Team:Paris Bettencourt/Engineering
Design & Results
Design
Minicell production
Part of the project requires controlling the production of minicells. There are three main advantages in being able to control when and how minicells are produced:
Tight control over the population increases safety;
Minicell producing E. coli are unhealthy cells, unable to divide properly, often becoming filamentous in which case they will grow until lysis is induced (1). By controlling minicell production, the starting population is kept as close to natural condition as possible to increase its fitness;
Healthy cells grow faster and in a controlled manner compared to cells constantly producing minicells allowing for a higher yield of minicell production once it is induced.
Minicells can be produced through two methods: by deleting the min operon or by over-expressing the ftsZ gene. We wanted to try multiple ways of producing minicells and determine which allows for tighter control over minicell production while having the highest yield possible. Induction was to be tested with the following methods:
Deletion of the min operon;
Over-expression of FtsZ;
Synergy between both.
To do so, min mutant cells, that are bacteria in which the min operon (genes minC, minD and minE) was knocked out and wild-type E. coli (K-12 MG1655) were compared. Two sets of constructs were designed for implementation in various plasmids, one IPTG induced and the other induced by a rise in temperature.
The first constructs were designed to induce minicell production by raising the temperature. Parts known to be temperature-sensitive:
pSC101ts: a pSC101 origin of replication derivative (low-copy number) which is temperature-sensitive. A point mutation (2) changes one amino acid in the regulatory protein repA from alanine to valine, changing its binding properties (3). It allows for normal plasmid replication at 30°C, limited replication at 37°C and complete curing at 42°C;
pR-pL and TcI: a promoter-repressor system. TcI is a derivative protein of cI857, a lambda phage repressor that is able to bind the operator regions upstream of the pR-pL promoters and to repress transcription (4). Transcription can start at temperatures above 38°C.
pTlpA and TlpA36: another promoter-repressor system. TlpA is a transcriptional autorepressor from the virulence plasmid of Salmonella typhimurium. In its low-temperature dimeric state, it blocks transcription from the 52bp pTlpA operator–promoter (5).
Figure 1) Design of plasmids for minicell induction. One plasmid is expressing min (1.1), one plasmid is expressing ftsZ (1.2) and the third construct is testing the synergy between both (1.3)
In the first construct design (Figure 1.1), the min operon is placed on a plasmid with a pSC101ts origin of replication. The plasmid must be transformed into an E. coli mutant depleted of its min operon. When the cells are grown at 30°C, the min operon is expressed. Expression is considerably reduced when temperature is raised up to 37°C due to a lack of replication of the plasmid, therefore inducing minicell production. Here, minicells are induced by the "deletion" of the min operon.
The second construct (Figure 1.2) focuses on producing minicells through the overproduction of FtsZ protein. The ftsZ gene is under the control of the pR-pL promoter. Downstream of the transcriptional unit (TU) is placed a second TU with the tcI gene under the control of the lac promoter. The plasmid is transformed into wild-type E. coli. At temperatures below 38°C, TcI is able to bind the pR-pL promoter and repress FtsZ expression. This repression can be enhanced with the addition of IPTG which prevents the inhibition of the TU controlled by the lac promoter. The entire constructs are placed on a plasmid with a p15A origin of replication (copy number around 10). Again, because of the absence of redundancy, the construct might be quite leaky allowing for FtsZ production before raising the temperature.
The last construct (Figure 1.3) combines the previous two for testing the synergy between min operon repression and FtsZ over-production. It is composed of two plasmids transformed in a min mutant E. coli. The first plasmid is the same as in the first construct: the min operon under the control of the pTlpA promoter and a temperature-sensitive origin of replication (pSC101ts). Like the second construct, the second plasmid is composed of two TU with ftsZ under the control of pR-pL and tcI under the control of the lac promoter. Downstream of the ftsZ gene, the sequence for TlpA36 is added. At 30°C, TcI binds pR-pL, preventing FtsZ and TlpA36 expression. The tightness can be increased by adding IPTG. When the temperature is raised, TcI is unable to bind pR-pL, there is FtsZ expression as well as TlpA36 which is able to bind pTlpA weakly. The repression can be leaky since there is redundancy with the temperature-sensitive origin of replication.
However, raising the temperature for minicell induction is not compatible with the minicell purification design of the project (see section on mother cell - minicell separation). As changing the induction of the minicells was easier than changing the induction of the phages, the plasmids were re-designed. To be able to combine the parts of the project, IPTG-induced constructs were created since they do not interfere with the heat-induced lysis.
