The staggering implications of one axon synapsing on another

It isn’t often that a single paper can change the way we think the brain works.  But such is the case for the paper described in the previous post (full copy below *** ) if the implications I draw from it are correct.

Unfortunately this post requires a deep dive into neuroanatomy, neurophysiology, neuropharmacology and cellular molecular biology.  I hope to put in enough background to make some of it comprehensible, but it is really written for the cognoscenti in these fields.

I’m pretty sure that some of these thoughts are both original and unique

Briefly, the paper provided excellent evidence for one axon causing another to fire an impulse (an action potential).   The fireror was from a neuron using acetyl choline as a neurotransmitter, and the fireree was a dopamine axon going to the striatum.

Dopamine axons are special.  They go all over the brain. The cell body of the parent neuron of the axon to be synapsed on uses dopamine as a neurotransmitter.  It sits in the pars compacta of the substantia nigra a fair piece away from the target they studied (the striatum). “Individual neurons of the pars compacta are calculated to give rise to 4.5 meters of axons once all the branches are summed”  — [ Neuron vol. 96 p. 651 ’17 ].”  These axons release dopamine all over the brain.  There aren’t many dopamine neurons to begin with just 80,000 which is 1 millionth of the current (probably unreliable) estimate of the number of neurons in the brain 80,000,000,000.

Now synapses between neurons are easy to spot using electron microscopy.  The presynaptic terminal contains a bunch of small vesicles and is closely apposed (300 Angstroms — way below anything the our eyes can see) to the post synaptic neuron which also looks different, usually having a density just under the membrane (called, logically enough, post-synaptic density).  Embedded in the postsynaptic membrane are proteins which conduct ions such as Na+, K+, Cl- into the postsynaptic neuron triggering an action potential.

But the dopamine axons going all over the brain have a lot of presynaptic specialization, but in many of the cases the post-synaptic neuron and its postsynaptic density is nowhere to be found (or the receptors for dopamine aren’t near the presynaptic specialization).  This is called volume neurotransmission.

However, in the nuclei studied (the striatum) dopamine synapses on dendrites of the major cell type (the medium spiny neuron) are well described and the 5 receptors for dopamine (called G Protein Coupled Receptors — GPCRs) are found there.  None of the GPCRs conduct ions or trigger action potentials (immediately anyway).  Instead, they produce their effects much more slowly and change the metabolism of the interior of the cell.  This is true for all GPCRs, regardless of the ligand activating them — and humans have 826 GPCR genes.

Note also that volume neurotransmission means that dopamine reaches nonNeuronal tissue — and there is good evidence that dopamine receptors are present on glial cells, pericytes and blood vessels.

The story doesn’t end with dopamine.  There are 3 other similar systems of small numbers of neurons collected into nuclei, using different neurotransmitters, but whose axons branch and branch so they go all over the brain.

These are the locus coeruleus which uses norepinephrine as a neurotransmitter, the dorsal raphe nucleus which uses serotonin and the basal nucleus of Meynert which uses acetyl choline.  There is excellent evidence that the first two (norepinephrine and serotonin) use volume neurotransmission. I’m not sure about those of the basal nucleus of Meynert.

What is so remarkable about the paper, that it allows the receiving neurons to (partially) control what dopamine input it gets.

All norepinephrine receptors are GPCRs, while only one of the 16 or so serotonin receptors conducts ions, the rest being GPCRs.

Acetyl choline does have one class of receptors (nicotinic) which conducts ions, and which the paper shows is what is triggering the axon on axon synapse.  The other class (muscarinic) of acetyl choline receptor is a GPCR.

Addendum 29 September — it goes without saying (although I didn’t say it) that any molecule released by volume neurotransmission doesn’t confine itself to finding targets on neurons.  Especially with norepinephrine, it could bind to receptors for it on the vasculature causing circulatory effects.  They could also bind to GPCRs on pericytes and glia.

Now the paper tested axon to axon firing in one of the four systems (dopamine) in one of the places its axons goes (the striatum).  There is no question that the axons of all 4 systems ramify widely.

Suppose axon to axon firing is general, so a given region can control in someway how much dopamine/serotonin/norepinephrine/acetyl choline it is getting.

Does this remind you of any system you are familiar with?  Perhaps because my wife went to architecture school, it reminds me of an old apartment building, with separate systems to distribute electricity, plumbing, steam heat and water to each apartment, which controls how much of each it gets.

Perhaps these four systems are basically neurological utilities, necessary for  the function of the brain, but possibly irrelevant to the computations it is carrying out, like a mother heating a bottle for her baby in water on a gas stove on a cold winter night.  The nature of steam heat, electricity, water and gas tell you very little about what is going on in her apartment.

The paper is so new (the Neuron issue of 21 September) that more implications are sure to present themselves.

Quibbles are sure to arise.  One is that fact that the gray matter of our brain doesn’t contain much in the way of neurons using acetyl choline as a neurotransmitter.  What it does have is lots of neurons using GABA which we know can act on axons, inhibiting axon potential generation.  This has been well worked out with synapses where the axon emerges from the neuron cell body (the initial segment).  However the different ionic composition of axons in the developing brain results in GABA having an excitatory effect.  Perhaps ionic composition varies in different parts of the neuron.

The work was done in living animals, so the paper contains no electron micrographs.  Such work is sure to be done.  No classical presynaptic apparatus may be present, just two naked axons touching each other and interacting by ephaptic transmission (the term does not appear in the paper).

So a lot of work should be done, the first of which should be replication. As the late Carl Sagan said “extraordinary claims require extraordinary evidence”.

Finally:

As Mark Twain said ” There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact.”

 

Memories are made of this ?

Back in the day when information was fed into computers on punch cards, the data was the holes in the paper not the paper itself. A far out (but similar) theory of how memories are stored in the brain just got a lot more support [ Neuron vol. 93 pp. 6 -8, 132 – 146 ’17 ].

The theory says that memories are stored in the proteins and sugar polymers surrounding neurons rather than the neurons themselves. These go by the name of extracellular matrix, and memories are the holes drilled in it which allow synapses to form.

Here’s some stuff I wrote about the idea when I first ran across it two years ago.

——

An article in Science (vol. 343 pp. 670 – 675 ’14) on some fairly obscure neurophysiology at the end throws out (almost as an afterthought) an interesting idea of just how chemically and where memories are stored in the brain. I find the idea plausible and extremely surprising.

You won’t find the background material to understand everything that follows in this blog. Hopefully you already know some of it. The subject is simply too vast, but plug away. Here a few, seriously flawed in my opinion, theories of how and where memory is stored in the brain of the past half century.

#1 Reverberating circuits. The early computers had memories made of something called delay lines (http://en.wikipedia.org/wiki/Delay_line_memory) where the same impulse would constantly ricochet around a circuit. The idea was used to explain memory as neuron #1 exciting neuron #2 which excited neuron . … which excited neuron #n which excited #1 again. Plausible in that the nerve impulse is basically electrical. Very implausible, because you can practically shut the whole brain down using general anesthesia without erasing memory. However, RAM memory in the computers of the 70s used the localized buildup of charge to store bits and bytes. Since charge would leak away from where it was stored, it had to be refreshed constantly –e.g. at least 12 times a second, or it would be lost. Yet another reason data should always be frequently backed up.

#2 CaMKII — more plausible. There’s lots of it in brain (2% of all proteins in an area of the brain called the hippocampus — an area known to be important in memory). It’s an enzyme which can add phosphate groups to other proteins. To first start doing so calcium levels inside the neuron must rise. The enzyme is complicated, being comprised of 12 identical subunits. Interestingly, CaMKII can add phosphates to itself (phosphorylate itself) — 2 or 3 for each of the 12 subunits. Once a few phosphates have been added, the enzyme no longer needs calcium to phosphorylate itself, so it becomes essentially a molecular switch existing in two states. One problem is that there are other enzymes which remove the phosphate, and reset the switch (actually there must be). Also proteins are inevitably broken down and new ones made, so it’s hard to see the switch persisting for a lifetime (or even a day).

#3 Synaptic membrane proteins. This is where electrical nerve impulses begin. Synapses contain lots of different proteins in their membranes. They can be chemically modified to make the neuron more or less likely to fire to a given stimulus. Recent work has shown that their number and composition can be changed by experience. The problem is that after a while the synaptic membrane has begun to resemble Grand Central Station — lots of proteins coming and going, but always a number present. It’s hard (for me) to see how memory can be maintained for long periods with such flux continually occurring.

This brings us to the Science paper. We know that about 80% of the neurons in the brain are excitatory — in that when excitatory neuron #1 talks to neuron #2, neuron #2 is more likely to fire an impulse. 20% of the rest are inhibitory. Obviously both are important. While there are lots of other neurotransmitters and neuromodulators in the brains (with probably even more we don’t know about — who would have put carbon monoxide on the list 20 years ago), the major inhibitory neurotransmitter of our brains is something called GABA. At least in adult brains this is true, but in the developing brain it’s excitatory.

