Cancer cells grow frantically and compete for limited space and energy. Resources diverted toward a specific goal take away from the resources dedicated to others. A system, such as a tumor mass, reaches an optimal state when no individual can be made better off without making others worse. Tumors optimize their gene expression by trading off non-essential tasks.
Evolutionary thinking is a powerful tool kit for studying development and disease. Prominent areas of relevance are the development of functional organs and the mismatch between their initial conditions and the current environments and lifestyles. These are much-explored research topics with important implications for health. However, an ever-pressing subject is the emergence of cancer from seemingly normal cells and the diverse behavior of tumors once they arise. It's not immediately evident that there is a strong connection between the two or that applying evolution to cancer study can be beneficial. Therefore, to look at cancer through an evolutionary lens, one has to make a plausible case for its applicability and suggest potential benefits for this way of thinking.
I want to turn to an interesting study from Uri Alon's lab. The study relied on a large dataset of gene measurements of cancer cells from different tissue. Based on these measurements, the study's authors could suggest several goals or tasks that each tumor strives for. They then classified various tumors into specialists and generalists based on their gene expression profiles using multi-task evolution. Specialists are the ones that optimize for a small number of tasks. On the other hand, generalists don't specialize but comprise profiles of gene expression that do not correspond to particular tasks above others.
Does evolution apply to the development of cancer?
For evolutionary advantages to count, two conditions must be met: a high enough growth rate and competition for limited resources. Growing tumors satisfy both needs. First, cancer cells grow at a frantic pace, especially at the beginning. The second condition is more subtle. Even though it appears to disregard all constraints, limitations of available energy and space limit how much a tumor can grow. Moreover, tumors have to prioritize functions other than endless growth at different stages or locations of the disease. Cancer cells with distinct functional profiles can therefore compete for the limited space and energy.
Cancer cells grow frantically and compete for limited space and energy.
Trade-offs between the cancer functions
Multi-task evolution is an appropriate way to describe the dynamics of competing cells. Cells cannot strive to multiply, invade other parts of the body or evade the immune system at all times. Consequently, any resources diverted toward a specific goal take away from the resources dedicated to others. This situation results in tumors concentrating on some tasks and only maintaining general tendencies toward others. At times of division, for example, cells can do little else. Cells that specialize in secreting hormones lose their ability to divide or grow. Pareto optimality captures in mathematical terms this exact state and describes its mechanics.
Optimizing for competing functions
Pareto optimality, named after the Italian engineer Vilfredo Pareto, characterizes the competing preferences of a group of individuals. A system reaches an optimal state in this way of thinking when no individual can be made better off without making others worse. Applied to competing cancer cells, trade-offs between the different biological tasks govern the distribution of resources. The same trade-off logic should apply to things like protein production and gene expression. Cells do not produce equal amounts of proteins at all times since they require only specific sets of proteins to achieve a biological function. So we should expect that in committed cells, a pattern of genes emerges differently from others that are not committed to the same task.
Cells do not produce equal amounts of proteins at all times since they require only specific sets of proteins to achieve a biological function.
Using gene abundance measurements, we can infer which functions direct a tumor by evaluating the particular set of genes expressed at a high level. We can also subject these tumors to different compounds and examine how the profiles of genes vary. Finally, we have crude measures for guessing which task drives a given tumor. The clinical characteristics of the patients from whom the tissues were isolated and their pathological gradation sometimes reflect these optimized tasks.
The study I referred to presented three key observations to support this model. First, the gene expression profiles of the different tumors corresponded to the known function of cancer cells. In addition, the tasks made sense in light of the clinical characteristics of the samples. Second, driver mutations pushed the expression profiles toward specialized tasks. Third, the tumors specialized in a given task were sensitive to drugs that disrupt it. Each of these three observations/predications deserves a separate treatment which I attempt next.
Tumors tune their gene expression for specific tasks.
Biologists use the term gene expression to refer to the abundance of a given gene inside cells. Genes are later translated into proteins to perform specific functions. These measurements indirectly provide an idea of the protein level and, more importantly, are easier to quantify. The numerically transformed profiles form a space with a given number of dimensions. On the corners of this shape lay the profiles with extreme values (archetypes), and in the middle, the nonspecialized generalists. For example, a cube has three dimensions with square faces; each has four corners. With more corners, we get a polyhedron. At the extremes of this shape is the highest abundance of a group of genes. By examining what functions they are known for, we can infer what role the set of genes is responsible for.
Some tumors had a higher expression of genes involved in cell division, energy production, lipogenesis, immune interactions, or tissue remodeling. Those five functions were among previously reported hallmarks of cancer that biologists agree are crucial. More invasive tumors were nearer to the invasion task. Tumors that are at an early stage were near the cell division task. Mutations drive changes in gene expression toward specific tasks.
Not all parts of the genomes code for proteins. When mutations occur in the coding regions, they could alter the structure or the amount of the protein. Mutations that cause changes in expression, we call drivers, and those that don't, we call passengers. The driver mutations alter the genes toward more specialized profiles (archetypes). For example, known mutations in the IDH1 and TP53 genes shifted glioma and breast cancer tissue expression toward the cell division profile. This observation is significant because it proposes a potential mechanism by which a cell can specialize even though, in theory, every cell contains the same set of genes. It's the change in the sequence or the structure of the genes that result in abundance changes, which eventually gives a cell the propensity to divide, invade, or process energy in a specific way. The totality of the cells in a given tumor gives it the characteristics of being more or less invasive, proliferative, or recurrent. Tumors are more sensitive to drugs that target their specialization.
Driver mutations alter the genes toward more specialized profiles.
If a given drug disrupts cell division, we should find a significant disruption in the tumors that multiply quickly. The study found precisely that. So, for example, tumors specialized in invasion and remodeling were sensitive to a drug called Trametinib. Those who specialized in cell division were sensitive to Ixabepilone. The first of the two drugs inhibits the pathway responsible for regulating cell proliferation, and the second stabilizes the microtubules essential for cell division. Evolutionary thinking can be instrumental in describing how different life forms arise and even help describe their behavior. Here, Hausser and colleagues used evolutionary logic to explain how various tumors optimize for various functions. Due to limited resources, a trade-off is a must. One manifestation of this logic is that the expression profiles of the different cells are optimized for a few tasks, be it cell division or invasion.
The gene expression profiles of the different cells are optimized for a few tasks, be it cell division or invasion.
These findings, on the one hand, show why different tumors may behave differently. For example, some are more prone to grow locally, and others tend to disseminate quickly and move to other parts of the body. Could it be also why specific mutations are prevalent in all tumors yet not relevant to all? The modifications that alter the gene expression appear to be very relevant here; they change the gene expression toward particular functions, for example. Finally, why drugs with a specific mechanism of action are effective in some cancer types, not others? The exciting study shows that when a drug disrupts a task in which the tumor is specialized, the tumor is more sensitive. These and others are questions raised by the study.
- Hausser J, Szekely P, Bar N, Zimmer A, Sheftel H, Caldas C, Alon U. Tumor diversity and the trade-off between universal cancer tasks. Nat Commun. 2019 Nov 28;10(1):5423.