Team:WHU-China/Engineering
The Design-Build-Test-Learn (DBTL) cycle has always been regarded as one of the major ideas of iGEM and synthetic biology. Due to the limited time of performing experiments this year (we didn’t return school until September actually), not much useful data was acquired from laboratory. However, we still endeavored to learn from what we obtained for improving our methodologies, and meanwhile paid extra efforts to dry lab work for demonstrating our designs. No matter how tough the situation was, the idea of engineering and DBTL was always kept in our minds, and finally we did generate expected results to a certain degree. Together with the results, four DBTL cycles are described in detail below.
Quorum quenching enzymes have been placed at the core position of our design. As there are two
classes of them (AHL lactonases and AHL acylases) and they have various substrate specificity
(prefer short-chain or long-chain AHLs or both), together with the fact that P.aeruginosa has two
quorum sensing systems composed of one short-chain and one long-chain AHL, we really need to select
the appropriate enzyme or enzyme combination to obtain better AHL degradation effects. After looking
through plenty of literatures for a long period, we learned that the products of AHL degradation
catalyzed by AHL acylases are more stable so the reaction is usually irreversible while the reaction
catalyzed by AHL lactonases is reversible. So we initially planned to choose several AHL acylases as
the candidates. However as most of the acylases prefer to degrade long-chain AHLs and there is not
much research on those that can cover a broad spectrum, we still chose one lactonase as the
candidate[1]. Finally, we decided to use three broad-spectrum acylases (AiiC[2], AiiO[3] and
PvdQ[4]) and one well-known broad-spectrum lactonase (AiiA) to conduct further experiments. Besides,
in order to utilize the experimental data more efficiently, we also designed a quorum dynamics model
to help us. By connecting the Km value we got from the experiment with the model, we can rapidly
decide the best enzyme(s) to use. For more details, you can see the model part of our wiki.
We designed four devices of these enzyme genes. But when we asked company to synthesize these genes,
the failed to synthesize pvdQ and aiiO genes for some reason. And we failed to get pvdQ gene from
P.aeruginosa directly by colony PCR. Therefore we only built the two devices in Figure 1. But again
the process was hard as the aiiC was so large that it was difficult to be ligated with a plasmid. So
we got many false positive colonies. And we couldn’t use colony PCR to check whether we built the
vector successfully as doing PCR on such a large gene was tough as well(only enzyme digestion
verification can be used). So although confirmed by the enzyme digestion test, the vector really
should be tested further. As for aiiA, which was successfully synthesized and put on pET28a(+)
plasmid by the company, it was much easier to test the proper ligation of it.
Fihure 1: the maps of pSB1C3+aiiC and pET28a (+)+aiiA.
Next step we wanted to test the functions of the two devices. We firstly used the commercial cell-free kit (Promega: E. coli S30 Extract System for Circular DNA) to express the AiiA. After purification and SDS-PAGE, we failed to see the target band (Figure 2).
Figure 2: SDS-PAGE to test AiiA after purification of the cell-free reaction system. L1: control group, without purification. L2-L5: purified and eluted by different concentrations of imidazole.
From the experiment we learned that the cell-free kit may not be used to express a large amount of proteins. So we then tried to transform the pSB1C3+aiiC into E.coli BL21(DE3), and used IPTG to induce the expression of the enzymes. Next, the standard AHL solutions and reaction with crude cell extract were tested by the method described in the measurement (used crude cell extract instead of purified enzymes) part of our wiki. And the HPLC results were as follows (Figure 3).
(A)
(B)
Figure 3: HPLC analysis of the standard solutions of C4-HSL(A) and 3-oxo-C12-HSL(B) (1mg/ml). A: the two peaks for C4-HSL occur at about 6 min and 8.4 min. B: the peak for 3-oxo-C12-HSL occurs at about 4.3 min.
These results showed that we successfully proved the availability of our method. As for the measurement of crude cell extract reacting with AHLs, the results were shown in Figure4. From these data (compared with the results of standard solutions above, certain amount not shown), we observed that the area of one of the C4-HSL peaks didn’t change much while the other peak reduced significantly after reacting with the crude cell extract (Figure4 A-B). And in the other reaction, the area of the 3-oxo-C12-HSL peak reduced to nearly the half (Figure4 C-D). So we assumed that the enzymes may be expressed by the cells and they could degrade AHLs with both short and long chains. However, as the height and area of all the peaks of our samples were too small, due to the small amount of AHLs we added to the reaction system maybe, the errors may be too large. So the data we got may not be convincing and we couldn’t tell whether the enzymes were successfully expressed. However, the preliminary experiments on both the standard solutions and our samples at least prove the possibility of using HPLC to quantify the AHLs which corresponds to the methods mentioned in other literatures. Still further experiments are needed to decide the amount of AHLs used in this experiment.
(A)
(B)
(C)
(D)
Figure 4: HPLC analysis of the crude cell extract reacting with AHLs. A: control group of C4-HSL reaction system: inactivated crude cell extract (heating at 95℃ for 5 minutes) mixed with C4-HSL. B: experimental group of C4-HSL reaction system: crude cell extract mixed with C4-HSL. C: control group of 3-oxo-C12-HSL reaction system: inactivated crude cell extract (heating at 95℃ for 5 minutes) mixed with 3-oxo-C12-HSL. D: experimental group of 3-oxo-C12-HSL reaction system: crude cell extract mixed with 3-oxo-C12-HSL.
Moreover, the relatively unsuccessful attempt of enzyme expression made us learn a lot. First, long genes and high GC content genes are not easy to get either from synthesis or colony PCR. Second, the traditional enzyme digestion and ligation method of molecular cloning is not that efficient. So in the next season, we may use Gibson assembly to do the molecular cloning and try to avoid cloning long genes. In addition, the first DBTL cycle made us think more---as we really wanted to find enzymes with broad specificity towards substrates, are there any ways to directly create one which will largely save our time? After doing another period of research, we put forward an idea that we could find the ancestral enzyme of the existing enzymes. Because ancestral enzymes usually went through less evolution and their substrate specificity was very broad. Although they may be less catalytically efficient, we could use directed evolution on the enzymes (introduce several mutations that are essential to their catalytical efficiency) to obtain both efficient and broad-spectrum ones. With the help of our mathematic model, we successfully found several ancestral enzymes of our candidates（Figure 5）. Due to the limited time, we only tested them in silico. But after optimization, the sequences we got can be used directly next season, which will greatly reduce the time of selecting candidate enzymes from such large amount of existing quenching enzymes. And the enzymes we created may be even more efficient than the existing broad-spectrum enzymes! For more details, you can see the model part in our wiki.
Figure 5: phylogenetic tree of PvdQ protein, generated by iTOL.
Above all, through the first DBTL cycle and initiating a second cycle, we are trying to tackle the enzyme selecting problem we meet, which we believe is approaching success!
Chemokines are the crucial effectors of the sensing module in our design, since we want to express them in our engineered bacteria to recruit immune cells, mainly monocytes and neutrophils to clear pathogens. We have chosen 7 chemokines (CXCL1, CXCL2, CXCL3, CXCL8, CCL2, CCL3 and CCL8) as our parts collection of chemokines, which are all known with chemotaxis effects on monocytes and neutrophils and have no reported glycosylated modification. Chemokines from CXCL family are known to have better chemotaxis effect on neutrophils and CCL family is known to have a preference to attract monocytes [6]. To evaluate the chemotaxis effects of the 7 chemokines and select out 2 to 4 chemokines which are suitable for our project, cytological experiments were conducted. In these experiments, we successfully cultured THP1 cell line as our model of monocytes and carry out the Transwell migration assay and flow cytometry to evaluate their chemotaxis effect on THP1 cells. Finally, we selected out CCL2 as the best chemokine part to attract monocytes at our given chemokine concentration (10 nmol/l)[Figure 6 and Figure 7].
Figure 6. Transwell migration assay result of CCL2 treatment, shown by cell number.(n=3)
Figure 7. Chemotactic index(CI) of 7 chemokines to THP1.(n=3)
In our plan, we will furtherly use differentiated HL-60 cells as the model of neutrophils and
evaluate the chemotaxis effect of these 7 chemokines to select out most suitable 1 to 2 chemokines
to attract neutrophils. Gradients of concentration are considered to be set as well, so that the
chemotaxis effect of all chemokines can be analyzed at different concentration levels and the most
suitable range of chemokine concentration can be identified. Nevertheless, the preference of
chemokines to one of the immune cells restricts their use, since both monocytes and neutrophils are
important in acute inflammatory reactions. Two chemokines may need to be produced simultaneously.
