Team:TUDelft/DesignTest

Sci-Phi 29

The Sci-Phi 29 Universal Toolkit


Inspired by iGEM’s great contribution in the creation and standardization of parts, our vision is to engineer a universal toolkit which can operate efficiently and predictably in any bacterial chassis (Adams, 2016). We envision that this will encourage iGEM teams and other synthetic biologists to use a wider range of organisms in their projects, saving time and resources in developing species-specific parts, consequently unifying the efforts of Synthetic Biology into host-independent genetic engineering (Kushwaha & Salis, 2015).


In order to create a transferable genetic toolkit whose parts present predictable and standardized behavior across a wide range of bacteria, we have addressed the two main processes in molecular biology: replication of genetic vectors and gene expression. By applying the engineering concepts of orthogonality (Liu, Jewett, Chin, & Voigt, 2018) and systems control (Ma, Trusina, El-Samad, Lim, & Tang, 2009), we aim to construct synthetic devices which decouple replication and gene expression levels from the varying conditions within different bacteria.


We thus have investigated the Φ29 replication system (Blanco & Salas, 1985) for in vivo orthogonal replication to create a universal vector to replicate independently of the host’s replication machinery. We have also engineered an orthogonal incoherent feed-forward loop to achieve stabilization of expression levels across different bacterial species .

  • Why is orthogonality so great? ﹀

    Much like how software developed for one operating system may fail to run in another operating system due to system-based specificities, genetic circuits seldom function effectively when transferred across two host organisms. Orthogonal systems are the “virtual machines” of Synthetic Biology: a network of processes which are insulated and independent from the host’s own native functions . Due to the minimal interactions between the orthogonal circuit and the rest of the cell, orthogonal systems can be highly predictable and transferrable, qualities which make them very promising in universal expression tools. Furthermore, little effect on the host means these systems can be heavily engineered to exhibit new functions (Liu, Jewett, Chin, & Voigt, 2018).


    (image explaining orthogonality)

    Orthogonal DNA replication has been adapted from a natural cytoplasmic selfish plasmid of Kluveromyces lactis and implemented in multiple species of yeast (Ravikumar, Arrieta, & Liu, 2014). However, no orthogonal replication systems have been successfully engineered in bacteria (Ma, Phan, Walsh, & Ye, 2015). We believe that establishing orthogonal replication in prokaryotes would be truly beneficial to Synthetic Biology in order to create a universal genetic vector which is truly transferable across highly divergent species. The further engineering of such a system would benefit Synthetic Biology efforts of every bacterium it works in.


    Orthogonal transcription is ubiquitous in Synthetic Biology, most prominently with the T7 bacteriophage RNA polymerase (T7 RNAP) and its corresponding promoter and terminator. The T7 transcription system has been extensively explored for heterologous expression in a wide range of organisms, as well as applied in logic circuits and cell-free systems (Wang, et al., 2018). They have also been used in universal expression tools, which allowed for effective expression in a diverse set of bacteria without the need for host-specific promoters (Kushwaha & Salis, 2015; Kar & Ellington, 2018; Liang, Li, Wang, & Li, 2018). While two of these systems rely on ribosome binding site (RBS) design for calibrating gene expression (Kushwaha & Salis, 2015; Liang, Li, Wang, & Li, 2018), Kar and Ellington demonstrated the use of different T7 promoter mutants with different strengths to regulate expression in a host-independent manner (Kar & Ellington, 2018). However, these systems are still prone to variations in translation rate which result from transferring the system between bacterial species, requiring constant recalibration for similar expression levels across organisms. To address this issue, we have designed a novel T7-based incoherent feed-foward loop.


    Orthogonal translation can be separated into two processes. The first consists of orthogonal translation initiation, with the engineering of orthogonal 16s rRNA (o-ribosomes) and orthogonal RBS sequences (Darlington, Kim, Jiménez, & Bates, 2018). The second process orthogonal translation elongation, related to codon with the engineering of orthogonal tRNA-aminoacid synthetases and orthogonal tRNAs (Chin, 2014). Unfortunately, even orthogonal translation initiation, which is far simpler to implement, is not appropriate for universal expression tools: o-ribosomes and o-RBSs are designed by computationally predicting Shine-Dalgarno and anti-Shine-Dalgarno sequences which have minimum cross-talk with the host’s translation machinery, making it necessary to re-design these tools for every new organism (Chubiz & Rao, 2008). Fortunately, with our designed incoherent feed-foward loop, it is possible to insulate expression levels from inter-species translational variations without the use of orthogonal translation.



