blog

Suppose you were taught that each species was created separately. In your anatomy class, you learned that a whale flipper is clearly designed to aid in swimming. A human hand is nicely adapted for grasping and manipulation. That all makes perfect sense.

Then the puzzle comes. Why is the complex bone structure in whale flippers and human hands so similar? That’s a good puzzle, because it tells you that something deeper is going on. Something is missing in your current view. Solve that puzzle, and a whole new way of looking at the world follows.

The story is always the same. Some pieces, at first hardly noticed, start to nag because they do not fit into the background picture. Once you see the puzzle clearly, you have a chance to solve it. But nothing happens until you see that there is a puzzle.

It is easy to see the puzzles that influenced the history of thought. It is not so easy to see the puzzles that are important right now—the puzzles that will shape the future direction of thought. I have for many years been struggling to see those puzzles of modern biology. Trying to see those modern puzzles clearly is much harder than for the nice stories of past work.

I started a series of essays to sharpen my focus. And, I hope, to stimulate others to solve these important puzzles, and to add their own puzzles to the list.

Of course, we want great puzzles. A great puzzle is one for which the solution connects up many disparate facts and seemingly different subjects into a more coherent whole. Usually we don’t get a great puzzle in one go. But we should set off in that direction.

Here are brief summaries of three examples from my series.

Why are so many males sterile?

Roughly 5% of human males fail to make good sperm. In many plant species, widespread pollen abortion and male sterility occur. In many different species, hybrid matings often produce sterile male progeny.

Natural selection prunes reproductive failure, thus widespread sterility is perhaps the most puzzling of all traits.

We do not know for certain why each type of male sterility occurs. In some cases, there are reasonable ideas and potential solutions. The point here is that these are good, perhaps great, puzzles. Why? Because exceptional failure almost always reveals something interesting about the way that organisms have been designed by evolutionary processes.

For example, widespread pollen abortion in plants often seems to be caused by mitochondrial genes. At first an unsolved puzzle, we have now figured out the reason. Mitochondria are passed to seeds through ovules. Mitochondria do not transmit through pollen. Mitochondrial mutations that abort pollen production and increase ovule production gain an advantage and become favored by natural selection.

Here, we have a conflict between different parts of the genome. Mitochondria transmit only through ovules. Nuclear genes transmit equally through ovules and pollen. Similar genomic conflicts appear to have shaped many aspects of the genome.

Plant male sterility is a great puzzle, because its solution provides broad insight into the evolutionary forces that have shaped genomes. Other cases of male sterility may also turn out to be great puzzles, because exceptional failure often reveals deep principles of evolutionary design.

Where does neurodegeneration start?

Until recently, the unstated view was that spatially separated neural tissues decay independently with age. Eventually, symptoms increase as the widespread burden of diseased tissue develops.  Alternatively, it could be that disease starts in a small piece of tissue and then spreads, as in cancer.

The distinction matters. Local initiation may begin by changes in a tissue microenvironment, by somatic mutation, or by various epigenetic or regulatory fluctuations in a few cells. A local trigger must be coupled with a mechanism for spread.

By contrast, independent decay across spatial locations cannot begin by a local change. Instead, parallel degeneration must depend on some global predisposition or spatially distributed change that leads to approximately synchronous decay.

To understand the causes of neurodegeneration, we must solve the puzzle of where neurodegeneration starts. Consideration leads to the deeper puzzle of how the mechanisms that initiate and spread disease within the body determine the relative importance of interacting local and global processes. The relative roles of somatic and inherited mutations must be understood in relation to this local versus global dichotomy.

Cancer provides an example of disease progression by local triggers and spatial spread, setting a conceptual basis for clarifying puzzles in neurodegeneration. Heart disease also has crucial interactions between global processes, such as circulating lipid levels, and local processes, such as the development of atherosclerotic plaques. The distinction between local and global processes helps to understand these various age-related diseases.

Why are genomes overwired?

Eukaryotic gene expression depends on transcription factors, methylation, histone codes, DNA folding, intron sequences, RNA splicing, noncoding RNA, and others. Gene expression controls cellular traits through this complex network of regulatory factors.

An engineer following classic principles of control theory would design a simpler system, one with fewer connections. Genomes are overwired. They have far more nodes and connections than classically engineered systems.

Why is regulatory wiring so complex? How does dense regulatory wiring influence trait expression and evolutionary processes of change? At present, there are no compelling answers. This is a great puzzle. The solution, I believe, will require synthesis of many disparate lines of thought.

Along one line, the strangeness of evolutionary dynamics, in which neutral accumulation of random changes interacts with powerful forces of selection that direct traits toward reproductive efficiency.

Along another line, the interplay between robustness, which protects genomic attributes from perturbation or corrects errors, and the inevitable evolutionary decay of those attributes that are protected from the direct action of selection.

Conceptually, one must contrast the simple direct logic by which one interprets function with the inductive dynamics by which natural selection actually accumulates those mechanistic changes that benefit function.

The relations between mechanistic architecture and the inductive dynamics of natural selection call to mind the spectacular recent breakthroughs in computational neural networks and deep learning that are revolutionizing modern life. In deep learning networks, the overwired architecture of broadly and densely connected computations interacts with inductive learning algorithms that seek general patterns.

Future puzzles

Scaling laws are puzzling. Geoffrey West introduces his recent book, Scale, by showing the incredible regularity of certain quantities in nature and in human activities.

For example, on a log-log plot, one gets a straight line between metabolic rate and body mass ranging from mice to elephants. We say that metabolic rate scales in a regular way with body mass.

In human activity, similar log-log scaling occurs between the number of patents created in a city and the population size of the city.

In each case, one can see why the variables may be related to each other. But why does the scaling follow an essentially perfect law-like pattern? Nearly all of the data across many orders of magnitude fall almost exactly on the scaling line.

West and his colleagues have given us a great puzzle. They have also suggested possible solutions, for example, fractal processes of growth. I suspect there is something deeper and broader, a solution that will illuminate many common patterns of nature. Opinions aside, this is a real challenge.

What are the other great puzzles of modern biology? I am always working on my list. Send me your ideas.