Vaccine Adjuvant Incorporation Strategy Dictates Peptide Amphiphile Micelle Immunostimulatory Capacity
Current vaccine research has shifted from traditional vaccines (i.e., whole-killed or live-attenuated) to subunit vaccines (i.e., protein, peptide, or DNA) as the latter is much safer due to delivering only the bioactive components necessary to produce a desirable immune response. Unfortunately, subunit vaccines are very weak immunogens requiring delivery vehicles and the addition of immunostimulatory molecules termed adjuvants to convey protective immunity. An interesting type of delivery vehicle is peptide amphiphile micelles (PAMs), unique biomaterials where the vaccine is part of the nanomaterial itself. Due to the modularity of PAMs, they can be readily modified to deliver both vaccine antigens and adjuvants within a singular construct. Through the co-delivery of a model antigenic epitope (Ovalbumin319–340—OVABT) and a known molecular adjuvant (e.g., 2,3-dipalmitoyl-S-glyceryl cysteine—Pam2C), greater insight into the mechanisms by which PAMs can exert immunostimulatory effects was gained. It was found that specific combinations of antigen and adjuvant can significantly alter vaccine immunogenicity both in vitro and in vivo. These results inform fundamental design rules that can be leveraged to fabricate optimal PAM-based vaccine formulations for future disease-specific applications.
KEY WORDSAdjuvant Co-localization Peptide amphiphile micelles Subunit vaccines
Vaccines have become a cornerstone of human health and disease prevention (1, 2, 3). Whole pathogen vaccines consisting of killed or inactivated infectious agents are the most commonly used formulations in the clinic. Despite the considerable efficacy achieved with these vaccines, they unfortunately can be associated with a number of deleterious side effects. Cases of injection site inflammation and unwanted host reactions along with storage difficulty and arduous production processes make traditional whole pathogen vaccines increasingly less appealing as novel alternatives emerge (4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15).
Within a whole pathogen vaccine, only certain components are directly targeted by the host immune response. These constituents termed antigens are most commonly peptides which lack the complexity of the entire pathogen but can facilitate a protective host response by themselves. Unfortunately, these subunit peptide vaccines have been found to be very weak immunogens since they lack the foreign immunostimulatory components found within whole pathogens which better stimulate host immune responses against the antigens. Thus, an effective delivery vehicle is required as a compensatory means for maximizing the prophylactic effects of peptide vaccines (16, 17, 18, 19).
Peptide amphiphile micelles (PAMs) have emerged as a promising vaccine carrier capable of inducing strong and durable prophylactic antibody responses (20,21). Our recent work has uncovered that certain physical properties of PAM vaccines including size and charge greatly influence their efficacy. Specifically, spherical and short cylindrical PAMs tens of nanometers in size with near neutral surface charge were found to best enhance antigen immunogenicity (22,23). While promising, PAMs alone may not induce strong enough host immune responses to be protective. Therefore, co-delivering molecular adjuvants, compounds with known immunostimulatory behavior, with antigen-based PAMs has the potential to yield a novel synthetic vaccine formulation with potent bioactivity. To test this theory, this paper focuses on creating and evaluating PAMs comprised of the model antigen OVABT-(KE)4 and the toll-like receptor 2 (TLR-2) agonist 2,3-dipalmitoyl-S-glyceryl cysteine (Pam2C) as these molecules have shown potent antigenicity and adjuvanticity, respectively (21, 22, 23, 24, 25, 26). From these efforts, new design rules which can be leveraged for the creation of future disease-specific PAM vaccines will hopefully be determined.
