INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Major histocompatibility complex class I (MHC-I) molecules present cellular and pathogen-derived peptides to antigen-specific receptors on CD8 T cells. The initial steps of MHC-I biosynthetic and transport pathways are well characterized. Proteolysis of intracellular proteins generates peptides, which are actively transported into the endoplasmic reticulum (ER) for assembly with MHC-I (36, 50). Key actors include the multicatalytic proteasome; ER chaperones (calnexin and calreticulin); TAPs (transporters associated with antigen processing), which translocate peptides across the ER membrane; and tapasin, a protein which bridges MHC-I and TAPs. MHC-I trimolecular complexes, which consist of a highly polymorphic heavy chain, 2-microglobulin, and the antigenic peptide, are then routed through the Golgi to the cell surface.

Post-Golgi trafficking steps are less understood. Although MHC-I molecules are stably expressed at the plasma membrane, a fraction is spontaneously internalized and recycled in T cells and in monocytes/macrophages (25, 38). The biological role of MHC-I recycling is not fully understood, but includes the optimization of peptide loading (1, 19). MHC-I internalization is accelerated by anti-MHC-I antibodies or during T-cell activation (25, 38, 46). In contrast, MHC-I endocytosis is barely observed in fibroblasts (25, 38). Internalized molecules are detected in early endosomes located close to or within the Golgi and are either recycled to the cell surface or end up being degraded (8, 19, 50). They may also enter classical MHC-II compartments (8, 19). MHC-I recycling occurs through clathrin-coated pits and involves determinants borne by the heavy chain cytoplasmic tail (13, 23, 46). However, none of the so-far-identified prototypic sorting signals mediating clathrin-dependent endocytosis are found in this region (21).

Prototypic sorting signals direct transmembrane proteins to various intracellular compartments, including the trans-Golgi network (TGN), endosomes, and lysosomes. Sorting signals are located in the cytoplasmic tail of proteins to be sorted and are recognized by adaptor protein (AP) complexes (5). AP complexes are involved in the formation of transport intermediates, such as clathrin-coated pits and clathrin-coated vesicles (CCVs). Four related AP complexes have been described so far: AP-1 and AP-2 target TGN and plasma membrane proteins, respectively, to endosomes (5); AP-3 participates in transport from the Golgi to lysosomes (43); AP-4 is localized at the TGN or a neighboring compartment (14). Sorting motifs located in cytoplasmic domains mostly consist of a leucine-based or tyrosine-based motif (L and YXX, respectively, where X represents any residue and  is an amino acid with a bulky hydrophobic chain, such as L or I). The medium (µ) chain of AP complexes binds tyrosine-based motifs (26, 28). The ligand of leucine-based signals may be the µ or  chains of AP-1 and AP-2 (6, 37).

In human immunodeficiency virus (HIV)-infected cells, MHC-I cell surface expression is down-regulated due to the action of the viral protein Nef (42). Synthesis and transport through the ER and cis-Golgi occurs normally, but MHC-I molecules from both the Golgi and the cell surface are misrouted towards the endosomal pathway (21, 42). MHC-I molecules accumulate in a perinuclear region which also contains proteins known to be abundant in the TGN (rab6 and -adaptin), as well as in more peripheral vesicles positive for endosomal markers (transferrin and clathrin). MHC-I molecules end up being degraded in lysosomes (42). Nef-responsive determinants are contained within the cytoplasmic tail of MHC-I (21). Tyrosine 321 (21), which is conserved in HLA-A and -B, and not in HLA-C and -E molecules, and two other residues (alanine 325 and aspartic acid 328) (9) lying within 7 amino acids (aa) of each other are necessary for Nef activity. Cohen et al. observed that these three residues would lie on the same face if they were displayed as an alpha helix, suggesting a potential interaction surface on the cytoplasmic tail of MHC-I molecules (9). With respect to the protective effect of HLA-C and -E against lysis by natural killer (NK) cells, the selective modulation of HLA-A and -B by Nef (21) allows HIV-infected cells to escape from both virus-specific cytotoxic T lymphocytes (CTLs) and NK cells (9, 10).

Tyrosine 321 of MHC-I may be considered as part of a degenerated YXX motif (YSQA). Since an interaction between Nef and the µ chain of AP complexes was revealed by the yeast two-hybrid system and in vitro with recombinant proteins (21, 32), a model was proposed in which Nef would disclose an otherwise cryptic signal (YSQA) to the AP-dependent sorting machinery. The validity of this model was supported by the resemblance of the effects of Nef on MHC-I and CD4. Indeed, Nef acts as a connector between CD4 and the clathrin-dependent sorting machinery (27, 34), and Nef mutants unable to bind AP complexes neither colocalize with clathrin nor down-regulate CD4 (7, 11, 17, 24, 39). However, by comparing Nef mutants for their ability to affect either CD4 or MHC-I expression, it was determined that Nef-induced CD4 down-regulation and MHC-I down-regulation constitute genetically and functionally separate properties (17, 24, 39). In particular, Nef mutants unable to bind AP complexes still modulate MHC-I, suggesting that this interaction may be not required for MHC-I down-regulation.

Here, we investigated this issue further by designing HLA-A2 molecules, termed A2-endo, which carry prototypic sorting signals (YSQI or YSQL) instead of the degenerated motif YSQA. We compared the spontaneous trafficking of A2-endo to that of wild-type (WT) HLA-A2 in the presence of Nef. A2-endo surface expression was constitutively reduced; molecules showed rapid internalization, recycling, and retention in a perinuclear region likely corresponding to the Golgi. However, we demonstrate that distinct trafficking pathways regulate clathrin-dependent and Nef-induced MHC-I modulation. We also show that the steady-state surface level of MHC-I, which can be modulated by rapid endocytosis or by the viral protein Nef, influences MHC-I-restricted lysis of target cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmid construction. The A2 WT vector, containing the HLA-A2 gene downstream of the cytomegalovirus promoter in pcDNA3 (Invitrogen), was described previously (21). HLA-A2 mutants were generated by PCR and inserted into pcDNA3. The sequence of HLA-A2 mutants was verified by sequencing. The Nef-FT, and Nef-mock vectors carrying the Nef LAI gene in a sense and antisense orientation, respectively, were described previously (21). The green fluorescent protein (GFP) vector (pEGFP) was obtained from Clontech. Dynamin-1 WT and K44A cDNAs were a gift from S. Schmid (University of California, San Diego) (12) and were subcloned in pcDNA3, yielding dynamin-1 WT and dynamin-1 K44A vectors, respectively. The HIV strain used is the BRU-HA infectious clone, and it contains a hemagglutinin (HA)-tagged integrase (31). BRU-HANef was constructed by inserting a frameshift mutation at the unique XhoI site of the provirus. The CD4 vector pCCD4 was obtained by inserting the CD4 cDNA in pcDNA3.

Cell analysis and reagents. Electroporation of HeLa cells, flow cytometry, and indirect immunofluorescence staining were performed as described previously (21). Anti-HLA-A2 monoclonal antibodies (MAbs) BB7.2 and MA2.1 provided by F. Lemonnier (Institut Pasteur) were used (42). An anti-transferrin receptor phycoerythrin (PE)-conjugated MAb (anti-CD71 PE) was obtained from Immunotech. Wheat germ agglutinin (WGA) conjugated to Alexa 488 was obtained from Molecular Probes. Nocodazole was obtained from Sigma. Secondary antibodies were obtained from Southern Biotechnologies.

FACS-based internalization assay. The fluorescence-activated cell sorter (FACS)-based internalization assay was performed as follows. HeLa cells were transiently transfected with the indicated HLA-A2 vectors, along with a GFP vector to distinguish the fraction of transfected cells. After 24 h, cells were incubated at 4°C with the MA2.1 anti-HLA-A2 MAb in phosphate-buffered saline (PBS)-1% bovine serum albumin, washed, and incubated at 37°C in culture medium (containing 20 mM HEPES [pH 7.4]). At different periods of time, MA2.1-bound HLA-A2 surface molecules were revealed by PE-labeled goat anti-mouse immunoglobulin G (IgG) antibody, and cells were analyzed by flow cytometry.

FACS-based recycling assay. The recycling of HLA-A2 was measured by using an assay adapted from reference (2). HeLa cells were transiently transfected with the indicated HLA-A2 vectors, along with a GFP vector to distinguish the fraction of transfected cells. After 20 h, cells were treated for 2 to 3 h with cycloheximide (100 µg/ml) and removed from the dish with EDTA (2 mM) in PBS. An aliquot of the cells was stained with the BB7.2 MAb, to define HLA-A2 steady-state surface levels. To remove cell-bound 2-microglobulin, the remaining cells were exposed twice for 1 min to an acidic buffer (50 mM glycine, 100 mM NaCl [pH 2.0]). Cells were then washed and resuspended in Dulbecco's modified Eagle's medium containing cycloheximide (100 µg/ml) plus HEPES buffer (pH 7.4) (20 mM). Fetal calf serum was omitted from the medium in order to avoid the presence of exogenous bovine 2-microglobulin. At different periods of time at 37°C, HLA-A2 surface levels were measured by flow cytometry after staining with the BB7.2 MAb. Steady-state surface levels were defined as 100%, and the fluorescence intensity at time zero, after the acidic wash, was defined as 0%.

Cytotoxicity assay. The HLA-A2-restricted CTL clone 161JXA14 recognizes HIV Gag aa 77 to 85 (peptide SL9, SLYNTVATL) (10). CTL assays were performed as described previously (45). Briefly, target cells were labeled with ⁵¹Cr for 45 min. After two washes, 10⁴ cells (50 µl) were mixed with synthetic peptide (50 µl) at various concentrations and incubated for 45 min at 37°C. Effector cells were then added for 4 h at 37°C, and radioactivity in cell supernatants was counted. The percentage of specific lysis was calculated as [(⁵¹Cr release due to peptide  spontaneous release)/(total release  spontaneous release)] × 100. Each experimental data point represents triplicate determinations. For experiments performed with HIV-expressing cells, cells were ⁵¹Cr labeled for 45 min 20 h after transfection.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

HLA-A2 molecules carrying consensus sorting signals (A2 YSQI or A2 YSQL) are constitutively down-regulated. Our previous experiments demonstrated that Nef-induced MHC-I modulation necessitates tyrosine residue 321, which is located in the cytoplasmic tail of HLA-A and -B heavy chains (21). Since Nef does not phosphorylate this tyrosine, it is likely that other amino acids contribute to the sorting signal revealed by Nef. Tyrosine 321 belongs to a conserved YSQA sequence (Fig. 1A), which almost fits the consensus sorting motif YXXI or YXXL. We therefore asked whether the alanine plays a role in MHC-I stability at the cell surface and susceptibility to Nef.

View larger version (33K): [in this window] [in a new window]   FIG. 1.   Surface expression and susceptibility of HLA-A2 mutants to Nef-induced modulation. (A) Amino acid sequence alignment of the cytoplasmic domain of HLA-A, -B, and -C consensus sequences and of A2 WT and point mutants. Dashes represent conserved residues. Consensus sequences and numbering are from reference 29. (B) Surface levels of WT and mutant HLA-A2. HeLa cells were electroporated with 4 µg of the indicated A2 vectors, along with 0.5 µg of a GFP vector. After 24 h, cells were stained with the anti-HLA-A2 MAb BB7.2, and the surface expression of HLA-A2 was measured in GFP-positive cells by flow cytometry. The data are the mean ± standard deviation of three independent experiments. (C) Surface levels of WT and mutant HLA-A2 in the absence or presence of Nef. HeLa cells were electroporated with 12 µg of Nef-FT (bold curves) or Nef-mock (thin curves) vector, along with 4 µg of A2 WT or mutant vectors and 0.5 µg of GFP vector. After 24 h, the surface expression of HLA-A2 was measured in GFP-positive cells by flow cytometry. Data are representative of three experiments.

View larger version (36K): [in this window] [in a new window]   FIG. 2.   Subcellular localization of HLA-A2 mutants in the absence or in the presence of Nef. HeLa cells were transfected with 12 µg of Nef-mock (left panels) or Nef-FT (right panels) vector, along with 4 µg of the indicated A2 vectors and a GFP vector. After 24 h, cells were fixed, permeabilized, and stained with an anti-HLA-A2 MAb. The localization of HLA-A2 in GFP-positive cells was examined by immunofluorescence staining and confocal microscopy analysis. Series of optical sections at 0.5 µm were recorded. A representative medial section is shown. Scale bar, 10 µm.

Effects of Nef and of consensus sorting signals on MHC-I down-regulation. Nef interacts with different components of the cellular trafficking machinery, including the µ chain of AP complexes (21, 32) and -COP (4, 33). Moreover, in the presence of Nef, MHC-I accumulates in the Golgi region and colocalizes with CCVs (18, 21). To investigate further how Nef facilitates recruitment of MHC-I by the sorting machinery, we performed a comprehensive and comparative analysis of A2-endo- and Nef-induced MHC-I trafficking pathways.

View larger version (33K): [in this window] [in a new window]   FIG. 3.   Susceptibility of A2 WT and A2 YSQL to nocodazole. HeLa cells were transfected with 12 µg of Nef-mock or Nef-FT vector along with 4 µg of A2 WT, or A2 YSQL vector. After 24 h, cells were incubated with (lower panels) or without (upper panels) nocodazole (10 µM) for 1 h before fixation. Localization of HLA-A2 was then analyzed as described in the legend to Fig. 2. To visualize the perturbation of the Golgi network induced by nocodazole, HeLa cells were stained with the Golgi marker WGA conjugated with Alexia 488 (right panels). Scale bar, 10 µm.

Analysis of A2-endo- and Nef-induced MHC-I internalization and recycling. (i) Cell surface stability of A2-YSQI or A2-YSQL. Consensus tyrosine-based sorting signals interact with the medium chain (µ) of clathrin-associated AP complexes and are ligands for the sorting machinery at the TGN and the plasma membrane. We examined whether surface A2-endo molecules were rapidly internalized. WT or mutant HLA-A2 molecules were transiently expressed in HeLa cells, and their stability at the cell surface was measured in a flow cytometry-based assay. Staining of surface molecules was performed with the MA2.1 anti-HLA A2 MAb, which does not induce internalization of HLA-A2. Staining of A2 WT, A2 YSQE, and A2 ASQA was high at time zero (MF of 500 U) and remained stable after 30 min at 37°C (Fig. 4A). In sharp contrast, A2-endo molecules displayed low steady-state levels (MF at time zero of 80 and 200 U, for A2 YSQI or A2 YSQL, respectively) and short half-lives at the cell surface (5 to 10 min; Fig. 4A). Therefore, A2-endo molecules are recruited by the sorting machinery, not only in the TGN, but also at the plasma membrane.

