Following Phase I, the KCL iGEM 2021 team primarily focused on further developing Renervate, and producing a proof of concept for our novel therapy involving a 3D-printed polycaprolactone (PCL) scaffold coated in a mussel-foot protein (MFP) bioadhesive. The main collective aim is to promote axonal regeneration in the spinal microenvironment after a spinal cord injury (SCI). Furthermore, to build upon Phase I, we have introduced the use of the Chondroitinase ABC (ChABC) enzyme in the form of microinjections to degrade chondroitin sulfate proteoglycans - inhibitory molecules that prevent axonal regeneration (Tran, Warren and Silver, 2021).
Unfortunately, the therapeutic nature of Renervate Therapeutics implies a limited validation of our project. In order to prove the holistic success of our therapy, we would need access to a spinal environment and an extended timescale to test our protocols (from successful axonal sprouting to biodegradability). Nevertheless, below we document the work achieved, to the best of our ability, over the course of this project.
As we were fortunate enough to gain access to the laboratory this summer - albeit limited due to the restrictions imposed by the COVID-19 pandemic- we were able to build upon our Phase I research and work towards a proof of concept study for Reneverate Therapeutics. Last year, we designed preliminary protocols for the expression and purification of PVFP-5 alongside protocols to study its cytotoxicity and adhesive strength. The majority of our Phase I protocols were based and built around “Recombinant mussel protein PVFP-5β: A potential tissue bioadhesive” - a paper by our Phase I collaborators, and team instructors and advisors: Professor Annalisa Pastore and Dr Caterina Alfano.
An overarching proof of concept study for Renervate Therapeutics would be the utilisation of a 3D printed PCL scaffold coated in our MFP - PVFP-5 - to allow axonal regeneration in the spinal microenvironment. This would be coupled with microinjections of our thermostabilised ChABC which CSPGs: inhibitory molecules which are expressed around the glial scar in spinal cord injuries (Tran, Warren and Silver, 2021). Unfortunately, the clinical and therapeutic nature of Renervate Therapeutics limits the scope in which our research can be conducted. Issues in simulating a reliable spinal cord microenvironment in vivo , and difficulties with incorporating a 3D printed scaffold to demonstrate axonal regrowth, have resulted in a systemic proof of concept which is difficult to achieve. Due to their tendency to fold into inclusion bodies, we have instead focused on proving that our PVFP-5 and ChABC proteins can be successfully synthesised recombinantly in a soluble form. We have also conducted a functional enzymatic assay to prove that our in silico mutations to confer thermostabilisation to ChABC at physiological temperatures have been successful. Lastly, we have developed proof that the architecture of our PCL scaffold can feasibly be 3D printed.
While our therapy provides the structural support for axonal regrowth, we need to combat the inhibitory microenvironment that prevents these damaged axons from reaching the scaffold. This is the role ChABC plays in our therapy, however, the wild-type 1HN0 ChABC (Huang et al., 2003) is not thermostable at physiological temperature. Therefore, we introduced mutations to the sequence to increase stability consequently enabling its use as part of our therapy. COVID-19 has limited our access to the lab so, instead of performing random mutagenesis ourselves, we utilised 8 mutations outlined in Dr. Hettiaratchi’s paper which conferred increased thermostability.
Before using this mutated ChABC in our experimental work, we needed to ensure that the thermostability did - in fact - increase. This was completed in collaboration with Phystech Moscow through a molecular dynamics simulation that they ran. The simulation was done on the wild-type sequence and on our mutated sequence to compare thermostability at physiological temperatures. They used GROMACS with the forcefield “c36m”. The simulation was done in a waterbox with a 0.15M NaCl concentration at a temperature of 310.15 K. Overall, we found that the mutations did actually increase the stability of the enzyme which allowed us to confidently synthesise our ChABC sequence for experimental wet lab work. Documentation regarding our simulations can be found on our model page
Following positive results from our molecular dynamics simulations to determine thermostability at 37 degrees Celsius of our mutated ChABC protein sequence, we decided to utilise the mutated protein sequence during synthesis of our parts. Through IDT, we synthesised our DNA sequences as composite parts directly (ChABC composite part: BBa_K3794005 (Figure 1)). This composite part (BBa_K3794005) is composed of an upstream LacI regulator (BBa_R0010), the ChABC coding sequence (CDS) with a 5’ 6xHis tag and TEV cleavage site, and a 3’ T1 Terminator (BBa_B0010).
