Direct cell-fate conversion of somatic cells: Toward regenerative medicine and industries

Abstract

Cells of multicellular organisms have diverse characteristics despite having the same genetic identity. The distinctive phenotype of each cell is determined by molecular mechanisms such as epigenetic changes that occur throughout the lifetime of an individual. Recently, technologies that enable modification of the fate of somatic cells have been developed, and the number of studies using these technologies has increased drastically in the last decade. Various cell types, including neuronal cells, cardiomyocytes, and hepatocytes, have been generated using these technologies. Although most direct reprogramming methods employ forced transduction of a defined sets of transcription factors to reprogram cells in a manner similar to induced pluripotent cell technology, many other strategies, such as methods utilizing chemical compounds and microRNAs to change the fate of somatic cells, have also been developed. In this review, we summarize transcription factor-based reprogramming and various other reprogramming methods. Additionally, we describe the various industrial applications of direct reprogramming technologies.

1. Introduction

All cell types in the bodies of higher multicellular animals have extremely diverse phenotypes despite their genetic identity. A single zygote can divide and differentiate into a considerable number and wide variety of cells with unique genome sequences during development. The differences in the phenotypic characteristics among cell types result in different transcriptome and epigenome variations. As Waddington predicted in the 1950s,¹⁾ the differentiation and identity of the cells are governed by a non-genetic system for the regulation of gene expression termed “epigenetics”. Although it has long been thought that terminally differentiated somatic cells cannot cross over the “epigenetic barrier” and trans-differentiate into other cell types, some somatic cells have plasticity for de- or re-differentiation, e.g., cardiomyocytes in zebrafish hearts^2,3) and pigmented epithelial cells in newt eyes.⁴⁾ These examples suggest that terminally differentiated animal cells can convert into other cell types.

In 2006, Takahashi and Yamanaka initialized the cell fate of fibroblasts from adult mammals and re-differentiated them into various cell lineages through the forced transduction of a defined set of transcription factors.⁵⁾ Their study provided information regarding the nature of cell differentiation as well as the possibility of induced pluripotent stem cell (iPSC)-based regenerative therapies. Indeed, many basic and applied studies have been performed for the realization of iPSC-based regenerative medicine. The successful application of this type of medicine has been reported, including for clinical trials of cell transplantation therapy for age-related macular degradation,⁶⁾ Parkinson’s disease,⁷⁾ and corneal epithelial stem cell deficiency.⁸⁾ At present, clinical studies of other diseases, such as spinal cord injury, severe heart failure, and stroke, are being undertaken with regards to iPSC-based regenerative medicine. Furthermore, several drug discovery studies employing patient-derived iPSCs have identified a number of hopeful pharmaceutical candidates for treating fibrodysplasia ossificans progressiva,⁹⁾ Pendred syndrome,¹⁰⁾ Parkinson’s disease,¹¹⁾ Alzheimer’s disease,¹²⁾ amyotrophic lateral sclerosis,¹³⁾ and muscular dystrophy.¹⁴⁾ In addition, the in vitro reconstruction of complicated and functional tissues and organs using iPSCs has also been reported, including in the brain,¹⁵⁾ intestines,¹⁶⁾ liver,¹⁷⁾ and kidneys.¹⁸⁾

Although iPSCs are a promising technology for future medicine, alternative methods for cell fate conversion, termed “direct reprogramming”, have also been developed in the last decade. Direct reprogramming involves the trans-differentiation of somatic cells directly into other cell types without complete initialization. Before the first study regarding iPSCs, only a few types of direct reprogramming had been reported, including the reprogramming of fibroblasts into myoblast and pancreatic β cells. In recent studies, various cell types have been directly converted from other somatic cell types. Most of the direct reprogramming methods applied the forced transduction of defined sets of transcription factors to convert terminally differentiated cells into specific target cells artificially. Initially, only single-transcription factor-based direct reprogramming methods were reported.^19–24) However, most recent studies for direct reprogramming employ a combinatorial transduction strategy similar to that of the iPSC, especially in the conversion between embryologically divided cell types. Although the combinatorial expression strategy of transcription factors with viral vectors is a powerful and stable technique for direct reprogramming, alternative methods are increasing because genomic integration of the transgenes causes genomic damage, and might induce unexpected functional attenuation, abnormal proliferation, and transformation. As explained above, the major applications of the direct reprogramming technologies are bedside clinical uses, such as cell transplantation and regenerative medicine. Meanwhile, other medical, e.g., drug discovery and disease study, and non-medical, e.g., livestock and pet industry, applications of the direct reprogramming technology have been used increasingly in recent years. Therefore, direct reprogramming technology will play a crucial role in both medical and non-medical industry.

2. Transcription factor-based direct reprogramming technologies

Similar to iPSC technology, most direct reprogramming methods employ forced expression of the defined set of transcription factors for cell-fate conversion (Table 1). Many are tissue-specific or developmental process-related genes; thus, it is considered that these transcription factors act as a “kickstarter” to drive the cell type-specific transcription networks. Furthermore, almost all of the reprogramming factor gene sets contain at least one pioneer factor, which is a transcription factor capable of opening closed chromatin sections. These factors may play a pivotal role in the machinery of direct reprogramming. In this section, examples and applications for various types of transcription factor-based direct reprogramming are introduced.

Table 1.

