Simultaneous Cell Death and Upregulation of Poly(ADP-Ribose) Polymerase-1 Expression in Early Postnatal Mouse Retina

Abstract

Purpose.: Poly(ADP-ribose) polymerase (PARP)-1 is a nuclear enzyme that transfers ADP-ribose units (PAR polymer) to nuclear proteins and has been implicated in caspase-independent cell death in different models of retinal degeneration. The involvement of PARP-1 in cell death occurring during normal postnatal development of the mouse retina was investigated. In addition, the expression of apoptosis-inducing factor (AIF), a caspase-independent cell death mediator, was explored because PARP-1 activation has been related to the translocation of a 57-kDa form of AIF into the cell nucleus.

Methods.: Cell death was determined in retinas of developing mice by both ELISA and TUNEL. PARP-1, PAR, and AIF were analyzed by immunocytochemistry and immunoblotting. Quantification of PARP-1 mRNA levels was also performed by real-time PCR.

Results.: PARP-1 upregulation and PAR polymer formation, indicative of PARP-1 activity, were observed during the first postnatal week simultaneously with the presence of abundant dying cells, some of which were not associated with active caspase-3. PARP-1 was downregulated and PARP-1 activity progressively declined in the retina during subsequent postnatal development, coinciding with the decrease in cell death. Truncated AIF (57 kDa) was present in the retina during the first postnatal week, gradually decreasing thereafter, and had a nuclear localization in some cells, which also showed strong PAR polymer nuclear staining.

Conclusions.: These results show that a caspase-independent cell death pathway exists during the normal development of the mouse retina and suggest that PARP-1 participates in this cell death pathway by mediating AIF translocation to the cell nucleus.

As in other regions, ¹ cell death in the retina can occur via a caspase-independent pathway. ^{2 –7} Caspase-independent death can be mediated by apoptosis-inducing factor (AIF), ^{8 –10} which is synthesized as a precursor protein (67 kDa) and inserted in the mitochondrial inner membrane as a 62-kDa mature form. In certain conditions (e.g., oxidative stress or DNA damage), the mature form is cleaved, giving rise to a 57-kDa truncated AIF form that translocates to the cell nucleus; once within the nucleus, truncated AIF is involved in cell death. ^{10 –15}  

Poly(ADP-ribose) polymerase (PARP)-1 is a nuclear enzyme critical for the maintenance of genomic integrity and the control of cell cycle and gene expression. ¹⁶ PARP-1 modulates the activity of a wide range of nuclear proteins by the transfer of ADP-ribose units (PAR polymers). ¹⁷ Proteins accepting PAR polymers include PARP-1 itself, DNA-binding proteins, and DNA repair proteins. ^{18 –20} PAR polymer formation by active PARP-1 is essential for AIF translocation to the nucleus. ^{9,10,15,21 –23}  

PARP-1 plays a pivotal role in multiple neurologic diseases and neurodegenerative disorders ^10,24 and in the retina. ^25,26 It was recently shown that inherited photoreceptor death depends on PARP-1 activation and AIF translocation. ²⁷  

Although PARP-1 has been linked to pathologic cell death in the retina, its possible role in physiological programmed cell death during retinal development is largely unknown. The present study shows that active PARP-1 and truncated AIF are upregulated in the mouse retina during the first developmental week, coinciding with the presence of abundant cell death. During the second week of development, when cell death is less frequent, PARP-1 and truncated AIF are downregulated. Hence, there appears to be a clear relationship between cell death and active PARP-1 expression, which may be linked to the induction of AIF translocation in caspase-independent cell death. 

Materials and Methods

Animals

C57BL/6 and BALB/c mice were provided by the Animal Experimentation Service of the University of Granada. Most experiments were performed on the C57BL/6 strain. Retinas were taken from mice at embryonic day (E) 18 and at postnatal days (P) 0 (day of birth), P3, P7, P14, P21, P28, and P60 (adulthood). Animals were maintained in a 12-hour light/12-hour dark cycle and had ad libitum access to food and water. All experimental procedures followed the guidelines of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the Research Ethics Committee of the University of Granada. 

