HIV-1 Nef Impairs Heterotrimeric G-protein Signaling by Targeting Gαi2 for Degradation through Ubiquitination*

Background: HIV Nef disrupts chemotaxis of immune cells by targeting multiple steps. Results: Nef recruits AIP4 E3 ligase, to ubiquitinate Gαi2 of heterotrimeric G-proteins for subsequent lysosomal degradation. Conclusion: Nef impedes chemokine-signaling events via degradation of Gαi2. Significance: Impaired G-protein signaling by HIV Nef-expressing lymphocytes probably contributes to immune dysfunction in AIDS. The HIV Nef protein is an important pathogenic factor that modulates cell surface receptor trafficking and impairs cell motility, presumably by interfering at multiple steps with chemotactic receptor signaling. Here, we report that a dominant effect of Nef is to trigger AIP4 E3 ligase-mediated Gαi2 ubiquitination, which leads to Gαi2 endolysosomal sequestration and destruction. The loss of the Gαi2 subunit was demonstrable in many cell types in the context of gene transfection, HIV infection, or Nef protein transduction. Nef directly interacts with Gαi2 and ternary complexes containing AIP4, Nef, and Gαi2 form. A substantial reversal of Gαi2 loss and a partial recovery of impaired chemotaxis occurred following siRNA knockdown of AIP4 or NEDD4 or by inhibiting dynamin. The N-terminal myristoyl group, 62EEEE65 motif, and 72PXXP75 motif of Nef are critical for this effect to occur. Nef expression does not affect a Gqi5 chimera where the five C-terminal residues of Gq are replaced with those of Gαi2. Lysine at position 296 of Gαi2 was identified as the critical determinant of Nef-induced degradation. By specifically degrading Gαi2, Nef directly subverts leukocyte migration and homing. Impaired trafficking and homing of HIV Nef-expressing lymphocytes probably contributes to early immune dysfunction following HIV infection.

The recirculation of naive and memory lymphocytes, the trafficking of lymphocytes into lymphoid organs, and the appropriate localization of effector cells depend upon a regulated configuration of cell surface adhesion molecules and chemoattractants. Ligand-occupied chemoattractant receptors, trigger G␣ subunits of the heterotrimeric G-protein G i to exchange GTP for GDP. This leads to the release of G␣ i -associated G␤␥ subunits, activation of downstream effectors, and directed cell migration (1). T lymphocytes employ several different chemokine receptors. The chemokine receptors CCR7, CCR9, and CXCR4 (chemokine (CXC motif) receptor 4) along with their cognate ligands regulate T-cell ontogeny in the thymus (2)(3)(4), whereas CCR7 regulates the migration of peripheral T cells to lymphoid tissue during adaptive immune response (5,6). Proinflammatory chemokine receptors like CXCR3 and CCR5 and their respective ligands regulate effector CD4 and CD8 T cell trafficking in peripheral tissues (7)(8)(9)(10). During an adaptive immune response, activated paracortical T cells use CCR7 and CXCR5 to migrate to the follicle edge and eventually into the lymph node follicle, where they help coordinate B lymphocyte responses (11,12). Impaired signaling through these different receptors can cause immune dysfunction and lymphoid tissue disorganization.
A hallmark of HIV infection is severe disruption of the cellular microenvironment of lymph nodes and the gastrointestinal associated lymphoid tissue. Although this occurs in part as a consequence of CD4 ϩ T cell depletion, a major contributing factor is compromised chemoattractant receptor signaling (13,14). Perinatal HIV infection results in profound T cell developmental abnormalities mimicking those observed in thymic dysplasia in DiGeorge syndrome (15). Implicating the HIV Nef protein as a major co-factor in the general pathology and immune dysfunction in HIV AIDS, transgenic mice expressing HIV-1 provirus or HIV Nef protein alone have a similar constellation of immune pathologies (16 -18).
Several mechanisms have been found by which Nef can interfere with chemoattractant receptor signaling. First, HIV and SIV Nef proteins have been shown to down-regulate chemokine receptor expression, presumably through enhanced receptor endocytosis (19,20). This has been associated with a marked chemotactic arrest, presumably due to loss of the cognate receptor (19). Second, Nef can interfere with signal transduction pathways downstream of the receptors by affecting different intracellular signaling proteins involved in these pathways. These include the Ser/Thr kinase PAK; small GTPases, such as Rac and CDC42; and guanine nucleotide exchange factors, such as Vav and DOCK2-ELMO1 (21)(22)(23)(24). In particular, Nef inhibited T lymphocyte chemotaxis toward CXCL12 (chemokine (CXC motif) ligand 12) by disrupting spatially organized cytoskeleton rearrangements through unregulated DOCK2-ELMO1-mediated Rac activation (24). Third, Nef has been linked to decreases in integrin expression, which impaired transmigration across endothelial cells (25). Fourth, Nef expression can alter the morphology of T lymphocytes and dendritic cells through changes in F-actin dynamics (26 -28) by deregulating cofilin through PAK2 interaction (29) and RhoA interaction with diaphanous interacting protein and p190RhoAGAP (30). With the exception of the studies implicating reduced receptor expression, the majority of these studies indicate that the prominent effect of Nef on chemokine and chemoattractant receptor signaling is downstream of the major signal transducer, the heterotrimeric G-protein G i .
Although Nef undoubtedly targets multiple sites in the chemoattractant signaling pathway, we have discovered a dominant effect of Nef because its expression results in a marked loss of steady-state levels of G␣ i2 protein but not of other G␣ or G␤␥ subunits. Nef triggered AIP4 (atrophin-interacting protein 4) E3 ligase-mediated G␣ i2 ubiquitination, leading to its endolysosomal sequestration and destruction. The consequence of the loss of G␣ i2 is a marked impairment in chemoattractant receptor signaling and directed leukocyte migration.

EXPERIMENTAL PROCEDURES
Expression Plasmids-Expression plasmids for CD4, CD8, WT and mutant Nef alleles, Nef Cerulean, and bicistronic IRES 2 plasmids encoding Nef and GFP have been described (31). Chemokine receptors, YFP-tagged G␣ i subunits, and the Gq i5 chimera have been described (32)(33)(34)(35)(36). Mutations replacing each one of the three lysines at positions 296, 307, and 314 of G␣ i2 (but not in G␣ i3 ) with arginines were engineered by PCR mutagenesis of YFP-tagged G␣ i2 (35). The GST-Nef construct was constructed by in-frame cloning of Nef ORF into the EcoRI-SalI sites of pGEX-4T1 (GE Healthcare). Plasmids expressing GST fused to full-length AIP4, AIP4⌬WW, and AIP4-WW-I-IV were generously supplied by Adriano Marchese (Stritch School of Medicine, Loyola University, Chicago) as were the expression plasmids for FLAG-ubiquitin, FLAGtagged WT AIP4, or c-Myc-tagged WT or C830A AIP4 mutant. The GST fusion proteins were expressed in Escherichia coli strain XL-1 Blue (Stratagene) and purified according to the manufacturer's (GE Healthcare) instructions.
