Aims

The aim of E.lectrode is to produce pili derived from Geobacter sulfurreducens with a skin-binding tag at the C-terminus of the pilus to enhance its functionality. This involves the following key steps:

1. Creation of Fimbriae-Deficient E. coli Chassis: Build a strain of ΔfimA E. coli NEB5a by performing gene deletion utilising a CRISPR-Cas12a system.
2. Vector Construction: Construct two vectors using Gibson assembly.
3. Expression and Functionality Assessment: Express modified e-pili by co-transformation of vectors into the ΔfimA strain, and test the binding affinity of the pili to collagen.

See Design for more detail.

Creation of Fimbriae-Deficient E. coli Chassis

CRISPR-Cas12a-mediated Deletion of a Gene Encoding for a Fimbrial Protein

The fimA gene (Ecocyc ID b0530) encodes the major type 1 pili structural subunit protein in Escherichia coli (E. coli) NEB5a. As our project involves the expression of a heterologous pili monomer in E. coli NEB5a, we needed to prevent the expression of the native type 1 pili to ensure maximal production of our desired type IV e-pili, and therefore we aimed to chromosomally delete fimA in E. coli NEB5a. To create this strain, we sought to employ a CRISPR-Cas12a system [1] that can be easily adapted to be used for chromosomal gene deletion in E. coli. This system employs two plasmids: the first plasmid, pSIMcpf1 (Addgene ID: 153034), carries the Cas nuclease and the Lambda Red recombination genes. The second plasmid (referred to as pTF) expresses the guide RNA (gRNA) and carries the repair template, composed of two 50-bp homologous arms. Because of its modularity, we only needed to introduce our target gene spacer and donor DNA (dDNA) sequence into the pTF plasmid, and could then carry out the gene deletion process with this modified plasmid along with the pSIMcfp1 plasmid.

However, we eventually discovered through genome analysis that the gene we deleted was instead the sfmA gene of this strain and that a labeling error in the databases we used (NCBI and Biocyc) led us to design our CRISPR-Cas12a system around this gene (see Engineering page for details). While this was not originally the plan, we quickly realised that deleting both sfmA and the true fimA would actually be beneficial, resulting in increased growth in minimal environments [2], which is beneficial because we will be expressing the e-pili with M9. However, we did not have time to complete the double deletion.

We were able to design a new fragment cassette with a gene spacer and dDNA for the actual NEB5a fimA gene, but we had too little time to experimentally delete the actual fimA gene in our NEB5a ΔsfmA strain.

Although it is reported to be cryptic, we do not think deleting sfmA was a wasted effort, as it is a putative fimbrial protein-encoding gene nonetheless and its deletion would further aid the growth rate of our host in minimal media [2]. Thus, this phase of our project was a success, achieving a 50% gene deletion efficacy as shown by our colony PCR (Figure 5).

Eventually, through multiple rounds of plasmid curing, we successfully procured a ΔsfmA strain of E. coli NEB5a (see Gene Deletion Lab Book in our Experiments page for details). After we discovered we would also need to delete the fimA gene to fully remove all native fimbriae formation, we originally aimed to repeat the gene deletion process with our newly designed pTF_fimA plasmid and produce an E. coli NEB5a ΔsfmAΔfimA strain to be used as the chassis of our e-pili production, but due to time constraints, we chose to directly use our ΔsfmA strain in the later stages of our project.

Vector Construction

Workflow

The proposed workflow for creating the pilus expression vectors involves several key steps. The type IV pilus assembly genes will be amplified from the E. coli K12 genome using PCR (Figure 6). These amplified fragments, along with a synthetic gene block containing the Geobacter sulfurreducens pilus (GspilA), will then be divided into two groups and assembled into two separate pBbE1k sub-plasmids using Gibson assembly (Figure 7). The genes in the first sub-plasmid will include GspilA, hofB-hofC, and hofM-hofN-hofO-hofP-hofQ, while the second sub-plasmid will contain ppdA-ppdB-ygdB-ppdC and gspO. Finally, these fragments will be amplified from the sub-plasmids and assembled into the final pBbE1c vector through Gibson assembly, resulting in a complete plasmid capable of expressing the pilus.

Our original vector design (Figure 7), where a single promoter controlled the expression of the entire pilus assembly system, raised some concerns. Redesigning the primers to insert a promoter in front of each operon was not feasible due to time constraints. However, because our initial plan involved subcloning, the overhangs are compatible with other BglBrick vectors. This compatibility allowed us to select a different BglBrick vector for one of the sub-plasmids, enabling co-transformation of the vectors. As a result, we could place a promoter in front of two operons (Figure 8). This approach provided a practical solution, improving the regulation of gene expression without requiring extensive redesign.

PCR Amplification of Type IV Pilus Assembly Machinery from K12 Genome and Plasmid Preparation for Gibson Assembly

The BglBrick vectors pBbE1c-RFP and pBbE1k-RFP were streaked out for single colonies. Red/pink colonies were observed, indicating successful expression of the RFP reporter gene (Figure 9). A random colony was chosen from one of the pBbE1c-RFP plates and the pBbE1k plate to inoculate liquid cultures and placed in the 37℃ incubator overnight.

The pBbA1k vector was streaked out for single colonies and red/pink colonies were observed, indicating successful expression of the RFP reporter gene (Figure 10). A random colony from the plate was selected to inoculate a liquid culture.

After the miniprep, we used a NanoDrop to measure the DNA concentration of the samples (Table 1), allowing us to calculate the volume of vector needed for the restriction digest with BamHI and NdeI.

Table 1. NanoDrop results after minipreparation of vectors from overnight cultures.

Vector	Concentration (ng/μl)
pBbE1c	179.5
pBbE1k	118.9

The digested vectors and PCR products containing the amplified type IV pilus assembly system gene fragments from K12 were loaded onto a 1% agarose gel and ran at 130V for 30 min. The results were largely as expected, except that no band was observed for pBbA1a (Figure 11). This sample, provided by an advisor, was intended to be used in place of pBbA1k if the digest had worked. However, due to the absence of a band, we proceeded with the original plan to use pBbA1k.

Additionally, no band was observed for the pBbA1a sample. This sample, provided by an advisor, was intended to be used in place of pBbA1k if the digest had worked. However, due to the absence of a band, we proceeded with the original plan to use pBbA1k.

The pBbA1k-RFP vectors were miniprepped from the two overnight cultures and a NanoDrop was conducted (Table 2).

Table 2. NanoDrop results after minipreparation of pBbA1k from overnight cultures.

Vector	Concentration (ng/μl)
pBbA1k-1	74.6
pBbA1k-2	85.6

The pBbA1k vectors were digested with BamHI and NdeI. The digested vectors were loaded onto a 1% agarose gel and electrophoresed at 130V for 30 min. The gel shows that the pBbA1k vectors were digested as expected (Figure 12). Both the pBbE1c vector and pBbA1k vector were digested and purified to be assembled with GspilA and the type IV pilus assembly system by Gibson assembly.

PCR Amplification of GspilA with His and Collagen-Binding Tags

After designing the primers to add tags to GspilA by PCR, we attempted to simulate the assembly using SnapGene but encountered issues. The problem arose with the assembly of GspilA and hofB. The original forward primer for the hofB-hofC fragment had an overhang that was compatible with GspilA without tags. After consulting with an advisor, we determined we could just shorten the original primer. Since GspilA was synthesised with Gibson overhangs, we do not need to engineer a Gibson overhang into the hofB-hofC forward primer (Figure 13).

PCR was conducted to amplify hofB-hofC using the modified primer and to incorporate the following tags at the C-terminus of the GsPilA protein: a His tag and two different collagen-binding peptides, TKKTLRT (C1) and LRELHLNNN (C2). The PCR products were then loaded onto a 1% agarose gel and run at 130V for 30 min. As shown in Figure 14, bands corresponding to hofB-hofC and GspilA with a His tag are present. The band for GspilA-His migrated below the lowest band of the DNA ladder, suggesting the need to use a 100 bp ladder for accurate identification. However, no bands were observed for GspilA with either of the collagen-binding peptides. This issue might be due to the long overhangs of the primers relative to the annealing regions. Using a different polymerase or increasing the primer concentration might improve amplification efficiency under these conditions.

