Team:Freiburg/Design
Design
We have designed our project to be modular. At the center of it are D- amino acids, whose use in synthetic biology we wish to boost by creating a series of tools and protocols, which are interconnected yet independent at the same time.
In each subproject, we have applied one or several engineering principles, which we describe in detail here below.
Easy production and detection of D-AAs
When creating the Biobricks for the enzymes for the production and detection of D-AAs (the racemase and the D-amino acid oxidase), we considered the principle of codon optimization. Since the goal is to purify the enzymes from E. coli cells, it is important to have high expression of the proteins. Codon optimization is a very important principle to consider when designing a desired gene for heterologous expression in a certain host1,2. Even for their use inside cells, good expression levels are crucial for having the enzymes function as desired.
Since the enzymes we selected come from the yeast Rhodotorula gracilis and the archeon Pyrococcus horikoshii, we had to order the genes codon-optimized for expression in E. coli.
In vitro incorporation of a D-AA into a protein of interest
This subproject was based on the principles of simplicity, cost-effectiveness and orthogonality.
Simplicity
In the modeling community, the so-called Occam’s razor principle is well-known. If two models can explain the data equally well, the simplest model is to be preferred. This principle can be applied in synthetic biology as well. If possible, select the simplest system that can function as desired. We applied this concept when trying to set up an in vitro system for the incorporation of a D-AA into a protein of interest and thus tested first the simplest system for in vitro protein production: the PURE system.3 After realizing that we would have to overcome several hurdles to use this system, we moved on to test cell lysates for the same purpose. Also here, we first tried to implement the simplest method, thereby we tested pure lysates without the addition of purified components as typically done in the literature.
Cost-effectiveness
This principle is closely related to the previous one, but has an economical touch to it. When creating a biological synthetic system, the costs associated to it are to be kept in mind. There would be no use of engineering a system that is extremely expensive to use. Therefore, we tested if we could obtain proteins in vitro with pure cell lysates without the addition of any other components like it is often done in with this approach.
Orthogonality
This is an extremely important principle in synthetic biology. Especially when introducing synthetic circuits inside cells, it is crucial to use components that do not interfere with the endogenous cellular machineries and processes. When working in vitro, too, it is essential to separate the tasks of the components inside the reaction tube. In this project, we used an orthogonal pair of tRNA synthetase/tRNA4 to incorporate a D-AA into super folder green fluorescent protein (sfGFP) at the position of the amber stop codon 5. Like this, the other tRNA synthetases and tRNA present in the reaction tube can carry on the normal translation process, leaving the duty to read the amber stop codon and to incorporate there the D-AA to the orthogonal pair.
In vivo incorporation of a D-AA into a protein of interest
In this subproject we used the principles of orthogonality, simplicity, modularity and compartmentalization.
Orthogonality
As explained above, we used an orthogonal tRNA synthetase/tRNA pair to read the amber stop codon in E. coli and incorporate there a D-AA instead of causing the translation to stop.
Simplicity
Genetic code expansion is possible in many organisms, such as Saccharmoyces cerevisiae, Pichia pastoris and mammalian cells. We worked with mammalian cells in other subprojects, thus we could have also performed in vivo D-AA incorporation into proteins using this model system. However, to follow the principle of simplicity, we opted for E. coli, since this is still the simplest organism to use and manipulate. Moreover, a plethora of plasmids and methods are available for protein purification out of E. coli cells. Since for this subproject we would need to purify the protein to prove the incorporation of the D-AA, this was an additional argument for selecting E. coli as model organism for this subproject.
Modularity
Modularity is another fundamental principle in synthetic biology. Modularity allows building complex systems from independent modules, which perform a specific task. When a system is modular, it is possible to easily apply changes by touching a specific module instead of the entire system. Here we used this principle and combined a strain that has been engineered to improve incorporation of non-canonical AAs 6 with a very recent method to achieve the same goal based on compartmentalisation of the orthogonal tRNA synthetase/tRNA pair and the mRNA of interest.7
Compartmentalisation
Cells have evolved compartments to carry out specific functions either to protect the rest of the cell from harmful byproducts or to protect the components of the compartment from unwanted reactions/encounters. Compartments are also helpful to increase reaction yields, as enzymes are at higher concentrations and closer to their substrates when inside a compartment. In this subproject, we decided to apply the principle of compartmentalization to improve the incorporation of D-AAs into a protein of interest. The inspiration came from a very recent paper in which this principle was applied in mammalian cells to increase the incorporation of other non-canonical AAs. By physically encapsulating the mRNA of interest with the orthogonal tRNA synthetase/tRNA pair into liquid droplets, the rate of incorporation of non-canonical AAs was shown to be significantly increased. We wished to establish this concept in E. coli cells for the incorporation of D-AAs.7
Safe expression of toxins from E. coli
In order to work with the MRSA toxin PSMα3 in an S1 safety lab, we had to chemically synthesize it and not produce it in E. coli. Indeed, a strain genetically engineered by the presence of a plasmid for the production of the toxin would be potentially harmful and would not be S1 level anymore. We wanted to find a way to bypass this issue all the while consenting the facile production of toxins from bacterial cells. To this aim, we applied the Divide et Impera (divide and conquer) approach: the toxin would be split in two parts, each harmless per se. Only when brought back together, the two parts would re-assemble the toxin, which would have its toxicity again. This principle can be applied thanks to split inteins, powerful proteins that catalyze a trans-splicing reaction whereby two independent protein fragments can be put back together with a peptide bond 8.
Mirror-image phage display to find D-peptide inhibitors of the MRSA toxin PSMα3
In order to perform mirror-image phage display we had to take several important decisions regarding which target to use, which peptide library to screen and how to perform the binding assay. We applied the principles of feasibility, compromise and diversification.
Feasibility
After deciding we would tackle Methicillin-resistant Staphylococcus aureus (MRSA), we had to select a molecular target. Because our goal was to create a therapy based on D-peptides, we first of all thought to target a protein which is secreted by the bacterium. This way, we would not have to worry about potential penetration issues of the D-peptides into the cells. Moreover, because we had to chemically synthesize the target in D-form, we had to be realistic in terms of the size of the protein we could manage to synthesize on our own. Our target should be feasible, but at the same time, truly important for fighting the toxicity of MRSA. Following this principle, we select the toxin PSMα3, which is secreted by MRSA and is relatively small (22 AAs).9
Compromise
It is well known that smaller peptides bind more specifically to a target, while longer ones are more likely to possess secondary structure features, but are prone to bind unspecifically to other targets. Longer peptides are also associated to a more diverse library, which is beneficial in the search of a binding ligand. We compromised between these two properties and selected a library of 12-AAs-long peptides.
Diversification
Phage display can be performed in solution or with the target immobilized on a surface. Each method has its own advantages and disadvantages. We applied the principle of diversification to increase the chances of finding good peptide binders for PSMα3 and performed both versions of the phage display: in solution and on surface.
Software to find D-peptide ligands to any target in silico
When writing the code of our software FinDr, we applied the common computer engineering principles for good code: modularity, versatility and user-friendliness. We additionally utilized the principle of Darwinian evolution and applied it to our in software.
Modularity
FinDr is composed of a sub-routine, the L-to-D converter, and of two modalities to find ligands to a given target. The L-to-D converter is indeed a standalone code, that users can run independently of FinDr to obtain the D-version of a protein of interest, provided a structure of the L-form is known. FinDr itself then has two modalities: one that represents the in silico mirror-image phage display and one that is a genetic algorithm. These two codes are independent from one another and can be executed separately.
Versatility
Our software can be applied to find L- and D-peptide ligands against any biomolecule as long as the 3D structure of the target is provided in PDB format. FinDr works without any limitation of protein size or type
User-friendliness
Using computational structural biology tools usually require some technical expertise. We implemented FinDr to be user-friendly. The user does not need to worry about how to handle the various input and output files of the several programs, as typical for docking and molecular dynamics simulations. FinDr is fully automated and only requires the user to provide a target PDB file and few additional parameters.
Evolution
We used this principle to either find de novo peptide binders to a target or to improve already known binders that were either computationally predicted or experimentally determined. As in natural evolution, a population of peptides undergoes mutations and recombination events during the course of the simulation. Peptides that bind to the target are selected by the algorighm and remain in the population for further directed evolution. As this process continues peptides with enhanced binding affinities towards the target emerge.
Modeling of D-proteins and of D-ligand/protein interaction
When modeling PSMα3, the peptide ligands in the in silico library and those found in the experimental mirror-image phage display, we applied the principle of realistic representation of nature. Peptides and proteins are not static entities. The side chains of the amino acids rotate in space and may change orientation with time. To represent this natural feature of the peptides and proteins studied in our project, we therefore performed molecular dynamics simulations prior to the docking simulations.
Chemical synthesis of D-proteins
Chemical synthesis of D-PSMα3 was performed following the principle of prototyping. We first synthesized L-PSMα3 to identify potential bottlenecks in the chemical synthesis. Only after having obtained good results with L-PSMα3, we proceed with the synthesis of D-PSMα3.
References
[1] Zhipeng Zhou et al. Codon usage is an important determinant of gene expression levels largely through its effects on transcription (2016). PNAS, 113 (41) E6117-E6125
[2] Novoa E.M., et al. Elucidation of Codon Usage Signatures across the Domains of Life (2019). Molecular Biology and Evolution, 36 (10): 2328–2339
[3] Shimizu Y, Inoue A, Tomari Y, Suzuki T, Yokogawa T, Nishikawa K, Ueda T. Cell-free translation reconstituted with purified components. (2001) Nat Biotechnol. 19:751–755.
[4] Chatterjee A. et al., A Versatile Platform for Single- and Multiple-Unnatural Amino Acid Mutagenesis in Escherichia coli. (2013) Biochemistry. 52(10)
[5] Ma H. et al., Genetic incorporation of d-amino acids into green fluorescent protein based on polysubstrate specificity (2015) RSC Advances. 5, 39580-39586
[6] Lajoie MJ. et al., Genomically recoded organisms expand biological functions. (2013) Science. 342(6156):357-60.
[7] Reinkemeier CD. et al., Designer membraneless organelles enable codon reassignment of selected mRNAs in eukaryotes. (2019) Science 363(6434)
[8] Li Y. et al., Split-inteins and their bioapplications (2015). Biotechnology Lett. 37 (11), 2121-2137
[9] Wang, R., et al. Identification of novel cytolytic peptides as key virulence determinants for community-associated MRSA (2007). Nature Medicine 13, 1510-1514.
