Viral infection upregulates NUP153 expression and causes NUP153 relocation to the viral amplification site
NUP153 has been shown to play a role in the life cycle of some viruses14,16, but its function during flavivirus infection is poorly understood. To shed light on NUP153’s role during flavivirus infection, we first analyzed the localization during tick-borne flavivirus infection as the protein is typically found in the nuclear envelope (NE). We infected cells with TBEV and Langat virus (LGTV; the low pathogenic model for TBEV) as representative tick-borne flaviviruses, and determined the cellular localization of NUP153 (Fig. 1A)17. We found that the infection affected the localization of a fraction of NUP153, which colocalized with the viral NS3 protein in the cytoplasm. We quantified the NUP153 signal in the whole cell, nucleus, and cytoplasm in mock-treated and infected cells (Fig. 1B). We found a 3-fold increase of NUP153 levels during TBEV and LGTV infection, both in the nucleus and the cytoplasm, as compared to the levels in the mock-treated samples (Fig. 1C). The increase in NUP153 level may have been caused either by a direct upregulation of NUP153 expression upon infection or an increased stability of the protein. To distinguish between these scenarios, we followed the RNA transcript and protein levels of NUP153 over time. We observed an increase in NUP153 RNA transcript levels already 24 hours post infection (hpi) with LGTV and TBEV, with an increase on the protein level detected at 48 hpi (Fig. 1D,E). These observations support an upregulation of NUP153 expression during infection. To verify the integrity of the nuclear pore and import/export function of the NPC, we quantified the staining of Mab414 (a monoclonal antibody that binds to phenylalanine-glycine (FG) repeat-containing NUPs) (Supp Fig. 1A,B) and polyA RNA levels (Supp Fig. 1C,D). We found higher signal of the Mab414 antibody in both the nucleus and cytoplasm during infection and no significant differences in the nuclear or cytoplasmic fractions of polyA RNA between mock-treated and TBEV-infected cells (Supp Fig. 1B,D). This indicates that tick-borne flaviviruses upregulate and recruit NUP153 to the cytoplasm without compromising the export of cellular mRNA.
During flavivirus infection, the viral protein NS3 is present in the host ER, along with viral dsRNA, at the replication vesicles18. To determine if NUP153 is recruited to specific membranous structures in the cell during infection, we conducted subcellular fractionation of LGTV-infected cells (Fig. 1F). The analysis revealed that the protein is recruited to the ER during infection (Fig. 1G and Supp Fig. 1E). Next, we determined the colocalization of NUP153 during TBEV infection. In these experiments, we used NS3 as a general marker of the viral amplification site and dsRNA, the viral RNA replication intermediate, as a specific marker of RNA replication vesicles (Fig. 1H)7. We observed that NUP153 colocalized in the cytoplasm with NS3 but not with dsRNA (Fig. 1H), indicating that NUP153 is recruited to the viral amplification site on the ER but not specifically to the replication vesicles. Collectively, the above data indicated that viral infection affects the expression and cellular localization of NUP153.
NUP153 is a proviral factor during early infection
NUP153 is an essential cellular protein. To investigate its role in LGTV infection, we therefore performed a transient CRISPR Cas9 knock-down of NUP153 in HEK293T cells and evaluated its impact on infection. We quantified the knock-down efficiency using flow cytometry analysis and observed a 20% reduction of NUP153 protein levels in NUP153 knock-down samples compared to those in the control (Fig. 2A). We used the proviral factor GBF1 as a positive control19. GBF1 knock-down resulted in a significantly reduced infection by 80%, as determined by flow cytometry. In comparison, the NUP153 knock-down resulted in a 70% reduction of infection, indicating that although the knock-down was not complete, it had a major impact on viral infection (Fig. 2B).
