|Home | About | Journals | Submit | Contact Us | Français|
Prevention of cranial radiation-induced morbidity following the treatment of primary and metastatic brain cancers, including long-term neurocognitive deficiencies, remains challenging. Previously, we have shown that inhibition of glycogen synthase kinase 3β (GSK3β) results in protection of hippocampal neurons from radiation (IR)–induced apoptosis and attenuation of neurocognitive dysfunction resulting from cranial IR. In this study, we examined whether regulation of the repair of IR-induced DNA damage is one of the mechanisms involved in the radioprotective effects of neurons by inhibition of GSK3β. Specifically, this study showed that inhibition of GSK3β accelerated double strand-break (DSB) repair efficiency in irradiated mouse hippocampal neurons, as assessed by the neutral comet assay. This coincided with attenuation of IR-induced γ-H2AX foci, a well characterized in situ marker of DSBs. To confirm the effect of GSK3 activity on the efficacy of DSB repair, we further demonstrated that biochemical or genetic inhibition of GSK3 activity resulted in enhanced capacity in nonhomologous end-joining–mediated repair of DSBs in hippocampal neurons. Importantly, none of these effects were observed in malignant glioma cells. Taken together, these results suggested that enhanced repair of IR-induced DNA damage may be a novel mechanism by which inhibition of GSK3β specifically protects hippocampal neurons from IR-induced apoptosis. Furthermore, these findings warrant future investigations of the molecular mechanisms underlying the role of GSK3β in the DSB repair of normal neurons and the potential clinical application of neuroprotection with GSK3β inhibitors during cranial IR.
Cranial irradiation (IR) often has long term or permanent neurocognitive effects, especially in the pediatric population.1,2 Some of these documented effects include intellectual impairment, reduction in performance IQ, memory loss, and dementia.3–6 We and others have reported that these effects may be due to IR-induced damage to the hippocampus, the functional center of the brain responsible for learning and memory.1,2,7 In addition, IR to the hippocampus has been linked with more significant cognitive deficits when compared with other parts of the brain.8 Novel strategies are thus needed to protect hippocampal neurons from radiation-induced damage to enhance the therapeutic index and alleviate the significant long-term toxicities of treatment.
Our previous studies have shown that glycogen synthase kinase-3β (GSK3β) is involved in IR-induced hippocampal neuronal apoptosis and subsequent neurocognitive decline.7,9 Specifically, we reported that IR activated GSK3β signaling in hippocampal neurons, as assessed by reduction of β-catenin and cyclin D1 levels following IR.9 These are well-known substrates of GSK3β that are targeted for proteolytic degradation after phosphorylation by GSK3β and thus served as biomarkers for GSK3β activity.10,11 Importantly, suppression of GSK3β-attenuated IR-induced hippocampal neuron apoptosis and protected neurocognitive function in irradiated mice.9,12
IR generates multiple forms of DNA damage, the most toxic of which is the chromosomal double-strand break (DSB). Even a single unrepaired DSB can be lethal.13–15 We therefore hypothesized that protection of hippocampal neurons from IR-induced apoptosis is, at least in part, due to enhanced repair of IR-induced DSBs. Repair of IR-induced DSBs occurs predominantly via the fast and efficient nonhomologous end-joining (NHEJ) DNA repair pathway.20,21 NHEJ occurs through all phases of the cell cycle, is dependent on the protein complex consisting of the DNA-dependent protein kinase catalytic subunit (DNA-PKcs), the Ku 70/80 heterodimer, and the XRCC4-ligase IV, and it ultimately rejoins the ends of DSBs with little or no homology. Functional assessment of the DNA-PKcs/Ku70/80/XRCC4/LigIV-dependent NHEJ activity has been previously established by the measurement of the efficiency of DSB end-rejoining in vitro and in vivo.16–21 Consistent with the role of NHEJ in the repair of IR-induced DSBs, cells deficient in any NHEJ protein have been shown to be hypersensitive to IR-mediated cytotoxicity.22–25
We recently have shown that lithium attenuates IR-induced apoptosis in neurons but not cancer cells by enhancing DNA repair through the NHEJ pathway.12 Lithium, however, is nonspecific, and the target(s) by which lithium enhances NHEJ-mediated DNA repair is not defined. One well-known target of lithium is GSK3β, a multifunctional serine/threonine kinase that is ubiquitously expressed in eukaryotic cells. It plays a pivotal role in a number of major processes, including embryonic development, cell differentiation, apoptosis, and insulin response.26–30 Specifically, in neuronal cells, inhibition of GSK3β has been shown to protect neural progenitor cells from apoptosis induced by genotoxic and other stresses.31,32 In addition, DNA damage was shown to activate GSK3β in neural cells.33
We have previously shown that inhibition of GSK3β attenuates IR-induced apoptosis of hippocampal neuroprogenitor cells and protects neurocognitive function in mice treated with cranial IR.9 In this study, we found a novel role of GSK3 in IR-induced DNA damage repair in hippocampal neurons. In particular, inhibition of GSK3 activity with the specific GSK3 inhibitors SB216763 and SB415286 accelerated the repair of IR-induced DSBs in irradiated neurons and hippocampus. Furthermore, inhibition of GSK3 enhanced the capacity of NHEJ-mediated repair of DSBs in treated hippocampal neurons. These observations were validated genetically by expressing a dominant negative GSK3β kinase–inactive mutant (GSK3βKI)9 and by silencing GSK3β expression using shRNA in hippocampal neurons. Importantly, none of these effects were observed in malignant glioma cells. Taken together, these findings revealed one novel mechanism by which inhibition of GSK3β attenuates IR-induced apoptosis of hippocampal neurons through enhanced end-joining of DSBs generated by IR.
The mouse hippocampal neuronal cell line HT22 was obtained from David Schubert (The Salk Institute; La Jolla, CA) and maintained in DMEM with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (Life Technologies). The mouse hippocampal neuronal cell line HN33 is an immortalized cell line derived from the somatic cell fusion of mouse hippocampal neurons and N18TG2 neuroblastoma cells.34–37 These cells were maintained in DMEM with 10% FBS and 1% penicillin/streptomycin.
To identify shRNA sequences that could knock down GSK-3β in HT22 hippocampal neuronal cells, we screened 5 MISSION shRNA clones NM_019827.2-1527s1c1 (Sigma) targeted against the mouse GSK-3β sequence. The effect of the shRNA clones on GSK-3β protein levels in HT22 cells was assayed by immunoblotting using anti-GSK-3β antibody (Cell Signaling). The clone (shHT22) that resulted in the greatest knockdown of GSK-3β was selected (targeting sequence 5′-CATGAAAGTTAGCAGAGATAA-3′), and a clone (cHT22) that gave no significant knockdown of GSK-3β (targeting sequence 5′-CGAGAAGAAAGATGAGGTCTA-3′) was used as a control.
The mouse glioma cell line GL261 and human glioma cell line D54 were obtained from Dr. Yancie Gillespie (University of Alabama-Birmingham) and maintained in DMEM with Nutrient Mixture F-12 1:1, 10% FBS, 1% sodium pyruvate, and 1% penicillin/streptomycin (Life Technologies). All cells were grown in a 5% CO2 incubator at 37°C. For irradiation of cells, the Mark I 137Cs irradiator (J.L. Shepherd and Associates) was used, delivering 1.84 Gy/min. A turntable ensured that the radiation was equally distributed. As a control, mock irradiation consisted of placing plates containing cells in the irradiator for the designated times without turning on the machine.
The GSK3 inhibitors SB216763 [3-(2,4-dichlorophenyl)-4-(1-methyl-1H-indol-3-yl)-1H-pyrrole-2,5-dione] and SB415286 [3-[(3-chloro-4-hydroxyphenyl) amino]-4-(2-nitrophenyl)-1 H-pyrrole-2,5-dione] were purchased from Tocris Biosciences.
Mouse wild-type GSK3β (pPRIG GSK3βWT) and dominant negative kinase-inactive GSK3β (pPRIG GSK3βKI) were constructed as previously described.9 In brief, GSK3βWT or GSK3βKI was subcloned into pPRIG, which is a bicistronic construct obtained from Dr. Patrick Martin (Université de Nice, Nice, France38). pdsRed2-N2 and peGFP-N1 were purchased from Clonetech Laboratories.
Cells were cultured and mounted onto sterile glass slides at 30% confluency. They were then treated with vehicle or 10 µM SB216763 for 16 h (timing and dose determined previously9). Following the treatment period, cells were exposed to either mock radiation or 3 Gy radiation using Mark I 137Cs irradiator (J.L. Shepherd and Associates) delivering 1.84 Gy/min. After the specified time point, immunohistochemistry for γ-H2AX was performed as described elsewhere.9,12 Rabbit anti-phospho-γ-H2AX antibody (Millipore) was used as the primary antibody at 1:500 dilution. Secondary antibody was 1:1000 anti-rabbit Alexa488 conjugated antibody (Invitrogen). Total cells were counted under a fluorescent microscope (original magnification, ×40; Carl Zeiss, Thornwood, NY), and cells containing >10 foci were scored for positive. At least 500 cells were counted.
