Proc. Natl. Acad. Sci. USAVol. 91, pp. 974-978, February 1994Neurobiology

Apoptotic photoreceptor cell death in mouse models ofretinitis pigmentosaC. PORTERA-CAILLIAU*, C.-H. SUNGt4, J. NATHANS*tf§, AND R. ADLER*§¶Departments of fOphthalmology, tMolecular Biology and Genetics, and *Neuroscience, and *Howard Hughes Medical Institute, Johns Hopkins UniversitySchool of Medicine, Baltimore, MD 21287-9257

Communicated by Lubert Stryer, October 19, 1993 (received for review August 2, 1993)

ABSTRACT Retinitis pnentsa (RP) is a group of in-herited human disease In which photoreceptor deerationleads to visual loss and eventually to blindness. Althoughmutatins In the rhodopsin, peripherin, and cGMP phpho-diesterase genes have been Idenf in some forms of RP, Itremains to be determined whether these mutations lead tophotoreceptor cell death o h necrotic or apoptotic mech-anisms. In this paper, we report a test of the hypothesis thatphotoreceptor cell death occurs by an apoptotic me sm inthree mouse models of RP: retinal degeneration slow (rds)caused by a peripherin mutation, retinal degeneration (rd)caused by a defect in cGMP phospiesterase, and tsgenicmice carrying a rhodin Q344ter mutation responsible forantosomal dominant RP. Two complementary techniques wereused to detect apop internueooml DNA frag-mentation: agarose gel electrophoresis and in situ labeling ofapoptotic cells by te al dUTP nick end labeling. Bothmethods showed extensive apoptosis of photoreceptors in allthree mouse models of retinal degeneration. We also show thatapopttc death occurs in the retina during normal develop-ment, that different mhanismscan cause photo-receptor death by activating an intrinsi death program in thesecells. These digs raise the possbility that retinal degener-ations may be slowed by interfering with the apoptotic mech-anism Itsef.

In humans, mutations in any of several genes encodingphotoreceptor-specific proteins have been shown to causeretinitis pigmentosa (RP), a disease characterized by loss ofphotoreceptor function and progressive degeneration of pho-toreceptor cells, eventually leading to blindness (1). Muta-tions in the rhodopsin gene are found in m25% of patientswith autosomal dominant RP (2-4); mutations in the geneencoding peripherin rds, an integral membrane protein that islikely to be involved in maintaining the structural integrity ofthe outer segment discs, are associated either with autosomaldominant RP (5, 6) or with a progressive macular degenera-tion (7, 8); finally, mutations in the gene encoding the (3subunit of cGMP phosphodiesterase have recently beenidentified in patients with autosomal recessive RP (9). Mousemodels of RP include the spontaneous mutants rds (retinaldegeneration slow; ref. 10) and rd (retinal degeneration; ref.11), in which photoreceptor cell death is triggered by nullmutations in the genes encoding peripherin/RDS (12) and theP subunit ofcGMP phosphodiesterase (13, 14), respectively.More recently, transgenic mouse models of dominant RPhave been constructed based on the observed humanrhodopsin mutations (refs. 15 and 16; C.-H.S. and J.N.,unpublished data). While defects in each of these geneswould be expected to affect phototransduction and/or thestructure of the outer segment, the pathway leading from theprimary defect to photoreceptor cell death is presently un-

known. One possibility is that the chemical alterations trig-gered by these mutations could cause cell death through anecrotic mechanism (i.e., toxic alterations of cellular homeo-stasis and/or disruptions ofplasma membrane integrity lead-ing to cell lysis). It has been proposed, however, that RP andrelated retinal degenerations could occur through reactiva-tion ofdevelopmental or physiological cell death mechanisms(17), a possibility also raised for other pathological neuronaldegenerations such as amyotrophic lateral sclerosis andAlzheimer and Parkinson diseases (18).

