Less than a decade ago the prospect for reprogramming the human somatic cell looked bleak at best. Despite the incisive studies of the previous forty years that included demonstrations by Briggs and King and Gurdon of the apparent reversibility of the differentiated state of simpler organisms such as the frog Rana Pipiens and the cloning of “Dolly” it seemed that the only methods at our disposal for the generation of human isogenic pluripotent cells would have to involve somatic cell nuclear transfer (SCNT, for an in depth review refer to 1). This procedure was technically demanding and inefficient in animals and was known to produce offspring of less than optimal quality when used for reproductive cloning but perhaps more importantly it had never been successfully applied to humans. In 2005, the first human embryo derived by SCNT was reported by our group  and this has been supplemented by the data of two other groups [3, 4] in the following few years but to date none of these have been successful in deriving the necessary embryonic stem cells (ESC) from such embryos. So what were we to do?
Enter Professor Shinya Yamanaka. In August 2006 his publication in Cell  promised to change everything by showing that it was apparently very simple to revert the phenotype of a differentiated cell to pluripotency by overexpression of four transcription factors in murine fibroblasts (Fig. 1). In the Autumn of the following year his group and Thomson's group published a highly similar procedure that worked on human fibroblasts [6, 7] and research into the mechanisms behind this cellular reprogramming protocol went into overdrive! A casual glance at the literature using “induced pluripotent” as a search term will highlight hundreds of articles from the development of in vitro models of human disease to studies of the mechanism by which as little as four transcription factors are able to change the epigenetic architecture of an entire genome. Theoretically this should be an enormously difficult task. The fact that almost no terminally differentiated cell types do this spontaneously is testament to this difficulty and indeed inducing pluripotency has often been described as the biological equivalent of making water flow uphill; however we all owe Yamanaka a debt of gratitude to pointing out how simple the induction of this process can be. Introduce the four factors then let the cell get on with it. Easy! In fact so easy there are probably several researchers out there thinking, “why didn't we think of that?” and maybe some of these were working on very similar ideas but as is always the case someone has to get there first.
The cells that emerged from the protocol devised by Yamanaka and Jamie Thomson became known as induced pluripotent stem cells or iPSC for short. The human versions are called hiPSC and although the technology that produced the initial cell lines was far from perfect (because it involved integrative retroviruses to introduce the four factors) the cells appeared to be highly similar to ESC and passed all the tests that determine pluripotency. Methods are now appearing that can generate iPSC without the involvement of such vectors and the most promising seem to suggest that it may be possible to produce these cells simply by exposing somatic cells to a combination of small molecules. The timeline shown in Figure 2 emphasizes the rapid progress made in iPSC research over the last 4 years. Therefore it would seem that we can produce pluripotent cells at will that moreover should carry the specific genomes on individual human beings. This has been hailed as an enormous development for regenerative medicine since transplant of the differentiated progeny of these individual iPSC should not be subject to immune rejection when transplanted back into the patient from whom the original somatic cells were obtained however, is the excitement surrounding iPSC entirely justifiable? Are they everything they promise to be or are we getting ahead of ourselves in the race to create the cells able to cure our ills? There are many outstanding questions with iPSC and solutions to these need to be found if these cells are ever to enter into human clinical application.
The Ten Most Pressing Questions About iPSC
The purpose of this editorial is to highlight issues we might need to sort out to realize the promise of iPSC in regenerative medicine. These views are not exhaustive but they are intended to be as accurate as we can be given our current state of knowledge. The following questions are (we believe) those that are most in need of satisfactory answers to allow iPSC technology to proceed.
1. How Similar Are iPSC and ESC?
Although at a first glance, iPSC and ESC seem to be very similar in terms of morphology, cell surface marker and gene expression levels, recent papers have demonstrated differences at the transcriptional level between the two cell types [8, 9]. Re-acquisition of pluripotency should suppress genes whose expression is specific to the somatic cell types from which iPSC were generated but in some cases expression of such genes is retained . This draws disturbing parallels with cloning by SCNT in which embryos and surviving offspring frequently express gene-specific to the somatic donor cell. It is difficult to determine if this is a normal feature of the iPSC derivation process because the aberrantly expressed genes are different for every cell line. At epigenetic level there seem to be some differences especially in their DNA methylation patterns particularly at the differentially methylated regions known as CpG shores . How these differences can and may affect their differentiation potential is currently unknown and needs to be investigated.
2. Are the Differentiation Abilities of iPSC Similar to ESC?
Several reports show that iPSC differentiate in a manner that is very similar to ESC in so far as their differentiated progeny seem to express similar marker genes and have similar morphology to the same cell types that differentiate from ESC [11–14]. Furthermore the limited number of transplantation studies performed suggest that iPSC derived somatic cells may be the functional equivalents of those from ESC [15, 16]. However, a recent paper suggests that although hiPSC use the same transcriptional network and development time course as the human ESC (hESC), their differentiation along the neural lineages is less efficient and more variable across the cell lines . It has been suggested by us and others [18, 19] that the reduced differentiation potential may be due to low levels of transgene expression, however the recent results obtained by Hu et al. seem to suggest that the variable differentiation efficiency was not due to presence of transgenes for similar results were obtained with hiPSC established with episomal transgenes . It is interesting to note that many of the differentially methylated regions referred to in question one are broadly involved in tissue differentiation so it is important to determine that the differences in DNA methylation patterns between iPSC and ESC have no significant effect on their ability to produce functional differentiated cell types. Even if we can prove that there is no difference between the differentiation properties of hESC and iPSC we are still in the same position as we would have been if ESC were our only source of cells since we still don't have completely effective methods to direct differentiation. All of this work remains to be done whether or not we choose iPSC.
3. Can We Make iPSC Technology Cost-Effective?
Generating and validating a hESC line under good manufacturing practice (GMP) conditions for clinical use is costly but if the cell line proves able to generate useful differentiated cells at least its growth in culture can be scaled up to industrial levels. This is the key to producing cells (for example pancreatic beta cells) in sufficient numbers to treat useful numbers of patients at a reasonable cost and some trials are currently underway in industry to demonstrate the effectiveness of this approach (Novocell Inc, San Diego, California. http://www.novocell.com/)). This process may not work so well with iPSC. If the objective of generating and using these cells is to produce patient specific cells the option of scale up is lost and we may have the situation where culture, differentiation and isolation of clinically useful cells has to be performed on an individual patient basis. Clearly this does not make economic sense; the cost of the resulting treatments would most likely be prohibitive so with present levels of technology, iPSC are more likely to make an impact in the development of in vitro disease models. This could work by producing iPSC from individuals affected by diseases for which there is a defined genetic component. Differentiation of these iPSC into the somatic cell types normally affected by the disease could enhance our understanding of its etiology that in turn could lead to exciting new drug leads. Another possibility is to create banks containing large numbers of iPSC lines. Sufficient numbers of these could provide some lines that have a close enough tissue match to the intended patient. It is not as perfect as complete patient specificity but at least it might be economically viable and it fits in with planned development of hESC banks along the same lines. Moreover, it would be much easier to generate iPSC lines with a wide genetic variety than hESC which will enable broader immune histocompatibility.
4. Are iPSC from Older Individuals As Good As Those from the Young?
A key facet of iPSC generation is that telomere lengths are reset to those normal for a pluripotent cell type of the species in question. Recent work  from Blasco's group shows that telomeres are elongated in iPSC and moreover that their telomeres acquire the epigenetic marks of ESC. This is important since clinical application of iPSC technology would be more likely to involve older individuals however cellular and organismal ageing doesn't just rely upon telomere shortening. Recent work from Lanza's group suggest that hemangioblasts and several other blood cell types derived from hiPSC show a limited expansion and early senescence, increased apoptosis and severely limited growth capacity . What about the accumulation of DNA damage that occurs during ageing? The mutation accumulation theory proposed by Peter Medawar in 1952  still finds favor with many workers in ageing research and it is difficult to see how the iPSC derivation process could repair or eliminate such excess errors. There are other non DNA sequence based errors that accumulate in the DNA with age such as the presence of 8-oxoguanine adducts and it is theoretically possible that some feature of pluripotent cells might be able to eliminate these lesions but it will still be necessary to confirm that the iPSC made from the elderly are an effective resource for regenerative procedures. This may not be an issue if we are restricting our focus to iPSC banks since only the best cell lines would be selected for inclusion in the bank.
