Derivation and characterization of an HIV-1 mutant that rescues IP6 binding deficiency

Background A critical step in the HIV-1 replication cycle is the assembly of Gag proteins to form virions at the plasma membrane. Virion assembly and maturation are facilitated by the cellular polyanion inositol hexaphosphate (IP6), which is proposed to stabilize both the immature Gag lattice and the mature capsid lattice by binding to rings of primary amines at the center of Gag or capsid protein (CA) hexamers. The amino acids comprising these rings are critical for proper virion formation and their substitution results in assembly deficits or impaired infectiousness. To better understand the nature of the deficits that accompany IP6 binding deficiency, we passaged HIV-1 mutants that had substitutions in IP6 coordinating residues to select for compensatory mutations. Results We found a mutation, a threonine to isoleucine substitution at position 371 (T371I) in Gag, that restored replication competence to an IP6-binding-deficient HIV-1 mutant. Notably, unlike wild-type HIV-1, the assembly and infectiousness of resulting virus was not impaired when IP6 biosynthetic enzymes were genetically ablated. Surprisingly, we also found that the maturation inhibitor Bevirimat (BVM) could restore the assembly and replication of an IP6-binding deficient mutant. Moreover, using BVM-dependent mutants we were able to image BVM-induced assembly of individual HIV-1 particles assembly in living cells. Conclusions Overall these results suggest that IP6-Gag and Gag-Gag contacts are finely tuned to generate a Gag lattice of optimal stability, and that under certain conditions BVM can rescue IP6 deficiency. Additionally, our work identifies an inducible virion assembly system that can be utilized to visualize HIV-1 assembly events using live cell microscopy. Supplementary Information The online version contains supplementary material available at 10.1186/s12977-021-00571-3.


Background
The HIV-1 Gag polyprotein which is composed of the matrix (MA), capsid (CA), spacer peptide 1 (SP1), nucleocapsid (NC), spacer Peptide 2 (SP2), and p6 domains, has central structural and functional roles in the HIV-1 replication cycle. During virion assembly, multimerization of the Gag polyprotein at the plasma membrane, primarily driven by the CA and NC domains, generates immature HIV-1 virions composed of radially oriented Gag hexamers [1][2][3][4][5]. Following assembly, and concomitant with or shortly after nascent particles are released, proteolytic processing of Gag by HIV-1 protease separates the aforementioned Gag domains [6]. The liberated CA protein undergoes a major structural rearrangement to form the mature conical core, composed of a lattice of CA hexamers with 12 CA pentamers, and is the salient feature of particle maturation [7]. Only after maturation are HIV-1 particles able to initiate new cycles of infection.
It has been previously shown that inositol phosphates play a critical role in both HIV-1 assembly and maturation. While assembly of HIV-1 Gag protein in vitro yields immature particles that differ in size and character Open Access Retrovirology *Correspondence: pbieniasz@rockefeller.edu 1 Laboratory of Retrovirology, The Rockefeller University, New York, NY, USA Full list of author information is available at the end of the article from authentic virions, addition of inositol phosphates to in vitro assembly reactions enables the production of particles that resemble authentic virions [8]. Further work identified inositol hexakisphosphate, or IP 6 , as the key mediator of this process. IP 6 , is a ubiquitous cellular polyanion containing 5 equatorial phosphates and a single axial phosphate, and facilitates formation of immature HIV-1 Gag lattice by binding to and stabilizing positively-charged rings of primary amines. These rings are formed by lysine residues at Gag positions 290 and 359 (K290 & K359) that are positioned at the center of the immature Gag hexamer [9]. Following the subsequent structural rearrangement of CA that accompanies maturation, IP 6 next binds to a second, distinct positively charged ring in the mature CA hexamer formed by arginine residues at CA position 18 (R18, Gag position R150). The R18 ring stabilizes the mature CA hexamer, and is required for viral DNA synthesis in newly infected cells [10][11][12]. It is thought that IP 6 is recruited into virions by interacting with K290 and K359 during immature particle production; this model is consistent with data demonstrating that HIV-1 K290A and HIV-1 K359A are significantly impaired in both viral production and IP 6 packaging, while HIV-1 K359I is assembly competent but generates poorly infectious particles [13].
The importance of each of the IP 6 -coordinating residues has been established, as mutagenesis of any such residue to an alanine (HIV-1 R18A, HIV-1 K290A , or HIV-1 K359A ) significantly impairs infectivity in either single cycle or spreading infection assays [9,13]. Additionally, yield of infectious virions is also substantially reduced in cells lacking key enzymes in the IP 6 biosynthetic pathway (IPPK or IPMK) or in cells overexpressing MINPP1, a phosphatase that dephosphorylates IP 6 [9,[13][14][15]. The IP 6 -coordinating amino acids are conserved among diverse lentiviruses, suggesting a general requirement for IP 6 [16].
While there is considerable evidence that perturbing IP 6 binding impairs HIV-1 replication, further investigation into the precise mechanisms underlying replication deficits is warranted. To better understand the role of IP 6 , we serially passaged virions containing substitutions in IP 6 -coordinating residues to identify second-site compensatory mutations that might rescue the resulting infectivity deficits. Accordingly, we found a single substitution that rescued the replication deficit observed in two IP 6 binding-deficient mutants. Using CRISPR/Cas9 knockout of IPMK, we show that the second-site substitution restored infectious virion yield despite loss of this IP 6 biosynthetic pathway. Strikingly, we also found that treatment with a maturation inhibitor Bevirimat (BVM) rescues infectivity of the IP 6 -binding-deficient mutant HIV-1 K359A . Indeed, using approaches in which the assembly of individual HIV-1 particles is imaged in living cells, we show that addition of BVM can induce the assembly of CA-mutant HIV-1 virions.

