Abstract

Biosafety and security is the key to the new generation of synthetic biology. Genetically Modified Organisms (GMOs) have shown us much potential for solving global issues. Yet, the release of GMOs in the natural environment has been a huge safety concern, inhibiting the implementation into the broader society. Also, in order to encourage innovation, it is crucial to protect intellectual property, in particular the gene sequence of GMOs, and GMOs themselves, from malicious interest and dual-use. This year, we present Optopass, an optogenetic passcode system for microorganisms. In this system, S. cerevisiae produces the target substances only when exposed to the correct color of light in the correct order. However, when the light is applied in the wrong order or the culture is exposed to the sunlight, a kill switch is activated. To implement this system, we used optogenetics for controlling S. cerevisiae by different colors of light, such as blue, red, green, and ultraviolet (UV-B), and site-specific recombination technology, such as Cre-loxP for storing the order of optical stimulus. Optopass proposes a foundational system for greater safety and security.

Why optogenetics?

Optogenetics is a technology that manipulates cellular functions using proteins activated by light of a specific wavelength. Optogenetics can be applied to turn on or off the expression of specific genes with various colors of light.

We aimed to construct a yeast optical passcode system by controlling gene expression with proteins activated by blue, red, green, or Ultraviolet-B (UV-B) light.

For more information, see our Results page.

Blue Light

For blue light acceptance, we used a system based on a protein called EL222. This is the same one that iGEM NUS Singapore 2021 used for its project. EL222, which consists of a light-oxygen-voltage-sensing (LOV) domain and a DNA binding domain, helix-turn-helix (HTH) domain, is connected to an activation domain (AD), VP16. In the dark, this complex does not bind to DNA because the LOV domain masks the HTH domain. When blue light (450 nm) hits them, the structure of the LOV domain changes, freeing the HTH domain, which binds to each other to form a homodimer that binds to the C120 region in the DNA. The VP16 domain then promotes the expression of the downstream gene [1][2].

In the previous study, a sequence of five C120 regions was used, but it was reported that the expression level did not change much when the number of these repeats was reduced under the conditions of the current experiment [3]. Considering the actual use in the real world, we decided to use only one C120 region because it is more convenient to use a simple construct that does not contain any repeats. If it becomes possible to easily construct five consecutive C120 regions in the future, the system may be improved not only by increasing expression but also by reducing basal expression.

Red Light

For the acceptance of red light, we used a system based on PhyB and PIF3 from A. thaliana. PhyB, a photoreceptor, undergoes a conformational change when it accepts red light (660 nm) in the presence of PCB, a chromophore, so that it can bind to a protein called PIF3. When it accepts far-red light (740 nm) or stops being illuminated, it dissociates back to its original structure [4]. As shown in the figure below, by attaching a DNA binding domain (DBD) to PhyB and an activation domain (AD) to PIF3, the AD comes near to the promoter only when PhyB and PIF3 are bound, and the transcription of the downstream gene is promoted.

Regarding the choice of DBD and AD to realize this system, we considered the following three options.

(1) Use GAL4DBD for DBD and GAL4AD for AD. [5].
(2) Use synTALE for DBD and VP64 for AD. (PhiReX) [6].
(3) Use zinc finger for DBD and VP64 for AD.

(1) was used by iGEM TU_Munich 2012 [7], and (2) was used by iGEM NUS Singapore 2021 [2]. Of these, we chose to use (2), PhiReX. There are two reasons for this. First, the yeast strain we use already contains GAL4, so using GAL4DBD and GAL4AD as in (1) could interfere with the yeast’s intrinsic system [8]. In addition, the synTALE used in PhiReX does not contain a nuclear transfer signal (NLS), so PhyB-synTALE does not bind to DNA until red light activates it, resulting in less leakage or basal expression. In a system where the encryption is broken only when light is applied in the correct order, the suppression of unwanted gene expression when light is not applied was a very important feature. After we discussed our project with iGEM Chalmers-Gothenburg in a partnership meeting, we got a new idea of using zinc finger instead of synTALE like (3). Yet, after comparing the effectiveness of DNA binding, between synTALE and zinc finger in our wet lab, we decided to use (2) system. See our Partnership and Results page for details.

