Engineering

Design 1 - Basic design

Ever since the initial discovery of IsPETase in 2016¹, there have been a multitude of papers devoted to improving the enzymatic capabilities of the enzyme through engineering. One of the most recent and prominent ones is FAST-PETase, which has been shown to exhibit superior capabilities compared to previous variants in terms of depolymerization rate². We thus chose it as the variant we sought to produce. We chose E. coli as our expression host as it is widely used and easily accessible. We also decided that we would want to secrete the enzymes, and found that a variant of the pelB signal sequence has been previously used to secrete IsPETase out of E. coli cells³. We hypothesized that the same signal sequence would similarly work for FAST-PETase.

As the activity of PETase is the bottleneck in the PET degradation pathway, we decided to use the wild-type MHETase coding sequence as the basis for our second part. As with PETase, we wanted to secrete the enzyme, and while initially we planned to use the same pelB signal sequence, we found literature suggesting it does not work with MHETase⁴. Instead, we adopted the lamB signal sequence which was shown to best secrete MHETase from E. coli.

As we wanted to eventually generate quantifiable data on the rate of PET depolymerization by enzymes produced from our parts, we made sure to choose a cloning strategy and a vector that would give our protein a C-terminus His-tag for purification. pET-22b was chosen as the expression vector as restriction cloning using NdeI and XhoI would yield an open reading frame (ORF) with a C-terminus His-tag as shown in Figure 1. Next, it was time to move into the wet lab.

Build 1 - Disappearing vectors

We acquired pET-22b and pET-28a backbones, and while the latter was not exactly compatible with our insert design, it proved useful in troubleshooting our first issue.

While attempting to linearize pET-22b with NdeI and XhoI for ligation with our inserts, we repeatedly ran into an issue where the plasmid would not appear on our gels. Figure 2 shows that while insert digestion and the digestion of our control vector pET-28a was successful, no bands were observed for digested pET-22b vectors.

While the restriction protocol for both plasmids was the same, they were purified from different E. coli strains. pET-22b was purified from BL21 (AI), while pET-28a was from TOP10. BL21 contains the endA1 gene, which codes for a non-specific periplasmic DNA endonuclease I, while in TOP10, this gene has been removed. We hypothesized that our purified pET-22b samples were contaminated with the endonucleases, which would activate when digesting the vector at 37°C, shredding our plasmids. This would be consistent with the gels as small DNA fragments would quickly move through the gel. To test our hypothesis, we added an additional washing step to our plasmid purification protocol.

Figure 3 shows that adding a washing step to our plasmid purification protocol solved the issue of disappearing vectors, meaning that there was most likely some type of contamination, possibly endonuclease I.

Despite now having the linearized backbone, multiple ligation attempts proved unsuccessful. We furthermore transformed the undigested pET-22b vector purified from BL21 (AI) into TOP10 and saw no growth. This led us to determine that the issue was most likely with the vector given that we were able to clone our inserts into pET-28a, transform successfully, and observe colonies with the same protocols. However, as stated before, pET-28a was used simply as a control plasmid, as the ORF formed with our cloning strategy was suboptimal in this vector.

It remains partially unclear to us why we had such issues with pET-22b. The vector had been used in other projects before, and supposedly had an insert cloned into the multiple cloning site (MCS) using the same restriction enzymes we tried using. Even though linearization was supposedly successful, it is curious that the 498 bp insert does not show in our gel on lane 2 in Figure 3.

Design 2 - Updated design

After giving up on pET-22b, we got our hands on pRSFDuet-1. The second MCS of this dual expression vector would be suitable for us as it included the NdeI and XhoI sites. However, this backbone does not include a His-tag downstream of the XhoI site, and as we had depleted our stocks of the original inserts after many cloning attempts, we decided to modify our design and order new inserts.

A His-tag and stop codon were added to both inserts. Before ordering the updated inserts from IDT, we again simulated the restriction cloning using Geneious as shown in Figure 4, and made sure translation rates were sufficient using the RBS calculator⁵.

For a more detailed description of the design process refer to the parts registry entries BBa_K4701300 and BBa_K4701301 for PETase and MHETase respectively, and our design page.

