Mechanisms of cellular and humoral immunity through the lens of VLP-based vaccines

Abstract

Introduction:

Vaccination can be effective defense against many infectious agents and the corresponding diseases. Discoveries elucidating the mechanisms of the immune system have given hopes to developing vaccines against diseases recalcitrant to current treatment/prevention strategies. One such finding is the ability of immunogenic biological nanoparticles to powerfully boost the immunogenicity of poorer antigens conjugated to them with virus-like particle (VLP)-based vaccines as a key example. VLPs take advantage of the well-defined molecular structures associated with sub-unit vaccines and the immunostimulatory nature of conjugate vaccines.

Areas Covered:

In this review, we will discuss how advances in understanding the immune system can inform VLP-based vaccine design and how VLP-based vaccines have uncovered underlying mechanisms in the immune system.

Expert Opinion:

As our understanding of mechanisms underlying the immune system increases, that knowledge should inform our vaccine design. Testing of proof-of-concept vaccines in the lab should seek to elucidate the underlying mechanisms of immune responses. The integration of these approaches will allow for VLP-based vaccines to live up to their promise as a powerful plug-and-play platform for next generation vaccine development.

1. Introduction:

The current Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) viral outbreak has highlighted the need for rapid vaccine development. Traditional vaccines are generally made of inactivated or live-attenuated viruses. As an example, the influenza virus is grown in chicken eggs before being purified and deactivated or killed as part of the vaccine formulation. Finally, adjuvants, stabilizers, and fillers are added before packaging. This is a labor and resource intensive process. The World Health Organization (WHO) estimates that it is a ~6-month process from identification of a new flu strain to commercial incorporation in seasonal flu vaccines[1]. This is for a well-studied virus such as influenza. If it is an entirely new viral pathogen, such as SARS-CoV-2, the process can be substantially slowed down partly because the process of attenuation is an inexact and time-consuming process to ensure safety of the vaccine. As a result, there have been attempts to reduce vaccines to just the key antigenic portions of the virus[2]. If only one or two proteins or carbohydrates are relevant to the protective immune response, recombinant expression or synthesis of these subunit antigens of interest presents a more homogenous construct, resulting in a simplified vaccine formulation.

Unfortunately, subunit vaccines are often associated with poor immunogenicity and short-lived immune responses[3]. Despite these setbacks, the appeal of a well-defined subunit vaccine has endured. To overcome the low inherent immunogenicity or stimulate long-lived immune responses, the antigenic targets have been conjugated to immunogenic carriers that boost the immunogenicity. These conjugate vaccines come in all shapes and sizes, from single antigens covalently conjugated to small molecule adjuvants to hundreds of copies of the antigen on nanoparticle carriers[4–8]. Boosting the immunogenicity of antigens and the magnitude of the immune response have allowed the possibility of vaccines targeting non-viral diseases such as cancer, high cholesterol, Alzheimer’s, and bacterial infections[9–16]. As a result, the field of conjugate vaccines is ever growing. This review will focus on a mechanistic understanding of the use of virus-like particle-based conjugate vaccines to elicit cellular and humoral immunity.

2. Search Strategy:

Publications were identified through Searches on PubMed, Google Scholar, and SciFinder Scholar and through the bibliographies/references of relevant literature. Searches were conducted using keywords including, virus-like particles, VLP, and conjugate vaccines, alone and in combination with specific terminology related to different parts of the immune response. There was no date restriction.

3. Virus-like particles as carriers in conjugate vaccines:

Virus-like particles (VLPs) are protein-based nanoparticles made up of repeating copies of a coat protein(s). These structures usually include the surface proteins and another macromolecule (RNA, DNA, or protein) encapsulated to aid in the assembly of a capsid[17,18]. Because of their viral origin, VLPs almost always contain a helper T cell epitope that can boost the immune response and bias it towards a long-lasting immunological memory[19]. Initial forays into the use of VLPs in vaccination strategies utilized human viral capsids such as Hepatitis B surface antigen (HBsAg) or Human Papilloma Virus (HPV) L1 Major Capsid Protein[20,21]. Commercial vaccines against HPV and hepatitis B have been approved using recombinant VLPs. These vaccines have been safe and widely effective[22–24]. Yet, there is much to learn based on clinical observations, such as the effect of age of vaccination on levels of antibodies elicited[25].

The use of fusion proteins with these recombinantly expressed capsids sparked the wave of interest in using VLPs as conjugate vaccine platforms. The drawbacks of these VLPs are that as the natural viruses can infect humans, the human immune system may be pre-exposed to the VLP. The previous exposure to the carrier may hinder the immune response towards the vaccine[26,27]. This underlying phenomenon is called carrier induce epitope suppression (CIES). While CIES is an observed phenomenon, it does not seem to be a major factor for every VLP in determining therapeutic efficacy[27–30].

Concerns over CIES have led to interests in VLPs based on viruses that do not infect humans. An example is the bacteriophage Qß, which naturally infects E. coli. In nature, Qß is a T=3 icosahedron made of 178 copies of a 14kDa coat protein (CP), one copy of a maturation/lysis protein (required for infection and cellular escape), and an RNA-polymerase[31,32]. A non-infectious version of the Qß VLP can be formed by recombinantly expressing just the coat protein in E. coli. In such a system, 180 copies of the CP make up the capsid and incorporates RNA from Qß and E. coli. The RNA content is consistent from batch to batch with an average of 3 strands with a length of ~800nt/strand [33]. Expression levels are also relatively high, reaching 30–50mg/L of culture media range in laboratory scale shake flasks[34,35]. The non-infectious capsid can accommodate a wide range of modifications, either with covalent conjugates, non-covalent packaging using the RNA/REV tag, or fusion proteins[34,36–39]. While the majority of these modifications have been aimed at eliciting humoral immunity, there has been evidence of a cellular immune response[40,41].

Similar to Qß, most VLPs are aided in assembly by interactions between RNA and proteins, usually via a hairpin loop in the RNA with the interior of the CP[18,42]. A small minority, typically double stranded DNA (dsDNA) viruses, utilize interactions between a scaffold protein and the CP. One of the most common examples is the enterobacteria phage P22. The recombinantly expressed VLP is a T=7 icosahedron made up of 420 copies of the CP and ~250 copies of the scaffolding protein[43]. The scaffold protein is amenable to a large range of modifications allowing for easy loading of the capsid interior[44].

Besides recombinant expression in E. coli as used in the previous two examples, another common expression vector is plants. For example, the Tobacco Mosaic Virus (TMV), which is a 304±5 nm long rod-shaped virus made up of >2,100 repeating copies of its 17.5kDa CP, can be expressed by transfecting N. benthamiana plants, a close relative of tobacco, in yields of >7mg/g of infected tissue[45,46].

There are other VLPs that are being explored as potential carriers. Information on them can be found in Table 1. As they come up in this review, more information about them is provided as well.

Table 1.

Common VLPs, their natural targets, structural features, production methods, and the antigenic targets explored.

VLP		Natural Targets	Recombinant Structural components	Shape and Size	Production Method	Antigens Targeted	Citations
Bacteriophage Qß		E. coli	180 copies of CP, RNA	Icosahedral (T=3)xs	Recombinant Expression of the CP in E. coli.	MUC1, p33, D2, Tn, Streptococcus pneumoniae, OVA, PCSK9, ApoB, CETP, HPV, Tau, Nicotine, GD2	[12,14,35,40,41,48,52–59]
Enterobacteria phage P22		Salmonella typhimurium	420 copies of CP, ~250 copies of scaffold protein, dsDNA	Icosahedral (T=7) 60nm diameter	Recombinant expression of CP and scaffold protein in E. coli	M/M2 proteins of RSV	[43,60]
Tobacco Mosaic Virus (TMV)		Plants related to the tobacco family	>2100 copies of CP, RNA	Rod-shaped 304±5nm in length	Transient transfection of N. benthamiana	OVA, GFP, p15e melanoma	[45,46,61]
Hepatitis B	Surface Antigen (HBsAg)	Homo sapiens		22nm diameter	Yeast expression	HPV E7, SARS-CoV-2	[20,62]
	Core Antigen (HBcAg)	Homo sapiens		29.4nm diameter	Cell Free Protein Synthesis	Lymphocytic Choriomeningitis virus (LCMV) p33, Toxoplasma gondii, FLAG, D2	[53,63–67]
Papillomavirus Family		Varies	Varies	Varies	Varies	HPV E7, HIV gp160, Hen egg lysozyme	[68–70]
Pseudomonas Phage PP7		Pseudomonas aeruginosa	180 copies of CP, RNA	Icosahedral (T=3) 25nm diameter	Recombinant Expression of the CP in E. coli.	HPV, Staph. aureus	[15,71,72]
Molony Murine Leukemia Virus (MoMLV)		Mus musculus	Gag, Env, RNA, lipid bilayer	200nm diameter	Transient expression in mammalian cells	Glycoprotein epitope 33–41 of LCMV, Zika virus	[73–75]
Rabbit Hemorrhagic Disease Virus (RHDV)		Oryctolagus cuniculus	180 copies of VP60, RNA	Icosahedral (T=3) 32–44nm diameter	Recombinant Expression of VP60 in E. coli. or Baculovirus	OVA, MART-1 melanoma antigen, Mel888 tumor lysate	[76–82]
Mosaic Virus Family		Varies	Varies	Varies	Varies	p33, gp100, Influenza M1, M2	[83–86]
MS2		E. Coli	180 copies of CP, RNA	Icosahedral (T=3) 26nm diameter	Recombinant expression in Yeast	Prostate Cancer, HPV	[87–89]
Human Immunodeficiency Virus-1 Gag		Homo sapiens				HIV gag, pr55 gag,	[90,91]
Parvovirus		Varies	Varies	Varies	Varies	OVA, LCMV	[69,92]

Besides enhancing the immunogenicity of epitopes conjugated to them, VLPs’ inherent stability can be an advantage over other vaccine technologies. There are a growing number of examples of VLP-based vaccines retaining their immunogenicity after drying and prolonged storage[47–50]. Unlike other next generation vaccine platforms, such as mRNA vaccines which require ultra-cold temperature and the associated cold-chain, dried VLP-based vaccines would be easier to deliver and administer in low resource environment[51]. Although this discussion has focused on the VLP-based vaccine stability in terms of storage, their stability in vivo may also play an important role in the efficacy of VLP-based vaccines. Unfortunately, there is a paucity of evidence about the stability and half-lives of VLPs in vivo. More research is needed to explore VLP stability in vivo and its effect on the immune response.

4. The Innate Immune Response:

4.1. A primer on the innate immune response:

As a field, innate immunity is well-reviewed[93–96]. This section will thus be a short primer on the key events in the innate immune response and its role in kickstarting adaptive immunity. Upon injury or insult, such as an injection, local cells release chemokines that recruit antigen presenting cells (APCs) to the site. Once there, APCs, such as dendritic cells (DCs) and macrophages (MØs), sample the extracellular and intracellular (released from leaking or dying cells) environment for potential antigens. They collect information from the external environment through a mixture of phagocytosis, pinocytosis, and receptor mediated endocytosis. Unlike the receptors we will discuss in relation to cellular and humoral immunity, the receptors on innate immune cells recognize broad patterns and are not unique to each individual cell. Activation or maturation of the APCs occurs through activation of pattern recognition receptors (PRRs), such as toll-like receptors (TLRs), stimulator of interferon genes (STING), or NOD-like receptors (NLRs) [97–100]. These PRRs have evolved to recognize highly conserved pathogen associated molecular patterns (PAMPs) or molecules released by damaged cells (the Damage-Associated Molecular Patterns—DAMPs). PAMPs range from small molecules, like lipopolysaccharides (LPS) or cyclic GMP-AMP (cGAMP), to large repetitive protein patterns. DAMPs can include lysosomal proteases, DNA, mitochondrial DNA, adenosine triphosphate, and High Mobility Group Box 1 (HMGB1). Different PRRs are found in different regions of the cell ranging from the cell surface (TLRs 1, 2, 4, 5, 6, & 11), to sub-cellular compartments (STING, TLRs 3, 7, 8, 9), and the cytosol (NLRs)[101]. Following antigen uptake, some activated APCs attempt to control the localized infection while others migrate to the lymph nodes to activate the adaptive immune response.

