Main

The coronavirus disease 2019 (COVID-19) pandemic is the third zoonotic coronavirus (CoV) outbreak in the last 20 years, following severe acute respiratory syndrome (SARS) in 2003 and Middle East respiratory syndrome in 2012, both of which were associated with severe clinical symptoms and high mortality rates1,2,3. SARS-CoV-2 infection and disease severity have been linked to pre-existing disease conditions such as obesity4,5,6,7, in addition to age and sex8. Despite great efforts to identify therapeutics for the effective treatment of patients with COVID-19, particularly those with obesity and other pre-existing conditions, treatment options remain limited. Many studies have targeted SARS-CoV-2 virus main protease (Mpro) as a major target for drug screening to identify new drugs for COVID-19 treatment9,10,11,12. To accelerate development of therapeutic drugs for COVID-19 treatment, other studies have performed high-throughput screening of clinical-stage or US Food and Drug Administration (FDA)-approved small molecules to repurpose known drugs13,14,15. This approach has led to the identification of remdesivir, a viral RNA polymerase inhibitor and its approval as emergency use authorization for the treatment of patients with COVID-19 (refs. 16,17). However, the clinical efficacy of remdesivir is limited based on clinical trial data, particularly for patients with severe COVID-19 (refs. 17,18).

Besides drugs targeting the viral proteins for their ability to inhibit viral infection and replication, recent studies have shown that host proteins are critically required for SARS-CoV-2 infection and replication19,20,21,22,23,24. A recent study showed that the drug plitidepsin (Aplidin), which targets the host protein eEF1A, exhibits potent anti-SARS-CoV-2 activity in cell cultures and mouse models25. Another US FDA-approved inhibitor topotecan (TPT) targeting topoisomerase 1 has been reported to suppress the lethal inflammation induced by SARS-CoV-2 and reduce morbidity and mortality in a preclinical mouse model26. Thus, identification of additional host biological pathways and corresponding drugs is urgently needed to facilitate new drug development for the treatment of the COVID-19 disease and prevent viral spread.

Although it is largely unknown why individuals with obesity are closely linked to the high risk of SARS-CoV-2 infection and severe disease, metabolic pathways (particularly the lipid metabolism) have been implicated in viral entry into the cells through viral protein palmitoylation and membrane fusion27,28,29. The SARS-CoV-2 virus can hijack the host cell metabolism by remodeling the host folate (one-carbon metabolism) and tricarboxylic acid (TCA) cycle metabolism to acquire the building blocks required for replication30,31. Understanding how SARS-CoV-2 alters the host cell metabolism may lead to identification of potential new therapeutic targets for COVID-19. Targeting metabolic pathways may particularly be a potential antiviral strategy for COVID-19 treatment29. In efforts to identify new host pathways and proteins that are critically required for SARS-CoV-2 infection, we screened a sublibrary of short hairpin (sh)RNAs targeting 95 metabolic genes in HEK293T-hACE2 (stably expressing human ACE2) to test for their ability to inhibit infectious-clone-derived mNeonGreenSARS-CoV-2 reporter virus (SARS-CoV-2-mNG) infection32. Each gene was targeted with an average of three shRNAs. HEK293T-hACE2 cells were transfected with shRNAs, followed by infection with SARS-CoV-2-mNG viruses32. SARS-CoV-2 infection efficiency was determined based on green fluorescence intensity (Fig. 1a). After two rounds of independent screenings, we analyzed the data in a correlation plot for consistency (z score) and found that the knockdown of several key genes (labeled in red) conferred the host cells’ resistance to SARS-CoV-2 infection (Fig. 1b and Supplementary Table 1). Similar results were obtained with knockdown of selected genes with specific shRNAs in HEK293T-hACE2 cells infected with SARS-CoV-2-mNG viruses (Fig. 1c,d). Among them, the knockdown of FASN and ACACA, which are key rate-limiting enzymes in fatty acid synthesis, markedly reduced SARS-CoV-2 infection. ACACA encodes acetyl-CoA carboxylase-α (ACC1), a rate-limiting enzyme that converts acetyl-CoA to malonyl-CoA. FASN encodes fatty acid synthase (FAS), a whole enzyme of FAS containing two identical multifunctional proteins. FAS synthesizes palmitate from acetyl-CoA to malonyl-CoA in the presence of NADPH. Therefore, both ACC1 and FAS act in concert to synthesize palmitate (Extended Data Fig. 1a), a key product of subsequent lipid synthesis.

Fig. 1: ShRNA library screen identifies FASN as a critical factor for SARS-CoV-2 infection.
figure 1

a, A schematic diagram of shRNA sublibrary screening against SARS-CoV-2 infection (created with BioRender.com). b, Correlation plot indicates activity (z score) of each gene in the two replicate screens. c,d, Validation of gene hits. HEK293T-hACE2 cells were transfected with indicated shRNAs. After 3 d, cells were infected SARS-CoV-2-mNG at an MOI of 0.01. Cells were subjected to fluorescence microscopy analyses (c) and qPCR analysis (d) at 24 h post infection (hpi). Scale bar, 200 μm. e, Generation of FASN KO HEK293T-hACE2 cells. Representative images of western blot analysis for FAS expression in HEK293T-hACE2 cells transduced with puro-guide control, FASN sgRNA1 (clone KO-1 and KO-2) or FASN sgRNA2 (clone KO-3 and KO-4) lentivirus. f, FASN KO inhibits SARS-CoV-2 infection. Control and FASN KO HEK293T-hACE2 cells were infected with SARS-CoV-2-mNG at an MOI of 0.01. The cells were imaged with fluorescence microscopy at 24 hpi. Viral infection was quantified by green fluorescence intensity and normalized by nuclear dye Hoechst 33342. g, FASN KO inhibits SARS-CoV-2 viral replication. Control and FASN KO HEK293T-hACE2 cells were infected with SARS-CoV-2 (USA-WA1/2020 strain unless otherwise stated) with an MOI of 0.01. Cells were subjected to qPCR analysis for viral RNA at 24 hpi. Data in d,f,g are representative of three independent experiments and are plotted as mean ± s.d. (n = 3 (f), n = 4 (d,g) per group). Experiments in c,e were independently repeated three times with similar results. Statistical analyses were performed with one-way analysis of variance (ANOVA) followed by Dunnett’s post-test compared with control cells (d,f,g). **P < 0.01, ****P < 0.0001; NS, not significant.

