Mechanism of antigen presentation after hypertonic loading of soluble antigens

Abstract

Hypertonic loading of proteins into cells has been used to introduce soluble proteins into the major histocompatibility complex class I pathway of antigen presentation followed by cytotoxic T-lymphocyte (CTL) induction. The precise mechanism for this pathway is not completely understood. The antigen is either processed and presented by/on the same cell or by professional antigen-presenting cells (APC) after taking up the antigen from damaged or apoptotic cells. After loading labelled ovalbumin (OVA), it could be co-precipitated with the proteasome complex, supporting the role of this pathway for antigen processing. The processing speed however, appeared to be slow since intact OVA could be detected inside the cells even after 18 hr. This corresponded well with the processing of OVA by isolated proteasomes. On the other hand, enough peptides for recognition of target cells by CTLs were generated in this reaction. One reason for the low level of processing might be that hypertonic loading may damage the cells and inhibit direct processing. In fact, at least 50% of the cells became positive for Annexin V binding after hypertonic loading which indicates severe membrane alterations usually associated with the progress of apoptosis. Annexin V binds to phosphatidylserine residues which also serve as ligand for CD36 expressed on monocytes and some immature dendritic cells. This may direct the phagocytic pathway to hypertonically loaded cells and thus enable professional APCs to present OVA-peptides. Therefore, in addition to the direct processing of OVA, CTLs can be primed by professional APC after uptake of apoptotic, OVA-loaded cells.

Introduction

The major histocompatibility complex (MHC) class I-restricted cytotoxic T-lymphocyte (CTL) response is usually reserved for the recognition of viral peptides which are produced and presented by infected cells.¹ Soluble proteins on the other hand are taken up by antigen-presenting cells (APC) and degraded in an endosomal compartment.² Peptides generated in this way are bound and presented by MHC class II molecules. For some soluble proteins, such as ovalbumin (OVA), it has been shown that cytoplasmic loading can generate a MHC class I-restricted CTL response.^3–5 In this regard cytoplasmic loading seems to mimic viral infection. For the presentation of virus-derived proteins two different pathways have been described. The intracellular processing of protein antigens by proteasomes with subsequent transport into the endoplasmic reticulum by the transporter associated with antigen presentation (TAP) followed by binding to MHC molecules is one possibility for direct presentation.⁶ In addition an indirect but very effective pathway has more recently been described where infected cells undergo apoptosis and are then taken up by dendritic cells which process and present the antigen.^7–10 Both pathways could also be effective in the presentation of OVA after hypertonic loading. To address this question, biotin-labelled OVA was hypertonically loaded into cells and its processing was followed. In addition, the influence of hypertonic loading on cell viability with special regard to the induction of apoptosis was investigated.

Materials and methods

Animals and cell lines

The murine thymoma cell line EL4 (H-2^b) was used throughout the experiments. C57BL/6 mice were purchased from Charles River (Sulzfeld, Germany).

Reagents and antibodies

OVA, sucrose, polyethylene glycol 1000 MW, and ethylenediaminetetraacetic acid (EDTA) were purchased from Serva Chemicals (Heidelberg, Germany). Streptavidin–agarose, Tris–HCl, phenylmethylsulphonyl fluoride (PMSF) were from Sigma (Deisenhofen, Germany). HEPES, RPMI-1640 and fetal calf serum were purchased from Gibco (Paisley, UK). Polyacrylamide and sodium dodecyl sulphate (SDS) were supplied by Pharmacia (Freiburg, Germany), and nitrocellulose was supplied by Schleicher & Schüll (Dassel, Germany). For biotinylation, biotin-ε-amino-caproic-acid-N-hydroxysuccinimide from Roche (Mannheim, Germany) was used.

Biotin labelling of OVA

OVA was biotinylated using biotin-succinimide in borate buffer at pH 8·8 according to a standard protocol. Unlabelled biotin was removed by extensive dialysis.

OVA loading of cells

EL-4 cells were grown in RPMI-1640 supplemented with glutamine, antibiotics and 5% fetal calf serum. Washed cells were loaded with biotinylated OVA as described previously.³ After extensive washing, the cells were either lysed immediately or cultured at 37° in a 5% CO₂ incubator. For lysis, cells were resuspended in a hypotonic solution [Tris–EDTA–dithiothreitol (DTT), 50–1–1 mm, containing azide and 0·2 mm PMSF] and homogenized on ice with a Dounce type homogenizer. The cell lysate was cleared by centrifugation at 15 000 g for 10 min. To the lysate, 50 µl of a preabsorbed 4% Streptavidin–agarose solution (Sigma) was added and the mixture was incubated at 4° under constant rotation for 2 hr. The agarose was spun down and washed three times with Tris-buffer. The pellet was resuspended in SDS–polyacrylamide gel electrophoresis (PAGE) sample buffer, boiled for 2 min and analysed on a 12% polyacrylamide gel under reducing conditions. After blotting to nitrocellulose, bands were analysed using a polyclonal rabbit anti-OVA (Cappel/ICN, Costa Mesa, CA) or antiproteasome antibody,¹¹ an alkaline phosphatase-labelled anti-rabbit antibody and developed with a highly sensitive chemiluminescent substrate (CSPD, Tropix, Bedford, MA).

Annexin V and propidium iodine labelling

At three time-points, immediately and 2 and 4 hr after loading, cells were washed and 1×10⁶ cells were incubated with propidium iodide and fluorescein isothiocyanate (FITC)-labelled Annexin V for 15 min at 37° according to the manufacturer's protocol (Annexin V Fluos, Roche, Mannheim, Germany). Cells were analysed on a fluorescence-activated cell sorter (FACS; FACSort, BD, San Jose, CA).

OVA processing in vitro

Proteasomes were isolated from EL4 as described.¹¹ Isolated 20S proteasomes were dialysed extensively against the reaction buffer (50 mm Tris–HCl/1 mm DTT pH 8·0). Proteasome (10 µg) and OVA (20 µg) were incubated in a 100-μl reaction for 4 hr at 37°. Thereafter the mixture was either analysed by PAGE/Western blot or separated by filtration through a filter with a molecular weight cut-off of 3000. All peptides smaller than this limit were used and tested in a conventional ⁵¹Cr-release assay with OVA-specific CTLs.

Cytotoxicity assay

The recognition of peptides produced by isolated proteasomes was tested in a standard 4-hr ⁵¹Cr-release assay. Briefly, 5×10⁶ target cells (EL4) were labelled with 100 µCi sodium ⁵¹chromate for 1 hr. Cytotoxicity was tested in the presence or absence of peptides eluted from the reaction above using CTLs specific for OVA. CTLs were induced by OVA-containing immunostimulating-ovalbumin complexes as described.^12,13 After 4 hr the supernatant was harvested with a filter system (Skatron, Lier, Norway), and the radioactivity was measured in a scintillation counter.

Results

Processing of OVA in vivo

Hypertonic loading of antigens has been used to introduce soluble antigens into the MHC class I pathway of antigen presentation. For direct presentation of the antigen by loaded cells the antigen has to be degraded and transported to the MHC complex. To follow the degradation of OVA in vivo OVA was hypertonically loaded into EL4 cells and the cells were incubated over a period of 4 hr and up to 18 hr. Thereafter, cells were lysed and equal amounts of total cell lysate (20 µg) were analysed on PAGE with subsequent blotting and development with an anti-OVA antibody. Using this protocol, a decrease in the intracellular OVA could be seen after 4 hr as well as after 18 hr, indicating the intracellular degradation of OVA (Fig. 1). Numerous proteolytic enzymes in the cells could be involved in this degradation. The proteasomes have been shown to be critically involved in antigen processing. To gain more information on their direct involvement we tested whether they form complexes with the loaded OVA.

Figure 1.

Ovalbumin (OVA) degradation in vivo was followed 2 and 18 hr after hypertonic loading of EL4 cells, The cell lysate was analysed on a 12% PAGE, blotted and analysed by polyclonal anti-OVA antibody. Marker lane 1, the OVA from the high molecular weight marker (Pharmacia) reacts with the antibody; lanes 2 and 4, control cell lysate immediately after loading; lane 3, after 2 hr; and lane 5, after 18 hr.

To test this, biotinylated OVA was loaded into the cytosol of EL4 cells. Immediately and 2 and 4 hr after loading, cells were lysed and the biotin labelled OVA was precipitated with Streptavidin–agarose, separated on PAGE and transferred to nitrocellulose. The nitrocellulose was then probed with an antibody which recognizes a subunit of the proteasome complex¹¹ and another one which detects OVA. As shown in Fig. 2, the tagged OVA could be co-precipitated with two subunits of the proteasome complex. The subunits can be detected in the precipitate immediately after loading, which is about 15 min after the start of the incubation, as well as after 2 and 4 hr. The control staining with the anti-OVA antibody clearly shows that intact OVA is present over the entire experimental time (Fig. 2, lanes 2–4). Within the 4 hr experiment the intensity of the OVA band decreased only slightly, indicating slow processing.

Figure 2.

