Randomized peptide assemblies for enhancing immune responses to nanomaterials

Abstract

Biomaterials capable of inducing immune responses with minimal associated inflammation are of interest in applications ranging from tissue repair to vaccines. Here we report the design of self-assembling randomized polypeptide nanomaterials inspired by glatiramoids, an immunomodulatory class of linear random copolymers. We hypothesized that peptide self-assemblies bearing similar randomized polypeptides would similarly raise responses skewed toward Type 2 immunity and T_H2 T-cell responses, additionally strengthening responses to co-assembled peptide epitopes in the absence of adjuvant. We developed a method for synthesizing self-assembling peptides terminated with libraries of randomized polypeptides (termed KEYA) with good batch-to-batch reproducibility. These peptides formed regular nanofibers and raised strong antibody responses without adjuvants. KEYA modifications dramatically improved uptake of peptide nanofibers in vitro by antigen presenting cells, and served as strong B-cell and T-cell epitopes in vivo, enhancing immune responses against epitopes relevant to influenza and chronic inflammation while inducing a KEYA-specific Type 2/T_H2/IL-4 phenotype. KEYA modifications also increased IL-4 production by T cells, extended the residence time of nanofibers, induced no measurable swelling in footpad injections, and decreased overall T cell expansion compared to unmodified nanofibers, further suggesting a T_H2 T-cell response with minimal inflammation. Collectively, this work introduces a biomaterial capable of raising strong Type 2/T_H2/IL-4 immune responses, with potential applications ranging from vaccination to tissue repair.

INTRODUCTION

Biomaterials and nanomaterials that elicit immune responses with a Type 2/T helper 2 (T_H2)/IL-4 phenotype are of increasing interest in a range of biomedical applications including vaccination and tissue repair. In both contexts, materials based on peptides have advantages owing to their specificity and ability to limit off-target effects. In vaccine design, peptide epitope vaccines are being explored against a variety of infectious diseases, cancers, and therapeutic targets. T-cell vaccines targeting infectious diseases[1–4], cancers [5,6], and several others[7,8] have achieved protection by activating and expanding epitope-specific CD4⁺ and CD8⁺ T-cell populations. Also, B-cell epitope vaccines are being investigated for allergy[9] and inflammation[10], infectious diseases [2,11], and a variety of cancers [6,12]. Although for many vaccination contexts Type 1/T_H1/IFNγ immune responses are highly desirable, there are immunotherapy applications where such phenotypes can be a liability. For example, for active immunotherapies against autologous targets such as cytokines, Type 1/T_H1/IFNγ immune phenotypes have been found to be associated with poorer performance, whereas Type 2/T_H2/IL-4 immune phenotypes have been found to be associated with improved performance[13,14]. In tissue repair, it is also increasingly appreciated that biomaterials eliciting T_H2/IL-4 immune profiles promote healing[15], whereas other phenotypes such as T_H1 or T_H17 are more associated with foreign body responses and fibrosis[16]. However, pioneering work in this area has primarily focused on complex decellularized tissue biomaterials, and there are few demonstrations of synthetic or epitope-based biomaterials rasing such pro-regenerative responses. Thus, peptide biomaterials promoting Type 2/T_H2/IL-4 immune phenotypes would be desirable in a wide range of applications.

To generate immune responses against specific epitopes, peptide-based immunotherapies and vaccines commonly rely on adjuvants and carrier proteins. While carrier proteins supply T cell epitopes and a physical scaffold that enhances multivalency, only a limited number of epitopes can be attached. Further, adjuvants can complicate vaccine formulation by inducing some degree of inflammation, manifested by swelling, redness, pain at the injection site, or systemic effects. Additionally, the resultant phenotype of the immune response is largely dictated by the adjuvant and may not be easily adjusted. Within epitope-based subunit vaccines, it is also challenging to select appropriate B-cell and T-cell epitopes. B-cell epitope identification usually involves predictive algorithms or using known antibodies to identify epitopes within antigens, a laborious process. To provide for T-cell help in epitope vaccines, broadly reactive “universal” T-cell epitopes such as PADRE[17] and VACV-WR, from the vaccinia virus[18], have been employed. Although these T-cell epitopes have been useful for boosting B-cell responses in a variety of platforms, they vary in the strength and breadth of induced immune responses between individuals and can also require extensive optimization. Thus, there is a need for designing immunologically active biomaterials that 1) have broadly reactive T-cell epitopes, 2) can induce desired immune phenotypes, particularly Type 2/T_H2/IL-4, 3) can boost responses to co-delivered antigens, and 4) can provide highly multivalent platforms.

To that end, we sought to create novel nanomaterials consisting of randomized peptide epitopes incorporated within a supramolecular platform, together acting as a Type 2/T_H2/IL-4 nanoscale adjuvant, a term that we will use here to describe materials whose adjuvanting properties depend on their nanoscale structure. Others have created nano-adjuvants by encapsulating toll-like-receptor agonists in poly(lactic-co-glycolic) acid nanoparticles[19,20] or by forming hydrogel depots[21,22] to name a few, but we chose to design a modular material that could be easily titrated with various co-delivered epitopes. To achieve this, we took inspiration from a class of materials termed glatiramoids. Glatiramoids are non-biologic complex drugs created by randomly polymerizing lysine, glutamic acid, tyrosine, and alanine into complex mixtures of polypeptides ranging in size from 4–12 kD. The first glatiramoid, glatiramer acetate[23] (trade name Copaxone), was clinically approved for reducing the frequency of relapses in multiple sclerosis in 1995[24]. It has been shown to act on antigen-presenting cells (APCs) [25,26], bias CD4⁺ T cells towards Th2 responses[27,28], and induce regulatory T cells (Tregs)[29]. Additionally, it is believed that glatiramoids act as universal altered peptide ligands (APLs) to promote the observed anti-inflammatory and Th2 immune cell populations[30]. APLs are epitopes with 1–2 key amino acid substitutions that when presented in class-II MHC lead to altered binding affinities with T-cell receptors (TCRs)[31]. This generally lower binding affinity favors production of anti-inflammatory populations. A therapeutic containing an immense number of different polypeptide sequences, like glatiramoids, is therefore highly likely to contain multiple APLs in a variety of species and genetic backgrounds. This makes glatiramoid-like antigens potentially useful as a universal T-cell epitope in a co-delivered peptide therapeutic. Glatiramoids continue to be actively investigated for the treatment of other autoimmune diseases such as inflammatory bowel disease[32], Huntington’s Disease[33], Alzheimer’s Disease[34], and macular degeneration[35]; however, no clear strategies exist to modify, optimize, or direct glatiramoids toward specific immune phenotypes tailored to distinct diseases because their complex mechanism of action and heterogeneous composition make systematic engineering challenging.

With these motivations, we created an immunomodulatory peptide system inspired by the randomized structure of glatiramoids, but in the form of a supramolecular nanomaterial. Supramolecular peptide therapeutics can co-deliver a variety of B- and T-cell epitopes in a highly multivalent fashion, and their modular design allows for the tuning of relevant physical parameters such as charge, size, and epitope density. Previously, supramolecular peptides have been developed towards vaccines and immunotherapies including KFE8 peptide nanofibers for West Nile Virus[36], Q11 peptide nanofibers for TNF-mediated inflammation[13] and epithelial tumors[37], micellar peptide amphiphiles for group A streptococcus[38], and α-helical coiled coil nanofibers for cancer immunotherapy[39]. In particular, Q11 has been studied as a fibrillizing peptide that when appended to selected B- and T-cell epitopes can raise strong immune responses without adjuvants[13,40]. Q11 is chemically defined, stable in both lyophilized powder and nanofiber form, and predictably forms nanofibers even when epitopes are attached at either terminus. Epitopes synthesized on the termini of Q11 can therefore be precisely co-assembled together, allowing for multi-epitope formulations[41]. While strongly immunogenic, Q11-based materials have been found to be minimally inflammatory[40,42]. They also do not delay healing when used in the context of dermal wounds, further suggesting their intrinsically immunogenic yet non-inflammatory character[42]. We hypothesized that combining the self-assembling and self-adjuvanting properties of Q11 nanofibers with the anti-inflammatory activity of glatiramoids could enhance the immunogenicity of selected peptide epitopes, and that the overall response to the materials would be biased towards a Type 2/T_H2/IL-4 phenotype.

Here, we report the development of a glatiramoid-Q11 analog, termed (KEYA)_nQ11 (Figure 1a). (KEYA)_nQ11 is composed of the peptide Q11 N-terminally functionalized with n amino acids randomly polymerized from mixtures of lysine (Lys, K), glutamic acid (Glu, E), tyrosine (Tyr, Y), and alanine (Ala, A). We found that (KEYA)₂₀Q11 (containing 20 randomized amino acids appended on the N-terminus of Q11) increases uptake of nanofibers by APCs, elicits Type 2/T_H2/IL-4 T-cell and B-cell responses, and amplifies responses to co-assembled epitopes. These results suggest a new strategy for augmenting immune responses to peptide-based therapeutics, especially those employing nanomaterials, and especially for applications where non-inflammatory responses are prioritized.

Figure 1. Reproducible synthesis and characterization of (KEYA)₂₀Q11.

(a) Hypothesized structure of (KEYA)₂₀Q11 self-assembled into a nanofiber. The variety of colors represent the 4²⁰ possible (KEYA)₂₀ sequences. (b) MALDI mass spectrometry indicating a range of molecular weights between the lowest (A₂₀Q11: 3416 g/mol) and highest (Y₂₀Q11: 5258 g/mol) possible with blue bars (Batch B). (c) Amino acid composition of Batches B, F, and G. (d) ThT assay with β-sheet peak at 480 nm, n = 3 experimental replicates per group. (e) Representative AFM of (KEYA)₂₀Q11 confirms nanofiber formation, whereas the peptide p(KEYA)₂₀ does not fibrilize. (f) Viability of DC2.4 dendritic cells and RAW264.7 macrophages after incubation with (KEYA)₂₀Q11 nanofibers with an alamar blue cell viability assay. ***p < 0.001, ****p < 0.0001 by two-way ANOVA with Tukey’s post hoc test. Mean +/− s.e.m. shown. n = 3 experimental replicates per group.

MATERIALS AND METHODS

Peptides.

All peptides listed in the Supplemental Information (Table 1) were synthesized using standard Fmoc solid phase peptide synthesis. The randomized additions for (KEYA)_xQ11 and p(KEYA)₂₀ were prepared by mixing the four Fmoc- and side-chain protected amino acids (lysine, glutamic acid, alanine, and tyrosine) at specific molar ratios prior to addition to the resin-bound peptide. Peptides were cleaved with standard TFA cleavage protocols, precipitated in diethyl ether, and dried overnight[43]. Fluorescent Q11 derivatives were synthesized on-resin by N-terminally conjugating SGSG-Q11 with 5-(and-6)-carboxytetramethylrhodamine (TAMRA) using 1-ethyl-3-(3-(dimethylamino)propyl)carbodiimide coupling. Q11, PADRE-Q11, NP-Q11, K3-Q11, E3-Q11, TAMRA-Q11, and TNF-Q11 were purified with reverse-phase HPLC, and all peptides were lyophilized and stored prior to use[43]. Peptide purity was confirmed using matrix-assisted laser desorption/ionization (MALDI) mass spectrometry on a Bruker Autoflex Speed LRF MALDI-TOF mass spectrometer using a-cyano-4-hydroxycinnammic acid as the matrix. Immunization solutions were prepared by first weighing out lyophilized peptides and vortexing the dry components for 30 minutes to facilitate co-assembly upon dissolution. Mixed peptides were dissolved in sterile cell culture water and incubated overnight at a concentration of 8 mM at 4 °C. Additional sterile water and sterile 10x PBS (Fisher) were added to generate a solution with a final concentration of 2 mM in 1x PBS that was then incubated at room temperature for 2–4 hours to allow for fibrilization into nanofibers. Immunizations involving (KEYA)₂₀Q11 were neutralized with 10 μL of 1 M sterile NaOH overnight to assist in fiber formation. Fluorescent nanofibers were created by vortexing 0.2 mM TAMRA-Q11 into the 2 mM formulation as dry powders, and proceeding as described above, with protection from light. For surface charge modification, 1 mM K3-Q11 and E3-Q11 were co-assembled into the nanofiber formulations by mixing as dry powders and proceeding as above, to add positive and negative net surface charge, respectively. For immunizations involving low concentrations of PADRE-Q11, PADRE-Q11 was first dissolved at 10 mM in sterile water and added to the dry peptide mixture after other dry peptides had been mixed and vortexed. Nanofiber formation then proceeded as described above.

(KEYA)₂₀Q11 structural analysis.

Amino acid analysis was performed by Biosynthesis, Inc. from lyophilized samples. Thioflavin T (ThT) binding was measured by mixing 20 μL of peptide nanofibers, prepared as described above, with 80 μL of 2.5 μM ThT in 1x PBS in triplicate samples in a 96 well plate. After five minutes, fluorescence emission was scanned from 460–600 nm at a fixed excitation of 442 nm on a Tecan Infinite 200 Pro plate reader. Signal peak at 482 is indicative of β-sheet formation. For AFM imaging, mica substrates (Electron Microscopy Sciences) were cleaved immediately prior to sample preparation, and nanofibers were diluted to 0.2 mM peptide concentration in water. 20 μL of diluted nanofibers were spotted onto the mica surface and allowed to adhere for 30 seconds before rinsing with ultrapure water and drying under a stream of nitrogen. Imaging was accomplished by Tapping Force AFM in air, using a Bruker MultiMode AFM and Bruker RTESPA-300 silicon cantilever tips

Cell Culture.

Cell lines:

DC2.4 dendritic cells (ATCC) were cultured in RPMI 1640 with 10% fetal bovine serum, 10 mM HEPES (Invitrogen), and 55 μM β-mercaptoethanol (Invitrogen) in T75 tissue culture flasks (Corning). Cells were passaged using trypsin (Sigma Aldrich) at 80% confluence and seeded for experiments at 300,000 cells/mL unless otherwise specified. RAW264.7 macrophages (ATCC) were cultured in DMEM with 10% fetal bovine serum, 55 uM β-mercaptoethanol, 1x non-essential amino acids (Cellgro), and 10 mM HEPES in 60 mm tissue culture plates (Corning #430166). Cells were passaged using cell scrapers at 80% confluence and seeded for experiments at 500,000 cells/mL unless otherwise specified. B-LCL human B cells (Astarte Bio) were cultured in RPMI with 10% FBS in T25 flasks and seeded for experiments at 200,000 cells/mL. Cell viability: DC2.4 and RAW264.7 cells were plated in 96 well plates and stimulated with peptide nanofibers, diluted to the working concentrations in cell media, and incubated for 24 hours. 10 μL alamarBlue (Invitrogen) reagent was added to each well and the plate was placed back in the incubator for four hours before the fluorescence was read on a Tecan Infinite 200 Pro plate reader using an excitation wavelength of 540–570 nm and an emission wavelength of 580–610 nm. Briefly, an alamarBlue dye was added to cells stimulated with (KEYA)₂₀Q11 or DMSO after which it becomes a fluorescent red color in the reducing environment of a healthy cell but remains blue in unhealthy cells. APC uptake: DC2.4, RAW264.7, and B-LCL cells were seeded in 12 well plates and allowed to adhere for 24 hours before being stimulated for 0.5–24 hours with 0.2 mM of fluorescently labeled nanofibers prepared as described above and diluted in media. Cells were washed twice with PBS and centrifuged before being resuspended in flow buffer (1xPBS containing 2% fetal bovine serum) containing 1 μg/mL of DAPI (ThermoFisher). Flow cytometry was performed on a BD FACSCanto II and analyzed using FlowJo software. Confocal imaging: DC2.4 cells were seeded at 10,000 cells per glass bottom dish (Nunc) and allowed to adhere for 24 hours before being stimulated for 24 hours with fluorescently labeled nanofibers. Media was removed and cells were washed twice with 1x PBS before being fixed with 4% paraformaldehyde. Cells were then blocked and permeabilized with 10% BSA in 1x PBS with 0.3% Triton X-100 before being stained with Phalloidin green (Biolegend) diluted in 1x PBS containing 1% BSA and washed twice in 1x PBS containing 0.1% BSA. Cell nuclei were then stained with DAPI solution before the cells were again rinsed in 1x PBS followed by ultrapure water. Anti-fade mounting medium (ThermoFisher) was carefully applied to the slide, and cells were visualized on a Zeiss 880 Airyscan inverted confocal microscope.

