Team:UofUppsala/Design

Design

Index

• Overview
• Nanobodies for detection
• First Generation
  • DNA-binding domain
• Steps towards an optimized system
  • Detection Module with an Integrated Amplification Step
  • Gram-positive host
  • Signal transduction and amplification
• Prototype
• References

How does our flexible biosensor work? Read more about the theory and the thought process behind our experiments here. To see the experimental results check Wetlab!

Overview

---

Our aim was to design a device that is able to recognize a target of interest and transduce a signal further. For this, we needed a membrane-bound sensory domain that generates a regulated response upon activation. For that, to started with a first generation system and gradually build up towards a more sophisticated system.

Nanobodies for detection

---

The use of therapeutic antibodies has gone through constant evolution during the last 50 years. The first generation began with the use of monoclonal antibodies (mAbs). Their immunogenicity and large size sparked the second generation, including antigen-binding fragments and single-chain variable fragments. Yet, these molecules had short half-life and upon aggregation they were immunogenic. The discovery of camelid heavy-chain antibodies initiated the third generation of antibodies. These molecules contain a single variable domain (VHH) that are named nanobodies, or single-domain antibodies (sdAb) (Figure 1).

Antibodies structurally contain two heavy and two light protein chains. Opposed to what it might be thought due to not having a full antibody structure, nanobodies have been shown to maintain antibodies’ capacity to selectively bind to a specific antigen, and in some cases are even more robust. Additionally, their small size makes them easier to transform into bacterial cells, hence they are ideal for research purposes.

Nanobodies possess great modularity, because they do not need extensive assembly or optimization, thus leading to a development of big diversity of fusion molecules, with a huge array of applications. In Figure 1 we can see a representation of a full-size antibody, a camelid heavy chain antibody and a nanobody. (1) (2) (3)

The small size and potential modularity of nanobodies inspired the initial ideas of our project.

Figure 1. From left to right: structure of an antibody, a camelid heavy chain antibody and a nanobody.

First Generation

---

A nanobody-coupled DNA-binding domain

Firstly, we chose the minimal (single component) system described in the article written by Chang et al. (4) using a fusion of a receptor to a DNA-Binding Domain (DBD) (Figure 2).

Figure 2. DBD fusion protein structure, including an N-terminal CadC-derived DNA-Binding-Domain (DBD), Juxtamembrane Linker (JM), Transmembrane Domain (TM) and a C-terminal caffeine-binding nanobody (VHH).

It contains a DBD naturally present in Escherichia coli, known as CadC. CadC is a protein found in the inner membrane as a monomer that dimerizes upon lysine binding in an acidic pH. This dimerization allows the binding of the intracellular domain to the operator region of the Cad operon on the chromosome, which promotes transcription downstream (Figure 3).

Figure 3. CadC-based DBD system, inactive on the left due to the absence of caffeine and active on the right leading to the expression of fluorescent proteins. Chang et al. used GFP as a reporter instead of our mRFP1 (4).

The receptor part is a caffeine-binding camelid nanobody (VHH-Caff) which is expressed on the inner membrane of E. coli in the periplasmic space. It dimerizes upon caffeine binding (5) thereby activating the fused DBD.

Although this system is limited to detection of small molecules that can transverse the E. coli outer membrane, generalizability to detection of extracellular large molecules such as proteins was demonstrated by using a different nanobody domain (that binds eGFP protein) and by removing the outer membrane and cell wall of E. coli to enable access to the ligand (eGFP) (4).

We have assembled two different plasmids that together produce the system shown in Figure 3. The first one expresses a codon-optimized version of the fusion protein designed by Chang et al. (4) shown in Figure 2. We cloned the gene in a medium copy plasmid instead of the low copy number plasmid of the Chang et al.

Our second plasmid encoded the pCadBA operator that contains the CadC-binding sequences Cad1 and Cad2 (7). Downstream, a reporter gene was encoded. We used mRFP (BBa_E1010) as the reporter for the testing of our system. The presence of caffeine in the environment triggered the dimerization of the monomers, the DBDs bound to the Cad1 and Cad2 regions, activating the transcription of the mRFP1 gene (Figure 3).

The fact that our reporter is on a different plasmid simplified the exchange and trial of different reporters and expression strengths. This contrasts with Chang et al. using a single plasmid for expressing the whole system. We also wanted to explore the effect of plasmid copy number on the system in order to regulate reporter expression.

