SEARCH

SEARCH BY CITATION

Keywords:

  • antigen cross-presentation;
  • dendritic cell;
  • MHC class I;
  • microspheres;
  • phagocytosis

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Experimental procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Cross-presentation is the presentation by MHC class I of antigenic peptides from exogenous proteins that have been internalized and processed by professional antigen-presenting cells, e.g. dendritic cells. We have investigated the influence of particle size and antigen load on cross-presentation following antigen delivery on microspheres (MS). Cross-presentation from small particles (0·8-μm) is sensitive to proteasome inhibition and the blockade of endoplasmic reticulum-resident MHC class I complex export, whereas cross-presentation from larger particles (aggregated clumps of 0·8-μm MS) is resistant to these antagonists. This observation may have been overlooked previously, because of the heterogeneity of particle size and MS uptake in unsorted dendritic cell populations. Larger particles carry more antigen, but we show that antigen load does not influence the cross-presentation pathway used. Whereas early endosome autoantigen 1 (EEA1) could be observed in all phagosomes, we observed endoplasmic reticulum SNARE of molecular weight 24 000 (ERS24) and cathepsin S in association with 3·0-μm and aggregated 0·8-μm MS, but not individual 0·8-μm MS. A potential mechanism underlying our observations may be the activation of β-catenin by disruption of E-cadherin-mediated adhesion. Activated β-catenin was detected in the cytoplasm of cells after phagocytosis of MS (highest levels for the largest particles). We propose that particle size can direct the use of different pathways for the cross-presentation of an identical antigen. Furthermore, these pathways have differing yields of MHC class I–peptide complexes, which is an important variable in designing vaccination strategies for maximal antigen expression and CD8+ T-cell priming.


Abbreviations:
BFA

brefeldin A

DC

dendritic cells

EEA1

early endosome autoantigen 1

ER

endoplasmic reticulum

ERS24

endoplasmic reticulum SNARE 20 000 molecular weight

MG132

carbobenzoxy-l-leucyl-l-leucyl-l-leucinal

MS

microspheres

OVA

ovalbumin

TAP

transporter associated with antigen processing

w/v

weight/volume

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Experimental procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Dendritic cells (DC) are believed to be the key immune activators of naive CD8+ cytotoxic T lymphocytes, leading to recognition and elimination of virus-infected cells or tumour cells. The DC accomplish immune priming by alerting cytotoxic T lymphocytes to foreign or altered self proteins, in a process known as antigen presentation, and by delivering co-stimulatory signals.1 All nucleated cells, including DC, display a snapshot of intracellular proteins as peptides bound to MHC class I molecules at the cell surface. The peptides derive from senescent proteins or defective ribosomal products synthesized by the host cell and degraded by proteasomes.2 A small fraction of these peptides escape further processing and enter the endoplasmic reticulum (ER) using the transporter associated with antigen processing (TAP), where the peptide-loading complex facilitates the binding of peptides to nascent MHC class I molecules.3 Loaded MHC class I complexes then reach the cell surface through the secretory pathway. Viruses may infect DC, allowing priming through direct presentation of endogenously synthesized foreign protein.4 However, DC are specialized in their ability to capture exogenous antigen for display on MHC class I molecules; this process is termed cross-presentation.5

Since its first description,6 the mechanisms of cross-presentation have been extensively investigated and two major pathways have been delineated, with others operating under certain experimental conditions. First to be described was the ‘phagosome-to-cytosol’ pathway, which required the action of the proteasome and TAP, and appeared to intersect with the classical direct presentation pathway.7 The second pathway to be identified, was both proteasome- and TAP-independent,8,9 and became known as the ‘vacuolar pathway’. Subsequent experiments showed the importance of intra-phagosomal cathepsin S in generating MHC class I-presented peptide complexes via this pathway.10

In the phagosome-to-cytosol model, antigenic peptides enter the ER using TAP and are assembled with MHC class I molecules, before transport to the cell surface through the secretory pathway. Egress of protein cargo, including peptide–MHC class I complexes, from the ER to the Golgi apparatus is blocked by the fungal antibiotic brefeldin A (BFA).11 However, a modified phagosome-to-cytosol model has been proposed, where the ER forms an active component of the phagosomal membrane, bringing the antigen-processing machinery to the phagosome;12–14 this mechanism is controversial.15,16 Here, antigenic peptides are returned to an ER phagosome, where they are loaded onto MHC class I molecules and transported to the cell surface. In this scenario, transport bypasses the Golgi apparatus, so BFA is not expected to block cross-presentation.14

The reported sensitivity of cross-presentation to BFA inhibition is variable, ranging from a complete block in bone-marrow-derived macrophages7,17 and DC2.4 dendritic-like cells,18 through only partial (35%) inhibition in BMA3.1A macrophages14 and minor inhibition with prolonged exposure to high concentrations in peritoneal macrophages,8 to no inhibition in peritoneal macrophages.8,19 Such studies have also made use of a wide range of particulate antigen, including 1-μm latex,19 2-μm latex,8 iron oxide,18 1-μm latex and 0·5-μm to 3-μm biodegradable17 and size-unspecified latex microspheres (MS).14 Often, experiments to isolate phagosomes for proteomic analyses use 0·8-μm particles,12–14,20 whereas microscopy studies use 3-μm particles.13,14 Meanwhile, Becker et al.21 found that ER membrane was selectively recruited to the phagosome when J774 macrophages phagocytosed large particles (3·0-μm and aggregated clumps of 0·8-μm MS), but not small particles (dispersed 0·8-μm MS). The authors suggested that the ER serves as a reserve to supply the membrane when cells ingest either extremely large particles, or a large dose of small particles.