Examples of the direct reprogramming with defined transcription factors

Source cells	Target cells	Transgenes and/or Treatment	Species	Cell transplantation	Published year	PMID
Neuronal (stem) cells
Astrocytes	Glutamatergic neurons	Pax6, Mash1, or Ngn2	Mouse		2007	1768704325)
Embryonic fibroblasts and Hepatocytes	Neuronal cells	Brn2, Ascl1, and Myt1l	Mouse		2010, 2011	2010743926); 21962918212)
Astrocytes	GABAergic neurons	Dlx2; Dlx2 and Ascl1	Mouse		2010	20502524213)
Embryonic fibroblasts and Adult skin fibroblasts	Dopaminergic neurons	Ascl1, Lmx1a, and Nurr1	Mouse & Human	YES	2011	2172532427)
Fetal fibroblasts and Postnatal foreskin fibroblasts	Neuronal cells	BRN2, ASCL1, MYT1L, and NEUROD1	Human		2011	2161764428)
Embryonic fibroblasts and Postnatal fibroblasts	Neuronal cells	ASCL1, BRN2, MYT1L, LMX1A, and FOXA2	Human		2011	2164651529)
Embryonic fibroblasts	Neural stem cells	Brn4/Pou3f4, Sox2, Klf4, c-Myc, and E47/Tcf3	Mouse	YES	2012	2244551730)
Embryonic fibroblasts and Fetal foreskin fibroblasts	Neural stem cells	Sox2	Mouse & Human	YES	2012	2268320331)
Sertoli cells	Neural stem cells	Pax6, Ngn2, Hes1, Id1, Ascl1, Brn2, c-Myc, and Klf4	Mouse	YES	2012	2206470032)
Fibroblasts (IMR90 cells)	Dopaminergic neurons	MASH1, NGN2, SOX2, NURR1, and PITX3 + A dominant-negative P53	Human	YES	2014	25129808214)
Non-sensory cochlear epithelial cells	Neuronal cells	Ascl1; Ascl1 and Neurod	Mouse		2014	2492835139)
Astrocytes	Neuronal cells	Brn4	Mouse		2014	24887289185)
Skin fibroblasts	Dopaminergic precursors	Brn2, Sox2, and Foxa2	Mouse	YES	2015	26224135215)
Adult skin fibroblasts	Motor neurons	NEUROG2, SOX11, ISL1, and LHX3	Human		2016	2672511246)
Fibroblasts (3T6 cells)	Neuronal cells	Ascl1, Brn4, and Tcf3	Mouse		2017	29137640216)
Umbilical cord blood cells	Neural stem cells	SOX2 and HMGA2	Human		2017	2884412733)
Fibroblasts (3T6 cells)	Neuronal cells	Ascl1, Brn2, and Foxa1	Mouse		2017	28855426119)
Resident glial cells	Neuronal cells	Ascl1, Lmx1a, and Nurr1	Mouse		2017	2884465841)
Fibroblast-like cells from retinal tissues	Neuronal cells	ASCL1 and PAX6	Human		2017	28697461217)
Pluripotent stem cell- derived cardiomyocytes	Neuronal cells	Brn2, Ascl1, Myt1l and Neurod	Mouse & Human		2017	28327614218)
Fibroblasts	Motor neurons	ASCL1, ISL1, NEUROD1, BRN2, HB9, LHX3, MYT1L, and NGN2	Human		2017	28099929219)
Fibroblasts	Neurocytes	SOX2, GATA3, and NEUROD1	Human		2017	2858733162)
Dermal fibroblasts	Neural precursor cells	SOX2 and PAX6	Human		2018	3045044034)
Embryonic fibroblasts and Newborn foreskin fibroblasts	Neural stem cells	Ptf1a	Mouse & Human	YES	2018	3003043435)
Adult fibroblasts	Neural precursor cells	SOX2; SOX2 and PAX6; SOX2 and LMX1A; SOX2, LMX1A, and FOXA2	Human		2018	3001847136)
Spiral ganglion non- neuronal cells	Neuronal cells	Ascl1 and Neurod	Mouse		2018	29492404220)
Umbilical cord mesenchymal stem cells	Neuronal cells	SOX2, ASCL1, and NEUROG2	Human		2018	29937717221)
Pericytes	Neuronal cells	ASCL1 and SOX2	Human		2018	29915193181)
Cord blood CD133(+) cells	Neuronal cells	FOXM1, SOX2, MYC, SALL4, and STAT6	Human		2018	2965319638)
Hepatocytes	Neuronal cells	Suz12, Ezh2, Meis1, Sry, Smarca4, Esr1, Pparg, and Stat3	Mouse		2018	2965319638)
Peripheral CD34(+) cells	Neural stem cells	AR, SOX2, SMAD3, MYC, JUN, WT1, TAL1, SPI1, and RUNX1	Human		2018	2965319638)
Urine epithelial-like cells	Neural stem cells	POU3F2, SOX2, BACH1, AR, PBX1, and NANOG	Human		2018	2965319638)
Muller glia cells	Neural stem cells	Bmi1, Spi1, Lmo2, and Cebpd	Mouse		2018	2965319638)
Astrocytes and Foreskin fibroblasts	Noradrenergic neurons	Ascl1, Phox2b, Ap-2a, Gata3, Hand2, Nurr1, and Phox2a	Mouse & Human	YES	2019	31315047222)
Bone marrow-derived cells, Fibroblasts, and Keratinocytes	Neural precursor cells	MSI1, NGN2, and MBD2	Human	YES	2019	3119617337)
Microglial cells	Neuronal cells	Neurod1	Mouse		2019	3063874544)
Cardiomyocytes
Cardiac fibroblasts	Cardiomyocytes	Gata4, Mef2c, and Tbx5	Mouse	YES	2010	2069189949)
Cardiac fibroblasts	Cardiomyocytes	Gata4, Mef2c, Tbx5, and Hand2	Mouse		2012	2266031850)
Cardiac fibroblasts and Embryonic fibroblasts	Cardiomyocytes	Mef2c and Tbx5 + Myocd or Gata4	Mouse		2012	2257576251)
Cardiac fibroblasts	Cardiomyocytes	GATA4, MEF2C, TBX5, MESP1, and MYOCD	Human		2013	2386149452)
Embryonic stem cells- derived fibroblasts	Cardiomyocytes	GATA4, MEF2C, TBX5, ESRRG, MESP1, ZFPM2, and MYOCD	Human	YES	2013	2431966053)
Adult tail-tip fibroblasts and Cardiac fibroblasts	Cardiomyocytes	Mef2c and Tbx5 + Bmi1 knockdown	Mouse		2016	26942853175)
Adult fibroblasts	Cardiomyocytes	Gata4, Hand2, Mef2c, Tbx5, and Znf281	Mouse		2017	28982760167)
Hepatocytes
Embryonic fibroblasts and Adult dermal fibroblasts	Hepatocytes	Hnf4a and Foxa1, Foxa2, or Foxa3	Mouse	YES	2011	2171629155)
Caudal fibroblasts	Hepatocytes	Gata4, Hnf1a, and Foxa3 + p19Arf knockdown	Mouse	YES	2011	2156249256)
Embryonic fibroblasts	Hepatocytes	HNF1A, HNF4A, HNF6, ATF5, PROX1, CEBPA, and MYC + TP53 knockdown	Human	YES	2014	2458292657)
Fetal and adult fibroblasts and Adipose tissue-derived mesenchymal stem cells	Hepatocytes	FOXA3, HNF1A, and HNF4A + SV40 large T antigen	Human	YES	2014	2458292758)
Fibroblasts (BJ and MRC-5 cells)	Hepatocytes	HNF1A and Any two of the three factors: FOXA1, FOXA3, and HNF4A	Human		2014	2496371559)
Liver cells in mouse models of chronic liver disease	Hepatocytes	Foxa3, Gata4, Hnf1a, and Hnf4a	Mouse		2016	2692320160)
Fetal lung fibroblasts	Hepatocytes	ATF5, PROX1, FOXA2, FOXA3, and HNF4A	Human		2017	2919229061)
Fibroblasts	Hepatocytes	OCT4, FOXA2, HNF1A, and GATA3	Human		2017	2858733162)
Embryonic fibroblasts	Hepatocytes	Foxa3, Hnf1a, and Gata4	Mouse	YES	2018	2995986763)
Embryonic fibroblasts	Hepatocytes	Hnf4a, Foxa3, Klf4, and c-Myc	Mouse	YES	2019	3063505464)
Islet-related cells
Liver cells in vivo	β cells	Pdx1	Mouse		2000, 2003	1080271420); 1277571421)
Pancreatic exocrine cells in vivo	β cells	Ngn3, Pdx1, and Mafa	Mouse		2008	1875401165)
Hepatocytes	Islet cells	Ngn3	Mouse		2009	1979388624)
Liver cells	β cells	PDX1, PAX4, and MAFA	Human		2014	2450446266)
Cultured adult pancreatic duct cells	β cells	Pdx1, Ngn3, and Mafa	Mouse		2015	2583666767)
Gallbladder cells	β cells	Pdx1, Ngn, Mafa, and Pax6	Mouse	YES	2016	2783304368)
Hematopoietic system-related cells
T precursor cells	Macrophages	Cebpa or Cebpb	Mouse		2006	1708808422)
T precursor cells	Dendritic cells	Pu.1	Mouse		2006	1708808422)
B cells	T cells	Pax5 knockout	Mouse	YES	2007	1785153273)
Fibroblasts (3T3 cells), Embryonic fibroblasts, and Adult skin fibroblasts	Macrophage-like cells	Pu.1 and Cebpa or Cebpb	Mouse		2008	1842455574)
B cells	Erythroid cells	Gata1, Scl, and Cebpa	Mouse		2012	22968040223)
Fibroblasts (3T3 cells) and Adult dermal fibroblasts	Megakaryocyte	Nfe2, Mafg, and Mafk	Mouse & Human	YES	2012	2285560975)
Skin fibroblasts	Monocytes	Spl1, Cebpa, Mnda, and Irf8	Mouse		2012	2242805876)
Embryonic fibroblasts and Adult ear skin fibroblasts	Hematopoietic progenitor cells	Erg, Gata2, Lmo2, Runx1c, and Scl	Mouse	YES	2014	2546624777)
Fibroblasts	Antigen-presenting dendritic cells	Pu.1, Irf8, and Batf3	Mouse & Human		2018	3053072778)
Bone-related cells
Neonatal foreskin fibroblasts	Chondrogenic cells	c-MYC, KLF4, and SOX9	Human	YES	2013	24146984224)
Dermal fibroblasts	Osteoblasts	OCT3/4 and OCT6 or OCT9 + L-MYC, c-MYC, or N-MYC	Human		2015	2649907479)
Fibroblasts	Osteoblasts	RUNX2, OCT4, OSTERIX, and L-MYC	Human	YES	2015	2591839580)
Gingival fibroblasts and Adult dermal fibroblasts	Osteoblasts	OCT4, OSTERIX, and L-MYC	Human	YES	2016	2699086081)
Embryonic fibroblasts	Osteoblasts	c-Myc, Oct4, and hLMP3	Mouse	YES	2019	3045309282)
Skeletal muscle-related cells
Fibroblasts (C3H10T1/2 cells)	Myoblasts	Myod	Mouse		1987	369066819)
Dermal fibroblasts	Myoblasts	MYOD1 and MYCL	Human	YES	2017	2850162383)
Embryonic fibroblasts	Skeletal muscle progenitor cells	Mef2b and Pitx1 + Pax3 or Pax7	Mouse	YES	2018	2880833984)
Adult fibroblasts	Skeletal muscle progenitor cells	Pax7, Mef2b, and Myod	Mouse	YES	2018	2880833984)
Other cells
Embryonic fibroblasts and Newborn foreskin fibroblasts	Brown fat cells	Prdm16 and Cebpb	Mouse & Human	YES	2009	1964149295)
Embryonic fibroblasts	Sertoli cells	Nr5a1, Wt1, Dmrt1, Gata4, and Sox9	Mouse	YES	2012	2295893196)
Iris-derived cells	Photoreceptor cells	CRX, RAX, and NEUROD	Human		2012	2255817585)
Embryonic fibroblasts and Adult tail-tip dermal fibroblasts	Melanocytes	Mitf, Sox10, and Pax3	Mouse & Human	YES	2014	2551021197)
Adipose tissue-derived stromal cells	Endothelial cells	SOX18	Human		2014	2529018993)
Dermal fibroblasts	Photoreceptor cells	CRX, RAX, OTX2, and NEUROD	Human		2014	2445616986)
Embryonic fibroblasts	Thymic epithelial cells	Foxn1	Mouse	YES	2014	2515098198)
Fibroblasts	Sweat gland cells	NF-κB and LEF-1	Human	YES	2015	2656686891)
Amniotic fluid stem cells	Pluripotent stem cells	OCT4	Human		2016	27019637225)
Cardiac mesenchymal progenitors	Adipocytes	Klf4 and c-Myc	Mouse		2016	2707780687)
Embryonic fibroblasts, Adult tail-tip dermal fibroblasts, Postnatal foreskin fibroblasts, and Fetal dermal fibroblasts	Renal tubular epithelial cells	Emx2, Hnf1b, Hnf4a, and Pax8	Mouse & Human	YES	2016	2782060099)
Endothelial progenitor cells	Smooth muscle cells	MYOCD	Human		2016	2687428189)
Embryonic stem cells	Trophoblast stem cells	Cdx2, Arid3a, and Gata3	Mouse		2017	28973471102)
Embryonic fibroblasts and Adult tail-tip dermal fibroblasts	Leydig cells	Dmrt1, Gata4, and Nr5a1	Mouse	YES	2017	28017657101)
Postnatal dermal fibroblasts	Endothelial cells	ER71/ETV2 (ETS variant 2)	Human	YES	2017	2800321994)
Dermal fibroblasts	Adipocytes	PPARG2	Human		2017	2898267988)
Embryonic fibroblasts and Umbilical vein endothelial cells	Intestine progenitor cells	Hnf4a, Foxa3, Gata6, and Cdx2	Mouse & Human	YES	2017	28943029100)
Embryonic fibroblasts and Adult dermal fibroblasts	Smooth muscle cells	Myocd, Gata6, and Mef2c	Mouse & Human		2018	3002627290)
Epidermal cells	Sweat gland cells	Foxc1	Mouse	YES	2019	3089451792)
Renal proximal tubular epithelial (HK2) cells	Nephron progenitor cells	SNAI2, EYA1, and SIX1	Human		2019	30827514103)

