Redirecting soluble antigen for MHC class I cross-presentation during phagocytosis

Phagocytosis enhances cross-presentation of independently endocytosed soluble antigens

We first studied the presentation of an antigen in the endocytic/MHC class II pathway during phagocytosis. We used OT-1/OT-II T-cell activation assays. DC2.4 cells (a C57BL/6 DC line), or C57BL/6 bone marrow DCs (BMDCs) were incubated with soluble OVA, in the presence or absence of MSU. We ascertained that adding solid structures did not alter OVA uptake by DCs (Fig 1A) within the first 2 hours of MSU/OVA/DC incubation. Extended presence of MSU or latex beads for 4 hours showed a similar degree of OVA uptake (Supporting Information Fig 1A). Non-specific binding of OVA to the DCs at 4°C was minimal, although some absorption was seen after 4 hours (Supporting Information Fig 1B). Antigen loading was then stopped by washing after 4 hours, and DCs were then incubated with OT-1 T cells that specifically recognize OVA peptide SIINFEKL/H-2Kb complex. CD69 expression and IL-2 production in response to soluble OVA in the presence of MSU were increased, regardless if DC2.4 (Fig 1B) or BMDCs (Fig 1C, percentage of CD69-positive cells are shown in bar graphs) were used as APCs. To see if this event was associated with DCs only, we also tested FLT3L (FMS-related tyrosine kinase 3 Ligand)-stimulated BMDCs and cultured primary macrophages. While FLT3L DCs (Supporting Information Fig 1C) behaved similarly to GM-CSF/IL-4 BMDCs and DC2.4 cells, macrophages, on the other hand, showed high variability in results from assay to assay in response to MSU, and were therefore not pursued further (data not shown). Adding MSU 20 minutes before or after OVA incubation with washings in between did not change the outcome, ruling out nonspecific binding of OVA to MSU (Supporting Information Fig 1D). In addition, we did not detect any binding of OVA to MSU under the microscope even when they were mixed (not shown). MSU treatment marginally decreased OT-II T-cell activation, although the difference was not statistically significant in multiple experiments (Fig 1D). We used a new preparation of finer MSU crystals made by one week of mechanical grinding with stainless steel balls. The resulting crystals were smaller (around 0.5–1 um in length) in comparison with crude crystals (10 to 20 um in average length). The fine MSU crystals had a much milder DC stimulation capacity and caused little cell death. This preparation, used in the duration of our assays, did not cause any change in apoptosis (Supporting Information Fig 2A) or MHC class I, class II, CD40, CD80, and CD86 expression in 6 hours (Fig 1E). Treatment of MSU crystals and latex beads did not cause any T-cell activation in the absence of OVA (Supporting Information Fig 2B). To demonstrate that our MSU preparation was not contaminated by microbial products or triggered innate immune response that led to the observed enhanced cross-presentation, we performed the cross-presentation assay with Nlrp3^−/−, Myd88^−/−/Ticam1^−/−, as well as Nfkb1^−/− BMDCs. The enhanced cross-presentation in response to MSU crystal remained intact despite these deficiencies (Supporting Information Fig 2C–E). Cross-presentation of soluble antigens can be mediated by mannose receptor (MR) [13]. To see if solid structure-treated cells have enhanced expression of MR, they were stained with an anti MR antibody. We detected no enhance expression of this receptor (Supporting Information Fig 2F).

Cross-presentation of endosomal antigen in the presence of solid structure is efficient

