Retroviral adapters hijack the RNA helicase UPF1 in a CRM1/XPO1-dependent manner and reveal proviral roles of UPF1

Abstract

The hijacking of CRM1 export is an important step of the retroviral replication cycle. Here, we investigated the consequences of this hijacking for the host. During HTLV-1 infection, we identified that this hijacking by the viral protein Rex favours the association between CRM1 and the RNA helicase UPF1, leading to a decreased affinity of UPF1 for cellular RNA and its nuclear retention. As a consequence, we found that the nonsense-mediated mRNA decay (NMD), known to have an antiviral function, was inhibited. Corroborating these results, we described a similar process with Rev, the functional homolog of Rex from HIV-1. Unexpectedly, we also found that, for HTLV-1, this process is coupled with the specific loading of UPF1 onto vRNA, independently of NMD. In this latter context, UPF1 positively regulates several steps of the viral replication cycle, from the nuclear export of vRNA to the production of mature viral particles.

Graphical Abstract

Figure 1.

HTLV-1 Rex alters the repartition of NES containing cargos, including the RNA helicase UPF1. (A) Confocal microscopy experiments were performed in HeLa cells transfected with the plasmids indicated on the left side. The factors revealed are indicated on the top (panel (i)). Objective x20. Scale bar: 20 μm. Quantitative analysis of the microscopy experiments (panel (ii)): mCherry signal was quantified in the cytoplasm and in the nucleus and expressed as a ratio. Wilcoxon, P-value ***< 0.005. (B) Z-stack analysis performed with IMARIS software on images from A with mCherry and a staining of the nucleus by DAPI. The nuclear localisation of mCherry is evaluated with the Mander’s correlation coefficient (MCC). Cells were obtained from three independent experiments. t-test P-value ***<0.005. Objective x63. Scale bar: 7 μm. (C) Confocal microscopy experiments were performed in HeLa cells transfected with the plasmids indicated on the left side (panel (i)). The factors revealed are indicated at the top. Objective x63. Scale bar: 20 μm. Colocalising pixels view is obtained as mentioned in the method section and only show the pixels simultaneously activated with the indicated channels. Quantitative analysis of the microscopy experiments (panel (ii)): on the left, UPF1 signal was quantified in the cytoplasm and in the nucleus and expressed as a ratio. Wilcoxon, P-value *** < 0.005; on the right, the percentage of Rex expressing cells that display nuclear retention of UPF1 is indicated. t-test P-value *< 0.05. (D) Z-stack analysis performed on Hela cells expressing or not Rex (without tag); UPF1 and the nucleus were stained. The nuclear localisation of UPF1 is evaluated with the MCC. Cells were obtained from three independent experiments. t-test P-value ***<0.005. Objective x63. Scale bar: 5 μm.

Figure 2.

Rex interacts with CRM1 and UPF1 leading to an enhanced UPF1/CRM1 interaction and the modification of UPF1 localisation. (A) PLA carried out in 293T cells transiently transfected with a HA-REX coding plasmid. PLA combined antibodies specific of the HA tag and FTSJ1, eIF5A or UPF1 were used as indicated. FTSJ1 (ribosomal RNA methyltransferase 1) is a nucleolar protein used as a negative control (no described interaction with Rex). eIF5A, a known partner of Rex, was used as positive control. Each dot corresponds to the colocalisation of the indicated protein with a resolution of 40 nm. Magnification of one representative cell is framed in the merge. Two examples of dots are pinpointed by a white arrow. For each condition, the number of spots per cell and their localisation in the cytoplasmic or nuclear compartment are represented in a bar plot on the right. Objective x60. Scale bar: 10 μm. (B) PLA experiment carried out in HeLa cells transiently transfected with HA-UPF1 with or without Rex. PLA combined antibodies specific of the Ha Tag and endogenous CRM1. Objective x20. Scale bar: 20 μm (C) coIP experiments on 293T cells extracts. Cells express the indicated combinations of Rex and RNA containing the RxRE motif. IP were performed using rabbit polyclonal antibodies against CRM1 (middle panel) or from a pre-immune serum. Proteins revealed by western blot were indicated on the side. (D) Same as B with an antibody targeting Rex. (E) Same as C) with an antibody targeting UPF1. CoIP were incubated for 30 min with RNAse A (0.1 mg/ml) (panel (ii)) or not (panel (i)). * indicates an aspecific signal. Rex specific signal corresponds to the lower band. (F) coIP experiments in 293T cells transfected with plasmids coding the indicated forms of Rex and revealed by western blot. An empty vector is transfected in lane 1. IP was performed with a rabbit polyclonal antibody against UPF1.

