This year's meeting of the International Society for Stem Cell Research in Philadelphia, Pennsylvania, included a jam-packed session on the standards and methodologies of creating induced pluripotent stem cells. But although excitement around advances in reprogramming somatic cells shows no signs of abating, new ideas regarding the field are surfacing.

Cellular memory

The meeting opened and closed with the notion that even genomes that have been reprogrammed to pluripotency retain some imprint of their former, differentiated selves. In the opening session, John Gurdon, of the Cancer Research UK Gurdon Institute in Cambridge, described how a frog nucleus taken from a differentiated cell retains certain gene-expression patterns even after being placed in the reprogramming elixir of an oocyte and undergoing multiple rounds of cell divisions. He posited that this cellular memory was maintained not so much by transcription factors but by histones capable of recruiting similarly functioning histones throughout cell divisions.

In the last session, Shinya Yamanaka of Kyoto University discussed his famous experiments showing that inserting a suite of genes can reprogram adult cells to pluripotency and that these induced pluripotent stem (iPS) cells can contribute to all tissues in adult mice. But he also presented new data showing that these cells remember their origins. Yamanaka's lab, and others, have now generated chimeric mice by mixing mouse embryos with reprogrammed cells from various sorts of specialized cells. Yamanaka had previously reported work showing that mice reprogrammed using the known oncogene c-Myc have unusually high rates of tumours and die early. However, the oncogene provided a ready explanation for these observations. During his talk at the International Society for Stem Cell Research (ISSCR), Yamanaka presented a puzzle to the other attendees. Though all the reprogrammed cells pass the tests of pluripotency, chimeric mice generated from reprogrammed stomach and liver cells die younger than mice generated from reprogrammed fibroblasts. Furthermore, mice generated from adult-derived fibroblasts die younger than those generated from fibroblasts collected from immature animals.

“There is just one clue” to the puzzle, Yamanaka said. His team found more methylation in iPS cells originally derived from liver and stomach cells, suggesting that reprogramming is still not complete. He expects that this partial reprogramming may be the cause of higher mortality. After the talk, one questioner remarked that the methylation patterns reminded him of what he'd seen in cloned animals. He wondered if cloned animals and those generated from iPS cells had been thoroughly compared.

Throughout the conference, scientists remarked on the differences between reprogramming processes driven by the egg (placing a differentiated nucleus in a reprogramming potion) and 'direct reprogramming' (inserting genes into a differentiated nucleus). The former happens almost instantaneously and apparently without the need for gene expression, whereas the latter requires continuous expression of introduced genes for a week or more. Perhaps, the question implied, there should be some hunt for the similarities between the two processes.

Exchanging cell fates

If the biggest news at last year's ISSCR meeting was that mouse cells can be reprogrammed completely back to pluripotency, the biggest news this year might have been that doing so might be unnecessary for directing cells down new lineages.

Two presentations gave new meaning to 'transdifferentiation' — a term that means one specialised cell type morphs into a different specialised type and refers to a notion that has been largely discredited over the past decade. A separate presentation brought to light data on cancer stem cells that casts doubt on a number of studies involving the cells. Although none of this information, the data or the studies, has been published in the peer-reviewed literature, they are already causing a stir. And of course stem cell scientists' fruitful obsession with how epigenetics controls cell fate remains a dominant theme.

Doug Melton's lab at the Harvard Stem Cell Institute, in Cambridge, Massachusetts, has been focusing on transcription factors that regulate the development of beta cells. At the meeting, Melton showed that expressing three key transcription factors in the pancreas of a living mouse converts exocrine cells directly into endocrine cells that closely resemble insulin-secreting beta cells. When transplanted into mice with induced diabetes, the cells relieved hyperglycaemia by secreting insulin. Melton invoked the Waddington model that depicts the tendency toward cell fate as a topological landscape where marbles take one of several downhill paths to a differentiated state. General opinion in the field presumes that for a specialized cell to take on a new fate, it must first erase its character, dedifferentiating completely. Melton thinks that rather than rolling uphill and moving down another path, reprogramming techniques can tunnel under the hill, so to speak, to move cells from one fate to another. Because these two types of cells are within the same organ, the cells may not so much be tunnelling under a mountain range as skipping over a very shallow ridge; nonetheless, the cells' appearance and behaviour changes drastically. These reprogrammed cells are not therapeutically useful because their insulin release is not regulated by glucose; however, the research sets the stage for new ways of manipulating cell fate to obtain clinically useful cells.

