Upon nuclear entry, HIV enhances the formation of CPSF6 puncta, where reverse transcription ends. The treatment with the reversible RT inhibitor nevirapine (NVP) can trap the viral RNA genome in these nuclear niches (Ay et al., 2025; Rensen et al., 2021; Scoca et al., 2023). Once NVP is removed, the trapped vRNA can resume RT entirely within the nucleus, a process we term nuclear reverse transcription. Here we demonstrate that this phenomenon depends on the presence of CPSF6 puncta, which are mainly composed of CPSF6 proteins and viral cores. Disruption of these puncta by high doses of PF74 (25 µM) significantly impairs the rebound of the RT, as evidenced by the absence of luciferase expression, similar to what is observed with complete NVP treatment (Figure 1A, B).

Formation of CPSF6 puncta upon HIV-1 infection hinges on two key events: the entry of the HIV-1 core into the nucleus and the binding of CPSF6 to the HIV-1 core (Blanco-Rodriguez and Di Nunzio, 2021; Blanco-Rodriguez et al., 2020; Buffone et al., 2018; Zila et al., 2021). To determine the contribution of CPSF6’s disordered domains to the formation of CPSF6 puncta upon HIV-1 infection, we correlated the binding of CPSF6 to the HIV-1 core with the formation of CPSF6 puncta. To this end, we first generated CPSF6 knockout (KO) THP-1 cells (Figure 1C, D) to eliminate the interference from the endogenous protein, which could affect the interpretation of results regarding the role of the analysed CPSF6 domains. CPSF6 depletion in THP-1 cells was performed by CRISPR–Cas9 technology. To completely eliminate the expression of the CPSF6 gene, we selected single clones by limiting dilution. We identified a clone that was completely KO for CPSF6, confirmed through western blot and immunofluorescence (Figure 1C, D), and we infected this clone and the control clone with HIV. CPSF6 puncta were detected only in the control-infected cells and not in the KO clone (Figure 1D; Figure 1—figure supplement 1). The viral integrase (IN) was observed within CPSF6 puncta, consistent with previous studies (Francis et al., 2020; Rensen et al., 2021; Scoca et al., 2023), but absent in CPSF6 KO cells where viral IN was observed in the cytoplasm (Figure 1D; Figure 1—figure supplement 1). Thus, we used KO cells for CPSF6 to assess the role of selected CPSF6 domains in HIV-induced condensates (Figure 1—figure supplement 2). We designed various CPSF6 deletion mutants (Figure 2A) to specifically assess the significance of the main disordered regions of CPSF6 protein (Figure 2—figure supplement 1) such as the FG (phenylalanine and glycine) motif, the low complexity regions (LCRs), and the mixed charge domain (MCD), in the ability of CPSF6 to bind to the core and facilitate the formation of CPSF6 puncta. We investigated the role of the FG peptide by generating a mutant that exclusively lacks the FG peptide (∆FG). Previous in vitro studies have shown that the FG peptide binds to the hydrophobic pocket formed between capsid hexamers (Buffone et al., 2018; Price et al., 2014). Here, we want to investigate the role of FG peptide in the context of the protein during the viral infection.

To further explore this, we developed an alternative plasmid by expanding the FG peptide deletion to include surrounding prion-like LCRs (∆FG ∆LCR). These regions, outside the CPSF6 context, have been identified as crucial for facilitating strong CPSF6 binding to capsid lattices (Wei et al., 2022). In our study, we aim to evaluate their role within a more physiological setting. Additionally, we assessed a CPSF6 variant that carries the 15-mer FG peptide flanked by non-LCR sequences, such as those derived from Beta-adducin (ADD2), kindly provided by Mamuka Kvaratskhelia (∆LCR + ADD2). These protein segments are known for their high flexibility, akin to the LCR of CPSF6. Furthermore, to elucidate the contribution of the LCRs of CPSF6 in the formation of CPSF6 puncta, we generated a mutant lacking both LCRs (∆LCR). Analysis of the MCD contribution to both the ability of CPSF6 to bind to the core and formation of CPSF6 puncta was achieved by deleting the MCD and adding 3 nuclear localization signals (3xNLS ∆MCD) since the deletion of the MCD results in a protein that localizes mainly into the cytoplasm (Figure 2A, B; Figure 2—figure supplement 2).

To correlate the ability of CPSF6 to bind to the HIV-1 core with formation of CPSF6 puncta, we expressed wild-type and mutant CPSF6 constructs in THP-1 cells knockout for CPSF6. Subsequently, we infected these cells with HIV-1 and analysed the presence or absence of CPSF6 puncta at 24 hr post-infection. Importantly, for the imaging experiment, we expressed CPSF6 WT and mutants without tags to avoid the formation of aggregates that could interfere with our conclusions. Our data show that HIV-induced CPSF6 puncta can form extremely rarely with the deletion mutant CPSF6 ADD2∆LCR and with the mutant lacking the FG or both the FG peptide and the LCRs (Figure 2B–D; Figure 1—figure supplement 2). However, when we analysed the role of the MCD domain in CPSF6 puncta formation, which was indicated to be important for condensing CPSF6 in NS (Greig et al., 2020), comparing the number of CPSF6 WT puncta induced by HIV infection with CPSF6 mutants revealed that the MCD domain does not play a critical role in HIV-induced CPSF6 puncta formation (Figure 2B–D). In addition, we observed that the majority of analysed CPSF6 3xNLS∆MCD puncta contain vRNA inside, similar to CPSF6 WT puncta (Figure 2E), thus corroborating the lack of a role for this intrinsically disordered domain in HIV-induced CPSF6 puncta. Since the NLS domain from SV40, which replaces the MCD, is highly basic and could potentially induce condensates, we fused CPSF6 with a non-basic NLS (PY-NLS) or removed the NLS entirely. Even though these two proteins do not efficiently enter the nucleus, the few that do manage to reach the nucleus can host viral particles, as evidenced by the presence of IN. Many viruses are typically blocked in the cytoplasm due to the presence of these mutants that are mainly cytoplasmic. However, because we used a high viral dose, the blockage in the cytoplasm was not complete. As a result, the viruses that successfully entered the nucleus induced the formation of puncta associated with CPSF6-deleted mutants, indicating that the MCD is not critical for the formation of HIV-induced CPSF6 puncta (Figure 2F). Similar to the MCD, when we compared CPSF6 truncated for the LCRs with CPSF6 WT, we observed that the LCRs do not contribute to CPSF6 puncta formation. Therefore, the FG peptide alone, without the LCRs, is the only CPSF6 domain required for their formation (Figure 2B–F; Figure 1—figure supplement 2).

Next, we tested the ability of the different CPSF6 deletion mutants for their ability to bind the viral core using a previously described capsid binding assay (Selyutina et al., 2018). Wild-type and mutant CPSF6 proteins were expressed in human 293T cells at similar levels (INPUT) (Figure 3A). Extracts containing wild-type and mutant CPSF6 proteins were incubated with stabilized HIV-1 capsid tubes for 1 hr at 25°C in the presence of 10 µM of PF74, which is a small molecule that competes with CPSF6 for binding to the hydrophobic pocket formed between hexamers that constitute the viral core (Buffone et al., 2018; Price et al., 2014). HIV-1 capsid stabilized tubes were washed, and the bound proteins were eluted using Laemli buffer (BOUND). For every construct, the percentage of bound protein relative to input in the presence or absence of PF74 is shown (Figure 3B). Our results revealed that the absence of the FG peptide (∆FG) entirely abolished CPSF6’s ability to bind to the viral core. In agreement, simultaneous deletion of the FG motif and LCRs (∆FG ∆LCR) resulted in a construct unable to bind to the viral core. Similar outcomes were observed when the LCRs were replaced with sequences derived from ADD2, even if the FG was present.

LCR-FG is notably more disordered than ADD2-FG, containing a high proportion of prolines (48 out of 98 residues), which makes it mostly non-foldable (Figure 4A–H). Since proline is a structure-disrupting residue, LRC-FG is not expected to adopt any secondary structure. In contrast, ADD2-FG contains fewer prolines (15 out of 98 residues) but has many charged residues. It is predicted to form two short α helices and a β strand, arranged as: α helix–FG–β strand–α helix (Figure 4E). ADD2-FG may form a flexible collapsed state, as its oppositely charged residues are evenly distributed, potentially allowing polyelectrostatic compaction. This suggests that FG within ADD2-FG may be less accessible for the interaction with the viral core’s hydrophobic pocket due to its involvement in this collapsed conformational ensemble (Figure 4A–E, Figure 4—figure supplement 1). This aligns with the inability of CPSF6 carrying ADD2 in place of the LCRs to induce CPSF6 puncta (Figure 2B). On the other hand, the deletion of only the two LCRs, while keeping the FG peptide intact, resulted in unexpected findings. The ∆LCR mutant exhibited a stronger binding affinity for the viral core when compared to the wild-type protein (Figure 3B). These results suggest that the LCRs surrounding the FG motif are modulating the affinity of CPSF6 to the viral core, which might be important for function. By contrast, deletion of the MCD (∆MCD) but retention of other regions, such as the FG peptide and the LCRs, demonstrated a binding affinity to the viral core similar to that of the wild-type protein. These results suggest that the MCD domain is not involved in the binding of CPSF6 to the viral core, which is not surprising since the CPSF6 (1–358), which does not have an MCD, binds to the viral core (Lee et al., 2010).

Thus, the viral capsid, through the FG peptide of CPSF6, constitutes the scaffold of HIV-induced CPSF6 puncta.

