A perspective of nanoparticle universal influenza vaccines

Abstract

Annual influenza infections cause massive economic loss and pose severe threats to public health worldwide. Seasonal influenza vaccines are the most effective means of preventing influenza infections but still possess major weaknesses: requiring updating vaccine strains every year, difficulty for accurate prediction of viruses that will be circulating and low protection efficacy due to mismatching, time-consuming egg-dependent production, and ineffectiveness of protection against influenza pandemics. The ultimate quest is for a universal influenza vaccine that could be used from year to year. One approach under investigation is to design influenza vaccine immunogens based on conserved amino acid sequences or conformational epitopes, rather than strain-specific ones of influenza viruses. Such vaccines can elicit broadly reactive humoral and cellular immunity. Universal influenza vaccine development has intensively employed nanotechnology because the structural and morphological properties of nanoparticles dramatically improve vaccine immunogenicity and induced immunity duration. Layered protein nanoparticles can eliminate the possibility of off-target immune responses, integrate epitopes into different regions of the particles for differentiated antigen-recognition and processing, and facilitate comprehensive immune response induction. Herein, we reviewed the designs of effective nanoparticle universal influenza vaccines, the recent discoveries of specific nanoparticle features that contribute to immunogenicity enhancement, and recent progress in clinical trials.

1. Introduction — An Urgent Need for an Affordable Universal Influenza Vaccine.

Seasonal influenza causes 3 to 5 million cases of severe illness and up to 650,000 deaths yearly according to the World Health Organization. ¹ Mismatched seasonal influenza vaccines are low protection efficacy and provide limited protection against circulating strains. For instance, by mid-October in 2017 the influenza vaccine effectiveness (VE) against H3N2 was estimated to be only 10%, which was closely associated with the predominant H3N2 activity in the southern hemisphere. ² Not surprisingly, vaccination with this H3N2-containing vaccine had not counteracted the H3N2-predominant epidemic in the United States during the following months. ³ During the 2017–2018 influenza season, the severity of influenza B outbreaks in certain regions of the northern hemisphere was associated with the lack of a vaccine strain from the influenza B Yamagata lineage in the traditional trivalent influenza vaccine. ^4–5

Influenza viruses contain three membrane proteins: hemagglutinin (HA or H), neuraminidase (NA or N), and matrix protein 2 (M2) (Figure 1). ⁶ Induction of protective immunity against influenza requires recognition of these surface proteins by the host. M2 is relatively genetically stable and possess low immunogenicity. HA and NA are much more immunogenic but also highly antigenically variable. There are 18 HA subtypes and 11 NA subtypes known for influenza A. ⁷ Antigenic drift — mutations in HA and/or NA — results in new influenza A virus strains over time. Antigenic shift — a major re-assortment of HA and/or NA genes — results in new subtypes of influenza A virus. As a result, influenza virus stays in a continuous state of genetic flux and displays different variants of HA and NA on its surface, enabling to evade preexisting immunity. A non-human influenza virus may also cause a pandemic in human populations by acquiring the capacity for transmission in humans. The recent infection of humans by highly pathogenic avian H5N1 and the outbreak of a novel avian H7N9 strain has reinforced this concern. ^8–9 The inherent variability of HA and NA surface proteins of influenza virus creates an intractable problem for the seasonal influenza vaccine approach. For this reason, there is an urgent need for a universal influenza vaccine that will induce broad cross-protection against divergent influenza viruses.

Figure 1. Schematic diagram of influenza A virus.

The antigen name encoded by each gene segment is labeled aside. HA, hemagglutinin; NA, neuraminidase; M1, matrix protein 1; M2e, matrix protein 2 ectodomain; vRNP, viral ribonucleoprotein; PB, polymerase basic protein; PA, polymerase acid protein; NP, nucleoprotein; NS, non-structural protein.

