Retinal Degeneration in a Rodent Model of Smith-Lemli-Opitz Syndrome: Electrophysiologic, Biochemical, and Morphologic Features

Objective  To assess the electrophysiologic, histologic, and biochemical featuresof an animal model of Smith-Lemli-Opitz syndrome (SLOS).

Methods  Sprague-Dawley rats were treated with AY9944, a selective inhibitorof 3β-hydroxysterol-Δ⁷-reductase (the affected enzymein SLOS). Dark- and light-adapted electroretinograms were obtained from treatedand control animals. From each animal, 1 retina was analyzed by microscopy,and the contralateral retina plus serum samples were analyzed for sterol composition.The main outcome measures were rod and cone electroretinographic amplitudesand implicit times, outer nuclear layer (ONL) thickness, rod outer segmentlength, pyknotic ONL nucleus counts, and the 7-dehydrocholesterol/cholesterolmole ratio in the retina and serum.

Results  By 10 weeks' postnatal age, rod and cone electroretinographic wave amplitudesin AY9944-treated animals were significantly reduced and implicit times weresignificantly increased relative to controls. Maximal rod photoresponse andgain values were reduced approximately 2-fold in treated animals relativeto controls. The ONL thickness and average rod outer segment length were reducedby approximately 18% and 33%, respectively, and ONL pyknotic nucleus countswere approximately 4.5-fold greater in treated animals relative to controls.The retinal pigment epithelium of treated animals contained massive amountsof membranous/lipid inclusions not routinely observed in controls. The 7-dehydrocholesterol/cholesterolmole ratios in treated retinas and serum samples were approximately 5:1 and9:1, respectively, whereas the ratios in control tissues were essentiallyzero.

Conclusions  This rodent model exhibits the key biochemical hallmarks associatedwith SLOS and displays electrophysiologic deficits comparable to or greaterthan those observed in the human disease.

Clinical Relevance  These results predict retinal degeneration in patients with SLOS, particularlythose with the more severe (type II) form of the disease, and may be morebroadly relevant to other inborn errors of cholesterol biosynthesis. Thisanimal model may also be of use in evaluating therapeutic treatments for SLOSand in understanding the slow phototransduction kinetics observed in patientswith SLOS.

Smith-Lemli-Opitz syndrome (SLOS) (Online Mendelian Inheritance in Man[OMIM] 270400)¹ is the first described exampleof a "multiple congenital anomalies syndrome" and the first in a growing listof human hereditary diseases involving inborn errors of cholesterol biosynthesis.^2-4 This often lethal, autosomalrecessive disease affects an estimated 1 in 20 000 to 1 in 60 000primarily white live births worldwide, although the incidence may be as highas 1 in 1590 to 1 in 13 500 (see Battaile et al⁵),arguably making it the fourth most prevalent autosomal recessive human diseaseknown. The key biochemical features associated with SLOS are abnormal (typicallygrossly elevated) levels of 7-dehydrocholesterol (7DHC), a penultimate precursorof cholesterol, and markedly reduced levels of cholesterol in blood and otherbodily tissues.^6-9 Thisabnormality in sterol composition is a consequence of defective activity inthe enzyme 3β-hydroxysterol-Δ⁷-reductase (EC 1.3.1.21),which catalyzes the conversion of 7DHC to cholesterol.^10-12 Theenzymatic defect, in turn, can be ascribed to any one of a myriad of mutationsfound to be associated with the DHCR7 gene (localizedto 11q12-q13), which encodes the Δ⁷-reductase enzyme.^13-15 Patients with SLOSexhibit a constellation of profound phenotypic and functional abnormalities,including microcephaly, holoprosencephaly, dysmorphic craniofacial features,micrognathia, limb asymmetries, syndactyly of the second and third toes, polydactyly,ambiguous genitalia, hypospadias, cleft palate, short nose with antevertednostrils, abnormal visceral development (especially the kidneys, liver, andlungs), hypotonia, failure to thrive, and mental retardation.^1-4

Reported ophthalmologic defects associated with SLOS include choroidalhemangiomas, absence of lacrimal puncta, blepharoptosis, pale optic discs,optic atrophy, optic nerve hypoplasia, sclerocornea and corneal endotheliumdefects, cataracts, and aniridia.^16-28 However,these findings were not uniformly observed in all of the relevant publishedcases and are not conclusive in the differential diagnosis of SLOS. In thesingle published case of retinal histopathologic features associated withSLOS, based on light and electron microscopic analysis of eyes obtained froma 1-month-old boy, Kretzer et al²³ describedextensive retinal ganglion cell and axonal dropout, particularly in the peripheralretina, with incipient optic nerve demyelination, as well as "mitochondrialdisintegration" in the retinal pigment epithelium (RPE) and the accumulationof "cytoplasmic masses" in the subretinal space proximal to the photoreceptorouter segments. However, the retina exhibited relatively normal histologicstratification and development, including the presence of grossly normal rodand cone photoreceptor cells.

Animal models for SLOS have been developed by treating rats with selectivepharmacologic inhibitors of 3β-hydroxysterol-Δ⁷-reductase,such as the experimental drugs AY9944 (trans-1,4-bis [2-dichlorobenzylamino-ethyl] cyclohexane dihydrochloride)^29-31 and BM15.766.^32-35 Inaddition, a genetic knockout mouse model also has been developed using homologousrecombination to delete much of the structural gene encoding the Δ⁷-reductase, but the affected progeny live only 18 to 24 hours postpartum, thus greatly limiting the utility of this animal model for experimentalstudies.^36,37 We previously describeda rat model of SLOS³⁸ created by the dietarytreatment of pregnant female rats with AY9944 during the second and thirdgestational weeks followed by the systemic injection of the progeny with thedrug during a 1-month postnatal period. Using this model, we found that despitemarked perturbation of cholesterol synthesis, including the gross elevationof 7DHC levels and reduction of cholesterol levels in the blood, retina, liver,and brain, the histologic and ultrastructural development of the retina proceedednormally, and the electrophysiologic competence of the retina was not substantiallycompromised. However, in subsequent studies³⁹ inwhich postnatal treatment was extended an additional 2 weeks, we observedreduced dark-adapted electroretinographic (ERG) b-wave amplitudes and increasedimplicit times in AY9944-treated rats, although there were no apparent histologicabnormalities in their retinas. These findings prompted us to speculate thatif we extend the treatment period sufficiently, we would observe even moresevere ERG deficits and frank histologic degenerative changes. Herein, weprovide compelling evidence demonstrating that long-term AY9944 treatmentin rats (up to postnatal age 10 weeks) causes profound retinal degeneration,correlating the changes in lipid metabolism with altered cellular physiologyand retinal structure. This model mimics the biochemical hallmarks associatedwith SLOS and is consistent with the recently reported ERG abnormalities associatedwith this devastating human disease.⁴⁰

Pregnant Sprague-Dawley rats (Harlan Sprague Dawley Inc, Indianapolis,Ind) (received 6 days after fertilization) and their progeny were used. Ratswere fed a cholesterol-free chow (Purina Mills TestDiet, Richmond, Ind), withfood and water provided ad libitum, and were maintained under dim cyclic lighting(20-40 lux; with a 12-hour light and 12-hour dark cycle) at standard roomtemperature (22°C-25°C). All procedures were approved by the SaintLouis University Animal Care Committee and were in accordance with the ARVO Resolution on the Use of Animals in Research and withthe NIH Guide for the Care and Use of Laboratory Animals.

AY9944 was obtained by custom synthesis and purified by recrystallizationto homogeneity (A. H. Fauq, PhD, and S.J.F., unpublished data, 2003), withits chemical, physical, and spectroscopic properties confirmed by comparisonwith an authentic sample of AY9944 (a gift from Wyeth-Ayerst Research, Princeton,NJ). The treatment protocol was essentially as described previously by Fliesleret al,³⁸ except the treatment period was extendedto 10 postnatal weeks. In brief, pregnant (6 days after fertilization) Sprague-Dawleyrats were fed cholesterol-free chow containing AY9944 (1 mg/100 g of chow;maximum of 40 g of chow daily) for the last 2 weeks of the 3-week gestationalperiod. Progeny were then injected on alternating days, 3 times per week,with an aqueous olive oil emulsion containing AY9944 (10 mg/mL; 25-30 mg/kgof body weight) and with a mixture of the fat-soluble vitamins A, D, and E.Controls were injected with vehicle alone and also were fed cholesterol-freechow.