Mice and immunizations.

Female C57BL/6 mice (Envigo) were housed in a centralized animal facility at Duke University. All procedures were approved by the Duke University Institutional Animal Care and Use Committee under protocol A264-18-11. Mice were anesthetized and 100 μL of the immunizations described above were delivered subcutaneously (50 μL to each flank) to each mouse. Booster immunizations were given at time points indicated, with the same formulation as the primary immunizations. Blood was collected from the submandibular vein; serum was isolated and used for ELISA analyses. Five to seven days after the last booster immunization, mice were sacrificed, and their organs were harvested. Intraperitoneal injections consisted of one 100 μL injection to the left side of the abdomen administered 2–12 hours before the mice were sacrificed. The lavage fluid was obtained by flushing the peritoneal space with 1–2 mL of cold PBS and removing 1 mL of fluid. For footpad inflammation measurements, anesthetized mice were injected with 30 μL of PBS, Aluminum Hydroxide (Alum, Milipore Sigma #239186), or (KEYA)₂₀Q11 nanofibers in their right hind paw. Swelling was measured with a digital caliper before injection and at various timepoints up to 72 hours.

Measurements of Immune Responses.

Antibodies:

Serum was analyzed for antigen-specific IgG antibodies as well as subclasses IgG1 and IgG2c by ELISA. Plates were coated with 2 μg/mL streptavidin solution, sealed to prevent evaporation, and incubated overnight at 4 °C. Plates were then washed in a 1xPBST (0.05% w/v Tween-20 in 1xPBS) solution and coated in biotinylated peptides to detect antibodies specific to the formulations in each experiment. After being washed and blocked with Superblock (ThermoFisher), serially diluted serum in 1% w/v BSA in 1x PBST was added to the plate and detected with horseradish peroxidase conjugated Fcγ fragment specific goat anti-mouse IgG, IgG1 or IgG2c (Jackson Immuno Research #115–035-003, #115–035-205, #115–035-208). T-cell activation: ELISpot plates (EMD Millipore) were coated overnight at 4 °C with anti-IFNγ (BD #551873) or anti-IL-4 (BD #551017) capture antibodies and then blocked with ELISpot medium (RPMI 1640, 10% fetal bovine serum, 55 uM β-mercaptoethanol, and 1x non-essential amino acids (Gibco)), 1 mM sodium pyruvate (Gibco). Mice were sacrificed as described above, and single cell suspensions were collected from the spleen that were then treated with ACK lysing buffer (Thermofisher) followed by lympholyte M (Cedar Lane) to isolate lymphocytes. Single cell suspensions prepared from draining lymph nodes collected as described above were washed and seeded directly onto the plates. Draining lymph nodes refers to the inguinal, axillary, and brachial unless otherwise stated. 2.5 or 5 million cells/well were seeded in duplicates, stimulated with 10 μM of the peptide of interest, and incubated at 37 °C for 48 hours. Plates were washed and incubated with biotinylated anti-IFNγ (BD #551873) or anti-IL-4 (BD #551017) detection antibodies, then with streptavidin-alkaline phosphatase (Mabtech), and lastly with Sigmafast BICP/NBT (Sigma Aldrich). Spots were allowed to develop for 5–20 minutes before being washed with ultrapure water and dried overnight. Plates were sent to Zell Net Consulting (New Jersey) for imaging and analysis. Spot count was determined by averaging the duplicates and subtracting the background (unstimulated) spots for each mouse from the stimulated spots before graphing. Any resulting negative number was corrected to 0. Any mouse without spots in the positive control well (stimulated with ConA) was removed from analysis. Cytokine production: Lavage fluid collected as described above was analyzed with a custom Luminex xMAP ELISA kit. Any cytokine levels at a not detectible range were given a zero for the purposes of graphing. T-cell population analysis: Flow cytometry was performed on splenocytes harvested and isolated as described above after a 24-hour incubation and 3-hour stimulation with a Cell Stimulation Cocktail (eBioscience). All fluorophores were purchased from Biolegend unless otherwise specified. Once in single cell suspension, the Fc receptor was blocked with 2.4G2 antibody (BD #553141) and cells were stained for CD3 (#100206), CD4 (#100411), CD8 (#100723), and CD25 (#101908). Cells were then fixed and permeabilized in a fixation/permeabilization buffer (BD #554714), stained with LIVE/DEAD Fixable Violet Dead Cell Stain Kit (Invitrogen), and stained with intracellular IL-4 (#504117). Flow cytometry was performed on a BD FACSCanto II and analyzed using FlowJo software.

In vivo imaging.

The back fur on CD-1 female mice (Charles River) was removed and mice were injected with TAMRA-labeled nanofibers on the right and left flanks of the back in hairless areas. An IVIS Lumina XR was used to track fluorescence correlating to nanofiber retention at injection sites over the course of 7 days. The dorsal view of each mouse was imaged using 535 nm excitation and a DsRed emission filter. PerkinElmer Living Image software was used to isolate radiant efficiency signals from each injection site and quantify the loss of signal over time. Sections of back skin around the injection site were harvested on Day 7, cryopreserved in O.C.T. (Tissue-Tek), and sectioned for confocal and H&E staining and imaging. All imaging analysis was performed in Imagej and any fluorescence level below the detectable limit was given a value of 0 for graphing purposes.

Statistical Analyses.

All data aren shown in the form of mean +/− standard error of the mean (SEM). Statistical significance was first identified with two-way analysis of variance (ANOVA) followed by Tukey’s post hoc test when all groups were compared, and Dunnet’s post hoc test when all groups were compared to a control. * indicators represent statistical significance at p < 0.05, ** at p < 0.01, *** at p < 0.001, **** at p < 0.0001.

RESULTS

Characterization of randomly polymerized nanofibers.

We generated Q11 peptides appended with randomized sequences of Tyr, Lys, Glu, and Ala using standard Fmoc solid-phase peptide synthesis (sequences in Supplemental Table 1). First we synthesized Q11 via typical sequential couplings, then performed up to 20 additional amino acid couplings using mixtures of the four Fmoc-protected amino acids for each coupling. This produced Q11 peptides N-terminally functionalized with randomized domains of up to 20 amino acids (Figure 1a). This approach allowed for the creation of a single batch of diverse peptide sequences attached to Q11, rather than several different sequences that would later require physical mixing.

We investigated the capacity of different peptide synthesis methods and amino acid ratios to predictably and reproducibly synthesize KEYA₂₀Q11. We initially used a CS Bio 136 peptide synthesizer and equimolar ratios of the four amino acids (Batches A and B, Supplemental Table 2). MALDI mass spectrometry indicated a broad distribution bounded by the lowest possible molecular weight (3416 Da, corresponding to (A)₂₀Q11) and the highest (5258 Da, corresponding to (Y)₂₀Q11) (Figure 1b). The distribution was skewed towards the lower range of the molecular weights, potentially indicating that alanine was favored in the coupling reactions on this instrument.

