Introduction

Foot-and-mouth disease (FMD) has long been known to be one of the most contagious and devastating diseases affecting cattle and other bi-ungulate species distinguished by fever and blister-like sores on the tongue and lips, in the mouth, on the teats, and between the hooves, as well as significant production losses, weakening, and debilitation of cloven-hoofed animals such as cattle, pigs, sheep, and goats, as well as more than 70 species of wild animals1. The etiological agent, Foot-and-mouth disease virus (FMDV), has a positive sense single stranded RNA genome encapsidated by sixty copies each of the four structural proteins of which VP1, VP2, VP3 and VP4 which are encoded by the N-terminal half of the ORF2. Nonstructural proteins include Lpro, 2A, 2B, 2C, 3A, 3B, 3C pro, and 3D pol which occupies about two-thirds of the ORF. Despite seven distinct serotypes named O, A, C, SAT 1, SAT 2, SAT 3 and Asia 13, the outbreaks experienced in Asia are solely from serotypes O, A and Asia 1 in recent years (FAO, 2016)4. Though influx of cattle from North Africa to the Middle-East during the season of sacrifice increases, risks introducing serotypes SAT1, SAT2, SAT3 in Asia, but till today serotype O is the most predominant agent behind the outbreaks in Bangladesh, although A and Asia1 are also found in circulation through demographic analysis. In the 53.89% of positive cases foot-and-mouth disease (locally known as “Khura rog”5), the fatality rates were higher (71.46%) in young (< 2 years age) calves, compared to the adult (2.27%)6. These lead to huge economic losses since about 70% of the human population is directly or indirectly dependent on the livestock sector with an estimated 52.8 million livestock, mostly food producing cattle and goats7. However, despite having the significant livestock population, the current production of meat and milk is inadequate to meet requirements and FMD largely contributes to the deficits causing estimated annual losses to be about US $125 million8.

Being a WOAH-listed disease and the first disease for which the World Organisation for Animal Health (WOAH, founded as OIE) established official status recognition, FMD outbreaks must be reported to the Organisation, as indicated in the Terrestrial Animal Health Code. According to the Department of Livestock Services (DLS) of Bangladesh, there were 25 FMD outbreaks reported in the country in 2020. The most affected regions were Rajshahi, Rangpur, and Khulna divisions9. The country must comply and implement initial measures described in the Global Foot- and- Mouth disease control strategy such as the presence of early detection and warning systems and the implementation of effective surveillance in accordance with the guidelines detailed in the Terrestrial Code.

With 548,817 cases recorded in cattle and buffalo between 2014 and 2017, Foot-and-Mouth Disease (FMD) is extremely endemic in Bangladesh. Higher rates are seen in eastern regions and during the post-monsoon season, highlighting the importance of targeted vaccination with minimum 80% coverage and continuous surveillance7.Another most important aspect required is precise, rapid confirmatory diagnosis so that the separation and treatment strategies can be brought to play. Basing on clinical signs only FMD cannot be differentiated from other vesicular diseases, such as swine vesicular disease, vesicular stomatitis and vesicular exanthema. Considering the endemicity of FMD, overlapping symptoms, large number of affected livestock and their economic impact, the availability and robustness of the serological surveillance and rapid diagnosis method for detection of viral infection at different stages can eliminate the shortcomings of disease control and prevention. The lack of low-cost standardized methods necessitates the development of a precise, disease-specific comprehensive FMD diagnosis kits10. Moreover, severe outbreaks that occur in Bangladesh every year are partly caused by immunogenically varied strains of FMDV that lack cross-protective immunity8,9,10,11,12. Since monovalent inactivated vaccinations are frequently used, it is important to identify the precise serotype causing the epidemic at a given moment. Serotype-specific diagnosis notifies us of the influx of any new serotype that might pose a concern in the future if early precautions are not taken. Thus, a time-sensitive serotype-specific diagnosis is required to confront FMDV cases. There is no need to underestimate the need of differentiating between vaccinated and infected sample since without doing so, the diagnosis would lose its significance and become invalid. Therefore, the presence of continuous circulation of more than one serotypes of FMDV and the risk of emergence, reemergence or infiltration of new serotype accentuates the need for successful diagnosis of FMDV not only for diagnosing serotype-specific seroprevalence but also for differentiating infected from vaccinated animals in a serotype dependent manner.

With this view, the project aims at developing novel indirect ELISA based prototype of diagnostic kit which will accommodate serotype specific VP1 antigen of circulating strains, serotype independent VP2 antigen to detect emergence of new serotypes and non-structural 3AB antigen to differentiate infected from vaccinated trait. This strategy using unique candidates will serve as multi-check diagnostic purpose as well as epidemiological study tool with post vaccination surveillance, which has never been reported or used before. A prototype model has been proposed in this article to simulate this system for the development of a single-use rapid dot-blot kit for detection of FMDV-specific antibody reflecting immune status of the animal. All this will ensure implementation of an effective time saving epidemiological surveillance (i) to distinguish vaccinated and non-vaccinated animals (ii) for early diagnosis of infected animals (iii) arrange vaccination for non-vaccinated animals, (iv) post-vaccination surveillance to conduct re-vaccination and (iv) treatment and isolation for animals infected with newer serotypes.

