Team:Wageningen UR/Description/Regulation
Regulation of toxin expression
A problem that is encountered when expressing Bt toxins is that overexpression can be lethal to the bacterial chassis, in our case Escherichia coli. We aimed to prevent premature lysis of BeeT and achieve higher toxin yield by separating the growth phase from the toxin-producing phase. Toxin expression in coupled to quorum-sensing: only when a sufficient number of bacteria are present, the toxin is produced. Furthermore, we aimed to create a subpopulation of bacteria that do not produce the toxin, even when the bacterial density is high. When the toxin-producing bacteria perish, the subpopulation survives. As the survivors are genetically identical to the rest of the population, they are able to initiate a new growth phase and subsequently new toxin production.
The high toxin expression is needed to induce significant damage to the Varroa population. It is known that longer-term, low-dose exposure of a pesticide to the target organism can cause resistance (Tabashnik, Brévault & Carrière, 2013). Ideally, the toxin is only expressed when the target organism is present. To further optimize toxin expression, two measures to regulate expression were explored. BeeT was designed to sense the presence of Varroa destructor through the use of two riboswitches: one senses guanine, the other senses vitamin B12. Both are normally not present in beehives, but indicate the presence of the mite as guanine is a major component of mite faeces and B12 is present in the haemolymph of the bees. When a mite attaches itself to a bee to feed on the haemolymph, BeeT will be able to sense it.
Lastly, a toggle switch for toxin expression was designed, by combining the riboswitch and part of the light killswitch employed in our safety system (hyperlink). A hybrid promoter was designed that both facilitates continuous toxin production after transient sensing of the mite, and shuts off production when BeeT escapes from the hive and is exposed to light.
Population Dynamics
Overhauling population dynamics to improve toxin production
In our quest to save the honey bees from Varroa destructor we envisioned using Bacillus thuringiensis cry toxins. When activated and concentrated near the surface of cell membranes, Bt-toxins form pores inside the membrane, lysing the cell as a result1. High level constitutive Bt-endotoxin expression in Escherichia coli is known to inhibit growth and possibly kill the producing cells2. The aim behind the population dynamics subproject is to deliver a system for toxin regulation where production does not impair growth nor population survival of the bacteria.Solutions to the problem of toxic expression are found in mechanisms that regulate protein production to minimize the negative effects on growth and survival3. Inducible expression for instance, is widely used to manually confine toxin expression in time to after the exponential growth phase of bacteria. We propose a regulation system that enables E. coli to separate the growth and the production phase by only expressing recombinant proteins when bacterial cell density is high. We cannot expect beekeepers to measure bacterial growth and induce expression themselves when the time is right.
Quorum Sensing: regulation based on bacterial density
For bacteria to regulate toxin production based on the density of bacteria, they need to be able to communicate bacterial density among them. Systems that enable bacteria to do so are called quorum sensing mechanisms. We have adopted one of the best known quorum sensing systems: the lux system originating form Vibrio fischeri and tested its working with a newly made GFP reporter (figure 1). The lux system consists of luxI and luxR two genes that allow for bacterial density communication. The LUXI protein is a synthase that produces acyl-homoserine lactones (AHLs). AHLs are small compound that diffuse across cell-membranes and function as autoinducers, the molecules that bacteria secrete to signal population density. A single cell produces insufficient autoinducer molecules to start quorum sensing. When there is a high density of bacteria, all producing AHL, the AHL concentration in the growth medium increases. AHL can then reach high enough cytoplasmic concentrations to effectively bind the LUXR protein. LUXR is a cytoplasmic receptor protein binding AHLs, regulating gene expression depending on whether AHL molecules are bound.
To test density dependent expression, a two plasmid quorum sensing system was assembled in and transformed to E. coli DH5alpha cells. One plasmid contains the actual quorum sensing system BBa_K546000 in the pSB4K5 backbone, while the other plasmid contains a newly made GFP quorum sensing reporter BBa_K1913014 in the pSB1C3 backbone. Bacteria that contain both plasmids should be able to communicate their cell density and produce GFP once cell density gets high enough. The bacteria were indeed found to color green in overnight liquid cultures, preliminary indicating functional quorum sensing and reporting hereof (figure 2). The cells that contain only the reporter plasmid do not give a green pellet, indicating that they do not substantially activate expression of GFP.
