Team:Wageningen UR/Engineering
Engineering
Driven to provide a solution to the harmful ammonia and methane emissions from cattle, we invented Cattlelyst. After extensive research and reaching out to various stakeholders, we decided that Cattlelyst would be a biofilter
containing bacteria that can convert these gasses into harmless compounds. We further built on this idea by designing our biological system, which is divided into three pillars: 1. Methane conversion,
2. Ammonia conversion and 3. Safety. For each of these pillars we went through one or multiple engineering cycles (Research - Design – Build – Test – Learn) until our final design.
We began by
researching the required conversions of methane (CH4) into carbon dioxide (CO2) and ammonia (NH3) into dinitrogen gas (N2) and how they are performed in
nature. More information about this can be found in Background. Next, we explored candidate bacteria for our biofilter. In this page we will discuss why we chose Escherichia coli (E. coli)
and Pseudomonas putida (P. Putida). Next, we will discuss how we went through the engineering cycles to design, build, test and learn from methane and ammonia conversions in these organisms and the
accompanying safety system of our project.
Chassis bacteria for co-cultures
While designing the biofilter we came across several requirements for candidate bacteria. The biofilter will be situated in an aerobic environment, so the bacteria should be well adapted for living in such an environment. Additionally, as we aim to reduce methane as well as ammonia emissions, both microbial conversions should happen. There is not any known bacterium capable of performing both conversions, and natural organisms with the required pathways are not equipped to perform the conversions in Cattlelyst, which will be explained further down. Therefore, bacterial chassis are needed in which the pathways can be engineered. These bacteria should be non-pathogenic and genetically accessible. After extensively reading relevant literature and orientating ourselves on possible mechanisms we have come up with two candidate bacteria that will be the chassis for our project.
Co-culture
Engineering the co-culture
Having selected the bacterial chassis for the methane and ammonia conversions in Cattlelyst, we needed to design our approach for building in these pathways and making them work. As mentioned above, we have organized our project into three pillars: 1. Methane conversion, 2 Ammonia conversion and 3. Safety. During the next chapters we will discuss the engineering cycles we have gone through with each pillar during our project
Cycle 1
Design
For our biofilter, we wanted to convert ammonia into dinitrogen gas. In the initial research phase, we read about newly-discovered bacteria that can complete this conversion aerobically, which is
perfect for its application in Cattlelyst. The mechanism they employ is referred to as coupled heterotrophic nitrification-aerobic denitrification, in short HNAD. Unfortunately, HNAD bacteria
have evolved to survive in completely different conditions to ours. Therefore, we needed to engineer the HNAD pathway into a microbe that does fit our conditions. At this moment, the exact mechanism
explaining the conversion is poorly understood. To gain insight in the HNAD phenomenon and build the groundwork for our future wet lab projects, we decided to develop a dynamic model. More explanation
can be found on the nitrogen conversion model page.
Build
First, the model structure was established based on pathways suggested by Wehrfritz et al. [11] and Joo et al. [12]. published experimental data for Pseudomonas stutzeri strains YZN-001
[13] and XL-2 [4], and (non-)enzymatic conversions reported by Caranto et al. [14]. Then, the mechanism was represented mathematically as a system of ordinary differential equations (ODEs).
Subsequently, the model was collectively fit to agree with time-series data for Pseudomonas stutzeri strains YZN-001 [13] and XL-2, validating this approach [4].
Test
The final model could capture the nitrogen dynamics for P. stutzeri strains YZN-001 [13] and XL-2 [4] well. We further validated the model by comparing the simulated nitrogen dynamics with
data for P. putida ZN1 [15]. Moreover, we found that when the cells are exposed to nitrate, the harmful intermediate nitrous oxide accumulates in the medium. Therefore, we identified
parameters responsible for this accumulation. Lastly, given that the data is obtained in high ammonia conditions, we simulated the system for low ammonia concentrations, to reflect the biofilter
conditions better.
Learn
We learned that the model was able to predict nitrogen dynamics for different conditions and different strains. This suggested that the proposed pathway structure underlies the HNAD phenomenon.
Moreover, we learned that with HNAD we can risk to accumulate nitrous oxide. With the model, we identified the maximum conversion rate of nitrous oxide reductase to be most interesting to optimize
for in vivo. Lastly, we learned that the ratio of nitrogen that goes to biomass and HNAD does not change in low ammonia concentrations. This suggests that HNAD could be a growth-coupled process,
which could explain the difference between autotrophic nitrifiers and heterotrophic nitrifiers. However, the HNAD behavior is rarely studied for in low ammonia concentrations, thus more research
is required to validate these model predictions.
