Designing of nanobodies against Dengue virus Capsid: a computational affinity maturation approach

Abstract Dengue virus, an arbovirus, is one of the most prevalent diseases in the tropical environment and leads to huge number of casualties every year. No therapeutics are available till date against the viral disease and the only medications provide symptomatic relief. In this study, we have focused on utilizing conventional nanobodies and repurposing them for Dengue. Computationally affinity matured, best binding nanobodies tagged with constant antibody regions, could be proposed as therapeutics. These could also be applied for drug delivery purposes due to their high specificity against the viral Capsid. Another application of these nanobodies has been thought to utilize them for diagnostic purposes, to use the nanobodies for viral detection from patient samples at the earliest stage using ELISA. This study may open a new avenue for immunologic study in foreseeable future with the usage of the same molecules for multiple purposes. Highlights Natural nanobodies against viruses were modified for use against Dengue virus Capsid conserved regions. Computational affinity maturation was performed making use of change in binding affinities upon mutating various residues in the complementary determining regions. Docking studies performed to inspect the docking groove, interface analysis and energy calculations. MM/GBSA calculations done to calculate binding free energy of the complex to determine stability of the complex. Communicated by Ramaswamy H. Sarma


Introduction
Dengue is a disease caused by Dengue virus, an arbovirus, i.e. an arthropod-borne virus. It belongs to the family Flaviviridae and genus flavivirus consisting of four serotypes. Although a fifth serotype was reported in 2013, no sequence level information is available. It is an enveloped virus with icosahedral symmetry and a single-stranded positive polarity genome (Salles et al., 2018). The virus has a high infectivity rate of approximately 100-400 million cases/year with a mortality rate of 10% for hospitalized patients and it increases to 30% for non-hospitalized ones. The mortality rate can be as low as 1% if proper care is available. The mode of treatment is purely symptomatic and no specific highly effective therapeutic is available till date (Salles et al., 2018). Due to the lack of treatment options, it is a major source of severe illness in several Asian and Latin American countries.
Modelling studies have revealed that almost 3.9 billion people are at risk to contract the disease in 127 countries and thus without some proper and quick-acting therapeutic the death toll could also be severe. The rate of infection has also increased almost eight fold in the past two decades making the disease one of the most important to counter in the world (https://www.who.int/news-room/fact-sheets/detail/ dengue-and-severe-dengue).
Antibodies have been used for over a century to treat several diseases. But their bottleneck comes in the fact that they are very large, fragile and have very poor cell penetrability thus making them non-suitable to target specific cells in the body. This is where the nanobodies hold the edge in this age where biomedical efforts are entering the sub-cellular level making use of these nanobodies, nanomedicine and nanotherapy all the more important. Nanobodies are antigen-specific, recombinant, single domain variable regions of primarily Camelid antibodies which have been shown to comprise only the heavy chains, though nanobodies have later been discovered from other sources as well (Hassanzadeh-Ghassabeh et al., 2013). Apart from their small size and greater penetrability, another feature that stands out for nanobodies is their refolding property following heat denaturation though nanobodies have also been shown to undergo heat denaturation dependent aggregation which can still be surpassed by forming disulphide linkages making nanobody engineering more important as well (Kunz et al., 2018). Thus, the nanobodies are an active area of research that are under trial against inflammation, breast cancer, brain tumour, lung diseases and others. Recently a bivalent nanobody called caplacizumab has been approved by FDA and European Medicines Agency for use against thrombotic thrombocytopenic purpura where there is excessive blood clotting in smaller blood vessels thus interrupting the blood flow (Jov cevska & Muyldermans, 2020).
In recent times, many studies have focused on drug repurposing to target novel targets using previously established drugs. This helps bypassing the efficacy, stability and toxicity-related questions for the drugs. Similarly, in this study, we have used established and verified nanobodies extracted from natural sources and targeting specific targets for targeting newer proteins. For this we considered the virus targeting nanobodies and used them to screen for their binding capacity against the conserved regions of the Capsid protein of dengue. We did not stop here, we tried to increase the binding affinity and energy besides the specificity of the nanobodies without affecting their overall framework which prompted us to use the CDR regions for affinity maturation. Affinity maturation is a natural process in our body where to increase binding affinity to a particular antigenic epitope the body mutates and selects antibodies with greater affinity. The same is done here, the difference being the thymus is replaced by our computer, the follicular cells and germinal centre is replaced by docking and simulation programs and the various enzymes replaced by servers and algorithms. Thus like the body, the ones which have a greater binding affinity come out on top while the rest are discarded.
This study would help us to design new modern therapeutics beyond the use of antibodies, drugs and vaccines as therapeutic agents. These are the upcoming fields to be utilized not only as therapeutics but also in diagnosis. These nanobodies which have been computationally affinity matured, can serve as an initial step for nanobody therapy, besides nanobody based diagnosis including nanobody based ELISA and nanobody based highly specific drug delivery system among many others.

Protein sequence retrieval and structure preparation
For this work, particular proteins need to be targeted which remain on the surface of the virus as antibodies cannot penetrate the cells. So the three structural proteins of Dengue virus were taken into consideration as Capsid, membrane and envelope. Of these, the envelope protein is 475 amino acids long thus providing plenty of room for the proposed nanobodies to bind but it also leads to the loss of specificity of the same. In case of membrane protein, no such stretch of conserved residues was found though the protein was only 81 amino acids in length. The Capsid protein was highly conserved as seen by aligning all reviewed sequences downloaded from Uniprot of the same using Multalin and the protein too was only 100 amino acids in length. From the conserved regions, stretches were identified where almost all residues remain conserved. The protein was also matched to the proteome databases using BLASTp and PSI-BLAST to ensure the uniqueness of the target and prevent any chance of cross-reaction to occur with other human proteins or other organisms which are present in the body. The structure of the same was prepared by a hybrid technique of homology modelling and threading using PHYRE2 in the intensive mode (Kelley et al., 2015). The structure was shown to have a highly stable structure with little disorderedness at the terminal region and the target regions were selected such that they fall in the structurally defined regions.

Nanobody retrieval and structure preparation
The nanobody sequences were retrieved from the Single Domain Antibody Database (Wilton et al., 2018). Here only those nanobodies were selected which are targeted towards viruses like Norovirus, Influenza virus, Arenavirus Juninvirus and MERS Betacoronavirus. Most of these 59 nanobodies have experimental evidence against them which include viral neutralization assays, ELISA and others. The nanobodies were also aligned using Multalin against each other to find out the unique nanobodies. All the nanobody structures were prepared using SWISS model homology modelling where both high coverage and identity were looked into for better prediction of the structures to be used in the study.

Docking
The nanobodies were docked to the protein using Cluspro docking server (Kozakov et al., 2017). In each case the first cluster with maximum number of models was selected as the frequency of the results point towards the greater chance of that interaction happening. All docking results were analysed for their interchain interactions and the interfaces were analysed so as to select those complexes which form interaction pairs involving the residues under consideration using PPcheck server and InterProSurf servers (Negi et al., 2007;Sukhwal & Sowdhamini, 2015). The interchain residues which occur within 8 Angstrom distance were also looked into as these residues primarily have the chance to form bonds and infer specificity to the nanobodies which would help in targeting the same towards specific proteins. The last step was conducted using the COCOMAPS server (Vangone et al., 2011).

Computation Affinity maturation
The selected complexes were analysed using various servers to predict the effect of mutations on binding affinity. For this the CDR1, CDR2 and CDR3 of the selected nanobodies were seen from the database based on their sequence features. The effect of mutations on these positions was first broadly screened using BeAtMusic server and the mutations which lead to increase in binding affinity were selected for (Dehouck et al., 2013). After this, the selected mutations were reanalysed by mCSM-AB and mCSM-AB2 servers but all these three servers only pointed out the single point mutations and their effect on binding affinity (Myung et al., 2020;Pires & Ascher, 2016). According to the independent model of mutation the final resultant affinity could be calculated by simple adding the affinities up for each position mutation. But as stated by the epistatic model of mutation this is not the case all the time and 2 or more mutations may have some additional relationship between them above the additive nature. So to analyse these mutations in combinations ProAffiMuSeq server was used where to conduct double mutation one mutation was fixed while the other conducted (Jemimah et al., 2020). Those combinations which increased the binding affinity were selected for and proceeded with to conduct triple mutations, finally using this data to chalk out all combinations of mutations possible which would result in increasing the binding affinity. All these combinations were selected based on the above method by fixing all other positions and targeting the mutation for a single position. This led us based on sequence features to our final set of computationally affinity matured nanobody sequences.

Conformational docking
The affinity matured nanobodies were put into Swissmodel to prepare their structures and docked in the same was as their native nanobodies. The results were selected solely based on their energy and cluster sizes. The docked structures were analysed based on the residue contacts to further shortlist the number of final nanobodies.

Affinity and energy analysis
The stability and binding affinity of protein-protein structures can be detected by observing their change in Gibb's free energy. So the same was analysed using PRODIGY web server and only those were selected where there is a decrease in the change in Gibb's free energy as it indicates towards increased stability of the complexes (Xue et al., 2016).
To further validate the same MM/GBSA simulations were carried out using the HawkDock server (Weng et al., 2019). The server conducts the simulation consisting of 5000 steps of which 2000 cycles correspond to steepest descent and the rest 3000 are conjugate gradients using ff02 forcefield and implicit solvent model. The binding free energies were considered for all the models.
Finally to sort out the issue of steric clashes and others all the structures were refined using GalaxyRefineComplex server (Heo et al., 2016). The interfacial residues were analysed using the InterProSurf server on these refined structures and the final shortlisting was done to arrive at the least number of nanobodies that have a marked level of specificity for the particular proteins.

Molecular dynamics simulation
GROMACS 2020.4 (Lindahl & Hess, 2020) was used to simulate capsid nanobody complexes. OPLS/AA forcefield (Robertson et al., 2015) with SPC water model and Dodecahedron simulation box was used in the simulation. The solvated protein complexes were neutralized by adding sufficient Na þ ions. A steepest descent algorithm (maximum no. of steps 50000) was used to energy-minimize the neutralized system. Then the energy minimized systems were subjected to 500 ps NVT equilibration at 300 K temperature and successively the NVT equilibrated system was subjected to 500 ps NPT equilibration at 1 atm pressure and 300 K temperature. During NVT and NPT equilibration period, the complexes were fully restrained with 1000. NPT equilibrated systems were then subjected to 50 ns production run with 2 fs time step with no restraint. The smooth Particle-Mesh Ewald method (Essmann et al., 1995) was used to calculate long-range electrostatic interactions and a 12 Å cutoff was used for both PME and van der Waals interactions. All the analyses of the simulated system were done using in build GROMACS commands. The simulation trajectory was used to quantify the Lennard-Jones, Coulombic and potential energies of the complexes.

Capsid nanobody binding free energy calculation using MM/PBSA method
Binding free energies of capsid nanobody complexes were computed applying the Molecular Mechanics/ Poisson-Boltzmann surface area (MM/PBSA) (Bakan et al., 2011;Vorontsov & Miyashita, 2011) method using g_mmpbsa tool (Kumari et al., 2014) with default parameters. 200 snapshots were taken from last 10 ns of the trajectory (at every 50th ps time) of each run for the calculation of binding energies of the complexes. The default solute dielectric constant of 2 was taken and the binding energies were calculated. Further optimization was done by increasing the solute dielectric constant to 4 and 70, of which the latter is the same as for that of the solvent model used (Li et al., 2013;Reddy & Berkowitz, 1989).

Protein selection and structure preparation
Dengue virus has 3 structural proteins. Of these the envelope being too big was not considered as it would compromise on the specificity of the nanobodies while membrane protein was rejected as no complete extensive stretches of conserved regions were found. This left us with only Capsid protein which has a small size of 100 amino acids and is highly conserved too. From the Uniprot database 20 complete reviewed proteomes of Dengue virus serotypes 1, 2, 3 and 4 were extracted and aligned against a Capsid protein having Uniprot ID P17763. From this the conserved residues were mapped and 2 stretches of conserved residues-14-25 and 59-66 taken for further studies. In these two stretches only one residue is not 100% conserved and that is residues number 19 (Figure 1).
The Capsid protein was matched against both the human proteome and the non-redundant protein sequence database. In both cases for both BLASTp and PSI-BLAST no results were found. It points to the direction that there is no similarity or homology of the considered protein with any other protein. This ensures the specificity of the nanobody to not attack other organism proteins. It also ensures that the incorporation of the nanobodies do not turn auto-immunogenic by targeting the human body proteins which is ensured by the lack of any similarity of the Capsid protein with all other proteins found in the human proteome. To construct the structure of the same no appropriate template was found so the model prepared using the PHYRE2 web server. While analysing the model it was seen that the first 10 residues of the model were disordered but it didn't affect the analysis as residues belonging to the structured region were taken into consideration. Also 90% of the model residues had a reliability of 100% which also made the model closest to the actual scenario if not the same (Figure 2).
3.2. Nanobody retrieval and structure preparation 59 nanobody sequences were retrieved and their structures were prepared. The models from the Swiss model server were selected keeping in mind the template coverage, identity and GMQE values. Thus the best models of the same were selected (Figure 3).

