Engineering Success

Ideation

We started with the idea of using bioadhesive proteins in a medical context, particularly in drug delivery systems. These proteins, known for sticking to biological surfaces, seemed perfect for creating a system that could deliver drugs in a controlled, targeted way. After some research, we saw the potential for bioadhesive proteins in slow, localized drug release [1], which inspired us to explore how we could develop a flexible, cost-effective drug delivery platform.

For testing, we chose cancer as our model disease because it remains one of the most prevalent and deadly diseases globally. One major problem in treating cancer is that most drugs, like Doxorubicin, cause severe side effects due to non-targeted delivery, sometimes making treatment more harmful than the disease itself. Doxorubicin is highly effective, but its off-target effects can be fatal, making it a great model drug to test our system. Additionally, Doxorubicin is naturally fluorescent, which simplifies the identification of drug location within cells and tissues, allowing us to track its distribution and verify its release from our LLPS structures more effectively.

By improving the targeting of drug delivery to cancer cells, we hope to minimize side effects and improve patient outcomes. If we can get the drug to reach its target more effectively, it could reduce mortality rates and improve treatment for many diseases beyond cancer.

Aim

Our goal is to develop a flexible drug delivery system using liquid-liquid phase separation (LLPS). This system will enable targeted, slow release of drugs, reducing the costs of current delivery methods. By using LLPS, we aim to improve precision in targeting specific sites, such as tumors, while also creating a platform that could be easily adapted for various medical applications, including cancer treatment.

Engineering Approach Using the DBTL Cycle

For our project, we utilized the design-build-test-learn (DBTL) cycle, a core principle of iGEM and synthetic biology. This iterative process started with defining the problem we wanted to solve and proposing a design that outlined the key requirements for success.

After establishing our design, we moved on to the build phase, transforming our ideas into actual constructs. Once our system was built, we conducted tests to evaluate its functionality. Finally, in the learn phase, we analyzed the data from our tests to see if our design met the initial criteria.

This cycle often led us to discover areas for improvement, encouraging us to refine our designs and complete the DBTL loop. Through this progressive approach, we effectively developed and fine-tuned our drug delivery system throughout the project.

DBTL Cycle 1: Target Component

Design

We started our design process by diving into the literature on liquid-liquid phase separation (LLPS) and its potential uses. We explored different ways LLPS could form, looking at both synthetic systems and natural biological processes. This helped us think about how LLPS could be applied to drug delivery. A common challenge in current drug delivery methods, like antibody-drug conjugates (ADCs), is their lack of specificity, which can lead to unwanted side effects and increased costs. [2,3]

Our approach was inspired by our advisor, Yin, and her research on bio-adhesive proteins.[4] We focused on two proteins: MFP-1 (mussel foot protein-1) and ADF-3 (Araneus diadematus fibroin-3), both of which Yin had used in her LLPS studies. MFP-1 is known for its strong underwater adhesion, while ADF-3 offers high tensile strength. By introducing Spy Tag/Spy Catcher domains, these properties are able to combine into one protein complex, allowing for precise assembly. Yin provided us with the plasmid containing these domains, which served as the starting point for our modifications.

Next, we focused on designing our Targeting Component. Initially, our plan was to modify Mfp-1 by adding the tumor-homing peptide p160, which would allow the system to specifically target cancer cells. [5] However, as we progressed and reviewed our design in more detail, we reflected on potential challenges. After further literature research and discussions, we realized that simply adding the homing peptide might not be enough. Once the drug reached the cancer cells, it could get trapped inside the endosome, where it would degrade before releasing the drug effectively.

This realization led us to rethink and redesign our approach. To overcome this issue, we decided to add a pH-dependent membrane-active peptide (PMAP) and an endosomal protease cleavage site (EPCS) to our design. The PMAP would help disrupt the endosomal membrane in response to the acidic environment inside the endosome, allowing the drug to escape. The EPCS, on the other hand, would ensure that once inside the endosome, proteases could cleave the site, releasing the drug payload at the right moment. [6] These adjustments were crucial for ensuring that the system could successfully deliver the drug to its target without getting degraded along the way.

Build

Once we finalized the design, we moved on to create the modified MFP-1 plasmid (pYY8aTv1). This plasmid incorporates specific modifications to the original backbone we received from Yin, which includes the sequence that facilitates liquid-liquid phase separation (LLPS).

