Research Article: Vaccines: From Empirical Development to Rational Design

Date Published: November 8, 2012

Publisher: Public Library of Science

Author(s): Christine Rueckert, Carlos A. Guzmán, Tom C. Hobman.


Infectious diseases are responsible for an overwhelming number of deaths worldwide and their clinical management is often hampered by the emergence of multi-drug-resistant strains. Therefore, prevention through vaccination currently represents the best course of action to combat them. However, immune escape and evasion by pathogens often render vaccine development difficult. Furthermore, most currently available vaccines were empirically designed. In this review, we discuss why rational design of vaccines is not only desirable but also necessary. We introduce recent developments towards specifically tailored antigens, adjuvants, and delivery systems, and discuss the methodological gaps and lack of knowledge still hampering true rational vaccine design. Finally, we address the potential and limitations of different strategies and technologies for advancing vaccine development.

Partial Text

Scourges of humanity, such as smallpox, polio, and measles, have been controlled by vaccination. Other epidemics, for instance tuberculosis, have yet to be sufficiently restrained by immunization. Accordingly, policy makers have given a high priority to the development of novel vaccines to induce protective immunity against selected pathogens. Most human vaccines contain attenuated or killed pathogens and were developed empirically, such as the yellow fever vaccine [1], [2]. Safety concerns were associated with undefined vaccine preparations based on whole pathogens (e.g., inactivated or attenuated bacteria or viruses). Thus, novel subunit vaccines are based on a restricted number of individual components (i.e., antigens) of the specific pathogen, which are able to confer protective immunity. Obviously, the chances of finding effective components of subunit vaccines empirically are low. Immunogenic parts of pathogens that provide antigens for B cell receptors (BCRs) and antigenic peptides that are presentable by MHC molecules to T cell receptors (TCRs) have to be identified. It is critical to compensate for excluded pathogen-associated molecular patterns (PAMPs), which activate the innate immune system to induce an appropriate adaptive immune response. Finally, vaccine delivery systems may be needed. Hence, the rational design of vaccines is mandatory.

Selecting the optimal antigen represents the cornerstone in vaccine design. With the advent of genomics, the traditional process of selecting candidate antigens one by one has been replaced by reverse vaccinology approaches. Namely, the coding potential of a pathogen’s genome is exploited by in silico selection, high throughput screenings, and profiling technologies (e.g., genomics, proteomics) to define promising antigens in relation to in vivo expressed genes and clonal variation [3]–[6]. Importantly, this approach is not suitable for nonproteinaceous antigens. Depending on the desired response, the antigenic protein should contain appropriate BCR epitopes and peptides that can be recognized by the TCR in a complex with MHC molecules. Synthetic peptides produced at comparably low cost can also be incorporated in subunit vaccines. This is relevant especially in epidemic situations when large amounts of vaccine doses need to be produced in a very limited period of time. A peptide-based vaccine meets high safety standards due to the possibility of excluding allergens, toxins, or other functional molecular domains of the pathogen. Restricting the immune response to defined antigenic regions can, furthermore, help avoid effects such as autoimmune responses, dominant responses against epitopes prone to antigenic drift, or responses against epitopes with specificity for a particular strain rather than multiple strains of the pathogen. However, the identification of immunogenic peptide sequences requires a considerable amount of experimental effort. Computational prediction methods can strongly reduce time and costs for vaccine development. Nevertheless, clonal variability and in vivo selection resulting in immune escape could render ineffective a vaccine based on short peptides encompassing a limited number of epitopes. Furthermore, there are technological constraints associated with this approach (e.g., synthesis of long polypeptides).

Subunit vaccines are likely to lack the molecular cues needed for efficient activation of the innate immune system, thereby failing to induce vigorous adaptive immunity. PAMPs can act as adjuvants, however many pathogen-derived products might exhibit toxic activity [39]. The only globally approved adjuvant for humans is alum. It facilitates TH2-dependent immune responses but promotes less effective cytotoxic responses and can cause side effects. A number of other adjuvants have been recently approved for use in defined human vaccines, such as MF59 and monophosphoryl lipid A-containing formulations [40], [41], and there are other candidates in the pipeline. Adjuvants are not licensed per se, but as part of vaccine formulations. This together with stringent requirements for reagents used on healthy individuals raise the costs of clinical development [41]. Considerable effort was invested in the development of adjuvants for mucosal immunization [42]. Vaccination via mucosal routes is known to elicit both mucosal and systemic immunity [43], fighting pathogens at the site of entry. However, safety issues were observed following intranasal vaccination with the heat labile toxin of Escherichia coli and its attenuated derivative [42], [44]. This will need to be considered for current candidate mucosal adjuvants, among them compounds with well-defined molecular targets, such as PAMPs, cytokines, and cyclic di-nucleotides [45]–[47]. For example, the TLR9-agonist CpG enhanced immune responses after vaccination against hepatitis B, anthrax, influenza, and malaria [48]–[51] and proved promising in vaccination of otherwise nonresponsive immune-compromised organisms [52]. However, many molecular mechanisms of adjuvanticity are still elusive. First insights were gained in receptors and signaling pathways involved in the recognition and processing of pathogenic factors and adjuvants in cells of the innate immune system [53]–[55]. Nevertheless, the discovered mechanisms of adjuvanticity do not translate to generally applicable strategies for rationally designed vaccines (see also [4]). Hence, to date, adjuvantation requires an additional solid theoretical background for systematic implementation in rational vaccine design.

