ABSTRACT
In contrast to many other viruses that escape the cellular immune response by downregulating major histocompatibility complex (MHC) class I molecules, flavivirus infection can upregulate their cell surface expression. Previously we have presented evidence that during flavivirus infection, peptide supply to the endoplasmic reticulum is increased (A. Müllbacher and M. Lobigs, Immunity 3:207–214, 1995). Here we show that during the early phase of infection with different flaviviruses, the transport activity of the peptide transporter associated with antigen processing (TAP) is augmented by up to 50%. TAP expression is unaltered during infection, and viral but not host macromolecular synthesis is required for enhanced peptide transport. This study is the first demonstration of transient enhancement of TAP-dependent peptide import into the lumen of the endoplasmic reticulum as a consequence of a viral infection. We suggest that the increased supply of peptides for assembly with MHC class I molecules in flavivirus-infected cells accounts for the upregulation of MHC class I cell surface expression with the biological consequence of viral evasion of natural killer cell recognition.
Many viruses have evolved mechanisms for the evasion of the immune responses of their hosts. These immune escape strategies frequently block the function of the major histocompatibility complex (MHC) class I antigen presentation pathway and hence prevent the recognition and killing of virus-infected cells by cytotoxic T (Tc) lymphocytes (for reviews, see references 33and 47). Escape from Tc cell surveillance is of particular importance for viruses which can establish persistent or latent infections. Given that Tc lymphocytes are the primary cellular mediators for the control and clearance of viral infections (reviewed in reference 50), it is paradoxical that infection with some viruses (flaviviruses, coronaviruses, and paramyxoviruses) elicits an increase in the cell surface expression of MHC class I molecules, the recognition elements for Tc cells (8, 10, 16, 22, 43). This phenomenon has been investigated in some detail for the flaviviruses, a family of enveloped, plus-strand RNA viruses which are mostly transmitted by arthropods (mosquitoes or ticks) to a vertebrate host (reviewed in reference 27). Many flaviviruses can cause disease in humans, ranging from nondescript febrile illness to encephalitis and hemorrhagic fever, with yellow fever virus, dengue virus, and Japanese encephalitis virus being of particular medical significance. Global warming may lead to extension of the range of arthropod vectors, and this plus transport of infected vertebrate hosts may cause increased human infection and disease, as in the recent emergence of West Nile virus (WNV) in New York (1) and Japanese encephalitis virus in Australia (7).
The upregulation of MHC class I expression at the surface of flavivirus-infected cells has been documented for different cell types (fibroblasts, trophoblasts, myoblasts, astrocytes, macrophages, B cells, and endothelial cells) of different species origins (human, mouse, and hamster) and is induced by flaviviruses from different serocomplexes (8, 19). It is not mediated by interferons or the increased biosynthesis of MHC class I molecules (10, 11, 16, 30). Furthermore, flavivirus infection can also lead to an increase of cell surface MHC class I expression in the mouse lymphoblastoid cell line, RMA-S, which is deficient in MHC class I-restricted antigen presentation due to a mutation in the transporter associated with antigen processing (TAP). TAP is an essential accessory molecule in the MHC class I pathway and functions in the transport of peptides from the cytosol into the lumen of the endoplasmic reticulum (ER) for loading of MHC class I glycoproteins (for reviews, see references 3 and 23). In RMA-S cells, which fail to express the TAP2 subunit of the peptide transporter (35, 46), flavivirus-induced upregulation of endogenous and transfected class I molecules takes place (30). A similar flavivirus-mediated effect was noted in cell lines where the endogenous MHC class I expression is undetectable or very low as a consequence of its developmental downregulation (Syrian hamster BHK and NIL-2 cells [20], mouse embryo fibroblasts, and trophoblasts [10, 11]). The increase of MHC class I molecules induced by flavivirus infection correlates with increased MHC class I-restricted antigen presentation, revealing the functionality of these molecules. This conclusion is based on the findings of augmented lysis of flavivirus-infected cells by alloreactive and virus-specific (influenza or vaccinia virus [VV] when used in double infections with a flavivirus) Tc cells and the increased recovery of peptides by acid extraction from the cell surface of flavivirus-infected cells in comparison to results with uninfected cells (30). Collectively, these results led to the proposition that flavivirus infection upregulates the cell surface expression of MHC class I by a mechanism which involves the increase in the supply of peptides to the ER. This interpretation is consistent with a model which predicts that the rate of peptide import into the ER mediated by TAP, but not the rate of biosynthesis of MHC class I molecules, dominantly controls cell surface expression of MHC class I (24, 34, 35).
The mechanism of how flavivirus infection increases the peptide supply into the ER has not been elucidated. The finding of partial repair of antigen presentation in RMA-S cells, which are nominally devoid of functional TAP, suggested that the flavivirus-induced effect was TAP independent (30). However, we obtained conflicting results with human T2 cells, in which both TAP subunits are deleted (40). In this cell line, flavivirus infection did not upregulate the cell surface expression of MHC class I and did not induce antigen presentation for Tc cell recognition. Here we have investigated the effect of flavivirus infection on peptide translocation into the ER and demonstrate for the first time an enhancement of TAP-dependent peptide import into the lumen of the ER as a consequence of a viral infection.
