Current concepts and progress in RSV vaccine development

Abstract

Respiratory syncytial virus (RSV) disease is an important cause of morbidity and mortality in children and debilitated adults and remains one of the major global unmet challenges for vaccine development. Several immunological issues have delayed the development of vaccines, especially the poorly protective response to natural infection and the enhancement of disease following administration of formalin inactivated vaccines during trials conducted in the 1960s. Advances in knowledge of the immune system, of the virus and its antigenic properties combined with new vaccine technologies are now injecting new hope into the field and have given rise to many promising vaccine approaches. Some of these may be optimal for use in children, while others may be more appropriate for pregnant women or vulnerable older adults. With a multi-pronged approach to prevention, we propose that it may be possible to destabilise community circulation of RSV and thus to significantly lessen the impact of RSV disease.

The burden of respiratory syncytial virus disease

Respiratory syncytial virus (RSV) is the leading cause of bronchiolitis in infants and a significant cause of morbidity and mortality in the elderly. It infects 65% of children in first year of life and continues to cause significant disease in children up to 5 years of age, affecting 34 million children each year of whom 3.4 million need admission to hospital and some 200,000 die; 99% of these deaths are in developing countries [1]. A recent prospective study of 56,560 children in Argentina showed that of 1293 children with respiratory infections, 61.6% were infected with RSV; of these, 13% had life-threatening disease [2]. The annual attributable mortality rate for RSV was 0.7 per 1000 infants, showing that life threatening and fatal RSV infections are common in the developing world.

In recent years, RSV has also been recognized as a significant problem in debilitated and elderly persons in whom infection may lead to cardiac failure and secondary bacterial pneumonia. Each year, 3-10% of adult colds are caused by RSV, and in the USA, 78% of the deaths due to RSV infection are in individuals over 65 years of age [3].

Despite measurable immune responses to primary and secondary infection, symptomatic RSV disease recurs throughout life [4]. Estimates of financial impact are largely limited to the direct costs of healthcare, but have been put at $680 million per year in the USA [5]. Although normally RSV causes only transient upper respiratory symptoms, severe RSV infection leads to lower respiratory tract illness such as pneumonia or bronchiolitis, resulting in cough, wheeze, respiratory obstruction and hypoxia [6]. The peak age of bronchiolitis is 2–4 months of age, perhaps reflecting the decline in protection afforded by maternal antibody, the maturation of the immune system and the physiological and anatomical features of the respiratory tract during this period of development [7,8]. Children reaching this susceptible age at the time of peak RSV circulation are most vulnerable. In addition, bronchiolitis appears to be causally related to wheeze in later life [9,10]. However, it remains unclear whether the delayed effects of early life infection are due to programming of a harmful adaptive response or due to other hitherto unrecognized factors.

The pathogenesis of RSV disease is diverse, but involves a dysregulation of immune responses in the airway with enhanced inflammation in the bronchial mucosa and surrounding tissue [4]. This is characterized by activation of innate immune responses and the appearance of neutrophils in the bronchial fluid [11–13]. The role of Th2 responses is not clear; in most cases, there is no lung eosinophilia (as there may be in some older children with asthma) [14]. Unlike asthma, viral bronchiolitis does not respond to steroids or bronchodilators [15,16]. In disease enhanced by prior vaccination with formalin-inactivated virus, the pathogenesis does appear to be Th2 related [17,18], and in some elderly persons with chronic lung disease (chronic bronchitis, emphysema and asthma), it is also possible that Th2 cytokines are involved [19].

Despite increased appreciation of the large global impact of RSV disease, there remain no specific antivirals and no licensed active vaccine. A number of significant hurdles remain; not least the many unanswered questions about the human immune response to RSV infection. RSV vaccine development is among the priorities for organizations such as the WHO and PATH as well as the pharmaceutical industry, which recognizes the unmet need. In addition, the Bill & Melinda Gates Foundation's current pneumonia strategy focuses on pneumococcus, influenza and RSV, and especially on maternal immunization to protect mothers and their newborn children [20]. Increasing investment therefore promises to accelerate both fundamental research and clinical development.