Source data

Full size image

To further substantiate the role of FAS in response to SARS-CoV-2 infection, we generated FASN knockout (KO) HEK293T-hACE2 cells (Fig. 1e). We showed that FAS-deficient cells markedly repressed the infection of the SARS-CoV-2-mNG virus (Fig. 1f) and the amount of viral RNA in the infected cells (Fig. 1g). Furthermore, we validated our observation in Caco-2 cells and showed that the KO of FASN significantly decreased the amount of SARS-CoV-2 viral RNA and infectious viral particles in the supernatant (Extended Data Fig. 1b–d). Taken together, these results suggest that FASN deficiency markedly inhibits SARS-CoV-2 infection.

As many FAS inhibitors are available, we next explore the therapeutic potential of FAS as a target for the treatment of COVID-19. To this end, we screened 22 FAS inhibitors for their ability to inhibit SARS-CoV-2 infection. HEK293T-hACE2 cells were pre-treated with dimethylsulfoxide (DMSO) control or FAS inhibitors for 16 h and then infected with the SARS-CoV-2-mNG virus. Among the 22 inhibitors tested at a final concentration of 5 µM, we found 11 FAS inhibitors that could markedly inhibit SARS-CoV-2 infection without obvious toxicity (Fig. 2a). Among them, orlistat is a US FDA-approved anti-obesity drug, while TVB-2640 is under phase II clinical trials as an anticancer therapy. Orlistat and TVB-2640 inhibit lipases to reduce the absorption of dietary fat and FAS to reduce lipid synthesis33. Next, we performed dose titration of seven inhibitors with the most promising inhibitory activity. Remdesivir was used as a positive control in our assay. TVB-3664 strongly inhibited SARS-CoV-2 with a half maximal effective concentration (EC50) value of 0.07 nM, which was five times more potent than remdesivir (EC50 = 0.4 nM) (Fig. 2b). Orlistat and TVB-2640 also markedly inhibited SARS-CoV-2 with EC50s of 390 nM and 4 nM, respectively. Similarly, the remaining FAS inhibitors TVB-3166, GSK-214069, FASN-IN-4 and FT113 inhibited SARS-CoV-2 with an EC50 of 11 nM, 0.9 nM, 18.6 nM and 17 nM, respectively (Fig. 2b). Furthermore, the cytotoxicity of these seven inhibitors was evaluated by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay in parallel with their antiviral activity. We did not observe significant toxicity at the working concentration of all seven inhibitors (Fig. 2b). Notably, the FAS inhibitor TVB-3664 (20 nM) still had a strong inhibitory effect even when the initial viral amount was increased by 100 times (multiplicity of infection (MOI) from 0.01 to 1). By contrast, remdesivir (20 nM) failed to inhibit SARS-CoV-2 infection at this level (Fig. 2c and Extended Data Fig. 2a).

Fig. 2: Screening of FAS inhibitors for anti-SARS-CoV-2 activity.
figure 2

a, In vitro screening of FAS inhibitors. HEK293T-hACE2 cells were pre-treated with 22 fatty acid metabolism-related inhibitors (at a final concentration of 5 µM) from 16 h before and during the infection. Inhibitor-treated cells were infected with the SARS-CoV-2-mNG virus at an MOI of 0.01. The cells were imaged with fluorescence microscopy at 24 hpi. Viral infection was quantified by green fluorescence intensity and normalized by nuclear dye Hoechst 33342. ☨ACSS2 is toxic and the treated cell died after treatment. b, Dose titration of selected inhibitors for anti-SARS-CoV-2 activity. Various concentrations of the selected inhibitors were used to assess their ability to inhibit SARS-CoV-2-mNG infection. Relative infection (black curve), cell viability (red curve) and EC50 value of antiviral ability are indicated. c, Direct comparison of FAS inhibitors with remdesivir by challenging with increased amounts of virus. HEK293T-hACE2 cells were treated with TVB-3664 (20 nM) and remdesivir (20 nM) and infected with SARS-CoV-2-mNG virus at MOI = 0.01, 0.1 or 1. Relative infection was quantified by the green fluorescence intensity at 16 hpi. d,e, Anti-SARS-CoV-2 ability of FAS inhibitors in HEK293T-hACE2 cells. Viral RNA (in the infected cells) and infectious viral particles (in the supernatant) were detected by qPCR (d) and plaque assay (e), respectively (orlistat 1 µM, TVB-2640 100 nM and TVB-3664 20 nM). f, Time-of-addition assay for the antiviral ability of FAS inhibitors in HEK293T-hACE2 cells. FAS inhibitors were added at different time points before and after SARS-CoV-2 infection. Viral RNA was detected at 24 hpi by qPCR. Ctrl, control. Data in a,c–f are representative of three independent experiments and are plotted as the mean ± s.d. (n = 3 (a–c), n = 4 (d–f) per group). Data in b are plotted as the mean ± s.d. for relative infection (quantitative fluorescence) and relative cell viability (MTT) performed in biological triplicate. EC50 value was calculated by a four-parameter logistic nonlinear regression model. Statistical analyses were performed using one-way ANOVA followed by Dunnett’s post-test (a,d–f) or two-way ANOVA followed by Tukey post-test (c). *P < 0.05, ****P < 0.0001 versus DMSO-treated cells (a,d–f) or remdesivir-treated cells (c). NS, not significant.

