Clinical and Genetic Characteristics of 18 Patients from 13 Japanese Families with CRX-associated retinal disorder: Identification of Genotype-phenotype Association

Inherited retinal disorder (IRD) is a leading cause of blindness, and CRX is one of a number of genes reported to harbour autosomal dominant (AD) and recessive (AR) causative variants. Eighteen patients from 13 families with CRX-associated retinal disorder (CRX-RD) were identified from 730 Japanese families with IRD. Ophthalmological examinations and phenotype subgroup classification were performed. The median age of onset/latest examination was 45.0/62.5 years (range, 15–77/25–94). The median visual acuity in the right/left eye was 0.52/0.40 (range, −0.08–2.00/−0.18–1.70) logarithm of the minimum angle of resolution (LogMAR) units. There was one family with macular dystrophy, nine with cone-rod dystrophy (CORD), and three with retinitis pigmentosa. In silico analysis of CRX variants was conducted for genotype subgroup classification based on inheritance and the presence of truncating variants. Eight pathogenic CRX variants were identified, including three novel heterozygous variants (p.R43H, p.P145Lfs*42, and p.P197Afs*22). A trend of a genotype-phenotype association was revealed between the phenotype and genotype subgroups. A considerably high proportion of CRX-RD in ADCORD was determined in the Japanese cohort (39.1%), often showing the mild phenotype (CORD) with late-onset disease (sixth decade). Frequently found heterozygous missense variants located within the homeodomain underlie this mild phenotype. This large cohort study delineates the disease spectrum of CRX-RD in the Japanese population.


Results
Participants. Eighteen affected subjects from 13 Japanese families with a clinical diagnosis of IRD and harbouring CRX variants were identified in this study. The detailed clinical information is provided in Table 1, and the pedigrees of the 13 families are demonstrated in Fig. 1.
Retinal atrophy at the macula was more evident on FAF images in eight subjects ( Table 1. Demographics and detected variants in 18 Japanese patients from 13 families with CRX-associated retinal disorder (CRX-RD). AD = autosomal dominant; AR = autosomal recessive; CORD = cone-rod dystrophy; F = female; CF = counting finger; LCA = Leber congenital amaurosis; LE = left eye; LogMAR BCVA = best-corrected Snellen visual acuity converted to the logarithm of the minimum angle of resolution visual acuity; LP = light perception; M = male; MD = macular dystrophy; No.=number; NA = not available; RE = right eye; RP = retinitis pigmentosa. All affected and unaffected subjects are originally from Japan and any mixture with other ethnicity was not reported. Age was defined as the age when the latest examination was performed. The age of onset was defined as either the age at which visual loss was first noted by the patient or when an abnormal retinal finding was first detected. Phenotype subgroup was defined based on clinical manifestations such as onset of disease, natural course, lesioned part on retinal imaging, and pattern of retinal dysfunction: LCA (including early-onset RP), a severe retinal dystrophy with early onset (<10 years) and extinguished retinal function; RP (including rod-cone dystrophy), a progressive retinal dystrophy often initially presenting peripheral atrophy with generalized rod dysfunction greater than cone dysfunction; CORD, a progressive retinal dystrophy often initially presenting macular atrophy with generalized cone dysfunction greater than rod dysfunction; MD, a progressive retinal dystrophy presenting macular atrophy with confined macular dysfunction despite no abnormalities in generalized cone and rod function. Syndromic findings of central nervous system abnormalities (described as multiple sclerosis-like changes) were reported in Patient 8. Marked preservation of the photoreceptor ellipsoid zone (EZ) at the fovea was identified in eight subjects (8/18, 44.4%; Patients 2, 5, 7, 10-14), and slightly preserved EZ at the fovea was seen in three subjects (3/18, 16.7%; Patients 9,16,17). Preserved foveal structure surrounded by parafoveal atrophy (i.e., bull's eye pattern) was observed in six subjects (6/18, 33.3%; Patient 1-left, 2-both, 10-both, 11-both, 12-both, 17-both).

