You are here

Keystone Symposia



The focus of the meeting was to examine recent developments in stem cell differentiation and dedifferentiation and one has to say that these aims were largely achieved. The scientific organisers (Shinya Yamanaka, Kyoto, Japan and Fiona Watt, Cambridge, UK) had invited several high quality presenters. Vast amounts of data were presented so naturally this has led us to produce quite a lengthy report. Our objective in producing this report is to give an overview of the conference proceedings rather than an exhaustive description of every presentation, so we apologise if anyone feels they have been missed out or if there are any omissions of data that were described.

Day One

The first day of the meeting saw the keynote addresses by Shinya Yamanaka (Kyoto University) and James Thomson (University of Wisconsin) which effectively set the tone for the entire meeting since the bulk of presentations were concentrated on pluripotent cells.

Day Two

On the second day we got down to the real business of the meeting. The first talk saw Shinya Yamanaka once again take the podium to describe some of his lab´s results in the derivation of induced pluripotent stem cells (iPSCs). However, in a manner that has become characteristic of Yamanaka, he was keen to stress the possible problems associated with iPSC rather than trying to underline their potential usefulness. The key theme of his talk was that human iPSC (hiPSC) have a degree of transcriptional heterogeneity, like their mouse counterparts, but that microarray studies have not shown any common misregulated genes. In short his argument was that we really know very little about the characteristics of hiPSC and the only measure we have of their pluripotency is that they are able to form teratomas. In view of this we need to define which qualities the best hiPSC lines should have to produce ´safe´ iPSC. This means we will need to specify the best donor somatic cells and reprogramming methods.

Next up was Rudolf Jaenisch who described some results from his reprogrammable iPSC mouse model. Many of these data have been published prior to the meeting so we can restrict our reporting of Prof Jaenisch´s talk to his treatment of X-inactivation. The most interesting observation was that although human embryonic stem cells (hESC) should normally have one inactive and one active X-chromosome (as opposed to mouse ESC (mESC) which often have both x-chromosomes active), culture under 5% oxygen conditions appears to repress the XIST gene which is involved in x-inactivation. However, if the cells are exposed to normal 20% oxygen first, they undergo x-inactivation that is not reversible by transferring the cells to 5% oxygen.

Helen Blau was next to present. She had gone back to an older method for studying the reprogramming of a somatic genome which involves fusion of the somatic partner with a pluripotent cell, in this case an mESC, to examine what happens to the epigenetic status of the somatic chromosomes. The ESC phenotype seems to be dominant in such heterokaryons. Fusions of human fibroblasts and mESCs are a good model because the RNA products of the human and mouse genomes can be distinguished by PCR. By this method, they showed that OCT4 and NANOG are rapidly upregulated from the human genome following fusion which was, perhaps not unexpectedly, accompanied by extensive DNA demethylation at the promoters of these genes. Such demethylation turned out to the principal theme of this talk as she went on to describe the AID (or AICDA) gene which may be important for DNA demethylation, since it can deaminate 5-methylcytosine to produce thymidine which is then detected as a mismatch by the DNA repair machinery of the cell. This results in replacement of thymine by cytosine which amounts to demethylation of the 5-methylcytosine at this position. The data presented included siRNA knockdown of AID in the heterokaryons which seemed to block the demethylation activity quite effectively. Helen Blau pointed out that their model still requires absolute proof particularly with regard to the exact mechanism of AID mediated demethylation and what targets this enzyme to one set of genes and not others. However, she did also point out that Wolf Reik in Cambridge has shown AID activity in primordial germ cells (PGCs) that are also capable of extensive DNA demethylation during their development. This point was to be addressed in a later presentation by Azim Surani.

After a brief description of methods for making pig iPSC by Jenny Liao there was an interesting talk from Christoph Bock who is trying to develop functional genomics assays to quantify the utility and safety of human hiPSC. Basically his method involved differentiating both iPSC and ESC as embryoid bodies for 16 days followed by genome wide DNA methylation mapping to see how differentiation affects this important epigenetic modification. Dr Bock argued that while there are other methods to measure epigenetic changes such as histone modifications patterns; this would require the use of techniques such as ChIP-on-chip or ChIP sequencing which although powerful, requires large numbers of cells. In contrast DNA methylation mapping can be done with a few as 5000 cells which lends itself nicely to the quantitative differentiation assay proposed in Dr Bock´s talk.