Source data

Full size image

We further tested the ability of FAS inhibitors to control viral replication. After the pre-treatment of selected inhibitors and the infection of SARS-CoV-2, the infectious viral particles in the supernatant and viral RNA in infected cells were quantified by qPCR and plaque assay, respectively. Consistent with SARS-CoV-2-mNG screening, TVB-3664, TVB-2640 and orlistat markedly inhibited the infectivity and replication of the SARS-CoV-2 virus (USA-WA1/2020 strain) (Fig. 2d,e). The replication of the SARS-CoV-2 virus was significantly reduced in all FAS inhibitor-treated HEK293T-hACE2 cells at the levels of viral RNA and infectious viral titers (Extended Data Fig. 2b,c). To further test the antiviral activity of the FAS inhibitors in other cell types, we treated human lung cell lines (NCI-H1355 and NCI-H1437), a human colon cell line (Caco-2) and mouse embryonic fibroblasts stably expressing hACE2 (MEF-hACE2) with the selected FAS inhibitors and found that these FAS inhibitors markedly inhibited SARS-CoV-2 in all the cells we tested, as indicated by virally infected green fluorescence-positive cells, viral RNA levels and infectious viral titers (Extended Data Fig. 2d–f).

To determine the antiviral activity of these FAS inhibitors after SARS-CoV-2 infection, we performed time-course experiments. TVB-3664, TVB-2640 and orlistat were added to HEK293T-hACE2 cells at different time points (−16, −1, 0, +1, +2, +4 and +6 h with infection time set as h 0). We found that all three inhibitors could potently inhibit SARS-CoV-2 even at 4–6 h post-infection (Fig. 2f). Taken together, our results suggest that FAS inhibitors can potently inhibit SARS-CoV-2 infection and replication in several tested cell lines.

A previous study shows that the palmitoylation modification of the SARS-CoV-2 spike protein is essential for viral infectivity34. FAS inhibitors reduced cellular levels of free fatty acids and palmitoylated proteins (Extended Data Fig. 3a,b). Furthermore, addition of BSA-conjugated palmitic acid (PA-BSA) reversed the antiviral effects of FASN KO and FAS inhibitors in a dose-dependent manner (Extended Data Fig. 3c,d). However, palmitic acid only partially reversed the antiviral effects of orlistat, probably because orlistat can also inhibit lipases in addition to its inhibitory activity on FAS.

Orlistat has been approved to treat obesity. Because of its clinical safety profile, we next tested whether orlistat has in vivo anti-SARS-CoV-2 activity. The K18-hACE2 mouse model has been successfully used for SARS-CoV-2 infection and COVID-19 diseases35,36. We infected hACE2 transgenic mice with SARS-CoV-2 virus (intranasal infection, 1 × 104 plaque-forming units (p.f.u.) per mouse), followed by intraperitoneal injection of orlistat (8 mg kg−1) daily starting at day 0 (Fig. 3a). DMSO/PBS buffer served as a control treatment. We found that SARS-CoV-2-infected mice treated with DMSO/PBS manifested a 15% loss of body weight and died between days 5–8 (Fig. 3b,c). In contrast, the orlistat treatment of SARS-CoV-2-infected mice markedly prolonged their survival, with 50% of the orlistat-treated SARS-CoV-2-infected mice gradually recovering their body weight and surviving (Fig. 3b,c). TVB-2640 also showed significant in vivo antiviral activity and protected SARS-CoV-2-infected mice for survival (Extended Data Fig. 4a,b). Furthermore, the capacity of orlistat to control local SARS-CoV-2 replication was evaluated in lung tissues from infected hACE2 transgenic mice. We found that orlistat treatment inhibited SARS-CoV-2 replication in lung tissues, with significantly decreased viral RNA amounts (Fig. 3d) and infectious viral titers (Fig. 3e). Consistently, histopathology analysis also revealed a reduction in lung inflammation in orlistat-treated mice compared with vehicle-treated mice (Fig. 3f,g and Extended Data Fig. 4c,d). Similar results were obtained using a different animal model, in which wild-type (WT) C57BL/6J (B6) mice were intranasally infected with adenovirus (Adv)-hACE2 to induce ACE2 expression in lung epithelial cells36. We showed that the administration of orlistat could significantly inhibited the disease’s progression with decreased histopathology in the lungs (Extended Data Fig. 4f,g), which is consistent with the observations in K18-hACE2 mice. To determine whether orlistat treatment has therapeutic effects on the disease’s progression after viral infection, we treated SARS-CoV-2-infected mice at day zero or day one after infection and found that orlistat treatment on day one after infection could also significantly prolong the animal survival compared with the control group, but failed to rescue infected mice (Extended Data Fig. 4e), suggesting that orlistat treatment has therapeutic effects even after SAR-CoV-2 infection. Taken together, these results suggest that orlistat treatment exhibits strong in vivo anti-SARS-CoV-2 activity, inhibits lung inflammation and disease progression and protects host survival after infection, thus providing a promising clinical candidate for the treatment of COVID-19.

