Background

Cells are responsive to a myriad signals under most conditions[1]. Upon receiving this cornucopia of signals, i.e. inputs from the external and internal environment, cells must adjust their own internal mechanisms in response in order to survive. This adjustment depends not only on processing a combination of current environmental signal inputs[1], but also on determining the cell’s current state, which is a result of a series of past inputs. In dig-ital circuit theory, this operating mode is known as sequential logic[3]. Sequential logic is a type of logic in which output is a function of the present value of inputs and the system’s internal state.


The field of synthetic biology aims to design biological systems to perform tasks to better understand analogous natural systems and to develop direct applications for research and medicine. Nowadays, a wide va-riety of tasks can be performed by synthetically engineered genetic circuits, mostly constructed using combinational logic. The output of combinational logic (also referred to as time-independent logic) is a function of the present input only[4]. In contrast to sequential logic, it has no internal state as an influence of the history of inputs. Without a memory of state, it is difficult to perform functions in a specific order, which has limited the widespread implementation of such systems[5].


The ability of sequential logic circuits to store modest amounts of information within living systems and to act upon them would enable new ap-proaches to the study and control of biological processes[6]. Indeed, the attention paid to this field has progressed far beyond scarce. For example, genetic toggle switches have proved to be effective in building synthetic memory modules[7]. In 2010, a bistable push-on push-off switch was constructed as a memory module by Lou et al[8][9]. Five years later, Pakpoom Subsoontorn et al. designed a toggle flip-flop switch whose state can be changed by the same signal[10]. Moreover, in 2012, Jerome Bonnet demon-strated a rewriteable recombinase-addressable data (RAD) module[11]. Re-cently, a state machine with two inputs and five states depending on the order of the input was realized in E. coli by Roquet, N. et al[12].

These studies indicate that more functions could be accomplished with a memory module embedded within the circuit[13]. Once the memory system is able to scale up to more states, a cell can be designed to do work that is more complex. In other words, we can reach a new dimensionality in designing synthetic life–time.

Introduction

This year, the Peking iGEM team is attempting to develop a frame of biological sequential circuits that are programmable. The envisioned circuit is capable of both storing states in DNA and automatically running a series of instructions in a specific order. More specifically, the sequential logic (or a real-state machine) that consists of a clock (trigger signal), flip flop (remembering device) and control unit (functional part) in bacteria. The clock is an oscillator with a repeated signal cycle that is utilized like a metronome to trigger actions of sequential logic circuits. Flip-flop is a memory device that can remember a state. Paired with a clock signal, it can realize state transition. The control unit is a functional part which can convert a signal from flip-flop into complex functions. With such a design, historical events are recorded and influence current cell behavior.

This work tries to point the way toward building large computational systems from modular biological parts—basic sequential computing devices in living cells, and ultimately, programming complex biological functions. Computers have thus become "alive". A unicellular organism itself cannot pack much computational power, but considered as a modular building block, its potential is impressive. 

References

1. Hansen, J., & Benenson, Y. (2016). Synthetic biology of cell signaling. Natural Computing, 15(1), 5-13.
2. Densmore, D., & Anderson, J. C. (2009, May). Combinational logic design in synthetic biology. In Circuits and Systems, 2009. ISCAS 2009. IEEE International Symposium on (pp. 301-304). IEEE.
3. Cavanagh, Joseph (2006). Sequential Logic: Analysis and Synthesis. CRC Press. pp. ix. ISBN 0849375649.
4. C.J. Savant, Jr.; Martin Roden; Gordon Carpenter. "Electronic Design: Circuits and Systems". 1991. ISBN 0-8053-0285-9 p. 682
5. Lou, C. et al. Synthesizing a novel genetic sequential logic circuit: a push-on push-off switch. Mol. Syst. Biol. 6, 350 (2010).
6. Zhang, H., Lin, M., Shi, H., Ji, W., Huang, L., Zhang, X., ... & Yang, Z. (2014). Programming a Pavlovian-like conditioning circuit in Escherichia coli. Nature communications, 5, 3102.
7. Subsoontorn P, Endy D (2012) Design and analysis of genetically encoded counters. Procedia Comput Sci 11:43–54. doi:10.1016/j.procs.2012.09.006CrossRefGoogle Scholar
8. J. Bonnet, P. Subsoontorn, D. Endy, Rewritable digital data storage in live cells via engineered control of recombination directionality. Proc. Natl. Acad. Sci. U.S.A. 109, 8884–8889 (2012). 10.1073/pnas.1202344109pmid:22615351 doi:10.1073/pnas.1202344109
9. Roquet, N., Soleimany, A.P., Ferris, A.C., Aaronson, S. & Lu, T.K. Synthetic recombinase-based state machines in living cells. Science 353, aad8559 (2016).
10. Burrill, D. R., & Silver, P. A. (2010). Making cellular memories. Cell, 140(1), 13-18.

Back to top

Flip-flop

Background

The basic building blocks of sequential logic circuits are flip-flops, which are devices that can store information. A flip-flop has two stable states (see more on "state") and its state can be changed by signals of arriving pulses. Thus, it and can store one bit of information and alternate its state following a trigger. It is the basic storage element in sequential logic. Click here to get a more specific description

When designing genetic-circuit counterparts of flip-flops, we should consider the following three criteria:
In a living cell, DNA is the natural medium for storing cell-state information and encoding functions. Recombinases, especially a subset called serine integrases and excisionases, are enzymes that can flip or excise specific fragments of DNA. They have been proved to be able to stably modify DNA sequences , which is the biological basis of our genetic flip-flop construct.

Large serine integrases reliably and irreversibly flip or excise unique fragments of DNA . DNA cleavage and re-ligation occur at the central crossover region at a pair of recombinase recognition sites (attB and attP), which allows the sequence to be flipped, excised, or inserted between recognition sites . After recombination, the original attB and attP sequences become reconstructed sequences - attL and attR. The resulting attL and attR sequences cannot be recognized and bound by integrases alone, so the state after integration is stable.

Another kind of recombinases, excisionases, are able to recognize and bind attL and attR sequences. With the help of excisionases, the state transitions become reversible.

Recombination Directionality Factors (RDF) are proteins which mediate reverse flipping to recover the sequence flipped by recombinases. By co-expressing the integrase with the corresponding RDF, or expressing the integrase-RDF fusion protein, recombination between the attL and attR sites can be induced. Thus, the flip-flop can be restored to the previous state.

Figure 1. Site-specific recombination either integrates, deletes or reverses a DNA sequence.

Aims

We would like to prove the feasibility of bio-flip-flops and optimize their function as parts. In order to do this, we have to:
We would first search for known recombinase sequences in the literature, construct the corresponding vectors, and select the best ones to make standardized parts. After confirming their expression and function, we would optimize these devices through the standard workflow of vector and RBS substitution. We would also prove integrase-RDF fusion protein function. Furthermore, we would con-struct half of the bio-flip-flop structure and experimentally show that it can function.

This would serve as a proof of concept for a novel design strategy for bio-flip-flops.

