Dormancy Optimises the Efficiency of the Bacterial Solar Battery
Photosynthetic bacteria draw energy from sunlight to drive the fixation of carbon, implying that the bacteria are not able to carry out carbon fixation in absence of light. However, the bacteria constantly metabolise and the carbon fixed during time of exposure to light, will be used as an energy source for the cells. To circumvent this, a photocontrol system created by the UT Austin iGEM 2004 team was used in combination with the RelE-RelB toxin-antitoxin system native to E. coli. In this way, a light-dependent dormancy system was implemented in E. coli.


The Photocontrol Device Mediates Light-Dependent Gene Expression
Plants and several photosynthetic microorganisms, such as cyanobacteria, contain signal transduction systems, which makes them capable of reacting to light Bussell AN, Kehoe DM. Control of a four-color sensing photoreceptor by a two-color sensing photoreceptor reveals complex light regulation in cyanobacteria. Proceedings of the National Academy of Sciences of the United States of America. 2013;110(31):12834-9.. This ability gives the organisms the opportunity to adapt and optimise the regulation of their metabolic rate in response to sunlight. This property is achieved by photoreceptors incorporated in their plasma membrane, of which phytochromes are the most abundant and well described Vierstra RD, Davis SJ. Bacteriophytochromes: new tools for understanding phytochrome signal transduction. Seminars in cell & developmental biology. 2000;11(6):511-21..
Several two-component signal transduction systems evolved in E. coli enables it to respond to various external conditions, such as osmotic stress, lack of metabolites and other external stress factors. Nothing indicates that light initiates such a two-component signal transduction pathway in wild type E. coli Alper T, Gillies NE. The relationship between growth and survival after irradiation of Escherichia coli strain B and two resistant mutants. Journal of general microbiology. 1960;22:113-28.. The UT Austin iGEM 2004 team applied the light sensing property of phototrophs to an E. coli. By aligning different phytochromes with the intrinsic kinase EnvZ from E. coli they revealed a way to create a two-component system consisting of a photoreceptor with an intracellular indigenous regulator system found in E. coli. By establishing this system the bacteria acquired the ability to respond to red light Levskaya A, Chevalier AA, Tabor JJ, Simpson ZB, Lavery LA, Levy M, et al. Synthetic biology: engineering Escherichia coli to see light. Nature. 2005;438(7067):441-2.. The photoreceptor from phytochrome known as Cph1 was isolated from the cyanobacteria Synechocytis PCC6803. Cph1 has a fusion site, which can be used to combine it with the kinase EnvZ, from the EnvZ-OmpR kinase-regulator system, to form a two-domain receptor known as Cph8. The chromophore phycocyanobilin (PCB) absorbs light in the red region with maximal absorbance at 662 nm Lamparter T, Esteban B, Hughes J. Phytochrome Cph1 from the cyanobacterium Synechocystis PCC6803. Purification, assembly, and quaternary structure. European journal of biochemistry. 2001;268(17):4720-30.. When heterogeneously expressed in E. coli, it can, in combination with the light receptor Cph8, be used to form a light-sensitive circuit, making E. coli able to respond to red light abor JJ, Levskaya A, Voigt CA. Multichromatic control of gene expression in Escherichia coli. Journal of molecular biology. 2011;405(2):315-324. .
In situations where no red light is present, the photoreceptor PCB activates the phytochrome Cph1, thereby promoting kinase activity through the EnvZ kinase, illustrated in figure 1. When the transcription factor OmpR is phosphorylated by EnvZ, expression of genes controlled by the OmpR-regulated promoter is initiated. Excitation of the PCB by red light, results in a situation, where EnvZ will not be able to phosphorylate the transcription factor OmpR. The lack of phosphorylated OmpR leads to no activation of the OmpR-regulated promoter, thereby preventing gene expression by this promoter.