5. What About the Reprogramming Method?
The first reprogramming protocols developed by Yamanaka, Thomson et al. while seemingly effective in achieving their aims would probably not have produced cells that could be applied in the clinical arena because they used integrative transgenes to deliver the four transcription factors. Insertion of the transgene vectors into numerous apparently random sites throughout the genome could, in principle, disrupt host genes, alter gene expression at nearby genomic loci or, if subsequently reactivated in the differentiated cells, result in these cells becoming cancerous. Fortunately iPSC generation can be achieved without viral vectors and without integration into the genome (see Fig. 2) but if anything these methods are even less efficient than viral protocols. The key to increasing efficiency of iPSC derivation and ensuring the safety of such cells for human application is a complete understanding of the reprogramming process. We know that reprogramming is almost always epigenetic since changes to the pattern of histone modifications and DNA methylation are required. For this reason histone deacetylase inhibitors can increase the efficiency of reprogramming in conjunction with only OCT4 and SOX2  but the precise roles of the four transcription factors are still obscure. The development of iPSC carrying inducible reprogramming factors  may allow us to work out the exact sequence of events taking place after transfection with the four factors. A possible alternative might be to examine “natural” forms of reprogramming such as that performed by the fertilized oocyte and by the primordial germ cells. Somatic cell nuclear transfer is the method of choice to examine the former and although it cannot be thought of as truly “natural” many of the processes applied by the oocyte to reprogramming the paternal genome seem to be conserved although this mechanism may well be quite different to that operative during iPSC generation. We already know that oocytes have very high concentrations of certain molecules such as nucleoplasmin and histones B4 and H3.3 that seem to be central to their reprogramming ability but although they are known to express OCT4 and SOX2 they contain very little NANOG gene product  and KLF4 and c-MYC are undetectable. It is therefore highly likely that other factors are also required to induce complete reprogramming and totipotency in oocytes but the eventual identification of oocyte-reprogramming molecules may well be able to enhance the efficiency of the iPSC derivation process. Useful insights can be also gained from study of heterokaryons that has recently led to the discovery that reprogramming towards the pluripotent state required AID-dependent methylation .
The value of understanding the reprogramming mechanism could lie in improving the quality of the iPSC product but it will be even better if this can be coupled to recent investigations that screened libraries of small molecules to identify compounds as potential alternatives to the retroviral or lentiviral vectors currently in use. This has led to some surprising leads which are particularly interesting since several of them fall outside known modulators of epigenetic modification processes such as histone deacetylase inhibitors. One of the most striking examples is RepSOX (a small molecule inhibitor of TGFβ signaling) from the laboratory of Kevin Eggan at Harvard University which as its name suggests can replace the SOX2 viral vector . Other small molecule hits are being identified that may replace OCT4 and KLF4 and since it is possible to derive iPSC using the combination of OCT4, KLF4 and RepSOX, there is some hope that small molecule based reprogramming may be commonplace in the near future.
6. What Happens to the Mitochondria?
Somatic tissues are often heavily reliant upon oxidative phosphorylation as their ATP source which means that they have considerable numbers of mitochondria. Conversely pluripotent cells seem to depend more upon glycolysis and it has been shown that ESC have comparatively few mitochondria. Pluripotent blastomeres of mammalian pre-implantation embryos are also characterized by limited oxidative capacity and great reliance on anaerobic respiration but moreover they restrict mitochondrial biogenesis so that the larger numbers of mitochondria present in the fertilized oocyte are partitioned between the blastomeres at each embryonic cleavage division. By the blastocyst stage this means that there are few mitochondria in each cell relative to the original oocyte and since we derive ESC from the inner cell mass at this stage then ESC have few mitochondria. The fibroblasts used to generate iPSC have lots of mitochondria so what happens to ensure that the resulting pluripotent cells have only the requisite numbers? Initial reports suggest that iPSC and hESC are similar in terms of their mitochondrial numbers and properties [18, 28] and their production of reactive oxygen species but how is such reduction in mitochondrial numbers achieved? Moreover if the cell is derived from an older individual it is possible that mutated mitochondrial genomes will present in the donor somatic cells. How can we be sure that the iPSC line does not contain these mutations or worse may even have selectively enhanced their presence?
7. What Is the Best Somatic Cell Type for iPSC Derivation?
The first iPSC were generated from dermal fibroblasts simply because these are easy to obtain and are commonly available in nearly laboratories. Some cell types (for example melanocytes, neural stem cells, hematopoietic stem cells and keratinocytes) can be reprogrammed to iPSC with higher efficiency than fibroblasts and in presence of fewer factors [29–31] and in shorter time. The number of reprogramming factors can also be reduced in some stem cell types for they may express some of those endogenously however with the possible exception of keratinocytes most of these alternatives are substantially more difficult to obtain than fibroblasts. A skin punch biopsy is much simpler than extraction of a bone marrow sample. Recent reports suggest that a combination of OCT4 and KLF4 can reprogram hair follicle dermal papilla cells into iPSC  and these cells should be even easier to obtain than fibroblasts. Of more practical value is also the reported derivation of iPSC from cord blood cells especially in view of their ease of access, fetal origin (hence probability of less acquired mutations with age) and existing cord blood banks . The real question when choosing a potential somatic donor cell is to balance ease of acquisition against propensity to reprogram. Some cell types appear to reprogram with consummate ease and it is possible (although not yet proven beyond all doubt) that some types may reprogram simply through an alteration of their culture conditions as highlighted by the work of Iqbal Ahmad leading to the apparently spontaneous reprogramming of rat limbal stem cells to something very like iPSC simply by exposing them to a specific set of growth factors intended to expand adult stem cell numbers and ESC conditioned medium to induced reprogramming . This is similar to the generation of apparently pluripotent stem cell lines from the germline stem cells of adult mouse testis simply by alteration in culture conditions  although whether these latter cells are the same as iPSC is open to debate.