The constructs for controlling minicell production by IPTG induction are very similar to the Temperature-Sensitive Constructs. The difference lies in the choice of promoters and repressors.
plac and LacI: the most well known promoter-repressor couple. LacI binds the plac operator preventing expression of the downstream gene until IPTG is added, preventing the binding and allowing for transcription.
ptet and TetR: another very well known promoter-repressor couple. TetR binds the ptet operator preventing the expression of the gene located downstream of it.
Figure 2. Construct with the combination of min repression and FtsZ overexpression for minicell formation.
This construct is intended to be transformed into a min mutant strain. In the absence of IPTG, LacI (constitutively produced), binds plac and prevents TetR expression. As the min operon is expressed, there is no minicell production. When IPTG is added, the production of TetR starts and prevents the expression of the min operon. Minicells are produced. This construct was designed with two different Anderson constitutive promoters to determine which promoter allows for limited leakiness without too much burden on the cell.
This construct should be transformed into a standard wild-type E. coli strain. Without IPTG, LacI prevents FtsZ expression. The addition of IPTG allows for FtsZ production and minicell production.
This two-plasmid system, to be implemented in a min mutant, is a combination of the two previous ones. It tests whether the combination of the deletion of the min operon and the overexpression of FtsZ allows for minicell production. According to Bi and Lutkenhaus (1990) (7), the combination of the two should neutralize each other. We wanted to confirm this assumption. In absence of IPTG, LacI binds plac and there is no FtsZ or TetR expression. The min operon is expressed normally to mimic the behaviour of a wild-type cell. The addition of IPTG allows for both FtsZ and TetR production, the latter inhibiting min operon expression.
Deletion of the min operon
A strain with the operon knock-out was used.
None of the plasmids previously presented for induction of min operon repression was created in the lab. Instead, strains in which the
entire min operon was knocked-out were used for constitutive minicell production. 2 strains were readily available:
TB43 obtained from iGEM Marburg 2015;
ALMin from the A. Lindner collection.
Overproduction of FtsZ
Minicell production is obtained by transforming wild type E. coli MG1655 with a plasmid containing the ftsZ gene obtained
directly from the genome. The details of the construct can be found in the previous section (Figure 2).
This plasmid was obtained thanks to Golden Gate assembly and transformed in both TB43 and MG1655.
Comparison of the minicell production depending on the method
The plasmid containing ftsZ was transformed in MG1655 and TB43 to study the synergy between both methods.
Minicell purification
One important part of the project is the separation of minicells and mothercells. The conventional minicell-purification methods include centrifugation at high speed and/or the use of antibiotics (8) (9) (10). To facilitate this process for labs without an ultracentrifuge, we created a novel method that does not require expensive technical equipment or the use of chemicals. Instead, cells can be lysed naturally with bacteriophages.
Bacteriophages are a type of virus that infect bacteria, but are harmless to humans.
They consist of a capsid or head that contains the DNA , followed by the collar. Below are a sheat and a baseplate with spikes and tail fibres, forming the tail of the bacteriophage.
Bacteriophages can bind to specific receptors on the surface of bacteria, a process known as attachment. Subsequently, penetration occurs, where the bacteriophages inject their DNA into the cytoplasm of the bacterial host cell.
Thereupon, the genetic material of the phage is transcribed and translated by the bacterial enzymes. Moreover, some phages can integrate into the bacterial chromosome. In this lysogenic cycle, their genome - the prophage - will replicate with the host genome, while having a low burden on the cell.
The endogenous, dormant phages can switch to the lytic cycle in response to external factors, such as heat or nutrients. In this active state, the prophages are transcribed and translated. Those viral particles are consecutively assembled in the bacterium to create many new phages.
The release of the new virions can be by budding or extrusion, but occurs in most cases by cell lysis.
The freed new bacteriophages can then infect other bacteria and the cycle starts again.
The dormant prophages are encoded in the genomic DNA of the bacteria. Consequently, only cells with genomic DNA will be lysed. This is the key point for the purification of minicells. They do not contain genomic DNA, therefore also do not contain the prophages. And as the change from lysogenic to lytic cycle implies death of the host bacterium, only minicells will remain as metabolic active cells in the sample.
However, the released phages pose a problem. As minicells have a very similar membrane composition as the E. coli parental cells, the phages can potentially infect the minicells. As minicells also contain the transcriptional machinery, the phages could be replicated and thus also lead to lysis of the minicells. To overcome this obstacle, the phages have to be made infection-defective, while they should keep their ability to lyse the host bacterium.