Through searching information, we found that chemokines may interact with each other and strengthen
or weaken their effectiveness [7]. Thus combined Transwell migration assay of 2 chemokines should be
carried out to test their interaction effect and select the best combination. Later, the selected
chemokine parts will be expressed in our engineered bacteria. Finally, purified products expressed
and secreted by our engineered bacteria will be examined by Transwell migration assay again to
validate their functional efficiency. However, because of time limitation caused by COVID-19, we did
not carry out these experiments, which are planned to be done next year.
Besides, in order to maintain the secretion system with sufficient sensitivity while not causing
serious immune factor storms due to leaked expression, we decided to add a secretion signal peptide
to the N-terminus of the cytokine, and add a TEV protease cleavage site between that and the
chemokine itself. In this way, the fusion chemokine can be cleaved by the co-expressed TEV protease
to remove its secretion signal peptide and inhibit its secretion. As you might expect, this system
has the risk of causing the expressed chemokines to accumulate in bacteria chassis, so we added the
ssrA rapid degradation tag to the C-terminal of the cytokine to antagonize its side effects. In our
plan, this part is an important part of safety. GFP will replace chemokines for expression and
secretion in E. coli to verify whether the cleavable secretion signal peptide and ssrA tag work as
expected compared to the form without them. Unfortunately, the PCR reaction of adding a homology arm
to GFP for recombination failed to obtain the target band. This error prevented our urgent
experiment from proceeding as expected to quantify the expression of intracellular and extracellular
proteins, so it was difficult for us to determine whether the system we designed works
effectively.
Although our experiments are not complete, we did other work to help prove and refine our design.
Our model work of inflammation storm helps us to evaluate the risk to causing inflammation storm
with in vivo chemokine treatment and may help direct dose and time of in vivo chemokine treatment.
Our hardware design of an organoid based on micro-fluidic chip will help us to validate and evaluate
the effectiveness of in vivo chemokine treatment by our engineered bacteria without using any living
organisms. Furthermore, based on the data we got from Transwell test, we have created a new part
collection composed of several chemokines to share with other teams for future usage. You can see
the parts section in our wiki for more
details!
The envisioned workflow of cell-free quorum sensing includes: (i) in vitro prototyping of quorum
sensing networks; (ii) in silico screening of quorum sensing inhibitors and in vitro rapid and
high-throughput testing; and (iii) in vivo testing for validation. In phase I, we planned to
complete in vitro prototyping of quorum sensing networks but failed. However, the idea of
design-build-test-learn cycle was still implemented throughout our cell-free experiments, and we
wanted to share our experiences.
To reconstitute the basic logic of las and rhl systems of P. aeruginosa in cell-free systems, four
plasmids were successfully constructed with pSB1A3 as the backbone (Figure 8), which include: (i)
Plasmid-T (K2965011+LasR; K2965011+RhlR) will overexpress LasR and RhlR in E. coli BL21(DE3) after
IPTG induction, and the bacteria will then be lysed to prepare cell-free systems; (ii) Plasmid-L
(J23100+LasI; K649000+LasI) will express LasI to synthesize OdDHL in cell-free systems, and achieve
a positive feedback by its second device; (iii) Plasmid-R (K649000+RhlI; R0071+RhlI) will express
RhlI to synthesize BHL in cell-free systems under the control of Plasmid-L, and achieve a positive
feedback by its second device; (iv) Plasmid-F (K649000+GFPmut3b; R0071+mCherry) will express
GFPmut3b and mCherry under the control of in vitro established las and rhl systems, and the green or
red fluorescence will be measured by a microplate reader to quantify the activated level of these
two systems. Another plasmid (C) was constructed (J23100+GFPmut3b) as a negative control in
subsequent drug screening process, for validating the function of potential inhibitory molecules.
Figure 8: Four recombinant plasmids constructed to reconstitute the basic logic of las and rhl systems.