Build 2 - Successful transformation

After the new inserts arrived, we restricted both pRSFDuet-1 and the inserts with NdeI and XhoI and ran a gel followed by gel purification. Next, we transformed the ligated constructs by electroporation into competent TOP10 E. coli. Unlike during previous attempts, this time we actually saw a tiny number of colonies on one of the plates with constructs containing our FAST-PETase insert. We proceeded to restreak them and saw considerable growth on kanamycin plates as shown in Figure 5. Unfortunately, none of the plates with the MHETase construct had any colonies at this point.

To gain confidence that the growing colonies indeed had our construct in them, we took samples from some of the colonies and conducted restriction analysis on purified plasmids as shown in Figure 6.

To confirm successful transformation, we purified plasmids from colonies 5 and 11 (Figure 5) and sent them out to Eurofins Genomics for Sanger sequencing with the DuetUP2 and T7term primers.

While we were waiting for the sequencing results, we decided to adjust our cloning strategy once more, as we were still working on completing the MHETase construct. We suspected that the largest reason for our troubles with transformation efficiency was related to the low DNA concentrations following gel purification, which were consistently as low as 2-5ng/μl despite changing our gel purification protocol on multiple occasions. Our new approach would be amplifying the inserts by PCR and subsequently restricting right after PCR purification, only running a control gel on some of the samples to retain high concentrations. This way we could attempt skipping gel purification.

We designed and ordered PCR primers for our inserts. Next, we ran PCR and successfully acquired concentrations above 300 ng/μl after PCR purification. To confirm successful amplification, some of the PCR product was run on a gel as shown in Figure 7 for MHETase.

Meanwhile, the sequencing results from Eurofins Genomics confirmed the successful cloning of the FAST-PETase insert. For colony 11, the sequencing matched perfectly with our design as shown in Figure 8. For colony 5, both reads called one silent mutation, although the fluorescent peak was not strong, nevertheless, we proceeded forward with colony 11. The construct from colony 11 was purified from the TOP10 E. coli and transformed into NEB T7 Express E. coli, which subsequently grew on kanamycin plates as shown in Figure 9. Thus the build-phase for FAST-PETase was completed.

We still needed to tackle cloning the MHETase insert. With our new strategy skipping gel purification, we were able to finally get colonies after transformation. Unfortunately, multiple restriction analyses indicated that the insert was just under 1000 bp in length while the desired length for the MHETase insert would have been 1872 bp. As the larger fragment was in each test the size of a correctly linearized pRSFDuet-1 vector, we could rule out incorrect restriction. We suspected that there may have been a mixup of the samples, as the restriction analysis would be consistent with the FAST-PETase construct. However, in the end, the reason for these results remained a mystery for us.

Test

With the FAST-PETase construct built, we were then ready to move on to the test phase. Here our goal would be to secrete the enzyme and attempt depolymerizing PET. The first goal could be checked by inducing protein expression with IPTG, followed by centrifugation and SDS-PAGE analysis to check for correctly sized product in the supernatant. The second goal can be studied by checking for difference in weight after exposing PET to the enzyme, or by checking for the presence of PET monomers by HPLC. Figure 10 shows the results of an SDS-PAGE run following expression induction with IPTG.

Learn

As no strong band of the correct size was seen on the SDS-PAGE gel, the next steps would include troubleshooting and possibly changing the design of the inserts. The pelB signal sequence should secrete FAST-PETase into the periplasmic space, where the leader sequence is subsequently removed by a signal peptidase³. Although this still leaves the outer membrane between the protein and the extracellular environment, PET depolymerization with supernatant from E. coli expressing IsPETase in this manner has been previously reported³.

For a second engineering cycle, we might consider ditching the idea of secreting the enzymes, instead opting for lysing the cells followed by enzyme purification by affinity chromatography. However, this approach is quite costly and does not scale up easily. Other possible approaches include using a genetically modified strain of E. coli that has a leaky outer membrane, or co-expressing bacteriocin release proteins that disturb the outer membrane⁶. Another study suggests that following the pelB sequence with a short aspartate tag can allow proteins to be secreted all the way across the outer membrane⁷. Lastly, we could change the expression host from E. coli to a host more readily capable of extracellular secretion. For example, IsPETase has been successfully secreted with the native signal sequence from Bacillus subtilis⁸.