2. Conserved Influenza Sequences as Universal Vaccine Immunogens

The suboptimal VE of conventional influenza vaccines is multifactorial. Lower VE is associated with previous vaccination with seasonal influenza vaccines ^10–12 and increasing age of the recipients. ¹³ Vaccinologists have attempted to improve the efficacy of the seasonal vaccine through different means. Increasing the dosage of conventional influenza vaccination induced higher vaccine-specific antibody titers and interleukin - 10 levels, but has little positive impact on the development of functional T-cell memory in older adults. ¹⁴ Recent studies revealed that insect cell-produced HA restores the reduced immunogenicity caused by egg-adaptation. ¹⁵ Researchers created a novel influenza vaccine strain by introducing eight interferon-sensitive mutations — point mutations identified through quantitative high-throughput genomics analysis — and found improved in vivo immunogenicity and protectivity. ¹⁶ Nevertheless, a universal influenza vaccine is still preferable and can be used without a yearly update of vaccine components.

A universal influenza vaccine would rely on conserved amino acid sequence and/or epitope conformation from influenza to generate broadly reactive humoral and cellular immunity. M2 is a relatively conserved influenza antigen but is expressed in low copy numbers on the virion. However, infected target cells abundantly express M2 on their surfaces where it is accessible for antibodies. ¹⁷ Protection against multiple influenza A virus subtypes elicited by M2 ectodomain (M2e)-based vaccines correlates closely with M2e-specific antibody affinity to Fc gamma receptors on target cells. ^18–21

The HA stalk domain is relatively more amino acid- and conformation-conserved than the HA head domain and can induce antibodies capable of binding multiple HA subtypes. ^22–28 Protein structural analysis has enabled the development of some recombinant HA stalk domain vaccines with effective in vivo immunogenicity. ^29–31 HA stalk-directed protective immunity prevents infection by inducing neutralizing antibodies, antibody dependent cellular cytotoxicity and antibody dependent phagocytosis. ³¹ T cell responses against internal influenza components and structural components correlate with cross-reactive protection and early clinical recovery from infection. ^32–34

Nucleoproteins (NP) enclose the viral RNA segments, are necessary for trafficking the viral genome, and are highly conserved in comparison to the surface glycoproteins. ^35–36 Blending NP with other conserved influenza components polymerase basic-2 and/or matrix protein 1 (M1) — in fusion constructs broadened the spectrum of immune protection induced by vaccine candidates. ^37–41

Vaccination with multiple conserved target antigens will have advantages over using single antigens. ⁴² A universal vaccine providing long-term protection against heterosubtypic influenza virus strains will benefit pandemic control and routine vaccination.

3. Different Types of Nanoparticles Accommodating Influenza Conserved Epitopes Benefit for Enhanced Immunogenicity

Conserved influenza antigens typically induce weak immune responses. There is a need to develop novel delivery systems and adjuvants that increase the immunogenicity of conserved components for next-generation vaccines. The application of nanotechnology in vaccinology has been increasing dramatically over the past decades. ⁴³ The common benefits of nanoparticles are prolonged circulation and directed targeting by modification of nanoparticle surfaces with dendritic cell or T cell targeting ligands. ⁴⁴ Nanoparticles are an efficient delivery system which facilitate the enhancement of immune cell uptake and the sustaining antigen release. ^45–46

Our recent study indicated that gradual intracellular disassembly of protein nanoparticles led to sustained specific immune responses. ⁴⁷ Physiologically controlled, stimuli-sensitive release is a major benefit of this nano system. An optimized eudragit-S/Trehalose polymer nanoparticle influenza vaccine released influenza antigens during in vitro simulation of the pH conditions of the mouse gastrointestinal tract. Mice experiments indicated that oral immunization with this nanoparticle vaccine induced protective immune responses. ⁴⁸ Thermo-sensitive, redox condition-sensitive, and near infrared light activated controlled release has recently been developed in the field of anti-cancer research. ^{44, 49} Comparatively, specific features of different nanoparticle types can benefit prophylactic universal influenza vaccine designs (Figure 2). One or multiple features can be synergistically integrated into one nanoparticle system for an optimal immune response. Divergent nanoparticle universal influenza vaccines under investigation are summarized in Table 1.

Figure 2. Nanoparticle universal influenza vaccine types.

(A) Polymer nanoparticle; (B) Virus like particle; (C) Gold nanoparticle; (D) Chemically assembled double-layered protein nanoparticle; (E) Naturally assembled protein nanoparticle (EMDB EMD-6332).

