It is quite unlikely that the word stroma will get many NNers excited, even those that are interested in cancer research. That is a pity since stroma plays a big role in many cancers.
One intriguing hypothesis is that stroma might be subject to evolution just as tumour cells are. One interesting type of stromal cell, myofibroblasts is rarely found in normal tissues but appears during wound healing. Myofibroblasts are also commonly found inside tumours and it is assumed that carcinomas leverage the wound healing response promoted by the myofibroblasts to sustain tumour growth.
An article (Nature Genetics, Vol 40 5 494-495) from Robert Weinberg in the 8th of May issue of Nature Genetics discusses precisely this topic. The idea that stromal and tumour cells could coevolve is very attractive to anyone interested in studying and modelling evolution. To many cancer biologists it also makes sense to think that recruited stromal cells might alter their genotype and phenotype in response to signals from the tumour cells.
For stromal/tumour coevolution to make sense we would need at least two pieces of evidence: that stromal cells do change their genotypes and phenotypes as the tumour progresses and that they do so in a way that makes some of these new phenotypes fitter and more capable of growth.
Unfortunately, the same issue of Nature Genetics carries an article that suggests that the first requirement does not happen, that is, the authors did not find genetic alterations in carcinoma associated fibroblasts. This result questions previous research that found genetic changes in carcinoma-associated fibroblasts. Furthermore, according to Weinberg, there are technical differences between the previous studies and this one that make this one more trustworthy.
Weinberg concludes the article saying that stromal/tumour coevolution, were it to be true, would complicate our view of tumour pathogenesis. I would add that it could also add a new agent in which we could act to alter the evolution of a tumour towards malignancy.
Great post David. I wonder though whether the stroma is really an appropriate target for pharmaceutical intervention as you suggest in the final paragraph. It seems to me (and I am by no means an expert!) that it would be very difficult to selectively target the abnormal stromal cells at the tumour site without causing wider systemic effects in healthy tissue. This is obviously also a problem when targeting the actual tumour cells, but in the latter case things like accelerated cell growth and defective DNA repair pathways make it somewhat easier to target only the malignant cells.
David, I think that first sentence was a set up for a response from me, but it can be more complete after the deadline on Weds!
In the context of Weinberg’s comments (and what I assume is the paper from Polyak’s group – no time to check right now – even reading this is procrastinating, let alone my perpetual and worried checking of realclearpolitics and fivethirtyeight – McCains chance of winning tomorrow currently estimated at 1.9% on 538 – which is a lot higher than I’d like!, Nate Silver, who runs the site is the guy who said, at the beginning of the season, that your own Devil Rays would make it to the World Series, got to give him kudos for his models). Sorry diversion!
I would note that while there was no evidence of genetic alterations in the stroma (at least per Polyak – other people have different opinions) this does not preclude epigenetic changes or phenotypic changes. Evolutionary selection, of course, takes place at this level, not at a genetic level. In terms of cellular morphology there is obviously a lot of plasticity – since all of the cells in our body have essentially the same genotype (except for a tumor I guess).
Anyhow this allows for functional evolution without mutation in this context – something that we need to talk about more. Getting back to Cath’s comment, I think that in some ways these interactions can be targeted, for example we have data suggesting that interactions between different stromal cell populations result in elevated SDF1 – which is essential for tumor progression, at least in some models. So this would suggest that a knowledge of the essential pathways acting between stromal cells might allow treatment of a genetically stable set of cell populations. So I think stroma does represent a good treatment target – but we need to understand the pathways better.
Got to go – 36 hours to finish this thing! See ya.
Cath: Thanks for your comment, When I made that statement I was only thinking that, by having more potential targets, it should be easier to find a weak point in tumour progression. Of course more complexity does not necessarily translate into more fragility but, via redundancy, it could also make tumours more robust. Still, I was reading tonight the following article from a group of people including Mina Bissell in The Scientist (quite readable and written for a non specialist readership like myself) and they also mention that as the tumour could be dependent on local factors that are extrinsic to the tumour cells themselves, the microenvironment could be a target for novel therapies. They mention antiangiogenic therapies as examples of such therapies although I believe that the promise they hold has yet to be delivered.
In any case, if there’s (Darwinian) evolution at the level of the stroma, it is likely that it will be proceeding at a slower pace than that of the tumour population and thus present an easier target.
Simon: I have to admit that I was thinking of you as I read Weinberg’s take on Qiu et al. (and yes, Polyak is in the author list although the last names are Izhak Haviv and Ian Campbell). I know that you advocate the idea of stroma coevolving with the tumour.
And you bring up a good point. Classical Darwinian evolution requires that the variation be inheritable but there’s room for some coevolution given that stromal phenotypes are quite plastic (capable of different behaviours in response to the environment) and that (correct me if I am wrong here) there are non-genetic mechanisms that could lead to the same result (I am thinking of DNA methylation here).
By the way, I hope that you have a good part of the grant finished and are missing only the finishing touches, I have the feeling that you’ll be busy tomorrow with slate.com, politico.com, news.bbc.co.uk, salon.com, huffingtonpost.com, thedailybeast.com, nyt.com, dailykos.com…. (I also know how to procrastinate in times like this!).
Good post and discussion. I must look into this paper now.
A wound that never heals…it has been said about Cancer.
@ Cath: if myofibroblasts are recruited in wound healing as well as in tumor tissue, there’s got to be some signaling involved with that. Interfering with the way that myofibroblasts get recruited at the site of the tumor, where they can support its growth, may open up an additional therapeutic handle. And it may be causing limited damage to normal tissue, as long as myofibroblasts happen to be very rare in normal tissue, and the signaling pathway to shut down is…unique.
Thank you for posting this, David! I am often referring to your content when I write on my italian blog.
Hola!
Remember, folks, that some forms of stroma give rise to mesenchymal stem cells (drop me a line if you don’t have access to any of these and would like me to give you a hand).
Pericytes, present in any vascularized tissue, can differentiate under the right circumstances into many things but firstly into cells that are indistinguishable from myofibroblasts.
So I think that tumoral signaling to endothelial cells, which has received a lot of research attention, has drawn eyes away from the signaling that must also be happening to the pericytes in the vicinity, which are poised to respond in all sorts of inappropriate ways.
I would maintain that their response is not “evolution” but rather “differentiation” – the phenotype of some stromal cells will change, and perhaps be selected for in the new tumor environment, but their genotype probably will not.
Massimo: Thanks for your comment. According to the article (from Bissell and co) tumours behave in a way comparable to drug junkies looking for the next fix from stromal cells. If Carcinoma Associated Fibroblasts are the dealers (and the comparison is reasonable), then any attempt to find how to target them will hardly be a waste of time.
Heather: There are some good articles there (although I could not get hold of the second one unfortunately). I understand that differentiation is likely to play an important role but my question would be if this differentiation can only proceed in one direction. If tumour cells can affect the way pericytes differentiate and if the range of phenotypes that can be obtained through this differentiation is not too small I think we could be able to frame the entire process as a coevolution of tumour with stroma. The usefulness of this approach would of course be limited by the capability of the cells to present a diversity of phenotypes (maybe via differentiation as you suggest).
I thought, though, that the major weak point of the argument is “they continually alter their genotype and phenotype in order to accommodate the needs of their ever-changing neoplastic neighbors.”
As a developmental biologist, I am quite used to the concept of cells inducing other cells to change phenotype. It’s the basis of embryonic development. For this, two conditions need to be met: the inducers need to emit signals, and the responders need to be competent to respond. We know the responders are, and the inducers probably do after their genetic alterations. I see no need for the responders to also have genetic alterations, although perhaps some epigenetic marks may be moved, as in development.
Hi Heather, thanks again. I like the idea that you don’t need necessarily genetic mutations but can tap in the normally not used parts of the gene-encoding DNA to alter the regular phenotype. In a stricter sense of the word you still need mutations to obtain open-ended evolution but there’s probably a lot of room for coevolution just allowing stromal cells to respond to signals from an abnormal environment.
Agreed overall, if the term “evolution” means “phenotypic change transmitted over cell generations in a new context”. While only some non-tumoral cells are poised to be multipotent responders (the pericytes, in my opinion), they share an identical genotype with the endothelial cells or any other still-normal cells in the environment of the tumor.
However, getting back to a more historical view of evolution and equating cells with organisms, it is only necessary that individuals with naturally occurring heritable variations that increase their chances for reproductive success (read, survival in the tumor environment) will produce progeny that inherit some of these variations. These progeny will end up representing a greater part of the genotypically normal cell population. In order to meet the criteria for the definition of evolution I carry around in my head, these changes could be epigenetic, but they should be passed on to future cell generations (eg. if you place a tumor pericyte back in a normal vascular environment, it should have lost its competence to revert to a completely normal phenotype). And I am also not sure that it is either genetic or epigenetic variation that is behind a differential response, but quite possibly it’s also a function of distance and cell density in the tumor microenvironment.
A very interesting line of thought, anyhow; thanks for bringing it up.
Heather,
You are right, this ‘special’ view of evolution requires something to act genetic material instead of the genetic material and accept the limitation that all the potential variation is reduced to combinations of the existing genetic possibilities.
So if we could abstract the process somewhat, assuming that the co-opted stromal cells have identical genes and that they lack genetic instability (as opposed to tumour cells), the epigenetic factors that make some of them adopt a particular phenotype are really what makes evolution possible and it is the transmission of those factors from the immediate microenvironment of a cell to the microenvironment of the daughter cell what would equate to the inheritable variance that is required for evolution to take place (together with selection of course).
I also think that this line of thought is very interesting so thanks for spurring the conversation in this direction.