Cells and Transfection-For this study, HeLa cells were used for microscopy, and T cell lines of Jurkat or CEM cells were used for phenotypic assays. HeLa cell transfection conditions have been described in detail before (69). The Department of Transfusion Medicine at the National Institutes of Health provided elutriated monocytes and leukocyte-enriched buffy coat from anonymous volunteers. Peripheral blood lymphocytes were purified as before (69), and cells were cultivated under standard conditions in RPMI medium or DMEM with 10% fetal calf serum (FCS) and 1% L-glutamine, as appropriate. Primary hematopoietic cells and cell lines in suspension (7-10 ϫ 10 6 cells/100 l) were nucleofected with 3-5 g of DNA using an Amaxa Nucleofector as recommended by the manufacturer. In all cases, GFP or CD8 expression plasmid was included to check transfection efficiency as described (31). Cell viability was checked using a FACS-based live/killed assay kit using a Guava Easycyte flow cytometer as described by the manufacturer (Millipore Corp.). Experiments were rejected if cell viability was less than 75%. For biochemical experiments, cell numbers were adjusted to equivalent viability and transfection efficiency. However, for most non-FACS-based experiments, transfectants were purified by positive selection using CD8 immunoaffinity beads from Stem Cell Technologies as recommended by the manufacturer.
HIV Infection-For single-cycle HIV infection, recombinant viruses expressing mouse CD24 in place of vpR (CD24), CD24 with a vpU deletion (CD24U), or CD24 lacking both vpU and Nef (CD24UM1T) were generated by 293-T cell transfection (Trans-IT LT1 transfection reagent, Mirus Corp.) with the respective HIV proviruses and a VSV-G expression plasmid. Infection was allowed to proceed for 48 -72 h to achieve Ͼ40% infection. Virus production, quantification, and T cell infection kinetics have been described before (33).
RNA-mediated Interference-G␣ i1 , G␣ i2 , G␣ i3 , and G␣ o subunits were knocked down using the recommended sequence (37). AIP4 and NEDD4 were knocked down using UUUC-AAUGCAGAAUUUCUGUGGUCC and UAGAGGAGAA-GGUUCUUGUUGUUGC, respectively (Invitrogen). Protocols for siRNA transfection followed by plasmid expression have been described before (31 ative ratio of fluorescence emitted at 510 nm after sequential excitation at 340 and 380 nm was measured as described (33), using Fura-2 AM (Invitrogen) as the calcium-binding dye in a fluorimeter (Photon Technology Inc., South Brunswick, NJ). Alternatively, intracellular Ca 2ϩ levels were measured using a FLIPR 3 calcium assay kit (Molecular Devices) with a Flex-Station III scanning fluorimeter as described (38), using Soft-max Pro (Molecular Devices) for data acquisition and analysis. Single cell analysis of calcium flux was performed using the Leica SP5, using a ϫ40 oil immersion objective, numerical aperture 1.25. HeLa cells in 8-well chambers (LabTek chambers, Thermo Scientific) were cotransfected with CerFP or Nef-CerFP and YFP-tagged WT G␣ i2 or G␣ i3 . At 18 -24 h post-transfection, cells were loaded with Hanks' balanced saline solution containing Ca 2ϩ and Mg 2ϩ (Invitrogen) with 1% FBS and 4 M Fluo-4 AM (Invitrogen). The cells were irradiated with a UV laser at 405 nm for Cerulean and with an argon laser at 488 nm for Fluo-4 and at 514 nm for YFP. Images were acquired in a 512 ϫ 400-pixel format with the pinhole set at 2 airy units every 1-1.4 s for ϳ200 s. The images were collected as line scan images, which provide the values of the calcium-dependent increase in fluorescence along a single spatial dimension as a function of time. For analysis, the regions of interest (ROIs) were drawn on the cells expressing Cerulean/Nef Cerulean. At least 10 ROIs were drawn for each field, and the calcium flux was measured in Ͼ5 fields for each experimental condition. The change in the intensities was analyzed using the Leica software, followed by graphing using EXCEL. Affinity Selections-GST pull-downs of cellular extracts (10 7 cell equivalents in 1 ml) in 25 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EGTA, 0.5 mM MgCl 2 , and 0.5% (w/v) Triton X-100 (lysis buffer) with protease inhibitors (1 Complete tablet/ml; Roche Applied Science) was done using purified GST or the indicated GST fusion proteins immobilized on glutathione-agarose beads (GE Healthcare) and incubated for 2 h. The beads were washed three times with lysis buffer and once with lysis buffer without Triton X-100. Bound proteins were resolved by SDS-PAGE and detected by immunoblotting using antibodies against the respective targets. His 6tagged Nef protein was affinity-selected from extracts of corresponding transfectants using Ni 2ϩ -nitrilotriacetic acid-agarose beads (Qiagen Corp.), and FLAG-tagged proteins were selected using agarose bead immobilized M2 mAb (Sigma-Aldrich).
cAMP Assay-Intracellular cAMP levels were assayed using the Direct cAMP Enzyme Immunoassay kit (Assay Designs), following the manufacturer's instructions. The optical density was read at 405 nm in a FlexStation II microplate scanner, and the results were analyzed using Softmax Pro 5 (Molecular Devices, Sunnyvale, CA).
Flow Cytometry-The fluorescent dye-conjugated antibodies for CD4, CD8, and the chemokine receptors were obtained from BD Biosciences, R&D Systems, or eBioscience. One million cells were incubated with the appropriate antibodies in 100 l of PBS containing 1% goat serum for 15 min at room temperature. The data acquisition was carried out using a FACSort TM (BD Biosciences) flow cytometer, and the analyses were done using FlowJo version 9.2. (Tree Star Inc., Ashland, OR). Flow cytometric detection of F-actin by phalloidin staining has been described (75).
Confocal Microscopy-Confocal microscopy was done as described (32) using a Leica TCS-NT/SP5 microscope (Leica, Exton, PA) equipped with a 100ϫ oil immersion objective, numerical aperture 1.32. Images were collected using digital zooms up to ϫ8. Co-localization experiments were done with live (fluorescent proteins) or fixed cells for most experiments as described (31). The fluorochromes were excited using a UV laser at 405 nm for Cerulean, an argon laser at 488 nm for Alexa 488 and YFP, and a DPSS laser at 561 nm for Alexa 568 and 647. Images were processed using the Leica TCS-NT/SP5 software, Leica Lite, and Adobe Photoshop CS4.