Key Achievements

Our Approach and Designs

Design of the bio-flip-flop

Using a pair of integrases and their corresponding excisionases, we designed the bio-flip-flop device to store state information as follows:

Figure 2. Schematic diagram of the bio-flip-flop mechanism.

Selection of Available Integrases

We first did a literature search and sequence alignment to mine the databases for available integrases for device development. Here we list the names and sources of the collected recombinases.


We finally chose to use TP901-1, Bxb1 and phiC31. These are well-known serine integrases, and all of them have been used to demonstrate static-input logic gates. Some have cofactors (equivalent to RDFs) that can reverse directionality. Previous work has shown that there exists a new set of 11 orthogonal integrases, greatly expanding the availability of integrases that can be formed into standardized parts. We hope that given enough time and resources, it is possible to construct more parts using our standard workflow described below.

Standard Characterization & Optimization Workflow

The viability of a bio-flip-flop relies on the performance of two integrases and their corresponding excisionases. To select integrases for the bio-flip-flop, we constructed expression vectors for different recombinases and tested their performance individually. Here we introduce a standard workflow to accomplish this task.

Figure 3. The workflow of bio-flip-flop optimization.

Different replication origins affect the copy number of plasmids, and it is highly probable that they also influence the expression levels of their encoded recombinases. To select integrases for the bio-flip-flop, we constructed expression vectors for different recombinases and tested their performance individually. This was carried out by co-transformation of E. coli Top10 with the expression vector and a reporter plasmid. The reporter plasmid expresses differen t fluorescent proteins before and after flipping of a constitutive promoter.



Figure 4. The standard genetic structure used to characterize the recombinases.


According to our design of the bio-flip-flop, the two inputs X and Y comprise a signal. Consequently, two distinct induction systems need to be constructed. For this reason, recombinases were cloned onto two different backbones, under regulation of differently induced promoters. The expression vectors were constructed by Gibson assembly. Plasmid type 1 is cloned from the repressor generator (RPG), which has a p15A replication origin, ampicillin resistance gene, lacI, an IPTG-induced pTac promoter, and a RiboJ insulator after pTac. The backbone of plasmid type 2 was cloned from pIntegrase_13, which was a gift from Christopher Voigt (Addgene plasmid # 60584). It has a ColE1 replication origin, kanamycin resistance gene, araC, and an arabinose-induced pBad promoter. RiboJ was added to Plasmid2 by cloning RiboJ and the recombinase from Plasmid1, and assembled after pBad.



Figure 5. Integrase expression vectors with different replication origins. The one shown on top has a relaxed ColE1 replication origin and recombinase expression is induced by arabinose via a pBad induction system. The one shown in the bottom picture has a relaxed p15A replication origin and recombinase expression is induced by IPTG.

For a given backbone and inducible promoter, RBS strength affects the expression level post-transcription. For this reason, we tested a series of RBS for different recombinases in different backbones.

In order to tune the RBS, an intermediate vector with a 20bp random sequence (45% GC), flanked by a pair of BsmbI recognition sites (introduced by PCR) was prepared for each recombinase. Vector construction was completed by adding RBS via Golden Gate assembly. The RBS calculator was used to generate RBS sequences with a wide range of translation initiation rates. Other sequences (B0029, B0030, B0032, B0034, B0035) were obtained from the iGEM website. The RBS oligos were obtained from the annealing of two complementary primers. Expression vectors for each recom-binase with different RBS were tested by flow cytometry after co-transformation with the reporter vec-tor. For each recombinase, an RBS correlating with low leakage (small RFP subset in the cultures with no inducer) and high recombination efficiency after induction was selected.

Measurement of Recombination Efficiency

We used a microplate reader to roughly measure the efficiency of the selected integrases. We used flow cytometry to conduct a more accurate characterization.

Microplate Readers

Microplate readers are instruments used to detect biological, chemical or physical events in samples in microtiter plates. We used a microplate reader to detect optical density and fluorescence intensity follwing the procedure as below:

1. Pick and inoculate single colony into 1ml of LB media with antibiotics in a V-bottom 96-well plate. Grow the culture overnight (about 12 hours) at 37°C, 220 rpm in a Thermo VARIOSKAN FLASH shaker.
2. Then, an aliquot comprising 2 μL of the culture was transferred into 1ml LB media with antibiotics and inducer (1mM IPTG or 10mM arabinose). Culture overnight (about 12 hours) at 37°C and 220 rpm.
3. Aliquot 1 μl of the cultures in the deep-well plates, then copy them to the 96-well plate filled with 200μl PBS. Don’t forget to leave at least 12 wells with only PBS in it as your negative conctrol.
4. Measure the samples (OD and Fluorescence). The former represents the density of bacteria, and the latter implies the efficiency of recombination. GFP/OD is calculated to measure recombination efficiency.

Flow Cytometry

For more detailed measurements,flow cytometry were used to evaluate the recombination efficiency. Our procedures are as follows:

1. Single colonies were picked and used to inoculate into 1ml of LB media with antibiotics in a V-bottom 96-well plate.
2. The cultures were grown at 37°C and 1000 RPM for 12h. Subsequently, an aliquot comprising 2 μL of the culture was transferred into 1ml of M9 glucose media with antibiotics and inducer (1mM IPTG or 10mM arabinose for RBS tuning, gradient concentration for transfer curve) in a V-bottom 96-well plate.
3. The cultures were grown at 37°C and 1000 RPM for 15h. An aliquot comprising 2μL of the culture was transferred into 198 μL of phosphate buffered saline (PBS) containing 2 mg/mL kanamycin in a 96-well plate. This mixture was incubated for one hour at room temperature before testing. Two lasers were used to excite GFP and RFP simultaneously. Single-cell fluorescence distribution at both emission wavelengths was recorded.

The counted cells were gated to eliminate the population which showed no fluorescence. The remaining cells were divided into two subsets by a diagonal: RFP sub-set and GFP subset. The recombination efficiency was estimated from the proportion of the RFP subset in the total fluorescent population.


Figure 6. A example of Gating the RFP and GFP subsets. Change of fluorescence after induction was seen. Left: no inducer. Right: 10 mM Ara for 15h. Change of fluorescence after induction was seen. Left: no inducer. Right: 10 mM Ara for 15h.

For the vector with ColE1 replication origin, we found proper RBS in a list of calculated RBSs for TP901-1.

We observed a shift in fluorescence level and wavelength in several trails, which indicated that recombinase can flip the promoter region and alter gene expression in our design. However, problems such as leakage and insufficient recombination efficiency were obvious. Tuning of our expression system seemed necessary before assembly into the bio-flip-flop device.

Characterization of Excisionases

To make sure that the sequence-reversing process is efficient and complete, we used two strategies: one encompasses combining the integrase and its corresponding RDF into a polycistronic structure, and the other is to construct fusion proteins comprising the integrases and RDFs in one polypeptide chain. We have been trying to improve the recombination efficiency by replacing the expression vec-tor with different replication origins or inducible promoters and changing the RBS sequence before the coding sequence.


