Introduction
We performed a range of modelling across each aspect of our project to gain an insight into how effective our therapy is and ultimately assess the feasibility of our design. In summary, we:
- Carried out molecular simulations on Chondroitinase ABC to prove that the mutated sequence is thermally stable, and hence suitable for our project
- Performed computational fluid dynamics on the scaffold to explore the effectiveness of the scaffold’s design, both in terms of the suitability of the micro- and macro-architecture, validate the design change, and provide future avenues for scaffold design optimisation
- Generated a structural model of the PVFP-5 protein using a combination of homology modelling an molecular dynamics, to determine 3D function of model and to develop a PVFP-5 and polycaprolactone (PCL) binding model.
- Explored the binding of PCL and PVFP-5 to validate that the mussel foot protein remains bound to the scaffold in the spinal cord microenvironment
- Calculated how much PVFP-5 coating is required for each personalised PCL scaffold for future reference.
Molecular simulation of Chondroitinase ABC
This year our team decided to look at Chondroitinase ABC (ChABC) as a potential therapeutic agent in SCI. The main use we are considering is injections along the spinal cord, which have the potential to stimulate neurite regrowth. The main issue with applying ChABC endogenously is that it becomes unstable at normal human physiological temperatures, 37°C. In order to solve this problem our team collaborated with the Phystech Moscow 2021 iGEM team to computationally model eight mutations on the 1HN0 ChABC structure.
Computational Mutation
A paper published by Dr. Marian Hettiaratchi in August of 2020 (Hettiaratchi et al., 2020) analysed the impact of directed mutagenesis at certain residue positions of ChABC. We were lucky enough to be provided by Dr. Hettiaratchi with a list of eight positions that were suspected to have a significant impact on the stability of ChABC. We were also given the mutations that might have the best impact at those positions.
In order to simulate those mutations in silico we compared the canonical 1HN0 structure (WT) with a homology model of the sequence with the mutated residues (MUT). The homology model was created using MODELLER 10.1. The two structures' sequences and overall structures were positionally identical, with the exception of the eight mutations.
Parameters of molecular dynamics simulation
The two structures WT and MUT were sent over to the Phystech Moscow team, who modelled and compared their thermostability in silico for 100 frames at 310.15°K (37°C). The simulation was performed in an explicit solvent that had been equilibrated for charge, pressure and temperature using the “c36m” forcefield and supporting ensembles through GROMACS.
Our Analysis
Figure 6: A comparison between the distributions for the frames of the total radius of gyration values of the wild-type and mutated ChABC protein models. The difference was statistically significant (p = 3.5e-9)
Figure 7: A comparison between the distributions of all the protein residues based on RMSF between the wild-type and mutated ChABC protein models. The difference was statistically significant (p = 1.1e-10)
Analysis of total results by Rgyr and RMSF
Two further ways the WT and MUT structures were analysed was through measuring the Radius of Gyration (Rgyr) for the whole protein over the modelling time and the Root Mean Square Flexibility (RMSF) of each residue. In Figures 6 and 7 we can see distribution plots for the values of those measurements. Overall, they also show significant positive improvement in stability. The mean Radius of Gyration went down from 34.79 to 34.51 with noticeable decrease in the maximum from 35.58 to 35.06 units. The RMSF values showed a decrease from 4.87 to 4.44 on average and 12.6 to 7.6 at the maximum. The latter part of this result shows that the mutations managed to greatly improve the most unstable regions in the protein.
Figure 8: A comparison of the RMSF values before and after mutation. The values are given for each residue position, with the mutated position in the centre.
Analysis of the local effect of mutations by RMSF
In Figure 8 we can see each of the mutations and how they affected the RMSF value at their own position and also the positions surrounding them. The effects can be split into three broad categories. ALA228LYS, ASN288ASP, SER343ASN, GLN 781 GLU are obvious improvements in the rigidity of the structure. SER274PRO and ARG670THR had what can be described as a variable effect. In the N-terminal direction the protein lost rigidity, which it gained in the C-terminal direction. LYS194GLU and LYS654ASP seem to have had an increase of flexibility. However, what also needs to be considered is not only the residues surrounding the mutation on the amino-acid chain, but also the residues that come into contact with it in three dimensional space. The massively decreased maximum mentioned in the section above was likely due to precisely LYS194GLU. Residue 194 sits opposite THR112. In the WT structure the RMSF value for the latter is 12.6, whereas in the MUT structure it is only 6.6. This decrease is best attributed to the effects of the mutation on the rigidity of the structure. Therefore, there was a slight increase in RMSF locally, it contributed to stabilising another area much more.
Conclusion
Overall, the results of the computer simulation of the combined effect of these eight mutations proved to be a moderate success. While it did not provide massive increases to the average stability of the whole 1HN0 structure, it did show possible solutions for several unstable local regions. It is also a good proof of concept that such an approach can provide valuable insight into the compound effect of several mutations.
Computational Fluid Dynamics (CFD)
What is Computational Fluid Dynamics (CFD)?
Computational Fluid Dynamics (CFD) is a mathematical method of modelling fluid behaviours so that they can be visualised in a digital simulation. Typically, fluid behaviours are modelled in relation to their movement within and around objects, in this case, a 3D printed scaffold. In order to model this behaviour, CFD makes use of various numerical analysis techniques to solve mathematical equations.
Naturally, modelling this behaviour requires consideration of many different factors such as pressure, mass and velocity, all of which are dependent on one another. In order to account for such variables, complex partial equations have been devised, such as the Navier-Stokes equations (Adair and Jaeger, 2019). Collectively, such equations are solved together to develop a more complete mathematical model of fluid behaviour.
In order to apply these equations, the object, fluid and region within which it is contained must be appropriately digitally represented through setting boundary conditions and meshing. Here, boundary conditions are seen as the ‘edges’ of the model, within which a closed system is formed, allowing modulation of the in and output of fluid volume. Within this system, the regions are meshed (“Mesh Types in CFD”, n.d.), wherein larger regions are split up into many smaller cells, the behaviour of which is dependent on that of other nearby cells. The coarseness of this mesh has a strong influence on the accuracy of the modelling, wherein a greater number of cells facilitates more accurate predictions of fluid behaviours as a result of accounting for more minute details (Liu et al., 2004). However, this can require an extremely large amount of computing power and take a long time to model. As such, one of the main features of CFD is identifying which elements of the model to simplify in order to maintain a balance between processing speed and accuracy.
CFD is an invaluable tool for understanding how fluids may behave in a given environment and has wide-ranging applications from anti-flooding architecture (Munoz and Constantinescu, 2018) to the behaviour of cerebrospinal fluid in relation to a 3D printed scaffold (Zhang et al., 2019), as is presently described. However, it is important to note that CFD does not provide 100% accurate predictions and should generally be seen as an estimate of behaviour due to a strong dependence on the validation of the chosen mathematical model and available processing power.
Why is CFD applicable to our project?
Within our current project, we are aiming to insert a 3D printed scaffold coated in mussel foot protein within the lesion at the site of complete thoracic spinal cord injury, aiming to promote axonal regrowth. Injections of ChABC will also be administered at either end of the scaffold to further facilitate regrowth.
Within the spinal cord, a vast variety of substances are transported via the cerebrospinal fluid (CSF), which acts as a medium for the circulation of nutrients and chemicals. Naturally, the implantation of a foreign object within the spinal cord will impact the flow behaviour of this fluid. In this regard, CFD allows us to investigate a variety of variables that relate to our scaffold, while being an effective alternative compared to complex and costly experimental tests to measure the same variables (Ali and Sen, 2018). In the case of the present project, we are specifically investigating wall shear stress (WSS) and permeability.
Permeability relates to the number of pores, interconnected or otherwise, that strongly determine the fluid flow through the scaffold (Singh et al., 2018). As such, permeability plays a key role in the scaffolds ability to facilitate waste removal, as well as oxygen and nutrient supply (Ochoa et al., 2009; Ali et al., 2018), thus directly influencing a scaffold's ability to promote axonal regeneration. However, there is a balance in permeability that must be achieved, where too great a permeability can result in too great flow rates, thus causing cell washout, whereas too low a permeability can result in poor nutrient supply to cells (Gómez et al., 2016; Truscello et al., 2012). As such, investigation of how the permeability of our scaffold will likely perform is an important initial test, again allowing us to adjust our design if necessary.
WSS, on the other hand, is defined as the ‘force per unit area applied by the wall surface on the fluid in a direction on the local tangent plane’, where its magnitude is directly proportional to the fluid velocity near a given wall (Katritsis et al., 2007). WSS has been shown to play a crucial role in a variety of variables, such as promoting mechanobiological responses and promoting cellular attachment (Kwon and Jacobs, 2007; Davies, 1995), two factors particularly relevant given our efforts to promote axonal regeneration. Additionally, WSS has a strong influence on the rate of shear-induced scaffold degradation (Chen et al., 2004; Chen et al., 2011; Lin et al., 2003), providing information as to how our scaffold may degrade over time and thus highlighting how we may improve our design.
Our Simulation
Prior to setting up the model, a literature review was conducted to obtain the relevant parameters required
to
complete our simulations. These values facilitated the modelling (with the highest possible accuracy) of the
cerebrospinal fluid that would be surrounding our scaffold once it is implanted in the place of the glial
scar in
the spinal cord.
