Phosphoinositides Direct Equine Infectious Anemia Virus Gag Trafficking and Release

doi:10.1111/j.1600-0854.2010.01153.x

Traffic. Author manuscript; available in PMC 2012 April 1.

Published in final edited form as:

Traffic. 2011 April; 12(4): 438–451.

Published online 2011 February 1. doi: 10.1111/j.1600-0854.2010.01153.x

PMCID: PMC3064743

NIHMSID: NIHMS266610

Phosphoinositides Direct Equine Infectious Anemia Virus Gag Trafficking and Release

Fiona Fernandes,¹ Kang Chen,² Lorna S. Ehrlich,¹ Jing Jin,³ Min H. Chen,¹ Gisselle N. Medina,¹ Marc Symons,⁴ Ronald Montelaro,⁵ Julie Donaldson,⁶ Nico Tjandra,² and Carol A. Carter¹^*

¹ Dept. of Molecular Genetics & Microbiology, S.U.N.Y, Stony Brook, NY 11727, USA

² Laboratory of Molecular Biophysics, NHLBI-NIH, Bethesda, MD 20892, USA

³ Section of Microbial Pathogenesis, Yale University School of Medicine, New Haven, CT 06536, USA

⁴ Feinstein Institute, Manhasset, NY 11030, USA

⁵ Dept. of Microbiology and Molecular Genetics, Center for Vaccine Research, University of Pittsburgh, Pittsburgh, PA 15261

⁶ Laboratory of Cell Biology, NHLBI-NIH, Bethesda, MD 20892, USA

^* Corresponding author: Carol A. Carter, Department of Molecular Genetics and Microbiology, Life Sciences Bldg., Rm 248, Stony Brook University, Stony Brook, NY 11794-5222. Telephone:(631) 632-8801; Telefax: (631) 632-9797; Email: ccarter/at/ms.cc.sunysb.edu

The publisher's final edited version of this article is available at Traffic

See other articles in PMC that cite the published article.

Abstract

Phosphatidylinositol 4,5-biphosphate (PI(4,5)P₂), the predominant phosphoinositide on the plasma membrane, binds the matrix (MA) protein of Human Immunodeficiency Virus type 1 (HIV-1) and Equine Infectious Anemia Virus (EIAV) with similar affinities in vitro. Interaction with PI(4,5)P₂ is critical for HIV-1 assembly on the plasma membrane. EIAV has been shown to localize in internal compartments hence the significance of its interaction with PI(4,5)P₂ is unclear. We therefore investigated the binding in vitro of other phosphoinositides to EIAV MA and whether intracellular association with compartments bearing these phosphoinositides was important for assembly and release of virus-like particles (VLPs) formed by Gag. In vitro, EIAV MA bound PI(3)P with higher affinity than PI(4,5)P₂ as revealed by NMR spectra upon lipid titration. Gag was detected on the plasma membrane and in compartments enriched in PI(3,5)P₂. Treatment of cells with YM201636, a kinase inhibitor that blocks production of PI(3,5)P₂ from PI(3)P, caused Gag to co-localize with aberrant compartments and inhibited VLP release. In contrast to HIV-1, release of EIAV VLPs was not significantly diminished by co-expression with 5-phosphatase IV, an enzyme that specifically depletes PI(4,5)P₂ from the plasma membrane. However, co-expression with synaptojanin 2, a phosphatase with broader specificity, diminished VLP production. PI-binding pocket mutations caused striking budding defects, as revealed by electron microscopy. One of the mutations also modified Gag-Gag interaction, as suggested by altered bimolecular fluorescence complementation. We conclude that phosphoinositide-mediated targeting to peripheral and internal membranes is a critical factor in EIAV assembly and release.

Keywords: PI(4,5)P₂; PI(3)P; PI(3,5)P₂; HIV-1; EIAV; VLP; multivesicular bodies; bimolecular fluorescence complementation (BiFC)

Retroviral assembly and release are directed by the structural precursor polyprotein Gag. The Gag polyprotein consists of multiple domains: matrix (MA), capsid (CA), nucleocapsid (NC) and smaller peptides [1]. These domains are cleaved by the virally-encoded protease in the final stages of Gag assembly to release mature forms of the structural proteins that reorganize to form infectious particles. Gag is the sole viral protein required for immature virus particle production and when expressed alone forms virus-like particles (VLPs) [1]. The MA domain in HIV Gag plays a dual role in viral particle assembly: It targets Gag to the site of viral assembly [2, 3] and facilitates Gag-membrane binding [4–6]. A cluster of basic residues within HIV-1 MA facilitates the association of Gag with phosphatidylinositol 4,5-biphosphate (PI(4,5)P₂), a phosphoinositide present on the inner leaflet of the plasma membrane. This association, as well as the self-association of Gag monomers, triggers exposure of a myristate moiety sequestered in MA, which further promotes membrane binding and Gag multimerization [7–10].

Like HIV-1, EIAV is a member of the lentivirus subgroup of retroviruses and replicates in macrophages. EIAV MA differs from HIV-1 MA in lacking a myristoyl signal but behaves similar to HIV-1 MA in existing in a monomer-trimer equilibrium in vitro and in binding to PI(4,5)P₂ with weak affinity [11, 12]. Thus, just as the interaction of PI(4,5)P₂ with HIV-1 MA is proposed to induce conformational changes that favor protein multimerization [7], binding of PI(4,5)P₂ to EIAV MA might similarly promote Gag assembly. Unlike HIV-1 Gag, which accumulates on the plasma membrane, EIAV Gag has been reported to localize to both the cell interior and to the plasma membrane [13, 14]. This suggests that the MA domain of EIAV Gag might target the protein to phosphoinositides present both at the cell periphery and on intracellular vesicles. Supporting this, we demonstrate in this study that, in vitro, phosphatidylinositol 3-phosphate (PI(3)P) a phospholipid that resides on early endosomes [15–17], binds EIAV MA with higher affinity than PI(4,5)P₂. Moreover, we show that, in cells, EIAV Gag co-localizes with markers of membrane compartments containing PI(3)P, phosphatidylinositol 3,5-biphosphate (PI(3,5)P₂) and PI(4,5)P₂ at steady-state. In contrast to HIV-1, where depletion of PI(4,5)P₂ from the plasma membrane has been shown to alter Gag localization and to inhibit particle release, similar treatment had little effect on EIAV Gag. However, depletion of additional phosphoinositide pools by co-expressing Gag with synaptojanin-2 (Sjn-2), a broad specificity phosphoinositide phosphatase [18, 19] or with YM201636, a PIKFyve kinase inhibitor [20], impacted both localization and budding. Mutation of K49, a residue in the phosphoinositide binding pocket of MA whose NMR chemical shift was affected by all phosphoinositides tested, inhibited VLP release from the plasma membrane. Mutation of PI binding pocket residues distal to K49 did not prevent intracellular multimerization or VLP release but altered Gag trafficking significantly. We conclude that interactions with phosphoinositides during assembly is a critical aspect of EIAV Gag trafficking and release.

Results

EIAV MA exhibits a preference for PI(3)P-containing phospholipids in vitro

We previously reported the interaction of EIAV MA with PI(4,5)P₂ [11]. To determine whether the protein recognized other phosphoinositides, PI-C₄ with phosphate groups in different positions of its inositol ring were tested by NMR as previously described [11]. Chemical shift changes of backbone amide resonances were observed in the ¹H-¹⁵N heteronuclear single quantum coherence spectra of MA upon titration of PI(3)P-C₄, PI(3,4)P₂-C₄ and PI(3,5)P₂-C₄. The chemical shift change profiles at 1:1 stoichiometry of protein:lipid were plotted. As shown in Figure 1, the residues exhibiting significant chemical shift changes were among the same as previously reported for PI(4,5)P₂ (i.e., K49, L104 and Y108[11].

Figure 1

NMR-observed chemical shift changes in EIAV MA induced by binding to PI(4,5)P₂-C₄, PI(3,5)P₂-C₄, PI(3)P-C₄ and PI(3,4)P₂-C_4.

The position of these residues in the EIAV MA structure is shown in Figure 2A. They mostly clustered into the surface area around S100 (Figure 2C) and K49 (Figure 2B). Though the binding profiles among the different phosphoinositides were similar (Figure 1), some residues were selectively perturbed, e.g. G88, a residue at the binding surface (Figure 2A), recognized only those with phosphate in the 4 position of the inositol ring, suggesting that the surface around S100 distinguished the different head-groups. Supporting this, the S100 residue was significantly perturbed by phosphatidylinositol 3,4-biphosphate (PI(3,4)P₂ and PI(3,5)P₂ phosphoinositides but not PI(3)P or PI(4,5)P₂ (Figure 1). This residue is located in the putative dimer interface [21] and may form part of the binding pocket as depicted in Figure 2C.

Figure 2

Schematic representation of mutations in the EIAV MA structure

Besides specific residues, the binding affinities varied among different phosphoinositides. Apparent K_d values were derived based on the observation of NMR chemical shift changes for the most strongly perturbed residues, K49 and Y108, as a function of PI concentration. Table 1 shows the average K_d based on residues K49 and Y108 binding with the difference used to define the K_d error range. Supplementary Figure 1 shows the curve fitting plots. PI(3)P bound EIAV MA with higher affinity than PI(3,5)P₂, PI(4,5)P₂ or PI(3,4)P₂. These data indicate that EIAV MA has a strong preference for phosphoinositides present in membrane compartments other than the plasma membrane.

TABLE 1

K_d Values (μm) for di-C₄-PI Binding to EIAV MA^*

EIAV Gag co-localizes with markers of internal and peripheral membranes

Since, in vitro, EIAV MA bound phosphoinositides present on endocytic compartments with significantly greater affinity than those on the plasma membrane (i.e. PI(4,5)P₂), we determined whether Gag co-localized with compartments containing the phosphoinositides or with phosphoinositide-interacting proteins that mark the membrane compartments. In all cases Pearson’s coefficient of correlation [22] was determined for multiple (10–15) Gag positive cells, as described in Material and Methods. A Pearson coefficient of 0.6 or higher was used to define significant co-localization under these conditions and the percentage of cells exhibiting this value is reported in Table 2A and andB.B. We used a GFP-tagged pleckstrin homology domain (PH) from phospholipase C δ (GFP-PH_PLCδ) to determine if EIAV Gag co-localized with PI(4,5)P₂. As reported previously [2, 23], the PH domain of PLCδ, which binds specifically to PI(4,5)P₂, localizes to the plasma membrane. EIAV Gag co-localized with GFP-PH_PLCδ on the plasma membrane in 35% of cells expressing Gag confirming its interaction with PI(4,5)P₂ (Figure 3A; Table 2A). This is consistent with the observation that Gag exhibited a predominantly dispersed punctate distribution with only ~25% of the cells showing plasma membrane localization exclusively. No significant level of co-localization of EIAV Gag with anti-PI(3)P antibody or with early endosome antigen 1 (EEA1), a protein with a FYVE domain that binds PI(3)P, was observed at 24–48 hours post-transfection [24] (Figure 3B and Figure 3C, respectively). However, we determined whether association might be transient by utilizing YM201636 to inhibit PIKFyve, the kinase that converts PI(3)P to PI(3,5)P₂ [20]. Inhibition of PIKFyve kinase blocks PI(3,5)P₂ production and induces the formation of swollen vesicles derived partly from endosomal material since EEA1 localizes to these vesicles [20]. As shown in panel 3D and Table 2B, in 45% of the Gag-positive cells, Gag was associated with aberrant compartments induced by the drug. Since PI(3,5)P₂ is found mainly in late endosome/multivesicular bodies (LE/MVBs), we tested for Gag co-localization with markers associated with both the internal and limiting membranes. Gag was detected on the limiting membrane of LE/MVB, as indicated by the Lamp-3 marker, in only 20% of the Gag-positive cells (Table 2A). However, in ~40% of the Gag-positive cells, Gag co-localized with lyso-bis-phosphatidic acid (LBPA), a marker for internal membranes [25] within the LE/MVB compartment (panel 3E; Table 2A), indicating that some of the protein was sorted through this compartment. These results indicate that EIAV Gag associates with several phosphoinositides under steady-state conditions, suggesting it is targeted to both internal and peripheral membranes of the cell.

TABLE 2A

EIAV Gag Association with Peripheral and Interior Membrane Compartments

TABLE 2B

Association with Interior Membrane Compartments: Effect of PI Lipid Inhibitors

Figure 3

Co-localization of EIAV Gag with internal and peripheral membrane markers

Depletion of PI(4,5)P₂ perturbs HIV-1 but not EIAV VLP production

Previous studies demonstrated that HIV-1 Gag and murine leukemia virus (MLV) Gag interact with PI(4,5)P₂ in cells [2, 26]. 5-ptase IV, a type IV phosphatase that is targeted to the membrane through a CAAX domain, depletes intracellular levels of PI(4,5)P₂ and PI(3,4,5)P₃ [27]. Co-expression with 5-ptase IV has been shown to inhibit both HIV and MLV Gag release [2, 26]. To determine whether PI(4,5)P₂ also plays a major role in EIAV assembly and release, we examined the effect of 5-ptase IV on EIAV VLP production. As reported previously [2], HIV-1 VLP production, as indicated by the intensity of the media-associated Gag signal in Western analysis, was inhibited by a low level of expression of 5-ptase IV (Figure 4A, lanes 1–4). In contrast, EIAV VLP production was not inhibited by 5-ptase IV expression (lanes 9–12). A mutated form of 5-ptase IV that lacked the catalytic domain (5-ptase IV-Δ1) did not have a detectable effect on HIV-1 or EIAV VLP production, as expected (panel 4A lanes 5–8, 13–16). Examination of cell lysates revealed that the expression of enzymatically active or inactive 5-ptase IV (Panel 4C lanes 2–4, 6–8, 10–12 and 14–16) did not diminish accumulation of the HIV-1 and EIAV Gag proteins (panel 4B). A quantitative analysis of VLP release efficiency, i.e., the amount of VLPs detected in the media divided by the sum of VLPs in media and Gag in the cell lysate, is shown in panel 4C. Under conditions where HIV-1 VLP release was inhibited in a dose-dependent manner, EIAV Gag release was minimally affected. At higher expression levels, 5-ptase IV inhibited EIAV Gag to a small extent (data not shown). Nevertheless, the results indicate that depletion of PI(4,5)P₂ does not interfere with EIAV Gag assembly and release to the same extent as it affects HIV-1.

Figure 4

5-ptase IV inhibits HIV but not EIAV VLP release

We used confocal microscopy to further investigate the effect of 5-ptase IV on HIV-1 and EIAV Gag subcellular localization. Consistent with previous findings, 5-ptase IV dramatically changed the distribution of HIV-1 Gag from predominantly plasma membrane localization in cells expressing Gag alone (Figure 5, panel A1 left cell) to mainly internal in cells co-expressing Gag and 5-ptase IV (panels A1–A2, right cell). This effect was specific, since plasma membrane localization was not disrupted following co-expression of HIV-1 Gag with the Δ1 mutant (panels A3 – A4). In contrast, under similar conditions, EIAV Gag distribution was not detectably altered whether it was co-expressed with active (panels A5 – A6) or inactive (panels A7–A8) enzyme. The same result was obtained in a cell where the viral protein exhibited exclusively plasma membrane accumulation (Figure 5B): EIAV Gag was found to completely rim the plasma membrane in the control cell in the field (top cell, panel 5B1) and was still predominantly located at the plasma membrane in a cell co-expressing the enzyme (bottom cell, panels B1–B3). This strong resistance to 5-ptase IV was observed in 90% of twenty Gag-positive cells.

Figure 5

5-ptase IV alters HIV but not EIAV Gag localization

Previous studies demonstrated that HIV-1 Gag is re-directed from the plasma membrane to internal compartments enriched in PI(4,5)P₂ when co-expressed with ADP-ribosylation factor 6/Q67L (Arf6/Q67L; panels 5C1 and C2) [2]. Arf6/Q67L is a constitutively active form of Arf6 which causes intracellular accumulation of PI(4,5)P₂-enriched endosomal structures [28, 29]. As shown in panels 5C3 and C4, an apparently similar level of Arf6/Q67L expression in cells containing EIAV Gag induced Gag clustering in some (solid arrow) but not other (open arrows) cells. The reason for this variability is unknown; nevertheless, these observations support the conclusion that EIAV Gag is significantly less impacted than HIV-1 Gag by depleting PI(4,5)P₂ from the plasma membrane.

The polyphosphatase synaptojanin-2 inhibits EIAV VLP production

As 5-ptase IV depletion did not result in significant inhibition of EIAV VLP release, we sought to determine if perturbation of other phosphoinositide levels affected EIAV Gag assembly or release. The phosphatase synaptojanin-2 (Sjn-2) contains two catalytic domains: a 5-phosphatase type II domain that specifically hydrolyzes PI(4,5)P₂, PI(3,4,5)P₃, Ins(1,4,5)P₃ and Ins(1,3,4,5)P₄ and a Sac 1 domain that hydrolyzes PI(3)P, PI(4)P and PI(3,5)P₂ [18, 19] (Figure 6, panel A). The enzyme can be directed to the plasma membrane through its Rac1-binding domain (RBD). Co-transfection of Cos-1 cells with DNA encoding EIAV Gag and wild-type Sjn-2 resulted in dose-dependent inhibition of EIAV VLP production (panel 6B, lanes 1–3). In contrast, no inhibition was detected following co-transfection of Gag with Sjn-2 containing inactivating mutations in both the 5-phosphatase and Sac 1 domains (Sjn-2**; panel 6B lanes 4–6). Quantitative analysis of VLP release efficiency (panel 6C) indicated that co-expression with Sjn-2 reduced VLP release at least 5-fold (n = 2 independent trials). Co-expression of Gag with the 5-phosphatase type II domain (i.e., the PD) alone had no effect on VLP release (data not shown). These results suggest that maintenance of steady-state PI(3)P, PI(4)P and/or PI(3,5)P₂ levels are important for efficient EIAV Gag VLP production. However, our attempts to delineate the role of the various phosphoinositides by engineering mutations in the catalytic site of either the Sac 1 domain alone or the PD alone in the context of the full length Sjn-2 protein yielded no reproducible effects. Possibly, the domains act cooperatively and cannot function independently.