For analysis of γ-H2AX foci in mouse hippocampal neurons, C57/BL/6J mouse pups were obtained from The Jackson Laboratory. SB216763 dissolved in DMSO at 0.6 mg/kg were given to mouse pups via intraperitoneal injection (i.p.) injection on postnatal days 7–10 daily for 3 consecutive days. DMSO was injected to vehicle-control mice. After this, mice were anesthetized, restrained, and treated with the 6 Gy cranial irradiation using a Therapax DXT 300 x-ray machine (Pantak) delivering 2.04 Gy/min at 80 kVP. For histological staining, mice were sacrificed 1 h after irradiation by cervical dislocation under isoflurane anesthesia. Immunohistochemistry for γ-H2AX was performed as described elsewhere.9,12 All animal procedures were approved by the Vanderbilt University Institutional Animal Care and Use Committee.
HT22 and HN33 neurons, and GL261 and D54 glioma cancer cells were seeded and treated with 10 µM SB216763 or vehicle for 16 h. Following the treatment period, cells were exposed to 3 Gy or mock irradiated and were incubated for the specified time points. Cells were then prepared and subjected to neutral comet assay following the manufacturer's instructions (Trevigen).12 Cells were then visualized using fluorescent microscopy (Carl Zeiss; Thornwood, NY). At least 100 cells were analyzed for each time point using the Comet Score software (TriTek Corp., Sumerduck, VA). Experiments were repeated in triplicate.
Analysis of NHEJ was performed as previously described.12 In brief, cells were treated with vehicle (DMSO), 10 µM SB216763, or 10 µM SB415286 for 16 h prior to transfection. A single DSB was generated in the plasmid substrate pEGFP-N1 by cleavage between the promoter and GFP reporter gene with EcoRI (Figure 3A). Linearized DNA was gel purified for cell transfection. Cells were cotransfected with a pdsRedN2 as a transfection efficiency internal control and cleaved substrate using Fugene6 (Roche). Cells were harvested 48 h later and subjected to 2-color fluorescence analysis. The green fluorescent cells represent the repaired DSB and restoration of the GFP expression. The red fluorescent cells represent the exogenous DNA transfection efficiency. No significant difference in transfection efficiency was observed between the different treatment groups. For each analysis, 100,000 cells were processed. The relative DSB rejoining activity was obtained by the ratio of green fluorescent cells to red fluorescent cells. At least 3 independent experiments were performed.
For pPRIG constructs with wild-type and kinase-inactive GSK3β, cells were first transfected with the various pPRIG plasmids alone. Twenty-four hours later, linearized pdsRed-N2 was transfected into the cells. A separate circularized pdsRed-N2 transfection was also performed as a transfection efficiency control, which revealed no differences in transfection efficiency due to the expression of the GSK3β constructs (data not shown). The relative DSB rejoining activity was obtained by the ratio of red fluorescent cells to green fluorescent cells. At least 3 independent experiments were performed.
Whole cell lysates were prepared using radioimmunoprecipitation lysis buffer (150 mM NaCl, 50 mM Tris, pH 8.0, 5 mM EDTA, 0.5% sodium deoxycholate, 0.1% SDS, 1.0% Nonidet P-40) with protease and phophatase inhibitor cocktails (Sigma) and subjected to SDS-PAGE analysis. GSK3β (Cell Signaling), β-catenin (Cell Signaling), and actin (Santa Cruz) levels were analyzed by Western blot.
The mean and standard deviation of each treatment group were calculated for all experiments. Statistical analysis was done using 2-way analysis of variance followed by a Bonferroni posttest using GraphPad Prism, version 4.02 for Windows (GraphPad Software).
The most critical lesion generated by IR is the DNA DSB. We previously found that lithium prophylaxis prior to IR protects hippocampal neurons from IR-induced apoptosis in part through the enhancement of the repair of IR-induced DSBs.12 However, the target(s) by which lithium enhances repair is unknown. We have also previously demonstrated that the activity of GSK3β, one of the major targets of lithium, is altered in irradiated hippocampal neurons, as determined by reduction of β-catenin and cyclin D1 levels after IR.9 These are well-known substrates of GSK3β that are targeted for proteolytic degradation after phosphorylation by GSK3β and thus served as biomarkers for GSK3β activity.10,11 Furthermore, we have also demonstrated that suppression of GSK3β attenuated IR-induced hippocampal neuron apoptosis and protected neurocognitive function in irradiated mice.9,12 We thus hypothesized that the neuroprotective effects of GSK3β inhibition may occur, in part, through enhanced repair of IR-induced DSBs.
