Determining the Antibody

The existence of four DENV serotypes (DENV 1-4) and the association of prior DENV infection with an increased risk for severe disease presents significant challenges to the healthcare infrastructure of subtropical countries, which have a huge burden of Dengue cases every year. People who are infected a subsequent time with a different serotype of the dengue virus may experience “antibody-dependent enhancement” (ADE), where the dengue virus of the different serotype takes advantage of pre-existing antibodies from the first infection to increase its virulence leading to a more severe form of DENV infection. ADE occurs due to the binding of human antibodies (like IgG) through their Fc region to immune receptors, and incomplete (non-neutralising) binding to the pathogen.


Ideation 1

IgY, the avian antibody, due to its non-reactivity with human immune receptors, can prevent ADE by binding specifically to the virus particle to neutralise it. We planned to switch the constant region of a full length IgG antibody to IgY. Goose-derived anti-DENV IgY has been shown to neutralise DENV infection without triggering ADE.

Problems

We faced problems with figuring out the mechanism of clearance and the potential immunogenicity of an antibody from a non-human, non-mammalian system. We also had multiple questions on the general immunotherapy techniques we could have used to modify IgY and make it a functional therapeutic. After thorough research and conversations with experts, we felt that we weren’t able to answer these questions adequately and decided to move onto another option.


Ideation 2

Another way to prevent ADE is to remove glycans from the DENV-targeting antibody, producing an aglycosylated IgG molecule. N-linked glycosylated is necessary for antibodies to bind to effector receptors, so having the antibody be aglycosylated would abrogate effector functions without affecting antigen-binding, hence making an effective therapeutic against Dengue without causing ADE.

Problems

The production of full-length antibodies is quite expensive. Bacteria cannot produce it in high yields, or with the necessary glycosylation. Yeast can produce the protein, but the differing glycosylation patterns mean that it is processed differently by the body’s machinery. Mammalian cell lines can produce antibodies, but are quite difficult to handle and grow.


Ideation 3

We settled on synthesising antibody fragments, specifically scFvs (single chain fragment variable), which would be a small protein, easy for bacteria to synthesise, with high affinity and specificity to the antigen. It would have no means of interacting with immune receptors, so it would not trigger ADE.

The issue arose that scFvs have quite short half-lives in serum, due to their size and non-interaction with the FcRn receptor, which is responsible for preventing the lysis of native antibodies in the body. We resolved this by adding a short FcRn-binding peptide into the protein, which mimics the interaction of IgG antibodies with the receptor.

Determining the Epitope

Ideation 1

The DENV non-structural protein 1 (NS1) is a component of the viral replication complex and can be found on intracellular membranes as well as on the cell surface. Endothelial hyperpermeability and vascular leak, pathogenic hallmarks of severe dengue disease, are directly triggered by DENV non-structural protein 1 (NS1). As such, anti-NS1 antibodies can prevent NS1-triggered endothelial dysfunction in vitro and pathogenesis in vivo.

Problems

We discovered that the NS1 protein only tends to accumulate once the dengue infection has already become quite severe. Additionally, anti-NS1 antibodies tend to cross-react with host proteins. They also target and destroy infected endothelial tissue, which could potentially lead to the worsening of the hemorrhagic fever.


Ideation 2

We then looked into the Fusion Loop Epitope (FLE), which is highly conserved across all the serotypes of dengue and is the target for several known neutralizing antibodies. It is important for the virus’ entry into the cell, so blocking it would effectively erase its pathogenicity.

Problems

The FLE, while conserved, is often hidden on the surface of the virus. The degree to which it is hidden varies across serotypes and stages of the virus life-cycle. So while it would be a great target for a neutralizing antibody, it might not always be accessible by one.


Ideation 3

We spoke to Dr Vidya Mangala Prasad and presented our candidates for epitope. She pointed out the issue with the FLE and suggested that if we were looking for a conserved epitope that is the target of neutralizing antibodies, we would be best off with the E-dimer epitope (EDE). EDE is a quaternary epitope that ticked all the boxes we were interested in. We looked into the topic ourselves after our conversation with Dr Prasad, and decided on a specific EDE-targeting antibody - C10.

After deciding our antibody and antigen, our goal now was to clone our c10-scFv insert DNA into our pET-21b vector. We made competent cells to transform our positive clone into. We then wanted to optimise the SHuffle system to produce our protein of interest by standardising certain factors like expression temperature, concentration of IPTG, and growth (OD) at the time of induction.

Cloning
We started by cloning our c10-scFv insert into our pET-21b vector. We had already transformed our insert (in a pUCIDT-KAN vector synthesised by IDT) into a DH5-𝞪 chemically competent cell to amplify it and eluted 50ul of the plasmid using plasmid prep.


Design Build Test Learn
We wanted to digest both vector and insert plasmid with NdeI and XhoI enzymes, with the final goal of cloning the insert into the vector. We set up a restriction digest for both the vector and the insert - a 30uL reaction with 1uL of each enzyme, left overnight. The reactions would be run on a 0.8% agarose gel and extracted to conduct the later steps. When the gel was imaged, we weren’t able to see a defined vector band - there was a lot of background. Insert band was visible but also had a huge smear. Star activity in vector DNA might have caused multiple fragments in the vector and the smear. We were advised to have a shorter digest period and a larger reaction volume.

First restriction digest with pet-21b vector and IgG-IgY-aMBP insert

Design Build Test Learn
We integrated the advice of our mentors from the first round into our experiment plan. We started with 1000 ng each of vector and insert DNA and set a 50 µL reaction using NEB high fidelity restriction enzymes. We ran the digest for three hours and merged wells in our gel to ensure proper loading of all our DNA. Vector and insert bands were visible and the DNA was extracted. After setting up ligation and transformation, there were more colonies on the control plate as compared to the 1:3 plate. It was suggested that our digest had not been complete. We also learned that DNA obtained from alkaline lysis is quite impure, and could possibly have interfered with the efficiency of our digest. Additionally, while it was sufficient this time, quite low amounts of DNA were obtained after extraction.

Plates after transformation with ligated product, control has more colonies than 1:3 plate

Design Build Test Learn
We used spin columns to isolate plasmids to ensure high purity. We began with 2000 ng of DNA so that we would have sufficient amounts following extraction. We set up a second control for ligation, this one without ligase, to determine how much of the vector had remained undigested. After determining the total amount of vector plasmid, restriction digest was set again in the manner learned previously. Extracted DNA was set for ligation, with two controls (Vector, Vector+ligase). There were no colonies seen on the first control plate (Vector). The second control plate (Vector+ligase) had about 30 colonies and there were almost 200 colonies on the 1:3 plate. Cloning was successful. Positive clones were found through alkaline lysis and agarose electrophoresis. The cloned plasmids were sent for sequencing.

Fig: L to R, ligation control, 1:3 plate, digestion control, 1:5 plate

Making competent SHuffle cells
In order to express our protein into SHuffle, we needed competent cells to transform our plasmid into. However, transforming SHuffle K-12 proved tricky and there were multiple rounds before we were finally able to successfully make chemically competent SHuffle K-12 cells.


Design Build Test Learn
Use chemical competency methods to make competent SHuffle cells. SHuffle K-12 cultures were grown and an ASB buffer (with bivalent cations) along with glycerol was added for ensuring chemical competency. No colonies were observed after transforming a plasmid into freshly made chemically competent SHuffle cells. Transformation was repeated, to obtain the same results. We were advised to monitor OD more thoroughly and use freshly prepared ASB buffer; we had re-used the buffer from a previous round of experiments.
Design Build Test Learn
Repeat the previous protocol, with fresh buffer and less lag time between optical density measurements New ASB buffer was prepared, cultures were kept on ice always and OD readings were taken with a higher frequency. No colonies were observed after transforming a plasmid Upon consulting lab members who had previously worked with this strain of SHuffle, we decided to try out the Inoue method. SHuffle K-12 was hypothesised to be incompatible with ordinary chemical competency.
Design Build Test Learn
Research the Inoue method. Inoue method of chemical competency was applied to SHuffle-K12 cells. Cells were cultured, washed with Inoue buffer, then dried. More buffer was added, along with DMSO. Colonies were observed after transforming our antibody into ultra-competent SHuffle K-12 cells SHuffle K-12 was made competent successfully.
To view the tables, please view on the desktop

Optimization of antibody production in a prokaryotic system
We wanted to optimise the production of our protein in our chassis. After research and conversations with experts, we settled on DOE, or statistical Design of Experiments. The goal was to find the optimum set of conditions so the yield of soluble protein was maximised.


Design Build Test Learn
We identified six factors worth investigating for influence on yield. We constructed a six-factor classical screening design in JMP. Found to be unfeasible at the scale and resources - the number of experiments was too high to do. We needed to find a more efficient way of identifying interactions - a six-factor model was far too laborious.
Design Build Test Learn
We spoke to experts and our mentors, and narrowed the model down to three factors that were deemed the most important. A response surface design on OD at induction, temperature and IPTG concentration was constructed - yielding 15 experiments. Due to erroneous execution, we lost a few cultures due to growing beyond the determined value of optical density. These experiments were repeated. SDS-PAGE gels were run in triplicate, and imaged to measure the amount of the protein in the supernatant.
Design Build Test Learn
ImageJ was used to set up the analysis. The software was standardised using gels run with multiple concentrations of BSA. The software was set up to analyse the gel images. The final response outputs were inputted into regression analysis in JMP Interaction between factors were identified.