MATERIALS AND METHODS
Cells.The human lymphoblastoid cell line LCL 721, its TAP-deficient derivative LCL 721.174, and the TAP-deficient human lymphoblastoid T2 cells were grown in RPMI 1640 medium supplemented with 10% fetal calf serum (RPMI-FCS). T2 cells stably transfected withH-2Kb genomic DNA (T2.Kb) (49), rat Tap1 plusTap2a cDNAs (T2.rTAP1/2a) (24), or rat Tap1 plusTap2u cDNAs (T2.rTAP1/2u) (26) were grown in RPMI-FCS supplemented with 0.8 mg of Geneticin (Gibco-BRL)/ml. L929 fibroblasts, natural killer (NK) cell-sensitive Yac-1 cells, HeLa cells, HeLa-S3 cells, and HeLa-S3 cells stably transfected with the herpes viral TAP inhibitor protein ICP47 (HeLa-S3.ICP47) were grown in Eagle's minimal essential medium supplemented with 5% FCS plus 0.8 mg of Geneticin/ml for the transfectant. HeLa-S3.ICP47 cells were a kind gift of H. Hengel, Berlin, Germany, and S. Kohlstädt, Heidelberg, Germany.
Viruses.Working stocks of the flaviviruses WNV (strain Sarafed), dengue-2 virus (Den-2) (strain New Guinea C), Murray Valley encephalitis virus (MVE) (strain MVE-1-51), and the alphavirus Semliki Forest virus (SFV) were prepared from suckling mouse brain and titrated by plaque formation on Vero cell monolayers. Virus stocks were diluted in Hanks' balanced salt solution containing 20 mM HEPES (pH 8.0) plus 0.1% bovine serum albumin (BSA) to obtain the required multiplicity of infection (MOI). VV recombinants encoding human Tap1(VV-hTAP1), human Tap2 (VV-hTAP2) or both humanTap genes (VV-hTAP1+2) (37) were kindly provided by J. Bennink and J. Yewdell, National Institutes of Health, Bethesda, Md. VV-Kk encodes theH-2Kk gene and was a gift from B. Arnold, DKFZ, Heidelberg, Germany. VV-TK− is a control virus with the thymidine kinase gene inactivated by homologous recombination. Virus stocks were crude lysates of VV-infected CV1 cells (18).
Chromium release assay.Secondary H-2Kb- and H-2Kk-restricted VV-immune Tc cells were generated from splenocytes from VV-infected C57BL/6 and CBA/H mice, respectively, as described previously (28). Target cell infection and51Cr labeling were as shown in the figure legends. Target cells (2 × 104) were incubated with effector cells at various effector/target ratios for 6 h, and the percent specific lysis was determined as described previously (28). The standard error of the mean (SEM) was always <2%, and spontaneous release was <20%. The NK cell assay, including phenotyping of effector cells, was performed as has been described (29).
Peptide transport assays.In vitro translocation assays using radioiodinated peptides and streptolysin O-permeabilized cells were performed essentially as described previously (12, 25). Briefly, flavivirus- or mock-infected cells (see below) were washed once with translocation buffer (TB) (130 mM KCl, 10 mM NaCl, 1 mM CaCl2, 2 mM EGTA, 2 mM MgCl2, 5 mM HEPES [pH 7.3]) and then permeabilized in streptolysin O (SLO) solution (50 μl/2.5 × 106 cells; 2.5 IU/ml in TB; Murex Diagnostics, Dartford, United Kingdom) for 20 min at 37°C. The cytosol was removed by 3 washes with 1 ml of TB. Permeabilized cells were resuspended in TB, and 2.5 to 5 μl of radioiodinated peptide in phosphate-buffered saline (PBS) (50 to 100 pmol; specific activity, 20 to 50 cpm/fmol) and 10 μl of ATP (100 mM) were added to a final volume of 100 μl. For a control, 10 μl of EDTA (100 mM) was used instead of ATP. The mixture was incubated for 20 min at 37°C, and peptide translocation was stopped by addition of 1 ml of ice-cold lysis buffer (1% NP-40 in 150 mM NaCl, 5 mM MgCl2, 50 mM Tris-HCl [pH 7.3]). Nuclei were removed by centrifugation at 10,000 × g , and the glycosylated peptide fraction was recovered with 30 μl of concanavalin A (ConA)-Sepharose (Pharmacia LKB, Uppsala, Sweden) slurry by overnight rotation. The ConA-Sepharose beads were washed four times with 1 ml of lysis buffer, and the radioactivity was quantitated by γ-counting. ConA-bound radioactivity is presented as a percentage of the input radioactivity.