Delivery systems become necessary when antigens are not efficiently transported to the inductive sites or presented to the immune system. For example, rapid degradation can result in weak or virtually absent responses to otherwise immunogenic antigens. The coding sequence of an antigen can be integrated into a live virus-vector, which infects antigen-presenting cells (APCs), preferentially dendritic cells (DCs) [56], [57]. The antigen is then directly presented by MHC molecules and can be recognized by TCRs. The continuous antigen expression leads to its persistent exposure to immune cells. Recombinant viral vectors can be modified with regard to effector cell targeting, expression promoters, and the type of antigenic transgene. Lentiviral vectors with improved safety and efficiency parameters have a comparatively high capacity for encoding transgenes, high transduction efficiency, low anti-vector host immunity, low genotoxicity, and persistent gene expression [58]. They proved promising in vaccination of mice with HIV-derived antigens and in nonhuman primates with SIV-derived antigens [59], [60]. In spite of the adenoviral vaccine vector’s known limited efficacy due to preexisting immunity in large populations [61], [62], it still induces protective immune responses with characteristic induction of CD8+ T cells in humans [63]. Recombinant adenoviral vectors derived from uncommon human serotypes, chimpanzee or human/chimpanzee chimeras can circumvent the problem of host immunity [64]–[66]. Human cytomegalovirus (hCMV) vaccine vectors are based on the ability of hCMV strains to superinfect individuals with persistent hCMV infection and immunity. Rhesus macaques developed specific CD4+ and CD8+ responses against SIV antigens delivered by a recombinant CMV vector [67], [68]. Elucidation of the molecular mechanisms leading to memory inflation during chronic hCMV infections might even lead to hCMV-based strategies to trigger life-long responses. Attenuated recombinant poxviruses are also intrinsically immunogenic, and insights in the promoted innate immune responses have accumulated [69]. The above-described vectors have considerable potential in human vaccination, especially in prime-boost regimens aimed at fine-tuning responses [70]. Different attenuated or commensal bacteria have also been successfully exploited for delivering vaccine antigens and biologicals [71]–[75].

Understanding what is needed to confer protection without side effects is a prerequisite to develop a tailored intervention. To date, characterization of human responses to vaccination relies mainly on measuring antibody titers or cellular responses from peripheral blood samples. This does not allow a comprehensive analysis of responses with regard to the effector cells or mechanisms stimulated and the status in all relevant compartments for acquired immunity. Efforts to tackle this problem link the regulation of transcription or protein activity to the prediction of vaccination outcomes [91]. Recent reports suggest the potential of systems vaccinology for the analysis of gene expression profiling experiments to identify patterns or signatures linked to a desired outcome of vaccination [92]–[94]. Human studies showed correlations of gene expression profiles or protein expression patterns with immune system activation upon vaccination against yellow fever and influenza in responders and nonresponders [95]–[97]. Others characterized transcription profiles after treatment of mice or murine DCs with adjuvant molecules [98], [99]. Correlations between successful immunization or toxic events and cellular expression profiles can be predictive for a particular vaccine. However, no general unambiguous markers were identified that would allow accurate prediction of efficacy or safety for vaccines in trials (introduced, for example, in [5]).

In this review we elaborate on recent achievements that facilitate rational vaccine design. There are many visions on the expected impact of reverse vaccinology, epitope prediction, structural vaccinology, systems vaccinology, and personalized medicine on the rational design of effective vaccines [3], [5], [6], [106]. However, the implementation of these concepts towards the development of new and more potent vaccines requires time and considerable financial investment. Rational vaccine design will rely strongly on the availability of clinical data on individuals with different clinical forms of disease or response to vaccination to learn what is needed for protection [107]. The gaps in knowledge on the immune system’s specific clearance mechanisms against many pathogens slow down the identification of the immune response that should be evoked by tailored vaccines in different population groups (Table 1). Many aspects of the host pathogen interaction and host immune status during persistent infection are also poorly understood, thereby hindering the development of therapeutic vaccines [108]. Further data from trials with empiric formulations are required to identify patterns or biomarkers that can reliably guide prediction of vaccine efficacy and safety at reasonable success rates (Figure 1). A widely accepted goal in vaccine development is the applicability to huge populations, if not all humankind. Nevertheless, there are reasons for more personalized approaches that consider specific preconditions in recipients, such as genetic background, pre-exposure to pathogens or vaccines, unique physiological background related to local culture/habits, age, and immunodeficiency.