Aalto-Helsinki 2020




  1. We were able to confirm BL21 cells survive in wastewater.
  2. We assessed functionality of the ermC gene.
  3. We succeeded in building our genetic circuit and showing that it gives an output.
  4. We improved the sensitivity of our genetic circuit by changing to an inducible promoter, testing a strain of E. coli with modified outer membrane and using an altered transcription factor.
  5. We tested several immobilization methods and found that egg white, matrigel and alginate might be suitable for our biosensor.

Dry-Lab: Modelling

  1. We estimated the range of our biosensor.
  2. We were able to build a mathematical model of the biosensor pathway.
  3. We designed improved ligand binding sites.
  4. We assessed the modified structure of FhuA with homology-modelling.


Plate-Reader Experiment

We have confirmed that the cells were able to survive in all experimental set-ups except for the negative control (milliQ). Surprisingly, we observed the best growth for samples that contained a mix of LB and wastewater (1:1 WW:LB and 3:1 WW:LB). Cells grown in LB and 1:3 WW:LB were also able to proliferate. As expected, wastewater was not an optimal medium: cells were able to survive, but we do not observe proliferation (Fig. 1).

Figure 1. Plate-reader experiments results.

Conclusion: E. coli cells seem to be suitable as an organism of-choice for a biosensor construction. We suspect the efficient proliferation in samples containing half or more wastewater may be a result of stimulation by some compounds present in wastewater.

Flow Cytometry Experiment for Cell Viability

To confirm results obtained from the plate-reader experiment, we performed a similar test using flow cytometry. Most viable cells were found in 1:1 WW:LB samples. This was followed by pure LB and WW samples. Pure milliQ had the least amount of viable cells (Fig. 2).

Figure 2. Flow cytometry experiments results.

Conclusion: The flow cytometry results confirmed that E. coli cells were capable of growing in media containing wastewater at least for the period of three hours.


As the concentration of macrolides in the influent and effluent wastewater is low, we used experimental data from literature to determine the sensitivity of our MphR-based biosensor. Experimental data from wildtype and best performing mutated E. coli was used to examine current range (Fig. 3).

Figure 3. MatLab model estimating the range of our biosensor.

Conclusion: Macrolide concentration in wastewater was found to be too low for the current MphR-based biosensor found in literature. To address this problem, we started looking into different methods of concentrating the sample. More about our concentrating research can be found in our design page.


Due to the fact that the goal of the biosensor was not only to detect the macrolides from wastewater but also quantify them, we used our mathematical model to predict how different concentrations affect the optical output. Scan of 10 different concentrations of macrolide in range: 0.1 nM-10μM was plotted to assess the behaviour of our biosensor in various concentrations (Fig. 4).

Figure 4. Output of the SINISENS biosensor with different macrolide concentration in wastewater.

Conclusion: Very little variation was found between outputs of different concentrations, especially in the designed working time-frame of the biosensor.


Level 1 MphR Assembly - ermC Functionality

Level 1 MphR (repressor cassette) and level 1 EGFP (output cassette) assemblies were constructed using MoClo protocol. Successful colonies were identified by colony PCR (Fig. 5). The assemblies were confirmed with sequencing.

Figure 5. Gel run of samples from colony PCR. Correct colonies were marked with ‘x’. First gel contains the output cassette with egfp. Second gel contains the repressor cassette.

When growing E. coli cells transformed with the repressor cassette, the lowest concentration of all of the antibiotics (10 ⲙg/ml) seem to not have a major effect on the cell growth. When it comes to cells grown in 50 and 100 ⲙg/ml, both transformed and non-transformed cells grown in spectinomycin showed no resistance towards this antibiotic. Media containing macrolide antibiotics seemed to only inhibit the growth of non-transformed cells (Fig. 7).

Figure 6. Survival of cells transformed with a repressor cassette expressing ermC. OD - optical density. SPE - spectinomycin. CLA - clarithromycin. ERY - erythromycin.

Conclusion: ermC gene is functional and provides resistance for both erythromycin and clarithromycin. Since both ermC and MphR are expressed under the same promoter and have RBS of the same strength, we suspect that MphR is also produced. However, final confirmation of MphR functionality can be proved by testing the full optical device (level 2 assembly).

Level 2 Optical Device Assembly

Level 2 assembly (optical device) was constructed using MoClo protocol. Successful colonies were identified by colony PCR (Fig. 7). The assembly was confirmed with sequencing.