Figure 6

Sjn-2 inhibits EIAV VLP release

Figure 7 shows the effect of Sjn-2 on HIV-1 and EIAV Gag subcellular distribution. As shown in panel A, HIV-1 Gag association with the plasma membrane (panel A1) was significantly disrupted by co-expression with Sjn-2 (panels A2 and A3). In contrast, the subcellular distribution of EIAV Gag appeared to be minimally perturbed (panel B, compare B1 to B2–B4). To determine whether the EIAV Gag on interior membranes was directed to a different compartment in the presence of Sjn-2 despite appearing minimally disturbed, we tested for Gag association with Lamp-3 in the presence and absence of Sjn-2 expression. As noted above and shown in panel 7C, most cells (80% of those expressing Gag) failed to exhibit co-localization of Gag and Lamp-3. However, the percentage exhibiting co-localization changed from 20 to 80% in cells co-expressing Gag and Sjn-2 (panel D; Table 2B). Taken together with the findings in Figure 6, the results suggest that Sjn-2-mediated depletion of phosphoinositides on internal membrane compartments alters trafficking and release of EIAV Gag.

Figure 7

Effect of Sjn-2 on EIAV and HIV-1 Gag localization

An inhibitor of PI(3,5)P₂ synthesis interferes with EIAV VLP release

To provide direct evidence that targeting to an intracellular membrane compartment is important for EIAV Gag release, we determined the effect of inhibitors of PI(3)P and PI(3,5)P₂ synthesis. Inconclusive results due to cell toxicity were obtained with LY294002 (data not shown), a widely-used and specific inhibitor of phosphatidylinositol 3 kinase, the kinase responsible for PI(3)P production [30]. However, YM201636, the inhibitor of PI(3,5)P₂ formation described above [20], diminished EIAV VLP production in a dose-dependent manner (Figure 8A). YM201636 did not diminish Gag accumulation inside the cell indicating the defect in VLP production occurred at the level of release. A quantitative analysis of VLP release efficiency (panel B) indicated that YM201636 inhibited release by about two fold. Based on these results and the results described above indicating that Gag was associated with a compartment induced by YM201636 treatment, we conclude that targeting to a compartment bearing PI(3,5)P₂ is necessary for efficient VLP release.

Figure 8

YM201636 inhibits EIAV Gag VLP production

Effect of PI-binding pocket mutations on Gag trafficking and release

We determined the effect of substituting Ala for residues in the PI binding pocket. K49 and S100 were selected initially based on their respective broad and selective recognition of phosphoinositides (c.f. Figure 1). As revealed by Western analysis (Figure 9A) and quantitative assessment of VLP release efficiency (panel 9B), the K49A mutant was as severely inhibited in VLP release as a mutant lacking the L domain, Δp9. In contrast, the S100A mutant was only slightly impaired. Mutation to alanine of T66, a residue predicted to form part of the dimer interface [21] had no apparent effect. Similar results were obtained in HeLa and equine dermal cells (data not shown). The transfected Cos-1 cells were examined by confocal and electron microscopy to further investigate the effect of the mutations. While the K49A mutant behaved similarly to the WT Gag protein with respect to all of the markers tested (compare Figure 10, panels A, D and F to Figures 3C, 3E and and7C)7C) and, similar to WT Gag, clustered with the YM201636-induced vesicles (compare Figures 3D and 10B), the S100A mutant exhibited a co-localization pattern that was significantly different from that of WT Gag or K49A-Gag: It co-localized extensively with EEA-1 (Figure 10 panel C) and Lamp-3 (Figure 10 panel G) and not with LBPA (panel E) {summarized in TABLE 2A}. To determine whether mutations in PI pocket residues near S100 would have similar effects, we made a radical change in L104 (L104A). A conservative change was made at position Y108 (Y108F) to minimize structural disruption. As shown in the Western analysis in Figure 9C and the quantitative assessment of VLP release efficiency in panel 9D, these mutants behaved like S100A in that neither was impaired in VLP release. Also like S100A and unlike WT Gag or K49A-Gag, co-localization with Lamp-3 was detected in >85% of the cells expressing the Y108F mutant and 100% of the cells expressing L104A (compare Figure 10, panels G, H and I to Figures 7C and 10F; summarized in Table 2A). Together these results indicate that residues S100, L104 and Y108 are all required for wild-type Gag targeting while residue K49 is required for efficient VLP release.

Figure 9

Release efficiency of phosphoinositide binding pocket mutants

Figure 10

Co-localization of K49A- and S100A-Gag mutants with internal membrane markers

Since our previous studies indicated that PI(4,5)P₂ binding triggered changes in MA oligomer formation that were altered by the K49A mutation [11], we examined the possibility that this mutation affected a structural change that is important for assembly and budding. We chose to test for changes in Gag-Gag interactions using bimolecular fluorescence complementation (BiFC) because previous studies demonstrated that EIAV Gag interacts homotypically in cells [31, 32]. In this assay, associations at a distance of 10 nm or less between non-fluorescent N- and C-terminal fragments (VN and VC, respectively) of the monomeric Venus fluorescent protein (VFP) result in reconstitution of VFP and fluorescence when VN and VC are fused to other proteins, such as Gag, that can multimerize. We engineered the K49A and the S100A mutations into the previously described WT EIAV VN and VC constructs, transfected these vectors into Cos-1 cells and then examined the cells by confocal microscopy.

