A Strong and Highly Significant QTL on 6 that Protects the Mouse from Age-Related Degeneration

Michael Danciger,1 Jessica Lyon,1 Danielle Worrill,1 Matthew M. LaVail,2 and Haidong Yang2

PURPOSE. BALB/cByJ (C) albino mice have significantly more etiology. Many studies have attempted to determine environ- retinal degeneration as they age than C57BL/6J-c2J (B6) albinos. mental risk factors that may be associated with AMD, but only To discover the genetic loci that influence age-related retinal smoking has been consistently demonstrated to be one of them degeneration (ARD), a quantitative genetics study was per- (for reviews, see 6–9). In contrast, twin studies and population- formed with 8-month-old progeny from an intercross between based familial aggregate studies have made it clear that these two strains. play a significant role in AMD. 10–13 It is also clear that AMD is 14,15 METHODS. The thickness of the outer nuclear layer of the retina a complex genetic disorder, one that first appears most was used as the quantitative trait. A genome-wide scan was commonly in elderly individuals, typically in those older than performed with 86 genetic markers at an average distance of 50 years. Because of the age of onset, informative family ped- 15.7 cM. Map Manager QTX was used to analyze the data. igrees of the size needed to identify genetic loci are difficult to find. Only one such (with a LOD score of 3.0) has been RESULTS. Three highly significant quantitative trait loci (QTLs) reported.16 Identifying AMD genetic loci with family-based were detected on mouse (Chrs) 6, 10, and 16. aggregate studies is difficult for the same reason and because The B6 alleles were protective against ARD in the first two, and the phenotype of AMD is heterogeneous. Is it one disease or a the C allele was protective in the third. Several suggestive, group of diseases? Are different phenotypes caused by the weak QTLs were also found, along with a gender-related effect. same or various genetic factors? To our knowledge, only one The strongest and most highly significant QTL on Chr 6 ac- study of this type has been reported. In the latest refinement of counted for 30% of the total genetic effect with a LOD score of this study, several loci were found in a very large cohort with 13.5. The RPE65/MET450 variant of major influence on con- LOD scores ranging from 2.0 to 3.16.15 stant light-induced retinal degeneration (LRD) in a previous study of these same two mouse strains had no influence on Therefore, when we observed that two albino mouse strains undergo significantly different rates of retinal degeneration as ARD, and only some of the weak, suggestive QTLs influencing 17 ARD were also observed in LRD. they age, we decided to perform a quantitative genetic study to find the chromosomal loci of the mouse genes responsible CONCLUSIONS. Because none of the ARD QTLs was homologous to for the difference. This would be the first step toward finding human chromosomal loci so far implicated in age-related macular the genes themselves—particularly the protective alleles of degeneration, each represents a new candidate for potential those genes—that influence age-related retinal degeneration. study. The gene represented by the Chr 6 QTL is of particular Compared with studies in humans, experiments with mice are interest because it has broad influence, very high significance, and much simpler to perform and are statistically more powerful. a B6 allele that protects against ARD. (Invest Ophthalmol Vis Sci. There are only two alleles to consider, one from each of the 2003;44:2442–2449) DOI:10.1167/iovs.02-1252 inbred strains. There is no problem with variation in pheno- type, because measurement of the thickness of the outer nu- ge-related (AMD) is the most com- clear layer (ONL) of the retina serves as a quantitative trait. Amon cause of severe, irreversible vision loss in developed There is no problem with confounding environmental influ- 1–5 nations, but little conclusive information is known about its ences because all the mice are exposed to the same conditions in the vivarium, and there is no problem recruiting subjects. Genes found to influence age-related retinal degeneration in From the 1Department of Biology, Loyola Marymount University, the mouse would be excellent candidates for study in human Los Angeles, California; and 2Beckman Vision Center, University of AMD (even though the mouse does not have a macula or a California San Francisco School of Medicine, San Francisco, California. fovea), because mouse and human genes involved in vision are Supported in part by NIH Grants EY13280 (MD) and EY01919 and often similar in effect. This is exemplified by the fact that EY02162 (MML), the Foundation Fighting Blindness (MD, MML), Re- mutations in several mouse vision genes such as Pdeb, Prph2, search to Prevent Blindness, That Man May See, Inc., and the Macula and Nr2e3 cause the same type of disease in mice that muta- Vision Research Foundation (MML). The Center for Inherited Disease tions in human orthologous genes (PDE6B, RDS-peripherin, Research is fully funded through a federal contract from the National 18–28 Institutes of Health to The Johns Hopkins University (Contract Number and NR2E3) cause in humans. N01-HG-65403). MML is a Research to Prevent Blindness Senior Scien- In this study, with a large F1 intercross between the mouse tist Investigator. strains BALB/cByJ (C) and C57BL/6J-c2J (B6), we identified a Submitted for publication December 6, 2002; revised January 21, number of quantitative trait loci (QTLs) containing genes that 2003; accepted February 13, 2003. influence age-related retinal degeneration. Among them were Disclosure: M. Danciger, None; J. Lyon, None; D. Worrill, three strong and highly significant QTLs and several weaker None; M.M. LaVail, None; H. Yang, None QTLs. Although most of the QTLs reflected B6 alleles that were The publication costs of this article were defrayed in part by page protective, a few were the opposite; C alleles were protective. charge payment. This article must therefore be marked “advertise- ment” in accordance with 18 U.S.C. §1734 solely to indicate this fact. In addition, although the net relationship between B6, F1, and Corresponding author: Michael Danciger, Department of Biology, BALB/cByJ age-related, retinal degeneration control animals Loyola Marymount University, Los Angeles, CA 90045-2959; was dominant for B6, QTLs with additive and recessive rela- [email protected]. tionships also were found once the loci were teased apart.