RelE and RelB Comprise a Toxin-Antitoxin System in E. coli
A toxin-antitoxin system is a system composed of two gene products, of which one specifies a cell toxin and the other an antitoxin, which neutralises the toxic effect caused by the toxin. In E. coli K-12 the cytotoxin RelE and antitoxin RelB comprise such a system Gotfredsen M, Gerdes K. The Escherichia coli relBE genes belong to a new toxin-antitoxin gene family. Molecular microbiology. 1998;29(4):1065-76.. Expression of the cytotoxin RelE inhibits translation in the cells, due to its ability to cleave mRNA found in the A-site of the ribosome Pedersen K, Zavialov AV, Pavlov MY, Elf J, Gerdes K, Ehrenberg M. The bacterial toxin RelE displays codon-specific cleavage of mRNAs in the ribosomal A site. Cell. 2003;112(1):131-40.. RelB neutralise the toxic effect of RelE through interaction between the two proteins. In situations of amino acid starvation, it is appropriate for the bacteria to halt the translation in order to avoid errors owing to absent amino acids. Consequently, one of the exciting factors for the expression of RelE is conditioned by amino acid starvation Pedersen K, Zavialov AV, Pavlov MY, Elf J, Gerdes K, Ehrenberg M. The bacterial toxin RelE displays codon-specific cleavage of mRNAs in the ribosomal A site. Cell. 2003;112(1):131-40..
Whether the cell lie dormant in response to expression of RelE depends on the ratio of RelB and RelE present in the cell. Several studies have shown that RelB RelE form a complex with RelB:RelE stoichiometry of 2:1 Overgaard M, Borch J, Gerdes K. RelB and RelE of Escherichia coli form a tight complex that represses transcription via the ribbon-helix-helix motif in RelB. Journal of molecular biology. 2009;394(2):183-96.Overgaard M, Borch J, Jorgensen MG, Gerdes K. Messenger RNA interferase RelE controls relBE transcription by conditional cooperativity. Molecular microbiology. 2008;69(4):841-57., When the RelB:RelE stoichiometric-ratio is lowered to 1:1, studies show that RelB is not able to protect the cells against the RelE-caused translational inhibition Overgaard M, Borch J, Jorgensen MG, Gerdes K. Messenger RNA interferase RelE controls relBE transcription by conditional cooperativity. Molecular microbiology. 2008;69(4):841-57.. To prevent free RelE circulating and discharging toxic effects in the cells under favorable conditions, studies in vivo have shown that RelB is present in 10x higher concentrations than RelE Overgaard M, Borch J, Gerdes K. RelB and RelE of Escherichia coli form a tight complex that represses transcription via the ribbon-helix-helix motif in RelB. Journal of molecular biology. 2009;394(2):183-96.. The heterologous induction of RelE could cause dissonance in the RelB:RelE ratio leading to serious consequences for the cells Christensen SK, Mikkelsen M, Pedersen K, Gerdes K. RelE, a global inhibitor of translation, is activated during nutritional stress. Proceedings of the National Academy of Sciences of the United States of America. 2001;98(25):14328-33.. The bacteria are not killed when RelE is present in abundance, but high expression of the RelE gene makes awakening of the bacterial cells a challenge Tashiro Y, Kawata K, Taniuchi A, Kakinuma K, May T, Okabe S. RelE-Mediated Dormancy Is Enhanced at High Cell Density in Escherichia coli. J Bacteriol. 2012;194(5):1169-76.. Hence, introducing a toxin to cells in a successful manner constitutes a challenge.

The Gillespie algorithm is a way to calculate the evolution of stochastic functions; in this case cell concentrations. To use the algorithm, two things are required:

1. Reaction rates of the system at a given configuration.
2. A random number generator.

For each time step, two things are calculated using the random number generator:

1. The time before next reaction.
2. Which reaction occurs.

The time before next step is given by:

Where S is the sum of the reaction rates and r₁ is a random number between 0 and 1. This gives the time, as if the system was one reaction with reaction rate S, using the random number to give an exponential distribution.
The reaction is chosen proportionally to each individual reaction rate using another random number, where reactions with high rates will occur most frequently.
As each reaction is carried out, the new time is the sum of the previous and added reaction times. These calculations are carried out until the time reaches the wanted limit, or a specific number of reactions has occurred. It is necessary to have a limit on the number of reactions, as it elsewise is possible for the calculations to continue indefinitely.

RelE and RelB Regulate Dormancy and Influence Their Own Expression

Modelling of the effects of different RelE and RelB expression levels were performed as an important aspect in the implementation of the RelE-RelB toxin-antitoxin system. The toxin RelE constrains bacterial growth by mRNA degradation, thereby inhibiting translation, whereas the antitoxin RelB inhibits this toxic effect by forming complexes with RelE. As seen in Figure 1, three different protein complexes are formed, namely RelB₂, RelB₂RelE, and RelB₂RelE₂, containing zero, one, and two RelE molecules respectively Guang-Yao Li, Yonglong Zhang, Masayori Inouye, Mitsuhiko Ikura, Structural Mechanism of Transcriptional Autorepression of the Escherichia coli RelB/RelE Antitoxin/Toxin Module, In Journal of Molecular Biology, Volume 380, Issue 1, 2008, Pages 107-119, ISSN 0022-2836.

The expression of both RelE and RelB is regulated by the relBE promoter, which is influenced differently by each of the complexes, as seen in Figure 2. When small amounts of RelE is present, RelB₂ and RelB₂RelE repress transcription through relBE by binding to the operator sequence. However, when high amounts of RelE are present, the toxin mitigates this repression by reacting with complexes bound to the operator sequence Cataudella I., Trusina A., Sneppen K., Gerdes K., Mitarai N. Conditional cooperativity in toxin-antitoxin regulation prevents random toxin activation and promotes fast translational recovery. Nucleic Acids Res. 2012;40:6424–6434. doi: 10.1093/nar/gks297.

During starvation, the half-life of RelB decreases significantly due to a Lon-proteaseOvergaard, M., Borch, J., Jørgensen, M. G. and Gerdes, K. (2008), Messenger RNA interferase RelE controls relBE transcription by conditional cooperativity. Molecular Microbiology, 69: 841–857. doi:10.1111/j.1365-2958.2008.06313.x, causing a shift in the equilibrium of RelB and RelE to a higher level of RelE. In a non-starvation situation, the interactions with the operator sequence keeps the amount of free RelE at a low level, thereby stabilising the system Cataudella I., Trusina A., Sneppen K., Gerdes K., Mitarai N. Conditional cooperativity in toxin-antitoxin regulation prevents random toxin activation and promotes fast translational recovery. Nucleic Acids Res. 2012;40:6424–6434. doi: 10.1093/nar/gks297. In our simulation, the shift in equilibrium is made by introducing additional expression of RelE.
Two different models were used with two different approaches:

1. How a given configuration of RelB and RelE production increases the RelE concentration and whether it could induce dormancy within 2 hours.
2. The time required for the bacteria to exit dormancy for each of the configurations, that is, how long it takes the levels of free RelE to decrease again.

