Come in Cell 42, your time is up

The view that cancerous cells are old cells that accumulate mutations which restore the hazardous gifts of immortality and replication is yielding to a more parsimonious explanation: that cancerous cells are mutant stem cells, that are 'born' immortal.

Professor Michael Clarke, associate director of the Stanford Institute for Stem Cell and Regenerative Medicine, told the Lorne Cancer Conference recently that when stem cells divide, their daughter cells initially retain their 'mother's' multipotency and replicative capacity, but acquire the ability to differentiate.

As they progress through the various precursor phases leading to final differentiation, they continue to exhibit degrees of 'stemness', but their capacity to proliferate is no longer open-ended. At each branch in the developmental 'tree', the number of replicative cycles is reduced, and falls to zero in the terminally differentiated cell. "So stem cells have some kind of counting mechanism," Clarke said.

The developmental trajectory of mutant stem cells mirrors that of healthy stem cells, but like Superman's mutant alter-ego, Bizarro, it's a dangerous caricature of the original, according to Clarke. He said a single, normal haematopoietic stem cell can reconstitute mouse blood cells for the life the mouse. Clarke's team lethally irradiated multi-potent progenitor cells and transferred them into mice.

"They differentiated and died, except for a subset of lymphocytes - memory B and T cells, which are essentially stem cells," he said. "They had all the 'stemness' characteristics of haematopoietic stem cells, but their differentiation potential was limited.

"The major point I want to make is that the ability to proliferate is not the same thing as self-renewal - a [partly differentiated] haematopoietic stem cell can make millions of cells, but eventually it will stop. So when we talk about stem cells' capacity for self-renewal, we must ensure the assay fulfils the criteria self-renewal."

Rube Goldberg machines

The question Clarke is asking is why do we have this system? Why are there stem cells? "It would be much more efficient if, when we suffer sunburn, all of the sunburned cells could renew themselves. Superficially, it would seem much better for a multicellular organism to replace and repair damage. This Rube Goldberg system evolved partly to help prevent multi-cellular organisms developing cancer."

Professor Robert Weinberg of the Whitehead Institute, discoverer of the first human oncogene, had suggested that multiple mutations - "between six and 50, depending on who you listen to" - were required to transform a fibroblast into a cancerous cell.

"If you do the mathematics, if every cell in a tissue was as long-lived, and accumulated that number of mutations, all of us would be dead of cancer by now," Clarke said. "It has long been recognised that, in leukaemias, there is a higher-order leukaemic cell - a stem cell - that drives the tumour's growth. The notion that this also occurs in solid tumours is best demonstrated in a teratoma - a germline cell from the testes that has turned cancerous.

"An immature teratoma is one of the worst cancers known, and is uniformly fatal within months of diagnosis. A closely related type, called a mature teratoma, looks histologically similar to the immature type, but they differ in that only the immature teratoma cells express early markers. The immature teratoma spreads like wildfire, but the mature teratoma's growth is self-limiting - it forms a tumour about a centimetre in diameter, then stops growing."

Clarke showed a slide of a tissue sample from a cancer, stained to highlight cells with early-lineage markers. "You see only a few cancerous cells in a sea of normally differentiated cells. Could it be that there is a cellular hierarchy, just as in the blood system and in normal tissues?"

In this system, the cancer stem cell would give rise to the heterogeneous population of cells seen in the tissue sample, most of which would lack its capacity for self-renewal - but retained some ability to proliferate.

"Eventually, these cells will stop proliferating and differentiate, so they don't spread," he said.

Stem cell markers

When Clarke's colleague, Dr Mahomed Al-Hajj, used flow cytometry to sort cells from different regions of a breast-cancer tumour, he found that some cell populations were enriched with cells capable of forming tumours; others were relatively depleted. Using as primary markers of stem cells, they found that only about 200 cells among tens of thousands expressing CD44 and CD24 proteins on their surface - primary markers of stem cells - formed tumours in a mouse model of breast cancer.

Samples of remaining cells expressing fewer stem-cell markers - indicative of their progress towards differentiation - were incapable of forming tumours.

CD44 proved a reliable marker for isolating cancerous cells from colon, neck and pancreatic tumours. CD44 cells lacking another marker, CD133, were strongly proliferative. Expressing CD133 at a high level in these cells abolished proliferation.

They subsequently verified their ability to identify cancerous stem cells in their mouse model of breast cancer. "We reasoned that the ability to read the genetic signature of stem cells from surface markers might be clinically valuable," Clarke said.

"With the [CD44/CD133] signature, we were able to predict the outcome in patients with breast cancer. The more closely related the signature of the patient's tumour to the full cancer stem-cell signature, the higher the risk of them developing metastasis and dying. The more distant the relationship, the lower the patient's risk."

