The Ningxia Hui Autonomous Region People’s Hospital Ethics Committee granted approval for the study (approval number 2022-KJCG-006), and everyone or legal guardian who took part in it signed an informed consent form and strictly followed the Declaration of Helsinki.

Referring to Ghose et al. (Ghose et al., 1985) who used ultrasonographically measured ocular axial length of less than 2 standard deviations from the mean value of normal subjects as a criterion, the patients with nanophthalmos whose axial length was less than 20 mm and who had no previous history of ocular surgery and were not accompanied by eyeballs and other systemic deformities were included in this study.

Five nanophthalmos pedigrees were recruited from the Ningxia Eye Hospital, People’s Hospital of Ningxia Hui Autonomous Region from 2020 to 2021. The patients and their major family members were asked about their ocular history, marital and family history. The detailed ophthalmologic examinations were performed, including best-corrected visual acuity (BCVA), slit-lamp microscopy, fundus examination with photographs (Daytona (P200T), OPTOS), optical coherence tomography (OCT, HDOCT4000, Carl Zeiss Meditec, United States), electroretinogram (ERG) and ocular axial measurement (SW-2100,SUOER) and ultrasound biomicroscopy (UBM,SW-3200L,SUOER). ERG was detected and analyzed with reference to the standards of the International Society for Clinical Electrophysiology of Vision (ISCEV).

Whole exome sequencing was performed on probands and their parents (Trio whole-genome exome sequencing model). Agilent SureSelect exome capture kit was used to perform the whole-genome exome capture, and high-throughput sequencer (Illumina) was used to perform sequencing subject to a depth of 100 ×. The raw sequencing data were processed by Illumina basecalling Software 1.7 and subsequently compared with the NCBI human genomic DNA reference sequence (NCBI build 37.1). Single nucleotide variation (SNV) was analyzed by SOAP software (http://soap. genomics. org.cn) while insert-deletion (Indel) was analyzed by BWA software (bio-bwa.sourceforge.net/) to obtain all the variants occurring in the DNA sequences in the samples. Common variants (which are usually considered that MAF (minor allele frequency) > 1%) were filtered from the database (db135). The variants that do not affect the structure and function of the protein were then filtered out. After stepwise filtering, the number of variants shared by all patients in the families was screened out, and then variants unavailable with the diseased relatives in the families were filtered to obtain the candidate pathogenic gene variants. Sanger validation was used to exclude false positives for candidate pathogenic genetic variants, and co-segregation was further verified in the normal family members.

The American College of Medical Genetics and Genomics (ACMG) established Standards and Guidelines for Interpretation of Sequence Variants in 2015, which were used to evaluate the pathogenicity of novel variations for genetic variation. MAF<0.005 was used as the criteria to exclude benign variants by reference to the databases for East Asian populations Allele frequencies available with 1000 Genomes Project (1000G, http://browser.1000genomes.org) and Exome Aggregation Consortium (http://exac.broadinstitute.org/). Polyphen2 (http://genetics.bwh.harvard. edu/pph2), SIFT (http://sift.jcvi.org),REVEL (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5065685/), CADD (https://cadd.gs.washington.edu/score) and Mutation Taster (http://mutationtaster.org/) were used for pathogenicity prediction. Measurements of the conservation of gene sequences across species in evolution have been made using websites like GERP++ (https://bio.tools/gerp). Variants were classified as uncertain clinical significance when at least one of four predictions had a benign outcome or when there was insufficient evidence of pathogenicity. When all predictions turned out to be accurate, variations were categorized as potentially pathogenic when used in conjunction with further data. Pathogenic variants were defined as frameshift, nonsense, and variants with experimental proof of causing loss of protein function. For the conservativeness study of variant loci, the online analysis tool Multalin (http://sacs.ucsf.edu/cgi-bin/multalin.py) was employed. Alphafold was used to construct the normal protein structure, and pymol software was used to make visualized analysis of the mutant protein.

