Level 0

‍Amplification of Level 0 parts
‍Due to issues of both length and synthesis complexity, the coding sequence for our chitinase of choice (ChiS) was ordered from IDT as two separate gblock segments, alongside anchor protein CotZ (ordered instead as a single fragment).

Figure 1. The design of our chitinase construct.

The anchor protein CotG was instead PCR amplified from the genome in two versions, one featuring the custom-designed linker and one without. Indeed, the same forward primer was used in both amplifications, but in one case the reverse one featured the linker sequence in addition to the standard overhang.
The two chitinase fragments and the primers used for the amplification of CotG were specifically designed to feature recognition sites of restriction enzyme BpiI, the toolkit’s enzyme of choice for L0 assembly, and appropriate overhangs to allow for correct directional assembly. This allowed the successful assembly of the anchor protein, linker and chitinase into a single fusion protein, contained in the toolkit’s L0 construct reserved for coding sequences (0c).

Figure 2.The gel image result of CotG with linker (The size is around 652 bp with primers oSP001 and oSP003)

As can be noted from the figure above, initially we were only successful in amplifying CotG-L not CotG. Noting the presence of primer dimers and the difference in melting temperatures of the two primer pairs, we adopted two different troubleshooting strategies:

1. Testing different primer concentrations
   
2. Testing a range of annealing temperatures in our thermocycler Phusion amplification protocol

Figure 3.The gel image result of CotG without linker and CotG_linker (The size is around 628 bp for CotG and 652bp with linker. We used primers oSP001 and oSP002)

Adopting these strategies, we were able to successfully amplify the CotG linear fragment for direct fusion to the ChiS coding sequence.

‍Storage of Level 0 Parts

‍ Whilst our amplified CotG and CotG-L fragments could be easily produced once we identified an effective protocol, we soon realized that our ChiS gblocks would run out if we attempted multiple L0 assemblies. Due to past cloning experience, we knew that we would likely have to perform multiple attempts at assembly before obtaining a successfully-assembled plasmid with no mutations. Given the cost and timescale required to obtain gblocks, we thus prioritized the design and construction of a storage mechanism for our two ChiS coding sequence parts.

Figure 4.The schematic design of our chitinase storage.

Our chitinase display plasmid design, consisted in the anchor protein and chitinase sequence coming together in a L0 plasmid. Thus we could not use the STK toolkit to store the ChiS CDS in a L0 plasmid. We opted instead to use the YTK system, a GG toolkit designed for yeast [1]. We designed primers to amplify our gblocks with recognition sites and overhangs appropriate for YTK. In the figure below you can note, a band corresponding to approximately 1000bp, confirming successful amplification.

Figure 5.The gel image result of chitinase part_1 and chitinase part_2 amplification (Size of Chis_part1 is around 978bp and the size of Chis_part2 is around 807bp; the primers are oPS051 and oPS052).

We performed a gel purification to obtain these linear fragments and perform a GG assembly into a L0 YTK backbone. We then transformed the product into E. coli and screened colonies using cPCR.

Figure 6.The gel image result of chitinase part_1 and chitinase part_2 colony PCR (inserts are about 1.1kb).

We inoculated successful colonies and made glycerol stocks, to ensure we had a safe and permanent supply.

‍ Assembly of Level 0 Plasmids
To create our L0 plasmid containing the fusion protein coding sequence, we performed a Golden Gate assembly using BpiI and 60 cycles. We tested both a 5:1 and 3:1 ratio of insert to backbone. Upon assembly, the L0 constructs were transformed into E.coli Top10 and screened using Phire colony PCR (Thermofisher). Bands of the correct length were observed in some of the screened samples, which were liquid cultured, miniprepped and sent for Sanger sequencing, which confirmed the successful assembly of all three designs. Then, correct constructs were cultured overnight in LB Cm, and used for a glycerol stock (1:1 in 50% glycerol).

Figure 7.The gel image result of L0 plasmid of CotG-Chis, CotG-L-Chis, and CotZ-Chis colony PCR.

Level 1

Assembly of Level 1 plasmids
‍ The newly obtained L0 CDS were subsequently assembled together with a promoter, RBS and terminator into a L1 construct (1A), containing an ampicillin resistance in E. coli, to yield a functional transcriptional unit. However, the L1A backbones in the toolkit do not feature neither an origin of replication functional in B. subtilis nor an antibiotic marker. Thus, it was necessary to use an EXP L1 vector, so dubbed in the toolkit, which did contain these features. However, one challenge present was the presence of the same antibiotic marker in the L0 vector and EXP L1, which made screening of correctly assembled L1 plasmids quite challenging.

Accounting for this feature of the toolkit we performed three different strategies in parallel:

1. Assembly with extensive screening rounds of L0 parts into STK108 (EXP L1)
   
2. Assembly of L0 parts into L1A backbone (change of antibiotic marker from chloramphenicol to ampicillin) and then from L1A to STK110 (EXP L2), featuring a change from Amp to Cm
   
3. Assembly in pMAD L2 vector for genomic integration – must be L2 to include homology arms, must also include an antibiotic marker TU to differentiate from L1s (both have Amp resistance)

In all cases, the composition of our transcriptional unit stayed the same, namely with:

• Constitutive promoter: (Hyperspank)
  
• RBS: tmRBS1 (stk45)
  
• CDS: L0 Fusion Protein CDS (CotG-ChiS/CotG-L-ChiS/CotZ-ChiS)
  
• Terminator: L3S2p21 (STK077)

Strategy 1
This approach proved to be incredibly time consuming and characterized by low assembly efficacy. Transformation plates displayed significant number of clones due to the antibiotic marker of the level zero building parts. Indeed, both the assembled product and the level zero parts share a Cm resistance cassette. Hence, in order to successfully screen the mutants, massive efforts were devoted towards several colony PCRs each screening upwards of 32 clones.

Initially, as can be noted from the figure below, most of our cPCR attempts resulted in negative clones or false positives. Though the backbone provided a challenge, this also became an opportunity as we realized there was room for optimization in cPCR protocol we were adopting, more specifically in the following areas:

1. Primer concentration – by using resuspended primer directly from the stock, we were using a very high concentration whilst also pipetting very small quantities, this resulted in both a greater risk of human pipetting errors and the prevalence of primer dimers. Upon consulting with other members of the lab, we elected to use a 10uM dilution of the primer stock, adding 1ul of each (forward and reverse) to the reaction. This enabled more high-throughput screening and better consistency in results.
   
2. Diluting the colony in water before adding Phire master mix – one issue we faced with cPCR was not knowing whether we had picked enough biomass and whether this had resuspended successfully. In response, we decided to pick colonies and dilute them in water first, before adding our master mix. The visual check of turbidity enabled us to better understand what went wrong when a cPCR failed.
3. Modifying our Thermocycler program – we realised that primer annealing temperatures failed to yield any amplification. To mitigate for discrepancies between the software prediction and the actual Ta of primer pairs we followed two strategies. Initially, touchdown PCR was adopted to optimise our amplification workflow. However, this proved to be unreliable with poor results. Hence, upon recommendation of our instructor Mo, we decided to use the AutoDelta function of our Thermal Cycler, an advanced setting that allowed us to closely vary the annealing temperature of the primer pair to our specifications. In particular, an initial melting temperature of 65 was selected. Then, for 10x consecutive PCR cycles, a 1C decrease from the initial Ta was programmed. At the end of these cycles, the machine would then perform unselective amplification with an annealing temperature of 50C. This protocol allowed us to resolve any discrepancies in primer pair and undesired interaction, leading to the obtainment of the PCR fragment of interest.

Figure 8.The gel image result of L1 plasmid of CotG-Chis, CotG-L-Chis, and CotZ-Chis colony PCR.

As can be seen by the figure above, after a few attempts we obtained positive band for CotG-ChiS and CotG-L-ChiS (≈ 2.7kb). We then inoculated these clones, miniprepped them and sent them for Sanger sequencing. Clones that aligned perfectly to the in-silico design (no mutations), were then used in subsequent assembly levels and stored in glycerol stocks.
For what concerns CotZ-ChiS, no clones were obtained. Hence, a new assembly was performed with the aim of screening more colonies.

Figure 9.The gel image result of L1 plasmid of CotZ-Chis colony PCR.

As it can be seen from the image, a single, positive hit was obtained.

Strategy 2
In the lab, we have successfully built and verified Chitinase display Level 1 transcriptional units (1A), this was done by screening using cPCR and then performing a restriction digestion.

Figure 10.The gel image result of L1A plasmid of CotG-Chis, CotG-L-Chis, and CotZ-Chis colony PCR.

Figure 11.The gel image result of L1 plasmid of CotG-Chis, CotG-L-Chis, and CotZ-Chis restriction digest with BsmbI.

Strategy 3
Consulting academics experienced in work with B. subtilis such as Graham Christie and Geoff Baldwin, we discovered B. subtilis prefers linear DNA to plasmids, all while boasting an efficient homologous recombination system. Considering the difficulties faced in building a L1 assembly that can be expressed in B. subtilis, we decided to attempt to perform a genomic integration so as to obtain spores displaying chitinase enzymes for our proof of concept. The STK toolkit provides parts for efficient genomic integration, namely:

1. pMAD vector - this is a L2 vector for expression in B. subtilis, that features a temperature sensitive origin of replication. Like all other L2 vectors in the toolkit, it provides space for 4 L1 transcriptional units. By having the first and last transcriptional unit correspond to a 5’ and 3’ homology arm respectively, the vector can be adapted for a double-crossover recombination for gene knockout of whichever gene the homology arms correspond to. This type of recombination would result in the L1B and L1C transcription units in the pMAD plasmid to be integrated in the place of the coding sequence of the gene being knocked out. Infact, this vector was designed to perform knockouts, enabling introduction of a marker such as an antibiotic resistance cassette to then screen for successful integration. In our case, we are using the same system but simply to integrate the transcriptional unit of our fusion protein.
   
2. LacA Homology Arms - The toolkit already contains L1A and L1D parts corresponding to the 5’ end and 3’ end homology arms of LacA, in other words the 1000 bases preceding and following the coding sequence of the LacA gene in the B. subtilis 168 genome. Using these, we can perform an integration at the LacA locus.
   
3. Antibiotic Resistance Casette - As outlined above, L1A to D vectors have an ampicillin resistance marker, so does the pMAD vector. This poses a challenge for screening of the successful pMAD construct, thus we elected to make use of the L1C transcription unit space to introduce a new antibiotic resistance cassette. The toolkit already contains a L1C Kanamycin resistance cassette functional in both E. coli and B. subtilis (STK 271). By utilizing this part in the assembly of pMAD, we can plate the screen transformants of the same on an LB-Kan plate.

In light of the above, our integration shuttle vector is designed with the following parts:

• Backbone: pMAD with sfGFP dropout cassette (STK011)
  
• L1A: LacA 5’ end homology arm (STK083)
  
• L1B: Our fusion protein (FP) transcriptional unit (CotG-ChiS, CotG-L-ChiS, CotZ-ChiS)
  
• L1C: Kanamycin resistance cassette (STK 271)
  
• L1D: LacA 3’ end homology arm (STK084)

Figure 12.The scheme of our pMAD design.

In terms of steps this strategy can be broken down into three stages:

• Assembly and verification of fusion protein transcriptional unit in L1B vector
  
• Assembly and verification of pMAD-2
  
• Linearization and Transformation into B. subtilis

Given the tightly controlled process of sporulation, we also wanted to test the difference in burden, expression level and functionality using an anchor protein specific promoter vs. a constitutive one. One concern was that using a constitutive promoter might result in excessive metabolic load, affecting the standard functioning of the cell including sporulation. Furthermore due to the timeliness of CotG expression, another worry was that the fusion protein might integrate into the wrong layer of the spore coat. Thus, we designed a new part for the STK toolkit, a CotG native promoter + rbs. We accomplished this by taking 200bp upstream from the CotG CDS in the B. subtilis genome and adding the appropriate recognition sites and overhangs on either end. Assembly using strategies 1 and 3 were attempted, given the poor results from strategy 2.

Figure 13.The gel image result of L1 plasmid of pHyp_CotG-ChiS (Hyperspank promoter) and pCotG_CotG-ChiS (CotG promoter) colony PCR.

Again, constructs were transformed in E.coli Top10 and screened via Phire colony PCR, which demonstrated the presence of potentially successful samples.

Transformation in B.subtilis

Before proceeding to transformation of the assembled plasmids into B. subtilis, restriction digestion of the constructs was performed to rule out the possibility of a false positive result in the screening stage with colony PCR.

Strategy 1
STK108 vectors containing our CotG and CotG-Linker were screened using SapI and EcorI. Inconsistent results were obtained.

Figure 14.The gel image result of STK108 vectors containing our CotG-Linker restriction digest with SapI and EcoRI.

It was then decided to repeat the golden gate and screening, more colonies were selected.

Figure 15.The gel image result of L1 plasmid of CotG-Chis, CotG-L-Chis and CotZ-ChiS colony PCR.

Screened again with EcoRI and NheI.

Figure 16.The gel image result of L1 plasmid of CotG-Chis and CotG-L-Chis restriction digestion with NheI and EcorI.

Seem to have positive clones.

For what concerns CotZ-ChiS, restriction digestions with EcorI, HindII or BsmbI were performed to validate the presence of our construct. Moreover, in the experimental setup, we also included the STK L0 CDS part to rule out the presence of any false positives.

Figure 17.The gel image result of using different restriction enzyme combinations of CotZ-Chis.

Strategy 2
Once correct samples were identified viar restriction digestion analysis, it was time to set up a higher order assembly with spacer sequences to allow for B. subtilis transformation.

Strategy 3
Having proceeded with all cloning strategies in parallel, we were able to successfully isolate and transform in bacillus CotG-ChiS, CotG-L-ChiS and CotZ-ChiS with a constitutive hyperspank promoter. However, after many colony PCRs, we still did not manage to verify a successful transformation. As an alternative, we considered characterizing burden of our transformed B. subtilis, but that experiment failed. Stay tuned for more news!

References

[1] Larroude M, Park YK, Soudier P, Kubiak M, Nicaud JM, Rossignol T. A modular Golden Gate toolkit for Yarrowia lipolytica synthetic biology. Microb Biotechnol. 2019 Nov;12(6):1249-1259. doi: 10.1111/1751-7915.13427. Epub 2019 May 31. PMID: 31148366; PMCID: PMC6801146.