Rates and Reactions

The skeleton of toxin-antitoxin system inherent to E. coli in the model was based on the study by Cataudella et al. 2012 Cataudella I., Trusina A., Sneppen K., Gerdes K., Mitarai N. Conditional cooperativity in toxin-antitoxin regulation prevents random toxin activation and promotes fast translational recovery. Nucleic Acids Res. 2012;40:6424–6434. doi: 10.1093/nar/gks297. In an E. coli cell with a size of 1-2 μm, each nM of concentration can be approximated to 1 molecule. Thus all units are converted to molecules^-1, as this fits the premises of the Gillespie algorithm. To simplify the model, the high affinity of RelE and RelB was used to ignore single RelB and only consider the dimer, RelB₂Cataudella I., Trusina A., Sneppen K., Gerdes K., Mitarai N. Conditional cooperativity in toxin-antitoxin regulation prevents random toxin activation and promotes fast translational recovery. Nucleic Acids Res. 2012;40:6424–6434. doi: 10.1093/nar/gks297UniProtKB - P0C079 (RELB_ECOLI). Thus, all mentions of RelB in the model refers to its dimer.
Whilst RelB has a relatively low half-life at about 3-5 minutesOvergaard, M., Borch, J., Jørgensen, M. G. and Gerdes, K. (2008), Messenger RNA interferase RelE controls relBE transcription by conditional cooperativity. Molecular Microbiology, 69: 841–857. doi:10.1111/j.1365-2958.2008.06313.x, RelE is rather stable and its half-life, here set to 43 min, is primarily an effect from dilution caused by the bacterial growthCataudella I., Trusina A., Sneppen K., Gerdes K., Mitarai N. Conditional cooperativity in toxin-antitoxin regulation prevents random toxin activation and promotes fast translational recovery. Nucleic Acids Res. 2012;40:6424–6434. doi: 10.1093/nar/gks297. However, when growth is restricted during dormancy, the half-life of RelE is increased to 2000 min, corresponding to approximately one day, as the dilution effect is no longer applicable. As the protein complexes are relatively stable, their half-life was set to the same as RelE. However, for RelE to become active in the inherent system under starvation, RelB in complexes must decayCataudella I., Trusina A., Sneppen K., Gerdes K., Mitarai N. Conditional cooperativity in toxin-antitoxin regulation prevents random toxin activation and promotes fast translational recovery. Nucleic Acids Res. 2012;40:6424–6434. doi: 10.1093/nar/gks297, for which reason the rate was set to a fourth of free RelB.
The transcription rates of RelE and RelB are based on the concentration of RelE and RelB under stable conditions where RelB is ten times more prevalent than RelEOvergaard M., Borch J., Gerdes K. RelB and RelE of Escherichia coli form a tight complex that represses transcription via the ribbon-helix-helix motif in RelB. J. Mol. Biol. 2009;394:183–196. doi: 10.1016/j.jmb.2009.09.006. Consequently, RelB has been given a transcription rate 100 times higher than RelE to make up for the higher half-life of RelE.
For the full set of reactions, read here
Each bacterial chromosome has two operators, each of which can bind one RelB₂ dimer (O(RelB₂)) or either one or two RelB₂RelE complexes (O(RelB₂RelE) and O(RelB₂RelE)₂)Overgaard, M., Borch, J., Jørgensen, M. G. and Gerdes, K. (2008), Messenger RNA interferase RelE controls relBE transcription by conditional cooperativity. Molecular Microbiology, 69: 841–857. doi:10.1111/j.1365-2958.2008.06313.x. Each cell is assumed to have four chromosomes with one relBE promoter each, as this is an average number of chromosomes for an exponentially growing E. coli cellOvergaard, M., Borch, J., Jørgensen, M. G. and Gerdes, K. (2008), Messenger RNA interferase RelE controls relBE transcription by conditional cooperativity. Molecular Microbiology, 69: 841–857. doi:10.1111/j.1365-2958.2008.06313.x. This was found to stabilise the inherent system considerably compared to a system containing one chromosome per cell, as the systems exhibited similar behaviour but with different amounts of noiseCataudella I., Trusina A., Sneppen K., Gerdes K., Mitarai N. Conditional cooperativity in toxin-antitoxin regulation prevents random toxin activation and promotes fast translational recovery. Nucleic Acids Res. 2012;40:6424–6434. doi: 10.1093/nar/gks297Overgaard, M., Borch, J., Jørgensen, M. G. and Gerdes, K. (2008), Messenger RNA interferase RelE controls relBE transcription by conditional cooperativity. Molecular Microbiology, 69: 841–857. doi:10.1111/j.1365-2958.2008.06313.x. The initial values in the model are listed below.

All values are integers as the Gillespie algorithm works with discrete numbers of molecules. The values were chosen based on a stable equilibrium found by letting the model run a simulation of the inherent system over 450 minutes with different starting values.
Ranging from 1-350 molecules per minute, the implemented expression rates of RelE and RelB in the model might seem too high, as the rates in the inherit system is effectively around 80-100 for RelB and 2-5 for RelE. However, the possibility of placing the relE and relB genes under regulation of controllable promoters makes the high total production values reasonable.
When the inherent toxin-antitoxin system is activated under starvation, 40-70 molecules of free RelE are found in each cell, making it reasonable to believe that the cells enter dormancy when a few tens of free RelE copies are present. This result obtained from the model is in agreement with literatureCataudella I., Trusina A., Sneppen K., Gerdes K., Mitarai N. Conditional cooperativity in toxin-antitoxin regulation prevents random toxin activation and promotes fast translational recovery. Nucleic Acids Res. 2012;40:6424–6434. doi: 10.1093/nar/gks297
For the full list of constants, see the attached Table 1 at the bottom of this page.

