Team:Lund/Engineering
Introduction
During our entire project, we made an effort to follow the engineering cycle guideline of Design, Build, Test, Learn and then repeat. We did this from the early brainstorming part to the practical DNA work. We faced more difficulties trying to produce our proteins than expected, resulting in a lack of time to wrap up our experiments before the deadline. We were, however, able to apply the engineering cycle throughout our many attempts of incorporating our construct into our vector, which we succeeded within the end. We decided to divide our attempts into three main cycles, constantly optimizing our design.
DNA and Cloning Work
General design: Choice of inhibitors
To start, we read up on different options for inhibiting curli and considered constructing a pathway to create plant metabolites, such as coumarins or epigallocatechin-3-gallate, that have shown promising results of curli inhibition [1][2]. In the end, we decided it would be more feasible if we produced one protein at a time instead of an entire pathway of enzymes as required by the metabolites.
We looked into proteins like the chaperone curli assembly protein C called CsgC since it has been described in the literature as a natural curli inhibitor. Curli-producing bacteria such as certain Escherichia coli strains use CsgC to hinder curli aggregation inside of themselves [3]. We also found an exciting peptide that we decided on calling the CsgF peptide (originally named PeptideNN17R) as it is derived from the CsgF protein naturally occurring in the curli system. The CsgF peptide blocks the secretion pore of the curli system [4]. When searching for amyloid-specific proteins, we came across the enzyme DegP. DegP can function as protease and a chaperone, a protein assisting other proteins to fold. DegP is specific for misfolded proteins, meaning it can break down amyloids [5][6]. Therefore, we decided to evaluate it in our experiments.
Further, we found one peptide, designated QFGGGNPP, that, according to a previous study, could inhibit curli directly. However, our computer model [insert link from model here] indicated that this was likely not the best option due to the high concentrations needed for curli inhibition [7]. Instead of making our own peptide inhibitor, we decided to study peptides that inhibit other amyloids. There are indications that cross-seeding fibrillation might occur between curli and human proteins such as amyloid-beta or alpha-synuclein. Cross-seeding fibrillation is when a small amount of amyloids, a seed, induces the formation of amyloid aggregates [8]. Cross-seeding indicates that the different amyloids consist of some similar parts, which could mean that the peptide inhibitors that work for one amyloid might also work for another amyloid. Looking at peptides that can inhibit amyloid-beta, which is involved in Alzheimer’s disease, we found two that seemed promising, and we wanted to test if they could inhibit curli as well [9]. We decided to study DB3DB3 and tANK6 due to them showing potential as amyloid inhibitors [10].
In summation, the five inhibitors chosen and eventually used in all cycles were CsgC, CsgF peptide, DegP, DB3DB3, and tANK6.
Cycle 1
Our project began when we decided to aim to suppress curli in the gut by engineering a probiotic. When browsing through our options of which probiotics could be suitable, we decided on Limosilactobacillus reuteri due to its many health benefits and because it has been researched in Lund [11][12]. The goal of the first iteration of the engineering cycle was to create a first implementation of the suggested solution using L. reuteri as our chassis. We needed an expression system for our five inhibitors described above to allow for expression and secretion by L. reuteri.
Firstly, the shuttle vector pTRKH3-ermGFP was selected since it has been previously tested in L. reuteri with positive results [13], and using L. reuteri was a central part of our project at this stage. The plasmid was constructed for a trial of protein expression using a GFP construct, marked in green in Figure 1 below, as such pTRKH3-ermGFP is not designed for use with other inserts. Therefore, we initially found no documentation of precisely where the promoter, terminator, and ribosome binding site were located. Consequently, we concluded that we would need to include these in our construct to have them at the desired place. Our promoter of choice was CP44, a standard part and the strongest promoter of the CP collection [14]. The terminator we used was the double terminator T7 and T1rrb. The terminator was chosen since the 2019 iGEM Oxford Team used this in an attempt to end transcription of the GFP construct in pTRKH3-ermGFP [15]. The selected ribosome binding site was Elowitz standard RBS, chosen for its well-documented use [16].
