The central pore of HIV-1 capsomers promotes sustained stability of the viral capsid

Abstract

The HIV-1 capsid, which orchestrates several key post-entry events to facilitate infection, is composed of hexamers and pentamers (capsomers) of the capsid (CA) protein. The capsomers are arranged in a closed, conical structure – the capsid – that protects the viral RNA genome and replicative enzymes reverse transcriptase (RT) and integrase (IN). Each capsomer contains a central pore lined with rings of positively charged amino acid side chains – Arg-18 (R18) and Lys-25 (K25). The R18 and K25 rings drive capsid assembly by binding the host polyanion inositol hexakisphosphate (IP6) and are proposed to mediate the import of dNTPs into the capsid to facilitate reverse transcription. Here we demonstrate that the R18 ring can be functionally replaced by the N21K substitution that establishes a new electropositive ring within the central pore. In contrast with previous studies in which R18 mutants were unable to adapt in culture, the N21K substitution facilitated the acquisition of second-site compensatory mutations that restored near-WT fitness to viral mutants lacking the R18 ring. Comparative analysis of several central pore mutants lacking the R18 ring revealed that particle infectivity was not correlated with IP6 binding or capsid assembly but rather with capsid stability and key post-entry events including reverse transcription and nuclear entry. Our results indicate that the central pore plays critical roles in both the assembly of capsids and their sustained stability post-entry.

Graphical Abstract

Introduction

HIV-1 particle assembly occurs at the inner leaflet of the plasma membrane of infected cells. This process is driven by the Gag polyprotein precursor, which assembles with the GagPol polyprotein precursor at a Gag:GagPol ratio of ~20:1 into a spherical lattice, called the immature Gag lattice. The oligomerization of Gag is primarily driven by its capsid (CA) domain, and upon completion of assembly, nascent virions bud from the cell as immature virus-like particles (VLPs). Concomitant with particle release, these VLPs undergo particle maturation in which the viral protease (PR), incorporated into virions as part of GagPol, cleaves the Gag and GagPol precursors into the mature forms of their constituent domains. The final step of productive maturation occurs when the mature CA protein assembles into a closed conical structure – the capsid – which serves as the protective shell of the viral core, enclosing the viral RNA genome as well as enzymes reverse transcriptase (RT) and integrase (IN). After viral entry, the capsid remains stable and intact^1,2, acting as a container for reverse transcription³ and interacting with host factors to mediate nuclear entry and viral DNA integration⁴.

The HIV-1 capsid is composed of a predominantly hexameric lattice with approximately 250 hexamers and exactly 12 pentamers of the mature CA protein. The quasi-equivalent assembly of CA into both hexamers and pentamers, collectively called capsomers, is crucial for the formation of a closed capsid during maturation. Specifically, the inclusion of pentameric “defects” at either end of the capsid lattice induces the curvature required to close the lattice, resulting in the capsid’s characteristic conical structure⁵. In the presence of high salt, recombinant CA protein assembles into helical tubes composed entirely of hexamers^6–9. However, in the presence of the host polyanion inositol hexakisphosphate (IP6), CA assembles into closed capsid-like particles (CLPs) that contain pentamers and are similar in size and shape to authentic HIV-1 capsids¹⁰. Indeed, IP6 is a critical co-factor for HIV-1 replication owing to its role in promoting the assembly of a stable capsid^10,11. IP6 also promotes virus assembly by binding and stabilizing the immature Gag lattice in a process that enriches virions with levels of IP6 necessary to build the capsid during maturation^10,12,13.

IP6 promotes the assembly of the HIV-1 capsid by binding to two amino acid side chains in the CA N-terminal domain – Arg-18 (R18) and Lys-25 (K25) – that form positively charged rings within a “central pore” in capsomers (Fig. 1A). Both the R18 ring and K25 ring can bind an individual IP6 molecule, and by neutralizing these positive charges lining the central pore, IP6 stabilizes capsomers. Mutations of either of these IP6-binding residues prevents proper capsid assembly and severely reduces particle infectivity^14–20. Consistent with the ability of IP6 to drive the in vitro assembly of CLPs, IP6 has been proposed to play a particularly important role in the formation of pentamers^{14–16,21,22}. Mutations at K25 prevent IP6-mediated CLP assembly in vitro, resulting instead in the exclusive assembly of helical CA tubes^14–16. Furthermore, our previous work demonstrated that K25A CLP assembly in vitro and closed capsid assembly in virions is restored by second-site compensatory mutations that likely restore pentamer formation¹⁴. In addition to binding IP6, the central pore also binds dNTPs in an interaction that is driven primarily by R18, suggesting that dNTPs could be transported through the central pore to facilitate reverse transcription inside closed capsids¹⁷.

Figure 1. The N21K substitution facilitates the acquisition of compensatory mutations by R18A.

(A) Cross section of the central pore of the CA hexamer (based on PDB 6R6Q) with IP6 molecules bound to the R18 and K25 rings. The location of N21 on the inner surface of the central pore is also shown. (B) Single-cycle infectivity of WT HIV-1 and the indicated mutants quantified by luminescence in infected TZM-bl cells. Virus stocks were normalized by RT assay prior to infection, and infectivity was quantified relative to WT virus produced in parallel. Error bars depict the mean ± s.e.m. from at least 7 independent experiments for each mutant. Infectivity measurements that did not exceed the background of the assay were assigned a value of zero. (C-E) Representative forced evolution experiment depicting spreading infection of WT and R18A/N21K after initial transfection of MT4 cells (C) and a second (D) and third (E) passage of virus collected from the preceding passage. (F, G) Representative forced evolution experiment depicting spreading infection of WT and R18S/N21K after initial transfection of MT4 cells (F) and a subsequent passage (G) of virus collected at the peak of replication. Arrows indicate the virus collected from a given passage that was used to initiate the subsequent passage. The Env(−) clone pNL4–3/KFS is included as a negative control. (H) Single-cycle infectivity of WT HIV-1 and the indicated mutants quantified by luminescence in infected TZM-bl cells. Virus stocks were normalized by RT assay prior to infection. Some data points for the R18A/N21K, S16L/R18S/N21K (LSK), R18S/N21K/A31T (SKT), R18S/N21K/T216I (SKI) mutants are the same as presented in Figure S3. Numerical infectivity values for each mutant are expressed as a percentage of WT infectivity above each corresponding bar. Error bars depict the mean ± s.e.m. from at least 3 independent experiments. Statistical significance was determined using Welch’s t-test (p-value summary: >0.05 = not significant; <0.05 = *; <0.01 = **; <0.001 = ***; <0.0001 = ****). Abbreviations for the mutants used throughout the study are listed here: R18S/N21K (SK); S16L/R18S/N21K (LSK); R18S/N21K/A31T (SKT); R18S/N21K/T216I (SKI); S16N/R18S/N21K (NSK); S16L/R18S/N21R (LSR).

Optimal stability of the HIV-1 capsid after viral entry is critical for several post-entry functions required for productive infection. Efficient reverse transcription of the viral genome requires a closed, stable capsid²³ to protect the viral nucleic acid from detection by host innate immune sensors^24–27, and to keep the RT enzyme associated with its nucleic acid substrate. Additionally, stable capsids are required for the core to efficiently dock at the nuclear envelope, engage the nuclear pore complex (NPC), and pass through the nuclear pore^28,29. After nuclear entry and the completion of reverse transcription, the core undergoes uncoating, a process characterized by the disassembly of the capsid leading to the release of viral DNA for integration into host chromatin^1,2,30,31. The timing of uncoating is crucial for productive infection, and premature uncoating induced by mutations that destabilize the capsid prevents efficient reverse transcription, nuclear entry, and integration^32,33. IP6 stabilizes pre-formed capsids in vitro²³, however its role in stabilizing capsids in cells is less certain. Several studies have shown that depletion of IP6 in target cells does not disrupt HIV-1 infection^13,34,35. In contrast, the infectivity of select CA mutants with hypostable capsids is acutely reduced by IP6 depletion, suggesting that IP6 stabilizes capsids post-entry but that this stabilizing effect is not required for WT HIV-1 to complete a productive infection. IP6 has also been proposed to play a role in dNTP import through the central pore of capsomers¹⁶. Thus, while the central pore and its interaction with IP6 are clearly important for capsid assembly, the significance of this interaction in infected cells post-entry and its implications for capsid stability, dNTP import, and reverse transcription are not well understood.

