key: cord-0856307-eju0kqlg
authors: Zhang, Liang; Wang, Tao; Yi, Yanyan; Song, Mengzhao; Jin, Mingxing; Guo, Kangkang; Zhang, Yanming
title: ARF1 with Sec7 Domain-Dependent GBF1 Activates Coatomer Protein I To Support Classical Swine Fever Virus Entry
date: 2022-03-23
journal: Journal of virology
DOI: 10.1128/jvi.02193-21
sha: 6e3faa9c83554b6ad79e793eb2e918c1729be392
doc_id: 856307
cord_uid: eju0kqlg

Classical swine fever virus (CSFV), a positive-sense, enveloped RNA virus that belongs to the Flaviviridae family, hijacks cell host proteins for its own replication. We previously demonstrated that Golgi-specific brefeldin A (BFA) resistance factor 1 (GBF1), a regulator of intracellular transport, mediates CSFV infection. However, the molecular mechanism by which this protein regulates CSFV proliferation remains unelucidated. In this study, we constructed a series of plasmids expressing GBF1 truncation mutants to investigate their behavior during CSFV infection and found that GBF1 truncation mutants containing the Sec7 domain could rescue CSFV replication in BFA- and GCA (golgicide A)-treated swine umbilical vein endothelial cells (SUVECs), demonstrating that the effect of GBF1 on CSFV infection depended on the activity of guanine nucleotide exchange factor (GEF). Additionally, it was found that ADP ribosylation factors (ARFs), which are known to be activated by the Sec7 domain of GBF1, also regulated CSFV proliferation. Furthermore, we demonstrated that ARF1 is more important for CSFV infection than other ARF members with Sec7 domain dependence. Subsequent experiments established the function of coatomer protein I (COP I), a downstream effector of ARF1 which is also required for CSFV infection by mediating CSFV invasion. Mechanistically, inhibition of COP I function impaired CSFV invasion by inhibiting cholesterol transport to the plasma membrane and regulating virion transport from early to late endosomes. Collectively, our results suggest that ARF1, with domain-dependent GBF1 Sec7, activates COP I to facilitate CSFV entry into SUVECs. IMPORTANCE Classical swine fever (CSF), a highly contact-infectious disease caused by classical swine fever virus (CSFV) infecting domestic pigs or wild boars, has caused huge economic losses to the pig industry. Our previous studies have revealed that GBF1 and class I and II ARFs are required for CSFV proliferation. However, a direct functional link between GBF1, ARF1, and COP I and the mechanism of the GBF1-ARF1-COP I complex in CSFV infection are still poorly understood. Here, our data support a model in which COP I supports CSFV entry into SUVECs in two different ways, depending on the GBF1-ARF1 function. On the one hand, the GBF1-ARF1-COP I complex mediates cholesterol trafficking to the plasma membrane to support CSFV entry. On the other hand, the GBF1-ARF1-COP I complex mediates CSFV transport from early to late endosomes during the entry steps.

FLAG-tagged expression plasmid was created, and a series of expression plasmids for the GBF1 truncation mutants was also generated: the GBF1-(1-885)-FLAG plasmid with the C terminus removed, the GBF1-(1-698)-FLAG plasmid with the C terminus and Sec7 domain removed, and the GBF1-Sec7-FLAG plasmid with the C terminus and N terminus removed (Fig. 1A) . These plasmids were transfected into SUVECs, and the expression of FLAG-tagged wild-type and mutant GBF1 proteins was confirmed by a Western blot assay (Fig. 1B) . In addition, we also assessed the cytotoxicity of two GBF1 inhibitors, namely, BFA and GCA (golgicide A). The results of the cell viability assay showed that 100 nmol/L BFA and 300 nmol/L GCA demonstrated no obvious toxicity to the cells (Fig. 1C) . Second, SUVECs were transfected with the indicated GBF1 plasmid [GBF1-FLAG, GBF1-(1-885)-FLAG, GBF1-(1-698)-FLAG, and GBF1-Sec7-FLAG] for 48 h, and cells were cotreated with CSFV (multiplicity of infection [MOI] of 1) and BFA (100 nmol/L). After 48 h, the cells were collected for viral genome copy number quantification. Our results showed that CSFV RNA copy numbers were significantly decreased in BFA-treated cells compared to those in CSFV-infected cells (P , 0.05) (Fig. 1D) , and viral genome copy numbers were partially rescued in GBF1-FLAG-, GBF1-(1-885)-FLAG-, and GBF1-Sec7-FLAG-transfected cells (P , 0.05) (Fig. 1D ). Unlike Sec7 domain-containing plasmids, GBF1-(1-698)-FLAG plasmid transfection did not rescue CSFV replication (P . 0.05) (Fig.  1D ). To verify the role of the GBF1 Sec7 domain in CSFV replication, we further examined the effect of GCA, a specific inhibitor targeting the GBF1 Sec7 domain, on CSFV replication. SUVECs were treated as described above, and similar results were obtained. GCA potently suppressed CSFV replication (P , 0.05); however, the Sec7 domain-containing plasmid significantly rescued GCA-induced CSFV replication inhibition (P , 0.05) (Fig.  1E ). Taken together, our results reveal that the Sec7 domain of GBF1 is a key factor in CSFV proliferation.