We next investigated amino acid content and batch-to-batch consistency using amino acid analysis (AAA). After subtraction of Lys and Glu residues contributing to Q11, the percentages of Lys, Glu, Tyr, and Ala were calculated using a MATLAB function (Supplemental Figure 1). We observed that for material produced on the CS Bio 136 peptide synthesizer there was considerable batch-to-batch variability, but batches synthesized using a CEM Liberty Blue microwave-assisted peptide synthesizer had much greater uniformity and fidelity to the ratios of the four reactant protected amino acids (Supplemental Table 2). Five batches were produced with the two different synthesizers and with varying amino acid contents (Batches A-E), and mice were immunized with each of the 5 batches. Despite their varying content, all batches elicited comparable antibody titers and T-cell responses (Supplemental Figure 2), indicating that batches with varying amino acid ratios were nevertheless immunologically similar to each other. Among Batches A-E, Batch B exhibited the most reliable nanofiber formation, so we sought to reproduce this formulation using the CEM Liberty Blue instrument, generating Batches F and G, whose amino acid content corresponded closely to Batch B (Figure 1c). These three batches (B, F, and G) were subsequently used for the remainder of the studies reported below.

By Thioflavin T (ThT) binding (Figure 1d) and atomic force microscopy (AFM, Figure 1e), the (KEYA)₂₀ randomized component appended to Q11 was not found to disrupt β-sheet supramolecular organization during 2 mM nanofiber formation. Furthermore, the addition of the (KEYA)₂₀ component did not affect the modulus or viscosity of the nanofibers (Supplemental Figure 3). Whereas the randomly polymerized peptide p(KEYA)₂₀ did not form discernable nanostructures, (KEYA)₂₀Q11 self-assembled into nanofibers in physiological conditions. This result was consistent with previous observations of other peptides appended to Q11, which has been shown to tolerate a wide range of attached peptide epitopes in its assembly[13,40,41,44,45]. Moreover these nanofibers were stable in physiological conditions and retained β-sheet organization after incubation with murine serum (Supplemental Figure 3).

Previous studies have found size- and composition-dependent cytotoxicity for various other nanomaterials[46], so ruling out such a consideration was an important early step towards utilizing these nanomaterials in vitro or in vivo. (KEYA)₂₀Q11 was not cytotoxic to cultured dendritic cells or macrophages at all working concentrations investigated (0.2–200 μM, Figure 1f). DMSO was used as a technical control, indicating the assay was sensitive enough to detect a dose dependent cytotoxic effect in both cell types.

Optimization of epitope length and nanofiber composition.

We next sought to understand how the KEYA domain’s length and density within the nanofibers influenced immune responses. Previously reported glatiramoids are about 4.7–11 kD in size[26], which is near the upper limit of reliable solid-phase synthesis when appended to the Q11 sequence. Furthermore, because of the polydispersity of the KEYA sequences traditional purification methods were not possible as they would reduce or eliminate the heterogeneity of the sequences. Therefore, we first sought to determine an effective randomized component length within confident synthesis limits, under 40 total amino acids, rather than increase opportunities for truncated sequences to be included in the formulation. Various lengths of the randomly polymerized domain in (KEYA)_nQ11 from n = 1–20 amino acids were synthesized, and their molecular weight distributions were analyzed with MALDI mass spectrometry (full spectra in Supplemental Figure 4, arranged on a single axis for comparison in Figure 2a). The three distinct peaks from (KEYA)₁Q11 correspond to A-Q11, K/E-Q11 (the molecular weights of Lys and Glu were not easily distinguishable), and Y-Q11. Amino acid analysis indicated that the most prevalent amino acid in (KEYA)₁Q11 was glutamic acid (Figure 2b). For (KEYA)₅Q11, (KEYA)₁₀Q11, and (KEYA)₂₀Q11, the MALDI exhibited progressively broader and more diverse mass distributions (Figure 2a), and their amino acid composition progressed from being enriched in glutamic acid to being enriched in alanine (Figure 2b). Mice were immunized with the four (KEYA)_n Q11 lengths and boosted every 2.5 weeks for the duration of the 14-week experiment (Figure 2c). Mice produced high (KEYA)_nspecific IgG antibody titers for n = 10–20 as well as a transient response against (KEYA) Q11, but no humoral response to (KEYA)₁Q11 was elicited (Figure 2c). We hypothesize this was due to the requirement for T-cell epitopes, typically between 8–16 amino acids long, for providing T-cell help to produce a strong B cell response. Previous reports indicated induction of antigen-specific antibody responses to 12–19 mer epitopes[13,14,40,41], and no humoral response to unmodified Q11[44]. We chose 20 KEYA additions as our maximun length for reasons stated above, and note longer sequences could be equally effective but likely unnecessary as they would be trimmed for MHC presentation. Additionally, IgG antibody titers were dominated by the IgG1 subclass when compared to IgG2c (Figure 2d). IgG1 and IgG2c antibodies are associated with Type 2/T_H2/IL-4 and Type 1/T_H1/IFNγ T-cell responses, respectively[47], signifying the production of a strong Type 2/T_H2/IL-4 response to (KEYA)₁₀Q11 and (KEYA)₂₀Q11. Of note, nanofiber formation was critical for humoral responses as mice immunized with p(KEYA)₂₀ did not raise KEYA-specific IgG antibodies (Supplemental Figure 5).

Figure 2. Epitope lengths of n = 10–20 amino acids in (KEYA)_nQ11 are required for immune engagement.

(a) MALDI spectra for (KEYA)_n for n = 1, 5, 10, and 20 exhibit increasing breadth and diversity in mass distributions. (KEYA)₁Q11 is indicated in yellow in each subpanel, (KEYA)₅Q11 in blue, (KEYA)₁₀Q11 in green, and (KEYA)₂₀Q11 in red, corresponding to Batch B. (b) Amino acid compositions for (KEYA)_nQ11 with n = 1–20 amino acid additions indicated on the x-axis. (c) Cartoon indicating the boosting schedule and graph of total IgG antibody titers against the immunizing peptide. (d) Anti-immunizing peptide IgG1 and IgG2c titers for week 14. (e) ELISpot of lymphocytes from draining lymph nodes collected at week 14 and re-stimulated with the immunizing peptide. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by two-way ANOVA with Tukey’s post hoc test. Mean +/− s.e.m. shown. n = 5 experimental replicates per group.

To measure T-cell responses, we harvested the lymph nodes of mice immunized with (KEYA)₁Q11, (KEYA)₅Q11, (KEYA)₁₀Q11, and (KEYA)₂₀Q11 and performed ELISpot. Draining lymph nodes were harvested at week 14, and lymphocytes were re-stimulated with their immunizing peptide in an ELISpot plate. Cytokines released from lymphocytes upon peptide re-stimulation were captured on the plate and developed into spots correlating to the magnitude of the T-cell response, with IL-4 and IFNγ production correlating to a Type 2/T_H2/IL-4 and Type 1/T_H1/IFNγ T-cell phenotype, respectively. (KEYA)₂₀Q11 stimulated significantly greater IL-4-producing lymphocytes than (KEYA)₁Q11 or (KEYA)₅Q11, while all of the groups had comparatively lower numbers of IFNγ-producing cells (Figure 2e). The T-cell results were consistent with the antibody responses in that (KEYA)₁Q11 and (KEYA)₅Q11 were considerably less effective at stimulating lymphocytes. Interestingly, nanofiber formation was necessary for the reduction in IFNγ production, as an ELISpot from mice immunized with p(KEYA)₂₀ had balanced IL4 and IFNγ production (Supplemental Figure 5). Taking this data together, we chose to continue investigating (KEYA)₂₀Q11 because it could be reproducibly synthesized and stimulated both an IgG1 antibody response and Type 2/T_H2/IL-4 biased T-cell production.