Identification of a second-site substitution that restores replication competence to IP 6 -binding deficient HIV-1 mutants
To identify second site changes that would rescue IP 6 binding deficient mutants, we passaged HIV-1 mutants encoding substitutions in IP 6 coordinating residues (HIV-1 R18A , HIV-1 K290A, or HIV-1 K359A ) in the highly-permissive MT4 T-cell line. Initial attempts, in which MT4 cells were infected with mutant viral stocks, were unsuccessful, likely due to the dramatically impaired fitness of these mutants and consequent inability to establish a sufficiently large population of infected cells to generate revertants. To overcome this problem, we instead cocultured MT4 cells with virus-producing 293T cells that had been transfected with HIV-1 R18A , HIV-1 K290A, and HIV-1 K359A proviral plasmids that encode GFP in place of nef. After removing the 293T cells, infected MT4 cells were co-cultured with uninfected MT4 cells, until most of the MT4 cells became infected (as monitored by visual inspection of GFP+ cells in the culture). Thereafter, cell-free supernatant was serially passed in MT4 cells (Fig. 1A).
For one mutant, HIV-1 K359A , observation of GFP positive cells suggested that an apparently compensatory mutation arose after approximately 2 weeks of passaging. PCR amplification and sequencing of Gag encoding sequences from this culture revealed the presence of a single nucleotide substitution in gag that resulted in a threonine to isoleucine substitution at Gag position 371 (Fig. 1B). No revertant mutants could be obtained for HIV-1 R18A or HIV-1 K290A . This finding may reflect a greater magnitude of impairment of these particular substitutions, making the generation of revertant mutants more difficult.
To determine whether the T371I mutant rescued the infectivity defect present in HIV-1 K359A , we generated a proviral clone, HIV-1 K359A/T371I , encoding both mutations and measured the infectious virion yield from proviral plasmid-transfected 293T cells. Addition of the T371I substitution to HIV-1 K359A restored infectious virion yield to wild-type levels (Fig. 1C). Although this secondsite, apparently compensatory change was identified only in the context of HIV-1 K359A , we asked whether the T371I substitution could rescue the HIV-1 K290A , given purported similar roles of K290 and K359 in binding IP 6 . Indeed, we found that HIV-1 K290A/T371I , unlike HIV-1 K290A , yielded similar levels of infectious HIV-1 virions to wild-type HIV-1.
To determine whether the effects of the T371I mutant were evident outside the context of transfected 293T cells, we performed spreading replication assays of HIV-1 WT , HIV-1 T371I , HIV-1 K359A , and HIV-1 K359A/ T371I in MT4 cells (Fig. 1D) and CEM cells (Fig. 1E). As expected, HIV-1 K359A replicated poorly in both cell types. HIV-1 T371I replicated poorly in CEM cells but well in MT4 cells, perhaps reflecting the greater permissiveness of MT4 cells. Importantly however, we found similar phenotypes for HIV-1 K359A/T371I in spreading replication assays in both MT4 and CEM cells; namely, addition of the T371I substitution to HIV-1 K359A restored replication, with the HIV-1 K359A/T371I double mutant exhibiting only a modest delay compared to HIV-1 WT in both cell types (Fig. 1D, E).