For the peptide import-export assay with nonglycosylatable peptide, flavivirus- or mock-infected T2.rTAP1/2a cells were permeabilized with SLO as described above. Permeabilized cells were washed twice in TB–0.1% BSA. Per sample of the translocation kinetics assay, a mixture of 2 × 106 cells in 10 μl of TB–0.1% BSA, 2 μl of the radioiodinated peptide RRYQKSTEL (20 μM stock in PBS), 5 μl of ATP (100 mM), or 5 μl of EDTA (50 mM) for the control in a final volume of 50 μl of TB–0.1% BSA was incubated for 0.5 to 16 min at 37°C. The translocation reaction was stopped by pipetting a 50-μl aliquot into 1 ml of ice-cold TB–0.1% BSA. The cells were pelleted for 10 min at 15,000 × g in the cold, the supernatant was carefully aspirated, and the cell-associated radioactivity was determined by γ-counting.
For flavivirus infection, cells were suspended at 0.5 × 107 to 1 × 107 cells/ml in Hanks' balanced salt solution containing 20 mM HEPES (pH 8.0) and 0.1% BSA in sterile 10- or 50-ml tubes, virus was added to obtain the required MOI, and tubes were left at room temperature for 30 min followed by incubation at 37°C in a CO2 incubator for 30 min. Medium was added to the infected or mock-treated cells to give a cell density of ∼106 cells/ml for suspension cultures or 2 × 105 cells/ml for adherent cells, and cells were transferred to tissue culture flasks and incubated in a CO2 incubator for time periods as indicated in the figure legends. For experiments shown in Fig. 4 and 5, infected and uninfected cells were maintained in RPM1 medium containing 1% BSA instead of FCS to prevent a differential in cell divisions between infected and uninfected cells (39). Adherent cells were recovered with trypsin from the culture flasks before use in the transport experiments. Cells were washed twice with PBS and counted to obtain equal cell numbers for use in transport assays. For treatment of cells with cycloheximide (CHX) (Sigma-Aldrich, Castle Hill, New South Wales, Australia) or actinomycin D (actD) (Sigma-Aldrich), the inhibitors were added at concentrations of 10 and 5 μg/ml, respectively, to the culture media after virus adsorption.
Immunoprecipitations.For comparative immunoprecipitation of molecules of the MHC class I pathway and WNV antigens, LCL 721 cells (2 × 107) were infected with WNV at a multiplicity of 15 PFU/cell or left uninfected. The duration of infection was 16 or 24 h including the labeling interval. Metabolic labeling was for 15 h with 35S-Pro-Mix cell labeling mixture (50 μCi/ml; Amersham Life Science, England) in 15 ml of Eagle's minimal essential medium containing one-fifth of the normal concentration of methionine and cysteine and supplemented with 1% FCS. This was followed by a further 1-h labeling interval in which cells were suspended in 1 ml of methionine- and cysteine-free medium containing 200 μCi of35S-Pro-Mix cell labeling mixture. After the labeling period, cells were washed twice with PBS, cell pellets were lysed in 1 ml of NP-40 lysis buffer, and lysates were precleared as described previously (41). For overexpression of TAP1 and TAP2, CV1 cells (2 × 106) were infected with VV-hTAP1, VV-hTAP2, or the control, VV-TK−, at a multiplicity of 10 PFU/cell. At 3 h postinfection the cells were starved for 0.5 h and metabolically labeled for 2.5 h, and lysates were prepared and precleared as described previously (41). Immunoprecipitation of TAP1 and TAP2 was carried out with rabbit antisera raised against peptides 1p3 and 2p3 (31), respectively, MHC class I was immunoprecipitated with an anti-HLA class I heavy chain antibody, and WNV-encoded polypeptides were immunoprecipitated with anti-WNV hyperimmune ascitic fluid. Immunoprecipitation, electrophoresis, and fluorography were performed as described previously (41).
RESULTS
WNV infection of TAP-deficient T2 cells does not repair antigen presentation via MHC class I.We have have previously shown that infection of the mouse B lymphoma cell line, RMA-S, with different flaviviruses upregulates the cell surface expression of MHC class I and recognition by virus-specific Tc cells despite a deficiency in TAP function due to the lack of expression of TAP2 (30). However, when similar experiments were performed in human T2 cells, in which both TAP subunits are deleted, no flavivirus-mediated upregulation of cell surface expression of HLA-A2 or -B5 was apparent by flow-cytometric analysis (data not shown). Also, Tc cell lysis by VV-immune, H-2Kk-restricted effectors could not be induced by double infection of T2 cells with VV recombinant for H-2Kk plus WNV (Fig. 1B). The same effectors efficiently killed T2.rTAP1/2a transfectants when infected with VV-Kk. A threefold upregulation of lysis of T2.rTAP1/2a cells was noted following double infection with VV-Kk and WNV. Infection of Kb-transfected T2 cells with WNV also failed to induce Kb-restricted presentation of viral Tc cell determinants (Fig. 1A). VV-immune Kb-restricted effectors efficiently lysed T2.Kbcells infected with a recombinant VV encoding human TAP1 and TAP2 (VV-hTAP1+2). However, T2.Kb target cells infected with WNV for 24 h followed by superinfection with VV did not show increased susceptibility to killing by VV-immune effectors in comparison to targets infected only with the control virus, VV-TK− (Fig.1A). The permissiveness of T2 cells to infection with WNV was verified by flow cytometry following internal staining of infected cells with a WNV-immune serum (data not shown) and expression of WNV-specific proteins detected by immunoprecipitation (see below). This result suggests that in contrast to the case with RMA-S cells, the complete TAP deficiency in T2 cells prevents flavivirus-mediated upregulation of MHC class I-restricted antigen presentation.
Effect of WNV infection on Tc cell recognition of TAP-deficient and TAP-transfected T2 cells. (A) T2 target cells stably transfected with H-2Kb (T2.Kb) were infected with VV-TK− or VV-hTAP1+2 for 16 h or infected with WNV for 24 h and superinfected with VV-TK− for 16 h. Target cells were labeled with 51Cr for 1 h followed by incubation with C57BL/6 anti-VV effectors for 6 h. (B) T2 and T2.rTAP1/2a cells were labeled with 51Cr for 1 h followed by infection with VV-Kk (MOI, ∼10) or WNV (MOI, ∼25) or double infection with VV-Kk (MOI, ∼10) and WNV (MOI, ∼25) for 16 h. The target cells were incubated with CBA/H anti-VV effectors in a 4-h 51Cr release assay. Each time point constitutes the mean of the percent specific lysis of three separate wells. The SEM was always <2%. Representative data from two experiments are shown. E:T, effector/target.
Upregulation of in vitro peptide transport in flavivirus-infected cells.To investigate the effect of flavivirus infection on the rate of peptide transport into the lumen of the ER directly, we employed an in vitro peptide transport assay (26). Infected or uninfected cells were permeabilized with SLO and incubated with radiolabeled peptides which contain an N glycosylation motif. Peptides translocated into the ER are modified by carbohydrate addition, allowing their recovery with ConA-Sepharose after cell lysis. Figure 2A demonstrates that the transport of two peptides (#64 and #968) is significantly increased in the human carcinoma cell line HeLa infected with WNV in comparison to results with uninfected cells. Infection for 12 h increased peptide transport by 40 to 60%. Interestingly, when the period of infection was extended to 24 h, a 2.5-fold reduction of peptide transport was noticed.
Dose and time dependence of flavivirus-induced modulation of in vitro peptide transport in human fibroblast and lymphoblastoid cells. (A) HeLa cells were infected with WNV (MOI, ∼5) for 12 or 24 h or left uninfected. The cells were permeabilized with SLO and incubated with radioiodinated peptide #64 (RYWANATRSL) or peptide #968 (TNKTRIDGQY) in a 20-min transport assay in the presence of ATP. Glycosylated peptides were recovered from cell lysates with ConA-Sepharose, and the radioactivity was counted. Transport in the absence of ATP gave <1,000 counts. (B) T2.rTAP1/2a cells were infected with Den-2 (MOI, ∼10) or MVE (MOI, ∼25) for 17 and 15.5 h, respectively, or left uninfected, and peptide transport was measured as described above. Means of two samples ± the SEM are presented. To test the dose dependence of flavivirus-mediated modulation of in vitro peptide transport, HeLa (C) and LCL 721 (D) cells were infected with WNV for 12 and 16 h, respectively, at the MOI indicated or left uninfected. Radiolabeled peptides #63 (RYWANATRSI) and #968 (TNKTRIDGQY) were tested in a 20-min transport assay. Means for two samples ± SEM are presented for the transport in HeLa cells.
The virus-mediated effect of augmented peptide import into the ER was also observed for other flaviviruses. Figure 2B shows that Den-2 or MVE infection of T2.rTAP1/2a cells for 17 or 15.5 h, respectively, gave ∼20% increased transport.
Flavivirus-mediated upregulation of in vitro peptide transport is dose dependent.To analyze whether the early stimulation and late inhibition of peptide transport found during the course of WNV infection was due to changes in host cellular pathways resulting from virus replication, the effect of the infectious dose on peptide transport was tested. The dose dependence of the virus inoculum on modulation of in vitro peptide translocation in HeLa cells and human B cells (LCL 721) is demonstrated in Fig. 2C and D. In HeLa cells a low MOI with WNV (1 to 5 PFU/cell) for 12 h increased, whereas higher doses reduced, peptide transport in comparison to that in uninfected cells (Fig. 2C). Peptide transport was also upregulated in LCL 721 cells when they were infected with WNV at a MOI of 5 to 25 PFU/cell for 16 h (16 to 29% increase for peptides #63 and #968, respectively; Fig. 2D). On the other hand, peptide transport was impaired when the virus dose was further increased. The results shown in Fig. 2 are consistent with a transient virus-induced intracellular process during the early phase of flavivirus replication which impacts the supply of peptides from the cytosol into the lumen of the ER. The events appear to take place at the end of the latent period or the early productive phase of virus replication (45), the duration of which inversely correlates with the virus dose used for infection.
