兽研摘要
目的
视黄醇脱氢酶5（RDH5）催化视网膜色素上皮（RPE）中视觉循环的最后一步氧化反应，再生11-顺式视黄醛以用于视觉色素再生。在人类中，RDH5突变导致白点状眼底，其特征为夜盲、眼底黄白色斑点积聚，部分患者还出现黄斑变性。目前，RDH5突变患者的治疗需求尚未得到满足。本研究利用RDH5相关疾病的大型动物猫模型（RDH5−/−），旨在通过开发能够挽救RDH5相关视网膜病变的基因增强疗法来满足这些需求。
方法
将携带猫RDH5 cDNA的腺相关病毒血清型2三重衣壳突变载体（AAV-VMD2-fRDH5）经视网膜下注射至12只RDH5−/−猫。为评估表达情况及治疗对视网膜功能和结构的影响，进行了免疫组织化学、高效液相色谱、视杆和视锥视网膜电图、谱域光学相干断层扫描以及半薄树脂包埋切片检查。
结果
RDH5转基因的表达改善了光照后视杆和视锥光感受器的功能恢复，并显示出延缓/阻止中央区（类似于人类黄斑）变性的前景。
结论
该基因增强疗法是治疗RDH5相关视网膜病变的一种有前景的方法。
关键词：RDH5，白点状眼底，AAV2，基因治疗，大型动物猫模型
视黄醇脱氢酶5（RDH5）催化视网膜色素上皮（RPE）内视觉循环的最后一步氧化反应，再生11-顺式视黄醛并转运至光感受器。在此，11-顺式视黄醛与视蛋白结合，以恢复视觉色素的光接受能力。缺乏11-顺式视黄醛时，视觉色素（如视杆细胞中的视紫红质）无法接受光线，因而不能启动光转导级联反应。RDH5功能缺失突变显著减慢11-顺式视黄醛和视紫红质的再生，表现为先天性夜盲。尽管最初认为主要影响视杆细胞，但RDH5突变患者确实会出现视锥细胞功能障碍1和变性2–9，并可进展为黄斑变性（MD）10–15。此外，RDH5突变患者通常出现眼底白色斑点，故名为白点状眼底。
尽管已培育出Rdh5基因敲除小鼠，但它们不能重现人类表型，其视杆细胞功能正常，仅在长时间漂白后表现视杆细胞恢复延迟16,17。为建立更接近人类RDH5相关视网膜病变的功能模型，培育了Rdh5−/−/Rdh11−/−小鼠。这些小鼠的表型更接近人类RDH5相关视网膜病变，已被用于评估潜在的转化治疗。
对双基因敲除小鼠口服补充9-顺式视黄醛可显著改善视杆和视锥细胞功能18。基于这些发现，已在临床试验中研究补充9-顺式β-胡萝卜素对RDH5相关视网膜病变的作用，但结果相互矛盾。这些试验使用富含9-顺式β-胡萝卜素的盐生杜氏藻（Dunaliella bardawil）粉末进行口服补充。尽管一项最初为期3个月的研究显示该治疗使视网膜电图（ERG）振幅有所改善（Clinicaltrials.gov：NCT00569023）19，但日本一项更长期的研究（UMIN临床试验注册号R000020490）20显示，口服D. bardawil粉末1年后对ERG振幅产生不利影响。因此，RDH5相关视网膜病变的有效治疗需求尚未得到满足。
如前所述，RDH5−/−猫作为一种大型动物模型，重现了RDH5相关视网膜病变表型的主要特征，为研究潜在疗法提供了高质量的模型物种。这些猫不仅因视杆细胞功能障碍而出现暗适应延迟，而且其中央区（AC）变性具有变异性（类似于人类MD的变异性）21。除视杆细胞功能障碍外，我们此前还强调了该疾病对RDH5−/−猫视锥细胞的影响，包括反复暴露于明亮闪光刺激后视锥细胞恢复延迟。尽管所有RDH5−/−猫均表现出这些视网膜功能影响，但仅一部分发生AC变性。在变性早期，光感受器延长并变形。随着进展，这可导致外核层（ONL）变薄和AC光感受器变性。在一些受累猫中，这种变性进展为ONL和光感受器几乎完全丧失22。RDH5−/−猫的AC变性和光感受器功能障碍与白点状眼底患者的疾病特征非常相似，使其成为RDH5相关视网膜病变的宝贵大型动物模型。
RDH5−/−猫为开发和测试潜在的基因增强治疗提供了独特的机会。腺相关病毒（AAV）载体为将治疗基因递送至视网膜提供了安全高效的技术，包括精确靶向特定细胞层的能力。此外，眼具有免疫豁免部位的特性，可能减少对病毒载体给药的不良免疫反应。美国食品药品监督管理局（FDA）批准的首个基因治疗产品voroetigene neparvovec-rzyl（Luxturna；Genentech，南旧金山，加利福尼亚州，美国）是一种AAV2基因治疗药物，用于治疗由RPE65双等位基因突变引起的视网膜营养不良。针对其他遗传性视网膜变性的基因增强治疗的临床试验正在进行中（综述见McClements等23）。RDH5相关视网膜病变是AAV基因增强治疗的一个有吸引力的靶标，原因有以下几点。RDH5 cDNA基因尺寸相对较小，使其能够被包装入AAV载体进行视网膜递送。视杆细胞功能受损但无显著视杆细胞丢失与黄斑营养不良缓慢进展之间的分离为治疗干预提供了宽广的时间窗口。我们报道了在RDH5−/−猫中成功实施的、具有潜在转化价值的基因增强治疗，该治疗既解决了夜盲症，又显示出阻止或延缓AC（黄斑）变性的潜力。
材料与方法
载体制备
载体由佛罗里达大学载体核心实验室（Powell基因治疗中心）制备。使用了两个治疗性基因构建体和一个报告基因构建体。最初，测试了由卵黄样黄斑营养不良（VMD2）启动子控制猫RDH5 cDNA的直载AAV2血清型，但未能在RPE中实现RDH5转基因表达。测试了包装于带有三个衣壳突变（Y444F+Y500F+Y730F）的AAV2中的报告基因构建体（VMD2-人源化绿色荧光蛋白[hGFP]）。最终治疗性构建体使用了包装VMD2-fRDH5的AAV2三突变血清型。
动物
所有动物实验程序均经密歇根州立大学机构动物护理与使用委员会审查批准，并遵循ARVO眼科与视觉研究中动物使用声明。猫饲养于密歇根州立大学兽医学院动物饲养设施中，保持12小时光照/12小时黑暗周期。本研究使用的所有猫均来自先前描述的繁殖群体。21两只RDH5−/−猫和一只对照猫被注射用于初步的载体和报告基因研究。十二只RDH5−/−猫接受视网膜下注射三衣壳突变治疗载体，剂量为5 × 1011 vg/mL，并随访评估治疗效果（补充表S1和S2提供完整的注射和操作信息）。其中，两只RDH5−/−猫注射较少载体体积以创建靶向AC的小疱，目的是评估治疗对变性的效果。另外两只RDH5−/−猫注射的目的是评估治疗区域与未治疗区域之间的类视黄醇差异。最后，八只RDH5−/−猫注射较大载体体积以创建大疱，目的是评估治疗对 retinal 功能的影响。小疱和大疱的示例可见于补充图S1。估计疱大小使用Heidelberg Engineering光学相干断层扫描（OCT）软件中的测量功能进行测量。ERG野生型（WT）数值见补充材料中的表格。
视网膜下注射
