Engineering Success

Design

During the initial planning process of our project, we decided to use an antimicrobial peptide called Sushi S1 which derives from the lipopolysaccharide (LPS)-binding region of Factor C of horseshoe crabs. Our literature research showed that Sushi S1 has one of the lowest minimum inhibitory concentrations (MIC) and minimum bactericidal concentrations (MBC) against gram-negative bacteria like E. coli in general [1] and our target pathogen Pseudomonas aeruginosa in particular [2]. By choosing a highly effective AMP, we want to make sure that as much bacteria as possible are killed upon delivery of the plasmid and induction of peptide expression.

In previous publications the functionality of Sushi S1 was only tested by applying the synthesized peptide to the bacteria [3]. It was shown that the peptide binds with high affinity to Lipopolysaccharides (LPS) on the surface of the outer membrane of gram-negative bacteria [4,5]. Since we have not found any papers trying to express Sushi S1 inside the targeted bacterium, we decided to first test what cellular localization of the peptide has the biggest impact on the bacterial growth.

We designed three versions of our plasmid allowing us to test the effect of Sushi S1 at three different locations: in the cytoplasm, the periplasm or the extracellular space. In the framework of iGEM we first focused on experiments with the gram-negative bacterium E. coli.

Build

Using the pET-22b(+) plasmid backbone we cloned three variants of Sushi S1:

• Cytoplasmic: pelB signal sequence was deleted from plasmid backbone
• Periplasmic: pelB signal sequence [6]
• Extracellular: insertion of heat-stable toxin II (HSTII) secretion sequence [7]

For cloning purposes and plasmid amplification we used E. coli Top 10. The nucleotide sequence of Sushi S1 was ordered as a synthesized G-block fragment from IDT. Our cloning strategy for the various plasmids consisted of a combination of restriction digest, Gibson assembly and Site Directed Mutagenesis (get more details about the cloning process in our Notebook).

Test

We performed growth curve experiments in E. coli BL21(DE3) and E. coli BL21(DE3)-pRARE2-LysS and measured OD₆₀₀ over time after induction of Sushi S1 expression (read more about our growth curve experiments on the Results page). We compared the three localizations with the growth of two different controls (empty pET-22b(+) plasmid and non-induced cultures).

Our initial findings revealed:

• Sushi S1 expression inhibited bacterial growth for all three localization sequences at least to some degree.
• The extracellular version HSTII‑Sushi S1 showed the strongest growth impairment.
• Induction of the empty pET-22b(+) plasmid also impaired growth. Therefore, a new control was required.

To address the issue of a new control and to verify our results we switched to pET-22b(+) plasmids containing the fluorescent protein mCherry as negative control. At this point we also changed our medium from LB‑medium to Mueller-Hinton broth (MH medium), as further literature research showed that it is recommended for testing antimicrobial susceptibility [8].

Although the results for HSTII-Sushi S1 were reproducible and bacterial growth was significantly reduced compared to mCherry expression, we observed that OD₆₀₀ values started to increase again after approximately 6 hours.

To test if we could eliminate this effect, we repeated the experiment in E. coli BL21(DE3) with pRARE2 LysS plasmid. In this strain, Sushi S1 expression inhibited growth for 24 h, but the mCherry control also showed strong growth inhibition.

Learn

From our expression experiments with Sushi S1 in E. coli we can draw two main conclusions:

• Effectiveness of AMP delivery concept
  Our chosen AMP Sushi S1 reduced bacterial growth at a significant level when produced within the target organism. Although the bacteria started growing again after about 6 hours, we were able to show that the concept of delivering a plasmid encoding a potent AMP to the bacteria works.

• Optimal peptide localization (signal peptide selection)
  The secreted form of Sushi S1 (HSTII-Sushi S1) showed the strongest impact on bacterial growth. This aligns with literature data, showing that Sushi S1 binds to Lipopolysaccharides (LPS) on outer bacterial membranes.

Based on these findings, we decided to focus on HSTII-Sushi S1 in all following expression experiments in E. coli.

Design

With the intention to test if growth inhibition can be observed for a longer period of time we replaced Sushi S1 with another antimicrobial peptide.

