IntroductionEffective vaccination relies on the ability to induce humoral and cellular immunity, with the nature of this adaptive immune response being highly dependent on the co-delivery of distinct antigen and adjuvant combinations to innate immune cells. Mimicking the diverse spectrum of immunostimulation elicited by attenuated viral and bacterial vaccines has remained a challenge for subunit vaccine formulations, which typically pair a single adjuvant and antigen for simplicity and scalability1,2,3. Furthermore, despite recent advances in creating hydrogels capable of sustained delivery of protein antigens4,5,6, approaches to combine antigens with physicochemically distinct adjuvants remain challenging. In particular for hydrophobic and charged macromolecular adjuvants in the form of lipids, small molecules, or nucleic acids, controlling the concentration and stability of each during multi-component loading within a hydrophilic network is rarely achieved7. To this end, nanocomposite and micelle-cross-linked hydrogels8,9 have been used to load hydrophobic cargoes10. These hierarchical materials are assembled via multi-length scale processes and frequently possess biomimetic properties that include multifunctionality, environmental responsiveness, and self-organization11. The fabricating process for the artificial hierarchical structures is, however, tedious and the applied synthetic components are complex.Here, we report on hierarchical hydrogels, assembled from a single simple homopolymer via a multi-stage process, that enable ratiometric loading and delivery of four physicochemically distinct adjuvants along with a protein-based antigen (Fig. 1). Starting with a poly(propylene sulfone) (PPSU)12 solution, the addition of water promotes interchain sulfone-sulfone bonding, yielding nanogels with short PPSU coronas capable of self-bonding or protein anchoring. When transferred into protein solutions in saline, the nanogels irreversibly aggregate into microgels as a secondary structure via surface fusion or protein bridging. Following subcutaneous injection, the microgels then bind to extracellular matrix components, mainly collagen fibers, via PPSU coronas to form stable in situ nanocomposite hydrogels as the tertiary structure. The delivery system takes advantage of efficient loading (>90%) throughout the process, allowing us to develop an optimized vaccine formulation that consists of monophosphoryl lipid A (MPLA)13, CL42914, CpG oligodeoxynucleotides (ODN)15, lipopolysaccharide (LPS)16, and the protein antigen ovalbumin (OVA). In vitro and in vivo experiments demonstrate enhanced humoral and cell-mediated immune responses for these five-component nanogel vaccines compared to formulation with a more standard single adjuvant and antigen pair.Fig. 1: Schematic of PPSU-based hierarchical hydrogels that enable ratiometric loading of multiple adjuvants for vaccine optimization.Hydrophobic adjuvants are encapsulated through network assembly of PPSU (1st gelation). Water-soluble adjuvants are adsorbed during the formation of microgels triggered by bridging nanogels with the antigen OVA (2nd gelation, only simple mixing steps that take less than 5 min before injection). Microgels anchor onto collagens upon subcutaneous injection (3rd gelation), allowing sustained release. Within the first 24 h after vaccination, Th1/Th2 cytokine levels increased. By day 7, anti-OVA IgM antibodies had risen, and over the subsequent four days, the vaccination effectively promoted the proliferation of adoptively transferred antigen-specific T cells. Anti-OVA IgG2b and IgG2c responses remained significantly elevated for up to one month, while significant levels of anti-OVA IgG1 antibodies persisted through day 84. Following a prime-boost administration on day 90 and a subsequent 15-day period, marked increases in neutrophils, CD11b+ conventional dendritic cells (cDCs), and B cells were observed.Full size imageResults and discussionRatiometric loading of five separate vaccine immunostimulantsThe preparation of PPSU nanogels has been described in our previous publications12,17. As prepared, the nanogels have a zeta potential between −30 and −40 mV12,17 that prevents aggregation in water. Importantly, the coronas of the nanogels consist of “living” PPSU that will undergo further sulfone-sulfone bonding on demand. This PPSU corona is also capable of anchoring proteins while preserving protein bioactivities at the interfaces17. We investigated nanogel aggregation in aqueous solutions, a requirement to inhibit cell uptake through the formation of