John Dick: Careful Assays for Cancer Stem Cells

by Monya Baker

The University of Toronto scientist calls for more controversy

John Dick identified the first cancer stem cell, in leukaemia. The widely used xenotransplantation assay that he developed can confirm the identity of prospective haematopoietic stem cells by demonstrating their ability to re-establish a human blood system in the mouse. He is a professor at the University of Toronto and its affiliated Princess Margaret Hospital and Director of the Program in Cancer Stem Cells at the Ontario Institute for Cancer Research. 

Tell me about the xenotransplantation assay.

Lucky for us that there is enough complementarity between what human stem cells need and what the mouse can provide so you can get human repopulation of the mouse's blood system.

In the blood system, there are a lot of steps that a cell has to go through to read out as a stem cell. You take a source of stem cells, you put it in a syringe and you squirt them into the blood stream. Essentially you are taking advantage of a beautiful property of these cells, which is [their ability] to find their way to the bone marrow. 

You can imagine that you could have a cell that's a perfectly good stem cell, but if a cell has lost one homing marker, it wouldn't go through these heroics and it would never read out in the transplant assay. It's a problem that plagued us [until we realized that] if we are already extracting bone marrow by tapping the femur of the mouse, we could do an interfemoral injection; that turned out to be really powerful in that it bypasses the heroics, and it's a more sensitive way to test stem cell function.

How can a stem cell scientist know whether to trust an assay?

What you really want to ask is: Will every stem cell in my test mixture read out in this assay and will every cell that's not a stem cell not read out? Look at [Sean] Morrison: just by changing the environment, he went from 1 in 1,000 to 1 in 4 [sampled cells reading as cancer stem cells]. He read out more of their stem cell potential. [Editor's note: See Cancer stem cells, becoming common.]

We ended up calling the cell a repopulating cell rather than an HSC [haematopoietic stem cell] just because we didn't have a full enough picture of what the cell did to really call it an HSC, and it seemed more cautious and perhaps more informative to define the cell based on what it did — which was repopulating a SCID mouse. And then we could apply whatever criteria, like this is a repopulating cell that will last at least 10 weeks and will generate these types of mature cells, and so you build layers on top of the definition. We hoped that that would eliminate some of the controversy about language. So people could start to agree about where they were similar and where they were different.

The blood field has a lot of squabbles. Is this why?

A problem in the field is that people don't ascribe to the same criteria. Two people stand up and talk about a stem cell, but one of them is a short readout and the other one is a long readout, and they can't agree on what the properties of those cells are. Essentially it's a language problem. We want to use shorthand. We use these words, 'stem cell' and 'progenitor cell', but the nature of the cell is essentially defined based on the assay by which it's been identified.

How did you move from studying blood stem cells to leukaemia stem cells?

One can't dissociate the study of leukaemia stem cells from normal blood development. It's one of the first tissues that was well studied. We had these assays [for normal cells], and we thought we might as well apply them to leukaemia.

[Editor's note: Others had previously established that quickly proliferating leukaemic cells are generated by more slowly dividing cells. To identify leukaemia stem cells, Dick used a xenotransplantation assay to show that a subset of these cells can regenerate the disease.]

Tell me about the reaction to your 1994 paper.

Benign neglect. It didn't have a lot of impact. People in the blood field were satisfied with that result. At least for the people who thought about it, it was a natural progression of what people had thought about from the previous work. There were these proliferative cells that first of all were assayed based on the colony assay, developed in the 1970s. We knew from the normal blood system that the colony assay did not read out a stem cell. And this was essentially saying there's another cell, that there was a hierarchy.

What does the field need to move forward now?

It needs controversy.

The field of cancer stem cells needs controversy?

That's a little tongue in cheek. But it does. Controversy sparks better and better science. What it does is it actually eliminates sloppy thinking. There's been a real rush onto the cancer stem cell bandwagon in the last couple of years. People are talking about cancer stem cells here, there and everywhere, and in any old cell line. There was a huge slippage in the kind of criteria and rigor. People were using this terminology without any thought or any rigor based on some cell-surface marker or something like that.

There was a lot of hype of CSCs [cancer stem cells] being the answer to everything. Now we are in the phase of asking, "How valid is it? How universal is it?"

People don't like to do the tough assays, the three-month assays that you transplant in an animal for tumour formation. It's much easier to go on a flow cytometer and find a marker and say it's a cancer stem cell marker and say "now I'm studying cancer stem cells."

But we know that differentiation gets mucked up [in cancer]. In normal subsets, you can have four or five markers and you know what that cell is. In cancer you have to be careful. In cancer more than anything, you won't be able to get away from doing these cumbersome assays.

You also can't sit down in your basement cowering in the dark, saying you can't do anything because there's a problem with everything. You have to go forward. You have to say, "This is the result that I found, and this is how I did my assay, and this is what I think I can interpret, and this is what I think I can't." We just have to be cautious. It's how progress gets made, just being very circumspect about what an experiment is telling you and what it isn't.

What's going to be the next tool that will be as useful as a FACS machine or a xenotransplant assay?

The short answer is doing biology on single cells. In the context of tissues and tumours, cells are heterogeneous, and we have to deal with that heterogeneity. The days of growing up a batch of cells and cracking them open and doing a bunch of biochemistry to work out pathways — that's brought us where we are today, but if we're interested in pathways or how molecules can interfere with pathways, we need to know how a single cell is responding.

How can understanding heterogeneity help with cancer therapies?

For years, the goal has been to achieve the highest tumour-kill frequency. The problem is some cells are more potent than other cells, and on top of that these cells have mechanisms that make them more resistant.

The whole idea of a cancer stem cell was raised to explain heterogeneity. If a tumour isn't heterogeneous in some function, the CSC model isn't wrong so much as irrelevant.

If you're saying that all cells can acquire these stem cell properties, that's really supporting the stochastic models — that's saying that depending on the external or internal environment, any cell could be a cancer stem cell. If there's randomness to it, you're left with the idea that every cell has equal potential, and that's a fundamentally different concept from the hierarchical model where it's more intrinsic to the stem cells.

Why does it matter if there's a hierarchy?

A hierarchy implies that you have some maturation process. You have a stem cell, you have transit-amplifying cells and [then] some kind of blood cell. If you have a hierarchy in the cancer, it implies that these stem cells are actually differentiating, they are doing it abnormally, and you can imagine over tumour evolution that the differentiability of the tumour cells becomes more and more impaired. It's possible that, with tumour evolution, the distances between the CSC to transit-amplifying to mature cancer cells might be relatively low compared to the bulk tumours, and the shape of the hierarchy becomes rather shallow. Over time a higher percentage of the tumour is cancer stem cells.

So cancer stem cells themselves are heterogeneous.

I think we need to add on another layer to the definition of those cells because there are two kinds of heterogeneity. LSCs [leukaemia stem cells] and maybe solid-tumour CSCs can have different repopulation potentials, and this is linked to differences in self-renewal potential. This is why there are long-term and short-term repopulating normal HSCs. That concept hasn't been well-entrenched in people who study normal solid-tissue stem cells, so it's easy to see why people who study cancer just assume that a cancer stem cell is a homogenous population.

Another kind of heterogeneity comes from CSCs' evolution. I don't think they are a static population. They could evolve to become more and more abnormal. I think that we ultimately need to have a hybrid model to accommodate that idea, and that doesn't mean that we have to say that the cancer stem cell model isn't true, but it's a more nuanced view.

There are a lot of tumour systems where you know a lot about their evolution, so ideally you'd like to get a sample of those at different stages and ask what the frequency is and what the properties are of their cancer stem cells.

What will be interesting then is to test tumours early and late in their evolution. Is there a difference in CSC heterogeneity during tumour progression? What would melanoma look like before it became metastatic?

Can the environment make a cancer stem cell?

This is where it's important to have an assay system where you have a microenvironment that is suitable to read out all stem cells.

When you purify the cells into different fractions, and you can capture all the stem cell activity in one fraction and not in the other fractions, that is the key evidence needed to support that idea [that cells have different potentials to form tumours].

[In the work that identified colon cancer stem cells,] we fractionated cells on the basis of CD133 expression on the cell surface [and put the cells in the kidney capsule], and we didn't see colon cancer arising with CD 133neg cells, only with CD133pos cells. But maybe the renal capsule isn't supportive of CD133neg cells and they died upon transplant. Se we looked at animals injected with the 133-negative cells, went back many months later, and found the cells were still alive but they never made a tumour. Of course it is still possible that if given a more supportive environment, maybe they would form a tumour - that is the unknown.

What do you think of all the similarities that are cropping up between stem cells and cancer stem cells?

The blood world and haematology/oncology developed independently from the solid-tumour world. These communities have grown up relatively independently, and stem cell ideas have been more entrenched in the blood system.

Most of what we know is coming from AML [acute myeloid leukaemia], and AML may be unique in some of the behaviours. In AML, there's biologically a lot of similarity between regular stem cells and AML stem cells. The cells are quiescent. We know through clonal tracking that they have long-term and short-term cells, they regulate their self-renewal machinery, they are still niche dependent. That there are stem cell similarities makes sense. What's interesting about Somervaille et al is that, for certain kinds of oncogenes, the stem cell program they pick hearkens back to an embryonic program rather than an adult program.