Figure 1 : ChABC composite part: BBa_K3794005.
Post-synthesis of our DNA sequences by IDT, we conducted a PCR to confirm the correct sequences were synthesised and also amplified. Our PCR products can be seen in Figure 2:
Figure 2 : PCR Products of BBa_K3794005 (lane 3 and 4).
Since we are working to develop a novel therapy, our extensive literature review revealed a lack of information and data availability in pre-existing scientific findings to base our project on. Consequently, we started studying and analysing our protein of interest. During Phase I, we decided to implement the use of PVFP-5β - a mussel foot protein mostly researched by Santonocito et al. - with whom we established a collaboration. However, due to the ongoing COVID-19 pandemic, our partnership was interrupted and we decided to change our sequence to PVFP-5 (UniProt U5Y3S6).
We then used the PVFP-5 UniProt U5Y3S6 (Guerette et al., 2013) sequence to design a new structural model using a combination of homology modelling and molecular dynamics through GROMACS. The structure of our protein was confirmed by our human practice outreaches, such as Dr Andrew Beavil. Validation also came from AlphaFold2 (Jumper et al., 2021), which demonstrated correct folding of the polypeptide. In fact, the latter software proved that all 9 of our predicted disulphide bridges had formed, to stabilise the protein.
Following the update to our PVFP-5 protein sequence from last year, we used IDT for synthesis of our PVFP-5 translational unit as a composite part (BBa_3794001). Similar to ChABC, an upstream LacI regulator (BBa_R0010) was synthesised alongside a 6xHis+TEV tagged PVFP-5 CDS. Our PVFP-5 composite part (BBa_K3794001) can be seen in Figure 3.
Figure 3 : PVFP-5 composite part: BBa_K3794001.
After synthesis of our Tyrosinase composite part (BBa_K3794003) (Figure 4), we conducted PCR on BBa_K3794001 and BBa_K3794003 to confirm proof of correct synthesis and amplification of DNA. Our PCR products for BBa_K3794001 and BBa_K3794003 can be seen in Figure 5.
Figure 4 : Tyrosinase composite part: BBa_K3794003
Figure 5 : PCR Products for BBa_K3794001 (lane 7) and BBa_K3794003 (lane 3 and 5) BBa_K3794003 PCR product is at 900bp band.
Our next step was to predict successful adherence between our mussel foot protein and the PCL scaffold; we discovered that this interaction had never been studied before in therapeutic applications. Therefore, the first means to achieve this was through studying the chemistry behind the individual compounds. We determined that hydrogen bonding would be very likely to occur between the DOPA residue of PVFP-5 due to the presence of diols and the oxygen of ester units in PCL. Our prediction was confirmed by Professor Herbert Waite from UC Santa Barbara and Dr Sarah Barry from King’s College London. Upon validation, we proceeded to showcase our findings in the form of a PyMOL video.
To further validate our prediction, we devised a calculator capable of approximating the quantity of MFP needed to coat a given area of PCL.
Before conducting an enzymatic assay for our thermostabilised ChABC (BBa_K3794005) - as part of our overarching proof of concept study - we needed to prove we are able to express it in a purified, soluble form at a respectable yield. After digesting and ligating BBa_K3794005 into pSB1A3, and transformation into competent BL21 (DE3) E.coli cells, we used IPTG-induction for protein expression at 37 degrees Celsius. The results from our protein expression can be seen below in Figure 6. Lanes 2 - 5 display ChABC expression at 0h induction, 3h induction (lanes 2 and 3, respectively), and the total cell lysate and soluble fraction of the cell culture (lanes 4 and 5, respectively). The SDS-PAGE gel of our ChABC expression results did not provide conclusive outcomes concerning the expression of ChABC in either the insoluble or soluble form.
