Multi-antigen spherical nucleic acid cancer vaccines

Abstract

Cancer vaccines must activate multiple immune cell types to be effective against aggressive tumours. Here we report the impact of the structural presentation of two antigenic peptides on immune responses at the transcriptomic, cellular and organismal levels. We used spherical nucleic acid (SNA) nanoparticles to investigate how the spatial distribution and placement of two antigen classes affect antigen processing, cytokine production and the induction of memory. Compared with single-antigen SNAs, a single dual-antigen SNA elicited a 30% increase in antigen-specific T cell activation and a two-fold increase in T cell proliferation. Antigen placement within dual-antigen SNAs altered the gene expression of T cells and tumour growth. Specifically, dual-antigen SNAs encapsulating antigens targeting helper T cells and with externally conjugated antigens targeting cytotoxic T cells elevated antitumour genetic pathways, stalling lymphoma tumours in mice. Additionally, when combined with the checkpoint inhibitor anti-programmed-cell-death protein-1 in a mouse model of melanoma, a specific antigen arrangement within dual-antigen SNAs suppressed tumour growth and increased the levels of circulating memory T cells. The structural design of multi-antigen vaccines substantially impacts their efficacy.

Main

Vaccination is an attractive strategy against cancers expressing targetable tumour-associated antigens and neoantigens. For melanoma, increasing efforts have been made to develop vaccines targeting identified tumour-associated proteins (for example, gp100, MAGE-A3, MART-1 and NY-ESO-1)^1,2,3,4. However, although these vaccines elicit some benefits (such as increasing activated melanoma-specific T cells), many are designed to primarily activate cytotoxic T cells. Tumours can have considerable heterogeneity and high mutational burdens^5,6 that allow for easy escape of immune surveillance⁷. Thus, vaccines that primarily rely on cytotoxic T cell activity are inadequate, necessitating vaccines containing antigens targeting multiple immune cell types to induce enhanced tumour remission.

Common approaches to elicit a multi-faceted immune response include the administration of ‘long peptides’ whose sequence covers multiple epitopes to activate both cytotoxic and helper T cells, or of multiple ‘minimal’ peptide antigens, each unique to T cell subclasses^8,9,10,11. However, many of these ongoing efforts involve pools of peptides, with or without an adjuvant, delivered in saline as a mixture. Recently, simple changes to the delivery of vaccine components using basic chemical linkages¹² or nanotechnology^13,14,15,16 have shown the potential of structuring vaccines to improve potency. This concept, termed ‘rational vaccinology’, offers a path to structurally optimize the placement of antigens targeting multiple immune cell types within a vaccine for broad antitumour immunity.

In this work, we explore the vaccine-design space involving multiple cell-targeting antigens. By employing structural changes in antigen placement, we elucidate the impact of the resulting immune response and harness it to drive success in translational efforts. Antigens use activated cytotoxic (CD8⁺) T cells to effectively kill tumours, as well as helper (CD4⁺) T cells to synergize immune interactions for long-lasting tumour rejection^17,18. CD4⁺ T cells maintain tumour-directed CD8⁺ functionality by recruiting them to the tumour site and enhancing their proliferation and effector functions^{19,20,21,22,23}. Therefore, the vaccines in this work consider the precise structural placement of both major histocompatibility complex (MHC)-I-restricted and MHC-II-restricted antigen targets (CD8⁺-activating and CD4⁺-activating, respectively) to prime the immune system most effectively.

Here we use the spherical nucleic acid (SNA) platform to elucidate the effect of nanoscale structure on multi-antigen immunological processes. The SNA comprises a nanoparticle core (such as a liposome) with a dense and radially arranged surface of oligonucleotides. SNAs are powerful tools to explore these complex relationships because of their biocompatibility²⁴, the ability to rapidly enter cells in high quantities^25,26, their potent immune activation when employing toll-like receptor 9 (TLR9) agonist DNA as the shell²⁷, and their modularity, which enables the defined nanoscale placement of components using well-known chemistry^28,29,30. In this work, we show how the structure of SNA vaccines carrying multiple immune-cell-targeting peptide antigens greatly influences immune activation. Changing the position of the antigen type within the SNA alters dendritic-cell processing, upregulates immune-cell pathways at the transcriptome level, enhances the production and secretion of cytokines and memory markers at the cellular level, and slows tumour growth at the organismal level. Collectively, these changes define vaccine potency against an aggressive B16-F10 melanoma tumour model and, importantly, elucidate design insights regarding the placement of multiple peptide antigens that can translate to other therapeutics and guide their development.

Results and Discussion

We sought to determine the optimal antigen-processing conditions for multi-antigen SNA vaccines to generate robust cytotoxic and helper T cell responses. In particular, we investigated how the delivery of peptides for two antigen classes (MHC-I-restricted and MHC-II-restricted) to dendritic cells (DCs) would change processing in vitro. DCs are critical professional antigen-presenting cells that induce signalling for effective T-cell priming. Previous work has shown the potential to enhance DC activation through the simultaneous delivery of both cytotoxic and helper antigens^31,32, but none have had a system capable of understanding the best way to present such antigens. We hypothesized that the simultaneous delivery of both antigen classes on the same nanoparticle, as opposed to their delivery on different nanoparticles, enhances the activation of both T-cell types, and that the structural location of the antigens markedly impacts vaccine performance.

Processing of multiple antigens in vitro based on antigen distribution on SNAs

To test this hypothesis, we designed and synthesized dual-antigen SNA vaccines (DA-SNAs) that contained both MHC-I-restricted and MHC-II-restricted antigens in different nanoscale locations (termed DA-SNA 1 and DA-SNA 2, based on the placement of each antigen; Fig. 1a). Owing to the modularity of SNAs, there are multiple different locations within the SNA construct where antigens can be placed. For this work, encapsulation and hybridization arrangements for antigen placement were selected and compared to one another. To assess how the distribution of antigens and delivery on different nanoparticles affect immune activation, formulations containing two individual SNAs, each presenting only one antigen class in the same position as in the DA-SNA vaccine, were synthesized (Supplementary Fig. 1). The formulations were termed ‘separate’ for the individual SNAs delivering a single antigen and ‘combined’ for the dual-antigen containing DA-SNA.

Fig. 1: The delivery of two classes of antigen from spherical nucleic acid (SNA) vaccines alters how the antigens are processed in vitro.

a, Dual-antigen SNA (DA-SNA) vaccines synthesized to alter the placement of MHC-I-restricted and MHC-II-restricted antigens within the same nanoparticle structure. b, The expression (measured through MFI) of co-stimulatory markers CD80 (left) and CD86 (right) on CD11c⁺ DCs based on delivery of the two antigen classes on either separate nanoparticles (dashed, separate) or a singular DA-SNA (solid, combined). c, CD8⁺ T cells specific for the OVA1 antigen (left) or CD4⁺ T cells specific for the OVA2 antigen (right) raised from a co-culture of treatment-pulsed DCs with naïve splenic T cells. d, MFI of CD69 activation marker signal within the population of antigen-specific CD8⁺ (left) or CD4⁺ (right) T cells. e, Fold change in T cell proliferation from OT1 splenocytes specific for the OVA1 antigen after co-culture with treatment-pulsed DCs. For all panels, mean ± s.e.m. shown, along with statistical significance between relevant comparisons. Significance was calculated using a one-way ANOVA with Sidak’s multiple comparisons test, with n = 3–4 replicates per group. NS, non-significant.

To synthesize DA-SNAs, a peptide from one antigen class was encapsulated into 50 nm 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) liposomes during their formation (Supplementary Figs. 2 and 3). In parallel, the peptide of the other antigen class was conjugated to a strand complementary to the CpG motif adjuvant DNA shell (‘CpG complement’) of the SNA using disulfide bond formation (Supplementary Fig. 4). DNA and peptide sequences used in this work can be found in Supplementary Tables 1 and 2. A hybridized duplex was formed by slow-cooling the CpG complement with an appended antigen to a complementary 3’-cholesterol-terminated CpG strand. The cholesterol moiety anchors the duplex to the surface of the liposome. These two strands have been previously shown to effectively duplex at temperatures below ~56 °C^33,34. This hybridized product was added to the liposomes to obtain an equimolar amount of each antigen. The liposome surface was backfilled with non-targeting T₂₀ DNA that did not contain antigen to obtain 75 total DNA strands per liposome, equivalent to a density of 1.6 pmol cm⁻² at which properties associated with SNAs that make them useful in biology are observed¹⁶. SNAs containing either a single encapsulated antigen or a single hybridized antigen (the ‘separate’ formulations) were synthesized following previous protocols¹⁶. SNA formation was confirmed using dynamic light scattering (Supplementary Fig. 5), and stability of the nanostructures was maintained over 45 d (Supplementary Fig. 6).