In summary, our results suggest that the FG peptide is the main determinant involved in the binding of CPSF6 to the viral capsid. Interestingly, our work implies that the LCRs may be modulating the affinity of the FG motif for the viral core (Figure 3B). Recognition motifs that mediate protein-protein interactions, such as the FG motif of CPSF6, are usually embedded within longer IDRs that can modulate affinity of the interaction (Karlsson et al., 2022). Taken together, our data show that the FG peptide coordinates both the binding to the viral core and the induction of CPSF6 puncta. This coordination suggests that the FG peptide plays a critical dual role in recognizing the viral capsid and facilitating the cellular clustering of CPSF6, which may be part of the cellular response to viral entry.

We aimed to investigate the role of the FG domain of CPSF6 in viral infection. To correctly perform the infectivity assay, we generated stable cell clones expressing protein levels comparable to WT cells (Figure 5A, B). Importantly, we assessed the reproducibility of our experiments by freezing and thawing these clones. Thus, we performed infectivity assays in THP-1 macrophage-like cells WT, KO for CPSF6 or complemented with CPSF6∆FG, using both a single-round virus and a replication-competent virus. The results, shown in Figure 5C, D, indicate that complete depletion of CPSF6 reduces infectivity, as measured by luciferase expression in a single-round infection (KO: ~65%; ΔFG: ~74%; compared to WT: 100% on average). Notably, a more pronounced defect in viral particle production was observed when WT virus was used for infection (KO: ~21%; ΔFG: ~16%; compared to WT: 100% on average). According to our results, depletion of full-length CPSF6, or expression of a CPSF6 variant lacking only the FG domain, both reduce viral infection compared to cells expressing full-length CPSF6, indicating that the FG domain plays an important role in HIV-1 infection.

The NS factor SC35, commonly used as a marker of NS, has been detected in HIV-induced CPSF6 puncta (Figure 6A). In this study, we investigated whether HIV-induced CPSF6 mutant puncta are associated with SC35. We infected cells expressing CPSF6 WT, CPSF6 3xNLS ∆MCD, CPSF6 ∆LCR, CPSF6 ∆MCD, and CPSF6 PY NLS ∆MCD, and examined whether the nuclear puncta formed by the various CPSF6 proteins associate with SC35. Imaging analysis revealed no significant difference in the association of SC35 with CPSF6 WT or the CPSF6 mutants (Figure 6B), confirming that the disordered domains, LCRs and MCDs, are dispensable for the formation of HIV-induced CPSF6 puncta that localize in the canonical nuclear niches marked by SC35.

NS factors, particularly those involved in their biogenesis, such as SON and SRRM2 (Ilik et al., 2020), have been identified as constituents of HIV-induced CPSF6 puncta. However, the specific role of NSs in these puncta remains unclear. Most of the existing results have been obtained through immunofluorescence experiments conducted several days post-infection. In this study, we conducted a time course experiment to investigate the biogenesis of HIV-induced CPSF6 puncta, which contain NS factors. Our goal was to capture the fusion event between the HIV-induced CPSF6 puncta and NS, providing insights into the dynamics of how HIV manipulates host nuclear structures during infection. We hypothesized that NS factors might either be recruited during the initial formation of HIV-induced CPSF6 puncta, shortly after the virus is released from the nuclear basket of the NPC, or later via the fusion between NSs and HIV-induced CPSF6 puncta. To investigate this, we performed live imaging to track CPSF6-mNeonGreen and NSs in cells expressing endogenous SRRM2 fused with a Halo tag, using CRISPR Paint (courtesy of Roy Parker) (Lester et al., 2021). We fixed and labelled samples at different time points post-infection, ranging from 6 to 30 h.p.i. (Figure 7A, Figure 7—figure supplements 1 and 2). At 6 h.p.i., 27% of HIV-induced CPSF6 puncta were still individual, compared to only 9% at 30 h.p.i. (Figure 7B). Concurrently, 61% of HIV-induced CPSF6 puncta were fused with NS at 6 h.p.i., rising to 75% at 30 h.p.i. (Figure 7B). This indicates a progressive increase in the number of HIV-induced CPSF6 puncta fusing with NS over time (Figure 7C). Overall, we detected individual CPSF6 puncta (green) and NSs (red) that quickly fused, confirming that this fusion occurs within the two independent puncta, CPSF6 and SRRM2, rather than during the formation of HIV-induced CPSF6 puncta (Figure 7A–C, Figure 7—videos 1 and 2).

Taken together, these results suggest that HIV-induced CPSF6 puncta first form independently of NS and later fuse with NS, causing an enlargement of NS as part of the hijacking process by HIV-1.

Macrophage-like cells, THP-1, were depleted for SON or SRRM2 using AUMsilence ASO technology (Gao et al., 2024; Marasca et al., 2022; Mazzeo et al., 2024; Zhang et al., 2023). The level of depletion of SON and SRRM2 was evaluated by immunofluorescence and western blot using antibodies against SON and SRRM2 (Figure 8A). Both depleted cells were analysed also for the presence of NS, labelled by SC35 (Figure 8—figure supplement 1). However, recent findings suggest that the primary target of the SC35 mAb is SRRM2 (Ilik et al., 2020). To confirm this, cells depleted of SRRM2 were labelled with antibodies against both SRRM2 and SC35 (Figure 8—figure supplement 1). We have observed that the reduction of SRRM2 resulted in a slight decrease in the mean intensity of SC35, whereas the depletion of SON (Figure 8—figure supplement 1) did not have the same effect.

