Report

Structure of a NHEJ Polymerase-Mediated DNA Synaptic Complex

See allHide authors and affiliations

Science  19 Oct 2007:
Vol. 318, Issue 5849, pp. 456-459
DOI: 10.1126/science.1145112

Abstract

Nonhomologous end joining (NHEJ) is a critical DNA double-strand break (DSB) repair pathway required to maintain genome stability. Many prokaryotes possess a minimalist NHEJ apparatus required to repair DSBs during stationary phase, composed of two conserved core proteins, Ku and ligase D (LigD). The crystal structure of Mycobacterium tuberculosis polymerase domain of LigD mediating the synapsis of two noncomplementary DNA ends revealed a variety of interactions, including microhomology base pairing, mismatched and flipped-out bases, and 3′ termini forming hairpin-like ends. Biochemical and biophysical studies confirmed that polymerase-induced end synapsis also occurs in solution. We propose that this DNA synaptic structure reflects an intermediate bridging stage of the NHEJ process, before end processing and ligation, with both the polymerase and the DNA sequence playing pivotal roles in determining the sequential order of synapsis and remodeling before end joining.

DNA double-strand breaks (DSBs) are a potentially lethal form of cellular damage, and failure to repair such breaks can lead to genomic instability (1, 2). In higher eukaryotes, the nonhomologous end joining (NHEJ) pathway is critical for the repair of DSBs (1). A functionally homologous repair system exists in many prokaryotes, where it is used to repair DSBs in stationary-phase and sporulating cells (3, 4). The bacterial NHEJ complex is composed of two proteins, Ku and a multifunctional DNA ligase (LigD) (310). In addition to a core ligase domain, LigD often possesses ancillary polymerase (PolDom) and nuclease domains (615). PolDom, a member of the archaeo-eukaryotic primase (AEP) superfamily (6, 16, 17), in turn has a variety of nucleotidyl transferase activities (69) as well as the ability to generate template distortions and primer realignment (12, 13). Here, we describe the crystal structure of a NHEJ polymerase–mediated synaptic complex, which reveals a DNA-directed mechanism used by repair polymerases to induce synapsis of noncomplementary ends through a dimeric arrangement.

PolDom can interact with a 3′-protruding DNA end containing a 5′-phosphate (5′-P) on the downstream strand (12), a probable first step in end joining, which is compatible with Ku binding near the ends. In the second step, PolDom may endeavor to connect the 3′-protruding DNA ends to configure a “gap-like” synaptic intermediate. When different DNA molecules were tested in polymerization assays, extension of the 3′-protruding strand by the Mycobacterium tuberculosis polymerase domain (Mt-PolDom) of LigD was observed by providing nucleoside triphosphates. However, template extension was restricted to the addition of a few nucleotides, suggesting that the specific nucleotides inserted may be templated, perhaps as a result of the pairing of the 3′ ends. In the DNA shown in Fig. 1A, the 3′-protruding nucleotide [deoxycytidine (dC)] is complementary to the base preceding the nucleotide (dC), which is adjacent to the 5′-P; therefore, a structure resembling a single-nucleotide gap can be formed, either by self-annealing (snap-back) or by connection of two ends (synapsis). In agreement with this mechanism, guanosine triphosphate (GTP) was preferentially incorporated by PolDom.

Fig. 1.

PolDom promotes synapsis of two 3′-protruding ends. Ribonucleotide insertion at 3′-protruding ends was carried out in the presence of 400 nM Mt-PolDom and 100 μM of each individually added nucleotide triphosphate. After incubation for 30 min at 30°C, primer extension products were detected by autoradiography. (A) When using a labeled DNA molecule in which the neighboring nucleotide to the 5′-P (X) was dC, only the expected insertion (G) is observed; however, this does not distinguish between the alternative modes depicted (“snap-back” or “synapsis”). (B) By adding a second 3′-protruding end molecule (also having a 5′-P but unlabeled) that provides a different nucleotide neighbor to the 5′-P (X = A, G, or T), a second prominent insertion is observed in each case, complementary to the X base provided “in trans.”

The specificity of the elongation reaction was analyzed using variants of the 3′-protruding oligonucleotide that differed in the deoxynucleotide (X) adjacent to the internal 5′-P. As predicted, the base preferentially added to the 3′-protruding strand varied as a function of the X nucleotide, which acts as a template (fig. S1). Having shown that the templating nucleotide is adjacent to the 5′-P, we measured whether synapsis of two ends was actually occurring. By simultaneously adding two distinct DNA “ends” (only one labeled) that differed in the nucleotide adjacent to the 5′-P, we observed that extension of the labeled DNA end occurred with either of the two nucleotides complementary to each of the possible templating nucleotides (Fig. 1B). Incorporation of GTP could reflect the snap-back reaction, but incorporation of the other nucleotides into the labeled DNA end was only possible if synapsis occurred with the unlabeled DNA. This establishes that Mt-PolDom is capable of mediating the bridging of two DNA ends. This conclusion is also supported by analytical ultracentrifugation and protein cross-linking studies (fig. S2).

PolDom is a functionally independent domain of LigD (11), which also contains nuclease and ligase domains that can interact with the DNA. As with PolDom, LigD catalyzed the templated elongation of a 3′-protruding DNA end, which could subsequently be ligated (fig. S3A). A 5′-P also stimulated this templated reaction catalyzed by LigD (fig. S3A), confirming that phosphate recognition is also critical for LigD. Moreover, LigD catalyzed 3′-terminal additions that were templated by a distinct DNA end, providing biological relevance to the finding that synapsis is mediated by the polymerization domain of LigD (fig. S3B).

To understand the molecular basis for DNA end binding by a prokaryotic NHEJ polymerase/primase, we elucidated the crystal structure (2.4 Å) of Mt-PolDom in complex with DNA containing a 3′-overhang and a recessed 5′-P. The structure revealed two PolDom-DNA complexes connected via the 3′-protruding DNA ends, forming a NHEJ polymerase–mediated synaptic complex (Fig. 2A). PolDom contains a prominent surface β-hairpin structure, loop 1 (Fig. 2, A to C), which is specific to NHEJ AEPs such as Mt-LigD and Pseudomonas aeruginosa LigD (Pa-LigD) (fig. S4). Conserved residues on loop 1 interact with the 3′-protrusion and orient the synapsis of the ends (Fig. 2D). The 3′-overhangs of the opposing template strands are brought together via a number of base-pairing and stacking interactions (Fig. 2E and fig. S6) (18). Each PolDom monomer makes intimate contact with the 5′-P on the downstream strand, which is bound in a positively charged pocket formed by Lys16 and Lys26 (Fig. 3A), two residues absolutely conserved in NHEJ AEPs (fig. S4). Notably, the N-terminal PolDom region containing Lys16 is absent in AEPs from Archaea and Eukarya (fig. S4).

Fig. 2.

Crystal structure of a polymerase-mediated DNA synaptic complex. (A) Ribbon representation of two Mt-PolDom–DNA binary complexes, in which each polypeptide and DNA chain is colored differently. (B) Electrostatic surface potential of the two PolDom monomers, emphasizing the symmetry of the synaptic complex and the positively charged atrium (blue area) nesting the synapsed 3′-ends. (C) Close-up view emphasizing the protruding β hairpin (loop 1) provided by each PolDom monomer (darker color) for dealing with the synapsis. (D) A view illustrating the residues (22) of loop 1 that bind to DNA in the synapsed region, in particular stabilizing the flipped-out base (C8) in the F chain (red). (E) Schematic view of the synapsed DNA, in which each DNA strand is differently colored as in (A) to (D). Nucleotides C12 and G13 of chain D (cyan) are depicted in light gray, as they are not seen in the crystal structure. The flipped-out base (C8 in chain F) is outlined in dark green. The bases A11 (chain D) and G13 (chain F), colored dark blue and magenta, respectively, form hairpin-like structures contributed by both 3′-ends.

Fig. 3.

Crystal structure of the Mt-PolDom monomer complexed with a 3′-protruding DNA end. (A) Ribbon representation of the Mt-PolDom complexed with a T/D molecule, emphasizing the most critical contacts or subdomains (loops) involved in DNA interaction. The first N-terminal 112 amino acids are colored in blue to outline domain similarities with the “8-kD” domain of Pol λ (18). The template strand (T) is shown in red and the downstream strand (D) in green. The location of an incoming GTP (pink; modeled from structure at PDB code 2IRX) supports the functional importance of the complex. (B) Ribbon representation of the ternary complex consisting of Pol λ, a single-nucleotide gap, and dideoxythymidine triphosphate (ddTTP) (PDB code 2BCV), emphasizing similar residues acting as DNA ligands at the downstream side of the gap; these are most likely critical for the NHEJ function of Pol λ. The region corresponding to the “8-kD” domain is depicted in blue. Red, template strand; orange, primer strand; green, downstream strand; pink, ddTTP.

