The very first step of replication initiation process is the recognition of specific motifs located within the origin region of DNA molecule (Figure 1) by replication initiation proteins (Figure 2, Stage I). The bacterial chromosome replication initiator DnaA protein consists of four domains, which play distinct roles (Sutton and Kaguni, 1997, Figure 3A). The best characterized DnaA is the Escherichia coli protein (EcDnaA), although structural data is limited only to domain I (resolved by NMR-analysis; Abe et al., 2007b) and IV (resolved in a nucleoprotein complex by crystallography; Fujikawa et al., 2003). Information concerning the structure of DnaA initiator is supplemented by structure of domains I and II of Mycoplasma genitalium DnaA (MgDnaA; Lowery et al., 2007), domains I and II of Helicobacter pylori DnaA (HpDnaA) in a complex with HobA protein (Natrajan et al., 2009), domains III and IV of Aquifex aeolicus DnaA (AaDnaA; Erzberger et al., 2002, 2006), domain III of Thermatoga maritima DnaA (TmDnaA; Ozaki et al., 2008), and domain IV of Mycobacterium tuberculosis (MtDnaA; Tsodikov and Biswas, 2011). Domain I of EcDnaA, located at the N-terminus of the protein, was shown to be involved in oligomerization of DnaA (Weigel et al., 1999; Simmons et al., 2003; Abe et al., 2007a), helicase loading (Sutton et al., 1998; Seitz et al., 2000), and interaction with DiaA (Keyamura et al., 2007), HU (Chodavarapu et al., 2008a), Dps (Chodavarapu et al., 2008b), and ribosomal protein L2 (Chodavarapu et al., 2011). The interaction with DiaA homologe, HobA protein, was shown for domains I and II of HpDnaA (Natrajan et al., 2007, 2009; Zawilak-Pawlik et al., 2007). In Bacillus subtilis, domain I of DnaA (BsDnaA) interacts with SirA, the sporulation-related protein (Rahn-Lee et al., 2011). However, the binding partner proteins can vary among DnaA orthologs, and replication initiator from one bacterium can interact with different partners compared to other orthologs, e.g., interaction of Thermoanerobacter tengcongensis DnaA with NusG protein, is not observed for BsDnaA (Liu et al., 2008). The second domain, forming a flexible linker, although it is not essential (Messer et al., 1999; Nozaki and Ogawa, 2008), was proposed to be involved in optimal helicase DnaB recruitment (Molt et al., 2009). The domain II, links domain I with domain III, which contains a common core structure of AAA+ proteins family members (Neuwald et al., 1999). Recent data showed that residues within this domain are engaged in interaction of DnaA (TmDnaA, EcDnaA, AaDnaA) with single-stranded DNA (ssDNA; Ozaki et al., 2008; Duderstadt et al., 2011). At the C-terminus of DnaA, domain IV (DNA Binding Domain, DBD) can be distinguished, which is responsible, via a helix-turn-helix motif (HTH), for interaction with double-stranded DNA (dsDNA) containing specific motifs named DnaA-boxes (Roth and Messer, 1995; Fujikawa et al., 2003). Interaction with these sequences is the very first step of the replication initiation process.

