IFITM3 Inhibits SARS-CoV-2 Infection and Is Associated with COVID-19 Susceptibility

SARS-CoV-2 has become a global threat to public health. Infected individuals can be asymptomatic or develop mild to severe symptoms, including pneumonia, respiratory distress, and death. This wide spectrum of clinical presentations of SARS-CoV-2 infection is believed in part due to the polymorphisms of key genetic factors in the population. In this study, we report that the interferon-induced antiviral factor IFITM3 inhibits SARS-CoV-2 infection by preventing SARS-CoV-2 spike-protein-mediated virus entry and cell-to-cell fusion. Analysis of a Chinese COVID-19 patient cohort demonstrates that the rs12252 CC genotype of IFITM3 is associated with SARS-CoV-2 infection risk in the studied cohort. These data suggest that individuals carrying the rs12252 C allele in the IFITM3 gene may be vulnerable to SARS-CoV-2 infection and thus may benefit from early medical intervention.

The severity of COVID-19 varies widely between infected individuals. Many are asymptomatic, some show mild symptoms, and a significant number of severe cases develop pneumonia, respiratory distress, and even death [3]. It is believed the polymorphisms of key genes in part underpin this broad range of COVID-19 symptoms, and some of these genes can be antiviral effectors that are induced by interferon (IFN), given the high sensitivity of SARS-CoV-2 to IFN inhibition [4]. Indeed, the study of the association between disease severity and single-nucleotide polymorphisms (SNPs) in IFN-stimulated genes (ISGs) has shed light on the critical roles of certain ISGs in host susceptibility to virus infections [5,6].

Subjects and Samples
Samples of 203 patients (with a median age of 51, 105 males, and 98 females) diagnosed with quantitative RT-PCR for SARS-CoV-2 infection were investigated in this study. All patients required hospitalization in JinYinTan Hospital or FangCang Hospital, Wuhan. Throat swabs and plasma samples of patients were collected for additional viral testing, cytokine and chemokine quantification, and DNA extraction. This study was approved by the Ethical Review Board of the Institute of Pathogen Biology, Chinese Academy of Medical Sciences and Peking Union Medical College. Consent was obtained from competent patients directly or from relatives/friends/welfare attorneys of incapacitated patients.

Definition of Mild and Severe Cases
According to the new coronavirus disease-19 (COVID-19)-related pneumonia diagnosis and treatment guideline issued by the National Health Commission, patients with mild clinical symptoms and negative lung imaging present were defined as mild cases. Patients were defined as severe cases who had pneumonia confirmed by chest imaging and had an oxygen saturation (SaO 2 ) of 94% or less while they were breathing ambient air or a ratio of the partial pressure of oxygen (PaO 2 ) to the fraction of inspired oxygen (FiO 2 ) (PaO 2 :FiO 2 ) at or below 300 mm Hg.

Sequencing and Genotyping of rs12252
The region encompassing the human IFITM3 rs12252 sequences was amplified by PCR from DNA obtained from throat swabs. The amplification was performed using the following forward and reverse primers, 5 -GGAAACTGTTGAGAAACCGAA-3 and 5 -CATACGCACCTTCACGGAGT-3 (Beijing Institute of Genomics, Shenzhen, China). The products encompassing IFITM3 rs12252 were purified and Sanger sequenced on an Applied Biosystems 3730 × 1 DNA Analyzer (GATC Biotech, Konstanz, Germany). SNP genotypes were identified by MEGA-X. Homozygotes were called based on high, singlebase peaks with high Phred quality scores, while heterozygotes were identified based on low, overlapping peaks of two bases with lower Phred quality scores relative to surrounding base calls. SNP genotype frequencies in our sequencing were compared with that of the Chinese Han population from the 1000 Genomes Project database.

Cytokines and Chemokines Quantification
Plasma cytokines and chemokines were measured using the Human Cytokine Standard 27-Plex Assays panel and the Bio-Plex 200 system (Bio-Rad, CA, USA) according to the manufacturer's instructions. Plasma samples from four healthy adults were used as controls for cross-comparison. These four healthy adults were tested negative for SARS-CoV-2 by RT-PCR and did not have respiratory symptoms or other health-related complaints before sample collection.

