Normoxic Activation of Hypoxia-Inducible Factors in Photoreceptors Provides Transient Protection against Light-Induced Retinal Degeneration

Abstract

Purpose.: Hypoxic preconditioning activates hypoxia-inducible transcription factors (HIFs) in the retina and protects photoreceptors from light-induced retinal degeneration. The authors tested whether photoreceptor-specific activation of HIFs in normoxia is sufficient for protection.

Methods.: Rod-specific Vhl knockdown mice were generated using the Cre-lox system with the rod opsin promoter controlling expression of CRE recombinase to stabilize HIF transcription factors in normoxic rods. Cell death was induced by light exposure and quantified by ELISA. Rhodopsin was quantified by spectrophotometry. Gene expression was analyzed by real-time PCR, and levels of proteins were determined by Western blotting. Morphology was investigated by light microscopy and retinal function tested by ERG.

Results.: The rod-specific Vhl knockdown stabilized HIF-α proteins and induced expression of HIF target genes in retinas of 10-week-old mice under normoxic conditions. Retinal morphology and function were normal. At 36 hours after exposure to excessive light, Vhl knockdowns showed significantly less photoreceptor cell death than did wild-type controls. Ten days after light exposure, however, photoreceptor degeneration in Vhl knockdowns was similar to that of control animals. Vhl knockdowns expressed Fgf2 at higher basal levels before light exposure. After light exposure, however, expression of Fgf2 was not significantly different from that of wild-type controls.

Conclusions.: Artificial activation of HIF transcription factors in normoxic photoreceptors results in an increased basal expression of Fgf2 that may contribute to a transient protection of rods against light damage. Full photoreceptor protection may require a hypoxia-like response in additional retinal cell types and/or the differential regulation of additional mechanisms.

Retinitis pigmentosa (RP) is characterized by the progressive loss of photoreceptors causing severe visual disturbances and blindness in humans. To investigate molecular mechanisms of photoreceptor degeneration, different mouse models have been developed. The model of light-induced photoreceptor cell death is widely used to analyze immediate changes in gene expression during degeneration or to test the efficacy of neuroprotective strategies ¹ such as hypoxic preconditioning. It has been shown that a short pre-exposure of mice to reduced levels of oxygen completely protects photoreceptors from light-induced degeneration. ²  

To cope with the stress imposed by reduced oxygen availability during hypoxia, cells regulate expression of genes for survival, growth, or metabolism. ³ This regulation is achieved mainly by special transcription factors called hypoxia inducible factors (HIFs). HIFs consist of two subunits, α and β. The HIF β-subunit (aryl-hydrocarbon receptor nuclear translocator [ARNT]) is constitutively expressed and located in the nucleus. HIF1A, HIF2A, and HIF3A, the three known isoforms of the HIF α-subunit, are expressed but quickly degraded in conditions of normal oxygen availability (normoxia). Hydroxylation of HIF α-subunits by prolyl hydroxylases (egl nine [EGLN] family) leads to their recognition by the von Hippel-Lindau (VHL) protein. VHL forms a multimeric protein complex together with Cullin-2, RBX1, and the elongins B and C, ^{4 –6} and acts as a E3-ubiquitinase targeting HIF α proteins to proteasomal degradation. ⁷ Another hydroxylase, the factor inhibiting HIF (FIH), hydroxylates HIF-α at an asparagine residue, inhibiting p300 coactivator recruitment and thus transcriptional activity. ⁸ Hydroxylation of HIF-α proteins by EGLNs and FIH depends on the availability of oxygen. ^8,9 Hence, hydroxylation is strongly reduced during hypoxia and HIF α-subunits are not recognized by VHL. Thus HIF α proteins are stabilized, enter the nucleus, bind HIF-β and p300, and participate in the regulation of gene expression. 

It has been shown that during hypoxic preconditioning, HIF1A and HIF2A are stabilized and that HIF target genes are differentially regulated in the retina. ^2,10,11 Erythropoietin (Epo) was shown to be strongly induced after hypoxic exposure, and injection of recombinant EPO was protective against light-induced retinal degeneration. ² Using a genetic approach, we recently showed that HIF1A is not essential in rods for protection by hypoxic preconditioning. ¹⁰ However, this does not exclude that rod-specific activation of HIF1A contributes to neuroprotection after hypoxic preconditioning. In addition, HIF2A and/or other factors that are expressed in rods and regulated by hypoxia might be involved in neuroprotection by hypoxic preconditioning. Alternatively, hypoxia might regulate factors in other cell types in the retina. Such factors might act in trans to protect rods. To address these questions, we investigated whether a hypoxia-like response in rod photoreceptors is sufficient to prevent retinal degeneration in normoxia. To achieve normoxic induction of a hypoxia-like response, we generated photoreceptor-specific Vhl knockdown mice using the Cre-lox system. Because the absence of VHL leads to increased stability of HIF α-subunits in normoxia, Vhl knockdowns may mimic the hypoxic response at normal oxygen levels. In case of protection, VHL protein may provide a target for tissue-specific pharmacological interventions in the future. 

Materials and Methods

Mice and Genotyping

Mice were treated in accordance with the regulations of the Veterinary Authority of Zurich and with the statement of The Association for Research in Vision and Ophthalmology for the use of animals in research. All mice were maintained as breeding colonies at the University of Zurich in a 12-hour/12-hour light/dark cycle (60 lux). 

Wild-type BALB/c mice were from a breeding colony at the University Hospital Zurich. 129S-Vhlh^tm1jae/J-mice (hereafter Vhl^flox/flox mice), which have loxP sites flanking exon 1 and part of the promoter of the Vhl gene, ¹² were purchased from Jackson Laboratory (Bar Harbor, ME). To generate rod photoreceptor-specific Vhl knockdowns, Vhl^flox/flox mice were crossed with mice expressing Cre recombinase under the control of the opsin promoter (LMOPC1, hereafter opsin-Cre mice). In these mice, Cre expression in rods starts at approximately postnatal day 7 and increases up to 6 weeks of age. ¹³ The breeding colonies were kept on the light-sensitive Rpe65_450Leu background. ¹⁴ A red fluorescence protein (RFP) reporter strain (ROSA-flox-RFP; kindly provided by Wolf-Dietrich Hardt, ETH, Zurich, Switzerland) was crossed with opsin-Cre mice to assay the expression pattern of Cre recombinase. 

The following primers were used to detect wild-type (wt) and Vhl-flox alleles: forward (forw) (5′-TGAGTATGGGATAACGGGTTGAAC-3′) and reverse (rev) (5′-AGAACTGACTGACTTCCACTGATGC-3′). The wt allele was identified as a 125-basepair (bp) PCR fragment and the Vhl-flox as a 317-bp fragment after gel electrophoresis. Presence of the opsin-cre transgene was tested by PCR using the following primer pair: forw (5′-AGGTGTAGAGAAGGCACTTAGC-3′) and rev (5′-CTAATCGCCATCTTCCAGCAGG-3′). In the presence of the transgene, the amplification reaction resulted in the production of a 411-bp fragment. To detect excision of floxed sequences in the Vhl gene, genomic DNA was isolated from retinal tissue and tested by PCR using the following primers: forw_un-excised (5′-CTGGTACCCACGAAACTGTC-3′), forw_excised (5′-CTAGGCACCGAGCTTAGAGGTTTGCG-3′) and rev_both (5′-CTGACTTCCACTGATGCTTGTCACAG-3′). The excised allele was identified as a 260-bp and the unexcised allele as a 460-bp fragment. Presence or absence of the RFP transgene was tested using the ROSA-tg forward primer: 5′-GCGAAGAGTTTGTCCTCAACC-3′ or the ROSA-wt forward primer: 5′-GGAGCGGGAGAAATGGATATG-3′, respectively, together with the ROSA-common reverse primer: 5′-AAAGTCGCTCTGAGTTGTTAT-3′. An amplification product of 300 bp or of 500 bp indicated the presence or absence, respectively, of the RFP locus. 

Laser Capture Microdissection

After mice were killed, their eyes were enucleated, immediately embedded in tissue freezing medium (Leica Microsystems Nussloch GmbH, Nussloch, Germany), and frozen in a 2-methylbutane bath cooled with liquid nitrogen. Retinal sections (20 μm) were fixed (5 minutes in acetone), air dried (5 minutes), and dehydrated (30 seconds in 100% ethanol, 5 minutes in xylene). Microdissection was performed using an Arcturus XT Laser capture device (Molecular Devices, Sunnyvale, CA). RNA was isolated using the Arcturus kit for RNA isolation (Molecular Devices) according to the manufacturer's directions including a DNase treatment to digest residual genomic DNA. Equal amounts of RNA were used for reverse transcription using oligo(dT) and M-MLV reverse transcriptase (Promega, Madison, WI). 

Semiquantitative Real-Time PCR

Retinas were removed through a slit in the cornea and immediately frozen in liquid nitrogen. Total RNA was prepared using a kit (RNeasy RNA isolation kit; Qiagen, Hilden, Germany) according to the manufacturer's directions including a DNase treatment to digest residual genomic DNA. Equal amounts of RNA were used for reverse transcription using oligo(dT) and M-MLV reverse transcriptase (Promega). Relative quantification of cDNA was conducted via real-time PCR using a kit (LightCycler 480 Sybr Green I Master kit; Roche Diagnostics, Basel, Switzerland), a LightCycler 480 instrument (Roche Diagnostics) and specific primer pairs (Table 1). Expression was normalized to Actb (β-actin) and relative quantification was calculated using the comparative threshold method (ΔΔCT) and commercial software (Light Cycler 480 software; Roche). 

Table 1.