Here we examine the role of the central pore of HIV-1 capsomers by determining the consequences of replacing the R18 ring with a new positively charged ring in the central pore. We report that replacing the R18 ring facilitates the acquisition of second-site compensatory mutations that substantially restore viral replication. Further manipulation of the central pore and comparative analysis of several mutants lacking the R18 ring reveal that, under this condition, the ability to bind IP6 and assemble capsids in vitro and in virions is not correlated with infectivity. However, we find that non-infectious central pore mutants lacking the R18 ring are defective for reverse transcription and nuclear entry, and that the mechanism underlying these defects is a lack of sustained capsid stability. Our findings indicate that, in addition to its well-established role in capsid assembly, the central pore functions to maintain the stability of capsids post-entry.

Results

The R18 ring in the central pore of capsid hexamers can be replaced by a new positively charged ring at position 21

We recently reported that propagation of an HIV-1 mutant lacking the K25 ring (K25A) resulted in the acquisition of second-site compensatory mutations that restore the ability of HIV-1 to assemble closed capsids¹⁴. However, our attempts to propagate several R18 mutants did not result in the acquisition of compensatory mutations¹⁴. Importantly in vitro, assembly assays indicated that R18 and K25 play separate roles in capsid assembly with R18 driving CA hexamer formation and K25 driving CA pentamer formation. To gain further insights into the function of the central pore, we investigated how HIV-1 could adapt to a loss of the R18 ring. In an additional in vitro selection, K25A acquired the N21K substitution (Figure S1). Interestingly, N21 is located in helix 1 of CA ~1 turn of the helix below R18 and above K25, and its side chain, like those of R18 and K25, is oriented toward the center of the central pore in the context of assembled capsomers (Figure 1A). We reasoned that the introduction of a lysine at position 21 could establish an additional positively charged ring capable of interacting with IP6, and may facilitate enhanced adaptability of HIV-1 mutants lacking an electropositive ring at amino acid 18 in CA.

To test the hypothesis that the R18 ring can be functionally replaced, we transfected 293T cells with infectious molecular clones harboring R18A, N21K, or R18A/N21K mutations to produce virions and measured their specific infectivity in TZM-bl cells. R18A displayed no measurable levels of infection in our assay, whereas N21K demonstrated a ~10-fold decrease in infectivity relative to wild-type (WT). Like R18A, R18A/N21K infectivity was often not measurable, however we were able to detect low levels of infection in some experimental replicates, suggesting that the infectivity of the R18A/N21K mutant is near the detection limit of our assay. Ultimately, R18A/N21K demonstrated a ~4-log decrease in infectivity relative to WT HIV-1 (Figure 1B). While these results indicate that the R18A/N21K mutant is severely defective, we postulated that this low-level infectivity may facilitate the acquisition of compensatory mutations upon propagation in T-cell lines. To test this, we transfected MT4 cells with pNL4–3-R18A/N21K and monitored replication kinetics by quantifying the RT activity in the supernatants. We observed a substantial (24–42 day) delay in replication of R18A/N21K relative to WT, suggestive of the acquisition of compensatory mutations. We subsequently re-passaged virus collected from the peak of replication until replication kinetics resembled those of WT virus (Figure 1C–E). We then extracted genomic DNA from infected cells collected at the peak of replication, PCR-amplified the Gag-coding region of the proviral DNA and subjected these amplicons to sequencing analysis to identify compensatory mutations. While we initiated more than a dozen independent R18A/N21K selection experiments, three resulted in the emergence of compensatory mutations. (Table S1).

Three mutations were observed in amino acid residues within or adjacent to the central pore. Notably, we observed a substitution of the alanine we introduced at position 18 to serine (R18A → A18S). We also observed S16L and H12Y, which are present within the central pore and N-terminal β-hairpin, respectively (Figure S2A). We also observed the A31T substitution which is present on the luminal surface of the CA hexamers and pentamers that comprise the capsid (Figure S2B). The acquisition of the A18S substitution indicates that a serine is better tolerated than an alanine at position 18 in the presence of the N21K mutation. Therefore, in subsequent experiments, we propagated R18S/N21K in MT4 cells (Figure 1F, G) and observed a rapid acquisition of A31T or T216I, which we previously reported as a compensatory mutation that restores CA pentamer formation and closed capsid assembly to K25A¹⁴. These findings indicate that HIV-1 can acquire mutations both proximal and distal to the central pore that are capable of restoring replication to a mutant lacking the R18 ring. During propagation the virus also acquired I15M, L56V, and E187K mutations (Table S1), but we did not investigate the effects of these mutations.

To test the ability of the selected mutations to compensate for the loss of the R18 ring in the presence of N21K, we introduced each mutation into infectious molecular clones harboring R18A/N21K or R18S/N21K substitutions and compared their specific infectivity in TZM-bl cells. We found that the addition of S16L, T216I, or the combination of H12Y and A31T (which emerged together in an individual experiment, Table S1) resulted in a substantial increase in R18A/N21K particle infectivity, though these mutants remained substantially (25–100-fold) less infectious than WT (Figure S3A).

Remarkably, the change from an alanine (R18A/N21K) to a serine (R18S/N21K) resulted in a ~100-fold increase in particle infectivity (Figure 1H; compare light blue and magenta bars). Furthermore, S16L (S16L/R18S/N21K), A31T (R18S/N21K/A31T), or T216I (R18S/N21K/T216I) restored R18S/N21K infectivity ~30-fold to ~30% that of WT (Figure 1H). While these mutations do not confer a complete rescue, they confer a ~3000-fold increase in infectivity over the R18A/N21K mutant that was used to initiate the selection experiments. The H12Y substitution did not increase R18S/N21K infectivity and did not enhance the ability of S16L or A31T to restore infectivity to R18S/N21K, suggesting that H12Y exerts only a subtle effect on virus replication. Moreover, the simultaneous addition of S16L and A31T to R18S/N21K antagonized the restorative effect conferred by each mutation individually (Figure S3B).

Finally, we engineered two additional mutants for comparison with S16L/R18S/N21K (hereafter called LSK). S16N/R18S/N21K (hereafter called NSK) was engineered because S16N has been previously reported to restore replication and infectivity to the replication-defective central pore mutant I15V³⁶. S16L/R18S/N21R (LSR) was engineered to test the effect of a different positively charged residue at position 21 in the absence of the R18 ring. Interestingly, despite only differing from the rescued LSK mutant by one amino acid, each of these mutants displays no measurable infectivity (Figure 1H). The summary infectivity figure of all mutants mentioned in this study with corresponding replicate numbers is displayed in Figure S4. Finally, we tested the replicative capacity of each of these mutant viruses in MT4 cells. The results were consistent with our infectivity measurements. Briefly, over 25 days in culture, WT, N21K, LSK, R18S/N21K/A31T (SKT), and R18S/N21K/T216I (SKI) all replicated with WT or near-WT kinetics, peaking at day 4–8 post-transfection. Consistent with selection experiments, R18S/N21K (SK) replicated with a ~10 day delay relative to WT, whereas R18A, R18S, R18A/N21K, NSK, and LSR showed no signs of replication (Figure S5). We also measured viral replication in peripheral blood mononuclear cells (PBMCs) isolated from four healthy donors (Figure S6). Whereas WT displayed rapid replication, peaking at day 4 post-infection, LSK, SKT, and SKI showed delayed replication kinetics, peaking between day 8 and 10. We did not observe any signs of SK replication in PBMCs. Overall, these infectivity and replication data demonstrate that while HIV-1 can adapt to the loss of the R18 ring, it remains extremely sensitive to changes in the central pore of capsomers.

Inositol phosphates bind and stabilize capsid hexamers lacking the R18 ring

Having confirmed that the N21K substitution restores infectivity to HIV-1 mutants lacking a positively charged ring at position 18 in CA, we next sought to understand how these changes to the central pore affect the ability of ligands IP6 and dNTPs to bind and stabilize CA hexamers. Our finding that N21K restores infectivity and replication to HIV-1 mutants lacking an arginine at position 18 in CA suggests that this change functionally replaces the R18 ring with a new positively charged ring of amino acids at position 21 (K21 ring). We therefore introduced the R18S and N21K mutations, along with either the S16L (LSK) or S16N (NSK) mutations, onto a WT CA hexamer structural model (PDB 6BHT), in which two IP6 molecules are resolved within the central pore – one bound above the R18 ring and the other adjacent to the K25 ring. (Figure 2A). The R18S mutation results in the loss of the electropositive ring that is responsible for binding the upper IP6 molecule. However, a new electropositive ring is established by the N21K mutation. This new K21 ring is positioned much closer to the lower IP6 molecule that is coordinated by the existing K25 ring. Because the upper IP6 molecule sits atop the R18 ring in the WT structure, we propose that the K21 ring is unlikely to replace this interaction. Rather, we hypothesize that the K21 ring participates in electrostatic interactions with the lower IP6 molecule, resulting in stronger coordination via combined effects of the K21 and K25 rings. Additionally, the introduction of the R18S mutation substantially opens the center of the pore, while the S16L and S16N mutations narrow the top of the pore. To quantify the extent to which S16 mutations narrow the top of the pore we measured the distance between Ser16 (WT), Leu16 (LSK), or Asn16 (NSK) and the nearest atom of the upper IP6 molecule on each side of the pore. These measurements suggest that the pore is narrowed by ~3.5 Å by the LSK and NSK mutations. These changes, along with the shifting of charge distribution toward the lower end of the pore, may have consequences for IP6 binding and occupancy and the potential movement of dNTPs.

