Vidyullatha Vasireddy,1 Monica M. Jablonski,2 Md Nawajes A. Mandal,1 Dorit Raz-Prag,3,4
Xiaofei F. Wang,2 Lesli Nizol,1 Alessandro Iannaccone,2 David C. Musch,1
Ronald A. Bush,3,4 Norman Salem, Jr,5 Paul A. Sieving,3,4 and Radha Ayyagari1
PURPOSE. To develop and characterize a heterozygous knock-inmouse model carrying the 5-bp deletion in Elovl4 (E_mut�/�)and to study the pathology underlying Stargardt-like maculardegeneration (STGD3).
METHODS. E_mut�/� mice were generated by targeting a 5-bpdeletion (AACTT) in the Elovl4 gene by homologous recombi-nation. E_mut�/� mice of age 2 to 18 months and age-matchedwild-type (Wt) littermate control animals were analyzed for theexpression of Elovl4 transcript, ELOVL4 protein, photorecep-tor-specific genes, and retinal fatty acid composition. Func-tional retinal changes were evaluated by electroretinography(ERG) and by morphologic and ultrastructural criteria.
RESULTS. E_mut�/� mice retinas showed the presence of bothWt and mutant Elovl4 transcripts and proteins. Morphologicevaluation revealed cone photoreceptor ultrastructural abnor-malities as early as 2 months of age, accumulation of lipofuscinin retinal pigment epithelium (RPE), and subretinal deposits atlater ages. Shortening of rod outer segments (OS) was observedat �10 months of age. Both cone and rod changes progressedwith age. Unlike rod-specific genes, expression of selectedcone specific genes was significantly reduced by 7 months ofage. Mixed rod–cone and light-adapted b-waves were higherthan normal at both 8 and 15 months. Levels of the fatty acids20:5 (P � 0.027), 22:5 (P � 0.040) and 24:6 (P � 0.005) were
found to be significantly lower in the retinas of E_mut�/� micethan in retinas of control subjects.
CONCLUSIONS. E_mut�/� animals display characteristic featuresassociated with Stargardt-like macular degeneration and serveas a model for the study of the mechanism underlying STGD3.(Invest Ophthalmol Vis Sci. 2006;47:4558–4568) DOI:10.1167/iovs.06-0353
Mutations in the gene elongation of very long-chain fattyacids-4 (ELOVL4) are implicated in human autosomal
dominant Stargardt-like macular degeneration (STGD3), whichis characterized by progressive loss of central vision, accumu-lation of retinal flecks, and window defects in the macula. Theage of onset of STGD3 is typically the teenage years, and thedisease is progressive.1,2
ELOVL4 is homologous to the ELO group of mammalian andyeast enzymes involved in the fatty acid chain elongation sys-tem.3 Like other ELO family members, the ELOVL4 proteincontains a distinct iron binding motif, a histidine cluster(HXXHH), and an endoplasmic reticulum (ER) retention signalat its C-terminal end (KXKXX). The highest amount of Elovl4expression is detected in the retina, and low levels of thistranscript were also detected in brain, skin, lens, and testes.4,5
In retinal tissue, the protein is localized to the ER of thephotoreceptor cells.6 All the mutations detected so far in thisgene result in premature truncation of the protein and conse-quent loss of ER retention signal.4,7,8 Heterologous expressionof mutant ELOVL4 in COS-7 cells demonstrated that the mutantproteins were misrouted and that they recruit the wild-typeprotein in the formation of aggresomes, a finding most likelyconsistent with a dominant negative effect.6,9
To assess the in vivo consequences of the 5-bp–deletionmutation (AACTT), we generated a heterozygous knock-in (KI)mouse model bearing the mutation in the Elovl4 gene(E_mut�/�) associated with human STGD3.2,10 In addition tocarrying one copy each of the mutant and wild-type Elovl4alleles, the expression of Elovl4 in these mice is controlled bythe native promoter. Herein, we report our observations on theocular phenotype of the E_mut�/� mice. We examined theeyes of these mice for photoreceptor degeneration, retinalpigment epithelial (RPE) changes, accumulation of lipofuscin,expression of photoreceptor-specific genes, electrophysiolog-ical response of the retina, and composition of total retinal fattyacids. The mice with a genotype homologous to STGD3 pa-tients showed predominant cone photoreceptor degenerationclosely mimicking the features observed in patients withSTGD3. Therefore, these mice can serve as a unique andbiologically relevant model for the study of the pathologicconsequences of the Elovl4 5-bp–deletion mutation.
MATERIALS AND METHODS
Reagents and Antibodies Used for the Study
Antibodies used in the study were as follows: anti-ELOVL4 antibody(1:500 dilution; Abcam, Cambridge, MA,); polyclonal anti-ELOVL4 af-
From the 1Ophthalmology and Visual Sciences, W. K. Kellogg EyeCenter, University of Michigan, Ann Arbor, Michigan; the 2Ophthal-mology, Hamilton Eye Institute, University of Tennessee, Memphis,Tennessee; the 3National Institute on Deafness and Other Communi-cation Disorders (NIDCD) and the 4National Eye Institute (NEI), Be-thesda, Maryland; and the 5Laboratory of Membrane Biochemistry andBiophysics, National Institute on Alcohol Abuse and Alcoholism(NIAAA), Rockville, Maryland.
Supported by National Eye Institute Grant EY13198 (RA); TheFoundation Fighting Blindness (RA); a research grant (RA) and anunrestricted grant to the Department of Ophthalmology, University ofTennessee (UTHSC) from Research to Prevent Blindness (RPB), Inc.; aNational Eye Institute Core Grant to UTHSC (P30EY13080); the Her-bert and Mary Shainberg Neuroscience Research Program (AI); anunrestricted grant from RPB to Ophthalmology at UTHSC, two NationalEye Institute Core Grants to the University of Michigan, Department ofOphthalmology (P30EY007003) and Vision Research (P30EY07060);and the Intramural Research Program of the National Institutes ofHealth, NIAAA, NIDCD/NEI.
Submitted for publication March 30, 2006; revised May 16, 2006;accepted August 21, 2006.
