INTRODUCTION

The basic concept of a vaccine is to trigger the host immune system and mount adaptive immune responses of sufficient magnitude and duration, including B-cell-mediated antibody production and/or specific T-cell-mediated cellular responses to a protective antigen(s) in order to prevent infection or reduce the related pathology. It is now well-known that successful vaccines should contain not only such a protective antigen(s), but also a good adjuvant that efficiently activates the innate immune system for optimal vaccine immunogenicity.

Table I. Adjuvants and Their Usage of Innate Immune Receptors
Full size table

It has been shown that toll-like receptor (TLR), one of the innate immune sensors, plays important roles not only in the initial proinflammatory responses, but also in the consequent antigen-specific immune responses, both of which are crucial for protective immunity against infectious diseases (1–3). A variety of immunostimulatory compounds, including protein, lipid, carbohydrates, and nucleic acids, have been shown to be TLR ligands and are currently being used experimentally or in clinical trials within vaccine formulations as an adjuvant. However, recent evidence has shown that conventional adjuvants such as aluminium hydroxide (Alum), incomplete and complete Freund's adjuvant (IFA or CFA), or unconventional adjuvant-containing vehicle such as apoptotic cells and virus, elicit efficient adaptive immune responses to vaccine in the absence of TLRs (4–6). Moreover, newly characterized intracellular innate receptors that sense a variety of immunomodulatory compounds, such as NOD-like receptors (NLR), RIG-like receptors (RLR) and yet unknown intracellular DNA receptors, have been demonstrated to activate the innate immune responses, and possibly the adaptive immune responses, in a TLR-independent manner (7–9). Thus, it is important for us to understand how these innate sensors or their downstream signaling pathway(s) mediate the adjuvant-induced innate and adaptive immune responses in order to develop potent, but also safe, vaccine or related immunotherapy. Here, we review recent advances in our understanding of the TLR-dependent and TLR-independent adjuvant activity of vaccine components.

Specific Delivery and Targeting of an Adjuvant to the Cognate TLR for Vaccine Potency and Safety

Most TLR ligands, especially, those that can be chemically synthesized or genetically modified, are now under development as candidate vaccine adjuvants ((10, 11) and Table I). These TLR agonists are very potent adjuvants in capacity of activating cells expressing the cognate TLR, in particular, dendritic cells (DCs), which are the key antigen presenting cells. DCs produce cytokines, chemokines, and interferons, and up-regulate their functions, including antigen processing and presentation to naïve T cells. It has been shown that coadministration of vaccine (antigen) and TLR agonist, in the form of direct conjugation, or incorporation into an efficient targeting vehicle (a delivery system such as a viral particle, liposome, or an attached antibody against a surface molecule on DCs) into antigen presenting cells via the endosomal pathway, is necessary for the optimal vaccine formulation (12, 13).

When we consider utilizing TLR-agonists as adjuvants in vaccine development, it will be important to appreciate that the intracellular localization of TLRs is quite distinct between subfamilies. While certain TLRs (TLRs 1, 2, 4, 5, 6, and possibly, 10 and 11) are expressed on the cell surface, others (TLRs 3, 7, 8, and 9) are found almost exclusively in intracellular compartments such as the endoplasmic reticulum and endosomes. Cell surface TLR1, TLR2, and TLR6 recognize lipoproteins, TLR4 recognizes lipopolysaccharide (LPS), and TLR5 (or TLR11, which is not functional in humans though) recognizes a pathogen-derived protein; by contrast, endosomal TLR3 and TLR7, TLR8, and TLR9 recognize nucleic acids (14, 15). The physiological meaning of these distinct expression patterns between cell types and intracellular compartments is yet to be elucidated, but it is thought that ligands easily liberated from pathogens, such as flagellin, lipoprotein, and LPS on a pathogen's surface, are recognized by the host's cell surface TLRs, while ligands hidden inside the pathogens, such as nucleic acids, are recognized in endosomes after lysosomal degradation of microbes or cells. Such evidence can be translated into vaccine formulation and delivery systems. Not only vaccines need to target antigen presenting cells, but also the antigen and adjuvant need to reside in the same vesicle in a cell for efficient antigen processing and presentation through the endosomal or phagosomal pathway, coupled with TLR-dependent DC activation/maturation in order to prime CD4 T cells (16).

In addition, expression of each TLR is also quite distinct among cell types. TLR2 and 4 are expressed on various immune cells including macrophages, DCs, B cells, granulocytes, NK cells, and T cells, and even on nonimmune cells such as fibroblasts and epithelial cells. TLR7 and TLR9 are largely expressed in the immune cells. In particular, these receptors are predominantly expressed in plasmacytoid DCs that produce a large amount of type-I interferon during viral infection. The intercellular crosstalk between these TLR-expressing cells may influence the outcome of adjuvant-induced adaptive immune responses. In the case of viral, DNA, or RNA antigens, which are expressed inside cells, cross-presentation of antigen to CD8 T cells is known to occur, during which TLRs in nonantigen presenting cells may affect the outcome. TLR expression can be altered in response to a variety of cytokines and environmental stresses induced by pathogens or vaccines. Thus, the efficient and specific delivery of vaccine antigen as well as adjuvant into antigen processing and/or presenting cells should be carefully considered for potent, but also safe, vaccine development.

TLR2/4 and NOD1/2 as Sensors for a Bacterial Cell-Wall-Based Adjuvant

TLR2 and TLR4 on the cell surface, and intracellular proteins such as NOD1 and NOD2, which contain a nucleotide-binding oligomerization domain, are known to recognize distinct components within bacterial cell walls; these include LPS (recognized by TLR4), lipoprotein (recognized by TLR2), peptideglycan (PGN) (recognized by NOD) and lipoteichoic acid (LTA) (recognized by TLR2). It is known that the adjuvant activity of bacterial cell walls is responsible for their ability to activate the innate immune system through cognate receptor(s), and purified components of bacterial cell walls have also been proven to be potent adjuvants. The TLR4 ligand LPS has been experimentally shown to be a potent adjuvant for vaccines, although its extreme toxicity prevents its use in humans (17, 18). The adjuvant effect of LPS is solely dependent on TLR4-mediated, MyD88-dependent signaling (19, 20). Efforts to eliminate the toxicity of lipid A led to the development of monophosphoryl lipid A (MPL) (17, 18). MPL-based adjuvant (monophosphoryl-lipid A/trehalose dicorynomycolate (“Ribi” adjuvant)) has been used in human clinical studies as a new-generation vaccine adjuvant against infectious diseases and seasonal allergic rhinitis, and was proved to be safe and effective (21, 22). MPL contains lipid A as a TLR4 ligand; however, it was recently shown that the dependency of TLR4 on adjuvant effect of MPL was surprisingly minor, at least for antigen-specific antibody responses (4), suggesting that there are yet unknown TLR-independent adjuvant factors within the MPL compound.

TLR2 mediates the adjuvant activity of its ligand, lipoprotein; for example, Mycoplasma macrophage-activating lipopeptide 2 (MALP-2) is recognized by a heterodimer of TLR2 and TLR6, and the synthetic bacterial lipopeptide PAM3CSK4 is recognized by a dimer of TLR2 and TLR1 (23, 24), both of which have been proven to be potent adjuvants in vivo (25, 26). Outer-surface lipoprotein (OspA) of Borrelia burgdorferi, which is used in vaccines for Lyme disease, and conjugate polysaccharide vaccines containing outer membrane protein complex derived from Haemophilus influenzae type b (Hib-OMPC) are both potent vaccine formulations. OspA and Hib-OMPC not only contain protective antigen, but also contain immunostimulatory cell wall components as an adjuvant mainly recognized by TLR2. In humans, low responders to OspA vaccine have impaired expression of TLR1, and TLR1−/− as well as TLR2−/− mice were unable to mount a protective response after OspA vaccination (27). However, recent evidence suggests that there may be other adjuvant factors within the OspA vaccine formulation as TLR2−/− mice were protected by a Pam3Cys-modified OspA vaccine (28). Similarly, Hib-OMPC vaccine-induced proinflammatory cytokines were TLR2-dependent; however, antigen-specific IgG titers were not dramatically reduced in the absence of TLR2 (29), suggesting the existence of other adjuvant factors in this vaccine formulation.

