A list of all strains and plasmids is provided in Supplementary Table S1. E. faecalis strains were routinely grown on BHI plates or in BHI medium with shaking at 37°C. E. coli was grown at 37°C in LB medium with shaking. Antibiotics were used at the concentrations given in Supplementary Table S1.

The pIP501∆traN in-frame knockout constructed by Kohler et al. (2018a) was complemented with the wild-type traN containing its native ribosomal binding site (RBS). TraN with its native RBS was thus amplified from pIP501 using the primers pEU327-BstYI_RBS traN fw and pEU327_SalI RBS traN rev. The PCR fragment was digested with SalI and BstYI and cloned into the shuttle vector pEU327. All primers used for cloning are listed in Supplementary Table S2.

Seven pEU327-RBS-traN variants were constructed for further complementation studies. They were designed using site-directed mutagenesis (New England Biolabs, Frankfurt am Main, Germany) according to the manufacturer’s instructions and transformed into competent E. coli DH5α cells. The sequence of the traN variants was verified by Sanger sequencing. For biparental mating assays, E. faecalis JH2-2 (pIP501∆traN) was transformed by electroporation with one of the seven pEU327-RBS-traN variants as described in Berger et al. (2022).

The biparental mating assays were performed as described previously (Berger et al., 2022). They were carried out in duplicates and repeated three times. pIP501 transfer rates are given as transconjugants per recipient cell. GraphPad Prism 8 statistical software was used for statistical analyses. Statistical significance was determined by using the unpaired two-tailed Mann–Whitney U test. Independent biological triplicates of knockout complementation by traN variants were compared with complementation by the native traN gene. p-values: p < 0.1 (*), p < 0.01 (**).

To add an N-terminal 7×His-tag in-frame to the traN sequence, the traN variants were subcloned into the 7×His-tag expression vector pQTEV. For improved readability, we refer to them without including the “His” prefix throughout the article. Thus, a Gibson Assembly Kit (New England Biolabs) was applied according to the manufacturer’s instructions. The primers for amplification of the pQTEV backbone (GA_pQTEV189-190 fw and GA_pQTEV189-190 rev) and traN (GA_pQTEV-traN fw and GA_pQTEV-traN rev) as well as those for sequencing are listed in Supplementary Table S2. The recombinant plasmids were transformed into E. coli DH5α (New England Biolabs). The sequence of the traN variants was confirmed by Sanger sequencing (Eurofins Genomics Germany). For expression of TraN and TraN variants (TraN_R23A-N24A-Q28A, TraN_G47A-G47A, TraN_H82A-R86A, and TraN_G100A-K101A), the recombinant plasmids were transformed into E. coli Lemo21 (DE3) cells (New England Biolabs).

To compare expression levels of the seven TraN variants with that of TraN, E. faecalis JH2-2 (pIP501∆traN) strains harboring pEU327-RBS-traN variants and pEU327-RBS-traN, respectively, were cultured in 4 mL BHI medium supplemented with respective antibiotics overnight under constant shaking at 37°C. E. faecalis JH2-2 (pEU327) and E. faecalis JH2-2 (pIP501∆traN) served as negative controls for verifying the antibody specificity. The overnight cultures were normalized to an optical density of 600 nm (OD₆₀₀) of 2 in a volume of 1 mL and harvested by centrifugation at 8,000 ×g for 15 min at 4°C. The cell pellets were washed three times in 1 mL phosphate buffer (50 mM KH₂PO₄/K₂HPO₄, pH 7) and resuspended in 300 µL SDS-PAGE loading buffer as per standard protocol (Laemmli, 1970). They were lysed by sonication through two pulses of 8 s followed by boiling at 95°C for 10 min. Identical amounts of the protein lysates were loaded onto a 12% SDS polyacrylamide gel. Western blots were probed with polyclonal anti-TraN antibody (BioGenes GmbH, Berlin, Germany) at 1:10,000 dilution followed by secondary horseradish peroxidase conjugated IgG antibody (Promega GmbH, Germany) at a dilution of 1:10,000. To check whether the TraN mutations impaired the affinity of the TraN antibody, we performed a Western blot with the heterologously expressed purified variants TraN_R23A-N24A-Q28A, TraN_G47A-G48A, TraN_H82A-R86A, and TraN_G100A-K101A, and the heterologously expressed TraN. We loaded 0.75 µg of each protein onto a 12% SDS polyacrylamide gel followed by immunoblotting, as described above.

The same expression and purification protocol was applied to TraN and its variants. A multiple sequence alignment of all variants is shown in Supplementary Figure S1A.

We prepared 50 mL overnight cultures (ONC) supplemented with 0.1 mg/mL ampicillin either from a single colony of an agar plate or from a frozen cryostock. Three-liter LB broth supplemented with 0.1 mg/mL ampicillin was inoculated with the ONC to an initial OD₆₀₀ of 0.05 and incubated under shaking at 37°C. Proteins were expressed after induction of the culture with 1 mM IPTG (final concentration) at an OD₆₀₀ of 0.4–0.6. Incubation continued for 3 h at 37°C under shaking. The cells were harvested by centrifugation at 9,000 ×g (JLA 8.100, Beckman Coulter Inc., Brea, United States) and stored at −20°C.

