Team:Hong Kong HKU/Poster

Beyond Unicellular: Recombinase circuits to develop monoculture heterogeneity with precise ratometric control

Abstract

Some natural populations of unicellular organisms exhibit uniquely high levels of cooperativity and labor division by sustaining phenotypic variation, which allows them to handle complex tasks efficiently. Similarly, achieving high degrees of compartmentalization in monocultures of engineered cells can bring tremendous benefits to multiple applications of synthetic biology, including healthcare, industrial use, and environmental remediation. Our project aims to develop a versatile set of tools that can help create and maintain stable phenotypic variation in cellular monocultures using bacteriophage-derived Cre recombinases, which recognize and invert a loxP-flanked genetic sequence in a stochastic manner. We introduced and assessed different mutations in the Cre enzyme to improve its accuracy and conducted pilot experiments for establishing ratiometric control of cellular phenotypes. To confirm the experimental phenotypic ratios, we built computational simulations that successfully predicted the phenotypic ratios and assessed phenotypic populations' stability and convergence with the coefficient of variation.

Project Background

Living systems are well known for achieving complex tasks by separating them between individual organisms on a population level. Our civilization has been able to achieve such a tremendous level of progress only because humans have decided to separate tasks and cooperate with each other. The goal that our team has put forward this year was to establish similar degree of task compartmentalization between cells to make any engineered cell population to be better at what it has been designed for. Ironically, while working on this project, we have realized that as a team we also had to 'divide and conquer' different tasks that iGEM has assigned to us, so we decided to design our poster to represent how our team members compartmentalized our work to develop the project.

Application Examples

The segregation of pathways has been proved to enhance the yield of protein production, especially in cases of severe metabolic burden. This could be extremely beneficial in the broad range of fields, from the medical industry to the environmental remediation research. Simple examples are shown above.

However, standard approaches at population level control need tight regulation and precise tuning to work, since the systems are not versatile and prone to intrinsic noise. Quorum sensing is just one of the examples.

Typical Quorum-sensing system

We decided to take another way and introduce single-cell randomness to achieve the stable probability distribution of a built set of phenotypes in order to control the ratio of phenotypically different subpopulations.
Our aim would be to achieve precise control of cellular subpopulations, where phenotypically different cells can be kept on stable subpopulation ratios, despite their metabolic differences. Thereby, even after a long time, subpopulations would maintain stable relative ratios within a monoculture. By contrast, cells that are designed with increased metabolic burden in the normal environment would grow slower and be overtaken by cells that face a much lower metabolic stress.

Establishing and maintaining control as opposed to natural divergence of populations

Project Design

Cre recombinases have major flaws, for example, the cross-reaction with genome off-target sites which lead to loss of function of circuits via deletion. That is why we are working hard to combat the drawbacks by introducing mutation sites (R32V, R32M) published by Dr Nikolai Eroshenko and Dr George Church. With the mutation sites, we hypothesize long term stability of the system would be achievable. With the interview with Dr Eroshenko, he said with Cre mutants we may be able to achieve our goal with the Cre-loxP system.

He also suggested adding a kill switch so to screen out any accidental mutants just in case, that's why we have the following designs for kill switches.
Besides Cre-loxP, we are conducting experiments throughout next year with Hin recombinases and Bxb1 excisionases and integrases, to evaluate the efficiency and applicability of different recombinase systems.

---

Cre recombinase rules

1. All lox sites are orthogonal
2. No overlapping of lox sites
3. Every bracket inside another bracket increase in level 1

---

Multiplexing

In designing phenotypic switching models, there are other problems to our novel system with a solution, like with metabolic differences and also intermediate zones of phenotypes.
In phenotypic switching models, it is hard to achieve traditional knock-out models mediated by traditional DNA editing like CRISPR knockdowns. Therefore we are introducing the crRNA system to downregulate native genes. CMR system is a native bacterial system that works like miRNA, while cas13 could knock out RNA transiently in bacteria (less efficient), and eukaryotes. By using different crRNA arrays in different phenotypes, we can swiftly knock down different sets of native genes and hence achieve complex phenotypes.
With the crRNA system, we could also introduce conditional asymmetric flipping of lox sites by downregulating recombinase gene in certain phenotypes, as well as to decrease the intermediate zone by downregulating the other set of phenotypic genes.

Moreover, our current design is one promoter on one side. What happens if we introduce more promoters in between the lox sites; move the terminator positions; use different recombinase systems with asymmetric flipping; use different ratios to counter different metabolic burden? That is why we have different computer simulation models for different situations. And we are in progress in building a large database for every possible combination of gene elements for researchers who wished to use our system.