MATERIALS AND METHODS
Peptide and Peptide Amphiphile Synthesis, Purification, and Characterization
OVABT peptide (ESLKISQAVHAAHAEINEAGRE) with an additional zwitterion-like repeat (KE)4 added to the C terminus (ESLKISQAVHAAHAEINEAGREKEKEKEKE) was synthesized on rink amide resin (Chem-Impex International, SC Wood Dale, IL) by solid-phase synthesis on a multiple peptide synthesizer (Advanced ChemTech 396 Omega, Louisville, KY) using Fmoc chemistry. The N terminus was then either acetylated with acetic anhydride or covalently coupled to 5,6-carboxyfluorescein (FAM, EMD millipore) to cap or fluorescently label the peptide, respectively. OVABT peptide amphiphile (PA) was synthesized similarly but Fmoc-Lys(Fmoc)-OH was added to the N terminus of OVABT on resin after which the Fmoc groups were deprotected with 25% piperidine in dimethylformamide (DMF). The two primary amines were then conjugated with palmitic acid (Palm) using a 1:5:4.2:10 Palm:HOBT:HBTU:DIPEA ratio in n-methyl-2-pyrrolidone (NMP). Fluorophore-labeled OVABT PAs were synthesized similarly except an additional Fmoc-Lys(ivDDE)-OH was positioned between the Fmoc-Lys(Fmoc)-OH and N terminus of the OVABT-(KE)4 peptide on resin. The ivDDE group was deprotected by 2% hydrazine in DMF allowing for FAM attachment to the primary amine side group. Adjuvant templated OVABT PAs were synthesized similarly as previously described except Pam2C instead of Fmoc-Lys(Fmoc)-OH was attached to the N terminus of OVABT on resin. All peptides and PAs were cleaved from resin and their side groups deprotected by a single reaction consisting of 2 h of exposure to the following mixture: TFA, thioanisole, phenol, water, ethanedithiol, and triisopropylsilane (87.5:2.5:2.5:2.5:2.5). Precipitation and washing with diethyl ether yielded crude products which were characterized and purified by mass spectrometry aided semi-preparative high-pressure liquid chromatography (HPLC, Beckmann Coulter, Fullerton, CA) using a C4 or C18 column (Milford, MA) and in-house solvent gradients. Pam2C-SK4 was purchased from InvivoGen (San Diego, CA) and 5(6)-carboxytetramethylrhodamine (TAMRA) modified Pam2C-SK4 was synthesized in-house according to the previously mentioned solid-phase synthesis technique with TAMRA attached to the N terminus of Pam2C-SK4 while the PA is still on resin. Similar to our previous work, micelle formation, morphology, size, and secondary structure were assessed by critical micelle concentration (CMC), transmission electron microscopy (TEM), dynamic light scattering (DLS), and circular dichroism (CD), respectively (22,23). Förster resonance energy transfer (FRET) was conducted similarly to a previously described protocol (24). In brief, 10% adjuvant supplemented peptide was formulated by directly mixing product solutions yielding 36 μM 2.1% FAM-labeled OVABT peptide with 4 μM TAMRA-labeled Pam2C-SK4. To form 10% adjuvant associated PAMs, methanol solubilized, air dried, and rehydrated mixture yielding 36 μM 2.1% FAM-labeled OVABT with 4 μM TAMRA-labeled Pam2C-SK4 heterogeneous PAMs. Single fluorophore-labeled monomers at the same concentrations (i.e., 0.756 μM for FAM or 4 μM for TAMRA) were included as controls. Fluorescence spectra were collected using a Cytation 5 fluorospectrophotometer for which laser excitation was set at 450 nm and emitted light was collected from 475 to 700 nm.
