How Xer-exploiting mobile elements overcome cellular control

Most strains of Neisseria gonorrheae (Ng), the causative agent of the sexually transmitted disease gonorrheae, and a few strains of Neisseria meningitidis (Nm), which is responsible for a large number of meningitides, harbor a 57-kb horizontally acquired genetic element, the gonococcal genomic island (GGI) (1–3). Certain versions of the GGI are associated with disseminated gonococcal infection (1, 4). In addition, the GGI encodes numerous homologs of type IV secretion system genes, which are necessary for DNA secretion and facilitate natural transformation of the Neisseria (1, 2, 4). GGI are found integrated at the chromosomal dimer resolution site of their host chromosome, dif, and are flanked by a partial repeat of it, dif_GGI (Fig. 1A) (1, 5). The dif site is the target of two highly conserved chromosomally encoded tyrosine recombinases, XerC and XerD, which normally serve to resolve dimers of circular chromosomes through the addition of a crossover between directly repeated dif sites (6). This reaction raises questions on how GGI could be stably maintained (5). The results presented by Fournes et al. (7) in PNAS shed a new light on this apparent paradox.

Fig. 1.

(A) Schematic of XerCD-mediated excision of the GGI. Black double lines, N. gonorrheae chromosomal DNA; red double lines, GGI DNA; black triangles, dif_Ng and red triangle, dif_GGI. (B) Sequence alignment of the Xer recombination site of mobile elements and their cognate partner chromosomal dif site; attP_CTX, CTXϕ attachment site; attP_VGJ, VGJϕ attachment site; cer, core dimer resolution site of plasmid ColE1; dif_Ec, E. coli dif; dif1_Vc, V. cholerae chr1 dif; and psi, core dimer resolution site of plasmid pSC101. Apart from attP_CTX, which is the stem of a forked hairpin from the single-stranded form of the genome of CTXϕ, a single of the two DNA strands is represented in the 5′ to 3′ orientation from left to right. Bases of cer and psi that differ from dif_Ec and bases from attP_CTX and attP_VGJ that differ from dif1_Vc are indicated in red. Plus (+) and minus (–) signs indicate whether the sites can engage in recombination pathways initiated by XerC or XerD strand exchanges; +^K denotes FtsK-dependent recombination pathways. (C) Schematic of Xer recombination. XerD and XerC are represented in magenta and green, respectively. Following the Cre paradigm, the active pair of recombinases are drawn with their extreme C-terminal domains contacting the partner recombinases in cis. Blue circles represent the hexamer of FtsK. (D) Sequence alignment of dif_Ng, the three different types of dif_GGI, dif1_Vc, and attP_TLC. Bases of the dif-like site of mobile elements that differ from their cognate dif partner are highlighted in color, with blue highlighting those that are identical in dif_GGI1 and dif_GGI2 and in attP_GGI3 and attP_TLC.

The Xer machinery is highly conserved in bacteria. The dif sites consist of 11-bp XerC- and XerD-binding motifs, separated by an overlap region at the border of which recombination occurs (Fig. 1B). Recombination is under the control of a hexameric DNA pump, FtsK (Fig. 1C) (8). FtsK is a powerful translocase (9) and strips DNA from most proteins (10). However, a direct interaction between its extreme C-terminal domain, FtsKγ, and the Xer recombinases stops it (Fig. 1C) (11, 12) and activates the exchange of a pair of strands by XerD catalysis when in the presence of a synaptic complex (Fig. 1C) (8, 11, 13). The exchange of a second pair of strands by XerC catalysis converts the resulting Holliday junction into product (Fig. 1C) (8, 13). FtsK belongs to the cell division machinery. It assembles at midcell when most of the chromosomal DNA has been replicated and segregated, which restricts recombination at dif to the time of cell division (14, 15) and to the chromosome replication terminus region (16, 17).

Numerous mobile elements have been shown to exploit Xer recombination. Plasmids use it for the resolution of multimers, the formation of which compromises vertical transmission from mother to daughters by reducing the number of independently segregating plasmid units (18). Integrating mobile element exploiting Xer (IMEX) use it to insert into the dif site of one of the chromosomes of their host (19). In both cases, the FtsK control imposed on Xer recombination must be overcome, because the replication/segregation cycle of plasmids and the integration/excision cycle of IMEX should be independent from the cell cycle. Moreover, Xer recombination leads to the formation of plasmid multimers when they harbor a dif site (17, 20) and to the excision of the intervening DNA between directly repeated dif sites (17, 21). Correspondingly, the central region of plasmid sites seems to prevent FtsK-dependent XerD catalysis (Fig. 1B) (22), and the central region of the attachment sites of most IMEX lacks the necessary homology to stabilize XerD-mediated strand exchanges with dif (Fig. 1B) (23, 24). This is not the case for the central region of the different alleles of dif_GGI (Fig. 1D). The problem was most striking for the most common of these alleles, dif_GGI1, which differs from the neisserial dif by only 4 bp (Fig. 1D).

In PNAS, Fournes et al. (7) observe that the Ng Xer recombinases efficiently bound to dif_GGI1, synapsed it with dif_Ng, and catalyzed complete recombination reactions between the two sites when activated by Ng FtsKγ. However, they noticed that recombination was reduced in the presence of the FtsK translocation module. The authors smartly hypothesize that FtsK translocation inhibited recombination by stripping Ng XerD from dif_GGI1, which they successfully verified in vitro.

It was previously suggested that GGI initially harbored true neisserial dif sites and that their stabilization resulted from mutations that occurred after their integration (5). Many different types of mutations, including mutations in the central region of the dif sites and mutations abolishing the binding of the recombinases to them, could impede Xer recombination. Why, then, should dif_GGI1 harbor mutations that blocked FtsK-dependent recombination without affecting XerC and XerD binding and synapse formation? One of the dif_GGI alleles found in Nm strains, attP_GGI2, harbors two out of four of the bases that differentiate dif_GG1 from dif_Ng, which suggests that these changes were not randomly picked up (Fig. 1D, blue bases of dif_GGI1 and dif_GGI2). Indeed, it is striking to note that dif_GGI2 is fully palindromic and carries two XerC-binding arms (Fig. 1D). In contrast, 8 out 11 of the bases of the XerD-binding arm of dif_GGI3 differentiate it from the XerD arm of dif sites (Fig. 1D). The attachment site of a V. cholerae IMEX, the toxin-linked cryptic phage (TLCϕ) harbors four of these bases (Fig. 1D, blue bases of dif_GGI3 and attP_TLC) (25). We previously demonstrated that XerD poorly bound to attP_TLC, which is sufficient to prevent XerD-mediated FtsK-dependent recombination (25). Thus, it is tempting to propose that GGI are IMEX and dif_GGI sites were selected not only to escape but also to overcome the normal cellular control imposed on Xer recombination by FtsK. GGI harboring dif_GGI3 probably belong to the TLCϕ class of IMEX, which integrate into and excise from the genome of their host via a XerD-first FtsK-independent recombination pathway (25). GGI harboring dif_GGI1 and dif_GGI2 probably define a new class of IMEX. Future work will need to address the Xer recombination pathway they exploit and if they can truly integrate independently of FtsK. In addition, it will be interesting to determine which factors encoded in the genome of GGI IMEX and/or in the genome of their host help them overcome the cellular control that is normally imposed on Xer recombination, as observed for plasmids (18) and the CTXϕ class of IMEX (26).