrII cloning sites of the pBAC-SARS-CoV-2 plasmid to generate the corresponding cDNA infectious clone. The integrity of the cloned DNA was verified by restriction analysis with the enzyme EcoRI and by Sanger sequencing.
The cDNA infectious clone of MERS-CoV strain EMC/2012, assembled in pBAC (pBAC-SA-FL) (13), was used to generate the mutants with a modified MERS-CoV E protein PBM. First, DNA fragments including nucleotides 27502 to 28319 of the MERS-CoV genome flanked by KflI and Pfl23II restriction sites were generated by PCR using the oligonucleotides indicated in Table 1 Growth kinetics. Subconfluent monolayers (around 90% confluence) of Vero E6 and Huh-7 cells were infected at an MOI of 0.001 with parental viruses or the respective mutant viruses without E protein PBM. Culture supernatants were collected at 24, 48, and 72 hpi, and virus titers were determined as previously described (12).
Mice. To determine mutant virus pathogenesis, SARS-CoV, MERS-CoV, and SARS-CoV-2 viruses were inoculated into the corresponding mice strains: 16-week-old BALB/c OlaHsd (Harlan), K18-hDPP4 (76), and K18-hACE2 (77) (Jackson Laboratories), respectively. Mice were anesthetized with isoflurane and inoculated intranasally using 100,000 or 10,000 PFU of the corresponding virus in a 50 mL volume of DMEM supplemented with 2% FBS. Manipulation of infected mice was carried out in a level 31 biological containment laboratory at CISA-CSIC (Algete, Madrid) equipped with the required containment infrastructure for animal and cell cultures work. Scores for clinical signs of disease were determined based on the animal physical appearance and behavior. They were classified in five grades of severity (0, normal strength; 1, less active; 2, dull or sluggish movements, ruffled fur, squinting eyes; 3, ataxia, tremors; and 4, motionless, lying down, or hunched posture). Virus titrations were performed as previously described (12). Viral titers were expressed as PFU counts per gram of tissue.
Lung histopathology. The entire left lung lobe was removed from each mouse and immersion-fixed in zinc formalin (Sigma-Aldrich) for 48 h. After the fixation period, samples were routinely processed and embedded in paraffin blocks that were then sectioned at a 4-mm thickness on a microtome, mounted onto glass slides, and routinely stained with hematoxylin and eosin (H&E). Lung sections were microscopically evaluated using an Olympus BX43 microscope by a single veterinary pathologist who was blinded to the identity of mice (Veterinary Pathology Department, Animal Health Research Center-CISA-CSIC, Valdeolmos, Spain). To assess the presence and severe histopathological lesions, lung inflammation parameters based on previous reports on SARS-CoV-2 infection in mouse models were used (78). The histopathological parameters evaluated were as follows: alveolar hemorrhage, alveolar edema, perivascular edema, alveolar septal thickening (interstitial pneumonia), inflammatory cell infiltration in alveoli, bronchi/bronchioles with epithelial necrosis, detached epithelium or inflammatory cells in the lumen (bronchitis/bronchiolitis), peribronchial/peribronchiolar and perivascular mononuclear infiltrates, and cytopathic effect in pneumocytes or syncytium formation. The histopathological parameters were  Role of Coronavirus E Protein PBM in CFTR Expression and Pathogenesis mBio graded following a semiquantitative scoring system as follows: 0, no lesion;1, minimal lesion; 2, mild lesion; 3, moderate lesion; and 4, severe lesion. The cumulative scores of histopathological lesions provided the total score per animal. In each experimental group, the individual scores were used to calculate the group average. Extraction and analysis of RNA. Cell culture or lung RNA samples were collected and incubated in RNAlater reagent (Ambion). To extract total RNA, lungs were homogenized in 2 mL of RLT lysis buffer (Qiagen) supplemented with 1% b-mercaptoethanol using the GentleMACS Dissociator system (Miltenyi Biotec) following the protocol proposed by the manufacturer. RNA was purified from supernatants using the rNeasy minikit reagent (Qiagen), as previously described. To generate cDNAs from the purified RNAs, a reverse transcription (RT) reaction was carried out using the High