The cholinergic system

Acetylcholine is not made evenly across the brain. Mesulam's group sorted the projection neurons into six sectors, Ch1 to Ch6, classified by where their axons go rather than by which anatomical nucleus their cell bodies happen to sit in, and that nomenclature is still the working map ^[3]. Ch1 and Ch2, in the medial septal nucleus and the vertical limb of the diagonal band, supply the hippocampus. Ch3 supplies the olfactory bulb. Ch4, the nucleus basalis of Meynert, supplies the entire cortical mantle and the amygdala, and it is the sector that matters most for everything below. Ch5 and Ch6, in the pedunculopontine and laterodorsal tegmental nuclei of the upper brainstem, project forward to the thalamus and to the midbrain dopamine cells instead of to cortex.

The nucleus basalis deserves a moment on its own. It sits at the confluence of the limbic system and the reticular activating system: it receives dopamine from the ventral tegmental area and substantia nigra, serotonin from the raphe nuclei and noradrenaline from the locus coeruleus, and it returns the principal cholinergic supply to cortex and amygdala. More than half of the varicosities along its cortical axons form conventional synaptic contacts rather than simply spilling transmitter into the surrounding tissue, so the signal is more addressed than the phrase "diffuse neuromodulation" suggests ^[4]. Its limbic wiring is the anatomical reason acetylcholine influences not only whether information arrives but how much weight that information is given.

Two more populations matter and are usually left out of the popular account. The striatum holds cholinergic interneurons that project nowhere outside it. They are only about one to two percent of striatal cells, yet they arborise so densely that the striatum carries the highest acetylcholine concentration anywhere in the brain, and their firing pauses at behaviourally salient moments ^[9]. Outside the brain entirely, acetylcholine is the transmitter at every neuromuscular junction, at every autonomic ganglion, and at the parasympathetic organ endings. That peripheral system is why cholinergic side effects are so recognisable: sweating, salivation, gut cramping, a slowed heart, blurred near vision.

Acetylcholine is chemically trivial: choline with an acetyl group esterified onto it. Choline acetyltransferase (ChAT) joins the two, a vesicular transporter packs the product into vesicles, and release is ordinary calcium-dependent exocytosis. Nothing in that chain is unusual. What is unusual is where the bottleneck sits.

The rate-limiting step is not the enzyme; it is getting choline into the neuron in the first place. That job belongs to CHT1, a sodium-dependent high-affinity choline transporter, and most CHT1 protein is not on the cell surface at all. It sits on endosomes and synaptic vesicles inside the terminal and is delivered to the membrane along with transmitter release, so a neuron that has just fired is a neuron that has just improved its own ability to recapture choline ^[10]. This is the mechanistic reason a choline supplement is a plausible lever at all, and simultaneously the reason it is a weak one; the transporter, not the substrate, is usually what limits supply.

Clearance is where acetylcholine is genuinely exceptional. Most transmitters are removed by being pumped back whole into the cell that released them. Acetylcholine is destroyed in the cleft instead, by acetylcholinesterase, which is among the fastest enzymes in biology and cuts a synaptic pulse down to roughly a millisecond. Only the choline half is recovered and reused. Two consequences follow. First, cholinergic signalling can be far more precisely timed than a slow neuromodulator has any right to be. Second, blocking that one enzyme is an unusually powerful intervention, which is why cholinesterase inhibitors are simultaneously a class of Alzheimer's drug and a class of nerve agent.

Choline itself is a nutrient with an awkward status. The body makes some of its own through the PEMT pathway, but not reliably enough for everyone, and common genetic variation moves the requirement around. The US Institute of Medicine formally recognised choline as an essential nutrient in 1998 and set an adequate intake rather than a full recommended allowance, because the evidence was not strong enough for one; measured intakes for men, women, older children and pregnant women all sit well below that level, with eggs and meat the densest ordinary sources ^[11]. See choline sources for how the supplemental forms differ from each other and from food.

The old picture of acetylcholine was a slow bath: tone rising and falling over minutes, setting a general state. Direct measurement has largely overturned that. Choline-sensitive microelectrodes fast enough to resolve subsecond events showed that in rats performing a cue-detection task, cues the animal actually detected evoked a sharp cholinergic transient in prefrontal cortex, and missed cues evoked nothing ^[5]. The transient tracked detection rather than the cue itself, it did not appear in a non-associational control region, and removing the cholinergic input to prefrontal cortex impaired detection specifically.

The reinterpretation that followed is worth stating plainly: forebrain cholinergic signalling that drives cognition is largely phasic, on the scale of milliseconds to seconds, and unlikely to be volume-transmitted ^[7]. Much of the older evidence for a slow diffuse signal came from methods that needed several minutes to produce one data point and that had to inhibit acetylcholinesterase in order to measure anything at all, which is close to studying a system after removing the thing that shapes it.

What the signal does once it arrives is best described as biasing a circuit toward taking new input seriously. Acetylcholine increases the strength of feedforward sensory input relative to internal feedback, contributes to theta rhythm, activates intrinsic mechanisms for persistent firing, and makes synapses easier to modify. In entorhinal cortex, perirhinal cortex and hippocampus that combination favours encoding new episodes over retrieving old ones ^[8]. That is the honest version of the highlighter metaphor. It is not that acetylcholine improves memory in general; it is that it shifts a circuit from replay toward recording, and the two are in competition.

