Introduction

With Reactive oxygen Species at the center of many diseases, harnessing them as a marker would enable innovative and broad research, diagnostics and therapeutic avenues. We have, thus, designed a cross-species ROS-detection platform aimed at detecting harmful ROS concentrations which naturally occur in diseases. Such a sensor could be paired with a detection and/or therapeutic output depending on the aim. As a proof of concept, we chose to focus on 2 diseases, each highly cumbersome: IBD in humans, and fireblight in plants. (see Description for more details)

Species and strain choice

Species choice is crucial since each organism presents different intrinsic ways of coping with ROS and oxidative stress and will, therefore, harbor a specific intracellular environment (pH, redox, electric, temperature variations…). These intrinsic factors will affect the platform’s characteristics once inserted. Within the same strain, even growth conditions will affect the system, but here we need to focus on the most important variables (ie. species, strain). Since our main work is to tune expression, it is important from the beginning to select a host which will be as close as possible to the end application host.

• For our medical case-study the choice was easy, as E. coli Nissle 1917 is the most engineered microbiota strain used in several studies for similar chronic inflammation therapies in the gut(1, 2). Comparison with classic lab-strains like E. coli DH5a and K-12 would be relevant to assess the importance of strain variability on the oxidative stress response.
• For our agritech application, we considered all the main bacterial biocontrol agents as potential vectors. Indeed, many currently used biocontrol sprays are bacteria-based (Pseudomonads and Erwiniaceae), like Pseudomonas fluorescens A506, EPS62e, Pseudomonas protegens Pf5 or Pantoea agglomerans/vagans E325(3–5). At the suggestion of Cosima Pelludat and Theo Smits, we looked into Pantoea vagans C9-1 “white” strain, which displays interesting characteristics(4, 6). Its multiple metabolic deficiencies restrict its lifespan in nature to a few days, alleviating biosafety concerns regarding GMOs and their dissemination. However, its carotenoid deficiency could impact its resistance to UV. All these strains would have been relevant to use as vectors, and we hope to be able to test them all with more time and collaborations. For now, we only managed to get P. protegens Pf5.
• Pseudomonas putida strains are used for biocontrol and many bioprocesses, showing strong tolerance of high ROS levels(7, 8). Therefore, we chose to include P. alloputida mt-2 kt2440 to our study, provided by the lab.

Bacterial plasmids design

oxyR-oxyS-gfp and oxyR-J23101-PL-oxyS-gfp

The reasons behind the design

This plasmid comes from the Panke lab, and was provided by Dr. Tsvetan Kardashliev. The native oxyRS regulon drives expression of oxyR in one direction, and of oxyS in the other. In the Panke plasmid, oxyS was replaced by the gfp gene. In this instance, oxyR negatively regulates its own expression, so a GFP peak will naturally decrease shortly, when oxyR stops being produced (through negative feedback). However, we want a stable expression of our output protein. To bypass oxyR’s impact on its regulation, we redesigned the promoter region to include a constitutive promoter for oxyR (part BBa_J23101 Berkeley 2006) and the sequence was adjusted next to the oxyR gene as done in BBa_K1104200 (NYMU-Taipei 2013). A polylinker (referred to as PL) was added between the new promoter (BBa_J23101) and the oxyRS regulon (named oxyS promoter), as shown in Figure 1.

Cloning design

The goal was to order synthetic DNA for the whole promoter region (J23101-PL-oxyS, see sequence in the Twist table below) but due to lab changes, we had to actually order a way larger fragment encompassing also oxyR and half of the GFP where ideal restriction sites were positioned (NcoI and BspHI). Due to issues with enzymes, the ligation did not work out and we had to amplify the whole backbone from the BspHI homology region to NcoI homology region (see Engineering details below).

oxyR-dmoxyS_OG-gfp and oxyR-dmoxyS_BS-gfp

The OG plasmids explanation behind the mutations

oxyR-dmoxyS_OG-gfp plasmids are the oxyR-oxyS-gfp but with designed mutations in the binding site region. These mutations were predicted computationally to affect binding more or less strongly (see Model page). However, we cannot predict beforehand the impact on the phenotype and ROS-sensing response in vivo. Therefore, we tested several combinations of predicted mutations (variation in number, association of strong and soft, different positions). Here is a table of designed sequences (from the most mutated to the least):

Understanding the aim behind the duplication

For oxyR-dmoxyS_BS-gfp, however, we duplicated the binding site region (BS). Why? If you look at this scheme (Figure 2), you can see that the oxyR binding site regulates the expression of two promoters. There is a positive regulation of the gfp gene, since the binding happens just before the -35 region where the polymerase binds, whereas it negatively regulates oxyR because the binding happens between the -35 and -10 regions, interfering with the transcription complex binding.

