Toxoids are toxins that are inactivated by chemical or physical treatment, but that nonetheless preserve their antigenic structure. This approach allows the formation of a protein that has completely lost toxic activity but is still capable of inducing the production of antibodies that will recognize and block the native toxin. This approach can be applied in case of infections in which the pathology exclusively depends on the damage induced by a single toxin produced by a pathogen, and, therefore, production of antibodies neutralizing the activity of the toxin or rapidly causing its elimination will be enough to avoid any disease. Vaccines against tetanus and diphtheria are based on toxoids; as in both cases, the pathology is caused entirely by a single toxic protein secreted by Clostridium tetani or Corynebacterium diphtheriae, respectively. Another commonly used toxoid is the inactivated pertussis toxin from Bordetella pertussis, which is one of the components of the acellular pertussis vaccine.

In case immunization with a single protein or a combination of proteins from a pathogen is sufficient to stimulate a protective immune response against that particular microorganism, the approach of a protein-based vaccine is appropriate. Proteins can be purified from in vitro cultures of a pathogenic microorganism. The resulting vaccine preparations contain different amounts of contaminants depending on the efficiency of the purification process. Licensed acellular pertussis vaccines currently available contain from two to four different proteins purified from B. pertussis and are able to confer protection against whooping cough comparable to that obtained with the whole cell vaccine. One of the most widely used subunit protein vaccines is the influenza vaccine composed of hemagglutinin (HA) and neuraminidase (NA) purified from the inactivated influenza virus.

Development of the recombinant deoxyribonucleic acid (DNA) technology has made possible the expression of protective protein antigens in heterologous expression systems such as E. coli, yeast, mammalian cells, or baculovirus. This technology avoids the problems related to growing and manipulating large amounts of a pathogen from which the antigen is purified. Moreover, recombinant proteins are generally better purified from cultured microorganisms resulting in cleaner vaccine preparations with a better safety
profile. A drawback of a clean vaccine preparation containing pure recombinant protein(s) is their reduced immunogenicity that may require the addition of an adjuvant to achieve enhanced efficacy.

VLPs are supramolecular protein structures formed by viral proteins in vitro in the absence of a viral genome, which are structurally similar to the natural viral infectious particles while they are not infectious. One example of a VLP is the vaccine used to prevent hepatitis B virus (HBV). HBsAg spontaneously assembles into spherical particles of 22 nm diameter when expressed in yeast or mammalian cells.^4,5,6 Many theoretical and practical aspects of VLP technology make this approach attractive to vaccine design. The viral antigens expressed in VLPs usually exhibit a conformational structure that is similar to what they present in the real virus, and this is usually associated with conservation of critical conformational epitopes that are targets of a neutralizing immune response. Another important characteristic of VLPs is that they express the antigenic epitopes in multiple repetitive structures in one single particle, which potentially can induce an efficient engagement of the antigen receptor on the surface of B cells, resulting in an improved antibody response compared to what is obtained with soluble protein antigens. As VLPs are not infectious, there are no issues related to manipulation of high amounts of infectious material. Moreover, no inactivation procedures, which may result in damage of important immune relevant epitopes, are required. VLPs are at the base of the two HPV licensed vaccines. The VLP approach has been applied to many vaccine candidates, some of which are currently being tested in clinical trials.^7,8 This approach has also been used for antigens that do not naturally form VLPs by developing chimeric VLPs using two different methods. In one case using DNA recombinant techniques, a candidate vaccine antigen is genetically fused with a protein that is able to form VLPs, resulting in its expression in the context of the VLP structure. Alternatively, an antigen is chemically linked to heterologous VLPs that will provide the multimeric structure able to induce an appropriate immune stimulation. When these chimeric particles are administered, they are able to elicit a strong immune response to all components in the VLPs, including the epitopes from the fused candidate vaccine antigen. The chimeric VLP approach has been evaluated for development of vaccines against viral infections such as hepatitis C virus (HCV)⁹ and HIV,¹⁰ or against malaria,¹¹ as well as of therapeutic vaccines for cancer or for addictive substances such as nicotine.¹² However, thus far, none of these experimental chimeric VLP vaccines have progressed beyond phase I/II clinical trials.

The surface of many pathogenic bacteria is covered by a capsular shell that is mainly assembled from ordered polymeric glycans. This extensive polysaccharide coat entirely shields the bacteria outer membrane, preventing other surface bacterial components from becoming a target of the host immune response. Nevertheless, antibodies to bacterial surface polysaccharides can clear the bacteria from the host by different mechanisms, such as complement-mediated killing and opsonophagocytosis. Hence, stimulation of an antibody response against the surface polysaccharide of pathogenic bacteria is a valid strategy for the development of vaccines against capsulated bacteria. Experiments using the capsular polysaccharide isolated from Streptococcus pneumoniae demonstrated for the first time the immunogenicity of a bacterial carbohydrate and a protective effect of a polysaccharide-based vaccination.¹³ The chemical structure of capsular polysaccharides varies not only between bacteria of different species but also between different strains within a single species, which are usually differentiated and typed based on their capsular polysaccharides. As a consequence, a limitation of polysaccharide-based vaccine is that the immune responses they elicit are often serotype specific. Thus, if disease burden is caused by multiple polysaccharide types, a vaccine would optimally contain multiple polysaccharides in order to confer broad protection. In addition to S. pneumoniae, for which a vaccine against 23 serotypes is available, polysaccharide-based vaccines have been developed for MenACWY, Hib, and Salmonella typhimurium.