Table 1: Parameters used in the computational fluid dynamics simulations accompanied with the
reasoning
or source justifying its use.
| Parameter |
Value |
Units |
Source/Reasoning |
| Inlet Velocity of Cerebrospinal Fluid at C4 |
10 |
cm/s |
(Haughton and Mardal, 2014) |
| Inlet Pressure |
1 |
mmHg |
(Gupta et al., 2010) |
| Dynamic Viscosity (of Water/CSF) |
10-3 |
Pa⋅s |
(Yu et al., 2020) |
| Mesh Element Size |
1.5 |
mm |
The value that generated the finest mesh quality that could be achieved |
After analysis of the spinal cord environment, and assessing the computational power available, appropriate
assumptions were made with the aim of facilitating successful simulations without compromising the validity
of the
model. This was confirmed through engagement with experts—specifically Dr Jack Lee, Dr Bryn Martin and Professor Sundararajan
Madihally
who work within the computational modelling and computational fluid dynamics fields, whose guidance is
outlined on
our Human Practices page .
Table 2: The assumptions made to carry out the computational fluid dynamics simulations
accompanied with
sources that have done the same.
| Assumption |
Source |
| CSF flow (i.e. flow through the scaffold) is laminar and viscous |
(Liu et al., 2017), (Ali and Sen, 2018) |
| CSF has similar properties to water |
(Martin et al., 2005) |
| CSF can be assumed to be an incompressible Newtonian Fluid |
(Kurtcuoglu et al., 2019), (Liu et al., 2017) |
| Rigid CFD domain boundaries |
(Kurtcuoglu et al., 2019) |
Next, the scaffold STL file created for the purpose of 3D printing on Fusion 360 was prepared
for
simulation. The scaffold’s cylindrical log structure was changed to cuboidal-shaped to overcome difficulties
in the
next stages of the simulation. The cuboidal logs create a larger number of points of contact between
crosshatched
logs than cylindrical logs meaning the process of obtaining a mesh of our scaffold geometry would be more
likely to
be successful. To ensure the scaffold geometry was of high quality and would therefore allow a smoother
simulation
process, its geometry was imported into Meshmixer. Within this software, values such as mesh density
and
offset distance were adjusted to ensure the solid body of our scaffold does not have any non-manifold
geometry. This
is when an edge in the solid is being shared by more than two faces, often implying there may be faces that
are
identical and overlaid in the scaffold geometry, which would cause errors during the meshing stage of the
simulation
process. Once this is completed the scaffold file was converted from an STL format to STEP
format on
Fusion 360 to allow for CFD simulation.
The ANSYS Workbench via the ANSYS Student 2021 R1 package was opened; this contains the
different
stages of the model. The Fluid Flow (Fluent) system was inserted into the Project Window of
the
Workbench via the Toolbox and the STEP file of the scaffold was imported into the
Geometry module. The scaffold geometry must undergo processing again to check and repair for issues
like
missing faces and gaps. Once the scaffold was error-free, the scaffold was split at two points of symmetry
and
portions were removed to leave a quarter of the scaffold— as this allows for simplification of the geometry
and,
therefore, less required computational power. As aforementioned, it is assumed that the fluid flow and
properties
remain the same for parts of the symmetrical parts of the scaffold. A cylindrical enclosure centred at the
original
origin of the whole scaffold was inserted to create boundary walls for fluid flow. It was also split into
quarters
to match the scaffold and a cuboidal portion of the enclosure was removed to leave an enclosure that was
only
slightly larger than the scaffold. A gap had to be present, otherwise the intersection would cause the
meshing
process to fail. The fluid region to be modelled was obtained by using the Combine tool, which
removed
regions of the enclosure that overlapped with the scaffold geometry to create negative spaces. The scaffold
is then
suppressed to prevent it from being involved in the simulation and removed from view. Different parts of the
remaining geometry were grouped with the labels: Inlet, Outlet, Wall, Symmetry1,
Symmetry2, Scaffold.
The Mesh module was then used to generate a mesh. The desired element size for the simulation
is
inputted. This defines how fine the mesh is and therefore affects the accuracy of the simulation results.
After
selecting the whole geometry, the Generate Mesh function was initiated.
The generated mesh was transferred to the Setup module and checked for quality of mesh to ensure that
the
simulation would run without any mesh-related errors. We changed the following settings from the default to
run our
simulations.
Table 3: *obtained from Fluent Database. ** converted from mmHg to Pa from the Inlet Pressure
parameter
as outlined in Table (1). Detailed information about the inputs for variables during the set up of
computational
fluid dynamics simulations.
| Setup variable |
Input |
| Models |
Viscous - Laminar |
| Materials >> Fluid |
Water-liquid (h20)* |
| Cell Zone Conditions >> Scaffold >> Edit |
Material Name: water-liquid |
| Cell Zone Conditions >> Scaffold >> Operating Conditions |
Operating Pressure: 133.322 Pa**
Gravitational Acceleration in Z: 9.81 ms-2
|
| Boundary Conditions >> Inlet >> Edit |
Velocity Magnitude: 0.1 ms-1
Supersonic/Initial Gauge Pressure: 133.322 Pa**
|
Following the setup of the simulation, the convergence conditions were set by altering the Residuals
section
of the Solutions module. The Absolute Criteria of the residuals continuity,
x-velocity,
y-velocity and z-velocity were set to 10-6 (as Ali and Sen (2018) used). The
simulation was
then initialised using 100 iterations and the Hybrid Initialization method. Once the initialisation process
was
completed, the calculation was run with 500 iterations.