Figure 1: Circular map of pTRKH3-ermGFP, made in Benchling [17].
Secondly, four signal peptides were selected to enable the secretion of our inhibitors: L. lactis Ups45, Lb. reuteri mub, B. licheniformis amyL, and SLPmod Lactobacillus Secretion Tag. All these have been previously shown to be effective in L. reuteri, although SLPmod has only been used in L. reuteri by one previous iGEM team, Oxford 2019 [18][19].
In order to assemble our construct, we decided to mainly focus on Gibson assembly and have restriction enzyme cloning as a second option. For the Gibson assembly, we intended to use the restriction enzyme MscI. When designing our inserts for this cycle, we included two sequences homologous to those present on the vector around the MscI restriction site. In addition, the restriction enzyme sites BamHI and NcoI were also included in the construct to allow for restriction enzyme cloning.
To initiate the Building phase of cycle 1, we started with digesting the plasmid in one location only, in the middle of the previous GFP sequence, to insert our new construct. We intended to disrupt the expression of GFP this way. Further, a Gibson assembly was conducted to put our construct together. However, we did not achieve our desired results and started re-evaluating our design. One of our concerns was that since the ermB (erythromycin ribosomal methylase) promoter, used in the original study [13], was not removed, short RNA strands might still be transcribed and translated. The ermB promoter of the pTRKH3-ermGFP plasmid would still be active since we would use erythromycin to select for the bacteria with our plasmid. While this would first have been determined upon having a successful transformation, we decided that the risk of this occurring was enough to terminate the first cycle. Thereby, we chose not to try restriction enzyme cloning and instead redesign our experiments for cycle 2.
Cycle 2
The purpose of the second iteration of the engineering cycle was to improve the expression of the inhibitors. With some advice from our supervisor, we changed the singular restriction site used previously to two sites, BamHI and EagI, present on either side of the GFP construct on the plasmid. Using these sites would completely remove the GFP construct, thus eliminating unnecessary disruptive transcription and translation caused by the ermB promoter. The restriction sites were introduced to the inserts via PCR amplification with primer overhangs. After that, we applied both Gibson assembly and restriction enzyme cloning to create our construct and build our system.
During both cycles 1 and 2, we faced many difficulties with our transformants with unpredictable growth on our plates, including growth on the negative controls, which should not carry an erythromycin resistance gene. Prior to Gibson assembly in cycle 1, erythromycin concentrations were optimized to ensure the significant difference in growth on positive and negative controls. However, throughout cycles 1 and 2, we saw that these results were not reliably replicable. It was becoming clear that the selection marker, erythromycin, was not efficient for cloning work in E. coli, as the strains we tried to use for genetic transformation appeared to be naturally resistant. Thereby, we decided to reevaluate our design once again.
Cycle 3
For the third iteration of the cycle, we aimed to work with a well-tested robust expression system: pET-11a. pET-11a is a high expression E. coli vector containing an ampicillin resistance gene, as opposed to an erythromycin resistance gene in the pTRKH3-ermGFP plasmid. Therefore, we could avoid the selection issues we had with erythromycin. However, switching to this vector meant we could not express our inhibitors in our intended chassis, L. reuteri. After considering the amount of time remaining for the project, we ultimately decided to aim for expression in E. coli instead. With higher chances of having a successful transformation, we could have the possibility to perform an inhibition assay with the inhibitor produced by the genetically modified E. coli. When working with pET-11a we performed subcloning in E. coli TG1 and E. coli XL1-blue and used E. coli BL21(DE3) as the expression host.
Switching the expression system to pET-11a meant that we had to redesign the inserts used for the genetic transformation. New restriction enzyme sites, NdeI and BamHI, present in the cloning region of the pET-11a vector, were introduced to the inhibitors using PCR amplification with primer overhangs. The restriction sites were added to the start and end part of the insert encoding the inhibitors, thereby removing promoter, RBS signal peptide, and terminator.
