Trex2 enables spontaneous sister chromatid exchanges without facilitating DNA double-strand break repair

Abstract

Trex2 is a 3' → 5' exonuclease that removes 3'-mismatched sequences in a biochemical assay; however, its biological function remains unclear. To address biology we previously generated trex2(null) mouse embryonic stem (ES) cells and expressed in these cells wild-type human TREX2 cDNA (Trex2(hTX2)) or cDNA with a single-amino-acid change in the catalytic domain (Trex2(H188A)) or in the DNA-binding domain (Trex2(R167A)). We found the trex2(null) and Trex2(H188A) cells exhibited spontaneous broken chromosomes and trex2(null) cells exhibited spontaneous chromosomal rearrangements. We also found ectopically expressed human TREX2 was active at the 3' ends of I-SceI-induced chromosomal double-strand breaks (DSBs). Therefore, we hypothesized Trex2 participates in DNA DSB repair by modifying 3' ends. This may be especially important for ends with damaged nucleotides. Here we present data that are unexpected and prompt a new model. We found Trex2-altered cells (null, H188A, and R167A) were not hypersensitive to camptothecin, a type-1 topoisomerase inhibitor that induces DSBs at replication forks. In addition, Trex2-altered cells were not hypersensitive to γ-radiation, an agent that causes DSBs throughout the cell cycle. This observation held true even in cells compromised for one of the two major DSB repair pathways: homology-directed repair (HDR) or nonhomologous end joining (NHEJ). Trex2 deletion also enhanced repair of an I-SceI-induced DSB by both HDR and NHEJ without affecting pathway choice. Interestingly, however, trex2(null) cells exhibited reduced spontaneous sister chromatid exchanges (SCEs) but this was not due to a defect in HDR-mediated crossing over. Therefore, reduced spontaneous SCE could be a manifestation of the same defect that caused spontaneous broken chromosomes and spontaneous chromosomal rearrangements. These unexpected data suggest Trex2 does not enable DSB repair and prompt a new model that posits Trex2 suppresses the formation of broken chromosomes.

Figure 1

Trex2 knockouts. (A1) Targeting Trex2 with the conditional vector in h2ax^−/− and ku70^−/− cells. The targeting vector contains floxed puΔtk (Chen and Bradley 2000) and the human short TREX2 isoform (Chen et al. 2007b). After gene targeting the mouse Trex2 coding sequences (red) were replaced with the human short TREX2 isoform (green). (A2) Screen for targeted clones. PCR amplification with TX2LR55 and hTX2Rev is shown. (A3) Cre-mediated deletion of human TREX2 and puΔtk. (A4) Screen for Trex2-deleted clones by PCR amplification with Cre1 and Cre2. (A5) Verification of Trex2 deletion by RT-PCR with mTx2For and mTx2Rev primers shown in A1. Detection of Rad51 mRNA was used as a loading control (Holcomb et al. 2007). WT, wild-type control. (B1) Replacing Trex2 in blm^−/− cells with puromycin acetyltransferase expressed by the phosphoglycerate kinase (PGK) promoter. (B2) Screen for targeted clones. PCR amplification with TX2LR55 and HATrev is shown. (B3) Verification of targeted clones with RT-PCR using mTx2For and mTx2Rev primers shown in A1. Detection of Rad51 mRNA was used a loading control.