We set out to investigate whether the reported variability in sensitivity of cross-presentation to BFA might reflect heterogeneity in the engagement of different antigen-processing pathways within experimental systems, influenced by the size and mode of delivery of model antigens. We have selectively analysed the influence of particle size in directing the cross-presentation pathway used by the antigen ovalbumin (OVA) for the generation of Kb-SIINFEKL MHC class I complexes. Furthermore, we have measured the epitope density of cross-presented complexes and compared the relative yield of cross-presentation between different antigen particle sizes.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Experimental procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Cell culture

The murine dendritic-like cell line, DC2.4,18 was cultured in RPMI-1640 medium supplemented with 10% fetal calf serum, penicillin, streptomycin and l-glutamine (R10). The B3Z T-cell hybridoma,22 which responds to the OVA peptide SIINFEKL bound to the murine Kb MHC class I molecule, was cultured in complete medium (R10 plus 1 mm pyruvate and 50 μmβ-mercaptoethanol).

Ovalbumin

Ovalbumin (Calbiochem, Nottingham, UK) was dialysed extensively against PBS at 4° to remove any contaminating SIINFEKL peptide, then frozen at −20° in single-use aliquots. Samples of dialysed OVA were tested for endotoxin contamination using Limulus Amoebocyte Lysate, according to the manufacturer’s instructions (Charles River Endosafe, Charleston, SC).

Microspheres

Yellow-green fluorescent latex 0·8-μm or 3-μm diameter MS were obtained from Polysciences Europe GmbH (Eppelheim, Germany). The MS were coated with dialysed OVA by passive adsorption and then opsonized with mouse anti-OVA antibodies (a gift from A. Tutt, Tenovus Research Laboratory, University of Southampton). Opsonized MS were tested for endotoxin contamination as described above. Opsonized MS were disaggregated by sonication immediately before use. The surface concentration of OVA on coated MS was estimated by extracting protein using PBS/1% weight/volume (w/v) SDS for 2 min at 95°, followed by a second extraction at room temperature. The extracted protein concentration was measured using a Micro BCA Protein Assay kit (Pierce: Perbio Science UK, Cramlington, Northumberland, UK) according to manufacturer’s instructions. MS were counted using a haemocytometer after sonication and the number of OVA molecules per MS was calculated (more details are given in Supporting Information).

Phagocytosis of microspheres

DC2.4 were plated in six-well culture plates and unless stated otherwise, given an OVA pulse consisting of either: 5 μl sonicated 0·8-μm MS, 10 μl sonicated 3-μm MS or 20 μl unsonicated 0·8-μm MS Plates were centrifuged for 3 min, 200 g at room temperature and immediately incubated at 37°, 5% CO2 for 30 min. After cooling and extensive washing to remove unbound MS, cells were chased for a further 4–6 hr at 37°, 5% CO2, as indicated. Cells were then lightly fixed with 1% w/v formaldehyde and incubated with B3Z reporter T cells O/N at 37°, 5% CO2. Hydrolysis of the chromogenic substrate chlorophenolred-β-d-galactopyranoside by β-galactosidase was used as a measure of antigen presentation and T-cell activation (more details are given in Supporting Information).

For inhibitor experiments, cells were pre-incubated as follows: 1 μm MG132 (Calbiochem, Nottingham, UK), 30 min, 37°, or 10 μm cytochalasin D (Sigma-Aldrich, Dorset, UK) 30 min, 37°, or corresponding volumes of DMSO. Inhibitors were also present during the chase incubation. The BFA (Sigma-Aldrich) was added at 10 μg/ml to the chase incubation only, as it was found to interfere, occasionally significantly, with phagocytosis. At concentrations > 1 μm, MG132 was toxic to DC2.4. To test the effect of the inhibitors on direct antigen presentation, 0·2 μm-filtered OVA was introduced directly to the cytoplasm of DC2.4 by hypotonic lysis of pinosomes.23

Sorting cells by flow cytometry

Fixed DC2.4 cells were sorted using a FACSAria flow cytometer (BD Biosciences, Oxford, UK) in the University of Southampton, School of Medicine Flow Cytometry Unit. Cells were interrogated with the 488-nm laser, and collected through a 100-μm nozzle into 15-ml polypropylene tubes coated with 1 ml PBS/2 mm EDTA/0·1% w/v BSA. Sorted cells were recovered by centrifugation in a swing-out rotor at 200 g for 5 min at 4°, and resuspended in 200 μl to 1 ml complete medium, before counting and plating in triplicate with B3Z in a 96-well plate.

Estimation of cross-presentation yield

DC2.4 cells were loaded with a range of concentrations of exogenous SIINFEKL peptide. Kb-SIINFEKL complexes were measured by (i) staining unfixed cells with the Kb-SIINFEKL-specific monoclonal antibody (a gift from J. Yewdell,24) and an Alexa Fluor 647 F(ab′)2 fragment, (ii) incubating fixed cells with B3Z overnight and (iii) using an epitope quantification kit (QIFIKIT, Dako UK Ltd, Ely, UK) to calibrate the Alexa Fluor 647 F(ab′)2 fluorescence to correlate antibody staining with B3Z response (full details are given in Supporting Information).

Microscopy

Images were prepared in the University of Southampton Biomedical Imaging Unit, using Leica SP2 and SP5 laser scanning confocal microscopes with × 40 and × 100 objectives. After pulsing DC2.4 cells with opsonized MS and chasing for the specified times, cells were fixed, permeabilized and stained as indicated in the figure legends. Primary antibodies were: rabbit anti-ERS24 (Osenses, Flagstaff Hill, Australia), goat anti-cathepsin S (Abcam, Cambridge, UK); goat anti-EEA1 and rabbit anti-c-fos (Santa Cruz Biotechnology, Heidelberg, Germany). Actin was stained with phalloidin conjugated to Alexa Fluor 647 (Molecular Probes, Invitrogen Detection Technologies, Renfrewshire, UK). Secondary antibodies were: goat anti-rabbit F(ab′)2 fragment-Alexa Fluor 488 and chicken anti-goat-Alexa Fluor 594 (Molecular Probes). Nuclear dyes were TO-PRO-3 and SYTOX Blue (Molecular Probes).