2.1. Neuronal cells.

By the end of 2018, the most reported target cells for direct reprogramming were neuronal lineage cells, including neural stem/progenitor cells (Fig. 1). The first example of the direct lineage conversion to neuronal cells was the reprogramming of astrocytes toward glutamatergic neurons with the transduction of single transcription factors in vitro: Pax6, Mash1, or Ngn2.²⁵⁾ However, the presynaptic output function of the astrocyte-derived neurons was insufficient, and the conversion was only achieved between closely related cells. In 2010, more successfully reprogrammed neuronal cells, termed induced neuronal (iN) cells, were reported. Combinatorial transduction of a set of transcription factors, Brn2, Ascl1, and Myt1l, induced functional neuron-like cells from mouse embryonic fibroblasts (MEFs) and hepatocytes in vitro.²⁶⁾ One year later, some studies generated iN cells from human fibroblasts.^27–29) Although the combination and member of the transcription factors in each study were different, only one transcription factor, Ascl1, was employed in almost every iN cell conversion method. This indicated that Ascl1 may be a vital regulator of neuronal reprogramming but can also be compensated for by other related factors. In addition to terminally differentiated neuronal cells, the direct induction of neural stem/progenitor cells, which are expandable and differentiate to a wide variety of neuronal lineage cells, was also reported.^30–37) The combinations of transcription factors for the induction of neural stem/progenitor cells were significantly different from those of the differentiated iN cells; thus, this might indicate that the transcription networks in the differentiated neuronal cells and the neural stem/progenitor cells are significantly different. In recent years, studies on the generation of multipotent neural stem/progenitor cells have increased because these cells can proliferate in vitro for use in cell transplantation therapies.^30–38) On the other hand, in vivo direct reprogramming studies for terminally differentiated iN cells are also increasing. In these studies, endogenous cells in the brain, such as glial cells, are converted into functional neuron-like cells in situ with plasmid transfection or viral vector infection.^39–44) Furthermore, iN cells have been applied for non-therapeutic applications; for example, direct neuronal reprogramming of cells from patients with severe nerve diseases, such as Huntington’s disease, amyotrophic lateral sclerosis, and myoclonus epilepsy associated with ragged red fibers, have been developed for pathological analysis and drug discovery studies.^45–48)

Figure 1.

The number of publications relating to direct reprogramming studies. The publications relating to direct reprogramming studies of different target cell types are shown as a bar chart. Each different color refers to an individual target cell type. In order to survey the direct reprogramming studies published previously, original articles were searched on PubMed with the following search formula: “direct reprogramming” [All Fields].

2.2. Cardiomyocytes.

Cardiomyocytes, striated muscle cells in the heart, are the second most frequently reported target cells in the field of direct reprogramming (Fig. 1). Cardiomyocytes are one of the most desirable cells for establishment in regenerative medicine because they are not expandable in vivo or in vitro. The first study of mouse-induced cardiomyocyte-like (iCM) cells was published in 2010,⁴⁹⁾ which was followed by other studies using different combinations of transcription factors.^50,51) Human fibroblast-derived iCM cells were established in 2013.^52,53) As well as iN cells, the combination of transcription factors for iCM cells shows wide variation, which might indicate redundancy of the transcription network of cardiomyocytes. However, only Tbx5 was included in all of the combinations of transcription factors for iCM cells. Therefore, it is considered that Tbx5 plays a central role in the transcription network of cardiomyocytes, similar to Ascl1 in neuronal cells. In previous studies, fibroblasts have frequently been employed as source cells for iCM cells because cardiac fibroblasts, which exist abundantly in the heart, are considered a potential candidate for source cells of in situ direct reprogramming. Some studies have achieved successful in vivo direct reprogramming of iCM cells in mouse hearts by using lentivirus vectors.^50,54) This technology may benefit patients with heart injury, such as myocardial infarction.