4.2. Virus-like particle-based activation of the immune system:

Canonically, the belief was that VLPs do not activate the immune system until they have entered immune cells. There is growing evidence that VLPs activate the immune system earlier than previously believed. Using Heat Shock Protein (HSP) VLPs, which do not contain any genetic material, Richert and colleagues reported that intranasal treatment of mice with the VLPs primed DCs and alveolar MØs in the lung[102]. Further studies using other VLPs including P22 revealed this priming was occurring via a TLR2 dependent mechanism (figure 1a)[103]. TLR2 is located on the cell surface and recognizes repeating protein patterns regardless of sequence. This expands to repeating protein structures regardless of whether it is in a spherical or linear repeating pattern. Although both spherical and linear patterns are recognized by TLR2, they activate separate activation pathways, by coactivation of TLR6 and TLR1 respectively[103]. Following the uptake of VLPs by APCs and activation of the APCs, the APCs migrate to the lymph nodes allowing for the interaction with the adaptive immune system (figure 1b). This includes skin-derived DCs which have been identified as the initial transporter of herpes simplex virus to lymph nodes for CTL priming[104].

Figure 1.

(A) The repetitive surface of virus-like particles (VLPs) first activate antigen-presenting cells (APCs) through toll-like receptor (TLR) 2/6 heterodimers located on the cell surface. This activation causes APCs to enter a temporary state of alert. Following uptake of the VLPs, into the endosome, the capsid is disrupted releasing the VLPs genetic material. The released genetic material can then activate TLR 9 homodimers (DNA) or TLR 7/8 heterodimers (RNA). Activation of APCs through TLR 9 or TLR 8/7 induces a more potent activation of the APCs. (B) Upon injection VLPs can be up taken by APCs which then migrate to the lymph nodes and present the VLPs or their conjugated antigens to the cells of adaptive immune system. VLPs can also efficiently drain from the injection site to the lymph nodes due to their size.

After the initial activation and recruitment of APCs, the VLPs are phagocytosed. This phagocytosis occurs through non-specific pathways including macropinocytosis, micropinocytosis, and phagocytosis[69,81]. In addition, VLPs derived from viruses that infect humans may enter the cells through receptor mediated pathways. In the phagolysosomes, the acidic environment and proteases break down the protein shell exposing the genetic material of RNA or DNA encapsulated, which is recognized by TLRs 7/8 or 9 respectively. Comparatively, the activation signals from TLRs 7/8 or 9 appear more important for the activation of an adaptive immune response. This has been evidenced by the short protective window obtained by the innate immune activation and weaker overall immune responses when using VLPs without the genetic material[103].

As some VLPs do not contain or can be prepared without the genetic material, there is an interest in overcoming the weak activation without TLR7/8 or 9 activation[105–108]. The simplest approach is to add an exogenous adjuvant to the vaccine formulation. To make sure the adjuvant affects the same cells that take up the VLPs, adjuvants have been covalently conjugated to or packaged in the VLPs. On the other hand, there is evidence that covalent linking of antigen and adjuvant to the same molecule may not be necessary, as the VLP is the major factor in which cells uptake the particles[41,109]. Conjugation to the same type of VLP should result in similar uptake levels of VLPs loaded with antigen vs. VLPs loaded with adjuvant. One of the more commonly utilized adjuvants, for cytotoxic T lymphocyte (CTL) based vaccines, has been synthetic oligodeoxynucleotides containing unmethylated cytosine and guanine (CpG). CpG activates TLR9, creating a T_H1 biased response that leads to strong B cell and T cell responses[110]. Another advantage of CpG is that, as a DNA analog, it can be used to replace the naturally occurring genetic material in VLPs[111]. This replacement can usually occur because for certain VLPs, such as Qß, assembly is dependent on interactions between the negatively charged phosphate backbone of the genetic material and the interior of the CPs. Substituting the genetic material with a well-characterized adjuvant has a further potential advantage. The genetic material packaged inside Qß may be well determined for size, length, and quantity, as compared to the host RNA encapsulated, the sequences of which are not well characterized[33]. That could pose a problem for regulatory approval for clinical applications due to batch-to-batch variability. A work around is replacing the genetic material with CpG. Although as with any solution, it has its pros and cons. CpG has a good safety profile. However, it can still lead to the overactivation of an inflammatory immune response causing immune dysregulation including cytokine storms which can be fatal[112]. Depending on the cells targeted, a potentially safer alternative is poly(I:C), a TLR-3 ligand[113,114].

Besides adding adjuvants that activate the entire immune response, with the concern of potential severe over-activation, there has been work on engineering VLPs to activate only one type of immune cells. Because of their large size, VLPs drain from the site of infection into the lymphatic system and end up in the draining lymph nodes[115,116]. This is part of what makes them excellent at eliciting strong humoral responses. It also can be advantageous for cellular immunity, as naïve T cells are located in the paracortex along with DCs[117]. There, VLPs that display activation ligands for co-stimulatory receptors on T cells can bind their cognate receptors boosting activation and differentiation. Derdak et. al. showed that Moloney murine leukemia virus (MoMLV) VLPs decorated with a TCR/CD3 ligand, CD80, and ICAM-1 can cause differentiation in T cells without the help of APCs[74]. Further experiments showed that the replacement of TCR/CD3 ligand with MHC class I molecules containing preprocessed antigens, can elicit antigen specific T cell responses. Although highly intriguing, this approach is specific to the VLP used.

