Psilocybin, autism and depression: the psychedelic experience is apparently not needed

 

I like today’s post because it draws together some reader experience, autism research from the 1960s, with modern day lab research and human trials.

A distinct subgroup of our readers fall into what is today level 1 autism, which many think was better described as Asperger’s. Even though cognition and language are not impaired, some can have severe struggles, including anxiety, depression and feeling they do not quite fit in. Remarkably, in some cases just getting an official diagnosis provides a boost, “it's not me, it's my autism”. If that is sufficient therapy, then great.

For the Aspies seeking a better life, some end up looking at their serotonin receptors.

I recently wrote a review of all of them, but the Aspie focus tends to be 5-HT2A.

Rethinking the role of serotonin receptor signaling in improving autism symptoms: Prucalopride for a sub-group?

Incidentally, the mother who prompted that post tells me that her adult son continues to show the behavioral benefit from prucalopride one month later.

5-HT2A was the target of Ivar Lovaas back in 1966 at UCLA in Los Angeles. Lovaas is best known today for his work in Applied Behavioral Analysis (ABA).

Back in 1966 he published this paper:

Modificationof autistic behavior with LSD-25

The study involved a pair of identical autistic twin boys, then around five years old.

The researchers were interested in whether LSD could modify behaviours that were particularly difficult to change in severely affected children.

 

Our reader comments

One of the most notable comments in this blog was written by an Aspie who found a single experience with stimulating 5-HT2A provided a lasting beneficial shift in his mood.

You might wonder how such an effect could occur, like flipping a switch.

The recent research now explains why this likley happened.

 

The 1966 LSD study was tiny and exploratory by modern standards. It cannot establish that LSD was an effective treatment for autism, and the behavioural changes reported were principally observed during the drug sessions. It is therefore important not to read modern ideas about long-lasting psychedelic effects back into the 1966 experiment.

Nevertheless, the study is historically remarkable.

Lovaas was involved in research asking whether altering brain function could alter autistic behaviour.

Then the science took a different path.

 

60 years later we are back to psilocybin – depression in humans, autism in mice

A new study has taken the question into a modern genetic model of autism.

The researchers used mice lacking Cntnap2, a gene associated with neurodevelopmental disorders in humans. These mice show reduced sociability together with hyperactivity and repetitive behaviour.

A single dose of psilocybin produced a persistent increase in social behaviour.

The effect was detectable one day later and remained for at least two weeks. Importantly, the mice did not simply become more active: their overall exploration and locomotion did not show the same change.

And psilocybin did not correct everything.

The hyperactivity and repetitive grooming remained elevated.

This is therefore not evidence that psilocybin "reverses autism."

It is evidence that a brief pharmacological intervention can produce a persistent change in one particular behavioural phenotype in a particular genetic mouse model.

There was another important finding.

The same treatment did not increase sociability in genetically normal control mice.

That suggests something much more interesting than a simple "socialising" effect.

Perhaps the drug is interacting with an abnormal neural state.

 

The 5-HT2A receptor

Psilocybin's psychedelic effects are primarily mediated through the serotonin 5-HT2A receptor.

The researchers blocked this receptor before administering psilocybin.

The persistent improvement in sociability disappeared.

This gives us an important mechanistic sequence:

 

Psilocybin

↓

5-HT2A activation

↓

persistent biological change

↓

altered social behaviour

 

But then came the most surprising experiment.

The mice did not have to experience the psychedelic state

The researchers administered psilocybin while the mice were under light anaesthesia.

The animals therefore experienced the acute pharmacological action of psilocybin while unconscious.

When tested later while awake, they showed the same persistent increase in sociability.

In other words, an awake psychedelic experience was not required for the lasting behavioural effect in these mice.

The psychedelic experience is not itself the fundamental therapeutic mechanism.

It is one consequence of activating 5-HT2A receptors, while another consequence is a longer-lasting biological change in the brain.

 

What could produce a lasting effect?

One possibility is that the brief receptor signal initiates a cascade of molecular and cellular changes.

There is increasing evidence from psychedelic research that psilocybin can influence gene expression, neuronal structure, synaptic organisation and brain network function.

Epigenetic mechanisms may also be involved.

In the Cntnap2 work, the researchers found evidence that DNA-methylation machinery is required for the persistent behavioural effect. 

Researchers found that blocking DNMT1, a key DNA-methylation enzyme, prevented this persistent effect.

Importantly, DNMT1 inhibition did not prevent the acute 5-HT2A-related response.

This suggests DNA-methylation machinery acts downstream of 5-HT2A activation to maintain the longer-term change.