Subsequently, we infected THP-1 depleted cells for SRRM2 and SON or control cells with HIV-1 and we fixed them at 48 hr post-infection for immunofluorescence. We calculated the percentage of CPSF6 puncta in HIV-infected THP-1 control cells (approximately 78%) and in HIV-infected THP-1 cells depleted for SRRM2 (about 43%) and SON (around 66%). Results from our experimental conditions indicate that the partial depletion of SRRM2 affects the formation of HIV-induced CPSF6 puncta, while the depletion of SON slightly reduces their establishment (Figure 8B).

CPSF6 contains several disordered regions (Di Nunzio et al., 2023) as well as NS factors (Ilik et al., 2020). Previous studies have established the role of IDR of SRRM2 in the biogenesis of NS (Ilik et al., 2020).

Here, we investigate the role of IDRs within NS factors, with a specific emphasis on SRRM2, in the fusion and stabilization of HIV-induced CPSF6 puncta. To address this inquiry, we utilized previously published HEK293 cell lines generated using the CRISPaint system (Lester et al., 2021), comprising two distinct lines: HaloTag SRRM2, SRRM2 full-length (FL) (1–2748 aa) fused with Halo tag (insertion at amino acid [aa] 2708), and a cell line lacking the C-terminal IDR of SRRM2, known as ∆IDR HaloTag SRRM2 (1–429 aa, with halo insertion at aa 430).

As anticipated, the truncated form of SRRM2 displayed a more diffuse distribution within the nucleus, without recruitment to NS (Figure 8—figure supplement 2), and consequently lacked nuclear puncta (Figure 8C, top panels). On the other hand, the number of SON puncta was highly similar between the two cell lines (Figure 8C, panels on the bottom, Figure 8—figure supplement 3).

Subsequently, we quantified the formation of CPSF6 puncta in both cell lines infected with HIV-1. No significant difference in HIV-induced CPSF6 puncta formation was observed between HEK293 and HEK293 carrying the SRRM2 halo tag (~27%). However, a substantial reduction in HIV-induced CPSF6 puncta was evident in the cell line carrying the SRRM2 form that lacks the IDR (~11%) (Figure 8D). Collectively, these results underscore the pivotal role of the SRRM2 IDR in the stabilization of HIV-induced CPSF6 puncta through their fusion with NSs (Figure 7A–C).

THP-1 cells (ATCC) are immortalized monocytic cells, which, once seeded, differentiate into macrophage-like cells under phorbol 12-myristate 13-acetate (PMA) treatment (160 nM). THP-1 cells were also engineered to knock out CPSF6. These cells were cultivated in RPMI 1640 medium supplemented with 10% foetal bovine serum (FBS) and 1% penicillin–streptomycin solution (100 U/ml). HEK293T cells (ATCC) are human embryonic kidney cells used to produce LVs. For Figure 2, we used two engineered HEK293 strains, Halo tagged SRRM2 HEK293 cells and Halo tagged SRRM2 ∆IDR HEK293 cells, kind gifts from Roy Parker’s lab (Lester et al., 2021), characterized by the Halo tagged SRRM2 protein. In Halo tagged SRRM2 ∆IDR HEK293 cells, Parker’s lab also deleted the SRRM2 sequence encoding for aa 430–2748. All the HEK293 strains were cultivated in Dulbecco’s modified Eagle medium supplemented with 10% FBS and 1% penicillin–streptomycin (100 U/ml). All cells used in the study were tested for mycoplasma and they were negative.

All Escherichia coli bacteria strains were grown in Luria-Bertani (LB) medium at 37°C. DHα competent cells and Stellar Competent Cells were used for molecular cloning, while E. coli One-Shot BL21star (DE3) cells were exploited for protein production.

To express the WT CPSF6/WT CPSF6-mNeonGreen and the mutant CPSF6/mutant CPSF6-mNeonGreen clones, the correspondent coding sequences were engineered in pSICO plasmids. The two original plasmids used were pSICO CPSF6-mNeonGreen and pLPCX CPSF6 ADD2, a gift from Mamuka’s lab. HIV-1ΔEnvIN_HA LAI (BRU) plasmid encodes the ΔEnvHIV-1 LAI (BRU) viral genome where the IN protein is fused to the HA tag while pNL4.3 ∆env ∆Nef IRES GFP plasmid encodes the ΔEnvHIV-1 NL4.3 viral genome and contains also a GFP sequence headed by an IRES. The pNL4.3 ∆env ∆NefLuc has the Luciferase cDNA as reporter gene.

