Team:IISc-Bangalore/Engineering
Engineering
In this page, we will introduce the Engineering Design Cycle that we followed in order to eventually accomplish our short-term goals. We followed this workflow for all four components that constitute CellOPHane: protein production, bacterial cellulose (BC) production, MCC sheet production, and hardware development. During this process, successful results were obtained, but also new challenges were discovered. Nevertheless, the drawbacks forced us to come up with new ideas and initiated new cycles of engineering, ultimately driving our project forward. While the exact experiments and results are described on the Results page, here we outline the underlying thought process that led to our final engineering success.
Synthetic biology is a discipline of science that differs a lot from the others. Usually, one would use the scientific method to obtain results in the majority of life sciences, centering one 's work on making observations and performing experiments. However, SynBio goes beyond that! It is all about constructing something new from parts (biological parts if you may). That is why, sometimes the engineering design process fits better with these projects, since a product has to be built or a "genetically engineered machine" has to perform a specific task. It reminds us that iGEM stands for International Genetically Engineered Machines.
Proteins
Cycle 1
After long search through literature, we selected dCBD and OpdA. The former was selected for its strong, synergistic binding with microcrystalline cellulose. The latter was selected for its broad substrate range that includes a couple of extremely hazardous pesticides which are heavily used in Indian agricultural fields. We chose sfGFP as the perfect reporter for our purpose.
Initially we tried to incorporate SpyDock in fusion protein C1 and SpyTag002 in fusion protein C2. Maintaining modularity was our first and foremost priority while designing the filter platform, and we planned to utilize noncovalent interaction in binding C1 with C2. SpyDock is a modified version of SpyCatcher which binds in 1:1 stoichiometry with SpyTag002 via noncovalent interaction, and not irreversible isopeptide linkage. SpyTag/SpyDock system is at the heart of Spy&Go resin that is used for protein purification. We planned to functionalize the bacterial cellulose (BC) sheet by incubating it with cocktail of C1 (SpyDock-dCBD) and C2 (OpdA-SpyTag002-sfGFP). After the functionalized filter gets installed (as a part of the associated hardware), the BC base of the filter would eventually get worn out with time. Thereafter, we can just flow 2.5M imidazole solution through our filter. It would elute out the C2 molecules from the filter, which can be purified and reused later. Thus, adopting this strategy would enable one of our two fusion proteins to be reusable. Most importantly, C2 being larger than C1 we hoped that this strategy would make our filtration process more economically viable.
Before stepping into wet lab work, we designed a first-order mathematical model on SpyTag/SpyDock interaction. We searched for parametric values required to run this model and fed those to it.
Upon running computer simulations, the model yielded plots regarding the extent of SpyTag/SpyDock complex dissociation at various volumetric flow rates.
The model showed that SpyTag/SpyDock interaction gets abolished after a short period of time and, as a result, the process of C2 getting washed away has a very low lifetime. Though we hoped that the noncovalent interaction-based strategy would decrease the cost as well as would make C2 reusable, we learnt from the model this strategy would take us nowhere. Thus, the model safely ruled out the incorporation of SpyTag/SpyDock into the constructs we would design next.
Cycle 2
The previous cycle made our focus shift to covalent protein ligation strategies instead. The SpyTag/SpyCatcher system readily attracted our attention. The interaction between the two is covalent, which means the binding is essentially irreversible. The reaction requires only mixing and is rapid, high yielding, robust to experimental conditions, and shows very good specificity. The system is functional over a wide range of temperature, pH and salinity. SpyTag002 and SpyCatcher002 were the preferred versions. We planned to incorporate SpyCatcher002 in fusion protein C1 and SpyTag002 in fusion protein C2. We also thought of some back-up plans. We included KpnI restriction sites on either end of the sfGFP CDS, should the need of removing sfGFP ever arises.
