The idea of engineering and DBTL(Design-Build-Test-Learn cycle) was always kept in our minds since it’s one of the major ideas of synthetic biology. With comprehensive efforts in the field of wet lab, model, hardware as well as human practice, we endeavored to learn from what we obtained for improving our methodologies to demonstrate our designs.

In all, we did generate expected results to a certain degree. DBTL cycles of our four parts designed are described in detail in the following contents.

DBTL cycle of fatty acid consumption module

In our design, consuming excess sebum on the skin surface to prevent it from blocking pores is one of the two important ways for our engineered bacteria to treat acne.

In the early stage of acne, the skin secretes a large amount of sebum which blocks the pores. The blocked pores form an anaerobic environment, which provides a more suitable environment for Propionibacterium acnes and proliferates Propionibacterium acnes in large numbers.

Over-proliferated Propionibacterium acnes will decompose sebum and produce a large number of free fatty acids. These fatty acids penetrate into skin capillaries through pores, causing local stress response,swelling, redness and acne, resulting in skin tissue damages. In view of this pathogenic mechanism, we proposed to consume the excessive sebum on the skin surface in order to block the excessive proliferation of Propionibacterium acnes and the immune response triggered by fatty acids.

In order to make our engineered bacteria consume the excess sebum on the skin surface more efficiently, we put our efforts on improving the β-oxidation ability of engineered bacteria (i.e. improving the decomposition ability of fatty acids). As shown in the following figure, fatty acids need to be activated to produce Fatty acyl-CoA before β-oxidation. Then, Fatty acyl-CoA enters the β-oxidation pathway. After oxidation, hydration, oxidation and cleavage, the fatty acyl-CoA breaks to form acetyl-CoA and fatty acyl-CoA with two carbon atoms less. When we inquired about the literature and the work of iGEM teams in previous years, we learned that overexpression of two enzymes in the β-oxidation pathway, fatty acyl-CoA synthetase (FadD) and fatty acyl-CoA dehydrogenase (FadE), in engineered bacteria can significantly improve the oxidation ability of engineered bacteria.

Figure 1. Fatty Acid β-oxidation pathway

Therefore, we tried to repeat this result, that is, overexpressing these two enzymes in our engineered.bacteria to improve their β-oxidation ability, so as to achieve the purpose of rapid consumption of excess sebum on the skin surface by engineered bacteria.

In order to overexpress FadD and FadE protein in engineered bacteria, we designed the following two plasmids. We first designed primers, amplified FadD and FadE genes from the genome of E. coli K12 strain by PCR, and added corresponding enzyme digestion sites on both sides of the gene; At the same time, we added six his tags at the N-terminal of the protein for subsequent purification and detection. After that, we connected our target fragment with the vector by enzyme digestion and enzyme linkage.

Figure 2. The maps of pBAD33-FadE and pDSW208-FadD

After the construction of the vector, we transferred the recombinant plasmid into E. coli DH5α Strain and induced expression. After a simple protein expression test, we began the functional test. We mixed the two-hour protein induced bacterial solution with equal volume of LB medium dissolved with fatty acid sodium and continued to culture at 37 ℃ for 4h. Then we detected the residual amount of fatty acids in the culture medium However, in this experiment, we did not get the expected results. The fatty acid decomposition ability of engineered bacteria which overexpresses FadD and FadE was not significantly higher than that of negative control.

Figure 3. The decomposition ability of engineering bacteria

In order to explore the reasons for the failure of our experiment, we tested the protein expression more carefully. Our protein was purified by nickel column and then detected by SDS-PAGE . The detection results show that the molecular weight of our target protein is about 20kDa less than our expected molecular weight - which explains why our constructed system does not achieve our expected results. Because the engineered bacteria we use this time is Escherichia coli DH5α. The internal protease system of the strain is not knocked out, so the overexpressed protein is easy to be degraded. For this reason, we replaced our host strain with E. coli BL21 strain for following experiments.

DBTL cycle of the signal peptide for PctA

The successful expression and secretion of PctA by our engineered bacteria is of vital importance to our project. In the expected situation, our engineering should transport PctA across plasma membrane and then outer membrane, letting it perform its duty in the surrounding environment.

The original PctA itself is a secretory protein purified from the supernatant of the culture. So in the first version of our design, we keep the original Sec signal peptide in the full sequence of PctA gene [1]. We adapted this signal peptide because the Sec system is also found in Escherichia coli . According to our prediction, the bacteria will first produce a 11 kDa precursor protein, then transport it into the periplasmic space and at the same time cut off the signal peptide after the Ala-Met-Ala sequence. Finally, a 7 kDa mature PctA protein will diffuse across outer membrane into the environment.

Figure 4. The full sequence of PctA gene

We adapt the original Sec signal peptide, and made no adjustment except adding a His tag. We put the sequence in plasmid pET26b(+) with T7 promotor to test the expression, secretion and the function of PctA.

In order to test the expression and secretion of PctA, we induced the engineered bacteria in different temperature and IPTG concentration, then analyze protein composition with SDS-PAGE (later tricine-SDS-PAGE) and Western Blot. Most of PctA form inclusion body and end up in precipitation; limited amounts of PctA in solution can be tested by WB; no PctA can be found in culture supernatant both by SDS page and Western Blot, even after the supernatant is concentrated for 40 times. This could result from the overpowerful of T7 promotor, or the original Sec signal peptide doesn’t fit in Escherichia coli expression system. PctA with Sec peptide is likely to end up in plasma or periplasmic space, which is different from our prediction.

Figure 5. The SDS page of induced engineered bacteria. PctA is successfully induced, but most end up in precipitation. A: the solution of bacteria lysis; B: the precipitation of bacteria lysis.

On the basis of former experiment, we begin the second turn of DBTL cycle. We look through papers and find a new way of transporting PctA. A new signal peptide, OmpA was chosen, because it is widely used in E. coli and at least one experiment can prove that protein with OmpA was successfully transported across outer membrane and entered the culture supernatant [2].

Figure 6. The redesigned OmpA-PctA recombined protein. OmpA is added to replace Sec signal peptide.

Due to the lack of time, we could not practice our new design to see if it works out. However, we are planning to finish this experiment in the future.

DBTL cycle of fatty acids sensing module

In the initial design of the project, there was actually no such module. At first, we hoped that the bacteriocin against Propionibacterium acnes could be expressed directly in a constitutive manner to achieve the optimal bactericidal effect. However, after consulting literature and communicating with medical college professors, we found that Propionibacterium acnes also existed on the skin surface of healthy people and played a certain role. It can produce protease and lipase to release arginine from the skin and degrade triglycerides to survive in the anoxic sebaceous gland. The released fatty acid also increases the adhesion between microflora [3]. In order to maintain a balanced, symbiotic relationship between the skin and skin microbiota, we designed this module and selected the fatty acid-sensitive promoter, in the hope of providing a relatively accurate control over the yield of bacteriocin.

However, we found that the existing fatty acids-sensitive promoters have a bad performance, taking a long time to better start the expression with high leakage expression. Therefore, we first used the existing fatty acid sensitive promoters and made artificial improvements on the RBS. The plasmids we designed is shown in the figure below.

Figure 7. The maps of pDSW208-99-1 and pDSW208-99-2. (a) pDSW208-99-2(pFadD_Lac+B0030+GFP) contains RBS without modification, and can test promoter function. (b) pDSW208-99-2(pFadD_Lac+ BBa_K3763000+GFP) Contains RBS with modification, and can test promoter function.

