Kinesin KIF4 Regulates Intracellular Trafficking and Stability of the Human Immunodeficiency Virus Type 1 Gag Polyprotein

Retroviral Gag proteins are synthesized as soluble, myristoylatedprecursors that traffic to the plasma membrane and promote viralparticle production. The intracellular transport of human immunodeficiencyvirus type 1 (HIV-1) Gag to the plasma membrane remains poorlyunderstood, and cellular motor proteins responsible for Gagmovement are not known. Here we show that disrupting the functionof KIF4, a kinesin family member, slowed temporal progressionof Gag through its trafficking intermediates and inhibited virus-likeparticle production. Knockdown of KIF4 also led to increasedGag degradation, resulting in reduced intracellular Gag proteinlevels; this phenotype was rescued by reintroduction of KIF4.When KIF4 function was blocked, Gag transiently accumulatedin discrete, perinuclear, nonendocytic clusters that colocalizedwith endogenous KIF4, with Ubc9, an E2 SUMO-1 conjugating enzyme,and with SUMO. These studies identify a novel transit stationthrough which Gag traffics en route to particle assembly andhighlight the importance of KIF4 in regulating HIV-1 Gag traffickingand stability.

INTRODUCTION

Top

INTRODUCTION

The Gag polyprotein is an essential structural component ofall retroviruses and lentiviruses, playing a leading role throughoutmultiple stages of the viral life cycle (1, 28). For a viralparticle to form, thousands of Gag proteins must multimerize,bind to membranes, and then assemble into a virion that is releasedfrom the cell. For human immunodeficiency virus type 1 (HIV-1)Gag, the components responsible for membrane binding have beenmapped to the N-terminal matrix domain (MA) and include theN-terminal myristate and a cluster of basic residues (35, 55,58). Regions within the capsid and nucleocapsid domains promoteGag-Gag multimerization, while the C-terminal p6 domain of Gagis needed for budding and release of the newly formed virion.Although Gag contains all the structural information neededto form a virus-like particle (VLP), assembly and release ofHIV-1 require recruitment and interaction of Gag with multiplehost cell proteins (4, 52).

A key unanswered question is how Gag proteins move from theirsite of synthesis on soluble, cytoplasmic polysomes to assemblypoints on cellular membranes. Recent studies point to the existenceof intracellular trafficking intermediates for HIV-1 Gag (38-40).However, the molecular basis for regulation of Gag movementwithin the cell is not known. We hypothesized that cellularmotor proteins might be important for Gag movement. This hypothesisis based on the finding that Gag proteins from HIV-1, simianimmunodeficiency virus, Moloney murine leukemia virus, and Mason-Pfizermonkey virus (MPMV) bind to KIF4, a kinesin family motor protein(21). The interaction was mapped to the N-terminal MA domainof Gag and the C-terminal tail of KIF4 (48). To date, no studieshave examined the functional importance of KIF4 interactionwith HIV-1 Gag.

KIF4A (hereby denoted KIF4) is a plus-end-directed, microtubule(MT)-stimulated ATPase in the kinesin motor family (24, 34).KIF4 contains a characteristic N-terminal motor domain, whichbinds MTs and ATP and is responsible for force generation alongMTs (10, 16, 45). It also contains a central stalk region involvedin dimerization and a C-terminal tail believed to mediate bindingof cargo. KIF4 localizes to both the cytoplasm and the nucleus(24) and has been implicated in organelle and vesicle transport,midzone formation, chromosome segregation during mitosis, andcytokinesis (22, 37, 59) as well as in regulation of programmedcell death in juvenile neurons (27). These studies establishKIF4 as a multifaceted motor protein that regulates the movementof multiple intracellular components.

Here we describe the first functional evidence linking KIF4to early stages of HIV-1 Gag biosynthetic trafficking. We showthat disruption of endogenous KIF4 slows progression of newlysynthesized Gag through its trafficking intermediates and resultsin decreased particle production. In addition, we report a novel,unanticipated regulatory function for KIF4, i.e., control ofGag protein stability. When KIF4 is disrupted, degradation ofintracellular Gag is dramatically increased. Finally, we identifythe earliest defined trafficking intermediate as a perinuclearcluster of Gag that also contains KIF4, Ubc9, and SUMO. Failureto progress beyond this stage results in degradation of intracellularGag protein. This knowledge could potentially be exploited asa means to develop therapeutic agents that block HIV-1 viralparticle production.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Antibodies and reagents. Polyclonal rabbit anti-actin was purchased from Sigma, St. Louis,MO; monoclonal mouse anti-CD63, polyclonal rabbit anti-Fyn,monoclonal mouse anti-caveolin-1, and monoclonal mouse anti-EEA1were purchased from BD Transduction Laboratories, San Diego,CA; monoclonal mouse anti-myc, polyclonal rabbit anti-extracellular-signal-regulatedkinase 2 (anti-ERK2), polyclonal rabbit anti-green fluorescentprotein (anti-GFP), monoclonal mouse anti-vimentin, monoclonalmouse anti-lamin A, and monoclonal mouse anti-gp120 were purchasedfrom Santa Cruz Biotechnology, Santa Cruz, CA; monoclonal mouseanti-Src was purchased from Upstate Biotechnology/Millipore,Billerica, MA; monoclonal anti-CD4 (leu3A) was purchased fromthe MSKCC Core Antibody Facility; anti-mouse Alexa fluor 594and anti-rabbit Alexa fluor 594 were purchased from MolecularProbes/Invitrogen, Carlsbad, CA; and HIV-1 p24 monoclonal antibodywas obtained from the NIH AIDS Research & Reference ReagentProgram, Germantown, MD. Polyclonal anti-p24CA was generatedin rabbits immunized with p24CA. Antibody to KIF4 was a kindgift from Mariano Bisbal and Alfredo Caceres (Mercedes y MartinFerreyra Institute for Medical Research, Cordoba, Argentina).Protein Plus A/G-agarose was purchased from Santa Cruz Biotechnology.Tran³⁵S Cys/Met was purchased from MP Biomedicals (Solon, OH).Chloroquine and MG-132 were purchased from Sigma.

Plasmids, transfections, and metabolic labeling. Plasmids encoding pCMV5 Gag, GagGFP, Gag( ${Delta}$ p6), pHXB2 ${Delta}$ BalD25S Gag,Fyn(10)Gag, c-Src, and Fyn have been described previously (15,26, 50). Plasmids encoding KIF4-T and KIF3A-T were generatedby reverse transcription of human A549 cell mRNA, PCR, and cloninginto Gateway (Invitrogen) mammalian expression vectors. KIF3A-TcDNA construction has been described previously (17). KIF tailconstructs were amplified by PCR from human A549 cells and clonedinto mammalian Gateway (Invitrogen) expression vectors as describedby the manufacturer. KIF4A-T (amino acids [aa] 741 to 1232)and KIF4A-TS (aa 875 to 1232) were amplified using forward primers5' GCTCGAGTGAAGAATTGGCTTGGA 3' and 5' TCCTCCAAAATACAGGTCAGCAAACTT3', respectively. The reverse primer was 5' TCAGTGGGCCTCTTCTTCGATAGGGGA3'. Lentiviral Gag was generated by cloning GagGFP into theViraPower lentiviral expression system (Invitrogen). The c-SrcGFPplasmid contains chicken c-Src fused to GFP with a 7-aa linkerbetween the C terminus of c-Src and the N terminus of GFP. pcDNA_vSrc( ${Delta}$ MA)Gag-mycwas a kind gift from Paul Spearman (Emory University Schoolof Medicine, Atlanta, GA). The pNL4.3 HIV-1 cDNA was a kindgift from John P. Moore (Weill Graduate School of Medical Sciencesof Cornell University, New York, NY). pEGFP-N2-CD4FL was a kindgift from John Wills (Pennsylvania State University Collegeof Medicine, Hershey). pCMV.Myc-Ubc9 was a kind gift from RobertWeldon (University of Nebraska, Lincoln, NE). The GFP-250 plasmidwas a kind gift from Elizabeth Sztul (University of Alabamaat Birmingham, Birmingham). pGFP-SUMO1 was a kind gift fromPier Paolo Pandolfi (Memorial Sloan-Kettering Cancer Center,New York, NY). KIF3 and KIF4 small interfering RNAs (siRNAs)were obtained from the HTS/siRNA core facility at Memorial Sloan-KetteringCancer Center. The KIF4 sequences used were as follows: KIF4_siRNA#1,5' GCUUCAAGAUUCUCUAGGA 3'; KIF4_siRNA#2, 5' CAAUUGAUUACCCAGUUAU3'; and KIF4_siRNA#3, 5' CCACUGAAGUUGGAGUCAU 3'.

COS-1 cells were seeded to 90% confluence and transfected withcDNA, using Lipofectamine 2000 (Invitrogen). Metabolic labelingwas performed after 24 h of incubation at 37°C as describedpreviously (49), using 50 µCi/ml Tran³⁵S label (MP Diagnostics).Briefly, cells were incubated in cysteine- and methionine-deficientDulbecco's modified Eagle's medium plus 10% fetal bovine serum(FBS) for 1 h. Cells were pulse labeled for 15 min at 37°Cand either maintained in medium containing label or chased forvarious lengths of time in chase medium (Dulbecco's modifiedEagle's medium plus 10% FBS plus 100 µM cysteine and methionine).VLPs were purified from the medium as described below. Cellswere harvested by being washed twice in cold saline-Tris-EDTAbuffer followed by lysis in 1x RIPA buffer (150 mM NaCl, 1 mMEDTA, 0.1% sodium dodecyl sulfate [SDS], 0.5% deoxycholate,1% Triton X-100, and 10 mM Tris, pH 7.4) containing proteaseinhibitors (0.5 mM 4-(2-aminoethyl)benzenesulfonyl flouride,0.5 µg/ml leupeptin, and 0.5 µg/ml aprotinin). Thelysate was clarified at 55,000 x g at 4°C for 15 min andimmunoprecipitated with rabbit anti-p24 antibody and proteinA/G beads overnight at 4°C. Samples were washed three timeswith cold 1x RIPA buffer and then subjected to SDS-polyacrylamidegel electrophoresis. Analysis of radiolabeled Gag was performedby exposure of the dried gels to a phosphorimager screen, whichwas scanned using a Fuji FLA500 phosphorimager. Quantitationand preparation of visual images were performed with ImageQuantsoftware (Molecular Dynamics).

VLP preparation. VLPs were purified as previously described (26). Briefly, mediumfrom transfected cells was clarified at 500 x g for 10 min at4°C and filtered through a 0.2-µm-pore-size filter.The filtrate was layered on top of a 20% sucrose cushion andcentrifuged for 2 h at 34,500 rpm in a Ti55 rotor. VLPs weresolubilized in 1x RIPA buffer and immunoprecipitated using rabbitanti-p24 antibody (as described above).

Proteasomal and lysosomal inhibition. COS-1 cells transfected with appropriate cDNAs and incubatedfor 24 h were then preincubated for 1 h with 10 µM MG-132(proteasomal inhibition) or 20 mM NH₄Cl plus 30 µM chloroquine(lysosomal inhibition). Cells were then pulsed with 50 µCi/mlTran³⁵S label, chased over a period of 7 h, and processed asdescribed above.

Microinjection. COS-1 cells were pressure microinjected intranuclearly withcDNAs in HKCl (10 mM HEPES, 140 mM KCl, pH 7.4), using a Narishigemicromanipulator (Narishige, Greenvale, NY). cDNA concentrationswere 20 µg ml^–1 for GagGFP and 5 to 10 µgml^–1 for KIF tail constructs. After injection, cells weremaintained at 37°C for 120 min to allow for expression ofcDNAs. Cells were transferred to recording medium (phenol red-freeHanks balanced salt solution with 20 mM HEPES, 1% FBS, 4.5 g/literglucose, and essential and nonessential amino acids). Cellswere incubated in a thermally controlled chamber (Harvard Apparatus)at 37°C (an open chamber was used to permit oxygen exchange)on a Nikon TE-2000U microscope (Nikon Inc., Mellville, NY),and GagGFP trafficking was monitored by time-lapse fluorescencemicroscopy. Images were collected at the indicated intervals,using a x40 objective and an ORCAII-ER microscope (HamamatsuPhotonics, Bridgewater, NJ). All devices were controlled byMetaMorph (Molecular Devices, Sunnyvale, CA).