The frozen cell pellets were resuspended in a triple amount of 50 mM HEPES pH 7.5, 100 mM (NH₄)₂SO₄ supplemented with protease inhibitors (0.001 mg/mL pepstatin, 0.002 mg/mL antipain, and 0.02 mg/mL leupeptin final concentration; Merck KGaA, Darmstadt, Germany) in relation to the total net cell weight. The cell suspension was homogenized (T18 digital ULTRA-TURRAX^®, IKA®-Werke GmbH & Co. KG, Staufen, Germany) and sonicated in a rosette cell on ice (T13 flat tip, 50% duty cycle, ∼60% amplitude, Sonopuls, UW/HD 2070, Bandelin electronic GmbH & Co. KG, Germany) twice for 10 min each. The cell debris were removed by centrifugation for 45 min at 4°C and 35,000 ×g.

The supernatants were filtered through 0.45 µm filters (Rotilabo® syringe filters, Carl-Roth GmbH + Co. KG, Karlsruhe, Germany) and loaded onto a HisTrap HP 5 mL column (Cytiva, Chalfont St. Giles, England) for affinity purification using the ÄKTA Pure system (Cytiva). The proteins were eluted with a 0.3 M imidazole step and a 0.3–1.5 M linear imidazole gradient. The quality of the samples was assessed by 12% SDS-PAGE, and the fractions containing most of the desired protein were pooled and loaded onto a 5-mL HeparinTrap column (Cytiva) to remove bound DNA. We applied 50 mM HEPES, 100 mM (NH₄)₂SO₄ pH 7.5 as running buffer. The proteins were eluted with a NaCl gradient from 0 to 1.5 M. The eluting peak fractions containing most of the desired protein were pooled and concentrated using filter-insert spin tubes (Amicon^® Ultra centrifugal filters, Ultracell^® 3000 MWCO, Merck Millipore Ltd., Darmstadt, Germany).

Additional analytical gel filtration was performed with the ÄKTA avant system (Cytiva) to determine the purity and oligomerization state of TraN and its variants. For this, 500 μL protein samples were loaded onto a Sephadex 75 10/300 GL increase column (Cytiva) using a 2-mL loop and 50 mM HEPES, 100 mM (NH₄)₂SO₄ pH 7.6 as running buffer. To check for DNA bound to the proteins, a chromatogram with two wavelengths was recorded at 280 nm for protein and 260 nm for DNA detection. Expression and purification were monitored by SDS-PAGE using 12% SDS polyacrylamide gels. To normalize the amount of loaded cell mass, we took samples equal to an OD₆₀₀ of 2 in 1 mL and resuspended them in 20 µL SDS loading buffer.

Purified TraN and its variants were changed into a phosphate buffer (50 mM NaH₂PO₄ pH 7.0, 50 mM NaF) by filter-insert spin tubes (Amicon^® Ultra centrifugal filters, Ultracell^® 3000 MWCO, Merck Millipore Ltd.). Protein concentrations between 0.08 mg/mL and 0.13 mg/mL were used. CD measurements were performed on a J-1500 circular dichroism spectrophotometer (JASCO Corporation, Tokyo, Japan). Far-UV CD spectra were recorded at 20°C from 260 to 190 nm with a data pitch of 0.2 nm and a bandwidth of 2 nm applying a scan speed of 50 nm/min. Ten spectra were recorded and accumulated for each sample. The resulting spectra were baseline-corrected by subtracting the signal of the buffer. The CD signal was converted to mean residue ellipticity [θ]_MRE, and the secondary structure was determined with DichroWeb using the CDSSTR algorithm (Compton and Johnson, 1986; Whitmore and Wallace, 2004; 2008; Miles et al., 2022) and SMP180t reference dataset (Abdul-Gader et al., 2011). A weighted-means smoothing algorithm over 25 datapoints was applied to all spectra.

Two pIP501-specific double-stranded DNA (dsDNA) oligonucleotides were created by annealing their complementary single strands ordered from Thermo Fisher Scientific. The melting temperature (T_m) of the oligonucleotides was calculated with SnapGene Viewer (SnapGene software, www.snapgene.com). The oligonucleotides corresponded to 1) a specific 34-mer (original TraN binding site (oBS), T_m = 56°C), and 2) a random 34-mer (oriT _pIP501 region, to which no specific binding had been demonstrated (random DNA), T_m = 66°C) (Goessweiner-Mohr et al., 2014a).

The oligonucleotide sequences are given in Supplementary Table S2. The forward and reverse strands of the oligonucleotides were annealed before using them in MST. To this end, 30 µL of a 1 mM solution of each ss-oligonucleotide was combined. The mixture was kept at 95°C for 1 h in a heating block (Drybath Stdrd 1blck, Thermo Fisher Scientific) and then slowly cooled down in an enclosed Styrofoam box yielding 60 µL of 500 μM ds-oligonucleotide each.

Monolith NT.115 (NanoTemper Technologies GmbH, Munich, Germany) was used to measure the protein-DNA interactions between TraN and TraN variants with the oBS and the random DNA. The generation of the ds-oligonucleotides is described above in 2.8.