Cell Death Detection by ELISA

Nucleosomes are released to the cytoplasm in dying cells; hence, their amount in the cytoplasm is a representative sign of cell death. To quantify free nucleosomes, both retinas of each animal were dissected, homogenized with a Dounce homogenizer to avoid damaging the cell nucleus, and centrifuged at 13,200 rpm for 10 minutes at 4°C. A portion of the supernatant, which contained cytoplasmic nucleosomes, was used to quantify protein concentrations by Bradford's method (Bio-Rad Protein Assay; Bio-Rad, Hercules, CA), whereas the remainder was analyzed using an assay kit (Cell Death Detection ELISA; Roche Diagnostics, Mannheim, Germany) containing anti–histone and anti–DNA antibodies to label nucleosomes, in accordance with the manufacturer's protocol. The intensity of the ELISA reaction, which indicates the amount of cytoplasmic nucleosomes, was measured in a reader (Multiskan Ascent; Thermo Fisher Scientific, Rockford, IL). 

Immunocytochemistry and TUNEL Assay

Primary antibodies, their sources, and their dilutions are summarized in Table 1. The anti–caspase-3 antibody recognizes the active form of caspase-3 but not the precursor inactive form. The anti–PARP-1 monoclonal antibody binds to an epitope located in the amino-terminal DNA-binding domain of PARP-1, the anti–PAR antibody was directed against PAR synthesized after activation of PARP-1, and the anti–AIF antibody was raised to a synthetic peptide corresponding to residues within the C terminus of AIF and recognizes the 67-, 62-, and 57-kDa forms. 

Table 1.

 

Sources and Dilutions of Primary Antibodies Used for Immunofluorescence and/or Western Blot Analysis

Antibody	Source	Catalog No.	Dilution for IF	Dilution for WB
Mouse monoclonal anti-PARP1	BD PharMingen (San Diego, CA)	556362	1:50	1:2000
Rabbit polyclonal anti-PARP1	Enzo Life Sciences (Plymouth Meeting, PA)	ALX-210–221	—	1:5000
Mouse monoclonal anti-PAR	Alexis Biochemicals (San Diego, CA)	804–220-R100	1:50	1:400
Mouse monoclonal anti-HMG-1 (HAP46.5)	Santa Cruz Biotechnology (Santa Cruz, CA)	sc-56698	—	1:1000
Rabbit polyclonal anti-AIF	Sigma (St. Louis, MO)	A7549	1:200	1:1000
Rabbit polyclonal anti-active caspase-3	R&D Systems (Minneapolis, MN)	AF835	1:50	—
Rabbit polyclonal anti-β-tubulin III	Sigma	T2200	—	1:10000

For immunostaining, the entire enucleated eye was fixed in periodate-lysine-paraformaldehyde ²⁸ for 6 hours at 4°C, cryoprotected in PBS containing 0.1% Triton X-100 (Sigma, St. Louis, MO) and 20% sucrose, and frozen in liquid nitrogen. Twenty-micrometer-thick cross-sections were permeabilized in 0.1% Triton X-100 in PBS and blocked with normal goat serum (Sigma) diluted 1:30 in PBS-1% bovine serum albumin (PBS-BSA) for 45 minutes. Then they were incubated overnight with the primary antibody at 4°C, washed in PBS, and incubated with the corresponding secondary antibody (Alexa Fluor 488-conjugated goat anti–mouse IgG [Molecular Probes, Eugene, OR] for monoclonal anti–PARP-1 and anti–PAR, and Alexa Fluor 488-conjugated goat anti–rabbit IgG [Molecular Probes] for anti-AIF) diluted 1:500 in PBS-BSA for 2 hours at room temperature. Sections were finally stained with either Hoechst 33342 (Sigma) or DAPI (Sigma), washed, and mounted (Fluoromount G; Southern Biotech, Birmingham, AL). As a negative control, primary antibodies were omitted in some sections. A similar procedure was used for active caspase-3 immunostaining, except that the eyes were fixed in 10% formalin and paraffin-embedded. Ten-micrometer-thick cross-sections were immunolabeled with the anti–active caspase-3 antibody and treated with the secondary antibody FITC-conjugated goat anti–rabbit IgG (Sigma) diluted 1:500 in PBS-BSA for 2 hours at room temperature. 

Pyknotic structures, revealed by Hoechst staining and either associated or not associated with active caspase-3 labeling, were counted in 20-micrometer-thick cryostat transverse sections of P7 retinas, including both central and peripheral areas, viewing 10 sections from four different animals under a microscope (Axiophot; Zeiss, Oberkochen, Germany) with a 40× objective. 

