Bacterial strains
All protein expression was done in strains derived from E. coli C321.A fimBE::tolC (strain A), where fimBE genes were deleted using λ-red recombination by replacing them with a tolC cassette. E. coli C321.A fimA::gent (strain B) was created by deleting fimA with gentamycin using λ-red recombination. fimA mutants in the fimBE::tolC background were created using multiplex automated genome engineering (MAGE)31. E. coli strain NEB5α was used for cloning and plasmid assembly (New England Biolabs).
Growth conditions
When necessary, chloramphenicol, kanamycin, and gentamycin were used at 30, 20, and 5 μg/mL, respectively. SDS was used at 0.005% w/v for fimBE::tolC selection. All C321 cells were grown in Luria-Bertani broth containing 5 g/L NaCl supplemented with relevant antibiotics or SDS. NEB5α cells were recovered in SOC medium as described in the NEB5α competent cell protocol (New England Biolabs).
Plasmids used
To create plasmid pSHDS.1, we cloned out via PCR the fimA-fimH pili operon from the chromosome of genetically recoded organism E. coli C321.A (C321) and then used Gibson assembly to insert fimA-H into a plasmid based on pZE21G23 under control of an aTc-inducible promoter. Plasmids pSHDS.80 and pSHDS.109 were made by creating mutations fimA A80TAG or A109TAG in pSHDS.1, respectively, using the Q5 site-directed mutagenesis protocol (New England Biolabs). Plasmid pAzFRS.2.t125 (addgene: https://www.addgene.org/73546/) was used to express the orthogonal translation system (OTS) that allows for the incorporation of nonstandard amino acids (NSAAs) into proteins.
Multiplex Automated Genome Engineering (MAGE) and λ-RED recombination
MAGE and λ-RED recombination were conducted as described elsewhere31. In short, liquid cultures were inoculated from frozen stock and grown overnight. These cultures were back-diluted 1:100 and grown to mid-logarithmic growth (OD600 ~0.6) in a shaking incubator at 34 °C. λ-red recombination proteins Exo, Beta, and Gam were expressed by keeping the cells shaking in a water bath at 42 °C for 15 minutes. Cells were immediately chilled on ice and moved to a 4 °C environment. 1 mL of cells was centrifuged at 16000 x g for 15 s. The supernatant was removed, and the cells were resuspended in milli-Q water. The cells were again spun down, the supernatant was removed, and the cells resuspended in fresh milli-Q water to wash. This process was repeated three times. After the final spin, the supernatant was removed, and either mutagenic MAGE oligos prepared at 5-6μM in DNase-free water or 50 ng of dsDNA were added directly to the cell pellet and mixed thoroughly. The oligo-cell mixture was applied to a pre-chilled 1 mm gap electroporation cuvette (Bio-Rad) and electroporated at 1.8 kV, 200 V and 25 mF. The cells were immediately resuspended in 2 mL LB broth and recovered at 34 °C in a shaking incubator for 4 hours. When the cells again reached mid-logarithmic growth additional MAGE cycles were conducted or the cells were plated for future analysis.
Successful incorporation of mutations in the fimA gene of C321 fimBE::tolC was screened using MASC-PCR as described eslewhere31.
dsDNA designed to replace fimBE with tolC and fimA with gentR contained the tolC or gentR cassettes with overhangs to regions directly before and after fimBE or fimA. Successful incorporation of tolC or gentR was confirmed by selection on SDS or gentamycin and sequencing of the fimBE and fimA regions.
Oligonucleotides
Oligonucleotides, including primers and MAGE oligos, were purchased from Keck Oligonucleotide Laboratory at Yale University.
Chromosomal pili expression
WT pili or those containing standard amino acid mutations (F, Y, or W) encoded genomically in strain A were expressed by inoculating three 500-mL flasks of LB from frozen glycerol stocks and growing the cultures at 34 °C without shaking. These cultures were grown in such conditions for 48 hours, after which cells were either imaged with TEM to visualize pili production, assayed for pili production using yeast agglutination, or used for pili protein harvesting and subsequent purification.