Cell death is a normal and necessary phenomenon duringembryonic development in general (19, 20) and in the nervoussystem in particular (21). Developmental cell death has beenfound to affect z50% ofthe cells that are generated in severalneuronal populations that have been analyzed in detail, suchas ciliary ganglion neurons, retinal ganglion cells, and spinalcord motor neurons (21); it occurs predominantly at the timeof synaptogenesis between neurons and their postsynaptictargets. It is generally accepted that, as part of their differ-entiation, developing neurons acquire an intrinsic capacity todie, which is normally inhibited by neurotrphic and otherfactors (21). Pathologic neuronal degenerations in the adultcould therefore result from abnormalities in the mechanismsresponsible for suppressing the expression of the intrinsicdeath program of the cells.

Recently, experimental tests of these hypotheses havebecome feasible due to the discovery that physiological celldeath usually occurs through apoptosis, which can be dis-tinguished experimentally from other forms of physiologicalcell death and from necrotic cell death (22, 23). Apoptosisaffects isolated cells, rather than patches of tissue, and is notaccompanied by inflammation; apoptotic cells are usuallyphagocytosed by adjacent cells, frequently without apparentinvolvement ofcirculating macrophages. A hallmark ofapop-tosis, which is not observed in necrotic cell death, is inter-nucleosomal DNA fragmentation, detectable by the appear-ance of a characteristic DNA ladder on agarose gel electro-phoresis (24). The fragmented DNA can also be detected byin situ labeling of apoptotic cell nuclei with a terminaldeoxynucleotidyltransferase (TdT)-mediated incorporationofbiotinylated nucleotides into the 3' ends ofDNA fragments[terminal dUTP nick end labeling; TUNEL (25)]. In thispaper, we report a test of the hypothesis that photoreceptordegenerations of genetic origin occur through an apoptoticmechanism, as defined by DNA fragmentation analysis.

MATERIALS AND METHODSExperimental Anias. Wild-type C57BL/6J mice were

purchased from The Jackson Laboratory or Harlan-Sprague-Dawley. Wild-type C3H, C3H rd/rd and C3H rds/

Abbreviations: RP, retinitis pigmentosa; TdT, terminal deoxynu-cleotidyltransferase; TUNEL, terminal dUTP nick-end labeling; RT,room temperature; P10, etc., postnatal day 10, etc.ITo whom reprint requests should be addressed at: The Wilmer EyeInstitute, Johns Hopkins University School of Medicine, 519Maumenee, Baltimore, MD 21287-9257.

974

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 26

, 201

9

Proc. Natl. Acad. Sci. USA 91 (1994) 975

rds mice were gifts of R. G. Foster (University of Virginia),J. Bennett (University ofPennsylvania), and S. Sanyal (Eras-mus University, Rotterdam). For construction of Q344terrhodopsin transgenic mice, an 11-kb mouse genomic DNAfragment containing the rhodopsin gene was engineered tocontain a stop codon in place ofthe Gln-344 codon (Q344ter),corresponding to a previously characterized humanrhodopsin gene mutation (4). Three independent lines oftransgenic animals carrying the Q344ter rhodopsin gene weregenerated by standard techniques (26). A detailed descriptionof these transgenic animals will be reported elsewhere.DNA Nick End Labeling by the TUNEL Method. Eyes were

fixed overnight at 4TC in phosphate-buffered saline (PBS)with 4% paraformaldehyde and embedded in paraffin. Four-micrometer-thick microtome sections were mounted onslides pretreated with Vectabond (Vector Laboratories) andwere subsequently deparaffinized by heating at 600C for 30min and washing twice in xylene for a total of 10 min. Thesections were then rehydrated through a graded series ofalcohols and double distilled water (ddH2O).The TUNEL technique was carried out as described (25)

with some modifications. Tissue sections were treated withproteinase K (20 pg/ml) in 10 mM Tris-HCI (pH 8.0) for 15min at room temperature (RT) and washed four times for 2min in ddH2O. Endogenous peroxidases were inactivated byincubating the sections for 5 min in 3% H202 at RT and thenwashing three times in ddH2O. Sections were preincubatedfor 10 min at RT in TdT buffer (30 mM Tris HC1, pH 7.2/140mM sodium cacodylate/1 mM cobalt chloride), and incu-bated for 1 hr at 37°C with 25-50 j1 ofTdT buffer with 0.5 unitof TdT per p1 and 40 ,uM biotinylated 16-dUTP in a moistchamber. The reaction was stopped by transferring thesections to 2x SSC buffer (300 mM NaCl/30 mM sodiumcitrate) for 15 min at RT. The sections were washed for 5 minin PBS and blocked in 2% bovine serum albumin in PBS for10 min at RT. After rinsing in ddH2O, the sections were againwashed in PBS and incubated for 1 hr at 37°C in VectastainABC peroxidase standard solution (Vector Laboratories),rinsed twice in PBS, and stained for 30-60 min at RT usingaminoethylcarbazole as a substrate. After the developingreaction was stopped with water, the sections were cover-slipped in Aqua-Poly/Mount (Polysciences). Positive con-trols were incubated with DNase I (1 pg/ml) in TdT buffer for10 min at RT before the incubation in biotinylated nucleo-tides. DNase, RNase, TdT, and biotin 16-dUTP were pur-chased from Boehringer Mannheim.

Quantitative Analysis. Four-micrometer-thick retinal sec-tions that included the ora serrata and the optic nerve (27)were stained by TUNEL and used for quantitative analysis.The number of labeled cells per section in each nuclear layerwas plotted for each time point, averaging the counts of fourto eight sections. One-third of all the slides (80/232 slides)were scored by two investigators, with 4.5% average vari-ability.DNA Gel Electrophoresis. Retinas were dissected away

from the retinal pigment epithelium and other ocular tissuesin calcium/magnesium-free Hanks' balanced salt solutionbuffer at 4°C. Five to seven retinas were pooled together,homogenized gently in 2 ml of extraction buffer (0.1 mg ofproteinase K per ml/100 mM NaCl/10 mM Tris HCl, pH8.0/25 mM EDTA, pH 8.0/0.5% SDS), and incubated over-night at 50°C in the same extraction buffer. Samples werephenol/chloroform (1:1) extracted, and the DNA was ethanolprecipitated and resuspended in TE buffer (10 mMTris HCl/l mM EDTA, pH 8.0). The DNA was then treatedwith 100 pg of DNase-free RNase per ml (Boehringer Mann-heim) for 1 hr at RT and incubated overnight in the presenceof proteinase K (0.1 mg/ml) at 37TC. The DNA was reex-tracted with phenol/chloroform, ethanol precipitated, andresuspended in TE buffer. Approximately 1-5 ug of DNA

was fractionated by 1.9%o agarose gel electrophoresis andstained with 0.05 pg of ethidium bromide per ml. For South-ern blot hybridization, genomic DNA was resolved on a 1.1%agarose gel, blotted onto GeneScreenPlus (DuPont), andhybridized with radiolabeled mouse genomic DNA.

RESULTSIn an initial experiment, we investigated whether the previ-ously described developmental cell death in the retina ofwild-type mice (27) involves apoptosis. The TUNEL methodspecifically labeled individual apoptotic nuclei in the gan-glion, inner nuclear, and photoreceptor cell layers of theretina during the first 2 weeks of development but not duringadulthood (Fig. 1B vs. Fig. 1C). TUNEL-positive cellsalways appeared surrounded by unstained cells, and therewere no indications of aggregation of apoptotic cells intomulticellular clusters. Comparative analysis of retinas atdifferent stages of development showed distinct temporalpatterns ofTUNEL labeling in the three cell layers (Fig. 2A),which correlated well with the patterns of pyknotic nucleidescribed previously by conventional histological techniques(27). We also observed the electrophoretic ladder character-istic of apoptosis when DNA isolated from postnatal day 10(P10) control retinas, at the peak of TUNEL labeling, wasresolved on a 1.1% agarose gel, blotted, and probed with totalmouse DNA (data not shown). In this case, the ladder ofbands was at or below the limit of detection by ethidiumbromide staining.To determine whether apoptotic cell death occurs in ge-