Purified TraN and its variants were changed into phosphate-buffered saline with tween (PBS-T) buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, 0.05% Tween 20 (v/v)) by filter-insert spin tubes (Amicon^® Ultra Centrifugal Filters, Ultracell^® 3000 MWCO, Merck Millipore Ltd.). Protein concentrations were determined using NanoDrop (ND1000, PEQLAB Biotechnologie GmbH, Erlangen, Germany). The proteins were diluted to 200 nM with PBS-T. As TraN and its variants featured an N-terminal 7×His-tag, a His-tag specific labeling method was used. The protein samples were mixed with RED-tris-NTA 2^nd Generation dye (NanoTemper Technologies GmbH) in PBS-T at a molar ratio of 2:1 to avoid unbound dye and were incubated in the dark for 30 min at RT. The fluorescently labeled proteins were centrifuged at 15,000 × g for 10 min. The supernatants were used for the binding assays.

The ds-oligonucleotides (oBS and random DNA, chapter 2.8) were diluted to 25.6 µM with PBS-T. A two-fold serial dilution of the oligonucleotides was performed with PBS-T in 500 µL reaction tubes as per the manufacturer’s manual (NanoTemper Technologies GmbH) resulting in final DNA concentrations of 12.8 µM to 0.781 nM (Supplementary Table S3). To prepare 15 different dilutions with DNA:protein ratios of 256:1 to 0.0156:1, the fluorescently-labeled protein samples (100 nM) were added to each dilution, resulting in a final concentration of 50 nM protein per sample. As a negative control, PBS-T buffer and 25 nM RED-tris-NTA 2^nd Generation dye was used. The samples were loaded into a capillary with a hydrophilic coating (NanoTemper Premium Capillaries, Cat# MO-K025, NanoTemper Technologies GmbH), placed into the capillary tray, and inserted into the Monolith NT.115 MST machine controlled by NTControl v2.1.31 software (NanoTemper Technologies GmbH). Capillary scans were performed at various LED power values before each measurement to ensure a sufficiently raw fluorescence signal (>200 counts). MST was performed at 60% LED and MST power with 5 s fluorescence detection before and after induction of local temperature gradient (Fluo before and Fluo after respectively) as well as with 30 s detection of the thermophoresis signal (MST On) and a 25 s delay between each measurement.

The data points were plotted using the MO. Affinity Analysis v.2.1.3 software (NanoTemper Technologies GmbH) and were fitted using the K_d model by setting the target concentration to 50 nM (final protein concentration). For calculation of the normalized fluorescence (F_norm) value, the T-Jump + Thermophoresis preset with an automatically determined best signal-to-noise ratio was chosen. The curve fitting parameters including the K_d were estimated and automatically calculated by the software. The results are presented as plots of ΔF_norm against the ligand (dsDNA) concentration in nM.

The E. coli/P. megaterium (previously known as Bacillus megaterium) shuttle vectors pRBBm59-RBS-traN-Strep and pRBBm59-RBS-gfp-Strep were constructed using Gibson Assembly (New England Biolabs). pRBBm59-RBS-gfp-Strep not encoding any pIP501 tra gene was used as a negative control in the pull-down assays. The backbone of pRBBm59-RBS-Strep was amplified using pRBBm59-RBS-traB-Strep (Berger et al., 2022) as a template using the primers GA_pRBBm59-RBS-Strep fw and GA_pRBBm59-RBS-Strep rev. The traN gene was amplified from pIP501 with GA_pRBBm59-traN fw and GA_pRBBm59-traN rev, and gfp was amplified from pRBBm59 by using the primers pRBBm59-GFP fw and pRBBm59-GFP rev (Supplementary Table S2). We chose Priestia megaterium as the expression host because it is a Gram-positive expression system and thus is better suited for a Gram-positive T4SS. Expression trials using E. coli hosts showed very low Tra protein expression. This approach has already been used recently by Berger et al. (2022).

P. megaterium MS941 (pMBGm19-RBS-traB-traO) (Berger et al., 2022) was transformed with pRBBm59-RBS-traN-Strep and pRBBm59-RBS-gfp-Strep, respectively, according to the protocol in Biedendieck et al. (2011). Transformants were selected on LB agar containing 35 μg/mL chloramphenicol and 10 μg/mL tetracycline and were screened for the presence of the respective pRBBm59 plasmids by colony PCR.

Expression and purification of Strep-tagged TraN and its interaction partners from P. megaterium MS941 (pMGBm19-RBS-traB-traO, pRBBm59-RBS-traN-Strep) lysate were performed as per Berger et al. (2022) with some modifications.

The expression cultures (1.2 L in 5 L baffled flasks) were inoculated to an OD₆₀₀ of 0.01 and incubated by shaking at 37°C to an OD₆₀₀ of 0.3–0.4. The expression of TraB-TraO and Strep-tagged TraN was induced by the addition of 0.5% (w/v) xylose for pMGBm19 and 0.5% (w/v) sucrose for pRBBm59. After 5 hours, the cells were harvested by centrifugation at 9,000 × g at 4°C for 60 min. The cell pellet was resuspended in three times the cell wet weight of complex binding buffer (100 mM Tris/HCl, 300 mM NaCl, 1 mM EDTA, pH 8), supplemented with protease inhibitor (1 μg/mL pepstatin, 2 μg/mL antipain, 20 μg/mL leupeptin in DMSO). The cells were then lysed by sonication for 1 h on ice (50% duty cycle, 60% intensity; Sonopuls, UW/HD 2070; Bandelin electronic GmbH & Co. KG, Germany). After 60 min of centrifugation at 4°C and 38,000 × g, the supernatant was filtered through a MF-Millipore 0.45 µm MCE membrane (Merck, Darmstadt, Germany) to eliminate residual cell debris. The filtered lysate was loaded onto a 5-mL StrepTrap™ HP column pre-equilibrated with complex binding buffer using an ÄKTA pure 25-L chromatography system. The proteins were eluted using the complex binding buffer supplemented with 2.5 mM desthiobiotin. Peak fractions were loaded onto 12% SDS polyacrylamide gels and subjected to immunoblotting. Primary polyclonal anti-Tra antibodies were used against TraB, TraE, TraF, TraG, TraH, TraJ, TraK, TraM, and TraN (BioGenes GmbH) with a dilution of 1:10,000, and TraO with a dilution of 1:1,000, to probe for coeluted Tra proteins. We used as secondary antibody a horseradish peroxidase conjugated antibody against rabbit IgG (Promega GmbH), diluted to 1:10,000.

Bands containing the elution fractions of the pull-down were excised from a silver stained 12% SDS polyacrylamide gel and subjected to in-gel reduction, alkylation, and trypsinization as per Gesslbauer et al. (2018). Supplementary Table S6 shows the resulting peptides. Extracted peptides were purified and concentrated using C18 spin columns (Pierce™ C18 Spin Columns; Thermo Fisher Scientific) as per the manufacturer’s protocol.

The concentrated purified peptides were applied to liquid chromatography (LC) through an Ultimate 3,000 RSLCnano System (Thermo Fisher Scientific). Peptides were separated with a flow rate of 300 nL/min on a C18 Aurora UHPLC column (25 cm × 75 µm ID, 1.6 µm) with CaptiveSpray Insert (IonOpticks). The mobile phases were A) 0.1% (v/v) formic acid in water and B) 0.1% (v/v) formic acid in acetonitrile. The HPLC gradient for separation was 2% B for 6 min, 2%–25% B for 90 min, 25%–40% B for 10 min, 40%–80% B for 10 min, 80% B for 10 min, and 2% B for 12 min. The LC system was coupled online to a timsTOF Pro mass spectrometer (Bruker Corporation, Billerica, Massachusetts, United States) with a CaptiveSpray nano-electrospray ion source (Bruker); the mass spectrometer was operated in PASEF mode. TIMS, PASEF, and calibration settings were applied as described by Meier et al. (2018).

Mass spectrometry raw files were processed with MaxQuant version 2.0.1.0 (Cox and Mann, 2008). MS/MS spectra were searched against the UniProt P. megaterium (strain ATCC 12872/QMB1551) (B. megaterium) reference proteome databank (Proteome ID: UP000000935). The amino acid sequences of TraA, TraB, TraC, TraD, TraE, TraF, TraG, TraH, TraI, TraJ, TraK, TraL, TraM, TraN, and TraO were manually included in the P. megaterium FASTA file. Search parameters were set as follows: trypsin was set as enzyme; a maximum of two missed cleavages was allowed; the minimum peptide sequence length was seven amino acids; the maximum peptide mass was 4,600 Da. Carbamidomethylation of cysteine was set as a fixed modification; oxidation of methionine and acetylation of protein amino-termini were set as a variable modification. All other MaxQuant search parameters were equivalent to those described by Meier et al. (2018).

TraN homologs were identified with an iterative PSI-BLAST and tBLASTn search using the National Center of Biotechnology Information (NCBI) nucleotide database (https://www.ncbi.nlm.nih.gov/). The criteria selected were: 1) only putative functional proteins; 2) the presence of a putative oriT checked by oriTfinder server (https://bioinfo-mml.sjtu.edu.cn/oriTfinder/); 3) presence of a relaxase gene; 4) presence of an additional signature gene of conjugative plasmids; 5) only plasmid encoded TraN homologs; 6) exclusion of unnamed plasmid entries; 7) exclusion of artificial plasmids such as shuttle vectors. As all putative TraN homologs identified by an iterative PSI-BLAST using TraN as query were annotated as hypothetical proteins, we incorporated a tBLASTn search as an additional criterion for pre-selecting plasmid-encoded TraN homologs. The tBLASTn algorithm is used to compare protein query sequences against a nucleotide sequence database, allowing the detection of potential homologous DNA sequences that might not be annotated as protein.

The web server oriTfinder (Li et al., 2018) was used to rapidly detect putative oriTs and transfer-associated modules such as relaxase genes, type IV coupling protein (T4CP) genes, or T4SS gene clusters within the selected plasmid sequences. The oriT site is an essential module of self-transmissible elements. A multiple sequence alignment of selected TraN homologs was performed with T-Coffee (version 11.0) using the Expresso algorithm with default settings (Armougom et al., 2006; Di Tommaso et al., 2011). Additionally, all pIP501 transfer proteins (TraA to TraO) were selected as query sequences for tBLASTn searches to identify homologs in putative conjugative plasmids encoding traN homologous sequences.

PlasmidFinder 2.1 (Carattoli et al., 2014) was used to identify the incompatibility group of the plasmids selected according to the criteria above. FASTA-formatted plasmid DNA sequences from NCBI were uploaded to the website of the Center for Genomic Epidemiology (CGE; https://cge.food.dtu.dk/services/PlasmidFinder-2.0/) and analyzed using these settings: 75% minimum identity threshold; 60% minimum coverage threshold. The database for Gram-positive plasmid replicon sequences was selected. Replicon families/incompatibility groups of the plasmids were identified based on the detection of specific replication (rep) initiator genes. Plasmid sequences for which PlasmidFinder found no match were not further analyzed and were denoted “not characterized”.