注射操作采用标准无菌手术入路。使用1%托吡卡胺（USP，Falcon Pharmaceuticals，美国德克萨斯州沃思堡）散瞳，并在整个操作过程中保持角膜湿润和受保护。手术在手术显微镜下进行，猫术前使用氯胺酮（2–6 mg/kg；MWI Animal Health，美国爱达荷州博伊西）和咪达唑仑（0.2–0.5 mg/kg；Gland Pharma Limited，印度特伦甘纳邦）进行前驱麻醉。通过面罩给予异氟烷（VetOne，法国巴黎）进行诱导，随后插管并维持异氟烷麻醉。在手术显微镜下摆好体位后，翻开一部分结膜，以便通过睫状体平坦部放置自封闭式25号巩膜切开端口。将注射套管置入玻璃体，并通过置于角膜上的玻璃体切除透镜观察后节。将注射器经巩膜端口推进，在直视下将39号套管推进穿透视网膜并进入视网膜下区域，在此注射载体以形成视网膜脱离区域（ bleb）。移除端口，缝合结膜（6-0 Vicryl；Ethicon，美国新泽西州萨默维尔）。结膜下注射类固醇/抗生素：地塞米松（0.2 mg；Bimeda US，美国伊利诺伊州绍姆堡）、醋酸甲泼尼龙（Depo-Medrol，4 mg；Zoetis，美国新泽西州帕西帕尼）和庆大霉素（4 mg；Fresenius Kabi，德国巴特洪堡）。采集眼底图像以记录 bleb 的位置。在10只RDH5−/−猫中，一只眼接受治疗，另一只眼作为未治疗对照用于比较。两只RDH5−/−猫接受双侧治疗。两只RDH5−/−猫的一只眼注射原始AAV2载体，一只对照猫用于报告基因研究。动物被密切监测直至从麻醉中完全恢复，随后返回动物饲养设施。注射前开始口服泼尼松龙（1 mg/kg，每日两次）（MWI Animal Health），并以递减剂量维持10天。所有视网膜下注射的目标是形成覆盖AC的 bleb；然而，由于该操作难以控制 bleb 扩展方向，并非所有猫均在AC上方形成 bleb（补充表S1）。
注射前后随访
如在全身麻醉下先前所述，动物在注射前接受了基线活体共聚焦扫描激光检眼镜（cSLO）和频域OCT（SD-OCT）成像（SPECTRALIS HRA+OCT；Heidelberg Engineering，德国海德堡）。在SD-OCT时使用Retcam II视频眼底相机（Clarity Medical Systems，美国加利福尼亚州普莱森顿）获取广角眼底检查图像。我们先前已表明，RDH5−/−猫的疾病相关改变呈双侧对称，因此每只动物的未治疗眼是合适的对照。
该治疗的主要结局指标是光照后视杆细胞恢复率的改善。这通过视杆细胞暗适应ERG方案测量。作为次要结局指标，使用其他ERG方案（强闪光、视锥细胞恢复、暗适应和明适应）对动物进行检查。主要和次要ERG结局指标、SD-OCT和Retcam图像在注射后3个月、6个月和1年时采集。在每个时间点测量AC的外核层（ONL）厚度，以确定治疗是否对视网膜结构有影响。每只眼测量三次并取平均值。仅使用AC上方有视网膜下 bleb 的RDH5−/−猫来确定治疗是否保留结构，因为AC是主要变性区域。
视网膜电图
通过ERG评估基因治疗对视网膜功能恢复的影响。动物暗适应过夜，使RDH5−/−猫达到完全暗适应以进行基线ERG。使用托吡卡胺滴眼液（USP 1%）散瞳。猫术前使用氯胺酮（2–6 mg/kg）和咪达唑仑（0.2–0.5 mg/kg）前驱麻醉。在暗红光照明下麻醉动物并放置电极。插管后，使用异氟烷维持动物麻醉。使用角巩膜缘周围结膜的固定缝线（4-0 734G PERMA-HAND Silk Suture；Ethicon）将眼固定于初级眼位。参考铂针皮肤电极置于外眦外侧5 mm处，接地电极置于肩胛骨之间。将ERG-Jet单极角膜接触电极（Fabrinal Eye Care，瑞士拉绍德封）加一滴Optixcare Eye Lube Plus（Aventix，加拿大安大略省伯灵顿）置于每只眼上。使用Espion E2或E3电生理系统配合ColorDome Ganzfeld（Diagnosys，美国马萨诸塞州洛厄尔）记录ERG。
暗适应恢复
为评估治疗后暗适应的恢复，采用先前描述的ERG方案。过夜暗适应和基线暗适应视杆细胞ERG记录后，将猫暴露于30 cd/m2的视杆细胞抑制光下10分钟。10分钟后，移除背景光刺激，将动物保持在黑暗中，在100分钟内定期进行视杆细胞ERG记录。结果报告为光照后100分钟时间点相对于基线的百分比。
反复强光暴露下的视锥细胞ERG（强闪光和视锥细胞恢复方案）
视锥细胞对较强光刺激产生反应，并且通常从刺激中快速恢复。我们此前已证明，在RDH5−/−猫中，一种在重复强光刺激后测量视锥功能的方案显示视锥恢复受损。22 该方案由交替的7.2 cd⋅s/m2闪光（记录ERG）与强光暴露（由四次1000-cd⋅s/m2闪光组成）交替进行。每组闪光之间间隔5秒，共重复25次，以评估该治疗对强光刺激累积效应的影响。该方案称为强闪光方案，结果以方案最后一步时强闪光基线的百分比报告。紧接强闪光方案之后，在抑制视杆的背景光存在下，通过每5分钟记录一次对7.2-cd⋅s/m2闪光的ERG反应，在120分钟内评估视锥功能的恢复。该方案称为视锥恢复方案，结果以120分钟时间点时强闪光基线的百分比报告。
暗适应与明适应比较
按照先前描述的方法，先进行标准的暗适应暗视亮度系列（20个步阶，范围从–4.5 log cd⋅s/m2至1.36 log cd⋅s/m2），然后暴露于30-cd/m2的抑制视杆白光，再进行明适应亮度系列（12个步阶，范围从–2.4至1.36 log cd⋅s/m2）和33-Hz 3-cd⋅s/m2闪烁光。25 分析时比较治疗眼与未治疗眼在2.482、7.2和23 cd⋅s/m2闪光强度下以及33-Hz闪烁光下的明视振幅和峰值时间。
波形建模
所有ERG的a波和b波振幅及峰值时间均使用国际临床视觉电生理学会（ISCEV）标准方案测量。26 按照先前描述，使用Naka–Rushton函数拟合暗视b波亮度-反应曲线的第一段。22 此外，使用Lamb和Pugh视杆光转导模型的Birch和Hood版本，对0.996至7.2 cd⋅s/m2闪光范围内的暗视视杆a波前缘进行建模。22
ERG的统计分析
使用Shapiro–Wilk检验对ERG数据进行正态性检验。如果数据呈正态分布，则进行治疗与未治疗个体之间的配对t检验，以及WT与治疗组之间和WT与未治疗组之间的非配对t检验。如果数据不呈正态分布，则进行非参数Wilcoxon符号秩检验。所有ERG统计分析均在RStudio 2023.06.1+524中完成。27 配对和非配对t检验以及Wilcoxon符号秩检验使用R包stats进行。在可能的情况下报告95%置信区间，或当无法计算95%置信区间时报告可达到的最高置信区间。由于样本量小、零差异和配对比较中的并列秩，置信区间并不总是可估计。由于视网膜治疗区域有限，有小疱的动物被排除在ERG功能统计比较之外。此外，双眼均注射的动物（无对照眼）被排除在ERG比较之外。参数数据的效应量使用Hedges’ g计算，非参数数据使用配对秩双列相关。效应量使用RStudio中的effectsize包计算。28
免疫组织化学和蛋白质印迹
在安乐死后，取出眼睛并固定在磷酸盐缓冲盐水（PBS；Sigma-Aldrich，St. Louis，MO，USA）中的4%多聚甲醛（PFA；Electron Microscopy Sciences，Hatfield，PA，USA）中，并按照先前描述包埋于最佳切割介质（TissueTek OCT；Electron Microscopy Sciences）中进行冰冻切片。29 对于IHC，使用穿过治疗视网膜的14-µm冷冻切片。使用浓度为1:100的RDH5兔源一抗（NBP2-15097；Novus Biologicals，Centennial，CO，USA），并使用浓度为1:500的Invitrogen Alexa Fluor 488山羊抗兔二抗（A11008；Thermo Fisher Scientific，Waltham，MA，USA）。所有玻片均使用配备pco.panda 4.2 sCMOS相机（70108074001；Excelitas，Pittsburgh，PA，USA）的Eclipse 80i荧光显微镜（Nikon Corporation，Tokyo，Japan）成像。
对于蛋白质印迹，解剖视网膜RPE脉络膜并置于含蛋白酶抑制剂的400-µL放射免疫沉淀分析（RIPA）缓冲液中，按照先前描述处理。21 然而，RDH5抗体稀释至1:500，α-微管蛋白稀释至1:2000，并使用化学发光底物（34076；Thermo Fisher Scientific）。视网膜RPE脉络膜取自同一眼的治疗疱区域和未治疗区域（n = 1）。收集一只WT猫的视网膜RPE脉络膜。使用ImageJ（National Institutes of Health，Bethesda，MD，USA）通过密度分析对蛋白质印迹进行定量。30 RDH5蛋白归一化至相应的上样对照（α-微管蛋白），并计算RDH5蛋白/α-微管蛋白比值。
树脂包埋半薄切片
按照先前描述的方法，实施安乐死后摘除眼球，并固定于含3%戊二醛和2%多聚甲醛的0.1 M二甲砷酸钠缓冲液（Electron Microscopy Sciences）中。固定后，从两只RDH5−/−猫接受治疗和未接受治疗眼的前房（AC）中采集直径3 mm的活检打孔样本，并保持结构完整。利用显示注射泡和血管 pattern 以识别前房的彩色眼底图像引导组织采集。活检打孔样本按先前描述的方法进行树脂包埋。随后获取1000 nm厚的切片，用环氧组织染色液（Electron Microscopy Sciences）染色，并在Nikon Eclipse 80i光学显微镜上成像。
类视黄醇的提取与分析
RDH5−/−（n = 2）和RDH5+/+（n = 1）猫暗适应过夜，随后短暂暴露于光（30 cd/m²，10分钟），并在安乐死前暗适应100分钟。在暗光下摘除眼球，从视网膜完整（包含RPE和脉络膜）的治疗区和未治疗区获取活检样本。视网膜打孔样本速冻并保存于–80°C。类视黄醇分析的所有操作均在暗红光下进行。
类视黄醇提取时，将样本转移至冰冷的缓冲溶液中，该溶液由10 mM磷酸钠（pH 8.0）、100 mM羟胺（Sigma-Aldrich）和50% v/v甲醇（Sigma-Aldrich）组成；用组织研磨器匀浆；并在22°C孵育15分钟。随后向样本中加入3 mL 100%乙酸乙酯（Thermo Fisher Scientific）和2 mL 3 M NaCl。剧烈振荡混合物后离心以促进相分离，收集上层有机相至硼硅酸盐管中。再加入3 mL乙酸乙酯并重复提取过程。合并含有提取类视黄醇的有机层，真空干燥。干燥样本用450 µL 100%己烷（Thermo Fisher Scientific）复溶，取100 µL部分通过正相高效液相色谱（HPLC；1260 Infinity II LC系统，配备二极管阵列检测器；Agilent Technologies，Santa Clara，CA，USA）使用Zorbax Rx-SIL柱（5 µm，4.6 × 250 mm；Agilent Technologies）进行分析。采用阶梯洗脱方案，溶剂流速为1.4 mL/min，首先用含0.6% v/v乙酸乙酯的己烷洗脱17分钟，然后用含10% v/v乙酸乙酯的己烷洗脱45分钟。在325 nm和360 nm处监测类视黄醇洗脱。使用ChemStation软件（Agilent Technologies）积分类视黄醇峰面积。类视黄醇标准曲线可将11-顺式-视黄醛肟和视黄醇酯峰面积转换为绝对质量。原始色谱图和定量类视黄醇种类用Prism（GraphPad Software，Boston，MA，USA）绘制和分析。
结果
AAV2-VMD2-fRDH5不导致转基因表达
将带有天然AAV2衣壳的初始构建体视网膜下注射到两只RDH5−/−猫的一只眼中。未检测到ERG拯救，也未检测到视网膜中RDH5的表达（数据未显示）。
带有AAV2三重衣壳突变的载体驱动RPE表达并改善经典视觉周期
为确认AAV2三重突变衣壳连同VMD2启动子能驱动转基因在RPE中的表达，使用对照猫进行了初步报告基因（hGFP）研究。这证实了GFP转基因存在且特异于RPE（补充图S2）。治疗眼的免疫组织化学和蛋白质印迹证实，RDH5转基因的表达仅发生在RDH5−/−猫治疗区域的RPE中（图1，补充图S3）。HPLC证明表达的RDH5具有功能。在短暂光漂白后暗适应100分钟，收集视网膜样本。来自WT猫的对照样本显示存在11-顺式-视黄醛，但与未治疗区域相比，RDH5−/−猫的治疗视网膜区域显示全反式视黄醇酯减少和11-顺式-视黄醛形成增加（图2，补充图S4，补充表S3）。与RDH5功能丧失相关的13-顺式-视黄醇酯异常存在也被基因治疗纠正。总之，数据表明基因治疗纠正了与RDH5功能丧失相关的视觉周期生化缺陷。
[图：在RDH5−/−猫视网膜下注射AAV2（三重Y>F）-VMD2-fRDH5载体。（A1）注射后眼底图像显示]
[图：基因治疗后RDH5−/−猫视网膜的类视黄醇分析。（A）HPLC色谱图（325 nm波长）]
基因治疗改善光照后光感受器恢复
过夜暗适应后的暗适应ERG在RDH5−/−和WT猫之间相当，仅在某些参数上有少数轻微变化，如我们先前报道。在本研究中，我们发现过夜暗适应后治疗眼和未治疗眼之间的Naka–Rushton和a波建模参数差异极少（差异在补充表S4中突出显示）。同样，对于常规明适应ERG，我们先前报道RDH5−/−和WT猫之间差异很小，并且不出所料，本文中治疗眼和未治疗眼之间的差异可忽略不计（补充表S5）。
在漂白处理后，经ERG评估，RDH5−/−猫在100分钟内的视杆功能恢复极少，而野生型猫几乎完全恢复。与未治疗眼相比，基因治疗眼的视杆恢复程度（图3）显著改善（3个月，P < 0.001；6个月，P = 0.008；1年，P = 0.03）（完整统计报告见补充表S6）。效应量分析表明治疗效果显著，注射后3个月Hedges' g值为2.21，1年为1.18，注射后6个月秩双列相关系数为1.00。由于仅治疗了部分视网膜，全视野闪光ERG记录的是整个视网膜（治疗区和未治疗区）的响应，因此预期不会显示与野生型猫相当的视杆功能恢复。
[图：RDH5−/−治疗眼与未治疗眼及野生型的暗适应恢复。（A）代表性ERG波形]
除了视杆光感受器光照后恢复延迟外，我们还评估了治疗RDH5−/−猫视锥细胞在反复暴露于强闪光并在强闪光间进行常规ERG记录的方案后的恢复能力。前述方案称为强闪光方案。在反复闪光暴露期间，明视ERG振幅出现下降。治疗眼与未治疗眼之间这些ERG振幅无显著差异（补充表S7）。在视锥功能被抑制后，在随后120分钟内无进一步强光刺激的条件下追踪视锥反应的恢复。该方案称为视锥恢复方案。与未治疗对照眼相比，治疗眼在所有注射后时间点以强闪光基线百分比评估的b波振幅恢复均显著更大（3个月，0.007；6个月，0.006；1年，0.03）（图4）。效应量分析表明治疗效果显著，注射后3个月Hedges' g值为1.50，6个月为1.38，1年为1.82。在任何时间点，治疗眼与未治疗眼之间a波占基线的百分比均无显著差异（补充表S7）。
[图：治疗与未治疗RDH5−/−眼及野生型的视锥恢复ERG。（A）6个月时的平均波形]
治疗提供AC的结构性保护