 

Primers Used for Real-Time PCR

Gene	Upstream (5′–3′)	Downstream (5′–3′)	Annealing Temperature (°C)	Product (bp)
Adm	ttcgcagttccgaaagaagt	ggtagctgctggatgcttgta	62	77
Actb	cgacatggagaagatctggc	caacggctccggcatgtgc	62	153
Bnip3	cctgtcgcagttgggttc	gaagtgcagttctacccaggag	60	93
Egln1	cattgttggcagaaggtgtg	caaaggactacagggtctcca	62	70
Epo	gccctgctagccaattcc	gccctgctagccaattcc	60	128
Gnat1	gaggatgctgagaaggatgc	tgaatgttgagcgtggtcat	58	209
Gnat2	gcatcagtgctgaggacaaa	ctaggcactcttcgggtgag	58	192
Mt1	gaatggaccccaactgctc	gcagcagctcttcttgcag	62	104
Mt2	tgtacttcctgcaagaaaagctg	acttgtcggaagcctctttg	62	94
Sag	ttacaagccttccaacctctgac	accagcacaacaccatctacag	64	189
Scl2a1	cagtgtatcctgttgcccttctg	gccgaccctcttctttcatctc	62	151
Rho	cttcacctggatcatggcgtt	ttcgttgttgacctcaggcttg	62	130
Vegfa	acttgtgttgggaggaggatgtc	aatgggtttgtcgtgtttctgg	60	171
Vhl	gagggacccgttccaataat	ttggcaaaaataggctgtcc	60	364
Fgf2	tgtgtctatcaagggagtgtgtgc	accaactggagtatttccgtgaccg	62	158
Edn2	agacctcctccgaaagctg	ctggctgtagctggcaaag	60	64
Stat3	caaaaccctcaagagccaagg	tcactcacaatgcttctccgc	62	133
Gfap	ccaccaaactggctgatgtctac	ttctctccaaatccacacgagc	62	240
Socs3	atttcgcttcgggactagc	aacttgctgtgggtgaccat	58	126
Lif	aatgccacctgtgccatacg	caacttggtcttctctgtcccg	60	216
Cntf	ctctgtagccgctctatctg	ggtacaccatccactgagtc	58	125

Western Blotting Analysis

Retinas were homogenized in 0.1 M Tris–HCl (pH 8) at 4°C and protein content was determined using Bradford reagent. Standard SDS-PAGE and Western blotting were performed. For immunodetection, the following antibodies were used: anti-HIF1A (no. Nb100–479; Novus Biologicals, Cambridge, UK, 1:1000), anti-HIF2A (no. Nb100–122, Novus Biologicals; 1:1000), anti-pSTAT3 (no. 9131; Cell Signaling Technology, Danvers, MA; 1:500) and anti-ACTB (no. A5441; Sigma-Aldrich, St. Louis, MO; 1:5000). Blots were incubated overnight at 4°C with primary antibodies followed by a 1-hour incubation with HRP-conjugated secondary antibodies. Immunoreactivity was visualized using the Renaissance-Western blot detection kit (PerkinElmer Life Sciences, Emeryville, CA). 

Electroretinography

ERGs were recorded according to previously described procedures. ^15,16 The ERG equipment consisted of a Ganzfeld bowl, a direct current amplifier, and a PC-based control and recording unit (Multiliner Vision; VIASYS Health Care GmbH, Hoechberg, Germany). Mice were dark adapted overnight and anesthetized with ketamine (66.7 mg/kg body weight) and xylazine (11.7 mg/kg body weight). Pupils were dilated and single-flash ERG responses were obtained under dark-adapted (scotopic) and light-adapted (photopic) conditions. Light adaptation was accomplished with a background illumination of 30 cd/m² starting 10 minutes before recording. Single white-flash stimulus intensity ranged from –4 to 1.5 log cd · s/m² under scotopic and from –2 to 1.5 log cd · s/m² under photopic conditions, divided into 10 and 8 steps, respectively. Ten responses were averaged with an interstimulus interval (ISI) of either 5 seconds (for –4, –3, –2, –1.5, –1, and –0.5 log cd · s/m²) or 17 seconds (for 0, 0.5, 1, and 1.5 log cd · s/m²). 

Light Exposure, Cell Death Detection, and Morphology

Mice were dark adapted overnight and pupils dilated with 1% ophthalmic solution (Cyclogyl; Alcon, Cham, Switzerland) and 5% phenylephrine (Ciba Vision, Niederwangen, Switzerland) 1 hour before exposure to white fluorescent light (13′000 lux) for 1 hour. 

After light exposure, mice were placed in darkness overnight. Thereafter, mice were kept in normal cyclic light until they were killed. To quantify apoptosis, mice were killed 36 hours after light exposure. Nucleosomal release was determined using a cell death detection (CDD) kit (Roche Diagnostics) according to the manufacturer's recommendations. 

For light microscopy, eyes were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.3) at 4°C overnight. For each eye, the superior and the inferior retinas were prepared, washed in cacodylate buffer, incubated in osmium tetroxide for 1 hour, dehydrated, and embedded in resin Epon 812. Sections (0.5 μm) were prepared from the lower central retina and counterstained with methylene blue. 

