The development of veterinary vaccines: a review of traditional methods and modern biotechnology approaches

REVIEW ARTICLE

The development of veterinary vaccines: a review of traditional methods and modern biotechnology approaches

Sérgio Jorge, Odir Antônio Dellagostin∗

Universidade Federal de Pelotas, Centro de Desenvolvimento Tecnológico, Núcleo de Biotecnologia, Pelotas, RS, Brazil

Received 29 September 2017; accepted 2 October 2017

Available online 16 October 2017

KEYWORDS

Vaccine antigen;

Reverse vaccinology;

Animal vaccination

Abstract The immunization of animals has been carried out for centuries and is generally accepted as the most cost-effective and sustainable method of controlling infectious veterinary diseases. Up to twenty years ago, most veterinary vaccines were either inactivated organisms that were formulated with an oil-based adjuvant or live attenuated vaccines. In many cases, these formulations were not very effective. The discovery of antigen/gene delivery systems has facilitated the development of novel prophylactic and therapeutic veterinary vaccines. To identify vaccine candidates in genomic sequences, a revolutionary approach was established that stems from the assumption that antibodies are more readily able to access surface and secreted than cytoplasm proteins; as such, they represent ideal vaccine candidates. The approach, which is known as reverse vaccinology, uses several bioinformatics algorithms to predict antigen localization and it has been successfully applied to immunize against many veterinary diseases. This review examines some of the main topics that have emerged in the veterinary vaccine field with the use of modern biotechnology techniques.

© 2017 Sociedade Brasileira de Biotecnologia. Published by Elsevier Editora Ltda. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/).

Introduction

Vaccinations are an effective method of preventing a wide range of animal diseases. The field of vaccinology has yielded several effective vaccines that have significantly reduced the impact of some important diseases in both companion animals and livestock. Today, the vast majority of licensed veterinary vaccines are in the form of live attenuated, killed/inactivated microorganisms, cell membrane compounds or toxoids (McVey & Shi, 2010; Unnikrishnan,Rappuoli, & Serruto, 2012). Live attenuated vaccines can be very effective because they induce both cellular and humoral immune responses (da Costa, Walker, & Bonavia,2015; Rizzi et al., 2012). However, a major concern that is associated with vaccines of this nature is the potential risk of reversion of the microorganism for a virulent phenotype (Shimoji et al., 2002; Unnikrishnan et al., 2012).Killed/inactivated vaccines are typically safer; however,they may be less effective than attenuated vaccines. The commercial vaccines based on toxoids (inactivated toxins)have some drawbacks since they require complex components in culture medium. The limitations of the three existing vaccine types in combination with the fact that several diseases have yet to be successfully treated with an efficient vaccine entails there is a need for better and safer vaccines that can prevent, control or eradicate animal diseases (Dunham, 2002; Redding & Weiner, 2009).

Recombinant vaccines represent an attractive strategy by which the limitations of conventional vaccines can be overcome, and a number of rationally designed and sub unit vaccines have already reached the veterinary market.Efforts to develop more effective vaccines against a large number of diseases using recombinant DNA technology are in progress around the world. Recombinant vaccines are developed based on rationally designed recombinant highly purified antigens through structure-based design, epi-topes focusing or genomic-based screening (Correia et al.,2014; Dellagostin et al., 2011). In addition to enhancing understanding of the genes responsible for virulence and facilitating the identification of the determinants of protective immune responses, these molecular approaches have provided new methods of developing novel vaccines against infectious, parasitic or metabolic diseases.

However, the inherent immunogenicity of recombinant antigens is often low in comparison to the more traditional vaccines, and there is a need for potent and safe vaccine adjuvants to ensure that recombinant vaccines can succeed. The low immunogenicity frequently observed in recombinant antigens occurs due to a lack of exogenous immune activating components. Recombinant antigens can be offered in different adjuvants, and the immunomodulatory effects are dependent upon the particular adjuvant used in conjunction with specific antigens.

In this review, we summarize the conventional and recombinant vaccines used in veterinary medicine and the molecular approaches that have led to the development of new vaccines in recent years. We have focused on vaccines that target infectious diseases.

Conventional veterinary vaccines

Historically, the development of veterinary vaccines was based on empirical trial - and - error approaches that were designed to mimic, by vaccination, the immunity induced by natural infection (Doolan, Apte, & Proietti, 2014). The conventional ‘‘isolate, inactivate or kill and inject’’ approach can induce protection against a wide range of bacterial and viral pathogens. The majority of the licensed veterinary vaccines that are currently in use are inactivated (killed),live-attenuated vaccines or toxoids. In fact, the widespread use of these vaccines has contributed considerably to the improvement of animal and public health. However, conventional vaccines are generally expensive to produce, and need to be administered multiple times to induce optimal immunity (Delany, Rappuoli, & Gregorio, 2014; Meeusen, Walker,Peters, Pastoret, & Jungersen, 2007).

Additionally, the whole - organism approach to vaccination is almost exclusively restricted to pathogens that can be cultured in vitro. Although this process has been successful for a number of ‘‘simple’’ pathogens with relatively low antigen variability, it has not been effectively applied to vaccinate against pathogens that have high antigenic diversity or/and are capable of evading or misdirecting the host immune response (Doolan et al., 2014). Also, traditional vaccine design is based on a strategy that involves mimicking the immunity induced by natural exposure; however, in the case of many pathogens, this is sub optimal and robust sustained protection may require inducing an immunity that exceeds the natural biological immunity while also ensuring the adverse effects associated with stimulating the inflammatory response are minimized (Zepp, 2010). This is especially true for chronic infections, in which the pathogen is able to co-exist with the host for an indefinite period of time despite the presence of immune responses induced by the host and targeted against the pathogen (Doolan et al.,2014).

Live - attenuated modified vaccines are capable of inducing both humoral and cell-mediated immune responses. In contrast, inactivated vaccines offer improved safety profiles but cannot provide effective long-term protection.They may also cause adverse side effects due to undesirable components. Toxoids induce reliable humoral immunity,but little or no cell-mediated immunity (Moreira et al.,2016). The types and key features of conventional and next-generation approaches to the development of veterinary vaccines are presented in Table 1.

Live-attenuated veterinary vaccines

Live attenuated vaccines are created by passage of viruses or bacteria in an unnatural host or cell. After multiple passages of the virus or bacterial strain in various media, the strain is administered to the natural host in the hope that random mutation has delivered a non-virulent and replicative infectious agent (Meeusen et al., 2007). However, the strains that are present in most of the existing live attenuated bacterial vaccines are not highly protective. In addition,they have many drawbacks. For example, they cause local inflammation and other unwanted reactions and they can revert to virulence. Additional issues include the inability to effectively culture the bacteria or virus, the possibility of inducing an autoimmune response, and the need for refrigerated storage (Babiuk, Pontarollo, Babiuk, Loehr, & Van Drunen Littel - van den Hurk, 2003; Meeusen et al., 2007).As the live attenuated organism can still infect target cells,these vaccines can replicate and induce both cellular and humoral immunity and, generally, do not require an adjuvant to be effective.

The process of producing virus vaccines is very complex because it uses living cells; as such, it is difficult to achieve standardization. Live-attenuated vaccines are also challenging to formulate because of the macromolecular complexity of viruses and bacteria; viruses can be enveloped or non-enveloped. In comparison to inactivated vaccines, live-attenuated viruses are easier to produce, do not require the use of adjuvants in the formulation, and only require minimal downstream processing (van Gelder& Makoschey, 2012). While naturally occurring attenuated viruses or viruses obtained after passage in different animal species or cell cultures were used as vaccine strains in the early vaccines, today, targeted mutagenesis can be applied to generate vaccine virus strains.