Table 1.

Overview of nanoparticle universal influenza A vaccines.

Overview of nanoparticle universal influenza A vaccines
Vaccine types	Vectors	Antigen types	Immunogenicity readout in animal models	References
Synthetic polymers	PLGA	Norovirus P particle containing M2e, Peptides of HA, NP and PA; outermembrane vesicles displaying M2e	Pigs, mice,	57, 58, 59
	poly-γ-glutamic acid/chitosan	M2, HA fusion peptide; H1N1 split vaccine	Mice	63, 64
VLPs	Virosome	Headless HA; M2e; M2e/FliC;	Mice	83, 84, 85, 86, 87
	Hepatitis B core	M2e	Mice, pigs, ferrets, human	18, 80
	MaMV	M2e	Canine	88
	Tobacco mosaic virus coat protein	M2e	Mice	89
	Papaya mosaic virus	M2e	Mice	92
	F88 bacteriophage	M2e	Mice	81
	Woodchuck hepatitis VLP vectored in Salmonella Typhimurium	M2e	Mice	94
	T7 bacteriophage	M2e	Mice	82
	Q-β	M2e	Mice	88
Self-assembling nanoparticles	Ferritin	Headless HA; M2e	Mice, ferrets	30, 97, 99
	Self-Antigen	M2e, headless HA; M2e-NA tetramer; NP	Mice	31, 47, 87, 100, 101
Metal nanoparticles	Gold nanoparticle	M2e/CpG	Mice	108

1). Polymer nanoparticles.

Synthetic polymers are widely used to fabricate nanoparticle vaccines due to their favorable properties such as biocompatibility and biodegradability. US Food and Drug Administration approved poly(D,L-lactic-co-glycolic acid) (PLGA) has been extensively characterized in animal models and is widely used in vaccine development as an encapsulation matrix ⁵⁰ to co-deliver ^51–52 and sustain controlled release of antigens. ^{46, 53} Other synthetic polymers with similar features include poly(D, L-lactide-co-glycolide), ⁵⁴ poly(D,L-lactic-co-hydroxymethyl glycolic acid) ⁵⁵ and polystyrene. ⁵⁶

In universal influenza vaccine development, PLGA nanoparticles encapsulating conserved influenza A virus components including norovirus P particle containing M2e and synthetic peptides of HA, NP and polymerase acidic protein, elicited strong prophylactic protection and eliminated influenza symptoms in pigs against swine H1N1 challenge, ⁵⁷ which significantly improves the previously reported immunization strategy with M2e and NP antigens in pigs. ⁵⁸ PLGA can be formulated into larger sizes and facilitate the release of nanometer-sized constructs. Recombinant outer membrane vesicles displaying M2e released from PLGA microparticles over 30 days, induced sustained protective M2e specific immunity in mice. ⁵⁹

Natural polymers based on polysaccharides have also been used to prepare nanoparticles, such as pullulan, alginate, inulin and chitosan. ⁴³ Non-toxic chitosan nanoparticles have been widely studied as vaccine vectors owing to its advantages, biocompatibility, biodegradability and amenability to ease modification into desired shape and size. ^60–62 Poly-γ-glutamic acid/chitosan nanoparticles containing truncated M2 and a fusion peptide of influenza HA and mucosal adjuvant cholera toxin subunit A1 acted as an effective mucosal vaccine against diverse influenza A viruses. ⁶³ Nanogel particles encapsulating pandemic H1N1 split vaccine antigen increased cross-protection against homologous and heterosubtypic influenza A viral infection. ⁶⁴ The diverse biocompatible polymers for nanoparticles expand our ability to accommodate different immunogens and adjuvants. However, because a major portion of the vaccine-loaded nanoparticles is the polymer materials, the low immunogen load of the polymer nanoparticles could be a limitation for some immunogens needing high dosage for a robust immune response.

2). Virus-like particles.

