Detection Engineering Cycle
One of the major challenges when working with PFAS is their detection. The current gold standard is mass spectrometry, a powerful but costly and technically demanding technique. Our objective was to develop a biological tool capable of detecting PFAS in water samples: a biosensor. Since the scope of the Fluorobreaker team focuses on TFA and PFOA, we aimed to develop two distinct biosensors, each dedicated to one of these molecules.
A biosensor is defined by two key properties:
- Specificity: the signal must be unique to the target molecule
- Sensitivity: the ability to detect the molecule at low concentrations.
Indeed, a biosensor that produces a positive signal for hundreds of unrelated molecules, or only responds at extremely high concentrations (e.g., 5 mol·L⁻¹), would be of little practical use. This engineering cycle therefore focuses on designing, building, testing, and characterizing a biosensor that meets these two requirements. In certain circumstances, a biosensor can provide a semi-quantitative indication of the target molecule’s concentration. In our project, achieving this feature would represent a valuable added advantage for the final product. To construct a biosensor, an easy-to-handle chassis is required. For this purpose, we selected E. coli MG1655, a well-characterized bacterial strain commonly used for transformation and heterologous protein expression.
A biosensor typically relies on two main components:
- A promoter that responds specifically to the target molecule.
- A reporter gene that generates a measurable signal.
Identifying promoters responsive to PFOA and TFA was therefore essential for our project. For PFOA, we relied on transcriptomic data from RNA sequencing performed by Wintenberg & al (2025) on E.Coli. Differential expression (DEG) analysis allowed us to identify candidate genes with a sufficiently high log₂ fold change (L2FC), i.e. greater than 1. Among these, we selected b0002 (thrA, L2FC = 5.28) and b3021 (mqsA, L2FC = 2.67). These two genes were chosen for complementary reasons: b0002 showed a high number of reads, while b3021 had a much lower read count. This contrast could help us determine which type of promoter is more suitable for a biosensor. Indeed, a relatively low expression level may limit background (leaky) activity but risks producing a signal too weak to detect, whereas high expression levels provide stronger signals but may also increase leakage. In addition, the two genes are involved in distinct metabolic pathways, further increasing the chance of identifying a promoter with a specific response. b0002 participates in L-homoserine biosynthesis, while b3021 is implicated in the antitoxin response. We then retrieved their promoters from EcoCyc, selecting approximately 200 bp upstream of their respective ATG start codons. However, since no comparable transcriptomic data were available for TFA, we had to adopt a different approach (see the TFA RNAseq Results).
Design 1.1
To monitor promoter activity, a reporter gene was required. After discussions with field experts, we learned that bioluminescence reporters are generally preferred over fluorescence. Indeed, bioluminescent signals can be easily detected with simple light-sensitive devices, such as a smartphone. Moreover, the relationship between protein expression and luminescence is typically more linear than with fluorescence, which makes bioluminescence more suitable for developing semi-quantitative reporters (Zhou & al, 2020). The luciferase operon is widely used for this purpose and is a common component of biosensors. We chose to work with the LuxCDEAB operon, previously characterized as a composite part in the iGEM database, while retaining its original RBS and scar regions.
To achieve a more specific response, we redesigned the system by splitting the Lux operon into two separate operons, each controlled by one of our selected promoters. This architecture enhances specificity, as luminescence will only be produced if both promoters are active, ensuring expression of the complete operon.
To troubleshoot potential issues with promoter activity, we incorporated additional fluorescent reporters: mCherry under the control of one promoter(promoter of the gene b3021 : BBa_25WX0GI9) and GFP under the other (promoter of b0002 : BBa_257DQD3R). Both reporters are well characterized, and their fluorescence is driven independently by their respective promoters. This setup allows us to pinpoint the source of a failure: if no luminescent signal is observed, the individual fluorescent outputs indicate which promoter is non-functional. Using SnapGene, we assembled the promoters and operons to generate the fragment of interest, which is our collection part:
Figure 1 : plasmid containing both PFOA sensible promoters with rapporter construct
As a proof of concept, we first designed a fragment carrying inducible promoters, in order to validate the functionality of the split operon strategy. For this purpose, we used pTet and pLac, along with their associated regulatory elements (TetR, TetO, LacI, and LacO) to build fully functional inducible promoters. This design enables the testing of a biosensor induced by IPTG and anhydrotetracycline (ATC). We generate this fragment :
Figure 2 : Plasmid to test whether the construct is reliable with an inducible promoter.
Both constructs (one with inducible promoters and one with our PFOA-responsive promoters) were further optimized by codon optimization (only for coding sequences) and removal of forbidden restriction sites.
The next step was to select a backbone to generate a complete and functional plasmid. We chose the pSEVA261 backbone, a medium–low copy number plasmid previously validated by our PI in his own research. A relatively low copy number helps limit basal expression, thereby reducing unwanted background signals from leaky promoters. This plasmid also carries a kanamycin resistance cassette as a selection marker (see below).
Figure 3 : Map of pSEVA261 used for cloning.
Since the backbone was already available in the lab (we designed primers to linearise the backbone), we only needed to order the gene fragments. To comply with IDT and Twist size and technical limitations, the constructs had to be divided into three parts. We decided to use Gibson assembly to reconstitute the full plasmid.
