Current Biology, Vol. 14, 1056–1064, June 22, 2004, 2004 Elsevier Ltd. All rights reserved. DOI 10.1016/j .cub.2004.06.021

ARF1 Regulates Nef-Induced CD4 Degradation

somes and lysosomes [16–19]. We further demonstratedJulien Faure,1,5 Romaine Stalder,2,5

Christelle Borel,2,5 Komla Sobo,2 that the Nef-induced lysosomal targeting of CD4 in-volves Nef-COP interactions [13]. This latter observationVincent Piguet,2,6 Nicolas Demaurex,3

Jean Gruenberg,1,* and Didier Trono2,4,* correlated with the previously reported association ofHIV-1 Nef with �COP [20]. However, although our results1Department of Biochemistry

2 Department of Microbiology and Molecular [13] revealed the importance of residues located in aC-terminal flexible loop of HIV-1 Nef in CD4 late-endoso-Medicine

3 Department of Cell Physiology and Metabolism mal migration, they also indicated that the Nef-�COPinteraction strongly depends on a then-unidentified co-4 Frontiers in Genetics Research Program

University of Geneva factor found in a COP-depleted fraction of the cytosol[13]. Because endosomal COP functions depend on1211 Geneva

Switzerland ARF1 in vitro [21], we reasoned that ARF1 might be thenecessary cofactor. The results presented here indicatethat this is the case.

Summary

ResultsBackground: The HIV Nef protein downregulates CD4through sequential connection with clathrin-coated pits

Nef Late-Endosomal Targeting and Interactionsand the COP1 coatomer, resulting in accelerated endo-with ARF1cytosis and lysosomal targeting.We previously reported that a Nef molecule mutated inResults: Here we report that the small GTPase ARF1two distal acidic residues (NefEE155QQ) inefficiently tar-controls the Nef-induced, COP-mediated late-endoso-geted CD4 to lysosomes [13]. To extend this observa-mal targeting of CD4. We find that Nef binds ARF1 di-tion, we expressed wild-type Nef, NefEE155QQ, or the non-rectly and can recruit the GTPase onto endosomal mem-myristoylated NefG2A mutant with a Myc-tagged form ofbranes. Furthermore, a complex comprising Nef, ARF1,ARF1 in BHK cells, prepared early-and late-endosomaland �COP can be immunoprecipitated from cells ex-fractions on a flotation gradient [22], and analyzed thepressing the viral protein. Residues in a C-terminal loopresulting material by Western blotting (Figure 1). Wild-of the viral protein facilitate both these interactions andtype Nef and NefEE155QQ were both found in early-endo-the targeting of Nef and CD4 to acidic late endosomes,somal fractions (lanes 3 and 7). However, only the wild-whereas other residues primarily involved in mediatingtype protein was detected in late-endosomal fractionsCD4 endocytosis are dispensable for this process. Fi-(lanes 2 and 6). As expected from its failure to bindnally, a dominant-negative ARF1 mutant blocks the mi-membranes, NefG2A did not copurify with endosomesgration of the Nef-CD4 complex to lysosomes.(lanes 10 and 11). ARF1 can interact with several targetConclusions: Our results support a model in whichmembranes, including endosomes, and—not surpris-ARF1 is the immediate downstream partner of Nef foringly—was found associated with early- and late-endo-CD4 lysosomal targeting.somal fractions under all conditions. However, ARF1appeared somewhat more enriched in late endosomesIntroductionof cells coexpressing wild-type Nef (Figure 1, lane 2),perhaps suggesting that it played a role in Nef-inducedThe early protein Nef of human (HIV) and simian (SIV)CD4 late-endosomal targeting. When CD4 was coex-immunodeficiency viruses downregulates CD4, the pri-pressed with wild-type Nef, its accumulation in late en-mary receptor of these viruses. This event is importantdosomes was not detected by this technique (data notfor preserving the infectivity of released virions [1] andshown). This contrasts with approaches in which theproceeds through two consecutive steps. At the plasmareceptor is coupled with antibodies at the cell surfacemembrane, Nef connects the cytoplasmic tail of CD4and then immunofluorescence microscopy is performedwith the adaptor protein complex (AP-2) of clathrin-(see reference [13] and results below). However, in thiscoated pits and the catalytic subunit of a vacuolar AT-latter case, the fraction of CD4 molecules that migratesPase, inducing the receptor to be rapidly endocytosed.from the surface is examined in a dynamic fashion,Then, in early endosomes, Nef interacts with the COP1within moments of its internalization, and the antibodycoatomer, which targets CD4 to lysosomal degradationused to tag the receptor likely survives the degradation[2–15].of CD4 in late endosomes/lysosomes. In contrast, thePrevious studies showed that an endosomal COPbiochemical analysis of purified endosomes providescomplex is involved in transport from early to late endo-information on the steady-state levels of CD4 moleculesassociated with these organelles; these levels are likely

*Correspondence: jean.gruenberg@biochem.unige.ch (J.G.), didier. to be low in late endosomes/lysosomes as a result oftrono@medecine.unige.ch (D.T.)

rapid degradation.5These authors equally contributed to this work.When Nef carrying the HA epitope was coexpressed6Present address: Clinic of Dermatology and Venereology, University

Hospital, 1211 Geneva, Switzerland. with Myc-tagged ARF1, the two proteins could be coim-

ARF1 Regulates Nef-Induced CD4 Degradation1057

Figure 1. Nef Associates with Late Endo-somes in an EE155-Dependent Manner

HA-tagged wild-type Nef or NefEE155QQ, orNefG2A were coexpressed with ARF1-Myc inBHK cells. Early- and late-endosomal mem-branes were separated on a sucrose flotationgradient and analyzed by Western blottingwith antibodies against HA (wild-type Nef andNefEE155QQ), Nef (NefG2A), Rab7 (a marker oflate endosomes, LE), and Rab5 (a marker ofearly endosomes, EE). PNS: postnuclear su-pernatant, HM: heavy membranes.