Nonstandard amino acid incorporation into pili and subsequent expression
Pili with inserted 4-propargyloxy-l-phenylalanine (PrOF) as a fimA A109PrOF mutation were expressed as follows. Strain B was transformed with both plasmid pSHDS.109 and pAzFRS.2.t1 and inoculated into LB containing chloramphenicol and kanamycin and grown overnight at 34 °C shaking at 225 rpm. The confluent culture was added as a 1:20 dilution to LB supplemented with chloramphenicol and kanamycin. To this culture, 20% w/v arabinose was added to a final concentration of 0.2% arabinose to induce the OTS encoded by pAzFRS.2.t1, 100 mM PrOF in 0.2 M NaOH was added to a final concentration of 1 mM PrOF, and 1 N HCl was added to a final concentration of 2 mM. The culture was grown at 34 °C shaking at 225 rpm for 3 h to induce the OTS encoded by pAzFRS.2.t1, after which anhydrotetracycline (aTc) was added to a final concentration of 60 ng/μL to induce pili containing fimA A109PrOF from pSHDS.109. Then, the culture was grown at 34 °C shaking at 135 rpm for an additional 8 h, after which cells were either imaged with TEM to visualize pili production, assayed for pili production using yeast agglutination, or used for pili protein harvesting and subsequent purification.
Yeast agglutination assay for pili production
100 μL of yeast grown overnight from frozen stock and diluted to an OD600 of ~1 in PBS pH 7.4 (−) MgCl2 (−) CaCl2 was mixed with 100 μL of E. coli grown in pili-producing conditions diluted to an OD600 of ~1 in PBS pH 7.4 (−) MgCl2 (−) CaCl2. The mixture was incubated at room temperature and shaken at 350 rpm for 15 min. 1.5 μL of mixture was pipetted onto a glass slide and the mixture imaged using a standard optical microscope. Visible agglutination of yeast cells indicated production of pili by E. coli cells. Agglutination was detected qualitatively, and pili production was confirmed with TEM imaging of E. coli samples which produced an agglutination phenotype. Adapted from Firon et al45.
Pili purification
Cells grown in pili producing conditions were spun down in 4 °C at 6000 rpm for 10 min and resuspended in 150 mM ethanolamine (ETA) pH 10.5. Cells were vortexed in a 50 mL conical tube for 2 min at vortex level 9 to shear pili from the cells, after which the cells were centrifuged in 4 °C at 10,000 rpm for 45 minutes. The supernatant was kept, and the cell pellet was discarded. The supernatant was then centrifuged in 4 °C at 23,000 × g for 1 h to remove remaining contaminants. The supernatant of this spin was kept, and the debris pellet was discarded. Saturated ammonium sulfate (SAS) (purchased from Thermo Fisher) was added to the supernatant to 15% of the final volume to precipitate out pili proteins. The solution was left static at 4 °C overnight, and then spun in 4 °C at 23,000× g for 1 h to collect precipitated pili. The supernatant was discarded, and the pili pellet was resuspended in ~600 μL 150 mM ETA pH 10.5. To this sample, 1 M MgCl2 was added to a final concentration of 100 mM MgCl2 and incubated at 4 °C overnight to precipitate out pili. The solution was then spun in 4 °C at 17,000 × g for 1 h to pellet pili. The supernatant was discarded, and the pili pellet was resuspended and solubilized in 150 mM ETA pH 10.5 by gentle pipetting to a volume of 150–500 μL depending on pellet size. The pili sample was then imaged or used for conductivity measurements. To determine purity of the sample via SDS-PAGE, the following steps were followed. To the pili sample solubilized in 150 mM ETA pH 10.5 in the previous step crystalline urea was added until the concentration was 6 M and the solution was then kept at room temperature for 4 hours. The solution was then eluted through a 20 mL Sepharose column using 6 M urea as the eluent. The first 5 mL was taken as the void volume, followed by 15 fractions, each being 1 mL in volume. Fraction 2 was found to contain the purified pili filaments. Fraction 2 was then concentrated to 60 μL using a 3kDa-cutoff Amicon Ultra-0.5 mL Centrifugal Filter. 15 μL of the concentrated fraction 2 sample was used for further gel analysis as follows. To the 15 μL sample 5μL of 4x Laemmelli buffer with 5% Beta-mercaptoethanol (BioRad) was added. The sample was then boiled for 25 minutes and then run on a SYPRO Ruby-stained SDS-PAGE gel. Once the presence and purity of FimA was confirmed, the remaining 45 μL volume of concentrated fraction 2 was then increased to 250 μL with 6 M Urea and subsequently dialyzed into 150 mM ETA pH 10.5. This protocol is partially inspired by a similar purification methodology46.