The reverse vaccinology approach to vaccine design can create recombinant vaccines that are generally safer and more immunologically defined than the traditional live - attenuated vaccines (Delany et al., 2014). When molecular approaches are employed, the obtained deletions and mutation can be identified. The targets for these deletions are the genes that are responsible for important metabolic processes, but that allow the development of immune response. Therefore, this approach represents a viable strategy by which some of the drawbacks associated with live-attenuated vaccines can be overcome.


suyatno rindang

Table 1    Characteristics of vaccines currently available for veterinary use.
Type of vaccines	Characteristics
Live-attenuated	Live strains are not highly protective;
	Reversion to virulence to a more virulent phenotype can occurs;
	Need for refrigerated storage;
	Induce both cellular and humoral immunity.
Inactivated (killed)	Inactivated vaccines offer good safety profiles;
	Cannot provide effective long-term protection due to the destruction of the pathogen replication;
	Many inactivated vaccines are unable to cope with the prevailing strains in the field;
	Frequently, new vaccines have to be generated from field strains with new outbreaks.
Toxoids	The amount of toxin produced in vitro is unpredictable;
	High levels of biosafety are required.
Recombinant subunit	Well-defined composition;
	No risk for pathogenicity;
	Can be produced in a variety of protein expression systems;
	Possibility for cost-efficient production and purification;
	Primarily humoral immune response; Need of adjuvant.
RNA/DNA-based	Humoral and cellular immune responses (antigen presentation by both MHC class I and II molecules);
	Challenges in adequate cellular uptake and expression;
	Long-term persistence of immunogen;
	Risk of integration into host genome not completely excluded;
	Unstable and quite expensive production (for RNA vaccines).
Vectored-based	Induce both cellular and humoral immune responses;
	In vivo amplification systems available;
	Some vaccines are commercially available with a well-known safety record;
	Viral vectors allow for efficient infection of target cells.

Inactivated veterinary vaccines

Inactivated vaccines currently consist of bacterins of one or more bacterial species or serotypes, or killed viral strains formulated most often in an oil or aluminum hydroxide adjuvant (Meeusen et al., 2007). Inactivated vaccines are stable in field conditions and less expensive to produce than live vaccines. The vaccine virus is usually grown in cell culture,either in roller bottles or bioreactors. The inactivation of the vaccine virus for the production of killed vaccines is achieved by physical or chemical treatments that cause denaturation of the proteins or damage to the nucleic acids.The inactivated antigen may be further purified and mixed with an adjuvant (van Gelder & Makoschey, 2012).

Inactivated vaccines offer improved safety profiles but cannot provide effective long - term protection due to the destruction of the pathogen replication (Cho, Howard, &Lee, 2002). A large number of viral infections are caused by viruses that have multiple serotypes (e.g., bluetonguevirus and influenza viruses). As a consequence, many of the existing viral vaccines are often unable to cope with the prevailing strains in the field, and new vaccines have to be generated from field strains in response to new outbreaks(Meeusen et al., 2007).

Toxoids

Vaccination is the best preventive measure available to control the diseases caused by bacterial toxins. The vaccines that are currently commercially produced consist of inactivated native toxins (toxoids) combined with conventional adjuvants, which, although efficient, present some production limitations. For example, the amount of toxin produced in vitro is unpredictable, and some of the toxins are potent biological toxins that require high levels of biosafety (Arimitsu et al., 2004).

The use of recombinant vaccines can overcome these limitations, since they can be produced efficiently in large amounts and usually present low reactogenicity and toxicity. As such, they represent promising alternatives to the current commercial vaccines. For example, the production of recombinant Escherichia coli toxins takes only 2-3 days using simple growth media and formaldehyde for inactivation. This production method does not require many bio safety precautions because the toxic domain of the protein can be removed (Moreira et al., 2016).

Conventional subunit vaccines

Subunit vaccines usually contain part of the target pathogen and provoke an immune response against that component only. Polysaccharide vaccines are a type of sub unit vaccine that is composed of long chains of carbohydrate molecules that make up the surface capsule of the bacteria. The absence of additional antigenic components that are capable of stimulating T cells means that purified polysaccharides are incapable of recruiting sufficient T-helper activity to mount a protective immune response. This problem has been overcome through polysaccharide - protein - conjugate technology via which the polysaccharide antigen is covalently linked to a carrier protein, typically an inactivated toxin (toxoid), like the tetanus or diphtheria toxoids. By using a conjugate vaccine, the immune responses to the polysaccharides are dramatically improved (Dintzis, 1992).

VLP vaccines are virus - like particles that do not contain replicative genetic material but permit presentation of antigens in a repetitive, ordered array similar to the virion structure, which is thought to increase immunogenicity (Jennings & Bachmann, 2008). Their close resemblance to native viruses in terms of the molecular scaffolds and absence of genomes entail that VLPs can effectively elicit both humoral and cell-mediated immune responses without requiring an adjuvant. However, these approaches have yet to be employed in a commercial vaccine (Liu et al., 2012).

information :

• Pathogen genome sequencing
• Identification of genes
• Gene of interest cloning
• Recombinant construct
• DNA vaccines
• Subunit vaccines
• Vectored vaccines

Figure 1  Biotechnological approaches to vaccine development using recombinant DNA techniques. The gene encoding the antigen is isolated and either expressed and purified from a protein-production system, or is expressed directly by the vaccine recipient following injection of an engineered plasmid or a live vector. Prime-boost strategies combine different antigen delivery systems to broaden the immune response.

Biotechnology applied to next generation vaccine development

Genomic analyses of pathogens and enhanced understanding of the mechanisms of pathogenesis has resulted in new antigen discovery and the development of recombinant veterinary vaccines. A large amount of draft and wholegenome sequencing of viruses, prokaryotes, and eukaryotes pathogens has been performed (Kremer et al., 2016; Pizzaet al., 2000; Tettelin et al., 2000; Vasconcelos et al., 2005).These advancements have also improved antigen discovery and the characterization of variability between viral pathogens, which typically contain fewer than ten genes,and eukaryotic pathogens, which often encode >10.000 genes (Aurrecoechea et al., 2007; Cho et al., 2002). The genome sequencing technologies and the approaches used to screen the genome and proteome of a pathogen have greatly improved the efficiency of antigen discovery (Seib, Zhao, & Rappuoli, 2012) because relevant antigenic structures can identify and produce recombinant vaccines that contain only the antigen necessary to elicit protective immunity.

Genomic databases generally contain whole genome sequences and the complete repertoire of encoded proteins from which vaccine screening is possible (Bagnoli et al.,2011). Surface-exposed antigens, secreted proteins, and toxins are commonly viable vaccine candidates against bacterial infections (Ravipaty & Reilly, 2010). However, further in vivo investigation of antigens is still necessary and desirable. Comparative genomic analysis software can be used to perform gene comparative analysis by basic sequence similarity searches. Sequence similarity algorithms facilitate the comparison of predicted coding sequences (ORFs) with known genes/proteins in public databases, and are commonly used to predict the degree of gene conservation among a bacterial population.

In silico analysis may also result in enhanced protein antigen qualities such as expression and solubility. As native gene sequences retain their own specific codon usage that reflects the composition of their respective genomic tRNA pools, gene sequences may be optimized for higher expression levels in any heterologous system (Bagnoli et al., 2011).One drawback of reverse vaccinology is that it cannot be used to predict polysaccharides or lipids, which are often included in vaccines as active compounds. Fig. 1 has shown a scheme of recombinant vaccine development strategies.

