Positive-strand RNA viruses stimulate hostphosphatidylcholine synthesis at viral replication sitesJiantao Zhanga,1, Zhenlu Zhanga,b, Vineela Chukkapallic, Jules A. Nchoutmboubed, Jianhui Lia, Glenn Randallc,George A. Belovd, and Xiaofeng Wanga,2

aDepartment of Plant Pathology, Physiology, and Weed Science, Virginia Tech, Blacksburg, VA 24061; bDepartment of Plant Protection, Fujian Agricultureand Forestry University, Fuzhou 350002, People’s Republic of China; cDepartment of Microbiology, The University of Chicago, Chicago, IL 60637;and dVirginia-Maryland Regional College of Veterinary Medicine, University of Maryland, College Park, MD 20742

Edited by Peter Palese, Icahn School of Medicine at Mount Sinai, New York, NY, and approved January 12, 2016 (received for review October 6, 2015)

All positive-strand RNA viruses reorganize host intracellular mem-branes to assemble their viral replication complexes (VRCs); how-ever, how these viruses modulate host lipid metabolism toaccommodate such membrane proliferation and rearrangements isnot well defined. We show that a significantly increased phospha-tidylcholine (PC) content is associated with brome mosaic virus(BMV) replication in both natural host barley and alternate hostyeast based on a lipidomic analysis. Enhanced PC levels are primarilyassociated with the perinuclear ER membrane, where BMV replica-tion takes place. More specifically, BMV replication protein 1a interactswith and recruits Cho2p (choline requiring 2), a host enzyme in-volved in PC synthesis, to the site of viral replication. These resultssuggest that PC synthesized at the site of VRC assembly, not thetransport of existing PC, is responsible for the enhanced accumula-tion. Blocking PC synthesis by deleting the CHO2 gene resulted inVRCs with wider diameters than those in wild-type cells; however,BMV replication was significantly inhibited, highlighting the criticalrole of PC in VRC formation and viral replication. We further showthat enhanced PC levels also accumulate at the replication sites ofhepatitis C virus and poliovirus, revealing a conserved feature amonga group of positive-strand RNA viruses. Our work also highlights apotential broad-spectrum antiviral strategy that would disrupt PCsynthesis at the sites of viral replication but would not altercellular processes.

positive-strand RNA viruses | viral replication complexes |virus–host interactions | virus control | phospholipids

All positive-strand RNA viruses [(+)RNA viruses], which in-clude numerous important human, animal, and plant path-

ogens, share similar strategies for genomic replication. A highlyconserved and indispensable feature of their replication is theproliferation and reorganization of host cellular membranes to as-semble viral replication complexes (VRCs). Despite this cen-tral importance, it is largely unknown how cellular membranesare rearranged by the viral replication proteins and how cellularlipid metabolism is modulated to accommodate membrane pro-liferation and remodeling.Brome mosaic virus (BMV) serves as a model for understanding

VRC formation of (+)RNA viruses (1). BMV is the type memberof the plant virus family Bromoviridae and a representative memberof the alphavirus-like superfamily, which includes many human,animal, and plant-infecting viruses (2). BMV encodes two repli-cation proteins, 1a and 2apol. 2apol serves as the replicase, whereas1a has an N-terminal methyltransferase domain (3, 4) and aC-terminal ATPase/helicase-like domain (5). Together, 1a and2apol are necessary and sufficient for BMV replication. BMVinduces vesicular structures in its surrogate host, the yeast Sac-charomyces cerevisiae, and its natural host, barley (6, 7). Thesestructures, termed spherules, have been shown to be the VRCs inyeast as 1a, 2apol, and nascent viral RNAs reside in the interiorof these compartments. Spherules are invaginations of the outerperinuclear endoplasmic reticulum (ER) membrane into the ERlumen and are about 60–80 nm in diameter (6). Remarkably,

expression of 1a alone in yeast induces spherule formation (6),which requires 1a’s amphipathic α-helix (1a amino acids 392–407)(8), helix A, and 1a–1a interactions (9, 10). In addition, severalhost proteins, including membrane-shaping reticulons (RTNs)(11) and an ESCRT (endosomal sorting complex required fortransport) component, Snf7p (sucrose nonfermenting7) (12), arerecruited by 1a to form spherules.Similar to other (+)RNA viruses, BMV promotes host lipid

synthesis and requires balanced lipids for the formation and ac-tivity of VRCs. Expression of 1a in yeast induces a ∼30% increasein total fatty acids (FAs) per cell (13). BMV also requires a highlevel of unsaturated FA (UFA) because a ∼12% decrease ofUFAs in the yeast ole1w mutant blocks its replication more than20-fold (13). In addition, deleting host ACB1 (Acyl-CoA-binding 1)gene results in formation of spherules that are smaller in sizebut are in greater number than in wild-type (WT) cells (14).ACB1 encodes acyl-CoA binding protein, which binds long-chainfatty acyl-CoAs and is involved in maintaining lipid homeostasis.Supplemented long-chain UFAs largely complement the BMVreplication defects in cells lacking ACB1, indicating that thealtered lipid composition is primarily responsible for BMV repli-cation defects (14).Cellular membranes are mainly composed of phospholipids, and

in particular, phosphatidylcholine (PC) constitutes ∼50% of totalphospholipids (15). PC is synthesized via the CDP–DAG (cytidinediphosphate–diacylglycerol) and Kennedy pathways in eukaryotes(16). PC synthesis is significantly enhanced during infection of

Significance

Positive-strand RNA viruses [(+)RNA viruses] include many im-portant human, animal, and plant pathogens. A highly con-served and indispensable feature of (+)RNA virus infection isthat these viruses proliferate and reorganize host membranesto assemble viral replication complexes (VRCs). We show thatbrome mosaic virus (BMV) stimulates phosphatidylcholine (PC)synthesis at the viral replication sites. BMV recruits a host en-zyme involved in PC synthesis to support proper VRC formationand genomic replication. We further show that hepatitis C virusand poliovirus also promote accumulation of PC at the viralreplication sites, revealing a feature common to a group of(+)RNA viruses. This virus-specific step can be targeted to de-velop a broad-spectrum antiviral strategy with the least sideeffects on host growth.

Author contributions: J.Z., G.R., G.A.B., and X.W. designed research; J.Z., Z.Z., V.C., J.A.N., andJ.L. performed research; X.W. coordinated the research; J.Z., Z.Z., V.C., J.A.N., J.L., G.R., G.A.B.,and X.W. analyzed data; and J.Z., G.R., G.A.B., and X.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1Present address: Department of Pharmacology and Toxicology, The University of Ari-zona, Tucson, AZ 85721.

2To whom correspondence should be addressed. Email: reachxw@vt.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1519730113/-/DCSupplemental.

E1064–E1073 | PNAS | Published online February 8, 2016 www.pnas.org/cgi/doi/10.1073/pnas.1519730113

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021

Dengue virus (DENV) (17), Flock House virus (FHV) (18), andpoliovirus (19, 20). A 70% and 35% increase of total PC levelswas recorded in DENV-infected mosquito cells (17) and FHV-infected Drosophila cells (18), respectively. At the peak time ofpoliovirus replication in HeLa cells, a ∼37% increase of PCcontent was observed after a 30-min chase (19). It was furtherfound that poliovirus promotes the import of FAs, which weresubsequently channeled to the viral replication sites. In addition,FAs were mainly incorporated into PCs (20).One critical question based on the aforementioned research is

whether the enhanced PC is synthesized in association with theVRCs or elsewhere in cells and subsequently transported intothe VRCs. If PC is produced in association with the VRCs, whatkey enzymes are recruited? We report here that several (+)RNAviruses, including BMV, hepatitis C virus (HCV), and poliovirus,promote significantly enhanced accumulation of PC content atthe viral replication sites, revealing a common feature of viralreplication among a group of (+)RNA viruses. We further dem-onstrate that BMV 1a interacts with and redistributes the hostenzyme, Cho2p (choline requiring 2), to the viral replication sites.As Cho2p converts phosphatidylethanolamine (PE) to PC in theCDP–DAG pathway, the relocalization of Cho2p suggests theVRC-localized PC synthesis. Deleting CHO2 inhibits BMV rep-lication up to 30-fold and results in formation of spherules that arelarger than those of WT cells. This work highlights the importanceof PC in VRC formation and the possibility of developing a noveland broad-spectrum antiviral strategy by specifically disrupting PCsynthesis at the viral replication sites but not general PC synthesis.