ARF1 was involved in CSFV infection in a Sec7 domain-dependent manner. GBF1 is a BFA-sensitive GEF for G proteins from the ARF family. The Sec7 domain has GEF activity, which activates class I and II ARFs (5, 18) . Since the Sec7 domain of GBF1 is involved in the replication of CSFV in SUVECs, the function of Sec7 domain effector ARFs (class I and class II) in CSFV infection is also worthy of attention. To determine whether the Sec7 domain promoted CSFV infection through the function of ARFs, we cloned the ARF1 to ARF5 genes into the pEGFP-N1 vector and verified their expression in transfected cells using Western blotting (data not shown). Following the characterization of the expression of ARFs in cells, the subcellular distribution of each of the ARFs was demonstrated in transfected cells. Using confocal microscopy, we observed that class I (ARF1, ARF2, and ARF3) and class II (ARF4 and ARF5) ARFs were located in the Golgi complex but not in the ER or mitochondria, indicating their mediator function in Golgi association trafficking ( Fig. 2A) . Furthermore, we screened which porcine ARF function is Sec7 domain dependent. SUVECs were transfected with the indicated plasmids and treated with or without BFA for 2 h. The subcellular distribution of different ARFs was disrupted with 100 nmol/L BFA as shown by using confocal microscopy (Fig. 2B ). In contrast, pEGFP-ARF1 to pEGFP-ARF5, but not pEGFP, clustered around the nucleus in untreated cells (Fig. 2B ). This result revealed that the porcine ARF1 to ARF5 proteins were sensitive to BFA, indicating that the functions of all ARFs were Sec7 domain dependent. Considering that all ARF functions were Sec7 domain dependent in SUVECs, we speculated that the activation of at least one ARF family member is required for CSFV replication. To this end, SUVECs were transfected with pEGFP-ARF1, pEGFP-ARF2, pEGFP-ARF3, pEGFP-ARF4, or pEGFP-ARF5 for 48 h and then cotreated with CSFV, BFA, or GCA for 48 h. We observed that ARF1 markedly rescued CSFV replication in Sec7 domain activity-defective cells (P , 0.05) ( Fig. 2C and D) . Together, these data suggest that ARF1 plays a major GBF1 Sec7-dependent role in CSFV replication.

ARF1 positively regulated CSFV propagation. To investigate the involvement of ARF1 in regulating CSFV infection, we examined the effect of Methyl 2-(4-fluorobenzamido)benzoate Exo-1, a specific inhibitor of ARF1, on CSFV propagation. A cell counting kit 8 (CCK-8) assay revealed that the safe concentration of Exo-1 for SUVECs was 5 mmol/L (Fig. 3A) . We also observed that Exo-1, at a concentration of 5 mmol/L, caused Golgi structure fragments as seen under a confocal microscope (Fig. 3B) , indicating that Exo-1 at 5 mmol/L could disrupt the function of ARF1. Using reverse transcriptionquantitative PCR (RT-qPCR) and 50% tissue culture infectious doses (TCID 50 ) per milliliter, we observed a dose-and time-dependent restriction of CSFV proliferation by Exo-1 (P , 0.05) (Fig. 3C to F). Next, stably ARF1-overexpressing (CMV-ARF1) cell lines were obtained by using recombinant lentiviruses and were utilized to functionally validate the correlation between ARF1 and CSFV infection. Our results showed that ARF1 overexpression promoted CSFV proliferation (P , 0.05) (Fig. 3G to I).

In light of the findings of the role of ARF1 in CSFV infection, we investigated whether viral production was also mediated by ARF1-T31N and ARF1-Q71L. The ARF1-T31N mutant mimics GDP-bound states, an inactive form of ARF1. ARF1-Q71I mimics GTP-bound states, an active form of ARF1. To this end, pEGFP-ARF1-T31N-and pEGFP-ARF1-Q71Ltransfected cells were infected with CSFV (MOI of 1). After 24, 48, and 72 h of infection, as depicted in Fig. 3J and K, viral genome copy numbers and infectious viral proliferation were decreased in cells expressing pEGFP-ARF1-T31N or pEGFP-ARF1-Q71L (P , 0.05).

Moreover, CSFV proliferation in pEGFP-ARF1-T31N-or pEGFP-ARF1-Q71L-transfected cells was verified using an indirect immunofluorescence assay (IFA) 24 h after CSFV infection. The ARF1-T31N and ARF1-Q71L protein expression levels in transfected cells were determined using Western blotting at the indicated times (Fig. 3L ). The CSFV plaques in pEGFP-ARF1-T31N-or pEGFP-ARF1-Q71L-transfected cells were smaller and less abundant than those in the control cells (Fig. 3M) . These results strongly support the positive role of ARF1 in CSFV replication.

COP I depletion inhibited CSFV invasion. COP I is a major substrate of ARF1, which coordinates cell component trafficking through vesicles in the early secretory pathway. Previous studies have demonstrated that COP I, ARF1, and GBF1 act as a complex to mediate cellular functions (19) . Given that ARF1 functions rely on the GBF1 Sec7 domain to participate in CSFV proliferation, we investigated whether COP I depends on the GBF1-ARF1 pathway to mediate CSFV infection. We utilized the CRISPR/Cas9 gene-editing system to generate COP I subunit knockout (KO) cell lines, designated COPa-KO and COP« -KO. COPa and COP« knockouts were verified using Western blotting (Fig. 4A) . Next, the COPa-KO and COP« -KO cell lines were infected with CSFV at an MOI of 1; cells treated with BFA (100 nmol/L) and GCA (300 nmol/L), which are GBF1 inhibitors, served as controls. After 24, 48, and 72 h of infection, cells were collected for viral genome copy number quantification, whereas the supernatants were collected for viral titration. The results showed that the genetic depletion of COP I expression resulted in a significant decrease in CSFV replication as well as in BFA (100 nmol/L)-and GCA (300 nmol/L)-treated cells ( Fig. 4B and C) (P , 0.05). Moreover, CSFV proliferation in cell lines was examined using an indirect IFA 24 h after CSFV infection. The CSFV plaques of the COPa-KO and COP« -KO cell lines and BFA (100 nmol/L)-and GCA (300 nmol/L)-treated cells were smaller and less abundant than those of the control cells (Fig. 4D) . Collectively, our findings suggest that COP I knockout inhibits CSFV proliferation.