SARS-CoV-2 Viral RNA Analysis
SARS-CoV-2 viral RNA content was tested using validated primers and probes targeting the nucleocapsid (N) gene, and data were acquired with the Bio-Rad real-time PCR system [31].

Enzyme-Linked Immunosorbent Assay
Enzyme-linked immunosorbent assay (ELISA) was used for detecting IgM and IgG antibodies against the SARS-CoV-2 N, S, and receptor-binding domain (RBD) as previously described [32]. Briefly, the purified RBD, S, and N proteins of SARS-CoV-2 were used as coating antigens. Horseradish peroxidase (HRP)-conjugated goat anti-human IgG (Sigma-Aldrich, St. Louis, MO, USA) and Horseradish Peroxidase-conjugated AffiniPure Goat Anti-Human IgM (Jackson ImmunoResearch Inc., West Grove, PA, USA) were used as the second antibody in 1:60,000 dilutions. The optimal coating concentrations of RBD, S, and N antigen were 10 ng/well, 20 ng/well, and 10 ng/well. The optimal plasma dilutions were 1:400.

Neutralizing Antibody Detection
Among the 203 patient samples in this cohort, we had sufficient plasma samples of only 81 patients to measure the neutralizing antibody level. The neutralization assay was performed with Vero E6 cells. Serial two-fold dilutions of plasma samples (starting at 1:10) were pre-incubated with SARS-CoV-2 at 100 TCID50 (50% tissue culture infective doses) for 2 h and then added to Vero E6 cells. The virus/plasma mixtures were removed after 1 h, and fresh growth medium was added to each well. The cytopathic effects were evaluated 5 days after incubation at 37 • C. For each plasma dilution, 4 duplicate wells were used. The neutralizing antibody titers were calculated using the Reed and Muench method [33].

Statistical Analysis
Statistical analysis of genetic data was performed using the chi-square test with SPSS software (version 23.0, SPSS Inc., Chicago, IL, USA). The statistical significance of allele frequencies was determined with Fisher's exact test. Statistical analysis of cytokine and chemokine data and antibody data was performed using Student's t test or the Mann-Whitney test, and the correlation analysis was performed using Pearson or Spearman correlation analysis with SPSS software (version 23.0). * indicates a significant difference p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, and **** indicates p < 0.0001.

IFITM3 Inhibits SARS-CoV-2 Entry
As IFITM proteins diminish the entry of different viruses, we next sought to examine the effect of IFITM3 on SARS-CoV-2 entry by BlaM-Vpr fusion assay. BlaM-Vpr is incorporated into SARS-CoV-2 S-pseudotyped lentiviral particles and delivered into target cells upon virus entry. Cleavage of CCF2-AM by BlaM-Vpr produces a fluorescence emission shift from green (520 nm) to blue (450 nm) ( Figure 2A). As shown in Figure 2B,C, IFITM3 expression reduced SARS-CoV-2 S-mediated virus entry in a dose-dependent manner. Compared with wild-type IFITM3, IFITM3(∆1-21) had minimal impact on virus entry ( Figure 2D,E). These data support that IFITM3 inhibits SARS-CoV-2 infection by preventing virus entry.
We further investigated the effect of IFITM3 on SARS-CoV-2 S-induced membrane fusion in the cell-cell fusion assay. SARS-CoV-2 S and Tat transfected HeLa cells were cocultured with HEK293T-ACE2 cells that expressed IFITM3 and Tat-inducible LTR-luciferase. After cell-cell fusion, Tat transactivates luciferase expression. Results of luciferase assay showed that IFITM3 inhibited SARS-CoV-2 S-mediated cell-cell fusion, while no effect was observed for IFITM3(∆1-21) ( Figure S2). Together, these data demonstrate that IFITM3 inhibits SARS-CoV-2 S-protein-mediated virion-cell fusion, as well as cell-cell fusion.

