Special Issue Reviews–A Peer Reviewed Forum
Transdifferentiation in developmental biology, disease, and in therapy
Transdifferentiation (or metaplasia) refers to the conversion of one cell type to another. Because transdifferentiation normally occurs between cells that arise from the same region of the embryo, understanding the molecular and cellular events in cell type transformations may help to explain the mechanisms underlying normal development. Here we review examples of transdifferentiation in nature focusing on the possible role of cell type switching in metamorphosis and regeneration. We also examine transdifferentiation in mammals in relation to disease and the use of transdifferentiated cells in cellular therapy. Developmental Dynamics 236:3208–3217, 2007. © 2007 Wiley-Liss, Inc.
DEFINITIONS AND CONCEPTS
Transdifferentiation and metaplasia are terms commonly used to denote changes in cellular phenotype. In the broadest sense metaplasia refers to the conversion of one cell type to another. Metaplasia includes conversion of one tissue stem cell to another, in which an undifferentiated cell is able to give rise to more stem cells (and progenitor cells) destined to differentiate toward specific cell types (Tosh and Slack, 2002). Transdifferentiation refers to the conversion of a differentiated cell of one developmental commitment into a differentiated cell of another lineage without first reverting to a more primitive stem cell or progenitor. This idea challenges our long held view that terminally differentiated cells are just that, in their final stage of differentiation, and that this state is fixed and cells are not capable of reversal or change (Diberardino, 1987). As relatively newly described biological phenomena, strict criteria for describing a true transdifferentiation event have been laid down. First, a lineage relationship between the original cell and the “new” cell must be substantiated, and, second, changes in phenotype must be accompanied by molecular changes, such as to allow the two cells before and after transition to be distinguished from one another on a molecular and biochemical level (Okada, 1986; Eguchi and Kodama, 1993). These criteria can be extended to include more rigorous inclusion of modern techniques. Analyses of the phenotype of the two cell types would be greatly facilitated by the use of large scale genomic or proteomic screening. Such screening yielding detailed information would be particularly helpful in elucidating whether the conversion is partial or complete (i.e., do the transdifferentiated cells retain any genes reminiscent of the parent cell?). For analyzing the ancestor–descendant relationship in the future, it is likely that the Cre/lox technique will be applied more often to monitor cell type conversion in vivo as it can be used to irreversibly label a lineage after transient activation of a tissue-specific promoter. However, there are caveats to using this technique; interpretation of the experiments depends on the stringency or leakiness of the promoters that are used.
Differentiation occurs during development and involves the coordinated activation and suppression of genes that give rise to the selective proteins that are specifically expressed in mature cell types (Slack, 2006). Changes in protein expression give rise to the specialized structure and function of the mature cell within the body. Differentiation occurs by means of several stages, beginning with a pluripotent stem cell, through any number of oligopotent or mulitpotent precursors until the mature cell type is observed. Transdifferentiation of mature cells can then be brought about either by (1) somatic mutation, (2) epigenetic changes, or (3) extracellular or environmental factors that mediate changes in gene expression.
Early experiments showed that the transfection of 5-azacytidine–treated mouse fibroblasts with cDNAs from myocyte cDNA libraries can induce transdifferentiation into myoblasts (Davis et al., 1987). cDNAs that induced a muscle differentiation have homology to the myc-similarity regions of MyoD and Myf-5 members of the myogenic regulatory factors (MRF) family of transcription factors later found to be essential for myogenesis (Kitzmann et al., 1998). It is thought that transdifferentiation may occur following changes in expression of such “master switch genes,” which encode transcription factors that induce a cell to differentiate along a particular developmental pathway (Table 1; Li et al., 2005b). During embryogenesis, tissues that develop as neighboring rudiments in a common cell sheet will have similar combinations of transcription factors defining their commitment. Neighboring tissues may differ by the expression of just one or a few transcription factors (Burke et al., 2007). It is likely that, if the two neighboring tissues differ in the expression of one or two transcription factors, then these are prime candidates for the master switch genes, that is, it is the overall combination of transcription factors that are important. Other examples of master switch genes include Pancreatic–duodenal homeobox gene 1 (Pdx-1), a homeodomain containing transcription factor, known to be essential in pancreatic development as Pdx-1 knockout mice are apancreatic (Offield et al., 1996). Overexpression of Pdx-1 in cells from neighboring regions of the endoderm (liver and intestine) induces a pancreatic phenotype and the ability to convert these tissues to pancreatic cells presumably reflects the close developmental relationship of these endodermal tissues (Ferber et al., 2000; Yoshida et al., 2002). Further examples of master switch genes are observed during adipogenesis, which is controlled by the transcription factors PPARγ and C/EBPα. Overexpression of PPARγ and C/EBPα in fibroblasts can induce expression of adipogenic genes such as aP2 and adipsin (Tontonoz et al., 1994). Foxa transcription factors are thought to play a coequal role in liver development (Rosen et al., 2002; Lee et al., 2005). Mouse embryos deficient in both transcription factors Foxa1 and Foxa2 show no liver bud development and expression of the hepatoblast marker α-fetoprotein (AFP) was absent; furthermore, cultures of endoderm derived from these embryos fail to express the liver markers albumin and transthyretin. It is thought that this finding is due to Foxa-mediated modification of chromatin structure during liver development, which normally allows transcription of liver specific genes (Lee et al., 2005).
