Titrating polyarginine into nanofibers enhances cyclic-di-nucleotide adjuvanticity in vitro and after sublingual immunization

Abstract

Effective sublingual peptide immunization requires overcoming challenges of both delivery and immunogenicity. Mucosal adjuvants, such as cyclic-di-nucleotides (CDN), can promote sublingual immune responses but must be co-delivered with the antigen to the epithelium for maximum effect. We designed peptide-polymer nanofibers (PEG-Q11) displaying nona-arginine (R9) at a high density to promote complexation with CDNs via bidentate hydrogen-bonding with arginine side chains. We co-assembled PEG-Q11 and PEG-Q11R9 peptides to titrate the concentration of R9 within nanofibers. In vitro, PEG-Q11R9 fibers and cyclic-di-GMP or cyclic-di-AMP adjuvants had a synergistic effect on enhancing dendritic cell activation that was STING-dependent and increased monotonically with increasing R9 concentration. Polyvalent display of R9 on assembled nanofibers was significantly more effective at promoting CDN-mediated DC activation in vitro than mixing nanofibers with an equimolar concentration of unassembled R9 peptide. Sublingual administration of nanofibers revealed a bell-shaped trend between increasing R9 concentration and enhancements to antigen trafficking and activation of DCs in the draining lymph nodes. Intermediate levels of R9 within sublingually-administered PEG-Q11 fibers was optimal for immunization, suggesting a balance between polyarginine’s ability to sequester CDNs along the nanofiber and its potentially detrimental mucoadhesive interactions. These findings present a potentially generalizable biomaterial strategy for enhancing the potency of CDN adjuvants and reveal important design considerations for the nascent field of sublingual biomaterial immunization.

Graphical Abstract

1. INTRODUCTION

Traditional vaccines against infectious pathogens offer limited ability to tune the specificity or phenotype of the protective immunity they elicit. Bespoke biomaterial vaccine platforms have emerged as alternatives to pathogen-based vaccines that offer greater control over the responses they raise^1–3. This engineerability provides a means of overcoming significant hurdles associated with raising certain types of immune responses, such as responses against peptide epitopes, mucosal delivery, and poor immunogenicity. While the use of peptide epitopes can direct a vaccine response against a precise target, generally they are poorly immunogenic compared with subunit or larger antigens^4–5. Mucosally delivered vaccines also typically have heightened immunogenicity requirements, and mucosal routes can degrade or damage antigens during administration.^6–8 The incorporation of adjuvants into biomaterial vaccines can be used to enhance immunogenicity, while the choice of material and design can significantly impact the effectiveness of mucosal transport^9–10. In recent years, biomaterial vaccines have been designed to promote mucosal delivery^11–17, to raise responses against highly specific peptide epitopes^18–21, and to elegantly incorporate adjuvants to tune response phenotypes^21–23.

We recently reported the design of self-assembling peptide-polymer nanofibers capable of raising B- and T-cell responses against peptide epitopes after sublingual immunization (delivery across the mucosal surfaces under the tongue)²⁴. These nanofibers contain a poly(ethylene) glycol (PEG) corona to shield them from mucus in the sublingual space, while relying on the combined effects of peptide self-assembly and adjuvants to overcome the immunogenicity barrier. Here, we report a simple strategy to enhance the co-delivery of cyclic-di-nucleotide (CDN) adjuvants and nanofibers to promote STING (stimulator of interferon genes)-mediated dendritic cell (DC) activation in vitro and after sublingual immunization. Utilizing the ability of polyarginine sequences to complex with CDNs via electrostatics and hydrogen bonding, we co-assembled nona-arginine (R9) ligands into our nanofibers to promote these non-covalent adjuvant-to-nanofiber interactions^25–26. We demonstrate the ability to titrate the level of R9 ligand into the nanofiber to modulate the extent of CDN-mediated DC activation in vitro, with activation increasing monotonically with R9 level, and show that our supramolecular approach elicits greater activation than mixing CDNs with unassembled R9 peptides. In vivo, we discovered a bell-shaped response of antigen trafficking and DC activation in draining lymph nodes when CDN-adjuvanted nanofibers of increasing R9 content were delivered sublingually, with 25% molar R9 in the nanofiber having the greatest effect, in contrast with the in vitro findings. These results highlight the complexity of designing mucosal vaccines to simultaneously promote delivery and immunogenicity and the importance of optimizing biomaterial parameters to maximize immune responses.

2. MATERIALS AND METHODS

2.1. Peptide-Polymer Synthesis and Nanofiber Preparation.

Peptides were synthesized using standard Fmoc solid-phase synthesis on Rink amide resins. PEG-peptides were synthesized by on-resin conjugation of 2000 MW mPEG-NHS (Creative PEGWorks PLS-214) to the N-terminus. Fluorescent peptides were synthesized by on-resin conjugation of 5 (6)-TAMRA (AnaSpec, AS-81120–01) to the N-terminus using DIC and HOBt as coupling reagents. Peptides were cleaved for 2 h at room temperature in a 95/2.5/2.5 TFA/triisopropylsilane/water cocktail, followed by washing with cold diethyl ether. Peptides were purified by reverse-phase HPLC using a C4 column (PEG-peptides) or C18 column (non-PEGylated peptides) and lyophilized. Peptide identity was confirmed using matrix-assisted laser desorption/ionization mass spectrometry on a Bruker Autoflex Speed LRF MALDI-TOF spectrometer using α-cyano-4-hydroxycinnamic acid (Sigma Aldrich, 70990) as the matrix.

To prepare nanofiber solutions, lyophilized peptides were dissolved at 8 mM in sterile water and incubated at 4 °C overnight. The solutions were then brought to the final concentrations in 1X PBS by addition of sterile water and sterile 10X PBS and incubated at room temperature for 3 h before use to allow for fibrillization.

2.2. Nanofiber Characterization.

To visualize nanofiber morphology by transmission electron microscopy, nanofiber solutions were diluted to 0.2 mM in 1X PBS and deposited onto Formvar/carbon-coated 400 mesh copper grids (Electron Microscopy Sciences, EMS400-Cu) for 1 min, rinsed with ultrapure water, and negatively stained for 1 min with 1% w/v uranyl acetate (EMS, 22400–1) prior to wicking away with filter paper. Samples were imaged on an FEI Tecnai G² Twin electron microscope at 120 kV.