We ordered codon-optimized DNA fragments for E. coli and amplified them using Phusion Plus high-fidelity polymerase enzymes. To integrate our targeting modifications, we employed Golden Gate Assembly, allowing us to add our changes without introducing unnecessary sequences or recognition sites for restriction enzymes.

After constructing the plasmid (Figure 1), we proceeded to chemically transform it into E. coli. However, we encountered a challenge: the transformation efficiency was unexpectedly low. This was a crucial point in our workflow, and we had to address the issue quickly. We decided to repeat the transformation with a new batch of competent cells. This adjustment significantly improved the efficiency, allowing us to move forward with the project. We then selected colonies from antibiotic plates and confirmed the build via colony PCR and whole plasmid sequencing. This verification was crucial to ensure the correct incorporation of our modifications.

Next, we induced the expression of the MFP-1-p160 construct in E. coli BL21 cells, initiating protein production.

Test

After initiating our protein expression, we aimed to verify the production of the MFP-1-p160 protein using SDS-PAGE analysis on samples collected from the induced E. coli BL21 cells. Initially, we didn’t see any bands on the gel, which raised concerns about our protein production (Figure 2). This prompted us to reflect on our approach.

We suspected that our lysis method might not have been effective enough. To troubleshoot, we consulted our advisor, Alba, along with external help from Michael. They offered helpful suggestions, including refining our sonication protocol and utilizing a mix of lysis techniques, such as sonication, while also recommending we increase the sonication amplitude.

After implementing these recommendations, we induced protein expression using 0.5 mM IPTG in LB and TB media. Following induction, we collected small samples, centrifuged them, and boiled them in Laemmli buffer before loading them into the SDS-PAGE gel This analysis successfully identified our target protein band at the expected molecular weight relative to the protein ladder. However, due to time constraints and incomplete optimization of protein production, we observed additional unexpected bands (Figure 3). Encouraged by this progress, we then attempted to purify our proteins using Ni-NTA spin columns. However, this initial attempt did not yield positive results, setting us back in our attempts to get our protein.

To tackle the challenges we encountered during the purification process, we thought about potential solutions and remembered that Michael, a protein purification expert from the University of Helsinki, could provide valuable guidance. We reached out to him, and he kindly offered training on FPLC techniques, which proved beneficial in refining our purification approach. Despite the training during our first FPLC attempt, we encountered challenges due to a broken UV detector and less-than-ideal resin (Figure 2) We addressed these issues by switching to a different resin and utilizing an ActaGo machine for our subsequent purification attempts. This change ultimately led to successful protein purification, allowing us to move forward in our project.

To further assess the functionality of our targeting component, we combined the Mfp-1-p160 protein with our other components and conducted fluorescence microscopy to evaluate the formation of liquid-liquid phase separation (LLPS). (Figure 3)

Learn

Our experience with protein production and purification involved several challenges that taught us valuable lessons. Initially, we had trouble detecting the expected protein bands during our SDS-PAGE analysis, which prompted us to rethink our protein production process. By making adjustments to the growth media, induction times, and IPTG concentrations, we managed to enhance both the yield and quality of our protein.

The purification stage also brought its own set of difficulties. Choosing the right resin was key; our first attempts with Thermo HisPur™ Ni-NTA resin didn’t provide the results we hoped for, so we switched to ProBond nickel resin, which worked much better. Although we began with simple spin column purification, we found that Fast Protein Liquid Chromatography (FPLC) was more effective in concentrating our protein solutions. Despite the challenges we faced, each hurdle helped us refine our approach and improve our overall methodology.

DBTL Cycle 2: Drug Component

Design

After we completed the targeting system, we moved on to the drug delivery component. Our main goal was to find a reliable way to attach Doxorubicin (Dox) to our liquid-liquid phase separation (LLPS) system. We initially explored a few different methods for conjugation, such as EDC/NHS chemistry and using proteases like LicP [5] to help release the drug by “scarless” cleavage. However, we quickly realized that these methods weren't precise enough and often led to non-specific conjugation, which made it tough to control the drug dosage.

To tackle this challenge, we designed a rapid self-ligating intein system [6] that would allow us to precisely conjugate Dox to our protein scaffold. We incorporated a Cfa^N intein at the C-terminal of ADF-3 and created a separate payload system using a Cfa^C-CGG8 sequence. This setup allowed us to link Dox to CGG8 through a pH-sensitive linker (EMCH) [7], giving us control over how many Dox molecules were attached to each protein. The Cfa^C-Cfa^N intein system enabled self-ligation between the ADF-3 protein and the Cfa^C-CGG8-Dox payload, ensuring we had consistent dosing in our delivery system.

