key: cord-0801515-gii9oywl
authors: Mekhail, Katrina; Lee, Minhyoung; Sugiyama, Michael; Astori, Audrey; St-Germain, Jonathan; Latreille, Elyse; Khosraviani, Negar; Wei, Kuiru; Li, Zhijie; Rini, James; Lee, Warren L.; Antonescu, Costin; Raught, Brian; Fairn, Gregory D.
title: Fatty Acid Synthase inhibitor TVB-3166 prevents S-acylation of the Spike protein of human coronaviruses
date: 2022-05-16
journal: bioRxiv
DOI: 10.1101/2020.12.20.423603
sha: 3402d501b3f0674edacc30130a10624ab690e9b5
doc_id: 801515
cord_uid: gii9oywl

The Spike protein of SARS-CoV2 and other coronaviruses mediate host cell entry and are S-acylated on multiple phylogenetically conserved cysteine residues. Multiple protein acyltransferase enzymes of the ZDHHC family have been reported to modify Spike proteins post-translationally. Using resin-assisted capture mass spectrometry, we demonstrate that the Spike protein is S-acylated in SARS-CoV2 infected human and monkey cells. We further show that increased abundance of the human acyltransferase ZDHHC5 results in increased S-acylation of the SARS-CoV2 Spike protein, whereas ZDHHC5 knockout cells had a 40% reduction in the incorporation of an alkynyl-palmitate using click chemistry detection. We also find that the S-acylation of the Spike protein is not limited to palmitate, as clickable versions of myristate and stearate were also found on the immunocaptured protein. Yet, ZDHHC5 was highly selective for palmitate, suggesting that other ZDHHC enzymes mediated the incorporation of other fatty acyl chains. Thus, since multiple ZDHHC isoforms may modify the Spike protein, we examined the ability of the fatty acid synthase inhibitor TVB-3166 to prevent the S-acylation of the Spike proteins of SARS-CoV-2 and human CoV-229E. Treating cells with TVB-3166 inhibited S-acylation of ectopically expressed SARS-CoV2 Spike and attenuated the ability of SARS-CoV2 and human CoV-229E to spread in vitro. Additionally, treatment of mice with a comparatively low dose of TVB-3166 promoted survival from an otherwise fatal murine coronavirus infection. Our findings further substantiate the necessity of CoV Spike protein S-acylation and the potential use of fatty acid synthase inhibitors.

The Spike protein of SARS-CoV2 and other coronaviruses mediate host cell entry and are S-acylated on multiple phylogenetically conserved cysteine residues. Multiple protein acyltransferase enzymes of the ZDHHC family have been reported to modify Spike proteins post-translationally. Using resin-assisted capture mass spectrometry, we demonstrate that the Spike protein is S-acylated in SARS-CoV2 infected human and monkey cells. We further show that increased abundance of the human acyltransferase ZDHHC5 results in increased S-acylation of the SARS-CoV2 Spike protein, whereas ZDHHC5 knockout cells had a 40% reduction in the incorporation of an alkynyl-palmitate using click chemistry detection. We also find that the Sacylation of the Spike protein is not limited to palmitate, as clickable versions of myristate and stearate were also found on the immunocaptured protein. and animals and cause various diseases targeting different tissues, including respiratory, enteric, renal, and hepatic (3, 4) . The current pandemic highlights the need for effective treatments of this family of viruses. There are enormous efforts to develop prophylaxis and therapeutic options for SARS-CoV2, including vaccination, protease inhibitors, and soluble decoy receptors (5) (6) (7) . Many putative countermeasures target the Spike protein, which is required to attach the virus to cells via binding to the angiotensin-converting enzyme (ACE2) and potentially other host proteins (6) .

Further, Spike protein proteolytic cleavage generates an S2 fragment capable of stimulating viral fusion and payload delivery into the cytosol (8) . However, the high mutation rates of positivesense (+) sense RNA viruses may result in resistance to some of these therapeutic approaches (9, 10) . Additionally, many potential zoonotic coronavirus species in bat and civet populations necessitate exploring all available therapies (11, 12) . Thus, identifying a pan-CoV infection therapy that targets a host enzyme is desirable for current and future potential zoonotic infections.

Although not truly synonymous, S-palmitoylation or S-acylation are often used interchangeably to describe the reversible covalent addition of palmitoyl or other fatty acyl chains to cysteine (Cys) residues via a thioester bond. This modification can occur spontaneously in high concentrations of acyl-CoAs or catalyzed by a family of zinc finger Asp-His-His-Cyscontaining protein acyltransferases or simply ZDHHC enzymes. In the Spring of 2020, Krogan and colleagues reported that ZDHHC5 and its binding partner Golga7 interact with the Spike protein of SARS-CoV2 using an affinity purification/mass spectrometry approach (13) . The results suggest that the cytosolic tail of the Spike protein may be post-translationally S-acylated.

A variety of labs have substantiated this, and collectively, these studies have demonstrated that multiple protein acyltransferases, including ZDHHC2, 3, 5, 6, 8, 9, 11, 12, 20, 21 and 24, directly or indirectly influence the S-palmitoylation/acylation of the Spike protein depending on the cellular context (14) (15) (16) (17) (18) (19) (20) . These findings likely result from the lack of specificity of some ZDHHC enzymes and the promiscuous nature of some substrates. Considering that the Spike protein contains ten Cys residues in proximity to the membrane, this would only increase the chances of modification by various transferases.

Previously, the Spike protein of the murine hepatitis virus (MHV) required S-acylation of its C-terminal tail to support virion production (21) . Specifically, preventing S-acylation prevented the MHV Spike protein from being incorporated into new virions (21) . Further, a mutant version of Spike unable to be S-palmitoylated displayed a reduced ability to stimulate membrane fusion vis-à-vis syncytium formation field (21, 22) . Additionally, S-acylation of the SARS Spike is critical for its ability to catalyze cell-cell fusion (23) . Here we aimed to extend the findings and investigate the role of ZDHHC5 in Spike S-acylation and its role in the human cold virus CoV-229E. Given the large number of ZDHHC enzymes that may potentially Sacylate the Spike protein, we also sought to examine the ability of the fatty acid synthase inhibitor TVB-3166 to limit S-acylation.

Alignment of the C-terminal cytoplasmic tails of several human and murine CoVs reveals that the 20 amino acids adjacent to the membrane are highly enriched in Cys residues (Fig. 1A) .

Indeed, in the case of SARS-CoV2 Spike, half of the first 20 amino acids are Cys residues, providing ample sites for potential S-acylation. To explore this possibility, we first established a model to express and study the SARS-CoV2 Spike protein using ectopic expression of Spike tagged at the C-terminus with a C9 epitope (TETSQVAPA) ( Fig. 2A) (24, 25) . Immunoblotting of the Spike-C9 revealed numerous reactive bands in transfected cells but none in controls (Fig.   2B ). Consistent with previous reports (25, 26) , we detected several Spike molecular species, including a full-length ~190 kDa and a ~100 kDa fragment compatible with the furin-cleaved S2 fragment. To determine if the ectopically expressed Spike-C9 protein is acylated, we incubated transfected cells with BSA conjugated alkynyl-palmitate (15-hexadecynoic acid; 15-HDYA), a "clickable" palmitate analog (27) . Spike-C9 immunocaptured using anti-C9 beads was reacted with cyanine 5.5 (Cy5.5) -azide using a copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC) (28) (29) (30) . As shown in Fig. 1B , the Spike-C9 from cells incubated with alkynylpalmitate and reacted with the Cy5.5-azide generated a similar banding pattern as the anti-C9 antibody, indicating that the full-length and S2 fragment of the Spike protein are both acylated.

Considering that the SARS-CoV2 Spike protein cytoplasmic tail contains ten cysteine residues, we sought to confirm that the modifications are on Cys residues and the degree of modification. To do this, we used an approach termed acyl-polyethylene glycol (PEG) exchange (APE) that involves substituting acyl groups attached to Cys residues with a 5-kDa PEG mass tag field (31) . The unreacted protein (input) and the protein treated with hydroxylamine to cleave the thioester bond and remove the fatty acyl chains, but not reacted with PEG (-PEG), have an apparent molecular weight of ~190 kDa (Fig. 1C) . Spike-C9 treated with hydroxylamine and reacted with a maleimide-functionalized five kDa PEG revealed five distinct bands consistent with an unmodified band and four sites of S-acylation. Given the proximity of the ten Cys residues within a 20 amino acid stretch, it is unclear if the addition of multiple five kDa PEG molecules could be sterically hindered. Another possibility is that the addition of the C9 epitope to the C-terminal end of the Spike may result in less or variable acylation. Regardless, these results demonstrate that the Spike protein can be modified on numerous sites and that some individual proteins are modified with a minimum of four acyl chains.

Next, we sought to confirm the results obtained using the heterologously expressed epitope-tagged Spike with the native Spike in cells infected with the SARS-CoV2 virus. We used a strategy termed acyl resin-assisted capture (acyl-RAC) (32) . In this workflow, free Cys residues are chemically blocked using N-ethylmaleimide. Next acyl-cysteine thioester linkages are then cleaved using hydroxylamine, and then treated lysates are incubated with thiol-reactive beads allowing for the capture of the previously S-acylated proteins by a mixed disulphide exchange reaction. This assay is readily coupled to mass spectrometry (MS) based protein detection and quantitation. As shown in Table 1, Table S1 , and S2, the Acyl-RAC MS workflow can identify numerous host proteins that are known to be S-acylated, including flotillin, calnexin, scribble and SNAP23 in both Vero and HEK293 stably expressing both the ACE2 receptor and TMPRSS2 protease (or simply HEK A2T2) (33) . Analysis of cells infected with SARS-CoV2 revealed the presence of the Spike protein and that it was one of the most abundant S-acylated proteins in both cell types. Furthermore, despite the infection, most of the other host S-acylated proteins were unaltered. These results collectively demonstrate that both the ectopically expressed Spike and Spike delivered during SARS-CoV2 infection are S-acylated.

The CoV Spike polypeptides are class I viral fusion proteins that mediate membrane fusion (34) . This function is essential to initiating viral infection. To investigate the role of S-acylation of the SARS-CoV-2 Spike protein in this process, we used a site-directed mutagenesis approach to mutate all ten Cys residues in the cytoplasmic tail to Ser (termed Spike C->S ) (Fig 2A) . Lysates of HEK293T cells ectopically expressing Spike-C9 and Spike C->S -C9 were immunoblotted and revealed that the full-length Spike C->S -C9 was expressed at comparable levels to the wild-type protein and that there is also the production of the S2 fragment (Fig. 2B, C) . Our studies noticed that the relative proportion of full-length compared to the S2 was variable but consistent within individual biological replicates. Whether this is due to a cell-autonomous factor, the efficiency of transfection, or perhaps even post cell lysis proteolysis was not determined. We next confirmed that the Spike C->S -C9 is not S-acylated by measuring the addition of the 15-HDYA (Fig 2D, E) .

Recent studies have demonstrated that the C-terminal region of Spike interacts with a variety of vesicular transport machinery, including COPI and COPII coatomers (35) . To determine if loss of the S-acylation sites alters the protein delivery to the plasma membrane, we used an antibody that detects an extracellular epitope together with immunofluorescence microscopy and flow cytometry. We confirmed that comparable amounts of both the Spike-C9 and Spike C->S -C9 traffic to the plasma membrane (Fig. 2F , G). Together these results demonstrate that mutagenesis of the cytoplasmic Cys residues to Ser and the concomitant loss of S-acylation have minimal impact on expression levels and transport to the cell surface.

Previous studies have demonstrated that the SARS-CoV-2 Spike protein can catalyze syncytium formation, provided that neighbouring cells express ACE2 (6) . To determine if Sacylation of the Spike protein is required for syncytium formation, we utilized a co-culture strategy where cells co-transfected with plasmids encoding Spike-C9 and soluble mCherry were co-cultured with cells co-transfected with plasmids encoding ACE2 and soluble GFP. Cocultures containing cells expressing mCherry alone and cells with ACE2/GFP showed no signs of cell-cell fusion, as expected (Fig. 3A, B) . However, co-cultures with Spike/mCherry and ACE2/GFP displayed numerous large syncytia containing soluble GFP and mCherry ( Fig. 3 A,   B ). In contrast, HEK293T cells expressing Spike C->S -C9/mCherry co-cultured with ACE2/GFP expressing cells displayed only a few smaller mCherry and GFP double-positive cells (Fig 3A,   B ). These results demonstrate that one or more membrane-proximal cytoplasmic Cys residues are required for the Spike protein to facilitate syncytium formation.

The human genome encodes a family of 24 ZDHHC (including ZDHHC11B) palmitoyltransferases (22) . ZHHC5 was reported to interact with the SARS-CoV2 Spike protein physically, as shown using IP-MS (13) . We confirmed these results using coimmunoprecipitation and found that Spike-C9 could transiently interact with HA-tagged ZDHHC5 (Fig. S1 ). Given that numerous ZDHHC enzymes have been demonstrated to S-acylate the Spike protein, we wanted to determine the relative importance of ZDHHC5. Thus, we assessed the contribution of ZDHHC5 to the S-acylation of Spike in parental HEK293T cells and genome-edited cells deficient in ZDHHC5 expression (Fig. S2 ). Spike-C9 was transiently expressed in wild-type and ZDHHC5 ko cells, incubated with 15-HDYA and subsequently processed as in Figure 1 to determine the extent of S-palmitoylation. Notably, the loss of ZDHHC5 resulted in nearly a 40% reduction in Spike S-palmitoylation (Fig 4A, B) . Thus, even though reports demonstrate that upwards of 11 ZDHHC enzymes can modify Spike, our experiments with the clickable palmitate suggest ZDHHC5 is responsible for a significant fraction of the post-translation modification.

As mentioned previously, the 15-HDYA experiment doesn't provide information about the number of sites being modified. Furthermore, 15-HDYA does not provide information on whether other fatty acids can be attached to the Spike protein. We sought to investigate this using complementary approaches. First, cells ectopically expressing Spike-C9 and 3xHA-ZDHHC5 were processed using the APE assay as in Figure 1C . The increased expression of ZDHHC5 resulted in a ~3-fold increase in S-acylation as detected by APE, along with the appearance of a 6 th band at a higher molecular weight (Fig 4C, D) .

Unlike the metabolic labelling with clickable fatty acids, the APE assay detects sites of Sacylation regardless of the source of the acyl chains. Next, we considered whether the Spike protein was modified by acyl chains other than palmitate and whether over-expression of the ZDHHC5 enhanced this modification. To investigate this possibility, cells were metabolically labelled with BSA-conjugated clickable fatty acid analogs of palmitate, stearate and myristate, followed by immunocapture and processing as in Figure 1B . As depicted in Fig 4E, F, the myristate analog (13-TDYA) and stearate analog (17-ODYA) are covalently attached to the Spike-C9 but less abundant. Curiously we also found that over-expression of ZDHHC5 had a modest impact on the labelling of Spike by 15-HDYA compared to control cells ( Fig S3) . We hypothesize that this is likely because either the uptake from the medium and synthesis of the 15-HDYA-CoA are limiting steps, and in the presence of de novo synthesized palmitoyl-CoA, it is inefficiently used to S-acylate proteins. Indeed when we repeat the experiment in the presence of the fatty acid synthase inhibitor TVB-3166, we see more efficient incorporation of the clickable label ( Fig S3) .

As part of the reaction cycle, ZDHHC enzymes are first auto-acylated with the acyl chains and subsequently transferred to the substrates (36). To our knowledge, no information is available on the fatty acyl specificity of ZDHHC5. Thus, taking advantage of the clickable fatty acids and protocol used in Fig 4E, we sought to determine which fatty acids are attached to ZDHHC5 as part of the reaction cycle. As shown in Fig 4G and H, ZDHHC5 displays a strong preference for 16-carbon acyl chains even compared to the 14-and 18-carbon chains. The results suggest that ZDHHC5 is responsible for a substantial amount of the palmitate attached to the Spike protein, but that other ZDHHC enzymes attach other fatty acyl chains to the Spike protein.

Next, we sought to investigate if our observations on the Spike protein and ZDHHC5 can be extended to another human coronavirus, namely CoV-229E, one of the viruses that cause the common cold. Although the Spike protein of CoV-229E recognizes a different host protein, CD13 (37), its cytosolic tail is similar to that of SARS-CoV2 (Fig. 1A) . We again generated a plasmid-borne C9 tagged Spike protein of the 229E virus (229E Spike-C9). Using the APE assay, we found that the 229E Spike-C9 protein is S-acylated on multiple sites when ectopically expressed in HEK293T cells. Curiously, the pattern obtained was different from that of the SARS-CoV2 Spike. First, we found that all the 229E Spike-C9 had at least one PEG attached ( Fig 5A) and that we could not resolve the individual bands. This raises the possibility that the 229E Spike is more heavily S-acylated than the SARS-CoV2 Spike. Again, the intensity of the PEGylated proteins is enhanced when HA-ZDHHC5 levels are increased (Fig. 5A, B) . This finding is consistent with the Spike proteins of SARS-CoV2 and likely 229E being substrates of ZDHHC5. To our knowledge, the role of other ZDHHC enzymes and the 229E Spike has not been investigated. Our results do not rule out the possibility that other ZDHHC enzymes may also modify the 229E Spike protein.

To investigate the role of ZDHHC5 in the generation of virulent progeny, we used two complementary cell-based assays. First, we used a previously described siRNA to silence ZDHHC5 in confluent monolayers of human MRC-5 fibroblasts (Fig. 5C, D) (38) . Next, we used these conditions and assessed the ability of CoV-229E to form plaques (Fig. 5E, F) . Cell monolayers transfected with a non-targeting siControl produced an average of ~10 6 PFU/ml, and monolayers transfected with the siZDHHC5 displayed significantly fewer plaques. From the plaque assay, it was unclear if the loss of ZDHHC5 impacted the initial infection or the subsequent spread to the neighbouring cells in the monolayer. We used a liquid culture assay to complement the plaque assay to investigate infection and dissemination. In this assay, cells were incubated in the presence of 229E virus at a low multiplicity of infection (MOI = 0.005) for 2 hours, followed by extensive washing. After 72 hours, cells were processed and stained with an antibody directed against the 229E Spike protein (39) and an AlexaFluor 488-conjugated secondary antibody. As shown in Fig. 5G and H, ≈50% of the siControl cells were infected as determined by immunofluorescence detection of the Spike protein. In contrast, only ≈15% of the ZDHHC5 silenced cells were positive for the Spike protein. These results are consistent with the notion that ZDHHC5 is required for the generation of virulent progeny and spread in vitro.

Based on our data, we sought to determine if ZDHHC5 and other ZDHHC enzymes could represent suitable therapeutic targets for treating CoV infections. Unfortunately, no specific inhibitor of ZDHHC5 exists. Previously, we found that the fatty acid synthase (FASN) inhibitor cerulenin prevented the S-palmitoylation of the ZDHHC5 substrates NOD1 and NOD2 (40) . This is logical since FASN produces a cytosolic pool of palmitoyl-CoA, the palmitate donor for ZDHHCs. However, cerulenin is not suitable therapeutically due to severe weight loss and offtarget effects in murine models (41) (42) (43) . Fortunately, first-in-class FASN inhibitors have been developed by Sagimet Biosciences, formerly 3-V Biosciences (San Mateo, USA). One of these inhibitors, TVB-2640, is orally available, well-tolerated, and a highly potent FASN inhibitor in clinical trials for cancer (44) and non-alcoholic fatty liver disease (45) . A commercially available related compound, TVB-3166, previously showed an IC50 in vitro of 42 nM and a cellular IC50 of 60 nM (46, 47) .

Since we wanted to achieve complete inhibition of FASN, we used higher concentrations of TVB-3166. Treating cells with 0.2 μM or 20 μM TVB-3166 did not impact cell viability, nor did it alter the expression of Spike-C9 in HEK293T cells (Fig 6A, B) . However, treating HEK293T cells expressing Spike-C9 with TVB-3166 or 2-bromopalmitate (2BP), a non-specific inhibitor of ZDHHC enzymes and other enzymes involved in fatty acid metabolism (48) , attenuated the Sacylation of the Spike protein as determined by the APE assay ( Fig. 6 C, D) . Additionally, we examined whether TVB-3166 is a direct inhibitor of ZDHHC5 by assessing the covalent attachment of 15-HDYA on the Spike protein in the presence of TVB-3166. Consistent with Fig   S3, we find that the exogenously added clickable fatty acid is incorporated more efficiently in the presence of the FASN inhibitor ( Fig 6E, F) , suggesting that there is competition between the endogenous acyl-CoAs and exogenous acyl chains. This also brings into question the mode of action of 2BP and whether it truly needs to be converted to 2BP-CoA to inhibit ZDHHC enzymes. Indeed, a previous study demonstrated that both 2BP and 2BP-CoA could directly modify Cys residues of ZDHHC enzymes and that 2BP can also irreversibly block Cys residues directly of target proteins such as Spike in our study (48) . This further demonstrates the need for a better pan-inhibitor of ZDHHC enzymes.

To evaluate the effect of TVB-3166 on the viral spread, HEK293 A2T2 cells were infected with SARS-CoV-2 strain SB3 (49) at a MOI=0.1 and then treated with TVB-3166 6 hrs postinfection for an additional 18 hours. Then, cells were fixed, permeabilized, stained for the SARS-CoV-2 nucleocapsid protein, and imaged with confocal microscopy. Notably, a ~70% decrease in infection was observed with TVB-3166 treatment (Fig 6 G) . These experiments collectively suggest that inhibiting de novo lipid synthesis is an efficient way of attenuating S-acylation and that this strategy effectively blocks the in vitro spread of the SARS-CoV2.

We confirmed that TVB-3166 treatment also inhibited the S-acylation of the 229E Spike-C9 protein. HEK293T cells were transiently transfected with 229E Spike-C9 and treated with TVB-3166; cell lysates were then subject to the ABE assay and revealed a decrease in the higher molecular weight species, thus poly-acylation (Fig. 7A, B) . We next measured the ability of 

We and others have now established that the Spike protein of SARS-CoV2 is S-acylated.

This modification supports its ability to catalyze membrane fusion and thus the infectivity of the SARS-CoV2 virus and experimentally useful pseudovirus (15, 18, 20) . Furthermore, these observations are similar to previous findings on MHV and SARS Spike proteins (21, 23) , demonstrating the importance of post-translational lipidation for the function of these fusogenic proteins. Collectively, various studies have now found that nearly half of the ZDHHC enzymes have the potential to S-acylate the Spike protein. Yet, there are no ZDHHC inhibitors, broad spectrum or selective, in clinical development. However, the findings that FASN inhibitors effectively block S-acylation of SARS-CoV2 Spike protein, attenuating the spread of the virus and even inhibiting the spread of other respiratory viruses, including respiratory syncytial virus, human parainfluenza 3 and rhinovirus (52) make this an area ripe for future investigations.

Mechanistically, the techniques that require cleavage of the thioester bond to detect sites of S-acylation, including acyl-biotin exchange, acyl-PEG exchange, and acyl-RAC, do not provide information as to the acyl chain species attached to the Cys residue(s). Using three alkyne-containing fatty acid analogs, we find that analogs of palmitate, myristate and stearate are attached to the Spike protein with a preference for the 16-carbon analog. We also found that ZDHHC5 has a strong preference for palmitoyl-CoA as a substrate and that when expressed in ZDHHC5 knockouts, the incorporation of palmitate was reduced by 40%. The recent studies on the S-acylation of the Spike protein have relied on a variety of cell lines and techniques to examine post-translational lipidation. Therefore, the choice of cell types and different techniques may be contributing to the heterogeneity of the collective results. Alternatively, with its ten Cys residues, it's also conceivable that the Spike protein is a promiscuous substrate for ZDHHC enzymes.

Considering the Spike protein contains ten Cys residues within the C-terminal domain and proximal to the transmembrane domain, it is worth considering the critical modification sites. A consensus from the various studies is that Cys1235 and 1236 are crucial for the infectivity of both the virus and pseudovirus. Additionally, a trend from these studies is that Cys1235, 1236, 1240 and 1241 are sites most highly modified by exogenously added 3 Hpalmitate or the clickable stearate analog 17-ODYA. Additionally, the results suggest that Cys1235 and 1236 are S-acylated in the endoplasmic reticulum by the ZDHHC20 before transiting to the Golgi apparatus and plasma membrane, where it can encounter additional ZDHHC isoforms. However, these studies rely on the uptake of fatty acids from the medium.

The growing body of literature has demonstrated FASN, and not exogenous fatty acids, is the primary source of acyl chains for S-acylation (40, 46, (53) (54) (55) (56) . Using an alternative approach, Zeng et al. replaced all ten Cys with Alanine residues and subsequently re-introduced individual Cys residues within the tail and examined the S-acylation using the acyl-biotin exchange. This study found that the individual Cys1235, 1236, 1240 and 1241 residues were not being S-acylated in isolation, yet the other six sites were being modified (18) . The significance of this finding is currently unclear. Still, it suggests that the recognition of the sites of S-acylation by the ZDHHC is more complex than just their proximity to the transmembrane domain. Another possibility is that thioesterases may more readily recognize some sites, especially in the absence of the other neighbouring S-acylated residue, so their steady-state S-acylation may appear lower. Indeed, how a de-acylation/re-acylation cycle may impact the findings of all these studies should be considered in the future.

This study, our recent studies on the peptidoglycan sensors NOD1 and NOD2 (40) , and various other studies have demonstrated that the inhibition of FASN greatly attenuates the Sacylation of proteins even in the presence of exogenous lipids and lipoproteins (53, 56, 57) . This observation is somewhat unexpected, although, to our knowledge, the source of fatty acids/fatty acyl-CoAs has not been rigorously examined. Treating cells with triacsin C a molecule known to block fatty acid uptake, does limit the uptake and incorporation of radiolabeled palmitate or clickable fatty acids from the medium (58,59) as well as prevent the formation of lipid droplets.

However, these studies did not investigate whether the proteins were still S-acylated by an independent method like the acyl-biotin exchange. Structurally unrelated FASN inhibitors, including the TVB compounds and cerulenin/C75, have comparable results; thus, this is unlikely the result of off-target effects. On the contrary, as we show in Fig 6F Fig S3, in the presence of TVB-3166, we see a tendency towards greater incorporation of the exogenously added clickable ligand.

One possible explanation is that exogenously added fatty acids are not internalized and converted to acyl-CoAs efficiently enough to supply the needs of the ZDHHC enzymes. Another possibility is that different acyl-CoA "pools" exist within the cytosol and that these may be channeled for further use and metabolism. Considering acyl-CoAs support phospholipid metabolism, triglyceride synthesis, protein acylation and can be imported into mitochondria and peroxisomes for β-oxidation, it may be difficult to resolve this question. Additionally, considering the concentration of CoA in the mitochondria is nearly two orders of magnitude higher than the cytosol(60), directly analyzing cytosolic CoA pools is not trivial. Our findings that the Spike protein can also be myristoylated and stearoylated (Fig. 4F ), yet FASN inhibition abolished S-acylation using the APE assay (Fig. 6C) , suggests that the FASN is also critical to generating the CoA species used for these reactions. In support of this notion, two recent studies have found that FASN generates an array of fatty acyl chains than just 16:0 and that FASN supports the generation of myristoyl-CoA and stearoyl-CoA and their subsequent attachment to proteins (61, 62) .

Does the loss of ZDHHC5 and FASN activity impact the SARS-CoV2 infection cycle by other mechanisms? In all likelihood, yes. First, S-acylation may be important for the function of additional viral proteins. For comparison, the Envelope protein of MHV also requires S-acylation to assemble virions (63) . Although we did not pick up other SARS-CoV-2 with the Acyl-RAC mass spectrometry analysis, it does not rule out the possibility that other viral proteins are, in fact, S-acylated. Indeed, in a previous study, we found it difficult to detect many of the SARS-CoV2 proteins, including those that could be potentially S-acylated (64) . Second, loss of ZDHHC5 will alter S-acylated proteins at the cell surface, including flotillin, and is also known to increase the rates of two actin-dependent endocytic processes, phagocytosis and macropinocytosis, through an unknown mechanism (38, 40) . Considering SARS-CoV2 is larger than the traditional clathrin-mediated endocytic cargo, other factors such as flotillin or actin machinery may play a role and thus be regulated by ZDHHC5. Additionally, since many proteins involved in vesicular transport and membrane fusion are S-acylated (65), including SNAP23 and SNARE proteins, treatment with FASN inhibitors would also impede these processes and thus attenuate the viral replication cycle. Indeed, many small molecules that interfere with phosphoinositides in the endocytic pathway and lysosomal pH are also known to disrupt the viral replication cycle (66) (67) (68) .

Alterations in lipid metabolism are documented in CoV infections. Sera from patients with COVID19 have altered apolipoproteins and lipid levels (69), while 229E infected cells have been shown to have increased levels of free fatty acids (69) . Given the fact that CoVs are enveloped viruses, they will require host lipids for replication. Indeed, CoV-induced remodeling of the cellular lipidome is necessary for robust viral replication. Specifically, recent studies have demonstrated that FASN activity is required to support SARS-CoV2 replication and that blocking fatty acid absorption using Orlistat reduced illness and symptoms in mice (67, 70) .

Additional experiments have demonstrated that the diacylglycerol acyltransferase 1 inhibitor (A922500) attenuates the production of viral progeny (71). This enzyme is critical for lipid droplet formation, and the same study found that SARS-CoV2 replications centers are in proximity to lipid droplets. Finally, the ectodomain of SARS-CoV2 Spike contains a binding pocket for another fatty acid, linoleic acid (72) times, and proteins were eluted with SDS Laemmli buffer and heated for 30min at 37°C before

Immunofluorescence: Cells were seeded on an 18mm-round coverslip in a 12-well plate. Cells Fluor 488 secondary antibody (Jackson ImmunoResearch) and DAPI (1 mg/mL) for 1h at room temperature, followed by mounting on slides with DAKO mounting media. Images were acquired with the WaveFX system. The total fluorescent signal was quantified with ImageJ. Spinning-disk confocal microscope. Images were processed using the Volocity Viewer v.6. Control (NaCl-treated) and hydroxylamine-treated replicates were compressed to 2 and a BFDR cut-off of 1% was used. 

Confocal images of mCherry and GFP expressing co-culture to assess syncytium formation.

HEK293T cells expressing mCherry and empty vector, Spike-C9 or Spike C->S -C9 were plated with HEK293T cells expressing soluble GFP and the ACE2 receptor. Representative two-color merged micrographs demonstrate that wild-type Spike and ACE2 co-culture form syncytium 

2020) A novel coronavirus outbreak of global health concern

The emergence of SARS, MERS and novel SARS-2 coronaviruses in the 21st century

Tissue and cellular tropism of the coronavirus associated with severe acute respiratory syndrome: an in-situ hybridization study of fatal cases

Coronavirus pathogenesis

Inhibition of SARS-CoV-2 Infections in Engineered Human Tissues Using Clinical-Grade Soluble Human ACE2

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

COVID-19 vaccines: breaking record times to first-in-human trials

A Multibasic Cleavage Site in the Spike Protein of SARS-CoV-2 Is Essential for Infection of Human Lung Cells

Rates of evolutionary change in viruses: patterns and determinants

Why are RNA virus mutation rates so damn high?

Origin and evolution of pathogenic coronaviruses

Natural mutations in the receptor binding domain of spike glycoprotein determine the reactivity of cross-neutralization between palm civet coronavirus and severe acute respiratory syndrome coronavirus

SARS-CoV-2 protein interaction map reveals targets for drug repurposing

2021) S-acylation of SARS-CoV-2 Spike Protein: Mechanistic Dissection, In Vitro Reconstitution and Role in Viral Infectivity

2021) Palmitoylation of SARS-CoV-2 S protein is essential for viral infectivity

2021) S-acylation controls SARS-Cov-2 membrane lipid organization and enhances infectivity

2021) S-acylation controls SARS-CoV-2 membrane lipid organization and enhances infectivity

The interactions of ZDHHC5/GOLGA7 with SARS-CoV-2 spike (S) protein and their effects on S protein's subcellular localization, palmitoylation and pseudovirus entry

2022) Palmitoylation of SARS-CoV-2 S protein is critical for S-mediated syncytia formation and virus entry

Identification of SARS-CoV-2 Spike Palmitoylation Inhibitors That Results in Release of Attenuated Virus with Reduced Infectivity

Palmitoylations on murine coronavirus spike proteins are essential for virion assembly and infectivity

Role of spike protein endodomains in regulating coronavirus entry

Palmitoylation of the cysteine-rich endodomain of the SARScoronavirus spike glycoprotein is important for spike-mediated cell fusion

1D4: a versatile epitope tag for the purification and characterization of expressed membrane and soluble proteins

Structural basis of receptor recognition by SARS-CoV-2

Characterization of spike glycoprotein of SARS-CoV-2 on virus entry and its immune cross-reactivity with SARS-CoV

Alkyne lipids as substrates for click chemistry-based in vitro enzymatic assays

Peptidotriazoles on solid phase: [1,2,3]-triazoles by regiospecific copper(i)-catalyzed 1,3-dipolar cycloadditions of terminal alkynes to azides

A stepwise huisgen cycloaddition process: copper(I)-catalyzed regioselective "ligation" of azides and terminal alkynes

Robust fluorescent detection of protein fatty-acylation with chemical reporters

Mass-tag labeling reveals site-specific and endogenous levels of protein S-fatty acylation

Site-specific analysis of protein S-acylation by resin-assisted capture

A simple protein-based surrogate neutralization assay for SARS-CoV-2

The coronavirus spike protein is a class I virus fusion protein: structural and functional characterization of the fusion core complex

Sequences in the cytoplasmic tail of SARS-CoV-2 Spike facilitate expression at the cell surface and syncytia formation

DHHC protein S-acyltransferases use similar ping-pong kinetic mechanisms but display different acyl-CoA specificities

Human coronavirus 229E binds to CD13 in rafts and enters the cell through caveolae

Targeted enhancement of flotillin-dependent endocytosis augments cellular uptake and impact of cytotoxic drugs

The human coronavirus HCoV-229E Sprotein structure and receptor binding

Palmitoylation of NOD1 and NOD2 is required for bacterial sensing

Reduced food intake and body weight in mice treated with fatty acid synthase inhibitors

Inhibitors of lipogenic enzymes as a potential therapy against cancer

Fatty acid synthase as a potential therapeutic target in cancer

Recent Advances in the Development of Fatty Acid Synthase Inhibitors as Anticancer Agents

Fatty Acid Synthase Inhibitor TVB-2640 Reduces Hepatic de Novo Lipogenesis in Males With Metabolic Abnormalities

FASN Inhibition and Taxane Treatment Combine to Enhance Anti-tumor Efficacy in Diverse Xenograft Tumor Models through Disruption of Tubulin Palmitoylation and Microtubule Organization and FASN Inhibition-Mediated Effects on Oncogenic Signaling and Gene Expression

Inhibition of de novo Palmitate Synthesis by Fatty Acid Synthase Induces Apoptosis in Tumor Cells by Remodeling Cell Membranes, Inhibiting Signaling Pathways, and Reprogramming Gene Expression

Profiling targets of the irreversible palmitoylation inhibitor 2-bromopalmitate

FASN Inhibitor) for the Treatment of Nonalcoholic Steatohepatitis: FASCINATE-1, a Randomized, Placebo-Controlled Phase 2a Trial

First-in-human study of the safety, pharmacokinetics, and pharmacodynamics of first-in-class fatty acid synthase inhibitor TVB-2640 alone and with a taxane in advanced tumors

Direct Inhibition of Cellular Fatty Acid Synthase Impairs Replication of Respiratory Syncytial Virus and Other Respiratory Viruses

Toll-like receptor mediated inflammation requires FASNdependent MYD88 palmitoylation

Inhibitory effects of cerulenin on protein palmitoylation and insulin internalization in rat adipocytes

Metformin alleviates inflammation through suppressing FASN-dependent palmitoylation of Akt

De novo lipogenesis maintains vascular homeostasis through endothelial nitric-oxide synthase (eNOS) palmitoylation

Fatty acid synthase modulates intestinal barrier function through palmitoylation of mucin 2

Inhibiting long chain fatty Acyl CoA synthetase increases basal and agonist-stimulated NO synthesis in endothelium

Coenzyme A: back in action

A bioorthogonal chemical reporter for fatty acid synthasedependent protein acylation

A Simple and Direct Assay for Monitoring Fatty Acid Synthase Activity and Product-Specificity by High-Resolution Mass Spectrometry

Envelope protein palmitoylations are crucial for murine coronavirus assembly

A SARS-CoV-2 Peptide Spectral Library Enables Rapid, Sensitive Identification of Virus Peptides in Complex Biological Samples

Examining the Underappreciated Role of S-Acylated Proteins as Critical Regulators of Phagocytosis and Phagosome Maturation in Macrophages

Inhibition of PIKfyve kinase prevents infection by Zaire ebolavirus and SARS-CoV-2

2021) Inhibitors of VPS34 and fattyacid metabolism suppress SARS-CoV-2 replication

On-target versus off-target effects of drugs inhibiting the replication of SARS-CoV-2

2021) Pharmacological inhibition of fatty acid synthesis blocks SARS-CoV-2 replication

Lipid droplets fuel SARS-CoV-2 replication and production of inflammatory mediators

Free fatty acid binding pocket in the locked structure of SARS-CoV-2 spike protein

Molecular Pathways: Fatty Acid Synthase

Touchdown PCR for increased specificity and sensitivity in PCR amplification

HTCC: Broad Range Inhibitor of Coronavirus Entry

Viral concentration determination through plaque assays: using traditional and novel overlay systems

Murine hepatitis virus strain 1 produces a clinically relevant model of severe acute respiratory syndrome in A/J mice

The Tie2-agonist Vasculotide rescues mice from influenza virus infection

SAINT: probabilistic scoring of affinity purification-mass spectrometry data

10 fields of view per condition for n=3 biological replicates were analyzed

Inhibition of fatty acid synthase abolishes S-acylation of Spike and attenuates SARS-CoV2 spread in vitro. (A) Immunoblot of ectopically expressed Spike-C9 in HEK293T cells treated with DMSO, 0.2 μM or 20 μM TVB-3166 for 16-18 hrs. (B) Quantification of panel 'A'. Data are the mean ± SEM n= 3 biological replicates normalized to GAPDH and analyzed using an unpaired t-test with Welch's correction: 0.2 μM P=0

Acyl-PEG exchange assay of Spike-C9 ectopically expressed in HEK293T treated with 0.2 μM or 20 μM TVB-3166 or 50 μM 2-Bromohexadecanoic acid (2-BP) for 16-18hrs

C' where PEG-modified bands were divided by the unmodified bands at ~180 kDa, multiple unpaired t-tests with Welch's correction, *** p=0.0001, ** p=0.0088 relative to DMSO control. Data are mean ± SEM of n= 3 biological replicates. (E) In-gel fluorescence (left) and immunoblotting (right) of Spike-C9 incubated with 15-HDYA following the treatment of DMSO or 20 μM TVB-3166 for 18h

H) Representative images and quantification of HEK293T A2T2 cells (stably expressing Ace2 and TMPRSS2) infected with SARS-CoV2 (strain SB3, MOI of 0.1), treated with either 0.1 μM or 1 μM TVB-3166 6hrs post-infection for 18 hours

Data are the mean ± SEM from n = 3 replicates with all microscope fields plotted as individual points. Data were analyzed using a one-way ANOVA, **** P<0vitro and in vivo. (A) Acyl-PEG exchange assay and quantification of 229E

Analysis of acyl-PEG assay where the PEG-modified bands normalized to unmodified bands at ~180 kDa, unpaired t-test with Welch's correction; data are mean ± SEM of n=3 biological replicates, **** p<0.0001. (C, D) Plaque assay of MRC-5 cells infected with 229E and treated with 0.2 μM TVB-3166 for 4 days

Days post-infection

The patent for TVB-3166, TVB-2640 and related compounds belong to Sagimet Biosciences (San Mateo, California). The authors receive no financial compensation or support from Sagimet Biosciences. The authors do not hold stock or interest in Sagimet Biosciences. The authors declare no competing interests.

In Gel Fluorescence Excitation of Cy5.5: 700nmIn Gel Fluorescence Excitation of Cy5.5: 700nm 

In Gel Fluorescence Excitation of Cy5.