Disclosure: V. Vasireddy, None; M.M. Jablonski, None; M.N.A.Mandal, None; D. Raz-Prag, None; X.F. Wang, None; N. Lesli, None;A. Iannaccone, None; D.C. Musch, None; R.A. Bush, None; N.Salem Jr, None; P.A. Sieving, None; R. Ayyagari, None
The publication costs of this article were defrayed in part by pagecharge payment. This article must therefore be marked “advertise-ment” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Corresponding author: Radha Ayyagari, Department of Ophthal-mology and Visual Sciences, #325, W.K. Kellogg Eye Center, 1000 WallStreet, Ann Arbor, MI 48105; [email protected].
finity-purified antibody (1:50 dilution, which is described elsewhere4);monoclonal anti-PDI antibody (1:5 dilution; Calbiochem, La Jolla, CA),anti-�-actin antibody (1:500 dilution; Sigma-Aldrich, St. Louis, MO),anti-S-opsin antibody (1:200 dilution; Chemicon, Temecula, CA), an-ti-M- opsin antibody (1:200 dilution; Chemicon); anti-rabbit Alexa Fluor555 (1:2500 dilution), anti-mouse Alexa Fluor 488 (1:400 dilution), andanti-rabbit Alexa Fluor 594 (1:400 dilution; all from Invitrogen-Molec-ular Probes, Carlsbad, CA).
For RT-PCR, a DNA synthesis system (Super Script First-Strand;Invitrogen) was used. For fatty acid transesterification, 14% (wt/vol)BF3 was prepared in methanol.
Construct Design and Development ofHeterozygous Elovl4 5-bp Deletion MutantKnock-in Mice
To construct a targeting vector, an 11.5-kb mouse genomic DNAfragment containing exons 3 to 6 of Elovl4 was cloned from the129/SV EV BAC genomic library. This fragment contains the sequencebetween 280 bp upstream of exon 3 and 1.4 kb downstream of exon6. The 5-bp deletion–mutation in exon 6 was generated on the longarm using site-directed mutagenesis. A selectable cassette containingthe Neomycin-resistant gene [Neo], flanked by two lox p sites wasinserted into the KpnI site (Fig. 1). The targeting vector was confirmedby restriction digestion and sequencing, using the forward (TGCGAG-GCCAGAGGCCAGTTGTGTAGC) and reverse (ATGTGTCA GTTTCATAGCCTGAAG) primers which can read the Neo gene cassette.
The targeting vector was linearized by NotI, and electroporatedinto 129/SV EV embryonic stem (ES) cells, and subjected to selectionwith G418 antibiotic and ganciclovir. PCR and Southern blot analyseswere used to identify recombinant clones. Correctly targeted ES cellswere injected into C57BL/6J blastocysts and implanted into pseudo-pregnant mice, to generate chimeras. These chimeras were mated withC57BL/6J mice to transmit the targeting allele. Mice carrying the 5-bpdeletion were identified by amplification of the tail DNA followed bysequencing (Fig. 1). The E_mut�/� mice were generated with assis-tance from Ingenious Targeting Laboratory, Inc. (Stony Brook, NY).
Animals
The mice were maintained according to the ARVO Statement for theUse of Animals in Ophthalmic and Vision Research and with protocolsapproved by animal care and Use Committee of the National Instituteof Health (NIH). All studies were performed using E_mut�/� mice andwild-type (Wt) littermate as control subjects. Animals were genotypedby amplification of mouse tail genomic DNA by testing for the pres-ence of the 5-bp deletion, as reported earlier.10 Mice were maintained
in a 12-hour dark–light cycle. All tissues used for analysis were col-lected from animals at the end of the dark cycle. E_mut�/� and Wtmice of ages 2 to 18 months were used to study histology; ultrastruc-ture; electroretinogram (ERG); photoreceptor gene expression; andfatty acid analysis of the retina.
Electroretinograms
To evaluate retinal function, full-field ERGs were recorded from five8-month-old E_mut�/� and six Wt control mice (all males) using aGanzfeld stimulus. Subsequently, five E_mut�/� mice and five Wt micefrom the same set of animals were used to record ERGs again at age 15months. Mice were kept in the dark for 12 hours before dark-adaptedresponses were recorded. Light-adapted responses were recordedagainst a continuous background of 34 cd/m2. All mice were anesthe-tized, and ERGs were performed as described earlier.11 These samemice were used for morphology, immunohistochemical analysis, anddetermination of expression of ELOVL4 at 18 months of age.
Retinal Histology and Ultrastructure
For evaluation of retinal histology, eyes were collected after perfusionwith a mixture of aldehyde consisting of 2% paraformaldehyde and 2%glutaraldehyde. Excised eyes were processed for histology and immu-nohistochemistry, as described earlier.4,12 Tissue sections were stainedwith toluidine blue and were viewed on a microscope (Eclipse E800;Nikon Inc., Tokyo, Japan), and the images were collected with image-processing software (MetaMorph; Universal Imaging Corporation,West Chester, PA).
For ultrastructural analyses, areas of interest that were identified atthe light level were thin sectioned and collected on 200-mesh grids.Sections were viewed on an electron microscope (model 2000; JEOLPeabody, MA). Images were captured using image-capturing software(Advanced Microscopy Techniques, Danvers, MA).
Immunohistochemistry
For evaluation of cone survival and localization of ELOVL4, immuno-histochemistry was performed using short (S)- and medium (M)-wave-length cone opsin–specific antibodies, anti-ELOVL4 antibody and anti-PDI (protein disulphide isomerase) antibody as described earlier.4,12
Sections were viewed under confocal microscope, and images werecaptured with the use of appropriate filters and lasers.
Expression of Elovl4 and Photoreceptor-SpecificGenes in E_mut�/� Retinas
Total RNA was isolated from 2-, 4-, and 7-month-old animal retinas(RNeasy kit; Bio-Rad Laboratories, Hercules, CA) according to the
FIGURE 1. Schematic depiction ofthe E_mut�/� targeting strategy. (A)Wild-type Elovl4 allele, with restric-tion enzyme sites (B, BamHI; E,EcoRI; R, EcoRV) and location of ex-ons (boxes). The restriction siteswere used to analyze the recombi-nants (data not shown). (B) The tar-geting vector, generated on the longarm of Elovl4 with a 5 bp-deletion(AACTT) and a Neo gene cassetteflanked by loxP sites. (C) The Neogene cassette with the 5-bp–deletionmutation were introduced in exon 6.N7 and N1 are the forward and re-verse primer sequences designed toread from the selection cassette intothe short arm (N1) and the long arm(N7). Forward (F) and reverse (R)primers were used to confirm thehomologous recombination.