After the calculations were run the Graphics tab was used to obtain various visualisations of the
fluid
simulations, such as pathlines of the fluid. The pressure drop was calculated using the inlet and outlet
pressures
obtained from the Reports tab under Surface Integrals as Area-weighted Averages. From
here, the
permeability coefficient of the scaffold could be calculated using the following equation (Yu et al., 2020):
(1)
This is where k is the permeability coefficient (m2), L is the length of the model (m), v is the
fluid
velocity (ms-1), ΔP is the pressure drop (Pa), and μ is the dynamic viscosity coefficient (Pa⋅s).
The wall shear stress values were obtained by selecting to visualise Contours under the
Graphics tab of
the Solve module. The wall shear stress at different points of the scaffold is displayed and the
maximum and
minimum values are returned.
This section presents the results for the permeability coefficient and wall shear stress of the modelled
scaffold obtained through the CFD simulations completed as outlined in the previous section.
Figure 9: A video showing the flow of the modelled fluid through the scaffold. The colours of the
pathlines and the particles represent the respective velocities.
Permeability
Table 4: The calculated permeability of the modelled scaffold from pressure drop values obtained
from computational fluid dynamics simulations run with meshes with element sizes ranging from 0.0015 m to
0.0030 m
| Mesh Element Size (m) |
Permeability Value (m2) |
| 0.0015 |
1.54 × 10-7 |
| 0.0020 |
1.56 × 10-7 |
| 0.0025 |
1.57 × 10-7 |
| 0.0030 |
1.63 × 10-7 |
Wall Shear Stress
Figure 10: Contour graphics of the modelled scaffold's wall shear stress in multiple views (click the arrows to change the image)
Table 5: The maximum and minimum wall shear stress of the modelled scaffold obtained from
computational fluid dynamics simulations run with meshes with element sizes ranging from 0.0015 m to 0.0030
m
| Mesh Element Size (m) |
Minimum Wall Shear Stress (Pa) |
Maximum Wall Shear Stress (Pa) |
| 0.0015 |
0.0019 |
1.10 |
| 0.0020 |
0.0140 |
2.10 |
| 0.0025 |
0.0200 |
0.87 |
0.0030 |
0.0130 |
0.79 |
Overall it was found that the scaffold has a permeability value that is higher than desired, however it is
within the range of scaffold permeabilities found in literature. Conversely, the WSS distribution found
suggests that the geometry is well-matched for its intended purpose, and reaffirmed that the change of
macro-architecture is justified. Further work should look at tuning the micro-architecture parameters
(such as pore size and geometry) to perfect the permeability constant of the scaffold. Other avenues of
investigation should also include the effects of surface roughness on the calculated parameters (Ali and Sen,
2018), as well as the effects of breathing and patient specific geometry (Pizzichelli et al., 2021).
However, the current studies provide a sound justification of the design choices and provide suggestions for
future adjustments to the scaffold design.
Modelling Perna viridis Foot Protein 5 (PVFP-5)
Introduction
The Perna viridis Foot Protein 5 (PVFP-5) was chosen as part of the KCL iGEM Renervate Therapeutics project as a way to adhere neurons to our scaffold. It possesses a very low cytotoxicity and immunogenicity, which makes it suitable as an implant coating (Santonocito et al., 2019). Mussel foot proteins (MFPs) also have the exceptional ability to adhere underwater after they have been secreted. This is because they have a very high L-DOPA content, which allows for bond formation with various surfaces. The L-DOPA is created as a result of an oxidation reaction involving the tyrosines of the protein, which make up to 20% of the overall mass of the protein’s residues. Upon secretion, proteins are required to maintain their structure under a variety of conditions, therefore they often utilise disulphide bonds for structural stability. Based on alignments and structural distances, we predicted the formation of nine such bonds. The purpose of the modelling was to ensure all the predicted disulphide bonds were formed correctly to create an accurate structural model that can be used to further characterise our protein in preparation for our wet-lab experiments.
Our Model
We used UniProt (Bateman et al., 2021) to find a PVFP-5 sequence that we would use. Last year’s team picked the sequence with the accession number U5Y6U9. We have since switched to U5Y3S6. This is because it has more tyrosines and therefore higher potential adhesive ability. The two sequences are shown here:
>tr|U5Y6U9|18-140 (old)
RDYYLNPCLPNPCRYGGTCKSIGLFGYKCFCTNGYKGKNCQFNACTPNPCLNGGTCALIYGPPYQCSCPYGYYGTKCEFKRHYYDRCGGCLNGGLCISDSYGKY
VCRCKPGYYGKRCIDPYY
>tr|U5Y3S6|18-139 (new)
VYYPNPCSPYPCRNGGTCKKRGLYSYKCYCRKGYTGKNCQYNACFPNPCLNGGTCGYVYGYPYYKCSCPYGYYGTKCEFKRHYYDRCGGCLNGGLCISDSYGKY
VCRCKPGYYGKRCIDPYY
Initially, we performed whole-sequence homology modelling using MODELLER 10.1 (Webb and Sali, 2021), which returned results that did not include the cysteine bond between the first and third cysteine. Under the advice of Dr. Andrew Beavil, the strategy was changed to instead individually model the domains that PVFP-5 is composed of. After running the protein sequence through PFAM (Mistry et al., 2021), we discovered that the protein likely contains three EGF-like domains. Each of those domains contains three of the nine cysteine pairs. The sequence was then split accordingly:
After the sequence was split according to its domains, we used the same workflow on each part for consistency. The workflow can be split into several parts as follows.
