.
Implementation of the synthetic biology engineering cycle was at the forefront of our minds when structuring our project for this year. The aim of our project was to use a modelling approach in designing inhibitory peptides against fungal effector proteins in stem rust. Following the de novo design of these peptides, our wet lab team intended to measure their binding with the fungal target proteins in the laboratory using a FRET assay. We had planned for the information gained from this test to feed back into the dry lab and allow further improvement in the peptide design.
Additionally, this cycle could be endlessly repeated to yield several strongly-binding peptides for implementation in future applications. Due to constraints surrounding the time required to design the inhibitory peptides, we were unable to synthesise them in the lab before the deadline. This could possibly be the task of a future iGEM team, allowing completion of our initially intended engineering cycle.
Despite this limitation, our team discovered that there were other aspects of our project that aligned with the criteria for engineering success. These included:
With the aid of advice from experts, our wet lab, dry lab and human practices teams worked collaboratively to weave various stages of the synthetic biology engineering cycle into the structure of our project this year.
.
.
With such a large portion of our project attributed to the modelling efforts of our dry lab team, we came to realise that the process of designing the inhibitory peptides themselves resembled an iteration of the engineering cycle. These short peptides were designed to prevent effector proteins in stem rust from binding to their corresponding protein binding partners in the plants’ immune system. In preventing this binding, the function of the plant immune system should remain intact and promote a more well-rounded immune response to the pathogen.
Unique fungal protein targets in stem rust were identified through research and a subsequent literature review of effector proteins in Puccinia rust species (Experiments). Further research of literature allowed us to find a corresponding protein in the plant immune system with which the target protein naturally binds. We then used these known proteins as known templates for the design of our own inhibitory peptides.
We found the sequence of each fungal protein through literature and used these as templates to produce a sequence with the GGSG linker and mVenus fluorescent protein attached. Two versions of each fungal protein sequence were created, one with the fluorescent protein attached at the N-terminus and another at the C-terminus (Part: BBa_K4467007, BBa_K4467008, BBa_K4467009, BBa_K4467010, BBa_K4467011, BBa_K4467012). We then produced 3D structure of both types of sequences using AlphaFold (for more information regarding our AlphaFold modelling, see Modelling)
The AlphaFold results were qualitatively assessed by inspecting the structure of each protein, ensuring the fluorescent protein, mVenus, wasn’t obstructing or interacting with the active site of the fungal protein. The confidence/ pLDDT scores of each AlphaFold structure were then assessed to ensure the structure was of adequate accuracy for further modelling simulations. The AlphaFold outputs of best structure and pLDDT scores were selected for subsequent analysis.
These pairs of structures were inputted in ClusPro (Porter et al., 2017) where the protein target was the receptor and the peptide was the ligand for docking simulations. We then took the top 10 balanced docking structures to record which regions of the peptide were interacting with the protein target. These regions were identified using Pymol to view peptide residues that were within 5 Angstroms of the target protein and polar contacts between the peptide and protein target. After determining the main interface areas of the peptide sequence we used this to design peptides.
To achieve the goal of producing a peptide to bind to the protein target, we must better understand how the template protein interacts with the protein targets. This was done by using the predicted active regions of the template peptide to design peptides that only possess different combinations of these regions. Furthermore, to better understand how the structure of the active site region behaves, we built peptides that possessed the same combinations but replaced the non active site regions with similar sequences that have known 3D structures.
These peptide sequences are then used to build 3D structures using AlphaFold. These structures were filtered by pLDDT score and by how well the active site regions are exposed for docking.
These filtered structures were then run as ligands with the target protein in the docking software, acting as a receptor to generate docking simulations.
After generating docking structure data, we learnt that we had multiple issues that needed to be addressed in order to engineer a peptide to interact with the target protein. These issues consist of:
Knowing this information, we need to produce scripts for active site annotation to handle the data analysis workload, use different docking software such as GalxyPepDock to generate more docking structure data to verify our findings and to assess which docking software is better suited for our protein. Furthermore, we should have also analysed the active sites of the peptide without the GGSG linker and fluorescent protein attached, so we can compare the differences in active sites.
.
.
As an initiative to contribute towards future iGEM teams, our wet lab aimed to study the interaction between the selected fungal and plant proteins. The purpose of studying this interaction was to uncover the nature of the native binding between fungal effector proteins and the plant immune system proteins. We hope future iterations of the engineering cycle will result in improved design of the fluorescent tags for FRET analysis and the application of quantitative experimental techniques, for example mass spectrometry.
Throughout our literature search for fungal effector proteins, we found four viable targets:
Pgt_IaaM (Part: BBa_K4467004),
PstHXT1 (Part: BBa_K4467003),
PstSCR1 (Part: BBa_K4467001) and
Pst_12806 (Part: BBa_K4467002).
However, due to time constraints and the integration of advice from fungal experts (see Megan Lenardon in
Integrated Human Practices
), progress on some of the fungal proteins was discontinued throughout the year. Consequently, the engineering cycle was only fully completed for the fungal protein
Pst_12806 (Part: BBa_K4467002) and plant protein
TaISP (Part: BBa_K4467005).
With the aim to investigate the interaction between our selected fungal effector proteins and their binding partners in plants, we began to plan possible methods for our experiments, eventually settling on a FRET assay. As FRET is highly dependent on the presence of fluorescent proteins, a variety of Puccinia effector proteins were modelled using AlphaFold to predict whether the addition of the fluorescent protein mCerulean and Histidine tag would disrupt the proteins’ structure when added to the N or C terminal. Likewise, AlphaFold modelling and inspection of the protein structure was performed on the corresponding plant proteins when mVenus and a Histidine tag was added to either terminal. The highest confidence proteins with little changes in structure following addition of the fluorescent tag were selected for synthesis in the lab (see Modelling). Each of the chosen proteins (Pgt_IaaM, PstHXT1, PstSCR1, Pst_12806, TaISP and SERK3B) were designed to allow for Gibson assembly into a pet19b plasmids (Figure 1 & 2). In the design of these inserts, a TEV site was added between the fluorescent protein and the protein coding sequence to ensure mCerulean/ mVenus was able to be cleaved if we decided to perform an experiment not requiring the fluorescent tags.
The DNA sequence we designed for the fluorescent fungal and plant fusion proteins were ordered as gBlocks, inserted into a pet19b plasmid through Gibson assembly and transformed into NEB TURBO E. coli . Single colonies were selected for colony PCR screening to ensure the correct length of insert was introduced into the plasmids. The plasmid DNA of successful colonies were isolated by Miniprep, sequenced at the Ramaciotti Centre for Genomics and transformed into NEB T7 Express E. coli cells for subsequent protein expression. The fluorescent fusion proteins for Pst_12806 and TaISP were expressed in E. coli and the lysate was utilised for FRET experimentation. Additionally, this method was also performed to express the fluorescent proteins, mCerulean and mVenus, by themselves. These protein samples were purified using gravity columns and the AKTA Start purification system, and were used as a negative control for the FRET experiment.
We tested the binding between proteins through a FRET assay, taking advantage of the overlapping excitation and emission spectra of mCerulean and mVenus to detect interaction between TaISP and Pst_12806 (Demir et. al., 2017). The samples from prior protein expression were lysed through sonication, centrifuged and filtered. The Nanodrop measured the concentration of protein in each sample, which was then utilised to calculate dilutions for standardising protein concentrations. Serial dilutions the proteins were prepared (see Modelling) in a 96-well FRET plate and analysed inside the plate reader. Light corresponding to the excitation spectra of mCerulean was used to excite the fluorescent proteins, while light with a wavelength corresponding to the emission spectra of mVenus was detected.
Our project was not without fault. In undergoing our design-build-test methodology, we have continued to evolve and learn from our mistakes, and plan for the future. Here, one of the first roadblocks we encountered involved the insertion of the gene sequences into the pET-19b plasmid. We used a plasmid backbone that already contained another insert, which had to be cut out. We initially used XbaI and BamHI to cut the insert, though XbaI cut further away from the insert, leaving a small overhang that would disrupt the Gibson assembly process when trying to insert our fungal protein sequence. As such, we decided to use a different restriction enzyme, NcoI, which cut closer to the insert site to avoid such issues.
Furthermore, the expression and purification of the fungal proteins and plant binding partners also presented a challenge. The expression of the TaISP plant protein, tagged with mVenus, realised some fluorescence, though this was after rounds of optimisation of the temperature and IPTG inducer concentration conditions to use. To overcome such low expression, naturally we upscaled the expression system to a larger culture volume to produce larger amounts. On the other hand, the Pst_12806 fungal protein proved even more difficult. Though there was nothing in the literature accounting for the expression of Pst_12806 in an E. coli host, we aimed to try, nonetheless. For this protein, there was no fluorescence seen in the sample, perhaps due to low levels of expression. In this case, we would need to upscale into larger culture volumes like we did for TaISP. More worrisomely, perhaps when modelling the proteins on AlphaFold and designing the mCerulean3 tag was there a significant structural change that impacted its ability to fluorescence, something that was unbeknownst to us. If this was the case, we would need to revisit our models and reiterate our design of the conjugate protein, one that has a longer linker sequence connecting the fluorescent tag and the fungal protein sequences. This may be true seeing how when we expressed mVenus and mCerulean3 by themselves as negative controls, we saw large amounts of fluorescence produced, indicative of a high protein yield.
Following on from this, perhaps it is best to heed the advice given by our academic consultants (see Human Practices), stating that it is best to use a model eukaryotic fungus like Saccharomyces cerevisiae. In doing so, we may account for post-translational modifications or specific folding that occurs in the presence of chaperone proteins. This may be a focus in the future cycle to generate usable amounts of the Pst_12806 protein to assay its binding with our peptide candidates.
Lastly, one of the focal points of our experimentation revolved around the use of FRET. When we performed our FRET assay, the presence of errors in our design ensured our results were inaccurate. Here, our lack of sufficient controls proved this true, mostly due to our inexperience with such a technique. In future, by ensuring we have all the necessary parameters to complete a reliable and accurate assay, only then can we measure the binding affinity of our peptide and the fungal protein target to explore its potential as a treatment of in controlling cereal rust infections
.
.
.
.