netically determined retinal degenerations, we studied threedifferent mouse models of RP. In rd/rd mice, rod photore-ceptor degeneration begins during the second postnatal weekand causes a rapid loss of photoreceptors that is essentiallycomplete by the end of the first month of postnatal life (11).During the first 12 postnatal days, the time course of appear-ance of TUNEL-labeled cells in the three nuclear layers inthe rd/rd retina resembled that seen in wild-type mice (seeabove). During the third week of postnatal life, however, therd/rd retina showed a dramatic increase in cell death that wasconfined to the photoreceptor layer (Fig. 2B). This is illus-trated in Fig. 1D, which shows the retina ofa P15 rd/rdmouse5 days after the onset of rapid photoreceptor degeneration(11). As in the normal developing retina, and in the otherretinal degeneration mouse models described below, noindications of aggregation of TUNEL-positive cells wereobserved. There was no increase in the number of apoptoticcells in other layers of the retina compared to the normalcontrols (Fig. 2B), in agreement with the previously reportedspecificity of photoreceptor degeneration in the rd/rd mouse(11). TUNEL analysis of retinas from adult rd/rd mice,obtained after degeneration of the rod photoreceptors wascomplete, showed only occasional labeled nuclei. Uponelectrophoresis in a 1.9%o agarose gel and ethidium bromidestaining, a ladder of fragments at intervals of 180-190 bp wasseen with DNA samples from the retinas of P15 rd/rd mice(Fig. 3A), but not from P19 wild-type mice (Fig. 3C) or P30rd/rd mice (Fig. 3B).

In rds/rds mice, photoreceptor degeneration begins afterthe second week of postnatal development and continues ata slow rate until all rod photoreceptors are lost at approxi-mately 1 year of age (10). Using the TUNEL method weobserved 10-20 labeled photoreceptor nuclei per 4-pm-thickretina section during the third week of postnatal development(Figs. 1F and 2C). The peak of photoreceptor degenerationwas significantly higher than that in age-matched controlretinas, where very few labeled cells were seen, but was notas great as that for the rd/rd mice (compare Fig. 2 B and C).In 4-, 6-, and 8-week-old rds/rds animals, we observedbetween 5 and 10 labeled photoreceptor nuclei per section, as

Neurobiology: Portera-Cailliau et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 26

, 201

9

976 Neurobiology: Portera-Cailliau et al.

- .

..W . .': ; ':. : I,|,

, A

..

....

D .. :.:is .....

FIG. 1. In situ labeling by the TUNEL method of cells undergoing apoptosis in the mouse retina during development and in retinaldegenerations. (A) Positive control retina treated with DNase I showing all nuclei labeled by TUNEL. OS, outer segment; ONL, outer nuclearlayer; INL, inner nuclear layer; GCL, ganglion cell layer. (B) Negative control retina from an adult C57BL/6J mouse showing an absence oflabeled nuclei by TUNEL. (C) C57BL/6J normal retina at P6 during the early period of developmental cell death in the retina. A small numberofTUNEL-labeled nuclei were observed in the three nuclear layers. One TUNEL-labeled cell is shown in each of the three nuclear layers. (D)rd/rd retina at P15 showing many TUNEL-labeled nuclei in the ONL but not in other layers of the retina. (E) Retina from a rhodopsin Q344tertransgenic mouse at P19 showing TUNEL-labeled nuclei exclusively in the ONL, with fewer nuclei labeled than in the rd/rd mouse. (F) Retinaofa rds/rds mouse at P23 showing an occasional labeled nucleus in the ONL. The different appearance ofTUNEL-positive cells in C-Fcomparedto DNase-treated cells in the positive control (A), and between cells in different layers ofthe positive control, is probably due to different degreesof chromatin condensation.