Running the Model

To give a stochastic view of the system, the Gillespie algorithm was run in the computer programming language MATLAB, utilising an implementation made by MATLAB user Nezar.
For the runs simulating dormancy, deterministic initial values were used and the system was run for 30 minutes without activation of the inserted toxin promoter. This resulted in a stochastic distribution of initial values mimicking variations between cells. Analysis showed, that 30 minutes was enough for the model to find a stable distribution, which is realistic considering the growth cycle of an E. coli cell.
For the runs simulating activation, the data generated at the end of a dormancy run was used as initial value and deactivated expression of RelE. When the concentration of free RelE decreases to below 15 copies, a cell is considered active. This value was probably set too low, but tests displayed marginal difference between 15 and 45 copies, where the lower limit was chosen to decrease uncertainty of the cell state.
All runs simulated 1000 cells, which should be sufficient to get stable averages and the model assumed well-mixed conditions in every cell and considered each cell independently. Furthermore, the model has no cut-off for maximum values of RelE, as the exact relation between RelE concentration and dormancy state is unknown, yet a functional cut-off was found through activation times.
Download of Matlab-scripts

Table 1

Results

As the bacteria only require a few tens of RelE molecules to enter dormancy, the threshold was placed at 40 copies to allow for lag in activation. This is of course an oversimplification, but it is not a problem that activation and dormancy are defined at different levels, since the exact number of RelE molecules required to induce dormancy is unknown.

The Impact of Promoters with Different Production Rates on RelE Expression
During the implementation of RelE, gene expression was simulated for promoters with different strengths, which were chosen through an iterative process. Promoters with production rates of 3.5 and 10.5 molecules per minute, both induce dormancy rather slowly, with the latter inducing dormancy in approximately 50 minutes. In cells cloned with a promoter producing 35 molecules per minute, the cells will enter dormancy in about 10 minutes, while promoters producing 105 and 350 molecules per minute both have a negligible timeframe. In the simulated dormancy system, the three strongest promoters exhibited similar results, indicating that a certain threshold value had been transcended. This means, that not only was the gene expression disproportional to the promoter strength, but the risk of overshooting was also increased tremendously.

RelB is Required for Activation of Bacteria after Dormant State
If RelB was not expressed, the bacteria remained dormant for hours after RelE production had ceased, making it clear, that production of the antitoxin RelB was necessary for activation. Considering the stability of RelE in non-growing conditions, it was not surprising to find that RelB production would be the primary element in sequestering free RelE.

Appropriate Ratio of RelE and RelB Expression is Essential
Different expression rates of RelB were combined with production rates of RelE at 35 and 100 molecules per minute, corresponding to relatively medium and strong expression levels respectively. This revealed that variation in RelB had a higher impact on the time required before the dormant state was reached when the RelE production is lowered. Out of the established configurations, the RelB₂:50-RelE:35 configuration showed promising results. Compared to the RelB₂:35-RelE:35 configuration, where only few bacteria reenter an active state within the modelled time set to 2.5 hours, the RelB₂:50-RelE:35 configuration revealed a high sensitivity to the expression level of RelB. This indicated a need for the expression of RelB to be higher than RelE, but as the best results were achieved at low production rates of RelE, it is important to stringently control the expression of RelB to ensure that the bacteria are able to enter dormancy.

Discussion of Model Regarding the Artificial Dormancy System

Model Limitations
As the model only simulates the dormancy for 2 hours, not all configurations have reached equilibrium, therefore these configurations might attain a higher concentration of free RelE than modelled. This could result in a prolonged phase reentering an active state. The activation model has a rather weak predictability, as the half-life of RelE is quite high. This essentially means, that the model is unable to reduce the total amount of RelE, given by RelE_tot=RelE_free+RelE_bound, because of the short simulated timespan. Thus, in activation runs, where the number of free RelE molecules reaches a high level, the amount of RelE_bound is equally high. The problem arises, since the model works under the assumption that bound RelB is not completely stable, causing RelE-RelB complexes to dissociate, whereby RelE is freed. This causes the induced RelB expression to approach equilibrium, implying that the decrease in free RelE is rather slow. It is therefore not only the amount of free RelE that determines the activation time, but also whether the amount of dissociating complexes is high enough to counter the RelB production. Hence, it has no relevance to further test high RelE production in this model, as a high number of complexes will easily be achieved.
Light Sensitivity
One thing the model does not include, is the actual sensitivity to light. Out in the open, the amount of light is rarely an on/off switch, which means there will be periods with varying degrees of activation. Since the bacteria should be active during overcast days, the system requires a threshold, both in the sensitivity of the light-regulated promoter, but also in the activation of the toxin-antitoxin system. Variation of light is implicitly modelled through variation in promoter strength, for instance a half probability of activation translates roughly to a half production rate in the individual bacteria. Because of this, it is important not only to find functioning configurations, but also to investigate the closely related configurations, so that the bacteria neither lay dormant at overcast days, nor make the dormancy system obsolete in moonlight.