Transmission electron microscopy (TEM) imaging of pili samples
Carbon film copper grids with mesh size 400 (Electron Microscopy Sciences) were cleaned with a PlasmaFlow plasma cleaner on medium for 30 s. 5 μL of pili sample was then dropcast onto the copper grid and left to adhere to the grid for 10 min, after which the remaining buffer was blotted off using filter paper. The samples were stained with a 1% PTA stain pH 6 by floating the grids on 50 μL droplets of stain for 30 s, removing the grid and blotting the stain off with filter paper, floating the grids on the stain a second time for 30 s, and finally removing the grids and blotting the stain off with filter paper. The grids were then air dried for 10 min before storage and imaging.
Mica sample preparation
A small square of mica (Electron Microscopy Sciences inc.) was attached to tape and the top layer was peeled off, leaving an atomically flat fresh layer of mica. Onto this layer of mica, 3–4 μL of pili solution was dropcast and left to dry in a desiccator at 20% humidity overnight. The buffer was removed by washing the electrode surface with 17 μL Milli-Q water. Washing was done by dropping 17 μL of water onto the mica surface, gently pipetting the bubble of water up and down a few times, pipetting off the water, and discarding the dirty water and tip. This process was repeated three times. After the third wash, the mica sample was left to air dry for at least 45 min, after which pili were imaged using an Asylum Cypher ES atomic force microscope (Asylum Research).
Fluorescence measurements
A time-based assay was conducted on a Cary 3E UV-Vis Spectrophotometer with excitation wavelengths at 280 nm and 295 nm. The excitation bandwidth was 2.5 nm, with the emission bandwidth at 10 nm. The emission scan range for the 280 nm excitation was 295 nm to 500 nm and the emission scan range for the 295 nm excitation was 310 nm to 500 nm. The step size for the scan was 1 nm. The scan rate was 100 nm/min. The sample used was 125 μL of FimA A80W A109W pili and 125 μL FimA WT pili, dissolved in milli-Q water (pH 7.0) or 150 mM ETA pH 10.5. The emission intensity at each wavelength in the scan range was measured in counts per second. The background emission of the solvent at 125 μL was collected and subtracted to yield the final results.
Electrode device sample preparation
Electrode devices were washed with acetone, isopropanol, ethanol, and Milli-Q water, in that order, two times. After the second water wash, the device was left to air dry. The device was then plasma cleaned with the electrode side facing up using a tabletop Harrick Plasma cleaner on low for 2 min. The electrode was taken out of the plasma cleaner, and 3 μL of pili sample was immediately dropcast onto the center of the electrode, which was then left to dry in a desiccator at 20% humidity overnight. The buffer was removed by washing the electrode surface with 17 μL Milli-Q water three times. After the third wash, the electrode was left to air dry for at least 45 min. Electrodes bridged by E. coli pili were located using (Asylum Research).
Atomic force microscopy (AFM) imaging
Soft cantilevers (AC240TS-R3, Asylum Research) with a nominal force constant of 2 N/m and resonance frequencies of 70 kHz were used. The free-air amplitude of the tip was calibrated with the Asylum Research software, and the spring constant was captured by the thermal vibration method. The sample was imaged with a Cypher ES scanner using intermittent tapping (AC-air topography) mode. Images were analyzed using Asylum research v.16 and Gwyddion v.2.55.
