SEARCH

SEARCH BY CITATION

Keywords:

  • mouse retinal development;
  • apoptosis pattern;
  • bax;
  • fasL;
  • gld;
  • visual functionality

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Apoptosis plays a major role in the development of the central nervous system. Previous studies of apoptosis induction during retinal development are difficult to interpret, however, because they explored different mouse strains, different developmental periods, and used different assays. Here, we first established a comprehensive sequential pattern of cell death during the whole development of the C57BL/6J mouse retina, from E10.5 to postnatal day (P) 21 by using the terminal deoxynucleotidyl transferase (TdT) -mediated deoxyuridine triphosphate (dUTP)–biotinylated nick end labeling (TUNEL) assay. We confirmed the existence of three previously described apoptotic peaks and identified another, later peak at P15, in both the outer nuclear layer, in which the photoreceptors differentiate, and the ganglion cell layer. Comparison of wild-type C57BL/6 mice, gld mice, defective in the death ligand fasL, and bax-/- mice, defective in the pro-apoptotic BAX protein, revealed a minor role for FAS ligand but a crucial role for BAX in both apoptosis and normal retinal development. The lack of BAX resulted in thicker than normal inner neuroblastic and ganglion cell layers in adults, with larger numbers of cells and an impaired electroretinogram response related to a decreased number of responsive cells. Our findings indicate that cell death during normal retinal development is important for the modeling of a functional vision organ and showed that the pro-apoptotic BAX protein plays a crucial role in this process. Developmental Dynamics 228:231–238, 2003. © 2003 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Apoptosis is an evolutionarily conserved process of cell death that plays an essential role in development, including in the development of a functional central nervous system (Oppenheim, 1991). The eyes are part of the central nervous system, and their development begins with the formation of bilateral evaginations from the prosencephalon. From approximately embryonic day (E) 10, the pit of each evagination hollows, leading to the formation of an inner and an outer layer (Kaufman, 1994). The inner layer of the optic vesicle, the neuroblastic layer (NbL), differentiates into additional layers. From approximately E17, the NbL consists of an inner neuroblastic layer (INbL), which later gives rise to the ganglion cell layer (GCL) and an outer neuroblastic layer (ONbL). These two layers are separated by the inner plexiform layer (IPL). The ONbL divides at around postnatal day (P) 6 to P8 to give the outer nuclear layer (ONL), in which the photoreceptor cells differentiate, and the inner nuclear layer (INL), which consists of horizontal, bipolar, Müller, and amacrine cells. The outer plexiform layer (OPL) forms between these two nuclear layers.

The first aim of our work was to carry out a comprehensive kinetic analysis of apoptosis throughout retinal development in a normal mouse strain. Previous studies have reported that apoptosis occurs at various stages of neural retina development (Young, 1984; Portera-Cailliau et al., 1994; Laemle et al., 1999; Trousse et al., 2001), but each group studied a different strain (C56BL/6, C57BL/6J, ZRDCT, and BALB/c, respectively) and used different techniques to detect apoptotic cells (numbers of pyknotic nuclei or terminal deoxynucleotidyl transferase (TdT) - mediated deoxyuridine triphosphate (dUTP)–biotinylated nick end labeling [TUNEL] labeling). Moreover, none of these studies covered the entire duration of eye development. In addition, studies that quantified apoptotic cells did so for half retina, without taking into account the growth of the retina or variations in the number of cells in each layer between stages P0 and P30 (Young, 1984; Portera-Cailliau et al., 1994). To assess cell death precisely at different stages, we counted both the number of apoptotic cells and the total numbers of cells in each layer and calculated the percentages of apoptotic cells.

A second aim of our study was to explore the molecular mechanisms involved in cell death during retinal development. Schematically, the induction of programmed cell death in mammals involves two major pathways: (1) the extrinsic pathway triggered by ligand-mediated engagement of the FAS/tumor necrosis factor (TNF)/TRAIL receptor family, leading to the activation of the caspase cascade; and (2) the intrinsic pathway triggered by growth factor deprival, cellular stresses, and p53 activation in response to genetic alterations or cell cycle dysregulation. This process leads to the mitochondrial release of caspase activators (cytochrome c, Smac/Diablo) and/or of caspase-independent death effectors (AIF; Hengartner, 2000; Martinou and Green, 2001), regulated by the anti- and pro-apoptotic BCL2/BAX protein family. We investigated the relative contributions of these two pathways by comparing retinal development, adult retinal morphology, and functional activity in wild-type mice, gld mutant mice, which have defective FAS ligand (FASL), and bax-/- mice, in which the bax gene has been deleted. The wild-type, gld, and bax-/- mice all had the same genetic C57BL/6 background. We found that BAX plays a major role in cell death during retinal development and is crucial for the establishment of a functional retina.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We used the TUNEL technique to study apoptosis in the retina, day by day, from E10.5 to P21 (each day corresponds to one stage) and in adults. The number of animals used and the complexity of the data obtained, due to the study of a kinetic in one layer giving rise to two then three retinal layers, meant that we have to choose a specific model for the statistical analysis: no simple model could be used for this data series. We chose a generalized linear model with a Poisson distribution and a logarithm link (Nelder and Wedderburn, 1972).

Stage E10.5, the time point at which the optic cup has just formed, is the first stage at which it is possible to count cells. At this stage, the percentage of apoptotic cells was high (Fig. 1). Most of the apoptotic cells were located in the middle of the retina (Fig. 2A). The percentage of dying cells decreased progressively from E11.5 to E13.5 and then remained at a constant low level until the end of embryonic development (Fig. 1). On E17.5, the NbL begins to divide into the INbL and the ONbL, separated by the future IPL. At this stage, only a few cells were apoptotic, and almost all of these cells were in the future INbL. By E18.5, the percentage of dying cells began to increase in the INbL, and in the IPL, whereas it was very low in the ONbL. After birth, the percentage of apoptotic cells continued to increase rapidly in the InbL and, at P2, this layer contained the highest percentage of apoptotic cells observed in any layer at any time point of retinal development (Figs. 1, 2B). The proportion of labeled cells remained high from P3 to P7 (almost 2.5%). During this period, few dying cells were observed in the ONbL. On P5, the OPL became visible as a thin line near the optic nerve. This separation extended toward the periphery from stage P5 and reached the ora serrata by P7. The ONL and INL were considered to be distinct from P7 onward. This stage corresponded to the start of an apoptotic peak in the INL, with the maximum percentage of dying cells observed on P9–P10. The proportion of dying cells gradually decreased thereafter (Figs. 1, 3, center left). The labeled cells were not distributed uniformly throughout the layer. They were more frequent in the middle of the INL, and 25% of the labeled cells were in the region adjoining the OPL. These latter cells probably corresponded to inner photoreceptor cells that failed to migrate through the OPL to the ONL (Young, 1984). The pattern of apoptosis also changed with the developmental stage. On P8, most of the labeled cells were located in the first third (from the optic nerve to the ora serrata) of the half retina. On P9, apoptotic cells were detected in the first two thirds of the half retina particularly in the second third. On P10, labeling was observed throughout the INL, mostly in the last two thirds of the periphery of the half retina. The apoptotic wave seemed to follow the differentiation wave from the optic nerve to the ora serrata. Between P8 and P13, a large number of INL cells underwent apoptosis, but the percentage of apoptotic cells decreased in the GCL and remained very low in the ONL. In the ONL, the dying cells were often adjacent to the OPL.

thumbnail image

Figure 1. Apoptosis throughout the development of the C57BL/6J mouse retina. Each layer is shown in a different color; each point corresponds to 1 day. Four peaks can be observed at stages embryonic day (E) 10.5, postnatal day (P) 2, P9, and P15. Error bars represent 95% confidence intervals of the percentages (see Statistical Methods section). NbL, neuroblastic layer; INbL, inner neuroblastic layer; ONbL, outer neuroblastic layer; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer.

Download figure to PowerPoint

thumbnail image

Figure 2. C57BL/6J retinas at different developmental stages, treated by the terminal deoxynucleotidyl transferase (TdT) -mediated deoxyuridine triphosphate (dUTP)–biotinylated nick end labeling (TUNEL) assay. Black arrows point to the apoptotic cells. A: Stage embryonic day (E) 10.5. Note that apoptosis is greatest in the middle of the retina. B: Stage postnatal day (P) 2: labeled cells are located in the INbL. C: Stage P9: many brown cells are visible in the middle of the INL. D: Stage P15: apoptotic cells are labeled in the ONL. NbL, neuroblastic layer; INbL, inner neuroblastic layer; ONbL, outer neuroblastic layer; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. Scale bars = 100 μm in A–D.

Download figure to PowerPoint

thumbnail image

Figure 3. Apoptosis levels in the three strains at the four important stages. Filled bars: C57BL/6J mice. Vertical hatched bars: gld mice. Horizontal hatched bars: bax-/- mice. Top left: NbL, stage embryonic day (E) 10.5. The two values are statistically different (P <0.034). ND: not determined. Top right: INbL, stage postnatal day (P) 2. Center left: INL, stage P9. Center right: ONL, stage P15. Bottom left: GCL, stage P15. Apoptosis level is lower in gld animals than in wild-type at stages E10.5, P2, and P15 but not P9, whereas it is lower in bax-/- animals than in wild-type for all stages studied. NbL, neuroblastic layer; INbL, inner neuroblastic layer; INL, inner nuclear layer; ONL, outer nuclear layer; GCL, ganglion cell layer; ND, not determined.

Download figure to PowerPoint

In addition to these three known apoptotic peaks, we observed a fourth event that has never been described before. This event occurred between P14 and P16 and concerned both the ONL and GCL. The percentage of apoptotic cells in the ONL, which remained low throughout postnatal development, increased on P14, reaching a peak at P15, and then decreased after P16 (Figs. 1, 2D). This unique apoptotic peak in the ONL coincided with a second peak in the GCL. After P18, as in adult retina sections, we observed very few apoptotic cells, if any, regardless of the layer considered (Fig. 1).

The animals begin to open their eyes at approximately P14–P15. An attractive hypothesis to account for the peak of cell death observed on P15 in the ONL, which contains the photoreceptor cells, is that the first exposure of the retina to light after the opening of the eyelid results in the positive selection of well-differentiated cells. We investigated this hypothesis by maintaining pups in the dark from P11 to P15 and compared the level of apoptosis in the three retinal layers with that in animals kept in normal light-cycle conditions over the same period. Similar numbers of dying cells were counted in the GCL and the ONL (data not shown). Thus, light exposure does not appear to play a role in the double peak of apoptosis in the ONL and the GCL on P15.

We then investigated the possible involvement of the BAX and FAS pathways in the apoptotic peaks (at stages E10.5, P2, P9, and P15) by studying retinas from bax knockout mice (Knudson et al., 2001) and from the gld strain, defective in the FASL (Lynch et al., 1994; Ramsdell et al., 1994) at these four stages.

The bax-/- strain was maintained by mating heterozygous parents; we, therefore, obtained bax-/-, bax+/-, and bax+/+ embryos. Due to their size, it was not possible to genotype these embryos at stage E10.5. Conversely, as all the gld mice were homozygous, no genotyping was necessary. For this reason, we compared only gld and B6 strains for experiments performed at stage E10.5.

At stage E10.5, there were half as many apoptotic cells in the NbL of gld mice as in the NbL of B6 mice (Fig. 3, top left). This difference is statistically significant. We also observed that the gld NbL contained more cells (approximately 1.5 times) than the B6 NbL. On P2, 0.99% of cells in the gld INbL and 0.07% in the bax-/- INbL were apoptotic, vs. 3.53% in the B6 INbL (Fig. 3, top right). The percentage of apoptotic cells in the INL at stage P9 was similar in gld and B6 (Fig. 3, center left). In contrast, the percentage of apoptotic cells at this stage in the bax-/- INL was one-sixth of that in the B6 INL. At stage P15, 0.23% of cells were apoptotic, in the ONL of gld mice and 0.16% in bax-/- mice, compared with 0.58% in B6 mice (Fig. 3, center right). In the GCL, the percentages of dying cell were also lower in the mutated mice than in the wild-type mice, with 0.42% of apoptotic cells in gld mice and 0.13% in bax-/- mice, vs. 2.23% in B6 mice (Fig. 3, bottom left).

These findings suggested that the BAX apoptotic pathway plays a more important role than the FASL pathway in the induction of apoptosis during normal retinal development. We, therefore, decided to study the expression of a major anti-apoptotic gene, bcl2. BCL2 prevents BAX-mediated apoptosis by forming heterodimers with BAX. It also prevents one of the death pathways triggered by FASL/FAS interactions, which results from BID change and its activation. We used in situ hybridization to study the expression pattern of bcl2. In B6 mice, bcl2 transcripts were detected in the optic cup and in the future brain at E10.5. At P2, low levels of bcl2 mRNA were detected in cells of the INbL of B6 mice (Fig. 4A). At P9, most of the B6 retinal cells containing bcl2 transcripts were located in the internal part of the INL, which is consistent with the location of TUNEL-labeled cells in the middle and external parts of the INL (Fig. 4B). The GCL cells expressed bcl2 at a very low level. At P15, bcl2 could be detected in the INL and cells of the ONL were lightly labeled (Fig. 4D). In adults, only the ONL cells contained bcl2 mRNA. These results are consistent with the TUNEL results.

thumbnail image

Figure 4. bcl2 expression pattern during the C57BL/6J retinal development. A: Stage postnatal day (P) 2. B: Stage P9. C: Stage P15. D: Adult. INbL, inner neuroblastic layer; ONbL, outer neuroblastic layer; GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. Scale bar = 200 μm in A (applies to A–D).

Download figure to PowerPoint

During retinal development, the INbL (then the GCL) and INL of bax-/- mice were thicker, after stages P2 and P9, respectively, than their counterparts in the control B6 retina. No such differences were observed between gld and B6 retinas.

This finding suggested that retinal cell death was effectively prevented in bax-/- mice, whereas it may only have been delayed in gld mice. We explored this possibility further by comparing the retina of adult wild-type B6, bax-/-, and gld mice. We studied the retinal morphology of several adults (6 months old) of each strain. The general morphology of the retinal layers in gld mice was similar to that in control B6 mice (Fig. 5). The thickness of the retinal layer and the total number of cells in one microscopic field were also similar, in each layer (Fig. 6, top left and right, and bottom left). We also counted the numbers of cells in the IPL, which is normally almost acellular. This layer contained only a few more cells in gld mice than in B6 mice (Fig. 6, bottom right). The morphology of bax-/- retinas was very different (Figs. 5, 6, top left and right, and bottom left). The INL of bax-/- mice was thicker than that of B6 mice and the GCL of bax-/- mice also contained more cells than did that of B6 mice. These observations were supported by the total number of cells in a microscope field, which indicated that the INL and GCL of bax-/- animals contained twice as many cells as did those of B6 animals. The IPL of bax-/- animals contained almost three times as many cells as did that of B6 mice (Fig. 6, bottom right).

thumbnail image

Figure 5. Photographs of adult retinas from the three strains at ×1,000 magnification. A: Control retina. B:gld retina. C:bax-/- retina. The GCL and INL of the bax-/- retina are thicker than those of the C57BL/6J retina, whereas the gld retina appears to be similar in size as the control. ONL, outer nuclear layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bar = 100 μm in A (applies to A–C).

Download figure to PowerPoint

thumbnail image

Figure 6. Number of cells in layers of normal and mutant adult retinas. Filled bars: C57BL/6J mice. Vertical hatched bars: gld mice. Horizontal hatched bars: bax-/- mice. Top left and right, and bottom left: Total number of cells in a field (magnification ×1,000) for the three strains studied in GCL, INL, and ONL. Bottom right: Mean number of cells per field in the IPL. Bax-/- GCL and INL contained twice as many cells as the C57BL/6J layers. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; IPL, inner plexiform layer.

Download figure to PowerPoint

We then investigated whether the bax-/- retina, which contained two times more cells in the GCL and INL than the B6 retina, was functional. We recorded and compared the electroretinogram (ERG) of B6, bax+/+ and bax -/- mice. In an ERG, the a-wave corresponds to photoreceptor cell activity, and the b-wave, to INL activity. The latency indicates the excitability of the cells, whereas the amplitude reflects the number of cells responding to the stimulus (Peachey et al., 1997). The ERGs for bax+/+ and B6 of the same age (6 months) showed similar latencies and amplitudes for the a- and b-waves. bax-/- animals of the same age also presented a latency similar to the controls for the a- and b-waves. However, the amplitude of the b-wave was much smaller in bax-/- animals than in controls (448.3 ± 45.6 mV in B6 mice, vs. 111.9 ± 32.4 mV in bax-/- mice for maximal flash intensity in the rod test protocol; 128.6 ± 5.6 mV for B6 vs. 44.5 ± 4.1 mV in bax-/- for maximal flash intensity in the cone test protocol; Peachey et al., 1997). These results indicated that the bax-/- INL cells that responded were as excitable as those in normal retina but that fewer INL cells were able to respond to the visual stimulus in these mice than in B6 mice.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Several groups have described the incidence of apoptosis during neural retina development, but each group used a different mouse strain, developmental stages, and techniques to detect apoptotic cells (Young, 1984; Portera-Cailliau et al., 1994; Laemle et al., 1999; Trousse et al., 2001). Thus, the first aim of our work was to provide a comprehensive kinetic analysis of apoptosis during the whole normal retinal development in the mouse. Considerable differences in the pattern of apoptosis have been reported between different individual animals at the same developmental stage (Vogel and Moller, 1980; Young, 1984; Portera-Cailliau et al., 1994). To reduce this variability, we used congenic B6 mice that were killed at the same time of day and we counted the cells in several retinal sections from each animal. Mixed model statistical analysis indicated that the main sources of variability were the developmental stage and the layer studied, and that interindividual variability was not statistically significant.

We confirmed and further explored the two previously reported peaks of apoptotic cell death at stages P2 and P9. These two peaks were proposed to correspond to important periods in the establishment of neuronal connections for ganglion cells (at P2) and interneurons (from P6 to P14; Young, 1984). We also report, for the first time in normal mice, a high level of cell death in the NbL of the optic cup, particularly in the middle of the NbL, at stage E10.5. It is known that the timing of NbL differentiation differs between pigmented and unpigmented embryos (Rachel et al., 2002). However, a similar pattern of apoptosis has been reported to occur at E10–E11 in the ZRDCT-N and BALB/c strains, which are pink-eyed diluted (p-/-) and albino tyr-/- (Laemle et al., 1999; Trousse et al., 2001). The invagination of the optic cup at E10.5 may involve the disappearance of cells in the middle of the retina, and this peak, therefore, may represent a morphogenetic type of cell death (Strettoi and Volpini, 2002). We also identified a previously unreported apoptosis peak occurring between P14 and P15 in both the GCL and ONL. The level of apoptosis in P15 retinas was similar in mice kept in the dark; therefore, exposure to light after the opening of the eye lids cannot be directly responsible for the cell death detected in the ONL. Apoptosis on P14 and P15 may result from a lack of photoreceptor connection, with only the well-connected photoreceptor cells remaining alive. The apoptosis occurring in the GCL and INL (Fig. 1) at this stage may be secondary to the death of rods and cones, with the neuronal cells connected to dying photoreceptor cell also being eliminated.

We also investigated the molecular mechanisms involved in apoptosis induction during retinal development. For this, we compared wild-type B6 mice, bax-/- mice, which lack the major pro-apoptotic BAX protein, which induces death at the mitochondria level, upstream the release of cytochrome c and Caspase 9 activation, and gld mice, which lack functional FASL, responsible for death signaling through the engagement of the FAS cell surface receptor leading to Caspase 8 activation upstream of the mitochondria. The absence of functional FASL or BAX resulted in a significant decrease in the percentages of apoptotic cells in the ganglion cells at P2, but this decrease was greater in bax-/- mice than in gld mice; almost no apoptotic cells were detected in the bax-/- mice. This finding is consistent with those for transgenic mice overexpressing bcl2, which encodes a major antagonist of BAX (Strettoi and Volpini, 2002). We also found that a lack of BAX led to much lower rates of apoptosis in the INL between P8 and P11, whereas a lack of functional FASL had no effect. Thus, the BAX-mediated pathway appeared to play a more important role than the FASL-mediated pathway in apoptosis induction during these crucial retinal developmental stages. Another major difference between these two pathways was the numbers of cells surviving in the retina after each developmental stage and at the end of development.

At each developmental stage studied, no morphologic differences were observed between the retinas of gld and B6 mice, and the numbers of cells in each layer of the adult gld and B6 retinas were similar. Thus, although apoptosis was less extensive at any given developmental time point in gld retinas, the same number of cells appear to die in total. These results suggest that the lack of a functional FAS pathway delayed the induction of apoptosis rather than actually preventing it. This observation is similar to those published for the interdigital epithelial cell death in apaf-1 knock-out mice (Cecconi et al., 1998). APAF-1 is involved in Caspase 9 activation, after the mitochondrial release of cytochrome c (Hengartner, 2000). Apaf-1-/- embryos display persistent interdigital epithelial cells, which normally undergo apoptosis at E15.5. However, in apaf-1-/- embryos, these die in the subsequent days by means of a caspase-independent necrotic pathway, which cannot be detectable by the TUNEL assay (Chautan et al., 1999). In several circumstances, caspase activation is known to only accelerate cell death and is not required for its execution (Wyllie and Golstein, 2001; Ameisen, 2002). However, if cell death occurs in the absence of caspase activation, the nuclear phenotype of apoptosis is usually lacking (Wyllie and Golstein, 2001; Ameisen, 2002), precluding detection by the TUNEL assay. A similar process may occur during retinal development in gld mice. Accordingly, the lack of engagement of the FAS pathway will delay the induction of cell death and prevent the induction of an apoptotic phenotype, but the cells may still die due to the limited availability of a survival factor, triggering BAX- or BAK-mediated mitochondria death pathways (Hengartner, 2000; Martinou and Green, 2001).

In contrast with gld mice, bax-/- mice displayed a completely different pattern of apoptosis during retinal development. Cells that did not die at the normal developmental stage seemed to remain alive until adulthood. It is difficult to determine the proportion of cells that die during the entire period of retinal development when studying the snapshots taken at given time points. One way of assessing the death rate is to compare the thickness of the B6 INL between stage P9 (approximately 100 μm) and adulthood (approximately 40 μm; Figs. 3, center left, 5A). This means that more than 50% of the cells produced during the retinal development in normal mice are destined to die. A similar finding was obtained for the brain, in which between 50% and 70% of cells die during development (Young, 1984). Comparisons of the thickness of the INL and the GCL in bax-/- and B6 adult animals (Fig. 5) showed that there were twice as many cells in these two inner layers in bax-/- mice than in B6 mice. This finding strongly suggests that the cells that survive abnormally in the retina in the bax-/- mice are not counterselected, as if these cells are no longer dependent on exogenous survival factors. The dysregulation of cell number, as shown by the abnormality of the ERG, affects the vision of adult bax-/- animals. The large number of excess cells may disturb the normal connections formed by the neuronal circuitry of the retina. Our findings in bax-/- mice are consistent with those recently published on bcl2-overexpressing mice: in adult retinas, the ONL was as thick as in B6 retinas and the INL and GCL contained two-times more cells than B6 INL and GCL (Strettoi and Volpini, 2002). BCL2 is an anti-apoptotic protein that prevents the mitochondria-dependent death induced not only by BAX, but also by BAK (Lindsten et al., 2000). The similar increase in cell number and the morphologic abnormalities in the retinas of bax-/- mice and of mice overexpressing bcl2 suggest that BAK may not play an important role in retinal cell death during development and that BAK cannot induce cell death in the absence of BAX in this organ. However, the results of our functional analysis of bax-/- adult retina differed from those obtained with bcl2-overexpressing mice. Indeed, adult bcl2-overexpressing mice seem to have normal vision and normal ERG results, despite having abnormal numbers of retinal cells and abnormal retina morphology (Porciatti et al., 1996; Gianfranceschi et al., 1999). In contrast, we observed abnormal ERG activity due to an abnormal number of cells responding to light. It is tempting, therefore, to speculate that the overexpression of bcl2 not only prevents BAX-mediated cell death, but may also confer additional differentiation and functional properties on the surviving retinal cells. For example, in other cell types, bcl2 overexpression has been reported to have effects other than the prevention of cell death, such as cell cycle down-regulation (Vairo et al., 2000) and changes in mitochondria energy production (Karlsson et al., 2002).

We show here that the BAX pathway plays an important role in developmental apoptosis in the ONL and GCL. However, the abnormal photoreceptor cell apoptosis occurring during the rd retinal degeneration in adult mice (Nambu et al., 1996; Mosinger Ogilvie et al., 1998) does not involve BAX (or FAS). Thus, the abnormal apoptotic processes involved in adult retinal degeneration do not appear to use the pathways involved in normal photoreceptor death during development. In GCL cells, the only available model of abnormal apoptosis induction in adult mice is after optic nerve transsection (axotomy). BCL2 can protect against this cell death (Chierzi et al., 1998). As axotomy may involve the same neurotrophic factor deprivation process that occurs in ganglion cells during development, a similar BAX-mediated pathway may be responsible for cell death in both circumstances.

In conclusion, our findings demonstrate that BAX plays a crucial role in normal retinal development and in the modeling of a functional adult retina. Cell death during retinal development, like that during brain development, may be crucial for the selection of the cellular differentiation and connection processes required for the emergence of visual function.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Animals

All animals were handled in accordance with the ARVO Statement for the Use of Animals in Ophthalmology and Vision Research. The reference mouse strain chosen was the C57BL/6J (B6) strain. Gld and bax-/- mouse strains with a C57BL/6J genetic background were also used (Lynch et al., 1994; Ramsdell et al., 1994; Knudson et al., 1995). For these two mutated strains, the Jackson Laboratory recommends the use of B6 as a wild-type control. Embryonic stages were numbered with day 0.5 taken as the morning of vaginal plug observation. P0 was the day of birth. Pregnant females and offspring were killed by lethal chloral hydrate injection or decapitation at the indicated time points, always at the same hour, to prevent variations due to the circadian cycle.

Breeding and Genotyping of Mice With Different Doses of bax

Mice heterozygous (+/-) for bax were mated to obtain bax+/+, bax+/- and bax-/- F1 animals. At chosen stages, a tail fragment was cut, DNA was isolated from this fragment and screened as previously described (Knudson et al., 1995). We found no differences between bax +/+ animals and wild-type B6, thus we only show the results for B6.

Tissue Preparation

Before the experiments, we compared the number of TUNEL-labeled cells in retinas from perfused and not perfused animals and found no differences (data not shown). After the animals were killed, E10.5 to E16.5 embryos were removed and kept intact. In the case of E17.5 and E18.5 embryos, the heads were cut, the skin removed, and the cornea perforated with a needle to allow the 4% PFA to enter the eye. In the case of the postnatal mice, eyes were dissected and cornea perforated as above. All samples were then fixed in 4% PFA at 4°C for at least 36 hr. The tissues were then embedded in paraffin and cut into 5-μm sagittal sections.

TUNEL

For each stage, we used at least two animals (from two different litters) and at least seven sections per animal. For the E17.5 to adult mice, we chose only sections including the optic nerve. For embryo sections, we chose the ones with the largest diameter. The TUNEL assay was performed with the DeadEnd Colorimetric kit (Promega), according to the manufacturer's protocol. Thus apoptotic cells were identified by the presence of a brown product in the nuclei and nuclear fragments. We used a slide from an E15.5 embryo as a positive control: the embryonic liver is a hematopoietic organ at this stage and contains many cells undergoing apoptosis. Negative control sections were incubated with the labeling mix lacking the TdT enzyme. Slides were counterstained for 3 min with 5% methyl green.

Quantitative Analysis

All slides were analyzed by the same individual. Two other individuals counted apoptotic cells independently on 30% of the slides and obtained similar results (less than 5% variation). The total number of cells in each layer was determined by counting the number of cells in a ×1,000 field and then multiplying by the number of fields for each section. All apoptotic cells were counted. We then calculated the percentage of apoptotic cells. In adults, we counted the total number of cells in a field of eight sections at ×1,000 magnification and calculated means.

Statistics

Apoptotic cell counts were analyzed using a generalized linear model with a Poisson distribution and a logarithm link (Nelder and Wedderburn, 1972). The log of the total cell count was included in the model as an offset. To take into account the clustering of the data by animal and by slide, we used the Generalized Estimating Equation approach to obtain a nonbiased variance of the estimates (Liang and Zeger, 1986). The covariates of the model were the layer and the day the animals were killed. Comparisons between layers at a fixed time or between times for a given layer were done by building and testing the corresponding contrast on the model parameters. To investigate the relative magnitude of the various sources of randomness, we also fitted a mixed model in which the animal and the slide were considered as random effects (Wolfinger and O'Connel, 1993). The GENMOD and GLIMMIX functions of the SAS 8.1 package were used for these analyses.

Study of P15 Animals Raised in Complete Darkness or Under a Normal Light Cycle

Litters of B6 mice obtained on the same day were maintained under a normal 12 hr/12 hr light cycle until P11. Half the pups were then transferred to complete darkness and half were kept in normal light cycle conditions until P15. All pups were killed at P15 and their eyes fixed as described above. For this study, we used retinal sections from animals of two different litters each.

In Situ Hybridization

Riboprobes were synthesized from the pGEM-T-easy plasmid (Qiagen), which contains a 920-bp fragment containing the entire coding sequence of the bcl2 cDNA. In situ hybridization was carried out as previously described (Mukhopadhyay et al., 2001), except that the conditions for anti-digoxigenin antibody incubation were changed (4°C overnight in this study). To ensure that the slides were comparable, we applied the same quantity of probe to each slide and treated all slides in a single experiment.

Electroretinogram

ERGs measure the electrical activity of rod and cone photoreceptors, which are the first cells to receive light signals, and of the INL cells, which treat the signals and transmit them to the brain by means of ganglion cells. C57BL/6J, bax-/-, and bax+/+ animals were analyzed. Mice were dark-adapted overnight before each recording. They were anesthetized with 20 μl/g of avertin. ERGs were recorded unilaterally, as previously described (Peachey et al., 1997).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Stanley Korsmeyer for generously providing us with the bax knockout mice. M.O.P. holds a MNERT fellowship.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES