Clinical And Genetic Insights Into X-linked Retinoschisis And The Exploration Of Gene Therapy And Molecular Mechanisms

Objective:X-linked retinoschisis（XLRS）is an X-linked recessive inherited retinal disease,with bilateral retinal involvement being common.Typical clinical manifestations in XLRS patients include significant vision loss,macular spoke-wheel pattern observed in the fundus,intraretinal schisis,and a negative-wave pattern.However,there is considerable heterogeneity in the structural and functional changes in the retina among different XLRS patients,which presents challenges for clinical diagnosis and treatment.To date,no clear genotype-phenotype correlation has been established.Furthermore,current clinical treatments for XLRS primarily revolve around regular follow-ups,carbonic anhydrase inhibitors,and other drugs,along with the treatment of complications.Effective prevention and therapeutic methods are still lacking.The inflammatory reactions observed in ongoing gene therapy clinical trials pose significant safety concerns.Therefore,the objectives of this study are:1)To analyze the clinical characteristics and molecular diagnosis of XLRS patients,explore three-dimensional indicators of SD-OCT,and apply them in the research of genotype-phenotype correlation analysis;2)To find new breakthroughs for the study of genotype-phenotype correlation by conducting molecular dynamics simulations of RS1;3)To further explore the cyst segmentation in OCT images of XLRS,a deep learning-based automatic segmentation system is proposed,combined with a data enhancement strategy optimized by deep reinforcement learning to improve its performance;4)Aiming to construct an XLRS-specific mouse model and patient-derived iPSCs to delve deeper into the pathogenic mechanisms of XLRS,and based on the animal model,to explore gene therapy,and develop new,safe,and efficient viral/non-viral vectors.Methods:1.This study enrolled a total of 37 patients diagnosed with X-linked Retinoschisis（XLRS）.All patients were confirmed to carry RS1 gene variants through genetic testing and underwent comprehensive ophthalmic examinations,including visual acuity（VA）,fundus examination,electroretinogram（ERG）,and SD-OCT.SD-OCT parameters of central foveal thickness（CFT）,cyst cavity volume（CCV）,and photoreceptor outer segment length were assessed.CCV was defined as the sum of the areas of cyst cavities in sequential bscans,multiplied by the distance between these scans,which is measured automatically by a self-developed software OCT-CCSEG.Structural changes of the protein associated with missense variants were quantified by molecular dynamics（MD）.The correlation between genotype and phenotype was analyzed.2.A total of 30 patients diagnosed with XLRS were enrolled,with all patients undergoing bilateral eye OCT imaging.Cyst labeling of OCT in the dataset was performed on the Zeiss APEER platform,manually annotated by two ophthalmologists with over five years of experience.Five state-of-the-art deep learning（DL）models-U-Net,U-Net++,Attention U-Net,Residual U-Net,and TransUNet-were employed to perform segmentation tasks on the dataset of 1,500 OCT images from 30 XLRS patients.Data augmentation optimized with deep reinforcement learning（RL）was utilized to enhance the performance of the models.Utilizing the CytoTune-iPS 2.0 Sendai Reprogramming Kit and the ReproRNAOKSGM Kit,induced pluripotent stem cells（iPSCs）were generated from patients with Xlinked retinoschisis（XLRS）.3.A missense variant Rs1R213W mouse model was generated,and its phenotype was identified,including mouse-tail sequencing,color fundus photography,OCT,ERG,protein expression,qPCR,retinal HE/immunofluorescence staining.The natural history of disease course was evaluated using OCT and ERG,with the quantification of mouse schisis cysts being carried out by a self-developed automated measurement platform for mouse,Deep-OCT-MCCSEG.The molecular mechanisms of the disease were explored using mouse retinal RNA transcriptome sequencing analysis,followed by cytokine detection,protein quantification,and immunofluorescence for further verification.4.A non-viral vector CS-Gua-RBITC-pEGFP was synthesized and its transfection efficiency and safety both in vivo and in vitro were validated in the previously established Rs1R213W mouse model as well as in ARPE and 661W cells,exploring the feasibility of non-viral vector application for XLRS gene therapy.Meanwhile,a novel AAV（Adeno-Associated Virus）vector was developed for intravitreal injection of model mice for gene therapy after completion of vivo and in vitro transfection efficiency and safety assessment.Posttreatment evaluations using OCT and ERG were conducted to analyze the improvement in retinal structure and function by the recombinant virus.Immunofluorescence staining was utilized to evaluate gene expression and immune inflammation levels after treatment.Results:1.27 different RS1 variants were identified,including a novel variant c.336₃37insT（p.L113Sfs*8）.The average onset age was 14.76±15.75 years,and the VA was 0.84±0.43 logMAR.The mean CCV was 1.69±1.87 mm3,correlating significantly with CFT(R=0.66,p<0.01].In the genotype-phenotype analysis of missense variants,CCV significantly correlated with the structural effect on the protein of mutational changes referred to wildtype,including root-mean-square deviation（R=0.34,p=0.04）,solvent accessible surface area（R=0.38,p=0.02）,and surface hydrophobic area（R=0.37,p=0.03）.The amplitude of scotopic 3.0 ERG a wave and b wave significantly correlated with the percentage change of P-strand in secondary structure（R=-0.58,p<0.01;R=-0.53,p<0.01）.2.RL-optimized data augmentation enriched data variance and model generalization.The DL model achieved human-level accuracy in segmenting schisis cavities.Notably,U-Net++outperformed others by attaining an accuracy of 0.9927 and a Dice coefficient of 0.8568.3.In our study,we generated the Rs1R213W mouse model,which adeptly replicates the human disease phenotype in terms of both retinal structural and