Thus, the advances in genomics and other ‘‘omics’’ have given rise to a ‘‘third generation’’ of vaccines that are developed through the use of novel technologies such as reverse vaccinology (Dellagostin et al., 2011; Rappuoli, Pizza, Giu-dice, & Gregorio, 2014). This approach allows identification of a broader spectrum of vaccines candidates, including proteins that had not been identified and/or no abundant.In addition, enable the identification of potential targets without the need to grow pathogens in the laboratory.The reverse vaccinology has been resulted in veterinary vaccines that protect against an increased range of vaccine-preventable diseases. These next-generation vaccines can be multivalent, are highly purified, deliver an improved safety profile, and offer a viable alternative to the more reactogenic whole cell vaccines (Oliveira et al., 2015; Rap-puoli, 2001).

Recombinant subunit

Sub unit vaccines contain short, specific proteins of a pathogen that are noninfectious because they lack the ability to replicate in the host. Protective antigens allow recombinant vaccines to be administered as safe, non replicating vaccines. There is currently a large amount of scientific interest in the identification of immunogenic and protective antigens for animal pathogens.

Cloning the gene coding for the antigen is often necessary to better characterize and produce the identified antigen.E. coli has been used extensively as a host for heterologous protein expression; however, this approach has some limitations relating to the yield, folding, and posttranslational modifications of the recombinant protein (Heinson, Woelk,& Newell, 2015; Simionatto et al., 2010). An alternative host to E. coli is the methylotrophic yeast, Pichia pastoris. This yeast strain has emerged as a powerful and inexpensive expression system for the heterologous production of recombinant proteins that facilitates genetic modifications,allows the secretion of expressed proteins, permits post translational modifications, and produces a high yield (Ghosh& Nagar, 2014; Hartwig et al., 2010).

The expression of antigens in heterologous systems enhances the safety of both the manufacturer and the user by eliminating the need for the use of a virulent or partially virulent microbe to induce immunity. The additional benefits of subunit vaccines are that they incorporate proteins in their most native form, thereby facilitating correct protein folding and the reconstitution of conformational  epitopes (Eshghi, Cullen, Cowen, Zuerner, & Cameron, 2009). By incorporating more than one protein into a subunit vaccine,it is possible to invoke immunity to more than one strain or serotype of a bacteria or virus pathogen (Dellagostin et al.,2011). The potential drawbacks of subunit vaccines are they offer only a moderate level of immunogenicity and require adjuvants to generate robust immune responses.

Vectored vaccines

The use of antigen/gene delivery systems has facilitated the development of novel prophylactic and therapeutic vaccine candidates. Vector vaccine technology uses a vector to deliver protective protein(s) to the immune system of the vaccinated host. These vectors are usually immunogenic and can display multiple antigens. Recombinant vector vaccines are classified as live vector vaccines and naked DNA vaccines. Plant vaccines are also vector vaccines that have significant potential in veterinary medicine.

Classical live vectors are attenuated bacteria or viruses that, in addition to inducing their own natural immunity,can also be used as carriers to express the immunogenic antigens of other pathogens. Poxviruses, which include the vaccinia, fowlpox, and canarypox viruses, have been successfully used as vectors for exogenous genes. Poxviruses can accommodate large amounts of foreign genes and can infect mammalian cells, resulting in the expression of large quantities of encoded protein. Currently, the canarypox virus vector system has been used as a platform for a range of veterinary vaccines including those against WNV, canine distemper virus, feline leukemia virus, rabies virus, and equine influenza virus. The bacterial attenuated vector BCG has been studied for several years. Recombinant BCG offers significant potential to express a large number of heterologous antigens and can induce solid immunity (Rizzi et al., 2012).

The use of plants to produce and deliver immunogenic antigens via food sources is highly beneficial. The use of transgenic plants represents an innovative development that has opened new avenues in the vaccine industries. In veterinary vaccinology, transgenic plants can produce and deliver immunogenic antigens via animal feed (Shams, 2005).

DNA and RNA

DNA vaccines induce antigen production in the host itself.DNA (or RNA) vaccine can be defined as a plasmid that contains a viral, bacterial, or parasite gene that can be expressed in mammalian cells or a gene encoding a mammalian protein (noninfectious diseases). The gene of interest is inserted into a plasmid along with appropriate genetic elements such as strong eukaryotic promoters for transcriptional control, a polyadenylation signal sequence for stable and effective translation, and a bacterial origin of replication. The plasmid is transfected into host cells and transcribed into mRNA, which is subsequently translated, resulting in the host cellular machinery producing an antigenic protein. The host immune system recognizes the expressed proteins as foreign, and this can lead to the development of a cellular and humoral immune response.

Immunization of animals with naked DNA encoding protective viral antigens would, in many ways, represent an ideal procedure for viral vaccines because it not only overcomes the safety concerns associated with live vaccines and vector immunity but also promotes the induction of cytotoxic T cells after intracellular expression of the antigens (Meeusen et al., 2007).

Adjuvants for recombinant veterinary vaccines

The low immunogenicity frequently observed in pure recombinant antigens occurs due to a lack of exogenous immune-activating components such as nucleic acids, lipids, lipopolysaccharides (LPS), proteins, cell membrane components. Recombinant antigens can be offered in different adjuvants, and there is frequently a need to enhance the immunogenicity (except DNA vaccines). The addition of adjuvants to vaccine antigens delivers several advantages, such as dose sparing, increased efficacy in the elderly, and broadening of the cell or/and humoral immune response.

Subunit recombinants are typically better tolerated than inactivated or live attenuated pathogens; however, they are generally less immunogenic and require the addition of an adjuvant to achieve protective immune responses (Soema, Kompier, Amorij, & Kersten, 2015). The immunomodulatory effects are dependent upon the particular adjuvant used in conjunction with specific antigens.

Several adjuvants have been evaluated for use in veterinary vaccines, such as mineral salts (aluminum) (Li,Aldavel, & Cui, 2014); emulsions (Montanide) (Miles et al.,2005; Peter, Men, Pantaleo, Gander, & Corradin, 2001); biodegradable polymeric microparticles, and nanoparticles. In addition, an alternative range of adjuvants has been described as ‘‘immune potentiators’’ because they exert direct effects on immune cells, thereby leading to their activation (Ott, Radhakrishnan, Fang, & Hora, 2000). Examples of these include Toll-like receptor (TLR) agonists such as monophosphoryl lipid A (MPL) (Garc¸on, Wettendorff, & Van Mechelen, 2011); saponins, and bacterial exotoxins (Marchioro et al., 2013).

Some adjuvants act by sequestering antigens in physically restricted areas, known as depots, to provide an extended time period of antigenic stimulation. Thus, several veterinary vaccines are in the form of emulsions in oil. This relatively old-fashioned technology is, nonetheless, a powerful approach that achieves a strong inflammatory response and slow antigen liberation, exactly what recombinant subunit vaccines lack. In contrast to the strongly immune-activating emulsion-type adjuvants, aluminum salt adjuvants are not capable of inducing Th 1 or cell-mediated immune activation to any significant degree; however, they are efficient Th 2 inducers, giving rise to high antibody titers in the vaccinated individual.

Several groups have independently proposed the use of nano or microparticles to develop controlled-release vaccines. Depending on their size, particles are internalized by either phagocytosis or endocytosis. The antigens are either adsorbed on the surface of the nanoparticles or encapsulated inside the nanoparticle matrix (Slütter et al.,2009). Currently, polymeric microparticles have not yet been successfully developed as a vaccine product. Micro particles generally enhance the induction of Th2-type, humoral immunity, while nanoparticles promote Th1-based,cell-mediated immune responses (Li, Aldayel, & Cui, 2014).