Given the role of COP I in CSFV infection, we conducted experiments to determine which steps of the CSFV life cycle were COP I mediated. First, COPa-KO and COP« -KO cell lines and BFA (100 nmol/L)-and GCA (300 nmol/L)-pretreated cells were used to determine the effect of COP I on CSFV binding and entry. To determine viral binding, SUVECs were pretreated with 100 nmol/L BFA and 300 nmol/L GCA for 24 h. Next, COPa-KO and COP« -KO cell lines and BFA-and GCA-treated cells were inoculated with CSFV in fetal bovine serum (FBS)-free medium for 1 h at 4°C. Unbound virions were then washed away using a precooled citrate buffer solution (pH 3). Total cells were collected for RNA extraction to determine the CSFV genome copy numbers. The results showed that the CSFV genome copy number was decreased in COPa-KO and COP« -KO cell lines and BFA-and GCA-treated cells, suggesting that COP I was required for CSFV binding (P , 0.5) (Fig. 4E ). To determine viral entry, SUVECs were pretreated with 100 nmol/L BFA and 300 nmol/L GCA for 24 h. Next, COPa-KO and COP« -KO cell lines and BFA-and GCA-treated cells were infected with CSFV in FBS-free medium for 1 h at 4°C to allow virion binding. Cells were then washed with a precooled citrate buffer solution (pH 3) to remove unbound virions and cultured for another 2 h at 37°C. The cells were extensively washed and collected for RNA isolation to determine the CSFV genome copy numbers. The results showed that CSFV genome copy numbers were decreased in COPa-KO and COP« -KO cell lines and BFA-and GCA-treated cells, suggesting that COP I was required for CSFV entry (P , 0.5) (Fig. 4E ). To further clarify the plasmid. Forty-eight hours after transfection, cells were fixed in 4% paraformaldehyde, and immunofluorescence staining was performed with anti-TGN46, ERP57, or TOM20. Cells were also counterstained with DAPI to label nuclei (blue). GFP, green fluorescent protein. (B) SUVECs were seeded onto glass coverslips and transfected with the pEGFP-ARF1, pEGFP-ARF2, pEGFP-ARF3, pEGFP-ARF4, or pEGFP-ARF5 plasmid. Forty-eight hours after transfection, cells were treated with or without BFA for 2 h. Next, cells were fixed in 4% paraformaldehyde and stained with DAPI to label nuclei (blue). (C and D) SUVECs were transfected with the indicated plasmids (pEGFP-ARF1, pEGFP-ARF2, pEGFP-ARF3, pEGFP-ARF4, and pEGFP-ARF5) for 48 h and cotreated with CSFV (MOI of 1) and BFA or GCA for 48 h. Next, cells were collected for RNA extraction with TRIzol lysis, and CSFV RNA copy numbers were measured by RT-qPCR. "*" represents the comparison between the BFA-or GCA-treated cells and the CSFV-infected cells, and "#" represents the comparison between the cells transfected with the indicated plasmids and the BFA-or GCA-treated cells. Exo-1 was measured by CCK-8 assays. (B) SUVECs were seeded onto glass coverslips and treated with or without Exo-1 for 2 h. Next, cells were fixed in 4% paraformaldehyde, and immunofluorescence staining was performed using an anti-TGN46 antibody. Cells were also counterstained with DAPI to label nuclei (blue). (C to F) SUVECs were cotreated with CSFV (MOI of 1) and 5 mmol/L Exo-1 or various concentrations of Exo-1 (5, 2.5, or 1 mmol/L), and cell pellets and culture supernatants were harvested for CSFV RNA copy number and CSFV viral titer determinations by RT-qPCR and TCID 50 result that COPa and COP« knockout could inhibit the invasion of CSFV, COPa and COP« knockout cells were transfected with the recombinant plasmid expressing the corresponding protein, and the entry of CSFV was then detected. The results showed that target protein expression could partially reduce the suppression of CSFV invasion caused by COPa and COP« knockout (P , 0.05) ( Fig. 4F and G). Altogether, these results demonstrate the involvement of COP I in CSFV invasion.

COP I inhibition impaired CSFV invasion by disrupting cholesterol transport to the plasma membrane. Previous studies have demonstrated that ARF1, which activates the assembly of COP I vesicles, mediates cell cholesterol trafficking (19, 20) . Intriguingly, a recent study on CSFV infection of PK-15 cells revealed that plasma membrane cholesterol is required for CSFV invasion (21) . Thus, we speculated that COP I was required for CSFV invasion by mediating cholesterol trafficking. We first ensured that the results showing that methyl-b-cyclodextrin (MbCD), which is primarily used as a cholesterol-depleting reagent, inhibits CSFV entry by impairing plasma membrane cholesterol were not cell specific. SUVECs were pretreated with different concentrations of MbCD for 12 h, and the influence of MbCD on CSFV binding and entry was evaluated as described above. As shown in Fig. 5A and B, we found that MbCD dramatically reduced viral entry in SUVECs in a dose-dependent manner (P , 0.05). Next, we determined whether COP I could regulate cholesterol trafficking by incubating cells with BFA and GCA or using COPa and COP« knockout cell lines. Cells were stained with filipin, which labels unesterified cholesterol, at room temperature for 30 min, and confocal laser microscopy revealed that cholesterol was mainly distributed on the cell membrane in control cells, whereas cholesterol was mainly concentrated in the perinuclear area in BFA-and GCA-treated cells and COPa and COP« knockout cells (Fig. 5C ). Importantly, cholesterol was not present at the plasma membrane after COP I inactivation (Fig. 5D ). To rule out the possibility that COP I depletion may affect the content of cholesterol in the cells, the total cholesterol contents in control cells and COP Idepleted cells were analyzed, and the results showed that COP I depletion did not affect the cell cholesterol content (P . 0.05) (Fig. 5E ). Moreover, the invasion efficiency of CSFV could be partially restored by replenishing cholesterol in COP I-inactivated cells (Fig. 5F ). Together, these results demonstrate that COP I depletion inhibits CSFV invasion by disrupting cholesterol trafficking to the plasma membrane.