Investigative Ophthalmology & Visual Science, June 2003, Vol. 44, No. 6 2442 Copyright © Association for Research in Vision and Ophthalmology

Downloaded from iovs.arvojournals.org on 09/27/2021 IOVS, June 2003, Vol. 44, No. 6 QTL Protective against Age-Related Retinal Degeneration 2443

Although the C57BL/6J-c2J albino isogenic to C57BL/6J was a the two most distal markers were 45 cM apart, Chr 8 where the two good choice to cross with the albino BALB/cByJ, it introduced most proximal markers were 39 cM apart, and Chrs 2, 5, 11, 12, and X potential epistatic confounds. The c allele (mutant tyrosinase where the most proximal markers were 48, 20, 17, 16, and 17 cM from gene) may have masked retinal degeneration gene alleles that the centromere, respectively. A list of markers used in crosses between would have been expressed on a pigmented C57BL back- C57BL and BALB/cByJ mice is available on the CIDR Web site (http:// ground or exposed alleles that would not have been expressed www.cidr.jhmi.edu/). A few additional dinucleotide repeat markers on that background. Last, because we used these same two within QTLs were analyzed in our laboratory. These were amplified by strains of mice in a previous quantitative genetic study of standard PCR methods and electrophoresed in 4% agarose gels for light-induced retinal damage,17 we were able to compare the allele determination by size. All map positions were based on the QTLs between the two studies to determine what influence the Encyclopedia of the Mouse Genome from Jackson Laboratories Mouse genes that modify light-induced retinal degeneration have on Genome Informatics (MGI; http://www.informatics.jax.org/; provided age-related retinal degeneration. in the public domain by Jackson Laboratories, Bar Harbor, ME). Mouse Genomic DNAs MATERIALS AND METHODS Genomic DNAs were isolated from liver tissue with a kit (Puregene; Mice Gentra Systems, Minneapolis, MN). BALB/cByJ and C57BL/6J-c2J albino mice were originally purchased Data Analysis from the Jackson Laboratories (Bar Harbor, ME) although some were maintained through many generations in our vivarium before study. Genotypes versus quantitative traits were analyzed with the Map Man- 30 C57BL/6J-c2J mice are derived from a C57BL/6J strain that underwent ager QTb17X program (http://mapmgr.roswellpark.org/mapmgr. a mutation that inactivated the tyrosinase gene (c) to make it albino. html/; provided in the public domain by Rosewell Park Cancer Insti- Therefore, the strain is isogenic with C57BL/6J. By convention, the tute, Buffalo, NY). With this program, a likelihood ratio statistic (LRS) abbreviation for C57BL/6J mice is “B6” or “B” and for BALB/cByJ mice was calculated for each of the 86 marker genotypes, with a probability is “C” or “CBy.” All mice were kept under a 12-hour light–dark cyclic inclusion level for further study of 0.05. Thus, any single-point LRS for light cycle with an in-cage illuminance of 2 to 7 ft-c. The temperature a marker that had only a 5% probability or less of occurring by chance of the vivarium was maintained between 18°C and 20°C. Cages were in this set of data was included for further study. Of these markers, the kept on four shelves of free-standing, five-shelf racks (never on the top one with the highest LRS was studied by interval mapping of all the shelf). Each week, the cages were rotated by shelf, by side of the rack markers on its chromosome. The marker at the peak of this QTL was (left or right), and by position on the shelf (seven positions from front put into the background for the next evaluation. Then peak markers to back). Mice were maintained on a low-fat diet (15001 Rodent from both the first and second interval maps were put into the back- Laboratory Chow; Newco Distributors, Rancho Cucamonga, CA) with ground for the next determination, and peak markers from the first, chow and water available ad libitum. second, and third QTLs were put into the background for the next For the quantitative genetic study, a nonreciprocal (BALB/cByJ x determination. However, after the three strong and highly significant C57BL/6J-c2J)F2 cross was made, and 268 F2 progeny were aged to 8 QTL markers were placed in background, the remaining LRSs were all months along with 30 BALB/cByJ, 23 C57BL/6J-c2J and 50 F1 control in the suggestive category and very close to one another. Therefore, mice. Because the mothers of all the F1s were BALB/cByJ, all F1 and F2 each of these was evaluated only with the three markers from the mice had the CBy mitochondrial genome and all F1 and F2 males had highly significant QTLs. To determine significance levels for this ge- the B6 Y chromosome. nome-wide screen, a test of 1000 permutations of all marker genotypes together was performed. For the quantitative trait of the entire retinal Ͻ Quantitative Traits section (the mean of 54 measurements), P 0.001 was 22.9 (highly significant [HS]); P Ͻ 0.05 was 16.0 (significant [S]) and P Ͻ 0.67 was After the mice were aged to 8 months, eyes were enucleated immedi- 9.1 (suggestive or sugg). A highly significant LRS of 22.9 or more had ately after death, fixed in a mixture of 2% formaldehyde and 2.5% a 99.9% probability or more of being real and would only occur by glutaraldehyde in phosphate buffer, embedded in an Epon-Araldite chance in 1 of 1000 genome scans such as this one. For the quantitative mixture, and bisected along the vertical meridian through the optic nerve head. A single 1-␮m section was taken from the cut surface of one of the half-orbs from each mouse and stained with toluidine blue, as described previously.29 On this section, measurements of the thick- ness of the ONL were made. Three measurements, each spaced 50 ␮m apart, were taken at nine 0.25-mm intervals, both in the superior and inferior hemispheres starting from the optic nerve head. The means of 54 measurements from the entire retinal sections were used to score the mice for the quantitative trait. For the purpose of comparison with results in a previous light-damage study,17 the means of 12 measure- ments from the posterior retina in the superior hemisphere were used for the quantitative trait (Fig. 4, areas marked 2–5). All procedures involving the mice adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the guidelines of the Loyola Marymount University Committee on Animal Research.

Genotyping Genotyping services were provided by the Center for Inherited Disease

Research (CIDR; The Johns Hopkins University, Baltimore, MD). For 2J FIGURE 1. Thickness of the ONL of the retina of C57BL/6J-c (F) and each chromosome, the most proximal marker genotyped was within BALB/cByJ (f) mice at various times after aging in dim cyclic light. The 15 cM of the centromere, internal markers were no more than 30 cM number of mice at 6, 8, 10, and 12 months were 6, 5, 5, and 3, apart, and the most distal markers were within 15 cM of the telomere. respectively, for B6 mice, and 6, 5, 3, and 4, respectively, for BALB/c The average spacing was 15.7 cM. The exceptions were Chr 1 where mice.

Downloaded from iovs.arvojournals.org on 09/27/2021 2444 Danciger et al. IOVS, June 2003, Vol. 44, No. 6

TABLE 1. Evaluation and Comparison of Quantitative Trait in Control and five weaker QTLs. Table 2 shows that based on 23 B6, 30 Mice at 8 Months of Age C, and 50 F1 control ONL retinal measurements (103 control animals), MAP Manager QTX calculated that 69% of the ob- Average ONL served age-related effect was due to genetics. Of that 69%, 21%, Thickness Genotype (␮m) n Comparison P* or 30% (21/69) of the total genetic effect was due to a gene acting in a QTL on mid-Chr 6 (Figs. 2a, 2b). The LOD score for BALB/cByJ (C) 39.15 30 C vs. B6 8.6 ϫ 10Ϫ18 this QTL was 13.5 and the B6 allele was dominant. QTLs on C57BL/6J-c2J (B6) 45.26 23 C vs. F1 4.2 ϫ 10Ϫ24 Chrs 10 and 16 were each 13% of the total genetic effect (Figs. F1 47.30 50 F1 vs. B6 NS (0.12) 2c–f; Table 2) with LOD scores of 6.4 and 7.1, respectively. The B6 allele for the Chr 10 QTL was additive with the C allele; * Probabilities (P) based on unpaired Student’s t-test. the B6 allele for the Chr 16 QTL was recessive/additive—that is, the Chr 16 B6 allele influenced the C allele but substantially trait of the posterior superior section (Fig. 4, positions 2–5) HS was less than 50%, or in other words, the B6 allele was recessive but 24.3, S was 15.2, and sugg was 9.2. The LRS was converted to an LOD not completely recessive. Purely additive alleles would have score by dividing by 4.6 (2 ϫ the natural log of 10). influence on one another equal to that of those acting in the Chr 10 QTL. Each of these first three QTLs was highly signifi- cant—that is, the probability that they would occur by chance RESULTS in a genome-wide screen of the size and complexity of this study was less than 0.001). However, although the QTLs of Age-Related Retinal Degeneration QTLs Chrs 6 and 10 reflected B6 protective gene alleles, the Chr 16 We used the thickness of the retinal ONL consisting of the QTL reflected a B6 allele that did not protect the retina from intercalated and packed photoreceptor nuclei as the quantita- age-related retinal degeneration. In this case, it was the C allele tive trait reflecting retinal degeneration. A thinner ONL corre- that was protective. sponds to fewer photoreceptor nuclei and a greater loss of The five weaker QTLs were all suggestive, with LOD scores photoreceptor cells. To determine the best age at which to ranging from 2.1 to 2.6, and equivalent to only 4% of the measure the quantitative trait, we measured the ONL in control genetic effect each. In four of these QTLs, the B6 alleles were BALB/c and B6 mice at ages 6, 8, 10, and 12 months (Fig. 1). protective; in the fifth on Chr 12, the C allele was protective The age of 8 months was selected for study because this was (Table 2). A suggestive QTL (P Ͻ 0.67) means that for every the earliest age showing a significant difference. With evalua- three genome scan studies similar to this one, two of these tion of additional 8-month-old control animals, it was observed (suggestive) associations will occur by chance. Therefore, in- that the average ONL thickness was not significantly different dependent confirmation is needed to verify these suggestive between B6 and F1 retinas, whereas both were significantly QTLs. thicker than those of BALB/c (Table 1). This suggested a net To determine whether any genes were acting together to autosomal dominant relationship between the aging retinal influence age-related retinal degeneration in a significant, syn- degeneration-influencing alleles of the two strains. For the ergistic way, we used the interaction function of Map Manager study, F1 mice from the B6 and C strains were intercrossed to QTX. For an intercross, this function tests every marker as an produce 268 F2 progeny (536 meioses) that were aged to 8 additive and dominant allele against every other marker as months in dim cyclic light. additive and dominant (four interactions per pair of markers). Using the Map Manager QT17X program30 to analyze the The interaction LRS (IX) needed for significance is approxi- ONL thickness data of the F2 progeny versus the genotype data mately 20 (LOD score of 4.35) for an intercross.31 When this from a genome-wide scan performed with dinucleotide repeat function was performed with an exclusion probability of 10Ϫ5 markers, we found three highly significant and strong QTLs or less (as the program recommends), only one interaction was

TABLE 2. QTLs from Age-Related Retinal Degeneration Intercross Study

Marker(s) at cM† from Mb‡ from LOD % Effect Best Fitting Sig* Peak of QTL Centromere Centromere Score (% Total Genetic Effect)§ Inheritance Model