We next investigated how the density of (KEYA)₂₀Q11 within nanofibers influenced their uptake in antigen presenting cells, finding that increasing amounts of (KEYA)₂₀Q11 significantly enhanced uptake in dendritic cells, macrophages, and B-cells (Figure 3a). It was straightforward to adjust the amount of (KEYA)₂₀Q11 within Q11 nanofibers by co-assembling varying amounts of Q11 and (KEYA)₂₀Q11 and further including 10% TAMRA-Q11 to fluorescently track the nanofibers. Dendritic cells were cultured with Q11 nanofibers containing 0% (KEYA)₂₀Q11 (unmodified Q11), 1% (KEYA)₂₀Q11, 10% (KEYA)₂₀Q11, or 100% (KEYA)₂₀Q11 for 0.5–24 hours (Figure 3a). Concentrations of 10% and 100% (KEYA)₂₀Q11, which correspond to 20 and 200 μM (KEYA)₂₀Q11 in a 200 μM total formulation with 20 μM TAMRA-Q11 respectively, were acquired by dendritic cells within 2 hours, and almost all dendritic cells had acquired the nanofibers by 24 hours. Macrophages exhibited a similar trend (Figure 3a), but with extremely rapid uptake of 100% (KEYA)₂₀Q11. After only 10 minutes, nanofibers containing 100% (KEYA)₂₀Q11 were acquired by almost 100% of macrophages indicating prompt innate immune cell engagement. This trend was also echoed in the B-cell uptake although over a much longer time course and never reaching 100% uptake (Figure 3a, representative flow cytometry shown in Supplemental Figure 6). In all cases, the highest concentrations of (KEYA)₂₀Q11 considerably accelerated the uptake of the nanofibers into dendritic cells, macrophages, and B cells, corresponding with the enhanced humoral and T-cell responses observed in vivo.

Figure 3. Increasing concentrations of (KEYA)₂₀ in Q11 nanofibers enhances APC acquisition and the strength of humoral responses while maintaining a Type 2/T_H2/IL-4 T-cell phenotype.

Uptake of TAMRA labeled (KEYA)₂₀Q11 in DC2.4 dendritic cells, RAW264.7 macrophages, and B-LCL human B cells stimulated for 0.1, 0.5, 2, 24, and 72 (B cells only) hours measured by flow cytometry (a). 100% refers to the molar percent of (KEYA)₂₀Q11 incorporated into a 0.2 mM Q11 nanofiber. 10% indicates 10% (KEYA)₂₀Q11 and 90% Q11, while 0% indicates the entire nanofiber contains only Q11. All groups included 10% fluorescent TAMRA-Q11. (b) Representative confocal images of fluorescent Q11- and (KEYA)₂₀Q11-stimulated DC2.4 cells. Nanofibers seen in red, cell nuclei in blue, and the actin cytoskeleton in green. Closed arrows show internalization of nanofibers, open arrow shows surface-associated nanofibers. (c) Populations of mouse APCs containing nanofibers collected 12 hours after i.p. injection. (d) Cartoon indicating the boosting schedule and graph of the anti-p(KEYA)₂₀ IgG antibody titers. 50% (KEYA)₂₀Q11 is represented darkest red, 33% (KEYA)₂₀Q11 in the second darkest, 10% (KEYA)₂₀Q11 in the second lightest, and 1% (KEYA)₂₀Q11 in the lightest color. (e) Week 16 IgG1 and IgG2c subclasses. (f) ELISpot results indicate that (KEYA)₂₀Q11 elicits considerable numbers of IL-4-producing spleen-derived lymphoytes and few producing IFN-γ. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by two-way ANOVA with (a) Dunnett’s multiple comparison test (# indicates when nanofibers containing 100% (KEYA)₂₀Q11 were significantly different from 2% and 0%) or (e-f) Tukey’s post hoc test. Mean +/− s.e.m. shown. n = 3–5 experimental replicates per group.

Nanofiber presence inside the cells was visually confirmed by confocal microscopy of dendritic cells (Figure 3b). Representative images show fluorescently labeled Q11 minimally internalized after 24 hours (Figure 3b, closed arrow) while fluorescently labeled (KEYA)₂₀Q11 was abundantly present both inside (Figure 3b, closed arrow) and decorating the surface of cells (Figure 3b, open arrow). Moreover, mice administered fluorescently tagged (KEYA)₂₀Q11 or Q11 intraperitoneally (i.p.) for 2 hours exhibited uptake of the nanofibers into almost 100% of the local macrophages (Figure 3c). About 60% of local dendritic cells also took up (KEYA)₂₀Q11 nanofibers, while only 25% of local dendritic cells internalized Q11, mirroring in vitro studies. By multiple measures, (KEYA)₂₀Q11 improved both the speed and extent of nanofiber uptake into APCs.

To explore how the improved APC uptake may have corresponded to the humoral and cellular responses, mice were immunized with Q11 nanofibers containing 1–50% (KEYA)₂₀Q11 and boosted every 2.5 weeks. The percentages shown in the legend of Figure 3d represent the molar concentration of (KEYA)₂₀Q11 in the total formulation. For example, 10% refers to 0.2 mM (KEYA)₂₀Q11 and 1.8 mM Q11 in a total 2 mM formulation. Mice exhibited a threshold for (KEYA)₂₀-specific IgG antibodies: greater than 10% (KEYA)₂₀Q11 was necessary to stimulate a humoral response (Figure 3d). This suggested that (KEYA)₂₀Q11 is capable of acting as a B-cell epitope at high but not low concentrations, consistent with APC uptake and again hinting at the potential role of epitope proximity on the nanofibers to engage B cells. Examining the IgG subclasses from week 16 serum revealed strong IgG1 biasing from the mice that produced high total IgG titers (Figure 3e), consistent with Type 2/T_H2/IL-4 immune responses. Lymphocytes harvested from the spleen at week 16 were processed and re-stimulated ex vivo with p(KEYA)₂₀ in an ELISpot. All groups produced high numbers of IL-4 spots, but the number of IFNγ spots was significantly lower in the 50%, 33%, and 1% (KEYA)₂₀Q11 groups (Figure 3f), suggesting that (KEYA)₂₀Q11 is a successful Type 2/T_H2/IL-4 T-cell epitope at all concentrations. With this information, the amount of (KEYA)₂₀Q11 in the nanofiber formulations can be titrated to include or exclude B-cell engagement while maintaining a strong Type 2/T_H2/IL-4 phenotype at all concentrations. It is important to note that the ELISpot assay did not have an inherent IL-4 bias, as an IFNγ bias could be evoked by adding CpG to the (KEYA)₂₀Q11 formulation (Supplemental Figure 7a). We have experimentally uncovered two important thresholding responses with these nanomaterials: a required epitope length and a required epitope concentration within nanofibers, both important for eliciting the strongest immune responses.

Enhanced responses to co-assembled T- and B-cell epitopes.