Figure 2. IP6 binds, stabilizes and promotes the assembly of LSK and NSK CA.

(A) A previously solved crystal structure of the WT crosslinked CA hexamer bound to IP6 (PBD 6BHT; left) was used to project the effect of the mutations present in the LSK (center) and NSK (right) variants. (B) Thermal stability measurements for the indicated crosslinked CA hexamers alone or in the presence of the indicated inositol phosphates, as measured by differential scanning fluorimetry. Error bars depict the mean ± s.e.m. from at least 3 independent experiments. (C) In vitro assembly kinetics of 100 μM of the indicated recombinant mature CA protein in the presence of 5 mM IP6, as determined by measuring absorbance at 350 nm over time. (D) Negative-stain EM images of the assembly reactions performed in (C). Scale bars 100 nm. Assembly kinetics and negative-stain EM images are representative of in vitro assembly experiments performed on at least three separate occasions. Abbreviations: S16L/R18S/N21K (LSK); S16N/R18S/N21K (NSK).

To more directly compare the ability of LSK and NSK to bind inositol phosphates, we expressed and purified crosslinked CA hexamers for three variants demonstrating vast differences in particle infectivity – WT (100% infectivity), LSK (30% infectivity), and NSK (non-infectious). We used nano differential scanning fluorimetry to quantify the extent to which IP6 and other inositol phosphates enhance the thermal stability of crosslinked CA hexamers. In the absence of inositol phosphates, we observed no difference in the stability of WT and LSK CA hexamers, whereas NSK CA hexamers were substantially more stable than either WT or LSK hexamers (Figure 2B, compare white bars). Thermal stability of CA hexamers was increased by the addition of the inositol phosphates IP2 (1,4), IP3 (1,4,5), IP4 (1,3,4,5), IP5 (1,3,4,5,6), and IP6 for each variant. As expected, CA hexamers were most potently stabilized by IP6. Peak thermal stability in the presence of IP6 was highest for NSK CA hexamers, though the shift in thermal stability induced by IP6 was greater for WT and LSK CA hexamers (Figure 2B, compare white and dark blue bars). This observation is likely due to the higher inherent stability of NSK CA hexamers, resulting in a lower thermal stability shift in the presence of IP6. These data are consistent with our structural projections, which did not reveal substantial differences in the ability of LSK and NSK to interact with IP6 in the context of CA hexamers. Together, these observations suggest that the ability to bind IP6 is not the basis for the observed differences in LSK and NSK infectivity.

NSK assembles capsids in vitro and in virions

Crosslinked hexamers, while useful for comparing inositol phosphate binding, do not allow the impact of mutations on capsid assembly to be assessed. We therefore used an in vitro turbidity assay, in combination with negative-stain electron microscopy (EM), to determine the ability of IP6 to drive the assembly of LSK and NSK CA protein into CLPs that broadly recapitulate the size and shape of authentic HIV-1 capsids. We mixed mature WT, LSK, or NSK CA protein with 5 mM IP6 and measured the kinetics of assembly by quantifying the increase in optical density at 350 nm (OD₃₅₀) (Figure 2C). While LSK and NSK CA protein assembled with slower kinetics than WT (with NSK assembling slightly faster than LSK), negative-stain EM revealed an abundance of CLP structures present across all samples (Figure 2D). While some LSK and NSK CLP structures appear slightly larger than WT CLPs we did not observe a substantial difference between LSK and NSK CLPs that could explain the large difference in the infectivity of these two mutants. Furthermore, this may be an in vitro-specific effect caused by changes in the kinetics of CLP assembly in bulk solution. Similar results were obtained when assembly reactions were performed in the presence of 1.5 and 3 mM IP6 (Figure S7). This finding suggests that although LSK and NSK CA assembles less efficiently than WT in the presence of IP6, each can assemble in vitro into the closed conical capsid structure required for infection. Furthermore, the assembly of LSK and NSK into closed CLPs suggests that pentamer formation is not prevented by these mutations.

Next, we sought to determine the effect of the LSK and NSK mutations on the ability of HIV-1 to properly assemble capsids in virions. We produced replication-incompetent WT, LSK, and NSK virions in 293T cells, and assessed capsid morphology using cryo-electron tomography (cryo-ET). For each virus, we observed a range of capsid morphologies and classified each imaged virion into one of four morphological categories: immature, mature (conical), mature (tubular), and mature (irregular) (Figure 3A and B). Roughly 75% of WT virions displayed the conical or tubular capsid morphologies typical of infectious particles, whereas only ~50% of LSK and NSK virions contained conical or tubular capsids. Notably, the number of LSK and NSK virions containing irregular capsid morphologies was twice that of WT virions. These findings are consistent with our in vitro assembly results showing that LSK and NSK CA can assemble into CLPs but do so less efficiently than WT. While we cannot rule out a slight maturation defect relative to WT, the similarity in the distribution of capsid morphologies found in NSK and LSK virions suggests that the drastic difference in their infectivity cannot be explained by differences in their ability to assemble mature capsids.

Figure 3. LSK and NSK mutants assemble capsids morphologically similar to those of WT in virions.

VSV-G pseudotyped WT, LSK, and NSK virions were purified and subjected to cryo-ET analysis. Tilt series were collected and reconstructions performed to assess capsid morphology. At least 100 virions for each were analyzed and classified as displaying immature morphology or conical, tubular, or irregular mature capsid morphology. Examples of sliced tomograms for each category are shown for each virus population analyzed (A) and the overall distribution for each virus population is shown in (B). Scale bars, 100 nm. Abbreviations: S16L/R18S/N21K (LSK); S16N/R18S/N21K (NSK).

NSK is defective for reverse transcription in cells and in vitro

The central pore of capsomers and its interaction with IP6 are crucial for the assembly of the HIV-1 capsid. However, our comparison of the LSK and NSK central pore mutants, in which we deliberately manipulated the IP6 binding site, reveals that the drastic differences in the particle infectivity of these mutants cannot be explained by differences in the ability of IP6 to bind and stabilize CA hexamers or facilitate capsid assembly. To understand the mechanism underlying the differences in particle infectivity between LSK and NSK, we quantified the accumulation of RT products in SupT1 cells infected with replication-incompetent WT, LSK, and NSK virions pseudotyped with VSV-G. We also included the R18S/N21K (SK) mutant and the rescued R18S/N21K/A31T (SKT) and R18S/N21K/T216I (SKI) mutants in this analysis for comparison. For the WT, early (minus-strand strong stop) and late (second-strand transfer) RT products accumulated rapidly during the first 6 hours post-infection (hpi) and continued to increase slowly between 6 and 12 hpi (Figure 4A and B). The rescued mutants LSK, SKT, and SKI showed reverse transcription kinetics similar to those of the WT, though the total copy numbers were ~1.5–3-fold lower than WT at each time point. In contrast, SK and NSK displayed only minor accumulation of early and late RT products over the course of infection. The levels of SK and NSK RT products decreased progressively relative to WT at each time point collected. At 12 hpi, SK early and late RT products were reduced 20- and 50-fold, respectively, relative to WT, and NSK early and late RT products were reduced 50- and 200-fold, respectively. We also quantified the accumulation of 2-LTR circles, which form only in the nucleus and thus provide a measure of nuclear import of viral DNA. Similar to the results observed for RT products, LSK, SKT, and SKI displayed only modestly reduced levels of 2-LTR circles, whereas SK and NSK displayed drastic reductions (50–300-fold) (Figure 4C). Importantly, the reductions in RT products and 2-LTR circles for each of these mutants relative to WT correlate well with the infectivity defects displayed by each of these mutants (summarized in Table S2): WT > LSK, SKT, & SKI > SK > NSK. This correlation suggests that the primary biological mechanism underlying the infectivity defects induced by replacing the R18 ring with the K21 ring is a decreased efficiency of reverse transcription and nuclear entry.

Figure 4. NSK virions are defective for reverse transcription in cells and in vitro.

