We observed that a naturally occurring mouse strain developed age-related retinal degeneration (arrd2). These mice had normal fundi, electroretinograms (ERGs) and retinal histology at 6 months of age; vessel attenuation, RPE atrophy and pigmentary abnormalities at 14 months, which progressed to complete loss of photoreceptors and extinguished ERG by 22 months. Genetic analysis revealed that the retinal degeneration in arrd2 segregates in an autosomal recessive manner and the disease gene localizes to mouse chromosome 10. A positional candidate cloning approach detected a nonsense mutation in the mouse double minute-1 gene (Mdm1), which results in the truncation of the putative protein from 718 amino acids to 398. We have identified a novel transcript of the Mdm1 gene, which is the predominant transcript in the retina. The Mdm1 transcript is localized to the nuclear layers of neural retina. Expression of Mdm1 in the retina increases steadily from post-natal day 30 to 1 year, and a high level of Mdm1 are subsequently maintained. The Mdm1 transcript was found to be significantly depleted in the retina of arrd2 mice and the transcript was observed to degrade by nonsense-mediated decay. These results indicate that the depletion of the Mdm1 transcript may underlie the mechanism leading to late-onset progressive retinal degeneration in arrd2 mice. Analysis of a cohort of patients with age-related macular degeneration (AMD) wherein the susceptibility locus maps to chromosome 12q, a region bearing the human ortholog to MDM1, did not reveal association between human MDM1 and AMD.
Retinal degenerations are a group of disorders that cause irreversible loss of vision. Clinically, a wide variety of inherited retinal degenerations have been described both in animal models and humans (1). Genes involved in causing more than 140 human retinal degenerations have been identified (RetNet: http://www.sph.uth.tmc.edu/Retnet/). Orthologs of some of these genes were found to be associated with retinal degenerations in animal models. Mouse models with retinal degenerations have provided insight into the etiology of human retinal degenerations and molecular pathology of disease progression (1). Some of these animals develop late-onset retinal degeneration, but a majority of them show slow progression of the disease with a less severe phenotype and onset before the age of 6 months (2). Animal models with a shorter life span serve as valuable tools to study human diseases particularly conditions that manifest late in life (3).
Degeneration of the central part of retina, the macula, is observed in age-related macular degeneration (AMD) that affects 1 in 20 Americans of 65 years of age or older (4). AMD is a progressive and chronic degeneration that in its late stage may involve the photoreceptors, the underlying retinal pigmented epithelium (RPE), Bruch's membrane and possibly the choriocapillaries in the macular region of the eye. Clinical features of AMD include hypo- and hyper-pigmentation, deposition of drusen, loss of photoreceptors, RPE abnormality and formation of choroidal neovascularization membrane (5–7). Several genes associated with AMD have been reported (8–11) implicating perturbations in the complement pathway and other genes in susceptibility to disease. Despite these significant advances, much remains to be discovered about the biological mechanisms that lead to late-onset degeneration in AMD.
In this study, we describe the characterization of a novel and naturally occurring mouse model with autosomal recessive late-onset severe retinal degeneration, arrd2, primarily involving the neural retina. We mapped the disease gene to mouse chromosome 10 and identified a nonsense mutation in a novel retina expressed transcript of the Mus musculus double minute 1 (Mdm1) gene. This gene was first identified in transformed mouse 3T3 cells along with the oncoprotein Mdm2. Both Mdm1 and Mdm2 are localized to mouse chromosome 10. Although Mdm2 is a well-studied gene, very little is known about Mdm1 (12,13). In this article, we present the characterization of a novel retina-expressed transcript of the Mdm1, expression profile of different transcripts of this gene, localization of its expression in the retina, molecular pathology and potential effect of the Mdm1 nonsense mutation leading to degeneration of retina in arrd2 mice. We also examine the relation of the human MDM1 to AMD.
Phenotype of arrd2 mice retina
Fundus examination of the homozygous arrd2 mice revealed no abnormality at 6 months of age (Fig. 1A, i and iii). The retinas of these mice at 14 months showed significant retinal hypo- and hyper-pigmentation (Fig. 1A, ii and iv). Signs of retinal degeneration such as vessel attenuation, alterations in the RPE and the presence of retinal dots were noted at 14 months of age. Electroretinogram (ERG) responses of arrd2 mice were found to be normal at 6 months (data not sown). The dark-adapted and light-adapted ERG response showed progressive amplitude loss from 8 to 16 months (Fig. 1B). The rod response is minimal at 16 months with some sparing of the cone responses (Fig. 1B). By 22 months, both rod and cone ERG responses were undetectable (data not shown). Histological analysis of the retina of arrd2 mice showed slight reduction in the length of the outer segment (OS) and thickness of outer nuclear layer (ONL) by 9 months (Fig. 1C) and the length of OS and the thickness of ONL decreased progressively with age. By age 19 months, the OS and ONL were nearly absent in arrd2 homozygous mice (Fig. 1C). The RPE cells at this age were found to be atrophic and hypo-pigmented.