To follow the dynamics of the quorum sensing, plate reader measurements were done for the previously mentioned 2 plasmid quorum sensing E. coli. Fluorescence intensity was assessed by exciting at 480 nm and measuring light intensity of 510 nm. In addition, the optical density (absorbance at 600nm) was measured to relate fluorescence intensity to cell density (figure 3). Fluorescence over OD600 is plotted to correct the fluorescence intensity for the amount of cells that produce the fluorescence. Starting from an OD600 of 0.8 there is a sharp increase in fluorescence intensity per cell for the quorum sensing system. A very small increase in fluorescence/OD600 is observed for the reporter only control cells as well.
The rough pattern (sharp fluorescence increase starting at OD600) is found in all three biological replicates but the exact reproducibility is limited. To provide future iGEM teams with a streamlined quorum sensing system that includes visualization, we assembled the actual quorum sensing system (BBa_K546000) together with the GFP reporter BBa_K1913014 in pSB1C3. This new part was tested in the same way as the two plasmid system (figure 3). This part behaves roughly the same as the system when divided between two plasmids. The difference is the backbone that accompanies the lux genes, resulting in different copy numbers. In conclusion, both tested systems function as was envisioned and provide a means of regulating gene expression based on bacterial density. In principle this allows us to express the Cry toxin if and only if bacterial density is sufficient (in this case OD600=0.8).
Subpopulation formation
Of course, population wide synchronized toxin overexpression is likely to kill all E. coli cells in the first wave of toxin expression. It would be more beneficial to increase the survival chances of some bacteria by keeping them as non-producers. Cells surviving the production phase would be able to initiate a new growth phase after death of the producing cells. The critical requirement for this is that while most cells display a certain behaviour (toxin production), some cells in the same population behave differently from the rest. A collection of cells that acts differently from the rest of the population is called a subpopulation.We hypothesized that a balance between two antagonistic genes can function as the basis for a subpopulation inducing circuit. The working of the dominantly expressed gene determines cell behaviour. We based our system (Figure 4) on a modified lambda Prm promoter. This promoter is regulated by two cI ‘repressor’ proteins that have opposite effects on transcription of downstream genes. Despite its name, the cI repressor protein from phage λ induces the promoter, whereas the cI repressor protein from phage 434 represses the promoter. These two phage derived genes thus work antagonistically and function as the basis of the subpopulation system reported here. To actually create a subpopulation it is essential that the levels of the two cI proteins (or at least the ratios between them) can differ from cell to cell. Inspired by persister cell formation, we investigated expressing the two cI genes in one operon behind the same promoter. Difference in the turnover of the two proteins assures that changes in promoter strength induce temporary changes in the ratio between the level of the proteins. Imagine a system where both cI genes are transcribed at intermediate levels. 434 cI has a much stronger ribosome binding site (RBS) than the λ cI gene and therefore is translated at a higher rate. This leads to a situation where there are more 434 cI than λ cI proteins: 434 cI is dominant. When the promoter regulating both genes is repressed, both genes will in time reach a new (lower) balance of protein levels. However, when 434 cI is degraded faster than λ cI, 434 cI protein levels will decrease more rapidly. Given the right tuning, this allows λ cI to briefly dominate the system. We hypothesize that small differences (cell age, metabolism etc.) between cells can determine whether λ cI indeed gets the chance to dominate the system. This principle of protein balance chances through changes in promoter strength is similar to how some cells become persister cells, while other continue growing5. We decided against an attempt to create actual persisters out of safety reasons. Persisters cells are a well studied yet poorly understood example of microbial subpopulations6. Persister cells make up around 1% of the population in the stationary phase7. In the main model for persister cells, persisters differ from other cells in the balance between toxin and antitoxin. In persister cells, the toxin in a toxin/antitoxin system dominates the antitoxin. This causes dormancy, allowing persisters to escape antibiotics and other environmental factors that might kill growing cells6.
To test the viability of the described subpopulation system, we constructed a plasmid where λ and 434 cI encoding genes are under control of pBAD, a promoter effectively inducible with L-Arabinose and repressible with D-Glucose. The 434 cI repressor protein encoded on the plasmids contains a C-terminal LVA tag, increasing the protein’s degradation rate. Another plasmid was constructed with the modified λ Prm promoter controlling expression of mRFP, which represents expression of the Cry toxin. The modified λ Prm promoter also controls an additional λ cI gene that provides positive feedback on the λ Prm promoter. This additional λ cI should be degraded more rapidly than the λ cI gene in the inducible operon as it is LVA-tagged. The two plasmids were combined into one plasmid with the two operons. We hypothesized that the system can only work when the balance between the cI proteins is in a certain ‘sweet spot’ of ratios. To improve our chances of hitting this sweet spot we incorporated a library of 18 possible RBS’s upstream of the 434 cI gene. The library was kindly designed by Daniel Gerngross using the Redlibs algorithm8 to provide a library of limited size representing a linear increase in predicted translation rates. This way we varied the translation rates of 434 cI between clones, resulting in different ratios between the λ and 434 cI protein levels for these clones. Transformation of the plasmid (including the RBS library) yielded clones with varying intensities of red coloring on agar (without L-Arabinose and D-Glucose plates (Figure 5). This diversity is a first indication that the assembled system works roughly as intended: variation in the 434 cI RBS should cause different ratios between λ and 434 cI protein levels, leading to differences in mRFP expression.
To provide a more quantitative measurement of the mRFP production and to see how this develops during bacterial growth, we analyzed 95 colonies (from the RBS library transformation) in an overnight plate reader experiment (figure 6.). The data from this experiment shows that the majority of the colonies display roughly the same pattern of mRFP production during growth. We chose to further analyze a selection of these colonies that either showed a unique response or represented a general response found in many colonies. For these colonies we compared the development of mRFP activity between samples grown on only LB with the selection antibiotic, on the same medium with L-Arabinose and on medium with L-Arabinose and Glucose (figure 7).
According to our expectation, glucose repression should lead to a subset of cells starting to produce substantial levels of mRFP. Therefore, addition of glucose should increase the total mRFP activity in the population. This is not clearly found in our results, but the development of mRFP activity in colonies B6, C10, E11 and H9 preliminarily suggests an increase in fluorescence in samples with Glucose starting after approximately 11 hours relative to the samples with Arabinose but lacking Glucose. Of course, subpopulations could only really be observed when fluorescence is assessed for individual cells. To this end we performed fluorescence microscopy on B6, C10, E11 and H9, comparing samples grown in LB (with selection antibiotics) with L-Arabinose to samples where we also added D-Glucose. In an effort to visualize the number of cells that show fluorescence among the other cells, we chose to make pictures where the cells were exposed to mRFP excitation with laser light, but also to low levels of white-light (figure 8).
At first glance this picture seems to confirm the hypothesis: it seems that when Glucose was added, a higher percentage of cells substantially produces mRFP. But this difference is already seen after 10m of incubation with Glucose. This time period is likely too short for protein degradation to play a role, let alone to influence first gene transcription and then even mRFP protein production. Additionally, this difference found after just 10m contradicts the preliminary results from the plate reader experiment, where glucose only seemed to give an effect after 10h! In an effort to clear-up the effect glucose has on the formation of subpopulations, we have performed a flow cytometry analysis for two of the clones: B6 and C10. This method clearly revealed that glucose addition has little or no effect on the formation of subpopulations in our cells. During the analysis we found that the choice of mRFP as reporter was a good one: we found a clear difference in fluorescence between the mRFP producing cells and cells that did not have a mRFP gene. Differentiation between GFP expressing cells and cells unable to produce GFP is known to be more troublesome. Therefore we advise future iGEM team planning to do flow cytometry of FACS to (if at all possible use mRFP as reporter rather than GFP).
Design of a mite sensing system based on riboswitches
Design of a mite sensing system based on riboswitches
To prevent toxin expression when the mite is not present, we have created a mite sensing system that can regulate toxin expression using riboswitches. Substances that indicate the presence of the mite are guanine, since 95% of the mite faeces consist of guanine1 and vitamin b12 since this vitamin is present in the haemolymph of insects. V. destructor feeds on the haemolymph of the honey bee and will leave traces of haemolymph in encapsulated larvae cells and through the hive1.
Guanine and vitamin b12 sensitive riboswitches were used to design the system. Riboswitches are pieces of mRNA that can regulate gene expression depending on if it is bound to a ligand. E. coli possesses a vitamin b12 riboswitch that regulates the expression of the btuB genes3. It also possesses b12 receptors on its outer membrane, allowing vitamin b12 uptake4. The vitamin b12 btuB riboswitch will be used to design a system that will start gene expression in the presence of vitamin b12. For the guanine riboswitch the xpt-buX operon of Bacillus subtilis has been used. E.coli possesses guanine receptors6, so no additional transporters are needed.
The riboswitches regulate translation of genes at the mRNA level2, so they need to be constitutively transcribed. Therefore, a promoter of the Anderson constitutive promoter family will be placed in front of the riboswitch. It is the consensus constitutive promoter family of iGEM, it is well documented and the amount of expression differs 2500 fold between different promoters. It can also be easily swapped for another promoter of the same family if a different amount of expression is needed7. This is convenient since the optimal expression level of the toxin gene is not known yet.
Riboswitches are known to be able to both start and stop gene expression upon binding to specific ligands. Both the guanine and vitamin b12 riboswitch will stop expression after binding to the ligands4,5. In order for the system to work, a component needs to be added that will invert this regulation. For this the TetR Quad-part inverter system has been used. Quad-part inverters, or QPI, are genetic regulatory inverters that consists of a ribosome binding site, a coding region for a repressor protein, a terminator and a promoter that is regulated by the encoded repressor protein8. The TetR system will be used since it is widely used and characterized as a well-functioning inverter8. Figure 1 shows a schematic overview of the system with the riboswitch, TetR QPI and monomeric Red Fluorescent Protein or mRFP.
An often encountered problem when creating a novel combination of genetic elements is the fact that even seemingly simple genetic functions behave differently in different settings9. Therefore, the behaviour of the designed construct might be different than expected. To make the system more stable or increase the chance that the system will behave as designed, the riboswitch system will be made bicistronic. This means that two genes are under the control of one promoter. In this case, the sequence of the riboswitches has been cloned together with fifty amino acids downstream of the riboswitch in the original genomic sequence of B. Subtilis and E. coli. Adding fifty amino acids to the system means an increase of protein production for the cell which can be stressful. Therefore a ssRA protein degradation tag has been added behind the additional fifty amino acids. The protein part in front of this tag will be degraded after translation. For the design of this system, the E. coli ssRA tag10, that consists of around thirty base pairs, has been used.
Cloning the mite sensing system based on riboswitches
The cloning of the system has been done with the use of a 3A-assembly out of two constructs. The first part is called B/Gribo and contains the riboswitch and constitutive promoter. The second part is the inverter part, consisting out of the TetR QPI and mRFP as a reporter gene.
Creating riboswitch part with PCR and special primers
A part containing the riboswitch has been created with the use of special primers that copies the correct riboswitch from genomic DNA. During the PCR, the primer has added the constitutive promoter and ssRA-tag. This part has been dubbed B-ribo and G-ribo (of vitamin b12 and guanine riboswitch). The expected length of B-ribo is 537 basepairs (Figure X, red arrow) and of G-ribo 468 basepairs (Figure X, red arrow)
Creating the inverter part out of bio bricks
The other part of the construct is the inverter-part, containing the TetR gene, a promoter that is inhibited by TetR and the reporter gene mRFP. This part is called the Inv-part. The Inv-part has been created out of standard part BBa_p0140 and BBa_I13521 from the iGEM kit as can be seen in figure X.
The PCR of the Inv-part shows a band between 2000 and 3000 basepairs where a band of 2122 base pairs is expected, as can be seen in figure X.
Combing the riboswitch part and the inverter part to clone the designed construct
Out of the ribo-part and the Inv-part, the designed system B/GRInv has been assembled via a 3a-assembly. The transformation of B/GRinv out of B/Gribo and Inv-part in psB1C3 has resulted in three different types of colonies: white, pink and red ones.
If no guanine or vitamin b12 are present, a colony containing the R/GRInv construct should have a white colour since TetR is expressed when no metabolite is present to bind to the riboswitch. TetR in its turn, will inhibit the promoter that normally expresses mRFP. However, in LB agar, small amounts of vitamin and guanine are present and therefore it is not possible to be sure which colony is the right one based on its colour: white, pink or red. PCR results show that only the pink colonies have the expected size of the B/GRInv construct as can be seen in figure X. Sequencing results confirm that the pink colonies contain the right construct.
Toggle Switch
Overview:
Although quorum sensing system could be a good strategy to keep BeeT’s toxin expression in a reasonable range, it is difficult to apply this system into realistic beehive conditions since BeeT may not growth to a certain density in the brood cells Therefore, we decided to engineer a regulatory system more suitable for the beehive context in beehive as a parallel strategy. In order to integrate mite sensing and light-control for regulating BeeT’s toxin expression, we designed a toggle switch. It consists of two repressors and allows controlling BeeT’s toxin expression under two different stimulus, guanine (or vitamin B12) and blue light. This toggle switch is based on a well-known genetic toggle switch developed by Gardner et al (Gardner T et al., 2000). On it the mutual repression of two repressor proteins results in bi-stability of the system, which could be switched between two stable steady states rapidly. These steady states are off-state and on-state. In our toggle switch, the on-state leads to the stable expression of BeeT’s toxin and it is only reached when the system detects that both BeeT is in the beehive (dark) and mites are present in the brood cells (guanine or vitamin B12). In conclusion, we constructed a regulatory system that could be applied into real beehive condition theoretically and it integrates mite sensing with light-control device, which could provide a stable output of toxin expression with fast respon-siveness.
How does light regulate BeeT?
In this part, we show the design of a whole regulatory system that connects the toggle switch and the light kill switch by means of light (Fig.1). The toggle switch controls expression of the BeeT’s toxin between an off-state and an on-state, which can be switched with two stimulus, guanine (or vitamin B12) and blue light. Our mite sensor is based on two types of riboswitches, one causes transcription termination and the other prevents transcription initiation by locating on the 5’ UTR region of the mRNA attenuating gene translation after binding a metabolite (Mandal, M., & Breaker, R. R., 2004).
The light sensor makes use of YF1 and FixJ. YF1 is the fusion of the LOV protein with a histidine kinase (Möglich A et al., 2009). In the absence of light, YF1 can activate FixJ by phosphorylation resulting in the activation of the Fixk2 promoter. In the presence of 480 nm wavelength light, YF1 can no longer phosphorylate FixJ due to the change position of a salt bridge on the LOV domain, leading to the deactivation of the Fixk2 promoter (Crosson S et al., 2003).
Changes on the guanine or vitamin B12 and light conditions result in different situations on the bee hive.
Situation 1:
Initially, in the brood cells, the expression of the BeeT’s toxin is inhibited by the LacI repressor since the phosphorylation process between YF1 and FixJ takes some time to activate Fixk2 and produce enough TetR repressor to inhibit ptet (promoter driving the expression of LacI). On this situation the system remains in the off-state (Fig.1 a). At the same time, MazE is constitutively expressed from the light Kill Switch preventing the death of the cells.
Situation 2:
When mites are present in the brood cells, they produce and release guanine and vitamin B12 into the microenvironment of the brood cells. The binding of these metabolites to the riboswitches on the 5’ UTR of the LacI mRNA will attenuate translation of LacI (Fig.1 b) resulting on the stop of the inhibition of the FixK2-plac hybrid promoter and initiating the expression of TetR and BeeT’s toxin (Fig.1 c). At this point the system has switched to on-state and any small perturbations will not influence the stable expression of BeeT’s toxin. As cells are still in dark, MazE continues gaining over MazF and cells are alive.
