Chimeric systems composed of swapped Tra subunits between distantly-related F plasmids reveal striking plasticity among type IV secretion machines

Bacterial type IV secretion systems (T4SSs) are a versatile family of macromolecular translocators, collectively able to recruit diverse DNA and protein substrates and deliver them to a wide range of cell types. Presently, there is little understanding of how T4SSs recognize substrate repertoires and form productive contacts with specific target cells. Although T4SSs are composed of a number of conserved subunits and adopt certain conserved structural features, they also display considerable compositional and structural diversity. Here, we explored the structural bases underlying the functional versatility of T4SSs through systematic deletion and subunit swapping between two conjugation systems encoded by the distantly-related IncF plasmids, pED208 and F. We identified several regions of intrinsic flexibility among the encoded T4SSs, as evidenced by partial or complete functionality of chimeric machines. Swapping of VirD4-like TraD type IV coupling proteins (T4CPs) yielded functional chimeras, indicative of relaxed specificity at the substrate - TraD and TraD - T4SS interfaces. Through mutational analyses, we further delineated domains of the TraD T4CPs contributing to recruitment of cognate vs heterologous DNA substrates. Remarkably, swaps of components comprising the outer membrane core complexes, a few F-specific subunits, or the TraA pilins supported DNA transfer in the absence of detectable pilus production. Among sequenced enterobacterial species in the NCBI database, we identified many strains that harbor two or more F-like plasmids and many F plasmids lacking one or more T4SS components required for self-transfer. We confirmed that host cells carrying co-resident, non-selftransmissible variants of pED208 and F elaborate chimeric T4SSs, as evidenced by transmission of both plasmids. We propose that T4SS plasticity enables the facile assembly of functional chimeras, and this intrinsic flexibility at the structural level can account for functional diversification of this superfamily over evolutionary time and, on a more immediate time-scale, to proliferation of transfer-defective MGEs in nature.

These in vitro structures have supplied important information about the architectures of several model T4SSs in their quiescent states.The structures, coupled with other biochemical and genetic findings, also predict regions of conformational flexibility that might be important for structural transitions associated with machine activation.These inherently flexible regions include the T4CP -channel interface, as suggested by difficulties in isolation of stable T4CP -channel complexes and recent evidence for spatial repositioning of VirD4-like TrwB to polar regions of donor cells where T4SS R388 channels are located upon contact with recipient cells [20].In the T4SS R388 structure, the IMC presents as an intrinsically stable substructure with well-defined contacts among its VirB3-, VirB4-, VirB5-, VirB6-, and VirB8-like components [5].However, there is also evidence for different conformations of T4SS IMCs and of VirB4 ATPases, possibly reflecting different functional states [5,21,22].The T4SS R388 possesses a central periplasmic stalk composed of pentamers of VirB6 with VirB5 with well-defined contacts [5], but the stalk lacks a central channel which presumably must form for substrate transfer or pilus production.Finally, the VirB7-, VirB9-, and VirB10-like components of the OMCC R388 assemble as well-defined inner and outer (I and O) layers, but these substructures are only sparsely connected to each other and adopt distinct rotational symmetries.Such features are also prominent among OMCCs of other T4SSs solved to date, leading to proposals that the OMCCs are highly structurally dynamic [2,[12][13][14][15].
Recently, we reported the structure of the OMCC encoded by the IncFV plasmid pED208 (hereafter OMCC ED ) [15].The OMCC ED has a central cone (CC; equivalent to the I-layer of T4SS R388 ) with 17-fold symmetry and an outer ring complex (ORC, equivalent to the R388 O-layer) with 13-fold symmetry.The CC is composed of 17 copies of the C-terminal -barrel domain of VirB10-like TraB and between 13 and 17 copies of the N-terminal domain (NTD) of VirB7-like TraV.The ORC is composed of 26 copies each of VirB9-like TraK and C-terminal domains (CTDs) of TraV.The CC and ORC are connected only by domains linking the TraV NTDs and CTDs, and short stretches of TraB below the -barrel domains that interact with TraK.In view of these sparse connections, we proposed that the CC and ORC substructures are capable of independent movement.Remarkably, results of structure-guided mutational analyses established, first, that deletion of the entire TraV linker domain is completely dispensable for function of the pED208-encoded T4SS (T4SS ED ).Second, deletion of TraK from the T4SS ED does not abolish substrate transfer, but does block production of the pED208-encoded (ED208) pilus [15].Overall, these findings suggest that the T4SS ED accommodates major structural changes without effects on substrate transfer, although some mutations can selectively block piliation.
In this study, we further interrogated the plasticity of F-encoded T4SSs by assessing the functionality of chimeric machines.The chimeras were generated by swapping individual Tra/Trb subunits between T4SSs encoded by the phylogenetically distantly-related F-like plasmids, classical F and pED208 [23][24][25].F is designated as IncF1B/MOB F12 group A and pED208 and its progenitor plasmid F 0 -lac are IncFV/MOB F12 group C plasmids [26].
In line with their distinct Inc/MOB classifications, the Tra/Trb homologs exhibit considerable sequence divergence (Fig. S1A).F and ED208 pili also differ serologically and render host cells susceptible to different male-specific bacteriophages [27,28].We systematically deleted each of the tra/trb genes required for high-frequency plasmid transfer from F and pED208, and then tested whether subunit swaps between these systems yielded functional chimeric T4SSs.We identified chimeric machines with swaps or mutations of specific machine components that conferred altered DNA substrate selection, or the capacity to build functional DNA transfer channels in the absence of detectable pilus production.We also determined that host cells with co-resident pED208 and F plasmids deleted of essential tra/trb genes assemble functional chimeric machines.We propose that the intrinsic capacity of F systems, and possibly other T4SSs, to acquire heterologous machine components not only expands and diversifies machine functions, but also serves as a mechanism for proliferation of non-selftransmissible MGEs in nature.

RESULTS AND DISCUSSION
Diversification of the substrate repertoire through plasticity at the T4CP -T4SS interface.
TraD swapping supports transfer of noncognate substrates.To begin interrogating the crossfunctionality of the pED208 and F systems, we first asked whether the DNA transfer and replication (Dtr) proteins involved in processing these plasmids for transfer recognize the noncognate substrates.Conjugative DNA transfer initiates by assembly of Dtr proteins at the MGE's origin-of-transfer (oriT) sequence, forming the relaxosome.The catalytic subunit of the relaxosome, the relaxase, nicks the DNA strand destined for transfer (T-strand) within the oriT sequence and remains covalently attached to the 5' end of the T-strand [29].Concomitantly, the VirD4-like T4CP recruits the DNA substrate through recognition of translocation signals (TSs) carried by the relaxase and other Dtr proteins [30,31].To determine if the pED208-encoded Dtr factors recognize F's oriT sequence and vice versa, we constructed poriT plasmids carrying the oriT sequences of pED208 or F. Host cells harboring pED208 and poriT ED transferred both plasmids at equivalent frequencies, and similarly for host cells harboring F and poriT F (Fig. S2A).
However, hosts with pED208 and poriT F failed to transfer poriT F , and similarly for hosts with F and poriT ED (Fig. S2A).These findings, coupled with results of T4CP swapping studies presented below, strongly indicate that the pED208-and F-encoded Dtr factors fail to bind or process the noncognate poriT substrates for transfer.
In F-like systems, the VirD4-like TraD receptors couple F plasmids to their cognate T4SS channels [32].To test whether F-encoded TraD (TraD F ) recognizes the pED208 substrate, we constructed donors carrying pED208traD and producing TraD F from a separate plasmid (Fig. 1A).Remarkably, TraD F supported transfer of pED208traD through the pED208-encoded T4SS (T4SS ED ) at frequencies only ~10 2 -fold lower than observed for TraD EDmediated transfer of wild-type (WT) pED208 (Fig. 1Bi).Hosts with FtraD and producing TraD ED from a separate plasmid also transferred FtraD through the F-encoded T4SS (T4SS F ), albeit at lower frequencies (~5 x 10 -7 transconjugants/donor, Tcs/D) than observed for TraD F -mediated F transfer (Fig. 1Ci).Both TraD T4CPs thus functionally interacted with the noncognate plasmid substrates and T4SSs, although TraD F coupled the pED208 substrate to the T4SS ED channel considerably more efficiently than TraD ED functioned in the F system.Both TraD proteins produced from separate plasmids, as well as TraD deletion mutants and chimeric proteins described below, accumulated at detectable levels, as shown by immunostaining of N-terminal Strep-tagged variants; we also confirmed that the Strep-tag did not affect TraD functionality (Fig. S2B,C).