Flavivirus-induced upregulation of peptide transport into the ER is TAP dependent.TAP-mediated peptide translocation across the ER membrane in permeabilized cells requires the presence of ATP (for a review, see reference 23). A similar dependence on ATP was also apparent in flavivirus-infected cells. Omission of ATP or addition of EDTA (Fig. 3A) in transport assays resulted in a 100- to 1,000-fold reduction in glycosylated peptides recovered in both infected and uninfected cells. Capture of divalent cations by EDTA completely abrogates the ATP-hydrolyzing function of TAP (12, 37). Flavivirus infection was never found to augment peptide transport significantly above this background level. It also failed to induce in vitro peptide transport in TAP-deficient LCL 721.174 (Fig. 3B) and T2 cells (data not shown). In both cell lines, infection with WNV at a wide range of multiplicities (1 to 100 PFU/cell) did not elevate peptide transport above background levels seen in the absence of ATP. Finally, to test the requirement for a functionally active TAP in flavivirus-induced upregulation of peptide transport into the ER, which was already suggested by the complete lack of in vitro peptide translocation in infected cells when ATP was omitted, we used HeLa cells transfected with cDNA for the herpes simplex viral TAP inhibitor protein, ICP47. ICP47 is a cytosolic protein which functions as a strong competitive inhibitor of peptide binding by human TAP and which is able to destabilize the human TAP1-TAP2 heterodimer (reviewed in references 33 and 47). WNV infection did not overcome the ICP47-mediated inhibition of peptide import into the ER, and transport levels in infected or uninfected HeLa-S3.ICP47 cells were not above background transport levels in the absence of ATP. As expected, WNV infection resulted in an upregulation of peptide transport in the parental HeLa-S3 cells (Fig. 3C).
WNV-mediated upregulation of in vitro peptide transport requires a functionally active peptide transporter. (A) Transport of radioiodinated peptide #968 in HeLa cells infected with WNV (MOI, ∼0.2) for 12 h or left uninfected in the presence of 10 mM ATP or 10 mM EDTA. Transport was for 20 min, glycosylated peptides were recovered, and radioactivity was counted. Means of two samples ± the SEM are presented. (B) Transport of peptides #63 and #968 in TAP-deficient LCL 721.174 cells infected with WNV for 16 h at the MOI shown or left uninfected. Radioactivity recovered following the 20-min transport assay was <1.5% of that recovered when TAP-expressing LCL 721 cells were used (see Fig. 2). (C) Transport of radiolabeled peptide #63 in HeLa-S3 and HeLa-S3.ICP47 cells infected with WNV (MOI, ∼2) for 16 h. Means for two samples ± the SEM are presented.
Flavivirus-induced modulation of peptide transport does not alter the substrate selectivity of TAP.From antigen presentation studies and in vitro peptide transport assays, it is apparent that TAP has a broad substrate specificity (23). Only the C-terminal peptide residue exerts a pronounced influence on in vitro peptide transport rates. To investigate whether the flaviviral effect on peptide import into the ER also alters the substrate specificity of TAP, peptides with the sequence RYWANATRSX (R10X, where the C-terminal residue X was one of 10 amino acids) were tested in transport assays using flavivirus-infected and uninfected cells. The results of two experiments with LCL 721 cells are compiled in Fig.4A. The transport pattern in the uninfected cells shows the nonselective transport phenotype of human TAP, which accommodates peptides with hydrophobic, small polar, or charged side chains at the C-terminal amino acid (26). Flavivirus infection upregulates the transport of all peptides independently of the nature of the C-terminal residue. Transport rates were increased by 30 to 50% (peptides R10G, R10S, R10T, R10P, R10Y, and R10F; experiment 1) and 10 to 20% (peptides R10I, R10N, R10E, and R10K; experiment 2).
Effect of WNV infection on transport of peptides differing in their C-terminal amino acid. LCL 721 cells were infected with WNV (MOI, ∼5 to 10) for 16 h or left uninfected, permeabilized, and used in a 20-min transport assay with glycosylated peptides of the RYWANATRSX series, where X is one of 10 amino acids shown. Means for two samples ± the SEM are presented.
The allelic rat TAP1/2u (rTAP1/2u) peptide transporter has a characteristically restrictive peptide transport phenotype with respect to the C-terminal amino acid of the peptide substrate (26). As expected, T2 cells transfected with rTAP1/2u efficiently transported peptides of the RYWANATRSX series when the C-terminal amino acid contained a hydrophobic or aromatic side chain but not peptides with charged or small, polar side chains at the C-terminal residue (Fig. 4B). WNV infection did not alter this transport pattern but increased the rate of transport of those peptides which are efficient substrates for rTAP1/2u.
Effect of flavivirus infection on peptide export from the ER.Translocation assays using flavivirus-infected cells had so far been performed with glycosylatable model peptides. To rule out the possibility that virus-induced alterations of the glycosylation machinery rather than increased peptide transport rates resulted in the increased recovery of glycopeptide, we carried out translocation assays with nonglycosylatable peptides. We have shown recently that such peptides only transiently accumulate in the ER before they are released through the Sec61p channel (12). Therefore, it was important to investigate whether flavivirus interferes with the ER export of nonglycosylated peptides.
T2.TAP1/2a cells were infected with WNV for 15 h, semipermeabilized with SLO, and incubated with the radioiodinated peptide RRYQKSTEL in the presence of ATP. The translocation reaction was stopped by rapid cooling in excess ice-cold buffer, and the cell-associated radioactivity was determined. Figure5A shows the typical biphasic retention of nonglycosylatable peptide with a short phase of net accumulation followed by a phase of net peptide release. WNV enhanced the initial accumulation of peptide RRYQKSTEL in T2.TAP1/2a cells by 22%. The kinetics of the export phase was not significantly altered in WNV-infected cells compared with results for mock-infected cells. We conclude that the flavivirus-induced upregulation of peptide transport that is observed with various glycosylatable peptides is also seen with a nonglycosylatable peptide.
Kinetics of peptide accumulation and requirement of viral protein synthesis for WNV-induced upregulation of peptide transport. (A) The nonglycosylatable peptide RRYQKSTEL was translocated in the presence of ATP for the indicated periods. After the reaction was stopped by rapid cooling, the cell-associated radioactivity was γ-counted. For control of unspecific peptide binding, the reaction was carried out in the presence of EDTA, which inhibits TAP-mediated peptide translocation. (B) LCL 721 cells were infected with WNV (MOI, ∼15) for 15.5 h or left uninfected. Following virus adsorption, the cells were maintained in medium containing CHX (10 μg/ml) or actD (5 μg/ml). Permeabilized cells were incubated with radioiodinated peptide #968 (TNKTRIDGQY) in the presence of ATP in a 20-min transport assay. Means of two samples ± the SEM are shown. Representative data from three experiments are shown.
Requirement of flaviviral but not host cellular macromolecular synthesis for upregulation of in vitro peptide transport.To investigate whether viral and host macromolecular synthesis is required for the induction of enhanced peptide transport early during flavivirus infection, CHX and actinomycin D (actD), inhibitors of protein and DNA-dependent RNA synthesis, respectively, were added to uninfected and WNV-infected LCL 721 cells used in an in vitro peptide transport assay (Fig. 5B). Treatment with CHX did not significantly reduce peptide transport in the uninfected cells compared to results with untreated cells but inhibited flavivirus-mediated upregulation of transport. The addition of actD to uninfected cells resulted in a reduction (∼10%) of peptide transport relative to that in CHX-treated and untreated cells. WNV infection gave a relative increase (23%) of peptide transport in the actD-treated cells. Given that viral RNA synthesis, which uses the virally encoded RNA-dependent RNA polymerase, is not significantly inhibited at the concentration of actD used, the inhibitor studies show that viral but not host protein synthesis is needed for augmented peptide transport in flavivirus-infected cells.
Effect of flavivirus replication on the expression TAP1, TAP2, and other molecules of the MHC class I pathway.The upregulation of in vitro peptide transport shown here and the demonstration of increased MHC class I cell surface expression and MHC class I-restricted antigen presentation in flavivirus-infected cells could be the result of augmented expression of molecules of the MHC class I pathway akin to the effects of gamma interferon (44). This was investigated by radioimmunoprecipitation. To this end, LCL 721 cells were left uninfected or were infected with WNV for 16 or 24 h and metabolically labeled. Using antisera against TAP1 or TAP2, both proteins could be detected in the cell lysates; their identity was confirmed by comigration with TAP1- or TAP2-specific bands precipitated from lysates of cells infected with VV recombinants (VV-hTAP1, VV-hTAP2) for overexpression of the human peptide transporter subunits (Fig. 6A and B). These bands were not seen in lysates from cells infected with control VV-TK−virus. No difference in the intensity of bands corresponding to TAP1 or TAP2 in WNV-infected LCL 721 cells compared to that of those for uninfected cells was noted. Likewise, there was no apparent difference in the expression of the MHC class I heavy chain in WNV-infected LCL 721 cells compared to results with uninfected LCL 721 cells (Fig. 6C). Productive infection of the cells with WNV was confirmed by immunoprecipitation of the virus-specific NS5 protein from infected but not uninfected cells (Fig. 6D). The intensity of the NS5-specific band increased with the longer infection time. Unaltered steady-state levels of TAP1 proteins and of class I heavy chains after 16 or 24 h of WNV infection were also apparent when cell lysates were analyzed by a Western blot (data not shown).
Effect of WNV infection on the biosynthesis of TAP1, TAP2, and MHC class I molecules. LCL 721 cells were infected with WNV (MOI, ∼15) or left uninfected. The duration of infection was for 16 or 24 h and included a metabolic labeling interval (see Materials and Methods). Metabolic labeling of infected and uninfected cells was for a period of 16 h and was followed by detergent lysis and immunoprecipitation with antisera against TAP1 (A), TAP2 (B), the MHC class I heavy chain (C), or WNV (D). VV infections of CV1 cells (MOI, ∼10) were for 3 h followed by metabolic labeling for 2.5 h, detergent lysis, and immunoprecipitation with antisera against TAP1 (A) or TAP2 (B). Proteins were analyzed by sodium dodecyl sulfate–10% polyacrylamide gel electrophoresis. Bands corresponding to TAP1, TAP2, the MHC class I heavy chain (HC), and the WNV nonstructural protein NS5 are labeled on the left, and masses (in kilodaltons) of marker proteins are shown on the right of the autoradiogram.
Effect of WNV infection on induction of NK cells, in vivo, and NK cell-mediated cytotoxicity, in vitro.NK cell activity is controlled by expression of inhibitory receptors interacting with self MHC class I molecules (for a review, see reference 13). To investigate if the increase in MHC class I expression after flavivirus infection has biological consequences for NK cell recognition, we tested WNV-infected and uninfected L929 cells and the classical NK target cell line, Yac-1, for lysis by splenic NK cells from naı̈ve, WNV-immune, and SFV-immune CBA/H mice. The splenocytes were harvested 2 days after immunization. Infection with the alphavirus SFV was employed as a control for a strong virus-induced NK cell response (29). In addition, effector splenocytes were phenotyped by treatment with anti-Thy1, anti-CD8, and anti-asialo GM1 antibodies plus complement (C′) prior to 51Cr release assays. Figure 7 shows the results of one such experiment. All splenic effector cells had NK cell activity, as shown by their ability to lyse the NK-sensitive target cell Yac-1 (bottom panels). Lytic activity was abrogated by treatment with anti-asialo GM1 and was partially sensitive to anti-Thy-1 treatment, but effectors were insensitive to treatment with antibodies to CD8, the classical Tc cell marker. Splenocytes from SFV-immunized mice had the greatest lytic activity, whereas splenocytes from WNV-immunized mice had lytic activity that was only marginally increased over that of control splenocytes. The important observation was, however, that L929 target cells infected with WNV for 24 h were at least three- to fivefold less sensitive to NK cell lysis than were mock-infected targets. The reverse is true for lysis by alloreactive Tc cells (19).
NK cell lysis of target cells infected with WNV. Splenocytes taken from naı̈ve CBA/H mice or at 2 days after intraperitoneal infection of mice with 106 PFU of WNV or 107 PFU of SFV were treated with complement (C′) only (■), anti-Thy-1 plus C′ (□), anti-CD8 plus C′ (○), or anti-asialo GM1 plus C′ (▵). Aliquots of effectors were adjusted to volumes of C′ control concentrations to give an effector/target ratio of 90:1. Target cells (Yac-1, uninfected L929, or L929 cells infected with WNV for 24 h with 10 PFU/cell) were labeled with 51Cr for 1 h, and assays were performed in triplicate with 4 threefold dilutions of effectors. Assay time was 6 h. The SEM of triplicates was never greater than 2%.
DISCUSSION
The mechanism for augmentation of MHC class I cell surface expression and antigen presentation induced in flavivirus-infected cells was investigated. Conflicting findings were obtained with two peptide transporter-defective cell lines with respect to the question of TAP dependence of the flavivirus-induced phenomenon. In an earlier report we demonstrated a pronounced increase of peptide-loaded MHC class I at the cell surface of TAP2-deficient RMA-S cells following infection with a flavivirus (30). This was interpreted as a consequence of the virus-induced and TAP-independent supply of peptides into the lumen of the ER for assembly with MHC class I. In contrast, flavivirus infection of T2 and LCL 721.174 cells, which both lack the genes for TAP1 and TAP2, did not increase antigen presentation via MHC class I, determined by flow-cytometric, biochemical, and immunological assays. This apparent discrepancy could be explained by a partial rather than complete antigen presentation deficit in RMA-S cells. Accordingly, some import of peptides into the ER is retained in RMA-S cells which can be enhanced by a flavivirus-mediated effect. A residual capacity to present cytosolic peptides in RMA-S cells has been observed by others (4, 9, 48) and is thought to be mediated by a TAP1 homodimer (6, 21). The possibility that T2 or LCL 721.174 lymphoblastoid cells were resistant to infection with flaviviruses was excluded based on the synthesis of virus-specific proteins in infected cells. In addition, enhanced Tc cell-mediated lysis was seen in WNV-infected T2 cells expressing rat TAP proteins, consistent with flavivirus-mediated upregulation of MHC class I-restricted antigen presentation also in this cell line when the peptide transporter is functional.
Here we present the first example of upregulation of TAP-dependent in vitro peptide import into the ER that is mediated by a virus infection. This biochemical assay complements our previous evidence for flavivirus-induced modulation of MHC class I antigen presentation (30) and localizes the impact of the virus infection to an early stage of the MHC class I pathway. It conclusively shows that flavivirus-induced augmentation of transport of peptide into the ER requires functionally active TAP, since it was not found in the absence of the peptide transporter, when peptide transport was prevented due to inhibition of ATP hydrolysis, or when TAP-mediated peptide transport was blocked in cells expressing the herpesviral TAP inhibitor, ICP47. In most experiments ER-specific peptide glycosylation was used as a read-out for peptide translocation. The stimulation of peptide glycosylation, per se, by flavivirus infection can be ruled out given that TAP mediated import of a nonglycosylatable peptide was also consistently augmented in flavivirus-infected cells.
