Structural and biochemical characterization of the relaxosome auxiliary proteins encoded on the Bacillus subtilis plasmid pLS20.

Bacterial conjugation is an important route for horizontal gene transfer. The initial step in this process involves a macromolecular protein-DNA complex called the relaxosome, which in plasmids consists of the origin of transfer (oriT) and several proteins that prepare the transfer. The relaxosome protein named relaxase introduces a nick in one of the strands of the oriT to initiate the process. Additional relaxosome proteins can exist. Recently, several relaxosome proteins encoded on the Bacillus subtilis plasmid pLS20 were identified, including the relaxase, named RelpLS20, and two auxiliary DNA-binding factors, named Aux1pLS20 and Aux2pLS20. Here, we extend this characterization in order to define their function. We present the low-resolution SAXS envelope of the Aux1pLS20 and the atomic X-ray structure of the C-terminal domain of Aux2pLS20. We also study the interactions between the auxiliary proteins and the full-length RelpLS20, as well as its separate domains. The results show that the quaternary structure of the auxiliary protein Aux1pLS20 involves a tetramer, as previously determined. The crystal structure of the C-terminal domain of Aux2pLS20 shows that it forms a tetramer and suggests that it is an analog of TraMpF of plasmid F. This is the first evidence of the existence of a TraMpF analog in gram positive conjugative systems, although, unlike other TraMpF analogs, Aux2pLS20 does not interact with the relaxase. Aux1pLS20 interacts with the C-terminal domain, but not the N-terminal domain, of the relaxase RelpLS20. Thus, the pLS20 relaxosome exhibits some unique features despite the apparent similarity to some well-studied G- conjugation systems.

Introduction

Horizontal gene transfer (HGT) is the exchange of genes between organisms not related to transmission of genes between parents and offspring and has important implications on evolution. Conjugation is one of the routes that allow HGT and is common in the bacterial realm [1], [2], [3], [4], [5], [6]. A membrane-embedded cellular machinery named the Type 4 Secretion System (T4SS) is used in conjugation for the transfer of a conjugative element from a donor to a recipient cell [7], [8], [9], [10], [11]. The relaxosome is another important component of T4SS-mediated conjugation and prepares the DNA for transfer [5], [12]. Conjugation also plays an important role in bacterial virulence, as the T4SS system is also used for the extrusion of virulence factor into the host environment (eg [13], [14]). Bacterial T4SS systems occur both in Gram positive (G+) and Gram negative (G-) bacteria and may be encoded in Mobile Genetic Elements (MGEs) such as plasmids [6].

Conjugation requires the initial preparation of the DNA that is to be transferred [5], [10], [11], [15]. This step involves an initial cut or nick at a specific site of one of the DNA strands, which then allows for the unwinding of this strand and subsequent transfer [5]. This critical step is performed by the relaxase [16], [17], [18], which have in common that they consist of an N-terminal endonuclease domain, followed by additional domains with a variety of functions [18], [19], [20], [21]. The endonuclease domain of the relaxase remains covalently attached to the ssDNA strand and this complex is transferred to the recipient cell. The process is strictly regulated at various levels, which includes the implication of auxiliary proteins that form part of the relaxosome complex [5], [22]. The auxiliary factors can be plasmid encoded or provided by the donor. In most cases, the auxiliary proteins are either indispensable or enhance conjugation frequency. The functions of the auxiliary proteins include specific DNA binding on the relaxosome, connecting the relaxosome with the T4SS channel, and interacting with the relaxase.

Recently, the relaxase of the Bacillus subtilis plasmid pLS20 was identified [23], which was named RelpLS20. Sequence comparison showed that the N-terminal domain of RelpLS20 contains sequence motifs corresponding to relaxase domains and was marked as the founding member of a new class of relaxases, named MOBL [23]. However, other authors suggest that this relaxase could also be classified as belonging to the MOBP subfamily, MOBMG [19], [24], [25]. In addition to the relaxase, two auxiliary proteins of plasmid pLS20 have been identified, which were named Aux1pLS20 and Aux2pLS20 [26]. They are encoded immediately upstream of RelpLS20 and were predicted to be Ribbon-Helix-Helix (RHH) proteins based on the occurrence of a β-strand followed by two α-helices at their N termini. RHH proteins are transcription factors that have a range of regulatory functions in prokaryotes and bacteriophages and have an interlaced, inseparable dimeric arrangement [27]. The well-studied Salmonella phage protein Arc is the founding member of this family.