Figure 11 shows cells transfected with DNA encoding WT Gag fused to VN alone (panels A and E) or co-transfected with plasmids encoding the VN and VC fragments fused to WT Gag (panels B and F); K49A-Gag (panels C and G); or S100A-Gag (panels D and H). By examining sections in the z plane of the cell, we determined the location of Gag multimerization. As previously described [31, 32], co-transfection of Gag-VN and Gag-VC together (panels B and F), but not transfection of either fragment alone (panel A and data not shown) reconstituted fluorescence, indicating that interactions between the two fragments mediated by Gag were necessary to achieve efficient VFP reconstitution. The BiFC signal was bright, punctate and was evident at the cell periphery (0 μm, panel B) and in the cell interior (1.2 μm, panel F) with some concentration in the perinuclear region. This is as previously reported, suggesting that these Gag-Gag-BiFC signals correctly represent intracellular Gag assembly. Similarly, BiFC signals were detected in plasma membrane-proximal and perinuclear regions of all cells expressing S100A-Gag (panels C and G, respectively). In cells expressing K49A-Gag, the punctate BiFC signal was detected at the cell periphery (panel D). However, in contrast to WT- or S100A-Gag, the signal was less robust in the cell interior (panel H). Since indirect immunofluorescence emanating from K49A-Gag was detected at both the cell periphery and the cell interior (c.f., Figure 10A, 10D and 10F), the failure to reconstitute VFP was not due to a lack of Gag accumulation in the cell interior. Rather, the reduced BiFC signal in the cell interior suggests that K49A-Gag was not associated in an arrangement that favors BiFC complex formation. We conclude that the substitution of Ala for residue K49 in the PI binding pocket interferes with the Gag-Gag interactions that take place in the cell interior but not those that occur on the plasma membrane.

Figure 11

K49A mutation alters Gag-Gag interaction revealed by BiFC

Cells transfected with DNA encoding the WT or mutated Gag-HA proteins were examined by electron microscopy (Figure 12). No cell-associated particles were detected in cultures transfected with DNA encoding WT Gag (not shown). This is a consistent finding in our studies [33]. In contrast, cell-associated VLPs ~100 nm in diameter were detected in cells that had been transfected with DNA encoding Δp9-, K49A-, or S100A-Gag. VLPs assembled from Δp9-Gag were detected in vesicles inside some of the cells (panel A, upper panel; enlarged in lower panel), however most were tethered to the cell periphery (panel B; 90% of 30 sections containing particles in two independent experiments). Similarly, K49A-VLPs inside cytoplasmic vesicles were detected in 10% of the sections with particles (panel C); VLPs tethered to the cell periphery were detected in most of the sections (panel D; 90% of 30 sections in two independent trials). Thus, the K49A defect resembled that exhibited by retroviral L domain mutants [34]. Consistent with its WT-like VLP release efficiency, S100A-VLPs inside cytoplasmic vesicles (panel E, upper panel; a VLP is enlarged in lower panel) or tethered to the cell periphery (panels F and G) were observed infrequently (each at 10% of 30 sections, n = 2). These observations, and the fact that S100A was released at essentially WT levels, support the conclusion that the tethered S100A particles did not comprise a significant proportion of the released VLPs. Overall, we conclude that residues in the PI binding pocket of EIAV MA are important determinants of Gag targeting, assembly and release.

Figure 12

Examination of cell-associated K49A and S100A particles by electron microscopy

Discussion

Previous studies have demonstrated that the Gag proteins of several retroviruses recognize PI(4,5)P₂ [2, 7, 26] and that the interaction of Gag MA with PI(4,5)P₂ can facilitate protein-protein interactions involved in HIV-1 Gag assembly and trafficking to the site of particle release [2, 35]. In this study we examined the role of phosphoinositides in EIAV Gag assembly and release. In vitro, EIAV MA bound several phosphoinositides with different affinities; MA exhibited the highest affinity for PI(3)P. In cells, EIAV Gag accumulated on vesicles enriched in PI(3,5)P₂ and PI(4,5)P₂. By depleting phosphoinositides through over-expression of 5-ptase IV and Sjn-2, we found that the steady-state levels of the PI(4,5)P₂ was not critical for Gag assembly, while depletion of phosphoinositides associated with internal membranes interfered with Gag release to a significantly greater extent. Additionally, inhibiting PI(3,5)P₂ production from PI(3)P using YM201636 reduced the efficiency of Gag release. Together these observations suggest that targeting to the endocytic compartments containing PI(3)P and PI(3,5)P₂ phospholipids is an important aspect of EIAV Gag trafficking and release.

Although HIV-1 is principally found on the plasma membrane, EIAV Gag was found predominantly associated with internal compartments. We showed that a subpopulation of WT EIAV Gag accumulates in endosomes. We speculate that EIAV Gag is normally targeted to endosomal compartments containing PI(3)P and PI(3,5)P₂ and sorted from these vesicles back to the plasma membrane by its affinity for PI(4,5)P₂. This notion is supported by observations that VLPs of bothΔp9-Gag, a mutant defective in a very late step in assembly, and K49A-Gag were detected at the cell periphery, as revealed by electron microscopy. The observation that K49A-Gag accumulated in endocytic compartments to ~the same extent as WT Gag (Table 2A) but did not multimerize like the WT in this location (Figure 11) and, moreover, exhibited reduced accumulation on the plasma membrane relative to WT Gag (Table 2A) and was defective for VLP release (Figure 9) suggests that binding the lipids on intracellular vesicles is a critical step. Previous studies demonstrated that PI(4,5)P₂ binding triggers major structural reorganization, including inducing the myristyl switch in HIV-1 MA [7] and changes in HIV-1 and EIAV Gag oligomer formation [11, 35, 36]. Perhaps EIAV Gag interaction with PI(3)P and/or PI(3,5)P₂ in particular induces conformational changes that facilitate productive EIAV Gag multimerization. After all, EIAV MA binds these phospholipids with high affinity while HIV-1 Gag does not bind them at all (7, 11). Presumably, changes resulting from mutation of S100 alter the route, but do not prevent, trafficking through these endocytic compartments. Indeed, compared to the WT Gag protein, relatively more S100A accumulated on the cytoplasmic face of PI(3)P- and PI(3,5)P₂-containing compartments (Table 2A).