In addition to the major role of TLR2 and 4 in the cell surface recognition of antigens and the subsequent activation of the innate immune system, NOD1 and NOD2, which are localized in the cytoplasm, have been shown to recognize PGN, a component of the bacterial cell wall (30, 31). Muramyldipeptide (MDP), a common structural component of PGN, is a ligand for NOD2, and other bioactive moieties of PGN, desmuramylpeptides (DMP) containing diaminopimelic acid (DAP), were found to be ligands for NOD1. Interestingly, MDP is a minimally required component of complete Freund's adjuvant (CFA), which is composed of a mycobacterial extract in an oil emulsion and is one of the most common adjuvants used experimentally (32). Although purified MDP is capable of inducing innate immune responses in human cells (but not in mouse cells) strong synergisms have been observed with the other TLR ligands (33, 34); these synergisms may contribute to the whole adjuvant activity of CFA, as CFA seems to contain TLR2 and/or TLR4 ligands. Similarly, the BCG vaccine, which is known to contain TLR2 and TLR4 ligands (as well as TLR9 ligand), has been shown to be able to induce adaptive immune responses in the absence of MyD88, a critical adaptor for TLR-mediated innate immune activations, suggesting that the BCG vaccine may contain TLR-independent adjuvant activity, probably NOD-like receptor ligands (35). Further studies should clarify which component(s) and host receptor(s) are critical for the adjuvant activity within these potent vaccine and adjuvant formulations.

TLR5 and NOD-Like Proteins Mediate the Flagellin-Induced Adjuvant Effect

TLR5 recognizes the bacterial protein flagellin, which is found in the flagellar structures of many bacteria (36). TLR5 is detected in epithelia in the lung and gut, and is also highly expressed in residual dendritic cells such as those in the lamina propria of the intestine (37). Flagellin is a potent immune activator, stimulating diverse biologic effects that mediate both innate inflammatory adaptive immune responses. The protein nature of flagellin is considered to be an advantage for many immuno-therapeutic applications mainly due to its ease of manipulation; for example, a DNA vaccine encoding a chimeric version of antigenic protein and flagellin has been developed (38, 39).

TLR5, however, appears not to be the only receptor that mediates the flagellin-induced adjuvant effect. Independently of TLR5 or MyD88, a member of the NOD-LRR protein family, neuronal apoptosis inhibitory protein 5 (NAIP5), has been shown to be involved in the detection of flagellin in the cytoplasm as well as in the caspase-1-dependent control of Legionella pneumophila infection by macrophages (40, 41). ICE protease activating factor (IPAF), another CARD-containing NOD-LRR protein, has been shown to recognize Salmonella typhimurium, whose infection also results in caspase-1 activation. Flagellin delivered to the cytosol activates caspase-1 via IPAF, and independently of TLR5 (42, 43). Although the mechanism by which these two proteins recognize the same ligand is not yet clear, NAIP5 and IPAF may cooperate in the recognition of such bacterial components as they can physically interact with each other. It will be of interest to clarify how the potent adjuvant activities of flagellin are mediated by these three flagellin receptors: cell-surface TLR5, the intracellular NOD-like protein NAIP5, and IPAF.