Purified polysaccharides are able to stimulate only B-cell responses because they cannot enter the major histocompatibility complex (MHC) cavity (which evolved to bind peptides) and cannot be presented to T cells. For this reason, they are classically known as “T-independent” antigens. Thus, the antibody response generated by a polysaccharide vaccine is a typical T-cell-independent response that presents two major disadvantages from a vaccine point of view. First, the response does not take place in germinal centers, and it is therefore not associated with affinity maturation of the antibodies or the generation of memory B cells. As a consequence, a polysaccharide vaccine will induce low- to intermediate-affinity antibody responses that is not followed by immunologic memory, resulting in the need for booster administration to maintain immunity. Moreover, a T-cell-independent antibody response is deficient or absent in young children (younger than 18 months of age) and in the elderly, two populations at risk of infection with capsulated bacteria such as pneumococcus or meningococcus, and for which a polysaccharide-based vaccine is not efficacious. A way to circumvent the limitations of these vaccines is to conjugate a polysaccharide to a carrier protein, such as TT or DT, in order to induce a carrier-specific T-cell response. Following vaccination with glycoconjugate vaccines, T follicular helper cells promote differentiation of the polysaccharide-specific B cells in the germinal center, where somatic mutation and affinity maturation take place, resulting in higher affinity antibody response and generation of long-lived memory B cells conferring long-term protection. The first demonstration that conjugation to a protein increases immunogenicity of a capsular polysaccharide was obtained in 1929; however, it took 50 years for this strategy to be applied to vaccine design, when a glycoconjugated vaccine for Hib was first tested in animals.¹ Compared to polysaccharide vaccines, glycoconjugated vaccines are indeed
more efficacious, being able to induce a long-lasting antibody response not only in adults but also in infants, young children, and the elderly. Similarly to the polysaccharidebased vaccines, the glycoconjugate vaccines are also serotype specific, and, therefore, multivalent formulations are required to achieve protection against multiple serotypes.

Priming of CTL responses requires presentation of the antigen by the MHC class I molecules, a pathway usually followed by endogenously synthesized proteins (eg, viral antigens) but not by exogenously administered antigens. For this reason, inactivated or subunit vaccines are not per se able to stimulate a CTL response, which is instead induced by live-attenuated viral vaccines. The observation that a gene encoding an external antigen could be inserted in the genome of the Vaccinia virus and expressed in the virus-infected cells has prompted the idea of genetically manipulating an attenuated or nonpathogenic virus to express an exogenous antigen from a pathogenic virus against which vaccines need to be developed. Because these antigens are synthesized by the host cells, they undergo all posttranslational modifications and acquire a correct conformation and should be able to stimulate a strong neutralizing antibody response. Moreover, antigens are presented in the context of the MHC class I pathway generating therefore a CTL response. Examples of attenuated strains of Vaccinia virus are the modified Vaccinia Ankara (MVA) strain obtained by repetitive passages through in vitro cell cultures and the New York Vaccinia strain obtained by genetic manipulation of the viral genome. These strains have been evaluated in both nonhuman primates and humans as vector vaccines for HIV, malaria, influenza, TB, and other infectious diseases.^16,17,18 Other poxviruses tested as vector vaccines include fowlpox and canarypox.¹⁹ Despite extensive testing of these vectors as vaccine candidates, so far, no successful vaccine has been developed using this approach.

Because of their immunogenicity and efficacy in the induction of a robust cellular response, replication-incompetent recombinant adenoviral vectors (rAds) have been successfully tested in many animal models as vaccine vectors for many pathogens such as Ebola, HIV, anthrax, influenza, and malaria.^20,21,22,23 Adenoviral vaccine vectors have also been evaluated in many human clinical trials, particularly for HIV and cancer. On the other hand, natural adenovirus circulates in the human population, and, therefore, preexisting immunity, in particular neutralizing antibodies, may be present in the subject who will receive the vaccine, compromising efficacy of these vaccines.²⁴ It is important to note that tens of adenovirus serotypes exist but most of the clinical studies have used a vector derived from a single serotype (Ad5). Serotype 5 is one of the most common serotypes spread in the human population, and a high frequency of subjects possesses anti-Ad5 antibodies. In order to overcome potential problems related to preexisting anti-Ad5 antibodies, new vectors based on different human serotypes²⁵ or on adenovirus from other species^26,27,28 were developed. At the present, the most promising are the chimpanzee-derived adenoviruses.²⁷

Alphaviruses are positive single-stranded ribonucleic acid (RNA) viruses in which the genome can be manipulated to obtain replication-defective infectious particles that express an exogenous antigen. Besides this favorable safety feature, upon entering the target cell, alphavirus RNA self-replicates, producing many copies of the coding RNA, which can then direct the synthesis of a very high amount of the antigen, making it a promising platform for a vector-based vaccine.²⁹ Alphavirus-based vaccines have been tested in animal models for several pathogens and have demonstrated excellent immunogenicity.^30,31,32,33 Some clinical trials have been conducted, but the results have not been published.

The prime-boost vaccination strategy using viral vectors has received wide evaluation in the vaccine field. Heterologous prime-boost regimens mainly use a viral vector or a DNA vaccine (see subsequent discussion) at priming
followed by a boost with a protein-based vaccine, although other options have also been used. This unconventional immunization schedule usually results in the induction of a strong cellular immune response, with generation of both CD4+ and CD8+ antigen-specific T cells. This is in turn associated with a higher and more specific antibody response against the vaccine target compared to homologous immunization with either DNA or viral vector. The immunologic mechanisms responsible for these improved responses are not entirely known. It is hypothesized that while DNA or viral vector priming is required to induce a strong T-cell response, boosting with only the target protein focuses the B-cell response on a single protein antigen for which high-affinity antibodies are desirable and at the same time avoiding interference by antivector immunity.³⁴