Ultrastructural analysis by transmission electron microscopy (TEM) indicated minimal changes in the retinas of arrd2 mice at 4 months. At this age, normal retinal architecture is maintained for the most part. In a few areas, minimal breakdown of OS architecture was observed suggesting increased numbers of phagosomes in the retinal pigment epithelium (Fig. 2A, black arrow). Inner segments, external limiting membrane and ONL are normal except for an occasional swollen synaptic terminal in the outer plexiform layer (Fig. 2B). The synaptic terminals have dense cytoplasm and large round mitochondria with complex, well-organized cristae (Fig. 2C). By 9 months of age, there is a prominent swelling and lucency of many synaptic terminals, some of which appear necrotic (Fig. 2D). In addition, swollen and necrotic cells are frequent in the inner nuclear layer adjacent to the outer plexiform layer, which is thinner than normal (Fig. 2E). The mitochondria of the synaptic terminals are swollen and demonstrate disorganization of the cristae (Fig. 2F, white arrows) and occasional malformation of the synaptic complexes (Fig. 2F, black arrow). The changes observed in the plexiform layers as early as 4 months indicate the possibility of early changes in the synaptic layers (Fig. 2C). OS fragmentation is a prominent feature (Fig. 2G, arrows) and there is a moderate increase in phagosome number in the retinal pigment epithelium. In addition, most of the apical processes of the retinal pigment epithelium are absent (Fig. 2H, arrow). Furthermore, the presence of lipofuscin-like granules was observed in the RPE (Fig. 2H, arrow-heads). Additional studies are needed to confirm lipofuscin accumulation in these mice.
Fundus examination of heterozygous arrd2 mice did not reveal abnormality even at older ages. The dark-adapted and light-adapted ERG response of arrd2/wt mice is normal at age 12 and 22 months (data not shown).
Genetic analysis showed that the late-onset retinal degeneration in arrd2 mice is an autosomal recessive trait. By linkage analysis of inter-crossed mice, we mapped the causative gene defect in arrd2 to mouse chromosome 10 in the genomic region flanked by D10Mit179 (116.7 m) and D10Mit267 (119.1 m) (Fig. 3A and B). Among the 35 genes localized to this interval, 9 were found to be expressed in retina. Sequence analysis of retina-expressed candidate genes-like Carboxypeptidase M (Cpm), Nucleoporin107 (Nup107), Transformed mouse 3T3 double minute protein2 (Mdm2), Solute carrier family 35 member E3 (Slc35e3), Glutamate receptor interacting protein 1 (Grip1), Kinesin family member C5C (Kifc5c), Interlukin 10-relative T cell-derived protein (Iltifb), Interferon gama (Infg) revealed no mutational changes.
Furthermore, a single base C to T nonsense substitution (CGA to TGA) at position 1195 (Arg 399 stop) in exon VIII of a novel transcript of the Mdm1 gene was observed in the homozygous state (GenBank ID: 1064299). The Mdm1 nonsense mutation is predicted to result in truncation of the protein by 320 amino acids due to a premature termination at codon 399 (Fig. 3C). This single base substitution was confirmed by sequencing additional arrd2 homozygous mice and this change was not detected in C57BL/6J, 129 Sv/J, CAST/EiJ and six additional wild strains of mice available at Jackson laboratories.
Characterization of the mouse Mdm1 gene
The mouse Mdm1 gene is known to have at least two transcripts, a longer transcript (LT) containing 14 exons (GI: 27229284) (Fig. 4A–I) and a shorter transcript (ST) (Fig. 4A, ii) containing only the first four exons (GI: 6754669). We observed a novel transcript in the retina by direct sequencing of mouse retinal cDNA, which is similar to the previously reported LT except for having an additional exon (total 15 exons) (GI: 1064299). The supplementary exon contains a 30 bp in-frame additional sequence (exon VIIa) and is located between exons VII and VIII. We refer to this transcript as the retinal transcript (RT) (Fig. 4A, iii). The Mdm1 RT consists of an ORF of 718 amino acids and could encode a protein of nearly 80.7 kDa, the LT consists of an ORF of 708 amino acids and could encode a protein of nearly 79.6 kDa, and the ST consists of an ORF of 222 amino acids and could encode a protein of nearly 25.1 kDa. We also identified a variant of the RT which lacks exon IV (Fig. 4A, iv) (GI: 1064301). This confirms the existence of multiple Mdm1 transcripts (12,13). As the arrd2 mutation is in exon VIII, the LT and the RT will be affected by this mutation, and the short transcript (ST) that contains only four exons will not be affected.
