Host‐ and pathogen‐derived adjuvant coatings on protein nanoparticle vaccines

Abstract Nanoparticulate and molecular adjuvants have shown great efficacy in enhancing immune responses, and the immunogenic vaccines of the future will most likely contain both. To investigate the immunostimulatory effects of molecular adjuvants on nanoparticle vaccines, we have designed ovalbumin (OVA) protein nanoparticles coated with two different adjuvants—flagellin (FliC) and immunoglobulin M (IgM). These proteins, derived from Salmonella and mice, respectively, are representatives of pathogen‐ and host‐derived molecules that can enhance immune responses. FliC‐coated OVA nanoparticles, soluble FliC (sFliC) admixed with OVA nanoparticles, IgM‐coated nanoparticles, and OVA‐coated nanoparticles were assessed for immunogenicity in an in vivo mouse immunization study. IgM coatings on nanoparticles significantly enhanced both antibody and T cell responses, and promoted IgG2a class switching but not affinity maturation. FliC‐coated nanoparticles and FliC‐admixed with nanoparticles both triggered IgG2a class switching, but only FliC‐coated nanoparticles enhanced antibody affinity maturation. Our findings that affinity maturation and class switching can be directed independently of one another suggest that adjuvant coatings on nanoparticles can be tailored to generate specific vaccine effector responses against different classes of pathogens.


| I N T R O D U C T I O N
Nanoparticle vaccine delivery systems have emerged as an attractive means of enhancing subunit vaccine adjuvancy. Particulate vaccine carriers can control release of soluble antigens to the immune system and protect them from degradation. 1 However, nanoparticles have been found to be more than just passive antigen depots, and certain types of particles exhibit their own immunostimulatory effects on antigen presenting cells. The exact nature of this nanoparticulate-mediated adjuvancy is unknown, and many fundamental studies have examined the immunological effects of nanoparticle properties such as size, 2 surface charge, 3 shape, 4 and material. 5 Generalized vaccine particle design principles are difficult to elucidate from these studies, however, due to our incomplete understanding of immunology of vaccination, and specifi-cally the type of immune response needed to successfully vaccinate against a particular pathogen. 6 The molecular adjuvants are a more predictable class of immunostimulants. Pathogen-associated molecular patterns (PAMPs) are macromolecules that interact with specific pattern recognition receptors (PRRs) on or inside antigen presenting cells. 1,7 Receptors that bind bacterially-derived or virally-derived macromolecules are hypothesized to initiate adaptive immune responses geared toward those particular classes of pathogens. 7,8 Toll-like receptors (TLRs) are a class of membrane-bound PRRs that have been extensively studied for vaccine adjuvant use. [9][10][11] However, safety concerns over administration of pathogen-derived compounds require thorough investigation. 12 Currently, several pathogen-derived vaccine adjuvants are undergoing clinical trials, but only two have been approved for use in humans. 13 V C 2017 The Authors. Bioengineering & Translational Medicine is published by Wiley Periodicals, Inc. on behalf of The American Institute of Chemical Engineers This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Flagellin (FliC) is a TLR-5 ligand shown to greatly enhance responses to influenza vaccination. 14,15 Given the strength of FliC as an adjuvant, vaccines have been proposed with genetic fusion of antigenic peptides with the FliC protein, 11,16 as well as nanoparticles decorated with FliC. 17,18 As of this writing, at least six clinical trials have been completed with FliC-fusion proteins. 19 The propensity of certain FliC-fusion proteins to aggregate, even at 48C, may decrease their efficacy, 11 and the sequence-dependent nature of FliC-fusion protein stability reduces its attractiveness as a platform technology. Nanoparticles with a stable, native FliC coat, or with native FliC admixed can combine the immunostimulatory properties of FliC with those of antigencontaining nanoparticles. The optimal location of antigen and adjuvant in nanoparticle vaccine formulations is still under active research, 9,20 and recent findings suggest that flagellated bacteria in the gut assist in TLR-5-mediated adjuvancy to subcutaneously administered influenza vaccines. 14 Using TLR ligands as adjuvants, however, poses the risk of safety issues 11 and immune responses against the adjuvant itself. 21 The use of host-derived proteins as vaccine adjuvants may be able to address some of the issues associated with pathogen-derived adjuvants. Antibodies, or immunoglobulins (Ig), coat pathogens during the immune response to an infection, and these proteins may be able to act as in situ adjuvants rendering nanoparticles more immunogenic in vivo.
While antibodies immobilized by affinity interactions on the nanoparticles' surface should remain bound, any soluble Ig in the formulation should be recognized as host protein and consequently nonimmunogenic, and would simply enter the host's circulating repertoire of antibodies. Additionally, the current, widespread good manufacturing practice production of humanized antibodies offers a pathway for largescale production of immunoglobulin-based adjuvants.
The idea of immunoglobulin-mediated adjuvancy has been explored through the use of antibody-bound antigen, or immune complexes, as vaccines. [22][23][24][25] IgG2a complexed with soluble ovalbumin (OVA) was able to enhance specific anti-OVA antibody concentrations and CD4 1 T cell responses by over an order of magnitude in comparison to soluble OVA. 26 Although several sources state that immunoglobulins enhance responses against soluble antigen and suppress them when bound to particulates, 27 this assertion was based on evidence of anti-Rh factor IgG suppressing immune responses against fetal erythrocytes in pregnant women. 28 Immunosuppressive responses against IgG-opsonized nanoparticulates have not been definitively reported. Moreover, a study comparing the inflammatory properties of soluble and insoluble immune complexes from rheumatoid synovial fluid found that the larger, insoluble immune complexes were more immunostimulatory than the soluble ones, 24 supporting the hypothesis that particle size and immunoglobulin opsonization may synergistically enhance immune responses.
The protein corona that forms on nanoparticles in serum in vivo consists of many protein types, and biomaterial-serum protein interactions are an active area of research. 29 Engineering biomaterial surfaces to bind antibodies can enhance immunogenicity by targeting the antigen particles to macrophages and dendritic cells via Fc receptors on these antigen-presenting cell types. 30 Furthermore, antibodyopsonized nanoparticles and microparticles provide a unique platform for activating the complement system, an inflammatory extracellular signaling cascade designed to neutralize infection, trigger local inflammation, and assist in the adaptive immune response. 7,31 The present study of adjuvant nanoparticle coatings looks at both pathogen-derived flagellin (FliC) and the host-derived antibody immunoglobulin M (IgM). IgM is the first antibody isotype made by antibody-producing B cells and is a stronger activator of the complement system than the more prevalent IgG. 32 It is possible that IgM enhances the adaptive immune response to the antigen to which it is bound. Given its lower affinity and different Fc structure than the more prevalent IgG, IgM likely serves an immunoregulatory function in addition to any neutralizing capabilities it may have. Although it has been proposed as a potential vaccine adjuvant due to its interactions with complement, B cells and T cells, 33 to the best of our knowledge, IgM has not been tested as part of any vaccine formulation yet.
Our vaccine nanoparticle core consists of model OVA protein nanoparticles (PNPs), which are nanoparticles composed entirely of cross-linked antigen protein. 10,34 Our immunization of mice with FliCand IgM-coated OVA PNPs examines (a) whether IgM could be used as a host-derived vaccine adjuvant, and (b) whether pathogen-derived adjuvants were more effective bound or unbound from antigen nanoparticles. Overall, our immunization study profiled differences in hostand pathogen-derived adjuvant responses.