Figure 6 : SDS-PAGE gel of E.coli cell samples taken at various points during protein expression of thermostable ChABC. Lane 1: ThermoFisher Prestained Protein Ladder. Lane 2: ChABC 0h Induction. Lane 3: ChABC 3h induction Lane 4: Total cell lysate. Lane 5: Soluble fraction. Lane 6 - 9: BBa_K3794003 samples.
Towards demonstrating the presence of ChABC, we conducted protein purification via a Ni-NTA spin column (Figure 7) and subsequent Western Blot (Figure 8) using a conjugated anti-His antibody. As seen below, despite confirming the presence of our thermostabilised ChABC composite part (BBa_K3794005), our protein yield remained considerably low.
Figure 7 : SDS-PAGE gel resulting from protein purification of expressed ChABC (BBa_K3794005). Lane 1: ThermoFisher Prestained Protein Ladder. Lane 2: Cell-lysate pellet resuspended with 6M GuHCl. Lane 3: Elution 1. Lane 4: Elution 2. Lane 5: Elution 3. Lane 6: Elution 4.
Figure 8 : Western blot of stained 6xHis tagged thermostable ChABC protein from BBa_K3794005. Lane 1: ThermoFisher Prestained Protein Ladder. Lane 2: Cell-lysate pellet resuspended with 6M GuHCl. Lane 3: Elution 1. Lane 4: Elution 2. Lane 5: Elution 3. Lane 6: Elution 4.
Therefore, in an effort to improve the solubility of ChABC, we attempted to express it in a different cell line (BL21). Following re-expression using BL21, we ran an SDS-PAGE gel to prove that it did improve the solubility. The results are shown below.
Figure 9 : SDS-PAGE gel of E.coli cell samples taken during expression of thermostable ChABC. Lane 1: ThermoFisher Prestained Protein Ladder. Lane 2: Total cell lysate 0h induction. Lane 3: Soluble fraction 0h induction. Lane 4: Total cell lysate 1h induction. Lane 5: Soluble fraction 1h induction. Lane 6: Total cell lysate 2h induction. Lane 7: Soluble fraction 2h induction. Lane 8: Total cell lysate 3h induction. Lane 9: Soluble fraction 3h induction.
While using BL21 did improve the yield of our protein considerably compared to our previous expression (Figure 6), our protein remained largely in the insoluble fraction of the cell culture. One main cause of this is the sheer size of our thermostabilized ChABC part (BBa_K3794005). It is over 3000bp long and ~115kDa large in size. Hence, it is a very large protein that tends to aggregate into inclusion bodies, and is therefore not as present in the soluble fraction relative to the insoluble fraction.
Following our expression, we experimented with our purification step. With the introduction of 1% Triton X-100 in our lysis buffer, we hoped it would improve the extraction and solubilisation of our expressed protein derived from BBa_K3794005. After a second protein purification - using a Ni-NTA resin rather than a spin column - we analysed the results using a SDS-PAGE (Figure 10).
Figure 10 : SDS-PAGE gel resulting from protein purification of expressed ChABC. Lane 1: Promega Broad Range Protein Molecular Weight Markers. Lane 2: Soluble fraction of cell-lysate pellet. Lane 3: Flow-through. Lane 4: Wash 1. Lane 5: Wash 2. Lane 6: Wash 3. Lane 7: Elution 1. Lane 8: Elution 2. Lane 9: Elution 3. Lane 10: Elution 4.
Having made various attempts to solubilise our thermostabilised ChABC, and optimise expression of BBa_K3794005, we can concentrate our bottleneck at the purification step. As seen in Figure 9, we are able to express a sufficient amount of protein - albeit in an insoluble form. It is the purification steps where our yields decrease which may be due to a variety of reasons: our protein derived from BBa_K3794005 is too large and cannot be sufficiently accommodate on the Ni-NTA resin, or the mutations we have introduced to confer thermostability may affect its solubility. In the future, we will continue to investigate the optimisation of BBa_K3794005 expression and purification to allow scaling up of synthesis for this enzyme in our therapy.