Distribution of the LT and RT was tested by PCR and restriction digestion of these transcripts with restriction enzymes, Nla III and Sex-A1 which specifically cut the RT and the LT without exon VIIa, respectively. Mouse tissue cDNAs were amplified with primers encompassing exon VI to exon IX and subjected to restriction digestion with the above enzymes. The retina expresses only the LT containing the novel additional exon, the RT (Fig. 4B, lanes: 10–12). Brain, lungs, anterior eye segments, and posterior eye segments (choroid-sclera, PS) express only the RT (Fig. 4B, lanes: 1–3, 4–6, 7–9, 13–15, respectively). The kidney, skeletal muscle, testis and uterine tissues express both RT and LT (Fig. 4B, lanes: 16–18, 19–21, 22–24, and 25–27, respectively). This analysis indicates that the novel transcript, RT, is expressed in several tissues including the retina. Sequence-based homology search revealed that this 30 bp region is also present in the human MDM1 transcript. The significance of this 30 bp additional exon is not known. Blast search did not reveal any significant homology with any other known nucleotide or protein sequences in the databases (except Mdm1 orthologs) indicating that this is a novel transcript.
We further studied the expression of different Mdm1 transcripts in mouse tissue in order to understand the potential mechanism underlying retinal degeneration due to the nonsense mutation in the Mdm1 gene. Using primers specific to the Mdm1 RT, LT and short transcript (ST), the level of expression of each transcript was determined in different mouse tissues by quantitative RT–PCR (qRT–PCR). Among all the tissues tested, the highest level of expression of the RT was detected in the retina (Fig. 4C, top panel). Significant level of expression of this transcript was also detected in the testis, whereas low level of expression is observed in other tissues, including iris-ciliary body, kidney, skeletal muscle, heart and brain (Fig. 4C, top panel). The previously reported ST of the Mdm1 gene showed significantly high level of expression in testis and very low level of expression in several tissues, including retina (Fig. 4C, middle panel). The previously known long transcript (LT) of the Mdm1 gene was also found to be expressed at very high levels in testis (Fig. 4C, bottom panel). Earlier we observed that this LT was not present in brain and eye tissues, including the retina (Fig. 4B). Minimal levels of expression of LT were observed in kidney, skeletal muscle and uterine tissues by qRT–PCR.
Similar to several other retinal genes, the highest levels of Mdm1 expression are observed in testis in addition to retina (14). Although the RT is the predominant transcript in the retina, low levels of Mdm1 ST were also observed. Expression of Mdm1 transcripts in other tissues tested was very low to insignificant compared to their levels in the retina and testis.
Retinal expression and location of Mdm1 transcripts
We studied the expression profile of Mdm1 in control (C57BL/6J) mice retina during post-natal development and aging. In developing eyes of normal mice, expression of this gene was detected from post-natal day 1 (data not shown). In the retina, a gradual increase in the expression of Mdm1 transcripts, RT and ST, was observed from 1 to 6 months of age (Fig. 5A). After 6 months, the expression level remained unchanged up to 20 months, the highest age point we have tested, indicating that the presence of high level of Mdm1 expression may be important for adult retinal tissue.
By in situ hybridization, expression of the Mdm1 gene (the RT) in wild-type albino mouse retinal sections was detected in all the nuclear layers of the neural retina: the ONL, inner nuclear layer and also the ganglion cell layer (Fig. 5B, arrows). Presence of Mdm1 was not detected in the RPE.
Molecular characterization of arrd2 retina
To gain insight into the pathology of degeneration, we studied the expression profile of different marker genes in arrd2 and age-matched wild-type control mice (C57BL/6J) retina at three time points: 5, 12 and 18 months of age. In Figure 6, we presented the expression profile of selected marker genes in arrd2 retina in comparison to controls, by taking the expression values of these genes in age-matched controls as 100%. At 5 months, the expression of rod OS proteins, rhodopsin and RDS, was not significantly different from wild-type retina, but reduced to ∼40% by 12 months and to insignificant levels by 18 months (Fig. 6A). Interestingly, rod inner segment protein arrestin (Sag) expression was higher (150–180% of wild-type) in arrd2 retina at 5 and 12 months (Fig. 6A). The cone photoreceptor marker, short-wave length cone opsin (Opn1-sw) was slightly reduced (70%) at 5 months and subsequently remained unaltered up to 12 months (Fig. 6B). Medium-wavelength cone opsin (Opn1-mw) was reduced to ∼45% at 5 months and gradually reduced to ∼25 and ∼5% by 12 and 18 months, respectively (Fig. 6B). Furthermore, the expression of cone inner segment protein cone transducin (Gnat2) showed gradual reduction and remained at ∼40% of the levels compared to controls at 18 months when the levels of most of the OS proteins are insignificant in arrd2 mice retina. Expression of Elongation of very long chain fatty acid 4 (Elovl4), which is an inner segment protein of both rod and cone photoreceptors, is reduced to ∼60% by 12 months and to ∼20% by 18 months (Fig. 6B). In summary, the expression profile of retinal cell marker genes suggests loss of both rod and cone photoreceptors cells between 5 and 12 months and sparing of a few cones at 18 months. The degeneration of cones appears to be slower than rods and these observations are consistent with the observations made by histology and ERG analysis (Fig. 1).