Distinct regions of TraD contribute to recruitment of cognate and noncognate substrates.F-like TraD subunits consist of an N-terminal transmembrane domain (NTD) implicated in establishing contacts with IMC components, a central nucleotide binding domain (NBD) thought to provide the energy for early-stage substrate processing and transfer reactions, and an unstructured C-terminal domain (CTD) involved in substrate binding [4,32,33].The NBDs of TraD ED and TraD F have conserved primary sequences and adopt similar structures as predicted by Alphafold [34], whereas their NTDs and CTDs are more divergent (Fig. S1B).Previous studies have determined that the C-terminal ~8 residues of TraD F specifically interacts with TraM F , a Dtr factor that binds the oriT F sequence and promotes formation of the catalytically-active relaxosome through recruitment of the TraI F relaxase [35,36].The TraD F -TraM F contact thus physically couples the F plasmid substrate to the T4SS F .
We confirmed that deletion of 15 C-terminal residues (C15) from TraD F and TraD ED attenuated transfer respectively of F and pED208 through their cognate T4SSs (Fig. 1Bii,Cii), likely due to loss of the TraD -TraM interactions.Importantly, even in the absence of the C-terminal discrimination motif, both TraD variants supported transfer of their cognate substrates, suggesting that other domains of these T4CPs also contribute to substrate binding and recruitment.Consistent with this proposal, deletions of the larger, nonconserved CTDs (166-residues for TraD ED , 148 residues for TraD F ; Fig. S1B) attenuated transfer of both plasmids through their respective T4SSs by ~10 1-1.5 -fold relative to effects of the C15 mutations (Fig. 1Bii,Cii).Importantly, however, even the CTD truncation mutants (designated NTD/NBDs) supported transfer of the cognate plasmids at frequencies well above threshold levels of detection (<10 -8 Tcs/D) (Fig. 1Bii,Cii).Thus, whereas the C15 motifs conferred 10 4-5 -fold stimulatory effects on transfer of the cognate substrates, the CTDs devoid of the C15 motifs also stimulated transfer by 10 1-1.5 -fold, and the NTD/NBD regions devoid of the CTDs stimulated transfer by 10 1-2.5 -fold.
Having shown that TraD F integrates into the pED208 system (Fig. 1Bi), we sought to identify the TraD F domains contributing to recognition of the pED208 substrate.As expected, deletion of the C15 discrimination motif had no effect on TraD F -mediated transfer of pED208 through the T4SS ED .Further analyses of truncation mutants established that TraD F 's CTD and NTD/NBD regions each conferred ~10 3 -fold stimulatory effects on pED208 transfer (Fig. 1Biii).Although TraD ED supported only a low level of F transfer through the F system, the corresponding analyses of TraD ED truncation mutants in the F system showed the same general pattern: TraD ED 's C15 motif did not contribute to F transfer and the CTD and NTD/NBD regions each stimulated transfer by ~10 0.5-1fold (Fig. 1Ciii).Sequence or structural motifs residing in TraD's CTD and likely in the NBD thus also contribute to recruitment of noncognate substrates.Finally, we analyzed the capacity of TraD chimeras bearing swapped C15 motifs to promote transfer.The TraD F C15 ED chimera elevated transfer of pED208 through the T4SS ED to levels similar to that conferred by native TraD ED , and the TraD ED C15 F chimera also efficiently conveyed F through the T4SS F (Fig. 1Biii, Ciii).These findings confirm the discriminatory role of the TraD C15 motifs for recruitment of cognate F-like plasmids.
We expanded these studies by testing whether a T4CP that naturally lacks a CTD also functionally integrates into the F and pED208 systems.The TraJ T4CP is required for transfer of the IncN plasmid pKM101 through the pKM101-encoded T4SS (T4SS KM ) [37]; it is only weakly related to TraD ED throughout its length, although the predicted NBD fold resembles that of TraD ED (Fig. S1B).Interestingly, donors harboring pED208traD and producing TraJ KM supported transfer of pED208traD, albeit at a low frequency of ~10 -7 Tcs/D (Fig. 1D).We appended TraD ED 's CTD to TraJ KM , which elevated pED208 transfer by ~10 1.5 -fold.We also substituted TraD ED 's NTD for that of TraJ KM , but this completely abolished pED208 transfer (Fig. 1D), possibly reflecting the importance of functional or structural interactions between the N-terminal transmembrane and nucleotide-binding domains of T4CPs [38][39][40].TraJ KM efficiently mediates the transfer of poriT KM , a plasmid harboring the pKM101 oriT sequence, through the T4SS KM [37].In the absence of TraD, TraJ KM also supported transfer of poriT KM through the T4SS ED and T4SS F channels at moderate frequencies (Fig. 1E, F), showing that TraJ KM functionally interacts with both T4SSs.We next asked whether TraD ED and TraD F recruited the poriT KM substrate.Although the fulllength T4CPs failed to support poriT KM transfer, variants deleted of the C15 or CTD motifs mediated transfer (Fig. 1E, F).TraD's NTD/NBD thus recruits a completely heterologous substrate, whereas the C15 motif blocks this recruitment.Finally, strains co-producing native TraD ED or TraD F and TraJ KM failed to transfer poriT KM , suggesting that the TraD T4CPs outcompete TraJ KM for engagement with the cognate T4SS channels (Fig. 1E, F).
Summary: With the sole exception of the structurally-defined TraD CT -TraM interaction [35], the nature of T4CP -relaxosome contacts is poorly understood for any conjugation system.Here, we supplied genetic evidence that TraD's C-terminal domain plays an important role in recruitment of cognate F or pED208 substrates.This domain also appears to actively block other TraD motifs from productively engaging with the heterologous pKM101 substrate.Outside of the C-terminal discrimination domain, TraD carries sequence or structural motifs in its CTD and likely NBD that contribute to recruitment of cognate F-like plasmids as well as noncognate DNA substrates.
Elsewhere, studies have shown that relaxases such as TraI F harbor internal translocation signals (TSs), while other relaxases can harbor C-terminal TSs [22,31,41].Additionally, T4CPs lacking a discernible TraD discrimination motif have been shown to bind relaxosomal components other than relaxases [37,42].It is conceivable that TraD carries distinct motifs capable of recognizing various TSs, enabling low-level promiscuous transfer of many different DNA substrates.Alternatively, TraD may engage promiscuously with many different DNA substrates through recognition of structural features universally shared by all relaxosomes by virtue of a common DNA enzymology.Even if TraD mediates transfer of heterologous substrates at low frequencies, the capacity to do so sets the stage for the evolved higher-frequency transfer of such substrates through mutation.IMC components are not swappable.The pED208-and F-encoded IMCs are composed of VirB3-like TraL, VirB4-like TraC, VirB6-like TraG, and VirB8-like TraE [43][44][45].The homologs exhibit between 24 to 55 % sequence identities, but are predicted to adopt similar structural folds (Fig. S1A, C).Deletion of IMC genes from both F-like plasmids abolished DNA transfer, and complementation by expression of the corresponding genes from separate plasmids fully restored conjugation proficiencies (Fig. 2 A, B).However, swaps of IMC genes between the pED208 and F systems failed to support DNA transfer at detectable levels (Fig. 2A,B).
Previous studies have shown that TraD T4CPs are not required for elaboration of F or ED208 pili [45,46].In contrast, the IMC deletion mutations abolished F and ED208 piliation, as evidenced by resistance of cells carrying the mutant plasmids to infection by male-specific bacteriophages (Fig. 2A, B, Fig. S3B).The IMC deletion mutants were resistant to single-stranded (ss) RNA phage MS2, which binds the sides of F pili but not ED208 pili, as well as to ssDNA phage M13, which binds the tips of both F and ED208 pili [46,47].We assessed MS2 sensitivity with a standard plaque assay, and sensitivity of non-lytic phage M13 using M13K07, which confers Kan r upon infection.Complementation of the deletion mutations with the cognate IMC genes restored sensitivity to both phages in all cases, but complementation with the swapped IMC genes failed to restore phage sensitivity (Fig. 2A,   B, Fig. S3B).In both systems, therefore, the swapped IMC subunits either fail to integrate into the heterologous T4SSs or when integrated the chimeric T4SSs are nonfunctional.
