The bacterial strains and plasmids used in this study are listed in Supplementary Table S1. Escherichia coli and Bacillus megaterium strains as plasmid hosts were cultured in lysogeny broth (LB) under shaking at 37°C, for B. megaterium strains baffled flasks were used. Enterococcus faecalis strains were routinely grown in brain heart infusion (BHI) broth at 37°C with shaking. 1% w/v agar was added for preparing solid media. For selection, antibiotics were added as indicated in Supplementary Table S1.

A pIP501∆traB in-frame deletion mutant was constructed in E. faecalis JH2-2 using the previously described markerless allelic exchange strategy (Kristich et al., 2007). Oligonucleotides were purchased from Thermo Fisher and are listed in Supplementary Table S2. Restriction enzymes were purchased from New England Biolabs, as well as T4 DNA ligase and Phusion DNA polymerase. The first recombinant construct for traB in-frame deletion was generated by amplifying traB upstream and downstream flanking regions (1,007 bp and 1,054 bp, respectively) using E. faecalis JH2-2 (pIP501) as template with primers containing PstI/XbaI restriction sites for the upstream fragment and BamHI/EcoRI for the downstream fragment. The fragments were sequentially cloned into pUC18 using the corresponding restriction sites to create pUC18-UPS-DWS-traB. E. coli DH5α was transformed with the recombinant DNA, clones were selected on LB agar with 100 μg/ml ampicillin. More than 96% of the traB coding region was deleted retaining only two intact N-terminal and C-terminal codons of traB to prevent polar effects on downstream tra genes. The fused flanking regions were excised from pUC18 by digestion with PstI and EcoRI and inserted into the equally digested suicide plasmid pKA, followed by transformation into E. coli EC1000, resulting in pKA-UPS-DWS-traB. Clones were selected on BHI agar containing 250 μg/ml erythromycin. The correctness of all plasmid constructs was confirmed by Sanger sequencing (Eurofins Genomics Germany GmbH, Ebersberg, Germany). Electrocompetent E. faecalis JH2-2 (pIP501) cells were transformed with pKA-UPS-DWS-traB by electroporation with a Bio-Rad GenePulser X-Cell using 0.2 cm gap width cuvettes and settings of 2,000 kV, 200 Ω, and 25 µF. Transformants were selected on BHI agar supplemented with 50 μg/ml fusidic acid, 20 μg/ml chloramphenicol, 20 μg/ml erythromycin, 100 μg/ml gentamicin, coated with 100 µl of a 20 mg/ml 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) solution. Blue colonies were screened for homologous integration of pKA-UPS-DWS-traB into pIP501 by PCR. Colonies with integrated pKA-UPS-DWS-traB were grown in BHI medium supplemented with 50 μg/ml fusidic acid and 20 μg/ml chloramphenicol three times to stationary phase and finally until an OD₆₀₀ of 0.4. Serial dilutions were spread on MM9YEG agar plates supplemented with 50 μg/ml fusidic acid, 20 μg/ml chloramphenicol and 10 mM DL-p-chlorophenylalanine, coated with 100 µl 20 mg/ml X-gal solution. White colonies were screened for traB deletion by PCR. The pIP501∆traB mutant was verified by sequencing of the deletion borders.