To test this hypothesis, we first analyzed DSBs generated by 3 Gy IR at various time points using the neutral comet assay12,39,40 in irradiated hippocampal neurons, with or without the GSK3 inhibitor SB216763. We have previously determined the optimal dose and timing of treatment with this inhibitor in conjunction with irradiation (3 Gy) in hippocampal neurons to achieve alteration of GSK3 activity in phosphorylating its substrates as well as the protective effect on IR-induced HT22 apoptosis.9,41 The maximal effective concentration of this inhibitor was found to be ~10 µM, and the optimal time of treatment was 16 h.9 The concentrations used have also been shown to act specifically on GSK3.41
Using the comet assay, cells with DSBs are evident by the presence of the characteristic “comet tail,” whereas cells without DSBs remain round (Figure 1A). The addition of the inhibitor alone did not alter the basal mean tail moment (Figure 1B) or the percentage of cells with comet tail (data not shown). Comparable levels of IR-induced DSBs were observed in both the inhibitor- and vehicle-treated HT22 (Fig. 1B) and HN33 (Fig. 1C) hippocampal neurons 15 min after IR, as calculated by the mean tail moment.39,40 Similarly, the percentage of cells with a comet tail were comparable 15 min after IR (data not shown). However, at 30 min following IR, repair of IR-induced DSBs was significantly accelerated in GSK3 inhibited HT22 and HN33 neurons, compared with vehicle-treated cells. By 8 h after IR, the level of DSBs has returned to baseline. These findings suggest that specific inhibition of GSK3 does not influence the initial formation of DSBs after IR, but it does accelerate the resolution of persistent DSBs in irradiated hippocampal neurons (Figure 1B and 1C).
DSBs can also be detected in cells via γ-H2AX foci, which are well characterized, widely used in situ markers for DNA DSBs (Figure 2A).20,21 To validate an effect of GSK3 inhibition on IR-induced DSB resolution, we assessed IR-induced γ-H2AX foci in hippocampal neurons with or without GSK3 inhibition. Similar to our observations with the neutral comet assay, treatment of HT22 neurons with SB216763 alone without IR did not affect the percentage of cells with γ-H2AX foci. When combined with IR, however, SB216763 significantly reduced the number of cells demonstrating γ-H2AX foci as early as 30 min after IR when compared with the vehicle (Figure 2B). A maximal 5-fold reduction—31% to 6%— was observed in the number of cells with elevated γ-H2AX foci 1 h after IR in the inhibitor treated cells versus the vehicle. This effect persisted even 4 h after IR. By 8 h following 3-Gy IR, basal DSB levels were achieved. Similar effects were also observed in HN33 neurons (Fig. 2C). These data further support the notion that neuroprotection of hippocampal neurons by GSK3 inhibition involves the accelerated repair of IR-induced DSBs.
We previously demonstrated that GSK3β inhibitors stabilize β–catenin and reduce apoptosis in the hippocampus of irradiated mice.9 To validate the effect of GSK3 inhibition on repair of IR-induced DNA damage in vivo, we next examined DSB repair of hippocampus by examining IR-induced γ-H2AX foci in hippocampal neurons of cranial-irradiated mice with or without GSK3 inhibition. Consistent with our findings in cultured hippocampal neurons, IR induced γ-H2AX foci in both vehicle and GSK3 inhibited mice. Importantly, inhibition of GSK3 accelerated the resolution of the elevated γ-H2AX foci 1 h after IR (44.6% vs. 13%; P < .01) (Figure 2D). A representative image of IR-induced γ-H2AX foci in the hippocampus can be found in Supplementary Figure 1. These results validate our finding in vitro that GSK3 inhibition accelerates the resolution of IR-induced DSBs.
Our results thus far revealed that GSK3 inhibition increases DNA DSB repair in irradiated hippocampal neurons. To determine whether this enhanced DSB repair is through an enhanced NHEJ mechanism, we utilized a well-established plasmid-based in vivo biochemical DSB end-joining assay that measures intramolecular NHEJ.12,42 This episomal nonreplicating plasmid reporter assay for end-joining has been shown to represent in vivo end-joining capacity of cells.43
In these experiments, a single DSB is generated in vitro in the plasmid substrate pEGFP-N1 between the promoter and the green fluorescent protein (GFP) reporter gene, which prevents the expression of the reporter in vivo. The linearized plasmid was subsequently co-transfected with the circularized plasmid pdsRed2-N2, which served as an internal transfection efficiency control. GSK3 inhibition did not significantly alter transfection efficiency, which was consistently 7%–10% of cells. Intracellular recircularization of the linear pEGFP-N1 plasmid by NHEJ-mediated end-joining allowed the expression of GFP, as depicted in Figure 3A. This was then measured using 2-color flow cytometric analysis. Again, the optimal dose and timing of treatment with this inhibitor for the kinase activity of GSK3 were determined in our previous studies.9 As shown in Figure 3B, a significant 2.5-fold enhanced level of rejoining was observed in SB216763 treated HT22 hippocampal neurons when compared with the vehicle. A similar effect on end-joining was observed using another specific GSK3 inhibitor, SB415286 (data not shown). Furthermore, an enhanced NHEJ capacity after GSK3 inhibition was also observed in another hippocampal neuron model, HN33 (Figure 3C). These results indicate that GSK3 inhibition may remove IR-induced DSBs by enhancing NHEJ-mediated repair of DSBs.
DNA end-joining–mediated repair occurs throughout the cell cycle, but is the predominant DSB pathway in cells in the G1 phase of the cell cycle. Because the GSK3β kinase is involved in cell proliferation pathways, it is possible that the enhanced end-joining capacity in GSK3 inhibited neurons may be due to increased cellular accumulation in the G1 phase of the cell cycle. We thus investigated the cell-cycle distribution of cells with and without GSK3 inhibition. As shown in Supplementary Figure 2, inhibition of GSK3 for 16, 24, or 48 h did not alter cell-cycle distribution of HT22 hippocampal neurons. A similar lack of cell-cycle redistribution was also observed in HN33 neurons and GL261 cancer cells (data not shown). These results suggest that the enhanced end-joining capacity in GSK3 inhibited neurons is not due to redistribution of cells in the cell cycle.
The data thus far suggest that biochemical inhibition of GSK3 enhances NHEJ-mediated repair of IR-induced DSBs in hippocampal neurons. To validate that this effect is specific to GSK3β inhibition, we assessed the effect of GSK3β on DNA repair with genetic approaches by silencing GSK3β via stably expressing shRNA specifically against GSK3β (shHT22), as well as inhibiting GSK3β kinase activity via expressing a dominant negative kinase-dead mutant of GSK3β (GSK3βKI) in HT22 hippocampal neurons.
To generate shHT22 cells, we first screened 5 MISSION shRNA clones targeted against the mouse GSK-3β sequence. The effect of the shRNA clones on GSK-3β protein levels in HT22 cells was assayed by immunoblotting. The clone (shHT22) that resulted in the greatest knockdown of GSK-3β was selected (targeting sequence 5′-CATGAAAGTTAGCAGAGATAA-3′), and a clone (cHT22) that gave no significant knockdown of GSK-3β (targeting sequence 5′-CGAGAAGAAAGATGAGGTCTA-3′) was subsequently used as a control.
As shown in Figure 4A, expression of GSK3β was not detected in shHT22 neurons but was seen in parent HT22 and HT22 neurons stably expressing a control shRNA (cHT22). In addition, because GSK3β targets β-catenin for degradation, shHT22 neurons exhibited increased β-catenin levels, compared with control cells as a consequence of reduced GSK3β expression (Figure 4A).
Having established a neuronal model with suppressed GSK3β expression, we next assessed the DSB end-joining repair capacity of these cells using the aforementioned in vivo intra-molecular DSB end-joining assay. As shown in Figure 4B, DSB end-joining capacity was significantly enhanced in shHT22 cells with GSK3β silencing, compared with control cHT22 cells. Furthermore, chemical inhibition of GSK3 enhanced end-joining capacity in cHT22 neurons, consistent with the observation in HT22 (Figure 3B), but did not further increase end-joining capacity in shHT22 neurons. These results support a role of GSK3β in modulating end-joining repair capacity in hippocampal neurons. Consistent with our previous findings, shHT22 cells with silencing of GSK3β exhibited a significant increased resistance to IR-induced apoptosis compared to cHT22 cells (Dr. D. E. Hallahan, personal communication).