Our results raised two questions: is this re-routing effect common to other crystals and how efficient is the endosomal cross-presentation compared with the phagosomal? To compare the two routes of cross-presentation, we fed DCs with either soluble or latex-bound Alexa-488 OVA. Considering these two routes of antigen uptake have different efficiencies, it was necessary to determine the amounts of antigen inside the cells. Due to extensive and different levels of proteolysis in cell lysates (not shown), western blotting-based quantitation was ruled out. We therefore selected FACS to provide a crude estimation of antigen uptake in BMDCs. Two hours into the incubation, 200 ug/ml soluble OVA was roughly equivalent to 15 ul of OVA beads as determined by similar mean fluorescence intensity (Fig 2A). Similar results were obtained from four-hour incubation (Supporting Information Fig 2G). Using this crude estimate as a starting point, we titrated both preparations for incubation with BMDCs, stopped the loading after 4 hours by washing, and mixed the treated cells with OT-1 T cells. Using CD69 and IL-2 as readouts, the same amount of OVA was cross-presented more efficiently via endosome than via phagosome (Fig 2B), i.e. undiluted latex OVA beads were cross-presented to a similar extent as the soluble antigen after a 40-fold dilution. In this assay, the particles used to deliver the antigen and for phagocytic stimulation were different (latex vs MSU), and the experiments were therefore imperfectly controlled. Some crystals such as MSU and CPPD (calcium pyrophosphate dehydrate) cannot be coated with OVA. We therefore studied latex beads both as phagocytic target carrying OVA and as solid structure that stimulated phagocytosis when OVA was given separately. Figure 2C shows that latex beads were fully capable of stimulating the soluble antigen cross-presentation. Because latex beads are an artificial stimulant, we therefore tested basic calcium phosphate (BCP), another pathogenic crystal, in addition to MSU. Figure 2D shows that BCP-stimulated cross-presentation was similar to that of fine MSU crystals and latex beads, suggesting that phagocytosis-induced soluble antigen presentation into the MHC class I pathway may be a common occurrence.

Phagocytosis reduces endocytic maturation and proteolysis

In our settings, phagocytosis appeared to alter endocytic MHC class II antigen processing. The maturation processes and proteolytic activities in both routes are similar given extensive cross talk and the shared fusion partners between the two vesicular systems [24]. It is possible that phagocytosis influences endosomal behavior. Endosomes and lysosomes can be traced by transferrin (TF) receptor and Lysotracker, respectively. Figure 3A shows that in the presence of MSU, endosome/lysosome overlapping was reduced by about 70%, suggesting a lack of progression in the endosomal maturation. The reduced co-localization was similarly detected with latex beads (Supporting Information Fig 3A). We used high resolution Electron Tomography (Fig 3B) and TEM (Fig 3C) to study the morphology and distribution of endosomes in the cytoplasm [25–27]. Interestingly, with MSU stimulation, there was an increased presence of unilamellar endocytic vesicles closely aligned with the plasma membrane. From the lack of any internal structures, these endocytic vesicles appeared to be at the beginning stage of endosomal fusion. These results suggested that in the presence of MSU there was an increase of early endosomes (or endocytic vesicles). Using TEM, we collected more representative pictures (6 images each group) from MSU-treated or -untreated DCs, and counted the numbers of these early vesicles defined as being smaller than 500 nm in diameter, within 1 um from the plasma membrane, and a void of internal structures. The comparison is shown in Figure 3C and the increase in the number of plasma membrane-proximal endosomes in the presence of MSU was apparent.

To see how phagocytosis altered intracellular proteolysis, we used standard fluorescent probes (DQ OVA) to measure protease activities. Figure 3D shows that in the presence of MSU, phagocytic (the time plot and the middle panel) and endocytic compartments (right panel) had reduced proteolysis particularly in the exponential range of the slope, which indicates the steady state proteolysis is reduced upon phagocytosis. Interestingly, the presence of MSU did not alter the acidification progression in the endocytic compartment (Supporting Information Fig 3B), suggesting that the accumulation of proteolytic enzymes through fusion is independent from the acquisition of vacuolar proton pumps (V-ATPase). The lack of strong proteolysis is regarded as a hallmark of DCs in comparison with macrophages and neutrophils [2]. Our results indicate that the “arrested” early endosomes have features that favor cross-presentation.