Localization of dead cells within the retina was investigated by the TUNEL technique. Cryostat sections were incubated in 2% TDT (Promega, Madison, WI) in TDT buffer (Promega) and 0.03% fluorescein-dUTP (Roche Diagnostics) for 1 hour at 37°C. After incubation, sections were washed with PBS and stained with nuclear dye Hoechst 33342 (Sigma). Some sections were double-labeled for active caspase-3 using Alexa Fluor 594-conjugated goat anti–rabbit IgG (Molecular Probes) as a secondary antibody and TUNEL. 

Confocal images were obtained with a microscope (TCS-SP5; Leica, Wetzlar, Germany). Images were stored in TIFF format and digitally prepared (Photoshop; Adobe Systems, San Jose, CA) by adjusting brightness and contrast. 

Western Blot Analysis

Whole protein extract of both retinas from one animal were used for each sample in PARP-1 and PAR polymer immunoblotting studies, whereas six retinas (from three mice) were pooled for each sample in AIF assays. Anti–β-tubulin antibody was used for normalization of the total proteins loaded. The primary antibodies used for Western blot analysis are summarized in Table 1. Retinas removed in PBS supplemented with protease inhibitor mixture (Roche Diagnostics) were resuspended in 50 to 100 μL lysis buffer (Tris HCl 50 mM, pH 8.0, EDTA 0.1 mM, Triton X-100 0.5%, β-2-mercaptoethanol 12.5 mM) for 30 minutes on ice and centrifuged at 13,200 rpm for 15 minutes at 4°C. After protein quantification (Bio-Rad Protein Assay; Bio-Rad), SDS PAGE-reducing sample buffer was added, and Western blot analysis was carried out using standard procedures; 25 μg (for PARP and PAR polymer) or 75 μg (for AIF) protein was loaded into each well of a SDS-polyacrylamide gel. The gels were run in a mini gel system (Bio-Rad), and proteins were transferred onto a polyvinylidene difluoride membrane (Immun-Blot PVDF Membrane; Bio-Rad). Blots were blocked with PBS containing 5% milk powder and 0.1% Tween-20 for 30 minutes and incubated overnight with the primary antibody. Blots were then incubated for 2 hours with the corresponding secondary antibody (peroxidase-conjugated anti–mouse IgG or peroxidase-conjugated anti–rabbit IgG; Sigma). Antibody reaction was revealed by chemiluminescence (ECL Western Blotting Detection System, Amersham Biosciences, Sunnyvale, CA). 

We also obtained nuclear protein extracts from six retinas to detect AIF translocation to the nucleus by using a previously reported method. ²⁹ Western blot scans were analyzed using ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). Densitometric analyses of Western blots were normalized with respect to β-tubulin. 

Determination of PARP-1 mRNA Expression by Real-Time PCR

Total RNA was isolated from retinas by reagent (Trizol; Invitrogen, Carlsbad, CA) extraction; 2 μg RNA was diluted in 20 μL retrotranscription reagent and used to synthesize cDNA with a cDNA synthesis kit (iScript; Bio-Rad). To quantitatively determine PARP-1 gene expression, real-time PCR analysis was performed (iQ SYBR Green Supermix and iCycler iQ detection system; Bio-Rad). Primers for PARP-1 detection were 5′-AGGCCCTAAAGGCTCAGAAT-3′ (sense) and 5′-CTAGGTTTCTGTGTCTTGAC-3′ (antisense). ³⁰ The expression of 18S rRNA (primers 5′-CGGCTACCACATCCAAGGAA-3′ [sense] and 5′-GCTGGAATTACCGCGGCT-3′ [antisense]) ³¹ ) was used as an endogenous control. Standard curves (coefficient of correlation >0.98) containing at least four concentrations (represented in triplicate) of a control cDNA were constructed for both the endogenous control gene and the gene of interest. For data analyses, the cycle threshold (CT) value was calculated in each case. Relative gene expression of PARP-1 was determined by the 2^–ΔΔCT method for real-time quantitative PCR. ³²  

Statistical Analysis

Quantitative data were expressed as mean ± SEM of at least three independent experiments for each time point. Statistical significance was determined by using the Student's t-test. 

Results

Increased Cell Death during the First Postnatal Week

Our results confirmed that extensive cell death, revealed by the amount of cytosolic free nucleosomes, occurs in the postnatal mouse retina during the first week after birth, as previously reported. ^33,34 The number of free nucleosomes remained elevated from P0 to P7, peaked at P7, and decreased at P14 and thereafter (Fig. 1). Hence, cell death levels in the retina significantly decreased after the first postnatal week. 

Figure 1.

 

Quantification of free cytosolic nucleosomes by ELISA throughout the postnatal development of the mouse retina. The amount of free nucleosomes, a sign of cell death, increases up to P7 and decreases thereafter. Data are expressed as mean ± SEM of three independent experiments. *P < 0.05 and **P < 0.01; significant differences between marked values and those in P14, P21, P28, and P60. OD, optical density.

Some Retinal Cells Die by an Alternative Caspase-Independent Pathway

Whereas Hoechst staining revealed nuclear morphologic changes typical of cell death, such as chromatin condensation and pyknosis, active caspase-3 indicated the occurrence of caspase-dependent cell death in the retina. Some pyknotic nuclei in P3 retinas were not associated with active caspase-3 (Fig. 2A). Similar observations were made in P7 retinas (Figs. 2B, 2C), in which 46.64% ± 1.49% (mean ± SEM; n = 10 sections from retinas of four animals) of the pyknotic structures were not related to active caspase-3 staining. Sections from P7 retinas double labeled with the active caspase-3 antibody and TUNEL assay confirmed that some TUNEL-labeled structures were not stained with anti–active caspase-3 (Fig. 2C). 

Figure 2.

 

Caspase-dependent and -independent pyknosis and TUNEL staining in confocal images of the postnatal mouse retina. (A, B) Active caspase-3 immunolabeling (green) and Hoechst nuclear staining (blue) in P3 (A) and P7 (B) retinas reveal the presence of pyknotic nuclei associated (white arrows) and not associated (red arrows) with active caspase-3. (C) Active caspase-3 labeling (CASP; red) combined with TUNEL assay (TUN; green) and Hoechst nuclear staining (HO; blue) in the inner nuclear layer of a P7 retina reveal that not all pyknotic or TUNEL-labeled nuclei are associated with caspase-3 activity. Merged images showing active caspase-3 and TUNEL (CASP+TUN), active caspase-3 and Hoechst (CASP +HO), and TUNEL and Hoechst (TUN+HO) signals demonstrate that though TUNEL labeling frequently coincided with caspase-3–positive pyknotic nuclei (white arrows), some of them were not related to active caspase labeling (red arrows). Images are representative of three retinas. GCL, ganglion cell layer; INL inner nuclear layer; ONL, outer nuclear layer. Scale bars: 5 μm (A), 50 μm (B), 10 μm (C).

These findings indicate that caspase-3–dependent apoptosis is present in the retina during the first postnatal week, but an alternative caspase-independent cell death mechanism also appears to operate. 

PARP-1 Expression in the Retina during the First Postnatal Week

AIF has been reported to participate in caspase-independent cell death in a way linked to the presence of PARP-1. ¹² Therefore, we analyzed the expression and activity of PARP-1 protein and mRNA. 

Immunoblots of retinal lysates showed that PARP-1 protein (113-kDa band) was present in high levels at P3 and P7 but decreased after P7 (Figs. 3A–D). Therefore, PARP-1 remained at high levels in the retina throughout the period of abundant cell death and decreased thereafter, though it was still present in the adult retina (Supplementary Fig. S1). Similar findings were obtained by using monoclonal (Figs. 3A–C) and polyclonal (Fig. 3D) PARP-1 antibodies and in both mouse strains (C57BL/6 [Figs. 3A–D] and BALB/c [Fig. 3E]). The presence of elevated PARP-1 protein levels at P3 and P7 coincided with enhanced PARP-1 mRNA levels in the retina during the first postnatal week (Figs. 3F, 3G), reflecting upregulation of the PARP-1 gene during this period. PARP-1 mRNA levels markedly decreased in P14 and P21 retinas, reaching values close to those found in adult retinas (Fig. 3F). 

Figure 3.

 

PARP-1 expression during postnatal development of the retina. (A) Western blot analysis of PARP-1 (by using the monoclonal anti–PARP-1 antibody) representative of three independent experiments. High levels of PARP-1 protein (113 kDa) are present at P3 and P7, decreasing thereafter. Bands at P28 and P60 are not seen because the immunoblot was exposed for a short time to avoid strong overexposure of bands at P3 and P7. Overexposed immunoblot showing the P60 PARP-1 band is seen in Supplementary Figure S1. The presence of PARP-1 fragments of 89 kDa at P3 to P7 indicates caspase activity (cleaving part of 113 kDa PARP-1). β-Tubulin was used as a loading control. (B) Densitometric analysis of complete PARP-1 form (113 kDa) in three different experiments. PARP-1 protein is increased in P3 and P7 retinal extracts, was strongly diminished in extracts older than P7, and was still present at P60. Densitometric values were expressed as relative levels with respect to the level at E18. *P < 0.01; significant differences between marked values. (C) Western blot analysis of PARP-1 expression (monoclonal anti–PARP-1 antibody) between P7 and P15 showing a gradual decrease in PARP-1 protein. The blot is representative of three independent experiments. (D) Western blot analysis of PARP-1 expression (using polyclonal anti–PARP-1 antibody) between P3 and P21 showed findings similar to those revealed with the monoclonal antibody. The blot is representative of three independent experiments. (E) Western blot analysis of PARP-1 expression in the retinas of P7 and P14 BALB/c mice. The blot is representative of three independent experiments and was obtained by using the polyclonal anti–PARP-1 antibody. Strong decrease in PARP-1 protein is seen in BALB/c mice between P7 and P14, similar to that observed in C57BL/6 mice. (F) Quantitative analysis of PARP-1 mRNA expression by real-time PCR using SYBR Green real-time analysis during retinal development. Histogram represents the CT (mean ± SEM) of three real-time PCR experiments conducted in triplicate; results were normalized to the expression of 18S rRNA. PARP-1 mRNA increases from P0 to P7, peaks at P7, and is subsequently downregulated. Note that the increase in PARP-1 mRNA coincides with increased PARP-1 protein levels. *P < 0.05 and **P < 0.0; significant differences between marked values. (G) Representative gel of three independent experiments on agarose gel electrophoresis of PARP-1 mRNA from P0, P7, P14, and P21 retinas. PARP-1 cDNA amplification products were run on 1.5% agarose gel, and the bands were digitalized. PARP-1 amplification products were referred to the corresponding 18S rRNA bands.

Immunofluorescence also revealed the presence of PARP-1 in the retina during the first postnatal week. PARP-1 immunolabeling was present in P3 to P7 retinas and was more abundant in the inner nuclear layer (Figs. 4A–D). PARP-1 immunolabeling was restricted primarily to the nucleus (Fig. 5). This immunolabeling pattern markedly changed at P14, when scarce PARP-1–positive retinal neurons were observed (Figs. 4E–H). Hence, PARP-1 immunoreactivity strongly decreased in the retina after the first postnatal week. 

Figure 4.

 

PARP-1 immunocytochemistry in confocal optical sections of postnatal retinas. (A–D) At P7, PARP-1–positive cells (A, C, green) are seen in both the inner (INL) and outer (ONL) nuclear layers, which are revealed by Hoechst staining (B, C, blue). (D) Negative control in which the primary antibody has been omitted. Right: Hoechst staining. (E–H) Scarce PARP-1 immunolabeling in the layers of P14 retinas; labeling of blood vessels is nonspecific and is present in the negative control (H). Sections of three different retinas were stained and analyzed for each time point. GCL, ganglion cell layer. Scale bars, 50 μm.

Figure 5.

 

Confocal optical sections of selected areas of the ganglion cell layer (GCL; A–C), inner nuclear layer (INL; D–F), and outer nuclear layer (ONL; G–I) in transverse sections of DAPI-counterstained, PARP-1–immunolabeled P7 retinas. Most of the PARP-1 immunolabeling is nuclear in the three retinal layers, though nuclei with no label are also seen, primarily in the ONL. Scale bar, 10 μm in all panels.

The presence of 89-kDa PARP-1 fragments in the blots (Fig. 3A) demonstrated that cleavage of a part of PARP-1 occurred in P3 to P7 retinas. This is in accordance with the presence of active caspase-3, which cleaves PARP-1 into two fragments of 24 and 89 kDa. ³⁵  

Taken together, these results show that PARP-1 expression in the retina was high during the first week of postnatal development and decreased during the second postnatal week, reaching a lower expression pattern in P14 that persisted into adulthood. 

PARP-1 Activation in the Early Postnatal Retina

PARP-1 activation was examined by immunoblots using the PAR antibody. These immunoblots showed the presence of poly(ADP)-ribosylated proteins during the first postnatal week and decreased thereafter (Fig. 6A). In addition, a poly(ADP)-ribosylated protein with a molecular weight of approximately 113 kDa, which most likely corresponded to the complete form of PARP-1, was strongly detected at P7 (Fig. 6A), indicating higher PARP-1 activation in the retina at this developmental age (Fig. 6B). The presence of the PAR polymer, as revealed by immunofluorescence (Figs. 7, 8), correlated well with the distribution of PARP-1 (Figs. 4, 5) and PAR Western blot data (Fig. 6A), with less PAR immunolabeling at P14 than at P7 (compare Figs. 7A–F with 7G–I). PAR immunolabeling was nuclear (Fig. 8), and some pyknotic nuclei showed intense PAR immunoreactivity (Figs. 7D–F). 

Figure 6.

 

Presence of poly(ADP)-ribosylated proteins in the developing retina. (A) Western blot analysis of PARP-1 activity (poly(ADP)-ribosylation of proteins) throughout postnatal development of the retina. The immunoblot is representative of three independent experiments. Variation in the presence of the active form of PARP-1 is indicated (PARP-1 activation). β-Tubulin antibody was used as a loading control. (B) Densitometric quantification of the active PARP-1 band; note that P28 was omitted in the histogram because this time point was not included in all experiments. Densitometric values were expressed as relative levels with respect to E18 values. *P < 0.05; significant differences between marked values.

Figure 7.

 

PAR immunocytochemistry (green) and Hoechst staining (blue) in confocal optical sections of P7 and P14 postnatal retinas. (A, B) PAR distribution (green) within the ganglion cell layer (GCL) and inner nuclear layer (INL) of a P7 retina. (B) Nuclear Hoechst staining. (C) Negative control at P7 in which the primary antibody was omitted. Right: Hoechst staining. (D–F) Higher magnification of the boxed area in B; note that intense PAR immunolabeling is sometimes associated with pyknotic nuclei, as seen with Hoechst staining (white arrows), though PAR staining is also observed in nuclei that do not show apparent pyknosis (red arrows). (G, H) PAR immunoreactivity (green) and nuclear staining (blue) in P14 retinas. PAR immunostaining is weaker at this stage than at P7. (I) Negative control at P14. Images are representative of results obtained in three different retinas for each time point. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. Scale bars: 50 μm (A–C, G–I); 20 μm (D–F).

Figure 8.

 

Confocal optical sections of selected areas of the ganglion cell layer (GCL; A–C), inner nuclear layer (INL; D–F), and outer nuclear layer (ONL; G–I) in transverse sections of DAPI-counterstained, PAR-immunolabeled P7 retinas. Most of the PAR immunolabeling is nuclear in the three retinal layers, though nuclei with no label are also seen. Scale bar, 10 μm in all panels.

Hence, PARP-1 protein and gene expression levels did not differ between P3 and P7, whereas PARP-1 activity was higher at P7 than at P3. For this reason, we centered our study on comparisons between P7 and older ages. 

Truncated AIF Expression in the Early Postnatal Retina and AIF Translocation

After the demonstration of increased PARP-1 expression and activation in the retina during the first postnatal week, we sought to determine whether truncated AIF was present in the retina at these developmental stages and whether it was translocated to the nucleus. 

Immunoblots of whole retinal extracts showed that 57 kDa AIF was present in the retina at P7 (Figs. 9A, 9B) and gradually decreased during the second postnatal week (Fig. 9C). Immunoblots of nuclear retinal extracts demonstrated that larger amounts of truncated AIF were present in the nucleus at P7 than at P14 (Fig. 9D). AIF translocation to the nucleus has been linked to caspase-independent cell death. In the present study, the presence of truncated AIF in the nucleus at P7 coincided with the peak of PARP-1 activity and PAR polymer formation, suggesting that this AIF form is involved in PARP-1–dependent and caspase-independent cell death in the retina. 

Figure 9.

 

Western blot analysis of AIF expression in the postnatal retina. (A) Immunoblot showing the presence of the 57-kDa truncated AIF form in P3, P7, P14, and P21 retinas. Note that the band is weaker after P7. β-Tubulin antibody was used as a loading control. This experiment was repeated three times. (B) Densitometric analysis of the amount of truncated AIF during retinal development (mean ± SEM). This AIF form is higher at P7 and decreases thereafter. *P < 0.05; significant difference between P7 and older ages. Values are expressed as relative levels with respect to P3. (C) Western blot analysis of 57-kDa AIF form, showing its gradual decrease between P7 and P15. (D) Immunoblot, representative of three experiments, showing that 57 kDa AIF is enriched in P7 nuclear extracts (Nucl P7) compared with P14 extracts (Nucl P14). Note that β-tubulin (a cytoplasmic protein) is absent from nuclear extracts; conversely, nuclear extracts show intense HMG1 (a chromatin-associated protein) labeling, whereas this is barely discernible in whole cell extracts (whole cell P7). An equal amount of protein was loaded in each lane.

The presence of AIF in the developing retina was also demonstrated by immunocytochemistry. Confocal optical sections of P7 retinas revealed that strong AIF immunolabeling was sometimes found in pyknotic nuclei that were also intensely immunolabeled by the anti–PAR antibody (Fig. 10A). Other cells showed discrete AIF labeling in their nuclei (Figs. 10B, 10C), suggesting that AIF had translocated from the cytoplasm to the nucleus, though much of the AIF remained cytoplasmic (Fig. 10C). In P14 retinas, nuclear AIF immunolabeling occurred less frequently (Fig. 10D). 

Figure 10.

 

Anti–AIF immunolabeling in postnatal retinas showing AIF expression and translocation to the nucleus. (A) AIF immunolabeling (AIF; red), PAR immunostaining (PAR; green), and Hoechst staining (HO; blue) in a P7 retina confocal optical section. Merged images show the localization of AIF and PAR (AIF+PAR), AIF and Hoechst (AIF+HO), and PAR and Hoechst (PAR+HO). Note that some pyknotic nuclei show intense AIF and PAR immunoreactivities (white arrows), which are absent in other pyknotic nuclei (red arrows). (B) Confocal optical section showing AIF staining (green) within the nucleus of one ganglion cell (arrow) in P7. (C) AIF immunolabeling in a confocal optical section of the inner nuclear layer (INL) at P7. Most of the staining is cytoplasmic, but AIF labeling appears in the nucleus marked by a white arrow; the red arrow indicates a cell showing intense AIF immunostaining in the cytoplasm and a pyknotic nucleus. (D) AIF immunostaining in a confocal optical section of the INL at P14, showing that nuclear labeling is scarce. (B–D) Nuclei stained with Hoechst (blue). Scale bars: 15 μm (A); 3 μm (B); 5 μm (C, D).

Discussion

Cell death in the postnatal developing mouse retina peaks at around P7. ³³ Immunofluorescence findings demonstrated that caspase-dependent cell death occurs in the developing mouse retina. However, the presence of alternative cell death programs during retinal development was suggested by the fact that not all pyknotic or TUNEL-positive nuclei were associated with active caspase-3 (Fig. 2). Although this result may be influenced by the inability to detect caspase activity during some stages of the death process, the consistent presence of pyknotic cells not associated with caspase-3 activity indicates that some of them die by a caspase-independent pathway. Other studies also sustain the occurrence of caspase-independent cell death in the postnatal retina. Thus, cell death was not blocked but only delayed in retinas of caspase 3^−/− mice, ³⁶ and several studies detected no caspase activation in different models of retinal degeneration or found that caspase inhibitors failed to protect against cell death. ^{4 –6,37,38} In short, although many retinal cells die by a caspase-dependent mechanism, caspase-independent pathways may also operate. 