To target CPSF6, three different crRNAs were used simultaneously (specific sequence: 5′-TCGGGCAAATGGCCAGTCAAAGG-3′, 5′-AGGACGGGGCCGTTTTCCAGGGG-3′, and 5′-CATGTAATCTCGGTCTTCTGGGG-3′, all ordered from Integrated DNA Technologies, IDT). Pre-designed unspecific crRNA was used as control (IDT). crRNA and tracrRNA were resuspended in IDT Duplex Buffer according to the manufacturer’s instructions. On the day of the nucleofection, duplexes were formed by mixing equimolar concentration of crRNA and tracrRNA, followed by 5 min annealing at 95°C. RNA duplexes were then mixed (1:2) with TrueCut Cas9 Protein v2 for 10 min at room temperature to generate ribonucleoprotein (RNP) complexes. 2 × 10⁵ THP-1 cells were resuspended in P3 Primary Cell Nucleofector Solution, mixed with RNP and Alt-R Cas9 Electroporation Enhancer (90 pmol, IDT), and nucleofected in a 4D-Nucleofector System using the P3 Primary Cell 4D-NucleofectorX Kit S (program FI-110). After nucleofection, cells were seeded in complete RPMI medium with 20% FBS. Three days after nucleofection, cells were plated for clonal selection.

Seventy-two hours post nucleofection, cells were diluted in RPMI medium containing 20% FBS and plated in five 96-well plates at 1 and 5 cells/well condition. After 1 month, selected microcolonies (50–100) are placed into 24-well plates. Once the wells were near confluence, cells were transferred into the well of a 6-well plate. After growing for another 1 month, cells were proceeded for western blot.

Proteins were extracted on ice from THP-1 cells using RIPA buffer (20 mM HEPES, pH 7.6, 150 mM NaCl, 1% sodium deoxycholate, 1% Nonidet P-40, 0.1% SDS, 2 mM EDTA, complete protease inhibitor), and protein concentration was quantified using a detergent-compatible (DC) protein assay (Pierce BCA Protein Assay Kit) with bovine serum albumin (BSA) as a standard. 90 µg of total protein lysate were loaded onto SDS–PAGE 4–12% Bis-Tris gels (Invitrogen); an Ab rabbit anti-CPSF6 (1:500) and an anti-rabbit HRP-conjugated (1:5000) were used for the detection of CPSF6, whereas the normalization was done by an Ab anti-actin HRP-conjugated (1:3000). Visualization was carried out using an ECL solution.

AUMsilenceTM 352 ASOs were synthesized by AUM BioTech, LLC (Philadelphia, USA). THP-1 negative control, THP-1 KD SRRM2, and THP-1 KD SON cells were differentiated with PMA (160 nM) for 48 hr then incubated 72 hr with 10 µM of a AUMsilenceTM ASOs complementary to the mRNA of SRRM2 and SON, respectively (scramble control AUMscrambleTM). All cells were kept in an incubator at 37°C and 5% CO₂. These cells were cultivated in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin–streptomycin solution (100 U/ml).

pSICO CPSF6-mNeonGreen was used to generate deletion mutants by excising different regions from the original sequence according to Supplementary file 1. The mutants were produced with and without the mNeonGreen tag, except for pSICO CPSF6-∆MCD and pSICO CPSF6-∆MCD PY NLS. All primers are specified in Supplementary file 2.

For pSICO CPSF6-mNeonGreen ∆LCR ADD2, CPSF6 ∆LCR ADD2 sequence was amplified by PCR from pLPCX CPSF6 ADD2 and specific primers were designed to add also BamHI restriction sites at the extremities. Phusion Flash High-Fidelity PCR Master Mix was used and the reaction was performed according to the manufacturer datasheet (100 ng template DNA and 56° annealing temperature). PCR products were treated with DpnI for 1 hr at 37°C and then digested with BamHI restriction enzyme at 37°C for 1 hr. 2 µl of the backbone pSICO CPSF6-mNeonGreen were digested with BamHI at 37°C for 3 hr and further treated with CIP. After gel extraction of the DNA and purification, insert and backbone were ligated with T4 Ligase for 2 hr at 22°C. 5 µl of the product were used to transform 50 µl of DH5α bacteria (30 min at 4°C, 45 s at 42°C, 2 min at 4°C, incubation in SOC medium for 1 hr at 37°C and plating on LB agar dishes).

All the other mutants were obtained using the In-FusionSnap Assembly protocol, which allowed us to amplify the original plasmid, deleting the small region of interest. The different primers used are reported in Supplementary file 2. As reported in the In-FusionSnap Assembly protocol , 5 ng of the original plasmid were amplified with PrimeSTARMax DNA polymerase in 35 PCR cycles (10 s at 98°C, 15 s at 55°C, 5 s/kb at 72°C). The PCR products were then digested for 1 hr at 37°C using DpnI to get rid of the original plasmid. After plasmid cleaning up, it was circularized through a ligation of 15 min at 50°C. 2.5 µl of products were used in Stellar Competent Cells’ transformation, following the same procedure used for DH5a transformation, previously explained.