(A-C) Early RT products (A), late RT products (B), and 2-LTR circles (C) were quantified by qPCR for WT, SK, LSK, SKT, SKI, and NSK mutants at 1, 2, 3, 6, 9, and 12 hours post-infection (hpi) of SupT1 cells. Each individual sample was quantified relative to CCR5 copy number to control for the cell number, and all data for each viral DNA species were normalized to copy number/CCR5 for WT at 12 hpi. Error bars depict the mean ± s.e.m. from 3 independent experiments. (D-F) minus-strand strong stop (D, MSSS), first-strand transfer (E, FST), and second-strand transfer (F, SST) DNA copies were quantified by qPCR in endogenous reverse transcription (ERT) reactions performed for 6 hours at 37°C with Streptolysin O (SLO), IP6, and dNTPs in the presence or absence of DNase. Control reactions lacking IP6 and dNTPs were subtracted to account for background levels of RT products, and experimental reactions that were quantified below background were assigned a value of zero. Error bars depict the mean ± s.e.m. from 3 independent experiments. (G) Comparison of mean LSK and NSK early (G) and late (H) RT products at 6 hpi in SupT1 cells (from A and C) relative to WT. Error bars for D-G depict the mean ± s.e.m. from 3 independent experiments. Statistical significance for D-G was determined by Two-way ANOVA with Tukey’s multiple comparisons test (p-value summary: >0.05 = not significant; <0.05 = *; <0.01 = **; <0.001 = ***; <0.0001 = ****). Abbreviations: R18S/N21K (SK); S16L/R18S/N21K (LSK); R18S/N21K/A31T (SKT); R18S/N21K/T216I (SKI); S16N/R18S/N21K (NSK).

There are multiple possible mechanisms that could explain how the changes that we made in the central pore could affect the efficiency of reverse transcription. First, we have previously shown that R18 plays a key role in binding to dNTPs, which have been proposed to translocate through the central pore to enable reverse transcription inside closed capsids^14,17. Therefore, the replacement of the R18 ring with a K21 ring may prevent efficient dNTP binding and import into the lumen of the viral capsid, depriving RT of the substrates required for viral DNA synthesis. We therefore measured the ability of ATP and dATP to stabilize WT, LSK, and NSK crosslinked CA hexamers. Whereas WT CA hexamers display increased thermal stability upon the addition of ATP and dATP, LSK and NSK CA hexamers were only marginally stabilized (Figure S8A). Consistent with this finding, fluorescence polarization (FP) assays revealed that the affinity of ATP for LSK (K_d= 161.5 μM) and NSK (K_d= 112.9 μM) CA hexamers is ~10-fold lower than its affinity for WT (K_d= 12.31 μM) CA hexamers (Figure S8B). We previously reported that the R18G mutation results in a ~1000-fold decrease in the affinity of ATP for CA hexamers¹⁴, suggesting that the introduction of the N21K mutation significantly restores nucleotide binding in the absence of the R18 ring, though not to WT levels. These data suggest that less efficient binding to dNTPs may contribute to the reverse transcription defects that we observed. However, the finding that the reduction in ATP binding is similar for LSK and NSK indicates that nucleotide binding to the central pore does not fully explain the vast differences in infectivity and reverse transcription by the LSK and NSK mutants.

A second mechanism that could explain how changes to the central pore may alter the efficiency of reverse transcription is perturbed capsid stability post-entry. While initial assembly of the capsid in virions is required for efficient reverse transcription and infection, capsids must also remain closed and stable after viral fusion, acting as a container for reverse transcription and a barrier to viral nucleic acid sensing by the innate immune system. We therefore tested the ability of WT, LSK, and NSK virions to reverse transcribe their genomes in an endogenous reverse transcription (ERT) assay. Briefly, virions were incubated with dNTPs, IP6, and streptolysin O (SLO), a pore-forming toxin that disrupts the integrity of the viral membrane, for 6 hours at 37°C. The pore-forming activity of SLO allows IP6 and dNTPs access to the interior of the virion to facilitate reverse transcription inside closed IP6-stabilized capsids. Consistent with previous studies^23,37, we observed reduced accumulation of intermediate (first-strand transfer, FST) and late (SST) RT products relative to early (minus-strand strong stop, MSSS) RT products. In agreement with our quantification of RT products in infected cells, the production of early, intermediate, and late RT products was slightly reduced for LSK relative to WT. NSK produced even fewer RT products, although this difference was less severe than in cells (Figure 4D–F, white bars; G–H, compare white and grey bars). Identical ERT reactions performed in the presence of DNase reduced the quantity of all RT products for each virus. Because DNase is presumably too large to enter a closed capsid, we interpreted this reduction in RT products to be a consequence of DNase-mediated digestion of reverse transcripts that were not protected by a closed capsid. WT RT products were reduced ~2–4-fold in the presence of DNase, and LSK RT products were reduced ~3–5-fold. In contrast, NSK RT products were reduced 17–25-fold in presence of DNase, suggesting that a large percentage of NSK RT products were not protected within closed capsids (Figure 4D–F, blue bars). The differences in the accumulation of MSSS and FST transcripts were statistically significant in the presence and absence of DNase for each virus. However, due to variability in the data, our SST results did not reach statistical significance (see Figure S9 for raw experimental data and background). Importantly, the production of LSK and NSK RT products relative to WT observed in the presence of DNase exhibits a similar trend to the results we obtained in cells at 6 hpi (Figure 4G). Together, our reverse transcription measurements in cells and in vitro suggest that NSK capsids are inherently unstable, leading to a loss of integrity upon disruption of the viral membrane. In cells, we posit that this loss of NSK capsid integrity occurs shortly after viral fusion and leads to rapid disassembly of the capsid and severely reduced reverse transcription due to the diffusion of the RT enzyme from its RNA substrate. In our ERT assay, we postulate that in the absence of DNase, the loss of NSK capsid integrity occurs after permeabilization of the viral membrane by SLO. However, because the overall architecture of the viral membrane remains intact, large-scale capsid disassembly and the loss of the RT enzyme is slowed, allowing reverse transcription to continue even after loss of capsid integrity. Finally, the milder sensitivity to DNase displayed by the LSK mutant relative to NSK suggests that LSK capsids are considerably more stable than NSK capsids, but less stable than WT capsids.

Mutants lacking the R18 ring demonstrate variable capsid stability, nuclear envelope docking, and nuclear entry

To test the inherent stability of the central pore mutant capsids after viral membrane permeabilization, we utilized a previously described¹⁴ TIRF microscopy assay (Figure 5A and B). Briefly, WT and several central pore mutant (R18S, N21K, SK, LSK, and NSK) virions were produced using a GagPol construct with EGFP inserted between MA and CA domains and flanked by viral protease cleavage sites. During viral maturation, EGFP is passively incorporated into capsids and serves as a fluid-phase capsid content marker that is released upon capsid rupture. EGFP-labeled virions were immobilized, permeabilized with SLO in the presence or absence of IP6 for 30 minutes and imaged to quantify the number of virions containing intact capsids. Consistent with previous results¹⁴, in the absence of IP6, we observed very low EGFP+ particle counts for WT and all central pore mutants relative to virus-only controls, suggesting that, in the absence of added IP6, capsids rapidly rupture after membrane permeabilization (Figure 5C). In the presence of IP6, we observed markedly increased EGFP+ particle counts for WT, N21K, and LSK virions, indicating that these capsids remained intact in the presence of IP6. In contrast, we observed substantially smaller increases in intact SK and NSK capsids in the presence of IP6, while R18S virions displayed no increase in intact capsids in this condition. These results are broadly consistent with differences in infectivity and reverse transcription measurements for these mutants and suggest that the inherent stability of capsids is the primary mechanism underlying these observed differences.

Figure 5. HIV-1 central pore mutants display variable capsid stability.

(A) Experimental design: virus particles containing EGFP as a capsid content marker were bound to glass dishes and permeabilized with SLO in the presence or absence of 50 μM IP6. Loss of GFP content marker provides a measure of capsid stability. (B) Images were acquired with a TIRF microscope and particles counted using Fiji. Representative masks generated during analysis of immobilized WT virus 30 min post-SLO or control treatment ± 50 μM IP6 are shown. Scale bar = 10 μm (C) Intact capsids were counted for the indicated mutants in untreated, and SLO-treated ± 50 μM IP6 conditions. Error bars depict the mean ± s.e.m. from counts acquired from 15 total images spanning 3 independent experiments. Statistical significance was determined by Two-way ANOVA with Dunnet’s multiple comparisons test (p-value summary: >0.05 = not significant; <0.05 = *; <0.01 = **; <0.001 = ***; <0.0001 = ****). Abbreviations: R18S/N21K (SK); S16L/R18S/N21K (LSK); S16N/R18S/N21K (NSK).