The ERGs were recorded from 4 control rats and 5 AY9944-treated rats.Animals were dark adapted overnight and then anesthetized by intramuscularinjection of ketamine hydrochloride (75 mg/kg) and xylazine hydrochloride(5 mg/kg). The pupils were dilated with 1% tropicamide and 2.5% phenylephrinehydrochloride, and the rats were placed on a regulated heating pad throughoutthe recording session. Strobe flash stimuli were presented using a Ganzfeldbowl (LKC Technologies, Gaithersburg, Md). The ERGs were recorded from botheyes using a thin stainless steel wire contacting the corneal surface througha thin layer of 1% methylcellulose. Platinum needle electrodes inserted intothe cheek below each eye and into the tail served as reference and groundleads, respectively. Responses were differentially amplified (0.5-1500 Hz),averaged, and stored using a signal-averaging system (model UTAS E-2000; LKCTechnologies).

In each recording session, a dark-adapted response series was obtainedfirst using strobe flash stimuli that ranged in intensity from −4.3to 0.5 log candela (cd)-sec/m², controlled by placing Wratten neutraldensity filters (Eastman Kodak Co, Rochester, NY) in the light path. Stimuliwere presented in order of increasing intensity, and at least 2 responseswere averaged at each flash intensity. A steady rod-desensitizing adaptingfield (0.6 cd/m²) was then presented in the Ganzfeld bowl. After7 minutes of light adaptation, responses were recorded to strobe stimuli rangingfrom 1.2 to 0.5 log cd-sec/m². At each intensity, responses to50 successive flashes presented at 2.1 Hz were averaged.

The a-wave amplitude was measured from the prestimulus baseline to thetrough of the a-wave, whereas the b-wave was measured to the positive peak,either from the trough of the a-wave or (if no a-wave was present) from thebaseline. Implicit times for a- and b-waves were measured from the time ofstimulus presentation to the a-wave trough and to the b-wave peak, respectively.

We also analyzed the leading edge of the dark-adapted a-wave in termsof a modified form of the Lamb and Pugh model of rod phototransduction^41,42

P3(i,t) = {1 − exp[−iS(t − t_d)2]} R_mP3,

where P3 representsthe mass response of the rod photoreceptors, the amplitude of which is expressedas a function of flash energy (i) and time (t) after flash onset; S, the gainof phototransduction; R_mP3, the maximumresponse; andt_d, a brief delay. This model was initiallyapplied to human control and patient data,^43,44 andit has been used since to model the leading edge of the rodent a-wave.^45-47 Application of thismodel requires the use of high-intensity stimuli.⁴⁰ Owingto this constraint, we only fit the model to responses obtained to the highest-intensitystimulus (0.5 log cd-sec/m²). To provide a basis of comparisonwith data reported by Elias et al⁴⁰ from patientswith SLOS, we converted 0.5 log cd-sec/m² to 16 520 R* usinga conversion factor of 1 cd-sec/m² approximately equal to 5012R*, where R* represents the number of rhodopsin photoisomerization eventsper rod.

Methods used for the saponification, extraction, chromatographic resolution(reverse-phase high-performance liquid chromatography), identification, andquantitative measurement of sterols from rat tissues (including serum, neuralretina, liver, and brain) were as described in detail elsewhere.^38,48 Allprocedures were performed under dim room illumination to minimize light-inducedisomerization and degradation of lipids. Tissue samples were harvested immediatelyafter ERG recording, while the animals were still under deep anesthesia. Bloodsamples were obtained by intracardiac puncture, and serum samples were preparedby centrifugation of whole blood after allowing for clotting to occur. Oneeye from each animal was harvested for histologic and ultrastructural analysis(see the following subsection); neural retinas (free of RPE) from contralateraleyes were rapidly dissected out and snap frozen in liquid nitrogen, as werethe livers, brains, and serum samples. Specimens were stored in darkness at−85°C until ready for analysis. Immediately before saponification,tissues were supplemented with an internal standard of [³H] cholesterol(American Radiolabeled Chemicals Inc, St Louis, Mo) to provide a means ofcorrecting for losses incurred during preparation of the nonsaponifiable lipidextracts and to act as an internal chromatographic standard for comparisonwith the mass elution profiles (UV detection, 205 nm). Identification andquantification of sterols and sterol mass were accomplished by comparing theindividual chromatographic peak retention times and integrated peak areaswith those of authentic sterol standards, particularly cholesterol and 7DHC(obtained from Sigma-Aldrich Corp, St Louis; recrystallized from methanol-watertwice before use). In addition, the chromatographic properties and detectorresponse factor (high-performance liquid chromatography integration unitsper nanomole of sterol) were determined for an authentic standard of 8-dehydrocholesterol(8DHC) (cholesta-5,8[9]-dien-3β-ol), obtained previously as a gift fromGeorge J. Schroepfer, Jr, MD, PhD (Rice University, Houston, Tex).

Retinas from the eyes of AY9944-treated and control rats were analyzedat the light and electron microscope levels essentially as described in aprevious publication.^49,50 Inbrief, 1 eye each from treated and control animals was immersed overnightat 4°C in buffered mixed aldehyde fixative (2% glutaraldehyde, 2% paraformaldehyde,in 0.125M sodium cacodylate buffer, pH 7.4, containing 0.025% calcium chloride),after removal of the superior cornea and the lens; the inferior cornea wasleft in place to serve as a geographic marker. After buffer rinses, osmification,and dehydration through a graded ethanol series, eyes were embedded in epoxyresin (Spurrs formulation), and 0.75-µm-thick sections were collectedonto glass microscope slides using an ultramicrotome (UltraCut-E; ReichertOphthalmic Instruments, Depew, NY). Eyes were sectioned along the verticalmeridian through the optic nerve head, from the superior ora serrata to theinferior ora serrata, and examined by light microscopy after staining with1% toluidine blue and coverslipping using a photomicroscope (model BH-2; Olympus,Melville, NY) with an oil-immersion lens (20× DPlanApo or 60×SPlanApo; Olympus). Digitized images were obtained using a digital camera(model DXM1200; Nikon Instruments Inc, Melville, NY), and images were storedusing Nikon software on an IBM-compatible personal computer. Thin (70- to80-nm) sections corresponding to retinal regions of particular interest werecollected onto copper mesh grids, counterstained with uranyl acetate–leadcitrate, and examined using an electron microscope (model 100CX; JEOL USA,Peabody, Mass) and an accelerating voltage of 60 keV.

Quantitative measurements of outer nuclear layer (ONL) thickness, rodouter segment (ROS) lengths, and pyknotic nuclei in the ONL were performedessentially as described previously.^51,52 Resin-embeddedblocks of hemisected eyes were sectioned for light microscopy along the verticalmeridian, as indicated in the previous subsection. Three serial sections werecut and mounted from each block and then examined and the section plane wasadjusted, as needed, for proper ROS alignment. From each section, digitizedimages were collected at 0.5-mm intervals, beginning at the optic nerve headand proceeding to the far periphery in the inferior and superior hemispheres.From each such image, 5 ONL thickness measurements and 10 ROS length measurementswere gathered from random locations within the field of view, and the datawere tabulated using a electronic spreadsheet (Excel; Microsoft Corp, Redmond,Wash). This provided 15 independent ONL measurements and 30 independent ROSmeasurements per locus; mean ± SD values for each locus were calculatedfrom eyes obtained from 3 to 5 rats per treatment group and plotted as a functionof distance from the optic nerve head, thereby generating a morphometric profilefor ONL thickness and ROS length across the vertical meridian for controland treated animals. Summary data averaged across the vertical meridian werefurther analyzed using a homoscedastic t test (Excel);significance was determined at the 95% confidence interval or higher. Usingthe same fields, pyknotic nuclei in the ONL were counted and recorded. Pyknoticcell counts were summed and tabulated for each linear millimeter of verticalmeridian length, and the resulting data were subjected to the same statisticalanalysis as used for ONL thickness and ROS length measurements.

Figure 1A presents a seriesof representative dark-adapted ERGs obtained from a control rat and a rattreated with AY9944. In each series, the responses to 5 differentstimulusintensities are superimposed. Responses obtained from AY9944-treated ratswere similar in overall waveform but were smaller in amplitude compared withthose from control rats. Figure 1Bcompares average intensity-response functions for a- and b-wave amplitudes.The responses of AY9944-treated rats were significantly reduced in amplitudebelow control levels for the a(F_1,7 = 41.5; P<.01) and the b-waves (F_1,7 = 12.3; P<.01). Across the stimulus-response range tested, mean ±SD a-wave amplitudes of AY9944-treated rats were reduced to 66.9% ±15.5% of control values, and b-wave amplitudes were reduced to 81.5% ±3.1% of control values across the stimulus-intensity range. Figure 1C presents the corresponding implicit time data. The implicittimes for rod responses were substantially greater in the treated group thanin controls, and these differences were statistically significant for thea-(F_1,7 = 28.4; P<.01) and b-waves(F_1,7 = 78.1; P<.01). These resultsindicate a significant perturbation of the rod-localized phototransductionprocess (reflected in the a-wave) and the bipolar cell response (mirroredby the b-waves).