In the scope of our integrated human practice, we had the chance to talk to Prof. William Wimley, who works on synthetically developing highly potent AMPs (more about our meeting with Prof. Wimley on the Human Practices page). He drew our attention to his most recent publication in which he characterized his new and very effective AMP D-CONGA Q7 [9]. Despite the fact that D-CONGA Q7 consists only of the D-enantiomers of amino acids, he encouraged us to try the peptide’s effectiveness when expressed in the targeted bacteria.

Based on our Sushi S1 results and discussion with Prof. William Wimley, we decided to test the peptide D-CONGA Q7 for our purpose. Since the expression of Conga by E. coli would produce the L-enantiomer of the peptide (L-CONGA), we additionally wanted to test the synthesized version of the peptide comparing its potency to D-CONGA Q7.

For the expression experiments we focused on two localizations: Cytoplasmic and extracellular (HSTII-Conga). However, our primary focus was on the extracellular version, given the results from our Sushi S1 experiments.

To be able to compare the effect of the two AMPs Sushi S1 and Conga, the experimental setup for both peptides was kept identical.

Build

Prof. Wimley kindly provided us with a small sample of his AMP D-CONGA Q7 to compare the effectiveness of both enantiomers of the peptide. L-CONGA was generously synthesized by the group of Prof. Maja Köhn and donated to our team.

For testing the expression of Conga in bacterial cultures we modified existing Sushi S1 plasmids using Site Directed Mutagenesis to insert the synthesized Conga sequence to generate two new constructs:

• HSTII-Conga for extracellular localization
• Conga without signal sequence for cytoplasmic localization

To analyze the effect of peptide expression E. coli BL21(DE3) standard strain and pRARE2 LysS strain were transformed with finished plasmids.

Test

Liquid bacterial cultures of E. coli BL21 and Pseudomonas fluorescens were incubated with synthesized Conga peptides and plated on MH plates to examine the impact of the AMP enantiomers on bacterial growth. After incubation with D-CONGA Q7 neither E. coli nor P. fluorescens showed any growth on plates. However, L-CONGA seemed to only impair growth of E. coli to a smaller degree than the D-enantiomer.

To analyze the impact of expressed Conga we performed growth curve experiments similar to Sushi S1. After the induction we measured OD₆₀₀ over time in both E. coli BL21(DE3) standard strain and pRARE LysS containing strain. We used mCherry expression and uninduced cultures as control.

Similar to the effect that we observed for Sushi S1, Conga inhibited growth for roughly 6 hours. Unfortunately our results of the pRARE LysS strain were inconclusive due to control issues (cultures of mCherry expression showed similar growth inhibition). The comparison between the intracellular and extracellular localization of Conga revealed that the extracellular HSTII-Conga showed stronger inhibition than the cytoplasmic version.

Learn

After testing the efficacy of two highly potent antimicrobial peptides in E. coli, we can conclude that the principle of expressing the AMPs within the targeted pathogen is promising.

• We observed that the L-enantiomer of Conga effectively inhibited growth of E. coli.
• Sushi S1 and Conga appeared to have an almost identical effect on bacterial growth. They acted as inhibitors for 6 hours, but lost their effect after a longer period of time.
• The extracellular localization was more effective for both AMPs.

Due to time constraints, we could only test these two peptides in two different bacterial strains. Further research is needed and could include:

• Identification of source of the control issues. During our experiments using the E. coli BL21(DE3) strain containing the pRARE2 LysS plasmid, we observed that the synthesis of our control protein mCherry significantly reduced bacterial growth. We suspect that mCherry secretion could cause problems that have negative effects on the viability of the cells. However, we did not have the chance to perform any tests to support this assumption or try the effects of another control protein.
• Testing AMPs with different mechanisms, e.g. peptides that attack the bacterium from the inside.
• Exploring alternative plasmid backbones to extend the effectiveness of the AMPs.

Design

Lipid-based nanocarriers are vesicular structures that can be synthesized in the lab and have been extensively researched and adapted for drug and gene delivery applications. Their properties, such as lipid composition, size, and surface modifications, can be tailored to enhance encapsulation, stability, and target specificity. There are several sub-categories of lipid nanocarriers, but we focused particularly on liposomes and lipid nanoparticles (LNPs) (see Figure 5). Liposomes (50 - 500 nm) have one or more lipid bilayers, and form a spherical structure to encapsulate the cargo in an aqueous solution. LNPs (100 - 400 nm), on the other hand, have a lipid monolayer and their payload is enveloped by reverse micelles. Their interior does not necessarily have to be aqueous.

Our liposomes and LNPs had to be able to encapsulate DNA and fuse with biological membranes in order to release the payload. In particular, we wanted to use cationic lipids as building blocks for the vesicles, as these would ideally ensure enhanced encapsulation of the negatively charged plasmid DNA and optimized fusion with the negatively charged bacterial membranes. Through research and consultation with experts (see our Integrated Human Practices page), we eventually decided on 2-dioleoyl-3-trimethylammonium propane (DOTAP), a cationic synthetic lipid that is commonly used for lipid-based carrier formulations for the delivery of genetic material. In addition to this, we also decided on using the neutral helper lipid 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), which is often used in combination with DOTAP. However, as DOTAP and DOPE were not yet available to us at the start of the project, we first started our work with 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), a phosphatidylcholine that is commonly used as a model lipid for biophysical experiments, as it mimics the mammalian phospholipid composition [10].

Build

We produced liposomes using a method called PVA swelling, followed by extrusion for size control. LNPs were prepared using a simple pipetting-mixing protocol or microfluidic mixing (MFM). The carriers were then tested and characterized in regards to size distribution, lipid composition and plasmid encapsulation efficiency.

Test

After developing the lipid-based nanocarriers, we conducted various tests to evaluate their effectiveness, focusing on size distribution, stability, and plasmid encapsulation efficiency.

Our test phase focused on three critical aspects of our lipid-based nanocarriers: size distribution, stability, and plasmid encapsulation efficiency. The results were both promising and insightful:

We successfully produced Giant Unilamellar Vesicles (GUVs) using PVA swelling and tested different lipid compositions. DOTAP/DOPE ratios of 52:48 and 26:74 showed larger diameters (up to 100 µm) and an enhanced stability compared to other lipid compositions (Figure 6).

GUVs were downsized to Large Unilamellar Vesicles (LUVs) of approximately 200 nm using extrusion (Figure 7). We selected DOTAP/DOPE (52:48) for further experiments due to optimal size and stability.

We successfully encapsulated plasmids into GUVs using PVA swelling with best encapsulation in POPC liposomes.

Plasmid-loaded pLNPs were produced using both simple pipetting-mixing and MFM, with average diameters of 150 nm and 170 nm, respectively.

During the process of pLNP production, we developed a novel, cost-effective Midori Green assay for quantification of DNA-encapsulation. Measured results closely aligned with the established PicoGreen assay, validating our new method (Details on our Measurement page).

Moreover, we could show that there is a positive correlation between DOTAP molar ratio and encapsulation efficiency (Figure 10) using PicoGreen assay. This finding highlights the importance of balancing lipid composition and N/P ratio for optimal performance (Figure 10) using PicoGreen assay. This finding highlights the importance of balancing lipid composition and N/P ratio for optimal performance.

It is critical to optimize both the lipid composition and the N/P ratio to achieve maximal encapsulation efficiency while preserving fusion capabilities and minimizing cytotoxicity, which will be discussed further in Cycle 2.

Overall, these tests confirmed the viability of both liposomes and LNPs as carriers, with significant encapsulation efficiency and stability, particularly with optimized lipid compositions and production protocols.

Learn

During our experiments, it became evident that LNPs are superior in terms of production efficiency compared to liposomes. While liposomes require labor-intensive processes such as PVA swelling and subsequent extrusion for size control, LNPs can be synthesized using simple mixing techniques, resulting in a significantly higher yield of particles.

As a result, we shifted our focus to LNPs, exploring different lipid compositions to optimize size distribution and encapsulation efficiency. Our findings indicated that lipid composition plays a crucial role in determining the performance of the nanocarriers:

LNPs with a higher molar ratio of DOTAP in the lipid composition exhibited increased encapsulation efficiencies, likely due to the cationic nature of DOTAP, which facilitates the interaction with the negatively charged plasmid DNA. POPC, appeared to potentially enhance the stability of the nanocarriers. Its inclusion in the lipid formulation contributed to more consistent size distribution and reduced aggregation over time.

Through further testing, we found that a combination of DOTAP, DOPE, and POPC in a molar ratio of 50:30:20 was an effective formulation. This mixture combined the advantages of each lipid: the high encapsulation efficiency from DOTAP, the membrane fusion properties of DOPE, and the stability imparted by POPC. This optimized composition was subsequently used in our later assays, to achieve robust and reliable results. However, while size, stability, and encapsulation efficiency are critical parameters, they are not the only considerations for the intended therapeutic application. An equally important factor is the fusion capability of the nanocarriers:

• Targeted Fusion: Ensuring effective fusion with the target bacterium, P. aeruginosa, is essential for delivering the therapeutic payload to the infection site.
• Minimizing Off-Target Effects: Avoiding unwanted fusion with human lung cells, which are in close proximity to the bacteria, is crucial to prevent potential cytotoxicity and ensure the safety of the delivery system.

This insight led us to the next phase of our project, where we will investigate the impact of different lipid compositions on the fusion behavior of LNPs, aiming to fine-tune the formulation for optimal therapeutic efficacy.

Design

The goal of Cycle 2 is to optimize our LNPs to selectively fuse with P. aeruginosa while avoiding fusion with human lung cells. This targeted fusion is essential for delivering therapeutic payloads directly to pathogens in the lungs, minimizing off-target effects and cytotoxicity to human cells. The LNPs should demonstrate high fusion efficiency with P. aeruginosa while exhibiting low interaction and fusion with human lung cells, and should not cause significant cytotoxicity at therapeutically relevant concentrations. To achieve this selective fusion, we employed a targeting mechanism inspired by the interaction of the bacterial lectin LecA on P. aeruginosa with the glycosphingolipid Gb3 on mammalian cells. Including Gb3 in the LNP formulations could enhance binding to bacteria that express such lectins, allowing selective targeting.

Build

The LNPs were formulated from DOTAP/DOPE/POPC in a molar ratio of 50:30:20, based on findings from Cycle 1. Gb3 was integrated into the lipid composition to enhance specificity for P. aeruginosa. This was inspired by the binding mechanism of LecA on P. aeruginosa with Gb3, aiming to enhance selectivity for bacterial cells [11,12,13]. To test the interaction, we added Atto647N-labeled DOPE, a red-fluorescent lipid, into the lipid composition, while the bacteria were labeled with GFP. This setup allowed us to observe interactions using confocal microscopy. Additionally, we conducted cytotoxicity assays (MTT and SYTOX Green) on human lung tissue cells and performed a fusion assay to evaluate the uptake of a GFP plasmid from pLNPs into human cells.

Test

Fusion Assays with P. aeruginosa aimed to observe interaction and fusion behavior between engineered LNPs and the target bacteria using fluorescence microscopy. Co-localization of fluorescence signals (Figure 11) for LNPs (red) and bacteria (green) suggests that our Gb3-incorporated LNPs can successfully target P. aeruginosa. This provides initial validation of our design strategy. However, further quantitative analysis is needed to determine the efficiency and specificity of these interactions compared to control LNPs without Gb3.

The MTT assay evaluated the viability of A549 cells after treatment with various concentrations of liposomes, eLNPs, and pLNPs. The results indicated cytotoxicity with liposomes, while LNPs showed a mild dose-dependent cytotoxic effect only at higher concentrations​.

The SYTOX Green assay was used to assess membrane integrity of lung cells following treatment with LNPs. Only a minimal impact of eLNPs on cell membrane integrity was observed, affirming their relative biosafety compared to liposomes​.

The eGFP Fusion assay investigated the potential fusion and plasmid uptake by human lung cells. The data indicated minimal fusion of pLNPs with A549 cells, supporting the hypothesis that these LNPs are less likely to fuse with mammalian cells, which is desirable for biosafety.

