The Motivation: A Self-Replicating COVID-19 Test Kit
When our team started to brainstorm what we wanted to spend this summer creating, we were in the midst of the COVID-19 shutdowns. We had all been watching the pandemic for weeks from our homes, observing shortcomings in our health care system that prevented efficient handling of the virus. One of those failures was lack of testing availability, which made isolation and contact-tracing difficult.
There were two main factors that led to the problems with testing. The first is that the standard tests require professionals to analyze samples and interpret results, but there are limited personnel available to do so. The second is that testing requires expensive reagents, hardware, and lab equipment - all of which are limited.
By creating a cell that is a live test that can infinitely replicate itself, be grown anywhere around the world, and wouldn't require any expertise or lab equipment to administer or read, we could solve both of these problems. By using the natural scalability of biology.
How Would Our Test Compare To Other Tests? [Source]
Compared to other tests, SEED has many advantages. Because the test would simply be a cell - which scale naturally simply by existing and growing - it would be scalable. To replicate, it would require nothing but sugar, water, and air to produce, and so it would be extremely affordable. It wouldn't require any lab equipment to administer or interpret, making it self-administrable. The way that the genetic constructs are designed make is interchangeable and allow for simple creation of a cell that detects any nucleic acid sequence. And finally, the test would be rapid, giving a readout at most a few hours after addition of a sample, less with signal amplification.
The possibility of creating a test like this, that could far exceed the capabilities and desirability of other tests, was strong motivation to our team, particularly within the context of the COVID-19 pandemic.
System Overview: Three Phases From Input To Output
At a high level, there are three phases in the way SEED works:
1. Nucleic acid uptake through natural competence of B. subtilis.
2. Intracellular detection through either RNA toeholds or homologous recombination.
3. Intercellular signaling through the quorum-sensing pathway.
The result is a rapid, colorimetric readout in response to a target.
Phase 1: Uptaking Nucleic Acids From The Environment
Once we had decided on our project, the next step was deciding which micro-organism to use for our live cell diagnostic. Ultimately, we ended up using Bacillus subtilis (a GRAS gram-positive bacterium) due to its unique natural ability to uptake DNA from its environment. By using a bacteria that is already able to uptake nucleic acids from its environment, we eliminated the need to engineer a new pathway to get DNA inside the cell.
How Does The Natural Competence Pathway Work?
Competent B. subtilis cells express specialized proteins that assemble into a DNA transport complex embedded into the cell membrane. Some of these proteins are ComG proteins, which must be present for DNA binding (the first step in the transformation process) to occur.
1. dsDNA irreversibly binds to the cell surface receptor ComEA
2. The dsDNA is cleaved by the NucA nuclease
3. The end of one strand initiates transport across the membrane through the channel protein ComEC and the ATPase ComFA
4. As the ssDNA enters the cell, special binding proteins are attached.
Once the DNA is fully transported into the cell, subtilis will integrate it into its main genome. This is another ability that is unique to Bacillus subtilis. In other competent cells such as E. coli, the eDNA will exist in the cell as a plasmid, separate of the main genome. (Maier et al).
Figure 1: DNA Transport Complex From Natural Competence Pathway In B. subtilis
How Is Competence Regulated?
It is believed that the natural competence ability of B. subtilis evolved as a response to nutritional stress. When nutrients are limited for prolonged periods of time, subtilis cells enter a stationary growth phase, where they release degradative enzymes like proteases that can break down surrounding resources to release nutrients. These enzymes are able to break down DNA and use it for the cell's metabolic processes. Thus, a pathway to bring DNA into the cell evolved, so that it could be broken down for metabolic purposes. (Hamoen et al).
Figure 2: Overview of Regulation in B. subtilis
The regulation of competence in B. subtilis is extremely complex, as seen in Figure 2. However, two important molecules are ComK and ComX. Expression of the DNA binding, uptake, and recombination proteins necessary for competence are controlled by the transcription factor ComK, which is expressed only when the exponential growth phase ends (van Sinderen et al). The expression of ComK can be stimulated by the pheromone ComX, which is sent from other cells via the quorum-sensing pathway, in response to increasing cell density. (Hamoen et al).
