Bacteria DNA conjugation: from the cellular to the community level

Bacterial conjugation, also referred to as bacterial sex, is a major horizontal gene transfer mechanism where the DNA is transferred from a donor to a recipient bacterium by direct contact. Conjugation is universally conserved among bacteria and occurs in a wide range of environments (soil, plant surfaces, water, sewage, biofilms and host-associated bacterial communities). Within these habitats, conjugation drives the rapid evolution and adaptation of bacterial strains by mediating the propagation of various metabolic properties, including symbiotic life-style, virulence, biofilm formation, or resistance to heavy metals and, most importantly, resistance to antibiotics. These properties make of conjugation a fundamentally important process at the center of extensive study. Here, we review the key steps of conjugation by following the life-cycle of the F plasmid during transfer from the donor to the recipient cell. We also discuss our current knowledge of the extent and impact of conjugation within an environmentally and clinically relevant bacterial habitat, bacterial biofilms.


Introduction
Conjugation was first discovered in 1946 by Edward Tatum and Joshua Lederberg, who showed that bacteria can exchange genetic information by unidirectional transfer of DNA, mediated by a socalled F (Fertility) factor, [1]. It was later realized that the F factor is a replicative extra-chromosomal genetic element, for which they later coined term plasmid, that can be transferred across the cell membrane of the parental strains. Since this seminal discovery, the identification of a plethora of conjugative elements, including plasmids, conjugative transposons and integrative conjugative element (ICEs), has revealed that conjugation is a universally conserved DNA transfer mechanism among Gram-negative and Gram-positive bacteria [2,3]. Conjugation was also shown to be a ubiquitous process that occurs in environment bacterial communities present in the soil, on plant surfaces, in water and sewage, as well as in biofilms and bacterial communities associated with plant or animal hosts [4]. Within these niches, conjugation facilitates the adaptation of bacterial strains by mediating the propagation of advantageous metabolic properties, such as symbiotic lifestyle, virulence, or resistance to heavy metals and antimicrobials. Conjugation is therefore a major driver of the rapid evolution of bacterial genomes [5,6]. Such fundamental importance has made of conjugation the focus of extensively study over the last decades. Experimental approaches have provided a detailed understanding of the molecular mechanism of conjugational DNA transfer, while systematic sequencing has uncovered the extent of conjugation at the ecological scale.
Conjugative plasmids generally carry all the genes required for their maintenance during the vertical transfer from the mother to the daughter cells, as well as the genes necessary for horizontal transfer during conjugation from the donor to the recipient cell. These functions are encoded by different regions or modules that compose what is generally referred to as the plasmid backbone.
Isolation and sequence analysis of an increasing number of conjugative plasmids has revealed considerable diversity in terms of genetic properties and organisation. This diversity also indicates that different plasmids might use various regulations, molecular reactions and strategies to achieve productive conjugational transfer and maintenance.
In this article, we review the key steps of conjugation by following the life-cycle of the plasmid during transfer from the donor to the recipient cell ( Figure 1). We focus on the first discovered and extensively described F plasmid, which we use as a paradigm to discuss other conjugative systems. The first chapter describes events occurring within the donor cell, i.e., the expression regulation of the plasmid tra genes required for conjugation; the processing of the plasmid by the relaxosome prior to transfer; the composition and function of the conjugative pilus in the mating pair formation process; the central role of the Type IV coupling proteins (T4CPs) and the transfer by the type IV secretion system (T4SS). The second chapter focuses on the dynamics of the newly acquired plasmid within the recipient cell, i.e., the plasmid establishment, which includes protection against host systems dedicated to foreign DNA elimination, early expression of leading genes and the conversion of the ssDNA plasmid into dsDNA; the plasmid maintenance, which includes plasmid replication and segregation; and the eventual phenotypic conversion of the transconjugant into a new donor cell with novel metabolic properties. In the third chapter, we review our current knowledge of the extent and impact of conjugation within an environmentally and clinically relevant bacterial habitat, bacterial biofilms.

Regulation of tra genes expression
The ability of the donor strain to perform conjugation requires the expression of the transfer genes clustered in the tra region of the plasmid. The transfer genes encode all the protein factors involved in the elaboration of the conjugative pilus and the T4SS required for formation of the mating pair, as well as the relaxosome components needed for the processing of the plasmid prior to transfer ( Figure 1, step i). The expression of tra genes is regulated by several factors, including plasmid and host proteins, cell-cycle progression and environmental conditions. Most tra genes are gathered in one operon under the control of the PY promoter, while traJ and traM genes are located upstream and controlled by independent promoters (Figure 2) [7]. Transfer gene expression follows a specific regulation cascade that starts with the production of TraJ protein ( Figure 2, Step1), which activates the PY promoter and the transcription of the tra operon ( Figure 2, Step2). The first gene to be transcribed, traY, encodes the TraY regulator protein that activates the PM promoter, resulting in the production of the relaxosome accessory protein TraM (Figure 2, Step3) [8]. Therefore, this regulation cascade results in the expression of all genes involved in the elaboration of the conjugative pilus, the T4SS system and the relaxosome composed of TraY, TraM, and TraI. It is observed that tra genes are normally repressed, most probably to prevent the fitness cost that would be associated with their constitutive expression [9]. It is interesting to remark that most regulation systems act by modulating cellular levels or activity of the primary activator TraJ. In most F-like plasmids (R100, R1, R6-5, ColB2-K77), the expression of traJ, and therefore that of other transfer genes, is repressed at the post-transcriptional level by the fertility inhibition system FinOP ( Figure 2) [10,11]. FinP is an antisense RNA complementary to the stem-loop structures of the traJ mRNA. FinP binding hides the ribosome binding site and prevents TraJ translation [12,13]. FinO is an RNA chaperon that protects FinP for degradation by RNase E and stabilizes the formation of the duplex FinP-mRNA traJ [14][15][16]. Besides, tra gene expression is also regulated by chromosomal-encoded host factors (reviewed in [17]). One such regulation involves the silencing of PY, PM, and PJ promoters by the chromosome encoded histone-like nucleoid structuring protein H-NS [18,19]. H-NS copy number per cell varies during growth [20], thus rendering F plasmid transfer rate growth-phase-dependent, i.e., maximum in exponential, reduced in mid-exponential and mostly abolished in stationary phase [21,22]. Yet, during the exponential phase, H-NS repression activity is itself counteracted by the cooperative binding of TraJ and the host protein ArcA (Aerobic respiration control of anoxic redox control) to the PY promoter [23]. In the case of the virulence plasmid pSLT of Salmonella enterica, H-NS repression activity was also reported to depend on Dam (DNA adenine methylase) methylation of the DNA [24]. Other examples of host factor-mediated regulation of tra gene expression includes the repression by the RNA binding protein Hfq that destabilizes both traJ and traM transcripts [18], and by the GroEL chaperone proteins that directly activate plasmid R1 TraJ proteolysis during the cellular heat shock response [25].
For some conjugation systems, tra gene expression is regulated by quorum sensing (QS) mechanisms. This is the case for the Tumor inducing conjugative plasmid (pTi), which allows Agrobacterium to infect and disseminate within plant hosts. At high cellular density, Agrobacterium produces agrocinopines molecules that activate different operons, including the arc operon that encodes TraR (unrelated to the F TraR protein), a LuxR-like protein. Binding of TraR to the QS molecule 3-oxo-octanoylhomoserine lactone (OOHL) triggers the transcription of trb and tra operons, resulting in the production of the T4SS and the relaxosome proteins. The QS lactonase BlcC is also produced, allowing the degradation of OOHL molecules in the stationary phase or during carbon and nitrogen starvation associated with the host plant death. Such regulation provides coordination between the pTi conjugation proficiency and the host state or to the bacterial density within the plant tissues during infection [26]. Pseudomonas aeruginosa uses QS as a defense mechanism against interspecies conjugation through the production of the QS molecule N-acyl homoserine lactone (AHL), involved in P. aeruginosa populational regulation of mechanisms such as virulence, biofilm formation, and metabolism [27]. AHL produced by P. aeruginosa can bind to the E. coli LuxR-like transcriptional factor SdiA, which then represses traI gene expression and prevents the conjugation of the RP4 broad-host-range plasmid integrated into the chromosome of the E. coli donors.
Regulating the expression of transfer genes is the chief strategy to modulate the transfer efficiency of conjugative plasmids. The above examples illustrate that tra gene expression is controlled by complex regulatory circuits, which involve the combined activities of plasmids and chromosomal host factors. This tight regulation allows controlling the transfer efficiency in connection with the plasmid life-cycle and the host physiology, in response to environmental conditions and populational interactions.

Superspreader mutations
Over the years, several studies have revealed genetic modifications, so-called superspreader mutations, which dramatically enhanced the conjugation efficiency of conjugative plasmids belonging to diverse incompatibility groups. The first superspreader mutation was characterized on the F plasmid, which carries an IS3 insertion sequence into the finO gene. FinO inactivation destabilizes the FinP-mRNA traJ duplex, thus resulting in the up-regulation of traJ and the constitutive expression of tra genes [28]. This naturally occurring mutation accounts for the enhanced transfer efficiency of the F plasmid compared to the related IncF plasmids R100, R6-5, and R1, in which the FinOP regulatory system is still active [29]. More recently, genetically induced superspreader mutations of several resistance plasmids have been isolated in laboratory settings. In the IncI plasmid pESBL associated with extended-spectrum beta-lactamases production in Enterobacteria, inactivation of the Hft locus triggered the overexpression of conjugative pili and 20fold enhancement of the transfer efficiency [30,31]. In the Citrobacter freundii IncM group plasmid pCTX-M3 that carries the blaCTX-M-3 gene, the deletion of two genes (orf35 and orf36) resulted in the enhanced expression of the tra genes and increased plasmid transfer [32]. Another example was reported in the Gram-positive broad-host-range (Inc18) plasmid pIP501 involved in the propagation of vancomycin resistance from Enterococci to methicillin-resistant strains of Staphylococcus aureus. In this case, the deletion of the traN gene encoding the small cytosolic protein TraN (unrelated to the F TraN protein) resulted in the upregulation of transfer factors and enhancement of the transfer efficiency [33].
Inducing the over-expression of plasmid transfer genes might not be the only way through which superspreader mutations increase the transfer efficiency of conjugative plasmids. It was shown that insertion of the Tn1999 transposon in the tir (transfer inhibition of RP4) gene of the IncL/M-type plasmid pOXA-48a, responsible for the dissemination of specific extended-spectrum β-lactamase genes in Enterobacteriaceae, increases the transfer efficiency by 50-100 fold without affecting traM expression levels [34]. The mechanism by which the inactivation of the Tir protein enhances transfer efficiency remains to be elucidated. These studies show that superspreader mutations can emerge by various mechanisms in different conjugative plasmids, and have the potential to aggravate the spread of drug-resistance plasmids among bacterial organisms.

F pilus structure and biosynthesis
Bacterial conjugation is a contact-dependent horizontal gene transfer mechanism that involves a conjugative pilus associated with a T4SS. Electron microscopy imaging was instrumental in analyzing the morphology of numerous conjugation pili encoded by plasmids belonging to different incompatibility groups [35][36][37][38]. These studies revealed that pili fall into two main morphological categories, thin flexible and thick rigid, which influence the ability to support conjugation in liquid or on a solid surface. F encodes a thin flexible pilus, which has a tubular structure ~8 µm in diameter and up to ~20 µm in length and which is constituted by the helicoidal arrangement of a unique protein subunit, the F pilin or TraA [39-44] (Figure 1, step i). The traA gene encodes a 121 amino acids pro-pilin peptide, that is subsequently maturated into a 70 amino acids F Pilin [45,46]. The maturation process involves TraQ and TraX proteins [47]. TraQ chaperone-like protein binds transiently to TraA pro-pilin precursor, thus allowing its accumulation into the inner membrane by an ATP-dependent pathway [48] and giving it the right conformation for a signal peptide cleavage. Processing of pro-pilin into matured pilin requires both the cleavage by the host leader peptidase B (LepB) and the N-terminal acetylation by TraX [49,50]. This maturation process ensures the availability of pilin subunits into the inner membrane before the assembly of the pilus by TraE, TraK, TraB, TraV, TraC, TraW, TraG, TraF, TraH, TraL, TrbC encoded by the transfer region [51][52][53]. Mutational experiments showed that this set of proteins can be separate in different functions. Briefly, TraE, K, C, G, L are responsible for the assembly of the tip, while TraB, V, W, F, and H are important for the pilus extension and TrbI for pilus retraction [52,53]. The pathway of F-pilus biosynthesis has been extensively reviewed [52,54] and will not be detailed here (Table 1).  Periplasm), and the homologues in RP4, pTI or R388 plasmids are shown.

Pilus biological function
The role of the F-pilus in conjugation has been actively debated. It was early proposed that the Fpilus extends to contact the recipient cell, then retracts to bring together the donor and recipient cells and form the mating pair [41,67,68]. This idea was convincingly supported by the direct visualization of the F pilus dynamics in live-cell, using a fluorescently labeled R17 bacteriophage that specifically binds along the pilus sides [69]. This work showed that donor cells produce flexible pili that continuously undergo cycles of extension and retracting, thereby probing the surroundings, regardless of the absence or presence of recipient cells. Yet, when contact is established with a recipient, the pilus retraction draws the cells together, resulting in the formation of mating pair [69]. In liquid culture, the pili mediate the formation of larger mating aggregates containing donor and recipient cells in tight wall-to-wall contact [70,71].
Allowing the formation of the wall-to-wall contact between the mating partners might not be the only role of the F-pilus, which was also proposed to serve as a channel conducting the singlestranded DNA during conjugation between distant donor and recipient cells [72]. Undeniably, the pilus axial hole has a diameter (30 Å) large enough to accommodate the DNA molecule [41,44]. Yet, only a few reports provide evidence for conjugational DNA transfer between mating partners spatially separated from each other. It was shown that DNA transfer could occur between donor and recipient separated by 6 microns membrane with pores 0.01 to 0.1 micron in diameter [73]. Also, microscopy imaging provided some evidence for DNA acquisition by recipient cells that are not in direct contact with a donor cell [74]. Yet, F-pilus ability to conduct DNA is still questioned and awaits for the clear visualization of DNA transfer between distant donor and recipient cells only connected by a pilus.

Factors involved in the specificity of donor-recipient interactions
The pilus ability to establish contact between donor and recipient cells can be considered as the first rate-limiting step in the conjugation process, and a key determinant of plasmid host-range specificity. In the seventies, numerous studies attempted to identify specific recipient receptor required for F plasmid transfer [70,[75][76][77][78][79][80]. The results revealed that mutations localized in the fourth external loop of the major porin OmpA or that alter the inner core composition of the lipopolysaccharide (LPS) affect the transfer of the plasmid F and other IncF-like plasmids such as R386, R538-1drd, and R1-19 but not the IncFII-type plasmid R100-1 [81][82][83][84]. Analysis of several ompA and LPS mutants revealed that they do not affect pili binding but result in defective mating-pair  [99]. Largescale mutants screening using the E. coli Keio collection or random E. coli insertion mutant library failed to identify recipient mutants affecting the transfer of the plasmid RP4 [100] or plasmid R388 [101].
Remarkably, some broad-host range IncP-like plasmids can also be transferred in archaea [102] and eukaryotes such as yeast [103] and mammalian cells [104]. Although the efficiency of conjugation varies among recipient type cells, these findings strongly suggest that plasmid transfer does not require any specific factors or active mechanism on the recipients side. Alternatively, a "shoot and pump" conjugation model envisages that the type IV secretion apparatus could act like a syringe able to inject DNA in any kind of membranes, using the pilus as a needle [105]. Perforation of the recipient bilayer cell-wall could be achieved by force or by dedicated enzymatic activity exposed at the pilus tip. The lack of requirement for specific receptors on the surface of the target cell is not an exception of the conjugative T4SS. Indeed, although the structural components of the Type VI secretion system machinery have been widely documented, no studies have yet characterized genetic factors that could act as target receptors on the prey cell surface.

Plasmid processing by the relaxosome
The initiation of conjugation requires the assembly and activity of a protein complex, the relaxosome, that processes the plasmid before DNA transfer ( Figure 1, step ii). The relaxosome introduces a site-and strand-specific DNA cut (nick) at the nic site located in the origin of transfer oriT and is required for the extrusion of the single-stranded DNA to be transferred (T-strand) [106][107][108]. In the F system, these two reactions are performed by the multifunctional TraI relaxase protein, which has both a transesterase domain that catalyzes the nic reaction and a DNA helicase domain that unwinds the plasmid DNA [64, [109][110][111][112][113][114]. Crucially, TraI recruitment and activity are governed by auxiliary proteins, including the integration host factor IHF and the plasmid-encoded TraY and TraM proteins, which have distinct roles in the relaxosome formation and activity at oriT [115][116][117].
Binding of IHF and TraY to their respective cognate binding sites ihfA/ihfB and sbyA located at oriT modulate the architecture of the DNA, thereby stimulating the loading of TraI [118][119][120][121][122]. TraM is a small protein that regulates its own expression by binding to sbmA and sbmB sites located in PM promoter, and that stimulates the DNA relaxation reaction through direct interaction with TraI after binding to sbmC site located in oriT region [123-127]. TraI nicking reaction involves a catalytic active Tyr residue [65,128] and results in the relaxation of the plasmid dsDNA, where the 5'-phosphate (P) end of the nicked strand (or T-strand) remains covalently bound to TraI ( Figure 1, step iii) [111,113,[129][130][131][132]. After the nicking reaction, the circular ssDNA conjugative plasmid is converted into dsDNA by Rolling Circle Replication (RCR) in the donor, while the linearized T-strand DNA bound to TraI at the 5' end is transferred into the recipient cell through the conjugative pore ( Figure  1, step iv).

Initiation of rolling circle replication in the donor cell
Rolling circle replication RCR mechanism is employed for the vegetative replication of some bacterial plasmids and has been very well-reviewed [133][134][135]. RCR is key to the transfer process of many Gram-negative and Gram-positive conjugative plasmids, but also to the infection cycle of other mobile genetic elements like DNA/RNA viruses and bacteriophages [128,136]. RCR reactions involved in vegetative replication or in plasmid transfer are very similar. The initiation and termination of RCR reactions performed by the Rep protein during vegetative plasmid replication are achieved by the relaxase protein during conjugation. Indeed, Rep and relaxases exhibit closely related functions, primarily allowing RCR initiation by nicking the double-strand DNA at the dso site (double-stranded-origin) or the oriT site, respectively [137]. The nicking reaction generates a 5'-P end that remains covalently bound to Rep or TraI and a 3'OH end used as a primer for the host DNA polymerase III. While the DNA polymerase III performs leading strand elongation, the parental double helix is unwound, and RCR ends by a second nicking reaction that releases the newly synthesized strand DNA ( Figure 1, step v). In the case of vegetative RCR, DNA unwinding is performed by a host DNA helicase recruited by the Rep protein, while Rep itself ensures termination and second nicking reaction. One major specificity of conjugation-associated RCR is that replication of the two ssDNA strands occurs in different cells, i.e., the leading strand is replicated in the donor, while the T-strand (lagging strand) is transferred and replicated in the recipient cell ( Figure 1). Because the relaxase that initiates the nicking reaction is transferred together with the T-strand [65, 138,139], a second relaxase protein is required in the donor to perform the DNA unwinding as well as the second nicking reaction [3,65,128]. Consistently, structure analysis by Cryo-EM shows that two relaxase molecules bind the oriT, one associated with the 5' end that is in open transesterase conformation and one associated with the 3' end that is in closed helicase conformation [140].

T4CP connects the relaxosome to the T4SS
After processing by the relaxosome complex, the nucleoprotein complex composed of the T-strand and the covalently bound TraI needs to be recruited to the conjugative pore for transfer ( Figure 1, step iii). This connection is mediated by the interaction between the relaxosome and the Type IV Coupling Protein (T4CP) located at the cell membrane [3,141]. All conjugative systems have their own T4CP, such as TraD, TraG, and TrwB, respectively, for the F, RP4, and R388 plasmids. T4CP are not required for pilus production or DNA processing, yet they are key to substrate specificity [142]. Our understanding of the molecular interactions required for specific substrate recognition and translocation is still incomplete. A great deal of information has been provided by comparing various T4SS involved in protein or nucleoprotein transport [143][144][145][146]. It appears likely that conjugation systems derivate from ancestral protein translocation machinery that evolved to translocate DNA coincidently. In this view, the T4CP would serve as the substrate receptor that interacts with one or several relaxosome components to recruit the T-strand to the T4SS. In the case of the F plasmid and some others plasmid systems, it is well-established that TraD interacts with TraM relaxosome protein [147][148][149][150][151]. Interaction between the T4CP and the relaxase has been demonstrated in the case of RP4 [152] and R388 plasmids [153]. Such interaction has been speculated in the F system but remains elusive [112,154].
T4CP are DNA-dependent ATPases anchored to the cell membrane via their N-terminal domain and have been shown to interact with the T4SS components in R27 [155] and R388 plasmid systems [153]. T4CPs show similarities with membrane-anchored ring DNA translocases, such as SpoIIIE and FtsK involved in chromosome DNA translocation during sporulation and cell division, respectively [141,156,157]. T4CP binds non-specifically to DNA, with a higher affinity for ssDNA [152,158,159], on which it forms oligomers with enhanced ATPase activity [160,161].
Altogether, these findings led to the model that membrane-anchored T4CPs interact directly with the relaxosome and form hexameric structures on the T-strand, which they actively translocate through the conjugation pore during transfer. Yet, it remains unclear whether a signal is required to activate the coupling function of the T4CP in a donor cell where the T4SS and the relaxosome are already assembled and functional [3]. It has been suggested that the stability of the TraD oligomeric complex depends on a yet unidentified F-encoded protein, which could then be a key regulator of plasmid transfer activation [160]. It has also been suggested that the formation of the mating pair could transduce a signal to activate the T4CP and trigger the transfer of the processed T-strand [3]. Of note, it was shown the relaxase has to be unfolded to be translocated into recipient cells [139]. In A. tumefaciens, as in many other Gram-negative and Gram-positive systems, unfolding of translocated proteins is proposed to be performed by the VirB11-like ATPase, absent from the F plasmid system [52, 145,162]. However, one can reasonably presume that TraI unfolding also required ATPase activity involving one of the ATPases of the F T4SS.

Plasmid circularization by TraI
The relaxase is transferred into the recipient, where it is refolded and able to perform several activities required for completion of the conjugation process ( Figure 1, step a) [65, 138,163]. The helicase activity of the internalized relaxase is thought to perform 5′ to 3′ tracking of the T-strand. Pulling by the relaxase from the recipient together with pushing by the T4CP from the donor presumably facilitates the transport of the T-strand through the conjugation pore [3]. Once both extremities of the oriT are brought together in the recipient, the relaxase performs the joining reaction, resulting in the recircularization of the ssDNA plasmid ( Figure 1, step b) [65, 138,163,164]. An alternative model proposes that the nicking of the newly synthesized oriT occurs in the donor cell before transfer [165]. There is no evidence for the requirement of additional host or plasmid factors in the circularization of the internalized T-strand. After completion of the recircularization reaction, the recipient cell possesses a single-stranded circular copy of the conjugative plasmid.

Avoiding host defense systems against foreign DNA
The newly acquired ssDNA conjugative plasmid might be considered as a foreigner DNA, against which host bacteria have developed defense mechanisms, such as restriction-modification, exonucleases, and recombination system or adaptive immunity as the CRISPR-Cas [166]. Despite these defense mechanisms, horizontal gene transfer plays an important role in genomic evolution (5-6% of bacterial genomes and up to 20% in some organisms) [6, 167,168], implying that transferrable plasmids have evolved adaptive mechanisms to counteract these host defenses.
Restriction-modification system (RM) is an ubiquitous defense mechanism found in 90% of sequenced bacterial genomes and other prokaryotes based on restriction enzymes and methylation [169]. Restriction-modification mechanisms are described as a « primitive immunity system » against exogenous DNA [170]. These systems are based on restriction enzymes that target specifically unmethylated DNA sequences located on the newly acquired mobile genetic elements, while the host DNA is protected by methyl groups added to specific adenine or cytosine [170]. Yet, plasmids have evolved various strategies to counteract enzymatic DNA degradation upon entry in the new host cell. IncN, IncI, and IncF plasmids encode ArdA and ArdB proteins (Alleviation of restriction of DNA) that directly inhibit the REase (Restriction Endonuclease) by mimic DNA sequences, thus competing for enzymes target [171][172][173]. IncW plasmids encode the ArdC protein that protects the transferred T-strand by transiently blocking the restriction sites [174]. More recently, it was shown that a hde operon (host defense evasion) of IncI plasmids encode 2 genes involved in anti-RM (vcrx089 and vcrx090) [175] . Besides the production of inhibitory proteins, some plasmids completely lost the restriction sites, as it is the case for the RP4 plasmid [176].
CRISPR-Cas immune systems (Clustered Regularly Interspaced Short Palindromic Repeats and the CRISPR-associated protein) represent another defense mechanism against foreign DNA. CRISPR-Cas systems, found in ~45% of bacterial and up to 84% of Archaea genomes [177,178], have been described to protect against infection by bacteriophages, and more recently against plasmids acquisition [179]. It was then discovered that some phages encode an anti-CRISPR Acr protein that inhibits the activity of the CRISPR-Cas system [180,181]. Importantly, anti-CRISPR Acr loci have been identified in conjugative elements and plasmids of Listeria, Enterococcus, Streptococcus, and Staphylococcus [182]. These loci encode for the CRISPR-Cas inhibitors AcrIIA16-19 that prevent exogenous DNA nicking mediated by the Cas9 enzyme in vivo. Mahendra and co-workers have also shown that conjugation of a Cas9-targeted plasmid of E. faecalis was possible in the presence of these CRISPR-Cas inhibitors. Encoding Acr-like proteins is therefore an efficient strategy for conjugative plasmids to facilitate their dissemination by avoiding degradation by the host CRISPR-Cas immune system. Another strategy is for conjugative plasmids to encode a Bet/Exo system able to repair double-strand breaks caused by CRISPR-Cas during conjugation, as recently reported for IncC plasmid pVCR94 [175]. Expression of these genes inside the recipient bacteria after the acquisition of the V. cholerae pCVR94 plasmid allows survival against exogenous DNA defense mechanisms without the involvement of anti-CRISPR-Cas proteins.

Early expression of the leading region genes
The leading region of conjugative plasmids is the first to be transferred into the recipient cell during conjugation ( Figure 1, step a) [57]. The F plasmid leading region is well conserved, of a size of 13kb and directly adjacent to the oriT and encodes for at least eight proteins [183], including a homolog to the chromosomal SSB (SSBC), PsiB and other proteins of unknown function. Importantly, plasmid ssb (ssbP) and psiB genes are expressed early upon entry of the plasmid in the transconjugant bacteria, but not in the donor cells [184]. This was demonstrated using lacZ fusions of the two genes inside the ColIb-P9 plasmid, and also by immunoassays concerning the F plasmid PsiB protein [184]. Similarly, RT-PCR studies showed that psiB and ardA genes of the IncI1 plasmid begin to be expressed 5 minutes after transfer initiation [185]. These observations suggested that leading region genes could be expressed rapidly from the newly acquired plasmid in single-stranded form, before its conversion into dsDNA. It was later shown that the leading region contains a specific 328 bp Frpo region (for F plasmid RNA polymerase), which, when single-stranded, can form a stem-loop structure presenting -10 and -35 double-stranded boxes recognized by the host RNA polymerase that initiates the synthesis of RNA primers in vitro [186]. It was therefore proposed that Frpo may serve as a single-stranded promotor that allows the early expression of the leading region genes ( Figure 1, step a).
Frpo was also proposed to direct the single-strand to double-strand conversion of the F plasmid. In vitro assays showed that the RNA primers synthesized by the RNA polymerase from Frpo persists as a RNA-DNA duplex that is recognized by host DNA polymerase III to initiate complementary strand synthesis [186]. Frpo-type sequences also termed ssi for single-strand initiation sequence are found on various conjugative plasmids including R6K, R100, ColE1, ColE2, Col1B and RSF1010 [187][188][189][190] and are functionally comparable to sso sequences (single-stranded origin) involved in the rolling circle replication mechanism [128, 134,191]. These findings are consistent with the previous observation that complementary strand synthesis of the ssDNA F plasmid inside the recipient bacteria involves a cooperative mechanism between host RNA and DNA polymerases III [192].
Altogether, these findings led to the model that the Frpo could help initiating early gene expression and the DNA synthesis reaction that converts the ssDNA plasmid into dsDNA duplex, immediately upon entry of the T-strand into the recipient cell ( Figure 1, step a-c). Whether Frpo performs these functions during conjugation in vivo remains to be demonstrated.

PsiB inhibits the SOS response
In the recipient cell, the presence of abnormal amounts of ssDNA, usually associated with DNA damages, results in the induction of the SOS response [193,194]. More precisely, the loading of RecA recombination protein on ssDNA results in the formation of the pre-synaptic filament, which stimulates the autocatalytic cleavage of LexA, the repressor of the SOS regulon. The SOS response triggers the induction of the division inhibitor SulA, resulting in cell filamentation and potentially the death of the transconjugant cell. SOS also induces the production of nucleases and other DNA processing factors that could provoke the degradation or mutation of the transferred ssDNA, or its processing as a recombination intermediate [195]. To counteract these effects, several conjugative plasmids, including F, encode the PsiB protein (Plasmid SOS inhibition), which inhibits SOS induction [196]. The depletion of psiB has mild effects on the efficiency of conjugation but increases the host SOS response up to six-fold [184]. PsiB interacts directly with RecA, thereby inhibiting several activities such as DNA binding, LexA cleavage, and strand exchange reaction [197,198]. SOS response inhibition by PsiB is even more potent in presence of SSBC protein coating the ssDNA. PsiB is well conserved among conjugative plasmids and is considered important for the early steps of plasmid establishment in the recipient, consistent with its early production in the transconjugant cells [57,196].

Roles of the host and plasmidic SSB proteins in plasmid establishment
Upon entry in the recipient cell, the transferred ssDNA plasmid is coated by the host SSBC protein.
SSBC is an essential conserved protein that binds non-specifically to ssDNA. It is involved in various mechanisms, including DNA replication, repair and recombination, SOS-induction, and other DNA metabolic processes (for review see [199,200]. SSB binding protects the ssDNA against enzymatic degradation and increases the processivity of DNA polymerase II and III [201] during the replication reaction converting the ssDNA strand into dsDNA helix. The rapid recruitment of the host SSBC protein onto the transferred ssDNA has recently been visualized by fluorescence microscopy [202]. This work revealed that SSBC proteins are rapidly recruited onto the ssDNA that penetrates the recipient cell presumably protecting it and facilitating its processing ( Figure 1, step a). Interestingly, the F plasmid, as many other conjugative plasmids encode their own SSBP protein homologous to the E. coli SSBC [203]. One can then wonder what is the benefit for conjugative plasmids to encode their own SSBP, and what is the specific function of SSBP compared to the chromosomal SSBC?
SSBP can bind ssDNA without specificity and SSBP of different incompatibility groups (IncF, IncI, IncY, Inc9, IncT, and IncB/O) can partially complement conditional mutations of E. coli ssbC gene [204][205][206][207]. However, although producing plasmid F SSBP in trans enables the growth of ssbC deletion mutant, complemented mutant exhibit some filamentation, growth rate reduction [208]. Moreover, a reduced affinity to ssDNA is observed for plasmid F SSBP in comparison to the E. coli SSBC, and plasmid F SSBP cannot stimulate the reaction of DNA synthesis by DNA polymerase III in vitro [209]. Sequence alignment revealed that SSBP proteins complementing E. coli ssbC mutant share high homology only with E. coli SSBC N-terminal part [209]. The SSBC N-terminal region contains the domains for ssDNA binding and monomer-monomer interactions to cooperatively maintains the binding of the tetrameric structure of SSBC to ssDNA. This structural conservation would explain the ability of SSBP to bind ssDNA. However, the C-terminal domains of SSBP are quite more homologous to each other than to SSBC [209]. Yet, as this domain is interacting with partner proteins constituting the SSBC interactome, one possibility would be that the interactome of SSBP and the reaction it is involved in might be different than SSBC.
Not all SSBP has been shown to complement E. coli ssbC mutants [205]. The latter study found that the inability of SSBP from IncP-like RK2 to complement E. coli ssb1 mutation could be attributed to ssbP gene repression by the RK2 kor genes and that a derepressed plasmid indeed complement the thermosensitive growth of E. coli ssb1 mutations [210]. It is thus to consider the possibility that the numerous SSBP could also complement ssbC mutant.
To date, the function of plasmid F SSBP in the context of conjugation is still unclear. SSBP protein could contribute to the protection of the transferred ssDNA by inhibiting enzymatic degradation or the recruitment of host inhibiting proteins. However, its expression timing instead supports the idea that SSBP could be involved in assisting the complementary strand synthesis of the transferred DNA, or simply increase the pool of available single-strand binding protein required for the first cycle of vegetative replication of the plasmid.

Plasmid maintenance: replication and segregation
Maintenance of the newly acquired dsDNA conjugative plasmid in the recipient cell lineage depends on two main active mechanisms, plasmid replication and segregation of the plasmid copies to daughter cells over generations.
The mechanisms of plasmid DNA replication in bacteria have been extensively studied and is the focus of well-referenced reviews [128,133,[211][212][213][214]. Here we would like to stress the role of replication in conjugation host-range specificity. Indeed, the ability of the transferred plasmid to replicate in the recipient cells is considered as the primary determinant of plasmid host-range specificity and efficiency of conjugation. The plasmid F transfers efficiently to E. coli and relatively closed enterobacteria while no transconjugant can be recovered after mating with more distant bacteria such as Vibrio or Pseudomonas. As early 1982, host-range restriction was attributed to the plasmid's inability to replicate in the recipient bacteria rather than to the inefficiency of plasmid transfer per se [215]. This was demonstrated by showing that mobilizable plasmids containing a cisacting origin of transfer of the plasmid F and an origin of replication active in the tested recipient can then be transferred by the F conjugation machinery in Pseudomonas [215], V. cholerae [216] and yeast [103]. The same approach was employed to show the ability of pCTX-M3, an IncI-like plasmid, to use its conjugation machinery to transfer a mobilizable plasmid in the host recipients where pCTX-M3 does not replicate [32]. The failure of F plasmid replication in P. aeruginosa comes in part from an inability of the plasmid replication protein RepE of the RepFIA replicon in complex with host DnaA-oriS to form a stable interaction with the host helicase DnaB [217]. In contrast, broadhost-range plasmids such as RK2 regulate their maintenance by modulating alternative strategies of replication depending on the host [218]. These findings indicate that the host-range of the conjugation machinery and the replication origins belonging to the same plasmid differ and can be mechanistically uncoupled. The specificity of narrow-host range plasmids appears limited by the specificity of their replicon rather than by their transfer range.
Maintenance of the newly acquired dsDNA plasmid also requires the timely segregation of the plasmid copies to the daughter cell during transconjugant cell division. To do so, low copy-number plasmids encode active partition systems, which mechanism, biological functions and conservation have been extensively reviewed [219][220][221][222][223]. In the context of conjugation, it is interesting to mention another maintenance strategy that involves the integration of the newly acquired conjugative DNA to the chromosome. Chromosome integration ensures the stable inheritance of the conjugative elements by vertical gene transfer over generations. The first characterized example was again the F plasmid, which uses insertion sequences (IS) and RecA-dependent homologous recombination to integrate into the E. coli genome [224]. The integrated plasmid can still initiate the conjugative transfer of the whole chromosome of the resulting "Hfr" strain (High frequency of recombination). This process of transfer is both progressive and oriented, and the order of chromosomic gene transferred depends on the position where the F plasmid was integrated. Alternatively, the integrated F plasmid can be excised out of the chromosome and recover its original autonomous form [225]. Chromosome integration is widely used by mobile DNA such as ICE (Integrative Conjugative Elements), transposons, phages, in both Gram-negative and Gram-positive bacteria, with all these systems having their particularities, especially regarding the recombination systems used for their integration/excision [226,227].

Phenotypic conversion of the transconjugant
Expression of the genes carried by the newly acquired genetic element results in the phenotypic conversion of the transconjugants into a novel type of cell that exhibits additional metabolic properties (Figure 1, step d). The expression of plasmid genes involved in DNA transfer converts the transconjugant into a new donor able to further transfer the plasmid in the population, thus accounting for the exponential rate of conjugative plasmid dissemination (see tra gene expression section 2.1). Other plasmid genes encoded by the tra region of the plasmid backbone are not directly involved in the transfer process per se. Indeed, conjugative plasmids often carry immunity (or exclusion) gene systems that are widespread in Gram-negative and Gram-positive organisms [93,228-231,231-235]. These immunity systems limit the plasmid-carrying cell ability to serve as a recipient for the same plasmid [42,54,62,236]. Preventing self-mating by surface exclusion is thought to avoid the metabolic cost and potential cell death associated with repeated plasmid transfer, but also to be important for plasmid stability and evolution [62,236].
The F plasmid immunity system relies on two factors, TraT and TraS proteins, none of which is required for F pilus synthesis or DNA transfer [56, [237][238][239]. These two exclusion factors work at different levels. TraT is an abundant outer membrane protein that is thought to span the cell surface [237,238]. TraT production reduces the recipient to form stable mating aggregates, presumably by interfering with the interaction between the pilus and recipient surface receptors. Consistent with this idea, it was reported that TraT interacts with OmpA, further suggesting that it could compete with TraN, which is key to mating pair stabilization [87,88,240]. TraS is an inner membrane protein, which production only slightly reduced aggregation of mating populations, but reduced DNA transfer frequencies by 100-200-fold. It is then proposed that TraS acts by preventing DNA transfer when stable mating aggregates have already formed [237,238,241]. In F-like plasmid, it is proposed that TraS interacts with TraG to achieve the entry exclusion process [92,93]. For these reasons, TraS is referred to as an entry exclusion protein (Eex) and TraT as a surface exclusion protein (Sfx). Like other tra genes, traS and traT expression are controlled by TraJ, implying that transconjugant cells acquire plasmid transfer ability immunity and concomitantly during the phenotypic conversion.
In addition to genes located on the plasmid backbone, conjugative plasmids may carry additional genes that are not directly involved in conjugation, but in a variety of biological functions such as virulence, biofilm formation, symbiotic life-style, membrane trafficking, resistance to heavy metal and, most importantly, resistance to antibiotics. Acquisition of these metabolic functions potentially facilitate the bacterial adaptation and survival in changing environments, and make of conjugation a major driver of the evolution of bacterial genomes. The successful maintenance of conjugative elements in bacterial populations shows that this selective advantage compensates for the metabolic burden associated with the metabolism of the newly acquired genetic information (highjacking of the host replication, transcription and translation machineries) [242]. The most prominent example being the acquisition of conjugative drug-resistance plasmids, which enables bacterial proliferation in microbial communities containing antibiotic-producing organisms, or in antibiotic-polluted and clinical environments. Indeed, analysis of commensal, environmental and clinical antibioticresistant pathogenic strains revealed a multitude of conjugative plasmids carrying one or more genes for resistance to most, if not all, classes of antibiotics currently used in clinical treatments. Conjugation is considered to be the most widespread intra-and inter-species resistance transfer mechanism, accounting for 80% of acquired resistance [243].

CONJUGATION IN NATURAL HABITATS, THE EXAMPLE OF BACTERIAL BIOFILMS
Gene transfer by conjugation is known to contribute to the genetic dynamics of bacterial populations living in a variety of environments, including the soil, on plant surfaces, in water and sewage, as well as in bacterial communities associated with plant or animal hosts [4]. Bacteria are generally considered as planktonic unicellular organisms, yet in natural and clinical environments, they often live in complex structures called biofilms. Biofilms shelter bacteria against external hazards, but have also been proposed to offer a niche that facilitates the dissemination of drug-resistance determinants by conjugation. Below, we review our current understanding of the interplay between biofilms and bacterial conjugation.

The biofilm as a niche promoting bacterial conjugation
In natural environments, bacteria predominantly live in spatially structured communities termed biofilms, in which a self-produced extracellular matrix holds the cells together [244]. Planktonic cells that initiate biofilm formation can adhere to a living or inert surface (surface-attached biofilms) or can be present at the air-liquid interface as a free-floating community (pellicle). Bacterial biofilms are found in virtually every ecosystem on Earth, from aquatic systems (sludge, rocks, and wastercater) to terrestrial environments (rhizosphere) and human organisms (skin, intestinal, urogenital and respiratory tracts). Furthermore, biofilms are associated with persistent and severe infections because of their ability to colonize medical devices and implants. Indeed, biofilm structures offer protection against hostile environments and, more worryingly, against antibiotic treatments. The biofilm architecture depends on the bacterial species, the surface colonized and environmental conditions, but this lifestyle is characterized by the production of extracellular polymeric substances (EPS) mainly composed of polysaccharides, proteins, lipids, and extracellular DNA. EPS matrix production is dynamic and continuous and mediates the formation of the biofilm architecture in which aggregates of microorganisms are trapped.
Many studies on bacterial conjugation show that plasmids transfer can occur in both natural and artificial biofilms, from the aquatic environment [245], phytosphere [246], animal and human host [247,248] or reactor-associated biofilms [249,250]. These researches studied plasmid transfer at the population level and mainly relied on limited cultivation-based assays that probably underestimate the extent of conjugation in natural biofilm. Further works elaborated conjugative plasmid expressing a fluorescent marker allowing direct in situ visualization of donors, recipients and transconjugants at the single-cell level within several types of biofilm formed at the liquid-air surface [251], on semi-solid agar surface [252][253][254], on filter [255], in flow-chambers [253,256,257] or in a reactor [258].
The biofilm environment provides a high cell density and close cell-to-cell proximity that may facilitate HGT through bacterial conjugation. In line with a view of biofilm as a hot spot niche of conjugation, several studies showed a higher frequency of plasmid transfer in biofilm than in the planktonic mode of growth (reviewed in [259][260][261][262]. Although the biofilm appears as a favorable environment to HGT by conjugation, many studies at single cells level have reported limited plasmid propagation inside a preformed established biofilm beyond the contact zone between the donors and recipient (reviewed in [263]. These observations are directly linked to the complex spatial structure of the biofilm, which might have a critical impact on horizontal plasmid spread within a biofilm.

Impact of the biofilm structure on conjugation
The biofilm is a complex structure sometimes composed of mixed bacterial species. The matrix shapes the spatial organization by clustering cells in microcolonies in an architecture characterized by non-uniform cell arrangements, open channels, pores, cavities, and different layers of living cells [264,265]. Such an organization determines the formation of cell clusters/aggregates and could therefore influence the efficiency of conjugative transfer [251]. This possibility can be addressed using microscopy to analyze the distribution of cells active in conjugation within biofilms. To date, most studies investigated the spread of the GFP-tagged Pseudomonas putida TOL plasmid. Analysis of P. putida recipient biofilm established in flow chambers revealed that transconjugants appear at the top surface of the biofilm, but not in the deep lower layers, reflecting a limited invasion of the transferred plasmid [253,257]. On agar plate, transconjugants only appear at the contact zone between TOL plasmid-donors and the recipient colonies [252,253,266]. Limitation of transfer to the outer layers of biofilm was also observed for IncF, IncI and IncW plasmids in E. coli [254]. It has been proposed that transfer can only be efficient over a short period between metabolically active cells growing at the donor-recipient interface [253,267]. Consistently, it was shown that plasmid invasion stop in non-dividing cells [266].
Within the biofilm structures, chemical gradients of oxygen, nutrients, temperature and pH, create microenvironments that influence the metabolic activity of bacterial cells [268]. This results in physiological heterogeneity between the cells surrounding the border of the biofilm and those embedded deep inside. Also, variations of the spontaneous mutation frequency within the biofilm parts result in the emergence of variant subpopulations with genetic heterogeneity [269]. Whether and how these factors impact the spatial pattern of conjugation within biofilms remain unknown.

Impact of conjugative plasmids on biofilm formation
Several studies investigated an implication of the presence of conjugative plasmid from diverse incompatibility group on biofilm formation ability [270][271][272][273][274]. To initiate biofilm formation, planktonic cells produce cell appendages like flagella and adhesion factors such as type IV pili, type 1 and curli fimbriae [275,276]. Genes coding for these types of accessory factors that promote attachment to biotic or abiotic surfaces are often found on conjugative plasmids, resulting in host biofilm formation increase (reviewed in [260]). Examples include type 3 fimbriae of the IncX1 plasmid pOLA52 [277]), non-conjugative type IV pili of the IncI1 plasmid pSERB1 [278] or pilus-like structure and surface adhesins of the Enterococcus faecalis plasmids pBEE99 and pCF10 respectively [279][280][281]. In 2001, Ghigo made the unexpected observation that conjugative pili itself of derepressed plasmid F can promote biofilm of E. coli cells initially unable to form such structure and revealed the pilin TraA as the main adhesion factor inducing biofilm formation [270]. Microscopic structure analysis of derepressed IncF plasmid R1drd19-and F-carrying E. coli biofilm showed the rapid formation of a dense and mature 3D mushroom-type biofilm similar to the P. aeruginosa biofilm architecture [282]. This peculiar architecture formation and biofilm maturation generated by derepressed plasmids overrides the need for cell surface appendages essential to E. coli biofilm such as flagella, type 1 fimbriae, Ag43 or curli [282]. In contrast, maturation of 3D mushroom-type biofilm structure depends of curli production induced in E. coli by natural plasmid F, which does not constitutively express F pili [283]. The presence of the plasmid R1drd19 also increase E. coli biofilm formation by decreasing the motility and increasing the level of quorum sensing inducer AI-2 [284,285] and IncP-9 TOL plasmid increases in P. putida the production of extracellular DNA known to structure the biofilm [286] and thus the biofilm formation capacity [287]. However, the genetic mechanism by which conjugative plasmids increase the biofilm formation has not been elucidated, but the presence of repressed or natural conjugative plasmids affects the global host chromosomal gene expression [283,285].
While the role of the conjugative pili has been mainly studied in E. coli biofilm, one may consider their impact differ depending on the host. Røder et al. indeed observed that the conjugative pili of the IncP-1 plasmids pKJK5 reduce the surface attachment of P. putida by increasing cell-cell adhesion resulting in reduced biofilm formation [288]. Further investigation will be necessary to decipher the complex interconnections between the conjugative plasmid and biofilm formation.

Inflence of antibiotic treatment on conjugation within biofilms
Dynamics of conjugation within biofilms need to be more investigated, regarding the specific interconnection where biofilm and conjugative plasmids act in synergy to promote both community-building and gene transfer. This cooperation raises serious questions about the contribution of HGT in the evolution and adaptation of biofilm-forming pathogens. Because of the increase of antibiotic-resistant infections, recent advances tried to bring a new understanding of biofilm responses to antimicrobial treatments.
Sub-minimal inhibitory concentrations (sub-MIC) of aminoglycosides enhance biofilm biomass of P. aeruginosa strain PAO1 and clinical isolates of E. coli through the response regulator Arr, a predicted phosphodiesterase altering the cyclic di-guanosine monophosphate (c-di-GMP) levels [289]. Linares et al. further demonstrated that besides aminoglycosides, sub-MIC of tetracycline and norfloxacin also increase the formation of P. aeruginosa biofilm. Still, no clear causal factors were identified [290]. Interestingly, a combination of tetracycline with cephradine has a synergistic effect on biofilm formation of a mixed culture of E. coli and P. aeruginosa [291]. Tetracycline also promotes biofilm of the pathogen Acinetobacter baumannii and whole genome sequence analysis revealed an increase of rate mutation under sub-inhibitory drug exposition such as SNPs but also insertions and deletions [292]. Besides the accumulation of genotypic variation, biofilm treatment with a low level of antibiotics produces changes in the gene expression profile, some of which may be linked to the increase biofilm formation [290,292,293].
Recently, Diaz-Pascual et al., investigated Vibrio cholerae biofilm at the community-scale level using a single-cell imaging system revealing changes in biofilms dynamics and architecture in response to antibiotic treatment [294]. After tetracycline exposure, they observed modifications in biofilm architecture and cell morphology, including a 2.5-fold increase in cell volume and a 29% decrease in cell density. This cell density decrease reflected a considerable alteration of the multicellular arrangement and the breakdown of the matrix within the biofilm. Furthermore, biofilms became susceptible to the colonization of their interior by new cells and the colonizer population increased until they invade the resident biofilm. Clearly, a sublethal dose of antibiotics influences the biofilm lifestyle, inducing significant modifications to the entire population. The biofilm matrix forms a shield to prevent penetration and diffusion of antimicrobials and the increase of biofilm formation in response to antibiotics, illustrated by an enhancing of biofilm matrix, seems to be a defense mechanism of the bacterial community.
Interestingly, when conjugation occurs in a free-antibiotics environment with a donor strain pretreated with sub-concentrations of antibiotics, conjugation frequency increased significantly [304,305]. The mechanisms by which antibiotics affect plasmid transfer remains unclear. In the literature, it is proposed that sub-MIC antibiotics treatment enhance the frequency of conjugation through an up-regulation of the tra gene expression in donors [298,[304][305][306][307]. However, in many studies, the increase of conjugation frequency was evaluated using the antibiotic which resistance gene is carried by the tested conjugative plasmid itself [295,297,298,300,301,307,308]. This approach makes it difficult to distinguish between the selection bias induce by the antibiotic on the mating population, and the actual effect on conjugation frequencies. Two studies rather support that antibiotics primarily acts through differential selection of the donor, recipient and transconjugant once gene transfer has occurred, rather that stimulating conjugation per se [202,309]. By using living cell microscopy, they were able to visualize in real time conjugation dynamics. Nolivos et al. showed that transfer frequency of the F plasmid, harboring the tetracycline gene resistance, is not increased by the presence of tetracycline. Lopatkin et al. also demonstrated that antibiotics, from six major classes, not significantly increase the conjugation efficiency of plasmids from five different incompatibility groups. The reports suggested that the direct contribution of antibiotics on gene transfer have been overestimated and proposed that antibiotics may act only as post-transfer selection drivers, favoring the growth of transconjugants over recipients. Even if a number of studies have advanced the potential stimulating effect of antibiotics on conjugation, their real impact needs to be further explored and antibiotics with different modes of action tested.
Undoubtedly, antibiotics play a role that must not be overlooked in the emergence of new multiresistant pathogenic strains. It is troubling to admit that antibiotic treatments amplify biofilm formation, then strengthened the difficulty to heal infections biofilm-associated. Not only, antibiotics induce biofilm formation, but they improve gene transfer within the community. Furthermore, profound changes induced by antibiotics allow the invasion of the biofilm by external microorganisms [294]. As biofilms are suitable environments for conjugative transfer, we can easily imagine that antibiotics could potentiate the invasion by a potential donor harboring gene resistance and its dissemination within the biofilm acting as a synergistic factor instead of an antagonist one. Microfluidic technology represents a promising method to investigate, in real-time and without disrupting biofilm structure, the dynamic of conjugation within communities. Recent studies used microfluidic devices combined with confocal microscopy to monitor in real-time plasmid RP4 transfer in mixed P. putida and E. coli biofilms and in activated sludge [310,311]. They were able to show that the structure and composition of the biofilm could modulate gene transfer routes. Indeed, in E. coli biofilms, the explosive spread of transconjugants illustrated the significant role of plasmid transfer, while in the sludge community, vertical gene transfer is more predominant. Using these advanced techniques, it is now more needed to understand how antibiotics can influence gene dissemination within these complex structures.

Conclusion
Our current knowledge of the sequence of reactions required for plasmid conjugation is well documented, especially for model plasmids such as the F factor, but also for other plasmids like RP4, R388 or pTi. The combination of genetic and biochemistry approaches has allowed describing the function of key Tra proteins in these reactions. However, even for these systems, the mechanistic functions of the most Tra proteins remain elusive. They have mainly been described in terms of essentiality for the mating pair formation or stabilization, DNA transfer and immunity, without further understanding of their activity at the molecular scale. Furthermore, as emphasized in the review, a number of major fundamental questions remains, such as the pilus ability to conduct the DNA during distant transfer, the existence and nature of a potential signal triggered by the mating pair formation that would activate conjugation, or the role of the leading genes in the early steps of plasmid establishment, for instance. Due to its intimate connection with the dissemination of drugresistance, conjugation is back in the focus of a global research effort worldwide. Modern experimental approaches should help gaining new insights into the mechanism of conjugation at the molecular and cellular scales, but also regarding the extend of conjugation into natural bacterial communities and its impact on the dissemination of bacterial metabolic traits.