One method to achieve this is by the modification of the terminase A gene. This gene is necessary to pack the viral DNA into the capsid. Deletion of this gene prevents the phages from successfully infecting bacteria.
An E. coli strain containing a lambda prophage with a deletion of the terminase A gene was kindly donated by Marianne De Paepe (INRAE). It enters the lytic cycle in response to heat, as it is placed under the temperature-sensitive promoter cI857. We transferred the genome of the prophage into our minicell-producting strain by P1 transduction.
A donor E. coli , containing the lambda prophage (yellow) in its genomic DNA (blue) is infected with P1 phages. The P1 phages insert their genome (red) into the bacterium and replicate. During this process they can also incorporate parts of the genomic DNA of the donor bacterium. Some of them will therefore contain the nucleotide sequence of the lambda phage. After lysis of the bacterium, the freed new P1 phages can infect other bacteria. If those phages do infect the recipient E. coli , they insert their genomic sequence into the bacterium. By homologous recombination, the DNA can consecutively also be incorporated into the recipient genome. To be able to select for the bacteria that incorporated the nucleotide sequence of the lambda phage, an antibiotic resistance cassette is located in the lambda prophage sequence. Only recipient bacteria that contain this part will be able to grow on selective media with the antibiotic. It is likely that they will contain the whole lambda prophage sequence as well.
The successful transfer of the whole lambda prophage was validated by PCR. Consequently, the antibiotic resistance cassette was removed by transformation with a plasmid containing an arabinose induced flippase. The flippase will be expressed after the addition of arabinose. It does recognize the FRT (flippase recognition target sites) sequences that flank the antibiotic resistance cassette and flips the two sequences. This recombination cleaves out the part in between the FRT sites, and leads therefore to the loss of the antibiotic resistance cassette.
To sum up, the lambda prophage was transferred from the host strain to the recipient strains TB43 and MinB by P1 transduction. After validation of the correct insertion, the antibiotic resistance cassette is removed from the genome with a flippase. The final lambda prophage construct in the new host bacterium is then analysed and characterized.
Protein production
Design
The purpose of the process is to induce the production of an enzyme of interest by the minicells. To produce the enzyme, we inserted the gene coding for it on a plasmid backbone. To ensure that this will be expressed only by minicells, its expression is induced by arabinose. The gene is placed under the control of the pBad promoter and the araC protein.
In the absence of arabinose, araC binds the pBad operator and prevents transcription of the downstream gene. Once arabinose is added, the conformation of araC changes thus releasing the promoter and allowing for transcription of the downstream gene. To ensure the tightness of the system, the araC gene is also placed on the plasmid, under the control of the pC promoter.
Figure 3) Plasmid map for enzyme production in minicells induced by arabinose. The characterization was done by inserting the gene coding for GFP.
The construct is placed in a pUC19 backbone, a high-copy number plasmid. Because many copies are present in a cell, there is a high chance for a minicell to take up at least one of these plasmids. Research indicates that high-copy number plasmids are not randomly segregated during cell division but tend to form clusters. These tend to be at the pole because of nucleoid exclusion (11). In this case, since minicells are more likely to form from the old pole (12), there is an even higher chance for the minicell to take up multiple copies of the plasmid.
Figure 4) Segregation of high copy plasmid leading to minicell containing multiple copies of plasmid of interest.
Assembly
Different approaches were used to create this plasmid.
Golden Gate assembly
Our first assembly strategy was to use Golden Gate to assemble the plasmid. The different parts (pBAD, RBS, CDS for GFP and B0015 and the araC transcriptional unit) were retrieved from the CIDAR MoClo kit. The backbone is the DVK backbone from the CIDAR MoClo kit, a high-copy number plasmid.
Circular Polymerase Extension Cloning
As we encountered difficulties in assembling the construct with Golden Gate assembly, another method with the main goal of obtaining similar results. CPEC method was used to include the coding sequence for GFP in 2 different backbones already containing the arabinose regulators. The first is a plasmid previously constructed by a Dr. Paul A Davison from Sheffield University (13) with an ampicillin resistance and a ColE1 origin of replication. The second is the pTS000 plasmid given by Dr. Jake Wintermute (14) from CRI.
Both plasmids and GFP genes were amplified for CPEC assembly and assembled following our CPEC protocol that can be found here.
