The serotonergic system

The serotonergic cell bodies sit near the midline of the brainstem along its whole length, in the raphe nuclei, and they split into two clusters that barely overlap in function ^[3][4]. The rostral group, in the midbrain and upper pons, projects forward to the entire forebrain; this is the half that psychiatric drugs and psychedelics act on. The caudal group, in the lower pons and medulla, projects backward and downward to other brainstem nuclei and to three parallel columns in the spinal cord, where it modulates pain, motor tone and autonomic drive.

Within the rostral group the dorsal raphe and the median raphe send parallel and overlapping projections to many of the same forebrain targets, but with structurally different axons: fine, highly branched fibres from the dorsal raphe, thicker beaded fibres from the median raphe ^[4]. That is not a trivia detail. The two populations respond differently to drugs and to damage, and a compound that changes serotonergic tone is not changing one thing evenly.

The most consequential anatomical fact is how much of the signalling is not synaptic. Serotonergic axons arborise over very large areas comprising functionally diverse targets, and while some terminals form classical chemical synapses, many do not; transmitter is released into the extracellular space and acts at a distance ^[4][5]. What that buys is reach, and what it costs is precision. Where the specificity actually comes from is the receptor: which subtype a given target cell expresses determines what a uniform bath of serotonin does to it ^[5]. That is why the receptor lesson beside this one is not optional reading.

Serotonin is made in two steps from the essential amino acid tryptophan. A hydroxylase converts tryptophan to 5-hydroxytryptophan (5-HTP), then a decarboxylase converts 5-HTP to serotonin. The first step is the slow one and the regulated one; the second is fast and almost indiscriminate.

For forty years it was assumed there was one tryptophan hydroxylase. There are two. TPH1 works in the periphery, chiefly in the gut, and TPH2 works in serotonergic neurons, and the second isoform was only identified in 2003 ^[6]. That discovery is the molecular reason gut serotonin and brain serotonin are genuinely independent systems rather than one pool with a barrier between them. Serotonin itself does not cross the blood brain barrier, so the 90 percent sitting in the gut and in platelets is unavailable to the brain no matter how much of it there is.

What does cross is tryptophan, and it crosses by competing for the large neutral amino acid transporter against several other amino acids that are far more abundant in dietary protein. This is why brain tryptophan availability tracks the tryptophan-to-competitor ratio rather than the absolute amount eaten, and it is the mechanism behind the acute tryptophan depletion experiments that dominate the next section. It is also why a tryptophan-rich food eaten as part of a normal protein meal does very little.

Clearance runs the usual monoamine route. The serotonin transporter (SERT) recaptures released 5-HT, and monoamine oxidase breaks down what is recaptured, producing 5-HIAA. SERT is the target of the SSRIs; MAO is the target of the oldest antidepressants and the reason for the single most dangerous drug interaction on this site.

Serotonin acts on seven receptor families and fourteen subtypes, of which all but one are G protein-coupled receptors and one, 5-HT3, is a ligand-gated ion channel. The full tour lives in serotonin receptors; the point to carry away here is that the receptor, not the transmitter, is what determines the effect. A single molecule that reaches everything can still produce completely different outcomes in two adjacent cells, purely because they express different subtypes.

Two subtypes account for most of what a reader of this site will encounter. 5-HT1A is broadly calming and mood-supporting, and it exists both as a somatodendritic autoreceptor on the raphe neurons themselves, where activation slows them down, and as a postsynaptic receptor on their targets. 5-HT2A is the receptor responsible for the effects of the classic psychedelics; essentially every tryptamine and phenethylamine psychedelic in this catalogue is at its core a 5-HT2A agonist ^[2]. See psychedelics and 5-HT2A.

One framework worth knowing ties those two together rather than treating them as unrelated. It proposes that the serotonin system supports two different adaptive responses to adversity: a 5-HT1A route that moderates stress and supports tolerating a difficult situation, enhanced by SSRIs, and a 5-HT2A route that increases plasticity and supports actively changing the situation or one's relationship to it, enhanced by psychedelics ^[13]. It is a hypothesis rather than a settled result, and it is the most useful available answer to the question of how two drug classes that both work on serotonin can feel so completely unalike.

There are seven ways to move this system, and they are not variations on one theme. The table sorts them; a few deserve comment.

Precursors are the gentlest and the most misunderstood. 5-HTP skips the regulated hydroxylase step entirely, which sounds efficient and is the problem: the decarboxylase that converts it is present throughout the body, not just in serotonin neurons, so a large share of an oral dose becomes serotonin in the periphery where it cannot help mood and can cause nausea. The controlled human literature is old, small, and thin on long-term safety ^[14]. Plain tryptophan is more physiological because the regulated step is left in place, and correspondingly less dramatic.

Reuptake inhibition is the best-evidenced lever in psychiatry and still a modest one. A network meta-analysis of 21 antidepressants across the acute treatment of major depression in adults found all 21 more effective than placebo, with effect sizes that were mostly modest and broadly similar to each other ^[11]. What that literature does not establish is a serotonin deficit being corrected; a drug working is not proof of the mechanism it was designed around.

Releasers are the opposite of gentle. MDMA and fenfluramine reverse the transporter and empty vesicles wholesale, which produces a signal no physiological firing pattern could, and which carries its own risk profile; see entactogens. And MAO inhibition, the oldest lever of all, is effective and is the single most dangerous combination partner in this catalogue, for reasons the last section covers.

This is the part most articles get wrong, and it is worth reading carefully because the honest answer is more interesting than either of the confident ones.

The classic experiment is acute tryptophan depletion: give a drink of amino acids with no tryptophan in it, which strips brain tryptophan by outcompeting it at the transporter, and see what happens to mood. The 1985 study that started the line found that healthy young men given a tryptophan-free mixture had significantly elevated depression scores five hours later and performed worse on a task while listening to a dysphoric distractor ^[8]. That result is real and it is repeatedly cited as proof that low serotonin causes low mood.

The meta-analysis of the whole literature says something more specific. Pooling 45 tryptophan depletion studies, depletion did not lower mood in healthy controls at all. It slightly lowered mood in healthy people with a family history of depression, moderately lowered it in patients with major depression currently in remission, and reliably induced relapse in remitted patients who were taking a serotonergic antidepressant ^[9]. The system therefore looks less like a mood thermostat and more like something whose failure is exposed by vulnerability. Low serotonin does not appear to make a healthy person depressed.

The 2023 umbrella review pushed further and is worth stating exactly. Across serotonin and 5-HIAA concentrations in body fluids, receptor binding, transporter imaging and post-mortem measurement, tryptophan depletion, and serotonin transporter gene association and gene-environment studies, it found no consistent evidence of an association between serotonin and depression, and no support for the hypothesis that depression is caused by lowered serotonin activity ^[10]. That does not mean antidepressants do not work; the trial evidence that they beat placebo is separate and stands ^[11]. It means the popular explanation of why they work has not been demonstrated, and the field has largely stopped claiming it has.

What serotonin does seem to do, on the positive evidence rather than the negative, is set how long you are willing to wait. Optogenetic activation of dorsal raphe serotonin neurons during a delayed reward task reduced premature give-ups and prolonged waiting, specifically during the deciding-whether-to-keep-waiting window, and control experiments ruled out simple motor inhibition or the stimulation itself being rewarding ^[12]. Patience, satiety and a sense of enough are a better summary of this system than happiness ever was ^[1].

Serotonin syndrome, more precisely serotonin toxicity, is the one hard safety rule on this page. Too much serotonergic activity at once produces a triad of altered mental state, autonomic instability and neuromuscular excitation: agitation and confusion, a fast heart rate and swinging blood pressure, sweating, tremor, hyperreflexia, and in severe cases hyperthermia, rigidity, seizures and death ^[15]. It is a drug effect rather than an idiosyncratic reaction, which means it is predictable from what is in the stack.

The most useful clinical formulation is the Hunter criteria, which reduced diagnosis to a small set of decision rules built around clonus as the cardinal sign: spontaneous clonus alone is sufficient, and inducible or ocular clonus plus agitation, sweating, tremor or hyperreflexia will do. Those rules turned out to be more sensitive and more specific than the older Sternbach criteria ^[16]. The practical read is that tremor and hyperreflexia after adding a serotonergic drug are not a curiosity to wait out.

The dangerous combinations are stereotyped. Any monoamine oxidase inhibitor plus any other serotonergic agent is the classic and the worst, and that includes combinations people do not expect: a MAOI plus an SSRI, plus tramadol, plus an entactogen like MDMA, plus 5-HTP. Among opioids the risk is specific rather than general; pethidine, tramadol, dextromethorphan and a few relatives have serotonergic activity while morphine, codeine and oxycodone do not, and that distinction is life-saving rather than academic ^[17]. The site's interactions and stacks tool flags exactly these pairs.

The second serious risk gets far less attention and is slower. Long-term agonism at the 5-HT2B receptor drives fibroblast proliferation in heart valves and causes valvular heart disease; that mechanism explains the withdrawal of fenfluramine and dexfenfluramine, and it applies to any serotonergic drug with meaningful 5-HT2B activity taken chronically ^[18]. It is the strongest argument for treating unfamiliar serotonergic research chemicals with more caution than their acute effects suggest, because nothing about the acute experience warns you.

Finally, the honest limits. Whether pushing serotonin upward helps a healthy person is not established, and the depletion literature suggests it mostly does not ^[9][10]. Precursors can flatten motivation or disturb sleep in some people, which is what a system that trades urgency for patience would be expected to do. And because the specificity in this system lives in fourteen receptor subtypes rather than in the transmitter, the single most useful thing a reader can do with a serotonergic compound is find out which subtypes it touches: serotonin receptors.