After consulting a large number of RBS data, we found that most RBS with a higher binding efficiency had the repetitive sequence of "aggg", or "aaa" after the start codon. B0030 was the one with the highest expression efficiency among the four existing RBS, with the sequence of "aaa" but no sequence of "aggg". Therefore, we modified the sequence to change "aagagg" to "aggagg", so that it could simultaneously have the characteristics of two sequences with high binding efficiency. In later experiments, we wanted to compare the performance of the two sequences.

Figure 8. The change in sequence

After the successful transformation of the plasmid, we used different concentrations of fatty acids for induction for different times. The expression level of eGFP was determined by fluorescence detected with a microplate reader, reflecting the performance of the promoter. The results are as follows:

Figure 9. Promoter and eGFP
Figure 10. BBa_K3763002(pFadD_Lac-RBS BBa_B0030-GFP-rrnB T1 and T2 terminator, pDSW208-99-2) contains RBS without modification
Figure 11. BBa_K3763003(pFadD_Lac-RBS BBa_K3763000-GFP-rrnB T1 and T2 terminator, pDSW208-99-2) Contains RBS with modification

According to the results, we can find that the performance of our fatty acids-sensitive promoter is improved to a certain extent. Taking 0.125× fatty acid concentration as an example, the fluorescence intensity of BBa_K3763003 is over 11000 after 6 hours, while the promoter of BBa_K3763002 is less than 10000 at three fatty acid concentrations. Although we can see that the optimized promoter still has a high expression leakage and low expression in a short time, the experimental results have been able to prove the function of the fatty acids-sensitive promoter, and partially prove the effectiveness and rationality of our design. For future optimization, after the first DBTL cycle, we have conducted a new round of design and experimental work.

We wanted to regulate the exact expression of bacteriocins from multiple perspectives. After the optimization of RBS, we turned our target to the improvement of FadR, and designed a directed evolution method to optimize the function of the repressor FadR corresponding to this promoter.

FadR is one of the regulatory enzymes in fatty acid biosynthesis in E. coli., which acts as the repressor in the fatty acid metabolism system. In the absence of fatty acids, FadR will attach to the gene fragment, and block the entire gene expression.

Figure 12. pFadD_Lac promoter and FadR

Directed evolution (DE) is a method used in protein engineering that harnesses the power of selection to evolve molecules, typically proteins or nucleic acids, from large, stochastically permuted pools or combinatorial libraries, to obtain a desired properties. It mainly comprises the following steps:

(1) gene diversification by random mutagenesis and/or gene recombination to generate a diverse library of variants;

(2) screening/selection to obtain variants with improved phenotypes.

(3)The improved variant will serve as a new starting point for the next round of gene diversification. [4]

The experimental process we designed is shown in the figure:

Figure 13. The process of directed evolution

We obtain the sequence of FadR directly by cell disruption of E. coli DH5α. After obtaining FadR, we further combined it with araBAD promoter. The following is our design of plasmid:

Figure 14. The map of pBAD33-FadR

After that, we designed to transfer the obtained plasmid into DH5α which already contained a plasmid with a fatty acids-sensitive promoter and GFP, induce it with different concentrations of fatty acids, and measure the fluorescence intensity.

At the same time, we obtained a mutation library of FadR by error-prone PCR, and performed the same operations as for FadR that were not subjected to error-prone PCR.

The gene sequencing results indicated that we successfully get FadR gene and introduce random nucleotide mutations into the parent sequence qualitatively because there are overlapping peaks.



Figure 15. Partial FadR sequencing results: a)results of PCR b) results of error-prone PCR

We haven’t finished this DBTL cycle due to the lack of time, but more work will be done in the future.

When discussing the problem of the time consumption of the directed evolution module, the students in the hardware group designed a microfluidic chip to help us. This is described in detail in the DBTL section of the physical model.

In summary, in this module, we have completed a complete DBTL cycle and obtained parts with acceptable performance, and we are performing a second cycle in order to make the engineered bacteria more sensitive to fatty acid. At the same time, we optimized our experiment by cooperate with hardware. Although some of the results are not completely obtained due to the lack of time, we believe that we are nearing success.

DBTL cycle of gene knockout module

In our design, a defect type of strain is constructed to prevent pollution in case of engineered bacteria escaping. In the literature review process, we found several common types of gene defects, including nutritional defects, cell wall defects and so on.[5][6][7] We chose to construct the deficient strain by knockout, because it is easier to ensure the normal growth of the deficient strain by replenishing certain nutrients compared to other methods. Among the common nutritional defects, the construction technology of amino acid deficient strains is relatively mature, and amino acids are often used in cosmetics as beneficial ingredients, so we chose to construct amino acid deficient strains.

Escherichia coli histidine operon is composed of 9 genes including hisL、hisG、hisD、hisC、hisB、hisH、hisA、hisF and hisIE. They are responsible for the synthesis of enzymes needed for histidine synthesis in E. coli. HisDCB gene in histidine operon is responsible for the last three steps of histidine synthesis: aminotransference, dephosphorization and dehydrogenation. They are essential components of histidine synthesis, and their absence will cause the nutritive deficiency type of histidine, making the growth of strain dependent on histidine.[7]

pKD46 plasmid was transfected into DH5α and the expression of λ -RED system was induced by arabinose. PCR amplification with pKD13 as template was used to obtain 1.4 kb DNA fragment. The 5'-end primer hisBCD-knockout-F included bases homologous to the upstream sequence of hisB gene of host DH5α chromosome and priming sequence of pKD13 plasmid. The 3'-end primer hisBCD-knockout-R includes bases homologous to hisD downstream of host DH5α chromosome and priming sequence of pKD13 plasmid. The PCR product was electrically transferred into DH5α receptor state containing pKD46 plasmid, and the recombinant was screened on kanamycin resistant LB plate with glycine.[7]

Figure 16. The template of hisBCD knock-out (1394bp) in DH5α

However, there was no colony growth on the resistance screening plate, indicating that no homologous recombination reaction occurred in the bacteria. We tried again but failed again with no colony growth on the resistance screening plate.

After analysis, it is speculated that probably the base sequence of the target gene is too long, resulting in the decrease success rate of gene knockout experiment.

Therefore, we tried to look for single-gene knockout strains that had been successfully conducted. And the Kerio collection appeared to be a good reference.[8][9] Based on this list, we selected glyA as the target gene. Gene glyA encodes protein serine hydroxymethyltransferase that converts serine to glycine, transferring a methyl group to tetrahydrofolate, thus forming 5,10-methylene-tetrahydrofolate (5,10-mTHF).

Furthermore, another problem rised our attention. Considering the fact that DH5α is genetically modified to amplify plasmids, probably it is not suitable for gene knockout. In addition to using the DH5α strain as the chassis, we decided to use Escherichia coli MG1655 strain simultaneously for gene knockout. Based on the chromosome gene sequences of the two strains, we redesigned the gene knockout primers. Similar to the previous experiments, the primers included a sequence of bases upstream or downstream of gene glyA and the priming sequence of the pKD13 plasmid.

Figure 17. The template of glyA knock-out (1395bp) in MG1655

We overcame many difficulties in the experimental process and repeated four rounds of gene knockout experiments. Unfortunately, due to bacillus contamination in the laboratory, the first round of knockdown experiment based on DH5α strain failed. Although there was colony growth on kanamycin resistant plate, it was most likely not DH5α strain due to the appearance of the colony. In the second round, DH5α was used as the chassis for knockout experiment, and no colony growth was found on kanamycin resistant plate, which may be due to problems in experimental operation, such as low sensitive state activity or high voltage during switching, etc. In the third round, MG1655 was used as the chassis for knockout experiment, but also ended in failure. In the fourth round of knockout experiment, we used DH5α as the chassis. Colony growth was found on kanamycin resistance plate, though gene sequencing results indicate that bacteria still had glyA gene. Considering that the colony is kanamycin-resistant, it is likely to have off-target effect in the knockout process.

Figure 18. Colony growth on kanamycin resistant plate. (a) The first round of knockdown experiment (b) The fourth round of knockdown experiment

Although off-target effect is very common in gene knockout, we still believe that some methods can be used to minimize the off-target rate and increase the success rate. Due to the lack of time, we did not carry out the next round of gene knockout attempt. But we planned to redesign the primers for gene knockout to make them more specific based on the lessons of previous experiments. Although there is a description of glycyrrhizin nutritional deficiency caused by glyA gene defects in Kerio Collection, we still retain the suspicion that glyA gene is so important for bacterial growth that it is difficult to knock out. In further validation, we still hope to try to construct E. coli strains with different genetic defects.

DBTL cycle of Hardware

Our ultimate goal is to obtain highly expressed promoters through directed evolution, so that our bacteria have high protein expression and further achieve the effect of acne treatment. While one of the biggest challenges in directed evolution is the need for high throughput. To solve this problem-raised from our design, an automated, high throughput microfluidic chip was developed for directed evolution.

Initially, I used a square array to try to capture cells and microbeads. (the structure is shown below in Figure 19)

Figure 19. The capture efficiency of square arrays
1. 80*80 square array; 2. Bird head channel; 3. Air chamber; 4. Microbeads

We designed a simple experiment: Vacuum the chip and solvate each array with 5 ml of bead loading buffer and use a 200 microliter pipette to transfer 200 microliter solution containing 100000 beads, in a drop-wise fashion, to the surface of each array, and finally observe it with the microscope. (For more detailed experimental procedures, please refer to Hardware 4.4)

But we cannot achieve only one microbead per array (often two or more, show in Fig.1). Moreover, great difficulties occurred in the matching between two layers of chips. Thus we redesigned the chip with smaller circular channels, hoping for better results. (The structure is shown below in Figure 20)

Figure 20. The circular array
Figure 21. Matching between two layers of chips

The results show that the improved circular array can capture one microsphere in one array, which lays a solid foundation for the subsequent directional evolution step (More detailed results can be found at Hardware).


[1] Faye T, Langsrud T, Nes IF, Holo H. Biochemical and genetic characterization of propionicin T1, a new bacteriocin from Propionibacterium thoenii. Appl Environ Microbiol. 2000 Oct;66(10):4230-6. doi: 10.1128/AEM.66.10.4230-4236.2000. PMID: 11010864; PMCID: PMC92290.
[2]Pechsrichuang P, Songsiriritthigul C, Haltrich D, Roytrakul S, Namvijtr P, Bonaparte N, Yamabhai M. OmpA signal peptide leads to heterogenous secretion of B. subtilis chitosanase enzyme from E. coli expression system. Springerplus. 2016 Jul 28;5(1):1200. doi: 10.1186/s40064-016-2893-y. PMID: 27516938; PMCID: PMC4963352
[3] Byrd AL, Belkaid Y, Segre JA. The human skin microbiome. Nat Rev Microbiol. 2018 Mar; 16(3):143-155. doi: 10.1038/nrmicro.2017.157. Epub 2018 Jan 15. PMID: 29332945.
[4] Wang Y, Xue P, Cao M, Yu T, Lane ST, Zhao H. Directed Evolution: Methodologies and Applications. Chem Rev. 2021 Jul 23. doi: 10.1021/acs.chemrev.1c00260. Epub ahead of print. PMID: 34297541.
[5] Liu, Daniel Yuen-Teh et al. “Global gene expression profiling of wild type and lysC knockout Escherichia coli W3110.” FEMS microbiology letters vol. 276,2 (2007): 202-6.
[6] 史新昌. 含真核表达质粒的营养缺陷型大肠杆菌工程菌株作为基因治疗口服递送载体系统的研究[D].第四军医大学,2014.
[8] Baba, Tomoya et al. “Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection.” Molecular systems biology vol. 2 (2006): 2006.0008.