Prion-like low complexity regions enable avid virus-host interactions during HIV-1 infection

Abstract

Cellular proteins CPSF6, NUP153 and SEC24C play crucial roles in HIV-1 infection. While weak interactions of short phenylalanine-glycine (FG) containing peptides with isolated capsid hexamers have been characterized, how these cellular factors functionally engage with biologically relevant mature HIV-1 capsid lattices is unknown. Here we show that prion-like low complexity regions (LCRs) enable avid CPSF6, NUP153 and SEC24C binding to capsid lattices. Structural studies revealed that multivalent CPSF6 assembly is mediated by LCR-LCR interactions, which are templated by binding of CPSF6 FG peptides to a subset of hydrophobic capsid pockets positioned along adjoining hexamers. In infected cells, avid CPSF6 LCR-mediated binding to HIV-1 cores is essential for functional virus-host interactions. The investigational drug lenacapavir accesses unoccupied hydrophobic pockets in the complex to potently impair HIV-1 inside the nucleus without displacing the tightly bound cellular cofactor from virus cores. These results establish previously undescribed mechanisms of virus-host interactions and antiviral action.

Fig. 1. Biochemical characterization of CPSF6, NUP153 and SEC24C interactions with HIV-1 cores and CA tubes.

Representative immunoblots showing CA nanotube mediated co-pelleting of endogenous CPSF6 (a), NUP153 (b) and SEC24C (c) from MT4 cell lysates. The proteins of interest were visualized by antibodies ab175237(Abcam) against CPSF6, NB100-93329 (Novus) against NUP153, ab122633 (Abcam) against SEC24C. Lane 1: cell lysate. Lane 2: supernatant or unbound fraction after pelleting in the absence of CA tubes. Lane 3: supernatant or unbound fraction after co-pelleting with pre-formed CA tubes. Lane 4: pelleted or bound fraction in the absence of CA tubes. Lane 5: pelleted or bound fraction in the presence of CA tubes. The experiments were repeated 3 times independently with similar results. Quantitation of GST-mediated affinity pull-down of native HIV-1 cores bound to indicated concentrations of GST-CPSF6_261-358(LCR-FG-LCR) vs GST-CPSF6(FG)/nonLCR (d), GST-NUP153_1306-1450(LCR-FG-LCR) vs GST-NUP153(FG)/nonLCR (e); and GST-SEC24C_196-314(LCR-FG-LCR) vs GST-SEC24C(FG)/nonLCR (f). The results were analyzed by Origin 2019 (v.9.6) software to determine binding K_d values. Each data point represents mean values + /− SD from three independent experiments. Source data are provided as a Source Data file. g Quantitation of GST-mediated affinity pull-down of isolated native HIV-1 cores vs crosslinked CA hexamers with 2 μM GST-CPSF6_261-358(LCR-FG-LCR) (left), GST-NUP153_1306-1450(LCR-FG-LCR) (middle) and GST-SEC24C_196-314(LCR-FG-LCR) (right). Mean values + /− SD from three independent experiments are shown. Source data are provided as a Source Data file. h Representative immunoblots of three independent experiments showing co-pelleting of GST-CPSF6_261-358(LCR-FG-LCR) (top), GST-NUP153_1306-1450(LCR-FG-LCR) (middle) and GST-SEC24C_196-314(LCR-FG-LCR) (bottom) with pre-formed CA nanotubes. The experiments were repeated 3 times independently with similar results.

Fig. 2. LCR sequences are critical for CPSF6 interaction with HIV-1 cores in infected cells.

