Engineering success

Cycle A

Characterizing promoters for optimal induction and overexpression systems in M. extorquens

The genetic engineering tools available for M. extorquens are not comparable with those available for model organisms like Escherichia coli. Hence, Cycle A focused on characterization of various constitutive and inducible promoters to determine their potential for application in Cycle B and Cycle D, which required constitutive and inducible promoters, respectively.

Design & Build

With regard to the phaC and groES/EL overexpression cycle (Cycle B), various constitutive promoters were assessed. The promoter with the highest known expression in M. extorquens is the mxaF promoter¹. PmxaF is the promoter for the mxaF gene in M. extorquens. This gene encodes the large subunit of methanol dehydrogenase¹, which is highly expressed in methanol-rich conditions², it has also been reported to be slightly upregulated on methanol³ and since we grew our cells on a methanol-based medium, we considered the expression of this promoter constitutive.

Since high levels of overexpression is not ideal in each situation, we also wanted to test promoters with lower expression rates than mxaF. Dr. Lennart Schada von Borzyskowski published a paper on several promoters that were characterized in M. extorquens that we identified as possible alternatives to the mxaF promoter¹. These promoters include the promoters to the fumarase coding fumC gene, the cytochrome c oxidase subunit II coding coxB gene and the transcription elongation factor thermo-unstable tuf gene. These promoters are referred to as PfumC (5-15%), PcoxB (40-50%) and Ptuf (70-100%) with increasing promoter strength relative to PmxaF (100%)¹. All three promoters are considered constitutive.

In terms of the autolysis system, we wanted to assess different inducible promoters. We chose to clone different inducible promoters that were hardly characterized for M. extorquens. We looked into the IPTG inducible promoters P_A1/O4/O3 and P_L/O4/A1⁴, cumate inducible promoters P_Q5 and P_Q2148⁵ and vanillate inducible promoter P_V10⁶ and and OHC14 inducible promoter pCinR^{7, 8}. As can be seen in Figure 3, the inducible promoters have a similar structure, where a repressor gene is constitutively expressed, which binds to the operator sequence. When the ligand is present, the repressor releases, and mCherry is expressed. For P_L/O4/A1, for example, the lacI repressor gene is constitutively expressed by the PlacI^q promoter, and binds to the lacO operator sequence inside the P_L/O4/A1 promoter region, thereby repressing the promoter. When the ligand IPTG is present, this binds LacI, releasing it from the operator. Then, P_L/O4/A1 is open for expression, leading to gene expression of the reporter gene mCherry. In general, repression is not complete, meaning there will always be some level of expression. The level of uninduced expression is indicated as the ‘leakiness’ of the promoter. In this cycle, we wanted to know how leaky the inducible promoters were, since we wanted to express a lethal gene construct, only at a desired time, so that it is able to grow normally before then.

A schematic overview of the mCherry testing cassette with different promoters is shown in Figure 3. An overview of all the characterized promoters is elaborated on in the Parts page .

As a backbone, we used pTE100 empty vector¹ containing an oriV-traJ’ origin with Tc^R. To clone the promoters and testing cassettes in the backbone, we used restriction cloning to insert the promoters, and Gibson cloning to insert the fluorescent reporter gene mCherry into the constructs. We were not able to construct, transform and test each plasmid given limited time. Some plasmids were available to us with either the promoter or the promoter and mCherry already. The PmxaF, PfumC, PcoxB and Ptuf plasmids had both the promoter and mCherry, P_L/O4/A1, P_A1/O4/03 and pCinR only needed to have mCherry put in. All plasmids were constructed and transformed into M. extorquens. However, due to time constraints we were unable to test P_Q2148 and P_A1/O4/03.

Test & Learn

We wanted to learn which of the constitutive promoters were most optimal for the phaC overexpression and which inducible promoters were most optimal for the autolysis system. Therefore we measured the mCherry fluorescence signal of the successfully cloned strains with a plate reader (Tecan Plate reader Infinite 200 PRO). Using the fluorescence signal and the cell density, the dose-dependent promoter strength was determined. You can find all the results on the Results page .