FRET Assay-FRET interactions between Nef-CerFP and YFP-tagged G␣ i2 or G␣ i3 were evaluated using the Leica acceptor photobleaching software to examine in live or fixed cells in the Leica SP5 confocal microscope with a ϫ60, numerical aperture 1.4 oil immersion objective in a 512 ϫ 512-pixel format using a DPSS laser at 442 nm for Cerulean and an argon laser at 514 nm for YFP. The exposure times were kept equal within each series of images and chosen such that all pixel intensities were within the linear range. A ϫ63 oil lens was used unless mentioned otherwise. ROIs were selected from individual fields visualized at zoom factors between 1 and 8 (Z1-Z8), and 50 -80% YFP photobleaching was achieved in the selected ROIs using the 514-nm laser at 100% power for 45 s at ϫ8 optical zoom. Each pixel of the image contains data corresponding to an emission spectrum resulting from both CFP and YFP. The fluorescence intensities of donor and acceptor were measured before and after photobleaching in the ROIs drawn at different cellular organelles and membranes. The diameter of the ROIs was kept uniform throughout the analysis at 8 m. 10 ROIs were drawn for each part of a single cell, and at least 10 different cells were taken for each FRET efficiency calculation. Only regions photobleached at Ͼ50% were considered for analysis. FRET efficiency was calculated using the equation E F ϭ (I postbleach Ϫ I prebleach )/I postbleach , where I is the average CFP fluorescence intensity after the subtraction of the background.
Chemotaxis Assays-End-point chemotaxis was determined using the transwell system with membranes of 6.5-mm diameter and 5.0-m pore size in RPMI containing 10 mM HEPES and 1% FBS as described (32). The ratios of migrated cells were determined from the number of cells in the lower and upper FIGURE 2. Nef markedly inhibited biochemical readouts of G␣ i2 activation. Nef inhibits chemoattractant-mediated calcium flux in Jurkat cells (A1) and the U937 cell line (A2). Cells were co-transfected with CD8 and Nef, co-expressers were purified by CD8-positive selection (Stem Cell Technologies), and Nef effect in Jurkat cells was evaluated by measuring cell surface CD4 expression. The time course of CXCL12 (10 nM)-initiated calcium flux profiles was obtained using FlexStation III and the recommended assay. Results are representative of four experiments. Agonist (CXCL12 followed by CCL2) dose (10 or 100 nM)-response profiles of intracellular calcium flux in U937 transfectants were analyzed by fluorescence ratiometry in a PTI fluorimeter (33). B1, time course of CXCL12-initiated Ca 2ϩ flux in HeLa cells expressing CerFP or Nef-CerFP (red) was monitored by video capture at 30 frames/s of Fluo-4 emission (green) up to 150 s after CXCL12 addition. The arrows denote CerFP or Nef-CerFP cells (expressed at ϳ10 -15% efficiency around 12-16 h post-transfection) to highlight their difference in Fluo-4 intensity. Fluorescent data were collected from ϳ10 ROIs for each field, the calcium flux was measured in Ͼ5 fields for each condition in an experiment, and the experiments were repeated three times (n ϭ 50 -60). The change in the intensities was analyzed using the Leica software followed by graphing using EXCEL. Ca 2ϩ flux profiles of a few (to avoid clutter) representative cells (ROIs) expressing CerFP (left) or Nef-CerFP (right) are shown in B2, with the ordinate showing relative intensity of Fluo-4 emission. CXCL12-initiated Ca 2ϩ flux is profiled in purified CEM cells co-transfected with CD8 and WT, null, or other Nef mutants. CEM cells were transfected with CD8, WT, null, or the indicated Nef mutants, and CD8(ϩ) cells were purified prior to measurement of Ca 2ϩ flux (as described above) (B3). Nef expression enhanced cAMP levels under basal conditions or after G␣ s activation by isoproterenol (C) or by forskolin stimulation of adenylyl cyclase (D). However, Nef did not further enhance the cAMP levels after isoproterenol treatment with transfectants overexpressing G␣ i3 (E). Jurkat cells were cotransfected with CD8 and Nef or vector (for C-E) and with a G␣ i3 expression plasmid (only for E). Transfected cells were purified by CD8 selection and assayed for cAMP production as described under "Experimental Procedures" (n ϭ 4; error bars represent S.E.; *, p Ͻ 0.05). NOVEMBER 30, 2012 • VOLUME 287 • NUMBER 49 chambers counted in a cell sorter after the addition of a known number of fluorescent reference particles (Spherotech, Inc., Libertyville, IL) (38). Flow cytometric detection of F-actin by phalloidin staining has been described (38).

HIV-1 Nef Induces G␣ i2 Ubiquitination and Degradation
Statistical Analyses-The statistical analyses were performed using GraphPad Prism version 5.0. A paired t test or one-way analysis of variance test was performed as appropriate, to determine the significance of the observed differences between the paired or unpaired samples. A value of p Ͻ 0.05 was considered to be significant in all of the analyses. The graphs were generated using either Microsoft Excel or GraphPad Prism, and the bars represent mean Ϯ S.E.

Nef-induced Chemotaxis Defect in T Cells and Monocytes-
Nef has been shown to impair chemotaxis toward CXCL12 in lymphoid cell lines, presumably through mechanisms other than depletion of cell surface CXCR4 (24,25,29,39,40). To investigate how Nef subverts chemotactic responses downstream of chemokine receptors, we selected cells for study in which Nef only modestly (Ͻ50%) reduced receptor expression. These included the T cell lines, CEM and Jurkat, human peripheral blood mononuclear cells (PBMCs), and human monocytes. Despite the relatively modest reduction in receptor levels (data not shown), Nef expression still markedly inhibited CXCL12-, CCL19-, CCL2-, or formylmethionylleucylphenylalanine-induced chemotaxis (Fig. 1, A-C).

Nef Markedly Inhibited Biochemical Readouts of Early G-protein Signaling from Agonist-stimulated Chemokine
Receptors-Most, if not all, chemokines activate G␣ i -coupled GPCRs. An agonist-bound GPCR exchanges GTP for GDP in the G␣ subunit of receptor-bound heterotrimeric G-protein. This results in G␣ and G␤␥ dissociation and the activation of downstream effectors. First, we examined the immediate consequence of G␤␥ dissociation by monitoring intracellular Ca 2ϩ flux from ER stores mediated by inositol 1,4,5-trisphosphate in response to activation of phospholipase C. We found that Nef expression markedly attenuated the rapid chemokine-induced mobilization of intracellular Ca 2ϩ normally observed in Jurkat cells and abolished the response in U937 cells (Fig. 2, A1 and A2). We also visualized single cell Ca 2ϩ flux in response to CXCL12 stimulation of HeLa cells expressing either Cerulean fluorescent protein (CerFP) or Nef-CerFP by real-time time lapse confocal video microscopy. CXCL12 increased intracellular Ca 2ϩ in the CerFP-expressing cells, but not the Nef-CerFP cells (Fig. 2, B1). This was confirmed by quantitative kinetic analysis of Ca 2ϩ flux of individual cells expressing CerFP or Nef-CerFP (Fig. 2, B2). These data suggested that Nef might perturb chemokine-mediated G-protein dissociation from cognate GPCR(s). In agreement with the corresponding effects on chemotaxis, alanine substitutions of four glutamates at 62 and at the PXXP motif partially reversed the Nef-induced inhibition of CXCL12-mediated calcium release from the ER (Fig. 2, B3).