To simplify the workflow and reduce fragment length, we first planned to assemble the IPTG/ATc-inducible biosensor. Once validated, this construct could then be modified by replacing the inducible promoters with the PFOA-responsive promoters. Accordingly, we ordered he inducible biosensor in three fragments (as shown below: “Insert1”, “Insert2” and “Insert3”), along with an additional small fragment containing the two PFOA-responsive promoters for later integration.
Figure 4 : Three fragments were incorporated into pSEVA261 by Gibson assembly. (A) Insert no. 1. (B) Insert no. 2.(C) Insert no. 3.
We planned to assemble the complete plasmid using Gibson assembly, designing homology regions at the ends of each fragment to enable seamless integration. The plasmid was constructed in SnapGene, and the design was validated through in silico Gibson assembly simulations to ensure successful reconstruction.
Figure 5 : Expected plasmid map after cloning
The assembly was to be verified by PCR using several primers spanning the junctions between fragments. Alternatively, next-generation sequencing could be used to confirm the exact sequence of the plasmid. Finally, bioluminescence and fluorescence emission could be measured using a Tecan plate reader.
Build 1.1
We performed Gibson assembly as described above, using the three designed fragments and the linearized backbone. The resulting constructs were transformed into heat-shock–competent E. coli MG1655, and transformants were selected on LB agar supplemented with kanamycin. We were able to obtain transformants.
Test 1.1
We tested our transformants by testing their fluorescent and luminescent signals, both negatively. PCR and sequencing of plasmids isolated from the transformants revealed only the empty backbone, indicating that the Gibson assembly had failed, probably a non-negligible part of the initial vector did not linearize. . Four colonies were tested, and in all cases the sequencing matched the empty backbone.
Figure 6 : Multiple alignment from miniprep and sequencing after transformation, showing that the integrated plasmid is the empty plasmid.
Multiple attempts to repeat the assembly under the same conditions were also unsuccessful; the transformation result consisted only of bacteria transformed with the empty plasmid.
Mini Learn
We identified several points that could be improved: linearization of the initial vector using a smaller amount of template, since this vector is what we consistently recovered from our transformants, indicating that it is at the core of the problem.
Following the first failure, our protocol included a DpnI digestion step to degrade methylated DNA, as this vector originated from a miniprep and E. coli methylates its genome at GATC sites.
A longer digestion time could help eliminate more of the initial vector.
Another parameter we decided to optimize was the incubation time of the Gibson Assembly reaction, originally set between 15 and 30 minutes.
Mini Design & Build
We redesigned our protocol by performing vector linearization through PCR using 1:100 of the initial DNA quantity (to reduce the amount of template DNA).
We extended the DpnI digestion time from 30 minutes to 1 hour and increased the Gibson Assembly incubation from 30 minutes to 1 hour.
The transformation protocol itself remained unchanged.
Mini Test
We again obtained colonies, but colony PCR and Sanger sequencing revealed that these were once more empty plasmids.
Mini Learn
Despite these optimization steps, the result remained unchanged.
This suggests that the issue likely lies in the design of the construct or assembly rather than in the cloning protocol itself, leading us to explore alternative strategies to save time and move forward efficiently.
Learn 1.1
Despite multiple attempts, the Gibson assemblies did not yield the expected plasmid.To explain this failure, we considered one main hypothesis: the complexity of the assembly with 4 long fragments.
To address this limitation, we had anticipated such difficulties and placed an order for a complete, ready-to-use plasmid from Azenta-Genewiz in order to validate the inducible plasmid, then perform a second assembly step to introduce our promoters. This strategy allowed us to move forward and proceed with plasmid characterization without being blocked by reconstruction issues (see Iteration 1.2).
Following these unsuccessful assembly attempts, we also realized that such a complex design could complicate the characterization of individual promoters. Since we had already struggled to successfully perform the Gibson assembly on this plasmid, we considered it safer and more efficient to run a separate, simplified experiment to characterize the promoters independently (see Cycle 2).
Design / Build 1.2
We ordered the same plasmid described in Design 1 from Azenta-Genewiz. This ready-to-use construct eliminated the technical limitations previously encountered. We then transformed MG1655 and successfully obtained transformants.
Figure 7 : Obtained transformants after Azenta-Genewiz plasmid transformation.
Test 1.2
The plasmid construction was verified by PCR and sequencing (despite being commercially synthesized), confirming that it corresponded exactly to the intended design.
Figure 8 : Alignment from miniprep and full plasmid sequencing after transformation Azenta - Genwiz transformation plasmid
To evaluate functionality, the transformants were incubated with IPTG (50µM) and ATC (10ng/mL) and fluorescence was measured using a Tecan (graph below). We monitored both mCherry and GFP signals, in addition to the bioluminescence signal. These measurements were primarily intended to validate our proof of concept. Therefore, we compared induced cultures to non-induced cultures and to non-transformed controls (lacking the plasmid). As the goal was to demonstrate differential induction rather than quantify expression levels, fluorescence data were not converted into absolute units. (image 9)
Figure 9: Normalized luminescence signal in E. coli DH5α transformed with either the biosensor construct or the empty plasmid, under different induction conditions: single induction with 10 ng/mL aTc or 20 µM IPTG, double induction with 20 µM IPTG and 10 ng/mL aTc, or without induction. * : p < 0,05 between the condition “Induction 20µM IPTG + 10 ng/mL aTc” and “non induction”, ** : p < 0,01 for the same conditions. # : p < 0,05 between the condition “Induction 20µM IPTG + 10 ng/mL aTc” and “Induction 10ng/mL aTc”, ## : p < 0,01 for the same conditions. † : p < 0,05 between the conditions “Induction 10ng/mL aTc” and “empty plasmid”, †† : p < 0,01 for the same conditions. (see Results for more detailed analysis)
Learn 1.2
We observed that luminescent output was only present under double induction, while single induction with IPTG alone did not trigger any detectable luminescence, as predicted by our design. Although induction with aTc alone also produced a luminescent signal (albeit weaker), this can be explained by the fact that the pLac promoter was found to leak significantly. As a result, the condition of induction with aTc alone effectively became a situation in which both promoters were activated simultaneously, which naturally led to luminescence emission according to our design. This indicates that the construct provides specific induction with minimal leakage, thereby confirming its functionality. Now that our plasmid is completed and functional, we can add in our PFOA-sensible promoters to create our PFOA biosensor. We also determined that both signals could be normalized using reference samples (not only with OD, but with the iGEM Measurement Kit) to ensure repeatable and comparable data, which is essential in the context of biosensor development. While not necessary for this artificially induced plasmid used as a control, this normalization strategy could be applied in future designs involving our PFOA-responsive promoters.