View larger version (25K): [in this window] [in a new window]   FIG. 4.   Effect of prototypic sorting signals and of Nef on the internalization of HLA-A2. (A) Kinetics of internalization of A2 WT and mutants. HeLa cells were transfected with 4 µg of the indicated A2 vectors and 0.5 µg of GFP vector. After 24 h, cells were labeled at 4°C with an anti-HLA-A2 MAb (MA2.1), washed, and incubated at 37°C for the indicated periods of time. Cells were then cooled at 4°C and stained with fluorescent antimouse IgG antibodies. Data are the ratio of the MF at different time points to the value obtained at time zero (400 U for A2 WT, A2 ASQA, or A2 YSQE and 130 U for A2 YSQI and A2 YSQL). Results from three independent experiments (mean ± standard deviation) are shown. (B) Kinetics of internalization of A2 WT in the absence or presence of Nef. HeLa cells were transfected with 4 µg of A2 WT, 12 µg of Nef or Nef-mock vector, and 0.5 µg of GFP vector. Analysis was performed as described for panel A. The surface levels of A2 WT were 400 and 130 fluorescence units, in the absence or presence of Nef, respectively. Results from three independent experiments (mean ± standard deviation) are shown. (C) Comparative analysis of A2 YSQI, A2 YSQL, and Nef-induced HLA-A2 internalization. Internalization rates of HLA-A2 were measured in HeLa cells as described for panels A and B. The 5-min time point is depicted. Results from three independent experiments (mean ± standard deviation) are shown. (D) Kinetics of internalization of HLA-A2 in HeLa clones expressing different steady-state surface levels of A2 WT or A2 YSQL. Independent HeLa clones stably expressing either high (HeLa A2 WT high, 400 U after staining with the MA2.1 anti-HLA A2 MAb) or low (HeLa A2 WT low, 120 U) levels of HLA-A2 WT or expressing A2 YSQL (HeLa A2 YSQL, 120 U) were isolated. Internalization of HLA-A2 was then measured as described above, except that data are presented as the fluorescence intensity of HLA-A2 staining at the indicated time points. For each cell type, results are the mean ± standard deviation of two independent clones.

(ii) Cell surface stability of HLA-A2 in Nef-expressing cells. We compared the rate of Nef-induced MHC-I endocytosis to those of molecules carrying prototypic sorting signals (Fig. 4B). In Nef-expressing cells, steady-state levels of A2 WT were low (MF at time zero of 200 U), and 70% of molecules remained at the cell surface after 30 min at 37°C. Therefore, although Nef slightly increased the endocytosis rate of A2WT, this effect was limited in comparison with the rapid constitutive internalization of A2 YSQI or YSQL (Fig. 4A). A2 ASQA and A2 YSQE were expressed at high levels at the cell surface and were not internalized in the presence of Nef (not shown). The data are summarized in Fig. 4C, which shows the 5-min time point of the internalization assay. In the absence of Nef, A2 WT was stably expressed at the cell surface, with less than 5% internalization. With Nef, this percentage increased to 25%. Stimulation of endocytosis was much more efficient for A2-endo (40 and 60% internalization for A2 YSQI and YSQL, respectively). The low efficacy of Nef-induced internalization of A2 WT suggests that Nef acts primarily on MHC-I molecules located in the Golgi region.

(iii) Steady-state surface level and internalization rate of HLA-A2 are independent parameters. It was important to rule out the possibility that A2-endo molecules were internalized rapidly because of a low number of molecules at the cell surface. We derived stable HeLa clones expressing either high or low surface levels of A2 WT and compared HLA-A2 internalization rates between these clones. Equivalent stabilities of A2-WT at the cell surface were observed in high (MF of 400 U after staining with the MA2.1 MAb)- or low (MF of 130 U)-expression clones (Fig. 4D). Thus, low surface expression of A2 WT was not associated with rapid internalization rates. In a HeLa clone stably expressing A2 YSQL, the steady-state surface level of A2 YSQL was equivalent to that observed in the HeLa A2 WT low clone (MF of 120 U). However, A2 YSQL was rapidly internalized in this clone, with a half-life at the cell surface of 5 to 10 min (Fig. 4D). Thus, the steady-state surface level and internalization rate of MHC-I are independent parameters.