The field does not yet have the terminology to describe these fate-shifting concepts precisely. Though Melton does not use the term, his work evokes the concept of transdifferentiation — a concept celebrated around eight years ago when scientists started observing switches in cell fate in Petri dishes or, more strikingly, upon injecting labelled cells into mice. Notable examples were bone marrow cells transdifferentiating to cardiomyocytes and neurons. Since then, most if not all of those events have been ascribed to cell-fusion events, and, owing to the uncontrolled and rare nature of the events, interest in transdifferentiation waned.

Though Melton prefers the term 'reprogramming', Thomas Graf from the Centre for Genomic Regulation in Barcelona embraced the word 'transdifferentiation' for his method of converting one type of blood cell to another. Graf's lab found they could convert immature B cells to macrophages with high efficiencies by engineering an avian leukaemia virus to insert certain genes into cells. Specifically, they used GATA1, which switches cells from one phenotype to another, and another factor, PU.1, which induces a lineage switch in the opposite direction. He found that this simple manipulation caused thousands of genes to change expression. However, when he searched for signs of de-differentiation, he found none. The cells did not show transient expression of stem cell–specific genes or signs of returning to an earlier progenitor-type stage. Instead, they activated a subset of genes for a specific lineage and then took on a new identity.

“It is not reprogramming,” he said at one point. [Editor's note: If you have views on the preferred terminology here, please email theniche@nature.com.]

Undiscovering cancer stem cells

Several talks pointed out that stem cells' behaviour depends on their environment, but none was as startling as that given by Sean Morrison, of the University of Michigan, in Ann Arbor, who presented results that severely undercut a series of cancer stem cell studies from recent years. Last year in Science, researchers Stephen Nutt and Andreas Strasser from the Walter and Eliza Hall Institute of Medical Research, in Melbourne, reported that the basis for an oft-used assay for cancer stem cells may just be an artefact of the hostile environment of grafting human cells to mice, but they looked only at a particular type of leukaemia1.

This year, Morrison provided evidence that, for at least one study in melanoma, their hypothesis holds true outside the blood. The assay in question asked if a human tumour possesses rare stem cells capable of re-initiating tumours upon transplantation. In these studies, human tumours were separated into individual cells; the cells were divided into groups by fluorescence-activated cell sorting based on the presence of particular markers on their surface; then, groups of sorted cells sharing similar markers were injected into mice. In past work, scientists have found that only rare cells possessing certain markers were capable of forming tumours. That, combined with their ability to be serially transplanted, gave these rare cells the hallmarks of a stem cell.

In leukaemia, the cells that could contribute to tumours were presumed to be as rare as one in a million cells. A key concept here is 'rare' — if half the cells in the tumour possess this property, then they really don't fit the definition of a stem cell, let alone a cancer stem cell. Several studies did identify 'rare' cells that seemed to fit the bill for a cancer stem cell identified from human tumours.

Morrison confirmed Strasser's hypothesis by showing that the rareness of human cancer-forming cells in the transplant assay may not be a consequence of stemness, but rather of their ability to thrive in a mouse. In other words, tumour-causing cells may not be so rare. Working with human melanoma cells of a type that had previously been shown to possess stem cells, Morrison's lab experimented with ways to pamper the cells before transplanting them to the mouse, trying to maximize their survival. They found that after this sort of manipulation, transplants of even single cells could lead to tumours in a mouse host more than ten per cent of the time. So if this tumour-forming ability is so common, can the cells that possess it be called stem cells?

Of course, the cancer stem cell field does not rest entirely on assays involving the transplantation of fractionated human tumour cells into immunocompromised mice. Studies using lineage tracing of stem cell markers to sites of tumour formation in mouse models support the cancer stem cell hypothesis. Other work supporting the cancer stem cell hypothesis focuses on the genetic similarities between embryonic stem cells and cells taken from human cancer samples.

On the other hand, cancer is a notoriously heterogeneous disease. Perhaps tumour-initiating cells are just as diverse. Surely these questions and more will be taken up at the meeting next year in Barcelona.

Related Articles

Stem cells not all to blame

Cancer stem cell sightings and slightings

Stem cell meeting 2007: routes and roadblocks on the way to cures

How do you know a reprogrammed cell is reprogrammed?

Stem cell society condemns unproven treatments