RECQL5/Recql5 helicase regulates homologous recombination and suppresses tumor formation via disruption of Rad51 presynaptic filaments

Abstract

Members of the RecQ helicase family play critical roles in genome maintenance. There are five RecQ homologs in mammals, and defects in three of these (BLM, WRN, and RECQL4) give rise to cancer predisposition syndromes in humans. RECQL and RECQL5 have not been associated with a human disease. Here we show that deletion of Recql5 in mice results in cancer susceptibility. Recql5-deficient cells exhibit elevated frequencies of spontaneous DNA double-strand breaks and homologous recombination (HR) as scored using a reporter that harbors a direct repeat, and are prone to gross chromosomal rearrangements in response to replication stress. To understand how RECQL5 regulates HR, we use purified proteins to demonstrate that human RECQL5 binds the Rad51 recombinase and inhibits Rad51-mediated D-loop formation. By biochemical means and electron microscopy, we show that RECQL5 displaces Rad51 from single-stranded DNA (ssDNA) in a reaction that requires ATP hydrolysis and RPA. Together, our results identify RECQL5 as an important tumor suppressor that may act by preventing inappropriate HR events via Rad51 presynaptic filament disruption.

Figure 1.

Cancer susceptibility phenotype of Recql5^−/− mice. (A) Cancer-free survival of wild-type (n = 32) and Recql5 knockout (n = 50) mice. The animals were monitored for up to 22 mo or until they succumbed to cancer. The two wild-type mice with cancer were identified when they were sacrificed at 22 mo of age. All tumor cases were determined based on the results of pathological analysis. (B) A table summarizing the cancers observed in Recql5^−/− mice, including the types of cancers, the frequency for individual type of cancers, and their ages (or average ages) of on-set. (C–F) Images of hematoxylin and eosin (H&E) histology of four representative cancers found in Recql5^−/− mice. (C) A well-differentiated squamous cell carcinoma with keratin formation and focal invasion. (D) A well-differentiated lung adenocarcinoma of the predominantly papillary type. (E) A breast carcinoma characteristic of malignant epithelial cells with poor gland formation by axillary mass histopathology. (F) A follicular lymphoma exhibiting characteristic mixture of small and medium-sized abnormal lymphocytes with features of centroblasts and centrocytes. Bars: C–E, 50 μm; F, 10 μm.

Figure 2.

Detection of γ-H2AX and Rad51 focus formation in MEFs by immunostaining. Detection of spontaneous Rad51 (A) and γ-H2AX (B) foci in MEFs. (C) Representative images of cells with γ-H2AX, and/or Rad51 foci. (D–F) The accumulation of MEF cells with Rad51 foci (D), γ-H2AX foci (E), and cells with γ-H2AX and Rad51 foci colocalization (F) at various time points after exposure to 50 nM CPT. In each data set, only cells with >10 foci were scored as positive. Each data point represents the mean ± SEM from at least 500 nuclei examined in two independent experiments. Bars, 10 μm.

Figure 3.

HR-mediated repair of I-SceI-induced DSBs in mouse ES cells. (A) Schematic illustration of the SCneo system for monitoring HR-mediated repair of I-SceI-induced DSBs. A single copy of the SCneo substrate was introduced into the Rb locus in both wild-type and Recql5 knockout ES cells by gene targeting. Repair of the DSB by HR results in the reconstitution of a functional Neo cassette and G418 resistance. (B) Recovery of G418-resistant colonies from cells containing the SCneo cassette. ES cells containing the SCneo cassette were cultured in medium containing G418 after transfection with either the expression vector for I-SceI or the corresponding empty vector by electroporation. G418-resistant colonies were scored at 10 d after transfection. The average values from three independent experiments are presented above the histograms.

Figure 4.

Effect of CPT treatment on chromosome stability in ES cells. (A) A metaphase spread from a wild-type cell. (B,C) Representative images of two metaphase spreads from CPT-treated Recql5 knockout ES cells. A number of abnormal features, including triradial (solid arrowheads), quadriradial (solid arrows), DNA fragments (asterisks), and end-to-end fusion (open arrowheads) are indicated. Bars, 5 μm. (D) Summary of the data from the metaphase spread analysis.

Figure 5.

Human RECQL5 binds Rad51 and inhibits the D-loop reaction. (A) Purified RECQL5 and Rad51 were mixed and subjected to immunoprecipitation with anti-RECQL5 antibodies. The reaction supernatant (S), wash (W), and eluate (E) were analyzed by SDS-PAGE. (IgH) Immunoglobulin heavy chains. (B) D-loop reaction scheme. (C,D) Rad51 K133R presynaptic filaments were incubated with RECQL5 or WRN and RPA (where indicated) before Hop2-Mnd1 and pBluescript form I DNA were incorporated. (Bl) Blank containing DNA substrates only. (E) The average values ± SEM from three or more independent experiments are plotted. (F) D-loop reactions with RPA and RECQL5 or RECQL5 K58R. (Std) Standard reaction with RPA but no RECQL5 or RECQL5 K58R. (G) Effects of order of addition of reaction components. Presynaptic filaments of Rad51 K133R were incubated with the rest of the components in the indicated orders. Hop2-Mnd1 was added together with the dsDNA to all the reactions. The concentration of RECQL5 and RECQL5 K58R was 45 nM. The average values ± SEM from three or more independent experiments are plotted in F and G.

Figure 6.

Turnover of the presynaptic filament as revealed by Topoisomerase-linked DNA topology modification. (A) The reaction scheme. (PK) Proteinase K. (B) Topologically relaxed duplex DNA was incubated with the indicated proteins, with or without calf thymus topoisomerase I. Only Rad51 K133R makes form U DNA. (C) Rad51 K133R presynaptic filaments were assembled and then incubated with RECQL5, RECQL5 K58R, or WRN, with or without RPA. Topologically relaxed duplex DNA and topoisomerase were subsequently added to complete the reaction. Form U DNA marker was run in lane 2. (Rl) Topologically relaxed duplex DNA; (ss) ssDNA.

Figure 7.

Analysis of presynaptic filament dissociation by electron microscopy. (A) Presynaptic filaments of Rad51 K133R on the 150-mer ssDNA. (B) Nucleoprotein complexes of RPA and the 150-mer ssDNA. The black scale bar in A and B denotes a length of 50 nm. (C) The relative abundance of Rad51 K133R presynaptic filaments and RPA–ssDNA complexes in reactions that contained the indicated protein components. Over 2500 nucleoprotein complexes were counted in each case.

Figure 8.

Model depicting the action mechanism of RECQL5 on Rad51 presynaptic filaments. RECQL5 utilizes the free energy from ATP hydrolysis to catalyze the dismantling of the Rad51 presynaptic filament. The ssDNA generated as a result of Rad51 removal is immediately occupied by RPA to prevent the reloading of Rad51.