Virus-like particle (VLP) formation is driven by the same mechanisms used during virus assembly — viral self-assembling proteins or domains. The morphological features of VLPs simulate the features of viruses that the host immune system has evolved to combat. This makes VLPs highly immunogenic. ^65–68

The first VLP vaccine for hepatitis B virus, developed by Glaxo Smith Kline, was commercialized in 1986 ⁶⁹ followed by commercialization of other VLP vaccines including Cervarix (human papillomavirus, Glaxo Smith Kline), Recombivax HB (hepatitis B virus, Merck & Co. Inc) and Gardasil (human papillomavirus, Merck & Co. Inc). Many others are currently in the clinical trial or research stage.

Because VLPs lack genomic components or contain premature termination codon, ⁷⁰ they are safe compared with other replicating vaccine vectors. Due to viral features like repetitive structures, naturally retained antigen conformation, and virion sizes, VLPs induce innate and adaptive immune responses. ^71–72 Influenza lacks a capsid. Instead, it has a core bridged with viral envelop by the matrix protein M1. VLPs are assembled in the physiological conditions on host membranes and released into the environment via budding. Our influenza VLPs are formed in SF9 insect cells by the same manner that influenza envelopes are formed: a budding process organized by the M1 protein. Enveloped VLPs can be produced by co-expression of structure proteins in mammalian cells, insect cells, or plants. ^73–79 Furthermore, the self-assembling biocompatible viral capsid proteins can be adapted to various cell systems, like mammalian cells, insect cells, plants, yeasts and Escherichia coli (E. coli) for VLP production. ^80–82

A vaccine incorporating PR8 (A/Puerto Rico/8/1934) headless HA into human immunodeficiency virus Gag-based VLPs expressed in 293T cells protected mice against influenza viral challenges. ⁸³ M2 VLPs have been produced in insect cells coinfected with recombinant baculoviruses expressing M1 and wild type M2 protein ⁸⁴ or multiple M2e. ⁸⁵ Supplemental M2 VLP immunization with inactivated H1N1 vaccine enhanced cross-protection against influenza viral challenges. ^84–85 Proper presentation of ligands or agonists in nanoparticles targeting immune cells potentially make antigens more immunogenic in vivo. ^86–87 Appropriate antigen presentation in a nanoparticle form and coadministration of adjuvant molecules, often targeting specific receptors of immune cells, elicits strong specific immune response. We designed a membrane-anchored fusion protein by replacing the hyperimmunogenic region of Salmonella enterica serovar Typhimurium flagellin with four repeats of M2e (4M2e-tFliC) and fusing it to the membrane anchoring domain of influenza HA. The fusion protein was incorporated into influenza M1-driven VLPs. Immunization with these VLP greatly enhanced the M2e specific immune responses. ⁸⁶

Non-enveloped VLPs have been widely developed with chimeric capsid proteins displaying influenza conserved epitopes which provide effective bystander T-helper responses and induce protective immune responses specific to such epitopes. ^{80–82, 88–94} Immunization with E. coli produced hepatitis B core VLPs displaying M2e cross-protected mice and ferrets against diverse influenza virus challenge and efficiently induced protective M2e-specific immunity in volunteers. ¹⁸ VLP-display are frequently investigated as a promising vaccine platform for presenting conserved surface proteins in a highly immunogenic form.

3). Self-assembling protein nanoparticles.

Naturally occurring self-assembling protein nanoparticles have been identified from a wide variety of sources. ⁹⁵ Self-assembling motifs can enable fusion proteins to assemble into protein nanoparticles. ^96–98 Ferritin can be self-organized into a nano scale structure (nanocage) with intracellular iron storage functionality. It serves as an ideal epitope presentation platform with repetitive symmetrical structure and an ordered matrix. It was previously reported that headless HA trimers and tandem copies of M2e displayed on ferritin nanoparticles retained native conformation and induced protective homosubtypic and heterosubtypic immunity in vivo. ^{30, 99} Because most of large self-assembling domains are adopted from non-human species, off-target immune responses are a major weakness of such self-assembling protein nanoparticles. ^{30, 96–97, 99}