Infectious HIV-1 K359A/T371I particle yield is not affected by reduction of IP 6 synthesis in virus producing cells
Because the HIV-1 K359A is defective for IP 6 binding we next asked whether HIV-1 K359A/T371I retained infectiousness when cellular IP 6 levels were reduced. Using CRISPR/Cas9 we generated 293T cell lines lacking IPMK, an enzyme in the IP 6 synthetic pathway. Previous work has demonstrated IPMK knockout cells have greatly reduced levels of both IP 5 and IP 6 [13]. To account for potential clonal variation in capacity to generate HIV-1 particles, we used 3 separate IPMK targeting sgRNAs or a corresponding empty vector to generate ten independent single cell clones of IPMK knockout and WT control 293T cells ( Fig. 2A). The loss of IPMK was confirmed by DNA sequencing of target loci, which revealed the introduction of frameshift mutations and the absence of intact IPMK alleles. In agreement with previous studies, the yield of infectious HIV-1 WT virions from IPMK-deficient 293T cells was significantly decreased, by tenfold (p = 0.0091, Fig. 2A). The yield of HIV-1 K359A from 293T cells was greatly reduced compared to wildtype HIV-1 as expected, and was not further reduced by IPMK deficiency (Fig. 2B). Yield of HIV-1 T371I was also slightly reduced compared to wildtype but not impacted by IPMK deficiency (Fig. 2C, p = 0.1374). Importantly, the yield of HIV-1 K359A/T371I was only marginally reduced compared to wild type HIV-1 and there was no difference in yield of infectious HIV-1 K359A/T371I from WT 293T cells versus IPMK deficient 293T cells (Fig. 2D, p = 0.178).
As IP 6 deficiency impacts both HIV-1 particle production as well as infectivity, we assayed particle levels in the same supernatants from Fig. 2A-D using detection of reverse transcriptase with the SYBR-PERT assay. Importantly, we see the same phenotype for particle production as we do with infectivity assays: production of HIV-1 WT particles is impaired in IPMK KO cells, but there are no production deficits for HIV-1 K359A , HIV-1 T371I , or HIV-1 K359A/T371I (Fig. 2E-H). However, while there was a tenfold reduction in HIV-1 WT infectivity from IPMK-deficient 293T cells, we only observed a fivefold reduction in particle production in the exact same supernatants.

