Transcription factor-based direct lineage reprogramming is a powerful tool to discover and study factors determining cell lineage fate. We propose that this methodology can be used to define the core transcriptional program directing red blood cell development. The aim of this study was therefore to identify the minimal set of transcription factors that allows direct lineage reprogramming of murine fibroblasts to erythroid progenitor cells.

A retrovirus library was created expressing the coding region of 63 transcription factors known to be involved in erythroid and blood development. Adult tail tip fibroblasts were obtained from erythroid lineage tracing mice, which express yellow fluorescent protein (YFP) in cells that have once expressed the erythropoietin receptor (EpoR) gene at any time of their development. Fibroblasts were depleted for hematopoietic lineage markers and passaged at least 3 times prior to transduction with different combinations of reprogramming factors. The readout for erythroid lineage conversion was formation of colonies of round YFP+ (EpoR+) cells, which were further subjected to extensive analyses to determine their resemblance with primary erythroid progenitor cells. Factor-subtraction experiments revealed a combination of 4 transcription factors, Gata1, Tal1, Lmo2 and c-Myc (collectively referred to as GTLM), capable of directly converting fibroblasts to erythroid progenitors in vitro. These induced erythroid progenitors (iEPs) emerged 5 to 8 days after transduction and displayed an erythroid progenitor-like morphology, featuring a characteristic central nucleus, coarse chromatin and deep blue cytoplasm after May Grünwald-Giemsa staining. When cultured in erythroid-promoting conditions, iEPs differentiated to Benzidine-positive normoblast-like cells. Flow cytometric analysis revealed that 28.9% ± 2.2 of live YFP+ cells collected at day 7 co-expressed CD71 and Ter119, and did not express CD45.

Bulk GTLM-transduced fibroblasts formed two types of colonies in methylcellulose assays, one with visibly red hemoglobinized cells and one with blast-like cells. Global gene expression analyses showed GTLM expression was generally higher in the hemoglobinized colonies, suggesting that complete reprogramming only occurs when each factor is expressed at a sufficient level. Hierarchical clustering and pairwise comparisons of gene expression data showed that iEP-derived hemoglobinized colonies correlated tightly with definitive erythroid (BFU-E) colonies from bone marrow and fetal liver. As expected, iEP-derived hemoglobinized colonies displayed large-scale downregulation of the fibroblast-specific program and extensive upregulation of genes specific to the erythroid lineage. In contrast to definitive erythroid colonies, iEP-derived hemoglobinized colonies did not upregulate Sox6 and Bcl11a; and predominantly expressed embryonic hemoglobin, similar to primitive erythroid cells. This could mean that additional factors are required to induce definitive erythropoiesis, which we are currently investigating.

Importantly, reprogramming was never successful using any combination of three of the GTLM factors, clearly demonstrating that all four factors are needed and that this is the minimal combination of required factors. Reprogramming of p53-null fibroblasts enhanced efficiency, but did not allow reprogramming without c-Myc. Additional experiments demonstrated that other factors known to be important for red cell development, including Klf1, Nfe2 and Myb, were not required and could not substitute for any of the 4 factors to induce erythroid fate.

To our knowledge this is the first successful direct conversion of non-hematopoietic cells to the erythroid lineage. Our results suggest that GTLM constitute the core of the erythroid program, capable of inducing expression of other transcriptional regulators such as Klf1, Zfpm1, Gfi1b, Nfe2 and Myb, which are necessary for normal red cell development. We anticipate that GTLM-induced direct erythroid reprogramming can be used as a new platform for understanding, controlling and studying erythroid lineage development and disease. Furthermore, this knowledge could potentially be applied to enhance methods for in vitro production of erythrocytes for personalized transfusion medicine.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.