![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Vol. 18, Issue 9, 3351-3365, September 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
*University of Cambridge, Cambridge Institute for Medical Research, Cambridge CB2 0XY, United Kingdom; and Institut Curie, 75005 Paris, France
Submitted March 8, 2007;
Revised May 24, 2007;
Accepted June 11, 2007
Monitoring Editor: Sandra Lemmon
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
; Collins et al., 1998). Nef is one of the so-called "accessory proteins" of both human and simian immunodeficiency viruses: although not required for viral replication in vitro, Nef plays a key role in the development of acquired immunodeficiency syndrome. Nef is a small protein, only 
200 amino acids, but it has been reported to interact with a large number of host cell proteins, including several proteins that are involved in membrane traffic. These interactions are thought to be responsible for the ability of Nef to modulate the surface expression of MHC-I and other molecules (for review, see Collins and Baltimore, 1999; Piguet et al., 1999; Doms and Trono, 2000; Roeth and Collins, 2006).
Among the binding partners that have been identified for Nef are the adaptor protein (AP) complexes. There are four AP complexes in mammalian cells, two of which, AP-1 and AP-2, are highly enriched in clathrin-coated vesicles (CCVs). AP-1 facilitates clathrin-mediated trafficking between the trans-Golgi network (TGN) and endosomes (although there is still some question about directionality), and AP-2 facilitates clathrin-mediated endocytosis (Robinson, 2004). AP-3, which seems to be able to act both in a clathrin-dependent and in a clathrin-independent manner (Dell'Angelica et al., 1998; Peden et al., 2004; Borner et al., 2006), participates in trafficking to lysosomes and lysosome-related organelles (Robinson and Bonifacino, 2001). The function of AP-4 is still unknown. Nef has been shown to bind to the AP complexes via a typical dileucine motif, ENTSLL (Janvier et al., 2003; Chaudhuri et al., 2007). However, mutations in this motif do not impair the ability of Nef to down-regulate MHC-I (Greenberg et al., 1998a). In addition, a dominant-negative mutant of dynamin, which blocks both clathrin-mediated and caveolae-mediated endocytosis, has no apparent effect on Nef-induced MHC-I down-regulation, indicating that Nef acts mainly by diverting intracellular MHC-I away from the plasma membrane rather than by promoting the internalization of MHC-I that is already at the plasma membrane (Le Gall et al., 2000; also see Kasper and Collins, 2003).
Mutagenesis studies on Nef have pinpointed a cluster of acidic residues, EEEE (amino acids 62–65), as being important for MHC-I down-regulation (Greenberg et al., 1998b; Williams et al., 2005). The current working model is that Nef uses this acidic cluster to bind to phosphofurin acidic cluster sorting protein (PACS) 1 (Doms and Trono, 2000). PACS-1 is a protein that was originally identified in a yeast two-hybrid library screen as a binding partner for furin, an enzyme that cycles between the TGN and endosomes (Wan et al., 1998). Retrieval of furin from endosomes is dependent on a cluster of acidic residues in its cytoplasmic tail, and PACS-1 has been shown to bind in vitro to the acidic cluster sequences of both furin and Nef (Piguet et al., 2000). In addition, antisense RNA directed against PACS-1, which causes an eightfold decrease in PACS-1 expression, partially inhibits Nef-induced MHC-I down-regulation (Piguet et al., 2000), and overexpressing a phosphorylation-deficient mutant of PACS-1 has a similar effect (Scott et al., 2003).
PACS-1 has been proposed to act in turn via adaptor complexes, in particular AP-1. Immunoprecipitation of PACS-1 brings down AP-1 but not AP-2, and this interaction has been mapped to a stretch of eight residues. Substituting these amino acids with eight alanines prevents coimmunoprecipitation with AP-1, and overexpression of the alanine-substituted construct has a dominant-negative effect on both furin sorting and Nef-induced MHC-I down-regulation (Crump et al., 2001). Thus, the model is that MHC-I binds to Nef, which binds to PACS-1, which binds to AP-1, which binds to clathrin, and that these interactions bring MHC-I into an endosome–TGN recycling loop and keep it off the plasma membrane (Doms and Trono, 2000).
Because dominant negatives rely on overexpression and can have indirect effects (e.g., by titrating out other binding partners), we decided to test the possible roles of AP-1, clathrin, PACS-1, PACS-2 (a homologue of PACS-1), and AP-2 in Nef-induced down-regulation of MHC-I by carrying out small interfering RNA (siRNA) knockdowns. The cells were treated with the siRNAs, transiently transfected with a Nef-encoding plasmid, and assayed for Nef expression and surface MHC-I by flow cytometry. While this work was in progress, Roeth et al. (2004) published a report in which they used siRNA knockdowns to investigate the role of AP-1 in Nef-expressing T cells and astrocytes. They found that depleting AP-1 inhibits the ability of Nef to reduce the amount of MHC-I on the cell surface. In the present study, we demonstrate that knocking down AP-1 has a similar effect in HeLa cells. In addition, we explore the functions of clathrin, PACS-1, and PACS-2 in Nef-induced down-regulation of MHC-I, as well as in the sorting of acidic cluster motifs in general. We also investigate the possible involvement of AP-2, and we find that unexpectedly, knocking down AP-2 dramatically enhances the ability of Nef to down-regulate MHC-I.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
) were used as the parental cell line for all of our experiments. The HLA-A2–expressing cell line was generated by Hewitt et al. (2002). The CD8-furinWT-expressing HeLa cell line was a kind gift from Matthew Seaman (University of Cambridge; Seaman, 2004). Two mutant HeLa cell lines, expressing CD8-furinAKGL and CD8-furinAKGL-AAAA, were generated for this study. A QuikChange mutagenesis kit (Stratagene, La Jolla, CA) was used to introduce the mutations into the CD8-furin plasmid (Seaman, 2004), and stable cell lines were selected as described previously (Motley et al., 2006). For the comparison between HeLa and T-cells, a viral-specific CD8 T cell line was used (MacAry et al., 2004).