Most VLPs reported in the literature are non-enveloped, meaning that they do not contain a lipid bilayer coating the exterior of the VLP. MoMLV VLPs have an envelope that the researchers used to anchor the ligands through glycosylphosphatidyl inositol (GPI)[118]. Without the bilayer, these ligands would have to be either covalently linked or fused to the CPs. Although VLPs are known for their stability, there are limitations to VLP stability to fusion proteins, and conjugates have to be evaluated on a case-by case basis, especially if more than one ligand is to be fused to the capsid[72,119]. This is not to exclude the possibility that injection and infiltration to lymph nodes could result in some VLPs providing co-stimulatory signals while APCs process and display the antigens. However, there is evidence that some co-stimulatory signals may be the result of size-dependent discrimination from the point of contact between the APC and the T cell[120,121]. As such, care should be taken in ligand choice. Fusion or covalent linkers could change the spacing between the VLP’s surface and cell surface effecting what ligands are excluded.

Another consideration to take is if the researcher is trying to elicit a total immune response, attaching T cell epitopes on the external surface may interfere with B cell responses, as B cell receptor signaling is highly dependent on an organized display of B cell antigens[122]. To overcome this, a general design is to use the surface of the VLP to display B cell antigens and co-stimulatory signals needed to enhance the B cell responses, while functionalizing the interior with CTL epitopes (CTLes) or adjuvants. An example of why this is important comes from the work of Alam et. al[56]. Aryl mannosides conjugated to the surface of Qß altered the fate of the immune response through activation of DC-SIGN and enhanced the cellular immunity. However, the boosting of the cellular immunity came at the expense of the humoral immune response, as one would expect due to the interruption of the B cell epitope display by the aryl mannoside ligand[56].

5. Cellular Immunity:

5.1. A primer on cellular immunity:

There have been many excellent reviews of cellular immunity and of its individual components[123–127]. As such, this section will only be a quick primer on the major steps of the cellular immune response. The first step in any immune response is the activation of innate immune cells including APCs. These APCs migrate to the site of infection and sample the extracellular and intracellular (released from apoptotic cells) environment for potential antigens. Afterwards some APCs, usually a mixture of MØs and DCs, migrate to the lymph nodes. The sampled antigens are processed in order to be displayed to naïve T cells in the lymph nodes. Once in the lymph nodes, the APC’s display the CTLes using major histocompatibility complex (MHC) class I molecules. If a T cell receptor (TCR) on a naïve T cell recognizes the MHC:CTLe complex, a maturation signal is received by the T cells. As a result, the activated T cells undergo clonal expansion creating a pool of effector T cells. These CTLs then migrate to the site of infection, which then sample the cells in the surrounding tissue. Nearly all cells in the body produce MHC class I complexes and use them to display intracellular peptides as a signal of the cell’s health status. When CTLs find cells displaying MHC:CTLe complexes that their TCRs recognize, they release granzyme and perforins to kill such cells by disrupting the cell membrane and causing apoptosis. After the initial infection is controlled, the large number of CTLs produced during clonal expansion is no longer needed. The CTL population undergoes a contraction, leaving behind memory subpopulations in case of repeat exposure to the same pathogen.

5.2. Presentation of cytotoxic T cell epitopes by antigen processing cells:

Because most VLP-based vaccines targeting T cell responses do not attempt to directly activate T cells, attention must be paid to the cross-presentation pathway in APCs. Cross-presentation is not as simple as the name suggests. It is not a single pathway but is a combination of multiple pathways that result in the same outcome, i.e., fragments of proteins of extracellular origin presented on MHC class I molecules. Figure 2 shows the complex nature of cross-presentation. The first step regardless of pathway is phagocytosis of the VLP by the APC. The phagosome then fuses with the lysosome creating a phagolysosome. Here is where the pathways diverge. There are three main cross-presentation pathways with each having minor pathways, which are still being fiercely debated in the literature[126,128–130].

Figure 2.

In conventional antigen presentation (left) cytosolic proteins are degraded by the proteosome. The resulting short peptides are transported to the endoplasmic reticulum (ER) by transporter associate with antigen processing (TAP) proteins. There the peptides are loaded onto new major histocompatibility complex (MHC) class I molecules and sent to the cell surface by way of the Golgi. During cross-presentation (right) extracellular proteins are phagocytosed. Inside the phagosome, enzymes can degrade the captured protein into small peptides. The peptides can then be loaded onto either recycled MHC class I molecules from the cell surface or new MHC class I molecules delivered to the phagosome by endoplasmic reticulum-Golgi compartments (ERGICs) containing Sec22b. Other pathways include the protein or peptides escaping to the cytosol where they are processed by the proteosome. Following proteasomal processing, the peptides can enter the conventional antigen presentation pathway or return to the phagosome via TAP proteins transported to the phagosome via ERGICs. Regardless of pathway once the MHC class I is loaded with the cytotoxic T lymphocyte epitope (CTLe) it is sent to the cell surface.

In the conventional antigen presentation (left side of figure 2) of MHC class I molecules, cytosolic proteins are processed into short peptides (~8–10 a.a.). Subsequently, they are transported by the transporter associated with antigen processing (TAP) proteins to the endoplasmic reticulum (ER), where the peptides are loaded onto MHC class I molecules and shipped to the cell surface.

The simplest form of cross-presentation (right side of figure 2) mimics the conventional antigen presentation process closely. After intake to the phagolysosome, proteins or peptides escape into the cytosol and are further processed by the proteasome. Then they are transported to the ER, loaded onto MHC molecules and displayed on the cell surface. While this seems straightforward, there are competing pathways out of the phagolysosome. pH changes and reactive oxygen species (ROS) can cause “leaky” endosomes allowing for passive escape. Meanwhile, active transport out of the phagolysosomes has been shown to occur through Sec61[131]. To aid in escape, there have been attempts to increase passive escape by utilizing cell penetrating peptides fused to the antigenic peptide[132].