It was shown that the pLS20 auxiliary RHH proteins are essential for pLS20 conjugation and that they bind to specific sites of the oriTpLS20. Aux1pLS20 binds to a 25 bp region, which contained an inverted repeat with sequence 5′-TGGTACCA-3′, which was proposed to be its binding site. Aux2pLS20 is able to bind to a much larger region of the oriT spanning several hundreds of bases. To explain this, it was proposed that Aux2pLS20 binds the TGTGCAT sequence which was the only sequence to be present three times in the oriT [26]. The oligomerization state of these proteins was studied by analytical ultracentrifugation (AUC) and was proposed to be tetrameric for Aux1pLS20 and hexameric for Aux2pLS20, although tetrameric forms were compatible with the sedimentation profile as well [26]. This study also used AUC to probe for interactions between the different relaxosome proteins, but none could be identified.

The exact function of the auxiliary proteins in the conjugative process of pLS20 is unknown. To gain knowledge on the function of these proteins, we characterize Aux1pLS20 and Aux2pLS20 using various biophysical techniques and have investigated the structures of these proteins. We present the low-resolution, SAXS envelope of the Aux1pLS20 and the crystal structure of the C-terminal domain of Aux2pLS20 at atomic resolution. In addition, we also study the interactions between the auxiliary proteins and the relaxase RelpLS20. Our results gave new insights into the relaxosome of pLS20, which has implications on similar G+ conjugative systems. Thus, we show that the C-terminal domain of Aux2pLS20 forms a four-helix bundle and with structural homology to TraM of plasmid F, thereby providing the first evidence of the existence of a TraMpF analog in G+ conjugative systems. Unlike other TraMpF analogs, Aux2pLS20 does not interact with the relaxase. Surprisingly, Aux1pLS20 did show an interaction with the C-terminal domain of RelpLS20. We discuss the implications of the results for the mechanism of relaxosome preparation and presentation to the T4SS channel.

Methods

Protein expression and purification

All protein constructs were expressed and purified using standard protocols. The cDNAs expressing the protein constructs were Aux1pLS20 without tags (Mw 9.02 kDa), Aux1pLS20 with C-terminal His tag (Mw 10.6 kDa), Aux2pLS20 (Mw 17.03 kDa), full length construct (Mw 48.75 kDa) and the C-terminal (residues 180–410, Mw 28.27 kDa) and N-terminal domain (residues 1–232, Mw 26.92 kDa) of RelpLS20. The pET28-based clone of Aux1pLS20 with C-terminal His tag was obtained as described in [26]. All the other cDNAs were chemically synthesized, adding an N-terminal HRV-3C protease cleavage site before the methionine codon of each protein construct. The cDNA synthesis and subcloning into pHTP1 vectors were purchased from NZYTech (Campus do Lumiar, Lisboa, Portugal). Escherichia coli BL21 (DE3) were transformed with pHTP1 carrying the insert of the different constructs of aux1, aux2, rel, relC-ter and relN-ter and were inoculated in fresh Luria-Bertani broth (LB) media complemented with 50 µg/ml kanamycin at 37 °C overnight. Then, the cells from the overnight culture were collected by centrifugation (4000 × g for 10 min) and suspended in expression media (typically 1 L of Terrific Broth (TB) with 50 µg/ml kanamycin), at a ratio of 15 ml of preculture per liter of medium. Cells in expression media were grown at 37 °C until an OD600 = 0.8–1 was reached. After that, protein expression was induced overnight at 20 °C by addition of 1 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG, Omnipur). After overnight induction, cells were centrifuged at 4000×g for 30 min, and pellets were frozen and stored at −80 °C. Before purification, the pellets were thawed and resuspended in lysis buffer at a ratio of 5 ml/g of cells. The lysis buffer contained 0.5 M NaCl, 50 mM Tris pH 7.5, 5 mM imidazole, 1 mM EDTA, 5% glycerol, and 0.2 mM of PMSF (phenylmethylsulfonyl fluoride) as protease inhibitor. The cell suspension was lysed by sonication, adding DNase I to a final concentration of 200 µg/ml and lysozyme to a final concentration of 100 µg/ml during sonication. Insoluble matter was precipitated by centrifugation (18 000×g, 30 min), and the supernatant was filtered through a 0.22 µm filter and applied to a nickel-charged His-Trap™ HP chelating column 5 ml (GE Healthcare Life Sciences). The column was washed with 10 column volumes (50 ml) of binding buffer (20 mM Tris pH 8.0, 500 mM NaCl, 5 mM Imidazole) to elute unspecific bound proteins. Bound proteins were eluted using sequential steps of 4, 10 and 25% of elution buffer (0.5 M NaCl, 50 mM Tris pH 7.5, 0.5 M imidazole). Fractions were analyzed on SDS-PAGE and fractions containing pure protein were combined and concentrated over an Amicon ultra 10 kDa MWCO (Millipore). The buffer was exchanged using three HiTrap 5 ml Desalting columns (Cytiva) connected in series to 20 mM Tris pH 8.0, 500 mM NaCl. After elution, proteins were incubated overnight at 4 °C with 1 mg of HRV-3C protease (in-house preparation). After incubation, HRV-3C, tags and uncut protein were removed by application of the incubation solution through a His-Trap™ HP chelating column. The flow-through was concentrated and further purified by Size Exclusion Chromatography to remove aggregates and other contaminants. The SEC column (Generon ProSEC 16/60 3–70 HR) was equilibrated in 20 mM Tris pH 8.0, 500 mM NaCl. Fractions containing the target protein were concentrated over an Amicon ultra 10 kDa MWCO (Millipore). Typically, a yield of ∼20 mg of Aux1pLS20, ∼10 mg for Aux2pLS20, ∼100 mg for full length RelpLS20, ∼20 mg for RelN-terpLS20 and RelC-terpLS20 were obtained from 10 g of pellet. Purity was assessed to be >95% by SDS–PAGE, followed by BlueSafe staining (NZYTech, Campus do Lumiar, Lisboa, Portugal). The protein concentration was determined by nanodrop, using the theoretical extinction coefficient to calculate the concentrations, immediately prior to usage where possible. Surplus aliquots were stored at −80 °C until usage.

Analytical size exclusion chromatography on Aux1pLS20, Aux2pLS20 and RelpLS20 mixtures

To determine the elution volumes of the separate proteins, 1.2 nmol of either full length RelpLS20, RelC-terpLS20 and RelN-terpLS20 were used for all the interaction assays. For complex-binding stoichiometry assays, molar ratios of 1:1, 5:1 and 10:1 were prepared for Aux1pLS20:RelpLS20, Aux1pLS20:RelN-terpLS20, Aux1pLS20:RelC-terpLS20 and Aux2pLS20:Aux1pLS20. All samples were incubated for 30 min on ice before injection. Thus, the maximum amount of Aux1pLS20 and Aux2pLS20 injected was 12 nmol in the 1:10 M ratio. Samples of the proteins alone and in complex were prepared in a final volume of 30 µL in 20 mM Tris pH 8, 300 mM NaCl and 25 µL were injected on a Superdex 200 increase 5/150 GL equilibrated with 20 mM Tris pH 8, 300 mM NaCl. The elution was run at a flow rate of 0.3 ml/min and the absorption of the elute was monitored at 260 and 280 nm. To estimate the molecular weights (MW) of the homo- and heterocomplexes, a calibration of the Superdex 200 increase 5/150 GL column was performed using proteins with known MW (i.e. BLC (239.64 KDa), RCO (81.28 KDa), BSA (69.29 KDa), MpARF3 (49.52 KDa), MpARF2 (44.63 KDa), AtARF5 (44.29 KDa), AtARF1 (41.13 KDa) and p69 (24.1 KDa), which were eluted using the same elution buffer used above. The derived relation between the elution volume (Vel) and MW was Vel = –0.6815 · log(Mw) + 5.1906, with an R2 = 0.933.

Isothermal titration calorimetry (ITC)

Calorimetric measurements were carried out using a VP-ITC instrument from MicroCal Inc. (Northampton, USA), in the Polymorphism and Calorimetry Unit of the Scientific and Technological Centers of the University of Barcelona. In the experiment, 1.4 ml of RelpLS20 solution at 87.2 μM (4.25 mg/ml) were titrated with 300 µL of Aux1pLS20 at 2770 μM (25 mg/ml) at 298 K. The reference cell was filled with double deionized water (ddH2O). All measurements were carried out in 20 mM Tris-HCl pH 8, 300 mM NaCl. The buffer solution and ddH2O were degassed at room temperature with stirring under vacuum for ≥30 min. Upon experimental setup, the RelpLS20 solution present in the sample cell was stirred at 300 rpm. The titration was initiated at 298 K after a stable baseline was achieved, with an initial injection of 2 µL of Aux1pLS20 during 4 s. The initial injection was followed by 29 injections of 10 µL during 20 s, each spaced by 300 s.

The calorimetric signal was integrated to obtain the enthalpy changes caused by complexation of RelpLS20 with Aux1pLS20. Data were analyzed using the software Origin 7.0 (January 2004) and fitted to one single-site binding model, subtracting the average heat of the last three measurements after saturation of the RelLS20 binding sites to correct experimental heats of the RelpLS20-Aux1pLS20 dilution. The enthalpy (), entropy (), binding constant () and number of binding sites () resulted from nonlinear least square data fitting. Dissociation constant () was calculated as the inverse of Kb and the standard Gibbs free energy value () was calculated using the equation:

where T = 298 K and ΔH0 and ΔS0 are the thermodynamic values resulting from experimental data fitting at the same temperature. The c value of the assay (c = 7.56) was calculated as , indicating a good shape of the binding isotherm to calculate accurate thermodynamic parameters from the calorimetry data [28].