Initially, we attempted to make a proof-of-concept of cell-free quorum sensing by a commercial cell-free kit (Promega: E. coli S30 Extract System for Circular DNA), with four plasmids co-expressed in one pot. However, the attempts failed and no fluorescence was detected by the microplate reader. Considering the regular method of expressing one plasmid per cell-free reaction, we deemed that the imbalance of plasmid expression and energy consumption led to the failure. Therefore, we turned back to the original plan, as preparing lysates with pre-expressed LasR and RhlR to express the rest three plasmids in a chronological and hierarchical order. Besides, we suspected the addition of SAM (an important co-factor and a substrate of acyl homoserine lactones) might sustain the biosynthesis of OdDHL and BHL, whereby helping to establish the systems. Subsequently, sixteen 20-μL cell-free reactions with different reaction conditions were established and incubated in 30℃ for 4 hours, and several reactions ‘turned’ green! We got crazy for quite a while, but later discovered blue light didn’t excite the expected green fluorescence, and the colour was more likely the colour of diluted E. coli lysates (Figure 9), totally different from the beautiful results in a research article [11].
Figure 9: No fluorescence observed in cell-free quorum sensing systems under white light (left) and blue light (right).
Then a SDS-PAGE assay was performed to answer our questions. It’s strange that the protein bands during IPTG inducing process (0, 0.5 hour, 1 hour, 1.5 hour, 2 hour) were very light, thus cannot serve as the expected negative controls. However, considering that the molecular weight of LasR was predicted as 26.62 kDa and RhlR 27.58 kDa, these two proteins seemed not to be pre-overexpressed in the lysates (Figure 10). Several factors might contribute to the failed overexpression. Firstly, to obtain efficient cell extracts, the protocols [9,10] varied in the appropriate incubation time after IPTG induction, while a protocol described ‘If enzyme overexpression is the objective, incubate in 30℃ shaker for 4 h, or ~8 doublings, postinduction for protein expression; If cell-free protein synthesis is the objective, incubate in 30℃ shaker to reach OD600 of 3’ [10]. However, in our design, we need to both make use of pre-expressed proteins and implement cell-free protein synthesis in the lysates, which means we should re-evaluate the inducing time and conduct a set of preliminary experiments. Also, the quality of IPTG might be a problem, as a senior in the laboratory mentioned this based on his experience on the bag of IPTG we’re using.
Figure 10: SDS-PAGE shows LasR and RhlR seem not to be pre-overexpressed in the lysates.
Unfortunately, there has been no time left for us to prepare cell-free systems for another round of DBTL process. Hopefully, after debugging these factors, we will prototype las and rhl systems in vitro, and facilitate massive screening of potential quorum sensing inhibitors whereafter.
Our ultimate goal is to construct engineered probiotics that will face Gram negative pathogens directly in lower respiratory tracts, being capable to quench their quorum sensing and expel them by recruiting immune cells. This firstly demands the survival of our probiotics in lower respiratory tracts: they should grow well and should not be excluded or poisoned by the presence of G- pathogens (e. g. Pseudomonas aeruginosa). So, how viable our probiotics are when they are exposed to G- respiratory pathogens has become a problem. To solve this problem raised from our design, a co-culture device was built (the structure is shown below in Figure 11; further information is available on Hardware page), aiming to find out whether our probiotics will be poisoned by G- pathogens in a lower-respiratory-tract-like environment.
Figure 11: the co-cultivation device and its components.
Control element:
1-3: Three-way valve; 4-9: Double pass valve; 10-13: Pressure-Controlled needle (including
pressure-controlled needle piston);
Sampling and ventilation elements:
14: Pressure regulating ventilation pipe; 15, 18: Sterile sampling tube; 16, 19: Bacteria sampling
tube (consisting of rubber pipe, disposable bacterial filter, double-pass valve, joint and
countersunk head of rubber pipe); 17, 20: Sampling mouth of bacterial liquid in the left culture
bottle (consisting of rubber pipe, double-pass valve, joint and countersunk head of rubber pipe)
Other components:
21, 22: Culture bottle; 23, 24: Multi-use bacterial filters; 25: Completely closed bulkhead; 26:
Connecting bridge
This system is basically composed of a culture of probiotics (E. coli Nissle 1917 (EcN)) and a culture of G- pathogens (Pseudomonas aeruginosa PAO1), which are separated by two sieves that allow controlled medium exchange but prevent cells to pass. EcN and P. aeruginosa PAO1 were cultured to log phase and added separately into this device, which was kept at 37℃ (similar to lower respiratory tract environment). Medium from P. aeruginosa PAO1 incubator was ejected into the EcN incubator to test that whether EcN would be affected by P. aeruginosa PAO1 metabolites.
