Phenotypes of Stop Codon and Splice Site Rhodopsin Mutations Causing Retinitis Pigmentosa

Transcription

1 Phenotypes of Stop Codon and Splice Site Rhodopsin Mutations Causing Retinitis Pigmentosa Samuel G. Jacobson,* Colin M. Kemp,* Artur V. Cideciyan,* Jennifer P. Macke,\ Ching-Hwa Sung\, and Jeremy Nathans^ Purpose. To understand the pathophysiology of retinitis pigmentosa caused by mutations in the rhodopsin gene that lead to truncation of the protein. Methods. Heterozygotes with the glutamine-64-to-ter (), the intron 4 splice site, and the glutamine-344-to-ter (Q344ter) mutations in the rhodopsin gene, representing families with at least three generations of affected members, were studied with clinical examinations and measurements of rod and cone sensitivity across the visual field, rod- and cone-isolated electroretinograms (ERGs), rod dark adaptation, and rhodopsin levels. Results. There was a range of severity of disease expression in each family, some heterozygotes having moderate or severe retinal degeneration and others with a mild phenotype. The mildly affected heterozygotes had normal results on ocular examination but decreased rod sensitivities at most loci across the visual field, abnormalities in rod-isolated ERG a- and b-waves, and reduced rhodopsin levels. Rod dark adaptation followed an approximately normal time course of recovery in patients with the mutation. Patients with the splice site or Q344ter mutations both had prolonged recovery of sensitivity, but the time course was different in the two genotypes. Conclusions. There is allele specificity for the pattern of retinal dysfunction in the, intron 4 splice site, and Q344ter rhodopsin mutations. The pattern of dysfunction in all three mutations suggests the mutant opsins interfere with normal rod cell function, and there is subsequent rod and cone cell death. Invest Ophthalmol Vis Sci. 1994;35: XVetinitis pigmentosa (RP) is a genetically heterogeneous group of retinal degenerations, some of which are caused by mutations in the gene encoding rhodopsin. 1 Most of the rhodopsin gene mutations responsible for RP are point mutations or small deletions, and all but one cause autosomal dominant RP (adrp) (for example, refs. 2-6). The exception is a stop codon mutation, glutamic acid-249-to-ter (E249ter), recently reported as a putative null allele that causes autosomal From the *Department of Ophthalmology, University of Miami School of Medicine, Batcom Palmer Eye Institute, Miami, Florida, and the ^Departments of Molecular Biology and (ienetics, mid Neuroscience, Howard Hughes Medical Institute, Johns Hopkins University School of Medicine, Baltimore, Maryland. Supported in part by Public Health Service research grant EY05627 (SGJ); the National Retinitis Pigmentosa Foundation, Inc., Baltimore, Maryland; The Chatlos Foundation, Inc., Longwood, Florida; and the Howard Hughes Medical Institute, Belhesda, Maryland. Dr. Jacobson is a Research to Prevent Blindness Dolly Green Scholar. Submitted for publication October I, 1993; revised November 5, 1993; accepted November 12, Proprietary interest category: N. Reprint requests: Dr. Samuel G. Jacobson, Bascom Palmer Eye Institute, 1638 N.W. 10th Avenue, Miami, FL recessive RP (arrp). Heterozygotes with the E249ter rhodopsin mutation had normal clinical examinations but mild rod photoreceptor-mediated functional disturbances. 7 This finding of null alleles carried in single dose in apparently unaffected individuals 7 prompted the hypothesis that the rod photoreceptor can remain healthy with only half the normal level of wild-type rhodopsin, whereas it cannot in the presence of abnormal rhodopsin due, for example, to a missense mutation. 8 A recent search for rhodopsin mutations in 282 patients with RP 6 revealed two families, one with patients heterozygous for a stop codon mutation, glutamine-64-to-ter (), and another with patients heterozygous for the intron 4 splice site mutation guanosine 43!<;> -to-thymidine, a mutation described previously in a possible carrier of arrp. 7 These potential null alleles, however, were found in families with RP that had at least three generations of affected members, 6 indicating that these alleles are not innoculiivcsiigativc Ophthalmology & Visual Science, April Vol. 35, No. 5 Copyright- Association for Research in Vision and Ophthalmology 2521

2 2522 Investigative Ophthalmology 8c Visual Science, April 1994, Vol. 35, No. 5 ous in single dose and that these heterozygotes have adrp. To understand more about the pathophysiology of rhodopsin gene mutations in RP that could lead to truncation of the protein, we studied the functional phenotypes of patients with the mutation and the intron 4 splice site mutation. We also compared the patterns of disease expression in these mutations with the pattern in heterozygotes with a mild phenotype who carry another rhodopsin stop codon mutation, glutamine-344-to-ter (Q344ter), reported previously MATERIALS AND METHODS Subjects The 19 patients in this study were from three families. Thirteen of 14 patients from the family with the mutation (all but patient 3, Table 1), the three patients from the family with the intron 4 splice site mutation, and both patients with the Q344ter mutation had previously participated in molecular genetics investigations that determined that they were heterozygotes with these mutations. 2 6 All patients underwent ocular examination, and most patients had Goldmann kinetic perimetry, dark- and light-adapted static threshold perimetry, and full-field ERGs using a clinical protocol. The patients with relatively mild disease also had dark adaptometry, measurements of rod-isolated ERG a- and b-waves, and fundus reflectometry. Informed consent was obtained from the patients and from normal subjects involved in the study after the nature of the procedures had been explained fully. The research procedures were in accordance with institutional guidelines and with the Declaration of Helsinki. Visual Function Tests Static threshold perimetry in the dark- and lightadapted states was performed using techniques previously described For dark-adapted perimetry, 75 loci (12 grid) in the visual field were tested with 650 nm and 500 nm stimuli (target size V). Photoreceptor mediation at each locus was determined from the sensitivity differences between the two stimulus colors, and rod sensitivity losses were calculated based on 500 nm test results in comparison to normal mean values. For light-adapted (10 cd-rrt 2 white background) perime- TABLE l. Clinical Characteristics of the Patients Patient No. Generation No. Age (yr)/ Sex RE Visual Acuity* LE Kinetic Visual Field ExtentfX (V-4e/I-4e) Fundus Appearance^ " Intron 4 splice site IV IV III III II III II II II II 11 I I III II I 1 1/M J3/F 24/M 32/F 34/M 35/F 30/F 37/F 43/F 64/M 65/F 71/M 73/F 75/F 76/M 21/F 46/M 76/F 20/25 20/40 20/20 20/40 20/20 20/50 20/40 2/200 20/60 20/60 LP 7/200U 20/20 20/50** HM 20/20 20/40 20/20 20/100 20/50 LP 20/100 20/60 LP 2/200 20/20 20/20 NLPH 41/35 37/5 98/28 42/21 92/72 15/2 94/75 91/43 86/<I 55/<l U/U <I/U <l/<] u 31/U# 94/94 95/60 U N N N N = no abnormalities; = pigmentary retinopathy; HM = hand motions vision; NLP = no light perception; LP = light perception; U = immeasurable. * Best corrected visual acuity. t Similar in the two eyes, unless specified. X Expressed as a percent of normal mean; 2 SD below normal equals 90% for V-4e and 88% for I-4e. 19 Visit in Visit in Glaucoma. # Nonglaucomatous eye. ** Strabismic amblyopia.