Another pathway is through recycling of MHC class I molecules. As the interior membrane of the phagolysosome used to be the cell surface, proteins from the cell surface including MHC class I molecules can be trapped in the phagolysosomes. In the phagolysosomes, proteins are processed into the short antigenic peptides, which can then be loaded directly onto MHC class I molecules in phagolysosomes and presented onto the cell surface again.

The final “main” pathway involves ER-Golgi intermediate compartments (ERGICs). These ERGICs are used to transport cargo from the ER to the Golgi[133]. These compartments are believed to be directed to the phagolysosomes through Sec22b, where they can deliver new MHC class I molecules, TAP proteins, and other transporters[128,130,134]. This allows the phagolysosomes to act similarly to the ER in MHC loading.

Which of the three aforementioned pathways dominates is up for debate, as is which pathway is important for antigens delivered by VLP-based vaccines[126,135]. There is very little work discussing cross-presentation of antigens delivered by VLPs. The scant information indicates the results may depend on the VLP utilized. Work from Win et. al. suggests it is through endosomal recycling of MHC class I molecules, by exploring cross-presentation of ovalbumin (OVA) and human melanoma-associated antigen (MART-1) delivered by rabbit hemorrhagic disease virus (RHDV)-based VLPs[81]. To study cross-presentation, inhibitors of specific pathways were utilized, which include lactacystin to inhibit proteasomal processing, US6 to prevent TAP-mediated transport, brefeldin A to inhibit vesicle secretion, bafilomycin A1 to prevent phagolysosome acidification, and primaquine to inhibit recycling of cell surface molecules. The final three inhibitors act at different stages of endosomal recycling pathway and virtually erased cross-presentation as measured by activation of antigen-specific CD8⁺ T cells. This was in direct contradiction to when soluble antigens were used. With soluble antigens, each inhibitor had some effects on the activation efficiency but did not completely abolish cross-presentation. This suggests that conjugation to VLPs shifted the importance of cross-presentation pathways to the endosomal recycling pathway. This was further supported by earlier work by Leclerc et. al. who used a similar lactacytsin based experiment to rule out proteasomal processing of antigens delivered by Papaya Mosaic Virus (PapMV) VLPs[86]. On the other hand, porcine parvovirus (PPV) VLPs loaded with OVA CTLes were cross-presented in a TAP and proteasome dependent, endosome-to-cytosol pathway[69]. Between the two extremes is the evidence provided by Ruedl et. al. They showed that DCs in TAP-deficient mice exhibited decreased cross-presentation efficiency compared to wild type mice when treated with Hepatitis B core antigen (HBcAg) VLPs but still retained the ability to activate potent T cell responses[64]. While each study used different sets of antigens overall, there were enough overlaps to suggest the different cross-presentation fates were likely due to the VLPs used in the studies. Taken together, these results indicate that the VLPs are biasing cross-presentation toward a certain pathway, but the exact mechanism is unknown. Size might be a determining factor as microspheres displayed size-dependent changes to cross-presentation efficiency when the proteosome was inhibited[136]. However, the smallest microspheres (d= 800nm) examined are the size of some of the largest VLPs known. This suggests that the differences described above are likely due to more than just size, as the RHDV (d= 32–44nm), PPV (d = 25–30nm), and HBcAg (d= 29.4nm) VLPs are an order of magnitude smaller than the microspheres[63,77,78,137].

Despite these initial studies, more research is needed into the best design practices for optimizing or directing cross-presentation. This is a currently underexplored area of the literature, probably because after delivery to APCs, cross-presentation is a fairly robust process due to redundant minor pathways. A deeper understanding may lead to better designed and thus more effective vaccines. There are some studies into linker effects and epitope length (full length protein, epitope only, or extended epitopes) [138–143]. For example, with RHDV VLPs loaded with gp100 CTLes, the two linkers utilized did not appear to play a role in the observed changes in immune response[138]. Instead, the differences were likely due to the decreased stability of the fusion proteins caused by increasing number of hydrophobic CTLe repeats. Studies on non-conjugated subunit vaccines show that extended epitopes work best because they can escape lysosomes better/faster than the full protein, and yet are large enough to possibly contain sequences that aid in trafficking to the ER while the minimal peptide epitope does not[141–143]. These studies were aimed at free peptides only, so how conjugation or fusion to VLPs affects their efficacy needs to be established. Current studies are an excellent start, but deeper dives into the mechanisms of VLP-based conjugate vaccine cross-presentation would be very beneficial to guide further design[64,81,86].

6. Humoral Immunity:

6.1. A primer on humoral immunity:

Similar to the previous sections on cellular and innate immunity, this section will just be a brief primer because of the availability of numerous reviews focused solely on each sub-topic[144,145]. The main effectors of humoral immunity are antibodies, which are produced by B cells. To activate naïve B cells, pathogens must migrate to the lymph nodes. There surface antigens can be recognized by antigen specific B cell receptors (BCRs) on the cell surface. As BCRs are activated, they cross-link magnifying the maturation signals, which stimulate the B cell to proliferate and secrete pentameric immunoglobulin M (IgM) antibodies[146]. These antibodies are relatively low affinity and short lived, which are only the first step in the humoral response. During the activation process, the B cell engulfs the antigen and processes it, with the end product being MHC class II antigens presented on the B cell surface. If the B cell encounters a matched CD4⁺ helper T cell with TCRs that recognizes the MHC Class II:peptide complexes on the B cell surface, it can undergo isotype switching, changing the subtype of antibodies produced from IgM to the other Ig isotypes. The specific isotype produced depends on the class of helper T cell, and the resulting cytokines they release. During isotype switching, the B cells also undergo somatic hypermutation. This is a rapid mutation of the gene encoding the variable region of the antibodies that leads to an increased binding affinity between the antibody and its cognate antigen. Following isotype switching, the B cell fully matures into a plasma cell and undergoes clonal expansion. The high affinity antibodies are then secreted into the bloodstream or mucosal barriers depending on the isotype. The effects the antibodies can have on the pathogen differ depending on their antigenic targets. They can prevent key receptor binding interactions (neutralizing), increase phagocytic clearance by MØs (opsonizing), trigger complement mediated cytotoxicity (CDC), or cause the formation of antigen-antibody plaques that are filtered from the bloodstream. Once the infection is under control, the plasma cells will undergo apoptosis. However, the memory subset of B cells will remain and can quickly reactivate and expand upon a secondary infection.