Figure 1: Model simulation of extracellular N-species for best parameter set. The continuous line corresponds to model simulations, the squares are time-series data. Upper panel details
dynamics for extracellular nitrogen species in condition for P. stutzeri strain YZN-001. Pink: [ammonia]ex , Red: [dinitrogen gas]ex, Green: [nitrate]ex,
Blue: [nitrite]ex, Yellow: [nitric oxide]ex, Black (top figure) [Nitrous oxide]ex. Lower panel details dynamics of the population volume in condition for P. stutzeri
strain YZN-001; Continuous line: population volume, Squares: data for population volume.
Cycle 2
Design
After gaining in-depth knowledge on HNAD through modeling, we designed this pathway as such that it could be built into the model organism P. putida. A schematic overview of the
pathway we built can be seen in Figure 2 below and more information about it can be found in Background.
Figure 2: Illustration of the pathway for the conversion of ammonia. The orange boxes indicate the enzymes. AMO = ammonia monooxygenase, HAO = Hydroxylamine oxidoreductase, Nap = periplasmic nitrate
reductase, Nir = nitrite reductase, Nor = nitric oxide reductase, Nos = nitrous oxide reductase.
Since the genes involved in heterotrophic nitrification are not completely clarified yet, we used the genes from the autotrophic nitrifier Nitrosomonas europea to design a synthetic nitrification pathway.
These genes, AMO and HAO, were codon harmonized for heterologous expression in P. putida.
For denitrification, multiple approaches were chosen to increase the chances of success. The first encompassed engineering all four denitrification enzymes separately, each with a synthetic regulation.
With this mosaic approach, the denitrification pathway can be optimized in such a way that no intermediates accumulate. Additionally, to ensure optimal conversion of nitrate/nitrite to dinitrogen gas,
we wanted to compare the activity of enzymes of different native aerobic denitrifiers.
The second approach, called the plug and play approach, included introducing the entire denitrification into P. putida at once. To prevent nitrous oxide formation, we designed a way to
redirect the electron flux from the aerobic respiration machinery to the denitrification machinery. By rewiring the electron transport system, the likelihood that insufficient electrons are supplied
to the denitrification machinery, which consequently results in nitrous oxide production, will be decreased. To achieve this, a CRISPR interference (CRISPRi) system was designed to downregulate the
terminal oxidases of P. putida to reduce expression of terminal oxidases, and whith that reduce the amount of electrons utilized in oxygen respiration.
Build
To engineer the enzymes for nitrification and denitrifications separately into P. putida, the encoding genes for each enzyme were cloned into a plasmid. To accomplish this, genes for
denitrification from multiple native denitrifiers were used.
The plug and play approach for denitrification was carried out by cloning all the required genes originating from one organism and placing them in a single plasmid, resulting in the construction of
a bacterial artificial chromosome (BAC). Subsequently, the entire denitrification pathway was integrated in the genome of P. putida. In parallel, various CRISPRi spacers targeting
the five terminal oxidases were made.
Test
Once all the plasmids for nitrification and denitrification were successfully constructed and transformed into P. putida. With all four denitrification enzymes, experiments were performed to quantify the activities in vivo, by measuring the amount of substrate consumed or the amount of product
made by P. putida. The first two enzymes of the denitrification pathway, Nap and Nir, showed the desired activities. Moreover, when combining these enzymes, no harmful intermediates
accumulated anymore, suggesting that the concatenated conversions work (figure 3). When testing the last two enzymes of the pathway, no activity was observed. However, due to time constrains only one
experiment was performed with these enzymes. Unfortunately, due to time constrains the nitrification enzymes
could not be tested.
Figure 3: NO2- production by P. putida EM42 ∆nasT containing the nap operon originating from P. denitrificans on plasmid (Nap) or containing the nir operon origination
from P. stutzeri on plasmid as well (Nap + Nir). The NO2- production was corrected for the change in OD over the time span of 24 hours.
For the plug and play approach, the entire denitrification pathway was tested by quantifying the amount of nitrate that was converted. However, this assay was not sensitive enough to conclude
whether the pathway was active. Therefore, experiments were performed to measure if intermediates of the pathway accumulated. These tests showed that a bit of nitrite and trace amounts of nitrous
oxide accumulated, suggesting that the first three enzymes of the pathway worked. However, more research is needed to derive conclusions on the efficiency of the pathway as a whole.
