Human Nucleoporins Promote HIV-1 Docking at the Nuclear Pore, Nuclear Import and Integration


Authors: Francesca Di Nunzio, Anne Danckaert, Thomas Fricke, Patricio Perez, Juliette Fernandez, Emmanuelle Perret, Pascal Roux, Spencer Shorte, Pierre Charneau, Felipe Diaz-Griffero, Nathalie J. Arhel.


The nuclear pore complex (NPC) mediates nucleo-cytoplasmic transport of macromolecules and is an obligatory point of passage and functional bottleneck in the replication of some viruses. The Human Immunodeficiency Virus (HIV) has evolved the required mechanisms for active nuclear import of its genome through the NPC. However the mechanisms by which the NPC allows or even assists HIV translocation are still unknown. We investigated the involvement of four key nucleoporins in HIV-1 docking, translocation, and integration: Nup358/RanBP2, Nup214/CAN, Nup98 and Nup153. Although all induce defects in infectivity when depleted, only Nup153 actually showed any evidence of participating in HIV-1 translocation through the nuclear pore. We show that Nup358/RanBP2 mediates docking of HIV-1 cores on NPC cytoplasmic filaments by interacting with the cores and that the C-terminus of Nup358/RanBP2 comprising a cyclophilin-homology domain contributes to binding. We also show that Nup214/CAN and Nup98 play no role in HIV-1 nuclear import per se: Nup214/CAN plays an indirect role in infectivity read-outs through its effect on mRNA export, while the reduction of expression of Nup98 shows a slight reduction in proviral integration. Our work shows the involvement of nucleoporins in diverse and functionally separable steps of HIV infection and nuclear import.


The nuclear pore complex (NPC) is a supramolecular protein assembly forming a highly selective channel embedded in the nuclear membrane. It regulates bidirectional nucleo-cytoplasmic transport for a large range of proteins and complexes too large to diffuse freely through the NPC [1][2][3]. They are composed of numerous copies of ∼30 different nucleoporins (Nups), which have a well-assigned localisation, function and half-life, and are present as multiples of eight reflecting the highly conserved eight-fold axial symmetry of NPCs [2][4][5][6].

The central substructure of the NPC is composed of transmembrane Nups that anchor the NPC to the nuclear envelope, scaffold Nups (e.g. Nup107/160 complex) that constitute cornerstones during NPC biogenesis, and FG-Nups (e.g. Nup98, Nup358/RanBP2, Nup214/CAN) so-called because they contain extensive repeats of phenylalanine-glycine (FG) domains that form an unstructured mesh at the centre of the channel [6]. Nup358/RanBP2 and Nup214/CAN have been mapped exclusively to the cytoplasmic side of the NPC, where 50–100 nm long flexible cytoplasmic filaments radiate from the NPC into the cytoplasm. Nup358/RanBP2 has been reported to be the major component of the cytoplasmic NPC filaments [7]. Nup98 is a symmetrical nucleoporin, located on both the cytoplasmic and nuclear sides of the NPC [8]. On the nuclear side of the NPC, Nups such as Nup153 and Nup98 associate with the nuclear basket and with the chromatin both in proximity of and away from the NPC [9].

Many viruses depend on access to the nuclear compartment for replication and have evolved unique strategies to translocate into the nucleus [10][11]. Retroviruses such as Murine Leukaemia Virus (MLV) enter the nucleus during mitotic nuclear membrane disassembly, however other viruses such as herpesviruses and adenoviruses dock their capsids at the NPC and release their genome into the nucleus, while still others (e.g. SV40 and baculovirus) enter in the nucleus with their capsid. The Human Immunodeficiency Virus type 1 (HIV-1), contrary to other orthoretroviruses, has evolved the ability to infect non-dividing cells through active nuclear import of its genome across the intact nuclear membrane through the NPC [12]. Although several viral elements have been proposed to act as determinants of HIV-1 nuclear import, most notably integrase (IN) and the central DNA Flap [13], it is commonly accepted that HIV-1 depends on host cell proteins to achieve translocation. Previous studies have shown the implication of several nucleoporins (Nup62, Nup85, Nup98, Nup107, Nup133, Nup153, Nup160, Nup214/CAN, and Nup358/RanBP2) in HIV-1 nuclear import and/or infectivity [14][15][16][17][18][19][20]. However, the mechanistic implication and individual contribution of these apparently redundant functions remain to be clarified.