Antigenic properties of RSV proteins

RSV is an enveloped, nonsegmented negative-sense, single-stranded RNA virus belonging to the Paramyxoviridae family. Its genome is made up of 10 genes that encode 11 proteins. With the exception of the attachment glycoprotein (G), the viral proteins are very well conserved. The three surface glyco-proteins of RSV, fusion (F), small hydrophobic and G proteins, are the main targets for humoral immune responses. G is an attachment protein that mediates virus binding and is important for efficient replication in vivo [11,12]. It is produced in abundance and is not only cell/virus associated but also secreted, a form that may act as a decoy for antibody [21]. In addition, it bears a conserved CX3C fractalkine-like motif, which is thought to modulate inflammatory responses [11]. Recent in vitro studies have demonstrated the inhibitory effect of this chemokine analog on type I/III interferon production by epithelial cell lines and plasmacytoid dendritic cells with additional effects on TNF expression by plasmacytoid dendritic cells and IFN production by T cells [22]. This may occur via the inhibition of toll-like receptor (TLR) 3/4-dependent mechanisms and may contribute to impaired viral clearance [23]. Immunization of mice with RSV G protein-induced neutralizing antibodies that block G protein-CX3CR1 interactions with both RSV A and B strains [24], prophylactic treatment with which could reduce respiratory disease [25]. However, G protein is highly glycosylated and poorly conserved across RSV strains. Therefore, the majority of neutralizing antibodies against G protein are against hypervariable serine–threonine-rich domains that are heavily O-glycosylated, although induction of antibodies against the central conserved region of G could offer an advantage by reducing RSV-induced tissue damage.

In contrast, F protein is well conserved and, unlike G, is essential for infectivity, being responsible for fusion of the viral and cellular membranes [26,27]. It also has immunomodulatory functions and can activate TLR4. In human monocytes in vitro, the production of IL-6 in response to F protein is dependent on TLR4 signaling, and TLR4 knockout mice are unable to clear the virus [28]. Furthermore, analysis of single-nucleotide polymorphisms in TLR4 showed that mutations that contributed to reduced responsiveness to RSV in vitro were associated with increased susceptibility to RSV infection in high-risk infants and young children [29,30]. Thus, the immuno-genicity of the F protein coupled with its high degree of sequence conservation across all RSV strains has dictated its selection as the major vaccine antigen.

Recent advances in the understanding of the structure of F protein have invigorated RSV vaccine development. F protein is metastable and synthesized in a prefusion form that is stabilized on the virion. Proteolytic cleavage leads to a conformational change that causes the fusion of adjoining membranes [27]. Until recently, only the structure of the postfusion F protein was known [26]. A number of antigenic sites had been described including that recognized by palivizumab [26]. However, studies in which human sera were adsorbed on columns coated with postfusion F indicated that the greater proportion of virus neutralization was not due to antibodies against epitopes in this confirmation [31]. Recent studies have elucidated the structure of prefusion F and identified a novel antigenic site, termed site Ø [32]. Antibodies that recognize this epitope offer theoretical advantages by preferentially targeting F protein on infectious virions. However, analysis of site Ø suggests a higher degree of sequence variation surrounding it, implying that while antibodies recognizing this site may have greater antiviral efficacy, they might also encourage viral mutation in an attempt to evade immune pressure.

The inner viral proteins are also well conserved and are major targets for cellular immunity. NS1 and NS2 are non-structural proteins that mediate inhibition of STAT2 expression and early IFN production [33]. Matrix (M) protein lines the inner surface of the viral envelope, having a central role in budding and formation of infectious virus [34]. Nucleoprotein (N), phosphoprotein (P) and polymerase (L) have roles in encapsidation, RNA replication and transcription, respectively [11]. It is of note that the minimal construct for an infectious virus-like particle (VLP) requires the M, F, N and P proteins [35]. We recently reviewed the role of cellular immunity in RSV, but comparatively little is known about the importance of human T cells in protection and immunopathology [4]. Further understanding of the targets of T cell immunity and the role of individual T cell subsets in viral clearance and B cell help will be necessary to direct the inclusion of RSV proteins or peptide epitopes in novel vaccine candidates.

Protective immunity against RSV

The majority of successful vaccines rely on replicating aspects of the protective immune response induced by natural infection without the pathological sequelae. However, since natural RSV infection does not induce complete immunity, simply imitating the response to infection is unlikely to be sufficient. There are two serogroups of human RSV (hRSV), termed A and B, in addition to multiple genotypes that can be distinguished by viral sequencing. Despite this, human sera and many monoclonal antibodies cross-react to both A and B serotypes, and immunity to RSV is only partially strain specific. In adult challenge studies, a single strain can reinfect with relative ease [36]. In addition, longitudinal studies show that reinfection can occur within even a single year, either with the same or a different serotype [37]. These findings suggest that RSV evades immunity in order to reinfect, but how it does this remains obscure.