Perspectives of the reverse vaccinology in animal health

The development of veterinary vaccines is a challenging task; however, reverse vaccinology is highly promising as a mechanism of veterinary vaccine development. Significant progress has been made in the field of vaccinology during the era of genomics, and next-generation vaccines are set to have an increasing impact on animal health. We can expect many more advances in vaccinology and the development of new effective veterinary vaccines that not only protect against infectious diseases but also against other diseases or chronic disorders. In fact, reverse vaccinology is now being applied to many bacterial, viral, and eukaryotic pathogens and, in all cases, has been successful in providing novel antigens for the design of new vaccines (Bagnoli et al., 2011; Buonaguro & Pulendran, 2011). More-over, the ability of rational design to improve candidate antigens can provide increased protection against antigenically variable pathogens (Seib et al., 2012). Table 2 shows the recombinant veterinary vaccines that are commercially available in 2017.

Genomics has catalyzed a shift in vaccine development toward sequence-based approaches, which use high-throughput in silico screening of the entire genome of a pathogen to identify genes that encode proteins with the attributes of immunogenic vaccine targets (Seibet al., 2012). The genes are expressed using foreign protein expression systems, including E. coli, yeast, and insect or mammalian cells, and are then purified and injected into a host to elicit immunity. In addition, the expression of recombinant proteins in plants could be a viable alternative to conventional expression systems and, therefore, they represent a versatile tool for the production of edible vaccines.

The low immunogenicity frequently observed can be overcome by the use of a specific adjuvant. Discovery and development of new adjuvants for recombinant targets is essential because purified protein antigens do not always induce the desired protective and sustained immune response against different target pathogens. For instance,there is an acute need to develop effective and safe vaccines against important veterinary diseases. As novel genome based technologies and new adjuvants continue to emerge, it is expected that new veterinary vaccines for important diseases will be within reach.

Novel effective veterinary vaccines are in high demand as a means of controlling new and re-emerging pathogens. A wide range of vaccine technologies has been applied to develop veterinary vaccines. Each approach has its inherent advantages and challenges. Almost all of the existing veterinary vaccines were developed using traditional vaccinology methods, which relied on screening a few candidates at a time based on the known features of the pathogen. However, over the last decade, there has been a significant acceleration in the advancement of biotechnological techniques and the ability to sequence a pathogen’s genome has provided vaccinologists with access to its entire antigenic repertoire. Such advances provide a great opportunity to create vaccines that are less dangerous but more effectively immunogenic than those developed by traditional methods.

Reverse vaccinology represents a promising approach to the discovery of recombinant vaccines against infectious,parasitic, and ever metabolic diseases. There is a distinct need to develop more potent, safer, better-characterized vaccines, in which different antigens can be combined, allowing for the development of vaccines against multiple strains of a pathogen. Recombinant vaccines fulfill this criterion and, as such, they are especially attractive for use as animal vaccines, for which vaccine cocktails are a useful vaccination option. The application of a biotechnological approach to the development of new effective veterinary vaccine candidates is fundamental and should be explored in depth.

Table 2      Recombinant veterinary vaccines available in 2017.
Animal species	Pathogens	Vaccine type
Cats	Feline leukemia virus	Vectored
Cats	Rabies virus	Vectored
Cattle	Ripcephalus (Boophilus)	Subunit
	microplus
Cattle	Ripcephalus (Boophilus)	Subunit
	microplus
Dogs	Canine distemper virus	Vectored
Ferrets	Canine distemper virus	Vectored
Fish	Infectious Hematopoietic	DNA
	Necrosis Virus
Horses	Influenza virus and	Vectored
	Tetanus toxin
Horses	Influenza virus	Vectored
Horses	West Nile virus	Vectored
Horses	West Nile virus	DNA
Poultry	Infectious	Vectored
	Laryngotracheitis virus
Poultry	Avian influenza virus	Vectored
Poultry	Marek’s disease virus	Vectored
Poultry	Newcastle disease virus	Vectored
Poultry	Mycoplasma	Vectored
	gallisepticum
Raccoons/coyotes	Rabies virus	Vectored
Sheep/goats	Echinococcus granulosus	Subunit
Swine	Classical swine fever	Vectored
	virus
Swine	Porcine circovirus	Subunit
Swine	Actinobacillus	Subunit
	pleuropneumoniae
Swine	Classical swine fever	Vectored
	virus
Swine	Porcinecircovirus	Subunit
Swine	Porcinecircovirus	Vectored

Conflicts of interest

The authors declare no conflicts of interest.

Acknowledgement

Financial support was provided by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq).

The development and early clinical testing of the ExPEC4V conjugate vaccine against uropathogenic Escherichia coli

The development and early clinical testing of the ExPEC4V conjugate vaccine against uropathogenic Escherichia coli

Objectives: In this ‘how it was done’ narrative review, we provide a description of, and context for, the early development of a conjugate vaccine targeting extra-intestinal, pathogenic Escherichia coli (ExPEC), from its creation in the laboratory to its testing in a large, first-in-human phase Ib trial.

Sources: We searched the Pubmed database for previous attempts to develop vaccines against ExPEC, and we provide data from laboratory and trial databases established during the development of ExPEC4V, the tetravalent conjugate vaccine candidate.

Content: Earlier attempts at ExPEC vaccines had mixed success: whole-cell or cell-lysate preparations have limited effectiveness, and though an early conjugate vaccine was immunogenic in animal models, its development stalled before extensive clinical testing could occur.

The development of the current conjugate vaccine candidate, ExPEC4V, began at a population level, with an epidemiological survey to determine the most common E. coli serotypes causing urinary tract infections (UTI) in Switzerland, Germany and the USA.

The O antigens of the four most prevalent serotypes were selected for inclusion in ExPEC4V. After its creation in the laboratory by means of an in vivo bioconjugation process engineered to occur within E. coli cells, ExPEC4V underwent toxicity and immunogenicity testing in animal models. 

It then underwent safety and immunogenicity testing in a first-in-human, phase Ib multicentre trial, whose population of healthy women with a history of recurrent UTI allowed for an additional, preliminary assessment of the candidate's clinical efficacy.

Implications: Laboratory development and early phase I testing were successful, as the vaccine candidate emerged with strong safety and immunogenicity profiles.

The clinical trial was ultimately underpowered to detect a significant reduction in vaccine-specific E. coli UTI, though it showed a significant decrease in the incidence of UTI caused by E. coli of any serotype. We discuss the findings, including the lessons learned.

A. Huttner, Clin Microbiol Infect 2018;▪:1

© 2018 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

Introduction

Escherichia coli and the need for a vaccine

Humans have never been without Escherichia coli. The divergence of the genera Escherichia and Salmonella occurred roughly 100 million years ago and coincided with the divergence of their hosts; the former is found in mammals and the latter in reptiles and birds [1]. Since then, Escherichia coli has evolved into one of the most diverse bacterial species: only about 20% of the genes found in a typical E. coli genome are shared by other E. coli strains [2], making some pathogenic to humans and many others beneficial [3]. This diversity is reflected in the roughly 180 serotypes of E. coli currently identified, and in turn reflects the species' ability to acquire novel genes encoded in plasmids and phage DNA through conjugation and transduction, respectively [4,5].

Some of these genes confer antibiotic resistance. Indeed, the clinical and economic burden of extra-intestinal pathogenic E. coli (ExPEC) is increasing in step with its resistance to antibiotics.