Secondary structure analysis was performed by assaying β-sheet content using thioflavin T (ThT; Alfa Aesar, J61043). Nanofiber or peptide solutions were diluted to 0.2 mM in a 0.05 mM solution of ThT and the fluorescence intensity was measured with an excitation wavelength of 440 nm and an emission wavelength of 482 nm using a Molecular Devices Spectramax M2 spectrophotometer.

Zeta potential was measured on fibrillized nanofiber solutions diluted to 0.2 mM in 1X PBS. Solutions were analyzed at 25 °C using an Anton Paar Litesizer 500.

2.3. In Vitro DC Activation Assays.

To assess DC activation in vitro, DC2.4 cells were seeded at 2.5 × 10⁵/well in a 12-well plate in complete DC media (RPMI + 10% heat-inactivated FBS + 50μM 2-mercaptoethanol + 10 mM HEPES (pH 7.4) + 1X non-essential amino acids + 1X penicillin-streptomycin). The next day, nanofiber solutions of PEG-Q11OVA and PEG-Q11(OVA/R9) were prepared at 2mM in the presence of c-di-GMP or c-di-AMP. 2X treatment solutions were prepared by diluting nanofiber solutions to 0.1 mM in complete DC media. Half of the culture medium in each well was replaced with 2X treatment media resulting in a final nanofiber concentration of 0.05 mM and final c-di-GMP or c-di-AMP concentrations of either 1 μg/mL or 5 μg/mL. Cells were cultured under treatment conditions overnight. For experiments involving inhibition of the STING pathway, cells were pretreated with the STING antagonist C-178 (Axon Medchem, 2923) for 4 hours prior to treatment at 5 μM, and 5 μM C-178 was additionally included in the treatment conditions. Following treatment, cells were trypsinized and collected for flow cytometric analysis. Cells were stained with anti-mouse CD80-PE (Clone: 16–10A1)(BioLegend, 104708) and anti-mouse CD86-PE-Cy7 (Clone: GL-1)(BioLegend, 105014) and DAPI and analyzed on a BD FACSCanto II cytometer. Data were analyzed using FlowJo software.

2.4. Assessments of DC Presentation and Cross-Presentation of Model Epitopes In Vitro.

To measure DC presentation of a model epitope to CD4 T cells, nanofiber solutions of PEG-Q11OVA and PEG-Q11(OVA/R9) were prepared at 2 mM in the presence of c-di-GMP or c-di-AMP. 2X treatment solutions were prepared by diluting nanofiber solutions to 0.1 mM in complete DC media. 2X treatment solutions were mixed with equal volumes of complete DC media containing DC2.4 cells and plated into a 96-well plate at 1 × 10⁴ cells/well resulting in a final nanofiber concentration of 0.1 mM and final c-di-GMP or c-di-AMP concentrations of 10 μg/mL. Cells were incubated at 37°C for 2 h. The plate was then centrifuged at 500 × g for 5 min, the supernatant was aspirated, and 100 μL of DOBW cells (T cell hybridoma) resuspended at 5 × 10⁵/mL in complete DMEM media (DMEM + 10% heat-inactivated FBS + 1X penicillin-streptomycin) were added to each well and incubated overnight at 37 °C. DOBW cells produce IL-2 when they encounter pOVA presented in MHC class II (I-Ab). The plates were centrifuged for 5 min at 500 × g, the supernatant was collected, and the IL-2 concentration in the supernatant was measured using an ELISA kit (BD Bioscience, 55148). DOBW cells were provided by C. Harding.

To assess DC cross-presentation of a model epitope, DC2.4 cells were seeded at 2.5 × 10⁵/well in a 12-well plate in complete DC media. The next day, nanofiber solutions of PEG-Q11SIINFEKL and PEG-Q11(SIINFEKL/R9) were prepared at 2mM. 2X treatment solutions were prepared by diluting nanofiber solutions to 0.2 mM in complete DC media and vortexed. Half of the culture medium in each well was replaced with 2X treatment media resulting in a final nanofiber concentration of 0.1 mM. Nanofibers were formulated such that the final SIINFEKL concentration in the treatment culture medium was either 100 μM, 1 μM, or 0.1 μM. Treated cells were incubated for 24 h at 37 °C for either 1 or 2 h. Cells were then trypsinized and collected for flow cytometric analysis. Cells were stained with anti-SIINFEKL peptide bound to H-2Kb-APC (Clone: 25-D1.16)(Invitrogen, 17-5743-82) and DAPI and analyzed on a BD FACSCantoII cytometer. Data were analyzed using FlowJo software.

2.5. DC Uptake Assay.

To assess DC uptake of various nanofiber formulations, DC2.4 cells were seeded at 2.5 × 10⁵/well in a 12-well plate in complete DC media. The next day, fluorescent nanofiber solutions were formed at 2 mM by co-assembling PEG-Q11OVA or PEG-Q11(OVA/R9) with 25% PEG-Q11TAMRA. For nanofiber formulations containing variable PEG-Q11R9 content, PEG-Q11OVA was used to backfill the nanofiber formulation such that the total nanofiber concentration of all tested conditions was equal. 2X treatment solutions were made by diluting the fluorescent nanofiber solutions to 40 μM. Half of the culture medium in each well was replaced with 2X treatment media resulting in a final nanofiber concentration of 20 μM. Cells were treated for either 1 or 4 h before trypsinization and collection for flow cytometric analysis. Cells were stained with DAPI and analyzed on a BD FACSCanto II cytometer to quantify the degree of TAMRA uptake. Data were analyzed using FlowJo software.

2.6. Animals and Immunizations.

For experiments investigating immune responses against ESAT6, female CBA/J mice were purchased from Jackson Laboratories (due to their compatibility with the T-cell epitope in ESAT6_51–70) and immunizations were initiated for mice aged 12 weeks. For in vivo assessment of DC trafficking and activation, female C57BL/6 mice were purchased from Envigo and immunizations were initiated for mice aged 12 weeks. Animal experiments were approved by the Institutional Care and Use Committee of Duke University. For sublingual immunizations, mice were deeply anesthetized by a cocktail delivering 100 mg/kg ketamine and 10 mg/kg xylazine. A micropipette with a 20 μL tip was used to apply 8 μL of the immunizing solution below the tongue, and the mouse’s heads were placed in anteflexion for 20 min following administration to prevent swallowing of the material. Vaccine grade cyclic-di-GMP (Invivogen, vac-nacdg) and cyclic-di-AMP (Invivogen, vac-nacda) adjuvants were mixed with nanofiber solutions at a dose of 10 μg/mouse.

2.7. Antibody Response Measurement by ELISA.

Serum was collected via the submandibular vein. For analysis of antigen-specific IgG by ELISA, plates were coated with 20 μg/mL of PEG-Q11ESAT6_51–70 overnight at 4 °C. Plates were washed with 0.5 g/L Tween-20 in PBS (1X PBST), blocked for 2 h with 1X PBST containing 20 g/L bovine serum albumin and 0.1% Tween-20 (PBST-BSA). Sera or mucosal secretions were diluted in PBST-BSA and added to the plate, and antigen-specific IgG was detected by horseradish peroxidase (HRP) conjugated Fcγ fragment specific goat anti-mouse IgG (Jackson Immuno Research, 15-035-071). Antibody titers were calculated using an absorbance cutoff of 0.2 OD. For measuring antibody subclasses, HRP conjugated goat anti-mouse detection antibodies were purchased from Southern Biotech (IgG1: 1071–05, IgG2b: 1091–05, IgG2c: 1078–05).

2.8. T-Cell Response Measurement by ELISPOT.

To analyze T cell activation by ELISPOT, mice were sacrificed 7 d after the final booster immunization and spleens were harvested. Briefly, 0.25 million splenocytes in 100 μL were plated in each well of a 96-well ELISPOT plate (Millipore, MSIPS4510). The cells were stimulated with 2.5 μM ESAT6_51–70 peptide, left untreated as negative controls, or stimulated with Concanavalin A (ConA) (Sigma, C5275) as positive controls. Biotinylated capture-detection antibody pairs for either IL-4 (BD, 551878) or IFNγ (BD, 551881) were used according to manufacturer’s guidelines in conjunction with streptavidin-alkaline phosphatase (Mabtech, 3310–0) and Sigmafast BCIP/NBT (Sigma, B5655). After development, plates were evaluated for spot count by Zellnet Consulting using a Zeiss KS ELISPOT reader.

2.9. DC Trafficking and Activation in Draining LNs After Sublingual Immunization.

To assess the degree of DC-mediated transport of antigen to the draining LN and the activation of antigen presenting DCs, draining lymph nodes were harvested 48 h after immunization with c-di-AMP-adjuvanted nanofibers. The draining LNs were defined as the mandibular LNs, accessory mandibular LNs, and superficial parotid LNs as per the nomenclature established by Van den Broeck et al²⁷. LNs were digested in a solution of 0.5 U/mL Dispase (STEMCELL, 07913), 0.1 mg/mL Collagenase P (Roche, 11213857001), and 0.1 mg/mL DNase (Roche, 10104159001) in serum-free RPMI for a total of 40 minutes with periodic agitation via repeated pipetting. The digested tissues were transferred to ice cold solutions of 10% heat-inactivated FBS + 5 mM EDTA in PBS before being strained through a 70 μM cell strainer. SIINFEKL presentation was detected using either anti-SIINFEKL peptide bound to H-2Kb-APC (Clone: 25-D1.16)(Invitrogen,17-5743-82) or anti-mouse SIINFEKL/H-2Kb-PE (Clone: 25-D1.16)(eBioscience, 12-5743-82). To identify DC subtypes and activation levels, cells were stained with anti-mouse CD11c-PE-Cy7 (Clone: HL3)(BD Biosciences, 558079), anti-mouse CD103-PerCP/Cy5.5 (Clone: 2E7)(BioLegend, 121415), anti-mouse CD80-PE (Clone: 16–10A1)(BioLegend, 104708), anti-Mouse CD86-APC/Cy7 (Clone: GL-1)(BioLegend, 105030), anti-mouse CD40-FITC (Clone: 3/23)(BioLegend, 124607), and DAPI. CountBright Absolute Counting Beads (Invitrogen, C36950) were introduced to the samples before running on the cytometer in order to calculate the absolute number of cells isolated from the draining lymph nodes. Stained cells were analyzed on a BD FACSCanto II cytometer and data were analyzed using FlowJo software.

2.10. Statistical Analysis.

Statistical analysis was performed as indicated in figure legends using GraphPad Prism version 8 software (GraphPad Software, La Jolla, CA). Data are presented as the mean ± standard error of the mean. Statistically significant differences are indicated in each graph as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Non-significant differences are indicated as n.s.

3. RESULTS

We have previously shown that self-assembling peptide-polymer nanofibers can be used to raise immune responses against peptide epitopes when mixed with a suitable mucosal adjuvant and delivered sublingually²⁴. These peptide-polymers contain the Q11 assembly domain, a peptide epitope on one terminus, and a PEG block on the opposite terminus to promote delivery through the salivary mucus layer. We sought to enhance the immunogenicity of our sublingual vaccine platform through co-delivery of cyclic-di-nucleotide (CDN) adjuvants and epitope-bearing nanofibers. CDN adjuvants activate dendritic cells through the cGAS-STING pathway, during which cytosolic interaction between STING and CDNs triggers downstream cell activation²⁸. The two negatively charged phosphate groups on each CDN molecule lead to poor intracellular delivery of free CDNs, suggesting that promoting nanofiber-CDN interactions could enhance co-delivery to promote stronger sublingual immune responses.

3.1. Co-Assembly with Polyarginine-bearing Peptides Titrates Nanofiber Charge and Increases DC Uptake.

We sought to promote nanofiber:CDN complexation by leveraging the polyvalent display of ligands enabled by assembled nanofibers to amplify the effects of two forms of noncovalent interaction. First, the negatively charged phosphate groups on CDNs allow for electrostatic interactions with positively charged molecules. Second, the guanidinium side chain of arginine residues has been shown to form bidentate hydrogen bonds with CDNs²⁶. Polyarginine sequences have been shown to interact more strongly with CDNs than other positively charged polymers, and this complexation has been utilized to enhance immune responses²⁵. We reasoned that by conjugating polyarginine sequences to PEG-Q11 peptides we could form a nanofiber with a dense array of polyarginine sequences to cluster CDNs to a greater extent than that previously achieved with unassembled polyarginine peptides. Further, the demonstrated ability to co-assemble multiple Q11 peptides into a single nanofiber²⁹ provided for the use of one nanofiber containing both a peptide epitope and polyarginine ligands, enabling co-delivery of antigen and adjuvant.

We initially co-assembled PEG-Q11 monomers containing the model OVA_323–339 (pOVA) peptide epitope (PEG-Q11OVA) and monomers containing a nona-arginine sequence (PEG-Q11R9) (Fig. 1). We chose to use R9 as the polyarginine ligand to maximize cellular uptake³⁰ and minimize potential cytotoxicity. PEG-Q11R9 assembled to form extended nanofibers morphologically similar to PEG-Q11OVA, although nanofibers containing the highly charged R9 ligand did appear slightly less regular via electron microscopy visualization (Fig. 1A,C). Co-assembly of PEG-Q11OVA and PEG-Q11R9 at a 1:1 molar ratio (PEG-Q11(OVA/R9)) appeared to restore this loss of regularity (Fig. 1B). To provide a quantitative assessment of these observations, we measured the binding of thioflavin T (ThT) – which recognizes β-sheet structures – to different nanofiber compositions (Fig. 1E). PEG-Q11OVA nanofibers had higher levels of ThT binding than PEG-Q11R9 nanofibers, with the co-assembled nanofibers exhibiting binding between these values. As expected, unassembled R9 peptides exhibited no ThT signal and formed no observable supramolecular structure (Fig. 1D–E). To test the robustness of PEG-Q11(OVA/R9) co-assemblies and assess their potential to co-deliver both peptide epitopes and R9-complexed adjuvant, we titrated PEG-Q11R9 into PEG-Q11OVA assemblies at molar percentages of 0, 25, 50, 75, and 100. Increasing R9 content in the nanofibers led to a consistent, linear increase in zeta-potential, suggesting the ability to smoothly gradate the total nanofiber charge (Fig. 1F).

Figure 1: Co-assembly allows for titration of positive charge into epitope-bearing peptide nanofibers.

PEG-Q11 peptides appended C-terminally to the OVA epitope (PEG-Q11OVA) or nona-arginine ligands (PEG-Q11R9) self-assembled to form nanofibers alone and when mixed at a 1:1 molar ratio (PEG-Q11(OVA/R9)) (A-C), as visualized by electron microscopy (all images 29000x magnification). Soluble R9 peptide without the Q11 assembly domain formed no observable supramolecular structure (D). The expected β-sheet secondary structure of PEG-Q11-containing peptides was confirmed by binding to Thioflavin T (E). Co-assembling PEG-Q11OVA and PEG-Q11R9 led to a dose-dependent increase in zeta potential with increasing PEG-Q11R9 content (F).

The known ability of polyarginine sequences to enhance cellular uptake³⁰ led us to ask whether PEG-Q11R9 might have additional effects on the immune response outside of CDN complexation. We fluorescently labelled PEG-Q11R9, PEG-Q11OVA, and pOVA peptides with the TAMRA fluorophore, then monitored their rate of acquisition by murine DC2.4 dendritic cells (DCs) in vitro (Fig. 2A). Nanofibers containing the R9 ligand showed significantly enhanced uptake compared to nanofibers without R9 or free peptide after as little as 1 hour of treatment. After uptake by antigen-presenting cells such as DCs, peptide epitopes must be processed and presented to T-cells on major histocompatibility (MHC) proteins. Likely owing to the ability of R9 to enhance uptake, treatment of DCs with PEG-Q11(OVA/R9) nanofibers led to significantly greater presentation of the CD4 epitope pOVA to T-cells in vitro, despite containing only half of the molar pOVA epitope content of PEG-Q11OVA nanofibers (Fig. 2B). This effect was independent of adjuvant, as the addition of the CDN adjuvants cyclic-di-GMP (c-di-GMP) or cyclic-di-AMP (c-di-AMP) did not impact the level of presentation measured. These findings suggest that PEG-Q11R9 is able to increase cellular uptake of nanofibers and enhance presentation of peptide epitopes, and could be useful for applications outside of CDN-mediated sublingual immunizations.

Figure 2: Nona-arginine promotes nanofiber uptake and antigen presentation independent of adjuvants.

(A) Positively charged PEG-Q11R9 nanofibers were rapidly acquired by dendritic cells in vitro, as compared with PEG-Q11OVA nanofibers or soluble pOVA peptide. TAMRA-labelled peptides or peptide nanofibers were incubated for 1 or 4 hours with murine DC2.4 dendritic cells and the uptake was measured by flow cytometry. *** p < 0.001, 2-way ANOVA with Tukey’s multiple comparisons test, n=6/group in two independent experiments. (B) Co-assembly of PEG-Q11OVA with PEG-Q11R9 promoted presentation of the pOVA epitope in MHC class II molecules, as measured by DOBW reporter T cells, which secrete IL-2 upon encountering DCs with pOVA loaded MHC II. IL-2 concentration in the supernatant was measured by ELISA. Inclusion of STING agonist adjuvants cyclic-di-AMP (c-di-AMP) or cyclic-di-GMP (c-di-GMP) did not alter presentation. **** p < 0.0001, 1-way ANOVA with Tukey’s multiple comparisons test, n=3/group. MFI = mean fluorescence intensity.

3.2. PEG-Q11R9 Promotes CDN Adjuvant-Mediated DC Activation.

To test our hypothesis that polyvalent R9 display on PEG-Q11 nanofibers would promote intracellular delivery of CDNs, we used DC activation as a simple and reliable readout for CDN delivery. The STING pathway is activated by the presence of CDNs in the cytosol of DCs and leads to DC activation, including the upregulation of the costimulatory molecules CD80 and CD86^{28, 31}. Non-complexed CDN is expected to enter cells poorly due to its negative charge. We found that treating DCs with a mixture of PEG-Q11R9 and c-di-GMP had a synergistic effect, leading to significantly greater expression of CD80 and CD86 than treatment with c-di-GMP alone (Fig. 3A–B). To rule out that non-specific interactions between PEG-Q11 nanofibers and CDNs were causing this effect, we treated DCs with PEG-Q11OVA and c-di-GMP and found no enhancement of DC activation beyond that induced by c-di-GMP alone. While non-adjuvanted PEG-Q11R9 had no effect on CD86 expression, it led to minor increases in CD80 levels.

Figure 3: Incorporation of nona-arginine (R9) into cyclic-di-AMP or cyclic-di-GMP adjuvanted nanofibers leads to dose-dependent enhancement of dendritic cell activation.

PEG-Q11R9 nanofibers mixed with cyclic-di-GMP (c-di-GMP) adjuvant led to significant upregulation of the dendritic cell activation markers CD80 (A) and CD86 (B). Murine DC2.4 dendritic cells were incubated with indicated treatments overnight and activation level was assessed by flow cytometry. **** p < 0.0001, 1-way ANOVA with Tukey’s multiple comparisons test, n=2–5/group. Titrating the PEG-Q11R9 content in nanofibers led to a PEG-Q11R9-dose dependent increase in CD80 and CD86 when mixed with (C) c-di-GMP or (D) the related STING agonist adjuvant cyclic di-AMP (c-di-AMP). n=4/group by two independent experiments of n=2/group (C) or n=3/group (D). To analyze the PEG-Q11R9-dependent increase on CD80 and CD86 levels, the fold-change in MFI was plotted against PEG-Q11R9 nanofiber content and a linear regression was performed (E, F). ** p < 0.01, *** p < 0.001, linear regression of MFI fold change vs. molar percentage of PEG-Q11R9 in nanofiber. MFI = mean fluorescence intensity.

We tested the robustness of the PEG-Q11R9 synergy with c-di-GMP by capitalizing on the ability to precisely titrate the R9 content in our nanofibers via co-assembly. We treated DCs with c-di-GMP plus PEG-Q11OVA nanofibers (as a non-R9 nanofiber control), PEG-Q11R9 nanofibers, or co-assembled PEG-Q11(OVA/R9) nanofibers at molar R9 ratios of 25%, 50%, and 75% of total peptide in the nanofibers. The total peptide concentration and adjuvant dose were constant across all groups, isolating R9 content as the tested variable. DC activation was increased in an R9-dose dependent manner, with significant linear correlations observed between the molar percentage of R9 in the nanofiber and the degree of CD80 and CD86 expression (Fig. 3C,E). This result highlights the specificity and modularity of our supramolecular strategy for R9:CDN complexation. We extended this finding by repeating this experiment while replacing the c-di-GMP adjuvant with a different STING-activating CDN adjuvant, c-di-AMP (Fig. 3D,F). As with c-di-GMP, there was a significant correlation between molar percentage of R9 in the nanofiber and CD80 and CD86 expression.

3.3. Enhanced DC Activation is the Result of a Specific R9:CDN Interaction.

To further demonstrate the specificity of using PEG-Q11R9 to enhance CDN delivery and DC activation, we synthesized PEG-Q11 components containing nona-lysine sequences (PEG-Q11K9) (Fig. S1). These peptide-polymers have the same theoretical net charge (+9) as PEG-Q11R9 but lack the guanidinium group on their side chain that allows for bidentate hydrogen bonding between arginine and CDNs. While treating DCs with c-di-GMP and either PEG-Q11R9 or PEG-Q11K9 both increased CD80 and CD86 levels, DCs treated with PEG-Q11R9 were significantly more activated at multiple adjuvant doses (Fig. 4). In further contrast to PEG-Q11R9, titrating K9 content into nanofibers did not result in a significant correlation with CD80 or CD86 expression, indicating inconsistent delivery of CDN to cells (Fig. S2). These results suggested that the synergistic effect of mixing c-di-GMP with PEG-Q11R9 is due to polyvalent hydrogen binding interactions, rather than simple electrostatic interactions.

Figure 4: Arginine incorporation into adjuvanted nanofiber formulations is significantly more effective than lysine at promoting DC activation.

PEG-Q11R9 nanofibers led to significantly greater activation of murine DC2.4 dendritic cells than PEG-Q11K9 when mixed with cyclic-di-GMP (c-di-GMP) adjuvant, as measured by upregulation of CD80 (A) and CD86 (B). DCs were incubated overnight with the indicated treatments and activation levels were measured by flow cytometry. ** p < 0.01, **** p < 0.0001, 1-way ANOVA with Tukey’s multiple comparisons test, n=3/group. MFI = mean fluorescence intensity.

We also examined the ability of PEG-Q11R9 to promote delivery of a different nucleotide adjuvant, the toll-like receptor 9 (TLR9) agonist CpG. As with the STING pathway, engagement of TLR9 occurs intracellularly and leads to downstream activation after delivery to DCs³². However, unlike c-di-GMP and c-di-AMP, CpG is an oligodeoxynucleotide and is not expected to form bidentate hydrogen bonds with guanidinium-containing arginine residues. Under the conditions tested, addition of CpG to PEG-Q11R9 did not promote greater upregulation of either CD80 or CD86, in contrast to robust increases when mixed with c-di-GMP or c-di-AMP (Fig. 5). This further points to the specificity of the interactions between PEG-Q11R9 and CDNs.

Figure 5: Incorporation of nona-arginine (R9) into nanofibers improves dendritic cell activation with cyclic dinucleotide adjuvants but not a different nucleotide adjuvant.

Mixing PEG-Q11R9 nanofibers with either cyclic-di-GMP (c-di-GMP) or cyclic-di-AMP (c-di-AMP) led to significant upregulation of the activation markers (A) CD80 and (B) CD86 on murine DC2.4 dendritic cells. By contrast, combining PEG-Q11R9 with a structurally different nucleotide adjuvant, CpG, did not increase DC activation compared with PEG-Q11R9 alone. DCs were incubated overnight with the indicated treatments and activation was measured by flow cytometry. ** p < 0.01, *** p < 0.001, **** p < 0.0001, 1-way ANOVA with Dunnett’s multiple comparisons test vs. PEG-Q11R9 only, n=3/group. MFI = mean fluorescence intensity.

3.4. PEG-Q11R9:CDN-Mediated DC Activation is STING-Dependent.

To confirm the mechanism of the enhanced DC activation promoted by PEG-Q11R9 and CDNs, we tested whether DC activation was STING-dependent. As a negative control for STING activation, we treated DCs with PEG-Q11R9 nanofibers and cyclic-di-UMP (c-di-UMP). This molecule is structurally similar to c-di-GMP and c-di-AMP, providing for bidentate hydrogen bonding with guanidinium groups and complexation with PEG-Q11R9. However, the pyrimidine of c-di-UMP is believed to inhibit stacking interactions that are important in binding to the STING protein, and it does not activate the STING pathway³³. While addition of c-di-GMP and c-di-AMP significantly enhanced DC activation above that induced by PEG-Q11R9 alone, addition of c-di-UMP had no effect (Fig. 6A,C). We also pre-incubated DCs with a selective small molecule inhibitor of the STING pathway (C-178), which exerts its effect by preventing a critical palmitoylation event needed for STING activation³⁴. Inclusion of the STING inhibitor dramatically reduced the CD80 and CD86 expression levels induced by co-treatment of DCs with PEG-Q11R9 and either c-di-GMP or c-di-AMP (Fig. 6B,D). Collectively, these results (Fig. 6E–F) confirm that the DC activation induced by PEG-Q11R9 and CDNs is due to enhanced delivery of CDNs to cells, where they activate the intracellular STING pathway.