It also remains unclear why some individuals suffer severe disease leading to respiratory failure, while others infected with the same virus show only minor or unapparent disease. Hypotheses include severe disease being driven by high viral load due to defective early control of viral replication. This might be compounded by RSV's ability to inhibit interferon (IFN) responses via early and abundant expression of viral NS proteins [38]. In addition, poor immune regulation may cause an overexuberant inflammatory response [4]. Thus, for a vaccine against RSV to be successful, it will need to induce immune responses beyond those stimulated by natural infection, but must also avoid the induction of harmful immunopathology.

Role of antibody in protection against RSV

Host responses to natural infection are only partially protective, but primary infection during infancy can be delayed by parenteral administration of palivizumab (Synagis), a humanized anti-RSV F protein monoclonal antibody. Palivizumab reduces hospitalization of at-risk infants by around 50% [39], suggesting that there are stable, serotype-independent neutralizing epitopes in the RSV F protein, and that antibody alone is capable of protecting against severe RSV disease [32].

A number of studies have investigated the role of serum antibody as a correlate of protection against RSV disease. Most have shown an association between low serum-neutralizing antibody titers and increased risk of RSV disease [40,41]. A number have also attempted to identify threshold values above which the likelihood of hospitalization with RSV falls significantly [42]. However, a lack of assay standardization makes it difficult to compare studies and populations. Furthermore, an upper threshold above which complete protection is seen has not been found. Therefore, in most RSV-experienced individuals, the variability in infection risk and disease severity is not fully determined by antibody levels. Thus, while the literature suggests that those with little or no neutralizing antibody are indeed at high risk of infection, even the highest levels of antibody induced by natural infection are insufficient to provide absolute protection. This is compounded by studies that indicate the short-lived nature of anti-RSV antibodies, which are only boosted temporarily on secondary infection and which drop rapidly with no sustained increment [43].

Palivizumab needs to be administered once a month during the RSV season in order to prevent disease and has no therapeutic efficacy in established infection [44]. In most countries, it is only given to children at greatest risk due to its high cost and the relatively low probability of preventing hospitalization in healthier individuals (since most children only suffer mild disease). Therefore, most current RSV vaccine strategies hope to replicate palivizumab's clinical efficacy through the induction of antibodies against F protein. Whether it will be possible to induce high-affinity neutralizing antibodies that persist at sufficiently high titer to provide long-term protection remains to be seen.

Cell-mediated immunity against RSV

The production of antibodies depends on T cell help [45]. Furthermore, if sterilizing immunity cannot be achieved by antibody, antigen-specific T cells may assist in the clearance of infected cells. In small animal models, depletion of CD4⁺ or CD8⁺ T cells impairs RSV clearance [4]. However, T cells have also been implicated in increased disease severity [46]. It therefore remains unclear whether induction of high frequencies of antigen-specific T cells by RSV vaccination is desirable. Recent studies in influenza have demonstrated the beneficial effect of pre-existing antigen-specific CD4⁺ and CD8⁺ T cells on subsequent disease severity, but no such evidence exists in RSV [45]. In infants hospitalized with bronchiolitis, CD8⁺ T cells dominate the adaptive response, which might suggest a role in pathogenesis [47]. However, children with defects in cell-mediated immunity suffer prolonged RSV infections, implying that induction of T cells may be beneficial as part of a balanced immune response [48]. Further analysis of T cell responses following infection is required to further delineate their role.

Impact of a failed vaccine

One of the key contributors to the slowing of RSV vaccine development was the failure in the 1960s of a formalin-inactivated vaccine candidate (FI-RSV). Formalin treatment is a common approach used for the inactivation of whole virus vaccines. It diminishes the replicative function, but the outer virion coat remains intact, which aids immune recognition. The method resulted in production of a safe and effective vaccine against poliovirus. Kim et al. developed a FI-RSV vaccine (Lot-100), which was tested in infants and children aged between 2 months and 9 years in the late 1960s [49]. FI-RSV vaccination was delivered intramuscularly in two or three doses separated by 1-3 months. Early results were promising; 43% of the vaccinees had a minimum fourfold increase in neutralizing antibodies and 91% had a minimum fourfold increase in complement fixing antibodies. However, upon natural RSV infection, 80% of the FI-RSV-vaccinated subjects were hospitalized, whereas only 5% of the parainfluenza-vaccinated control group required admission. Two children died and adverse effects were apparent in vaccinated infants, who were diagnosed with pneumonia, bronchiolitis, rhinitis and bronchitis.

The postmortem analysis of the lungs from the two fatal cases reported infiltration of mononuclear cells, neutrophils and eosinophils [49]. Further analysis of samples indicated that antibody–antigen complex deposition and complement activation were seen in the lung tissue. On more detailed investigation, serum antibodies directed against the surface proteins G and F in the sera of FI-RSV vaccinees were found to be poorly neutralizing, almost certainly contributing to decreased vaccine efficacy [50]. In addition, formalin inactivation was found to cause the formation of reactive carbonyl groups, which affected the quality of antibody responses [17].

Supporting the concept that cellular immune responses contributed to the vaccine-enhanced disease, in vitro studies showed exaggerated proliferation of lymphocytes from peripheral blood mononuclear cells (PBMCs) of the FI-RSV vaccinees compared with a nonvaccinated control group following natural infection [51]. A number of animal models of FI-RSV vaccination have subsequently shown skewing of virus-specific CD4⁺ T cell responses toward a Th2 phenotype [17,52]. Th2 cells are the main producers of the cytokines IL-4 and IL-5, which typically support eosinophilia, IgE production and airway hyper-sensitivity [53]. It has been proposed that this Th2 bias might be due to the absence of the normal Th1 response elicited by viral infection with paucity of cytotoxic CD8⁺ T cell responses related to nonreplicating virus. In addition, it has recently been shown that enhanced disease is associated with a profound loss of regulatory T cells during viral challenge [54].

Thus, animal studies suggest that, via a number of dysregulated adaptive mechanisms, FI-RSV vaccination promoted harmful immunopathology while providing no immune protection. Although these studies must be interpreted carefully, caution surrounding the risks of any novel vaccine causing similar harm has compelled more systematic approaches that have rightly prioritized safety over speed of advancement.

Animal models of human RSV infection

The adverse effects of FI-RSV vaccines highlight the importance of preclinical safety testing of vaccine candidates. However, the use of animal models in hRSV has been problematic and no single animal model adequately replicates all aspects of hRSV disease. For example, young children are typically most severely affected by RSV infection and the presentation of disease changes with age. In neonates, nonspecific presentation with symptoms such as periodic breathing and apnea is common, whereas older infants are more likely to exhibit upper and lower respiratory tract signs such as nasal discharge, wheeze and labored breathing [55]. In elderly adults, viral lower respiratory tract infections with RSV are less well studied, but it is likely that additional factors, such as bacterial superinfection, contribute to increased mortality [56].

RSV disease is usually confined to the respiratory tract. Although the receptor(s) for virus entry is not fully elucidated, RSV preferentially infects respiratory epithelial cells; infection of other cell types is seen in vitro only at high multiplicities of infection [57]. Studies restricted to systemic immune responses, including measurements of serum antibody, may therefore not address the most relevant compartment. Analysis of respiratory samples from natural RSV infection or RSV-challenged adult volunteers has shown production of various cytokines and chemokines, including IL-6, TNF-α, IL-8, IP-10, RANTES and MIP-1, which diverge markedly from those detected in the blood [58,59]. The activity of these early inflammatory mediators shapes the innate and adaptive immune responses. Furthermore, in the bronchoalveolar lavage (BAL) of naturally infected infants, neutrophils have been reported to be the most abundant cells, followed by macrophages, eosinophils and T lymphocytes, the relative importance of which remains unclear [60,61]. An appropriate animal model should therefore allow detailed examination of mucosal immunity, despite the current lack of understanding of equivalent responses in humans.

An ideal animal model should allow determination of which antigens to use in a candidate vaccine, the dose and route of delivery as well as any adverse side effects to the host. It should also mimic the clinical and pathological characteristics of hRSV infection. Practically, it is not possible to find all these features in one model, and it is therefore essential to choose the most appropriate model in order to study the cells, mediators and responses of interest. The main animal models used today include primates, bovine calves, mice and cotton rats.

Chimpanzees

Chimpanzees have a historical place in hRSV studies as the virus was first recognized as an infectious agent in this species as chimpanzee coryza agent [62]. Although chimpanzees allow hRSV replication upon experimental challenge, they only experience mild symptoms, including nasal discharge, sneezing and coughing, typically seen in upper respiratory tract diseases [63]. To date, there are no reports indicating lower respiratory tract disease in hRSV-infected chimpanzees. In addition, the maintenance, logistics, ethical issues and the limitation of reagents in working with these animals make them a less popular choice. Nevertheless, the similarity in their genetics and anatomy to man can have advantages for vaccine testing.

Cattle

Although sharing some of the logistic difficulties studying primates, studies of RSV infection in calves have considerable advantages. Cattle are a natural host of bovine RSV (bRSV), which does not infect humans, but is a ubiquitous pathogen in commercial herds and a leading cause of respiratory infection [64]. The spectrum of diseases it causes overlaps with hRSV disease. Infected calves may be asymptomatic, but may alternatively show severe respiratory signs (tachypnea, hyperinflation and cyanosis) and die of respiratory failure [65]. Antonis et al. describe the association of age with the development of severe bRSV disease: neonatal (1-day-old) calves are clinically less affected than young (6-week-old) calves but have greater lung pathology with lung neutrophilia and macrophage recruitment and an increase in the levels of IFN-γ, TNF-α and IL-6 in BAL fluid [66]. A disadvantage of experimentation on cattle is the considerable cost of experimental studies with large animals and the restricted range of laboratory reagents.

Cotton rats

Cotton rats are appreciated as important models for studying infectious human disease and are now used as standard for the evaluation of new vaccines, antivirals and prophylactic drugs in RSV. They provided the majority of the preclinical data that led to the licensure of palivizumab, with pharmacokinetics having been established in the model [67]. In cotton rats, RSV replication occurs in the lower airways and virus can be detected in both the upper and lower respiratory tract, leading to moderate bronchiolitis or pneumonia [68]. Upon experimental hRSV challenge, there is an induction of early IFN, chemokine and TLR responses [69].

The effect of FI-RSV vaccination has been studied extensively in cotton rats. Vaccine-enhanced disease occurs with infiltration of neutrophils, macrophages and lymphocytes. This has proven useful for the elucidation of disease mechanisms, but Shaw et al. highlight the role of nonviral components in these studies. FI-RSV is often a crude virus preparation contaminated with cell debris. Although priming and challenge with RSV-containing preparations lead to enhanced respiratory disease, virus-free cell culture components also cause sensitization [70]. To circumvent this problem, most laboratories use purified RSV stocks [71]. It is vital to include appropriate controls in these studies and to appreciate the complexities of the models.

Although cotton rats are clearly useful in preclinical studies of hRSV vaccination, the immunological reagents and techniques that are available to those using cotton rats are more limited than those available for mouse studies. There are no fully inbred or transgenic cotton rats, limiting studies of precise immunological pathways.

Mice

Mice are the most commonly used animals for experimental studies of hRSV infection. There are numerous reasons for this including the availability of immunological reagents, a complete genome sequence and the availability of transgenics, congenics and knockouts. The mouse genome is similar to humans with 99% of mouse genes having a human homolog [72]. Similar to cotton rats, mice are semipermissive hosts for hRSV replication. Using varying doses of inoculum between 10⁵ and 10⁷ pfu, a mild infection, moderate-to-severe lung pathology and illness can be induced [55]. It is important to highlight that differing substrains and isolates of RSV and mice result in different effects. The most commonly used combination is BALB/c mice and the A2 strain of RSV; such mice exhibit moderate-to-severe respiratory tract disease with signs of illness including weight loss, hunching, reduced motile activity, hypoxia and increased breathing rate [55]. Mice infected with hRSV have an increase in early innate cytokines, such as TNF-α, IL-6, MIP-1α and RANTES, in the BAL [73], which parallels findings from human infant studies [74]. Mice have been especially useful in studying the detailed responses of the innate and cell-mediated immune responses to RSV infection and in elucidating the effects of immune enhancement by prior vaccination [12].

Experimental human infection of adult volunteers with RSV

Despite the practical advantages and the availability of genetically modified animals that allow study of precise mechanistic pathways, there are major drawbacks in only using animals in experimental studies of RSV infection. Mice are mostly inbred and clean, whereas humans are outbred and have experienced a range of past and present infections and other environmental influences. Humans are continuously exposed to a variety of pathogens living as they do in diverse environments and suffering multiple infections throughout life, leading to complex antigenic histories that program their immune systems in a variety of ways. For these reasons, the immunobiology of humans may be radically different from that of mice. For example, it is known that mice that are older than 15 months are more permissive to RSV replication than younger animals [75]. Although there is no record of primary RSV infection occurring in adults or in elderly persons, this does not appear to be the case in man; most severe RSV cases occur before the age of 2 years. In addition, severe disease in mice leads to weight loss, which is not a major feature of human infection.

Preclinical studies in animal models may not adequately predict vaccine responses in human beings. Numerous vaccine candidates that have demonstrated excellent immunogenicity in animals have failed in man, while a number of vaccine candidates whose development was stopped due to poor preclinical efficacy might possibly have been effective in humans. However, clinical efficacy trials of RSV vaccines have a number of logistical difficulties to overcome. These include decisions on which patient groups to vaccinate and how to determine the outcome when the RSV incidence may be relatively low, particularly in adult populations.