Table 1. Examples of Naturally Occurring and Experimentally Induced Cell Type Conversions, and the Putative Master Switch Genes Involved
|Wolffian regeneration (iris to lens conversion)||Six3||During lens regeneration in urodele amphibians||(Grogg et al., 2005)|
|Barrett's esophagus (stratified squamous to columnar epithelium)||Cdx2||Possibly induced by bile and stomach acid reflux; predisposes to neoplastic adenocarcinoma||(Spechler, 2002; De Lott et al., 2005)|
|Liver fibrosis (hepatic stellate cells to myofibroblasts)||PPARγ and SRBP1c suppression||Following chronic hepatic tissue damage and inflammation||(Tsukamoto et al., 2006; Bachem et al., 1993)|
|B-lymphocytes to macrophages||C/EBPs and PU.1||Experimental conversion of cells isolated from bone marrow and spleen||(Xie et al., 2004)|
|Hepatopancreatic transdifferentiation (hepatocytes to β-cells)||Pdx1 (+Ngn3, NeuroD, MafA)||Experimental conversion in vivo and in vitro as a putative therapy for diabetes mellitus||(Ber et al., 2003; Kojima et al., 2003; Kaneto et al., 2005a, b; Imai et al., 2005)|
|Pancreas to liver||C/EBPs||Experimental conversion in vivo and in vitro as a potential therapy for liver disease||(Shen et al., 2000)|
TRANSDIFFERENTIATION IN NATURE
Although it may seem unconventional to think so, metaplasias are not necessarily always associated with errors made during development or tissue repair. Although rare, the phenomenon is observed across the metazoan phylogeny. However, it is important to stick to the strict definitions outlined above to identify which of these cell-type conversions embody a true transdifferentiation event. For example, during mammalian development, epithelial cell-type switches from stratified squamous to columnar during esophagus development and from pseudostratified to squamous during the development of the female reproductive system have been long identified (Cunha, 1976; Yu et al., 2005). However, such metaplastic events that occur developmentally will not be further dealt with as examples of transdifferentiation events in this review. In nature, transdifferentiation does occur and occurs under different situations. Transdifferentiation is thought to be a mechanism for tissue repair that excludes the need to resort to undifferentiated reserves of cells such as stem cells and is also used during vast tissue respecification processes associated with metamorphosis and the normal progression through lifecycles.
TRANSDIFFERENTIATION DURING METAMORPHOSIS
Organisms have evolved complex and varied life cycles with intermediate stages that eventually lead to a sexually mature individual capable of reproduction. Sponges begin life as parenchymella larvae that settle down for their sedentary adult lives. During this metamorphosis from larva to adult, the fate of the flagellated cells that function in locomotion of the free-swimming larva has long been debated (Boury-Esnault, 1976; Kaye and Reiswig, 1991). Amano and Hori (1996), used electron microscopy techniques to visualize natural markers of flagellated cells. During metamorphosis, flagellated cells of the larvae appeared to transdifferentiate into the choanocytes that line the internal cavity of the adult by means of an intermediate amoeboid cell stage (Amano and Hori, 1996). Unfortunately detailed lineage analysis was not available, so it is not clear whether this finding represents a true transdifferentiation event or not.