Figure 7. The two strategies used to characterize integrases and their RDFs. UP. The polycistronic structure containing the integrase and its corresponding RDF. Bottom. The fusion protein structure of integrase with its corresponding RDF.

The method is similar to the integrase test method. We constructed an input plasmid to express integrases and RDFs, and an output plasmid on which the promoter J23119 is located between the attL and attR sequences. The input plasmid expresses GFP before induction of integrase and RDF, and if an inversion occurs after induction, the output plasmid can express RFP. By comparing the changes in fluorescence types and intensities before and after induction, the efficiency of inversion can be estimated.

Figure 8.The standard genetic structure used to characterize the recombinases.

Construction of Bio-flip-flops

After having already optimized the expression systems of integrases and excisionases, we set out to construct a Bio-Flip-Flop. Since it is unknown whether the Bio-Flip-Flop is practical, we considered developing two hierarchical execution units supporting our Bio-Flip-Flop. We called these two units the forward latch and the backward latch. The state transition process of each unit consists of two stable and irreversible states. Nevertheless, the state transition proces s of both units consists of four stable and cyclic states. Such units support the verification of the feasibility of our Bio-Flip-Flop and further allow us to construct a complete and functional Bio-Flip-Flop.

Forward Latch

The structure of the forward latch consists of two plasmids with different induction systems of integrases (Figure 9). The state transition process of the forward latch consists of two stable and irreversible states (Figure 10).


Figure 9. The structure of the forward latch. Two plasmids constitute the forward latch. Each plasmid contains an integrase gene (Bxb1-gp35 or TP901) and a reporter gene (GFP or mRFP). To standardize the input, we used the pBAD promoter to regulate the transcription of Bxb1 and GFP, and the pTAC promoter to regulate the transcription of TP901 and mRFP. Moreover, each integrase recognizes and converts the attB and attP sites on the other plasmid.


Figure 10. The state transition process of the Forward latch. After induction with arabinose, the expression of integrase Bxb1 and GFP is initiated. The integrase Bxb1 recognizes and converts the attB and attP sites flanking the pTAC promoter, which leads to a change of orientation of the pTAC promoter. If we input IPTG next, the expression of integrase TP901 and mRFP will begin. The orientation of the pBAD promoter will change. As the GFP is degraded and mRFP is produced, the ratio of mRFP/GFP fluorescence rises.

Backward Latch

The structure of the Backward latch consists of two plasmids with different induction systems driving excisionases (Figure 11). The state transition process of the Backward latch consists of two stable and irreversible states (Figure 12). Because excisionases recognize attL and attR sites and convert them into attB and attP sites, the backward latch is capable of resetting the state of the Forward latch.


Figure 11. The structure of the Backward latch. Two plasmids constitute the Backward latch. Each plasmid contains an excisionase gene (Bxb1-gp47 or TP901) and a reporter gene (GFP or mRFP). To standardize the input, we used the pBAD promoter to regulate the transcription of Bxb1 and GFP, and the pTAC promoter to control the transcription of TP901 and mRFP. Moreover, each excisionase recognizes and converts the attL and attR sites on the other plasmid.


Figure 12. The state transition process of the Backward latch. After induction with arabinose, the expression of excisionase Bxb1 and GFP is initiated. The excisionase Bxb1 recognizes and converts the attL and attR sites flanking the pTAC promoter, which leads to a change of the orientation of the pTAC promoter. If we input IPTG next, the expression of excisionase TP901 and mRFP will begin. The ori-entation of the pBAD promoter will change. As GFP is degraded and mRFP is produced, the ratio of mRFP/GFP fluorescence rises.

Experiment

Because our final goal is to realize the state transition of our Bio-Flip-Flop, it is significant to characterize the efficiency of the primary state transition of the forward latch. We quantified the amounts of GFP and mRFP and considered the ratio of mRFP/GFP fluorescence as a direct measure of the efficiency of the state transition.

For all characterization experiments, co-transformed cells were first grown and subcultured overnight, after which we induced the cells with arabinose and IPTG in that order. Finally, we quantified the fluorescence strength using a flow cytometer.

Results

Primary Proof of Recombination

To prove the function of our integrase-RDF fusion protein in an illustrative way, IPTG was applied to the petri dish where E. coli Top10 co-transferred with expression and reporter plasmid were cultured. Compared to the RFP control (transferred with RFP expressing reporter plasmid only), green fluorescence was obvious when excited by visible light. Half red, half green colonies were seen, indicate incomplete recombination.


Figure 13. (left) RFP control. (middle) expression vector of TP901-RDF fusion protein co-transferred with reporter. White arrows: colonies with green fluorescence indicates recombination. (right) A example of half red, half green colonies, indicating incomplete recombination.

In a similar trail, Arabinose was applied to the petri dish where E. coli Top10 co-transferred with TP901-1 integrase expression and reporter plasmid were cultured. Compared to the GFP control (transferred with GFP expressing reporter plasmid and expression vector, but no inducer was added), We can see a transition of colony color from green (GFP expressin) to red (RFP expression). When excited by visible light, it’s distinguishable that the colonies nearly completely cease to express GFP. The expression of RFP and shut down of GFP expression shows that recombination has happened.


Figure 14. (Up, left) GFP control, colonies in green. (Up, right) GFP control, green fluorescence can be seen when excited. (Bottom, left) Expression vector of TP901-1 integrase co-transferred with reporter, red color instead of green indicates recombination. (Bottom, right) Expression vector of TP901-1 integrase co-transferred with reporter, green fluorescence is nearly completely shut down, indicating desirable recombination efficiency.

Fine-tuning of Recombination Efficiency

Selection of RBS for different replication origin

For the vector with ColE1 replication origin, we found proper RBS in a list of calculated RBSs for TP901-1 and Bxb1 gp35.


Figure 15. Bxb1 gp35 recombination efficiency with a variety of calculated RBS and RBS sequences from iGEM (B0030∼B0035).

T.I.R = Translation Initiation Rate


Figure 16. TP901-1 recombination efficiency with a variety of calculated RBS. A. At the vector with p15A.

B. At the vector with colE1.

T.I.R = Translation Initiation Rate


For the expression vector with p15A replication origin, a proper RBS for TP901-1 was selected.


Figure 16_b. TP901-1 recombination efficiency with a variety of RBS from iGEM (B0030∼B0035).

T.I.R = Translation Initiation Rate

Transfer curves

We evaluated the performance of phiC31 and Bxb1 gp35 on ColE1 expression vectors under a series of inducer concentration gradients. The results enabled us to determine the appropriate inducer concentration.


Figure 17. The transfer curves of the recombinases we utilized. A. The transfer curve of integrase phiC31. B. The transfer curve of integrase Bxb1.

Measurement of the Flipping Efficiency of Excisionases

We have tested the flipping efficiency of our Integrase-RDF fusion proteins with a group pf RBS. Flipping performance varies with different RBS. For TP901-RDF fusion proteins, recombination seems not obvious in some cases (RBS B0032, B0034, B0035, phiC 10000, Bxb 1000, int2 3000), while in others both flipping efficiency and leakage and significant.

Figure 18. pTac TP901-RDF fusion protein flipping efficiency with ten RBS. Proportion of GFP subset indicates flipping efficiency.

A second test was supplemented. Similar results support that our Integrase-RDF fusion proteins are able to flip the sequence between attL and attR site of the reporter, but much improvement are needed to reduce the leakage as well as increase the efficiency.


Figure 19. pTac TP901-RDF fusion protein flipping efficiency with two RBS. Proportion of GFP subset indicates flipping efficiency.

Integrase-RDF for Bxb1 and phiC31 were also constructed, tested. High flipping efficiency and leakage were seen.


Figure 20. Flipping efficiency for pTac integrase-RDF fusion protein of phiC31 and Bxb1. Proportion of GFP subset indicates flipping efficiency.

To conclude, we found that the Integrase-RDF fusion proteins are able to flip the sequence between attL and attR site (indicated by observed GFP subset), although tuning work remains to be done to construct a controllable, functional bio-flip-flop device.

Reference

1. Yang, L., Nielsen, A. A., Fernandez-Rodriguez, J., McClune, C. J., Laub, M. T., Lu, T. K., & Voigt, C. A. (2014). Permanent genetic memory with> 1-byte capacity. Nature methods, 11(12), 1261-1266.
2. Bonnet, J., Yin, P., Ortiz, M. E., Subsoontorn, P., & Endy, D. (2013). Amplifying genetic logic gates. Science, 340(6132), 599-603.
3. Baker, T. A., Bell, S. P., Gann, A., Levine, M., & Losick, R. (1970). Molecular biology of the gene.
4. Olorunniji, F. J., McPherson, A. L., Rosser, S. J., Smith, M. C., Colloms, S. D., & Stark, W. M. (2017). Control of serine integrase recombination directionality by fusion with the directionality factor. Nucleic acids research, 45(14), 8635-8645.
5. Siuti, P., Yazbek, J., & Lu, T. K. (2013). Synthetic circuits integrating logic and memory in living cells. Nature biotechnology, 31(5), 448-452.
6. Khaleel, T., Younger, E., McEwan, A. R., Varghese, A. S., & Smith, M. (2011). A phage protein that binds φC31 integrase to switch its directionality. Molecular microbiology, 80(6), 1450-1463.
7. Salis, H. M., Mirsky, E. A., & Voigt, C. A. (2009). Automated design of synthetic ribosome binding sites to control protein expression. Nature biotechnology, 27(10), 946-950.
8. Espah Borujeni, A., Channarasappa, A. S., & Salis, H. M. (2013). Translation rate is controlled by coupled trade-offs between site accessibility, selective RNA unfolding and sliding at upstream standby sites. Nucleic acids research, 42(4), 2646-2659.

T Flip-Flop

Basically, a real flip flop has two inputs. One input is a control input. For a T flip flop, the control input is labelled T. The other input is the clock. Positive edge-triggered flip flops can only change output values when the clock is at a positive edge. While negative edge triggered flip flops change on a negative edge, and level-triggered flip flops, that change only when the value is 1. When the clock is not at a positive edge, then the output value is held. That is, it does not change. A flip flop’s output is the bit it stored. Thus, the flip flop is always outputting the one bit of information.
Here's the characteristic table for a T flip flop.

T	Q	Q+	Operation
0	0	0	Hold
0	1	1	Hold
1	0	1	Toggle
1	1	0	Tohgle

The T flip flop characteristic table has 3 columns. The first column is the value of T, a control input. The second column is the current state, that is the current value being output by Q. The third column is the next state, that is, the value of Q at the next positive edge. It's labelled with Q and the superscript, + (the plus sign).

The T flip flop has two possible values. When T = 0, the flip flop does a hold. A hold means that the output, Q is kept the same. When T = 1, the flip flop does a toggle, which means the output Q is negated. Thus, in a T flip flop, you can either maintain the current state's value for another cycle, or you can toggle the value (negate it).

Why isn't it XOR?
First, XOR has two inputs, and one output. A T flip flop essentially has a control input and a control output. Second, the second column and the third column are really the same output, but at different points in time.

Difference between Flip flop and gate
The biggest difference between a flip flop and a gate is that a flip flop can hold its value. Even though holding a value is something very simple, it makes it different from logic gates, and allows us to design circuits that have cycles in them (i.e., feedback).

State

There is a generic definition of state:
State is a quantity that stores the previous history of what's happened in such a way that you can predict the output, given any future input.

From the definition, we can get two things. First, state refers to some recorded information. We can consider state as a kind of memory. For sequential logic circuits, this is some finite number of bits. Second, there is always a notion of time when we talk about state. Usually, we think of time being involved with state, because it may change over time.
Let’s think of an object, it starts off in an initial state, and over time, that state changes. For example, if the object is a piece of paper, when we are drawing we may draw or erase something (add or remove elements) over time. Our picture’s state may not keep track of all the things that have happened to it, nevertheless, the picture changes from state to state, as various operations are performing object.

Learn More

Back to top

Controller

Background

In our design, after having designed the flip flop, the device can remember the information of its state, the next step is to transform the state into an actual function. To achieve this transformation, we first needed a “reader” to read out the current state. At this point the control unit comes into play. A control unit is a DNA sequence with recombinase sites whose expression is controlled by recom-binase and RDF by reversing or deleting a promoter and/or a terminator. To make the control unit reliable and predictable, we first need to be able to predict the behaviors of its “building blocks” (or “elements” in electrical engineering), from which we weave our engineer’s perspective into the biolog-ical system. However, we need to make some adaptations and adjustments to these “elements” to make them usable.

Aim

Key Achievements

Our Approach and Designs

Characterization of Terminators

Terminators are responsible for disengaging the RNA polymerase from the DNA and stopping gene transcription in vivo. This characteristic makes terminators desirable as switches in synthetic genetic circuits. The terminators that are used to realize this function should be unidirectional. That is, they should be able to block the movement of RNA polymerase effectively in one orientation and hardly influence its movement in the other. With this “reader”, advantages include a shorter sequence, per-manently recorded message in the DNA and greater number of design plans. Consequently, it is es-sential that we characterize the different terminators. We first chose 24 terminators whose forward orientation strength is high, as previously reported. We then measured their strength in both the forward and reverse orientations in our testing system. Finally, we sifted out a set of six reliable unidirectional terminators without potential cryptic promoters in both orientations.

Given that the same part can perform differently in different genetic circuits, previous testing systems and results of terminators may not fit our project. We therefore constructed a rather simple but practical reporter plasmid (as shown below), based on which terminator strength can be measured via fluorescence intensity and simple calculations.

