DMT, LSD and psilocin made neural cells accumulate lactate. Their non-hallucinogenic look-alikes did not. Fuel, memory signal — or emergency brake?

There is a version of this story that is too good to be true, and it is worth naming it first so we can set it aside.

The seductive version goes: scientists have found the chemical fingerprint of a psychedelic experience. Drop DMT onto neurons and they light up with a molecule you can measure in a dish; drop on a compound that looks almost identical but doesn't make anyone hallucinate, and the signal stays flat. A biomarker of the trip, isolated at last.

That is not what happened. There was no consciousness in the dish. Nothing tripped. But something quieter, and in some ways stranger, did occur — and it is worth understanding precisely, because the honest version of this finding is more interesting than the hyped one.

What was actually found

Two papers from the same group are in play, and keeping them straight matters.

In November 2025, a team spanning Queen Mary University of London, the López-Neyra Institute in Granada and Virginia Commonwealth University posted a preprint mapping the signalling dynamics of psychedelics in a neural model — a phosphoproteomic survey, meaning they catalogued which proteins got switched on or off (via phosphorylation) after treatment. Out of thousands of events, they extracted a 73-site "signature" that separated hallucinogenic compounds from closely related non-hallucinogenic ones. Buried in it were proteins governing glucose handling and glycolysis, and one in particular — FOXK2, a transcription factor that ramps up aerobic glycolysis. When they ran a straightforward lactate assay to check whether this metabolic tilt was real, the hallucinogenic drugs raised extracellular lactate roughly two- to four-and-a-half-fold. Their sober structural twins did nothing.

So the raw observation — hallucinogens make lactate, their analogues don't — is already several months old. That is the correction worth flagging up front, because it is easy to write this as a bolt from the blue that landed three days ago. It didn't.

What is new is the peer-reviewed follow-up in Molecular Psychiatry (Taddei-Tardón and colleagues), which takes that metabolic observation and asks the next mechanistic question: what receptor signalling is required for it? And here the receptor-level answer is clean enough to be genuinely surprising.

The model is a mouse neural stem-cell line coaxed into becoming a mixed culture of neurons and glia — closer to real brain tissue than the engineered cell lines usually used for this work. Into it went a panel chosen for contrast:

All were compared at a single concentration — which makes the contrasts readable but rules out any claim about relative potency or full dose–response, a limitation the authors flag directly. It's worth holding that in mind before reading any absence of effect as an intrinsic property of a molecule.

Two results sit side by side, and their asymmetry is the whole story.

Plasticity was shared. Dendritic branching increased across both the hallucinogenic compounds and several of the non-hallucinogenic comparators. And that structural growth required two receptors working together: the classic psychedelic target 5-HT2A, and TrkB, the receptor for the growth factor BDNF. Silence TrkB and the dendritic response collapsed even for ketamine and pure TrkB agonists; silence 5-HT2A and the psychedelic-driven branching fell away specifically. This dovetails with the influential 2023 claim from Moliner and colleagues that psychedelics bind TrkB directly, and with the older Ly (2018) framework of psychedelics as "psychoplastogens." Growth, in other words, is not the hallmark of a hallucinogen. The sober twins grew dendrites too.

Lactate was not shared. Only the hallucinogenic four raised it. Knock down the 5-HT2A receptor and the lactate response vanished. Leave the receptor intact but chemically uncouple its downstream G-protein partners — Gq/11 with one blocker, Gi/o with another — and the lactate rise was generally abolished too. Both signalling arms feed it. (One rogue exception: blocking Gi/o actually boosted DOI's lactate output, a quirk the authors trace to the peculiar receptor selectivity of phenethylamines.)

Crucially, this is a claim about receptor signalling, not about plumbing. The study does not establish where the lactate physically comes from — which cell type made it, by what metabolic route, whether neurons or glia were responsible. The authors say plainly that the biochemical pathway connecting receptor activation to glycolytic flux and lactate release remains undefined. What they have pinned down is that the response needs 5-HT2A and leans on both G-protein arms to appear.

Put those together and you get the sentence that makes this study worth an essay:

In this system, lactate tracked whether a compound was hallucinogenic more faithfully than whether it grew dendrites.

