DNA polymerase θ (POLQ), double-strand break repair, and cancer

Abstract

DNA polymerase theta (pol θ) is encoded in the genomes of many eukaryotes, though not in fungi. Pol θ is encoded by the POLQ gene in mammalian cells. The C-terminal third of the protein is a family A DNA polymerase with additional insertion elements relative to prokaryotic homologs. The N-terminal third is a helicase-like domain with DNA-dependent ATPase activity. Pol θ is important in the repair of genomic double-strand breaks (DSBs) from many sources. These include breaks formed by ionizing radiation and topoisomerase inhibitors, breaks arising at stalled DNA replication forks, breaks introduced during diversification steps of the mammalian immune system, and DSB induced by CRISPR-Cas9. Pol θ participates in a route of DSB repair termed "alternative end-joining" (altEJ). AltEJ is independent of the DNA binding Ku protein complex and requires DNA end resection. Pol θ is able to mediate joining of two resected 3' ends harboring DNA sequence microhomology. "Signatures" of Pol θ action during altEJ are the frequent utilization of longer microhomologies, and the insertion of additional sequences at joining sites. The mechanism of end-joining employs the ability of Pol θ to tightly grasp a 3' terminus through unique contacts in the active site, allowing extension from minimally paired primers. Pol θ is involved in controlling the frequency of chromosome translocations and preserves genome integrity by limiting large deletions. It may also play a backup role in DNA base excision repair. POLQ is a member of a cluster of similarly upregulated genes that are strongly correlated with poor clinical outcome for breast cancer, ovarian cancer and other cancer types. Inhibition of pol θ is a compelling approach for combination therapy of radiosensitization.

Keywords: Alternative end-joining; DNA double strand breaks; DNA polymerase; DNA synthesis; MMEJ; Synthetic lethality.

Figure 1

A double-strand break can be repaired by cNHEJ with minimal end processing. If the break ends are resected to produce 3’ single-stranded tails, an altEJ pathway can be invoked (involving pol θ). Alternatively, homologous recombination (HR) can take place, depending on availability of a copy of the damaged gene.

Figure 2

Tripartite domain structure of human pol θ. The most conserved region of the helicase-like domain is highlighted in yellow. The DNA polymerase domain (blue) includes a nonfunctional exonuclease domain (“exo”), and three insertion loops designated ins1, ins2, and ins3. A central region (grey) connecting the two domains is predicted to be largely disordered.

Figure 3

A. Crystal structure of the polymerase domain of human pol θ. The polymerase domain harbors the canonical fingers (colored in blue), palm (red), and thumb (green) subdomains as well as an exonuclease-like domain devoid of proofreading activity (shown in grey). The crystal structure revealed that the polymerase domain comprises five unique insertions: three in the polymerase domain (insertions 1–3) and two in the exonuclease-like domain (exo1 and exo2). Disordered segments not visible in the crystal structure are indicated with dotted lines. B. Features of pol θ that grasp the primer-terminus. In this view of the ternary complex of the pol θ DNA polymerase domain, five Lys and Arg residues make specific contacts with phosphate residues (shown as nubs) in the primer DNA strand (orange). Two of these contacts are conserved in all A-family polymerases (shown with white carbons), and three of them are unique residues in pol θ (cyan carbons), with Arg2254 emerging from the distinctive Insert 2 (yellow). Additional unique contacts are made by Gln2384 and Tyr2387, which contact the major groove side of the base and the phosphate of the incoming nucleotide, respectively [from data in [56]]. For the sake of clarity, only the backbone of the primer strand is shown. Color coding of the protein subdomains is as in Figure 3A. C. Possible mechanism for alignment of terminal microhomologies, based on the dimerization of the pol θ helicase-like domain (POLQ-HLD) as observed in crystal structures [78]. Connectivity between the polymerase and POLQ-HLD, not shown here, could be in cis or trans.

Figure 4. POLQ is expressed in a cluster of genes predictive of breast cancer subtypes

The heatmaps show gene expression (high, dark red; low, dark blue) derived from the UCSC cancer genomics website (https://genome-cancer.ucsc.edu/proj/site/hgHeatmap/). Here, data were analyzed using the dataset “TCGA breast invasive carcinoma gene expression by RNAseq (IlluminaHiSeq), pancan normalized • N=1215”. The data are organized according to estrogen receptor (ER), progesterone receptor (PR), and HER2 amplification status (HER2) in the columns at the right. The orange color shows positive status, blue is negative, and grey shows undetermined status. Each row is a different breast cancer case. The area of the heatmaps containing triple-negative breast cancers is boxed with a solid line. The gene subset “a” in the top heatmap has a distinctive pattern of exceptionally high expression in triple-negative cancers (red boxed area) and exceptionally low expression in another set of cancers (blue starred region). The heatmap was produced with the subtype predictor PAM50 set of genes [90], with each column representing a gene in the order shown in the list. The bottom heatmap used POLQ and its 43 nearest expression neighbors where expression information is available in the target TCGA dataset. The expression neighbors were determined with the multi-experiment matrix tool (MEM, http://biit.cs.ut.ee/mem/index.cgi, output filtered with the text search “breast”) by querying AFFY44 human microarrays. Each column represents a gene in the order shown in the list. The heatmap pattern for the POLQ expression neighbors genes closely resembles the pattern for the subset of PAM50 genes shown at the left in the top panel, and five of the genes overlap between these sets as indicated in bold.