It is important to understand the cellular responses occurring in normal brain during chemo- and radiotherapy, to counter possible adverse effects, and increase the therapeutic ratio to improve the treatment of brain cancer. Neural cells derived from hESCs provide an excellent in vitro experimental system for studying the DDR in normal human brain. It is equally important to fully comprehend how genomic stability can be preserved in hESCs during in vitro propagation for future, safe applications in regenerative medicine. Herein, we show that the cellular response to DSBs changes as hESCs differentiate into non-cycling astrocytes, and have begun to characterize this dynamic relationship as it relates to the dependence of DNA damage sensors and effects on DSB repair.
Current understanding of the DDR in hESCs is that these cells have a reduced G1 phase and may lack the G1/S checkpoint altogether 
. Glioma cells show delayed resolution of γ-H2AX foci at later times after irradiation if they lack p53 and, hence, the G1/S checkpoint 
. Thus, one explanation for the slow resolution of foci in hESCs may be more robust cell cycle checkpoints (other than G1/S) and slower DDR recovery 
. Our results are consistent with previous reports suggesting that hESCs have a robust and active G2/M checkpoint and an absent G1/S checkpoint, which first appears in NPs after differentiation of hESCs 
. A recent report showed that KU-55933 was ineffective at inhibiting the G2 checkpoint after irradiation in hESCs unless extremely high concentrations were used despite the fact that phosphorylation of T68-CHK2, S15-p53 were significantly reduced with 10 µM 
. We show that KU-55933 is effective in hESCs at lower concentrations typically used to block the DDR in tumor cell lines since radiation-induced CHK2 and H2AX phosphorylation was inhibited 
. This consistent effect of KU-55933 on hESC would thwart the notion of any significant effect of a drug pump 
. The reason why we do not see any effect of KU-55933 on the formation of p-(S1981)-ATM foci after radiation of hESCs is most likely because this antibody recognizes a number of other DDR proteins phosphorylated by ATM and ATR, such as SMC1, in addition to auto-phosphorylated ATM 
. If ATR takes ATM's place SMC1 and other targets would still be phosphorylated. Thus, ATM seems to play a role in hESCs but does not affect DSB repair, suggesting that in these rapidly proliferating cells another PIKK besides ATM is responsible for regulating DSB repair. ATR is a likely candidate as we have shown herein. Recently, through genetic manipulation by targeted allele disruption an ATM knockout hESC line was established 
. These cells demonstrated a reduction of ATM downstream signaling after radiation including reduced phosphorylation of H2AX (S139), and CHK2 (T68), but no genetic instability by CGH analysis. Thus, these results are in line with our finding that ATM does not play a critical role in DSB repair in hESCs.
The initial number of γ-H2AX foci after irradiation was greater in hESCs compared to those in either NPs or astrocytes. Previously, when the cell cycle was correlated with the size of repair foci, it was found that irradiation of S-phase cells generated a greater number of smaller IRIF believed to represent a collapse of the replication machinery at sites of single-strand breaks 
. Thus, because a large fraction of hESCs is in S-phase at a given time, it is possible that a larger number of foci are seen in these cells compared to NPs and astrocytes. Furthermore, later times after irradiation showed more residual foci in hESCs in our study. The most reasonable explanation for this finding is that these cells depend largely on slower, more complex, DSB repair than either NPs or astrocytes.
Our results show that HRR is highly active in hESCs and declines as cells differentiate into NPs, and disappear entirely after differentiation into astrocytes. hESCs display a higher percentage of cells with RAD51 foci at early times after irradiation when compared to NPs, indicating more robust HRR, whereas in astrocytes, no RAD51 foci were seen. In support of this finding, we found that the expression of the 37-kDa form of RAD51 gradually decreased along the hESC-NP-astrocyte axis. Furthermore, a larger form of RAD51 was found in hESCs, which could be the result of alternative splicing, and RAD51 in astrocytes seems to undergo post-translational processing, both of which could affect HRR. A previous report showed that cleavage of RAD51 to a 31-kDa protein occurs in cells undergoing apoptosis 
. Thus, the smaller RAD51 species seen in the extract from astrocytes could perhaps be an inactive form of RAD51. Altogether, the level of full-length RAD51, and possible RAD51 processing could perhaps represent different layers of HRR control in these cell populations.
Previous work established that ATM is inhibited by caffeine more readily than ATR 
. We show that caffeine at doses ≥4 mM inhibit IRIF formation in hESCs. In line with this finding, knockdown of ATR reduced γ-H2AX and RAD51 IRIF supporting the idea that ATR is critical for HRR in hESCs. Furthermore, we noticed a significant increase in the level of γ-H2AX foci in unirradiated cells after ATR knockdown which seems to also occur in ATR-deficient mouse cells due to increased replicative stress induction of DSBs 
. If hESCs only require ATR but not ATM for DSB repair what could be the role of ATM in these cells? It is possible that ATR-dependent HRR is so dominant in hESCs compared to ATM-dependent repair and thus cannot be detected. It is also possible that ATM kinase targets, such as CHK2 and p53, are either sequestered or not fully functioning in hESCs 
. ATM could also be critical for apoptosis and for maintaining the progenitor cell phenotype 
. Regardless whether ATM is able to phosphorylate these proteins and KU-55933 able to inhibit or not, ATM does not seem to function the same way in hESCs as in somatic cells. Furthermore, regardless whether classical NHEJ occurs in hESCs or not, our results suggest that repair is largely, if not completely, DNA-PKcs-independent. Recently, the role of DNA-PKcs in NHEJ in mouse and human ESCs was examined 
. Whereas it was clearly demonstrated that in mouse ESCs DNA-PKcs is important for NHEJ it is unclear whether this is also the case in hESCs.
Since astrocytes seem to lack HRR, it should be possible to use these cells for examining casual relationships of proteins exclusively required for NHEJ. Indeed, we show that ATM kinase activity is strictly required for the formation and maintenance of repair foci in these cells. If KU-55933 was added prior to irradiation, a decrease of almost 90% in the number of γ-H2AX foci was seen, whereas when KU-55933 was added to cells after the formation of foci, they disassembled rapidly. There is a remote possibility that this represents faster than normal repair, however, this is highly unlikely since IRIF are not seen if the drug is added before irradiation. Furthermore, many, smaller IRIF remained in the KU-55933-treated cells which were only noticeable at high-power magnification. These foci were not resolved even at later times and eventually outnumbered the foci seen in the untreated control. Our results suggest that ATM plays a critical role in the formation and maintenance of repair foci in astrocytes. Furthermore, treatment of astrocytes with KU-57788 in part inhibited the rate of IRIF resolution. This finding indicates that while DNA-PKcs is involved to some extent in the repair of DSBs, ATM is vital for the initial signaling, formation and maintenance of repair complexes necessary for efficient DSB repair.
In summary, we have demonstrated that hESCs depend on HRR to a large extent whereas astrocytes do not and instead appear to exclusively utilize repair other than HRR. Importantly, hESCs requires ATR for HRR and not ATM, whereas astrocytes critically require ATM to keep repair factors assembled at DSBs and facilitate DSB repair. If astrocytes use classical NHEJ it is largely DNA-PKcs dependent. The contribution of NHEJ to DSB repair in hESCs is currently being investigated.