Figure 6: CDN adjuvant-mediated DC activation enhanced by PEG-Q11R9 is STING-dependent.

The synergistic upregulation of CD80 and CD86 on dendritic cells achieved by mixing PEG-Q11R9 and cyclic-di-nucleotide adjuvants is ablated by treatment of DCs with a small-molecule inhibitor of the STING pathway (C-178). Additionally, mixing PEG-Q11R9 with a non-STING activating control, cyclic-di-UMP (c-di-UMP), does not promote the same level of DC activation as cyclic-di-AMP (c-di-AMP) or cyclic-di-GMP (c-di-GMP). (A-D) Representative histograms showing CD80 and CD86 MFI plotted against mode-normalized cell counts. Murine DC2.4 dendritic cells were incubated overnight with the indicated treatments and CD80 and CD86 levels were measured by flow cytometry. STING-inhibited groups were incubated with C-178 for 4 hours prior to treatment with nanofibers and/or adjuvants. (E-F) Quantification of experiments depicted in A-D. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, 2-way ANOVA with Tukey’s multiple comparisons test; stars in “No Inhibitor” grouping indicates differences between “No Inhibitor” groups; stars in “+ STING Inhibitor” groups indicates differences between “No Inhibitor” and “+ STING Inhibitor” for a given treatment condition; n=3/group. MFI = mean fluorescence intensity.

3.5. Polyvalent Display of R9 by Nanofibers is More Effective than Unassembled R9 Peptides at Promoting CDN-Mediated DC Activation.

While our supramolecular strategy for enhancing CDN-mediated DC activation holds particular advantages for the Q11 system, it may also provide a more general strategy for increasing immunogenicity. In principle, any biomaterial platform that can be conjugated to peptide ligands at a high copy number could utilize polyvalent arginine display to promote CDN adjuvant delivery. To highlight the impact of polyvalency, we compared the abilities of free R9 peptides and PEG-Q11R9 nanofibers to activate DCs in vitro when mixed with c-di-GMP adjuvant (Fig. 7). At the same molar dose of R9, PEG-Q11R9 promoted significantly greater upregulation of CD80 and CD86 than unassembled R9 peptide. This suggests that a nanofiber with a dense array of R9 ligands along its axis is a more effective vehicle for delivering c-di-GMP to cells than unassembled R9 peptides.

Figure 7: Supramolecular assembly of PEG-Q11R9 promotes significantly greater CDN-mediated DC activation than unassembled R9 peptide.

(A) Hypothesized structure of assembled PEG-Q11R9 nanofibers or non-assembled R9 peptides interacting with cyclic-di-nucleotides. The densities of PEG and R9 ligands on Q11 are reduced for visual clarity. Polyvalent display of R9 ligands on PEG-Q11R9 nanofibers mixed with cyclic-di-GMP (c-di-GMP) or cyclic-di-AMP (c-di-AMP) led to significantly greater upregulation of (B) CD80 and (C) CD86 by dendritic cells than adjuvant mixed with unassembled R9 peptide. Murine DC2.4 dendritic cells were incubated overnight with the indicated treatments and activation was measured by flow cytometry. *** p < 0.001, 1-way ANOVA with Tukey’s multiple comparisons test, n=3/group. MFI = mean fluorescence intensity.

3.6. Sublingual Nanofiber Vaccine Elicits Systemic and Mucosal Responses against a Peptide Epitope from M. tuberculosis.

To test the ability of sublingually administered PEG-Q11R9 nanofibers to raise immune responses in the context of infectious disease, we utilized a preclinically protective peptide T- and B-cell epitope composed of residues 51–70 of the early secretory antigenic target (ESAT-6) of M. tuberculosis³⁵. We formed co-assembled nanofibers containing PEG-Q11 peptides appended to ESAT6_51–70 (PEG-Q11ESAT6) at molar epitope contents of 10% or 50%, with or without 50% or 90% molar PEG-Q11R9, respectively. We held the total nanofiber concentration constant across groups by backfilling with unmodified PEG-Q11. Mice were immunized sublingually with nanofiber formulations adjuvanted with either c-di-AMP or c-di-GMP adjuvant.

These immunizations yielded several interesting observations on B- and T-cell responses that may be useful for the nascent field of sublingually delivered nanomaterials. Nanofibers containing the higher epitope content (PEG-Q11(ESAT6_0.5/R9_0.5) and PEG-Q11(ESAT6_0.5)) raised strong serum antibody responses against ESAT6_51–70, while groups containing only 10% molar epitope content raised poor, inconsistent responses (Fig. 8A, S3A–D). The dominant antibody subclass was IgG2b, indicative of Th1 responses that are favorable for anti-tuberculosis immunity³⁶ (Fig. S3E). Mucosal vaccines against tuberculosis confer improved protection against the pathogen compared to vaccines administered subcutaneously due to the ability to raise local immune responses^37–38. To assess the mucosal response to our vaccine, we monitored antibody levels in the bronchoalveolar lavage fluid (BALF) of the lower respiratory tract and found that both PEG-Q11(ESAT6_0.5/R9_0.5) and PEG-Q11(ESAT6_0.5) elicited mucosal IgG (Fig. 8B). T-cell responses to the ESAT6 epitope were less dependent on epitope dose, with responses observed in all groups (Fig. 8C, S4). Notably, groups containing c-di-AMP adjuvant showed balanced IFNγ and IL-4 responses, while the c-di-GMP adjuvant promoted responses that were skewed towards IL-4 (Fig. S4C).

Figure 8. Titration of polyarginine into sublingually delivered nanofibers alters CD8 epitope trafficking and dendritic cell activation.