Changes in the physicochemical condition of protein solutions can drive the formation protein nanoparticles. Because changes in physicochemical condition are dynamic processes, nanoparticles assembled this way are generally not homogenous in size but do show a normally distributed range of sizes. We first used ethanol desolvation to assemble influenza M2e or influenza HA into protein nanoparticles. ^{31, 87, 100–101} Because these particles have solid structures without non-antigen components, they have the highest possible antigen-load for protein nanoparticles. Another advantage of these nanoparticles is that they can go through multiple cycles of particle assembly processes with different antigenic proteins to assemble proteins into different layers around the particles. Layered protein nanoparticles are particularly suitable for protein antigens with different stability in solution. We have generated double-layered protein nanoparticles by ethanol desolvation of M2e or NP peptides into nanoparticles as the cores and chemically crosslinking structure-stabilized influenza HA stalk antigens or M2e onto the outer layers. ^{31, 47} The fabrication process is summarized in figure 3. These physicochemical layered protein nanoparticles can maximize the different immunological role of the different antigens. ¹⁰⁰

Figure 3. Schematic diagram of double-layered nanoparticle fabrication process.

RT, room temperature; DPBS, Dulbecco’s phosphate-buffered saline.

The desolvated nanoparticle size is 50 to 300 nm, depending on the desolvent composition ¹⁰² and desolvated protein materials. ³¹ Because this technique is compatible with most proteins, other proteins such as innate signaling initiators (immune stimulators) or immune cell-targeting molecules can be crosslinked onto the outer layers to endow additional beneficial immunological features. ¹⁰³

The combination of dissolving microneedle patch technology with such protein nanoparticles enables convenient skin vaccination, which is syringe-free, painless, and can be self-administrated. Mouse skin immunization experiments demonstrated that this protein nanoparticle vaccine conferred at least four months of universal immunity against diverse influenza virus challenges. Furthermore, this approach allows for cold-chain independent storage and triggered long-lasting protection. ³¹

4). Metal nanoparticles.

Metal nanoparticles are rigid in structure and nearly non-biodegradable. The shape and growth of metal nanoparticles are controlled by fine-tuning the rates of surface diffusion and deposition. ¹⁰⁴ The inorganic nanoparticle is frequently investigated as a vaccine development platform in order to improve antigen immunogenicity and avoid antibody production against the platform materials. Gold nanoparticles have attracted attention in the nanomedical field due to their unique pros, their biocompatibility and easy fabrication in terms of size and shape. Gold nanoparticles less than 100 nm in diameter can be synthesized, which is preferentially recognized and engulfed by dendritic cells. Gold nanoparticles can also be formed into different shapes like star, spherical, cube and rod to control the induced, shape-dependent modulation of immune responses. ¹⁰⁵ However, the cons of gold nanoparticle vaccines include expensive manufacture cost and less drug loading capacity than polymer nanoparticles, VLPs and desolvated nanoparticles.

A preliminary study reported that M2e-immobilized gold nanoparticles formulated with CpG induced cross-protection against diverse influenza viral challenges. ^106–107 Our studies have demonstrated that gold nanoparticles can conjugate influenza HA and adjuvant protein flagellin and trigger strong immune responses conferring heterologous protection in mice. ^108–109

5). Important Physical Features Benefit to Enhanced Immunogenecity

The properties of nanoparticles, such as size and steric conformation, play an active role in mediating the biological effects and the adaptive immunity induction. ^110–113 Ultra-small nanoparticles less than 10 nm in diameter or soluble antigens can rapidly diffuse into and out of lymph organs, which decreases the opportunity for uptake by antigen presenting cells (APCs). ¹¹⁴ Nanoparticles larger than 100 nm can be trapped in the injection site, require active transport, and are eventually scavenged by tissue-resident APCs (Figure 4). ¹¹⁵ We found that intramuscularly injected fluorescent NP nanoparticles remained at the injection site much longer than fluorescent soluble NP protein (Figure 5). Smaller nanoparticles, optimally size around 50 – 200 nm in diameter, were shown to be engulfed more efficiently by dendritic cells while the larger, micrometer-scale nanoparticles were preferentially internalized by macrophages. ^{105, 116–117}

Figure 4. Nanoparticle transport route changes with nanoparticle sizes.

Nanoparticles with larger sizes over 100 nm are captured by dendritic cells from outside of lymphatic epithelia, in comparison, soluble protein antigen and smaller nanoparticles less than 10 nm are able to penetrate through lymphatic epithelia. Once antigens transported into lymph nodes via lymphatic vessels, they are presented to B cells and T cells by follicle dendritic cells. The larger nanoparticles are much more efficiently taken up by dendritic cells than smaller nanoparticles and potentially induce stronger and long-term specific immunity.