COP I depletion inhibited endosomal trafficking upon virus internalization. CSFV enters early endosomes and then fuses to late endosomes to enter porcine kidney 15 (PK-15) cells and porcine alveolar macrophages (PAMs) (22, 23) . Furthermore, previous studies have reported that COP I proteins may be important for endosomal trafficking (4). However, it remains unclear thus far whether COP I vesicles directly mediate the sorting of CSFV between different types of endosomes in the invasion step. Thus, we first used Rab5 and Rab7 small interfering RNAs (siRNAs) to investigate the role of early or late endosomes in CSFV infection. SUVECs were transfected with siRNA targeting Rab5 (siRab5) or siRab7, and the silencing efficiency was determined using Western blotting (P , 0.05) ( Fig. 6A and B) . Next, siRNA-transfected cells were infected with CSFV (MOI of 1), and the results showed that compared to nontargeting siRNA (NTsiRNA)-transfected cells, viral RNA copy numbers and the proliferation of CSFV were significantly decreased in Rab5-and Rab7-silenced cells (P , 0.05) ( Fig. 6C and D).

We examined the expression of Rab5(S34N) and Rab7(T22N) to further confirm the early and late endosomes in CSFV infection. The expression levels of Rab5(S34N) and Rab7(T22N) in transfected SUVECs were investigated using Western blotting ( Cell lines were infected with CSFV (MOI of 1), and SUVECs were cotreated with CSFV (MOI of 1) and an inhibitor (BFA or GCA) for 24 h. Next, cells were fixed in 4% paraformaldehyde and stained with anti-E2 antibody (red). (E) SUVECs were pretreated with BFA (100 nmol/L) or GCA (300 nmol/L) for 24 h. Next, inhibitor-treated cells and COPa and COP« knockout cell lines were infected with CSFV (MOI of 10) for 1 h at 4°C. Next, unbound virions were washed away using a precooled citrate buffer solution (pH 3), and cell pellets were harvested for CSFV RNA copy number determination by RT-qPCR to measure CSFV binding.

(Continued on next page) and F). To examine whether the blocking of Rab-mediated transport affected CSFV infection, SUVECs were transfected with Rab5(S34N) or Rab7(T22N). After transfection for 24 h, cells were infected with CSFV (MOI of 1), and viral genome copy numbers and viral titrations were determined at the indicated time points. We showed that CSFV proliferation was inhibited in Rab5(S34N)-or Rab7(T22N)-transfected cells (P , 0.05) (Fig. 6G and H) . As expected, we also observed that CSFV particles colocalized with Rab5 and Rab7 during invasion (Fig. 6I and Fig. 7J ), suggesting that early/late endosomes are required for CSFV endocytosis and subsequent productive infection.

To investigate whether COP I-depleted cells exhibited any defects in trafficking to early or late endosomes, SUVECs were pretreated with inhibitors of BFA and GCA to disrupt COP I function, and the colocalization of the virus with Rab5 or Rab7 was examined using confocal microscopy after CSFV infection for 2 h. We observed a significant decrease in the number of CSFV particles colocalized with Rab7 ( Fig. 6J) but not with Rab5 (Fig. 6I) , suggesting that virus particles exhibit defective trafficking to late endosomes in COP I-depleted cells. We also revealed that CSFV core (C) and E2 proteins colocalized with the COP I subunit COPb (Fig. 7) , which indicated that COP I vesicles were used for CSFV transport. Together, these findings strongly suggest that COP I vesicles are involved in virion trafficking from early to late endosomes in CSFV entry into SUVECs.

CSFV infection induces GBF1, ARF family, and COP I mRNA expression. Since the GBF1-ARF-COP I complex could regulate the proliferation of CSFV in SUVECs, whether CSFV infection of cells affects the expression of these host proteins is also worthy of attention. SUVECs were inoculated into 12-well plates and infected with CSFV at an MOI of 1, and the GBF1, ARF1 to ARF5, and COP I mRNA expression levels were measured using RT-qPCR. The results showed that CSFV infection significantly promoted GBF1, ARF1 to ARF4, and COP I mRNA expression at each time point (P , 0.05), whereas the effect of CSFV on ARF5 mRNA expression showed inhibition at 6 and 12 h and then promotion at 48 h (Fig. 8A) . Additionally, confocal laser microscopy revealed that the distribution of COP I-coated vesicles was disrupted in CSFVinfected cells (Fig. 8B) . These results indicate that CSFV infection induces the expression of the GBF1-ARF-COP I complex and mediates the distribution of COP I-coated vesicles.

In eukaryotic cells, the intracellular transport system is a complex and precise membrane structure composed of various organelles, cytoskeleton, and intracellular vesicles. The main function of the intracellular transport system is to accurately transfer the extracellular uptake of substances to various organelles or transfer substances synthesized in the ER to the target organelle or outside the cells. These special intracellular transport systems are indispensable for viruses to complete intracellular proliferation, including virus adsorption on the cell surface, invasion, genome replication, assembly, and the secretion of progeny viral particles (24) . Our previous studies have revealed that the Rab1b-GBF1 axis, an important regulator of intracellular trafficking, was required for CSFV proliferation (17) . Our present findings demonstrated that the Sec7 domain of GBF1 could rescue the inhibition of CSFV infection caused by BFA and GCA. This result revealed that the Sec7 domain is required for CSFV infection. The GEF activity of GBF1 is catalyzed by the Sec7 domain, which is responsible for activating class I and II ARFs (25) . Therefore, we speculated that the activation of at least one BFA-or GCA-sensitive ARF family member is required for CSFV replication. Subsequent results

SUVECs were pretreated with different concentrations of BFA (10, 50, and 100 nmol/L) and GCA (100, 200, and 300 nmol/L) for 2 h, and cells were incubated with CSFV (MOI of 10) for 1 h at 4°C. Next, cells were washed with PBS and cultured in fresh medium at 37°C for 2 h, and cell pellets were harvested for CSFV genome copy number determination by RT-qPCR to measure CSFV entry. (F and G) COPa and COP« knockout cell lines were transfected with pcDNA-COPa and pcDNA-COP« . At 48 h, cells were infected with CSFV (MOI of 10), the expression levels of COPa and COP« were measured by Western blotting, and virus entry was determined as described above for panel E. (B) SUVECs were pretreated with various concentrations of MbCD (5, 2.5, or 1 mg/mL) for 12 h and incubated with CSFV for 1 h at 4°C, unbound virions were washed away using a precooled citrate buffer solution (pH 3), and cell pellets were harvested for CSFV genome copy number determination by RT-qPCR to measure CSFV binding. SUVECs were pretreated with 5 mg/mL MbCD for 12 h, and cells were incubated with CSFV for 1 h at 4°C. Next, cells were washed with a precooled citrate buffer solution (pH 3) and cultured in the fresh medium at 37°C for 2 h, and cell pellets were harvested for CSFV genome copy number determination by RT-qPCR to measure CSFV entry. (C and D) Quantification of the cholesterol signal on the membrane relative to the cytosol. BFA (100 nmol/L)-and GCA (300 nmol/L)-treated SUVECs (for 8 h) and COPa and COP« knockout cells were fixed, and cholesterol was labeled by filipin. The intracellular cholesterol distribution was measured by confocal microscopy and quantified by using ImageJ. (E) BFA (100 nmol/L)-and GCA (300 nmol/L)treated SUVECs (for 8 h) and COPa and COP« knockout cells were harvested for total cholesterol determination by a Cholesterol/Cholesterol Ester-Glo assay kit from Promega. (F) BFA (100 nmol/L)-and GCA (300 nmol/L)-treated SUVECs (for 8 h) and COPa and COP« knockout cells were treated with cholesterol for 2 h, and cells were incubated with CSFV for 1 h at 4°C. Next, cells were washed with a precooled citrate buffer solution (pH 3) and cultured in fresh medium at 37°C for 2 h, and cell pellets were harvested for CSFV genome copy number determination by RT-qPCR to measure CSFV entry.

showed that the overexpression of the ARF family could rescue the inhibition of CSFV infection caused by BFA and GCA. Intriguingly, ARF1 showed a better effect than other ARF members, suggesting that ARF1 is a key effector of the Sec7 domain of GBF1 in CSFV infection. We further showed that ARF1 with GBF1 Sec7 domain-dependently activates COP I to support CSFV invasion. Previous studies reported that the GBF1-ARF1-COP I axis was required for the replication of various viruses, including dengue virus (DENV) (13), HCV (26), and chikungunya virus (CHIKV) (27) , which is consistent with our findings. However, Hazara nairovirus (HAZV) also utilizes COP I in an ARF1-independent manner to promote genome replication, possibly because the transport function of COP I-coated vesicles was not required for this step (28) .

In addressing the mechanism of the GBF1-ARF1-COP I function in CSFV invasion, we observed that plasma membrane cholesterol was reduced due to a cholesterol-trafficking defect in cells caused by COP I depletion, and this led to a significant defect in CSFV invasion. We also showed that supplementation with exogenous cholesterol restored CSFV invasion caused by COP I inhibition. Cholesterol is synthesized in the smooth ER and leaves the ER rapidly toward the plasma membrane to regulate the invasion of viruses (29, 30) . There seem to be two modes of cholesterol mechanisms involved in the virus invasion step. One mediates fusion between the virion and endosomes to support viral genome release into the cytosol, as in the cases of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (31), influenza A virus (IAV) (32) , and porcine reproductive and respiratory syndrome virus (PRRSV) (33) , whereas the other, which includes vaccinia virus (VACV) (34), human immunodeficiency virus (HIV) (35) , and West Nile virus (WNV) (36) , functions in a lipid raft-dependent manner and is involved in virus entry. Previously, studies have shown that cholesterol is required for CSFV infection, especially at the invasion step. Liang and colleagues reported that U18666A, which leads to cholesterol accumulation in the endosomal/lysosomal compartment through inhibition of cholesterol delivery to the membrane of late endosomes/lysosomes, inhibits CSFV fusion or uncoating (37) . Yu and colleagues demonstrated that the depletion of plasma membrane cholesterol with methylb-cyclodextrin (MbCD), which is primarily used as a cholesterol-depleting reagent, could cause a failure of CSFV internalization (21) . They also suggested that cholesterol was participating in CSFV invasion, possibly unrelated to lipid rafts (21) . Previous research showed that the depletion of plasma membrane cholesterol by MbCD causes an increase in membrane tension, and high plasma membrane tension correlates with endocytosis inhibition (38, 39) . The depletion of plasma membrane cholesterol or disruption of cholesterol trafficking results in changes in the fluidity, thickness, and intrinsic curvature of the membrane and affects the function of integral membrane proteins, including viral receptor or coreceptor conformation or distribution (40) . Thus, it will be meaningful to reveal the role of cholesterol in virus invasion from the perspective of the relationship between cholesterol and the function or conformation of plasma membrane proteins. The GBF1-ARF1-COP I axis is an important part of the early secretion pathway of host cells (24) . Moreover, it has been identified that COP I proteins mediate many additional effects beyond the well-studied trafficking roles within the early secretory pathways, including lipid homeostasis (41, 42) , cholesterol transport (43, 44) , and endosome maturation (45) . However, the reason for cholesterol relocalization in COP I-depleted cells remains unclear. Defining the underlying mechanism of COP I in cholesterol trafficking will be the focus of our next study.