Infection of target cells with impaired IP 6 synthesis by HIV-1 WT or HIV-1 K359A/T371I
It has been proposed that residues K290 and K359 recruit IP 6 into HIV-1 virions during assembly, thereby providing the source of the IP 6 that binds to and stabilizes the R18 ring in the mature capsid core. The rationale for this idea stems from previous studies which have demonstrated that reduction of cellular IP 6 levels in target cells does not impact susceptibility to incoming infection [13,14]. Because HIV-1 K359A/T371I is fully infectious despite encoding a mutation that is predicted to diminish IP 6 packaging into virions, we next asked whether HIV-1 K359A/T371I requires IP 6 in target cells to be maximally infectious. We generated twelve IPMKdeficient MT4 target cell clones and six control clones and performed single cycle infection assays using HIV-1 WT and HIV-1 K359A/T371I (Fig. 2D, E). In agreement with previous studies [13], there was no difference in the infectiousness of HIV-1 WT in WT or IPMK-deficient MT4 cells ( Fig. 2D p = 0.3863). Moreover, there was no deficit in the infectiousness of HIV-1 K359A/T371I in WT or IPMK-deficient MT4 target cells (Fig. 2E, p = 0.4331), suggesting that HIV-1 K359A/T371I either does not require IP 6 for replication, or that the T371I mutation rescues both replication and IP 6 incorporation.

Bevirimat rescues infectious virion formation by the IP 6 -binding deficient mutant HIV-1 K359A
Notably, The T371I mutation identified herein had been described previously in a different context. Specifically, this substitution was reported to stabilize the immature CA-SP1 lattice, mimicking the effect of maturation inhibitors (MI) [17,18]. Therefore, we next asked whether maturation inhibitors themselves could rescue the deficit in infectious virion yield exhibited by HIV-1 K359A . As a control, we included the previously described assembly-defective, maturation inhibitor-dependent CA mutant HIV-1 P289S [18], We found that that BVM indeed rescued the infectivity of HIV-1 K359A and HIV-1 P289S in both single-cycle and spreading replication. Specifically, in single cycle assays, BVM increased the yield of infectious HIV-1 K359A virions, up to 50-fold, and in a dose-dependent manner (Fig. 3A) from transfected 293T cells. In spreading replication assays, BVM restored HIV-1 K359A replication to levels similar to that of BVM-treated wildtype virus in MT4 cells (Fig. 3B). BVM also rescued the spreading replication of HIV-1 K359A in CEM cells, indeed in this context the effect of BVM on HIV-1 K359A spreading was greater than that on the previously described MI-dependent mutant HIV-1 P289S (Fig. 3C).