munoprecipitated with antibodies against HA (Figure Because these results suggested that ARF1 and Nefare part of the same protein complex, we used purified2A). Coimmunoprecipitation was less efficient with a Nef

mutant lacking the diacidic motif (NefEE155QQ) important recombinant versions of the two proteins (His-taggedARF1 and GST-Nef) to determine whether they canfor Nef-induced CD4 late-endosomal targeting [13].interact directly. After coincubation, ARF1 could be cap-tured onto glutathione beads via GST-Nef (Figure 2B).ARF1 also copurified with GST fused with the NefLL165AA

mutant, which is deficient in AP2-dependent inter-nalization [23]. In contrast, ARF1 did not associate toany significant extent with the purified GST-NefEE

155QQ mutant.

Nef-Mediated Recruitment of ARF1on EndosomesNef is tethered to the membrane by N-terminal myristoy-lation and might be able to recruit ARF1 onto endo-somes. However, when bound to GTP, ARF1 can alsointeract with the bilayer by exposing an N-terminal myri-stoyl head followed by a stretch of charged residuespresumably involved in phospholipid binding (a struc-tural modification known as a myristoyl switch) [24, 25].We therefore used an ARF1 mutant (ARF1�17) that can-not associate spontaneously with the lipid bilayer as aresult of a 17 residue N-terminal deletion that eliminatesthe myristoylation site and the stretch of charged aminoacids. After coexpression of CD4 and wild-type Nef orNefEE155QQ, endosomal membranes were prepared byflotation gradient [26] and incubated with recombinantARF1�17 in vitro (Figure 3A). ARF1�17 was selectivelyrecruited onto endosomes of cells expressing wild-typeNef. A small degree of ARF1�17 recruitment was ob-served in control cells, suggesting that this moleculemay also bind endogenous endosomal proteins. Of note,the same background level of binding was detected withendosomes of cells expressing NefEE155QQ.

Figure 2. Nef Interacts with ARF1 in an EE155-Dependent Manner Because ARF1 can be recruited by membrane-associ-(A) Lentiviral vectors expressing HA-tagged wild-type Nef or ated Nef and is involved in early- to late-endosomalNefEE155QQ, CD4 and Myc-tagged ARF1 were introduced into BHK transport, we investigated whether Nef was able to bindcells. Top: after immunoprecipitation with antibodies against HA, to the different forms of the GTPase. We first madesamples were analyzed by SDS-PAGE and Western blotting with

use of the well-characterized T31N dominant-negativeantibodies against HA and Myc. Bottom: the Nef and ARF1 contentsARF1mutation, which prevents GTP loading. After coex-of the samples were analyzed by SDS-PAGE and Western blotting.

Numbers represent the results of quantification, the signal obtained pression in vivo, ARF1T31N, much like wild-type ARF1,for each protein in the wild-type Nef-containing sample being given could be coimmunoprecipitated with wild-type Nef butthe arbitrary value of 1. not with NefEE155QQ (Figure 3B). Similarly, recombinant(B) E. coli-produced His-tagged ARF1 and the indicated GST-Nef ARF1T31N interacted in vitro with GST-Nef but not withproteins were coincubated in vitro before capture onto gluthatione

GST-NefEE155QQ (Figure 3C). Because the T31N mutationbeads and analysis by SDS-PAGE and Western blotting with anti-inhibits GTP binding, and thus also the GTP-dependentHis6 antibodies (top). Amounts of GST-Nef present on the blotting

membrane were determined by Ponceau red staining (bottom). myristoyl switch of ARF1, our observations that Nef in-

Current Biology1058

teracts with both ARF1T31N and ARF1�17 are consistent.Nef could also be coimmunoprecipitated with ARF1Q71L,an ARF1 mutant stabilized in the GTP bound, activeform (not illustrated), indicating that the association be-tween the viral protein and ARF1 is not affected by thenature of the nucleotide bound to the GTPase.

Nef-ARF1-�COP Trimolecular ComplexBecause Nef can recruit ARF1 to endosomal mem-branes and because previous studies demonstrated that(1) ARF1 recruits the � subunit of the COP1 complex[17, 27], (2) �COP is involved in endosomal transport[16–19], and (3) Nef can associate with �COP [13, 20],we asked whether Nef can form a ternary complex withendogenous ARF1 and �COP. For this, extracts of 293Tcells expressing a Myc-tagged form of �COP (�COP-Myc) with or without HA-tagged forms of wild-type orEE155QQ HIV-1 Nef (Nef-HA and NefEE155QQ-HA, respec-tively) were subjected to immunoprecipitation with aMyc-specific antibody (Figure 4A). Detectable amountsof endogenous ARF1 could be immunoprecipitated with�COP-Myc only in the presence of Nef, and the EE155QQmutation decreased the amount of Nef and ARF1 associ-ated with �COP by about 35% in this type of assay. Aswell, Nef-HA and endogenous �COP could be immuno-precipitated with antibodies against the Myc epitopeof ARF1-Myc in an EE155-enhanced fashion (Figure 4B).These results lend support to the existence of a trimo-lecular complex containing Nef, ARF1, and �COP, theformation of which is influenced by the Nef acidic dipep-tide. The persistence of a significant degree of Nef-ARF1-�COP coimmunoprecipitation with NefEE155QQ

suggests that independent interactions take place be-tween Nef and the other two proteins and that some ofthese interactions, namely between Nef and �COP, arenot influenced by this particular sequence motif.

