What are MFPs?
Auto-Oxidation and Polymerisation of PVFP-5
Post-Translational Modifications of L-DOPA
Structural Modelling of PVFP-5
Binding of PVFP-5 to PCL
MFPs in Therapeutics
Mussel foot proteins (MFPs) are a specialised group of organic polymers utilised by mussels to adhere to underwater surfaces (Danner et al., 2012). These organic polymers are typically organised into filamentous strands to form threads that subsequently end in a byssal plaque. The end of the byssal plaque is called the byssal foot, which is rich in L-3,4-dihydroxyphenylalanine (L-DOPA), a post-translationally modified version of the amino acid, tyrosine. L-DOPA contains a catechol group, which can form hydrogen and coordinate bonds with a range of surfaces, hence making it the main residue responsible for the adhesive capabilities in MFPs (Ahn, 2017).
In Phase I of our project we investigated the use of PVFP-5, an MFP found at the interface between the plaque and surface of the Asian green mussel species, Perna viridis, to coat a 3D-printed polycaprolactone (PCL) scaffold applied in the treatment of spinal cord injury (SCI). Following injury, this scaffold will be placed directly into the spinal cord lesion, where PVFP-5 will facilitate adherence of the scaffold to the spinal microenvironment. As PVFP-5 is highly biocompatible and can adhere to surfaces even in aqueous environments, it is an ideal candidate for a bioadhesive (Santonocito et al., 2019). After selecting PVFP-5 for this purpose, we primarily focused on researching the adhesion mechanism, redox behaviours, and protocols for its application. In Phase II we aimed to continue with this investigation, focusing on expression, adhesion chemistry and further applications of L-DOPA based polypeptides.
As mentioned previously, the main residue involved in adhesion in the mussel foot is L-DOPA. Our chosen protein, PVFP-5 was chosen due to its particularly high L-DOPA content which constitutes up to 21% of the amino acids present in this protein, hence providing significant adhesive properties (Bilotto et al., 2019). L-DOPA is formed via the hydroxylation of tyrosine into L-DOPA, which can be catalysed by the enzyme tyrosinase. Following this, L-DOPA can spontaneously oxidise further to form DOPA-quinone, as seen in Figure 1.
Figure 1: Reaction mechanism for the hydroxylation of tyrosine to form L-DOPA, followed by the oxidation of L-DOPA to form DOPA-quinone. The conversion of tyrosine to L-DOPA involves an oxidation reaction that results in the addition of a second hydroxyl group. L-DOPA can further be oxidised into a pair of ketones that result in the formation of DOPA-quinone.
DOPA-quinone plays a large role in cohesion between and within PVFP-5, however it has significantly reduced adhesive abilities to the target surface (Waite, 2017). Once L-DOPA has auto-oxidised to DOPA-quinone, it is used for sclerotisation, a form of tanning involving the cross-linking of proteins (Burzio et al., 2000). Cross-link formation is achieved via Michael addition, imine formation, covalent bonding with nucleophilic groups and radical generation (Burzio et al., 2000; Horsch et al., 2018). This method of polymerisation contributes to the structural integrity of the byssal thread. Additionally, during polymerisation cysteine residues are added to DOPA-quinone, forming cysteinyl-DOPA cross-links (Horsch et al., 2018). This is one of the most effective forms of cross-linking in DOPA-quinone polymers, and the cysteinyl-DOPA functionalities provide the thread with strong adhesive properties (Horsch et al., 2018).
One important consideration for the adhesive capabilities of PVFP-5 is the pH of the local environment. This is because L-DOPA is more likely to auto-oxidise into DOPA-quinone at a high pH. At the beginning of the adhesion process, the mussel pumps hydrogen ions into the byssus, reducing the pH to around 2. Once the base of the byssus is in close proximity to the target surface, the byssus is allowed to equilibrate to the pH of the sea water, (Ahn, 2017). However, our co-expression system will not have these same adaptations to combat the oxidation of L-DOPA to DOPA-quinone. As such, we investigated methods of circumventing the issue of L-DOPA being oxidized into DOPA-quinone through synthetic methods.