LVs and HIV-1 viruses were produced by transient transfection of HEK293T cells through calcium chloride co-precipitation. Co-transfection was performed as follows: for LVs, 10 µg of transfer vector, 10 µg of packaging plasmid (gag-pol-tat-rev), and 2.5 µg of pHCMV-VSV-G envelope plasmid; for VSV-G/HIV-1ΔEnvIN_HA LAI (BRU) -vpx viruses and VSV-G/pNL4.3 ∆env ∆Nef IRES GFP-vpx, 10 µg HIV-1ΔEnvIN_HA LAI (BRU) plasmid or pNL4.3 ∆env ∆Nef IRES GFP or pNL4.3 ∆env∆R∆Nef Luc, 2.5 µg of pHCMV-VSV-G plasmid and 3 µg of SIVMAC vpx (Durand et al., 2013) or full-length virus 10 µg HIV-1 NL4.3AD8 and 3 µg of SIVMAC vpx. After the collection of the supernatant 48 hr post-transfection, lentiviral particles were concentrated by ultracentrifugation for 1 hr at 22,000 rpm at 4°C and stored at −80°C. LVs and viruses were tittered by qPCR in HEK293T cells 3 days post-transduction.

THP-1 ctrl CRISPR clone 2 cells and THP-1 (duplex1-2-3 CRISPR) KO clone 4 cells were differentiated with PMA (160 nM) for 72 hr then transduced with different mutants of CPSF6 for 72 hr (MOI = 1) and then infected for 30 hr with VSV-G/HIV-1ΔEnvIN_HA LAI (BRU) -vpx (MOI = 10) in presence of Nevirapine (10 µM). The medium was always supplemented with PMA (160 nM).

For Halo tagged SRRM2 HEK 293 cells and Halo tagged SRRM2 ∆IDR HEK 293 cells, 2 × 10⁵ cells were seeded on coverslips coated with polylysine in complete growth medium (DMEM, GlutaMAX-I, 10% FBS, and 1% P/S) and incubated at 37°C (5% CO₂) for 24 hr. Cells were then infected with the VSV-G/HIV-1ΔEnvIN_HA LAI (BRU) (MOI 10) in complete growth medium supplemented with Nevirapine (10 mM) for 24 hr.

THP-1 control (scramble), THP-1 KD SRRM2 cells, and THP-1 KD SON cells were seeded on coverslips and differentiated with PMA (160 nM) for 72 hr. Then incubated for 48 hr with 10 µM of a FANA ASOs (scramble control FANA (SCR-FANA), SRRM2-FANA, and SON-FANA). ASOs used in this study were designed and synthesized by AUM LifeTech (Philadelphia, PA, USA). Next, cells were infected with the pNL4.3 ∆env ∆Nef IRES GFP-VPX (MOI 25) in complete growth medium and incubated at 37°C in 5% CO₂ for 4 days.

In the four time points experiment, THP-1 cells were differentiated with PMA (160 nM) for 72 hr. Cells were then infected with VSV-G/HIV-1ΔEnvIN_HA LAI (BRU) -vpx (MOI = 10) in the presence of Nevirapine (10 µM), in complete growth medium supplemented with PMA (160 nM), and incubated at 37°C in 5% CO₂ for 6, 9, 12, 30 hr post-infection.

The luciferase assay was performed using Luciferase Assay System (Promega, E4030). Infected differentiated THP-1 cells were washed with PBS and then incubated with Reporter Lysis Buffer at –80 °C overnight. After the thawing, cells were scraped, collected and centrifuged at 12,000 × g for 5 min to pull down all the cell debris. The supernatant was collected, and a part was used to react with the luciferase substrate. Data were acquired immediately at the luminometer.

THP-1 cells differentiate into macrophage-like cells within 48 hr under PMA treatment (160 nM). THP-1 cells were cultured at 37°C in 5% CO₂ in RPMI Medium 1640 supplemented with GlutaMAXTM-I, 10% FBS, 1% P/S (100 U/ml), MEM non-essential amino acid solution (0.1 mM), sodium pyruvate (1 mM), and HEPES buffer solution (10 mM). THP-1 cells differentiated into macrophages are infected with HIV-1 NL4.3AD8 with Vpx (MOI 6). Twenty-four hours after infection, the cells were washed and RPMI was added to the cells. Three days post-infection, the supernatant is collected and centrifuged. Total viral RNA was extracted from the supernatant via the QIAamp Viral RNA Mini Kit (QIAGEN, Hilden Germany) following the manufacturer’s protocol. Total viral RNA was eluted into 60 μl of DEPC. vRNA is reverse transcribed into DNA with the Maxima H minus reverse transcriptase (Thermo Scientific, EP0752) following the manufacturer’s protocol and the following primers were used (forward: GCCTCAATAAAGCTTGCCTTGA, reverse: TGACTAAAAGGGTCTGAGGGATCT). DNA copies for each sample were determined by qPCR and the SYBR green dye. The absolute DNA copies were calculated from a standard curve run in parallel using HIV-1 NL4.3AD8 plasmid. The qPCR cycle was as follows: 15 min at 95°C, 40 cycles (15 s at 95°C, 30 s at 58°C, 30 s at 72°C), 15 s at 95°C, 15 s at 55°C, 20 min melting curve until reaching 95°C, 15 s at 90°C.