Design / Build 1.3
We planned to incorporate our PFOA-responsive promoters following the design strategy below:
The previous plasmid would be linearized by restriction digestion (see “restriction sites” on the plasmid map). A fragment containing both promoters and homology regions was ordered for Gibson assembly.
After assembly, we expected to obtain the following construct, which was again modeled in silico using SnapGene.
Figure 10: Figure 10: Azenta-Genewiz synthesized the plasmid after cloning the PFOA-sensitive promoters.
Unfortunately, due to time constraints, we were unable to assemble or test this construct, and therefore could not collect experimental data or implement improvements based on it.Test 1.3 (future)
Here, we describe the planned tests that would have been performed if the PFOA-responsive plasmid had been successfully assembled. Luminescence and fluorescence signals would have been measured after normalization using a known fluorescent standard. Different concentrations of PFOA would have been tested under conditions similar to those described in DBTL 2.
Design 2.1
After encountering difficulties with the initial plasmid, we decided to individually characterize our PFOA-responsive promoters. To this end, we designed a simplified plasmid containing only one promoter directly controlling the expression of a fluorescent reporter. Since our PI already possessed a plasmid encoding RFP under the control of the pLac promoter (inducible with IPTG), we decided to use this backbone as a platform to test the responsiveness of our promoters to PFOA. The plasmid was designed as follows:
Figure 11 : Map of the plasmid used for cloning the PFOA-sensitive promoters individually to control mRFP1 expression.
This plasmid also has an intermediate copy number, which fits our selection criteria. It differs from the previous construct by its selection marker, carrying a chloramphenicol resistance gene (CmR) instead of kanamycin resistance. The mRFP1 fluorescent protein used in this plasmid is highly similar to mCherry, sharing almost identical excitation and emission spectra, which ensures compatibility with our detection setup previously used.
Build 2.1
We used this plasmid and transformed it into E. coli MG1655. The presence and integrity of the construct were verified by PCR using primers FR15 and FR14 (see plasmid map). Transformed bacteria were selected on LB agar supplemented with chloramphenicol, ensuring plasmid maintenance during growth.
Test 2.1
We assessed plasmid functionality and RFP induction by exposing cultures to 500 µM IPTG. Fluorescence was monitored over time using a Tecan plate reader, allowing us to follow both growth and fluorescence dynamics.
We did see a positive RFP signal for IPTG induced conditions, allowing us to conclud that the systeme is reliable.
Figure 12 : RFP fluorescence response after IPTG induction, showing that the construct is reliable for further experiments.
Learn 2.1
With this first design, we confirmed that our plasmid is functional and can therefore be used to characterize our promoters (b0002 and b3021). We also verified that the RFP signal was comparable to mCherry, allowing meaningful comparisons with our other plasmid constructs.
Design 2.2
Building on the previous plasmid, we planned to linearize it by PCR using primers IGEM1 and IGEM2, in order to remove the inducible pLac promoter and LacI (see the image below).
Figure 13 : Linearized plasmid, removinf Lac O, pLac and Lac I
Both promoters were amplified from the fragment previously ordered for the DBTL 1 plasmid, using two pairs of primers: b3021FW/b3021RV and b0002FW/b0002RV. Below is the design of the PCR strategy for the b3021 promoter
Figure 14 :Illustration of how a single promoter was amplified and inserted into the reporter plasmid, where LacI, pLac, and LacO were removed.
The primers for promoter amplification were designed with overhangs homologous to the linearized vector to enable Gibson assembly. Once again, SnapGene was used to model the PCR and Gibson assembly in silico, ensuring that the design would assemble correctly, we obtain this construction (shown with b3021):
Figure 15 : Final plasmid containing the promoter of interest, driving mRFP1 expression.
Build 2.2
We constructed the plasmid as previously described and transformed the Gibson assembly product into E. coli MG1655 DH5α. The assembly and integrity of the construct were verified by PCR and confirmed by sequencing.
Figure 17 : Colony PCR was performed to assess whether the cloning was successful. Wells 1–5 contained colonies transformed with the plasmid containing BBa_257DQD3R, while the following wells correspond to colonies transformed with the plasmid containing BBa_25WX0GI9.
Test 2.2
We assessed plasmid functionality by incubating subcultures with increasing concentrations of PFOA (image 14), as well as positive (previous plasmid induced with IPTG) and negative controls (MG1655 WT i.e. non-transformed).