ResultsBMV Specifically Promotes PC Accumulation at the Site of ViralReplication in Yeast. To determine whether phospholipid synthesisis modulated during BMV replication, total lipids were extractedfrom yeast cells in the absence or presence of BMV replicationand quantified by electrospray ionization tandem mass spec-trometry. In cells with BMV replication, the molar amount oftotal phospholipids and PC increased 19% and 28%, respectively,compared with cells without BMV replication (Fig. 1 A and B).The 28% increase in PC content (P < 0.05) was comparable to the35% and 37% increase in PC content seen during FHV (18) andpoliovirus (19) infections, respectively, but lower than the 70%increase detected during DENV infection in mosquito cells (17).In addition, the amount of phosphatidylserine (PS) increasedby 40%, whereas phophatidylglycerol (PG) increased by 56%(Fig. 1A). Conversely, a nearly twofold decrease in phospha-tidic acid (PA) levels was identified (Fig. 1A). The compositionalchanges of phospholipid species are shown in Fig. S1. Based onthe molar percentage of individual phospholipids (the percentageof molar amount of each lipid against total phospholipids), theincrease in PC levels due to BMVRNA replication was ∼8% (Fig.1C), which is subtle but statistically highly significant (P < 0.01).However, because the components necessary for BMV RNAreplication were expressed from plasmids, the changes in PCcontent are likely underestimated, as plasmids are lost at a 20%rate due to segregation inefficiency during mitosis (21).We next wanted to determine if the increase in PC levels was

global or specifically associated with the sites of viral replication.To do this, we used a PC-specific monoclonal antibody (mAb),JE-1, which does not recognize other phospholipids, such as PA,PE, PS, or phosphatidylinositol (PI) (22, 23). In the absence ofBMV components, using JE-1 as a primary antibody, PC showeda faint and diffused pattern throughout the cell by immunoflu-orescence (IF) confocal microscopy (Fig. 1D). In contrast, a strongPC signal was detected in cells expressing BMV components (Fig.1D), consistent with enhanced PC levels as observed in lipidprofiles (Fig. 1 A–C). In particular, the great majority of signalcolocalized with BMV 1a in the perinuclear ER membrane, thesite of viral replication (Fig. 1D). Moreover, among cells showing

the 1a signal (∼60% of the population), PC colocalized with 1a in>92% of cells.PE is the substrate of PC in the CDP–DAG pathway, and in

addition, PE is enriched at the viral replication sites of several(+)RNA viruses (24). To determine whether PE distribution isaffected by BMV, we used a PE biosensor, duramycin peptide(25). In the absence of BMV, PE is predominantly present in thecytoplasm. We found that a portion of PE signal, albeit veryweak, colocalized with that of 1a in the presence of BMV rep-lication (Fig. S2).

BMV Promotes PC Accumulation at the Site of Viral Replication inPlants. To verify that BMV promotes an increase in PC levelsduring natural infection, we analyzed phospholipid compositionsin mock- and BMV-inoculated barley systemic leaves at 7 d post-inoculation (dpi), a time when the systemic leaves started to showsymptoms. Similar to BMV-replicating yeast cells, the amountof total phospholipid and PC increased by ∼24% and ∼29%, re-spectively, in BMV-infected plants relative to mock-inoculatedplants (Fig. 1E). In cell wall-free protoplasts prepared from mock-inoculated leaves, the PC signal was evenly distributed throughoutthe cytoplasm but was excluded from the nucleus (Fig. 1F). A moreintense PC signal was occasionally detected surrounding the nu-cleus (Fig. 1F, Top Left). Agreeing with previous reports showingthat BMV 1a, 2apol, and nascent viral RNAs reside in the perinuclearregion in barley protoplasts (26), the 1a signal surrounded the nu-cleus in the majority of BMV-infected cells (Fig. 1F). Similar to yeastcells, the majority of the PC signal in BMV-infected barley cellscolocalized with that of 1a in the perinuclear ER (Fig. 1F). Thus,enhanced PC accumulation at the sites of BMV viral replicationoccurred both in yeast and plant host cells.

BMV 1a Stimulates Perinuclear Accumulation of the Host PC SynthesisEnzyme Cho2p. We next tested whether BMV has the ability tomodulate the distribution of enzymes involved in the productionof PC. PC is synthesized via two pathways in eukaryotes: theCDP–DAG and Kennedy pathways (Fig. 2A) (27). In routine ex-perimental settings with no choline present in the growth medium,PC is primarily synthesized via the CDP–DAG pathway in yeast viathree sequential methylations of the amino head group of PE (Fig.2A). The two enzymes catalyzing the conversion of PE to PC areCho2p, the PE methyltransferase, and Opi3p, the phospholipidN-methyltransferase (Fig. 2A). Cho2p converts PE to Phosphati-dylmonomethylethanolamine (PMME) by transferring a singlemethyl moiety to PE, and Opi3p adds two extra methyl groupsto PMME to produce PC (Fig. 2A). In the presence of 1a or BMVreplication, we observed a significant increase in Cho2p proteinlevels (Fig. 2B), suggesting that 1a is able to directly or indirectlypromote the expression of certain lipid synthesis enzymes.Cho2p localizes to ER membranes, both to the peripheral ER

(the larger outer ring in Fig. 2C) and perinuclear ER (the smallerinner ring). This localization pattern of Cho2p is similar to thatof Dpm1p, another ER protein, but differs from that of Rtn1p,which localizes mainly to the peripheral ER (Fig. 2C). In thepresence of 1a, Cho2p was depleted from the peripheral ER andenriched in the perinuclear ER (Fig. 2C). The ratio of Cho2psignal in the perinuclear ER against that of the whole cell (Fp/Ft)increased from 30% in the absence of 1a to 54% in the presenceof 1a (Fig. 2D). The Fp/Ft ratio of Rtn1p, a protein known to beredistributed from the peripheral to the perinuclear ER by 1a(11), increased from ∼15% to about 50% in the presence of 1a(Fig. 2 C and D). Conversely, the distribution of Dpm1p onlychanged slightly (Fig. 2 C and D). Thus, BMV 1a specifically re-distributes Cho2p from the peripheral ER to the perinuclear ER.Opi3p localized primarily to the perinuclear ER, and its distri-bution was not affected by the expression of 1a. Nevertheless,most of 1a colocalized with Opi3p, but not all Opi3p colocalizedwith 1a (Fig. 2C).

Zhang et al. PNAS | Published online February 8, 2016 | E1065

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021

We further verified that Cho2p was redistributed by and colo-calized with 1a biochemically. Lipid rafts or detergent-resistantmembrane (DRM) fractions are membrane microdomains thatare enriched with sterol and sphingolipids (28). Although themajority of DRMs are localized at the plasma membrane, they arealso present in ER membranes in yeast cells (29). Treatment ofcellular membranes with 1% Triton X-100 (TX100) at 4 °C dis-solves most of the membrane structures but not DRMs. After ul-tracentrifugation through a density gradient, DRMs and associatedproteins, such as Gas1p (29), float to the top fractions and thedissolved membranes as well as associated proteins (e.g., Dpm1p)sediment to the bottom fractions (Fig. 2E). Under such conditions,the majority of 1a fractionated to the top 3 fractions, suggesting that1a is associated with DRMs (Fig. 2E). Although a large fraction ofCho2p-HA was found in DRMs in the absence of 1a, a portion ofCho2p-HA also localized in the bottom four fractions (Fig. 2E). Inthe presence of 1a, however, the vast majority of Cho2p-HA floatedto the top three fractions, and no signal was detected in the bottom

four fractions (Fig. 2E). In addition, Cho2p-HA localized in thesame fractions as that of 1a, indicating that 1a enriched Cho2p atthe DRMs. This result was also consistent with the fact that 1aredistributed Cho2p to the perinuclear ER membrane and colo-calized with 1a (Fig. 2C).