3 Stop Codon Rhodopsin Mutations and RP 2523 try, the same 75 loci were tested with a 600 nm target. Cone sensitivity losses at each locus were determined by comparison with normal mean values. Dark adaptometry was tested with 500 nm and 650 nm stimuli (target size V) at 12 in the inferior visual field. Baseline dark-adapted thresholds were determined after at least 3 hours of dark adaptation on a day before exposure to any bright lights. A yellow bleaching light (wavelengths > 520 nm) was delivered with Maxwellian optics using a fundus camera (Carl Zeiss, Wetzlar, Germany); the 30 diameter field was centered on the test locus. For each patient, the recovery of sensitivity was measured after retinal exposures of 7.8, 6.9, and 6.3 log scot-td s. These exposures are expected to bleach about 99, 50 and 15% of the rhodopsin originally present, respectively. Further details of the method have been reported. 13 The time courses of dark adaptation in the patients were analyzed using a model shown to provide an accurate description of the kinetics of recovery of sensitivity in normals after adapting lights that bleach from as little as 1% to greater than 99% of the rhodopsin originally present Lamb's scheme postulates that the control of rod sensitivity results from the persistent presence of small amounts of R* (the activated form of photolyzed rhodopsin) after extinction of the adapting light. Lamb proposed that the R* is produced from relatively long-lived rhodopsin photoproducts Sj, via reverse reactions. The model thus consists of three sequential first-order reactions, each of which is weakly reversible and one of which saturates (i.e., becomes zero-order) after intense light adaptation: Rhodopsin light S 2 ^==± S k 21 k 32 where interconversion of S 2 and S 3 is rate limited, with a half-saturating value for S 2 of S 2sal. The model does not identify the individual photoproducts, which in principle could include one or more forms of phosphorylated opsin and/or opsin to which arrestin is bound. Though it does not have a comprehensive foundation of specified molecular reactions involving the rhodopsin molecule, it provides a relatively simple basis for describing quantitatively the kinetic abnormalities observed in several forms of RP, with an accuracy and level of detail that is unattainable using conventional mathematical schemes proposed for rod dark adaptation in the intact eye (for example, refs. 16 and 17). In particular, it enables various time domains (each associated with the relative abundance of one of the species S ls S 2, or S 3 ) during dark adaptation to be identified. As a result, the extent to which each of these is affected in patients with rhodopsin mutations can be individually assessed. Solutions of the set of first-order differential equations describing the model were obtained by numerical integration using the Runga-Kutta method, 18 and for each subject values for the parameters were obtained by minimizing the errors of the fits to the rod recovery data for all three bleaches. Full-field ERGs were performed using bipolar Burian-Allen contact lens electrodes and a computerbased system previously described. 19 ' 20 Suprathreshold stimuli were used to elicit a rod ERG (blue flash of 0.1 log scot-td *s, dark-adapted); a mixed cone and rod ERG (white flash of 5.4 cd s m~ 2, dark-adapted); and a cone flicker ERG (29 Hz white light flashes of 0.64 cd's«m~ 2, on a white background of 6.9 cd m~ 2 ). ERGs were also elicited in the dark-adapted state to different intensities of blue (Wratten 47B; Kodak, Rochester, NY) light flashes over a 3 log unit range (up to 1.8 log scot-td s). Waveforms were measured conventionally as follows: b-wave amplitude from baseline or the a-wave trough (when present) to the major positive peak; implicit time from stimulus onset to the major peak of the response; and for the cone flicker ERG, amplitude from negative to positive peak and timing to the positive peak. The Naka-Rushton equation [V = Vmax*I n /(I n + K n )] was fitted to the measured b-wave amplitudes from the intensity series to blue light flashes. In the equation, V is rod b-wave amplitude; V max, the amplitude at response saturation; I, the stimulus intensity; K, the intensity at half V m;ix ; and n, the exponent responsible for the slope of the function. Rod-isolated ERGs to high-intensity stimuli were performed using unipolar Burian-Allen contact lens electrodes and recording and analysis methods previously published. 21 In brief, pairs of scotopically matched waveforms to blue (Wratten 47A) and red (Wratten 26) flashes were digitally subtracted to give a cone ERG, that was then subtracted from the response to a photopically matched blue flash (double subtraction technique 22 ). Responses to a range of intensities from 2 log scot-td s to 4.5 log scot-td s in 0.3 log unit steps were recorded. The photoreceptor generated component of the rod-isolated ERG, PHI 23 or P3 24, was estimated by fitting a mathematical model to the leading edge of the a-waves in the intensity series. The model consisted of a family of delayed Gaussian functions of time and stimulus intensity 25 " 28 : t) = R r -exp[---i-<7-(t-t d ) 2 0) where R niax is the maximum response amplitude in nv; I, the energy of a brief flash in scot-td s; a, the sensitivity in scot-td" 1 s~ 3 ; t, the time after flash onset in sec-

4 2524 Investigative Ophthalmology 8c Visual Science, April 1994, Vol. 35, No. 5 onds; and t d, a brief time delay in seconds that approximates the initial stages of the transduction cascade as well as delays due to the recording apparatus. The sensitivity parameter a is equal to the product of k^, which is the number of isomerizations produced per rod per scot-td s of retinal illuminance, and A, which is the amplification constant in s" 2. The amplification constant A is the product of the rate of activated phosphodiesterase production per isomerized rhodopsin molecule, the rate of cgmp hydrolysis per activated phosphodiesterase, and the Hill coefficient governing the fraction of open channels. 25 " 27 The value of k^, is estimated to be approximately 5 in normal subjects. 25 Parameters of the P3 model (R max, o, and t d ) were determined in two steps. First, the waveforms were edited to make the pre-stimulus baselines coincide and were cut at the time when the b-wave intrudes. Next, the edited waveforms were used to find automatically (Matlab 4.0, The Math Works, Natick, MA) the two parameter values (a and t d ) that minimize the squared error between the model and the ensemble of waveforms. The parameter R max was set equal to the largest negative amplitude in the series. To permit independent comparisons of R nuix and a in patients and normal subjects, the response amplitude predicted by the P3 model was plotted against stimulus intensity for a fixed time. 2 '' For afixedtime after the time delay (t t d = T), the photoreceptor response shown in equation 1 reaches half-maximum response at the intensity 2 ln(2) 50% (2) On a graph of response amplitude versus log stimulus energy, a change in R max would correspond to a vertical scaling and a change in a would correspond to a horizontal shift. Imaging fundus reflectometry was performed with instrumentation and methods already described. 30 " Rhodopsin losses in the patients were determined by comparison of their double difference values (at 520 nm) with those from normal subjects at matched retinal locations. To study the relationship between the rhodopsin levels and rod-mediated sensitivity, dark-adapted static perimetric measurements with the 500 nm stimulus were made at 25 loci within the retinal region tested with fundus reflectometry RESULTS The schematic drawing of the rhodopsin molecule in Figure 1 shows the sites of the, intron 4 splice site, and Q344ter mutations. The mutation would encode a truncated protein missing six of the Q344ter G-T,bp4335 FIGURE l. Schematic drawing of the rhodopsin molecule. Amino acids are shown as circles. The sites of the mutations carried by the patients in this study are indicated. seven transmembrane domains, including the site of attachment of 11-cis retinal. 6 The substitution in the donor splice site of intron 4 could lead to an abnormal carboxy terminal region of the molecule. The Q344ter mutation encodes a protein missing the last five amino acids. 20 Table 1 shows some clinical characteristics of the patients in this study. In the family with the mutation, 12 of the 14 heterozygotes examined have ophthalmoscopic features of RP with attenuated retinal vessels, pigmentary retinopathy, and a waxy pale appearance to the optic nerve head. P5 has a normal ophthalmoscopic examination, and his sister, P7, has cystoid macular edema and only a few pigmentary changes in the peripheral retina. Visual acuities and kinetic visual fields range from normal or nearly normal to moderately or severely abnormal. In the family with the intron 4 splice site mutation, PI5 and PI6 have normal-appearing fundi, but PI7 has ophthalmoscopic evidence of an advanced stage of RP in both eyes. Visual acuities ranged from normal (PI5 in both eyes; PI6 in his non-amblyopic eye) to severely abnormal (PI 7 in her eye without glaucoma). Kinetic fields were normal in PI5, slightly subnormal in PI6, and reduced to a small central island in PI7. Records obtained from previous examinations of PI 7 indicated that Goldmann kinetic perimetry (V-4e target) 15 years earlier showed a central island and a temporal peripheral island separated by a nearly complete annular midperipheral scotoma; another field 5 years earlier showed only a small central island of vision. In our previous description of the phenotype of a family with RP caused by the Q344ter mutation, we noted that three siblings carrying the mutation had normal ocular examinations, normal visual acuities and kinetic fields, and abnormal rod function but normal cone function (patients 1 to 3 in ref. 10). In the present study, two of the patients, designated as PI8

5 Stop Codon Rhodopsin Mutations and RP 2525 and PI9 (representing patients 2 and 3, respectively, in ref. 10), were reexamined with further visual function tests to permit comparison with results obtained from the patients with mild phenotype from the families with and splice site mutations. Figure 2 shows results of kinetic and static perimetry in three family members with the mutation, representing different degrees of disease expression. P5 has a normal extent of visual field with kinetic perimetry using the V-4e target but a slightly reduced extent with the I-4e target (Table 1). There is rod sensitivity loss (mean loss, 11.3 db; SD 2.2 db) across most of the visual field. P4 has a kinetic visual field with reduced extent in the periphery with both target sizes (Table 1); rod sensitivity losses are far greater (mean of the 46 loci with measurable function, 32.9 db; SD 9.2 db) than in P5. The kinetic field of P2, the daughter of P4, using the V-4e target has a central island separated from an island in the temporal peripheral field by an incomplete annular midperipheral scotoma; with the I-4e target, the field is limited to only a central island. Rod sensitivity is measurable only centrally and in the temporal periphery and is reduced by between 2 and 3 log units at these loci (mean of the 22 loci with measurable function, 30 db; SD 11.2 db). Mean cone sensitivity losses across the visual field for P5 were 0 db (SD 2.1 db); mean of loci with measurable function for P4 were 9.1 db (SD 3.8 db; n = 30 loci) and for P2 were 6.0 db (SD 5.9 db; n = 15 loci). Figure 3 shows perimetric results in P7, the sister of P5, on two visits separated by about 7 years. In 1985, the patient had a normal extent of kinetic field with the V-4e target but a slightly reduced extent with the I-4e target (Table 1). Rod sensitivity loss at this time was 10.0 db (SD 4.8 db). In 1992, the kinetic field to the I-4e is more reduced in extent (Table 1), and rod sensitivity loss had increased (mean 17.6 db; SD 5.8 db). To determine if this progression of rod sensitivity PATIENT 5 PATIENT 4 PATIENT 2 KINETIC PERIMETRY ROD SENSITIVITY LOSS 0) 2, zuj o UJ N T N T N T ECCENTRICITY [deg] FIGURE 2. Kinetic perimetry (upper) and dark-adapted static threshold perimetry (lower) in three patients with the mutation. V (target area 64 mm 2 ) and I (target area 0.25 mm 2 ) targets at intensity 4-e (318 cd m~ 2 ) were used for kinetic perimetry. Results of static perimetry are displayed as gray scales of rod sensitivity losses. Gray scales have 16 levels representing 0 to 35 db (1 db equals 0.1 log units) sensitivity loss. White is 0 to 2 db loss, and black is greater than 35 db loss. Physiological blind spot is shown as a black square at 12 in the temporal field.

