THE LONG AND SHORT OF IT: The epithelium of the cervix consists of both flattened squamous cells (left, top), which occur in many layers and slough off as they are replaced by cells beneath them, and columnar cells (left, bottom), which form a mucus-secreting monolayer covered in microvilli.© P.M. MOTTA & S. MAKABE/SCIENCE SOURCE IMAGES

In 1998, University of Bath biologist David Tosh had a eureka moment. He noticed that some of the rat pancreatic exocrine cells he was working with had become unusually large and flat. After further testing, he identified the cause of the abnormality: the cells, which belonged to an established line called AR42j, were no longer pancreatic cells at all; they were hepatocytes, the principal cell type of the liver. The cells had changed their identity simply under the influence of the synthetic hormone dexamethasone, which had been added to the medium to enhance...

This wasn’t the first demonstration that cells could change type. Pathologists had long reported the transformation of one cell type into another in humans, a normally harmless process known as metaplasia.2 The transdifferentiation of the AR42j cells, however, offered a convenient in vitro model to study such cellular flip-flopping, and Tosh, others, and I believed that by better understanding this phenomenon, we could use it to our advantage. Indeed, in the past two decades, the study of cellular reprogramming has gone from basic science to applied bioengineering, with researchers now able to create a variety of important cell types.

The ability to generate cell types of interest in the lab has vast potential in both research and medicine. It is now possible to create cultures of human cells, such as neurons, hepatocytes, or cardiomyocytes, that would be difficult to sample from living subjects. Researchers can take skin or blood samples from patients suffering from diseases of interest and use those cells to derive disease-specific stem cell lines from which many cell types can be made. Cell reprogramming technologies may one day offer a new source of useful cells for therapeutic transplantation to treat diseases such as diabetes and Parkinson’s, among others. But first, researchers must better understand how cells switch type and be able to confirm that a cell expressing different genes or displaying a different morphology has indeed made a permanent identity change.

Reprogramming in nature

Classifying human or animal cell types traditionally means staining them with appropriate dyes and then viewing them under a light microscope. Any visually distinguishable cell type normally contains relatively large quantities of a few proteins that are associated with its specific function. Contractile proteins are found in muscle fibers, for example, while neurofilaments are found in neurons’ axons. Some differentiated cell types can persist for the life of the organism. Others have a finite lifetime and are replaced by new differentiated cells derived from progenitors.

For many years, researchers believed that mature cell types, stable as they were, could not be changed into different types of cells. In recent years, however, the engineered overexpression of specific transcription factors has yielded many dramatic transformations previously not thought possible—including the regression of adult fibroblasts to stem cell–like states and the direct conversion of mature skin cells to neurons.

Some cells are now even known to switch type during normal development. The epithelial lining of the esophagus, for example, is formed as a columnar epithelium, like the rest of the gut tube. Only later does it transform into a stratified squamous epithelium resembling that of the skin.3 Pancreatic endocrine cells are also seen occasionally popping up in the bile ducts of the liver. This change occurs in late embryogenesis and involves a switch from a ductal to an endocrine cell type.4

In the past two decades, the study of cellular reprogram­ming has gone from basic science to applied bioengineering, with researchers now able to create a variety of important cell types.

One useful model for understanding these cellular metamorphoses is the process of transdetermination in Drosophila. The imaginal discs of fly larvae are sometimes recoded during regeneration, such that an incorrect body part—a leg instead of a wing, say—develops due to aberrant expression of transcription factors.5 Comparable phenomena are found in other arthropod groups. It has long been known, for instance, that if a claw is amputated from a spiny lobster in genus Palinurus, it may regenerate not as a claw but as an antenna. And “serial heteromorphosis,” which involves the replacement of an appendage by one normally found elsewhere on the body, is observed in both crustaceans and insects.

TYPE CHANGE: This light micrograph shows a patch of squamous epithelial cells (flat, scale-like cells stained pink) in the human cervix, where normally there are only non-squamous cells. Such squamous metaplasia can be a protective response to an acidic vaginal environment, but metaplastic squamous epithelium can also be unstable and a potential site for cancer. © SPL/SCIENCE SOURCECellular identity switching is even seen, on rare occasion, in adult vertebrates. In some newts and salamanders, the lens of the eye can regenerate. After its removal, the cells of the dorsal iris proliferate, undergo depigmentation, and eventually redifferentiate to form a new lens.6 The two cell types are very different: iris cells are pigmented epithelial cells similar to those of the pigmented retina, while the lens is composed of modified keratinocytes containing high concentrations of crystallin proteins, which impart its characteristic transparency. Like the Drosophila discs, this system suggests a connection between wounding, a regenerative process, and the transformation of cell or tissue type.

In human tissues it is not unusual to find small bits of misplaced tissue, such as deposits of bone in the soft connective tissue, or patches of squamous (skin-like) epithelium in an epithelium that is normally glandular.7 Metaplasias are particularly common in the gastrointestinal and female reproductive tracts, where tissues may be subjected to chronic trauma, infection, or hormonal stimulation, leading to tissue regeneration. (See micrograph above.) Interestingly, the erroneous tissue type is often the same as that derived from cells that were embryonic neighbors of the cells that gave rise to the host tissue of the metaplasia. Patches of intestinal tissue sometime arise within the gastric mucosa, for example, and intestine and stomach develop from adjacent territories of the endoderm in the early embryo. Similarly, large intestine–type tissue can arise in the urinary bladder, a condition called cystitis glandularis. While the bladder is quite separate from the gut in the adult, it is derived from neighboring endoderm in the embryo.

Most metaplasias are harmless, but some increase the risk of cancer. For example, patches of squamous tissue in the columnar epithelium–lined lung bronchi, often seen in smokers, can sprout into lung cancer. And adenocarcinoma of the esophagus usually arises in areas of Barrett’s metaplasia, a condition in which the normally squamous epithelium of the lower esophagus becomes converted to a columnar epithelial type, with differentiation patterns typical of the intestine.

Pushed to the limits

CELLULAR REPROGRAMMING: For the better part of the past decade, researchers have been reprogramming adult cell types, either into induced pluripotent stem cells (iPSCs), which themselves can give rise to diverse cell types, or directly into other differentiated cell types through a process called direct reprogramming. Such approaches support the switching of diverse cell types once believed to be permanently locked in their differentiated form.
See full infographic: JPG | PDF
Early experiments aimed at altering cell fate involved moving cells from one part of an embryo to another, where they were exposed to different signaling factors. This type of experiment, which is very different from modern reprogramming involving the introduction of transcription factors into cells, made up much of classical experimental embryology and led to the discovery of a number of important developmental signaling systems, such as the Wnt and Hedgehog pathways, in the 1980s and ’90s.8

In 2006, Shinya Yamanaka’s group at Kyoto University announced a new strategy for cell reprogramming. By engineering the overexpression of a cocktail of pluripotency-associated transcription factors in mouse fibroblasts, the researchers generated the world’s first induced pluripotent stem cells (iPSCs).9 Several labs joined in, and before long iPSCs could be made using a variety of gene cocktails and from the cells of a range of species, including humans. (See illustration.)

iPSCs are very similar to embryonic stem cells. They can grow without limit in vitro and, when exposed to suitable media, can be prompted to differentiate into a wide variety of cell types. Their formation represents a cell-type transformation that is discrete and permanent. Mouse iPSCs will integrate into the mouse embryo, capable of contributing to all tissues, including the germ line. In vitro, human iPSCs have been redifferentiated into nearly all types of cells, including neurons, hepatocytes, and cardiomyocytes. They are seen as a valuable model for studying human development and disease; as a convenient system for obtaining useful cell types for drug testing; and, eventually, as a source of cells for therapeutic transplantation.

Since the discovery of iPSCs, researchers have refined Yamanaka’s techniques to increase transformation efficiency. By opening chromatin to increase the access of transcription factors to their target genes, the proportion of cells reprogrammed can be increased from about 1 in 100,000 to 1 in 1,000.5 Some researchers have even claimed to achieve nearly 100 percent efficiency. It is also now possible to make iPSCs from cell types other than fibroblasts. In particular, lymphocytes isolated from blood can serve as the starting material for cell dedifferentiation, making it easy to prepare iPSCs from individual patients. Stem-cell researchers are also beginning to forgo integrating vectors such as retro- or lentivirus, which incorporate the experimental transcription factor genes into the host genome, in favor of nonintegrating or excisable vectors or RNA, which have much lower risks of introducing harmful mutations. Various small molecules can now be substituted for some of the original reprogramming transcription factors, and two recent Nature studies demonstrated how pluripotency can be induced in differentiated adult mouse cells with a simple external stressor, such as low pH. (See “New Method for Reprogramming Cells,” The Scientist, January 29, 2014.)

However, despite their useful qualities and the fact that they closely resemble embryonic stem cells, iPSC lines commonly carry some “memory” of the cell type that they used to be, encoded as specific DNA methylation. These epigenetic marks may bias the cells toward differentiation into their parent type. By contrast, embryonic stem cells do not carry such biases, as they are derived from cells of the not-yet-differentiated inner cell mass of the early embryo.

Another important strand of reprogramming research is the direct conversion of one cell type to another via the overexpression of specific transcription factors. The first example of cells changing type without first passing through a pluripotent stage was the discovery of the myogenic factor MyoD by Harold Weintraub in 1989. Identified in the course of experiments on muscle development, MyoD alone can reprogram a variety of tissue-culture cell lines into multinucleated muscle fibers known as myotubes.10

Thanks to improvements in iPSC technology, it is now possible to derive an iPSC line from nearly any patient sample.

Since Weintraub’s time, researchers have achieved numerous other kinds of direct conversion, including the transformation of pancreatic exocrine cells into hepatocytes or endocrine cells; B lymphocytes into macrophages; and fibroblasts into neurons, cardiomyocytes, and hepatocytes. (See illustration.) The overexpression of just one to four transcription factors can achieve these transformations—typically, factors involved in the normal embryonic development of the cell type you wish to produce. These sets of transcription factors serve not simply to activate a group of target genes, but also to silence others and shift the cell into a new, stable state of gene expression.

These transformations appear to be genuinely “direct;” they do not involve reversion to a primitive cell state followed by a process of normal development. As such, they are very dramatic and have excited much interest. My own lab has had a long-standing interest in the possible generation of pancreatic beta cells by direct reprogramming, as it could have implications for the treatment of diabetes. The transplantation of islet cells is an existing and fairly successful therapy, but the number of donors fails to meet demand. Beta-like cells could fill this void, and if generated from cells of individual patients, could avoid the existing problem of immune rejection of the transplanted cells, although patients with Type 1 diabetes would still need some modest immunosuppression to deal with the autoimmunity of the disease itself.

Path to a new identity

JUMPING BETWEEN CELLS: While induced pluripotent stem cell (iPSC) biology surges ahead, some researchers aim to skip the pluripotency step altogether, directly coaxing adult cells of one type to become another. The overexpression of just one or a few transcription factors, typically those that are involved in the embryonic development of the desired cell type, can spur these dramatic transformations.
See full infographic: JPG | PDF
An important issue in cell reprogramming is the ability of introduced factors to find their gene targets in closed chromatin. Much of the DNA in vivo is wrapped up in nucleosomes and occluded by histones, and it can be further repressed by higher-order chromatin structures and by repressor complexes. Genome-wide localization studies show that most potential transcription factor binding sites are unoccupied, suggesting that most nuclear DNA is inaccessible.

But some transcription factors are able to locate their targets even in closed chromatin.11 MyoD was the first so-called “pioneer factor” to bring about direct reprogramming in intact chromatin. Moreover, if the MyoD transcription activation domain is added to the Oct4 gene sequence, then the resulting modified Oct4 protein, a key transcription factor for inducing or maintaining pluripotency, is much more active and generates iPSCs in less time than wild-type Oct4.12 The cocktails used for direct reprogramming usually include at least one pioneer factor. The ablation of Mbd3, the gene for a component of a major chromatin repression complex, can also substantially increase the efficiency of iPSC generation.13 Thanks to these and other improvements in iPSC technology, it is now possible to derive an iPSC line from nearly any patient sample.

The developmental relatedness of two cell types can also have an influence on direct reprogramming. In principle, the hierarchical nature of normal development might suggest that the chromatin states of related tissue types are more similar than those of distant tissue types. If this is true, the transcription factors used to induce expression of cell-specific genes should have easier access to their targets, so perhaps reprogramming would be more efficient. This is supported by the relative ease of generating pancreatic-like cells from the liver and of cardiomyocytes from cardiac fibroblasts. On the other hand, several recent examples of the direct reprogramming of fibroblasts suggest that it is possible to overcome any barrier with sufficiently potent transcription factors supplied at high doses.

It is critical to understand that reprogramming does not just represent upregulation of various genes by the introduced transcription factors. The transcription factors used for such processes normally have a number of differentiated gene products among their direct targets, but the genes are only expressed as long as the introduced transcription factors are present. Once these are lost, the cell phenotype will revert to the original. Real reprogramming involves an irreversible change of phenotype that persists despite the loss of the factors that originally provoked the change.

GROWING A NEW EYE: As a newt's eye regenerates, dorsal iris cells dedifferentiate (gray area, top; at higher magnification, right) before proliferating and redifferentiating into a new lens, which comprises modified keratinocytes that are morphologically distinct from the pigmented epithelial cells of the iris. COURTESY OF P.A TSONIS AND K. DEL RIO-TSONISIn the case of iPSCs, there is no doubt about the reality of reprogramming, because the cells will grow without limit, and the new state is stable. With direct reprogramming, however, it can be more difficult to ensure that the new gene expression pattern is really permanent. Use of nonintegrating vectors such as adenovirus to introduce the transcription factors leads to eventual loss of the viral DNA, but it can take a long time. Inducible systems, such as those that can be regulated via the administration of a particular drug, can help control the length of time reprogramming factors are expressed, but such systems can be difficult to turn off altogether and often show some expression in the absence of the drug. Another way to assess a cell’s type is to measure certain physiological characteristics, such as the elicitation of action potentials from induced neurons. This is important, but sometimes the physiological properties under study may be conferred by just a few gene products—perhaps those that are direct targets of the introduced factors—and therefore displayed without genuine reprogramming.

Interestingly, some experiments involving the administration of transcription factors to intact animals using gene-delivery viruses indicate that it may be easier to achieve transformations in vivo than in vitro. This is unexpected, as one might think that the complex organ environment, with cell types from different sources and a variety of external stimuli, would tend to stabilize a cell. However, it is possible to induce iPSCs in mice, where they tend to form tumors known as teratomas that contain pluripotent stem cells as well as a variety of differentiated derivatives.14 It is also possible to transform cardiac fibroblasts into cardiomyocytes in living adult mice,15 and to elicit at least a partial transformation of pancreatic exocrine cells towards pancreatic beta cells.16 Although naturally occurring cell-type transformations are rare in vivo, the ability to provoke them offers yet another opportunity for creating potential new therapies.

Jonathan Slack is an emeritus professor at the University of Minnesota (U.S.) and the University of Bath (U.K.). From 2007–2012 he directed the Stem Cell Institute at the University of Minnesota, and he has written several books on developmental and stem cell biology.


  1. C.N. Shen et al., “Molecular basis of transdifferentiation of pancreas to liver,” Nat Cell Biol, 2:879-87, 2000.
  2. J.M.W. Slack, “Homeotic transformations in man: implications for the mechanism of embryonic development and for the organization of epithelia,” J Theor Biol, 114:463-90, 1985.
  3. W.Y. Yu et al., “Conversion of columnar to stratified squamous epithelium in the developing mouse oesophagus,” Dev Biol, 284:157-70, 2005.
  4. J.R. Dutton et al., “Beta cells occur naturally in extrahepatic bile ducts of mice,” J Cell Sci, 120:239-45, 2007.
  5. L. Maves, G. Schubiger, “Transdetermination in Drosophila imaginal discs: a model for understanding pluripotency and selector gene maintenance,” Curr Opin Genet Dev, 13:472-79, 2003.
  6. P.A. Tsonis et al., “A newt’s eye view of lens regeneration,” Int J Dev Biol, 48:975-80, 2004.
  7. R.A. Willis, The Borderland of Embryology and Pathology, Butterworth, 1962.
  8. J.M.W. Slack, Essential Developmental Biology, 3rd ed., Wiley-Blackwell, 2012.
  9. K. Takahashi, S. Yamanaka, “Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors,” Cell, 126:663-76, 2006.
  10. H. Weintraub et al., “Activation of muscle-specific genes in pigment, nerve, fat, liver, and fibroblast cell-lines by forced expression of MyoD,” PNAS, 86:5434-38, 1989.
  11. K.S. Zaret, J.S. Carroll, “Pioneer transcription factors: establishing competence for gene expression,” Genes Dev, 25:2227-41, 2011.
  12. H. Hirai et al., “Radical acceleration of nuclear reprogramming by chromatin remodeling with the transactivation domain of MyoD,” Stem Cells, 29:1349-61, 2011.
  13. Y. Rais et al., “Deterministic direct reprogramming of somatic cells to pluripotency,” Nature, 502:65-70, 2013.
  14. M. Abad et al., “Reprogramming in vivo produces teratomas and iPS cells with totipotency features,” Nature, 502:340-45, 2013.
  15. L. Qian et al., “In vivo reprogramming of murine cardiac fibroblasts into induced cardiomyocytes,” Nature, 485:593-98, 2012.
  16. Q. Zhou et al., “In vivo reprogramming of adult pancreatic exocrine cells to beta-cells,” Nature, 455:627-32, 2008.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?