siRNA Knockdowns
Knockdowns were carried out on the CD8-chimera cell lines as described previously (Motley et al., 2003). An amended 6-d protocol was used for knockdowns in HeLa-HLA-A2 cells. The siRNA treatment was carried out as described above, but cells were transfected with plasmids on day 4 for 24 h followed by a shift to 26°C for a further 24 h on day 5. New siRNAs directed against PACS-1 were synthesized for this study with the following sequences: GACGAAGAUCUCCGGAAAG (PACS-1 siRNA1), AGCAUCCUCAGCACGCCAA (PACS-1 siRNA2), GAUCGUCCUUCCAGCUAGU (PACS-1 siRNA3), and GCUAGUGGACUGGUGGAAA (PACS-1 siRNA4). We also used a previously published PACS-1 siRNA (Crump et al., 2003). For knocking down AP-3 
, an siRNA was synthesized with the sequence GGACGAGGCAAAAUACAUA (Peden, personal communication). siRNA smartpools KIAA0602 and AP1G1 were used against PACS-2 and AP-1 
, respectively. Knockdowns of AP-1 µ1A, AP-2 
, AP-2 µ2, and clathrin heavy chain were carried out using siRNAs that have been described previously and shown to be effective (Hirst et al., 2003; Motley et al., 2003). All siRNAs were from Dharmacon RNA Technologies (Lafayette, CO), except for the siRNA, which was from Ambion (Austin, TX).
Plasmids
Transfections with pIRES plasmids were carried out using Mirus TransIT HeLa MONSTER transfection reagent (Cambridge Bioscience, Cambridge, United Kingdom) for 48 h as described above, and standard immunofluorescence labeling was performed as described previously (Motley et al., 2003). The control (NefSTOP) and wild-type (NefWT) plasmids have already been described (as pCG.NL43Nef*pIRESgfp and pCG.NL43Nef+pIRESgfp, respectively; Schindler et al., 2003). The two mutant plasmids (NefLL164-165AA and NefEEEE62-65AAAA) were constructed using a QuikChange mutagenesis kit (Stratagene).
Antibodies
Several different monoclonal antibodies against MHC-I were used. For flow cytometry and immunoprecipitation, we used BB7.2 (BD Biosciences, San Jose, CA) (Parham and Brodsky, 1981), a conformation-specific antibody that binds to HLA-A2 and HLA-A28. For Western blotting and immunogold electron microscopy (EM), we used HCA2, a kind gift from Emmanuel Wiertz (University of Leiden; Stam et al., 1990), which preferentially binds HLA-A isotypes. For immunofluorescence, we used W6/32 (Barnstable et al., 1978), which recognizes HLA-A, HLA-B, and HLA-C. When used on HLA-A2–expressing HeLa cells, all of the above-mentioned antibodies exclusively or primarily recognized HLA-A2, because it has been shown that expression of other MHC-I isotypes is greatly reduced when a particular isotype is overexpressed (Joyce, 1997). For flow cytometry on the parental HeLa cell line, which expresses HLA-A6802/B1503/C1203 (Peter Cresswell, personal communication), we used W6/32 (see above) and 4E (Yang et al., 1984; Zinszner et al., 1990), which binds HLA-B and HLA-C.
Other antibodies used in this study were in-house antibodies against µ3, clathrin heavy chain (CHC) (Borner et al., 2006), and Nef (MATG 020; Stumptner-Cuvelette et al., 2003); commercially available antibodies against µ2 and (both from BD Biosciences Transduction Laboratories, Lexington, KY), (mAb100.3; Sigma-Aldrich, St. Louis, MO), and CD8 (153–020; Ancell, Bayport, MN); a polyclonal antibody against AP-3 , which was a kind gift from Andrew Peden (University of Cambridge); and a polyclonal antibody against µ1A, which was a kind gift from Linton Traub (University of Pittsburgh). In-house antibodies were also generated against PACS-1 and PACS-2. Residues 428–493 (for PACS-1) and 348–419 (for PACS-2) were amplified by polymerase chain reaction (PCR) from IMAGE Clones 3091722 and 6371669, respectively, and cloned into pGEX-T1 to generate glutathione S-transferase fusion proteins. Expression and purification of the fusion proteins, the immunization protocol, and affinity purification of the resulting antisera were performed as described previously (Hirst et al., 2000). Secondary antibodies for immunofluorescence and flow cytometry were purchased from Invitrogen (Carlsbad, CA). Horseradish peroxidase-conjugated secondary antibodies (Sigma- Aldrich) were used for Western blotting.
Immunofluorescence and Western Blotting
Immunofluorescence labeling was carried out as described previously (Motley et al., 2003). For triple immunofluorescence labeling, cells were fixed with paraformaldehyde for 20 min, quenched with 0.05M NH4Cl in phosphate-buffered saline (PBS) for 5 min, and then they were stained with W6/32 followed by Alexa 555-conjugated anti-mouse immunoglobulin. Cells were subsequently permeabilized with 0.1% Triton X-100 for 9 min, and then they were restained with W6/32 followed by Alexa 647-conjugated anti-mouse immunoglobulin before imaging. Cells were viewed using a Zeiss Axiophot fluorescence microscope (Carl Zeiss, Jena, Germany) equipped with a charge-coupled device camera (Princeton Scientific Instruments, Monmouth Junction, NJ), and photographs were recorded using IP Lab software (BD Biosciences, Rockville, MD).
Western blots of whole cell homogenates were carried out as described previously (Borner et al., 2006). HeLa-HLA-A2 and T cell homogenates were matched for total protein concentration before electrophoresis using a Bio-Rad protein assay kit (Bio-Rad, Hercules, CA) according to the manufacturer's instructions.