Expression of glial fibrillary acidic protein (Gfap), a common retinal stress response marker, was found to be more than 20-fold higher (>2000% of control) even at 5 months (Fig. 6C). It gradually increased to >20 000% (200-fold) by 18 months (Fig. 6C). Expression of oxidative stress or redox-mediated stress markers Heme oxygenase 1 (Ho-1) and Ceruloplasmin (Cp) was not altered up to 12 months. Subsequently, expression of these genes increased significantly at 18 months, when the retina is appreciably degenerated (Fig. 6C). The apoptosis marker Caspase 3 (Casp3) and angiogenesis marker Vascular endothelial growth factor (Vegf) also showed expression pattern similar to Ho-1 and Cp (Fig. 6D).
Fate of Mdm1 transcripts in arrd2 retina
By qRT–PCR analysis of arrd2 mouse retina, the level of Mdm1 RT was found to be reduced to ∼40% of wild-type (C57BL/6J) retina by 5 months (Fig. 7A). By 12 months, the level of RT in the retina of arrd2 mice was <20% of the levels observed in C57BL/6J retina (Fig. 7A). The level of expression of RT was further reduced to ∼5% of the levels observed in C57BL/6J retina by 18 months (Fig. 7A). On the other hand, the Mdm1 small transcript (ST) showed elevated expression at 5 months: >250% of C57BL/6J retina (Fig. 7A). Subsequently, it was also reduced to ∼25 and ∼20% of the wild-type by 12 and 18 months of age, respectively (Fig. 7A). Significant reduction in the quantity of RT in arrd2 retina before the onset of retinal degeneration suggests degradation or specific down regulation of the mutant transcript.
To further assess the fate of Mdm1 RT, we analyzed arrd2 retina and testis samples at 6 months by northern blot analysis using a probe that recognizes Mdm1 RT and LT. The RNA samples isolated from C57BL/6J were used as controls for comparison. A strong ∼3.0 kb band representing Mdm1 transcript was observed in the RNA isolated from C57 retina and testis (Fig. 7B; lanes 3 and 7). The bands corresponding to Mdm1 LT or RT detected in arrd2 mice retina and testis are very faint and almost insignificant in comparison to the intensity of these bands in retina and testis from age-matched control animals (Fig. 7B; lanes 4 and 8). A set of doublet bands were detected in the testis (Fig. 7B; lane 7). Both RT and LT are expressed in testis (Fig. 4). Therefore, the doublet bands detected by northern blot analysis in the C57BL/6J testis are likely to represent these two transcripts of the Mdm1 gene. No significant expression is detected in kidney (Fig. 7B; lanes 1 and 2) and brain (Fig. 7B; lanes 5 and 6) tissues. The top and bottom panels of the Figure 7B showed the RNA quality and equal loading controls.
In agreement with the above observations, no significant labeling of Mdm1 was detected in arrd2 mouse retina by in situ hybridization at 3 months and at 10 months (data not shown). The mutation in the Mdm1 gene in arrd2 mice is a nonsense mutation in exon VIII which is 960 bp upstream of the original stop codon. Therefore, it is likely that this transcript is undergoing nonsense-mediated decay (NMD) in arrd2 mice.
Assay for nonsense-mediated decay of Mdm1 transcripts
To address the hypothesis that the reduction of the Mdm1 transcript in the arrd2 mouse retina is through NMD, we carried out NMD assay using retinal explants in culture and inhibiting translation with cycloheximide. When a nonsense mutation is detected by the translational machinery, through cellular proofreading, the nonsense mutation containing transcripts are degraded. Therefore, if the translation process in the cell is blocked, all transcripts get accumulated including the mRNA carrying the nonsense mutation.