除了基因增强治疗带来的功能改善外，还使用SD-OCT成像评估了RDH5−/−猫的结构性保护。一部分RDH5−/−猫在AC出现了双侧对称的进行性外层视网膜变薄。两只注射泡覆盖AC的猫在未治疗对照眼出现了AC变性。尽管两只动物均显示出治疗眼AC区域结构性保护的明确证据，但保护程度不同。在一只RDH5−/−治疗猫（RSA 20-053）中，治疗区AC与未治疗区AC相比几乎无变性。在另一只RDH5−/−治疗猫（RSA 20-054）中，治疗眼变性减慢但未停止。两只RDH5−/−猫AC的半薄电镜切片中也可观察到这种结构性保护（图5）。
[图：基因治疗导致的结构性保护。（A）RDH5−/−猫#20-053的RetCam图像显示治疗注射泡区域]
不良反应
注射耐受良好，除视网膜下注射所致的典型变化外无炎症。然而，两只治疗眼出现了初期非进行性白内障。一只在注射后3个月出现白内障，另一只在注射后1年出现。尚不清楚白内障是否由治疗直接导致。除注射视网膜切开术的瘢痕外，治疗注射泡区域无相关视网膜变性（补充图S5）。
讨论
目前RDH5相关视网膜病变患者的治疗需求尚未得到满足。尽管夜盲表型可能可控，但进行性MD的潜在风险威胁着正常日常生活活动所需的高锐度视力。AAV基因增强被认为是合适的方法。AAV载体正用于许多临床试验并取得有希望的结果23，且FDA已批准一种治疗另一种RPE视觉周期基因RPE65功能缺失的疗法31。该疗法应安全并改善光感受器11-顺式视黄醛的供应，理想情况下提供改善的视杆光感受器恢复并保护视锥和黄斑免于变性。在此，我们报告了一种有希望的AAV疗法，为治疗RDH5视网膜病变提供了概念验证。
我们通过报告基因研究鉴定出一种AAV血清型/启动子组合，经视网膜下注射可高效转导猫视网膜色素上皮（RPE）。该载体为携带三重衣壳Y>F修饰的AAV2载体，偶联VMD2启动子。利用该血清型/启动子组合构建了递送猫RDH5 cDNA的治疗性构建体。免疫组化（IHC）和蛋白质印迹证实，该治疗性构建体如预期仅在视网膜下疱区域内引起RDH5表达。对治疗视网膜样本的HPLC分析显示，异常13-顺式视黄酯被清除，11-顺式视黄醇向视紫红质再生所需的11-顺式视黄醛的转化改善。这反映在漂白后视杆细胞ERG恢复的改善上。
重复强光暴露（视锥恢复方案）导致视锥功能抑制后，治疗眼与未治疗眼之间b波恢复百分比的显著改善表明，补充RDH5也改善了视锥细胞功能。未治疗眼从经典RPE视觉循环获得的11-顺式视黄醛供应大幅减少，可能意味着它们依赖视网膜内视觉循环来重新激活视锥蛋白。随着持续强光暴露，视网膜内视觉循环提供的11-顺式视黄醛很可能将被耗竭。补充RDH5后，通过经典视觉循环可获得更多11-顺式视黄醛，使视锥细胞比未治疗的RDH5−/−视网膜中的视锥细胞恢复得更快。预期该治疗可改善RPE视觉循环的11-顺式视黄醛供应，但对视网膜内视觉循环无影响。
尽管治疗在某些ERG参数上导致治疗眼与未治疗眼之间存在显著差异，但治疗眼在这些参数上未达到野生型（WT）对照水平。治疗疱仅覆盖视网膜的一小部分，且所有ERG均采用全视野闪光Ganzfeld，汇总来自整个视网膜（治疗区和未治疗区）的输入。因此，即使在治疗疱内实现最佳转导，反应仍低于正常WT视网膜。
在两只成功靶向AC（前部脉络膜/睫状体区域）且在未治疗眼出现AC变性的RDH5−/−治疗猫中，观察到AC的结构性ONL（外核层）保留。有时难以预测注射疱的扩展方向，这意味着即使意图包括AC，疱也未必扩展到包括AC。此外，并非研究中所有RDH5−/−猫都出现AC变性。在两只治疗包括AC且未治疗眼显示AC变性的猫中，一只猫的AC变性减缓，另一只则停止。这些猫随访了4年。这两只RDH5−/−猫之间AC保留程度的差异可能反映了两只猫之间RPE转导效率的差异或背景遗传等因素。这些结果是初步的，需要进一步研究来探究治疗对光感受器的保护作用并理解结构挽救的差异。
该基因增强疗法改善了视杆和视锥细胞的功能恢复，并显示出能够减缓甚至可能阻止RDH5−/−猫AC变性的前景。该疗法的未来方向包括继续长期研究AC变性是被治疗阻止还是仅被减缓。总之，这项初步研究表明，RDH5 AAV2基因增强疗法提供了一种有前景且可能可转化的疗法，不仅能够改善视网膜功能，还能减缓部分RDH5相关视网膜病变患者中出现的致衰弱性MD表型。
补充材料
[内容请查看原文]
致谢
作者感谢Paige Winkler Smith在半薄切片和技术支持方面的帮助；Alicia Withrow博士在半薄和薄切片方面的帮助；Janice Querubin在部分动物处理和部分ERG/OCT数据采集方面的帮助；Lexi Daggy在部分动物处理和麻醉方面的帮助；Billie Beckwith-Cohen和Geneive Carrion的手术协助；以及密歇根州立大学兽医学院研究与教学技术支持部门和密歇根州立大学校园动物资源组在动物护理方面的帮助。我们还感谢佛罗里达大学的William Hauswirth博士提供RDH5载体生产。
由密歇根州立大学Myers-Dunlap犬健康 endowment 资助；George H. Bord和“Casper”猫科倡议 endowment；密歇根州立大学猫健康与福祉中心；美国国立卫生研究院资助（R01EY009339 to PDK）；美国退伍军人事务部资助（I02BX004939和IK6BX006800 to PDK）；UC Irvine眼科与视觉科学系的无限制Research to Prevent Blindness奖；以及美国国家眼科研究所UCI视觉研究中心核心资助（P30EY034070）。
数据可用性：数据可应要求提供。
利益披露：L.M. Ford，无；K. Sun，无；L.M. Occelli，无；M.E. Enfield，无；A. Daruwalla，无；R.F. Boyd，无；P.D. Kiser，无；S.M. Petersen-Jones，无
参考文献
[内容请查看原文]
关联数据
补充材料
[内容请查看原文]
查看英文原文▾
Abstract
Purpose
Retinol dehydrogenase 5 (RDH5) catalyzes the final oxidation for visual pigment regeneration of 11-cis-retinal as part of the visual cycle in the retinal pigment epithelium (RPE). In humans, RDH5 mutations cause fundus albipunctatus characterized by night blindness, accumulation of white-yellowish fundus spots, and, in a subset of patients, macular degeneration. Currently, there is an unmet need for therapies for patients with RDH5 mutations. Using a large-animal cat model of RDH5-associated disease (RDH5−/−), this study aimed to meet these needs through development of a gene augmentation therapy able to rescue RDH5-associated retinopathies.
Methods
An adeno-associated virus serotype 2 with triple capsid mutations vector packaged with feline RDH5 cDNA (AAV-VMD2-fRDH5) was subretinally injected into 12 RDH5−/− cats. To assess expression and the impact of treatment on retinal function and structure, immunohistochemistry, high-performance liquid chromatography, rod and cone electroretinograms, spectral-domain optical coherence tomography, and semi-thin resin-embedded sectioning were performed.
Results
The expression of the RDH5 transgene improved rod and cone photoreceptor functional recovery following light exposure and shows promise in slowing/halting area centralis (analogous to the human macula) degeneration.
Conclusions
This gene augmentation therapy is a promising approach for the treatment of RDH5-associated retinopathies.
Keywords: RDH5, fundus albipunctatus, AAV2, gene therapy, large-animal cat model
Retinol dehydrogenase 5 (RDH5) catalyzes the final oxidation of the visual cycle within the retinal pigment epithelium (RPE), regenerating 11-cis-retinal for transport to the photoreceptors. Here, 11-cis-retinal combines with opsins to renew the light-accepting ability of the visual pigments. Without 11-cis-retinal, the visual pigments (e.g., rhodopsin in rods) are unable to accept light and therefore do not initiate the phototransduction cascade. Loss-of-function mutations in RDH5 markedly slow regeneration of 11-cis-retinal and rhodopsin, manifesting as congenital night blindness. Although initially thought to primarily affect rods, cone dysfunction1 and degeneration do occur in patients with RDH5 mutations,2–9 and this can progress to macular degeneration (MD).10–15 Additionally, patients with RDH5 mutations typically develop white fundus spots, thus the name fundus albipunctatus.
Although Rdh5 knockout mice have been generated, they do not recapitulate the human phenotype, having normal rod function and showing delayed rod recovery only after prolonged bleaching.16,17 To develop a better functional model for human RDH5-associated retinopathy, Rdh5−/−/Rdh11−/− mice were generated. These mice have a phenotype that more closely resembles RDH5-associated retinopathy in humans and have been used to assess potential translational therapies.
Oral supplementation of the double-knockout mice with 9-cis-retinal significantly improved rod and cone function.18 Based on these findings, supplementation with 9-cis-beta carotene has been investigated in clinical trials for RDH5-associated retinopathy, although with contradictory results. The trials used oral supplementation with 9-cis-beta carotene–rich alga (Dunaliella bardawil) powder. Although an initial 3-month-long study showed that therapy resulted in some electroretinogram (ERG) amplitude improvement (Clinicaltrials.gov: NCT00569023),19 a longer-term study from Japan (UMIN Clinical Trials Registry R000020490)20 showed a detrimental effect on ERG amplitude after 1 year of oral dosing with the D. bardawil powder. There is therefore an unmet need for an effective therapy for RDH5-associated retinopathies.
As previously described, RDH5−/− cats, a large-animal model, recapitulate major features of the RDH5-associated retinopathy phenotype offering a quality model species for studying potential therapies. Not only do these cats have delayed dark adaptation caused by rod dysfunction, but they also have variability in area centralis (AC) degeneration (analogous to variable human MD).21 In addition to rod dysfunction, we previously highlighted the effects of this disease on cones in RDH5−/− cats, including delayed cone recovery after exposure to repeated bright flash stimuli. Although all RDH5−/− cats exhibit these effects on retinal function, only a subset develops AC degeneration. In the early stages of degeneration, the photoreceptors are lengthened and distorted. With progression this can lead to thinning of the outer nuclear layer (ONL) and degeneration of the photoreceptors of the AC. In some affected cats, this degeneration advances to an almost complete loss of the ONL and photoreceptors.22 RDH5−/− cats’ AC degeneration and photoreceptor dysfunction are very similar to the features of disease seen in patients with fundus albipunctatus, making them a valuable large-animal model for RDH5-associated retinopathies.
RDH5 −/− cats present a unique opportunity to develop and test potential gene augmentation therapy treatments. Adeno-associated virus (AAV) vectors offer a safe and efficient technology for delivery of therapeutic genes to the retina, including the ability for precision targeting of specific cell layers. Additionally, the eye offers an immune privileged site potentially reducing adverse immune responses to the viral vector dosing. The first U.S. Food and Drug Administration (FDA)-approved gene therapy product, voroetigene neparvovec-rzyl (Luxturna; Genentech, South San Francisco, CA, USA), is an AAV2 gene therapy for retinal dystrophy due to biallelic mutations in RPE65. Clinical trials for gene augmentation therapy for other inherited retinal degenerations are underway (for a review, see McClements et al.23). RDH5-associated retinopathy is an attractive target for AAV-based gene augmentation therapy for several reasons. The relatively small gene size of RDH5 cDNA enables it to be packaged into an AAV vector for retinal delivery. The disconnect between impaired rod function without significant rod loss and the slow development of MD provides a wide window of opportunity for therapeutic intervention. We report a successful, potentially translatable gene augmentation therapy in RDH5−/− cats that both resolves night blindness and shows potential to halt or slow AC (macular) degeneration.
Materials and Methods
Vector Production
Vectors were produced by the University of Florida Vector Core Lab (Powell Gene Therapy Center). Two therapeutic and one reporter gene constructs were used. Initially, a straight AAV2 serotype packaged with feline RDH5 cDNA under control of a vitelliform macular dystrophy (VMD2) promoter was tested and failed to result in RDH5 transgene expression in the RPE. A reporter gene construct (VMD2-humanized green fluorescent protein [hGFP]) package in an AAV2 with three capsid mutations (Y444F+Y500F+Y730F) was tested. For the final therapeutic construct, the AAV2 triple mutant serotype packaged with VMD2-fRDH5 was used.
Animals
All animal procedures were reviewed and approved by the Michigan State University Institutional Animal Care and Use Committee and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Cats were housed at the Michigan State University College of Veterinary Medicine vivarium and kept under a 12-hour light/12-hour dark cycle. All cats used in this study originated from a breeding colony described previously.21 Two RDH5−/− cats and one control cat were injected for a preliminary vector and reporter gene study. Twelve RDH5−/− cats were subretinally injected with the triple capsid mutation therapeutic vector at a dose of 5 × 1011 vg/mL and followed to assess efficacy of the therapy (Supplementary Tables S1 and S2 provide full injection and procedure information). Of these, two RDH5−/− cats were injected with less vector volume to create small blebs targeting the AC with the purpose of assessing the therapeutic effect on degeneration. An additional two RDH5−/− cats were injected with the purpose of assessing retinoid differences between treated and untreated regions. Finally, eight RDH5−/− cats were injected with larger vector volumes to create large blebs with the purpose of assessing the effect of the therapy on retinal function. Examples of a small and a large bleb can be seen in Supplementary Figure S1. Estimated bleb size was measured in the Heidelberg Engineering optical coherence tomography (OCT) software using the measure function. See tables in the Supplementary Materials for the ERG wild-type (WT) numbers.
