The Facts and Family Secrets of Plasmids That Replicate via the Rolling-Circle Mechanism

SUMMARY

Plasmids are self-replicative DNA elements that are transferred between bacteria. Plasmids encode not only antibiotic resistance genes but also adaptive genes that allow their hosts to colonize new niches. Plasmid transfer is achieved by conjugation (or mobilization), phage-mediated transduction, and natural transformation. Thousands of plasmids use the rolling-circle mechanism for their propagation (RCR plasmids). They are ubiquitous, have a high copy number, exhibit a broad host range, and often can be mobilized among bacterial species. Based upon the replicon, RCR plasmids have been grouped into several families, the best known of them being pC194 and pUB110 (Rep_1 family), pMV158 and pE194 (Rep_2 family), and pT181 and pC221 (Rep_trans family). Genetic traits of RCR plasmids are analyzed concerning (i) replication mediated by a DNA-relaxing initiator protein and its interactions with the cognate DNA origin, (ii) lagging-strand origins of replication, (iii) antibiotic resistance genes, (iv) mobilization functions, (v) replication control, performed by proteins and/or antisense RNAs, and (vi) the participating host-encoded functions. The mobilization functions include a relaxase initiator of transfer (Mob), an origin of transfer, and one or two small auxiliary proteins. There is a family of relaxases, the MOB_V family represented by plasmid pMV158, which has been revisited and updated. Family secrets, like a putative open reading frame of unknown function, are reported. We conclude that basic research on RCR plasmids is of importance, and our perspectives contemplate the concept of One Earth because we should incorporate bacteria into our daily life by diminishing their virulence and, at the same time, respecting their genetic diversity.

INTRODUCTION

Bacterial genomes are interspersed and enriched by mobile genetic elements (MGEs) that constitute ∼20% of the total DNA contributing to the genetic biodiversity of the bacterial world. MGEs are shared among bacteria, providing them with a common DNA pool, the so-called “mobilome” (1, 2). Bacterial species with the increased genetic traits provided by the acquired MGEs are equipped with flexible responses to fluctuating environmental challenges. These responses especially apply to the bacterial plasmids, the “plasmidome” (3), because they are autonomously replicating DNA molecules. Plasmids actively participate in bacterial evolution by providing their hosts with novel genetic traits that may be helpful under certain environmental conditions. Plasmids are a shared combination of mobile DNA able to spread among unrelated bacterial hosts by at least three ways, namely, conjugation, transduction, and natural transformation with free DNA.

Conjugation

Conjugation is a process by which a donor bacterium transfers one strand of DNA belonging to a plasmid or a chromosomally located conjugative element to a recipient bacterium of the same or different species. The dissemination of genetic information by conjugation is one of the mechanisms of horizontal gene transfer (HGT) in bacteria that play a key role in the spread of antibiotic resistance (Ab^r), and, hence, HGT is crucial in the selection and emergence of multiresistant pathogenic bacteria (“superbugs”). Conjugation requires intimate contact between donor and recipient cells, and it is mediated by a complex machinery that includes components encoded by two MGE regions: (i) the mating-pair formation (Mpf) region and (ii) the DNA transfer region (Dtr). The latter encodes a protein termed relaxase (generically designated Tra or Mob), which cleaves the supercoiled plasmid DNA at the phosphodiester bond of a specific dinucleotide located in a cis-acting DNA region termed the origin of transfer, oriT. After cleavage, the relaxase remains covalently bound to the DNA, and this complex is pumped out from the donor to the recipient cell through a type IV secretion system (T4SS) encoded by the Mpf region and that includes a T4-coupling protein (T4CP) and several more proteins, the most conserved being a VirB4-like ATPase (4). A given MGE can contain all the genes required for conjugation (conjugative elements) or only the Dtr region (mobilizable elements). Mobilizable elements use the Mpf genes from a helper MGE to be mobilized. A DNA structure-based alignment algorithm designed to find orphan oriT variants revealed potential oriT loci across thousands of sequenced plasmids (5). HGT mediated by these orphan oriTs would then be extended to many more MGEs than envisaged (5). HGT can also be mediated by other MGEs that are integrated into the bacterial chromosome and, upon their excision, transfer to other hosts by conjugation (6–9). They include integrative and conjugative elements (ICEs; also termed conjugative transposons) and integrative and mobilizable elements (IMEs). Other MGEs integrated in the host chromosome that participate in HGT include insertion sequences (ISs) and prophages (10, 11). They all participate in the spread of Ab^r genes, although the contribution of IMEs and ISs in the process is poorly understood and probably underestimated (12, 13). In the case of Firmicutes, a profusion of ICEs and IMEs has been reported and shown to cooperate in the spread of Ab^r traits, especially among streptococci, enterococci, and staphylococci (10, 14–16). Based on the gene content, nucleotide identities, and phylogenetic analyses, four most common families of ICEs have been reported in streptococci, their representatives being Tn916 (the best characterized), Tn5252, Tn1549, and vanG, this last one responsible for the spread of vancomycin resistance (16, 17).

Transduction

In Bacillus subtilis and Staphylococcus aureus, small plasmids replicating by the rolling-circle mechanism (RCR plasmids) have been shown to generate high-molecular-weight DNA, formed by linear plasmid concatemers, upon phage infection; this DNA was packaged into transducing phage particles (18–20). This particular mechanism of plasmid transduction is independent of the host major recombinase, RecA, indicative of the participation of phage-mediated functions (21). It was later discovered that transduction of RCR plasmids in B. subtilis mediated by phage SPP1 required a phage-encoded ATP-independent single-strand annealing recombinase protein; transduction frequencies were higher in RCR plasmids than in θ-replicating plasmids (22). Nevertheless, and despite these studies on phage-mediated plasmid transduction, the contribution of transduction to the spread of plasmid traits may be underrated and the role of transduction in HGT needs to be revisited by experimental and theoretical approaches (23). How general these processes are in different bacterial populations remains to be determined, although information on the influence of phage predation on the modulation of mammal gut microbiome indicates that they are (24).

Natural Transformation

The uptake of free DNA by bacteria has been studied in depth over the years, especially in the context of the differentiation processes involved in competence development (25–27). The role of free DNA in the environment has been greatly undervalued. Free DNA molecules participate in the matrix of bacterial biofilms (28) and in the spread of Ab^r without the need for physical cell-to-cell contacts (29). Genetic transformation was considered to be limited to a few species, namely, Streptococcus pneumoniae, B. subtilis, Haemophilus influenzae, and Neisseria gonorrhoeae, among the earliest studied bacteria that can enter the competent (“K”) state. Competence development results from a homogenous bacterial population that differentiates into two distinct subpopulations, one that enters the K state and another that does not. These processes are termed bistable, and they arise by the expression of a key regulatory gene, in this case, a specific sigma factor (30). Genetic transformation has been evolutionarily selected to increase the genetic variability of the part of the population that becomes competent. The K state relies on the transcription of regulatory circuits, which are diverse and tailored to the particular lifestyle of each bacterial species. Thus, competence can be thought of as a differentiation mechanism by which part of the bacterial population changes its gene expression patterns due to stress processes. As the result of stress, these bacteria trigger sets of otherwise silent genes whose transcription is directed by specific sigma factors (27, 31). Competent cells would commit “fratricide” (32) or “cannibalism” (33). They do so by killing the noncompetent subpopulation and releasing its content, which will be used as novel traits to incorporate into their genomes or as sources of food (34). The K state varies among species, and some bacteria, like S. pneumoniae, trigger competence to respond to DNA-damaging antibiotics. Since S. pneumoniae lacks the SOS response, it appears that the presence of SOS-inducing antibiotics like fluoroquinolones or mitomycin C leads to competence development in pneumococcal cells harboring the competence regulatory cascade (35). These antibiotics would play a role in the genome plasticity of at least some pathogens (36). The number of bacterial species developing competence for natural transformation is higher than imagined, and thus, the role of competence development in the bacterial lifestyle is of broader significance. Earlier, the genetic transformation was considered a way to incorporate exogenous material to cope with stress caused by fluctuating environments and acquisition of resistance to antibiotics or heavy metals (31, 37). However, more recent research regards genetic transformation also as a way to incorporate relatively short pieces of DNA that the competent cells would use to eliminate parasitic MGEs from their chromosomes through homologous recombination events (38). It appears that naturally transformable bacteria, at least S. pneumoniae and B. subtilis, harbor fewer MGEs than the nontransformable ones; the number of indigenous plasmids found in these two bacterial species seems to be lower than in other related nontransformable bacteria, like Staphylococcus, Enterococcus, or Lactococcus (31).

A somewhat more compromising and daring proposal considers transformation as a reversible mechanism used by many bacteria (the number of reported naturally transformable bacteria has increased up to nearly 70 species) to incorporate new traits in times of need and eliminate the chromosomally integrated MGEs when environmental conditions render them unnecessary (39). The proposal of a payoff by the MGEs considers that they could develop strategies to keep them as parasites (38). This proposal is supported by the following: (i) toxin-antitoxin (TA) operons are frequently associated with MGEs and (ii) TA systems contribute to the stability of the bacterial genome (40–42). Some TA pairs are associated with MGEs in streptococci, as follows: (i) the RelBE-like TA operon is associated with the IMESag-rpsI of Streptococcus agalactiae (this IME is widely distributed in streptococci despite not carrying any plain virulence trait) (15), (ii) the RelBE TA pair encoded by the pneumococcal chromosome appears associated with putative MGEs (43), and (iii) the PezAT operon is included within the PPI1 pathogenicity island of S. pneumoniae (44); acquisition of this island leads to loss of some functions (reduction in resistance to β-lactams and genetic competence) and the gain of other traits, like virulence (45). These observations have been extended to various transposons from Escherichia coli and proposed to play a “take me or die” mechanism to ensure the permanence of the MGEs (46).

At present, we are faced with an increasingly relevant role of genetic transformation in HGT (47), which was previously attributed mostly to plasmid-mediated conjugative transfer. Recent findings on the type IV competence pili indicate that the fraction of competent cells has a machinery that extends and retracts from the cell surface to seek and incorporate free DNA, at least in S. pneumoniae (48). In B. subtilis, competence pili are placed near the cell poles and are needed for the binding of DNA to the cell surface to initiate its uptake (49). Does that mean that competent cells (at least the Gram-positive [G+] ones) play an active role in finding and chasing a source of DNA? If this were the case, we could contemplate a different scenario in plasmid spread, which would be plasmids acting like baits waiting to be actively incorporated by the cell uptake mechanism (50). The discovery of defense systems, not only the antiphage systems but also the Wadjet antiplasmid system protecting against plasmid transformation in B. subtilis, has exposed new avenues that must be thoroughly explored (51, 52).

RCR PLASMIDS: FEATURES AND FAMILIES

The versatility of the mobilome is increased by small multicopy RCR plasmids (usually smaller than 10 kb) that can harbor one or two mob genes and one origin of transfer (the oriT). These plasmids can be horizontally transferred, employing functions provided by larger plasmids, and are the focus of this review. Replication by the RC mechanism was first reported in the E. coli bacteriophage ϕX174, and soon after in coliphages M13, f1, and fd, although these latter single-stranded DNA (ssDNA) phages showed differences from ϕX174, especially in the linkage between the Rep proteins (replication initiator proteins) and their target DNAs (53). It was later on that the pioneering work done in the laboratories of Dusko Ehrlich (France), Richard Novick (United States), and Saleem Khan (United States) demonstrated that small staphylococcal plasmids used the RCR mechanism (54–56). Many of these RCR plasmids spread easily by HGT and can cointegrate with other RCR plasmids (57) and with larger θ-replicating plasmids (58). Thus, RCR plasmids can be mobilized among bacteria either by conjugation, transduction, or transformation, making it more complicated to avoid the spread of Ab^r traits mediated by these plasmids. Several thousands of RCR plasmids have been sequenced, and they have been found in many bacterial species, being most abundant in G+ bacteria. However, RCR plasmids have been found in Gram-negative (G−) bacteria, Archaea, and even in the mitochondrial DNA of higher plants (59–61).

Plasmid Classification

Classification of plasmids has been the work of many groups over the years, but three main criteria have been pursued based upon (i) the incompatibility groups, (ii) the replicon, especially the Rep proteins, and (iii) the conjugative relaxases, either Tra or Mob proteins (9, 62–64). Incompatibility groups were established on the basis that if two plasmids share the genes involved in replication and its control, they are unable to coexist within the same host, i.e., they are incompatible (65, 66). Plasmid families based on inc determinants were shown to be limited because single mutations in an antisense RNA involved in the control of plasmid replication rendered the plasmids to be compatible (67). The other two methods of plasmid classification, albeit still widely used, have been questioned recently because they do not cover all of the more than 10,000 sequenced plasmids (5, 63).

New paths to classify plasmids include (i) considering them as word embeddings by the use of natural language processing approaches (68) and (ii) employing a clustering approach based on mash distance (69). In both cases, the results were promising in attaining global plasmid classification by taking the MOB families as a reference. The exceptions were the MOB_Q and MOB_V families, perhaps due to underrepresentation in the chosen training set. In addition to the updated data provided here for the MOB_V family, we believe that more information should be gathered concerning the plasmid families. Holistic approaches have provided broader views of the plasmidome by the employment of two types of network approaches, based on either plasmid shared k-mer content (70) or on plasmid average nucleotide identities (71). Both of them retrieved and considered more than 10,000 plasmid genomes and revealed the existence of discrete groups that gather plasmids with a common genomic backbone while pointed to HGT drivers.

In the case of RCR plasmids, their main peculiarity is that they have a few gene cassettes so that combinations of different modules would yield a new plasmid. Classification of these plasmids is related to their Rep proteins or, when present, their Mob proteins. In both cases, the family encompasses the proteins and their cognate DNA targets, namely, the origins of leading-strand synthesis (dso, double-strand origin of replication) and transfer (oriT), respectively (see “The Family, Yesterday and Today” below).

Features

The most common features of all RCR plasmids include the following: (i) replication by the RC mechanism, (ii) generation of ssDNA intermediates corresponding only to the leading strand, (iii) average copy number (N_av) of ∼20 to 30 copies per genome equivalent, and (iv) small size, ranging between the 846 bp of the Thermotoga plasmids pRQ7, pMC24, and pRKU1 (72–74) to the ∼30 kb of pCG4 from Corynebacterium glutamicum, which may be an unusual RCR plasmid because it harbors a class I integron (75). Generally, the maximum size of RCR plasmids is not greater than 10 kb (Table 1). Depending upon their size, RCR plasmids harbor one cassette for the replication of the leader strand (leading strand initiation and copy number control, LIC module), one or two loci involved in the replication of the lagging strand (sso, single-strand origin of replication), one module containing the Ab^r gene (antibiotic resistance determinant, DET), and one module involved in the plasmid mobilization (MOB). Exceptions can be found especially concerning the MOB region, which has been found in only a fifth of the RCR plasmids (see “The Family, Yesterday and Today”). The smallest plasmids, like those found in Staphylococcus aureus (plasmid pSN2, 1.3 kb [76]), Mycoplasma (plasmids pADB201, 1.7 kb [77], and pKMK1, 1.8 kb [78]) or in the haloarchaeon Natrinema sp. (plasmid pZMX201, 1.6 kb [79]), encode only the initiator of replication protein (generically termed Rep) and the components involved in copy number control, that is, the basic replicon.

TABLE 1.

Distribution of RCR plasmids in the RefSeq (version 90) database

RCR family	No. of plasmids	Size range in kb (median)	Bacteria or Archaea phyla (no. of plasmids)	Main taxonomic orders (no. of plasmids)
Rep_1	689	1.0–2,849 (9.9)	Firmicutes (368), Proteobacteria (272), Actinobacteria (26), Cyanobacteria (9), Bacteroidetes (9), uncultured bacteria (2), Spirochaetes (1), Euryarchaeota (1), Chlamydiae (1)	Bacillales (284), Enterobacterales (137), Lactobacillales (70), Rhizobiales (39)
Rep_2	163	0.87–1,770 (4.3)	Firmicutes (117), Tenericutes (28), Proteobacteria (14), Bacteroidetes (2), Cyanobacteria (1), uncultured bacteria (1)	Lactobacillales (88), Bacillales (24), Acholeplasmatales (16)
Rep_trans	224	1.5–2,849 (4.6)	Firmicutes (175), Proteobacteria (34), Cyanobacteria (6), Actinobacteria (4), Bacteroidetes (2), Spirochaetes (1), Thermotogae (1), Euryarchaeota (1)	Bacillales (93), Lactobacillales (82), Neisseriales (7), Pseudomonadales (7)

Replication by the RC Mode

Replication by the RC mechanism involves the following stages: (i) on a supercoiled double-stranded (ds) plasmid DNA molecule, the Rep initiator binds to a strand-specific DNA site within dso; (ii) binding of the Rep protein induces structural changes in the target DNA, like a DNA bend, and leads to protrusion of a stem-loop structure; (iii) an aminoacyl residue of Rep (a Tyr residue in most cases studied) produces a nucleophilic attack on a specific dinucleotide within dso, the nick site; (iv) a covalent aminoacyl DNA molecule is formed, whereas the free 3′-OH end of the cleaved strand is recognized by the host DNA polymerases to initiate leading-strand synthesis (60, 61, 80). Hence, RCR and conjugative DNA transfer are mechanistically similar processes since DNA-relaxing proteins (initiator of replication and relaxase) and recognition of a DNA site in the plasmid target (dso and oriT) participate in the initiation reactions (81, 82). The catalysis performed by the Rep initiators and Tra/Mob proteins is the same, although the identity of the nucleophilic residue may vary. Usually, the enzymes participating in this type of catalytic reaction belong to the superfamily of proteins known as HUH (His-bulky hydrophobic residue-His). Nucleases of this family participate in the breaking and joining of ssDNA in processes such as RCR initiation and termination (plasmids and coliphages), DNA transfer (plasmids and MGEs), transposition (nonreplicative MGEs), and replication of plant and animal viruses (83).