| Materials
Endotoxin-free EndoFit TM OVA was dissolved in sterile phosphatebuffered saline (PBS) for all nanoparticle formulations administered in vivo. OVA and endotoxin-free OVA were purchased from Invivogen (San Diego, CA). Antibodies were purchased from Thermo Fisher Scientific (Rockford, IL) unless stated otherwise.

| FliC expression and purification
The plasmid pET22b-flic was used to express recombinant FliC from Salmonella typhimurium. 35 The plasmid was transformed into E. coli  Figure S1).

| Nanoparticle synthesis and characterization
The 270-nm OVA PNP cores were made as previously described. 34 Briefly, 0.4 ml pure ethanol was added at a constant rate to 0.1 ml of 6.2 mg/ml OVA in PBS under constant stirring at 600 rpm. The amine- Nanoparticles were resuspended in water, air-dried, and sputter-coated with palladium prior to visualization with a Zeiss Ultra60 FE (Carl Zeiss Microscopy, Cambridge, UK) scanning electron microscope at 5.0 kV.  Nanoparticles were suspended in fresh DMEM 1 1% FBS at a concentration of 1 lg/mL and used to stimulate transfected cells for 8 hr.

| FliC coating characterization
Bright-Glo Luciferase Assay reagent (Promega, Madison, WI) was diluted 1:1 with serum-free DMEM and used to assess luciferase activity according to the manufacturer's instructions.

