Cells seem to decode their fate through optimal information processing | Hacker News Reader

Cells seem to decode their fate through optimal information processing

quantamagazine.org

153 pointsErrorcod333 comments

This is super cool! People underestimate how insanely chaotic a cell is at a molecular level. Often diagrams show blobs cleanly interacting but the reality is more like the images linked here:

https://mgl.scripps.edu/people/goodsell/illustration/public/

Everything touches everything. Everything is always moving around.

This work suggests that the cell has taken advantage of this enormous challenge. If everything is always in motion, that means you have trouble controlling things, but that gives you a chance to maximally sample your environment. This makes this sort of efficient data processing possible. Downstream are a number of mechanisms that help make sense of that signaling, denoising the chaos. One example from a lab I worked in briefly (old but still cool):

https://www.sciencedirect.com/science/article/pii/S009286741...

https://www.ncbi.nlm.nih.gov/pubmed/18599789

These illustrations are fantastic! These need to be included alongside the simplified models so students also get the big picture. Students of biochemistry should inherently get a sense of this, but the pictures communicate this point so well.

This is exactly why biochemistry is so hard. The smallest cell is a maelstrom of complexity.

See the book "The Machinery of Life", by David S Goodsell, the artist of the first illustration. It's an entire book of such illustrations, ordered to walk the reader though the different processes happening in a cell. I'm not a biologist, but I bought this book several years ago as it was just so fascinating and appealed to my engineering mind set of wanting to know how a system hangs together.

It makes sense in my mind that a noisy system will be, on average, operating at something approaching optimality. The noise jogs the system out of local minima and is effectively a form of annealing (in the numerical sense).

Off-topic, but thank for introducing me to the phrase “how a system hangs together” - it so neatly encapsulates the idea of the interactions between supporting elements of a system and the forces acting upon them.

Also, great connection to annealing.

Yes, my analogy is a blind man in a room with 1000 locked doors, and he has a key that fits only one of the locks and somehow, quickly, finds that door, uses his key, and opens it.

In recent years computer animation is being used to render astonishing bio-molecular vistas. Drew Berry seems to have pioneered this around 2010. Björk promptly used his work for her music video "Hollow".

Here is an example of the 'genre'

[1 - Harvard MCB : The Inner Life of the Cell ] https://www.youtube.com/watch?v=VdmbpAo9JR4

As my biology teacher put it: a cell is a factory in which workers are constantly kicking each other and workpieces around. Sometimes they bump into a workpiece they know what to do with, so they do some work and then throw it back into the crowd.

Hah, I just had a weird thought. I wonder if increasing motion/exercise speeds up intracellular action from wiggling it all around a bit and increasing the "random bump factor"? Or is the effect of macro movement negligible at a small scale?

Nope, but increasing temperature does speed up the chemical reactions within cells due to more vigorous motion of the constituent molecules. And this is true for not just biological processes but basically any chemical reaction: the higher the temperature, the faster the reaction.

And if I remember correctly, the elements in a cell may interact couple billion times per second.

Reminds me of an old school trading floor

Awesome images, thanks for sharing. I wish they'd have put these in science books growing up.

Are these to scale? It's interesting to see how much more particular (as in, composed of large particles) these cells look compared to the illustrations I'd seen growing up.

I always imagined the proteins that made up a cell wall would be more on the order of how grass looks to a human, in terms of size respective to the cell.

Or another analogy, I used to think of a cell like a room (the cell membrane) with furniture in it (the organelles). Now it seems more like just a big pile of furniture.

I believe the components are indeed drawn to scale! The cell is a mess, there is virtually no central coordination, just individual molecular signals. A protein gets phosphorylated by a receptor, then its on its merry way till it hits something it interacts with and maybe because it's phosphorylated it activates that thing to do something else, or maybe, luck of the draw it doesn't. It's amazing that it even works, but it does so because billions of years have shaped life to be resilient to chaos. Some context:

Unicellular life is roughly 3.5B years old

Eukaryotic life is roughly 2B years old. It took 1.5B years to figure out how to build and use organelles

Multicellular life is roughly 1B years old. Another billion years to figure out how to coordinate and specialize cells.