SAXS experiments on Aux1pLS20

SAXS experiments have been performed on the NCD-SWEET beamline at the synchrotron ALBA (Cerdanyola del Vallès, Barcelona, Spain) at 12.4 keV on Aux1pLS20 with C-terminal His tag. The data were collected on a Pilatus 1 M detector (with a pixel size of 172.0 × 172.0 μm2). The distance sample/detector was 2556.00 mm. 40 images were collected for each concentration (1.0 mg/ml, 2.5 mg/ml, 5.0 mg/ml, 7.5 mg/ml and 10 mg/ml) with an exposure time of 0.1 s. The q-axis calibration was obtained by measuring silver behenate [29]. The program pyFAI [30] was used to integrate the 2D data into 1D data. The 1D data has been averaged, subtracted, normalized by the concentration, extrapolated to zero concentration and merged with Primus [31] from the ATSAS package. The radius of gyration Rg and the maximum distance Dmax has been determined with GNOM [32]. The low-resolution envelop has been restored with DAMMIF [33].

Aux2pLS20 crystallization and structure solution

Purified Aux2pLS20 was concentrated to a final protein concentration of 10 mg/ml in 20 mM Tris (pH 8.0), 300 mM NaCl. Crystallization experiments of Aux2pLS20 were performed using the sitting-drop vapor-diffusion method at 18 °C, by equilibration of drops of 1 μL protein + 1 μL crystallization buffer against 100 μL of the crystallization buffer. Crystals were harvested from two different crystallization conditions. Cryo-cooling in liquid nitrogen was performed by soaking crystals on a cryo-protecting solution consisting in reservoir buffer complemented with 10% glycerol, followed by direct plunge-freezing in liquid nitrogen. Data was collected on two crystal systems on 21/09/2021 at the BL13-XALOC beamline of the ALBA synchrotron Light Source [34], [35], using MXCuBE software for data collection [36]. Crystals of one system belonged to the space group P4212 and were grown using 0.2 M Magnesium chloride hexahydrate and 20% w/v Polyethylene glycol 3,350 as crystallization buffer. Data for this system was processed to a resolution of 1.76 Å, with two protein monomers in the asymmetric unit. We will refer to this crystal structure as Aux2-S. Crystals of the other system belonged to space group P43212 and were obtained using 0.06 M D-Glucose, 0.06 M D-Mannose, 0.06 M D-Galactose, 0.06 M L-Fucose, 0.06 M D-Xylose, 0.06 M N-Acetyl-D-Glucosamine, 0.06 M Tris-Bicine pH 8.5, 20% v/v Ethylene glycol, 10 % w/v PEG 8000 as crystallization buffer. The data of this system was processed to a resolution of 1.89 Å with two protein tetramers in the asymmetric unit. We will refer to this crystal structure as Aux2-L. Data were processed with Autoproc v1.0.5 [37], [38], [39], [40], [41], [42]. See Table 1 for further statistics.

Table 1

Summary of the data processing and refinement statistics of the crystallographic analysis of the Aux2pLS20 structures.

	Aux2-S	Aux2-L
Data collection
Beamline	XALOC (ALBA)	XALOC (ALBA)
PDB code	7NUV	7QNQ
λ (Å)	0.9793	0.9793
Space group	P 4 21 2	P 43 21 2
Unit cell parameters (Å)	a = 44.840, b = 44.840, c = 141.153	a = 37.854, b = 137.854, c = 121.709
Resolution range (Å) a	47.05–1.759 (1.874 – 1.759)	97.48–1.892 (1.964 – 1.892)
# of reflections:
total	147,161 (5833)	2,316,344 (116090)
unique	12,790 (641)	87,560 (4373)
Completeness ellipsoidal (%)	90.2 (39.8)	96.1 (57.4)
<Ι / σ(Ι)>	10.0	14.3 (1.5)
Average multiplicity	11.5	26.5 (26.5)
Rmerge (%) b	36.5	0.19 (3.15)
Rmeas (%) c	38.3	0.20 (3.1)
CC(1/2) (%)	96.8	0.998 (0.522)
Structure Refinement
Rcrystd /Rfreee (%)	20.61 / 24.17	18.0 / 20.7
r.m.s.deviation from target values:
Bond lengths (Å)	0.007	0.005
Bond angle distances (Å)	0.912	0.662
Molprobity scores:
Clashscore (‰)	1.92	0.82
Poor rotamers (%)	0.85	0.36
Ramachandran Outliers (%)	0.00	0
Ramachandran Favoured (%)	99.21	100
Overall score (Å)	0.96	0.75
Isotropic B factor analysis
Average model B-factors (Å2)	27.0	44.16
B-factor from Wilson plot (Å2)	18.0	48.00

Throughout the table, the values in parentheses are for the outermost resolution shell.

Rmerge = Σh | Îh – Ih,i | / ΣhΣi Ih,i , where Îh = (1/nh) Σi Ih,i and nh is the number of times a reflection is measured.

Rmeas = [Σh (nh/[nh-1])½ Σi | Îh – Ih,i |] / ΣhΣi Ih,i , where Îh = (1/nh) Σi Ih,i and nh is the number of times a reflection is measured.

Rcryst = Σhkl | |Fobs| - k |Fcalc| | / Σhkl |Fobs|

Rfree = Σhkl⊂T | |Fobs| - k |Fcalc| | / Σhkl⊂T |Fobs| where T represents a test set comprising ∼ 5% of all reflections excluded during refinement.

To solve the structure of Aux2pLS20, a search for four alpha helices was performed using the program Arcimboldo lite [43]. The size of the helices was adjusted according to the secondary structure predictions computed with PSIPRED [44]. The initial model was built by auto-tracing in the output map with Buccaneer [45]. The structure was completed through alternate manual model building with Coot v0.8.9 [46] and refinement with PHENIX v1.9.2–4158 [47]. The model was validated and further adjusted and refined using MolProbity [48]. The crystallographic and refinement parameters are given in Table 1. Figures were prepared using PyMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC.)

Results

The relaxosome of pLS20 was shown to involve two RHH proteins [26], in addition to the recently identified relaxase, RelpLS20 and the oriT region [23]. To gain more insight into the role of these proteins in the relaxosome formation and their structures, we have performed biophysical and structural studies of the relaxosome proteins.

As a first step, we estimated the molecular weight (MW) of the pLS20 relaxosome proteins in solution using Size Exclusion Chromatography (SEC). The elution volumes of the SEC profiles of the individual purified proteins (Suppl. Fig. 1) were used to calculate the MW based on a calibration of the columns using standard proteins with known MW. The results of the MW estimation are shown in Table 2. The MW of the full length relaxase (RelpLS20), its N-terminal domain (RelN-terpLS20) and its C-terminal domain (RelC-terpLS20), respectively, correspond well to the theoretical values of a monomer, indicating that the protein is globular and monomeric in solution in the absence of other interaction partners. However, the estimated MW of Aux1pLS20 and Aux2pLS20 SEC both correspond to a pentamer and therefore deviate from the expected values based on the fact they are RHH proteins.

Table 2

Estimated molecular weights (MW) of the relaxosome proteins of pLS20.

Protein	Vel (ml)	Mw, monomer (kDa)	Estimated Mw based on Vel (kDa)	Calculated oligomerization state	Predicted oligomerization state
Aux1pLS20	2.05	9.02	46.94	Pentamer (5.20)	Tetramer
Aux2pLS20	1.82	17.03	85.78	Pentamer (5.03)	Tetramer
RelpLS20N-ter	2.26	26.92	27.06	Monomer (1.01)	Monomer
RelpLS20C-ter	2.21	28.27	30.85	Monomer (1.07)	Monomer
RelpLS20	2.10	48.75	41.17	Monomer (0.84)	Monomer

We then set out to map the interactions between respective pairs of the three pLS20 encoded proteins that have been shown to be essential for conjugation: Aux1pLS20, Aux2pLS20, and RelpLS20 [23], [26], performing SEC analysis with mixtures of Aux1pLS20, Aux2pLS20, full length RelpLS20, and the N- and C- terminal domains of RelpLS20. We found that Aux1pLS20 interacts with the full-length RelpLS20 and RelC-terpLS20 (Fig. 1 A-C), as evidenced by the appearance of new peaks with a Vel of 1.97 ml for the Aux1pLS20/RelpLS20 mixture and with a Vel of 2.00 ml for the Aux1pLS20/RelC-terpLS20 mixture. In contrast, no additional peak appeared in mixtures of Aux1pLS20 and RelN-terpLS20 (Suppl. Fig. 2). No other interactions were found between the different constructs of these three proteins (Fig. 1C and Suppl. Fig. 2). To confirm the Aux1pLS20/RelpLS20 interaction and to determine its strength, we measured the thermodynamic parameters of the interaction in solution using isothermal titration calorimetry (ITC, Fig. 1D). The Kd determined by the ITC measurements was 29.4 μM. The ITC results indicate that the interaction was favorable in terms of enthalpy and entropy, with a calculated ΔG0 of −6180.92 kJ mol−1, mainly driven by the exothermic component of the binding, with a ΔH0 of −5722 ± 120.4 kJ mol−1 and a ΔS0 of 1.54 kJ mol−1. Both parameters suggest that the interaction between Aux1pLS20 and RelpLS20 is strongly affected by hydrogen bond formation and van der Waals interactions [49]. The entropy component, although small, may be related to the burial of water-accessible surface area upon binding, resulting in the release of interfacial water molecules to the solvent, contributing favorably to the total entropy of interaction [50].

