Alpha-alumina nanoparticles induce efficient autophagy-dependent cross-presentation and potent antitumour response

Professional antigen presenting cells, such DCs, are capable of presenting exogenous antigens to cytotoxic T lymphocytes – a crucial process known as cross-presentation for the development of adaptive immunity to tumours and most infectious pathogens³. Since the magnitude of T cell expansion is regulated primarily by antigen-presentation and activation in vivo⁴, maximizing the efficiency of cross-presentation is likely the first key step for the successful development of therapeutic cancer vaccines⁵. Cross-presentation is a sequential, multi-step process that involves antigen internalization, protein degradation and loading of antigen-derived peptides into major histocompatibility complex class I molecules of antigen presenting cells^6,7. Optimal immune responses to most antigens require antigens to be administrated with an adjuvant^8,9. Currently, the microparticle precipitate of aluminum compounds (also known as Alum) is the only licensed adjuvant in the United States. Alum enhances antibody responses; however, the mechanisms are poorly understood and it has very limited ability to enhance cross-priming of cytotoxic T lymphocytes^10,11.

Here, we report that α-Al₂O₃ nanoparticles, in contrast to Alum, efficiently enhanced antigen cross-presentation and greatly improve the antitumour efficacy of tumour cell- derived autophagosomes. Soluble ovalbumin (OVA) was conjugated to aminophenol functionalized α-Al₂O₃ or other metal oxide nanoparticles via chemoselective ligation¹² between hydrazine-modified nanoparticles and OVA modified with aromatic aldehyde (Fig. 1a and Supplementary Fig.1 and 2). Transmission electron microscopy (TEM) analysis shows that the single crystalline α-Al₂O₃ nanoparticles with a clean surface (Fig. 1b) were coated with an amorphous layer after conjugation (Fig. 1c). Fluorescence microscope images show that α-Al₂O₃-OVA could be efficiently phagocytosed by DCs (Fig. 1d). Initially, the internalized α-Al₂O₃-OVA particles were near the plasma membrane and ultimately migrated to the perinuclear region of the DCs over an incubation time of 24 hrs. This is consistent with reported observation that the translocation of payload-carrying vesicles is along microtubules and towards the microtubule-organizing centre of DCs¹³.

Next, we used the antibody specific for the peptide and major histocompatibility complex class I molecule (K^b-SIINFEKL) complexes to evaluate the efficiency of cross-presentation of OVA¹⁴. The DCs loaded with α-Al₂O₃-OVA yielded a higher level of K^b-SIINFEKL complexes than the DCs loaded with OVA (Fig. 1e). We further examined the ability of DCs loaded with OVA, α-Al₂O₃ nanoparticles or α-Al₂O₃-OVA to stimulate naïve OVA-specific CD8⁺ T cells in vitro. DCs loaded with OVA or α-Al₂O₃ (60 nm or 200 nm)-OVA, but not α-Al₂O₃ nanoparticles (60 nm or 200 nm), induced a dose-dependent proliferation of OT-I T cells from naïve TCR transgenic mice, which recognize K^b-SIINFEKL complexes (Fig. 2a). DCs loaded with a mixture of OVA and α- Al₂O₃ nanoparticles were ineffective (Supplementary Fig. 3), suggesting that delivery of OVA and α-Al₂O₃ nanoparticles into the same intracellular compartment of DCs is a critical step for efficient cross-presentation. The half-maximal effective concentration (EC₅₀) for α-Al₂O₃ (60 nm)-OVA and α-Al₂O₃ (200 nm)-OVA were approximately 10 ng/mL and 1.3 ng/mL, respectively and they are approximately 500-1,000 folds more efficient than OVA (5 μg/mL).

DCs cross-presented OVA more effectively when OVA was conjugated to α-Al₂O₃ nanoparticles than to anatase TiO₂ nanoparticles or α-Fe₂O₃ nanoparticles regardless of their sizes, indicating that the chemical properties of α-Al₂O₃ nanoparticles play an important role. The DCs pulsed with α-Al₂O₃-OVA also stimulated T cells to release both IFN-γ and IL-2, indicating an efficient T-cell activation (Fig. 2b and 2c). Moreover, the efficiency of cross-presentation mediated by α-Al₂O₃ (60 nm)-OVA was significantly higher than OVA/anti-OVA immunocomplexes (IC)¹⁵ that deliver antigens via Fc receptors or DC stimulation with a TLR4 agonist (MPL) (Fig. 2d). The estimated enhancement of cross-presentation was 10-fold for IC and 100-fold for the TLR agonist. The level of similar enhancement of cross-presentation was reported for OVA/anti-DEC-205 immune conjugates and TLR2 agonist¹⁶.