The finding points to a molecular mechanism whereby a brief drug exposure can produce behavioural effects long after the psychedelic experience has ended.

This suggests that a short-lived pharmacological signal may be converted into a longer-lasting change in gene regulation.

We do not yet know that psilocybin permanently rewrites the epigenome in these mice.

Nor do we know that the behavioural change is caused by a permanent epigenetic modification.

A more cautious model would be:

 

brief 5-HT2A stimulation

↓

intracellular signalling

↓

changes in gene regulation and possibly epigenetic state

↓

altered capacity for neural plasticity

↓

changes in neural circuits

↓

persistent change in behaviour

 

The important word is persistent, not permanent.

 

Could this matter for depression in autistic people?

This brings us back to humans.

Depression and persistent low mood occur in some autistic people, and this may be a particularly interesting population for psychedelic research.

Psilocybin has already produced persistent antidepressant effects in studies of people with major depressive disorder.

But there is an important unanswered question:

 

Would an autistic person with depression respond in the same way?

We do not yet know.

Autism is heterogeneous, and depression in an autistic person need not have exactly the same biological basis as depression in a non-autistic person.

Nevertheless, the new mouse findings make the question more interesting.

The drug did not simply increase sociability in every animal.

And now the experiment has returned to humans

The circle is becoming particularly interesting because researchers are now asking the same mechanistic question in humans.

A Stanford Phase 2 study, SPACE, is investigating psilocybin administered under general anaesthesia in people with major depressive disorder. The rationale is to suppress the noticeable psychological effects of psilocybin, making it possible to investigate whether effects can occur without the conscious psychedelic experience.

This is not an autism study, and it is small and experimental.

But scientifically it is fascinating.

Can separate the pharmacological action of psilocybin from the subjective psychedelic experience?

If lasting antidepressant effects were observed despite anaesthesia, that would provide important evidence that the biological action of 5-HT2A stimulation can produce therapeutic effects independently of the psychedelic experience.

 

From LSD in autistic humans to psilocybin in autistic mice and depressed humans

This brings us back to the historical arc.

1966 — UCLA

Lovaas and colleagues studied LSD in autistic children.

The question was whether changing brain function could modify autistic behaviour.

The following decades

Lovaas's research increasingly concentrated on behavioural intervention.

The question became whether changing the environment and learning experience could change behaviour and development.

2026 — autism mouse model

Researchers can now manipulate the biology much more precisely.

They can alter a specific autism-associated gene, activate 5-HT2A receptors, block the receptor, examine molecular pathways and administer the drug while the animal is unconscious.

And they find that a brief exposure to psilocybin can produce a behavioural effect that persists after the drug and the psychedelic state have disappeared.

2026 — depressed humans

Researchers are now taking the question back into humans.

Can psilocybin produce a lasting antidepressant effect without the person experiencing the psychedelic state?

The historical arc is therefore almost a circle.

 

Why did they choose the CNTNAP2 model and how this relates to Pitt Hopkins

The researchers did not choose the Cntnap2 mouse because it represents autism as a whole. They chose it because it is an established genetic model with a measurable reduction in sociability. Cntnap2-knockout mice show reduced social interaction, together with hyperactivity and repetitive behaviour. This gave the researchers a way to ask a very specific question: can psilocybin change an abnormal social phenotype?

The result was quite specific. Psilocybin increased the mice's preference for the social stimulus without significantly increasing their general exploration or movement. It also did not correct the hyperactivity or repetitive grooming.

The researchers also tested genetically normal mice. They did not show the same beneficial increase in sociability. This is important because it suggests that psilocybin may interact differently with a brain that already has an altered neurodevelopmental state, rather than simply acting as a general social enhancer.

 

Why is this relevant to Pitt Hopkins?

CNTNAP2 is particularly interesting in relation to Pitt Hopkins because biallelic CNTNAP2 mutations cause a Pitt-Hopkins-like neurodevelopmental disorder. Classical Pitt-Hopkins syndrome, however, is caused by alterations in TCF4.

There is also a biological relationship between the two genes. TCF4 is a transcription factor that can regulate CNTNAP2 expression. Thus, CNTNAP2 sits within a neuronal gene-regulatory network that is relevant to Pitt-Hopkins biology.

This does not mean that a Cntnap2-knockout mouse is a model of classical Pitt-Hopkins syndrome. The primary genetic abnormalities are different, and the findings cannot be directly extrapolated from Cntnap2 mice to people with Pitt Hopkins.