To further investigate the role of NUP153 in TBEV infection, we used the auxin-induced degron (AID) system to rapidly degrade endogenous neonGreen-tagged NUP153 in DLD-1 cells [NUP153(NG)AID] without affecting the integrity of the NPC20. In this system, AID tagged proteins can be rapidly degraded conditionally upon addition of auxin. The neonGreen signal overlapped almost completely with the NUP153 signal in these cells, supporting that the neonGreen signal is a reliable marker for NUP153 localization (Supp Fig. 2A,B). Further, we verified the recruitment and colocalization of NUP153 with NS3 at the viral amplification site in DLD-1 cells (Fig. 2C,D and Supp Fig. 2C) and evaluated the degradation efficiency of the AID system. Full degradation of the NUP153(NG)AID protein occurred within 2 h of auxin treatment (Fig. 2E,F and Supp Fig. 2D,E). Using AID system, we first analyzed the effect of NUP153 depletion on LGTV infection, and found that the viral infection was reduced to 50% in the absence of NUP153 (Fig. 2G). Next, we depleted NUP153 at different timepoints during TBEV infection, to investigate the importance of NUP153 activity during the viral life cycle. In these experiments, the infection level was quantified by measuring the amount of viral E-positive cells. We observed that the addition of auxin 2 h prior to infection, and at 2 hpi or 4 hpi resulted in a 40% reduction of infection in NUP153(NG)AID cells but not in the parental DLD-1 control cells (Fig. 2H and Supp Fig. 2F). Auxin addition at 6 hpi led to no significant reduction in the number of infected NUP153(NG)AID cells, and had no detectable effect on the number of subsequently released viral particles (Fig. 2H,I). Taken together, these results indicated that NUP153 acts as a proviral factor during the early stages of viral infection.
TBEV infection affects NUP153 interactome
NUP153 thus plays an important role in viral replication and is recruited to the viral amplification site on the ER during infection. To understand the underlying mechanism, we employed an unbiased proteomic approach to examine the effect of infection on the NUP153 interactome. Cells transfected with enhanced green fluorescent protein (eGFP)-NUP153– or eGFP-encoding plasmids were infected with TBEV or mock-treated. The interacting proteins were enriched by protein GFP pulldown and analyzed by mass spectrometry. The analysis revealed subsets of proteins that specifically interacted (directly or indirectly) with NUP153 during either mock treatment or TBEV infection (Fig. 3A), as well as 242 proteins that interacted with NUP153 under both conditions. Pathway analysis (KEGG) of the shared protein set revealed that the NUP153 interactors were involved at different stages of protein production, from mRNA processing (spliceosome, 37 proteins) to nucleocytoplasmic transport (21 proteins) and translation (ribosome, 47 proteins) (Fig. 3B). Analysis of the data from mock-treated samples revealed that an additional set of proteins, involved in gene expression (spliceosome, mRNA surveillance, and ribosome), RNA degradation, and nucleocytoplasmic transport, was either lost or decreased during infection (Supp Fig. 3A). Although we did not observe differences in the distribution of polyA RNAs when comparing mock-treated and TBEV-infected cells (Supp Fig. 1C,D), we found that NUP153 interacted with fewer proteins involved in nucleocytoplasmic transport during infection, suggesting that this pathway might also be affected by infection. Statistical analysis of the 242 proteins found under both conditions revealed differential enrichment in mock-treated and TBEV-infected samples (Fig. 3C). Notably, during infection, we observed a shift in the cellular interactome of NUP153 towards actin cytoskeleton proteins, as well as a set of ribosomal and rRNA processing proteins (Fig. 3D). Several of these proteins participate in cell cycle regulation and proliferation, and are also involved in the translation process, either directly at the ribosome or by regulating ribosomal biogenesis21–28.
TBEV and other viruses induce translational shut-off in the host cell29,30. We investigated whether NUP153 is involved in this process, as we noted that infection altered its interactions with several proteins that are involved at different stages of gene expression. To quantify the absolute translation levels, we measured the incorporation of O-propargyl puromycin (OPP) into elongating polypeptides in the presence or absence of NUP153 using the AID system. We found that while TBEV infection resulted in a 40% reduction of global translation levels, NUP153 depletion did not affect these levels (Fig. 3E). This indicated that NUP153 is not involved in the translational shut-off caused by virus infection.
Cell cycle regulation is closely linked to cell proliferation. Because several proteins that we identified as interacting with NUP153 during viral infection are involved in this process, we also investigated the effect of NUP153 on cell proliferation. We determined the number of cells using DAPI staining as a readout in the AID-system in mock-treated and TBEV-infected cells. We found that the number of cells was reduced in the absence of NUP153, and this reduction was further decreased to 40% when the cells were infected with the virus (Fig. 3F). This suggested that NUP153 plays an important role in promoting cell proliferation during viral infection. To verify that this effect was not caused by cell death upon infection or NUP153 depletion, we determined the viability of these cells using Trypan blue. We observed no significant differences in cell viability following the different treatments (Fig. 3G). Collectively, we find that NUP153 is involved in ensuring effective cell proliferation during viral infection, potentially through interaction with several cell cycle regulating proteins, yet further studies are required to understand the mechanism.