We, thus, designed a version where there are 2 distinct oxyR binding regions (BR), as represented Figure 3. That way, we can try and alleviate negative feedback on the oxyR gene expression with strong mutations in BR2, and choose soft mutations for BR1 regulating GFP expression, in the hopes of tuning the sensing to high-ROS levels. It is also possible that BR1 promotes oxyR expression since it is now in a favorable position compared to the oxyR -35 site.
Since we never know what happens in vivo, we have designed 9 versions, testing various mutation strengths, numbers and combinations.

Here is a table of designed sequences (from the most mutated to the least):

Cloning design

All plasmids were obtained by Gibson Assembly (backbone amplification from GFP N-ter and oxyR N-ter).

oxyR-oxyS-katG-gfp

For the medical application, we chose to test the influence of catalase-expression on H202-induced stress and response, since the ultimate goal would be to counteract elevated levels of ROS in chronic inflammation. We aimed to use an intracellular H2O2 titration assay but planned to keep a fluorescent reporter in case it did not work (good call), so we added the catalase gene just before the GFP. We would expect GFP levels to soar less with katG than without, since the catalase would help the bacteria counter the stress more efficiently. H202 would therefore be degraded quicker, leading to oxyR being less activated and thus GFP less expressed. Note: GFP would be naturally less expressed since it is the second gene on the RNA.

Cloning design

The catalase gene in E.coli is katG, so we decided to genomically amplify it from our bacteria and clone it before the GFP through Gibson Assembly (parallel amplification of the backbone).

Yeast plasmid design

Yeast plasmids were based on the plasmid GFP-X0 we got from Addgene #115565 suited for bacterial expression (cloning phase) and yeast expression (sensor-testing phase). In yeast, the strong constitutive promoter adh1 induces expression of GFP, followed by the cyc1 terminator.

CaMV35S-gfp and trx2-gfp

To test TGA2- and yap1- mediated sensing, we had to add them to the GFP-X0, where GFP would be the reporter protein. The TF-regulated promoter should regulate GFP and not adh1 anymore, so an adh1-compatible terminator was added to close the ORF (padh1-tadh1), and added the CaMV35S/Trx2 promoter after (in front of the gfp gene, see Figure 4 below). However, we first added the adh1 terminator and the corresponding promoter, so that way we would obtain a plasmid which could give the background activation due to native yap1 or cellular environment.

Cloning design

We genomically amplified tadh1 with 2 PCRs, one for CaMV35S homology and the other for trx2 homology. We ordered the 2 promoters through Twist (see sequences in the Twist table below), and then did a 2-fragment Gibson Assembly (the terminator, the promoter, and the digested backbone on EcoRI+BamHI).

tga2-CaMV35S-gfp and yap1-trx2-gfp

Tga2 and yap1 were ordered with Twist and cloned through Gibson Assembly (SpeI+BamHI sites). Where tga2 worked on the first try, we never managed to integrate yap1. Since yap1 was too long for synthesis, it was separated into 2 parts. We thought maybe the homology between the 2 was too short, but ordering a new sequence did not work. We also have trouble preparing the backbone, with apparent restriction enzyme binding to the plasmid (higher bands than the plasmid), and switched to PCR amplification (see Results).

mScarlet versions of adh1-TGA2-tadh1-CaMV35S-GFP, adh1-tadh1-trx2-GFP and GFP-X0

Since the autofluorescence of yeast was impeding interpretation of fluorometric assays, we aimed to replace the GFP genes by the mScarlet gene (600 nm emission peak, also where plants do not autofluoresce).

Cloning design

We found a plasmid for ymScarletI expression with compatible restriction sites on Addgene called pDRF1-GW ymScarletI, and we planned for ligation.

mScarlet-XO V2

However, no expression was noticed, we thought it might be because the distance between the transcription start and translation start was longer for this design (also translation end). For GFP-X0, we thus designed primers for Gibson Assembly, in order to preserve the same spacing in the mScarlet-X0 as in the current GDP-X0 (between the adh1 promoter and cyc1 terminator) and managed to obtain mScarlet expression. Running out of time, we did not have time to clone mScarlet the same way in the other plasmids.

epPCR plasmids

We aim to amplify the promoters (CaMV35s, Trx2, oxyS, J23101-PL-oxyS) of all our key plasmids through error-prone PCR to obtain random mutagenesis of these regions. Flow cytometry would be then used for highthrough-put screening of mutants (GFP+ or GFP- screenings) to select those with the best ROS-detection range. The plasmids were designed to keep restriction sites between each genetic element, so we first tested Gibson Assembly of the epPCR library with XmaI+EcoRI digest of the plasmid. There was very low efficiency, therefore, we also tried backbone amplification from EcoRI (end of promoter) to XmaI (beginning) (see Results).

Engineering process

GFP-XO

GFP-XO is the backbone that was used for cloning of all yeast plasmids. It was ordered from Addgene (Plasmid #115565).

pDRF1-GW ymScarletI

pDRF1-GW ymScarletI was the source of the mScarlet gene we used to insert into our yeast constructs. It was orders from Addgene (Plasmid #118452).

oxyR-oxyS-GFP

The initial oxyR-oxyS-GFP plasmid was designed in the Panke lab (ETH Zürich) and was kindly provided to us by Dr. Tsvetan Kardashliev (University of Basel). As it already contained all components for ROS-dependent GFP expression, this plasmid provided an important baseline for measurements and comparison with further engineered plasmids. We first exchanged the promoter region of the original plasmid to modulate oxyR expression, creating the oxyR-J2311-PL-oxyS-GFP plasmid. To achieve desensitization of this ROS-reporter to desired thresholds, we employed multiple approaches for promoter engineering. For the oxyR plasmids, this mainly included error prone PCR and directed mutagenesis on the oxyS promoter.

oxyR-J2311-PL-oxyS-GFP

Exchanging the promoter began by excising the old promoter through restriction enzyme digestion with NcoI and BsphI. The digestion was gel purified and the backbone Gibson assembled with a new_oxyR DNA fragment from TWIST. The product (oxyR-J2311-PL-oxyS-GFP) was transformed into DH5ɑ and colonies were checked by sequencing. TECAN fluorescence assays comparing oxyR-J2311-PL-oxyS-GFP and oxyR-oxyS-GFP indicated that the new promoter did in fact perform worse than the original. For this reason we decided to focus our engineering efforts on oxyR-oxyS-GFP.

Error-prone PCR - random mutagenesis

Error prone PCR is a technique that was utilized to introduce random mutations into the oxyS promoter in order to influence the binding behavior of oxyR to oxyS. To increase the error rate of Taq polymerase during PCR amplification, MnCl2 was added in varying concentrations (0.1 - 0.8 mM). The oxyS promoter was then amplified using the primers old-oxyR-epPCR-fwd and old-oxyR-epPCR-rev. After PCR, DpnI was added to each reaction to selectively digest the methylated, original DNA. This should reduce the background of unmutated fragments. The obtained fragments were then cloned into the oxyR-GFP backbone, creating oxyR-epoxyS-GFP plasmids. Since we had to remove the original promoter from the plasmid to insert the potentially mutated version, we initially wanted to digest oxyR-oxyS-GFP with restriction enzymes. However, by that time we had realized that digestions do not work well on that plasmid, and decided to do backbone amplification instead. We therefore amplified the backbone from oxyR-oxyS-GFP, excluding oxyS, with oldBB-fwd and oldBB-rev primers and then performed Gibson assembly with the purified epoxyS fragments from epPCR. The products were then transformed into DH5ɑ and colonies screened by TECAN fluorescence measurements. Promising colonies, namely oxyR-epoxyS-GFP 0.6 MnCl2 col. 1 and col. 2 were later sequenced. Interestingly, they both did not contain mutations, despite showing considerably different responses to H2O2. EpPCR was also attempted for oxyR-J2311-PL-oxyS-GFP. However, we did not manage to generate mutated fragments.