Striated muscle cells of hydrozoan jellyfish may be multipotential with the ability to transdifferentiate into various cell types, including smooth muscle cells and nerve cells positive for the neuropeptide FMRFamide (Alder and Schmid, 1987; Schmid et al., 1988). Such transdifferentiation may be induced experimentally by alteration of cell–extracellular matrix (ECM) interactions, protein kinase C activation, and dihydrocytochalasin B–mediated cytoskeletal actin disassembly (reviewed in Schmid and Reber-Muller, 1995). However, transdifferentiation is also something more akin to the normal life cycle of some other hydrozoans for which life begins as sessile colonial polyps or hydroids that go on to produce solitary medusae by asexual reproduction. The free-living medusae undergo a growth phase leading to sexual maturity, liberate gametes, and succumb to organismic death by cell disintegration (Boero and Bouillon, 1993). It was hypothesized that the onset of sexual maturity was the point of no return in the ontogenic sequence of any living organism (Sterns, 1992). However, the hydrozoan Turritopsis nutricula is probably capable of ontogeny reversal simply by turning back into the hydroid form, even after having reached sexual maturity (Piraino et al., 1996). During this process of ontogeny reversal, transdifferentiation presumably plays a vital role in generating the new cell types of the hydroid form, from existing cells of the medusa. The exumbrella epidermis and the cells lining the radial canals (which are virtually devoid of any dedifferentiated cells or interstitial cells) give rise to the chitinous perisarc-secreting epithelium and transform into the hydroid endoderm, respectively. It is also possible that the other missing cell types, including sensory cells and myoepithelial cells could be derived by the transdifferentiation of existing cells (Piraino et al., 1996). By doing so, transdifferentiation could enable Turritopsis to evade death and attain potential immortality by a mechanism unparalleled by any other in the animal kingdom.
Tissue respecification is more common place during metamorphosis of arthropods. It is thought that transdifferentiation could partly be conducive to such transformations, particularly in cases where no apparent mitosis or cell death occur. Some classic studies have identified transdifferentiation, or dedifferentiation and redifferentiation of cells as it was defined then, as characteristic of normal arthropod development (Wigglesworth, 1940; David and Burnett, 1964; Hakim, 1976). Although they were convincing observations, they lack the evidence to fulfill the modern criteria that characterize a complete transdifferentiation event.
TRANSDIFFERENTIATION DURING REGENERATION
Many organisms are capable of restoring by regrowth, parts of their body that have been lost to injury or by autotomy. The origin of the cells that are involved in regeneration is of some debate. It is possible that regenerates arise by transdifferentiation of existing cells. It is equally possible that a reserve of undifferentiated cells, or in other words adult stem cells exist, which, when subjected to the right signals can be activated to give rise to a whole array of cell types (Brockes, 1998; Slack, 2003).
Remarkable regenerative capacity and histogenic plasticity are distinguishing features of the echinoderm phylum (Vickery and McClintock, 1998; Dolmatov, 1999). Holothurians or sea cucumbers use evisceration or the ejection of their gut as a means of evading predators, whereupon the whole gut has to regenerate from rudiments left behind. Evisceration can be induced under laboratory conditions by the injection of distilled water into the coelomic cavity of Eupentacta fraudatrix (Garcia-Arraras and Greenberg, 2001; Mashanov et al., 2005). Transdifferentiation may be used in the regeneration process. For instance, the endodermally derived differentiated cloacal epithelium generates the posterior luminal epithelium, while the mesodermally derived peritoneal and myoepithelial cells of the mesothelium could give rise to the anterior luminal epithelium of the gut, thus remarkably even crossing boundaries of the germ layer between mesoderm and endoderm (Mashanov et al., 2005). Despite the different origins of the cells, the anterior and posterior luminal epithelium are indiscriminate and indistinguishable once regeneration is fully complete (Mashanov et al., 2005).