Results

Sequence selection for VP1

The phylogenetic tree revealed that all the FMDV serotype O isolates of Bangladesh clustered in Ind2001d, Ind2001BD1 and Ind2001BD2 sub-lineages of Ind2001 lineage with exception of 3 isolates which were clustered and have been merged as outgroup along with other isolates (Fig. S1a in supplementary additional file 1). However, all isolates were under Middle East-South Asia (ME-SA) topotype. All of the isolates collected from recent outbreaks were found within Ind2001BD1 sublineage and this sub-lineage appeared as dominant sub-lineage over others in recent years. The circulation of more than one sub-lineage within a single lineage Ind2001 of serotype O was indicative of extreme genetic variation. From these clusters, isolate BAN/TA/Dh-301/2016 was considered to be a potential candidate for selection which was further validated by analyses of genetic distance among the isolates (Fig. S1a in the supplementary data). The phylogenetic tree of FMDV serotype A showed that all the respective isolates of Bangladesh were clustered in Genotype VII under Asia Topotype. These findings indicated less genetic variation within the serotypes. In this tree, BAN/Ch/Sa-304/2016 was initially selected as a potential candidate and representative of serotype. A circulating in Bangladesh (Fig. S1b in supplementary additional file 1). In phylogenetic tree of serotype Asia1, all the local isolates were found within the same cluster. However, this is the least circulating serotype in Bangladesh and no isolates of Asia1 serotype were found from recent outbreaks. Consequently, isolate BAN/Dh/Sa-318/2016 as representative of serotype Asia1 was selected as a potential candidate (Fig. S1c in supplementary additional file 1). The mean genetic distances of each of the selected sequences was found to be the lowest ones among the respective serotypes (Table S1 in supplementary additional file 1). This indicated that, the selected candidates more closely represented the serotypes than others in terms of genetic distance.

Sequence selection for 3AB

In order to analyze the diversity based on 3AB gene region, the aligned whole genomes were trimmed for the particular region in accordance to GCF_002816555.1 reference sequence and was subjected to phylogenetic analysis. The analysis also included the 3AB region of recently reported MYMBD21, belonging to novel sublineage under SA-2018 lineage11. The tree generated showed that 3AB sequence was not evolutionary very diverse across the serotypes specially circulating in Bangladesh (Fig. S2a in supplementary additional file 1) and nearby countries (Fig. S2b in supplementary additional file 1).

From the Wu Kabat variability plot generated on protein alignment sequences of the selected genes (Fig. 1), it was observed that, at most sites there was non-significant variation as variability was not above threshold 1.0 but at some sites higher variability existed. Mostly at amino acid position ranging from 125 to 150, significant variations were observed. A consensus sequence was prepared after analyzing the aligned protein sequences and the most occurring amino acids at positions of variability. .

Fig. 1
figure 1

Wu Kabat variability plot for 3AB protein alignment. The graph represents the Wu Kabat variability plot generated on protein alignment sequences of the selected genes. There is non-significant variation at most sites as variability is not above threshold 1.0 but at some sites higher variability exists. Mostly at amino acid position range from 125 to 150 more number of significant variations observed.

Full size image

Sequence selection for VP2

The VP2 selected protein of 219 amino acids long, incorporated the designed 130 amino acids long N-terminal consensus with six surface-exposed B cell epitopes, and suggested the possible potentiality of the protein for the development of a serotype-independent FMDV detection tool in Asia during our recent investigation13. This protein was synthesized and used for this holistic approach.

Plasmid designing

All the engineered sequences were further optimized and evaluated for heterologous expression in E. coli. The optimized sequences were sent for artificial synthesis in commercially available pET21a + plasmid vectors. The plasmids were named B4, B5 and B6 having the VP1 sequences of serotype O (BAN/TA/Dh-301/2016), serotype A (BAN/CH/Sa-304/2016) and serotype Asia1 (BAN/DH/Sa-318/2016) respectively. The plasmids P2 and P3 contained the designed sequences of VP2 and 3AB respectively. The codon optimization was done using OptimumGeneTM algorithm. The length of gene of B4 was 639 bp while being 633 bp for both B5 and B6, P2 was 669 bp and P3 was 684 bp, all of which were optimized for expression in E. coli. The native gene employed tandem rare codons that could reduce the efficiency of translation or even disengage the translational machinery. So, the codon usage bias was increased in E. coli by upgrading the Codon Adaptation Index. GC content (%) and unfavorable peaks were also optimized to prolong the half-life of the mRNA. Codon adaptation index (CAI) of more than 0.8 was considered and regarded as good for higher expression and that of value 1.0 was considered perfect. The CAI of B4, B5, B6 and P3 were increased to 0.97 for each and P2 increased to 0.95. Frequency of optimal codon before and after optimization when compared, indicated a prominent shift towards a higher percentage of codons with optimal quality and the same was observed for GC content percentages of the optimized sequences. The reports of OptimumGeneTM Codon Optimization analysis from GenScript were enclosed in supplementary additional file 2 Fig. s1).

Prediction of mRNA secondary structure

The evaluation of minimum free energy for mRNA of optimized sequences was carried out by the ‘RNAfold’ and ‘UNAFold’ web servers. The results were summarized in the Table 1 which showed greater stability of the optimized sequences as their minimum free energy are more negative.

Table 1 Evaluation of mRNA stability before and after optimization.
Full size table

Prediction of secondary structure of protein

Secondary structure of the B4, B5, B6, P2 and P3 proteins were predicted using GOR-IV (Supplementary Fig. S2). Secondary structure analysis of B4 protein from optimized sequence revealed that protein was made of 33 extended strands (15.21%), 79 helices (36.41%) and 105 (48.39%) random coils. B5 protein from optimized sequence showed that the protein i-was made of 56 extended strands (26.05%), 65 helices (30.23%) and 94 (43.72%) random coils. B6 protein was found to be made of 36 extended strands (16.74%), 78 helices (36.28%) and 101(46.98%) random coils. P2 contained 81 extended strands (36.99%), 23 helices (10.50%) and 115 random coils (52.51%). Secondary structure analysis of P3 protein from optimized sequence revealed that protein i-was made of 11 extended strands (4.91%), 100 alpha helices (44.64%) and 113 (50.45%) random coils. There was no ambiguous state found in any one of them. The results from Jpred 4 also corroborated well (report available from corresponding author if required).

Prediction of 3D structure, stability and physio-chemical parameters of the protein

3D models of unique designed proteins were shown in the Fig. 2 with both helix and surface structures for each protein. The major antigenic site of VP1, G-H loop were shown in red for B4, B5 and B6 and were shown to be present on the surface of the protein. Since, G-H loop contains RGD motif which binds to host cell receptor avβ3 integrin. Hence, presence of RGD motif on the surface of the proteins as observed when analyzed using PyMOL ensured the binding probability to the receptor. The 6 major antigenic sites for VP213 were shown in red for P2. The antigenic sites in 3AB protein14 were marked red in P3. The structures were also validated from Ramachandran plot (Additional file 2 Fig. s3). Stability and other physico-chemical parameters, analyzed via the Expasy’s ProtParam tool were shown in Table 2. The results from PredictProtein server were enclosed (Additional file 2 Fig. s4).

Fig. 2
figure 2

3D structure of 5 designed proteins with cartoon and surface model for each. B4.1, B5.1, B6.1, P2.1 and P3.1 represents cartoon model and B4.2, B5.2, B6.2, P2.2 and P3.2 represents the surface model of B4 (VP1 from serotype O), B5 (VP1 from serotype A), B6 (VP1 from serotype Asia1), P2 (designed VP2) and P3 (designed 3AB) accordingly. Major antigenic site, G-H loop that contains RGD motif for B4 (VP1-O), B5 (VP1-A), B6 (VP1-Asia1) and B cell epitopes for P2 (VP2) and P3 (3AB) has been shown in red. The G-H loops in the surface of all three VP1 proteins and the epitopes on P2 and P3 are exposed and hence suitable for antibody binding.

Full size image
Table 2 Physico-chemical Parameters of B4, B5, B6, P2 and P3 proteins analyzed by Expasy’s ProtParam tool.
Full size table

Expression and characterization of engineered proteins

All the plasmids after transformation into E. coli BL21 (DE3) strains were induced and from SDS PAGE of whole cell extracts the induction results were evident through thick bands at appropriate positions (Fig. 3). The VP1 proteins of serotype O, A and Asia 1 (B4, B5, B6 respectively) were in the range of 23–24 kD as their molecular weights were concerned. After SDS PAGE analysis of the whole cell extract, three distinct bands were evident near 22–23 kD in insoluble fraction but not in soluble fraction (Fig. 3a). The VP2 and 3AB protein (P2 and P3 respectively) were in the range of 23 kD and 24 kD respectively. After SDS PAGE analysis of the whole cell extract, 2 distinct bands were evident near 23–24 kD in insoluble fractions (lane 1,7 for VP2 and lane 2,8 for 3AB) but not in soluble fraction (lane 3,4) (Fig. 3b). Those induction bands were not present in lanes loaded with non-induced extracts (lane 5,6) as negative controls. Therefore, all proteins were induced and expressed in insoluble fraction of the cell.

Fig. 3
figure 3

SDS-PAGE analysis of whole extract cells after induction. (a) Lane 2,3,4 shows thick induction bands in insoluble fraction of B4, B5, B6 which belong to serotype O, A, Asia1 respectively. No band in soluble fractions observed. (b) Lane 2,8 shows induction band of P2 (VP2) and lane 3, 9 shows induction band of P3 (3AB) in insoluble fractions. Neither soluble fraction (lane 4, 5) nor non-induced fraction (lane 6,7) has any band. After purification based on His tag, the SDS PAGE shows single band for each (c) B4 (VP1-O), B5 (VP1-A), B6 (VP1-Asia1) inlane 2,3,4 respectively and (d)P2 (VP2) inlane1 and P3 (3AB) in lane 7. The range of 23–24 kDa approximately is applicable for each of B4, B5, B6, P2 and 25 kDa for P3.

Full size image

Protein purification and identification

Once the overexpression of desired protein was confirmed in the insoluble fraction, purification and desalting of the protein present in the cell pellets were done15. The purified and desalted proteins were again analyzed by SDS-PAGE which showed single band for B4, B5, B6, P2 and P3 proteins near 23–24 kDa (Fig. 3c,d). In Fig. 3c, lane 2,3 and 4 showed purified protein bands for B4 (serotype O), B5 (Serotype A) and B6 (serotype Asia1) respectively whereas in Fig. 3d, lane 1 and 7 showed bands for P2 (VP2) and P3(3AB) respectively.

Evaluation of immunogenicity of the expressed proteins

VP1: The western blot analysis was extended further to determine the antigenic reactivity against serotype specific anti VP1 antibody. Primary antibody for this purpose was obtained from cattle sera collected after the cattle were vaccinated with representative strains from which VP1 sequences were selected16. The bands at appropriate positions verified the binding of antibodies to the respective VP1 proteins (Fig. 4a).

Fig. 4
figure 4

Western blot analysis to determine antigenicity of expressed proteins. a) The thick bands at 23–24 kD area confirms the expression of our desired proteins VP1-O (B4), VP1-A (B5) and VP1-Asia1(B6) as well as their credibility to be detected by serotype specific antibody obtained from vaccinated sera with respective serotype. b) The presence of thick bands at all 3 positions VP1-O (B4), VP2 (P2) and 3AB (P3) refer to detectability of antibodies by the expressed proteins in infected cattle serum S466 and infection caused by serotype O.

Full size image

3AB and VP2: The western blot analysis was extended further to determine the antigenic reactivity against serotype non-specific anti 3AB and anti VP2. Primary antibodies for this purpose was obtained from cattle sera collected during 2021 FMD outbreak in Mymensingh and denoted as S46611,17 (Table S3 in supplement 1). In Fig. 4b 3 bands appeared, each one for B4, P2 and P3 inferring- (i) the sera contained antibodies against VP1, VP2 and 3AB, (ii) it was from infected cattle and (iii) the virus causing the infection belonged to serotype O.

This demonstrated the ability of the proteins to bind specific antibodies. VP1 of each serotype will bind anti-VP1 antibodies of that serotype only, making it a serotype dependent detection system. 3AB will bind anti 3AB, which will be present in infected cattle serum. Besides VP2 will bind to anti-VP2 antibodies, which will be present in cattle with active infection as well as in cattle vaccinated with whole-virus vaccines18. If the cattle is vaccinated with recombinant protein or, subunit vaccine, then the presence of anti-VP2 will indicate the presence of active infection only. Thus, the anti-VP2 immune response may provide a reliable detection of current infection in cases of subunit vaccination groups. Besides, the presence of anti-VP2 antibodies and absence of anti-VP1 antibodies in the same unvaccinated animal will indicate the transmission and emergence new serotypes (other than O, A and Asia1) in the susceptible population.

Dot blot hybridization

Figure 5 shows the results of dot blot hybridization when the membrane was treated with infected cattle sera S466. Earlier, a gradient ELISA was performed using the serum sample S466 which was serially diluted upto 1024 fold dilution. It was observed that 100 ng of each of B4 (VP1-O), B5 (VP1-A), B6 (VP1-Asia1), P2 (VP2), P3 (3AB) could detect their respective antibodies into serum diluted 1024 fold times (data not shown). On dot blot hybridization, three dots appeared, showing hybridization of anti-VP1 (serotype O), anti-VP2 and anti-3AB antibodies present in cattle sera with membrane bound immobilized proteins B4, P2, P3 respectively. A comprehensive large scale validation with field sera samples from different farms with government and non-government authorization will be required for conclusive conclusion of the experimental results.

Fig. 5
figure 5

Dot blot hybridization assessment for VP1, VP2 and 3AB. The infected serum S466 showed positive dots for all VP1-O (B4), VP2 (P2) and 3AB (P3) justifying successful hybridization of the anti-VP1, anti-VP2 and anti-3AB antibodies present in the serum with the immobilized proteins on the membrane.

Full size image

Predicted prototype of a prominent multi-diagnostic assay kit

The biomolecular engineering framework of the study stands on the fact that upon introduction of foreign antigen, the host produces polyclonal antibodies or antisera against the antigen. In case of FMDV vaccination, antisera will incorporate antibodies against only capsid protein VP1 of single or multiple serotypes but in case of natural infection antibodies against non-structural proteins will also be circulating in the serum. Now utilizing VP1, VP2 and 3AB as antibody capturing antigens on a common platform (Fig. 6) paves the principle of a prototype model of diagnostic kit with all desired attributes. Figure 6 represents the role of each protein and their cumulative interpretation upon implementation as diagnosis tool.

Fig. 6
figure 6

The schematic representation of the use of the expressed proteins in diagnosis along with the resultant interpretation. The proposed platform allows to specifically determine which serotype amongst O, A, and Asia1 is causing the infection (row 1,2,3), can assess the type of vaccination used inactivated or recombinant based on VP2 and 3AB (row 4,5) and can forewarn about emerging serotypes based on VP2 (row 6). Altogether, this proposes a comprehensive and innovative diagnostic approach.

Full size image

Discussion

The OIE and Food and Agricultural Agency (FAO) have legitimately suggested eradicating FMD globally by the year 203014, which cannot be achieved without precise, efficient surveillance of vaccinated and infected cattle. Although certain ELISA-based kits are available and imported kits are utilized19, all of them are either expensive or only applicable for particular serotypes20. Alternatively, the few established DIVA kits lack the ability of sero-surveillance21. Significant shortages of diagnostic tools and laboratory facilities for Foot-and-Mouth Disease (FMD) throughout the country like Bangladesh create a pressing need to develop more precise and reliable testing kits for commercial and private rural dairy farms. The current diagnostic kits aim to detect anti-NSP antibody and anti-VP1 antibody in separate experiment set-up, both of which requires collection of hundreds of samples to accommodate in a single ELISA plate. This causes delay in laboratory testing after collection, testing samples from unrelated area and use of expensive reagents. For efficient implementation of the aforementioned FMD control approaches in rapid and cost-effective manner, a comprehensive, single-use and point-of-care kit is required as per our study which can (i) differentiate vaccinated and infected cattle, (ii) identify FMDV serotype causing the infection, (iii) detect the emergence or re-emergence of novel or distant serotype in circulation, and lastly (iv) the post-vaccination surveillance for proper timing of booster doses. In this project, we focused on developing recombinant protein based diagnostic system for FMDV which will be easy-to-produce and user-friendly. There are a number of FMD kits available which are mainly ELISA based system and requires bulk number of samples (multiples of 96) and pre-programmed ELISA reader for quantitative detection of anti-FMDV antibody in susceptible livestock population. Besides, a number of lateral flow kits for FMDV diagnosis have been introduced in different countries, but no such kits neither available in this region, nor any study for detecting suitable candidates to develop a comprehensive FMDV have been reported. The proposed prototype will be applicable as Point-of-Care diagnostic platform at field level as well as local laboratories using single blood samples. The present study describes the efficient design of structural (VP1 and VP2) and non-structural (3AB) protein of FMDV in a prokaryotic system, conserving the surface exposure of neutralizing and non-neutralizing epitopes, confirmed by the dot-blot experiments. This study showed an effective approach for simultaneous detection of anti-FMDV immune response generated from different arrays of antigens. However, large scale validation of antigenic specificity and sensitivity with substantial amount of sera samples from different farms will follow-up in collaboration with government non-government authorization. This platform would be one of its novel kind so far reported or available.

VP1 sequences of three isolates namely BAN/TA/Dh-301/2016 (Serotype O), BAN/CH/Sa- 304/2016 (Serotype A) and BAN/DH/Sa-318/2016 (Serotype Asia-1) were selected as the candidates based on an earlier study of our group where the same isolates were selected for a trivalent vaccine formulation16. To validate the selection, phylogenetic tree and genetic distance analysis were performed along with the isolates from the most recent outbreaks9,11. All the local isolates were closely related to the selected isolates in the phylogenetic tree.

3AB sequence belongs to the region which is more or less conserved with few variations. The inference is also true for local isolates circulating in Bangladesh as our study incorporated the sequences of local isolates and also isolates from neighboring countries as import or cross border movement of animals take place quite frequently. The residues which appeared more frequently among the predominant strains were considered and thus the prepared consensus was suitable for use, irrespective of different serotypes. Besides, the consensus had even less genetic distance with local isolates’ sequences than the wild strains’ sequences had among themselves.

Thus the selection of the sequences for each protein was such that it encompasses all the circulating strains in Bangladesh even the newly emerged ones like MYMBD2111, and it also benefits neighboring countries such as India and Myanmar with whom we share border and transborder exchange often occurs. This type of all incorporating system is unique and one of its kind.

Before expression, the sequences were optimized in terms of codon usage, evaluated for mRNA secondary structure, protein secondary and tertiary structure. The three-dimensional structures of B4, B5, B6 revealed the G-H loop containing RGD motif to be exposed on the surface. The proteins therefore can bind to the antibody by G-H loop. For P3, it was better to use whole peptide sequence instead of using truncated or partial epitope sites only. This might yield more stable and functional antibody capture agent. Example of expression of truncated 3AB protein is also present which led to increase in solubility of the protein22. Since we employed ‘His’ tag to our constructs and purification of inclusion bodies from whole cell extracts was easily possible so truncation was avoided. Moreover, being a viral protein, the size of sequence was already manageable to proceed onto heterologous expression system. Moreover, the optimization of codons led to better stability of the mRNA secondary structures and tertiary protein structures than the wild sequences for all constructs. Therefore, the designed proteins were suitable for heterologous expression.

The conventional ELISA based on structural or non-structural proteins of FMDV uses partially purified antigens from infected cell cultures as diagnostic antigens, which require handling of live virus, posing risk of accidental virus escape from laboratories, and also these partially purified antigens often lack batch-to-batch uniformity23. The risk of virus release is the primary consideration for most researchers in the choice of virus antigen24. To address this limitation, recombinant proteins and peptide derived from FMDV have been developed as alternative to whole-virus and inactivated virus antigen. A study has been reported to express C-terminal half of VP1 of FMDV in E. coli followed by purification and confirmation of type specific reactivity. Thus E. coli expressed proteins and antibodies to the expressed proteins may form an alternative and cheap source of diagnostic reagents 25. Wang et al. (2007) subcloned the complete FMDV VP1 protein in an expression vector and transformed E. coli BL 21(DE3) to establish indirect ELISA to detect FMDV antibody in pigs26. A number of other FMDV proteins have also been expressed through heterologous expression in E. coli including VP227, NSPs28, and virus like particle29 of FMDV which is becoming increasingly popular, successful and cheap alternative to whole virus isolation.

Following purification of the desired targeted proteins, they were checked for their ability to capture antibodies through western blot and dot blot hybridization. Similar study with large number of sera samples was conducted and they also showed antibody capturing ability of 3AB (expressed in baculovirus expression system) in blocking ELISA format30. Though our sample number was small and expression system was E. coli but the obtained results synchronize with their study in justifying 3AB’s detection ability. Furthermore, structural protein VP2 was included in our investigation because anti-VP2 antibodies can provide information about infections that can arise from newly emerged and reemerged strains of FMDV. While VP2 is often conserved, infection with a new serotype will provide results that are VP1 negative but VP2 positive because VP1 varies greatly between serotypes. However, VP2 has changed quite little. Therefore, even if VP1 is unable to attach to anti-VP1 antibodies produced against the new serotype, the VP2 construct may still be effective against it indicating the emergence of novel strains, even serotypes. This can serve as a warning that a distant, new, or reemerging strain has entered the population against which natural or vaccine immunity is lacking, and can forewarn of an impending outbreak.

Considering the rational findings of the study, it is obvious that all of the component proteins VP1-O, VP1-A, VP1-Asia1, VP2, 3AB have been successfully examined as viable candidates for the diagnostic assay and have been efficiently expressed in heterologous low-cost system conserving the antigenicity of the proteins. Here we propose a comprehensive lateral flow assay platform (Fig. 6) using our unique and designed FMDV candidates for a conclusive assay kit. As demonstrated, if 3AB and VP2 are both positive, a positive VP1 would identify the serotype causing illness; if not, the result would suggest vaccination or previous immunization. Positive VP2 but negative VP1 circumstances can also serve as a warning sign for newly emerging serotypes. Salem et al. described the efficiency pf VP2 for serotype-independent screening of FMDV in Egypt, which indicates the reliability of VP2 as marker for identifying introduction of novel or distant serotypes like SAT 1,2,3, serotype C27.

Recently, different types of vaccines like inactivated whole-virus vaccine and subunit vaccine are also reported for prevention of FMDV. The anti-VP1 and anti-VP2 antibodies can differentiate among animals vaccinated with different types of vaccines. The negative anti-VP2 response indicates the recombinant protein or, subunit based vaccination, and the efficacy of the vaccine can easily be determined as long as the anti-VP2 is negative, indicating absence of infection. The anti-VP2 response can also aid into understanding the vaccination history and determining revaccination schedules for susceptible livestock population. Moreover, a 3AB positive response not only represents the recent infections, but also in cases of vaccinated, uninfected animals it indicates the poor manufacturing of vaccines and failure to remove NSPs. The prototypic model describe in this study can be simulated into a standard lateral flow assay strip, consisting of overlapping membranes on a firm plastic base. It is predicted that due to capillary action, the protein molecules embedded on the cellulose membrane will interact with the supplied serum, capturing the inherent antibody (if present). The assay will involve the use of colored or fluorescent particles as the tracer, which can be Goat anti-bovine IgG antibody (Invitrogen, USA) linked with horseradish peroxidase (HRP) to provide a signal. Color will be formed as a result of this antibody’s particular interaction with the target analyte and the subsequent chemiluminescent reaction. This unique approach of incorporating all our designed candidates in a single kit will particularly assess all circulatory serotypes responsible for FMDV in a specified sample for diagnostic and epidemiological purpose.

Conclusion

The primary goal of this research was to select, optimize, express, purify, and assess the qualitative potency of uniquely engineered VP1, 3AB and VP2 capsid proteins as serotype dependent and independent anti-FMDV antibody detection from bovine serum. Following an in silico biomolecular engineering of effective proteins, these were successfully expressed in E. coli BL 21(DE3) cells and purified (Fig. 7). The key accomplishment was the demonstration of specific immunogenicity of the produced proteins for serum antibody detection in a combined platform which will be applicable for any FMDV endemic areas. The high yield production of the proteins can aid in the development of a low-cost indirect detection approach in future. Besides, immobilized antigens were promising in qualitative, robust detection of FMDV within very short time implicating a unique holistic engineered platform for pre- and post-endemic screening as well as post-vaccination monitoring. We also proposed here a prototype model which is easy to develop employing the findings of the study. Although the assay validation was done using selected cattle sera, a comprehensive field level study with statistically correlated sera samples would be required upon collaboration with any government or non-government organization.

Fig. 7
figure 7

The figure illustrates the whole proposal of developing a comprehensive, multi-antigenic, point-of-care, novel diagnostic prototype. The process starts from FMDV genome study and selecting potential sequences for VP1 (serotypes O, A and Asia1), VP2 and 3AB followed by their optimization in stage 1. According to that, plasmid was designed and transformed into competent E. coli (BL21) cells. After inducing those cells overexpression of desired protein was obtained which was confirmed from SDS PAGE denoting completion of Stage 2. Successful purification of the proteins employing His tag affinity chromatography was stage 3. In Stage 4, the proteins were tested for their immunogenicity and hybridization and results of dot blot hybridization signifies the ability of immobilized proteins to specifically bind and detect specific antibodies in serum. Thus the proteins testify their role as forerunners of a holistic platform suitable for both sero-surveillance and differentiation between vaccinated and infected. The illustration was generated using BioRender (https://www.biorender.com/).

Full size image

Materials and methods

Sequence retrieval and dataset preparation

For the selection of suitable candidate sequences for three serotypes, respective sequences were retrieved from National Center for Biotechnology Information [NCBI] GenBank database and datasets were prepared accordingly31. Samples from the most recent outbreaks in different parts of Bangladesh were collected followed by amplification of VP1 region and sequencing in Microbial Genetics and Bioinformatics Laboratory (MGBL) at the Department of Microbiology, University of Dhaka9. All datasets with respective accession numbers are available in supplementary table S2.

Sequence analyses and selection of suitable VP1 sequence

Phylogenetic analysis

For Phylogeny, three different datasets were prepared for FMDV serotype O, A, and Asia1 (Table S2a,b,c in supplementary additional file 1). The dataset for serotype O included a total of 68 VP1 gene sequences among which 39 were sequenced in MGBL and 29 sequences were retrieved from NCBI database. For serotype A, the dataset was prepared by a total of 41 VP1 gene sequences (14 from MGBL and 27 retrieved from NCBI). In case of Asia1 dataset, a total of 50 VP1 gene sequences including 10 from MGBL were used. The sequences of all the datasets were first aligned using built in ClustalW program of MEGA732, then phylogenetic trees were constructed by the maximum likelihood method based on the Tamura-Nei model33. Gamma distribution and bootstrap values were set at 1 and 1000 respectively34. Topotypes, lineages and sub-lineages of the FMDV isolates were determined from the constructed phylogenetic trees comparing with reference sequences. The tree was then condensed using 5% cut off value. Phylogenetic analysis was constructed in MEGA 7.0. From the phylogenetic analysis and distance analysis (Table S1 in supplementary additional file 1) FMDVs O/BAN/TA/Dh-301/2016 (MK088170.1), A/BAN/CH/Sa-304/2016 (MK088171.1) and Asia1/BAN/DH/Sa-318/2018 (MH457186.1) isolates were selected as representative strains and these are also reported as vaccine candidate strains16.

Genetic distance analyses

Genetic distances of potential strains and the circulating local isolates in Bangladesh were calculated based on the genetic variations among the sequences. Analyses were conducted in MEGA7 using the Kimura 2-parameter model. The rate variation among sites was modeled with a gamma distribution (shape parameter = 2). The analysis involved 69 nucleotide sequences for serotype O, 21 for serotype A and 20 for serotype Asia1. Codon positions included were 1st + 2nd + 3rd to measure genetic distance for all possible reading frames. All ambiguous positions were removed for each sequence pair. There was a total of 639 positions in the final dataset. Finally, the mean genetic distances were calculated and compared by SPSS Program Version 20. For calculation and comparison of mean, 2-tailed T-Test with 95% confidence interval and 1000 bootstrap value was used.

Sequence analyses and consensus preparation for 3AB

Phylogenetic analysis

For Phylogeny, whole genome datasets were prepared for FMDV serotype O, A, and Asia1 (Table S2d,e in supplementary data). The dataset for Serotype Asia1 contained 11 whole genomes, A contained 7 whole genomes and most prevalent O contained 21 sequences and 1 reference genome accession number GCF_002816555.1 (https://www.ncbi.nlm.nih.gov/data-hub/taxonomy/12118/) which were reportedly circulating in the South East Asia region particularly in Bangladesh, India and Myanmar. Nine of these complete genomes were sequenced in the MGBL that were circulating in Bangladesh. These included two genomes each from serotypes Asia1 and A, as well as five genomes from serotype O9. The sequences of all the datasets were first aligned using built in ClustalW program of MEGA732, only the 3AB region of the aligned sequences were taken trimming off the rest portion of the whole genomes and phylogenetic tree was constructed by the maximum- likelihood method based on the Tamura-Nei model33. Gamma distribution and bootstrap values were set at 1 and 1000 respectively34.

Protein variability study and consensus preparation

PVS is a web-based tool that uses several variability metrics to compute the absolute site variability in multiple protein-sequence alignments. The translated alignment file was uploaded to study the Wu Kabat coefficient to understand the frequency of amino acid substitutions at each position in a protein sequence, based on a multiple sequence alignment of homologous proteins. A consensus sequence (P3) was prepared containing the most occurring amino acid residues at sites where variation appears based on the alignment study and Wu Kabat plot.

Codon optimization

The codons of the selected sequences were optimized for desired expression through a heterologous expression system. A wide variety of factors that can regulate and influence gene expression levels were considered, including (i) Codon usage bias (ii) GC content (iii) CpG dinucleotides content (iv) mRNA secondary structure (v) Cryptic splicing sites (vi) Premature Poly A sites (vii) Internal chi sites and ribosomal binding sites (viii) Negative CpG islands (ix) RNA instability motif (ARE) (x) Repeat sequences (direct repeat, reverse repeat, and Dyad repeat) (xi) Restriction sites that may interfere with cloning. To optimize these parameters, OptimumGeneTM algorithm was used which takes into consideration as many of them as possible, producing the gene sequence that can reach the highest possible level of expression.

Prediction of mRNA secondary structure

The structure of folding RNA messenger of the constructed gene was analyzed and after gene optimization, the energetic stability of the predicted structures were compared together by the UNAFold Web-based software (unafold.rna.albany.edu) and the RNAfold online server (rna.tbi.univie.ac.at)35.

Prediction of secondary and tertiary structures of the candidate proteins

The prediction of secondary structure of proteins was performed using GOR-IV (gor.bb.iastate.edu/) and Jpred 4 (compbio.dundee.ac.uk/jpred/) online servers36. In order to analyze the sequence and predict the protein structure and function, the percentage of random coils, alpha-helices and beta sheets, the solvent accessibility were evaluated by the PredictProtein server37. To build a three-dimensional tertiary structure of the proteins, SWISS-MODEL server was used. Suitable templates were found after searching the databases. The templates were then used to build and predict protein tertiary structures through homology modeling. For P3, I-TASSER software was used as suitable templates were unavailable. After the model was built, the suitable results were viewed with PyMOL molecular graphics systems38. 3D structures were evaluated using Ramachandran Plot Analysis39.

Prediction of stability of the candidate proteins

To predict the stability of the proteins, various physical and chemical parameters were computed using ExPASy-ProtParam tool40.The computed parameters included the molecular weight, theoretical pI, amino acid composition, atomic composition, extinction coefficient, estimated half-life, instability index, aliphatic index and grand average of hydropathicity (GRAVY).

Suitable consensus for VP2

An in-depth computational analysis of 14 compiled datasets was already conducted to determine the level of conservation in VP2 at different hierarchical levels of three FMDV serotypes (O, A, and Asia1) currently circulating in Asia, and a consensus VP2 protein was designed by our group previously13, that was used for the development of a serotype-independent FMDV detection tool.

Heterologous expression of designed plasmids

The selected protein sequences were designed for cloning and expression in E. coli BL21. So, commercially available compatible expression vector pET 21a ( +) (5.4 Kb) was chosen. B4 (O serotype VP1), B5 (A serotype VP1), B6 (Asia1 serotype VP1), P2 (VP2), P3 (3AB) plasmids were thus transformed into competent E. coli BL21(DE3) strains and induced using 1 mM IPTG (Isopropyl ß-D-1-thiogalactopyranoside).

Protein extraction and purification

Cells were harvested by centrifugation at 5000 rpm for 5 min at 4 ℃, resuspended in 5 mL of PBS and subjected to sonication at 100 W and 10 s pulse for 10 min. After sonication, centrifugation was done to separate soluble and insoluble fractions at 10,000 rpm for 10 min at 4 ℃. The protein-coding inserts were incorporated in the pET21a + vector with a 6X Histidine-tag located at the C-terminal of the protein, which aided in the purification of recombinant proteins using immobilized metal affinity chromatography. Once the overexpression of desired protein was confirmed in the insoluble fraction, purification was performed using Ni–NTA chromatography spin column (Thermo Scientific™, USA) according to manufacturer’s protocol15 and the final elution was performed using 500 mM imidazole.After purification, the remaining contaminants from elution buffer were removed using the ZebaSpin Desalting Column (Thermo Scientific™, USA) according the manufacturer’s protocol. After desalting and buffer exchange, the purified proteins were finally stored in PBS (pH 7.4).

SDS-PAGE analysis of proteins and western blot

Both the whole cell extracts as well as the purified proteins were subjected to SDS PAGE using Bio-Rad apparatus and conventional protocol. The purified separated proteins (5 µg) were transferred to 0.4 µm (by BIO-RAD) nitrocellulose membrane using Trans-Blot® SD Semi-Dry Transfer Cell (BIO-RAD). Semi-dry transfer was done by preparing a sandwich of 2 layers paper, nitrocellulose membrane, electrophoresed gel, and again two layers of paper and was immersed subsequently in transfer buffer, then placed in between two carbon plates (cathode and anode) in the blotting apparatus and connected to a power supply set at 25 V for 30 min. The protein-blotted membranes were incubated with SuperBlockTM Blocking Buffer (Thermo Scientific, USA) at 4 ℃ for 1 h to prevent non-specific background binding of the primary and/or secondary antibodies to the membrane. Next, the membranes were incubated overnight at 40º C on a rocker with primary antibody, the cattle sera were prepared at a dilution of 1:1000. For anti VP1 antibodies, sera from calves vaccinated with monovalent vaccines during a previous study16 were used and for anti VP2 and anti 3AB antibodies, naturally infected cattle sera used (Table S3 in supplementary additional file 1).

Dot blot hybridization

Dot blot hybridization of the purified proteins were done following Abcam protocol41 (Dot blot protocol, Abcam). With a narrow-mouth pipette tip, 2 µl purified protein containing 1 µg protein of proteins were blotted on a nitrocellulose membrane. After drying, blocking buffer was used to block non-specific sites for 30 to 60 min at room temperature. For 30 min at room temperature, the membrane was incubated with primary antibody (1:1000). It was incubated with an anti-bovine secondary antibody conjugated with HRP for 30 min at room temperature after being washed three times with TBS-T. After that, color development and result observation occured after TBST wash and the addition of colorimteric reagent within five minutes. Instead of ECL reagent, the Colorimetric chromogenic substrate (Thermo Scientific,USA) was used for detection of horseradish peroxidase (HRP) activity.