First, the sub-sequence was used as a BLAST query to search for homologues among the structures in the RCSB PDB database (Agarwala et al., 2018). A structure was considered a good template if it had a sequence with high query cover, homology, and identity. It also needed to have three cysteine bonds that matched those expected of an EGF-like domain. The alignments between the queries and the templates we selected can be seen in Figure 11 with all of the accompanying statistical information.
Figure 11: Color coded queries alignment to templates with accompanying statistics. From top to bottom the templates are: 5FMA (Weisshuhn et al., 2016), 1EDM (Rao et al., 1995), 6OFY (Dong, Anderson and Malkowski, 2019).
Second comes the initial modelling. This follows the steps usually taken for a standard homology modelling task. For it we use PyMOL (Schrödinger, 2020) as our visualisation software. To extend its capabilities we use the PyMOD plugin (Janson and Paiardini, 2021) coupled with MODELLER (Webb and Sali, 2021). To use them we import the query sequence and also the template structure. They are then aligned using the MUSCLE tool set at “Highest accuracy” (Edgar, 2004). From that alignment we then run MODELLER at the highest default setting using the PyMOD interface. At the end, we get a homology model to use as our initial structure.
Figure 12: Side by side comparisons of the cysteine bonds of the template (left) and model (right) structures.
Third, we run a molecular dynamics simulation. For that we used GROMACS (Abraham et al., 2015), which is a free software that is used predominantly on Linux. We have provided a general tutorial on how to use it HERE. Following those steps, we improve our domain models.
As a final step, each of the resulting structures were checked by MolProbity (Chen et al., 2010) for overall quality and clashscore. They are also checked in PyMOL to make sure that the cysteine bonds look correctly positioned. A side-by-side comparison of the template and the model cysteine bonds can be seen on Figure 12.
Now that each of the domains have been modelled and verified, their structures are aligned to the original full sequence using ClustalΩ (Sievers et al., 2011). Then MODELLER was used to combine the 3 domains into the full, correctly oriented protein structure. This full protein structure is then run through GROMACS for the full molecular dynamics simulation, to create the final protein model.
Figure 13: Model structure for the full PVFP-5 sequence with labelled tyrosine residues.
As we can see on the table below, GROMACS substantially improved the quality of the structures and removed any clashes.
Table 6: Quality scores for structural models of the templates, homology models, and refined models as asssigned by MolProbity.
For our final model, after we had run it through GROMACS we got an overall MolProbity score of 1.73 with a clashscore of 0. Both of those are excellent results that show promise for the accuracy of our model. The final model structure can be seen on Figure 13.
MFP-PCL Binding
We have identified a combinatorial approach to treating spinal cord injuries (SCI), involving the use of a polycaprolactone (PCL) based scaffold and our synthetic mussel foot protein bioadhesive. Prior to the implementation of our novel therapy, we needed to understand the surface interactions between these two main components of our therapy, i.e. whether the two chemical compounds were compatible and able to bind.
In developing our model, we first, conducted an extensive literature review to assess the evidence of binding between PVFP-5, our mussel foot protein and our scaffold. However, we were unable to identify any concrete resources to affirm our assumptions. Instead, we decided to study the individual chemistry of PCL and PVFP-5.
PCL is a polymer characterised by repeating units of aliphatic carbon chains with one ester functional group in each monomer. The ester is a fairly unreactive and stable compound, which contains a carbon double bonded to an oxygen. We discovered that this oxygen confers the intermolecular binding properties of PCL, as it has two available lone pairs of electrons and therefore holds particular relevance in the development of our therapeutic.
Figure 14: The overall structure of PCL can be summarised through its empirical formula, [C6H10O2]n, where n depends on the quantities of polymer available.
Following our polymer research, we identified the functionality of PVFP-5, which is based on the properties of the individual amino acids that construct its primary sequence, cysteines (C) and L-3,4-dihydroxyphenylalanine (DOPAs).
Cysteines are essential to the structure of our protein, as they are involved in disulfide bond formation and the stability of the polypeptide. Whilst DOPA, a post-translationally modified amino acid present in almost all MFPs, appears to hold a major role in the binding of our adhesive. It is composed of one phenyl ring with two adjacent hydroxyl substituent groups, which not only confers a degree of polarity to the molecule, but are also capable of hydrogen bonding.
By identifying the different properties involved in the functionalities of DOPA and PCL, we predicted that the carbonyl oxygen of our polymer might have the ability to hydrogen bond in a 1:2 ratio with the two hydrogens of the hydroxyl groups of the DOPA units present in our adhesive protein. Since, hydrogen bonds are regarded as strong bonds, we believe both chemical compounds are able to bind.
We approached Professor Herbert Waite, Distinguished Professor, co-leader of IRG-1 for the new Materials Research Science and Engineering Center (MRSEC) at UCSB funded by NSF, to validate our proposed binding model. He suggested that any free thiols present in the protein structure would be able to covalently bind to our scaffold structure. However, we were able to disprove this theory through our structural modelling. In fact, we determined that all cysteines should be involved in disulfide bridges, hence there would be no available thiol side chain suited for irreversible bonding. Additionally, he identified how different amino acid components could be involved in the binding between PCL - PVFP-5, such as cysteine. Additionally, Dr Sarah Barry suggested another main form of bonding: hydrophobic interaction between the straight aliphatic chain of PCL and the non-polar residues in PVFP-5.
After validation and incorporating his feedback into our predicted binding model, we developed a visual representation of our binding model using PyMOL and Adobe softwares to present our findings which can be found in the video below.
Figure 15: Visual representation of MFP-PCL binding model
Dynamic Calculator for MFP binding amount
We analysed our 3D model of the scaffold and protein respectively and noted their measurements. Through this, we were able to compute an estimate of the amount of PVFP-5 that would bind to a given surface area of our PCL polymer. The measurements obtained from the aforementioned investigation were then able to provide enough data to conclude that 1.4 micrograms of PVFP-5 would be required to supply a single layer coating of adhesive protein onto a 10 mm long scaffold with a diameter of 12 mm.
Given the nearly directly proportional relationship between the length of our 3D model and the required protein weight, we were able to make accurate predictions. These predictions could be of great use if implemented as guidance for future laboratory work as well as to direct any future manufacturing needs.
The calculator can be accessed directly on this page from the input field here:
References:
Molecular simulation of Chondroitinase ABC
- Hettiaratchi, M. H., O’Meara, M. J., O’Meara, T. R., Pickering, A. J., Letko-Khait, N., & Shoichet, M. S. (2020). Reengineering biocatalysts: Computational redesign of chondroitinase ABC improves efficacy and stability. Science Advances, 6(34). https://doi.org/10.1126/sciadv.abc6378
Computational Fluid Dynamics
- Adair, D., & Jaeger, M. (2019). An Efficient Strategy to Deliver Understanding of Both Numerical and Practical Aspects When Using Navier-Stokes Equations to Solve Fluid Mechanics Problems. Fluids, 4(4), 178. https://doi.org/10.3390/fluids4040178
- Ali, D., & Sen, S. (2018). Computational Fluid Dynamics Study of the Effects of Surface Roughness on Permeability and Fluid Flow-Induced Wall Shear Stress in Scaffolds. Annals Of Biomedical Engineering, 46(12), 2023-2035. doi: 10.1007/s10439-018-2101-z
- Ali, D., Ozalp, M., Blanquer, S., & Onel, S. (2020). Permeability and fluid flow-induced wall shear stress in bone scaffolds with TPMS and lattice architectures: A CFD analysis. European Journal Of Mechanics - B/Fluids, 79, 376-385. doi: 10.1016/j.euromechflu.2019.09.015
- Chen, H. B., Sundararaj. U., & Nandakumar, K. (2004). Modeling of Polymer Melling, Drop Deformation, and Breakup Under Shear Flow. Polym Eng Sei, 44(7), 1258-1266.
- Chen, V. H., Zhou, S. W., & Li, Q. (2011). Microstructure Design of Biodegradable Scaffold and its Effect on Tissue Regeneration. Biomaterials, 32(22), 5003-5014.