If the kinase activity of GSK3β is required for regulatory suppression of NHEJ function, we next reasoned that a similar enhanced DSB repair should be observed in cells expressing a dominant negative kinase-dead mutant of GSK3β (GSK3βKI), which has been shown to interfere with the endogenous GSK3β kinase activity previously.9 To test this, GSK3βWT, GSK3βKI, or vector control was ectopically expressed in HT22 and HN33 hippocampal neurons, and a similar intramolecular DSB end-joining assay was performed, because this well-established assay directly measures in vivo intramolecular NHEJ.12,42 In this assay, cells expressing the various GSK3β proteins can be detected through GFP expression via the parent bicistronic pPRIG vector, which allows for expression of any cDNA in the first cistron while keeping a high level of expression from its internal ribosome entry site–dependent second cistron encoding the enhanced GFP (eGFP).9 We previously confirmed inhibition of GSK3β activity by GSK3βKI as well as the expression of the various GSK3β constructs and GFP in hippocampal neurons.9 Rejoining of linearized pdsRed2-N1 was subsequently measured (Figure 5A). Ectopic expression of these GSK3β proteins did not alter transfection efficiency during this assay, which is ~10%. Only cells with expression of exogenous GSK3β, which was comparable in all groups, were collected for assessment of DSB end-joining efficiency. As shown in Figure 5B and 5C, a similar 2–3-fold increase in end-joining activity was observed in both HT22 and HN33 hippocampal neurons expressing GSK3βKI, compared with vector alone. Overexpression of GSK3βWT had minimal effect on end-joining. This is not surprising, because we previously showed that overexpression of the protein did not correlate with increased kinase activity.9 These results indicate a role of GSK3β inhibition in the enhanced NHEJ-mediated repair of IR-induced DSBs.
Our previous studies suggest that radioprotection by GSK3β inhibition does not occur in the mouse glioma cell line GL261.9 We next determined whether the observed effects of enhanced repair of IR-induced DSBs through GSK3β inhibition were specific to normal neuronal cells and not to tumor cells. At all time points tested following 3 Gy IR, no significant difference was seen in the resolution of IR-induced DSBs as measured by the mean tail moment using the neutral comet assay in GL261 (Figure 6A) glioma cells treated with either vehicle or 10 µM SB216763. Inhibition of GSK3β also did not have a significant effect on the percentage of cells with increased levels of IR-induced γ-H2AX foci at any of the time points that were studied (Figure 6B). Furthermore, end-joining capacity did not significantly change in GL261 (Figure 6C) tumor cells following GSK3 inhibition or GSK3βKI expression (Figure 6D). Similar results were observed in the D54 glioma cell line (Supplementary Figure 3).
We previously demonstrated that the activity of GSK3β is altered in irradiated hippocampal neurons as determined by reduction of β-catenin and cyclin D1 levels following IR, and blockade of GSK3β activity attenuates these effects.9 To gain insight into the differential effect of GSK3β inhibition in normal versus tumor cells, we examined the downstream GSK3β signaling in irradiated GL261 cancer cells with and without inhibitor. Interestingly, as shown in Figure 7, neither IR nor inhibition of GSK3β increased β-catenin levels, suggesting that in these glioma cells, GSK3β signaling may be dysregulated. This may explain the differential effects in normal versus tumor cells with respect to protection from IR-induced apoptosis. Additional studies are currently ongoing to support this hypothesis.
Taken together, these data suggest that inhibition of GSK3β protects hippocampal neurons but not glioma cancer cells by promoting DSB repair of IR-induced DSBs. This differential effect of GSK3β inhibition in glioma cancer cells versus mouse hippocampal neurons further supports the potential clinical application of GSK3β inhibitors in the protection of neurological functions during cranial radiation treatment.
This study provides biochemical and genetic evidence supporting that inhibition of GSK3 enhances the repair of IR-induced DSBs in mouse hippocampal neurons and, importantly, not in glioma cancer cells. Specifically, we found that inhibition of GSK3 accelerated the resolution of persistent IR-induced DSBs as measured by the neutral comet assay and IR-induced γ-H2AX foci both in cultured hippocampal neurons and in mice hippocampus. This was associated with an enhanced DSB end-joining capacity in GSK3-inhibited neurons. These results were validated via genetic manipulation of GSK3β activity using a dominant-negative kinase inactive GSK3β mutant and shRNA-mediated silencing of GSK3β expression. These findings demonstrate that one of the potential underlying mechanisms by which GSK3β inhibition protects IR-induced apoptosis of hippocampal neurons and attenuates neurocognitive function in irradiated mice, as demonstrated previously,9 may be through enhanced end-joining-mediated repair of IR-induced DSBs.
GSK3β is enriched in the brain and has been implicated in a number of prevalent disorders, such as Alzheimer disease, schizophrenia, Parkinson disease, and bipolar disorder.44–47 Inhibition of GSK3 has been shown to alleviate some of these processes. The enhanced repair of IR-induced DSBs by GSK3 inhibition may provide a novel mechanism by which neurons can survive from insults to the brain. It remains to be seen whether the neuroprotective effects of GSK3β inhibition outside of IR-induced apoptosis is similarly due to enhanced repair of DNA damage induced by other insults.
The GSK3 family consists of both GSK3α and GSK3β. The inhibitor used in this study, SB216763 (Tocris Biosciences), inhibits both isoforms of GSK. It is possible that the effects on repair observed in this study may also include a GSK3α-mediated response. However, our experiment using a kinase inactive GSK3β mutant and specific shRNA targeting GSK3β result in a 2–3-fold increase in end-joining capacity, which is similar to inhibitor-treated cells and suggests that the effects are most likely due to inhibition of GSK3β. Nevertheless, a role of GSK3α cannot be entirely ruled out. Whether GSK3α also plays a role in regulation DNA damage repair is an interesting question and needs to be investigated in future studies.
Although the present data from the neutral comet assay and γ-H2AX foci both suggest an enhanced repair by GSK3β inhibition, the kinetics of IR-induced DNA damage/repair are slightly different (ie, maximal comet tail moment detected 15 min after IR via the neutral comet assay versus delayed γ-H2AX foci peak at 1 h after IR). However, this may be explained by the fact that, although the neutral comet assay allows for the direct measurement of IR-induced DSBs, persistent or low-level (<50 strand breaks) DNA damage can go undetected, especially at lower radiation doses.39,40,48 In addition, the comet assay can be affected by changes in chromatin structure independent of DSBs. Futhermore, γ-H2AX foci formation typically requires the recruitment of the DNA damage response proteins by the DSB, which may “delay” foci formation. Alternatively, an effect on H2AX phosphorylation or subcellular localization cannot be ruled out. However, results from the in vivo plasmid-based NHEJ assay supports the notion that the observed effects of GSK3 inhibition on comet tails and γ-H2AX foci more likely represent persistent DSBs due to suppressed NHEJ-mediated DSB repair in cells without GSK3 inhibition.
The enhanced end-joining capacity of GSK3β inhibited neurons was shown via an end-joining assay of episomes. This assay has been shown to represent the end-joining capacity of cells and to be compatible with recently developed chromosomally integrated reporter substrates.42,43 This is in contrast to homologous recombination-mediated repair (HR), which is largely dependent on DNA replication processes and thus cannot be readily measured in nonreplicating episomal plasmids. Both biochemical and genetic data support a role of GSK3β inhibition and enhanced end-joining repair of DSBs. In contrast, however, GSK3β overexpression did not reduce end-joining capacity (Figure 5B). Because GSK3β is ubiquitously expressed and control of its function is not dictated by its expression level, this result is not unexpected.
It has been well established that radiation induces DSBs which is lethal if not repaired.13–15 We have recently observed that lithium-mediated neuroprotection from IR-induced apoptosis also occurs by increasing the NHEJ pathway.12 Our results support the hypothesis that GSK3β may be the target by which lithium enhances the repair capacity of neurons. GSK3β has also been reported to regulate the transcriptional activity of p53 through phosphorylation, which is an important regulator of DNA damage–induced apoptosis.49 Interestingly, the priming event of this regulation involves DNA-PK–mediated phosphorylation of p53. Additional investigation to dissect the role of GSK3 in the cross-talk between DNA repair pathways and apoptotic pathways is warranted. In addition, whether the enhanced repair capacity through GSK3β inhibition depends on cellular p53 status is an interesting avenue of future studies.
GSK3β is an enzyme that regulates glycogen synthesis in response to insulin. It is a ubiquitously expressed serine/threonine protein kinase that is a critical downstream element of the PI3 kinase/Akt cell survival pathway. Activation of this PI3 kinase/Akt pathway results in inhibition of GSK3β and subsequent cell survival or proliferation. The results from this study suggest a potential and interesting link between metabolic pathways and response to DNA damage. It is intriguing to hypothesize a model whereby cells under metabolic stress (nutrient deprivation) will activate the prosurvival Akt pathway to not only inhibit cell death but also activate DNA damage repair pathways. In addition, whether the mechanism by which neuroprotection occurs is through direct inhibition of GSK3β versus inhibition of downstream insulin or wnt signaling pathways is an interesting future study.
Radiation treatment approaches have been designed to spare the anatomic regions in the brain—namely, the hippocampus—which contain normal neural progenitor cells.50–54 These include tomotherapy and intensity-modulated radiotherapy techniques to specifically protect the hippocampus and reduce the neurocognitive toxicity following cranial irradiation.50–54 Many have proposed that the hippocampal neuron progenitor cells are the most affected during radiation-induced neurocognitive decline.55 Consistent with these findings, IR-induced cognitive deficit involves apoptosis in the neurogenic region of the hippocampus and the subsequent diminished neurogenesis in the hippocampus.56–60 Specifically, inhibition of GSK3β has been shown to protect neural progenitor cells from apoptosis induced by oxidative and other stresses.31,32 It would be interesting to investigate whether the radioprotective effects of GSK3ß inhibition occurs in normal neural progenitor cells.
Importantly, we did not observe enhanced repair in GL261 and D54 glioma tumor cells following GSK3 inhibition. This is consistent with our previous findings that lithium and inhibitors of GSK3 did not protect cancer cells from IR-induced apoptosis.7,9,12 One possible explanation may be the already enhanced repair capacity of cancer cells. In support of this notion, glioblastoma cells with hyperactivated Akt signaling, which can inhibit GSK3β activity, exhibit increased DSB repair capacity and radioresistance.61 In addition, the downstream targets of GSK3β in glioblastoma cells were not further altered after IR or GSK3β inhibition (Figure 7), suggesting a potential explanation for the differential effect of GSK3 inhibition between hippocampal neurons and cancer. In addition, several recent studies demonstrate the heterogeneity that exists in glioma cells with regard to their response to GSK3 inhibition, from blockade of glioma cell differentiation to induction of cell death.62,63 This differential response clearly illustrates the need for further investigation of the effects of GSK3 inhibition in a larger panel of gliomas, including glioma stem-like cells, which have been implicated to contribute to treatment resistance, to determine whether cancer cells are resistant to GSK3 inhibition–mediated radioprotection.
Another possible explanation may be the p53 status in these cancer cells. We previously reported that inhibition of GSK3β in irradiated neurons prevented IR-induced p53 accumulation, suggesting a potential link between p53, GSK3β, and neuroprotection.9 Interestingly, genetic knockout of NHEJ proteins in mice has also been shown to increase p53-dependent cell death of neurons and results in deficits in neurogenesis.64–67 In the cancer cells examined in this study, p53 was dysfunctional. It is possible that GSK3β inhibition can protect p53 proficient cancer cells. Future investigations to address the contributing factors and the underlying mechanisms for the lack of protection in cancer cells are warranted. In addition, it may be important to determine the p53 status of each individual patient's cancer prior to the use of these neuroprotective compounds for protection of normal tissues.
Our previous studies reveal similar effects with lithium. Lithium, however, requires a long 7-day prophylaxis, has a narrow therapeutic window, and lacks specificity. Because of these reasons, the use of GSK3β inhibitors as neuroprotectors provides obvious advantages over lithium. Prophylaxis with GSK3β inhibitors can start as early as 16 h before commencement of cranial IR, eliminating the need to wait 1 week before initializing radiation treatment that is necessary with lithium. In addition, a theoretical increased specificity can be achieved using these inhibitors, potentially decreasing the unwanted side effects of neuroprotectors from off-target effects.
Taken together, this study provides direct biochemical and genetic evidence supporting the notion that inhibition of GSK3β increases DNA repair in irradiated noncancerous neuronal cells. This not only links GSK3β signaling to DNA repair pathways but also generate novel targets for the development of neuroprotective drugs for use during whole brain radiation. Furthermore, these findings warrant future clinical investigations of neuroprotection with GSK3β inhibitors during cranial irradiation, especially in the pediatric population.
Conflict of interest statement. None declared.
This work was supported by the RSNA Holman Pathway Resident Research Seed Grant [#RR0813] from the Radiological Society of North America (to E.S.Y.), the Vanderbilt Institute for Clinical and Translational Research and the CTSA grant [UL1RR024975] (to E.S.Y.), and the University of Alabama-Birmingham School of Medicine IMPACT Award (to E.S.Y.).