6 2526 Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5 PATIENT 7,1985 PATIENT 7,1992 KINETIC PERIMETRY ROD SENSITIVITY LOSS 0) 3^ > o E z UJ o UJ T N T N ECCENTRICITY [deg] FIGURE 3. Kinetic perinietry and dark-adapted static threshold perimetry in P7, a patient with the mutation, on two visits separated by about 7 years. The data are displayed as in Figure 2. loss affected some regions of the visual field more than others, we divided the field into three regions and calculated the average of rod sensitivity losses within these regions. At eccentricities <30, there was about 5 db loss between visits; between 30 and 60, there was nearly 10 db loss; and at eccentricities >60, about 7 db loss occurred. This suggests that in the 7-year interval between visits, the midperipheral field had more sensitivity loss than peripheral and central fields. Data for cone sensitivity across the visual field was available for only the later visit; mean cone sensitivity loss was 3.7 db (SD 4.1 db). Figure 4 shows results of kinetic and static perimetry in two mildly affected patients representing two generations with the intron 4 splice site mutation (PI5, PI6) and a patient with the Q344ter mutation (PI9). PI5 and PI9 both have normal kinetic fields, whereas PI6 shows a slightly reduced extent with the I-4e target (Table 1). All three patients show some rod sensitivity losses across the visual field. Mean rod sensitivity losses were as follows: PI5, 7.4 db (SD 2.0 db); P16, 7.8 db (SD 2.7 db); and P19, 9.9 db (SD 3.1 db). Cone sensitivity losses were as follows: PI 5, 0.5 db (SD 1.6 db); P16, 2.7 db (SD 2.0 db); and P19, 0.2 db (SD 1.7 db). The two other siblings of PI 9 showed similar results with dark- and light-adapted static perimetry to their sister. 10 Figure 5 shows dark adaptometry results in two representative normal subjects and six patients, two from each genotype. For the light-adapting exposure used, which is expected to bleach ~99% of the visual pigments originally present within the test area, 13 the normal subjects recovered completely to their baseline dark-adapted sensitivity levels after about 55 minutes in darkness. P5 and P7, who carry the mutation, showed a similar time course, with no delay of either the appearance of the rod recovery branch or of the attainment of their baseline sensitivities (which are reduced by 7 db and 9 db, respectively). PI 5 and PI6, who carry the splice site mutation, both had slower recovery than in normal subjects, requiring about 90 minutes to return to within 1 db of their pre-bleach

7 Stop Codon Rhodopsin Mutations and RP 2527 SPLICE SITE PATIENT 15 PATIENT 16 Q344ter PATIENT 19 KINETIC PERIMETRY 3^ >- H O c h- Z UJ o UJ s ROD SENSITIVITY LOSS r 1. N T T N T N ECCENTRICITY [deg] FIGURE 4. Kinetic perimetry and dark-adapted static threshold perimetry in two patients who carry the intron 4 splice site mutation (PI 5, PI6) and a patient with the Q344ter mutation (PI 9). The data are displayed as in Figures 2 and 3. sensitivity levels. Recovery of rod sensitivity in the two patients with the Q344ter mutation is also prolonged and, in each case, takes more than 2 hours to return to within 1 db of the pre-bleach level. The curves fitted to the time course of recovery of rod function in the patients and the normal subjects are derived from the scheme proposed by Lamb. 14 ' 32 The kinetic parameters used to fit each set of data are shown in Table 2. When there were only small differences between the kinetics of the recovery curves for patients with the same mutation, a single set of parameters was used to generate the curve that describes them. In the case of the patients with the splice site mutation, curves were fitted to the data of PI5 and PI6 individually; with one exception (k 2 i), all abnormalities were similar in the two patients. Although some of the parameters in the data from the patients with the Q344ter mutation are similar to those in the data from patients with the splice site mutation, there are substantial quantitative differences between the sets associated with the two genotypes. Electroretinography using a clinical protocol showed that rod ERGs, mixed cone and rod ERGs, and cone flicker ERGs were abnormal to varying degrees in the seven members of the family with the mutation who were tested. Patients 4, 6, 8, 12, and 14 had no detectable responses to any of the stimuli. P5 and P7 had rod b-waves with reduced amplitude and normal implicit times. Cone flicker amplitude was normal, but timing was delayed in these patients. Serial data on P7 showed further reduction in rod b-wave amplitude and greater prolongation of cone flicker timing between visits, separated by 7 years. In the family with the splice site mutation, ERGs in PI5 and PI6 showed reduced rod b-wave amplitudes and normal or slightly delayed timing; cone ERGs were normal. ERGs in PI7 were not detectable on an examination 15 years earlier. Results with these stimuli in the patients with the Q344ter mutation have been published. 10 They showed mainly rod amplitude abnormalities. Figure 6A shows the first 15 ms of the rod-isolated responses in a normal subject and in three patients, each representing a different genotype. These patients had the mildest disease expression among those examined in their family. The P3 model has been fitted to

8 v < *» 2528 Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No I NORMAL CD K n Q S5o o "S&S»n) oo OUil)lM>0«>-00 > 10 U) Z 20 LJJ CO *, SPLICE SITE I v, Q344ter PB h PB 0 TIME FIGURE 5. Dark adaptometry results after bleaching of 99% rhodopsin at 12 in the inferior field in two normal subjects (upper left); P5 (unfilled squares) and P7 (filled squares) with the mutation (upper right); PI5 (unfilled circles) and PI6 (filled circles) with the splice site mutation (lower left); and PI8 (unfilled triangles) and PI9 (filled triangles) with the Q344ter mutation (lower right). Each panel also includes curves illustrating thefitof a model for kinetic analysis of rod dark adaptation to the data from the normal subjects and each genotype, using the parameters given in Table 2. PB, pre-bleach or baseline dark-adapted sensitivity level. Note the compressed time-scale used for times greater than 1 hour after the bleach. mm 3h the responses from an intensity series. P5, representing the family with mutation, and PI8, representing the family with the Q344ter mutation, have lower maximum amplitude than the normal subject, whereas PI5, with the splice site mutation, has a response with amplitude closer to that of the normal. Table 3 lists the P3 model parameters (R max, a, t d ) for six patients, two from each family, and, for comparison, the mean and standard deviation for a group of normal subjects. R max offiveof the six patients (except PI5) fell outside the range of the normal subjects; a and t d were within the normal range for all six patients. TABLE 2. Parameters* Describing the Kinetics of Rod Dark Adaptometry Patient No. 5 7 Intron 4 splice site /~\O A A *. ^;o44ter Normal *»t " k k l3 2sat * Parameters were obtained using the model proposed by Lamb 14, k 10 was treated as invariant from normal in all cases. f ky values are in units of s" 1. X S 25a t is the half saturating concentration of S 2 -