The coLAB AlphaFold2 server (Mirdita et al., 2017; Mirdita et al., 2019; Mitchell et al., 2020; Jumper et al., 2021; Mirdita et al., 2022) was used for 3D structure prediction of TraN and its putative homologs. In addition, the structures of the TraN variants from pQTEV constructs (TraN, TraN_R23A-N24A-Q28A, TraN_G47A-G48A, TraN_H82A-R86A, and TraN_G100A-K101A) were predicted to check whether the amino acid substitutions had an influence on the protein fold. Further structural predictions were made for a set of conjugative transfer proteins from different Firmicutes conjugative plasmids (indicated in subscript) to identify shared secondary structure elements: TrsR_pUC11B (GenBank accession number: WP_032398615.1), TrsR_pUC08B (WP_032398615.1), TrsR_pAF22 (WP_015062887.1), TrsR_p001F (WP_095348652.1), Tra20_pNP40 (WP_012477765.1), TraR_pNP40 (WP_012477778.1), Orf5_pMRC01 (WP_010890662.1), and TrsR_pIBB477a (WP_010890662.1). To visualize and determine root mean square deviation (RMSD) values, we performed alignments of the predicted structures to the crystal structure of TraN_pIP501 (PDB: 4P0Z, residues 1-117 are visible in the electron density) using PyMOL (PyMOL Molecular Graphics System, Version 2.6 Schrödinger, LLC). All alignments were performed using the “cealign” algorithm of PyMOL.

The transfer rate of the pIP501∆traN knockout has been shown to be two orders of magnitude higher than that of the original plasmid (Kohler et al., 2018a). However, this effect could not be restored to wild-type level by providing traN with the ribosomal binding site (RBS) of traJ _pIP501 in pEU327 in trans. In this study, we constructed the complementation plasmid pEU327-RBS _traN -traN containing the original RBS of traN. Henceforth, pEU327-RBS _traN -traN is referred to as pEU327-RBS-traN.

Providing pEU327-RBS-traN in trans to pIP501∆traN fully restored the wild-type pIP501 transfer rate in E. faecalis (Figures 1B,C). Thus, the polar effects of the traN knockout on the traO gene downstream in the tra operon can be excluded.

To investigate the impact of single alanine substitutions in TraN on pIP501 transfer, putative key residues were chosen based on their ability to form direct protein-to-base interactions which are defined as class 1 interactions (Kohler et al., 2018a) and direct protein-to-DNA-backbone interactions, respectively. Residues forming only water-mediated interactions were excluded. TraN key amino acid residues were identified based on the TraN/pIP501 DNA complex crystal structure (Kohler et al., 2018a) (Figure 1A). TraN contains two-winged helix-turn-helix (wHTH) motifs. Motif 1 is comprised of the amino acids interacting with the major groove of the DNA—residues R23, N24, and Q28—within the N-terminal TraN domain. Motif 2 is comprised of the residues H82 and R86 of the C-terminal domain, respectively. In addition, interactions with the minor groove of the DNA are exerted by the two wings of the wHTH motifs—residues G47–G48 and G100–K101 (Kohler et al., 2018a). To observe the effect of single alanine substitutions in TraN on pIP501 transfer, three amino acids were selected, each belonging to one DNA interaction site (Q28, H82, K101). Here, we exchanged these putative key residues against alanine by site-directed mutagenesis to disrupt the direct base contacts to the original TraN binding site upstream of the P _tra promoter of the pIP501 tra operon (Figure 1A). In trans complementation of pIP501∆traN was assayed with these variants cloned in the expression plasmid pEU327. Complementation effects were tested in biparental mating assays with E. faecalis.

Transfer rates of E. faecalis JH2-2 (pIP501∆traN) complementation strains (pEU327-RBS-traN_Q28A, pEU327-RBS-traN_H82A and pEU327-RBS-traN_K101A) were compared with pIP501, the complemented knockout with traN in trans, and the pIP501∆traN knockout strain. The single TraN amino acid substitution (H82A) revealed a small but significant increase in transfer rate (4.66 × 10⁻⁵) compared to complementation with traN in trans (1.17 × 10⁻⁵). Q28 and K101 substitutions led to non-significant changes in transfer compared to complementation with traN (Figure 1B). All transfer rates (mean values and standard deviations) are given in Supplementary Table S4.

As single amino acid substitutions of putative TraN key residues had only a minor effect on pIP501 transfer, three double alanine substitutions (G47A-G48A, H82A-R86A, G100A-K101A) and one triple substitution (R23A-N24A-Q28A) were generated. These residues are located on the DNA-binding α-helix and the wing of the wHTH motif of both N- and C-terminal DNA binding motifs. Variants of pEU327-RBS-traN with multiple alanine substitutions (G47A-G48A, R23A-N24A-Q28A, H82A-H86A, G100A-K101A) were constructed. E. faecalis JH2-2 (pIP501∆traN) was complemented with those variants. They were applied to biparental mating assays as donors and E. faecalis OG1X as recipient. Three of four multiple substitutions (R23A-N24A-Q28A, H82A-R86A, and G100A-K101A) revealed a significantly enhanced transfer rate. In each case, the differences were larger than one order of magnitude. The transfer rate was similar to that of the ∆traN knockout strain with a transfer rate of 1.72 × 10⁻³ (Figure 1C, Supplementary Table S4). Complementation of pIP501∆traN with traN-G47A-G48A showed the lowest effect (3.79 × 10⁻⁵) in comparison to complementation with traN (1.06 × 10⁻⁵) and the pIP501 transfer rate (1.72 × 10⁻⁵). Mean values and standard deviation of all transfer rates are given in Supplementary Table S4.