Next, we inquired whether Nef subverted activation of G-protein ␣ subunit(s). The major effectors of activated G␣ i subunits are adenylyl cyclase isoforms. GTP-bound G␣ i directly inhibits specific adenylyl cyclase isoforms, thereby reducing cAMP production by direct (by agents like forskolin) or hormone-induced (via activated G␣ s subunit) activation of adenylyl cyclase. Basal intracellular cAMP levels reflect a steady-state equilibrium of tonic inhibition and stimulation of adenylyl cyclase isoforms by activated G␣ i and G␣ s subunits, respectively. We found that Nef-expressing Jurkat cells had a 2-fold higher basal cAMP level than did the null expression controls, consistent with a loss of constitutive G␣ i signaling. Furthermore, Nef expression was associated with enhanced cAMP production in Jurkat cells treated with the ␤-adrenergic agonist, isoproterenol (ISOPRO), or forskolin, a direct adenylyl cyclase activator (Fig. 2, C and D). However, basal or hormone-induced cAMP levels were not enhanced in Nef(ϩ) cells co-expressing recombinant G␣ i3 (Fig. 2E). This suggested that Nef might solely target G␣ i2 because Jurkat cells and lymphocytes express G␣ i2 and G␣ i3 , not G␣ i1 .
Nef Induced a Marked Loss of Steady-state Levels of G␣ i2 but Not G␣ i3 or Other G␣ Subunits-To determine if Nef impaired G-protein functionality or induced a physical loss or sequestration of one or more G-protein subunits, we quantified the steady-state levels of G␣ subunits in Nef(ϩ) cells by immunoblotting. We found that Nef induced a dose-dependent decrease in G␣ i2 levels in Jurkat/CEM cells and monocytes, comparable in magnitude with CD4 down-regulation in Jurkat or CEM cells (Fig. 3A). In contrast, G␣ i3 in Jurkat cells, CEM cells (not shown), and monocytes or G␣ q and G␣ s in Jurkat cells were unaffected (Fig. 3A).
In single cycle HIV-1 infections, loss of G␣ i2 correlated with Nef expression. G␣ i2 levels were unaltered in cells infected with viruses deleted for Nef or Nef and vpU (Fig. 3B). Eight different HIV-1 Nef alleles induced a specific loss of G␣ i2 of comparable magnitude (Fig. 3C). The differences in the level of down-regulation might be dependent on critical Nef residues, and they are a subject of further investigation. G␣ i2 levels were unaffected in cells expressing Nef mutants with alanine substitutions in the acidic motif at 62 (E62A) or in the polyproline motif Protein bands were scanned for pixel density, and results are plotted as histograms with error bars representing S.E. (n ϭ 3; *, p Ͻ 0.03). B, loss of G␣ i2 in single cycle infections of CEM/Jurkat cells with identical reverse transcriptase unit equivalents of VSV-G pseudotyped Nef ϩ (wt), but not Nef Ϫ (M1T) or Nef Ϫ vpU Ϫ (⌬vpU M1T) NL4-3 HIVs expressing murine CD4 antigen in place of vpR. G␣ i2 , G␣ i3 , and Nef were detected by immunoblotting cellular extracts. C, HIV-1 Nef alleles induced a specific loss of G␣ i2 of comparable magnitude. Jurkat cells were transfected with various Nef alleles, and the levels of G␣ i2 down-regulation were assessed. Cellular extracts were immunoblotted for G␣ i2 , G␣ i3 , HA (Nef), or actin. D, certain Nef mutants had lost the ability to induce loss of G␣ i2 . Extracts of CEM/Jurkat cells transfected with the indicated Nef derivatives were analyzed by SDS-PAGE followed by immunoblotting for actin, G␣ i2 , and G␣ i3 . Relative pixel values in B and C represent averages from two experiments for each case. E, knockdown of G␣ i2 or G␣ i2 by cognate siRNAs partially inhibited CXCL12dependent calcium flux. Jurkat cells were co-transfected with CD8 and Nef or an empty vector following siRNA nucleofection and expression for 48 h. ϳ2 ϫ 10 2 transfectants were adjusted to reflect constant levels of CD8 and analyzed for CXCL12-initiated intracellular calcium flux. G␣ i2 and G␣ i2 in the respective siRNA transfectants were detected by immunoblotting as described above. NOVEMBER 30, 2012 • VOLUME 287 • NUMBER 49 at 72 (P72A) in Jurkat and CEM cells (Fig. 3D). We compared the effects of siRNA knockdown of individual G␣ i2 , G␣ i3 , or G o subunits on agonist-driven Ca 2ϩ flux. Knockdown of G␣ i2 or G␣ i3 resulted in a significant (ϳ60%) decrease in the agonist-mediated Ca 2ϩ flux (Fig. 3E). Thus, the Nef-induced defect in early G-protein signaling reflected by chemokinemediated Ca 2ϩ flux was attributed to the physical loss of the G␣ i2 subunit. . Nef-induced inhibition of G-protein signaling and G␣ i2 degradation were partially rectified by overexpression of recombinant G␣ i2 , and a Gq i5 chimera replacing the C-terminal five residues with those of G␣ I was resistant to Nef effect. CEM cells were nucleofected with CD8 and YFP-tagged G␣ i2 (A) or G␣ i3 (B) subunits with or without a molar excess of Nef. Purified CD8 transfectants were analyzed for CXCL12-driven intracellular Ca 2ϩ flux. Nef-induced loss of Ca 2ϩ flux in response to CXCL12 was partially reversed in cells expressing a 2-fold molar excess of G␣ i2 over Nef (A), but G␣ i3 co-expression did not ameliorate Ca 2ϩ flux deficit (B). At a 2-fold molar excess of G␣ i2 over Nef, there was much less relative loss in the steady-state levels of G␣ i2 -YFP (A, bottom). There was no loss of G␣ i3 -YFP with or without of Nef (B, bottom). Relative G␣ i2 -YFP pixel values shown in the immunoblot are averages from three experiments. C, Nef did not inhibit CXCL12-initiated calcium flux in cells expressing Gq i5 chimera replacing the C-terminal five residues with those of G␣ i2 , (36). Jurkat transfectants co-expressing CD8 and Gq i5 chimera with or without Nef were purified and analyzed as described above. D, Nef did not alter the stability of co-expressed Gq i5 chimera. Cell extracts of Jurkat transfectants were analyzed for expression of HA-tagged Gq i5 , co-expressed CD8, Nef, and actin. E, Gq i5 expression did not alter Nef-induced down-regulation of CD4 or CXCR4 at the PM. Black and gray histograms (with error bars (S.E.)) represent normalized data for CD8(Ϫ) and CD8(ϩ) gated cells (n ϭ 3; *, p Ͻ 0.05). IB, immunoblot.