TMPRSS2 Attenuated the Inhibition of IFITM3
TMPRSS2 allows SARS-CoV-2 entry at the cell surface [37,38], and has been shown to render SARS-CoV-2 resistant to the inhibition of IFITM3 [26,39,40]. Indeed, when TMPRSS2 and IFITM3 were co-expressed in 293T-ACE2 cells, infection of SARS-CoV-2 S-pseudotyped viruses was less inhibited by IFITM3 compared to that in cells not expressing TMPRSS2 ( Figure 3A). We further knocked down IFITM3 in TMPRSS2-positive Calu-3 cells, and did not observe any effect on infection of SARS-CoV-2 S-pseudotyped virus ( Figure 3B). These data are consistent with previous reports that TMPRSS2 counters the inhibition of IFITM proteins [26,39,40]. ure 2B,C, IFITM3 expression reduced SARS-CoV-2 S-mediated virus entry in a dose-dependent manner. Compared with wild-type IFITM3, IFITM3(Δ1-21) had minimal impact on virus entry ( Figure 2D,E). These data support that IFITM3 inhibits SARS-CoV-2 infection by preventing virus entry. S-pseudotyped viruses was less inhibited by IFITM3 compared to that in pressing TMPRSS2 ( Figure 3A). We further knocked down IFITM3 in TMPR Calu-3 cells, and did not observe any effect on infection of SARS-CoV-2 Svirus ( Figure 3B). These data are consistent with previous reports that TM ters the inhibition of IFITM proteins [26,39,40]. were transfected with siRNAs targeting IFITM3 (si2, si3) and then infected with len viruses pseudotyped with SARS-CoV-2 spike protein. Infection efficiency was dete ciferase activities and normalized to that of cells transfected with control siRNA (N of IFITM3 and actin was measured by Western blotting. Data in the bar charts ar mean ± SD of three independent experiments. Statistical significance was determi dent's t test. ns, not significant; *, p < 0.05; ***, p < 0.001; NC, negative control; si2, IF si3, IFITM3 siRNA3. Two SNPs of IFITM3, rs12252-C, and rs34481144-A ( Figure 4A) were reported to be associated with the disease severity of several virus infections [18][19][20][21][22][23]. To investigate the possible association of these SNPs with SARS-CoV-2 infection, we sequenced the IFITM3 locus of 203 hospitalized COVID-19 patients, which included 133 mild cases (65.5%), 43 severe cases (21.2%), and 27 deaths (13.3%) ( Figure 4B). Sequence analysis showed that 77 patients (37.9%) had the rs12252 CC genotype, 88 patients (43.4%) had the CT genotype, and 38 patients (18.7%) had the TT WT genotype ( Figure 4B). Compared with the prevalence rate of 25.4% CC in the healthy Chinese Han population (from the 1000 genome project), the frequency of the CC genotype in 203 COVID-19 patients (37.9%) was significantly increased (p = 0.04) ( Figure 4B,C), even though total C allele frequency did not show significant changes ( Figure 4D). This result suggests that the CC genotype may increase the risk of SARS-CoV-2 infection. We further explored the correlation between CC genotype and disease severity, and found that the distributions of CC and CT/TT were 36.1% and 63.9% in mild cases, 41.9% and 58.1% in severe cases, 40.7% and 59.3% in deaths ( Figure 4E,F). Although no statistically significant differences between these groups were found, the CC genotype in the severe and dead population tends to be over-presented. Next, the viral loads in throat swabs were determined, and the results showed no differences between CC and CT/TT genotypes in the entire patient cohort ( Figure 4G), patients with mild symptoms (Figure 4H), or patients with severe symptoms ( Figure 4I).  Furthermore, we investigated the association of another SNP rs34481144-A with SARS-CoV-2 infection. Only one patient in our cohort carries the AG genotype, which was consistent with 1000 genomes project data. Due to the limited number of the AG genotype in our patient cohort, we could not conclude the association of rs34481144 A with SARS-CoV-2 infection (Table S1).

The rs12252 CC Genotype Might Be Associated with SARS-CoV-2 Neutralizing Antibody Positivity
We next measured SARS-CoV-2 neutralizing antibodies in 81 patients (35 CC, 32 CT, and 14 TT genotypes) two to three weeks after the onset of symptoms. We found that 18 of 35 patients with the CC genotype (51.4%) were positive for neutralizing antibodies, and 34 were positive (73.9%) in the 46 patients of the CT or TT genotype ( Figure 6A). The neutralizing antibody positivity rate was significantly lower in the CC group (χ 2 = 4.356, p = 0.037), although no difference in neutralizing antibody titers in all positive cases was observed ( Figure 6B). These data suggest that the rs12252 CC genotype is probably associated with poor neutralizing antibody production. In addition, we measured the levels of anti-N antibody, anti-S antibody, and anti-receptor-binding-domain (RBD) antibody. There were no differences in the levels of anti-N antibody or anti-S antibody between patients with CC and CT/TT genotypes, whereas patients with the CC genotype had a lower level of anti-RBD antibody IgG ( Figure S4).

The rs12252 CC Genotype Might Be Associated with SARS-CoV-2 Neutralizing Antibody Positivity
We next measured SARS-CoV-2 neutralizing antibodies in 81 patients (35 CC, 32 CT, and 14 TT genotypes) two to three weeks after the onset of symptoms. We found that 18 of 35 patients with the CC genotype (51.4%) were positive for neutralizing antibodies, and 34 were positive (73.9%) in the 46 patients of the CT or TT genotype ( Figure  6A). The neutralizing antibody positivity rate was significantly lower in the CC group (χ 2 = 4.356, p = 0.037), although no difference in neutralizing antibody titers in all positive cases was observed ( Figure 6B). These data suggest that the rs12252 CC genotype is probably associated with poor neutralizing antibody production. In addition, we measured the levels of anti-N antibody, anti-S antibody, and anti-receptor-binding-domain (RBD) antibody. There were no differences in the levels of anti-N antibody or anti-S antibody between patients with CC and CT/TT genotypes, whereas patients with the CC genotype had a lower level of anti-RBD antibody IgG ( Figure S4).