Flow Cytometry
Cells were harvested 48 h after transfection using a trypsin-EDTA solution (Sigma-Aldrich). To assay for Nef expression, the cells were then fixed with 2% paraformaldehyde, permeabilized with 1% saponin, and blocked in 10% fetal bovine serum, and then they were labeled with anti-Nef followed by Cy5-conjugated goat anti-mouse immunoglobulin (Invitrogen). To assay for surface HLA-A2, unfixed cells were incubated at 4°C with anti-HLA-A2 (BB7.2) followed by the same secondary antibody, and 7-amino-actinomycin D (7-AAD) (Calbiochem, San Diego, CA) was used to distinguish live cells. The commercially available 9E10 (anti-myc) monoclonal antibody (mAb) (Santa Cruz Biothechnology, Santa Cruz, CA) was used as an antibody control for background labeling. Flow cytometric data were acquired using a two-laser, four-color FACSCalibur flow cytometer (BD Biosciences). Data analysis was done using CELLQuest (BD Biosciences) and FloJo software (Tree Star, Ashland, OR), and the mean of green fluorescent protein (GFP)-positive cells was used to compare and plot data. For electron microscopy and for the Western blotting experiment shown in Figure 7b, the GFP-positive cells were sorted using a MoFlo cell sorter (Dako UK, Ely, Cambridgeshire, United Kingdom) into fresh, chilled medium containing 10% fetal calf serum.
Flow cytometry data were analyzed using Flowjo (version 6.3.4; Tree Star). Cells were identified by their forward scatter/side scatter and 7-AAD profiles, and gates were used to distinguish live cells from debris for further analysis. To control for differences in overall surface MHC-I staining between samples, the mean fluorescence intensity (MFI) for the GFP+ and GFP– populations was expressed as a ratio (GFP+/GFP–). Nef function was then calculated (NefWT/NefSTOP). Transfection efficiency varied between experiments performed on different days; therefore, to allow for direct comparison of these experiments, test samples were standardized to the average for all NefSTOP samples assayed on one particular day. Knockdown and no knockdown samples were then compared, and differences were calculated. The average and SD was calculated for at least three independent repeats of each knock down. The values plotted relate to the amount of MHC-I retained at the cell surface in each condition; therefore, each value is inversely proportional to Nef activity.
Pulse-Chase Assay
For each time point, 1 x 106 HeLa-HLA-A2 cells were trypsinized, resuspended in 2% bovine serum albumin in PBS, and sorted for GFP-positive cells by using a MoFlo cell sorter (Dako UK). Cells were then washed with PBS and incubated in prelabel medium (DMEM lacking methionine and cysteine plus 2 mM L-glutamine) for 60 min at 37°C. The cells were pulsed for 9 min with 0.5 mCi/ml 35S-ProMix (GE Healthcare, Little Chalfont, Buckinghamshire, United Kingdom) and chase media were added (RPMI 1640 medium plus 10% fetal bovine serum, 2 mM L-glutamine, 10 mM HEPES, 3 mM methionine, and 60 µM cysteine) for the indicated period. The cells were washed with PBS and lysed with 1% Triton X-100 in Tris-buffered saline (TBS), pH 7.4, containing 4-(2-aminoethyl) benzenesulfonyl fluoride (AEBSF), and then they were precleared for 1 h with a control antibody (mouse IgG; Dako, UK) plus protein A-Sepharose (GE Healthcare). Immunoprecipitations were performed overnight with BB7.2 and fresh protein A-Sepharose. Samples were washed six times with TBS, 0.1% Triton X-100, followed by a final wash in TBS only. Samples were eluted in 2.5% SDS, 50 mM Tris, pH 8, and analyzed by SDS-polyacrylamide gel electrophoresis (PAGE), and then quantified using a PhosphorImager with ImageQuant software (GE Healthcare).
32P Labeling
For each condition, 7 x 106 HeLa-HLA-A2 cells were washed with TBS, pH 7.4, preincubated for 30 min in RPMI 1640 phosphate-free medium plus 2 mM L-glutamine, and labeled with 0.5 mCi of [32P]orthophosphate (PerkinElmer Life and Analytical Sciences, Boston, MA) for 3 h at 37°C. Cells were then washed three times with TBS plus AEBSF and phosphatase inhibitor cocktail 1 and 2 (Sigma-Aldrich). Cells were lysed with Nonidet P-40 lysis buffer (1% Nonidet P-40, 5 mM MgCl2, 50 mM Tris, pH 7.3, AEBSF, and phosphatase inhibitor cocktail 1 and 2), and HLA-A2 was immunoprecipitated using BB7.2 as described above, with the exception that samples were washed with Nonidet P-40 wash buffer (0.1% Nonidet P-40, 5 mM MgCl2, 50 mM Tris, pH 7.3, AEBSF, and phosphatase inhibitor cocktails 1 and 2) after immunoprecipitation. The precipitated samples were then split into two, and one fraction was digested with endoglycosidase H (endo H) according to the manufacturer's protocol (New England Biolabs, Ipswich, MA). Samples were separated by SDS-PAGE and transferred onto nitrocellulose before quantification using a PhosphorImager with ImageQuant software. Immunoblotting was subsequently carried out with HCA2 as described above to control for equal sample loading.
Coprecipitation Assay
Four hours before harvest, cells were washed with PBS and incubated with 25 mM NH4Cl. The cells were then washed again with PBS and trypsinized, and flow cytometry was used to assess transfection efficiency. Samples were adjusted to equal transfection efficiencies and treated with 2 mM dimethyl 3,3'-dithiobispropionimidate·2 HCl (Pierce Chemical, Rockford, IL) in 0.2 mM triethanolamine, pH 8, for 2 h at 4°C, followed by 50 mM glycine for 30 min to stop the cross-linking reaction. Cells were lysed with PBS, 0.3% 3-[(3-cholamidopropyl) dimethylammonio]propanesulfonate (CHAPS), 0.1% SDS, and AEBSF. One percent of the lysates (normalized for total protein) was set aside for input controls, and the remainder was precleared overnight with protein A/G-agarose (Calbiochem). Immunoprecipitation was carried out for 2–3 h by using BB7.2 and protein A/G-agarose. Samples were washed six times with TBS, 0.3% CHAPS, 0.1% SDS, and eluted with 2.5% SDS, 50 mM Tris, pH 8, followed by incubation with 150 mM dithiothreitol at 37°C for 30 min. Input controls were mock eluted with 150 mM dithiothreitol at 37°C for 30 min. Samples were analyzed by Western blotting as described above.
Electron Microscopy (EM)
For immunogold labeling of cryosections, control and AP-2–depleted HeLa-HLA-A2 cells, transfected with either NefSTOP or NefWT plasmids, were sorted for GFP-positive cells as described above. Fluorescence-activated cell sorting (FACS)-sorted cells in suspension were fixed by adding an equal volume of freshly prepared 4% paraformaldehyde/0.4% gluteraldehyde in 0.1 M phosphate buffer, pH 7.4. After 10 min of fixation at room temperature, cells were centrifuged at 1200 x g, and the pellet was gently resuspended in 2% paraformaldehyde/0.2% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4, and postfixed for 2 h at room temperature (Kleijmeer et al., 1996). Cells were embedded in 10% gelatin at 37°C, cooled on ice, trimmed, and infused with 1.7 M sucrose/15% polyvinyl pyrolidone overnight at 4°C. Ultrathin frozen sections were cut in a cryochamber attachment using a diamond knife mounted on a Reichert Ultracut S ultramicrotome (Leica, Milton Keynes, United Kingdom), collected with 2% methyl cellulose and 2.3 M sucrose in a 1:1 ratio, and mounted on Formvar-carbon–coated EM grids. Thawed 50- to 60-nm cryosections were labeled with HCA2, then incubated with rabbit anti-mouse antibody (Dako UK) followed by 10-nm protein A gold (Utrecht University, Utrecht, The Netherlands) (Slot and Geuze, 1983). The sections were observed in a Phillips CM 100 transmission electron microscope (Philips Electron Optics, Cambridge, United Kingdom) at an operating voltage of 80 kV.
Quantitative analysis of the intracellular distribution of HLA-A2 was determined by applying standard stereology methods (Weibel, 1979; Griffiths, 1993). The HCA2 antibody was first tested in serial dilutions on the parental cell line and HeLa-HLA-A2 cells to minimize nonspecific labeling. Grids with sections from each test sample, labeled with HCA2 at 1:75, were scrambled to allow for unbiased analysis. Grids were first examined at low magnification to find a grid square containing sections. At higher magnification (21,000x), photos were taken within randomly chosen cells. More than 100 pictures were captured of each test sample from two grids and two independent labeling experiments. The presence of gold particles associated with the plasma membrane, the nucleus, the cytoplasm, the Golgi region, lysosomes, electron-lucent vacuoles and associated tubulovesicular membranes, multivesicular bodies (MVBs) (both internal and limiting membranes), and MVB-associated tubulovesicular membranes were scored and expressed as a percentage of the total number of gold particles counted for each condition. More than 1200 gold particles were counted for each test sample, and a total of 6607 gold particles were analyzed. Golgi/TGN membranes were defined as clearly visible Golgi stacks plus tubular and vesicular membranes within a distance of 200 nm of the stack. Vacuoles were defined as possessing vacuolar morphology with an electron-lucent lumen ranging from 200 to 700 nm. MVBs were defined as typical round-membrane structures containing many tightly packed internal vesicles, whereas lysosomes were defined by their content of multilamellar ring-like structures. Tubular and vesicular membranes found within a distance of 500 nm of the electron-lucent vacuoles or MVBs were distinguished from tubulovesicular membranes found elsewhere in the cytoplasm.
Quantification of the percentage of area occupied by MVBs and electron-lucent vacuoles, their mean size, and number/100 µm2 was determined as described by Weibel (1979) and Griffiths (1993). The same photos used for gold particle analysis were overlaid with a 1.5-cm grid lattice to score points within the cytoplasm, the nucleus, MVBs, and the electron-lucent vacuoles. More than 6800 points were analyzed for each test sample, and a total of 32,272 points were analyzed.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
; Collins and Baltimore, 1999), for host cells we used a HeLa cell line that stably expresses HLA-A2, which has been shown to be particularly sensitive to Nef. siRNA knockdowns were started on day 1 and repeated on day 3. Transient transfections were carried out on day 4, by using an IRES plasmid encoding Nef and GFP as a bicistronic message so that we could identify the Nef-expressing cells (Schindler et al., 2003). As a control, we used the same plasmid in which a premature stop codon had been inserted into the Nef sequence at position 2 (NefSTOP). Mutations were also engineered into both the acidic cluster motif (NefEEEE62-65AAAA) and the dileucine motif (NefLL164-165AA). The cells were shifted to 26°C on day 5, because Nef-mediated MHC-I down-regulation in HeLa cells has been shown to be enhanced at this temperature (Kasper et al., 2005; Supplemental Figure S1). Assays were carried out on day 6.
|
To quantify surface expression of HLA-A2, we transfected the cells with NefSTOP, NefWT, NefEEEE62-65AAAA, and NefLL164-165AA plasmids, and then we labeled the cells after 48 h with anti-HLA-A2 at 4°C and analyzed them by flow cytometry. Figure 1e shows representative dot plots, with the GFP fluorescence on the x-axis and the surface HLA-A2 fluorescence on the y-axis. Histogram overlays are shown below the dot plots, comparing the surface HLA-A2 fluorescence in cells expressing NefWT, NefEEEE62-65AAAA, and NefLL164-165AA (solid gray) with the surface HLA-A2 fluorescence in cells expressing GFP only (i.e., NefSTOP; black line) and with the background fluorescence in cells labeled with an irrelevant primary antibody (gray line). The mean levels of surface HLA-A2 are shown in Figure 1f. Expression of wild-type Nef causes surface HLA-A2 to drop to 40–50% of control levels, which is typical of what others have found using cultured cell lines (Kasper and Collins, 2003). The dileucine mutant (NefLL164-165AA) is not significantly different from wild-type Nef, whereas the acidic cluster mutant (NefEEEE62-65AAAA) shows a marked reduction in its ability to modulate surface expression of HLA-A2. Interestingly, however, the acidic cluster mutant is still partially active, especially at very high concentrations. This is consistent with a report by Williams et al. (2005), who showed that mutating the four glutamic acid residues to glutamines impaired the ability of Nef to reduce surface expression of HLA-A2 in a human T cell line but did not abolish it. Together, the results confirm that in our system as in other systems, Nef down-regulates HLA-A2 in a concentration-dependent manner and that the acidic cluster contributes to the down-regulation, but the dileucine is dispensable.
siRNA Knockdowns
Having established a system for quantifying Nef-induced MHC-I down-regulation, we used siRNAs to knock down various proteins. Figure 2 shows representative Western blots of cells treated with a range of siRNAs. In every case, the signal from the target protein is reduced to almost undetectable levels after knockdown.
|
subunit or the µ1 subunit. Because AP-1 acts in a clathrin-dependent manner, we also looked at cells that had been depleted of CHC. The top and middle panels in Figure 3a show representative dot plots from cells transfected with the NefSTOP and NefWT plasmids, respectively. The characteristic downward slope of the Nef-expressing cells is much reduced in all three knockdowns, and the histogram overlays (Figure 3a, bottom) show much less separation between Nef-expressing (NefWT) cells and control (NefSTOP) cells. Each experiment was repeated at least three times, and the mean levels of surface HLA-A2 from all the experiments are shown in Figure 3b. Together, these data show that knocking down either AP-1 , AP-1 µ1, or clathrin strongly inhibits Nef-induced down-regulation of HLA-A2 in HeLa cells.
|
), then knocking down PACS-1 should have a similar effect to knocking down AP-1 or clathrin. However, when we depleted PACS-1, we saw no effect on Nef activity (Figure 4). This result was confirmed using four additional siRNAs directed against PACS-1, including one siRNA previously reported by Crump et al. (2003) to be effective for knockdowns. All of the siRNAs were able to deplete PACS-1 when analyzed by Western blotting, but no effect on down-regulation of HLA-A2 was observed (unpublished data). PACS-1 has a homologue called PACS-2, which is 54% identical in sequence. Because of the possibility that PACS-1 and PACS-2 might be functionally redundant, we also tried knocking down PACS-2, either alone or in combination with PACS-1. Again, we saw no effect on Nef activity (Figure 4), even though Western blotting showed efficient depletion of both proteins (Figure 2 and Supplemental Figure S4). Thus, neither PACS-1 nor PACS-2 seems to be required for the down-regulation of HLA-A2 by Nef.
|
). However, CPD may have more than one sorting signal in its cytoplasmic tail (Eng et al., 1999). Similarly, furin, the best characterized of the acidic cluster proteins, has at least two sorting signals: the acidic cluster and YKGL, a typical YXX
motif that acts as an internalization and intracellular sorting signal by binding to AP complexes (Voorhees et al., 1995). To examine the role of the acidic cluster motif without any contribution from the YXX motif, we made use of an existing cell line expressing a CD8 chimera with the wild-type furin tail (Seaman, 2004), and we also generated two new cell lines expressing CD8-furin chimeras with mutations in the cytoplasmic tail: one in which the YKGL sorting signal was rendered nonfunctional by mutating it to AKGL (CD8-furinAKGL), and another in which the acidic cluster was also mutated by changing an EEDE sequence to AAAA (CD8-furinAKGL-AAAA). The steady-state distribution of the three constructs is shown Figure 5, a–c. Both CD8-furinWT and CD8-furinAKGL have an intracellular perinuclear distribution. In contrast, CD8-furinAKGL-AAAA localizes mainly to the plasma membrane (Figure 5c), indicating that the acidic cluster is the essential sorting signal in the CD8-furinAKGL construct. Next we investigated the effects of knocking down clathrin, PACS-1, and PACS-2 on the steady-state distribution of CD8-furinWT and CD8-furinAKGL. Clathrin knockdowns caused both constructs to take on a more peripheral distribution, with increased labeling of the plasma membrane (Figure 5, d and h). However, knocking down PACS-1 and PACS-2, either alone or in combination, had no apparent effect (Figure 5, e–g and i–k). We also carried out a more quantitative assay for increased surface expression of the constructs by labeling cells at 4°C with a CD8-specific antibody and then analyzing them by flow cytometry. Surface expression of both CD8-furinWT and CD8-furinAKGL increased four- to sixfold in the clathrin-depleted cells, but no significant increase was observed in the PACS-1, PACS-2, or PACS-1 + PACS-2–depleted cells (Figure 5, l and m). Together, these data indicate that PACS-1 and PACS-2 are dispensable not only for Nef-induced down-regulation of MHC-I but also for the sorting of cargo proteins with acidic cluster motifs.
|
; Borner et al., 2006). Figure 5n shows that whereas the AP-1 subunit displays this type of behavior, PACS-1 is only found in trace amounts in CCVs prepared from control cells and is not depleted from mock CCVs. This indicates that the small amount of PACS-1 present in the control CCV preparation is a contaminant.
AP-2
Although the AP-1 and clathrin knockdown data support the hypothesis that Nef keeps MHC-I off the cell surface by exploiting an intracellular pathway, there is evidence that Nef also increases the rate of endocytosis of MHC-I (Schwartz et al., 1996), and it is thought that this pathway may be particularly important in HeLa cells (Blagoveshchenskaya et al., 2002). To investigate whether AP-2 might contribute to the down-regulation of MHC-I, we depleted both the and the µ2 subunit (Figures 2 and 6). We expected to find either a reduction in Nef activity or no effect. Surprisingly, however, we found that knocking down AP-2 actually enhances Nef activity (Figure 6 and Supplemental Figure S2). At high concentrations of Nef, surface labeling for HLA-A2 approaches the background levels observed when the cells are labeled with an irrelevant antibody (Figure 6a). The effect of knocking down AP-2 is entirely dependent upon coexpression of Nef; knockdown of AP-2 has no effect on the surface expression of HLA-A2 in cells transfected with the NefSTOP plasmid (Figure 6a).
|
), and we speculated that T cells might naturally express much less AP-2 than HeLa cells. However, when we probed Western blots of HeLa cells and T cells with antibodies against AP-1, AP-2, and AP-3 subunits, we saw similar expression levels of AP-1 and AP-2 in the two cell types and differences only in the expression level of AP-3 (Figure 7a).
|
Another possibility was that AP-2 depletion might alter the phosphorylation state of MHC-I. MHC-I can be phosphorylated in vivo by a cyclic AMP-dependent protein kinase (Guild and Strominger, 1984), and there is evidence that Nef only interacts with hypophosphorylated MHC-I (Kasper et al., 2005). If AP-2 were involved in the trafficking of the kinase, then knocking down AP-2 might lead to an increase in hypophosphorylated HLA-A2. To test this hypothesis, we labeled cells with 32P and immunoprecipitated cell extracts with anti-HLA-A2. Figure 7d shows that similar amounts of phosphorylated HLA-A2 come down in control and AP-2–depleted cells.
