Silk Production

The two major engineering challenges defining our project concern the synthesis of spidroin-derived proteins and binding graphene to the resultant fibres. Silk synthesis through the use of spider farms poses a variety of issues, from the associated ethical issues, to the likelihood of disease spread across an effective monoculture of spiders ^[1].

Therefore, deploying synthetic biology techniques to produce spider silk-like proteins in a microbial chassis would augment the efficiency and yield potential for a spider silk production operation. To maximise the ease of culturing and genetic engineering, we chose E. coli as the chassis, given its status as the most well-characterised model organism for use as a biomolecule production platform ^[2]. However, E. coli is limited by both size and absence of organelles capable of performing post-translational modifications required for folding spidroins into their final silk form^[3].

Our selected spider silk proteins (spidroins) are major ampullate silk proteins, MaSp1^[4] and MaSp2^[5] Both protein types comprise of repeated poly-alanine and glycine rich regions. Hydrophobic interactions between poly-alanine regions and hydrophilic interactions between glycine-rich regions confer the strength and elasticity of dragline silk respectively^[6]. The non-repetitive N- and C-terminal domains of these proteins are suggested to contribute towards protein solubility and self-assembly in the abdominal glands of spiders, where proteins are exposed to a pH gradient as they are drawn through an abdominal duct^[7]. Following the protocol for spontaneous fibre assembly from MaSps outlined in^[4], which occurs in the absence of a pH gradient, the N-terminal domains, which only contribute to fibre assembly via interactions with such a gradient, would be non-functional, so was removed from all MaSp constructs

Results reported in^[4] show the number of repeated poly-alanine and glycine rich regions controls fibre length, ranging from millimetre to metre scale. Results further show that removing the C-terminal domain causes formation of an amorphous aggregate of short fibres. Thus, MaSp constructs were designed containing various numbers of repeated regions, with and without C-terminal domains, with an aim to generate fibres of a range of structures, to explore the resultant differences in their mechanical properties. To increase the feasibility of target spidroin expression, all designed parts contained only a small number of repeated regions relative to the native sequences for the corresponding proteins.

Having reduced their size to a more appropriate length, the MaSp genes still posed a significant challenge, due to the poly-alanine regions present. These regions, encoded by GC-rich codons, resulted in a very high GC content for our genetic constructs, introducing associated complications arising from mis-priming or mis-annealing during genetic synthesis ^[8]. Thus, we were limited to synthesising the six MaSp constructs shown in the table below. Moreover, the poly-alanine regions and associated high GC content represent a translational issue, given the resultant abnormal codon usage. This places a strain on E. coli both to supply the required alanine as well as the corresponding tRNAs necessary for shuttling to ribosomes.

To determine the potential impact of poly-alanine regions on translational efficiency in our chosen chassis, we modelled MaSp synthesis as a function of the ability of E. coli to take up alanine from a surrounding growth medium and combine it with the corresponding tRNA (to form alanyl-tRNA), ready to be shuttled to ribosomes and incorporated into MaSp constructs (as shown in greater detail on the modelling page). Results indicated that whilst significant MaSp yield could be obtained using an unmodified base strain of E. coli (Rosetta DE3), availability of alanyl-tRNA would significantly limit maximum potential yield. Thus, we additionally modelled the impact of expressing CycA, an inner-membrane non-polar amino acid transport protein capable of uptaking alanine via a proton symport mechanism ^[11] and alanyl-tRNA synthetase, an enzyme which catalyses binding of alanine to its corresponding tRNA ^[12], as well as over-expressing the alanine tRNA. Results indicated that overexpression of tRNA elevates MaSp yield. Additionally co-expressing CycA and the alanyl-tRNA synthetase further augments MaSp yield. Modelling results informed further plasmid design, incorporating CycA, alanyl-tRNA synthetase and alanine tRNA as a 'helper plasmid' to be co-expressed alongside MaSp constructs.

To further mitigate the translational strain of MaSp production on the host and maximise overall yield, the Rosetta DE3 strain was selected. A BL21 derivative, Rosetta is deficient in the Ion and OmpT proteases, reducing degradation of heterologous proteins (including our recombinant MaSp proteins) ^[13]. Moreover, Rosetta overexpresses low usage frequency codons for amino acids including glycine, reducing the likelihood of translational errors arising from unavailability of glycine aminoacyl-tRNAs at the ribosome.

Finally, various existing processes for spider silk production involve an electrospinning component, whereby a spidroin protein ‘dope’ is pulled through a pH gradient, applying a sheer force and drawing the protein out into fibres ^[9]. This process can be entirely eliminated, increasing both the simplicity and cost of silk production, via the fusion of MaSp constructs with TrxA (iGEM part K3619001). Linkage of proteins to TrxA increases solubility of MaSp proteins in the E. coli cytoplasm and reduces likelihood of the formation of inclusion bodies ^[14] (which form when misfolded proteins aggregate in the cytoplasm due to overexpression of protein and associated translational errors ^[15]). Thus, synthesised MaSp proteins, fused to TrxA with a thrombin cleavage site between them, remain in solution after harvesting via cell lysis. Then, when thrombin is added and the solubility tag cleaved off, fibres spontaneously form, of strength comparable to native spider silk ^[4]. The overall workflow for silk fibre synthesis, and subsequent graphene binding, is shown in figure 2.

To assess the viability of our design outlined above, we attempted to separately transform into E. coli CycA, alanine tRNAs and the 6 MaSp constructs, both with and without the fused solubility tag TrxA. We were able to sucessfully transform and sequence-verify CycA, 4 alanine tRNAs (2 derived from humans, 2 from E. coli) MaSp constructs 1, 4 and 5 without fused TrxA and constructs 2, 3, 5 and 6 with fused TrxA. As is detailed on the results page, CycA characterisation indicated that its expression reduced net uptake of alanine, most likely due to the protein being non-functional. Whilst an assay was constructed to verify over-expression of the alanine tRNAs, false positive results showed the assay itself was non-functional and we did not have time to repeat the experiment. Therefore, we have not established whether alanine tRNA overexpression was achieved. Finally, no transformed MaSp constructs were seen in Western blotting so their expression cannot be confirmed.

Figure 2 - Schematic of the Workflow for the Production of Spidroin Fibres and Binding of RGO

CAD modelling

Hand model

In hopes of visualising the tensile strength and elasticity of our synthetic silk produced, we created a CAD model of a hand using the open-source software Blender. By binding the silk to the joints of the individual fingers, it is supposed to replicate the function of a tendon in a simplified way. For this purpose, we added pulleys according to where tendons are naturally fixated in place in the human anatomical structure ^[16]. As the thumb specifically only has three pulleys, which doesn't quite align with the way the synthetic silk will run, additional pulleys have been added to allow for the silk to exert the force in the right places. Realistically, tendons are further held in the right places by skin and muscles, which is not present in the CAD model, henceforth justifying the addition of further pulleys, even though they will slightly divert away from the anatomically correct structure of a hand.

Joints are mimicked by a hole-and-stick mechanism, where a 3D printed stick will run through holes at the location of the joitns to allow for movement. In addition, caps have been added at the back of the hand on each joint to limit the range of movement, mimicking the realistic movements of a hand where the phalanges are unable to bend backwards to an extreme degree.

To allow for the bending of individual fingers, all silk strings have their starting points at the tips of the fingers and will be fixed along the fingers down to the wrist of the hand by pulleys, where they all run into a single large pulley. As we were unable to synthesise synthetic silk due to time restraints on our wet lab, sewing threads were instead used to mimic tendons.

Please contact us if you would like to get the STL files of the hand for your own use.

Suggested Future Improvements

Beads could be fixed to the ends of the strings to both add weight to them and to allow for easy separation and control of the individual threads and fingers. To further prevent the beads from freely moving and thereby running the risk of tangling up, the beads are suggested to be magnetic, for which a corresponding metal plate is inserted in the arm part of the hand. This allows for the beads to stay ordered and to further bend the fingers at different angles.

As the silk threads won’t be working against any opposing forces with the current setups of the CAD hand, elastic strings are further recommended to be fixed to the back of the hand. This allows for both the testing and visualisation of the tensile strength of the tendons, as well as conveniently for the hand to stay in different positions.

References

[1] - Xu J et al. Mass spider silk production through targeted gene replacement in Bombyx mori. Applied Biological Sciences. 2018;115(35): 8757-8762. doi:10.1073/pnas.180680511