TLR3, 7, and 8, and RIG-Like Receptors Mediate the RNA-Induced Adjuvant Effect

TLR3 recognizes double-stranded (ds) RNA derived from the viral genome, or intermediates generated during viral replication, all of which have been shown to play an important role in antiviral responses. Poly-I:C, a synthetic version of dsRNA was one of the first therapeutic agents used to treat HIV and leukemia patients, but was abandoned due to its toxicity (44). Several studies have been undertaken to reduce the toxicity of poly-I:C, and this agent is currently undergoing clinical trials for breast cancer and ovarian cancer (45). Importantly, the dsRNA-induced, TLR3-mediated maturation of CD8 dendritic cells was shown to play an important role in the induction of antigen-specific CD4+ and CD8+ T cell responses via type I interferon-mediated cross-priming, suggesting that TLR3 is a good adjuvant target for inducing cellular immune responses (46, 47). However, dsRNA still stimulated dendritic cells in TLR3−/− mice, especially, when administered directly into the cytosol by transfection, indicating the existence of a TLR3-independent adjuvant receptor for dsRNA.

Recently, three homologous DExD/H box RNA helicases were identified as cytoplasmic sensors for viral infection and dsRNA (48, 49). Two family members, retinoic-acid-inducible gene I (RIG-I) (also called DDX58) and melanoma-differentiation-associated gene 5 (MDA5) (also called Helicard), share two N-terminal CARDs followed by an RNA helicase domain (48). RIG-I and MDA5 differentially sense invasion of a variety of RNA viruses by recognizing distinct features of the RNA genome or RNA products and trigger a TLR-independent signaling pathway through IPS-1, culminating in antiviral immune responses including type-I IFN production (50). Surprisingly, recent evidence suggests that MDA5, but not RIG-I, is essential for TLR3-independent, Poly-I:C-mediated innate immune responses, including type-I IFN production and dendritic cell activation (51, 52). These results also provide the important information that MDA5 may play a role not only in innate antiviral responses, but also in adaptive immune responses to virus or vaccine where dsRNA acts as an adjuvant.

Unlike dsRNA, single-stranded RNA had long been thought to be immunologically inert, because host cells are abundant with single-stranded RNA species. However, recent evidence suggests that single-stranded RNA is not inert, but rather very immunostimulatory unless heavily modified by methylation or with certain sequences (53–56) that may explain the ability of single-stranded RNA to reach endosomes (57). Single-stranded RNA genomes, oligoribonucleotides derived from HIV or influenza virus, some double-stranded short interference (si) RNAs developed for RNA interference (RNAi), and small synthetic compounds known as imidazoquinolins, are recognized by TLR7 in mice, and by both TLR7 and TLR8 in humans; this recognition activates various immune cells that produce type I IFNs and elicit cellular immune responses (58–60). In humans, TLR7, but not TLR8, is highly expressed in plasmacytoid DCs, and activation of TLR7 in these cells leads to the production of type-I IFNs. By contrast, TLR8 (but not TLR7) is highly expressed in monocytes, and activation of TLR8 in these cells leads to the production of proinflammatory cytokines, especially, IL-12 (61). TLR7 and possibly TLR8 utilize MyD88 as an essential adaptor to downstream signaling pathways. Several TLR7 agonists have been approved for clinical use in various viral infections (62). The TLR7 agonist imiquimod (5% cream) has been shown to be effective for external genital warts, basal cell carcinoma, and actinic keratosis (63–65), and is in a phase I clinical trial against human papillomavirus (22). Several other synthetic TLR7 agonist compounds have been in phase I or phase II trials against hepatitis B virus, hepatitis C virus, and cancer (22). The adjuvant activity of TLR7 ligand was also confirmed in nonhuman primates (66).

However, immunostimulatory single-stranded RNA derived from either RNA viruses, such as influenza, or synthetic oligoribonucleotides has been reported to stimulate the immune system in a TLR7/8-independent manner also. While immunostimulatory RNA and the RNA genomes of viruses such as influenza activate plasmacytoid DCs via TLR7, they were also able to activate myeloid cells, such as monocytes, conventional DCs, or fibroblasts, in a TLR7- or MyD88-independent manner (6, 55, 67). Moreover, recent studies suggest that RIG-I, in fact, recognizes 5′-triphosphate of single-stranded RNA (68, 69). Thus, it is important for us to know which innate immune receptors, TLRs and/or RIG-like receptors are critical for the induction of protective innate and adaptive immune responses during viral infection or vaccination with an RNA-based vaccine that may contain immunostimulatory RNA as an internal adjuvant. By knowing these details, we will be able to efficiently target such an RNA-containing vaccine to the right cells, and optimize their adjuvant activity depending on the innate immune receptors described above, to provide protective immune responses.