microgels before subcutaneous injection (Supplementary Fig. S1). The strategy involves the addition of salts to promote corona fusion by shielding the electrostatic repulsion among nanogels, or the use of low concentration protein solutions (e.g. hemoglobin, neutral charged) in water to bridge two nanogels through surface adsorption. By employing protein payloads as the crosslinking agent, the minimalist secondary microgels engineered here do not require additional cross-linking agents or stabilizers, allowing rapid and simple preparation of a protein-based microgel suspension that is amenable to facile injection.We explored the versatility and universality of this system by extending the study to other proteins, such as OVA. Although PPSU surfaces show high affinity to nonspecific proteins, the formation of OVA-based microgels was achieved in phosphate buffered saline (PBS) rather than in water (Fig. 2a) because the presence of salts is required to trigger the adsorption of highly negatively charged proteins17. That is, incubating PPSU nanogels in PBS with OVA leads to hybrid microgels through both corona fusion and crosslinking via protein antigen bridges (Fig. 2b, c and Supplementary Fig. S2). The microgels are not only bioactive based on the demonstrated protein adsorption mechanism17, but also support the pre-encapsulation of physicochemically distinct adjuvants given their high encapsulation efficiencies (>95%) for a wide range of organic cargoes12.Fig. 2: PPSU nanogels enable the ratiometric loading of multiple cargoes and subsequent assembly into microgels.a Comparison of the extinction spectra of the mixture of PPSU nanogels (1 mg/mL) with 1 wt.% of OVA in PBS or water. b CryoSEM of PPSU microgels following salt-induced gelation of nanogels via OVA bridges in PBS. c CryoTEM of PPSU microgels formed by mixing PPSU nanogels with OVA in PBS. (b-c) All experiments were independently repeated three times. d The complete self-quenching indicates the efficient capture of ICG during the formation of microgels in PBS. The inset shows ICG/PBS solutions before and after the addition of PPSU nanogels. The photos were taken after centrifugation. e Computational studies suggest that the PPSU surface is capable of molecular capture through hydrophobic interactions (Lennard-Jones). f Multiple adjuvants, including MPLA, CL420, CpG, and LPS can be co-loaded in a ratiometric manner while using the antigen OVA to bridge nanogels.Full size imageAmphiphiles can also to be loaded during the formation of microgels. We demonstrated, both experimentally and through simulations, the preserved affinity of PPSU surfaces to capture amphiphilic cargoes in aqueous solution during microgel formation. When we mixed PPSU nanogels with indocyanine green (ICG) in PBS, we detected by fluorescence measurements the efficient (~99%) and rapid ( 90% loading efficiencies for all these mentioned types of cargoes (Supplementary Fig. S4), including MPLA (non-ionic lipid amphiphile), CL429 (hydrophobic small molecule), CpG (hydrophilic nucleic acid), LPS (ionic lipid amphiphile), and OVA (protein). Because vaccine responses can be modulated and customized via the simultaneous delivery of multiple synergistic adjuvants22, we further demonstrated the ability of PPSU hydrogels to effectively co-load the antigen with all four adjuvants, showing that the resulting formulations can be easily controlled by the input ratios (Fig. 2f). Of note, these formulations mimic the full range of immunostimulation of an attenuated bacterium23,24,25.In situ formation of PPSU tertiary hydrogels upon administrationBecause the surfaces of the microgels were not exposed to excess protein for passivation17, we hypothesized that surface-accessible domains would be available for subsequent tissue attachment upon subcutaneous injection. This was confirmed by a model experiment in which we dropped the suspension of PPSU microgels onto an excised skin and observed an immediate immobilization of the microgels (Fig. 3a and Supplementary Fig. S5). We then subcutaneously injected PPSU microgels and extracted the in situ formed tertiary hydrogels 30 min after administration. Frequency-dependent oscillatory experiments conducted in a linear viscoelastic regime revealed that the gels exhibited a storage modulus comparable to the local biological tissues (Fig. 3b)26,27. The consistently higher maintenance of the storage modulus (G’) over the loss modulus (G”) across the evaluated frequency range indicated the solid-like properties necessary for robust hydrogel formation, thereby confirming a