For traB knockout complementation, the traB wild type gene with its ribosomal binding site (RBS) was amplified from pIP501 with primers containing BstYI/SalI restriction sites and inserted into the shuttle vector pEU327 followed by transformation into E. coli DH5α cells. Selection was carried out on LB agar with 100 μg/ml spectinomycin. For detection of C-terminally Strep-tag II-coupled TraB proteins and complementation studies a linker, a C-terminal Strep-tag II coding sequence (SA-WSHPQFEK) and a stop codon were inserted into pEU327 by using the site-directed mutagenesis kit (Q5 site-directed mutagenesis kit, New England Biolabs, Frankfurt am Main, Germany) according to the manufacturer’s instructions using the primers pEU327_Mut_C-Strep fw and pEU327_Mut_C-Strep rev. After DpnI treatment the nicked DNA was transformed into E. coli DH5α cells, resulting in pEU327 containing the C-terminal Strep-tag (pEU327-Strep). To incorporate the traB gene with its native RBS into pEU327-Strep, a Gibson Assembly Kit (New England Biolabs) was applied. The primers used for amplification of pEU327-Strep (GA_pEU327_CStrep fw and GA_pEU327-CStrep rev) and traB including its RBS (GA_RBS-traB fw and GA_RBS-traB rev) as well as all other primers used for cloning or sequencing are listed in Supplementary Table S2. The protocol was carried out according to the manufacturer’s instructions, resulting in pEU327-RBS-traB with a C-terminal Strep-tag (pEU327-RBS-traB-Strep). pEU327-RBS-traB was used to delete the traB _2-30 sequence by Q5 site-directed mutagenesis according to the manufacturer’s instructions using the primers Mut_pEUtraB31-110 fw and Mut_pEUtraB31-110 rev, resulting in pEU327-RBS-traB _31-110. For biparental mating assays, E. faecalis JH2-2 (pIP501∆traB) was electroporated with either pEU327-RBS-traB, pEU327-RBS-traB-Strep or pEU327-RBS-traB _31-110 as described in Construction of a traB in-Frame Deletion Mutant in pIP501 Selection for E. faecalis JH2-2 (pIP501∆traB; pEU327-RBS-traB), E. faecalis JH2-2 (pIP501∆traB; pEU327-RBS-traB-Strep) and E. faecalis JH2-2 (pIP501∆traB; pEU327-RBS-traB _31-110) was conducted on BHI agar with 20 μg/ml chloramphenicol, 20 μg/ml erythromycin and 500 μg/ml spectinomycin. Transformants were confirmed by PCR analysis.

Biparental mating assays were performed with isogenic E. faecalis (pIP501), E. faecalis (pIP501∆traB), E. faecalis (pIP501∆traB; pEU327-RBS-traB), E. faecalis JH2-2 (pIP501∆traB, pEU327-RBS-traB-Strep) and E. faecalis JH2-2 (pIP501∆traB, pEU327-RBS-traB _31-110) as donor strains and E. faecalis OG1X as plasmid-free recipient as described in Fercher et al. (2016). Transconjugants were selected on BHI agar supplemented with 1.5 mg/ml streptomycin and 20 μg/ml erythromycin. All mating assays were performed in quadruplicates. Transfer rates (number of transconjugants per recipient cell) are given with standard deviation. Significance and p-values were calculated with the Welch’s t test. Significance is indicated by asterisks: ***p < 0.0003, **p < 0.0038.

Based on the secondary structure prediction of TraB a truncation variant of traB was generated, lacking the secretion signal sequence. It was denominated TraB_31-110 referring to the amino acids present in the truncated protein. The 243-bp traB _31-110 PCR product was cut with BamHI and HindIII (New England Biolabs) and ligated into the expression plasmid pQTEV cut with the same enzymes. pQTEV contains a tobacco etch virus (TEV) protease-cleavable His-tag. The recombinant plasmid was transformed into E. coli BL21-CodonPlus (DE3)-RIL and BL21 star (DE3). The correctness of the traB _31-110 sequence was verified via Sanger sequencing by Microsynth AG (Balgach, Switzerland). All primers used for cloning and sequencing are listed in Supplementary Table S2.

TraB_31-110 was expressed in BL21 star (DE3) and E. coli BL21 CodonPlus (DE3)-RIL cells in LB medium supplemented with ampicillin (100 μg/ml) and in addition with chloramphenicol (32 μg/ml) for E. coli BL21 CodonPlus (DE3)-RIL under shaking with 180 rpm. At an OD₆₂₀ of ∼0.7 TraB_31-110 expression was induced by adding 0.5 mM IPTG. Incubation continued for 16 h at 30°C under shaking. Cells expressing TraB_31-110 were harvested by centrifugation at 9,000 × g for 40 min. The pellet was resuspended in lysis buffer (50 mM Tris/HCl, 150 mM NaCl, 20 mM imidazole, 2% Tween-20, pH 8.0, filtered using a MF-Millipore™ 0.45 µm MCE membrane (Merck, Darmstadt, Germany) containing protease inhibitor (Pierce protease Inhibitor Tablets, Thermo Fisher Scientific, Waltham Massachusetts, United States). After homogenization the cells were disrupted by sonication (50% duty cycle, 60% intensity, Sonopuls HD 2070 ultrasonic probe) on ice four times for 16 min each with 8 min breaks in between. The disrupted cells were centrifuged in a Beckman Coulter Avanti J-26 XP centrifuge (Brea, California, United States) using a JA 25.50 rotor at 38,000 × g for 40 min. The supernatant was filtered through a PVDF Rotilabo^® syringe filter with a pore size of 0.45 µm (Carl Roth GmbH + Co. KG, Karlsruhe, Germany) and loaded onto a 5-ml HisTrap™ HP column (Cytiva, Marlborough, Massachusetts, United States) and purified by an ÄKTA FPLC system. Test purification was monitored on a 12% SDS polyacrylamide gel using Lämmli standard buffer conditions (Laemmli, 1970).