9 Stop Codon Rhodopsin Mutations and RP ^ -300 LLJ Q -400 J 0 Q. 2 ** -100 B i 500 t; 400 < 200 I 100 NORMAL SPLICE SITE TIME [ms] L "" STIMULUS ENERGY [log scot Id s] FIGURE 6. (A) Rod-isolated a-waves to different stimulus intensities in a normal subject and patients with the (P5), the splice site (PI5), and the Q344ter (PI8) rhodopsin mutations. The smooth curves are a family of delayed Gaussian functions fitted to the leading edges of a-waves in the intensity series. The stimulus energies were O = 4.5, = 4.2, V = 3.9, T = 3.6, - 3.3, = 3.0, A - 2.7, = 2.4, 0 = 2.1 log scot-td s. (B) Graph of P3 model amplitude at the fixed time of 5 ms after time delay versus log stimulus energy. Solid lines represent the six patients whose data are in Table 3, and the dashed line represents the mean normal. Vertical lines correspond to the intensity producing half maximum response, I 50 %. Error bar on vertical axis is the mean normal R max 2 SD; bar on horizontal axis is mean normal I 50 % ± 2 SD. The relationship between R niax and a is shown in Figure 6B, which plots the P3 model amplitude at a fixed time versus the log stimulus energy for six patients, two from each genotype. The vertical lines denote the I 5()% values for the patients. In Figure 7 are graphs of rod ERG b-wave amplitudes at different intensities of blue light flashes in the dark-adapted state in six patients, two representatives of each genotype, compared to normal subjects. Both patients with the mutation show a reduced V nklx and an abnormal K. The serial data in P7 are notable in that they provide some information about the natural history of rod ERG change in the 2529 mutation; disease progression in this mutation appears to lead to more reduction of V niax and a further shift in K. The patients with the splice site mutation had abnormal V niax and K. One of the two patients with the Q344ter mutation also followed this pattern, whereas another fell just within the normal limits (outside the ± 1 SD range but inside the ± 2 SD range) for both V max and K. The parameters derived from the fitting of the Naka-Rushton equation to the rod ERG intensity series are given in Table 3. It is of interest that the ratio of b-wave V max to a-wave R niax is about 1.0 or greater for patients with and Q344ter mutations and for normal subjects. In the splice site mutation patients, however, the ratio of these parameters is about 0.5, and there were negative waveforms to all high-intensity stimuli. This suggests there is dysfunction not only at the rod outer segment but also at or proximal to the photoreceptor terminal region, such as has been recently demonstrated in patients with RP of unknown genotype. 21 Imaging fundus reflectometry was performed on P5 and P7 from the family with the mutation and on PI 5 and PI6 from the family with the splice site mutation. Figure 8 shows the relation between rhodopsin levels and psychophysically measured rod sensitivity losses in the four patients. In P5 and P7, measured pigment densities were considerably reduced from normal by a relatively constant amount. In PI5 and PI6, there was greater variation of densities within the measurement area. For all patients, the data points lie close to the line illustrating the predicted relationship for rod sensitivity losses caused by decreased light absorption as rhodopsin levels diminish. A similar pattern of results was found in patients with the Q344ter mutation. 10 DISCUSSION The ocular examination results in the patients heterozygous for the, intron 4 splice site, and Q344ter mutations in the rhodopsin gene showed that there was a range of severity of disease expression in each genotype. Some heterozygotes from each family had pigmentary retinopathy with reduced visual acuity and diminished visual field extent, whereas others had a normal ophthalmoscopic appearance with normal acuity and full kinetic fields. Rod-specific visual function test results indicated that even the patients who were apparently unaffected clinically had abnormal rod-mediated function. Dark-adapted perimetry has shown specific patterns of rod sensitivity loss, such as diffuse or altitudinal patterns, in previous studies of adrp patients with rhodopsin mutations. 10 ' 13 Regional retinal differences in disease severity were not discernible from the available data in the families with splice site and

10 2530 Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5 TABLE 3. Rod Isolated ERG Results A-wave* B-wave-f Patient No. Rmax (*V) 0" (scot-td~' -s~ 3 ) h (msec) Vmax logk (log scot-td -s) n 5 n7 Intron 4 splice site Q344ter Normals Mean ± SD NA ± NA ± NA ± ± ± ±0.13 NA = not available. * Parameters of the P3 model fit to the rod-isolated a-wave intensity series. f Parameters of the Naka-Rushton curve fit to rod-isolated b-vvave intensity series. + Visit in Visit in " n = 8 for a-wave; n = 57 for b-wave. Q344ter mutations. Serial measurements in one patient with the mutation suggested vulnerability of the midperiphery with disease progression, and this is consistent with the findings in more affected members of this family who had midperipheral scotomas and retained central and peripheral patches of rod function. A comparison of patients with the mildest phenotypes in the three genotypes indicates that, on average, patients with the mutation had the greatest degree of rod sensitivity loss and lowest levels of measurable rhodopsin, whereas those with the Q344ter mutation were intermediate, and splice site mutation patients had the least sensitivity loss and some of the higher levels of rhodopsin measured. The rod sensitivity losses in all patients were consistent with decreased probability of light absorption from the reduced levels of rhodopsin, as has been found in other rhodopsin mutations Thus, when dark adapted, these patients did not appear to have substantial quantities of photolyzed pigment products acting as a source of equivalent light in the photoreceptors, as has been proposed to occur in some forms of RP. 33 Rod ERG a-wave and b-wave analyses have been applied to patients with RP of unknown genotype , but the present study is the first to use this approach to interpret the waveforms from patients with rhodopsin mutations. The a-wave analysis in our patients showed that the maximum receptor response (R max ) was reduced compared to the normal mean result, and a-wave sensitivity (a) fell within our normal limits. The rod b-wave results showed varying amounts of reduction of V max and abnormalities in K. This ensemble of ERG findings is consistent with certain hypothesized underlying disease mechanisms, but not with others. For example, a disease process affecting the sensitivity of all rods across the retina in the same way, such as a decreased number of rhodopsin molecules in otherwise normal rods, 7 ' 8 would not explain the reduced a-wave and b-wave maximum responses and normal a-wave sensitivity. 24 ' 29 ' 37 A retina with well-functioning rods interspersed with nonfunctional receptors could lead to ERG findings such as we observed in our patients ' 37 It is also possible that the disease mechanism is more complex, and partially functional receptors may be contributing subnormally to the full-field response. 29 ' 37 Rod dark adaptation has been found to be abnormal in many different mutations of the rhodopsin gene that cause adrp, and kinetic analyses of the results have shown there can be specific abnormalities in the different genotypes. 38 Analyses of rod dark adaptation data in the present study indicate there are very different mechanisms of dysfunction in the three genotypes. The approximately normal time course of rod dark adaptation in patients with the mutation suggests that this function is essentially mediated by wild-type protein. However, the analysis indicates that at least one of the parameters, k 32, is abnormal, and by similar amounts in both P5 and P7 (Table 2). The in-