Host Independent Replication and the Phi 29 system


Mobile genetic elements will continue to be essential tools for Synthetic Biology: their flexible interchangeability across strains is crucial to reach a new multi-chassis genetic engineering paradigm (Jain & Srivastava, 2013).


One of the main constraints in genetically engineering novel and unconventional organisms is the need to identify vectors that work in these organisms. Natural broad host range plasmids can be replicated in a wider range of bacteria, but are often shown to be limited to a few specific groups of organisms, and their considerable size and complexity results in large fitness costs. Thus, substantial investigative efforts are put in the determination of host range and reduction of fitness costs of these vectors (Shintani Masaki and Suzuki, 2019).


Our team sees the potential of orthogonal systems in overcoming these issues. We thus have taken a bottom-up approach of engineering an independent DNA replication system to in live bacterial cells. We have turned to the Bacillus subtilis double-stranded DNA bacteriophage known as Φ29 (Blanco & Salas, 1985). A system such as this can be theoretically applied to any bacterium species as well as any other organism, once it only requires the expression of its own replication machinery (Liu, Jewett, Chin, & Voigt, 2018).


The Φ29 replication system has been shown to efficiently amplify DNA sequences in liposome synthetic cells with only 4 relatively small proteins in synthetic liposome cells (Nies, et al. 2018). The mechanisms of the system are explained visually and in text bellow (Blanco, Lazaro, De Vega, Bonnin, & Salast, 1994):


The linear, double-stranded DNA is flanked by two origin of replication sequences, named OriL and OriR. The terminal protein (TP) is linked to the 5’ end of each strand by a aminoacid/nucleotide covalent bond, which protects the DNA from exonucleases.




In initation, multiple double-strand binding proteins (DSB) bind along the origin sequences, changing the conformation change in the DNA and cause opening of the DNA duplex. The TP, forming a heterodimer with the single-subunit DNA Polymerase (DNAP) acts as a primer by recognizing the origins of replication. DNAP then catalyzes a covalent bond between the hydroxyl group of aminoacid Ser-232 of TP with an AMP molecule.


In elongation, DNAP is dissociates from TP and replicates DNA with high processivity, demonstrating helicase activity of strand displacement. Single-strand binding proteins (SSB) binds to the displaced single-strand for stabilization. During DNA synthesis, DNAP displaces both DSB and SSB proteins form double-strands and previously displaced single strands, respectively. When polymerization is completed, the DNAP dissociates from the DNA resulting in termination.


Reaching out to experts in orthogonal replication, we have come to learn that overexpression of the 4 proteins would not be sufficient for establishing the system in vivo, due to possible unintended interaction of the machinery with the host’s DNA. We consequently set out to find optimal concentrations of these proteins in E. coli cells. Furthermore, we explored the cell-free replication of our constructed linear DNA sequences to assess the functionality of replication the system when modified to our particular application.


  • Why do we use the phi 29 replication system? ﹀
    • With only 4 fairly small proteins and 2 replication origin sequences, the Φ29 replication machinery is quite compact and presents less burden to the cells than many replisomes (Nies, et al. 2018).
    • Φ29 is the most studied and understood double-stranded DNA bacteriophage (Blanco, Lazaro, De Vega, Bonnin, & Salast, 1994). The existing literature facilitates further engineering of our proposed plasmid in novel technologies.
    • The Φ29 DNA Polymerase has high processivity and fidelity (Blanco & Salas, 1985).
    • The Φ29 has been demonstrated a powerfull tool for synthetic cell studies (Nies, et al. 2018). In vivo replication of the system can be better engineered based on knowledge obtained with cell-free assays. Furthermore, a system which is easily transferable between in vivo and in vitro would be incredibly useful in many applications.
    • The existing DNA-protein covalent bonds offer many possibilities to engineer the terminal proteins with functional peptide sequences (Gella, Salas, & Mencı, 2016). We envision that the unique configuration of the double-stranded, protein-capped linear replicon will be a basis for many new engineered protein-DNA complexes.
    • Orthogonal replication not only enables replication independent from the host, but the ability to engineer the orthogonal DNA polymerase’s fidelity without introducing mutations in the cell’s genome makes in vivo directed evolution a possibility (Ravikumar, Arrieta, & Liu, 2014).
    • The necessity of the 4 proteins for efficient DNA replication, as well as the TP-based DNA priming, make the Φ29 system a viable option of a biologically contained system, contributing to the safety of its biotechnological applications. (Torres, Kr, Csibra, Gianni, & Pinheiro, 2016).

iFFL and control of gene expression across organisms


Whenever a new synthetic system is developed in a model organism like Escherichia coli, its implementation in another bacterium often requires complete rewiring and re-tuning with parts that function analogously in the new organism. This incompatibility arises from interspecies variations (Adams, 2016), such as copy number of plasmids (De Gelder, Ponciano, Joyce, & Top, 2007), transcription rates of promoters (Meysman, et al., 2014), translation initiation rates of ribosome binding sites (RBS) (Omotajo, Tate, Cho, & Choudhary, 2015) and the codon usage of coding sequences (Sharp, Bailes, Grocock, Peden, & Sockett, 2005).


Applying the concept of orthogonality, we approached the issue with the use of the T7 bacteriophage RNA polymerase (T7 RNAP), the most commonly used orthogonal transcription system. This system has been shown to enable host-independent transcription in a wide range of organisms (Wang, et al., 2018).


For host independent expression, we have investigated the implementation of autonomous T7 RNAP regulation in a portable expression system consisting of a Mixed Feedback Loop (MFL), known as Universal Bacterial Expression Resource (UBER) (Kushwaha & Salis, 2015). UBER and similar systems have are quite powerful tools as they do not depend on host-specific transcriptional elements (Kushwaha & Salis, 2015; Kar & Ellington, 2018; Liang, Li, Wang, & Li, 2018). However, UBER does not achieve reproducible expression levels across the organisms it is implemented in (as Rice iGEM (https://2018.igem.org/Team:Rice/Results#Tx) team 2018 has demonstrated), and changes in translation rates of the genes occur when transferring the system.


  • What are portable expression tools? Where do they fail? ﹀
    The mixed feedback loop of the UBER circuit consists of a TetR repressor under a T7 promoter, regulating expression of T7 RNAP, under an engineered T7 promoter with TetO operators. While the self-transcribing of the T7 RNAP accumulates in a positive feedback loop, the accumulation of TetR represses T7 RNAP transcription in a negative feedback loop. Control of T7 RNAP levels prevents cytotoxicity caused by overexpression of the high-processivity protein (Dubendorfft & Studier, 1991).


    The levels of TetR and T7 RNAP are calibrated with the use of different computationally designed RBSs (Salis, Mirsky, & Voigt, 2009) in both genes, resulting in different levels of T7 RNAP transcription rates for each calibration (Kushwaha & Salis, 2015).


    (image)

    Unfortunately, as is predicted in the RBS Calculator 2.0 (https://www.denovodna.com) (Salis, Mirsky, & Voigt, 2009), RBS strength is a species-specific parameter, thus recalibration of both genes as well as the genes of interest becomes necessary when transferring genes across certain bacteria.


    Other portable expression systems partly overcome this issue: the use of antisense RNA for regulation of T7 RNAP removes the need to also calibrate the TetR RBS to maintain (Liang, Li, Wang, & Li, 2018). The use of different T7 promoter mutant variants with different strengths allows the host-independent tuning of transcription of the gene of interest (Kar & Ellington, 2018). However, in these cases translation-based changes of expression will still occur in both the T7 RNAP and gene of interest levels. Therefore, in order to achieve gene expression levels independent of translational variations, we have designed a T7-transcribed incoherent feed-foward loop with same RBS in the repressor and gene of interest.



Further investigation of systems control engineering concepts has led us to the Incoherent Feed-Forward Loop (IFFL), a simple three-node network which was predicted to achieve perfect adaptation to input variations (Ma, Trusina, El-Samad, Lim, & Tang, 2009).


(iFFL diagram)

As observed above, whenever input (I) goes up or down, the output gene (O) changes along with it. However, we can add a repressor (R) that is affected in the same way by the input. When the input increases, the repressor rises in level and the expected increase in the output is counteracted by higher inhibition. When input decreases, repressor is lowered, and the expected decrease in the output is compensated by lower repression.


Segall-Shapiro and colleagues have implemented this circuit in order to achieve gene expression independent of copy number, genomic location and/or growth rate. Adaptation to such variations can be achieved by regulation with a strong repressor with uncooperative binding such as TALE proteins (Segall-Shapiro, Sontag, & Voigt, 2018).


(image/animation of stabilized system)

Any variation in plasmid copy number that would alter expression of the gene of interest also affects the repressor’s expression, which has an inverse effect on gene expression, thus compensating for that variation. (DROPDOWN EXPLAIN systems control, iFFL and TALE proteins?)


Through modelling predictions (link), we saw the opportunity to expand the control of gene expression to other factors that vary across organisms by modifying the system in multiple ways. In summary, what we discovered is that as long the ratio of expression levels between the (unrepressed) output gene and the repressor are kept the same between organisms, we can maintain control of the output gene


(image of improved stabilized system)

T7 promoter variants were used to drive transcription of both TALE and gene of interest. By using our IFFL in expression strains (inducible T7 RNAP expression in their genome) or portable systems like UBER, we can implement the system across different species.


The key advantages of having an orthogonal IFFL are two: firstly, we have predicted and experimentally verified that variations in T7 RNAP concentration will yield the same transcription of the gene of interest, since we can assume the ratio of transcription between two T7 promoter variants is kept the same even across organisms, having the same percentage change in both repressor and output gene. Secondly, the ever-increasing library of T7 promoter variants allows for precise host-independent tuning of gene expression by exchanging T7 promoters of both repressor and output. Moreover, we have demonstrated promoter strength independence of our IFFL by exchanging both promoter variants


We have used the same RBS in both repressor and output gene in order to maintain the same ratio of translation initiation between them. The RBS calculator 2.0 predicts that RBS strength across species only changes due to differing anti-Shine Dalgarno sequences, while other factors related to strength (such as mRNA secondary structures) do not change in different bacteria (Salis, Mirsky, & Voigt, 2009). Consequently we can assume that: although the same RBS yields different translation initiation rates when combined with different genes, these translation initiation rates change in the same way when transferred between bacteria


We have also constructed repressor and output gene (GFP) transcriptional units in an outward orientation (inverse from each other), reducing noise in GFP expression in response to readthrough transcription in terminators.


We have also engineered a broad host range promoter version of our IFFL, since we considered that our system could also be useful for less explored organisms that may simply not be compatible with T7 RNAP or other orthogonal transcription systems (Zhao, et al., 2017). For this goal, we have designed broad host range promoters which allow for expression in a wide range of organisms (Yang, et al., 2018). This approach, although not as readily tunable as with T7 promoters, is certainly a more versatile version of our system with less moving parts (no addition of portable T7 tools), which could be applied in a straightforward manner when fine-tuning of gene expression is not as crucial. We have successfully implemented this system in both Escherichia coli and Pseudomonas putida.


We have demonstrated our orthogonal expression level control system to be compatible with both orthogonal T7 expression systems and heterologous expression strains, as well as successfully engineering a broad host range promoter version which exhibited stabilization between two organisms, opening the possibility of finer tuning and further implementation of this system across non-canonical organisms.


Sci-Phi 29 Toolkit and the modular cloning of host-independent systems


We share with other iGEM teams (link to collab) the belief that Golden Gate Cloning is currently the most flexible and convenient technique for a standardized Synthetic Biology. It enables parallel one-pot assembly of full transcriptional units while not requiring the ordering of primers, as parts are interchangeable and construction is highly modular. This type of assembly fits perfectly with our and many other iGEM projects, where a multitude of genetic circuits with single-part changes can be easily constructed, enabling high-throughput assembly and characterization of parts.


We have designed our collection of parts to be Type IIS compatible, and all our genetic circuits were assembled by using the original MoClo Toolkit (Weber, Engler, Gruetzner, Werner, & Marillonnet, 2011; Werner, Engler, Weber, Gruetzner, & Marillonnet, 2012). Although being an older method of Golden Gate cloning, it was a great choice for our project. The toolkit presents a large collection of plasmids which provide a lot of options for assembly of tailor-made genetic circuits. Furthermore, the platform is also quite useful in the construction of Φ29 replisome-based plasmids with the one pot assembly of up to 7 transcriptional units at a time. However, MoClo also allows for the easy exchange of our iFFL circuitry between different backbones with its designed DraIII restriction sites in order to test it in other organisms, besides presenting both ColEI and the RK2 origins of replication for cloning in a broad host range.


Exploring multiple features of the MoClo Toolkit, we have developed a collection of essential parts for cloning of host-independent circuits. The collection consists of 21 novel parts, with 4 T7 promoter variants, 4 broad host range promoters, 7 riboinsulator and 5’ UTR parts, 5 coding sequences, 1 terminator, and 2 linear plasmid origin sequences, as well as using multiple parts already in the registry. Although the collection may not be extensive, our project was designed to allow quick expansion of our parts.


We also have expanded Golden Gate assembly in two ways: the modular construction of 5’ UTR parts by combining different ribozyme-based insulators and ribosome binding sites; and a method to make the two most used Golden Gate strategies compatible. With these novel techniques, we have done our part in expanding the usefulness, customizability and interchangeability of the Golden Gate strategies for further use by synthetic biologists.



References