Shift Happens (Part II)
(I'm rushing this a bit, so no voluminous cites; they're all available upon request if you don't want to take my word for something.)
The thermodynamic model of nerve impulses is badass on grounds of parsimony -- it enables us to make more predictions without making ad hoc assumptions or needing to know much about the fine chemical details -- but like any good theory it shows its real value by having immediate cash value in terms of solving outstanding problems in an elegant way.
For example, the fact that local cooling can induce action potentials even in the absence of sodium or potassium ions in the extracellular fluid, and local heating can inhibit them, falls out as a trivial consequence. So does the little-known fact that pulses can be induced by locally increasing the acidity of the membrane's environment: an excess of protons results in the (negatively charged) lipids and proteins becoming protonated, which increases the melting temperature of the membrane. Dumping calcium ions near the membrane does the same thing, for similar reasons. In general, anything that raises the phase transition temperature is predicted to be excitatory and anything that lowers it should be inhibitory.
Which brings us to the main event: the first non-trivial application the authors put their model to is in explaining how anaesthetics work. Given that they've been in use for a century and a half, the ongoing lack of a satisfactory general theory of anaesthesia is a bit of a scientific embarrassment. The one major discovery about them is almost as old as their use: anaesthetic potency correlates extremely well with lipophilicity, accounting for the overwhelming majority of the variance (called the Myer-Overton rule). This suggests that interactions with the lipid membrane are crucial to anaesthesia; the precise chemical nature of anaesthetics varies widely, but there's no other known feature of them that predicts their potency better than the MO rule.
The really interesting cases are krypton and xenon: noble gases like these don't specifically bind to anything at all, yet they still impeccably obey the same empirical rule as more complex anaesthetic molecules. In spite of all this, the current model has forced an obsession with explaining anaesthesia in terms of effects on ion channels. And there are some: anaesthetics do interact with proteins in the membrane, wherever they happen to have hydrophobic pockets that are the right shape and size. But the problem is that these interactions vary haphazardly from one anaesthetic to the next; none of them have the same protein interaction profile.
The empirical consistency and nonspecific nature of the MO rule demands an explanation that's independent of fine chemical details -- we want a theory that explains how all the different anaesthetics work and why the MO rule exists, not a different theory for each chemical class. In order to do this within a protein-centric framework, it'd have to be the case that all the anaesthetics had highly similar receptor-binding profiles, but they don't, so you end up forced into the biochemical equivalent of Ptolmaic epicycles.
From the POV of the more comprehensive thermodynamic model this all makes immediate sense: the potency of an anesthetic should be exactly proportional to their capacity to lower the melting point of lipid membranes. This refines and deepens the MO rule from an empirical generalization to a theoretically-grounded principle. So, prima facie this is pretty good. But hey, why stop there? Let's see what other fish we can drag in with this new net.
Here's a fun one, which is a bit of a mirror-image of the case of anaesthetics. It's been repeatedly observed that psychedelic potency *also* tends to correlate strongly with lipid partitioning, though the curve isn't such a nice tidy line. Check it:
Lipophilicity only accounts for about 60% of the variance, but still pretty "hmmmm"-worthy. If psychedelics expressed their effects via some membrane-related mechanism, that would explain the lack of a consistent protein-binding profile: for example, the 5HT2A receptor is a common binding site for classic psychedelics, but there are plenty of agonists at that receptor which aren't psychedelic and also several psychedelics that don't have significant affinity for it. And hey, it's not like we have a good theory of how they work their magic anyhow.
In the Barknecht et al (1974) paper I ganked that graph from, the authors note some interesting contrasts, like how DMA and mescaline have comparable lipid partition coefficients, yet mescaline is way more potent. The only difference between them is the an extra methoxy group on mescaline; same deal holds between amphetamine and TMA. So clearly there's more going on here, but a theory that makes partial sense is better than none, and we might be able to improve on it by considering other factors.
I have other priorities right now, but in a week or two I'll kick the tires of the soliton model a little and take it for a spin into some weird territory. Same Bat Time, Same Bat Channel.
