An Orthogonal Proteomic Survey Uncovers Novel Zika Virus Host Factors Pietro Scaturro1,2*, Alexey Stukalov1,2, Darya A

Home , ATP6V0C, BAZ1B, C7orf50, CCDC47, DDX24, DNAJC13

Letter https://doi.org/10.1038/s41586-018-0484-5

An orthogonal proteomic survey uncovers novel Zika virus host factors Pietro Scaturro1,2*, Alexey Stukalov1,2, Darya A. Haas1, Mirko Cortese3, Kalina Draganova4,5, Anna Płaszczyca3, Ralf Bartenschlager3,6, Magdalena Götz4,5,7 & Andreas Pichlmair1,2,8*

Zika virus (ZIKV) has recently emerged as a global health concern mutated in Alazami syndrome)4, LYAR (important for maintenance of owing to its widespread diffusion and its association with severe embryonic stem cell identity)5 and NGDN (a neuronal development neurological symptoms and microcephaly in newborns1. However, factor)6 (Extended Data Fig. 3a). NS4B has recently been implicated the molecular mechanisms that are responsible for the pathogenicity in the inhibition of neuronal development7. Compared to HCV-NS4B, of ZIKV remain largely unknown. Here we use human neural ZIKV-NS4B showed specific enrichment in cellular proteins associated progenitor cells and the neuronal cell line SK-N-BE2 in an integrated with the whole spectrum of ZIKV-associated pathogenesis, ranging proteomics approach to characterize the cellular responses to viral from neurodegenerative disorders and retinal degeneration (CLN6 infection at the proteome and phosphoproteome level, and use and BSG)8,9 to regulators of neuronal differentiation (CEND1 and affinity proteomics to identify cellular targets of ZIKV proteins. RBFOX2)10,11 and axonal dysfunction (CHP1 and TMEM41b)12 (Fig. 1 Using this approach, we identify 386 ZIKV-interacting proteins, and Extended Data Figs. 2–4a). Co-immunoprecipitation followed by ZIKV-specific and pan-flaviviral activities as well as host factors western blotting of transduced or ZIKV-infected cells verified that these with known functions in neuronal development, retinal defects and proteins specifically associate with arthropod-borne NS4Bs (such as infertility. Moreover, our analysis identified 1,216 phosphorylation those of dengue virus and ZIKV) and not NS4B of other Flaviviridae, sites that are specifically up- or downregulated after ZIKV infection, such as HCV; with a subset of proteins specifically binding to ZIKV- indicating profound modulation of fundamental signalling NS4B only (TMEM41b, CEND1 and CLN6; Extended Data Fig. 4b, c). pathways such as AKT, MAPK–ERK and ATM–ATR and thereby ZIKV-NS4B precipitated particularly well with ceroid-lipofuscinosis providing mechanistic insights into the proliferation arrest elicited neuronal protein 6 (CLN6), which is associated with a lysosomal by ZIKV infection. Functionally, our integrative study identifies storage disease that causes neurodegenerative late-infantile disorders ZIKV host-dependency factors and provides a comprehensive as well as retinal defects8. In human neural progenitor cells (hNPCs), framework for a system-level understanding of ZIKV-induced CLN6 redistributed to sites enriched in NS4B (Extended Data Fig. 3d). perturbations at the levels of proteins and cellular pathways. AP–LC–MS/MS experiments using CLN6 as bait identified common ZIKV, a flavivirus which is related to dengue virus, West Nile virus binding partners between NS4B and CLN6 (Extended Data Fig. 3e). and hepatitis C virus (HCV), has a single-stranded RNA genome of pos- Furthermore, CLN6 associated specifically with mTOR and TELO2, itive polarity, encoding a polyprotein that is co- and post-translationally important regulators of signalling pathways that are disrupted by processed into three structural (capsid (C), precursor membrane (prM) ZIKV7,13. Taken together, the ZIKV interactome revealed a number of and envelope (E)) and seven non-structural proteins (NS1, NS2A, new Flavivirus host-binding partners, including cellular proteins and NS2B, NS3, NS4A, NS4B and NS5)2. To comprehensively understand signalling pathway components involved in neuronal development. how ZIKV affects neuronal cells, we performed an unbiased proteomic In patients, during early stages of development, ZIKV predominantly survey to identify cellular proteins and associated complexes that inter- infects NPCs, causing microcephaly and other neurodevelopmental act with each of the ten ZIKV proteins expressed in human SK-N-BE2 injuries14. To gain insights into ZIKV-induced changes, we used a neuroblastoma cells (Extended Data Fig. 1a). Except for NS2A, all human induced pluripotent stem cell-derived neuronal differentiation ZIKV proteins were correctly expressed and processed (Extended model15. Analysis of the global proteomic changes that occurred during Data Fig. 1b, c). Affinity purification coupled with liquid chromatog- differentiation of hNPCs into neurons revealed significant upregula- raphy and tandem mass spectrometry (AP–LC–MS/MS), followed by tion of neuronal differentiation markers such as βIII-tubulin (TUBB3), Bayesian statistical modelling, identified 386 proteins specifically asso- microtubule-associated proteins 2 and 6 (MAP2 and MAP6), neuronal ciating with ZIKV proteins, resulting in 484 high-confident interactions cell adhesion molecule 1 (NCAM1), doublecortin (DCX) and ELAV- (Fig. 1, Extended Data Fig. 2 and Supplementary Table 1). We identi- like RNA binding protein 3 (ELAVL3) (Fig. 2a, Extended Data Fig. 5 fied previously reported proteins that showed bona fide interactions and Supplementary Table 2). Notably, ectopic expression of ZIKV- with flavivirus proteins, including subunits of the ATPase (ATP1A1, NS4B during differentiation specifically downregulated the expression ATP1A2, ATP1A3, ATP1B and ATP6V1H), voltage-dependent of a subset of proteins involved in neuronal differentiation (for example, anion-selective channel proteins (VDAC1, VDAC2 and VDAC3) as MAP2, MAP6, DPYSL3, DPYSL5 and CNTN2) as well as proteins asso- well as components of the cytochrome c oxidase complex (COX15, ciated with neurological diseases (for example, DOK3 and SUMO2), MT-CO2 and NDUFA4), confirming these processes as important suggesting disruption of specific developmental programs (Fig. 2b, c, targets of diverse flaviviruses3 (Fig. 1 and Extended Data Fig. 2). Extended Data Fig. 5c–e and Supplementary Tables 2, 3). Similar effects Notably, this analysis uncovered proteins linked to neurological could be seen in proteomic analysis of hNPCs infected with ZIKV in the diseases or development, particularly among the proteins that specifically presence or absence of differentiation stimuli (Extended Data Fig. 6a). interacted with capsid and NS4B (Fig. 1). For instance, the capsid- ZIKV infection led to a robust upregulation of type-I interferon- interacting proteins included LARP7 (involved in telomere stability and stimulated genes (for example, STAT1, MX1, OAS3 and IFIT1)

1Max-Planck Institute of Biochemistry, Innate Immunity Laboratory, Martinsried, Germany. 2Technical University of Munich, School of Medicine, Institute of Virology, Munich, Germany. 3Department of Infectious Diseases, Molecular Virology, University of Heidelberg, Heidelberg, Germany. 4Institute of Stem Cell Research, Helmholtz Center Munich, Neuherberg, Germany. 5Physiological Genomics, Biomedical Center, Ludwig-Maximilians-Universitaet, Munich, Germany. 6German Center for Infection Research (DZIF), Heidelberg partner site, Heidelberg, Germany. 7Synergy, Excellence Cluster for Systems Neurology, Biomedical Center, Ludwig-Maximilians-Universitaet, Munich, Germany. 8German Center for Infection Research (DZIF), Munich partner site, Munich, Germany. *e-mail: [email protected]; [email protected]

