Global Interactomics Uncovers Extensive Organellar Targeting by Zika Virus*

The geographical range of Zika virus (ZIKV) has expanded dramatically in recent decades, but no antiviral drugs or vaccines are available for the treatment or prevention of ZIKV infection. Here we map host cell protein interactions for each of the ten ZIKV polypeptides, generating a protein network comprising >3000 interactions. Mining this rich dataset, we highlight multiple examples of how ZIKV may usurp or disrupt the function of host organelles and uncover an important role for peroxisomes in ZIKV infection. Graphical Abstract Highlights BioID and IP-MS were conducted to generate a global ZIKV-host protein interactome Interactome consists of >3000 high confidence ZIKV-host protein interactions Data mining indicates that ZIKV proteins interact with multiple host cell organelles An important role for peroxisomes in ZIKV infection is uncovered Zika virus (ZIKV) is a membrane enveloped Flavivirus with a positive strand RNA genome, transmitted by Aedes mosquitoes. The geographical range of ZIKV has dramatically expanded in recent decades resulting in increasing numbers of infected individuals, and the spike in ZIKV infections has been linked to significant increases in both Guillain-Barré syndrome and microcephaly. Although a large number of host proteins have been physically and/or functionally linked to other Flaviviruses, very little is known about the virus-host protein interactions established by ZIKV. Here we map host cell protein interaction profiles for each of the ten polypeptides encoded in the ZIKV genome, generating a protein topology network comprising 3033 interactions among 1224 unique human polypeptides. The interactome is enriched in proteins with roles in polypeptide processing and quality control, vesicle trafficking, RNA processing and lipid metabolism. >60% of the network components have been previously implicated in other types of viral infections; the remaining interactors comprise hundreds of new putative ZIKV functional partners. Mining this rich data set, we highlight several examples of how ZIKV may usurp or disrupt the function of host cell organelles, and uncover an important role for peroxisomes in ZIKV infection.


Global Interactomics Uncovers Extensive Organellar Targeting by Zika Virus* □
Local transmission of ZIKV has now been reported in Ͼ50 countries (8).
Most ZIKV infections are asymptomatic, though ϳ20% of infected individuals develop flu-like symptoms (rash, fever, joint pain) that resolve after a few days. Notably, however, ZIKV can infect human neural progenitors and post-mitotic neurons (9,10). Consistent with these observations, recent ZIKV outbreaks have been linked to significant increases in Guillain-Barré syndrome ((11) an autoimmune disorder affecting neuromuscular function) and microcephaly (12)(13)(14)(15), leading the WHO to declare a Public Health Emergency of International Concern. To date, no antiviral drugs or vaccines are available for the treatment or prevention of ZIKV infection.
sh/siRNA and CRISPR-Cas9-based screens have identified several host factors required for ZIKV infection and replication (21)(22)(23). Importantly, however, very little is known regarding individual ZIKV protein -host protein interactions. Here, we present a BioID/IP-MS ZIKV-host interactome. Overall, the interactome is significantly enriched in proteins associated with polypeptide processing and quality control, vesicle trafficking, RNA processing machinery and proteins linked to lipid metabolism. Notably, the data set also highlights extensive organellar targeting by ZIKV proteins. For example, we report that: (1) the ZIKV Capsid protein is targeted to a variety of host endomembrane locations and can "remodel" nuclear/ER membranes; (2) the NS5 helicase protein is targeted to, and partially disrupts, Cajal Bodies; (3) the NS3-NS2B protease interacts with the HOPS/CORVET complex, and modifies lysosome volume, and; (4) NS2A is specifically targeted to peroxisomal membranes. Consistent with this observation, we demonstrate that peroxisome function is linked to efficient ZIKV replication in culture. Together, the ZIKV-host interactome thus provides a valuable resource for better understanding ZIKV biology and identifies numerous virus-host protein interactions that could be targeted in antiviral therapeutic strategies.