A cellular correlate that tracked the panel's hallucinogenic classification, and a measure of structural plasticity, came apart on the bench. That doesn't cleanly sever plasticity from human phenomenology — phenomenology was never in the dish to measure — but it hands researchers a model in which the two molecular shadows no longer perfectly overlap. The property everyone wants from these drugs — rewiring — showed up regardless of hallucinogenic status. The metabolic signal showed up with it.

Why lactate stopped being garbage

To feel the weight of this, you have to abandon the version of lactate you were taught. For most of the twentieth century lactate was the ash of exertion — the waste your muscles dump when oxygen runs short, the ache after a sprint. In the brain it was treated as an inconvenient by-product to be cleared.

That picture is dead. Over two decades, lactate has been recharacterised as an energy substrate neurons actively burn, a courier that carries instructions between cells, and a participant in the machinery of learning. The foundational demonstration came from Suzuki and colleagues in 2011: during memory formation, astrocytes break down their glycogen stores and hand lactate to neurons, and if you block that hand-off, long-term memory fails in rats — unless you supply lactate directly, which rescues it. Lactate is not the exhaust of thinking. On this evidence it is closer to a raw material for it.

Which is why a hallucinogen-specific lactate pulse is provocative rather than trivial. If lactate were mere waste, "psychedelics make more of it" would mean "psychedelics make cells work harder," full stop. But lactate does things. So the finding forces a question the authors are careful not to answer, and which the rest of this piece is about:

Is the lactate a by-product of the trip, a mechanism of it, or a defence against it?

There are three serious answers, and no current experiment can rule any of them out.

Interpretation one — lactate is the receipt

The deflationary reading: 5-HT2A activation drives intense, complex cortical signalling, and that costs energy. Cells lean harder on glycolysis, and lactate spills out as the metabolic receipt for an expensive state. On this view lactate is a correlate — it faithfully reports that a hallucinogenic drug is doing hallucinogenic work, without being part of the work itself.

This is the null hypothesis, and it has thirty years of human data behind it. Back in 1997, Vollenweider's PET studies showed psilocybin raising cerebral glucose metabolism, especially across frontal regions — the brain visibly burning more fuel under a psychedelic. We have been watching altered metabolism accompany altered consciousness for a generation; we simply read it as regional "activation." The new dish work may be the cellular echo of that old scanner image, though glucose uptake on a PET scan and lactate in a culture medium are not the same measurement.

The preprint authors themselves reach for a more startling framing of the "expensive state" idea: they note that hallucinogens seem to push cells toward glycolysis in a way that echoes how the brain responds to low oxygen — hinting at a shared metabolic mode between the psychedelic state and physiological anoxia. That is a hypothesis, not a result, but it is a vivid one.

Interpretation two — lactate is how the moment becomes lasting

The most elegant reading is that lactate bridges the acute experience and the durable rewiring clinicians actually care about.

The pieces are there. Lactate is required for long-term memory (Suzuki). It can switch on plasticity-related genes by potentiating NMDA-receptor signalling. Work from the Oslo group and others shows lactate raising neurotrophic factors — BDNF, GDNF — partly through a dedicated lactate receptor called HCAR1. And BDNF's receptor, TrkB, is precisely the second receptor this new study found to be essential for psychedelic-driven dendritic growth.

So you can sketch a tidy circuit: the hallucinogenic drug hits 5-HT2A, the metabolic arm produces lactate, lactate feeds neurotrophic and plasticity-supporting programmes that converge on the same TrkB machinery driving the structural changes. The trip and the rewiring would then meet in the metabolism.

It's a lovely story, and it is unproven in two important ways. First, the study explicitly does not show lactate causing the plasticity — the researchers call lactate a metabolic output associated with hallucinogenic signalling, not a demonstrated mediator of dendrite growth. Second, the endpoints didn't move together: lactate, dendrite branching, synapse formation and immediate-early gene induction rose and fell in only partially overlapping patterns. If lactate were the master switch for plasticity, you'd expect tighter coupling. You don't see it.

There is a sobering companion result. A 2021 study (Lundquist and colleagues) found that giving neurons exogenous lactate did boost astrocytic neurotrophic signalling — and yet this was not sufficient to enhance synapse formation or change behaviour in mice. A biochemical plasticity signal is not the same thing as meaningful rewiring. That distinction should be tattooed onto the wrist of anyone writing about psychedelics, a field with a chronic habit of treating "more dendrites" as automatic evidence of healing.

Interpretation three — lactate is the brake

Here is the reading that turns the intuition inside out, and it is my favourite because it is the least obvious.

