A Mannosylated Polymer with Endosomal Release Properties for Peptide Antigen Delivery

Abstract

Peptide cancer vaccines have had limited clinical success despite their safety, characterization and production advantages. We hypothesize that the poor immunogenicity of peptides can be surmounted by delivery vehicles that overcome the systemic, cellular and intracellular drug delivery barriers faced by peptides. Here, we introduce Man-VIPER, a self-assembling (40-50nm micelles), pH-sensitive, mannosylated polymeric peptide delivery platform that targets dendritic cells in the lymph nodes, encapsulates peptide antigens at physiological pH, and facilitates endosomal release of antigens at acidic endosomal pH through a conjugated membranolytic peptide melittin. We used d-melittin to improve the safety profile of the formulation without compromising the lytic properties. We evaluated polymers with both releasable (Man-VIPER-R) or non-releasable (Man-VIPER-NR) d-melittin. Both Man-VIPER polymers exhibited superior endosomolysis and antigen cross-presentation compared to non-membranolytic d-melittin-free analogues (Man-AP) in vitro. In vivo, Man-VIPER polymers demonstrated an adjuvanting effect, induced the proliferation of antigen-specific cytotoxic T cells and helper T cells compared to free peptides and Man-AP. Remarkably, antigen delivery with Man-VIPER-NR generated significantly more antigen-specific cytotoxic T cells than Man-VIPER-R in vivo. As our candidate for a therapeutic vaccine, Man-VIPER-NR exerted superior efficacy in a B16F10-OVA tumor model. These results highlight Man-VIPER-NR as a safe and powerful peptide cancer vaccine platform for cancer immunotherapy.

Graphical Abstract

2. Introduction

Cancer subunit vaccines utilize neoantigens to train the immune system to recognize and attack tumor cells¹. Among multiple tumor antigen sources, peptide antigens have garnered significant attention for multiple reasons: improved safety compared to whole antigens, facile incorporation of multiple epitopes, low cost, manufacturing ease, shelf stability and well-defined structures¹. Peptide antigen sequences can be identified empirically by mass spectrometry and sequencing or computationally designed and validated². Despite these salient features, peptide antigen vaccines have only met moderate clinical success, and even less so in the context of cancer vaccines due to peptides’ poor immunogenicity³.

Cytotoxic T-cells (CD8⁺) T-cells are most prominently credited for tumor-killing responses. As CD8⁺ T cell priming requires antigen presentation through the MHC class I (MHC-I) pathway (called “cross-presentation”), providing access to the MHC-I presentation is important to cancer peptide vaccines’ ability to induce tumor clearance⁴. There are three major barriers to MHC-I presentation for peptide subunit vaccine formulations – systemic, cellular and intracellular. Systemically, peptides need to preferentially localize into lymph nodes (LNs) where T-cell priming takes place⁵, while facing risk of degradation due to peptidases in circulation⁶. Cellularly, peptides need to be internalized by antigen-presenting cells (APCs) in order to be properly processed for antigen presentation. Intracellularly, peptides need to access the cytosolic compartment in order to access the MHC-I presentation pathway⁷. All of these pose significant barriers to generating efficient antitumor responses that peptide formulations alone cannot surmount.

Various drug delivery platforms have been proposed to surmount these barriers with impressive results. Antigens can either be conjugated to targeting moieties or formulated into sub-50 nm nanoparticles that can traffic to LNs from the interstitium^8–10. For instance, soluble antigen-mannose conjugates¹¹ and self-assembling antigen micelles¹² or antigen-protein complexes¹³ have both been demonstrated to preferentially accumulate in the LNs. Interestingly, the same technology could be leveraged to overcome the cellular barrier – APC-targeting moieties and nanoparticle formulation have both been shown to promote endocytosis^14,15. These can be in turn incorporated into controlled-release platforms such as hydrogels to impart additional benefits¹⁶. To surmount the intracellular barriers, strategies such as ‘proton-sponge’ cationic materials¹⁷, pH-responsive solubility-switching anionic polymers and combinations thereof^18,19, amphiphilic polymers^20,21 or membrane-perforating biomolecules²² have been employed to facilitate endosomal release. Successful integration of technologies to overcome these three barriers have resulted in preclinically efficacious formulations²³.

We previously reported two formulations for cancer peptide subunit vaccine applications. Our first report utilizes our well-developed Virus-Inspired Polymers for Endosomal Release (VIPER) platform²⁴. The VIPER platform is a self-assembling pH-responsive cationic amphiphilic polymer conjugated to the membranolytic peptide melittin²⁵. Melittin is sequestered in the self-assembled structure at physiological conditions, but is bioavailable in the acidic endosome where it can lyse the endosomal membrane to release disulfide-conjugated peptide cargo into the cytosol for MHC-I presentation. We coformulated a cationic VIPER conjugated to antigen peptide with the nucleic acid adjuvant poly(I:C) to form polyplexes²⁶. This formulation relied on non-specific, charge-mediated uptake to local dendritic cells after subcutaneous injection. The VIPER formulation improved antigen cross-presentation when tested in vitro and increased generation of antigen-specific cytotoxic T cells, but had modest effects on tumor growth rate and overall survival, possibly due to poor LN localization. In our second report, we focused on targeting dendritic cells (DCs) in the draining LN and synthesized a diblock copolymer that self-assembles into ~30 nm micelles with a poly(mannose) corona for targeting mannose receptor (CD206) expressed on DCs²⁷. We demonstrated that the targeted micelles were effective in LN localization and induction of DC maturation. In summary, our first formulation surmounted the intracellular and cellular barriers, but not the systemic barrier while our second surmounted the systemic and cellular barrier, but leaves the intracellular barrier unchallenged.

In this work, we addressed the aforementioned shortcomings with a mannosylated VIPER platform that self-assembles into sub-60nm micelles that allows for efficient LN draining and DC targeting. Upon endocytosis and acidification, the VIPER design with the pH-responsive, melittin-conjugated inner block induces endosomal rupture and antigen release into the cytosol and thus the MHC-I antigen presentation pathway in addition to MHC-II presentation from intact matured endosomes (Figure 1). We also used D-amino acids instead of L-amino acids to synthesize the melittin peptide (d-Melittin) to prevent antibody generation against the VIPER carrier²⁸. In addition, we explored stable conjugation of melittin to the polymeric backbone using a pentafluorobenzyl-thiol reaction to form a thioether bond to improve the endosomal release properties of VIPER²⁹ compared to our previously-employed disulfide conjugation strategy. Our previous work established that polymer-conjugated melittin is considerably more membranolytic²⁵, which can translate to improved endosomal release of antigen and MHC-I antigen presentation. Therefore, in this work, we evaluated a form of VIPER that retains melittin as a polymer-bound species within the endosome (VIPER-NR) as well as a form of VIPER for which melittin can be reduced and released in as a monomeric peptide (VIPER-R). The mannosylated VIPER formulation induced efficient endosomal release and MHC-I antigen presentation in vitro, and potent T-cell responses against a model antigen in vivo. Ultimately, this platform significantly slowed tumor growth and prolonged survival in an aggressive melanoma model, demonstrating the utility of our platform as a cancer peptide subunit vaccine formulation.

Figure 1.

Schematic illustration of Man-VIPER delivery systems. The Man-VIPER formulations consist of releasable disulfide conjugated OVA antigens, and either releasable disulfide conjugated membranolytic melittin (Man-VIPER-R) or non-releasable pentafluorobenzyl conjugated melittin (Man-VIPER-NR). The formulations self-assemble into well-defined micelles, which are preferentially endocytosed via the mannosylated hydrophilic segment. The micelles disassemble upon endosomal maturation, and either results in endosomal disruption and cross-presentation of MHC-1 epitopes, or lysosomal maturation and presentation of MHC-2 epitopes, to their own respective T-cell subsets.

3. Materials and methods

Materials

RAFT chain transfer agent 4-((((2-Carboxyethyl)thio)carbonothioyl)thio)-4-cyanopentanoic acid (CCC) was purchased from Boron Molecular. Azobisisobutyronitrile (AIBN), 4,4′-azobis(4-cyanovaleric acid) (ABCVA), cysteamine hydrochloride, piperidine, triisopropylsilane (TIPS) and ethanedithiol (EDT) were purchased from Sigma-Aldrich. Mannose ethyl methacrylate (ManEMA) was purchased from Omm Scientific. Diisopropylcarbodiimide (DIC) and ethyl cyanohydroxyiminoacetate (Oxyma) were purchased from Chem-Impex. Dimethoxybenzene (DMB), trifluoroacetic acid (TFA), acetic acid (AcOH), acetonitrile (ACN) and Ellman’s reagent were purchased from Fisher Scientific. Dithiothreitol (DTT) was purchased from Enzo Life Sciences. All were used as received. Diisopropyl ethyl methacrylate (DIPAMA) was purchased from Sigma-Aldrich. Pyridyl disulfide ethyl methacrylate (PDSEMA, also known as 2-(2-pyridinyldithio)ethyl methacrylate) and pentafluorobenzyl methacrylate (PFBzMA) was purchased from Tokyo Chemical Industries Ltd. These aforementioned monomers were purified by passing through a basic alumina column to remove any existing inhibitor prior to polymerization. Anhydrous N-methyl-2-pyrrolidone (NMP, 99.5%), and dimethylsulfoxide (DMSO, > 99%) were purchased from Sigma-Aldrich and stored with activated molecular sieves. Protected amino acids were purchased from Novabiochem and used as received. Buffers were prepared in-house using endotoxin-free water and salts purchased from Fisher Scientific.

Polymer synthesis

All polymers were synthesized by RAFT polymerization. Briefly, the mannose hydrophilic block (ManCTA) was synthesized by RAFT polymerization of ManEMA. ManEMA was combined with CCC, ABCVA (CCC:ManEMA = 1:44.4, CCC:ABCVA = 1:20) in dry DMSO (2.5 wt% monomer), purged with argon for 30 minutes, and reacted at 70°C for 3h with vigorous stirring. ManCTA was purified by 3x precipitation in 1:1 (v/v) acetone and diethyl ether, dissolving in DMF in-between, and dried in vacuo for 2 days. To synthesize CP-R and CP-NR, ManCTA was combined with AIBN (ManCTA:AIBN = 1:10), DIPAMA and either PDSEMA for CP-R (ManCTA:DIPAMA:PDSEMA = 1:8:45) or PFBzMA for CP-NR (ManCTA:DIPAMA:PFBzMA = 1:5:45) in NMP (20 wt% monomer), purged with argon for 30 minutes, and reacted at 70°C in NMP solvent (20 wt% monomer) for 18h with vigorous stirring. CP-R was purified through serial dialysis in NMP for 1 day and deionized (DI) water for 3 days. CP-NR was dialyzed in NMP for 1 day, and 30x AIBN used in polymerization was added directly to the solution and reacted at 70°C overnight to remove CTA head-groups³⁰, which can interfere with the conjugation reaction from aminolyzed thiols on the head-groups (data not shown). CP-NR is then purified in a similar manner to CP-R. All polymers were lyophilized post-dialysis and stored at −20°C until used.

Polymer Characterization

All ¹H NMR spectra were recorded on a Bruker AV 300 (Bruker Corporation, Billerica, MA) nuclear magnetic resonance (NMR) instrument in deuterated DMSO (DMSO-d6). Polymer molecular weights (Mw, Mn) and polydispersity index (Đ = Mw/Mn) were determined by gel permeation chromatography (GPC). The system consisted of an Agilent 1260 HPLC stack running heated LiBr-supplemented (0.1% w/v) DMF mobile phase at a flow rate of 1 mL min⁻¹ through a semi-prep three-column setup (Phenogel-TSKgel). Samples (5 mg/mL) were filtered through a 0.2 μm PTFE filter before analysis. We observed a difference between the theoretical Mn calculated by NMR and the Mn calculated by GPC, as well as a lower-than-expected shift in retention time for the block copolymers. This may be caused the presence of bimolecular termination products in the mannose macro-CTA owing to high conversion masking the retention time shift, as well as differences in the molecular weight-hydrodynamic radii relationship between mannose-CTA and CP polymers. The size distribution profiles of micelles were recorded on a dynamic light scattering (DLS) system (DynaPro NanoStar, Wyatt technology) at a micelle concentration of 1 mg/mL.

Peptide synthesis and conjugation

MHC-1 OVA epitope peptide (CSSSIINFEKL, or OVA-1), MHC-2 OVA epitope peptide (CSSISQAVHAAHAEINEAG, or OVA-2) and d-Melittin (GIGAVLKVLTTGLPALISWIKRKRQQC, or d-Mel) were synthesized using a Liberty Blue (CEM) microwave peptide synthesizer on Rink Amide resin (Millipore) using Fmoc-piperidine deprotection and DIC-Oxyma activation chemistry before cleavage and deprotection (5% DMB, 2.5% TIPS, 2.5% EDT, 0.5% H2O in TFA), and purification with reverse-phase HPLC (TFA-supplemented H2O:ACN).

AP-1, AP-2 and VIPER-R were synthesized by conjugating OVA-1, OVA-2 and d-Melittin to CP-R using disulfide exchange chemistry. Briefly, CP-R and peptides were dissolved in separate solutions of 2% (v/v) AcOH in DMSO at a CP-R concentration equivalent to 12.5 mM PDSEMA and 14-19 mM peptide. The two solutions were combined in equal volumes and allowed to react overnight at room temperature. The reaction solution was then dialyzed against DMSO (SpectraPor 6-8 kDa MWCO) for 2 days, against MilliQ water for 3 days to remove free peptides, and lyophilized over the weekend. Peptide loading was assessed by reacting conjugated polymers with DTT for 30 minutes at 37°C and comparing 2-mercaptopyridine release (Abs_max = 347 nm) compared to unconjugated polymers as previously reported by our group²⁵.

VIPER-NR was synthesized by conjugating d-Melittin to CP-NR using a recently-described pentafluorobenzyl-thiol (PFB-thiol) reaction²⁹. Briefly, d-Melittin was dissolved with CP-NR at a CP-NR concentration equivalent to 30 mM PFB and 10 mM peptide. The concentrations were adjusted to achieve a theoretical peptide loading of 19% by weight of the final conjugate. DBU (2 mmol) was added in excess to counteract acidic counterions present in d-Melittin and the reaction allowed to proceed overnight at room temperature. The reaction was quenched by addition of TFA, and dialyzed as above to remove free peptides. Peptide loading was assessed by reacting conjugated and unconjugated polymers with a 2-fold excess of cysteamine hydrochloride (50 mM in dry DMSO) in a similar manner to conjugation, and comparing cysteamine depletion using Ellman’s reagent to determine extent of PFB conjugation. ¹⁹F-NMR was also used to determine PFB substitution extent, but was disfavored due to low solubility of polymers in NMR solvents coupled with low mass sensitivity of ¹⁹F-NMR.

Micelle formulation

We utilized our previously-described pH-switch method to formulate our micelles with slight modifications²⁷. Briefly, individual polymer conjugates were dissolved in endotoxin-free water. The polymer solutions were combined at equal volumes. HCl (0.6% to 2.5% of 1 N solution) was added to the mixture until a clear solution was obtained. The solution was probe-sonicated for 30 seconds to ensure complete dissolution. A solution of 0.2 M pH 5.0 monobasic sodium phosphate was added, followed by 0.2 M pH 9.0 dibasic sodium phosphate to obtain the final pH in the range of 6.6-7.4. The formulations were prepared at ambient conditions. The critical micellar concentration (CMC) and transition point of the micelles were determined using the Nile Red assay³¹. Briefly, micelles with different concentrations or pH were mixed with 1μM Nile Red. Fluorescence intensity was measured using a plate reader (ex. 556 nm, em. 625 nm).

Hemolysis assay for endosomal rupture quantification

The polymers’ ability to disrupt endosomal membranes in a pH-dependent manner was evaluated using a hemolysis assay as previously described³². Briefly, red blood cells (RBCs) isolated from de-identified human whole blood samples (Bloodworks Northeast) were pelleted at 500g and washed three times with 150 mM NaCl and once with pH 7.4 PBS. The washed RBCs were resuspended in PBS of various pHs (pH 7.4, 6.2, and 5.6). RBCs were then treated with polymers at various concentrations of melittin or equivalent melittin-free polymer concentrations (polymer:RBC 10:190 v/v) in V-bottom well-plates and incubated at 37°C for 1 h. Cells were then pelleted and the supernatant was transferred to flat-bottom well-plates. The amount of hemoglobin leakage into the supernatant was quantified via absorbance spectroscopy (Tecan, λ = 575 nm) and normalized to 100% hemolysis with 20% Triton X-100 and 0% hemolysis with PBS at its respective pHs.

Cell lines and animals

The B16F10-OVA cell line (gift of Prof. Amanda Lund) was cultured in DMEM supplemented with 10% FBS and 1% P/S. The DC2.4-Gal8-GFP and Hela-Gal8-GFP cell lines were generated as previously described²⁶. DC2.4-Gal8-GFP cells were cultured in RPMI supplemented with 10% heat-inactivated FBS, 1% P/S, 1x nonessential amino acids, 10 mM HEPES buffer and 55 μM β-mercaptoethanol. Hela-Gal8-GFP cells were cultured in DMEM supplemented with 10% FBS and 1% P/S. The B3Z cell line (gift of Prof. Nilabh Shastri) was cultured in RPMI supplemented with 10% heat-inactivated FBS, 1% P/S, 1 mM sodium pyruvate and 50 μM β-mercaptoethanol. Cells were incubated at 37°C and 5% CO₂. Female C57BL/6 mice aged 6 to 8 weeks were purchased from Charles River Laboratories. All animal studies were approved by the University of Washington Institutional Animal Care and Use Committee (IACUC).

Gal8 endosomal disruption assay

DC2.4-Gal8-GFP and Hela-Gal8-GFP cells were cultured in TC-treated 96-well imaging plates overnight. Cells were incubated with polymers containing 2 μg/ml CSSSIINFEKL, 2 μg/ml CSSISQAVHAAHAEINEAGR, 38 μg/ml d-Melittin peptide for 18 hr. Media was replaced with Fluorobrite DMEM supplemented with 10% FBS, 25 mM HEPES. Hoechst 33342 was used to stain the nucleus and propidium iodide was used to stain the dead cells. Images were obtained using a Leica SP8X confocal microscope using a 20x objective.

DC2.4 Cross-Presentation and Viability Assays

The DC2.4 cell line was seeded into U-bottom TC-treated 96 well plates at 20,000 cells per well and incubated at 37°C and 5% CO2 overnight. To assess cross-presentation, cells were treated with the ovalbumin protein or polymers containing 2 μg/ml CSSSIINFEKL and 2 μg/ml CSSISQAVHAAHAEINEAGR with different amounts of d-Melittin peptide. After 4 hours, cells were washed twice with PBS and co-cultured with 100,000 B3Z cells for an additional 18-20 hours in the incubator. After co-culture, cells were pelleted by centrifugation and resuspended in 150 μL CPRG lysis buffer (0.15 mM chlorophenol red-β-D-galactopyranoside + 0.1% Triton-X100 + 9 mM MgCl2 + 100 μM β-mercaptoethanol + PBS). Cells were incubated at 37°C in the dark for 90 minutes. The supernatant was transferred to a flat-bottom 96-well plate and the absorbance was measured at 570 nm (reference 650 nm) using a plate reader. Cell viability was analyzed using the MTS/PMS assay 24 hours post-treatment (Promega).

In vivo dendritic cell activation

Female C57BL/6 mice were injected subcutaneously at the tail base with PBS or with 15 μg CSSSIINFEKL and 15 μg CSSISQAVHAAHAEINEAGR antigen peptides either in free peptide form or in Man-VIPER-R or Man-VIPER-NR formulations. Polymer formulations also contained 286 μg d-Melittin peptide (1:5 molar ratio of antigen and melittin). Twenty-four hours-post immunization, inguinal lymph nodes were harvested and digested in 50 U/mL type IV collagenase + 100 U/mL DNase I. Single-cell suspension was obtained by mashing the tissue through a 70 μm cell strainer. Cells were stained for viability with Zombie NIR viability kit, blocked with anti-CD16/32, and then stained with anti-CD86-BV510, anti-CD40-FITC, anti-CD11c-APC and anti-SIINFEKL-PE. An Attune NxT flow cytometer was used to analyze the data.

In vivo T cell responses

Female C57BL/6 mice were immunized with PBS, free peptides, AP, Man-VIPER-R and Man-VIPER-NR on day 0 and day 14. Spleens were harvested on day 21. Tissues were passed through 70 μm cell strainers and treated with ACK lysis buffer. Splenocytes were resuspended in IMDM and plated in separate 96-well plates for tetramer staining and intracellular cytokine staining (ICCS). For the tetramer staining, splenocytes were stained with Zombie NIR for viability and then stained with PE-labeled H-2Kb/SIINFEKL tetramer (NIH Tetramer Core). Cells were blocked with anti-CD16/32 and stained with anti-CD3-eFluor450 and anti-CD8-FITC. For ICCS, splenocytes were either restimulated with 5 μg/ml SIINFEKL peptide or 5 mg/ml OVA protein. 3 hr later, Protein Transport Inhibitor Cocktail (PTIC) was added and cells were incubated for another 3 hr. After incubation, cells were stained with Zombie NIR for viability, blocked with anti-CD16/32 and stained with either anti-CD3-eFluor450 + anti-CD8-FITC for CD8⁺ T cells or anti-CD3-eFluor450 + anti-CD4-FITC for CD4⁺ T cells. Cells were then treated with the BD Cytofix/Cytoperm Fixation/Permeabilization Kit and stained with either anti-TNFα-PE and anti-IFNγ-APC for CD8⁺ T cells and anti-IL-2-PE and anti-IFNγ-APC for CD4⁺ T cells. Data was collected using the Attune NxT flow cytometer and analyzed using Flowjo.

Tumor studies

Female C57BL/6 mice (N=8) were inoculated with 1.5 × 10⁵ B16F10-OVA cells in the right-hind flank on day 0. On days 3, 10, and 19, mice were immunized with PBS, free peptides containing 15 μg CSSSIINFEKL and 15 μg CSSISQAVHAAHAEINEAGR, AP or MAN-VIPER-NR containing the same amount of peptides. On days 4, 11, and 20, mice were injected with 100 μg of anti PD-1 ICB (Clone 29F.1A12, BioXCell) via intraperitoneal injection. Tumor length and width were then measured every other day using a vernier caliper, and volume was subsequently calculated using the formula: Volume = (Width² × Length) ÷ 2. Weight was also measured every other day. Mice were euthanized if total tumor volume exceeded 1500 mm³, visible discharge was observed on ulcers of the tumors, or the body weight loss exceeded 20%.

Statistical analysis

Statistical analysis was performed using the Graphpad Prism software. One-way ANOVA with post-hoc Tukey HSD test was used to compare more than two groups. Log-rank test was used for survival analysis.

4. Results and discussion

Polymer synthesis and characterization

Polymers were synthesized by RAFT polymerization in a two-step process. Mannose ethyl methacrylate (ManEMA) was polymerized first to make a macro-RAFT chain transfer agent (ManCTA), followed by copolymerization of diisopropylamino ethyl methacrylate (DIPAMA) with either pyridyl disulfide ethyl methacrylate (PDSEMA) or pentafluorobenzyl methacrylate (PFBzMA), to form well-defined diblock copolymers with low dispersity (Table S1, Figure S1–S2). The polymers without any conjugated peptides are denoted as CP, or control polymer, in accordance with previous nomenclature. To differentiate between polymers synthesized with the PDSEMA and PFBzMA monomers, which result in a labile disulfide or a nonlabile thioether bond upon peptide conjugation respectively, we denote polymers as CP-R (CP-Releasable) and CP-NR (CP-Non-Releasable).

We selected the model antigen OVA and the lytic peptide d-Melittin for conjugation. OVA is highly immunogenic and its MHC-I and MHC-II epitope sequences are well-defined, so we chose these peptide sequences with an N-terminal Cys-Ser-Ser sequence for conjugation (CSSIINFEKL and CSSISQAVHAAHAEINEAG, respectively). Melittin was chosen for its effective membrane-lytic capability and used in its d-amino acid form, which retains lytic activity without the adverse immunogenicity of L-melittin. ²⁸. Both antigens were conjugated through disulfide exchange with PDSEMA on CP-R, while D-melittin was conjugated through either PDSEMA on CP-R or a base-catalyzed thiol-PFBzMA substitution reaction on CP-NR²⁹. We decided to explore different conjugation strategies solely for d-Melittin because of its direct role in endosomal escape induction in VIPER, as well as the need to retain releasability for conjugated antigen. All peptides were conjugated with >50% efficiency as measured by the amount of conjugated peptides. The antigen-functionalized polymers without d-Melittin are denoted AP-1 and AP-2 for “Antigen Polymer”, demarcated by the MHC epitope identity. The melittin-functionalized polymers are denoted MP-R and MP-NR for “Melittin Polymer” in accordance with the CP-R and CP-NR nomenclature. The peptide conjugation results are displayed in Table 1. ¹⁹F-NMR was also performed to confirm conjugation for MP-NR (Figure S3).

Table 1.

