Like all stem cells, cancer stem cells produce cells like themselves and cells that form other, more mature cell types, but cancer stem cells lose the ability to hear signals that say either “you've proliferated enough” or “it's time to differentiate,” said Stanford University's Mike Clarke, a co-organizer of a 4 February symposium on cancer and stem cells held at Stanford, in Palo Alto, California. Researchers described their work in worms, flies, cell lines, mice and more.

Driving cells to differentiation

When stem cells divide, they can divide into two stem cells, two cells committed to a more differentiated fate, or one of each. Particularly in mammals, there's no good way to track which option stem cells 'choose' when they divide. Tannishtha Reya of Duke University, in Durham, North Carolina, described how she was able to watch this process at the single-cell level using a new imaging system, oncogenes and genetically engineered mouse cells1.

The basic message of the day was that the path from stem cell to fully differentiated cell follows three main phases, all of which are tightly regulated.

The Notch signalling pathway is usually shut off in mature cells. In fact, if the expression of green fluorescent protein (GFP) is linked to a Notch signal, haematopoietic stem cells appear green, and the colour dims as cells differentiate. Reya's team grew haematopoietic stem cells on a glass-bottom plate and imaged the cells every ten minutes for three days, a time frame that is usually long enough for the cells to divide twice. She saw divisions in which both daughter cells expressed GFP (meaning that they were both stem cells), divisions in which neither did (meaning both were committed) and divisions in which one daughter cell expressed GFP and the other didn't (meaning one daughter had committed).

Next, she investigated whether the cells's 'decisions' were preprogrammed or responsive to the environment. She found that the stem cells shifted their pattern of division based on the cells they were cultured with.

But she also tested genetic control by using different oncogenes. A less aggressive leukaemia showed greater growth and survival normal stem cells but no change in division pattern. In contrast, a more aggressive leukaemia favoured renewal, with both daughter cells more likely to become stem cells.

These discoveries raise the exciting possibility that if there were a way to identify a protein that would allow the cells to recover their ability to divide asymmetrically, it could turn an acute, aggressive and untreatable leukaemia into one that is chronic and manageable, or perhaps even halt its growth completely.

Limiting proliferation

Mike Clarke also thinks that driving differentiation might be a way to stall tumours. “Most cancer cells have limited proliferative capacity,” he said. “Cancer stem cells can make differentiated tumour cells” as well as more cancer stem cells.

He compared cancer cells that were able to drive tumour growth when transplanted in mice to those that were unable to and found that those expressing markers of mature cells could not cause tumour growth. Next, he explored the idea that the very cells that drive tumours are the ones that resist chemotoxic treatments. Indeed, radiation treatments largely spared breast cancer stem cells, whereas nontumourigenic cells were killed. Clarke found that both breast cancer stem cells and normal breast stem cells protected themselves from DNA damage in a variety of ways.

A separate set of studies conducted by Clarke looked at gene expression in human tumours; the more related the tumour was to a stem cell signature, the higher the rate of death and relapse. Clarke's group tracked this down to a set of microRNAs and has begun to identify their potential targets. One was expected: ZEB-2, implicated in neural crest stem cell functions. Another potential target, said Clarke, was “even more interesting.” This is a chromatin-remodelling protein (called BMI1) already implicated in self-renewal in multiple tissues. In one study of head and neck cancer, cancer stem cells expressed high levels of the protein, but the nontumourigenic cancer cells did not.

Tracing differentiation

The balance between differentiation and self-renewal is important in worms. Stanford's Andrew Fire, of RNA interference and Nobel fame, described alternative ways differentiation is controlled in worms. As the fertilized worm egg divides, a group of maternal RNAs are gathered into a single cell that eventually becomes the two germ cells. These RNA transcripts are actively degraded in the other cells, which are destined to become somatic cells. Although the future role of somatic cells can be shifted by moving their physical positions within the embryo, the germ cells's identity does not change. Other processes track only to the germ cells. In Fire's opinion, the worm's solution to separating germ and somatic cells is not to block transcription of certain genes, but to “not let somatic cells get enough information to become germ cells.”

Fire predicted multiple connections between germ cells, stem cells and tumour cells noting that reprogramming a differentiated cell into an embryonic-like state required the addition of oncogenes.

Understanding the regulation of tissue stem cells could help explain what goes awry in cancer. Stanford's Margaret Fuller gave a detailed description of the choreography in the Drosophila testes in which a hub of somatic cells keep a core of germ line stem cells in a state of self-renewal. Cell divisions in the testes hub always produce one cell that remains close to the hub and one that is pushed away. The cell closest to the hub stays a stem cell; the further one differentiates to produce spermatocytes. The orientation of the cell-division apparatus ensures this asymmetric outcome.

Knock out a gene called JAX, however, and the system falls to pieces. The first wave of cells differentiate normally, but then no more spermatocytes are produced. The stem cells fail to self-renew, so there are no additional spermatocytes after the first wave. “It's like turning off the tap,” Fuller says.

The protein encoded by JAX depends on the activity of another protein, a transcription factor named signal transducer and activator of transcription (STAT), and Fuller's team used a chromatin precipitation assay to figure out where this transcription factor attached itself in the genome.

The self-renewal switch

Irving Weissman, also of Stanford, described self-renewal as stem cells's “most fundamental property and most dangerous property.” He described his work homing in on the blood stem cells that could restore an irradiated mouse's blood system.

Conventional wisdom held that there was no communication between blood and bone, but this was contradicted by researchers's observations of leukaemia. Once established, the cancer can be found in the marrow of any bone in the patient, and the cells clearly descend from a single common ancestor. (In fact, leukaemia might elbow out the healthy cells in the niche, according to recent research.) If haematopoietic stem cells are infused into a mouse, they leave the bloodstream within five minutes, localizing to bone, spleen and liver. Working in mice, Weissman found that the niche could be cleared out using antibodies, allowing more successful transplants2. He proposed that stem cell migration could make cells besides leukaemia less tractable to surgery. Neural stem cells migrate, for example, and glioblastomas transplanted into animals move similarly to neural stem cells.

Next, Weissman described his work tracking leukaemia cells. Leukaemia stem cells arise from cells that normally function as multiprogenitors; these are not stem cells and are incapable of self-renewal, so they cannot normally generate multiple waves of proliferating cells. He has found evidence that a stage in chronic myelogenous leukaemia bestows an unregulated ability of self-renewal on the multiprogenitor cells, probably because the Wnt signalling pathway gets turned on. Most likely, a series of events occur so that β-catenin cannot be destroyed by the cells's machinery.

One necessary event is the activation of a marker called CD47 that prevents cells from being consumed by macrophages. This is a gene that is also turned on in healthy quiescent multiprogenitor cells that will encounter macrophages on their way to the bone marrow. In leukaemia, these cells can both avoid macrophages and divide repeatedly.

University of California, Los Angeles's Owen Witte wanted to understand how biopsies taken from areas of the prostate that are very close together can nonetheless show various cells in various stages from ordered growth to neoplastic growth to full-blown cancer. Mixing epithelial cells and mesenchymal cells can spur neoplasia in epithelial cells. To try to understand that influence between the cells, Witte's team started exploring the roles of fibroblast growth factors, already known to increase the amounts of the androgen receptor in epithelial cells. Not all of the growth factors had an effect, but fibroblast growth factor 10 did when expressed in mesenchymal cells. In some cases, cells from the induced carcinoma could start cancers when transplanted serially to other mice; this suggests a potential target for hormone-sensitive or refractory prostate cancer.

Tweaking apoptotic pathways

One way that the body prevents cancer is through cell death, or apoptosis of stem and progenitor cells that suffer unrepaired DNA damage. Two main pathways of apoptosis have already been firmly established, and Sam Sidi and Tom Look of the Dana Farber Cancer Institute, in Boston, think they may have found a more ancient pathway that is activated when cells are unable to stop cycling and repair damaged DNA. The two canonical apoptotic pathways converge on a highly conserved enzyme called caspase-3, whereas the new pathway relies on caspase-2. Sidi and Look worked with zebrafish lacking functional copies of a gene called p53, which triggers cell death in troubled cells and is dysregulated in perhaps half of all cancers. Using a type of antisense oligonucleotide known as a morpholino, Look's team knocked down the activity of a panel of genes that serve as 'checkpoints' in the cell cycle. For most genes, this made no difference in the developing fish embryos. But for one gene, encoding Chk1, knockdown caused the same pattern of death as in embryos with normal p53. Thus, Chk1 normally suppresses the new apoptotic pathway involving caspase-2. This implicates the new pathway in the death of cells that have defective Chk1 function and therefore would be susceptible to replicating damaged DNA and subsequent malignant conversion.

A small molecule that had similar effects against Chk1 as the morpholino was tested in human tumour cell lines lacking p53 that are able to survive despite high levels of DNA damage. The treated cells died, and caspase-2 was observed in its activated form. Hence the new cell death pathway can be activated by drugs that inhibit Chk1 that are now entering the clinic, which should provide a means to sensitize human tumour cells to undergo apoptosis after radiation and chemotherapy.

The basic message of the day was that the path from stem cell to fully differentiated cell follows three main phases, all of which are tightly regulated. To maintain a stem cell population, cells must self-renew; cells that go on to another fate divide a controlled number times to expand cell numbers, and these cells differentiate. Redundant processes and checkpoints keep these activities under control or destroy wayward cells. When these safeguards fail, cancer stem cells emerge.

Related articles

Cancer stem cells sightings and slightings