Gene expression

Clarke and his colleagues suspect that gene-expression patterns in some tumours are influenced by short-range interactions with the stromal cells in the connective tissues that surround and support the organs. Stromal cells may also actively guide differentiation of stem cells in organs.

"If you have fertile soil, and fertile seed - stem cells with cancer-derived signatures - there might be a synergistic interaction. We found this is indeed the case. Where patients tumours are relative unrelated to the stem-cell signature, and to the local stromal cells, there is a very low rate of relapse at eight years.

"If the cancerous cells' signature is related to that of either the stem cell or the stromal cell, there's an intermediate risk, and if it's related to both, there is a very high risk of relapse."

Clarke said there was even stronger evidence for interactions between normal cells and the tumour-cell population.

"CD44 is a very reliable marker for head and neck tumours. We found a striking association between moderately differentiated and well differentiated tumours, and CD44+ cell counts."

He said the polycomb protein BMI1 was absolutely required for the maintenance of many adult stem cells, making it reliable assay for stem cells - BMI1 was discovered by Professor Jerry Adams' research team at the Walter and Eliza Hall institute.

Clarke said cells immediately adjacent to tumours express high levels of CD44, as well as high levels of BMI1 for self-renewal. Cancerous cells more remote from CD44/BMI1-expressing cells lacked CD44 and BMI1 expression.

The Stanford Institute researchers then performed experiments to determine if other mouse models of cancer had a stem-cell component.

"About 10 markers are informative for tumours, and my feeling is that the reason we are seeing this differentiation in marker expression patterns in stem cells has something to do with the background of acquired mutations -the basic pattern is maintained in a large number of tumours."

Clarke said Walter and Eliza Hall institute researcher Dr Jane Visvader, whose research team isolated mammary gland stem cells in 2005, had identified a set of markers associated with normal breast cells. "When we use Jane's markers on our breast cancer cells, they all have high levels of CD29 expression - they resemble typical early progenitor stem cells in mice."

When Clarke's team isolated breast cancer cells displaying Visvader's diagnostic markers, and transplanted them serially into mice for five iterations, all formed breast tumours. Similarly, colon cancer cells with markers characteristic of early progenitor colon stem cells formed aggressive colon cancers in mice.

"If we take the 40 genes that are most highly expressed in cancer patients, they correlate with a poor prognosis. The 40 genes that are least expressed correlated with a good prognosis."

Progenitor stem cells

Clarke said that in chronic myeloid leukaemia (CML), normal and cancerous cells looked identical in chronic-phase patients. But when they took several billion cells from patients in terminal or blast phase of CML, and transplanted them into mice, they made a striking discovery: within a week, they vanished, and were replaced by what appeared to be progenitor stem cells for granulocytes and macrophages.

"How could this happen?" he asked. "Normally, blast cells can never go backwards and become self-renewing without a series of mutations. So we decided to experiment with a series of [knock-in] mutations in genes downstream of BMI1 in a transgenic mouse model of CML.

"We transplanted irradiated mice with progenitor stem cells that give rise to B and T lymphocytes, granulocytes and macrophages. The progenitor cells provide a long-lived engraftment, but myeloid cells are naturally short-lived, so they are lost after engraftment.

"We had earlier shown that BMI1 is absolutely required for normal renewal of haematopoietic stem cells - as a polycomb gene, its job is to repress expression of genes downstream from itself.

"BMI1 variously represses or activates genes linked to self-renewal, apoptosis, cell fate and differentiation."

Another colleague identified three very familiar tumour-suppressor genes - p16, p19, and p53 - as downstream targets of BMI1. Knock-in mutation in any one of these genes increased the frequency of stem cells in breast cancer tissues by up to twofold, but simultaneous mutations in all three increased stem cells by 12 to 15-fold.

Triple-mutant transgenic mice all had normal haematopoietic cell compartments, but the cells had a 10- to 20-fold increase in their ability to reconstitute bone marrow.

"We looked at the engrafted stem cells, and they were identical to the wild type. But when engrafted these multi-potent progenitor cells are capable of long-term engraftment. This suggests the three genes are part of a mechanism involved in a counting mechanism in multi-potent progenitor cells."

Clarke's group repeated the experiment in five triple-mutant mice, and again performed serial transplants of the resulting cells - all the transplants 'took' successfully.

"This suggests we have a cell population that normally lacks the ability to self-renew, but which is 20 times more abundant than haematopoetic stem cells. It's probably cycling faster than progenitor cells, so it can accumulate further mutations on the road to cancer.

"But the involvement of at least three oncogenes in cell renewal suggests no single pathway, if released by mutation, will allow a cell to renew. Only if they accumulate the right combination of mutations to you get a population that can be targeted by further mutational events.

"Mutations that release BMI-1 repressed genes can easily be overcome mechanistically by cells. No single pathway can drive self-renewal if a gene over-expressed or knocked out.

"Multi-cellular organisms had to evolve mechanisms to prevent that happening, or we wouldn't be here."