| Immunization
All animal work was compliant with the NIH Guide for the Care and Use of Laboratory Animals and all protocols and procedures employed were reviewed and approved by the Emory University Institutional Animal Care and Use Committee. Seven-week old female Balb/c mice (Jackson Laboratory, Bar Harbor, ME) were given 50 ll intramuscular (i. m.) injections into the right hind-leg of 0.2 mg/ml nanoparticle formulations as described in Table 1. Injections were repeated 21 days after priming for a boost administration.

| Sample collection
Blood was collected from immunized mice by submandibular venipuncture 2 weeks after prime and boost immunizations. Blood was allowed to clot at 48C for at least 30 min, and was centrifuged at 5,000 rpm for 5 min to collect serum. Serum samples were stored at 2208C for further analysis.
Following euthanasia on Day 39, splenocytes were prepared from mouse spleens. Briefly, spleens extracted from mice were homogenized manually with the plunger of a 1 ml syringe and cells collected by centrifugation at 3003g for 5 min. Cells collected were resuspended in red blood cell lysis buffer (150 mM NH 4 Cl, 10 mM NH 4 HCO 3 , 1 mM Na 2 EDTA, pH 7.4) for 5 min at room temperature, quenched with RPMI 1640 media (ATCC, Manassas, VA) and centrifuged for 5 min at 2,3003g. Splenocytes collected were resuspended in RPMI 1640 at

| Serum antibody assessment
OVA-specific IgG antibody titers were assessed by ELISA, as previously described. 10 Briefly, serial twofold dilutions of serum were analyzed using a standard ELISA procedure, with 1 lg/ml OVA in PBS as the capture antigen, 1% BSA in PBS as the blocking solution, and 1 lg/ml HRP-anti-mouse IgG in 1% BSA solution as the detection antibody. OVA-specific IgG1 and IgG2a concentrations were also assessed by ELISA as described above, using HRP-conjugated anti-mouse IgG1 and IgG2a, and monoclonal mouse IgG1-or IgG2a-anti-OVA to create a standard curve (Chondrex, Redmond, WA).

| Cytokine ELISpot
Splenocytes were seeded at a density of 2.5 3 10 6 cells/ml on interferon g (IFN-g) and interleukin 4 (IL-4) 96-well ELISpot membranes (R&D Systems, Minneapolis, MN). Splenocytes were stimulated with or without 50 lg/ml endotoxin-free OVA, and incubated at 378C in humidified air with 5% CO 2 for 36 hr. ELISpot membranes were developed according to the manufacturer's instructions. Wells were imaged using a dissection microscope (Olympus SZX16, Olympus Corporation, Tokyo, Japan), and spots were counted using ImageQuant TL's colony counting software (GE Healthcare, Pittsburgh, PA).

| Flow cytometry
Splenocytes were seeded at a density of 2.5 3 10 6 cells/ml on 96-well plates, and stimulated with or without 50 lg/ml endotoxin-free OVA, and incubated at 378C in humidified air with 5% CO 2 for 60 hr. Cells

| Affinity maturation
Affinity maturation of anti-OVA serum antibodies was measured using biolayer interferometry with the ForteBio Octet RED96 system (Pall

| Statistical analysis
Serum antibody titers were analyzed using the Mann-Whitney U test.
Antibody concentrations and T cell counts were analyzed using oneway analysis of variance (ANOVA) followed by Sidak's multiple comparisons test. Comparisons between two groups were performed using Student's t-test. All statistical analyses were conducted using GraphPad Prism 6 (GraphPad, La Jolla, CA). The p values of p < .05 were considered statistically significant (*p < .05, **p < .01). To test our hypotheses, statistical comparisons were assessed between G1 and G3, between G2 and G4, and for T cell counts, between G6 and all other groups.
Comparisons between these groups that were significant are noted in the figures, while comparisons that were not significant are not shown.

| Coated PNP synthesis and characterization
Monodisperse, 270 nm OVA nanoparticles were made as previously described. 34 Coating the nanoparticles did not significantly alter nanoparticle size (Figure 1a). IgM-coating the nanoparticles without a soluble OVA quenching step resulted in large, 1,000 nm particles, suggesting

| Coat activity
Coat activity was confirmed by testing FliC and IgM functionality. Since FliC is a TLR-5 agonist, FliC-coated nanoparticles were used to activate a TLR-5-dependent luciferase assay. FliC-coated OVA nanoparticles activated TLR-5 signaling, and did not significantly differ in activity compared to soluble FliC admixed with OVA nanoparticles (Figure 2a).
IgM's ability to activate complement was assessed by incubating IgMcoated nanoparticles with human serum and using ELISA to detect activated complement. 36 Uncoated OVA nanoparticles were found to activate complement, and the IgM coating on these particles did not significantly enhance complement activation (Figure 2b).

| Antibody production
Anti-OVA serum IgG titers were assessed 2 weeks after priming and boosting (Table 1)

| Affinity maturation
Anti-OVA antibody affinity was measured with the Octet RED system.

| DISCUSSION
Our previous work with OVA nanoparticles highlighted the importance of protein nanoparticle coating in altering dendritic cell inflammatory responses. 34 In addition to coating our nanoparticles with antigen, the current study explores the in vivo immune responses to pathogen-and host-derived adjuvant coatings on to nanoparticles.

| IgM as a host-derived adjuvant
Potential safety issues have been raised for TLR ligand-based adjuvants that may dissociate or diffuse away from the antigen. 6 Unlike FliC, host-derived IgM that may dissociate from the nanoparticles is probably not going to be seen as immunogenic as soluble FliC, and thus an OVA nanoparticle 1 soluble IgM group was not included in the study design.
Antibodies have been proposed as host-derived adjuvants before. 33,40 Most of these studies have been with soluble immune complexes consisting of soluble antigen bound to a cognate anti-body. 41,42 This strategy targets the antigen to Fc receptor-bearing antigen presenting cells, yet does not exploit a second feature of antibodymediated adjuvancy-the activation of complement.
Complement activation can be triggered by the proximity of two IgG Fc domains, or one IgM Fc domain exposed upon antigen binding. 32 Activation of complement is necessary for vaccination not only as an innate host defense mechanism, 7 but also for bridging innate and adaptive immune responses. 43

| Summary
In this work, we tested the efficacy of a host-derived adjuvant, IgM, as well as the use of a pathogen-derived adjuvant both on nanoparticles and admixed with them. Our results are summarized in Table 2. Our Other work has shown that delivery of two types of adjuvants in separate particles elicits greater effects compared to adjuvant co-delivery in the same particle. 9 Perhaps our most surprising finding was that antibody affinity maturation and IgG2a class switching did not correlate with one another.
While the two processes are normally associated with each other in the development of an antibody response, 7 we found that unadjuvanted OVA nanoparticles and FliC-coated OVA nanoparticles triggered affinity maturation, while IgM-and soluble FliC-adjuvanted nanoparticles did not. Our results stand in contrast to those by Corley et al., who showed that IgM-bound soluble antigen (IgM-ICs) accelerates affinity maturation responses to T-dependent antigens. 47 Future work should examine the differences in immune responses to soluble and nanoparticulate immune complexes, and whether such a difference can be exploited to tune the affinity of the humoral immune response. Affinity maturation is necessary for generating high affinity, neutralizing antibodies, which can be protective against highly conserved pathogens. 48 For pathogens that mutate or change yearly, such as influenza, however, the generation of high-affinity neutralizing antibodies results in a loss of antibody diversity, and can contribute to the phenomenon known as original antigenic sin, in which antibodies are only made to epitopes found on the first strain of virus the immune system encountered. 7 If vaccine adjuvants can delay the affinity maturation process while promoting diversification of antibody effector functions via class switching, it is possible that the memory B cell repertoire generated from the immunization will be more effective at combatting rapidly mutating pathogens.

| C O NC LU S I O N
As vaccination moves away from the isolate-inactive-inject paradigm 49 and toward more engineered vaccine formulations for directing the immune response, the interplay between particulate and molecular adjuvants needs to be understood. We examined the role of adjuvant location on model OVA PNPs with flagellin, and found that FliC location directs the affinity maturation response. To sidestep potential issues with pathogen-derived adjuvant toxicity, we also explored using immunoglobulins as a host-derived, immunostimulatory adjuvant coating on nanoparticles. We found that although IgM coating on OVA nanoparticles does not significantly enhance complement activation in vitro, it does enhance antibody and memory T cell responses in vivo, while not promoting affinity maturation. Further studies need to be done to investigate the effector functions of other classes of immunoglobulin adsorbed to nanoparticles, and if the delayed affinity maturation responses we see with our vaccine nanoparticles can translate to protective immune responses in in vivo challenge models of highly mutable pathogens.