At the beginning of the 1990s, experiments designed to evaluate new approaches for gene therapy were performed to test if intramuscular injection of a liposome formulation containing a plasmid DNA or a messenger RNA coding for a reporter protein could result in in vivo expression of the protein. Surprisingly, animals in control groups that were injected with pure DNA or RNA showed a significant and prolonged expression of the reporter protein in muscle cells.³⁵ This observation inspired a series of studies that demonstrated that mouse immunization with an antigen encoded by DNA could induce a strong antigen-specific immune response characterized by antibody production and generation of both T-helper and CTL responses.^36,37,38 One of these studies also established the protective effect of a DNA-mediated immune response, showing that mice injected with a plasmid encoding the nucleoprotein from a H1N1 strain of influenza A virus were completely protected from challenge with a virulent H3N2 strain.³⁸ Based on these promising results, a huge amount of experimental work has been performed in the last two decades to apply DNA vaccination to many disease targets with both prophylactic and therapeutic approaches. As with vector-based vaccines, the advantage of a DNA-encoded vaccine is that the antigen is synthesized within host cells, facilitating correct folding of conformational neutralizing epitopes and MHC class I antigen presentation, allowing generation of CTL responses. An additional advantage of DNA vaccines compared to viral vectors is that the former contains only DNA, therefore nullifying any effects of preexisting vector immunity. DNAbased vaccines are composed of plasmid DNA containing an antigen gene sequence. To be efficiently expressed in the host mammalian cells, the antigen sequence has to be under the control of a strong eukaryotic promoter, such as those present in many viral genomes (eg, the early promoter of the human cytomegalovirus [HCMV]). Classical plasmids used for DNA vaccination also contain antibiotic resistance genes and prokaryotic origin of replication sequences, both required for selection and replication in bacteria. These prokaryotic sequences may contain unmethylated CpG motifs that are recognized by toll-like receptor (TLR)9 and were originally shown to play an immunostimulatory role during DNA vaccination.³⁹ However, subsequent investigations have excluded a role for TLR9 and CpG motifs in the immunogenicity of DNA vaccines and have suggested that the immunostimulatory activity might be due to the double-stranded structure of DNA and its capacity to activate other innate immune receptors.^40,41,42,43 After intramuscular injection, DNA is primarily localized within myocytes where the encoded antigens are expressed, although expression can also be observed in antigen-presenting cells (APCs).^44,45 Indeed, several studies have demonstrated that expression of antigen by myocytes is critical for induction of a CTL response, although they do not directly present the antigen to T cells. Experiments using parental bone marrow transplantation into F1 mice established that generation of antigen-specific CTL responses by DNA immunization depends on a cross-priming mechanism. Transfected myocytes transfer the antigen to professional APC, driving its processing and presentation by MHC class I molecules, which then leads to triggering of a CD8+ T-cell response.^46,47 Manufacturing of DNA vaccines has several benefits over all other type of vaccines: 1) design of DNA construct is straightforward; 2) production of DNA plasmids is uncomplicated and cheap and does not require manipulation of infectious materials; and 3) plasmids are stable even at room temperature, reducing the complexity and the cost of distribution. Because of the potential advantages of a DNAbased vaccine, this approach has been extensively tested for many pathogen such as rabies, HBV, and HIV in different animal models with promising immunogenicity results.^48,49,50 Furthermore, the approach has also been evaluated in humans in several clinical trials.^51,52 The human studies have demonstrated that DNA vaccination is generally safe and able to stimulate an immune response; however, immunogenicity is usually lower than expected based on the results of animal studies, resulting in a diminished confidence in this approach. The initial failure of DNA vaccine in the clinical setting led to the exploration of a series of techniques to improve their immunogenicity. To improve expression in vivo, the antigen-coding sequences were modified in order to contain only codons that match transfer RNAs with high frequency in mammalian cells, whereas stronger promoters were included in the plasmids. In an attempt to increase stability of the administered plasmid, different formulations and delivery systems, such as liposomes or microparticles, have also been used.⁵³ In addition, several new injection techniques have been tested to improve in vivo transfection efficiency. A forced delivery of DNA into the skin can be obtained by the use of gold microparticles in gene gun techniques⁵⁴ or by a high-pressure liquid stream created by the Biojector device (Bioject Medical Technologies, Inc., Tigard, OR, U.S.A.),⁵⁵ whereas electroporation can greatly increase efficiency of intramuscular injection of DNA.⁵⁶

Although B cells can recognize linear epitopes within a protein, most of the epitopes recognized by a protective antibody response during a natural infection are conformational. Linear epitopes are usually not efficient at inducing a good neutralizing response. The goal of peptide vaccines is therefore to create T-cell epitopes capable of stimulating
T-cell responses and in particular CTLs, in cases of viral infections⁵⁷ or cancer,⁵⁸ where these type of immune responses are required to obtain a complete protection. Although the peptide approach is able to induce protection from viral infections and cancer in the mouse model, polymorphism of the MHC system has limited the application of peptides to human preventive vaccines. The use of peptides has been mostly restricted to therapeutic vaccines against cancer.⁵⁹