2.3. Hepatocytes.

Hepatocytes are parenchymal cells of the liver and have vital roles in this organ, i.e., lipid metabolism, glycogen accumulation, detoxification, and endocrine activities. As well as neuronal cells and cardiomyocytes, hepatocytes are also expected to be induced artificially for therapeutic uses because the specific functions of liver-derived hepatocytes cannot be maintained in canonical in vitro culture conditions and are unsuitable for cell transplantation therapy. Although iPSC-derived hepatocytes have been actively studied, the direct reprogramming of hepatocytes is also considered a promising alternative for regenerative treatments of the liver. The first studies of mouse-induced hepatocyte-like (iHep) cells were published by two independent groups simultaneously.^55,56) Although these studies used different protocols to induce iHep cells from fibroblasts, the resulting iHep cells showed similar phenotypes to those of endogenous hepatocytes, i.e., secretion of albumin, uptake of low-density lipoprotein, and xenobiotic metabolism. Furthermore, studies were also conducted using human and mouse iHep cells in a wide variety of reprogramming factor combinations.^55–64) The transcription network defining the cell fate of hepatocytes appears complex and may have redundancy. Among the transcription factors for iHep cell conversion, the Foxa family of transcription factors, i.e., Foxa1, Foxa2, and Foxa3, are considered primary factors for hepatic reprogramming because almost all iHep cell induction methods employ this family proteins as well as Ascl1 and Tbx5 in other direct reprogramming methods. iHep cells are expected to have potential in cell transplantation therapies because the induced cells are expandable and maintain their hepatic functions, unlike the liver-derived hepatocytes. Nevertheless, the human iHep cells required unfavorable factors, such as a virus-derived gene, to be expandable cells; therefore, further improvements are necessary for the medical use of iHep cells. In vivo reprogramming of the iHep cells with an adeno-associated virus (AAV) vector is also studied to treat liver disorders, such as cirrhosis,⁶⁰⁾ because the fibrous liver contains a massive number of fibroblasts as source cells for the direct reprogramming and exclusion target cells.

2.4. Islet-related cells.

Among the five types of endocrine cells in pancreatic islets, β cells are the most critical targets for regenerative medicine because they are the unique source of insulin and are destroyed in type 1 diabetes. Of note, all of the cells reported to be reprogrammed into β cells were endoderm-derived cells, e.g., hepatocytes, bile duct cells, pancreatic exocrine cells, and gallbladder cells, without exception.^{20,21,24,65–71)} The fact that only cells of closely related lineages have been converted into β cells might indicate the peculiarity of β cells from other lineages and plasticity among endoderm-derived cells. Relatively large numbers of studies of the in vivo direct reprogramming of these cells have been reported, because the generation of insulin-secreting cells in digestive organs is an easy and effective solution for type 1 diabetes.

2.5. Hematopoietic system-related cells.

Several terminally differentiated blood cells, e.g., dendritic cells, macrophages, and T cells, were easily converted from other blood cells by transducing with single transcription factors.^72,73) Therefore, these studies were reported earlier the other direct reprogramming studies. On the other hand, the direct reprogramming of blood cells from fibroblasts requires much more complex sets of transcription factors.^74–78) The long-distance lineage conversion appears to require a larger number of transcription factors than the short-distance lineage conversion to drastically change their transcription network. Toward practical use, the in vitro conversion of various types of blood cells has been studied vigorously to prepare cells for blood component transfusion.

2.6. Other types of cells.

Since the first report of the direct reprogramming of somatic cells into myoblasts in 1987,¹⁹⁾ a vast number of different direct reprogramming techniques have been reported; for example, osteoblasts,^79–82) myoblasts,^19,83) skeletal muscle progenitor cells,⁸⁴⁾ photoreceptor cells,^85,86) adipocytes,^87,88) smooth muscle cells,^89,90) sweat gland cells,^91,92) and endothelial cells^93,94) have been induced with several combinations of transcription factors. In addition, the direct conversion of various types of cells, such as brown fat cells,⁹⁵⁾ Sertoli cells,⁹⁶⁾ melanocytes,⁹⁷⁾ thymic epithelial cells,⁹⁸⁾ renal tubular epithelial cells,⁹⁹⁾ intestine progenitor cells,¹⁰⁰⁾ Leydig cells,¹⁰¹⁾ trophoblast stem cells,¹⁰²⁾ and nephron progenitor cells,¹⁰³⁾ has been achieved successfully, and this method is expected to be applied in medical treatment, drug discovery, and pathological studies. Indeed, the researchers of some studies employed the directly reprogrammed cells from patient-derived somatic cells for their disease research.^88,104,105)

2.7. Prediction of the combinations of transcription factors.

Although most direct reprogramming studies select transcription factor combinations from factor lists developed in previous studies in genetics, genomics, and developmental biology, this type of screening is unreliable and inefficient. Therefore, several tools for the prediction of reprogramming factors have been developed. For these studies, a large amount of comprehensive biological data, e.g., transcriptome and topologically associating domains, with computer algorithms, e.g., machine learning and network analysis, were employed to develop the prediction model.^106–110) These tools made the selection of transcription factor candidates and combinations easier. Therefore, the number and variation of direct reprogramming techniques using transcription factors will continue to increase.

3. Methods for direct cell fate conversion

Although the majority of direct reprogramming has been achieved through the forced transduction of transcription factors with viral vectors, these methods are expected to be replaced with alternative methods (Table 2). This is because frequently used viral vectors, retroviruses, and lentivirus vectors integrate the transgenes into the host genome and increase the risk of unexpected cell proliferation, transformation, and oncogenesis. Furthermore, although integration-free transduction methods, e.g., adenovirus vectors, AAV vectors, and plasmid vectors, are employed, the possibility of accidental genome integration of foreign DNA remains. Therefore, integration-free, DNA-free, and transgene-free methods are desired for medical applications. In this section, a wide variety of technologies used in direct reprogramming studies are introduced.

Table 2.

Examples of the direct reprogramming with various methods

Source cells	Target cells	Transgenes and/or Treatment	Species	Cell transplantation	Published year	PMID
Integration-free transduction
Fibroblasts (BJ and MRC-5 cells)	Hepatocytes	HNF1A and Any two of the three factors: FOXA1, FOXA3, and HNF4A (mRNA transfection)	Human		2014	2496371559)
Non-sensory cochlear epithelial cells	Neuronal cells	Ascl1; Ascl1 and Neurod (plasmid electroporation)	Mouse		2014	2492835139)
Cardiac fibroblasts	Cardiomyocytes	Gata4, Mef2c, and Tbx5 (peptide-enhanced mRNA transfection)	Mouse		2015	25834424116)
Astrocytes	Neural stem cells	Sox2 (Leu3p-αIPM system)	Mouse		2016	27148066121)
Fetal and embryonic fibroblasts and Brain cells in vivo	Neuronal cells	Ascl1, Brn2, and Myt1l (mRNA transfection with a GO-PEI-based reagent in vitro and in vivo)	Mouse & Human		2016	2814563140)
Peripheral blood mononuclear cells	Photoreceptor cells	CRX, RAX1, and NEUROD1 (SeV infection)	Human		2016	27170256112)
Gingival fibroblasts and Adult dermal fibroblasts	Osteoblasts	OCT4, OSTERIX, and L-MYC (plasmid transfection)	Human	YES	2016	2699086081)
Fibroblasts	Hepatocytes	OCT4, FOXA2, HNF1A, and GATA3 (protein transduction with HVJ-E)	Human		2017	2858733162)
Fibroblasts	Neurocytes	SOX2, GATA3, and NEUROD1 (protein transduction with HVJ-E)	Human		2017	2858733162)
Fibroblasts (3T6 cells)	Neuronal cells	Ascl1, Brn2, and Foxa1 (plasmid transfection with a polysaccharide-based reagent)	Mouse		2017	28855426119)
Mesenchymal stem cells	Hepatocytes	Hnf4a and Foxa3 (PiggyBac transposon)	Mouse		2017	28295042114)
Dermal fibroblasts	Endothelial progenitor cells	ETV2 (mRNA transfection)	Human	YES	2017	27778229117)
Fibroblasts (3T6 cells)	Neuronal cells	Ascl1, Brn4, and Tcf3 (plasmid transfection with a polysaccharide-based reagent)	Mouse		2017	29137640120)
Mesenchymal stem cells and Dermal fibroblasts	Neural stem cells	SOX2 (mRNA transfection)	Human		2018	29909688118)
Adult fibroblasts	Neural precursor cells	SOX2; SOX2 and PAX6; SOX2 and LMX1A; SOX2, LMX1A, and FOXA2 (plasmid transfection)	Human		2018	3001847136)
Dermal fibroblasts	Neural precursor cells	SOX2 and PAX6 (chemically modified mRNA or plasmid transfection)	Human		2018	3045044034)
Embryonic fibroblasts	Hepatocytes	Hnf4a and Foxa3 (modified mRNA transfection)	Mouse	YES	2018	30327781115)
Foreskin fibroblasts	Neuronal cells	ASCL1 + miR-124 + P53 knockdown (episomal plasmid transfection)	Human		2019	30539819122)
Bone marrow-derived cells, Fibroblasts, and Keratinocytes	Neural precursor cells	MSI1, NGN2, and MBD2 (plasmid transfection)	Human	YES	2019	3119617337)
Renal proximal tubular epithelial (HK2) cells	Nephron progenitor cells	SNAI2, EYA1, and SIX1 (PiggyBac transposon)	Human		2019	30827514103)
Chemical compounds (with or without transcription factors)
Fibroblasts (IMR90 or MRC5 cells)	Cholinergic neurons	NGN2 + Forskolin and dorsomorphin treatment	Human		2013	23873306130)
Embryonic fibroblasts and Neonatal tail-tip fibroblasts	Cardiomyocytes	CHIR99021, RepSox, Forskolin, Valproic acid, Parnate, and TTNPB treatment	Mouse		2015	26292833123)
Embryonic fibroblasts and Adult tail-tip fibroblasts	Neural stem cells	Valproic acid, Bix01294, RG108, PD0325901, CHIR99021, Vitamin C, and A83-01 treatment	Mouse	YES	2015	26788068124)
Fibroblasts	Neuronal cells	Valproic acid, CHIR99021, RepSox, Forskolin, SP600125, GO6983, and Y-27632 treatment	Human		2015	26253202125)
Fibroblasts	Neuronal cells	ISX9, SB431542, Forskolin, CHIR99021, and I-BET151 treatment	Mouse		2015	26253201126)
Embryonic fibroblasts	Neural stem cells	A83-01, Purmorphamine, Valproic acid, and Thiazovivin treatment	Mouse		2017	27207831131)
Urine-derived cells	Hepatocytes	FOXA3 and HNF1A or HNF4A + CHIR99021, RepSox, Valproic acid, Parnate, TTNPB, and Dznep treatment	Human		2018	30315254127)
Embryonic and adult fibroblasts	Skeletal muscle progenitor cells	Myod + CHIR99021, RepSox, and Forskolin treatment	Mouse	YES	2018	29742392128)
Adult dermal fibroblasts, Fetal pancreas fibroblasts, and Peripheral blood mononuclear cells	Neural plate border stem cells	BRN2, KLF4, SOX2, and ZIC3 + CHIR99021, ALK5 inhibitor, Purmorphamine, Tranylcypromine treatment	Human	YES	2018	30581079132)
Striatal neurons	Dopaminergic neurons	Sox2, Nurr1, Lmx1a, and Foxa2 + Valproic acid treatment (in vivo infection)	Mouse		2018	3031829242)
Glioblastoma cells (U87MG cells)	Neuronal cells	Forskolin, ISX9, CHIR99021, I-BET151, and DAPT treatment	Human		2018	30091580129)
Dermal fibroblasts	Neuronal cells	NGN2 and ASCL1 + Pyrintegrin, ZM336372, AZ960, and KC7F2 treatment	Human		2019	31099332133)
Culture condition and soluble factors (with transcription factors)
Liver cells in vivo	β cells	PDX1 + EGF and nicotinamide treatment	Human	YES	2005	15899968134)
Adult dermal fibroblasts and Neonatal foreskin fibroblasts	Multilineage blood progenitor cells	OCT4 + hematopoietic cytokines treatment	Human	YES	2010	21057492135)
Embryonic fibroblasts	Cardiomyocytes	OSKM + cytokines (including Bmp4) treatment	Mouse		2011	21278734136)
Adult skin fibroblasts	Neural stem cells	OSKM + neurosphere culture	Mouse	YES	2012	22467474137)
Amniocytes	Vascular endothelial cells	ETV2, FLI1, and ERG1 + TGFβ inhibitor treatment	Human	YES	2012	23084400139)
Embryonic stem cells-derived fibroblasts	Neural crest cells	SOX10 + Environmental cues including WNT activation	Human	YES	2014	25158936140)
Newborn fibroblasts	Hepatocytes	OSK + Soluble factors	Human	YES	2014	24572354226)
Dermal fibroblasts	Oligodendrocyte progenitor cells	Oct4 + OPC induction medium	Mouse	YES	2015	26497893141)
Embryonic fibroblasts	Keratinocytes	OSK + Bmp4 and RA treatment	Mouse	YES	2016	27473056142)
Skin fibroblasts	Neural stem cells	OCT3/4, SOX2, KLF4, L-MYC, and LIN28 + P53 knockdown + Neurotrophins treatment	Human		2016	27822179138)
miRNAs (with or without transcription factors)
Cardiac fibroblasts	Cardiomyocytes	miR-1, miR-133, miR-208, and miR-499	Mouse		2012	22539765147)
Embryonic fibroblasts	Neuronal cells	Ptb knockdown	Mouse		2013	23313552143)
Cardiac fibroblasts and Embryonic fibroblasts	Cardiomyocytes	Gata4, Mef2c, and Tbx5 + miR-133; Gata4, Mef2c, Tbx5, Mesp1, and Myocd + miR-133	Mouse & Human		2014	24920580148)
Adult dermal fibroblasts, Umbilical cord blood cells, and Senescent somatic cells	Neural stem cells	SOX2 + HMGA2 expression or LET-7B inhibition	Human	YES	2015	25600877144)
Fibroblasts	Cardiomyocytes	GATA4, MEF2C, TBX5, ESRRG, MESP1, MYOCARDIN, ZFPM2, and HAND2 + miR-1	Human		2017	28796841149)
Adult fibroblasts	Neuronal cells	miR-9/9* and miR-124	Human		2017	28886366145)
Adult fibroblasts	Spinal cord motor neurons	ISL1 and LHX3 + miR-9/9* and miR-124	Human		2017	28886366145)
Brain vascular pericytes	Cholinergic neuronal cells	ASCL1, MYT1L, BRN2, and TLX3 + miR-124	Human	YES	2018	29453933146)
Physical stimulation (with or without transcription factors)
Embryonic fibroblasts	Neuronal cells	Ascl1, Brn2, and Myt1l + Substrate topography	Mouse		2014	24709523153)
Fibroblasts	Cardiomyocytes	Myocardin, Tbx5, and Mef2c + Microgroove	Mouse		2015	26302234154)
Embryonic fibroblasts	Dopaminergic neurons	Ascl1, Pitx3, Nurr1, and Lmx1a + Nanogroove	Mouse		2015	25662493155)
Dermal fibroblasts	Endothelial progenitor cells	ETV2 + Hypoxia	Human		2016	27488544151)
Neonatal tail-tip fibroblasts	Cardiomyocytes	Gata4, Mef2c, and Tbx5 + Microgroove	Mouse		2016	27376554156)
Fibroblasts	Cardiomyocytes	Gata4, Mef2c, and Tbx5 + Hypoxia	Mouse		2016	26757100152)
Dermal fibroblasts	Chondrocytes	Electrical stimulation	Human		2019	31005261157)
CRISPR/Cas9-based methods
Embryonic fibroblasts	Neuronal cells	CRISPR/Cas9-based transcriptional activators + gRNAs for Brn2, Ascl1, and Myt1l	Mouse		2016	27524438158)
Liver cells in vivo	β cells	CRISPR/Cas9-based transcriptional activators + gRNA for Pdx1 (in vivo infection)	Mouse	YES	2017	2922478371)
Protein engineering (with transcription factors)
Embryonic fibroblasts	Cardiomyocytes	Gata4, Tbx5, and Hand2 + Mef2c fused with M3 domain of Myod	Mouse		2013	23794713159)
Astrocytes	Neuronal cells	Ascl1, Creb-VP16, and Icer	Mouse		2017	28321434160)
Other methods
Bone marrow stromal cells	Renal proximal tubular epithelial cells	HK2 cell-extracts treatment + streptolysin O	Human	YES	2015	25754206161)
Embryonic fibroblasts	Cardiomyocytes	Gata4, Hand2, Mef2c, and Tbx5 + miR-1 and miR-133 + TGFβ inhibitors or ROCK inhibitors	Mouse		2015	26354680162)
Pancreatic exocrine cells	β cells	Pdx1, Ngn3, and Mafa (synthetic mRNA transfection) + 5-Aza treatment	Mouse		2016	27187823163)
Neonatal cardiac and tail-tip fibroblasts	Cardiomyocytes	miR-1, miR-133, miR-208, and miR-499 + three-dimensional culture	Mouse		2016	27941896164)
Neonatal cardiac and tail-tip fibroblasts	Cardiomyocytes	miR-1, miR-133, miR-208, and miR-499 + Reprogramming medium	Mouse		2016	26975336165)
Adipose-derived stem cells and Neonatal foreskin fibroblasts	Neural stem cells	iPSC induction followed by TRA-1-60–/SSEA4–selection	Human	YES	2016	27569063166)
Fibroblasts	Adipocytes	OCT4 and C/EBPB (protein transduction with HVJ-E) + Adipogenic differentiation medium	Human		2017	2858733162)

3.1. Genome integration-free gene transduction.

Even now, transcription factors are one of the most powerful and useful tools for direct reprogramming. Therefore, several genome integration-free transduction methods have been applied for direct reprogramming instead of conventional viral vectors, such as retrovirus and lentivirus vectors. Sendai virus (SeV; hemagglutinating virus of Japan, HVJ) is an RNA virus that does not enter the nucleus and replicates in the cytoplasm. Therefore, vector systems using this virus are not at risk of genome integration of the transgenes. In addition to the induction of iPSCs,¹¹¹⁾ SeV vectors have been used for direct induction of photoreceptor cells from peripheral blood cells.¹¹²⁾ Additionally, a method for in vivo cardiac reprogramming using SeV vectors achieved high reprogramming efficiency.¹¹³⁾ These results suggested that the SeV vector system can be an alternative tool to induce cellular reprogramming based on the gene transduction method. Moreover, the envelope particle of the virus (HVJ-E) is a useful carrier of genes, siRNAs, proteins, and drugs. Human hepatocytes and neurocytes were induced from fibroblasts through the protein transduction technique with the HVJ-E carrier.⁶²⁾ The PiggyBac system is a transposon-based gene transduction technique. Although transgenes are integrated into the host genome, these sequences can be removed without any footprints with transposase. Some studies have employed this system and induced mouse hepatocytes¹¹⁴⁾ and human nephron progenitor cells from somatic cells.¹⁰³⁾ mRNA transfection is an alternative gene transduction technique that is safer than previous methods. Since higher animals, including humans, do not have reverse transcriptase, there are no risks of integration of the transduced mRNAs into their genome. Many studies using mRNA transduction have been reported, e.g., hepatocytes,^59,115) cardiomyocytes,¹¹⁶⁾ neuronal cells,⁴⁰⁾ endothelial progenitor cells,¹¹⁷⁾ and neural stem/precursor cells.^34,118) Nevertheless, several difficulties of this method, such as stability and transfection efficiency, remain. Solutions to these difficulties, such as the chemical modification of the mRNA^34,115) and specific transfection reagents,^40,116) have already been adopted in mRNA transfection studies. In addition to these methods, plasmid transfection has also been widely tested.^{27,29,32,78,116,117)} Although there is a risk of genome integration and a problem of low transduction efficiency, novel tools, such as a polysaccharide-based gene co-delivery system,^119,120) a Leu3p-αIPM system,¹²¹⁾ or episomal vectors,¹²²⁾ might improve the plasmid-based transduction technologies.

3.2. Chemical compounds.

Small chemical compounds that work as agonists or antagonists for various cellular reactions, such as signal transduction, the cell cycle, post-translational modification, apoptosis, epigenetics, and metabolism, have many advantages compared with gene transduction methods: 1) They are relatively more stable than biomolecules in vitro and in vivo. If they are not stable, chemical modifications can improve their stability. 2) Most can penetrate cells or organelles, unlike biopolymers. 3) There is no risk of genome integration of foreign sequences. 4) Suitable conditions for the chemicals are flexible and controllable.

Most chemicals used in the direct reprogramming studies have positive or negative effects for various signal transduction pathways, e.g., Wnt (CHIR99021),^123–129) cAMP (Forskolin),^{123,125,126,128–130)} TGFβ (RepSox, A83-01, SB431542),^{123–128,131)} Ca²⁺ (ISX9),^126,129) Shh (Purmorphamine),^131,132) Retinoic acid (TTNPB),^123,127) JAK/STAT (AZ960, SP600125),^125,133) Notch (DAPT),¹²⁹⁾ AMPK (dorsomorphin),¹³⁰⁾ PCK (GO6983),¹²⁵⁾ HIF (KC7F2),¹³³⁾ MAPK (PD0325901, ZM336372),^124,133) β1-integrin (Pyrintegrin),¹³³⁾ and Rho/ROCK (Thiazovivin, Y-27632)^125,131) signaling pathways, and epigenetic modifications, e.g., histone lysine deacetylation (Valproic acid),^{42,123–125,127,131)} histone lysine acetylation recognition (I-BET151),^126,129) histone lysine methylation (Bix01294, Dznep),^124,127) DNA methylation (RG108),¹²⁴⁾ and DNA demethylation (Vitamin C).¹²⁴⁾ Although several types of cells, such as neuronal cells, hepatocytes, cardiomyocytes, and skeletal muscle progenitor cells, have been induced with chemicals, half of the reported studies used gene transduction in conjunction with the chemical treatments.^{42,127,128,130,132,133)} Precise analysis of the functions of each transcription factor is required to compensate their functions with chemical compounds.