Subretinal Injection
Standard aseptic surgical approaches were utilized for the injection procedure. Pupils were dilated with tropicamide, USP 1% (Falcon Pharmaceuticals, Fort Worth, TX, USA), and corneas were kept hydrated and protected for the duration. The procedure was performed under an operating microscope, and cats were premedicated with ketamine (2–6 mg/kg; MWI Animal Health, Boise, ID, USA) and midazolam (0.2–0.5 mg/kg; Gland Pharma Limited, Telangana, India). They were induced with isoflurane (VetOne, Paris, France) delivered by mask and then intubated and maintained on isoflurane. After positioning under the operating microscope, a region of conjunctiva was reflected to allow placement of a self-sealing 25-gauge sclerotomy port through pars plana. The injection cannula was placed into the vitreous, and the posterior segment was viewed with a corneal-placed vitrectomy lens. The injector was advanced through the scleral port, and under direct visualization the 39-gauge cannula advanced to penetrate the retina and into the subretinal area, where the vector was injected to create a region of retinal detachment (bleb). The port was removed, and the conjunctiva was sutured (6-0 Vicryl; Ethicon, Somerville, NJ, USA). Subconjunctival steroid/antibiotics, dexamethasone (0.2 mg; Bimeda US, Schaumburg, IL, USA), methylprednisolone acetate (Depo-Medrol, 4 mg; Zoetis, Parsippany, NJ, USA), and gentamicin (4 mg; Fresenius Kabi, Bad Homburg, Germany) were injected. Fundus images were collected to document the position of the bleb. One eye was treated, and the other remained an untreated control for comparison in 10 RDH5−/− cats. Two RDH5−/− cats were treated bilaterally. One eye of two RDH5−/− cats was injected with the original AAV2 vector and one control cat for the reporter gene study. Animals were monitored closely until fully recovered from anesthesia, at which time they returned to the animal housing. Oral prednisolone (1 mg/kg twice daily) (MWI Animal Health) was started prior to injection and maintained on a tapering dosage for 10 days. The goal of all subretinal injections was a bleb encompassing the AC; however, as this is a difficult procedure to control direction of bleb expansion, not all cats received a bleb over the AC (Supplementary Table S1).
Pre- and Post-Injection Follow-Up
Animals underwent baseline in vivo confocal scanning laser ophthalmoscopy (cSLO) and spectral-domain OCT (SD-OCT) imaging (SPECTRALIS HRA+OCT; Heidelberg Engineering, Heidelberg, Germany) pre-injection under general anesthesia, as previously described.24 Wide-angle fundus examination images were also obtained at the time of SD-OCT using a Retcam II video fundus camera (Clarity Medical Systems, Pleasanton, CA, USA). We have previously shown that the disease-related changes in RDH5−/− cats are bilaterally symmetrical, making the untreated eye of each animal an appropriate control.21
The primary outcome measure for this therapeutic was an improvement in the rate of rod recovery following light exposure. This was measured using the rod dark-adaptation ERG protocol. As a secondary outcome measure, animals were investigated using other ERG protocols (bright flash, cone recovery, scotopic, and photopic). Both primary and secondary ERG outcome measures, SD-OCT, and Retcam images were taken at 3 months, 6 months, and 1 year post-injection. The outer nuclear layer (ONL) thickness of the AC was measured at each time point to determine if the therapy had an effect on retinal structure. Three measurements were taken and averaged together for each eye. Only RDH5−/− cats with a subretinal bleb over the AC were used to determine whether therapy preserves structure, as the AC is the main area of degeneration.
Electroretinography
The effect of the gene therapy on retinal functional recovery was assessed by ERG. Animals were dark adapted overnight to get RDH5−/− cats to full dark adaptation for a baseline ERG. Pupils were dilated with tropicamide ophthalmic solution (USP 1%). Cats were premedicated with ketamine (2–6 mg/kg) and midazolam (0.2–0.5 mg/kg). Animals were anesthetized and electrodes placed under illumination of a dim red light. Following intubation, animals were maintained with isoflurane. Eyes were positioned in a primary gaze using stay sutures in the perilimbal conjunctiva (4-0 734G PERMA-HAND Silk Suture; Ethicon). Reference platinum needle skin electrodes were placed 5 mm laterally from the lateral canthus, and the ground electrode was placed in-between the shoulder blades. An ERG-Jet monopolar corneal contact electrode (Fabrinal Eye Care, La Chaux-De-Fonds, Switzerland) with a drop of Optixcare Eye Lube Plus (Aventix, Burlington, ON, Canada) was placed on each eye. ERGs were recorded using an Espion E2 or E3 electrophysiology system with a ColorDome Ganzfeld (Diagnosys, Lowell, MA, USA).
Recovery of Dark Adaptation
To assess recovery of dark adaptation following therapy, a previously described ERG protocol was utilized.21 Following overnight dark adaptation and a baseline dark-adapted rod ERG recording, cats were exposed to a 30-cd/m2 rod-suppressing light for 10 minutes. After 10 minutes, the background light stimulus was removed, and the animals were maintained in the dark with periodic rod ERG recordings taken over the course of 100 minutes. Results are reported as a percentage of baseline at the 100-minute post-light exposure time point.
Cone ERG With Repeated Strong Light Exposure (Bright Flash and Cone Recovery Protocols)
Cones respond to stronger light stimuli and typically have rapid recovery from a stimulus. We previously showed that a protocol measuring cone function after repeated strong light stimulus showed impairment of cone recovery in the RDH5−/− cats.22 This protocol consists of alternating flashes of 7.2 cd⋅s/m2 from which the ERG is recorded interspersed with strong light exposure consisting of four 1000-cd⋅s/m2 flashes. This is repeated with 5 seconds between each group of flashes 25 times in total to assess the impact the therapy had on the accumulative effects of the strong light stimuli. This protocol is referred to as the bright flash protocol, and results are reported as a percentage of bright flash baseline at the final step of the protocol. Immediately following the bright flash protocol, recovery of cone function was assessed by recording ERG responses to 7.2-cd⋅s/m2 flashes every 5 minutes over a period of 120 minutes in the presence of a rod-suppressing background light. This protocol is referred to as the cone recovery protocol, and results are reported as a percentage of bright flash baseline at the 120-minute time point.
Scotopic and Photopic Comparisons
A standard scotopic dark-adapted luminance series (20 steps ranging from –-4.5 log cd⋅s/m2 to 1.36 log cd⋅s/m2) was followed by exposure to a 30-cd/m2 rod-suppressing white light, a light-adapted luminance series (12 steps ranging from –2.4 to 1.36 log cd⋅s/m2) and a 33-Hz 3-cd⋅s/m2 flicker as previously described.25 Photopic amplitudes and peak times at the 2.482-, 7.2-, and 23-cd⋅s/m2 flash strengths and 33-Hz flicker were compared between treated and untreated eyes for analysis.
Waveform Modeling
The a-wave and b-wave amplitudes and peak times for all ERGs were measured using standard International Society for Clinical Electrophysiology of Vision (ISCEV) protocols.26 As previously described, the Naka–Rushton function was used to fit the first limb of the scotopic b-wave luminance-response plot.22 Additionally, the leading edge of the scotopic rod a-wave was modeled with flashes ranging from 0.996 to 7.2 cd⋅s/m2 using the Birch and Hood version of the Lamb and Pugh rod phototransduction model.22
Statistical Analysis of ERGs
ERG data were tested for normality using a Shapiro–Wilk test. If the data were normally distributed, a paired t-test comparing treated and untreated individuals was performed and an unpaired t-test was performed between WT versus treated and WT versus untreated. If the data were not normally distributed, a non-parametric Wilcoxon signed rank test was performed. All ERG statistical analyses were performed in RStudio 2023.06.1+524.27 Paired and unpaired t-tests and Wilcoxon signed-rank tests were performed using the R package stats. The 95% confidence intervals, or the highest attainable confidence interval when 95% confidence intervals could not be calculated, were reported when possible. Due to low sample size, zero differences, and tied ranks in paired comparisons, confidence intervals were not always estimable. Animals with small blebs were excluded from ERG functional statistical comparisons due to limited retinal treatment regions. Additionally, animals injected in both eyes (no control eye) were excluded from ERG comparisons. Effect size for parametric data was calculated using Hedges’ g, and nonparametric data used matched-pairs rank biserial correlation. Effect size was calculated using the effectsize package in RStudio.28
Immunohistochemistry and Western Blot
Following humane euthanasia, eyes were removed and fixed in 4% paraformaldehyde (PFA; Electron Microscopy Sciences, Hatfield, PA, USA) in phosphate-buffered saline (PBS; Sigma-Aldrich, St. Louis, MO, USA) and embedded in optimal cutting media (TissueTek OCT; Electron Microscopy Sciences) for frozen sectioning as previously described.29 For IHC, 14-µm cryosections through the treated retina were used. A RDH5 rabbit primary antibody (NBP2-15097; Novus Biologicals, Centennial, CO, USA) at a concentration of 1:100 was used, and an Invitrogen Alexa Fluor 488 goat anti-rabbit secondary antibody was used at a concentration of 1:500 (A11008; Thermo Fisher Scientific, Waltham, MA, USA). All slides were imaged with an Eclipse 80i fluorescence microscope (Nikon Corporation, Tokyo, Japan) with a pco.panda 4.2 sCMOS camera (70108074001; Excelitas, Pittsburgh, PA, USA).