IOVS, October 2006, Vol. 47, No. 10 Photoreceptor Degeneration in Elovl4 Knock-in Mice 4559
manufacturer’s guidelines, and the reverse transcription reaction wasperformed (SuperScript-II; Invitrogen). The comparative Ct methodwas used to calculate the expression of total Elovl4, Wt, and mutantElovl4 transcripts, and photoreceptor-specific gene expression in dif-ferent samples of mouse retina, as described earlier.5 The primers usedfor the gene expression studies are listed in Table 1.
Protein Isolation and Western Blot Analysis
Western blot analysis was performed as described earlier, with anti-ELOVL4 antibodies and anti-�-actin antibodies.4,9 Levels of expressionof ELOVL4 were compared by loading an equal amount of protein andmeasuring the intensity of immunopositive bands by densitometry.
Lipid Extraction and Fatty Acid Analysis
At the end of a 12-hour dark cycle, retinas were collected from6-month-old E_mut�/� and Wt mice (n � 10 in each group). Theretinas were weighed, and lipid extraction was performed according tothe method of Bligh and Dyer.13 Butylated hydroxytoluene (BHT) wasadded to each sample, along with the internal standard 22:3 n-3 methylester. Retinas were subsequently transmethylated using the BF3-meth-anol method of Morrison and Smith14 as modified by Salem et al.,15
with the cosolvent hexane. The methyl ester samples were analyzed bygas chromatography, as previously described,15 except that the retinaswere injected using a splitless method. For the splitless injection of 2�L of the hexane extract, the purge flow to the split vent was at a rateof 14 mL/min and the valve opened 0.25 minute after injection.
Statistical Analysis
Data are presented as the mean � SD. Comparisons of means betweentwo experimental groups were performed using the two-tailed, inde-pendent Student’s t-test. Analysis of variance (ANOVA) was used forcomparisons of means among more than two groups. Adjustments formultiple comparisons made use of stepdown testing methods.16 ERGdata were analyzed with the Student’s t-test at each flash intensity, andwith the Mixed procedure (SAS, Cary, NC) for comparison of the entireintensity curve between these groups.
RESULTS
Expression of the Elovl4 in E_mut�/�
Mouse Retina
The KI mice carrying the 5-bp deletion in the heterozygousstate were viable and fertile. At age 2 months, the presence ofboth Wt and mutant Elovl4 transcripts was detected in theretinas of E_mut�/� mice using primers specific to the Wt andmutant alleles by qRT-PCR (Fig. 2A). Analysis of the 2-month-old retinal extract by Western blot also demonstrated thepresence of both Wt (37kDa) and mutant proteins (33 kDa; Fig.
TABLE 1. qRT PCR Primers Used for the Analysis of Photoreceptor Genes
FIGURE 2. (A) Evaluation of expression levels of Elovl4 in E_mut�/�
mice: Expression profile of Wt transcript (B) and mutant transcript (C)in E_mut�/� mice of 2 and 7 months age in comparison to theexpression of Wt transcript in age-matched control retina (A) as deter-mined by qRT-PCR. Both Wt and mutant transcripts were detected at2 and 7 months of age, indicating the expression of Wt and mutantalleles. (B) Immunoblot analysis of ELOVL4 protein in E_mut�/� micein comparison with the wild type. Total retinal lysates from E_mut�/�
mice and the Wt were subjected to SDS-PAGE followed by Westernblot analysis using anti-ELOVL4 antibodies. Bands corresponding to Wtand mutant ELOVL4 with expected molecular masses of 37 and 33 kDa,respectively, were observed in E_mut�/� mice, indicating the pres-ence of Wt and mutant proteins.
4560 Vasireddy et al. IOVS, October 2006, Vol. 47, No. 10
2B). These observations indicate that both Wt and mutantElovl4 transcripts and proteins are produced in the E_mut�/�
mouse retinas.
Evaluation of Expression of Photoreceptor-Specific Genes in E_mut�/� Retina
Real-time quantitative RT- PCR analysis was used to measurethe expression level of photoreceptor-specific genes: S-coneopsin (Opn1sw), M-cone opsin (Opn1mw), rhodopsin (Rho),RDS/peripherin, Rom1, Abca4, rod and cone transducin(Gnat1, Gnat2), rod and cone arrestin (Sag1, Arr3), andElovl4, in E_mut�/� mouse retinas (Fig. 3).
In the retina of 2-month-old E_mut�/� animals, there wasno significant change in the expression of either rod- or cone-specific genes or total Elovl4 (Wt and mutant alleles together)when compared with the expression in Wt mice (P � 0.05,data not shown). By 4 months, the expression of total Elovl4was found to be decreased, but it was not significant (P �0.207). There was a slight decrease in the expression of cone-specific markers such as Opnl-sw, Opnl-mw, and Arr3 inE_mut�/� mice when compared with Wt animals, but thedifference in the levels of expression was not yet significant atthat age (P � 0.1). There was no indication that levels ofexpression of other marker genes, Gnat2, Rho, Rom1, sag1, orGnat1 were altered in E_mut�/� mice compared with Wtcontrol mice (Fig. 3).
By 7 months, the expression of total Elovl4 was found to bereduced to less than half in the E_mut�/� retina comparedwith Wt retina (P � 0.001, Fig. 3). A significant decrease (P �0.043) in the expression of rod and cone expressed geneAbca4 was observed in the E_mut�/� retina, whereas theexpression of Rds was unaltered. Cone-specific genes Opn1sw,Arr3, and Gnat2 showed significantly lower levels of expres-sion in E_mut�/� (P � 0.010) retina compared with Wt,whereas the expression of Opn1-mw remained unaltered. Atthis age, also the rod-specific genes Rho and rod arrestin (Sag1)showed a significant decrease (P � 0.05) in their expression inthe E_mut�/� retina, whereas rod transducin (Gnat1) ap-peared to be slightly reduced but not to a statistically signifi-cant extent (P � 0.1). Expression of Rom1 was found to beunaltered in E_mut�/� mice (Fig. 3).
