Since its establishment, cell reprogramming of somatic cells into induced pluripotent stem cells (iPSCs) in vitro has provided a fantastic tool for disease modeling studies as well as presented potential applications in regenerative medicine1. However, in practice, the reprogramming of patient cells into iPSCs and subsequent redifferentiation into desired target cells is often an inefficient and time-consuming process. As an alternative approach, transdifferentiation directly converts one cell type into another, which bypasses many of the safety concerns associated with the pluripotent cell state. If this method can be applied in a safe and efficient manner in vivo, it can further help to eliminate the undesired issues that may arise during the in vitro culture and cell transplantation processes.

The idea of in vivo transdifferentiation is not a completely novel notion in the stem cell field and it has been developing rapidly in recent years due to the rise in the discovery of transcription factors that induce cell reprogramming. A particularly remarkable advance in this field was the direct reprogramming of adult pancreatic exocrine cells into β-cells in 20082. Utilizing a cocktail, the authors introduced different combinations of nine β-cell development-related transcription factors into the pancreas of adult mice by adenovirus. They found three factors (Pdx1, Ngn3, and Mafa) that were required to convert up to 20% of infected exocrine cells into insulin-producing β-cells. Lineage tracing experiments further demonstrated the exocrine cell origins of the induced β-cells. The absence of BrdU-labeled cells during the process excluded the possibility of a dedifferentiation stage. The induced β-cells were characteristically similar to the endogenous islet β -cells in cell morphology, ultrastructure, lineage marker expression and insulin secretion. However, the majority of the induced β-cells remained scattered or in small clusters without efficient integration into existing islets, which presumably resulted in their limited capacity to fully restore glucose homeostasis after chemically induced pancreatic injury. Nonetheless, this report provided an encouraging proof-of-principle for in vivo transdifferentiation and stimulated further studies. Using a similar method, Qian et al.3 generated induced cardiomyocyte-like cells from cardiac fibroblasts in vivo by retroviral delivery of three transcription factors (Gata4, Mef2c, and Tbx5) into the myocardium based on a previous in vitro experiment4; Torper et al.5 reported the successful conversion of resident astrocytes into mature neurons in situ by forced expression of Ascl1, Brn2a, and Myt11 in the adult mouse striatum; and other studies also reported the successful in vivo transdifferentiation6,7,8 (listed in Table 1).

Table 1 Examples for in vivo transdifferentiation (TFs: transcription factors)
Full size table

Although the methodologies were similar, the mechanisms involved in these in vivo transdifferentiation studies appeared to be different. In some cases, it seemed to be a direct cellular conversion, such as from exocrine cells to β-cells, but in other cases, the initial cells might need to dedifferentiate into an intermediate precursor stage before final conversion to a new fate. For instance, Zhang and colleagues reprogrammed astrocytes in the adult mouse striatum to neuroblasts, which are neural precursor cells9. In their study, they found that Sox2 alone was sufficient to reprogram the endogenous quiescent astrocytes to neuroblasts through a proliferative stage. The induced neuroblasts could subsequently differentiate into neurons with exogenous expression of Brain-derived neurotrophic factor (BDNF) and Noggin, or with VPA (a histone deacetylase inhibitor) treatment. Though the induced neuroblasts were proliferative, no tumor formation was detected during this study. More excitingly, neurons derived from induced neuroblasts exhibited an elaborate neural morphology, possessed functional voltage-gated sodium channels, and formed synapses with the endogenous neurons, a sign of efficient integration into the local neural network.

These fascinating results from the recent progress in in vivo transdifferentiation brought us new hopes and also raised new questions. For instance, does transdifferentiation go through a dedifferentiation stage or does it occur directly? What is the master regulator in the in vivo cell fate conversion process? Undisputedly, transcription factors play key roles during this process in which they build up a transition bridge between initial and target cells. It has been reported that using the same initial cells, astrocytes could be converted to neuroblasts and mature neurons by Sox2 alone9 and by combination of Ascl1, Brn2a, Myt115, respectively. With the appropriate combination of transcription factors, even terminally differentiated cells can be directly converted into another terminally differentiated cell type. For example, Rouaux et al.7 demonstrated that early post-mitotic callosal neurons could be converted into corticofugal neurons in vivo. Moreover, the type of initial cells selected also appears to have significant impact on transdifferentiation ability in vivo. The developmental origin of the initial and target cells may directly influence the efficiency and complexity of the transdifferentiation process. This is particularly evident in the transdifferentiation of pancreatic exocrine cells to β-cells, which occurred rapidly and efficiently2. This conversion benefited substantially from the few epigenetic differences between the two lineages as they share a common precursor.

On the other hand, as the underlying mechanisms of transdifferentiation remain largely unknown, a step-wise methodology for transdifferentiating cells cannot be defined. Meanwhile, the safety issues related to in vivo transdifferentiation have never been extensively investigated. A recent study by Serrano and colleagues detected iPSCs derived from hematopoietic cells or non-hematopoietic cells in the blood stream through overexpression of OSKM in vivo10. Those iPSCs migrate to different tissues via the blood stream and differentiate on site or form teratomas. Certainly these tumor-forming cells would not be desired targets in any in vivo transdifferentiation study. Therefore, the scale and extent of the dedifferentiation process need to be carefully monitored.

Considering the clinical advantages of in vivo transdifferentiation, the associated safety and efficacy issues are well worth careful investigation. Can the induced target cells maintain their new properties and functions in the long run? Can the unnecessary damages to native tissue or non-targeted cells be minimized in induced transdifferentiation? Do infected but un-reprogrammed target cells pose a safety concern? In order to meet the criteria of potential clinical application, it is also important to replace viral vectors without jeopardizing the efficacy of transdifferentiation. Screening small compounds that mimic the role of transcription factor overexpression is certainly a promising substitute for the use of virus. Furthermore, the recent report from Abad et al. suggested that in vivo cell reprogramming may also take place in the kidney, intestine, intracranial, stomach and other tissues10. Therefore, if the safety concerns can be solved, in vivo transdifferentiation may open a new avenue to generate various cell types for different clinical purposes.