So the authors of the paper worked on why this should be. GABA opens channels in the brain to the chloride ion. When it flows into a neuron, the neuron is less likely to fire (in the adult). This work shows that this effect depends on the negative ions (proteins mostly) inside the cell and outside the cell (the extracellular matrix). It’s the balance of the two sets of ions on either side of the largely impermeable neuronal membrane that determines whether GABA is excitatory or inhibitory (chloride flows in either event), and just how excitatory or inhibitory it is. The response is graded.

For the chemists: the negative ions outside the neurons are sulfated proteoglycans. These are much more stable than the proteins inside the neuron or on its membranes. Even better, it has been shown that the concentration of chloride varies locally throughout the neuron. The big negative ions (e.g. proteins) inside the neuron move about but slowly, and their concentration varies from point to point.

Here’s what the authors say (in passing) “the variance in extracellular sulfated proteoglycans composes a potential locus of analog information storage” — translation — that’s where memories might be hiding. Fascinating stuff. A lot of work needs to be done on how fast the extracellular matrix in the brain turns over, and what are the local variations in the concentration of its components, and whether sulfate is added or removed from them and if so by what and how quickly.

—-

So how does the new work support this idea? It involves a structure that I’ve never talked about — the lysosome (for more info see https://en.wikipedia.org/wiki/Lysosome). It’s basically a bag of at least 40 digestive and synthetic enzymes inside the cell, which chops anything brought to it (e.g. bacteria). Mutations in the enzymes cause all sorts of (fortunately rare) neurologic diseases — mucopolysaccharidoses, lipid storage diseases (Gaucher’s, Farber’s) the list goes on and on.

So I’ve always thought of the structure as a Pandora’s box best kept closed. I always thought of them as confined to the cell body, but they’re also found in dendrites according to this paper. Even more interesting, a rather unphysiologic treatment of neurons in culture (depolarization by high potassium) causes the lysosomes to migrate to the neuronal membrane and release its contents outside. One enzyme released is cathepsin B, a proteolytic enzyme which chops up the TIMP1 outside the cell. So what. TIMP1 is an endogenous inhibitor of Matrix MetalloProteinases (MMPs) which break down the extracellular matrix. So what?

Are neurons ever depolarized by natural events? Just by synaptic transmission, action potentials and spontaneously. So here we have a way that neuronal activity can cause holes in the extracellular matrix,the holes in the punch cards if you will.

Speculation? Of course. But that’s the fun of reading this stuff. As Mark Twain said ” There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact.”

The most interesting thing to an evolutionist is not that APOE4 increases the risk of Alzheimer’s disease

Neurologists were immensely excited by the discovery 25 years ago that the APOE4 variant of APOlipoprotein E increases the risk of Late Onset Alzheimer’s Disease (LOAD). 24,000 papers later (Google Scholar) we still don’t know how it does it. Should all this work have been done ? Of course ! !  Once we know the mechanism(s) by which APOE4 increases Alzheimer’s risk we’ll have new ideas to help us attack.

The APOE gene has 3 variants (alleles) APOE2, 3 and 4. The protein is average sized (299 amino acids). The 3 alleles differ at two positions (amino acids #112 and #158) where either cysteine or arginine can be found. The frequency of APOE4 is 14% in the adult white population, that of E3 is 78% and that of E2 is 8%.

Fascinating as this all is, it’s not what’s interesting from an evolutionary point of view.

[ Proc. Natl. Acad. Sci. vol. 113 pp. 17 – 18, 74 – 79 ’16 ] Postmenpausal longevity in females is not limited to humans. Humans, orcas and pilot whales are the only vertebrate species known to have prolonged postreproductive lifespans. Our fertility ends at about the same age that fertility ends in other female hominids (the great apes). However, apes rarely live into their 40s (even in captivity).

Unlike APOE4, APOE2 and APOE3 protect against late onset Alzheimer’s.

The fascinating point is that APOE2 and APOE3 aren’t found in the great apes. They are a human invention. Now LOAD occurs well past reproduction, so there should be no reason in terms of reproductive success for them to arise and be more common in human populations than the original APOE4.

Even more interesting is some work on another protein CD33, found on immune cells and glia in the brain [ Neuron vol. 78 pp. 575 – 577, 631 – 643 ’13 ] A minor allele (21% frequency in human populations) of CD33 (SNP rs 3865444) protects against Alzheimer’s. The allele is associated with reductions in CD33 expression in microglia, and also with reduction in levels of insoluble Abeta42 in (Alzheimer’s) brain. The numbers of CD33+ microglia correlate with insoluble Abeta42 levels and amyloid plaque burden. So decreasing (or inhibiting) CD33 function might help Alzheimer patients.