BMV 1a Interacts with Cho2p. Redistribution of Cho2p to the peri-nuclear ER and its enrichment at DRMs in the presence of 1a maybe due to an interaction between 1a and Cho2p or due to an in-direct and homeostatic redistribution because of ER reorganizationinduced by 1a. To verify which of the above hypotheses is correct,the potential Cho2p–1a interaction was first tested by a coimmu-noprecipitation (co-IP) assay. In yeast cells coexpressing 1a-His6and Cho2p-HA, the anti-HA antibody pulled down both Cho2p-HA and 1a-His6, and the anti-His antibody pulled down both 1a-His6 and Cho2p-HA (Fig. 3A). Conversely, Dpm1p-HA was notpulled down by the anti-His antibody and the anti-HA antibody didnot precipitate 1a-His6 (Fig. 3A) in cells expressing both 1a-His6

0 10 20 30 40 50 60 70

PA PS PE PC PI PG

Phos

phol

ipid

s (m

ol%

)

wt

0

2

4

6

8

Total PA PS PE PC PI PG Ph

osph

olip

ids

(nm

ol/O

D60

0 ce

lles)

wt wt+BMV wt+BMV

C

DAPI Merge BF

- BMV

DAPIAnti-PC(Alexa 488)

Merge BF

+ BMV

A

Anti-1a(Alexa 594)

Anti-PC(Alexa 488)

D

0 10 20 30 40 50

Total PA PS PE PC PI PG

Phos

phol

ipid

s (n

mol

/mg

DW

) Mock BMV

E

Merge BF

- BMV + BMV

DAPI Merge BF DAPIAnti-1a(Alexa 546) Merge BF

FAnti-PC(Alexa 488)

Anti-PC(Alexa 488)

Anti-1a(Alexa 405)

Merge BFAnti-PC(Alexa 488)

Anti-1a(Alexa 405)

80

100

120

140

160

Total PE PC

% o

f wt

wt wt+BMV

B

Fig. 1. BMV promotes PC accumulation at the sites of viral replication. (A) Molar amount of phospholipids in yeast cells in the absence (WT) or presence ofBMV replication (WT+BMV). (B) Percentage change in total phospholipids, PE, and PC content based on the molar amount. (C) The relative level of eachphospholipid class (molarity percentage, mol%) without or with BMV replication. *P < 0.05; **P < 0.01. (D) IF confocal microscopic images of PC distribution.PC was detected by using an anti-PC mAb (JE-1) and a secondary antibody conjugated to Alexa Fluor 488. BMV 1a was detected with an anti-1a antiserum anda secondary antibody conjugated to Alexa Fluor 594. The yellow color in the merged images indicates the colocalization of PC and 1a signals. The nucleus wasstained with DAPI. (Scale bar, 1 μm.) (E) Molar amount of phospholipids in mock- or BMV-inoculated barley plants. (F) PC distribution in barley protoplastsprepared from mock- and BMV-infected systemic leaves. 1a was detected using an anti-1a antiserum. The secondary antibodies for detecting 1a wereconjugated to Alexa Fluor 546 or 405 in the Top or Bottom panel, respectively. (Scale bar, 100 μm.)

E1066 | www.pnas.org/cgi/doi/10.1073/pnas.1519730113 Zhang et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021

and Dpm1p-HA. A previously reported interaction between Rtn1pand 1a was also confirmed in the co-IP assay (Fig. S3) (11).To verify the 1a–Cho2p interaction and determine the lo-

cation within the cell where the interaction occurs, we used aproximity ligation assay (PLA). If two proteins interact witheach other, a fluorescent dye will be deposited at the sites ofinteraction after performing an in situ PLA procedure (30).Using PLA, we detected perinuclear ER-localized dot signals(Fig. 3B) from two positive interactions, one between 1a-His6and 1a-FLAG based on a 1a self-interaction (9, 10) and theother between 1a-His6 and Rtn1p-HA (11) (Fig. 2C and Fig.S3). No such signal was detected in cells coexpressing 1a-His6and Dpm1p-HA (Fig. 3B) even though both proteins were

expressed well and localized to the ER membranes in yeast cells(Fig. 2C). In such a setting, fluorescent dots were present inperinuclear ER membranes in cells expressing 1a-His6 andCho2p-HA, suggesting that 1a interacts with Cho2p. Based onthe positive results from both co-IP and PLA, we concluded that1a recruited Cho2p to the viral replication sites most likely viathe 1a–Cho2p interaction.

Deletion of CHO2 Inhibited BMV Replication. Because BMV 1a re-cruits Cho2p for viral replication, deletion of the CHO2 geneshould significantly affect BMV replication. For the experimentsdescribed below, WT and cho2Δ cells expressing BMV compo-nents were grown in the absence or presence of choline. Because

G3P

PS

PI CDP-DAG PA

ChoPC

PMME

PECho2p

Opi3p

A

Psd2pPsd1p

MME

Eth

CD

P-D

AG

pat

hway

Kennedy pathway

DAPIAnti-HA(Alexa 594) Merge BF

Cho2p-HA

Rtn1p-HA

Dpm1p-HA

DAPI Merge BFAnti-His(Alexa 488)

- 1a-His6 + 1a-His6

Opi3p-HA

CAnti-HA(Alexa 594)

0 20 40 60 80

Dpm1p

Opi3p

Cho2p

Rtn1p

Fp/Ft (%)

-1a +1a

D

****

Cho2p-HA

1a

-BMV +BMV

Pgk1p

B

- 1a + 1a1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9

Cho2p-HA

Dpm1p

FLAG-Gas1p

1a

E

Cki1pPct1p

PDMEOpi3p

Cpt1p

Fig. 2. BMV redistributes Cho2p to the viral replication sites. (A) Diagram of the CDP–DAG and Kennedy pathways involved in PC synthesis. Key enzymes are listed.(B) Accumulated Cho2p, 1a, and pgk1p in the absence or presence of BMV replication. Cho2p was tagged with HA and expressed from its endogenous promoter incho2Δ cells, and Pgk1p served as a loading control. (C) IF images showing the distributions of host Rtn1p, Cho2p, Opi3p, and Dpm1p in the absence or presence of 1a.(Scale bar, 1 μm.) (D) Percentage of the fluorescent signals [Fp/Ft (%)] of each host protein localized to the perinuclear ER (Fp) versus total fluorescence in the cell (Ft)in the absence or presence of 1a. **P < 0.01. (E) Cho2p is enriched in the DRM fractions in the presence of 1a. Gas1p and Dpm1p serve as positive and negativecontrols, respectively.

Zhang et al. PNAS | Published online February 8, 2016 | E1067

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021

cho2Δ cells grew slowly after 5–6 generations in liquid culturewithout supplemented choline due to depleted PC content (31),we harvested cells after 20–24 h (2–3 generations) for analyzingviral replication in the absence of choline.BMV replication was tested under conditions that normally

lead to spherule or double-membrane layer replication com-plexes. Coexpression of 1a with low levels of 2apol (e.g., expressedunder the control of weak promoters) (Fig. S4) induces spherules.When 1a is coexpressed with high levels of 2apol, it leads to theformation of double-membrane layer replication complexes (32).Although morphologically different, spherule and layer replicationcomplexes support similar levels of BMV RNA replication (32).When cultured in the absence of choline, negative-strand RNA3levels in cho2Δ cells were only 3% and 18% of WT levels in layeror spherule conditions, respectively, indicating that BMV repli-cation was substantially affected as early as negative-strand RNAsynthesis. Positive-strand RNA4 accumulation in the cho2Δ cellssimilarly decreased by 16- or fourfold in layers or spherules, re-spectively (Fig. 4A). The addition of choline, which activates PCproduction via the Kennedy pathway, partially complemented theBMV replication defect in cho2Δ cells (Fig. 4A). Consistent withviral replication data, we observed a decrease in PC levels and anincrease of PE in cho2Δ cells compared with WT cells even in thepresence of choline (Fig. 5B). Nevertheless, we did not find anobvious difference in the localization of PE signals between WT

and cho2Δ cells as the PE signals were primarily detected in thecytoplasm without BMV but weakly associated with perinuclearER with BMV replication (Fig. S2). It has been reported thatsupersized lipid droplets (SLDs) form in cho2Δ cells due primarilyto an enhanced PE/phospholipids ratio (33). Although no SLDswere observed in WT cells, SLDs were formed in cho2Δ cells inthe absence or presence of BMV replication (Fig. S5). Never-theless, the relationship between LD formation and BMV repli-cation is currently unclear and awaits further investigation.

