For psychedelic-induced structural plasticity (e.g. spinogenesis, the production of spines) to occur, a net production of new biological material must take place: membranes, cytoskeleton, postsynaptic density proteins. That leaves a few possibilities: either there is synthesis of synaptic proteins, or there is less degradation of existing synaptic proteins, or both. mTOR is the leading candidate for the synthesis side of the equation.
For over a decade, the leading candidate for ketamine's rapid antidepressant effect beyond NMDA receptor blockade has been mTOR, the mechanistic target of rapamycin, a kinase that governs protein synthesis and, downstream, the growth of new dendritic spines and synapses.
The foundational study came from the lab of Ron Duman in 2010 (Li et al., 2010). In rats, ketamine triggered rapid synaptogenesis in the prefrontal cortex, measurable within hours as increased synaptic proteins and new dendritic spines. When the researchers infused rapamycin, an mTOR inhibitor, directly into the prefrontal cortex before ketamine, both the synaptogenesis and the antidepressant-like behavioral response disappeared. The implication was clean and appealing: if you block protein synthesis, you block the antidepressant effect. This became the mechanistic backbone for how the field talks about rapid-acting antidepressants, ketamine included.
It's worth flagging a limitation in the method itself. Rapamycin was delivered by cannula directly into the prefrontal cortex, not systemically, and a local infusion doesn't guarantee even exposure. Diffusion radius, which cell types actually take up the drug, and how long effective concentrations persist at the synapse are all somewhat opaque in this kind of preparation. In practice, that means the study reports a clean yes-or-no result (synaptogenesis happened, or it didn't), when the actual drug exposure behind that result may have been uneven across the tissue.
Here is a timing objection: ketamine's behavioral effects in patients show up within about two hours, which seems fast for anything described as protein synthesis. The answer is that dendritic protein synthesis doesn't always require transcription in the cell body and transport to the synapse. Neurons pre-position mRNAs locally at dendrites, so translation can happen within minutes at the synapse itself. That makes a two-hour window plausible.
The problem is that the original finding hasn't held up as a universal rule, challenged by the following studies.
Autry et al., publishing in Nature in 2011, proposed a different route entirely. Where Li et al.'s model centered on ketamine actively evoking new mTOR-dependent protein synthesis, Autry's group focused on a resting-state brake: spontaneous, low-level ("tonic") NMDA receptor signaling that occurs continuously, even without any deliberately evoked activity, normally keeps eEF2 kinase active and BDNF translation suppressed. Ketamine's blockade releases that brake, allowing a rapid rise in BDNF. This is often described as an mTOR-independent pathway, but that framing needs a caveat: BDNF itself signals through the TrkB receptor, which activates PI3K/Akt and, downstream of that, mTORC1. So the two pathways are not cleanly separable. What Autry's data show is that the initial trigger, the release of the brake on BDNF, doesn't require mTOR to happen. It doesn't mean the resulting protein synthesis avoids mTOR altogether once BDNF is out and signaling through TrkB. This didn't disprove Li et al.'s rapamycin result so much as complicate the picture: it suggested mTOR activation may be one of several convergent points in a shared cascade, not the sole upstream gatekeeper the original framing implied.
The enantiomer data add another layer. A 2018 study by Yang and colleagues comparing (R)-ketamine and (S)-ketamine found that intracerebroventricular rapamycin blocked the antidepressant effect of (S)-ketamine but not (R)-ketamine, which instead appeared to work through an ERK-dependent pathway that is largely mTOR-independent. It's tempting to connect this to the tonic-brake mechanism above, as if (R)-ketamine simply engages that resting-state NMDA disinhibition more strongly than (S)-ketamine does. But the data actually cut against that tidy story: (R)-ketamine is the weaker of the two at blocking NMDA receptors, yet it's the more potent and longer-lasting antidepressant in these models. If potency at the NMDA receptor doesn't track with potency at producing the effect, then the enantiomer difference isn't simply a matter of one form pressing harder on the same brake. We don't know why the two mirror-image forms of the same molecule lean on different kinase cascades. What the data do establish is that if mTOR dependence varies by stereoisomer of the same molecule, "protein synthesis inhibition blocks ketamine's effect" is not a property of ketamine so much as a property of one specific mechanistic route to a similar behavioral outcome.
The clearest complication, though, comes from the only published human test of the hypothesis. Abdallah and colleagues ran a double-blind, placebo-controlled, cross-over trial in which depressed patients received oral rapamycin or placebo before an IV ketamine infusion. If the rodent mTOR-blockade story translated directly, rapamycin should have blunted ketamine's antidepressant effect. It did not. At 24 hours, rapamycin pretreatment failed to block ketamine's response. At two weeks, the rapamycin group actually showed higher response rates (41 percent versus 13 percent on placebo) and higher remission rates (29 percent versus 7 percent). Systemic rapamycin in humans did not just fail to block the effect, it appeared to prolong it, the opposite direction from what the preclinical model predicted.
That result comes with an important caveat, though. Rapamycin is generally considered to have limited blood-brain-barrier penetrance, though it has been detected in human brain tissue at standard oral doses, and it's cleared slowly, hepatically metabolized with a half-life of roughly two to three days. A single oral pretreatment dose may simply not achieve the same central exposure as the direct cortical infusion used in rodents, which means this result doesn't cleanly settle whether the human brain's mTOR pathway was ever meaningfully blocked in the first place.
Taken together, mTOR looks like a real node in the synaptogenesis cascade behind rapid-acting antidepressants, but not the only one, and not a strict gatekeeper. BDNF/TrkB and ERK can drive similar plasticity through separate routes, even if they sometimes loop back into mTOR downstream. What mTOR-dependence you see also shifts with species, dosing route, and even which enantiomer you're using. The human rapamycin trial is the clearest reason for caution: rapamycin didn't block ketamine's effect, and the rapamycin group actually did better at two weeks. That result doesn't fit the simple "block mTOR, block the effect" model at all, though poor brain penetrance of oral rapamycin makes it hard to know what was actually blocked.
References
- Li N, Lee B, Liu RJ, Banasr M, Dwyer JM, Iwata M, Li XY, Aghajanian G, Duman RS. mTOR-dependent synapse formation underlies the rapid antidepressant effects of NMDA antagonists. Science. 2010 Aug 20;329(5994):959-964. doi: 10.1126/science.1190287.
- Autry AE, Adachi M, Nosyreva E, Na ES, Los MF, Cheng PF, Kavalali ET, Monteggia LM. NMDA receptor blockade at rest triggers rapid behavioural antidepressant responses. Nature. 2011 Jun 15;475(7354):91-95. doi: 10.1038/nature10130.
- Yang C, Ren Q, Qu Y, Zhang JC, Ma M, Dong C, Hashimoto K. Mechanistic target of rapamycin-independent antidepressant effects of (R)-ketamine in a social defeat stress model. Biological Psychiatry. 2018 Jan 1;83(1):18-28. doi: 10.1016/j.biopsych.2017.05.016.
- Abdallah CG, Averill LA, Gueorguieva R, Goktas S, Purohit P, Ranganathan M, Sherif M, Ahn KH, D'Souza DC, Formica R, Southwick SM, Duman RS, Sanacora G, Krystal JH. Modulation of the antidepressant effects of ketamine by the mTORC1 inhibitor rapamycin. Neuropsychopharmacology. 2020. doi: 10.1038/s41386-020-0644-9.