Vector construction

Figure 1. Vector construction to measure the terminator strength

As shown above (Figure 1.), the circuit used to measure terminator strength was constructed on the pSB4C5 vector. The terminator to be tested is located downstream of the Ptac promoter and followed successively by RiboJ, RBS, and sfGFP. To enable high throughput, we introduced a pair of BsaI cutting sites between Ptac and RiboJ so that Golden Gate assembly can be used to add different terminators efficiently. In addition, there is a random sequence (from a previous associated research, the author of which had designed it using the Sequence Manipulation Suite) between the two Bsal sites flanked by two RiboJ, which was constructed as a positive control.
The terminators to be tested were selected from the literature.

RiboJ acts as an insulator between the RBS and its upstream sequence in the genetic circuit. The insulating function of RiboJ is directly associated with its biochemical function. RiboJ can be defined as the DNA sequence containing a ribozyme called sTRSV- ribozyme and a 23-nt hairpin. The downstream hairpin helps to expose the RBS. mRNA can be cleaved by sTRSV autocatalytically at a defined residue. As a result, extraneous RNA leaders that arise from transcription from promoters with different start sites will be cleaved. Moreover, to make sure the sequence upstream of the RBS will not affect RiboJ’s efficiency, the transcripts of our genetic circuits were designed to have the same 5’-UTR sequence, to avoid the effect called “part junction interference”, in which neighboring sequences may alter the part’s character. For example, a barcode part was found to contain a sequence similar to the –10 region of a constitutive promoter.

Flow Cytometry

Vectors constructed with different terminators in both directions were tested via flow cytometry.
E. coli strains were grown for 12h in LB media containing antibiotics in culture tubes at 37°C and 250 rpm in an incubator. After 12h, the cultures were diluted 1:100 with M9 medium with antibiotics in 96-deep-well plates and grown for 2h at 37°C and 1150rpm in a shaker. Next, the cultures were diluted 1:100 with fresh M9 medium with antibiotics and different concentrations of IPTG. These cultures were grown for 6 h and then diluted 1:5 with phosphate-buffered saline (PBS) containing 2 mg/ml kanamycin or 35 ug/ml chloramphenicol to arrest protein production, and fluorescence was measured using a flow cytometry.

Cells were analyzed by flow cytometry using a BD biosciences Fortessa flow cytometer with blue (488-nm) laser. An injection volume of 15 μl and a flow rate of 1 μl/s were used. Cytometry data were analyzed using Flowjo. The gated populations consisted of at least 10 000 cells. The median fluorescence of the gated populations was calculated using Fowjo and used for all reporting. Autofluorescence of white cells was subtracted from all fluorescence measurements.

The Mathematical Formula

As shown below, a mathematical formula was also used to characterize the strength of the terminators. ＂Ts＂ is the abbreviation of ＂terminator strength＂. [GFP]_{random sequence} denotes the fluorescence strength of the plasmid with the random sequence. [GFP]_terminator denotes the fluorescence strength of the plasmid with a terminator.

Ideal unidirectional terminators should fit the following standards:

1. The Ts of both directions are larger than 0.9 (i.e. there is no cryptic promoter potential)
2. The Ts ratios (Ts of the forward direction/Ts of the reverse direction) are larger than 10
3. The larger Ts between the forward and reverse directions has to be larger than 10 (which ensures the blocking strength to be high enough)
4. The smaller Ts between the forward and reverse directions has to be no larger than 1.2 (i.e. terminators in reverse orientation no longer block the transcription process).

To remeasure the Ts of the unidirectional terminators

After the preliminary sifting of the terminators, a set of desirable unidirectional terminators were sifted out and re-measured using the plate reader to refine the parameters.

TOP10 cells were co-transformed with the system containing the terminator and one pair of attB/P sites and the expressing system. Cells then were cultured overnight on LB plates.

Colonies were picked and cultured in liquid media with antibiotics and inducer. The bacterial liquid culture was then diluted at the ratio of 1:100 into M9 media with both antibiotics and inducer into 96-well plate, cultured for 12-24 hours in the incubator. After the induction, OD600 and fluorescence were measured using the plate reader. Terminator strength was then calculated according to the formula above.

Characterization of Inversion Efficiency

After finishing these measurements, the main protagonist of our project, the recombinase, will come into play. When terminators are flanked by a pair of attB/P sites, the synthetic circuit will transit between on and off state, which is achieved by controlling the expression of the respective recombinases. Here the question becomes whether the attB/P sites flanking the terminators may alter their strength due to some unknown interaction. Thus, we needed to remeasure their strength. Additionally, the inversion efficiency of respective recombinases need to be characterized.

We introduced the testing system and expression system of the corresponding recombinases into the same cell to see if inversions happened, and if leaky expression of recombinases would impact the system. In the end, couples of terminators and recombinases with near-complete inversions and minimal leakage were selected.

Vector construction

Figure 2. Terminator flanked by a pair of attB/P sites

As shown above, we selected terminators for both directions that were sandwiched between attB/attP recognition sites, on the basis of the aforementioned terminator strength testing system. The constructed terminators are shown in the table below.


Those letters in pairs inside the brackets stand for different two-base core inside the sequence of recombinase recognition sites, which remained unchanged after inversion. The expression system includes the recombinase Φc31, Bxb1, TP901-1, or int2 under the control of the inducible pBAD promoter and optimal RBS, which can simultaneously ensure high efficiency and low leakage.

As a part working in genetic circuits, the terminator flanked with a pair of att B/P sites is a brand new terminator in a sense. It’s necessary for us developers to measure the Ts of the new terminator.

Remeasurement of Ts of the new terminators

The plate reader is which we use to remeasure. The protocol and the formula are same as the page above.

Characterization of inversion efficiency

Activation and induction Escherichia coli colonies were picked and grown for 12h in LB media containing antibiotics in culture tubes at 37°C and 250 rpm in an incubator. Next, the bacterial cultures were diluted at a ratio of 1:100 with fresh M9 medium with antibiotics and inducers, after which the bacteria were grown for another 12h.

Plate reader Fluorescence was measured using Varioskan© Flash. We measured the OD600 and green fluorescence (the excitation wave length is 485nm and the emission wave length is 510nm). The ratio of fluorescence intensity/OD600 was calculated using the net fluorescence intensity and net OD600. The transcription barrier strength of an attB/P-Terminator can be characterized quantitatively using the Ts formula.

The inversion efficiency was calculated according to the below formula.

F_terstands for the fluorescence intensity of the system containing both the terminator and the recombinase elements.
F_control stands for the fluorescence intensity of the system containing only the terminator element.
F_forward stands for the fluorescence intensity of the system containing both the forward terminator and the recombinase elements.
F_reverse stands for the fluorescence intensity of the system containing both the reverse terminator and the recombinase elements.
The difference between the fluorescence intensity of the system expressing a recombinase and the system not with recombinase expression, divided by the difference between the fluorescence intensity of the system containing the terminator in the forward direction and the system containing the terminator in the reverse direction.

Characterization of Terminator Flanked with Two Pairs of attBP Sites

Vector Construction Figure 3. Terminator flanked by two pairs of attB/P sites

As shown above(Figure 3.), the terminator is sandwiched between two different pairs of attB/attP sites, under the control of a constitutive J23119 promoter.