process of in situ gelation upon subcutaneous administration (Fig. 3c).Fig. 3: Characterization of tissue-bound hydrogels that formed following subcutaneous injection of PPSU microgels.a Dropping PPSU microgels (colored by ICG) onto an excised skin mimics the process of subcutaneous injection, anchoring the microgels onto tissues. b Frequency-dependent oscillatory experiments conducted in a linear viscoelastic regime. c The storage and loss moduli across the evaluated frequency range. (b-c) Rheological measurements were performed on excised samples 30 min post injection, demonstrating the in-situ formation of tertiary hydrogels. d SEM images of excised gels on day 0 (30 min post injection) and 5 days post injection. The SEM image of collagen fibers is included for comparison (ctr = control). e EDS elemental mapping of Fe3+-loaded hydrogels 30 min post injection. d, e All experiments were independently repeated three times.Full size imageWe tracked by scanning electron microscopy (SEM) imaging the in situ gelation of PPSU microgels within tissues after the injection (Fig. 3d and Supplementary Fig. S6). The results showed that the microgels adhered to the surface of collagen fibers 30 min post administration. Adhesion of the microgels towards the collagen fibers was also verified at day 5 post injection (Fig. 3d). Fusion between PPSU and other surrounding tissues was also found (Supplementary Fig. S7). The process of in situ gelation was accompanied by a decrease in the gel volume (Supplementary Fig. S8). As a further demonstration, we loaded the microgels with Fe3+ for energy-dispersive X-ray spectroscopy (EDS) elemental mapping (Supplementary Fig. S9) and confirmed SEM-EDS colocalization of Fe3+ that differentiated the tertiary PPSU hydrogels from collagen fibers (Fig. 3e).Having obtained evidence that the tertiary hydrogels form upon subcutaneous injection, we investigated the reversibility of the hierarchical structure and ruled out the possibility of releasing nanogels from the tertiary hydrogels (Supplementary Fig. S10). We then employed the tertiary hydrogels as depots for the sustained release of diverse cargoes. Real-time whole-body imaging showed that the administration of ICG-adsorbed microgels achieved prolonged release of ICG (Fig. 4a and Supplementary Fig. S11). Given the possibility that encapsulated cargoes could have slower release rates than their adsorbed counterparts, we proceeded to investigate the effect of our drug loading strategy on in vivo release kinetics. Förster resonance energy transfer (FRET)28 was applied as a tool in the comparison study to exclude the interference of concentration-induced fluorescence quenching. FRET imaging of the excised skins was used to monitor the in vivo release of a FRET pair consisting of encapsulated Rhodamine 6 G (Rh6G) and adsorbed Rh101. With the fluorescence of the encapsulated Rh6G nearly completely quenched by the adsorption of Rh101 (Fig. 4b and Supplementary Fig. S12), a burst release of Rh101 at 1 day post injection was revealed by the recovery of Rh6G fluorescence (Fig. 4c). Sustained release of the two dyes began on day 3, achieved through the liquefication of the tertiary hydrogels. These results demonstrated an erosion-controlled release process for the encapsulated dyes, whereas the release of the adsorbed dyes could be first triggered by protein replacement, a diffusion-controlled process that depends on the accessibility of nanogel surfaces. Not all the nanogel surfaces are accessible to proteins during the formation of the secondary hydrogels in PBS. This was confirmed by the inefficient FRET shown in Fig. 4d, where the adsorption of excess TNF-α only partially decreased the fluorescence intensity of the encapsulated MFT. The subsequent in vivo release demonstrated an erosion-controlled release process of up to 92 days for both the cargoes (Fig. 4e and Supplementary Fig. S13), regardless of the loading strategies. Note that the release of the adsorbed TNF-α was slower than that of the adsorbed Rh101 (Supplementary Fig. S14), likely due to the increased hydrophobic interactions between PPSU nanogels and higher molecular weight proteins relative to small molecule dyes.Fig. 4: In vivo release of model cargoes loaded within PPSU microgels.a Real-time whole-body imaging of adsorbed ICG after SC injection showed prolonged release of hydrogel cargo. The data are presented as mean values with the statistical significance determined by a two-sample t-test (n = 3 mice per group). b Efficient FRET demonstrates full accessibility (within ~10 nm) of the microgel