Capacity cDNA RT kit reagent (Applied Biosystems). To verify the sequence of the mutations introduced into the viral genome, 4 mL of the cDNA generated in the RT reaction was used as a template in a PCR using the enzyme Taq polymerase (Invitrogen). Cellular mRNA expression was analyzed by quantitative reverse transcription-quantitative PCR (RT-qPCR), using TaqMan technology with commercial probes (ThermoFisher Scientific) ( Table 2). In all cases, the reaction was performed with the FastStart Universal Probe Master reagent (Rox) (Roche). qPCRs were performed in a 7500 Real Time PCR System (ThermoFisher Scientific). qPCR data were analyzed using the 7500 software v2.0.6 (ThermoFisher Scientific). All experiments met the recommendations for analysis of gene expression by RT-qPCR (MIQE) (79). Each result is the average of three independent experiments in which each sample was tested in triplicate. The relative quantification of gene expression was performed from the mean values of CT (cycle in which the amplification curve cuts at the threshold level using the 2 2DDCT method) (80). The level of 18S rRNA was used as an endogenous control to normalize cellular mRNA levels in mouse lungs, while hydroxymethylbilane synthase (HMBS) was used to normalize cellular mRNA levels in cell cultures.
Deep sequencing of mRNAs. Total RNA extracted from the lungs of infected mice was used for rNaseq sequencing. The samples were digested with the DNAsaI (Roche) to eliminate possible DNA contamination. RNA was processed to remove rRNA, a major component in large RNA samples, and enrich the resulting sequences for less abundant expressed RNAs such as mRNAs. RNA libraries were constructed (Illumina method) and 300 nt paired-end reads were obtained in which the complementary strand was marked to distinguish the original sense of the sequences. Three replicates per variable and per day were sequenced. From each sample, 80 million sequences were obtained.
Quantification of mouse genes and differential expression determination. All mouse genes (ENSEMBL release 101) were quantified using featureCounts function of bioconductor package Rsubread (84) (ignoreDup=TRUE, requiredBothEndsMapped=TRUE, primaryOnly=TRUE). To calculate differential expression between pairs of replicated samples, DESeq and lfcShrink functions of bioconductor package DESeq2 (85) were used (cooksCutoff=FALSE, independentFiltering=TRUE, type="normal"). Differential expression results, including adjusted P values by FDR (86), were stored as text-tabulated files and converted to FIESTA format (https://bioinfogp.cnb.csic.es/tools/FIESTA; 2021) to facilitate the visualization of overexpressed and underexpressed genes in form of MA plots. For each comparison, genes with jlogRatioj $ 1.5 and FDR # 0.05 were considered differentially expressed.
Functional analysis of relevant genes. The DAVID tool (87) was used to determine overrepresentation of Gene Ontology terms (Biological Process) in differentially expressed genes (both overexpressed and underexpressed). The six most representative biological processes were selected based on the P value.
Treatment of mice with CFTR modulators. Mice (n = 11) were treated by intraperitoneal injection administration with an appropriate concentration of compounds 24 h before infection with 10,000 PFU of SARS-CoV-2. Mock-infected or infected animals were treated once daily with 12 or 6 mg/kg of Lumacaftor or Ivacaftor, respectively. All treatments ceased after day 8 postinfection. Three mice from each group were euthanized at 3 and 6 dpi to analyze virus production in the lungs. Cell viability assay. To determine the cytotoxicity of commercial compounds that modulate CFTR protein activity, cells were incubated with different concentrations (1 to 50 mM) of ivacaftor (VX-770, Merck) or lumacaftor (VX-809; Selleck Chemicals) and cell viability was determined by the MTT [3-(4,5dimethylthiazol-2-yl)-2,5-diphenyltetrazol bromide] reduction assay (88).
Statistical analysis. Two-tailed, unpaired Student's t tests were used to analyze the differences in mean values between groups. The statistical significances are indicated in the figures as follows: *, P , 0.05; **, P , 0.01; ***, P , 0.001.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.