A third possibility was that AP-2 might be acting as a sink for Nef binding. If Nef is able to interact with AP-2 as well as AP-1, then depleting AP-2 might increase the amount of Nef available to bind to AP-1. Roeth et al. (2004) have shown that AP-1 can be immunoprecipitated with antibodies against HLA-A2 in Nef-expressing T cells, and they suggested that Nef provides a link between HLA-A2 and AP-1. We were able to repeat this result using HeLa cells (Figure 7e). To find out whether Nef also causes other AP complexes to come down with HLA-A2, we probed blots of the same samples with antibodies against the AP-2 and AP-3 µ3 subunits. Figure 7e shows that the interaction seems to be specific for AP-1. We also carried out immunoprecipitation experiments using AP-2–depleted cells, to determine whether knocking down AP-2 increases the amount of AP-1 that coprecipitates with HLA-A2 in a Nef-dependent manner. However, we saw no differences in labeling intensity (Figure 7f). Thus, although Nef may well be binding in vivo to AP-2 (and AP-3) as well as to AP-1, our data indicate that the enhancement of Nef-induced down-regulation of HLA-A2 in AP-2–depleted cells is unlikely to be due simply to an increase in Nef availability.
Localization and Turnover of MHC Class I in Nef-expressing Cells
If AP-2 depletion does not affect the total amount of Nef in the cell, the phosphorylation state of MHC-I, or the interaction between MHC-I and AP-1, how else might it be acting? We postulated that knocking down AP-2 might have an indirect effect on the trafficking of MHC-I, by altering the membranes through which it passes. It has been reported that Nef causes an increase in the rate of degradation of HLA-A2 (Schwartz et al., 1996; Roeth et al., 2004), so first we tested whether there was even more degradation in cells that had been depleted of AP-2. The cells were pulse labeled with 35S and chased for up to 20 h, after which HLA-A2 was immunoprecipitated from cell extracts. Our results show that Nef expression causes a marked increase in the rate of degradation of HLA-A2 (Figure 8a) and also slows down the transport of HLA-A2 from the endoplasmic reticulum (ER) to the Golgi (Supplemental Figure S3). However, knocking down AP-2 did not have any obvious effect on HLA-A2 trafficking.
|
). The cells were fixed with paraformaldehyde and labeled for surface MHC-I, and then they were permeabilized with Triton X-100 and labeled again with the same primary antibody followed by a different secondary antibody to visualize total MHC-I. The immunofluorescence images show a clear reduction in the amount of MHC-I at the plasma membrane in Nef-expressing cells, with a concomitant increase in the labeling of intracellular membranes (Figure 8, f–i and n–q). However, at the light microscope level, we were unable to determine the exact identity of the intracellular membranes, or to distinguish any clear differences between AP-2–depleted and control cells. Therefore, we turned to electron microscopy to examine the ultrastructure of the HLA-A2–positive membranes.
For the EM studies, we used flow cytometry to sort for cells expressing high levels of GFP, and then we fixed the cells and labeled frozen thin sections with anti-HLA-A2 followed by colloidal gold. To avoid any bias in our morphometric analysis, the grids were scrambled before they were examined in the microscope. Representative micrographs are shown in Figure 9, a–d, and the percentage of label associated with different compartments in Figure 9e. Not surprisingly, the most striking difference between Nef-expressing and non-Nef–expressing cells is in the amount of HLA-A2 at the plasma membrane. In the control cells, the percentage of label at the plasma membrane goes down from 70% in cells transfected with the NefSTOP plasmid to 4% in cells transfected with the NefWT plasmid, and in the AP-2–depleted cells transfected with the NefWT plasmid, <1% of the label is at the plasma membrane. The reason that these differences are more dramatic than those we observe by flow cytometry is that we selected for high expression of GFP, instead of analyzing the entire population of GFP-expressing cells.
|
20% in the NefSTOP cells, was found to be associated with the "cytoplasm," which includes cytoplasmic membranes of unknown identity as well as cytoplasm where no membranes could be detected. Because the background labeling is extremely low (e.g., only 2–3% of the label is associated with the nucleus), and because HLA-A2 is an integral membrane protein, we suspect that nearly all of the cytoplasmic label is associated with membranes but that some of the membranes are out of the plane of section. These membranes are likely to include the ER, which is difficult to identify in frozen thin sections.
Although Nef has been reported to redirect MHC-I to the TGN (Piguet et al., 2000; Blagoveshchenskaya et al., 2002; Scott et al., 2003), we saw very little labeling of Golgi membranes, including the TGN (Figure 9, b and e). Instead, about half of the label in both sets of NefWT cells is associated with membranes that seem to be endosomal in origin. These include electron-lucent vacuoles (Figure 9c), MVBs (Figure 9, b and d), and tubules associated with the vacuoles and MVBs. Interestingly, we found that the relative amounts of MHC-I associated with vacuoles and MVBs are different in the two populations of NefWT cells: in the control cells, more label is associated with the vacuoles, whereas in the AP-2–depleted cells, more label is associated with the MVBs, in particular with internal vesicles. Table 1 shows that both of these compartments accumulate in Nef-expressing cells, as has been reported previously (Stumptner-Cuvelette et al., 2003). We found that the vacuoles are more abundant in the control NefWT cells than in the AP-2–depleted NefWT cells, which partially accounts for the increase in vacuole labeling in the control NefWT cells. In contrast, the MVBs are similarly abundant in both populations of cells, so the increased labeling of MVBs in the AP-2–depleted cells represents a real shift in the steady state distribution of MHC-I. The internal vesicles of MVBs are thought to be a "point of no return," from which proteins cannot recycle to the plasma membrane but are destined to be degraded in lysosomes (Gruenberg and Stenmark, 2004). Thus, the accumulation of MHC-I on these vesicles in the AP-2–depleted NefWT cells, as well as the slight increase in lysosomal labeling, helps to explain why there is less MHC-I at the plasma membrane.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
; Hewitt et al., 2002; Loureiro and Ploegh, 2006). HIV-1 is one of the most intensively studied of all viruses, and a role for HIV-1 Nef in the down-regulation of MHC-I was first demonstrated >10 years ago (Schwartz et al., 1996). However, there is still some uncertainty about the mechanism involved: whether endocytosis plays a significant role, where the MHC-I actually goes when it is not on the plasma membrane, and which of the host cell proteins contribute to the process. In the present study, we have used HLA-A2–expressing HeLa cells as a model system to investigate some of these issues. By knocking down various proteins with siRNA, we show that AP-1 and clathrin are both needed for down-regulation, that PACS-1 is dispensable, and that AP-2 actually impedes down-regulation.