Figure 3.

Rex expression leads to UPF1 reduced association to RNA, what is functionally link to NMD inhibition. (A) Coarse-grained molecular modelling with Martini force field of CRM1-UPF1 complex starting from UPF1 which CH domain open (PDB 2WJV) of UPF1 with CH domain closed (PDB 2XZL). UPF1 fragment is 77–916. Molecular dynamics simulation using Gromacs with Martini 2 force-field during 1 μs. UPF1 NES (89–103, dark blue) is bound to CRM1 (cyan) in both structures. UFP1 domains: CH: green; RecA2: wheat; RecA1: yellow, 1C: red, 1B: orange; stalk: black. The final state of simulations (1000 ns) of UPF1–CRM1 complexes are presented. The complete simulation movie is available as an additional data. The ‘radius of gyration’ (Rg (A)) and the ‘Buried Surface Area’ (BSA (A²)) from six independent simulations are indicated. (B) RIP experiments using a rabbit polyclonal antibody targeting UPF1 were performed with HeLa cells transfected with a Globin PTC coding plasmid and WT or NES mutant Rex coding plasmid. Immunoprecipitated RNA (Globin PTC) were quantified by RTqPCR. For each condition, the relative enrichment of Globin PTC RNA associated to UPF1 compared to conditions without Rex was displayed in the graph. n = 4 t-test P-value: ns > 0.05. and ***< 0.005). On the right panel, western blot controlling the levels of the expressed Rex and the immunoprecipitated UPF1. (C) Decay rate analysis of Globin PTC and Globin WT RNA. Globin RNA levels were quantified after HeLa cells were treated with DRB for 0 or 3 h. The rate of decay is expressed as ln (DRB3h/DRB0h). t-test P-value: ns > 0.05. and ***< 0.005. (D) Half-life evaluation of the Glob PTC mRNA in HeLa cells transfected with the indicated forms of Rex. mRNA half-lifes (t1/2 = ln (2)/λ with λ the time constant of the decay curves) are indicated in front of their respective conditions. n = 5 t-test P-value: *< 0.05. The expression of Rex at each time point after DRB treatment was evaluated by WB (lower panel). (E) Same as C with the Glob PTC mRNA and the indicated combinations of Rex and CRM1 overexpression. n = 3 t-test P-value: ns > 0.05, *< 0.05 and ***<0.005. (F) Confocal microscopy with Hela cells overexpressing the indicated combination of Rex and CRM1. UPF1 and Rex are stained. On the right, quantitative analysis of the microscopy experiments : UPF1 signal was quantified in the cytoplasm and in the nucleus and expressed as a ratio. Wilcoxon, P-value ***< 0.005.

Figure 4.

HIV-1 Rev promotes similar defects in UPF1 functions as HTLV-1 Rex. (A) (i) Confocal microscopy experiments were performed in HeLa cells transfected with the plasmids indicated on the left side. Rev was revealed with an anti Flag tag antibody. Magnification of the framed zone is shown in the lower panel (ZOOM). Objective x20. Scale bar: 50 μm. (ii) The same quantitative analysis was carried out as for Rex in Fig. 2C. (B) Half-life evaluation of the Glob PTC mRNA in HeLa cells transfected with the indicated forms of Rev. mRNA half-life (t1/2 = ln (2)/λ with λ the time constant of the decay curves) are indicated in front of their respective conditions. n = 3 t-test * P< 0.05. (C) Decay rate analysis of Globin PTC and Globin WT RNA. Globin RNA levels were quantified after HeLa cells were treated with DRB for 0 or 3 h. The rate of decay is expressed as ln (DRB3h/DRB0h). n = 3 t-test ns: P> 0.05; *** P< 0.005. (D) Same as C) except that HeLa cells were infected with HIV particles or control virus-like particles (CtrVLP) instead of being transfected with Flag Rev (left panel). GFP expressed from HIV or CtrVLP was observed with an epifluorescence microscope (right panel).). n = 4 t-test * P< 0.05 (E) RIP using a rabbit polyclonal antibody targeting UPF1 was performed with HeLa cells expressing Globin PTC mRNA as well as Rex or Rev as indicated. Immunoprecipitated RNA (Globin PTC) were quantified by RTqPCR. For each condition, the relative enrichment of Globin PTC RNA associated to UPF1 compared to control conditions (pCMV) was displayed in the graph. n = 4 t-test ns: P> 0.05; * P< 0.05, ** P< 0.01, *** P< 0.005 (F) Same as E with cells expressing globin WT mRNA. (G) same as E with the quantification of endogenous mRNA gadd45a, smg5, and gapdh. (H) Representative western blot controlling the RIP experiments: Rex and Rev proteins expression was monitored in the whole cell extract (INPUT) as well as UPF1 in immunoprecipitations.