So what were his conclusions? Essentially his data suggest that every ESC and iPSC line has some differentially methylated promoters but no single locus subject to this type of epigenetic modification can be used to distinguish and iPSC from an ESC. He also suggested that most of the genes that should not be expressed in ESC are more or less the same in iPSC lines. Such inactive genes tend to have hypermethylated promoters, however, although there was some variability in the numbers of genes showing hypermethylation between ESC and iPSC, there was a very good correlation between genes showing hypo-methylation (and therefore gene activation) in ESC and iPSC. This point was put forward as evidence that iPSC retain very little epigenetic memory of their somatic origins however, this doesn't explain why there should be variability of DNA hypermethylation at some genes. Dr Bock argued that there is possibly some degree of random hyper- or hypo-methylation at genes that are not immediately involved in the pluripotency maintenance network of ESC or iPSC.

After lunch (or more appropriately the skiing break!) Austin Smith (Cambridge University) presented some of his findings on the differences between mESC and the epiblast stem cells (epiSCs) and how these data relate to the ground state of pluripotency in mESC. Essentially the bulk of Austin´s presentation related to his recent publications on the subject but it was particularly well delivered and explained the subject matter clearly.

The next presentation from Ihor Lemischka (Mount Sinai School of Medicine) also described some of his recently published data on systems level analyses of fate changes in mESC. Ultimately he described an enormous amount of work performed on pluripotent and differentiating mESC to characterise their transcriptomes (mRNA), epigenomes (histone acetylation and RNA polymerase II binding maps) and their proteomes. This was all valuable data since it could be a useful tool to compare hESC and hiPSC, not only to each other but also to confirm that they differentiate in predictable ways to give cell types that do what they are supposed to do. However, the presentation covered a vast amount of published data and that we will not report here.

Amanda Fisher (Imperial College, London) was next to present some data that again put forward the idea that heterokaryons are a good model in which to study epigenetic reprogramming. Fusion of human B-cells with mESCs produces heterokaryons in which all the expected pluripotency genes are upregulated from the formerly somatic human genome. This is accompanied by downregulation of genes specific to the B-cell phenotype. The great advantage of this experimental system for studying reprogramming is that mESC with knockouts for various genes that might function as epigenetic modifier proteins can be used to observe what effect this has on the ability of ESC to reprogram a somatic genome in the heterokaryon. They showed that ESC lacking activity of Polycomb Repressive Complexes 1 and 2 (PRC1 and 2) were substantially less able to reprogram than wild type ESC and led them to hypothesise that the reprogramming effect might be due to PRC1 and 2 repressing genes that would otherwise interfere with reprogramming. They tested this by using RNAi to knockdown genes normally repressed by PRC1 and 2 but this failed to rescue a reprogramming phenotype in the PRC1 and 2 knockout cells. They did show that the gene JARID2 seems to be involved in reprogramming and demonstrated its enrichment at the promoters of genes known to have bivalent chromatin domains in pluripotent cells. Quite how this is able to mediate reprogramming they were unable to say. The presentation raised more questions than it provided answers, as is often the case in science, but it does provide a useful pointer for future research and the Fisher group are currently interrogating the factors required for successful reprogramming of human blood cells.

The first day ended with a couple of short talks from promising newcomers to the field. The last of these by Li Fang Chu (Baylor College) put forward an interesting hypothesis that primordial germ cells (PGCs) were actually the source of ESCs cultured from the inner cell mass (ICM). This was unusual, but certainly interesting, since there is documented evidence that PGCs are able to convert into embryonic germ cell lines, although these are a slightly different form of pluripotent cell. The hypothesis put forward here implies that PGCs, or at least a form of precursor cell type, exists in the ICM which is counter to the accepted wisdom that PGCs arise from the proximal epiblast at a later stage of development. However, Chu provides evidence for his idea by showing that clusters of cells expressing the PGC gene BLIMP1 (or PRDM1) appear in outgrowths of ICMs plated onto feeder cells. The cells were able to generate colonies of alkaline phosphatase positive cells whereas BLIMP negative areas of the ICM apparently could not do this. Moreover, he showed that these BLIMP1 positive cells were able to colonise the genital ridges of murine E8.5 embryos after in utero injection which lends credence to their PGC identity. This unusual idea was formulated into a hypothesis that ESC do not arise directly from pre-implantation epiblast cells but transition through an intermediate PGC stage. This is quite difficult to test from his current data as no other markers that are characteristic of the PGC phenotype was shown, but the idea seemed to stick with several other members of the meeting since it came up in subsequent presentations.

Day Three

The first slot of day three fell to Azim Surani (Cambridge University), which conveniently followed on from the germ cell topic of the last talk on day two. Surani is of course well known for his work on PGCs from mouse embryos and he began by describing new results from his investigations of the regulation of the Blimp1 gene by Lin28 and Let7. He also alluded to the idea that a form of PGC might be an ESC precursor by noting that Blimp1 negative epiblast cells are unable to produce PGCs in vivo. However he quickly moved on to more defined and published work on the changes in histone modification patterns and nuclear morphology during PGC specification in vivo. His novel results were concerned with the possibility that deamination of 5-methycytosine by Aid/Apobec genes, followed by base excision repair to replace the resulting thymidine with cytosine, might be the mechanism of the rapid DNA demethylation that occurs either soon before or immediately after arrival of PGCs in the genital ridge. This followed on from Helen Blau´s talk on day one, but Surani provided some nice data showing the coincident expression of Aid and Apobec genes with the known timing of DNA demethylation in ex vivo PGCs and the appearance of chromatin bound Xrcc1 which is known to be involved in the base excision repair process.

Surnani´s presentation resulted in a number of questions, but probably the most interesting one was that since base excision repair is a fundamentally error prone mechanism, how would this be acceptable in cells of the germline that are supposed to maintain genome integrity to a very high level? Also Helen Blau asked if loss of function studies should be performed but Surani countered both of these questions by saying that neither of these points could be easily examined with the small numbers of PGCs available from the mouse embryo.

The next slot was Hans Scholer (Max Planck Institute for Molecular Biomedicine, Germany) who began by asking how hESC relate to mESC and EpiSCs but in a short while the topic changed to Prof Scholer´s principal research focus of adult germline stem cells (GSCs). The argument of this section of the talk was that precise selection of culture conditions is able to convert GSCs from mouse testis into pluripotent cells that have a very similar transcriptome to mESC with the exception of their imprinted genes which show an androgenetic pattern as one might expect given the origin of the cells. In effect this amounts to cellular reprogramming by alterations in culture conditions without addition of extrinsic reprogramming factors such as viruses. Interesting enough in its own right but Prof Scholer then presented a critique of the one paper that claimed to have achieved similar results from human testis biopsies and showed that at a transcriptomic level the cells obtained in this latter paper were more similar to fibroblasts than to hESC. Clearly there is still much to be achieved in the field of pluripotent GSCs. Staying with the idea of non viral reprogramming, the next talk by Kevin Eggan (Harvard University) described his group’s efforts to replace viral reprogramming vectors with small molecules that alter the epigenome. His principal finding was the molecule he has named RepSOX due to its ability to replace SOX2 in fibroblast reprogramming experiments which seems to work as a TGFb inhibitor. These experiments still require the other vectors of course so the search is on for small molecules that can replace all of the Yamanaka factors and Eggan appears to have some candidates that can replace OCT4 and KLF4. Since RepSOX can replace c-MYC in the reprogramming process and a combination of the OCT4 and KLF4 vectors with this molecule is sufficient to generate iPSC, there are high hopes that we might be able to dispense with viral reprogramming in the near future.

Chad Cowan (Harvard University) is into fat! More precisely he is interested in obesity and how we can understand the pre-disposition of some individuals to accumulate excessive adipose tissue by examining the characteristics of adipocytes differentiated from iPSC of obsese individuals. This is an interesting model system and Dr Cowan´s presentation represents the first mention of the use of iPSC technology for disease modelling in this Keystone meeting. At present he has been able to establish methods to differentiate adipocytes from hESC and some limited data on hiPSC and these data formed the body of presentation.

The morning session continued with a couple of short talks from Azadeh Golipur (Samuel Lundefeld Research Institute, Canada) who was looking at RNAi screens to identify genes regulating the initial stages of reprogramming in iPSC generation and Colin Melton (USCF) who has examined the changes in miRNA expression during differentiation of mESC. To end we had a presentation by Larry Stanton (The Genome Institute of Singapore) describing the structure-function relationships of SOX2 and SOX17 and how these could be interconverted by removal of certain structural motifs.

