The creation of a biosensor for detecting cereulide levels in food samples is a big step forward in food safety and quality control. Cereulide is a poisonous chemical produced by certain strains of Bacillus cereus bacteria that causes health problems when consumed. This biosensor is intended to provide a quick and accurate means of detecting cereulide contamination in food products. To develop this biosensor, we went through three engineering cycles.
      To create a biosensor that could detect cereulide levels in food samples, we searched for detectable physical changes that could be used to trigger our circuit.
      It has been discovered that cereulide can act as a membrane channel-forming potassium ionophore, damaging mitochondria and blocking oxidative phosphorylation. With this discovery, past researchers have conducted experiments using boar semen to carry out a spermatozoon toxicity test[1]. Unfortunately, this method is ethically questionable and not viable commercially.
      In light of this, we created a biosensor to detect cereulide levels in food samples by sensing the potassium ion efflux that is caused by cereulide.
      In nature, cereulide binds and inserts itself into the inner mitochondrial membrane, acting as an ionophore to allow potassium ion diffusion. This disrupts the potassium ion concentration gradient, resulting in mitochondrial transmembrane potential disturbance, which impairs ATP synthesis and cell survival.
      The ionophore property of cereulide is due to its inner cyclic dodecadepsipeptide that contains amino acids and hydroxy acid residues, which enhance the complex formation with potassium ions, thus allowing the binding with and diffusion of potassium ions[2]. Our circuit utilizes this property to detect cereulide.
      KinC is a transmembrane sensor histidine kinase. It has a PAS-PAC domain that is responsible for sensing potassium ion leakage from the cell. The potassium ion leakage activates KinC to phosphorylate Spo0A, a protein involved in the regulation of biofilm formation[3]. Our circuit also utilizes this pathway to sense cereulide.
      Our sensing module makes use of the potassium-ion-selective ionophore property of cereulide and the endogenous potassium sensing pathway in B. subtilis to convert the cereulide signal to an intracellular signal of phosphorylated Spo0A.
      At a high cereulide level, cereulide binds to the cell membrane of B. subtilis, resulting in potassium ion leakage. This, in turn, activates KinC to phosphorylate Spo0A.
      Our processing module utilizes the promoter PsinI, which is under the control of Spo0A. When Spo0A is phosphorylated, it will bind to the Spo0A box in PsinI, resulting in PsinI transcription[4] and, thus, the downstream expression of RFP.
      At a high cereulide level, our sensing module responds to the signal, leading to an increase in the intracellular phosphorylated Spo0A level. This, in turn activates Psinl and thus the expression of RFP, which acts as an output signal to inform users that the sample contains cereulide.
      Our cycle 0 circuit has a PsinI promoter which is active when cereulide is present. Downstreams of the PsinI promoter are an RFP gene which will give our red fluorescence signal when cereulide is present.
      The construct is tested by exposing the transformed B. subtilis to different concentrations of cereulide and its structurally similar ionophore, valinomycin[5].
      In our first trial, we tested the circuit with 3 μM, 6 μM, and 9 μM of valinomycin. All three concentrations triggered RFP production at around 4 hours and showed a 60-fold increase in RFP fluorescence after 6 hours when compared to no valinomycin added.
      After our first trial, we wanted to find out the lower detection limit of our circuit, so we tested the circuit with 0.03 μM, 0.3 μM, 3 μM, and 4.5 μM of valinomycin in our second trial.
      In our second trial, concentrations of 0.3 μM above triggered RFP production at around 4 hours and showed a 20-fold increase in RFP fluorescence after 6 hours when compared to no valinomycin added. At the same time, we found out that 0.03 μM of valinomycin failed to trigger the cycle 0 circuit.
      Due to a limited stock of cereulide, we previously used valinomycin to test our cycle 0 circuit and assumed that valinomycin and cereulide would cause the same effect on our circuit as they have very similar structures and effects on gram-positive cell[5]. To validate our assumption, in the third trial, 0.3 μM of cereulide and valinomycin were used, respectively, to test our circuit. The intervals of sample collection are also shortened to 30 minutes to get more comprehensive data.
      The results of the third trial showed that the same concentration of cereulide and valinomycin cause very similar effects on our cycle 0 circuit, and therefore our assumption is validated.
      From cycle, 0 we found out that our circuit was able to produce a red signal when there is 0.3 μM or above valinomycin or cereulide, however, the emission of a red signal with the presence of cereulide would not be enough. The absence of RFP shall not be taken as a safe indicator; the absence of RFP may be caused by reasons other than the absence of cereulide, like cells dying during transit or due to improper storage. Therefore, to make our biosensor more user-friendly, GFP should be expressed when there is no cereulide as a negative control to prevent false negative results.
      To achieve this purpose, we employ a degradation system that makes use of inducible protein degradation in B. subtilis. It works by adding a modified ssrA degradation tag to the terminus of the protein of interest. SspB, an adaptor protein that delivers the modified ssrA-tagged protein to protease ClpXP, is essential for protein degradation. In the absence of SspB, the tagged protein is not degraded[6].