Immunoblotting

DC2.4 cells were seeded sparsely in six-well culture plates and allowed to grow undisturbed for 2 days. Opsonized MS were added to each well, but were not centrifuged onto the cells. After 1 hr at 37°, the plates were transferred to ice and the cells were gently rinsed with cold RPMI-1640 to remove excess MS. Cells were harvested on ice by gentle pipetting and were counted. A sample was reserved for flow cytometry analysis, and the remainder were incubated for 1 hr on ice in lysis buffer [Tris-buffered saline, pH 8·0 containing 0·1% w/v saponin, 1 × protease inhibitor cocktail (Roche, Hertfordshire, UK), 1 × phosphatase inhibitor cocktail III (Calbiochem, Merck Chemicals Ltd, Nottingham, UK)], then centrifuged at 20 000 g for 10 min at 4° to sediment nuclei, membranes and MS. The supernatant (cytosol) was removed for SDS–PAGE analysis. Supernatants equivalent to 300 000 cells were separated on a 12·5% polyacrylamide gel, transferred to nitrocellulose and probed with anti-active β-catenin, clone 8E7 (Millipore, Temecula, CA) and anti-GAPDH (Abcam). Blots were developed with SuperSignal West Femto ECL substrate (Pierce).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Experimental procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Heterogeneous uptake of microsphere-associated antigen by dendritic cells

We measured cross-presentation by the DC-like cell line, DC2.4, from OVA-coated 0·8-μm fluorescent MS. As previously reported,21 small MS are prone to form large aggregates, so we sonicated a portion to disperse them and compared cross-presentation and phagocytosis of equal numbers of unsonicated and sonicated MS (Fig. 1a,c,d). Treatment with BFA inhibited cross-presentation by 50% relative to the fixed cell controls when cells were supplied with sonicated MS, but not at all when cells phagocytosed the aggregated MS. By contrast, the proteasome inhibitor carbobenzoxy-l-leucyl-l-leucyl-l-leucinal (MG132) inhibited cross-presentation by both samples to a similar degree (85% sonicated; 78% unsonicated). Fixed cells could present exogenously loaded SIINFEKL (not shown), but when incubated with MS gave only background signals, indicating that the MS were not contaminated with significant levels of free SIINFEKL, and that serum proteases did not liberate SIINFEKL during the experiment. Dialysed OVA typically contained < 0·5 endotoxin units/mg, and the endotoxin content of opsonized microspheres was below 2 × 10–9 endotoxin units per 0·8-μm microsphere or 2 × 10−8 endotoxin units per 3·0-μm microsphere. Furthermore, DC2.4 cells are markedly less sensitive to stimulation by lipopolysaccharide compared with bone-marrow-derived DC (not shown). We checked that cross-presentation was dependent on phagocytosis. Cytochalasin D-treated live cells did not phagocytose MS, but did bind a number of MS at the cell surface for the duration of the assay (not shown). The relatively small number of Kb-SIINFEKL complexes detected in these samples may have arisen from the action of cell-surface proteases upon these surface-bound MS.25 Direct presentation of excess soluble OVA was completely inhibited by 10 μg/ml BFA, and partially inhibited by 1 μm MG132 (Fig. 1b).

image

Figure 1.  Heterogeneous uptake of 0·8 μm antigen. (a) DC2.4 cells were pulsed with either sonicated or unsonicated 0·8-μm ovalbumin (OVA) -coated microspheres (MS) equivalent to 5 μg OVA for 30 min and chased for 4 hr. Presentation of SIINFEKL was measured using the B3Z T-cell reporter line. Cells were treated with the indicated inhibitors, as described in Experimental Procedures. (b) The efficacy of brefeldin A (BFA) and MG132 in blocking direct presentation were tested by introducing an excess (10 mg/ml) of soluble OVA to the cytoplasm by hypotonic lysis of pinosomes. (c) Unsonicated 0·8-μm MS are phagoytosed as larger, aggregated particles, as seen by confocal microscopy and flow cytometry. MS are shown in green; nuclei in blue. Fluorescence peaks corresponding to cells containing 1, 2 or > 100 MS are marked on the histogram. Total antigen uptake (arbitrary fluorescence units) can be estimated by calculating the area under the histogram, and is shown in the corner. (d) Sonicated 0·8-μm MS are phagocytosed mainly as individual particles. (e) DC2.4 cells were pulsed with sonicated 3·0-μm OVA-coated MS and treated as above. (f) Cells containing MS were analysed by confocal microscopy and flow cytometry as in (c) and (d).

Download figure to PowerPoint

The observed sensitivity of cross-presentation to MG132 inhibition in Fig. 1(a) suggested that the OVA requires cytosolic processing by proteasomes, and BFA inhibition implied that SIINFEKL binds to nascent MHC class I complexes that traffic to the cell surface through the Golgi apparatus. Our initial results therefore suggested that sonicated MS were processed by a route with similar characteristics to the phagosome-to-cytosol pathway, whereas unsonicated MS appeared to be processed in a proteasome-dependent, but BFA-independent manner, consistent with the proposed ER-phagosome.

Confocal microscopy and flow cytometry showed a clear difference in the number and distribution of MS phagocytosed between aggregated (Fig. 1c) and sonicated (Fig. 1d) MS. Cells given aggregated MS phagocytosed more MS per cell, and in many cells, the MS were present as large clumps, whereas for the sonicated sample, the majority of phagocytosed MS appeared as single particles. The number of MS phagocytosed can be calculated from the total cell-associated fluorescence. In the experiment shown, DC given aggregated MS contained seven times as many MS as DC supplied with sonicated MS. It is clear that the uptake of small particle-associated antigen by a population of cells is highly heterogeneous in quantity and quality (particle size), depending on even such a simple factor as the aggregation state of the supplied antigen.