Approach

Balancing Bacterial Dormancy Requires Accurate Regulation of the System
The genes needed for inducing dormancy when the bacteria are not exposed to light, are found in the photocontrol device part, BBa_K519030. This part is composed of three genes named ho1, pcyA, and cph8, all of which are essential to ensure the cells ability to respond to red light. When the photocontrol device is exposed to light, a phosphorylation cascade activates the transcription factor OmpR, which in turn induces transcription through the OmpR-regulated promoter, BBa_R0082. This system is combined with the RelE-RelB toxin-antitoxin system in the endeavour to mediate light-dependent dormancy in bacteria. In the first considered design of the light-dependent dormancy system, the aim was to clone the photocontrol device, BBa_K519030, under control of a constitutive promoter, RelE under control of the OmpR-regulated promoter, BBa_R0082, and RelB under control of a constitutive promoter, all into one high copy BioBrick assembly plasmid pSB1C3, as seen on Figure 1.

The Photocontrol Device was Placed under Control of a Constitutive Promoter
From the constitutive promoter family the weak promoter, BBa_J23114, the two medium promoters, BBa_J23106, and BBa_J23110, and the strong promoter, BBa_J23102, were tested by fluorescence microscopy to determine which one to use for expression of the photocontrol device and RelB. Overnight cultures of the submitted parts expressed in E. coli TOP10 illustrated a clear gradient of increasing red fluorescent protein (RFP) expression correlated with the strength of the promoter, as seen on Figure 2.

On the basis of the data obtained by fluorescence microscopy, the strong constitutive promoter BBa_J23102, was chosen to control the photocontrol device. For still inexplicable reasons the photocontrol device emerged difficult to clone with the strong constitutive promoter. Although the molecular cloning of these parts was optimised several times very few successful clonings were accomplished, and the few times correct assembly was obtained, the BioBrick was not reproducibly purifiable. By sequencing, it was deduced that a region of the plasmid containing both the promoter and the BioBrick prefix had vanished. To circumvent this inconvenient cloning two other promoters, that are constitutive in E. coli, were examined, namely the PenI-regulated, BBa_R0074, and the Mnt-regulated, BBa_R0073, promoters. By performing fluorescence microscopy on composite parts of the promoters controlling yellow fluorescent protein (YFP), BBa_I6102, and BBa_I6103, respectively, the expression levels were assessed. The obtained results revealed that the PenI-regulated promoter facilitated very strong expression of the marker gene, whereas the expression controlled by the Mnt-regulated promoter was noticeably lower. Based on these findings, the photocontrol device was placed under the control of the PenI-regulated promoter instead of the strong constitutive promoter from the constitutive promoter family. As it turned out, this cloning likewise emerged difficult. After several attempts, it was decided to focus on the other aspects of the dormancy system.


Regulation of the OmpR-dependent Promoter Required a Low Copy Vector
The first construct containing the genes required for the light-induced dormancy was designed as shown in Figure 1. As the conducted modelling clarified, the necessity for stringent regulation of the RelE and RelB expression, the properties of the OmpR-regulated promoter were studied thoroughly. To assess the functionality of the OmpR-regulated promoter in practice, a reporter system containing the OmpR-regulated promoter controlling RFP was cloned into the E. coli strain MG1655 ΔompR. The phenotype of the resulting cultures revealed a dysregulation of the OmpR-regulated promoter. Thorough research lead to the finding that the OmpR-dependent promoter is not controllable when cloned on a high copy vector. As the modelling revealed, and which is evident from Figure 2-Main-Page, a relatively low expression of RelE is required to induce dormancy, whereas high expression levels quickly result in overshooting. Since the OmpR-regulated promoter is an integrated part of the light sensing system, replacement is not an option. Therefore, the variability of the relE gene copy number was studied, and it was found that the OmpR-regulated promoter should be cloned into the bacterial chromosome or a low copy vector to obtain proper regulation Levskaya A, Chevalier AA, Tabor JJ, Simpson ZB, Lavery LA, Levy M, et al. Synthetic biology: engineering Escherichia coli to see light. Nature. 2005;438(7067):441-2.. This intriguing finding let to the aspiration to investigate the controllability of the OmpR-dependent promoter on vectors with different copy numbers compared to the chromosome, thereby improving the characterisation of the promoter for the benefit to future iGEM teams.
To incorporate DNA onto the bacterial chromosome, homologous recombination with the red λ recombinase is a suitable approach Thompson JF, de Vargas LM, Skinner SE, Landy A. Protein-protein interactions in a higher-order structure direct lambda site-specific recombination. Journal of molecular biology. 1987;195(3):481-93.. Using this technique, a short fragment of chromosomal DNA at the bacterial attachment site attB Groth AC, Calos MP. Phage integrases: biology and applications. Journal of molecular biology. 2004;335(3):667-78. can be replaced with a linear DNA fragment encoding the OmpR-dependent promoter, RelE, and an chloramphenicol resistance cassette. Using polymerase chain reaction (PCR), the linear DNA sequence was flanked by sequences, which are homologous to part of the chromosome. The linear DNA fragment was electroporated into bacteria containing the pKD46 plasmid, encoding the red λ recombinase Thompson JF, de Vargas LM, Skinner SE, Landy A. Protein-protein interactions in a higher-order structure direct lambda site-specific recombination. Journal of molecular biology. 1987;195(3):481-93., which mediated the recombination. The fundamental concept of this approach is illustrated in Figure 3.