NUP153 is recruited by NS3 and NS5 to the cytosol
Approximately 50% of human proteins identified in the NUP153 pulldown experiment were endogenous RNA-binding proteins (Supp Table 1). In addition, we identified three RNA-binding viral proteins (C, NS3, and NS5) (Fig. 3A)31,32. To determine whether NUP153 requires RNA binding as a bridge to interact with these viral proteins, we performed co-immunoprecipitation assay in the presence or absence of RNase A (Fig. 3H). All three viral proteins co-immunoprecipitated with NUP153 in the presence of RNA, but the interaction between NUP153 and C was abolished after RNase A treatment, indicating that NS3 and NS5 but not C bind directly to NUP153. It has been reported that a fraction of NS3 and NS5 proteins is recruited into the nucleus during infection of flaivivruses2,33. We therefore hypothesized that their interaction with NUP153 might occur during nuclear import, in the nucleus, or the cytosol34–37. Immunofluorescence analysis revealed that, while the colocalization of NS3 and NUP153 was more pronounced in the cytoplasm than in the nucleus (Fig. 1H and Fig. 3I), NUP153 colocalized very strongly with NS5 in the nucleus and cytoplasm during infection (Fig. 3J and Supp Fig. 3B). Since NUP153 colocalized and interacted with both NS3 and NS5 in the nucleus and the cytoplasm, we hypothesized that these viral proteins are involved in the recruitment of NUP153 from the NE to the viral amplification site. Accordingly, we transfected cells with constructs encoding FLAG-tagged NS2B3, NS3, or NS5, and used immunofluorescence microscopy to analyze the intracellular distribution of these proteins and NUP153 (Supp Fig. 3C,D). We observed that the presence of NS3 and NS5 resulted in NUP153 level upregulation and recruitment to the cytoplasm, while the presence of NS2B3 only resulted in a minimal increase of the NUP153 signal in the cytoplasm. Accordingly, this could indicate that nuclear localization of the viral proteins might be required for upregulation of expression and recruitment of NUP153 to the cytoplasm. Furthermore, we did not observe any accumulation of NUP153 at specific sites following overexpression of NS3 or NS5, suggesting that additional viral components, such as viral RNA, might play a role in this process. Overall, we find that NUP153 interacts with viral proteins NS3 and NS5 in an RNA independent manner. These interactions could be involved in the upregulation and recruitment of NUP153 to the viral replication site that we observe during infection.
The RNA-binding domain of NUP153 is important for NUP153 recruitment to the viral amplification site and binds viral RNA
NUP153 is a large multi-domain protein38. The N-terminal domain contains NE-binding site (amino acids 1–144), NPC-targeting domain (amino acids 39–439), and an RNA-binding domain (amino acids 250–400). The middle domain contains four zinc fingers, and the C-terminal domain contains FG repeats (Fig. 4A)38,39. To investigate which region of NUP153 is important for the recruitment of the protein to the viral amplification site during infection, we transfected HEK293T cells with constructs encoding different eGFP-tagged NUP153 truncation variants, and then infected the cells with LGTV (Fig. 4B). Similar to what was previously reported, in uninfected cells, the variants containing the NE-binding site and NPC-targeting domain (eGFP-NUP1531–650 and eGFP-NUP1531–250) correctly localized at the NE, while the variant lacking these domains (eGFP-NUP153880–1475) did not and, instead, was detected diffused in the cytoplasm (Fig. 4B)40,41. Further, the cellular localization of eGFP-NUP1531–650, but not eGFP-NUP1531–250 (the variant lacking the RNA-binding domain), changed during infection (Fig. 4B), indicating that the RNA-binding domain is required for the recruitment of NUP153 to the viral amplification site.
Since the RNA-binding domain was necessary for recruitment of NUP153 to the viral amplification site, we hypothesized that NUP153 binds directly to viral RNA. Accordingly, we performed a crosslinking immunoprecipitation (CLIP) experiment to investigate whether NUP153 can interact with viral RNA (Fig. 4C,D). The analysis revealed that full-length NUP153, as well as the N-terminal (eGFP-NUP1531–650) and C-terminal (eGFP-NUP153880–1475) variants can bind viral RNA (Fig. 4C). As the FG repeat-containing C-terminal domain of NUP153 binds multiple proteins, including viral proteins, the interaction with viral RNA may have been mediated by indirect interactions involving additional binding partners42. The eGFP-NUP1531–250 variant, which lacks the RNA-binding domain, did not bind viral RNA, suggesting that the RNA-binding region43 also binds viral RNA (Fig. 4D).
Next, we investigated whether the interaction between NUP153 and viral RNA is RNA polarity-specific. We observed that NUP153 bound both, positive- and negative-strand viral RNA, although the interaction with the positive-strand was significantly more pronounced than that with the negative-strand (Fig. 4E). Together, these observations indicated that the RNA-binding domain of NUP153 is required both for recruitment to the viral amplification site and interaction with viral RNA in flavivirus infection.