There are only four places to push on this system, and they are not equally good. Two are worth knowing in detail here; the other two are questions about receptor subtypes and belong to acetylcholine receptors.

Adding substrate is the gentlest lever and the least reliable. The rodent work is real: alpha-glycerylphosphorylcholine given orally reverses scopolamine-induced amnesia in rats and partially restores acetylcholine in hippocampus and cortex, although at 600 mg/kg, a dose that does not map onto a human capsule ^[16]. In people the evidence is thinner and sits almost entirely in impaired populations. The Cochrane review of CDP-choline found a benefit on memory and behaviour in older adults with chronic cerebral disorders, no effect on attention, and a trial base too short to say anything about the long term ^[15]. Observational data is suggestive rather than causal: in the Framingham Offspring cohort, higher choline intake tracked with better verbal and visual memory and with less white-matter hyperintensity on MRI ^[17]. There is no convincing controlled trial of a choline source in healthy young adults.

Sparing the transmitter is the strongest lever by a wide margin. Inhibiting acetylcholinesterase leaves more acetylcholine surviving each pulse, and it is the only cholinergic intervention with a large randomised evidence base behind it. Donepezil, galantamine and rivastigmine are approved for Alzheimer's disease, and the pooled Cochrane analysis found them consistently better than placebo on cognition and global function; the effects are modest and symptomatic, and none of them slows the disease ^[14]. Huperzine A is the version sold as a supplement, and its evidence base is much weaker than the prescription drugs it is compared to.

The remaining two levers, direct receptor agonism and receptor blockade, cannot be evaluated at system level at all, because their effects depend entirely on which family and which subtype is engaged. The short version: pushing the nicotinic side tends to be alerting, pushing the muscarinic side tends to be autonomically noisy, and blocking the muscarinic side hard produces confusion and genuine hallucination rather than anything anyone enjoys.

Alzheimer's disease is where this system is best understood and most studied. Postmortem work published in 1982 showed that neurons of the nucleus basalis of Meynert undergo a profound and selective degeneration of more than 75 percent in Alzheimer's disease and senile dementia, which was the first documentation of a transmitter-specific neuronal population being lost in a major disorder of higher cortical function ^[12]. In the same year a review pulled the biochemical, electrophysiological and pharmacological strands together into the cholinergic hypothesis of geriatric memory dysfunction ^[13]. That hypothesis is the direct reason the first approved drugs for the disease were cholinesterase inhibitors.

The hypothesis has aged in an instructive way. It is no longer treated as the cause of Alzheimer's disease; the cholinergic loss follows tau pathology in the basal forebrain rather than initiating the illness. But the loss is early, and because Ch4 is a small population fanning out across the whole cortex, neurofibrillary degeneration in a comparatively small number of neurons perturbs neurotransmission in every cortical area simultaneously ^[4][2]. That amplification is why a modest lesion carries an outsized cognitive cost, and why propping up the surviving signal helps at all.

Cholinergic failure also shows up acutely and peripherally. Anticholinergic delirium is a standard clinical picture, blocking muscarinic transmission with scopolamine is the oldest pharmacological model of amnesia in healthy volunteers, and the cumulative anticholinergic burden of ordinary prescriptions is an active concern in older adults. Myasthenia gravis is an autoimmune attack on the nicotinic receptor at the neuromuscular junction, treated with cholinesterase inhibitors. Organophosphate pesticides and nerve agents are irreversible cholinesterase inhibitors, and their lethality is a direct consequence of how fast that enzyme normally has to work.

One arm of the system has nothing to do with cognition and is easy to miss entirely: the cholinergic anti-inflammatory reflex. Vagal cholinergic signalling restrains the acute inflammatory response in real time, in the same reflexive way the nervous system controls heart rate, which turned the vagus nerve into a therapeutic target rather than only a conduit ^[18]. It is a useful reminder that "cholinergic" describes a chemistry, not a function.

More acetylcholine is not better. Push the system hard, most easily by stacking a choline source with a cholinesterase inhibitor, and the result is cholinergic overload: headache, a heavy fogged feeling, low mood, nausea, gut cramping, tight muscles, and a wired but tired quality. Nothing in the biology promises a rising dose response, and the peripheral receptors have no interest in your reading comprehension. The interactions and stacks tool flags exactly these combinations.

The mirror image is the racetam headache, the most commonly reported effect in this corner of the catalogue and the least well documented. The usual account is that a racetam drives cholinergic circuits harder than the available transmitter supply can sustain, and that adding Alpha-GPC or CDP-Choline relieves it. That story fits the biology and has essentially no controlled evidence behind it. Treat it as folk pharmacology with a plausible mechanism, not as a finding. Pyrrolidones and the racetam family covers why only some members of that family have any cholinergic action at all.

Several things here are genuinely unsettled, and saying so is more useful than smoothing over the gaps. Whether raising choline intake changes acetylcholine synthesis meaningfully in a healthy, well-fed adult brain is not established; the transporter step argues against it and the human trials sit in impaired populations ^[15][10]. Whether phasic transients or slower background tone matter more for a given cognitive operation is still actively argued ^[7][6]. Why cholinesterase inhibitors help some patients clearly and others not at all has no accepted answer ^[14]. And the relationship between acetylcholine and dopamine in the striatum is a balance rather than a level, which is exactly the sort of thing a fixed daily dose is badly suited to tune ^[9]. See nootropic stacking for how that principle generalises across the rest of the catalogue.