BMV-Induced Membrane Rearrangements Are Affected in cho2Δ Cells.To identify the specific step(s) of BMV RNA replication that wereaffected by deleting CHO2, we tested several steps that occur be-fore negative-stand RNA3 synthesis. As shown in Fig. 4 A and B, weonly observed a minor decrease in 2apol accumulation associatedwith layer replication complexes in cho2Δ cells, which could becomplemented by the addition of choline. The accumulationand membrane association of 1a were not affected, based onWestern blotting and membrane flotation assay, respectively(Fig. 4). It should be noted that in the engineered BMV-yeastsystem, a stop codon was incorporated into the 5′ end of thecoat protein (CP) coding sequence, and thus, no BMV CPs areproduced and no viral particles are assembled (Fig. S4).We also tested the assembly of spherule and layer replication

complexes in cho2Δ cells. Among more than 100 spherulesmeasured, 1a-induced spherules were ∼72 ± 8 nm in diameter inWT cells (Fig. 6 A and F). However, in the absence of choline,larger spherules with a wider range in diameter formed upondeletion of CHO2 (Fig. 6 B–E, pointed to by the white arrow-head). In cho2Δ cells, the average spherule diameter was about90 ± 34 nm, an increase of ∼25% compared with that of spherulesformed in WT cells (P < 0.001; Fig. 6F).

A

B DAPI Merge BF

1a-FLAG1a-His6

Rtn1p-HA1a-His6

Cho2p-HA1a-His6

PLA

Dpm1p-HA1a-His6

IP His IP HATotal

+ + ++ --

+--

+ + ++ --

+--

+ + ++ --

+--

1a-HisCho2p-HADpm1p-HA

IB 1a

IB HA

1a-His

Cho2p-HA

Dpm1p-HA

Pgk1p

Fig. 3. BMV 1a interacts with Cho2p. (A) Co-IP of 1a-His6 and Cho2p-HA. Celllysates were subjected to IP using anti-His mAb (IP His) or anti-HA pAb (IP HA),followed by Western blotting using anti-HA (IB HA) or anti-His (IB His) anti-bodies, respectively. Dpm1p served as a negative control. (B) In situ detectionof the 1a–Cho2p interaction via PLA. In cells expressing various protein pairs,PLA signals (red dots) were detected by incubating fixed yeast spheroplastswith anti-His mAb and anti-FLAG pAb or anti-HA pAb following the protocoldetailed in Experimental Procedures. The interactions between 1a-FLAG and1a-His6 as well as Rtn1p-HA and 1a-His6 served as positive controls. TheDpm1p-HA and 1a-His6 pair served as a negative control. (Scale bar, 1 μm.)

A

% of wt+RNA4

% of wt-RNA3

100±16 6±1 100±11 25±3

100±18 3±1 100±6 18±6

wt cho2 wt cho2spheruleslayers

-choline +choline

wt cho2 wt cho2

100±14 30±2 100±8 37±2

100±16 27±9 100±2 52±1

spheruleslayers

18SrRNA

+RNA3+RNA4

-RNA3

1a

2apol

Pgk1p

1a

Pgk1p

Dpm1p

wt cho2

1 2 3 4 5 6 1 2 3 4 5 6B

Fig. 4. BMV replication is blocked in cho2Δ cells. (A) BMV RNA replicationwas measured in WT or cho2Δ cells under conditions that led to the formationof double-membrane layers or spherules in the absence or presence of choline.Positive- and negative-strand viral RNAs were detected using BMV strand-specific probes. 18S rRNA served as a loading control. Total proteins wereextracted from the same numbers of cells and subjected to Western blottingwith anti-1a or -2apol antibodies. Pgk1p served as a loading control. (B) 1aassociated similarly with membranes in both WT and cho2Δ cells. 1a was de-tected in the membrane fractions along with positive control Dpm1p after aniodixanol density gradient. Pgk1p served as a control for soluble proteins.

E1068 | www.pnas.org/cgi/doi/10.1073/pnas.1519730113 Zhang et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021

Although coexpression of 1a and a low level of 2apol inducesspherule formation, high levels of 2apol (compare 2apol signals inspherule and layer complexes in Fig. 4A) along with 1a support theformation of layer replication complexes, which are stacks of 1–4double-membrane layers surrounding the nucleus (32). As shownin Fig. 7A, three stacks of membranes were formed in WT cells.Consistent with decreased accumulation of PC (Fig. 5B) and 2apol

(Fig. 4A), the number of stacked membranes surrounding thenucleus was reduced to one (79% of the cells with layer structures)or two (21%) in cho2Δ cells (Fig. 7 B–E).

BMV Replication Promotes PC Synthesis Through the Kennedy Pathwayas Well. In the Kennedy pathway, choline is first converted tocholinephosphate by Cki1p (choline kinase) and, subsequently, toCDP-choline by Pct1p (cholinephosphate cytidyltransferase), andthen to PC by Cpt1p (choline phosphotransferase) (Fig. 2A). Inaddition, ethanolamine (Eth) and monomethylethanolamine (MME)can also be converted to PE and PMME, respectively, via the Ken-nedy pathway. PE and PMME are subsequently converted to PC viathe CDP–DAG pathway (Fig. 2A).Because yeast cells were grown in synthetic-defined medium

lacking choline, the Kennedy pathway does not contributesignificantly to PC production. To determine possible contri-bution of the Kennedy pathway to BMV replication, WT andcho2Δ cells were grown in media supplemented with choline tostimulate PC production (Fig. 2A). In the latter case, PC pro-duced from the Kennedy pathway accounted for most of the PCin cho2Δ cells.

WT cells expressing 1a or supporting BMV replication grewtwice as slowly as BMV-free cells (Fig. 5A). Supplementing sub-strates that stimulate the Kennedy pathway, like Eth, MME, orcholine (Fig. 2A), reduced the doubling time of BMV-replicatingcells by 25–40% (Fig. 5A), suggesting that the rate of PC synthesis isone of the limiting factors for cell growth during BMV replication.In the presence of choline and free of BMV, PC accounted for

30% of phospholipids in cho2Δ cells (Fig. 5B), a significant in-crease from the reported 10–20% of total phospholipids in theabsence of choline (31). Despite this discrepancy, PC levels werestill ∼9% lower in cho2Δ cells compared with WT. A corre-sponding ∼8% increase in PE levels was seen in cho2Δ cells (Fig.5B), suggesting that the Kennedy pathway can complement butnot completely replace the CDP–DAG pathway. Nevertheless,BMV RNA replication under both spherule and layer conditionspromoted a 28% increase in relative PC levels in cho2Δ cellssupplemented with choline (Fig. 5 B and C). Thus, BMV appearsto have the ability to modulate PC synthesis through both theCDP–DAG (Fig. 1) and Kennedy pathways (Fig. 5C). In addi-tion, this increase was much higher than the 8% increase in WTcells without choline during BMV replication (Fig. 1C). This isconsistent with significant increases in BMV replication in cho2Δcells supplemented with choline (Fig. 4A). This also agrees wellwith the results that BMV infection significantly stimulated theamount of PC in barley cells (Fig. 1 E and F), where the Kennedypathway is dominant in PC production (34).