NUP153 interacts with a conserved G-rich region on viral RNA if the G-quadruplex (G4) structure is destamilized
The RNA-binding domain of NUP153 has previously been shown to associate with G- and GC-rich regions of RNA molecule12,43. To determine the site of NUP153 interaction with viral RNA, we analyzed cells transfected with eGFP-NUP1531–650–encoding construct and infected with TBEV, using CLIP-sequencing (CLIP-seq) (Fig. 5A; Supp Fig. 4A). Approximately 1.4% of reads mapped to the TBEV genome (Fig. 5B), particularly the positive strand (Fig. 5C). Statistical analysis of the positive-strand reads revealed association with three viral RNA regions, depicted as peaks in Fig. 5C. The first peak was located at the end of the E protein coding region (peak 1, 2118–2264 nt), a region that is highly conserved among flaviviruses, while the two other peaks (peak 2, 5760–5948 nt; and peak 3, 6950–6970 nt) were located in the NS3 and NS4B coding sequences, respectively (Fig. 5C; Supp Fig. 4B–D).
Peak 1 contains two predicted G-quadruplex (G4) structures (Supp Fig. 5A)44, a type of four-stranded RNA secondary structure that can fold in certain G-rich sequences. The second of which is highly conserved among flaviviruses sequences (Fig. 5D)45,46. We used in vitro transcription to generate RNA fragments (585 nt; Supp Fig. 5B) containing the wildtype (WT) peak and a mutated construct (11G1C) lacking the second, highly conserved G4. We then transfected cells with these constructs and measured their ability to form G4s, by quantifying the mean intensity signal after detection with anti-G4 antibody (1H6). We detected G4 signal from both transfections, however signal intensity of WT RNA was significantly higher compared to 11G1C RNA, suggesting that the WT sequence has higher propensity to fold into stable G4 structures (Supp Fig. 5C,D). To verify the existence of the two different G4s, we then in vitro transcribed the peak (Fig. 5D) and analyzed the properties using circular dichroism (CD) spectroscopy. G4s formed in RNA oligonucleotides are reported to predominately form parallel G4s, with positive maximum around 264 nm, and their stabilization is dependent on potassium ions47. The CD analyses showed a positive maximum at 264 nm for the WT and the 11GC1 mutant peak 1in the presence of KCl, while in its absence, the maxima shifted to 268 nm for WT and 272 nm for the 11G1C (Fig. 5E). These data suggested that both the WT sequence and the mutant 11G1C may form parallel G4 structures in presence of KCl. The small shift in the positive maximum of the WT fragment in the absence of KCl could be explained by the formation of complex secondary structures along the RNA molecule, including RNA stem-loop structures, which produces overlapping CD spectra with RNA G4s48,49. In contrast, the 28G derivative, which lacked all (28) conserved guanines (including the two G4s) but contained 17 not conserved guanines showed a maxima at 272 nm in all conditions (Fig. 5E) indicating that 28G mutant cannot form RNA G4 secondary structures. To further confirm the effect of KCl in G4 formation and stabilization in these sequences, we conducted thermal stability analysis using temperature-induced CD melting experiments. These experiments showed that the melting temperatures (Tm) of both WT and 11G1C sequences were significantly increased upon KCl addition by around 3°C and 6°C, respectively, while no change in Tm was observed for 28G mutant in presence of KCl (Fig. 5F).
Even though the analysis indicated that peak 1 can form a G4, it was not clear whether the G4 was bound by NUP153. In vitro, several sequences are able to form G4s, yet most of these are found unfolded in vivo, potentially through a complex mechanism involving proteins that bind G-rich RNA regions47. We therefore quantified eGFP-NUP1531–650 binding to the different RNA fragments in vitro in the presence or absence of KCl. We found that eGFP-NUP1531–650 bound significantly more WT RNA in the absence of KCl than in the presence of KCl (Fig. 5G). Similarly, the absence of KCl improved the eGFP-NUP1531–650 binding to the 11G1C derivative, which lacked the second G4 motif. In contrast, only low levels of binding were observed with the 28G derivative. This interaction was not significantly affected by KCl (Fig. 5G). This indicated that peak 1 form G4 motifs in the presence of KCl but NUP153 interacted with peak1 in the absence of G4 secondary structure.