Hypoxic Preconditioning

Adult (8–10 weeks of age) BALB/c mice were exposed to reduced oxygen concentrations in a hypoxic chamber with food and water ad libitum. O₂ levels were reduced in steps of 2% by changing the O₂/N₂ ratio to reach a final O₂ concentration of 7% within 90 minutes. After 6 hours of hypoxia, mice were killed immediately to isolate retinal tissue for gene expression analysis. 

Rhodopsin Measurements

Mice were dark adapted overnight. Retinas were removed through a slit in the cornea under dim red light and placed in 1 mL of distilled H₂O for 1 minute. After 3 minutes of centrifugation at 15,000g, the supernatant was discarded and 700 μL of 1% hexadecyltrimethylammonium bromide (Fluka Chemie, Buchs, Switzerland) in H₂O was added to the pellet. Retinas were mechanically homogenized with a polytron (20 seconds, 3000 rpm), centrifuged for 3 minutes at 15,000g, and the supernatant was collected. The absorption at 500 nm was measured before and after exposure to bright white light (20,000 lux for 1 minute). The amount of rhodopsin present per retina was calculated using the following formula derived from the Lambert-Beer equation: ρ = vol × c = vol × Δabs500/(E500 × l × n). ρ is the amount of rhodopsin per retina (in moles); vol is the volume of the sample (in liters); c is the concentration of rhodopsin per retina (M); Δabs500 is the difference between absorption of the sample at 500 nm before and after bleaching; E500 is the extinction coefficient of rhodopsin at 500 nm (4.2 × 10⁴ cm × M); l is the path length of the cuvette (in cm); and n is the number of retinas. 

Results

Successful Knockdown of Vhl in Photoreceptors

Expression of Cre recombinase in our line of LMOPC1 mice was specific to the photoreceptor layer but showed a patchy pattern (Figs. 1A, 1B). Analysis of genomic DNA from retinal tissue of 10-week-old Vhl^flox/flox and Vhl^flox/flox ;opsin-cre mice showed excision of floxed DNA sequences in mice expressing CRE recombinase but not in control mice (Fig. 1C). Because excision is specific for rod photoreceptors, ¹³ whole retinal samples contain both the floxed (unexcised; from retinal cells without CRE expression) and the excised allele. Semiquantitative real-time PCR on samples obtained by laser capture microdissection revealed lower expression levels of Vhl mRNA in the ONL of Vhl^flox/flox ;opsin-cre mice (CRE+) compared to their control (Vhl^flox/flox ;CRE−) littermates (Fig. 1D). The reason for the upregulation of Vhl expression in the other cell layers is unclear, but might involve compensatory effects. 

Figure 1.

 

Knockdown of Vhl in photoreceptors of 10-week-old Vhl^flox/flox ;opsin-cre mice. (A) Expression pattern of Cre recombinase in ROSA-flox-RFP;opsin-Cre mice. Red, red fluorescent protein, expressed only in cells where Cre was active. Scale bar, 100 μm. (B) higher magnification of the section shown in (A). (C) PCR amplification of genomic DNA isolated from total retinal tissue of three different Vhl^flox/flox (CRE–) and three Vhl^flox/flox ;opsin-cre (CRE+) mice at 10 weeks of age. The amplified fragment from the floxed gene (floxed) has a length of 460 bp. CRE-mediated excision of the floxed sequence (excised) resulted in a 260-bp fragment. (D) Semiquantitative real-time PCR after laser capture microdissection showing relative Vhl mRNA levels in the different retinal layers (as indicated) of Vhl^flox/flox (CRE–) and Vhl^flox/flox ;opsin-cre (CRE+) mice. Values were normalized to Actb and expressed relatively to the respective values for each layer in Vhl^flox/flox mice, which were set to 1. Shown are mean values ± SD of three different mice amplified in duplicates. ONH, optic nerve head; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer.

Stabilization of HIF-α Proteins Leads to Increased Expression of HIF Target Genes in 10-Week-Old Vhl^flox/flox ;opsin-cre Mice