3.3. Partial reprogramming with specific culture conditions.

Some studies have utilized gene transduction combinations and external signals from culture media. In particular, initializing factors of iPSCs have frequently been used in combination with external signals. A wide variety of cells, e.g., β cells,¹³⁴⁾ blood progenitor cells,¹³⁵⁾ cardiomyocytes,¹³⁶⁾ neural stem cells,^137,138) vascular endothelial cells,¹³⁹⁾ neural crest cells,¹⁴⁰⁾ hepatocytes,¹⁴⁰⁾ oligodendrocyte progenitor cells,¹⁴¹⁾ and keratinocytes,¹⁴²⁾ were previously generated and transplanted into model animals. Although these protocols are time- and cost-effective compared with iPSC-employing methods, the risk of tumorigenesis caused by cell initialization remains to be elucidated.

3.4. MicroRNAs (miRNAs).

miRNAs regulate the expression of specific mRNAs post-transcriptionally. Therefore, miRNAs can be used for direct reprogramming instead of or in combination with gene transduction. In a previous study, somatic cells were converted into several types of neuronal cells^143–146) and cardiomyocytes.^147–149) miRNA induction avoids the problem of transgene genome integration because the miRNA genes can be induced into cells as RNA. However, most of the reported studies employed gene transduction simultaneously for conversion. Complete miRNA-based reprogramming methods are expected to be developed in future.

3.5. Physical stimulations.

Several physical stimulations have been used as additional stimulations for direct reprogramming. Hypoxia conditions have been frequently employed in stem cell biology and control stem cell functions through the HIF signaling pathway.¹⁵⁰⁾ In a study involving direct reprogramming, these conditions were also used to accelerate conversion.^151,152) On the other hand, micro- or nano-imprinted patterns, e.g., microgrooves, of cultured cell substrates are known as enhancers of reprogramming in biomaterials science. Biochemical and molecular biology studies are required to understand the effect of imprinted patterns for reprogramming.^153–156) Of note, the electrical stimulation-induced direct reprogramming of fibroblasts into chondrocytes without any soluble factors or transgenes was reported recently.¹⁵⁷⁾ Elucidating the molecular mechanism of the reprogramming phenomenon may aid in the development of direct reprogramming technology.

3.6. Other methods.

Several other innovative approaches have been applied in direct reprogramming studies, such as a CRISPR/Cas9-based method. A modified Cas9 protein, which has a domain for transcriptional activation and guides RNAs for the endogenous transcription factor genes, is introduced into source cells to activate endogenous transcription factors.^71,158) The genome integration-free protocols of the CRISPR/Cas9 system have already been established, thus, this method is also an integration-free method. Studies utilizing CRISPR-based direct reprogramming will increase because this method is straightforward and suitable for screening reprogramming factors.

Rational protein engineering of transcription factors is a unique approach translated from another field, i.e., biomolecular engineering. Functional protein domains, such as VP16 or M3 of Myod, are attached to transcription factors to improve their transcriptional activities. Neuronal cells and cardiomyocytes have been induced successfully by using artificial transcription factors.^159,160)

Additionally, the number of hybrid methods using the techniques above are increasing.^161–166) In future, engineering-based approaches optimizing the reprogramming process will be considerably more important compared with previous stages.

4. Mechanism analyses and future applications of direct reprogramming

To make direct reprogramming technology more efficient and safer for medical and industrial applications, analyses of the molecular and cellular mechanisms of cell fate conversion, i.e., the transition of the transcriptome, the behavior of transcription factors, epigenetic remodeling, and the heterogeneity of cell populations, are essential (Fig. 2). In addition, applied research on direct reprogramming technology in medicine and industry has increased in recent years toward practical use (Table 3). Therefore, representatives of molecular and cellular mechanism analyses are introduced in this section. In addition, novel applications of direct reprogramming, including non-medical uses, are described.

Figure 2.

(Color online) Different study types for direct reprogramming. Typical study types for direct reprogramming are indicated. Cutting-edge biochemistry and cell biology techniques have been employed for direct reprogramming studies. TF: transcription factor.

Table 3.

Applications of the directly reprogrammed cells

Source cells	Target cells	Transgenes and/or Treatment	Species	Published year	PMID
In vivo reprogramming (regenerative medicine)
Liver cells	β cells	Pdx1 (an adenoviral vectors)	Mouse	2000, 2003	1080271420); 1277571421)
Pancreatic exocrine cells	β cells	Ngn3, Pdx1, and Mafa (an adenoviral vector)	Mouse	2008	1875401165)
Cardiac fibroblasts	Cardiomyocytes	Gata4, Mef2c, Tbx5, and Hand2 (a retrovirus vector)	Mouse	2012	2266031850)
Cardiac fibroblasts	Cardiomyocytes	miR-1, miR-133, miR-208, and miR-499 (a lentiviral vector)	Mouse	2012	2253976554)
Non-sensory cochlear epithelial cells	Neuronal cells	Ascl1; Ascl1 and Neurod (in vivo plasmid electroporation)	Mouse	2014	2492835139)
Liver cells in mouse models of chronic liver disease	Hepatocytes	Foxa3, Gata4, Hnf1a, and Hnf4a (an AAV vector)	Mouse	2016	2692320160)
Brain cells in vivo	Neuronal cells	Ascl1, Brn2, and Myt1l (mRNA transfection with a GO-PEI-based reagent)	Mouse	2016	2814563140)
Bile duct cells	β cells	Pdx1, Ngn3, and Mafa (an adenoviral vector)	Mouse	2017	2836326969)
Hepatocytes	β cells	Pdx1, Ngn3, and Mafa (hydrodynamics tail vein injection)	Mouse	2017	2810087170)
Resident glial cells	Neuronal cells	Ascl1, Lmx1a, and Nurr1 (an AAV vector)	Mouse	2017	2884465841)
Liver cells in vivo	β cells	CRISPR/Cas9-based transcriptional activators + gRNA for Pdx1 (an AAV vector)	Mouse	2017	2922478371)
Striatal neurons	Dopaminergic neurons	Sox2, Nurr1, Lmx1a, and Foxa2 (a lentiviral vector) + Valproic acid treatment	Mouse	2018	3031829242)
Resident glial cells	Interneurons	Ascl1, Lmx1a, and Nurr1 (an AAV vector)	Mouse	2019	3125990143)
Microglial cells	Neuronal cells	Neurod1 (a lentiviral vector)	Mouse	2019	3063874544)
Disease model
Fibroblasts from Huntington's disease patients	Neuronal cells	PTB1 knockdown	Human	2014	2527553345)
Urine-derived cells from muscle diseases patients	Skeletal muscle cells	MYOD	Human	2016	27651888104)
TSC2 gene-inactivated fibroblasts (CRISPR knockout)	Neuronal cells	ASCL1, LMX1A, and NURR1	Human	2016	27857203207)
Adult skin fibroblasts from ALS patients	Motor neurons	NEUROG2, SOX11, ISL1, and LHX3	Human	2016	2672511246)
Dermal fibroblasts from metabolic disease patients	Adipocytes	PPARG2	Human	2017	2898267988)
Fibroblasts from BMD patients	Skeletal muscle cells	MYOD	Human	2018	30171539105)
Fibroblasts from MERRF patients	Neuronal cells	ASCL1 and BRN2 + REST complex knockdown	Human	2019	3079779847)
Other animals
Embryonic fibroblasts	Neuronal cells	ASCL1, BRN2, MYT1L, and NEUROD1	Marmoset	2014	24694048211)
Skin fibroblasts	Cardiomyocytes	GATA4, HAND2, TBX5, and MEF2C	Canine	2015	26681949209)
Cardiac fibroblasts	Cardiomyocytes	Gata4, Mef2c, and Tbx5 + miR-590	Rat, Porcine, & Human	2016	27930352208)
Fetal fibroblasts	Cardiomyocytes, Neurocytes, Oocytes, and Cumulus granulosa cells	OSKM + Lineage-specific medium	Goat	2017	27629151210)
Drug discovery
Embryonic stem cells	Neuronal cells	NGN2	Human	2019	3115548448)

4.1. Analyses of the molecular and cellular mechanisms of direct reprogramming.

Regarding the mechanisms of direct cell fate conversion, the reprogramming processes of cardiomyocytes,^{148,149,165,167–179)} neuronal cells,^{39,45,48,143,145,160,180–188)} and hepatocytes^189–192) have been well-studied among the various direct reprogramming methods. Additionally, several studies regarding the direct reprogramming of other cell types, e.g., neural stem/progenitor cells,^193,194) glial cells,¹⁹⁵⁾ photoreceptor cells,¹⁹⁶⁾ β cells,¹⁹⁷⁾ renal tubular epithelial cells,¹⁹⁸⁾ trophoblast stem cells,¹⁰²⁾ and mesodermal cells,^199,200) have been reported.