Once these different structures were synthesized, their ability to prime DCs was assessed by using MHC-I-restricted and MHC-II-restricted model antigens of ovalbumin (OVA), known as OVA1 and OVA2, respectively. The activation of DC cues and the ability to prime T cells were characterized using murine bone marrow-derived DCs (BMDCs) and antigens delivered as two separate SNA structures versus the combined DA-SNA vaccine. An overnight incubation of BMDCs with the different vaccine structural arrangements showed that the percentage of CD11c⁺ DCs expressing innate co-stimulation markers CD86 and CD80 did not significantly change (Supplementary Fig. 7). This is probably attributed to the fact that an extended incubation can lead to equivalent levels of adjuvant delivery between the combined and separate formulations and indicates that hybridization to CpG does not interfere with TLR9 activation. To maximize differences and evaluate the impact of early kinetics on immunostimulatory activity and the expression of the co-stimulatory markers, we pulsed DCs with the various structural formulations for 30 min and replaced the media before allowing the DCs time to express co-stimulatory signals. The results (Fig. 1b) illustrate that the arrangement of antigens with OVA1 hybridized and OVA2 encapsulated in either separate nanoparticles or one single DA-SNA construct leads to significant expression of CD80 and CD86. Upon analysis of a range of doses, elevated CD80 and CD86 expression occurs when higher concentrations are used, and it is not due to a preferred interaction of MHC-I with peptide that could lead to enhanced internalization. This trend is also exhibited, albeit to a lesser extent, when using SNAs containing a different set of peptides (Extended Data Fig. 1). The presentation of antigens to naïve splenic T cells to raise antigen-specific T cells, an indicator of the immune system’s ability to recognize tumour antigens³⁵, was not impacted by the way antigens were arranged; naïve T cells were able to clonally expand into OVA1- and OVA2-specific T cells (Fig. 1c; gating strategy in Supplementary Fig. 8). Indeed, ~0.6–0.7% of the live CD19⁻ population was double-positive for the OVA1-H-2k^b dimer and CD8⁺ marker. Similar trends were observed for OVA2-specific T cell differentiation, with ~0.9–1.1% containing double-positive markers for the OVA2-H-2-Ia^d tetramer and CD4⁺ antibody. Only the DA-SNA structures, and not the separate formulations, were capable of significantly elevating the amount of antigen-specific T cells above that of the T cell control baseline; this suggests that the combined delivery of both antigens to DCs leads to stronger T cell differentiation. This is more apparent when assessing expression of the early activation marker CD69 within either antigen-specific T cell populations. An increased amount of CD69 signal (as measured by median fluorescence intensity, MFI) is present when the antigens are delivered combined as one DA-SNA, with the DA-SNA 2 structure outperforming all tested groups (Fig. 1d). Moreover, delivery of antigens in DA-SNA structures regardless of the antigen placement translates to a 2-fold increase in T cell proliferation, an important step in antitumour responses, using OT1 splenocytes specific for the OVA1 antigen (Fig. 1e).

In vivo activation via DA-SNAs

On the basis of these promising in vitro findings using DA-SNAs, we evaluated how the arrangement of the antigens delivered affected the in vivo immune responses. C57BL/6 mice were immunized to delineate how the formulation of antigen on separate or the same nanostructures and how different placement of MHC-I-restricted and MHC-II-restricted antigens within DA-SNA vaccines affect immune activation. Mice were given three total injections (6 nmol by DNA and each peptide; Fig. 2a and Extended Data Fig. 2). On day 35, splenocytes were collected to assess raised specific immune responses towards both peptide antigens. After the 5-week period, CD8⁺ levels were significantly elevated for DA-SNA 2 immunization, reaching ~35% of the spleen population compared with when a simple mixture containing both peptide antigens and adjuvant DNA (termed ‘admix’), DA-SNA 1, and the separate equivalents of both DA-SNAs were used (Fig. 2b). CD4⁺ levels were only significantly changed when mice were treated with the separate equivalent of DA-SNA 1. Other treatment groups decreased the level of CD4⁺ splenic T cells, with the DA-SNA 2 group at the lowest with ~11.4% of the spleen cell population (Fig. 2b). DA-SNA 2 most significantly elevated the production of a key pro-inflammatory cytokine, IFN-γ, as well as degranulation marker, CD107a, upon restimulation with OVA1 peptide ex vivo. DA-SNA 2 immunization also generated the largest percentage of polyfunctional splenic CD8⁺ T cells (~15%, Fig. 2c, gating strategy can be found in Supplementary Fig. 9). Moreover, this correlated with an increase in the percentage of effector memory CD8⁺ T cells (CD44⁺CD62L⁻, ~54%) (Fig. 2d, left). The levels of pro-inflammatory markers produced in CD8⁺ T cells and the polyfunctionality of the population were significantly elevated for the combined DA-SNA 2 structure compared with its separate counterpart (Fig. 2c). Ex vivo stimulation of CD4⁺ T cells with OVA2 peptide showed an overall increase in these same parameters for both DA-SNAs and the separate formulations compared with admix treatment, further demonstrating the importance of the combined delivery of antigen and adjuvant to an immune cell. There was no significant difference observed between the DA-SNA 1 structure and its separate equivalent. These two constructs induced the highest levels of these pro-inflammatory markers in CD4⁺ T cells, although a statistically significant decrease was not observed between the two DA-SNAs (Fig. 2c). Moreover, the greatest elevation in OVA1-specific CD8⁺ T cells was observed for DA-SNA 2, ~3-fold higher than its separate equivalent (Fig. 2e). Ultimately, the contribution of the combined delivery of both antigen types in the DA-SNA structures led to greater IFN-γ secretion as measured through an enzyme-linked immune absorbent spot (ELISpot) assay. Levels of spot-forming cells (SFCs) upon ex vivo stimulation with the MHC-I restricted OVA1 antigen were highest with DA-SNA 2, and DA-SNA 2 generated 2.3-fold higher SFCs than its separate counterpart upon either MHC-I or MHC-II antigen (OVA2) ex vivo stimulation (Fig. 2f). An elevation in SFCs was observed for both DA-SNA structures compared with the admix; the largest enhancement overall was seen for DA-SNA 2. Splenocytes raised by DA-SNA 2 immunization led to ~2.2-fold and ~1.7-fold more SFCs than DA-SNA 1 immunization when stimulated ex vivo with either OVA1 or OVA2, respectively, demonstrating the potency using this arrangement of antigens on a DA-SNA to respond ex vivo to MHC-I-restricted or MHC-II-restricted antigen cues. The administration of OVA1 encapsulated and OVA2 hybridized antigens displayed differences between the separate and combined formulations, with the combined DA-SNA 1 structure elevating SFCs to a greater extent upon OVA1 ex vivo stimulation, and the separate formulation elevating SFCs to a greater extent upon OVA2 ex vivo stimulation. Taking this entire study holistically and given the role of CD8⁺ T cells in antitumour activity and the importance of activating both subsets of T cell responses effectively, we harnessed the combined delivery of antigen in the DA-SNA structures in further experimentation. Overall, when evaluating the differences in immune responses between the DA-SNAs, the results highlight that positioning the MHC-I restricted antigen in the hybridized architecture optimizes DC presentation for CD8⁺ T cell responses, while encapsulating the MHC-II-restricted antigen within the core induces modest enhancements in CD4⁺ activity while preserving cytotoxic function.

Fig. 2: Antigen placement within SNAs impacts immune responses after immunization.

a, Schedule of fortnightly immunization for C57BL/6 mice. Various treatment groups in study. Dose: 6 nmol per antigen; 6 nmol adjuvant. b, Change in CD8⁺ (left) or CD4⁺ (right) T cell populations in the spleen after vaccination scheme. c, Intracellular production of IFN-γ pro-inflammatory cytokine (left) or CD107a degranulation marker (middle) was assessed upon ex vivo restimulation with peptide antigen. Polyfunctional T cells (double-positive for both markers) were quantified (right) for CD8⁺ (top) or CD4⁺ (bottom) T cells. DA-SNA 2 significantly elevated production of all markers in CD8⁺ T cells, whereas differences were more subtle among production in CD4⁺ T cells, with both DA-SNAs observably elevating levels above those in mice immunized with an admix vaccine. d, Effector memory phenotype, measured through CD44⁺CD62L⁻ markers, was increased by DA-SNA 2 for CD8⁺ T cells (left). CD4⁺ effector function (right) was most elevated for Separate 1 Encap. 2 Hyb. immunization. e, Percentage of CD8⁺ T cells that are OVA1-specific, measured through staining with an antigen-specific dimer. f, Representative counts and images (left) of IFN-γ-secreting splenic T cells upon different ex vivo stimulations along with total SFCs measured by ELISpot assay (right). Mean ± s.e.m. shown. n = 3–6 mice per group. Statistical significance between relevant comparisons is shown. For all panels, significance was calculated using a one-way ANOVA with Tukey’s (b,c,d,f) or Dunnett’s (e) multiple comparisons test.