Two sets of expression systems, induced by IPTG and arabinose, were constructed. Same as before, a mathematical formula was also used to characterize the strength of the terminators. [GFP]_{random sequence} denotes the fluorescence strength of the plasmid with the random sequence. [GFP]_terminator denotes the fluorescence strength of the plasmid with a terminator. The inversion efficiency calculation formula is same as above.

Construction of Seven-Segment Display

After constructing two reliable elements with predictable behaviors, we aimed to directly implement and arrange them according to sequential logic, which comes in the form of a visible “1→2→3” transition on the seven-segment display. In this display, GFP expression is used instead of an LCD (liquid crystal display) to light up the panel. In our case, three simple and previously characterized parts were placed in segments except F. The system yields the predictable expression of recombinases which invert terminators between corresponding attB/attP sites either once (A,D,G) or twice (C,E).

Figure 4. The schematic of Seven-segment Display

Construction of pattern“2017”→“PKU”in 96-well plate

Cells were placed according to the pattern of “2017”and “PKU”, and cultured in the 96-deep-well plate. Then cells were transferred into clear-96-well plate, OD600 and fluorescence were measured using the plate reader. The transition from the pattern of “2017” to “PKU” was achieved using the inversion effect.


Figure 5. The schematic of “2017”→“PKU”

The holes marked “a” is always light. The ones marked “b” need to transform from the light state to the dark state. And the ones marked “c” perform in contrary.

Results

Characterization of Terminators

A total of 24 terminators were tested in both orientations (their code names[1] and abbreviations are shown in the table below). And a gradient of different concentrations of IPTG (0, 0.01, 0.1, 1mM) was used for induction. We found that the calculated Ts values varied with IPTG concentration. Commonly, the Ts value for the forward orientation increased with IPTG concentration and reached the highest value at 0.1mM, after which it declined slightly. In the reverse orientation there were no obvious trends. Consequently, 0.1mM of IPTG was the most suitable concentration for induction. After the test, we chose 6 terminators (221,836,435, 322,855 and 309), among which 435 and 322 showed the best performance and conformed best to the 4 standards described above.


(The code names and their abbreviations.)

















Characterization of Inversion Efficiency

Ts of the different attB/P+terminator combinations
Transformation with the testing system plasmid alone yielded the following results and conclusions:

Sole attB/P sites (without a terminator) can block the transcription to a certain extent. Such barrier strength varied with the types of the sites. Among the 5 pairs of attB/P sites tested, int2 had the strongest intrinsic barrier effect ("Ts"= 23.0), even exceeding that of the terminator 221F ("Ts"=16.6). The other sites had weaker barrier effects: int2＞ΦC31 ("Ts"=2.00)＞TP901-1(AG)("Ts"=1.42).



Combination of different attB/P sites and terminators can produce unexpected results

Int2 attB/P and 435F had the strongest barrier strengths among the recognition sites and terminators, respectively. However, the barrier strength of their combination was the weakest (“Ts”=5.37) among all combinations.
TP901-1(AG), attB/P and 221F had the weakest barrier strength among recognition sites and terminators, respectively, but the barrier strength of their combination was the highest.
In addition to the barrier effect, the combination of attB/P sites and terminators can also lead to the formation of cryptic promoters, which was the case for the TP901-1(AG) 221R combination.

The most excellent combinations of attB/P sites and terminators selected based on the four standards above were (with the corresponding “Ts” ratios in brackets): int2 221 (35.3), Φc31 435 (28.3), TPAG 221 (17.7), TPAG 435 (17.3), int2 855 (17.2), and BxGT435 (21.1).
Considering that the expression systems of int2 and TP901-1 are induced by IPTG, which causes a conflict with the reporting system, we used BxGT435 and Φc31 435, whose systems are induced by Ara, to conduct the reversal experiments.






Reversal Experiment

Co-transformation of E. coli with the testing system plasmid and corresponding recombinase expression system We observed successful reversal events after inducing the expression of the recombinase. As shown in the picture of the 96-well plate, the fluorescence difference is visible by naked eye. Using the microplate reader and analysis, we confirmed the existence of a significant fluorescence difference. In addition, we directly used the bacterial solution to conduct a PCR, and the sequencing results showed double peaks, which substantially proves that the reversal occurred.











Characterization of Terminators Flanked by Two Pairs of attB/P Sites

We built the following constructs:


Two sets of expression systems, induced by IPTG and arabinose, were constructed.



Ts of double attB/P+terminator combinations





Due to the introduction of double attB/P sites, which substantially extended the distance from the promoter to the GFP coding sequence, together with the barrier effect of each pair of attB/P sites, it became impractical to expect the Ts of the attB/P with a reverse terminator to be around 1. We there-fore lowered the standard and laid more emphasis on the ratio of Ts for both orientations. Based on the test results above, we chose TP901-1 BxGT435 and Φc31 BxbGT 435 to conduct the reversal experiment.



Reversal Experiment






Image Transition Demos

The E. coli groups with “Bxb1 T+B435F” were placed at the position c, and the E. coli groups with “Bxb1 T+B435R” were placed at the position b. The a position there are groups which always expressed sfGFP.

The E. coli colonies were picked and grown for 12h in LB media containing antibiotics in culture tubes at 37°C and 250 rpm in an incubator. Next, the bacterial cultures were diluted at a ratio of 1:100 with fresh M9 medium with antibiotics and without inducers, after which the bacteria were grown for another 12h. We measured the fluorescence intensity and OD600. Then the bacterial cultures were diluted with fresh M9 medium with antibiotics and 10mM arabinose at a ratio of 1:100 once again and were grown for another 12h. Finally, the fluorescence intensity and OD600 were measured once again.

The ratio of fluorescence intensity/OD600 was calculated using the net fluorescence intensity and net OD600. We regarded the ratio greater than 100 as “light” and the one less than 100 as “dark”. The ratio of every well is shown below.

A
Animation in 96-well Plate
Animation in 96-well Plate


B
Animation in 96-well Plate
Animation in 96-well Plate

Figure 6.The pattern which transformed “2017” to “PKU”.A. When the culture had begun for 12h, we got the pattern “2017”. B. When the culture had begun for 24h, we got the pattern “PKU”.

We have confirmed that our genetic circuits can work as designed through the “ 2017-PKU” pattern transition. Next we constructed the system named “7 segment display”. The details of the circuits in segments are shown in the table below.

Figure 7.The design of "1-2-3" transition


We got the figure below on cloud nine. The numbers in figure were fluorescence intensity of corresponding wells, which were measured using the microplate reader mentioned above and adjusted as the same methods. We defined the wells as “on” when its fluorescence intensity exceeds 50; otherwise defined as “off”. As plain as the nose on your face, we managed to make it. We realized the sequential logic!

Figure 8.The experiment result of "1-2-3" transition

References

1. Dyakonov T, Muir A, Nasri H, et al. Characterization of 582 natural and synthetic terminators and quantification of their design constraints[J]. Nature Methods, 2013, 10(7):659-664.
2. Brophy J A, Voigt C A. Antisense transcription as a tool to tune gene expression[J]. Molecular Systems Biology, 2016, 12(1):854.
3. Yang L, Voigt CA, et al. Permanent genetic memory with >1-byte capacity. Nat Methods. 2014 Dec; 11(12):1261-6.
4. Stothard P. The sequence manipulation suite: JavaScript programs for analyzing and formatting protein and DNA sequences. Biotechniques. 2000 Jun; 28(6):1102, 1104.
5. Lou C, Stanton B, Chen YJ, Munsky B, Voigt CA. Ribozyme-based insulator parts buffer synthetic circuits from genetic context. Nat Biotechnol. 2012 Nov; 30(11):1137-42. [PMID:23034349]

Back to top

Clock

Background

Clock Signal

A clock signal is a particular type of signal that oscillates between a high and a low state repeatedly. This signal is emitted by a clock generator and is utilized like a metronome to trigger actions of a sequential logic circuit. In one clock signal cycle, the state of the circuit, which is the basis of sequential logic circuits, remains unchanged after shifting. A sequential logic circuit designed with an inner clock can change its state automatically, even without inputs. Additionally, the time a state lasts can be controlled by the clock signal period.


Figure 1. Typical clock signal and fluorescence signals of the repressilator.

Repressilator Introduction

The repressilator is a synthetic genetic regulatory network consisting of a ring-oscillator with three genes -- tetR from the Tn10 transposon, cI from bacteriophage λ and lacI from the E. coli lactose operon, each expressing a protein that represses the next gene in the loop.


Figure 2. The structure of the original repressilator. It is encoded by a low-copy plasmid with three repressor genes and a high-copy plasmid containing a GFP gene driven by P_LtetO1.

The Repressilator as an Internal Clock

In our project, we aimed to use the repressilator to generate a clock signal. We chose it for the following reasons:

1. It can generate up to three oscillatory signals in different phases; two different signal peaks can drive the flip-flop to shift its state serving as a component of a period.
2. The period and amplitude of the oscillation can stay stable and robust through many generations.
3. The period of the oscillation is adjustable. We can modify the oscillation period by simply chang-ing the bacterial generation period and oscillator components.
4. It has the potential to be connected to a downstream circuit. Because of the oscillatory changes in promoter strength, we can add the promoters (P_R ,P_LtetO1 ,P_LlacO1) to the downstream circuit.

Aims

Compared with natural biological systems, synthetic circuits such as the repressilator have a lower accuracy of oscillation. There have been many efforts to enhance the robustness of oscillation of the original repressilator, such as improvements in the core cellular processes and the reporter systems . Thus, if we want to use the repressilator as an internal clock, we have to determine whether our construct would have a significant influence on the oscillation itself. To answer this question, we have to do the following experiments:

Key Achievements

Our Approach

Reproduction of the Oscillation of the Original Repressilator

We used a stereoscope, a microfluidic device and a plate reader to reproduce the oscillation experi-ments of the parent repressilator (single reporter oscillator and three-reporter oscillator). The main strains we used in the oscillation-related experiments are listed in Table 1. The clpXP genes in these strains were deleted to obtain more robust oscillations.

Table1. The strains used in the oscillation-related experiments.

Since one of the repressors is under the control of P_LlacO1, we can use IPTG (Isopropyl β-D-Thiogalactoside) to synchronize the oscillation in different cells. We also compared the synchronized oscillation and the unsynchronized oscillation. (see Results figure)

If we want to look at dynamic changes of the oscillation, population-based experiments are not enough, so we also designed a microfluidic device of the same type as the "mother machine" used in our lab to house single cells and their offspring. In this device, cells are trapped in narrow channels and newborn cells are washed away by fresh medium. (see Results video1)

Figure 3. Design of the "Mother Machine". It consists of dozens of thin channels holding single cells, and can be used to track individual cell line for generations.

Design of the Recombinase Oscillator

In order to test whether our modifications of the repressilator would severely affect its oscillation pattern, we combined the repressilator with recombinases (Bxb1 gp35) to obtain an oscillator-activated recombinase expression system. By adjusting this system, we can also determine whether the oscillation period is compatible with the recombination time.


Figure 4. Schematic diagram of the recombinase oscillator.

For the expression system of the recombinase oscillator, we made use of the plpt20 plasmid (Laurent et.al., 2016), and we cloned Bxb1 gp35 downstream of mVenus in plpt20. By doing this, the expression of recombinase was coupled to the expression of mVenus, which is under the control of the promoter P_LtetO1.
As for the reporter system used to detect recombinase expression, we constructed a revertible promoter surrounded by a Bxb1 attB site and a Bxb1 attP site on the sponge plasmid. When the recombinase is expressed, it inverts the promoter sequence and mRFP is expressed.
Both the expression and reporter systems of the recombinase oscillator were constructed using Gibson assembly.

A

B

Figure 5. Plasmids of the recombinase oscillator. (A) Plasmid encoding the oscillatory recombinase expression system (plpt20-gp35). (B) The plasmid encoding the oscillatory recombinase reporter system (plpt41-reporter).

Possible improvements

Design of the T7-coupled Oscillator

The repressilator is constantly working as an internal clock, but under most circumstances we would like to start the internal clock in a specific moment. Using IPTG to synchronize the internal clocks of all the cells in a population and keep them in the same phase seems plausible. However, through our experiments with the recombinase oscillator we found that using IPTG cannot prevent signal leakage of the internal clock. Thus, we would like to add an "AND" gate behind the clock signal, so that we can turn on or turn off the internal clock.
T7 RNA polymerase has been previously used to implement an "AND" gate. In that design, the first promoter controls the expression of a T7 RNA polymerase gene with two internal amber stop codons. The second promoter controls the amber suppressor tRNA supD. When both promoters are activated, T7 RNA polymerase is expressed and can act specifically on the downstream T7 promoter. What's more, T7 RNA polymerase can be used as an amplifier to increase the signal amplitude.

Our aim is to prove that T7 RNA polymerase can be combined with the repressilator and can serve as a potential on-off switch and amplifier. Therefore, we designed a T7 oscillator and recombinase amplifier.


Figure6. Schematic diagram of the recombinase oscillator.

A

B

C

Figure 7. Plasmids of the T7 polymerase oscillator. (A) The plasmid encoding the T7 oscillator expression system (plpt20-T7). (B) The plasmid encoding the T7 oscillator reporter system (plpt41-T7 test). (C) The plasmid encoding the recombinase amplifier system.

For the expression system of the T7 oscillator, we also used the plpt20 plasmid (Laurent et.al. , 2016). We cloned the T7 RNA polymerase right behind mVenus in plpt20. To detect the expression of T7 RNA polymerase, we added a T7 promoter before mRFP in the plpt41 plasmid. In order to prove that T7 polymerase can be used as an amplifier, we also constructed a recombinase amplifier. The re-porter system of the recombinase amplifier was adapted from the plpt41-reporter; we added a Bxb1-gp35 activated by a T7 promoter in the plpt41-reporter plasmid.