Although HeLa cells are a useful model system, they are not the normal targets of HIV-1, and there has been some question about their relevance to viral pathogenesis (Kasper and Collins, 2003). In particular, Nef-induced down-regulation of MHC-I in HeLa cells is much less dramatic than in primary T cells (Kasper and Collins, 2003), and endocytosis has been reported to play a more important role (Blagoveshchenskaya et al., 2002). Shifting HeLa cells to 26°C improves the efficiency of MHC-I down-regulation (Kasper et al., 2005; Supplemental Figure S1), and this is thought to be because at lower temperatures, MHC-I traffics more slowly through the secretory pathway and thus it is more available to be diverted away from the plasma membrane (Kasper et al., 2005). Consistent with previous studies (Kasper and Collins, 2003), we saw a slight increase in the rate of endocytosis of HLA-A2 in Nef-expressing cells (unpublished data); nevertheless, uptake was extremely slow compared with proteins that are internalized by clathrin-mediated endocytosis (Motley et al., 2003). Most importantly, knocking down AP-1 strongly inhibits the ability of Nef to down-regulate HLA-A2, similar to the effect seen by Roeth et al. (2004) when they knocked down AP-1 in T cells (also see Kasper et al., 2005). Thus, the same cellular machinery is used in both types of cells, indicating that insights gained from studies on HeLa cells should be widely applicable to other cell types, including T cells.
How does Nef interact with AP-1? Three regions on Nef have been implicated in MHC-I down-regulation: the acidic cluster, a proline-rich domain, and an amphipathic -helix near the N terminus containing a critical methionine residue (Roeth and Collins, 2006). In cells expressing a chimera consisting of Nef fused to the cytoplasmic tail of HLA-A2, AP-1 was shown to coimmunoprecipitate with the chimera, and deleting the -helix or mutating the methionine prevented AP-1 from coimmunoprecipitating, suggesting that this part of the molecule may interact either directly or indirectly with AP-1 (Roeth et al., 2004). Mutating the amphipathic -helix has also been shown to cause the most complete block in Nef activity. However, many studies, including our own, demonstrate that the acidic cluster is also needed for efficient down-regulation of MHC-I (Greenberg et al., 1998b; Piguet et al., 2000; Williams et al., 2005).
There are several reports claiming that the acidic cluster keeps MHC-I off the cell surface by binding to PACS-1, but for the most part, these studies have relied on overexpression of PACS-1 dominant-negative constructs (Blagoveshchenskaya et al., 2002; Crump et al., 2001; Scott et al., 2003), and there are several examples in the literature of such constructs producing indirect effects. For example, a truncated version of the AP-2 binding partner Eps15 has been shown to be a potent inhibitor of clathrin-mediated endocytosis (Benmerah et al., 1998). However, the much more modest effects of Eps15 knockdown (Huang et al., 2004), and even of a complete knockout in Caenorhabditis elegans (Salcini et al., 2001), suggest that the dominant-negative construct may work by titrating out key binding partners, such as AP-2. Similarly, overexpressing PACS-1 mutants may indirectly inhibit Nef activity by affecting other proteins. Antisense RNA directed against PACS-1 has also been reported to inhibit the down-regulation of MHC-I by Nef (Piguet et al., 2000). However, the extent of PACS-1 knockdown with the antisense RNA was less than we find with siRNAs, and off-target effects are more likely with full-length antisense RNA than with short, carefully chosen duplexes (Jackson et al., 2003). The present study is the first investigation into the role of PACS-1 in MHC-I down-regulation by using siRNA-induced silencing, and we saw no effect when we efficiently depleted PACS-1 by using a number of different siRNAs, nor when we depleted PACS-1 and PACS-2 together.
We also found that knocking down PACS-1 and/or PACS-2 did not affect the steady-state distribution of two CD8-furin chimeras, one of which was completely dependent upon its acidic cluster. In addition, when we probed Western blots of CCV preparations from HeLa cells with antibodies against PACS-1, we found that it was neither enriched in control CCVs nor depleted from mock CCVs prepared from clathrin-depleted cells, indicating that it is not part of the clathrin machinery. Thus, at present the function of PACS-1 is unclear, but it may play some sort of a regulatory role relating to the phosphorylation state of acidic cluster proteins. It is also not clear what the adaptor(s) for acidic cluster motifs might be, if not PACS-1, but candidates include the AP complexes, especially because acidic clusters can function as internalization signals as well as signals for intracellular membrane traffic (Voorhees et al., 1995).
Probably our most surprising result was the dramatic increase in HLA-A2 down-regulation when we depleted AP-2. We tested the role of AP-2 initially because we thought that knocking it down might inhibit Nef activity. MHC-I is normally internalized relatively slowly by a clathrin-independent pathway (Caplan et al., 2002), but at least one virus, Kaposi's sarcoma virus, drives MHC-I into the clathrin-dependent pathway by ubiquitinating its cytoplasmic tail (Hewitt et al., 2002). However, knocking down AP-2 had the opposite effect to the one we were expecting, and greatly enhanced the ability of Nef to down-regulate HLA-A2. This effect is not restricted to HLA-A2; we found that AP-2 knockdowns also enhanced the more modest Nef-induced down-regulation of other MHC-I isotypes endogenously expressed by the parental HeLa cell line (Supplemental Figure S2).
