Cancer: the wound that never heals?

The interaction between the immune system and cancer is a complicated and puzzling one. On the one hand, there’s evidence that the immune system can help to get rid of tumors.  On the other hand, there’s also growing evidence that an inflammatory environment is important for tumor survival and metastasis.  A recent paper (Feng et al. 2010.  Live imaging of innate immune cell sensing of transformed cells in zebrafish larvae: parallels between tumor initiation and wound inflammation.  PLoS Biology, 8 e1000562) aims to model very early events in tumor initiation, and indeed finds that the immune system both attacks and helps early tumors.  It appears that transformed cells send out chemoattractant signals in much the same way that wounded tissue does, leading to infiltration by leukocytes that attempt to kill the transformed cells; but this attack is usually insufficient to “heal” the wound, leading to a chronic inflammatory state that seems to help the tumor grow.

Feng et al. used a transgenic zebrafish line in which leukocytes are fluorescently labeled with GFP. They injected a gene construct carrying an oncogenic form of Ras labeled with mCherry, under a couple of different cell-type specific promoters, into the fish embryos. Because zebrafish are transparent, they could then watch the cells that express Ras (which are labeled in red) in real time — in other words, watch as a cell is transforms into an early cancer cell — and see what happens as the GFP-labeled leukocytes (labeled in green) encounter the Ras-expressing cells.  They did this at a stage of development when only the cells of the innate immune system (such as neutrophils and macrophages) have developed, allowing them to separate the effects of this part of the immune system from that of later-developing cells such as T and B cells.

The first surprise of this work is how rapidly leukocytes find the cancer cells: both neutrophils and macrophages accumulate near Ras-expressing cells usually before they’ve had a chance to divide.  They show interesting behaviors once they’re there: the cancer cell and the leukocyte reach out towards each other with filopodia and lamellipodia, often creating “tethers” that link the cancer cell to the leukocyte.  Feng et al. can also see that the immune system cells behave as if attacking the transformed cells: macrophages, true to their name, do this by engulfing whole cells, whereas neutrophils chop off smaller bits. However, it’s not clear that these are real attacks; it’s also possible that the leukocytes were just sampling the cancer cells.  The authors didn’t see any apoptosis in the cancer cells, which you might expect to see if a real attack was under way.

Comparing the pattern of leukocyte recruitment to the cancer with that of leukocyte recruitment to wounds, Feng et al. saw strong similarities in how often the leukocytes change direction and how long they stay in the affected area.  They therefore looked for molecular markers associated with wound healing in their cancer model, and see several markers of leukocyte activation.  As I’ve written about before, it’s recently been shown that one of the very earliest events in the response of the immune system to a wound is the generation of hydrogen peroxide, which seems to “call” leukocytes to the site of the wound (this is the work of Phillip Niethammer, who has just left the Mitchison lab to set up his own lab at MSKCC and is looking for post-docs, if anyone’s interested).  Feng et al. therefore looked for production of hydrogen peroxide by their tiny cancers and — rather amazingly — found it.  What’s more, they were able to show that blocking hydrogen peroxide production (using diphenyleneiodonium chloride, a chemical that inhibits NADPH oxidase enzymes) prevented leukocytes from migrating towards the  cancer cells.  Phillip’s earlier work had showed that the enzyme responsible for hydrogen peroxide production in the wound healing context is dual oxidase, DUOX: knocking down DUOX using morpholinos also prevented leukocyte homing to the mini-cancers.

Where does the hydrogen peroxide come from, is it the cancer cell itself or the surrounding host tissues?  Feng et al. generated cancer cells from DUOX knockout fish, and transplanted them into normal hosts; conversely, they took “normal” cancer cells and transplanted them into DUOX-knockout hosts.  Remarkably, leukocyte recruitment was reduced relative to normal but not absent in both cases, suggesting that both the cancer cells themselves and the host cells produce hydrogen peroxide. It’ll be interesting to find out what the trigger for hydrogen peroxide production is, and whether it’s different in the cancer cells versus the surrounding host cells.  When the cancer cells are the ones that lack DUOX, the leukocytes seem to have some difficulty recognizing them; videos show the leukocytes moving around the cancer cells instead of over them. Maybe this indicates that some oxidized protein on the cancer cell is important for recognition.

Does all this immune activity actually slow down the growth of the cancer?  Apparently not — in fact, it seems to encourage it.  Feng et al. used three different methods to prevent hydrogen peroxide generation and/or leukocyte recruitment, and found that in each case the number of transformed cells in the developing tumor was smaller without the leukocyte recruitment than it was in controls — despite the phagocytotic efforts of the neutrophils and macrophages.  The proliferative index of the transformed cells also goes down in these circumstances, suggesting that the presence of the leukocytes provides some kind of growth-factor-like signal.

The idea that wound healing is related to cancer goes back to the 1860’s, when Rudolph Virchow noticed that cancers often show up at the sites of old wounds, and proposed that cancer might result from chronic irritation.  This paper provides evidence that there’s something in that theory, and perhaps a step towards identifying an inflammatory factor that’s important in very early cancer growth.  But like Phillip’s earlier work on the role of hydrogen peroxide in wound-sensing, this paper leaves many questions unanswered.  The first is how good a model of human cancer this is: the cells are transformed in the very earliest stages of embryonic development, using a single oncogene (Ras).  Tim Mitchison points out that embryos don’t get cancer — which is kind of interesting in itself — and that it’s quite possible that this model is telling us something about the role of Ras signaling in inflammation instead.  Even if that were so, these results would still interesting. The normal functions of oncogenes are often poorly understood, and it wouldn’t be surprising if some of them are involved in inflammation. Perhaps Ras is involved in wound healing and inflammation that isn’t related to cancer, as well.

Another open question is the issue of how the hydrogen peroxide is sensed by leukocytes — is the sensing direct, or via some intermediary, e.g. an oxidized protein?  And then there is the question of how activation of Ras triggers the production of hydrogen peroxide. DUOX is positively regulated by Ca⁺⁺, so perhaps this pathway is involved. But Ras is a master regulator of many pathways, so the DUOX connection could be quite indirect.

The big question is whether signaling to leukocytes via DUOX and hydrogen peroxide important in human cancer, and would DUOX be a good target for cancer drugs?  The answer is not at all clear, unfortunately.  In humans, DUOX is mainly expressed in the thyroid (where it makes the hydrogen peroxide needed to oxidize iodide for incorporation into thyroid hormone), the lung (where it is thought to help kill invading microorganisms) and the gut (where its function is unclear). Tim Mitchison tells me that there is currently no evidence that DUOX is activated in human cancer cells. However other potential hydrogen-peroxide-producing enzymes, such as the NADPH oxidases NOX1 and NOX4, might be.  If there are any chemists reading this, may I suggest that you consider working on identifying improved inhibitors of enzymes in the NOX family?  They might or might not turn out to be potential anti-cancer or anti-inflammatory drugs — though you could hit the jackpot if they are — but in any case they’d be very useful to help probe the fascinating and potentially important roles hydrogen peroxide plays in cancer and inflammation in animal tissues. (I should warn you, though, that many people have probably thought of this already.)

Feng Y, Santoriello C, Mione M, Hurlstone A, & Martin P (2010). Live Imaging of Innate Immune Cell Sensing of Transformed Cells in Zebrafish Larvae: Parallels between Tumor Initiation and Wound Inflammation. PLoS biology, 8 (12) PMID: 21179501