PARP-1 activity appears to be linked to programmed cell death during postnatal development of the retina. It was elevated in the mouse retina during the first postnatal week, when abundant dying cells were also present, and decreased during the second week (Fig. 6), when cell death was markedly reduced. The decrease in PARP-1 activity was linked to transcriptional regulation of the PARP-1 gene (Figs. 3, 4). This conclusion contrasts with the idea that PARP-1 shows little transcriptional regulation during development (for example, see Fig. 3A in reference 6). To our knowledge, this is the first time transcriptional regulation of PARP-1 during development has been described, though a transient increase in PARP-1 transcription was reported in experimentally altered retinas. ³⁹  

Although PARP-1 strongly decreases after the first postnatal week, our experiments show that it continues to be present in the adult retina (Figs. 3B, 3F, and Supplementary Fig. S1), as previously reported in healthy rodents. ^{39 –42} Several studies ^{39,43 –45} and our own findings reveal that protein poly(ADP)-ribosylation takes place in the adult retina, though to a lower degree than during the first postnatal week (Fig. 6, P60, band pattern). Low amounts of PARP-1 in the adult retina would be responsible for this low protein poly(ADP)-ribosylation, though other enzymes (such as PARP-2) may also participate. ¹⁷  

In support of the relationship between PARP-1 activity and cell death, we found PAR immunoreactivity in some pyknotic nuclei (Fig. 7). In addition, it was previously demonstrated that cell death can be induced by PAR polymer, a product of PARP-1 activity, ²² and cell death in various retinal diseases was found to be caspase-independent and dependent on PARP-1 activation. ^{25 –27} However, this is the first report linking PARP-1 and cell death during the normal development of the retina. 

AIF participates in caspase-independent cell death and may mediate active PARP-1–dependent cell death in the normal developing retina. ¹⁰ As stated in the Introduction, translocation of the truncated AIF form (57 kDa) from the mitochondria to the nucleus is considered a key initial step in caspase-independent cell death. Nevertheless, translocation of the nontruncated AIF form (62 kDa) has also been described, ⁴⁶ and distinct forms of AIF appear to be involved in cell death phenomena resulting from different treatments. ²⁵  

We found elevated levels of truncated AIF (57 kDa) in mouse retinas during the first postnatal week, when abundant cell death, PARP-1 upregulation/activation, and PAR polymer formation were also observed. In addition, the relationship of AIF translocation with PAR polymer formation was reinforced by the finding of pyknotic nuclei containing both AIF and PAR polymer immunoreactivities (Fig. 10). Truncated AIF levels were reduced after the first week, coinciding with the downregulation of PARP-1 gene expression. Therefore, truncated AIF is likely involved in the PARP-1–dependent cell death pathway. 

Downregulation of PARP-1 and AIF in the retina during the second postnatal week would eradicate this caspase-independent cell death pathway. In this regard, it has been reported that the expression of proapoptotic members of the Bcl-2 family decreases during retinal development ⁴ and that some of these (e.g., Bid) have been related to AIF translocation. ⁴⁷ It is, therefore, possible that the downregulation of these Bcl-2 members would diminish the release of AIF from the mitochondria and thereby prevent AIF translocation to the nucleus. Caspase-3 (data not shown) and other key mediators of the intrinsic apoptotic pathway are also downregulated in the retina during the second week of postnatal development. ^{48 –50} Hence, cell death pathways may be blocked once cell death is no longer necessary for the correct development of the retina, preventing the inappropriate elimination of cells in the mature retina. 

Various authors have demonstrated that reactive oxygen species (ROS) can trigger death pathways that involve PARP-1 activation and AIF cleavage without caspase activation. ^26,51,52 It has been proposed that intracellular ROS induces peroxynitrite formation, causing DNA damage and PARP-1 activation ⁵³ with consequent PAR polymer formation. The end result is the activation of mitochondrial AIF, giving rise to a cleaved product (57 kDa AIF form) that promotes DNA fragmentation and nuclear condensation. ⁵² Studies are in progress to elucidate the molecular mechanisms involved in the AIF translocation that follows PARP-1 activation and to determine whether ROS play a role in PARP-1–dependent cell death during postnatal development of the mouse retina. 

Supplementary Materials

Figure sf01, PDF - Figure sf01, PDF 

Footnotes

 Supported by Spanish Ministry of Science and Innovation Grants BFU2007-61659 and BFU2010-19981 and by Junta de Andalucía Grant P07-CVI-03008.

Footnotes

 Disclosure: D. Martín-Oliva, None; R.M. Ferrer-Martín, None; A.M. Santos, None; M.C. Carrasco, None; A. Sierra, None; J.L. Marín-Teva, None; R. Calvente, None; J. Navascués, None; M.A. Cuadros, None

The authors thank Javier Oliver (CSIC, Granada) for helpful discussions and David Porcel (CIC, Universidad de Granada) and José A. Muñoz-Gámez (San Cecilio Hospital, Granada) for their useful assistance.