In bacterial chromosome origin, regions that are composed of a variable number of DnaA-boxes, can be identified (Ozaki and Katayama, 2009; Rajewska et al., 2012; Wolanski et al., 2014; Leonard and Grimwade, 2015). In the origin of E. coli chromosome (oriC), five 9-bp in length DnaA-boxes (R1–R5) were originally identified (Fuller et al., 1984; Matsui et al., 1985); in contrast, the origin of Caulobacter crescentus chromosome (Cori) possesses only two DnaA-boxes (named G-boxes; Shaheen et al., 2009). The studies with the use of in vivo and in vitro dimethylsulphate (DMS) footprinting as well as DNase I footprinting method showed that other, non–R DnaA binding sites are present in oriC, i.e., I (Grimwade et al., 2000; McGarry et al., 2004), C (Rozgaja et al., 2011), and τ sites (Kawakami et al., 2005). Such non-canonical sequences recognized by bacterial initiator were also found in oriC of C. crescentus (termed W-boxes; Taylor et al., 2011). The affinity of DnaA binding to R-boxes and non-R DnaA binding sites is different. Interestingly, binding of inititor to the DnaA-boxes in Cori of C. crescentus, both G-boxes and W-boxes, is lower compared to DnaA binding to the R-boxes in oriC of E. coli (Taylor et al., 2011), which might be characteristic for bacteria with a complex regulation of development. The DnaA binding sites, bound by initiator with affinity comparable only to interaction between DnaA and weak DnaA-boxes in E. coli oriC, were found in the origin of H. pylori (Zawilak-Pawlik et al., 2007; Charbon and Løbner-Olesen, 2011). In E. coli oriC three (named R1, R2, and R4) out of five DnaA-boxes are the widely separated, high affinity DnaA-boxes. They were found to be almost constantly bound by EcDnaA protein (Samitt et al., 1989; Nievera et al., 2006). The occupancy of only these three sites is insufficient for spontaneous origin opening and it was proposed that interaction of EcDnaA protein at high affinity binding sites may regulate conformation of the origin DNA (Kaur et al., 2014). Between the peripheral R1 and R4 sites, there are two arrays of low affinity binding sites, τ1 R5 τ2 I1 I2 and C3 C2 I3 C1, separated by one of high affinity—R2 (Rozgaja et al., 2011). EcDnaA molecules bound to the high affinity DnaA-boxes, termed bacterial Origin Recognition Complex (bORC), act as anchors and are required to assist in occupying weak sites by the EcDnaA protomers (Rozgaja et al., 2011; Kaur et al., 2014), and formation of replication-active pre-replication complex (pre-RC; Figure 2, Stage II). The binding affinity to particular sequences and replication activity of EcDnaA protein depend on nucleotide-bound state of protein. Although ADP-EcDnaA binds the high affinity DnaA-boxes and also R5 and C1 low affinity ones, the ATP-EcDnaA form is thought to be the replication-active one (Sekimizu et al., 1987; Leonard and Grimwade, 2011). ATP-EcDnaA form of initiator binds efficiently both high and low affinity binding sites (McGarry et al., 2004; Kawakami et al., 2005). Based on molecular docking, binding of ATP, instead of ADP, is presumed to cause changes in the EcDnaA protein conformation, thus leading to the formation of large oligomeric complex within the origin region (Saxena et al., 2015). The crystallographic data, when nonhydrolyzable ATP analog AMP-PCP was used, showed the formation of open-ended, right-handed helical filament of AaDnaA (Erzberger et al., 2006). Based on biochemical and genetic approaches it was found that there is an interaction between domain III (AAA + domain) of one DnaA (EcDnaA or AaDnaA) molecule and domain IV (DBD domain) of partner subunit (Duderstadt et al., 2010). It was proposed that during pORC and pre-RC complexes formation of the DBD domain is extended and the HTH motif is exposed, which results in the efficient binding of high and low affinity binding sites (Duderstadt et al., 2010). Occupation of the EcDnaA binding sites was shown to be sequential and polarized and DnaA protomers are released preferentially from the peripheral high affinity R1 and R4 boxes, through arrays of low affinity binding sites to the middle high affinity one—R2 (Rozgaja et al., 2011). The formation of DnaA oligomer within the oriC results in DNA destabilization in the DUE region (Speck and Messer, 2001; McGarry et al., 2004; Leonard and Grimwade, 2005, 2011; Duderstadt et al., 2010). Although two arrays of low affinity binding sites separated by high affinity sequences are occupied by EcDnaA protomers for efficient double-stranded DNA opening, binding of EcDnaA to a part of origin (containing only R1 high affinity box and τ1 R5 τ2 I1 I2 low affinity binding sites array) was shown to be active in DUE unwinding (Ozaki and Katayama, 2012). It was proposed that distinct DnaA multimers are formed on the left half (containing binding sites from R1 to I2) and the right half (containing binding sites from R2 to R4) of oriC (Ozaki and Katayama, 2012; Ozaki et al., 2012a).