Mechanistic understanding of DA-SNA-induced immune activation and propagation

To evaluate the possible factors driving these differences in in vivo activation, we evaluated the administration pathway for the structures upon immunization. We administered DA-SNAs subcutaneously with fluorophore labels for both peptide antigens and assessed biodistribution after 24 h (Supplementary Fig. 10). No significant differences in organ accumulation between the two DA-SNAs were observed for the hybridized antigen. The encapsulated antigen was primarily concentrated in the lymph node with very little detected in other major organs, and importantly, a significant increase in accumulation was observed as a result of DA-SNA 2 immunization. We sought to assess whether the lymph node trafficking could be explained by a difference in the release rate of the antigens from the DA-SNA structures (Supplementary Fig. 11). Release profiles performed in physiologically relevant solutions (10% FBS) ex cellulo showed relatively low levels of release for the hybridized antigen within 48 h, whereas the encapsulated antigen reached over 63% release for DA-SNA 1 and just under 50% release for DA-SNA 2. However, we have observed SNA uptake in as little as 30 min²⁵. When looking at the release profiles of the DA-SNAs within the first few hours of exposure to the serum, the structures exhibit the same release rates, which are also <25% of the total encapsulated antigen. The stability of the liposome shell, and hence the SNA, has been previously found to lead to differences in distribution^29,36. We postulate that the liposome stability, influenced by the peptide cargo^37,38, leads to the observed differences in DA-SNA encapsulated peptide distribution. Nonetheless, we emphasize that this parameter should be a design consideration for selecting peptide cargo, which varies on the basis of the antigen set, and this work illustrates important structural-based vaccine enhancements through multiple different antigen sets. We therefore conclude that the significant immune-wide changes observed in vivo cannot be attributed solely to differences in biodistribution or antigen release rates between the two nanoparticles. Moreover, when evaluating the trafficking of the antigens into splenic DCs (Supplementary Fig. 12), no differences can be resolved between the two DA-SNAs, and importantly, both generate a significant double-positive antigen population of DCs compared with naïve mice and mice administered with a simple mixture of the components. As a result of the minimal variation between the DA-SNAs based on distribution and organ-level trafficking, we performed dendritic-cell transcriptome analysis to evaluate how treatment with the differently structured vaccines impacts pathway enrichment and overall gene expression (Supplementary Fig. 13 and Dataset 1). The DA-SNA structures affect dendritic-cell pathways very differently compared with admix treatment, where few significant pathways were enriched. Pathway enrichment analysis between DA-SNA 1 and 2 suggests that DA-SNA 2 affects the internalization and pathway processing of components to a greater extent than DA-SNA 1.

We sought to resolve the processing pathways of DA-SNA 1 and 2 within dendritic cells in an effort to elucidate this impact on signalling kinetics. Using confocal microscopy (Fig. 3), we assessed the intracellular fate of DA-SNAs following uptake by BMDCs at pre-determined timepoints and compared co-localization of both the hybridized and encapsulated antigens with organelle markers for early endosome (early endosome antigen 1, EEA1), late endosome (Rab7), lysosome (lysosomal associated membrane protein 1, Lamp1), endoplasmic reticulum (protein disulfide isomerase, PDI), MHC-I and MHC-II. The data highlight that substantial processing occurs at earlier timepoints (that is, <1 h). Processing within the early endosome demonstrates comparable kinetics between the two structures (Fig. 3a). Significant differences between the two structures arise in the late endosome (Fig. 3b), where there is enhanced processing and trafficking for both the hybridized and encapsulated antigens for the DA-SNA 2 structure compared with both antigens for DA-SNA 1. Evaluation of the lysosome revealed key differences; there is decreased co-localization for the hybridized antigen of DA-SNA 2 (MHC-I restricted) and increased co-localization for the encapsulated antigen for DA-SNA 2 (MHC-II-restricted) (Fig. 3c), which suggests optimization of MHC-II loading for DA-SNA 2. We observed that the hybridized antigen of DA-SNA 1 (MHC-II-restricted) does exhibit increased co-localization with the lysosome at early timepoints but this quickly subsides, and no significant increases are observed later in the study. This also suggests that any DA-SNA 1 optimized processing routes are more transient, as increased co-localization of the DA-SNA 1 hybridized antigen (MHC-II-restricted) with MHC-II is also not observed. Analysis of processing at the endoplasmic reticulum (ER) showed no major differences between DA-SNA treatment for the hybridized antigens. This could be attributed to faster cross-presentation kinetics with the DA-SNA 2 structural arrangement for the hybridized antigen and thus less effective capturing of this process (Fig. 3d). Conversely, reduced co-localization of the ER and DA-SNA 2 encapsulated antigen (MHC-II-restricted), significant at the 1 h timepoint, could be attributable to decreased trafficking of the encapsulated MHC-II-restricted antigen for DA-SNA 2 to the ER. Importantly, the hybridized antigen of DA-SNA 2 (MHC-I restricted) exhibited increased co-localization with MHC-I at early timepoints (as early as 30 min), with significant increases observed up to 1 h (Fig. 3e), translating to an increased DA-SNA 2-driven OVA1 surface presentation on MHC-I (Supplementary Fig. 14). While not significant, there is a trending increase in the DA-SNA 2 encapsulated antigen (MHC-II-restricted) with MHC-II at later timepoints (Fig. 3f). As there is significant DA-SNA 2 encapsulated antigen (MHC-II-restricted) co-localization with the late endosome and lysosome, we suggest that the loading of MHC-II by DA-SNA incubation is a slower process than MHC-I loading. Moreover, differences between the DA-SNA treatments generally subsided by 6 h, suggesting that the immunological differences observed are driven primarily by early kinetics and processing of these structures by DCs.

Fig. 3: Time-dependent antigen processing differences driven by vaccine structure.

Representative confocal microscope images (top) of BMDCs incubated for 30 min with either DA-SNA 1 (left) or DA-SNA 2 (right) containing both Cy5-labelled hybridized antigen (red) and FITC-labelled encapsulated antigen (green). Nucleus was stained by DAPI (blue) and the following organelles were stained: a, EEA1 (early endosome); b, Rab7 (late endosome); c, Lamp1 (lysosome); d, PDI (endoplasmic reticulum); e, MHC-I; f, MHC-II. All organelles shown in yellow. Mander’s overlap coefficients (bottom) representing the fraction of Cy5 or FITC signal co-localized with respective organelles at 0.5, 1, 6 and 24 h are shown. Scale bar, 10 μm. g,i, Co-localization and h,j, flow cytometry analysis of DA-SNA 1 and DA-SNA 2 processing in BMDCs after a 1 h pulse and following 24 h treatment with chloroquine (g,h) and Brefeldin A (i,j). OVA1 co-localization with MHC-I (g) and ER (i) is shown. k,l, Co-localization of OVA2 with lysosome in the presence of leupeptin (k) and MHC-II in the presence of chloroquine (l). For a–f, the data show mean ± s.d. from 6–10 randomly selected fields of view. Statistical significance was calculated using a two-way ANOVA with Sidak’s multiple comparisons test. For g–l, data show mean ± s.e.m. for n = 3–6 replicates. Statistical significance was calculated using an unpaired t-test with Welch’s correction (g,i,k,l) and a one-way ANOVA with Tukey’s multiple comparisons test (h,j).

To further elucidate the pathways involved through DA-SNA 1 and 2 antigen presentation, we employed inhibitors to mechanistically assess antigen processing. The addition of a disrupter of endosomal acidification (chloroquine)³⁹ which has been shown to increase the extent of cross-presentation^39,40 impacts the co-localization of OVA1 with MHC-I complexes. Through this enhanced cross-presentation, we observe that DA-SNA 2 induces increased OVA1 co-localization with MHC-1 (Fig. 3g). These data suggest that DA-SNA 2 can take advantage of enhanced cross-presentation more effectively than DA-SNA 1. This architectural advantage ultimately translates to a greater surface presentation of OVA1 on MHC-I (Fig. 3h). The addition of Brefeldin A, which inhibits transport of assembled peptide-MHC-I complexes from the ER to the cell membrane⁴¹, and thus impedes cross-presentation, more negatively impacts DA-SNA 2. There is a significant decrease (4.7-fold) in co-localization of OVA1 peptide with the ER upon Brefeldin A treatment for DA-SNA 2 (Fig. 3i), translating to a complete loss of OVA1 surface presentation on MHC-I (Fig. 3j). The impact of pathway inhibitors on OVA2 processing was also analysed. The addition of leupeptin, an inhibitor of cysteine and serine proteases (for example, cathepsins B, S and L)⁴², leads to a drop in OVA2 co-localization with the lysosome (the site of MHC-II loading) for DA-SNA 2 (Fig. 3k). The antigen presentation through DA-SNA 2 is therefore more negatively impacted in MHC-II processing upon inhibitor addition. Moreover, the use of chloroquine, known to block MHC-II-dependent antigen processing^43,44,45, hinders the ability of OVA2 in DA-SNA 2 to co-localize with MHC-II (Fig. 3l). DA-SNA 2 is therefore more strongly impacted by the addition of inhibitors that disrupt MHC-I and -II processing, thus indicating that the presentation and kinetic profiles that DA-SNA 2 provides are more optimally suited for antigen processing.

To understand how the different DA-SNAs and admix treatments were inducing such varied downstream T cell responses, we collected and isolated splenic CD8⁺ and CD4⁺ T cells after immunization following the same schedule in Fig. 2 and performed bulk RNA sequencing (RNAseq). Principal component analysis (PCA) revealed holistically that the CD8⁺ and CD4⁺ T cell gene-expression profiles of mice immunized with admix formulations were most similar to those from naïve mice, suggesting that this is the cause of low overall activation (Fig. 4a). Mice immunized with DA-SNA 2 were most distinct from naïve mice in their CD8⁺ transcriptome, whereas both DA-SNA 1 and 2 differed from naïve mice in their gene-expression profiles in CD4⁺ T cells in a similar way. This suggests a rationale for the significant increases in DA-SNA 2-induced CD8⁺ T cell function but similar levels of CD4⁺ T cell function between both DA-SNAs observed in Fig. 2. Moreover, differentially regulated genes for DA-SNA 2 immunized mice exhibited greater absolute log fold changes (LFCs) in both T cell types compared with the other treatments, with at least double the number of differentially regulated genes as a result of DA-SNA 2 immunization compared with DA-SNA 1 (Fig. 4b). Differentially regulated genes were enriched in pathways involving inflammatory responses and upregulation of pro-inflammatory cytokines, chemotaxis and migration of key immune-cell populations (Fig. 4c and Supplementary Dataset 2). While some of the enriched pathways from DA-SNA 2 treatment were shared with the admix treatment and others with the DA-SNA 1 treatment, overall, the widespread activation induced at the transcriptome level for the DA-SNA 2 architecture correlated with enhanced immunological outputs.

Fig. 4: Immunization of mice with differently structured vaccines induces specific differences in gene expression among CD8⁺ and CD4⁺ T cells.

a, PCA plot from the full transcriptome of CD8⁺ (left) and CD4⁺ (right) T cells. b, Gene-expression changes represented by subsets of LFCs for CD8⁺ (left) and CD4⁺ (right) T cell populations as a result of different treatments. c, Selection of significantly enriched pathways calculated using GSEA analysis. Colours of squares correspond to the enrichment score for each pathway as a result of different treatments for CD8⁺ (left) and CD4⁺ (right) T cells. d, Gene signatures for CD8⁺ (top) and CD4⁺ (bottom) T cells. Colours refer to normalized (z-scored) gene-expression levels. Selection of relevant genes labelled. e, Volcano plots of CD8⁺ (left) and CD4⁺ (right) T cells between a pairwise comparison of DA-SNA 2 and DA-SNA 1. Coloured dots indicate significantly expressed genes; a positive LFC indicates an upregulation for DA-SNA 2 with respect to DA-SNA 1 (red) whereas a negative LFC indicates a downregulation for DA-SNA 2 with respect to DA-SNA 1 (blue).

Relevant gene signatures were identified for adaptive and innate immune activation and functioning across all treatments and include, for example, CXCR3, TNFSF9 and GZMK (Fig. 4d and Supplementary Dataset 3). These genes have particular relevance in T cell effector function and trafficking, antigen presentation and generation of cytotoxic T cells, and helper T cell cytolytic function, respectively. A particular comparison of DA-SNA 2 versus DA-SNA 1 demonstrates unique nanoscale-induced genetic differences simply by altering the placement of antigen class (Fig. 4e). A total of 452 and 229 overlapping significant genes in CD8⁺ and CD4⁺ T cells, respectively, were detected between both DA-SNAs. Specifically, DA-SNA 2 induced higher expression of IL2RA, CD44 and XCL1 in CD8⁺ T cells and LAG3, CCR7 and CCL9 in CD4⁺ T cells compared with DA-SNA 1. Ultimately, comparing gene signatures across all immunization treatments highlights the substantial impact that vaccine structure and in particular, nanoscale antigen placement, have on genome and expression patterns. These results underscore the immunological measurements that were detected, providing a mechanistic rationale that highlighted pathways leading to T cell activation and durable responses, and detail a framework for vaccine design using purposeful structure considerations.

DA-SNA structure-driven tumour inhibition and immune activation

To evaluate the therapeutic efficacy and immunological impact of DA-SNAs, we employed a murine E.G7-OVA lymphoma cancer model due to its stable expression of the OVA protein, expressing both the OVA1 and OVA2 epitopes used above⁴⁶. Briefly, C57BL/6 mice were inoculated subcutaneously with E.G7-OVA cells and immunized weekly with either DA-SNA or admix formulations (6 nmol of each OVA1 and OVA2 antigen, 6 nmol of adjuvant DNA) (Fig. 5a). Tumour-bearing mice immunized with DA-SNA 2 demonstrated a ~3-fold reduction in tumour growth compared with both control (saline-treated) and admix groups as soon as 5 d after the second immunization (day 15) and more than a 16-fold difference in tumour growth compared with saline-treated mice 22 d post-tumour inoculation (Fig. 5b and Supplementary Fig. 15). Importantly, DA-SNA 1 treatment did not effectively halt tumour growth compared with admix, unlike DA-SNA 2 treatment. Compared with the admix and DA-SNA 1, DA-SNA 2 produced a ~7-fold reduction in tumour growth at day 24, highlighting the pronounced impact of antigen positioning and, ultimately, translating to a significant extension in animal survival (median survival in days: PBS = 27; Admix = 24; DA-SNA 1 = 28; DA-SNA 2 = 35) (Fig. 5c). To further investigate the physical impact of treatment on tumour growth, tumours were excised from mice at day 15 following the same treatment regimen and subsequently weighed (Fig. 5d and Supplementary Fig. 16). Interestingly, at this point in the tumour growth curve, both SNA groups showcased a significant reduction in tumour weights compared with PBS-treated mice, suggesting that DA-SNA 1 is capable of raising an antitumour immune response, but this response is not as durable as that raised by DA-SNA 2.

Fig. 5: DA-SNA immune activation for enhanced tumor suppression.

a–c, C57BL/6 mice were subcutaneously inoculated (a) with E.G7-OVA cells (5 × 10⁵) in the right flank and provided weekly immunizations beginning at day 3 for a total of four vaccinations (6 nmol adjuvant, 6 nmol of each antigen). Average growth curves (b) and animal survival (c) are shown. P values shown for volume at day 24 of DA-SNA 2 compared to DA-SNA 1 and Admix (b). Data show mean ± s.e.m. from two independent experiments with n = 7–9. d, Weights following treatment schedule depicted in a (at day 15). e, Left: evaluation of immune CD8⁺ T cells in the spleen at the conclusion of the experiment. Right: ratio of CD8⁺/CD4⁺ T cells. f–i, Flow cytometric analysis at day 15 of PBMCs isolated from tumour-bearing mice under the schedule depicted in a. f, CD8⁺ T cells specific for the OVA1 antigen. g, Effector memory CD8⁺ T cells (CD44⁺/CD62L⁻) within this antigen-specific T-cell subset. h, CD4⁺ T cells specific for the OVA2 antigen. i, Effector memory CD4⁺ T cells (CD44⁺/CD62L⁻). Data show mean ± s.e.m. with n = 4–6 mice per group. For b, d–g and i, significance was calculated using a one-way ANOVA with Tukey’s multiple comparisons test. Panel h used a Welch ANOVA followed by Dunnett’s multiple comparisons test due to significant differences in s.d. between groups. Panel c was analysed using a log-rank test.

Source data

The immunological differences resulting from multi-antigen delivery were further elucidated by collecting spleens from E.G7-OVA-tumour-bearing mice and evaluating changes in splenic CD8⁺ and CD4⁺ T cells (Fig. 5e). The spleens of DA-SNA 2-treated mice generated a significantly higher percentage of CD8⁺ T cells compared with other treatment groups and also displayed an overall higher ratio of CD8⁺ to CD4⁺ T cells. To evaluate the immunological differences that contributed to tumour suppression, tumour-bearing C57BL/6 mice were assessed for circulating peripheral blood mononuclear cells (PBMCs) on day 15, when differences in tumour growth were first observed and when the impact of DA-SNA 2 treatment began to halt tumour growth while the other treatments had negligible impact. Notably, DA-SNA 2-treated mice showcased the highest level of circulating antigen-specific CD8⁺ T cells (Fig. 5f, gating strategy in Supplementary Fig. 17). This subset of CD8⁺ lymphocytes was further evaluated for their memory phenotype. In this case, DA-SNA 2 treatment significantly elevated the effector memory phenotype to over 60% of OVA1-specific circulating CD8⁺ T cells (Fig. 5g). Antigen-specific CD4⁺ T cells were also significantly elevated for mice treated with the DA-SNAs (Fig. 5h). As expected, due to the transcriptome profiles and immunological parameters previously explored herein for CD4⁺ T cells, there were negligible differences between the two DA-SNA groups. While there were not enough OVA2-specific CD4⁺ T cells to accurately delineate the memory phenotype within this subpopulation, the entirety of CD4⁺ T cells demonstrated an enhanced effector memory state when treated with DA-SNA 1 (~30% of CD4⁺ T cells) compared with treatment with DA-SNA 2, which matured ~10% of CD4⁺ T cells (Fig. 5i).