To further evaluate the effects of AY9944 treatment on rod photoreceptorfunction, we analyzed the leading edge of the a-wave in terms of the Lamband Pugh model^41,42 of rod phototransduction. Figure 2A compares the leading edge of thea-wave from each rat in response to the highest-intensity stimulus (0.5 logcd-sec/m²). In this format, responses have been normalized by dividingthe entire response by the maximal response amplitude (R_mP3) variable.The leading edge of the a-wave reached the trough more slowly for AY9944-treatedrats than for control animals, and there was no overlap between the 2 groups. Figure 2B displays the R_mP3 valuesobtained for each rat. The average value of R_mP3 for the AY9944-treatedanimals was reduced nearly 2-fold compared with that of controls (t₇ = 9.8; P<.001). Figure 2C plots the corresponding valuesof photoreceptor gain (S). The average value of S for the AY9944-treated ratsalso was reduced nearly 2-fold compared with that of controls (t₇ = 5.9; P<.001).

Figure 3A compares cone ERGsobtained from a representative control rat and an AY9944-treated rat. Theoverall amplitude of the cone ERG is reduced after AY9944 treatment. Figure 3B compares average intensity-responsefunctions for cone ERG b-wave amplitude; the responses of AY9944-treated ratswere significantly reduced below control levels (F_1,7 = 24.8; P<.01). Across the stimulus-response range tested, mean± SD cone ERG amplitudes of AY9944-treated rats were reduced to 60.8%± 1.0% of control values. Figure 3C presents the corresponding implicit times, which were significantlyprolonged in the treated group compared with controls (F_1,7 = 78.1; P<.01). These results indicate that cone photoreceptorfunction also was markedly affected by AY9944 treatment, as was the functionof the depolarizing class of cone bipolar cell, which makes a large contributionto the rat cone ERG under these stimulus and recording conditions.⁵³

These findings demonstrate that rod and cone function are markedly affectedby prolonged systemic treatment of rats with AY9944. Furthermore, they suggestthat visual information processing in the inner retina, at least at the levelof bipolar cell function, also is compromised by this treatment.

In good agreement with a previous study,³⁸ andconsistent with results obtained by other researchers^29,31,54-56 regardingthe effects of AY9944 on lipid metabolism, rats treated systemically withAY9944 for nearly 3 months exhibited grossly deranged sterol metabolism (Table 1 and Figure 4). This finding is evidenced by the marked accumulationof 7DHC (and lesser amounts of the 8-dehydro isomer, 8DHC) and the substantialreduction in the cholesterol content of retina and serum relative to age-matchedcontrols. In controls, cholesterol is the overwhelmingly dominant, if notexclusive, sterol present in all tissues examined, and steady-state levelsof 7DHC and 8DHC are barely detectable, if present at all. In fact, 7DHC wasthe dominant sterol detected in tissues from AY9944-treated animals, withthe 7DHC-cholesterol mole ratio being approximately 5:1 for retinas and nearly9:1 for serum samples. A similar trend in sterol composition was observedon analysis of whole brain and liver (data not shown). Although 7DHC is, byfar, the predominant sterol in the retinas and serum samples of AY9944-treatedrats, accumulation of 8DHC also was substantial, with the 8DHC-cholesterolmole ratio for retina being about 0.6 and the 7DHC-8DHC mole ratio being approximately8.7. On a per retina basis, the total sterol content of retinas from AY9944-treatedrats was approximately 80% that of age-matched controls, the difference beinglargely due to differences in eye size and, hence, overall retinal mass, sincetreated animals were considerably smaller than controls (eg, only approximately50%-60% of control body weight). The 7DHC-cholesterol mole ratio for serumsamples from AY9944-treated rats also was approximately 8.7, cholesterol levelswere reduced to approximately 23% of control levels, and total serum sterollevels also were markedly reduced (by approximately 75%) relative to controls.These latter findings are consistent with the well-known hypolipidemic effectsof AY9944.^29-31,54-56

In contrast to the previously reported lack of histologic alterationsinduced in rat retinas up to 1 postnatal month of treatment with AY9944,³⁸ extending the treatment duration up to 3 postnatalmonths (10 weeks) resulted in obvious histologic changes consistent with progressiveretinal degeneration (Figure 5).These changes, which included reduction in ONL thickness, pyknosis of ONLnuclei, dropout of photoreceptor cells, and reduction in ROS length, wereobserved in the superior (Figure 5Aand B) and inferior hemispheres (Figure 5C and D) along the vertical meridian and are in general agreementwith the noted reductions in ERG amplitudes. The changes observed in ONL thicknessand nuclear pyknosis (measurements relevant to overall photoreceptor viability)as a consequence of extended AY9944 treatment were quantified (Figure 6). Overall, ONL thickness was reduced by approximately 18%relative to controls (P<.01), from the optic nervehead region to the periphery, consistent with a loss of approximately 2 of10 rows of photoreceptor nuclei. Photoreceptor loss was symmetrical, witha comparable degree of ONL thickness reduction observed in the superior andinferior hemispheres. Pyknotic nucleus counts (measured per linear millimeterof retinal expanse along the vertical meridian) in each retinal region examinedwere consistently and substantially higher (approximately 4.5-fold, on average)in AY9944-treated animals compared with controls, consistent with enhancementof photoreceptor cell death and dropout in AY9944-treated rats.

