Experiments

We needed to amplify the sequences required for the ligation of our constructs (listed below) which were provided to us in iGEM Distribution. These sequences were in pre-cloned form, so we decided to transform them to Escherichia coli strains DH5α and NEB® 10-β and isolate them in higher concentrations viable for cloning. To examine if obtained plasmids contain the desired sequences, we made a restriction analysis by cutting these plasmids with BsaI, which was supposed to cut out sequences of interest. We then separated them using agarose gel electrophoresis.

Gel electrophoresis results:

Figure 2. Gel with restriction analysis of vector pUC, promoter, RBS, GFP and linkerDF

Figure 3. . Gel with restriction analysis of linkerDT4, linkerDT4tev, linkerT4T5, linkerT5E, linkerEF

Vector pUC and GFP sequences are long, so they could be observed as bands on the gel in figure 3. when cut with BsaI. Promoter, RBS and linkers are too short to be seen on the gel. We can deduce that BsaI cleavage occurred only because of the presence of a smear under main bands and the absence or fading of the 3 plasmid forms (linear, open circular and supercoiled), since after restriction mainly linear DNA should remain. However, it cannot be excluded that the smear might have been a result of using a greater sample volume than in the case of the uncut plasmids.

DNA concentration measurement results are presented in the Table 2.

Initially, we wanted to assemble all the expression vectors for overexpression of one esterase PnbA-DocScaB, four decarboxylases – Dcx1-DocXylY, Dcx2-DocXylY, Dcx3-DocXylY and Dcx4-DocXylY and scaffoldin ScfL, using Golden Gate assembly of all necessary parts (promoter, RBS, CDS, vector backbone and linkers). Later, we decided to order synthetized expression devices: PnbA-DocScaB device, EstJ6-DocScaB device (an esterase we included to the project later) and Dcx1-DocXylY device and use them instead. We could not afford to order a sequence as big as a scaffoldin with all the necessary regulatory sequences. Therefore, we tried to optimize Golden Gate assembly to obtain the expression vector of scaffoldin from following parts: Promoter (BBa_J435350), RBS (BBa_J435385), Scaffoldin CDS (BBa_K4695030), linkerDT4 (BBa_J435309), linkerT4T5 (BBa_J435349), linkerT5Ehis (BBa_J435369), linkerEF (BBa_J435361) and vector. In the third and final iteration, we used pJUMP 23-1A (BBa_J428347) as a vector backbone. For more information about previous cloning attempts, check the engineering cycle page.

Right after transformation, we did not yet have the PCR primers, so we could not check if transformed bacteria contain the desired sequences. Instead, we looked for transformants overproducing a protein with molecular weight equal to scaffoldin, namely 79 kDA.

Later on, we obtained primers allowing for verification of cloning with PCR and conducted the analysis. First, we did colony-PCR looking for just the ScfL CDS. Then we chose the best candidates and performed 14 PCR reactions with the following primer pairs:

• fJUMP_1_F + pJUMP_1_R
• fJUMP_1_F + RBS_348_R
• fJUMP_1_F + Scaf_R
• fJUMP_1_F + DT4_274_R
• fJUMP_1_F + T5E_332 _R
• fJUMP_1_F + Term_324_R
• Prom_296_F + pJUMP_1_R

Every pair was used in two reactions: one with template DNA and one without template DNA as a negative control. Sequences complementary to the primers are shown in Fig. 4. This way, we wanted to verify if all the parts assembled correctly.

Figure 4. . Scheme of pJUMP23-1A-scfL expression vector with primers used for verification of construct. The position of primers is shown in purple.

Figure 5. Gel from SDS-PAGE screening for ScfL production after IPTG induction on solid medium, gel 1.

Figure 6. . Gel from SDS-PAGE screening for ScfL production after IPTG induction on solid medium, gel 2.

We have not prepared negative or positive controls; therefore it is hard to analyze gels in Fig. 6. And 7. with certainty. Nonetheless, there are no significant differences between the lanes, especially around 79 kDa, so we can assume that overexpression did not occur in any of the samples.

Figure 7. . First gel from SDS-PAGE screening for ScfL production after IPTG induction in liquid medium.

