Regenerating functional new neurons in adult mammalian brains has been proven a difficult task for decades. Recent advancement in direct glia-to-neuron conversion in vivo opens a new field for neural regeneration and repair. However, this emerging new field is facing serious challenges from misuse of viral vectors to misinterpretation of conversion data. Here, we employ a variety of AAV vectors with different promoters and enhancers to demonstrate that astrocytes can be converted into neurons in a NeuroD1 dose-dependent manner in both wildtype (WT) and transgenic mice. Notably, astrocytes in WT mice were relatively easy to convert with higher conversion efficiency, whereas lineage-traced astrocytes in Aldh1l1-CreERT2 mice showed high resistance to reprogramming but were still converted into neurons after enhancing NeuroD1 expression with CMV enhancer. Furthermore, under two-photon microscope, we observed direct astrocyte-to-neuron conversion within 3 weeks of serial live imaging in the mouse cortex. We also demonstrated that high titre AAV reaching 1013 GC/ml caused severe neuronal leakage using a variety of AAV GFAP::GFP vectors, highlighting the necessity to inject low titre AAV into healthy brains to avoid artifactual results. Together, our studies suggest that lineage-traced astrocytes can be converted into neurons but require stronger conversion force such as enhanced NeuroD1 expression. Failure to recognize the difference between WT astrocytes and lineage-traced astrocytes in terms of conversion barrier will lead to misinterpretation of data.

In vivo astrocyte-to-neuron (AtN) conversion induced by overexpression of neural transcriptional factors has great potential for neural regeneration and repair. Here, we demonstrate that a single neural transcriptional factor Dlx2 converts mouse striatal astrocytes into neurons in a dose-dependent manner. Lineage-tracing studies in Aldh1l1-CreERT2 mice confirm that Dlx2 can convert striatal astrocytes into DARPP32+ and Ctip2+ medium spiny neurons (MSNs). Time-course studies reveal a gradual conversion from astrocytes to neurons in 1 month, with a distinct intermediate state in-between astrocytes and neurons. Interestingly, when Dlx2-infected astrocytes start to lose astrocytic markers, the other local astrocytes proliferate to maintain astrocytic level in the converted areas. Unexpectedly, while Dlx2 efficiently reprograms astrocytes into neurons in the grey matter striatum, it also induces partial reprogramming of astrocytes in the white matter corpus callosum. Such partial reprogramming of white matter astrocytes is associated with neuroinflammation, which can be essentially suppressed by the addition of NeuroD1. Our results highlight the importance of investigating AtN conversion both in the grey matter and white matter in order to thoroughly evaluate therapeutic potentials. This study also unveils a critical role of anti-inflammation by NeuroD1 during AtN conversion.