We keep assuming lactate must be part of the excitation — the accelerator of an intense neural state. But lactate has a receptor, HCAR1, and when you activate it, neurons get quieter. In 2022, Briquet and colleagues showed this directly: activating HCAR1 reduced neuronal excitability and cut the frequency of excitatory synaptic events — not only in rodent tissue but in living human brain slices resected from epilepsy surgery, where synaptic event frequency dropped by around 40%. Lactate, acting through this receptor, is a genuine neuromodulator, and its direction of effect is down. The same group has floated HCAR1 as a target for calming the over-excited epileptic brain.

Now re-read the finding. A hallucinogen throws the cortex into an intense, high-complexity state — and simultaneously triggers a surge of a molecule whose own receptor damps neural activity. What if the lactate pulse is not the fuel of the runaway state but the brain's attempt to contain it? Negative feedback. The system, sensing it has been kicked into overdrive by a 5-HT2A agonist, releasing its own soft brake.

That would be an extraordinary inversion: psychedelics generate the excitation; lactate helps stop it running away. But label the leap honestly. Briquet's study shows what activating HCAR1 can do; it does not show that the lactate measured in this psychedelic culture reached HCAR1, activated it meaningfully, or produced any inhibition. HCAR1 wasn't in these experiments at all. The brake is a hypothesis the data permit, not one they demonstrate — and it reframes the trip as a tug-of-war rather than a one-way push.

The DMT-shaped footnote

For a publication built around DMT, one detail deserves a spotlight. Among the hallucinogenic compounds, DMT raised lactate while, in the companion head-twitch work, evoking the weakest classic hallucinogenic behaviour in animals. DMT is also the shortest-acting of the bunch in the body — the molecule that, taken orally in ayahuasca, only lasts because harmala alkaloids protect it from breakdown. Work by Egger and colleagues on DMT together with harmine looked at how the harmala changes DMT's metabolism and at the resulting cerebral metabolic picture.

That does not yet show that harmalas reshape DMT-induced lactate signalling. But it makes the obvious next experiment unusually compelling: compare DMT alone, harmine alone and the combination in the same neural metabolic model, and see whether ayahuasca's pharmacology is a metabolic story as much as a pharmacokinetic one. The line has not yet been tested directly. It should be.

What the dish can and can't say

The temptation, having assembled all this, is to declare that consciousness has a measurable metabolic fingerprint. Resist it.

The cells were not tripping — a stem-cell-derived culture has no experience to have. What the researchers found is a molecular response that co-segregates with compounds that produce hallucinations in whole organisms. That is real and useful: a bench readout that separates psychedelic from non-psychedelic action more cleanly than the plasticity assays do, which could matter a great deal for screening next-generation compounds. But it is a correlation in a model, built on a single drug concentration, in mouse-derived cells, with the causal chain between receptor and lactate still undefined. The authors say so themselves, repeatedly.

And the central mystery survives all that caution. Two properties of these drugs, long assumed to be one thing, pulled apart on the bench: the rewiring the therapeutic hopes ride on, and the metabolic signal that tracks the trip. The compounds that don't make you hallucinate still grew dendrites. The ones that do carried an extra metabolic shadow.

So the question is not whether scientists have bottled the trip. They haven't. The question is quieter and better:

When a psychedelic alters a mind, is it also — necessarily, mechanically — altering the energetic economy that mind runs on? And is that metabolic shift the price of the state, the means of its lasting effect, or the brake that keeps it from consuming the system that hosts it?

Three answers, all still standing. That is a rare and honest place for a science story to end.

Sources drawn on: Taddei-Tardón et al., "Integrated 5-HT2A–TrkB and G protein signaling in serotonergic psychedelic responses," Molecular Psychiatry (2026), doi:10.1038/s41380-026-03746-6; Martin-Guerrero, Taddei-Tardón et al., "Quantitative phosphoproteomics uncovers the signalling dynamics of hallucinogenic psychedelics," bioRxiv 2025.11.24.690190 (Nov 2025); Suzuki et al., Cell 2011; Ly et al., Cell Reports 2018; Moliner et al., Nature Neuroscience 2023; Vollenweider et al., Neuropsychopharmacology 1997; Morland et al., Nature Communications 2017; Briquet et al., J. Cerebral Blood Flow & Metabolism 2022; Lundquist / Lev-Vachnish et al., 2021; Egger et al., 2023.