In-parallel with the plug and play approach, 15 CRISPRi plasmids that downregulate P. putida’s terminal oxidases were developed. The effect of downregulating these essential genes
was quantified by comparing growth curves. If downregulation of a terminal oxidase impairs growth significantly, it would be a good candidate to redirect the electron flux towards the
denitrification machinery. All spacers were tested thoroughly,but only for the spacer targeting the promoter of the Cyo oxidase, we found a growth defect during the exponential phase. Measuring the effect of downregulation of multiple terminal
oxidases at the same time should be the next step. However, due to time constraints, a so-called multiplex CRISPRi system could not be developed.
Learn
For the mosaic approach of denitrification we showed that the first two enzymes, Nap and Nir, are active separately and combined in P. putida, but the Nor and Nos did not work individually
yet. Nor was not yet tested with its accessory proteins due to time constrains. However, there are suggestions that Nor works for the
plug-and-play P. putida strain, given that trace elements of nitrous oxide accumulate. This slight accumulation does not exclude the possibility that Nos is active in the plug-and-play
strain. For the mosaic approach, we learned that even though Nos was expressed with accessory proteins, it is still not enough to result in an active enzyme. Lastly, we learned that downregulation
of the promoter of the Cyo oxidase with one CRISPRi spacer could impair growth during the exponential phase. All in all, more research is needed for successful nitrification and denitrification in P. putida.
Cycle 1
Design
As discussed in the Methanotroph section in Background, the consumption of methane is facilitated by two catalytic reactions. The enzymes sMMO and pMMO both
convert methane into methanol and Mdh subsequently converts methanol into formaldehyde (Figure 4). This formaldehyde can be assimilated into biomass. In this project, we used two E. coli
strains used that are capable of this [3], [16] Together with Mdh and pMMO or sMMO these E. coli strains are able to either grow solely on methane as carbon source or are auxotrophic for
methane. For a more in-depth explanation, consult the Wetlab MMO page.
Figure 4: Methane conversion pathway. Methane is converted to methanol via soluble or particulate methane monooxygenase (sMMO/pMMO). Methanol is subsequently converted to formaldehyde via
methanoldehydrogenase (Mdh). Formaldehyde/formate can be incorporated into biomass or converted into carbon dioxide.
In addition to using literature as a sole source of inspiration, we also used our in-house developed tool PIPE. The iGEM PIPE was used to model the conversion of methane into a high-value compound,
in this case the production of L-lactate, a commercially relevant compound [17]. Our tool also suggested using methane monooxygenase as well as methanol dehydrogenase in E. coli , when
this strain was forced to both grow on methane and produce L-lactate (see the PIPE test case).
Build
To build these different approaches we started in the lab with two tasks. First, sMMO and pMMO were constructed on plasmids and transformed into E. coli. For the pMMO we followed the method
as described in [18] and for the second enzyme, sMMO the building was unsuccessful and the testing phase was not reached. However, this was not essential for the completion of this pilar as we used
a parallel strategy, as pMMO and sMMO are similar in function.
Additionally, we engineered the strain already auxotrophic for formaldehyde to be auxotrophic for methanol, produced by MMO. This was established by incorporating Mdh into the pathway, which
converts methanol into formaldehyde and finally, biomass.
Test
The pMMO was tested in two different ways. First it was tested in vivo together with duroquinone as additional reducing power. No production of methanol was observed when E. coli
containing the pMMO was cultured in sealed bottles with methane in the headspace. Secondly, the pMMO was isolated and tested in an in vitro enzyme assay with duroquinone as additional
reducing power [18]. No production of methanol was observed when measuring with GC. We also tested the formaldehyde
auxotrophic strain containing the Mdh construct and we showed that this strain is now able to grow with methanol as sole C1 source. Thus, this strain is now auxotrophic for methanol, which is
produced by MMO.
Learn
What was the reason for not being able to clone the sMMO, and why was the pMMO not active? These very important questions remained unanswered as of yet, but there are several possibilities.
One is that the duroquinone was not activated properly. As the protocol used did not contain enough detail to fully replicate it, assumptions had to be made regarding exact quantities.