Summary: In the recently reported T4SS R388 structure, components of the IMC were shown to form a highlyspecific network of intersubunit interactions [5].It is reasonable to propose that F-encoded IMCs similarly have evolved specific networks of intersubunit contacts, to the extent that subunit exchanges even of these structural homologs are not tolerated.In comparing the sequences of the Tra/Trb homologs between the pED208 and F systems, we noted that sequence divergence of several subunit pairs was confined to short stretches rather than distributed throughout the proteins.For the IMC homologs, this was particularly evident for the TraL subunits (Fig. S1C, red lines).While such motifs might simply mark functionally-unimportant positions that accumulate mutations over time, they also may contribute to system-specific functions, for example, by mediating requisite contacts with cognate Tra subunits.Further mutation or swapping of such motifs between structural homologs should discriminate between these possibilities.

F-encoded OMCCs tolerate subunit or domain deletions and swaps. No high-resolution structures exist for
IMCs of F systems, but as mentioned earlier the OMCC ED structure was recently solved [5,15].The maps reveal intriguing features, including the presence of two distinct substructures, the ORC and CC, which are only sparsely connected and have different rotational symmetries (Fig. 3A).Remarkably, in testing the biological importance of various structural domains through mutational analyses, we discovered that the T4SS ED deleted of the major ORC component, TraK, retained the capacity to translocate pED208 but failed to elaborate the ED208 pilus [15].
Deletions and swaps of ORC components confer "uncoupling" phenotypes.To further interrogate the OMCC subunit and domain requirements for F-like systems, we first deleted and swapped ORC components (Fig. 3A).
Deletion of traK F phenocopied the traK ED mutation, rendering the T4SS F proficient for DNA transfer (~10 -6   Tcs/D) but unable to produce pili (Fig. 3B, C).As our previous assays, e.g., phage infection, only indirectly tested for F pilus production, we additionally deployed a sensitive fluorescent labeling procedure that enabled direct visualization of ED208 pili [25].Strains of interest were engineered to produce Cys-derivatized TraA ED pilins (TraA.C116), shown previously to support pED208 transfer at WT levels [25].The Cys118 residues are surfaceexposed on ED208 pili and accessible to labeling with fluorescent maleimide conjugates, e.g., FM488-mal.We and others also have deployed fluorescently-tagged ssRNA phages such as MS2 to decorate F pili [25,46,48].By fluorescent labeling of ED208 and F pili, we confirmed that cells carrying pED208traK or FtraK indeed fail to elaborate pili despite their capacity to transfer plasmids (Fig. 3D, E).The traK mutant hosts also were resistant to infection by M13K07, further showing that the traK machines do not even produce 'vestigial' pili with only their tips exposed on the cell surface (Fig. 3D,E, S3C).The traK mutations thus genetically "uncouple" the two major biogenesis pathways of T4SSs culminating in functional translocation channels or conjugative pili.
A swap of TraK F for TraK ED elevated pED208 transfer by ~10 2.5 -fold over that observed for the traK ED mutant machine, whereas the reciprocal swap elevated F transfer by ~10 4 -fold, approaching frequencies observed for transfer through the native T4SS F (Fig. 3B, C).Yet, neither chimeric machine elaborated detectable levels of pili (Fig. 3D,E, S3C).These findings firmly establish that F-like pili are completely dispensable for efficient DNA transfer through F-like channels.
VirB7-like TraV subunits have N-terminal lipid modifications that tether them at the OM, and an N-terminal helical domain (NTD) that binds laterally along the surface of the CC, forming contacts with three TraB -barrel domains [5,15].The C-terminal domain (CTD) assembles as two antiparallel -strands connected by a loop.In the assembled OMCC ED , the CTDs of two TraV monomers stack on top of each other, forming a 4-stranded -sheet.
These TraV CTDs connect laterally with adjacent TraV CTDs, so that collectively the 13 pairs of TraV CTDs build a belt that surrounds the 13 CTDs of TraK, presumably stabilizing the ORC substructure [15].In the pED208-and F-encoded TraV subunits, the NTDs and CTDs are well-conserved, but an intervening linker connecting the two domains is highly divergent (Fig. S1D).This sequence divergence is in line with our previous finding that deletion of the TraV ED linker had no discernible effect on T4SS ED function [15].Deletion of traV ED blocked all T4SS EDassociated activities (Fig. 3B,D, S3C) [15], but surprisingly the equivalent traV F mutation in the F system conferred the Tra + , Pil -"uncoupling" phenotype (Fig. 3C,E, S3C).
A swap of TraV F for TraV ED conferred the "uncoupling" phenotype, as evidenced by restoration of pED208 transfer without comparable effects on ED208 pilus production (Fig. 3B,D, S3C).In striking contrast, the swap of TraV ED for TraV F fully restored both F transfer through the T4SS F and F pilus production (Fig. 3C,E, S3C).

Mutations and swaps of the AP and C-terminal domains of the CC component, TraB, confer "uncoupling"
phenotypes.Next, we tested whether the CC substructures tolerated swaps or mutations.The VirB10-like TraB subunits are multi-domain proteins composed of NTDs that span the inner membrane (IM), proline-rich regions (PRRs) that extend across the periplasm, and C-terminal -barrel domains comprising the central cone or ring structures of OMCCs (Fig. 4A) [2,49].The TraB ED and TraB F homologs are only weakly related throughout their NTDs and PRRs, but their -barrel domains are well conserved and have similar predicted structures (Fig. S1D).A motif termed the antennae projection (AP; residues 306 -372, TraB ED numbering) associates with the distal region of the TraB -barrel (Fig. 4A).This motif consists of two -helices and an intervening loop (APL; residues 332 -346).The 17 APs associated with the 17 -barrels comprising the CC assemble as an OM-spanning channel, resulting in surface exposure of the APLs.The C-terminal ~80 residues (designated CTDs) of the 17 TraB subunits extend from the APs along the perimeter of the -barrels and then project into the central chamber of the OMCC (Fig. 4A) [5,15].Swaps of the TraB homologs did not restore functionality of either T4SS (Fig. 4B-E), which was not surprising given that members of the VirB10 family span the entire cell envelope and their NTDs form various contacts with T4CPs and other IMC components [50][51][52].To decipher contributions of specific TraB domains to T4SS ED and T4SS F functions, we constructed TraB F chimeras bearing the following TraB ED domains: NTD and PRR (designated TraB1), -barrel (TraB2), AP (TraB3), or CTD (TraB4) (Fig. 4A).All chimeras were stably produced (Fig. S4A).The TraB1 and TraB2 chimeras failed to support plasmid transfer or pilus production in either the pED208 or F systems, showing that the N-terminal regions and -barrel domains cannot be swapped (Fig. 4B, C).
The TraB3 (AP swap) and TraB4 (CTD swap) chimeras also failed to support pED208 transfer or ED208 pilus production (Fig. 4B,D), but we attribute this to the nonfunctionality of TraB F 's N-terminal region in the pED208 system.This is because the TraB3 and TraB4 chimeras supported F transfer through the T4SS F at WT levels, thus confirming that the APs and CTDs indeed are swappable (Fig. 4C).Remarkably, we detected F pili on <1 % of cells carrying FtraB and producing the TraB3 chimera and no F pili on isogenic host cells producing TraB4 (Fig. 4E).Consistently, both hosts also were resistant to M13K07 and MS2 infection (Fig. 4C).