The PCR to add collagen-binding peptides to GspilA was repeated using the PCRBIO HS VeriFi™ polymerase. The PCR products were analysed on a 1% agarose gel at 130V for 30 min (Figure 15). The bands appeared streaky, making it challenging to determine the precise cutting locations, therefore, we excised both bands from each sample to ensure comprehensive recovery. For improved resolution and more accurate differentiation between the bands, we will use a molecular weight marker tailored for small DNA fragments in future experiments.

DNA Purification

The PCR products were excised from the agarose gel and purified using the Zymoclean™ Gel DNA Recovery Kit. Following purification, the samples were quantified using a NanoDrop spectrophotometer or Quibit fluorometer. The digested vectors yielded relatively low concentrations. These values, though on the lower side, are consistent with the expected outcomes for large plasmid backbones after gel extractions (Table 3).

The absence of peaks at 260 nm for pBbE1c, pBbA1k-1, and pBb1k-2 were observed, which could be attributed to several factors, including potential issues with the gel extraction process, degradation of the DNA, or low DNA concentration. To confirm the presence of DNA, we used a Qubit fluorometer, which provides more sensitive and accurate quantification (Table 3).

Table 3. Comparison of NanoDrop and Qubit results after gel extraction of type IV pilus assembly system gene fragments and digested vectors. N/A indicates that the concentration was not measured by Qubit as a peak at 260 nm was observed.

Gene	Concentration (ng/μl) from NanoDrop	Concentration (ng/μl) from Qubit
GspilA with His tag	49.6	N/A
hofB-hofC	66.1	16.6
hofB-hofC using modified primer	16.5	N/A
hofM-hofN-hofO-hofP-hofQ	50.8	66.0
ppdA-ppdB-ydgB-ppdC	78.2	119
gspO	56.4	83.7
pBbE1c	25.6	38.0
pBbE1k	38.4	Not needed, as we are not using this vector.
pBbA1k-1	10.7	12.5
pBbA1k-2	19.2	22.4

The DNA fragments with the collagen-binding tags were purified from the gel using the Zymoclean™ Gel DNA Recovery Kit and subsequently quantified with a Qubit fluorometer (Table 4).

Table 4. Qubit fluorometer values for GspilA with collagen.

Collagen	Concentration (ng/μl)
C1	30.6
C1.2	32.6
C2	19.2
C2.2	18.7

Transformation of Assembled Constructs

The digested pBbE1c vector was assembled with GspilA-His, hofB-hofC, and hofM-hofN-hofO-hofP-hofQ, while the digested pBbA1k vector was assembled with ppdA-ppdB-ygdB-ppdC and gspO using Gibson assembly. The resulting constructs were transformed into NEB 5-alpha cells plated onto LB agar containing the appropriate antibiotics. The plates were incubated overnight at 37℃ (Figure 16). The negative control plate for pBbE1c, which lacked an insert, had more colonies than the experimental plate. This suggests that the control plate likely contains cells with re-ligated vector backbones. In contrast, the pBbA1k construct plate had significantly more colonies, indicating successful Gibson assembly of the plasmid.

The digested pBbE1c vector was assembled with either GspilA-C1 or GspilA-C2, along with hofB-hofC and hofM-hofN-hofO-hofP-hofQ. The resulting constructs were transformed into NEB 5-alpha cells plated onto LB agar containing the appropriate antibiotics. The plates were incubated overnight at 37℃ (Figure 17). The control plates (Figure 17A and 17B, left) had no colonies, while the experimental plates showed colony growth. This suggests that the transformations were successful only when the complete assembly was present (Fig 17A and 17B, right). On the C1.2 experimental plate (Figure 17C, left), the presence of colonies implies that the gel fragments might have contained the correct PCR product, or there may have been an issue with the assembly process leading to unintended DNA incorporation. The absence of colonies of the C2.2 experimental plate (Figure 17C, right) indicates the Gibson assembly failed, as expected, since it lacked the gspilA insert.

Verification of pBbE1c Vector Construction Containing His Tag

To verify the construction of the pBbE1c vector, colony PCR was performed on all four colonies from the plate. The PCR products were then analysed by running them on a 1% agarose gel at 130V for 50 min. As shown in Figure 18, only one of the colonies produced a band, indicating the presence of the desired construct.