- Davies, P. F. (1995). Flow-mediated endothelial mechanotransduction. Physiological reviews, 75(3), 519–560. https://doi.org/10.1152/physrev.1995.75.3.519
- Gómez, S., Vlad, M. D., López, J., & Fernández, E. (2016). Design and properties of 3D scaffolds for bone tissue engineering. Acta biomaterialia, 42, 341–350. https://doi.org/10.1016/j.actbio.2016.06.032
- Gupta, S., Soellinger, M., Grzybowski, D. M., Boesiger, P., Biddiscombe, J., Poulikakos, D., & Kurtcuoglu, V. (2010). Cerebrospinal fluid dynamics in the human cranial subarachnoid space: an overlooked mediator of cerebral disease. I. Computational model. Journal of The Royal Society Interface, 7(49), 1195–1204. https://doi.org/10.1098/rsif.2010.0033
- Gutierrez, R., & Crumpler, E. (2007). Potential Effect of Geometry on Wall Shear Stress Distribution Across Scaffold Surfaces. Annals Of Biomedical Engineering, 36(1), 77-85. doi: 10.1007/s10439-007-9396-5
- Haughton, V., & Mardal, K.-A. (2014). Spinal Fluid Biomechanics and Imaging: An Update for Neuroradiologists. American Journal of Neuroradiology, 35(10), 1864–1869. https://doi.org/10.3174/ajnr.A4023
- Katritsis, D., Kaiktsis, L., Chaniotis, A., Pantos, J., Efstathopoulos, E. P., & Marmarelis, V. (2007). Wall shear stress: theoretical considerations and methods of measurement. Progress in cardiovascular diseases, 49(5), 307–329. https://doi.org/10.1016/j.pcad.2006.11.001
- Kurtcuoglu, V., Jain, K., & Martin, B. A. (2019). Modelling of Cerebrospinal Fluid Flow by Computational Fluid Dynamics. In K. Miller (Ed.), Biomechanics of the Brain (pp. 215–241). Springer International Publishing. https://doi.org/10.1007/978-3-030-04996-6_9
- Kwon, R. Y., & Jacobs, C. R. (2007). Time-dependent deformations in bone cells exposed to fluid flow in vitro: investigating the role of cellular deformation in fluid flow-induced signaling. Journal of biomechanics, 40(14), 3162–3168. https://doi.org/10.1016/j.jbiomech.2007.04.003
- Lin, B., Sundararaj, Ü., Mighri, F., & Huneault, M. A. (2003). Erosion and Breakup of Polymer Drops Under Simple Shear in High Viscosity Ratio Systems. Polym Eng Sei, 43(4), 891-904.
- Liu, X., Luo, D., Hu, P., Yu, M., & Rong, Q. (2017). Modelling and Analysis of the Cerebrospinal Fluid Flow in the Spinal Cord. In M. Fei, S. Ma, X. Li, X. Sun, L. Jia, & Z. Su (Eds.), Advanced Computational Methods in Life System Modeling and Simulation (Vol. 761, pp. 77–85). Springer Singapore. https://doi.org/10.1007/978-981-10-6370-1_8
- Liu, Y., Pekkan, K., Jones, S. C., & Yoganathan, A. P. (2004). The effects of different mesh generation methods on computational fluid dynamic analysis and power loss assessment in total cavopulmonary connection. Journal of biomechanical engineering, 126(5), 594–603. https://doi.org/10.1115/1.1800553
- Mahammod, B., Barua, E., Deb, P., Deoghare, A., & Pandey, K. (2020). Investigation of Physico-mechanical Behavior, Permeability and Wall Shear Stress of Porous HA/PMMA Composite Bone Scaffold. Arabian Journal For Science And Engineering, 45(7), 5505-5515. doi: 10.1007/s13369-020-04467-w
- Martin, B. A., Kalata, W., Loth, F., Royston, T. J., & Oshinski, J. N. (2005). Syringomyelia Hydrodynamics: An In Vitro Study Based on In Vivo Measurements. Journal of Biomechanical Engineering, 127(7), 1110–1120. https://doi.org/10.1115/1.2073687
- Mesh Types in CFD. (n.d.). Retrieved August 23, 2021, from https://www.manchestercfd.co.uk/post/all-there-is-to-know-about-different-mesh-types-in-cfd
- Munoz, H. D., & Contantinescu, G. (2018). Application of a 3-D model to investigate flood-related engineering problems. River Flow, 40. https://doi.org/10.1051/e3sconf/20184006004
- O'Brien, F., Harley, B., Waller, M., Yannas, I., Gibson, L., & Prendergast, P. (2006). The effect of pore size on permeability and cell attachment in collagen scaffolds for tissue engineering. Technology And Health Care, 15(1), 3-17. doi: 10.3233/thc-2007-15102
- Ochoa, I., Sanz-Herrera, J. A., García-Aznar, J. M., Doblaré, M., Yunos, D. M., & Boccaccini, A. R. (2009). Permeability evaluation of 45S5 Bioglass-based scaffolds for bone tissue engineering. Journal of biomechanics, 42(3), 257–260. https://doi.org/10.1016/j.jbiomech.2008.10.030
- Pizzichelli, G., Kehlet, B., Evju, Ø., Martin, B., Rognes, M., Mardal, K., & Sinibaldi, E. (2021). Numerical study of intrathecal drug delivery to a permeable spinal cord: effect of catheter position and angle. Retrieved 16 October 2021, from https://doi.org/10.1080/10255842.2017.1393805
- Roe, P. L. (1986). Characteristic-based schemes for the Euler equations. Annual review of fluid mechanics, 18(1), 337-365.
- Seeni, A., Rajendran, P., & Mamat, H. (2019). A CFD Mesh Independent Solution Technique for Low Reynolds Number Propeller. CFD Letters, 11(10), 15-30.