Numerous subunit vaccines in use today, including those against HAV, HBV, HPV, anthrax, diphtheria, tetanus, MenC, tick-borne encephalitis, and pneumococcal conjugate vaccines, are adsorbed to aluminum salts.⁶⁴ The most used aluminum salts for licensed vaccines are aluminum hydroxide and aluminum phosphate, and they are generically called “alum.” Alum adsorption enhances the antibody responses to the coadministered antigens and, in some cases, can also increase antigen stability allowing for more extended vaccine shelf life. Only recently has the mechanism of action of alum been deeply investigated.^65,66,67 Early work suggested that alum is able to form a depot with the antigen at injection site increasing antigen persistency and availability. However, in the last 20 years, the depot effect of alum has been challenged by several studies that reported a similar half-life of antigen formulated in the presence or absence of alum.^68,69,70 Alum absorption increases antigen uptake by DCs in vitro, suggesting that it acts as an antigen delivery system.⁷¹ More recently, it has been shown that the interaction between alum and DCs requires membrane sphingomyelin and cholesterol.⁷² Moreover, alum injection results in cell recruitment to the injection site,^73,74 especially local recruitment of monocytes, which migrate to the draining lymph nodes and differentiate into inflammatory DC capable of priming T cells.⁷⁵ Altogether, these data suggest that alum has a dual function in promoting antigen uptake and a local proinflammatory environment at injection site. In vitro experiments conducted on mouse and human blood cells led to the identification of at least one of the molecular targets of alum inflammatory activity. Alum activates a cytoplasmic NLR protein called NLRP3, which associates with the adaptor protein ASC and the caspase-1 protease to form a protein complex called inflammasome.^76,77,78 Activation of NLRP3 ultimately results in the processing of pro-caspase-1 into its active form, which cleaves several proinflammatory cytokines including IL-1β and IL-18. It is still not clear how alum activates NLRP3; however, it has been proposed that phagosomal destabilization induced by aluminum and by other crystal structures (silica crystals) following phagocytosis plays an important role.⁷⁹

In vivo studies conducted in NLRP3-deficient mice have shown that direct NLRP3 activation contributes to alum adjuvanticity.^75,77,80 However, the requirement of NALP3 for the adjuvant activity of alum has been challenged by other studies using alternative antigens or different immunization routes.^76,81,82 Probably, other proinflammatory pathways triggered by alum exist and are independent from inflammasome activation. Recently, it has been proposed that host DNA released from dying cells acts as a danger signal that mediates alum adjuvanticity through the activation of several innate immune pathways including interferon regulatory factor 3 (IRF3). In addition, it has been demonstrated that (IRF3) is required for immunoglobulin (Ig)E antibody response to alum adjuvanted vaccines.⁸³

Oil-in-water emulsions are liquid dispersions of oil droplets stabilized by one or more surfactants. The first emulsion to be licensed for human use was MF59 composed of squalene
oil, emulsified with two surfactants (Tween 80 and Span 85). MF59 has been licensed in Europe since 1997 in an adjuvanted seasonal influenza vaccine for its capacity to increase flu immunogenicity in the elderly.⁸⁴ More recently, MF59 and a second squalene-based oil-in-water emulsion called AS03, composed of squalene, Tween 80 and α-tocopherol, have been licensed in Europe for pandemic influenza vaccines and have been widely used for the 2009 H1N1 pandemic flu campaign.⁸⁵ A third squalene-based oil-in-water emulsion called AF03 has been approved for pandemic influenza in Europe but has not been marketed yet.⁸⁶ Clinical trials demonstrated that avian H5 pandemic influenza vaccines containing oil-in-water emulsions are superior to alum-adjuvanted or nonadjuvanted H5 vaccines in the induction of protective antibody titers.⁸⁷ Both AS03 and MF59 allowed for antigen dose sparing and increased seroconversion and cross-protection.^88,89 It is important to notice that in the case of H5 avian flu, for which most of the population has not been previously exposed, nonadjuvanted vaccines did not reach the antibody titers considered to be the serologic correlate of protection, whereas emulsion-adjuvanted vaccines were able to elicit protective antibody titers after two doses. A phage display library approach demonstrated that MF59 can enhance the breadth and the affinity of the antibody responses to H1 and H5 pandemic influenza vaccines.^90,91,92 In particular, the addition of MF59 enhanced the recognition of the HA globular head region that contains most of the neutralizing epitopes. Two additional clinical trials have been conducted on individuals vaccinated several years before with H5 vaccines either nonadjuvanted or mixed with MF59 or AS03 by administering a new dose of oil-in-water emulsion-adjuvanted H5 vaccine (boost). These studies demonstrated that priming with avian H5 vaccines mixed with MF59 or AS03 allows for better and more rapid responses to a booster dose compared to individuals primed with nonadjuvanted vaccines.^93,94 Priming with adjuvanted vaccines was effective even when H5 antigens from different influenza clades were used for priming and boosting, inducing a rapid increase of cross-neutralizing antibody titers. These studies suggest that adjuvanted influenza vaccines may be used for a prepandemic vaccination strategy. Another recent study highlighting the strong impact that adjuvants can have on vaccine efficacy was recently performed on children between 6 and 72 months of age. The addition of MF59 to seasonal flu subunit vaccine increased the efficacy of the vaccine from 43% to 86% against polymerase chain reaction confirmed influenza-like illness.⁹⁵ Interestingly, in the age subgroup 6 to 24 months, nonadjuvanted influenza vaccine was unable to protect, whereas the efficacy of MF59-adjuvanted vaccine was 77%.

Several studies using the mouse model and human blood cells have addressed the mechanism of action of oil-in-water emulsion. In contrast to alum, which physically binds antigens through an adsorption process, oil-in-water emulsions are simply mixed with antigens. MF59 neither increases antigen persistence at the injection site nor induce any depot effect.⁹⁶ However, MF59 can enhance antigen uptake by activated DCs in mouse muscle⁹⁷ and phagocytosis in vitro in human peripheral blood mononuclear cells.⁹⁸ These data suggest that MF59 can act as an antigen delivery system; however, other studies have shown that it has also immunostimulating properties. In vitro analysis of human blood cells showed that MF59 induces the differentiation of monocytes toward DCs and can stimulate chemokine secretion (CCL2, CCL3, CCL4, and CXCL8) in human macrophages, monocytes, and granulocytes.⁹⁸ Intramuscular administration of MF59 in mice induced the recruitment of mononuclear cells partially dependent on chemokine receptor 2 (CR2).⁹⁹ The local effect of MF59 at an injection site has been evaluated in detail by combining DNA microarray and immunofluorescence techniques. Transcriptome analysis showed that the injection of MF59 in mouse muscle induced the activation of several innate immune genes, including Ccr2 and its ligands (CCL2, CCL7, and CCL8), further supporting the data described previously on CR2-dependent cell recruitment.¹⁰⁰ In addition, MF59 promoted the upregulation of many proinflammatory cytokines and other innate immunity genes and genes involved in leukocyte migration.¹⁰⁰ Two of the early genes modulated by MF59, JunB and pentraxin 3, were upregulated in the muscle, suggesting that the activation of muscle cells by MF59 might play some role in its adjuvant activity.

Intramuscular injection of fluorescently labeled MF59 and antigen revealed that MF59 colocalizes with the antigen in APCs in the muscle and enhances the number of antigenpositive DCs and B cells in the draining lymph node.¹⁰¹

Similarly to MF59, AS03 induces transient cytokine activation at the injection site and triggers the secretion of cytokines in human monocytes and macrophages.¹⁰² In addition, AS03 induces cytokine expression in the draining lymph node in a mechanism that is at least partially dependent on the presence of -tocopherol in the formulation.

In summary, the data described previously suggest that squalene-based oil-in-water emulsion function by enhancing antigen uptake and migration of antigen-positive APCs in the draining lymph nodes and by inducing an immunostimulatory environment at the injection site. However, the innate immune pathways required for the mechanism of action of emulsions are not well characterized. Unlike alum, MF59 does not activate NLRP3 in vitro and is not dependent on NLRP3 or caspase-1 in mouse immunization studies.^82,103 Despite the fact that MF59 adjuvanticity is inflammasome independent, adaptive responses induced by an H5 influenza subunit vaccine adjuvanted with MF59 largely depend on the inflammasome component ASC. Therefore, it has been proposed that ASC plays an inflammasome-independent function required for MF59 adjuvanticity.¹⁰³ Surprisingly, MF59 adjuvanticity to a recombinant MenB subunit vaccine requires another adapter protein called MyD88, which plays an essential role in TLR and IL-1R signaling. However, MF59 does not have any TLR agonist activity and the MyD88-dependent pathway required for its adjuvanticity is still unknown.⁸²

TLRs are key pathogen sensors that modulate the host immune system and play a fundamental role in response to microbial infection.¹⁰⁴ TLRs are PRRs expressed on innate immune cells including DCs that sense PAMPs. Upon pathogen recognition, DCs are activated and provide essential
signals for naive T-cell priming and development of an adaptive immune response. In addition to pathogen infection, microbial products from live-attenuated or heat-killed vaccines can also stimulate TLR pathways providing adjuvant activities.^105,106 For example, the yellow fever vaccine activates DCs by signaling through TLR2, 7, 8, and 9.¹⁰⁷ Many preclinical and clinical studies using purified TLR agonists have been conducted and demonstrated that all TLRs examined can be exploited to enhance adaptive responses to vaccination. However, despite the large amount of work performed, so far, only the TLR4 agonist 3-O-desacyl-4′-MPL, a component of lipopolysaccharide (LPS) purified from Salmonella minnesota has been licensed for human use. MPL is approved for two vaccines against HPV and HBV. In both vaccines, MPL has been adsorbed to alum to form a combination adjuvant called AS04. The HBV vaccine formulated with AS04 was developed for patients with renal insufficiency and was tested in several immunocompromised individuals including the elderly and patients on hemodialysis. The incorporation of MPL allowed for a more rapid increase of antibody titers and enhanced seroprotection rates using fewer vaccine doses.¹⁰⁸ HPV vaccine formulated in AS04 induced higher neutralizing antibody titers and memory B-cell responses compared to the same vaccine formulated only with alum.¹⁰⁹ A head-to-head trial was conducted to compare the alum-MPL adjuvanted HPV vaccine versus another HPV licensed vaccine that is adjuvanted only by alum.^110,111 Both vaccines exploit VLP technologies for the production of antigens and are highly efficacious in preventing HPV16/18-associated cervical intraepithelial neoplasia in women. However, the AS04-adjuvanted vaccine produced significantly higher neutralizing antibody titers to HPV16 and HPV18, and higher frequencies of memory B cells compared to the other vaccine.

The molecular mechanism of action of AS04 is well understood. The MPL component activates directly the LPS sensor TLR4 that is expressed by human innate immune cells including DCs and macrophages. Adsorption of MPL to alum does not interfere with its ability to activate human immune cells including DC.¹¹² AS04 injection in the mouse muscle lead to a transient production of interferon and proinflammatory cytokines at injection site and to an increased number of antigen-loaded DCs in the draining lymph node.¹¹² TLR4 stimulation on the surface of conventional DCs induces the production of IL-12 and activates the expression of MHC class II and costimulatory molecules leading to efficient CD4+ T-cell priming and to strong Th1 adaptive responses.

Synthetic TLR4 agonists have the advantage over MPL to be more pure and less expensive. Several synthetic TLR4 agonists such as RC-529 and GLA have been developed and tested as vaccine adjuvants.¹¹³ RC-529, a fully synthetic monosaccharide mimetic of MPL, is the most advanced and has been approved in Argentina for a vaccine against HBV.¹¹⁴

Virosomes were licensed for an influenza vaccine in 1994 and are composed of a mixture of phospholipids and cholesterol that are assembled in vitro with purified influenza HA and NA to form synthetic influenza viral-like structure. Influenza virosomes can be also used as adjuvants to incorporate additional antigens from other viruses or other pathogens in general. One formulation composed of H1N1 virosomes and inactivated HAV was licensed in 1997 to target HAV. The mechanism of action of virosomes is mainly linked to antigen delivery functions. Virosomes increase antigen uptake by APCs and display the antigen in draining lymph nodes in a multimeric form, facilitating B-cell-receptor engagement. The presence of preexisting immunity against influenza further enhance the adjuvant effect of virosomes probably through the formation of immune complexes that are rapidly internalized by APCs.¹¹⁵

Other liposome adjuvants made of phospholipid bilayers have been tested in preclinical models and in clinical trials in combination with various antigens and in some cases with other adjuvants, as described in the following paragraphs.

There is a great interest worldwide in exploiting TLRs as targets for vaccine adjuvants. Besides MPL, several additional TLR agonists are known to be effective adjuvants in preclinical studies.¹¹⁶ Among them, agonists of TLR2 and TLR9 and novel TLR4 agonists have been extensively tested in human as vaccine adjuvants (Table 43.3). Clinical data are also available for vaccine formulations including TLR5 and TLR7 agonists. Human data on TLR3 agonists are not yet published; however, clinical trials have already started.

Toll-Like Receptor 2. Bacterial lipoproteins targeting TLR2/1 or TLR2/6 dimers are validated vaccine adjuvants in humans and multiple TLR2 ligands have undergone clinical evaluation. The most advanced TLR2 agonist is the lipopeptide moiety of OspA, the surface protein of B. burgdorferi, the major component of a vaccine preventing Lyme disease.¹¹⁷ The vaccine was efficacious and licensed by the U.S. Food and Drug Administration (FDA) in 1998. However, despite the fact that the vaccine was safe and well tolerated, it was withdrawn by the manufacturer 3 years later following media coverage of possible autoimmune side effects.

Toll-Like Receptor 3. Agonists of the double-stranded RNA sensor TLR3 have not been extensively tested in human yet. However, a synthetic analogue of double-stranded RNA called polyinosinic:polycytidylic acid (Poly I:C) has been tested in multiple animal models with various antigens. Poly I:C can activate TLR3 and a second double-stranded RNA sensor called MDA-5, both leading to the production of type I IFN. In mice, Poly I:C was as effective as other TLR agonists in boosting antibody responses to HIV gag antigen and more efficient than any other TLR agonist tested in inducing specific IFN-γ+ CD4+ T cells.¹¹⁸ The same study showed that the adjuvanticity of Poly I:C depends on type I IFN. A study conducted with a Plasmodium falciparum antigen has shown that Poly I:C is effective in inducing specific antibodies and promoting multifunctional CD4+ T-cell responses in nonhuman primates.¹¹⁹

Toll-Like Receptor 5. TLR5 recognizes bacterial flagellin and is a validated adjuvant target in animal models. A vaccine composed of a recombinant fusion protein consisting of flagellin and four tandem copies of the ectodomain of the conserved influenza matrix protein M2e protects mice from a lethal challenge with influenza A virus.¹²⁰ Furthermore, a recombinant fusion protein made of S. typhimurium flagellin type 2 and the globular head of influenza HA could increase HA inhibition (HAI) titers by 12-fold over baseline in subjects over 65 years old.¹²¹ However, seasonal influenza vaccines are known to be immunogenic even in the absence of a coadministered adjuvant, and the absence of a control group in this vaccine trial makes it impossible to assess the real contribution of flagellin in HA responses.

Toll-Like Receptor 7/8. Preclinical data in mice and in nonhuman primates showed that small molecules agonist of the single-stranded RNA sensors TLR7 and TLR8 can improve the immunogenicity of various vaccine antigens if adequately formulated or directly conjugated to protein antigens.^122,123 Agonists of TLR7/8 have not been used as adjuvant in human vaccine formulations. However, a cream formulation of the TLR7 agonist imiquimod, which is licensed for dermatologic diseases, has been applied at the site of injection of several tumor antigens.¹¹⁶ The application of imiquimod-containing cream at the site of injection of recombinant human NYESO-1 protein antigen in melanoma patients induced measurable CD4+ T-cell and antibody responses in a subset of subjects. Unfortunately, the trial was designed without a control group that did not receive the adjuvant, and it is not possible to estimate the impact of imiquimod on the adaptive immune responses directed against NY-ESO-1.¹²⁴

Toll-Like Receptor 9. DNA oligodeoxynucleotides (ODNs) targeting TLR9 are in advanced development stage.¹²⁵ Two synthetic CpG-containing ODNs called 1018 and 7909 have been incorporated in various vaccine formulations and tested in humans. Both ODNs belongs to the B type CpG class; therefore, they are strong activators of B cells and induce large amounts of type I IFN by plasmacytoid DCs. ODN 1018 has been evaluated as an adjuvant to HBsAg vaccine and induced more rapid and sustained seroprotection in healthy adults and vaccine hyporesponsive populations.¹²⁶ The ODN 7909 has been tested with HBV, malaria, influenza, anthrax, and various cancer antigens.¹¹⁶ Clinical data demonstrated that 7909 enhances antibody responses to vaccines directed against infectious diseases. In addition, in some vaccine trials targeting cancer, 7909 had also a significant impact on CD8+ T-cell responses.¹¹⁶ Another class of TLR9 agonist adjuvant called IC31 consists of a cationic peptide KLKL(5)KLK complexed with the immunostimulatory ODN1a sequence.¹²⁷ IC31 was found to promote highly efficient Th1 responses through a MyD88- and TLR9-dependent manner.¹²⁸ The complex between the cationic peptide and the immunostimulatory ODN results in a depot effect at the injection site. Furthermore, the presence of the peptide reduces the amount of ODN needed to elicit an immune response. IC31 has been combined to the
Mycobacterium tuberculosis (Mtb) antigens Ag85B-ESAT-6 and tested in a phase I clinical trial.¹²⁹ IC31 significantly boosted antigen-specific IFN-γ-producing T cells.