Virus-Stimulated and Viral Replication Site-Localized PC AccumulationIs a Conserved Feature Among Some (+)RNA Viruses. AlthoughDENV and polioviruses stimulate PC synthesis (17, 20), it is notclear whether HCV infection increases total PC levels. Never-theless, certain PC species, such as PC 30:0, increased significantly,whereas some decreased (e.g., PC 35:1) upon HCV infection (35).To determine if the enhanced synthesis/accumulation of PC at thesite of viral replication is a common feature associated with(+)RNA virus infections, we checked distribution and signal intensityof PC and PE in cells infected with DENV, HCV, or poliovirus.PE is abundantly present in the cytoplasm primarily associated

with ER membranes but excluded from the nucleus in the ab-sence of viral infection (Fig. S6). Upon infection with HCV orDENV, there is no significant redistribution of the PE signal. Forinstance, in some infected cells, HCV NS5A and DENV NS3,which are the replication proteins and components of the VRCsof each virus, colocalized with PE signal (Fig. S6). However, inmany cells, we observed strong viral protein signal, but no cor-responding increase in PE signal was present (boxed areas in Fig.S6), suggesting neither virus modulates PE distribution. It shouldbe noted that PC is primarily synthesized by the Kennedy path-way in mammalian cells, and thus, choline, but not PE, is theprimary substrate for PC.As seen in yeast and barley cells probed with the JE-1 mAb,

the PC signal was weak and evenly distributed throughout mock-infected Huh-7.5 cells (Fig. 8 A and B). Upon HCV infection, thePC signal was not only more intense, but it colocalized with NS5Ain the perinuclear ER membrane (Fig. 8A). As shown in Fig. 8B,the increase in the intensity of the PC signal in HCV-infected overmock-infected cells was statistically highly significant (P < 0.001;Fig. 8B). Moreover, the colocalization of PC with NS5A at theperinuclear ER membrane had a correlation coefficient of 0.7 asanalyzed by both Manders’ and Peasons’ approaches, indicatingthat HCV promoted PC accumulation at the sites of viral repli-cation. Conversely, no enhanced PC signal based on IF micros-copy was detected during DENV infection (Fig. 8C).To monitor the synthesis of PC in poliovirus-infected HeLa

cells, we used a choline analog, propargyl-choline, which is ef-ficiently incorporated into PC without affecting the overall PCcomposition (36). In uninfected cells, propargyl-choline wasevenly incorporated into cellular organelles but excluded from

0 0.5

1 1.5

2 2.5

None Cho MME Eth wt

Gro

wth

Rat

e (fo

lds)

wt + BMVDoublingtime (h) 7.8±0.3 5.5±0.2 5.4±0.6 6.0±0.3 3.8±0.1

None

A

80 90

100 110 120 130 140 150 160

PC C

onte

nt

(% o

f No

BM

V) No BMV

Spherules Layers

cho2 wt

B

-S-L0

10

20

30

40

50

PA PS PE PC PI PG

Phos

phal

ipid

s (m

ol%

)

wt wt-S wt-L

C

Fig. 5. BMV induces PC synthesis via the Kennedy pathway. (A) Growth rateof WT cells in the media supplemented with choline, MME, or Eth in thepresence of BMV replication. (B) Phospholipid compositions (mol%) in WT orcho2Δ cells with supplemented choline in the absence and presence of BMVreplication. S and L, BMV replication as spherule- or layer-replication com-plexes. (C) Relative PC levels (mol%) in WT and cho2Δ cells without or withBMV replication in the presence of choline to activate the Kennedy pathway;**P < 0.01.

Zhang et al. PNAS | Published online February 8, 2016 | E1069

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021

the nucleus (Fig. 8D). Upon poliovirus infection, a substantiallyenhanced propargyl-choline signal was incorporated into cellstructures surrounding the nucleus (Fig. 8D). We also testedthe distribution of the poliovirus 2B protein, which serves as amaker of poliovirus VRCs. Detergent treatment, which per-meabilizes cells and enables antibodies to access the antigens,also affects PC detection. To preserve the integrity of the mem-branes and PC, we used 0.5% saponin in simultaneous detectionof poliovirus 2B and PC. Due to the suboptimal permeabilizationconditions used, the 2B signals were mainly detected in the pe-riphery of the PC-positive membranes (Fig. 8D, Middle). Never-theless, in the majority of cells where a 2B signal was detected, thepropargyl-choline signal colocalized with that of 2B (Fig. 8D),suggesting that poliovirus promotes PC accumulation at the viralreplication sites.

DiscussionIn this report, we examined how certain (+)RNA viruses mod-ulate cellular phospholipid synthesis during infection, includingBMV, DENV, HCV, and poliovirus. Among them, we found thatBMV, HCV, and poliovirus enhanced PC synthesis and/or accu-mulation at the viral replication sites. These three viruses infectplants or humans and belong to the alphavirus-, flavivirius-, orpicornavirus-like superfamily, respectively. They also induce verydifferent membrane-bound VRCs: BMV forms single-membranespherules that remain connected to the ER membrane (6),whereas HCV forms convoluted membrane webs and DMVs(double-membrane vesicles) (37) and poliovirus induces DMVsand tubular structures (38). Despite the topological differences inVRC structures, these viruses specifically promoted PC synthesis/accumulation at the viral replication sites, suggesting a commonfeature shared by them and possibly by other (+)RNA viruses.

Virus-Induced Local Synthesis and/or Accumulation of PC Is a StrategyShared by a Group of (+)RNA Viruses.Although (+)RNA virus RNAreplication invariably takes place in membrane-bound VRCs,

different viruses replicate in different organelle membranes andmay modulate and require different cellular lipids for the as-sembly and function of their VRCs. Host phosphatidylinositol-4-kinase III α (PI4KIIIα) and PI4KIIIβ, which convert PI toPI-4-phosphate (PI4P), are recruited by HCV (39–41) and Cox-sackievirus B3 (CVB3, a picornavirus) (42), respectively. AlthoughHCV stimulates the activity of PI4KIIIα to promote PI4P pro-duction for VRC formation (39, 40), PI4P produced by PI4KIIIβ isrequired for activating or recruiting CVB3’s replicase protein (42).Other (+)RNA viruses, like tomato bushy stunt virus (TBSV), en-rich and require high levels of PE at the viral replication sites (24).Although the expression of TBSV p33 replication protein alone issufficient to induce the enrichment of PE at the places colocalizingwith that of p33 (24), it is not clear whether p33 directly or indirectlyinduces the site-specific PE synthesis or transports PE to the site ofviral replication. Conversely, it has been well-documented that PCaccumulation is significantly enhanced during DENV infection inmosquito cells (17), in addition to the infection by FHV (18) andpoliovirus (19, 20). We did not observe obvious increases in PCaccumulation in DENV-infected Huh7 cells (Fig. S6) based on theIF confocal microscopy, which may reflect differences in cell type orother experimental conditions compared with ref. 17. Nevertheless,it is not clear where the virus-stimulated PC synthesis occursduring infections of DENV, FHV, and poliovirus.Our data indicated a localized PC synthesis and/or accumu-

lation at the viral replication sites of a group of diverse (+)RNAviruses (Figs. 1 and 8). In particular, our data support a workingmodel that BMV 1a redistributes host Cho2p to the viral repli-cation sites to promote a localized PC synthesis, and this extrapool of PC is likely designated for the VRC formation and re-quired for BMV replication.We have previously shown that poliovirus 2A protein stimu-

lates activity of host long chain acyl-CoA synthetase3 for theenhanced import of FAs, which are retargeted from lipid drop-lets to the viral replication sites. We have further shown thatthese FAs are incorporated into PCs (20). Our data here

Awt

200nm Cyto

Nuc

Bcho2

200nm

Cyto

Nuc

100nm

Dcho2

200nmCyto

Nuc

Ccho2

200nm

Cyto

Nuc

100nm100nm

100nm

Nuc

Cyto

200nm

Echo2

0

10

20

30

40

40 60 80 100 120 140 Diameter of spherules (nm)

wtcho2

Dis

trib

utio

n (%

)

F

Fig. 6. Enlarged spherules are formed in cho2Δ cells. Electron microscopic images of WT (A) or cho2Δ cells (B–E) expressing BMV 1a are shown. Arrows in Apoint to spherules, and arrowheads in B–E point out spherules with a wider diameter. Images on the Right are higher magnification micrographs of the boxedareas from the images on the Left. (F) Distribution of spherule diameters in WT and cho2Δ cells. ***P < 0.001. Cyto, cytoplasm; Nuc, nucleus.