The only colony from the pBbE1c plate that produced the expected fragment in the colony PCR was used to inoculate a liquid culture. The vectors were then miniprepped from the overnight culture and quantified with a Qubit fluorometer, resulting in a concentration of 562 ng/μl.

The remaining cells containing the assembled pBbE1c construct, were plated onto LB plates supplemented with chloramphenicol. The vectors were then purified from the overnight cultures using a miniprep procedure and quantified with a Qubit fluorometer (Table 5). Both pBbE1c samples were sent off for full plasmid Nanopore sequencing.

Table 5. Qubit quantification results of miniprepped assembled pBbE1c-His vectors.

Vector	Concentration (ng/μl)
pBbE1c-1	642
pBbE1c-2	549

Verification of pBbE1c Vector Construction Containing Collagen-Binding Tags

To set up overnight cultures, ten random colonies were selected from both the pBbE1c_GspilA-C1 and pBbE1c_GspilA-C2 plates. A total of 20 overnight cultures were prepared, miniprepped, and quantified. The pBbE1C vectors assembled with GspilA-C1 were quantified with a Qubit fluorometer, while the pBbE1c vectors assembled with GspilA-C2 were quantified with a NanoDrop spectrophotometer (Table 6).

Table 6. Qubit and NanoDrop results for assembled vectors with collagen-binding peptides. (Left) Qubit results for assembled pBbE1c vectors with GspilA-C1, hofB-hofC, and hofM-hofN-hofO-hofP-hofQ. (Right) NanoDrop results for assembled pBbE1c vectors with GspilA-C2, hofB-hofC, and hofM-hofN-hofP-hofQ.

Vector sample with C1	Concentration (ng/μl)	Vector sample with C2	Concentration (ng/μl)
1	196	1	256.2
2	255	2	215.1
3	250	3	268.4
4	232	4	246.6
5	195	5	232.6
6	190	6	311.0
7	114	7	193.8
8	192	8	142.0
9	204	9	148.9
10	222	10	254.6

All 20 samples of the pBbE1c vectors containing the collagen-binding peptides were loaded onto a 1% agarose gel and run at 130V for 30 min. The results were as expected, with all samples displaying three bands (Figure 19). However, there are variations in the brightness of the bands, which can be due to variations in DNA concentration. Since all samples were successfully verified by digestion, we selected the two samples with the highest concentration for each collagen-binding peptide to send for full plasmid Nanopore sequencing.

Verification of pBbA1k Vector Construction

Six random colonies from the pBbA1k plate were used to inoculate liquid cultures. The vectors were then miniprepped from the overnight cultures and quantified with a Qubit fluorometer (Table 7).

Table 7. Qubit quantification results of miniprepped assembled pBbA1k vectors.

Sample	Concentration (ng/μl)
1	37.2
2	18.1
3	21.0
4	18.4
5	45.3
6	63.1

To verify the construction of the pBbA1k vectors, a restriction digest was performed using BamHI and NdeI. The digested vectors were run on a 1% agarose gel at 130V for 30 min. The results were as expected, revealing two distinct bands at 2455 bp and 3513 bp (Figure 20).

Due to limited sample availability for both digestion and plasmid sequencing, 0.5 μl of the overnight culture from each pBbA1k construct was plated onto LB plates supplemented with kanamycin. The vectors were then purified from the overnight cultures using a miniprep procedure and quantified with a Qubit fluorometer (Table 8). The pBbA1k constructs with the highest concentration were sent off for full plasmid Nanopore sequencing.

Table 8. Qubit results of the assembled vectors after minipreparation.

Vector	Concentration (ng/μl)
pBbA1k-1	38.5
pBbA1k-2	44.0
pBbA1k-3	37.8
pBbA1k-4	39.6
pBbA1k-5	39.4
pBbA1k-6	33.7
pBbE1c-1	642
pBbE1c-2	549

Correction of Misassembled Vectors Following Sequencing Verification

All plasmids were sent for full plasmid Nanopore sequencing. When the results came back, the only vector that had assembled correctly was the pBbA1k vector containing ppdA-ppdB-ygdB-ppdC and gspO. In contrast, the pBbE1c vectors containing the various binding tags with the rest of the type IV pilus assembly system had misassembled. The vectors with the His tag and C1 contained a similar unexpected insertion between the end of the GspilA gene and the tag. Meanwhile, vectors containing C2 showed a large deletion in the GspilA gene. These findings confirmed the presence of a design flaw (see the Cloning Lab Book for more detail).