TLR9-Dependent and TLR9-Independent Adjuvant Effect of DNA

As a fundamental entity of most living organisms, DNA is normally tightly sequestered within the nuclear or mitochondrial membranes in eukaryotes, the cell wall in bacteria, or the envelope in viruses. However, in the circumstances of microbial infection or failure of host DNA clearance, DNA can be released from microbes or damaged host cells, and is detected by and modulates the innate immune system. Currently, TLR9, the only known receptor to detect immunostimulatory DNA such as CpG DNA, has been shown to play critical roles in mediating the protective immune responses to various infectious agents, allergic disorders, and cancer, and is implicated in a pathological role in certain autoimmune diseases (reviewed in (70–72)). Synthetic oligodeoxynucleotides (ODNs) that contain unmethylated CpG motifs trigger TLR9-mediated, MyD88-dependent signaling in macrophages, dendritic cells, and B cells to induce the production of proinflammatory cytokines, chemokines, and immunoglobulins. The robust innate immune response to CpG ODNs skews the host's immune milieu in favor of a strong cellular immune response, including induction of CD4 Th1 and CD8 CTL, an effect that underlies their use as vaccine adjuvants and anti-allergens. Preclinical studies provide evidence that CpG ODNs are effective for each of these uses and can modulate the immune response to coadministered allergens and vaccines (73, 74).

Plasmid DNA derived from bacteria contains immunostimulatory CpG motifs (72), which have been shown to stimulate the innate immune system; thus, these motifs can act as a “built-in” adjuvant for DNA vaccines (75). TLR9 is currently the only known receptor for the immunostimulatory CpG motifs in DNA, and TLR9-deficient antigen presenting cells, including dendritic cells, do not respond to CpG motifs (76). As expected, TLR9-deficient mice failed to mount Th1-biased antigen-specific immune responses to protein vaccines using CpG ODN as an adjuvant (76).

However, in the case of DNA vaccines, TLR9-deficient mice mounted a comparable amount of the encoded-antigen-specific IgG, including IgG1 and IgG2a, IFNγ secretion and CTL responses, to the amounts produced by wild-type mice (77, 78); another report showed a partial reduction of immune responses in TLR9-deficient mice (79). Moreover, recent evidence suggests that not only DNA derived from microbes, but also DNA derived from host cells, activates the innate immune system in a CpG motif-independent manner that is dependent on its double-stranded (ds) structure when it is introduced into the cytosol (80, 81) or if the homeostatic clearance of such DNA is hampered, this pathway is activated (82). Double-stranded (ds) DNA in the right-handed B-form (B-DNA), but to a lesser extent in the left-handed Z-form (Z-DNA), activates both immune and nonimmune cells to produce type I interferons (IFNs), cytokines, and chemokines through a TLR9-independent pathway, but as an yet undefined DNA recognition machinery, and a distinct signaling pathway in which TBK1, a noncanonical IkB kinase is involved (83, 84). These results suggest that the immunogenicity of DNA vaccines is controlled mainly by TLR9-independent and, possibly, CpG-motif-independent factors in the plasmid DNA that act as “built-in” adjuvants. It will be of interest to investigate whether TLR9-independent innate immune recognition of and regulation by DNA provide clues to the understanding of their physiological roles in the immunogenicity of DNA-based vaccines or immunotherapy.

Concluding Remarks

As TLR-related research on the innate immune system matures, TLR-independent pathway(s), which control not only innate, but also adaptive immune responses, have emerged. Further understanding of both pathways of innate immune recognition and regulation by many immunologically active compounds will hopefully facilitate the development of more potent and safer adjuvants, ultimately toward protective vaccines for applicable diseases.