The experimental hRSV infection model may have an important role in this regard. Experimental infection of volunteers with RSV has an extensive history with substantial work in the field having been carried out at the MRC Common Cold Unit from 1946 to 1989. In these studies, adults are inoculated with a fixed dose of a well-characterized virus, quarantined and sampled longitudinally during the course of infection. Compared with natural infection studies, this offers the advantages of standardized study populations, defined as virus inoculum, the ability to examine immune correlates both before and after inoculation, definite diagnosis of infection and the richness of longitudinal data analysis [58]. For the purposes of vaccine development, this allows the determination of correlates of protection and can demonstrate efficacy of protection by vaccine candidates without reliance on the relatively unreliable event of natural RSV exposure. Although infection of healthy adults does not necessarily recapitulate at-risk infants or immunosuppressed individuals, it may directly inform attempts to vaccinate other groups of adults including pregnant women and the healthy elderly.

Who to vaccinate against RSV infection

Although severe RSV primarily occurs in young children and the elderly, a variety of issues make these populations particularly difficult to target. Vaccine candidates are available that would be suitable to go into some of these populations (including live attenuated viruses in infants or young children and subunit vaccines for the elderly), and these may have both direct benefits to at-risk groups and interrupt transmission between susceptible groups, vectors and community reservoirs (Figure 1). However, high-risk populations are likely to have impaired immune responses, either due to immaturity, immunosuppression or immunosenescence. Vaccine safety is clearly an important issue, with the history and understanding of FI-RSV having caused extreme reluctance to vaccinate RSV-na'ive infants with an inactivated vaccine. In elderly adults, the efficacy trials that will be required for licensing may be impractical in view of the number of participants that would be required for an adequately powered study.

Figure 1. Targeting different populations with immune interventions against respiratory syncytial virus (RSV).

Multiple checkpoints may be targeted for the interruption of RSV infection. At-risk individuals may be protected directly through induction of protective immunity or indirectly via reduced circulation of infection in the community. Vaccination of healthy individuals aims to promote herd immunity and/or prevent transmission between school-aged children and adults with whom they are in contact. This would interrupt the amplification loop within households, hospitals and schools, reducing transmission to vulnerable populations (i.e., neonates, immunosuppressed individuals and the elderly). Vaccination of pregnant women during the last trimester aims to enhance maternofetal transfer of IgG to protective levels in term infants, but passive vaccination would probably still be required for preterm infants or immunosuppressed adults. Inactive or subunit RSV vaccines are most appropriate for RSV-experienced individuals in whom the risk of vaccine-enhanced disease is low. Live attenuated vaccines are likely to be safe in children, and the induction of mucosal immunity would have the added benefit of reducing virus shedding.

Recent emphasis has therefore fallen on maternal vaccination as a strategy for boosting protection in the neonate. IgG is actively transported across the placenta during the final weeks of pregnancy, and epidemiological studies have shown a negative correlation between the level of maternal neutralizing antibody and infectious disease incidence [76]. Recent studies examining maternal vaccination with trivalent inactivated influenza vaccine have suggested benefit to the mother and protection of infants up to 6 months of age. For these reasons, there has recently been renewed emphasis on the development of a maternally administered RSV vaccine in the hope of elevating the infant IgG level beyond the lower threshold of neutralizing antibody titer above which the risk of severe disease is reduced. A variety of vaccine strategies may be successful in this regard, with F protein subunit, VLPs and vectored vaccines all having shown success in the induction of neutralizing antibodies in preclinical models as well as some early Phases I and II trials [77–80]. Although this strategy would not protect premature infants, prolonging the window during which maternal antibody reduces the risk of infection before the immune system matures sufficiently for a normal anti-RSV response would be a major advance.

Current progress in vaccine development

Advances in our understanding of the structural biology of RSV F protein, estimates of global disease burden and the immune responses to infection outlined above, have led to widespread acceleration in the development of RSV vaccines, and these have recently been reviewed extensively elsewhere [81–83].

The spectrum of vaccine candidates in development ranges from live attenuated viruses to vector-based and subunit vaccines in a wide variety of forms, and it is likely that no single vaccine will be appropriate for all age groups. With the elucidation of the prefusion F protein structure, the number of particle-based and subunit vaccines that aim to deliver antigen in this form is on the rise. In early human studies, these appear immunogenic and well tolerated [77]. These have the potential to induce incremental rises in serum neutralizing antibody titer greater than natural infection, but the longevity of these responses and their ability to stimulate appropriate T cell responses are yet to be investigated in detail. Furthermore, inactive vaccines are unlikely to be suitable for use in RSV-naïve children for whom safety concerns will exist and may not induce significant local immunity unless delivered intranasally. A number of vector-based vaccines do seek to induce responses beyond serum neutralizing antibody, and these offer the promise of inducing a more balanced immune response that may be preferred.

With this increase in knowledge and enhanced incentives to vaccine development, there are now many potential vaccines at various stages of testing. The current pipeline is summarized in Figure 2, which is reproduced with kind permission of PATH. This 'snapshot' of RSV vaccine development is regularly updated and available online [84].

Figure 2. Vaccines currently in development against human respiratory syncytial virus infection.

Many candidate vaccines are at preclinical stage. However, there is live attenuated, particle-based, subunit and gene-based vectors being tested at Phase I or II trials. The respiratory syncytial virus vaccine snapshot is reproduced with permission from PATH and can be found at [84]. Reproduced with permission from the PATH [101].

Live attenuated virus vaccines

Live attenuated vaccine candidates have historically dominated Phases I and II trials. RSV can be attenuated by multiple strategies including cold passage, mutagenesis and reverse genetics. An early cold passaged, temperature-sensitive RSV vaccine (cpts-248/404) with mutations in N, F and L genes, when delivered intramuscularly to seronegative infants, was safe but poorly immunogenic [85]. Encouragingly, this vaccine showed no significant side effects and reverse genetics was used to try to generate a more immunogenic strain. By introducing changes in the small hydrophobic gene, a further live attenuated RSV candidate (rA2cp248/404/1030ΔSH or Medi-559; MedImmune/National Institute of Allergy and Infectious Diseases, Bethesda, MD, USA) was developed and tested in a Phase I clinical trials on healthy children and infants [86]. The majority of vaccinees made a serological response and/or developed viral shedding after inoculation, but an increase in the number of medical attendances led to further safety testing requirements. A Phase II trial to determine its ability to induce protective immunity upon natural exposure to RSV is underway [87,88]. Medi-559 has recently been improved for stability, resulting in a new formulation that is indistinguishable from MEDI-559 in terms of the replication and disease induction in seronegative chimpanzees. This may replace Medi-559 in subsequent clinical trials [89].

Although Medi-559 is the leading attenuated vaccine candidate, there are other promising strains in Phase I clinical trials. Medi-534 is a live, attenuated RSV/PIV3 chimeric virus vaccine candidate that has been tested in seronegative children between 6 and 24 months of age [80]. This vaccine candidate is thought to be safe, and administration of three doses over 4 months showed that the virus was infectious and immunogenic in the tested population [80]. A recent report examining local immune responses to a range of live attenuated viruses from Karron et al. showed increases in pro-, anti- and regulatory cytokines in nasal samples, suggesting that appropriate cytokine responses might contribute to additional protection [90].

Unfortunately, despite their good safety record and absence of vaccine-enhancement issues, poor immunogenicity, under-attenuation, genetic reversion and secondary transmissibility have dampened enthusiasm for live attenuated RSV vaccines [90]. Despite this, continuing advances in understanding of RSV disease and the technologies involved in virus generation may allow further development of optimally attenuated viruses for future study.

Particle-based vaccines

A number of VLP vaccines are in preclinical development. These multiprotein structures mimic the viral organization but lack the viral genome, providing noninfectious particles. The main advantage of VLPs is that the potency of the vaccine can be enhanced by exclusion of viral proteins with immunosuppressive function while presenting antigen in an optimum configuration. In addition, VLPs can bind pattern recognition receptors to trigger of innate immunity, which can promote B cell responses in the absence of adjuvants [91]. However, VLP manufacture may be difficult to optimize; despite the number of expression vectors that can be used to synthesize VLPs, the glycosylation and the correct folding of the proteins may be a limitation for their use in humans. Glycosylation cannot be achieved by bacteria and yeast is limited to high mannose glycoprotein modification that is inconsistent. Using mammalian culture systems is an option that favors the modification and assembly of the particles, but it is less controllable and more costly [92].

One leading VLP vaccine candidate against RSV is an F nanoparticle (Novavax, Inc., 9920 Belward Campus Drive, Rockville, MD 20850, USA). This particle is an insect cell-derived RSV F protein vaccine that is safe and protective in cotton rats [93]. As part of the Phase I clinical trials, it was administered intramuscularly to healthy adults [77]. The vaccine was found to be safe with no adverse effects, inducing a minimum sevenfold rise in anti-F IgG responses and achieving comparable to titers to those seen after administration of palivizumab.

Subunit vaccines

Immunization with purified proteins isolated from pathogens has a long and successful history. However, there are few subunit vaccines licensed for viral infections, and these are thought to assemble into VLPs [94]. Early clinical trials examined purified F protein (PFP) formulated with alum [95–98]. Gonzalez et al. evaluated the effects of immunization with the live attenuated virus cpts 248/404 followed by PFP-2 in adults in whom both vaccines were well tolerated. They induced a minimum fourfold rise in neutralizing antibody titers in both young and elderly participants [96]. Immunization of pregnant women with PFP-2 was also safe [95]. Following immunization, 95% of the recipients had a minimum fourfold increase in IgG levels. In addition, a modest increase in neutralizing antibody titers was observed in the babies at birth, 2 and 6 months after delivery. Anti-F IgA and IgG concentrations in breast milk were higher in mothers who received the PFP-2 vaccine [95].

More recently, a vaccine candidate containing a formulation of F, G and M subunits with or without alum has been trialed in elderly persons. All formulations were well tolerated, regardless of adjuvant. Interestingly, the nonadjuvanted formulation demonstrated most efficacy with 58% of the subjects demonstrating a minimum fourfold increase in the neutralizing antibody titers against RSV A [98].

Subunit F vaccines are well tolerated in seropositive individuals and induce neutralizing antibody, but in seronegative infants, there is a theoretical concern that immunization will induce vaccine-enhanced disease. Several groups are pursuing PFP subunit vaccines through Phase I clinical trials for the vaccination of adults, and particularly of pregnant women (Figure 2).

Vector-based vaccines

Viral vectors were initially developed for gene therapy, but their ability to induce innate and adaptive responses in mammalian hosts has also made them potential vaccine vehicles. Multiple vectors derived from viruses or bacteria exist for the delivery of genes-encoding proteins of interest. These vectors have the major advantage of mimicking natural infection in the host, and viral vectors allow transcription and translation via host cell machinery. This is not only important for processing and presentation of antigen to achieve priming of CD4⁺ and CD8⁺ T cells but also for the induction of relevant antibody responses. Moreover, vectors can be rendered replication incompetent and cannot evade immunity through selective gene expression, increasing their safety. Adenoviruses are a particularly attractive vector for vaccine delivery for many reasons: they can be grown to high titers in tissue culture, they can be inoculated systemically or at the mucosal surfaces and there are multiple serotypes available for use in humans [99]. Geraldine Taylor's group at the Pirbright Institute, in collaboration with Okairos (Basel, Switzerland), has recently presented work on a bRSV model where newborn calves without maternally derived RSV-specific antibodies were primed intranasally with a replication-defective chimpanzee Adenoviruses/RSV and boosted with Modified Vaccinia Ankara/RSV 8 weeks later. On Week 12 postpriming, the animals were challenged with bRSV and samples analysed on Week 13. Compared with control groups, prime/boosted animals had a significant rise in serum IgG, mucosal IgA and frequency of IFN-γ-producing CD4⁺ and CD8⁺ T cells, and calves were completely protected against bRSV infection. This strategy is now being evaluated in Phase I clinical trials [100].

Conclusion

Historically, RSV vaccine development has been hindered by two major stumbling blocks: limited immunogenicity and vaccine-enhanced disease. However, recent advances have increased likelihood that progress can be made in vaccination against RSV, which remains one of the greatest remaining challenges for vaccine development. This imperative is driving the research and development toward making rapid progress. We are of the opinion that the future for an RSV vaccine is now bright, and that the approaches outlined in this review have the potential to achieve a safe and effective vaccine.

Expert commentary

Of the major unmet challenges posed by infectious disease, prevention of RSV through vaccination is widely viewed as being among the most likely to be solved in the next decade, at least for certain target groups. Compared with tuberculosis, HIV or malaria, RSV infection would seem easier to prevent; however, attempts at developing an effective vaccine have been hampered by fundamental difficulties: poor immunogenicity, reactogenicity, fear of immune-mediated disease enhancement and antigen production.

The research agenda developed in academic laboratories has now been passed to small and large pharmaceutical companies with substantial records of success and well-developed vaccine divisions. Many lessons have been learned in other fields of vaccinology that can be applied to RSV vaccine research; with improved understanding of the key structural and antigenic requirements of RSV vaccination, we are optimistic that an RSV vaccine will be available within the next 10 years.

Five-year view

With the current level of increased activity in the field of RSV vaccine development, we anticipate that RSV vaccines may become available for use in one or more of the target groups. In addition to improved prophylaxis, we anticipate that it may be possible to interrupt RSV's community circulation by such interventions.

Key issues.

• Improved understanding of protective and harmful immunity to respiratory syncytial virus.

• Protein engineering of antigens that can induce broadly neutralizing antibody responses against respiratory syncytial virus.

• Tailoring prevention to specific vulnerable or important groups (infants, toddlers, adult care givers, pregnant mothers and the elderly).