Using this approach, it was attempted to assemble the OmpR-regulated promoter controlling RelE with the flanking regions by PCR. The assembly of the fragment derived from the plasmid with either one of the flanking fragments was achieved. However, the assembly of the complete chromosomal insertion fragment containing all three segments emerged problematic. Several attempts with numerous process optimisations were performed, but unfortunately without success. To circumvent this, another attachment site was chosen. The assembly of the complete fragment by PCR was achieved and chromosomal insertion by electroporation was attempted several times, but in vain. In consideration of the fact that RelE is a bacterial toxin, the cloning of this gene was inconvenient. The model indicated, that overexpression of the antitoxin RelB would bypass this difficulty. As the homologous recombination of this fragment emerged such a challenging task, it was decided to focus on the conduct of the OmpR-regulated promoter on different plasmids, as this was evidently the more convenient approach of the general implementation of the OmpR-regulated promoter.


An Inducible Promoter was Chosen to Regulate the Gene Expression of the Antitoxin RelB
Originally, it was thought to place the antitoxin RelB under a constitutive promoter of appropriate strength. However, as the modelling revealed, strict regulation of RelB is essential to counteract the toxic effect of RelE and enable a functional dormancy system. Thus, it was deduced that it would be more suitable to utilise an inducible promoter for this purpose and it was decided to put RelB under control of the LacI-regulated, lambda pL hybrid promoter, BBa_R0011. The William and Mary team found that the LacI-regulated, lambda pL hybrid promoter had a low level of noise, a measure of the variability in gene expression between cells in the population, when cloned into a low copy vector. Therefore, this regulated promoter was chosen.
To mimic the expression of RelB, a reporter system composed of the LacI-regulated, lambda pL hybrid promoter and GFP was assembled. Hereby, the appropriate concentration of IPTG required to induce the promoter on a low copy vector, was identified. During the cloning, it was discovered that the site between the LacI-regulated, lambda hybrid promoter and this particular construct under the tested conditions had formed a hotspot for transposons. In the light of this finding, another inducible promoter was chosen. The HKUST-Rice iGEM 2015 team demonstrated that induction of the AraC-regulated promoter, pBAD, caused a gradual increase in gene expression when cloned into the low to medium copy plasmid, pSB3K3, in contrast to an all-or-none behavior when cloned into the high copy vector, pSB1K3. Taking these results into consideration, the pBAD promoter was used in combination with the pSB3K3 vector to regulate the expression of the antitoxin RelB. Again, a reporter system was used to mimic the expression of RelB, as the part, BBa_I6058, containing YFP controlled by pBAD, was assessed on different vectors.


The Final Approach for the Three Components Comprised Three Different Vectors
Based on the modelling the system approaches reviewed in the preceding part, the final design, which is illustrated on Figure 4, was established. Ultimately, the dormancy system was composed of the photocontrol device controlled by the PenI-regulated promoter on a high copy vector, RelB controlled by pBAD, BBa_K2449031, on a low copy vector and RelE controlled by the OmpR-regulated promoter on either a low copy vector or the chromosome.

Carbon Fixation through the Calvin Cycle
Carbon fixation in autotrophic organisms is responsible for the net fixation of 7×10¹⁶ g carbon annually, thereby being the most imperative biosynthetic process in nature Berg IA. Ecological aspects of the distribution of different autotrophic CO2 fixation pathways. Applied and Environmental Microbiology. 2011;77(6):1925-36.. Six different autotrophic pathways for carbon fixation have been discovered in a variety of organisms Ducat DC, Silver PA. Improving Carbon Fixation Pathways. Current opinion in chemical biology. 2012;16(3-4):337-44.. The most widespread of these, is the Calvin-Benson-Bassham (CBB) cycle, found in photosynthetic eukaryotes, e.g. plants and algae, as well as in photo- and chemosynthetic bacteria B. Bowien MG, R. Klintworth, U. Windhövel. Metabolic and Molecular Regulation of the CO2-assimilating Enzyme System in Aerobic Chemoautotrophs. Microbial Growth on C1 Compounds: Proceedings of the 5th International Symposion. 1st ed. Institute for Microbiology, Georg-August-University Göttingen, Federal Republic of Germany: Martinus Nijhoff Publishers; 1987.. The Calvin cycle, as this pathway is also called, can proceed under aerobic conditions, and only three enzymes and one microcompartment involved are heterologous to the gram-negative bacteria E. coli, making this the most obvious choice for the implementation of a carbon fixation pathway. In contrast, the 3-hydroxypropionate pathway for carbon fixation would require the transfer of ten heterologous genes 1. Mattozzi M, Ziesack M, Voges MJ, Silver PA, Way JC. Expression of the sub-pathways of the Chloroflexus aurantiacus 3-hydroxypropionate carbon fixation bicycle in E. coli: Toward horizontal transfer of autotrophic growth. Metabolic engineering. 2013;16:130-9.. Furthermore, the reductive carboxylic acid cycle found in phylogenetically diverse autotrophic bacteria and archaea Hugler M, Wirsen CO, Fuchs G, Taylor CD, Sievert SM. Evidence for autotrophic CO2 fixation via the reductive tricarboxylic acid cycle by members of the epsilon subdivision of proteobacteria. J Bacteriol. 2005;187(9):3020-7. and the noncyclic reductive acetyl-CoA or Wood-Ljungdahl pathway Ragsdale SW, Pierce E. Acetogenesis and the Wood-Ljungdahl Pathway of CO(2) Fixation. Biochimica et biophysica acta. 2008;1784(12):1873-98. require strict anaerobic conditions.
The Calvin cycle involves eleven enzymes, of which eight are intrinsic to E. coli. The three heterologous enzymes are RuBisCo, SBPase and PRK. The latter phosphorylates ribulose-5-phosphate to ribulose-1,5-bisphosphate. This is the substrate for RuBisCo, which catalyses the carboxylation, whereby glycerate-3-phosphate is produced. Later in the cycle, SBPase catalyses the dephosphorylation of sedoheptulose-1,7-bisphosphate to sedoheptulose-7-phosphate, which is later converted to ribulose-5-phosphate, completing the circle. The net effect of three full cycles is the conversion of three CO₂ molecules into one molecule glyceraldehyde-3-phosphate, which can be used for energy production via glycolysis or polysaccharide biosynthesis. Separately, these enzymes have previously been heterogeneously expressed in E. coli using various donor species, such as wheat Dunford RP, Catley MA, Raines CA, Lloyd JC, Dyer TA. Purification of active chloroplast sedoheptulose-1,7-bisphosphatase expressed in Escherichia coli. Protein expression and purification. 1998;14(1):139-45., the algae Chlamydomonas sp. Tamoi M, Nagaoka M, Shigeoka S. Immunological properties of sedoheptulose-1,7-bisphosphatase from Chlamydomonas sp. W80. Bioscience, biotechnology, and biochemistry. 2005;69(4):848-51.Vira C, Prakash G, Rathod JP, Lali AM. Cloning, expression, and purification of Chlamydomonas reinhardtii CC-503 sedoheptulose 1,7-bisphosphatase in Escherichia coli. Preparative biochemistry & biotechnology. 2016;46(8):810-4., and the cyanobacteria Synechococcus Parikh MR, Greene DN, Woods KK, Matsumura I. Directed evolution of RuBisCO hypermorphs through genetic selection in engineered E.coli. Protein engineering, design & selection : PEDS. 2006;19(3):113-9..