Figure 7. Gel run of samples from colony PCR containing the assembled optical device. Correct colonies were marked with ‘x’.

Cells transformed with the optical device were grown in various erythromycin concentrations. The amount of erythromycin present in the media seemed to correlate with the amount of fluorescence produced. (Fig 8). However, the differences between fluorescence produced in different concentrations were not distinguishable enough for the biosensor to be useful for detection of very small macrolide amounts.

Figure 8. OD normalized fluorescence values produced by cells grown in different erythromycin (ERY) concentrations (0, 0.1, 0.5, 1, 10, 50, 100 µg/ml). The graph includes data from three repeats (each comprising 3 technical repeats).

Conclusion: It is apparent that the current design requires further fine-tuning. Initially it was considered that egfp is not expressed due to codon optimization. We suspected our changes might have led to formation of a secondary structure of mRNA. However, these results may also indicate that in spite of what our model previously suggested, the strength of the constitutive promoter was too high and there was always a MphR protein bound to DNA, regardless of the amount of antibiotic. For that reason we attempted to construct the full device again, replacing the constitutive promoter with an inducible one. This would allow us to select an optimal promoter strength.


The initial biosensor was designed with an constitutive promoter which was later suspected to be producing too much repressor protein, MphR. Scan of five different promoter strengths in range: 0-0.02 polymerase/second was plotted to visualise the behaviour of our biosensor with various promoter strengths (Fig. 10).

Figure 9. Scan of EGFP production in a constant erythromycin concentration. Different lines correspond to different promoter strengths. The lowest line represents the strength of the constitutive promoter used in our initial design, the highest line represents the output with no repressor present.

Conclusion: Promoter strength was found to have strong correlation with the output of the biosensor. As expected, less repressor present resulted in higher concentration of EGFP. As a result, we decided to swap our initial promoter to an inducible promoter, pBAD.


Before constructing the genetic circuit with the pBAD promoter, we decided to test different GFP options. Level 1 sfGFP assembly (sfGFP output cassette), comprising sfgfp reporter gene instead of egfp, was constructed using MoClo protocol. Successful colonies were identified by colony PCR (Fig. 10). The assembly was confirmed with sequencing.

Figure 10. Gel run of samples from colony PCR. Correct colonies were marked with ‘x’. The contains the output cassette with sfgfp.

To see if the output cassetete is capable of producing enough fluorescence, we have grown the cells transformed with output cassettes containing either sfGFP (Addgene) or EGFP were grown without a repressor. It was clear sfGFP produces stronger output (Fig. 11).

Figure 11. Comparison of fluorescent output produced by cells expressing two output cassettes (assemblies containing a reporter protein driven by an inducible pMphR promoter, see our experiments page), one containing sfGFP, second EGFP, grown in absence of the repressor. Fluorescence values were OD normalized. The graph includes data from one biological repeat (comprising 3 technical repeats).

Conclusion: The sfgfp seems to be a more suitable reporter protein for the biosensor as it produces stronger signals.

Inducible Optical Device Assembly

Level 2 inducible assembly (inducible optical device), comprising a pBAD promoter instead of the constitutive one, as well as sfgfp instead of egfp, was constructed using MoClo protocol. Successful colonies were identified by colony PCR (Fig. 12).

Figure 12. Gel run of samples from colony PCR. A. Colonies containing the inducible optical device. The assembly was confirmed with sequencing.

Optimizing Promoter Strength

The goals of initial experiments after construction of the inducible device were: (i) finding the necessary induction time before resuspending cells to experimental set-ups and (ii) selecting the optimal arabinose concentration in the experimental set-ups that would produce the best sensitivity. Since the macrolide resistance gene, ermC is driven by the same promoter as MphR (pBAD), we wanted to assess whether cells will survive being placed in media containing erythromycin or if they require prior induction. We have performed an experiment with cells that were not induced and induced with 1% arabinose 30 or 10 minutes prior to being placed in media containing 100 µg/m erythromycin and various concentrations of arabinose (0.1%, 0.5% and 1%). We observed the cell growth in all of the samples. We concluded that the induction prior to placing cells in erythromycin is not necessary.

In order to find the optimal promoter strength and achieve the most visible separation between fluorescence produced in different concentrations, the cells transformed with the inducible optical device were grown in three different arabinose concentrations. The observed results indicate that 0.1% is the most optimal induction (Fig. 13). It is worth noting that the separation between fluorescence produced in different erythromycin concentrations was also more distinguishable than in the optical device containing the constitutive promoter.