Having characterized (KEYA)₂₀Q11 and its ability to elicit humoral and cellular responses, we next investigated its effects on other epitopes coassembled together into nanofibers with it, exploiting the modularity of self-assembled peptide systems such as Q11[48]. Due to the ability of (KEYA)₂₀Q11 to stimulate T-cell responses even at low concentrations, we hypothesized that (KEYA)₂₀Q11 could influence the magnitude and quality of responses against other co-assembled T-cell epitopes. To test this hypothesis, mice were immunized and boosted once with nanofibers formed by coassembling (KEYA)₂₀Q11 with PADRE-Q11, consisting of Q11 N-terminally modified with the “universal” T-cell epitope, PADRE (Supplemental Table 1). Previously PADRE has been employed to raise strong T-cell responses in a broad range of platforms and applications[49–51], including peptide nanofibers[13]. All groups included 2.5% PADRE-Q11, previously determined to be the most effective concentration within nanofibers [13,41]. Owing to the finding that high concentrations of (KEYA)₂₀Q11 were found to induce antibody responses and low concentrations only stimulated T-cell responses (see above), we explored a wide range of (KEYA)₂₀Q11 in this experiment, studying how high and low concentrations of it influenced the response towards co-assembled PADRE. Nanofibers were prepared containing 97.5% (KEYA)₂₀Q11, 2.5% (KEYA)₂₀Q11, and 0% (KEYA)₂₀Q11 (Figure 4a). Lymphocytes were purified from spleens harvested at 3.5 weeks, and all groups were re-stimulated with both pPADRE (Figure 4a) and p(KEYA)₂₀ (Supplemental Figure 7b) in an ELISpot assay. Upon re-stimulation with pPADRE, the immunizations containing (KEYA)₂₀Q11 elicited more IL-4-producing cells than the immunization without (KEYA)₂₀Q11 (Figure 4a), suggesting that the addition of (KEYA)₂₀Q11 enhanced the response to PADRE-Q11 when co-assembled. IFNγ-secreting cells also increased, but not to a statistically significant degree. Upon re-stimulation with p(KEYA)₂₀, we observed that mice immunized with nanofibers containing (KEYA)₂₀Q11 further maintained their Type 2/T_H2/IL-4 T-cell bias, with statistically higher numbers of IL-4-producing cells than IFNγ-producing cells (Supplemental Figure 7b).

Figure 4. (KEYA)₂₀Q11 enhances other T-and B-cell epitopes.

(a) ELISpot from mice immunized and boosted with nanofiber formulations containing 2.5% PADRE-Q11 and either 97.5% (dark red), 2.5% (light red), or 0% (gray) (KEYA)₂₀Q11 co-assembled with Q11 into a 2 mM nanofiber. The ELISpot shows the spot count from cells re-stimulated with peptide-PADRE. (b) ELISpot from mice immunized and boosted with nanofiber formulations containing 50% NP-Q11 and either 50% (red), or 0% (gray) (KEYA)₂₀Q11 co-assembled with Q11 into a 2 mM nanofiber. The ELISpot shows the spot count from cells re-stimulated with peptide-NP (pNP). (c) Cartoon of immunization schedule for figures (d) and (e). All groups contained 50% TNF-Q11. (d) anti-TNF IgG antibody titers for groups described in (c). (e) anti-TNF IgG1 and IgG2c antibodies. *p<0.05, **p < 0.01, ***p < 0.001 by two-way ANOVA with Tukey’s post hoc test. Mean +/− s.e.m. shown. n = 3–5 experimental replicates per group.

(KEYA)₂₀Q11 was next combined with a different T-cell epitope specific for inducing CD4+ T-cell responses against a peptide epitope from influenza nucleoprotein, here termed NP-Q11. Mice were immunized and boosted once with nanofiber formulations containing 50% NP-Q11 and either 50% (KEYA)₂₀Q11 or 0% (KEYA)₂₀Q11 (Figure 4b). Again, spleens were harvested at week 3.5, and purified lymphocyte populations were used for ELISpot. Lymphocytes were re-stimulated with pNP (Figure 4b) and p(KEYA)₂₀ (Supplemental Figure 7c). NP-Q11 by itself induced only weak T-cell responses, but the addition of (KEYA)₂₀Q11 to the nanofibers significantly enhanced pNP T-cell responses without altering the balance between cells producing IL-4 or IFNγ (Figure 4b). Splenocytes from mice immunized with formulations containing (KEYA)₂₀Q11 additionally responded strongly to re-stimulation with p(KEYA)₂₀ (Supplemental Figure 7c) but without the strong IL4 biasing we had anticipated. As the NP epitope produces an IFN-γ biased response without (KEYA)₂₀Q11 (Figure 4b), this could highlight a limitation of the Type 2/T_H2/IL-4 biasing capabilities of (KEYA)₂₀Q11 in the face of natural epitope tendency. Interestingly, nanofibers bearing the NP epitope elicited greater numbers of cells producing IFN-γ compared to other nanofibers. It was not clear why this was the case overall, but even in nanofibers lacking (KEYA)₂₀Q11, there were greater numbers of cells producing IFN-γ, potentially indicating a bias for this particular epitope and demonstrating that (KEYA)₂₀Q11 may amplify an epitope’s natural bias instead of raising uniformly strong Type 2/T_H2/IL-4 responses for all epitopes. Moreover, when combined with commonly used adjuvants, NP-Q11 conformed to the established biasing of the added adjuvant (Supplemental Figure 8). The adjuvant CpG elicits T_H1-biased responses, and strong NP-specific IFN-γ secretion was produced from the addition of CpG to NP-Q11 (Supplemental Figure 8). Alum commonly elicits T_H2-biased responses, and when combined with NP-Q11 only NP-specific IL4 secretion was produced (Supplemental Figure 8). This demonstrates that the NP-specific immune response can be changed depending on the combined adjuvant and highlights (KEYA)₂₀Q11’s usefulness as a nano-scale adjuvant that maintains natural biases to co-assembled epitopes. Overall these findings indicated that (KEYA)₂₀Q11 is effective for augmenting responses to co-assembled T-cell epitopes, both those that already elicit strong responses by themselves (PADRE) and those that are less immunogenic alone (NP).

An important function of an effective T-cell epitope is to provide help to B cells to produce antibodies. In peptide nanofibers, B-cell epitopes require co-assembled T-cell epitopes to break immune tolerance[52], so we hypothesized that (KEYA)₂₀Q11 could act as an universal T-cell epitope for any co-assembled B-cell epitopes. To evaluate the ability of (KEYA)₂₀Q11 to provide T-cell help in this scenario, we titrated the amount of (KEYA)₂₀Q11 into co-assembled fibers containing 50% TNF-Q11. TNF-Q11 is a peptide B-cell epitope from the soluble form of the TNF protein shown previously to raise antibodies capable of reducing inflammation when a strong B-cell response is raised[13]. Mice were immunized and boosted three times with a final boost occurring 6 days before sacrifice (Figure 4c). Formulations including 37.5% or 50% (KEYA)₂₀Q11 raised antibodies against the peptide epitope (pTNF) by week 5 that remained stable for the duration of the 3-month experiment (Figure 4d). Antibodies were likewise raised against the p(KEYA)₂₀ epitope from the highest two concentrations of (KEYA)₂₀Q11 (Supplemental Figure 7d). The anti-TNF antibody IgG subclasses indicated a polyclonal population consisting of both IgG1 and IgG2c from the groups with high total IgG titers (Figure 4e), whereas the antibody IgG subclasses against p(KEYA)₂₀ remained primarily polarized towards IgG1 (Supplemental Figure 7d). Surprisingly, the lowest concentrations of (KEYA)₂₀Q11 did not raise anti-(KEYA)₂₀Q11 antibody responses, inconsistent with earlier findings (Figure 3). This may have been due to interference with the TNF B-cell epitope, decreasing engagement with T cells by relegating the (KEYA)₂₀Q11 epitope as subdominant. Overall, (KEYA)₂₀Q11 could be used to augment responses against different co-assembled T- and B-cell epitopes, while also continuing to raise responses against the (KEYA)₂₀Q11 component. While responses against (KEYA)₂₀Q11 were clearly polarized towards Type 2/T_H2/IL-4 in the absence of other co-assembled epitopes, we observed a greater range of phenotypes upon the introduction of additional epitopes. This unexpected phenomenon could occur when (KEYA)₂₀Q11 is the subdominant epitope, noted particularly when half the nanofiber formulation consists of a co-assembled epitope (Supplemental Figure 7c–d). However, responses still remained Type 2/T_H2/IL-4 or un-polarized, and not with a dominant Type 1/T_H1/IFN-γ response.

Immune activation is not mediated by inflammation.

