Tight-packing of large pilin subunits provides distinct structural and mechanical properties for the Myxococcus xanthus type IVa pilus

Significance To translocate on solid surfaces many bacteria use dynamic cell surface structures called type IVa pili (T4aP). T4aP undergo cycles of extension, adhesion to a surface, and retraction that generate a force sufficient to pull a cell forward. The T4aP filament is composed of thousands of copies of the major pilin. Here, using cryo-EM, we solve the structure of the T4aP of Myxococcus xanthus (T4aPMx), which has an unusual large major pilin called PilA. The large major pilins of T4aPMx are tightly packed and have extensive intersubunit contacts, making the T4aPMx more rigid and stronger than other studied T4aP. Interestingly, large major T4a pilins are found in several bacterial phyla and might represent an evolutionary adaptation to specific environments.


Supplementary Materials and Methods
Bioinformatics.Sequences of the K02650 (type IV pilus assembly protein PilA) were extracted from the KEGG SSDB database (1).To filter out highly homologous sequences we used the cdhit program with a threshold of 90% sequence identity (2).The 2308 obtained sequences were analyzed for presence of a class 3 signal peptide (SPIII) and subsequently processed into the mature pilin form using SignalP (6.0) (3).The taxonomic classification of the remaining 2071 pilin sequences was collected from KEGG SSDB database and sequences without a bacterial classification as well as sequences from bacterial phyla, only represented by one genome, were excluded from the analysis.The 1955 remaining pilin sequences (Dataset S1) were analyzed with the PROMALS3D multiple sequence and structure alignment server (4) to obtain aa and secondary structure consensus sequences.Alignments were generated using T-Coffee (5) and the ClustalW output format (6).They were shaded using the BoxShade Server.Residues on black are conserved in ≥60% of the proteins, and residues on gray are similar in ≥60% of the proteins.For the phylogenetic tree, the ANCESCON tool (7) of the MPI bioinformatics Toolkit (8) was used.The phylogenetic tree was annotated using iTol (v6) (9).The MIDAS motifs (DxSxS) were identified manually.The domains of PilY1 proteins were identified using Smart (10).
AlphaFold-Multimer model building.Structures were predicted using AlphaFold-Multimer modeling via ColabFold (Version 1.5.0)(11)(12)(13).ColabFold was run locally using colabfold_batch with default settings on an Nvidia A100 graphics card with 40 GB memory.Five models were generated and ranked according to their pTM and iPTM scores (11).The predicted Local Distance Difference Test (pLDDT) and predicted Alignment Error (pAE) graphs of the five models generated by ColabFold were visualized as vector-graphics using a custom Matlab R2020a (The MathWorks) script (14).Per residue model accuracy was estimated based on pLDDT values (>90, high accuracy; 70-90, generally good accuracy; 50-70, low accuracy; <50, should not be interpreted) (12).Relative domain positions were validated by pAE.The pAE graphs indicate the expected position error at residue X if the predicted and true structures were aligned on residue Y; the lower the pAE value, the higher the accuracy of the relative position of residue pairs and, consequently, the relative position of domains/subunits/proteins (12).PyMOL version 2.4.1 (Schrödinger LLC) was used to analyze and visualize the models.The PDB-text files of the AlphaFold-Multimer models of the two tip complexes (models 1, Fig. 4 A, B) are shown in Dataset S2.Structure superposition of the top PilA of the T4aP Mx with the bottom PilA of the AlphaFold-Multimer model was done using the align function in PyMol (root mean square deviation=0.841).The protein sequences of the mature minor pilins and PilY1 proteins without their signal peptides as reported earlier (15), were used for generating the models.