Large, 3-μm MS have been used in antigen presentation studies involving microscopy,13,14 and we chose them because their size is better defined (equivalent to approximately 52 of the 0·8-μm MS) than the aggregates of small particles. We incubated DC with OVA-coated 3-μm MS in the presence of BFA. In contrast to the notable effects of BFA upon sonicated 0·8-μm MS-based antigen delivery, this only led to a 17% reduction in cross-presentation (Fig. 1e). Proteasomal inhibition resulted in a 72% reduction in cross-presentation, an effect similar to that seen for the smaller MS. Although 3-μm MS are less prone to aggregation, we sonicated them before adding them to the cells. Microscopy and flow cytometry showed again that phagocytosis of the MS is heterogeneous, with cells containing single, two or multiple MS (Fig. 1f).

After these initial findings we tested BFA and proteasome inhibitors in multiple experiments, and found the effects on cross-presentation to be highly variable. For 0·8-μm MS, BFA inhibition ranged from zero to 89% (mean 43%, SD 26%, n = 12) and for 3-μm MS inhibition ranged from zero to 84% (mean 39%, SD 26%, n = 15). For 0·8-μm MS, MG132 and lactacystin inhibition ranged from zero to 39% (mean 16%, SD 15%, n = 6) and for 3-μm MS, from zero to 51% (mean 17%, SD 29%, n = 3). We reasoned that such wide ranges may relate to the heterogeneous nature of the DC population that had captured particles of differing sizes. In other words, multiple cross-presentation pathways may operate simultaneously within a population of cells, depending on the size of particles captured. In this scenario, an instance of apparent resistance of cross-presentation to BFA or MG132 may indicate the dominance of a vacuolar-like processing pathway within the DC population. We therefore thought it important to measure the effects of these inhibitors on highly homogeneous populations of DC that had phagocytosed defined numbers of MS.

Distinct cross-presentation phenotypes identified within populations of dendritic cells containing microsphere-associated antigen

To counter the variability in phagocytosis, we purified homogeneous populations of DC containing specified numbers of 0·8-μm MS and so specified antigen dose. The uniform size and fluorescence intensity of the MS enabled us to select cells through fluorescent sorting (Fig. 2a). We collected cells containing no MS (‘Null’), a single MS (‘Single’) and cells containing 100 or more MS (‘Aggregates’). The size of Aggregates was less defined, because it was not possible to say whether the MS were present as a single large particle, or several smaller ones. Microscopic analyses suggested that the former situation dominated. To enhance the number of cells in the Aggregates population, these samples received a larger starting dose of (unsonicated) MS.

image

Figure 2.  Sorting of dendritic cells (DC) into homogeneous populations. (a) DC2.4 cells were pulsed with sonicated and unsonicated 0·8-μm ovalbumin (OVA) -coated microspheres (MS) for 30 min and chased for 6 hr in the presence and absence of brefeldin A (BFA), fixed and then sorted by flow cytometry. Total: fluorescence profile of cells before sorting; Null: cells without MS; Single: sorted cells containing a single 0·8-μm MS; Aggregates: sorted cells containing 100 or more 0·8-μm MS. (b) Cross-presentation by the sorted cell populations described above, measured using the B3Z reporter T cells. (c) Cross-presentation by sorted cell populations after incubating DC2.4 cells in the presence and absence of MG132, as described in Experimental Procedures.

Download figure to PowerPoint

Cross-presentation by the Single population was sensitive to BFA inhibition, whereas cross-presentation by the Aggregates population was insensitive to BFA (Fig. 2b). These results were consistent over multiple experiments: for the Single population the mean inhibition by BFA was 82% (SD 16%, n = 7), and for the Aggregates, the mean inhibition was 22% (SD 19%, n = 10). Hence, the extremes of particle size are associated with distinct requirements for export of Kb-SIINFEKL complexes through the Golgi apparatus. Differing sensitivity to BFA inhibition suggests that newly synthesized MHC class I molecules are required for cross-presentation from single 0·8-μm MS and additional sources used for larger particles.

Cross-presentation by the Single population was completely blocked by MG132, whereas cross-presentation by the Aggregates population was unaffected (Fig. 2c). In four experiments, one using lactacystin instead of MG132, cross-presentation by DC containing 100 or more MS was inhibited by only 4·5% (SD 5%). The MG132 used in these experiments inhibited direct presentation of excess OVA (see Supplementary material, Fig. S1). In all experiments, fixed cell controls for non-specific SIINFEKL loading were undertaken, performing mock incubations with OVA-coated MS and, separately, supernatants from MS-pulsed live cells. In no instance were B3Z T cells activated by the control samples (not shown).

Using sorted, homogeneous cell populations, a distinct pattern of inhibition emerged. Sensitivity to proteasome inhibition and requirement for export from the ER suggests that OVA supplied on single, 0·8-μm MS may be processed by the phagosome-to-cytosol pathway. On the other hand, lack of a requirement for proteasomal processing and export from the ER for OVA supplied on very large particles is more consistent with processing by the vacuolar pathway. The misleading results from our initial experiments with unsorted cells, therefore, most likely arose from the operation of more than one antigen-processing pathway within the population.

Surprisingly, the Null populations were sometimes able to present SIINFEKL. Contamination of the samples with free SIINFEKL was unlikely, because fixed cell controls indicated background levels of free peptide (not shown) and because the presentation in these Null cells was sensitive to BFA and MG132. The SIINFEKL presentation signal and the cell background fluorescence intensity were both greater for the Null population of the unsonicated sample, in line with the larger number of MS added. It is possible that OVA immune complexes and fluorescent dye entered cells separately from the particles. Possible mechanisms include ‘nibbling’ of plasma membrane26 from cells containing MS, or direct entry of OVA to the lumen of the ER.27

