Feature: New ways to test an old idea

Faced with the plethora of data in the field of cancer biology supporting three apparently solid theories for how a melanoma might maintain and propagate itself, Dr Mark Shackleton and his colleagues at the University of Michigan set out some years ago to test these models, starting with the CSC idea, using an innovative model-based approach. They decided to start by improving the existing and classic assays of injecting melanoma cells into NOD-SCID immunocompromised mice.

“Much to our surprise, we found that a number of relatively minor assay modifications to the commonly used technique made a really big difference to the efficiency with which we could engraft human melanoma cells into these mice,” says Shackleton. For instance, instead of terminating the observation period at eight weeks after injecting the tumour cells into mice, Shackleton’s team continued to monitor the mice for up to six months.

“Unexpectedly, the vast majority of tumours that formed in the NOD-SCID model occurred after the eight week point. So just by waiting for a longer period of time to allow tumours to develop, the frequency of tumourigenic cells was roughly 10-fold higher than the one in a million or so previously reported.”

This was a fairly significant increase just by doing virtually nothing but feeding the animals and watching the tumours develop.

Buoyed by their quick success, the researchers continued playing around with assay conditions and in the end found two more that made a substantial difference to the engraftment success for human melanoma cells.

One involved using a slightly different mouse model – the more highly immunocompromised NOD-SCID IL2Rgamma-knockout mouse – and the other involved adding an extracellular matrix formulation called Matrigel to the cells prior to transplantation. Both these conditions had a highly significant effect on the frequency of tumour development from the injected cells.

“When we combined all three assay modifications – waiting longer, using the NOD-SCID IL2Rgamma mice and pre-mixing in Matrigel – we showed about a 5000-fold increase in our ability to detect tumorigenic potential within cancer cells, such that our average empirically determined frequency of melanoma-producing cells was in fact one in four! So we started off with about one in one million and ended up with one in four.”

Not surprisingly, these results (Nature, 2008) entirely altered the biological understanding of melanoma. Although not necessarily disproving the CSC model, its relevance in the context of that particular disease is far reduced because the tumourigenic fraction becomes almost identical to the entire tumour.

“Consequently, there are now serious questions as to whether human melanoma follows the CSC model,” Shackleton said. “It also raised questions for many of the other similar findings from other human cancers as to whether there may be aspects of the particular tumourigenesis assay used that are clouding our interpretation of the true biology of the cells and their underlying true malignant potential.”

Armed with their new and improved assay for studying how melanomas progress once they have formed, Shackleton and the team then moved on to test the tumour plasticity model. “This involved deriving tumours in our modified mouse model from cells purified using flow cytometry from phenotypically distinct subpopulations,” he said.

“For instance, from a given tumour, we could separate two cell populations each expressing a specific marker, A and B, and then establish tumours from those populations, and let’s say that the primary tumour showed about 50/50 expression of A and B.

“So, in our model system, the secondary tumours derived specifically from either A or B cells revealed an almost identical spread of expression, irrespective of their cell of origin. So marker A cells could make A or B cells, but also marker B cells could make marker A or B cells.”

This result was further verified by testing over 20 different markers, and in each case, the frequency of tumourigenic cells within those populations was very high and consistent with the previously established average of about one in four cells. In addition, the expression heterogeneity or profile within the secondary tumours was very similar to that within the primary cancer.

“These findings (Cancer Cell, 2010) demonstrated quite clearly that no hierarchical relationship existed among cells defined by those markers within the parental tumour and that melanomas at least possess significant tumour plasticity.” Essentially, this means that a very high proportion of melanoma cells have the potential to propagate and spread within patients.

What about clonal evolution?

Based on their new assay results so far, Shackleton and colleagues further reasoned that single cells isolated from those tumours showing a one in four malignant potential could grow tumours with about the same frequencies when transplanted into mice, and they shows that this was, in fact, the case.

“We have now transplanted several hundred single cells from multiple melanomas obtained straight from patients. We can now take multiple single cells from the same tumour sample and essentially compare sibling clonal tumours all derived from the same parental tumour to address all sorts of questions – biological, genetic and epigenetic.

It is a painstaking and time-consuming procedure, but it is also pretty cool, because by comparing the malignant and molecular properties of sister clonal tumours it enables us for the first time to study human cancer biology at the clonal level.”

The idea of clonal evolution in cancer biology basically predicts that single or multiple separate clones can evolve genetically in a sort of Darwinian-style process.

Thus, they acquire additional genetic mutations that alter their intrinsic biological behaviour and within those sibling tumours there should be some clones that are genetically and biologically distinct. “Our assay now allows us to test that directly,” Shackleton said.

“This is the real focus of our current work in Melbourne, and we are starting to generate some pretty interesting data using genetic comparisons of sibling clones.”

For example, they have identified clones that seem to have an increased metastatic potential compared to other siblings and that seem to be genetically distinct. Basically, Shackleton’s approach is enabling a whole host of results to be obtained for the first time by analysing the sibling tumours.

In continuing to study clonal evolution and by developing more clinically relevant models of disease progression, Shackleton hopes to identify new biomarkers of disease and specifically, of malignant progression, and to identify new and improved therapies for the treatment of melanoma, helped along by the $1 million in funding from Pfizer Australia Research Fellowship.