Our bodies rely on specialized cell types: brain cells compute information, red blood cells bind oxygen, and so on. Because almost all our cells have identical DNA, different patterns of gene and protein expression are needed to define these cell types. The selection and maintenance of these expression cascades were once thought to be irreversible after development. Over time, it emerged that cell identity could be changed, but it was often assumed that a cell could be converted into another cell type only if the two had a similar developmental origin. Ten years ago, Vierbuchen et al.1 overthrew this idea, by showing that connective-tissue cells called fibroblasts could be converted into functional neurons — which have a very different developmental origin — if they were engineered to express just three extra transcription factors.
This achievement was built on almost a century of visionary experiments in manipulating cell identity. In 1927, Hans Spemann showed that it was possible to change the fate of cells in a salamander embryo. The embryologist grafted ‘organizer’ cells (which drive early development of the body plan) from a donor embryo into a host embryo2, triggering the formation of a second embryo from the host cells. In 1962, the biologist John Gurdon showed that development can also be returned to the start3 — the nucleus of an adult cell can reacquire a state similar to that of cells in the earliest stages of development, and in this state it can give rise to an entire embryo.
In the 1980s, it became clear that cells can also be directly converted from one specialized cell type to another (Fig. 1a). The first example4 was the conversion of fibroblasts into muscle cells by inducing the cells to express the transcription factor MyoD. Some years later, a different transcription factor was used to turn non-neuronal cells of the brain called glia into neurons in vitro5. The first demonstration that this type of conversion could also occur in vivo in mice6 opened up a potential new branch of therapy based on converting reactive glia into new neurons after brain insults or neurodegeneration7.
In vitro, a wealth of other conversions was documented8, but all involved either reversion to an embryonic state9 or transformation into another cell type from within the same ‘germ layer’. Germ layers are the three layers of embryonic tissue (endoderm, mesoderm and ectoderm), which give rise to different organs and cell types. For instance, the gut tube and liver derive from the endoderm; muscle and connective tissue from the mesoderm; and neural tissue and skin from the ectoderm.
It was assumed that cells could be converted only to other cell types from the same germ layer, owing to their closely related developmental origins. This dogma was shattered by Vierbuchen and colleagues, who converted mesoderm-derived fibroblasts from mice into functional neurons by co-expressing the three transcription factors Brn2, Ascl1 and Myt1l in the fibroblasts (Fig. 1b). By showing that developmental barriers are not an unsurmountable hurdle to cell-type conversion, the paper had a tremendous impact.
First, it sparked a wave of interest in direct reprogramming to produce neurons. All of a sudden, fibroblasts — which are relatively easy to isolate from mouse embryos and are easy to grow in vitro — could be converted into a cell type of great therapeutic interest. The year after the paper’s publication, human fibroblasts were directly converted into neurons10, although this required more transcription factors than were needed for the conversion of mouse cells. It was only a few more years before transcription-factor cocktails had been defined to generate diverse neuronal subtypes11–13.
In 2015, it emerged that the ‘induced neurons’ produced using Vierbuchen and colleagues’ method retain their cellular age — if the fibroblasts come from a 60-year-old donor, the reprogrammed neurons show a corresponding cellular age14. Thus, direct neuronal reprogramming is well suited for obtaining neurons to study age-related neurodegenerative diseases such as Alzheimer’s disease, Huntington’s disease or motor neuron disease15,16.
Beyond these key impacts on translational research, the paper raised questions about how developmental origin affects the maintenance of cell identity. For instance, the work called into question whether sharing a germ layer would always ease direct reprogramming between cell types. The answer is no: skin cells, derived from the ectoderm, cannot be readily converted to neurons17 (Fig. 1c). Moreover, neurons can switch between subtypes only during development18. These findings called for reconsideration of models of cell-identity maintenance. Perhaps, instead of depending on developmental origin, the key factors in how easily cell types can interconvert relate to similarities and differences in gene regulation between the mature cell types19.
If this is the case, it should be possible to ascertain rules for interconverting cell types by altering specific gene-regulation parameters. However, no systematic studies to explore the potential of a given starter cell type to convert into different target cells have yet been carried out. This means that it is not yet possible to identify common rules for reprogramming — or, conversely, for the maintenance of identity. Filling this gap is now an important task.
The ease of direct programming hints at the fragility of the mechanisms that maintain cell fate. So what keeps cells stable over decades? Researchers are starting to investigate the mechanisms (passive and active) that regulate expression of the transcription factors involved in switches of cell type, and to ask whether long-lived cells are more difficult to convert because they have developed more-elaborate fate-maintenance mechanisms. This could also be the reason that human cells are much harder to convert into other cell types than are mouse cells.
The identification of these mechanisms would not only be a conceptual breakthrough, but would also help to overcome conditions in which cell identity becomes altered as cells deteriorate during ageing. For example, proteins that repress gene expression are involved in maintaining some aspects of cell identity in mature neurons20. However, little is known about whether or how these factors are depleted in ageing and neurodegeneration, and whether the loss of cell identity is a key contributor to ageing-related diseases.
Direct reprogramming has revolutionized the concept of what defines a cell type, and has allowed us to explore fascinating questions about development. It has also triggered a revolution in disease modelling. That this has taken place in just one decade is testament to the impact of Vierbuchen and colleagues’ discovery.
Nature 578, 522-524 (2020)