SAMHD1 restrains aberrant nucleotide insertions at repair junctions generated by DNA end joining

Abstract

Aberrant end joining of DNA double strand breaks leads to chromosomal rearrangements and to insertion of nuclear or mitochondrial DNA into breakpoints, which is commonly observed in cancer cells and constitutes a major threat to genome integrity. However, the mechanisms that are causative for these insertions are largely unknown. By monitoring end joining of different linear DNA substrates introduced into HEK293 cells, as well as by examining end joining of CRISPR/Cas9 induced DNA breaks in HEK293 and HeLa cells, we provide evidence that the dNTPase activity of SAMHD1 impedes aberrant DNA resynthesis at DNA breaks during DNA end joining. Hence, SAMHD1 expression or low intracellular dNTP levels lead to shorter repair joints and impede insertion of distant DNA regions prior end repair. Our results reveal a novel role for SAMHD1 in DNA end joining and provide new insights into how loss of SAMHD1 may contribute to genome instability and cancer development.

Graphical Abstract

The dNTPase SAMHD1 keeps dNTP levels at moderate levels, incompatible with aberrant overhang resynthesis and insertions at DNA repair joints.

Figure 1.

DNA Repair of plasmid substrates is skewed by SAMHD1. (A) A pool of linear DNA plasmids (substrate #1-#9, with the respective DNA end structures indicated) was transfected into SAMHD1-Flag expressing or non-expressing HEK293 cells. (B) Repaired plasmid junctions were PCR-amplified from extracted DNA 0 and 72 h post transfection from 3 independent experiments. A control PCR on TP53 is shown on the bottom. Bands corresponding in size to repair junctions are indicated with an asterisk. (C, D) Repair frequencies of inter- and intramolecular repair events (substrate #1–#9) are shown as circos plots for SAMHD1-Flag non-expressing and expressing HEK293 cells. The size of the segments reflects the occurrence of the nine different DNA plasmids serving as substrate for repair. Repair frequencies are indicated by size of ribbons (mean values from three independent experiments). Start of ribbon denotes forward primed arms; arrow of ribbon denotes reverse primed arm of substrates. Repair junctions significantly overrepresented (C vs D and vice versa) are colored red (P < 0.05) and light red (0.05 < P < 0.1), and the correspondingly underrepresented repair junctions are colored blue and light blue, respectively. Significances were calculated by two-tailed t-tests with unequal variances, n = 3. (E) Repair frequencies from C and D are shown separately for intramolecular (#1–#9) and intermolecular (IM) repair in HEK293 cells either expressing or not-expressing SAMHD1-Flag. (F) Repair frequencies from C and D are shown separately for intermolecular repair in HEK293 cells either expressing or not-expressing Flag-SAMHD1. Means are indicated; significances were calculated by two-tailed t-tests with unequal variances, n = 3.

Figure 2.

Repair junctions are modulated by SAMHD1. (A, B) PCR-amplified repair junctions from SAMHD1-Flag non-expressing (A) or expressing (B) HEK293 cells were sequenced and the frequencies of the lengths of the respective sequences are shown for each DNA plasmid substrate (#1-#9). The arrow within the graph for substrate #2 denotes the size of the repair junction corresponding to junctions repaired by MMEJ using a 6bp microhomology. (C) The respective DNA break structures for DNA plasmid substrates #1- #9 is shown. (D) The sequences of the most frequent junctions for cells not expressing Flag-SAMHD1 are shown and framed in blue. The most frequent repair junctions for Flag-SAMHD1 expressing cells (if different to non-expressing cells), are shown below and framed in red. The color of the frame allocates the sequences to the respective bars shown in (A, B). Bars are showing mean + SD from three independent experiments. (E) DNA break structure of DNA plasmid substrate #6 carrying cohesive 4 nt 5′ overhangs (top) and DNA sequences of two potential joining events (bottom) are shown. (F) The ratio of repair junctions deriving from direct cohesive joining (dcj) of DNA ends to junctions deriving from blunting prior joining by filling up ssDNA overhangs (bpj) for HEK293 cells either expressing or non-expressing SAMHD1-Flag is shown. (G) DNA break structure of DNA plasmid substrate #2 carrying a 6 bp microhomology (top) and DNA sequences of two potential joining events (bottom) are shown. (H) The ratio of repair junctions deriving from direct joining (dj) of DNA ends to junctions deriving from MMEJ for HEK293 cells either expressing or non-expressing SAMHD1-Flag is shown. (mean values are indicated; significances were calculated by unpaired two-tailed t-tests from three independent experiments).

Figure 3.

Repair of staggered ends with microhomologies by SAMHD1. (A, B) PCR-amplified repair junctions from SAMHD1-Flag non-expressing (A) or expressing (B) HEK293 cells were sequenced and the frequencies of the lengths of the respective sequences are shown for each DNA plasmid substrate (#11–#18). (C) The respective DNA break structures for DNA plasmid substrates are shown. (D) The sequences of the most frequent junctions for cells not expressing Flag-SAMHD1 are shown and framed in blue. The most frequent repair junctions for Flag-SAMHD1 expressing cells (if different to non-expressing cells), are shown below and framed in red (substrate #17 and #18). The color of the frame allocates the sequences to the respective bars shown in (A, B). (Bars are showing mean + SD from two independent experiments). (E) DNA break structure of DNA plasmid substrate #17 carrying cohesive 4 nt 5′ overhangs with 3bp microhomologies (top) and DNA sequences of two potential joining events (bottom) are shown. (F) The ratio of repair junctions deriving from direct cohesive joining (dcj) of DNA ends to junctions deriving from blunting prior joining (bpj) by filling up ssDNA overhangs (bpj) for HEK293 cells either expressing or non-expressing SAMHD1-Flag is shown. (G) DNA break structure of DNA plasmid substrate #18 carrying an internal 2bp microhomology (top) and DNA sequences of two potential joining events (bottom) are shown. (H) The ratio of repair junctions deriving from direct cohesive joining (dcj) of DNA ends to junctions deriving from blunting prior joining (bpj) for HEK293 cells either expressing or non-expressing SAMHD1-Flag is shown. (mean values are indicated; significances were calculated by unpaired two-tailed t-tests from two independent experiments).

Figure 4.

Induction of a chromosomal deletion at chr11q using nCas9. (A) Schematic representation of chr11 indicating nCas9 sites and primers for PCR amplification of breakpoint junctions. (B) Length of PCR-amplified amplicon-sequenced breakpoint junctions from HEK293 and HEK293^SAMHD1-KO cells transfected with the indicated SAMHD1 variants (K312A or wt; n = 4 for all samples). A length corresponding to joining of blunted (filled up) 5′ overhangs is set to 0 bp. (C) Western blots of HEK293 and HEK293^SAMHD1-KO cells (left), transiently transfected with the indicated constructs (right). For (B), median with interquartile range is shown in boxes, with whiskers extending the boxes with the largest/smallest value no further than 1.5 times of the interquartile range and other points plotted individually; significances were calculated by Mann-Whitney test and are indicated above the graph, medians are given above the x-axis.

Figure 5.

Elevated dNTP pool increases insertions at repair junctions. (A) HEK293^SAMHD1-KO cells were transiently transfected with constructs encoding SAMHD1 and SAMHD1 K11A mutants N-terminally fused to mCherry together with del11q inducing nCas9 constructs encoding zsGreen1 as separately transcribed gene. (B) Schematic representation of nCas9 induced del11q and primers for PCR amplification of breakpoint junctions. (C) Length of PCR-amplified amplicon-sequenced breakpoint junctions from HEK293 and HEK293^SAMHD1-KO cells transfected with the indicated SAMHD1 variants (K11A, K484T, K312A or wt) (n = 4 for all samples, except K484T n = 3) and (D) from HEK293 and HeLa cells grown in excessive dN to increase intracellular dNTP pools (n = 2 for HEK293 cells and n = 3 for HeLa cells). Median with interquartile range is shown in boxes, with whiskers extending the boxes with the largest/smallest value no further than 1.5 times of the interquartile range and other points plotted individually; significances were calculated by Mann–Whitney test and are indicated above the graph, medians are given above the x-axis. (E) Circos plots were generated from data in Figure 5D, which show the genome as circle with ribbons indicating the homologies of inserted DNA at the del11q repair junction to the respective distant genomic regions. Ribbons in red indicate that the insertions map to the genome ambiguously, with only one randomly chosen homology shown (detailed mapping results are summarized in Supplementary table S4; n = number of junctions with distant homologies/total number of unique reads analyzed). (F) Bars show the percentage of all amplicon-sequencing reads with insertions mapping to distant genomic sites or to transfected plasmids. Significances in (E, F) were calculated by Fisher's exact test.