Again the protective allele is only found in man. The great apes don’t have it just the major (nonprotective) allele.

Again, there is no way that having the allele directly improves your reproductive success. By the time it is protecting you, you’re infertile.

What in the world is going on? Why did alleles protective against Alzheimer’s arise in two very different proteins in the course of human evolution?

“There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact.” — Mark Twain.

The reason these alleles probably arose gets us in to an ancient battle in evolutionary theory — what is the actual unit of selection? It may be the group rather than the individual. Face it, human infants and children are helpless for longer than other primates, and need others to care for them, for at least 5 years. Who better than grandma and grandpa? So the fact that with granny around more children survive to reproduce constitutes group selection (I think).

As Theodosius Dobzhansky said “Nothing in Biology Makes Sense Except in the Light of Evolution”

Physics to the rescue

It’s enough to drive a medicinal chemist nuts. General anesthetics are an extremely wide ranging class of chemicals, ranging from Xenon (which has essentially no chemistry) to the steroid alfaxalone which has 56 carbons. How can they possibly have a similar mechanism of action? It’s long been noted that anesthetic potency is proportional to lipid solubility, so that’s at least something to hang your hat on.

Other work has noted that enantiomers of some anesthetics vary in potency implying that they are interacting with something optically active (like proteins). However, you should note sphingosine which is part of many cell membrane lipids (gangliosides, sulfatides etc. etc.) contains two optically active carbons.

A great paper [ Proc. Natl. Acad. Sci. vol. 111 pp. E3524 – E3533 ’14 ] notes that although Xenon has no chemistry it does have physics. It facilitates electron transfer between conductors (clearly a physical effect). The present work does some quantum mechanical calculations purporting to show that Xenon can extend the highest occupied molecular orbital (HOMO) of an alpha helix so as to bridge the gap to another helix.

This paper shows that Xe, SF6, NO and chloroform cause rapid increases in the electron spin content of Drosophila (probably another physical effect). The changes are reversible. Anesthetic resistant mutant strains (in what protein) show a different pattern of spin responses to anesthetic.

So they think general anesthetics might work by perturbing the electronic structure of proteins. It’s certainly a fresh idea.

What is carrying the anesthetic induced increase in spin? Speculations are bruited about. They don’t think the spin changes are due to free radicals. They favor changes in the redox state of metals. Could it be due to electrons in melanin (the prevalent stable free radical in flies). Could it be changes in spin polarization? Electrons traversing chiral materials can become spin polarized.

Why this should affect neurons isn’t known, and further speculations are given (1) electron currents in mitochondria, (2) redox reactions where electrons are used to break a disulfide bond.

Fascinating paper, and Mark Twain said it the best “There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact.”

Physics to the rescue

It’s enough to drive a medicinal chemist nuts. General anesthetics are an extremely wide ranging class of chemicals, ranging from Xenon (which has essentially no chemistry) to a steroid alfaxalone which has 56 carbons. How can they possibly have a similar mechanism of action? It’s long been noted that anesthetic potency is proportional to lipid solubility, so that’s at least something. Other work has noted that enantiomers of some anesthetics vary in potency implying that they are interacting with something optically active (like proteins). However, you should note sphingosine which is part of many cell membrane lipids (gangliosides, sulfatides etc. etc.) contains two optically active carbons.

A great paper [ Proc. Natl. Acad. Sci. vol. 111 pp. E3524 – E3533 ’14 ] notes that although Xenon has no chemistry it does have physics. It facilitates electron transfer between conductors. The present work does some quantum mechanical calculations purporting to show that
Xenon can extend the highest occupied molecular orbital (HOMO) of an alpha helix so as to bridge the gap to another helix.

This paper shows that Xe, SF6, NO and chloroform cause rapid increases in the electron spin content of Drosophila. The changes are reversible. Anesthetic resistant mutant strains (in what protein) show a different pattern of spin responses to anesthetic.

So they think general anesthetics might work by perturbing the electronic structure of proteins. It’s certainly a fresh idea.

What is carrying the anesthetic induced increase in spin? Speculations are bruited about. They don’t think the spin changes are due to free radicals. They favor changes in the redox state of metals. Could it be due to electrons in melanin (the prevalent stable free radical in flies). Could it be changes in spin polarization? Electrons traversing chiral materials can become spin polarized.

Why this should affect neurons isn’t known, and further speculations are given (1) electron currents in mitochondria, (2) redox reactions where electrons are used to break a disulfide bond.

The article notes that spin changes due to general anesthetics differ in anesthesia resistant fly mutants.

Fascinating paper, and Mark Twain said it the best “There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact.”