For western blotting, retina RPE choroid was dissected and placed in 400-µL radioimmunoprecipitation assay (RIPA) buffer with proteinase inhibitors and processed as previously described.21 However, the RDH5 antibody was diluted to 1:500, α-tubulin was diluted to 1:2000, and a chemiluminescent substrate was used (34076; Thermo Fisher Scientific). Retina RPE choroid was obtained from the treated bleb region and the untreated region of the same eye (n = 1). Retina RPE choroid from one WT cat was collected. Western blotting was quantified by densitometric analysis using ImageJ (National Institutes of Health, Bethesda, MD, USA).30 RDH5 protein was normalized to the corresponding loading control (α-tubulin), and the RDH5 protein/α-tubulin ratio was calculated.
Resin-Embedded Semi-Thin Sections
Following humane euthanasia, eyes were removed and fixed in 3% glutaraldehyde and 2% paraformaldehyde in 0.1-M sodium cacodylate buffer (Electron Microscopy Sciences), as previously described.25 Following fixation, 3-mm-diameter biopsy punches were collected from the AC of treated and untreated eyes of two RDH5−/− cats with preservation of structure. Color fundus images of the eyes showing the injection bleb and vascular pattern to allow identification of the AC were used to guide tissue collection. The biopsy punches were resin embedded as previously described.25 Then, 1000-nm-thick sections were obtained and stained with epoxy tissue stain (Electron Microscopy Sciences) and imaged on a Nikon Eclipse 80i light microscope.
Retinoid Extraction and Analysis
RDH5 −/− (n = 2) and RDH5+/+ (n = 1) cats were dark adapted overnight, followed by brief exposure to light (30 cd/m2 for 10 minutes) and 100 minutes of dark adaptation prior to euthanasia. Eyes were enucleated under dim light, and biopsies were acquired from treated and untreated sections with the retina intact (containing RPE and choroid). Retinal punches were flash frozen and stored at –80°C. Retinoid analysis procedures were all performed under dim red light.
For retinoid extraction, samples were transferred to an ice-cold buffered solution consisting of 10-mM sodium phosphate (pH 8.0), 100-mM of hydroxylamine (Sigma-Aldrich), and 50% v/v methanol (Sigma-Aldrich); homogenized with a tissue grinder; and incubated at 22°C for 15 minutes. Next, 3 mL of 100% ethyl acetate (Thermo Fisher Scientific) and 2 mL of 3-M NaCl were added to the samples. The mixture was vigorously shaken and then centrifuged to promote phase separation, and the upper organic layer was collected into a borosilicate tube. An additional 3 mL of ethyl acetate was added and the extraction procedure repeated. The combined organic layers, containing extracted retinoids, were dried in vacuo. The dried sample was reconstituted in 450 µL of 100% hexanes (Thermo Fisher Scientific), and a 100 µL portion was analyzed by normal phase high-performance liquid chromatography (HPLC; 1260 Infinity II LC system equipped with a diode array detector; Agilent Technologies, Santa Clara, CA, USA) using a Zorbax Rx-SIL column (5 µm, 4.6 × 250 mm; Agilent Technologies). A stepwise elution protocol was used at a solvent flow rate of 1.4 mL/min, first with 0.6% v/v ethyl acetate in hexanes for 17 minutes followed by 10% v/v ethyl acetate in hexanes for 45 minutes. Retinoid elution was monitored at 325 nm and 360 nm. Retinoid peak areas were integrated using ChemStation software (Agilent Technologies). Retinoid standard curves allowed conversion of 11-cis-retinal oxime and retinyl ester peak areas to absolute masses. Raw chromatograms and quantified retinoid species were plotted and analyzed with Prism (GraphPad Software, Boston, MA, USA).
Results
AAV2-VMD2-fRDH5 Does Not Result in Transgene Expression
The initial construct with a native AAV2 capsid was injected subretinally in one eye of two RDH5−/− cats. No ERG rescue was detected nor expression of RDH5 in the retina (data not shown).
Vector With an AAV2 Triple Capsid Mutation Drives Expression in RPE With Improvement in the Classic Visual Cycle
To confirm the AAV2 triple mutant capsid along with the VMD2 promoter could drive RPE expression of the transgene, an initial reporter gene (hGFP) study was performed using a control cat. This confirmed that the GFP transgene was present and specific to the RPE (Supplementary Fig. S2). Immunohistochemistry and western blotting of the treated eyes confirmed that expression of the RDH5 transgene occurred exclusively in the RPE of treated regions of the RDH5−/− cats (Fig. 1, Supplementary Fig. S3). HPLC demonstrated that the expressed RDH5 was functional. Retinal samples were collected after a brief photobleach followed by dark adaptation for 100 minutes. The control samples from WT cats showed the presence of 11-cis-retinal, but the treated retinal regions from RDH5−/− cats showed decreased all-trans-retinyl esters and increased 11-cis-retinal formation compared to untreated regions (Fig. 2, Supplementary Fig. S4, Supplementary Table S3). The abnormal presence of 13-cis-retinyl esters associated with RDH5 loss of function21 was also corrected by the gene therapy. Together, the data demonstrate that the gene therapy corrected the biochemical defect in the visual cycle associated with RDH5 loss of function.
[图：Subretinal injection of AAV2(triple Y>F)- VMD2-fRDH5 vector in a RDH5 −/− cat . ( A1 ) Post-injection fundus image sh]
[图：Retinoid analyses from RDH5 −/− cat retinas following gene therapy . ( A ) HPLC chromatograms (at 325-nm wavelength) of ]
Gene Therapy Improves Photoreceptor Recovery Following Light Exposure
Scotopic ERGs following overnight dark adaptation were comparable between RDH5−/− and WT cats with only a few minor changes in certain parameters, as we have previously reported.22 In the current study, we found very few differences in Naka–Rushton and a-wave modeling parameters following overnight dark adaptation between treated and untreated eyes (the differences are highlighted in Supplementary Table S4). Similarly, for regular photopic ERG, we previously reported little difference between RDH5−/− and WT cats,22 and, not surprisingly, the difference between treated and untreated eyes herein was negligible (Supplementary Table S5).
Following a bleach, the RDH5−/− cats showed very little recovery of rod function as assessed by ERG over a 100-minute period compared to almost full recovery in WT cats. The degree of rod recovery (Fig. 3) was significantly improved in gene therapy–treated eyes compared to untreated eyes (3 months, P < 0.001; 6 months, P = 0.008; 1 year, P = 0.03) (see Supplementary Table S6 for the full statistical report). Effect size analysis indicated large treatment effects, with Hedges’ g values of 2.21 at 3 months and 1.18 at 1 year post-injection, and a rank-biserial correlation of 1.00 at 6 months post-injection. As only part of the retina was treated, the fullfield flash ERG, which records from the entire retina (treated and untreated), was not expected to show rod functional recovery comparable to that in WT cats.