Downregulation of Vhl in photoreceptors led to a stabilization of HIF-α proteins in normoxic retinas. Increased levels of both HIF1A and HIF2A were detected in retinas of 10-week-old Vhl^flox/flox ;opsin-cre mice (CRE+; Fig. 2A). In addition to HIF-α stabilization, we noticed strong phosphorylation of STAT3. Expression levels of total STAT3 protein was also slightly elevated in Vhl knockdown retinas. Although STAT3 is reported to be part of the HIF-pathway, ¹⁷ it is unclear whether the increased phosphorylation of STAT3 is a consequence of the stabilized HIF-α proteins or whether it is due to other mechanisms in the Vhl knockdown retinas. Importantly, however, stabilization of HIF-α proteins and phosphorylation of STAT3 are features also found in the hypoxic retina. Therefore, the molecular response in the Vhl knockdowns may resemble the response in retinas of mice preconditioned by hypoxia. To determine whether the stabilized HIFs are transcriptionally active in normoxia, we analyzed expression of several hypoxia-responsive genes ¹¹ in retinas of 10-week-old Vhl^flox/flox ;opsin-cre mice (Fig. 2B) and in wild-type mice after hypoxic preconditioning (Fig. 2C). Expression of adrenomedullin (Adm), prolyl hydroxylase 2 (Egln1), solute carrier family 2 (facilitated glucose transporter) member 1 (Slc2a1, also called Glut1), and vascular endothelial growth factor (Vegf) was induced to a similar extent in normoxic Vhl^flox/flox ;opsin-cre mice and in hypoxic wild-type mice (after exposure to 7% oxygen for 6 hours). Regulation of these hypoxia response genes is largely attributed to HIF1. ^{18 –20} Expression of BCL2/adenovirus E1B 19-kDa interacting protein 3 (Bnip3), metallothionein 1 (Mt1), and metallothionein 2 (Mt2) was also upregulated in Vhl^flox/flox ;opsin-cre mice but not to the same extent as in hypoxic wild-type mice. Expression of erythropoietin Epo, which is reported to be activated by HIF2, ²¹ was induced in hypoxic but not in Vhl^flox/flox ;opsin-cre mice. This suggests that the stabilized HIF2 was not transcriptionally active, that induction of Epo during hypoxia is regulated by cells other than rod photoreceptors, or that Epo expression in photoreceptors is not controlled by HIF2. 

Figure 2.

 

Stabilization and activation of transcription factors in 10-week-old Vhl^flox/flox ;opsin-cre mice. (A) Western blot analysis showing protein expression of HIF1A, HIF2A, phosphorylated STAT3 (p-STAT3), and STAT3 in four different Vhl^flox/flox (CRE–) and Vhl^flox/flox ;opsin-cre (CRE+) mice at 10 weeks of age. ACTB served as a control for equal loading. (B) Total retinal RNA was prepared from 10-week-old Vhl^flox/flox ;opsin-cre mice and expression levels of individual genes (as indicated) were determined by semiquantitative real-time PCR. Levels were normalized to Actb and expressed relatively to the expression in Vhl^flox/flox control mice (set to 1). (C) Total retinal RNA was prepared from wild-type mice after hypoxic preconditioning. Expression levels were normalized to Actb and expressed relatively to expression in normoxic control mice (set to 1). Shown are mean values ± SD of four retinas amplified in duplicates.

Knockdown of Vhl in Rod Photoreceptors Does Not Affect Morphology and Function in Young Mice but Leads to Retinal Degeneration in Old Mice

To assess a potential spontaneous degeneration of photoreceptor cells in Vhl knockdown mice, we analyzed ERG responses (Figs. 3A–D) in young Vhl^flox/flox ;opsin-cre and Vhl^flox/flox littermates. ERG traces and scotopic and photopic b-wave amplitudes in 17-week-old Vhl knockdown mice did not differ from those in control Vhl^flox/flox littermates and were similar to those in wild-type mice. ¹⁶ Normal retinal structure in young Vhl knockdown animals was further supported by comparable retinal morphologies (see below) and by similar expression levels of the photoreceptor marker genes rod transducin (Gnat1), cone transducin (Gnat2), S-antigen (Sag), and rod opsin (Rho) in 10-week old Vhl^flox/flox ;opsin-cre and Vhl^flox/flox littermates (Fig. 3E). 

Figure 3.

 

Normal retinal function in 17-week-old Vhl^flox/flox ;opsin-cre mice. (A) Representative scotopic (dark-adapted) and (B) photopic (light-adapted) single-flash ERGs with increasing light intensities recorded from 17-week-old Vhl^flox/flox control and Vhl^flox/flox ;opsin-cre mice as indicated. The vertical line crossing each trace shows the time point of the light flash. (C) Scotopic and (D) photopic b-wave amplitudes from control (black, n = 3) and Vhl^flox/flox ;opsin-cre mice (red, n = 3) as a function of the logarithm of the flash intensity. Boxes indicate the 25% and 75% quartile range, whiskers indicate the 5% and 95% quantiles, and the asterisks indicate the median of the data. (E) Semiquantitative real-time PCR for photoreceptor-specific genes (as indicated) in 10-week-old Vhl^flox/flox (white bars) and Vhl^flox/flox ;opsin-cre (gray bars). Values were normalized to Actb and expressed relatively to the value of Vhl^flox/flox mice, which was set to 1. Shown are mean values ± SD of four animals amplified in duplicates.

ERG recordings from 1-year-old Vhl^flox/flox ;opsin-cre mice, however, revealed a reduction of a- and b-wave amplitudes under both scotopic and photopic conditions (Figs. 4A–D). The strong reduction in the a-wave amplitude suggested that photoreceptors might be injured or lost. This was confirmed by the morphologic analysis of 1-year-old retinas (Fig. 4E). Compared with the ONL of age-matched Vhl^flox/flox and opsin-cre control mice, the ONL of Vhl^flox/flox ;opsin-cre mice was thinned, with only three to four rows of photoreceptor nuclei remaining. In addition, the structure of inner and outer segments appeared severely disintegrated. Thus, knockdown of Vhl did not affect retinal function and morphology in mice up to 17 weeks of age but caused retinal degeneration in old mice. 

Figure 4.

 

Retinal degeneration in 1-year-old Vhl^flox/flox ;opsin-cre mice. (A) Representative scotopic (dark-adapted) and (B) photopic (light-adapted) single-flash ERGs with increasing light intensities recorded from 1-year-old Vhl^flox/flox control and Vhl^flox/flox ;opsin-cre mice as indicated. The vertical line crossing each trace shows the timing of the light flash. (C) Scotopic and (D) photopic b-wave amplitudes from control (black, n = 3) and Vhl^flox/flox ;opsin-cre mice (red, n = 3) as a function of the logarithm of the flash intensity. Boxes indicate the 25% and 75% quartile range, whiskers indicate the 5% and 95% quantiles, and the asterisks indicate the median of the data. (E) Representative morphologic sections from the lower central retina of 1-year-old Vhl^flox/flox , of opsin-cre and of Vhl^flox/flox ;opsin-cre mice. RPE, retinal pigment epithelium; OS, photoreceptor outer segments; IS, photoreceptor inner segments; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. Scale bar, 100 μm.

Knockdown of Vhl in Rod Photoreceptors Delays Light-Induced Photoreceptor Degeneration in Young Mice

To test whether the activation of the hypoxia-like response in photoreceptors during normoxic conditions is similarly neuroprotective to light exposure as hypoxic preconditioning, we used 10-week-old Vhl^flox/flox ;opsin-cre mice. At this age, retinal morphology (Fig. 5) and expression of photoreceptor-specific genes (Fig. 3) were still similar to controls. In addition, scotopic and photopic b-wave amplitudes in 17-week-old knockdowns were indistinguishable from controls, suggesting that also at 10 weeks, retinal function was comparable. Mice were exposed to 13′000 lux of white light for 1 hour and retinal morphology was analyzed and cell death quantified at 36 hours thereafter (Figs. 5A, 5B). At this time point, photoreceptor outer segments were better preserved (Fig. 5A, middle panels, black arrows) and fewer pyknotic nuclei (Fig. 5A, white arrows) were found in the ONL of Vhl^flox/flox ;opsin-cre mice. Reduced cell death was confirmed by the ELISA-based quantification of free nucleosomes generated by internuceleosomal DNA cleavage during apoptotic cell death. Although variable within groups, light-exposed Vhl^flox/flox ;opsin-cre mice showed significantly less cell death compared to their control littermates and to mice expressing solely opsin-cre (Fig. 5B). 

Figure 5.

 

Transient photoreceptor protection in Vhl^flox/flox;opsin-cre mice. (A) Morphologic sections of dark controls (DC), and at 36 hours (+36h) and 10 days (+10d) after light exposure from retinas of 10-week-old Vhl^flox/flox and Vhl^flox/flox ;opsin-cre mice. Shown are representative sections of the lower central retina, the most affected region in our light damage setup. Arrows indicate the layer of photoreceptor segments; arrowheads show examples of pyknotic nuclei. (B) Quantification of retinal cell death at 36 hours after light exposure of opsin-cre (black bars), Vhl^flox/flox (white bars) and Vhl^flox/flox;opsin-cre mice (gray bars). Values are presented as box blots representing the median and the lower and upper quartiles. Whiskers show the complete range of all data points. Values were calculated relative to the value of a positive control (provided by the kit), which was set to 100%. n = 1 (DC, opsin-cre); n = 7 (DC, Vhl^flox/flox ); n = 5 (DC, Vhl^flox/flox ;opsin-cre); n = 4 (+36h, opsin-cre); n = 9 (+36h, Vhl^flox/flox ); n = 6 (+36h, Vhl^flox/flox ;opsin-cre). (C) Relative rhodopsin levels of retinas at 10 days after light exposure (LE) in Vhl^flox/flox (white bars) and Vhl^flox/flox ;opsin-cre mice (gray bars). Values are expressed relative to the corresponding dark controls (DC), which were set to 100%. Shown are mean values ± SD of six retinas. (D) Basal rhodopsin levels in retinas of dark-adapted and not light-exposed opsin-cre (black bar), Vhl^flox/flox (white bar) and Vhl^flox/flox ;opsin-cre (gray bar) mice. Values are expressed relative to the values of Vhl^flox/flox mice, which were set to 100%. Shown are mean values ± SD of six retinas. *P < 0.05; **P < 0.01. Significance was calculated using an unpaired t-test. RPE, retinal pigment epithelium; OS, photoreceptor outer segments; IS, photoreceptor inner segments; ONL, outer nuclear layer; INL, inner nuclear layer. Scale bar, 50 μm.

At 10 days after light exposure, however, photoreceptors in Vhl^flox/flox;opsin-cre mice had degenerated to a similar extent as in Vhl^flox/flox controls (Fig. 5A, bottom panels). Accordingly, both genotypes showed a comparable reduction in rhodopsin levels (as a measure for surviving photoreceptors) to 58% (Vhl^flox/flox ) and 52% (Vhl^flox/flox ;opsin-cre) of their respective dark control levels, which were set to 100% (Fig. 5C). This suggests that the knockdown of Vhl delayed but did not prevent photoreceptor degeneration after light exposure. We noted that basal levels of dark-adapted rhodopsin levels (before light exposure) were reduced in untreated Vhl knockdown animals to 76% of the levels found in control Vhl^flox/flox or opsin-cre mice (Fig. 5D). Rhodopsin regeneration, however, was comparable to that in wild-type and reached 8 pmol/min (data not shown). This value is relevant for light damage susceptibility and close to data published for pigmented wild-type animals expressing the RPE65-450Leu variant. ¹⁴ Similarly, expression of Rho RNA was not affected (Fig. 3E), and intracellular localization of rod opsin was normal (determined by immunofluorescence according to ²² ; data not shown) in the knockdown. Thus, the reason for the reduced basal levels of rhodopsin in Vhl knockdown retinas is unclear. 

Because protection was only transient and not permanent, as observed after hypoxic preconditioning, mechanisms of protection might be different in Vhl knockdowns and the hypoxic wild-type retina. To address potential mechanisms for the transient protection against degeneration, we analyzed expression levels of several genes known to be involved in the retinal response to injury or stress. Of particular interest were genes implicated in the protective pathway controlled by leukemia inhibitory factor (LIF) after light exposure. ^23,24 Basal expression of several genes downstream of LIF (Fgf2, Edn2, Stat3, Socs3, and Gfap) was higher in knockdown animals (Fig. 6, triangles) than in controls (Fig. 6, squares). Basal expression of Cntf and Lif itself was not affected. After light exposure, however, expression levels of all these genes converged in the two strains and already at 12 hours after exposure, levels were no longer significantly different. The expression pattern of the individual genes after light exposure in wild-type mice is in accordance with data published previously. ^24,25  

Figure 6.

 

Gene expression after light exposure. Relative gene expression levels in retinas of 10-week-old Vhl^flox/flox (squares, blue) and Vhl^flox/flox ;opsin-cre (triangles, red) mice before (dark) or at 12, 24, 36, and 48 hours after exposure to 13′000 lux of white light for 1 hour. Values are expressed relative to values of dark-adapted Vhl^flox/flox mice which were set to 1. Shown are mean ± SD values of three mice. Expression in the two strains were compared at individual time points using a Student's t-test. *P < 0.05; **P < 0.01.

Discussion

Hypoxic preconditioning protects photoreceptors against light-induced degeneration. ² The exact mechanisms are not known but may include differential regulation of neuroprotective genes. Hypoxia-inducible transcription factors such as HIF1 and HIF2 are activated during hypoxic preconditioning in the whole retina ¹⁰ and the product of at least one of their target genes, EPO, has been shown to protect photoreceptors. ^2,26 Here we show that lack of VHL induced a sustained activation of HIF transcription factors in normoxic photoreceptors that led to a hypoxia-like response and provided protection against light-induced cell death. However, this protection was only transient and did not reach the level observed after hypoxic preconditioning. 

Knockdown of Vhl in rod photoreceptors stabilized HIF1A and HIF2A and induced phosphorylation of STAT3, leading to the increased expression of a set of target genes that were also activated in the wild-type retina after exposure to hypoxia (Fig. 2). Because the knockdown of Vhl was specific to rod photoreceptors, our data suggest that expression of these genes was regulated in rods by HIFs and/or pSTAT3. Epo was the only tested gene that was not induced in Vhl^flox/flox ;opsin-cre mice. Because lack of VHL is sufficient to induce Epo expression in the retina, as shown in Vhl^flox/flox ;α-cre mice, ²² our results suggest that upregulation of Epo in hypoxic wild-type mice may not occur in rods but in other cell types of the retina. Alternative explanations include the possibility that Epo expression in rods is not regulated by one of the factors activated here. 

Although retinas of Vhl^flox/flox ;opsin-cre mice had a normal retinal morphology and function, levels of rhodopsin were reduced. This cannot be explained by a reduced number of photoreceptor cells because expression of rod and cone marker genes were normal in 10-week-old Vhl^flox/flox ;opsin-cre mice. Rhodopsin regeneration was also not affected and reached similar kinetics as in wild-type control mice. In addition, we did not detect mislocalization of rod opsin in retinas of Vhl^flox/flox ;opsin-cre mice (data not shown). Thus, the reason for the reduced rhodopsin levels remains unclear and warrants further investigation. 

Transient Protection against Light Damage

Knockdown of Vhl in rods resulted in a transient protection (or delayed degeneration) of photoreceptor cells against light damage. Although light damage depends on light absorption by the visual pigment, ^{27 –29} light damage susceptibility is determined by rhodopsin regeneration kinetics rather than by the absolute levels of dark-adapted rhodopsin. ^14,30 Because regeneration rates were identical in both wild-type and Vhl knockdown mice, protection was thus not mediated by differences in photon absorption. 

The delayed degeneration suggests that rods lacking VHL have activated some mechanisms that can postpone but not prevent the death of injured photoreceptors. A potential factor involved in this process may be FGF2, which was shown previously to be neuroprotective for photoreceptors. ^31,32 The increased basal expression of Fgf2 in Vhl^flox/flox ;opsin-cre mice before light exposure (Fig. 6) may support survival of photoreceptors initially. After light exposure, however, expression levels were similar to levels in light-exposed control mice. Because photoreceptors in control mice degenerate, such levels are obviously not sufficient for a long-term rescue of injured photoreceptors. Interestingly, Lif knockout and wild- type mice have similar Fgf2 expression levels before light exposure, but whereas wild-type mice upregulate Fgf2 expression in response to photoreceptor injury, Lif knockouts do not. ²⁵ As a consequence, Lif knockouts show a light damage severity similar to that of wild-type mice early (unpublished data) but a stronger degeneration late (10 days) after exposure. ²⁵ Therefore, the expression profile of Fgf2 correlates with degeneration severity early or late after photoreceptor injury. 

The mechanism leading to increased expression of Fgf2 in the absence of VHL needs to be determined. However, Vhl knockdown mice also have increased basal levels of Edn2 (Fig. 6), and we recently showed that activation of the receptor for Edn2 (EDNRB) induces Fgf2 expression. ²³ Thus, it might be that the increased levels of HIF1 in the Vhl knockdown retinas upregulate expression of Edn2 in photoreceptors, as shown in other systems. ³³ EDN2 may then cause induction of Fgf2 via activation of EDNRB. Thus, increased expression of Fgf2 may be a consequence of the increased Edn2 expression through HIF1A stabilization in rods of the Vhl knockdown mice. Alternatively, lack of VHL might also directly regulate Edn2, as suggested for Edn1. ³⁴  

The restriction of the hypoxia-like response to rod photoreceptors might be another reason for the lack of full protection in the Vhl^flox/flox ;opsin-cre mice. In contrast to the knockdown mice, hypoxic preconditioning induces a systemic response and may thus involve factors induced in other cell types of the retina. Epo, for example, is strongly upregulated in the hypoxic retina and can protect photoreceptors against light damage ² and in the rds mouse. ²⁶ Epo, however, was not upregulated in the Vhl^flox/flox ;opsin-cre knockdown mice. Thus, at least one neuroprotective factor contributing to protection after hypoxic preconditioning was missing. The complete protection of photoreceptors in Vhl knockdown mice after hypoxic exposure (data not shown) suggests that retinas lacking VHL in rods were still able to react to acute hypoxia and further supports the conclusion that mechanisms in addition to rod-specific activation of HIF and STAT3 transcription factors are required for full protection. 

Thus, the rod-specific knockdown of Vhl mimics the response of the retina to hypoxic preconditioning only partially. Whether or not a more complete retinal knockdown of Vhl would result in increased protection cannot be tested because generation of such mice would require an inactivation of Vhl in all cells of the adult retina. Inactivation of Vhl already during retinal development (e.g., by using the α-cre mouse) results in severe abnormalities of the retinal vasculature and strong retinal degeneration. ^22,35 Because long-term lack of VHL and/or activation of HIF transcription factors leads to retinal degeneration (Fig. 4), the Vhl knockdown may not only trigger a protective reaction but simultaneously may also alter the physiology of rods and/or of the cellular environment, leading to a retinal stress counteracting the protection. Hence, only a short hypoxic period may be tolerated by the retina and be protective. Long-term hypoxia ³⁶ or a long-term hypoxia-like response (this work) may lead to degeneration. 

In summary, we show that a photoreceptor-specific Vhl knockdown induced a hypoxia-like response that protected photoreceptors against light-induced cell death in 10-week-old mice. However, protection was only transient and did not reach the level detected after hypoxic preconditioning. We hypothesize that the full and sustained protection of photoreceptors after hypoxic exposure requires regulation of factors in addition to the ones observed here, and/or that hypoxia-mediated protection of photoreceptors is controlled mainly by other retinal cells in a paracrine fashion. 

Footnotes

 Supported by Swiss National Science Foundation (Grant 3100A0-117760), Deutsche Forschungsgemeinschaft (DFG, Grants Se837/5-2, Se837/6-1, Se837/7-1), German Ministry of Education and Research (BMBF, Grant 0314106), Fritz Tobler Foundation, and H. Messerli Foundation.

Footnotes

 Disclosure: C. Lange, None; S.R. Heynen, None; N. Tanimoto, None; M. Thiersch, None; Y.-Z. Le, None; I. Meneau, None; M.W. Seeliger, None; M. Samardzija, None; C. Caprara, None; C. Grimm, None

The authors thank Coni Imsand and Hedwig Wariwoda for excellent technical assistance.