Figure 5.

CRM1 hijacking by Rex affects UPF1 functions in infected lymphocytes. (A) coIP experiments in the indicated lymphocytes cells extracts. IP were performed using rabbit polyclonal antibodies against UPF1. Proteins revealed by western blot are indicated on the side. (B) (i) Confocal microscopy experiments were performed in C91PL and C8166 lymphocytes. CRM1 and UPF1 were revealed. Objective x63. Scale bar: 10 μm. (ii) Quantitative analysis of UPF1 nuclear retention from confocal images (C) Z stack analysis performed on C91PL and C8166 with a staining of the nucleus by DAPI and UPF1. The nuclear localisation of UPF1 is evaluated with the MCC. Cells were obtained from three independent experiments. t-test P-value: *< 0.05. (D) RIP using a rabbit polyclonal antibody targeting UPF1 were performed with C8166 and C91PL lymphocytes. Immunoprecipitated endogenous RNA (sensitive or not to NMD) were quantified by RTqPCR. The relative enrichment of RNA associated to UPF1 in C8166 compared to C91PL was displayed in the graph. n = 3 t-test P-value: *<0.05; ***<0.005. (E) Graphical abstract of the consequences of CRM1 hijacking by Rex on the RNA helicase UPF1: Rex alters the nuclear export and favour the nuclear retention of NES containing cargoes, including UPF1. Concomitantly, Rex stabilises CRM1/UPF1 association. Altogether, this lead to a defect in RNA binding and NMD inhibition.

Figure 6.

Rex/CRM1/UPF1 complex drives the loading of UPF1 on vRNA. (A) RIP using a rabbit polyclonal antibody targeting UPF1 were performed with Jurkat, C8166 or C91PL lymphocytes. Immunoprecipitated unspliced vRNA was quantified by RTqPCR. The relative enrichment of vRNA associated to UPF1 in Jurkat and C8166 compared to C91PL was displayed in the graph. n = 5 t-test *** P< 0.005 (B) RIP experiments were carried out with the immunoprecipitation of UPF1, Rex or UPF2 as indicated in C91PL and in C8166 cells. The relative enrichment in vRNA in C91PL compared to C8166 is represented for each immunoprecipitated protein. n = 3 t-test * P< 0.05 (C) coIP experiments in the C91PL cells extracts. IP were performed from the same extract using UPF1, UPF2 or a control antibody (Ctr). Proteins revealed by western blot were indicated on the side. (D) Schematic diagram of the double RIP experiment performed in 293T cells co-transfected with the HTLV-1 WT molecular clone and a HA-UPF1 coding plasmid (upper panel). Quantification of the relative enrichment in vRNA after HA RIP targeting UPF1 (RIP1) and Rex RIP (RIP2) (lower panel) compared to RIP Ctr. n = 3 t-test * P< 0.05 (E) RIP targeting UPF1, Rex or using a control antibody (as indicated) were performed with 293T cells transfected with HTLV-1 R23L molecular clone complemented with either Rex NES mutant or Rex WT (REX NES and WT, respectively). As a negative control, cells were transfected with an empty vector (PCMV) . Immunoprecipitated RNA (vRNA) was quantified by RTqPCR. The graph displays the relative enrichment of vRNA associated to UPF1 or Rex compared to control antibody. n = 4 t-test ns P> 0.05, *** P< 0.005. (F) 293T cells are transfected with the indicated combinations of HTLV-1 molecular clone (pACH) and HA-UPF1 either WT or NES mutated (UPF1 NES). Quantification of the relative enrichment in vRNA after RIP was performed with HA antibody or Ctr antibody. n = 3 t-test ** P< 0.01.

