Dynamic remodelling of the human host cell proteome and phosphoproteome upon enterovirus infection

The group of enteroviruses contains many important pathogens for humans, including poliovirus, coxsackievirus, rhinovirus, as well as newly emerging global health threats such as EV-A71 and EV-D68. Here, we describe an unbiased, system-wide and time-resolved analysis of the proteome and phosphoproteome of human cells infected with coxsackievirus B3. Of the ~3,200 proteins quantified throughout the time course, a large amount (~25%) shows a significant change, with the majority being downregulated. We find ~85% of the detected phosphosites to be significantly regulated, implying that most changes occur at the post-translational level. Kinase-motif analysis reveals temporal activation patterns of certain protein kinases, with several CDKs/MAPKs immediately active upon the infection, and basophilic kinases, ATM, and ATR engaging later. Through bioinformatics analysis and dedicated experiments, we identify mTORC1 signalling as a major regulation network during enterovirus infection. We demonstrate that inhibition of mTORC1 activates TFEB, which increases expression of lysosomal and autophagosomal genes, and that TFEB activation facilitates the release of virions in extracellular vesicles via secretory autophagy. Our study provides a rich framework for a system-level understanding of enterovirus-induced perturbations at the protein and signalling pathway levels, forming a base for the development of pharmacological inhibitors to treat enterovirus infections.


Phosphorylation of host factors involved in formation of viral replication organelles
We observed an early increase / late decrease in phosphorylation (cluster 2) of GBF1 T1337 (Supplementary Figure 8b and Source Data File). GBF1 T1337 phosphorylation leads to Golgi disassembly 2,3 and the timing of T1337 phosphorylation increase coincides with Golgi disassembly during CVB3 replication, hinting to a possible connection. T1337 is phosphorylated by 5'-AMP-activated protein kinase (AMPK) 2 . We observed early increased (cluster 2) phosphorylation of activating phosphorylation events on T183 on the AMPK1 isoform and the corresponding residue T172 on AMPK2, and a reduced phosphorylation (cluster 3) of the inhibitory site S496 on AMPK1 (Supplementary Figure 8a). Collectively, these changes imply an early increase in AMPK activity following viral infection, which could explain the observed pattern of GBF1 T1337 phosphorylation.
Strikingly, AMPK-dependent GBF1 T1337 phosphorylation is also induced during infection with another positive-strand RNA virus, hepatitis C virus, and has been proposed to contribute to the formation of the membranous web (which is the name used for hepatitis C virus replication organelles) 4 . We investigated the importance of GBF1 T1337 phosphorylation for infection using a previously established assay 5 Figures 8c and 8d). Together these data suggest that (de)phosphorylation of T1337 is not essential for the function of GBF1 in replication. This finding is in agreement with the recent observation that the GBF1 region downstream of its catalytic sec7 domain is dispensable for virus replication 6 . PI4KB S266 phosphorylation (reduced late in infection, cluster 3, Supplementary Figure 8b and Source Data File), in combination with S258 phosphorylation (not detected), has been tentatively linked to Golgi integrity in one study 7 . Several mutants (T>A, T>E and T>D) of these sites, alone or in combination, supported replication in a previously-established assay using PI4KB knockout cells 8 (Supplementary Figure 8e and Source Data File), suggesting that S258/S266 phosphorylation is not important for the function of PI4KB in infection. However, fundamental knowledge about the PI4KB S258/S266 phosphosites is lacking and other explanations may be possible, including the option that a currently unknown cooperative effect with other sites exists that was not tested here. Single mutation of the other phosphosites that we detected in our analysis (S277, S428 and S511) nor of the wellstudied regulatory site S294 that is phosphorylated by PKD to activate PI4KB 9,10 also did not impair the ability of PI4KB to rescue CVB3 replication (Supplementary Figure 8e and Source Data File).
