The ER membrane protein complex is important for the biogenesis of flavivirus 1 polyproteins 2 3

Abstract Flaviviruses such as dengue (DENV), Zika (ZIKV) or West Nile virus (WNV) translate their genome as a single multi-pass transmembrane (TM) protein at the endoplasmic reticulum (ER) membrane. Several genetic knockout (KO) screens identified the ER membrane protein complex (EMC) as a critical host component for flavivirus infection. The EMC facilitates accurate insertion, topology and/or stabilization of specific cellular TM proteins including a subset of tail-anchored proteins and G protein-coupled receptors. Here, we show that deletion of EMC in human cell lines decreased infection with DENV, ZIKV and WNV by 100 to 10,000-fold. Using replicon and immunoblotting studies in EMC KO cells, we demonstrated that the EMC was essential for viral protein expression. Ribosome Profiling of DENV-infected wild-type and EMC KO cells revealed no drastic differences in translation efficiency. Instead, absence of EMC led to a large fraction of expressed viral proteins being targeted to the proteasome post-translationally. We detected a decrease in stability in NS4A-NS4B of the viral polyprotein, a region rich in transmembrane domains. Additionally, we identified non-synonymous point mutations in NS4A and NS4B by performing iterative passaging of DENV on EMC KO cells. Adaptive mutations rescued both viral replication and stable expression of the NS4A-NS4B polyprotein segment in EMC KO cells. Lastly, we showed a physical interaction between the EMC and DENV NS4B protein post-cleavage and rapid degradation of processed NS4B in the absence of EMC. Together, our results suggest that the EMC engages with DENV polyproteins to ensure proper biogenesis of the NS4A-NS4B region, and provide further evidence for the cellular function of the EMC in the stable expression of TM proteins.


55
The EMC was originally discovered in a genetic screen for yeast mutants modulating the unfolded 56 protein response (UPR) (Jonikas et al., 2009). Genetic interaction mapping and biochemical 57 purification revealed a six-subunit core complex (EMC1-6) conserved in yeast and mammals 58 (Christianson et al., 2011;Jonikas et al., 2009). Other than its more general association with UPR 59 and ER-associated degradation (ERAD), the EMC was shown to interact co-translationally and 60 enable biogenesis of a subset of multi-pass TM proteins (Shurtleff et al., 2018). Biochemical

68
Together, our data highlights how flaviviruses hijack important machinery for TM protein 69 biogenesis to achieve optimal expression of their polyproteins, which reinforces a role for the EMC 70 in stabilizing challenging TM proteins during synthesis.

73
To validate the enrichment of EMC components in the flavivirus CRISPR screens, we generated 74 isogenic EMC subunit KO cell lines in Huh7.5.1 and HEK293FT, two cell types that were used for 75 the genetic screens ( Figure S1A). Upon infection with DENV expressing Renilla luciferase 76 (DENV-Luc), we measured a ~10,000-fold reduction in viral replication in Huh7.5.1 EMC1-6 KO 77 cells ( Figure 1A) and a 10-1,000-fold reduction in HEK293FT EMC1-5 KO cells ( Figure 1B). The differences in the extent of the phenotypes may be attributed to the presence of compensatory 79 pathways or the degree of disruption of the entire protein complex between the two cell types 80 upon knockout of a single subunit. The strong deficit in DENV replication in the absence of the 81 EMC persisted over a 30-72h period ( Figure S1B). Complementation of EMC2 and EMC4 KO 82 cells with the respective cDNA completely restored levels of DENV infection ( Figures 1C and   83   S1C). Lastly, we tested the effect of EMC4 KO on several flaviviruses and saw a significant 84 decrease of viral RNA for DENV, WNV and ZIKV but not hepatitis C virus (HCV), which is a more 85 distantly related member of the Flaviviridae family ( Figures 1D and S1D). This is congruent with 86 results of an HCV CRISPR screen that did not identify the EMC as a critical host factor (Marceau 87 et al., 2016). This finding highlights that there is a specific, evolutionarily conserved dependency 88 on the EMC among mosquito-borne flaviviruses and not a universal disruption of trafficking to 89 and/or translation or replication at the ER membrane, which would affect a wider range of viruses.

91
Next, we sought to determine at which step of the viral life cycle the EMC is required. As the EMC 92 is important for the expression of a large set of TM proteins, its disruption leads to an altered cell 93 surface proteome, which may affect viral entry. We measured effects on viral uptake in WT and 94 EMC4 KO cells by quantifying the amount of internalized viral RNA at 2 and 6 hours post-infection 95 (hpi), where we did not observe a difference ( Figure 2A). Only at 24hpi, a timepoint when viral 96 replication had occurred, was a difference in RNA load evident between WT and EMC4 KO cells.