a PLAs showing association of HA-tagged ectopically expressed WT and chimeric CPSF6 proteins with incoming HIV-1 particles in HEK293T CKO cells. The cells were fixed at 6 hpi and PLAs were performed using anti-HA (ab236632, Abcam) and anti-HIV-1 CA specific antibodies (ARP-4121, NIH AIDS Reagent Program). The representative images show PLA puncta (red) and nuclei stained with DAPI (blue). Scale bar is 2 μm. b Quantitative results showing numbers of PLA puncta per cell by analyzing twenty-five cells for each sample. The averaged data (+/− SD) from three independent experiments are shown. Statistical significance of comparison of WT versus the chimeric CPSF6 proteins was determined by Student’s two-sample, two-tailed t-test. CPSF6/AD: p = 3.1E-12; CPSF6/NE: p = 2.4e-12; CPSF6/FU: p = 0.54; CPSF6/CD: p = 0.26; CKO: p = 8.8E-13. P > 0.05 was considered not significant (ns) and p < 0.0001 was considered highly significant (***). Source data are provided as a Source Data file. AD ADD2, NE NEURM, FU FUS, CD CDK19, WT wild-type CPSF6, CKO CPSF6 knock-out, LCR low complexity region.

Fig. 3. CPSF6 LCR effects on HIV-1 nuclear import and targeting to NS.

a–c HEK293T CKO cells were back-complemented with indicated HA tagged CPSF6 (CPSF6) proteins and infected with VSV-G pseudotyped HIV-1 viruses (MOI 1) labeled with INsfGFP (green) for 4 h. Cells were fixed and immunostained for SON nuclear speckle (NS) marker (red) and nuclei with SiR-Hoechst (blue). a Single Z-stack images. Scale bar is 5 µm. b Quantitation of the number of IN-labeled viral replication complexes (VRCs) >0.5 µm of the nuclear periphery. c Quantification of IN-VRCs colocalized with SON-NSs. Mean values obtained from each experiment and SEM from a total of 4 independent experiments are shown in b, c. Each independent experiment analyzed >30 nuclei for each construct. Statistical significances of comparison of CPSF6/WT versus CPSF6/AD, CPSF6/NE, CPSF6/FU, CPSF6/CD and CKO (in b, c) were determined by Student’s two-sample, two-tailed t-test. P-values of CPSF6/AD, CPSF6/NE, CPSF6/FU, CPSF6/CD and CKO are 1.1E-05, 1.7E-05, 0.6, 0.09, 1.7E-05 (in b) and 8.8E-05, 9.2E-05, 0.69, 0.22, 8.5E-05 (in c), respectively. P > 0.05 was considered not significant (ns) and p < 0.001 was considered highly significant (***). Source data are provided as a Source Data file. CPSF6/AD CPSF6/ADD2, CPSF6/NE CPSF6/NEURM, CPSF6/FU CPSF6/FUS, CPSF6/CD CPSF6/CDK19, CPSF6/WT CPSF6/wild-type CPSF6, CKO CPSF6 knock-out, LCR low complexity region.

Fig. 4. The CPSF6 LCR effects on integration site selection.

Distribution of HIV-1 integration sites in SPADs (a), gene dense regions (b), and LADs (c). CPSF6/AD, CPSF6/NE, CPSF6/FU, and CPSF6/CD constructs were expressed alongside WT CPSF6 (CPSF6/WT) in CKO cells. The random integration control (RIC) was previously described. With the exception of CKO cells, all experiments were done in duplicate. Data points from two independent experiment are shown and corresponding mean values are indicated by lines (also see Supplementary Table 2). Source data are provided as a Source Data file. CPSF6/AD CPSF6/ADD2, CPSF6/NE CPSF6/NEURM, CPSF6/FU CPSF6/FUS, CPSF6/CD CPSF6/CDK19, CPSF6/WT CPSF6/wild-type CPSF6, CKO CPSF6 knock-out, LCR low complexity region.

Fig. 5. Cryo-EM analysis of CPSF6(LCR-FG-LCR) binding to IP6-stabilized CA tubes.

Images of CA(A92E) tubes in the presence of WT GST-CPSF6_261-358 (a); GST-CPSF6_261-358(ΔFG), which contains LCR sequences but lacks the CA binding FG residues (b); and GST-CPSF6/AD(FG-nonLCR), which contains the CA binding 15-mer FG peptide but where flanking LCRs are replaced with nonLCR sequences from ADD2 (c). Partially ordered additional density (indicated by yellow arrow heads) along the surface of CA nanotubes was consistently observed in CA + WT GST-CPSF6_261-358(LCR-FG-LCR) samples (a, top). This additional density was also evident in 2D class averages (a, middle, yellow arrow heads) and in 1D density profiles (a, bottom). In contrast, no additional density was apparent after addition of GST-CPSF6_261-358(ΔFG) (b) or GST-CPSF6/AD(FG-nonLCR) (c) to CA nanotubes. The experiments were repeated 3 times independently with similar results. Front (d) and end (e) views of a cryo-EM map (~7.9 Å overall resolution) of a CA(A92E) nanotube in the presence of GST-CPSF6_261-358(LCR-FG-LCR), filtered by local resolution and colored by radial distance from the center of the nanotube (CA in gray, CPSF6 in orange, GST dimers in semi-transparent magenta to help visualize CA and CPSF6 density). Front views of the same cryo-EM map, sliced at the two planes indicated in e, show continuous CPSF6 density arising from LCR interactions (f), as well as CPSF6 discrete contact points with the CA hexamer lattice (g) and their correspondence to continuous LCR density (h) (the doted circles mark the positions of CPSF6-CA contact points shown in g).

Fig. 6. HDX-MS analysis of CPSF6(LCR-FG-LCR) binding to IP6-stabilized CA tubes.

a The 10 s HDX results for the CA peptide (aa 203-213) that is specifically protected upon addition of WT GST-CPSF6_261-358 but not GST-CPSF6_261-358(ΔFG). Gray and cyan bars show HDX values for the indicated CA peptide in the absence vs presence of added GST-CPSF6. Consolidated results from differential HDX-MS for protection in CA are shown in Supplementary Fig. 14a). The averaged data (+/− SD) from six independent experiments are shown. Statistical analysis of samples was carried out with Student’s two-sample, two-tailed t-test. Source data are provided as a Source Data file (b) Our X-ray structure of the CA_hex + IP6 + CPSF6_313-327 with protected CA regions colored according to the scale bar in Supplementary Fig. 14a and the regions showing no significant protection colored gray. For clarity, only one hydrophobic pocket formed by two adjoining CA monomers (colored gray) is shown. CPSF6 FG peptide is in magenta. The 10 s HDX results for CPSF6 peptides aa 316-358 containing the FG peptide + C-terminal LCR (c) and aa 266-315 containing N-terminal LCR (d) that are specifically protected in WT GST-CPSF6_261-358 but not in GST-CPSF6_261-358(ΔFG) upon incubation with IP6-stabilized CA(A92E) tubes. Gray and cyan bars show HDX values for the indicated CPSF6 peptides in the presence vs absence of IP6-stabilized CA(A92E) tubes, respectively. Consolidated results from differential HDX-MS comparing WT GST-CPSF6_261-358 vs GST-CPSF6_261-358(ΔFG) in the presence of IP6-stabilized CA(A92E) tubes are shown in Supplementary Fig. 14b. The averaged data (+/− SD) from six independent experiments are shown. Statistical analysis of samples was carried out with Student’s two-sample, two-tailed t-test Source data are provided as a Source Data file.

Fig. 7. HIV-1 infectivity assays probing a role of Pro residues in the context of CPSF6(LCR-FG-LCR).

a Schematic of WT and mutant CPSF6_1-358 constructs. Amino acid sequences of the CPSF6(LCR-FG-LCR) segment targeted by mutagenesis are shown. CPSF6₃₅₈/WT: native Pro residues are in bold and underlined. The FG containing peptide is highlighted in green. CPSF6₃₅₈(ΔFG): FG residues were deleted. CPSF6₃₅₈/mP-FGp: Pro residues in the FG-containing peptide were substituted with Gly or Ser. CPSF6₃₅₈/mP-LCR: Pro residues in the LCR segments were substituted with non-charged residues. CPSF6₃₅₈/mP-ED: Pro residues in the LCR segments were substituted with charged residues Glu or Asp. CPSF6₃₅₈/NE: the LCR segments of CPSF6₃₅₈ were replaced with non-LCR counterparts from NEURM_97-179. CPSF6₃₅₈/FU: the LCR segments of CPSF6₃₅₈ were replaced with alternative LCR from FUS_35-117. CPSF6₃₅₈(ΔFG)/FU: FG residues were deleted from CPSF6₃₅₈/FU. b Relative infectivity of VSV-G pseudotyped HIV-1 in the Hela cells stably overexpressing proteins shown in a. Infectivity was normalized to CPSF6₃₅₈(ΔFG). The averaged data (+/− SD) from three independent experiments are shown. Source data are provided as a Source Data file.