BVM increases release of HIV-1 K359A virions independently of the viral protease
The interaction between K359 amines and IP 6 likely stabilizes the immature Gag lattice, Similarly, maturation inhibitors are known to bind to the immature Gag hexamers at approximal site and stabilize the immature CA-SP1 lattice [19,20]. Therefore, we hypothesized that BVM rescue particle formation by HIV-1 K359A by stabilizing an otherwise destabilized lattice, effectively serving as a functional replacement for IP 6 . To test this idea, we measured the release of HIV-1 K359A virions from BVM-treated 293T cells by western blotting. BVM indeed increased the yield of HIV-1 K359A virions, in a dose-dependent manner (Fig. 4A). To confirm that this effect was not due to BVM-mediated inhibition of Gag proteolysis, we performed similar experiments in virions containing an inactivating mutation in protease. We observed a similar dose-dependent increase in immature particle release, even in the context of protease inactivation (Fig. 4B), suggesting that the effect of BVM on HIV-1 K359A assembly is due to the direct effect of BVM on the immature lattice, not through inhibition of proteolytic cleavage at the Gag-SP1 junction.
Visualization of BVM-induced HIV-1 assembly observed in real time using live cell fluorescence microscopy The above data strongly suggested that BVM rescues infectivity and release of HIV-1 K359A by facilitating particle assembly. To directly observe effects on virion assembly, we performed fluorescence microscopy using a novel imaging construct based on HIV-1 NL4-3, in which Pol has been replaced by an HIV-1 codon-mimicking mNeon-Green, and in which Env and Vpu bear inactivating mutations. The resulting construct, herein referred to as HIV-1 NG, generates Gag-mNeonGreen in place of Gag-Pol during a single cycle of infection, thus allowing visualization of particle assembly as punctae at the plasma membrane.
We generated HIV-1 vectors particle containing this reporter and derivatives (HIV-1 NG WT, HIV-1 NG K359A ,  (Fig. 5A). However, when infections were done in the presence of BVM, there were clearly increased numbers of membrane associated punctae in HIV-1 NG K359A and HIV-1 NG P289S infected cells (Fig. 5A) suggesting BVM is able to directly facilitate particle assembly by these mutants.
This ability to induce HIV-1 particle assembly via addition of an exogenous small molecule has potential applications in imaging and other studies, as an inducible Quantification of Gag-NG punctae per timepoint depicts mean ± SD for images from C particle assembly system. In order to test the possible utility of this approach, we performed live cell widefield imaging studies using HIV-1 NG P289S , as this mutant displayed a greater responsiveness to BVM-induced assembly (see Figs. 3A, B, 4B, and 5A). We infected TZM-bl cells in the absence of BVM and then, at 26 h after infection, added BVM 5 μM and began acquiring images at 30 min intervals. In the absence of BVM, few punctae are apparent in HIV-1 NG P289S infected cells, even after 12 h of imaging (Fig. 5B, Additional files 1, 2). However, in the BVM-treated cells, substantially more punctae were evident, as soon as 30 min after BVM addition (Fig. 5B,  Additional files 3, 4).
Given that such striking differences could be observed as early as 30 min post BVM addition, we repeated these experiments with increased time resolution. TZMbl cells were infected with HIV-1 NG P289S for 26 h and treated with BVM as above, followed by immediate image acquisition at 3 min intervals. Assembly of HIV-1 P289S was rapidly induced by BVM (Fig. 5C, Additional files 5, 6) with substantial numbers of punctae forming within 30-120 min. We performed similar experiments using TIR-FM imaging and quantified the presence of Gag-NG punctae over time, with similar results. Specifically, in the absence of BVM there were very few punctae evident at the plasma membrane (Fig. 6A, B, Additional files 7, 8). However, shortly after the addition of BVM, numerous punctae rapidly formed at the plasma membrane (Fig. 6, Additional files 9, 10). These data provide proof of principle that such a system could be used to experimentally manipulate HIV-1 assembly for imaging or functional studies.