To further elucidate why (KEYA)₂₀Q11 is an effective immunomodulatory nanomaterial we investigated the extent to which it may induce inflammatory responses at the site of delivery. While many adjuvants target local APCs and activate the innate immune system by causing inflammation[53], we hypothesized (KEYA)₂₀Q11 would not cause inflammation as it has been previously reported that Q11 nanofibers are not inflammatory[40,41,44]. Mice were injected sub-dermally within the hind footpad with (KEYA)₂₀Q11, Alum, or PBS to determine extent of inflammation caused by the nanomaterial (Figure 5a). Alum is a commonly used adjuvant associated with Type 2/T_H2/IL-4 immune responses that activates complement, eosinophils, and macrophages but can cause local reactions and inflammation[54]. Footpad swelling was measured over the course of 72 hours and normalized to a pre-injection footpad thickness measurement. An increase in footpad diameter was noticed in all groups at 3 hours, likely due to the liquid volume of the injection (Figure 5a). Once the injection volume dissipated, it became clear that Alum caused significantly larger swelling in the footpad and that the inflammation was sustained over the course of the experiment (Figure 5a). In stark contrast, there was no distinguishable swelling caused by the (KEYA)₂₀Q11 injection when compared to the PBS injection (Figure 5a).

Figure 5. (KEYA)₂₀Q11 does not cause injection site inflammation or increase production of inflammatory cytokines.

(a) Footpad swelling measured 3, 6, 12, 24, 48, and 72 hours following a single footpad injection of (KEYA)₂₀Q11 (red), Alum (dark gray), or PBS (light gray) and subtracted from baseline measurements. (b) Schematic for i.p.stimulation and lavage with (KEYA)₂₀Q11 (red), Q11 (blue), PBS (light gray), PBS+LPS (dark gray). (c) Inflammatory cytokine production following 4-hour i.p. stimulation. (d) Non-inflammatory cytokine production following 12-hour i.p. stimulation. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by two-way ANOVA with Tukey’s post hoc test. Mean +/− s.e.m. shown. n = 3–5 experimental replicates per group.

Delving further into the potential inflammatory response, mice were injected intraperitoneally with (KEYA)₂₀Q11, Q11, PBS, or PBS + lipopolysaccharide (LPS), and a multiplex cytokine analysis was performed on the lavage fluid (Figure 5b). Inflammatory cytokines IL1β, IL6, and IFNγ were elevated in response to the LPS injection but remained indistinguishable from PBS after (KEYA)₂₀Q11 or Q11 injections (Figure 5c). Interestingly, the anti-inflammatory cytokines IL-4 and IL-5 were significantly elevated only after (KEYA)₂₀Q11 injections, as compared with PBS (Figure 5d). IL-4 and IL-5 are often produced by Th2 cells and influence the proliferation and differentiation of T and B cells[55]. The remaining cytokines examined in the multiplex were not significantly different from PBS (Supplemental Figure 9). Taken together, the evidence supported the use of (KEYA)₂₀Q11 as a non-inflammatory adjuvant with the ability to stimulate IL-4- and IL-5- producing cells in vivo.

Elevated IL-4 production after (KEYA)₂₀Q11 immunizations.

Having established that (KEYA)₂₀Q11 does not mediate immune responses through inflammation, we next investigated the effect of (KEYA)₂₀Q11 on T-cell activation and proliferation as well as the IL-4 production of different effector T-cell populations. Mice were immunized and boosted twice with (KEYA)₂₀Q11, Q11, or PBS (Figure 6a) and their spleens were harvested 5 days after the last boost. After the splenocytes were stimulated overnight with their immunizing peptide, cells were stained and analyzed by flow cytometry. There was no difference in the total percent of CD3+ cells between groups (Figure 6b), but the cells stimulated with (KEYA)₂₀Q11 produced a significantly higher percentage of IL-4+CD3+ cells (Figure 6c). This result suggested that (KEYA)₂₀Q11 had no effect on the general proliferation of T cells but instead influenced their differentiation into IL-4 producing T cells. Moreover, there was a slight reduction in CD4+ T-cell number after (KEYA)₂₀Q11 stimulation (Figure 6d), but again, an increase in the percent of IL-4+CD4+ T cells(Figure 6e) and no significant difference in the number of CD8+ T cells between groups (Supplemental Figure 10). The CD4+ T effector population trend is comparable to the CD3+ T-cell population in terms of total T-cell production and IL-4 producing T cells, suggesting that (KEYA)₂₀Q11 was potentially only stimulating the expansion of specific IL-4 producing T-cell populations, at the expense of other T-cell populations. This data is also consistent with the previous ELISpot findings of strong IL-4 production from (KEYA)₂₀Q11 stimulated lymphocytes (Figures 2, 3, 4). Additionally, it was apparent that (KEYA)₂₀Q11 and even Q11 alone significantly increased the production of regulatory T cells (Tregs, CD4⁺CD25^hi) (Figure 6f). This analysis of the T-cell populations activated by the nanofibers was further evidence that (KEYA)₂₀Q11 is a strong Type 2/T_H2/IL-4 polarizing nanomaterial. Moreover, it was clear that (KEYA)₂₀Q11 did not influence expansion of T-cell populations, potentially beneficial in cases of inflammation. Rather, (KEYA)₂₀Q11 polarized the population toward a Type 2/T_H2/IL-4 T-cell phenotype, and nanofibers increased the production of Tregs over PBS immunizations.

Figure 6. Immunizations with (KEYA)₂₀Q11 lead to increased IL-4 production in CD4+T cells.

(a) Cartoon of s.c. immunization schedule with either (KEYA)₂₀Q11 (red), Q11 (blue), PBS (gray). Spleens harvested at week 6. Percent of (b) CD3+ cells and (c) CD3+IL-4+ cells from total number of live cells. Percent of (d) CD4+ cells and (e) CD4+IL-4+ cells from total number of CD3+ cells. (f) Percent of CD4+CD25hi cells from total number of CD3+ cells. *p < 0.05, ***p < 0.001, ****p < 0.0001 by one-way ANOVA with Dunnett’s post hoc test. Mean +/− s.e.m. shown. n = 4–5 experimental replicates per group.

Persistence at the injection site.

T-cell activation hinges on uptake and presentation of nanofiber components by APCs, so we hypothesized that high levels of APC uptake (Figure 3a) of (KEYA)₂₀Q11 were responsible for the strong T and B-cell response observed above. In subcutaneous immunizations, uptake by APCs occurs primarily at the injection site, so IVIS was used to image mice injected with fluorescently labeled nanofibers over the course of a week. Mice were injected with 1.8 mM Q11 co-assembled with 0.2 TAMRA-Q11 on their left flank and 1.8 mM (KEYA)₂₀Q11co-assembled with 0.2 mM TAMRA-Q11 on their right (Figure 7a), and images were taken daily to measure the retention of each nanomaterial at the injection site. Fluorescently labeled (KEYA)₂₀Q11 persisted an average of 5 days, whereas fluorescently labeled Q11 was no longer measurable after an average of 3 days (Figure 7b). The apparent flurorescence in the head of the mouse (Figure 7a) is due to background fluorescence of the nose cone and was not included in analysis. (KEYA)₂₀Q11 has a moderately negative surface charge by zeta potential (Supplemental Figure 11), so to determine whether surface charge influenced the retention time, we also compared surface-charge-modified Q11 nanofibers (see methods), including both K3-Q11 and E3-Q11, but these had no increased retention time compared to Q11 alone (Supplemental Figure 11). This implied that additional factors were responsible for the increased retention of (KEYA)₂₀Q11. The more intense radiant efficiency (Figure 7c) also indicated greater amounts of (KEYA)₂₀Q11 were retained at the site compared with Q11 alone.

Figure 7. (KEYA)₂₀Q11 persists at the injection site and maintains immunogenicity for up to 7 days.

Mice injected with TAMRA labeled Q11 on the left flank and (KEYA)₂₀Q11 on the right were (a) measured with IVIS immediately after injection and daily for 7 days. (b) The number of days the fluorescence was detected above background established by the control mouse is shown to compare the persistence between the two groups. (c) The radiant efficiency was calculated from the images collected and is shown over the course of 7 days. (d) Total amount of remaining nanofiber and (e) overlap between nanofibers and CD45+ cells calculated from confocal images. Undetermined percentages were assigned a value of 0. (f) Representative confocal images from injection site skin collected on Day 7. * p < 0.05 by unpaired t test (b, d, e) and two-way ANOVA (c), n = 5.