HS 103 controls ——26.1 69 (100) Dominant HS D6Mit209 33 76.4 13.5 21 (30) Dominant D6Mit284 37 93.4 HS D10Mit213 11 20.3 6.4 9 (13) Additive HS D16Mit139 43 66.1 7.1 Ϫ9(Ϫ13)࿣ Add/recess¶ D16Mit189 55 83.2 Sugg D14Mit126 5 17.0 2.7 3 (4) Recessive Sugg D18Mit208 38 61.2 2.6 3 (4) Recessive Sugg D12Mit60 16 28.9 2.6 Ϫ3(Ϫ4) Add/dom D12Mit236 22 39.5 Sugg D13Mit19 24 43.4 2.5 3 (4) Additive Sugg D8Mit47 53 106.7 2.1 3 (4) Add/dom D8Mit49 67 124.1

Analysis with Map Manager QTX 17b. Quantitative trait based on the average of all ONL thickness measurements. * HS, highly significant; S, significant; Sugg, suggestive. † cM positions are the distances from the centromere taken from linkage maps of the Encyclopedia of the Mouse Genome at the MGI web site. ‡ Mb positions were determined from the MGSCv3 sequence database and rounded to the nearest 0.1 million bases from the centromere. § The % total genetic effect ϭ % effect for this locus divided by total % effect of controls. ࿣ The “Ϫ” % genetic effect score indicates a C-protective allele; all other % genetic effect scores indicate B6-protective alleles. ¶ When two models are noted, both have similar % effects and similar LOD scores.

Downloaded from iovs.arvojournals.org on 09/27/2021 IOVS, June 2003, Vol. 44, No. 6 QTL Protective against Age-Related Retinal Degeneration 2445

The X and Y Chromosomes There was a gender-related effect on age-related retinal degen- eration. The average ONL thickness in 25 F1 control males was significantly greater than that in 25 F1 females (49.42 ␮m vs. 45.17 ␮m; unpaired Student’s t-test P ϭ 3.21 ϫ 10Ϫ10), and the average ONL thickness in 148 F2 males was significantly greater than that in 120 F2 females (45.17 ␮m vs. 43.41 ␮m; P ϭ 0.0016). Because the test intercross was nonreciprocal, all the mothers of the F1s were BALB/cByJ mice. For this reason, and for the reason that the Map Manager QT program does not distinguish between hemizygous male and homozygous female genotypes for the interval mapping function in intercrosses, we evaluated the influence of the loci on the X chromosome by other means. For each of the four X chromosome markers, we calculated the average ONL thickness for F2 males hemizygous for the C allele, F2 males hemizygous for the B6 allele (B), F2 females homozygous for the C allele, and heterozygous F2 fe- males (because of the breeding plan, there were no F2 females homozygous for X chromosome B6 alleles). Figure 3a shows the average ONLs of the F2 progeny with the various geno- types. The males consistently had ONLs thicker than those in females, and males hemizygous for C consistently had ONLs thicker than did males hemizygous for B. To evaluate the

FIGURE 2. (a, c, e) Interval maps of the LRS (converted to LOD scores) produced by the Map Manager QTX program of the three strong and highly significant quantitative trait loci (QTL). Dashed horizontal lines represent a 2-LOD distance from the peak of the QTL. The vertical projections from the points where the 2-LOD line crosses the graph of the LOD scores conservatively estimate the 95% CI. Each hash mark on the x-axis is 1 cM. (b, d, f) Histograms of the percentage of the total genetic effect compared with the model of inheritance. In addition, the peak markers from this age-related study are shown for the percentage of genetic effect they had in a constant light-induced retinal degener- ation study performed previously with the same two strains of mice.17 FIGURE 3. Comparison of average ONL thickness scores of F2 prog- eny of the same genotype for dinucleotide repeat markers on the X chromosome. MC, males hemizygous for the BALB/cByJ allele; MB, found, and it was of questionable significance. The interaction males hemizygous for the C57BL/6J-c2J allele; FC, females homozygous involved the markers D9Mit355 (mid distal Chr 9) and for the BALB/cByJ allele; CB, heterozygous females. (a) Histogram of D18Mit208 (middle Chr 18). The IX was 19.9 (LOD score ϭ average ONL thickness scores by genotype. (b) Plots of the negative 4.33). Otherwise, there were no other significant interactive logarithm of the probability of significant difference determined by the unpaired Student’s t-test between ONL thickness scores in males and gene effects, as represented by IX scores of less than 20 for all females of the same genotype for all four X chromosome markers. (f) marker (chromosomal loci) interactions of the genome-wide MC versus all females; (Ⅺ) MB versus all females; (Œ) MC versus MB scan. (males); (F) FC versus CB (females).

Downloaded from iovs.arvojournals.org on 09/27/2021 2446 Danciger et al. IOVS, June 2003, Vol. 44, No. 6

significance of this, we performed t-tests comparing average However, in the current aging study, we measured the ONL in ONL thickness in the two male and two female F2 genotypes at 18 positions across the entire length of the vertical retinal each marker. section. For the purpose of comparison, we recalculated the Figure 3b shows no significant difference between the fe- ONL thickness of the 268 F2 progeny of this aging study in only males homozygous for C and those heterozygous at any of the positions 2 to 5 in the superior posterior retina. We then four markers. Therefore, we compared all females with males calculated QTLs with this quantitative trait (Table 3). hemizygous for B and with males hemizygous for C, because F2 The same three QTLs that were strong and highly significant males on average had thicker ONLs than did F2 females. Males in the entire retinal section were strong and highly significant hemizygous for B was barely (at DXMit68) or nearly signifi- in just the superior, posterior hemisphere. Among the weaker, cantly different from F2 females (at the other three markers) suggestive QTLs, the Chr 8 locus was no longer significant suggesting that the B6 Y chromosome provides a small mea- when calculations were based on just the superior, posterior sure of protection against age-related retinal degeneration. This hemisphere, but a locus at mid-Chr 9 became suggestive. An was supported by the fact that 12 male 8-month-old B6 control additional locus on distal Chr 14 also became tentatively sug- mice had on average significantly thicker ONLs than did 11 B6 gestive. It was just a bit below the LRS cutoff for a suggestive 8-month-old female control animals (47.3 ␮m vs. 45.02 ␮m; QTL (9.1 vs. 9.2). The other four pairs of loci present when P ϭ 0.029). There was no significant difference in ONL thick- either form of the quantitative trait was used (including the Chr ness between 12 male and 18 female 8-month BALB/c control 12 QTL), were similar in strength (Tables 2, 3). animals (38.93 ␮m vs. 39.30 ␮m; P ϭ 0.463). The three strong age-related retinal degeneration QTLs had The males hemizygous for C were significantly different either no influence or very little influence on the genetics of from all females, with a peak at the marker DXMit216. They constant bright light-induced retinal degeneration. The QTL on were also significantly different from the males hemizygous for Chr 6 could not be assessed for its influence on light-induced B, but only at DXMit216 (Fig. 3b). This suggests that the retinal damage because of the possibility that the B6 allele of BALB/c X chromosome carries a gene near the DXMit216 locus the gene in this QTL was dominant in the previous study as it that contributes to protection against age-related retinal degen- was in the current age-related retinal degeneration study. A eration, but only when the B6 Y chromosome is present. If the QTL with a dominant B6 gene allele would have been hidden BALB/c X chromosome locus at DXMit216 could provide pro- in the genetic backcross of the light-damage study. Neverthe- tection without the B6 Y chromosome, the female F2 mice less, age-related retinal degeneration is significantly influenced homozygous for the X chromosome C allele at that locus would by at least two genes that have little or no influence on the type show average ONLs similar to those in males hemizygous for of constant bright-light–induced retinal degeneration we stud- the same allele. As shown in Figure 3, this was not the case. ied previously.17 In addition, the RPE65-MET/LEU450 variant on distal Chr 3, which accounted for nearly 50% of the genetic DISCUSSION response influencing light-induced retinal degeneration, had no influence on age-related retinal degeneration. Thus, substan- In a genetic study of the BALB/c and C57BL/6J-c2J albino strains tial genetic portions of these two causes of retinal degenera- of mice, we have identified three highly significant and strong tion are distinct from each another. QTLs that influence age-related retinal degeneration. Two of This is not to say that there is no overlap. There were six these loci on Chrs 6 and 10 represent genes with B6 alleles that suggestive QTLs in the aging study that were calculated using are protective, whereas the third on Chr 16 has a gene with a the average ONL thickness of the posterior superior retina. C allele that is protective. Three of these suggestive QTLs were in common with QTLs Because age has been a consideration in retinal light-in- from the light-damage study. The two QTLs on Chrs 9 and 12 duced damage in animal models,32–34 we compared the results that represented C alleles that protect the retina from light- of the present study with results from a previous light-damage induced retinal damage also protect the retina from age-related study in these same two mouse strains. The quantitative trait retinal degeneration. The one QTL on Chr 14 that was protec- used in the earlier light-damage study17 was based on ONL tive against light for B6 was also protective against aging. They thickness measurements taken from the more light-sensitive were also similar in strength: each of the three QTLs accounted superior, posterior retinal hemisphere (Fig. 4, positions 2–5). for only a few percentage points of the genetic effect in both

FIGURE 4. Average ONL thickness scores in retinas of control mice at each of the 18 measuring positions (3 measurements per position) on the retinal section. (Ⅺ) C57BL/6J-c2J, n ϭ 23; (‚) BALB/cByJ, n ϭ 30; (F) F1, n ϭ 50. The numbers (2–5) indi- cate the posterior superior region where the ONL thickness average was recalculated for the purpose of comparison with a previous light- damage study.

Downloaded from iovs.arvojournals.org on 09/27/2021 IOVS, June 2003, Vol. 44, No. 6 QTL Protective against Age-Related Retinal Degeneration 2447

TABLE 3. Calculation of Age-Related QTLs Based on Average Thickness of ONL 2–5 in the Superior, Posterior Retina

Marker(s) at LOD cM from Mb from % Effect Sig. QTL Peak of QTL Score Centromere Centromere (% Total Genetic Effect)

HS 103 controls — 25.3 —— 72 (100) HS Chr 6 D6Mit209 9.1 33 76.4 14 (19) D6Mit284 37 93.4 HS Chr 10 D10Mit213 8.0 11 20.3 12 (17) HS Chr 16 D16Mit139 7.5 43 66.1 Ϫ12 (Ϫ17) D16Mit189 55 83.2 Sugg Prox. 14 D14Mit126 2.6 5 17.0 3 (4) Sugg Chr 18 D18Mit208 2.1 38 61.2 3 (4) Sugg Chr 12 D12Mit60 2.3 16 28.9 Ϫ3(Ϫ4) D12Mit236 22 39.5 Sugg Chr 13 D13Mit19 3.0 24 43.4 4 (6) NS Chr 8 D8Mit47 — 53 106.7 — D8Mit49 67 124.1 Sugg Chr 9 D9Mit263 2.1 40 76.3 Ϫ3(Ϫ4) Sugg Dist. 14 D14Mit165 2.0* 52 97.7 3 (4)

* The LRS statistic for the suggestive category was 9.2 (LOD ϭ 2.0); the LRS for this marker was 9.1 (LOD ϭ 1.98). See Table 2 for descriptions of data. Sig., significance; Prox., proximal; Dist., distal.