(iv) Recycling of MHC-I and A2-endo molecules. We investigated the fate of intracellular MHC-I molecules and their ability to reach the cell surface in a FACS-based recycling assay (2). WT or mutant HLA-A2 molecules were transiently expressed in HeLa cells. Surface MHC-I complexes were disrupted by treating cells with an acidic buffer, which removed surface -microglubulin, resulting in the absence of staining with anti-HLA-A2 MAb BB7.2. Cells were then incubated at 37°C, and the appearance of surface HLA-A2 staining was measured by flow cytometry over time. Steady-state surface staining in the absence of acidic treatment defined 100% levels. Cells were pretreated for 2 to 3 h with cycloheximide (100 µg/ml) in order to suppress de novo protein synthesis. Under these conditions, small amounts of intracellular A2 WT or A2 ASQA gained access to the cell surface (Fig. 5), indicating that these molecules were inefficiently recycled in HeLa cells. In contrast, A2 YSQI staining reached 90% of steady-state levels after 20 min (Fig. 5). Similar results were observed with A2 YSQL (not shown). Thus, A2-endo molecules were constitutively internalized and recycled back to the cell surface. In the presence of Nef, A2 WT and A2 ASQA were still poorly reexpressed at the cell surface, whereas A2 YSQI was still rapidly and efficiently recycled towards the plasma membrane (Fig. 5). Thus, Nef does not affect MHC-I recycling.

View larger version (11K): [in this window] [in a new window]   FIG. 5.   Effect of prototypic sorting signals and Nef on the recycling of HLA-A2. HeLa cells were transfected with 4 µg of the indicated A2 vectors and 0.5 µg of GFP vector in the absence (left panel) or in the presence of 12 µg of Nef-FT expression vector. After 20 h, cells were pretreated for 2 to 3 h with cycloheximide (100 µg/ml) to eliminate de novo synthesis of MHC-I molecules. An aliquot of the cells was stained with the BB7.2 MAb to define HLA-A2 steady-state surface levels. To disrupt cell surface MHC-I complexes, the remaining cells were exposed to an acidic buffer that removed surface 2-microglobulin. Cells were then incubated at 37°C for the indicated periods of time, and the surface expression of HLA-A2, from a preexisting intracellular pool, was measured by flow cytometry in GFP-positive cells after staining with the BB7.2 MAb. Steady-state surface levels were defined as 100%, and the fluorescence intensity at time zero, after the acidic wash, was defined as 0%. The data show the ratio of the mean fluorescence at different time points to the value obtained for steady-state levels. In a typical experiment, steady-state levels were 750, 690, and 80 U for A2 WT, A2 ASQA, and A2 YSQI without Nef and 210, 680, and 50 U with Nef, respectively. Results from three independent experiments (mean ± standard deviation) are shown.

Distinct trafficking pathways regulate A2-endo and Nef-induced MHC-I modulation. We investigated the susceptibility of constitutive and Nef-induced down-regulation of mutant and WT HLA-A2 molecules to a dominant-negative dynamin-1 mutant. Dynamin-1 is an ~100-kDa protein involved in the formation of clathrin-coated vesicles, which acts in a GTP-dependent manner (41). Dominant-negative dynamin-1 mutant K44A fails to load GTP and blocks formation of clathrin-coated pits and CCVs at the plasma membrane without other obvious effects (12, 41).

View larger version (40K): [in this window] [in a new window]   FIG. 6.   Susceptibility of A2-endo and Nef-induced HLA-A2 and CD4 modulation to a transdominant-negative dynamin-1 mutant. (A) A2 YSQI and A2 YSQL are endocytosed through a dynamin-1-dependent pathway. HeLa cells were transfected with 4 µg of A2 WT, A2 YSQI, or A2 YSQL vector, along with 30 µg of dynamin-1 WT (white bars) or the transdominant-negative dynamin-1 K44A vector (grey bars), and 0.5 µg of GFP vector. After 24 h, the surface expression of HLA-A2 was measured in GFP-positive cells by flow cytometry. Surface levels of the transferrin receptor (Tf R) were measured as a control for the inhibition of the clathrin-dependent endocytic pathway by dynamin-1 K44A. The data show the steady-state surface levels of the indicated molecules, with 100% corresponding to the levels measured when dynamin-1 WT was expressed. The overexpression of dynamin-1 WT did not affect the surface level of transferrin receptor and of HLA-A2 WT or mutant molecules (not shown). The expression of Nef did not affect that of dynamin-1 WT or mutant proteins, as verified by Western blotting (not shown). (B) Nef-induced MHC-I modulation is not inhibited by dynamin-1 K44A. HeLa cells were transfected with 4 µg of A2 WT, along with either 30 µg of WT (white bars) or K44A dynamin-1 vector (grey bars), and with the indicated amount of Nef vector and 0.5 µg of GFP vector. After 24 h, the surface expression of HLA-A2 was measured in GFP-positive cells by flow cytometry. HLA-A2 steady-state surface levels, measured in the absence of Nef and dynamin-1 K44A, were defined as 100%. (C) Nef-induced CD4 modulation is inhibited by dynamin-1 K44A. HeLa cells were transfected with 2 µg of the CD4 vector of pCCD4, along with either 30 µg of WT (open bars) or K44A dynamin-1 vector (shaded bars), and with the indicated amount of Nef vector and 0.5 µg of GFP vector. After 24 h, the surface expression of CD4 was measured in GFP-positive cells by flow cytometry. CD4 steady-state surface levels, measured in the absence of Nef and dynamin-1 K44A, were defined as 100%. The results from three independent experiments (mean ± standard deviation) are shown.