Isolation of biotinylated OVA from EL-4 cells immediately after loading (lanes 2 and 5) and after 2 hr (lanes 3 and 6) and 4 hr (lanes 4 and 7). The OVA was precipitated from the cell lysate using Streptavidin–agarose and separated by 12% PAGE under reducing conditions. After blotting, nitrocellulose was assayed with polyclonal antibody to OVA (lanes 1–3) or proteasomes (lanes 4–6). Lane 1, size markers stained with India ink (67 000, 43 000, 30 000 and 20 000 MW).

Processing of OVA in vitro

To prove the direct involvement of proteasomes in OVA processing, isolated proteasomes from EL4 cells were incubated with defined concentrations of OVA and the processing was monitored by the disappearance of the OVA band on silver-stained gels. As shown in Fig. 3 the OVA band became slightly less intense after 2 hr (lane 1). Even after 18 hr the OVA band was still visible (lane 4) without a dramatic change in the intensity compared with the 2-hr appearance. Processing of OVA by isolated proteasomes was therefore not very effective in vitro.

Figure 3.

OVA processing in vitro: 12% PAGE with silver staining. Lane 1, OVA incubated with isolated proteasome for 2 hr; lane 4, after 18 hr incubation. Lanes 2 and 3 are OVA incubated only in the reaction buffer for 2 and 18 hr, respectively The OVA band at approximately 43 000 and the proteasome bands at 25 000 to 35 000 MW are clearly visible.

Recognition of peptides produced by isolated proteasomes by CTL

To confirm that proteasomes produce in vitro peptides which are able to sensitize target cells for cytotoxic attack by specific T lymphocytes, OVA was incubated with isolated proteasomes for 4 hr at 37°. This time was chosen since CTLs have been shown to recognize loaded cells within this time. Thereafter the reaction mixture was separated into proteasomes/intact OVA and processed peptides on the other hand by a membrane filter with a 3000-MW cut-off. All molecules smaller than 3000 MW were added in several dilutions to EL4 target cells. OVA-specific CTL were added and the cytotoxicity was measured in a chromium-release assay. As shown in Fig. 4 EL4 cells incubated with the peptides were readily lysed by OVA-peptide-specific CTLs. The reaction was effector/target (E/T) and concentration dependent and clearly shows that peptides produced from OVA by isolated proteasomes can sensitize target cells. Incubation of EL4 with intact OVA induced a specific cytotoxicity below 5% and was subtracted.

Figure 4.

Recognition of peptides produced by the incubation of isolated proteasomes from EL4 cells. EL4 cells were incubated with peptides isolated from the co-incubation of OVA with isolated proteasomes as described. In (a) the experiments were performed at different effector to target (E : T) ratios and in (b) the peptides were diluted. ▪ indicates a control with CNBr-digested OVA at 10 µg/ml.

Morphological changes after hypertonic loading of cells

In addition to direct processing and presentation of peptides there are several recent reports which support the view of cross-presentation as one main pathway in the induction of CTL by soluble antigens.^14,15 For this, the antigen has to be taken up by professional APC after cell lysis or in a particulate form. More recently it has been shown that after viral infection apoptotic cells are preferentially taken up by dendritic cells which are amongst the most potent APC.⁷ Since hypertonic loading is a stressful procedure for cells it may end with the induction of apoptosis in some of the cells. To test this, EL4 cells were hypertonically loaded and analysed after 2 and 4 hr for the binding of FITC-labelled Annexin V staining as an indicator of membrane disturbance and the beginning of apoptosis and propidium iodide, which indicates the more severe damage seen in apoptosis and necrosis. Figure 5 shows the changes in the side-scatter profile of cells after loading with OVA in the hypertonic solution. After 2 hr two clear-cut, different cell populations could be separated with an increase in side scatter. In the cells with a higher side scatter about one-half of the cells stained with FITC-Annexin alone or were double positive for Annexin V and propidium iodide (bottom row). After 4 hr in culture the two cell populations were even more pronounced and the cells with a higher side scatter were now mostly propidium iodide and Annexin V double positive, which is a hallmark of necrotic cells.

Figure 5.

Propidium iodide (PI) and FITC-Annexin V staining of EL4 cells immediately after OVA-loading (first column) and 2 hr and 4 hr after loading. Forward- and side-scatter patterns are shown in the first row and staining for R1 (first cloud) and R2 (second cloud) is illustrated in rows 2 and 3, respectively. Immediately after loading there were damaged cells in R1 which increased over time. The undamaged cells in R2 are PI and Annexin V negative and can be better separated after 2 and 4 hr.