ARF1 and Nef-Induced Transport of CD4 to AcidicLate-Endosomal CompartmentsWe investigated further the possible role of ARF1 in Nef-induced CD4 trafficking to acidic late-endocytic com-partments. We first measured the rates of CD4 endocy-tosis. In the absence of Nef, CD4 internalization wasonly marginally, if at all, affected by overexpression ofeither wild-type ARF1 or ARF1T31N. As expected, NefFigure 3. Nef Recruits ARF1 on Endosomal Membranes

overexpression greatly stimulated CD4 endocytosis,(A) HA-tagged wild-type Nef or NefEE155QQ and CD4 were coex-pressed in BHK cells. Endosomal membranes were prepared and and this increased uptake was largely preserved whenincubated in vitro with recombinant, purified His-tagged ARF1�17. either form of the GTPase was expressed (not illus-Then, membranes were retrieved by flotation in a sucrose gradient trated). We then examined CD4 late-endosomal trans-and analyzed as above, with antibodies against HA (Nef) or His6 port (Figure 5A). To identify late endosomes unambigu-(ARF1�17).

ously, we used lyso-bisphosphatidic acid (LBPA) as a(B) HA-tagged wild-type Nef or NefEE155QQ CD4 and Myc-taggedmarker [28]. When expressed alone, CD4 was only mar-ARF1T31N were coexpressed in 293T cells. After crosslinking and

immunoprecipitation with antibodies against Nef, samples were an- ginally detected within late endosomes containing LBPAalyzed with antibodies against HA and Myc. Numbers under bands (�10% of the internalized CD4 molecules colocalizedrepresent the results of quantification, values obtained for each with LBPA). In contrast, in the presence of Nef, endocy-protein in the wild-type Nef sample being assigned the arbitrary

tosed CD4 molecules were efficiently transported to thisvalue of 1.compartment (44% colocalization). The Nef-induced(C) Recombinant His-tagged ARF1T31N and wild-type or mutantlate-endosomal transport of CD4 was much less efficientGST-Nef proteins were produced in E. coli, purified, mixed, and

analyzed with antibodies against His. in cells expressing NefEE155QQ (19% colocalization).When wild-type ARF1 was overexpressed, some CD4appeared to be sequestered in regions adjacent to theplasma membrane, often in clusters of vesicles, presum-

ARF1 Regulates Nef-Induced CD4 Degradation1059

tosis of CD4 was slow in the presence of the AP binding-defective NefLL165AA mutant, internalized receptor mole-cules were targeted to late endosomes (Figure 5B, top),correlating this mutant’s ability to bind ARF1. A Nefmutant altered in the downstream acidic motif DD174,previously demonstrated to be crucial for binding thecatalytic subunit of a vacuolar ATPase and for accelerat-ing CD4 endocytosis [15], induced a similar phenotype(Figure 5B, bottom). With both of these mutants, eventhough a large fraction of CD4 molecules were retainedat the plasma membrane, those internalized clusteredin the paranuclear area, where they strongly colocalizedwith LBPA. This contrasted with the “starry sky” patternof CD4 distribution induced by NefEE155QQ; in that pat-tern, the majority of receptor molecules associated withearly/recycling endosomes [13]. These findings indicatethat the binding of Nef to the adaptor protein complex,a process in which the LL165 motif plays a crucial role,and to the vacuolar ATPase, another proposed mediatorof Nef-induced CD4 endocytosis, is not essential forCD4 late-endosomal targeting.

To monitor the pH of vesicles containing internalizedCD4, we labeled the receptor with FITC-coupled anti-CD4 antibody at the cell surface and allowed endocyto-sis to proceed at 37�C for 60 min. The pH of endosomescontaining the internalized antibody was then quantifiedby fluorescence microscopy [13, 29] (Figure 5C). In theabsence of Nef, CD4 was mostly found in vesicles witha mildly acidic pH (pH � 6.3), corresponding to that ofearly endosomes (not shown). In contrast, expressionof wild-type Nef caused CD4 to distribute within vesicleswith a low pH, consistent with that of late endosomesand lysosomes. Transport of CD4 to acidic compart-ments was inhibited when the diacidic motif requiredfor ARF1 binding was inactivated in the NefEE155QQ mu-tant, as previously observed [13]. Overexpression ofwild-type ARF1 did not prevent CD4 transport to acidiccompartments. However, consistent with our light mi-croscopy analysis (Figure 5A), the retention of someCD4 at or near the plasma membrane of cells overex-Figure 4. �COP-ARF1-Nef Interactionspressing the GTPase accounted for a significant fraction(A) HA-tagged Nef (wild-type or EE155QQ), CD4, and Myc-taggedof CD4-containing structures with a pH between 7 and�COP were coexpressed in 293T cells. Samples of extracts were

analyzed by SDS-Page and Western blotting with antibodies against 7.5. In cells producing the dominant-negative ARF1 mu-HA, Myc and ARF without (top) or with (bottom) prior immunoprecipi- tant T31N, wild-type Nef was no longer able to targettation with Myc-specific antibodies. CD4 to acidic vesicles. Of note, the acidification capacity(B) Same type of experiment, but with extracts of cells expressing of late-endocytic compartments was not compromised,Nef-HA and ARF1-Myc being immunoprecipitated with Myc-specific

as assessed with Lysotracker red (not shown). Together,antibodies and with antibodies against HA, Myc, and endogenousthese experiments support a model in which ARF1 is�COP (top and bottom) and ARF (top) being used for the Western

blot. involved in the Nef-induced transport of CD4 to acidicNumbers represent the results of quantification, the signals obtained late endosomes and lysosomes.in the sample containing wild-type Nef being given the arbitraryvalue of 1 for each protein. In (B), the �COP Western blot performedon the anti-Myc immunoprecipitates detects a background band Colocalization of Nef, ARF1, and �COP(top), also quantified as an internal control.