EXPERIMENTAL PROCEDURES
Experimental Design and Statistical Rationale-Both BioID and anti-Flag IP-MS analyses were performed on each of the ten ZIKV polypeptides, expressed individually in HEK293 T-REx cells, fused either with an N-terminal FlagBirA* or a C-terminal BirA*Flag epitope tag. MS samples prepared from two biological replicates were each analyzed twice (i.e. four MS runs) on a Thermo Q-Exactive HF quadrupole-Orbitrap mass spectrometer. For each prey protein, all four sample runs were compared against the two highest peptide counts among 16 control samples, prepared from HEK293 T-REx cells expressing the FlagBirA* tag alone (see below). Protein IDs with at least two unique peptides and an iProphet score Ͼ0.9 (corresponding to ϳ1% FDR) were defined as high-confidence interactors, when their Bayesian False Discovery Rate (BFDR) assigned by SAINT Express (v3.6) was below 1% (supplemental Table S1). Additional details provided below.
Zika Virus Infection-Cells at 70% confluence were rinsed with PBS, and virus diluted in growth medium supplemented with L-glutamine and 1% FBS. Virus inoculum was removed after 1 h and replaced with fresh media. Cells and supernatants were harvested at the indicated times post-infection for determination of TCID 50 .
Flag Affinity Purification-5 ϫ 150 cm 2 dishes of subconfluent (80%) 293 T-REx cells expressing the protein of interest were scraped into PBS, pooled, washed twice in 10 ml PBS, and collected by centrifugation at 1000 ϫ g for 5 min at 4°C. Cell pellets were stored at Ϫ80°C until lysis. The cell pellet was weighed, and 1:4 pellet weight/lysis buffer (by volume) was added. Lysis buffer consisted of 50 mM HEPES-NaOH (pH 8.0), 100 mM KCl, 2 mM EDTA, 0.1% Nonidet P-40, 10% glycerol, 1 mM PMSF, 1 mM DTT, and 1:500 protease inhibitor mixture (Sigma-Aldrich, St. Louis, MO). On resuspension, cells were incubated on ice for 10 min, subjected to one additional freeze-thaw cycle, then centrifuged at 27,000 ϫ g for 20 min at 4°C. The supernatant was transferred to a fresh 15 ml conical tube, and 1:1000 turbonuclease (BioVision, Milpitas, CA) plus 30 l packed, pre-equilibrated Flag-M2 agarose beads (Sigma-Aldrich) were added. The mixture was incubated for 2 h at 4°C with endover-end rotation. Beads were pelleted by centrifugation at 1000 ϫ g for 1 min and transferred with 1 ml of lysis buffer to a fresh centrifuge tube. Beads were washed once with 1 ml lysis buffer and twice with 1 ml ammonium bicarbonate (ammbic) rinsing buffer (50 mM ammbic, pH 8.0, 75 mM KCl). Elution was performed by incubating the beads with 150 l of 125 mM ammonium hydroxide (pH ϳ11). The elution step was repeated twice, and the combined eluate centrifuged at 15,000 ϫ g for 1 min, transferred to a fresh centrifuge tube, and lyophilized.
Biotin-streptavidin Affinity Purification-Cell pellets were resuspended in 10 ml of lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 0.1% SDS, 1:500 protease inhibitor mixture (Sigma-Aldrich), 1:1000 turbonuclease), incubated on an end-over-end rotator at 4°C for 1 h, briefly sonicated to disrupt any visible aggregates, then centrifuged at 16,000 ϫ g for 30 min at 4°C. The supernatant was transferred to a fresh 15 ml conical tube, 30 l of packed, pre-equilibrated streptavidin-Sepharose beads (GE, Boston, MA) were added, and the mixture incubated for 3 h at 4°C with end-over-end rotation. Beads were pelleted by centrifugation at 2000 rpm for 2 min and transferred with 1 ml of lysis buffer to a fresh Eppendorf tube. Beads were washed once with 1 ml lysis buffer and twice with 1 ml of 50 mM ammbic (pH 8.3). Beads were transferred in ammbic to a fresh centrifuge tube and washed two more times with 1 ml ammbic buffer. Tryptic digestion was performed by incubating the beads with 1 g MS grade TPCK trypsin (Promega, Madison, WI) dissolved in 200 l of 50 mM ammbic (pH 8.3) overnight at 37°C. The following morning, an additional 0.5 g trypsin was added, and the beads incubated for 2 h at 37°C. Beads were pelleted by centrifugation at 2000 ϫ g for 2 min, and the supernatant was transferred to a fresh Eppendorf tube. Beads were washed twice with 150 l of 50 mM ammonium bicarbonate, and washes pooled with the eluate. The sample was lyophilized and resuspended in buffer A (0.1% formic acid). One-fifth of the sample was analyzed per MS run.