We designed two constructs - C1 (SpyCatcher002-dCBD) and C2 (OpdA-SpyTag002-sfGFP). To know the efficiency of our system, both C1 and C2 proteins had to be synthesized, which was carried out using E. coli DH5 alpha as the cloning host and E. coli BL21(DE3) as the expression host. The specific genes for dCBD, OpdA, and sfGFP were derived from T. reesei, A. radiobacter, and A. victoria respectively. We codon optimized the genes for expression in E. coli with SnapGene 's codon optimizing tool. pETMCN-T7 (KanR) was the vector we intended to use in our experiments and it included features (e.g. T7 promoter, lac operator, and ribosomal binding site) that allowed for the inducible gene expression. Additionally, both constructs have a 10x-His tag for subsequent protein purification. We ordered the DNA sequences of both C1 and C2 from GenScript. They were shipped to us after being cloned into pUC57-Kan vectors.
After amplifying the constructs using suitable primers, we cloned each of them into pETMCN-T7 (KanR) using Gibson assembly. It yielded C1_pETMCN-T7 and C2_pETMCN-T7 which were subsequently transformed into E. coli DH5 alpha. Inoculation into liquid culture was followed by miniprepping. Plasmid obtained from positive clones (as per double digestion verification) was transformed into E. coli BL21(DE3) and secondary liquid cultures were induced with different working concentration of IPTG and under a range of temperatures.
SDS-PAGE of lysed cell pellets, conducted as the expression test, was inconclusive. Thus, we performed a western blot using anti-His primary antibody. The results showed prominent bands at ~25 kDa for C1 and at ~75 kDa for C2. It helped us to finalize the optimal induction conditions (namely 0.5 mM IPTG at 30°C) as well.
To perform solubility test, first we followed these conditions to induce a large scale E. coli BL21(DE3) cultures. Then, harvested cells were lysed and subjected to pulldown with Ni-NTA. SDS-PAGE was performed on the boiled samples taken from different culture fractions (viz., lysate, pellet, supernatant, washed and bead-bound). Bead-bound fraction for C1 showed prominent band at (~25 kDa), but C2 did not. Next, we performed two Western blots on the C2 pulldown samples, one using anti-His primary antibody and the other using anti-GFP primary antibody. Later, the pulldown of C2 was performed with an anti-GFP nanobody coupled to GST-tagged glutathione-sepharose beads and it was followed by SDS-PAGE.
Results from the expression test tell that both C1 and C2 are getting expressed by our expression host. No pulldown was completely successful for C2. Most probably, protein degradation was to blame. However, results from solubility test definitely tell that C1 is soluble. C2 might not have been soluble.
Cycle 3
The lack of solubility of C2 possibly stems from undesired protein-protein interaction. Therefore, we decided to make C2 less bulky by clipping out the sfGFP module. As mentioned previously, we already made a provision of doing the same in the design phase of the previous engineering cycle.
We found instances in the literature where other groups had used TB media in the secondary culture for OpdA production.
Restriction digestion of C2_pETMCN-T7 with KpnI followed by ligation yielded C2(2.0)_pETMCN-T7. This was our first modified construct in quest of functional C2 protein.
After verification, the plasmid was transformed into E. coli BL21(DE3) followed by Ni-NTA pulldown. We performed pulldown again using SpyDock-linked Ni-NTA bead.
We grew our expression host in Terrific Broth (TB) in addition to LB (i.e. the media used so far).
We conducted Bradford assay (using bovine serum albumin as the reference) and OpdA activity assay (using Coumaphos as the substrate) on the crude extract of the large-scale secondary culture.
No prominent band at ~50 kDa, the molecular weight of C2(2.0), could be observed in either of the pulldowns. Thus, like C2, C2(2.0) also could not be verified to be soluble.
Enzyme production seemed to be higher in cultures grown in TB than those grown in LB. Thus, we learnt to continue using TB as the growth medium for our protein 's expression.
