Contents
Phase 1 - OMV isolation and characterisation
Bradford assay
Using protein concentration as a substitute measure, OMVs can be indirectly quantified. A Bradford Assay was used to measure the amount of protein in the OMV samples. In order to determine the protein contents in OMV samples, a standard curve was made using bovine serum albumin (BSA). The regression line was used to calculate concentrations of OMVs in samples. E. coli Δnlp1 produced the most OMVs under the same growth conditions and isolation method. This might be the result of inherent variations in the physiological makeup of the five strains or variations in cell densities.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/omv-isolation-characterization/white-modern-mothers-day-presentation-43.jpg)
Table 1: This table shows the different protein yields present in OMVs of 5 hypervesiculating strains.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/omv-isolation-characterization/white-modern-mothers-day-presentation-43-1.jpg)
Fig1: This figure shows the comparison of OMVs yield from different hypervesicualting strains of E.coli. The highest is from ∆NlpI. The lowest is from ∆DegP.
DLS
A well-known, non-invasive method for determining the size and size distribution of molecules
and other particles often in the submicron range, and with the most recent technology, smaller
than 1nm, is dynamic light scattering (DLS), also known as quasi elastic light scattering (QELS).
Dynamic light scattering is frequently used to characterise particles, emulsions, or molecules
that have dispersed or dissolved in liquids. Different intensities of laser light are scattered due to
the Brownian motion of atoms or molecules in suspension. By analysing these intensity
variations, the Stokes-Einstein connection provides the Brownian motion's velocity and,
consequently, the particle size.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/1.png)
Fig2: This is the DLS data for BL21(de3) OMVs. The size of majority of OMVs that comes out of this data is around 120nm.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/white-modern-mothers-day-presentation-43-2.jpg)
Fig3: This is the DLS plot for ∆𝑇𝑜𝑙𝐴 OMVs. The size of majority of OMVs from this strain comes around 120nm.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/3.png)
Fig4: This is the DLS plot for ∆TolR OMVs. The size of majority of OMVs for this strain comes around 140nm.
SEM
The scanning electron microscope (SEM) produces a variety of signals at the surface of solid
specimens using a focussed beam of high-energy electrons. The signals produced by
electron-sample interactions include details about the sample's exterior appearance (texture),
chemical make-up, and crystalline structure and orientation of the constituent materials. In the
majority of applications, measurements are gathered over a chosen region of the sample's
surface, and a 2-dimensional image is created to show spatial variations in these qualities.
Here, we have used SEM to visualize the OMVs from different strains.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/omv-isolation-characterization/1.jpg)
Fig5: This image shows the OMVs from ∆TolR strain.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/omv-isolation-characterization/2.jpg)
Fig6: This image shows the OMVs from ∆TolA strain.
TEM
We initially photographed native tolR, tolA, and nlpI OMVs with a 120kV transmission electron microscope to confirm the existence of OMVs. The scans showed the presence of spherical heterogeneous structures with a diameter of 100-200 nm, which is compatible with the stated size range of OMVs.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/omv-isolation-characterization/4.png)
Fig10: BL21DE3 ISOLATED OMVs
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/omv-isolation-characterization/5.png)
Fig11: ∆TolR ISOLATED OMVs
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/omv-isolation-characterization/6.png)
Fig12: ∆𝑇𝑜𝑙𝐴 ISOLATED OMVs
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/omv-isolation-characterization/7.png)
Fig13: ∆NlpI ISOLATED OMVs
Comparison of growth curves
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/table-growth.png)
Table2: This table shows the OD600 values at different time intervals for five hypervesiculating strains
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/tola.jpg)
Fig14: This figure shows the change in OD with time for ∆TolA strain.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/tolr.jpg)
Fig15: This figure shows the change in OD with time for ∆TolR strain.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/degp.jpg)
Fig16: This figure shows the change in OD with time for ∆DegP strain.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/nlpl.jpg)
Fig17: This figure shows the change in OD with time for ∆NlpI strain.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/tolra.jpg)
Fig18: This figure shows the change in OD with time for ∆TolRA strain.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/dls-growthcurve/growth-curve.jpg)
Fig19: This figure shows the change in OD with time for all hypervesiculating strains. The fastest growing strain is ∆NlpI. The slowest growing strain is ∆TolA.
Phase 2 - Cas protein loading in OMV
Overview
One crucial aspect of our project is loading FnCas12a into OMVs to develop new CRISPR
delivery methods and biocontrol agents. We've replaced the commonly used Cas9 protein,
which is large and causes off-target effects, with FnCas12a. FnCas12a is smaller, has fewer
off-target effects, and can mature crRNA from the crRNA array, enabling multiplex gene editing.