Final design
Our pEnzyme was successfully cloned using golden gate cloning. What differed many between this successful golden gate and all of the failed ones were the parts used. For the final design, the majority of parts were from the MoClo kit which are reliable, and the arabinose operon was amplified from the pBAD-pR plasmid (University of Sheffield). The parts used from the MoClo toolkit were GFP (E0040_CD), the backbone (DVK_EF) and the terminator (B0015_DF).
Results
Minicell production
We determined the minicell production by NEB Turbo cells with microscopy.
Figure 5) Microscopy image of NEB Turbo cells transformed with our FtsZ constitutively expressed plasmid. 100x Oil Ph3 Objective Lens and TRANS wavelength in a Zeiss microscope with camera focus. The small, circular and refringent cells correspond to minicells.
The microscopy image clearly shows the presence of minicells which proves the constructed plasmid is able to induce minicell production. Further characterization is required in order to obtain quantitative data.
Minicell purification
In order to separate the E. coli from the minicells, an infection-defective lambda phage was inserted into the genome of the mother cells. This lambda phage contained a chloramphenicol resistance cassette. After P1 transduction, only cells that contain at least a part of the lambda phage are able to grow on chloramphenicol supplemented LBA plates. The presence of the lambda phage was then additionally verified by colony PCR. 3 colonies of TB43 and MinB that grew on LBA plates containing chloramphenicol were picked and analysed with two different primer pairs. One primer pair was located on the lambda prophage up- and downstream of attP and should result in an amplicon of 968bp. As for the other primer pair, one primer binds on a region of the lambda prophage and the other one on the genome of MG1655. This one should result in an amplicon of 729bp. Only the presence of both bands indicates the integrity of the prophage. PCR was performed with Taq-polymerase and 34 cycles.
Figure 6) Gel electrophoresis after colony PCR with Taq polymerase for the verification of the correct and complete insertion of the lambda phage in the genome by P1 transduction in strains TB43 and MinB.
The following gel-electrophoresis visualizes the respective fragments of the PCR as bands. TB43-2 and MinB-2 lack the 968bp band, indicating an incomplete insertion of the lambda phage. Both bands of 968bp and 729bp are present in the samples TB43-1, TB43-3, MinB-1 and MinB-3. The cells of those colonies can be used for the following experiments.
To characterize the lambda prophage construct in the new host bacterium,
several time-course OD measurements under different conditions were analyzed. In order
to save time and resources, we chose to only continue with the minicell producing strain
TB43, as the minicell yield was higher (according to the flow-cytometry results) and as
the MinB strain contains an unwanted kanamycin resistance.
Firstly, the lysis efficiency in the E. coli wildtype strain MG1655 and the minicell producing
strain TB43 was tested. For this, the OD of both strains with and without the integrated
lambda-phage was monitored. The cell cultures were grown overnight, washed 3 times, diluted
to an OD of 0.05 and plated on 96-well plates. They were incubated for 4h at 30°C before the
temperature was raised to 42°C. OD measurements were taken every 10 min. The mean of 3 biological
replicates with 8 technical replicates each was calculated and plotted with their standard deviations.
Figure 7) Timecourse experiment with monitoring of growth of MG1655 and TB43 with- and without the lambda prophage. Cells were incubated 4 hours at 30°C, then 6 hours at 42°C. The blue vertical line indicates the change of temperature. Mean of 3 biological and 8 technical replicates is displayed, with the standard deviation of the 3 biological replicates as errorbars.
All four strains show similar growth in the first 5 hours. Then, the strains without the lambda phage continue growing slowly, entering the lag-phase. The 2 strains containing the lambda phage on the contrary, show a decreasing OD from 5h on. This corresponds to cell lysis, occurring because the lambda phage is expressed in response to the raised temperature. After incubation of 6h at 42°C, the OD of those strains has almost returned to its initial level. The negative control sample without cells shows no changes in OD, indicating no contamination of the growth medium or plate.
In order to also analyse the infectiousness of our lambda phage, it was compared to a highly infectious P1 phage. MG1655 cells were grown with and without the addition of P1 or lambda phage lysate. An overnight culture of MG1655 was washed 3 times, diluted to an OD of 0.05 and plated on a 96-well plate. Phage lysate of P1 and lambda was added to a series of samples with 8 technical replicates each. The OD was measured every 10 min over 4h while the cells were incubated at 37°C.
Figure 8) Timecourse experiment with monitoring of growth of MG1655 with lambda- and P1-phage lysate. Cells were incubated 4 hours at 37°C. Mean of 8 technical replicates is displayed.