11 Stop Codon Rhodopsin Mutations and RP LU 100 Q Q_ LU ^ 400 cb o SPLICE SITE A Q344ter '85i '92 functional abnormalities (on the basis of this model) may thus be associated with relatively slow reactions, such as the binding of arrestin to bleached rhodopsin, and the reduction and removal of the au-trans retinal from the binding site. It has been suggested that the photoproducts involved in these reactions may play a role in setting the sensitivity of the rod. 34 ' 40 " 42 The rates of interconversions of one or more of them may depend on the integrity of the carboxy terminal region of rhodopsin. 45 In the splice site mutation, this region of the molecule is likely to be abnormal. Interestingly, both in patients with this mutation and in those with the Q344ter mutation, the value of S 2s;ll, the half-saturating concentration of S 2, is found to be about twice as large as normal, implying that the factor within the outer segment that limits the rate of this reaction after intense bleaching 1432 is present in abnormally high amounts relative to those of the expressed rhodopsin. Rod dark adaptometry in patients with the Q344ter mutation also showed a prolonged recovery of sensitivity, but this differed quantitatively from the abnormality in the splice site mutation. As expected from the extended period required for complete sensitivity recovery after a 99% bleach (Fig. 4), the kinetic -2-1 II 0T 1 STIMULUS ENERGY [log scot td s] FIGURE 7. ERG b-wave amplitude as a function of stimulus energy for normal subjects and the patients. {Top) Results from P5 (unfilled squares) and two visits for P7 (filled squares), both of whom have the mutation. (Bottom) Results from PI 5 (unfilled circles) and PI 6 (filled circles) with the splice site mutation, and PI 8 (unfilled triangles) and PI 9 (filled triangles) with the Q344ter mutation. Solid curves are the fits of the Naka-Rushton function to the patient data; arrows with symbols denote K for the patients. Dashed curve is a mean normal function; error bar on vertical axis is mean normal V, nax 2 SD; bar on horizontal axis is mean normal K ± 2 SD. 40 F crease in k 32, which is not seen in the splice site and Q344ter mutations, suggests the possibility of some factor that interferes with the control of rod sensitivity, at least during the later stages of recovery. 39 In patients with the intron 4 splice site mutation, unlike the mutation, there is prolonged recovery of sensitivity, which is reflected by the low values found in the kinetic analysis (Table 2) for k 34, the parameter that characterizes the last stage in the recovery of sensitivity (and is loosely identifiable with the regeneration of photo-activatable rhodopsin). Values were also abnormal for the other parameters associated primarily with the later stages of recovery. The % RHODOPSIN FIGURE 8. Relationship between rod sensitivity loss and rhodopsin levels in P5 and P7, who carry the mutation, and PI 5 and PI6, who carry the splice site mutation. Data from P5 and P7 are mean values for the central 10 X 10 rectangle of the measurement area; data points for PI 5 and PI6 are individual values from within the measurement area. The solid line describes the expected relationship if rod sensitivity loss was caused solely by the decreased probability of light absorption resulting from reduced levels of rhodopsin. Symbols as in Figure 7.

12 2532 Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5 analysis yielded values for k 23 and k 34 that were lower than those for patients with the splice site mutation (Table 2). The prolonged desensitization of the rod after bleaching may relate to defective reactions in the visual cycle of this mutant opsin because of the altered carboxy terminus of the molecule. 43 ' 44 The relationship between the results of noninvasive tests of visual function and the underlying photoreceptor pathophysiology is complex, but our findings in the mildly affected patients permit some speculation about the disease mechanisms in these three genotypes. The mutation, in theory, would be a functional null mutation. The measurable rhodopsin by fundus reflectometry and the rod-mediated function detected by psychophysics and electroretinography in these patients indicate that some rhodopsin (presumably only wild-type) has been synthesized, transported to the outer segment, and inserted into the disk membrane, and phototransduction does occur. A simple quantal catch model based on 50% of the rhodopsin molecules in each rod, which was used to explain the test results in other putative null mutations, 7 does not explain fully the results of the most mildly affected patients with the mutation. Of course, even these patients may already be at a later stage of this progressive retinal degeneration, which could have started with a 50% reduction in the number of rhodopsin molecules per rod. The unexpected finding of an abnormality in dark adaptation suggests that some factor, such as partial expression of the mutant protein or some aspect of the degenerative process of the disease, may be interfering with recovery of sensitivity of the wild-type protein after light activation. The intron 4 splice site mutation was also hypothesized to be a null allele based on examination of one heterozygote with this genotype whose clinical phenotype and rod function abnormalities were similar to those of heterozygotes with the E249ter mutation. 7 The finding of as much as 75% of normal rhodopsin levels in some retinal regions by fundus reflectometry and the abnormality in rod dark adaptation would not be expected from rods with only half the normal amount of wild-type rhodopsin. We speculate that both wild-type and mutant opsins are synthesized, transported to the outer segment, and inserted into the disk membrane and that, at very early stages of the disease, rod outer segment length and rhodopsin concentration may be normal. The abnormal carboxy terminal region of the mutant rhodopsin molecules would lead to the abnormal kinetics of recovery of sensitivity after light activation. The patients with the Q344ter mutation had rod dark adaptation results that lead to the speculation that this mutant opsin, like the splice site mutant, is synthesized and transported to the outer segment, where it causes a specific abnormality in the kinetics of recovery of rod sensitivity after light activation due to the truncation at the carboxy terminal region. Lending support to our hypothesis that this mutant opsin may reach the rod outer segment are the results of an investigation of the biochemical phenotype of the Q344ter mutant rhodopsin in vitro showing that this mutant was synthesized, regenerated with 11-cis retinal, and transported to the plasma membrane. 9 The steps leading from the different types of rod dysfunction in the, splice site, and Q344ter mutations to rod cell death are unknown. Further progress in the elucidation of the exact mechanisms of dysfunction and cell death of rod photoreceptors resulting from mutations in the rhodopsin gene will require the use of a number of different approaches, such as studies of the mutant opsins in vitro, 9 ' 4546 of transgenic animals, 47 " 50 and of donor retinas from patients with known genotypes. 51 An important issue concerning the genetic counseling of patients with rhodopsin gene mutations arises from the findings in this study. With this report, there are now five rhodopsin mutations in which some heterozygotes have been described as having a normal ophthalmoscopic appearance and relatively mild retinal functional abnormalities: Q344ter, 10 P23H, 13 E249ter, 7 intron 4 splice site (ref. 7 and present study), and (present study). Four of these five rhodopsin mutations have been associated with adrp; the exception is the E249ter mutation, in which heterozygotes were considered carriers of arrp. 7 Until we learn more about the basis for variation in disease expression in RP, caution dictates that all clinically unaffected heterozygotes with rhodopsin gene mutations should be counseled as if they have adrp. Even if not destined for severe visual loss themselves, they should be told of the chance of having children who could express a more severe form of the disease; and, considering the family with the E249ter mutation, 7 they should be made aware of the chances of producing a homozygote with a consanguineous marriage. Increasing recognition of the wide spectrum of disease expression in different genotypes of retinal degeneration makes the determination of the basis of this variation a topic of clinical and scientific importance warranting further study. 52 ' 53 Key Words null mutation, retinitis pigmentosa, rhodopsin, rod photoreceptor, stop codon Acknowledgments The authors thank Mrs. D. Slaughter, Ms. K. Stewart, and Mrs. B. Koernig for coordinating this study; Dr. X. Sun and