Preparation and Activation of Bone Marrow-Derived Dendritic Cells
Nine Different Vaccine Formulations used for the In Vitro BMDC Activation Study
BMDC stimulus groups
1.8 μM OVABT-(KE)4
1.8 μM Palm2K-OVABT-(KE)4
10% adjuvant supplemented peptide
1.8 μM Palm2K-OVABT-(KE)4 + 0.2 μM Pam2C-SK4
10% adjuvant templated PAMs
1.6 μM Palm2K-OVABT-(KE)4/0.2 μM Pam2C-OVABT-(KE)4
100% adjuvant templated PAMs
1.8 μM Pam2C-OVABT-(KE)4
0.2 μM Pam2C-OVABT-(KE)4
0.2 μM Pam2C-SK4
Four Different Vaccine Formulations Used for the In Vivo Immunization Experiment
10% adjuvant supplemented peptide
200 nmol OVABT-(KE)4 + 22.2 nmol Pam2C-SK4
100 nmol OVABT-(KE)4 + 11.1 nmol Pam2C-SK4
10% adjuvant associated PAMs
200 nmol Palm2K-OVABT-(KE)4/22.2 nmol Pam2C-SK4
100 nmol Palm2K-OVABT-(KE)4/11.1 nmol Pam2C-SK4
10% adjuvant templated PAMs
177.8 nmol Palm2K-OVABT-(KE)4/22.2 nmol Pam2C-OVABT-(KE)4
88.9 nmol Palm2K-OVABT-(KE)4/11.1 nmol Pam2C-SK4
100% Adjuvant Templated PAMs
200 nmol Pam2C-OVABT-(KE)4
100 nmol Pam2C-OVABT-(KE)4
Antibody Response Characterization
High binding, 96-well ELISA plates (Santa Cruz Biotechnology) were coated overnight with 4 μg/mL OVABT peptide in PBS. Wells were washed with PBS-T (0.05% Tween-20 in PBS) and blocked with 10% FBS in PBS (blocking buffer) for 1 h. Serum was serially diluted twofold in blocking buffer across the plate and incubated for 2 h. Wells were then washed with PBS-T and incubated with 1:3000 diluted detection antibody for 1 h. After additional washing with PBS-T, wells were incubated for 30 min with 100 μL TMB substrate (Biolegend) and optical density (OD) was measured at 650 nm absorbance using a Biotek Cytation 5 spectrofluorometer. End-point antibody titers were defined as the greatest serum dilution where ELISA OD was at least twice that of serum from mice vaccinated with PBS. If end-point titers were not reached with one plate, then additional titrations were utilized until ODs were diluted below detection.
Lymphocyte Isolation, Antigenic Challenge, and Stimulus Assessment
Mice were sacrificed 16-week post-primary vaccination after which draining lymph nodes and spleens from immunized mice were collected and grinded with a cell strainer pestle. Spleen cells were further treated with ACK lysis buffer in order to remove red blood cells. Single-cell suspensions of lymph node cells or ACK lysis buffer treated spleen cells were further prepared by filtering through a 70-μm nylon mesh cell strainer. Cells were plated in 96-well tissue culture treated plates at 1.7 × 105 cells/well and stimulated with 25 μg/mL OVABT peptide. After 72 h of incubation, cell culture supernatants were collected and stored at − 80 °C until further analyzed. Cell culture supernatants of pooled samples were screened by a multiplex cytokine kit (Biolegend) to determine any cytokine differences among the vaccination groups. Based on this screening, individual samples were analyzed for their IL-2, IFN-γ, and TNF-α concentrations by cytokine-specific ELISA kits (Biolegend).
JMP software (SAS Institute) was used to make comparisons between groups where an analysis of variance (ANOVA) was performed followed by Tukey’s HSD testing to determine pairwise statistically significant differences (p ≤ 0.05).
Intrinsic Non-immunostimulatory Behavior of PAMs
Antigen|Adjuvant Co-localization Affects PAM Immunogenicity
Without underlying immunostimulatory effects, PAM immunogenicity can be readily enhanced through the co-delivery of known molecular adjuvants. While this could be accomplished by simply mixing PAM and molecular adjuvants immediately before immunization, recent research has demonstrated that direct antigen|adjuvant co-localization can greatly improve vaccine efficacy (17,30, 31, 32, 33, 34, 35, 36, 37). Specifically, Pam2C has been shown to be able to co-localized with antigens via different methods including through hydrophobic association (24) or chemical conjugation (38, 39, 40). Though both strategies have shown promising outcomes, their differential impact on antigen immunogenicity has yet to be determined leaving open the question of which one is the optimal antigen/Pam2C co-localization strategy.
Antigen|adjuvant co-localization was further explored by comparing the efficacy of two different co-localization methodologies (i.e., adjuvant associated–hydrophobic driving force; adjuvant templated–covalent coupling). The data shows that 10% adjuvant templated PAMs induced significantly lower antibody titer IgG than 10% adjuvant associated PAMs at most time points post-boost vaccination (i.e., weeks 6, 10, 12, and 14). Even when the Pam2C adjuvant dose was increased by ninefold (i.e., 100% adjuvant templated PAMs), no improvement in antibody production was observed. IgG subtypes (i.e., IgG1, IgG2a, and IgG3) were similarly evaluated 2-week post-boost vaccination (i.e., week 6) (Fig. S2). Interestingly, IgG1 and IgG3 showed a similar trend as total IgG. For IgG2a, 10% adjuvant templated PAMs induced no detectable antibody production indicating that different co-localization strategies can alter antibody subtype polarization.