      Cycle 1 genetic circuit is composed of a cereulide-activated and cereulide-repressed component.
      The cereulide-activated component is our cycle 0 circuit. It has a PsinI promoter which is active when cereulide is present. Downstreams of the PsinI promoter are an RFP gene, SspB gene, and LacI gene. The SspB gene expresses SspB protein which will degrade ssrA-tagged GFP. The LacI acts as the repressor for the cereulide-repressed component so that the component is repressed when there is cereulide, and vice versa.
      The cereulide-repressed component has a hyper-spank promoter, which is repressed by LacI. When there is cereulide, the LacI expression would result in the repression of this promoter and thus the downstream gene expression, and vice versa. Downstream of the hyper-spank promoter is a ssrA-tagged GFP gene. The ssrA-tagged GFP is expressed when cereulide is absent so that the user is informed that the food samples do not contain cereulide. Additionally, a His-Tag is added to the GFP for validation of our circuit through western blotting.
In the absence of cereulide,
Eventually, there will be a high level of GFP. |
When cereulide present,
Eventually, there will be a high level of RFP. |
      The rise in RFP after valinomycin is added is not quite prominent if we look at the RFP fluorescence/OD data. Figure 8 shows that the RFP fluorescence/OD when 0 μM of valinomycin is added (i.e., basal expression) is at the same level as when 0.3-6 μM of valinomycin is added at 0 hours. Since the basal expression is relatively high, we will have to address this in the subsequent engineering cycle.
      For the GFP, it is found that the production of GFP is slower than anticipated, and the degradation of GFP also takes a considerable amount of time. Therefore, further tuning of the circuit is needed.
      The results from cycle 1 show that the GFP fails to fluoresce. We suspect that it may be because we added a degradation tag and a His tag at both termini, distorting the structure of GFP and rendering it unable to fluoresce. However, without the experimental data, we can use the modeling result from the dry lab (Check Circuit Modeling Cycle 1).
      In addition, the results from the previous experiments also show that the PsinI promoter may be leaky. As a result, there may still be red signals even when the sample does not contain cereulide. Additionally with the interpretation from dry lab modeling, the dry lab proposed a possible dual conditional degradation system for our circuit, which is a degradation system to regulate the level of RFP as well as that of GFP (Check Circuit Modeling Cycle 2).
      From the graph above, we can see that it is possible to reduce the RFP signal from the basal expression and is possible to set a threshold where the GFP would always be higher than RFP or vice versa.
      Our degradation system makes use of an inducible protein degradation in B. subtilis. It works by adding a modified ssrA degradation tag to the terminus of the protein of interest. SspB, an adaptor protein that delivers the modified ssrA-tagged protein to the protease ClpXP, is essential for protein degradation. In the absence of SspB, the tagged protein is not degraded [6]
      For our design, we use dual degradation to make sure the appearance of GFP and RFP will follow the order under two cereulide concentrations orderly.
      When there is no cereulide, the phosphorylation level of spo0A is low, resulting in a low activation of the sinI promoter. This leads to very low expression of SspB (E. coli), lacI, and RFP. These three proteins are attached with a ssrA (C. crescentus) tag.
      Since lacI serves as a repressor of the hyper-spank promoter, with a low lacI expression level, the hyper-spank promoter is not repressed, enabling its activation. Consequently, the downstream SspB (C. crescentus) and GFP will exhibit high expression levels. The two proteins are tagged by ssrA (E. coli) tag.
      Importantly, degradation occurs only when the ssrA tag and SspB are of the same bacterial protein type, that is, proteins with ssrA (E. coli) tag will be degraded by SspB (E. coli) and proteins with ssrA (C. crescentus) tag will be degraded by SspB (C. crescentus). Therefore, the high level of SspB (C. crescentus) targets only the proteins tagged with ssrA (C. crescentus) and delivers it to ClpXP for degradation rather than the GFP tagged with ssrA (E. coli) tag.
Eventually, there will be no RFP but a high level of GFP.(Check circuit modeling)
      On the other hand, in the presence of cereulide, the phosphorylation level of spo0A is elevated, leading to a higher activation of the sinI promoter. This leads to very high expression of SspB (E. coli), lacI, and RFP. These three proteins are attached with a ssrA (C. crescentus) tag.
      As a repressor itself, the high expression of lacI strongly inhibits the hyper-spank promoter. Consequently, the downstream SspB (C. crescentus) and GFP will express at very low levels. In the same manner, the ssrA (E. coli) is tagged to GFP.
      As a result, the remaining GFP tagged with ssrA (E. coli) is targeted by the high level of SspB (E. coli) and sent to ClpXP for degradation. In the end, there will be no GFP but a high level of RFP.
      We have ordered the gBlocks from IDT and have tried to clone them into E. coli. Unfortunately, due to time limitations, we did not have time to remove the His tag through Gibson assembly and test the circuit in B. subtillis. We plan to test the circuit in the future after we have successfully removed the His tag from the GFP and cloned the circuit into B. subtilis.