RcgA and RcgR, Two Novel Proteins Involved in the Conjugative Transfer of Rhizobial Plasmids

Extrachromosomal DNA elements, such as plasmids, allow for the adaptation of bacteria to new environments by conferring new determinants. Via conjugation, plasmids can be transferred between members of the same bacterial species, different species, or even to organisms belonging to a different kingdom. ABSTRACT Rhizobia are Gram-negative bacteria that are able to establish a nitrogen-fixing symbiotic interaction with leguminous plants. Rhizobia genomes usually harbor several plasmids which can be transferred to other organisms by conjugation. Two main mechanisms of the regulation of rhizobial plasmid transfer have been described: quorum sensing (QS) and the rctA/rctB system. Nevertheless, new genes and molecules that modulate conjugative transfer have recently been described, demonstrating that new actors can tightly regulate the process. In this work, by means of bioinformatics tools and molecular biology approaches, two hypothetical genes are identified as playing key roles in conjugative transfer. These genes are located between conjugative genes of plasmid pRfaLPU83a from Rhizobium favelukesii LPU83, a plasmid that shows a conjugative transfer behavior depending on the genomic background. One of the two mentioned genes, rcgA, is essential for conjugation, while the other, rcgR, acts as an inhibitor of the process. In addition to introducing this new regulatory system, we show evidence of the functions of these genes in different genomic backgrounds and confirm that homologous proteins from non-closely related organisms have the same functions. These findings set up the basis for a new regulatory circuit of the conjugative transfer of plasmids. IMPORTANCE Extrachromosomal DNA elements, such as plasmids, allow for the adaptation of bacteria to new environments by conferring new determinants. Via conjugation, plasmids can be transferred between members of the same bacterial species, different species, or even to organisms belonging to a different kingdom. Knowledge about the regulatory systems of plasmid conjugative transfer is key in understanding the dynamics of their dissemination in the environment. As the increasing availability of genomes raises the number of predicted proteins with unknown functions, deeper experimental procedures help to elucidate the roles of these determinants. In this work, two uncharacterized proteins that constitute a new regulatory circuit with a key role in the conjugative transfer of rhizobial plasmids were discovered.

other organisms, which revealed that they play a similar role, suggesting a conserved function for these proteins.

RESULTS
In silico characterization and taxonomic distribution of LPU83a_0145, LPU83a_0146, and LPU83a_0148. Two hypothetical genes were found in tandem between the Dtr and Mpf regions in all of the plasmids belonging to rhizobial group I-C (25). In the plasmid pLPU83a from LPU83, there is a third hypothetical gene in tandem, located downstream and in the same orientation as the other two, with the difference being that this gene is not conserved in all of the plasmids of group I-C. In pLPU83a, these genes are LPU83a_0145, LPU83a_0146 (conserved), and LPU83a_0148 (conserved). By means of BLAST and Pfam, only one domain corresponding to a putative a/b hydrolase was identified for LPU83a_0146 (Fig. S1A, available at http://sedici.unlp.edu.ar/handle/ 10915/140513). Proteins similar to LPU83a_0148 showed a transmembrane protein annotation. Thus, TMHMM-2.0 (31,32), a software for the prediction of transmembrane domains, was used to predict those domains. The results showed six transmembrane domains in LPU83a_0148 (Fig. S1B). LPU83a_0145 did not show any described domain. All three proteins were modeled with Alphafold (33). Despite the fact that the threedimensional (3-D) structures obtained in this way are prediction models, these structures were compared against the Protein Data Bank (PDB) database, using the Dali server (34) to find structural similarities. LPU83a_0145 showed alpha helix and beta sheet domains, but the structural comparison scores with other proteins were rather low. Thus, no inferences could be made (Fig. S2A, available at http://sedici.unlp.edu.ar/ handle/10915/140513). LPU83a_0146 also presented alpha helix and beta sheet domains (Fig. S2B), and it showed structural similarities to acetylxylanesterases. For LPU83a_0148, a-helix domains were predicted, and they would make it possible for this protein to anchor to the membrane. These domains showed similarity to insect olfactory receptors. The portion corresponding to the C terminus of LPU83a_0148 was modeled as a different domain (outside the a-helix domains), and it showed structural similarities to HTH domains, which could be related to transcriptional regulators (Fig. S2C).
In order to know if the hypothetical proteins were present in other organisms, we searched on the NCBI nr database for proteins with more than 30% identity with LPU83a_0145, LPU83a_0146, and LPU83a_0148, with at least 70% coverage of each protein. For LPU83a_0145, we found 1,142 homologous proteins, of which 47.55% corresponded to Proteobacteria (Fig. S3A, available at http://sedici.unlp.edu.ar/handle/ 10915/140513). Within this phylum, most of the organisms belonged to g-proteobacteria (16.99% of the total hits) and a-proteobacteria (16.55% of the total hits). For LPU83a_0146, we found 1,402 homologous proteins, of which 54.92% were encoded by Proteobacteria and 41.51% by Acidobacteria (Fig. S3B). Within Proteobacteria, almost all of the microorganisms belonged to a-proteobacteria (54.42% of the total hits), with Rhizobiales being the predominant order (48.28% of the total hits). In the case of LPU83a_0148, we found 786 homologous proteins, all of them belonging to Proteobacteria (Fig. S3C). For this protein, 99.49% of the total hits were found in a-proteobacteria. Within this class, 79.39% of the total hits belonged to Rhizobiales while 19.08% belonged to Rhodobacterales. It is also noteworthy that, in most cases, homologs to LPU83a_0146 were found downstream of LPU83a_0148 homologs. Moreover, the analysis of the genetic regions where homologous genes to LPU83a_0146 and LPU83a_0148 are present showed that determinants involved in conjugation are encoded among neighboring genes, allowing us to presume a role for these genes in CT. We also searched these proteins in the metagenomics proteins database. Interestingly, we found hits in the metagenomes of marine sediments, freshwater, and mine drainage, among others (Table S1, available at http://sedici.unlp.edu.ar/handle/ 10915/140513), extending the environments in which functions of these genes could be relevant.
Phylogenetic analysis of LPU83a_0146 and LPU83a_0148 homologs. Since many of the homologous proteins to LPU83a_0146 and LPU83a_0148 were found in tandem, we performed a phylogenetic analysis for homologs to LPU83a_0146 and LPU83a_0148 in order to evaluate whether both proteins showed a similar evolutionary history. For this analysis, we selected organisms with homologs to both proteins which displayed different percentages of identity. We also included several proteins belonging to plasmids found in group I-C (25). The results showed a similar topology between the phylogenetic trees corresponding to the LPU83a_0146 homologs (Fig. 1A) and LPU83a_0148 (Fig. 1B). In both trees, it was possible to observe two well- LPU83a_0148. (C) Genomic organization of conjugative genes in plasmids from groups X and Y. In pLPU83a, genes hyp1 and hyp2 correspond to LPU83a_0146 and LPU83a_0148, respectively. The accession numbers of each protein are listed in Table S2. Bootstrap values higher than 50 are shown on the branches. Proteins encoded in plasmids are annotated as the name of the plasmid. Black arrows highlight the hypothetical proteins of pLPU83a and the homologs found in pLPU88a and Shinella sp. DD12. The * symbol indicates that a sequence is annotated as a scaffold, meaning that it is not possible to determine whether it belongs to a plasmid. defined groups of proteins. One of them, group X, contains proteins from pLPU83a and proteins corresponding to plasmids belonging to group I-C. The other cluster, group Y, contains proteins encoded by plasmids corresponding to Sinorhizobium strains.
We analyzed the genetic neighborhoods of the analyzed genes in each group. Remarkably, all of them showed genes that were involved in CT. Nevertheless, the conjugative systems present in each group were different. The organization of the conjugative regions observed for the plasmids of group X is the same as that described by Castellani et al. (25) for the plasmids of group I-C. To point it out, traR is always located upstream and in the opposite direction of the LPU83a_0148 homologs. The trb operon, corresponding to the Mpf, is always located downstream of traR (Fig. 1C). In the plasmids of group Y, the Mpf is located downstream of the hypothetical genes, and it is composed by genes similar to those of A. tumefaciens (virB genes) (Fig. 1C). Another difference regarding group X is that the coupling protein of the plasmids of group Y (virD4) are associated with the Mpf, while in the plasmids of group X (traG), it is located within the Dtr region. In addition, Dtr genes and traR are absent from the analyzed region of the group Y members. Moreover, the rptA gene, whose product participates in CT via an unknown mechanism (35), is located upstream of the LPU83a_0148 homologs in group Y. This analysis allowed us to identify two groups among the plasmids that harbor homologs to LPU83a_0146 and LPU83a_0148.
Genetic organization of LPU83a_0145, LPU83a_0146, and LPU83a_0148. The conserved, in tandem organization of LPU83a_0146 and LPU83a_0148 could suggest that both of the genes are expressed as an operon. LPU83a_0145, the third gene located downstream of LPU83a_0146, in the same orientation, could be another element of the same transcriptional unit. A preliminary analysis, using data from a RNA-seq experiment (36), suggested that LPU83a_0146 and LPU83a_0148 are transcribed together and that LPU83a_0145 is transcribed independently (GEO accession number GSE141742). To confirm this organization, we designed specific primers to amplify fragments containing regions of contiguous genes, using cDNA as the template. Thereby, it would be possible to obtain polymerase chain reaction (PCR) fragments only if both contiguous genes were transcribed in the same mRNA. The results showed that LPU83a_0146 and LPU83a_0148 are transcribed as an operon, while LPU83a_0145 is transcribed independently (Fig. S4, available at http://sedici.unlp.edu.ar/handle/10915/140513).
Functional analysis of LPU83a_0145, LPU83a_0146, and LPU83a_0148. Separate in-frame deletions of the LPU83a_0145, LPU83a_0146, and LPU83a_0148 genes were generated by double crossing-over. Once the deletional mutants were obtained, the CT frequency of each plasmid was evaluated. The LPU83a_0145 mutant showed similar CT frequencies to the wild-type plasmid ( Fig. 2A). Deletion in LPU83a_0146 showed a remarkable increase in CT frequencies in comparison with the wild-type values (6.05 6 1.24 Â 10 25 versus 1.55 6 1.28 Â 10 26 ) while the mutant in LPU83a_0148 totally abolished the CT of pLPU83a. A deletion mutant of both LPU83a_0146 and LPU83a_0148 was also unable to perform CT.
To corroborate the CT phenotype observed for LPU83a_0146 and LPU83a_0148, we complemented these mutants with a complete wild-type copy of the deleted gene expressed in trans from a replicative vector (pBBR1MCS-5). For both mutants, the presence of the wild-type gene allowed CT frequencies similar to those of the wild-type plasmid (Fig. 2B), confirming that the observed phenotypes were due to the absence of each gene.
Due to the observed phenotype of each gene, we decided to name the operon rcg (for Rhizobial Conjugative Gene). Specifically, the gene whose product is completely necessary for conjugation (LPU83a_0148) is called rcgA, and the gene which reduces CT (LPU83a_0146) is called rcgR.
To evaluate whether the functions of rcgA and rcgR were conserved in other bacteria, heterologous complementation assays were done. For this, two representative strains carrying plasmids with homologous genes were selected: Sinorhizobium meliloti LPU88 (corresponding to group Y) and Shinella sp. DD12 (outside groups X and Y; both indicated with arrows in Fig. 1A). The homologous proteins encoded by S. meliloti LPU88 have 66.79% identity with RcgA and 73.8% identity with RcgR, while the proteins encoded by Shinella sp. DD12 have 56.99% and 64.76% identity, respectively. The results showed that the homologous genes encoded by S. meliloti LPU88 and Shinella sp. DD12 were able to restore the conjugative behavior of pLPU83a in the absence of rcgR (Fig. 2C) or rcgA (Fig. 2D), confirming the same biological activity in those organisms.
Functional analysis of rcgA and rcgR from different genomic backgrounds. Plasmid pLPU83a has different CT behaviors depending on the genomic background in which it is located (27). Considering the above, we evaluated the transfer of pLPU83aDrcgA and pLPU83aDrcgR from Sinorhizobium meliloti 20MP6 (Fig. 3A) and A. tumefaciens UBAPF2 (here, AtUBAPF2) (Fig. 3B). The transfer of pLPU83a from S. meliloti 20MP6 showed values that were close to those observed from the LPU83 background (2.5 6 1.5 Â 10 26 and 1.55 6 1.28 Â 10 26 , respectively). Plasmids with deletion on rcgA were unable to be transferred, and plasmids with deletion on rcgR showed a small but statistically significant increase in CT frequency (5.2 6 1.1 Â 10 26 ). When pLPU83a was transferred from AtUBAPF2, we observed CT with a frequency of 2.9 6 0.86 Â 10 29 , which is near the detection limit. As we expected, it was not possible to obtain the transconjugants of pLPU83aDrcgA, but pLPU83aDrcgR showed a remarkable increase in CT frequency compared with the wild-type plasmid (1.3 6 0.18 Â 10 26 ). Thus, rcgA seems to be completely necessary for the transfer of pLPU83a from different backgrounds, while rcgR acts as a conjugative repressor. In particular, the presence of rcgR in AtUBAPF2 represses conjugation in such a way that the transfer is almost undetectable.
Considering that the functions of rcgA and rcgR are conserved in the transfer of pLPU83a from different backgrounds and that homologous genes from other bacteria are able to complement mutants in the mentioned genes, we evaluated the behaviors of the plasmid and their derivative mutants in the genetic background from which the homologous genes were selected (S. meliloti LPU88 and Shinella sp. DD12). The transfer of pLPU83a from S. meliloti LPU88 was carried out with a relatively low frequency in comparison with the transfer from the native background (1.8 6 0.72 Â 10 27 and 1.55 6 1.28 Â 10 26 , respectively). The transfer of pLPU83aDrcgR showed similar values to those of the wild-type plasmid. For pLPU83aDrcgA, it was possible to obtain the transconjugants from S. meliloti LPU88 with a frequency of 9.4 6 0.57 Â 10 27 (Fig. 3C).
Conjugation of the wild-type plasmid from Shinella sp. DD12 was possible at low frequencies, showing no significant differences with the conjugative frequencies of pLPU83aDrcgR (2.1 6 1.85 Â 10 208 and 1.6 6 1.33 Â 10 208 , respectively). Nevertheless, despite the fact that the homolog to rcgA encoded in pDD12 showed the ability to complement our mutant, the transfer of pLPU83aDrcgA was not observed from this background (Fig. 3D). Shotgun proteomics of pLPU83aDrcgA and pLPU83aDrcgR. To further advance the characterization of rcgA and rcgR, based on the hypothesis that differences in CT frequencies could be due to changes in protein expression, a shotgun proteomic analysis of each mutant and the wild-type strain was performed. Shotgun proteomics allow for the identification of proteins based on the relative abundance of peptides. Three biological replicas of the wild-type, pLPU83aDrcgA, and pLPU83aDrcgR were analyzed and quantified through a Label Free Quantification (LFQ) method. Looking only for proteins with significant LFQ intensity values, we were able to identify a total of 1,638 proteins. Among them, 1,202, 1,186 and 1,250 proteins were found in the wild-type strain, pLPU83aDrcgA, and pLPU83aDrcgR, respectively (Table S3, available at http:// sedici.unlp.edu.ar/handle/10915/140513). Among the identified proteins, the abundance of proteins involved in the Dtr and Mpf of pLPU83a was quite low. The statistical analyses did not reveal proteins with different relative abundances in the comparisons between the wild-type and each mutant. Then, we analyzed the proteins that were detected in one sample but not in the other one, the ON/OFF proteins. When comparing the wild-type strain versus pLPU83aDrcgA, only 7 proteins were found exclusively in the wild-type strain, while 9 proteins were found exclusively in the pLPU83aDrcgA strain (Tables S3, red lines). The comparison between the wild-type strain and pLPU83aDrcgR, showed 3 proteins present exclusively in the wild-type strain, while 3 proteins were exclusive to pLPU83aDrcgR (Tables S3, yellow lines). Remarkably, two of the proteins present exclusively in pLPU83aDrcgR were TraR and TraC, the CT master regulator and a protein that is part of the Dtr, respectively. The remaining proteins showed no predicted relationship with CT, nor were they in the plasmid pLPU83a.
TraR expression analysis using transcriptional fusion. In plasmids whose transfer regulation depends on TraR, the expression of this protein is directly related with transfer frequencies. In these systems, TraR promotes the transcription of the Dtr and Mpf genes (21,22). As TraR was found in the proteomic analyses, we hypothesized that the phenotype generated by the deletion of each of the studied genes could be related to changes in TraR expression. To evaluate this hypothesis, we constructed a transcriptional fusion between traR and a gfp gene, and we analyzed it in our mutants. The construction was integrated into pLPU83a derivatives in such a way that traR remained functional and that the expression of both genes (traR and gfp) relied on traR promoter. For the strain pLPU83aDrcgA, no changes in traR transcript levels were observed (Fig. 4A), showing that the function of RcgA is not related to the levels of traR. For the strain pLPU83aDrcgR, we observed a significant increase in fluorescence levels (Fig. 4A), demonstrating that RcgR acts by repressing, in some way, traR transcription. We also evaluated the plasmid with a deletion in LPU83a_0145, but it did not show changes in traR expression. Moreover, the behaviors of pLPU83a and its mutants were evaluated in the AtUBAPF2 background. The plasmids with the transcriptional fusion were introduced into the mentioned strain, and the fluorescence levels were evaluated. As Fig. 4B shows, the fluorescence values were lower than those obtained from the LPU83 background. Nevertheless, the mutant in rcgR showed a significant increase in fluorescence levels, which correlates with the increase in CT frequencies.

DISCUSSION
Unraveling the molecular basis of plasmid transfer regulation provides an understanding of the behaviors of these molecules in their environments. Molecular mechanisms that regulate CT are diverse; the study of the regulatory networks related to this process aims to understand the dynamics of plasmid DNA dissemination. The plasmid transfer of antibiotic resistance genes has a notorious relevance due to its impact on human health (37,38). Symbiotic and pathogenic traits are also located in plasmids. In symbiotic bacteria, such as rhizobia, two main mechanisms that regulate CT have been described (18,19,(21)(22)(23). The characterization of hypothetical genes by biochemical and molecular biology approaches has allowed for the identification of new actors involved in the described mechanisms (28)(29)(30)39). In this work, we evaluated the function of three hypothetical genes, and presented evidence of two novel proteins involved in rhizobial plasmid transfer. One of the hypothetical genes, which we named rcgA, was predicted as a hypothetical protein with transmembrane domains by approaches from bioinformatics. Modeling with Alphafold (33) confirmed this prediction and showed also a putative HTH motif that is similar to those present in transcriptional regulators. Analyses showed that it is present almost exclusively in Alphaproteobacteria. The other hypothetical gene, rcgR, was predicted as a hypothetical protein with an a/b hydrolase domain. The model of this protein showed similarities to esterases. rcgR is mostly found in Alphaproteobacteria, but it is also present in Acidobacteria. The predominant distribution of both genes within the Rhizobiales order allows us to suppose that they have a specific function in rhizobia, which has not been extended to other organisms. The conserved organization and the closeness to conjugative genes could suggest a relationship between the functions of both genes and the conjugative process. It is also noteworthy that the environmental distribution is diverse, with the genes being detected in soils, the guts of aquatic organisms, and other environmental metagenomes.
Phylogenetic studies of RcgA and RcgR showed a similar evolutionary history for both proteins, with the obtained trees presenting a matching topology. Two welldefined groups of homologs were observed, with one of them containing proteins close to those encoded by pLPU83a (group X), and the other containing proteins from S. meliloti strains (group Y). It should be noted that group X contains the same plasmids that are present in group I-C, obtained from the TraR homologs (25) and showing an evolutionary relationship, and probably a related function, between the hypothetical genes and traR. Despite the plasmids from group Y lacking a traR gene, they harbor a complete Mpf set of genes and a coupling protein. Thus, they could be transferred, but the regulation of the CT might be different. Future studies on this system are important, considering that symbiotic plasmids (such as pSymA from Sinorhizobium meliloti RU11/001 and pSM11c from Sinorhizobium meliloti SM11) were found in this group. Until now, most of the described symbiotic plasmids are regulated by rctA/rctB (18)(19)(20), although a QS mechanism has also been described for a symbiotic plasmid from Rhizobium leguminosarum bv. viciae (40).
Regarding the biological functions of RcgA and RcgR, we confirmed their participation in the CT of the plasmid pLPU83a. Surprisingly, despite rcgA and rcgR being transcribed in an operon, they showed opposite effects on conjugation. The deletion of rcgA led to a nonconjugative phenotype, while the deletion of rcgR showed an increase in CT frequencies. Thus, rcgA encodes a protein that is essential for CT, while rcgR encodes a protein that, in some way, represses the process. The double mutant, DrcgAR, did not allow conjugation, indicating that rcgA is completely necessary for CT, even in the absence of the repressor. We also confirmed that the biological functions of both proteins are conserved in different organisms, as homologous proteins encoded in plasmids with a different evolutionary relationship and organization of conjugative genes were able to restore the CT phenotypes in complementation assays. Investigating the function of rcgA and rcgR homologs in the conjugation of plasmids without traR will be important in elucidating new regulatory networks.
From an ecological perspective, pLPU83a shows an interesting profile. It is able to use its own conjugative machinery or hijack the machinery of its host (26). Studies of pLPU83a and their derivative mutants in different genomic backgrounds confirmed that rcgA and rcgR are involved in the CT of pLPU83a when this plasmid is transferred from AtUBAPF2 and S. meliloti 20MP6. The transfer from AtUBAPF2 showed that the genomic background plays an important role in the CT of pLPU83a, as in this background, the wild-type plasmid is transferred at low frequencies, whereas the deletion of rcgR notably increases these values. A possible explanation is that in the AtUBAPF2 genomic background, the operon rcgA/rcgR has an enhanced transcription, resulting in a greater repression of pLPU83a transfer. Another possibility is a reduced transcription of the CT master regulation gene, traR, in this background. Experiments using a transcriptional fusion between traR and GFP in AtUBAPF2 confirmed that the increments in the CT frequencies of pLPU83aDrcgR are caused by an enhanced expression of TraR in comparison with the wild-type values. By analyzing the behavior of pLPU83a from other genomic backgrounds, we showed that this plasmid is able to conjugate from S. meliloti LPU88 and Shinella sp. DD12. Both strains contain homologous genes to rcgA and rcgR, and according to heterologous complementation assays, genes of both strains were able to complement the absence of rcgA and rcgR in the corresponding mutants of pLPU83a in the LPU83 genomic background. Despite the results of the heterologous complementation, it was not possible to obtain the transconjugants of pLPU83aDrcgA from Shinella sp. DD12. A possible explanation for this could be that the homolog to rcgA encoded by Shinella sp. DD12 is not expressed properly. The fact that there are no differences in CT between the wild-type and DrcgR plasmid in the S. meliloti LPU88 and Shinella sp. DD12 backgrounds could be explained by the rcgR homologs encoded in both organisms being functional and repressing CT to similar levels. This could also be supported with the lower CT frequencies observed for pLPU83a in these backgrounds.
Even though the shotgun proteomics assays show a low relative abundance of the proteins involved in CT, it is remarkable that TraR (the master regulator) and TraC (a protein member of the Dtr) were only detectable in pLPU83aDrcgR. It has been described that, usually, TraR directly activates the transcription of the Dtr and Mpf genes (22,41) and that, in some systems, TraR is involved in a positive feedback loop (22,23). In this way, TraR affects different nodes. Our proteomics results, together with the transcriptional fusion assays, allow us to affirm that RcgR is acting as a repressor, somehow affecting the production of TraR and TraC, thereby affecting CT. An interaction between RcgR and TraR (either direct or mediated by another protein) could be affecting the capacity of TraR to binding to its own promoter, thereby affecting the positive feedback loop, and to Dtr and Mpf promoters, thereby affecting the transcription of conjugative genes. In other rhizobial plasmids, a similar mechanism was described for the antiactivator TraM, which affects TraR function by impeding TraR dimerization, a necessary step for TraR function (42)(43)(44). A deletion mutant of traM in pLPU83a does not show any difference in CT frequencies (not shown). TraM and RcgR do not share any identity by BLASTP. The other regulatory mechanisms of conjugation are based on transcriptional interference in which there is competition between a repressor protein and a transcription factor for binding to the promoter (18)(19)(20). Even though there is no predicted DNA interaction domain in RcgR, the protein could be interacting with the traR promoter zone, thereby preventing the transcription of the gene.
The overexpression of TraC in the absence of RcgR could reflect an enhancement (but not one that is detectable in our proteomics approaches) in the levels of the proteins belonging to Dtr and Mpf. Thus, an interaction between RcgR and the promoters of Dtr and Mpf genes cannot be disregarded. TraR is able to induce Dtr and Mpf transcription by binding to a consensus sequence located in each promoter zone, called tra box (45). pLPU83a has tra boxes upstream of its Dtr and Mpf genes (26). Thus, this conserved sequences could be the target for RcgR action.
In QS-regulated systems, the mechanism for TraR activity implies the recognition of a signal molecule that is necessary for the dimerization and activity of the protein (21,46). In most cases, the recognized molecules are AHLs, which have the same homoserine lactone ring but differ in length and in the structure of the acyl chain (47). Some organisms are able to prevent the accumulation of AHLs by hydrolyzing the homoserine lactone rings of these molecules. For example, Agrobacterium tumefaciens A6 encodes an AHL-lactonase, attM, that is involved in the hydrolysis of 3OC8AHL, the molecule that mediates the QS response in this strain (48). Based on the predicted a/b hydrolase domain present in RcgR, this protein could be hydrolyzing a signal molecule that is hypothetically involved in TraR activation (either produced by the cell or imported through RcgA). Thus, the lack of this molecule would not allow the formation of the active form of TraR, and in consequence, the Dtr and Mpf genes would not be activated. With the same hypothesis that RcgR acts to hydrolyze a molecule, another possibility is that it participates in the processing of an unknown molecule, thereby generating a smaller version that abolishes TraR activation. This kind of small inhibitor resembles the inhibitor peptide I from Enterococcus faecalis pCF10, which represses CT (49,50).
The role of RcgA in CT is more difficult to decipher, as the mutant strain did not show any differences in fluorescence values or in the detected proteins. Although there is a high degree of conservation across type I rhizobial transfer systems that work well without RcgA, and although the proteins that form the Mpf have been widely described (51)(52)(53)(54), the presence of predicted transmembrane domains in RcgA allow us to hypothesize a structural role for this protein in group I-C of rhizobial plasmids, with it being an accessory protein of the Mpf or being necessary for Mpf assembly. Thus, the lack of RcgA would not allow Mpf assembly and would thereby prevent CT. Coupling proteins consist of a transmembrane and an ATPase domain, and they are essential for CT due to their role in the recruitment of DNA to the Type IV Secretion System (T4SS) (55,56). Based on this, despite pLPU83a encoding a coupling protein (traG), another hypothesis is that RcgA acts in a similar way to a coupling protein, being involved in the recruitment of a plasmid to the T4SS assembled in the membrane. Then, the absence of RcgA would not allow the interaction between the processed DNA and the Mpf. Another role could rely on the transport of a signal produced by the receptor strain. It has been recently described that the presence of a receptor strain increases the expression of conjugative genes in systems regulated by LuxR-like regulators (57). The fact that exogenous signals regulate CT has also been reported in pCF10 from Enterococcus faecalis through the transport of C peptide (49,50). Thus, RcgA could transport a signal that is needed for CT from the receptor strain. Indeed, as RcgA/RcgR are encoded in an operon structure, it could be hypothesized that RcgR inactivates the signal transported by RcgA, which is needed for CT. In this way, both genes would create a fine tuning of CT in the presence of the receptor strain. Hence, the data presented here shows evidence of new actors that expand the knowledge of conjugative plasmid regulation. Future efforts will be devoted to elucidating the complete molecular system, thus allowing for a better understanding of how plasmids are disseminated between soil microorganisms.
Bacterial matings. The bacterial matings were performed as described by R. Simon et al. (60). Briefly, overnight cultures were grown to stationary-phase. The donor and recipient strains were mixed in a 1:1 ratio, plated on TY plates, and incubated overnight at 28°C. The bacteria were resuspended in 1 mL of 10 mM MgSO 4 -0.01% Tween 40 (vol/vol). For quantitative conjugations, Agrobacterium tumefaciens UBAPF2 was used as the recipient strain. Serial dilutions were plated on selective TY medium supplemented with the corresponding antibiotics to quantify the number of donor, recipient, and transconjugant cells. The conjugation frequencies were calculated as the ratio of transconjugants per donor cell. In each case, plasmid profiles were done by Eckhardt-gels (61), as modified by Hynes & McGregor (62), to corroborate the desired plasmid transfer.
RNA extraction and synthesis of cDNA. For RNA isolation, R. favelukesii LPU83 was cultivated in TY media. Then, RNAprotect (Qiagen, Hilden, Germany) reagent was added, the cells were harvested, and the pellet was frozen in liquid nitrogen. The commercial RNeasy R Protect Bacteria Mini Kit (Qiagen, Hilden, Germany) was used for the RNA extraction, and the DNA was removed with DNase I (Qiagen, Hilden, Germany).
The total RNA was retrotranscribed to cDNA using a Moloney murine leukemia virus (MMLV) reverse transcriptase (Thermo Fisher Scientific, Buenos Aires, Argentina), using random hexamers as primers.
DNA manipulation, genetic constructs, and mutagenesis. The total DNA and plasmid preparations, restriction-enzyme analysis, cloning procedures, and E. coli transformation were performed according to previously established techniques (63).
PCR amplification was carried out with recombinant Taq DNA polymerase or with Phusion DNA polymerase as specified by the manufacturers. The primers used in this study are listed in Table S5 (available at http://sedici.unlp.edu.ar/handle/10915/140513).
The constructed vectors were introduced by conjugation into R. favelukesii LPU83-13. Double recombinants were selected in TY saccharose 10% wt/vol as Nm R , and Gm S . To corroborate the deletions, PCRs were carried out with primers 145-l-out and 145-r-out for deletion in LPU83a_0145 (581 bp), 146-l-out and 146-rout for deletion in LPU83a_0146 (657 bp), PCR3-r and 148-r-out for deletion in LPU83a_0148 (639 bp), and 146-l-out and 148-r-out for deletion in LPU83a_0146 and LPU83a_0148 (606 bp).
Plasmid constructions for complementation assays and generation of complemented strains. The vector for the LPU83a_0145 complementation assay was constructed as follows. A fragment containing the full LPU83a_0145 gene was amplified (757 bp) with Phusion polymerase and primers 145-l-Xba/145-r-Kpn. The fragment was cloned into the SmaI site of pK18mob. The resulting vector was digested with KpnI/XbaI, and the released fragment was cloned into the KpnI/XbaI sites of pBBR1MCS-5, yielding pBBR::0145 (5,453 bp).
The vector for the rcgR complementation assay was constructed as follows. A fragment containing the full rcgR gene was amplified (1,079 bp) with Phusion polymerase and primers 146-l-Xba/146-r-Kpn. The fragment was cloned into the SmaI site of pK18mob. The resulting vector was digested with KpnI/XbaI, and the released fragment was cloned into the KpnI/XbaI sites of pBBR1MCS-5, yielding pBBR::rcgR (5,775 bp).
The vector for the rcgA complementation assay was constructed as follows. A fragment containing the full rcgA gene was amplified (1,831 bp) with Phusion polymerase and primers 148-l-Xba/148-r-Kpn. The fragment was cloned into the SmaI site of pK18mob. The resulting vector was digested with KpnI/ XbaI, and the released fragment was cloned into the KpnI/XbaI sites of pBBR1MCS-5, yielding pBBR::rcgA (6,527 bp).
For the generation of complemented strains, the constructed vectors were transformed into E. coli S17-1, and then biparental matings to the respective strains were carried out. The transconjugants were selected as Nm r and Gm r Plasmid construction and the generation of Gm resistant strains. For the construction of the strain S. meliloti LPU88 Gm, an intergenic region between rptA and ORF4 (accession number JQ753316.1) from the plasmid pLPU88a was amplified with Phusion polymerase and primers Gm-88a-left/Gm-88a-right. The obtained fragment (239 bp) was cloned into the SmaI site of pG18mob2, yielding pG18mob2::88a (3,120 bp). To introduce pG18mob2::88a into S. meliloti LPU88, the pG18mob2::88a was first transformed in E. coli S17-1, and then biparental matings were carried out. The simple recombinants were selected as Sm r and Gm r . To corroborate the insertion, PCR was carried out with primers 88a-check/m13rv-40 (772 bp).
For the construction of the strain Shinella sp. DD12 Gm, an intergenic region between genes SHLA_RS09770 and SHLA_RS09775 from the plasmid pDD12c was amplified with Phusion polymerase and primers Gm-Sh-left/Gm-Sh-right. The obtained fragment (220 bp) was cloned into the SmaI site of pG18mob2, yielding pG18mob2::Sh (3,101 bp). To introduce pG18mob2::Sh into S. sp. DD12, the pG18mob2::Sh was first transformed in E. coli S17-1, and then biparental matings were carried out. The simple recombinants were selected as Sm r and Gm r . To corroborate the insertion, PCR was carried out with primers Sh-check/m13rv-40 (705 bp).
Plasmid construction for traR::GFP fusion and the generation of the strains carrying the construction. For the construction of pG18mob2::traR::GFP, the full traR gene from pLPU83a was amplified with Phusion polymerase and primers traR-Nter/traR-Cter. The obtained fragment (788 bp) was cloned into the SmaI site of pG18mob2, yielding pG18mob2::traR (3,669 bp). To introduce GFP in this plasmid, the full GFP gene was released from pMP6 with EcoRI and cloned into the EcoRI site of pG18mob2::traR, yielding pG18mob2::traR::GFP (4,446 bp). The orientation of GFP was confirmed by the restriction pattern generated by KpnI. To introduce pG18mob2::traR::GFP into R. favelukesii LPU83-13 and its derivative mutants, the pG18mob2::traR::GFP was first transformed in E. coli S17-1, and then biparental matings were carried out. The simple recombinants were selected as Nm r and Gm r . To corroborate the insertion, PCR was carried out with primers traR-out/m13rv-40 (1,890 bp). For the assays in A. tumefaciens UBAPF2, pLPU83a-13 and its derivative mutants containing the traR::GFP fusion were transferred to A. tumefaciens UBAPF2 by conjugation. The transconjugants were selected as Gm r , Nm r , and Rif r .
Bioinformatics 3-D structure predictions. For structural predictions, a simplified version of AlphaFold v2.1.0 (33) was used. PDB files were visualized with PyMOL 2.5 (The PyMOL Molecular Graphics System, Version 2.5, Schrödinger, LLC). The structural comparisons were done by using the Dali server (34) against the PDB database. The analyses of the transmembrane domains were done with TMHMM-2.0 (31,32).
Taxonomic analysis. A BLASTP analysis on the NCBI server was used to obtain the homologs. For the metagenomic proteins, we used the env nr database from NCBI. For the homologs in isolated bacteria, we used the nonredundant protein sequences (nr) database from NCBI. The data are actualized to February 2, 2022. Genetic regions were examined in the neighborhoods of the selected genes using the nucleotide visualizer tool from NCBI. As criteria for protein homology, we used the parameters of greater than 70% coverage and greater than 30% identity (64).
Phylogenetic analysis. For the construction of the RcgR (LPU83a_0146) and RcgA (LPU83a_0148) phylogenetic trees, the proteins were aligned with the module of MUSCLE implemented in MEGA5 (65). The models of protein evolution for our sequences were selected with Prottest2.4 (66). The best model was JTT 1 I 1 G 1 F for both trees. Maximum likelihood (ML) trees were inferred under the selected model using PhyML v3.1 (67). The robustness of the ML topologies was evaluated using a Shimodaira-Hasegawa-like test for branches, implemented in PhyML v3.1. We employed the best of the nearestneighbor interchange and shortest path routing algorithms to search the topologies of the trees and used 100 random trees as the initial trees.
The accession numbers for the proteins selected for the LPU83a_0146 and LPU83a_0148 phylogenies are listed in Table S2 (available at http://sedici.unlp.edu.ar/handle/10915/140513).
GFP transcriptional fusions and fluorescence assays. To evaluate traR promoter activity, we designed a strategy based on a transcriptional fusion between the GFP protein and traR, considering that an increment in traR transcription would lead to an increment in GFP expression. Details of the construction of pG18mob:traR::GFP are listed before.
Once the strains carrying the construction were obtained, fluorescence was measured as follows: cultures were grown on TY liquid media until stationary-phase. After an OD 600 measurement, a bacterial pellet normalized to OD = 1 was resuspended in 200 mL of NaCl 0.9% (wt/vol). Fluorescence was measured in a DeNovix QFX fluorometer. The values are expressed in arbitrary units (AU).
Proteomics analyses. For the protein isolation, 1.5 mL of cell cultures were centrifuged at 13,000 rpm for 5 min, and pellets were frozen in N 2 . Pellets were resuspended in 100 mL of 100 mM Ambic (ammonium bicarbonate [Honeywell Fluka]). Next, 100 mL of the organic solvent trifluoroethanol (TFE, Fluka biochemika) and 5 mL of 20 nM DTT were added. The tubes were mixed by inversion and incubated at 60°C for 60 min. In a second step, 20 mL of 200 mM iodoacetamid (IAA) was added, and the tubes were then placed in the dark at room temperature for 90 min. Then, 5 mL of 20 nM DTT were added again, and the tubes were then incubated at room temperature for 60 min. Half of the mixture was saved at -80°C as a backup. Then, 437 mL of 100 mM ammonium bicarbonate and 437 mL of bidistilled water were added to the remaining sample. In the next step, 10 mL trypsin gold (1 mg mL 21 ) (Mass Spectrometry Grade, Promega, WI, USA) were added, and the solution was incubated at 37°C overnight.
The protein digestion was purified the next day with Sep Pak C18 cartridges (Waters Corporation, Milford, MA, USA). The Sep Pak C18 cartridges were rinsed with 1 mL of solution B (65% Acetonitril, 35% bidistilled water; 0.1% TFA). The column was equilibrated with 1 mL of solution A (98% Acetonitril, 2% bidistilled water; 0.1% TFA). Next, the digest mixed with 1 mL of solution A was added and run through the column slowly. Subsequently, the cartridges were washed with 1 mL of solution A. The proteome was eluted with 100 mL of solution B in low protein binding collection tubes and then dried in a vacufuge concentrator (Eppendorf, Hamburg, Germany). Next, the dried peptide mixture was resuspended in 15 mL of solution A and quantified with a Nano-Drop 2000/2000c (Thermo Fisher, MA, USA).
LC-MS/MS measurements were carried out using a QExactive mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) that was online coupled to an LC system. The peptides were separated on a 25 cm steel column Acclaim PepMap 100 C18-LC-column with a particle size of 2 mm and a diameter of 75 mm (Thermo Fisher, MA, USA). Identification and a label-free quantification (LFQ) analysis were performed using the software MaxQuant with the default settings and a false discovery rate of p adj , 0.05 (68). In total, 253,381 MS/MS spectra were recorded, resulting in 18,114 identified peptide sequences corresponding to 2,375 proteins.
The statistical analysis of the LFQ data obtained from MaxQuant was performed with Perseus 1.6.10.43 (69). 2,078 proteins remained when at least 2 unique peptides were used as the threshold for identification. Among these, the normalized LFQ intensities became null in all of the samples for 440 of them. Thus, the final protein list contains 1,638 proteins.
Data availability. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (70) via the PRIDE partner repository (71) with the data set identifier PXD033583.