RETRACTED: Human DNA polymerase θ harbors DNA end-trimming activity critical for DNA repair

Abstract

Cancers with hereditary defects in homologous recombination rely on DNA polymerase θ (pol θ) for repair of DNA double-strand breaks. During end joining, pol θ aligns microhomology tracts internal to 5'-resected broken ends. An unidentified nuclease trims the 3' ends before synthesis can occur. Here we report that a nuclease activity, which differs from the proofreading activity often associated with DNA polymerases, is intrinsic to the polymerase domain of pol θ. Like the DNA synthesis activity, the nuclease activity requires conserved metal-binding residues, metal ions, and dNTPs and is inhibited by ddNTPs or chain-terminated DNA. Our data indicate that pol θ repurposes metal ions in the polymerase active site for endonucleolytic cleavage and that the polymerase-active and end-trimming conformations of the enzyme are distinct. We reveal a nimble strategy of substrate processing that allows pol θ to trim or extend DNA depending on the DNA repair context.

Keywords: DNA polymerase; Double-strand break repair; dideoxynucleotides; endonuclease; theta-mediated end joining (TMEJ).

Figure 1.. TMEJ is proposed to require an enzymatic activity to cleave 3′-unpaired bases.

Initial pairing of strands at a microhomology usually results in an unpaired 3′-end on one or both tails, which cannot be extended by DNA synthesis.

Figure 2.. Pol θ trims DNA ends.

DNA extension and end-trimming were observed in reactions initiated by mixing preformed binary complexes of pol QM1 and DNA oligonucleotides with dNTPs and Mg²⁺. Oligonucleotides were 5′- labeled (asterisks) with tetrachloro-fluorescein (TET, panels A, B, D-H) or 5′-³²P (panels C, I-K). Assays were conducted with 250 nM DNA, 125 nM enzyme and excess nucleotides; products were resolved on denaturing DNA sequencing gels. A) Substrates ds14AG/21C, ss13AA, ss14AG and ss14CC were assayed with 150 μM dNTPs over a time course spanning from 0 sec to 5 min. FLP = full-length product, +1 = terminal transferase product. B) Reactions with the metal-binding deficient pol QM1 variant D2330A, under the same conditions as in panel A. C) End-trimming of ss14GC takes place in reactions conducted with nucleotides and divalent metal ions Mg²⁺, Mn²⁺ or Co²⁺, but not Ca²⁺ (10 mM). The 5′-labeled marker ladder corresponds to the ss14AG sequence. Primer extension reactions with ds14AG/21T are included as a control. D) Dependence of end-trimming of ss14AG on dNTPs (5 μM to 500 μM each), monitored over a time course as in panel A. E) End-trimming of ss14AG is not observed in reaction mixtures using ddNTPs. F) End-trimming of ss14AG in reactions with 5 μM dNTPs is quenched by titrating ddNTP from 5 μM to 500 μM. G) Nucleotide concentrations vs. experimental rates of end-trimming (panel D) are plotted and fit to the kinetic model: $K_d{}^{app}=99\pm 41\mathrm{\mu }\text{M}$, k_endo = 0.37 ±0.051 sec⁻¹, catalytic efficiency = 0.0037 ±0.0012 μm⁻¹ sec⁻¹ H) Quantification of panel F, showing quenching of end-trimming by increasing concentrations of ddNTPs, in reactions supplemented with 5 μM dNTPs. Black lines show the best fit to a single-exponential model. I) End-trimming of ss14AG and ss14GC requires dGTP and dCTP. Reactions (t = 5 min) were conducted with 50 μM total dNTP concentration for each lane. J) Different ratios of dCTP:dGTP influence the end-trimming vs. elongation reactions (t = 2 min). K) An oligonucleotide terminated with ddCTP is refractory to end-trimming (t = 2 min). See also Figures S1 and S6.

Figure 3.. Conserved residues in the pol θ active site mediate end-trimming.

A) View of the ternary complex of pol QM1 bound to furan-containing DNA and ddATP (PDBID 4X0P (Zahn et al., 2015). Amino acid side chains critical for end-trimming and DNA synthesis are illustrated. The fingers, palm, thumb, and insert 2 regions are depicted in blue, red, green, and yellow, respectively. The DNA backbone of the template and primer strands is shown in pale yellow and orange, respectively, with the 3′-terminus becoming red. The divalent cation coordinated in the pol-active site is depicted as a sphere (magenta). B) Polymerase activity of variant DNA polymerases are compared with a primer-template substrate ds14AG/21T, with full-length product (FLP) marked by an asterisk. C) Reactions with ss14GC (top gel) or ss14AA (bottom gel) were monitored over 5 min with 500 μM dNTPs for pol QM1 and the primer grasp variant (see Fig S2A for other variants). The 30mer oligonucleotide template strand added to reactions in lane 7 with the nucleotides and Mg²⁺ shares limited complementary to the primers, but successfully primes DNA synthesis (see Table S1 or Fig S2A for sequence) D) End-trimming of ss14GC during the time courses in panel C and Fig S2A is quantified for each variant polymerase and fit to a single-exponential model (solid lines). E) The second step of the full reaction with ss14GC, in which the cleaved terminus is extended, is also quantified and fit to a single-exponential model (solid lines). F) Extension of ss14AA by variant polymerases was quantified by taking the summed intensity of all elongated species as the total product, and fit to a single-exponential model (solid lines). See also Figures S2 and S6.

Figure 4.. Pol θ reconfigures its active site.

Kinetic measurements by stopped-flow spectroscopy were conducted by mixing preformed pol/DNA binary complexes with solutions of Mg²⁺ and dNTPs. A) Reaction traces and fit to the kinetic models for three different substrates are plotted (ss14AG+ddCTP+dGTP, cyan, model #3; ss14AG+dNTPs, green, model #2; and ds14AG21T+dATP, magenta, model #1; see methods for details selecting kinetic models) to visualize different conformational trajectories of pol QM1 during DNA processing. B) Reaction traces during end-trimming of ss14AG with dNTPs (25 – 1000 μM). The solid black lines show the fit to a kinetic model #2 (double-exponential curve). C) Experimental rates are plotted in magenta and cyan, as indicated. Solid lines show the best fit to a hyperbola. D) Trapping of the end-trimming complex with ss14AG+ddCTP+dG is plotted and fit to kinetic model #3 (single exponential decay). E) The experimental rate of end-trimming complex capture is plotted vs. the concentration of ddCTP from 1 μM to 1000 μM (magenta, dGTP constant at 10 μM) or dGTP from 1 μM to 1000 μM (cyan, ddCTP constant at 10 μM). Solid lines show the best fit to a hyperbola. See also Figure S3.

Figure 5.. End processing can occur on double-stranded DNA with mismatched 3′-termini.

A) Strategy for pol QM1 reaction with substrates in vitro prior to sequencing analysis. The primer (blue) is annealed to a template (red) with a BamHI (also red) recognition site in the duplex region. Full extension of the primer by pol θ forms an EcoRI (purple) sequence. Following dual restriction digestion, products are ligated into pUC19. Various products may be formed (illustrated by the bulge), arising from mismatches left or editing performed by pol θ. B) PCR with m13 primers generates amplicons of approximately 100 bp which are then subjected to DNA sequencing. C) The substrates ds14′GA/30A (black) and ds14′AG/30A (green) are diagrammed D) Reactions of ds14GA/30A and ds14AG′/30A are compared on a denaturing gel as indicated. The FLP (*) marked by “n+1” is a single nucleotide longer with the ds14AG/30A substrate. E) The 28/54 series of substrates are described graphically to show that the only difference between ds28GA/54A (black) and ds28AG/54A (green) is the two 3′ bases of the primer at the 5′-overhang. F) The predominant ways by which pol θ processes the substrates are realignment of the terminus (left), direct repeat formation (center), and end-trimming (right). See also Figure S4.

Figure 6.. Pol θ end-trimming depends on the ability to transiently pair 3′-ends.

A) Denaturing sequencing gel showing consequences of alterations to the sequence of ss14AG. Lanes supplied with “dCTP*” contain α-³²P-dCTP, which was reacted with unlabelled, unphosphorylated DNA. The product apparent in lane 8 therefore represents labelling of the end-trimming product by subsequent incorporation of dCTP*. Below each substrate set, possible transient DNA pairings that could facilitate the observed reaction products are shown. B) Changes to ss14AG at the two 3′ bases are evaluated and compared in the presence of 10 mM Mg²⁺, or 10 mM Mg²⁺ supplemented with 1 mM Mn²⁺. Proposed transient DNA pairings are diagrammed as in panel A. C) Formation of protein/DNA complexes was monitored by electrophoretic mobility shift assays (EMSA) with 250 pM oligonucleotide. 100 μM ddCTP and 10 μM dGTP were supplied where indicated. D) EMSAs for substrate ss14GC (open circles) with supplemented ddCTP and dGTP were quantified and fit to a hyperbola (solid line). Error bars represent standard deviation for three independent replicate experiments. E) Purification of pol θ-FL yields a protein at the predicted MW, validated by immunoblotting. F) Primer extension assays conducted with pol θ-FL, ss14AG and 100 μM dNTPs recapitulate the end-trimming reactions observed with the polymerase domain construct pol QM1. The right two lanes show extension in the presence of a 30mer template, as employed with pol QM1 in Figure 3. See also Figure S5.

Figure 7.. Model for manipulation of 3′-ends in the polymerase active site of Pol θ.

(1) Initial pairing of strands at a microhomology often results in an unpaired 3′-end on one or both tails, which cannot be extended by DNA synthesis (2) Pol θ can transiently self-pair one 3′-end and catalyze nucleolytic end-trimming (2A), which can generate deletions within the flanks of repair junctions. Intramolecular “snapback” DNA synthesis is also possible at this step, primed by the transient self-pairing of the 3′-tail (2B). (3) Extension by the DNA polymerase activity of pol θ in the pol-active site is now possible. Both the end-trimming and DNA polymerase reactions depend on dNTPs, the 3′-OH of the DNA, and divalent cation, preferably Mg²⁺. Dideoxynucleotides inhibit both end-trimming and polymerase activities. See also Figure S7.