Characterization of peptide conjugation in constituents of Man-VIPER formulations

Polymer name	Conjugated peptide	Approx. peptide/polymera	Peptide wt%a
Man-AP-1	OVA MHC-I epitope	2.9	14.1
Man-AP-2	OVA MHC-II epitope	3.0	22.7
Man-MP-R	D-melittin	2.2	22.3
Man-MP-NR	D-melittin	1.8	19.4

^{a -}Calculated by determining residual conjugating groups - quantifying PDS release with DTT reduction (AP-1, AP-2, MP-R) or cysteamine depletion with Ellman’s assay (MP-NR)

The full formulation (visualized in Figure 1) of AP-1 coformulated with AP-2 and either MP-R (Man-VIPER-R) or MP-NR (Man-VIPER-NR) self-assembles into micelles with narrowly-distributed sizes at physiological pH (Figure 2a). Melittin-containing formulations (i.e. Man-VIPER-R and Man-VIPER-NR) had larger micelle sizes (51.6 nm and 40 nm in diameter, respectively) compared to non-melittin (i.e. Man-AP, 27.6 nm), likely due to additional peptide in the micelle core. The size difference is consistent with our previous work showing that formulations with higher peptide loading exhibit larger micelle sizes²⁷. Man-VIPER-NR formed smaller micelle sizes than Man-VIPER-R, possibly due to pi-pi stacking interactions from the pentafluorobenzyl linkers within the micelle core, which has been observed with other polymeric micellar systems in the literature³³. All formulations exhibited pH-dependent micelle disassembly with a transition point around pH 6.2 as determined by a Nile Red assay (Figure 2b). Interestingly, Man-VIPER-NR exhibits disassembly over a broader pH range compared to Man-VIPER-NR or Man-AP, also possibly due to pi-pi stacking interactions within the reversibly hydrophobic polymer blocks. To determine whether or not Man-VIPER-NR might have a lower critical micelle concentration (CMC) than Man-VIPER-R due to aforementioned additional pi-pi interactions within the core, we measured CMC by evaluating micellization at various polymer concentrations using the Nile Red Assay but observed no difference between the two formulations (Figure S4). These results demonstrate that Man-VIPER-R and Man-VIPER-NR exhibit desired self-assembly and pH-responsive properties for antigen delivery.

Figure 2.

Polymer characterization. (a) Size distribution of Man-AP, Man-VIPER-R and Man-VIPER-NR characterized by dynamic light scattering (d_mean = 27.6nm, 51.6nm, 40nm respectively). (b) pH-dependent micellization of Man-AP, Man-VIPER-R and Man-VIPER-NR (transition pH = 6.4, 6.3, 6.1, respectively).

Man-VIPER-R and Man-VIPER-NR enhances antigen cross-presentation through endosomolytic activity

The lytic properties of Man-VIPER-R and Man-VIPER-NR were examined using a hemolysis assay. The polymers were incubated with human red blood cells (RBCs) at various concentrations at pH 7.4, 6.2, and 5.6, representing the pH of the extracellular environment, early endosomes, and late endosomes/lysosomes (Figure 3a). Both Man-VIPER-R and Man-VIPER-NR showed no lytic activity pH 7.4, confirming full encapsulation and inactivation of D-melittin at physiological pH^24–26. Man-VIPER-R showed pH-dependent lytic activity at acidic endosomal pHs (HC₅₀ = 0.090 μg/ml at pH 6.2, 0.068 μg/ml at pH 5.6), while Man-VIPER-NR showed similar lytic activity at pH 5.6 and pH 6.2 (HC₅₀ = 0.045 μg/ml at pH 6.2, HC₅₀ = 0.043 μg/ml at pH 5.6). Man-AP showed negligible hemolysis compared to Man-VIPER-R and Man-VIPER-NR at pH 5.6, emphasizing the role of D-melittin in hemolysis induction.

Figure 3.

In vitro characterization of Man-AP and Man-VIPER. (a) Hemolytic activity of the micelles against human RBCs at pH 5.6, pH 6.2 and pH 7.4. (b-d) Endosomal disruption in DC2.4-Gal8-GFP cells. (b) Representative images of DC2.4-Gal8 cells. Hoechst 33342 was used to stain the nucleus in blue. Gal8-GFP in green was dispersed in the cytosol and localized to the disrupted endosomes. Propidium iodide in red stained dead cells. (c) Percentage of cells with disrupted endosomes. (d) Percentage cell viability (e) Endosomal disruption in Hela-Gal8-GFP cells. (f) Cross-presentation of Man-AP and Man-VIPER at different antigen to melittin ratios. DC2.4 cells were treated with vaccine formulations and co-cultured with B3Z cells. Production of β-galactosidase was quantified using a CPRG assay kit by measuring the absorbance at 570 nm. Data are represented as mean ± SD. N = 3 biological replicates. Statistical analysis was performed using one-way ANOVA with post-hoc Tukey HSD test (***p ≤ 0.001, **** p ≤ 0.0001).

Next, we directly evaluated the endosomal disruption ability of Man-VIPER-R and Man-VIPER-NR using a Gal8-GFP assay developed by the Duvall group³⁴. Gal8 is a protein that is dispersed in the cytoplasm but redistributes to the inner membrane of the endosomes upon its disruption, resulting in visualizable “punctaes” of redistributed Gal8-GFP. We used the dendritic cell DC2.4 cell line that stably expresses the Gal8-GFP fusion protein (DC 2.4 Gal8-GFP) and treated the cells for 18 hours with the nanoparticles before imaging Gal8 punctae by confocal microscopy. Man-VIPER-R-treated cells exhibit more punctae (~10% of cells) compared to Man-AP and Man-VIPER-NR (Figure 3b–d). All formulations showed no cytotoxicity at the tested concentrations. We observed a similar but more striking trend with the same experiment using a HeLa-Gal8-GFP cell line (Figure 3e). Most cells (>90%) treated with Man-VIPER-R had multiple disrupted endosomes. These results suggest that Man-VIPER-R is more capable of inducing endosomal disruption than Man-VIPER-NR, which is contrary to our initial expectations given previous data showing higher hemolysis with polymer-conjugated melittin²⁵.

