The Totally [Fe]rocious project aims to develop an alternative to antibiotics in order to deal with antibiotic-resistant Klebsiella pneumoniae strains. We propose a Trojan horse approach using aerobactin, a siderophore used by pathogenic K. pneumoniae [1], coupled to a gold nanoparticle. The conjugate, once excited with a laser, should be able to induce the death of cells [2]. In the lab, our goal was to express the synthesis pathway of aerobactin in Escherichia coli. This pathway, a nonribosomal peptide synthetase-independent siderophore (NIS) synthesis pathway [3], is composed of four enzymes IucA-D [4]. We ordered the sequence for all four genes in the form of gBlocks from IDT. The gBlocks were assembled in the vector by Gibson assembly and transformed in E. coli. The siderophore would then be produced and purified using a chromatography-based method and the final product would be quantified by a colorimetric method. The purity would be assessed by mass spectroscopy. Once we have obtained aerobactin in a sufficient amount and at an acceptable level of purity, we would proceed with the coupling of the gold nanoparticles to the siderophore. Ultimately, we would test the conjugate between the siderophore and the nanoparticle to see if, once excited with a laser, it can cause cell death of pathogenic K. pneumoniae strains.

Dry-lab – GEM

Our modelling and experimental results together provide a comprehensive understanding of how environmental and genetic factors influence citrate and siderophore production in Escherichia coli. Using genome-scale metabolic models and constraint-based simulations, we observed clear trade-offs between growth and metabolite secretion. While the wild-type strain achieved high growth rates under neutral conditions, enhanced citrate export occurred primarily in acidic and phosphate-rich environments, and siderophore production remained remarkably stable across most tested conditions. These findings reveal how specific nutrient and pH conditions can be exploited to guide metabolite yields without necessarily improving biomass formation[5].

These results have important implications for our project goals. They confirm that citrate and siderophore pathways can be selectively optimized depending on the environmental context, and that siderophore biosynthesis is inherently resilient—a desirable feature for targeted delivery strategies such as our siderophore–gold nanoparticle conjugate. Moreover, the ICDHyr knockout simulations revealed that siderophore synthesis can persist even in the absence of growth, opening possibilities for controlled, non-proliferative production systems.

Not everything went as expected. Some simulation runs initially produced unrealistic flux distributions due to boundary condition errors or missing reaction constraints, forcing us to review model definitions and verify reaction reversibility. Through this process, we learned the importance of model validation, parameter sensitivity testing, and cross-checking with experimental data. These challenges strengthened our understanding of metabolic modelling and improved the reliability of our results.

Overall, these results emphasize the value of combining computational predictions with experimental insight. They also highlight the necessity of transparency, reproducibility, and scientific honesty in modelling—acknowledging uncertainty is as important as presenting strong results. Our work not only supports the feasibility of siderophore-based delivery strategies but also lays the foundation for more predictive and sustainable approaches to metabolic engineering.

Wet-lab - gBlock Amplification

Our first step was to obtain a sufficient concentration of each gBlock to proceed with the Gibson assembly into the extracted and digested pBluescript SK- plasmid. We performed standard PCR protocols following the DNA fragment manufacturer’s recommendations. The results of these PCR amplifications are shown in Figure 1.

The results were satisfactory enough for our needs in the cases of gBlocks 1 and 2. They were both obtained in a sufficient amount to proceed with the Gibson assembly. They were purified on magnetic beads to rid the mixture of the PCR reaction components and the non-specific products. These two were then ready for the Gibson assembly. For gBlock 3, there was too much non-specific amplification and not enough of our goal fragment (around 2,000 bp), meaning that we would not be able to get enough product after purification.

We tweaked the PCR protocol for gBlock 3 trying Touchdown (TD) and Inverse-Touchdown (iTD) methods as well as removing the PCR Enhancer from the master mix (see Figure 2). Touchdown PCR consists in hybridizing the primers with the template at a higher temperature during the first cycles and gradually lowering it to Tm - 5°C. This technically leads to more specific amplification in the first few cycles. Inverse-Touchdown PCR, which is the opposite of TD, begins the PCR cycles at a lower hybridizing temperature to make sure the primers hybridize to the template and rising to Tm-5°C. This leads to less specific but easier primer hybridization. The aim of this method was to make sure to obtain an amplicon after the first few cycles. In the end, we found that the iTD protocol without enhancer yielded the best results, so we did multiple reactions with that protocol and pooled the results together after a gel-based purification (we still had too much non-specific amplification to do a column purification).

