A new paper from me, Lea Goentoro, and John Doyle published this week in Proceedings of the National Academy of the Sciences. The paper is titled 'Proliferation as a natural strategy to suppress neoplasia.' It’s about cancer resistance strategies in jellyfish and math models that could apply in people. On the way to these results, we borrowed and built on the developmental-immmune biology concept of tissue homeostasis and the control theory concept of robustness.
Cancer resistance: How do some animals, like jellyfish, achieve low rates of cancer, and what does that tell us about cancer in people?
Tissue homeostasis: How do organisms build and maintain functioning tissue that withstands injury, infection, mutation, and other stressors across uncertain environments?
Robustness: What rules describe implementable systems, like tissues, made up of specialized rule-bound cells that move, coordinate, adapt, and replicate?
Cancer resistance in jellyfish is an old mystery, but some recent results made it timely: the discovery that cancer can occur naturally in Hydra, another cnidarian; and the discovery, in Lea’s lab, that moon jellies selectively regenerate depending on lab-modulable environmental conditions. The biology of our distant jellyfish, we thought, cousins could tell us something about the evolutionary origins of multicellularity, development, regeneration, and cancer.
The ease with which the model mapped to biological prediction probably had to do with special things about jellies as a system (no vital organs, no adaptive immune system, plastic body plan, indeterminate growth, cold-bloodedness, and the previously mentioned lab-modulable regenerative capacity). But the phenomenon described by the model, we realized, seemed like it might occur in any tissue.
The paper reflects this shift in emphasis. I think we arrive at a pretty good, if partial, answer for why jellyfish don’t get cancer, and we arrive at some interesting questions about human cancers. For context, in humans, proliferative tissues tend to be cancer-prone (lung, liver, skin, colon). A common, plausible mechanistic explanation has been that cell proliferation potentiates mutation risk and therefore cancer risk. Yet this explanation faces the obvious problem that cancer does not appear, across all animals, to be inevitable. We show that it is possible for cell proliferation to have the opposite effect, suppressing cancer, and that this is essential to how jellyfish achieve cancer resistance. We call this strategy proofreading [1]: making cells in excess to allow for quality control through cell death. Because proliferation and cell death are coupled, decreasing proliferation decreases the flux through this selection mechanism.
Do human tissues resist cancer the same way? The idea is at least plausible. It turns out that when the proofreading strategy fails, the tissues with the highest baseline proliferation rates will still be the tissues with the highest cancer risk. Because aging slows down proliferation, we would expect to see the same pattern of lifetime cancer risk, with a skew towards old age consistent with observational data. To some extent this simply relocates the question to why humans have a global slowdown in old age. But it also gives us more tools to think about the exceptions, like pediatric cancers. Rather than attributing these early incidence peaks to environmental exposures or inherited mutations, we can look for how the physiological state of the tissue interacts with environmental exposures or inherited mutations. This might lead to different specific development-environment interactions for each tissue type or cancer subtype.
Tissue homeostasis has been an intriguing conceptual meeting point for many lines of biology research over the past decade. What does it take to build and maintain a tissue, as a whole? In the paper, we dial in on cell population size as a simple proxy for this question. By making this simplification, we consider the developmental biology, or stem cell biology, perspective on tissue homeostasis: homeostasis means renewing the cell population that makes up the tissue by making more cells.
Much of the remaining biological work of tissue homeostasis is the work of the immune system. The same white blood cells that fight infections also orchestrate injury response, wound healing, scarring, cancer defense, even metabolic waste product disposal. In fact, many of the roles we attribute to white blood cells are also taken on by tissue-local (parenchymal) cells. A vivid example of this is the role of T cells in killing cancer: what the T cells actually do is trigger programmed cell death pathways in the cancer cells. The cell death pathway precedes the evolution of T cells by hundreds of millions of years, yet we credit the T cell as the killer. The moon jellies use this cell death pathway to resist cancer without any T cells at all. So the jellies, along with the brewing synthesis of tissue homeostasis in the literature, helped us see these two problems, that we might typically think of as a development problem and an immune problem separately, as two facets of one problem. Likewise, the results could help us think about development and immunology differently, because these distinctions matter to us more than they matter to biology.
The tissue homeostasis problem has other facets, too. Tissue growth, regeneration, degeneration, and fibrosis are all important and intertwined biological processes that complicate our simple notion of population size homeostasis. So are metabolism and infection and autoimmunity. We needed mathematical models to get this far, and I think a strong mathematical foundation is essential for these more complex steps.
Robustness is the ability of a system to maintain its structure and function despite uncertainty in the real-world environment, or relatedly the ability of a model of the system to maintain fidelity to reality despite uncertainty in the modeling assumptions. Robustness is a structural property of a system. There are many ways to model systems that can develop cancer, and they can look pretty different when they fail. But if you want to model a tissue that achieves homeostasis and suppresses cancer, you have to build that functionality into the structure of the model itself. Infinitely many variations on the model arrive at roughly the same place. Some parameters can vary by orders of magnitude. An experiment that breaks the robustness structure can act on any of several pathways. The same robustness structure might be achieved by different molecular pathways in jellyfish and humans. The model is really, really simple, the smallest possible departure we could come up with from existing models. But we have good reason to believe, from testing varied other models and from our engagement with robust control theory, that more complex models would say the same thing.
This kind of two-level thinking about model structure is fundamental to control theory and multiscale modeling. In control theory, it can simultaneously be true that there is a unique input-output relationship that achieves an objective and that infinitely many circuits could implement the relationship. Similarly, biology can be enormously constrained in function while remaining infinitely varied in implementation. By analogy, consider language and grammar: language is infinitely expressive, even to the point that it can accommodate entirely new words, but most sequences of words are incompatible with meaning.
We relied on a mix of theory, experience, and careful testing, which is roughly the state of the art of this kind of biological modeling. There are mathematical cases where we can (by hand or by computer) efficiently and provably determine whether a given structure is robust. If correct and applicable, these determinations are as constraining as, say, an understanding of gravity in determining whether or not a bird can fly. The set of all possible systems might take until the end of the universe to search, but the set of systems that can implemented in reality might be searchable in an afternoon. So the work of theory is to define the set of systems we should search. Because the true set of tissue-homeostatic systems is so complex, there is much more work to be done here.