In our project, we conducted literature reviews to evaluate different methods available to preserve the catechol groups of L-DOPA and prevent oxidation into DOPA-quinone. This was done to maximise the number of adhesive bonds formed to our target surface. We found that there are two main ways to prevent oxidation: the first being to create a reducing environment through the use of acetic substances, and the second is to utilise catechol protecting groups to prevent L-DOPA from oxidising further. Within these two methods are a plethora of compounds that can serve to prevent oxidation.
One way of preventing the auto-oxidation of L-DOPA explored in a study conducted by Forooshani & Lee in 2016 was the use of catechol protecting groups, such as acetyl and acetonide. These act as protectors of the catechol moiety of L-DOPA, preventing it from oxidising into DOPA-quinone. However, literature on this topic is sparse, and mainly focuses on mussel-inspired bioadhesives or non-peptide L-DOPA molecules rather than mussel foot proteins, and specifically not PVFP-5.
Another avenue we explored this year was the use of acid, such as ascorbic or boronic acid. These are advantageous due to their reducing capabilities, thereby allowing them to reduce the ketone functionalities of DOPA-quinone back to the hydroxyl groups of L-DOPA to increase the adhesive ability of the protein. The use of acid was also suggested by Professor Herbert Waite, who recommended using either acetic or hydrochloric acid in our purification buffers for PVFP-5. However, even though none of these acids have previously been studied with PVFP-5, they have shown such strong reducing capabilities that they have prevented polymerisation completely in other mussel species such as A. ater and Mytilus edulis . As polymerisation is required for the tensile strength of our protein, we decided against the use of these acids.
During Phase I, we investigated ways of synthetically implementing the post-translational modification of tyrosine residues into L-DOPA, by using a tyrosinase enzyme. Tyrosinase is found in a wide range of species and, as mentioned above, catalyses the hydroxylation of tyrosine to L-DOPA (Shuster Ben-Yosef, Sendovski and Fishman, 2010). Tyrosinase was chosen for our project due to its various benefits, namely the fact that it is found in many organisms meaning it is commercially available, and it meets the requirements for material science applications (Horsch et al., 2018).
There are, however, many different types of tyrosinases that vary in activity and specificity. In Phase I, we proposed the use of a tyrosinase enzyme secreted by Bacillus megaterium, which was used by the iGEM UM Macau team in 2019 (BBa_K3033013). For our Phase II project we explored other candidates, such as the tyrosinase enzyme modified from the aquatic archaeon, Candidatus Nitrosopumilus koreensis (CTyr). An important consideration when deciding which tyrosinase to use was its monophenol/ diphenol activity. This measure is the ratio of enzymatic activity between the conversion of tyrosine to L-DOPA (monophenol activity) and the conversion of L-DOPA to DOPA-quinone (diphenol activity). This activity is crucial, as a low ratio will result in an excess of DOPA-quinone due to the quick hydroxylation of L-DOPA. As L-DOPA is the primary residue involved in adhesion, the ideal would be to ensure that L-DOPA is in high abundance as DOPA-quinone will naturally oxidise in residues that are not adherent to a surface. The two enzymes were compared in Table 1 below.
Table 1: A comparison of the tyrosinase enzymes secreted by Bacillus megaterium and Candidatus Nitrosopumilus koreensis.
The enzyme we decided to choose was Bamtyr, due to its extensive documentation. The fact that mTYR-CNK did not function in an in vivo environment added an additional amount of uncertainty about the suitability of this enzyme in our own experiment, as researchers were unable to ascertain the origin of this inactivity.
Since we decided to use a new sequence for PVFP-5 this year, we also needed to create a new structural model to visualise our protein in its native conformation before expressing and purifying it in the laboratory. We also decided to predict the binding mode of PVFP with PCL, to help validate the suitability of our protein for our scaffold, in order to optimise adhesion of the scaffold to the microenvironment of the spinal cord and promote axonal regrowth. Both of these considerations were important to research before we entered the lab, so we conducted in silico modelling methods for the structure and binding capabilities of PVFP-5.