Steady-state surface level of MHC-I influences MHC-I-restricted lysis of target cells. We investigated the functional consequences of MHC-I down-regulation, either induced by prototypic sorting signals or by Nef in terms of recognition and killing by specific CD8⁺ CTLs. We used as an effector the A2-restricted clone 161JXA14 (10), which recognizes a Gag epitope presented by HLA-A2 in HIV-infected cells (45). We first analyzed the influence of steady-state surface levels of MHC-I on the susceptibility of cells to CTLs. HeLa clones expressing various levels of A2 WT were used as targets, following sensitization with the cognate 9-mer peptide SL9. Experiments were performed with two clones expressing high steady-state surface levels of HLA-A2 (HeLa A2 WT high, MF of 1,010 and 900 U, respectively, after staining with the BB7.2 MAb) and with two clones expressing low levels (HeLa A2 WT low, MF of 210 and 260 U, respectively). No lysis was observed when parental HeLa cells, which do not express HLA-A2, were sensitized with the SL9 peptide (not shown). Peptide dose response analysis indicated that HeLa A2 WT high clones were efficiently recognized and lysed by CTLs (Fig. 7A). A maximal lysis of 78% (mean of the two HeLa A2 WT high clones) was observed when cells were preincubated with 1 µg of the peptide per ml. The SD₅₀, which is the peptide concentration giving 50% of maximal specific lysis, was 5 ng/ml. With HeLa A2 WT low clones, maximal lysis was 68% (mean of the two clones) and SD₅₀ was 38 ng/ml (Fig. 7A). This experiment indicated that cells expressing three- to fivefold less surface HLA-A2 molecules are less susceptible to lysis by the A2-restricted CTL clone and require about eightfold more peptide to reach half-maximum lysis. Therefore, the HLA-A2 steady-state surface level influences the susceptibility of target cells to lysis by CTLs. We then performed similar experiments with two HeLa A2 YSQL clones, in which steady-state surface levels (MF of 170 and 270 U, respectively) were close to those measured in HeLa A2 WT low clones. Maximal lysis was 58% (mean of the two HeLa A2 YSQL clones), and the SD₅₀ was 43 ng/ml (Fig. 6A). Therefore, the susceptibility of HeLa A2 YSQL cells to lysis by CTLs was reduced in comparison to that of HeLa A2 WT high, but was not significantly different from that in HeLa A2 WT low.

View larger version (26K): [in this window] [in a new window]   FIG. 7.   The HLA-A2 surface level influences lysis of target cells by specific CTLs. (A) Effects of HLA-A2 steady-state level and stability at the surface on CTL-mediated specific lysis of target cells. HeLa cell clones stably expressing high (A2 WT high) or low (A2 WT low) levels of HLA-A2 or expressing A2 YSQL were used as targets in a cytotoxicity assay with the HLA-A2-restricted Gag-specific CTL line 161JXA14. Various concentrations of the cognate Gag peptide (SL9) were added to 51Cr-labeled target cells and remained in the assay during the 4-h incubation of CTLs and target cells. The effector/target cell (E:T) ratio was 5:1. Two independent clones of HeLa A2 WT high (HLA-A2 surface levels of 1,010 and 900 [fluorescent] U after staining with BB7.2, respectively), HeLa A2 WT low (210 and 260 U, respectively), and HeLa A2 YSQL (170 and 270 U, respectively) were analyzed. For each cell type, the data are the mean ± standard deviation percent specific lysis of the two clones. HLA-A2 restriction of the CTL line was confirmed by its inability to kill parental HeLa cells incubated with SL9 or HeLa A2 cells incubated with an irrelevant A2-restricted peptide (not shown). (B) Down-regulation of HLA-A2 in HIV-expressing cells. HeLa cells were grown in six-well plates and transfected with 5 µg of the HIV provirus BRU-HA or BRU-HANef, along with 1 µg of A2 WT vector and 50 ng of GFP vector. After 20 h, the surface expression of HLA-A2 was measured in GFP-positive cells by flow cytometry. The proportions of transfected cells (expressing GFP) were identical for HIV and HIVNef and reached 55% (not shown). (C) Nef decreases the killing of HIV-expressing cells by the anti-Gag CTL clone. HeLa cells transiently expressing HIV or HIVNef and described for panel B were 51Cr labeled and incubated for 4 h with the HLA-A2-restricted Gag-specific CTL line 161JXA14 at the indicated E:T ratios. Data are the mean of triplicates from one representative experiment. The standard deviation of each experimental point was below 5%. The lysis of HeLa cells transiently transfected with A2 WT only was below 2% (not shown).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The down-regulation of MHC-I by Nef involves a region of the heavy chain cytoplasmic tail, which is highly conserved in HLA-A and -B. The YSQA sequence located at positions 321 to 324 of HLA-A2 can be considered as a degenerated version of tyrosine-based signals recognized by AP complexes regulating intracellular trafficking, the prototypic sequence of which is YXXI or YXXL. This observation has suggested a model possibly accounting for Nef-mediated down-regulation of MHC-I. According to it, the YSQA motif, which is ignored by the sorting machinery in the absence of Nef, would become recognized by virtue of Nef action. Nef, which physically interacts with µ chains of AP-1 and AP-2, either could make a bridge between µ and MHC-I or could induce modification of µ leading to the recognition of the YSQA motif (21). According to this hypothesis, WT HLA-A2 molecules would exhibit in the presence of Nef a behavior similar to that of mutant HLA-2 molecules displaying prototypic tyrosine-based endocytosis motifs. The aim of the present work was to test this hypothesis. A2-endo molecules, in which the WT YSQA motif was converted to prototypic YSQI or YSQL signals, were constructed with that purpose. Although A2-endo shared several features with those shown by WT HLA-2 in the presence of Nef, evidence exists that distinct mechanisms govern Nef-induced and constitutive AP-mediated endocytosis.