The OpdA activity assay tells us that OpdA, produced by TB-grown expression host, is not undergoing proteolytic cleavage under the expression conditions provided and is active. This serves as an extremely important aspect of CellOPHane: its Proof of Concept
Cycle 4
The lack of solubility of C2(2.0) might stem from the hydrophobic signal peptide attached to OpdA. Therefore, we decided to clip out the said signal peptide sequence.
We performed a PCR on C2(2.0)_pETMCN-T7 to amplify C2(2.0) minus the N-terminal signal. This PCR fragment was cloned into an empty pETMCN-T7 vector to yield C2(3.0)_pETMCN-T7. This was our second modified construct in quest of functional C2 protein.
As we are currently in the 'Build' phase of this engineering cycle, experiments are not performed yet.
As we are currently in the 'Build' phase of this engineering cycle, the progress on this is not significant enough to give us any conclusive findings.
Bacterial Cellulose Production
Cycle 1
Making bacterial cellulose (BC) sheet is integral to our project. Initially, we planned to achieve BC overproduction by knocking out celG gene in chvB- Agrobacterium tumefaciens C58 kindly provided by Prof. Ann G. Matthysse as a gift. We planned to adopt the FLP-FRT recombination to achieve the same.
Necessary primers were designed and ordered.
With help of gradient PCR, the lambda Red recombinase cassette of pKD46 (AmpR) was amplified and put under tetR promoter in pST-KT (KanR; TetR). Gradient PCR was again used to amplify the complement1-FRT-CamR-FRT-complement2 fragment present on pKD3 (CamR). Here, complement1 and complement2 denote the DNA sequence complementary to upstream and downstream region, respectively, of the genomic portion we intend to knock out.
The chvB- Agrobacterium tumefaciens C58 strain was revived from filter paper into LB agar plates.
The amplification of lambda Red recombinase system was successful, but that of complement1-FRT-CamR-FRT-complement2 was not. Based on the gel electrophoresis imaging, formation of primer dimer (PD) was probably the reason the latter failed. PD is a potential by-product in PCR reaction where two primer molecules hybridize to each other because of strings of complementary bases in them. The DNA polymerase amplifies the PD, leading to competition for PCR reagents, thus potentially inhibiting amplification of the DNA sequence targeted for PCR amplification.
No bacterial growth was found on the LB agar plates where we revived the mutant Agrobacterium strain. As per Prof. Matthysse 's words, the root of this problem might lie in the cultures being very old.
Cycle 2
We realized that we do not have time to re-design the pKD3 primers from scratch or to play with the gradient PCR conditions. Lack of mutant colony was another headache.
Thus, we gave up the knock-out strategy of ours and quickly devised an alternative. We ordered Komagataeibacter xylinus from NCL, Pune. Given appropriate conditions, it has been reported to produce BC in excessive amount.
Neither construction of a DNA sequence nor its implementation into a chassis was involved in this part (i.e. BC production) of our work.
The strain was inoculated from agar slant into liquid Hestrin Schramm (HS) media (pH 6). HS hard agar (HA) plates were streaked with the same liquid culture.
No growth was seen in either liquid culture or HA plates even after sufficient time was allowed for bacterial growth. The HS media was selected after going through available literature on BC overproduction by K. xylinus. However, the instruction sheet provided by NCL recommended sorbitol media with pH 6.2.
Cycle 3
We planned to use sorbitol media (pH 6.2) instead of HS media.
Neither construction of a DNA sequence nor its implementation into a chassis was involved in this part (i.e. BC production) of our work.
The strain was inoculated from agar slant into liquid sorbitol media. Sorbitol hard agar (HA) plates were streaked with the same liquid culture. We made glycerol stocks of the colonies obtained from HA plates.
Sorbitol HA plates exhibited growth but liquid culture did not show any turbidity. Fungal contamination, seen in the liquid culture, was the probable reason.
Cycle 4
Unless we get growth in liquid culture, we would not be able to derive BC. We also knew that HS media, not sorbitol media, is arguably the best at inducing BC overproduction in K. xylinus. Thus, after consulting experts we planned to culture the bacteria in modified HS media of pH 4.5.
Neither construction of a DNA sequence nor its implementation into a chassis was involved in this part (i.e. BC production) of our work.
The strain was inoculated from sorbitol HA plate into liquid modified HS media. HA plates could not be prepared with this modified HS media owing to shortage of time.
The test phase is still going on.
MCC Sheet
Cycle 1
As a back-up plan, we also tried to make a sheet of microcrystalline cellulose (MCC) from the commercially available powdered form. Though it is not directly synthesized by bacteria, such a sheet would help us with accomplishing our Proof of Concept goals. MCC and BC are almost similar in their characteristics.
As a starting point, we tried to find out a suitable solvent for dissolving MCC powder. Based on literature review, we shortlisted dimethyl sulfoxide (DMSO), 7% LiCl in dimethyl acetamide (DMAc), and N-methylpyrrolidinone (NMP) as candidates for further investigation. Drop casting was chosen as the characterization method of choice at the initial phase because it saves wastage of resources.
Neither construction of a DNA sequence nor its implementation into a chassis was involved in this part (i.e. sheet production) of our work.
MCC was mixed with these solvents and stirred overnight. To test whether it is possible to form a mechanically strong film of MCC from the solutions, we performed drop casting (on glass slide) at a small scale with each of the samples and also with different sample volumes.
In general, quenching gave better results compared to annealing. Drops derived from MCC+DMSO solution yielded a cracked surface upon quenching. Drops derived from MCC+NMP solution gave an excessively flaky residue. Sheet derived from MCC+LiCl/DMAc solution showed negligible presence of cracks, but it adsorbed moisture after it had returned to room temperature. The BC sheet we desire to make cannot be flaky. Thus, MCC+NMP solution was not subjected to further investigation.
Cycle 2
We sought to see whether the addition of a binder could influence the mechanical integrity of the film and strengthen it.
Neither construction of a DNA sequence nor its implementation into a chassis was involved in this part (i.e. sheet production) of our work.
We added 10% w/v polyvinyl alcohol (PVA) to MCC+DMSO as well as to MCC+LiCl/DMAc. Then, we performed drop casting (on glass slide) using these solutions, all well above 100°C as done before. One of the drop-casted films obtained from MCC+DMSO was subjected to atomic force microscopy (AFM).
All the drop-casted films turned out to be patchy and unusually brownish. Surprisingly, PVA is known to be strongly preferring the formation of mechanically strong films! However, we found out that the heating regime (above 100°C) used for the samples had exceeded the degradation temperature of PVA. It was the degradation of PVA that had caused the charred appearance of the films.
Usage of vacuum oven was causing the film to come out of the substratum (i.e. glass slide). It was interfering with the proper functioning of AFM.
MCC+LiCl/DMAc+PVA solution yielded highly flaky film, whereas MCC+DMSO+PVA film fairly maintained its integrity. Thus, MCC+LiCl/DMAc+PVA solution was not subjected to further investigation.
Cycle 3
The drop casting, thus, must be performed at a lower temperature.
Also, we ceased to use vacuum oven for drop casting purpose.
Neither construction of a DNA sequence nor its implementation into a chassis was involved in this part (i.e. sheet production) of our work.
We performed drop casting (on glass slide) using MCC+DMSO+PVA below 100°C.
The drop-casted film turned out to be homogeneous, just as we desire. We have thus finalized the optimal solvent, a beneficial additive, and the optimal conditions needed to come up with a homogeneous, crack-less, non-charred MCC sheet.
Hardware
Cycle 1
As the first step towards building a functional hardware as the implementation of our novel strategy, we decided to coat the inner surface of a microchannel with MCC. It would then serve as the proof of concept for a microchannel whose bore has been coated with the functionalized bacterial cellulose (BC) filter we seek to develop. Important aspects of this are the thickness of MCC coating, extent of uniformity in coat thickness along the length of microchannel, durability of the coating etc.
We first planned to fabricate a microchannel of square cross section whose one side is open. The Then, we can pour MCC+DMSO+PVA solution (suits our purpose, as determined by our work on MCC sheet development) on it through the open face. Upon drying, we would get a MCC coating on one of the faces. Lastly, a glass slab is to be used to seal the open face. As glass is transparent, it would help us in probing into the flow characteristics inside the microchannel by using fluorescent beads in future.
However, we anticipated many challenges with this design. First, only one of the faces ultimately happens to have the MCC coating. This means that in the prototype we seek to develop with functionalized BC sheet in place of MCC layer, only one side of the microchannel would host the embedded proteins. Thus, filtration efficiency of the system would decrease as most of the organophosphate molecules simply pass through the microchannel without coming into contact with a protein molecule for even once. Second, fabricating a microchannel in the clean room at requires special expertise and prohibitive amount of resources, given the scarcity of time.
Thus, we changed our plan. We wanted some simpler system as our Proof of Concept. We chose the tubes made up of polydimethylsiloxane (PDMS) whose diameter is in the order of millimetres. These are readily available, cheap, inert, non-toxic, and non-flammable. Being transparent, rheological studies would be quite easy to perform. Since PDMS is almost always cured with some cross-linking agent, it shows time-dependent elastic deformation. Low elastic modulus of PDMS enables these pipes to be flexible (this property might be valuable while designing a working prototype of our project). We tried to coat such pipes on the inner side with MCC.
Neither construction of a DNA sequence nor its implementation into a chassis was involved in this part (i.e. hardware development) of our work.
We collected MCC+DMSO+PVA solution (as described previously) in a syringe and injected that into a 10 cm long PDMS tube of about 5 mm inner diameter. The pipe was left to dry in an orientation horizontal relative to ground.
Even after 48 hours, the solution inside pipe did not dry off. Neither did it settle on the PDMS surface. It happened most probably because the PDMS surface is very hydrophobic, whether MCC+DMSO+PVA solution is less so.
Thus, we planned to modify the PDMS surface in order to increase its hydrophilicity.
References
- Khairil Anuar INA, Banerjee A, Keeble AH et al. Spy&Go Purification of SpyTag-proteins using Pseudo-SpyCatcher to Access an Oligomerization Toolbox. et al>. 2019;10(1734). doi: 10.1038/s41467-019-09678-w
- Hatlem D, Trunk T, Linke D, Leo JC. Catching a SPY: Using the SpyCatcher-SpyTag and Related Systems for Labeling and Localizing Bacterial Proteins. Int. J. Mol. Sci.. 2019 Apr 30;20(9):2129. doi: 10.3390/ijms20092129
- Zakeri B, Fierer JO, Celik E, Chittock EC, Schwarz-Linek U, Moy VT, Howarth M. Peptide Tag Forming a Rapid Covalent Bond to a Protein, through Engineering a Bacterial Adhesin. Proc. Natl. Acad. Sci. USA. 2012 Mar 20;109(12):E690-7. doi: 10.1073/pnas.1115485109
- Datsenko KA, Wanner BL. One-step Inactivation of Chromosomal Genes in Escherichia coli K-12 using PCR Products. Proc. Natl. Acad. Sci. U. S. A. 2000 Jun 6;97(12):6640-5. doi: 10.1073/pnas.120163297
- 5. Zhang C, Liu R, Xiang JF, Kang H. Dissolution Mechanism of Cellulose in N,N-Dimethylacetamide/Lithium Chloride: Revisiting through Molecular Interactions. The Journal of Physical Chemistry B. 2014 Jul;118:9507-9514. doi: 10.1021/jp506013c
- 6. Bracic M, Mohan T, Kargl R, Griesser T, Hribernik S, Köstler S, Stana-Kleinscheka K, Fras-Zemljica L. Preparation of PDMS Ultrathin Films and Patterned Surface Modification with Cellulose. RSC Adv. 2014;4:11955-11961. doi: 10.1039/C3RA47380E