Docking and selection of models
The nanobodies were docked with the Capsid protein using Cluspro docking server where the nanobodies were considered as receptors and the protein the ligand. The docking results were analysed based on interacting residues, interfacial residues and close contacts. All three criteria prompted us to reach the conclusion that 4 nanobodies are selected for further analysis. These are: VHH-58 which is used against MERS betacoronavirus, sdAb-DB id sdAb_1538_Cd, VHH-101 which is also used against MERS coronavirus, sdAb-DB id sdAb_7423_Cd, D4 used against influenza Nucleoprotein, sdAb-DB id sdAb_5011_Ca and J2.3 used against Juninvirus, sdAb-DB id sdAb_5733_Vp. The residues which were considered for the selection are given in the table. In the table only residues that were under consideration in this study have been mentioned. There are many other residues present in each criterion, but they have not been mentioned as they are beyond the scope and of no significance in this particular study. The CDR regions of the 4 nanobodies were also mapped by taking data from the database. The residues corresponding to CDR1, CDR2 and CDR3, respectively, are as fol  Figure 4).

Computational Affinity maturation
The process was carried out where in the first step of broad screening 65 mutations across 13 residues were identified for VHH-58 all belonging to the CDR regions. The number for VHH-101 is 85 mutations across 22 positions, for D4 it is 28 mutations across 7 positions and for J2.3 it is 105 mutations across 15 positions. Further analysis of the single mutants using the different servers and algorithms left us with 13 mutations across 7 residues for VHH-58, 20 mutations across 10 positions for VHH-101, 4 mutations across e positions for D4 and 24 mutations across 12 residues for J2.3.These final mutations were considered for several combinations to conduct affinity maturation. By the various steps, we finally arrived at five affinity-matured sequences for VHH-58, 16 for VHH-101, 3 for D4 and 18 for J2.3. All these were subjected to conformational docking to reach at the final number of nanobodies which would bind the target protein with high specificity and affinity, more than the original sequences.

Conformational docking and analysis
The docking procedure was the same as the previous one. Here, we analysed the structures based on their binding affinities which were interpreted using the values of change in Gibb's free energy. This analysis further brought down the numbers to 1 for VHH-58, 12 for VHH-101, 3 for D4 and 8 for J2.3. All these selected structures were simulated for MM/ GBSA calculation and their binding free energies determined. For all the structures the matured structures showed better energy than the original structures. But there was only one exception the matured structure of VHH-58. In the initial analysis, it was seen that this is the best model out of all the models as it had a very stable interaction with the protein with plenty of residues belonging to the selected region close to the nanobody residues apart from a large number of interacting and interfacial residues. On comparing the energy it was seen that the binding free energy of the same was lower than many other structures. Since we are considering them there was no reason to discard this as well. Finally, to get rid of the steric and other geometric and spacial clashes, the structures were refined and the interfacial residues mapped. This mapping was the final step of shortlisting and it led us to a total of 13 nanobodies, 1 each for VHH-58 and D4, 8 for VHH-101 and 3 for J2.3. In this analysis again it was seen that a large number of residues are present in the interface of the capsid protein and the matured sequence of VHH-58 reiterating on the fact that we have no particular reason to discard it from the final list in spite of the fact that it may not be better than the original structure unlike all other cases ( Figure 5, Table 2 andSupplementary 1, Table 3).

Molecular dynamics simulation analysis
We have conducted 50 ns MD simulation run to assess the dynamic characteristics of the nanobodies complexed with the capsid protein. From the trajectory of each of the capsid nanobody complexes, it was seen that the Coulombic and Lennard-Jones interaction energies were highly negative in nature for all the cases. The potential energies were also found to be highly negative reiterating on the stability of the complexes formed. Graphical representation of RMSD for all the complexes shows that the RMSD (Figure 6) of the backbone does not deviate too much throughout the trajectory. Average RMSD along with standard deviation (Table 4) for each set were calculated considering the plateau regions only. Other RMSD graphs are given in Supplementary material 2.