Antigen|Adjuvant Co-localization Influences Cellular Responses
Antigen|Adjuvant Co-localization Strategies Affect Pam2C Adjuvanticity
Though PAMs have shown tremendous promise as self-adjuvanting vaccine delivery vehicles (21,24,46), there is still much to be learned about which aspects of their design influence their immunogenicity. Our recent efforts have shown that PAM size and charge can be readily altered and this directly enhances or suppresses host immune responses to incorporated peptide antigen (22,23). Specifically, the most potent PAM formulation (i.e., Palm2K-OVABT-(KE)4) was found to possess the greatest capacity to cross multiple biological barriers including trafficking to the draining lymph nodes and being uptaken by APCs. While exciting, it is unknown whether PAMs possess any intrinsic immunostimulatory properties. Previous research has demonstrated that PAMs do not directly stimulate TLR-2 receptors (21). Though interesting, other potential pathways (e.g., other TLRs, NOD-like receptors, and RIG-like receptors) could be activated by PAMs. In order to evaluate the immunostimulatory capacity of PAMs, their capacity to generally activate APCs was explored. Remarkably, PAMs alone failed to significantly stimulate BMDCs compared to the potent response seen with a known molecular adjuvant (i.e., Pam2C) as evidenced by a lack of cell surface co-stimulatory marker expression changes (Figs. 1 and 4). Therefore, previously reported PAM immunogenicity appears to be a more related to its targeted delivery capacity (22) instead of an innate ability to activate APCs.
The lack of intrinsic immunostimulatory capacity allows for PAM vaccine immunogenicity to be enhanced through the incorporation of molecular adjuvants. Previously, the hydrophobic association of Pam2C-SK4 with PAMs at a 90/10 antigen/adjuvant molecular ratio was found to significantly enhance immunogenicity over adjuvant supplemented peptide and PAMs alone for group A Streptococcus peptide antigen (24). When a similar formulation was utilized in this work (i.e., 10% adjuvant associated PAMs), comparable improvements in antigen-specific antibody induction and isotype production were observed (Fig. 2 and Fig. S2). Cellular responses complemented these results as lymph nodes and spleens from mice vaccinated with 10% adjuvant associated PAMs possessed lymphocytes capable of producing a desirable cytokine profile in response to antigen re-stimulation (Fig. 3). Taken together, these results support the concept that adjuvant/antigen co-localization using PAM vaccines can be a powerful approach for improving subunit vaccination.
Although exciting, hydrophobic association is not the only method available for achieving adjuvant incorporation. Covalent coupling antigens to Pam2C have been previously shown to enhance peptide immunogenicity (40,47) and the chemical similarities between Pam2K and Pam2C potentially allow for direct adjuvant micelle templating. The results support this idea as Pam2C-OVABT-(KE)4 readily self-assembled in water at a low CMC (0.32 μM—Fig. S2a) into spherical and short cylindrical micelles (Fig. S2b) presenting peptide mostly in the β sheet conformation (95.0%—Fig. S2c). These micellar factors are very similar to those previously identified for Palm2K-OVABT-(KE)4 (e.g., 0.20 μM CMC and 91.4% β sheet content) eliminating some potentially confounding variables between the formulations (23). Interestingly, adjuvant coupling yielded a vaccine formulation (i.e., 10% adjuvant templated PAMs) with a weakened capacity to induce OVABT-specific antibody production (Fig. 2 and Fig. S3) and antigen sensitive lymphocytes (Fig. 3) compared to 10% adjuvant associated PAMs. This limitation was not even able to be overcome by increasing adjuvant content ninefold through the use of only antigen-adjuvant PAs (i.e., 100% adjuvant templated PAMs). The vaccination response with both adjuvant templated PAM formulations was found to actually be quite similar to 10% adjuvant supplemented peptide.