Build

After we completed the design for our drug delivery component, we moved on to construct the modified ADF-3 plasmid, incorporating the Cfa^N intein using Golden Gate Assembly. Given that the Cfa^C-CGG8 open reading frame (ORF) had a high GC content, we opted for Gibson Assembly to ensure our fragments amplified effectively via PCR and formed a scarless plasmid. (Figure 1)

In addition to the ADF-3 construct, we created separate eGFP plasmids linked through flexible linkers with Cfa^N, Cfa^C, and the Spycatcher domain. These eGFP plasmids served as a verification system since this was our first time working with inteins and related components. This design aimed to confirm that each component was non-toxic during protein expression, which we could verify through emission signals after excitation at 488 nm.

Once we completed constructing the plasmids and performed DNA precipitation, we proceeded to transform them into chemically competent E. coli cells. After plating on antibiotic plates, we selected colonies and confirmed the success of our constructs through colony PCR and whole plasmid sequencing (Figure 6). Each assembled sequence was annotated and compared to a theoretical map to check for any possible mutations.

However, when the sequencing results came back, we encountered an issue. Several different mutations had appeared, specifically affecting the number of CGG8 sites in our plasmids (5, 13, 3). Given the time constraints, we couldn't go back and redesign the plasmids, so we chose the plasmid with the CGG8 site count closest to our original design. This version was then used for all future work involving CGG8.

Test

After getting our modified ADF-3 plasmid and the Cfa^N intein all set up, we turned our attention to the drug delivery component. First up, we prepared a 10 mg/mL stock solution of Doxorubicin (Dox) and a 50 mg/mL EMCH solution. These were key ingredients for the next steps, where we aimed to link Dox to the CGG8 payload.

Using the Cfa^C-Cfa^N intein system, we combined the Dox-linked CGG8 payload with our ADF-3 scaffold. This setup allowed us to maintain precise control over how much drug was loaded into our liquid-liquid phase separation (LLPS) system, which was essential for ensuring that our delivery system would work effectively.

To make sure our proteins were actually being produced, we ran SDS-PAGE on the samples we obtained from the ADF-3 protein expression. This gel electrophoresis technique helped us visualize the protein bands, allowing us to check for distinct bands at the expected molecular weight. It was a reassuring step to confirm that our proteins were being expressed as planned. (Figure 3)

In addition to SDS-PAGE, we also turned to Fast Protein Liquid Chromatography (FPLC) to purify our samples. This method helped us isolate the proteins and ensure that we had high-quality preparations for our experiments. Having pure samples was crucial for the next testing phase.

Once we had our proteins ready, we moved on to testing the drug component. We confirmed the successful conjugation of Doxorubicin to the CGG8 payload by tracking Dox's natural fluorescence. Despite some challenges with multimerization, the system was functioning as expected.

To further validate our approach, we used a fluorescence microscope to observe the formation of LLPS droplets in the presence of the Cfa^C-CGG8-Dox payload. It was exciting to see the droplets form, confirming that the drug was properly integrated into the system and that everything was working as intended.(Figure 4)

Learn

During our experiments, we faced some challenges, particularly during the dialysis process. We initially used a 3.5 kDa membrane to separate free Doxorubicin (Dox) and excess EMCH linker from the conjugated protein. However, we found that this approach was quite tricky and did not yield the pure liquid-liquid phase separation (LLPS) droplets we were aiming for. Instead, we ended up with aggregates, which prompted us to reassess our methods.

This experience highlighted the importance of fine-tuning our dialysis technique to ensure effective separation. In hindsight, we realized that opting for ultracentrifugation might have been a better approach to achieve cleaner results.

Additionally, we discovered that proper filtration is crucial for obtaining a clean conjugation product. Impurities significantly impacted LLPS formation, affecting our system's efficiency. It was a valuable lesson when we noticed that even small contaminants could influence the outcome of our drug delivery system. Further, While the CGG8 protein wasn't entirely pure due to multimerization and the presence of free cysteine residues, we could possibly address this by adding TCEP to reduce the disulfide bonds.

From these experiences, we learned the need to optimize our purification and conjugation steps going forward. This means not only refining our dialysis and filtration techniques but also closely monitoring the entire conjugation process to ensure we achieve the desired product without unwanted side effects.

As we move ahead, we can apply these lessons and improve our protocols, which should enhance the efficiency and effectiveness of our drug delivery system. It's all part of the learning curve, and we're confident that we can build on our progress!