Nef-induced Inhibition of G␣ i2 Degradation Was Partially Rectified by Overexpression of Recombinant G␣ i2 , and a Gq i5 Chimera Replacing the C-terminal Five Residues with Those of G␣ I Was
Resistant to the Nef Effect-Co-expression of YFP-tagged recombinant G␣ i1 (not shown), G␣ i2 , or G␣ i3 failed to restore CXCL12mediated Ca 2ϩ flux in Jurkat cells expressing a 2-fold molar excess of Nef over the recombinant G␣ i2 subunits (Fig. 4, A and B). The failure to reverse CXCL12-mediated Ca 2ϩ flux by recombinant G␣ i2 reflected the destruction of endogenous and a substantial loss of the co-expressed G␣ i2 (Fig. 4A, bottom). Only by increasing the ratio of G␣ i2 to Nef expression could the signaling be partially restored. Although there was no loss of G␣ i3 -YFP co-expressed with a molar excess of Nef (Fig. 4B, bottom), G␣ i3 -YFP failed to restore CXCL12-initiated Ca 2ϩ flux (Fig. 4B, top).
A small C-terminal region of G␣ subunit(s) is the GPCR selectivity determinant and a recombinant Gq i5 , which exchanged five Gq residues at the C terminus for those of G␣ i2 , thus becoming functionally equivalent to G␣ i2 in chemokine responsiveness and pertussis toxin sensitivity (36). Nef failed to alter the expression of or inhibit CXCL12-induced intracellular Ca 2ϩ flux from the Gq i5 chimera (Fig. 4, C and D) while still down-regulating CD4 and CXCR4 (Fig. 4E). Because heterotrimeric G-proteins are bound to the receptor in an inactive state, it is possible that the Nef-induced lysosomal sequestration of G␣ i2 reflects aberrant trafficking of CXCR4. However, this is unlikely because cells co-expressing Nef and Gq i5 chimera retained CXCL12-driven Ca 2ϩ flux.
Nef Co-localized with and Promoted Sequestration of G␣ i2 , but Not G␣ i3 , in the Perinuclear Region Enriched for Endolysosomal Markers-To determine whether Nef interacted with G␣ i2 in vivo, we examined the subcellular distribution of YFP-tagged recombinant G␣ i2 or G␣ i3 in HeLa cells co-expressing CerFP or Nef-CerFP. In the CerFP-expressing cells, G␣ i2 and G␣ i3 were distributed at the PM irrespective of CXCL12 exposure. In Nef-expressing cells, a substantial fraction of G␣ i2 co-localized with Nef in the perinuclear region (co-localization correlation of Ͼ0.75), regardless of CXCL12 treatment, whereas G␣ i3 did not co-localize FIGURE 5. Nef co-localized with and promoted endolysosomal sequestration of G␣ i2 but not G␣ i3 . HeLa cells in coverglass chambers were co-transfected with YFP-tagged G␣ i2 (A1) or G␣ i3 (A2) with CerFP or Nef-CerFP (top or bottom row in A1 and A2) and treated for 10 min at 37°C with increasing concentrations of CXCL12 or not before processing for live microscopy. Cerulean is green, and YFP is red. A3, Nef-induced loss of YFP-tagged G␣ i2 occurred irrespective of whether or not the G-protein was membrane-associated. Cells were extracted in a buffer containing 1% (w/v) Triton X-100 (TXT-100) at 25°C (room temperature (RT)), conditions that are known to strip most if not all membrane-associated proteins. Membrane (pellet) and cytoplasmic (supernatant) extracts were resolved by SDS-PAGE followed by immunoblot detection of YFP, Nef, and CD8. Relative G␣ i2 -YFP pixel values are averages from three experiments. B, in Nef-expressing cells, YFP-tagged G␣ i2 (red) co-localized with Nef-CerFP (green) and endolysosomal markers, EEA1 and LAMP (blue) proteins, but not with markers (blue) for ER (GRP) or Golgi (GOLGIN). FIGURE 6. Nef associated with G␣ i2 but not G␣ i3 in vivo. Acceptor photobleaching FRET assay of HeLa cells co-expressing CerFP or Nef-CerFP with YFPtagged G␣ i2 or G␣ i3 . FRET assay was limited to G␣ i2 or G␣ i3 with CerFP or Nef-CerFP in the perinuclear area (A1) or at the PM (A2). Photomicrographs are representative of 10 fields (cells). Fluorescence intensities of 10 ROIs (arrows) corresponding to G␣ i2 or G␣ i3 in the perinuclear regions (B1) or at the PM (B2) in each cell were examined before (Pre) and after (Post) photobleaching, and calculated FRET efficiencies for all ROIs (ϳ100) from three independent experiments are shown as scatter plots with error bars. Donor emission of representative ROIs pre-and postphotobleaching is shown in the inset above the graph. The calculated statistical significance is represented (*, p Ͻ 01).
with Nef (correlation of Ͻ0.04) (Fig. 5, A1 and A2). Nef expression resulted in an equivalent loss of membrane-associated versus detergent-soluble fractions of G␣ i2 -YFP, indicating that Nef induced collective loss of G␣ i2 inside the cell (Fig. 5, A3). The G␣ i2and Nef-containing vesicular structures stained positively for endolysosomal markers EEA1 and LAMPs but not with the markers for ER (GRP) or Golgi (Golgin) (Fig. 5B).
Nef-mediated Lysosomal Sequestration of G␣ i2 Reflected in Vivo Interaction of Nef and G␣ i2 -Next, we performed live FRET microscopy to analyze nearest neighbor interactions between Nef and G␣ i . We measured FRET efficiency on the basis of the increase in the donor (CFP or Nef-CFP) fluorescence upon photobleaching the YFP-tagged G␣ i2 acceptor. FRET efficiencies generated by tandem CFP-YFP fusion protein served as the positive control. We analyzed Ͼ10 cells in each of three experiments, and photomicrographs of one representative cell are shown (Fig. 6, A1), with the calculated FRET efficiencies of ϳ100 ROIs shown in the graphs below (Fig. 6, A2). There was substantial FRET between Nef-CFP and G␣ i2 -YFP at the intracellular, perinuclear region (Fig. 6, B1) but not at the PM (Fig. 6, B2) or anywhere in the cell with G␣ i3 -YFP (Fig. 6, B1 and B2).
Nef Interacted with G␣ i2 in Vitro and Induced G␣ i2 Ubiquitination-To confirm the physical interaction between Nef and G␣ i2 , we performed in vitro GST pull-down assays of leukocyte extracts using GST or GST-Nef. Using extracts from Jurkat T-cells and primary monocytes, we found that GST-Nef bound G␣ i2 but not G␣ i3 in a dose-responsive manner (Fig. 7A).
To inquire whether the loss of G␣ i2 was a consequence of Nef-induced ubiquitination of G␣ i2 and its subsequent lysosomal proteolysis, we directly evaluated whether G␣ i2 was ubiquitinated. We co-transfected CEM cells with vectors expressing FLAG-tagged ubiquitin and either Nef or control. Nef expression resulted in significant loss of G␣ i2 (Fig. 7, B1 (left)). Cellular extracts were immunoprecipitated with FLAG mAb-coated beads, and the selected proteins were immunoblotted for G␣ i2 using mouse monoclonal antibody (Fig. 7, B1 (right)) or rabbit polyclonal antibody (Fig. 7, B2). Nef expression resulted in significant loss of G␣ i2 (Fig. 7B, left). Two immunoreactive bands of ϳ53 and ϳ63 kDa were seen in lysates of Nef-expressing cells, probably representing monoand diubiquitinated native G␣ i2 (Fig. 7B, right).