Figure 8. . Second gel from SDS-PAGE screening for ScfL production after IPTG induction in liquid medium

On the gels in figures 7. And 8. in lanes with PnbA-DocScaB and Dcx1-DocXylY we can see distinct thick bands confirming that overproduction occurred in the experiment conditions. We can also see a distinct band around 27.8 kDa in lane from a fluorescent colony, probably coming from sfGFP, which is absent in the rest of the lanes confirming that sfGFP gene was absent, or at least non-functional, in non-fluorescent colonies. There are, however, no bands around 79 kDa indicating ScfL production in any lane.

Figure 9. . Colony-PCR of 50 colonies in groups of 5.

Unfortunately, we have not added a stain to the marker, which is why it is not visible in figure 9. Nevertheless, we can see that product was created in every lane, so there were sequences complementary to the primers in at least one colony in every group. There is always only a single band in every lane and always on the same height, so even though we cannot check its length, it is likely that scaffoldin sequence is present in every group of colonies. In the lanes on the right, the negative controls were separated, and no band is visible, confirming that all the visible bands are the products of PCR reaction.

Figure 10. . Colony-PCR of colonies 11, 12, 13, 14 and 15.

In every lane we can see a band between 500 bp and 400 bp marker bands. The expected PCR product length is 446 bp, therefore it shows that bacteria in colonies 11, 12, 13, 14 and 15 harbor the pJUMP23-1A vector with scaffoldin sequence. In the lane before the marker on the right, the negative control was separated and no band is visible, confirming that all the visible bands are the result of PCR reaction.

Figure 11. . Simulation of gel after PCR verification of ScfL ligation.

Figure 12. . Simulation of gel after PCR verification of ScfL ligation.

On the gel in figure 12. every control lane is empty, showing that all the bands come from the PCR reaction. In the lane with pJUMP_1_R reverse primer, there is a band that is only slightly higher than the simulation predicts (figure 11.). In the lane with RBS_348_R reverse primer, there is no band. This primer gave a very faint band in the PnbA-DocScaB vector analysis (figure 19.), so it might indicate that the primer is suboptimally designed. In the lane with Scaf_R reverse primer, there is a band that is slightly higher than expected; the difference is non-negligible but might be a result of, for example, some secondary structure formation. In the lane with DT3_274_R reverse primer, there is a ladder. The thickest band is much lower than expected, but there is one band at the proper height. This primer had the highest melting temperature of 62.7 °C. If the sequence has low specificity, it might cause relatively strong non-complementary annealing. Together, it might result in most of the product sequences being unspecific. In the lane with T5E_332_R reverse primer, there is also a ladder, but the main band is only slightly higher than expected. The ladder might be caused by the primer’s low specificity to its sequences. In the lane with Term_324_R reverse primer, there is a band on the expected height. In the lane with Prom_296_F and pJUMP_1_R primer pair, there is also a band on the expected height.

In conclusion, although it is still possible that this vector was assembled properly, this gel has too many issues to prove it. This analysis should be repeated and the primers might need to be redesigned. Alternatively, sequencing could solve this ambiguity. Unfortunately, we did not have enough time for that.

As mentioned before, we had troubles with Golden Gate assembly, so we ordered complete sequences of PnbA-DocScaB device (BBa_K4695210), EstJ6-DocScaB device (BBa_K4695211) and Dcx1-DocXylY device (BBa_K4695220) from IDT. We wanted to test the expression of our enzymes from popular expression vectors, but our constructs were designed to have BsaI restriction sites for Golden Gate assembly. That limited our ability to clone it to multiple cloning sites of most vectors. Considering that, we decided to try blunt-end cloning to the pBAD24 vector, as the synthetized sequences had blunt ends. Next, E. coli strain Rosetta(DE3) cells were transformed with the obtained ligation mixture. The correctness of the cloning was assessed by restriction analysis.

There were very few colonies after transformation with Dcx1-DocXylY and PnbA-DocScaB, and there were no colonies of EstJ6-DocScaB. Therefore, we did not do restriction analysis for vector expressing EstJ6-DocScaB.

Figure 13. Gel after blunt-end cloning restriction analysis.

Plasmid containing Dcx1-DocXylY sequence was expected to have 5603 bp. After cutting with EcoRI, there should be 3838 bp, 1366 bp and 333 bp fragments visible. Plasmid containing PnbA-DocScaB sequence was expected to have 6605 bp. After cutting with EcoRI, there should be 1177 bp and 5428 bp fragments. On the gel in figure 13. in the lanes with uncut plasmid, bands are much higher than expected, which might indicate that ligation occurred incorrectly. Perhaps, multiple linearized vectors have connected to each other. There are no 1366 bp and 1177 bp bands, which indicates that there were no accessible EcoRI restriction sites in the plasmids, which suggests that cloning did not go as expected.

