Every day, the team went through engineering design cycle iterations in most experiments and protocols used. From decreasing the amount of DNA loaded on a gel so as not to saturate the imager, to increasing the extension time in a PCR to get more discrete and clearer bands. However, a few significant changes stood out as part of the design, build, test and learn approach.
At the early stages of the project, the team selected one of the JUMP backbones (pJUMP28-a) as the vector to make the constructs with. The team also intended to use Golden Gate cloning to build entire transcriptional units (TUs) comprising a promoter, ribosome binding site (RBS), coding sequence (CDS) and terminator, using only parts from the distribution kit.
Upon completion of the first attempt to assemble the first set of plasmids, each with their own 4 parts and vector, the team quickly realised that the method to screen colonies for the right plasmid we utilised (colony PCR), did not yield any meaningful results. What is more, pJUMP28-a contains a sfGFP by default and the intention was to clone in efasGFP (another green fluorescent reporter), which meant there was no way to differentiate the colonies with the ligation plasmid from the colonies with empty backbone visually. In theory both, successful ligation colonies and colonies from non-digested or recircularised pJUMP28-a would have appeared green. Additionally to this, the plates also showed non-fluorescent colonies, which made results even more puzzling.
The team tried to get around this problem by digesting pJUMP28-a and pJUMP29-a and swapping their inserts. This way the high copy number Ori from pJUMP28-a would have been preserved but it would have been combined with a lacZ gene, allowing for blue-white screening, potentially solving the problem. Nevertheless, when gel extraction was attempted to recover the correct insert and the correct backbone, a very low yield was produced. This was attempted a few times prior to switching tactics one last time.
The final decision was to use a pET28a vector instead, which encodes its own RBS, terminator and inducible pLac. Harnessing the Lac mechanism allowed for the attainment of different levels of gene expression by varying the concentrations of the inducer. This way, for fluorescent constructs it was very easy to tell which colonies had the right construct, since only correct ligation colonies would express a GFP. Anyhow, the team still screened them through colony PCR and sent them to sequence for a definitive check. For non-fluorescent constructs, colony PCR also yielded significantly better results than when using JUMP backbones. An explanation for a better success rate is that doing four ligations at once is considerably more error-prone than doing a single one.
Another example of a very significant learning outcome from the engineering design cycle pertains to the incorporation of DpnI digests.
To get the version of pET28a that was needed, inverse PCR was used to linearise, amplify and give BsaI restriction sites to the backbone. After the team attempted to ligate growth slowing genes (GS) into pET28a, with DNA from colonies that went through the screening process were sequenced as a last check. Nonetheless, the results indicated the DNA contained just the empty pET28a vector, without an insert.
This most likely means that some of the original circular backbone made its way through the purification process (including a gel extraction) and ended up mixed with the linearised backbone and transformed into the Escherichia coli that was subsequently mini-prepped. Most of the sequencing results received showed the empty pET vector.
The team immediately brainstormed to find a solution and thought of the DpnI enzyme, which cleaves DNA when its recognition site is methylated. This means products from a PCR (which are unmethylated) wouldn't be cleaved, but the original PCR template would be, producing only linearised pET28 as a result.
After making this single addition to the methods, 7 of our 8 following ligations were successful.
Another instance in which the engineering cycle was applied and reflected upon was after completion of the first two plate reading experiments, where fluorescence of vsfGFP in pET28a was measured under different concentrations of IPTG. In the first experiment the concentrations of IPTG ranged from 0 to 1000 µM in 100 µM steps and most cultures saturated the plate reader's fluorescence sensor. As a result, in the second trial the concentration range was reduced to 0 to 100 µM with smaller steps. Even in this trial, all the cultures with concentrations above 6 µM saturated the fluorescence sensor.
Hence, for the final fluorescence reading of pLac-driven plasmids, the range of IPTG concentrations employed was 0 to 5 µM. This ensured that all the curves remained within the sensor's working range and the data could be extracted and analysed. This iterative approach allowed for the acquisition of more useful data.