As the size of the aggregated 0·8-μm MS cannot be accurately defined, we repeated the above experiments using 3-μm MS (Fig. 3a), collecting Null and Single populations after treatment with BFA and MG132. The BFA did not inhibit cross-presentation by the Single population shown in Fig. 3(b), and from seven separate experiments, inhibition was modest (mean 23%, SD 27%). Inhibition by MG132 was maximally 88% (Fig. 3c), with a mean inhibition of 54% (SD 32%) in three separate experiments. We concluded that BFA and MG132 caused intermediate inhibition for cross-presentation of antigen supplied on 3-μm MS. This might suggest that a mixture of antigen-processing pathways, including phagosome-to-cytosol and vacuolar or ER–phagosome routes, might be adopted for particles intermediate in size between 0·8 μm and large aggregates.

image

Figure 3.  Sorting of cells containing single, 3·0-μm microspheres (MS). (a) DC2.4 cells were pulsed with sonicated 3·0-μm ovalbumin (OVA) -coated MS for 30 min and chased for 6 hr in the presence and absence of brefeldin A (BFA), fixed and then sorted by flow cytometry. Total: fluorescence profile of cells before sorting; Null: cells without MS; Single: sorted cells containing a single 3·0-μm MS. (b) Cross-presentation by the cell populations described above, measured by B3Z reporter T cells, in the presence or absence of BFA inhibition. (c) As (b), but cells were incubated in the presence or absence of MG132.

Download figure to PowerPoint

The effect of antigen load on microsphere-facilitated cross-presentation

The apparent relationship between particle size and cross-presentation phenotype prompted us to explore the role of antigen load as a possible contributing/confounding factor to our observations. We questioned whether high protein load, rather than particle size per se might be the reason for the observed insensitivity of cross-presentation by the Aggregates population. We prepared MS with ∼ 196 000 OVA molecules (High) and ∼ 3700 OVA molecules (Low) per MS. Both were supplied to DC without previous sonication, and cells were again sorted for populations containing High [OVA] and Low [OVA] aggregates (Fig. 4). Cross-presentation by DC containing High [OVA] aggregates was not significantly inhibited by BFA, whereas cross-presentation by DC containing Low [OVA] aggregates was inhibited by 29%. Over three independent experiments, the mean inhibition for highly loaded aggregates was 12% (SD 16%) and for reduced load aggregates it was 26% (SD 5%). Hence, a 50-fold reduction in the antigen load had only a minor effect on the sensitivity of cross-presentation to BFA, suggesting that nascent MHC class I molecules are not required for the processing of OVA supplied on very large particles, irrespective of protein load.

image

Figure 4.  Effect of decreasing antigen load on the sensitivity of cross-presentation to brefeldin A (BFA) and proteasome inhibition. (a) 0·8-μm microspheres (MS) were coated with ovalbumin (OVA) at the usual concentration (‘High’, ∼ 196 000 molecules per particle) or (b) at a reduced concentration (‘Low’∼ 3700 molecules per particle). DC2.4 cells were sorted by flow cytometry after a 20-min pulse and 5-hr chase in the presence/absence of inhibitors. Cross-presentation of OVA was measured using B3Z reporter T cells.

Download figure to PowerPoint

In the same experiment, cross-presentation from the High [OVA] aggregates was inhibited 18% by MG132 and the Low [OVA] by 40%. Reducing the protein load on the aggregated MS did not markedly alter the sensitivity of cross-presentation to the proteasome inhibitor, suggesting that antigen processing is more closely associated with particle size than absolute protein dose delivered in this system.

Immunofluorescence reveals different co-localization patterns between small and large microspheres

We asked whether our phenotypic observations were supported by differences in co-localization of endosomal/lysosomal marker proteins. We examined DC2.4 cells after pulsing with OVA-coated MS: sonicated 0·8-μm MS, sonicated 3·0-μm MS and aggregated 0·8-μm MS. Early endosome autoantigen 1 (EEA1) transiently associates with phagosomes and is required for phagosomal maturation.28 Actin polymerization and depolymerization under the plasma membrane contribute to driving phagocytosis.29 We could observe both EEA1 and F-actin, often co-localizing, around all three sizes of MS, as shown in Fig. 5.

image

Figure 5.  Confocal microscopy of DC2.4 after phagocytosis of microspheres (MS): early endosomal markers. Cells were supplied with MS coated with ovalbumin (OVA) and anti-OVA antibodies as described, for 30 min. Cells were fixed and stained for early endosome autoantigen 1 (EEA1, green) and F-actin (red). DIC: differential interference contrast; Merge: all three images superimposed. Arrows indicate examples of co-localization of EEA1, F-actin and microspheres. Scale bar represents 10 μm.

Download figure to PowerPoint

However, a different pattern emerged for cathepsin S, which generates peptides for MHC class I complexes in the vacuolar cross-presentation pathway.10 In Fig. 6, we observed a distinctive pattern of cathepsin S staining in control cell samples: a single, bright, concentrated area surrounded by weaker, diffuse staining throughout the cell. This was essentially similar for cells containing sonicated 0·8-μm MS and we could not discern any co-localization of cathepsin S with individual 0·8-μm MS. In contrast, there was a clear re-distribution of cathepsin S from the single bright area to co-localization with all 3·0-μm MS examined. Moreover, cathepsin S could be observed closely associated with a number of aggregated 0·8-μm MS. We also examined the localization of the 24 000 molecular weight ER SNARE (ERS24). Control cells showed a staining pattern consistent with ER and perinuclear location.30 Becker et al.21 have shown co-localization of ERS24 with 3·0-μm MS and aggregated clumps of 0·8-μm MS, but not individual 0·8-μm MS. We obtained similar results: we did not find ERS24 in association with single 0·8-μm MS, but we observed ERS24 in association with a number of 3·0-μm MS and aggregated clumps of 0·8-μm MS (Fig. 7).

image

Figure 6.  Confocal microscopy of DC2.4 after phagocytosis of microspheres (MS): cathepsin S. Cells were supplied with MS coated with ovalbumin (OVA) and anti-OVA antibodies as described, for 30 min. Cells were fixed and stained for cathepsin S (green). DIC: differential interference contrast; Merge, both images superimposed. Cells: cells incubated without MS. Arrows indicate examples of single 0·8-μm MS where no co-localization with cathepsin S could be observed. Scale bar represents 10 μm.

Download figure to PowerPoint

image

Figure 7.  Confocal microscopy of DC2.4 after phagocytosis of microspheres (MS): endoplasmic reticulum SNARE of 24 000 molecular weight (ERS24). Cells were supplied with MS coated with ovalbumin (OVA) and anti-OVA antibodies as described, for 30 min. Cells were fixed and stained for ERS24 (green). DIC: differential interference contrast; Merge, both images superimposed. 0·8-μm panel: arrows indicate examples of single 0·8-μm MS where no co-localization with ERS24 could be observed. 3·0-μm; 0·80-μm Agg panels: arrows indicate examples of co-localization of ERS24 with 3·0-μm and aggregated 0·8-μm MS. Scale bar represents 10 μm.

Download figure to PowerPoint

Activation of β-catenin correlates with the size of phagocytosed antigen

One mechanism by which antigen size might influence the processing pathway is by mechanical disruption of the cell during phagocytosis. Disruption of E-cadherin-mediated cell–cell interactions has been shown to induce DC maturation into a phenotype that induces T-cell tolerance.31 A key mediator of this pathway is the transcriptional co-factor, dephosphorylated (active) β-catenin. In resting cells, the majority of β-catenin is sequestered in a complex with E-cadherin,32 but upon disruption of E-cadherin-mediated interactions or after other signalling through the Wnt pathway, the level of active β-catenin in the cytosol increases and it migrates to the nucleus.31,32 We asked whether we could detect active β-catenin in the cytosol of DC2.4 cells isolated after phagocytosis of OVA-coated MS. We used a monoclonal antibody (clone 8E7) specific for the dephosphorylated form of β-catenin33–35 to probe an immunoblot of cytosolic extracts (Fig. 8). A major band with approximate molecular weight 50 000 and two smaller bands of approximately 28 000 and 25 000, were detected in the cytosolic fraction of DC2.4 that had phagocytosed aggregated 0·8-μm MS and 3·0-μm MS. A far weaker signal was obtained for cytosol from cells that phagocytosed sonicated 0·8-μm MS, and nothing was detected in cytosol from control cells. The expected size of β-catenin is 92 000.33 The three bands detected with the 8E7 monoclonal antibody may represent C-terminal protease degradation products formed during the 1-hr lysis procedure.

image

Figure 8.  Immunoblot of DC2.4 cytoplasm after phagocytosis of microspheres (MS). Cytoplasmic extract was prepared from control cells (lane 1), cells supplied with sonicated, 0·8-μm MS (lane 2), sonicated, 3·0-μm MS (lane 3) or unsonicated, aggregated 0·8-μm MS (lane 4). Extracts were separated by SDS–PAGE and subjected to immunoblotting and probing with anti-β-catenin and anti-GAPDH antibodies. The loading control, GAPDH, and a non-specific band at around 140 000 molecular weight were detected in all lanes; this band is described by the manufacturer. Molecular mass standards are indicated to the left of the blot (× 10–3). NSB: non-specific band, as described by the antibody manufacturer; *putative degradation products of β-catenin.

Download figure to PowerPoint

Use of homogeneous populations to estimate the yield of cross-presentation

We took advantage of the sorted DC populations with defined antigen content to deduce the relative yield from each pathway. A highly detailed analysis of the efficiency of direct presentation of Kb-SIINFEKL in fibroblasts, DC and macrophages has been carried out36 suggesting an efficiency of 0·07% for DC2.4 (one complex per 1406 OVA molecules degraded), but to date, no attempt has been made to calculate the efficiency of cross-presentation. To determine antigen input, we calculated the amount of OVA adsorbed to the surface of each MS (Table 1), then used B3Z cells and the 25-D1.16 antibody to determine the number of Kb-SIINFEKL complexes per DC, essentially as in refs 25 and 28 and as shown in Supplementary material, Fig. S2. At 4 hr after the antigen pulse, the yield of Kb-SIINFEKL complexes was 0·0024% (1 per 42 000 OVA molecules phagocytosed) and 0·0011% (1 per 91 000) for 0·8-μm and 3-μm MS, respectively (Table 2). We also estimated the yield of cells containing 100 or more MS, making a simplifying assumption that each cell contained one large particle, calculating the average size of that particle to be 266 × 0·8-μm MS. Cross-presentation yield at 4 hr was considerably less than for single particles, at 0·0004% (1 per 250 000).

Table 1.   Ovalbumin content of passively adsorbed microspheres
 3·0-μm MS0·8-μm MS
  1. Microspheres (MS) were coated with ovalbumin (OVA) as described in Experimental Procedures and Supporting Information.

Microspheres/ml1·53 × 1099·66 × 1010
OVA/microsphere (μg)4·75 × 10−79·4 × 10−9
OVA/microsphere (molecules)6·36 × 1061·26 × 105
Table 2.   Estimation of cross-presentation yield after a 4-hr chase
Antigen format0·8-μm single3·0-μm single0·8-μm aggregate
  1. OVA, ovalbumin.

Kb-SIINFEKL detection threshold, epitope number379383383
Kb-SIINFEKL per sorted cell, epitopes above detection threshold372117
OVA per particle, molecules1·26 × 1056·36 × 106(1·26 × 105) × 266
Yield, % Kb-SIINFEKL per molecule OVA0·00240·00110·0004

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Experimental procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Our results indicate that particle size plays a role in determining the pathway of antigen processing by DC. Becker et al.21 showed the importance of particle size on the recruitment of ER to the nascent macrophage phagosome, suggesting that the ER acts as a reserve when there is a demand for membrane. The mechanism by which large particles could regulate fusion of the ER with the plasma membrane, and whether this is cell-type-dependent, has yet to be determined. The functional importance of ER recruitment to the phagosome for antigen processing is keenly debated.12–16,37,38 Initial studies on ER-mediated phagocytosis suggested that the pathway is resistant to BFA inhibition, involves proteasomal processing and return of the antigenic peptide to the phagosome by TAP.12–14 We noted that the reported effects of BFA on cross-presentation were variable, and our own initial experiments have extended these observations. We also found that the cellular uptake of phagocytosed antigen was highly heterogeneous and influenced by aggregation of the MS.21

We examined different antigen delivery formats: sonicated, 0·8-μm MS that were phagocytosed as individual particles and unsonicated 0·8-μm MS that were taken up as larger aggregates. Whether each of these clumps was surrounded by one membrane instead of being formed of clusters of small phagosomes is not entirely clear; Fig. S3 (see Supplementary material) suggests that the latter may be the case, because actin appears to surround each of the MS in an aggregate. We also looked at sonicated 3·0-μm MS, because they have been used for imaging phagocytosis.13,14,21 By sorting DC into populations homogeneous for the number and size of phagocytosed antigen-loaded particles (cells containing a single 0·8-μm MS, a single 3·0-μm MS or > 100 × 0·8-μm MS), we attempted to unravel the association between particle size, antigen load and the sensitivity of cross-presentation to antigen-processing pathway inhibitors. The sorting technique also allowed for improved quantification of T-cell activation, because all DC in a sample contained comparable antigen doses.

We identified clear differences in the processing of OVA when supplied to cells on different sized particles. OVA delivered on single, 0·8-μm MS was processed in a BFA- and MG132-dependent manner, characteristics most consistent with the first described phagosome-to-cytosol pathway. Consistent with this observation, we did not find the ER SNARE ERS24 or cathepsin S by immunolocalization, although both actin and EEA1 were observed in association with 0·8-μm MS.

Cross-presentation of OVA by cells containing aggregates of 100 or more MS – irrespective of protein load – was largely resistant to BFA and MG132, features that more closely resemble the vacuolar pathway, where MHC class I is not sourced from the ER, but believed to be recycled from cell surface complexes. In line with this, cathepsin S was present in association with many of the aggregates and all of the 3·0-μm MS examined. Based on the findings of Becker et al.21 in macrophages, we might have expected cross-presentation from large aggregates of 0·8-μm MS to take place via an ER-phagosome, i.e. show resistance to BFA inhibition and sensitivity to proteasome inhibitors. Indeed, we also observed ERS24 associated with some aggregates of 0·8-μm MS and with some 3·0-μm MS. However, the observed lack of sensitivity to proteasome inhibitors and the presence of cathepsin S may indicate that ER may have been co-opted into the phagosome primarily as a source of membrane under circumstances of heavy demand, as suggested by Becker et al. Our data do not exclude the possibility of antigen processing in an ER-phagosome for OVA delivered by single, 3-μm MS: we observed, on average, modest inhibition by BFA and intermediate sensitivity to proteasomal inhibition. However, this processing phenotype might also arise from the simultaneous operation of a phagosome-to-cytosol and vacuolar pathway for antigen processing from a 3-μm MS.

In an attempt to understand how particle size could influence processing pathway, we asked whether uptake of particles might provide a stimulus similar to the mechanical disruption of E-cadherin-mediated cell–cell interactions,31 where β-catenin activation helps to shape maturation of DC into a tolerogenic phenotype. Such signals may also influence antigen processing. Importantly, Chen and Jondal39 have shown that mechanical disruption of clusters of bone marrow-derived DC activates a cathepsin S-dependent vesicular processing pathway for soluble OVA that is resistant to BFA. In our model system, in the absence of MS, no activated β-catenin was detected in the cytoplasm of DC2.4 cells, but increasing amounts were detected with increasing size of MS phagocytosed. Intriguingly, DC2.4 cells are mobile before, during and after phagocytosis, as illustrated in Supplementary material, Videos S1–S4, where cells can be seen forming and breaking cell–cell interactions, moving across the substratum and undergoing division (distributing phagocytosed particles evenly between daughter cells). Phagocytosis may provide more than one form of mechanical stimulus. Grembowicz et al.40 have demonstrated that disruption to the plasma membrane triggers migration of c-fos protein to the nucleus. We found that c-fos migrated to the nucleus during the course of antigen pulse–chase experiments, but that it occurred more slowly for cells supplied with sonicated 0·8-μm MS than those supplied with 3·0-μm MS or aggregated 0·8-μm MS (see Supplementary material, Fig. S4).

We propose that alternative antigen-processing pathways co-exist and dominate for the same antigen when delivered in different sizes, irrespective of antigen dose. Manipulation of cross-presentation systems to display pathways that may not be dominant under normal circumstances has been described previously for the vacuolar pathway. Shen et al.10 found that this pathway was only evident when biodegradable particles were supplied to TAP- bone marrow-derived DC.

Our experimental model has allowed us to make evaluations of the yield of cross-presented peptide derived from these distinctive presentation pathways for the first time. Calculations suggest that different antigen-processing pathways may yield different amounts of peptide complex per unit antigen captured, at the selected time-point. The lower yield from aggregates at 4 hr may be a consequence of the likelihood that much of the OVA is less accessible inside an aggregated particle. Indeed, if an aggregate of 266 × 0·8-μm MS is assumed to be spherical, approximately 85% of the adsorbed OVA would be inside the sphere, which could account for the six-fold reduction of yield observed. Alternatively, it may be that large numbers of particles, or large aggregates overload the cell, inducing transient nucleation of actin around the phagosome (actin ‘flashing’), which has been shown to induce a significant delay in phagosome maturation.41 Consequently, it might be predicted that cross-presentation from aggregated particles lasts longer than a similar dose of antigen delivered on multiple, single 0·8-μm MS, which may be beneficial when DC migrate to lymph nodes for priming of cytotoxic T lymphocytes. Our experiments form the basis for further studies to explore the kinetics of peptide generation, decay and antigen persistence for each size of antigen phagocytosed. Time–course experiments will enable us more directly to compare the efficiency of cross-presented OVA with the steady-state efficiency of directly presented OVA as measured by Princiotta et al.36

Cross-presentation of antigen by DC is believed to be an essential part of the priming reaction that leads to activation of cytotoxic T lymphocytes. We propose that particle size is an important factor in routing antigen processing, and that at early time-points, a pathway with phagosome-to-cytosol characteristics may yield relatively more MHC class I peptide complexes than a pathway with vacuolar characteristics, for a given antigen dose.

Recent murine studies have identified the importance of particle size in promoting distinctive cytokine responses following OVA vaccination. Particles of < 50 nm showed predominantly interferon-γ responses, whereas particles at least twice as large promoted interleukin-4 responses.42 Previous studies in human monocyte-derived DC have demonstrated an optimal size range of < 500 nm for acquisition by DC populations,43 whereas earlier mechanistic work suggested that particles of < 200 nm were predominantly taken up via receptor-mediated processes and those > 500 nm were taken up by macropinocytosis and phagocytosis.44 For human application, another key property relating to particle size is the trafficking within the body following administration. For mucosal vaccines requiring uptake from the nasal or gut-associated mucosa, formulations of 5–10 μm have been suggested for optimal induction of immunity.45 Finally, it is becoming apparent in model human systems that particle sizes that relate to pathogen size lead to qualitative immune responses ideally suited to viral or bacterial infection. Typically, particles of < 200 nm that are equivalent to virus promote robust T helper type 1 immune responses, whereas larger particles that are comparable to bacteria (i.e. 500–2000 nm) are associated with a humoral response.46

Future developments in vaccine design will require an understanding of particle size, composition and selective cross-presentation pathway targeting, to achieve maximal MHC class I peptide complex expression. This will enable improved T-cell priming and help the rational development of vaccines against viral disease and tumours.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Experimental procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

We thank J. Yewdell and A. Tutt for gifts of the 25-D1.16 and anti-OVA antibodies, respectively; Mr Richard Jewell for flow cytometry support and Dr David Johnston for assistance with confocal microscopy. This work was supported by a Wellcome Trust Clinician Scientist Fellowship to A.P.W.

Disclosure

  1. Top of page
  2. Summary
  3. Introduction
  4. Experimental procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

The authors state that there are no financial or commercial conflicts of interest.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Experimental procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Experimental procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure
  9. References
  10. Supporting Information

Figure S1. Control experiment to check the efficacy of MG132 in blocking direct presentation of cytosolic OVA. Both live and fixed cells were incubated with a hypertonic solution of OVA and subjected to a hypotonic shock.

Figure S2. Estimation of cross-presentation yield. (A) Calibration of the fluorescence intensity of an Alexa Fluor 647 F(ab′)2 fragment using QIFIKIT beads loaded with known numbers of mouse epitopes. Fluorescence was determined by flow cytometry. (B) Standard curve derived by plotting mean fluorescence intensity of the stained beads against the number of epitopes per bead. (C) Staining of SIINFEKL-MHC class I complexes on the surface of DC2.4 cells loaded with varying concentrations of exogenous SIINFEKL using the 25-D1.16 antibody. The number of complexes was calculated using the standard curve in (B). (D) Response of B3Z cells to DC2.4 cells loaded with varying concentrations of exogenous SIINFEKL. There is a degree of overlap between the upper range of the B3Z response curve and the lower range of sensitivity of the Kb-SIINFEKL-specific antibody. This overlap was used to translate B3Z response into Alexa Fluor 647 fluorescence intensity.

Figure S3. Distribution of actin around phagocytosed MS. DC2.4 cells were fed yellow-green fluorescent MS shown in green for 30 min, before fixing and staining cellular actin with a phalloidin-Alexa Fluor 647 conjugate (shown in red). Upper panels: cells were fed sonicated 3.0 μm MS. Lower panels: cells were fed aggregated 0.8 μm MS. Actin: single optical sections showing actin alone. Actin, MS: as before, but with the fluorescent MS overlaid. Actin z-stack: actin maximum projection of entire cells. Actin, MS z-stack: as before, but with the fluorescent MS overlaid. DIC: differential interference contrast image. Merge: Superimposition of fluorescence and DIC images. Arrows point to examples of phagosomes surrounded by actin. White bars represent 10 μm.

Figure S4. Translocation of c-Fos from cytoplasm to nucleus after phagocytosis of microspheres. 0.8 μm: DC2.4 cells were fed sonicated 0.8 μm MS according to Experimental Procedures. Cells were fixed and stained 1 hr (upper panels) or 4 hr (lower panels) after the start of the 30 min antigen pulse. C-Fos is shown in green and the nucleus in red. Areas of colocalisation appear yellow. DIC: differential interference contrast image. Merge: overlay of fluorescent and DIC images. 3.0 μm: cells were fed sonicated 3.0 μm MS. 0.8 μm Agg: cells were fed aggregated 0.8 μm MS. White bar represents 10 μm.

Video 1. Time-lapse animation of DC2.4 cells incubated for 24 hr in the absence of microspheres.

Videos 2–4. Time-lapse animations of DC2.4 cells incubated for 24 hr in the presence of OVA-coated 3.0 μm MS.

Data S1. Detailed Materials and methods.

FilenameFormatSizeDescription
IMM_3558_sm_FigS1.PDF21KSupporting info item
IMM_3558_sm_FigS2.PDF139KSupporting info item
IMM_3558_sm_FigS3.PDF1614KSupporting info item
IMM_3558_sm_FigS4.PDF2118KSupporting info item
IMM_3558_sm_Supplemental-Methods.doc34KSupporting info item
IMM_3558_sm_Video-3.avi5041KSupporting info item
IMM_3558_sm_Video-4.avi3106KSupporting info item
IMM_3558_sm_Video-1.avi3863KSupporting info item
IMM_3558_sm_Video-2.avi5037KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.