Figure 12: OD600 value versus time in co-cultivation device (experiment) and conical flask (control).
To confirm that the growth of our probiotics will not be interfered by the metabolites of
Pseudomonas aeruginosa and to test the basic function of our hardware, we designed a simple
experiment: First, we added 2ml bacteria solutions (which was cultured overnight) to 120ml fresh LB
medium (two flasks for EcN and one for P. aeruginosa PAO1) and cultured them in a shaker (37℃,
200rpm) for 2 hours to obtain an initial bacteria density. And then we poured one flask of EcN and
P. aeruginosa PAO1 in the left and right bottle separately. And we used the hardware to mix the
medium (without mixing the bacteria!) of the two bottles and took 2ml samples per 45 minutes to
measure the OD600 value in a spectrophotometer as the quantification of bacteria density since then.
The same procedure was also used on the other flask of our probiotics which was not mixed with the
medium of P. aeruginosa PAO1as the control group. And we used Matlab to draw the growth curve of the
bacteria in the three bottles. The mixing point was regarded as the 0 point at x axis. .
From Figure 12 we can see that the growth curve of our probiotics in the hardware is similar to that
in the control group. Although maybe due to the medium we used, the density of P. aeruginosa PAO1 is
not so high, both the pathogens and probiotics are still growing. This partially proves that our
probiotics can be cultured with P. aeruginosa PAO1 normally, which provides evidence for the
possibility of using our probiotics as chassis to fight against the P. aeruginosa PAO1 in our lungs.
Besides, the function of our hardware is verified again..
In this DBTL cycle, based on the first arising problem of the practicability of our project in
applicational situations—tolerance of EcN to pathogen metabolites, we designed a non-contact
co-cultivation device and tested the wild type EcN growth curve under a condition of limited cell
contact and controllable media exchange. The result revealed that it is possible for EcN to work in
the anticipated environment. Due to the limited time, further experiments, e. g. testing the
tolerance of fully engineered EcN to pathogen metabolites are required. Nevertheless, these works
still laid the first foundations of realization of our project in the next season.
1.Susanne Fetzner. Quorum quenching enzymes[J]. Journal of Biotechnology, 2015, 201:2-14.
2.Manuel R, Diggle S P, Stephan H, et al. Quorum quenching activity in Anabaena sp. PCC 7120:
identification of AiiC, a novel AHL-acylase.[J]. Fems Microbiology Letters, 2010(1):73-80.
3. Czajkowski R, Krzyzanowska D, Karczewska J, et al. Inactivation of AHLs by Ochrobactrum sp. A44
depends on the activity of a novel class of AHL acylase[J]. Environmental Microbiology Reports,
2011, 3(1):59-68.
4. Bokhove Marcel, Jimenez et al. The quorum-quenching N-acyl homoserine lactone acylase PvdQ is an
Ntn-hydrolase with an unusual substrate-binding pocket.[J]. Proceedings of the National Academy of
Sciences of the United States of America, 2010.
5. Yi-Hu, Dong, Gusti et al. Identification of Quorum-Quenching N-Acyl Homoserine Lactonases from
Bacillus Species.[J]. Applied & Environmental Microbiology, 2002.
6. Kenneth Murphy. Janeway’s Immunobiology[M]. New York: Garland Science, 2011. 99-103
7. von Hundelshausen P, Agten SM, et al., Chemokine interactome mapping enables tailored
intervention in acute and chronic inflammation. Sci Transl Med. 2017 Apr 5;9(384):eaah6650.
8.
Nuzzi PA, Lokuta MA, Huttenlocher A. Analysis of neutrophil chemotaxis. Methods Mol Biol.
2007;370:23-36.
9. Zhang Y, Huang Q, Deng Z, Xu Y, Liu T. Enhancing the effificiency of cell-free protein synthesis
system by systematic titration of transcription and translation components. Biochemical Engineering
Journal, 138 (2018): 47-53.
10. Karim AS, Jewett MC. Cell-Free Synthetic Biology for Pathway Prototyping. Methods in Enzymology,
608 (2018): 31-57.
11. Stark JC et al. BioBits™ Bright: A fluorescent synthetic biology education kit. Science
Advances, 4 (2018).
12. Merkl, R. & Sterner, R. Ancestral protein reconstruction: techniques and applications. Biol Chem
397, 1-21, doi:10.1515/hsz-2015-0158 (2016).