Figure 13. Selecting the optimal arabinose concentration. OD normalized fluorescence produced by cells transformed with an inducible optical device grown in different erythromycin (ERY) concentrations (0, 0.1, 1 and 100 µg/ml). A . 0.1% arabinose B. 0.5% arabinose C. 1% arabinose.

Conclusion: Construction of a new optical device containing a weaker promoter may improve sensitivity of the biosensor.


One of the methods we investigated to improve sensitivity was using transmembrane protein FhuA. We found proposed modifications to this protein in literature that would allow more molecules through it and wanted to assess their feasibility. The results from the SWISS-MODEL scoring were very promising and all the values of the modified FhuA protein were relatively close to the ones of the unmodified. More details can be found on our Modelling page.

In addition to that, we performed tests with GKCW104 strain obtained from Krishnamorthy and the colleagues [1], which has removed tolC gene (an element of many multidrug resistance pumps) and modified FhuA protein, leading to cell’s hyperporination. We transformed GKCW104 with the constitutive optical device (as the cells require arabinose induction to introduce above mentioned changes to the membrane and thus, using a pBAD promoter is not advisable). The results showed improved separation between fluorescence intensity produced in different erythromycin concentrations in comparison to TOP10 cells (Fig. 14). However, due to time constraint, we only replicated the experiment twice.

Figure 14. OD normalized fluorescence values produced by GKCW104 cells transformed with an constitutive optical device grown in different erythromycin (ERY) concentrations (0, 0.1, 1 and 100 µg/ml) and 0.1% of arabinose. The graph includes data from two repeats (each comprising 3 technical repeats).

Conclusion: The stability and protein structure were estimated with SWISS-MODEL and the obtained results suggest that these modifications might be feasible to work in practice. Additionally, using GKCW104 strain for the biosensor could potentially improve its sensitivity.


An improvement of the sensitivity was needed in order to detect macrolides, as the low concentration was found to be a problem. To address it, we decided to improve the binding affinity of MphR with macrolide molecules. Improved binding affinity would result in quicker and stronger binding of these molecules, thus initiating EGFP transcription earlier. Rosetta’s ligand binding site designing protocol was used to generate an improved MphR molecule. We decided to test the most promising mutants experimentally (Fig. 15).

Figure 15. MphR sequences with the mutations that were ordered to be tested experimentally.


We replaced natural MphR with new MphR sequences modelled in Rosetta. These modifications were likely to improve binding affinity of MphR to macrolides. We experimentally tested five of the sequences that showed best scores for erythromycin, clarithromycin or both. Level 1 and 2 assemblies containing MphR mutants were constructed using MoClo protocol. Successful colonies were identified by colony PCR (Fig. 16). The assemblies were confirmed with sequencing.

Figure 16. A. Level 1 assembly comprising pBAD and mphr mutants. B. Level 2 assembly comprising pBAD, MphR mutants and sfgfp. Bands were visualized on 0.7% agarose gel using SybrSafe. The samples with bands at correct lengths were marked with the x symbol.

One of them (third best score for erythromycin, see our modelling page) seemed to improve biosensor’s sensitivity and produce better separation between fluorescence measured in samples containing different macrolide concentrations (Fig. 17). We have only performed two repeats and these results should be replicated, but the initial data seems promising.

Figure 17. Fluorescence produced by cells transformed with the optical device comprising a MphR mutant in different erythromycin (ERY) concentrations [µg/ml]. Fluorescence values were OD normalized. The data includes an average of two biological repeats (each comprising three technical ones).


All the entrapment matrices; alginate beads, egg white and Cultrex, were able to immobilize the cells (Fig. 19-21). Viability dye was able to penetrate all of the matrices and the dyed cells are shown in the figures as magenta color. Viability of the cells was not significantly decreased during the incubation time based on the low amount of magenta colored cells. It was confirmed that the non-transformed cells showed magenta colored cells, but no fluorescence. When trying egg white and Cultrex entrapment by adding the cells on top of the matrix, immobilization could not be seen.

Figure 19. E. coli cells expressing sfGFP immobilized on alginate beads.

Figure 20. E. coli cells expressing sfGFP immobilized on egg white.

Figure 21. E. coli cells expressing sfGFP immobilized on Cultrex.

Figure 22. E. coli cells expressing sfGFP immobilized on Poly-L-Lysine.

Conclusion: When comparing these three different matrices, egg white has a non-uniform consistency, which is why it might not be the best option for our final product. Disadvantage of Cultrex is its more expensive price compared to alginate. We found a disadvantage of alginate beads to be that when measuring fluorescence, the light is reflected from the spheres, which makes it impossible to get fluorescence reads.

Polylysine immobilization experiments were unsuccessful as we saw movement of the cells in all of the samples (Fig. 22). In literature, research groups have documented successful immobilization. Thus, polylysine immobilization should be repeated.


We investigated using isoelectric focusing for concentrating the macrolide antibiotics from the waste water sample. The experiment was performed with erythromycin, thymol blue and cresol red. As expected, erythromycin was not visible on the gel. Thymol blue was chosen since it has the same pKa as erythromycin (8.9), but this did not give better results, most likely due to it being negatively charged both above and below its pKa. Finally, we repeated the test with cresol red, which has a pKa of 8.3. Cresol red did produce a small band close to the cathode, however the band did not have clear borders and was located very close to the end of the IPG gel, so it is questionable if it actually worked (Fig. 23).

Figure 23. Results of isoelectric focusing with cresol red. A small band without clear borders was visible next to the cathode.

Conclusion: Even though the initial tests were not successful, this technique may work with a more specialized device or gel optimized for small analytes.


  1. Performing viability tests with GKCW104 to confirm it can survive in wastewater and it is suitable microorganism for our biosensor.
  2. Transformation of GKCW104 with an optical device containing a MphR mutant in order to assess whether it may have a synergistic effect on the sensitivity.
  3. Testing sensitivity of the optical device containing MphR mutants to other macrolide antibiotics (clarithromycin and azithromycin) to assure initial results are replicable.
  4. Constructing a new optical device with a weaker promoter with similar strength that was achieved by pBAD induced by 0.1% arabinose.
  5. Improving the design of the genetic circuit to increase the strength of the output signal. It may include using a stronger RBS, changing the distance between RBS and initiation codon or adding an enhancer sequence, such as the mlcR gene from Dictyostelium discoideum, which has been shown to significantly increase protein expression levels. Additionally, it is easy to adjust and inexpensive [2].
  6. Constructing the optical device in a low copy number plasmid [3].
  7. Integrating the optical device into a bacterial chromosome to avoid plasmid loss and possibly improve the sensitivity.
  8. Further investigation of possibilities of concentrating the macrolide in the sample.
  9. Repeating immobilization tests with poly-lysine. Assessing suitability of other immobilization techniques, such as latex or silica layers [4,5].
  10. Constructing a biosensor with a more sensitive signal than an optical output. One option is using the Mtr pathway to build an electrochemical biosensor. More details about an electrochemical biosensor for macrolide detection can be found on our contributions page in the section Mtr pathway in Escherichia coli.
  11. Testing the biosensor’s performance in wastewater. The concentration of macrolides could be measured with liquid chromatography–mass spectrometry beforehand.
  12. Testing the biosensor at lower temperatures than 37°C, such as room temperature, to assure that it can function in wastewater treatment plant conditions.


1. Krishnamoorthy, G., Wolloscheck, D., Weeks, J., Croft, C., Rybenkov, V., & Zgurskaya, H. (2016). Breaking the permeability barrier of Escherichia coli by controlled hyperporination of the outer membrane. Antimicrobial Agents And Chemotherapy, 60(12), 7372-7381. doi: 10.1128/aac.01882-16
2. Kondo, T., & Yumura, S. (2019). Translation enhancement by a Dictyostelium gene sequence in Escherichia coli. Applied Microbiology And Biotechnology, 103(8), 3501-3510. doi: 10.1007/s00253-019-09746-7
3. Hicks, M., Bachmann, T., & Wang, B. (2019). Synthetic Biology Enables Programmable Cell‐Based Biosensors. Chemphyschem, 21(2), 132-144. doi: 10.1002/cphc.201900739
4. Asal, M., Özen, Ö, Şahinler, M., Baysal, H. T., & Polatoğlu, I. (2019). An overview of biomolecules, immobilization methods and support materials of biosensors. Sensor Review, 39(3), 377-386. doi:10.1108/SR-04-2018-0084
5. Bhardwaj, T. (2014). A Review on Immobilization Techniques of Biosensors. International journal of engineering research and technology, 3.

Special thanks to HSY for all their support

Kemistintie 1, Espoo, Finland