The samples with the added P1 phage lysate show little growth, before the OD is decreasing again. The samples with MG1655 and MG1655 + lambda phage lysate, on the contrary, show an exponential growth during the first hours. The negative control sample without cells shows no changes in OD, indicating no contamination of the growth medium or plate. To make sure that the lysis also does not occur at a later time point, the growth of MG1655+lambda was monitored for several hours more. However, no decrease in OD was observed which indicates that the lambda phages were not able to infect and lyse the bacteria.
Figure 9) Timecourse experiment with monitoring of growth of MG1655 with lambda phage lysate. Cells were incubated 4 hours at 37°C. Mean of 8 technical replicates is displayed.
Another question that was arinsing was the possibility of some resistant cells to remain. Those cells that did not lyse could potentially regrow. For this, the first lysis experiment with 8 technical and 3 biological replicates was further monitored over 10h at 42°C.
Figure 10) Timecourse experiment with monitoring of growth of MG1655 and TB43 with- and without the lambda prophage. Cells were incubated 4 hours at 30°C, then 18 hours at 42°C. The blue vertical line indicates the change of temperature. Mean of 3 biological and 8 technical replicates is displayed, with the standard deviation of the 3 biological replicates as errorbars.
Indeed, in both cell lines with the lambda phages, we could observe growth again. Some of the cells were probably containing a defective lambda phage or a lambda phage that is only transcribed at even higher temperatures. Those few cells are then able to replicate at 42°C, leading to a rising OD. As the OD was almost at basal level, the lysis efficiency was very high.
Protein production
After trying multiple cloning techniques such as digestion-ligation, Circular polymerase extension cloning etc. one technique finally ended up working : the famous golden gate!
Figure 11) Restreak of the NEB cells containing the enzyme-producing plasmid
We first wanted to clone our plasmid with a GFP construct to facilitate characterization studies. In the plate above, there was a concentration of arabinose at 13 mM. We performed a colony PCR for the colonies that were green after the Golden gate transformation plate for verification. The expected band size was 2.4 kB.
Figure 12) Gel image of 6 different colonies (from left to right) of the NEB turbo cells that were transformed with the enzyme-producing plasmid.
Here we can see that colonies 1, 2, 3, 4, and 6 have gel bands of the expected size (varying between 2 kB and 3 kB).
Therefore, these colonies were grown and miniprepped.
Once the high-copy number plasmid was assembled with GFP, it needed to be characterized. This was done by performing a timecourse
experiment in a plate reader with both OD600 for growth and fluorescence measurements. The plasmid was transformed into NEB Turbo
cells for optimal growth. The cells were plated with different arabinose concentrations (20mM - 10mM - 7mM - 5mM - 2mM - 0mM).
M9 medium supplemented with fructose is used as the arabinose system is dependent on the absence of glucose. Cells were then incubated at 37°C for 12 hours with measurements every 10 minutes. The settings of the plate reader were determined previously
by performing a fluorescein calibration curve (see Plate reader calibration section).
A)
B)
Figure 13) Timecourse experiment of NEB Turbo cells transformed with our GFP producing plasmid arabinose inducible and in presence of different arabinose concentrations, as a mean of 3 replicates. A) OD600 over 12 hours B) Fluorescence normalized to OD600 over 12 hours.
We can see all cells grow following a similar curve which fits our expectations. The difference can be due to the difference in initial cell concentration. The exponential phase ends at around 7 hours, as expected. We can also notice that the induction with arabinose at 20mM seems to affect growth. We hypothesize that such high concentrations should not be used. As expected, there is no increase in fluorescence for the control without arabinose. This also shows the tightness of the system. We can observe that the higher the concentration of arabinose, the higher the yield of GFP. It would seem the maximum is reached for all samples around 7 hours which also corresponds to the end of the exponential phase. We choose 7 hours as a time to further compare fluorescence between arabinose concentrations.
Figure 14) Fluorescence in NEB Turbo cells after 7 hours of incubation at 37°C in M9 fructose medium in the presence of various concentrations of arabinose.
We can clearly see the influence of the arabinose concentration on the GFP production and therefore, the fluorescence. The negative control (in absence of arabinose) shows little fluorescence, corresponding to
the leakiness of the construct and the autofluorescence of the cells. The highest induction is achieved with an arabinose
concentration between 10 and 20mM. However, we previously observed the concentration of 20mM to have an impact on cell growth.
Further fluorescence characterization studies could be required to find the optimal concentration between the ranges of 10 mM and 20 mM.
CPEC
Golden Gate assembly provided us with the wanted results. However, CPEC provided us with an alternative.
Even after 24 hours, no colonies showed GFP fluorescence. To verify if the uptaken plasmid contained the gene coding for GFP, we performed a colony PCR.
Figure 15) Gel image of 16 different colonies of the NEB Top 10 cells that were transformed with the enzyme-producing plasmid obtained with CPEC.
All of the colonies present a strong band corresponding to the GFP gene. It would seem the plasmid correctly incorporated the coding sequence but the GFP is not transcribed or translated correctly. This could be due to a point mutation during the amplification of the plasmid. To verify this hypothesis, the plasmids were sent for sequencing. The data was not retrieved before the wiki freeze.
REFERENCES
(1) Howard I. Adler and Alice A. Hardigree. “Growth and Division of Filamentous Forms of Escherichia Coli”. In:Journal of Bacteriology90.1 (1963), pp. 223–226.
(2) Karen A. Armstrong et al. “A 37×103 molecular weight plasmid-encoded protein is required forreplication and copy number control in the plasmid pSC101 and its temperature-sensitive derivativepHS1”. In:Journal of Molecular Biology175.3 (1984), pp. 331–348.issn: 0022-2836. doi:https://doi.org/10.1016/0022-2836(84)90352-8
(3) Lundin and iGEM12 Uppsala.Part:BBaK864051.http://parts.igem.org/Part:BBa_K864051.Accessed: 2021-09-02. 2012.
(4) Norma A Valdez-Cruz et al. “Production of recombinant proteins in E. coli by the heat inducibleexpression system based on the phage lambda pL and/or pR promoters”. In:Microb Cell Fact9.18(2010). doi:https://doi.org/10.1186/1475-2859-9-18
(5) Reini Hurme et al. “DNA Binding Exerted by a Bacterial Gene Regulator with an Extensive Coiled-coil Domain”. In:Protein chemistry and structure271.21 (1996), pp. 16626–12631. doi:https://doi.org/10.1074/jbc.271.21.12626.
(6) Piraner Dan, Abedi Mohamed, and Moser Brittany et al. “Tunable thermal bioswitches for in vivocontrol of microbial therapeutics.” In:Nat Chem Biol13 (2017), pp. 75–80. doi:https://doi.org/10.1074/jbc.271.21.12626.
(7) Bi, E., & Lutkenhaus, J. (1990). Interaction between the min locus and ftsZ. Journal of Bacteriology, 172(10), 5610–5616. https://doi.org/10.1128/jb.172.10.5610-5616.1990
(8) Gaastra, W., & Klemm, P. (1984). The Isolation of minicells. In Nucleic Acids (pp. 301–304). Humana Press. https://doi.org/10.1385/0-89603-064-4:301
(9) Jivrajani, M., Shrivastava, N., & Nivsarkar, M. (2013). A combination approach for rapid and high yielding purification of bacterial minicells. In Journal of Microbiological Methods (Vol. 92, Issue 3, pp. 340–343). Elsevier BV. https://doi.org/10.1016/j.mimet.2012.12.002
(10) Wei, E., Jackson-Smith, A., & Endy, D. (2020). Minicells as a potential chassis for engineering lineage-agnostic organisms. Cold Spring Harbor Laboratory. https://doi.org/10.1101/2020.07.31.231670
(11) Hsu, T.-M., & Chang, Y.-R. (2019). High-Copy-Number Plasmid Segregation—Single-Molecule Dynamics in Single Cells. Biophysical Journal, 116(5), 772–780. https://doi.org/10.1016/j.bpj.2019.01.019
(12) Rang, C. U., Proenca, A., Buetz, C., Shi, C., & Chao, L. (2018). Minicells as a Damage Disposal Mechanism inEscherichia coli. MSphere, 3(5). https://doi.org/10.1128/msphere.00428-18
(13) Song, Y., Cartron, M. L., Jackson, P. J., Davison, P. A., Dickman, M. J., Zhu, D., Huang, W. E., & Hunter, C. N. (2019). Proteorhodopsin Overproduction Enhances the Long-Term Viability of Escherichia coli. Applied and Environmental Microbiology, 86(1). https://doi.org/10.1128/AEM.02087-19
(14) Bongaerts, N., Edoo, Z., Abukar, A. A., Song, X., Sosa Carrillo, S., Lindner, A. B., & Wintermute, E. H. (2021). Low-cost drug discovery with engineered E. coli reveals an anti-mycobacterial activity of benazepril. https://doi.org/10.1101/2021.03.26.437171