6.2. B cell activation:

Naïve B cells reside mainly in the spleen and lymph nodes. In order to activate them, vaccines must be targeted to the lymphatic tissue. As discussed earlier, the sizes of VLPs are well-situated to track to lymph nodes (figure 1b). VLP’s fate can be further controlled by the method of administration. Cubas et. al. utilized SHIV VLPs (SIV Gag + HIV SF162 Env) to test the immune responses following injections at various common injection sites[116]. VLPs were labeled with an IR dye and lymph nodes were harvested 24hr post injection. Intraperitoneal injection showed no VLPs in the lymph nodes and the amount of VLP in the lymph nodes increased going from subcutaneous to intramuscular and finally intradermal injection. The intradermal injection had detectable VLP levels in all four lymph nodes harvested. VLP levels in the lymph node correlated positively with the IgG titers and CTL efficacy. Therefore, caution should be taken when comparing immune responses across various studies. A relatively weak immune response could be due to the choice of the injection method and the resulting fate of VLPs tracking to lymph nodes. Additionally, the injection method can skew the class of Ig elicited. Intranasal immunization using a Qß-based vaccine against influenza was the only immunization method that elicited strong local IgA titers in the lung[147].

Once in the lymph nodes, antigens must find naïve B cells that express BCRs capable of binding them. Binding of an antigen with a single BCR is often not sufficient for activation of naïve B cells (figure 3a). Cross-linking of multiple antigen-bound BCRs can induce a more potent response[144,148]. VLPs, because of their repetitive surface structures, allow for both high valency and high density of antigen display (figure 3b). Use of polymers loaded with various antigens have shown that there is a minimum valency (~10–20) for effective B cell activation[149]. Most VLPs contain >100 subunits so even at a 1:1 antigen:subunit ratio, the valency threshold can be readily reached. The issue comes from the density of the antigens on the VLP surface. There seems to be a sweet spot for antigen density, which makes sense because of the size of BCRs and their need to cross-link. At a certain point with increasing antigen density, there is a diminishing return as antigens become too clustered for BCRs to bind due to steric clashes. On the other hand, if the density is too low, the distance between antigen bound BCRs becomes too large for efficient cross-linking.

Figure 3.

Organized antigen display is critical for B cell receptor (BCR) cross-linking. (A) Soluble antigens activate BCRs in a random fashion leading to little BCR cross-linking and thus a weak immune response. (B) Conjugation of antigens to a virus-like particle (VLP), such as Q?, organizes the display of antigens. The organization increases the density of the activated BCRs leading to increased cross-linking and thus a stronger activation signal. This results in a more robust immune response. (C) Creation of immune synapse leading to BCR cross-linking and antigen processing. (1) Unbound BCRs do not interact with the native actin cytoskeleton; (2) BCR recognition of antigens leads to rearrangement of the actin cytoskeleton; (3) Myosin uses the remodeled actin cytoskeleton to pull antigen bound BCRs together; (4) Upon reaching a critical mass of antigen bound BCR, an intracellular vesicle containing a bud from the VLP envelope is formed; (5) The captured viral material is processed, and any captured major histocompatibility complex (MHC) Class II restricted antigens are presented on the B cell surface.

There are further biological reasons behind why antigens conjugated to VLPs can efficiently activate the immune system. Many viruses and bacteria have limited genomes, thus their structures are made up of densely repetitive structures. As a result, the immune system is sensitized toward such a PAMP, i.e., repeated patterns of epitopes spaced ~10 nm apart, or roughly the space between antigen binding sites on antibodies[150,151]. Because this PAMP is associated with many viruses, it makes sense that most VLP coats are ideally arranged to activate B cells[152]. This pattern is so strong that it is transitive to conjugated antigens, including antigens usually selected against in B cell development[40,52,67,70].

For enveloped VLPs, density may play a less important role. Recent efforts in B cell activation research have been directed at identifying the mechanism of B cell activation by APC mediated antigen presentation. APCs and B cells form an immunological synapse mediated by monomeric BCR antigen interactions. These monomeric interactions signal for reorganization of the B cell actin cytoskeleton. The rearranged cytoskeleton and myosin pull antigen bound BCRs towards the center of the synapse causing the oligomerization required for B cell activation (Figure 3c)[153,154]. It is not hard to imagine a similar mechanism being possible with enveloped VLPsx where the antigen is attached to the viral lipid bilayer, although this idea needs to be explored and validated[74,118].

Even more important to vaccines targeting humoral responses over cellular responses is the choice of the linker. Because they are processed prior to presentation, T cell epitopes are shown in a pristine manner. B cell epitopes on the other hand are usually presented to B cells prior to processing. This means that if the linker used is more immunogenic than the target epitope, the immune response may be redirected toward the linker instead of the target antigen. One prominent example is the triazole linker produced by the copper catalyzed azide-alkyne cycloaddition (CuAAC) reaction[10,155]. While the CuAAC reaction is a favorite for biological conjugation, including vaccines, due to its specificity and high efficiency, the triazole ring formed competed against the antibody generation toward the target cancer antigen[155]. As a general rule of thumb, acyclic and flexible linkers are preferred to reduce the anti-linker immune responses and the resulting interference to the desired immune responses.

A recent development in conjugation techniques are non-covalent conjugation approaches, taking advantage of peptide or RNA binding sites. A commercial phage display peptide library was used to identify a 7-mer peptide capable of binding to the surface of the cowpea mosaic virus (CPMV)[156]. Another example is the use of the HIV Rev tag to load proteins on the interior of Qß via the RNA involved in capsid formation[34,39].

7. Immunological memory and long-term efficacy:

7.1. Long-term cellular immunity

Once the initial immune response eliminates the infection, there is less of a need for the large number of CD8⁺ T cells. As a result, the pool of T cells undergoes a contraction[157]. While the pool of antigen specific T cells decreases, two main subsets of CD8⁺ T cells remain, i.e., effector memory (T_EM cells) and central memory (T_CM cells)[158]. T_EM cells circulate throughout the body. Upon subsequent infection, they are recruited to the site of infection and provide the initial defense through the classical contact mediated cytotoxicity, until a larger immune response can be mustered.

While the T_EM cells are preventing the establishment of a beachhead by the infection, APCs are sampling the infection and processing the antigens. When they reach the lymph nodes, they are able to activate T_CM cells. These T_CM cells are considered one of the more important factors in long-term cellular immunity. Upon reactivation, they undergo a process similar to clonal expansion, replicating the initial cellular immune response. The daughter cells from this population expansion are then able to differentiate into different T cell subsets. After the infection is under control again, the antigen specific T cell population undergoes another contraction leaving behind new/rejuvenated memory T cell sub-populations.

In the last decade a new subset of memory T cells, the tissue resident memory (T_RM), has been characterized[159]. The hallmarks of such cells are not cell surface markers but rather mainly the fact that they are found in barrier tissues, such as the lungs, skin, reproductive tracts, etc. and do not circulate[160,161]. In fact, T_RM cells can express different cell surface markers depending on the type of tissue they are in[162]. These cells are thought to play an important role in reactivation of the immune response to subsequent infections. VLP-based vaccines are able to elicit T_RM cells[163]. For example, intranasal immunization of mice with P22 loaded with M and M2 proteins from the Respiratory Syncytial Virus (RSV), fused to the P22 scaffold protein, was able to elicit T_RM cell populations in the lungs[60]. These cells were detected by flow cytometry analysis of bronchioalveolar lavage fluid up to 2 months post inoculation. Importantly while there was a small decrease in total cell counts for both M- and M2-specific T_RM cells, it was still protective as measured by lung viral titers, after re-challenge.

The T_RM mediated immune response to the secondary infection is due to a reverse of the traditional flow of information. As previously discussed, innate immune cells traditionally are the first cells to interact with pathogens through their recognition of PAMPs. Then they send signals to the adaptive immune system. Upon subsequent infection, T_RM cells recognize their cognate antigen. Instead of direct killing of the infected cells, they release cytokines (VCAM-1, IFN-γ, and TNF-α) to recruit innate and adaptive immune cells including DCs, T_EM cells, and memory B cells[161,164,165]. This creates a tissue-wide state of alert for infection[166].

While T_RM cells seemingly play an important role in long-term cellular immunity to infectious pathogens, there is evidence that they also may play a role in immune regulation of cancers. As their main function is not direct killing of infected cells, it makes sense that they would express higher levels of inhibitory ligands/receptors including PD-1 and TIM-3. However, after the use of anti-PD-1 antibodies, they regain their cytotoxic abilities. This could explain the outsized effectiveness of anti-PD-1 therapies in some individuals and the overall correlation between T_RM cell populations and cancer prognoses[165,167,168]. Elicitation of T_RM cells then should be measured post vaccination regardless of target, cancer or pathogen. As such, even though one of the draws to the use of VLPs as carriers in conjugate vaccines is the lack of a need for exogenous adjuvants, if the T_RM response is not strong enough for protection, adjuvants can be added to the vaccine formulation to boost the response. Of particular interest are zymosan, IL-1ß, or compounds to enable targeting certain DC subsets, most of which are being explored as adjuvants in VLP-based vaccines[169–172]. For further discussions about engineering vaccines to elicit T_RM cells, interested readers are referred to a recent review by Knight and Wilson[173].

Beyond initial T cell responses, there is a paucity of evidence about the length of memory and protection from cellular immunity elicited by VLP-based vaccines, especially when compared to long-term antibody-based immunity. This makes sense due to the methodologies available for measuring immune responses between the two. For antibodies, the most commonly analyzed isotype is IgG, which can be easily measured through blood draws. Other non-circulatory isotypes can be measured through non-lethal means[174]. In comparison, the measurement of antigen-specific T cells populations, particularly the important T_CM cells, requires sacrifice of the subject to collect and analyze the lymphatic organs. This prevents longitudinal studies of a subject’s immune response over time, and the need to collect samples from multiple time frames using multiple subjects can quickly escalate the population sizes and costs associated for the study. As such, there is a need to develop non-terminal methods of T cell analysis. Here lies an opportunity for VLPs to be utilized for such evaluations. One potential method for antigen-specific T cells could be to utilize VLPs functionalized with a tracking dye and MHC class I molecules loaded with the relevant peptide. Because VLPs preferentially drain to the lymph node, they should be able to find and bind to antigen specific T_CM cells. While further development is needed to yield quantitative analysis results, as a non-lethal alternative, it is intriguing.

7.2. Long-term humoral immunity

Initial activation of B cell leads to the expression of IgM, a pentameric isotype associated with the initial immune response and short lived-immune response. In order to generate a long-lived immune response, the B cells need to undergo isotype switching, which requires a costimulatory signal from a CD4⁺ helper T cell. Here is the advantage of VLPs over other multivalent vaccine carriers. As immunogenic proteins themselves, VLPs contain peptide sequences identified as helper T cell epitopes[175]. After formation of the BCR micro clusters, the antigens bound to the BCRs are endocytosed and further processed to present helper T cell epitopes. Helper T cell epitopes are traditionally restricted to peptides. If the desired antibody target is not a protein, then without an additional helper T cell epitope formulated into the vaccine, the immune response will be suboptimal[55,145]. VLPs avoid the need to include an additional factor, i.e., an exogenous helper T cell epitope, into vaccine design. Once the VLP is endocytosed and processed, the helper T cell epitope is presented on the B cell surface by MHC class II molecules. Follicular helper T cells in the lymph nodes then recognize the peptide:MHC complexes on the B cell. The resulting costimulatory signals trigger isotype switching, somatic hypermutation, and clonal expansion resulting in a large number of IgG secreting plasma and B cells.