Out of the constitutive promoters, PmxaF showed the most fluorescence compared to control (Fig. 4). Since we did not know what level of overexpression of phaC and groEL/ES would be beneficial for PHA production, and at what point the overexpression would start to burden the cells, we decided to use all the constitutive promoters for the phaC and groEL/ES overexpression.

For the inducible promoters, the results were not as promising as expected (Fig. 4). By mistake, the pCinR strains were induced with IPTG instead of OHC14, which explains the lack of induced effect seen, which is why this data is omitted. The P_V10 strain was first tested with freshly prepared vanillate stock, and an increase in mCherry signal was seen, and later, parallel strains were tested as well with these same dilutions. The absence of induction could be explained by the use of not freshly prepared inducers, which is why on the Results page the results of this replicate are omitted. However, to solidify conclusions about the characteristics of this promoter, the experiments need to be repeated. The P_Q5 promoter showed surprisingly little fluorescence. We omitted this part from our list since the signal was even lower than the negative control, indicating a technical problem for this measurement. The P_L/O4/A1 promoter showed the highest effect by the presence of the inducer IPTG. Since this promoter was also already characterized in literature⁴ and we found a strong effect of the inducer, we used this promoter for induced autolysis in Cycle D.

Redesign

For continuation of this cycle, we would redo the experiment for the strains we were not able to analyze yet. We also want to retest the strains mentioned here to solidify our conclusions and expand the parts collection of promoters for M. extorquens. Moreover, in the redesign of this experimental line, we would like to test even more promoters, such as this newly published variant of the cumate-inducible promoter P_Q2148⁹.

This would provide us with more knowledge and would expand the genetic engineering tools that are available for M. extorquens. For example, we made plasmids expressing the mCherry under cumate inducible promoters, but were not able to get good data on these strains. In the redesign of this experiment, we would like to test this again, integrating the knowledge from this cycle to be able to gather more data on this inducible promoter. Since IPTG is relatively expensive, it would be beneficial if an inducible system that has a cheaper and more sustainable ligand as inducer, such as cumate, could be used for large scale production. For example, we characterized the cumate inducible promoter with this in mind, as cumate is 10 times cheaper than IPTG, with cumate costing €52.10 per 5 gram and IPTG costing €492.00 per 5 gram¹⁰. On an industrial scale, this would allow for significant savings.

Cycle B

Increasing PHA production through overexpression of native genes in M. extorquens

The metabolism of wildtype (WT) M. extorquens is not optimized for large scale production of PHA¹¹. A powerful tool for enhancing the production of a compound such as PHA is the overexpression of native genes¹². Using our constitutive promoters from Cycle A, we designed a construct to constitutively express endogenous synthase genes to increase production levels of PHA.

Design & Build

Consulting with dr. Schada von Borzyskowski , we decided to overexpress the native phaC, groEL and groES genes. PhaC serves as the synthase for PHA, while the GroEL/ES complex is a chaperone involved in efficient folding of PhaC¹³. This complex also prevents the formation of aggregates and is hypothesized to increase PHA production¹³.

Based on the results of the characterization from Cycle A, we made various genetic constructs. These constructs are visualized in Fig. 5. Overall, we created constructs with four different promoters, with either phaC, groEL/ES or phaC and groEL/ES genes combined. In addition, we included a version with a histidine rich region (HIS tag) for the extraction of the protein.

Test & Learn

We tested the constructs by growing the bacterial strains in liquid culture and quantifying PHA production with Nile Red (NR) staining according to experiment 2: Overexpression of genes for increased PHA production allowed us to compare the amount of PHA produced by our strains to the WT.

The results of the PHA quantification of the different strains are summarized in Fig. 6. All constructs show increased PHA production in comparison to the WT at 48 hours. However, there was no promoter that was consistently better across all constructs. We learned that the optimal promoter differs between the different endogenous genes tested. The data is further elaborated on the Results page .

Redesign

In this cycle, we have learned that overexpression of phaC and groEL/ES genes together resulted in increased PHA production in comparison to the wildtype, where overexpression of phaC resulted in the biggest increase.

Based on these findings, in future experiments we could construct separate plasmids for the groEL/ES genes and the phaC gene, under the expression of Ptuf and Pmxaf respectively. We hypothesize that co-overexpression under different promoters would improve the synergistic effects. Furthermore, we want to combine the highest PHA producing overexpression strains with the carotenoid knockout strain of Cycle C. This could be achieved through transformation of the ΔcrtI strain with the overexpression plasmid. Theoretically, this could lead to a M. extorquens strain that allows for a higher influx of compounds into the PHB cycle and overexpression for even more production of PHA. The final strain would then be optimized according to all we learned in regard to the promoters, overexpression and the knockout in cycles A, B and C.

Cycle C

Enhancing PHA production of M. extorquens by knocking out a competing pathway to optimize the metabolic flux

To increase PHA production, we investigated the possibilities of changing the metabolic flux in favor of PHA production. This could be achieved by knocking out specific pathways that are superfluous for a production strain. In M. extorquens, a promising target for this is the carotenoid biosynthesis pathway¹⁴. This pathway creates pigments for UV protection that are not relevant in a microbial bioproduction setting, and can therefore be knocked out¹⁵.

Design & Build

To optimize metabolic engineering, we employed a genome-scale metabolic model of M. extorquens, identifying key knockout targets and assessing their impact on carotenoid biosynthesis. We predicted an increased carbon flux to the PHA production pathway (PHB cycle) to improve PHA production, see Figure 7. Three promising gene targets for a knockout were selected: dxr, ispA, and crtI. This was supported by literature where a colorless M. extorquens crtI knockout strain had been found to be viable¹⁴.

Various genetic tools, such as CRISPR-Cas based technologies, are available for gene deletion, but are not always efficient in non-model organisms, such as M. extorquens¹⁶. Therefore, we opted for a method that has been established for the use in alphaproteobacteria, including M. extorquens, which relies on two homologous recombination (HR) events between the genome and a non-replicating plasmid (pREDSIX) bearing DNA regions that surround the gene of interest¹⁷.

Test & Learn

Based on our validation criteria, we concluded that the crtI gene was successfully knocked out. However, knocking out the ispA gene and the dxr gene repeatedly failed. For the ispA gene this was due to cloning failure, because of time constraints we were not able to test a different type of cloning. We were not able to obtain a viable dxr knockout despite numerous transformation attempts. This may indicate that dxr is an essential gene, and that this gene deletion is lethal, which is supported by literature¹⁸. A more detailed overview of our results can be found on the Results page : Knockout of genes for increased PHA production. We aimed to quantify the amount of PHA that our crtI knockout strain produced, however, due to time constraints this was not possible. This would have been performed in the same manner as described in Cycle B.

We were not able to test the system, but our hypothesis was that by alleviating the metabolic load there would be more energy available for the PHB cycle, resulting in more PHA production. Furthermore, our computational predictions suggested that growing the knockout in nitrogen-deficient conditions may further increase PHA production. As such, validation of the strain and model would have also included growing the knockout in these conditions.

During our experiments, we faced challenges in gene knockouts within M. extorquens and the carotenoid biosynthesis pathway. The phenotype of our crtI knockout strain also did not match expectations, remaining pink despite literature reporting colorless mutants with crtI knockouts¹⁴. Moreover, since the restriction cloning of the ispA knockout continually failed, Gibson assembly could be an alternative route for obtaining this construct in future experiments.

Redesign

In this cycle, we have hypothesized that a knockout in the carotenoid biosynthesis pathway creates a change in the metabolic flux in favor of PHA production.To redesign this cycle, we would incorporate our learnings from both Cycle B (overexpression) and C (knockout). Drawing from the overexpression cycle, we would integrate the best-performing strains into the viable carotenoid knockout strain, creating an optimized M. extorquens strain for PHA production. Simulating the carotenoid biosynthesis pathway gene knockouts in the metabolic network model [Modeling page] of M. extorquens showed that the reaction flux through the PHB cycle is increased compared to the wild-type, except for the reaction synthesized by the PhaC enzyme. This highlights the need for combining a carotenoid knockout strain with a phaC overexpression strain to overcome this potential limitation.