Figure 7.

UPF1 controls different steps of the viral replication cycle. (A) Protein expression in cell extracts used for dual luciferase assay was monitored by western blot (left). Dual luciferase assay was performed on shCtr or shUPF1 293T cells transfected with a Rex WT plasmid or an empty vector. Firefly luciferase as well as Renilla luciferase were quantified. Firefly (FLuc)/Renilla (RLuc) ratio was plotted in a graph. n = 4 t-test * P< 0.05 (right). (B) Same as A with overexpression of MYC-UPF1 WT or mutated on the NES (MYC-UPF1 NES). To reduce the background of endogenous UPF1, cells were also treated with siRNA. Both Myc-UPF1 are siRNA resistant. (C) Schematic representation of the experiment. The same preparation of cell extracts and the corresponding extracellular fractions were used to perform western blot, qRTPCR and ELISA. FACS data are presented in Supplementary Fig. S5. (D) Western blot was carried out on cellular extract (i) and on the corresponding extracellular fraction (ii). The recognised proteins were indicated on the left. The relative levels of the capsid (CA) and the matrix (MA) compared to the precursor GAG (i) or Env (ii) were evaluated after densitometry quantification using Imagelab software. (E) ELISA spot assays (Zeptometrix) were performed with the cell extract of the indicated cells. MA (p19) quantification was plotted in bar plot. (F) Same as E with the extracellular fraction preparation. (G) vRNA quantifications were performed by qRTPCR on cell extracts. Primers used were localised in the gag coding region. n = 5 t-test * P< 0.05; ** P< 0.01 (H) Same as G with the corresponding ECF preparation. (I) FACS quantifications of p19 protein in 293T cells treated with Ctr, UPF1 or UPF2 siRNA (left). The same experiment was performed with DMSO or the NMD inhibiting compound known as SMG1i. n = 5 t-test ns P> 0.05, ** P< 0.01.

Figure 8.

UPF1 is loaded into the viral particles produced by the virus. (A) Confocal microscopy experiments were performed in C91PL and C8166 lymphocytes. CRM1, UPF1, and p19 were revealed by IF. White arrows show p19 foci. Objective x63. Scale bar: 10 μm. (B) ECF containing the viral particles was collected from 15.10⁶ of C91PL, C8166 and Jurkat cells and analysed (lanes 4, 8, and 12). The presence of UPF1, RRM2, and p19 was evaluated by WB and total protein was monitored with stainless procedure (upper and lower panels, respectively). Cells used to harvest the ECF were also lysed and monitored by WB after serial dilutions (equivalent to 4.10⁶, 2.10⁶, and 10⁶ cells; lanes 1, 5, 9; 2, 6, and 10; 3, 7, and 11, respectively). (C) RIP UPF1 on the ECF of C91PL and C8166. Western blot controls of the RIP on the left. Immunoprecipitated unspliced vRNA was quantified by RTqPCR. The relative enrichment of vRNA associated to UPF1 compared to the control RIP was displayed in the graph on the right. n = 3 t-test: *** P< 0.005. (D) Volcano plot of the log2FC (C91PL/CEM) for the ‘free particle’ fraction. UPF1 enrichment in the C91PL samples compared to CEM samples is indicated (n = 3). At the top, the distribution of the proteins only found in the C91PL samples is represented by their IBAC values: viral proteins are indicated. The stochiometric proportion of UPF1 compared to viral proteins from C91PL samples is also indicated (down). (E) Upper panel: schematic representation of the viral transmission experiment. Lower panel: bar plot representing the luciferase levels expressed by 150 × 10³ infected Jurkat after 24h of co-culture with 50 × 10³ C91PL. C91PL were treated with doxycyclin for 3 days (one treatment) or 6 days (two treatments) in order to express the shRNA (control or against UPF1). n = 6, t-test ns P> 0.05, *** P< 0.005. (F) Graphical abstract of the consequences of CRM1 hijacking by Rex on UPF1 interplay with vRNA.