On doors

Prompted by something about anaesthetics in reference 1, I have been poking around to try and find out something about how general anaesthetics work. With there being a long standing interest in whether this might inform how consciousness works – with the presumption being that it ought to.

Part of the answer seems to be that, despite some general principles, a large amount of work and a large number of facts, there is still a great deal to learn. We might know a good deal about the relatively small number of general anaesthetics in common use, a good deal about how to administer them and about how they affect the consciousness, awareness and responsiveness of the subject. But a lot of the molecular detail is still missing. How exactly is it that they reliably do what they do?

What follows is an attempt to bring some order into this layman’s knowledge of the matter, expressed, as is my wont, as a box model. Bearing in mind, as always, that these models are, to some extent at least, a matter of taste. There are usually lot of different ways to draw them.

I restrict myself to vertebrates, all organised in much the same way and for present purposes, perhaps all much the same.

Vertebrates are made up of cells. Cells can be organised into types, hundreds of them. Neurons are specialised cells and I believe that they come in around a hundred different varieties. Cells are often built into large assemblies, but that is not the whole story as there is also the extracellular medium, usually watery.

As with a machine or a ship, it is often convenient to break a vertebrate down into a hierarchy of parts, here called regions. So it might be convenient to break a body down into organs plus supporting structures. It might be convenient to break a brain down into a couple of hundred of regions – and lots of work has gone into defining regions of this sort in an effort to promote communications between the many teams working on brains.

In principle, there is a many to many map between regions and cell types. This region has that distribution of cell types. One might try to get statistical and specify the amount of variation one might reasonably expect between individuals. So, on average, A% of the cells of region B of species C will be of type D. And A plus or minus E will include F% of individuals. Or moving from percentages to numbers, on average, region B of species C contains A cells of type D. And A plus or minus E will include F% of individuals. Where A, E and F are numbers and B, C and D are names. In practise, we do not know many numbers of this sort, although eference 2 gives the flavour of how much we do know about how many neurons there are in a brain – without much regard to type of neuron or region within the brain.

In the case of neurons at least, it is convenient to think of regions of the cells, possibly four of them: dendrites (input side), cell body, axon hillock and axon (output side).

All cells have a cell wall behind which they do their business, a wall which can be characterised as a lipid bilayer. Such bilayers have a convenient tendency to form spheres – or something not that far removed. Spheres which are closed and more or less impermeable except at special places here called doors. For which see reference 3.

A door is a single molecule of a protein, a molecule which crosses the lipid bilayer from inside the cell to outside the cell, to the extracellular medium, sometimes in a tube-like configuration. Most cell input-output takes place at these doors. A cell which cannot do input-output for any reason will not last for long, if only because a cell is alive and needs energy, in some form or another, to stay alive.

By default, doors are shut as unrestricted input or output is a bad thing, usually fatal. But doors may be either opened or blocked by binding events between the door protein and special molecules called ligands – which might also be anaesthetics, or, contrariwise, toxins. I think that binding events are quite short, parts of a second rather than lots of seconds.

Again, one might try to get statistical and estimate how many doors a cell of any given type has. A number which, in the case of a neuron, might be many thousands. One might try to estimate the distribution of the duration of binding events. For the doors important to the brain called ligand gated ion channels, see reference 4.

Turning to the right hand side of the model, vertebrates might involve thousands of distinct proteins, some in large amounts, some in small. Some hundreds of these thousands might function as doors. As noted above, doors which are controlled by ligands, which might work from inside the cell or from outside the cell – this last being the case with anaesthetics. A door protein might have several sites at which a ligand might bind, perhaps working on the door in several different ways. In the case that we have several bindings on the same door, some kind of integration takes place to arrive at door behaviour.

Then any one binding event will take place at a binding site, a pairing of a host protein with a visiting ligand. Some, perhaps most, binding sites, will work with a number of ligands – some natural, some life enhancing and some toxic.

Additional information

We have only looked at doors connecting the inside of a cell with a medium which hosts the cell, the extracellular medium. For example, doors on the boundary surface between the walls of the intestines and the lumen, more or less the world outside. On the boundary surface between the alveoli of the lungs and the world outside. And of particular importance here, on the boundary surface between the interior of a neuron and the interstitial system within which the neuron is suspended, is situated. 

We have not looked at doors providing a direct connection between two cells and we have not considered the grouping of cells into sheets, larger assemblies, structures or organs.

Even so, inspection of the workings of even this part of the machinery in vivo can be difficult, in part because they happen very quickly, but it seems that computer simulations have got to the point where they can sometimes do better in silico. To which end one might use something like the CHARMM computer package (of the not very accessible reference 5) and one might need a large computer. Perhaps not so far from what the Deepmind people are doing with computing the folding of proteins, for which see the even less accessible reference 6.

Conclusions

The box model might be along the right lines, but there are not a lot of numbers to attach to it. Either because I have yet to find them or because they are not yet there, possibly because we do not yet have the necessary technology to make the estimates.

Notwithstanding, we know enough to know that the number of potentially interesting combinations is very large. Enough to keep lots of laboratories busy for a long time.

References

Reference 1: Feeling and knowing: making minds conscious - Antonio Damasio – 2021.

Reference 2: The human brain in numbers: a linearly scaled-up primate brain - Suzana Herculano-Houzel – 2009.

Reference 3: https://en.wikipedia.org/wiki/Lipid_bilayer. 

Reference 4: https://en.wikipedia.org/wiki/Ligand-gated_ion_channel. 

Reference 5: CHARMM: The Biomolecular Simulation Program - B.R. Brooks, C.L. Brooks and others – 2009.

Reference 6: Highly accurate protein structure prediction with AlphaFold - Demis Hassabis and many others – 2021.