The proband of family 1, a female, 7 years old, complained of reduced vision in both eyes for 1 year. Her parents denied family history and consanguineous marriage history (Figure F1-A). The BCVA was 0.05 in both eyes. The axial length of right eye and left eye were 15.15 mm and 14.71 mm, respectively. The anterior chamber depth of right eye and left eye were 2.36 mm and 2.23 mm, respectively (Figure F1-C). The fundus examination revealed pseudo-papilloedema in both eyes, no reflection in the macular fovea, thickened retinas at the posterior pole, tortuous and dilated retinal vessels, and indistinguishable arteries and veins (Table 1; Figure F1-C). OCT showed the macular mildly edematous, and the fovea centralis disappeared in both eyes (Figure F1-C). Binocular color ultrasonography suggested that the eye axes of both eyes were short and the eyeball walls were thickened. ERG findings suggested that the scotopic vision 0.01 and responses of flicker were in the normal range, whereas the b-wave amplitudes in the photopic 3.0 and scotopic 3.0 responses were significantly higher, and that the function of the amacrine cells in both eyes was basically normal (Table 2). As confirmed by whole exome sequencing and Sanger sequencing, the proband carried a homozygous missense variant c.1486G>A (p.Glu496Lys) in the MFRP gene, and both the phenotypically normal father and mother carried the same heterozygous variant, suggesting co-segregation of genotypes and clinical phenotypes (PP1_Supporting) (Figure F1-B; Table 3), which is consistent with an autosomal recessive pattern of inheritance. The variant has not been detected in the East Asian population database (ExAC_ EAS), gnomAD database and it has not been reported before (PM2_Moderate). A variety of bioinformatics computing software indicated the deleterious impact of this variant (PP3_Supporting). Moreover, proteomic conservation analysis suggested that the amino acid at site 496 was highly conserved across multiple species (Figure 1D), indicating that variants at such site are likely to have an impact on the structure and function of the MFRP protein. In the wild-type MFRP protein, site 496 was a polar negatively charged glutamic acid, and its amino acid backbone formed a hydrogen bond with the O atom of the uncharged polar valine at site 500 and the negatively charged glutamic acid at site 499, and such mutation resulted in the substitution of the polar negatively charged glutamic acid by the polar positively charged lysine at site 496 of the MFRP protein. These alterations in amino acid interactions did not lead to alterations in backbone interactions, while the mutation in the structure of the side chain led to alterations in interactions with 1,492, which resulted in alterations in protein structure and function (Figure 1E). According to the ACMG guidelines, the c.1486G>A (p.Glu496Lys) missense variant was rated as PP4 due to the phenotype of the missense variant carrier was highly consistent with nanophthalmos, and genetic testing report from credible sources suggested that the variant was pathogenic as PP5. Therefore, the c.1486G>A (p.Glu496Lys) missense variant was classified as a likely pathogenic (PP1+PM2+PP3+PP4+PP5).

The proband of family 2, a female, age of 17, complained of progressively reduced vision in both eyes for 10 years accompanied by night blindness for 8 years. Parents are consanguineous marriage (Figure 2A). She had horizontal nystagmus in both eyes. BCVA was 0.1 in the right eye and 0.15 in the left eye. The axial length of right eye and left eye were 15.57 mm and 15.59 mm, respectively. The anterior chamber depth of right eye and left eye were 2.26 mm and 2.23 mm, respectively (Figure 2C). Both eyes demonstrated shallow anterior, slightly smaller optic discs in both eyes with clear borders visible on fundus examination, pigmentation of peripheral retina 360° from the equator to the serrated rim showing a clear demarcation from the normal retina in the form of strips (Table 1; Figure 2C). OCT showed cystic dilatation of the fovea centralis, separation between retinal neuroepithelial layers, and visible small cystic cavities segmented by columnar connections (Figure 2C). ERG findings suggested severe decreases in the b-wave amplitude in the scotopic 0.01 and 3.0 responses, and mild decreases in the b-wave amplitude in the photopic 3.0 response and severe decreases in the flicker response (Table 2). As confirmed by whole exome sequencing and Sanger sequencing, the proband carried a homozygous nonsense variant c.271C>T (p.Gln91Ter)in the MFRP gene, and the phenotypically normal father and mother carried the same heterozygous variant, suggesting co-segregation of genotypes and clinical phenotypes (PP1_Supporting) (Figure 2B; Table 3), which was consistent with an autosomal recessive pattern of inheritance. Such variant was not detected or reported in the East Asian population database (ExAC_ EAS) and the gnomAD database (PM2_Moderate). Such mutation caused a change in the glutamine at site 91 to a termination codon, leading to a premature termination of polypeptide chain synthesis (Figure 3). The majority of proteins produced were inactive or lost their usual function (PVS1 - very Strong). Various bioinformatics computing software indicated the deleterious impact of this variant (PP3 - supporting). In addition, proteomic conservation analysis suggested that the amino acid at position 91 was substantially conserved across several taxa (Figure 2D), suggesting that variants at such site were likely to have an impact on the structure and function of the MFRP protein. Therefore, according to the ACMG guidelines, we consider the homozygous nonsense variant c.271C>T (p.Gln91Ter) of MFRP as pathogenic (PVS1+PP1+PM2+PP3).

