The mechanism for polar localization of the type IVa pilus machine in Myxococcus xanthus

ABSTRACT Type IVa pili (T4aP) are important for bacterial motility, adhesion, biofilm formation, and virulence. This versatility is based on their cycles of extension, adhesion, and retraction. The conserved T4aP machine (T4aPM) drives these cycles; however, the piliation pattern varies between species. To understand how these patterns are established, we focused on the T4aPM in Myxococcus xanthus that assembles following an outside-in pathway, starting with the polar incorporation of the PilQ secretin forming a multimeric T4aP conduit in the outer membrane. We demonstrate that PilQ recruitment to the nascent poles initiates during cytokinesis, but most are recruited to the new poles in the daughters after the completion of cytokinesis. This recruitment depends on the peptidoglycan-binding AMIN domains in PilQ. Moreover, the pilotin Tgl stimulates PilQ multimerization in the outer membrane, is transiently recruited to the nascent and new poles in a PilQ-dependent manner, and dissociates after the completion of secretin assembly. Altogether, our data support a model whereby PilQ polar recruitment and multimerization occur in two steps: the PilQ AMIN domains bind septal and polar peptidoglycan, thereby enabling polar Tgl localization, which then stimulates secretin multimerization in the outer membrane. Using computational analyses, we provide evidence for a conserved mechanism of T4aPM pilotins whereby the pilotin transiently interacts with the unfolded β-lip of the secretin monomer, i.e., the region that eventually inserts into the outer membrane. Finally, we suggest that the presence/absence of AMIN domain(s) in T4aPM secretins contributes to the different T4aPM localization patterns across bacteria. IMPORTANCE Type IVa pili (T4aP) are widespread bacterial cell surface structures with important functions in motility, surface adhesion, biofilm formation, and virulence. Different bacteria have adapted different piliation patterns. To address how these patterns are established, we focused on the bipolar localization of the T4aP machine in the model organism Myxococcus xanthus by studying the localization of the PilQ secretin, the first component of this machine that assembles at the poles. Based on experiments using a combination of fluorescence microscopy, biochemistry, and computational structural analysis, we propose that PilQ, and specifically its AMIN domains, binds septal and polar peptidoglycan, thereby enabling polar Tgl localization, which then stimulates PilQ multimerization in the outer membrane. We also propose that the presence and absence of AMIN domains in T4aP secretins contribute to the different piliation patterns across bacteria.

mechanisms, i.e., the extension/retraction of surface-exposed type IVa pili (T4aP) that enable cells to translocate across solid surfaces and the rotation of surface-exposed flagella that enable cells to swim through liquids or swarm across semisolid surfaces (2).Interestingly, the patterns in which these machines are positioned in cells vary between species (2)(3)(4).For both flagella and T4aP, these distinct patterns are important for efficient motility, biofilm formation, and virulence (3,5,6).How these patterns are established is poorly understood.
T4aP are highly versatile and not only important for motility but also for surface sensing, adhesion to and colonization of host cells and abiotic surfaces, biofilm formation, virulence, predation, and DNA uptake (4,7).The versatility of T4aP is based on their cycles of extension, surface adhesion, and retraction that are driven by the T4aP machine (T4aPM), a multiprotein complex that consists of at least 15 different proteins and spans from the outer membrane (OM) to the cytoplasm (Fig. 1) (4,(8)(9)(10)(11)(12).Cryo-elec tron tomography of the piliated and non-piliated forms of the T4aPM of Myxococcus xanthus and Thermus thermophilus revealed that both forms are multilayered structures (8, 9) (Fig. 1).However, while the architecture of the T4aPM is conserved, bacteria have adapted different piliation patterns.Specifically, in the rod-shaped cells of Pseudomonas aeruginosa (13,14) and M. xanthus (15,16), T4aP localization alternates between the two cell poles, in the rod-shaped Neisseria elongata cells, T4aP are unipolarly localized (17), while in the rod-shaped Thermosynechococcus vulcanus cells, they localize at both cell poles simultaneously (18).They localize in a "line along the long cell axis" (from hereon, lateral pattern) in the coccobacillus-shaped Acinetobacter baylyi cells (6), to the junctions between cells in the hormogonium of Nostoc punctiforme (19), and peritrichously in the rods of Burkholderia cepacia (20) and in the coccoid-shaped cells of Neisseria meningitidis  (8).The T4aPM is divided into five parts: (i) The OM secretin channel is formed by PilQ and stabilized by the LysM domain protein TsaP (29).
(ii) The PilN/-O/-P periplasmic alignment complex is anchored in the inner membrane (IM) and interacts with PilQ.(iii) The IM/cytoplasmic platform complex is composed of PilC/-M.(iv) The extension/retraction ATPases PilB/-T bind to the cytoplasmic base of the T4aPM in a mutually exclusive manner (8,31,32).
(v) The pilus fiber is formed by PilA subunits and a priming complex, composed of PilY1 and four minor pilins (blue: PilX, green: PilW, orange: PilV, and yellow: FimU), that remains at the tip of the extended T4aP (8,12).Tgl is an OM lipoprotein that is required for PilQ secretin assembly (28,30).Bent arrows indicate incorporation of and removal from the pilus base of PilA during extension and retraction, respectively.
Proteins labeled with single letters have the Pil prefix.
As noted, M. xanthus assembles T4aPM at both poles, but the T4aPM are only active at one pole at a time and this pole changes on average every 10-15 min (33,34).The assembly of the T4aPM at the two poles in M. xanthus depends on the OM secretin PilQ (Fig. 1) and follows an outside-in pathway (8,28).Without the secretin, the remaining components either do not accumulate or are not incorporated into the T4aPM (28).Moreover, assembly of the T4aPM was suggested to occur at the nascent and new cell pole during and immediately after the completion of cytokinesis (28).The secretin forms the conduit for the T4aP in the OM (8, 9) (Fig. 1).In contrast to canonical OM β-barrel proteins in which a single polypeptide forms the β-barrel, the β-barrel formed by a secretin is generated from 12 to 15 subunits and most of the secretin pore is periplasmic (8,(35)(36)(37)(38)(39). Secretin protomers comprise two major subdomains, an N-terminal species-specific region and the C-terminal conserved pore-forming region (35).The N-terminal region contains at least two N-domains and, in the case of T4aPM secretins, also occasionally one or more peptidoglycan (PG)-binding AMIN domains (25,35,37).The C-terminal region comprises the secretin domain and the β-lip region and forms most of the barrel and a gate that closes the pore (37,40).The periplasmic part of the secretin oligomer forms a large vestibule, which is open toward the periplasm and closed toward the OM by the gate (35).Secretins also facilitate substrate translocation across the OM in other T4P systems, in type II secretion systems (T2SS) and in type III secretion systems (T3SS) (35,36).For their assembly in the OM, secretins rely on a cognate pilotin protein that assists in (i) secretin monomer transport to OM (41)(42)(43), (ii) secretin insertion in the OM (40,44), (iii) secretin oligomerization in the OM (28,30,40,42,45), and/or (iv) protection of the secretin monomer from proteolytic degradation (46,47).Pilotins are OM periplasmic lipoproteins (36).In M. xanthus, the OM lipoprotein Tgl is the cognate PilQ pilotin (48-51).Lack of Tgl causes a defect in PilQ multimerization (28,30).Consequently, in the absence of Tgl, the remaining parts of the T4aPM do not accumulate or not assemble (28).Interestingly, the lack of PilQ assembly in Δtgl cells can be extracellularly complemented by Tgl + cells in a process referred to as OM-exchange, in which OM proteins are exchanged between cells (48, [51][52][53].
To understand how T4aPM becomes polarly localized in M. xanthus, we investiga ted when and how the PilQ secretin is recruited to the poles.We show that PilQ recruitment to the nascent pole initiates during cytokinesis, but most is recruited to the new poles in the daughter cells after completion of cytokinesis.We also demonstrate that Tgl is transiently recruited to the nascent and new poles during and after cytokinesis in a PilQ-dependent manner, and that Tgl dissociates after secretin assembly is completed.Based on a structure-function analysis of PilQ, our data support that its N-terminal PG-binding AMIN domains are crucial for its septal and polar recruitment, likely via binding to PG specific to the septum and cell poles.Our data support a model whereby PilQ monomers are recruited to the nascent and new cell poles by specific septal and polar PG via their AMIN domains, thereby enabling Tgl localization and, consequently, secretin assembly in the OM.Additionally, we propose a conserved mechanism for the T4aPM pilotin/secretin interaction, in which the pilotin transiently interacts with the unfolded β-lip of the secretin monomer, i.e., the region that eventually inserts into the OM.

The secretin PilQ is stably recruited to the nascent and new poles
Previously, PilQ was suggested to be recruited to the nascent and new poles of M. xanthus cells during and immediately after the completion of cytokinesis (28).In those experiments, a partially active PilQ-sfGFP fusion that accumulated at a reduced level was used (28).To reassess PilQ recruitment to the nascent pole, we used a strain in which pilQ at the native site is replaced by pilQ-sfGFP by double homologous recombination (5).In immunoblots, the active heat-and SDS-resistant PilQ-sfGFP multimer accumulated at close to native levels, while the monomer was only detected at a very low level (Fig. 2A; Fig. S1A).Of note, a small fraction of PilQ-sfGFP was cleaved to PilQ and sfGFP (Fig. 2A).
In agreement with previous observations that the M. xanthus T4aPM assembles at both cell poles (8,(27)(28)(29)(30), PilQ-sfGFP overall localized in a bipolar pattern (Fig. 2B) and ~19 ± 6% of the fluorescent signal is polar (5).However, we noticed that long cells had more symmetric bipolar PilQ-sfGFP clusters, while short cells had a higher degree of asymmetry and a few short cells even only had a unipolar signal (Fig. 2B and C).We did not reliably identify dividing cells with PilQ-sfGFP at the nascent poles at the constriction site at mid-cell.
To determine whether PilQ-sfGFP can be recruited to the nascent cell poles during cytokinesis, we treated cells with cephalexin to inhibit FtsI that catalyzes PG cross-linking at the septum (55) and blocks cytokinesis after the initiation of constriction in M. xanthus (56,57).In cells treated with cephalexin for 4-5 h, largely corresponding to one doubling time, the cell length had increased, the bipolar PilQ-sfGFP signals were more symmet rical, unipolar PilQ-sfGFP localization was not observed, and, importantly, PilQ-sfGFP localized at the constriction site at mid-cell in two-thirds of the cells (Fig. 2B).The cluster at the constriction site was stable when treated cells were followed by time-lapse fluorescence microscopy (Fig. 2D).These observations demonstrate that PilQ-sfGFP can be stably incorporated into the nascent poles during cytokinesis.Consistent with these findings, we observed by time-lapse fluorescence microscopy of untreated PilQ-sfGFPexpressing cells that ~20% of cells had a very faint PilQ-sfGFP cluster at the constriction site at mid-cell up to 50 min prior to completion of cytokinesis (Fig. 2E).However, most of the clusters only became clearly visible at the new poles after completion of cytokinesis, and on average, a polar cluster became reliably visible 20 min after completion of cytokinesis.In the daughter cells, the PilQ-sfGFP clusters at the new poles increased in intensity over time, and mostly during the first 60-90 min after completion of cytokine sis, ultimately resulting in the more symmetric bipolar localization pattern (Fig. 2E and F).
We conclude that recruitment of PilQ to the nascent poles initiates during cytokinesis, but most of PilQ is recruited over the first 60-90 min after completion of cytokinesis resulting in a symmetric bipolar localization of PilQ.We speculate that we did not detect a PilQ-sfGFP signal at the site of division in the analysis of snapshots (Fig. 2B) because the PilQ-sfGFP signal before completion of cytokinesis is too faint to be reliably detected and only becomes reliably detected when cells are followed in time-lapse microscopy experiments.

The pilotin Tgl is transiently recruited to the nascent and new poles
Next, we investigated the localization of the pilotin Tgl.We previously analyzed Tgl localization using a strain overexpressing an active Tgl-sfGFP protein and found that it localized to the cell envelope but not specifically at the cell poles or the division site (28).By contrast, Nudleman et al. found by immunostaining that Tgl localized unipolarly in ~30% of the cells (30).To resolve the localization of Tgl, we generated a strain expressing the active Tgl-sfGFP fusion (Fig. S1A) from the native site and at native levels (Fig. 3A) by integrating the tgl-sfGFP fusion construct at the tgl locus by double homologous recombination.Subsequently, we reevaluated its localization.
In all cells, Tgl-sfGFP localized along the entire cell periphery in a pattern typical of proteins localizing to the cell envelope (Fig. 3B).Moreover, in 14% of the cells, Tgl-sfGFP also localized in a unipolar cluster, and these cells were typically short in length (Fig. 3B  and C).Additionally, in 4% of the cells, Tgl-sfGFP localized at the constriction site, and these were typically long cells (Fig. 3B and C).In the remaining cells, Tgl-sfGFP did not form clusters (Fig. 3B).We note that Tgl-sfGFP localization is very different from the bipolar localization of the T4aPM in M. xanthus cells.We speculate that in our previous analysis of Tgl-sfGFP localization, its overexpression and the resulting strong cell envelope signal likely masked the weak Tgl-sfGFP clusters at the nascent and new poles.
Treatment of Tgl-sfGFP-expressing cells with cephalexin caused a significant increase in the fraction of cells with a mid-cell cluster, and while the fraction of cells with a unipolar signal remained unchanged, this signal was substantially weaker than in untreated cells (Fig. 3B).Of note, the fraction of cells with Tgl-sfGFP at mid-cell was significantly lower than in the case of PilQ-sfGFP in cephalexin-treated cells (Fig. 2B).When cephalexin-treated cells were followed by time-lapse fluorescence microscopy, we observed Tgl-sfGFP clusters appear at mid-cell in cells with constrictions, and these clusters disintegrated after ~100 min (Fig. 3D).Similarly, time-lapse fluorescence microscopy of untreated Tgl-sfGFP-expressing cells showed that the protein on average appeared at mid-cell in constricting cells ~10 min before completion of cytokinesis but in a few cells the cluster appeared up to 50 min prior to completion of cytokinesis (Fig. 3E).Upon completion of cytokinesis, the two daughters each inherited a cluster at the new pole that eventually disintegrated (Fig. 3E).The lifetime of a cluster from its first appear ance until it permanently disintegrated was ~70 min (Fig. 3E).Given a generation time of 5-6 h, this lifetime correlates well with the percentage of cells with unipolar and mid-cell clusters quantified in snapshots of cells expressing Tgl-sfGFP (Fig. 3B).Interestingly, the lifetime of a Tgl-sfGFP cluster coincides with the time (60-90 min) required for polar incorporation of PilQ-sfGFP at the nascent and new poles (Fig. 2E and F).
The observations that the PilQ-sfGFP cluster stably remains at mid-cell in cephalexintreated cells (Fig. 2D), while the Tgl-sfGFP cluster disintegrates in the presence or absence of cephalexin, support that Tgl is transiently localized to the nascent and new poles to promote secretin assembly in the OM and is not part of the fully assembled T4aPM.