Since the original signal peptide sequences we chose are inefficient in E. coli, these would have to be replaced if secretion in the new expression was to be achieved. To have the option to attempt this, we ordered primers introducing signal peptides for secretion in E. coli and oligonucleotides containing our inhibitors with the new restriction enzyme sites and new signal peptides. Due to time constraints, neither the primers with new signal peptides nor the new oligonucleotides were utilized. Instead, the inhibitors were only to be produced intracellularly with the option to lyse the cells to purify the inhibitors.
When applying the third cycle, we successfully built our system with one of our inserts, CsgC, for the first time. The first step in the Testing phase was to check if the inhibitor was being expressed. An expression assay was performed twice, but the results of both were inconclusive. Future experiments are required to fully provide evidence that our inhibitors are expressed and functional in curli inhibition. Our suggested approaches are further described in future experiments.
Curli Detection
As this part of the project does not entail building a system, it does not strictly follow the engineering cycle. However, the protocols used were constantly optimized, making the work cyclical similar to the engineering cycle, with the omission of the build step. In other words: Learn, Design, Test, and repeat.
During the initial Learning phase, appropriate strains and methods were identified via literature search, described under [insert link from experiments here]. However, since curli detection is not well described in the literature, different protocols had to be tested to perform different experiments to produce curli using E. coli MC4100 cells.
The most important parameters that changed during the experiments were the incubation time, temperature, sonication method of the culture, and the purification of the possible curli fibers. Standard protocols [insert a link from experiments here] were initially followed but later modified to get a clearer result when analyzing the fluorescence. In total, five experiments were performed.
Incubation time was not determined initially, as it is not known at what point the cells begin to produce curli. Hence, it was important to have a large time spectrum to be able to analyze our samples. When purifying the curli fibers, the protocol established by Y. Puwei [20] was followed, with some modifications. This protocol was chosen as it studied the same E. coli strain that was used in this project. The protocol was further modified in some experiments.
The first change in the protocol was the use of SDS (Sodium dodecyl sulfate). In the beginning, the SDS solution was used during the final steps to diminish impurities in the final sample according to literature [20]. After observing the results, SDS was avoided for the following experiments due to the risk of denaturing the curli fibers. If impurities were present in the samples for the following experiments, those were not considered when analyzing the spectrofluorometry data.
For the second experiment, apart from avoiding the SDS step, longer cultivation times were tested, as it was not known when the amyloid fibers were formed. The first experiment might not have given good results because of the short cultivation time. At this point, the cultivation time of the following experiments was increased up to a range of 100-200 h.
Furthermore, a temperature test with one culture was performed. Temperature conditions were changed because although the optimum growth temperature of the bacterium is 37ºC, the temperature at which the fibers are produced optimally may be another. As stressing the cells may affect the curli fibers production, two temperatures, 37°C, and 26°C, were tested. However, E. coli could be grown at other temperatures than these to study if the cells produce more curli while under stress. From this point, both temperatures were used in experiments.
A further change in the protocol was made regarding the amount of PBS buffer to resuspend the pellet. A smaller volume was used in order to obtain a more concentrated sample. The purpose of having a concentrated sample is that when measuring the ThT fluorescence, as measured in a cuvette of 1 mL, it is easier to detect curli fibers than having it in a more diluted one.
Finally, the sonication method was changed. From the beginning, an ultrasonic water bath was used, according to Puwei, Y [20]. As no fluorescence was detected, a possible explanation could be that the fibers were not released from the cells due to insufficient sonication. For that reason, an ultrasonic probe was used, and two different methods regarding intensity and duration were studied. The best results were the ones performed with 2x1 min of sonication at 80 % intensity. This method was used from that moment onward.
References
[1] Thakur S, Ray S, Jhunjhunwala S, Nandi D. Insights into coumarin-mediated inhibition of biofilm formation in Salmonella Typhimurium. Biofouling. 2020;36(4):479-491.