Discussion
Here, we report the identification of a single amino acid substitution (T371I) that rescues the replication of the defective, IP 6 -binding deficient mutant HIV-1 K359A . Despite several attempts we were unable to generate revertant mutants for HIV-1 K290A and HIV-1 R18A (although follow-up studies demonstrated that the T371I substitution rescues HIV-1 K290A as well as HIV-1 K359A ). The inability to generate revertant mutants for HIV-1 R18A and HIV-1 K290A is likely due to more substantial impairment. Indeed, previous groups have shown that HIV-1 K290A is impaired to a greater extent than HIV-1 K359A , potentially because K290 binds the 5 equatorial phosphates on IP 6 while K359 coordinates the single axial phosphate, suggesting a greater role for K290 in coordinating IP 6 [13]. The inability to generate a revertant mutant rescuing HIV-1 R18A could be explained by functions of this residue in addition to coordinating IP 6 , such as recruitment the cellular protein FEZ1 or as service as a conduit for dNTPs into the mature core [10,11,21,22].
HIV-1 K359A/T371I was fully infectious despite containing a mutation (K359A) that renders VLP assembly unresponsive to IP 6 in vitro and which substantially impairs IP 6 incorporation into virions [9,13]. Indeed, we found no significant reduction in yield of HIV-1 K359A/T371I from IPMK KO 293T cells, in contrast to WT HIV-1. This finding suggests that HIV-1 K359A/T371I is no longer dependent on IP 6 or requires substantially lower concentrations of IP 6 in virus producing cells.
In addition to its role in promoting immature particle assembly, IP 6 has also been implicated in stabilizing the mature lattice and promoting viral DNA synthesis by binding to a positively charged pore formed by R18 residues in the mature CA lattice [10,21]. Previously it was proposed that the source of the IP 6 required to stabilize the mature lattice in virions is selective recruitment by K290 and K359 residues, with IP 6 being liberated to bind R18 residues following disruption of the immature lattice after proteolysis [9,10]. In agreement with previous studies, we found that IPMK KO target MT4 cells were fully susceptible to infection by HIV-1 WT , indicating that IP 6 from target cells is not required to initiate a productive cycle of infection [13]. However, we also found no difference in the infectiousness of HIV-1 K359A/T371I in WT and IPMK KO MT4 target cells. This may reflect the possibility that other polyanions can fulfil a post assembly role. Indeed, recent studies demonstrated that other polyanions in mammalian cells such as glucose-1,6-bisphosphate can stabilize mature HIV-1 cores in vitro [23]. Alternatively, a very recent report has demonstrated that addition of the T371I substitution can rescue IP 6 incorporation in HIV-1 K359A virions to near WT levels [24]. Nevertheless, our data utilizing genetic ablation of the IP 6 synthetic machinery suggests that the T371I substitution confers reduced dependence on cellular inositol phosphate levels for virion assembly. Indeed, while production of HIV-1 WT was impaired in IPMK KO cells, there was no impairment of HIV-1 T371I or HIV-1 K359A/T371I in these cells. Thus HIV-1 T371I and HIV-1 K359A/T371I retains infectiousness even in the setting of reduced cellular IP 6 levels, implying at least some level of IP 6 independence, even if IP 6 is incorporated into virions in the context of the HIV-1 K359A/T371I double mutant. Alternatively, the lack of impairment of HIV-1 T371I or HIV-1 K359A/T371I in the setting of reduced cellular IP 6 levels maybe be due to the T371I substitution itself increasing affinity of the lattice for IP 6 , thus rendering virions less responsive to a reduction in cellular IP 6 levels. This notion is consistent with the recent report by Mallery et al. [24], demonstrating that the T371I substitution rescues incorporation of IP 6 in HIV-1 K359A. The T371I substitution, identified herein as a compensatory mutation that rescues infectivity deficit found in HIV-1 K359A , has previously been reported to rescue the infectivity of virions containing substitutions that confer resistance to maturation inhibitors (MIs). These substitutions render HIV-1 assembly-defective in the absence of MIs [18] and the T371I substitution was shown to stabilize the CA-SP1 lattice in this context, effectively mimicking the action of MIs [17]. Because the T371I substitution apparently mimics the effect of MIs, we hypothesized that MIs might also rescue the infectivity of HIV-1 K359A . Strikingly, we found that this was the case, and found that BVM can stimulate the assembly and release of HIV-1 K359A . That the stabilizing effects of both the T371I substitution and BVM can compensate for the lack of IP 6 coordination in HIV-1 K359A provides in vivo mechanistic support for the model proposed by Dick et al.: i.e. that binding of IP 6 to K290 and K359 residues stabilizes the immature lattice to drive particle assembly. Interestingly, while Mallery et al. [24] also demonstrated that MIs rescued HIV-1 K359A , they observed that PF-46396, but not BVM, rescued infection. This is potentially due to differences in the concentration of BVM used (5.0 µM vs 0.5 µM).
The ability to promote assembly via addition of a small molecule has potential utility in imaging and other studies of HIV-1 assembly. Indeed, when performing live cell time-lapse imaging, we observed BVM-induced assembly on the timescale of minutes. Such an experimental approach has potential utility for studying the sequence of events in HIV-1 viron assembly, such are RNA packaging, ESCRT protein recruitment, and protease activation, where virion assembly can be rapidly and (potentially) reversibly induced in real time simply by addition of a small molecule.
Together, these data support a model whereby stability of the immature CA lattice is finely tuned, with IP 6 coordinating and stabilizing the otherwise repulsive positive charges of K290 and K359 to drive assembly. Manipulations that cause IP 6 binding deficiency, either mutagenesis of K359 or decreasing IP 6 levels in producer cells, destabilize the immature lattice and decrease production of progeny virions. Conversely, manipulations such as the T371I substitution or treatment of HIV-1 WT with BVM, hyper-stabilize the immature lattice in the wild type context and decrease HIV-1 infectivity. However, either the T371I substitution or BVM treatment are able to rescue virion assembly and infectiousness in the context of IP 6 deficiency. Understanding how small molecules such as BVM or IP 6 can enhance Gag lattice stability can provide new tools to study virion assembly and potential avenues for antiretroviral therapeutics.

