Long Gene Expression and Non-CG DNA Methylation Identify Functional Neurons in microRNA-Mediated Direct Reprogramming

Reprogramming of non-neuronal cells into neurons is increasingly used to develop cellular models of human neuropathologies with patient-specific cells. However, many neuropathologies have age-related components that are erased when cells are reprogrammed to an induced pluripotent state. We and other have shown that direct neuronal reprogramming, as induced by the brain-enriched microRNAs (miRNAs), miR-9/9* and miR-124 (miR-9/9*-124), overcomes this barrier, effectively bypassing the pluripotent/multipotent stem cell stages and retaining the age of donor somatic cells. Understanding the mechanism driving direct miRNA-mediated neuronal reprogramming will allow the exploitation of this unique feature to further interrogate etiologies of age-related neuropathologies as well as of aging itself. Our recent investigation demonstrated extensive and dynamic remodeling of the epigenome induced by miR-9/9*-124 in human adult fibroblasts. One surprising feature that emerges from analysis of time series transcriptome profiling is enrichment of extremely long genes (> 100 kb from transcription start to termination sites), a phenomenon recently described in neuronal cells. We sought to understand the phenotypic consequences of long gene expression (LGE) as well as the molecular machinery underlying LGE in neurons. By analyzing publically available transcriptome data from other forms of neuronal reprogramming, we determined that LGE is a feature of neurons successfully reprogrammed by multiple methods. However, we found that reprogrammed neurons lacking electrical activity also lacked LGE despite having transcripts of popular neuronal “markers” (e.g. MAP2) and neuronal morphology, indicating that LGE may be reflective of neuronal functionality. In support of this, we found that by manipulating TOP1, a topoisomerase shown to be necessary for LGE, spontaneous electrical activity was dampened at the cell population level in direct relation with LGE genome-wide without affecting MAP2 or neuronal morphology. Finally, since LGE has been shown to be positively correlated with non-CG DNA methylation in the developing brain, we measured this feature in neurons reprogrammed from human fibroblasts by miR-9/9*-124. As predicted, reprogrammed neurons showed an enrichment for non-CG DNA methylation, which correlates with increased LGE. Together, this work implicates both the expression of extremely long genes and non-CG DNA methylation as molecular hallmarks of functional neurons, and provides a cellular platform to interrogate their potential roles in adult-onset neurological disorders.