KIF4 knockdown with siRNA. COS-1 cells were transfected on two consecutive days with 20nM siRNA in Lipofectamine 2000. On the second day of transfection,the siRNA was cotransfected with the indicated cDNA. KIF4 andGag mRNA levels were quantified from reverse-transcribed mRNAby TaqMan real-time quantitative PCR with appropriately designedprimers. Changes in KIF4 protein levels were quantified by comparingthe integrated fluorescence intensities obtained with anti-KIF4and anti-calnexin antibody staining in cells treated with KIF4_siRNAor LacZ_siRNA.

Immunofluorescence. Immunofluorescence analysis was performed as previously described(38). Briefly, at 24 h posttransfection, COS-1 cells were washedwith phosphate-buffered saline (PBS) and fixed with 4% formalin-PBSfor 15 min. Cells were incubated with a 1:250 dilution of primaryantibody in PBS plus 10% FBS for 1.5 h, washed in PBS, and thenincubated with a 1:500 dilution of secondary antibody for 45min. Cells were washed and mounted on microscope slides in Pro-Longmounting medium (Invitrogen). Laser scanning confocal microscopywas performed on a Zeiss LSM510 confocal microscope equippedwith an Axiovert 100 M inverted microscope, using a x63, 1.2-numerical-aperturewater immersion lens for imaging. Single confocal images wereprocessed in Adobe Photoshop.

RESULTS

Top

RESULTS

Disruption of endogenous KIF4 decreases VLP production and alters intracellular Gag stability. We first examined the effect of disrupting endogenous KIF4 onGag function, specifically VLP production. Previous studieshave established that constructs containing only the cargo-bindingtail domain of various kinesins function as dominant-negativeinhibitors of kinesin function (17, 46, 51, 54). We generatedKIF4-T, a tail domain cDNA construct encoding aa 741 to 1232of KIF4 fused to murine red fluorescent protein (mRFP) (17).As a control, we used a dominant-negative KIF3A tail constructencoding aa 601 to 702 of the KIF3A C-terminal tail (KIF3A-T)fused to mRFP. COS-1 cells were cotransfected with cDNAs encodingHIV-1 Gag and either myc-tagged KIF4-T or KIF3A-T. Cells wereincubated with [³⁵S]Cys/Met-containing medium 24 h after transfection,and incorporation of radiolabel into cell-associated Gag andGag in VLPs was quantified over 4 h (Fig. 1a). The rates ofradiolabel incorporation into Gag in cells expressing eitherKIF4-T or KIF3A-T were essentially equivalent during the 4-hlabeling time course. However, VLP production was significantlyreduced ( $~$ 10-fold) in the presence of the dominant-negative KIF4-Tconstruct compared with that in cells expressing the controlconstruct KIF3A-T (Fig. 1b).

View larger version (33K):
[in this window]
[in a new window]

FIG. 1. Effects of KIF4 disruption on VLP production and intracellular Gag stability. (a and b) COS-1 cells were cotransfected with 2 µg Gag and 2 µg of either KIF3A-T or KIF4-T cDNA and radiolabeled with Trans³⁵S label. Radiolabeled Gag from cell lysates and VLPs was immunoprecipitated with anti-p24 antibody, followed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and phosphorimager analysis. The amount of actin was determined by Western blotting. (c and d) COS-1 cells were cotransfected with Gag and KIF3A-T or KIF4-T, and levels of intracellular Gag were monitored over a 24-h period by Western blotting of cell lysates. (e) COS-1 cells were cotransfected with Gag and either empty vector, KIF4-T, LacZ_siRNA, or KIF4_siRNA. Transfected cells were then preincubated for 1 h in the presence or absence of either a proteasome inhibitor ([PI] MG-132) or lysosomal inhibitors ([LI] NH₄Cl and chloroquine). The cells were pulse labeled for 15 min in medium containing Tran³⁵S-labeled cysteine-methionine and then chased in nonradioactive medium with the respective inhibitor for the indicated times. The amount of labeled Gag remaining at each time point was quantified and plotted relative to the amount of Gag at time zero (100%). Each experiment was performed in duplicate a minimum of two times.

To better understand the effects of KIF4-T on Gag function,we monitored intracellular Gag levels over a 24-h period. COS-1cells were cotransfected with HIV-1 Gag cDNA and either KIF4-Tor control KIF3A-T cDNA, incubated for 24 h, and then radiolabeledin medium containing [³⁵S]Cys/Met for up to 24 h. At early timepoints (up to 5 h), Gag levels were equivalent under the twoconditions. However, at later time points, beyond 7 h, intracellularGag levels were significantly reduced ( $~$ 5-fold at 24 h) in cellsexpressing KIF4-T compared with those in cells expressing KIF3A-T(Fig. 1c and d). To determine whether the reduction in intracellularGag was due to decreased Gag stability, a pulse-chase experimentwas performed to monitor rates of Gag degradation. Under controlconditions, Gag degraded with a half-life of $~$ 6.5 h (Fig. 1e).In cells expressing KIF4-T, the half-life for Gag was considerablyshorter, i.e., $~$ 1 h. The addition of either proteasomal (MG-132)or lysosomal (NH₄Cl and chloroquine) inhibitors did not haveany effect on Gag degradation rates (Fig. 1e). In contrast,these compounds did inhibit epidermal growth factor receptordegradation (see Fig. S1 in the supplemental material). Thus,expression of endogenous KIF4-T leads to an increase in therate of Gag degradation, thereby reducing intracellular Gagprotein stability.