The proband of family 3, a male, age of 19, complained of sudden blurring of vision in the right eye for 11 days. He denied a family history of hereditary disease and consanguineous marriage history (Figure 1F3-A). BCVA was 0.02 in the right eye and 0.12 in the left eye. The axial length of right eye and left eye were 16.02 mm and 16.28 mm, respectively (Figure 1F3-C). The anterior chamber depth of right eye and left eye were 2.15 mm and 2.69 mm, respectively. There was no obvious abnormality in the anterior segment of both eyes. Fundus examination suggested that only the posterior retina in the right eye was in normal position as flat and attached, while the detached retina appeared to have a ring-shaped shape and was smooth and free of hiatus; in the left eye the optic disc was small and congested, with clear borders, and the retinal vessels were dilated, with indistinct arteries and veins (Table 1; Figure 1F3-C). OCT showed a wrinkled, elevated macular area and the fovea centralis disappeared of the left eye. Macular OCT could not be performed in the right eye due to retinal detachment (Figure 1F3-C). The results of binocular color ultrasonography showed both eyes to have short axes, and the right eye to have retinal detachment coupled with choroidal detachment. ERG findings suggested a severe decrease in b-wave amplitude in the scotopic 0.01, scotopic 3.0, and photopic 3.0 responses in the right eye, and a severe decrease in flicker response and amacrine cells function. In the left eye, the b-wave amplitude was mildly decreased in the scotopic 0.01 and scotopic 3.0, the b-wave amplitude was within the normal range in the photopic 3.0 response, the response of flicker was severely decreased, and the amacrine cells function was basically normal (Table 2). As confirmed by the whole exome sequencing and Sanger sequencing, the patient carried a homozygous missense variant c.1486G>A (p.Glu496Lys) in the MFRP gene, and both the phenotypically normal father and mother carried the same heterozygous variant, suggesting co-segregation of genotypes and clinical phenotypes (Figure 1F3-B; Table 3) (PP1_supporting), which is consistent with an autosomal recessive pattern of inheritance. Such variant was not detected in the East Asian population database (ExAC_ EAS), gnomAD database (PM2_Moderate). A variety of bioinformatics computing software indicated the deleterious impact of this variant (PP3_supporting). Proteomic conservation analysis revealed that the amino acid at site 496 was highly conserved across multiple species (Figure 1D). The protein structure suggested that the mutation will result in protein structure and function. (Figure 1E; Figure 3). According to the ACMG guidelines, the c.1486G>A (p.Glu496Lys) missense variant was classified as a likely pathogenic (PP1+PM2+PP3+PP4+PP5).

The proband of family 4, a female, age of 41, complained of reduced vision accompanied by occlusion for 3 months. She denied consanguineous marriage (Figure 4A). BCVA was 0.04 in the right eye and 0.15 in the left eye. The axial length of right eye and left eye were 18.25 mm and 18.30 mm, respectively. The anterior chamber depth of right eye and left eye were 2.00 mm and 2.17 mm, respectively (Figure 4C). The intraocular pressure was 27 mmHg in the right eye and 23 mmHg in the left eye (Table 1), and she had not received any treatment before joinning the study. Static gonioscopy showed closure of the anterior chamber angle in all four quadrants, all of which were narrow IV. Fundus examination revealed that small optic discs with clear borders in both eyes, C/D ratio was 0.8, with markedly dilated retinal veins in the right eye, but mildly dilated retinal veins in the left. A large number of scattered hemorrhagic spots were found on the retinal surface of the right eye at follow-up for up to 2 months. OCT showed the macular area was wrinkled and mildly edematous, and the fovea centralis disappeared in the right eye and was visible in the left eye (Figure 3C). ERG findings suggested b-wave amplitude higher than normal in the scotopic 0.01, scotopic 3.0 and photopic 3.0, responses of flicker were also higher than normal range in the left eye; b-wave amplitude and response of flicker in the normal range in the scotopic 0.01 in the right eye, the b-wave amplitude was higher than normal in the scotopic 3.0 and photopic vision 3.0, the amakrine cells function was basically normal (Table 2). As confirmed by the whole exome sequencing and Sanger sequencing, the patient carried a homozygous missense variant c.1411G>A (p.Val471Met)in the MFRP gene, and the phenotypically normal son carried the same heterozygous variant, suggesting co-segregation of genotypes and clinical phenotypes (Figure 3B; Table 3), which is consistent with an autosomal recessive pattern of inheritance (PP1_Supporting). Such variant was not detected in the East Asian population database (ExAC_ EAS), gnomAD database (PM2_Moderate). A variety of bioinformatics computing software indicated the deleterious impact of this variant (PP3_supporting). Proteomic conservation analysis revealed that the amino acid at site 471 was highly conserved across multiple species (Figure 4D), indicating that variants at such site are likely to have an impact on the structure and function of the MFRP protein. In the wild-type MFRP protein, site 471 was a nonpolar uncharged valine, and the O atom of its amino acid backbone formed a hydrogen bond with the O atom of the uncharged polar cysteine at site 474 and the uncharged nonpolar amino acid at site 475, and the p. Val471Met mutation resulted in the substitution of the nonpolar uncharged valine at site 471 of the MFRP protein for a nonpolar uncharged methionine. Such mutation led to the disappearance of the hydrogen bond formed between the O atom of the backbone and the uncharged nonpolar leucine at site 475, resulting in alterations in the structure and function of the protein (Figure 4E). According to the ACMG guidelines, the c.1411G>A (p.Val471Met)missense variant was rated as PP4 due to the phenotype of the missense variant carrier was highly consistent with nanophthalmos, and genetic testing report from credible sources suggested that the variant was pathogenic as PP5. Therefore, the c.1411G>A (p.Val471Met) missense variant was classified as a likely pathogenic according to the ACMG guidelines (PP1+PM2+PP3+PP4+PP5).

The proband of family 5, a female, age of 43, complained of reduced vision for 6 months. She denied a family history of hereditary disease and consanguineous marriage history (Figure 5A). BCVA was 0.15 in the right eye and 0.2 in the left eye. The axial length of right eye and left eye were 15.28 mm and 15.03 mm, respectively. The anterior chamber depth of right eye and left eye were 1.91 mm and 2.00 mm, respectively (Figure 5C). The intraocular pressure was 30 mmHg in the right eye and 28 mmHg in the left eye (Table 1), and she had not received any treatment before joinning the study. Static gonioscopy showed closure of the anterior chamber angle in all four quadrants, all of which were narrow IV. Fundus examination revealed that small optic discs with clear borders in both eyes, C/D ratio was 0.8, with mildly dilated retinal veins. Macular spectral-domain optical coherence tomography of both eyes shows a small dome-shaped papillomacular fold and nomacular edema (Figure 5C). ERG findings suggested that the b-wave amplitude in the scotopic 0.01 and 3.0 responses was within the normal range, the flicker response was severely decreased, and amakrine cells functioned normally. As confirmed by whole exome sequencing and Sanger sequencing, a novel compound heterozygous missense variants c.1486G>A (p.Glu496Lys) and c.602G>T (p.Arg201Leu) were detected on the MFRP gene of the proband (Figure 5B; Table 3) (PP1_supporting). The parents of the proband were deceased, and the genetic origin was unclear. However, it can be considered an autosomal recessive pattern of inheritance according to genetic test results and clinical phenotype. Such variant was not detected in the East Asian population database (ExAC_ EAS) and the gnomAD database (PM2_Moderate), and a variety of bioinformatics computing software indicated the deleterious impact of this variant (PP3_supporting). Proteomic conservation analysis revealed that the amino acid at site 496 and 201 were highly conserved across multiple species (Figure 5D). The protein structure suggested that the 201th position of wild-type MFRP protein is arginine with positive polarity, and the N atom in the main chain forms hydrogen bonds with the uncharged nonpolar valine O atom in the 236th position, and the O atom in the main chain forms hydrogen bonds with the valine N atom, and the side chains of amino acids form hydrogen bonds with the F199 position and E203. The mutation of p. Arg20Leu leads to the substitution of arginine at position 201 by nonpolar leucine, and the N atom of leucine main chain forms hydrogen bonds with uncharged nonpolar valine O atom at position 236, and the O atom of main chain forms hydrogen bonds with valine N atom, and the nonpolar functional groups of side chains do not interact with the surrounding amino acids. The changes of amino acid interaction and amino acid properties lead to the changes of protein structure and function after mutation. The missense variant c.1486G>A (p.Glu496Lys) resulted in the substitution of the polar negatively charged glutamic acid for a polar positively charged lysine at site 496 of the MFRP protein, which resulted in alterations in protein structure and function (Figure 5E). Therefore, the c.1486G>A (p.Glu496Lys) and c.602G>T (p.Arg201Leu) missense variant was classified as a likely pathogenic according to the ACMG guidelines (PM2+ PP1+PP3+PP4+PP5).