The stacked-X DNA Holliday junction and protein recognition

Abstract

The crystal structure of the four-stranded DNA Holliday junction has now been determined in the presence and absence of junction binding proteins, with the extended open-X form of the junction seen in all protein complexes, but the more compact stacked-X structure observed in free DNA. The structures of the stacked-X junction were crystallized because of an unexpected sequence dependence on the stability of this structure. Inverted repeat sequences that contain the general motif NCC or ANC favor formation of stacked-X junctions, with the junction cross-over occurring between the first two positions of the trinucleotides. This review focuses on the sequence dependent structure of the stacked-X junction and how it may play a role in structural recognition by a class of dimeric junction resolving enzymes that themselves show no direct sequence recognition.

Figure 1

Structural forms of DNA Holliday junctions. a. The parallel stacked-X junction initially proposed by Holliday as the recombination intermediate (Holliday, 1964). b. The extended open-X form of a DNA junction. c. The antiparallel stacked-X junction does not allow for migration of the junction along the DNA strands. d. Model of the antiparallel stacked-X junction proposed from solution studies (Duckett et al., 1988).

Figure 2

Single-crystal structures of Holliday junctions. a. The open-X junction in complex with the DNA repair protein RuvA (Hargreaves et al., 1998). b. Antiparallel stacked-X junction in the sequence CCGGTACCGG (Eichman et al., 2000). c. Model of the junction-resolving enzyme Hjc from Sulfolobus solfataricus bound to a stacked-X junction (Middleton et al., 2004).

Figure 3

Sequence effects on B-DNA, A-DNA and Holliday junctions. The sequence dependent stabilization of DNA structures has been determined using a crystallographic screen of the decanucleotide sequence CCnnnN₆N₇N₈GG, where each position in N₆N₇N₈ is allowed to be any of the four standard nucleotides, and the trinucleotide nnn is specified to maintain the inverted repeat symmetry of the motif (Hays et al., 2005). The N₆N₇N₈ trinucleotides that lead to formation of junctions from B-DNA, A-DNA from B-DNA, and junctions from A-DNA are indicated.

Figure 4

Stabilizing interactions in the Holliday junction. The formation of four-stranded DNA junction in the inverted-repeat sequence type d(CCnnnN₆N₇N₈GG) is dependent on the sequence and sequence dependent interactions at the N₆N₇N₈ trinucleotide. A hydrogen bond from the N4 amino of a cytosine at N₈ to the phosphate at the junction cross-over is seen to be essential, but not sufficient, to specify formation of the junction. Cytosine is favored over thymine at the N₇ position (when N₆=A and N₈=C) because the electrostatic interaction from the phosphate oxygens of N₆ to the N4 amino nitrogen of the cytosine base is stronger (distances range from 3.1 to 3.6Å) as compared to the C5 methyl group of the thymine base (distances range from 4.2 to 4.5Å). The preference is A > G > C (with N₇ and N₈=C) for the nucleotide at N₆.

Figure 5

Proposed model for indirect sequence recognition of stacked-X junctions. In this model, migration of the Holliday junction in free DNA requires a transition to the open-X form. This is consistent with models proposed from single-molecule studies on junction isomerization (McKinney et al., 2003) and translocation (Lushnikov et al., 2003). However, certain sequences such as the ACC-trinucleotide in an inverted repeat help to stabilize and stall the junction, thereby presenting a defined structure for recognition by a dimeric junction binding protein. In the complex, the protein induces a structural perturbation that either maintains the topology and symmetry of the stacked-X junction, or induces an open-X junction that can then migrate to a protein specific target site.