Fig. 3: FAS inhibitors show in vivo anti-SARS-CoV-2 infection and prolong the infected mouse survival.
figure 3

a, A schematic presentation of the experiment design for SARS-CoV-2 infection in the hACE2 transgenic mouse model (created with BioRender.com). b,c, Orlistat reduces body weight loss and prolongs the mouse survival after SARS-CoV-2 infection. K18-hACE2 transgenic mice were infected with SARS-CoV-2 virus (intranasal infection, 1 × 104 p.f.u. per mouse, USA-WA1/2020) and intraperitoneally (i.p.) injected with orlistat (8 mg kg−1 body weight) (n = 6) or DMSO/PBS control buffer (n = 8) daily from day 0 to day 5. Body weight changes (b) and survival curves (c) were observed and analyzed. d–g, Orlistat reduces viral replication and disease progression in lung tissues after SARS-CoV-2 infection. K18-hACE2 transgenic mice were infected with SARS-CoV-2 and i.p. injection with orlistat. d, Local viral production by qPCR analysis with lung tissue RNA samples on day 3 (n = 5 for each group). e, Viral titration by plaque assay with the supernatant of homogenized lung tissues on day 3 (n = 5 for each group). f, Histopathology of formalin-fixed and hematoxylin and eosin (H&E)-stained lung tissues on day 3 (n = 7 for control group and n = 8 for orlistat group). dpi, days post infection. g, Analyses of pulmonary lesion sizes and pathological severity scores, based on the percentage of affected area in lung tissues. Representative images of plaque assay (e) and immunohistochemistry (IHC) staining (f) were presented. Scale bar, 400 μm for ×10 and 1 mm for ×4. b and c show combined data of two independent experiments and are plotted as the mean ± s.d., d–g are representative data of two independent experiments and are plotted as the mean ± s.d. Statistical analyses were performed using two-way ANOVA followed by Sidak’s post-test (b), two-side unpaired Student’s t-test (d,e,g) and Mantel–Cox log-rank test (c). *P < 0.05; **P < 0.01.

Source data

Full size image

New SARS-CoV-2 variants α (lineage B.1.1.7), β (lineage B.1.351) and δ (lineage B.1.617.2), which were initially detected in the United Kingdom, South Africa and India, respectively, are rapidly spreading globally, raising major public health concerns. While the Alpha variant is associated with increased transmissibility37 and risk of death compared with other variants38,39, the Beta variant may also increase the risk of infection in immunized individuals by escaping from vaccine-induced antibody neutralization40,41. Recently, the Delta variant has become a predominant SARS-CoV-2 variant that accounts for over 90% of the new cases in the United States, thus raising major concerns about this surge and reducing protection through neutralizing antibodies in vaccinated individuals42,43. Consequently, identification of innovative drugs to inhibit new SARS-CoV-2 variants is urgently needed. To this end, we tested whether FAS inhibitors could inhibit the infection and replication of the new SARS-CoV-2 variants. We infected HEK293T-hACE2 cells with SARS-CoV-2 variants and found that FAS deficiency markedly reduced viral RNA levels in FASN KO cells, compared with control cells (Fig. 4a). Notably, FAS inhibitors (TVB-3664, TVB-2640 and orlistat) could strikingly inhibit SARS-CoV-2 new variants with similar inhibitory effects on the early-lineage SARS-CoV-2 (USA-WA1/2020) virus (Fig. 4b). Similar inhibitory activities of these FAS inhibitors were observed in NCI-H1355 and MEF-hACE2 cells infected with SAR-CoV-2 variants (Extended Data Fig. 5a,b). Taken together, our data suggest that FAS inhibitors effectively inhibit SARS-CoV-2 and its dominant variants, suggesting broad antiviral activity against early-lineage SARS-CoV-2 and new variants.

Fig. 4: FAS inhibitors show antiviral activity against SARS-CoV-2 new variants.
figure 4

a, FAS deficiency inhibits SARS-CoV-2 new variants. Control and FASN KO HEK293T-hACE2 cells were infected with α, β and δ strains with an MOI of 0.01. The cells were subjected to qPCR analysis for viral RNA at 24 hpi. b, Antiviral ability evaluation of FAS inhibitors against SARS-CoV-2 variants (B.1.1.7 and B.1.351). HEK293T-hACE2 cells were treated with orlistat, TVB-2640 and TVB-3664 at the indicated concentrations and infected with SARS-CoV-2 or variants with an MOI of 0.01. Relative viral RNA was detected by qPCR at 24 hpi. Data in a,b are representative of three independent experiments and are plotted as the mean ± s.d. (n = 4 (a,b) per group). Data in b were normalized to the average of DMSO-treated cells. Statistical analysis in a was performed using one-way ANOVA followed by Dunnett’s post-test, all experimental groups compared with scramble sgRNA-transduced HEK293T-hACE2 cells. ****P < 0.0001.

Source data

Full size image

The findings presented here provide direct evidence that the ACC1-FAS lipid synthesis pathway is critically required for SARS-CoV-2 infection and replication. Notably, we identified several FAS inhibitors, including the US FDA-approved drug orlistat that markedly reduce SARS-CoV-2 infection and disease progression in cell cultures and mouse models. During our work, several groups have shown that host proteins and factors are critically required for SARS-CoV-2 infection and replication19,20,21,22,23,24. Although a lipid metabolism, particularly cholesterol, has been demonstrated to play a role in SARS-CoV-2 infection, ACC1 and FAS in the lipid synthesis pathway have not been reported, especially for inhibiting the new variants.

Individuals with pre-existing conditions such as obesity have been closely associated with a high risk of SARS-CoV-2 infection, severe COVID-19 disease progression and even death, but the underlying mechanisms remain largely unknown. The infection and replication of the coronavirus involve multiple critical interactions with host cell membranes, including during viral entry and virus release44. Lipids represent the structural foundations of cellular and viral membranes, thus are important for viral replication by regulating the membrane fusion, envelopment and transformation44. Therefore, inhibition of the lipid synthesis pathways may lead to reduced viral infection44. FAS is one of the major targets for obesity and also a candidate target for anticancer therapy. The first FAS inhibitor, orlistat, is an orally bioavailable nonprescription drug for obesity management that is reasonably priced and generally well tolerated by healthy people45. It should be noted that the early treatment of orlistat as a preventive medicine for obesity and SARS-CoV-2 infection could be the key to inhibit viral replication, whereas the late treatment, after the onset of symptoms or severe disease, might be less efficacious. Although a previous study reports the antiviral potential of orlistat in multiple viruses, such as dengue virus, Japanese encephalitis virus, Zika virus and chikungunya virus46, but none of them has reached clinical treatment for viral pathogens. A recent work also shows that orlistat can suppress SARS-CoV-2 replication in an in vitro study47. TVB-2640 is the second clinical level FAS inhibitor and is now under phase II clinical study for cancer treatment. Both orlistat and TVB-2640 have established safety profiles. As TVB-2640 is ~100 times lower than orlistat in HEK293T-hACE2 cells, further studies on TVB-2640 as a promising therapeutic candidate are warranted. Taken together, our study has identified orlistat and TVB-2640 as potential anti-SARS-CoV-2 drugs with in vivo efficacy, therefore raising the possibility for further testing these drugs in other SARS-CoV-2 animal models and future human clinical studies to test their protective and therapeutic effects on infection of SARS-CoV-2 and its emerging variants.

Methods

Cells and viruses

HEK293T-hACE2 (BEI Resources, cat. no. NR-52511), Vero-E6 (ATCC, cat. no. CRL-1586), MEF-hACE2 (mouse embryonic fibroblasts transduced with human ACE2 were generated in our laboratory) were cultured in Dulbecco’s modified Eagle medium (DMEM; Thermo Fisher) supplemented with 10% FBS. Caco-2 (ATCC, cat. no. HTB-37) cells were cultured in Eagle’s minimum essential medium (Thermo Fisher) supplemented with 20% FBS. NCI-H1355 (ATCC, cat. no. CRL-5865) and NCI-H1437 (ATCC, cat. no. CRL-5872) cells were cultured in RPMI 1640 medium (Thermo Fisher) supplemented with 10% FBS. All these cell lines were supplemented with penicillin (100 U ml−1, Thermo Fisher) and streptomycin (100 μg ml−1, Thermo Fisher) and maintained at 37 °C in a humidified atmosphere with 5% CO2. In this manuscript, the SARS-CoV-2 virus specifically refers to the USA-WA1/2020 strain unless otherwise stated, which was originally isolated from an oropharyngeal swab from a patient with a respiratory illness who developed clinical disease (COVID-19) in January 2020 in Washington, USA. SARS-CoV-2 virus (USA-WA1/2020) was obtained from University of Southern California (USC) BSL3 Core and the original virus was requested from BEI Resources (NR-52281). SARS-CoV-2-mNG virus (icSARS-CoV-2-mNG) was a generous gift from P.-Y. Shi’s laboratory at the University of Texas Medical Branch. SARS-CoV-2 Alpha variant (lineage B.1.1.7), Beta variant (lineage B.1.351) and Delta variant (lineage B.1.617.2) strains were obtained from USC BSL3 (biosafety level 3) Core and the original viruses were obtained from BEI Resources (NR-54000, NR-54009 and NR-55611). Viral stocks used in this study were sequence verified by USC BSL3 Core. All SARS-CoV-2 virus-related BSL3 experiments were conducted in the Hastings Foundation and The Wright Foundation laboratories at USC. The standard operating procedure of the BSL3 level experiments has been validated and approved. Biological use authorization was approved by the biosafety committee of USC.

Animals and in vivo procedures

K18-hACE2 transgenic mice were purchased from Jackson Laboratory (cat. no. 034860), then bred and maintained in specific-pathogen-free facilities at the USC. Genotyping was performed with primers from Jackson Laboratory and hemizygous mice were used in our study. C57BL/6J mice were purchased from Jackson Laboratory (cat. no. 000664). In all experiments, mice were treated from 8–10 weeks old; and both males and females were used. Mice were housed under a 12:12 light:dark cycle and all the facilities are accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. All animal studies were approved by the Institutional Animal Care and Use Committee of the USC.

For in vivo SARS-CoV-2 infection, K18-hACE2 transgenic mice were first transferred into the ABSL3 facility of USC. All animal procedures (including infection, injection, weighing and killing) were performed after anesthesia by isoflurane. Mice were intranasally infected with 104 p.f.u. of SARS-CoV-2 virus (USA-WA1/2020) in 30 μl serum-free DMEM. Orlistat was first dissolved in DMSO and diluted in PBS for intraperitoneal injection. The same volume of DMSO/PBS buffer was used in control groups. Orlistat or TVB-2640 was intraperitoneally injected daily (8 mg kg−1 body weight) from the indicated starting time points to day 5 post-infection.

On days 3 and 5, a subset of mice was killed by isoflurane overdose and tissue samples were collected for histopathology and viral titer analyses. The left lung of each mouse was fixed by formalin for sectioning and H&E staining. The superior and middle lobes of the lungs were weighed, homogenized and lysed in 1 ml TRIzol for RNA isolation. The inferior and post-caval lobes of the lungs were weighed and stored at −80 °C until being homogenized in 2 ml DMEM and titrated by plaque assay. Briefly, supernatants of homogenized tissues were serially diluted in DMEM, 250 μl of diluted samples were added to monolayers of Vero-E6 cells, followed by the overlay of 1% low-melting agarose (prepared in 1× DMEM). Plaques were visualized on day 3 after formalin fixation and staining with 0.2% crystal violet. Triplicated wells were performed for each mouse for a mean value.

For the adenovirus (Adv)-hACE2 mouse model, C57BL/6J (B6) mice were intranasally infected with 2.5 × 108 p.f.u. of human ACE2-expressing adenovirus on day −5 before SARS-CoV-2 viral infection. On day 0, Adv-hACE2-treated B6 mice were intranasally infected with 1 × 104 p.f.u. of SARS-CoV-2 virus (USA-WA1/2020) in 30 μl serum-free DMEM. Orlistat (8 mg kg−1 body weight) or DMSO/PBS control buffer were administrated daily via intraperitoneal injection. Mice were killed on day 3 for tissue collection.

Screening of shRNA library against SARS-CoV-2

HEK293T-hACE2 cells were plated in 96-well plates (4 × 104 cells per well) and transfected with the shRNA vectors for metabolism genes (150 ng per well). At 72 h post-transfection, cells were subjected to SARS-CoV-2-mNG infection at an MOI of 0.01. Cells were subjected to fluorescence density analyses using FilterMax F5 microplate reader at 48 hpi. The z score was calculated with the average and s.d. of each plate. The screening was repeated in two independent experiments for analysis.

Generation of FASN KO cells

HEK293T-hACE2 and Caco-2 cells were transduced with Lenti-Cas9 (Lenti-Cas9-Blast, Addgene 52962) and selected with blasticidin (10 mg ml−1) for 14 d. The expression of Cas9 was validated by western blotting. The HEK293T-hACE2-Cas9 cells were then transduced with FASN sgRNA lentivirus and selected with puromycin (1 mg ml−1) for 7 d. The FASN KO bulk cells were plated into 96-well plates (0.3 cells per well) for FASN KO single-cell clone generation. The sequences of FASN sgRNA are as follows: 5ʹ-TTCTGGGACAACCTCATCGG-3ʹ, 5ʹ-TCCATCGTGTGTGCCTGCTT-3ʹ.

Plaque assay

Vero-E6 cells were plated as confluent monolayers in 12-well tissue culture plates. Medium was removed and wells was washed with DMEM without FBS. Then, 250 μl of serially diluted cell culture supernatant or homogenized tissue lysates was added into the wells, followed by an incubation of 45 min in 37 °C, 5% CO2 incubator with a gentle shake. After viral adsorption, supernatant was removed from the well and cells were overlaid with the overlay medium (1:1 mixture of 2% low-melting agarose solution with 2× DMEM). Three days post-infection, cells were fixed with 4% formaldehyde for 24 h. Agar plugs were then removed and cells were stained with 0.2% crystal violet to count plaques for the calculation of viral titers.

Fluorescence assay and quantification of SARS-CoV-2 infection

SARS-CoV-2-mNG virus (icSARS-CoV-2-mNG)-infected HEK293T-hACE2, Caco-2 or MEF-hACE2 cells were subjected to fluorescence-based imaging assay at 24 h after infection or indicated time point. Live cell nucleic acid fluorescent dye Hoechst 33342 was added to the infected cell medium at a final concentration of 1 μg ml−1 and was incubated at 37 °C, 5% CO2 for at least 30 min before imaging. Images of green fluorescence (FITC channel) and Hoechst 33342 (DAPI channel) were acquired using Revolve Fluorescence Microscope (ECHO). The fluorescence intensity of images was analyzed by ImageJ v.1.53e. Green fluorescence intensity was normalized by Hoechst 33342 fluorescence intensity of the same image. The relative infection was determined by the normalized green fluorescence intensity and compared with the control cells (DMSO-treated or other control group indicated elsewhere).

Inhibitor and palmitic acid treatments

One day before treatment, HEK293T-hACE2 (1 × 105 cells per well), Caco-2 (2 × 105 cells per well) or MEF-hACE2 cells (5 × 104 cells per well) were seeded in 24-well plates in culture medium and incubated at 37 °C, 5% CO2 cell culture incubator. Indicated concentration of FAS inhibitors was added to the cell culture medium at 16 h before infection and maintained in the medium during infection. Pre-treated cells were infected with the SARS-CoV-2 virus or other strains as indicated at an MOI of 0.01 (HEK293T-hACE2 and MEF-hACE2) or 0.05 (Caco-2). Viral RNA or supernatant samples were collected as described in the SARS-CoV-2 viral RNA isolation and quantitative analysis and plaque assay sections in Methods.

For inhibitor dose titration, we generated a fourfold serial dilution of selected eight inhibitors with different starting concentrations of 20 µM (orlistat, FASN-IN-4, FT113), 320 nM (TVB-2640, TVB-3166, GSK-214069) or 80 nM (TVB-3664, remdesivir). Diluted inhibitors were added to the cells 16 h before infection and maintained in the medium during infection. After being transferred to the BSL3 facility, pre-treated cells were infected with the SARS-CoV-2-mNG virus (icSARS-CoV-2-mNG) with an MOI of 0.01. Fluorescence assay and quantification of SARS-CoV-2 infection were performed at 24 h after infection as described in the fluorescence assay and quantification of SARS-CoV-2 infection section in Methods. A four-parameter logistic nonlinear regression model was used to calculate EC50 by GraphPad Prism v.8.0.2 software.

For palmitic acid treatment, BSA-conjugated palmitic acid (PA-BSA; cat. no. P0500, Sigma) was prepared as previously described48. In brief, PA-BSA was diluted in prewarmed cell culture medium and mixed in a water sonicator for 5 min before adding to the cells. After 2 h of incubation in a 37 °C cell culture incubator, PA-BSA-treated cells were infected with SARS-CoV-2. Viral RNA was analyzed 24 h after viral infection by qPCR.

Inhibitors were purchased form Selleck Chem: A769662 (S2697); orlistat (S1629); isoniazid (S1937); desoxyrhaponticin (S3300); fatostatin HBr (S8284); TVB-2640 (S9714); C75 (S8915); FT113 (S6666); EGCG (S2250); betulin (S4754); ACSS2 (S8588); kaempferol (S2314); praeruptorin B (S9392); TVB-3664 (S8563); cerulenin (C2389); G28U CM (5446); GSK-2194069 (5303); curcumin (S1848); and remdesivir (S8932). MedChemExpress supplied FASN-IN-2 (HY-112829); FASN-IN-1 (HY-111777); FASN-IN-4 (HY-12648); and TVB-3166 (HY-120394). TOCRIS supplied G28U CM (5446) and GSK-2194069 (5303). Cerulenin was purchased from Sigma (C2389).

Cytotoxicity assay

One day before inhibitor treatment, 2 × 104 HEK293T-hACE2 cells were seeded in 100 μl cell culture medium in 96-well plates. The next day, a series dilution of each inhibitor at indicated concentration was added to the cells (5 μl per well in cell culture medium). DMSO treatment was set up as negative control and baseline. Cell cultures were incubated for 2 d at 37 °C and 5% CO2 cell culture incubator. After the incubation period, 10 μl of MTT (5 mg ml−1) labeling reagent was added to each well and the microplate was incubated for 4 h in a cell culture incubator. Then, 100 μl of the solubilization solution (10% SDS in 0.01 M HCl) was added into each well. The plate was then allowed to stand overnight in the cell culture incubator. The 590 nm absorbance of samples was detected using a Synergy 2 Multi-Mode Microplate Reader (BioTek) and analyzed by the software Gene 5 (BioTek).

SARS-CoV-2 viral RNA isolation and quantitative analysis

Total RNA was collected from infected cells or mouse tissues at indicated time point using TRIzol reagent (Thermo Fisher Scientific) and removed from the BSL3 facility. RNA was extracted according to manufacturer’s instructions. Then, 1 μg of total RNA was used for RT–PCR using the SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific) and random hexamers. Complementary DNA was diluted by 1:10 and 1 μl diluted cDNA was used for qPCR analysis. The qPCR analysis was performed on the Applied Biosystems QuantStudio 6 Flex Real-Time PCR system (Thermo Fisher Scientific) with iTaq Universal SYBR Green Supermix (Bio-Rad). Each sample was run in three or four technical replicated wells. One primer set (SC2-S-F: GCTGGTGCTGCAGCTTATTA; SC2-S-R: AGGGTCAAGTGCACAGTCTA) targeting SARS-CoV-2 virus spike protein was used to measure the relative SARS-CoV-2 viral RNA and normalized to a human GAPDH primer set (hGAPDH-F: GACAGTCAGCCGCATCTTCT; hGAPDH-R: GCCCAATACGACCAAATCCGT) for human samples and a mouse GAPDH primer set (mGapdh-F: GAAGGGCTCATGACCACAGT; mGapdh-R: GGATGCAGGGATGATGTTCT) for mouse samples. An in-house constructed spike protein expression plasmid was used to generate the standard curve of absolute quantification PCR analysis. The results were analyzed using software QuantStudio Software v.1.3 (Thermo Fisher Scientific). Each assay was performed in triplicate with four technical replicates and each assay included no-template negative controls.

Time-of-addition assay

A number of 1 × 105 HEK293T-hACE2 cells were seeded in 24-well plates in culture medium (DMEM + 10% FBS + 1× penicillin-streptomycin) and incubated for 24 h at 37 °C, 5% CO2 cell culture incubator. Inhibitors were added to the medium at indicated time point (−16 h, −1 h, 0 h, 1 h, 2 h, 4 h and 6 h), then the cells were transferred into the BSL3 facility before infection. SARS-CoV-2 virus was added to medium at an MOI of 0.01 at the time point of hour 0. Before sample collection, cells were washed twice in PBS. To determine the SARS-CoV-2 virus inhibition ability, total RNA of infected cells was collected 24 h after infection by TRIzol reagent (Thermo Fisher Scientific) according to manufacturer’s instructions and analyzed by qPCR as previously described in the SARS-CoV-2 viral RNA isolation and quantitative analysis section in Methods.

Virus replication assay

A total of 1 × 105 HEK293T-hACE2 cells were seeded in 24-well plates in culture medium (DMEM, 10% FBS, 1× penicillin-streptomycin) and incubated for 24 h at 37 °C, 5% CO2 cell culture incubator. Inhibitors were added to the medium 1 d before infection (−16 h). Pre-treated cells were transferred into the BSL3 facility and washed twice with DMEM without FBS. SARS-CoV-2 virus was diluted into 200 μl of DMEM without FBS at an MOI of 0.01 and added to the washed cells. Cells with infectious medium were incubated on a rocker for 45 min at 37 °C, 5% CO2. After infection, infectious medium was removed and washed twice in PBS. Fresh cell culture medium (DMEM, 10% FBS, 1× penicillin-streptomycin) containing inhibitors at indicated concentrations was added to infected cells and incubated at 37 °C, 5% CO2. Samples were collected at the indicated time points after infection (0 h, 4 h, 8 h, 16 h, 24 h, 32 h, 48 h and 72 h). To determine viral RNA amplification, RNA samples of infected cells were collected using TRIzol reagent (Thermo Fisher Scientific) according to manufacturer’s instructions and analyzed by qPCR as described in the SARS-CoV-2 viral RNA isolation and quantitative analysis section in Methods. To determine viral particle generation, supernatant was collected at the indicated time points. Cells or debris were removed by centrifugation at 500g for 3 min. A plaque assay was performed to detect the infectious particle number as described in the plaque assay section in Methods.

Protein expression analysis

Whole-cell extracts of cultured cells were isolated in with lysis and extraction buffer (50 mM HEPES, 150 mM NaCl, 1 mM EDTA, 10% glycerol, 1.5 mM MgCl2 and 1% Triton X-100). Cell debris was removed by centrifugation for 20 min at 12,000g at 4 °C in a microcentrifuge. Lysate was mixed with 5× SDS loading buffer and heated for 5 min at 100 °C. Prepared samples were resolved by 10% SDS–PAGE and transferred to PVDF membranes (Bio-Rad). The membrane was blocked for 1 h at room temperature using blocking buffer (5% nonfat dried milk diluted in TBST). Blots were incubated with anti-FAS monoclonal antibody (3180S, clone C20G5, Cell Signaling Technology) diluted at 1:1,000. Protein expression was normalized to β-actin (sc-47778, clone C4, Santa Cruz Biotechnology) diluted at 1:2,000. The HRP-conjugated anti-mouse IgG antibody (31430, clone 31430, Thermo Fisher Scientific) and the HRP-conjugated anti-rabbit IgG antibody (32460, clone 32460, Thermo Fisher Scientific) were used to detect primary antibodies at 1:10,000 dilution. HRP was detected using chemiluminescent HRP substrate (Millipore). Digital images were acquired with the ChemiDoc XRS+ System and analyzed by Image Lab v.5.1 (Bio-Rad).

Free fatty acid assay

The free fatty acid assay was performed with the Free Fatty Acid Assay kit (cat. no. Ab65341, Abcam) according to the manufacturer’s instructions with some modifications. Briefly, SARS-CoV-2 virus-infected and non-infected cells were washed with PBS and fixed with 1% SDS solution for 1 h. Collected cell lysates were homogenized in chloroform and incubated on ice for 30 min. The organic phase was collected after centrifuge and vacuumed dry to remove trace chloroform. Dried lipids were dissolved in Fatty Acid Assay Buffer for further assay procedures. Measurement was performed on a microplate reader at Ex/Em = 535/587 nm for fluorometric assay immediately after the reaction was completed.

Palmitoylated protein assay

The palmitoylated protein assay was performed with the EZClick Palmitoylated Protein Assay kit-Red (cat. no. K416, BioVision) according to the manufacturer’s instructions. Briefly, cells were seeded in 96-well plates and treated with FAS inhibitors for 24 h before the assay. Next day, cell medium was changed with fresh medium containing inhibitors and 1× EZClick Palmitic Acid Label probe for another 24 h. After incubation, cells were fixed, permeabilized, stained with 1× EZClick Reaction Cocktail and DAPI and evaluated by fluorescence microscope imaging (ZOE Fluorescent Cell Imager).

Histology and microscopy

Fresh lung tissues were fixed with 10% formalin for 24 h and then sent to USC School of Pharmacy Histology Laboratory for further processing and H&E staining. Slides were evaluated by a pathologist under a microscope and fields were randomly selected for analysis. Pictures were taken at ×4 and ×10 magnification using the Olympus BX61 microscope along with DP71 digital camera (Olympus).

Histopathology score in SARS-CoV-2 model

After H&E staining, the size of pulmonary lesions was determined by the mean percentage of affected area in each section of lobes from each animal. Lung tissues were then scored on a 0–4 system for pathological severity in these infected mice: 0, no pathological change; 1, affected area (≤10%); 2, affected area (≤30% and >10%); 3, affected area (≤60% and >30%); and 4, affected area (>60%).

Statistical analysis

Descriptive statistics, including means, s.d., medians and ranges, were computed for each group. Data are represented as mean ± s.d. unless otherwise indicated. The sample size for each experiment is included in the figure legends. Statistical analyses were performed with unpaired Student’s t-test, one-way ANOVA followed by Dunnett’s post-test, two-way ANOVA followed by Tukey post-test and two-way ANOVA followed by Sidak’s post-test, as indicated in figure legends. Differences in animal survival were evaluated with Mantel–Cox log-rank test. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 versus control groups. NS denotes not significant (P value >0.05). All analyses were performed with GraphPad Prism v.8.0.2 (GraphPad Software).

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.