ATM localization and gene expression in the adult mouse eye

Animals

All animals were handled in strict accordance with the ARVO Statement for Use of Animals in Ophthalmic and Vision Research. C57BL/6J mice (Charles River, L'Arbresle, France) were maintained on a 12 h:12 h light-dark cycle at a room temperature of 21 °C.

RT–PCR and PCR

Atm mRNA in the cerebellum, neuroretina, RPE, ciliary body, and lens from 2-month-old adult mice were examined by RT–PCR. Total RNA was extracted from each sample of each mouse ocular compartment and from each mouse cerebellum by TRIzol method, according to the manufacturer’s recommendations (Invitrogen, Cergy-Pontoise, France). Next, 1 μg aliquots of total RNA were reverse transcribed with reverse transcriptase (SuperScript II, Invitrogen) and oligo dT primer, according to the manufacturer’s instructions.

PCR conditions such as MgCl₂ concentrations, annealing temperatures and number of cycles were optimized for Atm and cyclophilin A genes to perform PCR in the linear part of amplification [61]. We determined the appropriate number of cycles so that the amplification product was clearly visible on an agarose gel and could be quantified, but also so that amplification was in the exponential range and had not yet reached a plateau. The optimal number of cycles has to be in the same range for the specific RNA of interest (i.e., Atm) and the control, in this case cyclophilin A, to be measured by the comparison of the respective intensities of their distinct PCR amplified products on the same gel.

PCR reactions were performed in 25 µl of reaction mixture containing 2 µl cDNA, 0.2 µM of each Atm primer, and 0.04 µM of each cyclophilin A primer, 0.2 mM dNTP, 1.5 mM MgCl₂, 2.5 µl 10X PCR buffer, and 0.05 U Taq DNA polymerase (Invitrogen). Semiquantitative PCR was performed with a denaturing step of 94 °C for 4 min, followed by 30 cycles (optimal cycling) of 94 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min, then a final extension of 7 min at 72 °C. Atm primers (forward 5′- ATC CCT TGT GTG TTC TCT G −3′ and reverse 5′- CGC CTC TGC TGT CGT GTA T −3′) and cyclophilin A primers (forward 5′-TGG TCA ACC CCA CCG TGT TCT TCG-3′ and reverse 5′-TCC AGC ATT TGC CAT GGA CAA GA-3′) amplified 472 bp and 311 bp products, respectively.

PCR products were separated on 1% agarose gels and visualized by ethidium bromide staining under UV light. Each RT–PCR reaction was done in triplicate with three different samples for each microdissected tissue extract. The image of the ethidium bromide-stained gel was digitalized, using a digital video image analyzer ImagerTM (Appligene, Illkirch, France), and analyzed using ImageJ 1.40 g software.

We systematically verified that all the PCR reactions were performed during their exponential phase, also called the linear phase, and, thus, that we had not yet reached the plateau phase.

Statistical analysis

PCR signals were normalized as a function of the cyclophilin A gene expression, which reflects the initial amount of RNA, by making the ratio between the signal obtained for this Atm gene and that obtained on the control standard curve for the same quantity of cyclophilin A. Optical densities of the PCR bands were measured using ImageJ 1.40 g software. Statistical analysis was performed using ANOVA (ANOVA, Statview Software program, version 5) to detect significant intergroup differences. Values are expressed as mean±SEM, and a p<0.001 was considered statistically significant. Comparisons were made between retina and other ocular tissues using a Student–Newman–Keuls test.

Nonradioactive in situ hybridization

Eyes from adult C57BL/6J mice were fixed by incubation in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) at 4 °C, embedded in paraffin, and cut into 5 μm sections. The Atm PCR product was inserted into the pCRII-TOPO TA Cloning vector (Invitrogen). Products were checked by sequence analysis. Digoxigenin (DIG)-labeled riboprobes were synthesized by in vitro transcription using the standard DIG-labeling reaction protocol from Promega (Charbonnieres, France). T7 and SP6 RNA polymerase were used for the sense and antisense riboprobes, respectively.

Tissues were deparaffinized by incubation in xylene and rehydrated through a graded series of alcohol solutions. Next, 150 ng RNA probes were diluted in mRNA HIS Solution (Dako, Trappes, France) and incubated with sections at 60 °C overnight in a humidified chamber. After washing in Stringent Wash Concentrate 50X (Dako), labeling was detected by incubation with 1:500 alkaline phosphatase-coupled anti-DIG antibody in antibody diluent (Dako) for 1 h at room temperature. Atm mRNA hybridization signals were developed with the BCIP/NBT Substrate System (10–15 min at room temperature; Dako).