Significant accumulation (∼9-fold, P = 0.004) of the Mdm1 RT was detected by qRT–PCR when the translation is blocked by cycloheximide for 4 h in 2-month-old arrd2 mice retina (Fig. 8A; aR = arrd2 retina). These accumulated transcripts were degraded to near zero levels within 2 h of withdrawal of the cycloheximide (Fig. 8A). In the control C57BL/6J retina, cycloheximide treatment increased the quantity of Mdm1 transcript (not significantly) which returned back to normal level when the drug was withdrawn (Fig. 8A; CR = C57 retina). Similar results were observed when testis tissue of both the arrd2 and control mice was subjected to NMD assay (Fig. 8B; aT = arrd2 testis and CT = C57 testis). As an internal control, levels of β-Actin were tested in the arrd2 mouse retinal and testis samples. Slight accumulation (∼1.5-fold) of the β-Actin transcript was observed after 4 h of cycloheximide treatment in retinal samples, and after withdrawal of the drug the levels of β-Actin transcript returned to normal, but not to the near zero levels as observed for Mdm1 transcript (Fig. 8C). No change was observed in the levels of this gene in testis tissues (Fig. 8C). We also tested the level of the ST in arrd2 mice. It also showed slight accumulation with the treatment, but was significantly reduced after the removal of cycloheximide (Fig. 8D). These observations suggest that the stop codon mutation in the Mdm1 gene results in degradation of the RT through NMD.
Evaluation of the association of MDM1 gene with age-related macular degeneration in patients
To investigate the contribution of MDM1 to AMD risk in humans, an association analysis was conducted in the ‘Family Age Related Maculopathy Study (FARMS)’ cohort, a Caucasian family-based study population in which an AMD locus was mapped to a region on chromosome 12 that is syntenic to the region on mouse chromosome 10 harboring Mdm1. The human linkage peak on chromosome 12q is broad, spanning almost 36 cM and harboring 636 genes, 179 having known retinal expression according to BioMart and the GNF expression database. Three SNPs were chosen for analysis based on their predicted ability to represent the linkage disequilibrium (LD) structure at the MDM1 locus. In addition, two of these variants result in missense mutations within the gene. Testing was completed using a variance component family-based association model that included family and sibling effects. None of the three SNPs displayed significant associations with AMD severity. rs962976 and rs2306392, the missense variants, were not statistically significant under any model tested. rs973328, a variant slightly 3′ to MDM1, demonstrated the most significant finding of association with AMD severity of the tested SNPs (P = 0.08, dominant model).
Database of genome wide association studies (dbGaP) was consulted to see if MDM1 showed any significant findings in a sample of participants in the Age-Related Eye Disease Study (AREDS) cohort. The three SNPs examined in the FARMS cohort were also present in the AREDS analysis and were not reported to be significantly associated with AMD. In addition, the AREDS study reported non-significant findings of association between another six SNPs within MDM1 and AMD. Thus, we conclude that the Mdm1 locus is not the AMD locus on human chromosome 12.
We identified a naturally occurring mouse model with age-related progressive retinal degeneration (arrd2). These mice showed subnormal rod and cone ERGs at the age of 6 months progressing to undetectable ERG response by 22 months; slight decrease in photoreceptor nuclei at 9 months, to significant loss of photoreceptor layer by 20 months. We have mapped the disease locus to the mouse chromosome 10 and detected a nonsense mutation in the mouse double minute 1 (Mdm1) gene that is likely to result in NMD of the transcript. Our studies indicate that the late-onset progressive retinal degeneration observed in these mice is likely to be due to the loss of functional MDM1 in the retina of arrd2 mice.
The expression profile of photoreceptor-specific genes, electrophysiological responses and morphology of retina in arrd2 mice indicated late-onset progressive photoreceptor degeneration with severe degeneration of both rods and cones. Although the age of onset is late, the rate of degeneration is much faster after 8 months with cone degeneration slower than the rod degeneration. The phenotype of these mice with attenuation of retinal vessels, ERG abnormalities resembles late-onset, progressive rod-cone degeneration in humans. A large number of genes that cause retinal dystrophy in mice are also associated with retinal conditions in humans (15) and analysis of patients with clinical features similar to the phenotype observed in these mice may reveal the involvement of this gene in causing human retinal degeneration.
The retinal degeneration observed in arrd2 mice shares some similarities with AMD in humans, including late age of onset, fundus hypo-pigmentation and RPE atrophy (16). Furthermore, the critical interval of an AMD susceptibility locus localized to human chromosome 12 includes the human MDM1 gene (17,18). We have tested the association between MDM1 and susceptibility to AMD in two cohorts of patients, one a set of families with disease and the other a case–control study. Both of these analyses revealed no significant association between MDM1 and human AMD. AMD is a multigenic and genetically heterogeneous disease. Lack of association in particular cohorts of patients does not obviate the involvement of the MDM1 gene product in AMD pathogenesis. It is possible that this gene may not be directly associated with susceptibility to AMD but still may play a major role in the aging retina.
We detected four major transcripts of the Mdm1 gene including a novel RT. The novel transcript is the predominant transcript in the retina and the highest level of expression of Mdm1 RT in the retina suggests a critical role for the Mdm1 RT in retina. Absence of MDM1 protein may affect the retinal function and lead to degeneration. The Arg 399 Stop mutation found in the Mdm1 gene affects the RT and also the LT but not the ST. The Arg 399 Stop mutation observed in arrd2 mice is likely to result in nonsense-mediated decay of the transcript, resulting in a null allele or severely depleted truncated protein. The retinal pathology observed in the arrd2 homozygous mice and lack of pathology in heterozygous mice indicates that the absence of the Mdm1 gene product may underlie the mechanism leading to retinal degeneration in these mice. Presence of high levels of Mdm1 ST may compensate for the loss of LT and RT in the testis.
The decrease observed in the levels of expression of photoreceptor-specific genes in arrd2 mice could be due to the loss of photoreceptor cells rather than specific down regulation of the genes. However, the significant up-regulation observed in stress response gene Gfap at 5 months indicates that the biological events leading to degeneration of retina in arrd2 mice are active by this age. Consistent with these observations, abnormalities were noted in the retinal architecture of arrd2 mice by ultra structural analysis at this age. Finally, alteration of Casp3 mRNA is consistent with significant photoreceptor loss and indicates activation of apoptosis following initiation of retinal degeneration in arrd2 mice.
The biological function of the MDM1 gene is not known. Double minutes are chromosome-associated paired acentromeric extrachromosomal bodies that represent amplified sequences. Double minutes have been described in a variety of tumor cells and transformed cell lines. From the sequences that are amplified and over-expressed in a spontaneously transformed mouse 3T3 cell line, four double minute genes Mdm1, Mdm2, Mdm3 and Mdm4 (or MdmX) have been described. Mdm1, Mdm2 and Mdm3 are located on mouse chromosome 10, whereas Mdm4 is localized to mouse chromosome 1. Detailed studies were carried out on both Mdm2 and MdmX genes, whereas very limited information is available on the Mdm1 and Mdm3 genes. Mdm2 and Mdm4 (MdmX) negatively regulate the function of P53 in a synergistic manner (19,20). MDM2 is an ubiquitin ligase that catalyses ubiquitination of p53 and itself (21,22). MDM4 inhibits P53 by binding and masking the transcriptional activation domain. It has been demonstrated that MDM2 and MDM4 are required for proper development of the central nervous system. Loss of MDM2 and MDM4 results in a P53 dependent apoptotic phenotype in the developing CNS. In arrd2 mice with Mdm1 nonsense mutation, development of retina was found to be normal. Unlike Mdm2 and Mdm4 null mice, apoptosis in arrd2 mice appears to be activated much later and subsequent to initiation of photoreceptor loss. Furthermore, no sequence homology was observed between the Mdm1 and other Mdms. Therefore, although the sequence of Mdm1 is amplified in transformed 3T3 cells, the biological role of this gene may be different from other members of the double minute gene family. Neither the RT nor other transcripts of Mdm1 showed significant homology with any known sequences. However, presence of a coiled coil domain and reported localization of the Mdm1 protein to the nucleus suggest a role for MDM1 in transcriptional machinery like MDM2 (13). Additional studies are needed to determine the function of this gene.
Although the expression of Mdm1 was detected from very early stages of eye development, the onset of retinal degeneration phenotype in arrd2 mice was observed after 8 months of age. Expression of stress response genes and retinal ultrastructure analysis revealed significant changes in the retina of arrd2 mice as early as 5 months. Expression of Mdm1 was found to be increased with age and sustained high levels of expression were maintained at older age in the retina of control mice suggesting the presence of this protein is required in aging retina. The late-onset retinal degeneration observed in these mice is either due to that requirement of high level of MDM1 with age or to a mechanism yet to be determined. Null mutations in other genes expressed at early stages of development leading to late-onset phenotypes have been reported previously (23–25). High levels of Mdm1 expression are also observed in testis which includes high level of ST which is not affected by the non-sense mutation that might compensate to some extent the loss of LTs. Higher level of ST or low activity of the residual truncated transcript at early ages and/or cumulative effect of the absence of the Mdm1 gene may underlie the late onset retinal degeneration phenotype in these mice. Evaluation of the functional role of this gene and the mechanism underlying late-onset retinal degeneration in these mice may reveal the critical biological pathways that operate in maintaining the normal function of aging retinal tissue.
MATERIALS AND METHODS
The mice in this study were bred and maintained in standard 12 h dark and 12 h light cycle and standardized conditions in the Research Animal Facilities at the Jackson Laboratory, Bar harbor, ME, and University of Michigan, Ann Arbor, MI. They were maintained on NIH 31.6% fat chow and acidified water, with a 14 h light/10 h dark cycle in conventional facilities that were monitored regularly to maintain a pathogen-free environment. All experiments were approved by the Institutional Animal Care and Use Committee and conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.
Clinical retinal evaluation
One hundred and twenty-eight mice used in clinical characterization studies had pupils dilated with 1% atropine ophthalmic eye drops and were evaluated by indirect ophthalmoscopy with a 78 diopter lens. Fundus photographs were taken with a Kowa Genesis small animal fundus camera (Torrance, CA, USA) (26).
After at least 6 h of dark adaptation, mice were anesthetized with an intraperitoneal injection of normal saline solution containing ketamine (15 mg/gm) and xylazine (7 mg/gm body wt). ERGs were recorded as described earlier (27). Dark-adapted responses were recorded to short-wavelength (λmax = 470 nm; Wratten 47A filter) flashes of light over a 4.0-log unit range of intensities (0.3 log unit steps) up to the maximum allowable by the photic stimulator. Light-adapted responses were obtained with white flashes (0.3 steps) on the rod-saturating background after 10 min of exposure to the background light to allow complete light adaptation. Signals were sampled every 0.8 ms over a response window of 200 ms. The actual flash intensities used to obtain the representative ERG waves for dark-adapted responses were 0.6 and 2.1 log cd s/m2 and for light-adapted responses were 0.9 and 1.9 log cd s/m2, respectively. The ERG response obtained from same mouse at different light intensities and ages is presented in Fig. 1B.
Light and transmission electron microscopy
Both eyes from arrd2 mutant mice and age-matched C57BL/6J mice were enucleated and eyecups were prepared for light and electron microscopic examination as described previously (27). Two to six eyes were used for each age group. For each sample, 0.5 µm thin sections were stained with toluidine blue for light microscopy followed by ultrathin section preparation for TEM examination.
Gene mapping and sequencing
arrd2 mice were mated with B6 mice and the F1 mice, which exhibit no retinal abnormalities, were intercrossed to produce F2 mice. Tail DNA was isolated as previously reported (28). DNAs of 73 F2 offspring were genotyped using microsatellite markers to develop a structure map of the region. Initially, a genome scan of microsatellite (Mit) DNA markers was carried out on pooled DNA samples (29). After detection of linkage on chromosome 10, individual DNA samples were analyzed with microsatellite markers D10Mit96, D10Mit78, D10Mit70, D10Mit162, D10Mit135, D10Mit233, D10Mit179, D10Mit164, D10Mit267, D10Mit237, D10Mit297.
The candidate genes localized to the critical interval between markers D10Mit179 and D10Mit267 were identified by scanning the mouse genomic sequence available through NCBI. A total of 35 genes including putative genes have been localized to this interval and 9 of them were reported to be expressed in the retina. We selected these 9 genes for mutation analysis of arrd2 mice. PCR primers to amplify coding regions were designed using the mouse genomic sequence encompassing the candidate genes. PCR products were purified from agarose gels using a Qiagen kit (Qiagen Inc. Valencia, CA, USA). Sequencing reactions were carried out with automated fluorescence tag sequencing.
RNA isolation and real-time quantitative RT–PCR
To evaluate the transcripts of the Mdm1 gene and their expression by real-time qRT–PCR, RNA was isolated from different tissues of C57BL/6J mice. Mice were dark-adapted overnight and liver, lungs, brain, ciliary body and iris (eye anterior), posterior segment of the eye (containing RPE, choroid and sclera), retina, lens, cornea, heart, spleen, kidney, skeletal muscle, skin, testis and uterus were collected in TRIzol reagent (Invitrogen, Carlsbad, CA, USA) at the end of the dark cycle. RNA was isolated from these tissues following the manufacturer's protocol. All RNA samples were purified and single strand cDNA was prepared for qRT–PCR.
To study the age-related expression of Mdm1 in retina, cDNA was prepared from C57BL/6J retina of 20, 35, 50, 180, 300, 420, 500 and 600 days old mice. Similarly, for a comparative study of retinal cell-specific marker gene expression, cDNA was prepared from the retinal RNA of 5, 12 and 18 months old arrd2 and C57BL/6J mice. Primer pairs to amplify specific transcripts were designed for qRT–PCR (Fig. 4A, Supplementary Table 1). The levels of expression of four genes Gapdh, Hgprt, Actin-β and RpL19 were used as controls to normalize the Mdm1 gene expression and the data were analyzed using previously published protocols (14,30,31). Mean (± SEM) relative expression values calculated from at least three independent samples are presented.
In situ hybridization
A cDNA fragment (620 bp) that specifically recognizes the Mdm1 LT and RT was amplified from C57BL/6J mice retinal cDNA using primers located in exon 7 (forward) and exon 10 (reverse) and tagged to T7 and T3 promoters, respectively (Fig. 4A). The sequence of primers will be available from the authors upon request. The Mdm1 ribo-probes (sense and antisense) were generated by labeling the amplified cDNA fragment using a DIG-RNA labeling kit (Boehringer Mannheim GMBH, Mannheim, Germany). In situ hybridizations were performed following previously published protocols (32,33).