The completed system is shown in the figure below. When red light is irradiated, PhyB undergoes a conformational change and binds to PIF3, creating a protein complex. This protein complex moves into the nucleus and binds to the Jub1.1 region with the synTALE domain, where VP64 induces expression of downstream genes. When the complex is not bound to the region, transcription from the CYC1min promoter is kept low.

Consideration of the third color

We considered using green or UV-B light as the third color of light, which has little interference on proteins that react to blue and red wavelengths. We looked at the pros and cons of the two colors.

We were not able to conduct wet experiments, but based on our literature review and human practice, we determined that Ccas/CcaR, which can be regulated by green light, and UVR8+COP1, which can be regulated by UV-B, are suitable for our project.

Green

We adopted CcaS/CcaR as a system controlled by green light. See Integrated Human Practices page for other green light responsive proteins we have examined.

CcaS/CcaR undergoes an autophosphorylation-phosphotransfer reaction upon light irradiation of certain wavelengths. At 535 nm light, autophosphorylated CcaS transfers phosphate to CcaR, and CcaR activates transcription by binding its DNA binding domain to DNA. 670 nm light causes the reverse reaction. It is one of the most well-studied green photoreceptors, with understanding of the molecular mechanism and many cases of its use as an optogenetics switch. We have not found any eukaryotic applications, but we hypothesize that transcription can be physically inhibited if CcaR binds to the transcribed region by green light [9]. However, this would mean that red light would promote transcription, which is the same role as PhyB/PIF3. To separate the roles, we considered constructing a system in which CcaR inhibits the expression of a protein that inhibits the transcription of the target gene. In this case, the transcription is inhibited by red light, so it can be used separately from PhyB/PIF3. Once this issue is resolved, the ability of CcaS/CcaR to respond to both green and red will be useful in making passcodes more complex and creating a more robust security system.

UV-B

We adopted UVR8+COP1 as a system controlled by UV-B.

UVR8 is a homodimer in the dark, but when exposed to UV-B (280-315 nm) it monomerizes within minutes. COP1 is a light signaling regulator. These two elements interact primarily on the C-terminal VP motif of monomeric UVR8 and theC-terminalWD40 domain of COP1 within a few minutes [10].

UVR8 is very sensitive to UV-B, allowing optogenetic stimulation to occur below the safety threshold of cells [10]. Even after UV-B treatment for 3 or 5 s (inducing 55 and 70% dissociation of UVR8, respectively), cell viability was identical to dark controls [11]. In addition, UVR8 has little response to other wavelengths of light, such as blue and red. There is an example successfully regulating triple gene expression in a single mammalian cell with blue, red, and UV-B light. Hence, we can expect the accuracy of the response to the input [12].

Why order?

Our security system is novel in that it uses "order" rather than "combinations" of light colors. We often see attempts to control gene circuits using combinations, such as the "AND" circuit, in synthetic biology. Achieving sequencing is more difficult than that, but it is worth trying.

This feature contributes to the safety of the passcode. We aimed to incorporate the light order into passcodes for more robust security.

For more information, see our Modeling page.

Storing order

How to utilize the order of light stimuli? It seems to be difficult to utilize the order, compared to using a combination of stimuli. This is because in order to use "order" it requires information about which step (e.g. 1st, 2nd, ...) of input will be received next, and therefore it needs to store that "state" in the microorganism. It is also simply difficult to create a gene circuit that processes the sequence of stimuli. We looked for different ways to store the order. Strong candidates were dCas9 system and site-specific recombination system. After the literature review, human practice, and discussion with our partner team, Chalmers-Gothenburg, we decided to use site-specific recombinases such as Cre-loxP because of the following reasons:

In our system using recombinase, the state of which step of stimuli is received next will be stored in the genome sequence (The detailed description of construction is given below).