Immunostaining

Confocal microscopy and colocalization

Optically sectioned images were acquired by confocal laser scanning microscopy using a Leica SP5 confocal microscope and analyzed using Leica Microsystems LAS AF software (1.8.2). Two labels colocalized if they were too close in the tissue to be resolved optically. The extent of colocalization of two labels was measured using the “Colocalization” module of Imaris software (6.1.2; Bitplane, Zurich, Switerland), described previously [63]. Briefly, this program analyzes images of confocal sections acquired in two channels. Each confocal section consists of an array of square elements called pixels. A voxel is defined as a prism in which the base is the pixel and the height is the thickness of the confocal section. Imaris colocalization analyzes the entire confocal image by measuring the intensity of each label in each voxel. An intensity threshold (on a scale of 0–255 for pixel intensity) for each of the two labels is determined. In this study, we used a threshold of 30 (on a scale of 0–255) for each label. Voxel intensity values above this threshold were considered to be above background intensity. A voxel was defined as displaying colocalization when the intensity of both labels was above their respective thresholds. The percentage of material colocalized was based on the number of voxels displaying colocalization and the intensities (material) of the two labels in each voxel. The Pearson coefficient measures the correlation between the values of intensity for the two labels in voxels showing colocalization. The Pearson coefficient lies between +1 and −1, with positive values indicating a direct correlation and values near 0 indicating no correlation.

Western blot

Total protein lysates were prepared in a Potter-homogenizer with cold buffer composed of 150 mM Nacl, 10 mM Tris HCL, 1 mM EDTA, 1 mM EGTA, 2 mM NaOV, 190 mM NaF, 2 mM PMSF, 1% Triton X100, 0.5% NP-40, and protease inhibitor cocktail (Sigma-Aldrich). Protein concentration was determined using a Lowry protein assay kit (BioRad DC protein assay, Ivry-sur-Seine, France). Proteins were separated on 7% polyacrylamide gels. Each lane was loaded with 100 µg of protein extract. Separated proteins were transferred onto a nitrocellulose membrane and blocked with 5% skim milk for 1 h. Blots were incubated with either the monoclonal antibody against Mat-3 (generously provided by Professor Yosef Shiloh from Tel Aviv University, Tel Aviv, Israel) diluted 1:500 with 1% BSA in PBS-Tween for 3 h at room temperature or with the monoclonal antibody against ATMp (200–301–500) purchased from Rockland (Gilbertsville, PA) and diluted 1:500 with 1% BSA in PBS-Tween overnight. In parallel, membranes were incubated with 1:1,000 goat anti-β-Actin antibody (Tebu) for 2 h at room temperature. Proteins were detected by chemiluminescence (ECL, PerkinElmer Life, Courtaboeuf, France) on light-sensitive film (Biomax Light Kodak; Sigma-Aldrich). Atm^−/− mouse tissues were used as controls for nonspecific binding.

Expression and localization of DNA lesions in adult mouse ocular tissues

High amounts of 8-oxo-dG (a major oxidized form of guanine) are produced, both in genomic and mitochondrial DNA, by ROS in the retina, exposed to normal conditions of illumination and functioning in basal physiologic conditions. To evaluate the extent of oxidative damage of cellular DNA under basal conditions, we examined immunohistochemically the retinal and cerebellar distribution of 8-oxo-dG in the C57BL/6J mouse strain, using an anti-8-oxo-dG monoclonal antibody.

Strong immunoreactivity for 8-oxo-dG was easily detected in all retinal cell layers and in photoreceptor inner segments (PIS; Figure 1A). The PIS layer is highly enriched in mitochondria and peroxysomes. We further expanded our observations concerning the detection of DNA lesions by studying the 8-oxo-dG immunoreactivity in the control (Atm^+/+) mouse strain and the Atm-deficient mice (Atm^−/−). These strains have an sv129 genetic background. Both C57Bl/6J and sv129 mouse strains displayed the same 8-oxo-dG retinal immunostaining pattern (Figure 1B,C). Moreover, we also observed in cerebellar neurons an intense 8-oxo-dG immunoreactivity, mostly in the nuclei of the granule (GL) but also in the molecular layer (ML) of the cerebellar cortex (Figure 1F,G). Faint 8-oxo-dG immunolabeling was observed in the nuclei of the Purkinje cell layer (PL).

The physiologic oxidative stress existing in photoreceptor cells is very likely the source of DSBs [16] which, even if they might not be the major DNA lesions occurring in these neural sensory cells, constitute certainly an important percentage of the total DNA lesions affecting the mitochondrial and genomic DNA of these retinal cells specialized in phototransduction. We detected γH2AX, largely considered as a reliable marker of DSBs lesions, in the normal mouse retina (C57Bl/6J and Atm^+/+ mouse), in the Atm knockout mice retina (Figure 1 I-K), and in cerebellar neuronal cells of all mouse strains studied (cerebellar data not shown). The detection of this strong immunoreactivity in all retinal cells and especially in photoreceptor cells also strengthens the hypothesis suggesting that basal functional conditions in these cells are associated with a high oxidative stress linked to a huge production of ROS. This production is at the origin of DNA damage including DSBs both in genomic and mitochondrial DNA, since γH2AX immunoreactivity is strongly observed both in photoreceptors nuclei and PIS.