The cloning method for sMMO required several plasmids and as such, many cloning steps and Gibson assemblies are required. Unfortunately within the span of this project it was not possible
to clone sMMO. Yet, from literature we learnt that because of the high sensitivity of pMMO it is a better fit for our project as we deal with low methane concentrations. Although it is suggested
in literature that pMMO could be expressed recombinantly [18]. this has not been done in vivo , which is a requirement for our biofilter. As such it is important to continue engineering
both pMMO and sMMO.
Cycle 2
Design
After the months spent in the lab and the iGEM competition, we will continue the cloning of sMMO and chaperones and also make some alterations to the pMMO plasmid. The final approach is described
in this page, but we are confident that both enzymes are great options, each having their own merits and problems. Additionally, upon completion of the successful expression of this
enzyme it will be incorporated in the genome of an C1 growing strain and as such, create a completely synthetic methanotroph.
Test
After having recombinantly expressed pMMO in E. coli, in addition to testing in vivo, we will also retest this version in vitro and use additional reductants besides duroquinone.
Kill-switches: Cycle 1
Design
As we are using GMOs in our biofilter, we must ensure biocontainment. Therefore, we designed multiple safety circuits, which make sure both bacteria are killed outside the biofilter. Read more
about this on the Safety by design page! We started developing our first safety system in the synthetic methanotroph E. coli by basing its survival on the higher methane concentration
inside the biofilter than its surroundings. As there are no methane receptors in the cell, we had to look a bit further down the methane oxidation pathway and found out that E. coli can
detect the formaldehyde concentration in the cell sensitively by the FrmR protein [19]. We coupled this to a toxin-antitoxin system to create a system where a low formaldehyde concentration
(i.e. escape from the biofilter) triggers the production of a toxin that kills the cell. We used the hok/sok system, as this works on mRNA basis, lessening the metabolic burden on the cell [20].
A detailed explanation of this system can be found on the methane-dependent kill switch page.
Build
We started developing the methane-dependent kill switch by making a ‘toy model’ of the system. Using data from literature, we estimated part of the parameters of the full circuit. We then built the
full model using ordinary differential equations (ODEs) to test the dynamics of the system. Specifically, we tested if the system is sensitive enough to the small differences in formaldehyde
concentrations in the cell based on the methane concentrations in the real-life situation of the cow stable (Figure 5).
Figure 5: The biosafety circuit which is supposed to kill the microorganism if the formaldehyde concentration in the cell is low. When the formaldehyde concentration is high, formaldehyde
will bind to the FrmR protein which will then dissociation from the Pfrm promotor, thereby allowing transcription of LacI and antitoxin mRNA. LacI will repress the production of toxin mRNA. When
the formaldehyde concentration drops, there is more free FrmR that binds to the Pfrm promoters. This will inhibit the transcription of the antitoxin and LacI. When there is less LacI present,
the transcription of toxin mRNA will not be inhibited anymore. The organism will then be killed due to the production of the toxin that irreversibly damages the cell membrane of the organism.
Test
We used the model to simulate the dynamics of the methane-dependent kill switch and found out that the formaldehyde sensitivity of the circuit is too low. This means that the difference in formaldehyde
concentrations does not result in a sufficiently high hok/sok ratio to kill the cells if they escape. Therefore, we designed and modeled possible circuit extensions to improve the formaldehyde
sensitivity. We found out that some of these could achieve the desired increase in formaldehyde sensitivity, but not all of them would be possible to build in the lab.
Learn
From the model, we learned that the formaldehyde sensitivity of the methane-dependent kill switch needs to be improved to ensure cell death in low formaldehyde conditions. We found out that we need
a second input signal on toxin production in the methane-dependent kill switch to improve the dynamics of the circuit. This can be based on a second characteristic of our biofilter: cell density of
the bacteria. Inside the biofilter, there is a dense co-culture, so the cell density is higher than outside our biofilter.
Another possibility to increase formaldehyde sensitivity is to provide bigger differences in formaldehyde concentrations in the cell. In literature we found that E. coli strains with knockouts
of the formaldehyde detoxification pathway can tolerate lower formaldehyde concentrations. Therefore, creating these knock-outs could help in increasing the formaldehyde sensitivity of the
methane-dependent kill switch.
Kill-switches: Cycle 2
Design
We redesigned the methane-dependent kill switch using a hybrid promoter to control toxin production, in which both
methane concentration and cell density serve as input signal [21]. This has the
advantages of an expected increase in the formaldehyde sensitivity of the system, and additionally creates a solution to a problem we encountered during our human practices work. If the cows leave
the stable to graze outside, the methane concentration in the biofilter temporarily drops, but the methane-dependent kill switch should not be activated to kill E. coli. By using the
hybrid promotor, an “AND gate” is created, so that toxin production is only activated when both methane concentration
and cell density are low. This way, E. coli can survive a temporary drop in methane concentration, but is killed when escaping the biofilter, still warranting the safety of the system.
To ensure biocontainment of P. putida, we designed a cell density-dependent kill switch. Quorum sensing is a mechanism can be used to ‘sense’ the cell density [22], and will be coupled to the
hok/sok toxin-antitoxin system again. This mechanism relies on the production of the quorum sensing molecule AHL in the cell, which rapidly diffuses in and out of any cell and through the medium.
Therefore, the intracellular AHL concentration will only be high in high cell densities. Toxin production will be placed under the lux pL promoter, which is repressed by AHL. This way,
toxin production in repressed in high cell densities (inside the biofilter), and repression is lifted in low cell densities (outside the biofilter), leading to toxin production and cell death.
As the lux pR promoter is activated by AHL, antitoxin production is put under control of this promoter so antitoxin is produced in high cell densities.
With the design of these two kill switches and the hybrid promoter, we ensure cell death of both bacteria upon escape of the biofilter.
Build
We worked on building both kill switches as genetic circuits in the lab. We replaced the toxin and antitoxin by fluorescent reporters to prevent cloning difficulties and facilitate testing the
dynamics of the systems because fluorescent proteins are easier to measure and do not result in cell death. The proof of concept of the cell density dependent kill switch was built in both
E. coli and P. putida and the methane-dependent safety circuit was developed in E. coli.
Test
We tested the cell density dependent kill switch in the lab using plate reader experiments and a coculture experiment on agar plates and saw a high sensitivity of the circuit to the cell density
(Figure 6). This indicates that quorum sensing is a good input signal to use for the safety circuit. See more details here.
Additionally, we tested the toxicity of formaldehyde on strains with a knock-out for the formaldehyde sensing and detoxification pathways. We found out that in wild-type E. coli,
of the detoxification enzymes results in a higher toxicity of formaldehyde.
Figure 6: AHL sensitivity of the E. coli strain containing the reporter-pL-GFP and reporter-pR-RFP plasmids. The sensitivity is calculated by dividing the GFP/RFP ratio without AHL by
the GFP/RFP ratio with AHL.
Learn
We learned that the quorum sensing mechanism is a good input signal to base the cell density dependent kill switch in P. putida on, as the circuit was very sensitive to the AHL concentration.
Co-dependency
Design
We have designed our biofilter to harbor two genetically engineered microbes. Although we have designed safety mechanisms for both E. coli and P. putida, such as the quorum sensing,
we needed to add another layer of protection. We made the bacteria exchange vital nutrients to each other, so that if they are separated outside of the biofilter, both bacteria would die. More
specifically, we engineered E. coli. to grow on methane and secrete acetate, while P. putida needed to be able to use acetate as its carbon source. In addition, we designed both
microbes to be auxotrophic for one amino acid each. We did this by consulting the available literature and using the iGEM PIPE. Details on how we used this tool for the auxotrophy co-dependency can
be found at our Auxotrophy design page. 
Figure 6: Schematic overview of the three safety mechanisms in E. coli (right) and P. putida (left). 1: methane-dependent kill-switch, 2: proximity-dependent kill-switch, 3: co-dependency.
Build
With this tool we made the microbes auxotrophic and simulated their uptake and production rates for tryptophan (Trp) and arginine (Arg). We also checked that the auxotrophic models of E. coli and
P. putidawere able to grow on methane and acetate respectively.
After designing the strategy using PIPE, we went into the lab. Making knock-ins and knockouts as suggested by The iGEM PIPE in E. coli and P. putida.
Test
The iGEM PIPE predicted that that Arg needed to be supplied in relatively high amount compared to Trp. The simulation also indicated that P. putida can consume acetate while secreting
the amino acid needed to complement the auxotrophic E. coli. Similarly, E. coli could grow on methane while producing acetate and the amino acid complementing P. putida’s
auxotrophy.
Learn/Design
The main outcome of this application of the iGEM PIPE was that we then knew that Arg was likely not the best amino acid for establishing auxotrophy. This is because one of the two microorganism would
need to produce it in high amount for the other to survive. Nevertheless, we needed to test this experimentally in vivo (See our
Auxotrophy page for more details!). Therefore another cycle began!
Build
To establish the auxotrophy, we wanted to use amino acids that are only needed by the bacteria in low amounts, so the overproducing bacterium would be able to sustain the receiving bacterium.
To investigate which amino acids are suitable for this, we used the E. coli strains from the KEIO collection with knock-outs for specific amino acids. We grew these strains in medium supplemented
with the amino acids to see when wild type growth was restored.
Test
The experimental results confirmed the predictions obtained with the iGEM PIPE. For certain amino acids (e.g. Trp and Tyr) low concentrations were sufficient to sustain growth, in contrast with the
high concentrations needed for Arg-auxotrophs. Due to time constraints, we did not get to experimentally prove that the auxotrophic P. putida was able to grow on acetate. However,
we could test the relative abundance of the two microorganisms in a co-culture of P. putida and E. coli. The glucose assimilation pathway was knocked-out in P. putida
while E. coli was WT. Supplying only glucose as carbon source showed that the co-culture could be maintained by the same 1:1 ratio of the initial culture after 24 hours of growth.
Figure 7: The overproduction of Tyr by E. coli and Trp by the P. putida was shown to sustain minimal growth of auxotrophic E. coli strains for the respective amino acid.
Dashed lines indicate minimal concentration of the amino acid needed for growth.
Learn
In our final design, the amino acid pair of choice was Trp-Tyr, with P. putida auxotrophic for Tyr, and E. coli for Trp. Additionally, we learned that having a carbon source
co-dependency plays an important role in the maintenance of the ratio between the bacterial strains during co-culture.
Curious for more details on the final design of all our pillars? Visit the overview page!
Implementation
We developed Cattlelyst in the lab, but to implement the biofilter, we need to know more: how efficient will Cattlelyst be? What are the physical dimensions of our system? And what are the costs associated with running such an installation?
To answer these questions and understand the impact Cattlelyst will make, we modelled our biofilter from the microscale, standing for one cell, to the macroscale… the entire biofilter. By doing so, we could estimate the size and costs as well as show bottlenecks to improve our biofilter. We found out that decreasing the gas flow rate in the biofilter can help make our system even more efficient in the future. Please refer to the implementation page for more information on the physical design of our biofilter.
-
References
arrow_downward- D. J. Leak and H. Dalton, “Growth yields of methanotrophs - 1. Effect of copper on the energetics of methane oxidation,” Appl. Microbiol. Biotechnol., vol. 23, no. 6, pp. 470–476, 1986, doi: 10.1007/BF02346062.
- T. T, G. MT, and H. CA, “Development of a CRISPR/Cas9 System for Methylococcus capsulatus In Vivo Gene Editing,” Appl. Environ. Microbiol., vol. 85, no. 11, Jun. 2019, doi: 10.1128/AEM.00340-19.
- F. Y. H. Chen, H. W. Jung, C. Y. Tsuei, and J. C. Liao, “Converting Escherichia coli to a Synthetic Methylotroph Growing Solely on Methanol,” Cell, vol. 182, no. 4, pp. 933-946.e14, Aug. 2020, doi: 10.1016/J.CELL.2020.07.010.
- B. Zhao, D. Y. Cheng, P. Tan, Q. An, and J. S. Guo, “Characterization of an aerobic denitrifier Pseudomonas stutzeri strain XL-2 to achieve efficient nitrate removal,” Bioresour. Technol., vol. 250, no. August 2017, pp. 564–573, 2018, doi: 10.1016/j.biortech.2017.11.038.
- D. U. Lee, I. S. Lee, Y. D. Choi, and J. H. Bae, “Effects of external carbon source and empty bed contact time on simultaneous heterotrophic and sulfur-utilizing autotrophic denitrification,” Process Biochem., vol. 36, no. 12, pp. 1215–1224, 2001, doi: 10.1016/S0032-9592(01)00163-7.
- Q. Y. Zhang, P. Yang, L. S. Liu, and Z. J. Liu, “Formulation and characterization of a heterotrophic nitrification-aerobic denitrification synthetic microbial community and its application to livestock wastewater treatment,” Water (Switzerland), vol. 12, no. 1, 2020, doi: 10.3390/w12010218.
- M. Martin-Pascual et al., “A navigation guide of synthetic biology tools for Pseudomonas putida,” 2021, doi: 10.1016/j.biotechadv.2021.107732.
- F. Gubellini et al., “Physiological Response to Membrane Protein Overexpression in E. coli,” Mol. Cell. Proteomics, vol. 10, no. 10, Oct. 2011, doi: 10.1074/MCP.M111.007930.
- R. K. Bennett et al., “Expression of soluble methane monooxygenase in Escherichia coli enables methane conversion,” bioRxiv, p. 2021.08.05.455234, Aug. 2021, doi: 10.1101/2021.08.05.455234.
- S. Pande et al., “Fitness and stability of obligate cross-feeding interactions that emerge upon gene loss in bacteria,” ISME J. 2014 85, vol. 8, no. 5, pp. 953–962, Nov. 2013, doi: 10.1038/ismej.2013.211.
- J. M. Wehrfritz, A. Reilly, S. Spiro, and D. J. Richardson, “Purification of hydroxylamine oxidase from Thiosphaera pantotropha. Identification of electron acceptors that couple heterotrophic nitrification to aerobic denitrification,” FEBS Lett., vol. 335, no. 2, pp. 246–250, 1993, doi: 10.1016/0014-5793(93)80739-H.
- H. S. Joo, M. Hirai, and M. Shoda, “Characteristics of ammonium removal by heterotrophic nitrification-aerobic denitrification by Alcaligenes faecalis no. 4,” J. Biosci. Bioeng., vol. 100, no. 2, pp. 184–191, 2005, doi: 10.1263/jbb.100.184.
- J. Zhang, P. Wu, B. Hao, and Z. Yu, “Heterotrophic nitrification and aerobic denitrification by the bacterium Pseudomonas stutzeri YZN-001,” Bioresour. Technol., vol. 102, no. 21, pp. 9866–9869, Nov. 2011, doi: 10.1016/j.biortech.2011.07.118.
- NO is the product of hydroxylamine oxidoreductase Jonathan D. Caranto, Kyle M. Lancaster Proceedings of the National Academy of Sciences Aug 2017, 114 (31) 8217-8222; DOI: 10.1073/pnas.1704504114
- N. Zhang, H. Chen, Y. Lyu, and Y. Wang, “Nitrogen removal by a metal-resistant bacterium, Pseudomonas putida ZN1, capable of heterotrophic nitrification-aerobic denitrification,” 2018, doi: 10.1002/jctb.5863.
- O. Yishai, L. Goldbach, H. Tenenboim, S. N. Lindner, and A. Bar-Even, “Engineered Assimilation of Exogenous and Endogenous Formate in Escherichia coli,” ACS Synth. Biol., vol. 6, no. 9, pp. 1722–1731, Sep. 2017, doi: 10.1021/ACSSYNBIO.7B00086.
- C. A. Henard, H. Smith, N. Dowe, M. G. Kalyuzhnaya, P. T. Pienkos, and M. T. Guarnieri, “Bioconversion of methane to lactate by an obligate methanotrophic bacterium,” Sci. Reports 2016 61, vol. 6, no. 1, pp. 1–9, Feb. 2016, doi: 10.1038/srep21585.
- H. J. Kim et al., “Biological conversion of methane to methanol through genetic reassembly of native catalytic domains,” Nat. Catal. 2019 24, vol. 2, no. 4, pp. 342–353, Apr. 2019, doi: 10.1038/s41929-019-0255-1.
- D. Osman et al., “The Effectors and Sensory Sites of Formaldehyde-responsive Regulator FrmR and Metal-sensing Variant * □,” vol. 291, no. 37, pp. 19502–19516, 2016, doi: 10.1074/jbc.M116.745174.
- K. Gerdes, L. K. Poulsen, T. Thisted, A. K. Nielsen, J. Martinussen, and P. H. Andreasen, “The hok killer gene family in gram-negative bacteria,” vol. 2, no. 11, pp. 946–956, 1990.
- M. B. Miller et al., “Tuning the dynamic range of bacterial promoters regulated by ligand-inducible transcription factors,” Nat. Commun., vol. 9, no. 1, pp. 1–8, 2018, doi: 10.1038/s41467-017-02473-5.
- M. B. Miller and B. L. Bassler, “Quorum Sensing in Bacteria,” Annu. Rev. Microbiol., vol. 5, no. 1, pp. 165–199, 2001.