To determine the versatility and translatability of the design rules to guide structural placement for multi-antigen vaccination, we employed the MC-38 colon carcinoma model known for its high mutational burden⁴⁷. Briefly, C57BL/6 mice were inoculated subcutaneously with MC-38 cells and immunized weekly with DA-SNAs containing MHC-I neoantigen ‘Adpgk I’¹⁴ and ‘Adpgk II’, an antigen predicted to bind effectively to MHC-II (https://www.iedb.org and refs. ^48,49,50) (6 nmol each of Adpgk I and Adpgk II antigen, 6 nmol of adjuvant DNA) (Extended Data Fig. 3a–c). Tumour growth was similarly inhibited for mice immunized with DA-SNA 2, with a significant extension in survival conferred to these animals (median survival in days: PBS = 27; DA-SNA 2 = 38). These differences probably result from a combination of significant increases in the tumour microenvironment and among raised circulating immune cells. CD8⁺ and CD4⁺ T cells were increased in the tumour microenvironment (Extended Data Fig. 3d,e, gating strategy in Supplementary Fig. 18) and a significant reduction in Gr-1⁺CD11b⁺ myeloid-derived suppressor cells was observed (Extended Data Fig. 3f). Moreover, DA-SNA 2 treatment increased the percentage of Adpgk I-specific CD8⁺ T cells and induced splenic adaptive T cell immunity. Upon ex vivo stimulation of Adpgk I and II, DA-SNA 2-treated splenocytes secreted more interferon gamma than DA-SNA 1 or PBS-treated splenocytes (Extended Data Fig. 3g,h).

Structural impact of antigen placement in a clinically relevant melanoma tumour model

The findings learned regarding DA-SNA structure heretofore were used to assess the ability of dual-antigen placement to impact growth of B16-F10 melanoma tumours. This model has been established as highly aggressive with enhanced immunosuppressive properties⁵¹. The recently reported⁵² M27 and M30 neo-epitopes containing mutations present only in the tumour were selected as MHC-I and MHC-II antigens, respectively. Initially, C57BL/6 mice inoculated with B16-F10 tumour cells and receiving a weekly vaccination 3 d post-tumour inoculation of either DA-SNA 1 or DA-SNA 2 showcased an inhibition in tumour growth at day 17 compared with saline-treated mice (68% and 48%, respectively) (Fig. 6a,b). Indeed, a statistically significant ~4-fold increase in circulating effector memory antigen-specific CD8⁺ T cells was observed for mice treated with both DA-SNAs relative to saline treatment (Fig. 6c, gating strategy in Supplementary Fig. 17). Therefore, an antigen-specific immune response is produced, but the negligible impact this has on tumour growth indicates the likelihood of a highly immunosuppressive tumour environment that is inhibiting the therapeutic potential of these T cells.

Fig. 6: Inhibition using dual-antigen immunotherapy with immune checkpoint inhibitors.

a,b, C57BL/6 mice were subcutaneously inoculated (a, top) with B16-F10 cells (10⁵) in the right flank and provided weekly subcutaneous immunizations beginning at day 3 for a total of four vaccinations (9 nmol adjuvant, 9 nmol of each antigen). Average growth curves (a, bottom) and animal survival (b) are shown. c, CD8⁺ T cells specific for the M27 antigen and memory CD8⁺ T cell markers 44⁺/62⁻ in isolated PBMCs. Data show mean ± s.e.m. from two independent experiments with n = 6–7. d,e, B16-F10 tumour-bearing mice receiving weekly subcutaneous immunizations of DA-SNAs combined with an anti-PD-1 immune checkpoint inhibitor administered intraperitoneally at 3 and 6 d post DA-SNA immunization. Average growth curves (d) and animal survival (e) are shown. P values shown for growth comparing anti-PD-1 versus DA-SNA 2 + anti-PD-1 at days 17, 20 and 22. Data show mean ± s.e.m. from two independent experiments with n = 9–15. f–k, Flow cytometric analysis at day 17 of PBMCs isolated from tumour-bearing mice receiving the schedule indicated in d. f, Evaluation of circulating CD8⁺ T cells. g, Total effector memory CD8⁺ T cells (CD44⁺/62L⁻). h, M27-specific CD8⁺/CD19⁻ T cells. i, Quantification of circulating CD4⁺ T cells. j, Assessment of effector memory CD4⁺ T cells (CD44⁺/62L⁻). k, M30-specific CD4⁺/CD19⁻ T cells. The data show mean ± s.e.m. from n = 5–6 mice per group. For all panels except b and e, significance was calculated using a one-way ANOVA with Tukey’s multiple comparisons test. Animal survival (b,e) was analysed using a log-rank test. Hashtags in d indicate statistically significant differences between DA-SNA 2 + anti-PD-1 and anti-PD-1 treatment groups. ND, not detected.

Source data

Due to these observations and the inherent aggressiveness of this tumour model, DA-SNA treatments were combined with the immune checkpoint inhibitor anti-PD-1, a US Food and Drug Administration (FDA)-approved treatment for advanced melanoma, in an effort to overcome the tumour’s immunosuppression^53,54. These treatments began 3 d after B16-F10 tumour inoculation. Notably, when anti-PD-1 was co-administered with DA-SNAs, a significant decrease in tumour growth was observed beginning as early as 17 d post-tumour inoculation for animals treated with the combination DA-SNA 2 + anti-PD-1 therapy, while no substantial decrease in tumour growth was observed for mice in the other treatment groups (Fig. 6d and Supplementary Fig. 19). This translated to a 40% extension in median survival for these mice compared with those in the saline-treated or anti-PD-1 monotherapy-treated groups (Fig. 6e). The importance and the role that nanoscale antigen placement plays in driving synergistic and enhanced immune responses can be drawn from these results. Importantly, a comparison between DA-SNA combination treatment groups revealed an improvement in overall median survival for combination DA-SNA 2 + anti-PD-1 animals (P = 0.0507). The evaluation of immune cells isolated from peripheral blood further highlights this structure-induced difference where DA-SNA 2 appears to work synergistically with the checkpoint inhibitor. A significant increase in circulating CD8⁺ T cells for animals receiving a combination DA-SNA 2 + anti-PD-1 treatment was observed compared with all other groups (Fig. 6f). Interestingly, when the total CD8⁺ effector memory T cell population was evaluated among these circulating PBMCs, significant increases were detected for both combination DA-SNA + anti-PD-1 groups, with DA-SNA 2 + anti-PD-1 generating effector memory phenotypes in ~60% of circulating CD8⁺ T cells (Fig. 6g). Moreover, only this combination DA-SNA 2 + anti-PD-1 treatment was able to induce a robust antigen-specific CD8⁺ T cell response (Fig. 6h). Observations of CD4⁺ T cell circulation revealed a similarly high increase as a result of combination DA-SNA 2 + anti-PD-1 therapy (Fig. 6i), although both combination DA-SNA + anti-PD-1 groups significantly elevated the effector memory phenotype to ~28% of the CD4⁺ T cell population (Fig. 6j). The combination DA-SNA 2 + anti-PD-1 treatment increased levels of antigen-specific CD4⁺ T cell production along with anti-PD-1 monotherapy compared with DA-SNA 1 + anti-PD-1 treatment, although significant differences were not observed between groups (Fig. 6k).

Outlook

Although extensive research has explored the importance of adjuvants and antigens in the creation of new immunotherapies, so far, the importance of the structural presentation of multiple antigens within a specific construct and its role in eliciting a potent and desired immune response have not been explored. In this work we have shown that, for vaccine efficacy, antigen placement may be as critical as antigen choice. Indeed, when altering the placement of MHC-I-restricted and MHC-II-restricted antigens in two compositionally nearly identical vaccines, the treatment benefit against tumours is dramatically changed; one vaccine is potent and the other is ineffective. The origins of these differences may be due to antigen positioning affecting the pathway of processing that it undergoes in an immune cell, as well as its residence time in different cellular compartments. By changing the processing pathway and these kinetics of signalling, this affects the resulting immune response at the genetic, cellular and organismal levels. An encapsulated MHC-II and a hybridized MHC-I-restricted antigen upregulate genes specific for inflammatory responses, chemotaxis and migration of key immune cells, which together influence immune cell activity. These structurally defined genetic differences translate through to immunological behaviour upon repeated in vivo immunization, and ultimately define the tumour growth profiles in multiple tumour models, including E.G7-OVA lymphoma, and clinically relevant MC-38 colon carcinoma and B16-F10 melanoma mouse tumour systems. This is a key demonstration of the impact of vaccine antigen positioning across multiple cellular processes.

Vaccine development has focused on the composition, number and type of antigen, and the ratio of these components, with almost no attention paid to their structural presentation. It is clear that in addition to these important parameters, presentation of antigen must be a focus in future vaccine development. Having the power to optimize antigen presentation to match a desired signalling profile is critical to generating potent future vaccines, where small vaccine changes in antigen placement substantially elevate cell–cell communication, cross-talk and cell synergy. This insight can be harnessed for targets yet to be discovered, as well as those already in use. Taken together, the developments made in this work provide a path forward to rethink the design of vaccines for cancer and other diseases.

Methods

Materials and animals

Unless otherwise noted, all reagents were purchased commercially and used as received. Oligonucleotides were synthesized as described below. Peptides were purchased from Genscript or Northwestern’s Peptide Synthesis core. Chemicals were purchased from suppliers listed in parentheses. C57BL/6 mice and C57BL/6-Tg(TcraTcrb)1100Mjb/J (OT1, 003831) female mice aged 8–12 weeks old were purchased from Jackson Laboratory. Mice were used in accordance with all national and local guidelines and regulations, and protocols performed were approved by the Institutional Animal Care and Use Committee at Northwestern University. E.G7-OVA and B16-F10 cells were purchased from ATCC. MC-38 cells were kindly provided by Dr Bin Zhang. Antibodies were purchased and clones are provided in Supplementary Table 3.

Oligonucleotide synthesis and purification

Oligonucleotides were synthesized on an ABI 394 synthesizer using standard phosphoramidite chemistry with phosphate or phosphorothioate backbones as indicated (Supplementary Table 1). Following synthesis, the strands were deprotected using a 1:1 solution of 37% ammonium hydroxide/40% methylamine at 55 °C for 35 min, unless they contained a dye, in which case they were deprotected using 37% ammonium hydroxide at room temperature (r.t.) overnight. The strands were then purified using a C18 or C4 (for strands containing dye or cholesterol) column on reverse-phase HPLC, and the peaks were collected as fractions. The dimethoxytrityl (DMT) group was removed from the product strands by incubation in 20% aqueous acetic acid at r.t. for 1 h, followed by three washes with ethyl acetate to remove DMT. The final product was lyophilised and resuspended in deionized water (diH₂O). The concentration was measured using UV-vis absorption at 260 nm, with extinction coefficients calculated through the IDT OligoAnalyzer online tool (listed in Supplementary Table 1).

Oligonucleotide–peptide conjugate synthesis and purification

Thiol-functionalized oligonucleotides in diH₂O were reduced to generate a free thiol for future reactions. Reduction was done using dithiothreitol (100 mM, DTT, Sigma) dissolved in phosphate buffered saline (PBS) pH 8.5 at a final concentration of 100 mM at r.t. for 45 min. This solution was washed in a 3 kDa molecular weight cut-off (MWCO) spin filter (Amicon) at least 3 times with diH₂O. For OVA peptide conjugates, peptide was purchased on resin and washed three times each with dimethylformamide (DMF) and acetone before reacting 5 µmol at r.t. overnight with a solution of succinimidyl 2-(2-pyridyldithio)ethyl carbonate (SDEC, made using previous protocols⁵⁵) dissolved in DMF (10 equivalents with respect to the initial peptide loading on the solid support), with N,N-diisopropylethylamine (5 equivalents). The beads were subsequently washed three times with DMF and then acetone, and dried in air before being deprotected with 95% trifluoroacetic acid (2.5% triisopropyl silane, 2.5% diH₂O) for 1 h at r.t. The trifluoroacetic acid was blown off using nitrogen, and the beads were redissolved in DMF and filtered through glass wool. The peptide product was precipitated by adding ~5–6 times diethyl ether and was left at −20 °C for 1–2 h to further precipitate. The solution was centrifuged (2,000 × g, 3 min) to pellet the peptide, which was collected, dried and dissolved in DMF. Reduced DNA (0.5 µmol) was reacted overnight at r.t. with the dissolved peptide (5 µmol) in 70–75% DMF in water for a total reaction volume of ~1.5 ml. For peptide conjugates, the peptides were activated using 2,2′-dithiodipyridine (150 µmol) dissolved in 10 equivalents DMF under gentle agitation for 30 min at r.t.. The activated peptide was then washed 3 times in diethyl ether, pelleted by centrifugation (2,000 × g, 3 min) and allowed to dry. Reduced DNA (0.3 µmol) was reacted overnight at r.t. with the dissolved peptide (1.5 µmol) in ~70% DMF in water for a total reaction volume of ~1.5 ml.

Following conjugation of the peptide, the solutions were centrifuged at 17,000 × g for 2 min to pellet any precipitated peptide, and the supernatant was transferred to 3 kDa MWCO spin filters for 3–4 washes with diH₂O. The volume was concentrated to <500 μl, and the solutions were purified by preparatory scale denaturing (8 M urea) 15% PAGE gels (no more than 0.5 μmol by DNA loaded onto a single gel). The gels were run for 30 min at 175 V, then ~3 h at 350 V, and subsequently imaged using a UV lamp to cut out desired bands. Cut-out gel bands were crushed, and the product was collected by three washes with 1x Tris/Borate/EDTA buffer every ~4 h. The product mass was confirmed by electrospray ionization mass spectrometry (ESI-MS), and the concentrations were measured by UV-vis at 260 nm assuming an extinction coefficient of the DNA.

SNA synthesis

SNAs were synthesized as reported previously with slight modifications^16,56. Briefly, dried lipid films of 50 mg of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC, Avanti Polar Lipids) were hydrated with 3–4 ml PBS or PBS-containing peptide for encapsulated liposomes. (Note: solutions containing peptide included OVA1: 1 mg ml⁻¹ dissolved in PBS containing ~100 µl 1 M NaOH; OVA2: 1 mg ml⁻¹ dissolved in PBS containing ~60 µl 1 M NaOH; M27: 2.25 mg ml⁻¹ dissolved in PBS containing ~60 µl 1 M NaOH; M30: 1 mg ml⁻¹ dissolved in PBS containing ~100 µl 1 M NaOH; AgdpkI: 1mg/ml dissolved in PBS containing 120 μl 1M NaOH; Agdpk II: 0.65 mg/ml dissolved in PBS containing 50 μl 1M NaOH.) The solutions were subjected to 20 freeze-thaws in liquid nitrogen and then sonication at 37–40 °C. The liposomes were extruded using sequential high-pressure extrusion (Northern Lipids) using polycarbonate filters with pore sizes of 200, 100, 80 and 50 nm; liposomes were passed through each pore size 3 times. Following extrusion, the liposomes were concentrated down to ~2–3 ml using 100 kDa MWCO spin filters and dialysed overnight against 3.5 l PBS to remove unencapsulated peptide. The liposome concentration was determined using a phosphatidylcholine assay kit (Sigma, MAK049-1KT), assuming that a 50 nm liposome contains 18,140 lipids per liposome²⁴. Peptide concentration, if liposomes encapsulated peptide, was determined using a Pierce fluorescence assay kit (ThermoFisher, 23290) adding 1% sodium dodecyl sulfate (SDS) to rupture liposomes and release peptide for quantification, and using peptide dissolved in 1% SDS as a standard curve. The loading of peptide per liposome was calculated by dividing the peptide concentration by the liposome concentration. The amounts of each antigen per particle ranged from ~25 to 40 depending on the encapsulation yield of each batch, which was tuned using different starting concentrations.

Purified oligonucleotide–peptide conjugates were mixed in a 1:1 molar ratio with complementary 3’-cholesterol-terminated CpG DNA and centrivapped overnight. The next day, ~20–40 μl of duplex buffer (IDT) was added and the solution was slow-cooled to duplex the strands following the programme: 70 °C for 10 min, 23 °C for 1.5 h, 4 °C for ≥1 h. The duplex was added to a solution of synthesized liposomes at an equimolar amount to the peptide encapsulated within the liposome. To obtain the maximum 75 strands per liposome, the remaining space was filled with 3’cholesterol-terminated T₂₀ DNA. This mixture was incubated at 37 °C overnight and subsequently stored at 4 °C.

Ex cellulo-release study

DA-SNAs containing fluorophore-labelled antigens (2 μM in 1.5 ml volume) were placed into Slide-A-Lyzer MINI Dialysis Device in 50 ml falcon tubes with a 10K MWCO. The falcon tube solution was prepared to be 10% fetal bovine serum (FBS) in PBS. Samples were loaded and left on a rotator, and at specified timepoints, 200 μl of sample was collected and frozen at −20 °C until analysis using a BioTek plate reader.

Cell culture

All cells were maintained at 37 °C in a 5% CO₂ incubator. E.G7-OVA, MC-38 and DCs were cultured with RPMI 1640 media (Gibco, 11875093) containing 10% heat-inactivated (HI)-FBS and 1% penicillin-streptomycin, referred to herein as RPMI+/+. B16-F10 were handled using DMEM media (Gibco, 11965092) containing 10% HI-FBS and 1% penicillin-streptomycin.

BMDC collection

Bone marrow cells were collected from mice following a previous protocol⁵⁵. Briefly, red blood cells were lysed with 2–3 ml of ACK lysis buffer (Gibco, A1049201) for ~4 min and plated on 10 cm² cell culture dishes with 40 ng ml⁻¹ granulocyte-macrophage colony-stimulating factor (BioLegend, 576304) for 5–7 d before use to differentiate DCs from the population.

BMDC activation and cross-priming of T cells in vitro

The cells were collected from 10 cm² cell culture dishes, and DCs were isolated from the mixture using a magnetic biotin positive selection kit (Stemcell Technologies, 17665). A CD11c⁺ biotin-labelled antibody was used to select DCs (BioLegend) and, after separation, purified DCs were counted using a Vi-CELL BLU cell viability analyzer. For DC activation, 6 × 10⁴ DCs were cultured with SNA treatment in a final volume of 200 μl for a 30 min pulse. Cells were then washed twice with RPMI+/+ and left for 22 h in an incubator, after which cells were washed with PBS to end treatment, stained for 15 min at 4 °C using 0.5 μl of each antibody per tube (L/D, CD11c, CD86 and CD80), washed with PBS and fixed with 100 μl of fixation buffer (BioLegend, 420801). For studies using an MHC-I blocker, cells were incubated first for 30 min in 100 μl volume with 25 μg ml⁻¹ of anti-MHC-I (BioXCell, E0077). Subsequently, SNAs in 100 μl volume were added to the cell and blocker solution and the samples were pulsed for 30 min. The remaining steps follow the protocol above. To assess T cell specificity and activation, 1.6 × 10⁵ purified DCs were pulsed with SNA treatment for 30 min in the incubator in a final volume of 200 μl. After the 30 min pulse, the cells were washed twice with RPMI+/+ to remove any residual SNAs from the cell solution, and the cells were resuspended in 500 μl RPMI+/+. Concurrently, splenocytes were isolated from a naïve mouse. After dissociation of the spleen and lysis of the red blood cells, the cells were counted and resuspended to a concentration of 3 × 10⁶ cells per ml in warmed RPMI+/+, and 100 μl of this cell solution was transferred to each well in a 96-well round-bottom plate. To each well, 100 μl of treated DCs (3.3 × 10⁴ cells) were added so that the ratio of DC:splenocytes was 1:9. The cells were co-cultured for 3 d in the incubator, after which cells were washed with PBS and stained following the manufacturer’s instructions for either the DimerX Mouse H-2Kb:Ig Fusion Protein (BD, 552944) or OVA2 Tetramer (ProImmune). Staining antibodies in addition to the peptide-specific TCR markers included L/D, either CD8 or CD4, CD19 and CD69. After staining, the cells were fixed with 100 μl of fixation buffer. To assess T cell proliferation, 2.6 × 10⁵ purified DCs were pulsed with SNA treatment for 30 min in the incubator in a final volume of 200 μl. After the 30 min pulse, the cells were washed twice with RPMI+/+ to remove any residual SNA from the cell solution, and the cells were resuspended in 266.6 μl RPMI+/+. Concurrently, splenocytes were isolated from a C57BL/6-Tg(TcraTcrb)1100Mjb/J (OT1) mouse (Jackson, 003831). After dissociation of the spleen and lysis of the red blood cells, the cells were counted and resuspended to a concentration of 4 × 10⁷ cells per ml in PBS for staining with cell proliferation dye eFluor 450 (eBioscience, 65-0842-85), following the manufacturer’s instructions. After staining, the cells were washed, counted and resuspended in RPMI+/+ to a concentration of 3 × 10⁶ cells per ml; 100 μl of this cell solution was transferred to each well in a 96-well round-bottom plate. To each well, 33.3 μl of treated DCs (3.3 × 10⁴ cells) were added so that the ratio of DC:splenocytes was 1:9, and each well was brought up to 200 μl final volume with media. The cells were co-cultured for 3 d in the incubator, after which cells were washed with PBS, stained for CD8 (0.5 μl antibody per tube), washed and immediately analysed by flow cytometry using dilution of eFluor 450 as a measure of T cell proliferation. All samples were analysed using a BD A3 Symphony flow cytometer, with data analysed on FlowJo.

In vivo immunization to measure 5-week built-up immune responses

Female C57BL/6 mice were subcutaneously immunized fortnightly 3 times with different treatments. Treatments included: simple mixture (admix, 6 nmol of each peptide and 6 nmol CpG 1826 DNA), either DA-SNA (6 nmol OVA peptide and CpG 1826), or equivalent separate formulations of either DA-SNA (6 nmol OVA peptide and CpG 1826) or a double hybridized DA-SNA (termed ‘HH’; 6 nmol OVA peptide and 12 nmol CpG 1826). Volume of treatment injected was kept at 100 μl. One week after the final immunization, mice were killed and spleens were collected for subsequent immune assessment.

Collection procedure

Removed spleens were collected and held temporarily in 3–5 ml of RPMI+/+ until all spleens were collected, then they were passed through a 70 μm cell strainer with a constant flow of PBS. The cells were centrifuged at 1,200 r.p.m. for 5 min, after which supernatant was removed and the cells were resuspended in 2–3 ml ACK lysing buffer (Gibco, A1049201) for 4 min. To dilute the lysing buffer, PBS was then added to a final volume of 30 ml, and the cells were counted before centrifugation to resuspend in RPMI+/+ media at a concentration of 1 × 10⁸ cells per ml.

IFN-γ cytokine production

T cells were restimulated ex vivo to assess antigen-specific intracellular IFN-γ production. Splenocytes (4 × 10⁶) were cultured for 4 h at 37 °C in a 5% CO₂ incubator with 450 μl of RPMI+/+ media containing: either OVA1 or OVA2 peptide (10 μg ml⁻¹), monensin (2 μM), Brefeldin A (5 μg ml⁻¹) and CD107a antibody (0.5 μl). After the 4 h incubation, cells were centrifuged at 1,200 r.p.m. for 5 min, aspirated and washed with 600 μl PBS before 15 min of staining with surface antibodies (0.5 μl per sample each of: L/D, CD8 and CD4) at 4 °C. The cells were washed with 600 μl PBS, centrifuged at 1,200 r.p.m. for 5 min, aspirated and resuspended in 100 μl of Cytofix Fixation and Permeabilization solution (BD, 554722) for 20 min at 4 °C. The cells were then washed with 600 μl of Perm/Wash buffer (BD, 554723), centrifuged at 1,200 r.p.m. for 5 min, aspirated and resuspended in 100 μl of Perm/Wash buffer containing the intracellular antibody IFN-γ (0.5 μl per sample). The samples were stored at 4 °C before flow cytometry analysis.

T cell-memory phenotyping

T cells were assessed for effector memory phenotype. Splenocytes (3 × 10⁶) were washed with 600 μl PBS and stained for 15 min with surface antibodies (0.5 μl per sample each of: L/D, CD8, CD4, CD44 and CD62L) at 4 °C. The cells were washed with 600 μl PBS, centrifuged at 1,200 r.p.m. for 5 min, aspirated, resuspended in 100 μl of Fixation buffer (BioLegend, 420801) and stored at 4 °C before flow cytometry analysis.

ELISpot assay

ELISpot analysis was performed using the commercially available mouse INF-γ ELISPOT set (BD, 551083) following the manufacturer’s instructions. Briefly, the provided clean plate was coated with capture antibody overnight at 4 °C. Then the plate was washed with RPMI+/+ media and then blocked for 2 h at room temperature with 200 μl of RPMI+/+ media. The blocking buffer was removed by pipetting, being mindful not to let wells dry out, and quickly replaced with 2 × 10⁵ splenocytes in 100 μl RPMI+/+. To each well, an additional 100 μl of either antigen, non-specific peptide, media (negative control) or positive control solutions were added (antigen and non-specific peptide were added to a final concentration of 5 μg ml⁻¹; positive control was prepared as a mixture of anti-CD3 and anti-CD28 antibodies at a final concentration of 2 μg ml⁻¹ each). The solutions were left in an incubator at 37 °C in 5% CO₂ for 48 h. After this incubation, the plate was washed, and detection antibody, enzyme conjugate, and chromogenic substrate were added according to the manufacturer’s instructions. The dried plate was imaged and analysed using a CTL Immunospot imager.

Confocal microscopy

To evaluate the uptake and intracellular trafficking of DA-SNAs, BMDCs were collected from murine femurs and purified for CD11c⁺ (as described above) and subsequently seeded on 8-chambered slides (Lab-Tek, 155409) at 50,000 cells per well. Following overnight incubation, cells were treated with DA-SNAs (2.5 μM) containing fluorophore-labelled OVA1 and OVA2 antigens for either 0.5, 1, 6 or 24 h. For 6 and 24 h, cells were pulsed for 2 h with DA-SNAs and replaced with fresh media for the remaining time. After incubation at the indicated timepoints, cells were then fixed for 15 min (BioLegend) and blocked with 5% BSA (ThermoFisher) in PBS containing 0.1% Triton-X for 1 h. Cells were then stained for organelle markers using primary antibodies for EEA1 (ratio 1:700, Abcam), Rab7 (1:500, Abcam), LAMP1 (1:200, ABclonal), PDI (1:50, Cell Signaling), MHC-I (1 μg ml⁻¹, BioXCell) and MHC-II (1 μg ml⁻¹, BioXCell) overnight at 4 °C before secondary labelling with Alexa Fluor 555 (EEA1, Rab7, LAMP1 and PDI; Abcam ab150078) and Alexa Fluor 594 (MHC-I and MHC-II; ThermoFisher A48264) for 1 h at 4 °C. Cell nuclei were stained with DAPI for 1 min and stored in PBS until imaging. Imaging was performed using a Zeiss LSM 800 microscope, maintaining the same parameters for all images. Mander’s overlap coefficient was calculated using ZEN software (Zeiss). For inhibitor studies (including those done via flow), cells were seeded at 150,000 cells per well. Following overnight incubation, cells were incubated with chloroquine (5 μM, Sigma), Brefeldin A (2 μg ml⁻¹, BioLegend) or leupeptin (100 μM, Sigma) for 1 h. Following 1 h incubation, media containing DA-SNA (2.5 μM) and indicated inhibitors were added and pulsed for an additional 1 h. After pulsing with DA-SNAs, wells were washed with media and subsequently replaced with media containing inhibitors for a total incubation time of 24 h. Organelle staining was performed as described above.

Bulk RNA sequencing

Dendritic cells, CD4⁺ and CD8⁺ T cells were isolated from whole splenocytes from individual treatment groups following three fortnightly subcutaneous immunizations using magnetic positive selection kits (Stemcell Technologies, 17665, 18952 and 18953). From these isolated cell populations, RNA extraction was performed using an RNeasy Plus mini kit (Qiagen) in combination with QIAshredders (Qiagen) following the manufacturer’s specifications. RNA concentration was quantified using a NanoDrop 8000 (ThermoFisher), and RNA samples were stored at −80 °C until further use. Sequencing was conducted at the Northwestern University NUSeq Core Facility. Briefly, total RNA examples were checked for quality using RNA integrity numbers (RINs) generated from Agilent Bioanalyzer 2100. RNA quantity was confirmed with a Qubit fluorometer. The Illumina TruSeq Stranded mRNA Library Preparation kit was used to prepare sequencing libraries from 125 ng of high-quality RNA samples (RIN > 7). The kit procedure, including mRNA purification and fragmentation, cDNA synthesis, 3’ end adenylation, Illumina adapter ligation, library PCR amplification and validation, was performed without modifications. Libraries were sequenced using an Illumina HiSeq 4000 sequencer to generate 50 bp single reads at the depth of 20–25 million reads per sample. The quality of reads, in FASTQ format, was evaluated using FastQC. Reads were trimmed to remove Illumina adapters from the 3’ ends using cutadapt⁵⁷. Trimmed reads were aligned to the Mus musculus genome (mm10) using STAR⁵⁸. Read counts for each gene were calculated using htseq-count⁵⁹ in conjunction with a gene annotation file for mm10 obtained from Ensembl (http://useast.ensembl.org/index.html). Normalization and differential expression were calculated using DESeq2 that employed the Wald test⁶⁰. The cut-off for determining significantly differentially expressed genes was a false discovery rate (FDR)-adjusted P value less than 0.05 using the Benjamini-Hochberg method.

Gene set enrichment analysis (GSEA)

GSEA⁶¹ was performed to understand whether differentially expressed genes were connected to differentially enriched pathways. Genes detected with RNA sequencing were ranked on the basis of log₁₀-transformed nominal P values obtained from DeSeq2 analysis and were compared to naïve T cells. Pathway enrichment analysis was performed using the GSEA software (v4.0.3) and following the protocol in ref. ⁶². Gene sets were obtained from the Molecular Signatures Database and included Reactome and KEGG. The ranked list was remapped using a CHIP technology from the Molecular Signatures Database that used the Mouse Gene Symbol to remap to Human Orthologs (v7.1). A term was defined as differentially enriched if it had an FDR < 0.05. A subset of strongly enriched pathways was selected for visualization in R using the pheatmap package (v1.0.12). This selection included all pathways with an FDR < 0.05 and was relevant to immune responses in dendritic, CD8⁺ and CD4⁺ T cells.

Gene-expression profiles

Genes whose expression was significantly altered in both SNA treatment groups, as defined by FDR P < 0.05, were selected for visualization as heat maps. Gene-expression scores in FPKM were converted to z-scores across treatment groups and gene-expression values clustered using K-means clustering. Pairwise combinations were performed between two conditions of interest, setting naive CD4⁺ or CD8⁺ T cells as controls. Genes up or downregulated between groups as defined by FDR P < 0.05 and log₂ fold change of >0.5 (upregulated) or log₂ fold change <0.05 (downregulated) were visualized using a volcano plot.

In vivo efficacy studies

Female C57BL/6 mice aged 8–12 weeks were acquired from Jackson Laboratory. Tumour inoculation was performed by subcutaneously (s.c.) injecting animals with either 5 × 10⁵ E.G7-OVA, 10⁵ B16-F10, or 5 × 10⁵ MC-38 cells in the right flank. Immunizations were administered at a dose of either 6 nmol (OVA1/2 and Adpgk I/II) or 9 nmol (M27/30) of each antigen and CpG by s.c. injection in the abdomen. Immunizations were administered as listed in the treatment schedule provided in respective figures. For combination therapy with the immune checkpoint inhibitor anti-PD-1, mice were administered 100 μg anti-mouse PD-1 (clone RMP1-14, BioXCell) via intraperitoneal injection. Tumour growth was measured on pre-determined days and volume was calculated using the following equation: tumour volume = length × width² × 0.5. Animals were euthanized when tumour volumes reached either 2,000 mm³ (E.G7-OVA) or 1,500 mm³ (B16-F10, MC-38) or when the animal became moribund.

In vivo biodistribution and splenic uptake

Biodistribution of DA-SNAs to major organs was performed in female C57BL/6 mice (n = 3) using fluorophore-labelled OVA peptides and a single dose at 6 nmol administered s.c. After 24 h, whole organs were collected and stored briefly in PBS until imaged. Imaging was performed using an In Vivo Imaging System (IVIS) with excitation/emission filters set for 500/540 (FITC) and 640/680 (Cy5). Hybridized antigens were labelled using Cy5 fluorescent dye, while encapsulated antigens were labelled using FITC fluorescent dye. Data were quantified by measuring with the Living Image software v4.5. Spleens collected from biodistribution analysis were mashed through a 70 μm cell strainer after imaging. Cells were pelleted at 1,200 r.p.m. for 5 min, and cells were resuspended in ACK lysing buffer (Gibco) for 4–5 min. Cells were diluted in PBS up to ~25–30 ml and counted. Three million cells were put into flow tubes and washed with PBS. Cells were incubated with staining solution containing 0.5 μl each of: fixable live/dead-UV and CD11c (clone N418, PE) for 15 min at 4 °C in 100 μl PBS. Supernatant was removed following a wash and centrifugation, and cells were fixed in 100 μl of fixation buffer (BioLegend, 420801) and stored at 4 °C before flow cytometry.

Immunoactivation of PBMCs

For collection of PBMCs, animals were inoculated with cancer cells as described above. Treatment was performed following the same schedule and animals were euthanized on day 15 (E.G7-OVA), day 16 (MC-38) or day 17 (B16-F10). Blood was collected via cardiac puncture into EDTA-lined collection tubes (BD) and briefly mixed by inverting. Red blood cells were lysed using ACK lysing buffer (Gibco) and washed, and the remaining cells were subsequently stained using the methods described above with antibodies for L/D, CD4, CD8, CD19, CD44, CD62L and antigen-specific dimer or pentamer, and tetramer (E.G7-OVA, B16-F10) or L/D, CD8, CD19 and antigen-specific pentamer (MC-38).

Whole-organ immune assessment

Tumour weights and splenocyte evaluation were performed on C57BL/6 mice bearing an E.G7-OVA tumour in the right flank. At 3 d after tumour inoculation, the first immunization was administered, followed by an additional dose 7 d later (day 10). The tumours and spleens were excised from the animals at day 15 and subsequently analysed. Tumour microenvironment evaluation was performed on C57BL/6 mice bearing an MC-38 tumour in the right flank. At 3 d after tumour inoculation, the first immunization was administered, followed by an additional dose 7 d later (day 10). Tumours were excised from the animals at day 16 and subsequently analysed. To generate single-cell solutions, tumors were mechanically forced through a 70 μm cell strainer while maintaining hydration in PBS solution. The cells were subsequently centrifuged at 1,200 r.p.m. for 5 min. The spleen pellet was resuspended in ACK lysing buffer (ThermoFisher) for 4 min to lyse red blood cells and subsequently neutralized in PBS before centrifugation. Following centrifugation, the cells were labelled using the following antibodies: E.G7-OVA: CD4, CD8, CD19 and L/D; MC-38: CD4, CD8, CD45, CD11b, Gr-1 and L/D.

Statistical analyses

Statistics were calculated using GraphPad Prism 9 software, and specific statistical analyses used are highlighted in the respective figure captions. Comparisons between multiple groups were analysed with a one-way analysis of variance (ANOVA) with a Šidák or Tukey’s multiple comparisons test, or a Welch ANOVA with a Dunnett multiple comparisons test due to the lack of assumptions that could be made based on large differences in s.d. between groups. Statistics for animal survival were calculated using a log-rank test. Outliers for Fig. 5e,f and Fig. 6h,j were identified using the ROUT method with a Q set to 10% or 1%, respectively. For all cases, P values are depicted as follows: *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. The allocation of animals to each group, administered immunizations and measurements for studies were performed blind. Values in graphs are depicted as the mean ± s.e.m. or s.d., and this, as well as sample sizes, are indicated in the respective figure captions.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.