An alternative to the notion that PI binding induces conformational changes required for productive assembly and release is the idea that, in contrast to HIV, PI binding is NOT required for EIAV Gag trafficking. In this case, the defect in viral particle release following mutation of K49 would be attributable to its potential role in trimer stabilization or formation of higher-order assemblages of the trimer (c.f., Figure 2). We do not favor this interpretation (i) because EIAV and HIV MA exhibit structural and functional conservation and, although EIAV lacks N-terminal fatty acid modification, we expect that it conserves the PI-induced conformational change that occurs with HIV. Moreover (ii), although mutation of K49 to Ala did not alter Gag localization as revealed by confocal microscopy, it exhibited striking defects when examined by electron microscopy (c.f., Figure 12). Similar mutation of S100, a PI-binding pocket residue that is distal to K49 in the MA structure and not part of the trimer interface resulted in similar budding defects. Most likely, the mutations interfered with interactions between the lipid-interacting Gag protein assemblages and PI(4,5)P2-membrane microdomains at the budding site. S100A was released more efficiently than K49A; perhaps it is blocked at a later budding stage: We noted that its membrane tether appeared to be thinner. (iii) In the BiFC assay (c.f., Figure 11), K49A exhibited reduced Gag-Gag interaction in the cell interior but not at the periphery. If its release defect was exclusively due to disruptions in trimer stabilization or formation of higher-order assemblages of the trimer, defective protein-protein interaction should have been detected throughout the cell.

Further investigations to define the precise role of phosphoinositides in retroviral assembly, trafficking and release are warranted and may reveal new opportunities for development of antiviral strategies.

Materials and Methods

NMR spectroscopy and phosphoinositide phosphate (PIP) titration

The purification of ¹⁵N-labeled EIAV MA was described [11]. PI(4,5)P₂-C₄, PI(3,5)P₂-C₄, PI(3)P-C₄, and PI(3,4)P₂-C₄ were obtained from Echelon Biosciences Inc. (Salt Lake City, UT). Both PIPs and protein were dissolved in the same pH 4.4 10 mM sodium acetate buffer. PIPs were prepared in a stock concentration of 3–4 mM. The initial EIAV MA sample had a volume of 250 μL and a concentration of 200 μM. The titration was performed at 10 °C with sequential addition of PIPs at concentrations ranging from 30 to 600 uM. The ¹H-¹⁵N heteronuclear single quantum coherence spectra were collected for every titration. The normalized ¹H-¹⁵N chemical shift deviation and disassociation constant was calculated according to eq 1 and 2, respectively (c.f.,[11]). All NMR experiments were performed on a Bruker Avance 800 MHz spectrometer equipped with a cryo-probe and Z pulse field gradient.

Cell culture and transfection

Cos-1 cells were maintained in DMEM with 10% fetal bovine serum and penicillin streptomycin. Transfection of Cos-1 cells was performed using Fugene transfection reagent (Roche) as described by the manufacturer. Forty-eight hours after DNA transfection, tissue culture media was removed and saved for VLP isolation; cells were scraped in lysis buffer (50 mM Tris-HCL, pH7.5, 250 mM NaCl, 1mM EGTA, 5 mM MgCl2, 1% Nonidet P-40, 10% glycerol, 1× Roche Protease Inhibitor Cocktail) using a rubber policeman and pelleted. For isolation of VLPs, the culture media was passed through a 0.45 μm filter and then centrifuged through a 20% sucrose cushion at 36,000 rpm.

Plasmids, mutagenesis and DNA cloning

The EIAV Gag-HA construct and HIV Gag-HA constructs were described previously [31, 32]. The EIAV Gag-VN and –VC constructs for BiFC also were described previously [31, 32]. The K49A, S100A, L104A and Y108F mutants were constructed in the background of the EIAV Gag-HA or BiFC constructs using site-directed mutagenesis. The Sjn2-GFP and -Myc constructs were described previously [37]. The Sjn2**-GFP double mutant was made by swapping an AccI and Sufi fragment from a pre-existing Sjn-2 PD* single mutant to a Sjn-2 Sac* single mutant [38]. Constructs encoding 5-phosphatase IV tagged with Myc and a mutant lacking the catalytic site (5-ptase-IV and Δ1, respectively) were kind gifts from Drs. Ono and Freed [2]. The GFP-PH-_PLCδ and RFP-Arf6/Q67L constructs were previously described [28].

Antibodies and reagents

Reagents were obtained as follows: Phosphoinositide antibodies from Echelon Biosciences; anti-actin rabbit monoclonal antibody, anti-HA monoclonal antibody and anti-GFP monoclonal antibody from Sigma; Ride 800 anti-rabbit IgG from Rockland; Alexa Fluor 680 anti-mouse IgG and Texas Red goat anti-mouse IgG secondary antibodies from Molecular Probes. The PIKFyve kinase inhibitor YM210636 and the PI3 kinase inhibitor LY294002 were purchased from Symansin and Cell signaling, respectively.

Western analysis

Proteins were separated by electrophoresis through 8% SDS-polyacrylamide gels and electroblotted onto nitrocellulose membranes. Following incubation with appropriate primary and secondary antibodies, proteins were visualized using an infrared-based imaging system (Odyssey, LI-COR Biosciences). For analysis of particle release efficiency, the signal intensity on the Western blots corresponding to Gag in VLPs and Gag in cell lysates was determined. Release efficiency was defined as the ratio of the signal intensity value for the VLP-associated Gag to the sum of the values for VLP-associated Gag plus cell lysate-associated Gag (i.e., VLP/[cell lysate + VLP]).

Confocal Microscopy

Cells were plated on coverslips and fixed in 4% formaldehyde (Fisher) for 20 minutes. Permeabilization was done in 0.5% Saponin for 15 minutes or, where indicated, in 0.1% Triton-X-100. The cells were stained with either mouse or rabbit anti-HA to detect Gag-HA and appropriate secondary antibodies tagged with either FITC or Texas red. Hoechst stain was used to detect the nucleus. All images were captured on an inverted fluorescence/differential-interference contrast (dic) Zeiss Axiovert 200M deconvolving fluorescence microscope operated by AxioVision Version 4.5 (Zeiss) software. Ten to twenty optimal sections along the z axis were acquired in increments of 0.4 μm. Captured images show whole cell or, where indicated, the central section. The fluorescent data sets were deconvolved by using the constrained iterative method (AxioVision). To quantify relative co-localization of signal from two (red and green) channels, Pearson correlation coefficients were obtained using Image J quantification software down-loaded from NIH website http://rsbweb.nih.gov/ij/index.html. Co-localization analysis was performed using the co-localization finder plug-in. The significance level of correlation coefficients was assessed by reference to http://physics.mercer.edu/Younce/pearson.html.

Electron microscopy

Cultures grown on Permanox tissue culture dishes (Nunc Nalgene) were fixed with 2.5% glutaraldehyde, 1% paraformaldehyde in 0.12M sodium cacodylate buffer, pH 7.3–7.4, post-fixed with 1% OsO₄, and en-bloc stained with 1% uranyl acetate in H₂O, dehydrated in an ethanol series followed by propylene oxide and embedded in Epon. Thin sections cut parallel to the growth surface were stained with uranyl acetate and lead citrate and were viewed with a JEM-1200EX electron microscope (JEOL USA) equipped with an AMT XR-60 digital camera (Advanced Microscopy Techniques) or a FEI Tecnai12 BioTwinG² transmission electron microscope equipped with a AMT XR-60 CCD Digital Camera System

Supplementary Material

Supp Figure S1

Click here to view.^{(239K, jpeg)}

Acknowledgments

We thank W. Mothes for helpful discussion and critical reading of the manuscript and we thank G. Piszczek for sedimentation analysis of MA. This study was supported by NIH R56 and R01 awards AI068463 (to C. C.), and the Intramural Research Program of the NIH, National Heart, Lung, and Blood Institute (to J.D. and N.T.). We thank the NHLBI-NIH electron microscopy facility and the Stony Brook University Central Microscopy Imaging electron microscopy Core facility for their services.

References

1. Wills JW, Craven RC. Form, function, and use of retroviral gag proteins. AIDS. 1991;5:639–54. [PubMed]

2. Ono A, Ablan SD, Lockett SJ, Nagashima K, Freed EO. Phosphatidylinositol (4,5) bisphosphate regulates HIV-1 Gag targeting to the plasma membrane. Proc Natl Acad Sci U S A. 2004;101:14889–94. [PubMed]

3. Ono A, Freed EO. Binding of human immunodeficiency virus type 1 Gag to membrane: role of the matrix amino terminus. J Virol. 1999;73:4136–44. [PMC free article] [PubMed]

4. Facke M, Janetzko A, Shoeman RL, Krausslich HG. A large deletion in the matrix domain of the human immunodeficiency virus gag gene redirects virus particle assembly from the plasma membrane to the endoplasmic reticulum. J Virol. 1993;67:4972–80. [PMC free article] [PubMed]

5. Gallina A, Mantoan G, Rindi G, Milanesi G. Influence of MA internal sequences, but not of the myristylated N-terminus sequence, on the budding site of HIV-1 Gag protein. Biochem Biophys Res Commun. 1994;204:1031–8. [PubMed]

6. Ono A, Demirov D, Freed EO. Relationship between human immunodeficiency virus type 1 Gag multimerization and membrane binding. J Virol. 2000;74:5142–50. [PMC free article] [PubMed]

7. Saad JS, Miller J, Tai J, Kim A, Ghanam RH, Summers MF. Structural basis for targeting HIV-1 Gag proteins to the plasma membrane for virus assembly. Proc Natl Acad Sci U S A. 2006;103:11364–9. [PubMed]

8. Tang C, Loeliger E, Luncsford P, Kinde I, Beckett D, Summers MF. Entropic switch regulates myristate exposure in the HIV-1 matrix protein. Proc Natl Acad Sci U S A. 2004;101:517–22. [PubMed]

9. Paillart JC, Gottlinger HG. Opposing effects of human immunodeficiency virus type 1 matrix mutations support a myristyl switch model of gag membrane targeting. J Virol. 1999;73:2604–12. [PMC free article] [PubMed]

10. Resh MD. A myristoyl switch regulates membrane binding of HIV-1 Gag. Proc Natl Acad Sci U S A. 2004;101:417–8. [PubMed]

11. Chen K, Bachtiar I, Piszczek G, Bouamr F, Carter C, Tjandra N. Solution NMR characterizations of oligomerization and dynamics of equine infectious anemia virus matrix protein and its interaction with PIP2. Biochemistry. 2008;47:1928–37. [PMC free article] [PubMed]

12. Provitera P, Bouamr F, Murray D, Carter C, Scarlata S. Binding of equine infectious anemia virus matrix protein to membrane bilayers involves multiple interactions. J Mol Biol. 2000;296:887–98. [PubMed]

13. Puffer BA, Watkins SC, Montelaro RC. Equine infectious anemia virus Gag polyprotein late domain specifically recruits cellular AP-2 adapter protein complexes during virion assembly. J Virol. 1998;72:10218–21. [PMC free article] [PubMed]

14. Tanzi GO, Piefer AJ, Bates P. Equine infectious anemia virus utilizes host vesicular protein sorting machinery during particle release. J Virol. 2003;77:8440–7. [PMC free article] [PubMed]

15. Dumas JJ, Merithew E, Sudharshan E, Rajamani D, Hayes S, Lawe D, et al. Multivalent endosome targeting by homodimeric EEA1. Mol Cell. 2001;8:947–58. [PubMed]

16. Kutateladze TG, Ogburn KD, Watson WT, de Beer T, Emr SD, Burd CG, et al. Phosphatidylinositol 3-phosphate recognition by the FYVE domain. Mol Cell. 1999;3:805–11. [PubMed]

17. Odorizzi G, Babst M, Emr SD. Fab1p PtdIns(3)P 5-kinase function essential for protein sorting in the multivesicular body. Cell. 1998;95:847–58. [PubMed]

18. Guo S, Stolz LE, Lemrow SM, York JD. SAC1-like domains of yeast SAC1, INP52, and INP53 and of human synaptojanin encode polyphosphoinositide phosphatases. J Biol Chem. 1999;274:12990–5. [PubMed]

19. Schmid AC, Wise HM, Mitchell CA, Nussbaum R, Woscholski R. Type II phosphoinositide 5-phosphatases have unique sensitivities towards fatty acid composition and head group phosphorylation. FEBS Lett. 2004;576:9–13. [PubMed]

20. Jefferies HB, Cooke FT, Jat P, Boucheron C, Koizumi T, Hayakawa M, et al. A selective PIKfyve inhibitor blocks PtdIns(3,5)P(2) production and disrupts endomembrane transport and retroviral budding. EMBO Rep. 2008;9:164–70. [PMC free article] [PubMed]

21. Hatanaka H, Iourin O, Rao Z, Fry E, Kingsman A, Stuart DI. Structure of equine infectious anemia virus matrix protein. J Virol. 2002;76:1876–83. [PMC free article] [PubMed]

22. Manders EMM, Verbeek FJ, Aten JA. Measurement of colocalization of objects in dual color confocal images. Journal of Microscopy. 1993;169:375–82.

23. Varnai P, Thyagarajan B, Rohacs T, Balla T. Rapidly inducible changes in phosphatidylinositol 4,5-bisphosphate levels influence multiple regulatory functions of the lipid in intact living cells. J Cell Biol. 2006;175:377–82. [PMC free article] [PubMed]

24. Lawe DC, Chawla A, Merithew E, Dumas J, Carrington W, Fogarty K, et al. Sequential roles for phosphatidylinositol 3-phosphate and Rab5 in tethering and fusion of early endosomes via their interaction with EEA1. J Biol Chem. 2002;277:8611–7. [PubMed]

25. Kobayashi T, Startchev K, Whitney AJ, Gruenberg J. Localization of Lysobisphosphatidic Acid-Rich Membrane Domains in Late Endosomes. Biological Chemistry. [Short Communication] 2001 March;382:3. [PubMed]

26. Chan R, Uchil PD, Jin J, Shui G, Ott DE, Mothes W, et al. Retroviruses human immunodeficiency virus and murine leukemia virus are enriched in phosphoinositides. J Virol. 2008;82:11228–38. [PMC free article] [PubMed]

27. Kisseleva MV, Cao L, Majerus PW. Phosphoinositide-specific inositol polyphosphate 5-phosphatase IV inhibits Akt/protein kinase B phosphorylation and leads to apoptotic cell death. J Biol Chem. 2002;277:6266–72. [PubMed]

28. Brown FD, Rozelle AL, Yin HL, Balla T, Donaldson JG. Phosphatidylinositol 4,5-bisphosphate and Arf6-regulated membrane traffic. J Cell Biol. 2001;154:1007–17. [PMC free article] [PubMed]

29. Aikawa Y, Martin TF. ARF6 regulates a plasma membrane pool of phosphatidylinositol(4,5)bisphosphate required for regulated exocytosis. J Cell Biol. 2003;162:647–59. [PMC free article] [PubMed]

30. Vlahos CJ, Matter WF, Hui KY, Brown RF. A specific inhibitor of phosphatidylinositol 3-kinase, 2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4-one (LY294002) J Biol Chem. 1994;269:5241–8. [PubMed]

31. Chen C, Jin J, Rubin M, Huang L, Sturgeon T, Weixel KM, et al. Association of gag multimers with filamentous actin during equine infectious anemia virus assembly. Curr HIV Res. 2007;5:315–23. [PubMed]

32. Jin J, Sturgeon T, Chen C, Watkins SC, Weisz OA, Montelaro RC. Distinct intracellular trafficking of equine infectious anemia virus and human immunodeficiency virus type 1 Gag during viral assembly and budding revealed by bimolecular fluorescence complementation assays. J Virol. 2007;81:11226–35. [PMC free article] [PubMed]

33. Medina G, Zhang Y, Tang Y, Gottwein E, Vana ML, Bouamr F, et al. The functionally exchangeable L domains in RSV and HIV-1 Gag direct particle release through pathways linked by Tsg101. Traffic. 2005;6:880–94. [PMC free article] [PubMed]

34. Demirov DG, Freed EO. Retrovirus budding. Virus Res. 2004;106:87–102. [PubMed]

35. Shkriabai N, Datta SA, Zhao Z, Hess S, Rein A, Kvaratskhelia M. Interactions of HIV-1 Gag with assembly cofactors. Biochemistry. 2006;45:4077–83. [PubMed]

36. Alfadhli A, Barklis RL, Barklis E. HIV-1 matrix organizes as a hexamer of trimers on membranes containing phosphatidylinositol-(4,5)-bisphosphate. Virology. 2009;387:466–72. [PMC free article] [PubMed]

37. Malecz N, McCabe PC, Spaargaren C, Qiu R, Chuang Y, Symons M. Synaptojanin 2, a novel Rac1 effector that regulates clathrin-mediated endocytosis. Curr Biol. 2000;10:1383–6. [PubMed]

38. Rusk N, Le PU, Mariggio S, Guay G, Lurisci C, Nabi IR, et al. Synaptojanin 2 functions at an early step of clathrin-mediated endocytosis. Curr Biol. 2003;13:659–63. [PubMed]