Situation 3:
Once our BeeT escapes from the brood cell, the exposition to light would inhibit the phosphorylation process between YF1 and FixJ, resulting in the inactivation of FixJ and stopping the transcription from the hybrid promotor, FixK2 promoter on Kill Switch. TetR can no longer inhibit the ptet promotor and the system switch back to the off-state (Fig.1 d). In the meantime, MazF would start being expressed, which leads to programmed cell death reducing the the environmental impact of BeeT.
Hybrid promoter design:
In order to make Fixk2 promoter as an inducible promoter for the toggle switch, which could also be compatible with the light kill switch, we had to add additional repressor operator sequences into the Fixk2 sequence. Hence, the whole structure of Fixk2 should be figured out. However, we could not find any literature demonstrating the definite structure of the Fixk2 promoter (BBa_K592006) from the iGEM parts registry and even the original Fixk2 promoter from pDawn and pDusk system (Ohlendorf R et al., 2012) does not have a detailed structural analysis. So we decided to construct five different FixK2 hybrid promoters (BBa_K1913022, BBa_K1913023, BBa_K1913024, BBa_K1913025, BBa_K1913026) based on the sequence of the wild type FixK2 promoter of Bradyrhizobium japonicum (Nellen-Anthamatten D et al., 1998). Nellen-Anthamatten D et al made a definite sequence structure analysis of this Fixk2, showing the presence of two FixJ boxes between the -40 and -70 region and a -10 to -35 core element (Fig. 2, a). According to this sequence structure, we added two additional LacI operators, both upstream and downstream the Fixk2 sequence (Fig 2, b), which could be bonded by tetrameric Lac repressor resulting in the formation of a DNA loop and consequently on transcription repression (Oehler S et al., 1990). We also designed a ptet-Fixk2 hybrid promoter by inserting two TetR operators into the core element region so that Tet repressors could form a dimmer and bond to these operators, resulting in transcription repression (Lutz R et al., 1997). However, according to some previous iGEM projects (UNITN-Trento 2013, INSA-Toulouse 2013), the transcription activity of the wild type Fixk2 promoter is so weak that they all added an inverter part to control their target gene expression. Even the original pDusk system in darkness has only 5 times expression levels than in light conditions. Therefore, we decided to enhance the transcriptional activity of the Fixk2 promoter by changing the core element region of the wild type Fixk2 by a strong constitutive promoter (BBa_J23106) from iGEM part registry and by adding two typical FixJ boxes (Mesa S et al., 2005; Ferrières L et al., 2002) into the -40 to -70 region (Fig.3 a, b) In addition, we designed an additional plac-Fixk2 as a backup (Fig.3, d) via changing core element into -10 to -35 region of ompC promoter from another two component system (Mizuno T et al., 1986), because we couldn’t guarantee that FixJ boxes could be compatible with the core element of a constitutive promoter.
Hybrid promoters are activated in the dark in presence of YF1-FixJ
In order to test the hybrid promoters, we constructed five composite parts with mRFP gene as reporter (BBa_K19103027, BBa_K19103028, BBa_K19103029, BBa_K19103030, BBa_K19103031). These composite parts were co-transformed with the light sensor part (BBa_K19103034) into E.coli strain BL21, cultured under dark for 24h and their fluorescence was tested. We included as controls cells that only contained the Fixk2 composite parts. The results (Fig.4) illustrated that hybrid promoter BBa_K1913022 has 5 times significant different compared to control, which suggested that this promoter is the most sensitive one for being induces by FixJ. Hybrid promotores BBa_K1913023 and BBa_K1913024 showed relative higher florescence values than others, suggesting that they have certain leaky expression. Whereas there were no signifi-cant difference between these promotores and their controls, this means that these two pro-moters are very less sensitive to induction by FixJ than the others.
- EXAMPLE AUTHOR, EXAMPLE TITLE, ETC….
- EXAMPLE AUTHOR, EXAMPLE TITLE, ETC….