Our plan involves creating two plasmids: SpTorA-FnCas12a-6xHis and a Tat
machinery-expressing plasmid. The SpTorA-FnCas12a-6xHis plasmid tags FnCas12a with a
SpTorA signal peptide, localizing it in the periplasmic space for loading into vesicles derived
from the outer membrane with periplasmic proteins. The Tat plasmid exogenously expresses the
Tat ABCDE complex in the inner membrane, improving FnCas12a loading into the OMVs.
We also planned to perform two western blots with anti-His antibody. First on with bacterial lysate sample with induction and bacterial lysate without induction to confirm the expression of our FnCas12a circuit.
Second, We will perform a Western blot using three types of OMVs: Type 1, isolated from BL21DE3 bacteria co-transformed with SpTorA-FnCas12a-6xHis and Tat plasmid; Type 2, isolated from BL21DE3 bacteria transformed with only SpTorA-FnCas12a-6xHis; and Type 3, OMVs from BL21DE3 bacteria without any plasmid (control sample).
Cloning of SpTorA-FnCas12a-6xHis
The p-spTorA-GFP-H6C plasmid (Addgene 168517) was linearized (4542 bp) through PCR
using pBAD-spTorA-GFP-6xHis-FP and pBAD-spTorA-GFP-6xHis-RP.
To create FnCas12a tagged with the SpTorA signal peptide, we synthesized the sequence in
two fragments from IDT: F1-SpTorA-FnCas12a (1501 bp) and F2-SpTorA-FnCas12a (2615 bp)
within pUCIDT-AMP GoldenGate plasmids. F1-SpTorA-FnCas12a was obtained via PCR using
F1-spTorA-FnCas12a-FP and F1-spTorA-FnCas12a-RP. F2-SpTorA-FnCas12a was recovered
using F2-spTorA-FnCas12a-FP and F2-spTorA-FnCas12a-RP.
The three fragments were assembled with NEB HiFi assembly master mix to generate the
AraBad-SpTorA-FnCas12a-6xHis plasmid, and the Gibson product was transformed into
DH5alpha using the heat shock method.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/cas-loading-in-sptora/2.png)
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/cas-loading-in-sptora/4.png)
Fig 1: After recoving of each fragments was run in gel elctrophoresis and figure shows they are in proper location. Ladder used in first well is NEB 1KB+.
For cloning confirmation, bacteria from the transformation were spread on ampicillin plates, resulting in colony growth. Colony PCR was performed on 12 colonies using F1-spTorA-FnCas12a-FP, F1-spTorA-FnCas12a-RP, F2-spTorA-FnCas12a-FP, and F2-spTorA-FnCas12a-RP separately. The presence of appropriate bands in gel electrophoresis for both PCR products confirmed the successful insertion into the plasmid backbone, with the orientation maintained by Gibson assembly.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/cas-loading-in-sptora/3.png)
Fig2: Colony PCR result of first 6 colonies.
In Fig. 2, Colony PCR bands corresponding to F1-SpTorA primers (2615 bp) and F2-SpTorA primers (1501 bp) were observed in the correct position for every colony, while there were no bands for the control colony uper 2nd well was loaded with PCR product of control colony with F1-SpTorA primers and lower 2nd well was loaded with PCR product of control colony with F2-SpTorA primers (normal DH5alpha without transformation with the Gibson assembly product). Ladder in first well is NEB 1KB+.
SDS-PAGE
SDS-PAGE data showing no band around 130 Kd for uninduced bacteria but every induced bacterial lysate have a band at this position which indicates that FnCas12a protein may be express after induction. But the induce band is faint. It could be as FnCas12a is very big protein so we have optimise the induction and optimisation of SDS-PAGE itself needed.
![](https://static.igem.wiki/teams/4781/wiki/website-images/wetlab/cas-loading-in-sptora/sds.png)
Figure 3
TAT - ABCE Plasmid construction
Overview
The Tat circuit was planned so that we could hack the twin-arginine pathway of the bacteria that
exports folded proteins to the periplasm. In Escherichia coli and other Gram-negative bacteria,
TatA, TatB, and TatC are all essential for efficient translocation. The twin arginine translocase
(Tat) is a protein transport pathway that exists in archaea, bacteria, and plant chloroplasts. In
bacteria, it exports proteins across the plasma membrane and is important for many processes,
including energy metabolism, formation of the cell envelope, biofilm formation, heavy metal
resistance, nitrogen-fixing symbiosis, bacterial pathogenesis, and others. What makes this
protein transport system unusual compared to other transport systems (such as the general
secretory, or Sec pathway) is its ability to transport fully folded proteins across membranes. This
remarkable feat has no requirement for ATP as an energy source and relies solely on the proton
motive force (PMF).
• Upon receiving primers for linearizing the plasmid pSB1C3 for hi-fi DNA assembly
compatibility, gradient PCR was set at 68°C, 62°C, 64°C, 65, 66, and 67°C.
The bright band came in at 62°C and 64°C. The further PCR was done at 62°C.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-tat-gradient-12-09-2023-68-62-64-65-66-67.jpg)
• The next step was gel extraction, for which PCR was done and yielded a concentration of 19.4ng/ul and a 260/280 ratio of 1.78. • The next step was to do the assembly, which was then transformed into BL21DE3 competent cells and plated.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-cropped-tat-plasmid.jpg)
• The plasmid was isolated again from these cells, which gave a concentration of 123.5ng/ul and a 260/280 ratio of 1.85. This plasmid was then sent for sequencing.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-whatsapp-image-2023-10-12-at-17-42-27-fb4e9535.jpg)
![](https://static.igem.wiki/teams/4781/wiki/result-gfp-loading/cloning-results/tat-cloning.jpg)
Figure : Agarose Gel Electrophoresis to test successful assembly of TatABCE plasmid constructs with respect to pSB1C3 control. The size differences between those two shows that the assembly is successful.
Double Transformation
A double transformation of spTorA-GFP along with the Tat circuit was performed to check if our hypothesis that exogenous overexpression of the Tat ABCE complex will lead to greater transport of our desired protein into the periplasm, then eventually into the outer membrane vesicles, yielded successful results for double transformation.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-img20231009175743.jpg)
Figure: sptorA-Fncas12a-His6xTransformed in C41(DE3), plated on LB-Agar-Amp plates
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-img20231009175728.jpg)
Figure: sptorA-gfp-His6x Transformed in C41(DE3), plated on LB-Agar-Amp
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-cropped-tat-plasmid.jpg)
Figure: Tat-assembled Transformed in C41(DE3), plated on LB-Agar-Chl plates
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-img20231009175814.jpg)
Figure: sptorA-gfp+Tat-assembled Transformed in C41(DE3), plated on LB-Agar-Amp-Chl
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-img20231009175827.jpg)
Figure: sptorA-Fncas12a+Tat-assembled Transformed in C41(DE3), plated on LB-Agar-Amp-Chl plates
GFP LOADING IN OMV and Localization of gfp into nucleoid of pathogen:
Before loading FnCas12a into OMVs we tried loading a small fluro-protein GFP into the OMVs by using GFP tagged with SpTorA signal peptide and exogenously overexpressing TAT machinery protein complexes. Confocal imaging shows that using co-transformation with plasmid TatABCE and P-spTorA-GFP-H6C, GFP localization into the periplasm increases, indicating proper functioning of the Tat pathway overexpression plasmid.
![](https://static.igem.wiki/teams/4781/wiki/result-gfp-loading/cloning-results/judging-presentation-5-mins.jpg)
Figure : Control-uninduced shows no GFP signal, E.coli with only P-spTorA-GFP-H6C shows gfp signal in whole cell, But the E.coli co-transformed with TatABCE and P-spTorA-GFP-H6C shows GFP localization in periphery.
Next, from the co-transformed E. coli BL21(DE3) strain we isolated the OMVs after appropriate induction with 1mM IPTG and 1% (w/v) Arabinose inducer. Then we incubated the isolated OMVs with our pathogenic strain Xoo and tested using Confocal Imaging. Results show that OMV fuses with the Xoo bacteria and GFP not only loads into the OMV but after incubation, loaded gfp of the OMV localize into the nucleoid of the Xoo bacteria. This result is actually very crucial for our project which indicate that in same way if we load the Cas12a-gRNA complex then that might also localize to pathogen’s nucleoid and do the double strand break.
![](https://static.igem.wiki/teams/4781/wiki/result-gfp-loading/cloning-results/gfp-loading.jpg)
Figure: Confocal Images for GFP internalized into Xoo after GFP-loaded OMVs fuse with Xoo. (a) Xoo Plasma Membrane stained with FM4-64 (b) Xoo Nucleoid stained with DAPI (c ) Internalization of GFP loaded OMVs with Xoo (d) MERGED image of all three
Phase 3
Detection Circuit
Overview
The detection circuit was incorporated into the project after interaction with the farmers and understanding their lack of knowledge about the disease. The detection kit works as a reporter after the DSF sensor. It uses the two-component regulator system of RpfC- RpfG to sense the domain. The RpfG converts cyclic-di-GMP into cyclic -GMP once it senses DSF. The Vc2 riboswitch linearizes with low levels of cyclic diGMP. This leads to transcriptional activation of gfaspurple which produces the purple color.