97
Furthermore, we used a DENV replicon expressing luciferase, which bypasses entry and does 98 not produce viral particles, thus allowing the analysis of the effect of EMC KO specifically on 99 translation and replication. In this assay, we saw decreased luminescence in EMC4 KO cells 100 starting at 12 hours post-electroporation suggesting that the EMC is important during or prior to 101 viral replication ( Figure 2B). We additionally performed the replicon assay in the presence of MK-102 0608, a nucleoside analogue that inhibits the viral polymerase (Chen et al., 2015), and observed 103 that the luminescence signal in EMC4 KO cells was lower than in replication-inhibited WT cells suggesting that deletion of EMC has an effect on viral translation, which happens prior to 105 replication ( Figure S2A). To further dissect effects on translation or genome replication of DENV, 106 we conducted an immunoblot timecourse series, where WT and EMC4 KO cells were infected at 107 a high multiplicity of infection (moi=32) and lysates were harvested at early timepoints post-108 infection (6-12hpi), when no or only little DENV replication had occurred. Viral protein expression 109 was lower in EMC4 KO compared to WT cells as early as 6hpi further supporting a translation 110 defect ( Figures 2C and S2B). To rule out any contribution of low levels of replication to the 111 observed phenotype, we also performed the immunoblot experiment in the presence of MK-0608 112 and still saw different amounts of DENV protein accumulation between WT and EMC4 KO cells 113 indicating that the EMC is indeed important for viral protein expression ( Figures S2C and S2D).

114
Interestingly, while the EMC has been implicated in the stable biogenesis of TM proteins, we 115 found reduced expression of both TMD and non-TMD containing viral proteins. This suggests that 116 the defect likely occurs at the polyprotein stage thus affecting all individual, cleaved proteins.

118
The expression of flaviviral proteins is a complex process that requires several steps. DENV 119 translation is initiated in a cap-dependent manner, whereupon the single-stranded positive-sense 120 RNA genome is translated into a single ~3,400 amino acid long multi-pass TM protein. This 121 polyprotein is co-and post-translationally processed by host and viral proteases producing the 122 individual functionally active structural and non-structural proteins. We first tested whether the 123 defect occurs at the stage of polyprotein translation. We used ribosome profiling (Ribo-seq), 124 which measures the density of ribosomes on a given mRNA species with high positional 125 resolution, to compare translation efficiencies (TE) and successful full-length synthesis of both 126 cellular and viral proteins in WT and EMC4 KO cells (Ingolia et al., 2009 Figure 3D). Addition of BZ, however, did not restore 149 DENV replication arguing that the proteasome-targeted proteins are non-functional ( Figure S3C).

150
Together, this suggests that in the absence of EMC viral proteins are still translated but are unable 151 to achieve a stable conformation resulting in their degradation and inability to establish efficient 152 replication.

154
For cellular multi-pass client proteins, such as transporters and GPCRs, the EMC co-155 translationally engages with specific TMDs in order to support insertion into the membrane and allow stable expression of the entire TM protein (Chitwood et al., 2018;Shurtleff et al., 2018). We 157 used two orthogonal methods, a fluorescence-based protein stability assay and viral evolution, to 158 examine which region of the viral polyprotein requires the EMC for successful expression. First,

159
we utilized a dual-color fluorescent reporter as previously described (Chitwood et al., 2018). The

160
C terminus of different segments of the DENV polyprotein was fused to eGFP followed by a viral 161 T2A sequence and mCherry (G2C) ( Figure 4A). Translation of the fusion constructs generates 162 two protein products at a 1:1 ratio due to peptide bond skipping by the ribosome at the T2A site:

172
Similar results were obtained in HEK293FT ( Figure S4). NS4A has 3 TMDs with the third (termed 173 2K peptide) being highly hydrophobic and serving as a signal sequence for the translocation of 174 NS4B into the ER membrane (Lin et al., 1993). Subsequently, the NS4A-2K junction is cleaved 175 by the viral protease and the 2K-NS4B junction by the signal peptidase. As the 2K-targeted NS4B-176 5-G2C construct did not display a stability defect in EMC deficient cells, the results suggest that 177 EMC mediates stable expression by first engaging within NS4A prior to the 2K peptide and that a 178 defect in NS4A topogenesis is extended to the NS4B region. Stability of NS4A-4B-G2C was 179 restored to near WT levels by complementation of EMC4 KO cells with EMC4 cDNA ( Figure 4C).

180
Moreover, treatment of EMC4 KO cells with BZ increased the eGFP:mCherry ratio of NS4A-4B-181 G2C in EMC4 KO cells indicating that this region of the polyprotein is misfolded and targeted for degradation in the absence of EMC ( Figure 4D). Therefore, we linked the DENV replication 183 phenotype in EMC deficient cells to a specific molecular defect in the NS4A-4B region of the 184 polyprotein.

186
In a second, orthogonal approach, we performed viral adaptation to evolve EMC-independent 187 DENV mutants by passaging the virus consecutively on EMC4 KO Huh7.5.1 cells every 3-4 days.

188
After 20-22 passages we observed cell death in 3 out of 6 independent adaptation experiments.   Figure 5C). We also introduced the 209 two adapted mutations into NS4A-5-G2C, which showed increased stability in EMC4 KO cells 210 compared to the WT construct thus reverting the molecular destabilization phenotype ( Figure 5D).

273
The A6665G mutation is predicted to be in a TM helix of NS4A but A7558T is not in a TM region 274 (Miller et al., 2006(Miller et al., , 2007. Therefore, further studies are necessary to determine the exact 275 interaction between the EMC and the viral TMDs. Interestingly, the A6665G mutation also arose

345
Quantification of bands is shown in Figure S2B.

359
The DENV transcript is highlighted in red. The red line represents fold-change of TE=1.

362
The DENV transcript is highlighted in red. The red line represents fold-change of TE=1.

626
Distributions of fluorescence ratios were plotted as histograms using GraphPad Prism 8.