Anxiolytics & GABA

GABA is glutamate with its acidic tail removed, by an enzyme called glutamate decarboxylase (GAD). There are two forms of that enzyme and the difference between them is more interesting than it first appears. GAD67 is spread throughout the neuron and is almost entirely present as active holoenzyme, saturated with its cofactor. GAD65 sits mainly in axon terminals, and only about half of it is in the active form at any moment; the inactive reserve is thought to let GABA production track how hard the terminal is currently working ^[3]. In other words the brain keeps a bulk supply and a demand-responsive supply, and the demand-responsive one is at the synapse.

Inhibition then arrives in two flavours that are easy to conflate and are genuinely separate drug targets. Phasic inhibition is the classic version: GABA released into a synapse hits receptors clustered opposite the release site and produces a brief, sharp inhibitory current. Tonic inhibition comes from receptors sitting outside the synapse that are bathed in low ambient GABA and carry a persistent, non-decaying current ^[5]. The tonic current sets baseline excitability across a whole cell rather than timing a single event, and several important drugs, including some neurosteroids, act preferentially on it.

The most counterintuitive part is that inhibition is not a property of the receptor. GABA-A opens a chloride channel; whether chloride entering makes a cell harder or easier to fire depends entirely on the chloride gradient across its membrane, and that gradient is maintained by cation-chloride cotransporters, chiefly KCC2 pumping chloride out ^[6]. In the immature brain KCC2 expression is low, internal chloride is high, and GABA is depolarising rather than inhibitory; the same transmitter, the same receptor, an opposite sign ^[7]. That is a useful corrective to the mental model of GABA as a switch that always means "off", and it matters clinically wherever the gradient is disturbed.

Clearance is unglamorous and important: transporters called GATs pull GABA back into neurons and into surrounding astrocytes, and GABA transaminase then breaks down what is recovered. Nothing is destroyed in the cleft the way acetylcholine is. That recycling loop is the target of a small number of anticonvulsants and of almost nothing sold as a supplement.

Most fast-acting calming drugs act at the GABA-A receptor, a pentameric channel that opens to let chloride into the neuron when GABA lands on it, making the cell harder to fire ^[1]. What makes it such a rich drug target is that the protein carries several allosteric sites, extra dials that other molecules can grab in order to turn GABA's own effect up without being GABA themselves. Benzodiazepines, barbiturates, alcohol, neurosteroids and several general anaesthetics all use different dials on the same receptor.

That is why mixing them is dangerous in a way that is not merely additive in a vague sense. Several depressants converging on one channel from different sites produce a much larger effect than any of them alone, and the effect that stacks is respiratory depression. Combining alcohol with a benzodiazepine, or a benzodiazepine with an opioid, is the mechanism behind a large share of accidental overdose deaths, and no amount of tolerance to the subjective effect protects against it.

One result from this receptor is worth carrying at system level because it settled a long argument. Mice were engineered with a point mutation making a single GABA-A subunit insensitive to diazepam. In animals whose alpha1 subunit was made insensitive, diazepam lost its sedative and amnesic actions while keeping its anxiolytic-like effect ^[8]. In animals whose alpha2 subunit was made insensitive, the anxiolytic effect went and the sedation stayed ^[9]. Sedation and anxiolysis are therefore separable properties, carried by different subunits, which is the entire scientific basis for the search for a non-sedating benzodiazepine-like anxiolytic. That search has produced compounds that work in animals and has so far not produced a marketed drug, which is a fact worth sitting with. Subunit specifics belong to GABA receptors.

The neurosteroids deserve their own mention because they are endogenous. Allopregnanolone, a progesterone metabolite, is a potent and highly efficacious positive modulator of GABA-A, acting at its own site and with a marked preference for the extrasynaptic receptors carrying the tonic current ^[10]. The body makes its own benzodiazepine-like ligand, its level swings with the menstrual cycle and collapses after childbirth, and an intravenous formulation of it, brexanolone, produced a rapid and significant reduction in depression scores in a randomised placebo-controlled trial in severe postpartum depression ^[11]. See also ganaxolone, the orally active analogue.

Calming compounds form a ladder, and the rungs are not interchangeable. What changes as you climb is not only strength; it is how directly the compound potentiates GABA, and that is what predicts dependence. Indirect and non-GABAergic anxiolytics can be used casually. Anything that grabs an allosteric dial on GABA-A and turns it hard deserves genuine caution.

At the gentle end sit things whose mechanisms are often not GABAergic at all. L-Theanine from tea produces relaxation without sedation and its effects on stress and anxiety in acute settings have modest systematic-review support, with much weaker evidence for chronic use ^[21]. Adaptogens such as ashwagandha work on stress reactivity over weeks rather than on acute anxiety; a placebo-controlled trial in 64 chronically stressed adults found significantly lower scores on every stress scale used and substantially lower serum cortisol after 60 days ^[22]. See adaptogens and the HPA axis.