Regulation of Gag stability by KIF4 is mediated through the N-terminal region of Gag. To determine whether KIF4 disruption altered host cell proteinexpression, we monitored the levels of several endogenous proteinsin cells transfected with cDNAs encoding either KIF3-T or KIF4-Tor with an empty vector. Western blotting with anti-myc antibodyrevealed equivalent expression levels of KIF3-T and KIF4-T.We detected no changes in levels of endogenous actin, ERK-2,or caveolin-1 (Fig. 2a and b). Next, we examined expressionlevels of two transfected proteins, c-Src and Fyn. Like Gag,these two proteins are myristoylated and targeted to the plasmamembrane. Moreover, c-Src and Gag use a similar mechanism ofmembrane binding, a myristate-plus-basic motif in the N-terminalregion (41, 58). Despite these similarities, the expressionlevels of c-Src and Fyn were not affected by the presence ofKIF4-T (Fig. 2a and b).

View larger version (40K):
[in this window]
[in a new window]

FIG. 2. Expression of dominant-negative KIF4 causes a reduction in intracellular Gag level. COS-1 cells were cotransfected with either empty vector, KIF3A-T, or KIF4-T along with constructs encoding the indicated Gag proteins, c-Src, or Fyn or with pHXB2 ${Delta}$ BalD25S. The cells were incubated for 24 h. Levels of the indicated exogenous proteins, of endogenous actin, ERK-2, and caveolin-1 (in Gag-transfected cells), and of Env (anti-gp120) were monitored by Western blotting with the appropriate antibodies. (a) Representative panels (triplicates) showing expression of the indicated proteins. (b) Graphical representation of the combined experiments represented in panel a. Each experiment was performed a minimum of two times in triplicate.

The effects of KIF4-T on Gag derived from several differentexpression vectors were evaluated. We observed significant reductionsin Gag expression for both the GagGFP fusion protein and untaggedGag. Both of these constructs encode codon-optimized Gag drivenby a cytomegalovirus promoter. We also monitored Gag expressedfrom an HIV-1-derived lentiviral construct, pHXB2 ${Delta}$ BalD25S; inthis plasmid, Gag expression is driven by the HIV-1 long terminalrepeat. The presence of KIF4-T induced a dramatic decrease inthe level of Gag expression (Fig. 2a and b). Finally, the effectof KIF4-T was tested in cells transfected with an infectiousclone of pNL4.3 HIV-1 cDNA. Gag levels were strongly reducedby KIF4-T but not by KIF3-T (Fig. 2b). In contrast, no effecton another viral protein, Env, was detected (Fig. 2a and b).

The N-terminal matrix region of Gag has been shown to interactwith KIF4 (48). We tested the ability of KIF4-T to alter expressionlevels of the following three Gag mutants: Gag( ${Delta}$ P6), a constructwith a C-terminal deletion of the p6 region of Gag; Src( ${Delta}$ MA)Gag,where the matrix domain of Gag is replaced with the first 10aa from Src; and Fyn(10)Gag, a chimera where the first 10 aaof Gag are replaced by those of Fyn (7, 26). In the presenceof KIF4-T, Gag( ${Delta}$ P6) expression was reduced significantly, althoughto a slightly lesser extent than that observed for full-lengthGag. In contrast, KIF4-T had no effect on the level of Src( ${Delta}$ MA)Gagor Fyn(10)Gag (Fig. 2a and b). These findings imply that theeffects of disrupting endogenous KIF4 with dominant-negativemutants are specific to Gag and to the N-terminal region ofGag.

Intracellular Gag stability is reduced by KIF4_siRNA-mediated knockdown. To confirm that the reduction in Gag protein levels inducedby dominant-negative KIF4 was due to disruption of KIF4 function,we utilized siRNA technology to knock down endogenous KIF4.Three different siRNA sequences directed against human KIF4were tested. The levels of endogenous KIF4 protein in COS-1cells were undetectable by Western blotting. However, quantitativereverse transcriptase PCR (qPCR) revealed three- to fourfolddecreases in endogenous KIF4 mRNA levels compared to those inLacZ and no-siRNA controls (Fig. 3a). No significant effecton Gag mRNA levels was detected in cells treated with any ofthe KIF4_siRNAs (Fig. 3b). In addition, we observed an $~$ 60% decreasein KIF4 immunofluorescence in cells treated with KIF4 siRNAcompared with that in cells treated with a control LacZ siRNA.

View larger version (24K):
[in this window]
[in a new window]

FIG. 3. siRNA-directed knockdown of KIF4 reduces intracellular Gag levels. (a and b) COS-1 cells were transfected with one of three different KIF4_siRNA constructs, with LacZ_siRNA, or with no siRNA. Levels of KIF4 mRNA (a) or Gag mRNA (b) were quantified by qPCR following reverse transcriptase PCR with KIF4- or Gag-specific TaqMan probes. Each experiment was performed in triplicate. (c) COS-1 cells were cotransfected with either control LacZ_siRNA or a KIF4_siRNA (20 nM) along with constructs encoding the indicated Gag proteins, c-Src, or Fyn. For Gag only, cells were also cotransfected with KIF3_siRNA (40 nM). The cells were incubated for 24 h. Levels of the indicated exogenous proteins or of endogenous actin, ERK-2, or caveolin-1 (in Gag-transfected cells) were monitored by Western blotting with the appropriate antibodies. Each experiment was performed a minimum of two times in triplicate. Lenti-Gag*, Jurkat T cells expressing Gag via a lentiviral delivery system. (d) Gag-transfected COS-1 cells were cotransfected with either LacZ_siRNA or KIF4_siRNA#3 and with either empty vector (left axis) or a vector expressing full-length rat KIF4 (right axis). The amounts of Gag and rat KIF4 in cell lysates were determined by Western blotting (representative blots are shown below the graphs). (e) The amount of Gag in VLPs isolated from cells treated as described for panel d was quantified.