Then, we examined the nanoparticles mediated cross-priming in vivo using the naïve OT-I T- cells adoptive transfer model (Fig. 2e). When mice were injected with either 200 μg of OVA or 20 μg OVA mixed with the Alum adjuvant, the resulting OT-I T-cell expansion were similar. The larger nanoparticles, α-Al₂O₃ (200 nm)-OVA, which induced a strong T cell proliferation in vitro, failed to cross-prime naïve T cells in vivo. Whereas large quantities of α-Al₂O₃ (200 nm)-OVA aggregates were deposited near the injection sites, the smaller α-Al₂O₃ (60 nm)-OVA disappeared from the site of administration and drained into lymph nodes, in agreement with a previous report¹⁷. The number of Thy1.1⁺ OT-I CD8⁺ T cells in the lymph nodes and spleens of mice injected with α-Al₂O₃ (60 nm)-OVA was substantially higher than that induced by OVA/Alum. In addition, consistent with the in vitro test, both TiO₂ (100 nm or 25 nm)-OVA and α-Fe₂O₃ (25 nm)-OVA did not efficiently induce T cell proliferation either in the lymph node or in the spleen. These results show that α-Al₂O₃ nanoparticles with a diameter of around 60 nm were effective antigen carriers and have different functional properties when compared to traditional Alum adjuvant.

Multiple mechanisms have been proposed to explain the adjuvant activity of Alum¹⁸. Interestingly, it was shown that alum delivers soluble antigens to DCs without being phagocytosed¹⁹. To understand howα-Al₂O₃ nanoparticles induced efficient antigen cross-presentation, we used a confocal microscope and TEM to determine the subcellular localization of internalized α-Al₂O₃-OVA. When DCs were loaded with α-Al₂O₃-OVA for 6 hrs, we found that the majority of nanoparticles co-localized with the autophagosome marker, Atg8/LC3²⁰ (Fig. 3a), but not with the lysosome tracker (Supplementary Fig. 4). Such co-localization was verified with transfection of DCs with a plasmid DNA encoding the LC3-tdTomato fusion protein before loading with the conjugates. Interestingly, α-Al₂O₃ nanoparticle conjugates also co-localized with p62 (sequestosome-1), a cargo recognition and targeting molecule for the autophagy pathway²¹. The important role of p62 for selective autophagy of aggregated proteins, damaged organelles, or intracellular bacterial has been elucidated recently, and our results suggested p62 played a similar role in autophagy of internalized α-Al₂O₃ nanoparticles. By TEM analysis, we also found autophagosomes containing α-Al₂O₃ nanoparticles (Fig.3b and supplementary Fig.5).

As one of the major cellular pathways mediating degradation of proteins and organelles, autophagy is responsible for the major histocompatibility complex class II restricted presentation of endogenous antigens²². Recent reports, including ours, implicate autophagosomes of antigen donor cells for efficient cross-presentation^23,24, and that autophagy of antigen presentation cells also regulates the cross-presentation of antigens derived from herpes simplex virus type 1 and Bacille Calmette Guerin^25,26. To test whether autophagy affects cross-presentation of α-Al₂O₃-OVA, we inhibit autophagy by treating DCs with a phosphoinositide 3-kinase inhibitor: 3-methyladenine (3-MA) or wortmannin. Neither of these chemical inhibitors reduced the cross-presentation of OVA, but both inhibitors nearly abolished the cross-presentation of α-Al₂O₃-OVA (Fig. 3c). Furthermore, knockdown of the autophagy initiation gene, Atg6/Beclin 1, in DCs blocked the cross-presentation of α-Al₂O₃-OVA without affecting the cross-presentation of OVA (Fig. 3d and 3e). These data suggest that the functional autophagy pathway is required for the efficient cross-presentation of α-Al₂O₃-OVA but not OVA.