We also showed that the depletion of COP I impaired the colocalization of CSFV with the late endosome marker Rab7, suggesting that COP I depletion leads to a defect in CSFV invasion through inhibiting CSFV transport to late endosomes. Previous research indicated that COP I plays several roles in the life cycles of different viruses, including an involvement in endosome-dependent virus invasion, and COP I depletion perturbs cellular endocytic transport and thereby indirectly impairs vesicular stomatitis virus (VSV) entry (46) . COP I depletion does not perturb influenza virus binding but also inhibits virus transport to late endosomes, as previously reported (14) . One point worth noting is that CSFV invasion is sensitive to endosomal cholesterol, while its invasion defect in COP I-depleted cells is not due to the accumulation of endosomal cholesterol because cholesterol accumulation in COP I-depleted cells is different from U18666A, which leads to cholesterol accumulation in the endosomal/lysosomal compartment (47) . These data reveal that COP I mediates virus invasion by regulating virion endosomal trafficking, while the role of COP I in the maturation of endosomes may provide an explanation for our results. with CSFV (MOI of 1) for 6, 12, 24, and 48 h. Next, cells were collected for RNA extraction with TRIzol lysis at the indicated times, and the GBF1, ARF1, ARF2, ARF3, ARF4, ARF5, and COP I mRNA expression levels were measured by using RT-qPCR. (B) SUVECs were infected with CSFV (MOI of 1) for 24 h. Next, cells were fixed in 4% paraformaldehyde, and immunofluorescence staining was performed with the indicated antibodies. Cells were also counterstained with DAPI to label nuclei (blue).

In summary, our current results support a model in which COP I assists CSFV to enter SUVECs in two different ways, depending on the GBF1-ARF1 function. On the one hand, the GBF1-ARF1-COP I axis mediates cholesterol trafficking to the plasma membrane to support CSFV entry. This way may play a major role in CSFV invasion since in COP I-inhibited cells, supplementation with exogenous cholesterol almost restored CSFV invasion. On the other hand, the GBF1-ARF1-COP I axis mediates CSFV transport from early to late endosomes during the entry steps.

Cells and viruses. The SUVEC, COPa-KO, and COP« -KO lines were maintained in medium 199 (Gibco, Grand Island, NY, USA) containing 10% FBS (Gibco) and a 1% penicillin-streptomycin solution (Sigma-Aldrich, St. Louis, MO, USA). Porcine kidney 15 (PK-15) cells were maintained in Dulbecco's modified Eagle's medium (DMEM) (Gibco) containing 10% FBS and a 1% penicillin-streptomycin solution. All cells were cultured at 37°C in a 5% CO 2 incubator. The CSFV Shimen strain was obtained from the China Institute of Veterinary Drug Control (Beijing, China) and propagated in PK-15 cells.

Plasmids and siRNA. Cellular RNA was extracted from SUVECs using the AG RNAex Pro reagent (Accurate Biotechnology), and cDNA was synthesized using a PrimeScript first-strand cDNA synthesis kit (TaKaRa Bio). The swine ARF1 to ARF5 genes were amplified from cDNA using PCR and cloned into pEGFP-N1 or pCDH-CMV-513B vectors to obtain pEGFP-ARF1, pEGFP-ARF2, pEGFP-ARF3, pEGFP-ARF4, pEGFP-ARF5, and CMV-ARF1, respectively. The swine COPA and COPE genes were amplified from cDNA using PCR and cloned into pCDNA-3.1 vectors to obtain pcDNA-COPA and pcDNA-COPE. pcDNA-FLAG-GBF1, pcDNA-FLAG-GBF1(1-885), pcDNA-FLAG-GBF1(1-698), pcDNA-FLAG-GBF1-Sec7, ARF1-T31N, ARF1-Q71L, Rab5-S34N, and Rab7-T22N were constructed at Genecreate (Wuhan, China). The siRNA duplexes used in this study were siRab5 (catalog number sc-36344) and siRab7 (catalog number sc-29460). Plasmids pX458-COPA and pX458-COPE were constructed at GeneBiogist (Shanghai, China). The CRISPR/Cas9 knockout cell pSpCas9(BB)-2A-GFP vector was used to clone COPA and COPE single guide RNA (sgRNA) sequences between the BbsI sites. The primers used in this study are listed in Table 1 .

Cell viability assay. Cell viability was measured using the CCK-8 assay (catalog number CK04; Dojindo) according to the manufacturer's instructions. Briefly, cells were seeded into 96-well plates and inoculated with or without Exo-1, BFA, GCA, or MbCD at the indicated concentrations for 72 h in an incubator at 37°C with 5% CO 2 . Next, the cell viability reagent was directly added and incubated with cells for another 1 h at 37°C in a shaker (60 rpm). The absorbance values were recorded at 450 nm using an Infinite M200pro system (Tecan, Männedorf, Switzerland).

RNA extraction and RT-qPCR. RT-qPCR was performed to measure the expression levels of the indicated mRNAs using the specific primers listed in Table 2 . Total cellular RNA was extracted using the 

RNAex Pro reagent (catalog number AG21101; Accurate Biotechnology) and quantified using a NanoDrop One instrument (Thermo Fisher Scientific, Waltham, MA, USA). Next, cDNA was synthesized using the Evo Moloney murine leukemia virus (M-MLV) RT for PCR kit (catalog number AG11604; Accurate Biotechnology). CSFV RNA and mRNA expression levels were normalized to those of the housekeeping gene b-actin, estimated using the SYBR green premix pro Taq HS qPCR kit (catalog number AG11701; Accurate Biotechnology) according to the manufacturer's protocol and tested using the CFX Connect real-time PCR detection system (Bio-Rad, Hercules, CA, USA). Data were analyzed using the 2 2DDCT method.