Superimposed on the loss of photoreceptor cells was a substantial reductionin ROS length in the remnant photoreceptors. Analysis of well-aligned rodsin the superior hemisphere revealed an overall 33% loss in ROS length whenAY9944-treated rats were compared with controls (mean ± SD ROS length,22.1 ± 3.7 µm and 33.2 ± 2.7 µm, respectively) (Figure 7). A similar reduction in ROS lengthalso was observed in the inferior hemisphere (data not shown); hence, theretinal degeneration seems to be relatively symmetrical and uniform in bothhemispheres rather than exhibiting a geographic preference for a particularretinal region. In most other respects, however, the remnant photoreceptorsand their outer segments seemed histologically and ultrastructurally normal(see also the "Comment" section), except for the increased incidence of ONLpyknosis mentioned previously herein. Owing to the relative paucity of identifiablecone photoreceptors in the rod-dominant rat retina and the difficulties inobtaining optimal cone outer segment alignment from region to region alongthe vertical meridian, a similar analysis of cone outer segment lengths wasnot performed.

In addition to the observed histologic changes in the neural retina,examination at the ultrastructural level revealed that the RPE in AY9944-treatedrats was abnormal compared with that in age-matched controls (Figure 8). The RPE cytoplasm in treated rats was congested withnumerous membranous and osmophilic (presumed lipid-laden) inclusions, includingphagosomes, multivesicular bodies, and residual bodies, well beyond the normallevel of such inclusions in the RPE of control rats. Again, this pathologicalfeature did not exhibit any apparent regional preference but was consistentlyobserved along the entire vertical meridian in the superior and inferior hemispheres.

In addition, unlike in control rats, where the RPE exhibited an accumulationand then clearance of ingested ROS tips (phagosomes) as a function of thetime of day,^57,58 the congestionof the RPE in AY9944-treated rats seemed relatively invariant throughout theday (data not shown). Regardless, the RPE in treated rats maintained its normalpolarity, for example, distribution of mitochondria proximal to the basalplasmalemma and the Bruch membrane, extension of apical microvilli, and maintenanceof basolateral membrane and junctional complex integrity. Also, at this stageof treatment, there was no evidence of appreciable RPE hypertrophy or hyperplasia,RPE nuclear chromatin appeared comparable to that of controls, there was noobvious change in the thickness of the Bruch membrane, and there was no apparentincreased deposition of lipid inclusions in the Bruch membrane relative tocontrols.

We described the electrophysiologic (ERG), biochemical, histologic,and ultrastructural features of a progressive retinal degeneration in a rodentmodel of SLOS. The deficits in photoreceptor-mediated retinal function aregenerally consistent with the observed photoreceptor degeneration, as characterizedby a reduction in the ROS length with increased photoreceptor pyknosis andcell loss (ie, diminished ONL thickness). The definitive biochemical featuresof this animal model—an elevated 7DHC-cholesterol mole ratio (due tohigh 7DHC levels and low cholesterol levels) and a marked reduction in totalserum sterol levels relative to controls—are consistent with the hallmarksof the human hereditary disease, particularly as observed in the more severe(type II) form of SLOS.^1-4,6-11 However,a direct comparison between the retinal histopathologic features of our animalmodel and those of human SLOS must await analysis of a larger cohort of SLOSdonor eyes than currently exists.

As mentioned previously herein, the only published description of SLOSretinal histopathologic features is that provided by Kretzer et al²³ in a case report of an ocular specimen obtained froman affected 1-month-old boy. The neurosensory retina exhibited relativelynormal histologic stratification of the cellular layers and well-differentiatedrods and cones. Those findings are consistent with our previous study³⁸ of the AY9944-induced SLOS animal model, where theretina appeared relatively normal, histologically and ultrastructurally, inthe first postnatal month of life.