PARP2 PDAP1 NS1 HMBOX1

ESYT1 NS4A CEP250 E CHD8 ADAMTS12 NS5 Mitochondrial CLGN DHCR7 RPN2 ribosomal proteins ATM FAR1 NOA1 SEP15 TECR SC5D PGRMC1 MRPL42 MRPS23 MRPL2 CEP63 DMD TMEM33 MRPS22 MRPL35 PDS5A KRTCAP2 MRPL18 MRPL41 ALG8 GPRASP1 MRPS35 MRPS10 LPGAT1 MRPL21 MRPL13 MRPL24 TMX3 DDOST AAR2 Viral bait MRPS16 SCD5 MRPS34 MRPS5 MRPS25 MRPL17 PDE2A MRPL22 CCDC47 EI24 Q59GX9 MRPS7 MRPL45 ELOVL1 MRPL16 MRPL47 ERLIN1 MRPL1 CHP1 MRPS33 MRPL9 FADS2 XXYLT1 CLN6 NAP1L1 MRPS21 MRPS6 MRPS31 ICMT Prey protein KIAA0020 MRPL55 ERLIN2 MRPS11 VGF prM CEND1 MRPL3 MRPL19 AGPAT6 MRPS2 MRPS30 MRPL23 RPN1 B4DLN1 MRPL54 DERL2 Neurological MRPL10 TMEM41B B2RD90 MRPL39 MRPS14 MSH6 SV2A MRPL38 MRPL11 STT3A MRPS15 DAP3 SCD BSG disorders/ RCN2 MRPS9 MRPS27 ATP2A2 HSD17B12 MRPL37 MRPL28 neurogenesis AIFM1 MRPS26 CANX SPTLC1 HAX1 STOML2 ZC3HAV1 CALM1 AUP1 HADHB HADHA SSR3 DNAJC13 ICT1 SLC25A3 Localization F5H0B0 PEX19 LBR NS2B3 SEC61B LARP7 PTCD3 PHB GTPBP4 RAB18 EBP C7orf50 NOL6 SLC25A11 GNAI1 Cytosol DDX10 CMSS1 TUBB4A HACD3 SEC61A1 SRPK1 DIMT1 GRWD1 ZNF622 PHB2 VDAC1 DDX3Y SLC25A1 TUBA1A Plasma PPM1G DHX57 PGRMC2 VDAC2 ATP2B1 ATP1A3 SLC39A9 DNAJA1 MTDH SURF4 membrane FXYD6 NGDN BRIX1 VPS16 NAT10 CAMK2G IPO4 TSPO PPAN CEBPZ GNB2L1 HRAS NUP93 VDAC3 FAM134C SLC25A13 Nucleus MYBBP1A TIMM50 ABCB6 RFC1 DDX18 H1FX SQSTM1 NS4B RSL1D1 Capsid MTCH2 ZFR NUP205 SLC10A4 HLA-A DDX24 SET H2AFY SLC25A22 TM9SF2 Mitochondrion DKC1 POP1 TIMMDC1 TOP1 RBFOX2 CCDC137 RPSA SLC25A6 SLC38A1 LLPH RBM34 PRKDC ATP1B1 Endoplasmic NUP107 CHCHD3 SLC7A5 RRP12 RBM28 YIPF5 reticulum MAGED1 SMN1 SMPD4 SCAMP3 NIFK DDX21 KPNB1 IPO7 Golgi KRR1 GTF3C2 NUP85 LPHN2 CHPT1 BMS1 GNL3 EXOSC4 SLC3A2 YIF1A CDS2 apparatus ZCCHC8 VAPB MPDU1 DDB1 EXOSC1 FAU BOP1 NOP56 IPO9 EXOSC10 TIMM23 DHCR24 MT-CO2 EBNA1BP2 FTSJ3 EXOSC9 DDX20 TMEM165 Cytoskeleton HP1BP3 KRI1 XPOT LPCAT1 PRAF2 LYAR NUP160 TMCO1 RAB39A Mixed RPLP0 RPS15A SDAD1 RPL21 RPL5 SFXN3 NOL10 MRTO4 RPL34 RPS7 RPL11 NUP133 USMG5 ATP1A1 localization LARP1 NUP43 RPS3 RPL10 RPS3A RPL3 IMMT NCL RPS26 COPG2 ZNF512 SRRT RPL35A RPL23A SCAMP5 Mixed/unknown RPS8 RPL35 RPS12 NAP1L4 RPL29 RPL30 OXA1L NAP1L1 RPL36 TARBP2 localization RPL32 RPL7A NUP98 RPLP1 RPL9 SPNS1 SACM1L ATP2A1 GNL2 RPS16 RPS10 RPL22 TIMM17B RPLP2 RPL26 RPL24 RPL6 RPS21 SLC1A5 ZNF629 RPL23 COX15 RPL4 RPS2 SRSF5 RPL18A RPS6 RPL7 RPS19 DICER1 RPL27A RPS4X RPL8 GEMIN6 TMEM126A HDLBP RRP1B SLC25A12 TOMM22 HM13 SLC16A1 RPL13A RPL17 RPS5 EIF2B3 RPL31 RPS25 RPS24 RPS15 RPS17 PRSS12 RPL26L1 RPL18 RPS13 RPL12 RPL13 RPS20 RPL38 GEMIN4 SLC2A1 NDC1 DNAJC9 DDX50 RPS18 SERBP1 RPL28 RPL19 SURF6 TM9SF3 AAAS CKAP4 DNAJA2 RPS29 RPL27 RPL36AL RPL14 RPS4Y1 RPS28 RPL15 CDIPT RPS9 SMC2 ERAL1 TUBB3 TRIM26 SAAL1 B4DDU2 USP9X SMC4 MLLT11 Ribosomal proteins −5 Fig. 1 | A ZIKV proteins–host proteins interaction network in log2(fold change) ≥ 2.5; unadjusted one-sided P ≤ 5 × 10 . Interactions neuronal cells. Network representation of the high-confidence between ZIKV and host proteins are indicated by black lines. Published ZIKV–host interactome in SK-N-BE2 cells. There are eight ZIKV bait physical associations of the host proteins are indicated by dotted grey proteins (red squares) and 386 interacting host proteins (ellipses). lines. Subcellular localization (colour of nodes) of host proteins is taken n = 4 independent experiments, Bayesian statistical modelling. All from Gene Ontology. Proteins with known functions in neurogenesis or baits: log2(fold change) ≥ 2.5; unadjusted one-sided P ≤ 0.05; capsid: neurological diseases are circled in pink. alongside specific downregulation of neuronal factors under differen- mTOR targets (for example, ANKRD17, LARP1, PATL1 and EEF2K), tiation conditions (Extended Data Fig. 6b and Supplementary Table 4). the central kinase S6K and its main effector protein S6. Consistently, These experiments unveil the distinct effects of ZIKV and ZIKV-NS4B we found increased phosphorylation of proteins, such as the negative on the proteome of differentiating hNPCs, revealing specific targets that regulator of autophagy DAP, at residues known to inhibit its suppressor are differentially regulated during neurogenesis. function and consistent with the notion that autophagy is induced after Post-translational modifications of proteins, especially phosphoryl- ZIKV infection7,19 (Fig. 2e and Extended Data Fig. 7f). Notably, ZIKV ation, have crucial roles in controlling the complex reorganization of infection additionally downregulated the MAPK–ERK-signalling path- signal-transduction pathways upon virus infections16,17. We therefore way, since A-RAF and ERK2 as well as several substrates of the down- used LC–MS/MS-based time-resolved phosphoproteomics18 to inves- stream ERK1/2 map kinases (EIF4EBP1, BAZ1B and TWIST) were tigate cellular pathway deregulation in ZIKV-infected SK-N-BE2 cells. significantly dephosphorylated upon ZIKV infection (Fig. 2e, Extended This analysis identified 14,222 unique phosphorylation sites (locali- Data Fig. 8c, d and Supplementary Tables 5, 8). These data enable us to zation probability ≥ 0.75; Extended Data Fig. 7a–d), 1,216 of which quantitatively map individual phosphorylated residues onto signalling were modulated upon ZIKV infection (Fig. 2d and Supplementary pathways on a global scale, identifying complex interactions between Table 5). Analysis of enriched biological functions showed that path- different cellular pathways (Extended Data Fig. 7e). For instance, both ways involved in cellular assembly and organization, cell-cycle regu- ERK1/2 and mTOR pathways converge on S6K and modulate its phos- lation, nervous system development and neurological diseases were phorylation status20 (Fig. 2e), suggesting multiple possible mechanisms significantly modulated by ZIKV infection (Extended Data Fig. 8a that could contribute to the observed ZIKV-induced upregulation of and Supplementary Table 6). Notably, this approach allowed us to autophagy, decrease in hNPC proliferation and impairment of neu- map specific phosphorylation events along entire cellular pathways, rogenesis. Notably, upregulation of the ATM (ataxia-telangiectasia highlighting modulation of key processes such as ATM, AKT–mTOR mutated)/ATR (ATM- and Rad3-related) DNA-damage pathway, as and ERK–MAPK signalling cascades (Extended Data Figs. 7e, 8b and inferred from significantly increased phosphorylation of several sub- Supplementary Tables 7, 8). We identified strong downregulation of the strates of ATR (NPM1, CGGBP and GINS2), DNAPK (XRCC4) and AKT–mTOR signalling pathway, as indicated by dephosphorylation of downstream effector proteins (H2AFX), as well as proteins involved in known AKT1 substrates (for example, DNMT1, TBC1D4 and LARP6), cell-cycle regulation and DNA-damage checkpoints (TOP2B, CDK1)

a b Cellular function c ZIKV-NS4B/ HCV-NS4B 10 Upregulated Development/ Cytoskeleton Mock differentiation regulation Undiff. Diff. TUBB3 Cellular diff./undiff. Downregulated Neurological Wnt/Notch/ function

) PTN 2 DCX 8 disease hedgehog DOK3

e SMYD4 5 NCAM1 r

e CNTN2 MAP2 4B

S MAP6 Diff D ELAVL3 Differentiated 4 MYBPC2 MAP6 KIF5C 0 ZIKV-NS4BZIKV-NS4B

ZIKV-NS4B vs HCV-NS4B ZIKV-N DNAJB6 ENO2 MYH2 Differentiated DPYSL3 –5 –88 –4–4 4 8 TAGLN3 CNTNCCNTN22 ZIKV-NS4BZIKV-NS4B vs HCV-NS4BHCV KMT2C Median normalized LFQ (log UndifferentiatedUndifferentiate MAP2 DPYSL5DPYSL5 SUMO2 –4–4 DPYSL5 –10 MMAP2AP2 ITGB6 DPYSL3 –10 –5 0 5 10 MMAP6AP6 APBA1 Undifferentiated CDC42EP5 HCV-NS4B –8–8 DNM2 Median normalized LFQ (log2)

–3 0 3 –log10(P) de 25 50 log (fold change in LFQ) 75 Neuronal differentiation 2 100 Cell survival/proliferation S41 S209 S451 T207 RAS S447 S163 PPP2R5A S143 S167 LARP6 EMSY S154 244 S170 S257 S341 Upregulated ARAF DNMT1 127 MARCKS PKC T642 71 AKT TBC1D4 Actin cytoskeleton 24 h 48 h 72 h regulation MEK S74 S184 S72 S179 69 EEF2K PATL1 Y187 Y182 ERK2 S51 253 T185 T180 DAP mTOR Number of phosphosites MK2 p38 EIF4EBP1 S82 BAZ1B

Downregulated LARP1 S2045 S65 S158 ANKRD17 Autophagy HSP27 TWIST1 S2047 ATF7 T53 S766 764 S774 T51 S68 S6K T444 S236 S447 S6 S235 Axon S140 guidance DNA damage S517 S360 T137 Novel site Inferred upregulation S358 H2AX Known site DPYSL2 SMC1A Inferred DNA-PK Regulatory site downregulation ATR XRCC4 4 6 4 2 1.4 –1. –2 –4 –5 ATM S328 S94 GINS2 CGGBP S327 T180 EIF4EBP1 NPM1 S164 log2(fold change) S182 p53 S125

Fig. 2 | Impact of ZIKV and NS4B on hNPCs proteome upon transduction. Ratios of protein abundance in naive hNPCs (differentiated/ differentiation and global phosphoproteomic profiling of ZIKV- undifferentiated), differentiated hNPCs (ZIKV-NS4B/HCV-NS4B) or infected cells. a, Comparative proteomic analysis of hNPCs, undifferentiated hNPCs (ZIKV-NS4B/HCV-NS4B). Coloured squares undifferentiated or differentiated for five days in the presence of indicate functional annotations as in b. Up- or downregulated proteins ROCK inhibitor. LFQ, label-free quantification. Significantly up- or are displayed in shades of blue or green, respectively. Black circles downregulated proteins are shown in orange and blue, respectively. n = 4 indicate significant changes; sizes correspond to indicated P values. independent experiments, Bayesian statistical modelling. Unadjusted d, Total number of up- or downregulated class I phosphorylation sites two-sided P ≤ 0.01; |log2(fold change)| ≥ 1. Known markers of neuronal significantly regulated by ZIKV infection 24, 48 and 72 h post-infection. differentiation are highlighted. b, Impact of ZIKV-NS4B on differentiated n = 4, unadjusted two-sided P ≤ 0.01; |log2(fold change)| ≥ 1. e, Selected and undifferentiated hNPCs. The scatter plot displays ZIKV-NS4B cellular pathways and phosphorylation sites that are modified by ZIKV. specific changes (compared to HCV-NS4B). Proteins that were specifically Colours of protein nodes indicate measured (log2(fold change) of strongest modulated by ZIKV-NS4B are shown as large circles or squares. Colour phosphorylation site change) or inferred regulation of central proteins codes refer to functional annotations; squares indicate proteins that and substrates of canonical pathways. Individual phosphorylation sites are were specifically modulated by ZIKV-NS4B that are significantly up- or shown next to the corresponding molecules (colours indicate the highest downregulated during differentiation. n = 4 independent experiments, log2(fold change) over time). Circles, newly identified phosphorylation Bayesian statistical modelling. Unadjusted two-sided P ≤ 0.01; sites; squares, known phosphorylation sites; framed squares, regulatory |log2(fold change)| ≥ 1. c, Protein levels of cellular factors that were sites. Arrows and blunt-ended lines indicate positive or negative regulation differentially modulated upon hNPC differentiation and ZIKV-NS4B- of downstream proteins, respectively. could be observed (Supplementary Tables 5, 8). This analysis highlights identified 17 host factors for which silencing significantly reduced previously unreported ZIKV targets with known functions in neuro- ZIKV replication compared to non-targeting controls (Fig. 3a). In genesis. Among these targets, we note dephosphorylation of proteins, addition to confirming the essential role for recently described cellular such as p38 MAPK, MARCKS (one of the main PKC substrates) and factors in ZIKV replication, such as STT3A, MGGT1 and MSI124–26, we DPYSL2, activation of which positively modulates neurite outgrowth identified 14 additional host proteins that are required for ZIKV rep- and brain development21–23 (Fig. 2e and Extended Data Fig. 8e). lication. Among these were proteins, the cellular abundance of which To assess the functional relevance of the newly identified host factors, was strongly decreased upon NS4B expression and/or ZIKV infec- we selected cellular targets regulated either at the proteome (n = 26), tion in differentiating hNPCs (DOK3, XIRP2, YIPF4 and LMOD3), phosphoproteome (n = 8) or interactome (n = 21) level by ZIKV, and proteins that changed phosphorylation status upon ZIKV infection evaluated the consequences of virus infection on gene silencing (Fig. 3a, (LMO7, FXR1 and LARP7) and ZIKV-interacting proteins (LARP7 Extended Data Fig. 9a–c and Supplementary Table 9). This approach and LYAR, interacting with capsid, and BSG and CLN6, interacting

a 1 2 D C 3 X2 1 7 AR YAB AP1L1 P1A1 ATP6V0PRAFCDOK32 LARPSLC16A17 YIPFYIF14 LYA MSI1RRAGMMGT1BSGGCOM1CLN6LMO7XIRPFXR12 ICMTPOSTMYBPC2N SRLCR LMODSTT33 PRKDCA HAX1CALMLKIF21A5PDE1ZC3HTMEM41AVDPYSL2AFB LARP1SERPINI1UBA7NGDNCHP1AKTIEXOP AT DCDCEND1FAM64AMYH1S1PR2VGFNT RBFOIMPDH1MKI6PTMSSMN1DPYSL3SUMOAPOBEC2 CELSR2MARCKS

Proteome + + +++++ + +++++++++++ +++ + +++++ +++ + Phosphoproteome + + + +++ + + + ++ + + + Interactome 4B CC4B 4B C 4B4B4B 4B 4B 4B 4B CC4B C C 4B 4B 4B C 4B

Cell

viab.

Cell viability (%) r

eplicates

hRNA e 1 75% 2 50 66 100 150 200

Virus titr 3

Cell

Virus titre (%) viab.

eplicates

e 1 hRNA 10 25 100 400 1,000 2

Virus titr 3

bc200 Merge + DAPI CLN6–Flag NS4B d Merge + DAPI CLN6–Flag dsRNA

) CLN6:NS4B CLN6:dsRNA NT

150 –2 –3 –3 –3 of –6 –3 1.0 –4 Mock 1.0 × 10 Mock

(% 100 .8 × 10 0.8 0.8

e 8

tr P = 6.3 s R P = 1.3 × 10 s R ti P = 2.3 ×= 107.2 ×P 10 = P = 1.0 ×10 0.6 P = 2.2 × 10 P 0.6

s 50 on’ on’ s s ru 0.4 0.4 Vi ear H/PF/2013 H/PF/2013

0 P 0.2 Pear 0.2 3 b C NT

BS G 0 MSI1 0 CLN6 CHP1

CEND1 3 LMOD 3

66 R766 RBFOX 2 1 01 7 0 ATP6V0 2 MR766 TMEM41 / R M R766 F M M /P H H/PF/2

e f ZIKV

2B Merge + DAPI TMEM41b–Flag NS4B 2A

rM E

S C

S S

NS5

TMEM41b:NS4B NS3

NS4B NS4A

N N Interactors of host targets n = 26 1.0 Host targets

Mock mTOR Proteome 0.8 n = 386 n = 608 TELO2 EEF2K Interaction

s R 0.6 mTOR targets Measured in this study NCL CLN6 S6K

son’ 0.4 Previously published n = 11 H/PF/2013 0.2 n = 18 n = 35 Pear LMOD3 TUBB3 LARP7 0 YIF1A Phosphoproteome 13 6 n = 1,216 76 LMO7 sites MR766 20 MAP2 MARK n = 844 F/ MR 1/2/3 proteins /P n = 50 H Fig. 3 | shRNA screen identifies novel ZIKV host-dependency factors. mock-infected (mock) or infected with ZIKV H/PF/2013 or MR766 a, SK-N-BE2 cells were transduced with individual shRNAs that targeted (MOI = 5) for 24 h. Fixed cells were immunostained for dsRNA, ZIKV- genes selected from each of the proteomic surveys as indicated, and ZIKV NS4B and Flag-tagged proteins. Scale bars, 10 μm. n = 2 independent replication was measured by plaque-forming unit (PFU) assay. Results experiments. Representative images are shown. Graphs on the right display from each biological repetition and shRNA are presented as a heat map. Pearson’s correlation coefficients (mean ± s.d.; each dot represents a single Non-targeting (NT) shRNA-normalized values of individual repetitions cell). f, Intersection of ZIKV interactome, proteome and phosphoproteome are shown in each row. Grey circles indicate non-targeting controls and datasets highlighting characteristic proteins perturbed at multiple levels. shRNAs targeting known ZIKV host-dependency factors. Red squares ZIKV infection disrupts phosphorylation (green) of its host targets (for indicate cytotoxic shRNAs (normalized median cell viability <75% of example, LARP7) and along signalling pathways (for example, EEF2K and non-targeting controls; top rows). Green circles indicate significant hits S6K via CLN6–mTOR–TELO2; or MAP2 via MARK1/2/3). Protein levels (≥50% decrease in viral replication with both shRNAs each or ≥75% with (blue) are also affected, carrying signatures of ZIKV-induced changes to any shRNA). b, Validation of silencing efficiency and gene knockdown vesicle transport (YIF1A) and neuronal differentiation pathways (MAP2). effects on ZIKV replication on selected host factors from a. n = 4 Solid arrows represent protein interactions measured in this study; grey independent experiments, normalized mean ± s.d.; two-sided unadjusted dotted arrows represent published interactions. Numbers correspond to Student’s t-test. c–e, Confocal microscopy of hNPCs transduced with proteins significantly changed in a given dataset or intersection of datasets. CLN6- or TMEM41b-expressing lentiviruses (MOI = 3) and with NS4B) (Fig. 3a). Validation experiments confirmed that knock- Our data provide valuable insights into how ZIKV exerts its action down of CEND1, CLN6, CHP1, LMOD3, TMEM41b or BSG resulted on candidate proteins that have previously been identified in genetic in inhibition of ZIKV replication (Fig. 3b and Extended Data Fig. 9d, e). or drug-based screens. Among the recently identified ZIKV inhibi- Moreover, infection of hNPCs with two different ZIKV isolates tors27 is digoxin, an inhibitor of the ATP1A1 Na+/K+ ATPase, which (H/PF/2013 and MR766) recruited CLN6 and TMEM41b to sites that is identified here to interact specifically with NS4B. Fingolimod were enriched in double-stranded (ds)RNA and NS4B (Fig. 3c–e). inhibits the sphingosine-1-phosphate receptor, which is persistently We combined quantitative information at the proteome, phosphop- hypophosphorylated upon ZIKV infection. Mycophenolic acid roteome and interactome levels with publicly available interaction data- impairs inosine-5-monophosphate dehydrogenase 1; the cellular bases in order to obtain a holistic perspective on the effect of ZIKV on abundance of this protein is markedly downregulated in ZIKV- neuronal cells (Extended Data Fig. 10 and Supplementary Table 10). infected cells. Although some of the ZIKV-binding partners that This orthogonal intersection can be used to infer hotspots that are have been identified here may associate with orthologues of closely targeted by ZIKV at multiple levels. For example, LARP7, which related flaviviruses (for example, dengue virus and West Nile virus), interacts specifically with capsid, is additionally hypophosphorylated, their role in ZIKV-related pathogenesis might in addition rely on the whereas LMO7 is simultaneously downregulated at the protein level unique ability of ZIKV to cross the placental barrier and reach the and hyperphosphorylated upon ZIKV infection (Fig. 3f). developing brain28.