Fig. 1

In vitro interactions between relaxosome proteins. A) The SEC elution profile of RelpLS20:Aux1pLS20 mixtures at different stoichiometries. B) the SEC elution profile of RelC-terpLS20:Aux1pLS20 mixtures at different stoichiometries. C) The SEC elution profile of RelpLS20:Aux2pLS20 mixtures at different stoichiometries. In panels A-C, the green, orange and purple lines represent the profile of mixtures at 1:1, 5:1 and 10:1 M ratio, respectively, the blue line in A-C marks the peak of the corresponding RelpLS20 domain, the red line in all plots represents the Aux1pLS20 or Aux2pLS20 and the purple line marks the peak of the complex, D) Thermogram of an Aux1pLS20:RelpLS20 titration as determined by isothermal titration calorimetry. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

SAXS on Aux1pLS20

Aux1pLS20 is an RHH protein and is predicted to consist of 1 β-strand followed by three α-helices (Fig. 2A). The occurrence of the first sheet and two consecutive α-helices is consistent with the RHH motif. However, the third α-helix is an addition to RHH proteins, as it is predicted to span 50 amino acids. To probe the effect of this helix on the overall structure of Aux1pLS20, we performed SAXS measurements at different concentrations (Suppl Fig. 3). The data are consistent with a MW of 38.9 kDa, which corresponds well with the expected MW of 44.0 kDa for a tetramer, and is in accordance with AUC data [26]. The radius of gyration (Rg) was determined to be 34 Å, the maximum dimension (Dmax) as 130 Å and the Porod volume (VPorod) as 89 Å3. Reconstruction of the envelop using DAMMIN resulted in an elongated overall shape (Fig. 2B). Although the overall shape is overestimated by the SAXS analysis, it is consistent with the tetramer structure of the Arc RHH structure bound to DNA (PDB code 1PAR, see Fig. 2C) and the tetrameric configuration determined previously using analytical ultracentrifugation [26]. The Rg, Rmax and VPorod are also slightly overestimated, which is likely related with the formation of higher oligomerization states, which has also been observed in the SEC analysis, where a second species with lower elution volume appears (Suppl. Fig. 1). There is additional density in the SAXS volume compared to Arc, indicating that Aux1pLS20 forms a triangle instead of a tubular structure. The tip of this triangle is likely occupied by the additional helical residues present in the Aux1pLS20 sequence but not in Arc [51].

Fig 2

SAXS analysis of Aux1pLS20. A) Sequence of Aux1pLS20, showing the residues predicted as forming a β-strand in yellow and residues predicted to form α-helices in pink. B) Side (top panel) and top (bottom panel) view of the volume reconstruction based on the Aux1pLS20 SAXS data. C) Cartoon representation of the Arc tetramer in complex with the DNA (not shown) observed in the structure of Arc (PDB code 1PAR). The residues are rainbow colored, blue for the N-terminus changing to red at the C-terminus in a gradual fashion. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Crystal structures of the tetramerization domain of Aux2pLS20

The function of Aux2pLS20 has so far not been determined and cannot be easily inferred from the sequence. A high-resolution structure could provide insights into its function by structural homology to known proteins and we therefore set out to determine its structure. We were able to determine two crystal structures, named Aux2-L and Aux2-S, both containing essentially the same structure of a C-terminal fragment of Aux2pLS20 (Fig. 3), lacking the RHH domains. Since we attempted the crystallization of the full-length protein, these fragments occurred through in situ degradation, indicating flexibility between the RHH and C-terminal domains. Both structures show a tetrameric oligomerization state, where each protein chain folds into two consecutive alpha helices. The N-terminal helix consists of 12 helical turns and the C-terminal helix consists of 3 turns. The short and long alpha helices are connected by a loop of 16 amino acids (A98-E113). The first three N-terminal turns of the long alpha helices are kinked with respect the rest of the α-helix. The long α-helices form a coiled coil structure, interacting with the long α-helices of the other chains through extensive hydrophobic packing interactions, thereby forming tetramers. This creates a tubular structure with a narrow channel at the center. The short alpha helices act as a lock to stabilize the structure. They are oriented in a parallel fashion to the long helices, but interact with the long alpha helices of chains protomers n + 2 and n + 3, thanks to a long loop that connects both helices (Fig. 3C).