Pili conductance measurements for individual filaments
Conductance measurements were performed as described previously11. Electrode devices with 17 electrodes spaced 300 nm apart were imaged under AFM to find individual pili filaments bridging two electrodes. Devices with pili bridging two electrodes were re-hydrated by dropping 0.3 μL of 150 mM ETA pH 10.5 onto the electrode and waiting 45 minutes. Conductance G of pili was measured in a 2-electrode configuration inside a shielded dark box using an MPI Corporation probe station connected to a semiconductor parameter analyser (Keithley 4200A-SCS). DC voltages from −0.15 to 0.15 V in increments of 0.05 V were applied between electrodes bridged by pili and current A was measured until a steady state was observed, typically over a period of two minutes. The linearity of the I-V characteristics was maintained by applying an appropriate low voltage and the slope of the I–V curve was used to determine the conductance (G). IV linear fits did not always go through (0,0), however conductivity was derived from the slope of the IV linear fit. Measurements were performed at low voltages (<0.15 V) and over longer times (>120 s) to ensure a lack of electrochemical leakage currents or faradic currents as evidenced by the absence of significant DC conductivity in buffer. All analysis was performed using IGOR Pro v.7 (WaveMetrics Inc.). Data was graphed using Graphpad Prism v.8. To display representative current-voltage curves in Figs. 2 and 4, the y-intercept of the raw linear fit was subtracted from the value of each data point such that all linear fits for the data had a y-intercept of zero. This display method preserves the slope of the linear fit and thus the conductance measurement of the filament, and allows for easy comparison of the differences in the magnitude of the conductance between filaments. Raw current-voltage data for all filaments is reported in Supplementary Fig. 3 and the source data file.
Pili conductivity calculations for individual filaments
The conductivity σ of filaments was calculated using the relation described elsewhere33 ({{{{{rm{sigma }}}}}}=frac{{{{{{rm{GL}}}}}}}{{{{{{rm{na}}}}}}}) where G is conductance, L is length of the filament measured between the electrodes, (a={{{{{rm{pi }}}}}}{r}^{2}) is area of cross section of the filament with 2(r) as the height of the filament as measured by AFM, and n is number of filaments bridging the electrodes in series. For pili with gold nanoparticles attached, the height was measured as pili + AuNP, as in each case the gold nanoparticles completely covered the part of the filament between electrodes. All analysis was performed using IGOR Pro v.7 (WaveMetrics Inc.) Data was graphed using Graphpad Prism v.8.
Pili bundling by hexamethylenediamine
Three different concentrations of hexamethylenediamine, 0 M, 0.08 M, and 0.25 M, were mixed with pure samples of FimA A80W A109W pili in 150 mM ETA at pH 10.5. At this pH, both amines of the hexamethylenediamine molecule are still protonated47. These values were chosen based on the concentrations of hexamethylenediamine used elsewhere22. One original sample of FimA A80W A109W pili was divided among the three hexamethylene concentrations to normalize pili concentration between samples. These solutions were stored at 4 °C for 7 days in order to allow bundles to form. Afterwards, the solutions were imaged using AFM to determine if bundles were present. 0 M hexamethylenediamine was used as the negative control for no bundling, 0.08 M was found to be too small for bundling in WT pili, and 0.25 M was found to cause bundling in WT pili.
Pili conductance measurements for bundles
An interdigitated electrode (IDE) with a 5 μm spacing between the finger electrodes was cleaned with acetone and dried using nitrogen gas. 0.5 µL of the FimA A80W A109W pili sample was drop cast onto the IDE and the sample was placed in a desiccator for 30 minutes to dry. Then, 5 µL of milli-Q water was added to the area where the pili were dropcast. Water was left on the sample for 1 minute, after which it was wicked away with filter paper. The sample was placed in the probe station, and the current was measured between −0.2 and 0.2 V in increments of 0.05 V over 150 s using the same setup as for measuring individual filaments. The conductance G of the pili network was found using the same method as for individual filaments.
Pili conductivity calculations for bundles
Conductivity was calculated by using the thin-film measurement formula described elsewhere33, ({{{{{rm{sigma }}}}}}=frac{{{{{{rm{GL}}}}}}}{{{{{{rm{A}}}}}}}), where G is the conductance of the film, L is the length of the gap of the electrode, 5 microns, and A is the size of the drop that covers the electrode, which, after measurement, was estimated to be 0.926 µm2.
Synthesis of gold nanoparticle-pili (AuNP-pili)
The pili A109PrOF incorporating an NSAA, PrOF, containing an alkyne, were used to produce the AuNP-pili. 5 nm NHS-activated gold nanoparticles (Cytodiagnostics) were reacted with 11-Azido-3,6,9-trioxaundecan-1-amine (Sigma Aldrich) using the recommended protocol from Cytodiagnostics. In the last step, AuNP-linker-azide were obtained by buffer exchanging with PBS pH 7.4 (−) MgCl2 (−) CaCl2 using 100 kDa Amicon Ultra centrifugal filter units (three times, centrifuging at 14,000 x g for 15 min). Next, AuNP-linker-azide were coupled onto the pili A109PrOF through copper-catalyzed azide-alkyne cycloaddition reaction. Briefly, 20 µL of 50 mM THPTA and 10 µL of 20 mM CuSO4 were premixed for 30 min at room temperature. Later, 10 µL of 150 µM pili were diluted in 30μL of PBS pH 7.4 (−) MgCl2 (−) CaCl2 and 2.5 µL of AuNP-linker-azide were added to the mixture. Then, 1 µL of the mixture of THPTA/CuSO4 was added to the entire 40μL pili/AuNP mixture, are combined. Finally, 2.5 µL of 100 mM aminoguanidine and 2.5 µL of 100 mM sodium ascorbate are added. The reaction mixture was shaken at 500 rpm tabletop shaker (Santa Cruz Biotechnology) for 1 h. To stop the reaction and obtain the AuNP-linker-pili, the samples were buffer exchanged to 150 mM ETA pH 10.5 using 100 kDa Amicon Ultra centrifugal filter units (three times, centrifuging at 14,000 × g for 15 min).
MD Simulation of pili bundling
The atomic structure of the FimA monomer was obtained from the protein data bank (PDB ID 6C53). Polymerization of FimA into the type I pilus involves the insertion of an N-terminal beta strand formed by the first 20 residues of the neighboring subunit into an open groove in the immunoglobulin-like fold formed by residues 20 to 158 of the subunit of interest17. To construct our simulation model, we truncated the first 18 residues and added residues 3 to 20 of the neighboring subunit resulting in a complete immunoglobulin-like fold without an extraneous beta strand. Two of these monomer subunits were separated by a distance of 40 Å and rotated such that their outer surfaces faced each other. The hexamethylenediamine (HMD) molecule was parameterized using the CHARMM-GUI48 and the CHARMM49 general force field in its fully protonated state with a charge of +2. For our simulation with 250 mM HMD, we added 55 HMD molecules around the protein, calculated as the number of HMD molecules required to solvate the box with 250 mM. Simulation systems both with and without HMD were solvated in a TIP3P water box with dimensions of 101 × 88 × 61 Å using the VMD50 solvate plugin. This provides at least 12 Å between the edge of the box and the closest protein atom. The Particle-Mesh Ewald (PME) method51 was utilized to calculate long-range electrostatic interactions with 90 grid points in the x-direction, 80 grid points in the y-direction, and 54 grid points in the z-direction with a 12 Å cut-off. This spherical cut-off also extends to Lennard-Jones parameters. The temperature was maintained using a Langevin thermostat, and the pressure was maintained by using a constant Nosé-Hoover method in which Langevin dynamics is used to control fluctuations in the barostat. The velocity Verlet algorithm was used with a time-step of 1 fs.
MD simulations were performed with NAMD v2.1352 using the CHARMM3653 force-field parameters with periodic boundary conditions. First, each system was minimized, followed by a 500 ps simulation of the water box and any HMD molecules with fixed protein atoms. The models were then equilibrated to 310 K in the NVT ensemble for 3.5 ns under harmonic restraints with a force constant of 0.1 kcal/mol to the amino acid sidechains and a force constant of 1.0 kcal/mol to the protein backbone. Production runs were then performed in an NPT ensemble for 100 ns with frames being written to the trajectory every 2.5 ps. Analysis was performed by computing the distance between the geometric centers of each monomer subunit using a custom Tcl script available as part of the supplementary information. All of the simulations were done on the Grace supercomputing cluster, a computational core facility located at Yale University’s Centre for Research Computing.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