Figure 18: RFP fluorescence/OD response of E. coli MG1655 ΔRM transformed with the reporter construct BBa_25O14TR6, when exposed to different PFOA concentrations at t = 24h, when removing the water point. (n = 4).* : p < 0,05 between the condition “E.Coli transformed with BBa_25O14TR6” and “MG1655”, ** : p < 0,01 for the same conditions.
The results obtained for the b3021 promoter (shown below), and similarly for b0002, demonstrate that both promoters respond to PFOA in a dose-dependent manner, indicating that their activity correlates with PFOA concentration.
Learn 2.2
We assessed a dose-dependant effect with our RFP signal. It traduces a good response of both promoters regarding PFOA, which is a great proof of concept for using both promoters for a biosensor. We were also able to verify some of the characteristics we found in the transcriptomic data. Indeed, b0002 induces a higher signal than b3021, corroborating the transcriptomic data showing a higher number of reads for b0002. Finally, the b3021 promoter displayed the most discriminative RFP/OD response to increasing PFOA concentrations, with a pronounced difference between highly induced and non-induced samples after 24 hours of induction. This indicates that b3021 could be the most suitable candidate for further biosensor development.
Additionally, at this stage (in agreement with the conclusions from Learn 1.2, experiments performed at the same time), we determined that fluorescence signals could be normalized using reference samples to ensure repeatable and comparable data, an essential requirement for biosensor development. We had not performed this normalization earlier, as the promoters were not yet characterized. Implementing this step in future experiments would represent a significant improvement toward developing a reliable, ready-to-use biosensor.
Degradation Engineering Cycle
Design 1
Goal:
Demonstrate successful cloning of VNP plasmids with a fast and visible indicator.
Hypothesis:
Colonies with integrated constructs would fluoresce, enabling rapid screening.
Rationale:
Reduce dependency on lengthy sequencing at this early stage.
Build
Steps:
- Competent cells prepared freshly, transformed with ligation mix
- Plates incubated overnight
- Agarose gels run with digested constructs to confirm recircularization
Controls:
Included non-ligated backbone to assess background.
Equipment:
UV transilluminator for fluorescence, thermocycler for PCR verification.
Test
- Colonies were obtained with varying fluorescence intensities
- PCR confirmed the presence of insert in ~60% of fluorescent colonies
- Control backbone plate showed colonies (false positives), highlighting incomplete digestion/ligation
- Gel electrophoresis revealed faint unexpected bands, suggesting partial recircularization
Learn
- Fluorescence is powerful as a first-pass filter but not sufficient alone
- Re-circularization efficiency is critical
- Recommendation: Always pair fluorescence readout with confirmatory PCR
- Team identified need for stricter enzymatic digestion and purification steps to reduce false positives
Design
Objective:
Determine optimal ATC concentration for induction.
Hypothesis:
Higher ATC would produce stronger signals but potentially stress cells.
Experimental design:
Two levels chosen (20 ng/mL vs 100 ng/mL) to test balance between expression and growth.
Build
Methods:
- Cultures inoculated from identical starter overnight cultures to control variability
- ATC added at mid-log phase
- Two flasks for each condition to ensure reproducibility
Parameters logged:
- OD600 measured every hour
- Fluorescence intensity measured with plate reader
Test
- 100 ng/mL: High fluorescence but slower growth, some cells lysed prematurely
- 20 ng/mL: Healthy growth with sufficient signal for detection
- Data plotted as growth curves vs fluorescence intensity confirmed that moderate induction optimizes yield per biomass unit
Learn
- 20 ng/mL ATC balances induction strength and cellular fitness
- Excessive induction diverts resources to protein expression at the expense of viability
- This provided a benchmark concentration for downstream experiments
Design 3.1
Goal:
Improve plasmid yield for downstream applications.
Hypothesis:
Scaling culture volume and optimizing mini-prep kits would increase detection sensitivity.
Rationale:
Previous yields were insufficient for reliable PCR and assays.
Build 3.1
Approach:
- Shifted from 5 mL starter cultures to 50-100 mL mid-scale cultures
- Used high-quality plasmid kits with RNase treatment
- Added extra wash steps to enhance purity
Equipment:
Incubator shaker, refrigerated centrifuge.
Test 3.1
- Plasmid yield increased by ~8-fold compared to initial attempts
- Purity (A260/280 ratio) consistently ~1.9-2.0
- Downstream assays (restriction digestion, PCR, sequencing) performed more reliably
- Detection assays showed clearer signals due to higher input DNA
Learn 3.1
- Culture scale and plasmid integrity directly affect detection success
- Standard practice: Maintain a minimum scale of 50 mL per prep to ensure sufficient material
Design 4.1
Incident:
A mini-prep yielded no DNA.
Hypothesis:
Failure due to protocol omission rather than biological factors.
Aim:
Identify and correct error source.
Build 4.1
Actions:
- Retraced procedural steps, introduced checklist verification
- Repeated prep with careful timing, especially for neutralization and wash steps
- Ensured centrifuge properly cooled
- Included positive control plasmid prep to compare outcomes
Test 4.1
- Second prep produced expected DNA yields
- Checklist revealed initial run skipped a wash step
- Positive control confirmed equipment functioning correctly
- Data reinforced human error as root cause
Learn 4.1
- Procedural checklists became mandatory
- Critical: Emphasize training and discipline in routine methods
- Learning extended beyond plasmid prep to all workflows
Design 5.1
Problem:
False positives persisted due to uncut backbone plasmids during ligation.