In the examples discussed so far, it is usually difficult to establish a clear lineage relationship between the transdifferentiated cells and their origins, or delineate the molecular changes underlying the phenotypic conversions, due to poorly understood molecular genetics of these organisms. However, transdifferentiation events have also been observed in some more commonly used model organisms for regeneration studies. In the process of Wolffian lens regeneration in urodele amphibia, cells of the dorsal iris proliferate, undergo depigmentation, and redifferentiate into keratinocyte-like crystallin containing cells of the lens (Yamada and McDevitt, 1984; Tsonis et al., 2004). It is also possible to redirect the neuronal retina and cells of the outer cornea into a lens epithelium fate in this manner (Opas et al., 2001; Opas and Dziak, 1998). The homeodomain transcription factor Six3 may be the master switch gene in the iris–lens conversion process. Transfection of Six3 accompanied by inhibition of bone morphogenetic protein signaling can induce the transdifferentiation of the ventral iris (which does not normally take part in lens regeneration) into lens tissue (Grogg et al., 2005). Transdifferentiation has also been suggested to play a role in amphibian limb and tail regeneration, where epidermal, neuronal, and muscle tissue are replaced after amputation (Brockes and Kumar, 2002). Understanding the mechanisms underlying such regeneration processes in nature will cut new paths in combating degenerative diseases.
TRANSDIFFERENTIATION IN MAMMALS
Transdifferentiation is not a basic, primordial attribute of only a few vertebrates. Although less well understood, it has also been observed in mammals. Tissues that constantly renew throughout life such as hair follicles, the gut, and blood have an adequate reservoir of dedicated adult stem cells. However, many organs are not believed to maintain a resident undifferentiated stem cell population and either has to rely on the proliferation or transdifferentiation of existing differentiated cells to replenish the cell mass after injury. The lung for example, is an organ exposed to the external environment and, therefore, to pathogens and toxicants. Hence, the lung epithelium needs to respond rapidly and effectively to cell damage repeatedly caused by these agents. During repair and maintenance of the lung epithelium, in addition to mobilizing stem cells and promoting the proliferation of differentiated cells, transdifferentiation is also thought to occur (Rawlins and Hogan, 2006). After damage to the bronchiolar epithelium, ciliated cells respond by spreading and undergoing a sequential squamous to cuboidal to columnar transition and eventually transdifferentiate into distinct epithelial cell types to repair the damaged airway epithelium. During the process, expression of several key regulatory transcription factors characteristic of the developing lung, including β-catenin, Foxa2, Foxj1, and Sox family members were reactivated (Costa et al., 2001; Rawlins and Hogan, 2006). In the alveolar epithelium on the other hand, type I pneumocytes (flattened cells that participate in gaseous exchange) have been known to be restored by type II pneumocytes (thicker cells containing lamellar inclusion bodies and secreting surfactants) during development and also postnatal tissue repair (Flecknoe et al., 2000).
In the medical literature, the term conjunctival transdifferentiation is often used to refer to the ingrowth of the conjunctival epithelium onto the denuded cornea. Adjacent parts of the conjunctiva lose their goblet cell properties and become cornea-like during the process of corneal wound healing. This mechanism of corneal wound healing is only used when the wound extends beyond limbal structures called the Palisades of Vogt, where the limbal stem cells that maintain the cornea epithelium reside (Shapiro et al., 1981; Dua et al., 1990). Much controversy remains regarding the classification of this conjunctival–corneal transition as a transdifferentiation, because unlike a true transdifferentiation, conjunctiva-derived cornea do not have all the corneal biochemical and morphological properties and, furthermore, the process is readily reversed by vascularization (Sangwan, 2001). Therefore, “conjunctival transdifferentiation” is more likely an environmental modulation of a phenotype rather than a complete and irreversible transdifferentiation event.
During pancreatic regeneration in the Vervet monkey, foci of hepatocytes are found in the pancreas (Wolfe-Coote et al., 1996). It is likely that the ectopic hepatocytes could have arisen from the transdifferentiation of the exocrine pancreas as has previously been observed in other experimental systems such as in copper-depleted rats (Rao et al., 1986; Wolfe-Coote et al., 1996). Nonetheless whether transdifferentiation plays a hallmark role in normal mammalian pancreas regeneration, is open to debate and in need of further investigation. Similarly, the reverse transformation or foci of acini and pancreatic ductules in the human liver have been found; but is associated with severe posthepatitic cirrhosis, that is, it is not a normal phenomenon (Wolf et al., 1990).
TRANSDIFFERENTIATION AND DISEASE
There are numerous examples of metaplasia in human histopathology and some are more clinically significant than others in that they predispose to the development of neoplasia. Metaplasias are implicated in the development of various cancers, including lung, prostate, vaginal, and breast cancer (Auerbach et al., 1961; Nelson et al., 2002; Bjerkvig et al., 2005; Quinlan et al., 2007). The term transdifferentiation is sometimes used to refer to the origination of the elusive cancer stem cell (a cell that has the ability to renew itself and initiate a tumor) from a differentiated cell, which no doubt is a metaplasia but not necessarily a transdifferentiation. Therefore, the etiology of only a handful of these pathological conditions can be linked distinctly to transdifferentiation events from one differentiated cell type to another.
A clinical example of cell type switching is the premalignant condition of Barrett's esophagus or Barrett's metaplasia. Barrett's metaplasia is a known precursor to esophageal adenocarcinoma (Hameeteman et al., 1989). A patient is said to have developed Barrett's when the normal squamous lining of the esophagus is converted to an intestinal type columnar epithelium normally at the junction of the esophagus and stomach (Barrett, 1957). Much research suggests that the metaplasia is induced by chronic acid reflux from the stomach into the esophagus (Spechler, 2002). The master switch gene responsible for inducing Barrett's metaplasia is probably the Cdx2 gene, a member of the parahox cluster (Ferrier et al., 2005). Cdx2 is involved in intestinal epithelial differentiation and distinguishes the upper and lower epithelium of the alimentary canal; furthermore, Cdx2 expression has been found to be up-regulated in adenocarcinomas of the intestine (De Lott et al., 2005). Experiments have shown that ectopic expression of Cdx2 can induce intestinal metaplasia in the stomach (Li et al., 2005b). The exact mechanism by which metaplasia is induced in Barrett's is still unclear, and some debate remains as to whether Barrett's may be described as a true transdifferentiation event of the epithelium or simply the metaplasia of esophageal stem cells to intestinal stem cells and subsequent differentiation (Jankowski et al., 1999, 2000; Fitzgerald, 2005).
Other clinical examples where transdifferentiation has been implicated include liver fibrosis leading to cirrhosis. The liver has a remarkable ability to regenerate. After damage to hepatocytes, the inflammatory response initiated in consequence first removes the damaged hepatocytes, then repairs the damage and restored liver function by stimulating mitosis of existing hepatocytes. However, chronic tissue damage and the subsequent chronic inflammatory response can often lead to the formation of fibrotic tissue that prevents effective regeneration (Michalopoulos and DeFrances, 2005). The collagen-expressing and α-smooth muscle actin–expressing cells of the fibrotic tissue are predominantly derived by the proliferation of a fibrogenic myofibroblast population within the periportal parenchyma. However, the oxidative stress and cytokines released from Kupffer cells can induce the transdifferentiation of hepatic stellate cells (HSCs) to fibrogenic myofibroblasts, thereby contributing to fibrosis (Tsukamoto et al., 2006). A correlation between HSC to myofibroblast transition and adipocyte–preadipocytic fibroblast dedifferentiation has been suggested. Adipogenic transcription factors such as peroxisome proliferator-activated receptor-γ (PPARγ) and sterol regulatory element binding protein-1c (SRBP1c) maintain HSCs in a quiescent state (Miyahara et al., 2000; She et al., 2005). Wnt-and TNF-α–mediated suppression of PPARγ and SRBP1c adipogenic factors may initiate the HSC to myofibroblast transdifferentiation (Bachem et al., 1993; Tsukamoto et al., 2006). The nuclear receptor transcription factor pregane X receptor (PXR) modulates the expression of many genes necessary for hepatocytes function, particularly detoxification. Rifampicin and other activators of the PXR have been shown to inhibit HSC transdifferentiation and subsequent proliferation of myofibroblasts, thereby being effective as antifibrogenic agents (Wright, 2006).
Similarly, interstitial fibrosis that accompanies renal disease is partly induced by the transdifferentiation of the renal epithelium to myofibroblasts (Hay and Zuk, 1995). In vitro analysis of various renal cell lines have revealed the profibrotic actions of transforming growth factor-β (TGF-β). These observations suggest that TGF-β inhibition, along with type IV collagenase inactivation and basement membrane stabilization, could inhibit epithelial to myofibroblast transdifferentiation (Stahl and Felsen, 2001). Such findings could prove vital in the development of an efficacious therapy for renal fibrosis.
Transdifferentiation may also underlie some of the diseases found in animals but would undoubtedly have implications to human clinical conditions. For instance, in the adult mouse lung only a few mucous-secreting goblet cells are present outside the submucosal glands. But in mice exposed to aerosolized allergens, goblet cells are found lining the trachea and other large airways in abundance; this finding is a condition referred to as mucous metaplasia (Williams et al., 2006). Based on electron microscopy studies, the ectopic goblet cells have been shown to contain cilia, suggesting they could arise from a direct transdifferentiation of ciliated cells or Clara cells of the pulmonary epithelium (Hayashi et al., 2004; Tyner et al., 2006). However, genetic lineage tracing studies are necessary if only to convincingly exclude the possibility that they might also arise from epithelial progenitor cells.
A much celebrated example for which lineage analysis is available is the stepwise conversion of B-cells into macrophages. The CCAAT/Enhancer binding protein family transcription factors C/EBPα and C/EBPβ can act synergistically with the transcription factor PU.1 in converting isolated B-lymphocytes into macrophages (Xie et al., 2004). This conversion is accompanied by a loss of the B-cell specific marker CD19 and the induction of the macrophage marker MAC1. The presence of rearranged immunoglobulin genes in the reprogrammed macrophages indicates their origin from cells that have already undergone normal B-cell differentiation. Furthermore, in transgenic mice where Cre is driven off the CD19 promoter (which permanently labels B-cells with the eYFP reporter), the origin of the macrophages from B-cells can be followed in real time (Xie et al., 2004). Although this is an induced transdifferentiation rather than a naturally occurring one, such studies facilitate our understanding of the plasticity of mature cell types and could have implications in unraveling the pathogenesis of diseases such as leukemia.
TRANSDIFFERENTIATION AS A CELLULAR THERAPY
Knowledge obtained from studying transdifferentiation as a result of a natural phenomenon or as part of a disease is important for several reasons. Studying transdifferentiation will help answer questions regarding normal development, regeneration, and the etiology of some diseases. Furthermore, it will enable us to devise potential therapeutic modalities for various degenerative diseases, where in addition to alleviating the symptoms may also make it possible to permanently replace the damaged tissue. Inducing transdifferentiation of differentiated cells for such therapies, holds an advantage over using stem cell technology, as the utilization of stem cells in cellular therapy is often hindered by the ethical controversies and oncogenic concerns centered around the use of embryonic stem cells and the practical difficulties of identifying and isolating adult stem cells (Samson and Chan, 2006).
THERAPY FOR LIVER DISEASE
The term liver disease covers a wide range of conditions, including viral infections such as hepatitis A, B, and C, Wilson's disease, and liver disease caused by the excessive consumption of alcohol. Left untreated, liver disease produces scaring of the liver, fibrosis, or cirrhosis, and the secondary symptoms that arise from these such as portal hypertension and hepatic encephalopathy (Gressner et al., 2007). Current treatment for liver disease is based on treating the initial cause of the problem, but acute, long-term damage is thought to be irreversible. Recent developments have been made using artificial extracorporeal liver support for treatment of acute liver failure (Krisper et al., 2005). Based on kidney dialysis, this treatment, however, only represents a short-term bridge until transplantation or regeneration (in less severe cases) could occur. The clear problem with a treatment involving transplantation is the current lack of available donor tissue, despite that liver can be obtained from living donors (Ridgway et al., 2005). Rejection is another drawback of transplantation, and immunosuppressant drugs required to combat this rejection have their own side effects (Rolles et al., 1999).
Sources of hepatocyte-like cells for transplantation include salivary gland cells (Hui et al. 2001). SGP-1 cells derived from duct-ligated submandibular salivary glands can be induced to form hepatic (AFP and/or albumin positive) clusters when cultured on type-1 collagen (Okumura et al., 2003). Similarly, current work on transdifferentiation of pancreatic cells to hepatocytes (detailed below) raises the possibility that liver disease could be treated using transdifferentiated cells from the patients own pancreas. Expanding and transdifferentiating the cells in vitro, and transplanting them back into the patient reduces the risk of rejection and eliminates the need for a donor.
Pancreatic cells may have the potential to transdifferentiate to hepatocyte-like cells under a variety of experimental and pathological conditions. The ability to convert pancreatic cells to hepatocytes is probably due to the close developmental relationship between the two tissues (see below). Therefore, the number of genes that are expressed differentially between the two will be fewer in number and thus transdifferentiation may be achieved more easily than by using more distantly differentiated cell types (Hui et al., 2001). Pancreas and liver develop from a bipotential precursor population in the anterior endoderm of the developing embryo (Deutsch et al., 2001). It is thought that the anterior endoderm is destined to become pancreas. However, fibroblast growth factor signaling induces expression of Sonic Hedgehog, which inhibits pancreas development but not liver (Deutsch et al., 2001). Thus, the developmental pathways of liver and pancreas are similar; potentially similar enough that transdifferentiation may be induced between the two tissue types using ectopic expression of one or a few master switch genes.
The production of hepatocytes in the pancreas has been observed in vivo in rats treated with a copper-deficient diet (Rao et al., 1986). After 8–10 weeks of a copper-free diet, multiple hepatic foci were observed in the pancreas. These ectopic hepatocytes are negative for pancreatic enzymes and hormones, but express liver markers (Reddy et al., 1991). Further in vivo experiments have found that keratinocyte growth factor (KGF) under the control of the insulin promoter, generates hepatocytes in the islets of Langerhans of transgenic mice (Krakowski et al., 1999). Although it is likely that the β-cells produce the hepatic-like cells in the KGF model, it is still not known which pancreatic cells produce hepatocytes in the copper-deficient model.
The transdifferentiation of the pancreatic cell line AR42J-B13 (B13) to a hepatic phenotype can be induced following addition of the synthetic glucocorticoid dexamethasone (Dex; Shen et al., 2000). Treatment with Dex caused the down-regulation of the pancreatic marker amylase and induction of liver markers, including transferrin, α 1-antitrypsin, transthyretin, glucose-6-phosphatase, α-fetoprotein, cytokeratin 8, P-glycoprotein, and phenol sulfotransferase (Shen et al., 2000). At least some of the nascent hepatocytes appeared to be derived from pancreatic exocrine cells (Shen et al., 2000). Furthermore, the treatment with Oncostatin M enhances the transdifferentiation toward the hepatic phenotype. Later work demonstrated that not only could the transdifferentiated cells express early liver markers but also expressed genes involved with Phase I and II detoxification pathways, ammonia detoxification, and glucose metabolism, indicating that the transdifferentiated cells were expressing genes associated with a mature phenotype after treatment with Dex and Oncostatin M (Tosh et al., 2002; Wang et al., 2005; Burke et al., 2006).
It has been proposed that the molecular mechanism controlling the transdifferentiation of the B13 cells to hepatocytes involves the activation of the glucocorticoid receptor by Dex, which activates transcription of a master switch gene, a member of the C/EBP family of transcription factors. The hypothesis that C/EBPβ is an important player in the molecular basis of B13 transdifferentiation to hepatocytes is supported by the observation that liver inhibitory protein (LIP, an isoform of C/EBPβ mRNA that lacks a transactivation domain) can inhibit transdifferentiation. LIP presumably acts by heterodimerizing with C/EPBβ, producing an inactive form of the transcription factor (Shen et al., 2000). Interestingly, C/EBP (α and β) are expressed in the early liver rudiment but not in the pancreas (Westmacott et al., 2006), suggesting that C/EBPs may distinguish liver and pancreas during development. A similar up-regulation of C/EBPβ along with α-fetoprotein is seen during the Dex-induced transdifferentiation of primary rat pancreatic exocrine cells into hepatocytes (Lardon et al., 2004).
THERAPY FOR DIABETES MELLITUS
Type I diabetes is the result of an autoimmune disease in which the body's own immune system attacks and destroys pancreatic β-cells, causing a reduction in insulin production (Yamada and Kojima, 2005). Type I diabetes is currently treated with insulin injection. However, insulin therapy only represents a treatment and not a long-term cure for the disease (Samson and Chan, 2006). The ultimate cure would involve long-term replacement of the patient's cells with new β-cells. To achieve this goal, transplantation studies have been performed using either whole organ transplantation or islet transplantation using the Edmonton protocol, in which the islets of the donor are isolated using enzymatic digestion, before infusion into the patient's portal vein (Shapiro et al., 2001; Barrou et al., 2004). However, although these studies were ultimately successful in terms of restoring normoglycemia, they still require the use of immunosuppressant drugs and the 3-year organ survival rates are 70–80% (Shapiro et al., 2001). Furthermore, the source of cells is limited by the number of donors, which is exasperated by the fact that the Edmonton protocol requires cells from up to three donors (Ridgway et al., 2005). The scarcity of pancreatic tissue for transplantation has led to the search for alternative sources of β-cells for transplantation. Investigations have focused on using pancreatic ductal cells as a potential source for β-cell production. Hui and colleagues demonstrated that the human PANC-1 cell line and the rat ARIP cell line can be induced to differentiate into insulin-secreting endocrine cells by exposing them to Glucagon-like peptide 1 (GLP-1). However, this does require the expression of Pdx-1, which is not expressed in the PANC-1 or ARIP cells (Hui et al., 2001). Similarly, functional β-like cells have been generated from cultured adult exocrine pancreatic cells treated with leukemia inhibitory factor (LIF) and epidermal growth factor (EGF; Baeyens et al., 2005). The surrogate β-cell generated in this manner can rescue alloxan-diabetic mice after transplantation. As with many tissue culture systems, it is important to exclude the presence of contaminating progenitor-like cells.
Based on the concept that liver and pancreas arise from the same region of the developing embryo, recent work has focused on transdifferentiating hepatocytes to β-cells. Most approaches have been based on overexpression of Pdx-1, the master gene controlling pancreatic development. A modified form of the Pdx-1 homologue Xlhbox8 carrying the VP16 transcriptional activation domain from Herpes simplex has been used to convert liver to pancreas in transgenic Xenopus tadpoles. The Xlhbox8-VP16 construct was driven off the transthyretin (TTR) promoter to direct expression in the liver and contained a green fluorescent protein tag under the pancreatic elastase promoter (Beck and Slack, 1999; Horb et al., 2003). Transient expression of Xlhbox8-VP16 may induce formation of ectopic whole pancreas, in an otherwise normal animal, giving rise to both endocrine and exocrine cell types (Horb et al., 2003). It is thought that the appearance of ectopic pancreas in this case is the result of transdifferentiation of at least partially differentiated liver cells rather than simply changing the developmental fate of undifferentiated endodermal cells (Horb et al., 2003). This investigation also discovered that the same construct could also induce similar effects in the human HepG2 cell line (Li et al., 2005a).
Pdx-1 can also induce the expression of pancreatic genes in many other cell types, including fetal liver progenitor cells and in a liver stem line—the WB cells (Zalzman et al., 2003; Tang et al., 2006). In vivo transformation of hepatocytes into insulin-producing cells has been successfully achieved by the intravenous injection of adenoviruses encoding Pdx-1 (with and without VP16) and/or various other key regulatory pancreatic transcription factors, including NeuroD, Neurogenin3, MafA, and BetaA2 (Ber et al., 2003; Kojima et al., 2003; Kaneto et al., 2005a, b; Imai et al., 2005). The β-like cells produced from liver cells were able to rescue streptozotocin-induced diabetes in mice. Likewise, adult rat and human hepatocytes have been transformed in vitro and subsequently transplanted into diabetic mice, correcting hyperglycemia for various lengths of time (Sapir et al., 2005). Compared with the in vivo approaches, in vitro transformation of cells has the added advantage of overcoming the narrow tropism of the viruses, especially for organs such as the liver. It would also reduce vector toxicity by minimizing systemic infections. When Pdx-1 is directly introduced to the livers of rodents, it can trigger the development of pancreatic exocrine cells secreting digestive enzymes such as amylase, trypsin, chymotrypsin, and lipase (Kojima et al., 2003). Therefore, in vitro transformation of cells would also provide a greater quality control over the cells that can be selected for transplantation.
Transdifferentiation has attracted much attention in recent years mainly with respect to the potential to generate cells for therapeutic transplantation. At present, it is not clear whether transdifferentiated cells will fulfill their clinical potential. Harnessing the ability to transdifferentiate cells will complement the use of embryonic stem cells as therapeutic modalities. In addition, the change in phenotype associated with certain diseases may also be useful in the early diagnosis (e.g., diagnostic markers) or in treatment of the disease itself (reversing the transformation prior to the development of neoplasia). However, apart from the therapeutic importance, examples of transdifferentiation are remarkable because they ultimately demonstrate that the differentiated cells are not “fixed” but rather retain the ability to undergo genetic reprogramming.
We thank the Universities UK Overseas Research Studentship Committee, the Medical Research Council, Wellcome Trust, and the Department of Biology and Biochemistry for financial support.