ATM-independent, high-fidelity nonhomologous end joining predominates in human embryonic stem cells

Abstract

We recently demonstrated that human embryonic stem cells (hESCs) utilize homologous recombination repair (HRR) as primary means of double-strand break (DSB) repair. We now show that hESCs also use nonhomologous end joining (NHEJ). NHEJ kinetics were several-fold slower in hESCs and neural progenitors (NPs) than in astrocytes derived from hESCs. ATM and DNA-PKcs inhibitors were ineffective or partially effective, respectively, at inhibiting NHEJ in hESCs, whereas progressively more inhibition was seen in NPs and astrocytes. The lack of any major involvement of DNA-PKcs in NHEJ in hESCs was supported by siRNA-mediated DNA-PKcs knockdown. Expression of a truncated XRCC4 decoy or XRCC4 knock-down reduced NHEJ by more than half suggesting that repair is primarily canonical NHEJ. Poly(ADP-ribose) polymerase (PARP) was dispensable for NHEJ suggesting that repair is largely independent of backup NHEJ. Furthermore, as hESCs differentiated a progressive decrease in the accuracy of NHEJ was observed. Altogether, we conclude that NHEJ in hESCs is largely independent of ATM, DNA-PKcs, and PARP but dependent on XRCC4 with repair fidelity several-fold greater than in astrocytes.

Figure 1.. Description of the NHEJ-red repair cassette and processing of I-SceI-digested DNA.

(A) Schematic of the NHEJ-red cassette. (B) Oct3/4 (green) positive hESCs display DsRed (red) 48 h after infection with a multiplicity of infection (MOI) of 30 with Ad-SceI adenovirus. DAPI shows nuclear staining.

Figure 2.. Repair by NHEJ monitored by genomic DNA qPCR.

(A) Time course exhibiting an increase in SYBR green fluorescence after amplification by qPCR in hESCs (left panel). Polyacrylamide gel showing the NHEJ repair product at ~125 base pair fragment at the indicated times (right panel). (B) Relative NHEJ levels after infection with Ad-SceI adenovirus with 30 MOI at 24 h. Fold (x) and statistical significance indicates changes in the relative repair levels when compared to the Ad-SceI infected sample. The difference in increases in the relative quantity of NHEJ at 27 h in (A) compared to 24 h in (B) is mostly due to a difference in the values obtained from the samples without I-SceI between the two data sets.

Figure 3.. NHEJ occurs with faster kinetics after terminal differentiation.

(A) hESCs, NPs and astrocytes were seeded and 12 h later infected with Ad-SceI at an MOI of 100. Expression of HA-tagged I-SceI was examined in samples harvested 24 h after infection. (B) BG01V/-, NP/-, and astrocyte/NHEJ-red cells were infected with Ad-SceI and collected 24 h later. (Columns) Relative NHEJ levels were determined by genomic DNA qPCR and normalized to β-actin levels; (Error bars) SEM for data sets n = 3. Fold (x) indicates changes in the relative repair levels when compared to the hESC sample. *p < 0.05: **p < 0.01; ***p < 0.001. (C) (Top Panel) BG01V/-, NP/-, and astrocyte/NHEJ-red cells were infected with Ad-I-SceI at an MOI of 30. DsRed events were determined by FACS 48 and 72 h after infection. Fold (x) and statistical significance indicates changes in the relative repair levels when compared to the hESC sample. (Columns) % DsRed+ cells with 60,000 events collected; (Error bars) SEM for three independent experiments. (Bottom Panel) Representative FACS images of DsRed+ cells at 72 h after infection.

Figure 4.. Specific DNA-PKcs and ATM kinase inhibitors become more effective as hESCs differentiate.

(A) DNA-PKi and ATMi are functioning in hESCs. hESCs were harvested 5, 10, and 15 min after exposure to 6 Gy with or without ATMi (10 μM) and DNA-PKi (2.5 μM) or both. Drugs were added 15 min prior to radiation. Fold change depicts phosphorylation of KAP1 (S824) and H2AX (S139) after normalization to CHK1 (and GAPDH) which served as loading controls. (B) BG01V/NHEJ-red (C) NP/NHEJ-red and (D) astrocyte/NHEJ-red cells were infected with Ad-SceI and then treated with either ATMi at 10 μM or DNA-PKi at 2.5 μM 1 h after infection. Cells were collected at 24 h post-infection. (Columns) Relative NHEJ levels were normalized to β-actin; (Error bars) SEM for data sets n = 3. Fold (x) indicates changes in the relative repair levels when compared to the hESC sample. Differences in the scale of the separate cell populations (B-D) are due to variation in the uninfected sample PCR amplification from 3 separate experiments. Statistical significance of differences in NHEJ with respect to cells expressing I SceI with no inhibitor, are indicated.