Our study suggests that ZIKV evolved multiple mechanisms to 18. Humphrey, S. J., Azimifar, S. B. & Mann, M. High-throughput phosphoproteomics reveals in vivo insulin signaling dynamics. Nat. Biotechnol. usurp, exploit or perturb fundamental cellular processes, ultimately 33, 990–995 (2015). contributing to the broad spectrum of pathological abnormalities 19. Cao, B., Parnell, L. A., Diamond, M. S. & Mysorekar, I. U. Inhibition of autophagy observed in humans. In addition, this study provides a resource for limits vertical transmission of Zika virus in pregnant mice. J. Exp. Med. 214, 2303–2313 (2017). streamlining ZIKV research efforts and considering or excluding intra- 20. Ghouzzi, V. E. et al. ZIKA virus elicits P53 activation and genotoxic stress in cellular pathways for specific therapeutic interventions. human neural progenitors similar to mutations involved in severe forms of genetic microcephaly. Cell Death Dis. 7, e2440 (2016). 21. Morooka, T. & Nishida, E. Requirement of p38 mitogen-activated protein kinase Online content for neuronal diferentiation in PC12 cells. J. Biol. Chem. 273, 24285–24288 Any methods, additional references, Nature Research reporting summaries, source (1998). data, statements of data availability and associated accession codes are available at 22. Li, H., Chen, G., Zhou, B. & Duan, S. Actin flament assembly by myristoylated https://doi.org/10.1038/s41586-018-0484-5. alanine-rich C kinase substrate-phosphatidylinositol-4,5-diphosphate signaling is critical for dendrite branching. Mol. Biol. Cell 19, 4804–4813 (2008). 23. Yamashita, N. et al. Phosphorylation of CRMP2 (collapsin response mediator Received: 21 April 2017; Accepted: 20 July 2018; protein 2) is involved in proper dendritic feld organization. J. Neurosci. 32, Published online 3 September 2018. 1360–1365 (2012). 24. Marceau, C. D. et al. Genetic dissection of Flaviviridae host factors through 1. Miner, J. J. & Diamond, M. S. Zika virus pathogenesis and tissue tropism. genome-scale CRISPR screens. Nature 535, 159–163 (2016). Cell Host Microbe 21, 134–142 (2017). 25. Savidis, G. et al. Identifcation of Zika virus and dengue virus dependency 2. Neufeldt, C. J., Cortese, M., Acosta, E. G. & Bartenschlager, R. Rewiring cellular factors using functional genomics. Cell Rep. 16, 232–246 (2016). networks by members of the Flaviviridae family. Nat. Rev. Microbiol. 16, 125–142 26. Chavali, P. L. et al. Neurodevelopmental protein Musashi-1 interacts with the (2018). Zika genome and promotes viral replication. Science 357, 83–88 (2017). 3. Chatel-Chaix, L. et al. Dengue virus perturbs mitochondrial morphodynamics to 27. Barrows, N. J. et al. A screen of FDA-approved drugs for inhibitors of Zika virus dampen innate immune responses. Cell Host Microbe 20, 342–356 (2016). infection. Cell Host Microbe 20, 259–270 (2016). 4. Slomnicki, L. P. et al. Nucleolar enrichment of brain proteins with critical roles in 28. Shao, Q. et al. The African Zika virus MR-766 is more virulent and causes more human neurodevelopment. Mol. Cell Proteomics 15, 2055–2075 (2016). severe brain damage than current Asian lineage and dengue virus. Development 5. Li, H. et al. Ly-1 antibody reactive clone is an important nucleolar protein for 144, 4114–4124 (2017). control of self-renewal and diferentiation in embryonic stem cells. Stem Cells 27, 1244–1254 (2009). Acknowledgements We thank D. Mauceri, F. Sacco and L. Chatel-Chaix for 6. Jung, M. Y., Lorenz, L. & Richter, J. D. Translational control by neuroguidin, a discussions and R. Hornberger, I. Paron, K. Mayr and G. Sowa for technical eukaryotic initiation factor 4E and CPEB binding protein. Mol. Cell. Biol. 26, assistance. The work in the authors’ laboratory was funded by an ERC starting 4277–4287 (2006). grant (StG 311339, iVIP), the Max-Planck free-floater program, the German 7. Liang, Q. et al. Zika virus NS4A and NS4B proteins deregulate Akt–mTOR Research Foundation (PI1084/2, PI1084/3, PI1084/4, TRR 237 and TRR179) signaling in human fetal neural stem cells to inhibit neurogenesis and induce and the Federal Ministry for Education and Research (ERA-Net grant ERASe) autophagy. Cell Stem Cell 19, 663–671 (2016). all to A.Pi. R.B. was supported by the Deutsche Forschungsgemeinschaft 8. Radke, J., Stenzel, W. & Goebel, H. H. Human NCL neuropathology. Biochim. (BA1505/8-1). A. Pł. was supported by the Horizon 2020: Marie Skłodowska- Biophys. Acta 1852, 2262–2266 (2015). Curie ETN “ANTIVIRALS” (grant agreement 642434 to R.B.). M.G. was supported 9. Ochrietor, J. D. & Linser, P. J. 5A11/Basigin gene products are necessary for by the DFG (SFB871) and the advanced ERC grant ChroNeuroRepair. proper maturation and function of the retina. Dev. Neurosci. 26, 380–387 (2004). Reviewer information Nature thanks B. Berninger, I. M. Cristea, M. Evans and 10. Politis, P. K. et al. BM88/CEND1 coordinates cell cycle exit and J. MacKenzie for their contribution to the peer review of this work. diferentiation of neuronal precursors. Proc. Natl Acad. Sci. USA 104, 17861–17866 (2007). Author contributions P.S. and A.Pi. conceived the study. P.S. performed most 11. Gehman, L. T. et al. The splicing regulator Rbfox2 is required for both of the experiments. D.A.H. performed hNPC proteomics. K.D. performed hNPC cerebellar development and mature motor function. Genes Dev. 26, 445–460 cell culture. M.G. and R.B. intellectually contributed toward data interpretation. (2012). A.S. implemented the bioinformatic pipeline, statistical analysis and data 12. Tarlungeanu, D. C. et al. Impaired amino acid transport at the blood brain integration. A. Pł. generated phosphoproteomic samples. M.C. generated and barrier is a cause of autism spectrum disorder. Cell 167, 1481–1494 analysed immunofluorescence staining experiments. P.S. and A.Pi. wrote the (2016). manuscript with input from all authors. 13. Souza, B. S. et al. Zika virus infection induces mitosis abnormalities and apoptotic cell death of human neural progenitor cells. Sci. Rep. 6, 39775 Competing interests The authors declare no competing interests. (2016). 14. Tang, H. et al. Zika virus infects human cortical neural progenitors and Additional information attenuates their growth. Cell Stem Cell 18, 587–590 (2016). Extended data is available for this paper at https://doi.org/10.1038/s41586- 15. Watanabe, K. et al. A ROCK inhibitor permits survival of dissociated human 018-0484-5. embryonic stem cells. Nat. Biotechnol. 25, 681–686 (2007). Supplementary information is available for this paper at https://doi.org/ 16. Wojcechowskyj, J. A. et al. Quantitative phosphoproteomics reveals extensive 10.1038/s41586-018-0484-5. cellular reprogramming during HIV-1 entry. Cell Host Microbe 13, 613–623 Reprints and permissions information is available at http://www.nature.com/ (2013). reprints. 17. Söderholm, S. et al. Phosphoproteomics to characterize host response during Correspondence and requests for materials should be addressed to P.S. or A.P. infuenza A virus infection of human macrophages. Mol. Cell Proteomics 15, Publisher’s note: Springer Nature remains neutral with regard to jurisdictional 3203–3219 (2016). claims in published maps and institutional affiliations.

Methods were removed and fresh hNPC medium (basal or differentiation) was added. Mock Data reporting. No statistical methods were used to predetermine sample size. cells were inoculated with equal amounts of conditioned medium. The experiments were not randomized and the investigators were not blinded to For the generation of lentiviruses that expressed the individual HA-tagged ZIKV allocation during experiments and outcome assessment. viral proteins or Flag-tagged host proteins, synthetic cDNAs (sequences based on Cell lines and reagents. HeLa S3 cells (CCL-2.2) and Vero E6 cells (CRL-1586) recombinant ZIKV strain FSS13025, Asian lineage) were obtained in pUDC57 were purchased from ATCC. SK-N-BE2 cells were provided by R. Klein (MPI of vectors and the fusion gene was inserted into the lentiviral pWPI expression plas- Neurobiology, Munich). HEK293 cells were a gift from A. Bowie (Trinity College, mid. The lentivirus that expresses the HA-tagged HCV-NS4B of has previously Dublin). hNPCs were generated as described below. All cell lines were tested to been described30. Lentivirus stocks were produced and titrated following standard be mycoplasma-free and their identity verified by STR profiling. Primary anti- procedures. Nucleotide sequences of each construct are available from the bodies used in this study were: rabbit polyclonal anti-CEND1 (1:5,000 western corresponding authors upon request. blot; ab113076, Abcam), rabbit polyclonal anti-RBFOX2 (1:500 western blot; Lentiviral shRNA library production and shRNA screen. From the MISSION HPA006240, Sigma-Aldrich), rabbit polyclonal anti-CLN6 (1:500 western blot; TRC lentiviral library (Sigma-Aldrich), 54 MS hits and 4 controls were selected SAB4502281, Sigma-Aldrich), rabbit polyclonal anti-CHP1 (1:500 western blot; and shRNA were produced as follows. Two different shRNA-expressing lentivi- PA5-29876, Thermo-Fisher), rabbit polyclonal anti-BSG (1:2,000 western blot; ruses per gene (116 shRNAs in total; Supplementary Table 10) were individually HPA036048, Sigma-Aldrich), rabbit polyclonal anti-TMEM41b (1:250 western produced in HEK 293T cells (8 × 105) that were plated one day before transfec- blot; HPA014946, Sigma-Aldrich), mouse monoclonal anti-dsRNA J2 (1:400 tion. Transfections were performed as previously described4 using the packaging immunofluorescence; Scicons), mouse IgG2bκ anti-haemagglutinin (HA) (1:500 plasmids psPAX2 and pMD-VSV-G (provided by D. Trono) and shRNA-encoding immunoflorescence; 12CA5, Sigma-Aldrich), rabbit polyclonal anti-DENV-NS4B pLKO-puro plasmids. Viruses were collected at 48 and 72 h post-transfection, (dengue virus protein NS4B) (1:100 IFA; 1:500 western blot; GTX124250; Genetex; clarified by centrifugation and pooled before freezing. Controls included the non- this antibody cross-reacts with ZIKV-NS4B, as validated by immunodetection of targeting shRNA-encoding plasmids SHC001 and SHC002 (NT1 and NT2) as well ZIKV-NS4B in cells transduced with HA-tagged NS4B or ZIKV-infected cell lysates as the known virus restriction factors ATP6V0C and Musashi1. For quality-control (Extended Data Fig. 4b)). For detection of NS4B and capsid upon ZIKV infection purposes, approximately 25% of random samples were used to measure lentiviral by western blotting, the rabbit polyclonal anti-ZIKV-NS4B (1:1,000 western blot; titres. For screening purposes, 10,000 SK-N-BE2 cells per well were seeded in GTX133311; Genetex) and anti-ZIKV capsid (1:1,000 western blot; GTX133317, 96-well plates in 90 μl of complete DMEM and 24 h later cells were transduced Genetex) were used. For detection of phosphorylated and total protein abundance with 30 μl of each lentivirus (MOI ≈ 3) in triplicate wells/replica and three inde- by western blotting, mouse monoclonal anti-gammaH2Ax (ab22551, Abcam) and pendent biological replicates. Three days later, cells were infected with an MOI of the following rabbit monoclonal antibodies were used (Cell Signaling Technology): 0.1 of ZIKV H/PF/2013. Subsequently, 48 h post-infection, cell supernatants were anti-MARCKS (D88D11) and anti-phosphorylated MARCKS (D13E4), rabbit used for titration by PFU assay. Cell viability was measured by a resazurin-based monoclonal anti-mTOR (7C10) and anti-phosphorylated mTOR (D9C2), cell-viability assay, on plates which were transduced with shRNAs in parallel but p38–MAPK (D13E1) and anti-phosphorylated p38–MAPK (D3F9), S6 ribosomal were not infected with ZIKV. In brief, 50 μg ml−1 resazurin was added to each well protein (5G10) and anti-phosphorylated S6 (D57.2.2E). Where indicated, of a 96-well plate and incubated for 30 min at 37 °C, followed by measurement of to increase the basal phosphorylation of certain cellular proteins, fluorescence (535/590 nm) using an Infinite 200 PRO series microplate reader 12-O-tetradecanoylphorbol-13-acetate (TPA, Cell Signaling Technologies) or (Tecan). The normalization for the virus titres and fluorescence measurements DMSO, were added to the culture medium (32 nM), 15 min before collection of the were performed in two steps. First, the measurements in each replicate experiment cell lysates for western blot analysis. For characterization of hNPCs by immunoflu- were adjusted by the median factor of the replicate (the median of Xg,i/median(Xg,1, orescence, the following antibodies were used: mouse anti-GFAP (G3893, Sigma- Xg,2, Xg,3) across all the shRNAs (g = 1,2,...,116), where Xg,i is the titre of gth shRNA Aldrich), rat anti-Ki-67 (14-5698, EBioscience), mouse anti-MAP2 (M4403, clone in ith replicate). Second, the adjusted measurements were normalized, such that HM-2, Sigma-Aldrich), mouse anti-nestin (MAB5326, clone 10C2, Millipore) and the average signal of non-targeting controls was 100%. The following criteria rabbit anti-Sox2 (ab137385, Abcam). were applied to select hits: ≥50% decrease in median viral titres with two shRNAs Generation, maintenance and differentiation of hNPCs. Human induced or ≥75% if only one shRNA was efficient; cell viability ≥75% of non-targeting pluripotent stem cells (iPSCs) generated from healthy patient fibroblasts29 were controls. Validation experiments for selected shRNAs (Fig. 3b and Extended Data maintained in mTeSRTM 1 medium (STEMCELL Technologies) on Geltrex Fig. 9d, e) were performed in 24-well plates using the same overall protocol. (ThermoFischer Scientific) plates in the absence of feeders. The iPSCs were reg- Virus titration by plaque assay. Confluent monolayers of VeroE6 cells were ularly confirmed to be mycoplasma-negative. To generate hNPCs, iPSC colonies infected with serial tenfold dilutions of virus supernatants for 2 h at 37 °C. were collected with a cell scraper to ensure that relatively big colonies remained The inoculum was removed and replaced with serum-free MEM (Gibco, Life and these were cultured in suspension to form embryoid bodies in DMEM/F-12, Technologies) containing 1.5% carboxymethylcellulose (Sigma-Aldrich). Four 0.5% N2 supplement (Gibco), 5 μM dorsomorphin (Sigma-Aldrich), 10 μM days post-infection, cells were fixed for 2 h at room temperature with formalde- SB431542 (Sigma-Aldrich) and 10 μM Rho-associated kinase inhibitor (ROCK; hyde directly added to the medium to a final concentration of 5%. Fixed cells were Y-27632, StemCell Technologies). Embryoid bodies were grown for five days and washed extensively with water before staining with H2O containing 1% crystal the medium was changed every other day before gentle trituration and attachment violet and 10% ethanol for 30 min. After rinsing with water, the number of plaques on poly-ornithine–laminin-coated dishes in NPC medium, which consisted of was counted and virus titres were calculated. DMEM/F-12, 0.5% N2, 1% B27 supplement with 10 ng ml−1 bFGF (Peprotech Sample preparation for deep-proteomic and phosphoproteomic analyses. 100-18B-50) and 10 μM ROCK inhibitor. Three to four days after plating the For proteomic and phosphoproteomic analysis, SK-N-BE2 cells (107 cells per embryoid bodies, neural rosettes were manually isolated, dissociated mechanically biological replicate per condition; 4 biological replicates per condition), were to single cells and plated on fresh poly-ornithine–laminin-coated dishes in NPC mock-infected or infected with ZIKV H/PF/2013 at an MOI of 3. Cell pellets medium with 10 μM ROCK inhibitor. On the next day, the medium was changed were collected after 24, 48 and 72 h, lysed in 1 ml of lysis buffer (10 mM Tris- to NPC medium without ROCK and NPCs were propagated for up to 12 passages HCl (pH 7.5), 4% SDS and 0.05 mM DTT supplemented with complete pro- by splitting with accutase when 70–80% confluency was reached. For short-term tease and phosphatase inhibitor cocktails (Roche)), boiled 10 min at 98 °C, and differentiation experiments (Fig. 3c–e and Extended Data Fig. 4), 2 × 105 cells sonicated (4 °C for 15 min, or until a homogeneous suspension was formed). were seeded in six-well plates, transduced with empty lentiviruses (NT), lentivi- Clarified protein lysates were precipitated with acetone, and normalized protein ruses expressing HCV- or ZIKV-NS4B or infected with an MOI of 0.01 of ZIKV mixtures resuspended in 500 μl TFE digestion buffer. Protein digestion was per- H/PF/2013. After 48 h, differentiation was induced for five days by growth factor formed by adding 1:100 (protein:enzyme) trypsin and LysC with rapid agitation withdrawal and addition of 10 μM ROCK inhibitor, while undifferentiated cells (2,000 r.p.m.) overnight at 37 °C. An aliquot of the peptide mixtures was used were kept in NPC medium in the presence of bFGF. In both cases, the medium was for the determination of the total proteome (10%), while the remaining peptide replaced every other day. Cell pellets were collected, snap-frozen and processed for mixture (90%) was processed for phosphopeptide enrichment using the EasyPhos LC–MS/MS analysis as described below. protocol. For proteomic analysis of primary cells, hNPCs were seeded in six-well Virus strains, virus stocks preparation and generation of lentiviruses. ZIKV plates (2 × 105 cells per well) and either mock-infected, infected with an MOI strains H/PF/2013 and MR766 were obtained from the European Virus Archive of 0.01 of ZIKV or transduced with lentiviruses expressing HA-tagged ZIKV- (EVAg, France). Virus stocks were passaged once on VeroE6 cells and virus- NS4B or HCV-NS4B (MOI = 3). Two days after infection or lentivirus transduc- containing cell culture supernatants or conditioned medium from uninfected cells tion, cells were left in hNPC proliferation medium or neuronal differentiation were collected from day 3 to 8 post-infection. Supernatants were filtered through a was initiated by culturing the cells in hNPC differentiation medium. Five days 0.45-μM pore-size filter and stored at −70 °C. Titres of infectious virus were deter- later, cell pellets were collected and processed for label-free LC–MS/MS analy- mined by plaque assay as described below. For infection of hNPCs, virus stocks sis as described previously31. Each condition was performed in quadruplicate were diluted in hNPC medium and cells were inoculated for 2 h at 37 °C. Inocula biological replicates.

Deep-proteome and phosphoproteome LC–MS/MS analysis and data prior parameters τ1 = 1 and τ2 = 0.25 for model effects and variation in biological processing. Peptide mixtures were separated on a 50-cm reversed-phase column replicates, respectively. For the SK-N-BE2-cell dataset the corresponding statistical (diameter of 75 μm packed in-house with ReproSil-Pur C18-AQ 1.9 μm resin model was log(intensity) ∝ 1 + (after24 h + after48 h) × treatment with τ1 = 2 and 32 (Dr. Maisch GmbH)) as previously described . τ2 =.01 horseshoe prior parameters for proteomic data and τ1 = 1 and τ2 =.025 Raw MS files were processed within the MaxQuant environment (version for phosphorylation sites (phosphosites). The GLM model for AP–LC–MS/MS 1.5.7) using the integrated Andromeda search engine with false discovery rate experiments was log(intensity) ∝ 1 + bait, with τ1 = 1 and τ2 =.03 for ZIKV baits, (FDR) ≤ 0.01 at the protein, peptide and modification level. Proteome and phos- and τ1 = 2 and τ2 =.01 when host proteins were used as baits. phoproteome files were assigned to separate search parameter groups. The search The posterior distributions of model parameters were inferred using the included fixed modifications for carbamidomethyl (C) and variable modifications Hamiltonian Markov chain Monte Carlo (MCMC) method (Stan version 2.14), for oxidized methionine (M), acetylation (protein N-term), and phospho (STY) using eight parallel MCMC chains, 2,000 warmup and 2,000 sampling iterations for phosphoproteome files. Peptides with at least six amino acids were considered taking every fourth sample from each chain. The statistical significance of a for identification, and ‘match between runs’ was enabled with a matching time hypothesis that given model variable x is positive given data Y (one-sided P value), window of 0.7 min to transfer MS1 identifications between runs. Peptides and that is, the probability that a random sample from the posterior distribution of x proteins were identified using a UniProt FASTA database from human (UniprotKB would be non-positive (P(x ≤ 0|Y)), was calculated by approximating the posterior release 2015_08 including isoforms and unreviewed sequences) and ZIKV virus distribution from the MCMC samples using Gaussian kernel with Silverman’s rule- polyprotein corresponding to the H/PF/2013 strain (NCBI GenBank KJ776791.2; of-thumb bandwidth. For the x ≠ 0 hypothesis, the two-sided P value was defined individual viral cleavage products were manually annotated). as min(P(x ≤ 0|Y), P(x ≥ 0|Y)). The following criteria were used for reporting Sample preparation for affinity purification of ZIKV proteins and cellular pro- significantly regulated proteins and phosphosites: P≤ 0.01 and |median (log2(fold teins. For the determination of the ZIKV interactome, four independent affinity change))| ≥ 1 (Supplementary Tables 2, 5, 9). purifications were performed for each ZIKV HA-tagged viral protein. SK-N-BE2 To filter for specific interactions, the inferred posterior distributions of protein cells were transduced with an MOI of 3 of each lentivirus (8 × 106 cells per dish) intensities in each viral bait of the ZIKV AP–LC–MS/MS dataset were compared and 72 h later cells were scraped into 1 ml of lysis buffer (50 mM Tris (pH 8), with the average distribution of the same protein in all other baits excluding cap- 150 mM NaCl, 0.5% NP-40, cOmplete protease inhibitor cocktail, Roche) and sid protein (owing to its highly hydrophobic character and multiple subcellular HA-affinity purifications were performed as described previously33. In brief, localizations). The interaction was considered specific, if the one-sided P value clarified cell lysates were incubated with anti-HA-specific beads for 3 h at 4 °C, and (P(xbait ≤ xbackground|Y), where xbait is the protein intensity in the pull-downs of a non-specifically bound proteins removed by three washes with lysis buffer and given bait, and xbackground is its intensity in the pull-down experiments of the other five washes with washing buffer (50 mM Tris (pH 8), 150 mM NaCl). Bound pro- baits) was below 5 × 10−5 for capsid bait (since there were no other baits with teins were denatured by incubation in 20 μl guanidinium chloride buffer (600 mM the same biochemical properties) and 5 × 10−2 for the other baits, and median GdmCl, 1 mM TCEP, 4 mM CAA, 100 mM Tris-HCl (pH 8)). After digestion with protein enrichment against the background was more than 5.56-fold (= 22.5) for 1 μg LysC (WAKO Chemicals USA) at room temperature for 3 h, the suspension high-confidence and twofold for the lower-confidence set of ZIKV protein inter- was diluted in 100 mM Tris-HCl (pH 8), and the protein solution was digested with actions. The same filtering strategy was applied to AP–LC–MS/MS experiments trypsin (Promega) overnight at room temperature. Peptides were purified on stage of selected host targets; the P-value cutoff was 0.02 and log2(fold change) ≥ 1.5. tips with three C18 Empore filter discs (3M) and analysed by mass spectrometry For the representation of networks, subcellular localization of cellular proteins was as previously described34. For the analysis of CLN6 and other host-factor inter- extracted from Gene Ontology (version 2016-Sep-21). actomes, SK-N-BE2 cells were transduced with the corresponding Flag-tagged Merging of datasets. The integration of all AP–LC–MS/MS, proteomic and phos- lentiviruses and 72 h later cell lysates were processed as described above, using phoproteomic data was done by matching protein groups of individual datasets to agarose anti-Flag M2 beads (Sigma-Aldrich). each other using in-house scripts. The protein groups were matched if they shared Data processing of AP–LC–MS/MS samples. Raw mass-spectrometry data were a common protein accession code (‘Majority Protein ACs’ column in MaxQuant processed with MaxQuant (software version 1.5.3) using the built-in Andromeda output). In ambiguous situations (for example, one protein group of dataset A search engine to search against the human proteome (UniprotKB release 2015_08 shares different protein sequences with several protein groups of dataset B), the including isoforms and unreviewed sequences) plus ZIKV virus polyprotein (ZIKV priority was given to the reviewed protein sequences and canonical isoforms. strain FSS13025, Asian lineage; NCBI GenBank KU955593.1 with individual viral Integrating published protein–protein interactions, kinase–substrate relations cleavage products manually annotated), Gaussia luciferase (Q9BLZ2), HCV-NS4B and regulatory sites data. For overlaying our data with published protein–protein (E7ELX2) sequences, and the label-free quantification algorithm as described interactions, we used the IntAct interactions database (version 2017.04.08), IntAct previously35. In MaxQuant, carbamidomethylation was set as fixed and methio- complexes collection (version 2017.04.08) and CORUM protein complexes data- nine oxidation and N-terminal acetylation as variable modifications, using an base (version 2017.03.15). The interaction (physical association) was included if initial mass tolerance of 6 p.p.m. for the precursor ion and 0.5 Da for the frag- both interacting partners were significantly enriched in the same ZIKV bait. For ment ions. Search results were filtered with a FDR of 0.01 for peptide and protein published AP–MS interactions, we additionally required that one of the interacting identifications. partners was a bait. To integrate AP–LC–MS/MS datasets with proteomic and Statistical analysis of MS data. MaxQuant output files (proteinGroups.txt and phosphoproteomic screens, only published direct interactions (including kinase– ‘Phospho (STY) Sites.txt’ for proteome and phosphoproteome data, respectively) substrate relationships) were considered. were processed by a combination of in-house R (version 3.3), Julia (version 0.5) For phosphoproteomic analysis kinase–substrate relations and regulatory sites and Stan (version 2.14) scripts. were extracted from PhosphositePlus36 (kinase_substrate_dataset from 21 March To account for the variation in the analysed biological sample amounts and MS 2017 and regulatory_sites from 18 April 2017, respectively). performance, we inferred the normalizing multipliers for each MS run so that the Ingenuity pathway analysis. Proteome and phosphoproteome changes along normalized LFQ values of the representative set of proteins (around 500 randomly with unchanged proteins or sites, respectively, were submitted to the core values were selected from the ones quantified in most of the normalized samples) ingenuity pathway analysis (IPA). The following cut-offs were set for differentially do not change between the samples. First, this procedure was applied to normalize expressed proteins: P ≤ 0.01 and |log2(fold change)| ≥ 1.5. Ingenuity knowledge replicate MS runs within each condition, and then the normalization multipliers base was used as a reference dataset, while only experimentally observed findings were further adjusted to normalize the conditions between each other. For the were used for confidence filtering. No filters were applied at the level of species, analysis of proteome changes the LFQ intensities were fit to a Bayesian generalized tissues or cell lines. Input datasets were used for the functional analysis to identify linear model (GLM) for each protein group (raw intensities of each phosphosite biological functions and canonical pathways that were most significant. Right- for phosphoproteome data) individually. The model used horseshoe priors for the tailed Fisher’s exact tests were used to calculate P values. Biological functions or treatment effects, and the natural logarithm as a linking function for protein inten- pathways were considered significant when unadjusted P ≤ 0.05. Corresponding sities. Additional parameters with horseshoe priors were introduced to model the Benjamini–Hochberg-adjusted P values were calculated and are presented in biological variation within replicate samples. The MS measurements were set to the Supplementary Tables. follow the Laplace distribution. Technical MS runs from both LTQ-Orbitrap XL Co-immunoprecipitation assays. For validation of ZIKV-NS4B interacting and Orbitrap Q Exactive HF were used to estimate the signal-to-noise ratio and proteins, SK-N-BE2 cells were seeded into 15-cm2 dishes (8 × 106 cells per dish), sensitivity of protein quantification across the whole dynamic range of protein transduced with lentiviruses expressing ZIKV, DENV or HCV HA-tagged viral intensities. We used the obtained technical variation models to define the scale proteins or conditioned cell-culture medium. After 72 h of transduction, cell parameter of the Laplace distribution for each individual MS measurement as well monolayers were scraped into 1 ml lysis buffer (50 mM Tris-HCl (pH 8.0), 0.5% as scale and location parameters of the logistic distribution for missing MS meas- NP-40, 150 mM NaCl and protease inhibitor cocktail (cOmplete, Roche)), and urements. To analyse the hNPC proteome, we used the following GLM (in R GLM clarified lysates were processed for immunoprecipitation as previously described33. formula notation): log(intensity) ∝ 1 + differentiation × treatment with horseshoe Eluted proteins were further analysed by western blot as specified in the figure

© 2018 Springer Nature Limited. All rights reserved. RESEARCH Letter legends. For reciprocal co-immunoprecipitation experiments, SK-N-BE2 cells were Code availability. The analysis scripts written in R (version 3.3), Julia (version 0.5) transduced with lentiviruses expressing each Flag-tagged host protein and three and Stan (version 2.14) are available from the corresponding authors upon request. days later were infected with ZIKV at an MOI of 3. After 72 h of infection, cell mono Data availability. The mass-spectrometry-based proteomics data were deposited layers were processed for immunoprecipitation with anti-Flag M2 agarose beads at the ProteomeXchange Consortium (http://proteomecentral.proteomexchange. (Sigma-Aldrich) as previously described33. org) via the PRIDE partner repository with the following dataset identifiers: Immunofluorescence analysis. hNPCs (105 cells per well) seeded on poly- PXD009551, PXD009557, PXD009560 and PXD009561. The protein interactions ornithine–laminin-coated glass slides were transduced with lentiviruses (MOI = 3) from this publication have been submitted to the IMEx (http://www.imexconsor- expressing Flag-tagged proteins or empty control for 1 h at 37 °C. The next day, the tium.org) consortium through IntAct38 with the identifier IM-26452. medium was changed and 48 h post-transduction, cells were infected with ZIKV MR766 or H/PF/2013 at an MOI of 5 for 1 h at 37 °C. After 24 h, cells were fixed 29. Havlicek, S. et al. Gene dosage-dependent rescue of HSP neurite defects in with 4% PFA, permeabilized with 0.1% (vol/vol) Triton X-100 in PBS and unspe- SPG4 patients' neurons. Human Molecular Genetics 23, 2527–2541 (2014). cific binding sites were blocked with PBS containing 5% BSA for 1 h at room tem- 30. Paul, D., Bartenschlager, R. & McCormick, C. The predominant species of perature, and subjected to immunofluorescence staining using anti-Flag antibody, nonstructural protein 4B in hepatitis C virus-replicating cells is not anti-DENV-NS4B or J2 anti-dsRNA. Secondary staining was performed with don- palmitoylated. J. Gen. Virol. 96, 1696–1701 (2015). key anti-mouse Alexa568-conjugated and goat anti-rabbit Alexa 488-conjugated 31. Rappsilber, J., Mann, M. & Ishihama, Y. Protocol for micro-purifcation, antibodies. Nuclear DNA was stained with 4′,6-diamidino-2-phenylindole (DAPI) enrichment, pre-fractionation and storage of peptides for proteomics using (Molecular Probes). Coverslips were mounted in fluoromount-G mounting medium StageTips. Nat. Protoc. 2, 1896–1906 (2007). 32. Steger, M. et al. Phosphoproteomics reveals that Parkinson’s disease kinase (Southern Biotechnology Associates). Fluorescence images were acquired with a LRRK2 regulates a subset of Rab GTPases. eLife 5, e12813 (2016). Leica SP8 inverted confocal microscope using a 63× Plan-Apo NA 1.4 objective. 33. Scaturro, P., Cortese, M., Chatel-Chaix, L., Fischl, W. & Bartenschlager, R. Dengue Image analysis was performed using FIJI (http://fiji.sc/wiki/index.php/Fiji) and virus non-structural protein 1 modulates infectious particle production via Pearson’s correlation coefficients were calculated using the Coloc2 plugin37. interaction with the structural proteins. PLoS Pathog. 11, e1005277 (2015). Accession numbers. UniprotKB accession codes for all protein groups identified 34. Gebhardt, A. et al. mRNA export through an additional cap-binding complex consisting of NCBP1 and NCBP3. Nat. Commun. 6, 8192 (2015). by MS are provided in each respective Supplementary Table. The corresponding 35. Hubner, N. C. et al. Quantitative proteomics combined with BAC sequences were retrieved from UniprotKB (human; release 2015_08, including TransgeneOmics reveals in vivo protein interactions. J. Cell Biol. 189, 739–754 isoforms and unreviewed sequences). cDNA sequences corresponding to indi- (2010). vidual ZIKV open-reading frames used for generation of ZIKV lentivirus library 36. Hornbeck, P. V. et al. PhosphoSitePlus, 2014: mutations, PTMs and (KU955593.1) or related to the wild-type ZIKV H/PF/2013 strain (KJ776791.2) recalibrations. Nucleic Acids Res. 43, D512–D520 (2015). 37. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. were extracted from GenBank. Protein sequences of Gaussia luciferase (Q9BLZ2) Nat. Methods 9, 676–682 (2012). and HCV-NS4B protein (E7ELX2) were extracted from UniprotKB. 38. Orchard, S. et al. The MIntAct project—IntAct as a common curation platform Reporting summary. Further information on experimental design is available in for 11 molecular interaction databases. Nucleic Acids Res. 42, D358–D363 the Nature Research Reporting Summary linked to this paper. (2014).

Extended Data Fig. 1 | See next page for caption.