E1070 | www.pnas.org/cgi/doi/10.1073/pnas.1519730113 Zhang et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021

provided direct evidence that the enhanced PC content duringpoliovirus infection was produced at the viral replication sites(Fig. 8D).Our data also suggested that BMV stimulates PC production

via the Kennedy pathway (Fig. 5), the dominant pathway in highereukaryotes. Among the three enzymes in the Kennedy pathway(Fig. 2A), Pct1p (or CCT in other organisms) is the rate-limitingenzyme. Pct1p/CCT is inactive in its soluble form but is activatedupon binding to membranes (16, 43). FHV infection increasestranscription of the CCT-encoding Cct1 and Cct2 genes by∼twofold (18). It is possible that BMV, HCV, or poliovirus maytarget CCT to stimulate the Kennedy pathway-mediated PC pro-duction for their replications.

PC Synthesis Is Required for Proper Assembly of BMV VRCs. BlockedPC synthesis by the deletion of CHO2 inhibited BMV replicationup to 30-fold, showing the importance of PC synthesis in BMVreplication (Fig. 4A). However, this could be an overestimate ofPC’s contribution to BMV replication as deleting CHO2 pre-vented PC synthesis not just at the site of viral replication but alsothroughout the cell, which may affect BMV replication indirectly.Because BMV 1a induces spherule formation and is also in-

volved in recruiting Cho2p, the PC synthesized in the perinuclearER is most likely incorporated into the membranes that are usedto assemble VRCs. The fact that formation of both spherule andlayer replication complexes was affected in cho2Δ cells supportsthis notion (Figs. 6 and 7). For instance, only one double-mem-brane layer was formed in the majority of cho2Δ cells (Fig. 7). It ispossible that not enough PC is available to form such layers. Al-ternatively or in addition, altering the membrane lipid composi-tion might affect BMV 1a’s ability to interact with the membraneand/or 2apol. Stability and accumulation of 2apol are significantlyenhanced in the presence of 1a due to 1a-mediated recruitment of2apol into spherules or layers (32). Decreased 2apol accumulation

(Fig. 4) suggested that the changes in membrane phospholipidcomposition might affect the 1a–2apol interaction.Besides its role in maintaining membrane structure and compo-

sition, PC serves as a reservoir for producing secondary messengers:PA, DAG, and lysoPC. Enhanced PC levels may lead to the in-creased production of such signaling molecules that may stimulateviral replication. To support this notion, red clover necrotic mosaicvirus promotes PA production by recruiting phospholipase D-α and-β, which convert PC to PA by removing choline from PC (44).Interfering phospholipase-mediated PA production inhibited repli-cation of both red clover necrotic mosaic virus and BMV (44).The size of the VRCs is regulated by several factors. For

Semliki Forest virus, the length of the viral templates influencesthe size of VRCs (45). It has been previously shown that the size ofBMV spherules can be affected by several manipulations. Smallerspherules are formed in cells lacking ACBP and thus with thealtered lipid composition (14). Moreover, substitutions within themembrane-anchoring helix A of 1a also lead to the formation ofsmaller spherules (8), as does depleting the pool of host RTNs(11). In contrast, cho2Δ cells displayed spherules whose diameterwas about 25% greater than those in WT cells but were not fullyfunctional (Fig. 4A). These results provide further evidence thatthe composition of the membrane at the site of viral replication iscrucial for both proper assembly and function of VRCs.In conclusion, we revealed a common feature shared by a group

of (+)RNA viruses in that they promote significantly enhancedsynthesis and/or accumulation of PC content at the viral rep-lication sites. This pool of virus-stimulated PC likely contributesto the formation of VRCs. Although blocking general PC syn-thesis evidently affects host cell growth, looking for ways to blockthe viral replication site-localized PC accumulation could be anew way to specifically control viral replication with the leastpossible side effects on host growth.

Experimental ProceduresHosts, Viruses, Plasmids, and Antibodies. Yeast strain YPH500 and routineculture conditions have been described previously (14). BMV componentswere expressed from various constructs described in ref. 14 and illustrated inFig. S4. A 4-bp insertion in CP coding sequence results in an early stop codonso no CPs are produced. Choline chloride, Eth, or MME was added to a finalconcentration of 1 mM.

Barley plants (cv Rust) were grown in a growth chamber under 14 h light/10 h dark at 24 °C. The first expanded leaves of 6-d-old barley seedlingswere inoculated with BMV or buffer, and second expanded leaves wereharvested for lipid analysis and IF 7 DPI.

Rabbit anti-1a and mouse anti-2apol (provided by Paul Ahlquist, Universityof Wisconsin, Madison, WI), mouse anti-His (Genescript), anti-Dpm1p (LifeTechnologies), and anti-Pgk1p antibodies (Life Technologies) as well asrabbit anti-FLAG (Sigma) and anti-HA (Life Technologies) were used at 1:100dilution for IF and 1:10,000 or 1:3,000 dilution for Western blotting. Mouseanti-PC antibody (JE-1, a gift from Masato Umeda, Kyoto University, Kyoto)was used at 1:40 dilution for IF.

Lipid Analysis. Total lipids were extracted following the protocol provided bythe Kansas Lipodomics Research Center. Ten OD600 units of yeast cells wereprocessed. For barley, second expanded leaves were harvested at 7 DPI. Alllipid samples were analyzed on a triple quadruple MS/MS equipped for ESI atthe Kansas Lipodomics Research Center.

IF Assay and PLA. Yeast cells were fixed with 4% (vol/vol) formaldehyde andthe cell wall removed by lyticase and permeabilized with 0.1% TX100.Spheroplasts were incubated with specified primary antibodies overnight at4 °C and followed by a secondary antibody at 1:100 dilution for 1 h at 37 °C:anti-rabbit antibodies conjugated to Alexa Fluor 405 or 488 or anti-mouseantibody conjugated to Alexa Fluor 594 (Life Technologies). The nucleus wasstained with DAPI for 10 min. Barley protoplasts were isolated followingLoesch-Fries and Hall (46) and then processed using the same procedure as thatin yeast. Images were captured using a Zeiss LSM 510 confocal microscope. ForPE detection, spheroplasts were incubated with biotin-conjugated duramycin(Molecular Targeting Technologies) and then Texas Red-conjugated strepta-vidin (GenTex Corporation). Lipid droplets were detected by incubating live

A wt

B cho2

500nm 500nm

Nuc Nuc

Cyto Cyto

C cho2 cho2

500nm 500nmNuc

Nuc

Cyto Cyto

wt

cho2

# of stacks 2 31

19% 38% 43%

79% 21%

strain

E

D

Fig. 7. Decreased stacks of double-membrane layers are assembled in cho2Δcells. Electron micrographs of WT (A) or cho2Δ (B–D) cells expressing BMV 1aand high levels of 2apol are shown. (E) Percentage of cells with visible layerreplication complexes that contained a different number of stacked mem-branes. Cyto, cytoplasm; Nuc, nucleus.

Zhang et al. PNAS | Published online February 8, 2016 | E1071

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021

cells with Nile red for 10 min. Images for PE and LD signals (Fig. S2 and Fig. S5)were acquired with a Zeiss LSM 880 confocal microscope. To detect thecolocalization of two protein signals, signals were sequentially acquired toavoid any possible bleed-through.

Huh-7.5 cells were infected with HCV (Jc1) at an MOI (multiplicity ofinfection) of 0.05 for 48 h. Cells were fixed with 4% (vol/vol) para-formaldehyde, permeabilized with 0.1% TX100, and quenched with 50 mMammonium chloride. Cells were then blocked with 10% (vol/vol) goat serumand stained with NS5A mAb (9E10; a gift from Charlie Rice, RockefellerUniversity, New York) and then with anti-mouse IgG-Alexa Fluor 488. Cellswere then blocked again before staining for PC (JE-1) and detected withanti-mouse IgM (u chain) conjugated with Alexa Fluor 594. Cells weremounted with ProLong Gold anti-fade with DAPI (Life Technology) and vi-sualized using Olympus spinning disk microscopy. To detect PC distributionin DENV-infected cells, Huh-7.5 cells were infected with DENV-2 at an MOI of

1 and incubated for 48 h. The cells were processed as those infected by HCV.PC was detected using JE-1, and NS3 was detected using anti-NS3 antiserum.

For poliovirus, detection of PC was done by a “click chemistry” methoddescribed in Jao et al. (36). HeLa cells were infected with an MOI of 10 ofpoliovirus type I Mahoney. At 4 h postinfection, the incubation medium wasreplaced with Earls’ balanced salt solution with amino acids supplementedwith 200 μM of propargyl-choline and incubated for 1 h; the cells were thenfixed with 4% (vol/vol) formaldehyde in PBS. For immunostaining of the viralantigen, cells were incubated with mouse anti-poliovirus 2B antibody in PBSwith 0.05% saponin and followed by incubation with the anti-mouse sec-ondary antibodies conjugated with Alexa Fluor 594. Confocal images wereobtained with a Zeiss LSM 510 microscope. Colocalization analysis was per-formed using ImageJ software with colocalization plugin by Pierre Bour-doncle, Institut Jacques Monod, Service Imagerie, Paris, France.

PLA was performed following the standard procedure (Duolink, Sigma).Briefly, yeast spheroplasts were incubated with mouse anti-His6 and rabbit

PC NS5A DAPI Merge(Alexa 488)(Alexa 594)

PC(Alexa 594)

Uni

nfec

ted

HC

V-in

fect

edA

Uni

nfec

ted

DE

NV

-infe

cted

D

B

2Bpropargal-choline Merge

Uni

nfec

ted

polio

viru

s-in

fect

ed

(Alexa 594) (Alexa 488)

co-localizationC NS3 DAPI Merge

(Alexa 488)

***

Fig. 8. Newly synthesized PC colocalizes with viral proteins associated with the VRCs of HCV and poliovirus. (A) Huh-7.5 cells were infected with HCV (JC1) atan MOI of 0.05 for 48 h, fixed with 4% (vol/vol) paraformaldehyde, and permeabilized with 0.1% TX100. Cells were then incubated with anti-HCV NS5A mAb(9E10, IgG) and followed by anti-mouse IgG conjugated with Alexa Fluor 488. Cells were then reblocked, incubated with JE1, and followed by anti-mouse IgM(u chain) conjugated to Alexa Fluor 594. The yellow color in the merged images indicates the colocalization of PC and NS5A. DAPI stains the nucleus. (Scalebar, 13.5 μm.) (B) Comparison of PC signal intensity per cell in mock- and HCV-infected Huh-7.5 cells. ***P < 0.001. (C) Huh-7.5 cells were infected with DENV-2at an MOI of 1 for 48 h. Cells were processed as those of infected by HCV. PC was detected using JE-1, and DENV NS3 was detected by an anti-NS3 antiserum.(Scale bar, 13.5 μm.) (D) HeLa cells were infected with poliovirus type I Mahoney at an MOI of 10. Growth medium with 200 μM of propargal-choline wasadded at 4 h postinfection and incubated for 1 h. Propargal-choline–labeled PCs were detected by Alexa Fluor 594-conjugated azide. Poliovirus 2B wasdetected by an anti-2B mAb, and the nucleus was stained with Hoechst-33342. The colocalization panel shows the colocalized red and green pixels of 2B andPC identified with ImageJ software. (Scale bar, 10 μm.)

E1072 | www.pnas.org/cgi/doi/10.1073/pnas.1519730113 Zhang et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021

anti-HA or anti-FLAG at 1:100 dilutions overnight at 4 °C and then with sec-ondary antibodies conjugated with oligonucleotides (PLA probe MINUS andPLUS) for 1 h at 37 °C. After a 30-min ligation reaction and 100-min amplifi-cation reaction at 37 °C, samples were mounted with Duolink in situ mountingmedium with DAPI and observed using a Zeiss LSM Scanning 510 microscope.

Assay for DRMs. The yeast lysate in TNE buffer (50 mM Tris, pH7.4, 150 mMNaCl, 5 mM EDTA, pH 8.0) from 20 OD600 units of cells was subjected to adensity gradient at 201,000 × g at 4 °C for 5 h. The top fraction from themembrane flotation assay was incubated with TX100 at 1% for 30 min onice, adjusted to 40% (vol/vol) iodixanol, followed by the density gradient fora second time. The gradients were divided into nine fractions and analyzedby Western blotting for target proteins.

Co-IP Assay. The co-IP assay was performed as previously described in ref. 11.Briefly, 10 OD600 units of yeast cells were lysed in RIPA buffer (50 mM Tris at pH8.0, 1% Nonidet P-40, 0.1% SDS, 150 mM NaCl, 0.5% sodium deoxycholate,5 mM EDTA, 10 mM NaF, 10 mM NaPPi, and protease inhibitor mix). Aftercentrifugation to remove debris, the supernatant was mixed with Protein ASepharose beads (GE Healthcare) plus anti-His or anti-HA antibody and in-cubated overnight at 4 °C. Beads were washed three times with RIPA buffer,resuspended in 1× SDS gel loading buffer, and boiled for 10 min. The samples

were loaded onto a 10% (vol/vol) SDS/PAGE gel and used to detect targetproteins.

Electron Microscopy. Fixation, dehydration, and embedding of yeast cellswere performed as previously described (14). Images were obtained using aJEOL JEM 1400 transmission electron microscope at the Virginia-MarylandRegional College of Veterinary Medicine.

ACKNOWLEDGMENTS. We thank Drs. Glenda Gillaspy and Boris Vinatzer fordiscussions; Haijie Liu, Sarah Lau, and Alicia Campos for general assistance;Guijuan He for setting up the Nile red protocol; Drs. Yuji Hara and MasatoUmeda for providing JE-1 anti-PC antibody; Kansas Lipidomics ResearchCenter for processing lipid samples; and Drs. Arturo Diaz and JanetWebster for critical reading of the manuscript. We appreciate the helpfrom Drs. Zackary Adelman, Kevin Miles, and Glady Hazitha Samuel forproviding virus-infected mosquito cells. Research at the G.R. laboratory issupported by NIH Grant DK102883, and V.C. is supported by an AmericanHeart Association postdoctoral fellowship. Research at the G.A.B. labora-tory is supported by funds from the Maryland Agricultural ExperimentalStation and NIH Grant AI115383. Work in the X.W. laboratory is supportedby a Virginia Tech Startup fund, National Science Foundation Grant1265260, and the Virginia Agricultural Experiment Station and HatchProgram of National Institute of Food and Agriculture, United StatesDepartment of Agriculture.

1. Diaz A, Wang X (2014) Bromovirus-induced remodeling of host membranes duringviral RNA replication. Curr Opin Virol 9(12):104–110.

2. Wang X, Ahlquist P (2008) Brome mosaic virus (Bromoviridae). Encyclopedia of Virology,eds Mahy BWJ, van Regenmortel MHV (Elsevier, Oxford, UK), 3rd Ed, pp 381–386.

3. Ahola T, Ahlquist P (1999) Putative RNA capping activities encoded by brome mosaicvirus: methylation and covalent binding of guanylate by replicase protein 1a. J Virol73(12):10061–10069.

4. Kong F, Sivakumaran K, Kao C (1999) The N-terminal half of the brome mosaic virus1a protein has RNA capping-associated activities: Specificity for GTP and S-adenosylmethionine. Virology 259(1):200–210.

5. Wang X, et al. (2005) Brome mosaic virus 1a nucleoside triphosphatase/helicase domainplays crucial roles in recruiting RNA replication templates. J Virol 79(21):13747–13758.

6. Schwartz M, et al. (2002) A positive-strand RNA virus replication complex parallelsform and function of retrovirus capsids. Mol Cell 9(3):505–514.

7. Martelli GP, Russo M (1985) Virus-Host Relationships: Symptomatological and UltrastructuralAspects. The Plant Viruses, ed Francki RI (Plenum Press, New York), Vol 1, pp 163–205.