The afternoon session on day three began with Takashi Shinohara (Kyoto University, Japan) and his description of the positive and negative regulators of mouse GSCs (mGSCs) and how they used this information to develop defined culture media incorporating glial cell line derived neurotrophic factor and bFGF to maintain GSCs indefinitely in vitro.

Anthony Atala (Wake Forest University) was up next with an impressive review of his work over the last decade concerning the usefulness of multipotent cells derived from the amniotic fluid and placenta. These stem cells express OCT4 and SSEA4 but their profile of other surface markers suggests that they may represent a stage somewhere between ESC and lineage restricted adult stem cells. Despite this, they seem to have many characteristics of pluripotent cells including the ability to differentiate into cell types of all three germ layers but crucially they do not form teratomas in vivo and so have been suggested as a possible alternative to hESC/hiPSC.

We ended day three with a talk by Thomas Zwaka (Baylor College, Texas). This was a departure from the published programme but was interesting nevertheless because of the hypothesis that pluripotent cells rely more upon a tightly controlled, but dynamic, network of gene expression to maintain their phenotypes, whereas somatic cells may be more reliant upon the activities of their housekeeping genes. One of the principal genes in this network seems to be Ronin which has been shown to bind 866 sites throughout the mESC genome and seems to be an activator of the mTOR pathway. Approximately one third of Ronin target genes are also occupied by Oct4, so there is a possible link into the pluripotency gene network, although its principal functions seems to be regulating the key metabolic processes (such as ribosome biogenesis) in mESC. This may be one of the mechanisms by which they prevent their differentiation.

Day Four

The focus of the meeting changed today. Days one, two and three were concerned largely with pluripotent cells but today we switched our attention to adult stem cell types. The first talk of the day was given by Fiona Watt (Cambridge University, UK) describing her admirable studies of the epidermal stem cell niche. Much of her data on the identification of epidermal stem cells on the basis of Lreg1 expression has been aired before, but her description of collagen islands printed onto gold coated glass slides as a means of controlling epidermal stem cell differentiation was excellent and highlighted many interesting observations on the control of stem cell differentiation by shape and stiffness of the cells 3D environment. Data was also presented showing how the actin cytoskeleton can mediate these effects.

Haematopoietic stem cells (HSCs) were the focus of the next talk given by Toshio Suda (Keio University, Japan), wherein data were presented to show how hypoxic microenvironments are necessary to maintain long term haematopoietic progenitor cells in a quiescent and multipotent state. It has been suggested that increasing levels of reactive oxygen species (ROS) are one of the triggers causing differentiation of such HSCs and Suda´s data show that HIF1a (hypoxia inducible factor 1 alpha) may be involved in mediating this trigger, since HIF1a knockout mice show loss of quiescence in their HSC population and also decreases in HSC numbers as they age. This is the converse of normal ageing where HSC numbers actually increase (although they may not function so well as will see from the next talk) and suggests that the HSC pool size may be controlled at least in part by HIF1a.

In an interesting parallel with pluripotent cells, they showed that quiescent HSC are more dependent upon glycolysis than oxidative phosphorylation as their ATP source which fits well with the need to restrict the production of ROS. However a switch to oxidative phosphorylation occurs in cycling or expanding HSC. On a similar note Dr Suda also observed that the mitochondrial mass of quiescent HSC is quite low which draws another parallel with pluripotent cells.

The haematopoietic focus was maintained by the next speaker in what was perhaps one of the most though provoking lectures of the whole meeting. Amy Wagers (Harvard University) has recently published her findings on a possible contribution of serum borne factors from younger mice to a process that could be loosely termed “rejuvenation” of the stem cell population of older mice. These data arose from a set of heterochronic parabiosis experiments, which involves connecting the circulatory systems of one mouse to another to see what happens to the individuals concerned. However, when old mice were exposed to the circulatory environment of young mice, their skeletal muscle progenitor cells and HSCs began to behave in a similar fashion to those in the younger parabiont.

Decline in tissue function is a normal facet of ageing, with muscle atrophy and haematopoietic dysfunction being among the principal characteristics of this decline, and it is normally thought to be irreversible. Interestingly the work presented by Amy Wagers draws upon earlier developments by Thomas Rando and Irina Conboy (Berkelely CA) that showed some improvement of skeletal muscle satellite cell function in this heterochronic parabiosis model but this is the first demonstration that the effect may apply to other stem cell compartments. Wager´s data seem to imply that the circulating factors may affect the cells that define and create the HSC niche rather than the HSC themselves, since exposure of ageing osteoblasts to factors within young serum was needed to ensure that the HSC pool size reduced to youthful levels and that the skewing of myeloid versus lymphoid differentiation was reversed. Regarding the nature of such “rejuvenating” factors, rather than undertake a laborious analysis of the differences between old and young sera, the Wagers group undertook a small molecule screen to see if any promising candidates might be able to replicate this effect in vitro and in vivo. One ALK4 receptor kinase inhibitor did seem to be able to partially recapitulate the effect!

Like many developments this presentation raises more questions than it answers. A reasonably well tested theory of ageing is that many of its problems arise through the accumulation of harmful mutations in the DNA. If this is true how does parabiosis reverse this accumulation (if indeed it does so at all)? Similarly can it restore the telomeres of aged HSC? Even in the light of such questions the data is still a fascinating new addition to the investigation of ageing that might help us to elucidate the “genetic or epigenetic” basis of the ageing phenomenon.

Fred Gage (Salk Institute) had some interesting data on the expression of Sox2 in the self renewing cells of the dentate gyrus. Progeny of these cells differentiate into neurons within one month of the cells “birth” and this process appears to continue throughout adult life. The Gage group is studying the cellular, molecular and environmental influences that regulated neurogenesis in the adult brain as well as looking at methods to modulate the levels of adult neurogenesis. But perhaps the most interesting part of their work is using iPSC from humans and non human primates to see if there are differences in their neuron forming capabilities that could explain differences between the human and primate brains.

The afternoon session saw Deepak Srivastava (UCSF) discuss his work on miRNA regulation of cardiac stem cell fate. His group has identified individual miRNAs that govern differentiation of ESC and iPSC into mutlipotent cardiac progenitors, and some of these also direct subsequent differentiation into cardiomyocytes, endothelial and smooth muscle cells. These miRNAs appear to regulate transcriptional networks and signal transduction pathways central to the adoption of a cardiomyocyte cell fate and are amenable to manipulation which opens up more possibilities for directing the differentiation of pluripotent cells.

We had a brief reversal back to iPSC related topics for the talk by Hideyuki Okano (Keio UIniversity, Japan) who described the possible uses of iPSC for regenerating the damaged central nervous system. They found that miPSC could be induced to form neural stem cells using the same methods for mESC (Naka et al, Nature Neuroscience, 2008) and that these were able to contribute to the repair of damaged spinal cord in immunodeficient mice. Crucially these animals showed significant improvements in their ability to control their hindlegs after transplant and the only difference to transplant of cells derived from hESC was that fewer glial cells were produced from the iPSC. Hans Kierstead (UC Irvine) continued the spinal cord injury theme with similar data concerning the recent clinical trial by Geron using hESC to treat this condition.

The fourth day was finished off by Catherin Niemann (University of Cologne, Germany) who changed the focus completely with her description of sebaceous gland homeostasis by hair follicle stem cells.



Day Five

The last day of the meeting saw talks on a mixture of subjects ranging from differentiation of ESCs into cell types of potential clinical use to more immediate possibilities for applying ESCs to the discovery of new drug leads. The first talk of the morning was the one of the more industrial presentations of the meeting given by Emmanuel Baetge (Novocell, San Diego) who described the impressive progress made by Novocell in developing and validating protocols to produce pancreatic islets for treatment of type I and II diabetes.

Through a stepwise differentiation protocol modelled after pancreatic development, Novocell have been able to generate pancreatic endoderm progenitor cells which could produce glucose responsive endocrine cells upon transplantation into the epididymal fat pad, kidney capsule or subcutaneous or omental sites of SCID mice. These glucose responsive islet cells were capable of maintaining stable blood glucose levels in streptozotocin treated mice (which destroys their exisiting beta cells and induced hyperglycemia) for more than one year. Dr Baetge continued by describing Novocell´s efforts to scale up production of islets to deliver a safe and affordable treatment for diabetes.

Ron Mckay (NIH, Bethesda) was up next and although starting with some of his older data on neural stem cell differentiation from ESCs, he really wanted to make the point that hESC lines seem to differentiate in a predictable and uniform fashion and that hiPSC are basically the same except for one or two notable genes (such as glutathione-S-transferase (GST)). He debated the point at some length whether or not this might be due to individual genetic differences. but went on to form an interesting idea that iPSC might be a valuable way to answer questions about evolutionary biology since we have the opportunity to study an enormous number of iPSC lines derived from many individuals in a way that is somewhat impractical using hESC. After introducing this point he went back to the slightly more mundane, but undoubtedly valuable, differentiation of hESC to hepatocytes and indicated that some of the iPSC lines he was using were rather better for making hepatocytes than hESC.

Probably the real focus of the day was the possibility of translating hESC research into clinical realities. Consistent with this the next presentation came from Rita Perlingeiro (University of Minnesota) who discussed the differentiation of ESCs into muscle progenitors. Basic methods for differentiating mESC as embryoid bodies seem to be very inefficient for making myogenic cell types. To improve on this situation her group produced Doxycycline inducible Pax3- and Pax7-GFP constructs and stably transfected these into mESC. They differentiated these to give GFP-expressing PDGFRa–FLK1 cell populations that were able to engraft in the MDX mouse model for Duchenne muscular dystrophy, although in the case of Pax3-GFP, this nascent mesodermal population may have been transplanted too early after induction of differentiation, since teratomas were produced. The results with the Pax7-GFP line were rather more encouraging since not only were they able to contribute to repair and regeneration of dystrophic MDX muscle, but they also took up anatomical locations typical of the skeletal muscle satellite cells. Furthermore, these cells showed transcriptomic profiles similar to those of satellite cells isolated ex vivo. Interestingly miPSC showed almost identical Pax7 expression to mESC upon differentiation and preliminary evidence suggests that the Pax7 expressing cells may be able to engraft and repair MDX muscles albeit with a lower efficiency as the same cells derived from mESC.

An industrial focus continued with the talk given by Nobuko Uchida (StemCells Inc, Palo Alto) although the data was more involved with neuroprotection, that is using derivatives of ESC to either replace enzyme deficiencies or re-myleination strategies. The first example was of the neuronal ceroid lipofuscinoses, a rare group of fatal neurodegenerative diseases in which a lysosomal enzyme crucial to the removal of the products of lipid damage such as lipofuscin. Failure to remove this material effectively can lead to neuron death, but the therapeutic options are limited since the enzyme cannot cross the blood brain barrier if administered into the peripheral vasculature. To get around this problem, Stem Cells Inc are developing methods to transplant human central nervous system stem cells (Hu-CNS-SC) into mouse models of these diseases. The company has generated cell banks from CD133-positive cells enriched from the human CNS and suggests that unlike ESC or iPSC these cells do not require prior differentiation before transplant and they do not form teratomas. Another potential area of application for these cells is re-myelination of damaged axons. When injected into the shiverer mouse model (which has a myelination defect) these cells were able to contribute to axon repair. The company has recently received FDA approval to start a phase I study for the use of Hu-CNS-SC in Pelizaeus-Merzbacher disease, a fatal myelin disorder. A third potential application of this adult stem cell population is protection of the photoreceptors in an animal model of retinal degeneration. The company has hailed this as a possible treatment for age related macular degeneration.

The morning session ended with a return to more fundamental ESC studies with a presentation from Michael Drukker (Stanford medical school) on early ESC differentiation.

The first three talks of the afternoon were devoted to drug and biomarker discovery using ESC. Amy Sinor (Harvard Stem Cell Institute) was investigating spinal muscular atrophy using ESCs isolated from a mouse model of this disease and uncovered various pathways regulating the survival of motor neuron gene (SMN) in ESC derived motor neurons. In addition, her group has screened a large number of small molecules to find candidates capable of increasing SMN levels and decrease motor neuron death. Zhong Zhong (Glaxo SmithKline R & D, China) was doing broadly similar research aimed at drug discovery in neuroscience but Gabriela Cezar (University of Wisconsin) took a different approach based upon the identification and quantitative analysis of biochemical processes in neurodevelopmental disorders (a process she refers to as metabolomics) to identify new biomarkers for disease processes.



The meeting was very engaging since many of the presentations provided us with data that has not yet been published. Keystone meetings often provide the participants with the opportunity to reflect upon their current research projects in the light of findings from other workers in the same or similar fields and this meeting was no exception. We look forward to seeing more data from the labs of those who presented over the last few days and to some of the exciting developments, protocols and technologies that may well arise from their work.