Phenotype subgroups.
The mean age of onset of the two subjects with MD, 13 subjects with CORD, and three subjects with RP was 46. Five missense variants have been previously reported 16,[27][28][29][30][31][32]34,36 . Three variants were reported in the heterozygous state: p.R40W for CORD, p.R41W for CORD, and p.R43C for CORD. One variant was previously . Atrophic changes affecting the entire retina, including the macula, midperiphery, and periphery are found in Patient 5. Macular atrophy is more evident on FAF images in eight subjects (Patients 1, 3,8,11,12,15,17,18). A ring of high density AF is observed in 11 subjects to various degrees (Patients 1, 2, 3, 7, 8, 11-13, 15, 17, 18). Foveal appearance is relatively preserved in nine subjects (Patients 1-3 Together with the clinical features of the affected subjects and the model of inheritance in the pedigree, eight disease-causing variants in the CRX gene were determined.
In silico molecular genetic analysis. The detailed results of in silico molecular genetic analyses for the eight detected CRX variants are presented in Supplemental Tables 4 and 5. Schematic genetic and protein structures of CRX are shown in Fig. 4, and multiple alignment of seven species of CRX is presented in Supplemental Fig. 1.
General prediction, functional prediction, and conservation were assessed for the six missense variants and two single nucleotide deletion variants leading to frame shift and pathogenicity classification according to the American College of Medical Genetics and Genomics (ACMG) guidelines was performed. One pathogenic missense (p.R90W) and five likely pathogenic missense variants (p.R40W, p.R41W, p.R43C, p.R43H, and p.D65H) and the two likely pathogenic truncating variants (p.P145Lfs*42 and p.P197Afs*22) were revealed.
Overall, eight disease-causing variants in the CRX gene were identified in 13 families with ADCORD, ADRP, ADMD, and ARRP.
Genotype-phenotype association. Genotype subgroup classification was performed in the proband of 13 families (Patients 1, 3, 5, 6, 7, 10, 11, 13-18). There were eight subjects in genotype subgroup A (heterozygous missense), two in genotype subgroup B (homozygous missense), and three in genotype subgroup C (truncating variants). A distribution of the 13 families based on genotype subgroups and phenotype subgroups is shown in Table 2. There is a trend, but the number is not sufficient to statistically demonstrate a significant genotype-phenotype association.

Discussion
Detailed clinical and genetic characteristics of a Japanese cohort of 18 affected subjects from 13 families with CRX-RD are illustrated. Diverse clinical presentations with different inheritance patterns were identified in CRX-RD, including nine families with molecularly confirmed ADCORD, one family with ADMD, two families with ARRP, and one family with ADRP.
To our knowledge, these are data from the largest cohort of CRX-RD and includes the highest number of ADCORD cases to date, despite there being a well-characterized study of CRX-RD in 11 families from the UK 32 . Six out of 30 families (20.0%) diagnosed with CORD/MD/STGD and having a clear AD family history in the Japan Eye Genetics Consortium (JEGC) IRD cohort were associated with CRX-RD. The proportion of ADCRX-RD in molecularly confirmed ADCORD/MD/STGD in the JEGC cohort was considerably high (9/23 families, 39.1%) in comparison with European cohorts (e.g., 15.6% in the UK cohort) 7 . On the other hand, there were no families with LCA in our cohort, while four out of 11 families (36.4%) with CRX-RD in the UK cohort manifested the severe LCA phenotype. There could be a bias in the enrolment of IRD patients, although ethnic variation can also occur in CRX-RD.