The diminished immunogenicity caused by antigen-adjuvant coupling, especially when a greater quantity of adjuvant was used, was a quite surprising result. In order to further investigate why different co-localization strategies so greatly impacted immunogenicity, additional APC activation studies were conducted. The results revealed that both adjuvant templated PAM formulations induced less MHC II and CD86 surface marker expression on BMDCs compared to cells exposed to 10% adjuvant supplemented peptide and 10% adjuvant associated PAMs (Fig. 4b, d). The fundamental difference in these formulations is which peptide is tethered to Pam2C. Additional in vitro BMDC stimulation assessment using just Pam2C-SK4 and Pam2C-OVABT-(KE)4 (Fig. S4b, d) support these results, indicating the peptidic component of the Pam2C amphiphile influences its immunogenicity. Although PAM vaccine size has been shown to play an important role in influencing the host antibody response (22), immunogenicity differences observed for adjuvant associated PAMs and adjuvant templated PAMs were likely due to adjuvant incorporation method as they were found to be similar in size (Table S1).
The diminished or lack of immunostimulatory behavior seen with another formulation (i.e., adjuvant templated PAMs) likely stems from the requirement for proper TLR2 receptor-agonist binding. Protein crystallography has revealed that the lipid binding pocket in TLR2 and agonist association with TLR6 dictates receptor activation is dependent on a few key chemical features (48). Though Palm2K, as well as the previously utilized diC16, possess considerable similarities to the known TLR2 agonist Pam2C, their slight differences are quite important when it comes to stimulating the TLR2 receptor. The hydrophobic binding cleft of TLR2 is specifically designed for palmitoyl moieties that are separated by two hydrocarbons (48). While Pam2C satisfies this requirement, Palm2K possesses a four hydrocarbon spacer. This additional length alters the protein residue pocket alignment which is necessary to stabilize the lipid yielding diminished binding capacity similar to what has been shown with stearic acid modified glycerylcysteine (49). The presence of the thioether in Pam2C is also important as it has been shown to bind multiple residues in TLR2 as well as a residue in TLR6 helping to stabilize protein dimerization (48). The replacement of this with an ether significantly diminishes binding (50). Additionally, palmitoyl binding chemistry plays an important role in agonist-protein binding. While the replacement of one ester with an amide only partially inhibits agonist binding, replacing both groups, like what is done with Pam2K, completely prevents association (50). Therefore, the lack of APC activation by Paml2K is likely due to its inability to function as a TLR2 agonist.
The reduced immunostimulatory behavior found when Pam2C is bound to OVABT-(KE)4 instead of SK4 is probably caused by similar binding changes. Though the hydrophobic moiety is the same in this case, research has shown that agonist peptide sequence plays an important role in TLR2 activation as well (51,52). The presence of tetralysine has been shown to enhance binding fivefold (50), so its absence likely will at least somewhat diminish Pam2C bioactivity. More impactful though is the presence of the N-terminal serine which has been shown to undergo hydrogen bonding with a carbonyl group in the protein backbone of TLR2 (48). The lack of this group has been shown to diminish binding four to five orders of magnitude (50). The N-terminal glutamic acid of OVABT-(KE)4 is too long to facilitate this necessary hydrogen bonding likely greatly impacting its adjuvanticity. In addition, directly linking the antigen and the adjuvant may diminish their respective effects. Recent research has shown that decoupling antigen and adjuvant can actually maintain or improve vaccine immunogenicity (53,54). One rationale behind this theory is that antigens and adjuvants carry out their functions in different places in the APC. While TLR agonist adjuvants must interact with their corresponding TLR on the cell surface membrane (55, 56, 57) or early endosome (58, 59, 60), antigens need to be transported into late endosome or lysosomes and cleaved into small fragments before being presented by MHC II molecules on the cell membrane (61). Covalently tethering antigen and adjuvant together into one PA may require each biomolecule to carry out only a singular function. Therefore, while co-localizing antigen and adjuvant into a single PAM is attractive for delivering both molecules to the same APC, having the capacity to readily dissociate these from one another may allow for each to function optimally.