Steps towards an optimized system

---

Animated overview of the optimized system

T7 lysozyme is constitutively expressed and inhibits T7 RNA polymerase (T7 RNAP) in a non-induced system. Target recognition leads to dimerization of nanobodies (VHH) which induces transcription of T7 RNAP and TEV protease. As T7 RNAP is translated, it induces transcription of the final elements of the system, Qβ viral replicase and β-galactosidase. Qβ has a LAA degradation tag after the TEV-tag. When TEV protease is present, this degradation tag is cut, allowing Qβ to amplify β-galactosidase as the enzyme holds recognition sites for the replicase. As β-galactosidase cleaves its substrate, X-gal, 4-chloro-3-bromo-indigo is produced. This compound generates a strong output signal by changing the color of the media. The overview animation is based on DBDs as the detection module but the below mentioned NarX/NarL system would work analogously.

Click on the animation to view it again and continue reading this page for a detailed description of the elements of the optimized system.

Detection Module with an Integrated Amplification Step

Searching for alternatives to expand our first generation system, we came across the two-component system NarX-NarL, intrinsically present in E. coli. The histidine kinase NarX is located in the inner E. coli membrane. The presence of nitrate induces a conformational change when binding to NarX which leads to phosphorylation of the NarX kinase domain. Further, the response regulator NarL is phosphorylated and upon conformational change induces a promoter leading to anaerobic respiratory gene expression (Figure 4A). The reaction is naturally amplified as there are several NarLs activated by a single NarX, which was our reasoning for choosing the system as it would contribute to higher sensitivity.

To make our system flexible, we redesigned the sensory domain by fusing NarX to a nanobody (VHH) of choice. The idea, again, is that the nanobody can be exchanged depending on the target of interest. Due to the orientation of NarX on the membrane, the only requirement is that the C-terminal of the nanobody needs to be fused to the N-terminal of NarX. Beneficially, this is the most common direction of nanobodies.

As we cannot easily design the conformational change of the sensory domain that would induce NarX phosphorylation, a modified NarX based on a study by Mazé and Benenson (8) was used instead. In this design, phosphorylation is only possible if two, otherwise inactive kinases are brought into proximity by the fused VHH. We utilized this design by fusing the cytoplasmic part of the modified NarX with the membrane domain of natural NarX to a nanobody. The fusion protein would in this way be located in the membrane, with the nanobody as a detection arm. The system would then be activated when two nanobodies interact with binding sites on the target that are sufficiently close to each other to bring the fusion proteins in proximity, a process called forced dimerization.

Testing the module in Escherichia coli

The system was first designed for E. coli, to show that the dimerization and signal transduction pathway works.

Analogous to the DBDs experiment in the first generation system, the anti-caffeine nanobody (VHHac) was fused to the modified NarX. The idea is that the fusion protein will activate upon binding to a caffeine molecule, leading to phosphorylation of the response regulator NarL. To verify the activation, the promoter of NarL, PyeaR, with GFP will be introduced into the cell. Hence, during the presence of caffeine the modified NarX/NarL system will activate and GFP will be produced by the cell, giving a visible green light that is quantitative (Figure 4B).

Gram-positive host

The system in Bacillus subtilis

When thinking about more ways to optimize the system, we came across the idea of using B. subtilis, a gram-positive bacterium, as the chassis for the final system. Since this bacterium only possesses one membrane, the detection of non-permeable molecules would be more easily feasible, as long as the target ligand can cross the thick cell wall (in case the ligand in question is too large to access the membrane, we address potential solutions in our Contribution page). Additionally, transferring the final system to B. subtilis would be benifitial in regards to shelf life of the kit, as this bacterial species has the ability to sporulate, acquiring additional robustness in harsh conditions.

Figure 4. NarX/NarL two-component system, A: in natural form in E. coli, B: in modified form in E. coli, C: in rewired form in B. subtilis.

After testing the NarX-NarL two-component system in E.coli, the detection module would then be implemented in B. subtilis. However, when transferred into this new chassis, the NarX-NarL-PyeaR signaling cascade has been shown to be silenced. Schmidl et al. (9), after coming across this phenomenon, managed to restore the activity of the NarX-NarL system by swapping the DNA-binding domain of NarL for that of YdfI. Since this protein is a response regulator native to B. subtilis, this change effectively rewires the detection module to the promoter PydfJ115, which is functional in this host.

The ydfHI locus encodes for the two-component system responsible for natural activation of PydfJ115. Hence, nullifying the action of these genes through a knockout would reduce the noise in the signal transduction. This loss-of-function mutant would be created by taking advantage of the DNA recombination machinery present in B. subtilis, just as described in our handbook: by flanking our inserts with homology regions for the ydfHI locus, we are able to simultaneously integrate our constructs into the chromosome and induce a targeted knockout. Additionally, constructs integrated into the chromosome are more stable throughout generations when compared to those contained in replicative plasmids, contributing to the reliability of the detection device.

We assembled two constructs: one is an operon harboring both the VHH-NarX and the rewired NarL under the control of the same constitutive promoter; the other is composed of the promoter PydfJ115 coupled to a GFP. This last construct, just as in the E. coli system, serves as our proof of concept for the detection module (Figure 4C). Later on, when assembling all the modules together, this GFP would be swapped by the amplification module.