As antigen cross-presentation towards MHC-1 is desired, we evaluated in vitro cross-presentation using a DC2.4 and B3Z cell co-culture assay. B3Z is a T cell hybridoma that specifically recognizes the MHC I-SIINFEKL complex and produces measurable β-galactosidase upon binding³⁵. We tested Man-VIPER-R and Man-VIPER-NR with different antigen to melittin molar ratios to optimize for our in vivo studies (1:1, 1:3 and 1:5). Both Man-VIPER-R and Man-VIPER-NR demonstrated higher cross-presentation compared to Man-AP in a melittin dose-dependent manner (Figure 3f). Interestingly, no significant differences in the cross-presentation of Man-VIPER-R and Man-VIPER-NR was observed despite their different endosomal disruption properties. There are several possible explanations for this difference. For example, the Gal8 assay evaluated endosomal disruption at 18 hours post-incubation, and the two formulations may have different kinetics of peptide display and endosomal release unresolvable through the single timepoint. The cross-presentation assay evaluates display of antigen peptide on MHC-I complexes which is downstream of uptake and endosomal disruption and thus less susceptible to kinetic differences in endosomal release. The discrepancy in these two assays observed may also reflect differences in other steps in the cross-presentation process. We chose the 1:5 molar ratio of antigen to melittin for the in vivo studies due to the higher cross-presentation at this ratio.

Man-VIPER enhances DC maturation in vivo

DC activation is a fundamental step in mounting a T-cell-based immune response. We therefore assessed the effect of Man-VIPER formulations on dendritic cell (DC) activation in vivo by measuring expression of co-stimulatory molecules CD86 and CD40 in lymph node DCs³⁶. Mice were injected subcutaneously at the tail base with vaccine formulations (free peptides, Man-AP, Man-VIPER-R and Man-VIPER-NR), followed by cell isolation from the inguinal lymph nodes (ILNs) 24 hours later and analysis with flow cytometry. Both Man-VIPER-R and Man-VIPER-NR significantly increased CD86 and CD40 expression in CD11c⁺ DCs compared to Man-AP and free peptides (Figure 4, Figure S5), suggesting that Man-VIPER formulations exhibit adjuvant activity. Melittin has been shown to activate the NLRP3 inflammasome,³⁷. which in DCs triggers the secretion of interleukin-1β (IL-1β) and helps in priming CD8⁺ T cells³⁸. Our previous studies have shown that d-Melittin also exhibits similar activity³⁹, corroborating these results. DCs from animals treated with free peptides or Man-AP showed no difference in CD86 and CD40 expression compared to PBS. This is contrary to our previous study in which Man-AP increased the expression of CD86²⁷. The lower expression of the costimulatory molecules reported here might be due to the reduced antigen dosage employed in this study. We decreased the antigen dosage from 25 μg to 15 μg per mouse due to co-formulation with melittin-containing polymers for endosomal release.

Figure 4.

Evaluation of the expression of the co-stimulatory molecules in DCs in vivo. (a) Median fluorescence intensity of CD86 in CD11c⁺ DCs characterized by flow cytometry 24 h post immunization (b) Median fluorescence intensity of CD40 in CD11c⁺ DCs. Data are represented as mean ± SD. N = 4 biological replicates. Statistical analysis was performed using one-way ANOVA with post-hoc Tukey HSD test (*p ≤ 0.05, **** p ≤ 0.0001)

Man-VIPER induces superior in vivo adaptive immune responses compared to non-endosomolytic carriers

Encouraged by these results, we then assessed immunogenicity in a vaccination context. Mice were immunized twice with all formulations on days 0 and 14, and splenocytes were collected, isolated and analyzed on day 21 for antigen-specific T cell responses. The amount of antigen-specific cytotoxic T cells was assessed with SIINFEKL tetramer staining, while the cellular response to antigen was evaluated using intracellular cytokine staining upon restimulation with the antigens. All three polymeric nanoparticles, Man-AP, Man-VIPER-R and Man-VIPER-NR generated more antigen-specific CD8⁺ T cells compared to free peptides, suggesting that the mannosylated polymeric delivery vehicle enhances antigen delivery to DCs and promotes antigen-specific T cell proliferation (Figure 5). Notably, immunization with Man-VIPER-NR generated 2.7-fold more SIINFEKL-specific CD8⁺ T cells compared to Man-AP and Man-VIPER-R, which may indicate more robust CD8⁺ T-cell activation (Figure 5a). Particles with sizes around 40 nm enter the lymphatic system most efficiently in some reports⁹. Man-VIPER-NR has a diameter of 40 nm, while Man-VIPER-R has a slightly larger size of 51.6 nm, which might explain the higher number of antigen-specific CD8⁺ T generated by Man-VIPER-NR. As mentioned previously, we attribute the small diameter in Man-VIPER-NR nanoparticles to the presence of the pentafluorobenzyl linker. Animals treated with Man-VIPER-R or Man-VIPER-NR had significantly higher numbers of antigen-specific cytokine-producing CD8⁺ and CD4⁺ T cells compared to animals treated with Man-AP, although similar numbers of antigen-specific CD8⁺ T cells were observed in animals treated with Man-AP and Man-VIPER-R (Figure 5b, Figure S6). Man-AP generated an average of 6-fold more cytokine-producing T cells compared to free peptides, while Man-VIPER-R and Man-VIPER-NR generated 41- and 42- fold more compared to free peptides. This phenomenon can be attributed to the higher expression of co-stimulatory molecules CD86 and CD40 in DCs generated by Man-VIPER compared to Man-AP in the DC maturation study (Figure 4). It has been reported that CD28, the T-cell cognate receptor for CD80/CD86, can amplify early TCR signaling⁴⁰. Therefore, it is possible that T cells in Man-VIPER treated mice require less antigens to be activated. Both Man-VIPER-R and Man-VIPER-NR generated similar levels of IFNγ-, TNFα-producing CD8⁺ T cells and IFNγ-producing CD4⁺ T cells upon antigen restimulation (Figure 5c). Man-VIPER-R generated more IL-2-producing CD4⁺ T cells than Man-VIPER-NR, indicating that Man-VIPER-R generates a more robust CD4⁺ T cell response. The higher CD4⁺ T cell activation induced by Man-VIPER-R could be attributed to the slightly higher expression of CD40 on DCs (Figure 4b), which is critical for activation of CD4⁺ T cells⁴¹. However, as CD8⁺ T cells are directly associated with tumor cell elimination, we used Man-VIPER-NR in the tumor reduction study as it induced a higher antigen-specific CD8⁺ T cell proliferation compared to Man-VIPER-R.