All purification steps were performed on an ÄKTA FPLC system. Elution fractions of the nickel-based immobilized metal affinity chromatography (Ni-IMAC) were pooled and applied to dialysis using a dialysis tube (Spectra/Por^® 3, Repligen, Waltheim, Massachusetts, United States) with a molecular weight (MW) cutoff of 3.5 kDa. TraB_31-110 was cleaved by TEV protease and dialyzed against 50 mM Tris/HCl, 150 mM NaCl and 2 mM DTT for 2 days. Cleaved TraB_31-110 was separated from uncleaved TraB_31-110 and TEV protease by performing a reverse Ni-IMAC using a 5-ml HisTrap™ HP column at a flow rate of 3 ml/min. As last purification step, a size exclusion chromatography (SEC) using a Superdex 200 increase 10/300 (Cytiva) column at a flow rate of 0.3 ml/min was performed.

Polycistronic expression of traB-traO _pIP501 in B. megaterium MS941 was carried out with the B. megaterium/E. coli shuttle vector pMGBm19 containing traB to traO. traB with a C-terminal Strep-tag II and a ribosomal binding site was inserted into the B. megaterium/E. coli shuttle vector pRBBm59 to allow purification of TraB and its interaction partners via affinity chromatography.

traB to traO _pIP501 was amplified using Q5 polymerase and primers pMGBm19-traBO_fw and pMGBm19-traBO_rev. The BamHI/SacI-digested insert was ligated into pMGBm19 cut with the same enzymes and transformed into One Shot™ TOP10 Chemically Competent E. coli cells, resulting in pMGBm19-RBS-traB-traO. Gibson Assembly (Gibson Assembly Kit, New England Biolabs) was used to insert RBS-traB-Strep into pRBBm59. pEU327-RBS-traB-Strep served as template for amplification of the insert. Prior to Gibson Assembly, traB-Strep was amplified with an added stop codon, an artificial RBS (AAAGGGGGAAA) and a BamHI restriction site as spacer on the 5’ end using primers traB-pRBBm59 fw and traB-pRBBm59 rev, resulting in a 389 bp fragment. The amplicon was used as a template to produce the Gibson Assembly insert using the primers GA_traB CStrep prBB fw and GA_traB CStrep prBB rev. The backbone of pRBBm59 was amplified by using GA_pRBBm59 fw and GA_pRBBm59 rev primers and Q5 polymerase, generating a 5,750 bp fragment. The resulting construct was denominated pRBBm59-RBS-traB-Strep. The correctness of all plasmids generated was confirmed by Sanger sequencing. All primers used for cloning and sequencing are listed in Supplementary Table S2.

B. megaterium MS941 was co-transformed with pMBGm19-RBS-traB-traO, containing a xylose-inducible promoter and pRBBm59-RBS-traB-Strep, containing a sucrose-inducible promoter using a protoplast transformation protocol for B. megaterium (Biedendieck et al., 2011). Colonies were selected on LB agar containing 35 μg/ml chloramphenicol and 10 μg/ml tetracycline and were screened for the presence of the plasmids by colony PCR.