Surface expression of A2-endo was constitutively reduced in comparison with that of WT HLA-A2. The half-life of A2-endo at the surface was shorter than 10 min. Surface levels significantly increased when a transdominant-negative mutant of dynamin-1 (dynamin-1 K44A) was expressed. A2-endo accumulated intracellularly, mostly in perinuclear vesicles, which staining colocalized with that of Golgi markers and clathrin and was affected by drugs modifying the Golgi apparatus. A significant fraction of intracellular A2-endo was rapidly addressed to the cell surface. Taken together, these data indicate that A2-endo molecules were actively internalized through clathrin-coated pits, constitutively routed to the endosomal compartment, and recycled back to the plasma membrane. Thus, down-regulation of A2-endo is likely supported by a direct interaction of prototypic tyrosine-based sorting signals with AP complexes, both at the Golgi and at the plasma membrane. Our results indicate that the spacing of tyrosine-based sorting signals relative to the membrane (40) is adequate for interaction with the sorting machinery. Thus, the nonrecognition of the WT YSQA motif by the sorting machinery is due to the presence of alanine instead of isoleucine or leucine at residue 324.

In the presence of Nef, WT HLA-A2 exhibited cell surface down-regulation and intracellular accumulation reminiscent of the constitutive modulation of A2-endo. Cell surface molecules were actively internalized. Intracellular staining accumulated in perinuclear vesicles colocalizing with Golgi markers and modified by drugs affecting the structure of the Golgi apparatus. However, our study revealed several distinctive features indicating that molecular mechanisms mediating the action of Nef differ from those responsible for the down-regulation of A2-endo. For equivalent numbers of molecules present at the cell surface, internalization of surface MHC-I induced by Nef was less rapid than that of A2-endo (30 versus 70% internalized molecules after 30 min). MHC-I down-regulation was not affected by a dominant-negative dynamin-1 mutant. WT HLA-A2 molecules accumulating in intracellular vesicles in the presence of Nef were not efficiently addressed to the cell surface. Moreover, A2-endo surface expression was further reduced in the presence of Nef, indicating additional action of Nef on molecules bearing prototypic sorting signals. On the other hand, Nef did not affect the recycling of intracellular A2-endo towards the cell surface. These data indicate a limited effect of Nef on cell surface MHC-I, probably because this action only applies to neo-synthesized MHC-I molecules escaping intracellular retention. Accumulation of newly synthesized molecules in a region assimilable or closely related to the Golgi apparatus appears to be the predominant effect of Nef on MHC-I trafficking. These results render it unlikely that mechanisms supporting the effect of Nef on MHC-I are simple consequences of the disclosure of a cryptic endocytosis signal to the sorting machinery. It is noticeable that neither interaction between Nef and MHC-I, which would support the bridging hypothesis, nor interaction between MHC-I and AP complexes in the presence of Nef, which would support the µ modification hypothesis, has been reported so far.

These conclusions about Nef-induced MHC-I down-modulation contrast with the currently accepted model of Nef-induced CD4 down-modulation. Alteration of CD4 trafficking by Nef involves direct binding of AP complexes to a dileucine-based sorting motif located in the cytoplasmic tail of the molecule. Recognition of this motif by the sorting machinery is presumably responsible for directing cell surface CD4 towards clathrin-dependent endosomal compartments. It is mediated by the interaction of Nef with µ chains, which requires the presence of another dileucine signal in the C-terminal loop of Nef (7, 11, 17, 24). In contrast with CD4, mutation in the dileucine motif of Nef does not affect MHC-I down-regulation (17, 24, 39). Our finding that Nef-induced CD4 but not MHC-I down-regulation is affected by the dynamin-1 K44A transdominant mutant provides additional evidence for distinct mechanisms supporting these effects.