In summary, the expression of cone-specific genes wasfound to be altered in the early stages of degeneration, and inlater stages the decrease in expression of cone-specific geneswas found to be greater than the decrease in expression ob-served in rod-specific genes. Among the rod-specific genestested, the expression of structural components of rod photo-receptors, such as Rds and Rom1 were not altered even inlate-stage degeneration (7 months), whereas the genes Rho,Sag1, and Gnat1, that are involved in the visual cycle showeddecreased levels of expression (Fig. 3). In E_mut�/� mice,reduction in the expression of selected photoreceptor genesindicates either downregulation of these genes or loss of pho-toreceptors, predominantly S-opsin expressing cones.
ERG Findings
At 8 months of age, b- and a-wave amplitudes were consistentlyhigher in dark- and light-adapted responses in E_mut�/� mice(Fig. 4). This difference was statistically significant (P � 0.05)in dark-adapted responses at flash intensities to which bothrods and cones contribute and at all flash intensities in thecone-driven, light-adapted ERG. When considering the set of alllog b-wave amplitudes across all stimulus intensities in thedark-adapted responses (inclusion criteria: amplitude �15 �V),there was a significant difference between Wt and E_mut�/�
mice (P � 0.004). The implicit time of the b-wave in thedark-adapted responses of the E_mut�/� mice was signifi-
cantly delayed relative to the Wt (P � 0.05). The b-wave of thelight-adapted responses was also slightly delayed in theE_mut�/� mice, though this difference was not statisticallysignificant.
The ERG at 15 months showed similar results though am-plitude differences between the groups were smaller. The b-and a-wave amplitudes again were consistently higher in mixedrod-cone dark adapted responses in the E_mut�/� mice; how-ever, the differences were no longer statistically significant.The b-wave amplitudes were significantly higher in theE_mut�/� mice in cone-driven light-adapted responses (P �0.05; Fig. 5).
Retinal Structure and Evaluation of PhotoreceptorCell Loss
To assess the effect of the 5-bp deletion–mutation in Elovl4 onthe retina, we performed several analyses to evaluate cellularstructure (Fig. 6) and cytoarchitecture (Fig. 7), along with anassessment of the number of cones (Fig. 8) and rods (Fig. 9) atvarious ages from 2 to 18 months. Light microscopic evaluationof retinal sections from 2- to 15-month-old E_mut�/� mice
FIGURE 3. Expression profile of photoreceptor-specific markers inE_mut�/� mouse retina. Quantitative expression of Elovl4, the cone-specific markers cone opsin (Opn1sw, Opn1mw), transducin (Gnat2),and cone arrestin (Arr3) and the rod-specific markers rhodopsin,transducin (Gnat1), and arrestin (Sag1) were determined by qRT-PCR(iQ SYBR Green Supermix and iCycler; Bio-Rad, Hercules, CA). A set offour housekeeping genes: hypoxanthine guanine phosphoribosyl trans-ferase (Hgprt), �-actin, ribosomal protein L (Rpl19), and succinatedehydrogenase (Sdha) were measured for normalization of the expres-sion values. Expression of different genes are presented (o) as apercentage of the expression of the gene in Wt mice (f). Data calcu-lated from at least three independent samples, each of which wasanalyzed at least in three replication reactions on an arbitrary scale.The results are presented as the mean � SD.
IOVS, October 2006, Vol. 47, No. 10 Photoreceptor Degeneration in Elovl4 Knock-in Mice 4561
revealed progressive changes in the outer retina (Fig. 6). Sev-eral abnormalities were present in the RPE of the E_mut�/�
mice. For example, vacuoles were present in the RPE at all agesfrom 2 to 15 months (Figs. 6B–G, 7D). A higher-magnificationview of the vacuoles is shown in Figure 7D. Moreover, thethickness of the RPE was considerably greater in E_mut�/�
mice at 2 months than in age-matched Wt control mice (com-pare Figs. 7D and 7A), a finding that has been described inabcr�/� mice.17 Numerous packets of phagocytized OS andirregularly shaped pigment granules were present throughoutthe cytoplasmic compartment of the RPE. Based on the criteriaused by Mata et al.,17 the appearance of these granules isconsistent with lipofuscin deposits. In addition, in E_mut�/�
mice 8 months of age and older, there were expansive areas ofdebris in the subretinal space (Figs. 6D and 6G). The subretinaldebris comprises primarily RPE fragments and undigestedouter segments (Fig. 6H).
As early as 2 months, structural abnormalities were presentin the outer segments of some photoreceptors at the outer/inner segment (OS–IS) interface (Fig. 6B). This localizationalong with the periodic spacing of the disrupted OS stronglysuggested that these were the OS of cone photoreceptors. Anexample of a dysmorphic cone OS is seen more clearly at theelectron microscopic level (Fig. 7E). To assess the effect of the5-bp deletion in Elovl4 specifically on cone photoreceptors,
cryosections taken from the posterior pole of eyes obtainedfrom E_mut�/� and Wt mice at various ages were immuno-stained with S- or M-opsin-specific antibodies. The number ofimmunopositive M- or S-cones per 250-�m field was deter-mined in central and peripheral areas across each retinal sec-tion. There was a significant decrease in the number of conesexpressing S opsin in E_mut�/� mice aged 6 months or olderthan in control animals (P � 0.023; Fig. 8A). In contrast, noage-related decline in the number of cones expressing M opsinwas found (P � 0.062; Fig. 8B). At 18 months, the number ofcones expressing S-opsin continued to decline compared withcontrol mice (P � 0.001; Fig. 8A). In contrast, the number ofcones expressing M-opsin showed no difference in numberwhen compared with control retinas.
Rod photoreceptor OS structure was normal at all ages.However, by 10 months, the length of rod OS and the thicknessof the outer nuclear layer (ONL) appeared thin compared withthat in younger mice. To determine whether the loss of rodswas significant, we measured the length of OS and counted thenumber of cell layers in the outer nuclear layer. Although wefound no difference compared with control subjects in eithermeasure at 8 months (data not shown), at 10 months a signif-icant difference in the ONL cell count in the far peripheralregion was observed (P � 0.03), whereas there is no significantchange in the mid periphery and central region. Rod outer
FIGURE 5. Representative (A) dark- and (B) light-adapted responses ofan E_mut�/� and a Wt control mouse at 15 months. Intensity–re-sponse curves of (circles) b- and (squares) a-wave amplitudes of the(C) dark and (D) light-adapted responses of E_mut�/� (n � 5, filledsymbols) and Wt (n � 5, open symbols) mice at 15 months. Standarderror bars are shown. b-Wave amplitudes were significantly differentbetween the groups at all intensities of the light adapted responses in(D) (inclusion criteria, �15 �V).
FIGURE 4. Representative (A) dark- and (B) light-adapted responses ofan E_mut�/� mice and a Wt control mouse at age 8 months. Intensity–response curves of (circles) b- and (squares) a-wave amplitudes of the(C) dark- and (D) light-adapted responses of E_mut�/� (n � 5, filledsymbols) and Wt mice (n � 6, open symbols). Standard error bars areshown. *Significant amplitude differences between the groups (P �0.05). Amplitudes are significantly different between the groups at allintensities in the light-adapted responses in (D) (inclusion criteria,�15 �V).
4562 Vasireddy et al. IOVS, October 2006, Vol. 47, No. 10
segment length at the far periphery and mid periphery wasfound to be significantly shorter at 10 months of age whencompared with that in control subjects (P � 0.012, 0.005),whereas no significant difference was found in the centralregion.
The final structural abnormality manifest in the eyes ofE_mut�/� mice involved the outer plexiform layer (OPL). Atthe light microscopic level, the OPL contained gaps and washeterogeneously stained. At the electron microscopic level,there were areas surrounding the rod spherules that appeared
FIGURE 6. Light microscopic evalua-tion of the retina in wild-type (A) andE_mut�/� mice (B–G) at variousages: (A) 2-month-old Wt mouse; (B)2-month-old E_mut�/�mouse; (C)4-month-old E_mut�/�mouse; (D,G) 8-month-old E_mut�/�mouse; (E)10-month-old E_mut�/�mouse; and(F) 15-month-old E_mut�/�mouse.The retinal area illustrated in (D) isshown at lower magnification in (G)to demonstrate the extent of the sub-retinal debris. A similar area is shownat higher magnification in (H) andindicates that the debris is composedprimarily of RPE fragments and undi-gested OS. Black arrows: disruptedcone OS; white arrows: vacuoles inthe RPE (high-magnification imageshowing vacuoles, SupplementaryFig. S1, is available online at http://www.iovs.org/cgi/content/full/47/10/4558/DC1). Black asterisks indi-cate gaps in the OPL. Scale bar, 10 �m.
FIGURE 7. Electron microscopic eval-uation of the retinas in Wt (A–C) andE_mut�/� mice (D–F) at 2 months.White arrows: undigested OS; blackarrows: small, irregularly shapeddense bodies (lipofuscin); black as-terisks: vacuoles within the RPE;black arrowhead: disrupted coneOS; white arrowhead: gaps sur-rounding rod spherules in OPL. Highmagnification image of (D) is avail-able as Supplementary Fig. S1, onlineat http://www.iovs.org/cgi/content/full/47/10/4558/DC1. Scale bar, 2 �m.
IOVS, October 2006, Vol. 47, No. 10 Photoreceptor Degeneration in Elovl4 Knock-in Mice 4563
to be empty and lacked the extracellular matrix present in theWt control subjects (compare Figs. 7F and 7C).
Our recent studies in which we evaluated the localization ofELOVL4 in COS-7 cells demonstrated that the presence ofELOVL4 with the 5-bp deletion caused mislocalization of theprotein away from the endoplasmic reticulum.6,9 Similar ob-servation was reported by Grayson and Molday.6 and us.9Todemonstrate whether a similar phenomenon occurred in themouse retina, we evaluated the localization patterns of boththe ELOVL4 protein and PDI, an ER marker. In the retina of4-month-old Wt animals, ELOVL4 was localized in the expectedlocation (Fig. 9A). ELOVL4 was found in greatest abundance inthe ER of photoreceptors along with a minor component in thephotoreceptor inner segments outside of the ER, as observedpreviously. Minimal immunopositive labeling was found in theOPL, consistent with previously reported localization ofELOVL4 (Figs. 9A, 9B).4,6,9 With this anti-ELOVL4 antibody,directed toward the N terminus of the molecule, we alsoobserved immunoreactivity in the inner nuclear layer (INL).This finding differs from the findings of Grayson and Molday,6
who used an antibody directed toward the C terminus ofELOVL4. This methodological difference probably accounts forthe discrepancy in the pattern of immunoreactivity. InE_mut�/� mice of the same age, there were two strikingdifferences in the localization pattern of ELOVL4 comparedwith that in control mice. First, the quantity of immunopositive
labeling in the inner segments appeared to be much greater.Second, there appeared to be a substantially greater amount ofELOVL4 localized to the OPL. To verify these differences inELOVL4 amount, we quantified the amount of signal capturedwith the ELOVL4-specific channel from the confocal micro-scope. To make the images directly comparable, all immuno-histochemistry experiments were performed simultaneouslyand all images were taken on the same afternoon using iden-tical microscope settings. Image intensity was determined withimage-analysis software (MetaView; Update Software, ChevyChase, MD) on areas of the same size over the inner segmentand OPL regions. The integrated intensity values were normal-ized such that the values obtained from Wt retinas were set to1.0. The amount of immunopositive label of both the innersegments (P � 0.015) and OPL (P � 0.003) were significantlygreater in the E_mut�/� mice than in the Wt littermate controlsubjects. Consistent with these observations, semiquantitativeWestern blot analysis of ELOVL4 protein in the retinas ofE_mut�/� and Wt mice also indicated a higher amount of totalELOVL4 protein in the E_mut�/� (Fig. 9D). The anti-ELOVL4antibodies used in these experiments recognize both the Wtand mutant proteins.
Retinal Fatty Acid Profile of Elovl4 5-bp–DeletionMutant KI Mice
To determine the effect of the Elovl4 5-bp deletion–mutationon the retinal fatty acid composition, we analyzed the fatty acidcomposition of retinas of 6-month-old mice. In addition, the
FIGURE 9. Confocal immunohistochemical localization of ELOVL4(red) and PDI (green) in retinas from Wt (A) and E_mut�/� (B) miceat 4 months. Blue: nuclei. (C) The average normalized integratedintensity values for the IS region and OPL region. The immunolabelingintensity in the IS (P � 0.015) and OPL (P � 0.003) was significantlyhigher in the E_mut�/� mice. (D) Western blot analysis of ELOVL4protein and �-actin in the retina of E_mut�/� mice in comparison tocontrol retina. Scale bar, 10 �m.
FIGURE 8. Analysis of the number of cone photoreceptors at variousages. Cone photoreceptors were analyzed from 4-, 6-, and 18-month-old E_mut�/� mice retinas immunolabeled with anti-S or -M opsinantibodies and compared with the Wt. (A) The number of S-opsin-containing cones at 4 (P � 0.065), 6 (P � 0.023), and 18 (P � 0.00)months. (B) The number of M-opsin-containing cones at 4 (P � 0.013),6 (P � 0.062), and 18 (P � 0.472) months. (f) Wt; (u) E_mut�/�.Data are the mean � SD.
4564 Vasireddy et al. IOVS, October 2006, Vol. 47, No. 10
fatty acid composition of the whole eyeballs of pups on the dayof birth (P0) with no fully formed photoreceptors was deter-mined.
There were no significant differences in the composition ofretinal fatty acids of 6-month-old E_mut�/�mice and Wt con-trol subjects, with the exception of some of the n-3 fatty acids.A significant alteration in the levels of eicosapentaenoic acid(EPA), 20:5n-3 (P � 0.028), docosapentaenoic acid (DPA),22:5n-3 (P � 0.04), and 24:6n-3 (P � 0.005) was observed inthe E_mut�/� retina compared with Wt retinas (Table 2). Asimilar trend of alteration in long-chain fatty acids in the wholeeyeballs was observed in P0 E_mut�/� pups when comparedwith the Wt (Table 3). There is a significant alteration in thelevels of 20:0 (P � 0.01), 20:1 n-9 (0.038), EPA, 20:5n-3 (P �0.006), DPA, 22:5n-3 (P � 0.001), 24:5 n-3 (P � 0.003), and24:6 n-3 (P � 0.046) in the eyes of E_mut �/� P0 pupscompared with that of the fatty acid composition of controlanimals. The observed changes in long-chain fatty acid contentin adult and P0 pups indicates the possible involvement ofElovl4 in long-chain fatty acid metabolism.
DISCUSSION
In this study, we developed a KI mouse model containing anElovl4 5-bp deletion–mutation, shown previously to be associ-ated with human dominant Stargardt-like macular dystrophy(STGD3). The genotype of heterozygous KI (E_mut�/�) miceresembles humans affected with STGD3, as they express bothWt and mutant Elovl4 alleles. In these mice, cone photorecep-tor degeneration was evident from age 2 months, progressing
to significant cone photoreceptor loss by 6 months, with afurther decrease by 18 months. This effect was stronger on S-than M-cones. Rod photoreceptor loss was not evident until 10months. There was significant loss of both photoreceptor nu-clei and outer segment length after this time. Substantial RPEchanges are evident as early as 2 months including undigestedOS packets, lipofuscin, vacuoles, and deposits of pigmenteddebris in the subretinal space. These observations are consis-tent with the pathology observed in STGD3 patients includingprimary RPE atrophy, loss of central vision with visual acuitiesranging from 20/20 to 20/300, and worsening of color visionwith acuity loss.1,2,18 These changes have been found to beprogressive over three or four decades in STGD3 patients. Lossof photoreceptors and accumulation of lipofuscin were alsodescribed in Elovl4 5-bp–deletion transgenic mice with two toeight copies of the transgene.19 The morphologic changesdescribed in the transgenic mice appear to be more severethan in our KI animals, but these models have not been com-pared directly. The severity of retinal degeneration in theElovl4 5-bp deletion transgenic mice could be because of themultiple copy number of the transgene, the site of insertion ofthe transgene and also the presence of the two copies of theWt alleles that remain in the background in addition to thetransgene.
The structural aberrations observed in the photoreceptor ofE-mut�/� mice suggest retinal disease by age 2 months. Accu-mulation of a large amount of undigested OS material withinthe RPE and debris in the subretinal space of these miceindicates impaired digestion of phagocytized OS. Impairmentof the RPE function itself is likely to be involved in photore-
TABLE 2. Fatty Acid Profiles of Retinas from E_mut�/� and Wild-Type Mice at 6 Months
IOVS, October 2006, Vol. 47, No. 10 Photoreceptor Degeneration in Elovl4 Knock-in Mice 4565
ceptor cell loss that is commonly observed in macular degen-eration.20–22
The ERG response is the sum of all retinal potentials andcomprises negative and positive components. Increased ampli-tudes in a diseased retina may therefore result from eliminationof an underlying negative-going component. For example, ERGresponses increase after administration of tetrodotoxin (TTX)to remove spiking activity originating in the proximal reti-na.11,23 It is therefore possible that a suppression of a proximalretinal component of cone pathway function contributes tothe increased response amplitudes. However, elimination of acomponent is usually accompanied by altered waveforms, aswas seen when TTX is administered, whereas the ERG re-sponses of the E_mut�/� mice had normal waveforms.
Generally, in partial death of the photoreceptor populationERG amplitudes decrease; only a few studies report the oppo-site. One study reported slightly increased L-cone–driven ERGresponses in a patient with autosomal recessive Stargardt mac-ular dystrophy-fundus flavimaculatus24 associated with muta-tions in the ABCA4 gene. Other instances of supernormal conefunction were reported in the rhodopsin knockout mouse, aswell as in a transgenic pig model of retinal degeneration(P347L rhodopsin mutation) where they were attributed to anabnormal retinal development.25,26 In theory ERG amplitudescould also increase if resistivity of the RPE increases,27 as mightoccur with diffuse accumulation of lipofuscin. A plausiblefinding to explain the persistently supernormal ERG ampli-tudes could be the mislocalization of ELOVL4 immunoreactiv-ity to the OPL, where its presence may have induced, via an asyet to be determined mechanism, an increase in the photore-ceptor-to-bipolar cell synapse.
The fatty acid analysis showed significantly altered levels oflong-chain n-3 fatty acids not only in the retina of 6-month-oldanimals but also in the eyeballs of P0 pups with no significantphotoreceptors. The altered levels of fatty acids in the 6-month-old E_mut�/� animal retina could be due simply to the pho-toreceptor cell loss, but the alteration in the n-3 levels of P0pups with no photoreceptors may indicate a more definite rolefor ELOVL4 in n-3 fatty acid metabolism and the influence ofthe presence of the 5-bp deletion–mutation on the n-3 fattyacid profile. The level of docosahexaenoic acid (DHA, 22:6n-3), the major n-3 fatty acid present in the retina was notaltered either in the retinal tissue of 6-month-old E_mut�/�
mice or in the eyeballs of P0 pups, indicating that Elovl4mutation may not directly affect the DHA metabolism. The dietof these animals contained somewhat higher amounts of DHA,EPA, and linolenic acid (1% DHA, 1% EPA, 20:5 n-3; 1% linole-nic acid,18:3 n-3; and 20% linoleic acid, 18:2 n-6) comparedwith the mean amount in the American diet, and this may havemasked some of the effects of the Elovl4 mutation. Dietaryintake of DHA and EPA were suggested to effect favorably theseverity of phenotype in STGD3.28,29 Additional studies areneeded to evaluate the specific role of Elovl4 in fatty acidmetabolism.
The levels of expression of Wt Elovl4 were found to besignificantly lower in E_mut�/� animals than in control reti-nas. The retina of knock-out mice carrying the Elovl4 null allelein the heterozygous state did not develop significant photore-ceptor degeneration by 16 months, suggesting that haploinsuf-ficiency of Elovl4 may not cause photoreceptor degenera-tion.30 Therefore, the reduction in the levels of Elovl4 Wt allele
TABLE 3. Fatty Acid Profiles of Eyeballs from E_mut�/� and Wild-Type Mice at P0
4566 Vasireddy et al. IOVS, October 2006, Vol. 47, No. 10
alone may not be responsible for the photoreceptor degener-ation observed in the E_mut�/� mice.
In control mice, expression of Elovl4 is evident from mouseembryonic day 7, suggesting that Elovl4 may play a role innormal development.5 In the E_mut�/� mice, the develop-ment of the retina was observed to be normal with no signif-icant morphologic abnormalities until 2 months. This is verysimilar to what has been described in STGD3 patients in whichretinal development appears to be normal, and no vision ab-normalities are reported typically until the teenage years.1,2
Absence of developmental abnormalities in the retina ofE_mut�/� mice and patients with STGD3 and later onset ofretinal degeneration both in E_mut�/� mice and patients mayindicate that the effect of mutant Elovl4 does not immediatelyfollow the presence of mutant protein as such, but could bedue to a secondary process that is induced by the presence ofmutant ELOVL4 protein.
Morphology and gene expression profile of E_mut�/� miceindicated progressive photoreceptor cell loss, with significantcone cell degeneration, followed by loss of ROS length. In theretina of E_mut�/� mice accumulation of Elovl4 protein wasobserved in the photoreceptor and outer plexiform layers (Fig.9). In heterologous cultured cells expressing the wild-type andmutant ELOVL4 proteins, the mutant protein was found tointeract with the Wt protein resulting in the accumulation ofthe Wt-mutant protein aggregates. Therefore, it is likely thatthe retinal degeneration observed in the E_mut�/� mice andin patients with STGD3 could be due to a dominant negativeeffect exerted by the mutant protein.
The mouse retina contains two types of cone pigments,short-wavelength (S)-sensitive and medium-wavelength (M)-sensitive.31 Unlike humans and other species, the majority ofthe cones in the mouse express both cone pigments. TheM-cone opsin is expressed in every cone of the mouse retina,whereas the S-cone opsin is expressed in most of the cones.32
In E_mut�/� mice, we observed significant structural abnor-malities in cone photoreceptors, primarily resulting in a de-crease in the number of cones expressing both S-opsin proteinand S-opsin mRNA (Fig. 8). Although the specific type ofphotoreceptors that are affected in patients with STGD3 is notyet known, the loss of visual acuity and the worsening of colorvision with visual acuity loss reported in these patients areconsistent with the cone photoreceptor–associated changesobserved in the E_mut�/� mice. Selective loss of cone photo-receptors has been reported in mice lacking the cone-specificgene CNGA3.33 Predominant cone photoreceptor abnormali-ties in E_mut�/� mice and ELOVL4-associated macular degen-eration in humans suggest that cones are more susceptible tothe damage caused by Elovl4 mutations than are rods.
These KI mice with cone photoreceptor degeneration pre-ceding rod degeneration provide a valuable model for studyingdifferential molecular pathways that operate in cone and rodphotoreceptors and the mechanism underlying selective de-generation of macula and specifically to characterize the mech-anism(s) underlying degeneration in STGD3.
Acknowledgments
The authors thank Austra Liepa (University of Michigan) for genera-tion, maintenance of the animals, and assistance in the preparation ofmanuscript; Mitchell Gillett (University of Michigan) and KathyTroughton (University of Tennessee) for sectioning of histologic spec-imens; and Elisheva Reese (University of Tennessee) for assisting withimmunohistochemistry.
References
1. Stone EM, Nichols BE, Kimura AE, Weingeist TA, Drack A, SheffieldVC. Clinical features of a Stargardt-like dominant progressive mac-
ular dystrophy with genetic linkage to chromosome 6q. ArchOphthalmol. 1994;112:765–772.
2. Griesinger IB, Sieving PA, Ayyagari R. Autosomal dominant macu-lar atrophy at 6q14 excludes CORD7 and MCDR1/PBCRA loci.Invest Ophthalmol Vis Sci. 2000;41:248–255.
3. Tvrdik P, Westerberg R, Silve S, et al. Role of a new mammaliangene family in the biosynthesis of very long chain fatty acids andsphingolipids. J Cell Biol. 2000;149:707–718.
4. Ambasudhan R, Wang X, Jablonski MM, et al. Atrophic maculardegeneration mutations in ELOVL4 result in the intracellularmisrouting of the protein. Genomics. 2004;83:615–625.
5. Mandal MN, Ambasudhan R, Wong PW, Gage PJ, Sieving PA,Ayyagari R. Characterization of mouse orthologue of ELOVL4:genomic organization and spatial and temporal expression.Genomics. 2004;83:626–635.
7. Bernstein PS, Tammur J, Singh N, et al. Diverse macular dystrophyphenotype caused by a novel complex mutation in the ELOVL4gene. Invest Ophthalmol Vis Sci. 2001;42:3331–3336.
8. Maugeri A, Meire F, Hoyng CB, et al. A novel mutation in theELOVL4 gene causes autosomal dominant Stargardt-like maculardystrophy. Invest Ophthalmol Vis Sci. 2004;45:4263–4267.
9. Vasireddy V, Vijayasarathy C, Huang J, et al. Stargardt-like maculardystrophy protein ELOVL4 exerts a dominant negative effect byrecruiting wild-type protein into aggresomes. Mol Vis. 2005;11:665–676.
10. Zhang K, Kniazeva M, Han M, et al. A 5-bp deletion in ELOVL4 isassociated with two related forms of autosomal dominant maculardystrophy. Nat Genet. 2001;27:89–93.
11. Rangaswamy NV, Frishman LJ, Dorotheo EU, Schiffman JS, BahraniHM, Tang RA. Photopic ERGs in patients with optic neuropathies:comparison with primate ERGs after pharmacologic blockade ofinner retina. Invest Ophthalmol Vis Sci. 2004;45:3827–3837.
12. Jablonski MM, Dalke C, Wang X, et al. An ENU-induced mutation inRs1h causes disruption of retinal structure and function. Mol Vis.2005;11:569–581.
13. Bligh EG, Dyer WJ. A rapid method of total lipid extraction andpurification. Can J Biochem Physiol. 1959;37:911–917.
14. Morrison WR, Smith LM. Preparation of fatty acid methyl estersand dimethylacetals from lipids with boron fluoride–methanol. JLipid Res. 1964;53:600–608.
15. Salem N Jr, Reyzer M, Karanian J. Losses of arachidonic acid in ratliver after alcohol inhalation. Lipids. 1996;31(suppl):S153–S156.
16. Holm S. A simple sequentially rejective Bonferroni test procedure.Scand J Stat. 1979;6:65–70.
17. Mata NL, Tzekov RT, Liu X, Weng J, Birch DG, Travis GH. Delayeddark-adaptation and lipofuscin accumulation in abcr�/� mice:implications for involvement of ABCR in age-related macular de-generation. Invest Ophthalmol Vis Sci. 2001;42:1685–1690.
18. Lagali PS, MacDonald IM, Griesinger IB, Chambers ML, Ayyagari R,Wong PW. Autosomal dominant Stargardt-like macular dystrophysegregating in a large Canadian family. Can J Ophthalmol. 2000;35:315–324.
19. Karan G, Lillo C, Yang Z, et al. Lipofuscin accumulation, abnormalelectrophysiology, and photoreceptor degeneration in mutantELOVL4 transgenic mice: a model for macular degeneration. ProcNatl Acad Sci USA. 2005;102:4164–4169.
20. Strauss O. The retinal pigment epithelium in visual function.Physiol Rev. 2005;85:845–881.
21. Duncan JL, LaVail MM, Yasumura D, et al. An RCS-like retinaldystrophy phenotype in mer knockout mice. Invest OphthalmolVis Sci. 2003;44:826–838.
22. Boulton M, Dayhaw-Barker P. The role of the retinal pigmentepithelium: topographical variation and ageing changes. Eye.2001;15:384–389.
23. Bui BV, Fortune B. Ganglion cell contributions to the rat full-fieldelectroretinogram. J Physiol. 2004;555:153–173.
24. Scholl HP, Kremers J, Vonthein R, White K, Weber BH. L- andM-cone-driven electroretinograms in Stargardt’s macular dystro-phy-fundus flavimaculatus. Invest Ophthalmol Vis Sci. 2001;42:1380–1389.
IOVS, October 2006, Vol. 47, No. 10 Photoreceptor Degeneration in Elovl4 Knock-in Mice 4567
25. Banin E, Cideciyan AV, Aleman TS, et al. Retinal rod photorecep-tor-specific gene mutation perturbs cone pathway development.Neuron. 1999;23:549–557.
26. Jaissle GB, May CA, Reinhard J, et al. Evaluation of the rhodopsinknockout mouse as a model of pure cone function. Invest Oph-thalmol Vis Sci. 2001;42:506–513.
27. Rodieck RW. The Vertebrate Retina: Principles of Structure andFunction. San Francisco: W. H. Freeman; 1973.
28. Hubbard AF, Askew EW, Singh N, Leppert M, Bernstein PS. Asso-ciation of adipose and red blood cell lipids with severity of dom-inant Stargardt macular dystrophy (STGD3) secondary to anELOVL4 mutation. Arch Ophthalmol. 2006;124:257–263.
29. MacDonald IM, Hebert M, Yau RJ, et al. Effect of docosahexaenoicacid supplementation on retinal function in a patient with autoso-
mal dominant Stargardt-like retinal dystrophy. Br J Ophthalmol.Feb 2004;88:305–306.
30. Dorit Raz-Prag RA, Avvagari R, Fariss RN, et al. Haploinsufficiencyis not the key mechanism of pathogenesis in a heterozygous Elovl4knockout mouse model of STGD3. 2006;47:3603–3611.
31. Szel A, Rohlich P, Caffe AR, van Veen T. Distribution of conephotoreceptors in the mammalian retina. Microsc Res Tech. 1996;35:445–462.
32. Applebury ML, Antoch MP, Baxter LC, et al. The murine conephotoreceptor: a single cone type expresses both S and M opsinswith retinal spatial patterning. Neuron. 2000;27:513–523.
33. Biel M, Seeliger M, Pfeifer A, et al. Selective loss of cone functionin mice lacking the cyclic nucleotide-gated channel CNG3. ProcNatl Acad Sci USA. 1999;96:7553–7557.
4568 Vasireddy et al. IOVS, October 2006, Vol. 47, No. 10