Goal:
Eliminate background by re-linearizing backbone with PCR.
Rationale:
PCR linearization prevents re-ligation without insert.
Build 5.1
Method:
- PCR amplification of backbone with outward-facing primers
- Gel extraction of linearized fragment
- DpnI treatment used to degrade template plasmid
- Constructs rebuilt with insert ligations
- Proper controls included
Test 5.1
- Colony PCR confirmed >85% correct clones post-linearization, compared to ~40% previously
- Fluorescence intensity uniform, reflecting construct accuracy
- False-positive background drastically reduced
Learn 5.1
- Adopting backbone PCR re-linearization became standard
- Recommendation: Always use PCR linearization + DpnI cleanup in cloning workflows
Design
Need:
Ensure sterile and stable filtrates for assays.
Hypothesis:
Smaller pore size improves retention of contaminants.
Variables:
Pore size (0.45 μm vs 0.22 μm vs 0.1 μm).
Build
Method:
- Filtration conducted in parallel with identical volumes
- Flow rates measured, filter clogging recorded
- Outputs collected for purity testing
Equipment:
Sterile syringe filters, vacuum manifold.
Test
- 0.1 μm MCE filters: Best retention, though slower flow
- 0.22 μm: Allowed some breakthrough
- 0.45 μm: Inadequate for sterility
- Clogging manageable with pre-filtration of large particulates
Learn
- Adopt 0.1 μm filters as standard
- Trade-off in flow time acceptable for assurance of sterility and reproducibility
Design
Objective:
Formalize all validated strategies into standardized workflows.
Hypothesis:
Codified procedures reduce variability and increase reproducibility.
Scope:
Induction, cloning, plasmid prep, filtration.
Build
Implementation:
- Created Standard Operating Procedures (SOPs) incorporating optimal ATC, PCR linearization, mandatory checklists, 0.1 μm filters
- Training given to all team members
- Documentation shared in lab wiki for transparency and reproducibility
Test
- Replicated experiments under new SOPs showed reduced variability in growth curves
- Consistent plasmid yields and reproducible fluorescence
- Cross-validation between team members confirmed transferability of procedures
Learn
- Moving from exploration to standardized practice strengthens project foundations
- Codified methods ensure continuity for future cycles and teams
Design
Objective:
Verify the integrity of cloned constructs before scaling.
Hypothesis:
Some clones may carry mutations or defective regulatory regions.
Build
- Mini-preps performed on selected clones
- PCR used to amplify the 1.5 kb insert
- Products sent for Sanger sequencing
Test
- Most clones carried the correct insert
- Some (e.g., B1) contained point mutations
- Others (I1, J1) had altered RBS regions
Learn
- Sequence verification is essential before large-scale cultures
- Defective clones were re-transformed and resequenced
Design
Goal:
Increase colony yield and viability.
Hypothesis:
Host strain impacts transformation efficiency.
Build
- Compared DH5α with K16MG55
- Adjusted competent cell volumes to reduce resource use
- Negative controls included
Test
- K16MG55 produced significantly more viable colonies
- However, some clones (M, L, I, K) still failed to grow or were contaminated
Learn
- Host strain is critical for transformation efficiency
- Non-viable clones must be identified and repeated early
Design
Objective:
Determine whether protein is intracellular or secreted.
Hypothesis:
Vesicle release may contribute to extracellular protein.
Build
- Cultures centrifuged
- Supernatants filtered (0.45 μm, 0.1 μm)
- Fluorescence measured in pellets and supernatants
Test
- Strong fluorescence in pellets
- Very little in supernatants
- 0.1 μm filters frequently clogged
Learn
- Proteins remain mainly intracellular
- Purification efforts should focus on pellets
Design
Objective:
Maximize protein purity from pellets.
Hypothesis:
Wash number impacts yield vs purity.
Build
- Ni-NTA purification performed with 3, 4, or 5 washes
- Fractions analyzed via SDS-PAGE and Western blot
Test
- Three washes: Higher yield, lower purity
- Five washes: Excellent purity, low yield
Learn
- Trade-off required: five washes for analytical work, three for large-scale production
Design
Goal:
Optimize TEV protease usage to minimize contamination.
Hypothesis:
Lower protease volumes reduce residual contamination while maintaining cleavage.
Build
- Cleavage performed with 2, 1, 0.5, and 0.2 μL TEV
- Residual protease removed using MagneHis beads
Test
- 2 μL: Excessive protease contamination
- 0.5-1 μL: Good cleavage, minimal contamination
- 0.2 μL: Insufficient cleavage
Learn
- Optimal TEV concentration ~0.5-1 μL
- Removal step essential
Design
Objective:
Identify most productive clones under optimal conditions.
Hypothesis:
Direct TB induction outperforms LB→TB transfer.
Build
- Compared direct TB induction with LB→TB transfer
- Both induced with aTc
Test
- Direct TB induction: Strong production in clones A1, U1, O1, W2, F1, R1, E1
- LB→TB transfer: No overproduction
- Clone U1: Basal expression without induction (leaky promoter)
Learn
- Direct TB induction is optimal
- Promoters with background expression must be monitored
Design
Goal:
Select best clone based on protein yield.
Hypothesis:
Certain clones outperform others significantly.
Build
- Protein concentrations measured with Thermo Scientific quantification kit
- BSA standards and spectrophotometry used
Test
- Clone A1: ~150 μg/mL, highest yield
- O1, M1, U1: Much lower yields
Learn
- Clone A1 chosen as best compromise between yield and purity for functional studies
Design
Goal:
Improve protein band visibility on SDS-Page gels by adjusting the amount of culture material loaded per well.
Hypothesis:
Increasing the loading from 0.02 ODu to 0.4 ODu per well will enhance protein detection without saturating the gel.
Rationale:
The first trials at 0.02 ODu resulted in gels where no bands were visible, possibly due to insufficient protein amounts.
Build
- Cultures: MG1655 WT, MG1655 A1, and MG1655 U1 were grown and monitored by OD.
- Induction performed at OD ~0.5 for all strains.
- SDS-PAGE gels prepared with 0.4 ODu per well (20× higher than previous loading).
- Computations of cracking buffer volumes were performed to normalize protein loading across samples.
- Equipment: spectrophotometer for OD monitoring, SDS-PAGE system for gel electrophoresis, Coomassie blue staining.
Test
- Growth curves showed consistent profiles for WT, A1, and U1, with induction around OD 0.5.
- SDS-PAGE gels now revealed clear overproduction bands:
- Enzyme bands visible from ~5 h post-induction onward.
- Protein quantities appeared to peak after ~7 h of growth.
- No visible overproduction in WT, validating its role as a negative control.
- Compared to the first iteration, enzyme expression was now detectable and quantifiable on gels.
Learn
- Loading 0.4 ODu per well provided sufficient protein for visualization, solving the detection issue from Iteration 1.
- Induction around OD 0.5 led to clear enzyme overproduction without altering bacterial growth, suggesting non-toxic expression or secretion via vesicles.
- Recommendation: Maintain induction at OD ~0.5 and load at least 0.4 ODu per well for reliable protein visualization on SDS-PAGE.
Iteration 16 – Optimization of Sonication Protocol - Change lysis method to purify a higher enzyme quantity
Design
Goal:
Identify the best sonication parameters to lyse bacterial cells efficiently while preserving enzyme integrity.
Hypothesis:
Sonication settings (power, duration, duty cycle, pellet resuspension volume) impact lysis efficiency and enzyme quality.
Build
- Cultures: A1, pellets resuspended in PBS.
- Conditions tested:
- Test 1: 50 mL pellet in 5 mL PBS, 10 cycles × 10 pulses, power 8, duty 30%.
- Test 2: 250 mL pellet in 25 mL PBS, 10 cycles × 10 pulses, power 8, duty 30%.
- Test 3: 50 mL pellet in 5 mL PBS, 10 cycles × 10 s continuous, power 10, duty 30%.
- Test 4: 50 mL pellet in 5 mL PBS, 10 cycles × 10 pulses, power 4, duty 30%.
- Test 5: 50 mL pellet in 5 mL PBS, 10 cycles × 10 pulses, power 8, duty 50%.
- Test 6: 100 mL pellet in 10 mL PBS, 10 cycles × 10 pulses, power 8, duty 30%.
- Test 7: 100 mL pellet in 5 mL PBS, 10 cycles × 10 pulses, power 8, duty 30%.
- Test 8: 100 mL pellet in 10 mL PBS, no sonication.
- Test 9: 50 mL pellet in 5 mL PBS + Cell Lytic.
- Pause of 30 sec between cycles.
- SDS-Page used for protein visualization.
Test
- Test 1 vs Test 2 vs Test 4 vs Test 5: No major differences; dilution volume and reduced power had minimal impact.
- Test 3 (continuous power 10): Lower total protein, but enzyme still produced. However, overheating likely caused denaturation.
- Test 6 vs Test 8 (same culture, with vs without sonication): No differences visible, suggesting some proteins were already soluble.
- Test 9 (Cell Lytic): Comparable to sonication, but no significant advantage observed.
Learn
- Sonication is effective but very time-consuming; fewer cycles (5 instead of 10) will save time without reducing yield.
- Continuous high-power sonication risks enzyme denaturation (overheating).
- Cell Lytic provides no major advantage over sonication.
- Recommendation: standardize to 250 mL pellet in 25 mL PBS, 5 cycles of 10 pulses at power 4, duty 30% for balance of efficiency and practicality.
Design
Goal:
Compare protein extraction efficiency between different lysis methods: sonication, Cell Lytic and lysozyme-only.
Hypothesis:
Sonication would release more protein than Cell Lytic and lysozyme-only.
Rationale:
Since Cell Lytic is very expensive, we are looking for another solution to lyse bacteria.
Build
- Cultures: WT, A1, U1, and O1 from several batches.
- Lysis conditions tested across experiments:
- Sonication: multiple powers, cycles, and pellet-to-buffer ratios.
- Cell Lytic: applied to pellets as per supplier protocol.
- Lysozyme-only control.
- Samples analyzed by SDS-Page (pellet, supernatant fractions, purified fractions).
Test
- Test 1 (Sonication tests with Cell Lytic control, 19/09): Cell Lytic yielded a protein profile almost identical to sonicated lysates, but no advantage was seen.
- Test 2 (Direct purification, 22/09): Two pellets processed in parallel (one with sonication, one with Cell Lytic). Gel comparison suggested higher yield with sonication.
- Test 3 (Lysis protocol comparison, 24/09–26/09):
- Pellets lysed with: (a) sonication, (b) Cell Lytic, (c) lysozyme-only
- Result: Sonication yielded more soluble protein in the supernatant than lysozyme-only and Cell Lytic.
Learn
- Sonication reproducibly increased protein yield in the soluble fraction.
- Recommendation: Adopt sonication as the sole lysis method for future purifications.
Design
Goal:
Quantify protein concentrations after purification using BSA assays.
Hypothesis:
A standard BSA calibration curve would allow accurate quantification of purified enzymes.
Rationale:
To evaluate yields, we tried both spectrophotometric absorbance and SDS-Page gel-based BSA quantification, each performed twice.
Build
- Protocols tested:
- Absorbance (ThermoScientific kit): Calibration at 560 nm, samples compared to BSA standards.
- Gel-based BSA assay: BSA standards loaded on gel, band intensities quantified with ImageJ.
- Both methods were applied on multiple purified samples from different dates (24/09, 26/09, 01/10).
Test
- Test 1 (Absorbance, first run):
- Problems: incubation extended to 45 min instead of recommended 30 min → values inconsistent, especially at low concentrations.
- Controls missing for some sets → hard to calculate true protein concentration.
- WT unexpectedly gave high absorbance despite no visible proteins on gel (interference suspected).
- Test 2 (Absorbance, second run with better controls):
- Results inconsistent with gels, with controls giving a higher concentration than the samples.
- Test 3 (Gel-based BSA assay, first run):
- Clearer visualization, enzyme bands compared to BSA standards.
- Gave higher estimated concentrations than absorbance assay, possibly overestimation.
- Test 4 (Gel-based BSA assay, second run with ImageJ quantification):
- Produced quantifiable ranges (e.g., A1.1 ~678 µg/mL, O1.3 ~1242 µg/mL, U1.1 ~1223 µg/mL).
- However, discrepancies remained with absorbance-based values.
Learn
- Absorbance assay is sensitive to timing and interfering compounds; it produced inconsistent results in our hands.
- Gel-based BSA quantification gave results more consistent with visible SDS-Page bands, though less precise.
- Recommendation: Rely primarily on BSA gel quantification with ImageJ.
Design
Goal:
Monitor defluorination activity of enzymes via pH-dependent color change.
Hypothesis:
RPA1163 (A1) should catalyze fluoroacetate defluorination, releasing protons and lowering pH, observable by phenol red or bromothymol blue.
Build
- Test 1 – Inductor Petri Dishes: Fluoroacetate (0, 10, 100 mM) added to MG1655 WT, A1, O1 cultures.
- Test 2 – Liquid Culture Assay: Cultures supplemented with fluoroacetate (0, 2, 4 mM).
- Test 3 – TECAN Assay with Clear Lysates & Purified Enzymes: Phenol red assay, absorbance measured at 540 nm. Conditions: 0, 5, 10 mM fluoroacetate.
Test
- Test 1: Yellow halos appeared at high FA (100 mM), but inconclusive (medium acidification). Overnight, all plates turned purple due to bacterial alkalinization.
- Test 2: No visible differences after short incubation; overnight cultures also turned basic.
- Test 3:
- Clear lysates: No activity detectable (low enzyme concentrations).
- Purified enzymes: A1 showed acidification at 5 & 10 mM FA.
- O1 behaved like WT (no activity on FA, consistent with design).
- Enzymatic activity of A1 quantified as ~0.045–0.055 µmol/min/mg, consistent with literature (Khusnutdinova et al.).
Learn
- Plate and liquid culture assays confounded by bacterial metabolism (alkalinization). A fluoride probe test would be more relevant.
- TECAN assay with purified enzymes was successful and reproducible.
- A1 enzyme confirmed active on fluoroacetate with activity matching published data.
- Recommendation: Use TECAN-based colorimetric assay as standard for activity measurements; extend to PFAS substrates.
Engineering Cycles for Analytical Method Development
Goal: Establish a reliable HPLC-UV analytical method to detect and quantify N-[(4-chlorophenyl)methyl]-2,2,2-trifluoroacetamide (cycled TFA) and 4-chlorobenzylamine, enabling verification of the enzymatic ring addition to TFA catalyzed by SpL lipase.
Iteration 1.1
Design
Based on literature precedent from Zeng et al. (2018), we designed our initial HPLC protocol to separate and identify the components of the enzymatic reaction: TFA (substrate), 4-chlorobenzylamine (reagent), and cycled TFA (product). Since TFA itself lacks a UV-absorbing chromophore, we focused on detecting the amine reagent and the amide product.
Initial Protocol Design:
- Column: Hypersil™ ODS-2 C18 (250 mm × 4.6 mm, 5 µm)
- Mobile Phase:
- Solvent A: Methanol (without additives)
- Solvent B: Water (without additives)
- Gradient Profile (30 minutes):
- 0 min: 99% B, 1% A
- 20 min: 1% B, 99% A
- 24 min: 1% B, 99% A
- 26 min: 99% B, 1% A
- 30 min: 99% B, 1% A
- Detection: UV at 230 nm (optimal for aromatic compounds) and 210 nm and 250 nm (we began with three wavelengths to determine which one is the most relevant to the further steps)
- Injection Volume: 5–10 µL
- Test Compounds: Toluene (as column verification standard), 4-chlorobenzylamine
The 30-minute gradient was designed to cover a wide polarity range, ensuring separation of both highly polar and less polar compounds. We chose to start with pure methanol/water without buffer additives to establish baseline chromatographic behavior.