Conclusion
We identified a single-site revertant mutation, T371I, that rescues replication competence of the IP 6 -binding-deficient mutant HIV-1 K359A . Using CRISPR/ Cas9 to genetically ablate IP 6 biosynthesis, we showed that these resulting HIV-1 K359A/T371I virions are less dependent on cellular inositol phosphate levels. Remarkably, we also found that the maturation inhibitor BVM could restore the assembly and replication of HIV-1 K359A and developed an inducible particle assembly system using BVM-dependent HIV-1 mutants. In addition to providing insight on Gag-Gag and Gag-IP 6 interactions during HIV-1 assembly, our work also identifies an inducible virion assembly system that can be used in investigating HIV-1 assembly events in living cells.

Cells and media
293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Gibco) supplemented with 10% fetal calf serum and gentamycin. MT4 cells were maintained in Roswell Park Memorial Institute (RPMI) 1640 Medium (Gibco) supplemented with 10% fetal calf serum and gentamycin. Cells were maintained at 37 °C and 5% CO 2 . All transfections with viral plasmids were performed with polyethyleneimine.

Plasmid construction
All full-length proviral plasmids used in this study were based on the HIV-1 clone NHG, a previously described HIV-1 clone that encodes GFP in place of Nef (Accession number: JQ585717) [25]. Mutant viruses were derived from this parental plasmid using primer mutagenesis with fragments assembled into NHG digested with SpeI and SbfI using NEB HiFi  AAG CGC GCA CGG CAA GAG GCG AGG and dPol Xba  AS 5ʹ-CTA CTA TTC TTT CCC CTG CAC TCT AGA CTA  CTA CTT TAT TGT GAC GAG GGG TCG C; dPol Xba S  5ʹ-GCG ACC CCT CGT CAC AAT AAA GTA GTA GTC  TAG AGT GCA GGG GAA AGA ATA GTA G and dVpu-dEnv NheI AS 5ʹ-CTC CTC GCT AGC GTA CTA CTT ACT GCT TTG ATAG). The sequence encoding neon green (NG) was codon optimized to have nucleotide composition and codon usage similar to that of Pol using the Codon Optimization On-Line Tool from Singapore University (http:// cool. syncti/ org) and was synthesized by GeneArt. Neon Green was fused into the p6* frame of Pol through overlap PCR and inserted via SphI and XbaI Mutagenesis of this construct was accomplished using the same primers as above.

Viral stock production
293T cells were seeded at 6 × 10 6 cells per 10 cm dish and transfected the next day using polyethylenimine.