The Carboxysome Increases the Efficiency of the Carboxylation by RuBisCo
Many photosynthesising bacteria have developed CO₂ concentrating mechanisms to increase the efficiency of the carbon fixation process. Cyanobacteria and many chemoautotrophic bacteria utilise organelle-like polyhedral bodies, that increase the internal concentrations of inorganic carbon 4000-fold compared to external levels Mangan NM, Brenner MP. Systems analysis of the CO(2) concentrating mechanism in cyanobacteria. eLife. 2014;3.. These microcompartments, called carboxysomes, appear to have arisen twice during evolution and have undergone a process of convergent evolution Rae BD, Long BM, Badger MR, Price GD. Functions, Compositions, and Evolution of the Two Types of Carboxysomes: Polyhedral Microcompartments That Facilitate CO(2) Fixation in Cyanobacteria and Some Proteobacteria. Microbiology and Molecular Biology Reviews : MMBR. 2013;77(3):357-79.. The two types, designated ɑ and β, share main structural and functional features Rae BD, Long BM, Badger MR, Price GD. Functions, Compositions, and Evolution of the Two Types of Carboxysomes: Polyhedral Microcompartments That Facilitate CO(2) Fixation in Cyanobacteria and Some Proteobacteria. Microbiology and Molecular Biology Reviews : MMBR. 2013;77(3):357-79.. The ɑ-carboxysome consists of a proteinaceous outer shell composed of six different shell proteins designated CsoS1ABCD and CsoS4AB, and encloses RuBisCo, the shell associated protein CsoS2, and the enzyme carbonic anhydrase CsoS3. On average, ~250 RuBisCo molecules are localised within each carboxysome, and these are organised into three to four concentric layers Iancu CV, Ding HJ, Morris DM, Dias DP, Gonzales AD, Martino A, et al. The structure of isolated Synechococcus strain WH8102 carboxysomes as revealed by electron cryotomography. Journal of molecular biology. 2007;372(3):764-73.. The carbonic anhydrase converts HCO₃^-, which diffuses passively into the carboxysome, to CO₂, thereby driving the continued diffusion of HCO₃^- into the microcompartment Mangan NM, Brenner MP. Systems analysis of the CO(2) concentrating mechanism in cyanobacteria. eLife. 2014;3.. The increased CO₂ concentration in the vicinity of RuBisCo increases the rate of carbon fixation by saturating the RuBisCo enzyme and increasing the CO₂ to O₂ ratio, enabling carboxylation to prevail over oxygenation Mangan NM, Brenner MP. Systems analysis of the CO(2) concentrating mechanism in cyanobacteria. eLife. 2014;3.. The shell associated protein is essential for the biogenesis of the ɑ-carboxysome Cai F, Dou Z, Bernstein SL, Leverenz R, Williams EB, Heinhorst S, et al. Advances in Understanding Carboxysome Assembly in Prochlorococcus and Synechococcus Implicate CsoS2 as a Critical Component. Life (Basel, Switzerland). 2015;5(2):1141-71.. The genes encoding the enzymes and shell proteins forming the ɑ-carboxysome from Halothiobacillus neapolitanus are clustered into the cso operon. This operon has been heterogeneously expressed in E. coli and its transcriptional regulation Cai F, Heinhorst S, Shively JM, Cannon GC. Transcript analysis of the Halothiobacillus neapolitanus cso operon. Archives of microbiology. 2008;189(2):141-50. and functionality Bonacci W, Teng PK, Afonso B, Niederholtmeyer H, Grob P, Silver PA, et al. Modularity of a carbon-fixing protein organelle. Proceedings of the National Academy of Sciences of the United States of America. 2012;109(2):478-83. has been studied.