AFM tip functionalization and FS.
Gold AFM probes (PNP-TR, NanoWorld) were incubated overnight in a 1mM ethanolic 1-dodecanethiol solution to modify them with methyl groups and render them hydrophobic, then rinsed with ethanol and sterile water and kept in milliQ water at 4°C until use (no longer than 48 hrs).Early exponential-phase M. xanthus cultures (OD550 ~ 0.5) grown in CTT at 32°C in the dark, were diluted in MC7 buffer (10 mM MOPS pH 7.6, 1 mM CaCl2) and passaged gently through a 26 gauge needle to dissolve cell aggregates before seeding cells in a 35mm untreated polystyrene Petri dish (Corning).After 30 min incubation to allow cells to adhere, they were adequately adhering for high-quality AFM-FS, gently rinsed with MC7 buffer, and immediately used for AFM.Prior to any AFM measurements, the spring constant of the probe's cantilever was determined as reported previously (16) allowing for the accurate correlation between measured cantilever deflection and tensile force in stretched T4aP.AFM recordings were done at room temperature using a NanoWizard® 4 NanoScience AFM (JPK Instruments) in force mapping (volume) mode, which allows the recording of F-d curves in a pixel-by-pixel fashion over a defined surface area in a raster array.Single adherent M. xanthus cells were visualized with an inverted microscope and force probed in MC7 buffer with the hydrophobic AFM tip.For each tip-cell pair, a large (10×10 µm, 32 × 32 pixels) force map was first recorded to generate a topographical image of a whole cell and to find pilus signatures localized to one of the cell poles (17).Subsequently, a smaller (3×3 µm, 32×32 or 16×32 pixels) map was recorded over a piliated pole.Sample height could be determined from the approach section of each F-d curve, while the retract portion provided pilus forced extension and adhesive information.F-d curve analysis was performed using the JPK data processing software.The nanospring constant (given as k pilus below) was calculated using the serial spring equation: , with keff (effective spring constant) equal to the slope of the linear region of a nanospring extension profile in an F-d curve and ks (spring constant of the cantilever sensor).Graphs pertaining to AFM data was generated using R Studio.
T4aP purification and T4aP shearing assays.For structural analysis and persistence length measurements, T4aP were sheared off from WT and the hyper-piliated ∆pilT strain using a modified protocol (15) based on the procedure of (18).Briefly, cells grown on 1% CTT, 1.5% agar plates (10 plates, 12x12 cm) for 2-3 days were gently scraped off the agar and resuspended in 4 ml/plate pili resuspension buffer (100mM Tris-HCl pH 7.6, 150mM NaCl).The pooled suspension was vortexed for 10 min at the highest speed and centrifuged for 20 min at 13,000 g at 4°C to remove cell debris.The supernatant was centrifuged twice for 10 min at 13,000 g at 4°C.T4aP in the cell-free supernatant were precipitated by adding 10× pili precipitation buffer (final concentrations: 100mM MgCl2, 500mM NaCl, 2% PEG 6000) for at least 3 hrs at 4°C.The solution was centrifuged for 30 min at 13,000 g at 4°C, and the pellet resuspended 1ml pili resuspension buffer.The pili solution was loaded on top of a centrifuge tube containing a 10-70% sucrose gradient (29 ml) of pili resuspension buffer.After 15 hrs centrifugation at 115,000xg in a swing bucket rotor (SW72Ti) at 4°C, the tube was punched at the bottom, and 1.5 ml fractions harvested and analyzed by SDS-PAGE using SDS-lysis buffer (10% (v/v) glycerol, 50mM Tris-HCl pH 6.8, 2mM EDTA, 2% (w/v) SDS, 100mM DTT, 0.01% bromphenol blue).To remove the sucrose, PilA-containing fractions were diluted 13.5 fold in pili resuspension buffer, and the solutions precipitated again with pili precipitation buffer (s.a.).The pili were resuspended in pili resuspension buffer.Only T4aP purified from the hyper-piliated ∆pilT strain were of sufficient purity and quantity for structural analyses.
For T4aP shearing assays, 60 mg cells grown on 1% CTT, 1.5% agar plates for 2-3 days were gently scraped off the agar and resuspended in pili resuspension buffer.Cell suspensions were vortexed for 10 min at the highest speed.Cells from a 100 µl aliquot were harvested, the pellet solved in 200 µl SDS lysis buffer, and immediately denatured at 95°C for 5 min.This represents the cellular fraction.The remaining suspension was centrifuged for 20 min at 13,000 g at 4°C.
The supernatant was removed and centrifuged twice for 10 min at 13,000 g at 4°C to remove cell debris.T4aP in the cell-free supernatant was precipitated by adding 10× pili precipitation buffer for at least 3 hrs at 4°C.The solution was centrifuged for 30 min at 13,000 g at 4°C, and the pellet was resuspended in SDS lysis buffer (for immunoblotting) or in pili resuspension buffer for negative staining (1µl per mg vortexed cells).T4aP sheared and purified from the same amount of cells of the respective parent strain (WT or ∆pilT) were loaded and separated by SDS-PAGE.For WT and its derivatives T4aP sheared from ~15 mg cells and 40 µg of protein from cellular fractions (~0.4 mg cells) were loaded per lane.For the hyperpiliated ∆pilT and its derivatives T4aP sheared from ~0.6 mg cells and 10 µg of protein from cellular fractions (~0.1 mg) were loaded per lane.
T4aP-dependent motility assays.Cells from exponentially growing M. xanthus cultures were harvested and resuspended in 1% CTT to a calculated density of 7×10 9 cells ml -1 .5µl of cell suspension were spotted on soft agar CTT plates (0.5% casitone, 10mM Tris-HCl pH 8.0, 1mM KPO4 pH 7.6, 8mM MgSO4, with the indicated concentrations of select agar (Invitrogen)) and incubated at 32°C for 24 hrs.Colony edges were imaged using a Leica MZ75 stereomicroscope with a Leica MC120 HD camera.Colony diameters are measured at time point zero and after 24 h up to the longest flares.Numbers in mm represents the increase in colony diameter within 24 hrs.

Transmission electron microscopy.
For negative-staining of pili ~5 µl of sheared T4aP solutions were applied on 300 mesh Formvar/carbon copper-grids.After 10 min, grids were washed twice with water and negative-staining was done with a solution based on an Organotungsten compound (Nano-W, Nanoprobes).For negative-staining of cells 300 mesh Formvar/carbon copper-grids were pretreated for 1 hr with each 50 µl of a freshly prepared chitosan dilution (100-fold dilution in water of chitosan stock solution=15mg/ml chitosan in 2M acetic acid).After 1 hr the chitosan solution was blotted away with filter paper and the grids were dried.An aliquot of exponentially growing M. xanthus cultures (5-20 µl) were put on top of a pretreated grid and cells were allowed to attach to the grid at 32°C in a humid chamber.After 3 hrs, grids were washed twice with water and negative-staining was done with a solution based on an Organotungsten compound (Nano-W, Nanoprobes).Grids were inspected with a JEM-1400 electron microscope (JEOL) at 100 kV.

Persistence length determination.
To determine the flexibility of different T4aP, persistence length measurements were performed using micrographs of negatively stained T4aP from the indicated strains.For each strain, 30-50 filaments were traced using the ImageJ analysis tool (19).Persistence length (L) is determined by the statistical relationship of cos (θ) and contour length (λ), according to exp(-λ/L)=<cos(θ)>.

Pilus length determination.
To determine the length of T4aP from WT and PilA-variants, micrographs of negatively stained cells from indicated strains were analyzed using the MetaMorph application (Meta Imaging Series 7.8).For each strain, length of T4aP from three cells were measured.The mean T4aP length ± standard deviation (STDEV) is shown in µm.
Student's t-test was used to test whether the mean T4aP length of the variants is significantly different from WT mean T4aP length (p-value < 0.05).

Supplementary Fig. 1. Phylogenetic tree of 226 large major T4a pilins.
Phylogenetic tree of 226 large major pilins, as listed in SI Appendix, Table S1.The large pilins of the Betaproteobacteria are found especially in Burkholderiales, and those are shown in bold with the abbreviation Burk.at the end of the locus tag.The numbers at the outer edge of the circle indicate large pilins with six or more cysteine residues.
The SF was further purified using sucrose-gradient centrifugation and fractions were collected.A-D.The sequences of the mature minor pilins, and PilY1 proteins without their signal peptides as reported earlier (15), were used for generating the models.length of the segment (Å).PL (µm) is calculated from the slope of the linear trendline as in Fig. S5A.The mean contour length ± standard deviation (STDEV) is shown in µm in the upper right corner.Negative staining micrographs of sheared pili from ∆pilT strains expressing PilA WT and indicated PilA variants as shown in Fig. 5F were used for this analysis.

