Requirement for Atm in Ionizing Radiation-Induced Cell Death in the Developing Central Nervous System

See allHide authors and affiliations

Science  15 May 1998:
Vol. 280, Issue 5366, pp. 1089-1091
DOI: 10.1126/science.280.5366.1089


Ataxia telangiectasia (AT) is characterized by progressive neurodegeneration that results from mutation of the ATMgene. However, neither the normal function of ATM in the nervous system nor the biological basis of the degeneration in AT is known. Resistance to apoptosis in the developing central nervous system (CNS) ofAtm−/− mice was observed after ionizing radiation. This lack of death occurred in diverse regions of the CNS, including the cerebellum, which is markedly affected in AT. In wild-type, but not Atm−/− mice, up-regulation of p53 coincided with cell death, suggesting that Atm-dependent apoptosis in the CNS is mediated by p53. Further, p53 null mice showed a similar lack of radiation-induced cell death in the developing nervous system. Atm may function at a developmental survival checkpoint that serves to eliminate neurons with excessive DNA damage.

Ataxia telangiectasia is a hereditary multisystemic disease resulting from mutations ofATM (1) and is characterized by progressive neurodegeneration, cancer, immune system defects, and hypersensitivity to ionizing radiation (2-4). ATM is similar at its COOH-terminal region to the phosphoinositol 3-kinase family, suggesting a role in signal transduction (2). Moreover, in vitro studies have shown that ATM is a component of the cell cycle checkpoint machinery that causes growth arrest after ionizing radiation–induced DNA damage (2,5). Although defects in ATM lead to neurodegeneration, its physiological role in the nervous system is unclear (6). The neurological defect or defects in AT become apparent early in life, suggesting that they originate during development. Furthermore,Atm is highly expressed in the developing nervous system, but only at low levels in the adult CNS (7).

We generated an Atm-knockout mouse by deleting a region of Atm that included exons 57 and 58 (8). These Atm−/− mice had a similar phenotype to that of other recently reported Atm−/− mice, including sterility and proneness to lymphoma (9). In many regions of the CNS in Atm−/− mice, ionizing radiation failed to induce cell death (10). There was an almost complete absence of irradiation-induced cell death in the hippocampal dentate gyrus, retina, cerebellum, and cerebral cortex compared with wild-type (WT) tissues (Fig.1). In WT mice, irradiation of the external granule layer (EGL) of the cerebellum resulted in widespread cell death, whereas the EGL of Atm−/− mice showed only occasional dead cells. In contrast, the thymus showed similar susceptibility to irradiation in both WT andAtm−/− mice. As an independent measure of cell death, in situ end labeling (ISEL) was used (11). ISEL-positive cells were seen in the WT, but notAtm−/− CNS after irradiation (Fig. 1). Apoptosis in the thymus occurred in both the WT (Fig. 1K) andAtm−/− (Fig. 1L) mice, indicating that the cellular machinery responsible for apoptosis-associated DNA degradation is present in Atm−/− mice.

Figure 1

Absence of ionizing radiation–induced apoptosis in regions of the Atm−/− CNS. Neutral red–stained sections of P3 Atm−/− cerebellum (C) and P5 retina (E) show a pronounced lack of pyknotic nuclei compared with similar sections from irradiated WT tissues (B and D). (A) Unirradiated WT cerebellum. (F toL) ISEL staining in the EGL of unirradiated WT (F), irradiated Atm−/− (G), and irradiated WT mouse (H and I) at P3. ISEL staining in (I) coincides with cells in the EGL having the morphological characteristics of apoptosis. In contrast, apoptosis is absent from unirradiated thymus (J), but present in irradiated WT (K) and Atm−/− (L) mice. Magnification: ×85 (A to H); ×170 (I to L). Differences in apoptosis in the P3 CNS of Atm−/− and WT mice after irradiation were also found for the dentate gyrus and regions of the cerebral cortex.

To establish that the Atm−/− phenotype was independent of genetic background (12), we performed the irradiation experiment with mice from each contributing genotype, 129/svj and C57BL/6. Similar spatial and temporal patterns and levels of death were observed after irradiation in each strain. We also studied cell death in the dentate gyrus of irradiatedAtm−/− mice up to 48 hours after irradiation and still did not detect cell death, indicating that apoptosis was not simply delayed in the Atm−/− mice (Fig.2). Taken together, our data support the conclusion that Atm mediates ionizing radiation–induced apoptosis in the developing CNS.

Figure 2

Absence of cell death in the dentate gyrus of Atm−/− mice up to 48 hours after irradiation. After ionizing irradiation, the developing P5 dentate gyrus of WT mice (+/+) show typical pyknosis after 24 hours (A) and extensive cell depletion after 48 hours (B). However, P5 Atm−/− dentate gyrus demonstrates marked resistance to irradiation for up to 48 hours (C and D), with no detectable pyknosis or obvious cell loss.

The p53 gene product is required for ionizing radiation–induced apoptosis in thymocytes and cerebellar granule neurons (13, 14), and p53 induction in irradiated cell lines derived from AT individuals and Atm−/− mice is delayed or absent (15). Therefore, we examined whether p53 is involved in Atm-dependent apoptosis. We found reduced p53 induction in the CNS of Atm−/− mice after irradiation as measured by protein immunoblot analysis and immunohistochemistry (16). Protein immunoblot analysis showed a lack of p53 stabilization in the cerebral cortex, olfactory bulb, and hippocampus and only a weak induction in the cerebellum fromAtm−/− mice compared with marked up-regulation in WT mice (Fig. 3A). However, induction of p53 was apparent in the thymus of both WT andAtm−/− mice, consistent with p53-dependent, but Atm-independent, death after γ-irradiation in this organ. Additionally, immunohistochemistry showed p53 induction after irradiation localized to regions of the cerebellum that subsequently undergo apoptosis (Fig. 3B). p53−/− mice (17) were also resistant to irradiation-induced cell death in the developing nervous system, supporting the notion that Atm-dependent death in the CNS requires p53 (Fig. 3Cb). Because ATM interacts with c-Abl (18) and c-Abl can regulate apoptosis after ionizing radiation (19), c-Abl null mice (20) were also examined. Cell death in the CNS of c-Abl null mice was indistinguishable from that in WT mice after irradiation (Fig.3C). Thus, Atm is an upstream modulator of apoptosis in the CNS after γ-irradiation and most likely signals by way of p53, but independently of c-Abl.

Figure 3

Atm-dependent apoptosis coincides with p53 stabilization. (A) Protein immunoblot analysis of p53 after ionizing radiation in the P5 CNS of WT andAtm−/− mice. In all WT (+/+) CNS tissues, p53 stabilization is apparent 2 hours after ionizing radiation, whereas Atm−/− tissues (−/−) show a pronounced deficit in p53 stabilization, although p53 stabilization in the thymus is similar in WT andAtm−/− mice. +IR: Tissue from irradiated animals. (B) Immunohistochemical localization identifies p53 stabilization in granule cells of the cerebellar EGL in WT, but not Atm−/− mice after irradiation. p53 stabilization in the cerebellar EGL coincides with cells in these regions that subsequently undergo apoptosis (Fig. 1). (a) Unirradiated (−IR) WT, (b) irradiated (+IR) WT, and (c) irradiatedAtm−/− cerebellum. (C) Neutral red–stained cerebellum from irradiated WT (a), p53 null (b), and c-Abl null mice (c) show that WT and c-Abl mice undergo irradiation-induced apoptosis, whereas p53 null mice do not. (Insets) ISEL-stained cerebellum of the respective genotypes. A similar lack of apoptosis was found in the retina and dentate gyrus after irradiation inp53−/− mice, but notc-Abl−/− or WT mice.

Although no widespread cell death was observed in the developing postnatal Atm−/− CNS after γ-irradiation, two populations of neural cells still underwent apoptosis in theAtm−/− mice. In the irradiated WT eye, apoptotic cells were seen throughout the periphery and center of the retina, whereas apoptosis was seen only in the periphery of irradiatedAtm−/− retina. Cell death was also apparent in the subventricular zone (SVZ) of both WT andAtm−/− mice, although it was reduced in theAtm−/− mice. Both the SVZ and the marginal zone of the retina contain relatively undifferentiated, multipotent precursors capable of giving rise to both neurons and glia (21). In contrast, the cerebellar EGL and the dentate gyrus harbor neuroblasts that generally give rise to a single lineage (22). Thus, as cells move from a multipotent, less-differentiated state to their terminal phenotype, Atm function after irradiation becomes apparent. BecauseAtm−/− and p53−/− mice appear neuroanatomically normal (23), Atm- and p53-dependent apoptosis is probably distinct from programmed cell death occurring during nervous system development (24).

These data establish a role for Atm during radiation-induced apoptosis in select cell populations in the developing CNS. It is possible, at this stage, that neurons require Atm as a component of a survival checkpoint so that developing neural cells that have genomic (or other) damage can be eliminated. Thus, defective Atm may allow genomically compromised neurons to survive, and their accumulated mutations lead to functional deficits later in life. In specific cell populations such as Purkinje or granule cells, this process could lead to selective neurodegeneration as seen in AT (3). Further, because disruption of Atm function is protective against irradiation, these findings may have therapeutic implications for attenuating the serious neurological sequelae after craniospinal irradiation for pediatric brain tumors.

  • * To whom correspondence should be addressed. E-mail: peter.mckinnon{at}


View Abstract

Navigate This Article