functional alterations.Specifically,during the observation and quantification of schisis cavities in mice,we devised a computational method for determining the mouse retinal CCV and pioneered an automated computation platform for mouse CCV,named Deep-OCT-MCCSEG.Moreover,our observation of the natural disease progression revealed that schisis manifests in the model mice between postnatal days 12-14,with the cavities subsequently enlarging,reaching their peak at 8 weeks,and thereafter diminishing with age.The ONL thickness in the model mice exhibited a declining trend from the age of 2 weeks,reflecting a progressive decrease in photoreceptor cell count with age.Concurrently,ERG evaluations corroborated the ageassociated decline in retinal function in the model mice.We observed that the expression of the RS1 gene in the Rs1P213W model mice was significantly reduced compared to wildtype mice and identified synaptic functional impairments.Immunofluorescence revealed damage to the OPL in the model mice,with mislocalization of some PSD95 and mGluR6 to the ONL.Additionally,transcriptome sequencing and inflammatory cytokine assays indicated activation of microglial cells and the related inflammatory cascades in the model mice.Beyond the animal model,this study also accomplished the establishment and validation of RS1 iPSC lines derived from two XLRS patients.4.From cellular transfection and toxicity assays,we identified the optimal grafting ratio of the non-viral vector CS-Gua,which was then subjected to animal trials.In evaluations of transduction efficiency and safety,CS-Gua exhibited impressive penetrative capabilities.Immunofluorescent staining highlighted the drug’s ability to penetrate and express within the photoreceptor layer of the retina.However,retinal toxicity became evident in fundus photography,marked by white patch-like lesions and whitening of retinal vessels postinjection in mice.Upon assessing the retinal structure,function,and RS1 protein expression following the introduction of the novel AAV gene therapy,it was discerned that the recombinant virus,when delivered via intravitreal injection to the Rs1R213W mouse model,effectively mitigated retinal schisis and fostered the preservation of both rod and cone cells.Concurrently,an enhancement in the electrophysiological responses of photoreceptor and bipolar cells was observed,aligning with therapeutic objectives.Moreover,the recombinant virus successfully triggered RS1 protein expression in the model mice,with exogenous Ra1 gene expression offering a mild inhibitory effect on the activation of retinal microglial cells.Conclusions:1.CCV is a promising indicator to quantify the structural disorganization of XLRS retina.The software OCT-CCSEG calculated CCV automatically,potentially facilitating the prognosis assessment and personalized treatment.Moreover,MD-involved genotype-phenotype analysis implied an association between the protein structural alteration and XLRS severity measured by CCV and ERG.2.By utilizing reinforcement-learning-based automatic data augmentation,deep learning segmentation models demonstrate a robust and precise method for the automated segmentation of schisis cavities in OCT images.These findings are a promising step toward enhancing clinical evaluation and treatment planning for XLRS.3.In this study,we established the Rs1r213W mouse model and generated RS1 iPSCs derived from two XLRS patients,completing the identification of both the mouse model and iPSCs.Concurrently,we explored the natural disease progression of the model mice,documenting the temporal changes in their retinal structure and function,laying a solid foundation for subsequent gene therapy research.Additionally,we observed synaptic dysfunction in the model mice,accompanied by the activation of microglial cells and associated inflammatory cascades.This suggests that interventions to inhibit microglial activation and related inflammatory responses should be implemented at early stages.4.In in vivo experiments,CS-Gua exhibited a certain degree of penetration and transduction efficiency,yet its retinal toxicity was notably pronounced.This drawback curtails its potential application as a gene therapy vector for XLRS.Conversely,the novel AAV vectors consistently demonstrated favorable transduction efficiency and safety in both in vitro and in vivo studies.In gene therapy experiments using the model mouse,the recombinant virus ameliorated retinal structure and function,achieving therapeutic objectives,and mildly suppressed the activation of microglial cells in the retina.Innovative points:1.This study pioneered the exploration of three-dimensional metrics in SD-OCT for quantifying schisis cavities and identified CCV as an effective indicator for assessing structural changes in the XLRS retina.With our proprietary OCT-CCSEG software,we achieved automated cyst identification and CCV computation,ensuring precise and efficient measurement.2.For the first time,we delved into the association between the phenotype and genotype of XLRS from the perspectives of three-dimensional metrics in SD-OCT and molecular dynamics simulation.3.We introduced deep learning to the segmentation of schisis cavities in OCT images of both XLRS patients and model mice,providing a reliable tool for disease monitoring and crafting personalized therapeutic strategies.4.We developed the Rs1R213W mouse model and established RS1 iPSCs derived from two XLRS patients,followed by thorough characterization.Observations throughout the natural disease course documented the temporal variations in retinal cyst volume in RS1 model mice for the first time.Moreover,the study advocates for early-stage intervention to inhibit the activation of microglial cells in XLRS and the associated inflammatory cascade.5.We explored the feasibility of employing the novel non-viral vector CS-Gua in XLRS gene therapy applications.Simultaneously,we innovatively constructed a new AAV vector and assessed its potential in gene therapy experiments using the model mouse.