Generation of IPMK KO cell lines
The IPMK-targeting guides g1: ATG TAC GGG AAG GAC AAA GT; g2: GGT GGA CTC GAT CGC CGG TG; or g3: CCG GCC ACC TGA TGC GAG AG were designed using the Broad Institute GPP Web Portal and cloned into len-tiCRISPRv2 bearing a Hygromycin resistance cassette digested with BsmBI. lentiCRISPR v2 was a gift from Feng Zhang (Addgene plasmid # 52961; http:// n2t. net/ addge ne: 52961; RRID:Addgene_52961). VLPs were prepared as above, with the exception that 1 × 10 6 293Ts/ well were seeded in a 6 well plate and transfected the next day with 1 µg lentiCRISPRv2, 1 µg of Gag-Pol, and 0.2 μg of VSV-G. At 48 h post transduction of target cells with lentiCRISPRv2, cells were placed in selection with 100 μg/mL Hygromycin for ~ 10-14 days. Single cell clones were obtained by limiting dilution, and editing was verified by amplifying and sequencing target loci using primers: IPMK Seq F: 5ʹ-CGC TTC TGC TCT CCG TTA TG and IPMK Seq R: 5ʹ-GGA TTT GGC GTG CAC ACC AG and assessment using Synthego ICE, which identifies Indel frequency in Sanger sequencing data (Synthego Performance Analysis, ICE Analysis. 2019. v2.0.). Control cells were obtained similarly, using a lentiCRISPRv2 plasmid not harboring a sgRNA cassette.

Single cycle infectivity assays
WT control or IPMK KO 293T cells were seeded at 2.5 × 10 5 cells/well in a 24 well plate and transfected with 625 ng of HIV-1 WT, HIV-1 K359A , or HIV-1 K359A/T371I proviral plamids. Virions were prepared as above and titrated on MT4 cells. 24 h post infection, Dextran Sulfate was added (50 µg/mL) to limit infection to a single round. 48 h post infection, cells were fixed with 4% PFA and assessed via flow cytometry.

Spreading assays
5 × 10 4 cells per well were seeded in a 96 well plate and infected at an MOI of 0.001. 16 h post infection, cells were washed three times and placed in 5 µM BVM or DMSO control. Supernatants were collected at indicated timepoints, and levels of reverse transcriptase were quantified using the SYBR-PERT assay as previously described [26].

Western blotting
293Ts were seeded 5 × 10 5 cells/well in a 12 well dish and transfected the next day with 1.25 µg proviral plasmid. 48 h post transfection, cell lysates and virions pelleted through 20% sucrose (14,000xg for 90 min at 4 °C) were separated on a NuPage 4-12% Bis-Tris Gel (Invitrogen) and subsequently blotted onto a nitrocellulose membrane. Blots were blocked with Intercept Blocking Buffer (Li-Cor) and probed with primary antibody along with a corresponding IRDye 800CW-or IRDye 680-conjugated secondary antibody. Images were acquired using an Odyssey scanner (Li-Cor Biosciences). HIV-1 CA was detected using a human monoclonal anti-p24 (NIH AIDS Reagent Catalog #530). Imaging 5 × 10 4 TZM-bl cells per well were plated in a Lab-Tek Chamber Slide and infected the following day with indicated imaging construct at an MOI of 1. For fixed samples, cells infected in the presence or absence of 5 µM BVM were fixed 48 h post infection and imaged on a DeltaVision OMX SR imaging system using a 60X Widefield oil immersion objective (Olympus) with an exposure time of 50 ms, 10% Transmission, A488 nm laser. For live-cell samples, image acquisition began 26 h post infection, with cells placed in the presence or absence of 5 µM BVM at the time of image acquisition. Images were acquired at 37 °C, 5% CO 2 at indicated timepoints using a 60X Widefield oil immersion objective with an exposure time of 45 ms, 5% Transmission A488 nm laser. For TIR-FM, cells were imaged approximately 28-30 h post infection in the presence or absence of 5 µM BVM at 37 °C, 5% CO 2 . Images were acquired every 1 min for 90 min using a 60X RING-TIR-FM objective (Olympus Apo N 60X 1.49 Oil) with an exposure time of 100 ms, 10% Transmission A488 nm laser. Representative images were acquired, and all images were analyzed using Fiji (https:// fiji. sc/). Briefly, images were auto thresholded and Gag-NG punctae quantified using the Analyze Particles function in Fiji. Reported is the mean ± SD for displayed images (n = 2 per condition).

Graphing and statistical analysis
All graphs and corresponding statistical analyses were produced and analyzed with Graphpad Prism.