in Early EndosomesWe then examined the possible colocalization of Nef,ARF1, and endosomes. For this, we made use of theably reflecting the increased activity of the GTPase

(arrows). However, CD4 still migrated to LBPA-con- chimeric protein 44Nef, which comprises the extra-cellular and transmembrane domains of CD4 fused withtaining vesicles (20% colocalization). In contrast, this

process was strongly inhibited upon expression of Nef. HeLa cells were transfected with both 44Nef or44NefEE155QQ and ARF1-gfp or ARF1T31N-gfp. Cell sur-ARF1T31N, with the receptor remaining essentially within

LBPA-negative peripheral endosomes (10% colocaliza- face CD4 determinants were labeled with a specific anti-body, and endocytosis was allowed to proceed for 40tion). CD4 migration was also inhibited after expression

of ARF1Q71L (data not shown). Of note, although endocy- min. Cells were then fixed and incubated with antibodies

Current Biology1060

Figure 5. Transport to LBPA-Containing Acidic Late Endosomes

(A) CD4 molecules at the surface of transfected Hela cells were labeled with T4-4 monoclonal antibody and allowed to internalize at 37�C for60 min. The distribution of antibody-tagged CD4 molecules was then analyzed by confocal microscopy, and cells were further labeled withantibodies against the late-endosomal lipid lysobisphosphatidic acid followed by Alexa 568- (for LBPA) and 488- (for CD4) conjugated secondaryantibodies. For each transfection, one representative confocal plane, representative of more than 100 cells examined, is shown. Percentages

ARF1 Regulates Nef-Induced CD4 Degradation1061

against the early-endosomal marker EEA1 before exami- more, our data indicate that, although the EE155QQ muta-tion severely affects Nef-induced CD4 lysosomal tar-nation by indirect immunofluorescence and confocalgeting, it only partially abolishes the association of Nefmicroscopy (Figure 6). Wild-type 44Nef, but not thewith �COP. This suggests that several interactions may44NefEE155QQ mutant, showed a significant degree oftake place within the Nef-ARF1-�COP complex, not all ofcolocalization with ARF1-gfp and EEA1 (Figures 6A andwhich are affected by the Nef acidic dipeptide mutation.6B). In the absence of Nef, ARF1T31Ngfp was fully cyto-

The C-terminal loop of Nef is involved in recruitingsolic and leaked from the methanol-fixed cells (not illus-the proposed cellular mediators both of CD4 endocyto-trated). As expected from the results of our biochemicalsis, i.e., the adaptor complex and the vacuolar ATPase,analyses, 44Nef weakly stabilized the mutant form ofand of CD4 late-endosomal targeting, i.e the COP1 co-the GTPase onto endosomes (Figure 6C). In contrast,atomer. The bindings of these two sets of mediatorsno such effect was observed with 44NefEE155QQ because,are most likely mutually exclusive because of spacein this case, most of the mutant ARF1 molecules leakedconstraints. This fits well with the sequential involve-out of the fixed cells (Figure 6D). As well, colocalizationment of these cellular complexes in Nef-mediated CD4of Nef, ARF1-gfp, and �COP was dependent on thedownregulation; adaptins and the vacuolar ATPase arepreservation of the Nef EE155 sequence (Figures 6E andrecruited at the plasma membrane for CD4 internaliza-6F). Finally, in the presence of brefeldin A, a drug thattion via clathrin-coated pits, and the COP1 coatomer isprevents the formation of ARF-GTP, CD4 late-endoso-recruited in early endosomes for CD4 lysosomal tar-mal targeting was inhibited, and the colocalization ofgeting. However, there is not a complete overlap in the44Nef, ARF1gfp, and early-endosomal markers was lostresidues involved in binding either complex or in the(not illustrated). Together, these results strongly supportfunctional consequences of their mutation. For instance,a model in which ARF1 is involved in the Nef-inducedmutating EE155 in Nef does not alter Nef binding totransport of CD4 to an acidic late compartment.adaptins or the acceleration of CD4 endocytosis butinterferes with CD4 lysosomal targeting [13]. Recipro-

Discussioncally, the LL165 and DD174 mutants can target CD4 to lateendosomes, but without initially inducing CD4 internal-

Four lines of evidence point to ARF1 as the immediate ization.downstream partner of Nef for the lysosomal targeting Our attempts to generate a multimolecular complexof CD4. First, Nef binds ARF1 directly and is capable comprising Nef, ARF1, and �COP by using recombinantof recruiting the GTPase onto endosomal membranes. ARF1 synthesized in E. coli have been unsuccessful,Second, these interactions are influenced by residues perhaps because of posttranslational ARF1 modifica-in a C-terminal flexible loop of Nef; these residues are tions not recapitulated in vitro. Alternatively, such aalso important for Nef association with late endosomes complex may require other factors or may be very tran-and targeting of CD4 molecules to this compartment. sient. Similarly, in the biosynthetic pathway, ARF1 isThird, a complex comprising Nef, ARF1, and �COP can known to bind coatomer in close proximity to the cargobe detected in cells expressing the viral protein. Fourth, binding site [31], but the nature of possible complexesexpression of a dominant-negative ARF1 mutant or of cargo molecules, ARF1, and coatomer is not clear.treatment with brefeldin A blocks the lysosomal tar- Some biosynthetic cargo molecules were proposed togeting of CD4. These data complement our previous modulate ARF1 GTPase activity [32], but whether thisdemonstration of an EE155-promoted, yet largely indirect modulation depends on coatomer remains an unsolvedinteraction between Nef and �COP and of the COP1- issue [33–35].dependent, Nef-induced lysosomal transport of CD4 It is tempting to postulate that Nef, like the ARF1[13]. Altogether, our results support a model in which exchange factor ARNO [36], stabilizes ARF1 at the mem-ARF1 connects the CD4-Nef complex with this coat brane independently of the nucleotide state of thestructure and thereby plays a crucial role in the late- GTPase. Although we were unable to reveal GDP-GTPendosomal targeting of CD4 by the viral protein. exchange activity for Nef toward ARF1 in vitro, we can-

The Nef EE155 sequence was found not to influence not rule out the possibility that the viral protein exertsthe binding of Nef to �COP in the yeast two-hybrid this action in vivo, for instance if it requires that Nef besystem or of GST-Nef to an in vitro-translated C-terminal bound to CD4. Alternatively, Nef may prime ARF1 forfragment of �COP [30]. However, Nef-mediated recruit- subsequent GTP exchange. The dominant interferencement of COP1 constituents is most likely indirect and imposed by ARF1T31N suggests that, although dispens-depends on a critical cofactor [13]. The present work able for Nef binding, ARF1 activation is required for CD4-

Nef transport to acidic late endosomes. These observa-strongly suggests that ARF1 is this cofactor. Further-

represent the proportion of CD4-specific green signal overlapping with the 6C4-specific red signal. In cells overexpressing ARF1, some CD4is sequestered in the juxtamembrane region (arrows). However, CD4-LBPA colocalization is far more pronounced than in the presence ofARF1T31N.(B) The same experiment with the endocytosis-defective NefLL165AA NefDD174AA mutants. In these cases, a large fraction of CD4 moleculesstay at the plasma membrane, but those internalized are predominantly found colocalizing with LBPA in the perinuclear region.(C) CD4 � Nef or NefEE155QQ, CD4 � Nef � ARF1, or CD4 � Nef � ARF1T31N were coexpressed in HeLa cells, and cell-surface CD4 moleculeswere labeled with anti-CD4 antibody conjugated to FITC. After reincubation at 37�C for 60 min, FITC fluorescence was used for measuringthe pH of vesicles containing the internalized antibody, as previously described [13]. Mean pH is indicated in each case, with the number inparentheses indicating the number of vesicles included in each analysis.

Current Biology1062

Figure 6. Colocalization of Nef and ARF1 on Early Endosomes

44Nef (A, C, and E) or 44NefEE155QQ (B, D, and F) was coexpressed with ARF1gfp (A, B, E, and F) or ARF1T31Ngfp (C and D) in HeLa cells. CD4molecules at the surface of the cells were labeled with T4-4 monoclonal antibody and allowed to internalize at 37�C for 40 min. Cells werefurther labeled with antibodies against the early-endosomal marker EEA1 (A–D) or �COP (E and F) and examined by confocal microscopy.

EH1 clones, all obtained from the AIDS Research and Referencetions fit well with our previous findings that ARF1 isReagent Program, Division of AIDS, NIH), anti-Nef polyclonal [37],involved in the in vitro formation of multivesicular trans-anti-LBPA (6C4) [28], anti-�COP (Sigma, clone maD), anti-EEA1port intermediates destined for late endosomes [21].(Transduction Lab), and anti-ARF (which recognizes ARF1, 3, 5, and

Because ARF1 recruits endosomal COPs [21], which 6) (1D9, Alexis).are required at the same transport step [16–19], andbecause these are also involved in Nef:CD4 downregula- Plasmidstion [13], an attractive model is that Nef acts by nucleat- The ARF1 cDNA (gift from V. Faundez, UCSF, San Fransisco) was

amplified by PCR to encode an ARF1-Myc fusion protein and wasing the transport machinery via ARF1 binding. By hi-inserted into a pCB6 vector [38]. ARF1T31N-Myc and ARF1Q71L-Mycjacking ARF1, Nef may have evolved an efficient way towere generated with the Quickchange Stratagene kit. For obtainingovercome endosomal controls in order to divert CD4ARF1-GFP fusion expression vectors, the appropriate cDNA was

into the pathway leading to its degradation in late endo- subcloned into pEGFP-N (Clontech). For generating an ARF1-His6

somes and lysosomes. fusion protein, ARF1 cDNA was introduced into pET20b (Novagen)vector. ARF1�17-His6 was obtained by mutagenesis of the first ATGcodon into a TTG codon. The wild-type Nef encoding sequenceExperimental Procedureswas derived from the HIV1 R7 recombinant clone [39], with twodifferences relative to the published sequence (E151D and E158K).Cell Culture and Immunological Reagents

Baby hamster kidney cells (BHK-21), Hela cells, and 293T cells (pro- CMV-based expression plasmids producing CD4, both wild-typeand mutant forms of Nef, and CD4-Nef chimeras were describedvided by G. Nolan, Stanford University) were grown and maintained

in DMEM medium with 10% FCS, 100 �g/ml of penicillin and strepto- [2, 13]. In some vectors, three copies of the influenza hemagglutininepitope were added to Nef C terminus by oligo-directed mutagene-mycin, and 2 mM glutamine. The following antibodies were used:

anti-His6 C-ter (Invitrogen), anti-HA (3F10, Roche; HA.11, Covance), sis. BHK cells stably expressing Nef and CD4 were obtained bylentiviral vector-mediated transduction with a second-generationanti-Myc-tag (9E10, Roche), anti-CD4 PE or FITC-conjugated

(MT310, Dako), anti-CD4 (T4-4), anti-Nef (AE6, HIV1 JR-CSF, and packaging system [40]. For generating the pHR�CMV-CD4 vector,

ARF1 Regulates Nef-Induced CD4 Degradation1063

the GFP sequence from pHR�CMV-GFP [41] was replaced with the AcknowledgmentsCD4 sequence from pCMX-CD4 [2]. The pHR�EF-Nef and pHR�EF-ARF1 vectors express the nef and ARF1 reading frames downstream We thank Feng Gu for her participation in the initial phase of this

study, Joe Curran for kindly providing expert advice, and Marie-of the elongation factor 1� promoter [42]. For generating the bacte-rial expression vectors pGEX-Nef wt and pGEX-NefEE155QQ, the cor- Helene Beuchat and Monique Beulet for technical assistance. This

work was supported by grants from the Swiss National Scienceresponding nef sequence was inserted into pGEX-2T (Amersham-Pharmacia Biotech). The �COP-Myc expression vector was a gift Foundation (to D.T., J.G., and N.D.), from the International Human

Frontier Science Program (to J.G.), and from the Gabriella Giorgi-from T. Kreis.Cavaglieri Foundation (to D.T.).

Lentiviral Vectors Production Received: May 29, 2003Lentiviral vector stocks were generated by transient transfection of Revised: March 3, 2004293T cells as previously described [43, 44], with pCMV�R8.91 and Accepted: March 30, 2004pMD-G as packaging and envelope plasmids, respectively. Published: June 22, 2004

ReferencesProtein StudiesGST-Nef fusion proteins were purified with glutathione-Sepharose

1. Lama, J., Mangasarian, A., and Trono, D. (1999). Cell-surfaceaccording to the manufacturer’s guidelines (Pharmacia). His-taggedexpression of CD4 reduces HIV-1 infectivity by blocking Envfusion proteins were expressed in the BL21 (DE3) E. coli strainincorporation in a Nef- and Vpu-inhibitable manner. Curr. Biol.and purified on Ni-NTA agarose beads (Quiagen). The eluate was9, 622–631.concentrated (Biomax membrane, Millipore) and purified on a Seph-

2. Aiken, C., Konner, J., Landau, N.R., Lenburg, M.E., and Trono, D.adex G75 column in 10 mM HEPES (pH 7.4), 5 mM MgCl2, and(1994). Nef induces CD4 endocytosis: requirement for a critical150 mM NaCl. The eluate was pooled and concentrated. The GST-dileucine motif in the membrane-proximal CD4 cytoplasmic do-capture assay was performed as described [13]. The procedure formain. Cell 76, 853–864.measuring ARF1�17-His6 binding to endosomal membranes was

3. Bresnahan, P.A., Yonemoto, W., Ferrell, S., Williams-Herman,previously described [21]. Endosomal fractions [26] were preparedD., Geleziunas, R., and Greene, W.C. (1998). A dileucine motiffrom cells pretreated with 20 �g/ml brefeldin A for 1 hr at 37�C soin HIV-1 Nef acts as an internalization signal for CD4 downregu-they would be depleted from Golgi membranes [21, 45]. Subcellularlation and binds the AP-1 clathrin adaptor. Curr. Biol. 8, 1235–endosomal fractionation was performed as described [46]. In brief,1238.transfected BHK cells were homogeneized, and a post-nuclear su-

4. Rhee, S.S., and Marsh, J.W. (1994). Human immunodeficiencypernatant was prepared. This post-nuclear supernatant was ad-virus type 1 Nef-induced down-modulation of CD4 is due tojusted to 40.6% sucrose, 3 mM imidazole (pH 7.4), loaded at therapid internalization and degradation of surface CD4. J. Virol.bottom of an SW41 tube, and overlaid sequentially with 35% and68, 5156–5163.25% sucrose solutions in 3 mM imidazole (pH 7.4) and then with

5. Rhee, S.S., and Marsh, J.W. (1994). HIV-1 Nef activity in murinehomogenization buffer (HB; 8,5% sucrose, 3 mM imidazole [pH 7.4]).T cells. CD4 modulation and positive enhancement. J. Immunol.The gradient was centrifuged for 90 min at 35,000 rpm with an SW41152, 5128–5134.rotor. Early- and late-endosomal fractions were collected at the

6. Craig, H.M., Pandori, M.W., and Guatelli, J.C. (1998). Interaction35%/25% and 25%/HB interfaces, respectively. For Nef-specificof HIV-1 Nef with the cellular dileucine-based sorting pathwayimmunoprecipitations (IP), 100–200 �g protein of postnuclear super-is required for CD4 down-regulation and optimal viral infectivity.natant was mixed in IP buffer (0.5% NP-40, 0.25% sodium deoxy-Proc. Natl. Acad. Sci. USA 95, 11229–11234.cholate, 0.05% SDS, 4% sucrose, 1.5 mM imidazole [pH 7.4], PBS

7. Foti, M., Mangasarian, A., Piguet, V., Lew, D.P., Krause, K.H.,0.5) with 0.5 �g of a mix of anti-Nef monoclonal antibodies (AE6,Trono, D., and Carpentier, J.L. (1997). Nef-mediated clathrin-EH1, HIV1 JR-CSF) for 2 hr at room temperature, followed by 30coated pit formation. J. Cell Biol. 139, 37–47.min onto A/G agarose beads (Santa Cruz). In most experiments,

8. Greenberg, M.E., Iafrate, A.J., and Skowronski, J. (1998). The293T postnuclear supernatant was incubated with 3 mM DTSSPSH3 domain-binding surface and an acidic motif in HIV-1 Nef(Pierce) during 30 min at room temperature. The crosslinker wasregulate trafficking of class I MHC complexes. EMBO J. 17,inhibited by the addition of 20 mM Tris over 15 min. The IP was2777–2789.then performed in 25 mM NaCl, 1% Triton, and 0.5% SDS, as de-

9. Lock, M., Greenberg, M.E., Iafrate, A.J., Swigut, T., Muench,scribed above. For HA-directed IP, extracts were incubated withJ., Kirchhoff, F., Shohdy, N., and Skowronski, J. (1999). TwoDynabeads (Dynal) carrying HA-specific antibody. Myc-specific im-elements target SIV Nef to the AP-2 clathrin adaptor complex,munoprecipitations were performed with 250 �g of postnuclear su-but only one is required for the induction of CD4 endocytosis.pernatant (lysis in 10 mM KCl, 20 mM HEPES [pH 6.5], and 0.05%EMBO J. 18, 2722–2733.Triton, then addition of 150 mM NaCl), incubated with 50 �M DTSSP

10. Mangasarian, A., Foti, M., Aiken, C., Chin, D., Carpentier, J.L.,(Pierce) for 30 min at RT. The crosslinker was inhibited by the addi-and Trono, D. (1997). The HIV-1 Nef protein acts as a connectortion of 100 mM glycine for 1.5 hr at RT. The Myc immunoprecipita-with sorting pathways in the Golgi and at the plasma membrane.tions were performed in RIPA buffer with c-Myc monoclonal anti-Immunity 6, 67–77.body-agarose beads (Clontech) for 2.5 hr at RT, followed by one 10

11. Mangasarian, A., and Trono, D. (1997). The multifaceted role ofmin wash in 0.5 M LiCl and three 10 min washes in RIPA buffer.HIV Nef. Res. Virol. 148, 30–33.

12. Piguet, V., Chen, Y.L., Mangasarian, A., Foti, M., Carpentier,J.L., and Trono, D. (1998). Mechanism of Nef-induced CD4 en-Trafficking Studies

Immunofluorescence and confocal microscopy analyses were per- docytosis: Nef connects CD4 with the mu chain of adaptorcomplexes. EMBO J. 17, 2472–2481.formed as previously described [13] with a Zeiss LSM510 micro-

scope. Colocalization was measured with Metamorph 4.6 software 13. Piguet, V., Gu, F., Foti, M., Demaurex, N., Gruenberg, J., Carpen-tier, J.-L., and Trono, D. (1999). Nef-induced CD4 degradation:(Universal Imaging Corporation, Westchester, PA), with a 35% and

30% intensity threshold on the red (LBPA, Alexa 568) and green a diacidic-based motif in Nef functions as a lysosomal targetingsignal through the binding of b-COP in endosomes. Cell 97,(CD4, Alexa 488) channels, respectively. The pH measurements were

carried out as described [13] after endocytosis of a FITC-conjugated 63–73.14. Schwartz, O., Marechal, V., Danos, O., and Heard, J.M. (1995).CD4 antibody (clone MT310) for 1 hr at 37�C. 44Nef, �COP, and

ARF1gfp colocalization were performed with either anti-CD4 anti- Human immunodeficiency virus type 1 Nef increases the effi-ciency of reverse transcription in the infected cell. J. Virol. 69,body (806) followed by Alexa 568 (red) and anti-EEA1 antibody

(Transduction Lab) or anti-�COP antibody (Sigma) followed by Alexa 4053–4059.15. Lu, X., Yu, H., Liu, S.H., Brodsky, F.M., and Peterlin, B.M. (1998).633 (blue).

Current Biology1064

Interactions between HIV1 Nef and vacuolar ATPase facilitate 35. Lanoix, J., Ouwendijk, J., Stark, A., Szafer, E., Cassel, D., Dej-gaard, K., Weiss, M., and Nilsson, T. (2001). Sorting of Golgithe internalization of CD4. Immunity 8, 647–656.

16. Daro, E., Sheff, D., Gomez, M., Kreis, T., and Mellman, I. (1997). resident proteins into different subpopulations of COPI vesicles:a role for ArfGAP1. J. Cell Biol. 155, 1199–1212.Inhibition of endosome function in CHO cells bearing a tempera-

ture-sensitive defect in the coatomer (COPI) component epsi- 36. Beraud-Dufour, S., Paris, S., Chabre, M., and Antonny, B. (1999).Dual interaction of ADP ribosylation factor 1 with Sec7 domainlon-COP. J. Cell Biol. 139, 1747–1759.

17. Whitney, J.A., Gomez, M., Sheff, D., Kreis, T.E., and Mellman, and with lipid membranes during catalysis of guanine nucleotideexchange. J. Biol. Chem. 274, 37629–37636.I. (1995). Cytoplasmic coat proteins involved in endosome func-

tion. Cell 83, 703–713. 37. Aiken, C., Krause, L., Chen, Y.L., and Trono, D. (1996). Muta-tional analysis of HIV-1 Nef: identification of two mutants that are18. Aniento, F., Gu, F., Parton, R., and Gruenberg, J. (1996). An

endosomal bcop is involved in the pH-dependent formation of temperature-sensitive for CD4 downregulation. Virology 217,293–300.transport vesicles destined for late endosomes. J. Cell Biol.

133, 29–41. 38. Brewer, C.B. (1994). Cytomegalovirus plasmid vectors for per-manent lines of polarized epithelial cells. Methods Cell Biol. 43,19. Gu, F., Aniento, F., Parton, R., and Gruenberg, J. (1997). Func-

tionnal dissection of COP-I subunits in the biogenesis of multi- 233–245.vesicular endosomes. J. Cell Biol. 139, 1183–1195. 39. Aiken, C., and Trono, D. (1995). Nef stimulates human immuno-

20. Benichou, S., Bomsel, M., Bodeus, M., Durand, H., Doute, M., deficiency virus type 1 proviral DNA synthesis. J. Virol. 69, 5048–Letourneur, F., Camonis, J., and Benarous, R. (1994). Physical 5056.interaction of the HIV-1 Nef protein with beta-COP, a component 40. Zufferey, R., Nagy, D., Mandel, R.J., Naldini, L., and Trono, D.of non-clathrin-coated vesicles essential for membrane traffic. (1997). Multiply attenuated lentiviral vector achieves efficientJ. Biol. Chem. 269, 30073–30076. gene delivery in vivo. Nat. Biotechnol. 15, 871–875.

21. Gu, F., and Gruenberg, J. (2000). ARF1 regulates pH-dependent 41. Zufferey, R., Dull, T., Mandel, R.J., Bukovsky, A., Quiroz, D.,COP functions in the early endocytic pathway. J. Biol. Chem. Naldini, L., and Trono, D. (1998). Self-inactivating lentivirus vec-275, 8154–8160. tor for safe and efficient in vivo gene delivery. J. Virol. 72, 9873–

22. Bomsel, M., Parton, R., Kuznetsov, S.A., Schroer, T.A., and 9880.Gruenberg, J. (1990). Microtubule- and motor-dependent fusion 42. Salmon, P., Kindler, V., Ducrey, O., Chapuis, B., Zubler, R.H.,in vitro between apical and basolateral endocytic vesicles from and Trono, D. (2000). High-level transgene expression in humanMDCK cells. Cell 62, 719–731. hematopoietic progenitors and differentiated blood lineages

23. Craig, H.M., Reddy, T.R., Riggs, N.L., Dao, P.P., and Guatelli, after transduction with improved lentiviral vectors. Blood 96,J.C. (2000). Interactions of HIV-1 nef with the mu subunits of 3392–3398.adaptor protein complexes 1, 2, and 3: role of the dileucine- 43. Dull, T., Zufferey, R., Kelly, M., Mandel, R.J., Nguyen, M., Trono,based sorting motif. Virology 271, 9–17. D., and Naldini, L. (1998). A third-generation lentivirus vector

24. Antonny, B., Beraud-Dufour, S., Chardin, P., and Chabre, M. with a conditional packaging system. J. Virol. 72, 8463–8471.(1997). N-terminal hydrophobic residues of the G-protein ADP- 44. Naldini, L., Blomer, U., Gallay, P., Ory, D., Mulligan, R., Gage,ribosylation factor-1 insert into membrane phospholipids upon F.H., Verma, I.M., and Trono, D. (1996). In vivo gene delivery andGDP to GTP exchange. Biochemistry 36, 4675–4684. stable transduction of nondividing cells by a lentiviral vector.

25. Goldberg, J. (1998). Structural basis for activation of ARF Science 272, 263–267.GTPase: mechanisms of guanine nucleotide exchange and 45. Rojo, M., Pepperkok, R., Emery, G., Kellner, R., Stang, E., Parton,GTP-myristoyl switching. Cell 95, 237–248. R.G., and Gruenberg, J. (1997). Involvement of the transmem-

26. Huber, L., Fialka, I., Paiha, K., Hunzinker, W., Sacks, D.B., Bahler, brane protein p23 in biosynthetic protein transport. J. Cell Biol.M., Way, M., Gagescu, R., and Gruenberg, J. (2000). Both cal- 139, 1119–1135.modulin and the unconventional myosin myr4 regulate mem- 46. Aniento, F., Emans, N., Griffiths, G., and Gruenberg, J. (1993).brane trafficking along the recycling pathway of MDCK cells. Cytoplasmic dynein-dependent vesicular transport from earlyTraffic 1, 494–503. to late endosomes. J. Cell Biol. 123, 1373–1388.

27. Fischer, K.D., Helms, J.B., Zhao, L., and Wieland, F.T. (2000).Site-specific photocrosslinking to probe interactions of Arf1with proteins involved in budding of COPI vesicles. Methods20, 455–464.

28. Kobayashi, T., Stang, E., Fang, K.S., de Moerloose, P., Parton,R.G., and Gruenberg, J. (1998). A lipid associated with the anti-phospholipid syndrome regulates endosome structure/func-tion. Nature 392, 193–197.

29. Gagescu, R., Demaurex, N., Parton, R.G., Hunziker, W., Huber,L.A., and Gruenberg, J. (2000). The recycling endosome ofMadin-Darby canine kidney cells is a mildly acidic compartmentrich in raft components. Mol. Biol. Cell 11, 2775–2791.

30. Janvier, K., Craig, H., Le Gall, S., Benarous, R., Guatelli, J.,Schwartz, O., and Benichou, S. (2001). Nef-induced CD4 down-regulation: a diacidic sequence in human immunodeficiencyvirus type 1 Nef does not function as a protein sorting motifthrough direct binding to beta-COP. J. Virol. 75, 3971–3976.

31. Zhao, L., Helms, J.B., Brunner, J., and Wieland, F.T. (1999).GTP-dependent binding of ADP-ribosylation factor to coatomerin close proximity to the binding site for dilysine retrieval motifsand p23. J. Biol. Chem. 274, 14198–14203.

32. Springer, S., Spang, A., and Schekman, R. (1999). A primer onvesicle budding. Cell 97, 145–148.

33. Goldberg, J. (2000). Decoding of sorting signals by coatomerthrough a GTPase switch in the COPI coat complex. Cell 100,671–679.

34. Szafer, E., Pick, E., Rotman, M., Zuck, S., Huber, I., and Cassel,D. (2000). Role of coatomer and phospholipids in GTPase-acti-vating protein-dependent hydrolysis of GTP by ADP-ribosyla-tion factor-1. J. Biol. Chem. 275, 23615–23619.