Learn

Our preliminary findings indicated some interaction between the LNPs and P. aeruginosa; however, further investigation is required to validate the efficacy and specificity of the targeting mechanism. These preliminary findings were constrained by time limitations and the need for collaboration with a partner laboratory, as our team was not certified to handle the S2-level pathogen P. aeruginosa, while the partner lab possessed the necessary certification. Future work should focus on optimizing the incorporation of Gb3 into the lipid formulation to enhance selectivity for the target bacterium, while ensuring stability and minimizing off-target effects.

Our pLNPs exhibited no significant cytotoxicity at lower concentrations, but a dose-dependent increase in cytotoxicity was observed at higher concentrations, highlighting the necessity for careful dose optimization in therapeutic applications. Importantly, the minimal fusion observed with A549 cells in the eGFP assay suggests that the current pLNP formulation is effective in minimizing unintended interactions with human cells.

The lipid composition should be further optimized to minimize cytotoxic effects. Cytotoxicity can be assessed by testing different lipid formulations and determining which lipids, and at what concentrations, contribute most to the formulation’s toxicity. Subsequently, the formulation can be further modified, carefully balancing other critical properties such as fusogenicity, specificity, and encapsulation efficiency against potential adverse effects. This process requires extensive data collection. Computational modeling may be employed to reduce the data needed, identifying correlations between lipid ratios and the desired LNP properties.

Design

In the development phase of our project, we realized that our liposomal delivery system would entail significant production costs that could harm overall market adoption. To circumvent this bottleneck, we decided to look into a promising alternative for plasmid delivery: outer membrane vesicles (OMV).

OMVs are vesicles produced by Gram-negative bacteria [14], among other things encapsulating genetic material to mediate horizontal gene transfer, this fact alone makes ideal candidates for It forms the basis for our efficient and cost effective delivery solution. We have used the E. coli BL21(DE3) omp8 strain (Human Practices) , a variant of the BL21(DE3) strain with the following deletions: OmpA, OmpC, OmpF and LamB, in order to isolate a larger amount of OMVs that display a greater proportion of our outer membrane protein eCPX [15].

eCPX or the enhanced circularly permuted OmpX, derived from the native E. coli OmpX, was genetically fused to a SpyTag peptide. This construct is controlled via an inducible LacI promoter and allows us to functionalize our OMVs with multiple phage tail proteins that act as targeting ligands(see our human practices). This approach is tailored to maximize targeting efficiency while limiting specificity towards P. aeruginosa.

We hypothesized that when using an inducible promoter to express eCPX-SpyTag, OMVs produced pre-induction without the eCPX-SpyTag alongside the OMVs displaying eCPX-SpyTag would lead to a heterogenous OMV population that would be impossible to separate. We decided to circumvent this issue by replacing the inducible promoter with one of 2 constitutive promoters; LacIq (BBa_K3257003) and AmpR (BBa_K2796022).

Build

We were able to clone both promoters into our plasmid via PCR using primers containing our promoter regions. This was followed by a Gel-electrophoresis and subsequent transformation into E. coli BL21(DE3) omp8 strains. We then evaluated all 3 promoters in terms of cell viability by measuring growth curves, after which, we isolated OMVs from the aforementioned cultures to measure protein expression and functionality by performing standardized SDS Gels and Western Blots.

Test

Overexpression of outer membrane proteins typically overloads protein translocation machinery [16], leading to cell stress and sub-optimal bioproduction capabilities. However, overexpression of eCPX has been shown to not inhibit cell viability [17]. Our growth curves confirmed these findings, however the Amp promoter seemingly showed higher reduction of cell viability compared to the LacIq promoter.

However, cell viability does not give the full picture with respect to protein expression. Therefore, we performed SDS-PAGEs followed by Western Blots with OMVs that had been isolated with optimized protocols in order to maximize both OMV production and protein expression (See our Results).

Our SDS Gels and Western Blots illustrated that the LacIq promoter lead to higher levels of eCPX-SpyTag expression and functionality compared to the Amp promoter, however comparing it to the inducible Trc Promoter (induced with 100 µM IPTG) the same level of protein expression was not achieved.

Learn

We concluded this cycle with the knowledge that the LacIq constitutive promoter used in place of the inducible Trc Promoter could achieve similar levels of protein expression without the need for induction, this however wasn’t the case with what was observed with the Amp promoter.

Our results could be partly attributed to the reduced cellular impact associated with eCPX overexpression compared to other outer membrane proteins [17] and partly due to the higher expression levels associated with LacIq compared to the Amp promoter.

Design

OMVs can mediate horizontal gene transfer among other processes (See our Descriptions). Our goal was to engineer OMVs to deliver AMP coding plasmids while displaying specificity towards P.aeruginosa. This specificity could be achieved through the functionalization of phage-tail proteins on the OMV surface. In order to ascertain the efficiency of OMV-mediated plasmid delivery, numerous “non-specific fusion” experiments were carried out with OMVs that had not been functionalized with targeting ligands.

Build

OMVs encapsulating plasmids with diverse copy numbers were extracted from various donor strains using standard extraction protocol for use in fusion experiments (See our Notebook/Protocols).

Test

We conducted “non-specific” fusion experiments with differing OMV concentrations alongside dissimilar recipient strains using divergent experimental protocols[18][19][20], in an effort to measure the efficiency of OMV-mediated transformation (See more on the Results).

OMVs were screened for plasmid encapsulation via DNAse coupled PCR. (See more on the Results) prior to their use within the fusion experiment. After numerous attempts at non-specific fusion, we were not able to transform our recipient E. coli BL21(DE3) strains using OMVs encapsulating plasmids. However, an interesting phenomenon was observed when E. coli BL21(DE3) strains were incubated with OMVs produced by a strain carrying an ampicillin resistance plasmid. Bacteria would seemingly grow when incubated overnight in LB with ampicillin only to “regain” susceptibility on further inoculations in the same media.

Learn

After having observed the same result in experimental repeats, we thought about possible reasons. Our final hypothesis hinges on the periplasmic localisation of β-Lactamase[21], the enzyme responsible for degrading β-Lactams such as ampicillin. We speculate that OMVs encapsulating β-Lactamase alongside the plasmid, are able to degrade the ampicillin present, thereby allowing bacterial growth without plasmid uptake.

Design

This hypothesis was thought up by eliminating possibilities such as contaminations through heavy use of controls. However, in order to prove our hypothesis, we had to perform new experiments that would definitively prove that false-positive results arose out of an encapsulation of β-Lactamase.

Build

In order to restrict the cause of antibiotic degradation to OMVs, the fusion protocol was modified in order to remove OMVs prior to incubation in LB with ampicillin. After stationary incubation of bacteria and OMVs, the solution was centrifuged down in order to remove the OMVs present within the supernatant. The bacterial pellet was then resuspended and incubated further in LB media containing antibiotics.

A second experiment was required in order to verify the ampicillin degrading capabilities of OMVs, wherein OMVs would be incubated overnight in LB containing antibiotics, followed by the inoculation of a bacterial culture in the same media.

Test

No growth was observed at all proving the susceptibility of the wild type strain.

In contrast, growth was observed when a wild type culture was inoculated in media that had previously been inoculated with OMVs indicating the lack of ampicillin within the media post-OMV incubation. OMV controls showed no change in optical density; were not contaminated.

Learn

Taken together, these results confirm our theory regarding β-Lactamase encapsulation and the ability of OMVs to degrade ampicillin present in the media. Unfortunately, these results invalidate our ability to perform further “non-specific fusion” experiments using our isolated OMVs, due to the exclusive encapsulation of plasmids carrying ampicillin resistance.

After discussing with our primary investigators, we devised a strategy in order to circumvent ampicillin resistance that relied upon replacing the β-Lactamase gene with one coding for chloramphenicol acyltransferase Due to the fact that chloramphenicol acyltransferase inhibits chloramphenicol present within the cytoplasm, we would see lower encapsulation of antibiotic degrading enzymes into the OMVs. However, due to time constraints, we were not able to integrate these improvements into our project.