In the RCR mechanism, the initiation of plasmid replication is independent of the synthesis of a primer RNA (pRNA), a step that is required in the strand displacement replication and the θ-replication modes, the other two common replication mechanisms of circular plasmids (60, 61, 80). The next stage, elongation, is performed by the host machinery, while Rep protein remains covalently attached to the 5′ end of the parental strand; DNA polymerase III (DNA Pol III) performs the synthesis of newly replicated DNA, the helicase PcrA opens the strands in the replication fork, and the ssDNA-binding protein (SSB) covers and protects the cleaved nontemplate strand (60, 61, 80). Replication of the leading strand is finished through a new nucleophilic attack after the nick site is reconstituted, a reaction that is probably performed by the Rep molecule that initiated replication and that would have moved together with the replication machinery (84). The final reaction would bring together the initial 5′ end with the 3′ end thus generated, through a trans-esterification reaction that releases the parental strand in an ssDNA configuration and the newly synthesized leading strand as relaxed dsDNA. A by-product of this stage is one subunit of the Rep protein bound to a short segment of ssDNA (∼10 nt long) that is unable to reinitiate replication, which does not happen in the case of coliphages (85). The generation of the strand-specific ssDNA replication intermediates is the tell-tale sign of all RCR plasmids, and its experimental detection in a newly identified plasmid is the demonstration of its replication by the RCR mode (56, 86, 87). The ratio of the ssDNA intermediates to the dsDNA plasmid forms indicates the replication efficiency: the higher the ratio, the lower the plasmid replication efficiency (86). The ssDNA intermediates are later converted into dsDNA plasmid molecules by the host RNA polymerase (RNAP), DNA Pol I, and, most likely, DNA Pol III upon recognition of a specific region (the single-strand origin, sso), as described below. Finally, DNA ligase and DNA gyrase would seal and supercoil, respectively, the two circular DNA molecules resulting from this mode of replication.

RCR Plasmid Families

Plasmids replicating by the RC mode were first reported for the staphylococcal plasmids pC194 and pT181, because of the generation of ssDNA forms by the former plasmid, and to the specific nicking-closing activity of the Rep protein encoded by the latter plasmid (55, 56, 88). RCR plasmids were soon reported to exist in several bacterial species, such as Staphylococcus (pC221, pUB110, pE194), Streptococcus (pMV158), Streptomyces (pIJ101), and Mycoplasma (pHPK255), and in many other bacteria and archaea (61, 89). When the number of RCR plasmids grew substantially, a small informal gathering was held during the Third Plasmid Biology Meeting (Spain, 1992), where the generic names of the various genes and loci of RCR plasmids were defined and accepted. However, the first serious attempt to elaborate a database with the known RCR plasmids and family classifications was initiated by Chris M. Thomas, under the umbrella of the European Science Foundation-sponsored meetings of the Network on Plasmid Molecular Biology and Ecology (1995–1997) and by Mark Osborn, who built the first database of plasmid replicons (90), later discontinued. It was the replicon cassette that defined the various RCR plasmid families, as early noticed by Koonin and Ilyina. They performed an in-depth comparison of the Rep and the Mob proteins that allowed a detailed overview of the features of these proteins (91, 92).

According to the replicon cassette, the RCR plasmids have been grouped into several families, the three most studied ones being the following: (i) Rep_1 (PF01446), whose prototypes are the staphylococcal plasmids pC194 and pUB110; (ii) Rep_2 (PF01719), of which the streptococcal plasmid pMV158 and the staphylococcal plasmid pE194 are the representatives, and (iii) Rep_trans (PF02486), represented by the staphylococcal plasmids pT181 and pC221 (60, 93–96). The prototypes of these three families harbor all the genetic modules (LIC, DET, and MOB; see above) except plasmid pC194, which contains neither the MOB module nor the ssoU origin, the latter being present in plasmids pUB110 and pMV158 (81). A fourth RCR family of replicons, exhibiting a Rep_trans domain linked to a helix-turn-helix (HTH) motif at the N-terminal region of the protein, has been reported. These replicons have been grouped within the MOB_T family of relaxases, whose prototypes are ORF20 of Tn916 and Nick of ICEBs1. These elements, despite being ICEs, may also use their MOB_T relaxase as a Rep initiator and their oriT as an origin of replication (97). Other RCR plasmid families (up to a total of 17) have been proposed to exist, although less experimental knowledge on these replicons has accumulated over the years. These include at least the families represented by the staphylococcal plasmid pSN2 (family RepL; PF05732) (94), the corynebacterium plasmid pCG1 (family Replicase, PF03090) (98), and plasmid pGRB1 from archaea (family DUF1424; PF07232) (79, 99). However, it is still premature to classify these plasmids in so many families because there is a lack of information on their biochemistry and molecular biology features. Integrated versions of RCR plasmids can be also found in the chromosomes. Such is the case of IMESag-rpsl from S. agalactiae. It harbors, in addition to a mobilization region (a mobM gene and an oriT highly homologous to those of pMV158), plasmid genes involved in RCR and its control that belong to the Rep_1 family (15). This particular IME is kept as an integrated copy in the bacterial chromosome, but it can excise and produce circular copies that can replicate and be transferred to other bacteria, constituting one more example of the faint lines dividing RCR plasmids from other MGEs (15).

Through a bioinformatics survey of the RefSeq (version 90) database (Table 1), we found that the RCR plasmids of the Rep_1 family were, by far, the most abundant (689 hits), and they populated the Bacillales. The Rep_trans family (224 hits) exhibited an even distribution between Lactobacillales and Bacillales. The Rep_2 family represented by pMV158 retrieved 163 RCR plasmids, hosted mostly by the Lactobacillales (81).

PLASMID pMV158 AND RELATIVES

Plasmid pMV158 (GenBank accession no. NC_010096.1 and MZ396393; the coordinates used in this review refer to the sequence associated with the latter accession number) (Fig. 1A) is a most promiscuous RCR plasmid first isolated from S. agalactiae (100) that has been the subject of our interest for the past 30 years. During this time, we have investigated the plasmid’s replication and control mechanisms, as well as its conjugative transfer among bacterial species; it has also been used as a model for structure-function studies (80). Plasmid pMV158 has 5,540 bp and G+C content of 37%, which agrees with the similar G+C content of its streptococcal host. There are, however, some regions located at the dso and oriT origins where the G+C content increases up to 45%, which is a common feature for plasmids of the pMV158 family (101). The pMV158 plasmid, or some of its derivatives as shown by our records, has been transferred by natural or artificial transformation and by conjugative mobilization with helper plasmids to several bacterial genera, such as streptococci, lactococci, bacilli, corynebacteria, clostridia, and enterococci, as well as Listeria monocytogenes, S. aureus, and the G− host E. coli (Fig. 1B). Although we have not conducted a systematic study, pMV158 has been mobilized between G+ bacteria by plasmids of the Inc18 family (pIP501, pAMβ1) as helper plasmids (102), whereas IncP and IncW plasmids mobilize pMV158 to E. coli (103). Many of the studies between 1986 and 1997 were done with a derivative of pMV158, termed pLS1, which was constructed by the deletion of a 1.1-kb EcoRI fragment (Fig. 1A). The deletion, at the time unknowingly, resulted in a plasmid lacking the lagging-strand origin ssoU and mobilization functions (104). Later on, we extended our research to the pMV158 parental plasmid to study its conjugative mobilization (105).

FIG 1.

Features of plasmid pMV158 (GenBank accession no. MZ396393). (A) Schematic map of the plasmid (scale 1:18.47) showing its relevant features, as described in the text. The positions of the different known genes are depicted within the rectangle by arrows, with arrowheads pointing to the direction of transcription. Promoters are indicated by red arrows, whereas the positions of the bent sequences are depicted by horizontal arrowed blue lines. Hairpins are shown below the rectangle. The positions of the plasmid origins of replication (dso, ssoU, ssoA) and transfer (oriT) are indicated by colored rectangles. LIC module, replication and control; DET module, constitutive resistance to tetracycline; MOB module, conjugative mobilization. The deletion in the pLS1 derivative of plasmid pMV158 is indicated. (B) Hosts in which plasmid pMV158 or its derivatives have been shown to replicate.

The experimentally determined N_av value of pMV158 was always higher than that of pLS1, a circumstance that was overlooked over the years until we discovered unsuspected cross talks between plasmid modules (106) (see “The Unexpected: Cross Talk of MobM Relaxase with RepB Initiator” below). The N_av of pMV158 varies among the hosts in which it replicates: ∼30 copies in S. pneumoniae, ∼20 in Enterococcus faecalis, and ∼7 in B. subtilis and E. coli. However, whereas the plasmid is stably inherited in the absence of selective pressure in the three G+ hosts, it is not in E. coli. In this host, pMV158 has a high experimental rate of plasmid loss per cell and generation: less than 0.1% of E. coli cells harboring pMV158 retain the plasmid after growing in the absence of selective pressure for 50 generations (107).

The Family, Yesterday and Today

We have previously inspected the pMV158 Rep_2 family attending to the replicon (81). At the time, a PSI-BLAST analysis was done using the RepB initiator of pMV158 as the initial query. Seventy-eight plasmids were retrieved. Most of them (55 replicons) did not encode any relaxase, whereas the other 23 harbored a putative relaxase of the MOB_V family. Many of the plasmids of the pMV158 family did not seem to encode a clear marker, but resistance to several antibiotics and even a gene encoding arsenate resistance were detected.

In this review, we decided to revisit the relaxase MOB_V plasmid family represented by pMV158 (9) and to perform an in-depth study of the present family status. To this end, we have explored a data set of 13,798 archaeal and bacterial plasmids contained in NCBI RefSeq (release 90). These data were analyzed for the presence of MOB_V relaxases by using MOBscan (108), which led to the detection of 434 MOB_V plasmids. To compare these plasmids and evaluate their distribution, we assessed the gene families shared between genomes by using AcCNET (109). This bioinformatics tool generates bipartite networks connecting two types of nodes, some representing plasmid genomes and others the homologous protein clusters. If a plasmid encodes a member of a given protein cluster, a link connects them. As a result, plasmids encoding related proteins assemble in dense groups. A total of 13,948 proteins contained in the 434 MOB_V plasmids were used in the analysis. Proteins were included in the same cluster if they were >80% identical and covered >80% of the protein alignment, producing a total of 8,937 homologous protein clusters. By applying the ForceAtlas2 algorithm (110) implemented in Gephi (111) with null edge weight influence, the resulting bipartite network of MOB_V plasmids (Fig. 2) was visualized and the plasmid nodes were colored by their host taxonomy. Independent groups of dense clusters of MOB_V plasmids emerged in the bipartite network, indicating that a great variety of plasmid backbones can be found in the MOB_V class. At least 12 plasmid taxonomic units (PTUs) were previously identified within the MOB_V class (71). MOB_V plasmids were distributed in at least six phyla: Firmicutes (mostly in orders Bacillales and Lactobacillales), Proteobacteria (mostly Rhodobacterales, Rhodospirillales, and Pseudomonadales), Cyanobacteria (mostly Nostocales), Bacteroidetes, Spirochaetes, and Tenericutes, showing that MOB_V plasmids are not restricted to monoderm hosts. Plasmids from different phyla and even genera are clustered apart. Only 21 homologous protein clusters connected plasmids from two different phyla, and 43 linked plasmids from two different orders within a phylum. We interpret this finding as an indication of low propagation of specific MOB_V backbones through high taxonomy rank hosts.

FIG 2.

Taxonomic distribution of MOB_V plasmids. AcCNET bipartite network with >80% protein identity and >80% protein alignment coverage. Networks were visualized with Gephi, using the ForceAtlas2 algorithm (110). Homologous protein clusters (8,937) are colored in gray. Plasmid nodes (434) are colored by the host taxonomy: phylum (A) or order (B).

MOB_V RCR plasmids are <100 kb and especially abundant in the size range of <10 kb (Fig. 3A). Plasmids of the MOB_V class are enriched in RCR initiators, since 186 of the 434 MOB_V plasmids were found to encode at least an RCR initiation protein, either of the Rep_1 (93 hits), Rep_2 (44 hits), or Rep_trans (55 hits) superfamilies (Fig. 3B). Seventy RCR protein clusters (35 from Rep_1, 21 from Rep_2, and 14 from Rep_trans) are present in the network (Fig. 4), distributed in MOB_V plasmids from Firmicutes and Proteobacteria. Curiously, a high rate of MOB_V plasmids encodes more than one relaxase (46 out of the 434; 38 plasmids encode two, 7 encode three, and one encodes 4), mostly from the MOB_V, MOB_P, and MOB_Q classes. This could be an indication of frequent cointegrate formation by the MOB_V plasmids.

FIG 3.

Overlap between MOB_V and RCR plasmids. The presence of MOB_V relaxases was evaluated with MOBscan (108), while RCR initiation proteins of the Rep_1 (Pfam PF01446.17), Rep_2 (PF01719.17), and Rep_trans (PF02486.19) families contained in the RefSeq90 plasmid data set were detected by using HMMER 3.1b2 (301). (A) Distribution of replication initiators of MOB_V plasmids according to plasmid size, depicted with the ggplot2 R library. (B) Venn diagram showing the distribution of RCR initiators Rep_1 (pink), Rep_2 (green), and Rep_trans (yellow), as well as MOB_V plasmids (blue). It was built by using the online tool http://bioinformatics.psb.ugent.be/webtools/Venn/.

FIG 4.

Distribution of RCR initiator proteins in the MOB_V plasmid data set. AcCNET bipartite network with >80% protein identity and >80% alignment coverage. Protein clusters of RCR initiators of the Rep_1, Rep_2, and Rep_trans families are highlighted in red, green, and yellow, respectively. The rest of the protein clusters are colored in gray, while nodes corresponding to MOB_V plasmids are in black.

As explained above (see Introduction), besides the relaxase, a T4CP and an Mpf system are also required for a plasmid to be self-transmissible by conjugation. Eight Mpf types are currently defined (112). Only 23 of 434 MOB_V plasmids present in the analyzed RefSeq data set also contained T4CP and Mpf components, according to MacSyFinder (113) (Table 2). MOB_V seems to be a MOB relaxase class specialized in mobilizable plasmids. The exceptions were the MOB_V plasmids hosted in Cyanobacteria: 14 out of 37 are potentially conjugative. One hundred fifty-two out of the 8,937 protein clusters contained MOB_V relaxases, anticipating the wide variety of MOB_V sequences that is illustrated in the phylogenetic tree (Fig. 5). Distinctive relaxase clades included plasmids mostly hosted in a single phylum with only a few incursions of phylogenetically close relaxases from a different taxonomic origin, in congruence with the low interphylum connectivity described above. Nevertheless, relaxases from plasmids encoding different RCR initiator families usually appeared intermingled at short evolutionary distances, suggesting the occurrence of recombination events between plasmid modules.

TABLE 2.

Conjugative systems in MOB_V plasmids

Plasmid accession no.	Size (bp)	Plasmid host (phylum)	MPFa
NC_017961.1	36,262	Enterococcus faecium (Firmicutes)	MPFFA
NC_005024.1	46,445	Staphylococcus aureus (Firmicutes)	MPFFATA
NC_013320.1	50,429	Staphylococcus aureus (Firmicutes)	MPFFATA
NC_012547.1	54,000	Staphylococcus aureus (Firmicutes)	MPFFATA
NZ_CP026963.1	61,643	Staphylococcus aureus (Firmicutes)	MPFFATA
NZ_CP022484.1	65,180	Enterococcus faecium (Firmicutes)	MPFFA
NZ_CP028186.1	66,602	Campylobacter jejuni (Proteobacteria)	MPFT
NZ_CP020115.1	92,326	Nodularia spumigena (Cyanobacteria)	MPFC
NZ_CP026693.1	121,770	Nostoc sp. (Cyanobacteria)	MPFC
NC_019739.1	146,045	Microcoleus sp. (Cyanobacteria)	MPFC
NZ_CP024790.1	152,318	Nostoc flagelliforme (Cyanobacteria)	MPFC
NZ_CP025539.1	185,327	Vibrio harveyi (Proteobacteria)	MPFF
NZ_AP018211.1	190,194	Calothrix brevissima (Cyanobacteria)	MPFC
NZ_AP018235.1	194,818	Fremyella diplosiphon (Cyanobacteria)	MPFC
NC_016983.1	204,052	Photobacterium damselae (Proteobacteria)	MPFF
NC_010632.1	254,918	Nostoc punctiforme (Cyanobacteria)	MPFC
NZ_AP018219.1	255,223	Anabaena variabilis (Cyanobacteria)	MPFC
NZ_AP018300.1	312,401	Fischerella sp. (Cyanobacteria)	MPFC
NZ_AP018186.1	320,476	Nostoc sp. (Cyanobacteria)	MPFC
NZ_AP018187.1	328,926	Nostoc sp. (Cyanobacteria)	MPFC
NZ_AP018185.1	366,120	Nostoc sp. (Cyanobacteria)	MPFC
NZ_AP018217.1	391,340	Anabaena variabilis (Cyanobacteria)	MPFC
NC_003276.1	408,101	Nostoc sp. (Cyanobacteria)	MPFC

^aMPF, mating-pair formation type according to MacSyFinder (113).

FIG 5.

Phylogenetic tree of MOB_V relaxases. The relaxase domains of the 334 nonidentical MOB_V proteins (N-terminal 300 residues) were aligned using MUSCLE v3.8.31 (302). Positions with gaps in more than 20% of the proteins were removed from the alignment with Trimal 1.2rev59 (303). ModelFinder was used to identify the substitution model with the best fit to the data, according to the Bayesian information criterion (304). Using the VT+F+R9 model, a maximum-likelihood phylogenetic reconstruction was inferred with IQ-TREE version 1.6.1 (305). Branch support was estimated by performing 1,000 replicates of the SH-like approximate likelihood ratio test (306) and 1,000 ultrafast bootstrap (UFBoot) approximation replicates (307). The resulting tree was visualized and annotated using the online tool iTOL (308). The TraI relaxase of the MOB_P plasmid RP4 was used as an outgroup for rooting the tree, and its branch is shown as a discontinuous line. UFBoot support values ≥95% are shown. The prototype MOB_V relaxase, the MobM protein of pMV158, is highlighted in green. The inner ring indicates the presence of RCR initiators in the corresponding plasmids. The outer ring indicates the phylum of the plasmid bacterial host.

Beyond the RefSeq data set, MOB_V relaxases are also abundant in environmental, animal, and human metagenomes. Querying the MGnify database (https://www.ebi.ac.uk/metagenomics/) with the N-terminal 300 amino acids of MobM_pMV158 (accession no. WP_012218462.1) (https://www.ebi.ac.uk/metagenomics/sequence-search/search/phmmer) (E value = 0.01) retrieved hits in soil (3), marine (193), and freshwater (271), as well as the animal (400), human nondigestive (646), and human digestive (7,766) microbiomes.

GENETIC ORGANIZATION AND FEATURES

Like most RCR plasmids, pMV158 shows extremely compact information (Fig. 1): only 30 bp separates the transcriptional terminator sequence of the copG-repB operon and the −35 region of the tetL promoter, and 60 bp separates the UAA termination codon of copG (formerly repA) and the AUG initiation codon of repB, and even this short region encodes (on the complementary strand) the 46-nt-long antisense RNA II. Furthermore, the two components participating directly in the control of plasmid replication, the transcriptional repressor protein CopG and the antisense RNA II, are the smallest plasmid-encoded replication control molecules so far reported, this being a general feature of RCR plasmids. We were surprised to find two “silent” regions of about 300 bp each that might keep some family secrets still to be fully understood (see “Family Secrets: the Unknown ORF109”). Plasmid pMV158 and relatives are composed of, at most, three distinct gene cassettes involved in (i) replication and control (LIC module), (ii) constitutive resistance to tetracycline of the tetL type (DET), and (iii) conjugative mobilization (MOB; see “Features” above). The LIC module includes three genes: copG, which encodes the 45-amino-acid CopG protein (Cop in the case of pE194 [114, 115]); repB, which is cotranscribed with copG in a single copG-repB mRNA; and rnaII, encoding a small antisense (countertranscribed) RNA that participates with CopG in the control of RepB synthesis and, thus, in the N_av value of the plasmid. The efficiency of transcription/translation of these genes by the bacterial machinery influences the plasmid copy number in different hosts. The DET module includes only the gene encoding the Ab^r protein, a type L tetracycline determinant. The MOB cassette includes the mobM gene that encodes the MobM relaxase, which autoregulates its synthesis and participates in the mobilization of pMV158 and also in the control of its N_av. Furthermore, pMV158 (Fig. 1A) harbors the following loci: (i) the origin of leading-strand replication (dso), within the LIC module; (ii) the origin of transfer (oriT), within the MOB cassette; and (iii) two lagging-strand origins of replication (ssoU and ssoA) (81, 89). Cross talks between elements participating in pMV158 replication and mobilization have been described previously (106, 116). In sum, the known genes harbored by pMV158 encode four proteins (coding strand) and two RNAs (complementary strand). The proteins are the transcriptional repressor CopG, the initiator of replication RepB, the antibiotic resistance determinant TetL, and the relaxase MobM. The two RNAs are RNA II, antisense to the copG-repB mRNA, and RNA I, for which we have little information (117), although some possible roles are proposed (see The Mysterious RNA I).

Promoters, Terminators, and Ribosome Binding Sites

The promoters and transcription initiation sites of the pMV158 genes have been experimentally determined, except for the promoter involved in the transcription of the tetL gene that remains putative, although indirect evidence points to the predictions made earlier (104) (Fig. 1A and Table 3). The promoters of the copG-repB, tetL, and mobM genes exhibit the −10 extended sequence (118, 119), which is present in several pneumococcal promoters (120). The canonical sequences 5′-TTGACA-3′, and 5′-TATAAT-3′ for the −35 and −10 regions, respectively, were found in the promoter of the rnaI gene (Table 3), although its spacer was suboptimal (18 nt), reducing its transcription rate (117). Further, the promoters directing the synthesis of the copG-repB and mobM mRNAs and the promoter P_ctII for the synthesis of the antisense RNA II are subjected to transcriptional regulation (106, 121–123). Given these facts, we consider that all plasmid promoters have been selected as optimized for their function, according to the early concept raised by Deuschle et al. (124).

TABLE 3.

Transcription signals in plasmid pMV158

Gene(s)	Promoter(s)	−35 sequencec	Spacer length (nt)	−10 sequencec	Reference
copG-repB	Pcr	TTGAtt	17	TATGCTActAT	255
tetL	Ptet	TTGACA	16	TGTGGTAaAAT	104
mobM	E. coli: Pmob2	TgGAag	17	TGTGCTAaAcT	122
	L. lactis: Pmob1	aTGAat	16	TGTGtTATAcT	311
	S. pneumoniae a	Both promoters		Both promoters	123
rnaI	PI	TTGACA	18	TGtTATAAT	117
rnaII	PctII	TTGctt	17	TATAAT	121
ssoA b	PA	TTGACA	16	TAgcgT	147
ssoU b	PU	TaGAaA	17	aAattT	199

^aThe pneumococcal SigA protein recognized both promoters.

^bOn ssDNA.

^cNonconsensus nucleotides in the sequence are shown in lowercase letters.

Termination of transcription by RNAP at Rho-independent sequences (intrinsic termination) is achieved without the need for accessory factors (125–127). This mode of termination requires that the transcribed RNA folds and generates a stem-loop structure, followed by a set of uridines at the 3′ end of the nascent RNA. These two facts facilitate the release of the RNAP at the intrinsic termination sites. In pMV158, evidence of transcription terminators derives from direct in vitro studies in the case of the T_II terminator of RNA II (128) and from in vivo measurements in the case of the tetL termination sequences (129). In the former case, a termination efficiency of ∼85% was experimentally determined (128), coinciding with the predicted efficiency (130). There are no experimental data on transcriptional terminators for the other pMV158 RNAs, except those derived from theoretical calculations (130).

Concerning the signals for the initiation of translation, all the predicted ribosome binding sites (RBS) fall within the topmost-used Shine-Dalgarno (SD) motifs and are located at the most used distance of 6 nt upstream from the initiation codon (131). The SD sequence of copG has a consensus hexanucleotide (5′-GAGAGG-3′). The tetL mRNA has two SD sequences (5′-GGAGGA-3′ and 5′-GGAGGG-3′) due to the existence of a truncated leader peptide (see The Antibiotic Resistance Genes). In the case of mobM, the consensus spans 8 nt (5′-AAGGAGGG-3′), making it a canonical pneumococcal SD sequence (132). This is not the case for repB translation, where no clear RBS sequence was observed (104). It was later proposed that the sequence 5′-AUUUCU–4-5 nt–UAUA–9-10 nt–AUG-3′ (coordinates 830 to 855), which is also present in other pneumococcal genes, could be the signal recognized by the ribosomes so that it was considered an atypical ribosome binding site (ARBS) (133). The ARBS sequence was also found in the pneumococcal dpnM gene, but it was later discovered that, in this case, it was an extended −10 promoter region (120). The region was revisited in light of the interactions between the copG-repB mRNA and RNA II (134), and a weak SD sequence (5′-GGGU-3′; coordinates 837 to 840) was proposed (116). This signal would account for the observed low levels of RepB protein synthesis, as expected for a protein subjected to dual (CopG and RNA II) regulatory control (135). Lacking further experimental data, like detailed mutagenesis of the region, ribosome toeprinting (136), or ribosome profiling (137), the existence of such a weak SD region is reasonable at our present degree of knowledge.

Supercoiling and Inverted Repeats

Supercoiling of plasmid DNAs reflects, in general, the supercoiling of the host chromosome. It represents underwinding of DNA strands, i.e., negative supercoiling. However, transient regions of positive supercoils (overwinding) can occur depending upon several plasmid functions, like replication or transcription (138). These changes in supercoiling would affect many DNA-protein transactions, since bound proteins can also affect the conformational changes in the overall supercoiling of the plasmid molecules. The presence of a negative DNA supercoiling conformation in RCR plasmids leads to torsional stresses that must be released at regions containing inverted repeats that generate stem-loop (hairpin) structures. Such tensions may affect replication because of the need for helix unwinding at regions close to the places where replication initiates. The release of these stresses would lead to a dynamic process of generation of local hairpin secondary structures that could, somehow, condition plasmid replication (139). In the case of the RepB initiator of pMV158, it has been shown that in vitro cleavage of cognate or related RCR plasmid DNAs (e.g., pE194 and pFX2) depends upon the degree of plasmid supercoiling (101). In most RCR plasmids where dso has been mapped, transcription units that run parallel or divergent of such an origin of replication generate waves of negative supercoiling that favor extrusion of hairpins around the region.

The generation of secondary structures on supercoiled pLS1 DNA (deletion derivative of pMV158) (Fig. 1A) was determined in vitro by susceptibility to endonuclease S1, because this enzyme digests preferentially unpaired DNA like the loops of hairpins (140). Mapping of the resulting fragments revealed the existence of three major stem-loop structures placed at coordinates ∼450 (hairpin I), ∼410 (hairpin II), and ∼100 (hairpin III) (Fig. 1A). Hairpin I locates near the RepB-mediated cleavage site, which is located at the dinucleotide 5′-GpA-3′ (coordinates 448 to 449). The stem of this hairpin consists of a stretch of seven G’s paired with seven C’s, making it one of the highest G+C contents of the entire plasmid. The base of the stem of hairpin II locates just 5 bp upstream of hairpin I, whereas hairpin III locates further upstream and is the main hairpin of the entire plasmid. This latter secondary structure was generated on supercoiled DNA but also coincides with the putative terminator sequence of RNA I (117). Partial or total removal of the ssoA region resulted in plasmids with reduced N_av, increased instability, and a high ssDNA/dsDNA ratio, because plasmid pLS1 lacks ssoU. Furthermore, plasmids with intact ssoA but lacking hairpin III exhibited twice the number of copies than the parental plasmid, but they did not show changes in replication efficiency (measured as ssDNA/dsDNA ratios) (140). It is thought that the generation of hairpin I exposes the RepB nick site in an ssDNA configuration, which is the substrate required by the initiator RepB to cleave its target DNA. The role of hairpins II and III would be as competitors of hairpin I, because the extrusion of either of the former would hinder the extrusion of the latter, regulating in cis the initiation of replication. Alternatively, divergent transcription from promoter P_I could generate waves of supercoiling directed toward hairpin I, affecting the overall structure of the plasmid dso and, consequently, the plasmid copy number (60).

The dso (origin of replication) and oriT (origin of transfer) regions of pMV158 accumulate most of the DNA secondary structures. In addition to the hairpins I, II, and III located in the vicinity of dso, three inverted repeats (IRs) located in the oriT region (coordinates 3564 to 3606) were experimentally mapped. Such IRs are located upstream of the mobM gene. Unlike the hairpins of dso, the three IRs of oriT partially overlap and are distributed in such a way that the extrusion of one of them would preclude the extrusion of the other two, placing the dinucleotide cleaved by MobM (5′-GpT-3′, coordinates 3595 to 3596) in different positions (141).

Direct Repeats

In pMV158, there are three consecutive 11-bp direct repeats at dso (5′-TCGGCGACTTT-3′; coordinates 533 to 565); they were termed downstream direct repeats (DDRs). It was considered unlikely that DDRs performed a role similar to the iterons present at the origin of replication of θ-replicating plasmids, which participate in the control of replication by the “handcuffing” mechanism (142). DDRs at the dso were found in all members of the Rep_2 family of plasmids, although their numbers and lengths differed, from two DDRs and 7 bp (pHPK255 from Helicobacter pylori [143]) to three DDRs and 21 bp (pADB201 from Mycoplasma mycoides [77]). The pMV158 DDRs are located 84 bp downstream of the RepB-generated nick site, and they encompass three DNA-helix turns. DDRs are placed in the same face of the DNA helix with respect to the nick site, and they are the preferential binding site of the RepB initiator. The region was termed the bind locus, whereas the region that contains the nick site was termed the nic locus (144). Attempts at introducing half a helix turn between the bind and nic loci failed, strongly suggesting the need for proper positioning of the RepB initiator to its cleavage site (M. Moscoso and M. Espinosa, unpublished). More recent attempts showed that an 11-bp insertion between DDRs and the nick site could be introduced but not 4- or 5-bp insertions (145). The contacts of RepB with its target DNA have been established by high-resolution footprinting experiments (146). These assays revealed that RepB also contacts two consecutive 7-bp direct repeats (5′-GTGCCPuA-3′) at coordinates 461 to 474, albeit with lower affinity. The new locus, termed the proximal direct repeats (PDRs), is located in phase with the nic locus. It has been proposed that PDRs were part of the locus bound by the catalytic RepB protein (146). Computer-assisted analyses of the dso region of the pMV158 plasmid family showed that they all share a similar organization: the highly conserved nick sequence followed by two and three direct repeats (146).

Other Repeats

Two sequences within the pMV158 “silent region” are repeated in other regions of the plasmid. First, there is a 14-bp sequence at coordinate 281 (5′-AAAGGGCGCGTTTG-3′) that is identical to the one starting at coordinate 3194. The latter sequence is placed at the beginning of the ssoU origin and is removed in plasmid pLS1 (Fig. 1A). We found these sequences in lactococcal plasmids pLd9 (GenBank accession no. MF150537.1; 15,313 bp) and pLd8 (accession no. MF150536.1; 47,015 bp) and shorter (∼12 bp) ones in many chromosomes of Firmicutes. The role of these sequences remains unknown. Second, there is an 8-bp sequence at coordinate 154 (5′-ACGCTAAA-3′) that is complementary to a sequence (coordinate 5294) found at the ssoA origin (Fig. 1A). Six nucleotides of this sequence (5′-TAGCGT-3′) constitute the so-called CS-6 sequence, which is conserved in the ssoA of many RCR plasmids (86). The CS-6 sequence was shown to participate in transcription termination of the primer RNA (pRNA) that initiates lagging-strand synthesis (147) (see The Lagging-Strand Origins). In a region (coordinates 3230 to 3500) that includes ssoU and is just upstream of oriT, the sequence 5′-TTGCTGA-3′ is repeated five times, scattered along the region. This sequence is also present in plasmid pUB110, where it was reported to be important as a binding site for associating the plasmid with the B. subtilis membrane (148). Both pMV158 and pUB110 have in addition a 9-bp direct repeat (5′-TTTTGAGTG-3′) in such a region (coordinates 3249 and 3377 of pMV158). Whether these repeated sequences play any role in the biology of these plasmids is presently unknown.

DNA Curvatures and Protein-Induced DNA Bends

DNA sequences exhibiting at least four successive A-T pairs (A-tracts) are frequent in most genomes, and they contribute to several processes, like replication, recombination, transcription, and, in general, many protein-DNA interactions because of their mechanical features and angles at the base-pair stacking (149). When two or more A-tracts are located in phase with the DNA helical angle, a local curvature in the region is attained (150, 151).