Overnight cultures of B. megaterium MS941 (pMGBm19-RBS-traB-traO; pRBBm59-RBS-traB-Strep) were grown in baffled flasks in LB medium supplemented with chloramphenicol (35 μg/ml) and tetracycline (10 μg/ml) at 37°C and 250 rpm. Expression cultures (1 l in 5 l baffled flasks) were inoculated to an OD₆₀₀ of 0.05 in the same medium and incubated until an OD₆₀₀ of 0.3–0.4 at 37°C. Expression of TraB-TraO and Strep-tagged TraB was induced with 0.5% (w/v) xylose and sucrose, respectively, followed by incubation for 4 h. Cells were harvested at 9,000 × g at 4°C, lysed in 70 ml TraB-TraO/TraB-Strep binding buffer (50 mM Hepes, 300 mM NaCl, 1 mM EDTA, pH 7.5, 0.5% Nonidet^® P 40 Substitute (NP-40) Proteomics Grade (AMRESCO, VWR Life Science) filtered through a MF-Millipore 0.45 µm MCE membrane (Merck, Darmstadt, Germany) supplemented with additional NP-40 up to 2% and protease inhibitor (Pierce protease Inhibitor Tablets, Thermo Fisher Scientific, Waltham Massachusetts, United States) using the SONOPULS HD 2070 ultrasonic homogenizer (BANDELIN electronic GmbH & Co. KG, Berlin, Germany) on ice for 60 min at 60% intensity and 50% duty cycle. After 60 min centrifugation at 4°C and 38,000 × g, the supernatant was passed through a 0.45 μm PVDF syringe filter (Merck Chemicals and Life Science GmbH, Vienna, Austria) prior to loading onto a 5-ml StrepTrapTM HP column pre-equilibrated with TraB-TraO/TraB-Strep binding buffer at 4°C using an ÄKTA pure 25 L chromatography system. The protein was eluted in a 60% gradient of TraB-TraO/TraB-Strep elution buffer (50 mM Hepes, 300 mM NaCl, 1 mM EDTA, 2.5 mM desthiobiotin). 2 ml fractions were collected. Peak fractions were loaded onto a 12% SDS polyacrylamide gel and subjected to immunoblotting using polyclonal anti-Tra antibodies directed against TraF, TraG, TraH, TraK, and TraM (Biogenes, Berlin, Germany) at a 1:10,000 dilution to probe for coeluted Tra proteins. As secondary antibody a horseradish peroxidase conjugated antibody against rabbit IgG (Promega GmbH, Mannheim, Germany) was used at a dilution of 1:10,000.

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 180 nm with a data pitch of 0.2 nm and a bandwidth of 2 nm applying a scanning speed of 100 nm/min. Each spectrum was recorded as an average of 20 scans. The resulting spectra were baseline-corrected by subtracting the signal of the buffer. TraB_31-110 was prepared with a concentration of 0.15 mg/ml in 50 mM Tris/HCl, 150 mM NaCl, pH 8.0, 0.05% DDM. The CD signal was converted to mean residue ellipticity [θ] and secondary structure was determined with dichroweb using the CDSSTR algorithm (Compton and Johnson, 1986; Miles et al., 2022).

50 µl samples of 10 µg TraB_31-110 in SEC buffer containing 0/0.001/0.01/0.05/0.1% glutaraldehyde in 50 mM Bicine, 150 mM NaCl, 1 mM DTT were incubated for 20 min at room temperature. Cross-linking was stopped by adding glycine to a final concentration of 140 mM followed by 5 min incubation at room temperature. 400 µl of −20°C-cold acetone was added and the samples were precipitated at −20°C for 2 h, followed by centrifugation at 16,100 × g for 15 min. Pellets were resuspended in 10 µl distilled water and 10 µl SDS-PAGE loading buffer. 15 µl samples were loaded onto a 12% SDS polyacrylamide gel. Unstained protein MW marker (Thermo Fisher Scientific) was used to assess the size of the cross-linked oligomers.

Putative TraB_31-110 bands were excised from a silver-stained 12% SDS polyacrylamide gel and subjected to in-gel reduction, alkylation and trypsinization as described previously (Gesslbauer et al., 2018). Extracted peptides were purified and concentrated using C18 spin columns (Pierce™ C18 Spin Columns; Thermo Fisher Scientific) according to 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 (IonOptics). 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 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 1.6.15.0. MS/MS spectra were searched against the UniProt E. coli (K12) reference proteome databank (Proteome ID: UP000000625) (Cox and Mann, 2008). The TraB_31-110 sequence was manually included in the E. coli 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, and 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).

