One of the earlier attempts at transdifferentiating liver cells into pancreatic beta cells was reported by Ferber and colleagues in 2000. Using a “gain-of-function” strategy, they transferred a constitutively active Pdx1 cassette to recipient mice by means of an adenoviral vehicle . Ectopic expression of the gene was mainly observed in the liver, where it activated the expression of the endogenous genes Insulin 1 and 2 and prohormone convertase 1/3 (PC 1/3). These genes are typically active in beta cells, but not in liver tissue. Plasmatic insulin levels were substantially elevated in treated mice compared with controls treated with an empty virus alone. More strikingly, ectopic insulin expression was found to reduce glucose levels in streptozotocin-treated mice. In a series of follow-up experiments, the same team reported that ectopic Pdx1 expression in the liver persisted well beyond the few weeks during which adenoviruses (which do not integrate into the host genome) maintain their activity. Indeed, they found that the exogenous Pdx1 was able to induce the activation of its endogenous counterpart, thereby priming self-sustainable regulatory networks leading to long-standing “transdifferentiation,” as evidenced by ectopic insulin production and maintenance of normoglycemia in streptozotocin-treated animals up to 8 months after the initial treatment. Like other master regulators of development and stem cell self-renewal, Pdx1 exerts a positive feedback over its own promoter. Since liver cells already express Pdx1 transcriptional partners such as HNF1beta and 3beta, the authors speculated that the transient administration of the exogenous genes resulted in permanent effects. It was also observed that the insulin-producing cells were located mainly around the hepatic central veins, a distribution that was hypothesized to allow systemic hormone release without harmful effects in liver function.
The molecular basis of this phenomenon remained unclear until very recently, when Meivar-Levy et al. presented compelling evidence that Pdx1-mediated transdifferentiation requires an intermediate “de-differentiation” step. Pdx1, but not other pancreatic genes delivered using the same system, induced a substantial down-regulation of C/EBPbeta and LAP, two redundant proteins of a family of transcription factors known to play important roles during liver embryonic development as well as in adult hepatocytes. Overexpression of LAP in primary cultures of human hepatic cells prevented the Pdx1-mediated de-differentiation and activation of the pancreatic program. However, down-regulation of C/ EBPbeta was insufficient by itself to trigger the activation of pancreatic genes, with the exception of Ngn3. As expected, the simultaneous down-regulation of C/ EBPbeta and the administration of Pdx1 had a synergistic effect in inducing transdifferentiation.
Liver transdifferentiation approach as reported by Ferber et al. (2000) . A Pdx1 cassette is delivered systemically through an adenoviral vector. Expression of this cassette in the liver yields insulin-producing cells. When the procedure is done in streptozotocin (STZ)-treated diabetic mice, it results in a very significant reduction of blood glucose levels.
The question remains of whether this approach is targeting bona fide hepatocytes or perhaps more undifferentiated progenitors that might be more amenable to transdifferentiate. In vitro experiments are not conclusive because hepatocyte cultures become fibroblastic in appearance very rapidly, perhaps because of the adaptation of the cells to attachment in plastic. In addition, it is very likely that the insulinproducing cells generated in this manner are not true beta cells, but rather hybrids between hepatic and pancreatic cells.
Parallel experiments conducted by Horb and colleagues confirmed that transdifferentiation is indeed feasible in other contexts. In short, the authors created transgenic frogs where a Pdx1 (Xlhbox8)-VP16 fusion cassette is expressed under the control of the liver-specific promoter transthyretin (TTR) . The rationale for the use of VP16 – a potent transcriptional transactivator from the herpes simplex virus – is that nonpancreatic cells may lack the appropriate molecular partners for Pdx1 to exert its biological function. An additional marker was added to screen for successful transdifferentiation, namely the green fluorescent protein (GFP) under the control of the elastase promoter, a pancreas-specific regulatory element. Up to 60% of transgenic tadpoles showed partial or total conversion of liver to pancreas, as evidenced not only by the expression of GFP but also by that of pancreatic endocrine (insulin and glucagon) and exocrine (amylase) markers. It is important to note that no transdifferentiation was observed when Pdx1, without VP16, was used. This observation suggests that the assertion that Pdx1 is necessary, but not sufficient to promote pancreatic differentiation, remains true for the liver.
Experimental conversion of liver to pancreas in a frog transgenic model, as described by Horb et al. The Xlhbox8 gene, the amphibian homolog of Pdx1, was fused to the VP16 transactivation domain, and placed under the control of the liver-specific promoter TTR. An additional marker was added to follow up successful conversion events (GFP driven by the promoter of elastase, a gene expressed in the pancreas but not the liver). Transgenic tadpoles exhibited various degrees of liver transdifferentiation into pancreas.
Given that the timing was such that the onset of the ectopic Pdx1 expression was coincident in time with the initiation of liver development, this event could in theory be considered an induced redirecting of early organogenesis, rather than a proper transdifferentiation event. The authors, however, used the same construct to transfect immortalized human hepatocytes (HepG2), which led to elastase activation in approximately 65% of the cells that received it. Of these, approximately 15% were insulin positive. These results were subsequently expanded by characterizing the transdifferentiated cells. Confirming the observations of Ferber and colleagues, it was found that the hepatic phenotype was lost upon ectopic expression of Pdx1; that the requirement for the transgene was not permanent, as an initial trigger was sufficient to activate the pancreatic lineage; and that the insulin-positive cells obtained through this approach had PC 1/3, C-peptide, and glucagon-like receptor 1 (GLP-1), among other functional markers of true beta cells. These cells were glucose responsive and increased insulin expression upon treatment with GLP-1 and beta-cellulin.
Using a lentiviral vector, Tang et al. were able to transdifferentiate rat hepatic stem-like WB cells into pancreatic beta cells both with Pdx1 and Pdx1–VP16. This study was the first to systematically compare the transdifferentiation potential of the two versions of the gene. While they found that cell lines expressing either Pdx1 or Pdx1–VP16 long-term had comparable gene expression profiles as well as a similar capacity to correct hyperglycemia in recipient diabetic mice, short-term expression gave a marked edge to the VP16-fused version. Additional studies were significantly consistent with the first set of data first published by Ferber and colleagues, and showed the cumulative effect of adding other pancreatic endocrine factors to the mix, such as Ngn3, NeuroD, or MafA. Interestingly, Wang et al. reported their inability to induce liver-to-pancreas transdifferentiation in vivo when using adeno-associated viruses as vectors to co-deliver Pdx1 and Ngn3. However, when they delivered these cassettes using plasmid vectors with an irrelevant adenoviral vector, they reported correction of hyperglycemia in diabetic rodents. The authors postulated that the antigen-dependent immune response elicited by the adenoviral capsid (but not other viruses) was instrumental in the induction of transdifferentiation.
The concept of liver transdifferentiation has attracted significant attention for several reasons. First, liver cells are easier to obtain and expand than those derived from the pancreas. Therefore, they provide an easily accessible source (a biopsy might provide enough cells to manipulate ex vivo) that could be extracted from the very same patients we want to treat, thus eliminating the risk of allogeneic rejection . Second, it is becoming clear that, despite some degree of functionality, transdifferentiated liver cells are not true beta cells. Some evidence indicates that the ability of these cells to appropriately regulate insulin secretion in a glucoseregulated manner might not stand comparison with that of true beta cells. However, clinical therapies could be devised even if these cells worked just as a “pump” (i.e., continuously secreting a basal amount of insulin in a nonregulated fashion). Most importantly, in type I diabetic patients, the immune system is poised to attack and destroy any cell that resembles a beta cell. From this perspective, these “hybrids” might have a selective advantage over native beta cells, because they could be able to elude the autoimmune response. Of course, this hypothesis hinges on the assumption that the autoimmune response will spare non-beta cells that express insulin, which might not be the case in view of the fact that insulin has been shown to be an auto-antibody in type I diabetes.
A more recent attempt at transdifferentiating non-endocrine tissue into beta cells used a different starting material, one that – at least in theory – should be more closely related to the desired end product. Based on the screening of at least 20 transcription factors (of which nine gave rise to gross phenotypic changes in beta cells when knocked out) expressed either in terminally differentiated beta cells or their progenitors Zhou et al. were able to reprogram pancreatic exocrine tissue into islet cell types using a combination of three genes (Pdx1, Ngn3, and MafA) delivered by means of adenoviral vehicles. New insulin-producing cells were detected as early as 3 days after the injection of the adenoviral mix into the pancreata of Rag1 −/− mice, a strain typically used to minimize the occurrence of viral-induced immune responses such as those described earlier by Wang et al. The number of these cells kept expanding for up to 3 months, long after the adenoviruses had been cleared from the recipients. They were indistinguishable from native beta cells in terms of size, morphology, presence, and distribution of insulin granules and molecular markers. Unlike in other transdifferentiation settings, the original exocrine phenotype appeared to be completely abrogated (i.e., they were not “hybrids”). Diabetic mice subjected to the treatment showed a significant and permanent improvement in blood glucose levels, even if diabetes was not completely reversed. The latter observation could be explained by the fact that the newly created beta cells remained isolated and did not cluster to form islets. Indeed, beta cell communication is essential to stimulate glucose-mediated insulin secretion.
As promising as this rapidly evolving field is, safety concerns may still preclude its immediate clinical translation. The observation that the ectopic genes need only be expressed transiently in order to activate transdifferentiation is encouraging. However, the use of adenoviruses may have serious side effects by eliciting immune responses in the host. Also, the ability of Pdx1 to induce exocrine tissue as well as endocrine derivatives proved harmful by inducing fulminant hepatitis in animal models. In a transgenic setting, ectopic expression of the gene resulted in widespread liver dysmorphogenesis, with abnormal lobe structures and polycystic lesions.Certainly, the systemic infusion of viral vectors containing master pancreatic regulators into human patients does not seem a clinical possibility in the near future. However, the extraction of liver tissue for ex vivo transdifferentiation and subsequent reimplantation in the patient appears to be a more reasonable course of action. It is likely that in this setting the adenoviruses would have already been cleared from the cells at the time of transplantation, increasing the overall safety of the procedure.
Tags: amphibian homolog of Pdx1, down-regulation of C/EBPbeta and LAP, Ectopic expression of the gene, hepatic central veins, liver tissue, Liver transdifferentiation, Pdx1-mediated transdifferentiation, Plasmatic insulin levels, prohormone convertase 1/3, streptozotocin-treated mice, transdifferentiating liver cells, Xlhbox8 gene