To investigate further the effect of IP6 on capsid stability of CA mutants lacking the R18 ring, we measured the effect of IP6 depletion in target cells on particle infectivity. Several studies have reported that while WT HIV-1 infectivity is unaffected by the depletion of IP6 in target cells^13,34,35, infectivity of certain CA mutants that assemble inherently unstable capsids is severely reduced in IP6-depleted cells³⁵. We therefore compared the infectivity of WT and several central pore mutants (N21K, SK, LSK, SKT, SKI) in WT 293T cells (parental) and cells in which IPPK, the host enzyme that phosphorylates IP5 to synthesize IP6, was knocked out (IPPK KO). As a positive control we also measured the effect of IP6 depletion on the hypostable CA mutant virus P38A. Consistent with previous reports^13,34,35,38, WT HIV-1 infectivity was similar in parental and IPPK KO cells, whereas P38A infectivity was reduced ~30-fold in IPPK KO cells relative to parental cells (Figure S10). Relative to P38A, our central pore mutants showed much milder sensitivity to target cell IP6 depletion. SK and LSK infectivity was reduced ~3-fold, and N21K and SKT displayed a 2-fold reduction in infectivity in IPPK KO cells relative to parental cells. We also observed a statistically significant, but minor, increase in SKI infectivity in IPPK KO cells. We were unable to test NSK in these experiments because its infectivity was below the detection limit of our assay in both cell lines. The mild sensitivity to IP6 depletion displayed by LSK may be indicative of a slight reduction in capsid stability and is consistent with the mild DNase sensitivity observed in our ERT experiments. However, the SK mutant, which demonstrates significantly reduced capsid stability in our TIRF assay, displays the same mild sensitivity to target cell IP6 depletion. This suggests that the effects of target cell IP6 depletion on the infectivity of CA mutants do not necessarily correlate directly with capsid stability and should be interpreted with caution.

We next sought to determine the ability of our central pore mutant capsids to dock at the nuclear pore and enter the nucleus. We produced WT, LSK, and NSK virions from 293T cells expressing YFP-labeled APOBEC3F that is resistant to Vif-mediated degradation (A3F-E289K-YFP), and infected HeLa cells with these virions. A3F-YFP is packaged into virions and incorporated into HIV-1 capsids via its interaction with the viral ribonucleoprotein complex, providing a marker for the subcellular localization of capsids in infected cells²⁹. Infected cells were fixed 6 hpi, and discrete YFP-positive particles were quantified in the nucleus and at the nuclear envelope (NE) using confocal microscopy (Figure 6A). The percentage of NSK particles at the NE and in the nucleus was severely reduced relative to WT, and comparable to the level of WT particles in the presence of the capsid inhibitor PF74, which destabilizes the capsid and prevents NE docking and nuclear entry¹ (Figure 6B and C). LSK capsids docked at the NE with WT-efficiency, but displayed a modest (2-fold) decrease in nuclear entry relative to WT. These data are consistent with the reduction in 2-LTR circles that we observed for the NSK and LSK mutants. Based on the preceding results, and because optimal capsid stability is required for NE docking and nuclear entry^28,29, we propose that NSK capsids are inherently unstable, resulting in essentially no NE docking, nuclear entry, or infection. In contrast, LSK capsids are stable enough to dock at the nuclear pore complex (NPC), but do not pass through the nuclear pore as efficiently as WT capsids, leading to a modest decrease in infectivity.

Figure 6. NSK capsids are defective for nuclear envelope docking and nuclear entry.

VSV-pseudotyped Δenv A3F-E289K-YFP-labeled WT, LSK, and NSK virions were used to infect HeLa cells for 6 hours. Infected cells were then fixed, stained with BioTracker 650 Red Nuclear Dye, and imaged by confocal microscopy to determine the subcellular localization of HIV-1 capsids. An additional condition in which WT-infected cells were treated with 10 μM PF74 was used as a negative control. (A) Representative confocal images showing HIV-1 capsids labeled with A3F-E289K-YFP at the nuclear envelope (open triangle) and inside the nucleus (filled triangle). Scale bar, 10 μm. (B) YFP-positive particles at the nuclear envelope were quantified as a percentage of the total particles in the cytoplasm and expressed as a percentage of WT in untreated cells. (C) YFP-positive particles in the nucleus were quantified as a percentage of the total particles in the cell and expressed as a percentage of WT in untreated cells. Statistical significance was determined by paired t-test (p-value summary: >0.05 = not significant; <0.05 = *; <0.01 = **; <0.001 = ***; <0.0001 = ****). Abbreviations: S16L/R18S/N21K (LSK); S16N/R18S/N21K (NSK).

Discussion

Each capsomer in the HIV-1 capsid lattice contains a central pore that binds IP6 to facilitate capsid assembly¹⁰. The central pore also binds dNTPs, which are proposed to translocate through the pore to enable DNA synthesis inside the capsid^16,17. These binding events are coordinated by two positively charged rings (R18 and K25) that line the inner rim of the central pore. Our previous work produced two findings that are relevant to the current study. First, we found that HIV-1 is unable to adapt to the loss of its R18 ring, whereas mutation of the K25 ring can be readily compensated for by mutations in CA that do not re-establish a positively charged ring in the central pore. Second, we found that the K25 ring is particularly important for the assembly of closed capsids during maturation. This finding is consistent with several other studies showing that mutations within the central pore prevent proper capsid assembly^{18,32,39–41}.

In this study, we engineered a mutation (N21K) in combination with R18A designed to replace the R18 ring with a new positively charged ring in the central pore (K21 ring). In contrast to single R18 mutants described previously¹⁴, R18A/N21K acquired second-site compensatory mutations in CA that markedly increased its infectivity and replication upon propagation in MT4 cells (Figure 1). These mutations did not introduce any new charged rings in the central pore, indicating that the R18 ring can be functionally replaced by the K21 ring (Figure S2). Importantly, structural projections suggest that the K21 ring does not replace the R18:IP6 interaction but rather facilitates enhanced engagement between the K25 ring and IP6. Comparative biochemical analysis of a rescued mutant, S16L/R18S/N21K (LSK), and a completely non-infectious mutant, S16N/R18S/N21K (NSK), revealed no differences in IP6-mediated stabilization (Figure 2B), in vitro CLP assembly (Figure 2C–D), or core formation in virions (Figure 3), suggesting that capsid assembly can occur at near-WT efficiency even in the absence of the R18 ring and the molecule of IP6 with which it interacts. In contrast, we observed a severe defect in the ability of the NSK mutant to undergo reverse transcription in cells (Figure 4A–C). This finding, combined with NSK’s sensitivity to DNase in the endogenous reverse transcription assay (Figure 4D–F) and reduction in particles upon viral membrane permeabilization in the TIRF assay (Figure 5), suggests that NSK capsids are inherently unstable and break open shortly after fusion. This hypothesis is supported by confocal microscopy analyses demonstrating that NSK capsids are defective in nuclear envelope docking and nuclear entry. Overall, these findings establish a role for the central pore in promoting the sustained stability of HIV-1 capsids that appears to be independent of its ability to bind IP6.

The finding that HIV-1 can replicate in the absence of the R18 ring is noteworthy in light of early structural studies that described the R18 ring as the only binding site for IP6^10,11. While several subsequent studies have confirmed that the K25 ring can also coordinate an IP6 molecule^{15,19,21,42,43}, the R18 ring likely binds IP6 with greater affinity owing to the tighter packing of its side chains and its closer proximity to the outer surface of the capsid. Based on our structural projections of LSK and NSK hexamers we propose that the R18:IP6 interaction is lost upon replacement of the R18 ring with the K21 ring, which redistributes the positively charged side chains in the pore away from the upper IP6 molecule that sits atop the R18 ring in WT hexamers and toward the lower IP6 molecule that is positioned closer to the K25 ring. Thus, we postulate that the R18 ring and its interaction with IP6 are not absolutely required for HIV-1 replication. We previously came to a similar conclusion regarding the importance of the K25 ring based on the ability of second-site compensatory mutations to restore infectivity and replication to K25A. However, a key difference highlighted here in the relative importance of these two positively charged rings is that while the R18 ring can facilitate replication as the sole electropositive ring within the central pore, the K25 ring requires the introduction of an additional electropositive ring in the absence of R18. We speculate that this is due to lower affinity between IP6 and the K25 ring relative to the R18 ring, leading to a requirement for additional clustering of positively charged side chains (i.e. K21) to efficiently coordinate IP6.

While a systematic exploration of IP6 binding to the mature capsids of lentiviruses has not been completed, all lentiviruses⁴⁴ encode lysines analogous to the K158 and K227 residues that are responsible for enriching IP6 into immature HIV-1 particles^10,45. Furthermore, IP6 is required for the assembly of the immature VLPs of several lentiviruses in vitro⁴⁶, suggesting that enrichment of IP6 into virions to support capsid assembly during maturation is likely a conserved feature of lentiviruses. Indeed, the presence of positively charged amino acid side-chains within helix 1 of CA at or near positions 18 and 25 is also well-conserved in lentiviruses^44,47. However, the placement of a positively charged amino acid side chain between these two positions (as established by the N21K mutation described in this study) is rare. An exception is bovine immunodeficiency virus (BIV), which encodes a lysine at position 21 along with a lysine at position 18 and an arginine at position 29. Other retroviruses also commonly encode positively charged residues at analogous positions. The alpharetrovirus Rous sarcoma virus (RSV) encodes an arginine at position 21, and cryo-ET data of RSV CLPs assembled in the presence of IP6 suggests that R21 (along with a lysine at position 17) coordinates an IP6 molecule at the center of RSV capsomers⁴⁸. Taken together with our findings, existing structural and sequence data suggest that IP6 is required for the assembly and stability of capsids from diverse retroviruses, but with considerable flexibility in its binding site.

Our finding that NSK capsids are inherently unstable but not assembly-defective indicates that the central pore plays a role in maintaining stability of the HIV-1 capsid post-entry. Most reported central pore mutations that reduce particle infectivity are unable to efficiently assemble capsids during particle maturation. Notable exceptions are R18G and R18L, which are capable of assembling into normal cones, but also display aberrant capsid morphologies^14,17,20. Given the importance of the central pore in binding IP6 and the ability of IP6 to stabilize capsids, our finding that mutations within the central pore can destabilize capsids seems intuitive. However, we were unable to detect notable differences in the ability of IP6 to facilitate CLP assembly or stabilize LSK and NSK CA hexamers (Figure 2B–C), suggesting that the differences in the inherent stability of LSK and NSK capsids is unrelated to their ability to bind IP6. One caveat regarding this conclusion is that our thermal stability measurements were performed with isolated, crosslinked CA hexamers. It is therefore possible that these experiments do not fully recapitulate the stabilizing effect of IP6 in the context of a complete capsid lattice composed of native hexamers and pentamers.

The LSK and NSK mutants differ at a single amino acid position (S16L vs. S16N). Why a leucine (LSK) at position 16 restores infectivity ~30-fold, whereas an asparagine (NSK) completely blocks infectivity in the context of the R18S/N21K mutations is unclear. This is especially surprising because the Ser-to-Asn substitution is a more conservative amino acid change than Ser-to-Leu. One possible explanation for this disparity in S16L vs. S16N is effects on dNTP import. The central pore is the channel through which dNTPs are putatively imported into the capsid to facilitate viral DNA synthesis. Importantly, the R18 and K25 rings have each been proposed to play a role in this process^16,17,49,50. Notably, it is the polar side-chains at residue 16 that negatively affect infectivity (S16 or N16) whereas the non-polar leucine (L16) increases it. It is therefore tempting to speculate that the hydrophobic substitution at position 16 is needed to promote the import of nucleotides into the capsid interior and disfavor dNTP dissociation back out of the pore. Whilst there is no direct evidence for the role of position 16 in nucleotide import, NSK is clearly impaired for reverse transcription compared to LSK. However, we did not observe substantial differences in the ability of ATP to bind or stabilize LSK and NSK CA hexamers (Figure S7). Consistent with the key role that R18 plays in dNTP binding to the central pore, both LSK and NSK exhibited a ~10-fold reduction in ATP binding affinity (Figure S7). This reduction in ATP binding is less severe than the ~100-fold reduction exhibited by R18G CA hexamers¹⁷, suggesting that the N21K mutation partially restores dNTP binding in the absence of the R18 ring. Because LSK remains capable of reverse transcribing its genome with near-WT efficiency (Figure 4A–C), we propose that this reduction in nucleotide affinity is not sufficient to alter the kinetics of viral DNA synthesis. Importantly, the lack of differences in nucleotide binding displayed by LSK and NSK support our conclusions that capsid stability, rather than dNTP import, is the primary mechanism driving the differences in LSK and NSK infectivity.

While we did not pursue biochemical studies to investigate the mechanisms by which the A31T and T216I compensatory mutations restore infectivity to R18S/N21K, we suspect a similar mechanism to that of S16L. A previous study found that T216I was able to restore capsid stability in target cells to the well characterized hypostable CA mutant P38A⁵¹. Furthermore, we previously showed that T216I restores infectivity to another central pore mutant (K25A) that displays a reduced interaction with IP6, likely by enhancing the stability at the interface between capsomers to compensate for reduced stability within capsomers¹⁴. Mutations at position A31 (to Gly, Asn, or Ser) are associated with viral escape from CD8⁺ T-cell responses in HLA-B*5703-positive individuals⁵². While these A31 mutations are generally associated with a reduction in fitness, their effect on capsid biology is not well understood. Further studies will be required to elucidate the importance of this residue in the structure and function of the capsid.

In summary, our findings highlight the flexibility of positively charged ring placement within the central pore of HIV-1 capsomers while simultaneously demonstrating the sensitivity of HIV-1 to changes within the central pore. Furthermore, our findings demonstrate the critical roles that the central pore plays in both the assembly of the HIV-1 capsid during particle maturation and its sustained stability post-entry.

Materials and Methods

Cells

MT4 and SupT1 T-cell lines were acquired from American Type Culture Collection (ATTC) and cultured in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, penicillin (100 U/mL), and streptomycin (100 μg/mL). HEK 293T, HeLa, and TZM-bl cells were acquired from ATCC or the NIH AIDS Reagent Program (TZM-bl) and cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented as above. The previously described 293T parental and IPPK KO cells⁴⁵ were cultured in DMEM supplemented as above. Whole blood was acquired from healthy donors via the NCI-Frederick Research Donor Program, and PBMCs were isolated using a Ficoll gradient (Histopaque-1077; Sigma no. 10771). PBMCs were cultured in RPMI 1640 supplemented as above and stimulated with 2 μg/mL phytohemagglutinin P (PHA-P) for 5 days prior to infection. Culture medium was supplemented with 50 U/mL IL-2 during replication.

Plasmids

pNL4–3, a lab-adapted, subtype B infectious molecular clone⁵³, was used to transfect MT4 cells for replication and forced evolution experiments and used to transfect 293T cells to produce virus for infectivity assays. The Env-defective molecular clone pNL4–3/KFS⁵⁴ was used as a non-replicating control for replication and forced evolution experiments. pNL4–3/KFS was also used to transfect 293T along with a plasmid expressing VSV-G to produce virus used to infect SupT1 cells for quantification of viral DNA synthesis. For confocal microscopy experiments, virions labeled with YFP-tagged HIV-1 Vif-resistant APOBEC3F mutant⁵⁵ were prepared by co-transfection of 293T cells with pNL4–3/KFS containing WT or mutant CA, pA3F-E289K-YFP, and a VSV-G expression plasmid⁵⁶. The A3F-E289K-YFP construct was generated by site-directed mutagenesis of pA3F-YFP²⁹. An env-/vpr-, luciferase encoding pNL4–3-based plasmid (pNL4–3.Luc.R-.E-)⁵⁷ was used to co-transfect 293T cells along with VSV-G to generate virus used for infectivity experiments in 293T parental and IPPK KO target cells. HIV-1 CA mutant pNL4–3 plasmids were generated by subcloning DNA fragments purchased from Twist Bioscience. For cryo-ET and endogenous reverse transcription experiments, virions were produced in 293T cells via co-transfection of pMDG2, a packaging plasmid encoding VSV-G (Addgene #12259), pCRV GagPol⁵⁸, and pCSGW⁵⁹ as described previously. For TIRF experiments, virions were produced as above with a pCRV GagPol iEGFP plasmid⁶⁰.

Replication Kinetics

3 × 10⁶ MT-4 cells were transfected with 3 μg pNL4–3 DNA in 0.7 mg/mL DEAE dextran dissolved in Dulbecco’s Phosphate Buffered Saline (DPBS) for 15 minutes at 37°C. The transfection mix was removed by centrifugation and the transfected cells were resuspended in complete RPMI medium. MT-4 cultures were split at a 1:1 ratio and supernatant samples were collected every 2–4 days for up to three months. To determine the kinetics of viral replication, HIV-1 reverse transcriptase (RT) activity in the supernatant was quantified using a previously described ³²P-based assay⁶¹. PBMCs were activated with PHA-P for 5 days and subsequently infected with 3 RT counts per minute (cpm) per cell for 2 hours at 37°C. PBMCs were then centrifuged and resuspended in complete RPMI medium supplemented with 50 U/mL IL-2 for ~2–3 weeks. Supernatant samples were collected every 2 days, and media were replenished without splitting the cells. Viral replication kinetics were determined as above.

Forced-evolution experiments

MT-4 cells were transfected with pNL4–3 harboring mutations coding for amino acid substitutions in the central pore of CA, and supernatants were collected every 2–4 days for up to three months. Cells and supernatant were collected at the peak of replication and genomic DNA was isolated using the QIAmp DNA Blood Mini kit (Qiagen – Catalog # 51106). Genomic DNA was subjected to polymerase chain reaction (PCR) using primers specific for the HIV-1 GagPol coding region. Amplified GagPol fragments were treated with ExoSAP-IT PCR Cleanup Reagent (Thermo Fisher Scientific Catalog # 78201.1.ML) and submitted for Sanger sequencing. RT activity in viral supernatants collected at the peak of replication was quantified, and 3 × 10⁵ RT cpm of virus was used to inoculate 3 × 10⁶ fresh MT-4 cells. Replication kinetics were measured to confirm reversion to WT replication, and genomic DNA isolation, PCR, and sequencing was repeated as above.

Single-cycle infectivity

293T cells were transfected with pNL4–3 plasmids overnight using linear polyethylenimine (PEI) and media were changed the following day. Virus in the supernatant was collected at 24 hours post media change, filtered through a 0.45 μm syringe filter, and stored at −80°C. Virus was quantified by ³²P RT assay and equal amounts of virus were used to infect TZM-bl cells. At 48 hours post infection, TZM-bl cells were lysed with britelite plus reagent (Revvity #6066766) and luminescence was quantified to determine infectivity. All measurements were normalized to signal acquired from WT-infected samples. For single-cycle infectivity measurements in IP6-depleted target cells, equal numbers of 293T parental and IPPK KO cells were seeded in matched plates and infected with VSV-G-pseudotyped, luciferase-encoding, ΔvprΔenv (pNL4–3.Lu.R-.E-) virions. Luciferase was quantified as above to determine the infectivity of each virus in each cell line.

Protein production and purification

The capsid (CA) proteins were expressed in E.coli C41 cells overnight at 25°C, lysed in lysis buffer (50 mM Tris-HCl (pH 8.0), 200 mM NaCl, 20% BugBuster, Protease inhibitor tablets, 1 mM DTT) and centrifuged (24,000 rpm, 1 hour). The supernatant was precipitated with 25% ammonium-sulphate (wt/vol) followed by centrifugation (13,000 rpm, 20 min). The precipitated CA was resuspended and dialyzed against 50 mM MES (pH 6.0), 20 mM NaCl, 1 mM DTT. The CA protein was further purified via a cation-exchange column with a gradient from 20 mM–1 M NaCl followed by size exclusion chromatography with Tris pH 8.0, 20 mM NaCl, 1 mM DTT, concentrated and snap frozen. Disulfide-stabilized CA A14C/E45C/W184A/M185A hexamers were expressed in E.coli C41 cells, lysed with sonication in 50 mM Tris-HCl (pH 8.0), 200 mM NaCl, 20% BugBuster 20 mM 2-mercaptoethanol, Protease inhibitor tablets. Lysate was then cleared by centrifugation (24,000 rpm, 1 hour). The supernatant was precipitated with 25% ammonium-sulphate (wt/vol) followed by centrifugation (13,000 rpm, 20 min). The pelleted material was resuspended in 50 mM citric acid (pH 4.5) and dialyzed against 50 mM Citric acid (pH 4.5), 20 mM 2-mercaptoethanol. The protein was further dialyzed with multiple changes into 50 mM Tris-HCl (pH 8.0), 1 M NaCl, 20 mM 2-mercaptoethanol. Reducing agent was removed by dialysis against 50 mM Tris (pH 8.0), 1 M NaCl, and then finally into 20 mM Tris (pH 8.0), 40 mM NaCl. Assembled hexamers were identified by non-reducing SDS–PAGE and were further purified via anion-exchange (2X 5ml HiTrap^® Q Fast Flow, GE Healthcare) in 20 mM Tris-HCl (pH8.0) 40 mM-1M NaCl gradient followed by size exclusion chromatography (Superdex 200) in 20 mM Tris (pH 8.0), 40 mM NaCl. Peak fractions were pooled and concentrated before snap freezing in 100μl aliquots and storage at −80°C.

Structural Projections

The HIV-1 CA hexamer structural model 6BHT¹⁰ was loaded into PyMol and the S16L, S16N, R18S, and N21K mutations were introduced via PyMol’s “Mutagenesis” function to generate structural projections for LSK and NSK hexamers. Measurements were taken via PyMol’s “Measurement” function from the β-carbon of Ser16 (WT) and the terminal amide group of Asn16 (NSK) to the nearest atom of the top IP6 molecule on each side of the central pore. For Leu16 (LSK) we generated measurements from each δ-carbon to the nearest IP6 atom to determine the shortest distance. Images were generated using PyMol’s “png” command and the measurements made in PyMol were added manually to the images generated.

Nano-differential scanning fluorimetry

DSF measurements were performed using a Prometheus NT.48 (NanoTemper Technologies) over a temperature range of 20–95°C using a ramp rate of 2.5°C / min. CA hexamer samples were prepared at a final concentration of 7.5 μM hexamer in Intracellular buffer (110 mM Potassium gluconate, 25 mM KCl, 5 mM NaCl, 2 mM MgCl₂, 10 mM HEPES). Ligands were added at 15 μM. DSF measurements were performed in triplicate.

Fluorescence polarization anisotropy

Fluorescence anisotropy measurements were performed at 22 °C on a Cary Eclipse Fluorescence Spectrophotometer (Agilent). Fluorescein-labelled ATP was obtained from Perkin Elmer and used for saturation binding experiments at a concentration of 2 nM. Proteins were diluted to give a range of two-fold dilutions with highest final hexamer concentrations of 4 μM for mutants with ATP and 400 nM hexamer for all other samples. Proteins were mixed with either FITC-ATP or FITC-HCB to give final probe concentrations of 2 nM and 1 nM respectively. Proteins and probes were mixed at the indicated concentrations and allowed to equilibrate at room temp. Fluorescence polarization was measured using Pherastar FSX (optic module FP 485 520 520). Samples were measured in triplicate with at least two independent runs. Binding curves were fit using GraphPad Prism (GraphPad Software, Inc.).

In vitro assembly assays

CA proteins were dialyzed against 50 mM MES (pH 6.0), 40 mM NaCl, 1 mM DTT. CA proteins at a final concentration of 100 μM were mixed with IP6 at 25°C (final concentration 1.25 – 5 mM). The apparent increase in absorbance reflecting increased light scattering (OD₃₅₀) was measured using a PHERAstar FSX Plate reader (BMG Labtech) in 384-well plate with shaking between each measurement at 25°C.

Negative-stain electron microscopy

Samples were assembled at room temp with 200 μM CA, 3 mM IP6 for mutants and 400 μM CA and 2.5mM IP6 for WT. 4 μL of sample from the assembly was put onto a glow discharged carbon coated grid (Cu, 400 mesh, Electron Microscopy Services), washed and stained with 2% Uranyl-acetate. Micrographs were taken at room temperature on a Tecnai Spirit (FEI) operated at an accelerated voltage of 120 keV and Gatan 2k × 2k CCD camera. Images were collected with a total dose of ~20 e/Å⁻² and a defocus of 1–3 μm.

Cryo-electron tomography

Virus-like particles were produced in HEK293T cells as previously described¹⁴. Supernatants were filtered through a 0.45 μm and 0.22 μm filter. Particles were concentrated by ultracentrifugation over a 20 % (wt/vol) sucrose cushion (2 hours at 28,000 rpm in a Beckman SW32 rotor; Beckman Coulter Life Sciences). The pellet was resuspended in PBS. Grids were prepared by the plunge-freeze method in liquid ethane using a FEI Vitrobot Mark II. Data collection and analysis was performed as previously described¹⁴ but briefly: Tomographic tilt series of LSK and NSK were acquired between −60° and +60° with increments of 3° using a dose symmetric scheme using Serial-EM⁶². Tomograms were reconstructed using IMOD 4.9⁶³. The alignment of 2D projection images of the tilt series was performed using gold beads as fiducial markers. Tomograms were reconstructed by back projection.

Reverse transcription in cells

2.5 × 10⁶ 293T cells were seeded into 10 cm dishes, cultured overnight, and co-transfected with pNL4–3 and a VSV-G expression vector. Virus-containing supernatant was collected and filtered as above and treated with Benzonase nuclease (Millipore Sigma #E1014) to remove carryover plasmid DNA. Virus was then pelleted through an 8% OptiPrep cushion and resuspended in complete DMEM containing amplification grade DNase I and 10 mg/mL MgCl₂. Virus was aliquoted, quantified by ³²P RT assay, and stored at −80°C until use. 5 × 10⁵ SupT1 cells per timepoint were infected with 10 RT cpm per cell of each virus for 2 hours at 4°C. Cells were harvested at 1, 2, 3, 6, 9, and 12 hours post-infection, washed once with PBS, lysed, and frozen for subsequent processing. DNA was isolated using the QiaAMP DNA Blood Mini kit according to the manufacturer’s instructions. Viral transcripts were quantified by qPCR of 5 μL of extracted DNA with 10 μL Taqman Fast Advanced Master Mix (Applied Biosystems #4444557) and 5 μL primer:probe mix on a QuantStudio 3 Real Time PCR system (Applied Biosystems). Previously described primer pairs and probes were used to detect early (minus-strand strong stop; MSSS) and late (second-strand transfer; SST) RT products and 2-LTR circles⁶⁴. CCR5 DNA was also quantified to control for cell number. Probes were labeled at the 5’ end with 6-carboxyfluorescein (FAM).

Endogenous reverse transcription

For ERT reactions HEK293T cells were seeded on 10 cm dishes and transfected as above but without VSVG-expressing plasmid. Four dishes were used per virus. After 6 hours, cells were washed with PBS to remove excess DNA and then incubated for 3 days in DMEM with 2% FBS. Supernatants were filtered through 0.45μm filters and incubated with 1/1000 benzonase (70746, Millipore) at room temperature for 1 hour. Supernatants were then spun through 20% sucrose cushion for 2 hours at 28k in SW28 and resuspended on ice with 500 μL 100 mM Tris pH 8, 150 mM NaCl. Aliquots were frozen for further use. ERT reactions were set up in 96-well plates in 20 μL volumes. 10 μL of freshly thawed VLPs were mixed with buffer containing 100 mM Tris pH 8, 150 mM NaCl, 2 mM MgCl₂, 0.5 mM TCEP and 1 mg/mL of BSA (B8667, Sigma). Reactions were supplemented with desired concentrations on IP6 (Sigma, 593648), dNTPs (28–4065-51, GE Healthcare), DNase (Sigma, DN25) at final concentration 0.1mg/ml and 50 nM Streptolysin O (SLO, Sigma, S5265). Plate was then covered with adhesive film, spun down and incubated at 37°C for 5 hours. To purify DNA we used Blood and Tissue Qiagen kit as follows: add 20μl reaction volume to 180 μL PBS and then mix with 200 μL lysis buffer, continue as per manufacturer instructions and elute DNA in 40 μL of water. DNA was quantified using Power SYBRGreen (Applied Biosystems) ABI StepOnePlus PCR System using standard SYBRGreen protocol. For standards we used CSGW vector that was used to generate VLPs. Primers were for early products (MSSS: minus strand strong stop), intermediate products (FST: forward strand transfer) and late products (SST: second strand transfer). Reactions were run as duplicates.

Streptolysin O (SLO) production

The 54 kDa (residues 101–571) version of the SLO was tagged with 6xHis by cloning into pOPTH-Tev vector for production in E.coli C41 strain. Cells were lysed with 20% BugBuster in 25 mM Tris pH8, 150 mM NaCl, 1 mM DTT, benzonase and lysozyme. Lysate was loaded onto a Ni column in 25 mM Tris pH 8, 150 mM NaCl, 1 mM DTT, 10 mM Imidazole, washed with buffer with 10 mM Imidazole, followed by 40 mM imidazole and then eluded with 200 mM Imidazole. 1 mL fractions were then checked on a polyacrylamide gel, peak fractions combined and dialyzed to 25 mM Tris pH 8, 150 mM NaCl and 1 mM DTT. Protein was then aliquoted and stored at −70°C.

TIRF microscopy

TIRF microscopy was carried out as previously described¹⁴. pCRV GagPol iEGFP construct was first generated by inserting EGFP between MA and CA of Gag with EGFP sequence flanked by PR cleavage sites for EGFP release from Gag upon maturation⁶⁰. To generate virions, constructs with corresponding capsid mutations were transfected at a ratio of 1 to 4 of pCRV GagPol iEGFP to pCRV GagPol. 6 hours after transfection, cells were washed and then incubated for 3 days with 2% FBS containing media. Virus containing supernatants were filtered as above and quantified by RT activity. Supernatants were diluted in PBS to equalize to RT units across viruses then placed on 8-well glass Ibidi dishes precoated with poly-L-lysine (Sigma, P4707) for binding. After two PBS washes, viruses were incubated for 30 min with 20 nM SLO in the presence or absence of 50 μM IP6 in in reaction buffer (SRB, 100 mM Tris-HCl pH 8, 150 mM NaCl). Untreated wells were used as control for normalization. After incubation wells were washed 2 times with PBS then fixed with 4% formaldehyde for 20 min. Images were acquired with a Nikon TIRF inverted microscope with a 100x/1.49NA oil-immersion objective, a 1.5x intermediate magnification and Prime95B sCMOS camera from Photometrics resulting in a 74nm pixel size. Particle numbers were analyzed in Fiji, where images were median filtered, and background subtracted. An intensity threshold was used to create a mask and a watershed step allowed separation of touching particles. ROI were filtered by area within 1 to 200 pixels.

Quantifying nuclear envelope docking and nuclear entry

Virions labeled with A3F-E289K-YFP were prepared by co-transfection of 293T cells with pNL4–3/KFS encoding WT or mutant CA, pA3F-E289K-YFP, and VSV-G expression vector²⁸. Supernatants from the transfected cells were collected 24 hours post-transfection, clarified using a 0.45 μm membrane filter, and concentrated by ultracentrifugation (100,000 × g) for 1.5 hours at 4°C through a 20% sucrose cushion (wt/vol) in 1X Dulbecco’s Phosphate-Buffered Saline with calcium and magnesium (PBS). The concentrated virus was resuspended in 500 μl DMEM and the number of YFP-labeled virions was quantified by single virion analysis⁶⁵. To quantify nuclear envelope association and nuclear import of HIV-1, HeLa-based cell lines were seeded in ibiTreated μ-slides (3 × 10⁴ cells/well; Ibidi) one day prior to infection. Cells were challenged with an equal number of A3F-E289K-YFP-labeled virus via spinoculation (1,200 × g for 1 hour at 15°C) in the presence of polybrene (10 μg/mL). Following spinoculation, the medium was replaced with prewarmed medium containing BioTracker 650 Red Nuclear Dye (0.5 μL/mL; Millipore). At 6 hours post-infection, the cells were fixed with 4% paraformaldehyde in 1X PBS for 10 minutes. Confocal z-slices at the equatorial plane of the cells were acquired using a Nikon Eclipse Ti-E microscope equipped with a Yokogawa CSU-X1 spinning disk unit and a Plan-Apochromat 100× N.A. 1.49 oil objective, using 514-nm (YFP) and 647-nm (BioTracker650 dye) lasers for illumination. Images were captured using a TwinCam system (Cairn) equipped with a 565-nm splitter and two iXon Ultra (Andor) cameras and analyzed with Nikon Elements or ImageJ⁶⁶. Each field of view (88 μm²) contained an average of 8 cells, and 7 fields were acquired per sample. A custom MATLAB program¹ was used to quantify the efficiency of nuclear envelope association and nuclear import of YFP signals. First, YFP signals were detected using Localize⁶⁷. Second, nuclear envelope and nuclear masks were generated from the BioTracker 650 signal. Finally, colocalization of each YFP spot with the nuclear envelope and nuclear masks was assessed. The percentage of cytoplasmic YFP spots colocalized with nuclear envelope masks and the percentage of total YFP spots colocalized with the nuclear masks was calculated. For display purposes, a pixel-averaging filter was applied to the images, and contrast was adjusted.

Statistical Analysis

All experiments were repeated at least three times unless otherwise stated. The number of repeats (N), determination of error, and statistical analysis is listed for each specific experiment in the figure legends.

Research Highlights.

• HIV-1 can adapt to the loss of its IP6-binding Arg-18 ring in capsomers upon introduction of a new positively charged ring within the central pore.

• The location of IP6-coordinating residues within the central pore of HIV-1 capsomers is flexible.

• The central pore of HIV-1 capsomers is critical for promoting both capsid assembly in virions and sustained capsid stability in target cells.