We further interrogated the contribution of TraB's AP domain to T4SS function.To this end, we deleted the AP of TraB ED (TraBAP ED ) or substituted its APL with 5 Gly residues (TraBAPL5G ED ).Both variants were stably produced (Fig. S4A).TraBAP ED failed to support ED208 transfer or pilus production (Fig. 4B, D), establishing the essentiality of the OM channel for all T4SS ED functions.By contrast, TraBAPL5G ED supported pED208 transfer at high levels (10 -2 Tcs/D), but abrogated ED208 pilus production, as evidenced by M13K07 resistance and detection of ED208 pili on <1 % of cells examined (Fig. 4B, D).We conclude that the 5G motif suffices for proper insertion of the AP -helices into the OM as a prerequisite for substrate transfer.However, specific residues in the APL appear to be critical for pilus production (see Fig. 4D), as evidenced by the "uncoupling" phenotypes of the TraB3 chimera in the F system (Fig. 4C, E) and the TraBAPL5G mutation in the pED208 system (Fig. 4B, D) Summary: Our findings underscore the remarkable plasticity of the F-encoded OMCCs in accommodating major structural perturbations without loss of DNA transfer functions.These perturbations include deletions of the TraK and TraV components of the ORC, swaps of both subunits, swaps of the AP and CT domains of TraB, and substitution of the APL with 5xGly residues.Intriguingly, with the exception of the TraV ED swap for TraV F in the F system, all of these mutations conferred the Tra + , Pil -"uncoupling" phenotype, as evidenced by resistance to male-specific phages and production of very few or no pili.Thus, with respect to pilus biogenesis, the F-encoded OMCC is considerably less flexible in accommodating the perturbations imposed here.To account for these findings at a mechanistic level, we propose that the subunit or domain swaps or mutations conferring the "uncoupling" phenotype, disrupted system-specific contacts with other T4SS subunits that are selectively required for pilus production but not for substrate transfer.The peripheral ORC, as well as TraB's AP domain, are wellsituated to form contacts with one or more F-specific proteins described below that might selectively regulate pilus production.On the other hand, both the surface-exposed APL and TraB's CTD positioned within the central chamber of the OMCC central chamber might selectively contribute to piliation through direct interactions with TraA pilins or the growing pilus.Interestingly, the N-and C-terminal domains of the two TraV homologs, as well as the APL and CTD motifs of the TraB homologs, possess only a limited number of divergent sequences that might be responsible for the "uncoupling" phenotypes that accompanied subunit swapping (Fig. S1D).
More broadly, our findings strongly indicate that the F-like OMCCs are not simply passive players in the biogenesis pathways for F-like pili, for example, by providing the OM channels through which growing F-like pili extrude.This view is also supported by our recent in situ studies of the T4SS ED , which supplied structural evidence that the F pilus nucleates assembly on an OM as opposed to an IM platform [45].How the OMCC orchestrates nucleation of F pili and rounds of pilus extension and retraction remain intriguing questions for further study.
Effects of F-specific subunit swaps.F-like plasmids encode at least eight F-specific subunits that enhance or are required for DNA transfer and pilus assembly.Recently, we characterized effects of F-specific gene deletions from pED208 and F, which resulted in assignments of the F-specific subunits into one of three classes: i) Class I factors (TraF, TraH, TraW) are required for DNA transfer and piliation, ii) Class II factors (TrbC, TraU, TraN) are not required but contribute significantly to both T4SS F functions, and iii) one Class III factor (TrbB) is essential for F pilus production but not for plasmid transfer [25].One F-specific factor, TrbI, could not be classified because the trbI mutations exerted polar effects on expression of other tra/trb genes.As shown for other Tra subunits, the Fspecific homologs encoded by pED208 and F adopt similar predicted structural folds (Fig. S1E).Sequence identities range from 33 -68 %, but the alignments show that a few F-specific homologs diverge in only a few short regions (see Fig. S1E) that might account for the observed phenotypes accompanying subunit swapping.
Class I subunit swaps.Deletions of TraF, TraH, or TraW rendered the pED208 system completely nonfunctional (Fig. 5) [25] Swaps of F-encoded TraF, TraH, and TraW for their pED208 counterparts failed to restore functionality of the pED208 system, as shown by lack of DNA transfer, resistance to M13K07 infection, and absence of ED208 pili (Fig. 5A,B, S3D).The corresponding exchanges of TraF ED and TraH ED for their counterparts in the F system yielded similar outcomes (Fig. 5C,D, S3D).In the F system, the traW F mutant strain retained a low level of DNA transfer, was at least partially sensitive to M13K07 and MS2, and a few cells in the population (~5 %) elaborated detectable pili (Fig. 5C,D, S3D) [25].Interestingly, the TraW ED swap for TraW F restored F transfer through the T4SS F nearly to WT levels, but did not enhance F pilus production or confer sensitivity to phages beyond levels observed with the traW mutant (Fig. 5C,D).
Class II subunit swaps.Deletions of the Class II factors (TraU, TrbC, TraN) strongly attenuate plasmid transfer and production of ED208 and F pili (Fig. 5A-D) [25].Intriguingly, reciprocal swaps of TraU and TrbC in both systems restored transfer of pED208 and F nearly to WT levels (Fig. 5A,C) without corresponding enhancement of pilus production (Figs.5B,D).In contrast, both TraN swaps fully restored DNA transfer and pilus production in both systems to WT levels (Fig. 5A-D).These findings, and several other properties described below, distinguish TraN from the other Class II subunits and, indeed, all other Tra/Trb subunits.
Class III subunit swaps.TrbB, adopts a thioredoxin fold, possesses a catalytic CxxC motif, and functionally substitutes for the thioredoxin isomerase Dsb [53].In the pED208 system, TrbB ED confers only a ~10 2 -fold enhancement of DNA transfer (Fig. 5A), but is essential for pilus production (Fig. 5B) [25].Despite evidence that TrbB F can shuffle disulfide bonds even among substrates that are not involved in conjugation [53], the swap of TrbB F for TrbB ED failed to restore full function of the pED208 system (Fig. 5A, B).Curiously, in the F system, TrbB F is completely dispensable for DNA transfer, but contributes quantitatively to F pilus production both in terms of the number of pili produced per cell and the number of cells in a population with pili [25].The TrbB ED swap for TrbB F had no discernible effects on F plasmid transfer or the number of cells producing F pili relative to the trbB F mutant (Fig. 5C, D).

Summary:
The F-specific subunits have been proposed to assemble as one or F-specific complexes based on results of two-hybrid screens [43,[54][55][56].Notably, TraF F and TraH F interact with multiple F-specific partners as well as components of the OMCC, prompting models depicting these Class I factors as forming a central node that connects the F-specific complex(es) to the OMCC to regulate its activity [54][55][56].Such a central role might explain why the TraF and TraH swaps failed to restore activity of the pED208 and F systems: the swapped subunits likely do not form the requisite contacts for functionality of the heterologous systems.TraW also was reported to interact with TraF F and TraH F , as well as other F-specific subunits [54,56,57], which might account for its essentiality in the pED208 system.In the F system, however, TraW F appears to play a more peripheral role than TraF F or TraH F , possibly because other interactions among the F-specific components compensate for its absence.Also of interest, TraW ED integrated functionally into the F system, restoring DNA transfer but not enhanced production of F pili.
The central (residues ~70-100) and C-terminal (~140-218) regions of the TraW homologs (Fig. S1E) diverge in their primary sequences; conceivably these regions carry motifs specifying binding partner interactions selectively required for pilus production.
Results of our swapping studies suggest that the pED208-and F-encoded TrbC and TraU homologs share certain structural motifs or binding partner interactions, which enabled integration into the heterologous systems and restoration of DNA transfer.However, TrbC or TraU swapping did not restore pilus production, further implying that other motifs possibly within regions of sequence divergence orchestrate pilus production only in the context of the cognate T4SS.The TrbC homologs diverge in their sequences throughout their lengths, but the TraU homologs possess only a few small clusters of divergence (Fig. S1E) that might mediate such system-specific contacts.
Although phenotypes of the traN mutations supported assignment of TraN as a Class II factor [25], TraN has a number of features that are unique among components of conjugation machines, including its integral OM topology [59,60].Additionally, a large extracellular domain (ED) promotes formation of mating pairs through binding of outer membrane proteins (OMPs) displayed by recipient cells [59,[61][62][63].Different TraN subunits elaborate structurally distinct ED's that bind different OMPs, and this results in selective transfer of the TraN-encoding F plasmids to specific enterobacterial species.In the case of TraN F , for example, the ED mediates binding to E. coli OmpA, while the ED of TraN ED adopts a distinct structural fold consistent with binding of OmpW [24,61].Thus, while TraN ED and TraN F are swappable between the pED208 and F systems, in fact, they mediate formation of stable mating pairs through distinct ligand -receptor interactions.TraN subunits also possess large, Cys-rich periplasmic regions that contribute in unspecified ways to DNA transfer and pilus production [25,64].Intriguingly, no TraN interactions with other T4SS components have yet been identified [54,56], leaving open the question of how TraN F and TraN ED can function interchangeably in the pED208 and F systems.
Moreover, TrbB subunit swapping failed to restore activities of the heterologous systems, despite the likelihood that both proteins possess similar catalytic functions.The data thus favor the notion that TrbB subunits establish systemspecific partner interactions of critical importance for piliation but not translocation, independently of their disulfide isomerase activities.Such contacts might be mediated by sequence-divergent terminal regions of the TrbB homologs located outside of the thioredoxin folds (Fig. S1E).
TraA F pilins functionally integrate into the pED208 T4SS.The last Tra subunit required for assembly of Fencoded T4SSs is the TraA pilin.TraA is composed of four domains, with domains II and IV consisting of hydrophobic -helices and domain III of a short stretch of basic residues (Fig. S1F) [66,67].Newly synthesized TraA propilins insert into the IM via domains II and IV, the long leader peptide is cleaved by leader peptidase I (LepB), and the resulting pool of IM-integrated pilins is used for reiterative cycles of F pilus extension and retraction [66].In the assembled pilus, domains II and IV form the stacking interfaces of adjacent pilins, hydrophilic domain I and the extreme C terminus are surface-exposed, and domain III is located in the pilus lumen in a 1:1 stoichiometric association with IM-derived phospholipids [16].
The TraA ED and Tra F pilins are 44 % identical, possess the same domain architecture, and adopt similar predicted tertiary structures (Fig. S1E) as well as quaternary structures in the assembled ED208 and F pili [16].Despite these similarities, reciprocal swaps of the TraA pilins failed to restore T4SS ED or T4SS F functions (Fig. 6A,B).We discovered one reason for the lack of cross-functionality, namely, host strains carrying pED208traA and expressing traA F or FtraA and expressing traA ED failed to accumulate either of the TraA pilins at detectable levels (Fig. 6C).These findings suggested that the pED208 and F systems lack factors respectively required for stabilization of TraA F and TraA ED .In the F system, the membrane-bound chaperone TraQ was shown to direct insertion and stabilization of TraA F in the IM [68,69].As pED208 lacks a discernible traQ gene [24], we coexpressed traA F and traQ F in a strain harboring pED208traA.Indeed, this strain accumulated abundant levels of TraA F (Fig. 6C), and efficiently transferred the pED208 plasmid to recipients (Fig. 6A).Remarkably, however, this strain failed to elaborate F pili as shown by resistance to both M13K07 and MS2 phages and absence of pilus labeling with fluorescent MS2 (Fig. 6A, S3E).
We envisioned that another pED208-encoded factor might supply the TraQ-like chaperone function necessary for stabilization of TraA ED in the IM.We analyzed our collection of pED208 tra/trb mutations for effects on TraA ED production, but all mutant strains accumulated the pilin at detectable levels (Fig. S4B).pED208 carries several uncharacterized orfs in the tra region that might encode a TraQ-like function.Three (orfX1-3) are embedded in the OMCC gene cluster and one (orfX4) is in the F-specific gene cluster, which is where traQ resides in F (Fig. S4C) [24].We deleted the OMCC and F-specific gene clusters from pED208, but strains carrying the resulting plasmids also accumulated TraA ED pilin at abundant levels, arguing against contributions of the uncharacterized orfs, the entire OMCC, or F-specific complexes to TraA ED stabilization (Fig. S4D).It remains possible that a factor encoded elsewhere on pED208 is required for stabilization of TraA ED , although we present evidence below that the F system fails to incorporate TraA ED even when it does stably accumulate as a pilin pool in the IM.
Summary: Presently, it is not known how pilin subunits of T4SSs contribute to channel assembly or activity, or where they nucleate to build the extracellular filament.We recently solved in situ structures of the pED208encoded channel without and with the associated ED208 pilus in the native environment of the bacterial cell envelope [45].Several features of these structures support a model depicting the assembly of F-like pili on a platform near or at the OM.In the solved structure of the T4SS ED channel, there is a clearly defined cylindrical tube that extends from the IM and projects up through the OMCC, but does not cross the OM.Intriguingly, this central tube is present even in the solved structure of a traA ED mutant machine, establishing that it is not built from TraA ED pilins [45].Moreover, in the structure of the T4SS ED channel associated with the ED208 pilus, the pilus attaches to the OMCC at the OM junction, but several densities of unknown composition block the pilus lumen from joining with the cylindrical tube in the periplasm [45].Here, our finding that TraA F integrates into the pED208 system to restore DNA transfer without elaborating F pili, suggests that TraA F engages productively with the T4SS ED channel, but fails to form interactions with other T4SS subunits necessary for nucleation of the pilus.
Conceivably, the IM-integrated form of TraA F comprises an essential part of the IMC.Alternatively, upon extraction from the IM, TraA F might be delivered to the distal region of the channel where it comprises part of the OM gate.Further mutational studies of the divergent sequences between the TraA ED and TraA F homologs (Fig. S1F) should unveil domains or residues required for the distinct activities of these pilins in controlling channel function vs piliation.
Strains harboring co-resident pED208 and F mutant plasmids naturally assemble functional chimeric machines.Our findings to this point established that swaps of several tra/trb genes yielded chimeric machines functional at least for DNA transfer.The potential for chimeric T4SSs to assemble in natural settings is medically important in view of recent evidence that enterobacterial clinical isolates can carry two or more F-like plasmids [70][71][72].Here, we surveyed the COMPASS database [70] for strains carrying one or more F plasmids, and determined that indeed an appreciable fraction of sequenced strains (~15%) harbored two or more F plasmids (Fig. 7Ai), Recently, it was also reported ~25 % of 'classical' F plasmids (those designated as group A) lack one or more genes required for F transfer [26].We extended this analysis to include 5,664 F-like plasmids in the COMPASS database [70], which surprisingly showed that many are missing one or more tra/trb genes (Fig. 7Aii).Among the F-like plasmids lacking one tra gene, ~85 % are missing traQ (Fig. 7Aiii).Like pED208, these F-like plasmids might rely on another chaperone T4SSs for pilus biogenesis.However, among the remaining mutant F plasmids, many lacked genes encoding functions shown here to be complementable by subunit swapping between the distantly related F and pED208 systems (Fig. 7Aiii).
In view of these findings, we hypothesized that strains with multiple F plasmids might disseminate them widely even if one or more are non-selftransmissible.To test this idea, we constructed E. coli donors with all combinations of pED208 and F variants deleted of genes for the core VirB-and VirD4-like functions.We then assayed for transfer of both plasmids and at least the production of vestigial pili as monitored by M13K07 phage sensitivity.
The DNA transfer and phage infection data are presented in Fig. 7B, and a subset of the notable findings are summarized schematically in Fig. 7C.
Overall, results were consistent with data presented for the tra/trb gene swaps.For strains harboring one traD mutant plasmid, for example, both plasmids were conjugatively transferred even in cases when the second plasmid carried mutations, e.g., IMC deletions, that rendered the encoded T4SS nonfunctional (Fig. 7B, Ci).Moreover, both plasmids were transferred at frequencies expected of the chimeras predicted from results of our TraD swapping studies.Strains capable of assembling the T4SS F TraD ED chimera transferred pED208 variants at high frequencies (10 -1-2 Tcs/D), but F variants at low frequencies (10 -5-6 Tcs/D) because TraD ED functions poorly in recruitment of the F substrate (Fig. 1C).Conversely, strains competent for assembly of T4SS ED TraD F transferred both plasmids at high frequencies because TraD F efficiently recruits both F and pED208 substrates (Fig. 1Bi, Bii, Ci).Because TraD does not contribute to production of F-like pili [46], strains harboring at least one traD mutant plasmid elaborate pili and thus were sensitive to M13K07 infection (Fig. 7Biii).A strain carrying pED208traD and FtraD is transfer-defective but M13K07 sensitive, the latter presumably due to production of both ED208 and F pilus receptors (Fig. 7Biii, see legend).
Strains with combinations of pED208 and F plasmids deleted of IMC genes were invariably transfer-minus and resistant to M13K07 infection (Fig. 7B, Cii).These findings strengthen our proposal that the IMC homologs cannot be exchanged between these systems due to the evolution of system-specific interaction networks among IMC substructures.
With respect to the OMCC components, strains with co-resident plasmids bearing combinations of traK and traV mutations transferred both plasmids at moderately high frequencies of ~10 -4 Tcs/D (Fig. 7Bi, Bii, Ciii).These frequencies were considerably higher than the transfer frequencies observed for T4SSs lacking TraV or TraK (10 -8-6 Tcs/D) (Fig. 3B,C), confirming that TraV and TraK are freely exchanged even when the cognate machines are produced.Remarkably, nearly all strains carrying pED208traK and an F plasmid capable of donating TraK to the T4SS ED were sensitive to M13K07 infection.Similarly, strains carrying pED208traV, FtraK, or FtraV along with another F or pED208 plasmid capable of donating the missing TraV or TraK subunits were M13K07 sensitive (Fig. 7Biii).These findings contrast with earlier results showing that, among the combinations of OMCC swaps, only the exchange of TraV ED for TraV F in the F system supported production of F pili (Fig. 3E).To account for these findings, we propose that TraV or TraK are more readily donated to the heterologous T4SS when they are first able to adopt their folded structures in the context of their cognate machines.Regardless of the underlying mechanism, the finding that strains with combinations of F-like variants harboring OMCC mutations are minimally capable of elaborating 'vestigial' pili is suggestive of a functional synergism between coresident mutant machines.