Signal transduction and amplification

Pathogen biomarkers can be present in the sample in very low concentration. Upon detection with a biosensor, it is likely that low concentrations will result in a low signal. This signal could be masked by the noise of the system itself, thus yielding a high amount of false negative results. With the addition of an amplification system, it is possible to increase the signal-to-noise ratio, and therefore decrease the amount of false negative results and increase sensitivity (10).

β-galactosidase as a reporter

Perhaps the simplest way of signal amplification is to exchange a fluorescent reporter for an enzyme. Compared to fluorescent proteins, a lower number of enzyme molecules needs to accumulate to give rise to a noticeable signal. As one molecule of enzyme can process several molecules of substrate, the dynamics of the system also changes, which can make the biosensor faster.

β-galactosidase was chosen due to its ability to cleave its substrate, X-gal, giving a read-out signal visible to the naked eye in the way of a bright blue color. In the past, this reaction was widely used for monitoring cloning approaches in a method called blue-white screening (11). X-gal is a permeable, colorless compound that can be used to detect the presence of the enzyme β-galactosidase (12). Upon reaction with β-galactosidase, X-gal is cleaved and thereby oxidized into the blue product 5'-dibromo-4 4'-dichloro-indigo which gives a visible color change (13).

As an alternative reporter molecule to X-gal, ortho-Nitrophenyl-β-galactoside (ONPG) was considered which shows quicker reaction times according to the literature (14). This detection approach is thought to be fast, easy to incorporate with the residual biosensor system and safe to use.

Qβ viral replicase at the mRNA level

On top of using the enzyme as a reporter gene, the signal could be amplified even more by the activity of Qβ viral replicase (15). This RNA-dependent RNA polymerase amplifies the transcript of our reporter enzyme, β-galactosidase. This amplification is specific due to the presence of Qβ recognition sites at both the 3’ and 5’ prime ends of the β-gal mRNA. As a result, there is more β-galactosidase transcript available for translation, which means that more enzyme is produced and the colour change is stronger. The amplification efficiency can be further increased by knock-out of a gene coding for RNase III, an enzyme which is responsible for the degradation of double-stranded RNA (15).

Noise reduction elements

Due to the presence of this amplification system, a small leakage in the expression of Qβ replicase would also result in the amplification of the β-galactosidase transcript, and thus a false positive signal. Therefore, additional layers are included in the design to allow tight control of Qβ replicase expression.

The first layer of stringent Qβ-expression control is incorporating a T7 RNA polymerase into the system. This enzyme is known to be very specific for its promoters and it can be inhibited by its specific lysozyme. In the proposed system, the T7 RNA polymerase gene is induced by the detection module, and then T7 promoters control Qβ and β-galactosidase transcription. The basal expression of T7 RNA polymerase (caused by leakage or random activation of the detection module) is controlled by constitutively expressed lysozyme. Therefore, we will only see a positive signal when the expression of T7 RNA polymerase is higher than what the lysozyme can inhibit. The expression level of lysozyme will be fine-tuned with respect to the level of noise while taking into account single-cell variations in the population of biosensor cells.

The second layer of tight control over the amplification system is based on the degradation of the Qβ enzyme, due to the LAA degradation tag (BBa_M0050) present at the C-terminal end of Qβ. This causes the basal levels of Qβ to be rapidly degraded as the LAA tag triggers the fastest degradation among SsrA tags. The degradation tag can be cleaved off by TEV protease due to the presence of a TEV tag between the Qβ enzyme and the degradation tag. The TEV-protease expression is triggered upon activation of the detection module, in parallel with T7 RNA polymerase induction. This results in simultaneous expression and stabilization of the Qβ enzyme, which increases the signal to noise ratio.

Based on signal-to-noise experiments, each layer can be either included or taken away. Alternative layers, such as antisense transcription or the degradation tag on β-galactosidase can be added to further reduce noise if needed.

Type IIS assembly was used for the cloning of each part of the amplification system. This method was chosen because it easily allows us to construct combinations of promoters and RBSs with different strengths. Generating many combinations is essential in order to fine-tune the final system.

iGEM Type IIS guide

Prototype

---

In order to explore the implementation of NANOFLEX, where the kit can be distributed to point of care we had to think about how it would be distributed. Thus we developed a hardware prototype. The main points we kept in mind while designing our prototype was users safety, easy use of the kit not requiring expertise, manufacturing process and cost, the possibility to recycle the kit and of course correct functionality. The final prototype is a three-part chamber where one chamber is for lyophilized bacteria, one for media that activates bacteria and a sample chamber that also locks the system making the contents inaccessible for the user. You can read more on the prototype development here and on future improvements here.

References

---