Results
Our objective regarding biotechnology was to engineer an oxalic acid producer that is
- Safe
- Easy to handle and
- Produces high quantities of oxalic acid for bioleaching of PGMs.
Proof of concept
Creating an oxalate producer
The first big goal was to genetically engineer Corynebacterium glutamicum to produce oxalic acid. In order to achieve this, we introduced the codon-optimized oahA-gene originating from Aspergillus niger into the expression vectors pEKEx2 and pPBEx2 by DNA assembly using NEB’s assembly protocol. OahA encodes the oxaloacetate acetylhydrolase which converts oxaloacetate into oxlaate and acetate. Since oxaloacetate is a key metabolite in the TCA cycle, it was important to check whether the cells could even survive this interference.
The two plasmids were transformed into E. coli and subsequently electroporated into C. glutamicum. Colony PCR followed by gel electrophoresis confirmed the successful transformation (Fig. 1).
The positive clones were sent for sequencing which confirmed the positive gel electrophoresis results.
First cultivation experiment
Viability and production-capability of the plasmid-containing mutants C. glutamicum pEKEx2_oahA and C. g. pPBEx2_oahA were tested in a first cultivation:
Five 50 mL cultures on CGXII-medium in baffled shake flasks were inoculated with
- C.g. wildtype
- C.g. pEKEx2_oahA
- C.g. pEKEx2_oahA (+)
- C.g. pPBEx2_oahA
- C.g. pPBEx2_oahA (+)
and cultivated until growth stagnated. Cultures 3) and 5) were induced with 2 mM IPTG after 3 hours to start oxalate production. Optical density was measured every 1.5 hours and oxalate levels were quantified using a commercial enzymatic oxalate assay kit.
Results showed that the growth rate of plasmid containing cultures was only slightly decreased in comparison to the wildtype. This prooves viablity and good growth performance on CGXII-medium.
| culture | µmax |
|---|---|
| WT | 0.43 |
| pEKEx2 | 0.42 |
| pEKEx2+ | 0.37 |
| pPBEx2 | 0.40 |
| pPBEx2 | 0.38 |
As expected, growth rates of induced cultures are lower than those not induced. This is a first hint that oxalate production was indeed induced upon addition of IPTG which increases the metabolic burden. Oxalate quantification unfortunately showed no reliable results.
Fig. 3 shows that oxalate was produced in the induced cultures but not/only to a low extend in the uninduced controls as well as the wildtype. However, oxalate levels after 28 h in the induced cultures show extremely high variance (156235.3 µM vs. 1239.0 µM for pEKEx2_oahA and pPBEx2_oahA, resprectively). This can only be attributed to errors in measurement, since it is highly unlikely that two plasmids so similar behave so differently at the end of cultivation. In conclusion, oxalate production can be proved qualitatively but no quantitative results are obtained at this point.
Conclusion
The key findings for this first stage are:
- C. g. pEKEx2_oahA and pPBEx2_oahA are not only viable but show excellent growth performance
- Oxalate production is possible by introducing just one new enzyme in C. glutamicum’s metabolism
- Oxalate is exported out of the cells
- Oxalate quantification proves difficult
Open questions are:
- How much oxalate can be produced?
- How much oxalate remains intracellular?
- How can oxalate production be increased?
Scale-up DASGIP
DASGIP bioreactors are stirred-tank reactors with a working volume of 1.5 liters. All reactors were equipped with Rushton turbines. In this experiment, Corynebacterium glutamicum, a model organism commonly employed in bioprocess engineering, was cultivated in six parallel reactors with CGXII medium to examine the impact of oxalate production on bacterial growth kinetics. The wild type was cultivated in three reactors, while the remaining three reactors contained the bacterium with the plasmid pPBEx2_oahA for oxalate production. This plasmid contains a repressor system in which the gene for oxalate production is only transcribed upon addition of IPTG. After 3 hours of cultivation, two pPBEx2_oahA cultures were induced with 2 mM IPTG to investigate the effects of oxalate production on growth, while one wild-type culture was induced with IPTG as a control to assess the effects of the inducer itself on growth. Samples were taken hourly throughout the experiment to determine the OD₆₀₀, biomass dry weight, and oxalate concentration. The experiment was terminated when exhaust gas analysis indicated an increase in O₂ content in the off-gas, signifying that the bacteria no longer consumed the supplied oxygen. This occurred after approximately 10 hours for the wild type and after approximately 14 hours for the plasmid-bearing cultures.