compared to none in age-matched retinas of control or rd/rdmice. No labeled cells were observed in other cell layers inthe retina in any of the older rds/rds animals (Fig. 2C). Byethidium bromide staining, a DNA ladder was observed insamples from rds/rds retinas obtained at P17 retinas (Fig.3C), as well as at P15 and P16 (data not shown).As a third model of RP, we investigated several lines of

transgenic mice carrying a mouse rhodopsin gene with aQ344ter mutation, a mutation in the human rhodopsin geneassociated with autosomal dominant RP (4). In three inde-pendent transgenic lines, the rod photoreceptors degenerateduring the first several months of postnatal life, with somevariability among littermates in the time course of photore-ceptor loss as determined by conventional histologic methods(C.-H.S. and J. N., unpublished data). In these three lines,the ratio of transcripts derived from the transgene to thosederived from the endogenous rhodopsin gene are 0.8, 1.3, and1.5 as determined by a quantitative reverse transcriptase/PCR assay. Immunostaining with an antibody specific for theQ344ter rhodopsin shows that the Q344ter protein accumu-lates exclusively in the photoreceptor layer in each of thelines (C.-H.S. and J. N., unpublished data). To quantitate thelevel of apoptotic cell death, we determined the number ofTUNEL-labeled nuclei in four transgenic littermates at P19and in a single transgenic animal at P14. Both control andtransgenic retinas showed a similar low number of labeled

nuclei in the inner nuclear and ganglion cell layers at P14 andP19. However, the transgenic retinas showed a large excessoflabeled nuclei in the photoreceptorlayer (Figs. lEand 2D).A similar excess of TUNEL-labeled photoreceptors wasobserved in two littermates from a second transgenic line atP24 (data not shown). The variability in the number ofTUNEL-labeled nuclei observed among transgenic litter-mates is likely to reflect the variability in the time course ofphotoreceptor loss noted above. The apoptotic nature of thephotoreceptor death in these transgenic animals was con-firmed by the presence of a DNA ladder upon agarose gelelectrophoresis and ethidium bromide staining ofDNA sam-ples from P19 retinas (Fig. 3B).

DISCUSSIONApoptotic Cell Death During Retinal Deveopment. The

most detailed description ofdevelopmental neuronal death inthe mouse retina was provided by Young (27), who quanti-tated the frequency of pyknotic nuclei in various retinallayers during the first weeks of postnatal life. He observedpyknotic nuclei not only in the ganglion cell layer, wheredevelopmental neuronal death has been well documented inseveral species (28), but also in the inner nuclear and thephotoreceptor cell layers. In the latter, a peak of cell deathwas observed around P7-P8; a few pyknotic nuclei could be

.--.'W"=AzWA

Proc. Nad. Acad Sci. USA 91 (1994)

Dow

nloa

ded

by g

uest

on

Dec

embe

r 26

, 201

9

Neurobiology: Portera-Cailliau et al. Proc. Nati. Acad. Sci. USA 91 (1994) 977

postnatal age (days)

LO 0 LO 0 0pO C(day n

postnatal age (days)

c

0U,

U)

0)

.0

-j

Ln

zI-

U) uO 0 ef 0 0 0

EL V- Ir- 04 CO) (0X. EL 0. CL 0. 0.

postnatal age (days)c0

0

11

0..C.,0

0

U)=

-I

z

I-

CO)RCD-oU C.)~I

CD C"

FIG. 2. Quantitation ofthe time course ofapoptotic cell death in control and retinal degeneration mice. (A) Control C57BL/6J mice. A similarpattern of TUNEL-labeled cells was seen with a C3H-derived control strain in which the rd gene was wild type. (B) rd/rd mice. (C) rds/rdsmice. (D) Rhodopsin Q344ter transgenic and control mice at P14 and P19. Tg1 to 4, transgenic littermates. Error bars represent SDs.

seen through the second week, but they were essentiallyundetectable thereafter. Although there were minor quanti-tative differences between our results and Young's, probablydue to the use of different mouse strains, our studies byDNAgel electrophoresis and the TUNEL technique are consistentwith Young's observations and support a model in which

A B0

-l C

0. 0C

M CMM

C 1bIL

0

0.

M

4,

_ ..

FIG. 3. Genomic DNA ladders from mouse retinas. (A) P15 rd/rdmouse. (B) P30 rd/rd mouse (lane 1) and P19 rhodopsin Q344tertransgenic mouse (lane 2). (C) P17 rds/rds mouse Oane 1) and P19control mouse (lane 2). Ethidium bromide-stained gels are shown asnegative photographic images. Lanes M, DNA markers (multiples of123 bp).

developmental cell death occurs through an apoptotic mech-anism in all layers of the retina, including the photoreceptorlayer.