Saponins are natural products extracted from plants, which have been known to exert adjuvant activities for more than 80 years. Most of the saponins used as vaccine adjuvants today are extracted from the bark of Quillaja saponaria. The QS21 is a fraction of Quillaia saponaria extract purified by reverse-phase chromatography.¹³⁰ QS21 is known to promote cross-presentation and stimulate specific CD8+ T-cell responses to subunit vaccines in preclinical models and has been formulated with liposomes, emulsions, MPL, and CpG to obtain various combination adjuvants that have been tested in human.¹³¹ Two of these combinations, AS01 and AS02, have been successfully adopted for a vaccine preventing malaria transmission.

Saponins have also been formulated with cholesterol and phospholipids to obtain a particulate adjuvant called Iscomatrix® (CSL, Victoria, Australia).¹³² Iscomatrix® has been formulated with several vaccine antigens and tested in various animal species including mice, guinea pigs, cattle, and nonhuman primates.¹³² In addition, experimental vaccines containing Iscomatrix® targeting HCV, influenza, and HPV16 have been tested in humans.¹³³ More recently, a subunit vaccine composed of a recombinant NY-ESO-1 testis cancer antigen formulated with Iscomatrix was tested in a phase I trial.¹³⁴ In animal models, Iscomatrix® is an efficient adjuvant in boosting antibody and both CD4+ and CD8+ T-cell responses even to recombinant antigens.¹³³ Mechanism of action studies showed that Iscomatrix® increases antigen translocation from the endosome to the cytosol of DCs in vitro and can promote cross-presentation in both proteasome-dependent and tripeptidyl peptidase II-dependent pathways.¹³⁵ Accordingly, when Iscomatrix® was combined to ovalbumin and injected in mice, it potently enhanced the number of DCs in draining LN able to activate OT-I CD8+ T cells ex vivo.¹³⁶ More recently, it was shown that Iscomatrix® injection in mice induced cytokine expression and antigen-independent cell recruitment in draining lymph nodes and enhanced antigen uptake by CD8+ and CD8- DCs; however, CD8+ DCs were more efficient in cross-presenting ovalbumin peptides to CD8+ T cells.¹³⁷ Unfortunately, CD8+ T-cell responses induced in human by subunit vaccines are in general less strong compared to those induced in animal models, suggesting that cross-presentation in humans is more difficult to achieve. For example, the experimental vaccine using HCV core protein antigen formulated in Iscomatrix® induced high antibody titers in all patients and specific CD4+ T-cell responses in seven out of eight patients; however, in only two of eight patients with HCV could core-specific CD8+ T cells be detected.¹³⁸

Water-in-oil emulsions were developed by Freund in 1937 and are still largely used in preclinical models. Incomplete Freund adjuvant is composed of mineral oil emulsified with mannide monooleate and is a potent adjuvant with a large number of antigens. However, when incomplete Freund adjuvant was tested in humans, it produced several local adverse reactions including sterile abscesses. Later, novel water-in-oil formulations were developed by using less reactogenic oils and surfactants. The most advanced are Montanide ISA 51 and ISA 720 (Seppic, Paris, France) that have been used in malaria, HIV, and cancer vaccine trials.¹³⁹ Montanide ISA 51 is composed of mineral oil and surfactants, whereas ISA 720 is made of squalene oil emulsified in mannide monooleate to obtain droplets of 1 µm of diameter. The different oil composition affects the reactogenicity profiles of this class of adjuvants: Mineral oils persist at injection site for a long time, whereas squalene is more rapidly metabolized. The mechanism of action of water-inoil surfactants has been linked to their ability to 1) increase the persistence of the antigens that are slowly released at the injection site (depot effect), 2) increase antigen uptake by APCs like other antigen delivery systems, and 3) induce local inflammation that promotes blood cell recruitment at injection site. In a clinical trial, Montanide enhanced antibody and cellular adaptive responses to coadministered antigen. Cellular responses could be further enhanced by the inclusion of CpG ODNs.

In some cases, a single adjuvant may not be sufficient to achieve a protective immune response. For this reason, several combinations composed of various classes of adjuvants have been tested in preclinical and clinical settings. The most successful example of adjuvant combination is AS02 composed of MPL, QS21 saponin and an oil-in-water emulsion. When P. falciparum recombinant antigen RTS,S was formulated with AS02, a 37% reduction of malaria infection was observed in a clinical trial.¹⁴⁰ Recently, the same vaccine has been optimized with the adjuvant combination AS01, in which the oil-in-water emulsion is substituted with liposomes, further increasing the protection rate.¹⁴¹ The AS02 combination has been also used for the non-small cell lung cancer antigen MAGE-A3. The vaccine gave a significant antibody and CD4+ T-cell responses. However, only a small subset of patients developed antigen-specific CD8+ T-cell responses.¹⁴² In the future, adjuvant combinations may be further improved by using different TLR agonists. In vitro data on human cells showed that TLR3 and TLR4 agonists can synergize with TLR7, TLR8, and TLR9 inducers for cytokine production and DC activation.¹⁴³ Accordingly, preclinical data in mice showed that combining TLR3 with TLR9 or TLR2 agonists resulted in a synergistic effect on the development of antigen-specific CD8+ T cells.¹⁴⁴ Furthermore, the small molecule TLR7 agonist imiquimod/R837 synergized with the TLR4 agonist MPL in mice inducing more persistent germinal centers, increased antibody titers, and increased frequencies of antigen-specific CD4+ and CD8+ T cells and plasma cells.¹⁴⁵

