Inositol hexakisphosphate (IP6) and inositol pentakisphosphate (IP5) are required for viral particle release of retroviruses belonging to the primate lentivirus genus

Inositol hexakisphosphate (IP6) potently stimulates HIV-1 particle assembly in vitro and infectious particle production in vivo. However, knockout cells lacking the enzyme inositol-pentakisphosphate 2-kinase (IPPK-KO), which adds the final phosphate to inositol pentakisphosphate (IP5) to produce IP6, were still able to produce infectious HIV-1 particles at a greatly reduced rate. HIV-1 in vitro assembly can also be stimulated to a lesser extent with IP5, but it was not known if IP5 could also function in promoting assembly in vivo. IPPK-KO cells expressed no detectable IP6 but elevated IP5 levels and displayed a 20-100-fold reduction in infectious particle production, correlating with lost virus release. Transient transfection of an IPPK expression vector stimulated infectious particle production and release in IPPK-KOs but not in wildtype cells. Several attempts to make an IP6 and IP5 deficient stable cell line were not successful, but transient expression of multiple inositol polyphosphate phosphatase-1 (MINPP1) into IPPK-KOs resulted in the near ablation of IP6 and IP5. Under these conditions, HIV-1 infectious particle production and virus release were essentially abolished (1000-fold reduction). However, other retroviruses including a Gammaretrovirus, a Betaretrovirus, and two non-primate Lentiviruses displayed only a modest (3-fold) reduction in infectious particle production from IPPK-KOs and were not significantly altered by expression of IPPK or MINPP1. The only other retrovirus found that showed a clear IP6/IP5 dependence was the primate (macaque) Lentivirus Simian Immunodeficiency Virus (SIV-mac), which displayed similar sensitivity to IP6/IP5 levels as HIV-1. Finally, we found that loss of IP6/IP5 in viral target cells had no effect on permissiveness to HIV-1 infection. However, because it was not possible to generate viral particles devoid of IP6 and IP5, we were not able to determine if IP6 or IP5 derived from the virus producer cells is required at additional steps beyond assembly. Author Summary Inositol hexakisphosphate (IP6) is a co-factor required for efficient production of infectious HIV-1 particles. The HIV-1 structural protein Gag forms a hexagonal lattice structure. The negatively charged IP6 sits in the middle of the hexamer and stabilizes a ring of positively charged lysines. Previously described results show that depletion of IP6 reduces, but does not eliminate, infectious virus production. This depletion was achieved through knock-out of inositol-pentakisphosphate 2-kinase (IPPK-KO), the enzyme responsible for adding the sixth and final phosphate to the molecule. Whether IP6 is required, another inositol phosphate can substitute, or IP6 is simply acting as an enhancer for virus production was unknown. Here, we show that loss of IP6 and inositol pentakisphosphate (IP5) abolishes infectious HIV-1 production from cells. We do this through a cell-based system using transiently expressed proteins to restore or deplete IP6 and IP5 in the IPPK-KO cell line. We further show that the IP6 and IP5 requirement is a feature of primate lentiviruses, but not all retroviruses, and that IP6 and IP5 is required in the producer but not the target cell for HIV-1 infection.


61
The HIV-1 structural protein (Gag) is produced in the cytoplasm and traffics to the plasma 62 membrane where it assembles into a viral particle that buds from the host membrane [1]. Gag is a 63 polyprotein consisting of the Matrix (MA), Capsid (CA), Spacer 1 (SP1), Nucleocapsid (NC), Spacer 2 64 (SP2), and p6 domains [1,2]. During assembly, the Gag protein assembles into an 'immature' hexagonal 65 lattice, driven primarily by interactions involving the CA and SP1 domains [1,3]. The C-terminal CA and 66 SP1 domains contain an alpha-helix that forms a six-helix bundle with the other Gag proteins in the 67 hexamer [4,5]. This bundle is important in formation and stabilization of the immature lattice [4,5].