Energy and Electrons are Required for Carbon Fixation
For the Calvin cycle to proceed, energy in the form of ATP and electrons carried by NADPH are required. In photosynthesising organisms, such as plants and cyanobacteria, these constituents are provided by photophosphorylation performed by the photosystem complex. When engineering E. coli to perform photophosphorylation, 13 genes for the biosynthesis of chlorophyll a and 17 genes for the biosynthesis of photosystem II need to be heterogeneously expressed. Several attempts have been made at expressing part of it, such as the psbA gene Efimov VA, Reverdatto SV, Beilinson BA, Fradkov AF, Chakhmakhcheva OG. [Expression of the gene coding for the D1-protein of barley photosystem II in Escherichia coli]. Bioorganicheskaia khimiia. 1994;20(5):524-35. and the 18-kDa protein of photosystem II Kuwabara T, Takeuchi M, Honda S, Nakajima N, Watanabe A, Kondo N. Expression in Escherichia coli of the extrinsic 18-kDa protein of photosystem II of spinach. Plant & cell physiology. 1995;36(3):435-9., both of which was successful.
An alternative process, in which a diverse array of phototrophic bacteria and archaea harvest energy from light, is through a retinal-containing protein, called proteorhodopsin, that catalyse light-activated proton efflux across the cell membrane and thereby drive ATP synthesis. In contrast to photosystems, the process involving proteorhodopsin is anoxygenic and generates no NADPH vital for the Calvin cycle to proceed Walter JM, Greenfield D, Bustamante C, Liphardt J. Light-powering Escherichia coli with proteorhodopsin. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(7):2408-12.. The heterologous expression of this light-powered proton pump in E. coli enabled photophosphorylation when the bacteria were exposed to light Martinez A, Bradley AS, Waldbauer JR, Summons RE, DeLong EF. Proteorhodopsin photosystem gene expression enables photophosphorylation in a heterologous host. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(13):5590-5. , and even generated a proton motive force, which turned the flagellar motor, yielding light-dependent motility Walter JM, Greenfield D, Bustamante C, Liphardt J. Light-powering Escherichia coli with proteorhodopsin. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(7):2408-12..

Approach

Engineering E. coli to Perform the Calvin Cycle
In order to engineer E. coli in the outer chamber to turn atmospheric CO₂ into cellulose, the carbon first needs to be fixated by the bacteria. This requires the heterologous expression of the genes encoding the three enzymes RuBisCo, SBPase, and PRK. In correspondence with the 2014 Bielefeld iGEM team, who worked with a similar subpart of their project, we discussed the cloning of these genes and received two crucial parts. The first of this was BBa_K1465214, containing RubisCO from Halothiobacillus neapolitanus and PRK from Synechococcus elongatus under the control of a composite promoter controllable by IPTG. The second was BBa_K1465228, which contained SBPase from Bacillus methanolicus. Through the concurrent expression of these parts in E. coli, all enzymes required to fixate CO₂ through the Calvin cycle were present in the cells.
The first approach was to assemble both parts on one plasmid by molecular cloning, as shown on Figure 1. In doing this, it was discovered that the BBa_K1465214 part was missing roughly 1 kbp when digested with standard restriction enzymes with recognition sites within the BioBrick prefix and suffix. Sequencing of the part revealed that the part was missing the entire promoter sequence of 1239 bp. Consequently, the assembly of the parts additionally required the cloning of a promoter. For this purpose, the Tac-promoter (Ptac) from the part BBa_K864400 was chosen, as this is commonly used to overexpress genes in E. coli. With the objective to place the two Calvin cycle parts on different vectors, it was attempted to clone Ptac in front of both parts. However, this did not succeed and it was decided to prioritise other aspects of the project and therefore keep this part theoretical henceforth.
Implementing the Carboxysome in E. coli to Increase Carbon Fixation Efficiency
The implementation of the microcompartment carboxysome in the test organism can increase the efficiency of the carbon fixation process substantially. As for the Calvin cycle parts, we corresponded with the 2014 Bielefeld iGEM team on their experience of the implementation of the carboxysome, and received the two parts that together contained the cso operon from Halothiobacillus neapolitanus. These parts were BBa_K1465204, containing csoS2, and BBa_K1465209, containing csoS3, csoS14, and csoS1D. As part of the optimisation, we aimed to combine these parts into one part containing the entire cso operon, as shown on Figure 1. Furthermore, a promoter was required to drive the gene expression. For this purpose, Ptac was chosen, as for the genes encoding the Calvin cycle enzymes.

We succeeded in assembling both carboxysome parts in the composite part, BBa_K2449030, but for some inexplicable reason, this composite part likewise emerged difficult to clone with the Tac-promoter. Therefore, it was decided to focus at other aspects of the project and, as for the Calvin cycle part, keep this part theoretical henceforth.
A fitting approach to verify the expression of the genes involved in carbon fixation, could consist of matrix-assisted laser desorption/ionization mass spectrometry (MALDI MS) and quantitative PCR (qPCR). For this, one could build upon the experience obtained by the 2014 Bielefeld iGEM team.
In order to test the functionality of the parts we had hoped to assemble, we had reached out to Nordcee, which is a major research group within the Department of Biology at SDU. Our hope was to compare the carbon fixation rates for cultures expressing the Calvin cycle enzymes alone and in combination with the carboxysome, whereby the effectivity of the microcompartment also would have been assessed. In cooperation with Nordcee, measurements of CO₂ decrease in an airtight culture would have been carried out with an infrared CO₂analyser. Another test approach would have been using ¹⁴CO₂, and then measure the amount our organism had taken in.