Escherichia coli is the leading cause of both bloodstream infections

and urinary tract infections (UTI) [6,7]. The proliferation of multidrug-resistant strains in recent years has further increased the incidence of E. coli-related hospitalizations, treatment failures and mortality [8]. The pipeline of antibiotics with novel mechanisms of action is severely limited; in addition, evidence is mounting that antibiotics' indiscriminate killing of ‘good’ bacteria is harmful to their hosts [9]. There is a growing need for alternative strategies to prevent and treat selected, pathogenic ExPEC strains causing UTI and abdominal infections with or without bacteraemia.

Although bacteriophage therapies and monoclonal antibodies are garnering increasing interest [10,11], these approaches remain early-stage and will not reach the clinical phase of development for some years. Vaccine development is further along and so more clinically relevant at this time. This narrative review chronicles the early development of the current leading ExPEC vaccine candidate, ‘ExPEC4V’, which was designed to prevent UTI and associated complications such as bacteraemia, from its creation in the laboratory to its phase Ib testing in humans.

Previous attempts at an extra-intestinal E. coli vaccine

Although there are several vaccine candidates for intestinal pathogenic E. coli strains, there are not many for extra-intestinal strains. One approach is whole-cell-based or cell-lysate-based formulations containing multiple E. coli strains; they have had only limited success. The best known is Uro-Vaxom®, capsules of which contain the crude lysates of 18 E. coli strains. Clinical studies dating from the 1980s reported significant reductions in UTI recurrences in women taking Uro-Vaxom® versus placebo. These studies have been questioned, however, for their methodology and reporting: among other criticisms, they were not randomized and it is unclear whether and how concomitant antibiotic therapywas taken [12,13].

Similar formulations include Strovac®, Urvakol® and Urostim®; all of these ‘uropathogen cocktails’ are of limited effectiveness (see

Brumbaugh and Mobley for a recent review [14]).

Other candidate vaccines have specifically targeted the O antigen of the E. coli lipopolysaccharide: the O antigen is the major surface component of outer membrane-encircled Gram-negative bacteria and is therefore highly exposed to the host immune system.

Most of these vaccine candidates have proven unsuccessful, however. The O antigen is generally aweak immunogen [15], which is not surprising given the organism's long co-existence with humans, and its associated lipid-A portion activates Toll-like receptor 4, thereby inducing significant local and systemic inflammatory reactions [16].

Detoxification through the removal of lipid A successfully reduces reactogenicity, but it also further reduces immunogenicity. A protein-conjugated vaccine targeting the O antigen was therefore attempted in the early 1990s; this candidate used the well-known, highly immunogenic Pseudomonas aeruginosa exoprotein A as its carrier protein and targeted the O18 polysaccharide [17].

The candidate proved immunogenic and efficacious in mice, protecting them against lethal O18 challenge.

Subsequently, Cross et al. tested a conjugate vaccine containing purified polysaccharide from 12 O serotypes chemically conjugated to P. aeruginosa exoprotein A in healthy human volunteers; vaccine immunogenicity was variable, with different serotypes showing different immunogenic potential [18].

Since then, the progress of conjugate E. coli vaccines has stalled: the process of chemically combining multiple individually conjugated serotypes in a single vaccine vial is technically challenging and costly [15,19].

The laboratory: creation and pre-clinical development of ExPEC4V

Selecting the target

As described above, E. coli's O antigen is highly exposed to the host immune system, making it an attractive vaccine target. The production of anti-O antigen antibodies has been shown to confer protective effects against E. coli infections. Moreover, the high clinical efficacy of multivalent glycol-conjugate vaccines against infections further supports the use of surface O antigen polysaccharides in conjugate vaccine development [20e23].

Additionally, animal studies indicate that IgG targeting the O antigen are the likely correlate of protection against extra-intestinal E. coli; IgG passively transferred to septic mice challenged with lethal doses of E. coli O18 provided a significant degree of protection [17].

Because the number of serotypes that can be included in a single vaccine is limited, their correct selection is crucial. Despite the rich variety of O antigens on E. coli surfaces, only a limited subset of serotypes have been observed to manifest extra-intestinal pathogenicity, and even fewer have shown uropathogenicity [24].

Although many studies have evaluated the occurrence of virulence factors in clinical isolates from individuals with UTI, only a few studies have comprehensively analysed the O serogroups of E. coli associated with these infections [7].

Thus, as a first step in the identification of appropriate vaccine targets, more than 2000 E. coli urinary isolates were collected in community and hospital settings, primarily in Switzerland but also in Germany and the USA, to identify the most prevalent E. coli serogroups causing UTI in women, including those carrying antibiotic resistance genes. There was remarkable consistency in prevalence: the top 12 E. coli serogroups associated with UTI remained stable across geographic locations, seasons and symptom manifestations (unpublished). Serotype O25, which is strongly associated with multidrug resistance, was particularly prevalent, with a consistent ranking in the top four.

Though the conjugated ExPEC4V vaccine candidate could hold as many as 12 O antigens in order to have broad coverage against the most infections caused by pathogenic E. coli, only four serotypes were included in this first-generation vaccine based on prevalence and antibiotic resistance. The four most common serotypes, O1A, O2, O6A and O25B, with a total coverage of around 30%e35% of UTI caused by E. coli, were selected for the ExPEC4V model, which underwent toxicity testing in animals and first-in-human, phase I testing.

Determining the vaccine type: a bioconjugate candidate

Pure polysaccharide vaccines had proved weakly immunogenic and chemically (in vitro) conjugated vaccines were technically challenging. The bioconjugation technology developed at Glyco-Vaxyn (now LimmaTech) was therefore considered a suitable technology for the production of such a multivalent vaccine [15].

Bioconjugation refers to the biosynthesis of polysaccharide and carrier protein within E. coli cells, with their subsequent in vivo coupling by means of the oligosaccharyltransferase PglB from the N-linked protein glycosylation system, originally identified in Campylobacter jejuni and subsequently transferred to E. coli [19,25].

The technology allows the in vivo conjugation of an antigenic polysaccharide, in this case pathogenic O antigens from E. coli, to a protein of choice. Serotypes O1A, O2, O6A and O25B were conjugated in vivo to the genetically detoxified P. aeruginosa exoprotein A; this proteinwas chosen because of previous experience with it as a protein carrier [17,26].

Pre-clinical testing of the candidate

The resultant tetravalent candidate vaccine was tested, along with a monovalent O6 precursor, in various non-clinical studies to establish the biological properties of the material and to evaluate potential toxicities (Table 1). The information obtained also contributed to the identification of the optimal composition and formulation process. The monovalent and tetravalent formulations of the vaccine were tested in different animal species (mice, rats, rabbits) to identify effects on IgG levels and antibody functionality, the latter through the opsonophagocytic killing assay. These immunological parameters have been shown to predict clinical efficacy for licensed vaccines preventing similar infections [22,23].

The usefulness of an aluminium hydroxide adjuvant was also explored; no consistent additional benefit to the immune response was observed [27]. Promising results were obtained showing a strong IgG immune response and robust antibody functionality in in vitro assays, and the first clinical trial was planned.

The clinic: how ExPEC4V was tested for safety,

immunogenicity and efficacy

Outcomes and trial population

A phase I, first-in-human trial must have safety as its primary outcome and must be composed of a population of healthy volunteers who, it is assumed, will be better prepared to handle unexpected toxicity. Phase I vaccine trials assess immunogenicity as a close secondary outcome; clinical efficacy can rarely be evaluated in people who must be healthy at baseline for study entry.

In the case of the ExPEC4V candidate, however, there was indeed a healthy population in whom clinical efficacy could be assessed. Adult womenwho were healthy at baseline but who had a recent history of recurrent UTI due to E. coli were chosen as the trial population; they could then be evaluated for incidence of UTI after vaccination [28]. Thus women who had had either two UTI in the preceding 6 months or three UTI in the preceding year were eligible for the study. To increase the probability that we were targeting the relevant population, we further required documentation of at least one urine culture with growth of E. coli in the preceding 5 years.

Though the trial's primary outcome was the incidence of adverse events, it was powered to assess efficacy trends so as to provide a clinical proof of concept. The sample size was calculated to detect a 64% decrease in the incidence of vaccine-serotypespecific UTI in the treatment group: assuming a vaccine-serotype UTI incidence of 25% in the placebo group and at 80% power, just under 100 women in each group would be necessary. Thirteen clinical sites throughout Switzerland, most of them within major hospitals but some consisting of outpatient clinics, were recruited for the study.

The initial plan for volunteer recruitment relied heavily on physicians in clinical practice to disseminate information on the

trial to their patients. The plan proved insufficient, with only sluggish recruitment in the first months after trial launch; physicians and their teams had limited time and were not always in

regular contact with those patients who might be eligible. When advertisements were placed online and in the local newspapers of participating cities, recruitment increased dramatically, allowing original targets and timelines to be met.

Determining the trial design and length of follow up

Safety concerns in this first-in-human trial had to be balanced with methodological concerns. Therewas universal agreement that the trial should be randomized and placebo-controlled, yet blinding was another issue. Although some investigators were in favour of classic double blinding, this design was not strongly recommended in an early Scientific Advice Meeting between the study sponsor (Glycovaxyn, Schlieren, Switzerland) and The Swiss

Agency for Therapeutic Products (Swissmedic). Ultimately, the trial was conducted in single-blind fashion, with trial physicians unblinded as an additional safety measure.

Another safety measure was a staggered rollout, with the first eight participants randomized 1: 1 to either reduced-dose vaccine or placebo. Reduced-dose vaccine contained only 1 mg of each surface polysaccharide, while target-dose vaccine contained 4 mg of each. In the absence of safety signals in this preliminary group, a subsequent group of eight participants was randomized, this time to target-dose vaccine or placebo. After another observation period, all remaining participants were randomized to target-dose ExPEC4V or placebo.

Most phase I vaccine trials do not last more than 6 months; safety issues are typically apparent early on, and durability of the immune response can be evaluated in later-phase studies. In this trial, participants were followed for 9 months post-treatment; the extra months of observation primarily targeted clinical efficacy, though safety and immunogenicity at these time points were also evaluated.

The findingsdincluding lessons learned

The vaccine candidate emerged with a strong safety profile; there were no related severe or serious adverse events. Immunogenicity results were also promising: vaccination induced significant IgG responses for all four included serotypes, higher by fourfold to nine-fold on day 30 compared with baseline, and opsonophagocytic killing activity showed antibody functionality [28].

Clinical outcomes were surprising. First, the prediction of a baseline UTI incidence of 25% among placebo recipients in the 9-month follow-up period proved to be an overestimate. In both arms there were too few UTI in general, and certainly too few caused by vaccine-serotype E. coli, to show a difference in the incidence of vaccine-serotype E. coli UTI; in the placebo group, only 11% had a UTI caused by a vaccine-specific serotype. A significant reduction, however, in UTI caused by any E. coli serotype was seen: vaccinees had nearly a third fewer of these as placebo recipients. Antibody cross-reactivity among serotypes may exist, but needs additional confirmation. In post hoc analyses assessing UTI with high ( 105 CFU/mL) bacterial counts, the difference in UTI reduction grew wider [28]. Infections with higher bacterial loads might require a stronger or different immune response: it may be that a parenteral E. coli conjugate vaccine will be useful in preventing more severe cases of UTI and more invasive E. coli infections, such as those in the bloodstream, while conferring less protection against milder colonization and less invasive infections. A parallel example can be seen with parenteral pneumococcal conjugate vaccines, which had only intermediate efficacy against otitis mediadlike cystitis, also a mucosal infectiondbut high efficacy against invasive pneumococcal disease [29]. More data are needed to confirm this hypothesis, however.

Women do not experience UTI at regularly spaced intervals; recurrences tend to cluster, sometimes after long, symptom-free interludes [30,31]. During the trial's follow-up period, we were well aware of the increasing likelihood that 9 months could prove to be too short an exposure time for the determination of vaccine efficacy. Yet no interim analysis for efficacy end points had been laid out in the trial protocol: good clinical practice required that the database remain locked until the last volunteer's last follow-up visit. An extension of the follow-up period was discussed, but its importance could not be determined until identification and serotyping of the urinary isolates could occur. By the time the data could be analysed, volunteers had dispersed and dedicated site teams had been disassembled. The trial itself was effectively over.

Preparation for, and ethics and regulatory approvals of, a multicentre follow-up study could take another year; during that time of presumed heightened immunity in the vaccine group, any additional UTI events in all trial participants would be missed. It was ultimately decided to let later-phase trials with larger populations and longer follow up assess efficacy outcomes more definitively; the safety and immunogenicity results provided by this phase Ib trial clearly supported moving the product to advanced development.

Current and future work

During the preparation of the trial, Glycovaxyn exclusively licensed the ExPEC4V bioconjugate vaccine, nowknown as JNJ-1860, to Janssen Pharmaceuticals, Inc. (Johnson and Johnson, New Brunswick, NJ, USA), which is continuing its clinical development. A dosefinding phase I trial for safety and immunogenicity has recently been completed in Japan (NCT02748967), and a phase II trial including some 850 volunteers is underway in the USA (NCT02546960), with preliminary results showing strong immunogenicity and an acceptable safety profile at the dose of 4 mg each of O1A, O2 and O6A antigens and 8 mg of O25B antigen (Table 1) [32].

Transparency declaration

VG is an employee of LimmaTech Biologics AG, a spin-out entity from GlycoVaxyn, which created the ExPEC4V vaccine and sponsored the pre-clinical and phase Ib studies. AH declares no conflicts of interest.

Acknowledgements

The authors wish to thank the trial volunteers once again, as well as all investigators involved in the laboratory and clinical development of the x4V vaccine.

Funding

No funding was received for the writing of this manuscript

Author contributions

AH and VG co-designed and co-wrote the manuscript.

Quantitative structure-activity relationship for estrogenic flavonoidsfrom Psoralea corylifolia

Quantitative structure-activity relationship for estrogenic flavonoidsfrom Psoralea corylifolia

a b s t r a c t

A combination of in vitro and in silico approaches was employed to investigate the estrogenic activities of flavonoid compounds from Psoralea corylifolia. In order to develop fluorescence polarization (FP) assay for flavonoids, a soluble recombinant protein human estrogen receptor ligand binding domain(hER-LBD) was produced in Escherichia coli strain. The competition binding experiment was performedby using coumestrol (CS) as a tracer. The result of FP assay suggested that the tested flavonoids can bindto hER-LBD as affinity ligands, except for corylin. Then, molecular modeling was conducted to explorethe binding modes between hER-LBD and flavonoids. All the tested compounds fit into the hydrophobicbinding pocket of hER-LBD. The hydrophobic and hydrogen-bonding interactions are dominant forcesto stabilize the flavonoids-hER-LBD binding. It can be speculated from molecular docking study that thehydroxyl groups and prenyl group are essential for flavonoid compounds to possess estrogenic activities. Both methylation of hydroxyl group and cyclization of prenyl group significantly diminish the estrogenicpotency of flavonoids. Furthermore, quantitative structure-activity relationship (QSAR) analysis was performed by the calculated binding energies of flavonoids coupled with their determined binding affinities.Comparison between the docking scores and the pIC50values yields an R-squared value of 0.9722, indicating that the estrogenic potency of flavonoids is structure-dependent. In conclusion, molecular docking canpotentially be applied for predicting the receptor-binding properties of undescribed compounds basedon their molecular structure.