Virus titration by an indirect IFA. An indirect IFA was used for viral titration using a previously described method (17) . Briefly, cells were seeded into 96-well plates and inoculated with CSFV-containing cell supernatants for 72 h in an incubator at 37°C with 5% CO 2 . Next, the cells were fixed with 4% paraformaldehyde (PFA) for 20 min at room temperature and washed with cold phosphate-buffered 

saline (PBS) three times. After fixation, the cells were permeabilized with 0.2% Triton X-100 for 10 min at room temperature and washed three times with PBS. After blocking with 3% bovine serum albumin (BSA) for 2 h, the cells were incubated with mouse anti-E2 CSFV antibody (1:200) (Ab-mart) at room temperature for 2 h. After three washes with PBS, the cells were incubated with goat anti-mouse IgG(H1L) (Alexa Fluor 594) antibody (1:200) (catalog number ab150116; Abcam) for 1 h at 37°C. Fluorescence-positive wells were observed and recorded using a fluorescence inversion microscope (Nikon, Tokyo, Japan). Viral titration was determined as the 50% tissue culture infectious dose (TCID 50 ) per milliliter using the method proposed by Reed and Muench (48) . Measurement of virus replication using an indirect IFA. CSFV replication (Fig. 4B, Fig. 5C , and Fig.  7B ) was measured using an indirect IFA according to a method described previously (17) . Briefly, CSFVinfected cells or inhibitor-treated cells were fixed with 4% paraformaldehyde for 20 min at room temperature and washed three times with cold PBS. After fixation, the cells were permeabilized with 0.2% Triton X-100 for 10 min at room temperature and washed three times with PBS. After blocking with 3% BSA for 2 h, the cells were incubated with mouse anti-E2 CSFV antibody (1:200) (Ab-mart) at room temperature for 2 h. After three washes with PBS, the cells were incubated with goat anti-mouse IgG(H1L) (Alexa Fluor 594) antibody (1:200) (catalog number ab150116; Abcam) for 1 h at 37°C. Fluorescence-positive wells were observed using a fluorescence inversion microscope (Nikon, Tokyo, Japan).

Overexpressing cell lines were established using lentiviruses. ARF1-overexpressing lentivirus was produced as previously described (17) . Briefly, overexpression plasmids with three other plasmids (pGag/Pol, pRev, and pVSVG) were cotransfected into HEK293T cells using Lipofectamine 3000 transfection reagent (catalog number L3000015; Thermo Fisher Scientific). After 16 h, the medium was replaced with advanced DMEM containing 2% FBS, 0.01 mmol/L cholesterol (catalog number C8667; Sigma-Aldrich), 0.01 mmol/L L-a-phosphatidylcholine (catalog number P443; Sigma-Aldrich), 1:1,000-diluted chemically defined lipid (catalog number 11905031; Invitrogen), and 4 mmol/L L-glutamine (catalog number G7513; Sigma-Aldrich). After 48 h, the supernatants containing the lentivirus were collected. Lentiviral titers were determined as the TCID 50 per milliliter in HEK293T cells. Lentiviruses (MOI of 1) were used to infect SUVECs. After 12 h of infection, SUVECs were cultured in fresh medium with puromycin (5 mg/mL) (catalog number P8833; Sigma-Aldrich) to select stable cell lines. The cytomegalovirus (CMV) empty vector was treated as a control.

Western blotting. Cells were collected, total proteins were extracted using radioimmunoprecipitation assay (RIPA) lysis buffer containing 1% protease inhibitor cocktail (catalog number HY-K0010; MedChem Express), and the protein concentration was measured using a bicinchoninic acid (BCA) protein assay kit (catalog number 23227; Thermo Scientific). Next, the same amount of protein sample was separated using 8% or 10% SDS-PAGE, followed by transfer to polyvinylidene difluoride (PVDF) membranes (catalog number ISEQ00010; Merck Millipore). After blocking with 5% milk, the PVDF membranes were incubated with the following primary antibodies at room temperature for 2 h: mouse anti-FLAG antibody (1:5,000) (catalog number 66008-3-Ig; Proteintech), rabbit anti-ARF1 antibody (1:1,000) (catalog number AF2479; Beyotime), rabbit anti-COPa antibody (1:4,000) (catalog number ab192919; Abcam), rabbit anti-COP« antibody (1:4,000) (catalog number ab180946; Abcam), rabbit anti-Rab5 antibody (1:4,000) (catalog number ab18211; Abcam), rabbit anti-Rab7 antibody (1:4,000) (catalog number ab74906; Abcam), or rabbit antiactin antibody (1:5,000) (catalog number ab8227; Abcam). Subsequently, the PVDF membranes were incubated with horseradish peroxidase-conjugated goat anti-mouse IgG (catalog number SA00001-1; Proteintech) or goat anti-rabbit IgG (catalog number SA00001-2; Proteintech) secondary antibody (1:5,000) at room temperature for 1 h. Finally, the signal was detected using an enhanced chemiluminescence analysis system; b-actin served as an internal control protein.

Binding and entry assays. Drug-pretreated cells or COPa-KO and COP« -KO cell lines were infected with CSFV (MOI of 10) and cultured for 1 h at 4°C to allow virus binding without internalization, followed by washing with a precooled citrate buffer solution (pH 3) three times to remove unbound virions (binding assay). Cells were cultured in fresh culture medium at 37°C for another 2 h to allow virus internalization. After inoculation, the cells were washed with a citrate buffer solution (pH 3) to remove the noninternalized virions on the surface of cells, and the cells were then harvested after washing with precooled PBS three times (entry).

Confocal microscopy. To determine the ARF distribution ( Fig. 2A and B) , SUVECs were seeded onto glass coverslips in 35-mm cell culture dishes and cultured overnight. Cells were then transfected with the indicated plasmids (pEGFP, pEGFP-ARF1, pEGFP-ARF2, pEGFP-ARF3, pEGFP-ARF4, and pEGFP-ARF5) for 48 h and treated for 2 h with BFA. To determine Golgi structures (Fig. 3B) , SUVECs were treated with Exo-1 (5 mmol/L) for 2 h. To visualize the colocalization of CSFV, Rab5, and Rab7 ( Fig. 6I and J) , SUVECs were pretreated with the inhibitors BFA (100 nmol/L) and GCA (300 nmol/L) for 8 h to disrupt COP I function and infected with CSFV (MOI of 10) for 2 h. The colocalization of the virus with Rab5 or Rab7 was examined using confocal microscopy. To visualize the colocalization of CSFV C and E2 with COPb (Fig. 7) , SUVECs were infected with CSFV (MOI of 10) for 12 h.