8. Direct Conversion of One Cell Type to Another Without Reverting to Pluripotency
Can we dispense with pluripotent cells altogether? Some reports suggest it is possible to convert one differentiated cell type into another without having to reverse differentiation all the way back to a pluripotent state by using methods broadly similar to that used to induce reprogramming by Yamanaka. Using a pool of lentiviral vectors encoding 19 genes specific to the neural phenotype, Thomas Vierbuchen et al. . were able to convert fibroblasts into functional neurons and were eventually able to define only three of these that were necessary and sufficient for this conversion. Neither was this the first report of such “transdifferentiations” for as early as 2002, Phillipe Collas claimed to have induced fibroblasts to adopt a T-cell like phenotype by incubating the former with a cell free extract made from T-cells . Similarly Doug Melton's group reported that transfection of pancreatic exocrine cells with three β-cell specific genes was sufficient to convert them into cells indistinguishable from natural beta cells . In one sense we can rationalize these transdifferentiation experiments by considering the epigenetic landscape concept of Conrad Waddington . Pluripotent cells are at the top of a slope and differentiation represents their journey down that slope. It is really hard to climb all the way back up the hill to pluripotency (unless you're Shinya Yamanaka who made his own ski lift pass!) and in practice it might be easier simply to climb a little way up the hill then drop back down into a different but equally stable epigenetic state that represents a transdifferentiated cell. Only time will tell how successful this approach might be.
9. Ethics and Cloning
A major driving force behind the success of iPSC technology is their supposed ethical neutrality. Embryonic stem cell derivation involves the destruction of surplus embryos so quite naturally this has raised significant ethical objections to the practice of deriving hESC and since iPSC derivation avoids this they have been hailed as an ethical alternative. On a similar note, the earlier methods proposed for making patient specific pluripotent cells such as nuclear transfer (NT, often known as therapeutic cloning) were likely to have destroyed even more human embryos per derived cell line than simple hESC derivation thus at a first glance it seems that iPSC solves all of these issues at a stroke. However, it is theoretically possible to clone a human being using NT (reproductive cloning) and although this is formally outlawed by most nations, reports of attempts to do this have still appeared. One thing that participants in pluripotent stem cell research seem not to consider (or at least don't mention) is that iPSC technology may actually make cloning easier if they can be differentiated into primordial germ cells (which early evidence suggests that they can) and then prompted to develop further into gametes. Development of completely functional gametes from hESC/hiPSC has not succeeded as yet but it may be simply a matter of time until such cells are produced. A recent change to the embryology and fertility laws of the United Kingdom has tried to pre-empt such developments by prohibiting the creation of embryos from Human artificial gametes made by these or a variation of these techniques but this does not say that someday someone may attempt to implant such an embryo and thus clone a human being from iPSC. All strictly illegal – but then again so was reproductive cloning.
10. Are There Cells We Can't Derive iPSC from?
We stated earlier that the principal short term benefit if iPSC would be to develop disease models from affected individuals but what if the molecular problems that lead to the disease prevent the derivation of iPSC from the somatic tissues of that individual? This is more likely than might be first imagined. Fanconi anemia is a disease characterized by hematological abnormalities resulting from defects in the DNA repair mechanism of affected individuals and generation of iPSC from the skin fibroblasts of Fanconi patients has proved elusive to date . We are uncertain why this should be the case but it implies that DNA repair is essential to the epigenetic reprogramming mechanism (it may actually be involved in DNA demethylation) and so if this is defective the cells cannot reprogram. It may be possible to repair the genetic defect by introducing a non mutated copy of whichever of the 13 genes are known to be affected and contribute to the development of Fanconi anemia but if a disease results from complex multigenetic origins this may be very difficult to do. In any case if our aim was to generate a disease model why would we seek to correct the mutations in the first place? This could be useful if cell transplantation is a realistic treatment option (which for Fanconi anemia it could be) but it will not help us to use iPSC differentiation as a drug discovery system.
Induced pluripotent stem cells are a revolutionary tool for generating in vitro models of human diseases and may help us to understand the molecular basis of epigenetic reprogramming. However, further studies are necessary to improve the technology of hiPSC derivation, to determine their similarity to hESC and to improve targeted differentiation of pluripotent stem cells in general. The progress of the last four years has been truly amazing, almost verging on science fiction, but if we can learn to produce such cells cheaply and easily, and control their differentiation, our efforts to understand and fight disease will become more accessible, controllable and tailored.