A new Nature paper makes the striking claim that neural precursor cells of the brain migrate all the way to prostate tumors. Furthermore, the team led by Claire Magnon claims that these long-distance cellular travelers from the brain not only take up residence in the prostate but also strongly drive progression of the prostate cancer via neurogenesis there.
From what I know, this claim goes counter to general principles of animal physiology so if it’s accurate, it’s a huge deal. Other than blood cells, most cells in certain tissues don’t just wander far off to other distant tissues even in disease states. Still, in science we need to challenge general assumptions and cells do sometimes wander.
The paper, Mauffrey, et al., entitled “Progenitors from the central nervous system drive neurogenesis in cancer”, is mostly based on a mouse model of prostate cancer. The tumors are driven in these mice by exogenous Myc, although the team also analyzed some human prostate tumor samples in correlative work.
The claim of neural precursors globe-trotting from the brain to prostate is not addressed in the human context, but the authors seem to think it also occurs there. Since most of the paper is on mice, we need to invoke #inmice here somewhat. If true in humans too, this is an even bigger discovery.
As a Myc, cancer, and stem cell researcher, this paper caught my attention. I also am a prostate cancer survivor myself.
I’m somewhat skeptical of the claims here and I think that the authors went out on a limb with how strongly they made their claims. In particular I strongly doubt that this occurs as part of the normal disease process in human prostate cancer development. It’s not that I see any technical red flags on first glance in the pub, but rather the core claim just doesn’t make sense to me. Of course, that doesn’t mean it’s wrong, but I’m surprised it hasn’t made more press.
So what’s going on here?
The paper largely relies on the assumption that a marker of central nervous system (CNS) precursor cells called DCX definitively marks only these CNS cells. Thus, when the authors see DCX+ cells inside prostate cancers (which is convincing), they claim that they must come from brain. However, for me a much simpler explanation is that instead local cells take on a DCX+ fate or perhaps a few already have it such as in a prostatic nerve fiber (note that DCX+ cells are found in the peripheral nervous system too) and are attracted into the tumor locally. I don’t know that the paper rules out these simpler possibilities so that’s a concern.
Even so, let’s say the paper is right that the prostatic tumor DCX+ cells are from the brain, how would brain cells get all the way from the brain to the prostate? That’s like going from Earth to the Moon, in cellular terms. The authors can detect a few DCX+ cells in the blood of their prostate cancer mice, but not many. Still, they argue that the cells go through the blood-brain barrier (on its own, a big hurdle) and then all the way from brain to prostate cancer in the vasculature. That’s possible, but surprising.
If this does happen in people, when do the brain precursor cells go to the prostate relative to the tumor initiation and growth? If the brain cells go there prior to tumor initiation, then why do they even go there in the first place? If as seems more likely they go later, perhaps attracted by the tumor, how do they contribute to the progression of the already existing tumor? The authors argue the brain cells definitely contribute to the tumor’s progression. They also suggest brain cells go to other kinds of tumors, at least in mice.
This is one of those papers that strongly shakes things up. I saw an accompanying News & Views piece focused on it, but unfortunately it doesn’t really discuss limitations of the work or ask probing questions.
In a way, I hope I’m wrong in being so skeptical. The new finding, if replicated and more definitively proven, would be very cool and I like it when there are paradigm-challenging or breaking discoveries. Still, for me at least, it’s going to take a lot more proof to be convinced.
Just saw this paper. My first reaction was that they are using Tamoxifen inducible system.
A system based on Estrogen receptors in a prostate cancer study? This chemical is a well established Anti-cancer agent and they give animals 5x the dose used for cancer treatment in patients?
On-top of this. The methods are written in such a way it is ambiguous if the control mice where even given tamoxifen.
“DCX-creERT2;loxp-EYFP Hi-MYC mice were injected at weeks 3 to 4 after birth”
In the end I found a less than convincing control experiment tangentially adressing the cytotoxic effects of Tamoxifen, albeit with an n=2. This was very likely requested by a reviewer which to me suggests the problem was raised.
In my book, this one is on nature.
Dr. Knoepfler, I love your articles. This one reminded me of the journal club I had during my PhD days. I wonder if lineage tracing is possible to indicate some information to the comments above?
Thanks!
The authors did some tracing via fluorescent protein viral transduction in the brain. It’s just not clear to me how clean that kind of system is. I don’t know if you mean more of a genetic system, but that came to mind for me too like incorporating a reporter into specific loci that would turn on only in brain cells. Maybe that could have drawbacks as well.
I highly doubt the veracity of this paper, it is exactly as you said, over-reliant on a single marker.
Now I happen to know from my wife, who currently works on single cell prostate cancer analysis, that she finds a lot of different cell types in >1 million single cells from ~50 prostate biopsies using CyTOF, and she did not find any neuronal cell clusters. Given that the CyTOF marker panel is limited (and obviously no DCX marker was used for prostate samples), it is possible that the method would not allow to find neuronal cells. However, given that amongst the million single cells she observed, she did not find unexplained clusters, meaning that all cells roughly map towards either tumor, epithelial or immune cell clusters (and subtypes of those). No neuronal cells. She could however show that prostate cancer cells have an increased dysregulation of protein expression, meaning their proteomic identity tended to drift. So my money would be that it is a fluke abarrent expression and the editors/peer review really messed up.
My wife’s paper is currently in preparation so it will still take some time before it is published (Nature Med.).
I am a pathologist (and a prostate cancer survivor). My default interpretation of this sort of finding is that some prostate cancers may aberrantly express this marker. Aberrant expression of various markers is not uncommon in cancers. The authors need to show that the DCX positive cells in these cases are indeed neural precursor cells and not something else.
Also, since DCX expression is found in the peripheral nervous system, in humans, at least, perineural invasion by prostate cancers is common, so that is one possible explanation of this finding. As always, more research is needed. It sounds to me that this paper’s conclusion my be unwarranted.
I don’t know why the paper didn’t address the points you raised.
They did. Clearly neither the author of this article nor mdfinfer read the article. In the study, they steriotactically injected the SVZ with TdT expressing lentivirus and showed that the DCX+ NPCs found in the prostate tumor express tomato (thus SVZ neural progenitors egressed and established home in the prostate). Sadly, this is an example of the all-to-common tendency for people (e.g. mdfinfer and this article’s author) to come to their own conclusions based on exactly no data.
I read the article, but I think there could be other explanations for what is reported such as artifacts.