Wet-lab - Gibson Assembly & Transformation

With all gBlocks in sufficient concentration, we proceeded with the Gibson assembly into pBluescript SK-, followed by a chemical transformation into competent NEB5α E. coli cells. The transformed cells were plated on XGAL medium for white/blue screening. A few different positive mutants were selected and they were incubated overnight. We then isolated their plasmid and digested each of them using BamHI and EcoRI. When we did the gel electrophoresis to verify the results, the fragments did not correspond with the ones we were expecting (Figure 3). However, we had multiple repeated digestion patterns indicating that we were able to obtain a certain product in a repeatable fashion.

In our interpretation, this means that during the Gibson assembly, a part of our fragments was successfully integrated to the plasmid in a repeatable fashion. However, somewhere in our gBlock sequences was an element that interfered with the designed assembly and led to an unwanted final product that does not contain the full sequence of all three gBlocks. As we also had trouble with the amplification of gBlock 3, we suspect that that fragment is our problem child. To get some insight into what is happening, we decided to send one of our repeated digestions to Flow Genomics for DNA sequencing. The plasmid from clone 4 was chosen for sequencing.

We suspected that it might be ambitious to attempt a Gibson assembly of three 2kb gBlocks at once on an unestablished protocol on limited time, so we also attempted a different assembly method, opting for a PCR Stitch. However, we only had time for one attempt with this method and the stitch of gBlocks 1 and 2 yielded very little product, and not at our desired length of 4 kb (see Figure 4). The protocol we followed for this stitch is the exact same as the protocol for the amplification of gBlocks 1 and 2, except we used the forward primer of gBlock 1 and the reverse primer of gBlock 2.

Wet-lab - DNA sequencing

Once the sequencing results came in, we were not shocked to see that the sequenced plasmid was not the expected construction containing all three gBlocks. However, the sequencing did give us some insight on what could have gone wrong. The plasmid extracted from clone 4 comprised 5514bp whereas the expected construction was around 8750bp long. Upon closer inspection, we realized that the sequenced plasmid contained the beginning of gBlock 2 and the beginning of gBlock 3 (see interactive map below). The gBlock 1 and the beginning of gBlock 2 were inserted successfully, but something went wrong and the beginning of gBlock 3 was inserted in the middle of gBlock 2. Since the PCR products used in the Gibson Assembly were all of the right size, the assembly itself was probably not the problem. What could have happened is a recombination event after the transformation of the assembly. The presence of repeated sequences long enough to induce homologous recombination could explain the loss of gBlock 2 and gBlock 3’s sequences.

Future Outlook

These results represent what we were able to obtain in the Wet-Lab for this year’s iGEM cycle. The team’s objective was to tackle the antibiotic-resistant Klebsiella pneumoniae strain using a Trojan horse approach. To further develop the analogy, aerobactin was our wooden horse and gold nanoparticles the soldiers hidden inside it. Our unsuccessful attempts at building an expression vector for the enzymes responsible for aerobactin’s synthesis halted us in our steps. The expression of the enzymes and purification of the siderophore represented the first steps in the Totally [Fe]rocious project. Although we did not manage to meet all our objectives, we managed to gain insight on what went wrong and what can be changed in order to go forward with the next steps for this project. As we mentioned earlier, maintaining the expression vector in E. coli is most likely where we tripped. Because we were not able to express the enzymes to produce the siderophore, we could not try the purification protocol nor the nanoparticle coupling method. We remain hopeful that we will eventually manage to produce aerobactin and that we will get to try to couple the nanoparticles to the siderophore. Ultimately, the goal would be to test to see whether or not the conjugate between the siderophore and the nanoparticle can act as an alternative to antibiotics in the fight against multiresistant K. pneumoniae.

In terms of our next steps, the first thing we will have to solve is the expression system. The Gibson assembly method did not allow us to build a plasmid containing the genes coding for the enzymes required for in vivo synthesis of aerobactin. Since the assembly of all three gBlocks in the vector at once was complicated, we think that assembling them two by two is a more promising way to deal with our issue than redesigning a Gibson assembly from scratch. This is why a PCR stitching method seems like a viable option to us. We tried this method with our current set of primers but it did not work. The use of this method would require us to redesign primers for this specific use. We would assemble the gBlocks two by two, eventually leading to a final assembly step where we would have all three gBlocks lined up in a single fragment and the backbone. This final assembly step can be done by Gibson assembly. The assembly of two fragments instead of four increases our odds of success. Another possible avenue would be to assemble the gBlocks in the backbone by utilizing the recombination ability of Saccharomyces cerevisiae. Although this method could work, it would require us to utilize a backbone compatible with the yeast. Once the assembly is done, we would extract the plasmid, amplify the region containing the genes iucA-D to clone it in a vector compatible with E. coli and transform it into E. coli. If our problem really is related to recombination, we would try to transform the assembly in a cloning strain that has limited recombination activity, such as Stbl from Thermo Fischer or NEB Stable from New England Biolab [6,7].