68
During or shortly after budding from the cell, the viral protease cleaves the Gag polyprotein into its 69 constitutive components, which separates CA from SP1 and eliminates the six-helix bundle [1,6]. The 70 liberated CA protein then assembles into a structurally distinct 'mature' lattice which forms the viral 71 core [1,7].

72
Early attempts to assemble full length HIV-1 Gag protein in vitro revealed that proper assembly 73 required the presence of cell lysate [8]. This pointed to an assembly co-factor that catalyzed viral 74 assembly in cells. Further research revealed that inositol phosphates were sufficient to stimulate proper 75 assembly, but the mechanistic basis for this effect was poorly understood [3]. Recently, a Cryo-EM 76 reconstruction of in vivo-produced immature HIV-1 particles revealed a small density above the CA-SP1 77 six-helix bundle that was coordinated by two rings of lysine residues, suggesting the presence of a 78 negatively charged molecule inside the particle that helped stabilize the bundle [4]. This evidence for 79 such a molecule, in conjunction with previous data that inositol phosphates stimulate assembly, 80 prompted further evaluation of the role of inositol phosphates as HIV-1 assembly co-factors [3,8,9]. In 81 particular, Inositol hexakisphosphate (I(1,2,3,4,5,6)P 6 or IP6), which is a hexagonal six-carbon ring with a 82 negatively charged phosphate at each position, seemed like a likely match for the density identified in 83 particles [10].

84
In assembly experiments in vitro, the presence of IP6 was found to potently promote immature 85 assembly and even to modulate whether particular Gag proteins assembled into immature or mature 86 lattices [10]. Mutation of the lysine residues in Gag believed to coordinate the negatively charged 87 molecule made the Gag proteins 100-fold less responsive to IP6 in in vitro assembly reactions [10].

88
When a crystal structure of the HIV-1 CA CTD SP1 protein in the presence of IP6 was solved, a density was 89 observed associated with the six-helix bundle that precisely matched the density described in the Cryo-90 EM reconstruction [4,10]. These biochemical and structural data strongly support the conclusion that 91 the density observed in HIV-1 particles is indeed IP6, but the data could not reveal whether IP6 is a 92 requirement for HIV-1 assembly in vivo.

93
IP6 is found in mammalian cells at concentrations of 10-100uM [11], and is synthesized by a 94 series of host enzymes through a complex and not fully resolved process (Fig 1A)

119
We previously showed that IP6 stimulates immature in vitro HIV-1 particle assembly [10]. We 120 further showed that knock out of IPPK, the only enzyme known to catalyze addition of the final 121 phosphate in the generation of IP6 (Fig 1A and B (Fig 2A-C). In contrast, the IPMK-KO cells had 139 residual levels of both IP6 and IP5 (Fig 2A-C). This was consistent with a previous report that showed 140 that cells from an IPMK knockout embryo were also shown to produce very low levels of IP5 and IP6 141 through an unknown mechanism [21,22].

142
Next, we measured infectious HIV-1 virus particle production from HEK293FT cells and its two 143 derivative knockout lines. An HIV-1 ΔEnv provirus containing a GFP reporter (HIV-CMV-GFP) was co-144 transfected with a VSV-G expression construct into the three cell lines in parallel, and the media was 145 titered on fresh target cells ( Fig 2D). As we reported previously, infectious particle production from the 146 IPPK-KO cell line was reduced ~20-100 fold compared to HEK293FT cells ( Fig 2E) [10]. The IPMK-KO cells 147 displayed a more modest ~5-fold reduction in infectious particle production that corresponded with the 148 residual IP6 levels found in the cells ( Fig 2E). We then tested whether the block in infectious virus 149 particle production was due to a block in virus release or to reduced infectivity of released virus. To 150 accomplish this, we measured the Gag/CA protein level in producer cells and the supernatant. Western 151 blots with an antibody against p24 CA revealed that HEK293FT, IPPK-KO, and IPMK-KO cells all produced 152 full length Gag at relatively equal levels ( Fig 2F,

177
In contrast, exogenous expression of MINPP1 reduced, but did not ablate, IP6 and IP5 levels ( ). As before, infectivity was reduced 20-100-fold from IPPK-KO cells ( Fig 5B). Addition of IPPK to 203 HEK293FT cells had no appreciable effect on this number, but addition to IPPK-KO cells enhanced 204 infectious particle production by approximately 10-fold ( Fig 5B). Likewise, addition of MINPP1 to 205 HEK293FT cells also had no appreciable effect on infectious particle production, but addition to IPPK-KO 206 cells further reduced infectivity by approximately 10-fold, which approached background levels ( Fig 5B).

207
Western blotting was again used to determine at which step in infectious virus particle production was 208 blocked ( Fig 5C).

235
To test the role of IP5 depletion in susceptibility of target cells, we next transduced the MINPP1-236 IRES-GFP vector or an empty IRES-GFP control ( Fig 4A) into HEK293FT or IPPK-KO cells, and then tested 237 their susceptibility to infection (Fig 6B). Two days after transduction with MINPP1 IRES-GFP or IRES-GFP, 238 cells were transduced with VSV-G pseudotyped HIV-1 ΔEnv virus containing a CD4 reporter (Fig 6C). Two 239 days after virus transduction, surface CD4 was stained with an APC-conjugated antibody to score for 240 successful virus transduction (Fig 6D).   (Fig 7A). Infectious MLV 258 particle production from IPPK-KO was reduced a few-fold compared to HEK293FT cells; however, this 259 reduction could not be modulated further by addition of IPPK or MINPP1. Expression of IPPK actually 260 caused a small but statistically insignificant reduction in infectious particle production compared to 261 empty vector (Fig 7A). With expression of MINPP1 in IPPK-KO cells, there was no difference in virus 262 output compared to empty vector ( Fig 7A). These data suggest that the 3-fold reduction in virus particle 263 release with MLV does not reflect a direct IP6 or IP5 requirement.

264
We next tested the Betaretrovirus Mason-Pfizer Monkey Virus (MPMV). As with HIV-1 and MLV, 265 expression of IPPK and MINPP1 in HEK293FT cells did not affect infectious particle release (Fig 7B).

266
Infectious particle production was again slightly decreased from IPPK-KO cells, but neither IPPK nor 267 MINPP1 expression altered this output ( Fig 7B). As with MLV, these data suggest that MPMV does not 268 have a strict IP6 or IP5 requirement for infectious particle production. Additionally, amino acid 269 sequence alignment between HIV-1, MLV, and MPMV CA proteins shows no homology to the K290 and 270 K359 residues that interact with IP6 and IP5 in HIV-1 (Fig 7C) [10].

272
The IP6 and IP5 requirement is conserved across primate lentiviruses. had similar outputs to HIV-1 (Fig 8A). Infectious particle production was reduced over 20-fold from 278 IPPK-KO cells relative to HEK293FT cells. Importantly, infectious particle production was partially 279 restored with addition of IPPK and precipitously reduced by the addition of MINPP1. As with MLV and 280 MPMV, transfection of IPPK-KO with EIAV and FIV produced about 3-fold fewer infectious virus particles 281 than HEK293FT cells (Fig 8B-C). However, neither virus was significantly affected by introduction of IPPK 282 or MINPP1 (Fig 8B-C). Comparison of the protein sequence alignments shows homology between all four lentiviruses at K290 and K359; however, prolines upstream and downstream of K290 are not 284 conserved for FIV and EIAV (Fig 8D). Together, these data point toward an IP6 and IP5 requirement for 285 primate lentiviruses but not lentiviruses of other species.

286
We next wanted to determine if the sensitivity observed in infectious particle production is 287 reflected in an assembly assay in vitro with purified proteins. We have reported previously that IP6 288 stimulates assembly of EIAV particles, despite the lack of dependence in infectivity assays [40]. We 289 therefore chose to test the stimulation of HIV-1, SIV, FIV, and EIAV in in vitro assembly reactions at pH8 290 and different IP6 concentrations (Fig 9). As expected, addition of as little as 5 µM IP6 stimulated robust 291 assembly of immature, spherical virus like particles (VLPs) for HIV-1 and SIV-mac (Fig 9A-B