We first evaluated the effects of Dynasore (41), a small molecular weight dynamin inhibitor, and the proteosome inhibitor, epoxomycin. We found that pretreatment with the inhibitors substantially reversed the Nef-induced loss of G␣ i2 (Fig. 7C).
Nef Co-localizes with E3 Ubiquitin Ligases AIP4 and Nedd4-Ubiquitination of agonist-occupied CXCR4 by the Nedd4-like E3 ubiquitin ligase AIP4 (42) primes it for traffic through the ESCRT pathway, ultimately resulting in lysosomal proteolysis (43,44). We inquired whether Nef-induced endolysosomal trafficking of G␣ i2 was facilitated by these E3 ligase(s). We stained HeLa cells co-transfected with plasmids expressing YFP-tagged G␣ i2 or G␣ i3 and Nef-CerFP or CerFP with antibodies against AIP4 or NEDD4. There was extensive co-localization of G␣ i2 with Nef and AIP4 in the perinuclear area (Fig.  8A). There was a similar co-distribution of Nef and G␣ i2 with the NEDD4-positive region (Fig. 8B). In contrast, G␣ i3 was mostly at the PM, with no evidence of co-localization with Nef, AIP4, or NEDD4 (Fig. 8, A and B (bottom)).
AIP4 Bound G␣ i2 through the HECT Domain of the E3 Ligase and to the Tetraglutamate and Polyproline Motif of Nef, Presumably through the WW Domain(s)-We evaluated the binding potential of AIP4 for Nef and G␣ i subunits in vitro and in vivo. In GST-Nef pull-down assays, AIP4 was recovered from Jurkat cell extracts in a specific and quantifiable manner (Fig.  9A, top). Like other members of the NEDD4 family of HECT domain E3 ligases, AIP4 has four WW domains (Fig. 9A, center), which preferentially bind to proline-rich PY motifs (45)(46)(47) or hyperphosphorylated C termini of PM receptors (44). Accordingly, GST-AIP4 bound poorly to Nef mutants with sub-FIGURE 7. Nef interacted with G␣ i2 in vitro and induced G␣ i2 ubiquitination. A, varying amounts of purified GST or GST Nef were incubated with cellular extracts of Jurkat cells or human monocytes. Bound proteins were resolved by SDS-PAGE, and G␣ i species were detected by immunoblotting. B, Nef-induced ubiquitination of native G␣ i2 in Nef-expressing CEM cells. CEM cells were co-transfected with FLAG-tagged ubiquitin and Nef or null plasmid. Cellular extracts were directly immunoblotted for G␣ i2 (B1 and B2, left panels) or immunoprecipitated first with anti-FLAG polyclonal antibody (B1) or mAb (B2) followed by immunoblotting for G␣ i2 . Asterisks denote G␣ i2 (left) and mono-and diubiquitinated G␣ i2 (right). Numbers (kDa) refer to molecular mass markers. Data are representative of three experiments. C, Nef-mediated G␣ i2 breakdown was partially rectified by Dynasore, a small molecular weight inhibitor of dynamin, or the proteosome inhibitor epoxomycin. Cellular extracts were analyzed by immunoblotting for G␣ i2 . G␣ i2 bands are shown pairwise for Nef(ϩ) versus Nef(Ϫ) for each treatment with relative (%) density (average values from three experiments) denoted below. NOVEMBER 30, 2012 • VOLUME 287 • NUMBER 49 stitutions of the polyglutamate at position 62 (E62A) or the polyproline tract at position 72 (P72A) (Fig. 9A, bottom left). GST fusion proteins containing WT AIP4 or a deletion mutant that excised all four WW domains but retained the HECT domain bound G␣ i2 (Fig. 9A, bottom right).

HIV-1 Nef Induces G␣ i2 Ubiquitination and Degradation
Next, we co-transfected Jurkat cells with expression vectors for FLAG-AIP4 and Nef-HA or a control plasmid. Immunoblotting FLAG immunoprecipitates prepared from lysates from co-expressing cells for Nef-HA revealed an interaction between them (Fig. 9B, top). Furthermore, mutations at the Nef motifs judged to be critical for in vitro binding to AIP4 were also impaired for in vivo binding (Fig. 9, B (bottom) and C).
From these observations collectively, we concluded that Nef facilitated G␣ i2 ubiquitination by recruitment of the NEDD4 class of HECT domain E3 ligases, presumably as a ternary complex of Nef, G␣ i2 , and E3 ligases(s).
Nef-induced Loss of G␣ i2 Was Partially Reversed by Expression of Enzymatically Defective AIP4 or by siRNA Knockdown of AIP4 or NEDD4-To firmly establish the role of AIP4 in the Nef-induced G␣ i2 ubiquitination and degradation, Jurkat cells were co-transfected with Nef or an empty vector and WT or an enzymatically defective HECT domain mutant, AIP4-C830A (43,48). AIP4-C830A overexpression reversed the Nef-mediated loss of G␣ i2 (Fig. 10A).
We then compared the effect of siRNA knockdown of AIP4 or NEDD4 on Nef-induced loss of endogenous G␣ i2 in Jurkat and CEM cells. siRNA-induced reduction of AIP4 or NEDD4 E3 ligase reversed the Nef-induced G␣ i2 loss (Fig.  10B). Simultaneously, we evaluated chemotaxis of Nef versus non-Nef transfectants after siRNA knockdown of AIP4 or NEDD4. Nef-mediated chemotactic inhibition was partially rectified by NEDD4 knockdown, although not nearly as well as by the AIP4 knockdown (Fig. 10C). Silencing the E3 ligases did not affect cell viability or curtail other activities of Nef, such as CD4 (Fig. 10D) down-regulation.
Lysine at Position 296 Is the Critical Determinant of G␣ i2 Ubiquitination and Degradation in Nef-expressing Cells-Finally, we identified the potential lysine residue in G␣ i2 that may be targeted for ubiquitination by AIP4 in Nef-expressing cells, FIGURE 8. Nef recruits HECT domain E3 ligases, AIP4, or NEDD4 and facilitates G␣ i2 ubiquitination, presumably as a ternary complex. A, Nef co-localized with G␣ i2 but not with G␣ i3 in the AIP4-enriched perinuclear region. HeLa cells co-transfected with Nef-CerFP and YFP-tagged G␣ i2 or G␣ i3 were fixed, permeabilized, and stained with murine mAb against AIP4 followed by Alexa 647-conjugated anti-mouse IgG. B, Nef co-localizes with G␣ i2 but not G␣ i3 in NEDD4-positive regions in HeLa cells. Experimental details are as above except for the use of rabbit polyclonal antibody against NEDD4. Monochromatic images on the right in each row correspond to Nef-CerFP, YFP-tagged G␣ i2 or G␣ i3 , and anti-AIP4 or -NEDD4 fluorescence with two-channel (like AIP4/NEDD4 and G␣ i2 or AIP4/NEDD4 and Nef, etc.) composite images shown on the left. 4ϫ cropped RGB images (Nef-CerFP (R), G␣ i2 -or G␣ i3 -YFP (G), and AIP4/NEDD4 (B)) with arrows denoting vesicles showing maximal co-localization are shown on the far left in each row.