Discussion
IFITM3 has been reported to inhibit a number of pathogenic viruses in the families of orthomyxoviruses, flaviviruses, filoviruses, and coronaviruses [7][8][9][10][11][12][13][14]. Interestingly, IFITM3 was reported to restrict the entry mediated by the spike proteins of SARS-CoV, MERS-CoV, hCoV-229E, and hCoV-NL63, yet it enhances the infection of hCoV-OC43 [8,9,15,41]. IFITM1, IFITM2, and IFITM3 were reported to inhibit SARS-CoV-2 infection, particularly in cells not expressing the TMPRSS2 protease, which allows SARS-CoV-2 entry at the plasma membrane [26,39,40,42,43]. In this study, we further showed that IFITM3 inhibited SARS-CoV-2 infection by preventing virus entry. The first 21 amino acids of IFITM3, which carry the endocytic sorting signal 20YEML23 and the Y20 phosphorylation site and have been shown essential for the restriction of VSV and influenza A virus, are also important for inhibiting SARS-CoV-2. In addition, results of BlaM-Vpr fusion assay and cell-cell fusion assay support that IFITM3 inhibited SARS-CoV-2 S-protein-mediated virion-cell fusion and cell-cell fusion, consistent with a previous report [40].
The rs12252 C allele of IFITM3 has been reported to be associated with the disease severity of several viral infections, including influenza virus, Hantaan virus, and HIV-1 [18][19][20][21][22][23]. We now show that our COVID-19 cohort has a significantly higher presentation of the rs12252 CC homozygous genotype compared to the general control population, which is in agreement with previous reports [44][45][46]. The rs12252 C allele was predicted to affect the splicing of IFITM3 RNA, resulting in the loss of the first 21 amino acids of IFITM3 and the expression of a truncated version of IFITM3 protein ∆(1-21), which did not inhibit influenza A virus (A/WSN/1933) [16]. Our study showed that this ∆(1-21) IFITM3 mutant also lost the ability to inhibit SARS-CoV-2 infection, which at least partially underlies the increased risk of acquiring SARS-CoV-2 for those carrying the rs12252 CC genotype. Given the current lack of strong evidence supporting the expression of ∆(1-21) IFITM3 in the rs12252 CC carriers [47,48], the molecular mechanisms behind the association of rs12252 CC with viral infections warrant further investigation. In any case, the rs12252 C may serve as one genetic marker for vulnerability to SARS-CoV-2 infection. IFITM3 SNP rs12252 has been reported to be associated with SARS-CoV-2 infection risk, COVID-19 disease outcome, and mortality rate in some cohorts [44][45][46][49][50][51], but not in others [52,53]. In this cohort of Chinese COVID-19 patients, we did not observe a statistically significant difference in the frequency of the CC genotype between the mild, severe, and fatal COVID-19 cases, nor the viral loads, but did find significantly higher presentations of the CC genotype in SARS-CoV-2 positive individuals, suggesting an association of IFITM3 with the acquisition of SARS-CoV-2 infection. These results together suggest that the ultimate effect of IFITM3 polymorphisms on SARS-CoV-2 infection and COVID-19 disease severity might depend on the genetic background of the studied populations.
IFITM3 SNP rs34481144 was reported to be associated with the risk for hospitalization after SARS-CoV-2 infection [50,51]. In this study, we also investigated the SNP rs34481144 at the IFITM3 core promoter, and found only one patient carrying the AG genotype in our cohort. This is consistent with 1000 genomes project data, i.e., Asian descents have a very low frequency of rs34481144 AA and AG genotypes, which precludes the further study of the possible association of this SNP with COVID-19 using our cohort.
COVID-19 is characterized by high levels of pro-inflammatory cytokines and chemokines, including IL-2, IL-7, IL-10, G-CSF, IP10, MIP-1A, MCP-1, and TNFα, especially in patients cared for in ICU [3]. Early studies have shown that increased levels of pro-inflammatory cytokines in plasma, such as MCP1, were associated with pulmonary inflammation and severe lung damage in SARS-CoV patients [54]. In our COVID-19 cohort, we observed increased levels of IL-2, IL-4, IL-15, IL17, and IP-10, which likely activate both T-helper-1 (Th1) and T-helper-2 (Th2) cells and lead to lung inflammation and damage, similar to what has been reported in SARS-CoV and MERS-CoV infections. In addition, we detected higher levels of RANTES and lower levels of IL-2, IL17, and basic FGF in CC carriers, which suggests that Th2 and T-helper-17 (Th17) cell functions might have been compromised in the CC patients. A detailed follow up of cytokine levels in COVID-19 patients at all stages of the disease will help to illustrate a more accurate map of host inflammatory responses to SARS-CoV-2 infection. Only 21 patients had sequential samples, and we therefore chose to evaluate the levels of cytokines and chemokines in these patients. Moreover, for the reason of strict blood sample control, we did not obtain sufficient samples to measure the levels of cytokines and Viruses 2022, 14, 2553 13 of 16 chemokines in some patients. Due to the enormous pressures doctors and nurses faced in the early phase of the COVID-19 pandemic, we were unable to have regular periods of sample collection nor detailed clinical information, which limits the further interpretation of our data on cytokine and chemokine levels and association with IFITM3 polymorphisms in the context of clinical characteristics other than disease severity.
Polymorphisms in IFITM3 have been reported to modulate antibody responses to virus infection. Individuals with the rs12252 CC genotype have been shown to have higher pre-existing immunity to pdm09H1N1 than those with the TT genotype in a young people cohort [55]. Another study reported that the CC carriers presented lower seroconversion (SCR) and HI antibody titer 28 days after inoculation of the trivalent inactivated vaccine (TIV) [56]. In our study, we found that patients with the CC genotype had a lower positivity frequency of neutralizing antibody and lower levels of RBD antibody IgG than that in patients with CT/TT genotypes ( Figure S4). These findings indicate that the rs12252 SNP in IFITM3 may modulate antibody responses, especially IgG production. A previous study showed that IFITM3 deficient mice had reduced the number of activated B cells and dysfunction of the germinal center to plasma/memory transition [56], which may in part explain the association of deficient antibody response with the rs12252 SNP in IFITM3.
In summary, our data demonstrate that IFITM3 is able to inhibit SARS-CoV-2 by impeding virus entry. We further observed the association of the rs12252 SNP in IFITM3 with a higher risk of SARS-CoV-2 acquisition, thus revealing one genetic factor underpinning the susceptibility to COVID-19.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Wuhan Infectious Diseases Hospital (protocol KY-2020-02.02, January 2020).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.