Atchaneeyasakul et al²⁸ described variableocular findings in a group of 8 children (aged 1 week to 5 years) with well-documentedSLOS ranging from mild to moderately severe disease phenotypes. In addition,the sterol composition of ocular tissues (including the neural retina, RPE,lens, cornea, sclera, and ocular muscle) from a spontaneously aborted fetus(32 weeks' gestation) affected with SLOS was reported in that study, withthe corresponding tissues from the eyes of a nonaffected 3-month-old childserving as a control. All ocular tissues from the fetus with SLOS exhibitedgrossly elevated levels of 7- and 8-dehydrosterols compared with normal eyes(which did not exhibit any detectable 7DHC or 8DHC), with 7DHC-cholesterolmole ratios ranging from approximately 0.4 for the retina, cornea, sclera,and muscle to 1.48 for the lens, with the ratio for the RPE being 0.93. Qualitatively,these results are consistent with the biochemical findings in our animal model,although our animals exhibited more profoundly deranged cholesterol biosynthesis.

The recent study by Elias et al,⁴⁰ whichrepresents the first ERG study of patients with SLOS, demonstrates significantlydelayed rod activation and deactivation kinetics, as well as reduced postreceptorsensitivities, compared with unaffected controls. In that study, the sensitivityvariable S (see Hood and Birch⁴⁴) for patientswith SLOS was found to be only approximately 61% of the value determined forcontrols. This variable reflects relative mobilities of and efficiency ofinteractions between the components of the phototransduction cascade, frominitial photon absorption by the visual pigment rhodopsin to the closing ofthe cyclic guanosine monophosphate–gated ion channels in the outer segmentplasma membrane that govern the "dark current" in the rod cell.⁴² Thosefindings are consistent with the ones reported in the present study sincewe observed an approximately 2-fold decrease in the value of S in retinasfrom AY9944-treated rats compared with controls. Elias et al⁴⁰ hypothesizedthat the presumed decrease in cholesterol content of the ROS membranes, asinferred from the known inhibition of cholesterol biosynthesis and from theblood sterol analysis of their patients, may explain the slow phototransductionkinetics. However, we consider this unlikely given that (1) ROS membranesare notably cholesterol deficient naturally relative to other mammalian plasmamembranes⁵⁹ and, (2) if anything, a decreasein the cholesterol content of the membrane would be expected to increase thelateral mobility of the constituent membrane proteins, thereby increasingtheir interactions and, concomitantly, enhancing the rate of phototransductionbecause cholesterol tends to restrict molecular motions (decrease membranefluidity) above the phase transition temperature of the membrane lipids.^60,61 At this point, the reason for theslow kinetics of phototransduction remains unclear. However, the experimentalmodel described herein offers the ability to examine structure-function relationshipsin a systematic manner, with the potential of answering this question.

In contrast to the findings of Elias et al,⁴⁰ wherepatients with SLOS were found to have, on average, only a slight decrease(approximately 16%) in the saturated amplitude of the rod response (R_mP3), our SLOS rat model exhibited nearly a 2-fold reduction in theaverage R_mP3value relative to controls. The profound reductionin rod responses in the AY9944-treated rat may be due, in part, to the lossof nearly one third of the length of the ROS and the nearly 20% reductionin the total number of remaining viable rods. Because there was no correlativeassessment of retinal histologic characteristics in the study by Elias etal,⁴⁰ and, to our knowledge, there are no age-matchedSLOS retinal histologic specimens reported in the literature, further directcomparisons between our experimental animal study and the recently reportedstudy of human patients with SLOS cannot be made.

In addition to a-wave abnormalities, the timing of dark-adapted b-wavesof AY9944-treated rats was significantly slower than in control rats. Althoughthe b-wave represents the mass response of rod bipolar cells⁶² andis thus affected by changes at the photoreceptor level, a computational modelof the ERG (see Hood and Birch⁶³) indicatesthat the implicit time changes noted here cannot be replicated by a decreasein either S or R_mP3. Both of these changes will shift the implicittime function to the right along the stimulus-intensity axis but will notcause delays at low stimulus intensities. Instead, these results indicatethat there is an additional defect at the level of b-wave generation, at thesynaptic level, or in bipolar cell signal transduction. Cone dysfunction associatedwith SLOS was not evaluated in the study by Elias et al⁴⁰ orin any other published study, to our knowledge. In the present study, we showedthat light-adapted ERGs of AY9944-treated rats are dramatically altered comparedwith those of normal rats, that the average maximum cone response amplitudewas reduced by at least 40%, and that implicit times were substantially increased.Assuming that the rat model mimics the human disease, these results wouldpredict that patients with SLOS would exhibit cone dysfunction as well asthe previously documented rod dysfunction.

Why should derangement of cholesterol biosynthesis, with concomitantaccumulation of 7DHC, cause photoreceptor cell death and retinal degeneration?In brief, it has been proposed that cytotoxic "oxysterols" derived from 7DHCmay be involved (for a detailed discussion, see Fliesler⁶⁴).Hence, this may represent yet another (albeit somewhat specialized) exampleof the lipid peroxidation mechanisms that have been implicated in the pathobiologicfeatures of retinal degenerations, including experimental retinal light damage,^65-67 age-related maculardegeneration,^67-69 anduveitis.⁷⁰ This hypothesis is supported furtherby a series of studies in our laboratory that have shown that (1) SLOS ratsare markedly more susceptible to retinal light damage than are normal albinorats,³⁹ (2) treatment of SLOS rats with a systemicantioxidant before intense light exposure can protect against retinal lightdamage,⁷¹ and (3) steady-state levels of lipidhydroperoxides in the retinas of SLOS rats are approximately 2-fold higherthan those in controls, and exposure to retinal light damage conditions producesan additional 3-fold elevation in retinal lipid hydroperoxides, with concomitantlygreater histologic damage than observed in light-exposed normal rats.⁷² In addition, cholesta-5,7,9(11)-trien-3β-ol(a compound generated by the decomposition of 7-hydroperoxy-cholesta-5,8-dien-3β-ol,a sterol hydroperoxide formed by the photo-oxidation of 7DHC) has been identifiedin the plasma of patients with SLOS.⁷³ Furthermore,by-products of 7DHC oxidation have been shown to retard the growth rate ofcultured rat embryos, a fact that may have particular significance with respectto the in utero developmental abnormalities associated with SLOS.⁷⁴

Regarding the ultrastructural abnormalities observed in the RPE of ratstreated with AY9944, we speculate that the observed accumulation of membranousinclusions and lipid deposits may be due to inhibition of lysosomal enzymesrequisite for phagosome digestion. Consistent with this finding, Sakuragawaet al⁷⁵ described a Niemann-Pick rodent modelproduced by AY9944 administration, with "lamellar inclusion bodies" appearingin the retina, lens, and other ocular and nonocular tissues, including gliaand neurons in the brain. Although the mechanism underlying these observationsis speculative and has been challenged by subsequent studies,^76,77 toour knowledge, the possibility that oxysterols derived from 7DHC are involved,either primarily or secondarily, has not been proposed or examined by otherresearchers. Studies are currently under way in our laboratory to test thishypothesis directly.

Finally, in addition to offering an experimentally accessible systemto study the mechanisms underlying the retinal degeneration and electrophysiologicdysfunction associated with SLOS, this animal model provides a valuable toolfor examining possible therapeutic interventions, such as dietary cholesterolsupplementation. This may allow further design optimization of therapeuticconditions (eg, combined cholesterol-antioxidant regimens) to be used in clinicaltreatment trials in addition to those currently in progress.^27,78-81

Correspondence: Steven J. Fliesler, PhD, Department of Ophthalmology,Saint Louis University School of Medicine, 1755 S Grand Blvd, St Louis, MO63104-1540 (fliesler@slu.edu).

Submitted for publication October 22, 2003; final revision receivedJanuary 20, 2004; accepted January 20, 2004.

This study was supported in part by grant EY07361 from the US PublicHealth Service, Washington, DC (Dr Fliesler); by an unrestricted departmentalgrant from Research to Prevent Blindness, New York, NY (Dr Fliesler); andby funding from the Department of Veteran Affairs, Washington, DC (Dr Peachey).

This study was presented in part at the Annual Meeting of the Associationfor Research in Vision and Ophthalmology; May 5-10, 2002; Ft Lauderdale, Fla;the Xth International Symposium on Retinal Degeneration; September 30-October5, 2002; Burgenstock, Switzerland; the National Institutes of Health (NICDH)Symposium on Inborn Errors of Cholesterol Synthesis; November 14-15, 2002;Bethesda, Md; and the Fifth Scientific Symposium on Smith-Lemli-Opitz Syndrome;June 27, 2003; Denver, Colo.

We thank Ellen R. Elias, MD, Anne B. Fulton, MD, and Ronald M. Hansen,PhD, for helpful discussions and for providing the results of their ERG studyof patients with SLOS before publication.