studies. The presence of these three QTLs in two separate fashion, a locus was identified at 1q25-q31. More recently, in a studies and with similar (but small) effects in each study veri- large genome-wide scan study of a cohort of 391 families with fies them and shows that at least a small part of age-related a minimum of two individuals with AMD, three categories of retinal degeneration may be influenced by the same genes that disease were established to make the genetics more precise. influence light-induced retinal degeneration. The result was the identification of four potential AMD loci In humans, the question of sunlight exposure as an envi- with LOD scores between 2.0 and 3.16.15 The loci were 1q31 ronmental factor in AMD has been investigated extensively, but (matching the family study), 17q25, 9p13, and 10q26. The no clear answer has emerged.6–9,35–39 The genetic factors that ABCA4 gene, which causes several types of autosomal reces- influenced light-induced retinal damage in our studies ac- sive retinal degeneration including Stargardt’s macular dystro- counted for a small percentage of the genetic effect in age- phy, has been implicated in AMD. In this case, individuals with related retinal degeneration (not including the unknown influ- damaging mutations in only one allele are thought to be sus- ence on light-induced damage of the B6 dominant Chr 6 QTL). ceptible to the disease. However, the evidence is controversial Perhaps, in the same way, sunlight exposure has only a small and, if true, would influence only a small percentage of AMD influence on the genetic predisposition to AMD, although in- cases.40–48 A second gene implicated in AMD is APOE. The ⑀4 dividuals with different alleles at the relevant loci might differ allele of this gene has been associated with a small protective from each another. effect against AMD.49–52 Ctsdtm1Cptr, a disrupted allele of the There was no gender difference in the amount of retinal mouse gene that expresses cathepsin D, produces progressive damage present in BALB/c retinas compared with B6 retinas age-related changes in the mouse retina similar to those of after constant light exposure.17 However, there was in age- AMD when homozygous.53 Human chromosomal regions ho- related retinal degeneration. Because the cross we produced mologous to the three highly significant QTLs are 2p and 3p was nonreciprocal, all mothers of the F1 mice were BALB/c. (Chr 6 QTL), 6q (Chr 10 QTL), and 3q and 21q (Chr 16 QTL). This allowed us to see a difference in age-related retinal degen- None of the QTLs found in our study are in mouse loci homol- eration between F1 males (all hemizygous for the BALB/c X ogous to the AMD-associated human loci or the loci of the chromosome and all carrying the B6 Y chromosome) and F1 AMD-associated genes cited herein. Therefore, these QTLs rep- females. The gender difference carried through to the F2 gen- resent loci of genes that previously, were not known to influ- eration as well. Based on the results of F2 progeny with differ- ence age-related retinal degeneration and may serve as candi- ent X chromosome genotypes, we hypothesized that a gene in dates for study in AMD once identified. These same genes may a region of the BALB/c X chromosome near the marker modify the monogenic inherited retinal degenerations as well. DXMit216 conferred some resistance to age-related retinal de- The mouse age-related retinal degeneration QTLs on Chrs generation, but only in the presence of the B6 Y chromosome. 10 and 16 cover broad regions and come with large 95% This was deduced from the following: (1) the ONL of retinas confidence intervals (CI) that include many genes. Based on from male B6 mice were thicker than those of B6 females; (2) the 2-LOD support intervals shown in Figures 2c and 2e and there was no difference in the ONLs between male and female the MSGCv3 mouse genome sequence map (www.ncbi.nlm. BALB/c mice; (3) F2 males hemizygous for the C allele of nih.gov/genome/sequence/ provided in the public domain by DXMit216 were protected from age-related retinal degenera- the National Center for Biotechnology Information, Bethesda, tion compared with F2 females homozygous C for the same MD), we estimate the distances spanning the QTLs at 30 Mb or marker; (4) F2 males hemizygous for the DXMit216 C allele more for both the Chr 10 and Chr 16 QTLs. The QTL on were protected from age-related retinal degeneration com- mid-Chr 6 has a 2-LOD support interval of approximately 10 cM pared with F2 males hemizygous for the B6 allele of DXMit216; (Fig. 2a). Using the bootstrap analysis function from Map Man- (5) F2 males hemizygous for the B6 allele for any of the four ager QTX (not shown), the 95% CI lies between D6Mit209 and markers genotyped on the X chromosome appeared to be D6Mit284 (ϳ8.5 cM in our cross), which are placed at 76.4 and protected from age-related retinal degeneration compared with 93.4 Mb, respectively on the MSGCv3 map, a distance of 17 F2 females. Further aging studies using F1s from B6 mothers Mb. There are well over 200 genes in this region, and many are are needed to test this hypothesis. expressed in retina. Evaluating the many good candidate genes In a family study by Klein et al.,16 with 10 individuals from in this Chr 6 QTL region will be assisted by refinement of the two generations with AMD inherited in an autosomal dominant locus. To do this, additional studies must be performed with

Downloaded from iovs.arvojournals.org on 09/27/2021 2448 Danciger et al. IOVS, June 2003, Vol. 44, No. 6

additional F2 progeny from the same intercross, and/or with 19. Pittler SJ, Baehr W. Identification of a nonsense mutation in the rod recombinant inbred C57 x BALB/c (CxB or BxC) strains with photoreceptor cGMP phosphodiesterase beta-subunit gene of the “mosaicized” chromosomes, and/or with similar QTLs from rd mouse. Proc Natl Acad Sci USA. 1991;88:8322–8326. other crosses that may overlap. This task is made much easier 20. Travis GH, Brennan MB, Danielson PE, Kozak CA, Sutcliffe JG. by the fact that we are starting with a relatively narrow region Identification of a photoreceptor-specific mRNA encoded by the (for a QTL) because of the strong influence of this Chr 6 gene. gene responsible for retinal degeneration slow (rds). Nature. Discovery of this and the other QTL genes and their protective 1989;338:70–73. alleles may open avenues of study that contribute to the de- 21. Akhmedov NB, Piriev NI, Chang B, et al. A deletion in a photore- ceptor-specific nuclear receptor mRNA causes retinal degenera- velopment of gene or pharmaceutical therapies for age-related tion in the rd7 mouse. Proc Natl Acad Sci USA. 2000;97:5551– and other retinal degenerations. 5556. 22. Haider NB, Naggert JK, Nishina PM. Excess cone cell proliferation Acknowledgments due to lack of a functional NR2E3 causes retinal dysplasia and degeneration in rd7/rd7 mice. Hum Mol Genet. 2001;10:1619– The authors thank Ken Manly for his very generous assistance in 1626. interpretation of Map Manager QTX analyses and Joseph Jabbra for 23. McLaughlin ME, Ehrhart TL, Berson EL, Dryja TP. Mutation spec- helping to get the Loyola Marymount vivarium outfitted. trum of the gene encoding the beta subunit of rod phosphodies- terase among patients with autosomal recessive retinitis pigmen- tosa. Proc Natl Acad Sci USA. 1995;92:3249–3253. References 24. Danciger M, Blaney J, Gao YQ, et al. Mutations in the PDE6B gene in autosomal recessive . Genomics. 1995;30: 1. Kahn HA, Leibowitz HM, Ganley JP, et al. The Framingham Eye 1–7. Study. I. Outline and major prevalence findings. Am J Epidemiol. 25. Farrar GJ, Kenna P, Jordan SA, et al. A three-base-pair deletion in 1977;106:17–32. the peripherin-RDS gene in one form of retinitis pigmentosa. 2. Klein BE, Klein R. Cataracts and macular degeneration in older Nature. 1991;354:478–480. Americans. Arch Ophthalmol. 1982;100:571–573. 26. Kajiwara K, Hahn LB, Mukai S, Travis GH, Berson EL, Dryja TP. 3. Klaver CC, Wolfs RC, Vingerling JR, Hofman A, de Jong PT. Age- Mutations in the human retinal degeneration slow gene in autoso- specific prevalence and causes of blindness and mal dominant retinitis pigmentosa. Nature. 1991;354:480–483. in an older population: the Rotterdam Study. Arch Ophthalmol. 27. Haider NB, Jacobson SG, Cideciyan AV, et al. Mutation of a nuclear 1998;116:653–658. receptor gene, NR2E3, causes enhanced S cone syndrome, a dis- 4. Hakkinen L. Vision in the elderly and its use in the social environ- order of retinal cell fate. Nat Genet. 2000;24:127–131. ment. Scand J Soc Med Suppl. 1984;35:5–60. 28. Gerber S, Rozet JM, Takezawa SI, et al. The - 5. Martinez GS, Campbell AJ, Reinken J, Allan BC. Prevalence of specific nuclear receptor gene (PNR) accounts for retinitis pig- ocular disease in a population study of subjects 65 years old and mentosa in the Crypto-Jews from Portugal (Marranos), survivors older. Am J Ophthalmol. 1982;94:181–189. from the Spanish Inquisition. Hum Genet. 2000;107:276–284. 6. Hawkins BS, Bird A, Klein R, West SK. Epidemiology of age-related 29. LaVail MM, Gorrin GM, Repaci MA, Thomas LA, Ginsberg HM. macular degeneration. Mol Vis. 1999;5:26. Genetic regulation of light damage to photoreceptors. Invest Oph- 7. Hyman L, Neborsky R. Risk factors for age-related macular thalmol Vis Sci. 1987;28:1043–1048. degeneration: an update. Curr Opin Ophthalmol. 2002;13:171– 30. Manly KF, Cudmore RH Jr, Meer JM. Map Manager QTX, cross- 175. platform software for genetic mapping. Mamm Genome. 2001;12: 8. McCarty CA, Mukesh BN, Fu CL, Mitchell P, Wang JJ, Taylor HR. 930–932. Risk factors for age-related maculopathy: the Visual Impairment Project. Arch Ophthalmol. 2001;119:1455–1462. 31. Kruglyak L, Lander ES. A nonparametric approach for mapping quantitative trait loci. Genetics. 1995;139:1421–1428. 9. Klein R, Klein BE, Moss SE. Relation of smoking to the incidence of age-related maculopathy: The Beaver Dam Eye Study. Am J Epide- 32. LaVail MM, Kumar NN, Gorrin GM, Yasumura D, Matthes MT. Age miol. 1998;147:103–110. and mononuclear enucleation as potential determinants of light damage in the mouse retina. In: Hollyfield JG, Anderson RE, LaVail 10. Klein ML, Mauldin WM, Stoumbos VD. Heredity and age-related macular degeneration: observations in monozygotic twins. Arch MM, eds. Retinal Degenerative Diseases and Experimental Ther- Ophthalmol. 1994;112:932–937. apy. New York: Plenum Press; 1999:317–324. 11. Meyers SM, Greene T, Gutman FA. A twin study of age-related 33. Lai YL, Jacoby RO, Jonas AM. Age-related and light-associated macular degeneration. Am J Ophthalmol. 1995;120:757–766. retinal changes in Fischer rats. Invest Ophthalmol Vis Sci. 1978; 17:634–638. 12. Gottfredsdottir MS, Sverrisson T, Musch DC, Stefansson E. Age related macular degeneration in monozygotic twins and their 34. Weisse I, Stotzer H, Seitz R. Age- and light-dependent changes in spouses in Iceland. Acta Ophthalmol Scand. 1999;77:422–425. the rat eye. Virchows Arch A Pathol Anat Histol. 1974;362:145– 13. Klaver CC, Wolfs RC, Assink JJ, van Duijn CM, Hofman A, de Jong 156. PT. Genetic risk of age-related maculopathy: population-based 35. Mitchell P, Smith W, Wang JJ. Iris color, skin sun sensitivity, and familial aggregation study. Arch Ophthalmol. 1998;116:1646– age-related maculopathy: The Blue Mountains Eye Study. Ophthal- 1651. mology. 1998;105:1359–1363. 14. Gorin MB, Breitner JC, De Jong PT, et al. The genetics of age- 36. Delcourt C, Carriere I, Ponton-Sanchez A, Fourrey S, Lacroux A, related macular degeneration. Mol Vis. 1999;5:29. Papoz L. Light exposure and the risk of age-related macular 15. Weeks DE, Conley YP, Tsai HJ, et al. Age-related maculopathy: an degeneration: the Pathologies Oculaires Liees a l’Age (POLA) expanded genome-wide scan with evidence of susceptibility loci study. Arch Ophthalmol. 2001;119:1463–1468. within the 1q31 and 17q25 regions. Am J Ophthalmol. 2001;132: 37. Frank RN, Puklin JE, Stock C, Canter LA. Race, iris color, and 682–692. age-related macular degeneration. Trans Am Ophthalmol Soc. 16. Klein ML, Schultz DW, Edwards A, et al. Age-related macular 2999;98:109–115; discussion 115–117. degeneration; clinical features in a large family and linkage to 38. Taylor HR, West S, Munoz B, Rosenthal FS, Bressler SB, Bressler chromosome 1q. Arch Ophthalmol. 1998;116:1082–1088. NM. The long-term effects of visible light on the eye. Arch Oph- 17. Danciger M, Matthes MT, Yasamura D, et al. A QTL on distal thalmol. 1992;110:99–104. chromosome 3 that influences the severity of light-induced dam- 39. Harlan JB, Weidenthal DT, Green WR. Histologic study of a age to mouse photoreceptors. Mamm Genome. 2000;11:422–427. shielded macula. Retina. 1997;17:232–238. 18. Bowes C, Danciger M, Kozak CA, Farber DB. Isolation of a candi- 40. De La Paz MA, Guy VK, Abou-Donia S, et al. Analysis of the date cDNA for the gene causing retinal degeneration in the rd Stargardt disease gene (ABCR) in age-related macular degenera- mouse. Proc Natl Acad Sci USA. 1989;86:9722–9726. tion. Ophthalmology. 1999;106:1531–1536.

Downloaded from iovs.arvojournals.org on 09/27/2021 IOVS, June 2003, Vol. 44, No. 6 QTL Protective against Age-Related Retinal Degeneration 2449

41. Souied EH, Ducroq D, Gerber S, et al. Age-related macular degen- 47. Webster AR, Heon E, Lotery AJ, et al. An analysis of allelic variation eration in grandparents of patients with Stargardt disease: genetic in the ABCA4 gene. Invest Ophthalmol Vis Sci. 2001;42:1179– study. Am J Ophthalmol. 1999;128:173–178. 1189. 42. Stone EM, Webster AR, Vandenburgh K, et al. Allelic variation in 48. Rivera A, White K, Stohr H, et al. A comprehensive survey of ABCR associated with Stargardt disease but not age-related macular sequence variation in the ABCA4 (ABCR) gene in Stargardt disease degeneration. Nat Genet. 1998;20:328–329. and age-related macular degeneration. Am J Hum Genet. 2000;67: 43. Zhang K, Kniazeva M, Hutchinson A, Han M, Dean M, Allikmets R. 800–813. The ABCR gene in recessive and dominant Stargardt diseases: a 49. Simonelli F, Margaglione M, Testa F, et al. Apolipoprotein E poly- genetic pathway in macular degeneration. Genomics. 1999;60: morphisms in age-related macular degeneration in an Italian pop- ulation. Ophthalmic Res. 2001;33:325–328. 234–237. 50. Schmidt S, Saunders AM, De La Paz MA, et al. Association of the 44. Bernstein PS, Leppert M, Singh N, et al. Genotype-phenotype apolipoprotein E gene with age-related macular degeneration: pos- analysis of ABCR variants in macular degeneration probands and sible effect modification by family history, age, and gender. Mol siblings. Invest Ophthalmol Vis Sci. 2002;43:466–473. Vis. 2000;6:287–293. 45. Shroyer NF, Lewis RA, Yatsenko AN, Wensel TG, Lupski JR. 51. Klaver CC, Kliffen M, van Duijn CM, et al. Genetic association of Cosegregation and functional analysis of mutant ABCR (ABCA4) apolipoprotein E with age-related macular degeneration. Am J alleles in families that manifest both Stargardt disease and age- Hum Genet. 1998;63:200–206. related macular degeneration. Hum Mol Genet. 2001;10:2671– 52. Souied EH, Ducroq D, Rozet JM, et al. ABCR gene analysis in 2678. familial exudative age-related macular degeneration. Invest Oph- 46. Guymer RH, Heon E, Lotery AJ, et al. Variation of codons 1961 and thalmol Vis Sci. 2000;41:244–247. 2177 of the Stargardt disease gene is not associated with age- 53. Rakoczy PE, Zhang D, Robertson T, et al. Progressive age-related related macular degeneration. Arch Ophthalmol. 2001;119:745– changes similar to age-related macular degeneration in a transgenic 751. mouse model. Am J Pathol. 2002;161:1515–1524.

Downloaded from iovs.arvojournals.org on 09/27/2021