This study was approved by the Ethical Review Board of the First People’s Hospital of Yunnan Province. Informed consent was obtained from all participants and the participants’ legal guardians before the collection of clinical data and genomic samples (Approval No: KHLL2025-KY026).

The proband visited our outpatient clinic because of high myopia, and her parents were eager to prevent the recurrence of the same situation in the offspring. Information regarding the family history was obtained through interviews with the patient’s parents. A detailed physical examination and developmental assessment were performed. The proband was subsequently tested via refractive error measurement, best-corrected visual acuity (BCVA), slit lamp examination, and fundus photography. Moreover, audiometry and spine X-ray examinations were performed.

Genomic DNA was extracted from the peripheral venous blood of the patient and her parents. WES was performed as previously described (Mamanova et al., 2010). Firstly, the DNA was interrupted and the library was prepared, and then the DNA of the target gene exon and adjacent cut region was captured and enriched by Roche KAPA HyperExome chip (KAPA Biosystems, Boston, MA, United States of America). Finally, the DNA was screened using NGS assay based on the DNA sequencer of MGISEQ-2000 (BGI, China). The average depth of sequencing in the target region was ≥180X, and the proportion of sites with average depth >20X in the target region was >95%. The sequencing fragments were compared with the human reference genome (UCSC hg19) by BWA to remove duplicates. GATK was used for base mass value correction for SNV, INDEL and genotyping. Exon level copy number variation was detected using ExomeDepth. Variants were annotated and screened based on clinical information, population databases, disease databases, and biological information prediction tools. The pathogenicity of the variants was determined according to the ACMG guidelines.

The candidate COL11A1 variant was validated by Sanger sequencing for the patient and her parents. Primers were designed for the variant of the splice site variant in intron 53 of the COL11A1 gene. The forward (AAG​CTA​GCA​CTG​GAC​TTT​TGA​C) and reverse primers (AGG​TAT​CTG​AAA​TAG​GGC​TGA​G) were used for amplification. The PCR products were further subjected to Sanger sequencing.

Four in silico splice site prediction programs, MaxEntScan (https://www.genes.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html), NNSPLICE 0.9 (https://www.fruitfly.org/seq_tools/splice.html), NetGene2 (https://services.healthtech.dtu.dk/services/NetGene2-2.42), and alternative splice site predictor (http://wangcomputing.com/assp), were used to predict the splice variants to evaluate variant pathogenicity. The protein structure models of AlphaFold (https://alphafold.ebi.ac.uk/entry/P12107) and SWISS-MODEL (https://swissmodel.expasy.org/) were used to explore the effect of pathogenic variant in the COL11A1 gene on protein structure. Subsequently, the three-dimensional structure of the protein models was analyzed and visualized using PyMOL. The prediction method was carried out using standard procedures.

Using genomic DNA from the proband and his parents as a template, primers were designed to amplify the COL11A1 gene, including the exon 53, intron 53, exon 54, part of intron 54, and exon 55 regions. Primers were designed via seamless cloning to amplify heterozygous genomic DNA carrying the c.4069-1G>T variant site to obtain two target insertion gene fragments, COL11A1-A and COL11A1-B. The sequences of primers used were as follows: AF: AAG​CTT​GGT​ACC​GAG​CTC​GGA​TCC​GGT​CAA​GAT​GGT​GTT GGT​GGT​GAC​AAG​G; AR: CCT​GAG​TGT​GGA​GTT​AAT​GTT​CAG​GAT​ACC​ACC​AAA​GTC; BF: CAT​TAA​CTC​CAC​ACT​CAG​GGT​GGC​GAT​TTT​AAT​TC; and BR: TTA​AAC​GGG​CCC​TCT​AGA​CTC​GAG​CTT​AGC​ACC​TTT​TTC​ACC​TTG​TCT​TCC​CTC​T.

The vector pMini-CopGFP0 was digested via the BamH I/Xho I double endonuclease, and the digested products and the COL11A1-A fragment and COL11A1-B fragment were recombined. The correct wild type (WT) and mutant type (MT) minigene plasmids were selected, and 293T cells were subsequently transfected with the WT and MT plasmids via Lipofectamine 3000 according to the instructions. After transfection into 293T cells for 48 h, the transcripts were analyzed via RT‒PCR. The primers used for RT‒PCR amplification were as follows: primer sequence F: GGC​TAA​CTA​GAG​AAC​CCA​CTG​CTT​A; R: CTT​AGC​ACC​TTT​TTC​ACC​TTG​TCT​TC. The PCR products for both the WT and MT variants were subjected to 1.0% agarose gel electrophoresis, followed by Sanger sequencing.