Figure 2

Camptothecin (CPT)-induced broken chromosomes. (A) Trex2-altered cells exhibited the same dose response to camptothecin as control cells (AB2.2). Four clones of Trex2^hTX2, Trex2^H188A, and Trex2^R167A were compared to the parental clones of trex2^null and trex2^null-2 cells (Dumitrache et al. 2009) and control cells. These experiments were done three times. The four clones from each experiment were averaged and then the three experiments were averaged. Standard deviation is shown. SF, survival fraction. (B) The percentage of metaphase spreads (MPS) with microscopically visible spontaneous and camptothecin-induced broken chromosomes. AB2.2, control cells. (C) Summary of results. Shown is the number of MPS with zero broken (br) chromosomes, one broken chromosome, two broken chromosomes, or three broken chromosomes. No metaphase spread had more than three broken chromosomes. Statistics (Student’s t-test) refer to lanes in B for comparisons: 1 and 2, P = 0.0001; 1 and 3, P < 0.0001; 1 and 4, P = 0.0003; 2 and 3. P = 0.0141; 2 and 4, P = 0.026; 3 and 4, P = 0.398; 5 and 6, P = 0.061; 5 and 7, P = 0.0003; 5 and 8, P = 0.0008; 6 and 7, P = 0.033; 6 and 8, P = 0.013; 7 and 8, P = 0.391; 9 and 10, P = 0.708; 9 and 11, P = 0.321; 9 and 12, P = 0.481; 10 and 11, P = 0.517; 10 and 12, P = 0.847; 11 and 12, P = 0.608; 1 and 5, P = 0.037; 1 and 9, P = 0.0002; 2 and 6, P = 0.872; 2 and 10, P = 0.027; 3 and 7, P = 0.511; 3 and 11, P = 0.206; 4 and 8, P = 0.546; 4 and 12, P = 0.872; 5 and 9, P < 0.0001; 6 and 10, P = 0.037; 7 and 11, P = 0.569; and 8 and 12, P = 0.411.

Figure 3

Camptothecin-induced 53BP1 foci. (A) trex2^null cells exhibit the same level of camptothecin-induced 53BP1 foci as control cells (IB10). (B) Summary of results.

Figure 4

Exposure to γ-radiation. (A) Trex2-altered cells exposed to γ-radiation. Four clones of Trex2^hTX2, Trex2^H188A, and Trex2^R167A were compared to the parental clones of trex2^null and trex2^null-2 cells (Dumitrache et al. 2009) and control cells. These experiments were done three times. Standard deviation is shown. SF, survival fraction. (B) trex2^null h2ax^−/− cells exposed to γ-radiation. Average of two h2ax^−/− trex2^null clones is shown. These experiments were done three times and standard deviation is shown. TC1, control cells. (C) trex2^null ku70^−/− cells exposed to γ-radiation. Average of two trex2^null ku70^−/− clones is shown. These experiments were done three times and standard deviation is shown. J1, control cells.

Figure 5

I-SceI–specific DSBs. (A) The HDR-specific substrate, DR-GFP. The I-SceI site interrupts the GFP sequence while the BcgI site is the wild-type sequence at this location. After generation of an I-SceI DSB, the free ends invade the adjacent truncated GFP substrate that contains the wild-type BcgI site. Nonreciprocal homologous recombination (gene conversion) restores GFP. (B) The NHEJ-specific substrate, EJ5-GFP. There are two I-SceI sites separated by a puromycin phosphotransferase (puro) selection cassette. After generation of two I-SceI–induced DSBs, distal end joining restores the GFP cassette. There are two potential outcomes. First, the I-SceI site is restored. This detects classical NHEJ (Ku70 dependent). Second, the I-SceI site is lost. This detects Alt-NHEJ (Ku70 independent). (C) trex2^null cells repaired I-SceI–induced DSBs more efficiently than control cells (AB2.2) for the DR-GFP (DR, P < 0.001, Student’s t-test) and EJ5-GFP (EJ, P < 0.0001) substrates. The mean of six to eight experiments is shown. (D) I-SceI digestion of PCR-amplified DNA from GFP⁺ cells with the EJ5-GFP substrate. After I-SceI digestion, the amplified DNA is either uncut (U) or cut with I-SceI (S).

Figure 6

The average number of spontaneous SCEs per metaphase spread (MPS). (A) Spontaneous SCEs in control (AB2.2) and trex2^null cells. Trex2 deletion reduced SCEs in control AB2.2 cells (Yates-corrected chi-square test, P < 0.0001). (B) Spontaneous SCEs in control (AB2.2), blm^−/−, and trex2^null blm^−/− cells. Trex2 deletion did not reduce SCEs in blm^−/− cells (P = 0.11). Total numbers: AB2.2, 374 SCEs from 1965 chromosomes (61 metaphase spreads); trex2^null, 354 SCEs from 4020 chromosomes (106 metaphase spreads); blm^−/−, 1348 SCEs from 1900 chromosomes (50 metaphase spreads); and trex2^null blm^−/−, 1348 SCEs from 1894 chromosomes (50 metaphase spreads).