surfaces by small molecule fluorophore Rh101. c The recovery of Rh6G fluorescence at day 1 indicated quick desorption/replacement of adsorbed Rh101 upon SC injection. λem/λex = 465/640 nm for Rh101, λem/λex = 465/560 nm for Rh6G. The data are presented as mean values with the statistical significance determined by a two-sample t-test (n = 5 mice per group). d Not all of the microgel surfaces are accessible to proteins, as suggested by the inefficient FRET from encapsulated MFT in response to incubation with TNF-α. e Sustained release was achieved for the encapsulated MFT and adsorbed proteins. λem/λex = 465/600 nm for TNF-α, λem/λex = 465/520 nm for MFT. The data are presented as mean values with n = 3 mice per group.Full size imageEmploying PPSU hydrogels for multi-component vaccine optimizationThe sustained release system investigated here takes advantage of three levels of physical binding (sulfone binding, then protein adsorption, and ultimately tissue binding), leading to hierarchical architectures from a simple homopolymer structure. It would allow us to design multi-component vaccines that typically require a tedious or impractical process for optimizing the concentration and ratio for each adjuvant. To this end, we used RAW-blue macrophages to assess retention of adjuvant bioactivity upon loading by PPSU at a low, mild, and high concentration, based on their published effective ranges (Supplementary Fig. S15)29,30,31,32,33. The results showed that PPSU alone had no effect on RAW-blue macrophages, whereas the adjuvant-loaded nanogels (microgel precursors) induced the activation of NF-kB. They have either equally effective or even enhanced activities compared to the free form adjuvants of LPS and CL429, respectively.To determine the immunogenicity of the 5-component formulation (OVA-bridged PPSU microgels with four adjuvants, termed PPSU-4Ad) on antigen-presenting cells (APCs), PPSU-4Ad, CpG-loaded PPSU (PPSU-CpG), and OVA-loaded PPSU (PPSU-OVA) were incubated with primary C57BL/6 bone marrow-derived dendritic cells (BMDCs). Of note, for all OVA-based immunizations, each formulation contained an identical amount of OVA antigen. After 24 h, activation marker expression of the BMDCs, including MHC-II, CD40, CD80, and CD86, were assessed by flow cytometry. Low immunogenicity for PPSU-OVA was confirmed, as indicated by the same level of expression for all tested markers as that of the PBS group. But PPSU-4Ad strongly activated CD40, CD80 and CD86 expression on BMDCs, whereas the single adjuvant-loaded group (PPSU-CpG) only marginally activated CD40 and CD86 (Supplementary Fig. S16). Moreover, PPSU-4Ad stimulated MHC-II expression. These results verify that simultaneous delivery of antigen and adjuvants by PPSU-4Ad to APCs can induce a strong immunostimulatory response.C57BL/6 mice were then immunized subcutaneously with either PPSU-CpG or PPSU-4Ad (Fig. 5a, LPS was not included as an adjuvant for cytokine and antibody experiments). Blank PPSU and PBS groups were included as controls. At 24 h post-injection, PPSU-4Ad induced both Th1 and Th2 serum cytokines (Fig. 5b). The elevated Th2 cytokine levels indicate a tendency for PPSU-4Ad to favor humoral responses34. Seven days post-immunization, mice immunized with PPSU-4Ad showed a significant increase in anti-OVA IgM antibodies compared to those immunized with PPSU-CpG or the blank PPSU and PBS controls (Fig. 5c). Additionally, C57BL/6 mice were subcutaneously injected with either PPSU-LPS or PPSU-4Ad. The hydrogels were retrieved 30 min post-injection and incubated in transwell inserts with RAW-Blue macrophages to assess the sustained release of adjuvants. NF-κB/AP-1 activation remained lower for both hydrogels compared to nanogels and bolus formulations on days 1 and 3, with fresh medium replaced on the day of observation (Supplementary Figs. S14 and S17). However, activation was still detectable with no significant change on day 3 (Supplementary Fig. S17).Fig. 5: Validation of multi-adjuvant/antigen-loaded PPSU hydrogels as a subunit vaccine.a Schematic schedule of in vivo blood analysis and adoptive T cell transfer. (created with BioRender.com). b Cytokine fold changes 24 h after subcutaneous injection of the multi-antigen/adjuvant (LPS excluded) loaded PPSU relative to PBS control in mice. n = 3 mice per group (mean ± s.d). c Blood anti-OVA IgM antibody levels 7 days post subcutaneous administration of the multi-antigen/adjuvant (LPS excluded) loaded PPSU relative to single adjuvant, blank PPSU and PBS controls. n = 3 mice per group (mean ± s.d), with *p (0.0355)