Anatomy of the cell’s brain

For some time, I have been wanting to write a post on the ”hardware” of information processing in the cell. AK’s Rambling Thoughts has however beat me to it, in a series of four blog posts describing certain core principles. If I try to sum up the most important points, here they are:

  • Enzymatic reactions can function, in principle, as analog computer circuits.
  • The sheer number of interlinked reactions, both enzymatic and involving the regulation of protein transcription, allow us to view the cell as a computer with a huge processing power.
  • Many of the analog circuits are wired with positive and negative feed-back mechanisms that enable them to give a digital, all-or-none response. This entails a loss of processing power, which is in AK’s opinion offset by advantages of speed.
  • The computing is “modularised” in the cell, meaning that some reactions occur only in specialised compartments (a synaptic bouton, for example), and communication within the cell can take the form of chemical gradients occurring as a consequence of reactions taking place only in one location.

AK gives many specific and detailed examples of the mechanisms that are involved in each of these processes. But to me, the real power of the discussion lies in the presuppositions that are not very explicitly stated, but very strongly supported by AK’s examples.

The cell deals in processing information
It is easy enough to see that cells process matter. They take up glucose and other nutrients and transform them into energy and structural components. It is not intuitive for many to think of these processes as manipulation of information, but here’s how I see it:

If the cell is going to be successful at all, it needs very fine-tuned regulation of its diverse activities. Any regulation that takes place in response to external factors can be thought of as an internal representation in the cell of its exterior. A map, if you wish, in many dimensions – oxygen tension, concentrations of nutrients, activities of hormones, and so on. This map will then be the input into the cell’s “decision machinery”, which will output some sort of behaviour.

Historically, the decision system has often been conceptualised as a set of enzymes acting linearly and without a lot of interconnections. With more sophisticated models derived from computer theory, it is far more likely that we will be able to capture some of the complex goings-on and generate accurate predictions of how these systems will behave.

Important principles emerge even from only a very general knowledge of the signalling networks
We do not have to know exactly which enzymes do what in order to make some pretty powerful inferences about the signalling network. Based only on a few known classes of interaction, like positive and negative feedback, it’s possible to envision properties of the whole system. Prime among these is the strong nonlinearity that must be expected when there are so many parameters.

In a way I envy physicists. If I poke my teacup, they will be able to predict exactly where it will move to, and when, and in what position it will stop. But if I poke a member of one of my cell cultures (gently), or Zelda, my mother’s Chinese Crested, there is no telling what will happen.

Zelda. Yes, she is cute.

Zelda. Yes, she is cute.

Suddenly the system might go into one of these digital, all-or-none responses. I do not believe they are motivated by speed, by the way, because I do not see why an analog response could not be as fast. Sometimes they are probably motivated simply by the necessity to decide some things one way or another. In particular, this concerns “cell fate decisions”, when the cell decides whether to replicate, or to commit apoptosis. There can be no half-measures with some things.

The cell is not a unit
Some things appear to us at first to be one and indivisible. For example, it came as a small shock to me when I learned that my brain sleeps in regions – it’s not at all necessary for the whole brain to sleep at once. I have also discovered an interesting subdivision when I am daydreaming. Rather often, I start constructing arguments or writing a speech in my head. No visual imagery attached. That’s left hemisphere activity. And then I find myself having hummed a tune with no lyrics, almost without noticing, perhaps for several minutes. That’s the right hemisphere entertaining itself when I am not watching.

The “I” of the cell can be similarly elusive. Some cells are several decimetres long, making it impossible for diffusion of small molecules to carry information from one part of the cell to another. Besides the specialised long-distance information transfer systems (such as axonal transport mediated by the cytoskeleton), such a cell is a set of functionally semi-independent decision-making centres. It’s a bit like an 18th-century colonial empire with faraway patches of land, only intermittently connected by slow sailing vessels.

There are times when I feel a pang of jealousy over discoveries already made, that other people have lived to experience. But then I remember that the pace of discovery has only increased and keeps increasing, and then I feel the same sense of wonder as one might before the sky on a starry night, when I think of what we will know about these systems in one or two decades.

3 Responses to Anatomy of the cell’s brain

  1. AK says:

    Great summary. I’ll admit I probably didn’t spend enough time on this aspect of the idea. As it was I bit off more than I could chew with part IV.

    As for your pangs, there’s plenty waiting to be discovered as the detailed discoveries are integrated into more global paradigms. Personally I think the field is ripe for a paradigm shift, with whole worlds to be discovered in the new paradigm(s).

  2. AK says:

    I didn’t notice at first…

    This entails a loss of processing power, which is in AK’s opinion offset by advantages of speed.

    […]

    I do not see why an analog response could not be as fast.

    Actually, it’s a tradeoff between speed and power. In order for a phosphorylated enzyme system to react very rapidly, both phosphorylation and dephosphorylation have to be happening at a high rate. Since this is an energy-expensive process, the enery cost of a rapid analog response is high.

    This would also be true of a simple digital system, but a more sophisticated system could be arranged so it only used energy while switching. The most common example is a trigger system, in which a self-catalysing phosphorylation cycle runs quickly to completion, after which it has to be reset through a more complex process.

    There are fairly fast analog processes (AFAIK), for instance in the dendrites, but these tend to be fairly expensive.

  3. evolvingideas says:

    Thanks AK for your comments!

    It’s probably correct that the all-or-none systems are fairly expensive in many cases.

    As always, I am biased by my background in cancer research – in fact, all the digital signalling processes with which I am familiar in any detail pertain to apoptosis signalling, in which case the energy expenditure is not much of an issue anyway. (The reason being that the enzymes effecting apoptosis are synthesised some time before the process is initiated, and are then activated mostly by cleavage rather than phosphorylation, thus without requiring much ATP.)

    Let the paradigm shifts come! And let us sing of them in praise!

    (And let me not blog at this hour of night, lest I become slightly bombastic.)

%d bloggers like this: