Signal-Driven Approaches

A natural first attempt at directing the differentiation of huES cells toward beta cells was to expose them to the natural milieu of signals that drive the process in vivo. Given the practical difficulties of working with human fetal tissue, Brolen et al. transplanted spontaneously differentiated huES cells containing subpopulations of pancreatic (Pdx1 + /Foxa2 + ) and endocrine (Pdx1 + /Isl1 + ) together with murine fetal pancreatic explants under the kidney capsule of recipient immunocompromised mice. Beta cell-like clusters expressing not only insulin, but also many transcription factors known to participate in the maintenance of the beta cell phenotype, could be consistently observed in the experimental group, but not in control animals that were transplanted with the huES and nonpancreatic tissues, such as the liver or the telencephalon of mouse embryos. The results later reported by Vaca et al. in an in vitro setting are consistent with these observations. As for purely in vitro approaches, the adaptation of the protocol of Lumelsky et al., based on the generation and expansion of intermediate-stage nestinpositive cells, was another logical step. It was first reported by Baharvand et al., who were able to observe glucose-mediated insulin release in vitro, but failed to detect insulin secretory granules. As is the case with mouse ES cells, even the cells that produce bona fide insulin as a result of nestin-based differentiation protocols from huES cells are more related to neuroectoderm than to endodermderived beta cells.

The first reports on “canonical” ES cell differentiation into beta cells came shortly after the description of the conditions for definitive endoderm differentiation in murine ES cells.Using a method based on the administration of Activin A (an analog of Nodal) in the presence of a very low percentage of serum, D’Amour and colleagues obtained cultures consisting of up to 80% of cells with definitive endoderm characteristics (Sox17, Goosecoid, Foxa2, and Mixl1 expression). These cells could be further enriched by sorting for CXCR4, a chemokine receptor expressed in mesoderm and definitive endoderm, but not in primitive/visceral endoderm. Transplantation of these cells into immunodeficient mice resulted in further progression of the endoderm differentiation program, as evidenced by expression of intestinal and liver markers in histological sections of grafts. Arguably, the most critical contribution of this study to the field was the finding that Activin A was most effective at low serum concentrations. Serum-borne factors are known to interfere with specific differentiation pathways, including that leading to the generation of definitive endoderm. The same team of investigators contributed a breakthrough follow-up study shortly after the publication of the first one. Capitalizing on their ability to efficiently differentiate huES cells toward definitive endoderm, they applied FGF7 (a powerful inducer of Notch signaling during pancreatic development ) and cyclopamine (an inhibitor of Shh signaling ) to obtain cells that exhibited features similar to those of the primitive gut tube (chiefly expression of HNF-1beta and HNF-4A). Subsequent addition of retinoic acid (which had been previously found by others to help both the patterning of endoderm and the progression of pancreatic development ) in defined medium resulted in the appearance of Pdx1- positive cells with a expression profile consistent with that of posterior foregut, first, and pancreatic endoderm/endocrine progenitors later. A final maturation step led to the formation of pancreatic endocrine cells with a messenger RNA pattern comparable to that of human adult islets . All of the endocrine cell types found in adult islets were represented in these cultures. Given their importance for prospective cell therapies for diabetes, the authors focused on the characterization of putative beta cells, which constituted approximately 7% of the total population. These cells had very high insulin content, almost equivalent to that of adult islets and much more elevated than in earlier protocols.They had well-defined secretory granules and were shown to secrete insulin/C-peptide in response to a variety of secretagogues, but not glucose. An explanation for the latter observation was that these beta-like cells might not be completely mature.

Another multistep protocol described around the same time showed some degree of glucose-responsiveness, but comparable differentiation efficiency. Shortly thereafter, the authors of the original study completed an unparalleled triad of breakthrough reports. Seemingly accepting defeat at obtaining therapeutic yields in vitro, they decided to transplant huES cell derivatives into immunodeficient mice midway throughout the protocol, in the hope that they would complete their maturation in vivo. The procedure was successful at preventing the development of diabetes in streptozotocin-treated recipients, but the implanted cells took several months to mature upon implantation, and teratogenic lesions were observed in at least 15% of the animals.

The entire field is presently at a crossroads: is there anything else we can do to significantly increase the efficiency of full differentiation in vitro, or should we abandon such efforts and focus instead on making the transplantation of immature progenitors safer and faster acting? Only time will tell, but judging from the almost frenetic pace of research on the subject, a definitive answer might be just around the corner.

Genetic Manipulation

Following the demonstration that constitutive Pax4 expression improves the outcome of pancreatic differentiation in mouse ES cells, Liew et al. selected huES cell clones stably transfected with Pax4. Late-stage EBs generated from these cells were plated on Matrigel ™ and treated with low glucose and nicotinamide, a set of conditions previously found to enhance the differentiation of beta cell progenitors in vitro. They yielded a higher number of cells that stained positive for Newport Green (NG), a zinc-fluorescent probe commonly used to label and sort beta cells.

FACS-enriched NG-positive cells had higher levels of Insulin, C-peptide, and Pdx1 expression than those derived from their nontransfected counterparts. However, they were irresponsive to glucose stimulation. In another recent report, Lavon et al. found that constitutive expression of Foxa2 in huES cells did not significantly alter the pattern of pancreatic specification in spontaneously differentiating EBs. Pdx1 overexpression, in contrast, resulted in an overall acceleration in the onset of the downstream gene Isl1 and the up-regulation of downstream genes such as Ngn3 and Pax4. However, insulin-expressing cells could not be detected unless the cells were allowed to further differentiate in vivo teratomas, and the differences between the genetically modified cells and the wild-type controls were statistically insignificant.

Protein Transduction

The only example thus far of the use of this novel approach to aid in the pancreatic differentiation of huES cells was reported only recently. The authors subjected huES cell-derived EBs to TAT-Pdx1 treatment for 7 days, using Pdx1 protein as a control. Despite the well-known ability of Pdx1 to penetrate cells by virtue of an antennapedia-like protein transduction domain, samples treated with TATPdx1 showed a very significant up-regulation of Pdx1 targets compared with the controls. These included its own endogenous Pdx1 counterpart ( ~ 20-fold), Insulin ( ~ 30-fold), or islet amyloid polypeptide ( ~ 12-fold). C-peptide could be detected by immunofluorescence, but GLUT-2 levels remained unchanged when compared with those of samples treated with Pdx1 protein.

A caveat of these studies – which could explain the partial effects observed – is that Pdx1 was applied at a very early differentiation stage, prior to the initial specification of primitive gut/posterior foregut cells. At this point, in the absence of the molecular partners that would normally act in concert with Pdx1 during pancreatic specification the protein alone might be insufficient to fully trigger by itself the onset of such program. It is plausible, however, that the use of TAT-Pdx1 at the appropriate huES cell differentiation stage might improve the yield and function of beta cells in the context of signal-driven approaches. In this context, it would be interesting to test whether TAT-Pdx1-VP16, a recently described version of transducible Pdx1 that includes the VP16 transactivation domain would improve ES cell differentiation efficiency.

iPS Cells

With the cloning of Dolly the sheep and the advent of SCNT the possibility of tailoring ES cells to each patient was immediately acknowledged. The nuclei of easily harvested somatic cells from any given donor would undergo a reprogramming process by microinjection into enucleated oocytes. The activation of the reconstituted embryo would lead to the in vitro development of blastocysts from which ES cells – genetically identical to the donor – would be harvested. The principle of “therapeutic cloning” was readily tested in several species, including nonhuman primates but, despite initial reports suggesting otherwise huES cells could never be derived from SCNT-derived embryos. Alternatives to SCNT included the use of cellular extracts or fusion with ES cells but none of these approaches was completely successful. Very recently, however, the group of Takahashi and Yamanaka were able to reprogram murine somatic cells by introducing four critical components of the core ES cell circuitry, namely Oct3/4, Sox2, c-Myc, and Klf4. Similar results with human cells were presented shortly thereafter by two independent groups – the second using Sox2, Oct3/4, Lin28, and Nanog in the reprogramming mixture . The biotechnological feat of reprogramming adult somatic cells in such simple fashion (the top scientific breakthrough of the year 2008 of the journal Science ) might have enormous medical implications, provided that the newly reprogrammed cells are comparable to blastocyst-derived ES cells; and the procedure can be done safely. While all preliminary evidence seems to confirm that these induced pluripotent stem (iPS) cells are functionally indistinguishable from their embryo-derived counterparts, the use of retroviral vectors to deliver the critical genes is still unsafe in the context of human therapies . Ongoing efforts at addressing this concern include the use of nonviral delivery methods and protein transduction. The use of iPS cells for directed differentiation is now a fertile field, as dozens of laboratories around the world attempt to replicate results previously obtained with huES cells.Of special interest is the recently reported proof of principle that iPS cells can undergo pancreatic differentiation using a protocol similar to that used by Jiang et al.

Safety of ES Cell Therapies

The translation of basic ES cell research into clinical therapies remains hindered by a number a safety concerns that arise from their very nature. Indeed, these cells are characterized by an unlimited proliferation potential under the appropriate conditions. If unchecked, even minute percentages of undifferentiated cells could keep dividing upon transplantation in immunosuppressed recipients, yielding teratomas. These tumors develop either from carry-over undifferentiated cells or from cells that de-differentiate upon transplantation. Some groups have approached this problem by screening the number of undifferentiated cells in each transplantable preparation. If such number is below the threshold known to produce teratomas in nude mice, they argue, these preparations should be considered safe for clinical use. However, this method does not take into account the risk of dedifferentiation after transplantation.Other teams have addressed the teratogenic potential of ES cells by integrating suicide genes into them. Such elements sensitize ES cells to specific pro-drugs, which can be used to induce their selective ablation both in vitro and in vivo. However, if teratomas were to form in vivo due to de-differentiation of implanted cells, administration of the pro-drug would kill the entire graft. The risk of accumulation of genomic instabilities as a result of long-term cell culture might be subtler, but not less dangerous. In view of later observations, the first reports on the karyotypic stability of huES cells turned out to be inexact. As it happens with other cells, their adaptation to proliferative conditions in vitro invariably results in the selection of traits – perhaps not surprisingly – similar to those responsible for malignant transformation.These include chromosomal and subchromosomal alterations not always detectable in standard G-banding tests.Rushing ES cell therapies to the clinic would not be advisable until steps are taken to minimize these risks.

Line-to-Line Variability

Despite the overall similarities between different huES cell lines when kept under conditions that help maintain their pluripotency, several groups have reported differences in a number of parameters, ranging from population doubling time to susceptibility to spontaneous differentiation. These differences become even more striking when the cells start to specify. In a comprehensive study of huES cell lines allowed to spontaneously differentiate, Osafune et al. found >100-fold line-to-line variations in the expression of markers specific for the three germ layers. Three cell lines, for example, had a higher propensity to develop mesodermal lineages; two exhibited a marked preference for ectodermal or neural genes; another two showed a tendency to give rise to endodermal cell types; and so on. Out of the latter, line huES turned out to be the best in terms of differentiation potential into pancreatic cells. One potential conclusion from these results is that we ought to look for the right cell line for each differentiation protocol. This would argue in favor of banking an extensive number of ES cell lines and use only the ones that are better suited for each purpose. The alternative view, however, is that the results manifest the need for developing more robust protocols that will work in a cell line-independent fashion. After all, prospective personalized therapies where ES-like (iPS) cells are derived from the patient might not afford the clinician the freedom to choose the most appropriate cell line for the desired developmental outcome.

Spontaneously differentiated ES cells containing endocrine and exocrine progenitors are able to terminally mature in vivo when engrafted together with fetal pancreatic tissue. Controls in which other tissues unrelated to the pancreas were transplanted together with the ES cells did not yield insulin-producing cells.

huES cell differentiation protocol described by D’Amour and colleagues (2006). FCS fetal calf serum; FGF fibroblast growth factor; RA retinoic acid; Cyc cyclopamine; Ex-4 exendin-4 IGF-1 insulin-like growth factor 1; HGF hepatocyte growth factor.

huES cell differentiation protocol reported by Jiang et al. . EGF epithelial growth factor; bFGF basic fibroblast growth factor; IGFII insulin-like growth factor II.

Original protocols for the generation of iPS from adult fibroblasts. Retroviruses were used for the delivery of two different sets of reprogramming genes. Both combinations included the master regulators Oct3/4 and Sox2.

Undifferentiated ES cells can proliferate indefinitely under the appropriate conditions. Upon differentiation and transplantation, undifferentiated escapees may keep proliferating, especially if the recipient is immunosuppressed. These cells will give rise to a tumor termed teratoma . There is evidence that such tumors may also originate by de-differentiation of differentiated cells . Strategies to prevent this undesired outcome may involve the engineering of suicide genes in the genome of the donor cell line ( black nuclei ), which will destroy the host cell in the presence of a pro-drug . Current research focuses on the design of strategies to activate these suicide genes only in the cells that remain undifferentiated, as opposed to the entire graft.

Embryonic stem cells • iPS cells • Nestin • Embryoid bodies • Teratomas

Embryonic stem (ES) cells were first derived from the early mouse blastocyst more than two decades ago.  Under appropriate culture conditions (which at the time included the use of fibroblast feeder layers), these cells could be propagated indefinitely and had the potential to differentiate into derivatives of all three embryonic layers, as evidenced by their ability to both form teratomas in immunocompromised recipients and extensively contribute to the development of the embryo when injected in recipient blastocysts. ES cells exist only transiently in the inner cell mass of the early embryo . Specific culture conditions keep them cycling rapidly ex vivo (doubling time ranges from 24 to 48 h ) without loss of pluripotency for extended periods of time.

The molecular machinery behind the “stemness” of ES cells is very well conserved across species. The genes Oct3/4 , Sox2 , and Nanog lie at the core circuitry that imparts pluripotency and self-renewability.. In fact, at least two of them ( Oct3/4 and Sox2 ) have proven indispensable in the reprogramming of somatic cells into ES-like induced pluripotent stem (iPS) cells.With some exceptions, the down-regulation of these genes is associated with the activation of specific differentiation pathways. In addition to the above gene expression signature, human ES (huES) cells are characterized on the basis of their lack of stage-specific embryonic antigen 1 (SSEA-1) and presence of SSEA-3, SSEA-4, TRA-1-60, TRA-1-81, telomerase, and alkaline phosphatase.

As mentioned in the preface of this work, for decades, ES cells were chiefly considered a tool for the creation of targeted/knockout mice. However, the contribution of ES cells from non-murine species to the germ line could never be demonstrated. The first reports on somatic cell nuclear transfer (SCNT) in 1996 put and end to this search, as it was now possible to generate targeted animals of virtually any species of interest to humans in one single step.

In the meantime, steady progress was being made in refining the culture settings that would allow for the survival of nonhuman primate blastocysts up to the point where isolation of their ES cells was possible.This seminal report was followed shortly afterwards by the first report on the generation of human ES cells, 1 which opened the door to the possibility of devising regenerative therapies by means of expanding ES cells in vitro and then differentiating them into the tissue of interest. In this chapter, we review the attempts that have been made thus far to convert them into pancreatic endocrine cell types of potential use in the treatment of type I diabetes.

human embryonic stem (ES) cell colony grown over an inactivated mouse embryonic fibroblast feeder layer . When grown on Matrigel ™ , human ES cells form a monolayer . A higher magnification picture shows the morphological characteristics of ES cells: polygonal shape, high nucleus:cytoplasm ratio and refractive nucleoli.

A Venn diagram representing the overlap of the three critical regulators of the ES cell phenotype (Oct3/4, Sox2, and Nanog) promoter-bound regions. A large number of sites are co-bound by the three genes.Autoregulatory loops formed by Oct3/4, Sox2, and Nanog.

ES Cells and Gene Targeting

During the last 25 years, the application of gene targeting techniques to ES cells has become a routine procedure to generate genetically modified mice. The availability of large populations of these immortal cells makes it feasible to target specific genes, despite the low frequency of homologous recombination in mammalian cells. Targeted clones can be easily selected in vitro and used to generate chimeric mice by aggregation or injection into blastocysts. If the host blastocyst and the donor ES cells belong to different strains of mice, chimerism can be visually assessed by the mixed coat color of the resulting animals. Typically, up to 70% of injected blastocysts are overtly chimeric. The degree of chimerism varies widely from barely detectable to complete ES coloration. Since ES cells retain the potential to contribute to all embryonic lineages, some of them may partially colonize the germ line. Because of the fine-grained nature of ES cell chimerism, the germ line is usually a mixed population of donor- and host-derived cells. In order to obtain the highest possible number of ES-derived gametes, ES cells for blastocyst injection are of male genotype. The introduction of male ES cells into female host blastocysts normally results in the generation of fertile intersex animals, which only transmit the ES genotype.This is a consequence of the expression of the Y chromosome- linked sry gene, which controls mammalian sex determination. In these cases, backcrossing chimeras with the strain from which the ES cells were originally derived renders animals with the original genetic background, with an ideal 50% of the offspring carrying the modified allele. A 25% germ line transmission (one out of the first four male chimeras tested) is not unusual, although this percentage varies from clone to clone.

If conceived as a sequential approach, however, the notion of adding several genes to progressively specify stem cells in vitro would pose daunting technical difficulties. It would be very difficult to replicate the native expression pattern of most of these genes, which is either transient or fluctuating during development . Stepwise clonal derivation of cells transfected with one inducible gene at a time would be utterly impractical. Moreover, uncertainty about the copy number and site of integration of these cassettes would raise the aforementioned safety concerns. Strategies based on homologous recombination for targeting the integration of reprogramming genes into specific loci would be safer, but quite unworkable due to the number of genes involved. Also, it might not solve the problem of subsequent reactivation of these genes. Still, the in vitro behavior of cells typically defies conventional wisdom about natural development. Ngn3, for instance, is thought to act only transiently at the time of endocrine specification, and its down-regulation is supposed to be necessary for differentiation to proceed. However, ectopic constitutive expression of Ngn3 seems to be permissive for differentiation in other settings. Similar arguments can be made about ES cell reprogramming factors Oct3/4 and Sox2, whose permanent expression would be theoretically incompatible with the activation of differentiation pathways – yet differentiation is not impeded when iPS cells constitutively expressing these two factors are allowed to form teratomas. A plausible explanation for these observations is that the forced expression of some of these genes will trigger endogenous networks that will irreversibly take over the process from that point on, perhaps even deregulating the ectopic gene.Other forms of genetic manipulation are not intended to induce phenotypic changes, but to favor the in vitro identification (and possible sorting/selection) of desired differentiation outcomes, as well as posttransplantation tracing. These include plain tagging, the generation of cell lines expressing fluorescent reporter genes under the control of tissue-specific regulatory sequences and gene-trap strategies for the specific ablation of cells that do not have the desired characteristics.