The DUE melting is the consequence of DnaA binding to arrays of DnaA-boxes (Figures 1A, 2, Stage II). The location of particular binding sites suggests that DnaA, bound to sequences of the high affinity DnaA-boxes (R1, R2, R4), could cause the bending of DNA molecule via interaction through domain I of already bound three protomers (Rozgaja et al., 2011; Kaur et al., 2014; Leonard and Grimwade, 2015). The model of constrained loop formed by EcDnaA bound to the high affinity binding sites was proposed (Kaur et al., 2014). The bending of oriC containing DNA molecule is supported by accessory histone-like proteins HU and integration host factor (IHF). A binding site for IHF was found within the oriC region (Polaczek, 1990) and it was shown that IHF can enhance the unwinding of DNA by DnaA (Hwang and Kornberg, 1992; Ryan et al., 2002). It was demonstrated that HU has the same effect on DUE destabilization (Hwang and Kornberg, 1992), although its mechanism of action is different (Ryan et al., 2002). Data obtained with ELISA (Enzyme Linked Immunosorbent Assay) showed that HU interacts with domain I of EcDnaA, which was proposed as an interaction which stabilizes the DnaA oligomer (Chodavarapu et al., 2008a). The Fis protein, identified originally as factor for inversion stimulation in site-specific DNA recombination, was also shown to have an influence on DNA unwinding (Wold et al., 1996). Specific binding sites for Fis were identified in oriC (Gille et al., 1991). Although Fis, in contrast to IHF, negatively regulates DNA replication initiation, when the origin lacks some DnaA binding sites resulting in altered non-functional conformation of origin, both Fis and IHF can work together to correct these alterations (Kaur et al., 2014). This joint action is achieved by inducing bends in oriC and establishing functional origin conformation (Kaur et al., 2014).

The formation of DnaA oligomer with synergistic action of architectural proteins can introduce torsional strain into DUE, facilitating the melting of the double-stranded DNA helix. The binding of DnaA to DUE region was also thought to introduce DNA melting, and ATP-DnaA-boxes were distinguished within the oriC DUE sequence (Speck and Messer, 2001). Recent studies showed direct binding of EcDnaA and AaDnaA protein to formed single-stranded DNA within the DUE (Ozaki et al., 2008; Duderstadt et al., 2010, 2011; Cheng et al., 2015). Studies with DnaA mutants (Ozaki et al., 2008; Duderstadt et al., 2010), as well as crystallography (Duderstadt et al., 2011), showed that this interaction occurs through residues located within the AAA+ domain III of bacterial initiator. The AaDnaA protomers form a helical filament on ssDNA (Duderstadt et al., 2011), however, it differs from the filament formed on the dsDNA (Erzberger et al., 2006; Duderstadt et al., 2010). It was proposed that protomers in this oligomer are more compact when compared to the extended DnaA molecules in dsDNA-DnaA complex (Duderstadt et al., 2010, 2011). The binding of ssDNA concerns just one T-rich strand of DUE and depends on sequence of 13-nucleotide sequences, which can be distinguished within the DUE. In oriC three 13-mers are present (Bramhill and Kornberg, 1988a) and the binding of EcDnaA occurs at least two 13-mers. EcDnaA does not form a complex with ssDNA containing just one 13-mer (Ozaki et al., 2008). Formation of this nucleoprotein complex is achieved only by ATP-DnaA protein (Ozaki et al., 2008) and one AaDnaA protomer binds three nucleotides of ssDNA (Duderstadt et al., 2011; Cheng et al., 2015). Studies with the use of single-molecule fluorescence assays showed that the formation of this nucleoprotein complex is highly dynamic and that AaDnaA molecules assemble on ssDNA in the 3′ to 5′ direction (Cheng et al., 2015). The presence of dsDNA region containing DnaA-boxes, adjacent to ssDNA DUE, stabilizes the DnaA (EcDnaA and AaDnaA) filament on ssDNA (Ozaki and Katayama, 2012; Cheng et al., 2015). Recently published data revealed presence of a new origin element, termed DnaA-trio, composed of repeated trinucleotide motif that stabilizes DnaA filaments on the ssDNA (Richardson et al., 2016). What is important, binding single strand of DUE region is required for origin activity (Ozaki et al., 2008, 2012a,b; Duderstadt et al., 2011).

Similarly as during bacterial chromosome replication, the first step in open complex formation in many theta-replicating plasmids, especially in iteron-containing plasmids, is the binding of plasmid replication initiator, Rep protein, to specific sequences within origin region (Figure 2, Stage I). Rep proteins are structurally different from bacterial DnaA protein and consist of winged-helix (WH) domains (Figure 3B, Komori et al., 1999; Díaz-López et al., 2003; Sharma et al., 2004; Swan et al., 2006; Nakamura et al., 2007a,b; Pierechod et al., 2009). The crystal structures of nucleoprotein complexes of π protein from plasmid R6K (Swan et al., 2006), RepE protein from plasmid F (Komori et al., 1999; Nakamura et al., 2007b), and a DNA binding domain of Rep protein from ColE2–P9 plasmid (Itou et al., 2015) as well as N-terminal domain of RepA protein from plasmid pPS10 (Giraldo et al., 2003) were obtained. Furthermore, homological models for plasmid Rep proteins: RepA from P1 (Sharma et al., 2004), RepA from pSC101 (Sharma et al., 2004), and TrfA from RK2 (Pierechod et al., 2009) were shown. Plasmid Reps are composed of two WH domains, of which one is responsible for oligomerization and the role of a second one is the protein's interaction with DNA (Giraldo et al., 1998; Nakamura et al., 2004; Pierechod et al., 2009). Plasmid replication initiators are present as dimers in solution, however, an exception is known i.e., RepE protein from pAMβ1 plasmid is present as a monomer (Le Chatelier et al., 2001). Although the Rep dimers interact with DNA (Filutowicz et al., 1985; Ingmer et al., 1995; Komori et al., 1999), they are replication-active in the monomeric form (Kawasaki et al., 1990; Wickner et al., 1992; Sozhamannan and Chattoraj, 1993; Konieczny and Helinski, 1997). Conformational activation of plasmid replication initiators is carried out by chaperon proteins (Kawasaki et al., 1990; Wickner et al., 1992, 1994; Sozhamannan and Chattoraj, 1993; Konieczny and Helinski, 1997). In contrast to bacterial replication initiator DnaA, the domain responsible for binding of nucleotide was not identified in Reps' structures. There is also no evidence showing if Rep proteins can form helical filaments on DNA similar to that formed by the AaDnaA protein. For some Reps, e.g., TrfA protein from RK2 plasmid, two forms of protein, different in length, occur: the shorter 33 kDa (TrfA-33) and longer 44 kDa (TrfA-44). There are different requirements for each particular form depending on the host bacterium. In E. coli both forms of TrfA can initiate the plasmid replication, whereas in Pseudomonas aeruginosa only the longer form is active (Caspi et al., 2001; Jiang et al., 2003; Konieczny, 2003; Yano et al., 2013, 2016).

During plasmid replication initiation, monomers of Reps bind to specific repeated sequences, named iterons, present within origin region (Figures 1B, 2, Stage I). The number of iterons varies among plasmid origins, from two iterons in plasmids ColE2 and ColE3 (Yasueda et al., 1989), three iterons in pSC101 (Churchward et al., 1984), and some plasmids from IncQ incompatibility group (Loftie-Eaton and Rawlings, 2012), four iterons in origin of plasmid F and pPS10, up to five (origin of plasmids RK2 and P1) or even seven such sequences in oriƔ of plasmid R6K (Rajewska et al., 2012). Iterons are short sequences, in length ranging from 17-bp in RK2 plasmid (Stalker et al., 1981), 19-bp in plasmids F (Murotsu et al., 1981), and P1 (Abeles et al., 1984), to 22-bp in R6K (Filutowicz et al., 1987), and pPS10 (Nieto et al., 1992). But in some plasmids the iteron sequences which are present in one origin can differ in length and apart from short sequences, significantly longer iterons [up to even 76-bp in plasmid R478 from IncHI2 incompatibility group (Page et al., 2001)] can be present. The binding of Rep proteins to iterons is sequence-specific and mutations in these motifs disrupt binding of plasmid initiation protein. Changes in a sequence of iterons abolished binding of π protein within the oriƔ of plasmid R6K and thus replication activity in vivo (McEachern et al., 1985). Negative effects on replication was also observed for mutants in a sequence of P1 plasmid iterons (Brendler et al., 1997). The sequences separating particular iterons are also important for Rep nucleoprotein complexes formation and proper replication activity of origin. It was shown in case of the RK2 plasmid that in vitro the TrfA protein has a high preference for binding to DNA containing at least two out of five binding sites, when compared to the formation of nucleoprotein complex with DNA containing just one iteron (Perri et al., 1991). The requirement for the presence of more than just one iteron sequence for TrfA binding was also shown in vivo (Perri and Helinski, 1993). Rep proteins bind to iterons in a cooperative manner (Perri and Helinski, 1993; Xia et al., 1993; Bowers et al., 2007) and the cooperativity of binding depends on the spatial location of iterons, since separation of two iterons by a half helical turn abolished cooperativity (Bowers et al., 2007). These results suggest the possibility of formation of higher order nucleoprotein structure on plasmid iterons bound by Reps. It was shown that WH domains of Reps contact three nucleotides in DNA. In π protein from R6K plasmid, WH1 domain contacts wGwnCnT motif, and WH2 domian contacts GAG sequence (Swan et al., 2006). Similarly, the WH2 domain of RepE monomer also contacts three nucleotides of top (GTG sequence) and three nucleotide of bottom strand (GtCA sequence) of double-stranded molecule containing iteron sequence (Nakamura et al., 2007b). However, unlike for the bacterial DnaA protein, to date there are no evidence showing that strong and weak binding sites for Reps are present within plasmid origins. There were just predictions of potential binding sites, other than iterons, for π protein in R6K plasmid and suggestions on potential role of such sites (Rakowski and Filutowicz, 2013). Certainly like DnaA, Rep proteins can bind within single-stranded region of melted DUE, and this binding is sequence-specific, since binding concerns a particular strand. Nucleoprotein complexes formation with the ssDNA DUE was detected for TrfA (bound with A-rich strand) and RepE (bound with T-rich strand) proteins (Wegrzyn et al., 2014). Within the DUE of plasmid origins, repeated sequence, similar to 13-mers distinguishable in oriC, can be found. There are four 13-nucleotide sequences in plasmid RK2 DUE region (Doran et al., 1998) and all of them are required for TrfA-ssDNA DUE complex formation. Lack of even one 13-mer hinders plasmid replication (Wegrzyn et al., 2014). Also, even a point mutation within this region affects plasmid replication since the lack of DUE melting was observed for some of the changed sequences (Kowalczyk et al., 2005; Rajewska et al., 2008).

The Rep protein encoded by plasmids, can be accompanied by host DnaA initiator during open complex formation and DUE melting within a plasmid origin (Figure 2, Stage II). DnaA binding sites have been found in replication origin of many plasmids including plasmids P1 (Abeles et al., 1984, 1990; Abeles, 1986), F (Kline et al., 1986; Murakami et al., 1987; Kawasaki et al., 1996), RK2 (Doran et al., 1998; Caspi et al., 2000), pSC101 (Sutton and Kaguni, 1995). The number of DnaA-box sequences differs among plasmid origins, the position and orientation of these binding sites are as important as position and orientation of the iterons (Doran et al., 1998, 1999). The inversion of one out of four DnaA boxes in origin of RK2 plasmid abolished plasmid DNA replication, despite the fact that three remaining DnaA boxes were bound by the host initiator (Doran et al., 1999). Although the DnaA protein is not required for replication initiation for some plasmids, e.g., R1, binding of DnaA increased the plasmid replication efficiency (Bernander et al., 1991, 1992) and mutations within a binding site for DnaA decreased the R1 plasmid replication (Ortega-Jiménez et al., 1992). In bacteria, ATP-DnaA form is essential for chromosomal DNA replication (Sekimizu et al., 1987; Leonard and Grimwade, 2005, 2011). Interestingly, studies with ATP-binding mutant of DnaA, which was inactive in oriC replication, showed that bacterial initiator lacking an ability to bind a nucleotide was effective in open complex formation within plasmid R6K oriƔ (Lu et al., 1998). Also in the presence of ATPS, a non-hydrolyzable analog of ATP, the pattern of bands in KMnO₄ footprinting assay with DnaA and TrfA proteins and plasmid RK2 DNA showed no significant differences, when compared to opening reaction containing ATP (Konieczny et al., 1997). Thus, the DnaA is suspected to play a different role in plasmid replication initiation, compared to its role in chromosome replication. A direct interaction between plasmid and host replication initiators was shown (Lu et al., 1998; Maestro et al., 2003) and the interaction was detected in the N-terminus of π (between 1 and 116 aa) protein of R6K plasmid (Lu et al., 1998) and RepA protein of pSC101 (Sharma et al., 2001) and domain I and IV of host initiator (Sharma et al., 2001). The mutations in RepA protein from pPS10 plasmid were introduced, which enhanced the interaction of RepA with DnaA protein and resulted in changes in host range of pPS10 plasmid (Maestro et al., 2003).

Similarly to bacterial chromosome replication initiation, the binding of DnaA protein to DnaA-boxes within plasmid origins can be enhanced by the presence of architectural proteins IHF, and HU (Shah et al., 1995; Fekete et al., 2006). The binding of IHF to its binding site in oriƔ region significantly enhanced binding of bacterial DnaA to R6K plasmid origin (Lu et al., 1998). In pSC101 plasmid binding IHF to its cognate binding site is required for plasmid replication initiation and mutations within this sequence disrupts plasmid replication (Stenzel et al., 1987). For plasmid P1 the binding of IHF to its site, located downstream of one out of two arrays of DnaA-boxes (the second array is located upstream of DUE) is required only when the nearby DUE array of DnaA-boxes is not active and the second DnaA-boxes array serves as a secondary origin compensating the function of the first one (Fekete et al., 2006). The P1-mini derivative was just slightly unstable in IHF E. coli mutant (Ogura et al., 1990). The mutations in gene for IHF protein did not affect plasmids F (Ogura et al., 1990) and RK2 (Shah et al., 1995) replication. In contrast, the lack of HU protein in vitro results in significant decrease in mini-F plasmid DNA synthesis (Zzaman et al., 2004) and in vivo KMnO₄ reactivity of P1 plasmid origin (Park et al., 1998) as well as abolishment of plasmid F replication in vivo (Ogura et al., 1990). During plasmid RK2 replication initiation, HU could functionally replace DnaA protein, although it could not enhance DUE melting as efficiently as DnaA (Konieczny et al., 1997). It was proposed that one of the DnaA functions could be the stabilization of origin melting induced by Rep protein. The other DnaA role during replication initiation is its function in helicase loading. Interestingly, for some plasmids, e.g., RK2, DnaA assists Rep during plasmid replication initiation only in particular hosts, while in others DnaA is dispensable [DnaA P. aeruginosa is dispensable for RK2 plasmid replication initiation, but required in E. coli (Caspi et al., 2001; Konieczny, 2003)].

Following the primer synthesis, clamp loader complex loads the ring-shaped β-clamp (discussed below in details), that encircles dsDNA, tethers DNA polymerase, and slides along dsDNA, thus significantly increasing speed (up to 100-fold), and processivity (up to 5000-fold) of DNA replication (Kelch et al., 2012). The E. coli clamp loader is composed of γ, τ, δ, δ', χ, and ψ subunits, albeit only γ, δ, δ' are crucial for β-clamp loading (reviewed in details in Kelch, 2016). The γ subunit is a truncated version of τ subunit, encoded by dnaX gene, and arises from translational frameshift (Flower and McHenry, 1990). Both γ and τ subunits have AAA⁺ domain, however, γ subunit lacks τ subunit domain responsible for DnaB helicase and Pol III core binding (Tsuchihashi and Kornberg, 1989; O'Donnell and Studwell, 1990; Flowers et al., 2003). Before clamp loader binds DNA, it adopts appropriate, ATP-driven conformational state that increases its affinity for the β-clamp (Podobnik et al., 2003). It is under debate whether the ring structure of β-clamp is actively opened or captured in open conformation. The T4 bacteriophage trimeric clamp is the least stable sliding clamp and it was found to dissociate from DNA by monomerization, thus no force in opening of the ring is required (Soumillion et al., 1998). The dimeric clamps (bacterial, e.g., E. coli) are regarded as stable, hence more active ring-opening mechanism is expected to be in demand. On the basis of a crystal structure of single subunit of E. coli clamp loader (namely δ subunit) in complex with β-clamp, it was proposed that δ subunit is a molecular wrench, that induces rearrangements of β-clamp at dimerization interface, albeit without ATP hydrolysis (Jeruzalmi et al., 2001). With the use of real-time fluorescence-based clamp binding and opening assays, it was shown that clamp loader binds closed β-clamp in solution, prior to β-clamp opening (Paschall et al., 2011). Yet, the deuterium exchange coupled to Mass Spectrometry experiments revealed that most sliding clamps are dynamic at their monomers' interfaces (Fang et al., 2011, 2014). Therefore, it is also probable that β-clamp is trapped in an open conformation by clamp loader.

The crystal structure of clamp loader complex was solved from T4 bacteriophage (Kelch et al., 2011), E. coli (Simonetta et al., 2009), and its eukaryotic homolog, Replication Factor C (RFC), from Saccharomyces cerevisiae (Bowman et al., 2004). Each of the clamp loader complex reveals pentameric structure. Since AAA + ATPases usually adopt circular hexamers, it was proposed that sixth subunit was lost during the evolution (Indiani and O'Donnell, 2006). Indeed, the gap between the first and fifth clamp loader subunits is favorable, because it provides the mechanism of specific accommodation of the primer-template junction structure (Kelch, 2016). It was suggested that clamp loader recognizes minor groove and thus it binds at the 3′ primer-template junction specifically. However, the crystal structure of the clamp loader:DNA complex revealed that clamp loader contacts template DNA exclusively (Bowman et al., 2005; Simonetta et al., 2009). Despite the fact that the DNA synthesis may be initiated only from 3′ OH primer end, the clamp loader can assemble in vitro at either 3′ or 5′ primer terminus forming a stable complex (Park and O'Donnell, 2009). While clamp loader binds only DNA template, β-clamp interacts with both RNA primer, and DNA template within the RNA-DNA hybrid and it was shown that the β-clamp distinguishes between the 5′ and 3′ primer end (Park and O'Donnell, 2009). Consistently, it was demonstrated that SSB hampers clamp loading on the 5′end of primer (Hayner et al., 2014). The ATPase activity of the clamp loader is lower when it is assembled at the 5′ terminus, comparing to the ATPase activity of clamp loader located at 3′ terminus (Park and O'Donnell, 2009). ATP hydrolysis triggers β-clamp closing on DNA and the release of clamp loader from β-clamp:DNA nucleoprotein complex (Pietroni and von Hippel, 2008). Thereby, the 3′ primed end loading preference also arises from the higher rate of clamp closure and clamp loader dissociation (Park and O'Donnell, 2009). The β-clamp must be closed in the ATP hydrolysis-dependent manner, to release clamp loader (Hayner et al., 2014). Clamp loader must free the β-clamp to allow the Pol III core to bind, since they accommodate the same binding site within the β-clamp, namely the hydrophobic cleft.

β-clamp crystal structures were obtained from various organisms i.e., E. coli (Oakley et al., 2003; Burnouf et al., 2004), P. aeruginosa (Wolff et al., 2014), Streptococcus pyogenes (Argiriadi et al., 2006), M. tuberculosis (Gui et al., 2011; Kukshal et al., 2012; Wolff et al., 2014), B. subtillis (Wolff et al., 2014), T. maritima (structure 1VPK), Eubacterium rectale (structure 3T0P), Streptococcus pneumoniae (Argiriadi et al., 2006). The crystal structures of β-clamp homologs—Proliferating Cell Nuclear Antigen (PCNA)—are also available from Eukaryotes and Archea (to name a few: Homo sapiens (Punchihewa et al., 2012), S. cerevisiae (Krishna et al., 1994), Sulfolobus solfaraticus (Williams et al., 2006). All of them adopt ring shaped homodimer (e.g., E. coli) or homotrimer (human PCNA, Pyrococcus furiosus PCNA), albeit the exception is PCNA of Archea, S. solfataricus, which exists as a heterotrimer (Dionne et al., 2003). β-clamp monomers bind in a head to tail manner (Kelman and O'Donnell, 1995). The β-clamp and PCNA structure is conserved among all kingdoms of life, in contrast to amino acid sequence (Jeruzalmi et al., 2001). However, the amino acid sequence of region termed hydrophobic cleft was found to be highly conserved (Jeruzalmi et al., 2001). The hydrophobic cleft is a site for interaction with β-clamp binding partners (Jeruzalmi et al., 2001). β-clamp forms a protein interaction hub and serves as a platform for multiple protein interactions crucial in various cellular processes, i.e., DNA elongation in every living organism (Hedglin et al., 2013), regulation of DNA replication in E. coli, B. subtilis, C. crescentus (Katayama et al., 2010), DNA repair in E. coli (Rangarajan et al., 1999), toxin-mediated replication fork collapse in C. crescentus (Aakre et al., 2013). All described β-clamp interaction partner proteins share similar motif, the Clamp Binding Motif (CBM; Dalrymple et al., 2001).

Interestingly, clamp binding motif was also identified in plasmid replication initiators, including RK2 plasmid initiator-TrfA (Kongsuwan et al., 2006; Dalrymple et al., 2007). It was shown that TrfA protein lacking the leucine 137 and phenyloalanine 138 within the clamp binding motif is unable to bind β-clamp (Kongsuwan et al., 2006). The TrfA ΔLF mutant facilitated the determination of biological relevance of this interaction. The complex of TrfA and β-clamp was found to be the key feature for replisome assembly and thereby for oriV-dependent DNA replication of both supercoiled dsDNA plasmid and ssDNA plasmid in vitro, albeit the clamp loader complex is still crucial (Wawrzycka et al., 2015). Hence, the question arises—how do the Rep and clamp loader cooperate to load the β-clamp at plasmid origin? Three hypothetical models to explain the mechanism of Rep-mediated β-clamp loading could be considered (Figure 4). In the “β-clamp hand-off model” TrfA binds to the bottom strand of ssDNA close to the 3′ end of synthesized primer, recruits β-clamp, and hands it off to the clamp loader complex (Figure 4A). Then, the δ subunit of clamp loader opens the β-clamp and clamp loader positions it onto primer-template junction, as it is thought to occur during replisome assembly at E. coli oriC. This model is consistent with the results of the in vitro DNA replication experiments performed with the use of ssDNA, containing sequence of RK2 plasmid oriV (Wawrzycka et al., 2015). It was demonstrated that TrfA interacts with specific strand of ssDNA of DUE, i.e., the bottom strand, which serves as the site for replisome assembly (Wegrzyn et al., 2014; Wawrzycka et al., 2015). It can be further speculated that TrfA may assist the clamp loader in recognition of the 3′ end of primer-template junction within the oriV. Another possible role of TrfA is illustrated in Figure 4B (second model, “β-clamp:clamp loader recruitment model”). Once TrfA is bound to bottom single strand of DUE, it recruits the β-clamp, which is in complex with clamp loader, to the RK2 plasmid origin. Thus, the local concentration of β-clamp:clamp loader complex increases, the clamp loader can assemble β-clamp onto the 3′ end of a primer within the plasmid origin. Because TrfA-β-clamp interaction was shown in the absence of DNA [using both ELISA and SPR (Surface Plasmon Resonance) technique (Kongsuwan et al., 2006; Wawrzycka et al., 2015)], the third model may also be justified (Figure 4C, “β-clamp directed to oriV model”). In the third model the TrfA that is not bound to DNA forms complex with β-clamp associated with the clamp loader, then directs it to the plasmid origin, oriV. Next, the clamp loader:β-clamp:TrfA complex binds to the bottom strand of DUE via TrfA. TrfA passes the β-clamp bound to clamp loader on the primer-template junction. Although ATP binding to clamp loader (namely γ and τ subunit) is required for β-clamp opening, it cannot be excluded that TrfA—whose ATPase activity has not been revealed—substitutes the clamp loader's function at this stage. TrfA may capture β-clamp in open conformation and load it onto primed DNA. Since ATP hydrolysis is required for β-clamp closing (Trakselis et al., 2001), here may participate the clamp loader.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.