(A-C) CBA/J mice (n=5/group) were primed on Week 1 and boosted on Weeks, 3, 5, and 7 with sublingually-delivered nanofiber vaccines containing 50% molar PEG-Q11ESAT6 and adjuvanted with cyclic-di-nucleotides. Vaccination raised antibody responses in the serum (A) and BALF (B). In (A), bold lines represent the mean titer and faded lines represent the titers of individual mice. n.s. = p > 0.05, unpaired, two-tailed t-test. (C) Nanofiber vaccines elicited robust cellular immunity with both Th1 and Th2 responses detected. An ELISPOT was performed on splenocytes harvested from mice on Week 8. SFC: spot-forming cells. n.s. = p > 0.05, multiple t-tests with Holm-Šídák correction. (D-H) Analysis of DCs within draining LNs after sublingual immunization of C57BL/6 mice with a model epitope, SIINFEKL, shows the effects of polyarginine content on cross-presenting DC trafficking and activation. Sublingual nanofiber vaccines were adjuvanted with c-di-AMP. (D) Gating strategy used to identify CD11c⁺ DC subsets (CD103 +/−) and SIINFEKL-presenting DCs within a representative sample. (E) Quantification of the percentage of DC populations presenting the SIINFEKL epitope. (F) Number of SIINFEKL-presenting CD11c⁺ CD103⁺ cells in the draining LNs. (G,H) Quantification of the expression of co-stimulatory molecules CD80 and CD86 on the surface of SIINFEKL⁺CD11c⁺CD103⁺ cells. (D-H) * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, n.s. = p > 0.05 by 2-way ANOVA with Tukey’s multiple comparisons (E) or 1-way ANOVA with Tukey’s multiple comparisons (F-H), n=5/group. MFI = median fluorescence intensity.

The results of these immunizations show the ability of our sublingual nanofiber vaccine platform to raise relevant antibody and T-cell responses both systemically and mucosally, and also highlight the importance of epitope dose and adjuvant choice in designing sublingual biomaterial vaccines. These sublingual immune responses were not significantly enhanced by co-assembly with polyarginine ligands, a surprising finding given the striking synergy of PEG-Q11R9 and CDN adjuvants in boosting DC activation in vitro. This unexpected finding prompted a final experiment, described below.

3.7. Titrating R9 into Nanofibers Enhances DC Trafficking and Activation in the Draining Lymph Nodes After Sublingual Immunization.

To investigate the difference between the in vitro and in vivo effects of combining CDN adjuvants with polyarginine-bearing nanofibers, we monitored the trafficking and activation of dendritic cells after sublingual immunization. We hypothesized that excess positively charged polyarginines of PEG-Q11R9 fibers could lead to adhesive interactions with mucus in the salivary layer and hinder transport to the epithelium, thus limiting DCs’ ability to acquire fibers and actively transport them to the draining lymph nodes.

To track nanofiber-acquiring DCs, we co-assembled the model CD8 epitope OVA_257–264 (SIINFEKL) into nanofibers and measured SIINFEKL-presenting cells in the draining lymph nodes²⁷ (mandibular, accessory mandibular, and superficial parotid) 48 hours after sublingual administration. We titrated the level of R9 in PEG-Q11(SIINFEKL/R9) nanofibers, forming fibers containing 0%, 25%, 50%, or 75% molar ratios of R9. Consistent with intranasal Q11 cross-presentation³⁹, we observed CD11c⁺CD103⁺ DCs to be the major SIINFEKL-presenting population, with no presentation by CD11c⁺CD103⁻ cells (Fig. 8D–E). The percentage of SIINFEKL-presenting CD11c⁺CD103⁺ cells was significantly elevated only in mice receiving nanofibers containing intermediate amounts of R9 (25%, 50%), and not in mice that had either 0% or 75% R9 (Fig. 8E). To rule out that R9-bearing nanofibers are simply intrinsically better at promoting cross-presentation, we measured SIINFEKL presentation in vitro and found R9 co-assembly to have no effect (Fig. S5). When adjusted for cell count, only mice receiving PEG-Q11(SIINFEKL/R9_0.25) co-assemblies had significantly greater total numbers of SIINFEKL⁺CD11c⁺CD103⁺ cells, and a bell-shaped trend was observed, with intermediate levels of R9 maximizing the response (Fig. 8F, S6A–B). These results are consistent with the hypothesis that the CDN/R9 synergy is still present in vivo, but excess positive charge beyond the level necessary for complexation can have a deleterious effect on DC trafficking and activation.

A key finding in vitro was the ability of CDNs and R9-bearing nanofibers to increase DC activation. To test the in vivo relevance of this finding, we measured the levels of the costimulatory markers CD80 and CD86 in SIINFEKL-presenting CD11c⁺CD103⁺ DCs in the draining lymph node (Fig. 8G–H, S6C–D). CD80 levels in SIINFEKL⁺ DCs were significantly greater among mice receiving PEG-Q11(SIINFEKL/R9_0.25) nanofibers than in all other groups, highlighting the ability to optimize R9 content via co-assembly to maximize DC activation. CD80 levels again showed a bell-shaped trend, peaking with intermediate levels of R9. Consistent with the in vitro findings, a simple mixture of peptide R9 with PEG-Q11SIINFEKL nanofibers did not increase either SIINFEKL⁺CD11c⁺CD103⁺ cell numbers or CD80 expression, validating polyvalency and supramolecular assembly as strategies for enhancing the effects of CDN/R9-mediated immune responses. Together, these results suggest that intermediate amounts of R9 in the nanofibers enable optimal complexation and delivery of CDNs, a finding facilitated by the modularity of the platform.

4. DISCUSSION

In this study, we report a strategy for enhancing CDN-mediated immunogenicity in vitro by complexation with highly polyvalent displays of polyarginine ligands on peptide nanofibers. We further show that the ability to smoothly gradate polyarginine concentration via nanofiber co-assembly allows this approach to be optimized for sublingual delivery, where excess positive charge appeared to negatively impact efficacy. Our findings have significant implications both for the nascent field of sublingual biomaterial immunization and for the design of immune-modulating strategies more broadly.