To correct the pBbE1c vectors with the His tag, we designed primers that amplified the vector in two fragments, removing the insertion, and then reassembled the vectors using Gibson assembly. As we got the sequencing results for the vectors containing the collagen-binding peptide after confirming the correct assembly of the pBbE1c vectors with the His tag, we decided to use the His-tagged vector as a template to assemble the vectors with the collagen-binding peptides in a similar manner to correct the His-tagged vector. We also designed a primer to remove the tag and produce WT pili to serve as a baseline for conductivity testing.

The two samples of the pBbE1c vectors with the His tag sent for sequencing were both amplified by PCR and run on a 1% agarose gel at 130V for 30 min (Figure 21). The gel appeared successful, except for one missed well. Fortunately, we had two samples expected to yield the same results. We excised the bands and purified the DNA using the Zymoclean™ Gel DNA Recovery Kit. The purified DNA was then assembled using Gibson assembly to generate the plasmid.

The assembled vectors were transformed into NEB 5-alpha cells and plated onto LB agar plates supplemented with chloramphenicol (Figure 22). The growth of the plate suggests successful assembly of the vectors. However, a negative control was not included, making it difficult to rule out the possibility of background growth.

To set up overnight liquid cultures, three random colonies were selected from the plate. The plasmids were then isolated using the QIAprep Spin Miniprep Kit and quantified using a Qubit fluorometer (Table 9).

Table 9. Qubit concentrations of the corrected, miniprepped pBbE1c-His vectors.

Sample	Concentration (ng/μl)
2	151
3	89.5

The vectors were digested with BamHI and run on a 1% agarose gel at 130V for 30 min (Figure 23). Both samples showed the expected three bands, indicating that the vectors likely assembled correctly. As a result, the vectors were sent for Sanger sequencing.

To assemble the vectors containing the collagen-binding peptides and vectors without the His tag, we adapted the corrected pBbE1c-His vector. The amplified fragments were run on a 1% agarose gel at 130V for 30 min (Figure 24). The gel showed clear single bands for each sample, indicating successful amplification.

The fragments were assembled using Gibson assembly to produce the plasmids. The plasmids were then transformed into NEB 5-alpha cells and plated onto LB agar supplemented with chloramphenicol (Figure 25). Colonies can be seen on all three plates, indicating successful assembly. However, a negative control was not included, making it difficult to rule out the possibility of background growth.

From each plate, three random colonies were selected to set up overnight cultures. The plasmids were purified using the QIAprep Spin Miniprep Kit and quantified with a Qubit fluorometer (Table 10). Samples 2 and 3 of the pBbE1c-C2 vector showed notably lower concentrations compared to the other samples. This may be due to factors, such as suboptimal bacterial growth, plasmid instability, or inefficiencies during the purification process.

Table 10. Qubit results for the assembled vectors with the collagen-binding peptide and vectors without tags.

Sample	Concentration (ng/μl)
pBbE1c-C1 (1)	406
pBbE1c-C1 (2)	582
pBbE1c-C1 (3)	408
pBbE1c-C2 (1)	321
pBbE1c-C2 (2)	177
pBbE1c-C2 (3)	187
pBbE1c-WT (1)	336
pBbE1c-WT (2)	487
pBbE1c-WT (3)	252

To verify the assembly of the vectors, they were digested with BamHI, and the resulting digests were analysed on a 1% agarose gel at 130V for 30 min. If the vectors had assembled correctly, each sample would show three distinct bands. This pattern was observed in the pBbE1c-C1 vector samples, and in one of the pBbE1c-WT samples. In contrast, the other samples displayed a single band, resembling the band of the undigested vector. The digest was repeated using a shorter incubation time and the high-efficiency NEB BamHI-HF enzyme, but similar results were obtained (Figure 26). This confirmed that the issue does not lie in enzyme efficiency or DNA degradation, but rather suggests an incorrect assembly of the plasmids.

All three C1 samples, C2 sample 2, and WT sample 2 were sent for Sanger sequencing. The main objective was to verify the end of GspilA to ensure the correct tag has been added or that the WT sequence is present, with perfect alignment. While there are some point mutations elsewhere, we have previously checked the plasmid with Nanopore sequencing, so these mutations might be due to limitations of Sanger sequencing. For more information on the sequencing results, see Cloning Lab Book.

Expression and Functionality Assessment: Collagen Binding

The following strains were constructed and tested for their ability to bind to collagen: E. coli ΔsfmA (negative-control strain) E. coli ΔsfmA A1k S1a (control strain lacking GspilA) E. coli ΔsfmA A1k S1a E1c-WT (control strain expressing the unmodified GspilA) E. coli ΔsfmA A1k S1a E1c-C1 (strain expressing GspilA, modified to encode the collagen-binding tag “TKKTLRT” at its 3’ end) E. coli ΔsfmA A1k S1a E1c-C2 (strain expressing GspilA, modified to encode the collagen-binding tag “LRELHLNNN” at its 3’ end) The strains were grown on LB-agar containing the appropriate antibiotics, resuspended in phosphate-buffered saline (PBS), and transferred to wells of 96-well microtitre plates that are coated with rat tail collagen 1 (Thermo Fisher Scientific ; A1142803) and regular non-collagen coated plates. After an incubation period the wells were washed to remove cells that had not bound to the well surface. Subsequently, RFP fluorescence was quantified using a plate reader. For a detailed protocol, see Protocols.

The fluorescence reading of a control plate with unwashed wells was measured to find out the baseline fluorescence level of the cultures.

Table 11. Absolute fluorescence units of cultures after the incubation step. Separate wells of the non-coated plate were not washed and contain the original culture. The first four strains were added to the wells at an OD600 of 2.0. The strain carrying E1c-C2 was added at an OD600 of 0.5. Number of technical replicates: 4.

100 μL culture	Average	stdev
ΔsfmA	43.0	4.0
ΔsfmA A1k S1a	48445.3	284.8
ΔsfmA A1k S1a E1c-WT	33558.8	399.5
ΔsfmA A1k S1a E1c-C1	35240.8	422.5
ΔsfmA A1k S1a E1c-C2	6558.3	144.0

As seen from Figure 27, the results from the collagen binding assay might suggest that the bacteria generally bind better to the control plate than to the collagen-coated plate. We have various theories of why this is so; see our Engineering page for more detail.

Project Achievements

Project Successes

• Successfully deleted the sfmA gene from E. coli NEB 5-alpha.
• Constructed vectors for pili expression.
• Co-transformed the constructed pili expression vectors into ΔsmfA E. coli NEB 5-alpha.

Project Challenges

• Gene deletion accuracy: A labelling error in databases led to the unintentional deletion of sfmA gene instead of the intended gene.
• Design flaw: The seqeuncing results revealed that the pili expression vectors was misassembled due to incorrect design. This misassembly resulted in additional time spent fixing the vectors before proceeding.

Future Directions

Please see our Implementation page for the potential ways our project can be utilised in the future.

References

• Jervis AJ, Hanko EKR, Dunstan MS, Robinson CJ, Takano E, Scrutton NS. A plasmid toolset for CRISPR‐mediated genome editing and CRISPRi gene regulation in Escherichia coli. Microbial Biotechnology. 2021 Mar 12;14(3):1120–9. Available from: https://enviromicro-journals.onlinelibrary.wiley.com/doi/10.1111/1751-7915.13780
• Qiao J, Tan X, Ren H, Wu Z, Hu X, Wang X. Construction of an Escherichia coli Strain Lacking Fimbriae by Deleting 64 Genes and Its Application for Efficient Production of Poly(3-Hydroxybutyrate) and L-Threonine. Applied and Environmental Microbiology [Internet]. 2021 May 26 [cited 2024 Apr 25];87(12). Available from: https://journals.asm.org/doi/10.1128/aem.00381-21