The Mt-PolDom mutant (Lys16 → Ala) was unable to bind to DNA and had very reduced polymerase activity (fig. S5), whereas gap filling was normal. Other interactions with DNA are indicated in figs. S4 and S6. The PolDom-DNA interactions are reminiscent of the contacts observed in the structure of the evolutionary unrelated NHEJ polymerase, Pol λ–gapped DNA complex (18, 19) (Fig. 3).

The apical loop 1 (β5-β6) interacts with the 3′-protruding strand, thus constituting a potentially important element for maintaining the synapsis between two 3′-protruding DNA ends (Fig. 2, C and D). To analyze the functional importance of these interactions, we mutated loop 1 residues (83 to 85) to alanine and evaluated the DNA binding and polymerization capacity of the resulting mutant (mut-loop). On a gapped DNA substrate, the DNA binding potential of mut-loop was equivalent to that of wild-type PolDom (fig. S7). Therefore, the presence of a 5′-P appears to be enough to ensure enzyme-DNA stability in a gap, and loop 1 is dispensable when the primer terminus, the template, and the 5′-P are physically connected and not discontinuous. However, the integrity of loop 1 was critical to forming a synaptic complex of two 3′-protruding DNAs. Electrophoretic mobility shift and 3′-extension assays showed that mut-loop was very inefficient at forming a synaptic complex (fig. S8). An analogous loop-like structure may play a related role in eukaryotic NHEJ polymerases (20, 21).

The importance of PolDom, and loop 1 in particular, in mediating DNA synapsis was further probed by fluorescence resonance energy transfer (FRET) using DNA with a 3′-overhang identical to that present in the crystal structure. The steady-state fluorescence spectra of doubly labeled 3′-protruding DNA (3′-fluorescein donor and 3′-rhodamine acceptor) with increasing amounts of wild-type Mt-PolDom showed a marked concentration-dependent increase in emission of the rhodamine fluorophore at 605 nm (Fig. 4) due to FRET from fluorescein. The presence of a PolDom-dependent FRETemission peak signifies a close approach of the 3′-overhang with the duplex region of another DNA, indicative of a stable protein-mediated interaction between two DNA ends. In contrast, the mut-loop mutant exhibited a markedly reduced FRET signal, indicating that loop 1 plays a critical role in stabilizing the synaptic complex. This conclusion is further supported by protein cross-linking studies (fig. S2).

Fig. 4.

PolDom-mediated DNA synapsis probed by FRET. (A) Steady-state fluorescence spectra of 3′-FAM (donor)/3′-ROX (acceptor)–labeled, 3′-protruding DNA with varying ratios of Mt-PolDom; ratios of PolDom to DNA are indicated at the right. The doubly labeled 3′-protruding DNA was incubated with increasing amounts of either wild-type (WT, left panel) or loop-mutant (mut-loop, right panel) Mt-PolDom. The presence of a FRET emission peak signifies a close approach of the 3′-overhang with the duplex portion of another DNA, mediated by PolDom. The difference in maximal FRET signal between WT and mutant protein indicates that the loop residues play a critical role in synaptic complex formation. (B) Relative FRET differences between WT and mut-loop (AAA) Mt-PolDoms in the presence of DNA (at ROX emission maxima, λϵm = 605 ± 2 nm) showing that, at a ratio of 20:1, WT PolDom exhibits a >4-fold increase in fluorescence emission due to FRET relative to the mutant PolDom-DNA. The DNA concentration is 50 nM.

The structure presented here establishes that NHEJ polymerases can promote the formation of end-bridging complexes, thereby directing the break alignment process (fig. S9). The limited number of contacts made between the enzyme and the 3′-protrusions suggests that PolDom, and presumably other NHEJ polymerases, allow a large degree of rotational freedom that enables the termini to search for sequence complementarities on the opposing break. This “homology” searching process acts, together with Ku, to align the break by forming presynaptic bridging structures, promoted by favorable microhomology-directed base pairing, that nucleate the formation of the synaptic complex (fig. S9). Thus, final end synapsis, like that shown in the crystal structure, may require a certain degree of mispairing, template dislocation, or realignment facilitated by base flipping, and the eventual formation of hairpin structures at the terminal ends. The hairpin-like structures observed—located in a large, solvent-accessible channel within the PolDom complex—could conceivably accommodate the small 3′-exonuclease domain of LigD (NucDom) (6, 15), facilitating the controlled resectioning of the ends. This may possibly explain the preference of NucDom for recessed 3′-ends (6, 15) and suggests that the nuclease resection process may be regulated by the conformation of the ends within the synaptic complex.

Supporting Online Material

www.sciencemag.org/cgi/content/full/318/5849/456/DC1

Materials and Methods

SOM Text

Figs. S1 to S9

Table S1

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article