Skin sections were collected from the injection site on day 7, and immunofluorescence microscopy was used to image CD45, a common lymphocyte marker (Figure 7d–f). The total amount of fluorescently labeled nanofibers remaining was also quantified by exclusively analyzing the injection site area for TAMRA+ signal (Figure 7d), and it was confirmed that more (KEYA)₂₀Q11 remained at the injection site when compared to Q11. Moreover, about 6% of the nanofibers overlapped with lymphocytes (Figure 7e), indicating interaction between nanofibers and lymphocytes. Representative images are shown with their corresponding H&E stained section (Figure 7f), demonstrating considerable cellular infiltration of the injection site material and signifying little to no capsule formation around the material. Furthermore, evidence of CD45⁺ cells at the injection site suggested ongoing immunogenicity of (KEYA)₂₀Q11 over the course of the weeklong study. Clearly, the addition of the randomized (KEYA)₂₀Q11 component to the nanofibers increased retention time at the injection site, critical for maximum APC uptake and downstream immunomodulation. Moreover, a lack of capsule formation allows for full infiltration of the injected material allowing (KEYA)₂₀Q11 to continue to interact with lymphocytes and maintain immunogenicity for the duration of the experiment.

DISCUSSION

It has been established that while peptide-based immunotherapeutics can have finely directed specificity for chosen epitopes, they generally lack sufficient immunogenicity to provoke suitable immune responses. Therefore, it has been an important goal to develop vaccine components with the capacity to augment peptide immunogenicity while minimizing any inflammatory side effects. Here, we developed a novel immune-modulating nanomaterial, (KEYA)₂₀Q11, with Type 2/T_H2/IL-4 immune properties and the ability to enhance the response to other co-assembled peptide epitopes. This property arose from both the glatiramoid-inspired (KEYA)₂₀ segment as well as its ability to self-assemble into nanofibers. (KEYA)₂₀Q11 was reproducibly synthesized and maintained β-sheet supramolecular assembly characteristics when fibrilized in physiological conditions. (KEYA)₂₀Q11 also strongly activated APCs, induced humoral antibody responses at high concentrations, and induced Type 2/T_H2/IL-4 phenotypes at all concentrations. The dose-dependent behavior of (KEYA)₂₀Q11 echoed previous findings where epitope content and ratios between T- and B-cell epitopes modulated the strength and phenotype of the response raised[13,41,56,57]. More importantly, this result indicated that (KEYA)₂₀Q11 can be used at various concentrations within nanofibers to modulate B-cell responses, adjusting them depending on the intended application, which could be useful for several applications in which antibody responses are undesirable. For example, the development of antidrug antibodies to adalimumab and other monoclonal antibody therapies can be associated with long term failure of treatment[58,59]. Another relevant example is the potential for antibody dependent enhancement (ADE) of viral entry of flaviviruses[60] and coronaviruses[61] into cells following vaccination. Thus, employing low concentrations of (KEYA)₂₀Q11 may present an opportunity to exclusively promote a cellular response upon vaccination.

Upon co-assembly with additional T-cell epitopes, (KEYA)₂₀Q11 augmented T-cell responses without changing the Th1/Th2 polarization raised by these epitopes themselves; when mixed with B-cell epitopes, (KEYA)₂₀Q11 promoted B-cell responses without needing to add additional T-cell epitopes such as PADRE. The ability of (KEYA)₂₀Q11 to increase cellular responses while maintaining the co-assembled epitope’s inherent biasing capabilities makes (KEYA)₂₀Q11 a useful and novel nanoscale adjuvant. We hypothesize that this capability could be due to (KEYA)₂₀Q11 acting as a sub-dominant epitope and envision future experiments to test this hypothesis. Additionally, we observed that (KEYA)₂₀Q11 provided the strong T-cell epitope help required to produce a humoral response to an otherwise nonimmunogenic B-cell epitope. It is significant to note that when using (KEYA)₂₀Q11 in an adjuvant-like context, no discernible inflammation was present at the injection site. This is unlike commonly used adjuvants such as Alum. Moreover, we demonstrated the effectiveness of (KEYA)₂₀Q11 as a co-delivered nanoscale adjuvant for two disease-specific epitopes, influenza and anti-TNF, with the implication that (KEYA)₂₀Q11 could be applied in a variety of contexts.

The concentration of (KEYA)₂₀Q11 used throughout the manuscript is variable because when co-assembled with other epitopes the formulation is generally dependent on the co-assembled epitope concentration. For example, B-cell epitopes generally require concentrations of 50% or greater in the nanofiber forumlations to be effective, so the (KEYA)₂₀Q11 concentration in that instance could be no higher than 50%. A variety of concentrations was explored when (KEYA)₂₀Q11 was used alone. This highlights the versatility of (KEYA)₂₀Q11 to provide benefit in a variety of conditions.

The complete mechanism of immune modulation by (KEYA)₂₀Q11 remains to be fully elucidated, yet the present work suggests a number of components of this full mechanism. Studies have shown that APLs, minor alterations to epitopes familiar to immune cells, can shift the phenotype of the immune response in an anti-inflammatory direction[31,62,63], potentially due to low binding strength between the epitope and the TCR. Others have postulated that glatiramoids act as a universal APL, presenting millions of sequences to immune cells who in turn partially recognize a subset of these sequences and activate an anti-inflammatory response[64]. Given the glatiramoid-like composition of (KEYA)₂₀Q11, these processes may be relevant to the peptide nanofibers reported here as well. We hypothesize when co-assembled with other epitopes, (KEYA)₂₀Q11 could be assuming a sub-dominant role through weakly binding APLs, allowing the other, normally weak epitopes to be contextually dominant and thus amplifying their immune response[65]. While we provide evidence for the ability of (KEYA)₂₀Q11 to stimulate expansion of IL-4 producing T cells as well as Tregs, the underlying mechanisms driving these responses remain to be articulated. We can, however, speculate on reasons for the immunogenicity of (KEYA)₂₀Q11. The ability of (KEYA)₂₀Q11 to persist at the subcutaneous injection site for extended periods of time would give it further opportunities to engage local APCs, increasing those traveling to draining lymph nodes and stimulating resident T and B cells there. Furthermore, persistence at the inject site may allow the primed T and B cells in the lymph node to migrate back to the injection site and be directly activated by the remaining epitopes, reinforcing their Type 2/T_H2/IL-4 polarization.

Future studies aimed at tracking nanofiber movements throughout the body could help to increase understanding of the mechanisms by which (KEYA)₂₀Q11 operates. Moreover, further modulation of the identity and concentrations of multiple co-assembled epitopes could be investigated to continue to test the hypothesis regarding the sub-dominance of glatiramoid-like epitopes. Overall, supramolecular glatiramoid materials such as (KEYA)₂₀Q11 may have potential applications in vaccine development, towards infectious diseases and towards non-infectious applications such as inflammatory autoimmune diseases, wound healing, or graft rejection.

CONCLUSIONS

We investigated the immunogenicity of a glatiramoid-inspired component within a supramolecular platform. We characterized both the reproducibility of the synthesis process and the non-inflammatory immune response raised upon vaccination in mice. (KEYA)₂₀Q11 was found to be a potent nanoscale adjuvant that enhances responses to a broad range of co-assembled epitope peptides. (KEYA)₂₀Q11 was also capable of increasing injection site retention and APC uptake, properties critical for augmenting T- and B-cell activation. These findings indicate that this supramolecular glatiramoid-inspired platform may be a promising strategy for increasing the immunogenicity of peptide-based therapeutics.

Data Availability

All data are either provided in the figures, supplementals or are otherwise available upon request.