The ability to generate neurons from other, easily accessible cell types from patients with a neurological or psychiatric disease is something of a holy grail for neuroscience, as such neurons would enable researchers to model the disease in vitro or to develop cell replacement therapies. Now, Tursun et al. show that germ cells can be directly converted into specific types of neuron in vivo.

During development, specific transcription factors induce the differentiation of cells to a particular fate, but it is thought that they can only do so in a specific cellular context. To study this context-dependence, Tursun et al. focused on the Caenorhabditis elegans transcription factor CHE-1, which is known to induce ASE neurons (a type of chemosensory neuron), and asked whether CHE-1 is necessary and sufficient to induce an ASE cell fate. They found that, after expressing CHE-1 throughout the C. elegans body — both in larvae and adults — the ASE markers gcy-5 and ceh-36 were still only expressed in head sensory neurons.

The authors hypothesized that a factor might prevent CHE-1 from reprogramming other cells into ASE neurons, and performed an RNA interference (RNAi) library screen in which they identified LIN-53 as a potential candidate. They found that, after knockdown of lin-53 using RNAi, CHE-1 expression could induce gyy-5 and ceh-36 in germ cells. Differential interference microscopy revealed that the cells now resembled neurons, even expressing axo-dendritic projections. This effect was decreased in mutant worms with a reduced mitotic pool of germ cells, but not in mutant worms in which entry into meiosis was blocked, indicating that CHE-1 can reprogramme only the mitotic pool of germ cells to a neuronal cell fate.

The newly converted cells expressed several pan-neuronal reporter genes as well as ASE-specific genes — including gcy-7 and eat-4 — but not genes that are normally exclusively expressed in other neuron types, suggesting that that the reprogrammed cells had acquired an ASE phenotype rather than a generic neuronal phenotype.

Importantly, other neuron types could also be induced from germ cells: knockdown of lin-53 in combination with ectopic expression of unc-30 or unc-3 — which are the transcription factors required for the terminal differentiation of GABAergic and cholinergic motor neurons, respectively — reprogrammed germ cells into cells expressing the GABAergic marker unc-47 and the cholinergic marker arc-2, respectively. Thus, lin-53 loss seems to endow germ cells with the capacity to 'redifferentiate' into various types of neuron depending on the transcription factor that is expressed.

How does removing lin-53 do this? LIN-53 is known to recruit histone-modifying complexes to histones H3 and H4. Two of these complexes, NURD and Sin3a, contain histone deacetylase (HDAC) components, raising the possibility that removing lin-53 prevents HDAC activity, enabling gene transcription. Indeed, in worms treated with the HDAC inhibitors valproic acid or trichostatin A, expressing che-1 triggered germ cells to adopt an ASE fate.

The authors speculate that LIN-53, by regulating chromatin structure, prevents transcription factors from accessing genes and inducing (re)differentiation. These findings extend a previous study showing that expressing a combination of three transcription factors could convert adult mouse fibroblasts into functional neurons in vitro. These studies indicate the potential of both somatic and germline cells from adult animals to convert to a distinct neuron type without having to enter a pluripotent stage.