Apoptotic Cell Death in Mouse Models of RP. The principalresult of this study is that photoreceptor cell death occursthrough an apoptotic mechanism in three different mousemodels of genetically determined retinal degeneration. In allthree cases, the appearance of large numbers of TUNEL-positive cells began and/or extended beyond the stage atwhich developmental neuronal death reaches a peak innormal animals and was restricted to the photoreceptor layer.The apoptotic nature of cell death in these retinal degener-ation mouse models is supported by two lines ofexperimentalevidence. Internucleosomal DNA fragmentation was ob-served by agarose gel electrophoresis as described for apop-totic cell death in other tissues (23, 24). TUNEL-positivecells were scattered throughout the retina and were notclustered in large patches, as would be the case if cell deathoccurred through a necrotic mechanism in these animals (23).It is known that apoptotic cells are quickly removed throughphagocytosis by neighboring cells in the same tissue (23), andthis may explain the apparent lack of accumulation ofTUNEL-positive cells in retinas examined at increasinglylonger times after the onset of degeneration. These observa-tions support the hypothesis that photoreceptor cells have anintrinsic death program that is tightly controlled under normalcircumstances (17) but that can be reactivated by the effectsof qualitative and/or quantitative abnormalities in photore-ceptor proteins such as rhodopsin, peripherin/RDS, orcGMP phosphodiesterase. It seems reasonable to supposethat an analogous activation of the apoptotic program occursin the diseased human retina.

C0

0CO

(n

U

0)

._(0-l

w

I-

z

C

C.)U)

U)

C.)

'zI-

Dow

nloa

ded

by g

uest

on

Dec

embe

r 26

, 201

9

978 Neurobiology: Portera-Cailliau et al.

The mechanisms through which the mutations studied inour experiments cause an activation ofthe apoptotic programremain to be elucidated. Uncertainties include the degree towhich similar molecular mechanisms operate in the threemodels of retinal degeneration and in normal embryonicdevelopment. In any event, our results suggest the possibilitythat photoreceptor degenerations of genetic origin could beslowed by interfering with the apoptotic mechanism itself,even if the mutant genes remain within the cells (29, 30).Because the degenerative changes in RP occur over decades,even modest decreases in the rate of cell death could signif-icantly increase the number of years of useful vision. Thiscould be particularly important in those cases in which theprimary molecular defect is confined to rod photoreceptorsand a cone degeneration occurs as an apparent secondaryresponse to the loss of rods. Treatments that interfere withthe apoptotic mechanism could therefore have importantimplications for the design of therapeutic strategies for hu-man RP and related blinding disorders, for which no treat-ments are currently available.

Note added in proof. After submission of this manuscript, Chang etal. (31) reported patterns of apoptotic photoreceptor cell death inmouse models of retinitis pigmentosa similar to those reported here.

We are grateful to Drs. J. Bennett, R. G. Foster, and S. Sanyal forexperimental animals, and to D. Golembieski for secretarial assis-tance. This work was supported by National Eye Institute GrantEY5404 and a Senior Investigator Award from Research to PreventBlindness, Inc. (R.A.), and the National Retinitis Pigmentosa Foun-dation and the Howard Hughes Medical Institute (C.-H.S., J.N.).

1. Heckenlively, J. R. (1988) Retinitis Pigmentosa (Lippincott,Philadelphia).

2. Dryja, T. P., McGee, T. L., Reichel, E., Hahn, L. B., Crow-ley, G. S., Yandell, D. W., Sandberg, M. A. & Berson, E. L.(1990) Nature (London) 343, 364-366.

3. Dryja, T. P., Hahn, L. B., Cowley, G. S., McGee, T. L. &Berson, E. L. (1991) Proc. Natl. Acad. Sci. USA 88, 9370-9374.

4. Sung, C.-H., Davenport, C. M., Hennessey, J. C., Maumenee,I. H., Jacobson, S. G., Heckenlively, J. R., Nowakowski, R.,Fishman, G., Gouras, P. & Nathans, J. (1991) Proc. Natl.Acad. Sci. USA 88, 6481-6485.