Sea sponges, essentially "animals" are 700M years old

Animals that can move around are about 500-600M years old, backbones in animals show up around the same time, then we're off to the races.

This sort of patterning is maybe in the 700-800M year old window. It took billions of years for life to figure this out, you can do a lot by evolving that long.

https://www.newscientist.com/article/dn17453-timeline-the-ev...

> A protein gets phosphorylated by a receptor, then its on its merry way till it hits something it interacts with and maybe because it's phosphorylated it activates that thing to do something else, or maybe, luck of the draw it doesn't.

There are proteins that spread via diffusion, but there is also transport along microtubules in transport vesicles. Not entirely sure how the routing/navigation question is solved though, aka how does some vesicle know it needs to go to the nucleus vs golgi apparatus.

Oh yes, of course there are many motorized processes, but the point is there's no inherent logic to them. A myosin grabs onto a filament and goes to the plus end of the actin filament (usually, not some of them, again it gets complicated :) but it doesn't know where that tube is going. It'll just happily move to the end and fall off/get blocked. If the destination "wants" or "needs" supra-diffusion transport, it needs to recruit the tubes to come to it. Every entity in the cell essentially acts as its own agent, and it's only through interactions that it does one thing or another.

EDIT: probably more accurate to call actin "filaments" not tubes, typo

Actually, there’s quite a bit of inherent logic to actomyosin traffic networks, as we’re now beginning to discover. For example, myosin 10 has evolved to walk along specific bundles of actin, but not so well along single actin filaments. The bundle has mechanical integrity and leads to a distant location in the cell, and the myosin 10 can detect that the actin is bundled. So in some sense it knows that it is walking along the right structure that leads to that distant location. There are some other examples, but this is all at the cutting edge of motility research.

I believe they are roughly to scale. I remember reading how Goodsell does some back of the envelope calculations when planning an illustration.

You hear in pop-science how the human body is ~50-70% water as if that is a lot. Really it’s a rather low level of solvent to run chemistry reactions in. I did some undergraduate research in a biochemistry lab and when we would perform an experiment with a protein we were studying, the proteins and other chemicals would always be very dilute, maybe ~99.9999% water or higher. Cells are very “sludgy” and it’s surprising to me that anything works at all the proteins don’t just immeadietly stick together and precipitate out of solution like egg-whites when cooked.

It’s a tricky number to estimate because the orders of magnitude are so stark. Proteins are at very, very low concentrations but each protein is made of hundreds or thousands of atoms. The number of protein molecules is small but the they are almost big enough that their volume is not insignificant. There are other things that are unintuitive about chemical reactions. A lot of chemical reactions are faster than the movement of molecules. A protein maybe able to catalyze a chemical reaction faster than it can diffuse through water so it sits around most of the time waiting to collide with the molecule it will catalyze. Basically like a CPU waiting an eternity for data from a “fast” SSD to show up.

> You hear in pop-science how the human body is ~50-70% water as if that is a lot. Really it's a rather low level of solvent to run chemistry reactions in.

For reference, gelatin is about 85% water[0], and still manages to be a solid glob you can balance utensils on top of.

0: The stuff I had on hand is 20g powder to 118g water.

Cells are extremely crowded. In fact, some processes like protein folding require crowdiness, if it's missing it can misfold.

Realistic animation: http://www.cellimagelibrary.org/images/28234

Also worth keeping in mind just how fast the motion really is. That whole clip happens in 15 micro seconds.

Another famous example: The magic of the Golgi stain, which first produced pictures of neurons, was that it stained only a tiny fraction of the cells. That's why you can obtain a picture with cells reaching out and occasionally connecting. Whereas in reality all that empty space is other cells, doing the same thing.

More modern pictures are like so: http://www.conte.harvard.edu/news/2015/8/18/imaging-the-brai...

literally the first thing we did in grad school (biophysics) was a journal club covering the idea that cells are so concentrated, basic things like the free energy of ATP hydrolysis doesn't match what you measure in free solution, and this has huge implications for cells. I forget the term (order parameter?) but the experiment was funny because it involved packing some lipid membranes full of PEG.

In your statement, based on your understanding, do you think either of the following wording substitutions are accurate?

Instead of: > how insanely chaotic a cell is Perhaps: "how insanely high a cell's spatial and temporal entropy is" ?

And instead of: > denoising the chaos. Perhaps: > decompressing (interpreting) the spatio-temporal activity ?

If you put cells in a blender, their entropy goes up. And a crystal has lower entropy. So what’s interesting about them isn’t that their level of entropy is high or low.

It sort of is related to being in between though. It's that their complexity is high. Which is essentially at the phase transition between high and low entropy states of matter. Cells regulate a system full of energy to keep themselves in that messy but information rich state.

I spent some time in the Gregor lab during grad school. I'm obviously biased but I think the work represents some of the most original happening right now in biophysics. These papers were extremely rewarding to read and represent almost a decade of work on part of many members of the lab.

For those interested, I recommend diving into some of the lab's earlier work as well as the work of Bill Bialek, Thomas's advisor, who formulated a lot of these theories for photon sensing in the eye decades ago.

Have you got any references for Bialek’s theory of photon sensing?

Bialek is an amazing explainer, you'll be left convinced it's all so simple and straightforward. He wrote a book and taught a graduate level class around it, you'll find a draft of the book available at [1]

Chapter 1 cover photon counting:

``` 1. Photon counting in vision (Lectures W 8 Feb through W 22 Feb 2012)

In this Chapter, we will see that humans (and other animals) can detect the arrival of individual photons at the retina. Tracing through the many steps from photon arrival to perception we will see a sampling of the physics problems posed by biological systems, ranging from the dynamics of single molecules through amplification and adaptation in biochemical reaction networks, coding and computation in neural networks, all the way to learning and cognition. For photon counting some of these problems are solved, but even in this well studied case many problems are open and ripe for new theoretical and experimental work. The problem of photon counting also introduces us to methods and concepts of much broader applicability. We begin by exploring the phenomenology, aiming at the formulation of the key physics problems. By the end of the Chapter I hope to have formulated an approach to the exploration of biological systems more generally, and identified some of the larger questions that will occupy us in Chapters to come.

```

[1] http://www.princeton.edu/~wbialek/PHY562.html

Thank you immensely!

These biological computers are intrinsically based on vibrations. The vibrations aren't just a source of diffusion and Brownian motion; they cohere into meaningful harmonic structures. Alan Turing described morphogenesis in terms of inhibition and excitation loops, which gives rise to banding patterns due to oscilatory harmonics and resonances [1]. We are so accustomed to thinking about things in terms of discrete, separable parts, we have a hard time imagining emergent temporal structures. Living organisms, from cells to brains to cities, are composed of interacting waves and harmonic structures. (I'm emphasizing a hippie-style "resonance and harmony" language here because it really is so critical for understanding these systems.

[1] Yang, L., Dolnik, M., Zhabotinsky, A. M., & Epstein, I. R. (2002). Spatial resonances and superposition patterns in a reaction-diffusion model with interacting Turing modes. Physical review letters, 88(20), 208303.

William Bialek is my favorite scientific speaker (he has some good ones on youtube). His depth in such a wide range of sciences and topics is remarkable.

I think Thomas Gregor has some of the most precise biological measurements at the single molecular level.

The combination of the theory and precision measurements in studying the fly embryo by these people have resulted in very unique and creative progress in the field. From what I hear when they first started this work, the old-school developmental biologists thought what they were doing was absurd. They have successfully put a much more quantitative perspective back into biology.

This reminds me of the work over at Levin lab.

https://ase.tufts.edu/biology/labs/levin/

Is there an information equivalent to gravity, e.g. some sort of gradient is formed that the cell simply follows like a bowling ball on a sheet?

Reply on news.ycombinator.com