Introduction

Apoptotic cell death is an essential component of normal development, particularly in the central nervous system (CNS), but apoptosis has also been implicated in disease states including neurodegenerative diseases. The bcl-2 family has recently emerged as central in regulating apoptosis through the balance of its pro- and antiapoptotic members.1 Bax and Bak, the only known systemically expressed multidomain proapoptotic bcl-2 family members, play an important role in both developmental and pathological apoptosis. Because of impaired developmental apoptosis, mice deficient in Bax and Bak retain interdigital webs, imperforate vaginas, and supernumerary CNS neural progenitor cells as well as rod photoreceptors in the retina.2, 3 Because of impaired pathological apoptosis, cultured mouse embryonic fibroblasts deficient in Bax and Bak are resistant to numerous apoptosis-inducing stimuli in vitro, including staurosporine exposure, growth factor withdrawal, and ultraviolet (UV) irradiation.4

Mice subjected to experimental retinal detachment undergo photoreceptor apoptosis, but Bax-deficient mice are completely rescued from this apoptosis.5 The cell death in this model is most likely hypoxia induced, as breathing high oxygen content air can partially protect photoreceptors.6, 7 Similarly, hypoxia-induced activation of the apoptotic pathway in vitro is rescued by Bax deficiency, and Bak activation in this system is dependent upon the presence of Bax.8 In contrast to the cell death induced by hypoxia, Bak can be activated in the absence of Bax when cell death is induced by staurosporine exposure, growth factor withdrawal, and UV irradiation.4 Similarly, in the retina, Bax deficiency is insufficient to protect photoreceptors from death induced by a mutation in the cyclic guanosine monophosphate (cGMP) phosphodiesterase,9 which may induce photoreceptor death by causing intracellular calcium overload.10

To determine whether combined Bax/Bak deficiency would protect photoreceptors in vivo in response to oxidative stress compared to Bax deficiency, we used the light damage model, which results in photoreceptor degeneration.11 In wild-type mice, light damage induces photoreceptor apoptosis in a c-fos-dependent manner12, 13, 14 detectable with terminal deoxynucleotidyl transferase dUTP nick end-labeling (TUNEL) and DNA-laddering analysis.15 While the precise mechanism of light damage is not yet known, the protective effects of various antioxidants16, 17 and the appearance and/or upregulation of markers of oxidative stress following light damage18, 19 have implicated photo-oxidative damage.

Light damage results in photoreceptor degeneration and has been used as a model to study some features of retinal degenerations including age-related macular degeneration (AMD) and retinitis pigmentosa (RP), blinding diseases that currently have no successful treatments. Because of the laminar organization of the normal retina with its defined number of photoreceptors, retinal light damage is an ideal system to investigate potentially protective in vivo effects of Bax/Bak deficiency.

It has been hypothesized that apoptotic cell death in neurodegenerations represents a pathologic reactivation of developmental apoptotic pathways. Overexpression of the antiapoptotic gene bcl-2, shown to reduce developmental apoptosis of neurons including ganglion cells,20 protects ganglion cells from apoptosis following pathologic in vivo axotomy.21 We have recently shown that deficiency of Bax and Bak blocks normal developmental apoptosis of rod photoreceptors,3 and overexpression of bcl-xL, an antiapoptotic bcl-2 family member, blocks developmental lead exposure induced photoreceptor cell death.22

Bcl-2 family members may similarly be involved in additional photoreceptor degenerations. Overexpression of the antiapoptotic gene bcl-2 protects photoreceptors from pathologic stimuli in vivo, including genetically induced retinal degenerations23 and photic injury.23, 24 Protection from genetically induced retinal degeneration is augmented by coexpression of bcl-2 binding protein BAG-1.25 Bcl-2-mediated rescue is incomplete and transient, however, possibly because of the toxic effects of bcl-2 overexpression on photoreceptors.23 Overexpression of antiapoptotic gene bcl-xL is not toxic to photoreceptors but does not rescue photoreceptors from photic injury or genetic mutations.24 Bcl-2 inhibits the proapoptotic activity of Bax and Bak,26 suggesting that, similar to overexpression of bcl-2, deficiency of Bax and Bak may protect against light damage. Unlike bcl-2 overexpression, deficiency of Bax and Bak does not cause photoreceptor degeneration3 and may confer a more complete rescue than bcl-2 overexpression. The purpose of this study is to investigate whether deficiency of Bax and Bak protects photoreceptors from light-induced cell death in vivo.

Results

Protection from DNA cleavage

To determine if deficiency of Bax and Bak protected retinas from light damage-induced cell death, bax−/−bak−/− (n=3) and bax−/− (n=3) retinas subjected to bright light exposure were analyzed by TUNEL 24 h later. Both bax−/− and bax−/−bak−/− retinas contained TUNEL-positive photoreceptors (Figure 1). To quantify the difference in cell death, the numbers of TUNEL-positive photoreceptors were counted in three different sagittal sections per retina through the optic nerve head (ONH). Bax−/−bak−/− retinas contained significantly fewer TUNEL-positive photoreceptors than bax−/− retinas, with mean±standard error of the mean (S.E.M.) of 53.0±9.7 versus 100.9±8.2, P<0.0007 (Figure 1c).

Figure 1
figure 1

Fewer TUNEL-positive photoreceptor nuclei 24 h following light damage in bax−/−bak−/− retinas compared to bax−/− retinas. (a, b) Fluorescence photomicrographs of TUNEL-labeled retinas in the sagittal plane at the ONH demonstrate fewer TUNEL-positive nuclei (red) in bax−/−bak−/− (a) compared to bax−/− (b) retinas 24 h following light damage. All nuclei are counterstained with DAPI (blue). Photoreceptor nuclei reside in the outer nuclear layer (ONL). Bar: 100 μm. (c) Histogram comparing numbers of TUNEL-positive photoreceptors from bax−/−bak−/− (n=3) and bax−/− (n=3) mice. The histogram displays the mean (±S.E.M.) of total numbers of TUNEL-positive photoreceptors counted in three different sections per retina cut in the sagittal plane

Full size image

Preservation of photoreceptor nuclei and inner/outer segments

Light-damaged bax−/−bak−/− versus light-damaged bax−/−

To determine whether the decrease in the number of TUNEL-positive photoreceptors 24 h following light damage corresponded to a longer-term preservation of photoreceptors, morphologic analysis was performed 1 week following light damage. Photoreceptors from noninjured bax−/−bak−/− and bax−/− mice have no significant differences.3 At 1 week following light damage, bax−/−bak−/− eyes demonstrated substantial morphologic preservation of photoreceptors compared to bax−/− eyes; bax−/− retinas compared to bax−/−bak−/− retinas had profound thinning of the photoreceptor nuclear layer and disorganization and shortening of the photoreceptor inner/outer segments (Figures 2 and 3, Table 1). Both the bax−/− and bax−/−bak−/− retinas were most severely affected centrally near the ONH, with relative preservation of their peripheral retinas. The bax−/−bak−/− photoreceptors appeared normal beyond approximately 750 μm from the ONH, while the damage to bax−/− retinas extended further to approximately 1500 μm from the ONH. The bax−/−bak−/− photoreceptors within 1500 μm from the ONH were significantly protected compared to bax−/− photoreceptors; the bax−/−bak−/− photoreceptor nuclear layer within 1500 μm contained 2.86±0.84 more nuclei per column compared with bax−/− retinas in the same region (P<0.0001) and 11.5±3.68 μm longer inner/outer segments (P=0.002) (Table 1).

Figure 2
figure 2

Near complete morphologic protection 7 days following light damage in bax−/−bak−/− but not bax−/− retinas. (a) Photomicrographs of plastic sections from bax−/−bak−/− and bax−/− retinas 7 days following light damage. Sagittal plane sections pass through the ONH. Both the inner and outer retinas of bax−/−bak−/− mice are thicker than that of bax−/− mice. The bax−/−bak−/− inner retina is thicker because of impaired developmental neuronal regression, while the bax−/−bak−/− outer retina, namely photoreceptors in the outer nuclear layer (ONL) and inner/outer segments (IS/OS, labeled with *), are protected from light damage-induced thinning observed in the bax−/− outer retina. Bar: 150 μm. (b, c) Plot of the thickness of the ONL, measured in numbers of photoreceptor nuclei per column (b), or of the thickness of the inner/outer segments (IS/OS), measured in micrometers (c). Measurements were made in triplicate every 150 μm away from the ONH. Injured bax−/−bak−/− (n=3, dark blue) and bax−/− (n=2, red) retinas 7 days following light damage and noninjured age-matched bax−/−bak−/− retinas (n=3, light blue) are displayed as mean values (±S.E.M.). Noninjured age-matched bax−/− retinas (not shown) are no different from noninjured bax−/−bak−/− retinas

Full size image
Figure 3
figure 3

Protection against photoreceptor cell loss and inner/outer segment shortening and disruption in bax−/−bak−/− compared to bax−/− retinas 7 days following light damage. (a–f) Photomicrographs of 1 μm thick plastic sections at the indicated distances superior to the ONH from bax−/− (a, c, e) and bax−/−bak−/− (b, d, f) retinas 7 days following light damage. At 150 μm from the ONH (a, b), bax−/−bak−/− and bax−/− retinas have similar numbers of photoreceptor nuclei (ONL) and inner/outer segment (IS/OS) lengths (indicated by double-headed arrow), although the inner/outer segments in the bax−/− retina are disorganized, unlike those in the bax−/−bak−/− retina. At 1000 μm from the ONH (c, d), protection conferred by deficiency of Bax and Bak is demonstrated by the marked photoreceptor cell loss and IS/OS shortening and disruption not observed in the bax−/−bak−/− retina. Light damage is minimal in the periphery, and by 2000 μm from the ONH (e, f), bax−/− retinas have a normal number of ONL photoreceptors and a normal IS/OS length and organization, similar to bax−/−bak−/− retinas. Bar: 25 μm

Full size image
Table 1 Comparison of numbers of photoreceptor nuclei per column and photoreceptor inner/outer segment length (μm)
Full size table

Injured bax−/−bak−/− versus noninjured bax−/−bak−/−

To determine whether deficiency of Bax and Bak afforded complete protection against light damage, bax−/−bak−/− retinas exposed to light damage were compared to nonexposed bax−/−bak−/− retinas. While there was no statistically significant difference in inner/outer segment lengths, injured bax−/−bak−/−retinas had 0.9±0.47 fewer photoreceptor nuclei (P=0.04) within 1500 μm from the ONH compared to noninjured bax−/−bak−/− retinas (Figure 2, Table 1).

Retinal function assessed with electroretinography

Compared to age-matched wild-type controls (n=5), adult bax−/− (n=2) and bax−/−bak−/− (n=6) eyes had no significant loss of rod phototransduction as indicated by the saturating amplitude of the rod a-waves. Mean±S.E.M. scotopic a-wave amplitudes were 324±61 μV for wild-type, 271±24 μV for bax−/−, and 256±96 μV for bax−/−bak−/− retinas. Consistent with the nearly normal function of the rod photoreceptor cells, scotopic b-wave amplitudes, which are the field potential driven mainly by rod bipolars downstream of photoreceptors, were not significantly different in bax−/− (272±68 μV) and bax−/−bak−/− (202±105 μV) retinas compared to wild-type retinas (301±75 μV). While the number of animals used for the electroretinogram (ERG) analysis prohibits detailed statistical analysis, the electroretinographic data of the knockout mice show that retinal function of these mice is near to normal and establish that the absence of the bax and bak genes does not necessarily lead to loss of physiological function.

Discussion

Combined deficiency of proapoptotic bcl-2 family members, Bax and Bak, imparts resistance to developmental apoptosis2, 3 and in vitro pathologic apoptosis.4 Bax deficiency alone is sufficient to protect completely photoreceptors from hypoxia-induced death caused by retinal detachment5 and to protect cells from hypoxia in vitro,8 but is not sufficient to protect cells in culture4 and photoreceptors in vivo from other types of apoptotic insult.3, 9 We demonstrate here that combined deficiency of Bax and Bak in murine retinas significantly protects photoreceptors from degeneration induced by acute light damage relative to Bax deficiency; bax−/−bak−/− retinas exposed to light damage had decreased photoreceptor cell death with resulting preservation of photoreceptor morphology compared with bax−/− retinas.

The photoreceptor protection observed with deficiency of Bax and Bak is consistent with previous studies, which have demonstrated protection from light damage by overexpression of the antiapoptotic gene bcl-2.23 Bcl-2 overexpression, however, was toxic to photoreceptors, and resulting protection was incomplete with 3–4 rows of nuclei remaining after 14 days of constant light exposure. In contrast, deficiency of Bax and Bak does not cause photoreceptor death.3 Instead, it results in nearly complete protection 7 days following acute light damage; bax−/−bak−/− photoreceptors exposed to light damage retained approximately 9–10 rows of nuclei, and morphometric analysis was required to detect a small difference of only 0.90±0.47 fewer nuclei in injured compared to noninjured bax−/−bak−/− retinas.

Protection of photoreceptor morphology and number of nuclei in bax−/−bak−/− retinas was more significant than reduction of TUNEL-positive photoreceptors. This result can be explained by the observation that following light damage more photoreceptors become TUNEL positive than ultimately die.27 Thus, many of the TUNEL-positive photoreceptors in bax−/−bak−/− retinas may have nonlethal DNA nicks, perhaps resulting from photo-oxidative stress.

While the precise mechanism of light damage is unclear, functional rhodopsin mediates its pathological consequences.28 We performed ERGs on 2-month-old bax−/−bak−/−, bax−/−, and wild-type animals to determine whether the protection observed in the bax−/−bak−/− retinas was due to decreased phototransduction. There were no significant differences in the rod a-waves among animals, suggesting that differences in photoreceptor phototransduction do not account for the photoreceptor protection in bax−/−bak−/− mice. Another ERG study on 6-month-old bax−/− and wild-type mice similarly demonstrated normal photoreceptor responses.29 This study also detected decreased b-wave amplitudes in bax−/− mice, indicative of decreased numbers of responsive INL bipolar cells. The b-wave amplitudes in our bax−/− and bax−/−bak−/− mice were similar to wild type. This discrepancy is most likely a result of age-related changes that reduce bipolar responsiveness in the older, 6-month-old mice but not the 2-month-old mice in our study. This study and our current study are consistent in the finding of normal a-waves, suggesting that changes in bipolar responsiveness are independent of any changes in rod photoreceptor phototransduction.

With normal photoreceptor phototransduction, the decreased photoreceptor cell death in bax−/−bak−/− retinas most likely resulted from loss of the proapoptotic functions of Bax and Bak. Deficiency of both Bax and Bak in cultured MEFs, but not either one individually, imparts resistance to apoptotic stimuli in vitro,4 and increases in Bax and Bak proteins have been observed in PC12 cells exposed to H2O2-induced oxidative stress.30 Resistance of bax−/−bak−/− photoreceptors to photo-oxidative injury is consistent with these in vitro results.

The bax gene is closely linked to pigment-related loci pink-eyed dilution (p) and tyrosinase (Tyr), and mice homozygous for bax deletion (bax−/− and bax−/−bak−/− mice) are lightly pigmented despite their C57BL/6 background (Jax Bulletin Number 2 (1999), from http://jaxmice.jax.org/info/bulletin/bulletin02.html). Consequently, we were unable to compare light damage in the lightly pigmented bax−/−bak−/− and bax−/− mice to heavily pigmented wild-type C57BL/6 mice and bak−/− mice. Melanin pigment in the iris, retinal pigment epithelium, and choroid absorbs light and protects the retina from light damage, and these wild-type mice are thus highly resistant to light-induced photoreceptor cell death. Because we compared light damage in bax−/−bak−/− and bax−/−, but not wild-type or bak−/− retinas, we were unable to make any conclusions about the individual roles of Bax and Bak in mediating light damage-induced photoreceptor apoptosis; we can only conclude that Bax deficiency is insufficient to protect completely photoreceptors from light damage and that combined Bax/Bak deficiency affords greater protection. Bax and Bak serve overlapping functions in developmental photoreceptor apoptosis such that loss of both is required for maximal impairment of developmental cell death, and it is likely that they serve similar roles in light damage-induced apoptosis.

The bax−/−bak−/− mice have poor viability because of severe developmental abnormalities, and less than 10% of pups survive beyond the neonatal period. Thus, only a small number of adult mice were available for study. We are currently generating conditional, retina-specific knockout mice. Their improved viability will allow us to dissect the individual roles of Bax and Bak in light damage and determine the extent of protection afforded by Bax/Bak deficiency compared to wild type.

We have shown previously that Bax and Bak are important mediators of developmental photoreceptor apoptosis,3 and this current study provides evidence that Bax and Bak are similarly important in mediating light damage-induced cell death. Apoptosis represents the final common pathway of many retinal and other neurodegenerations, and targeting both Bax and Bak is therapeutically attractive. The protection of Bax/Bak-deficient mice from light damage demonstrates a role for these proapoptotic factors in an in vivo form of pathological death and warrants investigation into the involvement of Bax and Bak in other neurodegenerative processes.

Materials and Methods

Mouse light damage

The bax−/− and bax−/−bak−/− mice were generated in a C57BL/6 background as previously described.2 Mice were reared with a light–dark cycle of 12 h. Adult (2–3 months) mice were dark adapted for 12 h, pupils were dilated with topical atropine 1%, and mice were exposed to 10 k lux of cool white fluorescent light in a well-ventilated room for 7 h from 0200 to 0900 as described.19 Eyes were enucleated immediately following killing after either 24 h following light damage for TUNEL analysis or 7 days following light damage for morphologic analysis. All procedures concerning animals were approved by the University of Pennsylvania Institutional Animal Care and Use Committee and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

TUNEL analysis

Eyes enucleated 24 h following light damage were immersion fixed in 4% paraformaldehyde for 2 h. Eyecups generated by removal of the anterior segment were cryoprotected overnight in 30% sucrose and then embedded in Tissue-Tek OCT (Sakura Finetek, Torrance, CA, USA). Cryosections 10 μm thick were cut in the sagittal plane through the ONH. The fluorescein-conjugated TUNEL in situ cell death detection kit (Roche, Mannheim, Germany) was applied to sections according to the manufacturer's protocol. TUNEL-labeled sections were analyzed by fluorescent microscopy using a Nikon Eclipse TE-300 microscope (Nikon Inc., Melville, NY, USA), and digitized images were acquired with a Spot RT Slider camera (Diagnostic Instrument Inc., Sterling Heights, MI, USA) with ImagePro Plus v.4.1 software (Media Cybernetics, Silver Spring, MD, USA). For each retina, the number of TUNEL-positive photoreceptors was counted in three sagittal sections through the ONH. The mean number of TUNEL-positive photoreceptors per bax−/−bak−/− retina (n=3) was compared to that from bax−/− retinas (n=3) by the Student's t-test.

Morphologic analysis

Eyes enucleated 7 days following light damage were immersion fixed in 2% paraformaldehyde/2% glutaraldehyde overnight. Eyecups generated by removal of the anterior segment were embedded in JB-4 plastic according to the manufacturer's protocol (Polysciences Inc., Warrington, PA, USA). Semi-thin sections 1 μm thick were cut in the sagittal plane through the ONH and stained with 30% Richardson's methylene blue/azure II.

Digitized images of each section were acquired with a light microscope as described above. The number of nuclei per column of outer nuclear layer (ONL) photoreceptors was counted in triplicate at 150 μm intervals from the ONH to 2100 μm from the ONH. The thickness of the inner/outer segments was also measured in triplicate at the same intervals, using image analysis software (ImagePro Plus, ver 4.1; Media Cybernetics) to calculate distances from manually set lengths.

Statistical comparisons were performed by calculating the means of each measurement (photoreceptor nuclei count and inner/outer segment thickness) for injured bax−/−bak−/− (n=3) and bax−/− (n=2) retinas and for noninjured bax−/−bak−/− retinas (n=3) at each distance from the ONH. Injured bax−/−bak−/− mice were compared to either injured bax−/− mice or noninjured bax−/−bak−/− mice. The S.E.M. and the differences in each measurement, along with their 95% confidence intervals and P-values, were calculated from the generalized equation estimate (GEE) with adjustments for inter-eye correlations and correlations among repeated measurements from the same eye.31 Because light damage injured the region within 1500 μm on each side of the ONH, statistical analyses were performed on data gathered from this region. All data analyses were performed in SAS 8.2 (SAS Institute Inc., Cary, NC, USA).

Electroretinography

Retinal function of adult (2–3 months) bax−/−bak−/− (n=6), bax−/− (n=2), and wild-type (n=5) mice was assessed by full field ERGs as previously described.32 Briefly, mice were anesthetized with xylazine/ketamine and their eyes dilated with 1% tropicamide (Alconox, New York, NY, USA). These mice were placed on a holding unit preheated to 38°C. A platinum wire electrode was placed on the cornea using 2.5% hydroxypropyl methylcellulose (Goniosol; Ciba Vision, Duluth, GA, USA) as a protector, and a reference tungsten wire electrode was placed into the mouth. These mice were placed inside a Faraday cage and exposed to a series of stimuli to test the field potential of rods and rod-driven retinal bipolar function.