5. Farrar, J. L., Kenna, P., Jordan, S. A., Kumar-Singh, R.,Humphries, M. M., Sharp, E. M. & Sheils, D. M. (1991) Na-ture (London) 354, 478-480.

6. Kajiwara, K., Hahn, L. B., Mukai, S., Travis, G. H., Berson,E. L. & Dryja, T. P. (1991) Nature (London) 354, 480-483.

7. Nichols, B. E., Sheffield, V. C., Vandenburgh, K., Drack,A. V., Kimura, A. E. & Stone, E. M. (1993) Nature Genet. 3,202-207.

8. Wells, J., Wroblewski, J., Keen, J., Inglehearn, C., Jubb, C.,Eckstein, A., Jay, M., Arden, G., Bhattacharya, S., Fitzke, F.& Bird, A. (1993) Nature Genet. 3, 213-218.

9. McLaughlan, M. E., Sandberg, M. A., Berson, E. L. & Dryja,T. P. (1993) Nature Genet. 4, 130-134.

10. Sanyal, S., Hawkins, R. K. & Zilmaker, G. H. (1988) Curr. EyeRes. 7, 1183-1190.

11. LaVail, M. M. (1981) Invest. Ophthalmol. Visual Sci. 21,630-657.

12. Travis, G. H., Brennan, M. B., Danielson, P. E., Kozak,C. A. & Sutcliffe, J. G. (1989) Nature (London) 338, 70-73.

13. Bowes, C., Li, T. S., Danciger, M., Baxter, L. C., Applebury,M. L. & Farber, D. B. (1990) Nature (London) 347, 677-60.

14. Pittler, S. J. & Baehr, W. (1991) Proc. Natl. Acad. Sci. USA 88,8322-8326.

15. Olsson, J. E., Gordon, J. W., Pawlyk, B. S., Roof, D., Hayes,A., Molday, R. S., Mukai, S., Cowley, G. S., Berson, E. L. &Dryja, T. P. (1992) Neuron 9, 815-830.

16. Naash, M. I, Hollyfield, J. G., Al-Abaidi, M. R. & Baehr, W.(1993) Proc. Natl. Acad. Sci. USA 90, 5499-5503.

17. Adler, R. (1986) in The Retina: A Model for Cell BiologyStudies, eds. Adler, R. & Farber, D. (Academic, Orlando, FL),Part I, pp. 112-150.

18. Appel, S. (1981) Ann. Neurol. 10, 499-505.19. Glucksman, A. (1951) Biol. Rev. 26, 59-86.20. Raff, M. C. (1992) Nature (London) 356, 397-400.21. Berg, D. K. (1982) in Neuronal Development, ed. Spitzer,

N. C. (Plenum, New York), pp. 297-331.22. Waring, P., Kos, F. J. & Mullbacher, A. (1991) Med. Res. Rev.

11, 219-236.23. Vaux, D. L. (1993) Proc. Nat!. Acad. Sci. USA 90, 786-789.24. Wyllie, A. H. (1980) Nature (London) 284, 555-556.25. Gavrieli, Y., Sherman, Y. & Ben-Sasson, S. A. (1992) J. Cell

Biol. 119, 493-501.26. Hogan, B., Constantini, F. & Lacy, E. (1986) Manipulating the

Mouse Genome: A Laboratory Manual (Cold Spring HarborLab. Press, Plainview, NY).

27. Young, R. W. (1984) J. Comp. Neurol. 229, 362-373.28. Cunningham, T. J. (1982) Int. Rev. Cytol. 74, 163-186.29. Koike, T., Martin, D. P. & Johnson, E. M. (1989) Proc. Nat!.

Acad. Sci. USA 86, 6421-6425.30. Bissonnette, R. P., Echeverre, F., Mahboubi, A. & Green,

D. R. (1992) Nature (London) 359, 552-554.31. Chang, G.-Q., Hao, Y. & Fong, F. (1993) Neuron 11, 595-605.

Proc. Nad. Acad. Sci. USA 91 (1994)

Dow

nloa

ded

by g

uest

on

Dec

embe

r 26

, 201

9