Fig. 3

Crystallographic structure of the tetramerization domain of Aux2pLS20. A) Sequence of the full-length protein, predicted β-strands are colored yellow and α-helices pink. The green box indicates the fragment crystallized. B) Cartoon representation of the Aux2pLS20 structure, showing the four monomers in different colors. The N-terminal helices are the inner and longer helices, the C-terminal helices are the external, shorter helices shown at the top. C) Top view of the Aux2pLS20 structure along the α-helices, coloring as in B). D) Structure of the C-terminal domain of the C-terminal domain of plasmid F TraM (PDB code 3D8A). The TraMpF chains are colored in the same order as the Aux2pLS20 inner helices whereas the peptide molecules of the coupling protein TraDpF are colored red. Note the relative shift of the C-terminal helices with respect to the inner helices compared to the Aux2pLS20 structure shown in panel C. E) Graphical summary of the results described in this article. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Interestingly, the structure is similar to the C-terminal domain of TraMpF protein from plasmid F (PDB codes 3D8A and 2G7O) and that of pED208 (PDB code 3ON0) as shown in Fig. 3D [52], [53], [54]. TraM also consists of two α-helices that fold into a similar tetrameric stalk structure, consisting of a central α-helix followed by a second antiparallel helix that laterally packs against the central helices. The latter structure reveals the complex of the full-length protein bound to DNA. Comparison of the Aux2pLS20 four-helix bundle with that of the 3D8A and 3ON0 structures shows that the tetramerization domain in Aux2pLS20 is considerably longer than in the other analyzed structures, suggesting that the N-terminal turns of the long alpha helices in Aux2pLS20 may form part of the DNA binding domain. This region is not included in the crystal structures of the TraMpF in absence of DNA, indicating that the density is not well defined in that region and that these residues are disordered. The structure of the DNA complex of TraMpF shows that this region becomes ordered upon binding of DNA. Therefore, the DNA binding may induce a conformational change in this region that would displace the N-terminal part of the long alpha helix, allowing the RHH domain to change position to adjust to the configuration of the DNA binding sites in the oriT.

When comparing the structure of the tetramerization domain of Aux2pLS20 with that of TraMpF (PDB codes 3D8A and 3ON0), the most pronounced difference is in the position of the C-terminal helices with respect to the long helix. The loop connecting the small and long helices is shorter in TraMpF. As a result, the interactions between the C-terminal and N-terminal helices in the four-helix bundle occur between adjacent monomers, in contrast to Aux2pLS20, where the short helix interacts with the n + 2 monomer. It should be noted that the positions of the shorter helices are equivalent with those observed in the TraMpF structure despite the swap in position.

TraMpF interacts with the plasmid F coupling protein, TraD, and the structure of the complex has been determined (PDB code 3D8A) [53]. The TraDpF peptide binds to the loop region that connects the N- and C-terminal helices (Fig. 3D). The helices in the 3ON0 structure TraMpF, not bound to the peptide, show a similar layout to the structure of the TraDpF/TraMpF complex, with only minor adjustments of the residues in the loop between the helices. In contrast, in Aux2pLS20, the loop between the helices of Aux2pLS20 traverses the region corresponding to the binding site of TraDpF, which is due to the swap of the position of the short C-terminal helix described above.

Discussion

In this paper, we report the low-resolution structure of Aux1pLS20 and two high-resolution structures of the tetramerization domain of Aux2pLS20. In addition, we analyze the possible interactions between the auxiliary proteins and the relaxase using ITC and SEC, which showed that Aux1pLS20 and RelpLS20 interact, but the other proteins do not.

Our results indicate that the RHH protein Aux1pLS20 tetramerizes and forms an elongated structure, reminiscent of the RHH proteins, of which the transcriptional repressor Arc from the Salmonella phage P22 is a founding member [27]. The sequential length, function and structural features of Aux1pLS20 are similar to protein TraY of IncF plasmids: both TraY and Aux1pLS20 are RHH proteins and are able to bind to inverted sequences on their respective oriTs [55], [56]. Our SAXS data suggest that Aux1pLS20 is a tetramer like TraYpF, which is in agreement with the AUC data previously published [26]. From the SEC experiments, however, a pentameric configuration was derived. It should be noted that the elution volume of the proteins in SEC can vary according to their shape. The discrepancies of the MW derived from SEC data compared to SAXS and AUC can be explained by the elongated, non-globular shape of Aux1pLS20. Given that RHH domains are intrinsically dimeric in nature, it follows that the oligomerization state of Aux1pLS20 should be a multiple of dimers. Our combined data and previously published data show that the tetrameric arrangement is most likely.

The data obtained for the relaxase suggests that the protein is monomeric in solution, which is consistent with previous dynamic light scattering and AUC data [23]. Many relaxases are found in a monomeric state, both in solution and in the crystal and single particle electron microscopy structures, which is particularly true for G+ relaxases [21], [57].

The structures of the C-terminal domain of Aux2pLS20 show that it consists of two alpha helices connected by a short loop (Fig. 3). The N-terminal helix is kinked and forms a four-helix bundle, combining four monomers through extensive and tightly packed hydrophobic interactions. As a result, the structure of Aux2pLS20 shows a tetrameric oligomerization state. However, the SEC data indicate a pentamer. As stated above, RHH domains oligomerize as multiples of dimers, and a pentameric structure for the Aux2pLS20 is highly unlikely given that the protein contains dimeric RHH domain. Interestingly, the oligomerization of Aux2pLS20 was previously determined as a hexamer using AUC data [26]. From the structures, it is hard to envisage how the C-terminal stalk domain could reorganize into a hexameric structure, since the four-helix bundle is held together by a tightly packed hydrophobic core. This packing would likely be lost when adding an additional two helices to form a hexamer. However, we cannot exclude the possibility of the formation of a hexamer, and additional work will be required to determine the oligomerization states of this protein at different stages of the conjugation process.

The domain architecture and the C-terminal structure of Aux2pLS20 are reminiscent of the TraM protein of plasmid F and pED208 [52], [53], [54], [58]. Thus, Aux2pLS20 is a structural analog of the TraMIncF proteins. This type of protein is a common factor in gram negative relaxosomes of plasmids from different incompatibility groups [5], [59]. In plasmid F and other systems, one of the functions of this protein is to interact with the coupling protein through the C-terminal domain and the relaxase, functioning as a connector protein. Thus, the structural homology of the Aux2pLS20 C-terminal domain with that of TraMpF suggests that one of the functions of Aux2pLS20 is to interact with the coupling protein of pLS20.

Unlike TraMpF, which was proposed to also interact with the relaxase TraIpF [60], no interactions could be detected between Aux2pLS20 and RelpLS20. This is consistent with previous AUC data which did not detect any interaction between these proteins [26]. Thus, it seems that the interaction between the T4SS and the pLS20 relaxosome is not accomplished by a direct interaction between Aux2pLS20 and the relaxase and it is likely that Aux2pLS20 bridges the relaxosome and T4SS channel through interactions with DNA regions of the oriT [26].

The comparison of the structure of the tetramerization domain of Aux2pLS20 with that of TraMpF also provides clues on how different connector proteins select their substrates. The swapped position of the two helices with respect to TraMpF and the concomitant differences in the position of the loop between the two helices, result in a substantially altered binding pocket of the peptide fragment of the coupling protein. As a consequence, the interaction between the coupling protein of pLS20 and Aux2pLS20, if it exists, is likely different from that observed for plasmid F-like connector proteins. Future work should elucidate if Aux2pLS20 interacts directly with the pLS20 coupling protein, and if the TraMpF binding pocket is preserved or is located in a different region of the protein.

We found that Aux1pLS20 is able to interact with the C-terminal domain of RelpLS20 in SEC and in ITC. This is to the best of our knowledge unique among the relaxosomes studied so far. The interaction is consistent in that it occurs with the full length RelpLS20 as well as its C-terminal domain, but not the N-terminal domain (Fig. 1 and Suppl Fig. 2). It should be noted, however, that this interaction was not observed in AUC data previously published [26]. An explanation of this apparent discrepancy may lie in the different experimental conditions used in both techniques and that the interaction is weak as determined by ITC measurements. It is likely that it can only be detected under favorable conditions, which may explain why AUC experiments failed to detect this interaction. Finally, it should be noted that the experiments described above were performed in absence of DNA, and it is expected that the presence of DNA sequences with affinity for these proteins may induce stronger binding and stabilization of the complexes. Future experiments are needed to investigate these questions further.

The findings in this paper are summarized graphically in Fig. 3E. The combined data presented here show that Aux1pLS20 and Aux2pLS20 likely bridge the various components of the relaxosome and mediate the interaction with the T4SS channel. We identify Aux2pLS20 as a structural analog of TraMpF, which suggests that Aux2pLS20 links the relaxosome to the T4SS channel through interactions with the oriT. Furthermore, we show that Aux1pLS20 interacts with RelpLS20, which suggests that it stabilizes the binding of the relaxase to the DNA sequence, as both that RelpLS20 and Aux1pLS20 contain DNA binding domains. In addition, the interaction provides a possible function to the C-terminal domain of the relaxase.

Conclusions

We present the envelope of Aux1pLS20 and two X-ray structures of Aux2pLS20. The combined data presented here show that Aux1pLS20 and Aux2pLS20 are likely tetramers in solution. We identify Aux2pLS20 as a TraMpF analog and show that Aux1pLS20 interacts with the C-terminal domain of RelpLS20.

Funding

This work was supported by Ministry of Economy and Competitiveness of the Spanish Government grants BIO2016-77883-C2-1-P and PID2019-108778GB-C21 to W.J.J.M, PID2020-117028 GB-I00 (AEI/FEDER, EU), BIO2016-77883-C2-2-P (AEI/FEDER, EU) and FIS2015-72574-EXP (AEI/FEDER, EU), which also supported N.B., to R.B. Funding for open access charge: Ministry of Economy and Competitiveness of the Spanish Government PID2020-117028 GB-I00. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author contributions

IC, NB and AC designed and performed the experiments, BMC, XC and FGO assisted in the technical preparation of the experiments. JVC, AMA, and WJJM provided pET28-based clones and cell culture material of some of the protein samples (or subdomains) used in the initial and/or final studies. MM analyzed the SAXS data, RB designed the research and wrote the paper.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.