Nef interacts with various components of the cellular trafficking machinery, including -COP, a component of non-CCVs, and NBP-1, the catalytic subunit of the vacuolar ATPase associated with AP-2 complexes (4, 22). Binding to -COP or NBP-1 requires diacidic-based motifs located in the C-terminal disordered loop of Nef (EE155 and ED165, respectively) (22, 33). Mutation of these motifs does not affect Nef-induced MHC-I down-regulation (S. Le Gall, unpublished observation), suggesting that -COP and NBP-1 are not involved in this process. Nef residues associated with the capacity to induce MHC-I down-regulation are located in the N-terminal domain and the polyproline helix of the SH3-binding domain (18, 24). The latter region mediates interactions between Nef and tyrosine or serine and threonine kinases (3). The guanine nucleotide exchange factor Vav also binds to the polyproline motif of Nef, thus activating Vav and subsequent cytoskeletal rearrangements (16). Whether Vav or another SH3-containing protein might be a downstream partner of Nef in MHC-I modulation remains to be determined. Recently, Nef has been reported to bind to the cellular protein PACS-1 (35). PACS-1 was initially described as directing TGN localization of furin by binding to its phosphorylated cytosolic domain (47). Nef interaction with PACS-1 and Nef-induced MHC-I down-regulation are dependent on a cluster of acidic amino acids located in the N-terminal domain of the viral protein. Moreover, MHC-I down-regulation by Nef is partially inhibited in PACS-1 antisense cells. Therefore, Nef may act as a connector between MHC-I and the PACS-1-dependent sorting pathway (35). Our findings that Nef-induced MHC-I down-regulation takes place mostly in the Golgi region and is not affected by a negative transdominant dynamin-1 mutant support this model.

In contrast with HeLa cells and fibroblasts, in which MHC-I is stably expressed at the cell surface, T cells and macrophages exhibit active internalization and recycling of MHC-I in CCVs (23, 25, 38). In B cells, MHC-I molecules are spontaneously internalized and found in endosomes, from which they enter classical MHC-II compartments and are transported back to the plasma membrane (8, 19, 50). Thus, MHC-I molecules lacking so-far-identified prototypic sorting signals can interact with the sorting machinery in the absence of Nef, at least in certain cell types. Mechanisms responsible for cell-type-specific regulation of MHC-I trafficking have not been elucidated. Interestingly, the intensity of Nef-induced MHC-I modulation also varies depending on the cell type. Nef-expressing or HIV-infected HeLa CD4 cells show a threefold decrease in MHC-I steady-state surface levels 21; this study). In contrast, lymphoid cells, such as CEM or Jurkat cells stably or transiently expressing Nef, show a 10-fold decrease in surface MHC-I (18, 42), while a 100-fold decrease has been reported in HIV-infected primary T cells (10). Although part of the observed differences may be attributed to experimental conditions, such as the expressed levels of Nef (21), Nef-induced MHC-I modulation is presumably more efficient in T cells than in other cell types. This observation allows speculation about a possible relationship between constitutive recycling of MHC-I molecules and susceptibility to Nef action.

We have analyzed the functional consequences of MHC-I modulation in terms of recognition and killing by specific CTLs. The epitope density required for a half-maximal cytolytic response by CTLs varies from several thousand peptide-MHC complexes per target cell to fewer than 10, with different combinations of MHC-I, peptides, and CTLs (44). Interestingly, the density of the naturally occurring viral epitope on HIV-infected cells is low compared to the entirety of host cell peptides presented by MHC-I. The abundance of naturally processed HIV peptides was estimated to be in the range of 10 to 400 molecules per infected cell (45). We have shown that a three- to fivefold decrease in the MHC-I steady-state surface level was sufficient to render target cells more resistant to lysis by a specific anti-Gag CTL clone. These results confirm that partial reduction of surface MHC-I in Nef-expressing or HIV-infected cells reduces the efficacy of CTL-mediated cell destruction (10). Thus, Nef might help the virus to evade the CTL response in vivo. HeLa cells expressing A2-endo were more resistant to lysis by specific CTLs than those expressing WT HLA-A2. One can therefore speculate that not only the amount of cell surface HLA antigens, but possibly also the duration of their presence at the cell surface may be significant in determining the susceptibility to CTL-mediated lysis.

The biological role of endocytosis and recycling of MHC-I has not been fully established. Optimization of peptide loading by MHC-I recycling through early endosomes or classical MHC-II compartments has been reported (1, 8, 19). Endocytosis of MHC-I may also play a role in MHC-I-restricted presentation of exogenous antigens (20, 48). A2-endo molecules provide a valuable tool for studying the various biological aspects of MHC-I post-Golgi trafficking, including internalization and recycling.