The intracellular levels of several transfected and endogenousproteins were compared in COS-1 cells treated with KIF4_siRNAversus LacZ_siRNA controls. Treatment of cells with KIF4_siRNAresulted in decreased intracellular levels of Gag, pNL4.3 Gag,pHXB2 ${Delta}$ BalD25S Gag, and Gag( ${Delta}$ P6) but had no effect on the intracellularlevels of transfected Src( ${Delta}$ MA)Gag, Fyn(10)Gag), c-Src, and Fynor endogenous actin, ERK-2, and caveolin-1 (Fig. 3c). As anadditional control, we monitored the effects of KIF3A_siRNAtreatment of COS-1 cells. No change in levels of Gag were observedcompared to those after LacZ_siRNA treatment (Fig. 3c). We alsomonitored the effects of KIF4_siRNA treatment in Jurkat T cells.In this case, Gag was expressed via a lentiviral infection deliverysystem (Lenti-Gag*). Again, KIF4_siRNA caused a strong reductionof Gag levels in Jurkat T cells, an effect equivalent to thatobserved in COS-1 cells (Fig. 3c).

The rat KIF4 sequence is 87% identical to that of human KIF4and is predicted to be resistant to the effects of KIF4_siRNA#3directed against human KIF4. Expression of full-length rat KIF4rescued the Gag degradation phenotype and the defect in VLPproduction in KIF4_siRNA#3-treated cells (Fig. 3d and e). Moreover,intracellular Gag levels were increased in the presence of full-lengthrat KIF4 (Fig. 3d). Taken together, these data support the hypothesisthat KIF4 regulates Gag protein stability. Knockdown of KIF4induces Gag degradation, while increased KIF4 expression resultsin higher levels of intracellular Gag protein.

Gag intracellular trafficking is delayed when KIF4 function is disrupted. The GagGFP fusion protein has been exploited extensively asa stable, accurate reporter of Gag localization in transfectedcells (15, 38). We microinjected cDNA encoding GagGFP into COS-1cells and used time-lapse fluorescence microscopy to followthe intracellular trafficking of newly synthesized GagGFP. GagGFPfluorescence progressed through at least four phenotypicallydistinct stages, termed stages I to IV, over a period of 2 to6 h postmicroinjection (Fig. 4a). The stages were characterizedbased on the number of GagGFP puncta and their mobility. GagGFPwas initially dispersed diffusely throughout the cytoplasm (stageI). GagGFP puncta soon became apparent in the cytoplasm (stageII). There were typically 3 to 20 puncta per cell, and the punctawere relatively stationary. With time, the number of GagGFPpuncta increased (>20 per cell), and they exhibited movementand began to disseminate throughout the cytoplasm (stage III).By 6 h, there were numerous, highly mobile GagGFP puncta (stageIV). The stage IV pattern was typical of the Gag and GagGFPdistribution at steady state (31, 38). Quantification revealedthat >50% of the cells had reached stage IV and only 10%remained at stage I by 6 h postinjection (Fig. 4b and c) (n= 56 cells).

View larger version (62K):
[in this window]
[in a new window]

FIG. 4. Expression of dominant-negative KIF4 delays Gag trafficking. (a) COS-1 cells were microinjected with GagGFP cDNA and monitored for 6 h. Four distinct phenotypic stages were observed and were classified as stages I to IV, based on the number of GagGFP puncta and their movement within the cell. TMTC, too many to count. Puncta in stage II are indicated by white boxes; the inset represents an enlargement of one of the puncta. Bar = 10 µm. (b) At each time point, the percentage of cells with GagGFP at each of the four stages was quantified in the absence (left; n = 56) or presence (right; n = 59) of coinjected KIF4-T. (c) GagGFP was coinjected with one of three controls (empty vector, RFP, or KIF3A-T) or one of two dominant-negative KIF4 constructs (KIF4-T or KIF4-Ts). The percentage of cells with GagGFP at stage I (left) or stage IV (right) at the indicated time points was quantified.

The appearance of GagGFP fluorescence depends on folding andmaturation of the GFP fluorophore. To determine whether additionalGag intermediates had formed but were not detected by GFP fluorescence,we analyzed GagGFP-microinjected cells that were fixed and stainedwith anti-GFP antibody. The staining patterns were identicalto those observed by live imaging. Confocal imaging revealedthat Gag and Gag puncta at early stages were localized withinthe interior of the cell and that by 6 h, many of the punctawere associated with the plasma membrane (see movies with three-dimensionalimage reconstruction in the supplemental material). Moreover,the stages observed in cells microinjected with GagGFP cDNAstrongly resemble those reported recently for cells transfectedwith GagGFP (38). These results indicate that a series of intracellularGag intermediates are visible before Gag appears at the plasmamembrane.

We next examined the effect of interfering with KIF4 functionon the kinetics of Gag trafficking. When GagGFP and mRFP-KIF4-TcDNAs were coinjected, Gag trafficking was significantly delayed.Nearly 80% of the cells remained at stage I and only 15% hadprogressed to stage IV by the 6-h time point (Fig. 4b and c).We obtained similar results when cells were coinjected withGagGFP and a shorter KIF4-T cDNA (KIF4-TS, encoding aa 875 to1232) (Fig. 4c). In contrast, coexpression of mRFP or RFP-KIF3A-TcDNA did not affect GagGFP trafficking (Fig. 4c). These datasuggest that KIF4 is involved in intracellular transport ofHIV-1 Gag.

Disruption of endogenous KIF4 results in the accumulation of intracellular Gag in perinuclear clusters. To further investigate the mechanism by which KIF4 disruptionaltered Gag levels, we performed a more detailed time courseof GagGFP trafficking in COS-1 cells coinjected with GagGFPand mRFP-KIF4-T cDNAs. Time-lapse recordings revealed an accumulationof GagGFP into large clusters in the perinuclear region, whichwe termed Gag perinuclear clusters (GPCs) (Fig. 5b), comparedto Gag at steady state (Fig. 5a). GPCs were also visible inGagGFP-transfected cells treated with KIF4_ siRNA (Fig. 5c and d)or cotransfected with KIF4-T (Fig. 6a). Quantitative analysisrevealed a five- to sixfold increase in the number of cellswith GPCs when Gag-GFP-expressing cells were treated with KIF4_siRNAor cotransfected with KIF4-T (Fig. 6a). Typically, GPCs appearedto be large structures of 0.5 to 1 µm in diameter andshowed a clear outline by differential interference contrastimaging (Fig. 5d).