Fig. 8. HIV-1 infectivity assays probing roles of N- and C-termini CPSF6 LCRs.

a Schematic of WT CPSF6₃₅₈ and corresponding mutant proteins stably overexpressed in Hela cells. CPSF6₃₅₈/m(N + C): both N- and C-termini CPSF6 LCRs were replaced with non-LCR sequences from NEURM_97-179. CPSF6₃₅₈/mN: the N-terminal LCR was replaced with non-LCR sequences from NEURM_97-179. CPSF6₃₅₈/mC: the C-terminal LCR was replaced with non-LCR sequences from NEURM_97-179. b Relative infectivity of VSV-G pseudotyped HIV-1 in the Hela cells stably overexpressing proteins shown in a. Infectivity was normalized to CPSF6₃₅₈/m(N + C). The averaged data (+/− SD) from three independent experiments are shown. Source data are provided as a Source Data file. LCR low complexity region.

Fig. 9. Interplay between LEN, CPSF6 and HIV-1.

a Antiviral activities after LEN addition to HEK293T cells at indicated time points post-infection. The normalized and averaged data (+/− SD) from three independent experiments are shown. Source data are provided as a Source Data file. b Representative immunoblotting to show the expression levels of CPSF6 in parental, CPSF6 KO (CKO) and CPSF6 overexpressing HEK293T cells. The experiment was repeated 3 times independently with similar results. c Antiviral activities of LEN in indicated cells with varying levels of CPSF6. VSV-G pseudotyped HIV-scarlet viruses (MOI = 0.5) were used to infect indicated cells. The infection levels in different cell lines without LEN treatment were 43.9 ± 2% (WT), 38.5 ± 0.9% (CKO) and 38.4 ± 1.6% (WT + CPSF6). The normalized and averaged data (+/− SD) from three independent experiments are shown in c. Source data are provided as a Source Data file. d TZM-bl cells were infected with INmNG-labeled fluorescent HIV-1 pseudoviruses for 8 h to allow virus nuclear import. At 8 hpi, cells were treated with DMSO or indicated concentrations of LEN for 30 min, fixed with PFA and immuno-stained for endogenous CPSF6. Single Z-slice images show nuclear IN-puncta (green) and endogenous CPSF6 (red). Yellow dashed circles and white arrows point to nuclear IN-puncta colocalized or not-colocalized with CPSF6, respectively. Insignificant nuclear import occurs within 30 min of LEN treatment. Scale bar is 5 μm. e Analysis of background subtracted CPSF6 fluorescence associated with nuclear IN puncta in d. All data points (n = 553 for DMSO; n = 740 for 0.5 nM LEN; n = 389 for 5 nM LEN; n = 746 for 50 nM LEN; n = 309 for 500 nM LEN; n = 412 for 5000 nM LEN) obtained from a total of three independent experiments are overlayed. Statistical significance of comparison of indicated concentration of LEN versus DMSO control was determined by non-parametric Mann-Whitney two-tailed rank sum test. P-values for 0.5, 5, 50, 500 and 5000 nM LEN were 0.21, 0.66, 1.3E-24, 1.7E-22, 2.5E-28, respectively. ***p < 0.0001. ns: not significant. Source data are provided as a Source Data file. f Isolated HIV-1 cores were incubated with 2 μM GST-CPSF6_261-358 for 20 min. DMSO control or indicated concentrations of LEN were added to the mixture and incubated for 20 min. The HIV-1 cores pulled-down by GST-CPSF6_261-358 through glutathione sepharose beads were detected by ELISA. The averaged data (+/− SD) from three independent experiments are shown. Source data are provided as a Source Data file. LEN lenacapavir, a.u. arbitrary units.