For the detailed reasoning process why we employed site-specific recombination system, please visit Integrated Human Practices.

Site-specific recombination system

When there are two specific recognition sequence sites in the same sequence, recombinases carry out "cutting out" sequences flanked by specific recognition sequences.

In the case of Cre, for example, two Cre proteins recognize and bind to the loxP site, forming a dimer. The two Cre-loxP dimers attach so that the two loxP sequences face in opposite directions to form a tetramer. Finally, the DNA is cleaved at the center of the loxP site, causing recombination by crossover. This can be used to remove DNA sequences between two recognition sequences.

Consideration of recombinases that can be used

There are 9 recombinases that are known not to interfere with each other. (Cre, Dre, VCre and Vika in the Cre system and Flp, Kw, KD, B2 and B3 in the Flp system [13]) For Cre, Dre and Flp, we designed the sequences so that we could actually be used in experiments. Cre and Flp have previous studies used in yeast, and we were inspired by iGEM team Hong Kong HKUST 2017 to use Dre.

Recognition sequences

Recombinase cuts out the sequence sandwiched between two recognition sequences. There are some mutant recognition sequences that recombine by the same recombinase, but they only recombine with the same mutant recognition sequences. Two kinds of recognition sequences are required for each recombinase (as described in the Details below), because if there are the same recognition sequences, there is a risk that undesirable recombination may happen between recognition sequences which are recognized by the same recombinase. For each recombinase, the recognition sequences that only recombine among identical sequences, which are used to prevent such unintended recombination, are as follows.

• Cre: Recombination occurs in loxP, lox2272, loxN, lox511 and loxFAS when the two recognition sequences are identical [14][15].
• Dre: Recombination occurs in rox, rox2 and rox12 when the two recognition sequences are identical [16].
• Flp: Recombination occurs in frt and f3 when the two recognition sequences are identical [17].

Recombinase-based cryptosystems

Site-specific recombination system allows the creation of gene circuits, as shown in the diagram above, to realize an ordered cryptosystem. As mentioned above, Flp and Cre are recombinases, and FRT, loxP and lox2272 are recognition sequences. In the gene circuits, Flp removes sequences between the FRTs and Cre removes sequences between the loxPs and between the lox2272s. It is necessary to use two recognition sequence variants, loxP and lox2272, for Cre. If the same one is used, there is a risk that unintentional recombination may occur between those four recognition sequences.

Consider the case where light is applied in the correct order (Red light→Blue light→Third light, note that Third light is described as Green light in the animation). Red light causes Flp to be expressed and the DNA sequence between FRTs to be removed. Blue light then causes Cre to be expressed and the sequence sandwiched between loxP and lox2272 to be removed. Subsequent exposure to the third light causes Reporter expression.

Consider the case where light is applied in the wrong order. For example, if the third light (green light in the animation) is applied after Red Light, the region between lox2272 is not removed and the kill switch is expressed and the yeast dies. In this way, the gene circuit was designed so that the kill switch is triggered if the light is applied in an incorrect order.

Extensibility

The cipher length is determined by the number of mutually non-interfering pairs of recombinase and the corresponding recombinase-binding sequence.

The example given in the previous section is for a cipher length of 3. In this example, when the inputs are applied in the correct order, Flp, Cre, and Reporter are expressed at each step in this order. For a longer cipher, more recombinases are required: another recombinase is required in addition to Flp and Cre for a cipher length of 4. As noted, 9 recombinases can be used, so the cipher length can be extended up to 10.

For example, if the cipher length is set to 4 so that Red light→Blue light→Third light→Blue light is the correct answer, the circuit would be as follows, by using Dre/rox in addition.