The middle of the ladder is where the trade begins in earnest. Phenibut is 4-amino-3-phenylbutanoic acid, a GABA analogue that kept both charged ends of the molecule and therefore genuinely reaches GABA-B receptors; it is effective for anxiety and it carries a real dependence and withdrawal syndrome ^[18]. It is the instructive counterexample to the racetams, which cyclised the same skeleton, lost both charges and lost the GABA activity along with them; see pyrrolidones and the racetam family. At the top of the ladder sit the benzodiazepines and the Z-drugs, which are effective, well documented, and the reason this page has a section on dependence.

A bottle labelled GABA is one of the most common calming supplements sold, and it is the clearest case on this site of a name doing all the persuading. GABA is a small, highly polar, doubly charged molecule at body pH, and molecules like that cross the blood brain barrier poorly. Whether any meaningful amount of an oral dose reaches the brain in an adult human is still not resolved, and the evidence that it does is limited and conflicting ^[20].

The clinical literature reflects that. A systematic review of 14 placebo-controlled human trials of natural or biosynthetic oral GABA concluded there is limited evidence for a benefit on stress and very limited evidence for a benefit on sleep ^[19]. Some individual studies do report changes, typically small shifts in EEG alpha and beta power or in mood questionnaires within half an hour of dosing, which is faster than a brain-penetration story comfortably explains and points toward peripheral or enteric mechanisms instead.

That is not the same as saying nothing happens. GABA receptors exist in the enteric nervous system and on peripheral tissue, and a gut-to-brain signal is a perfectly plausible route for a small subjective effect. What the evidence does not support is the implied mechanism on the label, which is that swallowing the brain's inhibitory transmitter tops up the brain's inhibitory transmitter. Picamilon, which joins GABA to niacin specifically to get it across the barrier, is the more interesting version of the idea, and its human evidence base is almost entirely Russian-language and old.

The practical conclusion is that the gentle end of the ladder is worth using, but not for the reason it is usually sold. L-Theanine, glycine, magnesium and the adaptogens have small, real effects through mechanisms that are mostly not direct GABA-A potentiation, and that is exactly why they do not cause tolerance.

The catch with strong GABAergic anxiolytics is that the brain adapts to being braked. Lean on it hard enough for long enough and the system compensates: receptor number and subunit composition shift, coupling between the benzodiazepine site and the GABA site uncouples, and glutamatergic signalling rises to meet the new inhibition ^[15]. The result is tolerance, needing more for the same effect, and then a rebound worse than baseline when the drug is removed, because the compensations are still in place and the thing they were compensating for is gone.

The rebound is not a mild inconvenience. Benzodiazepine withdrawal produces rebound anxiety and insomnia, and at the severe end confusion, psychosis and seizures; alcohol withdrawal in a physically dependent person can be fatal. Guidance has recommended limiting benzodiazepines to 2 to 4 weeks for decades, and doctors worldwide still prescribe them for months or years, which has produced large populations of long-term dependent users, many of them older adults taking hypnotics ^[17]. Withdrawal is achievable, and the two things that make it work are gradual dose tapering and psychological support; mental and physical health and cognitive performance improve after withdrawal, particularly in elderly patients ^[17][16].

Tolerance does not develop evenly across effects, and that unevenness is dangerous. Tolerance to the hypnotic and sedative actions of a benzodiazepine builds within days to weeks, while tolerance to the anxiolytic action builds more slowly and tolerance to respiratory depression builds least of all ^[15]. A person escalating the dose because sleep stopped working is escalating past a respiratory effect that has not adapted with them.

This is why the site's interactions and stacks tool treats stacking two strong depressants as a serious flag, and why the calming compounds worth reaching for first are the ones that do not form dependence. See tolerance and dependence for how the same logic applies across the rest of the catalogue.

It is tempting to close the loop and say anxiety is a GABA deficit. The evidence is more partial than that, and the gaps are worth naming.

There is real human evidence that GABA is low in some psychiatric states. Proton magnetic resonance spectroscopy found a 52 percent reduction in occipital cortex GABA in medication-free depressed patients compared with healthy controls, which was the first in vivo demonstration of abnormally low cortical GABA in depression ^[12]. Reviews of anxiety disorders describe a picture of disturbed GABAergic modulation rather than a simple shortage, with changes in receptor binding as well as in transmitter level ^[2][13][14].

What has not been shown is the step everyone wants. There is no demonstration that raising GABA levels in a person without a disorder produces a durable benefit, no established way to raise brain GABA with a supplement, and no accepted biomarker that would tell an individual whether their anxiety is GABA-related at all. The GABA-A receptor's own therapeutic ceiling is a related unsolved problem: the alpha1 and alpha2 knock-in experiments proved two decades ago that anxiolysis can be separated from sedation, and translating that into a marketed non-sedating anxiolytic has repeatedly failed ^[8][9][1].

Two further honest caveats. First, GABA is not only an inhibitory transmitter; during development it acts as a trophic factor influencing proliferation, migration, differentiation and synapse maturation, and its signalling properties then are different from and in some ways opposite to those in the adult brain ^[7]. Long-term GABAergic drug exposure is therefore not a neutral background condition, especially in the young. Second, this system is one half of a pair. Every claim about inhibition is implicitly a claim about the excitation it is balancing, which is the subject of NMDA, glutamate and memory, and the compensations that produce benzodiazepine withdrawal happen on the glutamate side ^[15].