Interestingly, flavivirus-mediated upregulation of in vitro peptide transport was strictly time and virus dose dependent and took place during the latent or early productive phase of virus replication. This strongly suggests that during this period of infection, viral translation products are synthesized, or intracellular changes are induced, which increase the net import of peptides into the ER. The flavivirus genome encodes a single polyprotein precursor which is cleaved into at least 10 viral proteins (36). The polyprotein processing events during the latent period could, however, yield different gene products than are found at later stages of virus infection (14, 18). These products may be essential in the regulation of flavivirus gene expression, exemplified by the switch from minus-strand to predominantly plus-strand RNA synthesis later in infection (2). Thus, a flavivirus (precursor) protein produced early in infection could be responsible for the transient upregulation of TAP-dependent peptide translocation into the ER. Since viral protein synthesis is barely detectable during the latent period of flavivirus infection, the presence and nature of alternative polyprotein cleavage products remain poorly defined.
In addition to the characterization of the putative viral gene product responsible for the induction of increased peptide import into the ER, the detailed cell biological mechanism for this phenomenon has not been fully resolved. Several putative mechanisms were ruled out as playing a major role in the flavivirus-mediated effect on the MHC class I pathway. First, we have been unable to find an increase in the biosynthesis and steady-state levels of TAP1 or TAP2 proteins following flavivirus infection, which could account for the sometimes >50% increase of in vitro peptide transport rates. Second, we failed to find an association of flavivirus-encoded proteins with the peptide transporter in coimmunoprecipitation experiments with antibodies directed against TAP using mild detergent conditions that were amenable for immunoprecipitation of TAP-MHC class I complexes (32, 42; M. Lobigs, unpublished results). Third, it is conceivable that flavivirus infection induces TAP to undergo posttranslational modifications or to associate with cofactors that could result in increased rates of peptide transport. Recently, it was suggested that the transport function of TAP can be blocked by the phosphatase inhibitor okadaic acid (15). However, we were unable to detect an inhibitory effect of okadaic acid treatment on peptide transport by uninfected or WNV-infected T2.rTAP1/2a cells (M. Lobigs and F. Momburg, unpublished results). Fourth, the flavivirus effect on peptide transport could occur after TAP-mediated peptide translocation and influence the stability or export rates of translocated peptides. Net peptide accumulation in the ER is a function of TAP-mediated peptide import and peptide export through the Sec61p channel into the cytoplasm (12). Therefore, it could be envisaged that, for instance, delayed processing of the flavivirus polyprotein (14, 18, 41) may lead to a temporary inhibition of Sec61p channels that could result in enhanced peptide accumulation. However, most experiments with nonglycosylatable peptide showed an uninhibited export from cells in the early phase of flavivirus infection. Only occasionally did we observe a slightly retarded kinetics of peptide export (F. Momburg and M. Lobigs, unpublished results). Furthermore, only a small fraction of glycosylatable peptides, which we have shown here to be imported into the ER at an increased rate in flavivirus-infected cells, undergo retrotranslocation from SLO-permeabilized cells or microsomes (12). Therefore, we do not favor inhibition of peptide export as the major cause for the flavivirus-induced increased recovery of transported peptide.
The early increase of peptide import into the ER of flavivirus-infected cells is followed by the later reduction of the recovery of glycosylated peptide compared with that seen in uninfected cells. This late downregulation is consistent with the effect of infections with other viruses on peptide transport, probably due to the inhibition of host cell macromolecular synthesis or the cytopathogenicity of the infections. We have found, for instance, that infection of HeLa-S3 cells with VV-TK− for 16 h at a MOI of ∼2 reduces transport by >50% compared to that in uninfected cells; however, when a recombinant VV encoding human TAP1 and TAP2 was used in the same experiment, peptide transport was augmented by up to 80% of that in uninfected cells (M. Lobigs, unpublished result).
The pronounced decrease of peptide translocation in cells infected with flavivirus for 24 to 48 h contrasts with the upregulation of peptide-loaded MHC class I molecules on the cell surface seen at these time points (30). We suggest that the transient increase of peptide import early during infection augments the rate of peptide loading of MHC class I molecules in the ER and, in turn, the concentration of MHC class I on the plasma membrane and that the latter is apparent for a prolonged time period due to the long half-lives of peptide-loaded class I molecules at the cell surface (17). The biological function of upregulation of MHC class I-restricted antigen presentation in flavivirus-infected cells is not immediately apparent in terms of an advantageous strategy for the virus. The increased concentration of MHC class I-peptide ligand is expected to result in the more efficient elimination of virus-infected cells by Tc cells. On the other hand, an increase of class I MHC on the cell surface reduces susceptibility to NK cell-mediated lysis (5, 38). Consistent with the latter is the observation that flaviviruses are poor inducers of NK cells and that flavivirus infection reduces the susceptibility of cells to NK cell lysis. An escape of the mosquito-borne flaviviruses from the early NK cell immune response may be crucial for the production of a viremia of sufficient magnitude and duration for the transmission of the virus from an infected host to the arthropod vector during the blood meal.
ACKNOWLEDGMENTS
We thank M. Pavy and R. Tha Hla for excellent technical assistance.
F.M. was a recipient of an Australian Research Council International Fellowship, and M.L. was a recipient of a Travel Fellowship from the Australian Academy of Science.
FOOTNOTES
- Received 1 February 2001.
- Accepted 29 March 2001.
- Copyright © 2001 American Society for Microbiology