Figure 5.

Evaluation of the antigen-specific T cell responses in vivo. (a) Percentage of SIINFEKL-tetramer⁺ CD8⁺ T cells in spleens after immunization. (b-c) Intracellular cytokine staining of the splenocytes. (b) Splenocytes were restimulated with SIINFEKL peptide and percentage of IFNγ⁺ and TNFα⁺ CD8⁺ T cells were characterized by flow cytometry (c) Splenocytes were restimulated with OVA protein and percentage of IFNγ⁺ and IL-2⁺ CD4⁺ T cells were characterized by flow cytometry. Data are represented as mean ± SD. N = 5 biological replicates. Statistical analysis was performed using one-way ANOVA with post-hoc Tukey HSD test (**p ≤ 0.01, ***p ≤ 0.001, **** p ≤ 0.0001).

Man-VIPER-NR slowed tumor growth and improved survival in an antigen-expressing melanoma model

We evaluated the utility of Man-VIPER-NR as a therapeutic vaccine against an aggressive, poorly-immunogenic B16F10 murine melanoma model. Mice were subcutaneously injected with B16F10 cells engineered to express the OVA antigen (B16F10-OVA) on day 0. On days 3, 10 and 19, mice were immunized subcutaneously at the tail base with either PBS, free peptides, Man-AP or Man-VIPER-NR. Anti PD-1 immune checkpoint blockade (ICB) was given to some of the mice one day after each immunization through intraperitoneal injection (Figure 6a). ICB had no effect on tumor growth compared to PBS, likely because the B16F10-OVA tumor model is poorly immunogenic and inefficient in activating antigen-specific cytotoxic T cells⁴². Man-VIPER-NR significantly delayed tumor growth and improved survival compared to free peptides and Man-AP, while Man-AP exhibited modest tumor growth delay and survival benefits consistent with our previous data²⁷ (Figure 6b,c, Figure S7). Co-treatment with ICB interestingly did not result in increased therapeutic efficacy (Figure S8). The immunosuppressive tumor microenvironment, T cell exhaustion and other inhibitory signaling can all lead to the resistance in the ICB therapy⁴³. The tumor response to immunization might be further improved by administration of TGF-β blockade⁴⁴, blocking other inhibitory pathways by combining with anti-CTLA-4⁴⁵, and increasing T cell infiltration⁴⁶. We also noticed 5%-10% weight loss in mice treated with Man-VIPER-NR two days after vaccination, but the mice recovered weight later and demonstrated no other side effects (Figure S9). These results establish Man-VIPER-NR as a promising therapeutic cancer peptide vaccine platform. There exists avenues to improve this Man-VIPER-NR, which includes co-formulation with known adjuvants such as STING, TLR or NLR agonists.

Figure 6.

Evaluation of the antitumor effect in a B16F10-OVA tumor model. (a) Study timeline. Mice were inoculated with B16F10-OVA cells on day 0, vaccinated on days 3, 10, 19, and treated with ICB on days 4, 11, 20. (b) Tumor growth curve. Tumor was measured every other day. Data are represented as mean ± SEM. N = 8 biological replicates. Statistical analysis was performed using unpaired t test on day 21 (**p ≤ 0.01). (c) Kaplan-Meier survival curve. Survival analysis was performed using the log-rank test (**p ≤ 0.01, **** p ≤ 0.0001).

5. Conclusions

In this work, we developed a mannosylated endosomolytic cancer peptide vaccine platform building from our previous mannosylated peptide delivery work and our established Virus-Inspired Polymers for Endosomal Release (VIPER) system. Man-VIPER-NR self-assembles into micelles suitable for lymph node trafficking and efficiently delivers peptide antigens to the cytosol through endosomal lysis induction, resulting in superior antigen cross-presentation, cytotoxic T-cell priming and potent antitumor response. This culminates into superior efficacy as a therapeutic vaccine compared to non-endosomolytic formulations in an aggressive, poorly immunogenic murine melanoma model.