Why does knocking down AP-2 potentiate Nef activity? We were able to rule out a number of possibilities, and in fact the only clear difference we saw between control and AP-2–depleted cells was in the steady-state distribution of HLA-A2. This difference was not obvious by light microscopy, but by electron microscopy, we found that although HLA-A2 was localized to endosomes in both populations of cells, there was preferential localization to different endosomal subcompartments. Interestingly, almost none of the HLA-A2 was associated with the TGN, in spite of reports claiming that this is where MHC-I is localized in Nef-expressing cells (Piguet et al., 2000; Blagoveshchenskaya et al., 2002; Scott et al., 2003). However, all of the previous studies have relied on fluorescence microscopy, and it can be difficult to distinguish the TGN from other perinuclear membranes at this level of resolution. The present study is the first to localize MHC-I in Nef-expressing cells by electron microscopy, and our observations clearly indicate that Nef directs MHC-I from the TGN to endosomes, rather than the other way around.
Our current working hypothesis is that knocking down AP-2 affects the protein composition of endosomes and that this in turn influences the fate of MHC-I, which is rerouted to endosomes in Nef-expressing cells. Several studies have shown that membrane proteins whose steady-state distribution is mainly endosomal and/or lysosomal, including the transferrin receptor, lysosomal-associated membrane proteins (LAMPs)-1 and -2, CD63, and the invariant chain of MHC class II, redistribute to the plasma membrane in AP-2–depleted cells (Dugast et al., 2005; Janvier and Bonifacino, 2005; McCormick et al., 2005). It has also been reported that knocking down AP-2 changes the dynamics of early endosomes, causing them to mature more rapidly into late endosomes (Lakadamyali et al., 2006). These changes are consistent with our own observations. Nef expression induces the formation of electron-lucent vacuoles and MVBs, both of which are thought to be endosomal (Stumptner-Cuvelette et al., 2003). However, there are fewer vacuoles relative to MVBs in AP-2–depleted cells. The vacuoles are likely to be an early population of endosomes because they do not contain any internal membranes, whereas the MVBs are filled with the internal vesicles characteristic of later endosomes (Gruenberg and Stenmark, 2004). Knocking down AP-2 alters the steady state distribution of HLA-A2 in Nef-expressing cells, causing a shift from vacuoles to MVBs. Together, these findings suggest that the HLA-A2 may be targeted initially to the vacuoles, which then mature into MVBs. The acceleration of this maturation process in AP-2–depleted cells would help to keep the HLA-A2 off the cell surface, because it would limit the amount of time spent in an early endosomal compartment, from which it could be transported to the plasma membrane, and sequester it inside a late endosomal compartment, from which it would go on to be degraded in lysosomes.
Is MHC-I also localized to endosomes in Nef-expressing T cells, and if so, are they early or late? So far, this question has only been addressed by light microscopy, but Roeth et al. (2004) showed partial overlap of MHC-I with LAMP-1, a marker for late endosomes and lysosomes, and they also demonstrated an increase in the rate of lysosomal degradation. Thus, it seems likely that in T-cells, Nef also targets MHC-I to a late endosomal compartment as a prerequisite for degradation in lysosomes, which helps to explain why HeLa cells become more T cell-like in their response to Nef when AP-2 is depleted. In the future, it will be important to confirm and extend our findings on HeLa cells using T cells, and in particular to localize HLA-A2 in Nef-expressing T cells at the electron microscope level. Other important questions include how the HLA-A2 is sorted to the internal vesicles of MVBs (i.e., whether it is by ubiquitination and interaction with ESCRT complexes, and if so, which E3 ubiquitin ligases are involved), and precisely why knocking down AP-2 stimulates this pathway. Because AP-2 knockdowns change the steady-state distribution of so many proteins normally residing in endosomes and lysosomes, establishing which of these proteins might contribute to the enhancement of Nef activity will not be easy. However, double knockdowns of AP-2 and clathrin show that the clathrin knockdown phenotype is the dominant phenotype, indicating that the enhanced Nef activity in AP-2–depleted cells is still clathrin dependent (Supplemental Figures S4 and S5). Interestingly, preliminary observations indicate that knocking down AP-3 also enhances Nef-induced down-regulation of HLA-A2 (Supplemental Figure S5), and because T cells naturally express relatively low levels of AP-3, this provides a possible clue as to why Nef exerts a stronger effect on T cells than on HeLa cells.
Although we embarked upon this study to try to understand the mechanism of action of Nef, our findings also shed light on the mechanism of action of AP complexes. The endosomal localization of HLA-A2 in Nef-expressing cells supports a role for AP-1 in forward rather than retrograde trafficking between the TGN and endosomes. The lack of any effect on Nef-induced down-regulation of HLA-A2 when we knock down PACS-1 means that the role of PACS-1 as an adaptor for the AP-1 pathway needs to be reevaluated. The potentiation of Nef by the depletion of AP-2 (and possibly AP-3), and the inhibition of Nef by the AP-1 knockdown, suggests that different types of adaptors may have fundamentally different functions, rather than all of them contributing in a similar manner to lysosomal delivery (Pelham, 2004). The change in the steady-state distribution of HLA-A2 in the AP-2–depleted, Nef-expressing cells and the enhancement of HLA-A2 down-regulation in these cells highlight the importance of the AP-2 complex in endosomal dynamics. Thus, like many viral proteins, Nef is proving to be a valuable tool for investigating the membrane trafficking machinery of the host cell.
|
ACKNOWLEDGMENTS |
|---|
; Simon McCallum and Rachael Walker for help with the flow cytometry; Nick Bright for advice on electron microscopy; and Siân Piper, Louise Boyle, Paul Luzio, John Kilmartin, and members of the Robinson laboratory for helpful discussions. This work was supported by grants from the Wellcome Trust and the Medical Research Council. |
Footnotes |
|---|
The online version of this article contains supplemental material at MBC Online (http://www.molbiolcell.org). 
Address correspondence to: Margaret S. Robinson (msr12{at}mole.bio.cam.ac.uk).
Abbreviations used: AP, adaptor protein; CCV, clathrin-coated vesicle; CHC, clathrin heavy chain; CPD, carboxypeptidase D; LAMP, lysosome-associated membrane protein; MVB, multivesicular body; PACS, phosphofurin acidic cluster sorting protein.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