While previous studies have shown that PAMs can be utilized to improve subunit vaccine efficacy (21,22,24), design rules associated that govern this behavior are sorely lacking. This work expands on previous research, revealing that PAM immunogenicity is likely dictated by their targeted trafficking ability (i.e., lymph node accumulation and APC uptake) instead of directly stimulating APCs (e.g., by interactions with toll-like receptors or mannose receptors). PAM immunogenicity can be further enhanced through the co-localization of a molecular adjuvant (i.e., Pam2C). Interestingly, the method by which the adjuvant is incorporated was found to make an appreciable difference in peptide immunogenicity. Specifically, hydrophobic association was found to enhance both antibody and cellular responses over adjuvant supplemented peptide whereas covalent tethering showed no improvement even when the quantity of adjuvant delivered was greatly increased. Upon further analysis, these results correspond to established molecular features that govern receptor agonist activity. The results reported help inform future adjuvant incorporated PAM formulation design by allowing for better and more rapid optimization of this platform technology for a variety of different types of vaccines. Further studies are in necessary to create a more comprehensive tool box for Pam2C incorporation including the use of different peptide sequences with modified conditions (e.g., dose sparing, different peptide epitopes, and stability). These together would allow for the establishment of more comprehensive vaccine adjuvant optimization strategies as well as the creation of vaccines designed against a variety of emerging and re-emerging diseases.
We thank Professor Thomas Phillips, Professor Jeffrey Adamovicz, Alexis Dadelahi, and Dr. Curtis Pritzl for their useful input on this work. We also thank Biolegend technical support team for their assistance on flow cytometry and cytokine multiplex assays.
This work is supported by the University of Missouri start-up funding, the University of Missouri research council board, and the PhRMA Foundation.
- 1.Fenner F, Henderson DA, Arita I, Jezek Z, Ladnyi ID, Organization WH. Smallpox and its eradication. 1988.Google Scholar
- 3.Mast E, Mahoney F, Kane M, Margolis H. Hepatitis B vaccine. Vaccines, 4th ed Philadelphia: WB Saunders Company 2004:299–338.Google Scholar
- 13.Ross KA, Loyd H, Wu W, Huntimer L, Ahmed S, Sambol A, et al. Polyanhydride-based H5 hemagglutinin influenza nanovaccines elicit protective virus neutralizing titers and cell-mediated immunity. Synthetic nanoparticle-based vaccines against respiratory pathogens 2013:149.Google Scholar
- 14.An M, Liu H. Dissolving microneedle arrays for transdermal delivery of amphiphilic vaccines. Small. 2017;13(26)Google Scholar
- 20.Barrett JC, Ulery BD, Trent A, Liang S, David NA, Tirrell MV. Modular peptide Amphiphile micelles improving an antibody-mediated immune response to group A Streptococcus. ACS Biomater Sci Eng. 2016;Google Scholar
- 22.Zhang R, Smith JD, Kramer Jake S, Allen BN, Martin S, Ulery BD. Peptide amphiphile micelle vaccine size and charge influences immunogenicity. ACS Biomater Sci Eng 2018;Submitted.Google Scholar
- 23.Zhang R, Morton LD, Smith JD, Gallazzi F, White TA, Ulery BD. Instructive design of tri-block peptide amphiphiles for structurally complex micelle formation. ACS Biomater Sci Eng 2018;Accepted.Google Scholar
- 31.Kuai R, Ochyl LJ, Bahjat KS, Schwendeman A, Moon JJ. Designer vaccine nanodiscs for personalized cancer immunotherapy. Nat Mater. 2016;Google Scholar
- 43.Dowling JK, Dellacasagrande J. Toll-like receptors: ligands, cell-based models, and readouts for receptor action. Toll-Like Receptors: Springer. 2016:3–27.Google Scholar
- 57.Triantafilou M, Gamper FG, Haston RM, Mouratis MA, Morath S, Hartung T, et al. Membrane sorting of toll-like receptor (TLR)-2/6 and TLR2/1 heterodimers at the cell surface determines heterotypic associations with CD36 and intracellular targeting. J Biol Chem. 2006;281(41):31002–11.CrossRefPubMedGoogle Scholar
- 59.Takeshita F, Gursel I, Ishii KJ, Suzuki K, Gursel M, Klinman DM, editors. Signal transduction pathways mediated by the interaction of CpG DNA with toll-like receptor 9. Semin Immunol; 2004: Elsevier.Google Scholar