Abstract
Purpose :
Reprogramming Muller glia cells (MG) into neurons in vivo represents a potential therapeutic strategy that could repair damaged neural circuits and restore visual function. Adeno-associated viruses (AAVs) have been widely used in combination with a short GFAP promoter to deliver reprograming factors into MG, with high-impact studies recently showing MG-to-neuron conversion with a high efficiency. However, the authentication of glia-derived neurogenesis has been questioned due to the leaky expression of AAV-GFAP-mediated transgene(s) into endogenous neurons. We therefore hypothesize that modifying the AAV constructs could decrease their leakage and enhance gene delivery efficiency into MG.
Methods :
ShH10 (an AAV serotype used for transducing rodent MG)-based delivery systems were modified for intravitreal injection of neuronal reprogramming factor Neurod1, Math5, Ascl1, or Neurog2 (reporter GFP or mCherry was used as control) into mouse MG. Stringent fate mapping experiments (tamoxifen-induced Glast-CreERT;CAG-LSL-Sun1-GFP mice at the age of 4-6 weeks, 3-5 retinas per group) were performed to trace the lineage of MG and their progeny. Retina sections were used for immunohistochemistry analysis with cell-type-specific markers.
Results :
While ShH10-GFAP-mediated intravitreal delivery of GFP or mCherry exhibits high specificity (~98%) for MG, various levels of leaky expression into amacrine and retinal ganglion cells were observed after gene delivery of Neurod1 (80.2%), Math5 (80.3%), Ascl1 (31.0%), or Neurog2 (9.9%). The transgene-dependent leakage cannot be corrected after lowering the AAV titers, using alterative serotypes or injection routes (PHP.eB for subretinal injection). We then developed a new ShH10-GFAP-based tool that dramatically reduces neuronal leaky expression (only 10.3% for Neurod1 and 0.7% for Math5) by positioning the factor after mCherry reporter. The second new AAV tool we developed is a ShH10-CAG-FLEX dual system that allows high-specificity delivery of Neurod1 (97.4%) and Math5 (95.1%) into MG upon genetic-based activation of tamoxifen-inducible Cre expression. Using both new tools, we provide evidences that Neurod1 or Math5 alone fails to convert MG into retinal neurons.
Conclusions :
Our study indicates that highly efficient expression of reprogramming factors in stringently lineage-traced MG is the key to carrying out MG-to-neuron reprogramming research.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.