Experiments

In wet lab, due to covid-19 constraints, we were only able to conduct some pilot experiments due to lack of human resources. We constructed a simple reporter plasmid with inversion property, and cloned Cre recombinase construct, conducted site directed mutagenesis, verified and tested the constructs in vitro and in vivo. We were able to obtain preliminary data with positive result for maintenance of equilibrium position.

Our next steps in the future months, as second phase project, would be construct bxb1 and Hin recombinase mediated inversion circuits. Also to include different variables, like metabolic differences into our testing.

1. We successfully assembled lox-reporter constructs: loxP and lox2272.

   

   Figure 1: Verification of lox-reporter construct.

2. We successfully created Cre mutants (verified by Sanger sequencing) and assembled them with other components (e.g. pBAD and strong/ weak RBS).

   

   Figure 2: Verification of Cre mutants via sanger sequencing.

   

   Figure 3: Verification of Cre mutants' backbone change.

3. We performed in vitro Wild-Type Cre shuffling and showed the flipping event (50.547% inversion rate), instead of deletion event, on the lox-reporter construct by Cre mutants.

   

   Figure 4: Digestion check on the flipping event by wildtype Cre recombinase on loxP-reporter in vitro

4. We performed first in vivo Wild-Type Cre shuffling showed that Cre mutants have much lower gene targeting activity as well as the off-target activity, as compared to wildtype Cre.

   

   Figure 5: Digestion check on the flipping event on loxP-reporter by Cre WT and Cre mutants in vivo.

To further optimize the Cre expression level, we induced the Cre with different arabinose to evaluate the activity and stability of the cre mutants.

Figure 6. Optimisation of Cre mutant expression level. Showing restriction digestion check of Cre (co-transformed with reporter construct) activity in vivo. The bands expected were produced by reporter construct and are not present in Cre constructs.

On the left, we could observe inversion bands in R32V mutant at 0.01% but not 0.1% or 0.001% of arabinose expression. Too much Cre inhibits the inversion rate via blocking of lox sites, while too few Cre decreases the rate of reaction. We could also see the wild type Cre has no pattern in all arabinose concentration, either inverted or initial position, indicating loss of genetic information by wild type.

Table 7. R32V 0.01% arabinose flipping percentage measured via imageJ

Band Intensity	Corresponding bp	Normalized intensity	Corresponding flipping rate (flipped)
27610	2480	11.13	55%
23960	1750	13.69

After we determined the optimal arabinose concentration, we ran all mutants with wild type for digestion (Same batch and same samples).

Figure 8. Showing the digestion pattern (0.01% arabinose) of all mutants in vivo extending from figure 7.

Orange shows the wild type, which shows that the reporter construct has disappeared or integrated into the genome via nonspecific interactions (Orange). Meanwhile, the mutants retain either initial or deleted conformation. However, it is of our surprise we were able to observe inversion patterns but restriction digestion with EcoRI and PstI shows a deleted position compared to Y324F control (Green). On the right, we could observe inversion position bands (Blue), and the initial position bands (Red). While the last figure 7 and this figure 8 uses the same sample and restriction digestion, and we are able to observe both band patterns in mutants in figure 7, the DNA bands may have overlapped for the mutants in figure 8‘s gel.
From these experiments, we could conclude several things. Cre mutants have overwhelming advantages compared to Cre wild type in terms of specificity and accuracy of inversion, as we see Cre wild type has a loss of reporter construct in all in vivo experiments. Meanwhile, we could observe inversion and initial patterns in Cre mutants. However, there are still minor deletion events shown and we would improve our project by implementing our proposed solutions.

---

First is to include a kill switch for deletion mutants, since plasmids are shorter, this deletion overtakes the whole population very quickly. The deletion rate might be much lower, but the ripple effect caused would lead to quick domination of plasmid population in the cell.

Second is to also test other recombinases. We have bought human and prokaryotes optimized Bxb1 integrase/excisionase systems, as well as Hin based invertases for prokaryotes to further see which recombinase systems work the best.

Modelling

Where does metabolic burden come from?

'Metabolic burden' can be defined as the 'portion of a host cell's resources in the form of energy (ATP/GTP), or raw materials such as amino acids.

When cells express foreign proteins, it often utilizes a significant amount of the host cell's resources, leading to a decrease in the growth rate of host cells. Production of more waste products like acetate can also alter host cell physiology and biochemistry.

Overproduction of polypeptides is also deleterious for the host cells.

Due to protein-associated burden, plasmid-carrying cells will have lots of selective pressure. The cells without the difficult tasks will soon spread among the population.

Kaleta and colleagues computed the metabolic burden of protein production based on human biochemical pathways.

The group represented the metabolic burden of each amino acid in terms of ATP consumption, and hence were able to calculate the metabolic burden for protein production given the amino acid sequence.

Translational errors of E. coli cells occurs at a rate of about 2 x 10 -3 to 2 x 10 -4 errors per cell per generation.

The accumulated mutation may also create batches of cells with lower metabolic burden, affecting the entire population.

How could we solve the problem?

studies point out that division of labour (DOL) is favoured over the single cell (SC) in expression of metabolic pathways with increased metabolic burden and toxicity of intermediate products

Red area = single culture favoured; blue area = division of labour favoured

---

Modelling part 2

Chaotic beginning - when bacteria with 2 different phenotypes are put together → population dominated by bacteria with less metabolic burden within 400 minutes

(tf:tr = 1:2)

(tf:tr = 1:1.1))

---

Resolving chaos with our genetic circuit:

(tf:tr = 1:2) → 3:1

(tf:tr = 1:10) → 4:1

Our system is also stable over time with less than 2% error from a predicted number of cell even when the two types of cells divide 10 times faster

(tf:tr = 10:1)

Equilibrium is still achieved overtime despite having different initial phenotypic population and with changing probability of phenotype flipping

(nf0=100; nr0=1000 (1:10), ps=0.05, tf=10, tr=20)

We can also achieve stable phenotypic ratio even when the probability of phenotype switching from F (green) to R (red) was the same as from R to F

Mutation is inevitable in natural condition → We, therefore, developed a mutation model to show conditions where deletion or point mutations happen in our system, leading to development of cells with less metabolic burden

cells with shorter division time will eventually dominate the population

---

Hence, we developed a kill switch model:

Cells that have mutated will be killed either because of the expression of a toxin, or the lack of expression of an antitoxin within a toxic environment

With a 99% confidence, we can largely delay the population growth of mutant cells and therefore maintain population stability.

Project Parts

To conduct pilot testings, our team wishes to maintain a specific phenotypic ratio of red and green fluorescence gene by going through flipping recombination events carried out by the action of Cre recombinase. To achieve this function, two types of constructs has been cloned. The recombinase/mutants constructs with different strengths, and reporter construct.

---

Reporter construct has the following parts:

1. Lac Promoter
2. Strong RBS
3. Recombinase site pairs
4. GFP reporter gene
5. Inverted RFP reporter gene

Construct for the expression of CRE/CRE mutant recombinase has the following parts:

1. pBAD promoter
2. Weak/Strong RBS
3. Mutagenised CRE recombinases

Other Recombinase systems

1. Bxb1 integrase/excisionase (Prokaryotes and eukaryotes)
2. Hin recombinase (Prokaryotes)
3. PhiC31 recombinase (Prokaryotes and eukaryotes)

Summary of our newly created parts

Human Practices

For applications of our project, we have approached multiple gene editing experts in the synthetic biology industry and also engaged the general public by conducting an online synthetic biology course. We have also held an online symposium in which all HK iGEM teams, ex-iGEM participants, and 4 guest speakers participated.

Integrated human practice:

1. We were able to receive valuable comments and advice from multiple experts in the synthetic biology field. Dr. Eroshenko, one of the co-authors of the Cre-mutant paper, provided us with some constructive suggestions, which led us to change the direction of our project.
2. We have also had a virtual meeting with the Koc team, in which we had interactive feedback sessions on each other’s projects. We also exchanged connections to approach respective personnel to improve our projects.
3. Dr. Karen Yuen, an expert in yeast genome recombination and shuffling techniques, provided us with advice in our early project development stage. Meeting with her has also opened new doors to apply our project to yeast cultures.
4. We also had an opportunity to present our project at Prof. JD Huang’s lab. This has not only taught us how to present our data more effectively and concisely but also opened up potential applications of our project.
5. Our project does not stop here. We plan to meet Marcelo from Global Yeast, which is a company that works on advanced digital fermentation. We hope that our recombinase system can help in smart fermentation and industrial yeast strains.

Synthetic biology symposium:

1. We held the first-ever HK virtual symposium on ‘Advances in Synthetic Biology’ on Aug 29. Every HK iGEM team participated to share fascinating ideas on applying synbio in different fields, such as environmental remediation and medicine. ex-iGEMers from undergraduate teams were also invited to give comments to presentations by current iGEM teams. Their advice was very rewarding and inspired us for the future development of our project. 4 experts in the synbio field, Mr Philipp Boeing, Prof. Hieke Sederoff, Mr ZhiPeng Qiu, and Prof. Angela Wu, shared their projects and perspective on synbio. We not only learnt about several top-notch current research, but also how to set up a start-up based on our projects. The Symposium has nearly reached 500 audiences.