Figure 5. In vivo visualization in mice.

The fabrications of fluorescent NP protein and fluorescent NP nanoparticle were described previously. (46) Mice were fed alfalfa-free diets for two weeks to reduce autofluorescence background before intramuscular injection with five micrograms of fluorescent antigen each mouse. Biodistributions of intramuscularly injected Alexa Fluor 700 succinimidyl ester dye-conjugated fluorescent NP nanoparticle and fluorescent soluble NP protein were analyzed in vivo using Perkin-Elmer IVIS Spectrum In Vivo Imaging System at the following time points: pre-injection, 0-hour post injection (hpi), 24-hpi and 48-hpi. (46) Mice injected with DPBS were used as negative control.

The strength of induced immunity is also affected by the efficiency of dendritic cells migration towards the draining lymph nodes (LNs) and the length of antigen retention period in LNs. ¹¹⁸ We recently found that ~200 nm protein nanoparticles had an increased chance to be presented by APCs due to their efficient drainage into inguinal LNs and spleens and relatively long residence at those sites compared with soluble antigens. The soluble molecules diffused rapidly from the injection sites and disappeared earlier from the LNs. ⁴⁷

We also found that nanoparticle vaccines delivered by dissolvable microneedle patches conferred mice long-lasting anti-influenza A virus immunity. We speculated that the entrapment in immune tissues and the intracellularly-activated disassembly of nanoparticles within APCs allowed for the sustained processing and releasing of peptides for a longer period versus soluble antigens, shaping memory T cells and long-term immunity. Sustained antigen release is a critical characteristic for long-lasting memory immune responses to nanoparticle vaccines. ¹¹⁴

Particle shape also plays important roles in the mediation of immunity induction, especially in the phagocytosis of particles. ^119–120 The local shape at the interface between particles and phagocytes determines the efficiency of particle uptake. ^121–123 For example, spherical gold nanoparticles (40 nm) more efficiently induced antibody responses against delivered antigens than did cube- and rod-shaped nanoparticles with similar sizes. ¹²⁴

A phase I clinical trial of the universal influenza A nanoparticle vaccine, M2eHBc (ACAM-FLU-A™) sponsored by Acambis (later acquired by Sanofi Pasteur), generated interesting results. The intramuscularly injected ACAM-FLU-A™, adjuvanted with QS21, was well tolerated and able to stimulate anti-M2e seroconversion in up to 90% of healthy volunteers. However, the induced M2e-specific antibody titers dropped rapidly over 10 months. ¹⁸ Resolving the short-term persistence issue of the M2e-specific immune responses is the last piece of the puzzle in M2e-based vaccine designs. We found that desolvated M2e nanoparticles have an order of magnitude higher levels of M2e antigen loading than the M2eHBc construct, potentially inducing stronger and longer sustained protective M2e-specific T cell immune responses in humans.

Concluding remarks

Even if proof-of-concept has been demonstrated in pre-clinical data, a vaccine product candidate still be potentially incompetitive in the market if costly or inscalable manufacturing process. The baculovirus expression vector/insect cell (BEVS/IC) system constitutes an optimal expression system for VLP manufactures, through which Glaxo Smith Kline’s Cervarix human papillomavirus VLP cancer vaccine is produced. Insect cell large-scale suspension cultures established in either stirred or rocked bioreactors have been shown to be one of the best VLP production system. Furthermore, improved insect cell lines has been ‘humanized’, which perform mammalian-like post-translation glycosylation modifications. ¹²⁵ There are no other marketed types of nanoparticle vaccines.

Because of a series of appreciated features, nanoparticles are potent for the development of an affordable universal vaccines. They are safe, biocompatible, flexible to be fabricated in variable sizes, relatively high surface area for immune ligand-receptor ligation-recognition, capable of accommodating high load of different conserved antigens in the same particle, and sustaining longer period of antigen-supplement, processing and presentation. All these features will be interpreted into a broadly protective universal influenza vaccine.