13 Stop Codon Rhodopsin Mutations and RP 2533 Mr. D. Azevedo for help with data acquisition; and Mr. B. Eisner for assistance with data analysis. Dr. D. Hood kindly provided Matlab scripts used for automated fitting of the a-wave model. References 1. Humphries P, Farrar GJ, Kenna P. Autosomal dominant retinitis pigmentosa: Molecular, genetic and clinical aspects. In: Osborne N, Chader G, eds. Progress in Retinal Research. Oxford: Pergamon Press; 1993: Sung C-H, Davenport CM, Hennessey JC, et al. Rhodopsin mutations in autosomal dominant retinitis pigmentosa. Proc Natl Acad Sci USA. 1991; 188: Dryja TP, Hahn LB, Cowley GS, McGee TL, Berson EL. Mutation spectrum of the rhodopsin gene among patients with autosomal dominant retinitis pigmentosa. Proc Natl Acad Sci USA. 1991; 88: Sheffield VC, Fishman GA, Beck JS, Kimura AE, Stone EM. Identification of novel rhodopsin mutations associated with retinitis pigmentosa by GCclamped denaturating gradient gel electrophoresis. Am J Hum G<??i<?n 991; 49: Inglehearn CF, Keen TJ, Bashir R, et al. A completed screen for mutations of the rhodopsin gene in a panel of patients with autosomal dominant retinitis pigmentosa. Hum Mol Genet. 1993; 1: Macke JP, Davenport CM, Jacobson SG, et al. Identification of novel rhodopsin mutations responsible for retinitis pigmentosa: Implications for the structure and function of rhodopsin. Am J Hum Genet. 1993;53: Rosenfeld PJ, Cowley GS, McGee TL, Sandberg MA, Berson EL, Dryja TP. A null mutation in the rhodopsin gene causes rod photoreceptor dysfunction and autosomal recessive retinitis pigmentosa. Nature Genet. 1992; 1: Mclnnes RR, Bascom RA. Retinal genetics: A nullifying effect for rhodopsin. Nature Genet. 1992; 1: Sung C-H, Schneider BG, Agarwal N, Papermaster DS, Nathans J. Functional heterogeneity of mutant rhodopsins responsible for autosomal dominant retinitis pigmentosa. Proc Natl Acad Sci USA. 1991;88: Jacobson SG, Kemp CM, Sung C-H, Nathans J. Retinal function and rhodopsin levels in autosomal dominant retinitis pigmentosa with rhodopsin mutations. Am J Ophthalmol. 1991; 112: Jacobson SG, Voigt WJ, Parel J-M, et al. Automated light- and dark-adapted perimetry for evaluating retinitis pigmentosa. Ophthalmology. 1986; 93: Jacobson SG, Apathy PP, Parel J-M. Rod and cone perimetry: Computerized testing and analysis. In: Heckenlively J, Arden GB, eds. Principles and Practice of Clinical Vision Testing. St Louis: Mosby-Year Book; 1991: Kemp CM, Jacobson SG, Roman AJ, Sung C-H, Nathans J. Abnormal rod dark adaptation in autosomal dominant retinitis pigmentosa with pro-23-his rhodopsin mutation. Am f Ophthalmol. 1992; 113: Lamb TD. The involvement of rod photoreceptors in dark adaptation. Vision Res. 1981; 21: Kemp CM, Jacobson SG, Cideciyan AV, Sung CH, Nathans J. Kinetic analysis of abnormal rod dark adaptation in autosomal dominant retinitis pigmentosa with rhodopsin mutations. ARVO Abstacts. Invest Ophthalmol Vis Sci. 1992;33: DowlingJE. Chemistry of visual adaptation in the rat. Nature. 1960; 188: Rushton WAH. The Ferrier lecture 1962: Visual adaptation. Proc R Soc Lond (B). 1965; 162: Press WH, Flannery BP, Teukolsky SA, Vetterling WT. Numerical Recipes. London: Cambridge University Press; Jacobson SG, Yagasaki K, Feuer WJ, Roman AJ. Interocular asymmetry of visual function in heterozygotes of X-linked retinitis pigmentosa. Exp Eye Res. 1989;48: Jacobson SG, Roman AJ, Cideciyan AV, Robey MG, IwataT, Inana G. X-linked retinitis pigmentosa: Functional phenotype of an RP2 genotype. Invest Ophthalmol Vis Sci. 1992;33: Cideciyan AV, Jacobson SG. Negative electroretinograms in retinitis pigmentosa. Invest Ophthalmol Vis Sci. 1993;34: Sandberg MA, Miller S, Berson EL. Rod electroretinograms in an elevated cyclic guanosine monophosphate-type human retinal degeneration. Invest Ophthalmol Vis Sci. 1990;31: Granit R. The components of the retinal action potential in mammals and their relation to the discharge in the optic nerve. / Physiol. 1933; 77: Hood DC, Birch DG. A computational model of the amplitude and implicit time of the b-wave of the human ERG. Vis Neurosci. 1992;8: Lamb TD, Pugh EN. A quantitative account of the activation steps involved in phototransduction in amphibian photoreceptors. / Physiol. 1992; 449: Pugh EN, Lamb TD. Amplification and kinetics of the activation steps in phototransduction. Biochim Biophys Acta. 1993;1141: Breton ME, Schueller AW, LambTD, Pugh EN. Analysis of ERG a-wave amplification and kinetics in terms of the G-protein cascade of phototransduction. Invest Ophthalmol Vis Sci. 1994;35: Hood DC, Birch DG. Light adaptation of human rod receptors: The leading edge of the human a-wave and models of rod receptor activity. Vision Res. 1993;33: Hood DC, Shady S, Birch DG. Heterogeneity in retinal disease and the computational model of the human-rod response./ Opt Soc Am. 1993; 10: Faulkner DJ, Kemp CM. Human rhodopsin measure-

14 2534 Investigative Ophthalmology 8c Visual Science, April 1994, Vol. 35, No. 5 ments using a TV-based imaging fundus reflectometer. Vision Res. 1984;24: Kemp CM, Faulkner DJ, Jacobson SG. The distribution and kinetics of the visual pigments in the cat retina. Invest Ophthalmol Vis Sci. 1988; 29: Lamb TD. Dark adaptation: A re-examination. In: Hess RF, Sharpe LT, Nordby K, eds. Night Vision. Cambridge: Cambridge University Press. 1990: Fain GL, Lisman JE. Photoreceptor degeneration in vitamin A deprivation and retinitis pigmentosa. Exp Eye Res. 1993; 57: Massof RW, Wu L, Finkelstein D, Peny C, Starr SJ, Johnson MA. Properties of electroretinographic intensity-response function in retinitis pigmentosa. Doc Ophthalmol. 1984; 57: Arden GB, Carter RM, Hogg CR, et al. Rod and cone activity in patients with autosomal dominantly inherited retinitis pigmentosa: Comparisons between psychophysical and electrophysiologic measurements. Br] Ophthalmol. 1983;67: Birch DG, Fish GE. Rod ERGs in retinitis pigmentosa and cone-rod degeneration. Invest Ophthalmol Vis Sci. 1987;28: Hood DC, Shady S, Birch DG. Interpretation of Naka- Rushton parameters from patients with ADRP and CRD. In: Technical Digest on Non-Invasive Assessment of the Visual System. Washington, DC: Optical Society of America; 1993: Kemp CM, Jacobson SG, Cideciyan AV, Wu S. Kinetic analyses of rod dark adaptation in normals and in autosomal dominant retinitis pigmentosa with rhodopsin mutations. ARVO Abstracts. Invest Ophthalmol Vis Sci. 1993; 34: Fain GL, Cornwall MC. Light and dark adaptation in vertebrate photoreceptors. In: Shapley R, Lam DK, eds. Contrast Sensitivity: From Receptors to Clinic. Cambridge, MA: MIT Press; 1993: Corson DW, Cornwall MC, MacNichol EF, et al. Sensitization of bleached photoreceptors by 11-cis-locked analogues of retinal. Proc Natl Acad Sci USA. 1990;87: Cornwall MC, Fain G. Bleaching of rhodopsin in isolated rods causes a sustained activation of PDE and cyclase which is reversed by pigment regeneration. ARVO Abstracts. Invest Ophthalmol Vis Sci. 1992; 33: Hofmann FP, Pulvermiiller, Buczylko J, Van Hooser P, Palczewski K. The role of arrestin and retinoids in the regeneration pathway of rhodopsin. J Biol Chem. 1992;267: Hargrave PA, McDowell JH. Rhodopsin and phototransduction: A model for G protein-linked receptors. FASEBJ. 1992;6: Dolph PJ, Ranganathan R, Colley NJ, Hardy RW, Socolich M, Zuker CS. Arrestin function in inactivation of G protein-coupled receptor rhodopsin in vivo. Science. 1993; 260: Khorana HG. Rhodopsin, photoreceptor of the rod cell. J Biol Chem. 1992;267:l Min KC, Zvyaga TA, Cypess AM, Sakmar TP. Characterization of mutant rhodopsins responsible for autosomal dominant retinitis pigmentosa: Mutations on the cytoplasmic surface affect transducin activation./ Biol Chem. 1993;268: Olsson JE, Gorden JW, Pawlyk BS, et al. Transgenic mice with a rhodopsin mutation (Pro-23-His): A mouse model of autosomal dominant retinitis pigmentosa. Neuron. 1992;9: Naash MI, HollyfieldJG, Al-Ubaidi MR, Baehr W. Simulation of human autosomal dominant retinitis pigmentosa in transgenic mice expressing a mutated murine opsin gene. Proc Natl Acad Sci USA. 1993; 90: Huang P, Gaitan A, Hao Y, Peters RM, Wong F. Cellular interactions implicated in the mechanism of photoreceptor degeneration in transgenic mice expressing a mutant rhodopsin gene. Proc Natl Acad Sci USA. 1993;90: Chang G-Q, Hao Y, Wong F. Apoptosis: Final common pathway of photoreceptor death in rd, rds and rhodopin mutant mice. Neuron. 1993; 11: Li Z-Y, Jacobson SG, Milam AH. Autosomal dominant retinitis pigmentosa caused by the threonine-17-methionine rhodopsin mutation: Retinal histopathology and immunocytochemistry. Exp Eye Res. In press. 52. Humphries P. Hereditary retinopathies: Insights into a complex genetic aetiology. Br J Ophthalmol. 1993;77: Moore AT, Fitzke F, Jay M, et al. Autosomal dominant retinitis pigmentosa with apparent incomplete penetrance: A clinical, electrophysiological, psychophysical and molecular genetic study. Br J Ophthalmol. 1993;77: