The overall goal of the project was to produce and test two populations of modified OMVs: affi-OMVs expressing an anti-HER2 affibody and THP-OMVs expressing a tumour homing peptide (THP) against CX3CR1. Our hypothesis was that both affi-OMVs and THP-OMVs would be selectively and simultaneously internalised in HER2+ and CX3CR1+ breast cancer cells, when compared to non-cancerous cells. The project was divided into three phases with distinct goals, to enable us to track our progress and achieve important milestones. Details of particular experiments can be found in their corresponding protocols or in the lab notebook.
Phase 1 of the project focused on isolating unmodified native OMVs from different hypervesiculating E.coli strains. This involved selecting an appropriate E.coli strain for OMV production, optimising the protocol for OMV isolation, verifying the isolation of OMVs, and characterising OMV size, morphology, and concentration.
Standard laboratory strains of E.coli have minimal production of OMVs in culture, and are unsuitable for purposes requiring large amounts of OMVs [1]. Four E.coli K12 single gene knockout strains (ΔtolA, ΔtolR, Δnlp1, ΔdegP) were previously reported to have improved yields over the parental K12 strain [1][2][3][4]. These strains were obtained from the Keio collection [5]. We compared growth curves and final OD600 of the strains to gain a rough estimate of the fitness cost of the gene knockouts. All cultures were standardised to OD600 = 0.001 and grown for 18h with orbital shaking in a 96-well plate. OD600 was measured every 15 minutes by a microplate reader.
ΔdegP showed delayed growth kinetics and the lowest final OD600 (Fig 1, 2, Table 1). From our literature review, we found that ΔtolR produces more OMVs than ΔtolA. Therefore, the remainder of the project was restricted to E.coli Δnlp1 and E.coli ΔtolR.
Fig 1: Comparative analysis of growth curves of four hypervesiculating E.coli strains.
Fig 2: Individual growth curves of (A) ΔtolA (B) ΔtolR (C) ΔdegP and (D) Δnlp1. Error bars indicate mean ± standard error of mean of three biological replicates, measured in duplicates.
Table 1: Final OD600 of four hypervesiculating strains. Values are reported as mean ± standard error of mean of three biological replicates, each measured in duplicates.
| Strain | Mean final OD600 ± standard error of mean |
|---|---|
| ΔtolA | 0.71 ± 0.015 |
| ΔtolR | 0.71± 0.022 |
| Δnlp1 | 0.70 ± 0.015 |
| ΔdegP | 0.67 ± 0.012 |
The next step was to isolate OMVs from E.coli ΔtolR and E.coli Δnlp1. OMVs were isolated by serial centrifugation of cell-free supernatants of E.coli cultures, first at 48,000g and then 100,000g. The complete protocol is available on the experiments page.
We went through many iterations of optimisation where we tried different rotors, increased the duration of centrifugation, added alternate rounds of centrifugation to concentrate the sample, and switched from syringe filtration to vacuum filtration to process higher culture volumes.
OMVs were isolated from E.coli ΔtolR and E.coli Δnlp1, under similar conditions and with the same protocol. We used these samples to confirm the presence of OMVs in the isolated fraction, and to compare vesiculation between ΔtolR and Δnlp1.
OMVs can be detected indirectly by probing for the presence of outer membrane proteins that are present in most gram negative bacteria. The most common protein used for this purpose is Outer Membrane Protein A (OmpA). OmpA is a 37kDa constitutively expressed protein.
To reliably use OmpA as a marker for OMVs, we verified that OmpA was present in all strains we utilised. We also confirmed the presence of OmpA in the OMV fraction isolated from E.coli ΔtolR (Fig 3).
Fig 3: Anti-OmpA Western blot of OMVs isolated from ΔtolR, and bacterial pellets of parental strain E.coli K12 and hypervesiculating strains ΔtolA, ΔtolR, Δnlp1 and ΔdegP. Overnight cultures were standardised to OD600 = 1, and equal volumes of culture were loaded in each lane to control for cell numbers, and probed with anti-OmpA anitbody. Bands were detected at the expected size of 37 kDa. An additional lower molecular weight band was noted in the bacterial pellets, that was absent in the OMV fraction.
Subsequently, we confirmed the presence of OmpA in the OMV fraction isolated from E.coli Δnlp1, and compared it to OMVs previously isolated from E.coli ΔtolR. OmpA was detected in both samples, indicating the presence of OMVs (Fig 4).
Fig 4: Anti-OmpA Western blot of OMV fractions from E.coli ΔtolR and E.coli Δnlp1. Equal volumes were loaded for the OMV samples. Cell pellets of both strains were also analysed, but the presence of excess sample in these lanes hindered the proper resolution of proteins during electrophoresis, causing a hazy and diffused appearance (Fig 5, duplicate samples were analysed on SDS PAGE).
During the isolation of OMVs, cell debris may co-sediment, causing inflated protein concentration values, as well as introducing unwanted bacterial cell components. Isolated OMVs were found to be free of viable cells by plating on LB agar. SDS PAGE analysis revealed that OMVs were highly enriched for proteins ~35 kDa in size, which we hypothesise are members of the outer membrane protein (Omp) family (Fig 5). Very faint bands were noted for other molecular weights, indicating the general absence of cell debris. Overall, the OMV isolation protocol was effective and primarily yielded OMVs, with minimal contamination.
Fig 5: SDS PAGE of OMVs are their corresponding source strains. Excess sample for bacterial pellets caused the appearance of a smear. OMV samples had intense bands ~35 kDa. This SDS PAGE gel is a duplicate of the samples analysed by Western blot (Fig 3).
To confirm the presence of OMVs, we first imaged native ΔtolR OMVs using a 120kV transmission electron microscope (Fig 6). The images clearly indicated the presence of spherical heterogeneous structures, 100-200 nm in diameter, consistent with the reported size range of OMVs [7].
Fig 6: Transmission electron microscope image of OMVs isolated from E.coli ΔtolR. Spherical structures were noted but the lipid layer was not visible. Stain used was 4% phosphotungstic acid stain and the ΔtolR OMVs were resuspended in distilled water.
To better visualise the morphology of these structures, OMV sample was imaged by cryo-TEM.
Fig 7: (A, B) Cryogenic electron microscope images of E.coli ΔtolR OMVs suspended in 1X PBS. The lipid bilayer is distinct, confirming the presence of OMVs.
OMVs can be indirectly quantified using protein concentration as a proxy measure [2] [6]. Other methods of quantification are nanoparticle tracking analysis and electron microscopy. Protein quantification of the OMV samples was performed via a Bradford Assay. A standard curve was constructed using Bovine Serum Albumin (BSA) and the regression line was used to compute protein concentrations of OMV samples (Table 2, Fig 8). When subjected to the same culture conditions and isolation procedure, E.coli Δnlp1 yielded a higher amount of OMVs. This may be due to intrinsic differences in the physiology of the two strains, or due to differences in cell numbers.
Table 2: Protein concentration of OMVs fractions isolated from E.coli ΔtolR and E.coli Δnlp1. Measurements were carried out in duplicates.
| Sample | Mean adjusted absorbance at 595nm (A) | Protein concentration (ug/ml) (C) | Total volume (ml) (T) | Total protein (P = C*T) (ug) |
|---|---|---|---|---|
| ΔtolR OMV | 0.028 | 100.64 | 0.11 | 11.06993 |
| Δnlp1 OMV | 0.011 | 48.18 | 0.31 | 14.93495 |
Fig 8: Comparative analysis of the OMV yield from two E.coli strains. OMV yield is measured as total protein in the sample, in micrograms.
The size distribution of nanoparticles such as OMVs can be measured via dynamic light scattering [9]. We compared the size distributions of OMVs isolated from E.coli ΔtolR and E.coli Δnlp1.
Fig 9: Comparison of the size distribution of OMVs from E.coli ΔtolR and E.coli Δnlp1. The X axis indicates diameter (in nm) and the Y axis indicates the percentage. There is a strong correspondence between the two distributions.
Average diameter of E. coli ΔtolR OMVs = 40 nm
Average diameter of E. coli Δnlp1 OMVs = 39 nm
Summary
We acquired four E.coli hypervesiculating strains: ΔtolA, ΔtolR, Δnlp1, ΔdegP. From our bacterial fitness experiment and literature review, we restricted our experiments to E.coli ΔtolR and E.coli Δnlp1. We verified the presence of OMVs in the isolated fractions of ΔtolR and Δnlp1 by TEM and probing for OmpA. We compared the size distribution and OMV yield from ΔtolR and Δnlp1, and found that Δnlp1 produced more OMVs, but the size distribution did not differ. We chose Δnlp1 E.coli to be the chassis for the project. We confirmed that the isolated OMV were free from viable cells and cell debris.
The goal of phase 2 was to produce affi-OMVs and THP-OMVs. This involved designing and obtaining DNA constructs for our composite parts, cloning them into E.coli expression vector pGEX-4T1, inducing the expression of the proteins and isolating affi-OMVs and THP-OMVs.
We designed two alternate sequences for the affibody: ClyA-Myc-Affi (BBa_K4359006) and ClyA-V5-Affi (BBa_K4359008) . We first received ClyA-Myc-Affi cloned into a pGEX-4T1 backbone from Dr. Vipul Gujrati, TUM Germany. This backbone was chosen because the multiple cloning site (MCS) was downstream of the IPTG-inducible tac promoter, enabling control over protein expression. We verified the identity of the plasmid and the insert via a PCR with backbone-specific primers (Fig 10A) and Sanger sequencing (Fig 10B). Following confirmation, we transformed the plasmid into E.coli Δnlp1.
Fig 10: (A) PCR amplification of ClyA-Myc-Affi produced an amplicon of the expected size (1.5kb). (B) Chromatogram of Sanger sequencing of the sample pGEX-4T1 ClyA-Myc-Affi. Three sequencing reactions were carried, two on the top strand and one on the bottom strand. On average, 300-400 bp of reads were obtained per reaction.
Next, we tested the expression of the protein. An overnight culture was diluted 1/100 into 25 ml of LB ampicillin and grown at 37 degree celsius till mid-log. The culture was induced with IPTG (final concentration 1 mM) and grown for 18 hrs at 20 degree celsius. Cells were pelleted by centrifugation and run on SDS PAGE, followed by Coomassie Brilliant Blue staining (Fig 11). A band at 45kDa was visible, corresponding to the molecular weight of ClyA-Myc-Affi.
Fig 11: SDS PAGE gel image comparing protein profiles of cells before and after IPTG induction. A band ~45 kDa in size is visible following induction, suggesting the presence of ClyA-Myc-Affi.
Affi-OMVs were isolated from E.coli Δnlp1 expressing ClyA-Myc-Affi. OMVs obtained were analysed by SDS PAGE to detect the presence of ClyA-Myc-Affi in the OMV fraction (Fig 12).
Fig 12: (A) SDS PAGE gel image indicating expression of ClyA-Myc-Affi (red arrow) in the bacterial cells following IPTG induction. A faint band is also visible prior to induction, in line with the known leaky behaviour of the lac operon. A faint band of the same molecular weight (black arrow) is visible in Affi-OMVs but is absent in native OMVs. Lower molecular weight bands are also visible in both cells post-induction and Affi-OMV (green arrow) but absent in cells pre-induction. (B) SDS PAGE gel image of Affi-OMV sample from (A). Higher amount was loaded to visualise the bands more clearly. ClyA-Myc-Affi is visible at 45 kDa (black arrow), along with the lower molecular weight band noted earlier (green arrow).
The presence of ClyA-Myc-Affi in affi-OMVs was definitively confirmed via Western blot (Fig 13).
Fig 13: (A) Western blot showing the presence of ClyA-Myc-Affi in E.coli cell pellets before (0 hrs) and after (18 hrs) of IPTG induction. The membrane was probed with rabbit anti-myc tag primary antibody. A band was detected at the expected size of 45 kDa. An additional higher molecular weight band was also noted. (B) The presence of ClyA-Myc-Affi was also noted in Affi-OMVs. Signal detected was proportional to the amount of OMVs loaded.
As determined from our engineering cycle 3, we found that the myc tag is incompatible for use with mammalian cell lines due to cross-reactivity of anti-myc antibodies with the myc-tag and components of mammalian cell lysate. We improved our design to replace this tag with the V5 epitope tag instead. We obtained our composite parts ClyA-V5-Affi (BBa_K4359008) and ClyA-3XFLAG-THP (BBa_K4359009) by DNA synthesis. The parts were flanked by EcoN1 and BamH1 restriction sites which were used to ligate them into the plasmid pGEX-4T1. We successfully digested and ligated these parts into the backbone to produce pGEX-4T1 ClyA-V5-Affi and pGEX-4T1 ClyA-3XFLAG-THP (Fig 14). We transformed these plasmids in E.coli DH5a and confirmed our clones by colony PCR, restriction digestion (Fig 15) and Sanger sequencing (Fig 16).
Fig 14: Agarose gel image of the pGEX-4T1 backbone digested with EcoN1 and BamH1. Linearised backbone with cohesive ends was obtained at the expected size (4.3 kb). (B) PCR amplification of gBlocks dsDNA gene fragments of ClyA-3X-FLAG (1049 bp) and ClyA-V5-Affi (1217 bp) using the Fantom high-fidelity polymerase. The PCR fragments were digested with EcoN1 and BamH1 and ligated (with a 1:7 ratio of insert:backbone) into the linearised backbone.
Fig 15: (A) Agarose gel images of colony PCR to identify the positive clones. We screened 5 colonies each for pGEX4T-1 ClyA-V5-Affi and pGEX4T-1 ClyA-3XFLAG-THP, and identified 3 positive clones for the former and 2 positive clones for the latter. (B) Restriction digestion using the EcoN1 and BamH1 restriction sites released inserts corresponding to their expected sizes: ClyA-3XFLAG-THP (1 kb) and ClyA-V5-Affi (1.2 kb).
Fig 16: Representative chromatogram of Sanger sequencing of the sample pGEX-4T1 ClyA-V5-Affi and pGEX 4T1 ClyA-3XFLAG-THP. For pGEX-4T1 ClyA-V5-Affi, three sequencing reactions were carried: two on the top strand and one on the bottom strand. For pGEX-4T1 ClyA-3X-FLAG, two sequencing reactions were carried: one on the top strand and one on the bottom strand. On average, 300-400 bp of reads were obtained per reaction.
Following successful cloning, we transformed E.coli Δnlp1 to produce two populations: one containing pGEX-4T1 ClyA-V5-Affi and the other containing pGEX-4T1 ClyA-3XFLAG-THP, to produce Affi-OMVs and THP-OMVs respectively. Prior to isolating OMVs, we tested the expression of the proteins, similar to Fig 11. We verified that full length ClyA-V5-Affii and ClyA_3XFLAG_THP were being produced, corresponding to the appropriate band sizes.
Fig 17: SDS PAGE gel image indicating expression of ClyA-V5-Affi (green arrow) and ClyA-3XFLAG-THP (red arrow) in the bacterial cells following IPTG induction. Lower molecular weight bands (black), similar to the size seen in Fig (11) and Fig (12) are also visible in both cases post-induction.
Affi-OMVs containing ClyA-V5-Affi and THP-OMVs containing ClyA-3XFLAG-THP were isolated. We investigated the localisation of these proteins to the OMV fraction by Western blot.
Fig 18: Anti-FLAG Western blot of bacterial cells (before and after induction) and THP-OMV fraction. Equal volumes were loaded for the induced and uninduced cell samples. A band corresponding to ClyA-3XFLAG-THP at 37kDa is visible in all lanes, suggesting the production and localization of ClyA-3XFLAG-THP to the OMV fraction. A lower weight band is visible in the induced cells. From earlier SDS PAGE analysis (Fig 9), we noted low levels of expression occurring prior to induction as well.
Fig 19: Anti-V5 Western blot of bacterial cells (before and after induction) and Affi-OMV fraction. Equal volumes were loaded for the induced and uninduced cell samples. A band at the appropriate corresponding to ClyA-V5-Affi is visible in all lanes, suggesting the production and localization of ClyA-V5-Affi to the OMV fraction. A lower weight band is visible in the induced cells. From earlier SDS PAGE analysis (Fig 9), we noted low levels of expression occurring prior to induction as well.
Previous Western blots indicate that the recombinant proteins are present in the OMV fraction but do not provide information about the orientation of the protein. For Affi-OMVs and THP-OMVs to be internalised, it is important for the affibody and THP respectively to be present on the external surface of the OMV, and not in the lumen or embedded in the membrane.
A proteinase K test was carried out for Affi-OMVs to investigate this. Proteinase K is a non-specific protease. If the affibody and the tag are exposed on the outer surface, we expect them to be degraded or truncated following proteinase K treatment. If the affibody and tag are in the interior of the OMV, they should be protected from Proteinase K by the membrane, and should be intact. If OMVs are lysed with SDS prior to proteinase K treatment, then regardless of the orientation, we expect protein degradation. The following results were obtained for this test:
Fig 20: Results of the Proteinase K test performed on Affi-OMVs. 7.5ug of OMVs were incubated with or without SDS and with or without Proteinase K, as indicated. The entire volume of the reaction was analysed by Western blot against the V5 tag.
The MTT assay is a colorimetric assay for assessing cell metabolic activity, and can be used to measure cell viability. In this case, we aimed to measure cell viability following treatments of different concentrations of THP-OMVs expressing ClyA-3XFLAG-THP, as compared to a control treatment of 1X PBS. SK-BR-3 cells were utilised for this assay. The mean percentage viability is reported as the viability of THP-OMV treated SK-BR-3 cells normalised to the mean viability following PBS treatment (Fig 21).
Fig 21: Cytotoxicity of THP-OMVs as measured using the MTT assay with SK-BR-3 cell cultures. The mean percentage viability is reported as the viability of THP-treated SK-BR-3 cells normalised to the mean viability following PBS treatment. Viability is reported as mean ± standard error of mean of three biological replicates. Concentrations of THP-OMVs indicated are working concentrations.
Summary
In Phase 2, we successfully cloned ClyA-V5-Affi and ClyA-3XFLAG-THP into the backbone pGEX-4T1. We tested the expression of all three proteins and verified that they translocated to the OMV fraction. We investigated the safety of THP-OMVs by the MTT assay and did not find significant cytotoxicity. We performed a Proteinase K test to verify the display of ClyA-V5-Affi on the outer surface of OMVs but did not obtain any conclusive results.
In phase 3 of the project, we aimed to test the internalisation of Affi-OMVs and THP-OMVs in the HER2+ breast cancer cell line, SK-BR-3, compared to a control cell line HEK 293T. The results of these experiments can be found on the proof of concept page.