We have previously shown that effective sublingual immunization with peptide nanofibers requires both surface modification with mucus-inert materials such as PEG or PAS (peptide sequences composed of proline, alanine, and serine) and formulation with a mucosal adjuvant, such as c-di-AMP or CTB (the nontoxic B subunit of cholera toxin)²⁴. Our working hypothesis is that the adjuvant is required to recruit dendritic cells to the sublingual space, where steady state DC levels are low⁴⁰, while surface modification reduces rapid mucus-mediated nanofiber clearance²⁴ to allow recruited DCs to actively transport nanofibers to the draining lymph nodes and spleen.

The results reported here are consistent with this hypothesis. We show in vitro that PEG-Q11R9 nanofibers promote intracellular delivery of CDN adjuvants to DCs in an R9 dose-dependent fashion (Fig. 3), activating the STING-cGAS pathway and upregulating costimulatory markers (Fig. 6). While DC activation increases monotonically with increasing R9 concentration in vitro, the effect during sublingual immunization is complicated by the additional factor of delivery through the salivary mucus layer. We show that at a moderate molar R9 concentration of 25% within co-assembled PEG-Q11 nanofibers there is a maximum effect on antigen trafficking to the draining lymph nodes (Fig. 8F). The increase in total number of antigen-presenting DCs in the nodes with PEG-Q11(SIINFEKL/R9_0.25) nanofibers supports the hypothesis that better CDN co-delivery enhances DC recruitment to the sublingual space, increasing active cell-mediated transport. This is further supported by the increased CD80 levels in SIINFEKL-presenting DCs for mice in the 25% R9 group, which suggests greater acquisition of CDN adjuvant by these DCs (Fig. 8G).

These results suggest an optimization between two opposing effects of polyarginine; when no R9 is present there is no enhancement of CDN co-delivery with the nanofibers toward the sublingual epithelium, while at overly high R9 concentrations, mucoadhesive interactions begin to counter the mucus-penetrating effects of PEG, impeding nanofiber transport to the epithelium. We have previously shown that in vitro measurements of mucin adhesion to nanofibers correlates strongly with in vivo efficacy as sublingual vaccines²⁴, and many cationic polymers are known to promote mucoadhesion through electrostatic interactions with negatively charged mucin groups^41–42. It is therefore reasonable to hypothesize that the increasing positive charge that occurs with increasing R9 nanofiber content (Fig. 1F) is accompanied by stronger mucoadhesive properties, although further experiments are needed to address the potentially complex interplay between adjuvanticity, mucoadhesion, and mucus penetration. Still, these results highlight the benefits of flexible supramolecular vaccine platforms versus more traditional attenuated pathogen or subunit approaches. By controlling the extent of R9 modification, we illustrate the ability to incorporate a ligand that might otherwise be considered simply unamenable to sublingual delivery, and use this control to enhance sublingual antigen trafficking and DC activation.

We report here that our sublingual nanofiber vaccine raises both systemic and mucosal antibody responses and splenic T-cell responses against a peptide epitope from M. tuberculosis, but that these responses are not enhanced by co-assembly with 50% molar R9 (Fig. 8A–C). While not studied here, several additional modifications could further enhance the effects of the vaccine by optimizing the immunogenicity/delivery balance. In addition to a careful titration of R9 content within PEG-Q11(ESAT6/R9) nanofibers, a shorter polyarginine ligand or longer PEG chain may also help to maximize CDN complexation while minimizing mucoadhesion.

There exist some alternate explanations and caveats to this study. First, the loss of effectiveness observed at high R9 concentrations in vivo could be related to morphological changes to the nanofibers caused by effects of positive charge on assembly. We did observe some alteration of nanofibers qualitatively (Fig. 1A–C) and a reduction in β-sheet secondary structure quantitatively (Fig. 1E) when appended to R9, although this concern is mitigated by our experience with nanofibers of varying morphologies, lengths, and secondary structures raising immune responses via the subcutaneous route^{24, 43–46}. We also observed some cell death in vitro after extended treatment with nanofibers containing 100% PEG-Q11R9 (Fig. S7). This is not unexpected, as the cytotoxicity of polyarginine polymers increases with increasing molecular weight⁴⁷, and a similar trend may exist with polyvalency of polyarginines. Consistent with this, recently reported polyarginine-rich nanocomplexes also showed some in vitro cytotoxicity⁴⁸. It is possible that there is some influence of cell death-related products on DC activation in our study. However, our results strongly suggest that the enhanced DC activation observed is a direct result of synergistic effects between polyarginine and the cyclic-di-nucleotide adjuvants, as robust DC activation was not observed after treatment with either polyarginine-bearing nanofibers or adjuvant alone. Further, in vitro cytotoxicity assays are not directly relevant to sublingual application with nanomole quantities of material, and R9-mediated cell death is unlikely to explain the in vivo trends observed in this study.

The strategy reported here could have generalizable benefits outside of sublingual immunization. The polyvalent display of R9 on nanofibers was significantly more effective than simple mixing with a polyarginine peptide both in vitro and in vivo (Fig. 7–8). Cyclic-di-nucleotide adjuvants are also used for other mucosal routes, such as intranasal immunization^49–51, and for parenteral immunizations, particularly within the cancer field^52–54. Our findings may be complementary to those of other research groups that have recently developed biomaterial-based strategies for CDN delivery. Encapsulation of CDNs within a number of nanoparticle or microparticle platforms has been shown to enhance delivery to the lymph nodes with multiple routes of administration^55–58. Biomaterial platforms with the ability to multivalently display ligands could utilize polyarginine sequences to promote CDN complexation to further promote immunogenicity. In contexts such as intranasal administration with thinner mucus layers where mucoadhesion is well-tolerated or even preferred⁵⁹, the combination of polyvalent R9 display and CDN adjuvants could have additional synergies for enhancing immune responses.

5. CONCLUSIONS

We developed a strategy for combining polyvalent arginine display on peptide-polymer nanofibers with cyclic-di-nucleotide adjuvants to enhance DC activation in vitro and promote antigen trafficking and DC activation after sublingual immunization. This may be a generalizable strategy for enhancing the potency of CDN adjuvanticity with biomaterial vaccine platforms that are amenable to polyvalent display of ligands. Our ability to titrate the concentration of R9 within nanofibers allowed us to discover different optimal polyarginine levels for in vitro and in vivo conditions. We found that in vitro DC activation increased monotonically with increasing R9, while moderate (25%) concentrations of R9 were most effective for sublingually delivered nanofibers. These findings highlight the merit of a nuanced, gradated, and modular approach to vaccine design to address the complex material and immunological challenges of effective immunization.