Kinetic and yield parameters of the reactors
To quantitatively characterize and compare the growth behavior of the different cultures, the maximum growth rate (µmax), the biomass yield on substrate (glucose) (YX⁄Glc), and the specific glucose uptake rate (qGlc) were calculated from the measurement data and are given in the following table.
Maximum growth rate (µmax)
The wild type achieved a maximum specific growth rate of 0.41 h-1. The non-induced plasmid-bearing culture exhibited a reduced growth rate of 0.33 h-1, representing a 19.51% decrease. The IPTG-induced plasmid cultures displayed a further reduction of 4.88%, corresponding to a maximum specific growth rate of 0.31 h-1 and thus a total reduction of 24.39% compared to the wild type.
This reduction can be attributed to the increased metabolic burden imposed by the plasmid. The bacteria must allocate additional energy and resources for plasmid replication and expression of the kanamycin resistance gene encoded on the plasmid, as well as, following IPTG addition, for the expression of the oahA gene responsible for oxalate production.
The IPTG-induced wild-type culture exhibited a growth rate of 0.40 h-1, comparable to that of the non-induced wild type. This indicates that IPTG itself does not exert a significant influence on the growth of C. glutamicum.
Biomass yield per glucose (YX⁄Glc)
During the determination of the biomass yield on glucose, relatively high standard deviations were observed for both the wild type and the plasmid-bearing strain. To assess whether the mean values differ significantly, a Student's t-test was performed. The resulting p-value did not indicate significance, suggesting that the biomass yield on glucose does not differ between the strains. This outcome contrasts with our initial hypothesis that the yield would decrease in the plasmid-bearing strains, as part of the substrate is redirected towards oxalate production. The reason for this is presumably that the amount of oxalate produced by the bacteria is too low to cause a measurable change in biomass substrate yield.
Specific glucose uptake rate (qGlc)
The specific glucose uptake rate was 0.19 g⁄(g·h) for the wild type and 0.14 g⁄(g·h) for the plasmid-bearing cultures, representing a 19% reduction that corresponds to the decreased maximum specific growth rate.
Oxalate Production
To measure oxalate production in samples from Reactor 1 and Reactor 3, an enzymatic oxalate assay kit from Sigma-Aldrich (MAK315) was used. This assay employs HRP and OX enzymes to catalyze a colorimetric reaction, with color intensity being directly proportional to oxalate concentration. The optical density values of samples from Reactor 1 (Fig. 4) demonstrate qualitatively that bacteria harboring the induced pEPEX2_oahA plasmid exhibited increasing oxalate production over the course of cultivation. No oxalate production was detected in the wild-type strain.
Due to an imprecise standard curve and the limited number of samples that could be analyzed with this assay, duplicates or triplicates were not possible, and the quantitative results lack validity. Only a rough estimate can be provided: approximately 582.8 µM oxalate was produced after 15 hours.
Due to the high costs, limited sample numbers, and unsatisfactory results in the evaluation of this and previous cultivation experiments, the development of an in-house oxalate assay was initiated concurrently.
Conclusion
The results of this bioreactor cultivation experiment provide clear evidence for the metabolic cost of oxalate production in C. glutamicum. Plasmid-bearing strains showed a 19.51% reduction in maximum specific growth rate and a 19% decrease in specific glucose uptake rate compared to the wild type. Notably, IPTG induction further reduced the growth rate by 4.88%, indicating that active oxalate production imposes an additional burden beyond plasmid maintenance. The cell dry weight over time, depicted in the following figure, visually confirms this growth impairment, with plasmid-bearing cultures requiring approximately 4 additional hours to reach stationary phase. The absence of a significant change in biomass yield on glucose suggests that the amount of oxalate secreted by the bacterium was too low to exert a measurable effect. These findings provide a quantitative baseline while revealing substantial optimization potential for future strain engineering to enhance oxalate production.