Directed mutagenesis

After not achieving convincing results with epPCR, we switched to a directed mutagenesis approach. This happened in tight conjunction with the computational part of our project, as a model was fitted to predict specific mutations in the promoter sequences that would decrease the binding affinity to the transcription factor oxyR. A set of 19 DNA fragments containing combinations of these mutations in oxyS were ordered from IDT. For cloning, we amplified the backbone from oxyR-oxyS-GFP using BB-PCR-GA-F2 and Backbone-PCR-GA_R primers and then inserted the dmoxyS_OG and dmoxyS_BS fragments via Gibson assembly. DH5ɑ colonies from transformation were sequenced to confirm cloning success, before being analyzed in TECAN fluorescence assays. In the end we successfully cloned and tested 5 oxyR-dmoxyS_OG-GFP and 7 oxyR-dmoxyS_BS-GFP variants. For the remaining 7 fragments we did not have enough time left.

Tga2-CaMV35S-GFP and CaMV35S-GFP

CaMV35S-GFP was constructed by adding tAdh1 and CaMV35S to the GFP-XO backbone. Tga2-CaMV35S-GFP was then assembled from CaMV35S-GFP by inserting Tga2. GFP-XO was digested with BamHI and EcoRI. tAdh1 was amplified from yeast BY4743 genome by PCR using tadh1-FW and CaMV35S-Rv primers, thereby adding compatible ends for Gibson assembly. Digested GFP-XO backbone was then assembled with tAdh1 and CaMV35S through Gibson assembly, followed by transformation into DH5ɑ. Colonies were checked for successful cloning by restriction enzyme digestion of extracted plasmid. Digestion with EcoRI and SphI was expected to produce fragments of 6845 bp, 626 bp and 331 bp for CaMV35S-GFP. Gel electrophoresis of the digested plasmids revealed expected fragment sizes only for CaMV35S-GFP colony 4, indicating successful cloning in that colony. (Figure 5)

To add Tga2 to CaMV35S-GFP, the backbone was first digested by BamHI and SpeI. Tga2 DNA fragment (TGA2_SpeI_BamHI) was synthesized by TWIST and inserted into CaMV35S-GFP via Gibson assembly followed by transformation into DH5ɑ. Colonies were once again checked by digestion and gel electrophoresis of obtained plasmids. Digestion was performed with SphI and the expected fragments (7337 bp, 1616 bp) were observed for Tga2-CaMV35S-GFP colonies with varying intensity. (Figure 6)

After receiving the relevant primers, both Tga2-CaMV35S-GFP and CaMV35S-GFP were sequenced, further validating cloning. For testing and quantifying the function of these constructs, they were transformed into yeast BY4741.

yap1-trx2-GFP and trx2-GFP

trx2-GFP was cloned by adding tAdh1 and Trx2 to the GFP-XO backbone. Similar to the construction of CaMV35S-GFP, tAdh1 was amplified from yeast genome using tadh1-FW and trx2-Rv primers, the latter making it compatible for cloning with the Trx2 DNA fragment (UAS_pTRX2, synthesized by TWIST). tAdh1 and Trx2 were then inserted into BamHI-EcoRI-digested GFP-XO via Gibson assembly. The product was transformed into DH5ɑ and the colonies checked by digestion and gel electrophoresis of the obtained plasmids. Digestion with EcoRI and SphI should produce fragments of 6845 bp, 626 bp and 331 bp length for trx2-GFP, which was observed on the gel, thereby confirming cloning success (Figure 5). Later on, the trx2-GFP plasmids were validated by sequencing. For testing and quantifying the function of trx2-GFP, it was transformed into yeast BY4741. Once we had successfully cloned the trx2-GFP plasmid, we attempted to add Yap1, the transcription factor that binds Trx2, to the construct. However, all of the approaches we tried to achieve this, failed. Initially, we tried restriction enzyme digestion of trx2-GFP with SpeI and BamHI, followed by Gibson assembly with yap1-Spel and yap1-BamHI. Yap1 was ordered as two halves from TWIST, since it was too large to be synthesized as one DNA fragment. Transformation into DH5ɑ yielded only one colony which was checked by plasmid digestion with SphI and gel electrophoresis (Figure 6). The gel revealed that cloning had been unsuccessful. Digestion and Gibson assembly have been repeated. The Gibson assembly product was amplified by PCR and checked by gel electrophoresis, indicating issues with assembly or digestion. We then tried a new Yap1_BamHI fragment with 21 bp longer (43 bp total) homology to the second fragment to improve assembly. However, this also failed. After testing different combinations of restriction enzymes with different conditions and doing gel electrophoresis, we found out that BamHI was not cleaving correctly. Using a fresh batch of enzyme also did not deliver correct fragments. We subsequently switched to backbone amplification PCR of trx2-GFP. Unfortunately, we had difficulties amplifying the backbone with correct length (~7900 bp) and high enough concentration. First, the pTRX2_BamHI_F and pTRX2_SpeI_R primers were used and failed. Then, Yap1y-BB-fwd2 and Yap1y-BB-rev primers were designed and tested. With the new primers, we managed to obtain fragments of correct size, but could not purify it in sufficient amounts. Gibson assembly with Yap1 DNA fragments did not succeed. We successfully amplified Yap1 from yeast BY4741 genome, yet Gibson assembly with the backbone did not work. This was likely due to the poor quality of the backbone amplification.

Trx2 epPCR

In an effort to reduce affinity of Yap1 to the Trx2 promoter, we performed error prone PCR (epPCR) on the promoter to introduce mutations. For this we used trx2-epPCR-fwd and trx2-epPCR-rev primers and performed the reactions at different Mn2+ concentrations ( 0.1 - 0.8 mM) to tune the error rate. Afterwards, the PCR products were incubated with DpnI to digest the original, non-amplified DNA. The amplified and potentially mutated Trx2 promoter was inserted into XmaI-EcoRI-digested trx2-GFP (Trx2 removed) and transformed into DH5ɑ. However, we only obtained non-mutated colonies through this method (determined by sequencing).

mScarlet plasmids

After measuring strong autofluorescence of yeast during TECAN fluorescence assays, we exchanged GFP to mScarlet to separate the emission wavelength from that of yeast’s autofluorescence. For this purpose, GFP had to first be removed from all of the yeast GFP plasmids by restriction enzyme digestion using EcoRI and SalI and subsequent gel purification. The mScarlet fragment was obtained by digestion of pDRF1-GW ymScarletI with the same enzymes (gel purification) and then ligated with the backbone. Ligations were transformed into DH5ɑ, colonies sequenced and correct plasmids transformed into BY4741. TECAN fluorescence assays with mScarlet plasmids however did not show any signal, regardless of condition. The original pDRF1-GW ymScarletI worked as expected. Due to the relative locations of restriction sites on the plasmids, cloning positioned mScarlet relatively far from the promoter and terminator in our target plasmids (79 bp and 72 bp respectively). We suspected this could lead to it not being expressed properly. To address this problem, we amplified mScarlet from pDRF1-GW ymScarletI with mScarletX0-F and mScarletX0-R primers, thereby adding compatible ends for Gibson assembly closer to the gene itself. The amplified mScarlet was subsequently assembled with EcoRI-digested GFP-XO (GFP removed) and transformed into DH5ɑ. In the new construct, mScarlet is 49 bp and 27 bp away from the promoter and terminator respectively. After confirmation by sequencing, mScarlet-XO V2 was transformed into BY4741 and tested for fluorescence in a TECAN plate reader. This time, mScarlet fluorescence could be detected. Due to time constraints, we did not manage to reclone, transform, and test all yeast plasmids (trx2-GFP, Tga2-CaMV35S-GFP, CaMV35S-GFP) with mScarlet V2.

mScarlet-XO

The mScarlet-XO plasmid was constructed from GFP-XO by exchanging GFP to mScarlet. First, mScarlet was excised from pDRF1-GW ymScarletI by restriction enzyme digestion with EcoRI and SalI. The mScarlet fragment was purified from agarose gel after gel electrophoresis. GFP-XO was digested with EcoRI and SalI to remove GFP. The backbone was gel purified as well. Digested GFP-XO (GFP removed) was then ligated with the extracted mScarlet fragment using ligase. Ligations were transformed into DH5ɑ. Emerging colonies were sequenced to confirm cloning.

Primer sequences