Tgl is important for multimerization and stability of PilQ and PilQ is impor tant for polar recruitment of Tgl
Our fluorescence microscopy analyses showed that Tgl-sfGFP on average formed a visible cluster at mid-cell slightly earlier than PilQ-sfGFP (Fig. 2E vs Fig. 3E).We, therefore, hypothesized that Tgl could be responsible for recruiting PilQ to mid-cell during cytokinesis.To this end, we analyzed the protein accumulation and localization of each fluorescent fusion in the absence of the other.
In agreement with previous observations (28,30), only the monomer fraction of PilQ and PilQ-sfGFP accumulated in the Δtgl mutant (Fig. 2A and 3A), confirming that Tgl is important for multimerization of PilQ.We also noticed that the total level of the PilQ  variants, and especially of PilQ-sfGFP, was reduced in the absence of Tgl (Fig. 2A and 3A), arguing that Tgl is also important for PilQ stability.Accordingly, the PilQ-sfGFP fluorescent signal was strongly reduced, and as reported (28), polar and mid-cell clusters were not detected (Fig. 4A).In previous immunofluorescence studies using a Δtgl::tet r strain, PilQ was reported to localize to the poles in the absence of Tgl (30); however, we also did not observe PilQ-sfGFP clusters in this strain background (Fig. S1B and C).
In the inverse experiment, we observed that Tgl and Tgl-sfGFP accumulated at the same level in the presence and absence of PilQ (Fig. 2A and 3A ) and that Tgl-sfGFP localized at the cell envelope (Fig. 4A).However, Tgl-sfGFP neither formed unipolar nor mid-cell clusters in the absence of PilQ (Fig. 4A).Because Tgl-sfGFP accumulates at native levels but does not form polar or mid-cell clusters in the absence of PilQ, these data support that PilQ recruits Tgl to mid-cell and the poles rather than the other way around.We note that the time-lapse fluorescence microscopy analyses showed that Tgl-sfGFP on average formed a visible cluster at mid-cell slightly earlier than PilQ-sfGFP (Fig. 2E vs Fig. 3E).Because cells expressing PilQ-sfGFP also accumulate a fraction of untagged PilQ monomer (Fig. 2A), we speculate that the slight delay (30 min) in the average timing of PilQ-sGFP recruitment to the nascent/new poles relative to the average recruitment of Tgl-sfGFP could originate from a preference of untagged PilQ for the constriction site.

PilQ recruitment to the nascent and new poles depends on the AMIN domains
Next, we addressed how PilQ is recruited to the nascent poles.In P. aeruginosa, localiza tion to the nascent poles of the inner membrane (IM) protein PilO (Fig. 1), and therefore the T4aPM, depends on the PG-binding AMIN domains of PilQ (25).The M. xanthus PilQ contains three AMIN domains, one of which is sufficient for the correct assembly and polar localization of the T4aPM (8).Therefore, we speculated that deleting all three AMIN domains would prevent PilQ from being recruited to the nascent and new poles.
To this end, we generated a strain expressing a PilQ variant lacking all three AMIN domains fused to sfGFP (PilQ ΔAMIN×3 -sfGFP) (Fig. 4B) by deleting the region encoding the AMIN domains in pilQ-sfGFP at the native site.In immunoblots with α-GFP antibodies, monomeric PilQ ΔAMIN×3 -sfGFP accumulated at a moderately lower level than the fulllength protein, although a significant fraction of the protein was also cleaved to generate free sfGFP, and no heat-resistant multimers were detected (Fig. 4C).Consistently, cells expressing PilQ ΔAMIN×3 -sfGFP were non-motile (Fig. S1A).Because PilQ ΔAMIN×3 -sfGFP was not detected by the PilQ antibodies (Fig. 4C), we suggest that the epitopes detected by these antibodies are within the AMIN domains.Consistent with the three AMIN domains being essential for polar PilQ recruitment, PilQ ΔAMIN×3 -sfGFP did not generate mid-cell and polar clusters (Fig. 4A).We conclude that the three AMIN domains are required for recruitment of PilQ to the poles and may play a role in multimer formation.
Next, to test whether the three AMIN domains are sufficient for polar recruitment, we generated a fusion in which the three AMIN domains were fused to sfGFP (PilQ AMIN×3 -sfGFP) (Fig. 4B) and replaced pilQ with pilQ AMIN×3 -sfGFP by double homologous recom bination.However, immunoblot analysis and fluorescence microscopy revealed that PilQ AMIN×3 -sfGFP did not accumulate, thus precluding further analyses (Fig. 4C; Fig. S1D).

Cell division-independent polar recruitment of PilQ
Generally, septal and polar PG contains fewer stem peptides, is considered metabolically mostly inert, and modifications acquired during cytokinesis are retained at the poles indefinitely (58,59).Moreover, it has been shown that the AMIN domain of the cell division protein AmiC in Escherichia coli binds to septal PG during cytokinesis (60)(61)(62).Based on these considerations, and because the three PilQ AMIN domains are required for polar recruitment of PilQ, we hypothesized that the old cell poles would have the properties needed to recruit and incorporate the PilQ secretin in the OM independently of a cell division event.To test this hypothesis, we expressed PilQ-sfGFP in a ΔpilQ mutant under the control of the vanillate-inducible promoter (P van ) and then followed its polar recruitment (Fig. 5A; Fig. S2A).Remarkably, we observed for 100% of cells that PilQ-sfGFP was recruited to both poles independently of cell division as well as to mid-cell when cells started constricting (Fig. 5A and B).As expected, neither polar nor mid-cell PilQ-sfGFP recruitment was observed in the absence of Tgl (Fig. 5A and B; Fig. S2A).Consistently, we also observed that upon induction of untagged PilQ synthesis in the ΔpilQ strain additionally expressing Tgl-sfGFP from the native site, Tgl-sfGFP transiently formed clusters at both poles in 93% of cells (Fig. 5C and D; Fig. S2B).Finally, to determine whether PilQ recruited to the poles independently of cell division was competent to guide the assembly of the remaining components of the T4aPM, we repeated the PilQ induction experiment in a strain in which pilM at the native site is replaced by mCherry-pilM by double homologous recombination (12).The cytoplasmic PilM protein (Fig. 1) is the last component to be incorporated into the polar T4aPM in M. xanthus (28).Before induction of PilQ synthesis, the active mCherry-PilM fusion (12) localized diffusely to the cytoplasm; importantly, upon induction of pilQ expression, mCherry-PilM also localized in a bipolar pattern in 93% of cells (Fig. 5C and D; Fig. S2C).
Based on these observations, we conclude that the polar recruitment and OM incorporation of PilQ can occur at both poles independently of cell division, and that these secretins support the assembly of the remaining components of the T4aPM.Because this incorporation depends on Tgl, we also conclude that the cell divisionindependent PilQ incorporation into the OM follows the same mechanism as in the case of its incorporation at nascent and new poles.

Tgl is not important for PilQ transport across the periplasm
To evaluate whether OM localization of Tgl is important for its function, we generated a strain expressing Tgl C20G -sfGFP (using the numbering of the full-length protein), in which the conserved Cys residue (+1 in the mature protein) (Fig. S3A) was substituted to Gly to prevent its acylation and, therefore, transport to and anchoring in the OM.Additionally, because an Asp in position +2 of mature lipoproteins in E. coli can cause their retention in the IM (63), we also generated a strain expressing Tgl S21D -sfGFP.Expression of Tgl C20G -sfGFP from the native site or under the control of P van as well as Tgl S21D -sfGFP under the control of P van only resulted in very low levels of accumulation of the proteins (Fig. S3B), thus precluding their further analyses.Therefore, to obtain more insights into the function of Tgl, we determined the subcellular localization of PilQ in cell fractionation experiments in the presence and absence of Tgl.In wild-type (WT) cells, the PilQ monomer and multimer were enriched in the membrane fraction (Fig. 6A).Similarly, in Δtgl cell extracts, monomeric PilQ was enriched in the membrane fraction.Control proteins previously shown to localize to the IM or OM were enriched in the membrane fraction and a cytoplasmic control protein was enriched in the soluble fraction documenting that the fractionation procedure worked properly (Fig. 6A).
To determine whether monomeric PilQ is integrated into the IM or OM, we isolated the OM of WT and Δtgl cells after osmotic shock with sucrose and EDTA treatment.Monomeric PilQ was detected in the OM fraction of both strains (Fig. 6B).As expected, in WT extracts, the heat-and detergent-resistant oligomers were also enriched in the OM fraction, while control proteins fractionated as expected (Fig. 6B).These results demonstrate that Tgl is not required for the transport of monomeric PilQ across the periplasm to the OM, and that monomeric PilQ is at the OM.

A computational structural model of the Tgl/PilQ complex
To evaluate how Tgl interacts with monomeric PilQ to promote its stability and multime rization in the OM, we analyzed the two proteins in silico.While the sequences of T4aPM pilotins are not well conserved (42) (Fig. S3A), it was previously suggested that they all share a similar superhelix structure composed of six TPR motifs (48, 66), which are typically involved in protein-protein interactions (67).In agreement with this suggestion, a high-confidence AlphaFold-based structure of monomeric Tgl includes 13 anti-parallel α-helices, among which helices 1-to 12-fold into six TPR motifs forming a superhelix (Fig. 7A; Fig. S3C).Additionally, the Tgl structural model could readily be superimposed on the solved structures of PilF and PilW (pilotins of the secretin of the T4aPM in P. aeruginosa and N. meningitidis, respectively (66, 68)) (Fig. 7B).While PilF does not contain disulfide bridges and PilW contains one, which is crucial for its function and connects TPR domains 3 and 4 (Fig. S3A) (66,69), Tgl is predicted to contain three disulfide bridges that link TPR domains 5 and 6 as well as TPR 6 and α-helix 13 (Fig. 7A; Fig. S3A).Conservation analysis of the amino acid sequence of Tgl homologs using ConSurf revealed two conserved hydrophobic surfaces, one in the N-terminal TPR1 (from hereon CS1) and one within the concave groove of Tgl close to CS1 (from hereon CS2) (Fig. 7C).
In secretins, the two or more conserved N-terminal N-domains are involved in oligomerization (70), interaction with IM components of the T4aPM (71), and also form part of the periplasmic vestibule, while the AMIN domains, if present, bind PG (25, 35, 37) (Fig. 7D).In the C-terminal region, the secretin domain (PF00263) forms β-sheets, which in the secretin oligomer form the gate, part of the periplasmic vestibule, and the amphipathic helical loop (AHL) (37, 40) (Fig. 7D).The hydrophobic surface of the AHL associates with the inner leaflet of the OM (Fig. 7D and E) (40).The amphipathic β-lip region in the C-terminal region forms part of the β-barrel (37), but mainly forms the βstranded region with which the secretin spans the OM (Fig. 7D and E) (40).
Similar to other T4aPM secretins, PilQ from M. xanthus (PilQ Mxa ) is divided into four main regions: the three AMIN domains connected by flexible linkers, the N0-and N3domains, the β-lip region, and the C-terminal secretin domain (Fig. 7F).Monomeric PilQ Mxa was modeled with high confidence using AlphaFold (Fig. 7F; Fig. S3C) and could readily be superimposed on a protomer from the cryo-electron microscopy-based structure of P. aeruginosa PilQ (PilQ Pae ) (38) (Fig. 7D; Fig. S4A).Similar results were obtained when using monomeric PilQ Pae modeled with AlphaFold (Fig. S3C and S4A), supporting that the predicted structures are modeled with high confidence.
It is currently not known how pilotins of T4aPM secretins interact with their cognate secretin monomer.Therefore, to gain insights into how T4aPM secretins and their pilotins interact, we started with the M. xanthus proteins and used AlphaFold-Multimer to predict heterodimeric structures of Tgl and monomeric PilQ Mxa .Surpris ingly, in the high-confidence heterodimer model, the amphipathic β-sheet of the β-lip observed in the structural model of the PilQ Mxa monomer (Fig. 7F; Fig. S3C) is unfolded into (i) a hydrophobic α-helix and (ii) an unstructured region (Fig. 7G).Remarkably, Tgl is modeled to specifically interact with this hydrophobic α-helix and this unstructured region via the conserved hydrophobic surfaces CS1 in TPR1 and CS2 in the concave groove, respectively (Fig. 7C and G).Underscoring the validity of this structural model of the heterodimer, PilF and PilW are also modeled with high confidence to associate with their partner secretin in the same way, i.e., using the same interfaces as in Tgl to contact the unfolded β-lip (Fig. S4B through D).Moreover, these specific interactions appear to depend on the cognate pilotin-secretin pair, because heterodimer modeling of Tgl with the P. aeruginosa or the N. meningitidis secretin yielded structural models of lower confidence, and in which some of these interactions were lost (Fig. S4E through G).
In conclusion, we suggest that T4aPM pilotins by associating with the unfolded βlip of their cognate monomeric secretin keep this region, part of which will ultimately be inserted into the OM, in a conformation optimal for oligomerization and OM insertion.Once the secretin monomers multimerize and the correctly folded β-lip integrates into the OM, the interaction with the pilotin would be lost, thus explaining why the pilotin only transiently associates with the secretin.

Survey of the presence and absence of AMIN domains in T4aPM secretins and piliation pattern
As shown here, septal and polar recruitment of the PilQ secretin, and therefore the T4aPM, in M. xanthus depends on its AMIN domains.Because AMIN domains are not universally conserved in T4aPM secretins (37), we wondered whether their presence or absence correlated with the localization pattern of the T4aP/T4aPM in other species.To this end, we selected bacteria with different T4aP localization patterns and studied the domain architecture of their secretin.

DISCUSSION
Here, we focused on the polar incorporation of the T4aPM in the rod-shaped cells of M. xanthus to understand how different localization patterns of T4aP are ultimately established.M. xanthus is an ideal system to address this question because the T4aPM assembly pathway is well-understood and initiates with the PilQ secretin in the OM (8,28).Thus, using PilQ as a proxy for the T4aPM allowed us to examine how the specific localization of the T4aPM is determined.We demonstrate that PilQ is recruited to and begins to assemble in the OM at the nascent pole during cytokinesis, and these processes continue for 60-90 min in the two daughter cells after completion of cytokinesis.The recruitment and assembly eventually result in the symmetric localiza tion of PilQ at the two cell poles.Consistent with the pilotin Tgl being important for PilQ multimer formation (28,30), we observed that Tgl transiently associated with the nascent and new cell poles in a PilQ-dependent manner and largely in parallel with PilQ recruitment and OM assembly.Moreover, we demonstrate that PilQ recruitment to and assembly at the nascent and new poles depends on its PG-binding AMIN domains.
How, then, does PilQ assemble at the nascent and new poles?Several lines of evidence support that this assembly is a two-step process that crucially depends on the PG-binding AMIN domains in PilQ in concert with Tgl.First, a PilQ variant lacking all three AMIN domains accumulated but was not recruited to the nascent and new poles and did not assemble to form multimers, while a PilQ variant with only one AMIN domain is sufficient for correct assembly and polar localization of the T4aPM (8).In agreement with this observation, recruitment of the T4aPM in P. aeruginosa to the nascent poles depends on the PilQ AMIN domains (25).AMIN domains also have crucial functions in recruiting the amidases AmiB and AmiC to the site of cell division in E. coli (60,72), and the AmiC AMIN domain specifically binds to septal PG during cytokinesis (60)(61)(62).Second, previous work concluded that PG modifications acquired during cytokinesis are retained at the poles indefinitely (59).Third, the pilotin Tgl, which is an OM lipoprotein, is important for PilQ stability and multimerization in the OM.Fourth, in cells lacking PilQ, Tgl still accumulated; however, it did not localize to the nascent and new poles.Fifth, the PilQ multimerization defect in Δtgl cells can be extracellularly complemented by the transfer of Tgl from Tgl + cells (48, [51][52][53].Based on these lines of evidence, we suggest a model for the polar recruitment and OM incorporation of PilQ and, consequently, the complete T4aPM.In this model, PilQ monomers and Tgl are translocated to the OM independently of each other (Fig. 9, step 1).During and immediately after cytokinesis, PilQ monomers specifically recognize and stably bind to septal and polar PG via their AMIN domains (Fig. 9, step 2).These PilQ monomers either bring along Tgl or recruit Tgl to the poles (Fig. 9, step 2).The high local concentration of PilQ/Tgl complexes at the nascent and new poles, facilitates Tgl-dependent OM incorporation and multimerization of PilQ, eventually resulting in the release of Tgl (Fig. 9, steps 3 and 4).Upon assembly of the PilQ multimer in the OM, the remaining components of the T4aPM are incorporated (Fig. 9, step 5).In agreement with our model, PilQ can be recruited to both poles and assemble multimers in a cell-division independent manner, supporting that both cell poles have the properties required for recruitment and incorporation of the PilQ secretin in the OM independently of a cell division event.We note that not all PilQ is localized to the cell poles.We, therefore, suggest that PilQ monomer recruitment to the nascent and new poles represents an example of a diffusion-and-capture mechanism for polar protein localization (73,74).In this mechanism, OM-associated PilQ monomers and Tgl diffuse in two dimensions until the PilQ monomers recognize and bind septal and polar PG that serve as a polar landmark, thereby enabling polar Tgl localization (Fig. 9, steps 1 and 2).We speculate that the number of assembled PilQ multimers in the OM at the cell poles, and therefore the number of T4aPM at the cell poles, is limited by the availability of the specific septal and polar PG recognized by the PilQ AMIN domains.In the future, it will be important to determine the exact PG recognized by the PilQ AMIN domains.Similarly, it will be important to determine whether monomeric PilQ and Tgl are recruited as a complex or sequentially.The assembled T4aPM in M. xanthus has a width of 15-20 nm (8, 12), while the average pore size of PG has been estimated to ~2 nm (75).While the incorporation of T4aPM in parallel with cytokinesis is compatible with these different dimensions, it is more difficult to understand how the T4aPM would be incorporated after completion of cytokinesis and independently of cell division.Thus, it will also be important to address how the T4aPM is assembled post-divisionally.
How then does Tgl stimulate PilQ multimer formation in the OM? Tgl is important for PilQ stability and multimerization in the OM.In cellular fraction experiments, we found that the PilQ monomer is associated with the OM in a Tgl-independent manner demonstrating that Tgl is not required for translocating monomeric PilQ from the IM across the periplasm to the OM.We note that whether PilQ is associated with the OM or integrated into the OM cannot be distinguished based on these experiments.High confidence in silico structural models of monomeric Tgl, monomeric PilQ, and heterodimeric Tgl/PilQ complexes, support that Tgl interacts with monomeric PilQ via hydrophobic interfaces.Specifically, our structural models suggest that two conserved hydrophobic surfaces, i.e., CS1 in TPR1 and CS2 in the concave groove, in Tgl interact with the hydrophobic parts of the unfolded amphipathic β-lip of monomeric PilQ.Therefore, our results suggest that Tgl at the OM binds the OM-associated PilQ monomer thereby (i) stimulating multimerization by maintaining an oligomerization-ready conformation of the PilQ monomer, (ii) protecting monomeric PilQ from proteolytic degradation, and (iii) ensuring that the assembled secretin only forms at the OM.Because Tgl is associated with the OM via its acylated N-terminus, CS1 and CS2 are close to the OM and, therefore, ideally positioned to assist in PilQ secretin integration into the OM.Once PilQ monomers multimerize and integrate into the OM, the interaction with Tgl would be lost because the β-lip is integrated into the OM and the interaction surfaces no longer available for interaction with Tgl (Fig. 9, steps 3 and 4).In silico structural models of PilF/PilQ and PilW/PilQ heterodimers support that they form complexes similar to that of Tgl/PilQ.These models of the heterodimers are also supported by the observations that PilF interacts with the C-terminal region of monomeric PilQ Pae and that loss-of-function PilF variants have substitutions in TPR1 (76).We, therefore, propose that T4aPM pilotins and their cognate monomeric secretin use the same conserved mechanism in which the pilotin interacts with the unfolded β-lip of the monomeric secretin to aid its OM integration.Lending further support for this generalized mechanism of the cognate T4aPM pilotin/secretin pairs, cognate pilotin/secretin pairs of the T2SS and T3SS interact via the so-called S-domain at the C-terminus of the secretin monomer (35,36).However, T4aPM pilotins are structurally different from T2SS and T3SS pilotins (77), and T4aPM secretins lack the S-domain (35,36).
Interestingly, AMIN domains are not universally conserved in T4aPM secretins (37).A 10 species survey generally supports that the presence and absence of AMIN domain(s) in a T4aPM secretin overall correlates with a polar and peritrichous piliation pattern, respectively.However, three species did not follow this overall correlation.Specifically, the coccoid N. meningitidis and N. gonorrhoeae assemble peritrichous T4aP and both have secretins that contain two AMIN domains.Notably, it has been suggested that these two species have a rod-shaped ancestor and that the emergence of their coccoid cell shape derives from relatively recent gene losses from the genome of this ancestor (17,78).Thus, we speculate that the AMIN domains in the N. meningitidis and N. gonorrhoeae T4aPM secretins are remnants from the rod-shaped ancestor.Interestingly, the rod-sha ped cells of N. elongata have polarly localized T4aP (17) and, similar to N. meningitidis and N. gonorrhoeae, a secretin with two AMIN domains.However, an N. elongata mutant with a coccoid cell shape assembles peritrichous T4aP (17), as do N. meningitidis and N. gonorrhoeae, thus supporting the hypothesis that the AMIN domains in the PilQ secretins of N. meningitidis and N. gonorrhoeae could be remnants from a rod-shaped ancestor.The third species in which the piliation pattern and absence/presence of AMIN domain(s) correlation did not match was A. baylyi.This species has a unique lateral piliation pattern, and its T4aPM secretin contains two AMIN domains.Interestingly, the lateral localization of the T4aPM, but not the assembly of the T4aPM, depends on the FimV protein (6).In the future, it will be interesting to test whether the AMIN domains of the A. baylyi secretin are also important for the localization of the T4aPM and whether A. baylyi may potentially accumulate a specific form of PG laterally that is recognized by its T4aPM secretin.Thus, it remains possible that two pathways converge to determine the lateral localization of the T4aPM in A. baylyi, i.e., one that depends on FimV and one that depends on the AMIN domains in the secretin.Along the same lines, FimV in P. aeruginosa, in addition to being important for the accumulation of the PilQ secretin, is important, but not essential, for polar localization of PilQ (25,79).By contrast, the PilQ AMIN domains are essential for the polar localization of the T4aPM in P. aeruginosa (25).Again, these observations suggest a scenario in which two pathways may converge to ensure polar localization of PilQ, and, therefore, the T4aPM, and in which the PilQ AMIN domains are essential.Importantly, not all bacteria with T4aPM possess a FimV homolog, e.g., M. xanthus is not predicted to encode a FimV homolog.Interestingly, the secretins of T2SS and T3SS lack AMIN domains, and these secretion systems have been reported to have a dispersed localization (80)(81)(82).Thus, the overall correlation between the presence/absence of AMIN domains in the relevant secretin and the localization of the relevant macromolecular structure may extend beyond the T4aPM.

Plasmid construction
All oligonucleotides used are listed in Table S1.All constructed plasmids were verified by DNA sequencing.
For pLC220 (plasmid for replacement of tgl with tgl-sfGFP in the native site): the tgl-sfGFP fragment was amplified from pSC104 (28) using primers tgl_fw_hin diii/sfgfp_rv_xbaI.The downstream fragment was amplified from genomic DNA from M. xanthus DK1622 using the primer pair tgl_ds_fw_xbaI/ tgl_ds_rv.To generate the full-length insert, both DNA fragments were digested with XbaI and ligated.Next, the insert was digested with HindIII and EcoRI, and cloned into pBJ114.For pMH111 (plasmid for replacement of tgl with tgl C20G at the native site): upand downstream fragments were amplified using genomic DNA from M. xanthus DK1622 as DNA template and the primer pairs tgl_CtoG_A_HindIII/tgl_CtoG_Bov and tgl_CtoG_Cov/ tgl_CtoG_D_BamHI, respectively.To generate the full-length insert, an overlapping PCR using the two fragments as DNA templates and the primer pair tgl_CtoG_A_HindIII/ tgl_CtoG_D_BamHI was performed.Subsequently, the fragment was digested with HindIII and BamHI, and cloned into pBJ114.
For pMH119 (plasmid for expression of tgl-sfGFP from the 18-19 site under the control of the vanillate promoter): tgl-sfGFP was amplified using genomic DNA from M. xanthus SA12016 (tgl::tgl-sfGFP) as DNA template and the primer pair Pvan_tgl_fw_NdeI/ sfGFP_rv_tgl_EcoRI.The fragment was digested with NdeI and EcoRI, and cloned into pMR3690.
For pMH120 (plasmid for expression of tgl C20G -sfGFP from the 18-19 site under the control of the vanillate promoter): tgl C20G -sfGFP was amplified using genomic DNA from M. xanthus SA12035 (tgl C20G ::tgl-sfGFP) as DNA template and the primer pair Pvan_tgl_fw_NdeI/sfGFP_rv_tgl_EcoRI.The fragment was digested with NdeI and EcoRI, and cloned into pMR3690.
For pMH121 (for generation of an in-frame deletion of the AMIN × 3 domains of native pilQ): up-and downstream fragments were amplified from genomic DNA from M. xanthus DK1622 using the primer pairs PilQ_dAMIN_A_XbaI/PilQ_dAMIN_B and PilQ_dAMIN_C/pilQ_dAMIN_D_HindIII, respectively.Subsequently, the up-and downstream fragments were used as a template for an overlapping PCR with the primer pair PilQ_dAMIN_A_XbaI/pilQ_dAMIN_D_HindIII, to generate the AD fragment.The AD fragment was digested with XbaI and HindIII, and cloned in pBJ114.
For pMH122 (for generation of an in-frame deletion of tgl): up-and downstream fragments were amplified from genomic DNA of SA6053 (Δtgl) (28) using the primer pair tgl-A_XbaI/tgl-D_EcoRI to generate the AD fragment as described in reference (83).The AD fragment was digested with XbaI/EcoRI and cloned in pBJ114.For pMH125 (for replacement of pilQ with pilQ AMINs×3 (1-475) -sfGFP in the native site of the pilQ::pilQ-sfGFP strain): up-and downstream fragments were amplified from pMH118 using the primer pairs PilQAMIN_A_KpnI/PilQAMIN_sfGFP_overlay_rev and PilQamin_sfGFP_overlay_fwd/sfGFP_rev_pilQ_EcoRI, respectively.Subsequently, the upand downstream fragments were used as a template for an overlapping PCR with the primer pair PilQAMIN_A_KpnI/sfGFP_rev_pilQ_EcoRI, to generate the AD fragment.The AD fragment was digested with KpnI and EcoRI, and cloned in pBJ114.
For pMH127 (plasmid for expression of tgl S21D -sfGFP from the 18-19 site under the control of the vanillate promoter): tgl S21D -sfGFP was amplified using pMH119 as DNA template and the primer pairs Pvan forw/Tgl_S21G_overlay_rev and Tgl_S21G_over lay_fwd/sfGFP_rv_tgl_EcoRI to introduce the point mutation.Subsequently, both PCR fragments were used as a template for an overlapping PCR with primer pair Pvan forw/sfGFP_rv_tgl_EcoRI, to generate the full-length fragment.The fragment was digested with NdeI and EcoRI, and cloned into pMR3690.
For pMP183 (plasmid for expression of pilQ from the 18-19 site under the control of the vanillate promoter): pilQ was amplified using pMH118 as DNA template and the primer pair Pvan_PilQ_fwd_NdeI/PilQ_rev_EcoRI.The fragment was digested with NdeI and EcoRI, and cloned into pMR3690.

Epifluorescence microscopy
Cells were visualized following a slightly modified protocol (90).Briefly, exponentially growing cells were placed on a glass coverslip attached to a metal frame.Cells were covered with a thick 1% agarose pad supplemented with 0.2% (wt/vol) Bacto Casitone and TPM (10 mM Tris-HCl pH 8.0, 1 mM K 2 HPO 4 /KH 2 PO 4 pH 7.6, and 8 mM MgSO4), and supplemented with vanillate or cephalexin as indicated.For long time-lapse microscopy, the pad was additionally sealed with parafilm to reduce evaporation.Additionally, to prevent motility during time-lapse microscopy, strains contained an in-frame deletion of gltB (ΔgltB), which encodes a component of the M. xanthus gliding motility machine (54).Cells were imaged using a DMi8 inverted microscope and a Hamamatsu ORCA-Flash4.0V2 Digital CMOS C11440 or a DFC9000 GT (Leica) camera.Images were analyzed using Metamorph v 7.5 (Molecular Devices) and ImageJ (91).Image segmentation was done using Omnipose (92) and cell outlines were transformed to Oufti-compatible meshes using Matlab R2020a (The MathWorks).Segmentation was manually curated using Oufti (93).For signal detection and background correction, a previously published Matlab script was used (94).Because the Tgl-sfGFP fluorescent clusters have low fluorescence intensity, the script was modified to detect the strongest pixel intensity in each cell segment assigned by Oufti.Specifically, each pixel intensity in each segment was normalized to the maximum pixel intensity within the cell.Next, to identify cells with one or more fluorescent cluster(s), cells were only considered to have a cluster if less than 10% of the selected pixel intensities had a normalized fluorescence above 0.75.Hence identifying cells with an intense and narrow fluorescent peak.

Fractionation of M. xanthus cells
To fractionate M. xanthus cells into fractions enriched for soluble or membrane proteins, 20 mL of an exponentially growing M. xanthus suspension culture was harvested by centrifugation (8,000 × g, 10 min, RT) and concentrated to an optical density at 550 nm (OD 550 ) of 28 in resuspension buffer [50 mM Tris-HCl pH 7.6, 250 mM NaCl supplemented with Complete EDTA-free protease inhibitor (Roche)].Cells were lysed by sonication with 5 × 30 pulses, pulse 60%, amplitude 60% with a UP200St sonifier and microtip (Hielscher), and the lysate was cleared by centrifugation (12,000 × g, 5 min, RT).As a sample for total cellular protein, an aliquot of the cleared lysate was taken and mixed with 4 × SDS lysis buffer (200 mM Tris-HCl pH 6.8, 8% SDS, 40% glycerol, 400 mM DTT, 6 mM EDTA, and 0.4% bromphenol blue).A 200 µL aliquot of the remaining supernatant was subjected to ultracentrifugation using an Air-Fuge (Beckman) (100,000 × g, 20 min, RT).The resulting supernatant is enriched in soluble proteins and a sample was taken and mixed with 4 × SDS lysis buffer.The pellet was washed by resuspension in 200 µL resuspension buffer and was subjected to ultracentrifugation as above.The remaining pellet, which is enriched in IM and OM membrane proteins, was resuspended in 100 µL 1 × SDS lysis buffer.All samples were heated for 10 min at 95°C, separated by SDS-PAGE and analyzed by immunoblot.

OM protein enrichment
As a sample for total cellular protein (total fraction), 2 mL of an exponentially growing M. xanthus cell suspension was harvested by centrifugation (8,000 × g, 3 min, RT) and concentrated to an OD 550 of 7 in 1 × SDS lysis buffer.To isolate a fraction enriched for OM proteins, 50 mL of the cell suspension was harvested (4,700 × g, 25 min, 4°C), and the pellet was gently resuspended in TSE8-buffer [100 mM Tris-HCl pH 8, 1 mM EDTA, 20% (wt/vol) sucrose, protease inhibitor cocktail (Roche)] to a concentration corresponding to OD 550 = 50.The sample was incubated for 30 min at 4°C to release the OM, followed by centrifugation of the samples (16,000 × g, 30 min, 4°C).The resulting supernatant is enriched in OM and periplasmic proteins and was recovered for the following steps, while the pellet, containing cells without OM or where the OM had not been released, was frozen at −20°C.Next, 150 µL of the supernatant was ultra-centrifuged using an Air-Fuge (Beckman) (~133,000 × g, 1 h, RT) to separate the OM from periplasmic proteins.The resulting supernatant was discarded and the OM-enriched pellet (OM fraction) was resuspended in 150 µL 1 × SDS lysis buffer.The frozen pellet was thawed, resuspended to OD 550 = 50 in resuspension buffer (50 mM Tris pH 7.6, 10 mM MglCl 2 ) and lysed by sonication.Cell debris was removed by centrifugation (16,000 × g, 15 min, 4°C).The cell-free supernatant (~150 µL) was subjected to ultra-centrifugation as described above.The resulting supernatant contained cytoplasmic proteins and was mixed with 4 × SDS lysis buffer (cytoplasmic fraction).All samples were boiled for 10 min at 95°C, separated by SDS-PAGE, and analyzed by immunoblot.

Bioinformatics
Full-length protein sequences or sequences in which the signal peptide was identified with SignalP 6.0 (97) and removed, were used for AlphaFold and AlphaFold-Multimer modeling via ColabFold (1.3.0)(98-100).The predicted Local Distance Difference Test (pLDDT) and predicted Alignment Error (pAE) graphs of the five models generated were made using a custom Matlab R2020a (The MathWorks) script (101).Ranking of the models was performed based on combined pLDDT and pAE values, with the best-ranked models used for further analysis and presentation.Per residue model, accuracy was estimated based on pLDDT values (>90, high accuracy; 70-90, generally good accuracy; 50-70, low accuracy; and <50, should not be interpreted) (98).Relative domain positions were validated by pAE.The pAE graphs indicate the expected position error at resi due 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, conse quently, the relative position of domains/subunits/proteins (98).PyMOL version 2.4.1 (http://www.pymol.org/pymol)was used to analyze and visualize the models.Structural alignments were performed using the PyMOL Alignment plugin with default settings.Hydrophobicity was calculated in PyMOL according to the hydrophobicity scale (102).Conservation of Tgl residues was assessed using the ConSurf server with default settings (103).Protein domains were identified using the Interpro server (104) and the predicted AlphaFold structures.The alignment of Tgl, PilF, and PilW was generated using Muscle5 (5.1) (105).

FIG 1
FIG 1 Architectural model of the T4aPM.Architectural model of the cell envelope-spanning non-piliated (A) and (B) piliated T4aPM (B) of M. xanthus with the 15 core proteins (8).The T4aPM is divided into five

FIG 2
FIG 2 Accumulation of PilQ variants and localization of PilQ-sfGFP.(A) Immunoblot detection of PilQ/PilQ-sfGFP.Protein from the same number of cells from exponentially growing suspension cultures was loaded per lane.Blot was probed with the indicated antibodies.The blot was stripped before applying a new antibody.LonD served as a loading control.Monomeric and oligomeric forms of PilQ/PilQ-sfGFP are marked with an asterisk.Calculated molecular weights of proteins without signal peptide (if relevant) are indicated.(B) Localization of PilQ-sfGFP in the presence and absence of cephalexin.Left panels, representative epifluorescence images of cells expressing PilQ-sfGFP.The percentage of cells with a unipolar (blue), bipolar (orange) cluster localization pattern, or no cluster (gray) is indicated.Blue, orange, and white arrowheads indicate unipolar, bipolar, and mid-cell clusters.Percentage of cells with a mid-cell cluster is indicated in white.Scale bar, 5 µm.Right panels, normalized fluorescence profiles of cells, for which a cluster was detected, as a function of the relative cell length.Mean (black line) and standard deviation (SD) (orange) of the relative fluorescence along the normalized cell are depicted.Cell length was normalized from 0 to 1, where 0 is assigned to the pole with the highest fluorescent value.(C) Localization of PilQ-sfGFP in short (<6.5 µm) and long (≥6.5 µm) cells.Same cells analyzed as in untreated cells in (B).Signals are shown as in (B), right panel.(D) Time-lapse microscopy of a PilQ-sfGFP expressing cell treated with cephalexin.Epifluorescence and phase-contrast (Continued on next page)

FIG 2 (
FIG 2 (Continued) images are shown.Arrow indicates first time point at which the PilQ-sfGFP cluster is clearly visible at mid-cell.Time indicates time point after the addition of cephalexin (t = 0).(E) Polar recruitment of PilQ-sfGFP during the cell cycle.Left panels, epifluorescence and phase-contrast images from time-lapse microscopy of cells expressing PilQ-sfGFP.PilQ-sfGFP clusters are visible at the nascent poles during (upper panel) or after completion (lower panel) of cytokinesis.Arrowheads indicate first time point at which the PilQ-sfGFP cluster is clearly visible.Right panel, analysis of appearance of the PilQ-sfGFP cluster relative to completion of cytokinesis for each daughter cell.The first time point after completion of cytokinesis is defined as t = 0 and indicated by the gray vertical bar.The black line and error bars represent the mean ± SD.The appearance of PilQ-sfGFP clusters was studied in 196 daughter cells.(F) Analysis of polar incorporation of PilQ-sfGFP after completion of cytokinesis.The cellular fluorescence was quantified at different time points after completion of cytokinesis and the relative fluorescence along a normalized cell was plotted.Mean (black line) and SD (orange) are indicated.Cell length was normalized from 0 to 1, where 0 was assigned to the old pole.n = 25.In (D-F), to follow cells on hard agar by time-lapse microscopy for extended periods of time and avoid that they move out of the field of view, all strains contain an in-frame deletion of gltB (ΔgltB), which encodes a component of the M. xanthus gliding motility machine (54).

FIG 3
FIG 3 Accumulation of Tgl variants and localization of Tgl-sfGFP.(A) Immunoblot detection of Tgl/Tgl-sfGFP.Protein from the same number of cells from exponentially growing suspension cultures was loaded per lane.Blot was probed with the indicated antibodies.The blot was stripped before applying a new antibody.LonD served as a loading control.Monomeric and oligomeric forms of PilQ are marked with an asterisk.Calculated molecular weights of proteins without signal peptide (if relevant) are indicated.(B) Localization of Tgl-sfGFP in the presence and absence of cephalexin as in Fig. 2B.Scale bar, 5 µm.(C) Localization of Tgl-sfGFP in short (<6.5 µm) and long (≥6.5 µm) cells.Same cells analyzed as in untreated cells in panel (B).Signals are shown as in Fig. 2B, right panel.(D) Time-lapse microscopy of cells expressing Tgl-sfGFP treated with cephalexin.Left panel, epifluorescence and phase-contrast images are shown.Arrows indicate time points at which the Tgl-sfGFP cluster is clearly visible.Time indicates time point after the addition of cephalexin (t = 0).Right diagram shows lifetime of Tgl-sfGFP clusters.The black line and error bars represent the mean ± SD. n = 54.(E) Recruitment of Tgl-sfGFP to the nascent and new poles during the cell cycle.Left panel, epifluorescence and phase-contrast images from time-lapse microscopy of a cell expressing Tgl-sfGFP.Tgl-sfGFP clusters generally appear at the nascent poles during cytokinesis.Arrows indicate time points, at which the Tgl-sfGFP cluster is clearly visible.The boxed areas are shown below in a higher magnification.Right (Continued on next page)

FIG 3 (FIG 4
FIG3 (Continued)    panels, analysis of appearance of a Tgl-sfGFP cluster relative to completion of cytokinesis for each daughter cell and lifetime of the Tgl-sfGFP cluster relative to completion of cytokinesis for each daughter cell.The first time point after completion of cytokinesis is defined as t = 0 and indicated by the gray vertical bar.The black line and error bars represent the mean ± SD.The appearance of Tgl-sfGFP clusters was studied in 190 daugther cells.In (D and E), strains analyzed contain the ΔgltB mutation.

FIG 5
FIG 5 Cell division-independent assembly of T4aP machines.(A, C) Induction of expression of pilQ-sfGFP (A) or pilQ (C) from P van in the indicated strains followed by time-lapse epifluorescence microscopy.Time indicates the interval after the addition of vanillate (t = 0).In (A), upper panel, the arrows indicate the first appearance of fluorescent clusters of PilQ-sfGFP.In (A), lower panel, cells are shown after 24 h of PilQ-sfGFP induction in the Δtgl mutant.In (A), 10 µM vanillate was used for inducing PilQ-sfGFP accumulation at WT levels in the ΔpilQ background (Fig. S2A), and 500 µM vanillate was used to highly induce PilQ-sfGFP accumulation in the Δtgl ΔpilQ background (Fig. S2A).In (C), upper panel, 20 µM vanillate was used for inducing PilQ accumulation at WT levels in ΔpilQ cells expressing tgl-sfGFP (Fig. S2B).Arrows indicate time points at which the Tgl-sfGFP cluster is clearly visible.In (C), lower panel, 1 mM vanillate was used to rapidly induce pilQ expression in cells co-expressing mCherry-pilM (Fig. S2C).Arrows indicate the first appearance of fluorescent clusters of mCherry-PilM.Localization of mCherry-PilM at the poles is used as a proxy to study correct assembly of the T4aPM.In (A and C), strains analyzed contain the ΔgltB mutation.(B and D) Percentage of cells in which PilQ-sfGFP (B), Tgl-sfGFP or mCherry-PilM (D) clusters appeared at the old poles of cells in (A and C) upon induction of PilQ-sfGFP/PilQ synthesis; n = 30 cells for all strains.

FIG 7
FIG 7 Structural characterization of Tgl alone and in complex with its PilQ secretin partner.(A) AlphaFold model of mature Tgl.Upper panel, TPR domains 1-6 in Tgl are indicated to scale.Lower panel, AlphaFold structure of Tgl.Cys residues (Cys152/ Cys182, Cys172/Cys195, and Cys212/Cys227, numbering for unprocessed protein) engaged in disulfide bridge formation are indicated in red.Amino acids are indicated using the numbering of the full-length protein.(B) Superposition of the AlphaFold Tgl structure with the solved structures of PilF (PDB: 2HO1) (42) and PilW (PDB: 2VQ2) (66).(C) Surface representation of the sequence conservation calculated with ConSurf using 279 homolog sequences (upper panel) and hydrophobicity of Tgl (lower panel).The conserved hydrophobic surfaces CS1 in TPR1 and CS2 in the concave groove are marked with dashed squares.(D) Cryo-EM structure of the tetradecameric PilQ secretin of P. aeruginosa (PDB: 6VE2) (38).Upper panel, domain architecture of PilQ Pae .Lower panel, Cryo-EM structure, in which the different domains of one PilQ Pae protomer are indicated as in reference (37) are colored as N3 (marine blue), secretin (green), and β-lip region (yellow).The N0 and two AMIN domains, which are not resolved in the structure, are represented by a cyan cylinder and magenta boxes, respectively.In the part of the secretin marked by the dashed box, the front part of the barrel structure has been removed to show the inside of the barrel with the gate.(E) Surface representation of the hydrophobicity of the cryo-EM solved structure of the tetradecameric PilQ Pae (Continued on next page)

FIG 7 (
FIG 7 (Continued) secretin.Note the amphipathic AHL and β-lip.In the part of the secretin marked by the dashed box, the front part of the barrel structure has been removed to show the inside of the barrel with the gate.Regions in which the protein structure was sliced are colored black.(F) AlphaFold model of M. xanthus PilQ monomer.Upper panel, domain architecture of PilQ.Lower panel, AlphaFold model with regions colored according to the domains and as described in panel (D).For clarity, AMIN domains are not shown.(G) AlphaFold model of heterodimer of M. xanthus PilQ monomer and Tgl.Inset: Interaction interface between Tgl (surface representation) and PilQ (cartoon) colored according to hydrophobicity.The conserved hydrophobic surfaces CS1 in TPR1 and CS2 in the concave groove are marked with a dashed square.In (A, B, and G), the acylated N-terminal Cys residue of mature Tgl (residue Cys20 in the unprocessed protein) that places the protein at the inner leaflet of the OM is indicated by a green circle.

FIG 9
FIG 9 Model of polar incorporation and OM assembly of the PilQ secretin as well as polar assembly of the T4aPM in M. xanthus.See main text for details.Note that in step 1, PilQ monomers and Tgl away from the septal and polar PG are shown not to interact; however, it is possible that the two proteins interact prior to their polar localization.In steps 1 and 4, the arrows indicate that the proteins can diffuse in the OM.In steps 1-3, PilQ is shown to associate with the OM via its AHL domain; however, it is not known how PilQ monomers associate with/integrate into the OM.In step 3, while the secretin oligomerizes from 12 to 15 PilQ monomers, only two are shown for illustration purposes.In step 5, all T4aPM components except for PilQ are colored as in Fig. 1.

TABLE 1 M
. xanthus strains used in this work a The allele encoding the fusion construct was integrated by double homologous recombination to replace the WT allele.

TABLE 2
Plasmids used in this work