Kill Switch

Biosafety is a central issue in synthetic biology. It is essential to reduce the potential for leakage and make it significantly more difficult for the GMOs to survive in the uncontrolled environment so that they do not impact nature. While many kill switches are intended to "make organisms commit suicide," our kill switch also works in a "confidentiality" role. This is the novel usage of the kill switch. Our kill switch, which fragments the DNA sequence, is under the control of a light-activated promoter, and works when exposed to incorrect light stimuli including sunlight. This not only prevents GMOs from surviving in the natural environment, but also makes it difficult to recover the gene sequence from the dead microorganisms.

For more information, see our Safety and Security page.

Mechanism of Apoptosis Induction

To realize the system for safety and security, we decided to use Bax as a kill switch; Bax is a protein that induces apoptosis, and cells in which it is expressed cannot produce the target substance. It binds to mitochondrial voltage-dependent anion channels (VDAC) and promotes their opening and releases apoptosis-inducing substances such as cytochrome C from inside mitochondria. The released apoptosis-inducing substances activate caspase-3 and other enzymes, resulting in DNA fragmentation and cell death [18].

Why Bax?

Bax expression induces apoptosis by fragmenting the DNA sequence into multiples of 180 bp each [19]. So it not only kills the yeast, but also acts as a shredder, making it difficult for the sequence information to be read from the dead cells. For example, when a 10,000 bp DNA sequence is fragmented into 180 bp pieces, there are $56!×2^{55}>10^{91}$ possible original sequences that must be examined one by one, making a complete recovery virtually impossible.

As possible proteins to kill yeast, we also considered other proteins that disrupt the cell membrane or cleave DNA sequences, such as Cas3, Cas9, and nucA, for example [20]. However, Cas3 and Cas9 only cut specific parts of the DNA sequence and do not shatter the sequence as Bax does. Therefore, if they are incorporated into our security system as kill switches, the sequence information can be stolen from the yeast carcass, just as if they were killed by disrupting the cell membrane, leading to a loss of cryptographic reliability. Although nucA fragments DNA sequences in the similar way to Bax, we decided to use Bax because it has been studied more, and the reaction rate parameters were available from the literature during our design phase.

As described above, Bax is an ideal protein that works perfectly with the kill switch we are looking for in terms of social implementation as a security system.

Dummy System

Based on the human practice with Tohoku Medical Megabank Organization, we developed an advanced form of Optopass called a "Dummy System" and modeled it. The Dummy System aims to protect useful microorganisms on a "strain" basis, whereas the current Optopass method could only protect information gene-by-gene, by controlling the competitive relationship between a target strain (with the target gene) and a dummy strain (without the target gene) through an optical passcode system. The objective is still to protect confidentiality, but the Dummy System is more secure and versatile than the standard Optopass system.

For more information, see our Modeling page.

Encoding

First, let the target strain and the dummy strain coexist in one environment and place them in an ecologically competitive relationship. For example, it has long been known that the competition equation of Lotka Volterra, which describes the competitive relationship between two species in yeast strains, holds true [21].

Decoding

When an optical cipher is input into this target-dummy coexisting environment, the equilibrium is broken by the Optopass system pre-installed on the dummy side, and the stable coexistence state is terminated. If we look at this in the phase space of the dynamical system, the solution trajectory converges to a different stable point than before the input according to the parameter change in the competition equation of Lotka-Volterra with the input. As a result, if the input is correct, we obtain a colony consisting only of the target strain, whereas if it is incorrect, we obtain only a colony filled with dummy strains. The advantage is that if the input is incorrect, the amount of target microorganisms can be reduced to zero more reliably than with kills witch, thus achieving more robust protection. It is also more robust in that it is more difficult to isolate the target strain and read the sequence because the dummy strain is overwhelmingly larger than the target strain even before the light is applied. In addition, the standard Optopass system requires genetic modification of the microorganism to produce the substance to be protected each time, and protection is limited to substances that can be produced by microorganisms such as yeast that can be genetically modified with the Optopass system, whereas with the Dummy System The Dummy System, on the other hand, does not require major recombination of the target microorganisms themselves, and can be applied to a wide range of existing microorganisms, making it a widely applicable protection target.