1. IntroductionAs an erect annual herb, Psoralea corylifolia (Leguminosae)has been used in traditional practices of Ayurvedic and Chinesemedicine [1–5]. It is widely distributed and considered as a naturalalternative remedy due to its diverse beneficial effects, including hepatoprotective, estrogenic, antidepressant, antimicrobial,antioxidant, and antitumor activities [6,7]. The flavonoid com-pounds derived from the fruits of Psoralea corylifolia can be dividedinto isoflavones, flavanones and chalcones [8]. Multiple biologicalactivities of these components have been confirmed, demonstrating their potential for treating diseases [6,9]. As phytoestrogens,flavonoids share similar structure with endogenous estrogens(such as 17 -estradiol). They can interfere with endocrine regulations in the human body through binding to estrogen receptors(ERs) [10–14]. Estrogen receptors are the transcriptional factors playing important roles by binding and activating estrogen response elements(EREs) on target genes, subsequently controlling cell proliferation and survival in normal mammary tissue [15]. They belong tothe nuclear receptor (NR) superfamily [16] and participate in theregulation of reproduction, development, metabolism, and homeostasis [17,18]. Two isoforms of estrogen receptors (ER and ER )have been identified and share the similar crystallographic structure [19–22]. Current available endocrine therapies for ER-positivebreast cancers mainly focus on the selective estrogen receptormodulators (SERMs) [23]. They exert dual agonistic or antagonistic effect on ER transcription and have been applied for treatinghormone responsive breast cancers for decades [24,25]. Estrogen mimetics including both natural and synthetic chemicals have been reported to selectively activate ERs [26–29].Phytoestrogens, a group of plant-derived compounds with estrogenic properties [30], can structurally or functionally mimicmammalian estrogens [31,32]. They are considered as a naturalsource of SERMs eliminating the side-effects of hormone replacement therapy. Furthermore, they are also known to exert widelybenefits to human health, especially against cancer, osteoporosis,irregular menopause syndrome, cardiovascular disease, etc. [33,34]. Genistein is a typical phytoestrogen isolated from soybeans andbelongs to isoflavonoids [35], a class of secondary metabolites thatmainly occur in Leguminosae [36]. It has been confirmed to bindto the human estrogen receptors and disrupt normal estrogenicsignaling. However, the estrogenic potential of many other naturally occurring flavonoid compounds and the underlying molecularmechanism of their pharmacological activities are still unclear.Hence, the present work focuses on the estrogenicity of flavonoidsisolated from the fruits of Psoralea corylifolia.A combination of in vitro and in silico approaches was employedto investigate the estrogenic activities of flavonoid compounds,including two isoflavones corylin and neobavaisoflavone, three flavanones bavachin, isobavachin and bavachinin, three chalconesbavachalcone, isobavachalcone, and 4 -O-methylbavachalcone. Inorder to develop fluorescence polarization assay for flavonoid compounds, a soluble recombinant protein human estrogen receptor ligand binding domain (hER-LBD) was produced first. The competition binding experiment was performed by using coumestrol (CS)as a tracer. Based on the determined binding affinities of hER-LBDwith flavonoids, molecular docking was conducted to explore theirbinding modes, in an attempt to establish a quantitative structure-activity relationship (QSAR) model for evaluating and predictingthe estrogenic potential of flavonoid compounds.

2. Materials and methods

2.1. Materials and chemicals

Isopropyl -D-1-thiogalactopyranoside (IPTG), dimethylsulfoxide (DMSO), and coumestrol (CS) were purchased fromSigma-Aldrich (St. Louis, MO, USA) and TCI (Tokyo, Japan). Corylin Corylin(≥98%), neobavaisoflavone (≥98%), bavachin (≥98%), isobavachin(≥98%), bavachinin (≥99%), bavachalcone (≥98%), isobavachalcone(≥98%), and 4 -O-methylbavachalcone (≥98%) were purchasedfrom Yuanye Biotechnology Co., Ltd. (Shanghai, China). The structures of these flavonoid compounds are shown in Fig. 1. All otherreagents used were of analytical grade.


2.2. Expression and purification of hER˛-LBD

The coding sequences of human estrogen receptor ligandbinding domain (hER-LBD) and glutathione S-transferase (GST)were inserted into the pGEX-4T-1 vector at restriction sites BamHIand XhoI. The expression plasmid pGEX-4T-1-hER-LBD was introduced into Escherichia coli strain BL21(DE3)pLysS. Cells weretreated with 0.5 mM IPTG overnight at 20◦C to induce the expression of hER-LBD. A 0.22 m membrane filter (Millipore, Bedford,MA, USA) was used to remove all the bacterial cells from suspension. Afterward, the supernatant was loaded onto an IDA-Ni2+column (Novagen, Madison, WI, USA) to purify the target protein.


2.3. Fluorescence polarization assay

In this work, an autofluorescent exogenous estrogen coumestrol(CS) was employed as a probe. The protein hER-LBD (250 nM)and the probe (10 nM) were mixed in a total volume of 290 Land titrated with various concentrations of flavonoids (10 L). Themicroplate was subjected to FlexStation 3 (Molecular Devices, Sunnyvale, CA, USA) after being incubated at room temperature for 2 h.The excitation and emission wavelengths were 355 and 405 nm,respectively. The IC50value (concentration of flavonoid for 50%inhibition of binding between CS and hER -LBD) was calculatedaccording to a four parameter logistic equation Y = (A − D) / [1 + (X/ IC50)B] + D, where Y and X correspond to the polarization value andthe tanshinone concentration, A and D are the polarization valuesat zero and an infinite concentration respectively, and B is the slope parameter. Data analysis was performed using GraphPad Prism 5(GraphPad Software, USA).


2.4. Molecular docking

The crystal structure of hER-LBD complexed with diethylstilbestrol (DES) was obtained from the Protein Data Bank (PDBID: 3ERD) [37]. The initial structures of flavonoid compoundswere constructed with GaussView 5.0.9 and optimized using theB3LYP/6-31G(d) method with Gaussian 09W. The molecular lengthand Connolly solvent-excluded volume (CSEV) of flavonoids werecalculated by using AutoDockTools-1.5.6 and Chem3D Ultra 8.0,respectively. Then automated docking with grid-based energy evaluation was carried out by AutoDockTools-1.5.6 to explore theflavonoids-hER-LBD binding modes. The center of the grid boxwas identified by the location of the synthetic nonsteroidal agonist DES with its size adjusted so as to enclose all the key residues.We built the box around hER-LBD with 38 points cube coverage and set a spacing of 0.375 Å between the grid points. Thetested compounds were docked into rigid receptor structure usinga Lamarckian genetic algorithm (GA) provided by AutoDockTools-1.5.6. All other docking parameters, such as number of GA runs(10), population size (150), maximum number of evals (medium),maximum number of generations (27,000), rate of gene mutation(0.02), and rate of crossover (0.8), were set to defaults. The predicted binding energies (kcal mol−1) were calculated based on thescoring function. For each flavonoid, 10 independent docking runswere conducted and the first-ranked conformation with the lowest binding energy was chosen to provide insights into its bindingmodes toward hER-LBD. The intermolecular interactions werevisualized by using the program PyMol.