The thermodynamic model of nerve impulses is badass on grounds of parsimony -- it enables us to make more predictions without making ad hoc assumptions or needing to know much about the fine chemical details -- but like any good theory it shows its real value by having immediate cash value in terms of solving outstanding problems in an elegant way.
For example, the fact that local cooling can induce action potentials even in the absence of sodium or potassium ions in the extracellular fluid, and local heating can inhibit them, falls out as a trivial consequence. So does the little-known fact that pulses can be induced by locally increasing the acidity of the membrane's environment: an excess of protons results in the (negatively charged) lipids and proteins becoming protonated, which increases the melting temperature of the membrane. Dumping calcium ions near the membrane does the same thing, for similar reasons. In general, anything that raises the phase transition temperature is predicted to be excitatory and anything that lowers it should be inhibitory.
Which brings us to the main event: the first non-trivial application the authors put their model to is in explaining how anaesthetics work. Given that they've been in use for a century and a half, the ongoing lack of a satisfactory general theory of anaesthesia is a bit of a scientific embarrassment. The one major discovery about them is almost as old as their use: anaesthetic potency correlates extremely well with lipophilicity, accounting for the overwhelming majority of the variance (called the Myer-Overton rule). This suggests that interactions with the lipid membrane are crucial to anaesthesia; the precise chemical nature of anaesthetics varies widely, but there's no other known feature of them that predicts their potency better than the MO rule.
The really interesting cases are krypton and xenon: noble gases like these don't specifically bind to anything at all, yet they still impeccably obey the same empirical rule as more complex anaesthetic molecules. In spite of all this, the current model has forced an obsession with explaining anaesthesia in terms of effects on ion channels. And there are some: anaesthetics do interact with proteins in the membrane, wherever they happen to have hydrophobic pockets that are the right shape and size. But the problem is that these interactions vary haphazardly from one anaesthetic to the next; none of them have the same protein interaction profile.
The empirical consistency and nonspecific nature of the MO rule demands an explanation that's independent of fine chemical details -- we want a theory that explains how all the different anaesthetics work and why the MO rule exists, not a different theory for each chemical class. In order to do this within a protein-centric framework, it'd have to be the case that all the anaesthetics had highly similar receptor-binding profiles, but they don't, so you end up forced into the biochemical equivalent of Ptolmaic epicycles.
From the POV of the more comprehensive thermodynamic model this all makes immediate sense: the potency of an anesthetic should be exactly proportional to their capacity to lower the melting point of lipid membranes. This refines and deepens the MO rule from an empirical generalization to a theoretically-grounded principle. So, prima facie this is pretty good. But hey, why stop there? Let's see what other fish we can drag in with this new net.
Here's a fun one, which is a bit of a mirror-image of the case of anaesthetics. It's been repeatedly observed that psychedelic potency *also* tends to correlate strongly with lipid partitioning, though the curve isn't such a nice tidy line. Check it:
Lipophilicity only accounts for about 60% of the variance, but still pretty "hmmmm"-worthy. If psychedelics expressed their effects via some membrane-related mechanism, that would explain the lack of a consistent protein-binding profile: for example, the 5HT2A receptor is a common binding site for classic psychedelics, but there are plenty of agonists at that receptor which aren't psychedelic and also several psychedelics that don't have significant affinity for it. And hey, it's not like we have a good theory of how they work their magic anyhow.
In the Barknecht et al (1974) paper I ganked that graph from, the authors note some interesting contrasts, like how DMA and mescaline have comparable lipid partition coefficients, yet mescaline is way more potent. The only difference between them is the an extra methoxy group on mescaline; same deal holds between amphetamine and TMA. So clearly there's more going on here, but a theory that makes partial sense is better than none, and we might be able to improve on it by considering other factors.
I have other priorities right now, but in a week or two I'll kick the tires of the soliton model a little and take it for a spin into some weird territory. Same Bat Time, Same Bat Channel.