Finally, we observed a late increase (cluster 1) in OSBP S240 phosphorylation ( Supplementary   Figure 8b), which has been linked to Golgi dissociation 11 . OSBP S240 is phosphorylated by protein kinase D (PKD) 11 . We detected an increased phosphorylation of two known activating sites on PKD1 and/or PKD3 (the peptides could not be unambiguously ascribed to a specific isoform), while no known regulatory sites on PKD2 were identified. The strongest increase in S240 phosphorylation occurred between 4 and 6 h.p.i, which is also the period around which the Golgi falls apart and replication organelles are formed. Unfortunately, we could not study the importance of OSBP phosphorylation as OSBP knock-out cells are not viable and hence a suitable experimental system is not available.

Immunoblotting
After the incubation time medium was aspirated, cells were carefully washed with ice-cold PBS and lysed in the plate for 15 min on ice in lysis buffer (50mM Tris-HCl pH7.5, 150mM NaCl, 1% Nonidet P40) supplemented with Complete EDTA-free protease inhibitor cocktail (Roche) and PhosStop phosphatase inhibitor mix (Roche). Cells were scraped with a rubber policeman and transferred to 1.5ml vials on ice and lysis was continued for another 15 min. Cell debris was pelleted by centrifugation for 10 min at 4°C at full speed in a tabletop centrifuge. Supernatants were transferred to clean vials and stored at -20°C.
Lysates were mixed with Laemmli sample buffer and heated 5 min at 95°C to denature proteins.
Denatured samples were stored at -20°C. Samples were analysed by SDS-PAGE gel electrophoresis and transferred to nitrocellulose membranes by wet-blotting. Membranes were blocked with 2% BSA in TBS-T (TBS (Tris-buffered saline; 10mM Tris pH7.5, 150mM NaCl) with 0.1% Tween-20) for 1 h at room temperature and incubated overnight at 4°C with primary antibodies diluted in 2% BSA in TBS-T. Blots were washed three times with TBS-T, incubated with infrared dye-conjugated secondary antibodies diluted in 2% BSA in TBS-T for 1 h at room temperature, washed three times with TBS-T and rinsed in TBS and then PBS. Blots were scanned using a Li-Cor Odyssey Fc near-infrared fluorescence imaging system.
For the immunoblotting of LC3 and GAPDH the following alterations were made to the protocol. Cells were detached using trypsin, washed in PBS and resuspended in RIPA buffer (40 mM Tris-HCl pH8, 150 mM NaCl, 0.5% sodium deoxycholate, 1% Triton X-100, 0.1% sodium dodecyl sulphate) supplemented with protease inhibitor cocktail (Roche). Lysis was performed for 30 min on ice and cell fragments were removed by centrifugation at 15,000xg for 15 min. Protein concentration was determined by Pierce BCA assay kit (ThermoScientific, Waltham, MA) according to the manufacturer's instructions. Cell lysates (7.5-10 µg) and positive controls for LC3I and LC3II (Nanotools #1041/PC3/LC3I and #1042/PC3/LC3II) were subjected to SDS-PAGE gel electrophoresis. Membranes were probed as described above with the exception that 0.25% fish skin gelatin (FSG, Sigma-Aldrich) in PBS-T was used as blocking buffer. Blots were washed five times in blocking buffer prior to secondary labelling using an antibody conjugated to horseradish peroxidase (HRP) and washed five times in PBS-T and three times in PBS prior to imaging. Blots were incubated with ECL solution (SuperSignal West Dura Extended Duration Substrate, ThermoScientific) and imaged using a Bio-Rad ChemiDoc imaging system and the accompanying Image Lab software (Bio-Rad).

Quantitative reverse-transcriptase PCR (qRT-PCR)
At the indicated time points, medium was aspirated and cells were lysed in lysis buffer + 1% tris(2carboxyethyl)phosphine (TCEP). In some experiments, cell lysates were stored at -20°C. RNA was isolated using the NucleoSpin mini RNA isolation kit (Macherey-Nagel) according to the manufacturer's instructions and mRNA concentrations were checked using the Nanodrop. cDNA was generated using the Taqman Reverse Transcription Reagents (Applied Biosystems) with random hexamer primers according to the manufacturer's instructions. qRT-PCR was performed using the LightCycler 480 SYBR Green I master mix (Roche) for 45 cycles (10 sec at 95°C, 5 sec at 58°C, and 30s at 72°C) in a LightCycler 480 (Roche). Primers against lysosomal target genes (TFEB, LAMP1, TPP1, SCPEP1, GNS, BLOC1S1 and BLOC1S3) are described in 1 , against actin in 13 and against CVB3 viral RNA in 14 .