Several methodologies have been employed for the molecular biological analysis of direct reprogramming. Comprehensive transcriptome analyses using RNA-seq have been performed frequently, and differential expression gene analysis between source and target cells has provided a considerable amount of information regarding essential regulator genes and signaling pathways during conversion.^{143,148,176,179,180,200)} Epigenetic remodeling studies during cell-fate conversions have also been performed in various reprogrammed cells.^{44,48,171,175,177,182,188,191,194,195)} Histone modification and DNA methylation data clarified the drastic and rapid change in chromatin states in reprogrammed cells. In addition, detailed analysis of transduced and endogenously upregulated transcription factors was a critical task for understanding direct reprogramming. Several methods, such as loss-of-function, ChIP-seq, and mass spectrometry, have been used to analyze the function and behavior of transcription factors.^{102,145,160,167,169,172,173,179,184–186,190,195,199)} Transcription factor studies indicated that almost all of the reprogramming-related transcription factors played vital roles during the reprogramming process, e.g., by remodeling histone and DNA modifications, initiating the target transcription network, and erasing the transcriptional signature of the source cells. Recently, the activities of specific transcription factors, termed pioneer factors, seem pivotal for remodeling the chromatin states of cell conversion.²⁰¹⁾ For example, Ascl1 was identified as a pioneer factor for the induction of iN cells, which can preferentially bind to chromatin and allowed the binding of other reprogramming factors used for iN cell induction.²⁰²⁾ Additionally, in the case of iHep cell induction, well-known pioneer factors from the Foxa protein family are indispensable.^{55,56,58–64,114,115,127,189)} Therefore, understanding the roles of pioneer factors will improve direct reprogramming technology. Additionally, the function of miRNA in iCMs has been thoroughly analyzed, and, therefore, miRNA-based reprogramming technologies of iCMs have been developed.^148,165,177) The analysis of miRNAs in other direct reprogramming methods may result in their improvement. Other molecular and cellular analyses, such as signal transduction,^174,183,197) metabolic remodeling,¹⁹⁸⁾ time-lapse imaging,¹⁷⁰⁾ and cell morphological studies,¹⁶⁸⁾ have been performed. Recently, single-cell analyses, such as single-cell RNA sequencing (scRNA-seq), have been performed in many direct reprogramming studies.^{149,181,187,192,199)} These studies have clarified many features of direct reprogramming processes. For example, a limiting step and an obstruction factor,¹⁸⁷⁾ a stem cell-like transition state,¹⁸¹⁾ and multiple trajectories¹⁹²⁾ of the direct cell fate conversions were discovered using scRNA-seq data. In the near future, other single-cell analysis techniques, such as ATAC-seq,²⁰³⁾ ChIL-seq,²⁰⁴⁾ CUT&RUN,²⁰⁵⁾ and Hi-C,²⁰⁶⁾ will reveal the chromatin dynamics and epigenetic remodeling processes in single converting cells during direct reprogramming processes.

4.2. Medical and industrial applications of direct reprogramming.

As described above, most applied research for direct reprogramming has focused on clinical medicine, i.e., cell transplantation and in vivo direct reprogramming therapies. Meanwhile, some non-medical applications have been reported in recent years (Table 3). Direct reprogramming studies of disease model cells, which are induced from patient-derived cells, are necessary for basic studies of different disorders, and these cells can be employed for in vitro experiments for drug discovery. Indeed, a human iN cell type, instead of primary neuronal cells, was previously applied to in vitro drug treatment experiments.⁴⁸⁾ Additionally, several directly induced neuronal and skeletal muscle cell types from patients with various diseases have been reported.^{45–47,104,105,207)} Direct reprogramming technology has the potential to change the pharmaceutical industry. In direct reprogramming studies, cells from different species, e.g., rat,²⁰⁸⁾ pigs,²⁰⁸⁾ dogs,²⁰⁹⁾ goats,²¹⁰⁾ and marmosets,²¹¹⁾ have been generated. Some of the animals studied were industrial animals, such as livestock or pets, since direct reprogramming technology might contribute to animal industries, such as veterinary medicine.

5. Summary

Direct reprogramming is a promising technology that can convert somatic cells into various terminally differentiated or somatic stem/progenitor cell types for medical and industrial applications. Therefore, the number of studies utilizing this technology has increased rapidly in the last decade. Artificial cell fate conversions can be achieved with the forced expression of defined sets of transcription factors, treatment with chemical compounds as agonists or antagonists for various bioreactions, culturing with soluble factors or specific culture conditions, the induction of miRNA sets, appropriate physical stimulations, among other methods. Various cell types have been induced from somatic cells through direct reprogramming. Among them, neuronal cells, cardiomyocytes, and hepatocytes have been well-studied and their reprogramming mechanisms have been thoroughly analyzed. These technologies are expected to be applied in a variety of industries such as agriculture, biomaterial, healthcare, and medical industries.

Biographies

Profile

Kenichi Horisawa was born in Tochigi prefecture, Japan, and graduated from University of Tsukuba in 1997. After 2 years in a scientific publishing company, he entered the Graduate School of Science and Technology, Keio University, and received his Ph.D. degree under the supervision of Prof. Hiroshi Yanagawa in 2005. He worked as a postdoctoral fellow in Prof. Hiroshi Yanagawa’s laboratory from 2005 to 2010, and as an Assistant Professor in Prof. Nobuhide Doi’s laboratory from 2010 to 2013 in the Faculty of Science and Technology, Keio University. In 2013, he joined Prof. Atsushi Suzuki’s laboratory at the Medical Institute of Bioregulation, Kyushu University, as an Assistant Professor. His research interests are molecular mechanisms for the regeneration and direct reprogramming of hepatic cells, primarily on the molecular function and behavior of the master transcription factors for hepatic development.

Atsushi Suzuki was born in Gunma prefecture, Japan, in 1974 and graduated from Tohoku University in 1998. Then, he moved to University of Tsukuba with an aim to contribute to the development of innovative medicine using the knowledge and experience obtained from the study of basic biology. Subsequently, he succeeded in the prospective isolation of hepatic stem cells from the developing mouse liver and unraveled the mechanisms controlling the properties of hepatic stem cells during liver development. He made his Ph.D. thesis defense in the summer of 2002 and then spent time at the Salk Institute for Biological Studies, San Diego, U.S.A., to learn novel experimental techniques. After being awarded his Ph.D. from University of Tsukuba in March 2003, he continued his work at the Salk Institute until returning to RIKEN/CDB in Japan in 2005. Two and half years later, he moved to Kyushu University to join the tenure track program of “Kyushu University Research Superstar Program (SSP)” and became a principal investigator in his own laboratory. In that time, he decided to start new research projects, and finally succeeded in the induction of direct conversion of skin-derived fibroblasts to hepatocytes using defined transcription factors. He subsequently became an Associate Professor in 2011 and then Professor at Kyushu University in 2013. His ongoing studies are focused on not only liver development, regeneration, and diseases but also the determination and conversion of the fate of cells in various digestive organs, with the hope that these will provide new insights into therapies for diseases in organs of the digestive system.