Mass Spectrometry Analysis-High performance liquid chromatography was conducted using a 2 cm pre-column (Acclaim PepMap 50 mm ϫ 100 m inner diameter (ID)) and 50 cm analytical column (Acclaim PepMap, 500 mm ϫ 75 m ID; C18; 2 um; 100 Å, Thermo 1 The abbreviations used are: ZIKV, Zika virus; BioID, proximity-dependent biotin identification; CB, Cajal body; CRISPR, clustered regularly interspersed short palindromic repeats; C, capsid protein; DENV, Dengue virus; Env, ZIKV envelope protein; GO, gene ontology; IF, immunofluorescence; NS1-5, ZIKV nonstructural viral proteins 1-5; PreM, ZIKV pre-membrane/membrane protein; sh/siRNA, short hairpin/small interfering RNA; UPS, ubiquitin-proteasome system. Fisher Scientific, Waltham, MA), running a 120 min reversed-phase buffer gradient at 225 nl/min on a Proxeon EASY-nLC 1000 pump in-line with a Thermo Q-Exactive HF quadrupole-Orbitrap mass spectrometer. A parent ion scan was performed using a resolving power of 60,000, then up to the twenty most intense peaks were selected for MS/MS (minimum ion count of 1000 for activation) using higher energy collision induced dissociation (HCD) fragmentation. Dynamic exclusion was activated such that MS/MS of the same m/z (within a range of 10 ppm; exclusion list size ϭ 500) detected twice within 5 s were excluded from analysis for 15 s. For protein identification, Thermo .RAW files were converted to the .mzXML format using Pro-teoWizard (v3.0.10800; 4/27/2017) (25). Using the Trans-Proteomic Pipeline (TPP v4.7 Polar Vortex rev 1, linux build 201705011551) converted files were searched using X!Tandem (Jackhammer TPP 2013.06.15.1) (26) and Comet (2014.02 rev. 2) (27) against the Human RefSeq Version 45 database (containing 36113 entries). Search parameters specified a parent ion mass tolerance of 10 ppm, and an MS/MS fragment ion tolerance of 0.4 Da, with up to 2 missed cleavages allowed for trypsin. No fixed modifications were used. Variable modifications of ϩ16@M and W, ϩ32@M and W, ϩ42@N terminus, and ϩ1@N and Q were allowed. Proteins identified with an iProphet cut-off of 0.9 (corresponding to Յ1% FDR) and at least two unique peptides were analyzed with SAINT Express v.3.6. Control runs (16 runs for BioID; 16 runs for IP-MS; all from cells expressing the FlagBirA* epitope tag only) were collapsed to the four highest spectral counts for each prey, and high confidence interactors were defined as those with BFDRՅ0.01. Each bait protein was analyzed with both an N-and C-terminal BirA* tag. Two biological replicates (i.e. separate transfections) were each subjected to two MS runs (two technical replicates). Each of the ten ZIKV bait proteins was analyzed using both BioID and Flag IP, for a total of 160 MS runs. Average R 2 between technical replicates: 0.976. Average R 2 between biological replicates: 0.888 (supplemental Table S1).
Cajal Body Quantification-HeLa cells were transfected with the FlagBirA*-NS5 vector, incubated 24 h, then incubated an additional 24 h on coverslips with biotin (50 M). Cells were fixed and permeabilized as above, then stained with anti-COIL and Streptavidin AF594 (as a reporter of positive FlagBirA*-NS5 cells). Cajal body (CB) volumes (red objects gated between 0.1-5 m 3 ) were quantified using the Volocity 6.0 measurement tool. In each experiment, CB volumes (COIL-positive nuclear bodies) were quantified in DAPI positive objects (gated between 100 -500 m 3 ) of the NS5-positive gated population versus the NS5 low/negative gated population. Data (raw data supplemental Table S4) of 25 stitched fields for each of the two independent experiments were merged and plotted using Excel. For calculation of the number of CB/cell, three nonoverlapping fields per experiment were used, and the number of CB observed in each subpopulation was averaged according to the number of nuclei (supplemental Table S3). Student's t test was performed with Excel T.TEST function (Microsoft, Redmond, WA).
Lysosome Quantification-HeLa cells were transfected with the indicated GFP-tagged constructs and incubated 48 h on coverslips at 37°C. Cultured cells were stained with 1 g/ml of Hoechst 33342 dye (Fisher Scientific) for 15 min at 37°C. Cells were washed twice with pre-warmed PBS, before incubating in DMEM with no pH indicator (GIBCO, Invitrogen). Lysotracker Red DND-99 (Molecular Probes, Eugene, OR) was used at 50 nM for 30 min. Coverslips were fixed with 4% paraformaldehyde for 15 min, then washed three times with PBS, and mounted. Lysosome volumes (red objects gated between 0.1-5 m 3 ) were quantified using the Volocity 6.0 measurement tool. Data from 16 stitched fields per condition were merged. ANOVA was performed with the FUNCRES.XLAM Excel add-in. Student's t test was performed with Excel T.TEST function (Microsoft). Graph and data distribution were assembled with Excel.
Peroxisome Microscopy-Cells were fixed in paraformaldehyde and permeabilized with 0.1% Triton X-100 in PBS, followed by incubation with the indicated antibodies. A Zeiss LSM 710 laser-scanning confocal microscope with a 63 ϫ 1.4 NA oil immersion objective was used to acquire fluorescence images of cells. Z-stacks series were collected for measuring cell and peroxisome volume. Volocity 5.0 software (Perkin Elmer, Waltham, MA) was used to quantify cell volume, peroxisome number (peroxisome density) and peroxisome volume. ImageJ (NIH, Bethesda) was used to adjust brightness and contrast of images.

RESULTS
The ZIKV genome encodes a single large polyprotein, which is cleaved by viral and host cell proteases to generate ten distinct polypeptides (Fig. 1A). To better understand ZIKV-host interactions, the ten ZIKV proteins were fused in-frame with an N-or C-terminal Flag-BirA R118G tag (constructs based on ZIKV H/PF/2013 ANO46305.1 (28); individual proteins based on cleavage sites reported in (18)) and expressed in HEK 293 cells, a system that displays characteristics of immature neurons (29) and which can be infected with ZIKV (30) (supplemental Fig. S1A, S1B). Bait protein localization was queried using anti-Flag immunofluorescence (IF) confocal microscopy (supplemental Fig. S1C; additional IF available at ProHits-web.ca). Localizations were consistent with several previous reports on ZIKV and/or related Flaviviral proteins (supplemental Fig. S1C). Colocalization of otherwise identical N-and C-terminal-tagged ZIKV proteins indicated that the tagging moieties did not significantly affect intracellular localization (supplemental Fig. S1D).
Using these cells, the virus-host protein interaction landscape was characterized using immunoprecipitation coupled with mass spectrometry (Flag IP-MS) and proximity-dependent biotin identification (BioID (31), supplemental Table S1, supplemental Fig. S2A-S2B; searchable data set available at ProHits-web.ca; raw data available at massive.ucsd.edu, accession #MSV000082311). Using a Bayesian false discovery rate of Յ1%, 3033 high confidence interactions were identified among 1224 unique human proteins.
To better understand the ZIKV-host interactome, a selforganized force-directed bait-prey topology map was generated, in which map location is determined by the number and abundance (i.e. total peptide counts) of host interactors (Fig.  1B). As expected, this approach closely clustered all six predicted ZIKV transmembrane proteins. This "core" map region encompasses the most highly interconnected preys (i.e. those that interact with the highest number of baits; Fig. 1C) and is enriched for nuclear, endoplasmic reticulum (ER), Golgi and other membrane polypeptides. Capsid protein (C) interactors were enriched in both membrane and nonmembrane proteins; this bait polypeptide thus occupies a distinct territory connected to, yet slightly removed from, the core. The three nonmembrane ZIKV proteins (NS1, NS3, and NS5) occupy peripheral areas of the map (Fig. 1B). These areas are enriched for e.g. DNA and RNA-associated proteins (Fig.  1C). Previously described virus protein interactors and putative novel interactors are represented throughout the map (Fig. 1C).
Three hundred forty-seven physical or functional interactors previously implicated in Flavivirus infection, and an additional 434 interactors linked to other viruses, are represented in the ZIKV interactome ( Fig. 1D; supplemental Table S1), comprising Ͼ60% of the interaction network and highlighting the quality of the data set. 738 polypeptides were detected as high confidence interactors for a single ZIKV bait protein, revealing a high degree of specificity for the viral-host interactions. Relevant to the link between ZIKV infection and fetal brain development, more than 85% of the ZIKV interacting proteins identified in HEK 293 cells are expressed in the human cerebral cortex (32) (supplemental Fig. S2C, supplemental Table S2).
One hundred sixty-five proteins were detected by both BioID and IP-MS (Fig. 1E). This protein population was significantly enriched for host components localized to soluble cellular fractions (e.g. RNA binding proteins (73/165; GO: 0003723) and nucleolar (54/165; GO:0005730) components) but was depleted for membrane proteins (supplemental Table  S2). Only ten interactors in this group were annotated as intrinsic to intracellular membranes (GO:0031300).
Although the two techniques identified a roughly equal number of high confidence interactors (701 versus 688 proteins identified by IP-MS and BioID, respectively), BioID identified Ͼ3x more membrane-associated polypeptides ( Fig. 1E; see supplemental Table S3 for all GO annotations). Consistent with this observation, although both data sets are enriched for e.g. ubiquitin proteasome system (UPS) components, the IP-MS data set is uniquely enriched in cytoplasmic polypeptides and RNA processing proteins, whereas the BioID data set is specifically enriched for nuclear, ER and Golgi membrane proteins, cytoplasmic vesicle and lipid droplet components (Fig. 1F). Like previous reports (33,34), our data thus suggest that IP-MS and BioID sample largely discrete protein populations and intracellular environments, and thereby identify complementary interactor sets.
Consistent with previous CRISPR-and sh/siRNA-based screens identifying host proteins required for Flavivirus replication (21-23, 35, 36), multiple ZIKV polypeptides interact with components of the ER signal peptidase complex, ER membrane complex, glucosyltransferases and the BAT3 tailanchored membrane protein insertion complex ( Fig. 2A). This large suite of host proteins is thus likely to be required for proper ZIKV polypeptide folding, processing, posttranslational modification, membrane localization and virion assembly.  supplemental Table S2 for full GO category analysis.) Individual ZIKV protein interactomes revealed targeting to host proteins with a variety of biological functions. For example, the Capsid (C) protein IP-MS identified interactions with multiple nucleolar proteins (Fig. 2B). Consistent with this observation and previous reports demonstrating that the related Dengue virus (DENV) capsid protein is localized to the nucleolus (37), IF confocal microscopy of ZIKV C-expressing cells revealed colocalization with the nucleolar marker fibrillarin (Fig. 2C). BioID identified C proximity interactions at the nuclear, ER and other membranes (Fig. 2B). C is also localized to DAPI-poor nuclear structures not associated with fibrillarin, and which are not observed in cells expressing other ZIKV proteins (Fig. 2D, supplemental Fig. S1C). These structures were also detected by antibodies directed against lamin B1, suggesting that they are derived (at least in part) from the nuclear membrane. Surprisingly, these structures were also detected by antibodies directed against the ER markers prolyl 4-hydroxylase subunit B (P4HB) and calnexin (CANX; Fig. 2E). Consistent with this observation, when Flag-FAM134B (a Cinteracting protein required for ER-specific autophagy (38) and which inhibits ZIKV replication (39)) was coexpressed with GFP-Capsid protein, it was also recruited to these nuclear structures (Fig. 2E). Like other Flaviviruses, ZIKV extensively remodels the host endomembrane system to create viral replication factories (40). Although additional study will be required to better understand these C-associated structures, our observations are consistent with previous reports (e.g. (41)) suggesting that ZIKV promotes the inward budding of host membranes.
In the cytoplasm, C is localized to the ER and vesicular structures (Fig. 2F). The C interactome includes several proteins associated with lipid storage and metabolism (Fig. 2B) and, consistent with previous work demonstrating that both DENV and Hepatitis C virus capsid proteins localize to the surface of lipid droplets (42), cytoplasmic ZIKV C colocalizes with the lipid droplet marker Bodipy 493/503 (Fig. 2F). Together, these findings reveal multiple putative roles for C in remodeling host cell membranes and highlight the advantages of using both IP-MS and BioID for a more complete characterization of protein-protein interactions.
The other ZIKV proteins are targeted to a number of additional host cell systems and organelles. For example, consistent with a recent report (43), our Flag IP-MS links ZIKV NS5 to the STAT2 signaling protein (supplemental Table S1) as well as DDA1, DDB1 and VPRBP/DCAF1 (Fig. 3A), components of a host cell ubiquitin E3 ligase previously linked to several other viruses (44). IP-MS conducted on both DDB1 and VPRBP/DCAF1 in the presence of NS5 validated these interactions for the ZIKV NS5 protein (Fig. 3A, supplemental Table S4).
NS5 BioID also identified components of the Cajal Body (CB; Fig. 3B), a nuclear organelle with crucial roles in spliceosomal snRNP maturation (45). The influenza A virus nucleoprotein is also localized to CBs (46) but little is known regarding the role of this organelle in the Flavivirus life cycle. Consistent with our data, both NS5 (but no other ZIKV proteins; Fig. 3C, supplemental Fig. S3) and biotinylated proteins in NS5-expressing cells (Fig. 3D) colocalize with the CB marker coilin. BioID conducted on coilin in the presence and absence of NS5 (Fig. 3B, supplemental Table S5) validated this interaction. Notably, NS5 expression was also uniquely associated with significantly increased numbers of CB/cell, accompanied by a significant decrease in CB volume (Fig. 3E, supplemental Table S6). NS5 thus appears to disrupt CBs.
ZIKV infection has been linked to significant defects in centrosome function (54,55). Consistent with these observations, our data reveal ZIKV polypeptide interactions with multiple centrosome and cilia regulatory proteins. For example, NS3 (Fig. 3H) interacts with UNC119B, a myristoyl chaperone required for recruiting NPHP (nephronophthisis) proteins to cilia (56); CEP192, a major regulator of centrosome and spindle assembly; OFD1, a centrosome/centriolar satellite protein (57); and CEP85, a protein linked to centrosome biogenesis

FIG. 2. Individual ZIKV proteins interact with both shared and unique subsets of host proteins linked to a variety of biological functions.
A, ZIKV proteins interact extensively with the host endomembrane system. Bait proteins are represented by larger nodes, interactors represented by smaller, color-coded nodes, and grouped according to intracellular localization. Interactions detected by BioID are indicated with blue edges (dark blue detected with N-terminal tagged ZIKV bait protein, light blue detected by C-terminal tagged bait protein); interactions detected by IP-MS are indicated by red and orange edges. Prey proteins previously linked to Flavivirus function are encircled in solid gray, prey previously linked to other viruses are highlighted with a broken gray circle. B, C interacts with nucleolar proteins, nuclear and ER membrane proteins, and polypeptide components of lipid droplets. C, C colocalizes with the nucleolar marker fibrillarin. Immunofluorescence microscopy of GFP-tagged C protein and endogenous fibrillarin. All scale bars 10 m. D, Nuclear C protein occupies DAPI-poor regions, and colocalizes with the nuclear membrane protein lamin B1. E, Nuclear C colocalizes with the ER markers P4HB, CANX and FAM134B. F, Cytoplasmic C is localized to the ER and lipid droplets.  (58,59). The NS3-OFD1 interaction was validated via BioID (Fig. 3H, supplemental Table S8) and the NS3-CEP85 interaction was confirmed by co-IP western (Fig. 3I). Consistent with these data, NS3 and its biotinylated proximity partners colocalize with pericentrin (PCNT), a component of the pericentriolar material (Fig. 3I). NS3 also localized to the abscission sites of dividing cells, suggesting an interaction with shared components of the centrosome and midbody (Fig. 3J). Although no gross defects were observed for primary cilia or centrosomes in ZIKV-infected Vero cells (data not shown), structured illumination microscopy revealed large-scale disruption of centriolar satellite clusters (Fig. 3K), consistent with disruption of centrosome function by ZIKV.
The peroxisome has important antiviral functions, and proteins from multiple viral classes localize to this organelle (60). Lipids synthesized exclusively in peroxisomes also appear to be required for viral replication: e.g. decreased peroxisomespecific ether lipid synthesis impairs influenza virus replication (61). Consistent with these studies, our data revealed extensive ZIKV protein interactions with peroxisome-associated polypeptides (Fig. 4A). For example, several ZIKV proteins established high confidence interactions with the peroxisomal membrane protein ABCD3/PMP70; ACBD5, VAPA and VAPB, a set of ER-peroxisome membrane contact site proteins required for lipid transfer (62,63); and the peroxisome biogenesis protein PEX11B. ZIKV NS2A uniquely interacts with both components of the peroxisome protein import complex, PEX3 and PEX19 (Fig. 4A). Consistent with these data, GFP-NS2A (but no other ZIKV proteins) localizes to peroxisomes (Fig. 4B, supplemental Fig. S4). To begin to explore the role of the peroxisome in ZIKV biology, GFP-NS2A was expressed in a wild type (wt) human skin fibroblast cell line and fibroblast lines derived from Zellweger syndrome patients deficient for PEX19 (PBD399 (64)) or PEX3 (PBD400 (65)) and thus lacking functional peroxisomes. As expected, GFP-NS2A colocalized with peroxisomes in wt fibroblasts (Fig. 4C). Notably, however, it was mislocalized to the cytoplasm in both PBD399 and PBD400 cells, suggesting that NS2A is specifically targeted to peroxisomes (Fig. 4C).
ABCD3/PMP70 is significantly decreased in response to ZIKV infection (as compared with control proteins GAPDH and ATP5A; Fig. 4D). Consistent with this observation, significant decreases in peroxisome density (peroxisomes/unit cell volume) were also observed in cells expressing GFP-NS2A (Fig.  4F, supplemental Table S9) and in ZIKV-infected cells (Fig. 4E, 4G, supplemental Table S10). Finally, consistent with an important role for the peroxisome in ZIKV biology, ZIKV yields were significantly lower in Zellweger fibroblasts than in WT cells (Fig. 4H). DISCUSSION Together, our global ZIKV interactome identifies Ͼ1200 putative host interacting protein partners, physically linking the ten individual viral proteins to specific human polypeptides, and revealing a broad array of putative accessory functions for both the structural and nonstructural Zika polypeptides in the virus life cycle (Fig. 5; for detailed enrichment analysis, see supplemental Table S2). For example, the NS3 protease has been linked to ZIKV polyprotein cleavage and is essential for virus replication (66). Our data suggest that coexpression of NS3 and its cofactor NS2B can also impact lysosome function. Interactions between NS3 and three components of the HOPS/CORVET complex suggest that the decreased lysosome volume observed specifically in response to the coexpression of NS2B and NS3 could be because of a defect in late endosome-lysosome fusion, potentially supporting virus-containing endosome escape from lysosomal degradation and clearance. Our analysis also revealed a clear association of NS4A, NS4B and PreM with the 26S proteasome. Although these interactions remain to be investigated in an infectious context, the data suggest a putative role for this set of ZIKV proteins in e.g. blocking viral protein degradation, or the targeted degradation of host innate immune response proteins. As expected, C, PreM and Env were localized to the ER, where virus factories are created and viral particles assembled (67). We also detected NS2A, NS2B, NS4A, and NS4B at this organelle. These four nonstructural proteins are also likely to be involved in the formation of virus-induced membrane structures at the ER, by homology with West Nile Virus proteins (68). NS2A and NS2B are also likely to play roles in vRNA replication, as reported for the homologous proteins in the Japanese encephalitis virus (69).