- Shi, C., Lu, N., Qin, Y., Liu, M., Li, H., & Li, H. (2021). Study on mechanical properties and permeability of elliptical porous scaffold based on the SLM manufactured medical Ti6Al4V. PLOS ONE, 16(3), e0247764. https://doi.org/10.1371/journal.pone.0247764
- Singh, S., Shukla, M., & Srivastava, R. (2018). Lattice Modeling and CFD Simulation for Prediction of Permeability in Porous Scaffolds. Materials Today: Proceedings, 5(9), 18879-18886. doi: 10.1016/j.matpr.2018.06.236
- Truscello, S., Kerckhofs, G., Van Bael, S., Pyka, G., Schrooten, J., & Van Oosterwyck, H. (2012). Prediction of permeability of regular scaffolds for skeletal tissue engineering: A combined computational and experimental study. Acta Biomaterialia, 8(4), 1648-1658. doi: 10.1016/j.actbio.2011.12.021
- Venton, J., Bouyagoub, S., Harris, P., & Phillips, G. (2017). Deriving Spinal Cord Permeability and Porosity Using Diffusion-Weighted MRI Data. Poromechanics VI. doi: 10.1061/9780784480779.180
- Zhang, S., Vijayavenkataraman, S., Lu, W. F., & Fuh, J. Y. (2019). A review on the use of computational methods to characterize, design, and optimize tissue engineering scaffolds, with a potential in 3D printing fabrication. Journal of Biomedical Materials Research Part B: Applied Biomaterials, 107(5), 1329-1351.
PVFP-5
-
Abraham, M. J., Murtola, T., Schulz, R., Páll, S., Smith, J. C., Hess, B., & Lindahl, E. (2015). GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX, 1-2(C), 19-25. doi:10.1016/j.softx.2015.06.001
-
Agarwala, R., Barrett, T., Beck, J., Benson, D. A., Bollin, C., Bolton, E. et al. (2018). Database resources of the national center for biotechnology information. Nucleic Acids Research, 46(D1), D8-D13. doi:10.1093/nar/gkx1095
-
Bateman, A., Martin, M., Magrane, M., Agivetova, R., Ahmad, S., Alpi, E. et al. (2021). UniProt: The universal protein knowledgebase in 2021. Nucleic Acids Research, 49(D1), D480-D489. doi:10.1093/nar/gkaa1100
-
Chen, V. B., Arendall III, W. B., Headd, J. J., Keedy, D. A., Immormino, R. M., Kapral, G. J. et al. (2010). MolProbity: All-atom structure validation for macromolecular crystallography. Acta Crystallographica. Section D, Biological Crystallography., 66(1), 12-21. doi:10.1107/S0907444909042073
-
Dong, L., Anderson, A. J., & Malkowski, M. G. (2019). Arg-513 and leu-531 are key residues governing time-dependent inhibition of Cyclooxygenase-2 by aspirin and celebrex. Biochemistry (Easton), 58(38), 3990-4002. doi:10.1021/acs.biochem.9b00659
-
Edgar, R. C. (2004). MUSCLE: Multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research, 32(5), 1792-1797. doi:10.1093/nar/gkh340
-
Janson, G., & Paiardini, A. (2021). PyMod 3: A complete suite for structural bioinformatics in PyMOL. Bioinformatics, 37(10), 1471-1472. doi:10.1093/bioinformatics/btaa849
-
Mistry, J., Chuguransky, S., Williams, L., Qureshi, M., Salazar, G., Sonnhammer, E. L. L. et al. (2021). Pfam: The protein families database in 2021. Nucleic Acids Research, 49(D1), D412-D419. doi:10.1093/nar/gkaa913
-
Rao, Z., Handford, P., Mayhew, M., Knott, V., Brownlee, G. G., & StuartZ, D. (1995). The structure of a Ca2+-binding epidermal growth factor-like domain: Its role in protein-protein interactions. Cell, 82(1), 131-141. doi:10.1016/0092-8674(95)90059-4
-
Santonocito, R., Venturella, F., Dal Piaz, F., Morando, M. A., Provenzano, A., Rao, E. et al. (2019). Recombinant mussel protein pvfp-5β: A potential tissue bioadhesive. The Journal of Biological Chemistry, 294(34), 12826-12835. doi:10.1074/jbc.RA119.009531
-
Schrödinger, L.The PyMOL molecular graphics system, version 2.7 [computer software]
-
Sievers, F., Wilm, A., Dineen, D., Gibson, T. J., Karplus, K., Li, W. et al. (2011). Fast, scalable generation of high‐quality protein multiple sequence alignments using clustal omega. Molecular Systems Biology, 7(1), 539-n/a. doi:10.1038/msb.2011.75
-
Webb, B., & Sali, A. (2021). Comparative protein structure modeling using MODELLER. Current Protocols in Bioinformatics, 54(1), 5.6.1-5.6.37. doi:10.1002/cpbi.3
-
Weisshuhn, P., Sheppard, D., Taylor, P., Whiteman, P., Lea, S., Handford, P., & Redfield, C. (2016). Non-linear and flexible regions of the human Notch1 extracellular domain revealed by high-resolution structural studies. Structure (London), 24(4), 555-566. doi:10.1016/j.str.2016.02.010