Fold change of expression of a gene compared to first time point and normalised to actin in the series was calculated according to the Ct method.

Replication rescue experiments
HeLa cells grown in a 96-well plate were transfected with plasmids for the expression of GBF1 phosphorylation mutants in the context of a BFA-resistant A795E mutation in the Sec7 domain 5 . The next day the cells were transfected with the CVB3 replicon RNA with the Renilla luciferase gene replacing the capsid coding region. After replicon RNA transfection the cells were placed in the medium containing the cell permeable Renilla luciferase substrate EnduRen (Promega) and the indicated amount of BFA. Luciferase signal was recorded from live cells every hour for 18 hours after replicon transfection. Wild-type (WT) and PI4KB-knockout (PI4KB KO ) HeLa cells grown in 96-well plates were transfected with plasmids for the expression of PI4KB (wild-type or mutants) 8 . As controls, kinase-dead (KD) PI4KB or GFP targeted to the Golgi through the transmembrane domain of galactosyltransferase (GalT) were used. The next day, cells were infected with RLuc-CVB3 virus at MOI 0.1 for 8 hrs cells were lysed and luciferase activity was determined using the Renilla luciferase assay (Promega).

Supplementary Figure 3. Viral protein abundance in the proteome dataset.
Ranked protein abundances from the highest to the lowest. The average iBAQ 15 value across the biological triplicates is shown. During infection, viral protein levels increase dramatically in line with the high levels of viral protein production. Because the viral proteins are derived from a polyprotein that is stepwise proteolytically processed, some peptides produced upon the LysC/trypsin digestion cannot be unambiguously assigned to a precursor or mature viral protein. Therefore, the least processed form to which the detected peptide is unique is indicated (e.g. VP4-2-3 or POLYP [polyprotein]). Source data are provided as a Source Data file.    Figure 5. Inhibition of mTORC1 and activation of TFEB during CVB3 infection. a. Phosphorylation of eEF2 T56 is induced by CVB3 regardless of the cellular background. Different cell lines (human HuH7, A549 and HAP1 cells; monkey BGM and Vero E6 cells) were infected with CVB3 at MOI 10. Cells were lysed at indicated time points and lysates were analysed by Western blotting as in Figure 4a. b. CVB3 induces eEF2 T56 phosphorylation, but the cardiovirus EMCV does not. HeLa cells were infected with either CVB3 or EMCV at MOI 15, cells were lysed at different time points and lysates were analysed by Western blotting as in Figure 4a. Mengovirus capsid, which recognises multiple cleavage forms, serves as an infection control for EMCV. c. TFEB targets are upregulated upon CVB3 infection. HeLa cells were infected with CVB3 at MOI 50 as in Figure 4a. At different time points, cells were lysed, RNA was extracted, cDNA was made and the expression levels of a set of genes that are known to be under control of TFEB were determined by qRT-PCR. In parallel, viral genome levels were determined as a control for infection (see panel d). Shown are means, error bars represent ± SEM (n=3 independent experiments, each with three biological replicates). A plot showing the individual data points is available in the Source data file. d. Increase of CVB3 genomic RNA levels analysed by qRT-PCR from the same samples as in panel c serves as an infection control. Shown are means, error bars represent ± SEM (n=3 independent experiments, each with three biological replicates) A plot showing the individual data points is available in the Source data file. e. The TFEB target protein LAMP1 is increased in infected cells. Lysates of CVB3-infected cells from Figure 4a were analysed by Western blot using an antibody against lysosome-associated membrane protein 1 (LAMP1). f. Upregulation of lysosomal genes upon CVB3 infection depends on TFEB. HAP1 WT and TFEB KO cells were infected with CVB3 at MOI 10, cells were lysed at the indicated time points and RNA levels were determined by qRT-PCR as in panel c (mRNA levels of LAMP1 and SCPEP1) or panel d (CVB3 genomic RNA levels). Shown are means, error bars represent ± SD (n=3 independent experiments, each with three biological