[图：Recovery of dark adaptation in RDH5 −/− treated versus untreated eyes versus wild-type . ( A ) Representative ERG wavefo]
In addition to a delay in rod photoreceptor recovery from light exposure, we assessed the ability of the cones in treated RDH5−/− cats to recover following a protocol involving repeated exposure to strong flashes of light with a regular ERG recording between strong flashes. The previously described protocol was termed the bright flash protocol. Over the period of repeated flash exposures, there was a decline in the photopic ERG amplitudes. There was no significant difference between these ERG amplitudes between treated and untreated eyes (Supplementary Table S7). Following this suppression of cone function, the recovery of cone responses was followed over the subsequent 120 minutes with no further strong light stimulus. This protocol was termed the cone recovery protocol. The treated eyes had significantly greater recovery of b-wave amplitude assessed as a percentage of bright flash baseline at all post-injection time points (3 months, 0.007; 6 months, 0.006; 1 year, 0.03) compared to their untreated counterparts (Fig. 4). Effect size analysis indicated large treatment effects, with Hedges’ g values of 1.50 at 3 months, 1.38 at 6 months, and 1.82 at 1 year post-injection. There was no significant difference in the a-wave percentage of baseline between treated and untreated eyes at any time points (Supplementary Table S7).
[图：Cone recovery ERG for treated versus untreated RDH5 −/− eyes versus wild-type . ( A ) Average waveforms from 6 months po]
Therapy Provides Structural Preservation of the AC
In addition to functional improvement resulting from the gene augmentation therapy, RDH5−/− cats were assessed for structural preservation using SD-OCT imaging. A subset of RDH5−/− cats developed a bilaterally symmetrical progressive outer retinal thinning in the AC. Two cats in which the injection bleb encompassed the AC developed AC degeneration in the untreated control eye. Although both animals showed clear evidence of structural preservation in the AC region of the treated eye, the degree of preservation differed. In one RDH5−/−-treated cat (RSA 20-053) there was virtually no degeneration in the AC of the treated region compared to the AC of the untreated region. In the other RDH5−/−-treated cat (RSA 20-054), there was slowing, but not halting, of the degeneration in the treated eye. This structural preservation could also be seen in semi-thin electron microscopy sections across the AC in both RDH5−/− cats (Fig. 5).
[图：Structural preservation resulting from gene therapy . ( A ) RetCam image of RDH5 −/− cat #20-053 showing treated bleb re]
Adverse Effects
The injections were well tolerated with no inflammation beyond the typical changes seen as a result of a subretinal injection. However, incipient and non-progressive cataracts developed in two of the treated eyes. One developed the cataract 3 months post-injection, and the other developed the cataract 1 year post-injection. It is unclear if the cataract developed as a direct result of treatment. Beyond the scar from the injection retinotomy, there was no retinal degeneration associated with the treated bleb region (Supplementary Fig. S5).
Discussion
There is currently an unmet need for therapy for patients with RDH5-associated retinopathies. Although the night-blindness phenotype may be manageable, the potential of progressive MD represents a threat to the high-acuity vision needed for normal activities of daily living. AAV gene augmentation is proposed to be an appropriate approach. AAV vectors are being used in many clinical trials with promising results,23 and an FDA therapy for treating lack of function of another RPE visual cycle gene, RPE65, has been approved.31 The therapy should be safe and improve the supply of 11-cis-retinal to photoreceptors, ideally providing improved rod photoreceptor recovery and protecting the cones and macula from degeneration. Here, we have reported a promising AAV therapy that provides a proof of concept for treating RDH5 retinopathies.
We identified an AAV serotype/promoter combination that efficiently transduced feline RPE from a subretinal injection using a reporter gene study. This was an AAV2 vector with a triple capsid Y>F modification coupled with a VMD2 promoter. A therapeutic construct using this serotype/promoter combination was constructed to deliver feline RDH5 cDNA. IHC and western blotting confirmed that this therapeutic construct resulted in RDH5 expression just in the region of the subretinal bleb, as expected. HPLC analysis of treated retinal samples showed clearance of abnormal 13-cis-retinyl esters and improved conversion of 11-cis-retinol to the 11-cis-retinal required for rhodopsin regeneration. This was reflected in an improvement in rod ERG recovery following bleaching.
Significant improvement in the percent of b-wave recovery after cone functional suppression from repeated bright-light exposure (cone recovery protocol) between treated and untreated eyes indicated that RDH5 supplementation also improves cone functionality. The untreated eyes had a very reduced 11-cis-retinal supply from the classic RPE visual cycle, likely meaning they rely on the intraretinal visual cycle to allow reactivation of the cone opsins. With continued strong light exposure, it is likely that the 11-cis-retinal supply from the intraretinal visual cycle will become depleted. With RDH5 supplementation, there is more 11-cis-retinal available via the classic visual cycle for the cones to recover more quickly than those in the untreated RDH5−/− retina. Therapy is expected to improve 11-cis-retinal supply from the RPE visual cycle but will have no effect on the intraretinal visual cycle.
Although treatment resulted in significant differences between the treated and untreated eyes for some ERG parameters, treated eyes did not reach WT control levels for these parameters. Treatment blebs encompassed only a small portion of the retina, and all ERGs used were fullfield flash with a Ganzfeld with summated input from the entire retina, treated and untreated. Therefore, even with optimal transduction within the treatment bleb responses are lower than for normal WT retina.
Structural ONL preservation in the AC was seen in the two RDH5−/−-treated cats in which the AC was successfully targeted and which developed AC degeneration in the untreated eye. It can sometimes be difficult to predict the direction in which the injection bleb extends, meaning that even when the intention was to include the AC the bleb did not necessarily extend to include it. Furthermore, not all the RDH5−/− cats in the study developed AC degeneration. Of the two cats that had therapy that included the AC and showed AC degeneration in the untreated eye, the AC degeneration was slowed in one cat and halted in the other. These cats were followed for 4 years. The difference in degree of preservation of the AC between these two RDH5−/− cats could reflect differences in RPE transduction efficiency between the two cats or factors such as background genetics. These results are preliminary, and further studies are required to investigate the photoreceptor preservation from therapy and understand the differences in structural rescue.
This gene augmentation therapy treatment improves functional rod and cone recovery and also shows promise in being able to slow and potentially halt AC degeneration in RDH5−/− cats. Future directions of this therapy include continued long-term investigation into whether the AC degeneration is halted or just slowed by treatment. In conclusion, this preliminary study showed that RDH5 AAV2 gene augmentation therapy offers a promising and potentially translatable therapy able not only to improve retinal function but also to slow the debilitating MD phenotype seen in a subset of humans with RDH5-associated retinopathies.
Supplementary Material
[内容请查看原文]
Acknowledgments
The authors thank Paige Winkler Smith for her help with semi-thin sectioning and technical support; Alicia Withrow, PhD, for her help with semi-thin and thin sections; Janice Querubin for her help with some of the animal handling and some of the ERG/OCT data collection; Lexi Daggy for help with some of the animal handling and anesthesia; Billie Beckwith-Cohen and Geneive Carrion for surgical assistance; and the Michigan State University College of Veterinary Medicine Research and Teaching Technical Support and Michigan State University Campus Animal Resources group for help with animal care. We also thank William Hauswirth, PhD, of the University of Florida for the RDH5 vector production.
Supported by the Michigan State University Myers-Dunlap Endowment for Canine Health; George H. Bord and “Casper” Endowment for Feline Initiatives; Michigan State University Center for Feline Health and Well-Being; a grant from the National Institutes of Health (R01EY009339 to PDK); grants from the Department of Veterans Affairs (I02BX004939 and IK6BX006800 to PDK); an unrestricted Research to Prevent Blindness Award to the UC Irvine Department of Ophthalmology and Visual Sciences; and a National Eye Institute UCI Center Core Grant for Vision Research (P30EY034070).
Data Availability: Data will be made available upon request.
Disclosure: L.M. Ford, None; K. Sun, None; L.M. Occelli, None; M.E. Enfield, None; A. Daruwalla, None; R.F. Boyd, None; P.D. Kiser, None; S.M. Petersen-Jones, None
References
[内容请查看原文]
Associated Data
Supplementary Materials
[内容请查看原文]