• Upon receiving the DNA fragments for the circuit, they were run on 1% gel electrophoresis to check if the synthesis led to the correct fragment size.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-det-1-2-3.jpg)
The fragments were synthesized correctly.
• We then performed gradient PCR to check the optimum temperature of melting for the
primers we ordered to linearize the plasmid and make it compatible with Hi-Fi DNA
Assembly. To make it compatible, we ordered primers such that it had 30 bp overlapping
regions with the first fragment and the last fragment.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-det-tat-23-8-23-58-60-62-64-66.jpg)
The most bright band was at 62 °C. Hence, we decided to perform the PCR for linearizing the
plasmid at 62°C.
• We then performed PCR and did the gel extraction, which yielded a concentration of
6.798ng/ul and a 260/280 ratio of 1.8.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-det-gel-extraction-conc-1.jpg)
• We then moved on to NEB Hi-Fi DNA Assembly. After assembly, we transformed it directly into chemically competent DH5alpha cells and plated it in LB-Chloramphenicol plates. After 12-14 hours of incubation at 37°C, we noticed colonies:
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-whatsapp-image-2023-10-12-at-15-42-34-dce8aac5.jpg)
• After this, we picked a colony from the plate and gave primary culture for plasmid isolation. The plasmid isolation resulted in a concentration of 168.5ng/ul. It was loaded in well B1.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-det-plasmid-isolation-conc.jpg)
• The plasmid isolated again went through PCR with the forward primer of the first fragment and the reverse primer of the last fragment to check if our circuit was cloned successfully. Then it was run in 0.8% agarose gel. The band was supposed to come around 4578bp.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-det-26-9-23-pcr.jpg)
This proved that our cloning was successful. Further checking was to be done by sequencing the sample. • The plasmid was further transformed into BL21DE3 cells for protein expression and testing of the circuit.
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cropped-det-bl21de3.jpg)
DSF ISOLATION(HRMS Analysis)
After isolating Diffusible Signal Factor (DSF), we used High-Resolution Mass Spectrometry to confirm the isolation of the compound. The total yield obtained after rotavapor and vacuum was 40mg. The expected mass of isolated DSF was approximately 211kDa, as determined through various protocols and theoretical calculations. However, our measured mass was around 209kDa ±1.
![](https://static.igem.wiki/teams/4781/wiki/reults/1.png)
Fig - Isolated DSF peak obtained from HRMS.
![](https://static.igem.wiki/teams/4781/wiki/reults/2.png)
Fig - Impurities and noise found in HRMS peak.
Although we successfully isolated DSF, we encountered a significant amount of impurities. This could be attributed to the LB medium and analytical grade reagents used in the process. To address this issue, we consulted with Postdoctoral Fellow Abhishek Singh from the Department of Chemical Sciences. Abhishek recommended employing LC-MS for better verification and noise reduction. He also highlighted potential limitations associated with column chromatography.
DETECTION TESTING
The detection circuit needed 2 inducers to function- DSF and IPTG. Of these, IPTG could be
easily prepared by mixing appropriate amounts of IPTG powder in distilled water. On the other
hand, we planned to isolate DSF from Xoo Culture itself.
The Xoo cells were first pelleted down and the supernatant was lyophilized. The powdered
supernatant contained DSF. The supernatant was then resuspended in methanol to generate
different concentrations of the lyophilized stock.
Then E.coli BL21DE3 containing Detection circuit was cultured and at OD600 value between
0.6-0.8, DSF solution was added along with IPTG.
We expected to get purple color due to the chromoprotein gFas-Purple, but unfortunately we did
not see any color change.
Possible reason could be that since lyophilized stock had a lot of impurities (as shown by HRMS
data), DSF concentration may not have been adequate enough to induce the chromoprotein
expression. Furthermore, methanol and its oxidized product are cytotoxic to cells, so we are not
sure if it had any effect on the system [2].
FUSION
It is widely established that Gram-negative bacteria utilize outer membrane vesicles (OMVs) for both intercellular and intracellular communication [3]. Given that our target pathogen, Xanthomonas, is also a Gram-negative bacterium, we proposed the fusion of E. coli vesicles with Xanthomonas. However, there is a lack of direct evidence for this interaction in the existing literature, prompting us to pursue experimental validation. The first step in fusion verification was performed using FACS and following results were obtained:
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cloning-results/control.png)
The control Xanthomonas unstained showed relatively less Mean Fluorescent Intensity
![](https://static.igem.wiki/teams/4781/wiki/cloning-results/cloning-results/omv-fusion.png)
A 56-fold median rise was seen when OMVs were incubated in a 2:1 volume ratio with
Xanthomonas. Conclusion the Mean Fluorescent increase in the fusion sample suggests that
there is some kind of interaction taking place between OMV’s and Xanthomonas bacteria.
Our initial step involved isolating and purifying OMVs from E. coli. Subsequently, we labelled
these OMVs with a suitable dye and incubated them with Xanthomonas cells. If fusion occurs
between the vesicles and the Xanthomonas cell membrane, the dye will diffuse throughout the
membrane. The resulting membrane signal can then be detected using Confocal Microscopy.
An extremely crucial part of our project is the internalization of OMV constituents into the target
pathogen. To test the feasibility of this process, we expressed GFP in E.coli and loaded it into
OMVs via SpTorA signal peptide. We expected that if this loading was successful and E.coli
OMVs did fuse with Xanthomonas, then GFP would be internalized and we would get GFP
signals from Xanthomonas cells. And we did obtain a positive result.
With this we have proven our hypothesis that bacterial OMVs can indeed act as a delivery
system for functional proteins into other microbial systems. This is true for any protein that
comes within the size limit of Tat secretary pathway I.e. around 150kDa [4]
Proof of Concept
1. Fusion of Outer Membrane Vesicles with Xoo
Our project was completely based on the assumption that the outer membrane vesicles will
naturally fuse with the Xanthomonas. This assumption came from several literature surveys
where they mentioned about gram negative bacteria communicating with each other through
outer membrane vesicles , the communication can be intra-species or inter-species. [1]
To prove our hypothesis, we designed an experiment where we exogenously expressed GFP in
E.coli BL21DE3. We then translocated this GFP into periplasm by hijacking the TAT secretary
pathway of Gram-negative bacteria, which transport folded proteins from cytosol to periplasm.
This GFP would then be internalized into OMVs.
These OMVs were then isolated from E.coli culture and purified. Parallely, we set up a
Xanthomonas culture and incubated the OMVs with it. This would allow the OMVs to fuse with
Xanthomonas plasma membrane and the GFP to be internalized. The GFP signal would then
be recorded via Fluorescence Microscopy.
2. Protein loading in OMV using SpTorA signal peptide and Exogenously expression of Tat
GFP loading in OMV (phase-2 result) proves that SpTorA signal peptide and Tat machinery can be used to load protein. In future we will do experiment to test the Cas (bigger protein) is loading in OMV by Tor signal peptide and exogenous expression of Tat machinery by western blot method mention in Phase-2 result.
Future prospects:
Positive results from fusion experiments opened up new possibilities for research in this arena
as we have verified that OMVs can act as delivery systems for functional proteins in bacterial
systems. In place of GFP, there can be other important proteins such as FnCas12a which can
be loaded and delivered via OMVs. Loading of Cas protein may be difficult owing to its large
size, but as smaller Cas proteins get discovered, our results may open door to new
opportunities of CRISPR/Cas delivery across biological systems. This can therefore have
enormous implications in the therapeutic sector.
Delivery of vesicular content into mammalian systems is well known. Now, we have also verified
the same across bacterial systems. Superiority of functional protein delivery DNA delivery lies in
the fact that there is no issue of lack of protein expression (since we are already introducing
functional proteins).
Further, we can also express target specific ligands on the surface of OMVs for targeted
delivery.
If we do any further work related to this project, we'll update it here: Link
References
- 1. Kesty NC, Kuehn MJ. Incorporation of heterologous outer membrane and periplasmic proteins into Escherichia coli outer membrane vesicles. Journal of Biological Chemistry. 2004 Jan 16;279(3):2069-76.
- 2. Patel R, Smith SM, Robinson C. Protein transport by the bacterial Tat pathway. Biochimica et Biophysica Acta (BBA)-Molecular Cell Research. 2014 Aug 1;1843(8):1620-8
- 3. Zhou, Lian, et al. "Extraction, purification and quantification of diffusible signal factor family quorum-sensing signal molecules in Xanthomonas oryzae pv. oryzae." Bio-protocol 7.6 (2017): e2190-e2190
- 4. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7904680/#:~:text=coli%...
- 5. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5308417/
- 6. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6708511/
- 7. Berleman, James, and Manfred Auer. "The role of bacterial outer membrane vesicles for intra‐and interspecies delivery." Environmental microbiology 15.2 (2013): 347-354.