We noticed that compared to knockdown of Beclin 1, Atg 12 silencing was less effective at suppressing cross-presentation of α-Al₂O₃-OVA (Fig. 3e) and α-Al₂O₃-OVA did not significantly increased the production of LC3-II over the basal level of LC3-II in untreated DCs (Supplementary Fig. 6a). We hypothesized that α-Al₂O₃-OVA might utilize an Atg12-independent autophagy pathway for the efficient cross-presentation; such a non-canonical autophagy pathway has been reported recently²⁷. Consistent with this hypothesis, we found that cross-presentation of α-Al₂O₃-OVA was blocked by brefeldin A, an inhibitor of ER-Golgi function that was used to distinguish the non-canonical from the canonical autophagy pathway (Fig. 3f, Supplementary Fig. 6b). Cross-presentation of exogenous antigens can be enhanced by blocking antigen degradation by lysosomes. Treating DCs with ammonium chloride resulted in more efficient cross-presentation of OVA, without any effect on the cross-presentation of α- Al₂O₃-OVA (Fig. 3c). These findings reveal that the α-Al₂O₃-mediated non-canonical autophagy diverted a significant amount of antigens into autophagosomes, thus delaying acidification and degradation (Supplementary Fig. 7)²⁸.

Next, we examined whether α-Al₂O₃-OVA could elicit an endogenous T cell response capable of eliminating established tumours. Mice were injected with B16F10-OVA tumour cells and on day 7, tumour-bearing mice were injected subcutaneously with α-Al₂O₃-OVA, α-Al₂O₃ nanoparticle alone, or soluble OVA mixed with Alum. We used intracellular IFN-γ staining to enumerate the frequency of the OVA-specific CD8⁺ T cells in spleens of naïve or tumour-bearing mice 7 days post vaccination. We found a high level of OVA-specific T cells in mice vaccinated with α-Al₂O₃-OVA (Fig. 4a). Remarkably, the mice injected with α-Al₂O₃-OVA completely rejected tumours and remained tumour free for more than 40 days, while all other groups succumbed to tumour burden (Fig. 4b).

To further demonstrate the adjuvant activity of α-Al₂O₃ nanoparticles to boost the T cell response to vaccines containing a limited amount of antigens, we examined the ability of α-Al₂O₃ nanoparticles (60 nm) to increase the cross-presentation of tumour-associated antigens enriched in autophagosomes. Although whole tumour cells are a good source of antigens, we recently showed that tumour cell-derived autophagosomes (Fig. 4c and supplementary Fig. 8 and 9) are more efficient antigen carriers and capable to inducer a broader antitumour responses than whole tumour cells ²¹.

The tumour cells were engineered to express short-lived intracellular OVA so that successful cross-presentation could be assessed with OT-I CD8⁺ T-cell proliferation assay. Compared to the DCs pulsed with naked autophagosomes (Fig 4e, top panel), the DCs pulsed with α-Al₂O₃ nanoparticles (60 nm) that are conjugated with autophagosomes (Fig. 4d) were more efficient for cross-priming OT-I CD8⁺ T cells (Fig. 4e, bottom panel). Next, we evaluated the therapeutic efficacy of autophagosomes (derived from 3LL tumour cells without expressing intracellular OVA) and α-Al₂O₃ (60 nm)-autophagosome conjugates in C57BL/6 mice bearing experimental metastases 3LL lung tumours (Fig. 4f). The subcutaneous injection of α-Al₂O₃-autophagosomes but not naked autophagosomes significantly suppressed the formation of lung metastases as compared to PBS treated control mice. To improve the therapeutic efficacy of α-Al₂O₃-autophagosomes, we co-administrated the vaccine with anti-OX 40 antibody to promote the proliferation and survival of antigen-specific T cells³⁰. This combined treatment led to zero metastases in 3 out of 5 mice; no effect was observed in mice treated with OX40 antibody alone.

Our findings indicate that α-Al₂O₃ nanoparticles are novel and efficient carriers for delivery of antigens to the autophagosome-related cross-presentation pathway in antigen presenting cells. They are also capable of boosting the antitumour efficacy of tumour-derived autophagosomes containing unknown tumour-specific antigens in a limited amount. The demonstrated prototype of α-Al₂O₃-based cancer therapeutic vaccine may lead to re-engineering aluminum adjuvant for highly effective therapeutic vaccines for cancers and chronic infections.