The following online services were used to search for transmembrane motifs in the TraB sequence and potential orthologous proteins: PSIPRED 4.0 (Buchan and Jones, 2019; Moffat and Jones, 2021), MEMPACK (Nugent et al., 2011), DomPred (Bryson et al., 2007), MEMSAT-SVM (Nugent and Jones, 2009). PSIPRED was used to predict the secondary structure content of TraB and orthologous proteins. After identifying seven TraB orthologs by an iterative PSI-BLAST using the NCBI database, a multiple sequence alignment with T-Coffee was performed using the PSI/TM-Coffee function for membrane proteins (Notredame et al., 2000; Chang et al., 2012). The multiple sequence alignment and the structural categorization were evaluated with the implemented TCS tool (Chang et al., 2014; Chang et al., 2015). General features of the recombinant TraB_31-110 constructs were assessed with ProtParam (Gasteiger et al., 2005). RoseTTAFold was used for 3D structure prediction of TraB and its putative orthologs (Baek et al., 2021). PyMOL Molecular Grahics System, Version 2.5 was used for the presentation of the theoretical structures and their alignment using the method “cealign” (Shindyalov and Bourne, 1998).

Subcellular fractionation of E. faecalis JH2-2 (pIP501), E. faecalis JH2-2 (pIP501∆traB), and E. faecalis JH2-2 (pIP501∆traB, pEU327-RBS-traB-Strep) was performed as described in Goessweiner-Mohr et al. (2013) with some modifications. Cultures were diluted to an OD₆₀₀ of 0.05 in BHI medium containing 20 μg/ml chloramphenicol, for the pEU327 derivative, additionally supplemented by 500 μg/ml spectinomycin and incubated at 37°C until an OD₆₀₀ of 0.5 was reached. The culture was kept on ice for 15 min and centrifuged for 10 min at 4°C at 6,000 × g. The pellet was washed twice with 10 ml 50 mM KH₂PO₄/K₂HPO₄ buffer (50 mM, pH 7) and resuspended in 1 ml lysis buffer (50 mM KH₂PO₄/K₂HPO₄, pH 7, 1 mM EDTA, 1 mM MgCl₂, 100 μg/ml RNAse, 1 U/µl DNAse, 1:1000 diluted protease inhibitor [stock solution: 1 mg/ml pepstatin, 2 mg/ml antipain, 20 mg/ml leupeptin], 1 mg/ml lysozyme). Unlysed material was removed by centrifugation at 1,000 × g for 2 min at 4°C. The supernatant was centrifuged at 16,000 × g for 20 min at 4°C. The resulting pellet corresponding to the cell wall fraction was resuspended in 100 µl potassium phosphate buffer (50 mM KH₂PO₄/K₂HPO₄, pH 7). The supernatant was applied to ultracentrifugation at 163,000 × g for 2 h at 4°C. The resulting pellet representing the membrane fraction was resuspended in 100 µl potassium phosphate buffer (50 mM KH₂PO₄/K₂HPO₄, pH 7) containing 1% Triton X-100. The supernatant containing the cytoplasmic fraction was reduced to a final volume of 100 µl by vacuum centrifugation in a centrivap concentrator (LABCONCO, Kansas City, Missouri, United States). TraB was probed in the cell wall, cell membrane, and cytoplasmic fraction by western blotting using a polyclonal antibody directed against TraB_31-110 (ProteoGenix SAS, Schiltigheim, France) in a dilution of 1:8,000. The same protocol was used as described in (Expression and Purification of Strep-Tagged TraB and its Interaction Partners from B. megaterium MS941 (pMGBm19-RBS- traB-traO; pRBBm59-RBS-traB-Strep) Lysate), with the difference that no detergent was used processing the membrane.

To study the role of the transmembrane protein TraB in pIP501 conjugative transfer, biparental mating assays were conducted with the E. faecalis JH2-2 (pIP501∆traB) deletion mutant as donor and E. faecalis OG1X as recipient. The transfer frequency of the deletion mutant was below the detection limit of the assay (<4.4 × 10⁻⁸ transconjugants per recipient). As traB deletion resulted in a total loss of plasmid transfer, TraB proved to be essential for pIP501-mediated conjugative transfer (Figure 1).

Complementation of the traB knockout by supplying the wild type traB gene in trans showed full recovery of transfer with transfer frequencies close to wild type level thus excluding polar effects of the deletion on downstream tra genes. Similar transfer rates were observed, when supplying the wild type traB gene with a C-terminal Strep-tag (traB-Strep) in trans. Thus, the Strep-tag did not impact the functionality of TraB. Complementation of the deletion mutant by supplying a truncated traB _31-110 variant lacking the N-terminal secretion signal sequence in trans resulted in a significantly diminished transfer rate in comparison to the wild type pointing to a crucial role of the signal sequence as correct localization of TraB is substantial in the transfer process.

Combining the secondary structure prediction of PSIPRED and the analysis with MEMPACK, MEMSAT-SVM and SignalP5.0 a scheme of the TraB domain composition was built (Nugent and Jones, 2009; Nugent et al., 2011; Almagro Armenteros et al., 2019; Buchan and Jones, 2019; Moffat and Jones, 2021) (Figure 2). Although TraB is predicted to contain three α-helices, the mature TraB_31-110 only contains two membrane spanning helices as the first helix spanning from lysine 2 to phenylalanine 23 belongs to a secretion signal sequence, which is cleaved off between alanine 30 and alanine 31.

The first transmembrane helix (TMH1) starts at proline 34 and stops at proline 70. It is predicted to enter the membrane at asparagine 43 and leaves the membrane at isoleucine 64. Lysine 67 might interact with the phospholipid headgroup and tyrosine 68 as a membrane anchor. The loop connecting TMH1 and TMH2 is facing into the cytosol. TMH2 starts with proline 75, is predicted to enter the cell membrane at leucine 85 and to leave it at tyrosine 106. Tryptophane 104 probably serves as membrane anchor. The possible function of some particular amino acids will be discussed in more detail in the following chapters.

Subcellular localization by cell fractionation revealed that TraB_pIP501 is found in the cell wall fraction and in the cell membrane fraction. This has been shown for E. faecalis JH2-2 (pIP501) as well as for the complementation strain E. faecalis JH2-2 (pIP501∆traB, pEU327-RBS-traB-Strep). The traB deletion strain E. faecalis JH2-2 (pIP501∆traB) served as negative control. The Western Blot is shown in Figure 2, panel C.

We used the NCBI database to search for TraB_pIP501 orthologs. The results are shown in Table 1. Interestingly, TraB orthologs were detected on numerous bacterial strains belonging to the phylum Firmicutes, including nosocomial pathogens such as MRSA, S. aureus, E. faecalis as well as biotechnologically relevant bacteria such as Lactococcus lactis and environmental species like Clostridium algoriphilum. Their identities range from 42 to 100% identity on amino acid level. All traB orthologs were found in gene clusters encoding a putative T4SS, several of them on conjugative plasmids, namely on the broad-host-range inc18 plasmids, pRE25 and pAMβ1 (TraB orthologs with 100% identity on amino acid level; the other Tra orthologs show identities in the range of 74–100%) from E. faecalis, on the pSK41/pGO1-family plasmid pV030-8 from S. aureus and on the pNP40-like plasmid pUC11B from Lactococcus lactis (Ortiz Charneco et al., 2021). All of them are small proteins containing 101 to 114 amino acids in their unprocessed form. Their mature form is even more similar regarding their size ranging from 80 to 83 amino acids.

Figure 2, panel B shows a multiple sequence alignment generated by PSI/TM-Coffee of TraB_pIP501 and its orthologs presented in Table 1. The accession numbers of the respective proteins are given. Table 1 gives both, the accession number, and the name of the respective gene product. The multiple sequence alignment suggests which parts of the proteins are inserted into the membrane. All TraB orthologs show the same pattern according to their insertion into the membrane, with the N- and C-terminal region pointing to the extracellular space and a loop region connecting two transmembrane helices facing into the cytoplasm (note: The amino acids 2–23 of TraB comprise the signal sequence). The evaluation with the in the T-Coffee package implemented calculation of the transitive consistency score (TCS) shows a high score for every protein sequence, rendering the prediction and the multiple sequence alignment reliable (Chang et al., 2014; Chang et al., 2015). The mating channel formation protein from pUC11B shows the lowest overall identity with TraB_pIP501, which is also reflected by the TCS evaluation. The largest sequence differences are located at the N-terminal part containing the secretion signal sequences. The best-conserved region is the part facing into the cytoplasm, although it has a difference of two amino acids in length. The second transmembrane helix of TraB from Staphylococcus aureus 3688STDY6124879 and the CagC family T4SS protein of Clostridium algoriphilum DSM 16153 seem to be longer than the second transmembrane helix of the other ones.