Mucosal vaccines elicit protective immune responses at the site of pathogen infection and have the additional advantage to be needle free, thereby improving accessibility and
cost-effectiveness. Out of five licensed mucosal vaccines, four are live attenuated (intranasal flu, oral rotavirus, OPV, and typhoid) and one is made of killed V. cholerae given with cholera toxin (CT) B antigen. Notably, no subunit vaccine has been licensed for mucosal administration due mainly to the absence of safe and effective human mucosal adjuvants. In fact, the administration of antigens in the absence of adjuvants in respiratory or gastrointestinal tracts can induce tolerance. Mucosal adjuvants are required to 1) protect antigens from degradation by mucosal proteases, 2) enhance internalization of vaccine antigen by intraepithelial DCs or M cells, and 3) overcome regulatory mechanisms that prevent the development of effector B- and T-cell responses in the mucosal-associated lymphoid tissue. Mucosal immunity can protect from pathogen infection by various mechanisms including local production of IgA, IgG, and IgD antibodies by mucosal B cells; activation of mucosal CD4+ T cells producing IL-17 (Th17); and activation of mucosal CD8+ T cells.¹⁴⁶ A large number of animal studies showed that some TLR agonists such as CpG ODNs, MPL, and flagellin can act as mucosal adjuvants; however, they have not been tested in humans yet.¹⁴⁷ The most advanced mucosal adjuvants are CT and heat-labile toxin (LT) from E. coli. Both are adenosine diphosphate (ADP)-ribosyl tranferase enterotoxins composed of a pentameric B subunit and a catalytic A subunit. The B pentameric subunit binds the ganglioside GM1 expressed on the surface of several host cells including mucosal M cells. The binding of the B pentamer promotes internalization of LT and CT into target cells followed by processing of the complex and release of the A subunit into the cytosol, where it catalyzes the ADP-ribosylation of trimeric G proteins.¹⁴⁸ Both CT and LT induce potent mucosal responses to coadministered antigens when administered intranasally in mice. However, the use of wild-type toxins as adjuvants in human has been prevented by a nonfavorable safety profile. LT was included in an inactivated virosome influenza intranasal vaccine licensed in Switzerland. After licensure, the vaccine was associated with a 19-fold greater risk of facial nerve paralysis (Bell palsy) compared to unvaccinated controls and therefore was withdrawn by the manufacturer.¹⁴⁹ In order to improve the safety profile of CT and LT, several nontoxic mutants have been designed. LTK63 and CTK63 have a single amino acid substitution in the nicotinamide adenine dinucleotide-binding site of the subunit A of LT and CT, respectively, which results in a complete loss of enzymatic activity. Both protein complexes assemble into a holotoxin and induce mucosal immune responses.^150,151 LTK63 was superior to CTK63 in inducing mucosal responses in mice and was selected for further investigation in human clinical trials. In a phase I clinical trial, LTK63 was tested as mucosal adjuvant with an intranasal trivalent inactivated influenza vaccine. LTK63 was simply coadministered with influenza HA antigens or formulated in a nanoparticle called biovector made of polysaccharide, phospholipids, and cholesterol. Both intranasal formulations were safe and enhanced IgA mucosal responses to HA in nasal secretions. The presence of the biovector enhanced both IgA responses and HAI titers.¹⁵² Later, LTK63 was also administered intranasally with Mtb Ag85B-ESAT6 fusion protein or with HIV glycoprotein 140 in two phase I clinical trials. In both trials, one subject developed a transient form of Bell palsy, questioning further development of LTK63 in intranasal administration.¹⁵³ It is important to report that LT has been used also for transdermal immunization using a skin patch system. In a first approach, the LT patch has been used to induce anti-LT immune responses and protect from enterotoxigenic E. coli infection. In a phase II placebocontrolled field trial, the LT patch protected 75% from moderate-to-severe diarrhea and 84% from severe diarrhea.¹⁵⁴ The same patch has been used as an adjuvant to enhance immune responses to a coadministered intramuscular H5 influenza vaccine. The application of the LT patch to the vaccine injection site increased HAI titers and seroconversion.¹⁵⁵ The mechanism of action of CT and LT as mucosal adjuvants is not well characterized. However, the data on genetically modified toxins showed that the enzymatic activity is not required. Experiments conducted in mouse lungs demonstrated that LTK63 activated the expression of several innate immune genes that may be involved in adjuvanticity by activating mucosal APCs.¹⁵⁶ The same study demonstrated that LTK63 induced recruitment of DCs and activation of alveolar macrophages in the airways. These events may increase antigen capture and presentation in the lung. Recently, one study demonstrated that LT and LTK63 can have direct effects on DC in vitro. Mouse bone marrow -derived DC stimulated ex vivo with LT or LTK63, and antigen have an increased capacity to stimulate specific IL-17, IL-10, or IFN-γ-positive CD4+ T cells in recipient mice. The same study showed that LT and LTK63 can stimulate the production of MIP-2 by mouse DCs and synergize with TLR agonists for the production of other cytokines including IL-1α and IL-1β.¹⁵⁷ Similarly to what has been previously observed for alum (see previous paragraph), LT and LTK63 increased IL-1β production in vitro by activating the NLRP3 inflammasome complex. In agreement with in vitro data, IL-1R has been implicated in IL-17 and IFN-γ CD4+ T-cell responses elicited by parenteral immunization with LT, LTK63, and keyhole limpet hemocyanin antigens. It remains to be demonstrated if the same pathway is involved in mucosal adjuvanticity.