Intrigued by the apparent redundancy of Nups potentially assisting HIV-1 translocation through the nuclear pore, and challenged by the lack of mechanistic implications for these Nups, we set out to determine the involvement of four key Nups (Nup358/RanBP2, Nup214/CAN, Nup98 and Nup153) in functionally separable steps of HIV-1 infection and nuclear import. We found that although all four Nups induced defects in HIV-1 infectivity when depleted, only Nup153 actually showed any evidence of participating in HIV-1 translocation through the nuclear pore, probably assisting HIV-1 to exit from the nuclear basket and integrate in the host chromosomal genome possibly in concert with Nup98. Further, we provide evidence that Nup358/RanBP2, identified in a functional genomic screen [17], mediates HIV-1 core docking at the nuclear pore by interacting with capsid. Investigation of the mechanistic role in HIV-1 infection of Nup214/CAN, which was found in another screen [18], revealed that it only plays an indirect role through its effect on mRNA export. Our work sheds light on the participation of key nucleoporins in HIV-1 infection in functionally separable steps of nuclear import.


Efficient Knock-down of Human Nucleoporins using Lentiviral Vector-encoded shRNAs

We constructed lentiviral vector-based small hairpin RNA (LV-shRNA) by inserting shRNA sequences (targeted against Nup358/RanBP2, Nup214/CAN, Nup98 or Nup153, Table S1) downstream of the H1 promoter in the U3 region of the 3′ LTR of the HIV-1-derived vector TRIP-CMV-eGFP. Efficiency of RNA interference was assessed by Western blotting using specific anti-Nup antibodies in LV-shRNA transduced cells compared to non-transduced cells (WT) and cells transduced with LV alone (C) (Fig. 1A). Based on considerations of nucleoporin half-life and stability, as well as cell viability, all knock-down Nup cells were used at 5 days post-transduction (p.t) except for Nup153 knock-down (KD) (2 days p.t), at which time points GFP expression, used to identify transduced cells, was >95%. Contrary to other nucleoporins, Nup153 depletion was deleterious to cell viability after 7 days. Since efficient knock-down was already possible at 2 days post-infection (Fig. 1A), we chose to use all Nup153 knock-downs at this point comparing them with the corresponding control. Other nucleoporins required a further 5 days after transduction for efficient knock-down possibly because they have a longer half life.

To investigate the effects of nucleoporin depletion on the expression and distribution of other nucleoporins, we carried out a series of immunolabelling reactions followed by confocal fluorescence microscopy (Fig. 1B). These were carried out as a set and in the same conditions to allow comparisons of nucleoporin intensity as well as localisation between different LV-shRNA transduced cells. In control cells, nucleoporin labelling generated a characteristic nuclear rim fluorescent signal. Nup98 was also detected in the nucleus, which is concordant with previous studies showing nuclear localisation of Nup98 [21][22][23]. Because of high background of the RanBP2 antibody in immunofluorescent labelling, we also used labelling of RanGAP-1, known to form a complex with Nup358/RanBP2 during interphase and metaphase [24][25][26][27], as an indirect marker for Nup358/RanBP2 localisation. As Nup358 provides a binding site for sumoylated RanGAP at the NPC [24][25], depletion of Nup358 leads to a concomitant loss of RanGAP from the nuclear pore. As expected, the knock-down of Nup358/RanBP2 led to a strong reduction in RanGAP1 perinuclear labelling (Fig. 1B). On the whole, depletion of specific nucleoporins had limited impact on the expression of the other nucleoporins tested and their correct incorporation in NPCs. In some cases, however, some perturbation in nuclear rim labelling could be observed. For instance, Nup98 depletion induced slight perturbations in Nup214 and Nup153 nuclear rim staining, with increased detection of Nup214/CAN in the cytoplasm and of Nup153 in the nucleus (Fig. 1B). This may be due to the concomitant depletion of Nup96, since Nup98 and Nup96 are translated as a polyprotein from the same messenger RNA [28] and Nup96 is a component of the Nup107–160 complex. Moreover, Nup214/CAN depletion led to a slight reduction of RanGAP1 at the nuclear envelope, which is concordant with a previous report that Nup214/CAN stabilises Nup358/RanBP2, and that RNA interference of Nup214/CAN reduces Nup358/RanBP2 levels at the nuclear envelope [29].

We next tested the effect of transduction and nucleoporin depletion on cell viability using an MTT colorimetric assay that measures mitochondrial activity in living cells, through the conversion tetrazole into formazan salts by intracellular NAD(P)H-oxidoreductases. We found no notable differences between knock-down and control cells at the time of infection, except for Nup214 KD cells which had 50% reduced viability (Fig. 1C). Results discussed subsequently provide some explanation for this cytotoxicity. Furthermore, since some nucleoporins are involved in mitotic progression [5], we tested the effect of nucleoporin knock-down on cell cycling. Flow cytometry profiles following propidium iodide labelling revealed no cell cycle arrest at 4, 5 or 7 days post transduction (p.t.) for any of the transduced cells (Fig. 1D and data not shown). A slight increase in G1/S population cells was noted for Nup98 knock-down cells and may be accounted for by the concomitant knock-down of Nup96 since depletion of components of the Nup107–160 complex affects progression through mitosis [30]. Taken together, LV-shRNA-mediated knock-down of Nup358/RanBP2, Nup214/CAN, Nup98 and Nup153 led to an efficient knock-down of the targeted nucleoporins and to minimal cytotoxic or cytostatic effects in the time frame of our experiments (up to 7 days p.t).

Nucleoporin Depletion Disrupts Infectivity of HIV-1

We investigated the importance of the targeted nucleoporins on the early steps of HIV-1 infection in one-cycle infectivity assays. P4-CCR5 cells, which express ß-galactosidase under the control of the HIV-1 long terminal repeat (LTR) promoter transactivated by the viral tat protein, were infected at different MOI of wild-type and vesicular stomatitis virus G protein (VSV-G) pseudotyped HIV-1. Infectivity was measured 48 h post infection (p.i) and luminescence values were systematically normalised for live cell count using protein quantification. Depletion of all tested nucleoporins led to considerable reduction in HIV-1 infection, and with both viral envelopes thus ruling out possible variations due to the viral entry pathway (Fig. 2A and 2B). Infectivity defect ranged from 2-7-fold for Nup98 KD to over 2-log reduction for Nup153 KD in some experiments. Since knock-down of Nup98 leads to a concomitant knock-down of Nup96, which is derived from the same precursor [28], we verified the phenotype obtained with Nup98 KD cells using Nup98 knock-out (KO) mouse embryonic fibroblasts (MEFs), previously shown to specifically target Nup98 and not Nup96 [31]. Transduction of MEFs with TRIP-CMV-eGFP revealed a similar infectivity defect in specific Nup98 KO cells as that seen in Nup98 KD cells (∼4-fold, Fig. S1A), suggesting that the effects we observed with Nup98 knock-down cells are specific to depletion of Nup98. We conclude that depletion of all tested nucleoporins induces considerable disruption of infectivity as measured by single cycle HIV-1 infectivity assays. These results, importantly obtained in a stable knock-down system that offers greater versatility and longer testing conditions, are concordant with previous observations in transient KD cells [14][17][18][15][19][20][16][32]. The efficiency of shRNA technology provided the opportunity to investigate in depth, and in robustly controlled and reproducible conditions, the involvement of each studied nucleoporin in experimentally separable steps of nuclear import (docking, translocation, integration).

Like other lentiviruses, HIV-1 has evolved an active import across the nuclear pore [12] which enables it to infect non-dividing cells. In contrast, the access of retroviruses such as the Murine Leukaemia Virus (MLV) to the host cell chromatin is dependent on mitosis and the associated disassembly of the nuclear envelope. To determine the specificity of each nucleoporin for HIV-1 infection, we investigated the effects of their depletion on MLV infection. P4-CCR5 cells were infected with either HIV-1-Luc or MLV-Luc at the same MOI and luciferase activity measured 48 h p.i. MLV infection was entirely unaffected by the depletion of Nup358/RanBP2, or Nup153 (Fig. 2C). Surprisingly however, knock-down of Nup214/CAN and Nup98 decreased both HIV-1 and MLV infectivity equally (Fig. 2C).

Although All Studied Nucleoporins Perturb HIV-1 Infection when Depleted, Only Nup358/RanBP2, and Nup153 Affect its Nuclear Entry

Having confirmed that all studied nucleoporins are involved in HIV-1 infection, we next sought to confirm the implication of these in HIV-1 nuclear import. Circular forms of non-integrated HIV-1 DNA containing two long terminal repeats (2-LTR) are found exclusively in the nucleus of infected cells and constitute convenient markers of nuclear import. We used quantitative PCR to measure 2-LTR circles in nucleoporin-depleted cells following infection [33]. Results revealed that only two out of the four tested nucleoporins, Nup358/RanBP2, and Nup153, led to a defect in HIV-1 nuclear import when depleted, while Nup214/CAN and Nup98, on the other hand, had no effect on HIV-1 nuclear import (Fig. 3A). No notable differences were observed between wild-type envelope and VSV-G pseudotyped HIV-1.

In addition to their involvement in nuclear import, nucleoporins have also been shown to participate in nuclear events such as chromatin remodelling and regulation of gene expression [22][23]. In the case of HIV-1, recent studies suggest that nucleoporin knock-down can affect the efficiency of provirus integration and/or the selection of chromosomal sites for integration [18][34]. We therefore tested the ability of HIV-1 to integrate within host chromatin in wild-type and knock-down cells using Alu-PCR in infected cells at 24 h p.i. Results showed a significant decrease in integrated HIV-1 DNA in cells depleted of Nup358/RanBP2, Nup98 and Nup153, but not Nup214/CAN (Fig. 3B). For most nucleoporins, this decrease was highly correlated to the decrease observed in 2-LTR circles (Fig. 3B and 3C), thus illustrating a general nuclear import defect rather than a specific block of integration following nuclear import. In the case of Nup98, however, depletion led to a 2-fold decrease in Alu-PCR signal compared to control, whereas no effect was observed for 2-LTR signals (Fig. 3B and 3C), suggesting that Nup98 might play a role in an HIV-1 integration step rather than translocation through the nuclear pore. This is discordant with a previous report suggesting a direct involvement of Nup98 in HIV-1 nuclear entry [14] but concordant with König et al[18]. The observed defects in HIV-1 nuclear import and integration upon Nup153 depletion (<10-fold) are considerably more modest than the general defect in infectivity that we observed (>100-fold decrease). The depletion of Nup153 is associated with pleiotropic effects that likely account for this difference. Firstly, because of its involvement in nuclear basket integrity [35], Nup153 might be critical for the nucleocytoplasmic transport of host co-factors required for HIV-1 transcription or mRNA processing. Secondly, the potential involvement of Nup153 in the spatial organisation of chromatin [36] could mean that its depletion results in an overall decrease in transcription or in the integration of HIV-1 provirus in genomic sites unfavourable for transcription.

Several nucleoporins are known to be involved in mRNA export [37][38][39]. Since Nup214/CAN had no role in HIV-1 nuclear import or integration, we tested whether it might be involved in RNA export. We isolated and compared levels of cytoplasmic and nuclear RNA from control and knock-down cells. Cells depleted in Nup214/CAN and Nup153 had a 70% and 50% reduction, respectively, in cytoplasmic RNA relative to the nuclear fraction (Fig. S1B), confirming their role in a generic nuclear export [27][40]. We conclude that, even though Nup214 was identified as a potential factor involved in HIV infection [18], the reduction on HIV-1 infectivity is in fact linked to a non-specific inhibition of RNA export, and consequently ß-galactosidase or luciferase mRNA export. The inhibition of RNA nuclear export accounts for the partial cytotoxicity measured in Nup214/CAN cells by MTT assay (Fig. 1C) and explains why both HIV-1 and MLV infections are equally affected by Nup214/CAN depletion (Fig. 2C).

Nup358/RanBP2 Mediates HIV-1 Docking at the Nuclear Pore

At present, it is not known how HIV-1 docks at the nuclear pore, nor which nucleoporins if any are responsible for mediating this interaction. Since Nup358/RanBP2 and Nup214/CAN are both cytoplasmically oriented nucleoporins, we hypothesised that they might play a role in assisting HIV-1 docking at the NPC. Our previous work suggests that HIV-1 docks at the nuclear membrane as a complex still containing the viral capsid and that uncoating occurs at the nuclear pore upon completion of reverse transcription [41][42]. Recent work also suggests that HIV-1 capsid may be important for mediating interactions with the nuclear transport machinery or the nuclear pore itself, and that capsid mutations can disrupt these interactions and HIV-1 nuclear import [43][44][19][20]. At 6 h p.i, capsid (p24) fluorescent signal appears as a punctate pattern throughout the cytoplasm and at the nuclear envelope (Fig. 4A, left-hand panels). These bright spots are distinct from the weak and hazy p24 signal that may occasionally be observed in cell nuclei and that likely corresponds to primary antibody non-specific background signal. Our previous work supports that p24 signal at the nuclear rim corresponds to intact capsid cores [41]. A similar punctate pattern is observed with FlAsH-labelling of integrase at 6 h p.i. [45]. To assess whether Nup358/RanBP2 or Nup214/CAN mediate HIV-1 docking at the nuclear envelope, we measured the presence of p24 signal in LV-shRNA transduced and control cells at 6 h p.i. To obtain statistically robust comparison of control and knock-down cells, we used a spot detection software following the location of the nuclear and plasma membranes, and quantified individual p24 spots within 2 pixels of the nuclear membrane (480 nm, Fig. 4A). Around 100 cells per sample from 3 independent experiments were analysed. Using this approach, we found that ∼30% of all cytoplasmic p24 signal was perinuclear in control and Nup214/CAN knock-down cells at 6 h p.i (Fig. 4B). In contrast, depletion of Nup358/RanBP2 led to a decrease in perinuclear p24 signal (∼15% of total cytoplasmic signal), suggesting that Nup358/RanBP2 is involved in HIV-1 docking at the nuclear envelope. Although Nup358/RanBP2 has been shown to associate with microtubules for kinetochore assembly during mitotic nuclear envelope breakdown [26][46], we do not think that this can account for the changes in HIV-1 localisation that we observe upon Nup358/RanBP2 depletion. Indeed, Nup358/RanBP2 does not play a role in cargo movement along microtubules and our microscopy observations were limited to interphasic cells when Nup358/RanBP2’s main location is at the cytoplasmic side of NPC.