Cells were then washed with precooled PBS three times, fixed with 4% paraformaldehyde for 20 min, and permeabilized with 0.5% Triton X-100 in PBS for 10 min at room temperature. After three washes with PBS, the cells were blocked with 3% BSA in PBS for 2 h at room temperature. The cells were then incubated with the indicated primary antibodies overnight at 4°C. After three washes with PBS, the cells were incubated with goat anti-mouse IgG(H1L) (Alexa Fluor 594/488) antibody (1:200) for 1 h at room temperature in the dark. Subsequently, the cells were counterstained with 49,6-diamidino-2-phenylindole (DAPI) to label cell nuclei (blue) at 37°C for 5 min and washed with cold PBS. Finally, images were captured using a confocal laser scanning microscope (LSM510 Meta; Zeiss, Germany).

BFA (100 nmol/L)-and GCA (300 nmol/L)-treated SUVECs (for 8 h) and COPa-KO and COP« -KO cell lines were washed with precooled PBS three times and fixed with 4% PFA for 15 min at room temperature. Cells were then stained with 50 mg/mL filipin (Sigma) for 30 min at 4°C and analyzed using confocal microscopy. Cholesterol measurement. BFA (100 nmol/L)-and GCA (300 nmol/L)-treated SUVECs

Cells were lysed, and the total cholesterol content was assayed using a Cholesterol/Cholesterol Ester-Glo assay kit from Promega (catalog number J3190) according to the manufacturer's instructions. Statistical analysis. All experiments were performed at least three times, and data were expressed as means 6 standard deviations (SD). The data were analyzed by a t test with GraphPad Prism 6 software

Studying classical swine fever virus: making the best of a bad virus

Structure and mechanism of COP I vesicle biogenesis

An endosomal beta COP is involved in the pH-dependent formation of transport vesicles destined for late endosomes

Rab1b interacts with GBF1 and modulates both ARF1 dynamics and COP I association

Dominant inhibitory mutants of ARF1 block endoplasmic reticulum to Golgi transport and trigger disassembly of the Golgi apparatus

A redundant mechanism of recruitment underlies the remarkable plasticity of the requirement of poliovirus replication for the cellular ArfGEF GBF1

Hijacking of multiple phospholipid biosynthetic pathways and induction of membrane biogenesis by a picornaviral 3CD protein

Mouse hepatitis coronavirus RNA replication depends on GBF1-mediated ARF1 activation

Functional and physical interaction between the Arf activator GBF1 and hepatitis C virus NS3 protein

Hepatitis C virus triggers Golgi fragmentation and autophagy through the immunity-related GTPase M

ARF1 and GBF1 generate a PI4P-enriched environment supportive of hepatitis C virus replication

Dengue virus uses a non-canonical function of the host GBF1-Arf-COP I system for capsid protein accumulation on lipid droplets

Dissecting the role of COP I complexes in influenza virus infection

The guanine nucleotide exchange factor GBF1 participates in rotavirus replication

Identification of class II ADP-ribosylation factors as cellular factors required for hepatitis C virus replication

Rab1b-GBF1-ARFs mediated intracellular trafficking is required for classical swine fever virus replication in swine umbilical vein endothelial cells

The Sec7 guanine nucleotide exchange factor GBF1 regulates membrane recruitment of BIG1 and BIG2 guanine nucleotide exchange factors to the trans-Golgi network (TGN)

9Å structure of the COP I coat reveals that the Arf1 GTPase occupies two contrasting molecular environments

Aster-B coordinates with Arf1 to regulate mitochondrial cholesterol transport

Engagement of cellular cholesterol in the life cycle of classical swine fever virus: its potential as an antiviral target

Rab5, Rab7, and Rab11 are required for caveola-dependent endocytosis of classical swine fever virus in porcine alveolar macrophages

Entry of classical swine fever virus into PK-15 cells via a pH-, dynamin-, and cholesterol-dependent, clathrin-mediated endocytic pathway that requires Rab5 and Rab7

Coats, tethers, Rabs, and SNAREs work together to mediate the intracellular destination of a transport vesicle

ER-to-Golgi transport: COP I and COP II function (review)

Role for ADP ribosylation factor 1 in the regulation of hepatitis C virus replication

Key components of COPI and COPII machineries are required for chikungunya virus replication

Hazara nairovirus requires COP I components in both Arf1-dependent and Arf1-independent stages of its replication cycle

The role of host cholesterol during flavivirus infection

Cholesterol 25-hydroxylase suppresses SARS-CoV-2 replication by blocking membrane fusion

The antiviral effector IFITM3 disrupts intracellular cholesterol homeostasis to block viral entry

Interferon-induced transmembrane protein 3 is a virus-associated protein which suppresses porcine reproductive and respiratory syndrome virus replication by blocking viral membrane fusion

The lipid raftassociated protein CD98 is required for vaccinia virus endocytosis

HIV gp41-mediated membrane fusion occurs at edges of cholesterol-rich lipid domains

West Nile virus entry requires cholesterol-rich membrane microdomains and is independent of alphavbeta3 integrin

U18666A inhibits classical swine fever virus replication through interference with intracellular cholesterol trafficking

Membrane tension: a challenging but universal physical parameter in cell biology

The secretion-coupled endocytosis correlates with membrane tension changes in RBL 2H3 cells

The role of cholesterol in membrane fusion

GBF1-Arf-COP I-ArfGAP-mediated Golgi-to-ER transport involved in regulation of lipid homeostasis

COP I complex is a regulator of lipid homeostasis

b-COP as a component of transport vesicles for HDL apolipoprotein-mediated cholesterol exocytosis

RNAi screen of Salmonella invasion shows role of COP I in membrane targeting of cholesterol and Cdc42

Endosome maturation

Genetic inactivation of COP I coatomer separately inhibits vesicular stomatitis virus entry and gene expression

Identification of NPC1 as the target of U18666A, an inhibitor of lysosomal cholesterol export and Ebola infection

A simple method of estimating fifty percent endpoints

This work was supported by the National Natural Science Foundation of China under project number 31772765.We report no conflict of interest.