by evaluating the susceptibility to Nef of individual G␣ i2 -YFP mutants that exchanged unique lysines for arginines at positions 296, 307, and 314. Jurkat cells were co-transfected with wild type or the respective LYS/ARG mutants with or without Nef. Immunoblotting for YFP showed that the LYS/ARG mutant, M1, was resistant to Nef-induced steady-state loss (Fig. 11A). In agreement, CXCL12-driven Ca 2ϩ flux in M1 mutant-expressing cells was relatively more resistant than in the WT G␣ i2 -YFP expressers (Fig. 11B). We then examined the subcellular distribution of WT G␣ i2 -YFP versus the M1 mutant in HeLa cells co-expressing CerFP or Nef-CerFP. In the CerFP-expressing cells, WT and M1 were distributed at the PM. In Nef-expressing cells, a substantial fraction of WT but not M1 co-localized with Nef in the perinuclear region (co-localization correlation of Ͼ0.75 versus Ͻ0.125) (Fig.  11C).

DISCUSSION
Through multiple criteria, we have shown here that HIV Nef impairs G-protein signaling from chemokine receptors by selectively targeting G␣ i2 for ubiquitination and rapid endolysosomal destruction. Nef-induced loss of steady-state levels of G␣ i2 was observed in many cell types in the context of gene transfection, HIV infection, or Nef protein transduction. Lys/ Arg substitution at each of the three unique lysines near the C terminus of G␣ i2 identified lysine at 296 as the crucial determinant of Nef-induced degradation. Nef-mediated loss of G␣ i2 was shown to be dependent on ubiquitination by AIP4, because it was reversed by overexpression of catalytically defective C830A AIP4 mutant, by siRNA knockdown of AIP4, or by use of proteosomal inhibitors or a dynamin antagonist. Previous reports have shown that, following agonist activation, CXCR4 is ubiquitinated by the NEDD4 class HECT domain E3 ubiqui- FIGURE 9. Biochemical and genetic analysis of Nef interaction in vivo with G␣ i2 and E3 ligases. A, Nef and G␣ i2 interacted with AIP4 in vitro. Shown is a schematic illustration (top) of GST-tagged WT AIP4 and mutants deleted for the WW or HECT domains (44). Jurkat cells were transfected with WT, M20A, E62A, P72A, or L164A Nef mutant, and cellular extracts were incubated with ϳ22 pmol of GST or GST-AIP4 immobilized on agarose beads. GST-bound fractions and unselected lysate (2%) were analyzed for Nef by immunoblotting (left). Cellular extracts of Jurkat cells were reacted with equimolar amounts (ϳ22 pM) of GST-AIP4, GST-AIP4 ⌬WW, GST-WW-I-IV, or GST alone. Bound fractions were analyzed by immunoblotting with anti-G␣ i2 mAb (right). Input, an aliquot of extract immunoblotted without selection. Numbers in both cases refer to bound fraction (%) averaged from two experiments. B, in vivo interaction of Nef with AIP4. Extracts of Jurkat cells cotransfected with FLAG-AIP4 and HA-tagged Nef or empty HA vector and immunoprecipitated with FLAG-mAb and Nef were identified by immunoblotting with rabbit anti-HA (top). Relative binding affinity of Nef mutants for AIP4 was evaluated in Jurkat cells cotransfected with GFP, FLAG-AIP4, and His 6 -tagged WT or mutant Nef (bottom). Cellular extracts were bound to Ni 2ϩ -nitrilotriacetic acid beads, and the bound FLAG-AIP4 was detected by immunoblotting with anti-FLAG mAb. C, alternatively, Jurkat cells were cotransfected with GFP, FLAG-AIP4, and HA-tagged WT or mutant Nef. Extracts were immunoprecipitated with FLAG mAb, followed by immunoblotting with HA antibody. Data represent results from three experiments.
tin ligase AIP4 (but not NEDD4), which primes it for traffic through the ESCRT pathway, resulting in lysosomal proteolysis (43,44). Nef recruits either E3 ligase for G␣ i2 ubiquitination without need for agonist treatment.
Nef bound G␣ i2 much better than other G␣ species in vitro. Given the level of sequence homologies among G␣ i isoforms, it is not surprising that Nef bound to G␣ i3 at 4.8 pM concentration. But it occurred at relatively much higher concentrations of G␣ i3 compared with G␣ i2 , indicating that Nef preferably binds to G␣ i2 rather than G␣ i3 . We repeated the experiment with increasing concentration of GSTNef, and we found binding of G␣ i3 to Nef (data not shown). But this occurred at a concentration that is physiologically not relevant. Nef and G␣ i2 co-localized in vivo in perinuclear vesicles enriched for endolysosomal markers, which was further validated by increased FRET intensity between Nef-CFP and G␣ i2 -YFP in the intracellular vesicular structures. There was also extensive co-localization of Nef with G␣ i2 (but not G␣ i3 ) and AIP4 (or NEDD4) in the same perinuclear vesicles. Taken together, these findings strongly implied that Nef-induced G␣ i2 ubiquitination occurs in a ternary complex of Nef, E3 ligase, and G␣ i2 . AIP4 binding to substrates is mediated by its WW domain, which targets polyproline motifs like PPPY or PPXY (47,49,50) or phosphothreonine and phosphoserine residues (51). Our studies showed that AIP4 bound G␣ i2 through the HECT domain and Nef, presumably through the WW domain(s). Accordingly, AIP4 poorly bound Nef mutants that substituted the polyglutamate at 62 (E62A) or the polyproline tract at position 72 (P72A). This suggested that the Src homology 3 binding domain of Nef, with the PXXP motif, which has been demonstrated to be important for viral replication (52,53), might be crucial for the down-regulation of G␣ i2 . Additional residues surrounding the PXXP motif have been shown to affect the HIV replication and disease progression (54) and might account for the variation in the ability of various Nef alleles to down-regulate G␣ i2 .
Ubiquitin-dependent degradation of G-proteins(s) was originally shown for the yeast Gpa1, a G␣ equivalent (55)(56)(57), which was routed to multivesicular bodies for proteolysis. Mammalian G␣ o (58), G␣ i (59, 60), G␣ s (61,62), and G␤␥ (63, 64) subunits are presumed to be regulated by the proteasomal pathway; however, only G␣ s (61,62), transducin G␤␥ (64), and G␥ 2 (63) were shown to be ubiquitinated. To our knowledge, Nef-induced G␣ i2 loss represents the first instance where ubiquitination routes the G-protein for endolysosomal proteolysis. A systematic analysis of essential genes in yeast important for regulating G-protein signaling identified a preponderance of genes involved with protein degradation (65). Our results indicate that mammalian cells must also carefully control expression of the essential components in the G-protein signaling pathway and that HIV Nef has probably co-opted this mechanism that regulates G␣ i2 stability.