8. Liu L, et al. (2009) An amphipathic alpha-helix controls multiple roles of brome mosaicvirus protein 1a in RNA replication complex assembly and function. PLoS Pathog 5(3):e1000351.

9. Diaz A, Gallei A, Ahlquist P (2012) Bromovirus RNA replication compartment forma-tion requires concerted action of 1a’s self-interacting RNA capping and helicase do-mains. J Virol 86(2):821–834.

10. O’Reilly EK, Paul JD, Kao CC (1997) Analysis of the interaction of viral RNA replicationproteins by using the yeast two-hybrid assay. J Virol 71(10):7526–7532.

11. Diaz A, Wang X, Ahlquist P (2010) Membrane-shaping host reticulon proteins playcrucial roles in viral RNA replication compartment formation and function. Proc NatlAcad Sci USA 107(37):16291–16296.

12. Diaz A, Zhang J, Ollwerther A, Wang X, Ahlquist P (2015) Host ESCRT proteins arerequired for bromovirus RNA replication compartment assembly and function. PLoSPathog 11(3):e1004742.

13. Lee WM, Ahlquist P (2003) Membrane synthesis, specific lipid requirements, and lo-calized lipid composition changes associated with a positive-strand RNA virus RNAreplication protein. J Virol 77(23):12819–12828.

14. Zhang J, Diaz A, Mao L, Ahlquist P, Wang X (2012) Host acyl coenzyme A bindingprotein regulates replication complex assembly and activity of a positive-strand RNAvirus. J Virol 86(9):5110–5121.

15. van Meer G, Voelker DR, Feigenson GW (2008) Membrane lipids: Where they are andhow they behave. Nat Rev Mol Cell Biol 9(2):112–124.

16. Henry SA, Kohlwein SD, Carman GM (2012) Metabolism and regulation of glycer-olipids in the yeast Saccharomyces cerevisiae. Genetics 190(2):317–349.

17. Perera R, et al. (2012) Dengue virus infection perturbs lipid homeostasis in infectedmosquito cells. PLoS Pathog 8(3):e1002584.

18. Castorena KM, Stapleford KA, Miller DJ (2010) Complementary transcriptomic, lip-idomic, and targeted functional genetic analyses in cultured Drosophila cells high-light the role of glycerophospholipid metabolism in Flock House virus RNAreplication. BMC Genomics 11:183.

19. Vance DE, Trip EM, Paddon HB (1980) Poliovirus increases phosphatidylcholine bio-synthesis in HeLa cells by stimulation of the rate-limiting reaction catalyzed by CTP:Phosphocholine cytidylyltransferase. J Biol Chem 255(3):1064–1069.

20. Nchoutmboube JA, et al. (2013) Increased long chain acyl-Coa synthetase activity andfatty acid import is linked to membrane synthesis for development of picornavirusreplication organelles. PLoS Pathog 9(6):e1003401.

21. Hegemann JH, et al. (1999) A fast method to diagnose chromosome and plasmid lossin Saccharomyces cerevisiae strains. Yeast 15(10B):1009–1019.

22. Fujimoto K, Umeda M, Fujimoto T (1996) Transmembrane phospholipid distributionrevealed by freeze-fracture replica labeling. J Cell Sci 109(Pt 10):2453–2460.

23. Nam KS, Igarashi K, Umeda M, Inoue K (1990) Production and characterization ofmonoclonal antibodies that specifically bind to phosphatidylcholine. Biochim BiophysActa 1046(1):89–96.

24. Xu K, Nagy PD (2015) RNA virus replication depends on enrichment of phosphati-dylethanolamine at replication sites in subcellular membranes. Proc Natl Acad Sci USA112(14):E1782–E1791.

25. Zhao M, Li Z, Bugenhagen S (2008) 99mTc-labeled duramycin as a novel phosphati-dylethanolamine-binding molecular probe. J Nucl Med 49(8):1345–1352.

26. Restrepo-Hartwig MA, Ahlquist P (1996) Brome mosaic virus helicase- and polymer-ase-like proteins colocalize on the endoplasmic reticulum at sites of viral RNA syn-thesis. J Virol 70(12):8908–8916.

27. Carman GM, Henry SA (1999) Phospholipid biosynthesis in the yeast Saccharomycescerevisiae and interrelationship with other metabolic processes. Prog Lipid Res 38(5-6):361–399.

28. Simons K, Gerl MJ (2010) Revitalizing membrane rafts: New tools and insights. NatRev Mol Cell Biol 11(10):688–699.

29. Bagnat M, Keränen S, Shevchenko A, Shevchenko A, Simons K (2000) Lipid raftsfunction in biosynthetic delivery of proteins to the cell surface in yeast. Proc Natl AcadSci USA 97(7):3254–3259.

30. Söderberg O, et al. (2006) Direct observation of individual endogenous proteincomplexes in situ by proximity ligation. Nat Methods 3(12):995–1000.

31. Summers EF, Letts VA, McGraw P, Henry SA (1988) Saccharomyces cerevisiae cho2mutants are deficient in phospholipid methylation and cross-pathway regulation ofinositol synthesis. Genetics 120(4):909–922.

32. Schwartz M, Chen J, Lee WM, Janda M, Ahlquist P (2004) Alternate, virus-inducedmembrane rearrangements support positive-strand RNA virus genome replication.Proc Natl Acad Sci USA 101(31):11263–11268.

33. Fei W, et al. (2011) A role for phosphatidic acid in the formation of “supersized” lipiddroplets. PLoS Genet 7(7):e1002201.

34. Li-Beisson Y, et al. (2013) Acyl-lipid metabolism. Arabidopsis Book 11:e0161.35. Diamond DL, et al. (2010) Temporal proteome and lipidome profiles reveal hepatitis C

virus-associated reprogramming of hepatocellular metabolism and bioenergetics.PLoS Pathog 6(1):e1000719.

36. Jao CY, Roth M, Welti R, Salic A (2009) Metabolic labeling and direct imaging ofcholine phospholipids in vivo. Proc Natl Acad Sci USA 106(36):15332–15337.

37. Romero-Brey I, et al. (2012) Three-dimensional architecture and biogenesis of membranestructures associated with hepatitis C virus replication. PLoS Pathog 8(12):e1003056.

38. Belov GA, et al. (2012) Complex dynamic development of poliovirus membranousreplication complexes. J Virol 86(1):302–312.

39. Berger KL, et al. (2009) Roles for endocytic trafficking and phosphatidylinositol 4-kinaseIII alpha in hepatitis C virus replication. Proc Natl Acad Sci USA 106(18):7577–7582.

40. Reiss S, et al. (2011) Recruitment and activation of a lipid kinase by hepatitis C virusNS5A is essential for integrity of the membranous replication compartment. Cell HostMicrobe 9(1):32–45.

41. Berger KL, Kelly SM, Jordan TX, Tartell MA, Randall G (2011) Hepatitis C virus stim-ulates the phosphatidylinositol 4-kinase III alpha-dependent phosphatidylinositol4-phosphate production that is essential for its replication. J Virol 85(17):8870–8883.

42. Hsu NY, et al. (2010) Viral reorganization of the secretory pathway generates distinctorganelles for RNA replication. Cell 141(5):799–811.

43. Cornell RB, Northwood IC (2000) Regulation of CTP:phosphocholine cytidylyl-transferase by amphitropism and relocalization. Trends Biochem Sci 25(9):441–447.

44. Hyodo K, et al. (2015) Phosphatidic acid produced by phospholipase D promotes RNAreplication of a plant RNA virus. PLoS Pathog 11(5):e1004909.

45. Kallio K, et al. (2013) Template RNA length determines the size of replication complexspherules for Semliki Forest virus. J Virol 87(16):9125–9134.

46. Loesch-Fries LS, Hall TC (1980) Synthesis, accumulation and encapsidation of individualbrome mosaic virus RNA components in barley protoplasts. J Gen Virol 47:323–332.

Zhang et al. PNAS | Published online February 8, 2016 | E1073

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

021