Surprisingly, the arrangement of the components in Type Ib systems is the only common aspect shared with the aforementioned systems. The interactions between their main components are different, and so may be the segregation mechanism. The centromere site localizes upstream of the par operon and consists of direct and inverted repeats. However, in plasmid pCXC100 the centromeric site contains only direct repeats (Yin et al., 2006; Huang et al., 2011). The CBPs, which also functions as repressors (Carmelo et al., 2005; Weihofen et al., 2006) are small proteins that share the arrangement into N- and C-terminal domains (Figure 2C). The N-terminal domain, which shows a highly divergent sequence, is flexible and unstructured, and includes a conserved arginine finger that has been implicated in the activation of ATP hydrolysis in the motor protein (Barilla et al., 2007). The C-terminal domain topology is β1-α1-α2 and includes a ribbon-helix-helix (RHH) DNA-binding motif (Murayama et al., 2001; Golovanov et al., 2003; Huang et al., 2011). The β1 strand from two different molecules pairs into an antiparallel β-ribbon, meaning that these CBPs are also present as dimers in solution (Barilla and Hayes, 2003; Golovanov et al., 2003).

Plasmid pSM19035 harbors a unique partitioning system: rather than being encoded in a single operon, each gene is transcribed separately from different promoters. The full centromere contains 3 separate parS sites, consisting of 9, 7, and 10 iterons that occur twice in the plasmid genome (parS1, parS1′, parS2, parS2′, parS3, parS3′, de la Hoz et al., 2000, 2004; Dmowski et al., 2006). However, parS2 appears to be the main centromeric sequence (Dmowski and Kern-Zdanowicz, 2016). Interestingly, each parS overlaps with the promoters of genes involved in plasmid copy number and maintenance: parS1 with Pδ, parS2 with Pω and parS3 with P_copS (de la Hoz et al., 2000). The CBP, ω, binds to each parS with different affinities, depending on the number of iterons (de la Hoz et al., 2004). This feature may be crucial to fine-tune repressor affinity for different promoters (Weihofen et al., 2006). The nucleoprotein complex is a left-handed protein helix that wraps the DNA (Weihofen et al., 2006) covering only the parS site (Pratto et al., 2009). Protein binding to both direct and inverted repeats involves comparable interactions, due to the pseudo-symmetry of the dimer (Weihofen et al., 2006, Figure 2C). Binding induces minor structural changes mainly affecting the loop connecting α1 and α2. In contrast to other RHH DNA-binding proteins, there is no DNA bending (Pratto et al., 2009). Because the DNA is not curved, ω first makes contact with the DNA major groove via base specific interactions with residues on the β-sheet and then the N-termini of the α2 helices clamp the phosphate backbones (Weihofen et al., 2006). Assuming nearly straight DNA, the number and orientation of repetitions will affect the distances between helices α1 of adjacent ω dimers, thereby modulating the cooperativity. The motor protein, δ, binds non-specifically to DNA but is recruited to the location of the segrosome (Pratto et al., 2009) to form a ternary complex, giving rise to intermolecular pairing of parS regions (Pratto et al., 2008, 2009). This bridging may increase the local concentration of ω, in turn increasing the ATPase activity of the motor protein and thus inducing detachment of this protein and promoting mobility (Pratto et al., 2009). This system may combines both DNA wrapping mechanisms (during segrosome formation) and DNA bridging mechanisms (when the motor protein participates in segregation).

In plasmid TP228, the centromere (parH) is continuous and consists of direct and inverted repeats separated by AT-rich regions. A DNA region between the operon genes and the centromere (O_F) comprises more repeats that play important roles in partitioning and transcription regulation (Zampini et al., 2009; Wu et al., 2011). Binding of the CBP, ParG, to parH occurs via the RHH motif, but unlike in ω, ParG is also dependent on the protein's N-terminal tail, which modulates binding affinity (Golovanov et al., 2003). Apparently, the AT-enriched spacers may increase the binding cooperativity of ParG to DNA (Wu et al., 2011). Like plasmid pSM19035, the centromere site is not curved and ParG binding does not induce DNA bending. The motor protein, ParF polymerizes into filaments and does not bind to DNA (Barilla et al., 2005; Schumacher et al., 2012). The N-terminal domain of ParG is not only important for the activation of the ATPase but also facilitates ParF filament nucleation and bundling (Barilla et al., 2007). Furthermore, in contrast to all other described systems, the ParF-ParG interaction is not dependent on the formation of the segrosome. This suggests that pSM19035 and TP228 despite sharing the same type Ib partitioning system employ distinct segregation mechanisms.

The centromeric site (parC) consists of tandem repeats localized in a single locus upstream of the operon. The arrangement can be continuous (plasmid pSK41, Schumacher et al., 2007a) or split into two regions (plasmid R1), with the par cassette promoter in the middle (Dam and Gerdes, 1994). However, the resulting segregation complexes are very similar. The CBP, ParR, contains two domains, an N-terminal domain with a RHH DNA binding motif (as seen in type Ib systems), and a C-terminal domain that is involved in the interaction with the motor protein. The domain topology of the ParR N-terminal domain is β1-α1-α2-α3-α4-α5 (Figure 2A). The β1-strands from two monomers combine in an antiparallel fashion and the α1-α2 helices come together to form an extensive dimer (Moller-Jensen et al., 2007; Schumacher et al., 2007a). The C-terminal domain includes a 3-helix cap that reinforces the tight dimerization of the N-terminal domain and an unstructured C-terminal tail with a high degree of sequence conservation (Moller-Jensen et al., 2007).

The nucleoprotein complex forms a discrete helical arrangement with a diameter of 15-nm (Moller-Jensen et al., 2007; Hoischen et al., 2008). The structure of the nucleoprotein complex (Schumacher et al., 2007a) reveals a continuous helical array in the crystal packing (Figure 2A). Each turn consists of 6 symmetrical pairs of dimers (involving the assembly of 12 ParR dimers), producing distinct negative and positive electrostatic on the inner and outer surfaces of the helix. The DNA wraps ParR by interacting with the outer, positively charged surface of the super helix, with each dimer binding one parC iteron. When the centromere is split in two, the promoter region forms a DNA loop that protrudes out of the ParR-parC ring structure (Hoischen et al., 2008; Salje and Lowe, 2008), repressing the promoter (Jensen et al., 1994; Breuner et al., 1996), and regulating transcription of the partition genes (Salje and Lowe, 2008). The DNA is bent by 46° and widened so that the major groove grows from 11 to 14Å (Schumacher et al., 2007a). The groove enlargement allows the insertion of the RHH motif, as described for other DNA-binding RHH proteins (Somers and Phillips, 1992; Raumann et al., 1994; Gomis-Ruth et al., 1998). Interestingly, the phosphate contacts cluster at the 5′ ends of the 10-bp repeats, creating the closest physical associations between ParR and the DNA. Full-length ParR from plasmid pB171 crystallized in a helical superstructure in the absence of DNA, with a diameter very similar to that measured in the nucleoprotein complex (15 vs. 18 nm) (Figure 2A). Moreover, the protein arrangement into dimers and the electrostatic distribution are also similar (Moller-Jensen et al., 2007). These observations lead to the question; which event occurs first? If ParR assembly into a super-helical structure occurs first, then the macromolecular complex may recruit parC. Otherwise, the centromere might function as a scaffold for ParR oligomerization.

For ParR, the segrosome structure positions the conserved C-tails clustered on the inside surface of the helix, where they mediate binding to the motor protein, ParM (Schumacher et al., 2007a; Salje and Lowe, 2008). The ParR tail binds to a hydrophobic pocket in ParM in an interaction resembling that described for actin polymer modulators and the barbed end of actin filaments (Gayathri et al., 2012). Furthermore, the segrosome binds only at the growing end of the polar ParM double helical filament favoring filament growth via a formin-like mechanism (Gayathri et al., 2012). Why does the ParR-ParM interaction require the clustering of so many ParR tails? It is possible that several tails bind to a single ParM molecule with distinct affinities, regulating ParM filament dynamics. Alternatively, the presence of free tails may be necessary to explore the space around the filament end and to facilitate the addition of ParM molecules to the growing filament while remaining attached at all times.

The type III system were the most recently discovered partitioning systems (Larsen et al., 2007). For TubZRC, the centromeric site (tubC) is localized upstream of the operon and contains several direct repeats in a single locus that can be split into two (pBtoxis) or three (pBsph) blocks, resembling discontinuous parC sites (Aylett and Lowe, 2012; Ge et al., 2014a). During partitioning, the CBP (TubR in this case) mediates the assembly of the segrosome nucleoprotein complex and acts as a repressor of tubRZ transcription (Tang et al., 2006; Larsen et al., 2007; Ge et al., 2014a). TubR is a small winged-helix DNA-binding protein with a high degree of structure conservation. The topology is β1-α1-α2-α3-α4-β2-β3-α5, where the α3-α4 helices form the HTH motif (α4 is the “recognition helix”) and the loop between β2-β3 forms the wing (Figure 2B, Ni et al., 2010). Interestingly, TubR forms a highly intertwined dimer involving the canonical HTH motif, resulting in an atypical protein-DNA binding (Aylett and Lowe, 2012). The N-termini of both recognition helices in a dimer protrude into the major groove of the DNA, while the acidic patch in the wing complements the DNA backbone phosphate in the minor groove. The nucleoprotein complex takes the shape of a flexible filament, with TubR wrapping helically around both sides of tubC (Aylett and Lowe, 2012, Figure 2B). The filamentous complex closes to form 18-nm wide ring-like structures (Aylett and Lowe, 2012). However, the structure of plasmid pBM400 TubR, with no DNA bound, reveals a helical arrangement, resembling the ParR super-helical complex (Figures 2A,B). It thus remains unclear whether the segrosome complex is formed by TubR wrapping of the DNA or by DNA wrapping of the TubR oligomer, which could lead to different interacting mechanism with the motor protein (TubZ).

TubR binds to TubZ C-terminal tail (Ni et al., 2010). However, the interaction is only possible following formation of the filamentous segrosome. Neither TubR alone (Oliva et al., 2012) nor TubR bound to either of the two-iteron clusters are capable of interacting with TubZ (Aylett and Lowe, 2012; Fink and Lowe, 2015). Therefore, the clustering of TubR may generate the binding site for TubZ. Differently to type II systems, the segrosome tracks the shrinking minus end of the TubZ filament, suggesting a pulling segregation mechanism (Fink and Lowe, 2015).

Type III partition systems involve a third protein with a predicted HTH DNA-binding motif and a long coiled-coil domain (TubY), located downstream of the partition operon (Oliva et al., 2012). TubY seems to be a regulator protein that modulates TubZ assembly (Oliva et al., 2012) and also acts as a transcriptional activator (Ge et al., 2014b) but the exact molecular mechanisms remain elusive.

It is still common for new partitioning systems to be discovered in plasmids, phages, and on chromosomes. Together with a growing body of molecular insights these will help to broaden our understanding of DNA trafficking during bacterial cell division and in particular how DNA is attached to the CBP during segrosome formation and then to the motor protein through the segrosome.

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The handling Editor declared a shared affiliation, though no other collaboration, with the author and states that the process nevertheless met the standards of a fair and objective review.