Figure 5.. DNA-PKcs knockdown partially reduces NHEJ in hESCs.

(A) Western blot showing DNA-PKcs expression 48 and 72 h after transfection of BG01V cells with GFP control siRNAs or siRNAs targeting DNA-PKcs or ATR [5]. The fold change in DNA-PKcs was calculated after normalization to ATR which served as a loading control together with a non-specific (N.S.) band. (B) Western blot showing HA-SceI levels in BG01V cells 48 h after infection which occurred 48 h after knockdown. (C) BG01V/NHEJ-red cells were infected with Ad-I-SceI at 30 MOI, 48 h after knockdown. DsRed events were determined by FACS 48 h after infection. (Columns) % DsRed+ cells with 10,000 events collected; (Error bars) SEM for data sets n = 3. (D) BG01V/NHEJ-red cells were infected with Ad-I-SceI at an MOI of 30 48 h after knockdown. Cells were collected at 24 h post-infection. (Columns) Relative NHEJ levels were determined by genomic DNA qPCR and normalized to β-actin levels; (Error bars) SEM for three samples. Fold (x) and statistical significance indicates changes in the relative repair levels compared to the siGFP sample.

Figure 6.. XRCC4 knockdown and expression of a XRRC4 decoy partially reduces NHEJ in hESCs

(A) XRCC4 knockdown and NHEJ in hESCs. Western blot analysis of extracts with XRCC4 antibody was carried out 48 and 72 h after transfection of BG01V/NHEJ-red cells with non-targeted control siRNAs or siRNAs targeting XRCC4. The fold change in XRCC4 levels was calculated after normalization to GAPDH which served as a loading control. (B) BG01V/NHEJ-red cells were infected with Ad-I-SceI at 30 MOI, 48 h after knockdown. Cells were collected at 24 h post-infection for genomic DNA qPCR to determine repair. (C) XRCC4 decoy reduces NHEJ in hESCs. Immunocytochemistry (top panel) and western blot (bottom panel) of BG01V/NHEJ-red cells 48 h after infection with the Ad-Flag-XRCC4_115-293 virus described previously [31], or an EGFP expressing adenovirus. (D) BG01V/NHEJ-red cells were infected with either adenovirus for 48 h and then infected with Ad-SceI and harvested 24 h later. (Columns) Relative NHEJ levels were determined by qPCR and normalized to β-actin levels (Error bars) SEM of three samples. Fold (x) and statistical significance indicate changes in the relative repair levels as compared to those in the I-SceI-expressing cells treated with non-targeting control siRNA.

Figure 7.. High-fidelity NHEJ decreases through differentiation

BG01V/-, NP/-, and astrocyte/NHEJ-red cells were infected with Ad-SceI and collected 24 h after infection. DNA was amplified with Amplitaq Gold and was digested with PsiI endonuclease where indicated. (Columns) High-fidelity NHEJ levels were determined by the relative level of the digested portion (PsiI-sensitive) of the PCR DNA fragment as a fraction of uncut DNA; (Error bars) SEM for three samples. Fold (x) indicates changes in relative repair levels when compared to the astrocyte samples.

Figure 8.. PARPi functions in hESC and induces repair foci but does not affect NHEJ

(A) PARPi inhibits PARP in hESCs. Images (top panel) and graphical depiction (bottom panel) of γ-H2AX foci in hESCs and NPs aftercells were treated with PARPi at 3 μM for 16 h. (B) PARPi does not affect NHEJ in hESCs. BG01V/NHEJ-red cells were treated with PARPi for 16 h and collected at 24 h after Ad-SceI infection. (Columns) Relative qPCR levels was normalized to β-actin; (Error bars) SEM of three samples. No statistical significance was found between treated and untreated sample.