Discussion

There are several strategies to introduce soluble proteins into the MHC class I restricted pathway of antigen presentation. These include the application of the antigen in a particulate form or together with lipids.^1,12,13,16 All of these methods facilitate the uptake of the antigen by professional APC which then present the antigen. The hypertonic loading of cells seems to be somehow different from these protocols in that usually non-professional APC can be used for loading and CTLs recognize the antigenic peptides directly on the target cells. The cells must therefore have processed the antigen properly and presented it correctly. This processing pathway includes the proteasomes, as has been shown by the sensitivity to blockade by proteasome inhibitors.¹⁷ The co-precipitation of OVA with proteasomes after hypertonically loading the antigen as shown here is a further substantiation of the functional role of proteasomes in the processing of the antigen. This co-precipitation shows some specificity as hypertonic loading of human as well as rat immunoglobulin was not nearly as efficient in forming a co-precipitate. The follow-up after loading shows that the amount of intact OVA declines after 4 hr, although intact OVA is still visible after 18 hr in intact cells. Hence, this conventional processing seems not to be very effective, which corresponds to the data on the in vitro processing of OVA by isolated proteasomes from these cells. There are reports on the processing of ovalbumin by isolated proteasomes from bovine erythrocytes which show that the correct peptides are produced although no clear data on processing speed can be extrapolated.¹⁸ In most cases isolated proteasomes were tested not with intact proteins but with fragments of ovalbumin not exceeding 30 amino acids or with model substrates.^19–21 Another explanation for the slow processing in vitro might be that isolated proteasomes have to be activated by extensive dialysis against the reaction buffer under reducing conditions. DTT, like other reducing agents, is quite unstable, it is therefore possible that the proteasomes become inactive over a prolonged period of incubation. An additional reason for the slow reaction in vitro might be that the 20S proteasomes show a preferential degradation of denatured or elongated proteins because denatured proteins fit best into the barrel-shaped proteasome.^22–25 It is not clear however, which alterations of the protein, e.g. digestion or denaturation, occur within the cell after hypertonic loading.

Although the processing speed in vitro is not very high, enough peptides for sensitizing the target cells were produced. In addition, loaded cells can be lysed by CTLs directly, which indicates that processed antigen must be presented on the cell surface.^3,5 The question is therefore whether this is the only pathway for antigen presentation after hypertonic loading.

More recent data suggest that cross priming is the major pathway for antigen presentation after hypertonic loading of OVA into cells.^15,26 In these experiments however, loaded spleen cells were given intraperitoneally which bears both possibilities for antigen presentation directly by loaded cells as well as uptake of loaded cells by professional APC. In this regard it is interesting that most immunization protocols include restimulation in vitro where either irradiated OVA transfected cells or OVA-loaded cells were used for stimulation in the presence of spleen cells.

This combination of cells therefore consists of dendritic cells and T cells in the spleen cell population plus the irradiated EL4-OVA which probably undergo apoptosis. The uptake of these cells might be the important trigger for the induction of the CTL response. The uptake of apoptotic cells by professional APCs has been shown to induce a strong CTL response in virus-infected cells and the hypertonic loading might mimic this in several ways.⁶ The antigen is therefore not only like a viral protein present in the cytosol of the infected cell but also the hypertonic stress may induce the start of the apoptotic programme. In the experimental design presented here, about 50% of the loaded cells show signs of apoptosis after 4 hr. It is interesting to note that many of these cells bind Annexin V, which shows preferential binding to the phosphatidylserine residues exposed in cells undergoing apoptosis. Phosphatidylserine residues on the other hand are a ligand for CD36 which is expressed on macrophages and some dendritic cells. Macrophages use this receptor interaction in phagocytosis whereas dendritic cells seem to prefer the usage of the α_vβ₃-integrin for the uptake of apoptotic bodies.^27–29 For the presentation of OVA to CD4 T lymphocytes it has been shown that dendritic cells sensitized in vitro are extremely powerful APC.³⁰ This pathway might therefore be effective, especially after immunization in vivo. From the receptors mentioned neither CD36 nor the scavenger receptor is expressed on the EL4 cells as shown by polymerase chain reaction (data not shown) so other receptors might be involved. The uptake of antigen released by apoptotic or necrotic cells by macropinocytosis could be an additional form of antigen uptake ending in presentation as originally suggested for antigen loading.^8,16,31,32 In addition it has been proposed that lysed cells release endogenous adjuvants which help to stimulate CTL, although this process has not been exactly defined.³³ For antigen processing, a protease different from proteasomes might even be involved.^34,35 In conclusion, the data presented here show the processing of OVA by proteasomes after hypertonic loading. The induction of apoptosis and necrosis in loaded cells offers a distinct mechanism for antigen presentation in this set-up where uptake of loaded and apoptotic cells by professional APCs is a good stimulus of the sensitivity of the CTL response, at least in vivo.