MM/PBSA analysis of the complexes
To quantify the strength of interaction between capsid proteins and nanobodies, MM/PBSA methods were applied by taking 100 snapshots from 40 to 50 ns from the trajectories. We have taken the last 10 ns of the trajectory as the trajectories reach a plateau indicating the convergence of the simulation. Using the default dielectric constant value of 2 resulted in 7 cumulative positive binding free energies (Supplementary material 3) (vdW þ electrostatic þ polar solvation energy þ SASA) out of the total 13 stable complexes. The rest of the seven values were highly negative in nature (Figure 7). The most anticipated reason behind the positive binding energy could be the unusual positive values of both electrostatic energy and polar solvation energy. Contemplating the scenario, another set of MM/PBSA calculations were executed taking the solute dielectric constant of 70 which is almost equal to the solvent dielectric potential. This result reported that 3 out of 7 had changed from positive to negative binding energy (Supplementary material 4) but by a very small extent as compared to the 6 stable complexes (from 1st MM/PBSA set). We have also conducted the   -58 17, 20, 59 20, 59, 61 14, 15, 16, 17, 18, 19, 20, 21, 24, 25, 59, 60, 61, 62, 65 À11.3 À98.55 same calculation using the dielectric constant of 4 and in this case complex 58.1 alone converted into negative binding energy (À61.054 ± 82.017 kJ/mol). Thus, the 6 nanobodies i.e. 101.2, 101.5, 101.11, J2.3.3, J2.3.8 and J2.3.9 showing negative binding energies were selected to be having a proposed action in binding to the capsid in a highly stable manner.

