SulFind 2021


Several approaches were considered when developing a biosensor with adequate sensing capability for detecting hydrogen sulfide H2S. After extensive research the two most promising approaches were brought into the lab to be tested. One of the ideas was a whole cell sensor, the other was a molecular sensor with heme proteins implemented in a microfluidic chip. Both projects went through the engineering cycle, illustrated in Figure 1, which consists of the following steps: design, build, test and learn. On this page we describe the engineering cycle for both projects.


Figure 1: The four steps of the engineering cycle.

Engineering of heme-based sensor

We were interested in designing a molecular sensor because it offers immediate and selective detection. These are some of the most important sensor qualities because H2S-levels in the water tanks can rise quickly. In our initial research phase we investigated several approaches to using heme-proteins as a basis for a molecular H2S sensor. This was largely motivated by the known interactions between many heme-proteins and H2S, such as with cytochrome c oxidase, hemoglobin and myoglobin [1]. This interaction is non-covalent and reversible, which offers a possibility for continuous detection. Early on, we considered the use of cantilevers to detect interaction in a heme-protein with high H2S-affinity, as well as variations on heme-based oxygen sensors. As we continued our investigation, we discovered a seemingly more robust way of detecting H2S-heme interactions, that is by using myoglobin conjugated with a fluorophore [2].

After deciding to use myoglobin as the basis of our sensor, we needed to find a way to create a detectable signal for when H2S binds to the protein. For myoglobin, this is possible using the principle of Förster Resonance Energy Transfer, also known as FRET. This is a method commonly used to study protein interactions based on an energy transfer that occurs between an energy donor and acceptor, usually fluorophores, in close proximity [3]. This energy transfer produces a fluorescent signal that can be detected. In our sensor system, we can use FRET to detect H2S-binding because the absorption spectrum of myoglobin shifts when interacting with H2S [4]. The particular method of energy transfer we implemented is called absorption into fluorescence, and is achieved by labeling the N-terminus of myoglobin with a fluorophore that has an emission peak corresponding to an area of interest in the absorption spectra of myoglobin [5]. This area of interest can be seen in Figure 2 around 550nm. When exciting the fluorophore, energy from the fluorescence will be absorbed into the myoglobin, and therefore changes in the absorption can be detected by fluorescence.


Figure 2: Absorption spectrum of ferric myoglobin, before and after addition and reaction with Na2S. The line in black shows the absorption spectrum before the reaction, and the blue after the reaction. The area of interest for our heme-based H2S sensor is the hump around 550 nm, where there is an increase in absorption for the reaction between ferric myoglobin and H2S. Figure from (Jensen & Fago, 2018) [6]

Our main objective in the testing phase for our myoglobin-fluorophore approach was to investigate the optimal concentrations of fluorophore-bound myoglobin using different H2S concentrations. The first stages of testing were done outside of the microfludic chip to assess the different sensor parameters. As presented in our Results Pagehigher fluorophore-myoglobin concentrations seemed to increase the variance in the measurements. Additionally, we found lower variance when slightly increasing the distance between the excitation and emission measurement bands used for measuring the fluorescence signal in the myoglobin-fluorophore. Correcting for these factors, we managed to get a signal at very low H2S concentrations. However, there were still some issues with variance, lack of a general trend and linearity, potentially caused by different factors which are further described in the discussion segment of our Results Page. The main factors here were the use of non-isolated measurement wells, no degassing and potential oversaturation of the reaction mixture. From our testing, we learned that the further development of this approach would therefore require a fluorescence measurement instrument with closed measurement wells or chambers, to prevent H2S diffusion, and a degasser to deoxygenate the sample before and after to prevent H2S oxidation. This would however require more resources than currently available to us, but it seems probable that this would increase the effectiveness of the sensor significantly. On the other hand, these limitations could be reduced when moving the sensor system to a microfluidic chip.

Design of Microfluidic Chip

The design of our microfluidic chip was a quite simple one, and it was for a flow focusing device. It consisted of three fluid inlets; one for the myoglobin solution, one for the H2S solution and one final inlet for the carrier oil. In the first phase of the microfluidic chip was the droplet generator (Figure 3). There, the inlets for the myoglobin solution and the H2S solution merged together. Right afterwards this new mix flowed past the carrier oil inlet, making individual and separated droplets from the flow. The second part of the chip was the mixing phase, where a mixing channel in a zigzag shape ensured proper mixing of myoglobin and H2S to make sure fluorescence would occur (Figure 4). Last phase was the collection chamber, a linearly expanding channel that caught all of the droplets within it. These droplets would then be collected and the fluorescence measured off-chip (Figure 5).