As with T cells, once the infection is eliminated most of the antibody generating plasma cells are eliminated. This leaves behind a small population of memory B cells capable of quickly responding to secondary infection. Even after the contraction, there is a small population of plasma cells that remain and secrete antibodies. VLP-based vaccines generate particularly high long-term titers, suggesting they are uniquely capable of eliciting long-lived plasma cells, although the exact reasoning is unknown. There is some evidence that factors beyond valency and density may be the cause[176,177]. Upon re-infection, the memory B cell population can expand and differentiate into new plasma and memory cells. Memory B cells elicited by VLP-based vaccines are able to respond to secondary challenges as measured by IgG titers following a long-term booster[59].

8. Concluding thoughts:

Overall, VLP-based conjugate vaccines pose an attractive solution to current issues in vaccine development. The variety of different VLPs, conjugation strategies, antigenic targets, and adjuvant choices present a great opportunity to fine-tune the immune response. However, there is a paucity of mechanistic understanding to guide the best practices in VLP-based vaccine design. A majority of the articles published in this field are of the proof-of-concept genre. The development of new techniques and knowledge about cellular mechanisms in the general field of immunology offer an exciting opportunity in VLP-based vaccine design. It is time to move beyond proof-of-concept to mechanistic investigations. We look forward to the availability of a powerful plug-and-play VLP-based vaccine platform, which can be rapidly deployed to address future disease outbreaks similar to SARS-CoV-2 or to personalized cancer vaccines. There is much for this field to explore but the future is bright.

9. Expert Opinions:

While recent work has led to amazing advances in the field of virus-like particle-based vaccines, there is still more work to be done. It is time for the field to move beyond the current trend of empirical research design towards rational design based on solid theoretical and mechanistic understandings. An excellent example being the work done exploring cross-presentation of antigens delivered by VLPs [64,69,81,86]. Another example being the work exploring the effect of stereochemistry on antigen stability[57]. General advances in the field of immunology have allowed for a deeper understanding of the complex interactions of immune cells. VLPs have the potential to probe the immune responses due to their ability to accommodate a wide variety of functionalization approaches. Furthermore, the use of enveloped VLPs could act as “simplified” cells for investigating specific receptor interactions, while care would have to be taken to account for multi-receptor/ligand interactions, including size discrimination-based interactions, like the need for CD45 exclusion from the binding site, for efficient opsonization of by macrophages[121].

Due to their sizes, VLPs are well suited for imaging of the lymphatic system. Conjugation of ligands/receptors to the surface and functionalization with a wide variety of imaging probes to the interior could be utilized to monitor changes in the immune system in vivo. VLPs have been shown to be stable to functionalization with a wide range of biologically relevant molecules. Particularly, some VLPs are able to get large numbers of copies onto each capsid, in excess of 500 copies/VLP particle[35]. Development of carbohydrate analogues that exhibit increased specificity to one immune receptor than their natural form, such as HA, in conjunction with VLP particles could provide potent targeted imaging of the immune system[178]. The excellent safety profile of VLPs combined with higher specificity would potentially mitigate some of the toxicity associated with current imaging dyes/procedures by lowering the required dose of the imaging agent. This would be especially beneficial for the analysis of T cell responses, enabling longitudinal monitoring for each subject. The use of radioactive PET tracers, near-infrared (NIR) dyes, RAMAN dyes, or other imaging modalities that have more tissue penetrating depth than standard fluorescent spectroscopy, could allow for semi-quantitative comparison of the magnitude of antigen specific T cell population[179]. A major hurdle to overcome would be non-specific interactions between VLPs or conjugated ligands with random cells. There is some work looking at maximizing/minimizing VLP interactions with certain cells[180].

While the possibility of universal design rules for VLP-based vaccines are an attractive idea, looking at some of the contradicting reports in the literature and the complexity of known immunological pathways, there are likely to be design rules specific to each VLP[27,29,30,64,81,82,86,128–130,134]. Having knowledge of how different VLPs engage different immune pathways can allow for finer tailoring of the immune response. A pressing need is in the realm of personalized cancer vaccines. Some cancers downregulate MHC class I expression. Patients with such a cancer would benefit from a more robust humoral response, which could inform the decision on which VLP platform to build the vaccine on.

Previous reviews have called for the finding of a niche commercial application to drive virus-like particle-based vaccines through the rigorous process of clinical trials and regulatory approval[181]. Although three years is not a long time to see substantive change from when the review was published, the COVID-19 pandemic was a golden opportunity, seeing as it opened the door for another new type of vaccine platform (mRNA vaccines). As of January 2021 of 15 known preclinical VLP-based vaccines against SARS-CoV-2, 2 made it to clinical trials (Trial IDs: ACTRN12620000817943, NCT04450004) [62,182]. All great journeys start with a single step. Hopefully these candidates will open the door for more VLP-based vaccines to make the jump from the lab to the clinic.

Article Highlights:

• Virus-like particles (VLPs) are protein-based nanoparticles derived from viruses and recombinantly expressed in various systems (insect, mammalian, plant, and bacterial cells)

• VLPs are powerful carrier proteins capable of potently boosting immune responses against antigens loaded onto/into the VLP.

• Advances in understanding the immune system have helped explain why VLPs are capable of generating potent immune responses. At the same time advances in VLP-based vaccines have further elucidated the mechanisms underlying immune responses. This allows for the rationale design of next generation VLP-based conjugate vaccines.

• VLPs contain huge potential as a next generation plug-and-play vaccine platform allowing for rapid responses to novel diseases. Anti-SARS-CoV-2 VLP-based conjugate vaccines have made it to clinical trials suggesting these vaccine platforms are starting to gain traction.