We summarized the phenotypes and compiled the incidence of different phenotypes of STL from a previous review (Boothe et al., 2020), systematic review (Boysen et al., 2020; Frederic et al., 2012), and large-scale studies (Alexander et al., 2020; Bath et al., 2021; Rose et al., 2001; Zimmermann et al., 2019). To comprehensively identify previously published patients with COL11A1 splice site alteration variants, we used ClinVar to identify pathogenic and likely pathogenic COL11A1 variants.

The proband, a 6-year-old girl, was the first child of nonconsanguineous parents. No family history and medical history was observed within her family. Her mother had a normal pregnancy, and a cesarean section was performed at term. The patient was found to have high myopia at the age of 3 years. Ophthalmologic assessment confirmed bilateral highly myopic astigmatism in the right eye (RE): −8.00/–1.00 × 180 and left eye (LE): −5.50/–1.00 × 35. Slit lamp examination revealed no abnormalities in the conjunctiva, cornea, or lens. Fundus examination revealed leopard fundus (Figure 1). The results of the spinal X-ray examination revealed mild scoliosis (Figures 2A,B). Her height was 114 cm, which was below the 10th percentile. Her motor and mental development were normal. No clinical characteristics of orofacial abnormalities were detected. Furthermore, audiometric examination revealed no significant abnormalities.

A de novo heterozygous splice site variant, c.4069-1G>T, in the COL11A1 gene was identified via WES. This variant was confirmed to not be previously annotated in ESP databases, the 1,000 Genomes Project, the ExAC Browser, or the gnomAD database. Sanger sequencing of the COL11A1 gene confirmed the c.4069-1G>T heterozygous variant in the patient, and neither of her parents carried the variant (Figure 2C). According to the ACMG guidelines for variant pathogenicity, c.4069-1G>T was classified as “likely pathogenic” (PVS1_Moderate + PS2_Moderate + PM2).

The prediction scores for acceptor and donor splice sites in the wild type and mutated type are shown in Figure 3. A higher score predicts a strong splice site. The scores of the acceptor splice site of intron 53 were markedly decreased from 8.49 to −0.10 for the c.4069-1G>T variant determined via MaxEntScan. The other three prediction programs showed that the acceptor splice site mutated from G to T resulted in the disappearance of the acceptor splice site. These analyses revealed that the variant weakened the acceptor splice site.

To determine the splicing pattern and verify the pathogenicity of the splice site variant (Figure 4B), we constructed a minigene vector via an in vitro splicing assay for both the WT and MT COL11A1 gene sequences from exons 53 to 55. RT‒PCR revealed that the cells containing the WT plasmid produced an amplicon of 233 bp. In contrast, the cells transfected with the c.4069-1G>T variant sequence produced two fragments, a 254 bp band and a 179 bp band (Figure 4A). Subsequently, Sanger sequencing revealed that the two mutant fragments presented exon 54 skipping and a retention of 21 bp in intron 53 (Figure 4C). These findings suggest that the variant in the COL11A1 gene produced two aberrantly spliced cDNA.

The three-dimensional (3D) structure prediction of the wild and mutant models of the COL11A1 protein were shown in Figure 5. Two aberrantly spliced cDNA resulted in one protein variant with a deletion of 18 native amino acids (p.Gly1357_Arg1374del) (Figures 5b1,b2,b3), and another variant with an insertion of seven amino acids (p.Pro1356_Gly1357insValPhePheHisIleLeuTyr) (Figures 5c1,c2,c3). Compared with the wild type sequence, both variant proteins had altered amino acid sequences. Further structural analysis suggested that both mutant proteins form new hydrogen bonds in the mutated regions, leading to abnormal helix folding.

Owing to the high phenotypic heterogeneity of STL, we summarized the clinical phenotypes of partially reported STL cases. As shown in Table 1, the most common phenotypes included myopia (88%) (Boysen et al., 2020) and hearing impairment (82.5%) (Frederic et al., 2012) in STL type 2. In our patient, myopia was the main presentation, but hearing abnormality was not observed.

Other common clinical phenotypes of STL type 2 included cataract (59%), retinal detachment (38%), cleft palate (24.6%), and joint hypermobility (33.8%). Other relatively rare symptoms, such as Pierre-Robin sequence (1.5%), Bifid uvula (1.5%), high-arched palate myasthenia (1.5%), and scoliosis (1.5%), were reported in less than 10% of STL type 2 cases (Zimmermann et al., 2019). Among the rare symptoms, our patient presented with mild scoliosis.

ClinVar has reported 186 pathogenic and likely pathogenic COL11A1 variants, with a total of 84 sites with splicing variants (Figure 6). For the sites with splicing variants, most variants affect +1 or +2 residues at the 5′donor splice site and −1 or −2 residues at the 3′acceptor splice site.