Finally, strains carrying pED208traA and transfer-defective F variants, e.g., IMC deletion mutants, were proficient for transfer of both plasmids (Fig. 7Bi, ii, Civ).These results are consistent with our findings that the pED208 system is capable of recruiting TraA F pilins from a TraQ F -stabilized IM pool (Fig. 6A).Additionally, strains capable of assembling T4SS ED TraA F chimeric machines were resistant to M13K07 infection (Fig. 7Biii), showing that these chimeras fail to elaborate F pili or even display pilus tips on the cell surface.In contrast, strains carrying FtraA and transfer-defective pED208 variants failed to transfer either plasmid (Fig. 7B), which also is consistent with our above findings that TraA ED pilins fail to integrate productively into the F-encoded T4SS (Fig. 6B).

OVERALL CONCLUSIONS: T4SS chimeras constitute a mechanism for functional diversification and MGE dissemination.
By systematically analyzing effects of subunit or domain exchanges between distantly-related F and pED208 systems, we identified regions of the encoded T4SSs that accommodate sequence or structural perturbations without loss of DNA transfer proficiency.Remarkably, many of the chimeric machines were deficient in pilus production, suggesting that the requirements for elaboration of dynamic F pili are considerably more constrained than for elaboration of functional translocation channels.Mechanistically, the ability to genetically "uncouple" pathways for elaboration of functional channels vs biogenesis of F pili sets the stage for further definition of structural features and Tra/Trb partner interactions required for one but not the second of these dynamic processes.
On an evolutionary scale, it is reasonable to propose that intrinsic flexibility at the substrate/T4CP and T4CP/T4SS interfaces has contributed to diversification of T4SS substrate repertoires.Flexibility within the OMCC also can be predicted to account for the remarkable structural diversity recently shown to exist among OMCCs [2].Although it is hypothesized that this structural diversity endows T4SSs with specialized functions, such structure -function relationships await definition.On a more immediate time scale, the capacity of co-resident MGEs to build functional chimeric systems offers a general mechanism that can account for the persistence and spread of nonselftransmissible elements in natural settings.That many of the F-like chimeras characterized here were proficient in DNA transfer but not in pilus production further indicates that T4SS chimerism can prove beneficial to both the bacterial host and resident F plasmids.While such strains are proficient for dissemination of MGEs and associated fitness traits, they are resistant to host cell lysis and plasmid loss induced by male-specific bacteriophages.

Plasmid constructions. Plasmids and oligonucleotide primers are listed in Tables
F and pED208 mutations.E. coli strain HME45 were used to generate deletions of the tra genes from pED208 or pOX38 (the F plasmid) by recombineering as previously described [25,73].A FRT-Kan r -FRT cassette from pKD13 was amplified with primers listed in Table S2 to carry homologous sequence to the upstream and downstream regions of the region targeted for deletion.The λ red-gam system in HME45 cells carrying pED208 or F was induced by growth at 42°C, and the FRT-Kan r -FRT amplicon of interest was introduced by electroporation, with Kan r selection for recombinants.Complementing plasmids were introduced into recombinant strains for conjugative transfer of the pED208 or F plasmids harboring FRT-Kan r -FRT cassettes into MC4100 cells carrying pCP20, which expresses the Flp recombinase.Transconjugants were grown at 42°C overnight to induce Flp recombinase expression for excision of the FRT-Kan r -FRT cassette and to cure pCP20.E. coli strain MC4100 carrying pKD46 were used to generate deletions of the F-specific or OMCC gene clusters from pED208 or F. The λ red-gam system from pKD46 cells carrying pED208 or F was induced with 0.2% arabinose, and the FRT-Kan r -FRT amplicon of interest was introduced by electroporation, with Kan r selection at 42 °C to generate recombinants and cure cells of pKD46.pCP20 were introduced into the recombinant strains for excision of the FRT-Kan r -FRT cassette.Substitutions of tra/trb genes with the FRT scar were confirmed by sequencing across the recombination junctions.
Cloning of tra/trb genes.The pED208 and F tra/trb genes were amplified by PCR using primers listed in Table S2.
Amplicons were digested with NheI and HindIII, and the resulting fragments were inserted into similarly digested pBAD24.Plasmid pKKF083 expressing traA F from the nahG promoter was constructed by PCR amplification, digestion of the PCR product with NdeI and BamHI, and insertion of the resulting fragment into similarly digested pKG116.Plasmids pKN9 -pKN12 expressing different traB chimeras were constructed by Gibson assembly (NEB).Partial traB DNA fragments were amplified by PCR using pED208 or F as a template and primers listed in Table S2.The resulting fragments were inserted into pBAD24 digested with NheI and HindIII by Gibson assembly.
traD, traJ, traB constructions.pYGL343 and pYGL351 expressing N-terminally Strep-tagged traD ED and traDC15 ED , respectively, from the nahG promoter were constructed by PCR amplification using pED208 as the template, digestion of the PCR products with NdeI and BamHI, and insertion of the resulting fragments into similarly digested pKG116.pYGL342 and pYGL348 expressing N-terminally Strep-tagged traD F and traDC15 F , respectively, from the nahG promoter were constructed by PCR amplification using pOX38 as the template, digestion of the PCR products with NdeI and BamHI, and insertion of the resulting fragments into similarly digested pKG116.pYGL491 expressing traDC166 ED from the nahG promoter was constructed by deleting the C166 sequence from pYGL343 by inverse PCR.pYGL492 expressing and traDC148 F from the nahG promoter was constructed by deleting the C148 sequence from pYGL342 by inverse PCR.pYGL353 expressing traD ED C15 F from the nahG promoter was constructed by changing the C15 sequence of pYGL343 by inverse PCR.pYGL352 expressing traD F C15 ED from the nahG promoter was constructed by changing the C15 sequence of pYGL342 by inverse PCR.pYGL511 expressing traD ED C148 F from the nahG promoter was constructed by PCR amplifications using pYGL343 and pOX38 as templates, digestion of the two PCR products with KpnI and HindIII, and ligation of the resulting products.pYGL510 expressing traD F C166 ED from the nahG promoter was constructed by PCR amplifications using pYGL342 and pED208 as templates, digestion of the two PCR products with KpnI and HindIII, and ligation of the resulting products.pYGL493 expressing Strep-tagged traJ KM from the nahG promoter was constructed by PCR amplification using pKM101 as the template, digestion of the PCR products with NdeI and BamHI, and insertion of the resulting fragments into similarly digested pKG116.pYGL494 expressing traJ KM C166 ED from the nahG promoter was constructed by PCR amplifications using pKM101 and pYGL343 as templates, digestion of the two PCR products with NdeI and KpnI, and ligation of the resulting products.pYGL528 expressing N1-134 ED traJ KM from the nahG promoter was constructed by PCR amplifications using pKM101 and pYGL491 as templates, digestion of the two PCR products with NsiI and SpeI, and ligation of the resulting products.pYGL529 expressing N1-134 ED traJ KM C166 ED from the nahG promoter was constructed by PCR amplifications using pKM101 and pYGL343 as templates, digestion of the two PCR products with NsiI and SpeI, and ligation of the resulting products.pPK020 expressing traBAP ED from the P BAD promoter was constructed by inverse PCR amplification using pPK019 as the template and ligation of the resulting product.pPK021 expressing traBAPL5G ED from the P BAD promoter was constructed by inverse PCR amplification using pPK019 as the template and ligation of the resulting product.pYGL683 and pYGL684 expressing traBAp strep and traBAPL5G strep , respectively, from the P BAD promoter were constructed by PCR amplification using pPK020 and pPK021 as the templates respectively, digestion of the PCR products with NheI and HindIII, and insertion of the resulting fragments into similarly digested pBAD24.poriT plasmid constructions.pCGR97 carrying the pKM101 oriT sequence and expressing traK-traJ-traI KM was constructed by PCR amplification using pKM101 as the template, digestion of the PCR products with NheI and HindIII, and insertion of the resulting fragments into similarly digested pBAD24.pYGL490 carrying the pKM101 oriT sequence and expressing traK-traI KM was constructed by deleting traJ from pCGR97 by inverse PCR.pYGL248 carrying the F oriT sequence was constructed by PCR amplification using pOX38 as the template, digestion of the PCR products with NotI and HindIII, and insertion of the resulting fragments into similarly digested pBAD101.pYGL249 carrying the pED208 oriT sequence was constructed by PCR amplification using pED208 as the template, digestion of the PCR products with NotI and HindIII, and insertion of the resulting fragments into similarly digested pBAD101.
Conjugation assay.Donor and recipient cells were grown overnight at 37C in presence of the appropriate antibiotics, diluted 1:50 in fresh antibiotic-free LB media, and incubated with shaking for 1.5 h. 10 µl of donor and recipient cell cultures were mixed, spotted onto nitrocellulose filters placed onto LB agar supplemented as necessary with 0.2% arabinose for induction, and incubated for 5 h at 37C.The filters were suspended in 1 ml LB, serially diluted with LB and plated on LB agar containing antibiotics selective for transconjugants (Tcs), recipients, and donors, incubated at 37C overnight.The frequency of DNA transfer is presented as the number of transconjugants per donor (Tcs/D).All matings were repeated at least three times in triplicate; a representative experiment is shown with replicate data points and the average transfer frequencies as bars along with standard deviations as error bars.For matings with donors bearing coresident pED208 and F plasmids, donor and recipient cells were grown overnight at 37C in presence of the appropriate antibiotics, diluted 1:50 in fresh antibiotic-free LB media, and incubated with shaking for 2 h.75 µl of donor and recipient cell cultures were mixed in 96 well plate and incubated for 3 h at 37C, and the mating mix was serially diluted with LB and plated on LB agar containing antibiotics selective for transconjugants acquiring each of the two plasmids and donors.
Phage infection.Sensitivity of plasmid-carrying host cells to the M13KO7 and MS2 bacteriophages was assessed as previously described [25].Briefly, strains carrying F or pED208 plasmids were grown overnight in the presence of the appropriate antibiotics, diluted 1:50 in fresh antibiotic-free LB medium, and incubated for 3.0 h.A 1 ml aliquot of cells was incubated with 1 μl M13KO7 (10 11 plaque-forming units per ml, pfu/ml) (NEB) for 10 min on ice to allow attachment, and then at 37°C for 10 min to allow infection.Cells were washed once with fresh LB and incubated for an additional 30 min at 37°C.Cells were serially diluted and plated onto LB agar containing antibiotics selective for Kan r transductants and for total colony-forming units (CFUs).M13KO7 sensitivity was calculated as the number of Kan r transductants/total CFUs.M13KO7 phage infection experiments were performed at least three times in triplicate, with results presented for a single representative experiment showing replicate data points, average transfer frequencies as horizontal bars, and standard deviations as error bars.For M13K07 infections, source data are presented in Fig. S3, and presented in the manuscript as '+' (sensitive, defined as >10 -4   Kan r colonies/total CFUs), '-' (resistant, defined as <10 -6 Kan r /total CFUs), or '+ p ' (partially sensitive, 10 -4-6 Kan r /total CFUs).For determinations of MS2 phage sensitivity, a 500 μl aliquot of cells subcultured as described above was mixed with 5 ml LB soft (0.75%) agar.A 5 μl aliquot of MS2 (~10 11 pfu/ml) (kindly provided by L. Zeng, Texas A&M) was spotted and the plate was incubated overnight and examined for plaque formation.Results are presented as '+' (sensitive, clear plaques), '+ p ' (partially sensitive as evidenced by turbid plaques) and '-' (resistant, no plaque formation) [25].
Detection of Tra proteins by immunostaining.Overnight cell cultures (5 ml) were diluted 1:100 in fresh LB and incubated for 1.5 h at 37 o C. Sodium salicylate (1 M final concentration; VWR) was added for induction of traD, traJ, or traB variants, and cells were incubated for 1 h at 37 o C. Cultures (1 ml) were centrifuged at 5,000 x g for 5 min to pellet cells and resuspended in 50 l of physiologically buffered saline (PBS).An equivalent volume of 2x Laemmli buffer was added, and samples were boiled for 5 min prior to electrophoresis through SDS-12.5% polyacrylamide (30:0.8acrylamide/bis-acrylamide) gels.For detection of TraA pilins encoded by pED208 or F, overnight cell cultures (5 ml) were diluted 1:100 in fresh LB, incubated at 37 o C for 2 h, and 1 ml aliquots of cell cultures were harvested by centrifugation at 5,000 x g for 15 min.Cell pellets were resuspended in 50 l of PBS and 50 l of 2x Laemmli's sample buffer and boiled for 5 min.Samples were electrophoresed through SDS-16.5% polyacrylamide (30:0.8acrylamide/bis-acrylamide) gels.Electrophoresed material was transferred to nitrocellulose membranes and blots were developed with primary antibodies against the Strep-tag (Genscript) for detection of Strep-tagged TraD, TraJ, or TraB variants, or with anti-TraA polyclonal or JEL-92 monoclonal antibodies specific for TraA ED or TraA F , respectively (kindly provided by L. Frost).As a loading control, all blots were also developed with antibodies against E. coli RNA polymerase  subunit (RNP).Blots were developed with horseradish peroxidase (HRP)-conjugated secondary antibodies for detection of immunostained proteins by chemiluminescence.The entire protocol was repeated at least three times, and representative immunoblots are presented.

Detection of ED208 pili by AF488-maleimide labeling and F pili by decoration with MS2-GFP. ED208 pili
were fluorescently labeled as previously described [25].Briefly, host cells carrying pED208 variants and pKKF005 (expresses traAcys) were grown overnight, diluted 1:50 in fresh antibiotic-free LB medium, and incubated for 1.0 h at 37 o C. As appropriate, cells were induced for traAcys expression with 1mM sodium salicylate (final concentration), and incubated for 2.5 h at 37°C [25].Alexa Fluor™ 488 C 5 Maleimide (Thermo Fisher Scientific) was added to a 100 l aliquot of the cell culture at a final concentration of 25 g/ml followed by incubation for 30 min on ice.Cells were washed once in physiologically-buffered saline (PBS), and resuspend in 100 l PBS.
Labeled cells (5 l) were placed on a 1% PBS agarose pad for imaging with Nikon A1 confocal microscope with a Plan Apo 100x objective lens (Oil Immersion), and NIS-Elements AR software with FITC and DIC filters).F pili were visualized by adsorption of fluorescent MS2-GFP [25,46].Host cells carrying F variants were grown overnight, diluted 1:50 in fresh antibiotic-free LB medium, and incubated for 3.5 h at 37 o C. A 5 l aliquot of cells was mixed with MS2-GFP (10 10 pfu's/ml final concentration) and incubated for 30 min on ice.Labeled cells were placed on a 1% PBS agarose pad for imaging with a Nikon A1 confocal microscope.At least 250 cells per strain were examined to quantitate the percentage of cells in a population with visible ED208 or F pili.
Pairwise Structure Alignment.The TraA pilin and TraB ED structures were obtained from the RCSB Protein Data Bank web server (wwpdb.org)[74].Structures of the Tra and Trb homologs encoded by F and pED208 were predicted using the AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/) [34,75].Predictions of the TraJ KM structure were generated using standard settings and databases for ColabFold (https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb#scrollTo=kOblAo-xetgx) [76].The highest-ranking TraJ protein structure was depicted.Comparative model constructions of the Tra and Trb homologs were performed with the Pairwise Structure Alignment tool (https://www.rcsb.org/alignment) on the RCSB Protein Data Bank web server with the parameter set to jFATCAT.The generated pairwise structure alignments were modified using ChimeraX-1.5 [77].Bioinformatics analyses.We used the COMPASS database [70] to identify the numbers of F plasmids carried by bacterial strains identified as having at least one F-like plasmid.We identified a total of 908 bacterial strains as carrying at least one plasmid categorized as IncF.The fractions of strains harboring one or more F-like plasmids was presented in a pie chart format.We generated a presence/absence profile for the tra/trb genes using the IncF plasmid dataset (5,664 sequenced plasmids) obtained from PLSDB (v.2021_06_23_v2) [78].We carried out tblastn searches against a nucleotide database constructed from the IncF plasmid sequences dataset using the following Tra/Trb proteins as queries: TraA, TraL, TraE, TraK, TraB, TraV, TraC, TraF, TraW, TraU, TrbC, TraN, TraQ, TrbB, TrbI, TraH, TraG and TraD carried by pOX38 (MF370216.1),pED208 (AF411480.1),pKpQIL-UK (KY798507.1),and pOZ172 (CP016763.1).
We selected these plasmids as queries because ~90 % of the F plasmids analyzed here were from one of four enterobacterial species and the plasmid queries also were originally identified in these species (F, E. coli; pED208, Salmonella spp.; pKpQIL Klebsiella spp.; pOZ172, Citrobacter spp.).By default, hits greater than an e-value of 10 -3 were considered as positive hits.When there was a hit using at least one query, the gene was defined as present.Pie charts are presented for the fractions of F plasmids with complete sets of tra/trb genes vs those missing 1 or more, and for the fractions of plasmids missing specific tra/trb genes among the collection of F plasmids shown to lack 1 tra/trb gene.Upper: Schematics depicting chimeric T4SSs with substituted OMCC subunits (blue/green checkered); these systems translocate substrates and may or may not elaborate pili.Lower: pED208 or F transfer frequencies by donors bearing deletions of OMCC genes or isogenic strains expressing the complementing cognate or noncognate genes.Plasmid transfer frequencies and phage susceptibilities are presented as described in Fig. 2    Upper: Schematics depicting chimeric T4SSs with substituted F-specific subunits (blue/green checkered); these systems may or may not translocate substrates or elaborate pili (denoted as lighter coloration).Lower: pED208 or F transfer frequencies by donors bearing deletions of F-specific genes or isogenic strains expressing the complementing cognate or noncognate genes.F-specific subunits are grouped according to phenotypes of corresponding deletion mutations (Classes I, II, III, see [25]).Plasmid transfer frequencies are presented as described in Fig. 2 [70] and PLSDB (v.2021_06_23_v2) [78].i) The number of F plasmids per strain among all strains shown to carry at least one F plasmid.Colors represent the number of F plasmids per strain.A total of 908 strains carrying at least one F plasmid were identified and analyzed.ii) The prevalence of F plasmids with completely intact or missing tra/trb genes.Colors represent the fraction of F plasmids with the depicted number of missing tra/trb genes.A total of 5,664 F plasmid sequences were analyzed from PLSDB (v.2021_06_23_v2).The tblastn searches were performed against these F plasmid sequences using Tra and Trb proteins encoded by the following F plasmids as queries: pOX38 (MF370216.1),pED208 (AF411480.1),pKpQIL-UK (KY798507.1),pOZ172 (CP016763.1)(see Materials and Methods).iii) The fractions of specific tra/trb genes that were missing among F plasmids shown to lack one tra/trb gene.B) Heat maps showing transfer frequencies of Bi) pED208 (blue) and Bii) F (green) variants deleted of the tra genes shown by donor strains carrying the pairwise combinations of mutant plasmids depicted.
Transfer frequencies (Tcs/D) are color-graded as depicted.Biii) M13K07 susceptibility of host strains carrying the mutant plasmids depicted.M13K07 susceptibilities (Kan r colonies/total CFUs) are color-graded as depicted.The color shades reflect our predictions of what pili are produced enabling M13 infection; blue, ED208 pili; green, F pili; blue/green triangles, both pili are produced.C. Schematics and summaries of transfer and piliation among donors co-harboring pED208 and F plasmids with the mutations indicated.Strains with co-resident mutant plasmids are grouped to highlight: Ci) exchangeability of TraD T4CPs, Cii) lack of functional exchanges of IMC subunits, Ciii) exchangeability of OMCC subunits TraV and TraV, Civ) functional incorporation of traA F into the pED208 system, resulting in the Tra + , Pil -"uncoupling" phenotype.
Alignments of Tra/Trb protein homologs.Panel A: Alignment of the F and pED208 tra/trb regions.Blue: VirB/VirD4-like scaffold subunits that are conserved among T4SSs.Red: F-specific proteins that contribute to assembly or function of F-like T4SSs.Yellow: F-specific proteins involved in surface exclusion.Green: Nonconserved proteins.Lines connect genes encoding protein homologs; numbers correspond to the percent amino acid identities across the lengths of the homologs.Accession numbers: pED208 (NCBI Bioproject PRJNA871772); F (NZ_MF370216).Panels B -F (following pages): Similarities among the Tra/Trb homologs at the primary sequence and structural levels, grouped as indicated (orange highlights).Upper: Sequence alignments of the Tra/Trb proteins indicated, generated with Multalin [79].Known domains and percent identities are shown for regions indicated.Red lines denote regions of sequence divergence, one or more of which might confer the distinct phenotypes accompanying subunit swapping described in the text.Lower: Structures of the Tra/Trb homologs predicted from the AlphaFold Protein Structure Database [75].The solved TraA and TraB ED structures were from RCSB PDB.The structure of pKM101-encoded TraJ was predicted by ColabFold [76].Structures of the homologs are presented separately and superimposed.For TraB, only the -barrel and AP domains comprising the OMCC and AP channel are shown.Percent identities of the structurally aligned sequences, Root Mean Square Deviation (RMSD), TM-score, and number of equivalent residues relative to the entire sequence lengths were derived from the RCSB PDB Pairwise Structure Alignment website (https://www.rcsb.org/alignment).S2 Fig. Functionality of poriT plasmids and traD variants.A. Transfer of poriT ED and poriT F plasmids by donor cells carrying pED208 (blue bars) or F (green bars).B. Functionality of Strep-TraD in donor cells with pED208 (blue bars) or F (green bars) variants.Donor cells carried pED208 or F or the isogenic traD mutant plasmids without or with plasmids expressing the corresponding traD or strep-traD genes.Panels A & B: Transfer frequencies are presented as transconjugants per donor (Tcs/D).All matings were repeated at least three times in triplicate; a representative experiment is shown with replicate data points and the average transfer frequencies as horizontal bars along with standard deviations as error bars.Panels C & D: Host cells carrying pED208traD were assayed for production of the strep-tagged TraD ED , TraD F or TraJ KM variants shown.Total cellular proteins normalized on a per cell equivalent basis were subjected to SDS-PAGE and immunostaining of western blots with α-strep antibodies for detection of the T4CP variants or -RNP antibodies against E. coli RNA polymerase βsubunit as a loading control.S3 Fig. Quantitation of strain sensitivities to M13K07.M13K07 phage sensitivity is shown for host cells carrying F or pED208 or gene deletion variants without or with a plasmid expressing the corresponding genes from F or pED208.Phage sensitivity is reported as the number of kanamycin-resistant (Kan r ) transductants per total colony-forming units (CFUs).Panels: A. traD T4CPs; B. IMC subunits; C. OMCC subunits; D. F-specific components; E. traA pilins.All infection assays were repeated at least three times in triplicate; a representative experiment is shown with replicate data points and the average transfer frequencies as vertical bars along with standard deviations as error bars.Data in the manuscript figures are presented as '+' (sensitive, defined as >10 -4

Figure 3 .
Figure 3. Chimeric F-like T4SSs with substituted OMCC components can confer the Tra + , Pil -"uncoupling" legend.D & E. Visualization of Cys-derivatized ED208 pili by labeling with AF488-mal or F pili by labeling with MS2-GFP.Representative static images are shown for cells carrying pED208 or F, plasmid variants deleted of OMCC genes, or deletion plasmids along with plasmids expressing the complementing cognate or noncognate genes shown.Numbers correspond to percentages of cells with detectable pili in the cell population; ND, none detected.Scale bars, 2 m.

Figure 4 .
Figure 4. F-like T4SSs with substituted TraB chimeric proteins establish the functional importance of AP
legend.M13 or MS2 columns: Susceptibility of donor cells to M13 (M13K07) or MS2 phages: +, sensitive; + p , partially sensitive; -, resistant.Quantitative data for M13K07 infections are presented in Fig. S3.B & D. Visualization of Cys-derivatized ED208 pili by labeling with AF488-mal or F pili by labeling with MS2-GFP.Representative static images are shown for cells carrying pED208 or F, variants deleted of F-specific genes, or isogenic strains expressing the complementing cognate or noncognate genes.Numbers correspond to percentages of cells with detectable pili in the cell population; ND, none detected.Scale bars, 2 m.

Figure 6 .
Figure 6.A T4SS ED TraA F chimeric machine translocates pED208 in the absence of F pilus production.A & B. Upper: Schematics depicting T4SSs with substituted TraA pilin subunits (blue or green); the TraQ chaperone