Through understanding the genetic circuits involved with regulation of competence expression, researchers have been able to develop strains with inducible supercompetence in response to a variety of different inputs. In our project, we took advantage of this and used strains that had easily inducible supercompetence.
Strain SCK6 : Xylose-Inducible
In our toehold detection experiments, we used SCK6. In this strain, the ComK gene has been placed under the control of the xylose-inducible promoter PxylA. Adding xylose to B. subtilis cultures at 1% (w/v) causes the expression of DNA transport proteins, effectively "turning on" competence ability. (Zhang et al).
Strain REG19 : Mannitol-Inducible
When testing our recombination-based system, we used REG19. In this strain, ComK has been placed under the control of the mannitol-inducible promoter PmtlA, along with another competence gene, ComS. With REG19, adding mannitol to B. subtilis cultures at 3% (w/v) turns on competence ability. (Rahmer et al).
We chose the bacteria B. subtilis to make our live cell diagnostic in, in order to take advantage of its natural competence. The specific strains that we used (SCK6 and REG19) have been engineered to have inducible super-competence, where addition of certain sugars can act as on "on-switch" for the competence machinery.
Phase 2: Detecting the Target Sequence Inside the Cell
Once the DNA has traveled inside of the cell, the next step is getting intracellular detection. We tried two different detection methods inside of our cells, one using RNA toehold switches, and the other using B. subtilis' ability to integrate environmental DNA into the main genome through homologous recombination.
Detection Method 1: RNA Toehold Switches
RNA toehold switches are simple nucleic acid detectors able to be produced by cells, and operate both in vitro and in vivo. They are simply strands of RNA, but engineered in such a way that the strands fold in on themselves to form a hairpin loop. The RNA strand is composed of five parts: a trigger, RBS, start codon, linker, and reporter.
Under normal circumstances, when the RNA strand is folded into the hairpin loop, the RBS and start codon are blocked so that translation of the RNA strand cannot occur. However, when the target sequence comes near to the RNA strand, it binds complementary to the trigger, causing the hairpin loop to unravel. With the RBS now exposed, ribosomes are able to bind to the RNA toehold, resulting in transcription of the reporter. To see a video visualization of this process, click on this link>
Mechanism of the Toehold Switch
1. Easy Interchangeability
It is very simple to adapt RNA toeholds to detect different sequences, and to give different outputs. To change the nucleic acid sequence that the RNA toehold is looking for, all one has to do is swap the trigger sequence. Similarly, to change the visual readout of the toehold switch, all one has to do is swap out the reporter gene.
2. High Specificity
RNA toehold switches have been shown to detect target sequences with high specificity. (Isaacs et al). This is because the toehold is engineered so that binding to itself and creating the hairpin loop is a very thermodynamically favorable conformation. However, in order to unravel, the RNA strand binding to the target sequence must be more thermodynamically favorable than it binding to itself. That level of favorability is difficult to reach, and is often only possible if the target sequence is an exact complement of the trigger sequence. In fact, toeholds have been shown to have single-nucleotide sensitivity (Hong et al).
3. Can Detect Multiple Targets In One Cell
Due to the relatively low metabolic tax of production, cells can express multiple toeholds with various different targets and readouts, allowing for more complex analysis of environmental nucleic acids than if the cell was only able to detect one target sequence at a time.
4. Produced In Vivo
Toeholds have been shown to work before in vitro and in vivo. They are produced in vivo by a cell's normal metabolic and genetic processes. This is especially relevant to our project, where our goal is to create a cell that produced everything necessary for diagnosis of a virus, or detection of a nucleic acid sequence. It is important to note, however, that while RNA toeholds have been shown to detect accurately inside of cells, these were E. coli and not B. subtilis. Our research would be the first proof-of-concept for RNA toeholds working in vivo in B. subtilis.
Our Toehold Design
We designed our toeholds for integration on the pVeg YFP plasmid we created (description on Parts Registry here). The full toehold sequence is documented here.
The trigger strand is designed to be complementary to a section of the KanR gene, which we chose as our target sequence due to the availability of kanamycin resistance genes for us to use for free. For the RBS, we used a sequence specific for ribosome-binding in B. subtilis. For our reporter gene, we chose YFP due to its fast diffusion time and bright yellow fluorescence.
We plan to insert the toehold sequence into the plasmid right after the pVeg promoter and right before the YFP sequence. Because pVeg is the most active promoter in B. subtilis, this means toehold expression will be maximized. Putting YFP directly after the toehold allows it to be the reporter sequence.
We chose to test RNA toehold switches as one of our detection methods due to their interchangeability, specificity, in vivo production, and B. subtilis' ability to produce multiple at a time. We were able to start testing one toehold, which we had designed to integrate into our pVeg YFP plasmid and express YFP in the presence of the KanR gene, which we chose as a target arbitrarily due to availability.
Detection Method 2: Homologous Recombination
Another approach we investigated is based on the recombination system used in the miniBacillus project and outlined in Altenbuchner 2015. This approach takes advantage of B. subtilis' natural ability to integrate environmental DNA into the main genome through homologous recombination.
First, we transform a plasmid containing a negative selection marker that is flanked by regions of homology to the target sequence. When the cell takes up the target sequence, a recombination event is triggered, causing the gene encoding the negative selection marker to be excised and replaced by the incoming target sequence.
If cells don’t uptake the target sequence, then recombination does not occur. And if recombination does not occur, then the negative selection maker will remain in the cell, leading to its death when exposed to mannose. Therefore we can tell if the target sequence is present by whether or not the cell dies when exposed to mannose.
What is the negative selection marker, and how does it lead to cell death?
Our negative selection marker is ManP (the Part can be found described in further detail in the Registry here). It is a good selection marker for us to use because it provides a binary signal (the cells grow if it's not present and don’t if it is) and its activity as a selection marker is well understood and described in this paper: Altenbuchner 2015.
ManP acts as a selection marker by phosphorylating mannose that enters the cell, which prevents mannose from leaving the cell. Usually this would be fine, because manA could metabolize the mannose by turning it into fructose. However, when there is no manA expression in the cells (like in the case of the REG19 strain we used, which had manA knocked out), the phosphorylated mannose never gets metabolized, and the buildup of mannose causes the cell to die.
How is ManP excised from the genome in the presence of the target sequence?
The presence of a target sequence triggers a homologous recombination event, because we have engineered our strain of B. subtilis with two regions of homology to the target sequence incorporated into its genome on either side of the ManP cassette. When the target sequence is brought into B. subtilis through its competence system, it is brought towards B. subtilis’ chromosome for integration, where it then binds to regions of homology. After the target sequence binds to the regions of homology in the genome, a recombination event is triggered.
ManP is excised from the genome due to this recombination event. The target sequence binds to the arms of homology on either side of the ManP cassette, which results in the target sequence replacing the cassette and excising it out of the genome. This recombination event can only be triggered in response to the target, because the homology arms surrounding the ManP cassette (which we designed and put in the Parts Registry here) are designed to match the target sequence.
What are the benefits of the recombination-based method?
This method takes advantage of a process that is already occurring inside of our cells. B. subtilis naturally wants to recombine DNA it brings in via with competency machinery. Because the method relies on a natural process that is already present in un-engineered B. subtilis, we are confident that this system will work.
The recombination-method is also very versatile in that it is possible to excise any gene in you want in response to nearly any target sequence you want (limitations discussed below).
What are the downsides of the recombination-based method?
There are a couple of potential downsides to this method.
1. Much of the literature suggests using long homology arms to trigger recombination events like the ones this system relies on. This makes a recombination-based system for detection unable to detect target sequences that are less than a few hundred base pairs.
2. It is unclear at this time how specific a recombination-based system can be. For example, homologous recombination can be triggered even when sequences are not exactly identical, meaning it is almost very difficult to have single nucleotide specificity for detection systems like this. These issues may result in higher false positive rates for systems like this one.
3. Using manP for a reporter system, as we have done here, limits the speed at which the test can give results because the system modulates cellular growth as its readout (this modulation is only apparent after at least hours). This, however, could be addressed by using a different reporter system in conjunction with a recombination-based system, such as excising a repressor for a fluorescent protein, which would result in a colorimetric readout.
ManP Plasmid Design
Target Sequence (To Trigger Recombination)
We used a region in the pOpen Yeast plasmid from Free Genes between the AarI and MspA1I cut sites. We used this region because it was readily available to us in quantity - the sequence can theoretically be anything you want.
Each region of homology flanking the negative selection maker is around 500 bp. 500bp is the length of homology conventionally used when looking for high efficiency recombination in B. Subtilis, even though some report relatively high efficiency with shorter regions of homology.
In addition to RNA toeholds, we also decided to test a detection method based on homologous recombination, which we were more confident would work because it uses a system that is well-documented in the literature, and shown to work. It works by excising the ManP cassette only in the presence of the target sequence. If the cells continues growing after exposure to mannose, then the target sequence was detected and integrated into the genome.
Phase 3: Amplifying the Signal For a Faster Readout
The next phase after we are able to get a visible readout from detection - whether it is with the toehold or ManP systems - is trying to amplify the signal to be as fast as possible. Over the summer, we developed an idea of how to do this and designed and ordered genetic constructs to do so. However, because we had limited time in the lab (we were not able to get in until August), we were only able to show a proof-of-concept for Phase 1, and had begun but not finished testing the detection systems in Phase 2. We will describe Phase 3 of SEED here, but we were not actually able to test out the ideas or constructs that we ordered.
Hijacking the Quorum-Sensing Pathway
A signal amplifier is important because it causes the system to produce a measurable readout (i.e. colorimetric, pH-change, conductivity) throughout the entire bacteria colony upon detection of a small amount of target nucleic acids. Our research question in designing an amplifier was: how can we get entire bacterial colonies on the same page with one another if only a few bacteria detect a target sequence? The answer is to use a bacterial communication system which already exists: the quorum sensing pathway.
Normally, once B. subtilis cells reach a critical population density, their quorum sensing molecules activate global pathways to produce community-wide goods. One such quorum molecule is ComX which, at a critical density, activates pathways that result in the activation of the srfA-promoter (PsrfA). PsrfA activation upregulates the downstream production of surfactant molecules, resulting in a biofilm that is contributed to by every cell in the colony.
The novel pathway we engineered takes the positive feedback loop of ComX from the quorum-sensing pathway, and rewires it to cause cells to produce fluorescent proteins instead of surfactant molecules. The result is a rapid, colony-wide colorimetric readout.
Novel Pathway Design
The first step was knocking out endogenous ComX in our B. subtilis strain. By doing that, we were then able to put ComX under the expression of whatever promoter we wanted. In the Signal Amplification construct that we designed (linked here in the Parts Registry), the ComX and luxABCDE genes are under the control of the srfA-promoter, PsrfA.
The reasoning is this: once a critical amount of cells are exposed to the ComX molecule, their PsrfA genes are activated to produce even more ComX and luxABCDE (a bioluminescent reporter of our choice). The positive feedback loop would cause bioluminescence to increase exponentially, resulting in a rapid readout.
Integrating ComX with Toeholds and ManP
The system is integrated by plugging the detector into the signal amplifier. Upon detection of a specfied DNA or RNA sequence, whether by the toehold-switch or recombination method, comX will be conditionally-translated protein. If the detection method is the toehold-switch, upon target-sequence binding to the hairpin will unfold and allow for the translation of comX. If the detection method is homologous recombination, the target sequence will recombine on landing pads to excise a terminator upstream of comX, allowing for comX transcription. Because the signal amplifier pathway allows comX to upregulate its own production in addition to a colorimetric reporter, once a small amount of cells to detect a target sequence, a quorum sensing cascade spreads throughout the whole colony to produce a visual readout./p>
We have not yet been able to start testing the constructs for Signal Amplification, but the idea is still an important part of our project and the SEED system. The new pathway would be able to amplify the detection signal from either the Toehold or ManP system by using ComX (originally to the quorum-sending pathway) to stimulate a uniform response across the entire colony.