3. Results and discussion

3.1. Determination of binding potency between hER˛-LBD andflavonoids
For competitive fluorescence polarization assay, a soluble protein hER-LBD and an exogenous estrogen coumestrol (CS) wereemployed as recognition element and fluorescent tracer, respectively. The recombinant protein was expressed in Escherichia coliand purified with immobilized metal affinity chromatography(IMAC). At the beginning of the determination, the receptor andthe tracer form a CS-hER-LBD complex. With a large molecular volume, CS-hER -LBD rotates slowly and produces a high FPvalue. Then, if the added flavonoid can compete for the bindingsite of receptor, CS is displaced from the complex. The unboundtracer molecule, with decreasing size, rotates remarkably fast andcauses a low FP value. Therefore, the FP signal can be monitored todifferentiate the bound and unbound tracer.
As can be observed in Fig. 2, all the tested flavonoids (exceptfor corylin) exhibited dose-dependent binding to hER-LBD andtheir IC50values obtained from the competition curves werelisted in Table 1. The flavonoid compounds from Psoralea corylifolia demonstrate distinct binding potency toward hER-LBDwith IC50values ranging from 5.6 nM to 949.6 nM, except forcorylin. Their binding affinities with hER -LBD is in the order ofneobavaisoflavone > isobavachin > bavachalcone > isobavachalcone > 4'-O-methylbavachalcone > bavachin > bavachinin > corylin. For theisoflavones derivatives investigated herein, neobavaisoflavonewith a prenyl group exhibits remarkably greater estrogenicpotency than corylin, whose prenyl group is cyclized. It has beenreported that the presence of prenyl group appears to be crucialfor estrogenic effects [38], which is confirmed in this work. In summary, the flavonoid compounds from Psoralea corylifolia canbind to hER-LBD as phytoestrogens.

3.2. Structural basis for estrogenicity of flavonoids

It is generally assumed that phytoestrogens and xenoestrogens exert their stimulatory effects on ERs through binding atthe same site as that occupied by endogenous estrogens such as17-estradiol [39,40]. Previous studies have revealed the crystal structures of hER-LBD complexed with several flavonoid-likecompounds, such as genistein (PDB ID: 1 × 7R and 2QA8) andbiochanin A (PDB ID: 5JMM) [41,42]. However, the protein structures co-crystallized with the flavonoid compounds from Psoraleacorylifolia have not been reported yet. In this work, moleculardocking was conducted to explore the binding modes betweenhER -LBD and flavonoids.
The hydrophobic binding pocket of hER -LBD possesses a probeaccessible volume of 450 Å3, nearly twice as large compared todiethylstilbestrol (DES), a synthetic nonsteroidal agonist ligand[43]. The molecular length and Connolly solvent-excluded volume(CSEV) of flavonoid compounds were calculated and compared tothat of DES (Table 2). The sizes of flavonoids are close to that ofDES, making the hydrophobic pocket large enough to accommodate these compounds. As shown in Fig. 3, flavonoids completely fitinto the cavity without disruption of the coactivator binding site,known as a transcriptional activation function 2 (AF-2). Furthermore, it can be observed that all these flavonoids fit into the bindingsite similar to that of DES. As the core helix of AF-2, H12 is stabilized by flavonoids to pack against H3 and H11, constraining theirmovement (Fig. 4). As a result, the receptor seals flavonoids in thecavity and forms a hydrophobic groove to facilitate co-activator(CoA) recruitment.

Molecular docking results suggest that hydrophobic andhydrogen-bonding interactions are the dominant forces to stabilize the flavonoids-hER -LBD binding. As shown in Fig. 4, withregard to isoflavones, neobavaisoflavone makes significant morehydrogen-bonding interactions than that of corylin. The two phenolic hydroxyl groups of neobavaisoflavone form hydrogen bondswith polar residues located at the two ends of the cavity, whichsignificantly help to improve the stability of neobavaisoflavone-hER-LBD binding. However, only a hydrogen bond can beobserved between corylin and hER-LBD, which may be attributedto the cyclization between hydroxyl and prenyl group. For three fla-vanones investigated in this work, both bavachin and isobavachinform hydrogen bonds at two active sites, namely H3 and H11.Unfortunately, the phenolic hydroxyl group at A-ring of bavachininis methylated, resulting in the loss of hydrogen-bonding interaction with Gly521 in H11. Similarly, for the tested chalcones,4 -O-methylbavachalcone loses the hydrogen bond toward Gly521(H11) compared to bavachalcone and isobavachalcone, due to themethylation of one of its phenolic hydroxyl group at A-ring. Hence,it can be speculated that the hydroxyl groups and prenyl group are essential for flavonoid compounds to possess estrogenic activities. Furthermore, as shown in Fig. 4 and Table 2, all the testedcompounds form a hydrogen bond with a water molecule, exceptfor corylin. Loss of hydrogen bonds with the water molecule andthe key residues in H3 greatly diminish the estrogenic potencyof corylin, which is consistent with aforementioned result in thefluorescence polarization assay.


3.3. Quantitative structure-activity relationship (QSAR) model

For each docked ligand, the binding energy (score) were calculated by AutoDock. Among these flavonoids, neobavaisoflavoneseems to be the most potent ER ligand with maximum binding energy (−9.92 kcal mol−1), as shown in Table 1. Due tothe cyclization between hydroxyl and prenyl group, corylindisplays a minimum binding energy (−3.53 kcal mol−1).The order of the calculated binding energies is as follow:neobavaisoflavone > isobavachin > bavachalcone > isobavachalcone> 4 -O-methylbavachalcone > bavachin > bavachinin > corylin.Interestingly, the predicted binding potency for flavonoids withhER -LBD is in agreement with their experimentally determinedbinding affinities. As shown in Fig. 5, comparison of the docking scores with the pIC50values, namely -log10(IC50) values[44], yields an R-squared value of 0.9722, indicating that theestrogenic potency of flavonoids is structure-dependent. Thena quantitative structure-activity relationship (QSAR) model was established for evaluating and predicting the estrogenic potentialof flavonoid compounds. Based on the structures of undescribedcompounds, molecular docking may be helpful for predicting theirreceptor-binding properties.



4. Conclusion

The present work aims to investigate the estrogenic activitiesof flavonoid compounds from Psoralea corylifolia by a combinationof fluorescence polarization and molecular docking approaches.Both in vitro and in silico studies indicate that the tested flavonoidcompounds from Psoralea corylifolia can bind to hER-LBD asaffinity ligands, except for corylin. The hydrophobic and hydrogen-bonding interactions are the dominant forces to stabilize theflavonoids-hER-LBD binding. Structure-activity relationship analysis of estrogenic flavonoids suggests that both methylation ofhydroxyl group and cyclization of prenyl group significantly diminish their estrogenic potency. Therefore, it can be speculated thatthe hydroxyl groups and prenyl group are essential for flavonoidcompounds to possess estrogenic activities. Additionally, the correlation analysis between the docking scores and the pIC50valuesindicates that the estrogenic potency of flavonoids is structure-dependent. This work may be helpful for in silico screeningof selective estrogen receptor modulators (SERMs) from naturalbioactive compounds based on their molecular structures.


Acknowledgements

This work was supported by the National Natural ScienceFoundation of China (31871717, 31601534 and 31701349), theNational Key Research and Development Program of China(2018YFD0300200 and 2016YFD0300103), the China PostdoctoralScience Foundation (2018T110249 and 2017M621213), and theScience and Technology Development Project Foundation of JilinProvince (20180520102JH, 20180201062SF, and 20150519010JH).