Cycle D

Uncovering the optimal inducible autolysis system for M. extorquens

Current methods for PHA extraction are environmentally hazardous, costly, and dangerous for workers^{19, 20}. To address this, we used cell lysis systems from bacteriophages. Phages are viruses that infect bacteria, and due to this, they have cell lysis systems for killing and escaping their host cells²¹. If these lysis systems can be expressed in M. extorquens, at the desired time, this could benefit the extraction processes.

Design & Build

The optimal choice for a phage-based cell lysis system would be a phage specific to M. extorquens, however, these have not been found yet. Therefore, we selected 5 lysis genes and gene systems from phages that infect bacteria that are phylogenetically closely related to M. extorquens. See Table 1 on the Results page for an overview of the lysis genes and their host organisms.

Due to the highly lethal nature of these genes, we used the inducible promoter P_L/O4/A1, as characterized in Cycle A. This was done to allow selective activity of the lethal genes.

As exogenous DNA is not always functional in a new host organism, we also optimized several aspects of our genetic constructs to ensure their functionality in M. extorquens using the De Novo DNA transcription level prediction software and the De Novo DNA combined synthesis success tool and codon optimization tool²². The genetic constructs can be seen in Figure 9.

See our Results and Notebook pages for details on the cloning of these constructs.

Test & Learn

Firstly, we tested the systems by comparing the effect of the gene inserts on colony forming unit frequency in both the induced and uninduced state, as can be seen in fig. 10. This showed a statistically significant decrease in the number of surviving colony forming units for all cell lysis systems. The reduction was especially notable for Salmonella phage endolysin Gp110 (GP110) and Pneumococcus EJ1 Holin/Endolysin (EJ1), with these showing high lethality. However, there was no difference between the induced and uninduced conditions, indicating that the genes were functional, but the inducer did not work.

Secondly, to confirm the effects of the inserted genes, we studied the effect of the genes on the cell wall integrity of M. extorquens cells. This was done by growing controls and cell lines under exposure to butanol, an organic solvent. M. extorquens is generally highly resistant to solvent toxicity, and this trait is closely linked to cell wall structure and integrity²³. If the lysis genes had a cell wall compromising effect, this would be noticeable under solvent stress. Both Gp110 and EJ1 showed a significantly decreased tolerance to butanol in comparison to the other strains and to the control. This supports our hypothesis that these inserts compromise the cell wall of M. extorquens. See Results page for more information. Based on this, we could conclude with a good degree of certainty that two of the phage lysis gene constructs were functional in M. extorquens.

However, the lysis system did not work entirely as anticipated. Our results indicate that the P_L/O4/A1 inducer was either too leaky, or did not repress our specific genes well. Despite this, all lysis systems showed statistically significant effects in either the survival study, the solvent exposure experiment, or both.

Redesign

This cycle shows that the P_L/O4/A1 promoter is not fully repressed in the absence of IPTG. We found that especially Gp110 endolysin and EJ1 Holin/Endolysin show statistically significant lethality.

As found in Cycle A, the IPTG inducible promoter P_L/O4/A1 had a high level of leakiness, and to iterate on the design of this cycle, we would need to employ an alternative inducer system. Due to the leakiness, we were unable to characterize the effect of the auto-lysis system on extraction efficiency. Future iterations for the auto-lysis system should include screening multiple inducers to identify an optimally non-leaky and affordable inducible system for large-scale industrial use.

Conclusion

Our engineering cycles allowed for optimization of PHA production of M. extorquens, eventually leading to improved yields and cost-efficiency. The cycles combine data analysis, modeling, experimentation, and iterative refinement to achieve our goal. Going through the DBTL cycles allowed us to obtain results, negative and positive, learning from them and then adjusting the process. Despite not being able to complete every cycle due to time constraints, each cycle did help us obtain important results that could be used in other cycles. Overall, all of the experiments allowed us to gain more knowledge about how to produce cost-effective and sustainable PHA.

Future experiments

As the engineering cycle is an iterative process, it is never fully completed. Since an iGEM project only lasts a few months, there is a limited amount of experiments that can be performed. Therefore, there are various experiments and steps that PHAse Out can take in the future. These include assessing the reproducibility of the results obtained here, and then incorporating all of the learnings from the different cycles to engineer an optimized M. extorquens strain that includes overexpression, knock-out, and autolysis.