The median age of onset for CRX-RD was in the fifth decade in our cohort, although it varied from teenage years to the 8 th decade, which is considerably later than that of other CORD/MD/STGD patients (e.g., 19.0 years for ABCA4-associated retinal disorder) 37 . In addition, over half of patients with late-onset disease (>45 years) have preserved favourable VA, and two maintained VA even after 10 years of disease history (Patients 13 and 15). Fundus and FAF showed variable findings; however, the severity of macular atrophy was generally associated with the severity of VA decline, and a characteristic ring of increased AF signal was observed in most subjects (>70%). Two-thirds of the subjects demonstrated preserved foveal structure often surrounded by parafoveal atrophy (i.e., Electrophysiological findings of CRX-RD were also mildly affected in our cohort. Ten of 16 subjects (62.5%) had no or mild dysfunction both in generalized rod and cone systems, all of whom were classified into MD or CORD. In contrast, two subjects with CORD showed moderate retinal dysfunction, one subject with CORD and three subjects with RP had severe retinal dysfunction. Interestingly, a lower b to a ratio in DA bright flash responses was identified in approximately half of the subjects. Although an interaction with the phototransduction cascade was suggested in a previous study of CRX-and OTX2-transfected iris-derived cells 38 , the molecular mechanism to support this phenomenon is unknown. This electronegative finding was also observed in the early stage of other CORD/MD/STGD and was not specific for CRX-RD 13,39-41 ; however, this characteristic feature can be helpful to consider CRX-RD in patients with early maculopathy.
Phenotype subgroups were associated with disease severity in our cohort. The subjects with MD had later-onset disease, maintained VA, and normal generalized retinal function. In general, the subjects with CORD had late-onset disease but VA decline, and the subjects with RP presented early-onset disease, VA decline, and severe generalized retinal dysfunction. Thus, determining phenotype subgroups with comprehensive clinical assessments provides crucial information directly related to disease severity and progression.
Eight pathogenic/likely pathogenic CRX variants were identified in our cohort, including three novel variants. One novel missense variant (p.R43H) located within the homeodomain of the CRX protein was found in two affected subjects with MD in a single family. A novel de novo truncating variant (p.P145Lfs*42) was revealed in a patient with early-onset RP. A novel recurrent truncating variant (p.P197Afs*22) was detected in two families with CORD. Comprehensive high-throughput gene screening of both affected and unaffected members was effective in obtaining a genetic diagnosis of CRX-RD manifesting AD or AR inheritance, as well as identifying de novo variants.
Four recurrent CRX variants were identified in our cohort, and two of these with available allele frequency in the general population revealed considerably high frequency in East Asia (p.R41W and p.D65H). Several cases with ADCORD caused by the former variant (p.R41W) have been reported in East Asian 16,28,35 , and the phenotype was the same as that observed in our two families (Families 4 and 5). Jin et al. reported only one Japanese RP case homozygous for the latter variant (p.D65H), and the phenotype was the same as that observed in our two families (Families 8 and 9) 29 . Given these facts, these two CRX variants with higher frequency are major causes of CRX-RD in the East Asian population, leading to CORD and RP, respectively.
The patients with heterozygous missense variants located within the homeodomain frequently associated with CORD (7/8 families; 87.5%) are consistent with previous studies 34 . A postulated dominant-negative effect can be considered for these heterozygous missense variants within the homeodomain, as reported for p.K88N 42 . Two families with homozygous missense variants (p.D65H) showed a severe phenotype, and the molecular mechanism is uncertain, unlike the well-studied homozygous missense variants (p.R90W), in which the mutant homeodomain showed a significantly reduced ability to transactivate the rhodopsin promoter and lower synergistic activation with the transcription factor NRL 36 . Three families with heterozygous truncating variants showed CORD (2/3, 66%) or RP (1/3, 33%). Notably, nonsense-mediated decay could possibly modify the phenotype in such variants 43 .
There are limitations in this study. The selection bias related to the disease severity should be inherent, since it is unusual for genetically affected subjects with good vision to visit clinics/hospitals. In addition, this cross-sectional retrospective case series study does not include longitudinal information; thus, natural history studies in a larger cohort could provide more accurate information regarding the disease progression of CRX-RD. The molecular mechanisms of AD missense, AR missense, and AD truncating variants have not yet been clarified in CRX-RD, and further functional investigation for each variant is required to conclude disease causation. The samples of affected and unaffected subjects of families with CRX-RD are still small to conclude the molecularly confirmed inheritance and genotype-phenotype associations/correlations in such a diverse disorder; thus, larger cohort studies are required for further analyses.
In conclusion, this large nationwide cohort study delineates the clinical and genetic characteristics of CRX-RD in Japan. A high proportion of ADCRX-RD was determined in Japan, which manifests late-onset ADCORD. The frequently found missense variants located within the homeodomain of the CRX protein can explain the mild phenotype of CRX-RD. In contrast, a relatively severe RP phenotype was associated with homozygous CRX  www.nature.com/scientificreports www.nature.com/scientificreports/ missense variants in a small number of patients. This information will help to monitor and counsel patients, as well as design future therapeutic trials.

Methods
The protocol of this study adhered to the tenets of the Declaration of Helsinki, which was approved by the ethics committee of the participating institutions from Japan: National Institute of Sensory Organs, National Hospital Organization Tokyo Medical Center (Reference; R18-029). Informed consent was received from all participants for the tests after an explanation of the procedures, and permission was obtained to use their medical data for research.
Participants. Participants with a clinical diagnosis of IRD and available genetic data were studied between 2008 and 2018 as a part of the Japan Eye Genetics Consortium Studies (JEGC studies; http://www.jegc.org/) 44  Clinical investigations. Medical history was obtained for all affected subjects and unaffected family members (where available). The onset of disease was defined as the age when any visual symptom was first noted by patients or parents or when the subject was first diagnosed.
Comprehensive ophthalmological examinations were performed in all affected subjects and unaffected family members (where available), including measurements of decimal VA converted to LogMAR units, ophthalmoscopy, fundus photography, FAF imaging, SD-OCT, kinetic and static visual field testing, and electrophysiological assessments according to the international standards of the International Society for Clinical Electrophysiology of Vision (ISCEV) 45-48 . Phenotype subgroups. For the purpose of this study, phenotype subgroups were defined based on clinical manifestations such as onset of disease, natural course, lesioned part on retinal imaging, and pattern of retinal dysfunction, partially according to a previous report 34 : LCA (including early-onset RP), a severe retinal dystrophy with early onset (<10 years) and extinguished retinal function; RP (including rod-cone dystrophy), a progressive retinal dystrophy often initially presenting peripheral atrophy with generalized rod dysfunction greater than cone dysfunction; CORD, a progressive retinal dystrophy often initially presenting macular atrophy with generalized cone dysfunction greater than rod dysfunction; MD, a progressive retinal dystrophy presenting macular atrophy with confined macular dysfunction despite no abnormalities in generalized cone and rod function.
Genetic screening of the CRX gene. Genomic DNA was extracted from affected subjects and unaffected family members (where available for co-segregation analysis). Whole exome sequencing with target sequence analysis of 301 retinal disease-associated genes (based on RetNET; https://sph.uth.edu/retnet/home.htm; accessed on 1 July 2017) was performed according to a previously published method and through the Phenopolis platform (www.phenopolis.org) 44,49 . The identified variants were filtered on their allele frequency (less than 1%) in the Human Genetic Variation Database (HGVD; http://www.genome.med.kyoto-u.ac.jp/SnpDB/about.htm; accessed on 1 July 2017), which provides allele frequency of the general Japanese population. Depth and coverage for the target areas were examined with the integrative Genomics Viewer (http://www.broadinstitute.org/igv/) to detect structural variants.
Disease-causing variants were determined from the called/detected variants in the 301 retinal disease-associated genes, in consideration of the clinical findings of the affected subjects, the model of inheritance in the pedigree, and the results of co-segregation analysis.
Genotype subgroups. Genotypic subgroup classification was performed based on the heterozygous/ homozygous state of missense variants and presence of truncating variants: Genotype A-subjects with heterozygous missense variants; Genotype B-subjects with homozygous missense variants; and Genotype C-subjects with heterozygous truncating variants.