Discussion
Dengue virus, a mosquito borne flavivirus, causes dengue fever and Dengue shock syndrome across more than 100 countries round the world. The number of cases has been consistently rising around the globe. Till date no therapeutic is available against dengue which allows for cross-protection though the process started in around 1929. In some countries, vaccine is available for dengue but it has been seen to work on people who have suffered from the disease earlier while for absolutely new people who have never contracted the disease the situations have worsened. This may be due to the inability of the vaccine for cross protection which may lead to enhanced effects when a different serotype attacks due to Antibody-Dependent Enhancement. Several tetravalent live attenuated virus vaccines are presently in clinical trials while some are showing promising results in the same (http://www.denguevirusnet.com/vaccine-research.html). Using antibodies as therapeutics, a method of passive immunization is used to combat several infectious diseases such as rabies, diphtheria, tetanus, hepatitis B, respiratory syncitial virus and others for over a century. Though the effects are short-lived in comparison to vaccines but their effects are instant, work on immuno-compromised people and their development time is shorter (Sparrow et al., 2017). But using antibodies has its own set of complications. Their large size makes it very tough to target particular cells besides its lack of specificity. It is in these cases where the importance of nanobodies lie. They are small, highly soluble, highly stable and excellent tissue penetration ability. So they can be used to get to places where normal antibodies fail to reach (Bannas et al., 2017).
In this study, we have taken into consideration nanobodies with already available experimental evidence and plan to use them to target our particular protein of interest. By comparing the protein sequences we have been able to find a completely conserved region in the protein irrespective of serotype which is highly specific to the protein in consideration and shows no matches in BLAST whether it be in other microbial population or the human proteome. The nanobodies were docked and selected based on their interacting and interfacial residues and these selected nanobodies were gone through with computational affinity maturation. In our body in the lymphoid organs once a particular clone of cells identify the antigen by their binding to the variable region of their membrane-bound antibody, affinity maturation begins in the germinal centre where follicular cells are      present. The maturation happens involving the hyper-variable regions in the variable region of the antibodies. Which mutation is beneficial and which is not is determined by binding of the matured antibodies to the antigen present on the surface of the follicular B cells. In the process, most of the mutations are rejected while only those where the binding affinity gets improved from the parent antibody are produced in huge quantities to act against the antigen. Similar to this approach we conduct multiple mutations in the CDR regions of the nanobodies which are essentially the variable regions of the antibody. Only mutations that increase the binding affinity are taken into consideration for further analysis where we find out the right combinations of mutations which would effectively increase the binding affinity of the nanobodies when they bind the antigen leading to the formation of a highly stable complex as can be seen from the calculation of the binding energies of the same from the 50 ns simulation trajectories, also taken into consideration is the fact that the nanobodies would bind to the areas of interest as off-target nanobodies would lead to a non-specific response which is of very less use in targeting the viral Capsid protein.
According to the conventional independent model of mutation, it is said that each mutation occurs independent of the rest and their effect is also independent of any other mutation. But a new model called the epistatic model considers the interrelationship between these mutations. According to this, the relation between two mutations or more is not always additive in nature. The co-occurrence of multiple mutations affect the protein in a variety of ways which can never be explained satisfactorily by the independent model. This is where the need for conducting double or triple mutation at the same time arose. So to do this first we considered one mutation to be fixed and studied the other mutations with respect to it. Thus all the pairwise inter-relationships between the mutations were considered and combinations were formed such that in a combination no two mutations are present where there is a lowering of binding affinity. The same process was done till the entire combination was validated at a sequence level which still doesn't reveal the entire picture. That is why the docking and simulation was performed to validate which of the results hold good in all the scenarios. Also multiple mutations need to be taken into consideration as there is a question of target specificity and by conducting mutations according to the independent model this specificity related details can never be deduced as the specificity is a dimensional aspect not just a one-dimensional sequence aspect. Thus for this the epistatic model was adopted where multiple mutations were considered at the same time for the purpose of affinity maturation. The short-listing of the data points towards more specificity and better binding affinity and energy. It ensures that the final resulting nanobodies do not interact the protein by chance, the interaction is highly specific, stable and robust which is further validated by the molecular dynamic simulation results.
In our body, the effector action is performed by the Fc regions of antibodies be it by the help of the complement system or other modes. So if the Fc region of a foreign antibody is the same in origin to a human then the antibody can behave similarly to a self antibody. Here comes the use of the proposed nanobodies which can be tagged to the heavy chain Fc regions and used in the human body. It would be a combination of the specificity incurred by the nanobody which would help to recognize the target Capsid protein while the humane constant region would ensure proper effector action. So by a little bit of modification essentially addition this antibody can be used for therapeutic purposes.
A second use of the nanobody is to act as a target delivery system where the nanobody can be tagged to a drug or some toxin and this would be targeted to the viral Capsid due to the fact that the nanobody is specific for the protein. But in this process, it has to be ensured that the delivered item is only released or becomes functional when there is specific strong and stable interaction between the two.
A third function in which this nanobody could be used is for the detection or diagnosis of the disease. At present the diagnostic methods include IgM test but the issue with it is crossreaction with other flaviviruses, RT-PCR within the first week or NS1 test till 12 days (https://www.cdc.gov/dengue/healthcareproviders/diagnosis.html). But for IgM test the body needs to elicit an immune response that does not happen instantly or doesn't happen at all in immuno-compromised patients. In case of NS1 test, the protein is detected only when the NS1 protein is released from the infected cells. Thus the latter test is positive only in patients where the infection cycle has proceeded a bit. Thus the reliable test for the detection is RT-PCR which is quite costly and time-consuming. So apart from these ELISA can be done using these nanobodies. This is capable of giving results at the earliest even before infection into the cells as happened as here a structural protein is targeted which can be readily recognised by the nanobody from the patient blood and thus may provide a much more efficient, fast and cheap mode of detection of the disease which would lead to early diagnosis and the same tagged with the constant region can also be used for treatment purpose as well.
Thus, this study provides a starting material for development of passive immunizing therapeutic against this disease which claims hundreds of lives each year. This study aims at designing the presently available nanobodies for newer applications. This would only provide a lead in the direction of developing novel therapeutics against this virus as also novel detection mechanisms. Further in vivo and confirmatory experiments need to be done to look into the efficacy of the nanobodies.

Conclusion
Dengue fever leads to numerous deaths across the world specially in the tropical countries. Till date no generalized drug or antibody is available which can be used against all the serotypes of the virus. Here we have attempted to make use of the already established nanobodies against various human infecting viruses to use them and/or their modifications against the Dengue virus structural protein, Capsid.
Binding with consensus region was tested for and subsequently, the CDR regions of the selected nanobodies were computationally affinity matured. Mutated versions of the nanobodies were predicted and analysed to be more specific to the target regions in the Capsid protein. This study may serve as a beginning in designing novel nanobodies as a therapeutic option by tagging them with not only humanized antibody constant regions but also with drugs to deliver them to the target cells with higher specificity. These nanobodies could also be used in ELISA for better, faster and efficient detection of the virus in patient samples before the infection has even proceeded enough for immunogenic reaction or virus proliferation inside the host body as detected by NS1 positive tests. Thus, these modified nanobodies could be used in the field of therapeutics but diagnostics too.