It was of interest that, although Nef spared G␣ i isomers other than G␣ i2 , early G-protein signaling and chemotaxis defects perpetrated by Nef implied that other G␣ i isomers were not recruited for signaling in the different cell types we have used. This was further confirmed by siRNA knockdown of either G␣ i2 or G␣ i3 , which resulted in a significant (ϳ60%) decrease in the agonist-mediated Ca 2ϩ flux. Notwithstanding the paramount role(s) of G␣ i in immune cell trafficking, chemokine receptors can couple to other G-proteins to facilitate chemotaxis (66 -68) in monocyte/macrophages. In addition, an alternative chemokine receptor pathway has been described for certain receptor/cell combinations in dendritic cells and neutrophils (69, 70) that is activated by G␣ i2 but was dependent on G q proteins for optimal G-protein signaling and chemotaxis. . Nef-induced loss of G␣ i2 was partially reversed by enzymatically defective AIP4 or by siRNA knockdown of AIP4 or NEDD4. A, the HECT domain mutant of AIP4 (C830A) reversed Nef-induced degradation of G␣ i2 . Jurkat cells were nucleofected with CD8, FLAG-tagged WT, or C830A AIP4 mutant and HA-tagged Nef or empty HA vector. Extracts from cells adjusted for equivalent CD8 expression were analyzed by immunoblotting using mAb against G␣ i2 and anti-FLAG antibody for detecting AIP4. Numbers refer to relative (%) G␣ i2 amounts for the respective transfectants adjusted for equivalent CD8 expression. B, Nef-induced loss of G␣ i2 was partially reversed by siRNA knockdown of AIP4 or NEDD4. Transfected Jurkat cells were disrupted in 0.5 ml of lysis buffer, and extracts were resolved by SDS-PAGE followed by immunoblotting for actin, NEDD4, AIP4, G␣ i2 , or G␣ i3 . Numbers denote relative (%) pixel densities of G␣ i2 averaged from three experiments. C, siRNA knockdown of AIP4 or NEDD4 partially reversed Nef-induced inhibition of chemotaxis toward CXCL12. After a 48-h treatment with the respective siRNAs, Jurkat cells were cotransfected with GFP and Nef(Ϫ) or Nef(ϩ) plasmid. Cells were evaluated for chemotaxis toward 20 nM CXCL12. Histograms represent pairwise comparison of relative (%) chemotaxis efficiency of GFP(ϩ) Nef(Ϫ) versus Nef(ϩ) (n ϭ 3). D, siRNA knockdown of AIP4 or NEDD4 did not interfere with Nef-mediated CD4 down-regulation. Jurkat cell transfectants were analyzed for CD4 expression by flow cytometry. Relative (%) CD4 mean fluorescence values for Nef(ϩ) and Nef(Ϫ) transfectants are plotted pairwise for each condition as histograms (with error bars). The mean fluorescence value for Nef(Ϫ) cells in each case was arbitrarily assigned as 100 (n ϭ 3); *, p Ͻ 0.05 when compared with their respective plasmid (mock)transfected controls.
Given that Gq stability and functionality were unaltered by Nef, migration of certain leukocyte subtypes, such as DCs and monocytes, might be differentially regulated in vivo.
Nef is a short-lived protein produced early in the HIV life cycle; Nef protein and fragments are packaged within nascent HIV particles and may be delivered to newly infected cells (71). Non-infected cells may also be subject to Nef effects either by its secretion by infected cells or by its transfer via intercellular nanotubular conduits from infected to non-infected cells, which can alter cellular function (72,73) and membrane dynamics causing a transfer of infected cell signaling to recipient cells (74). The targeting of G␣ i2 for destruction by Nef in infected as well as in nearby non-infected cells provides several potential advantages for HIV. First, disabling responsiveness to chemoattractants prevents cells from leaving a nidus of infection, thereby promoting cell-to-cell spread of the virus. Second, because of the importance of T cell migration for T cell function, disruption of directed T cell migration will impair the induction of effective immunity. Third, a sharp reduction in G␣ i2 will impair the sequestration of G␤␥ subunits. Freed G␤␥ subunits can activate downstream effectors that can provide signals that promote viral replication as well as effect cell survival. Fourth, decreased G␣ i2 removes its tonic inhibition of adenylyl cyclase isoforms, increasing cAMP levels and PKA activity, which can also enhance HIV replication (75)(76)(77). Fifth, because heterotrimeric G-proteins also function in GPCR-independent signaling pathways involved in intracellular trafficking and cell division (78 -80), Nef-induced G␣ i2 loss may impact T cell function by affecting these pathways.
Although the Nef-induced chemotaxis defect may derive in part from its effects on other signaling events, such as PAK2-mediated cofilin deregulation (29) or inappropriate DOCK2/ELMO1driven Rac activation uncoupled from chemokine receptor signaling (24), we have shown that Nef-induced loss of G␣ i2 probably trumps these because it impacts the earliest events in chemokine signaling. HIV, by expressing Nef, has chosen to target G␣ i2 for destruction. Nef probably does so by usurping a normal cellular mechanism that regulates G␣ i2 stability. Nef-induced loss of G␣ i2 in lymphocytes will profoundly affect their function and may impact signaling pathways that impact HIV replication.

Acknowledgments-We thank Philip Murphy (LMI, NIAID, National
Institutes of Health) for critical review of the manuscript. We thank Steven Becker (Biological Imaging Section, RTB, NIAID, National Institutes of Health) for technical advice. We thank Adriano Marchese (Stritch School of Medicine, Loyola University, Chicago) for the generous supply of numerous reagents and technical advice and Bruce Conklin (University of California, San Francisco) for the Gq i5 chimera. FIGURE 11. Lysine at position 296 of G␣ i2 is the critical determinant of Nef-induced degradation. A, the C-terminal 66 residues of G␣ i2 are shown at the top, with all lysines shaded and the three lysines unique to G␣ i2 denoted by asterisks. M1-M3, the respective lysine to arginine mutations engineered into the YFP-tagged G␣ i2 . CEM cells were transfected with CD8, WT, or the indicated G␣ i2 mutant and HA-tagged Nef or empty HA vector. Extracts of transfected cells were adjusted for equivalent CD8 expression and analyzed by immunoblotting using mAb against YFP for detecting G␣ i2 . Numbers at the bottom are relative (%) pixel densities of G␣ i2 averaged from two experiments. B, M1 mutant is more efficient than the WT G␣ i2 -YFP in rectifying Nef-induced G-protein signaling (by intracellular Ca 2ϩ flux) defect. CEM cells were nucleofected with CD8, and YFP-tagged WT or M1 G␣ i2 CD8 transfectants were analyzed for CXCL12-driven intracellular Ca 2ϩ flux as described in the legend to Fig. 3. C, YFP-tagged WT, but not the M1 G␣ i2 , was internalized and co-localized with Nef-enriched vesicular structures. HeLa cells in coverglass chambers were co-transfected with YFP-tagged WT or M1 G␣ i2 mutant with Nef-CerFP or CerFP and processed for live microscopy. Cerulean is green, and YFP is red.