Halo tagged SRRM2 HEK293 cells were seeded and transduced with CPSF6-mNeonGreen lentiviral vector (MOI 0.5) for 24 hr. 0.3 × 10⁶ cells were then transferred on poly-l-lysin coated Ibidi-dishes and infected with VSV-G/HIV-1ΔEnvIN_HA LAI (BRU) (MOI = 40) in presence of Nevirapine (10 µM) for 2 hr. After having changed the medium and labelled with HaloTagTMR Ligand (5 µM) (see ‘HaloTag labelling’ section) and Hoechst 333342 diluted 1:80000 in complete DMEM medium, 4D movies were acquired.

3D movies were acquired using a Nikon Ti2-E Confocal Inverted Spinning Disk microscope, using a 63x objective (Plan Apochromat, oil immersion, NA = 1.4) and a sCMOS Hamamatsu camera, Orca Flash 4. Pixel size 6.5 µm, 2048 × 2044 pixel, QE 82%. For the Z stuck, a Z piezo stage was used, with 0.33 µm interval. For live imaging acquisitions, cells were placed in an environmental chamber with 37°C temperature, 5% CO₂ and 21% O₂.

pET-11a vectors were used to express the HIV-1 capsid protein. Point mutations, A14C and E45C, were introduced using the QuikChange II site-directed mutagenesis kit (Stratagene) according to the manufacturer’s instructions. All proteins were expressed in E. coli one-shoot BL21star (DE3) cells (Invitrogen). Briefly, LB medium was inoculated with overnight cultures, which were grown at 30°C until mid log-phase (A₆₀₀, 0.6–0.8). Protein expression was induced with 1 mM isopropyl-β-d-thiogalactopyranoside overnight at 18°C. Cells were harvested by centrifugation at 5000 × g for 10 min at 4°C, and pellets were stored at −80°C until purification. Purification of capsid was carried out as follows. Pellets from 2 l of bacteria were lysed by sonication (Qsonica microtip: 4420; A = 45; 2 min; 2 s on; 2 s off for 12 cycles), in 40 ml of lysis buffer (50 mM Tris pH = 8, 50 mM NaCl, 100 mM β-mercaptoethanol and Complete EDTA-free protease inhibitor tablets). Cell debris was removed by centrifugation at 40,000 × g for 20 min at 4°C. Proteins from the supernatant were precipitated by incubation with 1/3 of volume of saturated ammonium sulfate containing 100 mM β-mercaptoethanol for 20 min at 4°C and centrifugation at 8000 × g for 20 min at 4°C. Precipitated proteins were resuspended in 30 ml of buffer A (25 mM MES pH6.5, 100 mM β-mercaptoethanol) and sonicated two to three times (Qsonica microtip: 4420; A = 45; 2 min; 1 s on; 2 s off). The sample was dialysed three times in buffer A (2 hr, overnight, 2 hr). The sample was sonicated and diluted in 500 ml of buffer A and was chromatographed sequentially on a 5-ml HiTrap Q HP column and on a 5-ml HiTrap SP FF column (GE Healthcare), both pre-equilibrated with buffer A. The capsid protein was eluted from HiTrap SP FF column using a linear gradient from 0 to 2 M of NaCl. Absorbance at 280 nm was checked to take the eluted fraction that had higher protein levels. Pooled fractions were dialysed 3 times (2 h, overnight, 2 h) in storage buffer (25 mM MES, 2 M NaCl, 20 mM β-mercaptoethanol). Sample was concentrated using centricons to a concentration of 20 mg/ml and stored at –80 °C.

1 ml of monomeric capsid (3 or 1 mg/ml) was dialysed in SnakeSkin dialysis tubing 10,000 MWCO (Thermo Scientific) against a buffer that is high in salt and contains a reducing agent (buffer 1: 50 mM Tris, pH 8, 1 M NaCl, 100 mM β-mercaptoethanol) at 4°C for 8 hr. Subsequently, the protein was dialysed against the same buffer without the reducing agent β-mercaptoethanol (buffer 2: 50 mM Tris, pH 8, 1 M NaCl) at 4°C for 8 hr. The absence of β-mercaptoethanol in the second dialysis allows formation of disulphide bonds between cysteine 14 and 43 inter-capsid monomers in the hexamer. Finally, the protein is dialysed against buffer 3 (20 mM Tris, pH 8.0, 40 mM NaCl) at 4°C for 8 hr. Assembled complexes were kept at 4°C up to 1 month.

Human HEK293T cells were transfected for 24 hr with a plasmid expressing the specified CPSF6 variant tagged with mNeonGreen. Cell media was completely removed and cells were lysed in 300 μl of capsid binding buffer (CBB: 10 mM Tris, pH 8.0, 1,5 mM MgCl₂, 10 mM KCl) by scrapping off the plate. Cells were rotated at 4°C for 15 min and then centrifuged to remove cellular debris (21,000 × g, 15 min, 4°C). Cell lysates were incubated with stabilized HIV-1 capsid tubes for 1 hr at 25°C. Subsequently, stabilized HIV-1 capsid tubes were washed by pelleting the complexes by centrifugation at 21,000 × g for 2 min. Pellets were washed using resuspension in CBB or PBS. Pellets were resuspended in Laemmli buffer 1X and analysed by western blotting using anti-p24 or anti-mNeonGreen antibodies.

Intrinsic disorder propensity of CPSF6 was evaluated using the Rapid Intrinsic Disorder Analysis Online platform (RIDAO) (https://ridao.app/) designed to predict disordered residues and regions in a query protein based on its amino acid sequence (Dayhoff and Uversky, 2022). RIDAO yields results by combining the outputs of several commonly used per-residue disorder predictors, such as PONDR VLXT (Romero et al., 2001), PONDR VL3 (Peng et al., 2006), PONDR VLS2B (Peng et al., 2005), PONDR FIT (Xue et al., 2010), as well as IUPred2 (Short) and IUPred2 (Long)(Dosztányi et al., 2005a; Dosztányi et al., 2005b). RIDAO also computes a mean disorder score for each residue based on these. In the resulting intrinsic disorder profile, the disorder score of 0.5 is the threshold between order and disorder, where residues/regions above 0.5 are disordered, and residues/regions below 0.5 are ordered. The disorder score of 0.15 is the threshold between order and flexibility, where residues/regions with the disorder scores above 0.15 but below 0.5 are flexible, and residues/regions below 0.15 are highly ordered.

Amino acid compositions of the intrinsically disordered C-terminal domain (residues 261–358) of human CPSF6 and its different variants (CPSF6 ΔFG ΔLCR, CPSF6 ΔLCR, CPSF6 ΔFG, and CPSF6 ADD2 ΔLCR) were analysed to evaluate the relative abundance of prion-like low complexity region (LCR) defining uncharged, charged, and Pro residues in these protein regions. The corresponding values of the relative abundance of these residue groups were calculated by dividing numbers of prion-like LCR defining uncharged (Ala, Gly, Val, Phe, Tyr, Leu, Ile, Ser, Thr, Pro, Asn, Gln), charged (Asp, Glu, Lys, Arg), and Pro residues by total number of amino acids in the corresponding protein fragments. As references, we used the corresponding data for protein sequences deposited to the UniProtKB/Swiss-Prot database that provides information on the overall distribution of amino acids in nature (Bairoch et al., 2005; PDB Select 25; Berman et al., 2000), which is a subset of structures from the Protein Data Bank with less than 25% sequence identity, biased towards the composition of proteins amenable to crystallization studies; and DisProt 3.4 that is comprised of a set of consensus sequences of experimentally determined disordered regions (Sickmeier et al., 2007). Per-residue intrinsic disorder propensities of the LCR-FG and ADD2-FG sequences were evaluated by PONDR VLXT (Romero et al., 2001), which is sensitive to local peculiarities of the amino acid sequences freely available at http://www.pondr.com/ (accessed on August 3, 2024). Linear distribution of the net charge per residue within the LCR-FG and ADD2-FG sequences was evaluated by CIDER (Holehouse et al., 2017), which is a webserver for the analysis of a wide range of the physicochemical properties encoded by IDP sequences freely available at http://pappulab.wustl.edu/CIDER (accessed on August 3, 2024). Secondary structure propensities of the LCR-FG and ADD2-FG sequences were evaluated by PSIPRED (McGuffin et al., 2000), which is a highly accurate secondary structure prediction method freely available to non-commercial users at http://globin.bio.warwick.ac.uk/psipred/ (accessed on August 3, 2024).

All images were analysed using Fiji Software. More in detail, for the count of the NSs, a macro was computed to segment cellular nuclei and to select and count the NSs. For the nuclei segmentation, a Gaussian Blur with sigma = 2 was used and for the channel related to the NS, the set threshold was at 7000–46,012 (min and max). For the counting, a ‘size = 10 – Infinity summarize’ was used for the nuclei, whereas for the NSs the size was reduced to ‘size = 0 – Infinity summarize’.

For live imaging analysis, Arivis software was used to reconstruct the 3D movies.

All data were statistically analysed with GraphPad Prism 9 (GraphPad Software, La Jolla, California USA, https://www.graphpad.com). Calculations were performed and figures were drawn using Excel 365 or GraphPad Prism 8.0. Statistical analysis was performed, with Wilcoxon matched paired t-tests or Mann–Whitney unpaired t-tests. Spearman correlation coefficients (r) were calculated using GraphPad Prism.