Similar to peak 1, peak 2 (in the NS3 coding sequence) is conserved among flaviviruses and contains a predicted G4 structure (Supp Fig. 4C, 5A)44; however, the G4 is not conserved (Supp Fig. 5F). The total percentage of guanines in peak 2 is similar to that in peak 1 (31% vs. 32%) (Supp Fig. 5E) but, while 56% of these residues are conserved in peak 1, only 31% are conserved in peak 2. No conserved guanines were located in peak 3 (Supp Fig. 4D). Although several of the guanines in peak 2 are not conserved, we observed a pattern of guanine-adenine (GA) nucleotides that are highly conserved among flaviviruses that could be important for NUP153 binding. To investigate whether the conserved guanines or GAs identified in peak 2 are important for the binding to NUP153, we performed an in vitro binding assay with WT and mutated RNA sequences. We found that in the presence of KCl, peak 2 WT, 8G8A, and 19G RNA derivatives (Supp Fig. 5F) bound eGFP-NUP1531–650 to a similar degree as peak 1 WT in the absence of KCl (Fig. 5G,H). This indicated that the binding of NUP153 to peak 2 was not mediated by the conserved guanines or GAs.
Taken together, the above findings indicated that NUP153 binds positive-stranded RNA, and specifically, conserved guanines in peak 1, in the absence of a G4, while its binding to peak 2 is mediated by non-conserved residues.
NUP153 plays an important role in flavivirus infection by regulating translation of viral proteins
Since the guanines needed for NUP153 binding in peak 1 are conserved among all flaviviruses, we next sought to investigate the importance of NUP153 during infection with other prominent flaviviruses. We first evaluated the intracellular localization of NUP153 during infection with ZIKV, JEV, WNV, YFV, and DENV2. We found that, similar to TBEV and LGTV infection, infection with other flaviviruses induced the expression of NUP153 (Supp Fig. 6A) and recruited NUP153 to the cytosolic fraction (Fig. 6A,B). Furthermore, depletion of NUP153 using the AID-system resulted in a reduced number of infected cells in all the tested flaviviruses, indicating that NUP153 is an important conserved host protein in flavivirus life cycle (Fig. 6C and Supp Fig. 6B).
Flaviviruses have a single-stranded RNA genome that encodes a polyprotein that is subsequently processed into 10 viral proteins. Many viruses employ mechanisms to regulate viral protein translation, increasing the translation of specific viral proteins, while decreasing the translation of other proteins to ensure optimal replication. However, how flavivirus protein translation is regulated and whether all proteins are expressed equally during different stages of the viral amplification cycle is unknown50. Since we found that NUP153 interacts with numerous ribosomal proteins (Supp Table 1), is required early in the infection (Fig. 2H), and interacts with the viral mRNA at the end of the coding region of the structural proteins, we hypothesized that it could play a major role in regulating viral RNA translation to increase the ratio of viral structural proteins compared to non-structural proteins, leading to enhanced viral particle production during the early stages of infection. To test this hypothesis, we determined the levels of E and NS1 in TBEV-infected cells, and E and NS2B in WNV-infected cells using the AID-system. We observed that NUP153 depletion resulted in an approximately 20% reduction in E levels of both, TBEV and WNV (Fig. 6D,E). However, the expression of TBEV NS1 and WNV NS2B increased in the absence of NUP153, leading to higher NS1/E and NS2B/E ratios (Fig. 6F). This indicated that NUP153 interacts with the RNA corresponding to the E terminal region, and thereby modulate the expression of downstream non-structural proteins to favor the expression of the structural proteins.
Efficient production of progeny virions early in infection is crucial for viral spread and for preventing the cellular antiviral type 1 interferon signaling leading to enhanced infection51–53. Accordingly, we measured the efficiency with which the TBEV and WNV viruses spread to new cells in the presence or absence of NUP153. Brefeldin-A (BFA), which disassembles the Golgi complex and prevents virus release54, was used as a reference of initial infection, and was added at 16 hpi. In the presence of NUP153, both viruses spread to new cells between 16 hpi and 24 hpi. However, in the absence of NUP153, no additional cells became infected in the first 24 hpi for both TBEV and WNV, compared to the BFA-treated cells (Fig. 6G and Supp Fig. 6C). At 48 hpi, many new cells become infected even in the absence of NUP153, indicating that NUP153 is relatively more important early in infection.
Taken together, our results suggest that NUP153 binds to flaviviral RNA in the end of the E coding region thereby balancing the protein translation in favor for the structural proteins, leading to increased virion production and release. In the absence of NUP153 the regulation of viral protein translation is lost, decreasing the E protein levels in the cells and thereby delaying the virion release (Fig. 6H).