replicates). Plots showing the individual data points are available in the Source data file. g. CVB3 replication is not affected by TFEB knockout. HAP1 WT and TFEB KO cells were infected with Renilla luciferase-expressing CVB3 reporter virus at MOI 0.01 as described in Figure 1a. At the indicated time points cells were lysed and luciferase levels were determined as a quantitative measure of genome replication. The replication inhibitor guanidine was included as a control that shows that infection and translation levels were comparable between cell lines. Shown are means, error bars represent ± SEM (n=3 independent experiments, each with three biological replicates). A plot showing the individual data points is available in the Source data file. Source data are provided as a Source Data file for all panels.    Separation of naked and EV-enclosed CVB3 based on buoyant density. 100,000xg pelleted material from CVB3 infected cells was treated with 0.1% Triton-X100 to disrupt any membranes or left untreated before floatation into a density gradient. A shift of the infectivity from low-density fractions (1.10-1.04 g ml -1 ) to higher density fractions (1.35-1.15 g ml -1 ) upon detergent treatment is observed, indicating that infectious virus present in low-density fractions is enclosed in extracellular vesicles. c. Representative plot from three independent experiments showing the infectivity in different density gradients fractions for HAP1 WT and TFEB KO HAP1 cells. Virus was at isolated from cell culture supernatants 8 hpi by pelleting at 100,000xg and floated into a density gradient. Source data are provided as a Source Data file for panels b and c.
In case a site has multiple phosphorylation forms, the pattern for the lowest detected phosphorylation multiplicity is shown. Not significantly changed  Normalised phosphorylation of selected phosphosites on host factors that are linked to Golgi structure, based upon the data in Supplementary Data 4. Red points are the mean values of the normalised phosphosite intensities from the 4 replicates, filled grey area corresponds to 95 % confidence interval, and the blue line is the fitted regression curve. c. GBF1 A795E expression restores CVB3 replication upon BFA treatment. HeLa cells transiently expressing the BFA-resistant GBF1 mutant A795E (with or without the T1337A mutation) were transfected with CVB3 replicon RNA containing a Renilla luciferase reporter gene and incubated in the presence or absence of 1 µg ml -1 of BFA. Renilla luciferase signal was recorded from live cells every hour. Empty vector pUC was used as a negative control. Shown are means, error bars represent ± SD (n=2 independent experiments, each with 16 biological replicates). d. Non-phosphorylatable T1337A and phosphomymetic T13317E GBF1 mutants similarly rescue CVB3 replicon replication from BFA inhibition. The experiment was performed as in panel c with phosphosite mutants in the background of the BFA-resistant A795E mutant. Shown are means, error bars represent ± SD (n=2 independent experiments, each with 16 biological replicates). e. PI4KB phosphomutants similarly rescue RLuc-CVB3 replication in PI4KB KO cells. Wild-type HeLa cells (WT) and PI4KB KO cells transiently expressing PI4KB mutants were infected with RLuc-CVB3, cells were lysed and the Renilla luciferase signal was determined. As negative controls Golgi-targeted GFP with the membrane anchor of galactosyltransferase (GalT) and kinase-dead PI4KB (PI4KB KD) were used. For S277 and S428 we studied the S>A, S>D and S>E mutants as specific mutants of these sites have not been described to our knowledge. For S511 we used only the previously described non-phosphorylatable S>A mutant but not the S>D and S>E mutants as these do not faithfully mimic the phosphorylated state of S511 (which is S496 in most common isoform) 16 . Additionally, we studied mutants the well-known regulatory site S294 of which the non-phosphorylatable S>A mutant has been previously described 9,10 . Shown are means, error bars represent ± SD (n=3 independent experiments, each with three biological replicates). Plots showing the individual data points are available in the Source data file. Source data are provided as a Source Data file for b, c, d, and e.