To exclude the observed changes in transfer rate of pIP501∆traN complemented with traN variants cloned into pEU327 being due to lack of expression of the proteins, we verified their expression in E. faecalis JH2-2. Expression of TraN and seven TraN variants was investigated in E. faecalis JH2-2, harboring pIP501∆traN and pEU327 carrying traN or one of the traN variants.

TraN as well as all TraN variants were expressed in E. faecalis JH2-2 (Supplementary Figure S2). However, the level of expression appeared to differ. TraN_H82A-R86A and TraN_G100A-K101A particularly showed lower expression than TraN and the other variants (Supplementary Figure S2), so we verified whether the alanine substitutions changed the binding affinity of the anti-TraN antibody. The four multiple-substituted TraN variants and TraN (TraN, TraN_R23A-N24A-Q28A, TraN_G47A-G48A, TraN_H82A-R86, and TraN_G100A-K101A) were expressed in E. coli Lemo21 (DE3), purified as described in chapter 2.6 and applied to immunoblotting (Supplementary Figure S1C), resulting in signals with differing intensity for TraN and the variants.

Especially TraN_H82A-R86A, with the lowest signal intensity in the immunoblot checking the expression (Supplementary Figure S2), and also TraN_R23A-N24A-Q28A, and TraN_G100A-K101A yielded a weaker signal in the immunoblot than TraN. As we were using a polyclonal anti-TraN antibody, we assumed that it binds to different TraN epitopes. If one of these epitopes is changed due to alanine substitution(s), it may result in a weaker signal in the immunoblot. Thus, it can be argued that the lower signal intensity for these TraN variants in E. faecalis might be due to the mutation(s) of epitopes recognized by anti-TraN antibodies and not due to lower expression levels. In addition, we want to emphasize that, due to application of the overexpression plasmid pEU327, sufficient protein levels of the TraN variants TraN_H82A and TraN_H82A-R86 will be present, although expression levels of these two TraN variants seem to be lower—as evidenced in the immunoblot (Supplementary Figure S2).

TraN and the four variants (TraN_R23A-N24A-Q28A, TraN_G47A-G48A, TraN_H82A-R86A, TraN_G100A-K101A) were expressed in E. coli Lemo21 (DE3) and purified by HisTrap, followed by HeparinTrap affinity chromatography. To assess purity and oligomerization states, an analytical size-exclusion chromatography (SEC) was performed. All variants were well expressed and exhibited similar final yields of purified proteins. Corresponding SDS polyacrylamide gels of all purification steps are shown in Figure 2 and the chromatograms showing the elution profile of purified TraN and its variants in Supplementary Figure S1B. TraN and its variants showed a pronounced overexpression comparing the samples before induction with the samples at harvesting. The elution fractions of the analytical SEC runs showed that we retrieved TraN and all its variants in high purity. The 260/280 ratio indicates that TraN and its variants were all free from bound DNA, which is a crucial prerequisite for DNA binding studies (Supplementary Figure S1B).

Circular dichroism (CD) measurements were performed to compare whether multiple alanine substitutions affect the fold and stability of TraN. All TraN variants showed almost identical CD spectra compared to TraN, displaying two minima at 207 nm and 216 nm. This suggests that there are no significant changes in secondary structure between the variants (Figure 3A).

We used the CoLab AlphaFold2 server for structure prediction of the TraN variants (TraN_Q28A, TraN_H82A, TraN_K101A, TraN_H82A-R86A, TraN_G47A-G48A, TraN_G100A-K101A, TraN_R23A-N24A-Q28A). The alignment of multiple-substituted TraN variants with the experimental TraN structure resulted in overall RMSD values between 0.588 Å (TraN_G47A-G48A) and 0.643 Å (TraN_H82A-R86A). The alignment of single-substituted variants to TraN gave even lower RMSD values of 0.519 Å to 0.540 Å (Supplementary Table S5). Alignments of the double- and triple-substituted variants are shown in Figure 3B. In addition, we provide close-up images of the regions showing the mutated residues (Figures 3C–F). The low RMSD values indicate that the mutations do not affect the protein fold in the predicted structures. In comparison, the alignment of the crystal structures of free TraN (PDB: 4P0Z) with TraN in complex with oBS (PDB: 6G1T) yielded a RMSD value of 0.837 Å.

To investigate the sequence-specific binding affinities of the TraN variants (TraN, TraN_R23A-N24A_Q28A, TraN_G47A-G48A, TraN_H82A-R86A, and TraN_G100A-K101A), we used MST with dsDNA substrates corresponding to the original TraN binding site (oBS) upstream of the pIP501 tra operon, and a random pIP501 sequence of the same size (34-mer). The oligonucleotides are identical to the 34-bp dsDNA (oBS) identified and used by Gössweiner-Mohr et al. (2014a) for isothermal titration calorimetry (ITC) measurements. Dose-response curves of all protein-DNA combinations are shown in Figure 4A. Additionally, the sites where mutations were introduced are shown on the TraN-DNA complex crystal structure (PDB: 6G1T) (Kohler et al., 2018a). Close-up images of the mutated sites are also depicted in Figure 3. The dissociation constants for the interaction of TraN and its variants with the ds-oligonucleotides are shown in Figure 4B. All fitted curves exhibit a sigmoidal shape from which the K_d was calculated. It was not possible for the interaction of the double-substituted proteins TraN_H82A-R86A and TraN_G100A-K101A with the random DNA to fit the data points by a sigmoidal curve, suggesting an extremely low binding affinity. In the presence of the random DNA, the K_d values for all other TraN variants were in the low micromolar range, suggesting a much lower affinity than the binding of the specific 34-mer (oBS).

TraN showed the strongest binding affinity to the oBS 34-mer sequence (1.17 × 10¹ ± 2.65 nM) followed by the double-substituted variant TraN_G47A-G48A (4.88 × 10¹ ± 9.75 nM). The triple-substituted variant TraN_R23A-N24A-Q28A showed an almost 13-fold increase in the K_d value (1.54 × 10² ± 1.95 × 10¹ nM) compared to TraN, suggesting a strong negative effect on DNA binding affinity. Both TraN_H82A-R86A (4.25 × 10² ± 1.02 × 10² nM) and TraN_G100A-K101A (4.06 × 10² ± 4.55 × 10¹ nM) variants showed approximately 36-fold higher K_d values yielding the lowest binding affinities to the oBS 34-mer of all variants compared to TraN.

The binding affinities determined through MST exhibit a strong correlation with the results obtained from the biparental mating assays. When the pIP501∆traN knockout strain was complemented with traN_G47A-G48A in trans (3.79 × 10⁻⁵ transconjugants/recipient), the impact on pIP501 transfer was minimal compared to complementation with traN-G100A-K101A (6.86 × 10⁻⁴) (Figure 1B). Consequently, the binding affinity of TraN_G47A-G48A was found to be higher than that of TraN_G100A-K101A, indicating its strongest capability in sequence-specific binding on pIP501 oBS amongst the tested variants.

Previously, yeast two-hybrid studies showed interactions of TraN with TraE, TraG, TraH, and TraL (Abajy et al., 2007). Therefore, we decided to perform a pull-down assay i) to confirm these interactions and ii) to elucidate further direct or indirect TraN interactions. Consequently, P. megaterium MS941 was co-transformed with pMGBm19-RBS-traB-traO and pRBBm59-RBS-traN-Strep. This resulted in co-expression of C-terminally Strep-tagged TraN (TraN-Strep) and pIP501 transfer proteins TraB to TraO in P. megaterium MS941.

Utilizing affinity chromatography, Strep-tagged TraN and the proteins interacting with it were purified from the clarified supernatant of P. megaterium MS941 (pMGBm19-RBS-traB-traO, pRBBm59-RBS-traN-Strep) lysate.

To detect coeluted Tra proteins, elution fractions from TraN pull-downs were subjected to immunoblotting with anti-Tra antibodies directed against TraB, TraE, TraF, TraG, TraH, TraJ, TraK, TraM, TraN, and TraO. The elution fraction was compared to the flow-through and lysate fraction. We could demonstrate the coelution of TraE, TraF, TraG, TraH, TraJ, TraK, TraM, and TraO, as evidenced by the presence of the respective bands (Figures 5A–J).

TraB was not coeluted, suggesting no interaction with TraN. In contrast with the calculated molecular weight (MW) of 8.3 kDa, the expected MW on the SDS polyacrylamide gel and the corresponding immunoblot is around 12 kDa (Berger et al., 2022). The membrane incubated with anti-TraO antiserum showed two bands which most likely present two differently processed forms of the protein: one with its signal sequence and one without it. The cleavage results in a size difference of 2.5 kDa (Berger et al., 2022). The single band representing TraG had a MW of around 20 kDa instead of the calculated MW of 40.4 kDa. It is hypothesized that the T4SS protein complex includes only the CHAP domain of TraG, as the observed band corresponds to its calculated MW of 22 kDa (Arends et al., 2013; Berger et al., 2022). Likewise, TraF was detected at around 60 kDa, which was slightly above the calculated 52.8 kDa. TraE was coeluted in rather small amounts. The nitrocellulose membrane incubated with the anti-TraK antibody showed two bands (Figure 5G), indicating the presence of two TraK variants. This can be attributed to the existence of two different translational start codons for TraK (Goessweiner-Mohr et al., 2014b). The coelution of TraM might be because TraN and TraM are both DNA binding proteins. In contrast to its C-terminal domain, the N-terminal domain of TraM showed unspecific DNA binding (T. Berger and W. Keller, unpublished data).

To confirm that the coelution of proteins together with the Strep-tagged TraN is highly specific and not the result of unspecific binding, we conducted a negative control experiment. A GFP pull-down was performed followed by immunoblotting using anti-TraN or anti-TraF antibodies (e.g., for a Tra protein which was coeluted in the TraN pull-down). TraN and TraF could be detected only in the lysate and flow-through fraction but not in the elution fraction (Figure 6), suggesting that there is no unspecific binding of pIP501 proteins.

Next, we analyzed the elution fractions from TraN pull-downs with high-resolution mass spectrometry. An in-gel tryptic digest was performed from the elution fractions, and the resulting peptides were analyzed by nano-LC-MS/MS. For the pull-down elution fractions, the MS spectra yielded positive hits for all Tra proteins except for TraA (not included in the expression vector, negative control), TraC, TraG, TraL, and TraO. Looking at the LFQ intensity signals of the samples (blue labels in Supplementary Table S7), TraE, F, H, J, K, M, and N show strong signals. TraN occurs in a huge surplus over the other Tra proteins as it carries the Strep-tag and binds to the affinity column whether attached to other Tra proteins or as a monomeric protein. Notably, TraB and TraD are detected but not clearly quantifiable, indicating their low abundance in the pull-down fraction.

Homology searches with TraN were performed to find putative similar proteins in other bacterial species or even genera.

A previous search for TraN homologs was performed by Kohler et al. (2018a). Despite their common presence in mobile genetic elements from Firmicutes, TraN homologs have been rarely investigated. As many new plasmid sequences have been submitted to NCBI in the last 5 years, we conducted a new search for putative TraN homologs. Additionally, we not only used a larger dataset but also assigned the selected plasmid sequences to putative conjugative and non-conjugative plasmids. According to the criteria defined in chapter 2.12, 54 plasmid sequences and one transposon sequence were selected for further analysis. All protein sequences displaying over 87% identity on amino acid level to TraN originated from Firmicutes.

According to the tBLASTn alignments, 26 out of 54 homologs displayed 100% identity on protein level with TraN, and 39 out of 54 homologs exhibited 100% query coverage with protein identities between 88% and 99% (Supplementary Table S8).

An oriT was predicted for 44 out of the 54 plasmids (see Supplementary Table S8). For the remaining ten, a tBLASTn search with the sequence of the pIP501 relaxase TraA, the ATPase TraE, and the coupling protein TraJ also revealed no hits. A putative TraJ_pIP501 homolog was identified for the oriT-lacking plasmids pSM19035 (AY357120.1), pF104_1 (CP072892.1), pEGM181-1 (CP050484.1), and pGMrib (CP045673.1). Plasmid pEGM181-1 additionally encodes for a TraE_pIP501 homolog but not for a TraA_pIP501 (Supplementary Table S9). Among the 54 plasmids with a predicted oriT, p2 (CP041740.1), plas1 (CP064340.1), and pVEF3 (AM931300.1) lack traA-/traE-/traJ-like sequences, with p2 being the only one encoding for a TraJ_pIP501 homolog (Supplementary Table S9).

Supplementary Table S8 provides additional information on the presence of a pIP501-like T4SS for the initially selected 54 plasmid sequences and one transposon sequence. The NCBI tBLASTn search employed the pIP501 transfer proteins TraA to TraO as queries. Of these MGEs, 65% encode a complete T4SS (traA—traO), underlining the conjugative potential of the TraN-encoding MGEs. Interestingly, four putative conjugative plasmids (p3-38, pEFS108_1, pL15-B, pW3) were identified as having a nearly complete T4SS cluster, lacking only a traO-like gene (Supplementary Table S8).

In total, at an amino acid level, 11 different TraN sequences were detected in 28 different plasmids.

Eleven plasmids, each representing a distinct TraN_pIP501 homolog and containing a putative oriT, were selected from the initial set of 28 (Table 1). Each TraN homolog shares ≤99% protein identity with TraN_pIP501. Table 1 displays homologs of three pIP501 signature proteins—TraA_pIP501, TraE_pIP501, and TraJ_pIP501—in the 11 selected plasmid sequences, indicating the likely presence of a T4SS in these plasmids that encode traN homologous sequences. All plasmids encoding a TraN_pIP501 homolog exhibited high-level protein identity (≥96%) to the pIP501 TraA, TraE, and TraJ T4SS proteins (Table 1). The 11 different TraN homologs were applied to a multiple sequence alignment using T-Coffee Expresso (Figure 7A).

Sequence alignments with T-Coffee Expresso were performed to determine the extent to which TraN homologous sequences deviate from TraN_pIP501. As depicted in Table 1, five TraN homologs are truncated: that encoded on p3-38 (TraN_1-97), the homolog from pEFS108_1 (TraN_1-111), and those from pP47-61, pFYY063-optrA-70K and pR05720-3 (TraN_1-114). The coLAB AlphaFold2 server was used to predict the structures of the 11 distinct TraN_pIP501 homologs. The structures were aligned with the solved structure of TraN (PDB: 4P0Z). As the sequence identity is quite high, the overall structure of all homologs is likely very similar to TraN_pIP501. The respective RMSD values are listed in Table 2. The most notable difference is observed in the homolog encoded on p3-38 (Figure 7B), where not all key residues are conserved. This discrepancy is attributed to its smaller size (99 aa instead of 122 aa), leading to the absence of Gly100 and Lys101. In all other homologs, the corresponding amino acids (Arg23, Asn24, Gln28, Gly47, Gly48, His82, Arg86, Gly100, and Lys101) were not altered, indicating that the mode of DNA recognition is conserved among all homologs compared.