Northern blot analysis
Northern blots were generated using 5–7 µg of purified total RNA isolated from kidney, brain, retina and testis of C57BL/6J and arrd2 mice. Mdm1 cDNA probe was generated from the same region of the gene that is used for generating the in situ hybridization probe. Probe for human β-ACTIN gene which recognizes its mouse ortholog was obtained from Clontech. Hybridization was carried out using Express Hyb (Clontech) following previously published protocols (14).
Assay for nonsense-mediated decay
Retinal and testis explants of C57BL/6J and arrd2 mice were cultured in retinal medium as described by Hatakeyama and Kageyama (34) on millicell cell culture membrane (Millipore, Bedford, MA, USA) for 12 h at 37°C with a 5% atmospheric CO2 concentration (34). After 12 h, the media was replaced with fresh retinal media in one set of samples which served as controls. The media of the second set of samples was replaced with fresh media containing 100 µg/ml of cycloheximide and incubated for 4 more hours. After 4 h incubation, tissues were harvested from the control sample and one set of the cycloheximide treated samples. The second set of cycloheximide treated replica samples were washed with RNAse free phosphate buffered saline to remove the cycloheximide and incubated for an additional 2 h with fresh retinal medium. Testis tissues from both C57BL/6J and arrd2 mice were also treated with cycloheximide in a similar manner. Total RNA was isolated from each sample, converted to first-strand cDNA and used for qRT–PCR. Quantitative expression of the Mdm1 transcripts was determined by qRT–PCR. Expression of the β-Actin gene was measured as an internal control for NMD in arrd2 mice retina and testis samples. C57BL/6J samples served as independent controls for the Mdm1 nonsense-mediated decay in arrd2 mice.
Association of Mdm1 with AMD susceptibility
A cohort of patients from the Family Age-Related Maculopathy Study (FARMS) was examined for evidence of association between MDM1 and AMD. Subjects were recruited through the Retina Clinic at the University of Wisconsin, Madison and evaluated as previously described (17). The standardized clinical evaluation did not include fluorescein angiography, preventing direct comparison with the angiographic features identified in the current mouse model. Families were identified through severely affected index cases; in all, 293 individuals from 34 families were included in the current study. The human MDM1 gene was identified by orthology to mouse chromosome 10 and the gene LD structure was analyzed in the HapMap data (CEU population: Centre d'Etude du Polymorphisme Humain) using Haploview (35). We identified two LD blocks in this gene (see Supplementary Material). Three SNPs were chosen to cover the LD blocks, with two SNPs spaced across the larger block. Two of the chosen SNPs code for missense variants. Combining the data on these three SNPs effectively captures most of the common variants within MDM1 (HapMap pairwise Tagger, r2 ≥ 0.8, MAF ≥ 0.2) (36).
Genotyping was conducted as part of a custom GoldenGate Illumina panel. One sample was excluded from the analysis due to a call rate of less than 90% for the entire panel. No inconsistencies in genotyping were detected using MARKERINFO (S.A.G.E. v5.3) (http://genepi.cwru.edu/) and comparison of 28 replicates. Each SNP was tested for association with a 15 step AMD phenotype severity scale based on standardized grading of stereoscopic 30° color fundus photographs, using a variance component family-based model incorporating both sibling and family effects in ASSOC (S.A.G.E.). SNPs were individually tested under additive, dominant and recessive inheritance models.
As a virtual replication, the database of Genotype and Phenotype (dbGaP) (37) was queried to identify any MDM1 variants analyzed in a genomewide association analysis conducted by the National Eye Institute Age Related Eye Disease Study (AREDS) (38). The AREDS study examined a Caucasian cohort of approximately 400 severely affected AMD patients and 200 controls using 100 K Affymetrix and Illumina panels, and conducted single marker tests for association with AMD status.
This study was supported by the grants from the National Institutes of Health grants EY13198, U10EY06594, EY015286, EY13438, EY10605, EY015810, NIH T32-EY07157, T32 GM007250, EY07758, RR01183, Research to Prevent Blindness, Inc., New York, NY (R.A.) and EY07758 (B.C.). Foundation Fighting Blindness (R.A., J.R.H., B.C.), Retina Research foundation. Institutional shared services are supported by National Cancer Institute, Cancer Center grant CA34196. The results of this paper were obtained by using the software package S.A.G.E., which is supported by a US Public Health Service Resource Grant (RR03655) from the National Center for Research Resources. This research was supported by the Gene Expression and Genotyping Facility of the Comprehensive Cancer Center of Case Western Reserve University and University Hospitals of Cleveland (P30CA43703).
The authors thank Austra Liepa for maintenance of the mice, and technical assistance, and Mitchell Gillett for tissue preparation for morphological analysis.
Conflict of Interest statement. None declared.