B.
WT and PilA-variants were imaged by TEM and negative staining electron micrographs of cells from indicated strains were analyzed.For each indicated strain, the length of T4aP in µm from three cells was measured (diamond, circle, star).
SF and PilA-containing fractions (5-8) were separated by SDS-PAGE and visualized by Coomassie protein staining (A) and probed with α-PilA antibodies (B).The calculated molecular mass of PilA and positions of molecular markers are indicated.C. Representative micrograph of negatively stained T4aP Mx .Scale bar, 200nm.D. Representative cryo-EM micrograph of T4aP Mx .Scale bar, 500nm.E. Averaged power spectrum from T4aP Mx .Supplementary Fig. 3. Comparisons of T4aP Mx and PilA Mx to other T4aP structures and T4a pilins.A. Helical arrangement of depicted T4aP with N and N+4 subunits shown in the same color in a side view (top row) and viewed from top (2 nd row).T4aP characteristics (resolution, diameter, rise, twist, subunits/turn and pilin length were taken from this study (Mx_8TJ2) and previously solved T4aP structures (N.gonorrhoeae, Ng_5VXX (20); N. meningitidis, Nm_5KUA (21); P. aeruginosa PAK, Pa_5VXY (20); T. thermophilus, Tt_6XXD and Tt_6XXE (22); enterohemorrhagic E. coli, Ec_6GV9 (23), G. sulfurreducens, Gs_6VK9 (24); S. sanguinis, Ss_8PFB (25)).T4aP are shown as 12mers with the exception of Ss_8PFB, which is shown as a 9mer.Note that the major pilin of G. sulfurreducens is heterodimeric and composed of PilA-N (61 aa) and PilA-C (104 aa) (24), and the T4aP from S. sanguinis is heteropolymeric and composed of two very similar major pilins PilE1 and PilE2 (139 and 132 aa) (25).The last row indicates persistence length as measured and depicted in Fig. S5A.B. Ribbon representation of PilA Mx and depicted pilins from previously solved T4aP structures as in A with helical elements (α) shown in red, β-stranded elements (β) shown in blue and lessstructured areas (loops) in grey.The first and last residues of the helices of α1-N and α1-C are shown as well as the number of residues (aa) in the melted region.Pilin characteristics (length of pilin and globular domain (aa) and % structured (α+β) were taken from this study and previously solved T4aP structures as in A. C. Multiple sequence alignment of PilA Mx and pilins from previously solved T4aP structures as in A. The top row indicates the structural elements of PilA Mx as in Fig. 3B.The blue shaded areas indicate extra residues of PilA Mx in two areas (between β1 and α3 and between β2 and β4).Residues on black are conserved in ≥60% of the proteins, and residues on gray are similar in ≥60% of the proteins.The green shaded box at the end of the alignment depicts numbers of cysteine residues and disulfide bridges in the aligned T4a pilins.4A.B. Plots generated by AlphaFold-Multimer for the complex of cluster_1 proteins (FimU1, PilV1, PilW1, PilX1, and PilY1.1) and PilA as in A. The color code of the six proteins is as in Fig. 4B.C. Predicted protein-protein interaction by β-strand addition between PilY1.3 (purple) and PilX3 (blue) by the AlphaFold-Multimer model.The boxed area shows the area of the β-strand addition without α1 of PilX3.D. Predicted protein-protein interaction by β-strand addition between PilY1.1 (purple) and PilX1 (blue) by the AlphaFold model of cluster_1 proteins.The boxed area shows the area of the βstrand addition without α1 of PilX1.
for cloning purposes are indicated in bold.Restriction sites are underlined.267268 The number represents the total number of large pilins in the corresponding phyla or 251 probacterial class, respectively.*The number represents the mature major pilin length in aa.
#1Plasmids used for construction of pilA-variants at the endogenous locus are indicated in brackets.