Education and public engagement:

Advising the younger generation of scientists

1. We mentored 2 high school iGEM teams this year, the Hong_Kong_HCY 2020 team and HKIS 2021 iGEM team. In the HCY mentorship, one of our members, Kenneth, is working closely as an advisor to give suggestions to his alma mater team. The mentorship goes back to 2018. Kenneth brought back iGEM competition to his high school to provide a jumping board to university applications and research opportunities for students for his alma mater. For HKIS 2021 mentorship, he provided several weeks of basic lab training to two HKIS team members.

2. Online course on synthetic biology:

   We believe that students from all backgrounds (science or non-science) could join us to share the joy of modelling and engineering microorganisms to serve different purposes in the synbio field. Therefore, we launched the Online Synthetic Biology course for students from different schools and departments to join. Participants engaged and interacted with us in the lecture. We introduced some basic biology and synthetic biology knowledge and applications of synbio. We have also briefly touched upon the ethics of synbio. We plan to conduct more online courses in the future.

Acknowledgements

Attribution :

1. Dr. Alan Wong/ Dr.Jiangwen Zhang
2. Dr. Aixin Yan for allowing us to use her lambda red plasmid.
3. Dr. Karen Yuen for giving advice during early project development stages and contributing Yeast codon optimised Cre plasmids.
4. Professor J.D. Huang for allowing us to present in his lab meeting.
5. Dr. Nikolai Eroshenko for giving advice about recombinase gene circuits.
6. KOC, HKUST, and CityU iGEM 2020 teams, in various collaborations.
7. HKU KBSB Staff, and Lab technician Eva Chiu in helping and setting up equipment.
8. HKU faculties and CEDARS for funding

Wet lab lit review references:

1. Bonifield HR, Hughes KT. Flagellar phase variation in Salmonella enterica is mediated by a posttranscriptional control mechanism. Journal of bacteriology. 2003 Jun 15;185(12):3567-74.
2. Ackermann M. A functional perspective on phenotypic heterogeneity in microorganisms. Nature Reviews Microbiology. 2015 Aug;13(8):497-508.
3. Livet, J., et al., Transgenic strategies for combinatorial expression of fluorescent proteins in the nervous system. Nature, 2007. 450(7166): p. 56-62.
4. Eroshenko, N. and G.M. Church, Mutants of Cre recombinase with improved accuracy. Nat Commun, 2013. 4: p. 2509.
5. Stephens K, Bentley WE. Synthetic Biology for Manipulating Quorum Sensing in Microbial Consortia. Trends in Microbiology. 2020 Apr 24.
6. Farzadfard, F. and T.K. Lu, Genomically encoded analog memory with precise in vivo DNA writing in living cell populations. Science, 2014. 346(6211).
7. Friedland, A.E., et al., Synthetic Gene Networks That Count. Science, 2009. 324(5931): p. 1199-1202.
8. Sheth, R.U. and H.H. Wang, DNA-based memory devices for recording cellular events. Nature Reviews Genetics, 2018. 19(11): p. 718-732.
9. Yang, L., et al., Permanent genetic memory with> 1-byte capacity. Nature methods, 2014. 11(12): p. 1261-1266.
10. Roquet, N., et al., Synthetic recombinase-based state machines in living cells. Science, 2016. 353(6297).
11. Chiu, T.-Y. and J.-H.R. Jiang, Logic synthesis of recombinase-based genetic circuits. Scientific reports, 2017. 7(1): p. 1-13.

Dry lab lit review references:

1. Tsoi R, Wu F, Zhang C, Bewick S, Karig D, You L. Metabolic division of labor in microbial systems. Proceedings of the National Academy of Sciences. 2018 Mar 6;115(10):2526-31.
2. Kaleta C, Schäuble S, Rinas U, Schuster S. Metabolic costs of amino acid and protein production in Escherichia coli. Biotechnology journal. 2013 Sep;8(9):1105-14.
3. Seo JH, Bailey JE. Effects of recombinant plasmid content on growth properties and cloned gene product formation in Escherichia coli. Biotechnology and Bioengineering. 1985 Dec;27(12):1668-74.
4. Medina MG, Carbonell X, Villaverde A. Connection between gene dosage and protein stability revealed by a high‐yield production of recombinant proteins in an E. coli LexA1 (Ind−) background. Biotechnology and bioengineering. 2002 Jun 30;78(7):722-30.