Here are some of our parameters for the flow setup:

  • Flow rate of heme solution = 30 µl/hr
  • Flow rate of H2S solution = 30 µl/hr
  • Flow rate oil phase = 250 µl/hr
  • Droplet generation frequency = 400 Hz
  • Droplet volume = 41 pl


Figure 3: Droplet generator. The two first inlets are for the myoglobin and H2S solution respectively. These merge and meet the carrier oil, making droplets.


Figure 4: Mixing channel to ensure proper mixing for myoglobin and H2S.


Figure 5: Collection chamber for the droplets.

The SQR approach

We began our research by looking into sulfide-binding molecules, and came across the transcriptional repressor protein SqrR, which showed promise as a component in a whole-cell sulfide sensor. This repressor protein and the way it functions in transcriptional regulation became the foundation for the “SQR” part of our project. Even though we ultimately ended up ending the SQR approach early to focus on the Heme approach and microfluidic chip design, we went through one full iteration of the engineering cycle, designing genetic constructs that would act as sulfide sensors, building the designs by cloning the constructs into bacteria, testing our sensors with varying levels of H2S, learning from our mistakes, and designing new constructs.

Design → Build → Test → Learn → Design

SqrR is a repressor protein that is involved in sulfide-dependent gene expression in the bacterium Rhodobacter capsulatus [7]. When no sulfide is present, SqrR binds to the Psqr promoter, inhibiting the expression of the gene encoding Sulfide:quinone oxidoreductase (SQR). SQR is an enzyme that oxidises sulfide to zero-valent sulfur (S0), which can then conjugate with multiple different nucleophiles or low molecular weight thiols to form reactive sulfur species (RSS) [7]. Upon reaction with a RSS, a di-, tri-, or tetrasulfide bond forms between two cysteine residues of SqrR, resulting in a conformational change that reduces its DNA binding affinity. RSS thereby functions as inducers of the genes under control of the Psqr promoter that is repressed by SqrR [7].

Design 1.0

We made two different genetic designs that utilize the SqrR protein. The first insert was designed to be modular and allow for the steps of sulfide oxidation, and RSS sensing to be separated on a microfluidic chip. First H2S would be reacted with Glutathione disulfide (GSSG), so that glutathione persulfide (GSSH), a RSS, is formed. The GSSH would then diffuse into engineered cells with the genetic insert, where SqrR is inhibited, resulting in expression of the reporter gene. This process is illustrated in Figure 6. The insert consisted of a fluorescent reporter gene placed under control of the Psqr promoter, sqrR, the gene that encodes SqrR, placed under control of a constitutive promoter, as well as biobrick ribosomal binding sites and terminators. The insert design is illustrated in Figure 7.


Figure 6: Schematic representation of hydrogen sulfide detection by the GSSG-dependent design.


Figure 7: Components of composite parts for hydrogen sulfide detection by the GSSG-dependent design. Figure designed using Pigeon [8].]

Three different inserts were designed like this. In two of the parts, CreiLov was used as a reporter gene. CreiLov is a flavin-binding fluorescent protein that does not require oxygen for fluorescence, and can therefore be used in anaerobic environments [9]. The protein was used so that integration into a microfluidic chip might be feasible. We also varied the constitutive promoter used for SqrR expression, so that the relative strength of the promoters were different. This was done because we suspected that the expression level of SqrR would affect the sensitivity of the sensor.

We also designed inserts that would function as a whole-cell sensor without the need for any additional components. In this design, the bacterium would also express the SQR enzyme, and therefore oxidise H2S inside the cell. The process is illustrated schematically in Figure 8, and the part design is shown in Figure 9. These designs were in part inspired by the Nanjing-China iGEM team in 2017. Link to their design is here.


Figure 8: Schematic representation of hydrogen sulfide detection by the whole-cell design.


Figure 9: Components of composite parts for hydrogen sulfide detection by the whole-cell design. Figure designed using Pigeon [8].]

As with the previous design, we also made variants where the reporter protein was either CreiLov or GFP, and the strength of the constitutive promoters varied.

Read more about the design on our Parts Page.


We ordered all the designs fully synthesized as linear fragments from Integrated DNA Technologies (IDT). IDT was unable to synthesize the construct containing both SQR and GFP encoding genes, likely because of the length of the construct.

We inserted the constructs into the plasmid vector pENZ004, a plasmid with a mid-range copy-number and Amp resistance genes, and used Gibson assembly to clone the vector into E. coli DH5-alpha cells. We then confirmed the success of the cloning by sequencing, to confirm the presence of our constructs in the bacterial plasmids. The entire procedure, as well as troubleshooting steps, is detailed on our Lab Notebook Page.

We were unable to successfully clone one of the constructs for the whole-cell design. After doing some troubleshooting, and attempting several Gibson assemblies and transformations we decided to move on with testing the four constructs that were successfully cloned into our bacteria, so that we would have sufficient time for testing.

Read more about the cloning procedure on our Results Page.


The bacteria transformed with the GSSH-dependant construct were tested by measuring the absorbance and fluorescence of transformed bacteria on a Tecan plate reader after incubating the bacteria with varying concentrations of freshly prepared GSSH (a RSS). The bacteria containing the SQR-construct was tested by measuring the absorbance and fluorescence of transformed bacteria on a Tecan plate reader after incubating the bacteria with varying concentrations of freshly dissolved Sodium sulfide (Na2S). Fluorescence was measured at 450 nm for excitation and 495 nm emission for the CreiLov parts, and 475 nm for excitation and 545 nm emission for the GFP parts.

We tested the design with the following concentrations of H2S ranging from 0 to 1000 ug/L, based on results we received from Carlo Lazado about the H2S levels dangerous to fish. Read more about this on our Human Practises Page.

We found no significant difference in observed fluorescence between the bacteria that had been incubated with GSSH or H2S and the control. The fluorescence measurements are presented on our Results Page.


We troubleshooted our experiment design in the lab by varying the medium the bacteria were grown in, increasing the H2S concentrations, and treating the bacteria with GSSH/H2S earlier in their growth phase. Time did not allow further troubleshooting in the lab, as we needed to focus on the Heme and microfluidic chip part of our project, which were showing more promise.

Our advisor Alex raised the topic of whether GSSH would be able to cross the cell membrane. The molecule is too big to diffuse across the cell membrane. This is a major flaw in our design, and made us reconsider the modular approach where sulfide is oxidised outside the cell.

Upon further analysis of the genetic sequences we had ordered from IDT we became aware of two mistakes. When inserting the one of the biobrick RBS (Part:BBa_B0030), was accidentally copied from the registry as dsDNA instead of ssDNA, resulting in the sequence being inserted twice and inverted. This could likely have an effect on the ribosome binding to the transcript. Perhaps more disastrous, we had removed upstream purification and cleavage sites from the CreiLov gene, and omitted inserting an ATG codon back into the sequence. These mistakes occurred because we were pressed for time when ordering our constructs, but are unfortunate and could have been prevented if more time was allocated for double-checking our constructs.

Design 2.0

Even though we did not have time to order and test new genetic constructs, we made a second design of the construct where SQR is expressed within the cell, as this approach alleviates the need for GSSH crossing a cell membrane. We fixed the mistakes with the RBS and the missing start codon for CreiLov in the construct where SQR is expressed in addition to the sensor sequences. In addition, we also changed a part of DNA at the beginning of the sequence that could form hairpin structures, and potentially caused problems with the PCR primers.

If time had permitted, we would have ordered this improved design, and tested in the lab. We would order the part as smaller parts that we could assemble, so that it would be less likely that problems would occur during synthesis. We would also have tested CreiLov alone with a constitutive promoter, to add another layer of modularity when troubleshooting.


1. Pietri, R., E. Román-Morales, and J. López-Garriga, Hydrogen sulfide and hemeproteins: knowledge and mysteries. Antioxid Redox Signal, 2011. 15(2): p. 393-404.

2. Strianese, M., et al., Myoglobin as a new fluorescence probe to sense H2S. Protein Pept Lett, 2011. 18(3): p. 282-6.

3. Sekar, R.B. and A. Periasamy, Fluorescence resonance energy transfer (FRET) microscopy imaging of live cell protein localizations. J Cell Biol, 2003. 160(5): p. 629-33.

4. Nicholls, P., The formation and properties of sulphmyoglobin and sulphcatalase. Biochem J, 1961. 81(2): p. 374-83.

5. Strianese, M., et al., Absorption into fluorescence. A method to sense biologically relevant gas molecules. Nanoscale, 2011. 3(1): p. 298-302.

6. Jensen, B. and A. Fago, Reactions of ferric hemoglobin and myoglobin with hydrogen sulfide under physiological conditions. Journal of Inorganic Biochemistry, 2018. 182: p. 133-140.

7. Shimizu, T., et al., Sulfide-responsive transcriptional repressor SqrR functions as a master regulator of sulfide-dependent photosynthesis. Proc Natl Acad Sci U S A, 2017. 114(9): p. 2355-2360.

8. Bhatia, S. and D. Densmore, Pigeon: A Design Visualizer for Synthetic Biology. ACS Synthetic Biology, 2013. 2(6): p. 348-350.

9. Zou, W., K. Le, and M.L. Zastrow, Live-Cell Copper-Induced Fluorescence Quenching of the Flavin-Binding Fluorescent Protein CreiLOV. ChemBioChem, 2020. 21(9): p. 1356-1363.