The simplest experimental determination of curved regions in DNA fragments involves detecting anomalous migration when such fragments are electrophoresed on native 6% polyacrylamide gels at different temperatures; under these conditions, the center of the curved region can be calculated by using circular permutation assays (152) or electron microscopy (EM) (153, 154). The employment of such techniques in plasmid pMV158 revealed the existence of four regions exhibiting curvatures. Three of them were placed within the LIC region (153), whereas the fourth mapped to the MOB region (C. Nivet and M. Espinosa, unpublished). In LIC, two curved regions (Fig. 1A, bend 1 and bend 2) were located around coordinate 635, a region that includes DDRs and the P_cr promoter. Between coordinates 550 and 650, there is a long stretch of A-T tracts that are located at every helix turn through five consecutive turns. This region was later shown to contain the targets of RepB (bend 1) and CopG (bend 2). The DNA target of RepB is included within the bend 1 region, and, upon binding, the initiator was shown to bend the DNA, so that a nucleoprotein complex would be generated in a way that allows RepB to approach the nick site to initiate replication of the leading strand (146). In the case of CopG, the bend 2 region was defined by EM (154). This, in addition to the three-dimensional structure of CopG complexed with its DNA target (155) and to high-resolution footprints (156), demonstrated that CopG introduces a strong DNA bend of ∼120° at its binding region. The bend 2 region spans 48 bp where four CopG dimers bind and bend the DNA, hindering the binding of RNAP to the P_cr promoter. The third curved region within the LIC module (bend 3) (Fig. 1A) was located around coordinate 820, close to promoter P_ctII, although no A-T tracts were located around the region (153). The meaning of this curvature remained unknown until we recently discovered a MobM-mediated cross talk between the replication and mobilization modules of pMV158 (106). The MobM conjugative relaxase was shown to bind not only to its target oriT but also to a region encompassing promoter P_ctII (around coordinates 835 to 866). The binding of MobM to this region reduced the synthesis of the RNA II inhibitor and, as a consequence, relieved the negative control exerted by this element on the synthesis of RepB, the net aftermath being an increase in the N_av value (106). We now propose that the bend 3 region is the place where MobM would bind and silence, at least partially, the synthesis of RNA II. The fourth curved region observed in pMV158 (bend 4) (Fig. 1A) was located upstream of oriT, although its position was not precisely determined (C. Nivet and M. Espinosa, unpublished). Lacking more information, we speculate that the bend 4 region might facilitate the binding of MobM to its target oriT, either as the initiator of DNA transfer or as the transcriptional repressor of its synthesis (122).

Partition and Segregational Stability

It has been postulated that high (∼20)-copy-number plasmids neither encode nor need a partition system (94, 157, 158), and there are no indications that pMV158 is an exception. Moreover, multimer resolution sites have not been identified, as expected because termination of replication by the RC mode would not generate multimeric molecules. Generally, plasmids with the pMV158 replicon are stably inherited in nearly all hosts tested except for E. coli (107). However, several puzzling observations have been reported but not yet fully explained. They pertain to the pMV158 stable inheritance in different hosts. Whether or not the findings can be considered faithful segregation of plasmid copies between daughter cells is unclear (to quote Stanley Cohen at one of the Plasmid Biology meetings, “true partition is like true love”). In S. pneumoniae, pMV158 was shown to be fully stable (159). Segregational stability was related to the presence of a functional sso region, since the deletion of the two pMV158 sso (ssoA and ssoU) led to the accumulation of large numbers of ssDNA intermediates and loss of segregational stability in the absence of selective pressure (160, 161). However, segregational stability and accumulation of ssDNA molecules were separated into two different phenomena when some mutations in the ssoA origin of plasmids that lack the ssoU origin led to stable plasmids that, however, accumulated ssDNA intermediates (162).

Maintenance of pMV158 in B. subtilis was reported to be unstable in various strains of this host, a finding that was attributed to unknown functions or sequences placed within the mobM gene (163). However, it was found that a region encompassing the pMV158 ssoA could stabilize the inheritance of ΔssoU plasmids, despite ssoA being not functional in this host (160, 161). In E. coli, pMV158 and its derivatives have an N_av value high enough to be, at least theoretically, stably inherited by random distribution. However, such plasmids are extremely unstable and generate large amounts of ssDNA (159). This latter defect, but not the plasmid instability, could be corrected by the cloning of the lagging-strand origin from the ssDNA coliphage f1 (107).

Curiously, it has been reported that plasmid pLS1, a ΔssoU derivative from pMV158, confers a measurable genetic load to S. pneumoniae, since cells harboring pLS1 showed a reduction of ∼10% in cell growth rate compared with cells harboring the parental plasmid, even though the two replicons were segregationally stable and pMV158 has a higher N_av (164). A further interesting finding is the presence of toxin-antitoxin modules in some RCR plasmids and IMEs, which might account for stabilization of plasmid inheritance after transfer to a recipient host (15, 165).

THE MYSTERIOUS RNA I

One of the mysterious regions of pMV158 is the one that encompasses roughly coordinates 5330 to 5540 and 1 to 400 (Fig. 1A). It has been considered a “silent” region because it does not encode any polypeptide. With pMV158 having such a compact genetic structure in which there is little space between one gene and the next, it struck us as a “weird” piece of empty DNA. It was, however, found that at the middle part of the region (roughly between coordinates 80 and 200) is located the gene rnaI (117). It encodes RNA I, which is transcribed in the opposite direction concerning the plasmid mRNAs. RNA I cannot be defined as “antisense” because a “sense” target has not been found or as countertranscribed (166) because no mRNA has been detected in the corresponding region of the complementary strand. The rnaI gene, including its promoter, spans coordinates 194 to 82 of pMV158 (Fig. 1A). The existence of the gene was early predicted and proposed to participate in the control of plasmid replication because its putative secondary structure resembled the CopA RNA of plasmid R1, which participates in the control of replication of this plasmid (104, 167). However, it was later demonstrated that replication control of pMV158 is exerted by the combined action of the transcriptional regulator CopG and the antisense RNA II (121) and that RNA I did not participate, at least directly, in the control of replication of plasmid pLS1 (117). Transcription of rnaI is directed from a nearly canonical promoter, P_I, that has a spacer of 18 nt rather than the usual 17 nt found in pneumococcal promoters (120); however, the −35 and −10 boxes are canonical and, in addition, the −10 region shows the TGN extension that is present in most of the promoters identified so far in pMV158 (Table 3). The transcription initiation point of RNA I, its size (80 nt in length), and in vivo expression have been determined (121). Several mutations at the rnaI gene were performed, and the results showed no differences between the mutants and the wild type in terms of copy number, stability, or generation of ssDNA. Furthermore, by type I incompatibility tests, it has been ruled out that RNA I is an incompatibility determinant (168). We are left with the unanswered question of what is the role of RNA I in the biology of pMV158? It was earlier speculated that one role of this element might be to serve as a primer for a RecA-dependent lagging-strand replication when no sso is present in the plasmid (169), although to our knowledge, no experimental proof has been published. However, it was shown that (i) expression of the rnaI gene did not influence lagging-strand replication in plasmids with an efficient sso (117) and (ii) the presence or absence of rnaI did not lead to variations in the efficiency of lagging-strand replication in plasmids lacking the two sso loci of pMV158 (161), making the role of RNA I as a primer unlikely. Spanning coordinates 83 to 120 of pMV158 there is a perfect 16-bp inverted repeat that would generate the RNA I transcription terminator on RNA but also could generate hairpin III (170) on supercoiled plasmid DNA (Fig. 1A). This hairpin was mapped in vitro as the major secondary structure generated on supercoiled pLS1 DNA. Moreover, a region with intrinsic curvature was identified upstream of the rnaI gene (153). It was proposed that transcription of RNA I would generate waves of negative supercoiling directed toward the dso region, which, in turn, would facilitate the extrusion of hairpin I (117). Since hairpin I harbors the sequence recognized by the RepB initiator, its extrusion would facilitate the cleavage reaction to initiate leading-strand replication. It is worth stating that the experiments on RNA I were performed with plasmid pLS1 but never with the parental pMV158, and we know now that cross talks between the MOB and the LIC modules do exist (106). New approaches could be used to study the role of RNA I in the biology of pMV158, such as employment of “complexomics” like that recently employed to detect RNA-protein interactions in S. pneumoniae through the technique called GradSeq, which is the fractionation of RNA-protein complexes by sucrose gradients, followed by protein analysis and RNA sequencing (RNA-Seq) (171). Indeed, the role of small RNAs is far more complex than envisaged in many bacteria (172).

THE ANTIBIOTIC RESISTANCE GENES

Bacterial plasmids constitute one of the most relevant vehicles for the spread of genetic traits, including Ab^r. It is of importance to know why some genes, like Ab^r genes, are located on plasmids and not on chromosomes. Mathematical modeling supports the idea that the prior acquisition of a gene may condition its location and that once it becomes a fixed trait, it is unlikely to lose it (173). It follows that the location of Ab^r genes in plasmids could be ascribed to a higher rate of acquisition of these traits, perhaps due to the higher number of copies (targets) of plasmids than of chromosomes. The model also predicts that the probability of encountering plasmid-located Ab^r genes depends upon the frequency of interspecific plasmid transfer events and the number of species among which the gene and plasmid are shared (173). These findings hold especially true for small RCR plasmids because of their ubiquity, promiscuity, high copy numbers, and recombination abilities: intermediates of RCR are ssDNA molecules that are highly recombinogenic.

Usually, RCR plasmids bear just one Ab^r gene, except the smallest replicons, like those isolated from Mycoplasma, which are devoid of any resistance cassette (77, 174). The promiscuity and diverse strategies to spread Ab^r exhibited by the RCR plasmids make them versatile vehicles for the dissemination of resistance among bacteria. The mRNAs of some of these Ab^r genes fold into secondary structures that would lead to translation inhibition by occlusion of the SD sequence (RBS2) to the ribosomes. However, translation of a leader peptide starting from an upstream SD sequence (RBS1) would result in refolding of the mRNA and exposure of the translation initiation signals to the ribosomes (Fig. 6A). In all these inducible genes, the mRNAs fold in such a manner that RBS2 becomes occluded, which is an efficient manner to inhibit translation (175). The two best-characterized Ab^r genes of RCR plasmids are those encoding resistance to chloramphenicol (Cm^r) and tetracycline (Tet^r) and, to a lesser extent, resistance to erythromycin (Erm^r).

FIG 6.

Leader peptides in plasmid-encoded tet and cat genes. (A) Schematic representation of a DNA fragment that contains the promoter, the leader peptide-encoding region, and the RBS sequences. (B) Nucleotide sequences of the leader peptide regions of cat genes from plasmid pUB112 (GenBank accession no. X02872.1), plasmid pC221 (GenBank accession no. X02166.1), the prototype cat-86 from the chromosome of B. pumilus (GenBank accession no. K00544.1), and plasmid pC194 (GenBank accession no. K01998.1). (C and D) Nucleotide sequence of the tetL (C) and tetK (D) genes from pMV158 and pT181 (NCBI accession no. NC_010096.1 and GenBank accession no. CP001783.1), respectively. The promoters and translation initiation signals are depicted. Putative promoter (boldface), initiation (green) and termination (red) codons, and proposed RBS sequences (blue, underlined) are indicated.

Chloramphenicol Resistance

Resistance to Cm is frequently due to the acetylation of the antibiotic by the enzyme chloramphenicol acetyltransferase (CAT), resulting in an inactive form of the drug (176). In G+ bacteria, the expression of the cat genes is usually inducible by Cm (177). The mechanism of induction by Cm has been studied for several cat genes of staphylococcal RCR plasmids, namely, pC194 (178–180), pC221 (181), and pUB112 (182). In addition, the chromosomally encoded cat-86 gene harbored by Bacillus pumilus (183) has been studied in detail (184). In all these cases, the translation of the Ab^r gene is hindered by the generation of mRNA secondary structures that occlude RBS2 (Fig. 6A and B). In the presence of Cm, ribosomes that have bound the drug would translate a short leader peptide (of ∼20 residues) that starts at RBS1 and that precedes the secondary structure in the mRNA. Translation of the leader peptide would result in the stalling of the ribosomes at a specific site in the leader mRNA. These leader peptides seem to be rRNA-binding molecules that target the peptidyl transferase center of the 23S rRNA. The nascent peptide would alter these translating ribosomes, which would become stalled (184). Ribosome stalling would lead to either mRNA refolding or resolution of the stalled transcript by RNase J1 (185). In any of these cases, ribosome stalling would lead to RBS2 becoming fully accessible to unmodified ribosomes, which, in turn, translate the cat gene. Inducible features, in conjunction with the poor expression of the pC194 cat gene in S. pneumoniae, have led to (i) the selection of constitutive versions of the cat gene, (ii) the selection of deletions that couple two promoters to increase the synthesis of the cat gene, and (iii) the isolation of gene mutations that increase Cm^r. All these mutations led to high expression of the cat gene in the pneumococcal host (178, 179).

Tetracycline Resistance Determinant

The TetL protein of pMV158 belongs to the major facilitator superfamily (MFS) of efflux proteins found in G+ bacteria (186). They are energy-dependent, membrane-associated proteins that confer Tet^r by pumping the antibiotic outside of the bacterial cell, reducing the intracellular concentration of the drug. The MFS proteins are highly hydrophobic and show 14 membrane-spanning regions separated by short central hydrophilic stretches (186). The efflux proteins exchange a proton for a Tet-cation complex against a concentration gradient. The majority of the MFS proteins are located in the lipid bilayer with the hydrophilic amino acid loops protruding into the periplasmic (G− bacteria) or the cytoplasmic (G+ bacteria) space (187), and they have been proposed as suitable targets for the design of novel antibacterials (188). In the case of TetL_pMV158, the protein exhibits anomalous electrophoretic mobility on SDS-polyacrylamide gels, indicative of membrane association (104). A 4-fold increase in the copy number of plasmids with the pMV158 replicon (from ∼25 to ∼100) resulted in a moderate reduction in Tet^r in S. pneumoniae because selection for plasmid-containing cells dropped from 1.0 to 0.5 μg/mL. This reduction was concomitant with a diminution in the cell growth rate (doubling time from ∼25 to ∼35 min in the presence of selective pressure) and size of the colonies in plates, demonstrating a deleterious effect of increasing the number of protein molecules in the cell membrane (189).

The tetL gene from pMV158 is highly homologous to other tet genes carried by staphylococcal plasmids like pT181 (annotated as tetK; GenBank accession no. CP001783.1) (190). Homologies between tetL and tetK extend along the entire sequence of both genes, being 69% and 61% for the nucleotide and amino acid sequences, respectively (104). However, whereas the expression of the tetK gene is inducible by Tet, the expression of tetL is constitutive (104). Although the mechanisms involved in the inducibility of the tetK gene have not (to our knowledge) been addressed in detail, that may be due to the translation of a short leader peptide (Fig. 6D) that would lead to the accessibility of the ribosomes to the translation initiation signals of the gene, as has been shown for the cat (Cm^r) genes. The tetL gene would not encode a leader peptide (Fig. 6C). The main differences between tetL and tetK were found at the 5′ region of the genes. In the tetL gene, there is a promoter sequence with a near-consensus −35 region (5′-TTGAaA-3′ [a lowercase letter indicates a nonconsensus nucleotide]) and an extended −10 element (5′-TGTGGTAaAAT-3′, coordinates 1583 to 1610) separated by 16 nt. This extended −10 element is more conserved in G+ bacteria (the so-called 5′-TRTG-3′ motif) than in E. coli, in which the −10 extension is 5′-TG-3′ (120, 191). Downstream of the tetL promoter, there are RBS1 and RBS2 sequences, separated by about 80 nt, albeit a TAA stop codon would prevent the synthesis of a leader peptide (Fig. 6C). The tetL gene has an inefficient Rho-independent transcriptional terminator (129, 130, 192).

We have performed a BLASTn search of the NCBI database, providing the nucleotide sequence of the tetK gene of pT181 (coordinates 2701 to 4321) as the query. The search retrieved 123 hits with identities higher than 95%, all of them belonging to S. aureus. When the same search was done with the tetL gene of pMV158 (coordinates 1580 to 3112), 110 hits with an identity higher than 95% were retrieved; however, in this case, the distribution of the gene among bacterial species was much broader and included, among others, staphylococci (34 hits), enterococci (26 hits), Bacillus (10 hits), and streptococci (4 hits).

Erythromycin Resistance

In the case of the RCR plasmid pE194 (193), the ermC gene provides Erm^r by methylation of the bacterial ribosomes. This gene is also inducible by low doses of the antibiotic, but in this case, folding of the mRNA seems to be more complex than that of the two previous ones: large secondary structures could be predicted within an ∼140-bp region that could encode a 19-residue leader peptide (194). Induction of Erm^r would be the result of different configurations of the translating mRNA, as in the cases of Tet^r and Cm^r discussed above. However, a more recent report points to the existence of two leader peptides rather than only one, at least in the case of the ermC gene, suggesting a complex regulation in the expression of these genes (195).

Distribution of Ab^r in the Family

We also used the AcCNET approach to inspect the distribution of Ab^r traits among the MOB_V family of plasmids. Fifty-seven clusters of antimicrobial resistance proteins were identified in 104 MOB_V plasmid genomes, 89 present in Firmicutes, 13 in Proteobacteria, one in Bacteroidetes, and one in an undefined host. Most of these plasmids were gathered in a single connected group in the network (Fig. 7). The number of Ab^r genes carried by these plasmids ranged from 1 (64 plasmids) to 11 (a single plasmid). The most abundant Ab^r protein clusters were those that provided resistance to lincosamides, tetracyclines, aminoglycosides, glycopeptides, β-lactams, and macrolides. The highest degree of connection of an Ab^r protein node was 27, corresponding to a Tet^r protein cluster, showing that 27 plasmids harbor members of that protein cluster (Fig. 7). All together, Tet^r protein clusters (11 in total) contained members encoded by 70 MOB_V plasmids. Curiously, the Ab^r genes do not seem to be preferentially transferred between plasmids at short evolutionary distances. The most abundant recently transferred Ab^r determinants between MOB_V plasmids, estimated as those included in closely related families (proteins clustered at 95% identity and 90% coverage), were aminoglycoside resistance determinant ANT(4′)-Ib (shared by eight plasmids) and a Tet^r determinant of the TetM type (connecting five plasmids).

FIG 7.

Distribution of Ab^r traits in the MOB_V plasmid data set. AcCNET bipartite network with >80% protein identity and >80% alignment coverage. Clusters of antimicrobial resistance proteins (red in the left panel) were selected by blastp search against the Comprehensive Antibiotic Resistance Database (CARD) (309) by considering hits with >90% identity and >80% query coverage. The group containing most Ab^r clusters is zoomed at the right, where they are colored according to drug class.

THE LAGGING-STRAND ORIGINS

Lagging-strand origins of RCR plasmids and ssDNA phages are highly structured regions where, on ssDNA replicative intermediaries, intrastrand DNA pairings take place (86, 87). This generates hairpins (most of them with internal bulge-loops) that play an important role in the initiation of lagging-strand DNA synthesis (147, 162, 196). Among RCR plasmids, several types of sso have been described, namely, ssoA, ssoU, ssoT, ssoW, and ssoL, although the two first are the best-characterized ones. Previous work (86, 161), followed by a collaborative effort between the laboratories of S. A. Khan and M. Espinosa, permitted the dissection of these two origins from plasmid pMV158. The ssoA region was first reported in plasmid pLS1 (ΔssoU ΔmobM) (Fig. 1A) as a specific orientation-dependent sequence that participated as a signal involved in the conversion of ssDNA to dsDNA forms (86). This region was highly structured and could generate a single hairpin by intrastrand pairing on ssDNA spanning coordinates ∼5220 to 5350 (Fig. 8A). The ssoA region was also present in other RCR plasmids, and although they did not show a high degree of homology, two conserved sequences were identified: a site proposed to participate in plasmid cointegration termed recombination site B (RS_B) and a 6-nt consensus sequence (5′-TAGCGT-3′; CS-6) located in the terminal loop of the hairpin. A so-called ssDNA promoter (196) was later located in the vicinity of RS_B (147). This ssDNA promoter was shown to be a binding site for RNAP to transcribe an ∼17-nt-long pRNA that terminated at the CS-6 (147). The pRNA was extended some 80 to 100 nt by the host DNA Pol I that was later replaced by DNA Pol III to complete lagging-strand synthesis (147). The transition from pRNA to lagging-strand replication was investigated later, and a transcription elongation protein complex linking the transcription and replication systems was found (197).

FIG 8.

Scheme of sso in pMV158. (A) The two lagging-strand origins mapped and identified in pMV158 (ssoA and ssoU) and the proposed alternative sso (alt_ssoA) located within the tetL gene. (B) Plasmids harboring deletions in these regions are affected in their transformation abilities. S. pneumoniae (Spn) and B. subtilis (Bsu) were made naturally competent; E. coli (Eco) and E. faecalis (Efa) were electrotransformed. In all cases, plasmid DNAs were prepared by CsCl gradient purification. The number of transformants per microgram of plasmid DNA is indicated. NT, not tested.

The ssoU region was shown to be a very efficient signal for the conversion of ssDNA to dsDNA molecules in a variety of G+ bacteria but not in E. coli (86). Moreover, this region plays an essential role in plasmid promiscuity (198). The ssoU origin spans ∼210 nt (coordinates 3210 to 3420) and is located upstream of the oriT region and the mobM gene (Fig. 1A and 8A). Mapping of unpaired sequences showed that unlike ssoA, the ssoU region has a complex structure where at least three possible hairpins (I, II, and III) could be detected. At the base of hairpin II, an RS_B sequence was located where the host RNAP binds to synthesize an ∼45-nt-long pRNA to initiate lagging-strand replication (147). There is substantial homology between the different types of lagging-strand origins (ssoA, ssoU, and ssoT), especially around the RS_B sequence, a nonsurprising fact, as this sequence is the primary RNAP binding site to synthesize the initiating pRNA (198, 199). Between ssoU and oriT, there is another ∼150-bp “silent” region in which there is a 109-residue-long open reading frame (ORF109) of unknown function (see “Family Secrets: the Unknown ORF109”).

More light is thrown by the collection of data accumulated over the years, which has now allowed us to compile the roles of the ssoA and ssoU regions (Fig. 8B and Table 4). It was previously found that conjugative mobilization between S. pneumoniae and E. faecalis (mediated by pIP501 as the helper plasmid) was the most efficient transfer method when the plasmid harbored an intact ssoU (198). When pneumococcal intraspecies transfers were tested (Table 4), no significant reduction in the efficiency of transfer (ν value) was observed for plasmids impaired in the functionality of either ssoA or ssoU, indicative that both origins are functionally replaceable. However, an 80% reduction was observed for plasmids lacking both origins. In interspecies transfer experiments from S. pneumoniae to E. faecalis (Table 4), the ν values obtained for pMV158wt were nearly identical to those observed when the plasmid carried a nonfunctional ssoA. However, there was an ∼97% reduction in the transfer efficiency when the plasmid had a defective ssoU, with a further decrease (∼99%) when the plasmid tested lacked both origins. It was predicted (198) that plasmids defective in both origins should be unable to replicate, and this was the case in E. coli but not in the G+ hosts tested (Fig. 8B and Table 4). It was hypothesized that some other sequences could act as sso, although with less efficiency, and they would be functional only in the G+ hosts. A search for unpaired CS-6 sequences within a secondary structure revealed the presence of one possible stem-loop structure that was placed within the coding region of the tetL gene (Fig. 1A) and that could act as an alternative to ssoA, thus named alt_ssoA (86) (Fig. 8A). Replacement of the tetL gene by a cat gene, in which no such alternative ssoA could be detected, failed to rescue plasmids defective in both origins (ssoA and ssoU) in any of the hosts tested. We took these results as indicative that pMV158 has not only two well-defined and functional sso regions but an alternative one (alt_ssoA) as well (Fig. 8A). This one could be used as a backup signal for lagging-strand synthesis in the G+ hosts. In the case of B. subtilis (natural competence), the efficiency of transformation was always lower than in S. pneumoniae (Fig. 8B), which can be ascribed to the failure of B. subtilis to be transformed with monomeric supercoiled DNA plasmid forms (200–202).

TABLE 4.

Efficiency of transfer (ν), copy number (N_av), and replicative indicators (ssDNA/dsDNA) of pMV158 and derivatives in different hosts

Host/plasmid	S. pneumoniae a			E. faecalis a			B. subtilis b			E. coli c
	νd	N av	ssDNA/dsDNA ratioe	ν	N av	ssDNA/dsDNA ratio	ν	N av	ssDNA/dsDNA ratio	ν	N av	ssDNA/dsDNA ratio
wt	100	34 ± 5	1	100	17 ± 4	1	100	12 ± 4	1	100	8 ± 1	1
ΔssoA	70	22 ± 4	1	100	10 ± 3	1	30	8 ± 3	1	80	8 ± 2	1
ΔssoU tetL	70	22 ± 2	1	3	5 ± 2	20	30	5 ± 2	10	80	5 ± 2	1
ΔssoU cat	70	22 ± 2	1	3	5 ± 2	20	30	5 ± 2	10	80	5 ± 1	10
ΔssoA ΔssoU tetL	20	5 ± 1	100	1	3 ± 1	50	15	3 ± 1	100	0	—f	—
ΔssoA ΔssoU cat	0	—	—	NTg	NT	NT	0	—	—	0	—	—

^aThe method used for transfer was mobilization.

^bThe method used for transfer was natural competence.

^cThe method used for transfer was electrotransformation.

^dThe efficiencies of transfer (ν) were normalized to 100% for the values obtained for pMV158wt in the various hosts.

^eThe ssDNA/dsDNA ratio was normalized to 1 for the pMV158wt in the various hosts.

^f—, not measurable.

^gNT, not tested.

In addition to the above, it was apparent that the N_av values depended strongly upon the host and the presence of intact sso regions. Plasmid pMV158wt exhibited the highest N_av in all hosts tested, and these values were reduced as the sso regions were deleted (Table 4). In all the hosts tested, except in S. pneumoniae, it was clear that the adaptive value of ssoU was greater than that of ssoA on plasmid replication, as was evident from the N_av values and the ratios of ssDNA/dsDNA (Table 4). Furthermore, when the fitness of the plasmids was assessed, it was found that the presence of pMV158wt in S. pneumoniae did not affect cell growth (doubling time) nor their colony size or appearance, whereas cells harboring pLS1 (ΔssoU) exhibited an ∼10% reduced growth rate despite the reduced N_av (from 34 to 22) and a reduced number of the TetL membrane-bound molecules (86, 164).

We can conclude that the presence of two fully functional sso origins and one backup sso origin in pMV158 (Fig. 8A) provides the plasmid with a panoply of possibilities to colonize new hosts. The ssoA region is host dependent and would require more specific host-encoded proteins, whereas alt_ssoA would be a backup alternative. Concerning ssoU, its participation in broad-host-range replication was predicted (86) and later demonstrated (198). The sequences and organization of the sso origins would be important for interaction with different RNAPs and, likely, with other host proteins (DNA polymerases, SSBs). We performed a BLASTn search for homologies, using a 269-nt-long query encompassing ssoU from pMV158 (coordinates 3246 to 3514); it retrieved 278 hits in all bacteria, 258 (92.8%) of them in Firmicutes. All of these hits had an E value of 5e−138 and exhibited 100% identity lengthwise with the entire DNA fragment. Many of the sequences were found in plasmids from Staphylococcus and Enterococcus, and several were chromosomally encoded, although we did not investigate whether the chromosomally encoded ones were located on mobile regions. We propose two alternative hypotheses, still to be experimentally tested: (i) ssoA might have evolved from ssoU in a process of domestication by the bacterial host or (ii) ssoU (and perhaps also ssoT) might have evolved from ssoA through the acquisition of sequences that allowed the stable maintenance of the plasmid in the new colonized hosts. If this latter assumption were correct, we could envisage a model in which RCR plasmids harboring only an ssoA type of lagging-strand origin would be confined to a given host and close relatives, with little chance to colonize evolutionary unrelated hosts. Acquisition of an ssoU region would permit these plasmids to expand their repertoire of hosts, which could be achieved by mobilization, transduction, or transformation, as shown for plasmid pUB110, which can populate not only staphylococci but also bacilli (21). The “acme” of the promiscuity of RCR plasmids would be represented by pMV158, because by having ssoA and ssoU (and the extra alt_ssoA as a backup), it would be able to colonize a wide variety of hosts.

THE INITIATOR AND TERMINATOR PROTEINS

Many RCR plasmids encode two types of initiator/terminator proteins, the Rep proteins involved in replication and the Mob proteins involved in their conjugative mobilization; both types support the two lifeguards of plasmid lifestyle, namely, vertical and horizontal spread. These types of proteins usually belong to the superfamily of HUH endonucleases, which are involved in plasmid, phage, and viral replication and also in conjugation and transposition, although the HUH motif is not present in the Rep_trans family of RCR plasmids (83, 92). Whereas Rep proteins have ∼250 amino acids, Mob proteins are twice as big, making a substantial difference for proteins performing a similar biochemical function. Rep proteins initiate and terminate replication, oligomerize, and interact with other host-encoded proteins, like SSB and PcrA helicase. Mob proteins initiate and terminate mobilization, associate with the host membrane, and interact with other proteins (dimerization, T4CP). In addition, it has been shown that MobM_pMV158 acts as a transcriptional repressor to control its synthesis and that of the RNA II (106, 122). This scenario applies to pMV158 and also to plasmids pUB110, pE194, pC221, and pT181, to name the most studied ones in both replication and mobilization. All these plasmids encode two DNA-cleaving proteins that act on different sites and specifically recognize different DNA sequences to initiate and terminate replication or transfer. Rep and Mob proteins require their DNA substrate to be in an ssDNA configuration (81). The binding of these proteins to their target bends DNA and promotes hairpin formation, leading to exposure of the nick site in an unpaired configuration (54, 105, 140, 141, 146, 203, 204). Differences do exist in the location, configuration, and recognition of their DNA target as well as in the catalytic residues involved in the initiation processes. While most nucleases use water molecules to perform the nucleophilic attack on their DNA target (205), the HUH endonucleases use either a nucleophilic Tyr or His residue.

Rep initiators use a Tyr residue, as ascertained by mutational, biochemical, and structural analyses (206–209). In addition, one or more cofactors are needed to complete the reaction, namely, metal ions and even small auxiliary proteins (210–212). In the case of plasmid pMV158, a Tyr residue (Y99) is the active residue in RepB (213, 214), whereas a His (H22) is the catalytic residue employed by MobM (215). Proteins RepC_pT181, RepD_pC221, and MobA_pC221 require Mg²⁺ (216, 217). However, RepB_pMV158 and MobM_pMV158 require Mn²⁺ for their activity, although this cation could be replaced, at least partially, by Mg²⁺ or Ca²⁺ (210, 211, 218, 219). Whether the Mn²⁺ requirement is a streptococcal feature is not known at present, but streptococci have acquired specific proteins to sense Mn²⁺ availability, which is crucial in their infective processes. These bacteria have specific regulators, like the MtsR metalloregulator, which react to the levels of Mn²⁺ and, apparently, to the iron levels (220).

The dso Regions and Their Cognate Rep Initiators

The dso region is located at different positions relative to its cognate rep gene (Fig. 9): upstream of the rep gene (pMV158 and pFX2) or placed within the 5′ region of the rep gene (pT181 and pC221). This latter organization is probably also found in pC194 and pUB110. Whatever the location, all dso regions can be divided into the bind and nic loci. The former defines the binding site of the Rep initiator and the latter where the cleavage occurs (the nick site). Both loci can be contiguous (Rep_1 and Rep_trans families) or separated by an intervening sequence of up to 100 bp (Rep_2 family) (Fig. 9). Whereas the nic locus is conserved among the members of the same family, the bind locus is the one that confers plasmid specificity to the initiation by the Rep protein (101). Nevertheless, purified RepB_pMV158 and RepC_pT181 proteins cleave supercoiled DNA of related plasmids, provided that a minimal degree of supercoiling is achieved; this suggests that in vitro, but not in vivo, binding of the Rep protein to the bind locus is not required (101, 221, 222). As stated above, RepB_pMV158 binds with high affinity to the three DDR sequences but it can also bind, albeit with weaker affinity, to the two PDR sequences, which are closer to the nick site (Fig. 9) (146). RepC_pT181 and RepB_pMV158 were shown to induce a bend at dso and, subsequently, to melt the DNA to promote hairpin extrusion and cleavage of the nick site exposed in an ssDNA configuration to initiate replication (101, 146, 203).

FIG 9.

The leading strand and dso region of several RCR plasmids. The dso regions contain the bind and nic loci, which correspond to the sites where the initiators bind and cleave their target DNA, respectively. They can be placed upstream of the rep gene (pMV158) or within the 5′ region of the rep gene (pT181). In some plasmids, the nic and bind loci are perhaps not separated (pC194). The dso regions are depicted in green, whereas promoters and regions coding the initiator proteins are shown in red. In pMV158, the PDRs (proximal direct repeats) and DDRs (downstream direct repeats) are indicated. IR, inverted repeat.

The Rep proteins of the RCR plasmids have cleavage and joining (DNA strand transfer) activities on ssDNA, which are mediated by a Tyr residue that can perform a 5′-phosphotyrosine covalent linkage with the DNA substrate, leaving a free 3′-OH end at the cleaved dinucleotide. The 3′-OH end acts as the replication primer, and it also serves as the nucleophile during the breakage of the phosphotyrosine intermediate in the strand transfer reaction at the replication termination. Biochemical studies have been performed with the purified Rep proteins from plasmids pT181, pC221, pUB110, and pMV158. Whereas the first three initiators were shown to be dimers, RepB_pMV158 purified as a hexamer (144, 146). The dimeric configuration of the former Rep proteins allowed creation of a coliphage ϕX174-based model to initiate replication, in which a specific Tyr residue belonging to one Rep protein subunit cleaves the scissile phosphodiester bond at the nick site, remaining covalently bound to the 5′ end (85). Assembly of the replisome would take place at the nicked DNA molecule by the participation of host-encoded proteins, namely, PcrA helicase, SSB, and DNA Pol I, which would be later replaced by DNA Pol III (89, 206, 217, 223, 224). Termination of replication by RepC_pT181 was cleverly solved in the laboratory of Richard Novick after the observation that a heterodimer composed of RepC-RepC-ssDNA appeared as one product of replication termination. They proposed a model in which, after one round of replication, the replisome extended the leader strand ∼10 nt past the nick site until the hairpin where replication initiated was fully regenerated. Termination would be achieved through nucleophilic attacks performed by Tyr residues and the free 3′-OH DNA end generated during replication initiation, releasing the RepC-DNA heterodimer (85, 225). No such molecules were detected in RepA_pC194 replication, and mutational analyses suggested that a Tyr residue participated in the initiation, whereas a Glu residue would be involved in the termination (226). Covalent intermediates between the replication initiator and its target DNA were also not found in the case of the filamentous bacteriophage f1, a puzzle that was solved by using short incubation times (227). In the case of RepB_pMV158, mutational analyses showed that residue Tyr99 performed the cleavage of the supercoiled plasmid DNA, but no covalent RepB_pMV158-DNA molecules were detected (101). However, determination of the chirality of the phosphate involved in the cleavage and joining (strand transfer) reactions demonstrated the existence of a covalent intermediate (213). More recent work has permitted the detection of covalent RepB_pMV158-DNA adducts by modulation of the activity of the initiator through changes in the temperature and time of incubation as well as by reducing the pH, which increased the half-life of the covalent complex from 1 h (at pH 7.0) to more than 6 h (at pH 4.5) (145, 214).

From the structural point of view, the landscape is discouraging, because high-resolution structures are available only for DNA-free representatives of the Rep_2 (RepB_pMV158) (206) and Rep_trans (Rep_pSTK1 and plasmid pC221 chimeric RepDE and RepDN proteins) (207) families, pointing to the lack of structural understanding of how plasmid replication initiators interact and process their cognate DNA. In the case of RepB_pMV158, limited hints of the protein-DNA structure of the complex were obtained at 24 Å resolution by negative-staining EM (206). However, the DNA in the structure could not be visualized, because the EM structure had lower resolution than those obtained by DNA footprinting experiments (146). More information gave the high-resolution crystal structures of the hexameric full-length RepB_pMV158 without DNA. In these structures, each of the protomers exhibited two different domains separated by a short hinge region: the origin-binding domain (OBD) and the oligomerization domain (OD) (206). The overall structure of the hexamer shows a crevice and a central channel through which a single DNA strand, but not dsDNA, could pass (Fig. 10A and B). The EM data of the complex, when compared to the protein-only data, showed additional electron density that occludes the positively charged crevice and central channel, suggestive of the proper positioning of the DNA at this region (206). The two main domains of RepB_pMV158 have been purified separately, and their properties have been analyzed by analytical ultracentrifugation and by small-angle X-ray scattering (224). From these data, it was apparent that the hinge region played an essential role in the plasticity exhibited by the hexamer, facilitating multiple orientations between OBDs and ODs and thus permitting the orientation of the hexamer to the target DNA (145, 224). The OBDs were detected in nine different orientations relative to the rigid hexamer ring formed by ODs. Therefore, the domain configuration in the hexamer could be related to the role of the protein in the recognition of the direct repeats at the plasmid dso. It was further noticed that the hinge region could exist in all the members of the Rep_2 family of proteins, enhancing the relevance of this region. RepB_pMV158 showed homology with several replication initiators from viral origins, like adenovirus, nanoviruses, and circoviruses, although the orientation in the hexameric rings was different (224). The overall structural information of RepB_pMV158 led to a model in which the hexameric initiator would first bind to its cognate bind locus, inducing a DNA bend that favors its loading to the nic locus and the extrusion of the dso hairpin. Cleavage mediated by the initiator and, perhaps, generation of a small gap would allow the replisome assembly in which the first stages would be performed by DNA Pol I (228). Interestingly, the PcrA helicase is not needed for pMV158 replication (see Specialized Host Functions), suggesting that RepB could assemble as a hexamer around the nicked strand (206). After assembly, RepB could act, perhaps, like the hexameric adeno-associated replication initiator or the papillomavirus E1 multifunctional initiator helicase (229). Hexamerization of the replication initiator seems to be common in small plant and animal viruses, although these proteins have a C-terminal helicase domain that is absent in RepB_pMV158 (224). Be that as it may, the interactions between RepB_pMV158 and the pneumococcal PcrA protein (if they interact) remain to be ascertained, and only the resolution of the three-dimensional structures of both proteins in a single complex might clarify this point.

FIG 10.

Electrostatic surface representation of the RepB_pMV158 hexamer (A and B) and the RepDE chimeric dimer (C). (A) RepB full-length hexamer [FL(1-210)] “C3 structure” (Protein Data Bank [PDB] ID 3DKY, left) and “C2 structure” (PDB ID 3DKX, right) viewed from the N terminus (top view) and the C terminus (bottom view). The potential DNA-binding surface (blue), the flexibility of the N-terminal origin-binding domain (OBD) (top view), and the rigidity of the oligomerization domain (OD) hexameric ring (bottom view) are also indicated. (B) Side view of the RepB “C2 structure.” (C) Structure of the hybrid chimera initiator of replication RepDE (PDB ID 4CWC). Blue, red, and white represent positively charged, negatively charged, and neutral surface, respectively.

The pC221 RepDE and RepDN chimeras consist of the 21-kDa central domain of RepD_pC221 initiator (carrying the active Tyr191) and the 14-kDa C-terminal fragment of the RepE_pS194 or RepN_pSN2 initiator, respectively. The crystal structures indicate that these initiator proteins form horseshoe-shaped dimers harboring a positively charged channel that could accommodate dsDNA; such a channel includes the six amino acids involved in protein specificity (221, 222) as well as the catalytic residues. Amino acids participating in the interaction between the initiator and the PcrA helicase were located to the open end of the horseshoe (Fig. 10C). Recent determination of the crystal structures of viral HUH initiator-ssDNA substrate complexes has permitted the construction of Rep chimeras, in which the sequence specificity for the substrate has been modified by changing four amino acids, facilitating the construction of tools to handle viral infections (230).

The oriT Regions and Their Cognate Mob Relaxases

Several RC replication initiators have been studied in detail, but there is a lack of information on the conjugative relaxases encoded by RCR plasmids, except those encoded by pMV158, pC221, and pC223 (a pC221 relative). The mobilization cassette of pMV158 (MOB_pMV158) includes, most likely, a single mobM gene and its cognate origin of transfer, oriT_pMV158 (Fig. 11A and B), although the existence of a second ORF (see “Family Secrets: the Unknown ORF109”) has been another unrecorded secret in the pMV158 public life. The apparently simple organization of the MOB_pMV158 cassette is also present in pUB110 and pE194, but, to our knowledge, the biochemistry of the mobilization machinery from these latter plasmids has not been analyzed. Nevertheless, it is worth pointing out that oriT_pUB110 and its cognate MobU relaxase have been used to create a new high-frequency recombination-like system dependent on RecA and a helper plasmid to mobilize regions of the B. subtilis chromosome (231).

FIG 11.

Organization of the oriTs of RCR plasmids. (A) The oriT of pMV158 (accession no. MZ396393; coordinates 3564 to 3606) and of pUB110 (accession no. NC_001384) are located upstream of mobM and mobU, respectively. The three inverted repeats, IR1, IR2, and IR3, are indicated by horizontal arrows, and the MobM nick site (5′-G/T-3′) is indicated by a vertical arrow (105). The two promoters of mobM are indicated by red arrows. The N-terminal moiety of MobM contains the DNA binding and relaxase activities (motifs I, II, and III). (B) Potential hairpins generated at the oriT_pMV158. (C) The oriT of plasmids pC221, pC223, and pS194, showing the five IRs and the three genes involved in their mobilization (216). I, II, and III in the Mob proteins indicate the three conserved regions in the HUH proteins (83).

The organization of the mobilization region in plasmids pMV158, pUB110, and pE194 is different from that present in pC221, pC223, and pS194 (216). In the latter plasmids (Fig. 11C), in addition to their oriT, three protein-encoding genes were found, namely, mobC, mobA, and mobB; they belong to the MOB_P family (9). MobC is an auxiliary protein, whereas MobA and MobB (both genes overlap) are needed for relaxosome formation and transfer (212, 232). At the oriT region of these plasmids, up to five inverted repeats (IRs) have been found, whereas three have been reported for oriT_pMV158 (Fig. 11A and B). These IRs seem to have DNA sequences shorter than the well-studied ones of plasmids F, RP4, and R388, although only two IRs are present in the oriT of the latter plasmids (233, 234). The nick introduced by MobA-MobB was located upstream of a potential promoter that could transcribe the three mob genes in a single mRNA subjected to self-regulation by MobC (212). MobC was later shown to belong to the ribbon-helix-helix (RHH) DNA-binding proteins (235). A model was proposed in which the auxiliary MobC would facilitate the MobA-mediated nicking, probably by generating a high-order complex that would permit the relaxation reaction and, perhaps, the transcription of mobC (216). MobC and MobA of plasmid pC221 were purified as His-tagged proteins, and several of their biochemical properties were studied (236). Whereas MobC purified as a dimer, the oligomerization degree of MobA could not be determined due to polydispersion in the samples. Relaxation kinetics experiments showed that the reaction reached equilibrium after 60 min of incubation at 30°C, achieving a maximum of nicking-closing products of ∼55% of the total input DNA. The optimal pH was ∼8.0, and the reaction required Mg²⁺; all these conditions resembled those found earlier for MobM_pMV158 (105), although Mn²⁺ was the preferred cation in the case of MobM_pMV158 (210).

In pMV158, a detailed analysis of the 87-bp region between coordinates 3564 and 3650 showed an impressive compact organization, especially if we consider that this region (with a 69% A+T content compared to 62.3% of the entire plasmid) encompasses (i) oriT, (ii) the nick site, (iii) the Pmob1 and Pmob2 promoters (Table 3), and (iv) the MobM binding site (Fig. 11A) (122). Three partially overlapping inverted repeats (IR1 to IR3) would generate three mutually exclusive hairpins in which the 5′-GpT-3′ dinucleotide cleaved by MobM would be situated at diverse places (Fig. 11B) (141). The position of these IRs is such that IR1 is placed 8 nt upstream of the nick site, which is surrounded by the arms of IR2. IR3 includes IR1 and, when extruded, would generate a hairpin that would place the nick site at the 3′ end of its right arm. MobM was shown to induce a shift in hairpin extrusion, from IR3 to IR1, in which the nick site would be exposed in ssDNA configuration (237). This binding would hinder the recognition of the mobM promoter region by RNAP (122) and would preclude the extrusion of hairpins IR3 and IR1. In this mechanism, expression of the mobM gene and initiation of plasmid transfer would be suppressed through a single MobM-DNA interaction (141). Mutations at the left arm of IR3 led to plasmids with altered secondary structures. Plasmids harboring these mutations showed a decrease in the intracellular amount of relaxed forms, as well as an ∼50-fold reduction in their frequency of conjugative transfer (Table 5). The interplay between the different hairpins and MobM could be the result of the stochastic distribution of protein molecules at their target DNA. This would ensure that once IR3 is extruded, binding of MobM to it would preclude the generation of IR2; conversely, binding of MobM to IR2 would hinder IR3 extrusion. Under these circumstances, DNA cleavage by MobM would occur only when the protein binds to IR3. This cleavage would lead to DNA relaxed molecules, where no hairpins could be generated, and thus these molecules would be productive for conjugative transfer. Such a molecular switch would generate a relay by which the system would be turned off when MobM binds to IR2 and turned on when the protein binds to IR3 (141).

TABLE 5.

Plasmid relaxed forms in total DNA extracts and mobilization frequencies in plasmid pMV158 and derivatives^a

Plasmid	Relaxed forms (%)	ν (transconjugants/recipients)b,c
pMV158	77 ± 4	102.6 ± 3.5
pMV158mutIR3L	57 ± 5	2.4 ± 0.4
pMV158ΔIR3L	65 ± 7	2.1 ± 0.6
pLS1 (ΔoriT)	<8.3	<10−10d

^aThe derivatives of pMV158 carry base changes at the left arm of IR3 (mutIR3L), a small deletion at IR3 (ΔIR3L), or a total deletion of the oriT region (pLS1, used as a control).

^bAverage result of 4 experiments.

^cν, mobilization frequency × 10⁻⁷.

^d10⁻¹⁰ is the detection limit.

The Relaxase Function of MobM

Protein MobM is a 494-residue dimeric protein that exhibits a relatively high content of Glu and Asp residues (∼20%), which provides an isoelectric point of 6.38. Two main domains in MobM, separated by a short α-helical hinge, have been defined (215). The N-terminal domain (roughly the first 200 residues) is the most conserved among the relaxases of RCR plasmids and contains the DNA binding and relaxase activities, exhibiting the characteristic alpha/beta fold present in all HUH relaxases (210). MobM was considered the prototype of the MOB_V1 subfamily of relaxases (9), and all members of this subfamily exhibited three conserved motifs (233): (i) motif I, HXXR, which was later shown to be the catalytic center (215); (ii) motif II, NY(D/E)L, which was proposed to be the catalytic domain but was later shown not to be the case, at least in MobM, and (iii) motif III, HXDE…PHXH, which is the metal coordination site (83, 215). Interestingly, the subfamily MOB_V2, represented by the relaxase of plasmid pBBR1, shows motifs I and III but lacks motif II. A plasmid mutational analysis that tested all possible Tyr-to-Ala changes within the relaxase domain revealed that none of the Tyr mutants suffered from lower apparent plasmid mobilization frequency (238). In the case of MobM_pMV158, two truncated versions of the protein were constructed: MobMN199, harboring the first 199 residues, and MobMN243, which kept the first 243 residues (210, 219). Both proteins have lost their dimerization ability, and whereas MobMN243 was able to cleave supercoiled plasmid DNA and ssDNA oligonucleotides, MobMN199 retained the activity against the first but not the second substrate. These findings indicate that the hinge domain that follows the catalytic relaxase domain is needed for proper positioning of the target DNA in the active center. There was an intrinsic instability of protein MobMN243 that resulted in the breakdown of MobMN243 into two fragments, the breaking point being mapped around residue 199; such instability was attributed to a flexible region between two domains (219, 239). These observations indicated that the MobM region between residues 200 and 243 is important in DNA processivity and led to the questions of whether (i) protein fracture has a role in the transfer of the relaxase-ssDNA complex into the recipient cell and (ii) only the first 200 residues of the relaxase are transferred because this N-terminal moiety of MobM would suffice for the closing reaction taking place in the recipient cell. In any case, it was apparent that the N- and C-terminal domains of MobM are loosely connected. MobM would, then, have a relaxation domain from residues 1 to 200, whereas the hinge domain (residues 200 to 243) would act as a nicking modulator that facilitates a change in the relative orientations of the N- and C-terminal domains during substrate processing. Despite the different sizes and oligomerization states of MobM_pMV158 and RepB_pMV158, both proteins seem to exhibit similar global structures in the sense of the flexible positioning of their N- and C-terminal domains. Whether this is a feature of all members of the pMV158 plasmid family remains unknown.

Determination of the three-dimensional structure of MobMN199 by X-ray crystallography (215) permitted performance of mutational analyses at critical residues within the N-terminal region. Some of the resulting proteins were purified and tested for their ability to relax supercoiled plasmid DNA. In addition, three pMV158 derivatives harboring mutations in the critical residues Tyr44 (Y44F, predicted to be involved in DNA cleavage) and His22 (the catalytic residue as seen in the crystals, H22A and H22Y; see Fig. 12B) were constructed and tested for their mobilization ability. The results showed that Tyr44 was involved in neither DNA relaxation nor in plasmid transfer, whereas either of the mutations in His22 abolished both relaxation and transfer. It was impressive to realize that a single C-to-T transition at coordinate 3791, which changes the codon CAU to UAU (H22Y), abolished not only the in vitro but also the in vivo activities of protein MobM (215). The C-terminal moiety of MobM exhibits a high alpha-helical content (218). Within this region, at least two functions are located: (i) a putative leucine zipper spanning residues 317 to 338, which was proposed to be involved in dimerization and interactions with the T4CP of the helper plasmid, and (ii) a putative coiled-coil (a nearly all α-helical) region located around residues 420 to 460, which was predicted to be involved in the plasmid association with the cell membrane. Fractionation of cell extracts and Western blot hybridizations showed that the membrane fraction contained most of the MobM native protein. Disruption of the α-helical region, followed by fractionation, failed to show the mutated MobM in the membrane fraction, and the plasmid carrying the mutations was refractory to mobilization (218).

FIG 12.

Crystal structure of MobM catalytic domain bound to oriT DNA, the Nic0-SPO structure (PDB ID 4LVL). (A) Overall structure. The active site contains the histidine triad and E129 (blue) that coordinate an Mn²⁺ ion and the nucleophilic H22 (orange). (B) Zoom into the active site, with bases and residues shown as stick representations. Ray-like arrow points to the phosphate cleaved by residue H22.

Structures of the MobM Relaxase

In the last 2 decades, crystal structures of the relaxase DNA-binding domains bound to their cognate DNAs have been determined for relaxases of the MOB_F, MOB_Q, and MOB_V families, providing crucial insights into understanding the mechanism of DNA binding and processing by these enzymes (215, 240–243). The MOB_F TrwC_R388 relaxase-DNA complex structure was the first one to be solved and revealed an alpha/beta HUH endonuclease fold consisting of five central antiparallel β-strands flanked by α-helices in which the central β-sheet hosts the metal ion coordinating the His triad and the most N-terminal helix contains the catalytic Tyr residue (240). The structure of TrwC explained how HUH relaxases bind the stem of the oriT DNA hairpin and the downstream single-stranded region that forms a crucial U-turn just before the active site. Concerning relaxase-DNA structures of other families, the structures of MOB_Q NES_pLW1043 and MOB_V MobM_pMV158 (215, 243), showed that despite very low sequence homology, both proteins exhibited an overall fold similar to that of TrwC_R388. In addition, the structures revealed that both NES and MobM have a more compact catalytic domain than TrwC and are, respectively, ∼80 and ∼100 residues shorter than TrwC. However, the most striking difference highlighted by the MobM structure was that in contrast to other HUH relaxases, MobM uses a His (and not a Tyr) residue as the catalytic nucleophile. Notably, the use of a His nucleophile greatly facilitates the catalytic reaction initiation but, on the other hand, provides lower stability of the relaxase-DNA covalent adduct (215, 243). The overall structure of the relaxase domain of MobM can be compared to a left hand (Fig. 12A), where the palm (the central β-sheet with the flanking helices) serves as a platform for ssDNA docking and metal ion coordination by a His triad and, unique to MOB_V relaxases, a Glu residue (Glu129 in MobM). At the top of the protein, the index and the middle finger (short upper elements made of a helix, a β-sheet, and a loop) grab the stem of the DNA hairpin, and heading down, the thumb (an elongated loop and a short helix at the C terminus) wraps the ssDNA and directs it into the active site. Finally, the ring and the little fingers (His22-bearing helix and the following loop), together with the bottom of the palm (Glu129-bearling loop), lock the DNA in the active site (Fig. 12B). The NES structure and two alternative MobM structures termed “mono3” structure and “ortho” structure, in which either the thumb is missing (NES) or the little finger is moved away from the active site proximity (NES and MobM), reflect the importance of the thumb and the little finger in DNA delivering and locking in the active site. In these three structures, the DNA does not form the crucial U-turn, and the nucleotide just before the nick site is placed farther away from the active site (Fig. 13) (215, 243, 244).

FIG 13.

Electrostatic surface representation of the MobM (A, C, and D) and NES (B) relaxase structures. The DNA backbone is presented as a yellow tube, and the scissile phosphate (if present) as sticks. MobM (residues 2 to 199) bound to nic0-spo DNA (PDB 4LVL) (A), dna26 DNA (“mono3” structure) (215) (C), and dna26 DNA (“ortho” structure) (D) with the Mn²⁺ ion from the Nic0-SPO structure being shown as a cyan sphere (PDB ID 5N2Q). (B) NES (residues 2 to 195) bound to OriTHP30 DNA with the Ni²⁺ ion being shown as a cyan sphere (PDB ID 4HT4). Blue, red, and white represent positively charged, negatively charged, and neutral surfaces, respectively.

THE REPRESSORS

Plasmid replication (vertical) and mobilization (horizontal) are gene transfer processes subjected to strict control in which plasmid- and chromosome-encoded factors (proteins and/or small RNAs) would participate. In the Rep_2 family of RCR plasmids, control of replication is exerted by the combination of an antisense RNA and a Cop protein (116, 134, 245, 246) with the unexpected participation, at least in plasmid pMV158, of the MobM relaxase (106). In the two other RCR plasmid families, only one or two antisense RNAs have been reported to participate (115, 116, 247).

The Antisense RNA II

The synthesis of the copG-repB mRNA is under the control of the repressor CopG, which binds to the P_cr promoter (transcriptional repression) (Fig. 1A). Moreover, the synthesis of the RepB initiator is controlled at the translational level by the interaction of the antisense RNA II with the copG-repB mRNA and the occlusion of the RBS (121). This organization is present in all the members of the Rep_2 family of RCR plasmids analyzed so far (89). A similar replication control circuit based on a Cop protein and antisense RNAs was reported for the Inc18 family of θ-replicating plasmids (248), which are those that mobilize Rep_2 plasmids more frequently (249). In pMV158 (Fig. 14A), there is an overlapping of the sequence comprising the last codons of copG and the terminator sequence of RNA II (complementary strand). Folding predictions and mapping of in vitro-synthesized RNA II (247) indicated a simple structure with a single hairpin at the end of the molecule. It is composed of an 8-bp stem having a 1-nt bulge and closed by a 6-nt loop followed by a 5′ unpaired tail; the termination sequence contains a stretch of U residues typical of prokaryotic Rho-independent terminators (Fig. 14). The copG-repB mRNA is predicted to have a more complex structure, with at least three hairpins near the 5′ end of the repB coding region (Fig. 14B). The signals to initiate the translation of repB were proposed to include a putative atypical ribosome binding site, ARBS (104, 135). Site-directed mutagenesis was performed at the −10 region of rnaII, which nearly overlaps (but by 1 nt) the proximal ARBS region. Specifically, the sequence 5′-TTAT-3′ was substituted for 5′-CCGG-3′ (Fig. 14A). Unexpectedly, the derivative plasmid showed the same N_av as the wild type; this was due to a coincidental reduction in the levels of the repB mRNA and in the amounts of RNA II (135). Further deletion analyses showed that (i) translation of repB and copG are uncoupled, the former carrying its initiation signals, and (ii) alterations of the ARBS sequence reduced translation of repB, pointing to the relevance of this sequence in the translation efficiency of this gene (250). An in silico analysis of the region showed its conservation among plasmids of the Rep_2 family, leading us to propose that the 5′-UGGG-3′ sequence located upstream of the repB coding region could act as a weak SD (wSD) (Fig. 14A), a hypothesis that needs to be explored in more detail (116).

FIG 14.

Antisense RNA II of pMV158. (A) The DNA sequence between coordinates 577 and 896, encompassing the copG and rnaII genes, is shown. High-resolution footprints on DNA by CopG repressor are indicated in blue letters. The 13-bp symmetric element (SE; within the target of CopG) is boxed in green. The initiation of transcription (+1, arrow), the copG initiation and termination codons, and the repB initiation codon are shown in boldface letters. The −35 and −10 regions of promoters P_cr and P_ctII are shown in red letters. The putative ribosome-binding site (SD) for copG and the putative weak SD sequence (wSD) for repB are underlined. The palindromic sequences (arrows) corresponding to the CopG binding site, the MobM binding site, and the transcription terminator of rnaII are indicated. (B) Schematic representation of the secondary structures generated in the copG-repB mRNA (green) and in the antisense RNA II (red). After duplex formation (right), the mRNA would refold and binding of the translating ribosomes would be hindered. Modified from reference 116.

Concerning the mechanism of translation inhibition, pairing between RNA II and copG-repB mRNA involves part of the 3′ moiety of the 5′ tail of RNA II, probably playing an important role in the first steps of the RNA-RNA interactions. Unexpectedly, the loop of the single RNA II hairpin played only a secondary role in the RNA interactions (134). Rather than the usual hairpin-kissing interactions through unpaired loop regions (251), the initial duplex started by pairing the 5′ end of RNA II with its cognate sequence in the mRNA. A computational model of three-dimensional interactions (schematized in Fig. 14B) supported these results and showed structural rearrangements in the region encompassing the ARBS/wSD sequence, perhaps accounting for the inability of the ribosomes to bind to this region and initiate RepB translation (116).

Transcriptional Repressor Protein CopG

Cop repressors encoded by the Rep_2 family of RCR plasmids are small, dimeric DNA-binding proteins exhibiting the RHH motif (252). Cop proteins bind to a specific region that encompasses the single cop-rep promoter. In the case of pMV158, repressor CopG (45 residues, formerly RepA) was purified and shown to be a spherical homodimer; its biochemical and biophysical properties were determined (253). High-resolution footprinting assays demonstrated that the target of the protein extends 48 nt, to which four dimers bind, generating an overall bend of ∼120° along this region (254, 255) (Fig. 15A). CopG was chemically synthesized and shown to have biochemical properties indistinguishable from its biological counterpart (256). Moreover, a chemical CopG was employed to show that protein folding and dimerization were concomitant processes (257). Cop, the CopG homolog encoded by plasmid pE194, was purified and deduced to be a trimer because of its behavior under exclusion chromatography (258), but to our knowledge, no further characterization of the protein has been performed.

FIG 15.

Structure of CopG bound to its target DNA. (A) Three-dimensional structure of the solved protein-DNA complex, in which two dimers of the protein are bound to the 13-bp symmetric element through two antiparallel β-sheets (arrows). (B) Proposed model of four dimers of CopG bound to its entire DNA target (48 bp) covering successive helices and introducing a global bend of ∼120° (155). The structures were generated by PyMOL (310) from PDB ID 1B01 (A) and 1EA4 (B).

The three-dimensional structures of CopG alone and in complex with its DNA target were solved (155) (Fig. 15A). The protein has an RHH conformation resembling the Arc repressor of phage P22 (252, 259). In complex with DNA, the protein structure was that of a tetramer composed of two dimers, related by a crystallographic dyad; this dimer of dimers is considered the functional unit of the protein. CopG interacted with DNA bases via the N‐terminal β‐ribbon, and with backbone phosphate groups, inducing an ∼60° DNA bend due to compression of both minor and major grooves confronting the protein (155). Raman spectra analysis performed with the Arc repressor, alone and in complex with its target DNA, showed that the protein was bound to the major groove, establishing direct contacts and causing some changes in the deoxyribose rings, consistent with Arc-mediated bending of its target DNA (260). Similar studies in solution were done with CopG, and preliminary results confirmed those obtained by crystallography (L. Dostál, M. Espinosa, G. del Solar, and H. Welfle, unpublished).

Nowadays, CopG is one of the representatives of the RHH family that includes 5,864 proteins from 3,006 species and 82 structures (Pfam accession no. PF01402 [https://pfam.xfam.org/family/RHH_1]). The 48-bp-long DNA region to which CopG binds encompasses the P_cr promoter regulated by the protein, and a model in which four CopG dimers bind to four successive DNA helix turns that contains the operator region was proposed (155) (Fig. 15B). Mutational analyses of the operator region later confirmed the model and established the preferred order of DNA binding (156). The mechanism of repression by CopG was also revealed by competition experiments between purified CopG, RNAP, and operator DNAs and showed that cooperative binding of CopG to its target hindered RNAP binding. Furthermore, CopG was able to displace the promoter-bound RNAP by binding to the opposite helix of the DNA target and introducing a strong DNA bend that expelled the already bound RNAP from its promoter (156). A thought-provoking experiment was attempted on the basis that the coding sequences of genes copG and repB are continuous except for a single UAA stop codon (coordinate 790), which is the stop codon of copG (Fig. 14A). Attempts to create a single CopG-RepB chimeric protein by changing the UAA stop codon to AAA (Lys) failed, leaving us with no explanation for it (M. Moscoso and M. Espinosa, unpublished).

Conversion of CopG into a transcriptional activator was accomplished by the replacement of the catabolite activator protein (CAP) DNA-binding site for the CopG-binding site (261). In this case, the E. coli fur operon, involved in iron transport and metabolism, was chosen because its transcription is under the control of two CAP-regulated tandem promoters located on opposite sides of the DNA helix. The CAP-binding sites were substituted for the CopG-binding site, and copG (inserted into the E. coli chromosome) was fused to an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible promoter. In cultures induced with IPTG, a 5-fold increase in the expression of fur was observed, revealing that properly positioned CopG targets and CopG-induced bends were able to activate the interactions of RNAP with a promoter. Based on their small size and DNA-binding properties, the question of whether some small prokaryotic transcriptional repressors could be converted into activators by accelerated evolution experiments was recently addressed (262). The answer to this was “yes”: employment of the lambda Cro repressor (63 residues) allowed demonstration that this small protein could also perform a binary role (repression or activation). Thus, engineering of transcription proteins and their DNA binding targets seems a promising avenue in synthetic biology because of the feasibility of converting the DNA sites into loci to sequester the transcription proteins (“decoy binding sites”) as a mechanism to regulate the transcription rate of any desired gene (263).

The Unexpected: Cross Talk of MobM Relaxase with RepB Initiator

Cross talks between different plasmid-encoded regulatory proteins were shown early by the labs of Chris M. Thomas and David Figurski by using the promiscuous plasmid RK2 as a model system. Among key proteins participating in regulatory circuits, KorA was demonstrated to regulate a transcriptional switch between operons involved in replication and conjugation (264–266). However, in the case of small RCR plasmids, no indications of such interactions between modules were reported, until we got a clue by comparing the N_av values of pMV158 and several derivatives in different hosts (Table 4). A thorough search for a possible antisense RNA around the region encompassing the oriT and ssoU origins (Fig. 1A) did not show any indication of such a regulatory element (A.M. Hernández-Arriaga, G. del Solar, and M. Espinosa, unpublished). The first observation that protein MobM could bind to a site different than oriT derived from EM analyses of MobM bound to pMV158 supercoiled DNA, followed by linearization of the plasmid molecules and measurement of the contour lengths of the DNA regions between MobM positions and DNA ends. The distribution of the MobM positions on the pMV158 DNA revealed the existence of two MobM-binding sites, the main site (site I) coinciding with the position of oriT and a secondary binding site (site II) located just adjacent to the −10 element of the P_ctII promoter of the antisense RNA II (106). DNase I footprinting experiments on supercoiled plasmid DNA confirmed that MobM recognizes a site (site II) between the positions +2 and −30 of the P_ctII promoter. Such a site includes the sequence 5′-TCTTTA-GCC-ATAAAGTATAATATA-3′ (coordinates 863 to 840; −10 box underlined, identical nucleotides highlighted in bold), which has homology with the minimal oriT sequence (5′-ACTTTATGAATATAAAGTATAGTGTG-3′; coordinates 3570 to 3595) (210).

Using a pMV158 derivative that lacked oriT and carried the mobM gene under the control of an inducible pneumococcal promoter, we demonstrated that MobM represses transcription from the P_ctII promoter in pneumococcal cells (106). Also, using transcriptional fusions based on the lacZ reporter gene, we showed that MobM decreases the activity of the P_ctII promoter in E. coli (106). Further analyses of the effect of MobM on the copy number of the indicated pMV158 derivatives supported that by binding to the P_ctII promoter, MobM reduces the intracellular levels of the antisense RNA II and, as a result, increases the plasmid copy number. We explained these findings in light of a regulatory switch in which the main role of MobM is to act as a conjugative relaxase and RepB functions as a replication initiator, both proteins needing a supercoiled DNA substrate. Replicating plasmid molecules could not participate in the conjugative transfer, whereas MobM-relaxed plasmid forms would be unable to replicate, posing a threat to plasmid vertical inheritance: plasmid mobilization would reduce the number of supercoiled plasmid molecules available for the replication machinery in the donor cells. In this scenario, a MobM-mediated decrease in the amount of the replication inhibitor RNA II would provide an increase in the N_av of the plasmid and, as a consequence, in the number of plasmid molecules usable for the replication machinery (106).

Family Secrets: the Unknown ORF109

One of the most puzzling observations, still unsolved, relates to the presence of ORFs upstream of the mobM gene of pMV158. Checking the possible ORFs, we found one ORF that would encode a protein of 109 amino acids. This ORF has two internal ATG codons, which could be translation initiation sites of proteins of 72 and 48 residues, respectively. This sequence extends from coordinates 3460 to 3790. To check whether any of these possible ORFs were transcribed/translated, two experimental approaches were followed. First, we performed in vitro-coupled transcription/translation assays with PCR products placed under the control of the Φ10 promoter of phage T7 and employing T7 RNAP in S30 extracts of E. coli. No translation products were detected. A second approach included testing in vivo protein synthesis. Specifically, the region under study was cloned into vector pET24b, which places the putative genes under the control of promoter Φ10 and provides a C-terminal His₆ tag. Analysis of whole-cell extracts by Western blotting using anti-His antibodies did not provide positive results. Thus, we concluded that either these ORFs are part of a pseudogene or they are the result of recombination processes and loss of function. BLASTp searches were done versus the entire NCBI database. In this case, we retrieved 6 top hits that were highly homologous with ORF109: three of them were proteins of unknown function encoded by S. mutans (accession no. QFG42102.1), S. thermophilus (CAD0146092.1), and S. aureus (T44131); the other three were reported to be putative recombinase/mobilization proteins of staphylococcal origins (ESR29106.1, ASR83112.1, and CAE18143.1), the homologies extending to the first 50 N-terminal residues. A BLASTn search (from coordinates 3460 to 3790) against all the bacterial databases retrieved 108 top hits, all of them belonging to plasmids isolated from Streptococcus, Enterococcus, and Staphylococcus. Most of the homologies extended over the first 160 nt. Exceptions were found for the enterococcal plasmid pVA380 (L23803.1) (267) and pRGI00402 (HG796317.1) from the metamobilome of rat cecum (268), in which the homologies extended from coordinates 3551 to 3790. Furthermore, two other plasmids were found in which the homology extended throughout the sequence: plasmid pRW35 from S. pyogenes isolate 9116-03 (EU192194.2) and plasmid pRW35-like from S. pyogenes strain TSPY556 (CP032701.1). What could be the meaning, if any, of these sequences? Taking into account the role of small ORFs in bacteria and eukaryotes (269), we could elaborate on some hypotheses. It has been shown that the translation of some ORFs per se may have a regulatory role that is independent of the peptide produced (270, 271). For example, the translation of small ORFs in the 5′ untranslated regions (UTRs) (termed upstream ORFs [uORFs]) may decrease the translation efficiency of the downstream gene (272). Such a decrease would be a sensible response to downregulate the synthesis of MobM (in addition to its self-regulation), as it is expected for conjugative and mobilizable plasmids from G+ bacteria (273). Alternatively, ORF109 might be considered either an untranslated upstream region of gene mobM that would reflect the existence of an extinct accessory factor such as TraK_pKM101 or MobC_pC221 (212, 274) or a useless “leftover” of recombination processes. If the latter is correct, the MobM relaxase would be enough for self-regulation and relaxation activities, as has been reported for the TraB protein encoded by Streptomyces conjugative plasmids (275).

SPECIALIZED HOST FUNCTIONS

Despite all the years of research on RCR plasmids, little is known about the participation of host-encoded functions in their replication and conjugative transfer. Exceptions are the DNA Pol I and DNA Pol III proteins, for which indirect evidence has been reported (147, 228, 276). More direct information has been obtained for RNAP (147, 277), and recently we have shown that the main pneumococcal sigma factor, SigA, recognizes the promoters of the copG-repB (replication) and mobM (conjugative mobilization) genes from plasmid pMV158 (123). Also, as discussed below, the participation of the PcrA helicase in the replication of some RCR plasmids has been investigated in detail.

PcrA Helicase

The replisome of RCR plasmids requires a helicase to unwind the dsDNA. Usually, PcrA is the helicase used by RCR plasmids from G+ bacteria, whereas those from G− bacteria would use the UvrD helicase. The two proteins share ∼40% sequence identity and are structurally similar (278, 279). The role of the staphylococcal PcrA helicase in the replication of pT181 has been studied in depth at the genetic, biochemical, and molecular levels (223, 280). Furthermore, interactions between the highly conserved C-terminal region of PcrA and the host RNAP have been reported (281). However, there has been a lack of information on how the replisome is loaded onto the dso region of pT181 until more recent findings clarified this process. The crucial stage of RCR initiation is the generation of the nick by the plasmid-encoded RepC initiator, which, in turn, would allow the loading of the helicase (282). These interactions have been elegantly solved for the PcrA helicase of S. aureus and the RepC initiator of plasmid pT181. The three-dimensional structure of the RepDE chimera (see “The dso Regions and Their Cognate Initiators” above) showed the presence of a horseshoe-like cavity (Fig. 10) in which the catalytic Tyr and DNA-binding residues are located inside and outside this region, respectively (207). Amino acids participating in the interaction with PcrA were located at the open end of the horseshoe. This observation permitted the construction of a model in which PcrA helicase would bind in front of the RepDE dimer to generate a ring-shaped arrangement that would clasp the already extruded hairpin loop within the dso (207). The stability of the interactions between PcrA and RepC was addressed by the employment of magnetic tweezers, which permitted the unveiling of the DNA-protein-protein interactions at the plasmid dso, especially those related to the nicking and unwinding reactions (208, 282). These PcrA-RepC communications were concomitant with alterations in the DNA topology at the initiation stage. In the presence of RepC alone, the target DNA was twisted around −1.5 helix turns, a change that was insufficient to promote the dso hairpin extrusion. However, it was observed that PcrA and RepC were able to interact with each other on dso before the nicking reaction started, and this cross talk led to the introduction of additional negative DNA twists. Since the two proteins would be already bound at dso, cleavage by RepC and unwinding by PcrA would be a near-simultaneous reaction. The addition of the SSB protein resulted only in a slight increase in the unwinding rate, which was assigned to inhibition of ssDNA rehybridization. Despite the above results, the question whether all RCR plasmids require the participation of PcrA remains open, because it was shown that pMV158 (283) and pC194 (J.A. Ruiz-Masó and G. del Solar, personal communication) replicate in a pcrA3 genetic background, where pT181 is unable to replicate (284, 285). These findings suggest that PcrA could be involved in the replication of the Rep_trans plasmid family whereas another helicase would participate in the replication of Rep_1 and Rep_2 plasmids. A role of the B. subtilis PcrA helicase in conjugation rather than in replication has been proposed (286), pointing to a more complex role of PcrA, at least in B. subtilis (287).

DNA Gyrase

Obvious reasoning dictates a role for DNA gyrase in RC replication because relaxed plasmid molecules must be converted into supercoiled molecules to initiate the next round of either replication or mobilization. However, at least to our knowledge, direct experimental evidence on this role is lacking. In pMV158, we found that recognition by purified RepB and MobM proteins of their targets depends upon the degree of supercoiling of the plasmid DNA substrate (219, 288). In vitro relaxation of DNA from plasmids pLS1 (pMV158 derivative), pFX2, and pE194 (the three of them belonging to the Rep_2 family) by RepB_pMV158 was tested, and it was found that whereas the former two plasmids were efficiently relaxed, pE194 was resilient to relaxation even at high concentrations of RepB_pMV158. The degree of supercoiling of the plasmids was tested and found to be lower for pE194 than for pLS1 or pFX2, a finding that was attributed to the source of the plasmid DNA: pLS1 and pFX2 were prepared from S. pneumoniae, whereas pE194 was from B. subtilis (288). Plasmid DNA molecules with different degrees of supercoiling were prepared by relaxation with topoisomerase I in the presence of increasing concentrations of ethidium bromide. When such plasmid molecules were then treated with RepB_pMV158, it was found that the topoisomerase I-relaxed pE194 could be made sensitive to RepB_pMV158 but only at a relatively high ethidium bromide concentration (>8 mM). These observations indicate that the host DNA gyrase plays an important role in the degree of plasmid DNA supercoiling and, consequently, in the ability of a plasmid to replicate in a given host. We have been unable to establish pE194 in S. pneumoniae, despite the ermC gene being expressed in this host (189). Similar results were found when MobM_pMV158 was tested in its ability to recognize and cleave DNA from other plasmids of the MOB_V1 family, and in this case, pE194 was again the plasmid-less prone to be cleaved by MobM_pMV158 (219). Be that as it may, it would be interesting to perform experiments of pMV158 replication in pneumococcus strains affected in DNA gyrase B (289). Also, it would be interesting to extend to plasmids the approaches in which the bacterial chromosome was shown to be organized into topological domains with fixed locations (290).

Architectural Proteins

Little is known about the role of the architectural proteins in replication and mobilization of the RCR plasmids, although two observations made on pMV158 may be worthy of comment. First, the binding of RepB to its cognate dso resulted in substantial changes in the configuration of the helix, suggesting the generation of a DNA loop on supercoiled plasmid DNA. Second, between the two RepB binding sites (DDR and PDR) (Fig. 9), there is a tract of 10 successive A-T pairs, conserved in the Rep_2 plasmid family, which may function as a binding site of the nucleoid-associated protein HU. Now that the pneumococcal HU protein has been purified (291), it would be most interesting to perform experiments in this direction, especially after the finding of the weak DNA-binding ability of the E. coli HU protein and its role in chromosome dynamics (292). Other architectural proteins, such as H-NS, are not encoded by the pneumococcus, suggesting that they do not play an essential role in the biology of the RCR plasmids, at least in this bacterium.

RecA

Generation of highly recombinogenic ssDNA replicative intermediates is a feature of the RCR plasmids. This contributes to the ease with which these plasmids can recombine among themselves (293). Many of the RCR plasmids have been shown to spread easily by HGT and can cointegrate (i) with other RCR plasmids (57), (ii) with larger θ-replicating plasmids (58), and (iii) into the host chromosome (294). This last case has been reported for the staphylococcal plasmid pE194, in which short homologous sequences (6 to 14 bp) between the plasmid and the chromosome of B. subtilis were enough to yield a chromosomally integrated copy of the plasmid that was bordered by short direct repeats (294). The recombinogenic ability of the RCR plasmids may have led to the presence of “remnants” of their genes (e.g., Ab^r genes) in the chromosome of their hosts. However, an in-depth study of the influence of RecA and other recombination pathways (287) is still due. The pMV158 derivative plasmid pLS1 was unable to yield Tet^r colonies in a recA-deficient E. coli strain, neither the wild type nor a high-copy-number (N_av, ∼50) derivative (104). Nevertheless, when the high-copy-number derivative harbored the pC194 cat gene (instead of the original tetL), a small number of transformants were recovered, all of them having a very reduced number of copies (N_av, ∼1). When the same plasmid was introduced into a thermosensitive E. coli recA strain, it exhibited a normal copy number (N_av, ∼5) at 30°C, but only a faint band could be detected by Southern hybridization at the restrictive temperature of 40°C; no indication of plasmid integration into the host chromosome was obtained (104). It was concluded that an intact RecA function was very important but probably not essential for plasmid replication in the G− host. New findings on RecA-independent rearrangements between repeated sequences point to the participation of RarA (a 447-amino-acid AAA⁺-ATPase) in RecA-independent recombination events in E. coli, although its specific role remains to be fully clarified (295). To our knowledge, the only role of RecA that has been proposed for RCR plasmids has been a hypothetical RecA-dependent lagging-strand replication when no sso regions were present (see The Mysterious RNA I) (169). However, we have never tested the role of the pneumococcal RecA or RarA proteins on pMV158 replication, and the early observations (104) on the extremely low number of plasmid copies in the E. coli recA(ts) mutant strain were never understood nor was the subject pursued.

CONCLUSIONS AND PERSPECTIVES

Being small, RCR plasmids easily spread among bacteria because they represent a low genetic cost to the hosts and, many times, a huge profit for bacterial survival. Thousands of these plasmids have been sequenced, but studies on their role in the bacterial lifestyle are still scarce. Knowledge of the prevalence of the different plasmid families, their associations with different resistance genes, their host range, and their mobilization functions can be of utmost interest to understand the epidemiology and transmission of Ab^r traits by RCR plasmids. This, in turn, will lead to developing novel tools to find new replicons in bacterial species and to design intervention strategies to prevent the dissemination of Ab^r genes. There is also a need for in-depth knowledge of RCR plasmids to understand (i) their ubiquity, (ii) the mechanisms by which they spread, (iii) their influence on the bacterial world because they are shared among different bacterial species and, consequently, in complex communities, (iv) their relevance as vehicles to spread genetic information, and (v) their role in the appearance of novel traits in community- and hospital-acquired bacterial infections (296).

The strategy of “travel light” used by RCR plasmids (i.e., low genetic load to their hosts) has allowed them to colonize a wide variety of bacteria. RCR plasmids have also become part of the whole bacterial mobilome because they may convert themselves into permanent, and yet mobile, entities when they become IMEs (15). The panorama of the mobilome, as we see it today, is an amalgam in which we try to discriminate and classify a wide collection of elements. We have discovered that relaxases of the MOB_V family have representatives in large and small replicons (Table 1), without any clear reason for this distribution. Research on RCR plasmids has contributed to the knowledge of their replication mechanisms, but much is yet to be investigated. We know little about the structures of the replication initiators and the mobilization relaxases, and we are ignorant of possible “hidden” functions encoded by these replicons. But, more importantly, little is known (and too much is assumed) about the host functions participating in the vertical and horizontal inheritance of RCR plasmids, aside from the interactions between the staphylococcal PcrA helicase and the pT181-RepC initiator. Moreover, even in this case, it is still poorly understood how plasmids like pMV158 can become independent of PcrA. Is it because RepC is a dimer and RepB a hexamer? Assembly of the replisome must be necessarily different in both cases, but just with one example of each mechanism, and with only two structures of the initiators solved (one chimera and one natural), we cannot elaborate further on this hypothesis. RCR plasmids have a high copy number, and they overreplicate when they first colonize a new host to reach at least twice the N_av value before cell division (115, 116). This indicates that to be stably maintained, the incoming plasmids must either hijack the host functions they need or make use of host function “surpluses” of the abundant proteins, like RNAP, DNA polymerases, gyrase, SSB, and helicases. Be that as it may, there are too many gaps to be filled before we can consider RCR plasmids a closed case.

The spread of RCR plasmids by any of the HGT processes (see Introduction) should be viewed as a wealth of genetic biodiversity that contributes to the evolution of the bacterial species and their adaptation to changing milieus. Thus, to maintain our environmental equilibrium, we must consider all biological entities as a single unit, including bacteria and their MGEs. It is within this framework that we want to define the concept of One Earth, because we should not consider bacteria as an enemy that we must fight. Such a belligerent attitude toward our microbiological world is erroneous, a view also recently shared by others (297). Under the One Earth umbrella, we propose aiming for an entente cordiale with our bacterial world and thus understanding how bacterial populations behave and strive for resources in the different competitive niches they inhabit (298). A peaceful coexistence with bacteria should aim to know and attenuate their virulence traits rather than inhibit HGT processes. Antivirulence therapies do not hinder bacterial growth but limit the production of virulence factors. To cope with the colonization of new niches, bacteria have to reprogram their gene expression in response to external stimuli (the nichome), and this can be achieved by specialized global regulators yet to be precisely known (299). The large number of MGEs constituting the mobilome (1) enriches the genetic biodiversity found in bacterial populations. Their coding ability contributes not only to the Ab^r (the resistome) (300) but also to the procurement of novel traits that may be beneficial for a given population under different environmental conditions. These approaches will help our comprehension of life as a whole and will help to preserve a balanced equilibrium within our biosphere.

Biographies

M. Pilar Garcillán-Barcia received a degree in Biochemistry and a Ph.D. in Biology. She is a CSIC Tenured Scientist at the Institute for Biomedicine and Biotechnology of Cantabria, where she leads the Functional Plasmidomics group. Her main research interests are the study of mobile genetic element evolution and diversity and plasmid functional adaptations in bacterial communities, with a special focus on plasmid-encoded secretion systems.

Radoslaw Pluta is a Marie Sklodowska-Curie postdoctoral fellow at IRB Barcelona. He received an M.Sc. degree in Molecular Biology (University of Wroclaw) and a PGCert in Protein Crystallography (University of London). Before pursuing his doctoral degree at IRB Barcelona, he gained additional international research experience at the University of Groningen and EMBL Grenoble. He earned his Ph.D. in 2014 from Pompeu Fabra University, Barcelona, for his work elucidating the structural basis of the MobM relaxase-mediated mechanism for gene transfer and spread of antibiotic resistance. During his first postdoc, he worked on the structural biology of gene regulation by riboswitches (IIMCB Warsaw, 2014–2018). Currently, he focuses on structural determinants and small-molecule targeting of gene regulation by TGF-β signaling SMAD complexes. He has published several articles, coorganized two conferences, and serves as a review editor for Frontiers in Molecular Biosciences and an ad hoc reviewer for Frontiers in Microbiology, Nucleic Acids Research, and Nature Communications.

Fabián Lorenzo-Díaz studied Biology at the University of La Laguna (ULL, Canary Islands, Spain). ln 2004, he started his Ph.D. at the CIB Margarita Salas (CSIC, Madrid), working on the molecular mechanisms of horizontal gene transfer supervised by Manuel Espinosa. During that period, he performed stays at the Molecular and Cell Biology Institute (Rosario, Argentina, 2006) and at the Johns Hopkins University (Baltimore, MD, USA, 2007). After obtaining his Ph.D. (summa cum laude), he returned to the Canary Islands, working in different groups at the University of La Laguna and the University Hospital NS Candelaria. He performed a postdoctoral stay at the Institute ZIK-Septomics (Jena, Germany, 2014) to work on deep-sequencing projects. In 2015, he got a position as Assistant Professor in the Genetics Department (ULL). Currently, he is involved in research projects aiming to analyze the role of the human microbiome and genetic variation in asthmatic patients.

Alicia Bravo has been a group leader at the Centro de Investigaciones Biológicas Margarita Salas (CIB), Madrid, since 2012. She obtained her academic degrees in Biological Sciences from the Complutense University, Madrid. Her Ph.D. studies at the CIB led to the discovery of the plasmid-encoded Kis-Kid toxin-antitoxin system. From 1988 to 1990, she was a postdoctoral fellow at the Max-Planck Institut für Molekulare Genetik, Berlin. During this tenure, she was supported by the European Molecular Biology Organization and worked on bacteriophage SPP1-DNA packaging and recombination-dependent plasmid DNA replication. From 1991 to 2005, she worked on bacteriophage phi29-DNA replication at the Centro de Biología Molecular Severo Ochoa, Madrid, and was awarded a research contract from the Ramón y Cajal Programme. In 2006, she moved to the CIB and established her independent research line on global transcriptional regulators in Gram-positive bacteria. At present, her group aims to understand how opportunistic bacteria adapt to new environments.

Manuel Espinosa is Professor ad honorem at the Centro de Investigaciones Biológicas Margarita Salas, CSIC, Madrid. He was born in Los Realejos, Tenerife, Canary Islands, and has studied general sciences, physics (Complutense University, Madrid, UCM), and biology (UCM). He received his degree in 1966 and his Ph.D. in 1969 (UCM). Postdoctoral stays were spent at Groningen University (The Netherlands, 1974) and Institute of Biochemistry and Biophysics (Polish Academy of Sciences, Warsaw, 1979–1980). He moved to the Biology Department, Brookhaven National Laboratory, USA, in 1980. He participated in developing the first cloning system in Streptococcus pneumoniae. Back in Spain he worked on rolling-circle replication (1985) and DNA transfer (1997) of plasmids from Gram-positive bacteria and on pneumococcal toxin-antitoxins (2006). He was director of the CIB Margarita Salas (1992–1993). He was elected a member of the European Molecular Biology Organization (EMBO, 1997) and served on EMBO Committees of the Young Investigator Programme (2007–2010) and Fellowships (2008–2012). He has been the principal investigator of numerous national and international grants. His current research interests include streptococcal conjugative relaxases and pneumococcal toxin-antitoxin systems.

This article is dedicated to our colleague and friend Dhruba K. Chattoraj on the occasion of his official retirement.