ROS and cathepsin S are involved in soluble antigen cross-presentation

Our data suggested that OVA can be processed in the endocytic compartment. To rule out the possibility of antigen translocation to the cytosol, we tested whether the redirected MHC class I antigen required TAP for presentation. Figure 4A shows that TAP-deficient BMDCs sensed MSU crystals and showed the enhanced cross-presentation, suggesting that cytosolic antigen processing was not essential for the soluble antigen cross-presentation. In DCs, phagocytosis activates NOX2 (NADPH oxidase), and its associated ROS inhibits endosomal peptidases [28, 29]. This step is critical in protecting antigenic MHC class I epitopes from destruction. To see if this mechanism was employed here, we generated DCs from CYBB (cytochrome B β chain) and NCF (Neutrophil cytosolic factor 1) deficient (Cybb^−/− and Ncf1^−/−) mice. CYBB (Fig 4B) and NCF (Supporting Information Fig 4A) deficient DCs showed a reduced cross-presentation. These DCs when pulsed with OVA peptide showed no defect in activating OT-1 cells. Importantly, the effect of MSU treatment on the cross-presentation was diminished. This suggests that ROS generation during phagocytosis is important in dampening proteolysis and MSU uptake actively utilizes this effect to enhanced cross-presentation.

From endosomes, there are potentially several ways whereby antigenic peptides can be loaded onto MHC class I molecules. For vacuolar antigen processing, endosomal cathepsins are critical. cathepsin L and B-deficient DCs showed no reduction in this mode of antigen presentation as both OT-1 CD69 expression and IL-2 production were intact in response to MSU (Fig 4C and D). However, cathepsin S deficiency led to reduced OT-1 IL-2 production and CD69 expression (Fig 4E). Importantly, cathepsin S-deficient cells no longer responded to particle stimulation for higher levels of cross-presentation (Fig 4E and Supporting Information Fig 4B). In some experiments and for unknown reasons, cathepsin S-deficient DCs stimulated slightly stronger CD69 expression but not IL-2 activities at the basal level, yet they still failed to show an enhanced response to MSU stimulation in both cases (Supporting Information Fig 4B). The results were consistent with the understanding that cathepsin S is the only major protease active in early endosomes (pH range 6.0–7.5) and is responsible for antigen processing [30].

Phagocytosis triggers MHC class I cross-presentation of soluble antigen in vivo

To study the in vivo relevance of our findings, we tested this phenomenon in mice. C57BL/6 CFSE-labeled OT-1 (Thy1.1) and OT-II (Ly5.1) cells with congenic marker mismatch were injected i.v. into C57BL/6 hosts (Thy1.2 or Ly5.2) and their proliferation was analyzed 6 days later. Injection of MSU s.c. triggered strong proliferation of OT-1 cells when OVA was delivered i.p. (Fig 5A), and this proliferation was absent without OVA (Supporting Information Fig 5A). OT-II activation was not altered (Fig 5A). To see how cathepsin S deficiency affected this phenomenon, we used Ctss^−/− mice to perform the same assay. Supporting Information Figure 5B shows that as in the in vivo setting, Ctss^−/− mice had a slightly higher basal OT-1 proliferation, however adding MSU did not result in a higher response, similar to the in vitro observation. We also tested whether antibody production, a result of CD4 T-cell activation, was affected by MSU. Figure 5B shows that in the presence of MSU, IgG1 antibody production was reduced. Interestingly, IgG2a response remained intact. To test this change in CD4 T cells directly, we injected OT-II cells (Ly5.1) i.v. into recipient mice (Ly5.2) and delivered OVA with or without MSU s.c.. After two weeks, isolated OT-II cells were gated via Ly5.1 and stained for intracellular IL-4 and IFN-γ production. With MSU injection, while IFN-γ expression was slightly higher, the overall number of positive cells remained very small. The IL-4 positive cells were reduced by about 80% (Fig 5C). This shift away from a dominant Th2 phenotype was in line with the notion that a balancing CD8 activation was in place.

Mice and media

C57BL/6, Tap1^−/−, OT-II and OT-1 mice were purchased from Jackson laboratories and Ctss^−/−, Ctsl^−/−, Ctsb^−/−, Cybb^−/−, Ncf1^−/−, Myd88^−/−/Ticam1^−/−, Nlrp3^−/− and Nfkb1^−/− mice were housed at University of Calgary Animal Research Centre [33, 37]. Cathepsin-deficient mice were gifts from Dr. Kenneth Rock of UMass Medical School. All mouse experiments were approved by the Animal Protocol Committee of University of Calgary. DC2.4 cells were maintained with 5% FBS media. BMDCs were grown with IL-4 and GM-CSF (Biosource and Gibco inc, Invitrogen) (51). On day 6 of DC culture, cells were gently washed and replaced with fresh cell culture media before use.

Reagents

Ovalbumin (OVA) was purchased from Sigma. Monosodium urate and basic calcium phosphate crystals, both used at 250 ug/ml for antigen presentation assays, were made from materials from Sigma as described previously [33, 37]. For MSU, the crude crystal preparation was tumbled with 5 mm stainless steel beads in a 50 ml conical tube for one week to produce finer crystals about 1um in length. Latex beads (3 um) were purchased from Polyscience Inc. Complete Freund’s adjuvant (CFA) and incomplete Freund’s adjuvant (IFA) were from Rockland. All IL-2 ELISA Ready-Set-Go kits were from eBioscience. Alexa-488 OVA, Lysotracker (deep red and green) and Alexa-594 Transferrin receptor were bought from Invitrogen. Anti mouse CD4 was from BD. Anti mouse antibodies for CD8, CD40, CD45.1, CD69, CD80, CD86, CD90.1, MHC I (H2Db), MHC II (Ia/Ie), IL-4 and IFN-γ were from eBioscience Inc. Secondary antibodies and Prolong Gold Antifade were bought from Molecular Probes (Invitrogen).

Flow cytometry

For the OVA uptake experiments, DCs were incubated for indicated durations with and without crystals (250 ug/ml) along with Alexa-488-conjugated OVA. In some assays, DCs were directly incubated with Alexa-488 conjugated OVA-coated latex beads. The cells were scrapped, washed with PBS and fixed using 2% PFA and read using an Attune® Acoustic Focusing Flow Cytometer. Apoptotic assays were done using apoptotic detection kits based on manufacturer’s instructions (eBioscience). All analyses were done using Treestar Flowjo 10.0.

Antigen presentation assays

OVA antigen (soluble or OVA coated beads) was added with or without phagocytic stimuli to BMDCs grown in 24 well plates. After 4 hours, cells were given fresh media and 2 × 10⁶ OT-1 or OT-II cells and allowed to interact. 18–24 hours later the cells were collected and stained for CD8/CD4 and CD69. The supernatant collected at the same time was used to measure IL-2 levels. A similar protocol was followed for all gene deficient BMDCs.

Electron microscopy

For TEM analyses, the cultured cells were processed in situ for fixation, dehydration, infiltration and embedding in the culture dish. The cells were prefixed with 1.6% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.3 for 1 hour and post fixed with cacodylate buffered 1% osmium tetroxide for 1 hour at room temperature. Cells were then dehydrated through graded ethanol and embedded in Epon mixture. After polymerizing the hardened Epon layer containing the embedded cells was separated from the plastic culture dish. Under a light microscope a representative area was selected, trimmed and glued to resin stub for sectioning. Ultra-thin sections were cut with a diamond knife on an ultramicrotome (Ultracut E, Reichert-Jung, Vienna, Austria) and collected on single hole grids with Formvar supporting film. The sections were stained with aqueous uranyl acetate and Reynolds’s lead citrate and observed under a Hitachi H7650 TEM at 80 kV. The images were acquired with an AMT16000 digital camera mounted on the microscope. Unilamellar vesicles lacking electron dense contents and that were under the appropriate size of early endosomes (300–500 nm diameter) were visually counted and quantified using Graphpad Prism 6.0. For SEM, dried crystals were laid on prefabricated standard aluminum stubs with carbon tape. The samples were coated with gold (Techniques Hummer II, Anatech). Samples were then inspected under a Scanning Electron Microscope (ESEM– XL30, FEI).

Electron Tomography

Thick sections (about 250 nm) were cut and stained with 2% aqueous uranyl acetate and Reynold’s lead citrate and then placed on one side of a TEM Slot Grid (1 × 2 mm slot) that was covered with a ~40 nm thin continuous formvar film (EMS, Hatfield, Pennsylvania) and left to dry for several minutes. Colloidal gold particles (10 nm diameter) were placed on the both sides of the grid to serve as fiducial markers. Finally, a thin carbon coating was applied to both sides of the grid for mechanical stabilization and to reduce electric charging in the microscope. The images were captured on a Gatan UltraScan 4000 CCD (Gatan, Pleasanton, California, USA) at 2048x2048 pixels. Dual axis TEM tomography was carried out by taking one image every degree for a range between 120 and 130 degrees with the program Serial EM. The tomographic reconstruction was done by weighted back-projection with the IMOD software package, resulting in a tomographic dataset of approximately 4 nm resolution [56, 57].