Optimization
Cultivation parameters
To elevate oxalate production, different induction times and inducer concentrations were tested to find the optimal conditions. However, results were inconclusive (data not shown) and need to be repeated in future experiments.
Genome integration
Another approach to enhance oxalate production is to create a more stable production strain by integrating oahA into the genome. This way, no more antibiotics are needed for prevention of plasmid-loss. If integration is successful, the next step is to introduce the plasmid again to see how the combination of plasmid and genome integrated oahA influences the production.
For genome integation, we chose pK19mobsacB as the integration vector. pK19 contains an origin of replication for E. coli, which means that it can be amplified in E. coli for subsequent integration in C. glutamicum‘s genome via 2 step homologue recombination. After extensive literature research, we identified the region between cg3228 and IldA as a suitable locus [Lange et al., 2017].
Three different attempts were carried out:
-
Integration of oahA
Figure 6: Integration of oahA into cg3228/IldA. -
Integration of oahA with addition of a lacI repressor system
Figure 7: Integration of oahA with lacI repressor system into cg3228/IldA. -
Integration of oahA with lacI repressor system and terminator
Figure 8: Integration of oahA with lacI repressor system and terminator into cg3228/IldA
-
Integration of oahA
In the first trial, oahA with tac promoter and lac operator was assembled into pK19 together with Upstream and Downstream adapters homologue to the C. glutamicum genome integration locus flanks.
Figure 9: Integration vector pK19_oahA containing oahA under Ptac and with lac operator. Upstream and Downstream flanking regions are homologue to the integration locus in the C. g. genome. pK19 contains SacB and KanR for two-step selection. After assembly, pK19_oahA was introduced into E. coli via heat shock. After selection on Kan50 agar plates, colonies were tested with cPCR for correctly assembled plasmids and sent for sequencing. However, no positive clones could be confirmed.
Figure 10: Sequencing results aligned with the expected sequence. Each line shows one tested sequence. Red indicates matches, blank indicates mismatches/missing basepairs. The sequencing results all have in common that no tested sequence contains a functional tac promoter. This led us to conclude that the plasmid is too high of a burden for E. coli if not regulated since the expressed oahA gene constantly removes oxaloacetate from the TCA cycle. This means that no cell containing the whole Ptac-oahA sequence could survive which is exactly what we see in the sequencing results.
-
Integration of oahA with addition of a lacI repressor system
To solve this problem, we decided to try integrating an additional lacI-repressor system alongside the oahA and Ptac to inhibit constant OAH-expression. Without addition of the inducer IPTG, the repressor system consisting of lacI, its promoter lacIq and the lac operator should inhibit the expression of oahA. LacI is a DNA binding protein which binds to the lac operator and inhibits transcription of subsequent genes [Bell, Lewis, 2001].
After assembly, the reactions were directly transformed into E. coli by heat shock. After selection on agar plates containing Kanamycin overnight, the surviving colonies were checked with colony PCR and subsequent gel electrophoresis. Unfortunately, no positive colonies had formed:
Figure 11: cPCR results. Expected size of the correctly assembled fragment was 4170 bp. However none of the 40 tested colonies contained the right fragment; most had a size of 500-1000 bp. The results proved our previous theory wrong; integrating a repressor system alone does not lead to survival of correctly transformed colonies. We suspected that one cause could be that we did not add a terminator to our oahA gene. To eliminate this problem we decided to try including the T7 terminator and see how this influences colony formation.
-
Integration of oahA with lacI repressor system and terminator
To design a functional integration vector, we decided to add the T7 terminator to our oahA unit. The terminator is supposed to end transcription after reaching the end of the protein coding sequence oahA which should decrease energy demand.
This time, the whole fragment was ordered from TwistBiosciences to eliminate any assembly mistakes. The fragment was amplified using Q5-PCR and cloned into the linearized pK19 vector. This time, pK19 was linearized with HindIII and BamHI.
The assembled pK19_lacI_oahA_TE was tested by transforming it into E. coli and testing for positive colonies with cPCR after selection on Kan50 agar plates. However, the gel electrophoresis of the PCR-products showed no positive results.
Figure 12: Results of E. coli pK19_lacI-oahA_TE cPCR. Expected length was 3642 bp -> no positive clones. We learned that adding a terminator alongside the lac-repressor system did not help with decreasing the metabolic burden on E. coli. After some further literature research, we found that a possible problem could be the high copy number of pK19 (~500). In comparison, the expression vector pEKEx2 we used in previous experiments only has a copy number of 30. Possible solutions could be choosing a different integration vector with medium copy number for E. coli or directly electroporating the assembly product into C. glutamicum for direct genome integration. Although not optimal due to low plasmid concentrations, this would eliminate the need for a plasmid that is suitable not only for C. glutamicum but also for E. coli. This will be tested in further experiments.
Conclusion
The failed genome integration remains to be solved. Possible solutions are:
- Using a different integration vector with lower copy number
- Directly electroporating the integration vector in C. glutamicum
Cultivation parameters need to be further tested to increase oxalate production. Apart from inducer concentration and induction time, we want to test different supplements (e.g. ethanol) and their influence on production.
Our main problem however is oxalate quantification. With the commercial oxalate assay kit, no reliable results could be obtained. To address this issue, we tried measuring oxalate using HPLC and even designed our own oxalate assay with a precipitation based approach.
Oxalate measurements
The graph displayed below illustrates a representative test using standards with known oxalate concentrations created in CGXII medium as their matrix. It compares both the expected values with those measured by our RevION assay.
The mean absolute error (MAE) of this calibration set was approximately 6,82%, which is an excellent outcome for the first successful validation. Moreover, the calibration equation achieved an R2 value of 0,9692, indicating that the model describes oxalate concentrations with a precision of roughly 97% relative to the measured absorbance data.
Leaching results
Our leaching experiments were performed under three different conditions A, B and C with oxalate as the leaching agent. After the leaching, the three samples and a reference samples as standard were sent to the labs of the polymer chemistry department of the university and analyzed there via ICP.
ICP, short for Inductively Coupled Plasma, is an analytical method which relies on spectroscopy to detect several metals of interest at once, like the PGMs rhodium, platinum and palladium in our case.
For doing so, the samples are placed into a plasma flame, which is very hot and electrically conductive. The heat disrupts all chemical bonds and so isolating the single molecules. So the PGM oxalate complexes are destroyed and the PGMs are then free to be detected. For detection and quantification, the PGMs are analyzed via mass spectroscopy.
| Sample | A | B | C |
|---|---|---|---|
| Leaching duration [h] | 5 | 6 | 8 |
In the reference sample, 0,02 mg⁄L rhodium, 0,94 mg⁄L platinum and 0,19 mg⁄L palladium were detected. That means a weight percentage of 0,03 of rhodium, 1,35 of platinum and 0,27 of palladium.
In the sample treated under condition A, 0 mg⁄L rhodium and palladium and 0,13 mg⁄L platinum were detected, which matches a weight percentage of 0,16 of platinum.
In both samples treated under conditions B and C, 0 mg⁄L rhodium and palladium and 0,03 mg⁄L platinum were detected, which matches in both conditions a weight percentage of 0,04 of platinum.
Statistical analysis of the results revealed for all three conditions revealed a standard error of around 0,02 and a r-value of nearly one.
Regarding our results so far, the need for further optimization of our experimental set-up becomes obvious. What immediately catches the attention is that for every condition, only one experiment and so one analysis has been performed. This of course makes our results look vague and unproven. In the time given, we couldn't perform more experiments. In the future, we will conduct more experiments and together with our consultants in the chemistry department improve our experimental set-up. We want to extend our set-up to specifically detect rhodium, palladium and other PGMs and to improve our leaching and analysis of each PGM.
References
1. Lange et al. (2018): Harnessing novel chromosomal integration loci to utilize an organosolv-derived hemicellulose fraction for isobutanol production with engineered Corynebacterium glutamicum, Microb Biotechnol., 2018 Jan;11(1):257-263 URL: https://doi.org/10.1111/1751-7915.12879
2. Bell, Lewis (2001): The Lac repressor: a second generation of structural and functional studies, Current Opinion in Structural Biology, Volume 11, Issue 1, 1 February 2001, Pages 19-25 URL: https://doi.org/10.1016/S0959-440X(00)00180-9