Extended Data Fig. 1 | Design rationale, expression and subcellular non-specific binding to the anti-HA beads as well as organelle-dependent localization of HA-tagged ZIKV proteins. a, Schematic representation enrichment artefacts in AP–LC–MS/MS analysis. b, Intracellular levels of ZIKV HA-tagged viral proteins and controls used in this study and of ZIKV proteins in lentivirus-transduced SK-N-BE2 cells. SK-N-BE2 experimental set-up used to generate the ZIKV cellular interactome. cells were transduced with lentiviruses encoding each of the ZIKV The full-length ZIKV genome is shown at the top, with the 5′ and 3′ proteins at an MOI of 3, and 72 h later cell lysates that had been clarified untranslated regions depicted with their putative secondary structures. by centrifugation were used for western blotting against HA and β-actin. Polyprotein cleavage products are separated by vertical lines and labelled Numbers on the left refer to molecular weight standards given in kDa. as specified. The individual open reading frames (ORFs) of each of the Asterisks mark proteins expressed at a lower level, which can be observed ZIKV proteins were fused with an HA epitope either at their N terminus upon longer exposure of the membrane. Note that C-terminally tagged or C terminus as depicted. Numbers above each sequence refer to the NS2A and N-terminally tagged NS2B-3 could not be detected, whereas the nucleotide sequence of the ZIKV isolate FSS13025 used as reference C-terminally tagged NS2B-3 fusion protein was expressed at the expected (GeneBank: KU955593.1). To ensure the correct subcellular localization molecular weight, and was therefore used for further experiments in this and protein topology, the sequence of the capsid anchor (Canch), the study. The bottom lane indicates the predicted molecular weight of each transmembrane domain of prM (prMTM) and the transmembrane domain viral protein in kDa. n = 2 independent experiments. A representative of Envelope (ETM) were fused at the N termini of prM, envelope and blot is shown. c, Subcellular localization of ZIKV proteins in lentivirus- NS1, respectively. Additionally, given the complex topology of NS2A, the transduced SK-N-BE2 cells. SK-N-BE2 cells were seeded in 24-well plates transmembrane domain of envelope, and a fusion protein encoding the and transduced as described above. After 72 h of transduction, viral first and the last 50 amino acids of NS1 was fused at its N terminus. proteins were detected using rabbit anti-HA antibody and Alexa-Fluor488- As additional controls, an empty pWPI-lentiviral vector (NT) or conjugated secondary antibody and imaged by confocal microscopy. Scale HA-tagged non-structural protein 4B of hepatitis C virus (NS4B-HCV) bar, 10 μm. n = 1 experiment. and Gaussia luciferase (G. luciferase) were included to monitor for

Extended Data Fig. 2 | Volcano plots of individual ZIKV-interacting and multiple subcellular localizations, a more stringent P value cut-off proteins. Volcano plots of label-free AP–LC–MS/MS of individual ZIKV was used (unadjusted one-sided P ≤ 5 × 10−5) and ribosomal proteins proteins. Each volcano plot displays all identified proteins (median were not individually labelled. The respective viral bait of each AP is log2(fold change) of bait-specific protein enrichment in comparison shown in black, black circles represent known DENV target proteins to the background plotted against the corresponding −log10(P value). while proteins with roles in neurogenesis, neurodegenerative diseases or Dotted grey lines represent the log2(fold change) and P value cut-offs infertility are labelled in red. n = 4 independent experiments, significance used. High-confidence interacting proteins (log2(fold change) ≥ 2.5, testing (P values) results from Bayesian statistical modelling (see Methods, unadjusted one-sided P ≤ 0.01) are coloured according to their subcellular ‘Statistical analysis of MS data’). localization. In the case of capsid, due to its highly hydrophobic character

Extended Data Fig. 3 | See next page for caption.

Extended Data Fig. 3 | Validation of NS4B-interacting proteins SMN1, LYAR, LARP7 and NGDN were infected as described above and upon ZIKV infection. a, Reciprocal co-immunoprecipitation of NS4B Flag-immunoprecipitated proteins probed with a ZIKV capsid-specific with Flag-tagged host factors. SK-N-BE2 cells transiently transduced antibody. n = 2 independent experiments. Representative blots are shown. with empty lentiviruses (NT) or lentiviruses expressing Flag-tagged d, hNPCs were transduced with lentiviruses expressing Flag-tagged CLN6 CEND1, CLN6, TMEM41b, RBFOX2 and CHP1 were mock-infected alone (mock) or co-transduced with lentiviruses expressing HCV-NS4B– or infected with ZIKV H/PF/2013 (MOI = 1) and three days later Flag- HA or ZIKV-NS4B–HA (MOI = 3), stained with Flag- and HA-specific immunoprecipitated proteins were probed with a ZIKV-NS4B-specific antibodies and imaged by confocal microscopy. Boxed areas are enlarged antibody. The bottom lane indicates the predicted molecular weights of on the right. Data are mean ± s.d. of Pearson’s correlation coefficients each viral protein in kDa. n = 3 independent experiments. Representative (each dot represent a single cell). Scale bars, 10 μm. n = 2 independent blots are shown. b, Validation of anti-DENV-NS4B-specific antibody experiments. e, Network representation of CLN6-interacting cellular for immunodetection of ZIKV-NS4B. SK-N-BE2 cells were transiently proteins and comparison with the ZIKV-NS4B interactors. AP–LC–MS/ transduced with empty lentiviruses (NT) or lentiviruses expressing HCV- MS analysis of Flag-tagged CLN6 in SK-N-BE2 cells revealed novel CLN6- NS4B–HA, DENV-NS4B or ZIKV-NS4B–HA, and 72 h later were probed specific interacting proteins (displayed in magenta; including TELO2 and consecutively with mouse anti-HA (left) or rabbit anti-DENV-NS4B mTOR) as well as shared cellular interactors with ZIKV-NS4B (displayed (right) specific antibodies. ZIKV-infected SK-N-BE2 cell lysates, shown in yellow, including ICMT and STT3A). Solid lines represent specific on the right side, were included as control. Asterisks indicate unspecific interactors identified by AP–MS/MS analysis (log2(fold change) ≥ 1.5, bands. n = 1 experiment. c, Reciprocal co-immunoprecipitation of capsid unadjusted one-sided P ≤ 0.02), grey dotted lines represent previously with Flag-tagged host factors. SK-N-BE2 cells transiently transduced published protein–protein interactions. n = 4 independent experiments, with empty lentiviruses (NT) or lentiviruses expressing Flag-tagged significance testing results from Bayesian statistical modelling.

Extended Data Fig. 4 | Volcano plots of ZIKV-NS4B and HCV-NS4B- transduced with empty lentiviruses (NT) or lentivirus-expressing interacting proteins. a, Volcano plot comparing the specificity of protein HA-tagged HCV-, DENV- or ZIKV-NS4B proteins were used for enrichment in ZIKV-NS4B versus HCV-NS4B AP (median log2(fold HA-immunoaffinity purification and probed with the indicated change) plotted against −log10(unadjusted P) for ZIKV-NS4B versus antibodies. n = 3 independent experiments. Representative blots are HCV-NS4B). Specific NS4B-interacting proteins are labelled as described shown. c, Reciprocal co-immunoprecipitation of NS4B with Flag-tagged in Extended Data Fig. 2. Previously reported interactors of HCV-NS4B, CLN6. SK-N-BE2 cells transiently transduced with empty lentiviruses or the ZIKV closely related DENV-NS4B are shown in blue and green, (NT) or lentivirus-expressing Flag-tagged CLN6 were infected with ZIKV respectively. n = 4 independent experiments, significant testing results (H/PF/2013) and Flag-immunoprecipitated proteins probed with NS4B- from Bayesian statistical modelling. b, Co-immunoprecipitation of ZIKV- specific antibodies. n = 3 independent experiments. Representative blots NS4B–HA with endogenous host proteins. Cell lysates of SK-N-BE2 cells are shown.

Extended Data Fig. 5 | See next page for caption.

Extended Data Fig. 5 | Effects of ZIKV infection or NS4B transduction proliferative state in the presence of bFGF. No neuron-specific protein on undifferentiated or differentiated hNPCs. a, Experimental approach expression (MAP2) or astrocyte marker expression (GFAP) was detected used for hNPC differentiation. hNPCs were transduced with empty (NT), under these conditions. After differentiation for five days in vitro in the HCV-NS4B–HA- or ZIKV-NS4B–HA-expressing lentiviruses (MOI = 3), presence of ROCK inhibitor, the neuronal marker MAP2 was upregulated mock-infected or infected with ZIKV (H/PF/2013) (MOI = 0.01). After whereas the proliferation marker Ki-67 was downregulated, suggesting 48 h, differentiation was induced for five days by growth factor withdrawal commitment towards the neuronal lineage. Scale bars, 20 μm. n = 3 and addition of 10 μM ROCK inhibitor, while undifferentiated cells were independent experiments. Representative images are shown. kept in NPC medium in the presence of bFGF. Cell pellets were then c, Cellular processes enriched in NS4B-ZIKV changes in proliferating used for quantitative label-free LC–MS/MS-based profiling of the global or differentiating hNPCs. Right-tailed Fisher’s exact test, Benjamini– proteome. b, Characterization of hNPCs used for proteomic analysis. Hochberg-corrected P values are shown (see Supplementary Table 3). Undifferentiated hNPCs kept in the presence of bFGF were stained with d, Total number of proteins identified across biological replicates and (i) SOX2- and GFAP-, (ii) Ki-67- and MAP2-, and (iii) nestin-specific conditions. e, Heat map of all significant changes occurring upon antibodies. Alternatively, hNPCs were differentiated for five days in the proliferation and differentiation conditions in NS4B-ZIKV-transduced presence of a ROCK inhibitor and were stained with (iv) Ki-67- and cells in comparison to NS4B-HCV-tranduced cells and mock transduction. MAP2-specific antibodies. Under proliferating conditions, hNPCs Unadjusted P ≤ 0.01, |log2(fold change)| ≥ 1.0; significance testing results expressed Ki-67, SOX2 and nestin, confirming the maintenance of their from Bayesian statistical modelling. c–e, n = 4 independent experiments.

Extended Data Fig. 6 | Global proteomic analysis of ZIKV-specific n = 4 independent experiments. Bayesian statistical modelling, effects in differentiating or proliferating hNPCs. a, hNPCs were |log2(fold change)| ≥ 1; unadjusted two-sided P ≤ 0.01. b, Venn diagram mock-infected or infected with ZIKV (H/PF/2013) (MOI = 0.01) and displaying the total number of significantly up- or downregulated proteins cultured under proliferating or differentiating conditions as described in every experimental condition (|log2(fold change)| ≥ 1; P < 0.01) and in Extended Data Fig. 5a. Volcano plots display proteins that were the total number and gene names of significantly modulated proteins significantly up- or downregulated by ZIKV-infection in differentiated similarly regulated by ZIKV infection or ZIKV-NS4B transduction (left) or undifferentiated (right) hNPCs. Viral proteins are labelled upon differentiation (diff) or proliferation (undiff). n = 4 independent in red. Proteins with functions in antiviral immunity are labelled in experiments. green. Significance cut-offs are indicated by the dashed grey lines.

Extended Data Fig. 7 | See next page for caption.