Proteolysis assays

These assays were performed as previously described [29, 58]. For phagosomal analyses, IgG- DQ OVA conjugated particles used to evaluate proteolysis were prepared as previously described [59]. To measure endocytic proteolysis, soluble DQ OVA (Molecular probes) was used as the substrate. Relative fluorescent units (RFUs) were obtained from the equation: RFU=SFRT/CF, where SFRT is substrate fluorescence in real time, while CF represents average calibration fluorescence and was represented relative to time expressed in minutes. After normalization against the control fluor, the range of the curve where the signal increased exponentially was selected. For the plot in Figure 3 D showing phagosomal proteolysis, this range corresponds to 60–80 min. Relative rates, as defined by the increase in RFU over time, were calculated for each assay. Hydrolytic capacities were evaluated by plotting the slopes (as described by the equation y=mx+c, where y=RFU, m=slope and x=time) of the linear portion of the relative substrate fluorescence. The numbers obtained from six independent wells were averaged as the hydrolytic capacity for the conditions listed. The process was repeated for all treatments and plotted relative to untreated/negative controls. The selected range of the experimental groups plotted this way was compared by one-way analysis of variance (ANOVA) with Sidak’s multiple comparisons test.

Immunostaining

Endosomal tracking: DCs were plated at low density (less than 50% confluence) on sterile cover slips and grown overnight. 1 ul/ml of Alexa 594 tagged Transferrin receptor and 1 ul/ml of Lysotracker DND 26 were added to cells in the presence and absence of crystals or latex beads. After 5 or 10 min of incubation, the cells were washed and fixed. Cells were fixed in 2% PFA in PBS for 15 minutes, permeabilized in methanol and mounted with Prolong Gold with DAPI. All imaging was done on a Delta vision microscope followed by deconvolution of the images. Image J was used for the correlation of Lysotracker and Transferrin. More than 50% pixel overlap was documented as a positive correlation.

In vivo assays

200 ug OVA with or without 1 mg MSU crystals were injected s.c. in footpads and thighs of C57BL/6 and Ctss^−/− mice. BL6 control mice were injected with CFA admixed with 200 ug OVA. After 24 hours, isolated OT-1 and OT-II T cells (using CD8 and CD4 kits from Stemcell Technologies.) labeled with CFSE (Invitrogen) were injected intravenously. Local draining lymph nodes and the spleen were collected on day 6 and analyzed for CFSE proliferation. The populations were selected based on congenic markers. The data was collected on BD LSR-II flow cytometer and analyzed with Flowjo 10.0.

Intra-cellular staining was done with a similar protocol: the mice were injected with OVA (with and without MSU crystals) s.c. in footpads and thighs. After 24 hours, isolated CD4 cells from OT2 mice were injected intravenously. The mice were boosted with the same regimen after 1 week of initial injections. At the end of two weeks, spleen and draining lymph nodes were collected and CD4 cells were isolated as described earlier. The cells were processed with permeabilization buffer and IC fixation buffer from eBioscience according to manufacturer’s instructions. The cells were then stained with CD4, Ly5.1, IL-4 and IFN-γ and read with BD LSR II.

Antibody induction was performed as previously described [33], the mice were boosted after 2 weeks with the same regimen except incomplete Freund’s adjuvant was used instead of CFA. Two weeks after final immunization, sera were collected and OVA specific antibody titers were determined using ELISA. IgG1 titers were analyzed as described [33]; IgG2a titers were measured using a kit from Alpha Diagnostics International.

Statistical analysis

All experiments were repeated at least three times with no less than three replicates per sample and three mice per group. All statistical analysis was done using Graphpad Prism 6.0. A one-way ANOVA was done to compare all data sets when there were more than 3 sets. Sidak’s multiple comparisons test was done as a posttest on selected columns. In cases where there were only 2 sets of data, student T test was used. P value legend *: <0.05, **:<0.01, ***”<0.001, and N.S: not significant.