View larger version (43K):
[in this window]
[in a new window]

FIG. 5. KIF4 disruption induces the formation of GPCs. (a and b) COS-1 cells were microinjected with GagGFP and either control vector (a) or vector encoding KIF4-T (b). The inset (b) shows the expression of RFP-tagged KIF4-T. (c and d) The differential interference contrast image of one cell transfected with GagGFP cDNA and KIF4_siRNA was magnified three times to visualize the size and outline of a single GPC (arrows). (e and f) COS-1 cells were cotransfected with cDNAs encoding c-SrcGFP and either KIF3A-T (e) or KIF4-T (f) constructs. (g and h) COS-1 cells were cotransfected with CD4 cDNA and either LacZ_siRNA (g) or KIF4_siRNA and GagGFP cDNA (h). Bar = 10 µm (a and c to h) or 15 µm (b).

View larger version (33K):
[in this window]
[in a new window]

FIG. 6. GPCs form during the natural progression of intracellular Gag trafficking. (a) The percentage of GagGFP-positive cells (total number of cells counted is shown in parenthesis) that contained GPCs after 24 h was determined in COS-1 cells cotransfected with GagGFP and either KIF3A-T, LZ_siRNA, KIF4-T, or KIF4_siRNA. (b) COS-1 cells were transfected with either pHXB2 ${Delta}$ BalD25S or pNL4.3 (full-length infectious clone) and stained with anti-p24 antibody and Alexa 488 to detect GPCs (green). (c) Sequential images showing that GagGFP perinuclear clusters form (arrows) and disperse (*) over a period of 2 h in untreated cells. (d) GagGFP-expressing cells were stained with anti-KIF4 antibody to detect endogenous KIF4 (red). Bar = 20 µm.

To determine whether formation of GPCs was specific to Gag,the effects of KIF4 disruption on localization of two otherproteins were evaluated. No obvious differences in the localizationof c-Src were observed in the presence of KIF4-T compared tothat in the KIF3A-T control (Fig. 5e and f). Moreover, localizationof CD4, a protein that traffics to the plasma membrane via thesecretory pathway, was unaffected by knockdown of KIF4 withKIF4_siRNA (Fig. 5g and h). Thus, the host cell traffickingpathways for c-Src and CD4 appear to remain intact when KIF4function is disrupted.

We next examined whether GPCs formed during the natural progressionof Gag intracellular trafficking. Cells were microinjected withGagGFP cDNA, and 4.5 h later, they were transferred to recordingmedium containing cycloheximide. Over the next 2 h, transientformation and dispersion of multiple GPCs were clearly evident(Fig. 6c). GPCs were also identified in cells transfected withpHXB2 ${Delta}$ BalD25S Gag and in cells transfected with a full-lengthinfectious clone of pNL4.3 (Fig. 6b). Staining with anti-KIF4antibody revealed significant colocalization between GPCs andendogenous KIF4 (Fig. 6d). These data imply that newly synthesizedGag transiently forms clusters that also contain KIF4.

Gag colocalizes with the E2-sumoylating enzyme Ubc9 in perinuclear clusters. The next set of experiments was designed to identify the cellularstructure or compartment(s) that contained the GPCs. Stainingfor lamin A revealed that GPCs, while often adjacent to thenucleus, were not enclosed by the nuclear membrane (Fig. 7a).Another possibility is that GPCs are aggresomes. These structuresare sometimes formed when misfolded proteins accumulate in thecytoplasm and aggregate near the MT organizing center; proteinsin the aggresome are then degraded by the proteasome. Sinceaggresomes are known to induce the reorganization of vimentinfilaments around the MT organizing center (18), we monitoredvimentin staining in the presence of KIF4_siRNA. No apparentrestructuring of vimentin filaments around the GPCs was evident(Fig. 7b). Moreover, no colocalization between GagGFP GPCs andGFP-250, an aggresome marker (9), was noted in the presenceof KIF4 siRNA (Fig. 7c).

View larger version (27K):
[in this window]
[in a new window]

FIG. 7. GPCs colocalize with the E2-SUMO-conjugating enzyme Ubc9 but not with endocytic markers. (a, b, d, e, and f) COS-1 cells were cotransfected with GagGFP and KIF4_siRNA and stained for lamin A (red) (a), vimentin (red) (b), Lysotracker (red) (d), EEA-1 (red) (e), or CD63 (red) (f). (c) Imaging of COS-1 cells that were triply transfected with GagRFP, GFP-250, and Kif4_siRNA. (g) COS-1 cells were cotransfected with GagGFP, myc-hUbc9 (red), and KIF4-siRNA. Cells were stained with anti-myc and Alexa 594 antibodies. (h) COS-1 cells were cotransfected with GagRFP, SUMO-1 GFP, and KIF4-siRNA and imaged directly. Bar = 10 µm.

Gag has been shown to utilize compartments of the endocyticpathway for trafficking and assembly (11, 32, 38, 47). A subsetof Gag proteins colocalize with CD63, a marker for the multivesicularbody (MVB) (36, 38). Accordingly, we stained COS-1 cells cotransfectedwith GagGFP and KIF4_siRNA with Lysotracker, a marker of cellularacidic, endocytic compartments, or with antibodies to the earlyendosome marker EEA-1 and the late endosome/MVB marker CD63.No significant colocalization between any of these endocyticmarkers and GPCs was detected (Fig. 7d to f).

A previous study reported colocalization of MPMV Gag with theE2-sumoylating enzyme Ubc9 in structures described as Gag perinuclearfoci (53). Ubc9 localizes to the nucleus and to the cytoplasmand functions as an E2-sumoylating enzyme for multiple sumoylatedproteins (57). Given the similarity between these perinuclearfoci and HIV-1 Gag-containing GPCs, we examined the localizationof Ubc9 and HIV-1 Gag GPCs in cells cotransfected with GagGFP,myc-tagged Ubc9, and KIF4_siRNA. As depicted in Fig. 7g, GPCscolocalized with perinuclear cytoplasmic clusters containingUbc9. In addition, in the presence of Ubc9, there was significantcolocalization of GPCs with SUMO-GFP or myc-SUMO (Fig. 7h).In contrast, no colocalization between c-Src and myc-Ubc9 wasdetected in KIF4 siRNA-treated cells (see Fig. S2 in the supplementalmaterial). Taken together, these data establish the GPC as aUbc9- and SUMO-positive compartment that participates in theintracellular trafficking pathway for HIV-1 Gag.

DISCUSSION

Top

DISCUSSION

KIF4 regulates Gag stability and trafficking. In this study, we identify a novel role for the cellular motorprotein KIF4 as a regulator of HIV-1 Gag protein stability andintracellular trafficking. Our results imply that Gag accumulatesin GPCs as part of its intracellular trafficking pathway. WhenKIF4 is functional, GPC formation is transient. When KIF4 functionis disrupted, a subpopulation of Gag accumulates in GPCs, andthe remainder is degraded. Our data imply that GPCs, KIF4, andpotentially other cellular components in GPCs play a role incontrolling the stability of intracellular Gag protein.

Regulation of Gag protein stability. We and others have shown previously that a subpopulation ofnewly synthesized HIV-1 Gag is degraded shortly after synthesis(44, 49). This population first appears in detergent-resistant,cytoplasmic complexes that fractionate in the middle of a 0to 18% Optiprep gradient (49). Thus, degradation is part ofthe natural course of Gag biosynthesis. The studies in thisreport reveal a mechanism for increasing Gag degradation. Disruptionof KIF4, either with KIF4-T or KIF4_siRNA, resulted in a lossof intracellular Gag protein due to an increased rate of Gagdegradation (Fig. 1 to 3). The protease responsible for Gagdegradation has yet to be identified, but the mechanism is neitherproteasome nor lysosome dependent (Fig. 1e). Because KIF4 disruptiondid not alter levels of endogenous proteins (actin, ERK2, andcaveolin-1), other exogenous, transfected proteins (c-Src andFyn), or Gag chimeras with altered N termini, we concluded thatthe effect of KIF4 is specific to the N-terminal MA region ofHIV-1 Gag. This is the region of Gag that has been shown tointeract directly with KIF4 (48).

At least three other examples of increased Gag turnover havebeen documented. First, overexpression of a chimeric proteincontaining the cytoplasmic tail of the HIV-1 Env protein gp41fused to β-galactosidase (6) was shown to induce HIV-1Gag downregulation. Complexes of Gag and the β-galactosidase-Envfusion protein were detected in the perinuclear region of thetransfected cells. Second, in HIV-1-expressing cells, siRNA-directedknockdown of heterogeneous nuclear ribonucleoprotein A1 resultedin reduced Gag expression (25). Third, siRNA-mediated depletionof SOCS1, or suppressor of cytokine signaling 1, decreased levelsof intracellular Gag and caused Gag to accumulate in perinuclearaggregates (43). However, these Gag aggregates differ from theGPCs reported here, as they colocalized with lysosomal markersand were subject to lysosome-mediated degradation. Regardlessof the mechanism of Gag degradation, strategies that reduceGag protein levels are of great interest because they have thepotential to serve as a novel form of antiviral therapy.

GPCs as an intracellular trafficking intermediate. Formation of perinuclear clusters of intracellular Gag was previouslydocumented by our laboratory, using live- and fixed-cell imagingof Gag- and GagGFP-transfected cells (38). Perinuclear clustersof Gag were generated early during Gag trafficking, within 1.5h after cells were released from a cycloheximide block. Theseclusters then dispersed into multiple puncta, which spread throughoutthe cell and ultimately concentrated at the plasma membrane.The results reported here, obtained using microinjected GagGFPcDNA, confirm this finding and establish the GPC as the earliestdetectable Gag intermediate. We do not know what percentageof the Gag in the GPC ultimately traffics to the plasma membrane.It is possible that a fraction of Gag traffics straight fromthe cytosol to the plasma membrane without passing through theGPC. It is also not known if Gag molecules multimerize withinthe GPC or, alternatively, are transported as monomers and multimerizeat the plasma membrane.

What is the identity of the GPC-containing compartment? GPCsare unlikely to represent aggregated, misfolded Gag in an aggresome,since no colocalization between GPCs and GFP-250, an aggresomemarker, was detected (Fig. 7b). Moreover, no colocalizationof GPCs with markers of the endocytic pathway was noted (Fig.7). GPCs were present, although transiently, even in cells whereKIF4 function was unperturbed, arguing that their formationis part of the natural course of Gag intracellular trafficking.The presence of a discrete number of GPCs per cell (typicallythree or four) suggests that GPCs represent a distinct subcellularcompartment that forms and is visible during Gag biosynthesis.

To assemble an infectious particle, HIV-1 Gag needs to recruitother viral proteins as well as viral RNA. In COS-1 cells expressingfull-length, wild-type HIV-1 Gag, Gag colocalizes with viralgenomic RNA in the perinuclear region (39). Thus, it is possiblethat GPCs comprise the region of Gag-RNA interaction. Recruitmentof other viral components could occur as Gag continues alongits trafficking pathway. Several studies have documented thepresence of Gag in endocytic compartments, particularly theMVB (11, 32, 36, 38, 47). At least three different models forGag trafficking have been proposed. One school of thought arguesfor the MVB as a Gag trafficking intermediate. Several linesof evidence support this model. HIV-1 Gag as well as infectiousHIV-1 particles localize within the MVB (11, 32, 36, 38, 47).MVBs contain the ESCRT components that are required for virionegress (1). Moreover, Gag binds to the ${delta}$ subunit of AP-3. Whenthis interaction is disrupted, Gag no longer associates withMVBs, and viral particle formation is inhibited (7). A secondmodel postulates that Gag accumulates and buds particles fromtetraspanin-enriched microdomains. These regions of the plasmamembrane are enriched in ESCRT components and tetraspanin proteinsand are thought to be derived from MVBs as a result of fusionwith the plasma membrane (19, 33). A variation on this themeis a third model proposing that newly synthesized Gag transitsdirectly to the plasma membrane, where it assembles and budsparticles (8, 14, 20). The presence of Gag in MVBs is attributedto endocytosis of virus particles, regulated by the viral accessoryprotein Vpu (14, 31). Differences in the data supporting thesethree models and their interpretations have not yet been reconciled.Regardless of the model(s) that ultimately proves to be mostaccurate, the GPC could serve as an early intermediate stepof Gag trafficking in each scenario.

Ubc9 interacts with several retroviral Gags, including thoseof HIV-1, MPMV, and Moloney murine leukemia virus, and eachof these Gag proteins is modified by sumoylation (12, 53, 56).Both MPMV Gag (53) and HIV-1 Gag (this study) colocalize withUbc9 in perinuclear clusters. It is interesting to speculatethat GPCs might provide a staging area for Ubc9 to interactwith and sumoylate Gag. Sumoylation of Gag is not required forassembly, since mutation of the SUMO-1 attachment site on HIV-1or murine leukemia virus Gag does not affect particle production(12, 56). However, the particles released from cells expressingthese mutant Gag proteins exhibit reduced infectivity. It ispossible that during intracellular trafficking Gag acquiresposttranslational modifications that facilitate entry or postentrysteps.

KIF4: motor or scaffold? How does KIF4 function to regulate Gag stability and trafficking?An obvious explanation is that Gag utilizes the motor functionof KIF4 to propel assembly intermediates along MTs. Since KIF4is a plus-end-directed motor, movement would proceed centrifugallyfrom the cell center toward the plasma membrane, the directionof viral egress. It is well established that MTs are also neededto transport MVBs and secretory lysosomes out to the plasmamembrane (3).

Thus, Gag that assembled in these exocytic organelles wouldalso be dependent on MTs for release from the cell. A role forMTs in assembly is supported by the finding that treatment ofT cells with inhibitors of MT polymerization prevents formationof Gag-containing assembly platforms on the plasma membraneand reduces cell-to-cell transmission of HIV-1 across the virologicalsynapse (19). On the other hand, no apparent interaction ofGag with tubulin was noted by fluorescence resonance energytransfer (39), and treatment of cells with the MT-depolymerizingagent nocodazole did not affect particle release (19, 20). However,it is important that cells contain a stable pool of MTs thatare nocodazole resistant (13). In addition, transport eventsthat normally occur along MTs can be bypassed in the absenceof MTs (42). The mechanism underlying this bypass is thoughtto involve centrifugal reorganization of secretory and endocyticmembrane compartments. The lack of an effect of nocodazole istherefore not sufficient to rule out a role for MTs in HIV-1Gag trafficking. Moreover, it was recently shown that stableMTs are required for HIV-1 infection (30). Thus, Gag might preferentiallyutilize the stable pool of MTs for transport, and this selectionmay be enhanced by association with KIF4.

An alternate possibility is that KIF4 acts as a scaffold tocoordinate Gag assembly and movement. Kinesins are often linkedto their cargo via scaffolding or adaptor proteins (2). Interactionwith such scaffolding proteins may promote assembly and/or packagingof protein complexes. Moreover, KIF4 interacts with multiplecellular components, including chromosomes, L1-containing vesicles,retroviral Gag proteins, and the cellular proteins PRC1 andPARP1 (22, 23, 27, 37, 48, 59). Thus, KIF4 could serve as ascaffold to recruit newly synthesized Gag and Gag-associatedmolecules.

It is important that disruption of KIF4 inhibits cytokinesis(59). Thus, the effects we observed on Gag stability and traffickingcould somehow be related to a cytokinesis defect. At present,there are no reports directly linking HIV-1 assembly and cytokinesis.However, two proteins that participate in HIV-1 budding, Tsg101and the ESCRT-associated protein ALIX, are recruited to themidbody during cell division and are required for the abscissionstep of cytokinesis (5, 29). One could postulate that the cytokinesisdefect in KIF4-disrupted cells results in retention of ESCRTcomponents at the midbody and a functional depletion of thesecomponents at sites of HIV-1 budding. We consider this possibilityunlikely. Loss of ESCRT proteins causes tethered HIV-1 particlesto accumulate at the plasma membrane. In contrast, in KIF4-depletedcells, Gag accumulates in the GPC in the perinuclear region.We cannot exclude other indirect effects of KIF4 depletion onGag function.

Model for Gag trafficking and KIF4 function. Based on previously published reports and the data presentedhere, we propose the following model. HIV-1 Gag is synthesizedon soluble polysomes. Upon release into the cytosol, the newlysynthesized protein accumulates in GPCs. Formation of GPCs istransient. Dispersion of GPCs is mediated, either directly orindirectly, by KIF4. When KIF4 function is blocked or when KIF4levels are depleted, Gag begins to accumulate in GPCs. We proposethat GPCs are discrete structures that have a limited capacityfor Gag. When this capacity is reached, newly synthesized Gagproteins can no longer associate in GPCs and are instead targetedfor degradation. This model explains the ability of KIF4-T andKIF4_siRNA to block Gag trafficking at early time points aftersynthesis and also for these treatments to promote downregulationof Gag. Induction of Gag degradation is a unique antiviral strategythat could be exploited as a therapeutic target for interventionduring HIV infection.

ACKNOWLEDGMENTS

We thank Agnes Viale for assistance with qPCR, John Moore foruse of biosafety level 2 laboratory facilities, Alan Hall andTatiana Omelchenko for assistance with microinjection, and RaisaLouft-Nisenbaum for technical assistance. The following reagentwas obtained through the NIH AIDS Research and Reference ReagentProgram, Division of AIDS, NIAID, NIH: HIV-1 p24 monoclonalantibody (183-H12-5C), from Bruce Chesebro and Kathy Wehrly.

This research was supported by NIH grant CA72309 to M.D.R.,NIH MSTP grant GM07739, NCI grant P30-CA08748 (which providespartial support for the Molecular Cytology Core), and AmericanCancer Society grant RSG-06-142-01-CSM to G.K.

FOOTNOTES

* Corresponding author. Mailing address: Cell Biology Program, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, Box 143, New York, NY 10065. Phone: (212) 639-2514. Fax: (212) 717-3317. E-mail: m-resh{at}ski.mskcc.org

Published ahead of print on 6 August 2008.

Supplemental material for this article may be found at http://jvi.asm.org/.

REFERENCES

Top