Extended Data Fig. 7 | Time-resolved phosphoproteomic analysis of experiments, right-tailed Fisher’s exact test, Benjamini–Hochberg- ZIKV-infected SK-N-BE2 cells. a, Total number of phosphopeptides, adjusted P ≤ 0.05). f, Profile plots of significantly changing phosphosites, phosphosites and phosphoproteins identified in this study. b, Total mapped to pathways in Fig. 3d, and their corresponding total protein number of phosphosites identified across biological replicates and levels at 24, 48 and 72 h after ZIKV infection (orange for phosphorylation conditions (n = 4). c, Relative percentage of class I, II and III phosphosites levels, yellow for protein levels) or mock treatment (grey). Numbers next (localization probability >0.75) and phosphorylation of specific Ser, Thr to protein names refer to MaxQuant protein group IDs (Supplementary and Tyr residues. d, Venn diagram of significantly changing phosphosites Table 5). Points are normalized intensities of individual replicates, solid at 24, 48 and 72 h post-infection (h.p.i.). e, Network analysis of cellular line is median, filled area corresponds to 25–75 percentiles, dashed lines processes significantly modulated by ZIKV infection at phosphorylation mark 2.5–97.5 percentiles of the posterior distribution. n = 4 independent level (IPA). Nodes are canonical pathways identified with Fisher’s exact experiments; Bayesian statistical modelling. test; edges are shared proteins between the pathways (n = 4 independent

Extended Data Fig. 8 | Biological functions and pathways modulated by R, regulation; S, signalling. c–e, Immunoblot analysis of mock- or ZIKV- ZIKV at the phosphorylation level and validation by phospho-specific infected SK-N-BE2 with phosphorylation- or virus-specific antibodies. antibodies. a, b, Biological functions (a) and canonical pathways (b) Where indicated, samples were treated with 12-O-tetradecanoylphorbol- modulated by ZIKV infection (IPA, n = 4, right-tailed Fisher’s exact test, 13-acetate (TPA; 32 nM) 15 min before lysis. n = 2. Representative blots unadjusted P ≤ 0.05. Corresponding Benjamini–Hochberg-adjusted are shown. P values are provided in Supplementary Tables 6 and 7). P, pathway;

Extended Data Fig. 9 | See next page for caption.

Extended Data Fig. 9 | Effect of shRNA-mediated silencing of used for titration by PFU assay. The cellular viability was assessed with ZIKV-modulated host proteins on viral replication. a, Schematic a resazurin-based assay that was performed in parallel. b, c, ZIKV titres representation of the experimental set-up used for the shRNA screen. and cell viability upon knockdown of selected host factors. Viral titres, as Selected target genes were chosen among the cellular host factors determined by PFU assay, and cell viability, as determined by resazurin specifically binding ZIKV-capsid or -NS4B (interactome; n = 26), assay, are expressed as a percentage of the non-targeting controls. n = 3 significantly regulated at the phosphoproteomic level by ZIKV infection independent experiments. The box plot middle line corresponds to (phosphoproteome; n = 6) or significantly regulated at the proteomic the median, the hinges represent the first and third quartiles, the lower level both by ZIKV infection and ZIKV-NS4B-expression in hNPCs (upper) whisker extends from the lower (upper) hinge to the minimum (proteome; n = 23). Controls included non-targeting shRNA of two (maximum) value no further than 1.5× the interquartile range. d, Cell different pLKO-based lentivirus generations (NT1 and NT2) used as viability of SK-N-BE2 cells upon knockdown of selected host factors reference; and shRNAs targeting ATP6V0C and Musashi-1, which impair (n = 4 independent experiments, normalized mean ± s.d.; related to flavivirus pH-dependent viral entry and ZIKV replication, respectively. Fig. 3b). e, Validation of the silencing efficiency of selected shRNA Two individual shRNAs per gene were selected from the MISSION TRC targeting ZIKV host factors. Gene silencing was evaluated for each library (Sigma-Aldrich), and used for lentivirus production in HEK293T cellular protein via western blot detection using the antibodies specified cells as described in the Methods. SK-N-BE2 cells were transduced with on the right. Asterisk indicates non-specific bands, numbers on the left individual lentiviruses (three wells per shRNA) in three independent indicate molecular weight markers expressed in kDa. n = 2 independent experiments. Three days post-transduction, cells were infected with ZIKV experiments. Representative blots are shown. H/PF/2013 (MOI = 0.1) and 48 h later virus-containing supernatants were

Extended Data Fig. 10 | Integration of data from orthogonal proteomic analysis, grey lines represent published protein–protein interactions from screens and protein–protein interaction databases. Integrated network IntAct and CORUM databases. Up- or downregulation at the proteome of ZIKV-interacting proteins, CLN6-interacting cellular proteins, proteins level is marked with filled circles (hNPCs: up, red; down, dark blue; changing at proteome and/or phosphoproteome levels plus measured or SK-N-BE2: up, orange; down, light blue). Phosphorylation changes in published interactions between them. Baits are shown as large red squares. SK-N-BE2 are presented as red (up) or blue (down) circle borders. Solid lines represent specific interactions identified by AP–MS/MS

Andreas Pichlmair & Pietro Scaturro; Corresponding author(s): 2017-03-04150D

Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.

Statistical parameters When statistical analyses are reported, confirm that the following items are present in the relevant location (e.g. figure legend, table legend, main text, or Methods section). n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement An indication of whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistics including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated

Clearly defined error bars State explicitly what error bars represent (e.g. SD, SE, CI)

Our web collection on statistics for biologists may be useful.

Software and code Policy information about availability of computer code Data collection Raw spectral data were extracted with the MSFileReader (v3.0.31, Thermo Fisher Scientific) or MSConvert (v3.0.11537, ProteoWizard), and proteomics MS Raw files were processed with MaxQuant (version 1.5.7 for proteome and phosphoproteome datasets; version 1.5.3 for Interactome datasets; version 1.5.6 for the Proteomic dataset) for identification and quantification.

Data analysis The analysis scripts used for analysis of proteomic data were written in R (version 3.3), Julia (version 0.5) and Stan (version 2.14) and are available from the corresponding author upon request. For processing and analysis of immunofluorescent images, FiJi (version 1.50i) was used.

For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information. March 2018 Data

1 Policy information about availability of data nature research | reporting summary All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability

Raw MS files and MaxQuant output files that support the findings of the study have been deposited to ProteomeXChange repository (http:// proteomecentral.proteomexchange.org) via the PRIDE partner repository with the following data set identifiers PXD009551, PXD009557, PXD009560 and PXD009561. The protein interactions from this publication have been submitted to the IMEx (http://www.imexconsortium.org) consortium through IntAct37 with the identifier IM-26452. The results of the statistical analysis of MS data are provided in the Supplementary Tables.

Field-specific reporting Please select the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Life sciences Behavioural & social sciences For a reference copy of the document with all sections, see nature.com/authors/policies/ReportingSummary-flat.pdf

Life sciences

Study design All studies must disclose on these points even when the disclosure is negative. Sample size No statistical methods were used to predetermine sample sizes. Sample sizes are always indicated in figure legends or related "Methods" section.

Data exclusions No samples were excluded from the studies. No animals were excluded from studies.

Replication For MassSpectrometry, in vitro viral replication experiments and co-IP-WB analyses, a minimum of three biological experiments were performed independently. Additionally, for virus replication and cell-viability measurements two to three technical replicates per each independent biological experiment were included. For experiments involving production of lentiviruses (shRNA or over-expression constructs) at least two independently produced batches of lentivirus preparation were used in biologically independent experiments. All attempts at replication were successful.

Randomization N/A. No randomization was used given the small number of samples and the lack of influence of randomization on the experimental design and experimental approach used. (no animal experiments were performed in this study).

Blinding N/A. Investigators were not blinded to experimental groups (in vitro experiments required prior knowledge for data interpretation)

Materials & experimental systems Policy information about availability of materials n/a Involved in the study Unique materials Antibodies Eukaryotic cell lines Research animals Human research participants

Antibodies Antibodies used rabbit polyclonal anti-CEND1 (#ab113076; Clone#EPR3739, 1:5000; Abcam), rabbit polyclonal anti-RBFOX2 (#HPA006240;

lot#A104051, 1:500; Sigma-Aldrich), rabbit polyclonal anti-CLN6 ( #SAB4502281; lot#3115062, 1:500;Sigma-Aldrich), rabbit March 2018 polyclonal anti-CHP1 (#PA5-29876; 1:500; Thermo-Fisher), rabbit polyclonal anti-BSG (#HPA036048; lot#R33347, 1:2000; Sigma- Aldrich), rabbit polyclonal anti-TMEM41b (#HPA014946; lot#A100914, 1:250; Sigma-Aldrich), mouse monoclonal anti-dsRNA J2 (1:400 IFA; Scicons), mouse IgG2bκ anti-HA (12CA5; Sigma-Aldrich), rabbit polyclonal anti-DENV NS4B (#GTX124250; Genetex), rabbit polyclonal anti-ZIKV NS4B (#GTX133311; GeneTex), anti-ZIKV Capsid (#GTX133317 Genetex), mouse monoclonal anti- gammaH2Ax (#ab22551; Abcam), mouse anti-GFAP (#G3893; lot#096M4844; 1:200; Sigma-Aldrich), rat anti-Ki67 (#14-5698; lot#C148504, 1:200; EBioscience), mouse anti-MAP2 (#M4403, lot#076M4853V, 1:400; clone HM-2; Sigma-Aldrich), mouse anti- Nestin (#MAB5326, clone 10C2; lot#246548, 1:100; Millipore), rabbit anti-Sox2 (#ab137385, lot#GR245326-3, 1:250; Abcam) and the following rabbit monoclonal antibodies (all from Cell Signalling Technology): anti-MARCKS (#D88D11) and anti-phospho- MARCKS (#D13E4), rabbit monoclonal anti-mTOR (#7C10) and anti-phospho-mTOR (#D9C2), p38-MAPK (#D13E1) and anti-

2 phospho-p38 MAPK (#D3F9), S6 ribosomal protein (#5G10) and anti-phospho S6 (#D57.2.2E). Secondary antibodies detecting

mouse or rabbit IgG (Jackson ImmunoResearch, Dako) were horseradish peroxidase (HRP)-coupled. nature research | reporting summary

Validation Antibodies against cellular proteins from Sigma-Aldrich, GeneTex, Abcam, Ebioscience and Thermo-Fisher have been validated in this study by either knock-down (RBFOX2, BSG, MARCKS, CEND1, CLN6, CHP1, TMEM41b), over-expression (DENV-NS4B, ZIKV- NS4B, ZIKV-Capsid, HA), virus infection (dsRNA) or subcellular localization experiments (GFAP, Sox2, Ki67, Nestin, MAP2). Additionally, phospho-specific antibodies and respective basal unphosphorylated isoforms have been thoroughly validated by the respective manufacturers (CST, Cell Signalling: https://www.cellsignal.com/).

Eukaryotic cell lines Policy information about cell lines Cell line source(s) HeLa S3 (CCL-2.2) and Vero E6 cells (CRL-1586) were purchased from ATCC. SK-N-BE2 cells were kindly provided by Rüdiger Klein (MPI of Neurobiology, Munich). HEK293 cells were a gift from Andrew Bowie (Trinity College, Dublin). Human Neural Progenitor (hNPC) cells were generated as described in Methods section from Human induced pluripotent stem cells (iPSC) derived from healthy patient fibroblasts.

Authentication The identity of all the immortalized cell lines used in this study was confirmed by STR-profiling (Eurofins Medigenomix). Official certification can be provided upon request. The identity of iPSC-derived hNPC was confirmed by mass-spectrometry and immunofluorescence-based analysis of know neuronal proliferation and differentiation markers.

Mycoplasma contamination All cell lines were tested to be mycoplasma free by standard PCR-based assay.

Commonly misidentified lines None of the cells listed in the ICLAC database were used in this study. (See ICLAC register) Method-specific reporting n/a Involved in the study ChIP-seq Flow cytometry Magnetic resonance imaging March 2018