The BioID approach revealed a number of associations between specific Zika polypeptides and host-cell lipid metab- olism machinery, which is significantly reprogrammed upon Flaviviral infection (reviewed in (70)). Like the DENV capsid polypeptide (42), the ZIKV Capsid protein was also detected at lipid droplets, where it could highjack host-cell lipid metabolism for viral replication. We also found that ZIKV NS2A is uniquely targeted to peroxisomes, an organelle that plays a critical role in lipid processing, and which acts as an important innate immune response signaling platform (60,61). ZIKV infection or expression of NS2A alone decreased peroxisome density (organelle number/cell volume) in host cells, and ZIKV replication was significantly impaired in peroxisome-deficient cells. Together, these data suggest that NS2A targets peroxisomes to remodel host-cell lipid populations and/or to effect viral escape from the host innate immune response (as previously shown for DENV and WNV (71)).
We also detected many interactors involved in DNA and RNA metabolism. For example, ZIKV NS5 interacts with coilin, a major component of the Cajal Body (CB), a nuclear membranefree organelle required for spliceosome maturation. We speculate that the ability of NS5 to fragment CBs could impair the processing of host-cell mRNAs, and thereby increase the availability of host ribosomes for viral RNA translation, and/or ZIKV NS5 could (1) The NS3-NS2B protease interacts with HOPS/CORVET complex components, accompanied by an apparent inhibition of late endosome-lysosome fusion. NS3-NS2B function could thus be important to protect endosomal ZIKV from lysosome-mediated degradation; (2) PreM, NS4A and NS4B are associated with the 26S proteasome. This host cell complex is commonly highjacked by viruses for e.g. innate immune system evasion; (3) multiple ZIKV proteins localize to the ER, consistent with roles in viral factory formation and viral replication; (4) like other Flaviviruses, the ZIKV capsid protein is addressed to lipid droplets, where it could e.g. highjack host-cell lipid metabolism for viral replication; (5) NS2A is uniquely localized to peroxisomes, and negatively impacts peroxisome density; (6) NS5 is localized to and appears to disrupt Cajal bodies, possibly to impair host cell splicing or usurp CB activities for viral RNA maturation; (7) ZIKV capsid protein-induced inward nuclear/ER membrane budding is likely important for the creation of viral factories; (8) Deficiencies in centrosomal proteins, including centriolar satellite components, are responsible for congenital microcephalies. ZIKV infection results in the dispersion of centriolar satellites, possibly explaining the surge in microcephaly cases associated with infected human populations. be important for usurping CB activities required for vRNA maturation.
Notably, expression of the C protein alone induced the inward budding of nuclear/ER membranes into the nucleoplasm. The ability of C to interact with both host endomembranes and RNAs has been well documented, properties that are critical for vRNA encapsidation in Flavivirus virions (reviewed in (72)). Its positively charged carboxy-terminal domain makes it very efficient at interacting with RNAs, and we hypothesize that it could sequester host mRNAs in the nucleus to promote vRNA translation. Consistent with a previous report (73), a subpopulation of the ZIKV capsid protein is localized to the nucleolus. Capsid nucleolar localization has been documented for other Flaviviruses (74), but the function of the viral protein in this host organelle remains unclear. Addressing the capsid protein to nucleoli could be important for maintaining host ribosomal RNA production during infection, and thus preserving functional host cell translation machinery. Finally, we report that ZIKV infection is accompanied by centriolar satellite dissolution. Deficiencies in several different centrosomal proteins, including satellite components (75), have been linked to congenital microcephalies (76). Together with the specific interactions we report between ZIKV NS3 and multiple centrosome components (e.g. CEP192, CEP85, OFD1), this observation could help to explain the surge in microcephaly cases observed in infected human populations. The relationship between NS3 and centriolar satellites will be important to investigate.
It is important to note that the current data set: (1) does not include interactions between viral proteins; (2) describes interactions that occur under relatively high ZIKV protein expression levels, and; (3) likely lacks some virus-host protein interactions that can only be observed in the context of a viral infection. Nevertheless, our data highlight the astounding versatility of ZIKV polypeptides to simultaneously target numerous host cell organelles and provide important molecular insights into how ZIKV proteins could usurp and manipulate the host-cell machinery to the benefit of the virus. As we demonstrate here, this data set represents a rich resource that can be mined to better understand how ZIKV exploits and/or disrupts specific host cell systems and organelles, and thus paves the way for a better understanding of ZIKV biology and the identification of new antiviral targets.