Build
We prepared test solutions:
- Toluene: Diluted to 10% and 1% (v/v) in methanol
- 4-Chlorobenzylamine: 10 mM solution in 50:50 water/methanol mixture
All analyses were performed on an Agilent 1260 Infinity HPLC system. Initial system pressure ranged from 108-117 bar depending on solvent composition.
Test
Toluene Results:
- First injection (10% solution): tr = 15.020 min
- Second injection (1% solution): tr = 14.970 min
- Excellent reproducibility confirmed proper column function and gradient stability
4-Chlorobenzylamine Results:
- Retention time: ~9 minutes
- Critical observation: Two high-intensity peaks appeared in concentrated samples, suggesting either:
- Molecular/ionic association
- Concentration-dependent effect on the column
- Presence of impurities
This double-peak phenomenon created ambiguity in peak identification and could interfere with accurate quantification.
Learn
Several important lessons emerged from these initial tests:
- Peak multiplicity issue: The appearance of double peaks for 4-chlorobenzylamine at high concentrations indicated that our method needed optimization to prevent this artifact. The phenomenon disappeared upon dilution, confirming it was concentration-dependent rather than a true impurity.
- Buffer requirement: For basic compounds like 4-chlorobenzylamine, the absence of buffer in the mobile phase may contribute to peak tailing or splitting due to ionization state variations.
- Gradient length: The 30-minute gradient, while comprehensive, was unnecessarily long for our analytical requirements and would reduce throughput.
- Need for buffer: We decided to add 20 mM ammonium acetate to both solvents to improve peak shape and resolution for ionizable compounds.
Iteration 1.2
Design
Based on learnings from Iteration 1.1, we redesigned the HPLC method with the following improvements:
Optimized Protocol:
- Column: Same (Hypersil™ ODS-2 C18)
- Mobile Phase with Buffer:
- Solvent A: Methanol + 20 mM ammonium acetate
- Solvent B: Water + 20 mM ammonium acetate
- Buffer preparation: 0.776–0.782 g ammonium acetate per liter
- Optimized Gradient (20 minutes):
- 0 min: 80% B, 20% A
- 10 min: 10% B, 90% A
- 15 min: 10% B, 90% A
- 18 min: 80% B, 20% A
- 20 min: 80% B, 20% A
- Detection: UV at 230 nm, 210 nm, and 250 nm (finally we focused on 210 nm because toluene does not absorb at 230 nm and it is the internal standard we used to plot the calibration curve of cycled TFA)
- Sample preparation strategy: Dilute concentrated samples 5-fold to avoid peak-splitting artifacts
The new gradient starts at 80% water (more realistic initial conditions for polar analytes) and completes the separation in 20 minutes instead of 30, improving efficiency without sacrificing resolution.
Build
We prepared fresh solutions with the buffered mobile phase system:
- 4-Chlorobenzylamine: 10 mM stock solution, analyzed both undiluted and diluted 5×
- Cycled TFA (from chemical synthesis): 10 mM solution, diluted 5×
- Co-injection mixture: Equimolar solution of both compounds
We also tested toluene again with the buffered system to verify any changes in retention behavior due to the ammonium acetate addition.
Test
Toluene with buffered solvents:
- tr = 15.225 min (slight increase compared to non-buffered system)
- Single, sharp peak maintained
4-Chlorobenzylamine (optimized conditions):
- tr = 8.026 min (5× diluted solution) – single, well-defined peak
- tr = 7.601 min (in mixture with cycled TFA)
- Critical success: The double-peak artifact disappeared with dilution and buffered mobile phase
Cycled TFA:
- tr = 9.721 min (5 µL injection)
- tr = 9.170 min (10 µL injection)
- tr = 9.2 min (in mixture with 4-chlorobenzylamine)
- Well-defined peak with sufficient intensity at 230 nm
Co-injection results:
- 4-Chlorobenzylamine: tr = 7.601 min
- Cycled TFA: tr = 9.2 min
- Excellent baseline separation with ~1.6 min difference in retention times
- No peak overlap, confirming the method can reliably distinguish reagent from product
Learn
The optimized method successfully addressed all issues identified in Iteration 1.1:
- Buffer benefits confirmed: The addition of 20 mM ammonium acetate provided:
- Better peak shape for ionizable compounds
- Improved resolution
- Elimination of concentration-dependent peak splitting when combined with appropriate dilution
- Gradient optimization validated: The 20-minute gradient proved sufficiently effective for separating all target compounds while reducing analysis time by 33%. This improvement is critical for processing multiple samples in degradation kinetics studies.
- Sample preparation protocol established: 5-fold dilution of 10 mM stock solutions provides optimal signal intensity while avoiding column overloading effects.
- Method robustness: The reproducibility of retention times across multiple injections (±0.1 min for cycled TFA) demonstrated method stability suitable for quantitative analysis.
- Ready for calibration: With confirmed separation and reproducible retention times, the method was now suitable for constructing calibration curves with an internal standard (toluene).
Next steps defined:
- Construct calibration curve using toluene as internal standard
- Validate linearity range (0–10 mM)
- Apply method to analyze enzymatic reaction samples
- Experimental protocols were developed under the supervision of Pr. Chambert (INSA Lyon) and Dr. Karine Faure (ISA), who provided the initial gradient and solvent protocol and guided the methodological adjustments throughout the two-week development period.