Having completed the structural modelling of our new PVFP-5 sequence - and confirming the 9 disulphide bonds we had predicted via AlphaFold2 (Jumper et al., 2021) - we synthesised our translational unit for expression of PVFP-5 as a composite part (BBa_K3794001). With the knowledge that we have a high disulphide bond content in our protein structure, we designed an expression system in which we utilised E.coli strains that have been optimised for the formation of disulphide bonds e.g. Rosetta-Gami B (DE3).
Similar to ChABC/BBa_K3794005, our proof of concept was focused on providing evidence that PVFP-5, BBa_K3794001, can be expressed and purified in its soluble form. Current literature has largely taken an approach to denature, and later refold PVFP-5 and similar analogues during the purification steps.
BBa_K3794001 was expressed in E.coli BL21 (DE3) and Rosetta-Gami B (DE3). Both these cell lines were used to compare which cell line is most optimal for synthesis of PVFP-5 in its soluble form. Following expression in these two cell lines with IPTG-induction at 37 degrees Celsius, we conducted SDS-PAGE analysis to visualise the results of our expression (Figure 11).
Figure 11 : SDS-PAGE gel of E.coli cell samples taken at various points during protein expression of PVFP-5 (BBa_K3794001). Lane 1: ThermoFisher Prestained Protein Ladder. Lane 2: PVFP-5(Rosetta-gami B) 0h induction Lane 3: PVFP-5 (Rosetta-gami B) 3h induction Lane 4: PVFP-5 (Rosetta-gami B) Total cell lysate Lane 5: PVFP-5 (Rosetta-gami B) Soluble fraction Lane 6: PVFP-5 (BL21 DE3) 0h induction Lane 7: PVFP-5(BL21 DE3) 3h induction Lane 8: PVFP-5(BL21 DE3) Total cell lysate Lane 9: PVFP-5(BL21 DE3) Soluble fraction.
As seen in Figure 11 - around the 15kDa band - we had suspected strong expression of PVFP-5 in both BL21 (DE3) and Rosetta-Gami B (DE3) due to the strong bands present in Lanes 4 and 5 (Rosetta-Gami B (DE3)), and 8 and 9 (BL21 (DE3)). This is because our PVFP-5 protein is ~16kDa. However, our suspected expressed PVFP-5 was largely found in the insoluble fraction/total cell lysate in the form of inclusion bodies or aggregates. Nevertheless, we could not be sure of this expression prior to purification. Following this, we conducted a Ni-NTA spin column purification to determine if these bands were, in fact, PVFP-5 derived from the expression of BBa_K3794001.
After protein purification, we ran an SDS-PAGE gel to see the results of our purification. The yield of our protein was very low in our eluted samples but, unfortunately, we were not able to take a photo of this SDS-PAGE gel. Despite having strong bands in our SDS-PAGE of expression samples, we were not able to isolate and purify our protein. We suspect that this is due to the intrinsic adhesive nature of PVFP-5 which meant that it was adhering to E.coli proteins and the membrane leading to aggregation.
To further our proof of concept in optimising the expression of PVFP-5/BBa_K3794001, we experimented with another E.coli cell line which has been optimised for disulphide bond formation: SHuffle (DE3). Following the expression of BBa_K3794001 in SHuffle (DE3) with IPTG-induction at 37 degrees Celsius, we ran the results from our expression on an SDS-PAGE gel (Figure 12).
Figure 12 : SDS-PAGE gel of E.coli cell samples taken during expression of PVFP-5 in SHuffle (DE3)(BBa_K3794001). Lane 1: ThermoFisher Prestained Protein Ladder. Lane 2: Total cell lysate 0h induction. Lane 3: Soluble fraction 0h induction. Lane 4: Total cell lysate 1h induction. Lane 5: Soluble fraction 1h induction. Lane 6: Total cell lysate 2h induction. Lane 7: Soluble fraction 2h induction. Lane 8: Total cell lysate 3h induction. Lane 9: Soluble fraction 3h induction.
As seen in Figure 12, we had suspected PVFP-5 expression around the 15kDa band. However, it was - again - largely found in the form of aggregates/in the total cell lysate. This could, as aforementioned, be attributed to the adhesion nature of PVFP-5 sticking to various E.coli proteins and membranes thus preventing it from being synthesised in the soluble form.
To confirm successful expression of BBa_K3794001 into a soluble form of PVFP-5, we conducted a protein purification once again. This time, however, with 1% Triton X-100 in an attempt to solubilise PVFP-5 from the components of the E.coli cell-lysate. Following purification via a Ni-NTA resin, we ran an SDS-PAGE gel to visualise the results of our purification (Figure 13).
Figure 13 : SDS-PAGE gel of E.coli cell samples taken during expression of PVFP-5 in SHuffle (DE3) (BBa_K3794001). Lane 1: Promega Broad Range Protein Molecular Weight Markers. Lane 2: Soluble fraction of cell-lysate pellet. Lane 3: Flow-through. Lane 4: Wash 1. Lane 5: Wash 2. Lane 6: Wash 3. Lane 7: Elution 1. Lane 8: Elution 2. Lane 9: Elution 3. Lane 10: Elution 4.
Following analysis of Figure 13, we learnt that - irrespective of our efforts - we could not significantly improve the yield of our purified protein . Despite this, the team at KCL iGEM 2021 have completed our proof of concept and demonstrated that PVFP-5, BBa_K3794001, can be recombinantly expressed in E.coli in its soluble form without having to refold its structure: a task that has not typically been achieved in wider research.
Despite having a low yield of purified ChABC following our updated protein expression of BBa_K3794005, we attempted to conduct an enzymatic assay at physiological temperatures as part of our proof of concept to demonstrate that our purified ChABC, derived from BBa_K3794005 is functional.
After concentrating ChABC from its elution samples from protein purification, an enzymatic assay at 37°C measuring absorbance at 232nm was conducted 10 minutes after mixing ChABC with the substrate chondroitin sulphate, using the BMG LabTech FluoSTAR OMEGA 96-well plate reader.
There were four simultaneous reactions running, each conducted in triplicate. The wells were set out as follows:
Table 1: Section of 96-well plate that ChABC enzymatic assay was conducted in.
(A1 - A3): 100uL Buffer A (Tris-HCl, 15mM, pH 8.8)
(B1 - B3): 20uL chondroitin-sulphate sodium salt (3mg/mL) + 80uL Buffer A
(C1 - C3) 20uL ChABC (0.014mg/mL) + 80uL Buffer A
(D1 - D3) 20uL chondroitin-sulphate (3mg/mL) + 20uL ChABC (0.014mg/mL) + 60uL Buffer A
(E1 - E3) Empty wells
The results from our enzymatic assay can be seen below (Figure 14):
Figure 14 : ChABC enzymatic assay absorbance graph.
When ChABC is exposed to chondroitin sulfate it engages in a hydrolysis reaction of the sugar polymer present in the compound, resulting in the formation of a C4-C5 alkene. This C4-C5 alkene forms a weak chromophore detectable by spectrophotometry (Clemons et al., 2013). The average absorbance of both the chondroitin-sulphate-sodium salt (0.130 nm) and the thermostabilized ChABC (0.107 nm) are both below the average absorbance of combined chondroitin-sulphate sodium salt and thermostabilized ChABC combined (0.157 nm). The increase in absorbance can be attributed to an increased concentration of glycosidic alkenes which indicates successful activity of our enzyme. In the future, we aim to conduct an assay with a thermostabilished ChABC enzyme with a higher concentration and purity, combined with increased assay repeats to demonstrate statistical significance of our results.
We gained access to Professor Trevor Coward’s lab at Guy’s and St Thomas’ Hospitals which allowed us to use the 3D printers to print a prototype. Having redesigned our scaffold to a cross-hatch design, we sent our STL file into the lab to be sliced. This was then processed by the Cubicon printer and printed. However, initial attempts proved unsuccessful. The PCL material was melting at a much lower temperature than expected and this led to the strands of material fusing between layers. After changing the printing parameters, the scaffold was printed successfully with a printing time of 90 minutes.
Figure 15 : PCL scaffold being printed in the Cubicon printer.