Expression and localization of Atm mRNA in adult mouse ocular tissues

We investigated Atm mRNA levels in four distinct murine ocular tissular compartments and in the cerebellum by semiquantitative RT–PCR (Figure 2). As expected, a 472 bp band, corresponding to Atm mRNA, was observed in all mouse samples tested (cerebellum, neuroretina, RPE, ciliary body, and lens; Figure 2A). Cyclophilin A cDNA band intensity (used as an internal control) varied less than 5% between each lane of electrophoresis migration in all our experiments and was used to normalize variations in Atm expression. Atm mRNA was detected in each mouse ocular tissue sample examined; bands of the same size were observed in both ocular and cerebellar tissues. Atm mRNA levels were significantly higher in the neuroretina than in any other ocular tissues (Figure 2B).

To determine the cellular sites of Atm mRNA expression in the adult mouse ocular compartments, we analyzed paraffin-embedded tissue sections of adult C57BL/6J mouse eyes (P60) by nonradioactive in situ hybridization using DIG-labeled sense and antisense riboprobes [64]. As expected, no specific labeling was detected in any ocular tissue sections hybridized with the Atm sense probe (Figure 3A). Atm mRNA was detected in all neuroretinal nuclear layers such as ganglion cell layer (GCL), inner and outer nuclear layer (INL and ONL) and in PIS (Figure 3B). No significant Atm hybridization signal was detected in the photoreceptor outer segments (POS) as well as the inner and outer plexiform layers (IPL and OPL). Expression of Atm mRNA seemed higher in the PIS, INL, and GCL than in the ONL. Atm mRNA was also expressed in the RPE cell layer, which is adjacent to the neural retina and is also derived from the CNS (outer layer of the optic cup; Figure 3C). The systematic bioinformatic analysis of ESTs databases and tissue specific cDNA profiling databases allowed us to uncover very interesting data obtained independantly by other teams about Atm gene expression. The most interesting data are released in the BioGPS website and correspond to an Atm gene expression/activity chart. They are in complete adequation with the results presented in this report. This chart confirms the significant levels of expression and activity of Atm detected in cornea, iris, ciliary body, RPE, retina, and eyecup. This chart indicates strikingly that the ratios of Atm gene expression in most compartments of the eye and especially in the retina are comparable to that detected in the cerebellum.

Non-neuronal ocular tissues, including choroidal (Ch) and scleral (Sc) cells, also produced Atm transcripts (Figure 3C). We detected significant levels of Atm mRNA in the corneal epithelial cells (CEp), corneal stromal keratocytes (CS) and corneal endothelial cells (CEn) (Figure 4A). In the lens, Atm mRNA was readily detected in the lens epithelium (Le), the transitional zone (Tz), and in elongating fiber cells (Lf), in which the nucleus was still visible (Figure 4B). Iridal cells also showed significant Atm mRNA hybridization signals (Figure 4C). The pigmented ciliary epithelial (PCE) and nonpigmented ciliary epithelial (NPCE) cells of the ciliary body both displayed Atm mRNA hybridization signals (Figure 4D).

We also characterized Atm expression in normal mouse cerebellum tissue sections. This served as a positive control for our experiments on adult mouse ocular tissues. As expected, no specific labeling was detected in any cerebellar tissue sections hybridized with the Atm sense probe (Figure 5A). Atm mRNA was mainly detected in the granule GL and ML of the cerebellar cortex (Figure 5B,C). Strong labeling was also detected in the PL (Figure 5D).

Expression and localization of ATM protein in adult mouse ocular tissues

To confirm our findings from the nonradioactive in situ hybridization experiments, we used immunohistochemistry to investigate the presence of unphosphorylated ATM in adult mouse ocular paraffin-embedded tissue sections, especially in the retina. The specificity of the anti- ATM antibody was confirmed in western blots of proteins homogenates from mouse neuroretina or brain. Identical single bands corresponding to a protein of a molecular weight around 350 kDa were reproducibly detected for each tissue extract tested. This band corresponded to the expected size of ATM, reported in previous publications [65]. This band was not detected in Atm-deficient mice proteins homogenates (Figure 6). These results demonstrate the specificity of the monoclonal anti-ATM antibody used in our experiments.

The presence of ATM protein in eye tissue sections from adult mice was investigated by immunohistochemistry (Figure 7). No ATM immunostaining was observed in the appropriate control experiments (Figure 7A). We detected ATM protein in the nuclei of the ONL, INL, and GCL (Figure 7B-G). ATM immunoreactivity was also observed in the RPE cell layer and in non-retinal cells such as the Ch cells, including uveal and vascular endothelial cells, and Sc cells (Figure 7B,H). Interestingly, ATM immunostaining in the ONL, INL, and GCL appeared to be of similar intensity (Figure 7B-D). In contrast, the intensity of the in situ hybridization signal appeared to be weaker in the ONL than in the INL or GCL (Figure 3B). ATM protein in the ONL might be subject to post-transcriptional regulation; further study is required to confirm this hypothesis. These results are consistent with a previous study suggesting that ATM is post-transcriptionally regulated in proliferating cells such as peripheral blood mononuclear cells [66]. Thus, ATM may be post-transcriptionally regulated in both dividing and post-mitotic cells.

ATM immunostaining was also detected in other non-neuronal ocular cells, including the nuclei of CEp, CS, and CEn (Figure 8A). ATM immunolabeling was observed in the nuclei of the Le, the Tz, and the Lf (Figure 8B). ATM was slightly detected in all iridal cells, with signal in both pigmented layers of the iris and iridal uveal cells (Figure 8C). It was also present in the nuclei of ciliary processes. Both the PCE and the NPCE cell layers had marked ATM immunostaining (Figure 8D). All cell types observed in the ciliary body showed significant levels ATM immunoreactivity, including those located at the root of the ciliary body and at the iridocorneal angle.

In cerebellum tissue sections, we detected ATM immunoreactivity in the nuclei of the GL and ML of the cerebellar cortex (Figure 9B,C). Faint but significant ATM immunolabeling was observed in the nuclei of the PL (Figure 9D,E). No ATM immunostaining was observed in the appropriate control experiments (Figure 9A).

Detection of phosphorylated ATM in adult mouse ocular tissues

Given the potential role of ATM in the maintenance of cell integrity and its broad distribution throughout the adult mouse eye, we studied the expression of its active phosphorylated form (ATMp) in adult ocular tissues using immunohistochemistry and western blot analysis. For immunohistochemistry experiments, we used three antibodies specific for mouse phosphorylated ATM at serine 1987 (pS1987; see Methods). The same pattern was observed with these antibodies. We used one of these pS1987 antibodies (Rockland) for western blotting. We detected a specific band in each lane of DNA electrophoresis migration, which corresponded to the size of ATMp. This band was not observed in Atm-deficient mice (Figure 6). We detected strong ATMp immunoreactivity in all cell layers—GCL, INL, ONL, IPL, and OPL—and in the PIS in adult retinal sections (Figure 10B-D). ATMp immunoreactivity showed a similar pattern to that of ATM in GCL and INL (Figure 10E,F; Figure 7E,F). ATMp seemed to be mainly, but not exclusively, detected in the nuclei of these neuroretinal layers, whereas it was mostly detected in the cytoplasm of photoreceptor cells (Figure 10G). We also detected ATMp immunostaining in the RPE cell layer and in Ch and Sc cells (Figure 10B,H). No immunostaining was observed in the appropriate control experiments performed with each ATMp antibody (Figure 10A).

These highly reproducible observations shown in Figure 10, which strongly support the spontaneous activation of ATM in both retina and cerebellum, especially in photoreceptor cells and cerebellar granule cells, combined with the detection of DNA damage - using specific 8 oxo-dG and γ-H2AX antibodies - in the same cells previously shown above in Figure 1 contribute to establish the existence of a clear link between the occurrence of high spontaneous levels of oxidative stress induced DNA lesions and ATM phosphorylation in these very cells.

Among the non-neuronal cells of ocular tissue sections, we observed strong ATMp immunostaining in the nucleus and cytoplasm of CEp, CS, and CEn cells (Figure 11A). Whereas ATM cytoplasmic immunostaining was faint in many corneal cells, ATMp signal was strong in both the nucleus and cytoplasm of all corneal cells. ATMp immunostaining of the keratocytes illustrates quite clearly the detection of the strong ATMp immunoreactivity, both in the nucleus and cytoplasm of these cells, with an intense and widespread ATMp immunolabeling evident throughout the cytoplasm (arrow in Figure 11A). ATMp immunostaining was observed in the nucleus and cytoplasm of Le, Tz, and elongating nucleated Lf cells (Figure 11B). We detected strong ATMp immunostaining in iridal cells, whereas a faint ATM immunoreactivity was evident in the same cells (Figure 11C). We detected ATMp immunoreactivity in PCE and NPCE cells of the ciliary body (Figure 11D). ATMp appeared to be present in the nucleus and cytoplasm in all iridal, PCE, and NPCE cells.

We observed strong nuclear ATMp immunolabeling in the PL (Figure 12D,E), but weak immunolabeling in the cytoplasm of the GL (Figure 12B,C); this was in contrast to the nuclear immunolabeling observed for ATM (Figure 9B,C). No immunostaining was seen in the appropriate control experiments performed with each antibody (Figure 12A).

Subcellular localization of ATM and ATMp in adult mouse photoreceptors

Our findings revealed ATM immunolabeling in the nuclei of all retinal cell layers, but a different pattern of localization was observed for ATMp. We detected ATMp immunostaining, mostly in the cytoplasm of photoreceptor cells, but mainly in the nuclei of other retinal cells. We also observed the same distinct and specific patterns of ATM and ATMp localization in the cerebellar GL. ATM protein was detected mostly in nuclei, whereas ATMp seemed to be localized in the cytoplasm of granule cells. To confirm this, we determined ATM and ATMp distribution and their subcellular localization using high-resolution confocal imaging of adult mouse retinal and cerebellar tissue sections.

Confocal imaging showed ATM immunolabeling to be present in the nuclei of the ONL. Colocalization of ATM and nuclear DNA staining (TO-PRO-3) was particularly marked at the peripheral nucleoplasmic euchromatin, which surrounds the central densely stained heterochromatin [67] (Figure 13A). We used an anti-recoverin antibody to further investigate the subcellular localization of ATM in the ONL. Recoverin is a cytoplasmic protein involved in the visual transduction cascade in both types of photoreceptor cells (cones and rods). The anti-recoverin antibody revealed a specific immunostaining exclusively in the area surrounding the nuclear region, which is immunoreactive for ATM (Figure 13B). We analyzed confocal images using the “Colocalization” module of the Imaris software package (Bitplane) to quantify the extent of ATM and recoverin colocalization. Results of the correlation analysis for the staining intensity of both antibodies were expressed as Pearson correlation coefficients. The Pearson coefficient is a number between +1 and −1, with positive values indicating a direct correlation and values near 0 indicating no correlation. A representative image of ATM and recoverin staining is shown in Figure 13B. The correlation coefficient of overlap between them was −0.28. Thus, ATM and recoverin did not colocalize. These data provide strong evidence that ATM protein is localized mostly in the nuclei of photoreceptors cells. To confirm the localization of ATMp immunolabeling in the ONL, we performed a double immunostaining for ATMp and recoverin proteins. The double staining strongly suggested that recoverin colocalized with ATMp protein in the cytoplasm of ONL (Figure 13C). A representative image of the ATMp and recoverin immunostaining is shown in Figure 13C. The Pearson coefficient in voxels with colocalization of ATMp and recoverin was 0.9, with 50% and 65% colocalization of ATMp and recoverin immunostaining (percentage of total intensity of green and red labels), respectively. There was a high degree of colocalization between ATMp and recoverin. These data provide strong evidence that ATMp protein is primarily localized in the cytoplasm of photoreceptors cells.

In cerebellar tissue sections, ATM immunostaining in PL appeared to be predominantly nuclear, with faint but detectable labeling in the cytoplasm (Figure 14A). ATM immunoreactivity was also detected in the nuclei of GL (Figure 14B). The Pearson coefficient in voxels for colocalization of ATM immunolabeling with nuclear DNA staining by propidium iodide (PI) was 0.42, with 70% colocalization of ATM immunostaining and 72% colocalization of PI staining. These data provide strong evidence that ATM protein is essentially localized in the nuclei of granule cells. ATMp immunostaining was also detected in the cytoplasm of GL (Figure 14C). A representative image of ATMp and PI staining is shown in Figure 14C. The Pearson coefficient in colocalized voxels was close to zero, showing that ATMp and PI were not colocalized. These data confirm that ATMp was mainly in the cytoplasm of granule cells.

We can summarize our findings as follows: ATM and ATMp were distributed in the same cells in all ocular compartments, with the exception of the ONL. In most, if not all, ocular compartments, ATM cellular immunoreactivity was detected primarily in the nucleus. ATMp cellular immunoreactivity was also detected primarily in the nucleus in most ocular cells, but was also detected in the cytoplasm. Indeed, we observed ATMp immunoreactivity mostly, if not exclusively, in the cytoplasm of all photoreceptors, in the PIS layer, as well as in the OPL and in the IPL. ATM appeared to be localized exclusively in the nucleus of these cells. These contrasting distribution patterns for ATM and ATMp were also observed in the granule cells of the cerebellum.

ATM and ATMp expression in non-neural tissues of the adult mouse eye

The cornea and lens are the first tissues to receive and absorb the light directly. ATM immunoreactivity was mostly detected in the nucleus, with a fraction observed in the cytoplasm of all corneal epithelial cells. ATMp showed strong staining both in the nucleus and cytoplasm of all epithelial corneal cells. It has been demonstrated that Atm gene expression is upregulated in response to a mitogenic stimulus, resulting in an increase in ATM protein levels [66]. Moreover, during cell cycle progression in physiologic conditions, ATM is activated during mitosis and localizes at centrosomes with p53 [80]. As mitosis frequently occurs in the corneal epithelium [81], it is not surprising that both ATM and ATMp immunoreactivities are detected in these cells. Similar observations were made in all dividing cellular and fiber compartments of the adult mouse lens. ATMp immunostaining suggests the possible involvement of ATM in the differentiation and cataractogenesis of the lens. Interestingly, mice heterozygous for Atm gene are more sensitive than wild-type animals to the cataractogenic effects of IR [82,83]. IR-induced cataracts result from the accumulation of abnormally differentiated progeny of the lens germinative cells, which are located in the equatorial region of the lens and are in complete continuity with the lens epithelium. Moreover, the targeted disruption of Nbs1 in the lens altering the nuclear localization of the MRN (Mre11/Rad50/Nbs1) complex contributes to the early stages of cataract development due to impaired lens fiber cell differentiation [57]. The MRN complex is essential for the cellular response to DNA DSBs and links DNA repair to activation of checkpoint signaling through the protein kinase ATM [84]. Additionally, many proteins involved in DNA repair seem to be involved in cataractogenesis [85]. Indeed, overexpression of DNA polymerase β in the lens epithelium results in the early onset of severe cortical cataracts. This mechanism seems to be linked to DNA damage of lens epithelial cells, which can result in defective cell cycle control and apoptosis. Thus, ATM is present in its activated form in lens and corneal cells to regulate cell cycle and eliminate DNA damage.

Light as well as UV radiation cross the cornea and aqueous humor of the anterior chamber to meet iridal and ciliary body cells. As may be expected, ATMp shows strong immunostaining in these cells. Indeed, the iris is a potential site of free radical damage as it is rich in polyunsaturated fatty acids, which are particularly susceptible to peroxidation [86]. Furthermore, many antioxidant enzymes, such as Cu/Zn superoxide dismutase, catalase, and acidic glutathione S-transferase are present in the iris. These enzymes are capable of coping with high amounts of ROS and other free radical species [87]. Similarly, several classical antioxidant enzymes, as well as 2-Cys peroxiredoxins (PRDX), are detected in the ciliary body [88]. These enzymes are all involved in the neutralization of hydrogen peroxide and protect the cellular components against oxidative stress. These studies suggest a physiologic function for 2-Cys PRDXs in the protection of cells in the human ciliary body [89].This suggests that ATM and ATMp are present to protect iridal and ciliary body cells from oxidative damage.

ATM and ATMp expression in the adult mouse retina

Photoreceptor cells continuously produce high levels of ROS as a byproduct of their various functions throughout their lifespan. Thus, as may be expected, we detected ATM and ATMp immunoreactivities in these cells. The POS may be highly susceptible to lipid peroxidation by ROS due to their high concentration of docosahexaenoic fatty acyl chains and to the proximity of the likely sources of ROS production: peroxisomes and mitochondria in the adjacent PIS and RPE cells during POS phagocytosis. In this study, we detected significant levels of ATMp in PIS and in the IPL and OPL. However, the subcellular localization of ATM and ATMp in photoreceptors differed substantially from that observed in other retinal cells. Indeed, ATM immunoreactivity in the ONL was mostly, if not exclusively, confined to photoreceptor nuclei and did not colocalize with recoverin immunoreactivity. In stark contrast with ATM staining in photoreceptor cell bodies, ATMp had a predominantly cytoplasmic localization.

The function of ATM in the cytoplasm is not clear, but previous findings in mice have demonstrated that Atm deficiency leads to abnormalities of cytoplasmic organelles [38]. The different patterns of intracellular localization reported for ATM may result from tissue-specific functions of ATM or from cell-cycle arrest inherent to the post-mitotic nature of the cells [90]. It is widely known that cells in which ATM is localized in the nucleus undergo cell division. By contrast, neurons are post-mitotic and undergo cell-cycle arrest. It remains to be established whether the differences in intracellular localization depend on the nature of the tissue as well as the cell type producing ATM, or on any cell cycle arrest.

In the neuroretina, the activated form of ATM is present in the cytoplasm, where it may protect against oxidative damage resulting from the high levels of ROS production by peroxisomes and mitochondria-major sites of oxidative metabolism. A pioneering study suggested that some ATM protein was localized outside the nucleus in cytoplasmic vesicular structures, more precisely in the microsomal fraction [91]. Additionally, it has been demonstrated that a portion of the extranuclear ATM was localized to peroxisomes. It was indeed shown that ATM might be bound to peroxisome membranes, but ATM internalization in the peroxisomal matrix still remains to be confirmed [92]. Another study described a marked preferential distribution of ATM immunoreactivity in the endosomes of the cerebellar granule neurons, with a much lower intensity of the ATM immunohistochemical staining in the Purkinje and molecular layers [76]. Cytoplasmic ATM may be required in the mouse brain for preventing lysosomal accumulation [38]. A similar function of ATM may be required in retinal cells.

The reason for cerebellar degeneration in A-T is not clear. It has been attributed, by several investigators, to cytoplasmic functions of ATM that are not necessarily linked to the DNA damage response. PIS are highly enriched for peroxisomes and mitochondria. These organelles are sources of ROS but contain specialized antioxidant systems responsible for detoxifying reactive oxygen intermediates. However, when faced with excessively high oxidative stress and DNA damage, these systems cannot carry out their normal detoxification functions. ATM signaling is required to sense and initiate DNA DSBs repair. Therefore, nuclear genomic instability resulting from loss of this function has so far been considered as the major mechanism underlying the pathology of A-T. However, this disease presents a wide variety of symptoms, not all of which are readily explained by nuclear genomic instability. The study of cells and animal models of A-T has led to much speculation about additional pathogenic mechanisms.

Moreover, recent studies have demonstrated that ATM is involved in other important functions. Two novel intertwined roles have been proposed for ATM: the regulation of ribonucleotide reductase, the rate-limiting enzyme in the de novo synthesis of deoxyribonucleoside triphosphate, and the control of mitochondrial homeostasis [93,94]. ATM could modulate mitochondrial respiratory rates directly or indirectly, or through some DNA repair functions. Alternatively, ATM may regulate mitochondrial function in a DNA-repair-independent, hence metabolic, capacity [95]. These new findings suggest that A-T pathophysiology is more complex than anticipated and may be closely related to the field of mitochondrial diseases. The photoreceptor mitochondria are confined within the inner segment and presynaptic regions of the OPL. ATMp protein was also readily detected within the OPL of the retina, where photoreceptors form synapses with neuronal bipolar cells. Most mitochondria in the presynaptic regions of the rod bipolar neurons are concentrated in the IPL. Indeed, axonal presynaptic terminals of the bipolar cells are either synaptically connected to AII amacrine cells—which, in turn, are connected to the cone cells circuitry—whereas cone bipolar cells are directly synaptically connected to the retinal ganglion cells of the cone pathway [96]. Mitochondria are primary sites of ROS generation. Previous studies have shown that one characteristic feature of cells from patients with A-T is their state of continuous oxidative stress [31]. A-T cells display intrinsic mitochondrial dysfunction, implicating ATM in the regulation of mitochondrial function [95]. It is thus possible that the activated form of ATM reduces or prevents the deleterious effects of ROS production in the mitochondria-rich regions of retinal neurons, resulting from the normal functioning of the electron transport chain coupled with phosphorylative oxidation.

Although the highest oxygen concentrations are detected in the neural retina, one should not underestimate the importance of oxygen consumption and ATP production and thus the high amounts of ROS physiologically produced in RPE cells [97,98]. ATM protein showed immunostaining in the nucleus of RPE cells, and strong ATMp labeling was observed in these cells. The RPE plays a central role in the physiology of the retina [99]. RPE has numerous functions extremely important for the neural retina. It regulates the ionic environment of the subretinal space. It phagocytoses shed outer segments of photoreceptors. It converts the 11 trans retinal to the photosensitive 11 cis retinal, which is a key molecule of the visual cycle and phototransduction cascade. It also constitutes the outer hemato-ocular barrier of the eye. In addition to its light absorption properties, which generate free radicals, RPE is exposed to high oxygen concentrations, which may promote the generation of ROS, inducing oxidative damage in the event of impaired antioxidant defense mechanisms [68]. Indeed, POS phagocytosis subjects RPE cells to an oxidative event of the same order of magnitude as that measured in macrophages [100]. The event is not an extracellular macrophage-type respiratory burst, but may be due to intracellular hydrogen peroxide resulting from NADPH oxidase activity in the phagosome or from β-oxidation of ROS lipids in peroxisomes [101]. The significant levels of ATMp immunoreactivity observed in RPE cells may prevent the detrimental consequences, in particular DNA damage, caused by the frequent occurrence of oxidative events during physiologic RPE functions.

Distribution of ATM and ATMp in photoreceptor and cerebellar granule cells

The cytoplasmic localization of ATMp in photoreceptors is intriguing, as some form of retinal degeneration or dysfunction would be expected to occur in Atm-deficient mice and A-T patients. So far, no such abnormalities have been reported. It remains to be determined why distinct ATM and ATMp immunostaining patterns are observed in granule cells and photoreceptors but not in other neuronal cells.

One possible explanation for this discrepancy may be related to the specific neuronal cell types targeted in various neurodegenerative diseases involving either the cerebellum or retina, or both. Indeed, studies performed on postmortem cerebellum samples from A-T patients unequivocally demonstrate that cerebellar granule cells are major targets of the A-T neurodegenerative process [102]. Whatever the molecular genetics cause might be for any type of inherited retinal degenerations, photoreceptor or RPE cells are the major retinal cell types that undergo degeneration (RetNet). This highlights a specific susceptibility of these cells to oxidative stress and DNA damage. The molecular basis of this cell type-specific susceptibility remains to be elucidated. The Harlequin mutant mouse strain (Hq) is an important mouse model, combining both cerebellar granule and photoreceptor cell degeneration [103]. This model has enabled the establishment of a link between oxidative damage, cell cycle reentry, and cell death. Reduced levels of apoptosis-inducing factor (Aif) were obtained in this mutant by proviral insertion. The cerebellar neurons and retinal cells in this mouse thus display increased ROS-induced DNA damage accompanied by the appearance of S-phase markers, preceding apoptotic death, cell cycle activation in neurons, and DNA synthesis. The particular vulnerability of cerebellar granule and photoreceptor cells to ROS cannot be fully understood without consideration of the crucial functions of ATM in cell survival. The consequences of ATMp cytoplasmic localization in these cell types open up new possibilities in the study of ATM-associated molecular pathways in cell survival, both in physiologic conditions and in neurodegenerative disorders. Notably, the links between the cytoplasmic localization of ATMp and the major role played by mitochondria and peroxisomes in ROS production, including vicious circles of ROS overexpression in neurodegenerative processes, have not been clearly established.

Another alternative, but not exclusive, explanation of these distinct patterns of ATM and ATMp localization and their possible link to the apparent sparing of photoreceptor cells in A-T patients might be provided by findings obtained in the mutant Purkinje cell degeneration mouse model (pcd) [104]. Pcd mutant mice undergo selective degeneration of specific neuronal populations: cerebellar Purkinje cells, photoreceptor cells, and olfactory mitral cells. A recent study has identified predegenerative changes, including DNA damage/repair foci in target neurons with a well preserved general cytology, in the early stages of the neurodegenerative process in these mice [105]. The substantial vulnerability of photoreceptors in pcd mice contrasts with the apparent resistance of the same cells in Atm-deficient mice. Indeed, no retinal degeneration or dysfunction has been reported so far both in A-T patients and in Atm-deficient mice [106-108]. However, A-T patients and Atm-deficient mice ophthalmic phenotypes must nowadays be reassessed by emerging noninvasive techniques increasingly available for exploring both human and mouse retinal structure and functions. This is mandatory for eliminating the possibility that A-T patients and Atm-deficient mice photoreceptors might actually display some subtle or manifest retinal dysfunctions corresponding to predegenerative alterations possibly reversible, such as those detected at the earliest stages of the neurodegenerative process occurring in the olfactory mitral cells of pcd mutant mice. Most A-T patients are affected by oculo-motor abnomalities. These abnormalities prevent in most cases the stable fixation of any visual stimulus. The absence of fixation often does not allow the ophthalmologists to perform complete and accurate visual examinations. Many A-T patients display a characteristic oculomotor apraxia (difficulty in the initiation of voluntary eye movements) frequently preceding the development of telangiectases [109]. Moreover, many A-T patients present a strabismus and/or a nystagmus [106]. It is important to emphasize that a nystagmus presumed to be of oculomotor origin might well be a consequence of retinal dysfunction or degeneration.

Another possible explanation for the discrepancy observed between ATM and ATMp cellular immunostaining patterns may involve the second master regulators of the DNA damage response: ATR protein kinase [110]. Most ATM substrates can also be phosphorylated by ATR, and the major functions of ATR and ATM in cell-cycle control are overlapping and redundant. However, ATM and ATR respond to different types of DNA damage: ATM responds to DSBs, and ATR responds to replication stress [111]. Although ATR is primarily a replication stress-response kinase, it is also activated by DSBs. A major established fact must be emphasized: both ATM and ATR are interdependent [112,113]. In A-T cells, shared ATM and ATR substrates are not phosphorylated efficiently in response to IR because ATM is absent, causing a delay in end resection. Eventually, ATM-independent end resection does occur, allowing ATR to recognize the damage [114]. The consequence of this interdependency is that Atm-deficient cells have severe cell-cycle checkpoint defects in response to IR exposure. The overlapping and redundant functions of ATM and ATR, as well as their specific distinct functions, may vary according to neuronal cell type. Further experiments based on conditional knockout mice for ATR as well as ATM are needed to test this hypothesis.

The importance of maintaining genomic stability in neurons cannot be overemphasized. Their finite number, long life, high metabolic rate, and continuous exposure to oxidative stress, together with high levels of gene transcription, require stringent control of genomic integrity. Our findings show that ATM is present in the adult mouse whole cerebellum and retina. ATM and its activated form seem to be required for preventing the accumulation of oxidative damage in the eye, especially in the photoreceptors of the retina. The ATM/ATMp system is involved in the protection from abnormal ROS production and irreparable lethal DSBs and DNA deletions. Retinal antioxidants and DNA repair systems seem to be particularly efficient in the adult retina; indeed, retinal malignancy is rare in adult patient retinas despite daily exposure to UV and visible light radiations and metabolism-induced damage. Our study provides further insight into the molecular building blocks and pathways underlying the highly efficient retinal systems protecting against oxidative stress, DNA damage, retinal degeneration, and malignancies.