[2] Hengge R. Targeting Bacterial Biofilms by the Green Tea Polyphenol EGCG. Molecules. 2019;24(13):2403.
[3] Evans M, Chorell E, Taylor J, Åden J, Götheson A, Li F et al. The Bacterial Curli System Possesses a Potent and Selective Inhibitor of Amyloid Formation. Molecular Cell. 2015;57(3):445-455.
[4] Yan Z, Yin M, Chen J, Li X. Assembly and substrate recognition of curli biogenesis system. Nature Communications. 2020;11(1).
[5] Fang K, Jin X, Hong S. Probiotic Escherichia coli inhibits biofilm formation of pathogenic E. coli via extracellular activity of DegP. Scientific Reports. 2018;8(1).
[6] Hauske P, Meltzer M, Ottmann C, Krojer T, Clausen T, Ehrmann M et al. Selectivity profiling of DegP substrates and inhibitors. Bioorganic & Medicinal Chemistry. 2009;17(7):2920-2924.
[7] Cherny I, Rockah L, Levy-Nissenbaum O, Gophna U, Ron E, Gazit E. The Formation of Escherichia coli Curli Amyloid Fibrils is Mediated by Prion-like Peptide Repeats. Journal of Molecular Biology. 2005;352(2):245-252
[8] GtR [Internet]. Gtr.ukri.org. 2021 [cited 11 October 2021]. Available from: https://gtr.ukri.org/projects?ref=BB%2FM02427X%2F1
[9] Perov S, Lidor O, Salinas N, Golan N, Tayeb- Fligelman E, Deshmukh M et al. Structural Insights into Curli CsgA Cross-β Fibril Architecture Inspire Repurposing of Anti-amyloid Compounds as Anti-biofilm Agents. PLOS Pathogens. 2019;15(8):e1007978.
[10] Schartmann E, Schemmert S, Niemietz N, Honold D, Ziehm T, Tusche M et al. In Vitro Potency and Preclinical Pharmacokinetic Comparison of All-D-Enantiomeric Peptides Developed for the Treatment of Alzheimer’s Disease. Journal of Alzheimer's Disease. 2018;64(3):859-873.
[11] Chen L. Exploring the Propanediol Utilization Pathway in Lactobacillus reuteri. Lund: Division of Biotechnology, Lund University, 2018. 180 p.
[12] Forskningsoutput - Lunds universitet [Internet]. Portal.research.lu.se. 2021 [cited 11 October 2021]. Available from: this link
[13] Lizier M, Sarra P, Cauda R, Lucchini F. Comparison of expression vectors in Lactobacillus reuteri strains. FEMS Microbiology Letters. 2010;308(1):8-15.
[14] Part:BBa K1033225 - parts.igem.org [Internet]. Parts.igem.org. 2021 [cited 11 October 2021]. Available from: http://parts.igem.org/Part:BBa_K1033225
[15] Part:BBa K3183021 - parts.igem.org [Internet]. Parts.igem.org. 2021 [cited 11 October 2021]. Available from: http://parts.igem.org/Part:BBa_K3183021
[16] Part:BBa B0034 - parts.igem.org [Internet]. Parts.igem.org. 2021 [cited 11 October 2021]. Available from: http://parts.igem.org/Part:BBa_B0034
[17] Benchling [Biology Software]; 2021 Retrieved from https://benchling.com.
[18] Wu C, Chung T. Green fluorescent protein is a reliable reporter for screening signal peptides functional in Lactobacillus reuteri. Journal of Microbiological Methods. 2006;67(1):181-186.
[19] Part: BBa BBa K3183009 - parts.igem.org [Internet]. Parts.igem.org. 2021 [cited 11 October 2021]. Available from: http://parts.igem.org/Part:BBa_K3183008
[20] Puwei, Y. Characteristics and chemical properties of Escherichia coli Curli. (2015).