Microfluidic Image Analysis

Firstly, we opened the files from the microscope generated in nd2 format in NIS element viewer or image J plugins NDi6d. We were then able to change them to videos or pictures in Tag Image File Format. Finally, we used image J to analyze the intensity of the images in TIF format.

Results

Recurrence of the Oscillation in the Repressilator

Figure 8. Stereoscopic image of colony oscillation. Triple reporter Plpt107+titration sponge. We repeated the experiments of the original repressilator (John et al., 2016,) using the LPT117 strain. By using a stereoscope, we discovered concentric yellow fluorescence circles in the colony of LPT117. This spatial pattern is the reflection of temporal pattern of the oscillation. It is formed when the fluorescence signal oscillates during cell division, and inner parts of cells in the colony cannot divide and remain the fluorescence signal.
We repeated the experiments of the original repressilator (John et al., 2016,) using the LPT117 strain. By using a stereoscope, we discovered concentric green fluorescence circles in the colony of LPT117.

A

B

Figure 9. Typical time trace of a single cell for original repressilator. A. Time trace of LPT117 repressilator (plpt107+titration sponge). B. Time trace of single reporter repressilator (plpt20+titration sponge). Both traces were normalized to their means. The oscillation period of LPT117 is around 510 minutes in LB medium,14 generations ( cell division period is about 35 minutes). The oscillation period of the single reporter oscillator is also around 15 generations, 450 minutes (cell division period is about 30 minutes).

Recombinase Oscillator

Figure 10. Optical microscopic views of a bacteria colony with a ' disk-like ' pattern. An optical microscopic view of a bacterial colony with a ' disk-like ' pattern, the red parts of the colony are distributed in wedge form. When recombinase Bxb1-gp35 expresses, the promoter of mRFP surrounded by Bxb1 attB and attP sites is inverted, which leads to the expression of mRFP.Some cells shift from yellow to red because of the expression of Bxb1-gp35, and its offsping, which are distributed in wedge form, kept expressing mRFP..


Figure 11. Time trace of a single cell for recombinase oscillator. The oscillation period of recombinase oscillator is around 300 minutes, 14 gene in LB medium (cell division period is 20 minutes), which is similar to the oscillation period of LPT117 and single reporter repressilator. These results shows that our modification on the parent repressilator (combine recombinase with repressilator) will not affect the clock itself and clock signal can be passed from the repressilator to downstream structure (invert of mRFP promoter).


Video 3. Microfluidic video of Recombinase Oscillator.

References

1. Elowitz, M. B., & Leibler, S. (2000). A synthetic oscillatory network of transcriptional regulators. Na-ture, 403(6767), 335-338.
2. Potvin-Trottier, L., Lord, N. D., Vinnicombe, G., & Paulsson, J. (2016). Synchronous long-term oscilla-tions in a synthetic gene circuit. Nature, 538(7626), 514-517.
3. Wang, P., Robert, L., Pelletier, J., Dang, W., Taddei, F., Wright, A., & Jun, S. (2010). Robust Growth of Escherichia coli. Current Biology, 20(12), 1099–1103.

Back to top

Demostrate

Background & Design

We attempts to develop a framework of biological sequential circuits that are programmable. The envisioned circuit is capable of both storing states in DNA and automatically running a series of instructions in a specific order. More specifically, the sequential logic (or a real-state machine) that consists of a clock (trigger signal), flip flop (remembering device) and control unit (functional part) in bacteria. The clock is an oscillator with a repeated signal cycle that is utilized like a metronome to trigger actions of sequential logic circuits. Flip-flop is a memory device that can remember a state. Paired with a clock signal, it can realize state transition. The control unit is a functional part which can convert a signal from flip-flop into complex functions. With such a design, historical events are recorded and influence current cell behavior.

Methods

The bio-flip-flop is constructed as two sequences each with an invertible promoter flanked between recombination sites and expressing recombinase or recombinase-RDF fusion protein mediating inversion of another sequence.

Fig.1. Bio-flip-flop construction.

We first tested and optimized recombinase and recombinase-RDF fusion protein functions and we standardized these parts into BioBrick devices. Then we separated the function of bio-flip-flops and built two circuits: the Forward and Backward latch. Each of them can implement state change when receives a group of trigger signal consisting of two inducing events. We then introduced these latches in vivo and proved their functionality.
Check the FLIP-FLOP module for more details.




Fig.2. Construction of Forward and Backward latches of the bio-flip-flop and their function.

Unidirectional terminators whose directions are changed by recombinases or recombinase-RDF fusion proteins are used to control transcription of the target gene. We measured the strengths of 24 terminators in both directions, and we elected six suitable unidirectional terminators. We measured the strengths of several terminators paired with attB/P sites and the inversion efficiency of three recombinases.
Check the CONTROLLER module for more details.



We used repressilator, a classic genetic oscillator, as an internal clock to drive our bio-flip-flop. To do so, we first observed and evaluated the repressilator system in our own experimental environment. We then constructed an oscillator-induced recombinase expression system and observed promoter inversion, which shows that repressilator has the possibility to act as an internal oscillatory trigger signal generator.
Check the CLOCK module for more details.

Results

Following the methodology of modularity in synthetic biology, we divided the system into three key components. We showed the feasibility of their assembly into a whole system by a set of simulation results.

First, an ordinary differential equation system describing recombinase and RDF function was built. Parameters were estimated from experimental data. Then we assembled four components (two recombinases and two fusion proteins) as the one in flip-flop design. Inducing of the system is in the form of square waves to simulate manual induction like changing of medium.





Fig.3. Simulation of the whole flip-flop function. X_F denotes the fraction of forward direction (right) of the upper promoter pBAD. Y_F denotes the fraction of forward direction (right) of the lower promoter pTAC.

We also proved that the system can also be triggered by repressilator as internal clock signal. Substituting the upper promoter (repressed by tetR protein) to PLtetO1 and the lower one (repressed by cI protein) to PR, we then did another simulation of forward latch and proved that the flip-flop can change its state and then recover back to initial state after arrival of two adjacent trigger signal groups.

Discussion

In this simulation experiment, the flip-flop lost approximately 10% of its initial state after two trigger signal groups. If we use manual and discrete induction, the loss is roughly 5%. These results have reached the expectations. However, in practical experiment, there are still many problems to solve and further research is needed. Though we constructed and configured a set of recombinase and recombinase-RDF fusion protein tools, we observed leaky expression and other problems when assembled. Further modification of repressilator (adding one more recombinase) has to be tested to satisfy the need of in vivo automatic state transition.

Potential Applications

This work tries to point the way toward building large computational systems from modular biological parts—basic sequential computing devices in living cells—and ultimately, programming complex biological functions. We also illustrated the possibility of self-triggered state transition. Possible application would be an automatic cell factory that can complete a series of manufacturing procedures with just one trigger at the start. Furthermore, our bio-flip-flop can act as an in vivo memory device and aid in recording cellular events, which can provide deep insight into cell differentiation and development.




Back to top