To obtain a more detailed picture of the 3D structure of the membrane-inserted TraB protein and its orthologs we generated structural models using the Robetta server (https://robetta.bakerlab.org). The recently available deep learning-based modeling method RoseTTAFold enables the ab-initio generation of structural models, which are quite accurate and do not require known homologous structures (Baek et al., 2021). The structures of TraB and the seven orthologs showed a very similar overall structure with the 2 TM helices arranged in an antiparallel manner and connected by a short and structurally defined loop. The signal helix, which is connected to TMH-1 via a longer, flexible loop exhibited the largest deviations in the alignment shown in Figure 3, panel A. The pairwise alignment to TraB_pIP501 was performed with the alignment function using the cealign algorithm of PyMOL 2.5 and yielded r.m.s.-deviations between 2.17 Å for the MPF protein from pUC11B over 80 atoms and 4.62 Å for TraB from Staphylococcus aureus 3688STDY6124879 over 104 atoms (Schrödinger and DeLano, 2020), PyMOL was retrieved from http://www.pymol.org/pymol. After removal of the secretion signal sequence the mature TraB protein consists of the two antiparallel TM helices, with a short flexible stretch at the N-terminal side of TMH-1 and the C-terminal end of TMH-2. The length of the two helices differs slightly (55 Å for TMH-1 and 50 Å for TMH-2) and matches quite well the secondary structure prediction and the expected length of a membrane spanning helix (Phillips, 2018). TMH-1 is straight, whereas TMH-2 features a pronounced kink in its C-terminal part, which is due to a conserved proline residue (P103 in TraB_pIP501). The two TMHs are connected by a tight loop (P70 to P75), which constitutes one of the highest conserved regions in the multiple sequence alignment (Figure 3B, panel B).

The TM helices exhibit a distinct distribution of hydrophilic and hydrophobic residues: in the middle part (colored in salmon in Figure 3, panel B, C and D) there are mainly hydrophobic amino acids and no charged ones; the outer and inner parts (slate and yellow in Figure 3, panel B, C and D) consist of mainly hydrophilic residues, a large portion of which is charged. Both the outer and the inner parts carry a net positive charge, which is more pronounced for the inner part—including the TMH connecting loop, there are 5 lysines and 3 acidic residues (2 aspartates and 1 glutamate) in this region. The lysines are oriented closer to the central part of the TMHs consistent with their proposed membrane anchoring function (Landolt-Marticorena et al., 1993; Planque et al., 1999). The C-terminal part of TMH-2 has distinct features as it contains no charged amino acids but 3 aromatic residues, 2 tryptophanes and 1 tyrosine, which are clustered around the kink caused by P103. The two tryptophanes (W100 and W104 in TraB_pIP501) are conserved among all compared TraB-like proteins and most likely play an important role in membrane anchoring (Landolt-Marticorena et al., 1993; Planque et al., 1999) and/or protein-protein interaction within the MPF complex. The completely conserved residues including the two tryptophanes are shown in Figure 3, panel B. For better visualization, we zoomed into the regions containing the conserved residues and labeled them referring to their position within TraB_pIP501. This is shown in Figure 3, panel C for the outwards facing parts of TraB_pIP501 and in panel D for the cytosolic loop region.

TraB_31-110 encoded on pQTEV-traB _31-110 was expressed in two different E. coli expression strains, BL21-CodonPlus (DE3)-RIL and BL21 star (DE3). A test purification, where the protein was isolated by a Ni-IMAC was performed to evaluate expression levels. As the isolated protein migrated slower than expected for its calculated MW, we performed MS analysis confirming that the band at 16 kDa corresponds to TraB_31-110. The examined bands are shown in Figure 4, panel A. As the TraB_31-110 band was more prominent using BL21-CodonPlus (DE3)-RIL cells for expression, they were used for large-scale expressions and purification. Figure 4, panel B shows the different purification steps of TraB_31-110. Lanes 1 and 2 show the pooled elution fractions from the Ni-IMAC step before and after dialysis and TEV cleavage, respectively. TEV cleavage was performed for 2 days, because detergents might compromise the cleavage efficiency of the TEV protease. This appeared to be the case for the used Tween-20 as only about 50% of TraB_31-110 was cleaved even after 2 days. Next, a reverse Ni-IMAC was performed to separate TEV-cleaved TraB_31-110 from uncleaved TraB_31-110 as well as from the TEV protease. Lanes 3 and 4 show the elution fractions of the reverse Ni-IMAC, containing impurities binding nonspecifically to the column, uncleaved TraB_31-110 and TEV-protease. The flow through was collected, concentrated, and loaded on to a Superdex 200 increase column to obtain TraB_31-110 of higher purity (lanes 6–10 of Figure 4, panel B). Lane 6 represents the peak eluting in the void volume of the column, containing highly aggregated impurities. TraB_31-110 eluted as a single peak at an elution volume of 12.3 ml (lanes 7–10) corresponding to a calculated MW of 172 kDa. This represents the micelle bound TraB_31-110. For further analysis like in vitro cross-linking or CD measurements only the main fractions containing TraB_31-110 in the highest purity were used. The CD spectrum in Figure 4, panel C indicates mainly α-helical content and no β-strand which agrees with the structural features obtained from the secondary structure predictions of TraB_31-110 (see TraB is an α-Helical Transmembrane Protein).

The results of the in vitro cross-linking are shown in Figure 5. At a glutaraldehyde concentration of 0.01% a dimer and trimer band become clearly visible, and a faint band of a tetramer appears. At a crosslinker concentration of 0.05% the distribution is shifted towards the trimer and tetramer. At the highest glutaraldehyde concentration (0.1%) unspecific cross-linking occurred leading to an indistinguishable variety of aggregates of different sizes. These results indicate that TraB_31-110 forms oligomers in Tween-20 micelles in a range from dimers to tetramers.

As a predicted membrane protein, TraB is a likely candidate to interact with other T4SS proteins to assemble the translocation channel. A pull-down approach was used to elucidate the TraB interaction partners. The co-transformed B. megaterium MS941 strain expressed both, C-terminally Strep-tagged TraB (TraB-Strep) and the tra _pIP501 operon proteins TraB to TraO. The C-terminal Strep-tag does not impact the functionality of TraB, as demonstrated by wild type transfer frequencies for the pIP501 traB knockout strain complemented with TraB containing a C-terminal Strep-tag (E. faecalis JH2-2 (pIP501∆traB, pEU327-RBS-traB-Strep) (Figure 1). TraB-Strep and interacting proteins were purified from the clarified supernatant of B. megaterium MS941 (pMGBM19-RBS-traB-traO; pRBBm59-RBS-traB-Strep) lysate via affinity chromatography. The putative MPF proteins TraF, TraH, TraM and TraK coeluted with TraB-Strep and were detected by immunoblotting (Figure 6). The cytosolic protein TraN and the putative surface adhesion protein TraO were used as negative controls. None of them was coeluted, as depicted for TraO (Figure 6). The two bands visible on the membrane incubated with anti-TraK antibody are due to two different translational start codons for TraK, resulting in the expression of two TraK variants differing by 4.1 kDa in size (Goessweiner-Mohr et al., 2014b). The two bands representing TraO are probably showing two differently processed forms. Unprocessed TraO has a MW of 29.9 kDa containing a putative secretion signal sequence with a cleavage site between alanine 24 and aspartic acid 25. Secreted TraO shows a MW of 27.4 kDa. The membrane incubated with anti-TraG antibody showed a band at around 20 kDa, which migrated significantly faster than the calculated MW of 40.4 kDa for TraG. This could mean that only the CHAP domain of TraG is part of the isolated protein complex as the band would correspond to its calculated MW of 22 kDa (Arends et al., 2013).