For Phase II of our project, we decided to use a new protein sequence for PVFP-5 which is 40 amino acids longer than our Phase I sequence, and contains an additional 7 tyrosine residues to increase its adhesive strength. Our new sequence also contains a high number of cysteine residues (18), all of which we predicted would be involved in disulphide bonding. To create our structural model, we began by replicating our structural modelling approach from Phase I by using PyMol to create a full structural homology model, followed by a molecular dynamics simulation using GROMACS and YASARA to settle any atomic irregularities within our protein structure. However, we discovered that this method did not work this year due to the large number of disulphide bonds required within our structure (9). Following PFAM analysis of our protein sequence, we discovered that PVFP-5 is composed of 3 EGF-like domains, and after consulting with Dr Andrew Beavil, we decided to change our modelling approach. Following Dr Beavil’s advice, we created a new modelling approach in which we homology modelled each EGF-like domain individually, and then ran molecular dynamics simulations on the whole structure using GROMACS, after bringing all three domains together to ensure the correct formation and orientation of our desired disulphide bonds. To validate our new modelling workflow, we used the highly acclaimed AlphaFold2 artificial intelligence software to produce a structural prediction simulation on our protein, which confirmed the order of our predicted bonds. More detail surrounding our structural model can be found on our Modelling Page, and a guide to troubleshooting models with high numbers of disulphide bonds using our new modelling approach can be found on our Contributions Page.
To ensure PVFP-5 is an appropriate bioadhesive for our therapy, it needs to be able to adhere to two surfaces: tissue surface and our scaffold, which is made of polycaprolactone (PCL). Our protein’s adhesive properties towards tissue surfaces is already well-characterised in investigations of mussel foot proteins in therapeutic contexts (Wei et al., 2015). However, the binding of PVFP-5 to PCL has not previously been reported in any literature, and we therefore considered it important to predict this binding using the chemistry behind both materials. PCL is a plastic composed of repeating units of aliphatic carbons with an ester functional group present within each monomer. As seen previously, L-DOPA is a diphenol, meaning it contains two hydroxyl groups. We therefore predicted that the main binding mechanisms between PVPF-5 and PCL would be hydrogen bonding and hydrophobic interactions. This theory was then confirmed and built upon by Professor Herbert Waite and Dr Sarah Barry. Following these meetings, we created a digital visualisation of this binding using Pymol, which can be found here.
Mussel foot proteins (MFP) provide a large flexibility in their applications due to their biocompatibility and low immunogenicity. Due to these properties, they have been utilised in replacing fibrin sealant towards wound closure. The original fibrin sealants are synthesised using donated plasma, which pose a risk to the immunogenic response and in the worst cases, lead to viral infections (Spotnitz, 2014). Studies performed on rabbit wounds have shown that the MFP-fibrin hybrid compounds are capable of stopping the bleeding as well as having benefits in tissue regeneration and healing (Quan et al., 2019).
Another potential field of application for MFP is dentistry. Polyetheretherketone (PEEK) is commonly used in tooth repair due to its superior mechanical strength. However, it does not interact extensively with its surrounding biological environment and it has an increased risk for rejection or infection at the interface between implant and tissue. To overcome this issue, researchers have developed a mussel-inspired protein possessing a library of L-DOPAs, that would coat the bone implant in a similar way to how our scaffold is coated with PVFP-5 (Li et al., 2021). The protein library contains an azido group that allows for the addition of antimicrobial compounds ideal for cell growth.
Similarly, the addition of an azido group on mussel-inspired proteins could help prevent Internal Stent Restenosis (ISR) in vascular stents as it could carry functional groups or larger molecules with preventative purposes. ISR is a disorder that damages the endothelial layer triggering platelet aggregation and activation on the stent. However, mussel foot proteins could play a key role in regenerating the endothelial layer through the suppression of smooth muscle and platelet migration as well as the acceleration of endothelial cell migration and proliferation (Yang et al., 2020).