To confirm these results using a biochemical assay, and to better distinguish between cytoplasmic and nuclear membrane-associated HIV-1 p24, we carried out cell fractionation in control and all knock-down cells at 6 h p.i, and tested for p24 content in cytoplasmic and nuclear fractions using Western blotting (Fig. 4C). At this time point, p24 signal detected within the nuclear fraction at 6 h p.i corresponds to viral complexes docked at the nuclear membrane and not to intranuclear p24 (Fig. 4A[41][42]. Successful fractionation was monitored using cytoplasmic and nuclear markers (tubulin and lamin A/C, respectively) in addition to actin loading control. Furthermore, we verified that the fractionation protocol does not perturb the nuclear rim localisation of Nup358/RanBP2 or Nup214/CAN using confocal fluorescence microscopy (Fig. S2). At 6 h p.i, nuclear membrane p24 signal was detected for control cells and cells depleted for Nup214/CAN, Nup98 and Nup153 (Fig. 4C). In knock-down cells for Nup358/RanBP2, in contrast, nuclear signal was distinctly low, indicating a defect in docking at the nuclear membrane. To confirm that p24 signal detected in the nuclear fraction is associated with the nuclear membrane and co-localises with NPCs, we labelled fractionated nuclei with appropriate antibodies and monitored protein localisation using microscopy. In control cells, p24 signal formed a ring of punctate labelling coincident with the Nup214 ring (Fig. 4D). In Nup214 knock-down cells, p24 signal was also found as a ring of fluorescent spots around the nucleus. In contrast, p24 signal in nuclear fractions of Nup358 depleted cells was strongly reduced, highlighting the ability of Nup358/RanBP2 to mediate intracellular HIV-1 docking at the nuclear membrane (Fig. 4D). Furthermore, a kinetic measurement of 2-LTR circles in Nup358/RanBP2 depleted cells compared to wild-type controls at 24 h, 48 h, and 72 h p.i. indicated that HIV-1 nuclear import is not delayed but blocked (Fig. 4E), suggesting that Nup358/RanBP2 is necessary for HIV-1 docking.

Nup358/RanBP2 Interacts with HIV-1 in vitro Assembled CA-NC

Based on these results, we next asked whether Nup358/RanBP2 interacts with HIV-1 cores. For this purpose, we tested the binding of Nup358/RanBP2 to in vitro assembled CA-NC complexes that recapitulate the architecture of the bona fide HIV-1 core [47] and were previously used to demonstrate the interaction between rhesus TRIM5α (TRIM5αrh) and the HIV-1 core [48]. In particular, Nup358/RanBP2 contains a cyclophilin homologous domain [49] that bears a high degree of homology with cyclophilin A (Fig. 5A), a well-established HIV-1 capsid interactant [50]. We therefore hypothesised that HIV-1 capsid docks at the nuclear pore via an interaction with the cyclophilin domain of Nup358/RanBP2. To test the implication of the RanBP2 CypA homology domain in interaction with capsid, we deleted the C-terminal residues 2787–3224 encompassing the cyclophilin-homology domain and RanBP homology region 4 (RBH4). We incubated 293T cell lysate expressing the full-length fusion protein GFP-RanBP2 or the GFP-RanBP2-ΔCyp deletion mutant with HIV-1 CA–NC complexes assembled in vitro, as previously described [47][51]. We verified that fusion of Nup358/RanBP2 to GFP did not disrupt localisation at the nuclear membrane and that the folding of the C-terminal deletion mutant GFP-RanBP2- ΔCyp allowed correct localisation at the nuclear membrane (Fig. 5B).