Vector pUC (BBa_J435330) was the first vector we had chosen as a potential expression vector for our devices, as it is a part of the E. coli expression kit collection in the iGEM Distribution Kit. We cloned sequences of each device encoding Dcx1-DocXylY, PnbA-DocScaB and EstJ6-DocScaB to the pUC vector using Golden Gate assembly. Then, we transformed E. coli strain BL21(DE3) with the obtained plasmid.

During proper Golden Gate assembly sticky ends of ligated DNA fragments should assemble into sequences recognized by SapI restriction enzyme. Therefore, we assessed the correctness of the ligation by restriction analysis with SapI. It should result in the formation of two bands on the agarose gel with the following lengths:

• Dcx1-DocXylY: 1764 bp and 1042 bp
• PnbA-DocScaB: 1764 bp and 2044 bp
• EstJ6-DocScaB: 1764 bp and 1417 bp

We used two negative controls, one was an undigested plasmid, and the other was treated with BsaI which should not recognize any sequence on the vector after proper assembly.

Figure 14. Restriction analysis of the vectors for overexpression of PnbA-DocScaB, Est6J-DocScaB and Dcx1-XylY with vector pUC backbone.

On the gel in figure 14. we can see all the expected bands in lanes treated with SapI which indicates, that SapI restriction sites were successfully formed. Lanes with uncut plasmid and treated with BsaI look the same which indicates that BsaI restriction sites were successfully removed from the vector pUC. In figure 2. we can see that the unmodified vector pUC is cut by BsaI. We can conclude, that the assembly was carried out correctly.

Originally we planned to conduct overexpression of our devices from vector pUC (BBa_J435330), but we came to conclusion, that this high copy number plasmid can be suboptimal for overexpression, so we decided to try another vector. While looking for alternatives we found JUMP part collection in iGEM Distribution Kit, consisting of various plasmids and decided to test a few of them. We chose the plasmids listed in table 9. All of them had constitutively active sfGFP gene. We transformed them do DH5α and inoculated overnight liquid culture with transformants to see how intensely they would fluoresce.

Table 9. pJUMP plasmids we chose and transformed

Figure 15. Fluorescence comparison between bacteria with different pJUMP vectors. From the left: pJUMP23-1a, pJUMP26-1A, pJUMP28-1A, pJUMP29-1A.

The most intense fluorescence was observed in pJUMP28-1A and the least intense in pJUMP23-1A. Because sfGFP has been expressed constitutively and is a protein with relatively low requirements, we assumed, that fluorescence intensity in this case is correlated with plasmid copy number. We were looking for a low copy number vector, as we expected it to work better for overexpression. Therefore for future use we chose pJUMP23-1A.

We had concluded that pJUMP23-1A would be the most promising available vector for the expression of our devices, therefore we cloned the devices encoding Dcx1-DocXylY, PnbA-DocScaB and EstJ6-DocScaB ordered from IDT into this plasmid. Then we transformed E. coli strain BL21(DE3) which is tailored for gene overexpression. Next, we isolated the plasmid and conducted fast verification by restriction analysis followed by more complex PCR verification. We performed 16 PCR reactions with the following primers pairs:

• fJUMP_1_F and pJUMP_1_R
• fJUMP_1_F and RBS_348_R
• fJUMP_1_F and Est_1_R
• fJUMP_1_F and DT4_357_R
• fJUMP_1_F and Ock_2_R
• fJUMP_1_F and T5E_332_R
• fJUMP_1_F and Term_324_R
• Prom_296_F and Est_1_R

Every pair was used in two reactions: one with template DNA and one without template DNA as a negative control. Sequences complementary to the primers are shown in figure 14. This way we wanted to verify if all the parts were assembled correctly.

Figure 16. Scheme of pJUMP23-1A-pnbA-docscaB expression vector with primers used in this experiment shown in purple.

Figure 17. Verification of proper pJUMP23-1A vector constructs assembly with restriction analysis

We can observe all of the expected bands, except for 303 bp in PnbA-DocScaB, which probably were too short to be visible or have been washed out of the gel. This suggests that all three vectors were assembled correctly.

Figure 18. Simulation of gel with PCR analysis of pJUMP23-1A-pnbA-docscaB vector assembly in SnapGene

Figure 19. 1% agarose gel with PCR analysis of pJUMP23-1A-pnbA-docscaB vector assembly

All the expected bands appear on the gel at the correct height confirming, that the vector has been assembled correctly.

We managed to obtain E. coli clones with vectors encoding PnbA-DocScaB, EstJ6-DocScaB and Dcx1-DocXylY with vector pUC (BBa_J435330) or pJUMP23-1A (BBa_J428347) backbones. We tried overproducing these proteins overnight at 16 °C on a small scale to see if there was any difference in the abundance of enzymes between vector pUC and pJUMP23-1A on the gel. After discovering that pJUMP23-1A gives better yield, we also tried more intense 3-hour overproduction at 37 °C with CaCl2 supplementation, since dockerin domains have binding pockets for calcium ions, determining their folding. Therefore we wanted to see if it could enhance the overproduction.

Figure 20. SDS-PAGE after overnight overexpression from vector pUC.

Figure 21. SDS-PAGE after overnight overexpression from pJUMP23-1A.

Figure 22. SDS-PAGE after 3.5 hour overexpression from pJUMP23-1A.

Our fusion proteins are expected to have following molar mass:

• PnbA-DocScaB – 64,6 kDa
• EstJ6-DocScaB – 42,0 kDa
• Dcx1-DocXylY – 28,4 kDa

Comparing gels from overnight overexpression we can see that expected bands are much thicker in samples from pJUMP23-1A than those from vector pUC. In EstJ6-DocScaB case there isn’t even a visible band in samples from vector pUC. Also, bands heights seem much more accurate in pJUMP23-1A which might indicate that enzymes in cells with vector pUC didn’t have correct conformation. It might be caused by high copy number of vector pUC, which is often disadvantageous for protein overexpression. Meanwhile pJUMP23-1A has medium copy number. Coming to these conclusions, we decided to use vector with pJUMP23-1A backbone for protein production.

Comparing gels from overnight and 3.5-hour overexpression we can see that induced bands are much thicker in samples from overnight overexpression. Again, there is no visible band for EstJ6-DocScaB in samples from 3.5-hour overexpression. The OD600 after overproduction reached 0.7, which was quite low. Perhaps we added too much of CaCl2. Therefore, we decided to carry out overnight overexpression for production of these enzymes.

After obtaining bacteria with functional expression vector encoding PnbA-DocScaB and Dcx1-DocXylY we needed to induce overexpression of those sequences and harvest proteins on a big scale. After that, we set to purify the protein using immobilized metal affinity chromatography (IMAC). We used Ni-NTA resin, since our proteins were His-Tagged at the C-terminal end.

Figure 23. SDS-PAGE gel containing samples before and after induction of a large scale (3 liters each) overproduction of PnbA-DocScaB and Dcx1-DocXylY proteins. Every sample is doubled on the gel.

Induction of both cultures was successful, as additional bands appear on the gel after adding IPTG and incubating overnight. The thickness of the bands is not as big as in the test overproduction, most probably because of the late induction. The PnbA-DocScaB band is on the expected height (64,6 kDa), but the band expected to be Dcx1-DocXylY (28,6 kDa) is much lower, around 20 kDa. This indicates, that the protein was not expressed properly. Lack of around 7 kDa can correspond do the molecular weight of the dockerin domain, which is 7 kDa as well.

Figure 24. SDS-PAGE gel containing samples of lysates, supernatants and pellets of the bacteria harboring overproduced proteins PnbA-DocScaB and Dcx1-DocXylY. This was done to establish which fraction should be loaded onto the Ni-NTA resin. The gel used was flawed, but it is still possible to gather data from the results.

Results indicate, that most of the PnbA-DocScaB is located in the supernatant and most of the overproduced protein from the pJUMP23-1A-dcx1-docxylY is located in the pellet. We opted to purify the supernatant with PnbA-DocScaB and due to the lack of time needed for optimization and purifying proteins from pellet we also decided to try to load the potential Dcx1-DocXylY supernatant onto the Ni-NTA resin, as there is a small band of overproduction in the supernatant sample.

Figure 25. SDS-PAGE gel containing samples from different stages of Ni-NTA purification of PnbA-DocScaB.

It can be observed, that the PnbA-DocScaB band (64,6 kDa) visible in the load sample is much less abundant in the flow-through fraction, which indicates binding of the PnbA-DocScaB to the resin. Wash (50 mM imidazole) eluted most of the non-specifically bound proteins. Imidazole gradient (50-500 mM) proved to be effective in eluting the desired protein. At a concentration of around 150 mM of imidazole, a thick band with height corresponding to the overproduction band in the first path started to appear on the gel. The purity and the concentration of eluted samples are decent.

Figure 26. SDS-PAGE gel containing samples from different stages of Ni-NTA purification of Dcx1-DocXylY.

It can be observed, that most of the overproduced protein is still visible in the flow-through fraction. This, in addition to the lack of any specific bands at the expected kDA in next steps, shows the absence of desired protein in the supernatant load. In 10th and 12th fractions two bands at around 70 kDa height can be observed being eluted by higher concentrations of imidazole. Those are undesired proteins, which bind more specifically to the Ni-NTA resin, possibly due to higher histidine content.

Since we incorporated TEV protease recognition site between the enzymatic domain PnbA and dockerin domain DocScaB we wanted to obtain the native form of PnbA. It was supposed to be used in the enzymatic analysis to compare its activity with the fusion protein. Unfortunately we didn’t have enough time to conduct such analysis. After the TEV protease cleavage we tried to separate the native enzyme from cleaved dockerin domain on the Ni-NTA resin taking advantage of the fact that the His-tag remained linked to the dockerin.

Figure 27. SDS-PAGE gel with fractions obtained from enzymatic digestion of PnbA-DocScaB cut with TEV protease.

In the path with purified PnbA-DocScaB we can see a thick band of the protein with little contaminations. In the path with the post-reaction mixture (PnbA-DocScaB + TEV protease) we can see a thick band of PnbA and a band of cleaved DocScaB, both on the expected height, proving that protein was successfully cleaved. We can also see a distinct band at the same height as uncut enzyme, meaning that the cleavage was not complete and in the mixture remained significant amounts of whole PnbA-DocScaB. In the path with Ni-NTA flow-through we can see all of the bands that we could observe in the post-reaction mixture, meaning that the purification was not very efficient. In the path with Ni-NTA elution we can once again see all 3 bands of PnbA-DocScaB, PnbA and DocScaB, but thinner, especially the PnbA. It means that even though the purification was not satisfactory some of the cleaved dockerin domain was removed. Surprisingly PnbA was also present in the elution, meaning that, even though His-tag should have been removed together with dockerin domain, native form of PnbA seems to still have some affinity to the Ni-NTA resin.

Analysis of the purified protein on SDS-PAGE is conducted in denaturing conditions, so no information about protein folding and activity is obtained. After purification of PnbA-DocScaB, we wanted to characterize the acquired enzyme in terms of its stability and tertiary structure.

Usage of Tycho is very simple. According to the protocol, that guides you when using the system, in order to measure our sample we simply left it to reach the room temperature, and then transferred a fraction of the purified protein solution to the dedicated capillary and inserted the capillary into the device. The analysis was done, data were collected and analyzed.

Figure 28. Graph showing the relation of temperature with the ratio of intrinsic fluorescence at 350 nm and 330 nm, which directly translates to the folding state of measured protein.

Tycho allows to see how the rise of the temperature affects the tertiary structure of proteins in the sample, by measuring the intrinsic fluorescence of tryptophan and tyrosine residues. The higher the ratio, the more residues are emitting light and the more unfolded the protein is. Results of this particular experiment show a big unfolding event occurring at around 54 °C and a smaller one at around 63 °C. Event at 54 °C might correspond to the unfolding of the enzymatic PnbA domain, and the 63 °C can be associated to the unfolding of the smaller dockerin domain. This shows, that the purified PnbA-DocScaB protein indeed has a functional tertiary structure and loses stability at >50 °C.

In order to measure the concentrations of substrates and products of enzyme reactions in High-Performance Liquid Chromatography (HPLC) we needed to plot a calibration curve. We worked on two different chromatographs, therefore we made the calibration curve twice.

We needed to estimate low concentrations near the end of the calibration curve concentration range. Therefore, considering that no peak on the chromatogram was observed with no substance added, we decided to set the calibration curve intercept at (0,0).

Figure 29. Calibration curves for the first chromatograph with intercept set at (0,0).

Table 13. Properties of calibration curves

Calibration curves plotted for PA. MBP and BA on the first chromatograph have coefficient of determination (R2) >0.99, giving quite a good fit. Calibration curves plotted for DBP and BB on the other hand have quite low R2 therefore values calculated using them would not be very reliable.

Figure 30. Calibration curves for the second chromatograph with intercept set at (0,0).

Table 14. Properties of calibration curves

All calibration curves plotted on the second chromatograph have R2 equal or higher than 0.99 giving a decent fit. Surprisingly, peak areas for BB are higher than on the first chromatograph, even though they were measured at wavelength of 205 nm, at which BB has lower molar absorption coefficient than at 225 nm used in the first chromatograph.

After purifying PnbA-DocScaB we wanted to test its activity. The analyses were done in 20% ACN in 10 mM PBS (pH = 7.4) to increase substrate solubility. We used dibutyl phthalate (DBP) and butyl benzoate (BB) as substrates. This experiment was carried out in triplicate for both DBP and BB. Enzymatic reaction was stopped by centrifugation on a membrane with 10 kDa weight cut off. This also separates proteins from the rest of the sample, preventing them from being loaded onto the HPLC column. Concentrations of the compounds were calculated using the second callibration curve (figure 30.)

Figure 31. Mean DBP concentration in samples after centrifugation on Amicon, measured at 205 nm.

Figure 32. Mean MBP concentration in samples after centrifugation on Amicon, measured at 205 nm.

Figure 33. Mean BB concentration in samples after centrifugation on Amicon, measured at 205 nm.

Figure 34. Mean BA concentration in samples after centrifugation on Amicon, measured at 205 nm.

We can see that the concentration of substrate in control sample is around ten times lower than 400 µM in DBP in time 0 and almost four times lower after 30 minutes. In the case of BB it even dropped below the calibration curve concentration range. It indicates that this method requires redesigning. Amicons might be the potential cause of the concentration decrease. It requires investigation if these compounds interact with the filter. This experiments should be repeated with different method of reaction stopping and enzyme removal. We can see that in the control sample there is no MBP but in the samples with PnbA-DocScaB there has been a clear signal from MBP on the chromatogram. Similarly in the experiment with BB as a substrate in the control sample there have been minimal peak from BA, but there have been a significant signal from BA in the sample with the enzyme. It should be kept in mind that we cannot say with confidence what was the concentration of the product in the sample, since substrate concentration measurements in control samples turned out to be unreliable. Nonetheless this data proves that PnbA-DocScaB catalyzes reactions of ester bond hydrolysis in both DBP and BB. Though there was no signal from PA, therefore we cannot say if this enzyme catalyzes hydrolysis of MBP in the experiment conditions.

We can see that with sampling time equal 0 minutes product was already created. It indicates that reaction stopping by centrifugation on Amicon was too slow to observe the change in reagents concentration in time. If we were to repeat this experiment we would try stopping the reaction with acid as it was dome by Xu et al. (Xu Youqiang, 2021). Reaction mixture setting and sample preparation have been done in room temperature, therefore we can see that enzyme was functional in this temperature. Concentration of substrate significantly changes after 30 minutes in both control and research sample. Moreover it increases for DBP and decreases for BB. These unexpected changes in substrate concentration might be connected to the temperature change during incubation, since DBP and BB did not form homogenous mixture in 20% ACN in PBS, so temperature rise might have changed the distribution of the substrates between phases.

It is important to point out that after 30 minutes of incubation white flakes precipitated from the reaction mixture. It was probably denatured enzyme. It shows that PnbA-DocScaB has rather low stability in 20% ACN in PBS in these temperature conditions. Fast denaturation of the enzyme also gives plausible explanation for lack of product concentration rise after 30 minutes of incubation.

In conclusion, although this experiment failed to give data about reaction kinetics and enzyme activity it still confirms that PnbA-DocScaB catalyzes ester bond hydrolysis in MBP and BB in room temperature.

u Youqiang (2021) ‘An efficient phthalate ester-degrading Bacillus subtilis: Degradation kinetics, metabolic pathway, and catalytic mechanism of the key enzyme’, Environmental Pollution, 273, p. 116461. Available at: https://doi.org/10.1016/j.envpol.2021.116461.