There is another important difference. Reduced social motivation is a clear phenotype of the Cntnap2 mouse, but it should not automatically be assumed to be the principal social problem in Pitt Hopkins syndrome.

People with Pitt Hopkins can have profound communication and social-communication difficulties, but many also actively seek attention, enjoy being with other people and show considerable social interest. In other words, the ability to communicate socially and the motivation to engage socially are not necessarily the same thing.

This means that if psilocybin were investigated in Pitt Hopkins, increasing social motivation would not necessarily be the most appropriate primary outcome.

Instead, researchers could ask whether it changes social communication, spontaneous communication, mood/affect, anxiety, repetitive behaviours, adaptive functioning or other measurable Pitt Hopkins phenotypes.

 

The research question

The Cntnap2 finding nevertheless raises an intriguing question for Pitt Hopkins:

If 5-HT2A activation can produce a persistent improvement in a behavioural phenotype caused by disruption of CNTNAP2, could it also modify some downstream consequences of TCF4 dysfunction?

The next logical experiment would therefore be to test psilocybin in a Tcf4 haploinsufficient Pitt Hopkins mouse model.

Researchers could examine not only behaviour but also the underlying biology:

• Does 5-HT2A activation alter the synaptic abnormalities associated with TCF4 deficiency?
• Does it affect gene expression or epigenetic regulation?
• Does it alter neuronal plasticity?
• Are any behavioural effects persistent after the drug has disappeared?
• Does the effect require the conscious psychedelic state, or could it occur under anaesthesia as in the Cntnap2 experiment?

The most interesting possibility would be a downstream compensation: the TCF4 mutation would remain, but a transient pharmacological intervention might alter some of the neuronal consequences of that mutation and allow the system to function differently.

 

Pitt-Hopkins is already being approached through epigenetics

There is an especially interesting parallel here because Pitt-Hopkins researchers are already testing an epigenetic strategy in humans. The RVL-001 trial, being conducted by Unravel Biosciences with the Pitt Hopkins Research Foundation in Colombia, is testing Vorinostat, a histone deacetylase (HDAC) inhibitor, in people with genetically confirmed Pitt Hopkins syndrome. The current exploratory study is small and is designed to examine safety and efficacy, as well as changes in the transcriptomic profile.

Vorinostat is approaching the problem from a very different direction than psilocybin. Rather than activating 5-HT2A receptors and potentially initiating downstream changes in neuronal signalling and plasticity, vorinostat directly alters epigenetic regulation by inhibiting HDAC enzymes, thereby changing how tightly DNA is packaged around histones and influencing gene expression.

There is a particularly strong rationale for this in Pitt Hopkins. In Tcf4-haploinsufficient mice, HDAC inhibition with vorinostat normalized abnormalities in hippocampal long-term potentiation and memory recall. Molecular studies also found changes in gene expression and DNA methylation associated with the treatment.

This creates an intriguing contrast:

Pitt-Hopkins genetic defect → TCF4 haploinsufficiency → abnormal gene regulation

Vorinostat: attacks the problem directly at the epigenetic/transcriptional level

versus

Psilocybin: 5-HT2A activation → intracellular signalling → potentially altered gene regulation/plasticity → possible downstream compensation.

 

In other words, these are two very different ways of trying to change the functional consequences of the same developmental genetic disorder.

The important point is that neither approach corrects the underlying TCF4 mutation. Both are attempting to modify downstream consequences of TCF4 deficiency.

 

Gene therapy

As was discussed in the recent post on gene therapy, there is now a fundamentally different approach being investigated in Pitt Hopkins: gene therapy.

Pitt Hopkins is caused by loss of function of one copy of TCF4, so gene therapy attempts to address the problem at its source. The investigational therapy MZ-1866 uses an AAV9 vector to deliver a functional copy of TCF4 to the brain.
It is being tested in a Phase 1/2 first-in-human clinical trial. The first participant was dosed in February 2026.
The current study is investigating safety and tolerability as well as potential clinical effects.
Unlike vorinostat or psilocybin, this approach is attempting to restore the missing genetic function itself.

Conclusion

Some Aspie readers of this blog have already established that a single dose of Psilocybin can produce long lasting improvements in social behavior. They skipped the mouse models and adopted the Nike approach.

The LSD trials in children in the 1960s assumed that the psychedelic experience was the therapy, now we see that this is not the case. You can sleep through it and still get the long lasting effect.

The beneficial effect in CNTNAP2 mice does indicate that the experiment should be repeated in the closely related Pitt Hopkins model. Of course, it should also be checked in idiopathic autism models like the BTBR and the maternal immune activation (MIA) model

In the prenatal valproate (VPA) model, psilocybin was previously tested and it did rescue the social-behavioural abnormalities. 

Note

Psilocybin is the main psychoactive compound found in “magic mushrooms.” It is converted in the body into psilocin, which produces the psychedelic effects. Research using purified psilocybin therefore studies the key active psychedelic component rather than the whole mushroom.

How the UCLA Rapamycin Study Changes Our Understanding of Autism: From mTOR to Precision Medicine

 

  

Rapamycin is not a new subject in this blog, but it did reappear in the recent Yale paper about what drugs “should” help in some autism. One of our Chinese readers did trial it years ago, based on a published case history from China. Some readers of Epiphany ASD have children with TSC and receive everolimus, a close relative of rapamycin. Everolimus is prescribed by specialist physicians for approved indications associated with TSC. Other parents have trialed rapamycin more recently.

For more than a decade, researchers have reported that the drug rapamycin can improve autism-like behaviours in animal models. At first glance, the latest study from UCLA appears to be just another example. Adult mice exposed to inflammation before birth received a single dose of rapamycin, and within two hours many of their autism-like features—including sensory hypersensitivity, repetitive behaviours, neuronal hyperexcitability and abnormal brain network activity—were dramatically improved.

https://www.uclahealth.org/news/release/drug-reverses-autism-like-brain-changes-adult-mice-within

It would be easy to conclude that this is simply another story about a promising drug.

I think that would miss the real significance of the work.

The UCLA study is not really about rapamycin.

It is about what rapamycin has taught us about the autistic brain.

For decades autism has been viewed primarily as a disorder of abnormal brain development. If neural circuits develop incorrectly before birth, the assumption has been that they cannot later be corrected. Treatment therefore focused on helping individuals adapt to those differences rather than attempting to restore the function of the underlying circuits.

The UCLA study challenges that assumption.

The remarkable finding was not simply that rapamycin improved behaviour.

It was how quickly the improvements occurred.

Within approximately two hours, neurons that had been firing excessively returned towards normal activity. Functional communication between brain regions improved. Sensory over-responsivity diminished. Repetitive behaviours became less pronounced.

Two hours is far too short for the brain to rebuild abnormal neural connections that developed before birth.

Instead, the study suggests something much more exciting:

Even when abnormal brain anatomy remains, the function of those circuits may still be remarkably modifiable in adulthood.

That is a profound change in thinking.

The implications extend far beyond rapamycin itself.

They suggest that future autism therapies may not need to "repair" the autistic brain. Instead, they may simply need to restore healthier patterns of neuronal activity.

If this principle also proves true in humans, it could fundamentally change how autism is treated.

 

Five take-home messages

Before exploring the details, here are the five ideas that I believe make the UCLA study important.

1. Adult autistic brain circuits remain surprisingly plastic

The study demonstrates that functional improvements can occur rapidly, even when structural developmental abnormalities remain unchanged.

2. Different causes of autism may converge on common biological pathways

Genetic disorders such as tuberous sclerosis, environmental models such as prenatal valproic acid exposure, and maternal immune activation all appear capable of producing excessive neuronal activity through related signalling pathways involving mTOR.

3. Rapamycin is probably not the future treatment

Its greatest contribution may be demonstrating that adult brain circuits can still be rebalanced.

4. Precision medicine is replacing one-size-fits-all autism treatment

Rather than treating "autism," future therapies are likely to target the specific biological pathway that is abnormal in each individual.

5. Autism clinical trials may need redesigning

If some therapies produce meaningful physiological improvements within hours or days, measuring outcomes only after three months may miss important information about how they actually work.

 

 

 

 

A timeline of how our understanding has evolved

The UCLA study did not appear in isolation.

It represents the latest step in a scientific journey that has gradually transformed our understanding of autism biology over the past fifteen years.

Table 1. How our understanding of rapamycin and autism has evolved

Year	Discovery	Major Insight
2012	Rapamycin improves autistic-like behaviour in adult mouse models of tuberous sclerosis (TSC).	Excessive mTOR signalling can directly contribute to autistic behaviours, establishing mTOR as a potential therapeutic target.
2019	Rapamycin improves social interaction in mice exposed prenatally to valproic acid (VPA).	mTOR dysregulation is not confined to rare genetic disorders and may contribute to some forms of non-syndromic autism.
2022	Italian researchers demonstrate that rapamycin restores neuronal excitability without correcting the original potassium channel defect.	Behaviour can improve through rapid functional normalization of neuronal circuits rather than structural repair of the developing brain.
2026	UCLA researchers show that a single dose of rapamycin rapidly normalizes neuronal activity, sensory processing and functional brain networks within approximately two hours.	Adult autistic brain circuits remain functionally modifiable even after abnormal development, shifting the focus from structural repair to physiological rebalancing.
Future	Precision medicine identifies the disrupted biological pathway in each individual rather than treating autism as a single disorder.	The goal becomes restoring healthy physiology rather than simply suppressing symptoms, using individualized combinations of targeted therapies.

Viewed together, these studies reveal a gradual evolution in scientific thinking.

Originally, researchers hoped rapamycin might become a treatment for tuberous sclerosis.

Today, it appears to have served an even more important purpose.

It has become the experimental tool that demonstrated a fundamental principle:

Adult autistic brain circuits remain remarkably plastic.

 

How the story began

The rapamycin story started with tuberous sclerosis complex (TSC).

TSC is caused by mutations in either the TSC1 or TSC2 genes. These genes normally act as brakes on the mTOR signalling pathway. When they are defective, mTOR becomes chronically overactive, resulting in abnormal cell growth, epilepsy, benign tumours and a high incidence of autism.

Rapamycin was therefore an obvious therapeutic candidate because it directly inhibits mTOR.

Remarkably, adult mouse models of TSC showed substantial improvements in social behaviour following treatment.

This immediately raised an intriguing question.

Was rapamycin correcting something unique to tuberous sclerosis?

Or had researchers stumbled upon a biological pathway that might also contribute to other forms of autism?

That question led to the next important discovery.

In 2019, Japanese researchers investigated the prenatal valproic acid (VPA) model, one of the best-established environmental models of autism.

Valproic acid exposure during pregnancy has long been known to increase the risk of autism in humans. In mice, prenatal exposure produces many autism-like features including altered social interaction, repetitive behaviours and abnormal neuronal development.

The researchers found that rapamycin again improved behaviour while simultaneously reducing excessive activation of the mTOR pathway.

This suggested that mTOR dysregulation might not be restricted to rare genetic syndromes.

Instead, very different causes of autism might converge on the same downstream biological pathway.

The next breakthrough came in 2022.

An Italian research group studied neurons within the nucleus accumbens, a brain region involved in motivation, reward and social behaviour.

They discovered that rapamycin rapidly restored normal firing patterns in hyperexcitable neurons and improved social behaviour.

Perhaps the most surprising observation was what did not happen.

The underlying potassium channel abnormality remained.

Neuronal structure remained largely unchanged.

Rapamycin had restored neuronal function without correcting the original developmental abnormality.

That finding hinted that brain circuits might remain physiologically flexible long after development had finished.

The UCLA study has now taken that idea much further.

 

The major advance is not that rapamycin works

The UCLA researchers deliberately chose a completely different autism model.

Instead of using a genetic disorder or prenatal exposure to valproic acid, they exposed pregnant mice to mild maternal inflammation, mimicking one of the environmental risk factors associated with autism.

The offspring later developed many familiar autism-like features:

• sensory over-responsivity
• repetitive behaviours
• neuronal hyperexcitability
• abnormal communication between brain regions
• increased susceptibility to seizures
• excessive activation of the mTOR pathway

Once again, rapamycin improved nearly every abnormality.

But this time the researchers looked far more carefully at how quickly those changes occurred.

The answer was astonishing.

Within approximately two hours, neuronal firing patterns, sensory responses and functional brain connectivity all shifted towards normal.

That is simply too fast for structural rewiring of the brain.

Instead, the findings suggest that rapamycin acted primarily by rebalancing the physiology of existing neuronal circuits.

That distinction may prove to be one of the most important conceptual advances in autism research for many years.

It shifts the question from:

Can we repair the autistic brain?

to something much more achievable:

Can we restore healthier function to autistic brain circuits?

That change in perspective forms the foundation of modern precision medicine—and it is the subject of the next section.

 

 

Convergence, Precision Medicine and Why Not Every Autistic Brain Has the Same Biology

One of the biggest obstacles to developing effective autism treatments has been the enormous biological diversity hidden behind a single clinical diagnosis.

Two children may both receive a diagnosis of autism while sharing remarkably few biological abnormalities.

One may have a mutation affecting calcium channels.

Another may have excessive activation of the PI3K-Akt-mTOR pathway.

A third may have altered GABA signalling.

A fourth may have neuroimmune dysfunction.

Yet all four satisfy the diagnostic criteria for autism.

For many years this heterogeneity has frustrated drug development. Clinical trials often enrol hundreds of autistic individuals whose underlying biology is completely different. Even if a treatment works extremely well in one biological subgroup, the benefit may disappear statistically when averaged across everyone else.

The UCLA study strengthens an alternative way of thinking.

Instead of treating autism as one disorder, perhaps we should think of it as many different disorders that converge on a relatively small number of disturbed brain circuits.

That concept lies at the heart of precision medicine.

 

Different roads lead to the same destination

One of the most remarkable observations from the rapamycin literature is that completely different causes of autism produce surprisingly similar abnormalities.

Consider the three major animal models discussed so far.

 

Tuberous sclerosis

Here the starting point is genetic.

Mutations in TSC1 or TSC2 remove the normal brakes on mTOR signalling, producing chronic overactivation of the pathway.

 

Prenatal valproic acid exposure

Here the initiating event is environmental.

Exposure to valproic acid during pregnancy alters gene expression during brain development, ultimately producing abnormal neuronal excitability and excessive mTOR signalling.

 

Maternal immune activation

The UCLA model begins differently again.

Mild inflammation during pregnancy alters fetal brain development, eventually producing neuronal hyperexcitability, sensory dysfunction and abnormal functional brain networks.

At first sight these appear to be completely unrelated disorders.

Yet they converge on remarkably similar downstream abnormalities:

• excessive mTOR signalling
• neuronal hyperexcitability
• disrupted excitation/inhibition balance
• abnormal functional connectivity
• sensory over-responsivity
• repetitive behaviours

Perhaps even more strikingly, all three respond to rapamycin.

This pattern suggests that mTOR may function as a common biological hub where several apparently unrelated pathways intersect.

That does not mean mTOR is the cause of autism.

Rather, it may represent one of the final common pathways through which many different causes disturb brain function.

 

 

The figure illustrates an important principle.

Most autism therapies do not directly target mTOR.

Instead, they influence different parts of a biological network that ultimately converges on the same endpoint:

normal neuronal circuit function.

This systems approach represents a major shift away from searching for a single "autism gene" or a single "autism drug."

 

Pitt-Hopkins syndrome reminds us that biology can move in both directions

Whenever a new biological pathway becomes fashionable there is a temptation to assume that every patient has too much or too little of it.

The mTOR pathway is no exception.

After reading about tuberous sclerosis and the UCLA study, it would be easy to conclude that autism is caused by excessive mTOR activity.

That would be a mistake.

One of the clearest examples comes from Pitt-Hopkins syndrome.

Pitt-Hopkins syndrome results from mutations in the TCF4 gene, a transcription factor that regulates hundreds of genes involved in brain development.

Experimental studies suggest that reduced TCF4 activity may decrease signalling through the PI3K-Akt-mTOR pathway, impairing neuronal growth and protein synthesis.

Although this area is still evolving, the important principle is clear.

If mTOR signalling is already reduced, further suppressing it with rapamycin could theoretically worsen the underlying biological abnormality.

The treatment strategy becomes almost the mirror image of tuberous sclerosis.

Instead of inhibiting mTOR, researchers have explored ways of restoring trophic signalling.

These include:

• IGF-1, which activates the PI3K-Akt-mTOR pathway.
• Trofinetide, a synthetic analogue of the IGF-1-derived peptide glypromate (GPE), developed to support synaptic function.
• BDNF-enhancing approaches, which indirectly support PI3K-Akt-mTOR signalling and neuronal plasticity.
• Gene therapy, aiming to restore normal TCF4 expression and thereby correct multiple downstream pathways simultaneously.

The objective is not to maximise mTOR activity.

Neither is it to suppress it.

The objective is to restore normal physiological signalling.

 

The Goldilocks principle

 

 

The Goldilocks principle comes from the children's story Goldilocks and the Three Bears, where Goldilocks rejects porridge that is too hot or too cold, choosing the one that is just right.

In biology, the term is used to describe systems that work best within an optimal range. Too much activity is harmful, but so is too little. The objective is to restore the system to its normal physiological set point.

Many signalling pathways operate within an optimal range.

Too much activity causes disease.

Too little activity causes disease.

mTOR appears to behave in exactly this way.

Excessive activity may produce:

• abnormal protein synthesis
• neuronal hyperexcitability
• impaired autophagy
• abnormal sensory processing

Insufficient activity may impair:

• dendritic growth
• synapse formation
• neuronal plasticity
• learning and memory

Both extremes can disturb the development and function of neuronal circuits.

The goal is therefore not "high mTOR" or "low mTOR."

It is normal mTOR.

 

Which autistic individuals might benefit from mTOR modulation?

This is the practical question every family eventually asks.

Current evidence suggests the answer is:

probably only some of them.

The strongest rationale currently exists for conditions already known to involve excessive activation of the PI3K-Akt-mTOR pathway.

Examples include:

• tuberous sclerosis complex
• PTEN-related syndromes
• PIK3CA-related disorders
• certain forms of macrocephalic autism
• some individuals with prenatal valproic acid exposure
• some individuals affected by maternal immune activation

Other features may also raise interest, although much more research is needed.

These include:

• epilepsy
• epileptiform EEG abnormalities
• marked sensory hypersensitivity
• evidence of neuronal hyperexcitability

Importantly, none of these features proves that mTOR is abnormal.

They simply identify groups in which further research may be particularly informative.

Conversely, disorders such as Pitt-Hopkins syndrome remind us that not every autistic individual is likely to benefit from mTOR inhibition.

Some may ultimately require therapies that gently enhance trophic signalling instead.

 

Precision medicine replaces one-size-fits-all treatment

This way of thinking represents one of the biggest changes in autism research.

Historically, the question was:

Does this drug treat autism?

Precision medicine asks a different question.

Does this drug correct the biological abnormality present in this particular individual?

That may sound like a subtle distinction.

In reality, it changes everything.

A future clinic might begin not by prescribing medication, but by asking:

• Which genes are affected?
• Which signalling pathways are abnormal?
• Is neuronal excitability increased or decreased?
• Is mTOR signalling excessive, insufficient or normal?
• Is neuroinflammation contributing?

Only then would treatment be selected.

This is already common practice in oncology, where cancers are increasingly classified by their molecular biology rather than by their location in the body.

Autism may eventually follow a similar path.

 

A systems biology approach

One lesson from the UCLA study is that no single pathway operates in isolation.

The brain functions as an interconnected network.

Changes in one signalling pathway inevitably influence many others.

That is why Figure 2 is so important.

It illustrates that treatments such as:

• statins,
• metformin,
• pioglitazone,
• IGF-1,
• trofinetide,
• calcium channel blockers,
• bumetanide,
• NAC,

all act at different points within the network.

Yet many ultimately converge on the same physiological endpoint:

restoring healthier neuronal circuit function.

This systems biology perspective moves us away from asking:

"Which drug is best?"

towards a far more useful question:

"Which combination of interventions is most likely to restore normal physiology in this individual?"

That question naturally leads to the next issue.

Even if rapamycin can rapidly rebalance neuronal circuits, is it actually the right drug for long-term treatment?

As we shall see in the next section, the UCLA study itself provides a compelling reason to think the answer is probably not.

 

Why Rapamycin Is Probably Not the Final Answer – Combination Therapy, Clinical Trials and the Rise of N-of-1 Precision Medicine

At this point, readers might reasonably ask a simple question.

If rapamycin can normalize brain function so dramatically in mice, why not simply use rapamycin to treat autism?

The UCLA researchers themselves provide the answer.

Ironically, one of the most important findings in the study was not the initial improvement.

It was what happened afterwards.

When rapamycin was given repeatedly, its effectiveness gradually diminished.

The brain adapted.

That observation may ultimately prove just as important as the initial two-hour response.

 

The brain fights back

The nervous system is remarkably good at maintaining stability.

Neuroscientists refer to this as homeostasis.

Whenever one signalling pathway is pushed strongly in one direction, numerous compensatory mechanisms begin working to restore equilibrium.

This phenomenon is familiar throughout medicine.

Patients taking opioids gradually require higher doses to achieve the same effect.

Tolerance develops to benzodiazepines.

Many antidepressants require weeks before compensatory changes produce their full clinical benefit.

Rapamycin appears to trigger a similar biological response.

Initially, suppressing excessive mTOR activity restored healthier neuronal function.

With prolonged treatment, however, compensatory mechanisms gradually reduced its effectiveness.

The lesson is important.

It suggests that completely suppressing one signalling pathway is unlikely to represent the ideal long-term strategy.

Instead, future therapies may need to work with the brain's natural homeostatic mechanisms rather than continually fighting against them.

 

Rapamycin may have fulfilled its real purpose

This leads to what I believe is the central message of the UCLA study.

Rapamycin's greatest contribution may not be as a treatment. It may be the drug that proved adult autistic brain circuits can still be rebalanced.

Whether rapamycin itself ultimately becomes a routine autism therapy is almost secondary.

The important discovery is that rapidly restoring normal neuronal physiology appears possible.

The challenge now becomes finding safer and more sustainable ways of achieving exactly the same result.

 

Could combination therapy outperform rapamycin?

Modern medicine has repeatedly discovered that several modest interventions often outperform one powerful drug.

Hypertension provides an excellent example.

Doctors rarely prescribe the maximum dose of a single blood pressure medication.

Instead they commonly combine two or three lower-dose drugs that act through different mechanisms.

Blood pressure is controlled more effectively.

Side effects are reduced.

Compensatory physiological responses are minimised.

The same philosophy may eventually apply to autism.

Instead of strongly inhibiting one signalling pathway, future treatment may involve gently nudging several interconnected pathways towards normal physiological function.

No individual intervention would need to produce a dramatic effect.

Together, however, they might restore healthy neuronal network activity while avoiding the biological adaptation observed with chronic rapamycin.

 

 

One of the most useful ways of thinking about autism therapies is to group them according to where they act within the biological network rather than according to their traditional medical uses.

Upstream modulators

These therapies influence signalling before it reaches mTOR.

Examples include:

• Statins, which reduce Ras and Rho signalling through inhibition of protein prenylation.
• Metformin, which activates AMPK, indirectly reducing excessive mTOR activity.
• Pioglitazone, which reduces inflammatory signalling while influencing the PI3K-Akt pathway.
• IGF-1, which increases PI3K-Akt signalling when trophic support is inadequate.
• Trofinetide, which enhances IGF-1-related neurotrophic signalling.

 

Direct mTOR modulators

These drugs act directly on the pathway itself.

• Rapamycin
• Everolimus

 

Downstream modulators

Many familiar autism interventions act downstream by altering neuronal excitability rather than directly influencing mTOR.

Examples include:

• bumetanide
• calcium channel blockers
• memantine
• benzodiazepines
• N-acetylcysteine (NAC)
• neuromodulation

Although these therapies appear very different pharmacologically, many ultimately converge on a common physiological objective:

 

Restoring healthier neuronal circuit function.

That systems biology perspective is likely to become increasingly important over the coming decade.

 

Do autism clinical trials need redesigning?

One aspect of the UCLA study has received surprisingly little attention.

The improvements occurred within approximately two hours.

That observation immediately raises an important question.

Are autism clinical trials measuring outcomes at the right time?

Traditionally, participants return after several months before researchers assess whether treatment has been successful.

That makes perfect sense for therapies expected to produce gradual developmental changes.

It makes much less sense for therapies that appear to work by rapidly restoring neuronal physiology.

The UCLA study suggests that meaningful changes in brain function may occur long before structural changes become possible.

If so, waiting three months before making the first assessment may overlook valuable biological information.

 

Some autism interventions appear to act rapidly

Rapamycin is probably not unique.

Several interventions have been reported to produce relatively rapid improvements in at least some autistic individuals.

Examples include:

• mTOR modulation
• calcium channel blockers in selected channelopathies
• bumetanide
• N-acetylcysteine (NAC)
• memantine
• benzodiazepines
• corticosteroids in selected immune-mediated cases

These drugs have very different pharmacology.

Nevertheless, many ultimately influence:

• neuronal excitability
• excitation/inhibition balance
• functional brain network activity

If these mechanisms are correct, improvements may begin within hours or days rather than months.

 

Long-term studies remain essential

None of this means that three-month or six-month follow-up should disappear.

Quite the opposite.

The UCLA study demonstrated that biological adaptation can reduce the effectiveness of prolonged treatment.

Long-term follow-up therefore remains essential for determining:

• durability of benefit
• development of tolerance
• adverse effects
• functional outcomes.

The point is simply that early measurements and late measurements answer different scientific questions.

Early assessments tell us whether the drug is biologically active.

Later assessments tell us whether the benefit persists.

Future trials may therefore benefit from measuring both.

A typical design might include:

• baseline
• 24–48 hours
• one week
• one month
• three months
• six months

Such studies would distinguish:

• rapid responders,
• delayed responders,
• non-responders,
• and individuals who initially improve but later develop tolerance.

 

The Chinese sirolimus case: an example of an N-of-1 trial

One human case illustrates both the promise and the limitations of this approach.

Chinese physicians reported treating an 8-year-old boy with non-syndromic autism using low-dose sirolimus (rapamycin).

Rapamycin/Sirolimus Improves the Behavior of an 8-Year-Old Boy With Nonsyndromic Autism Spectrum Disorder

The child had previously received behavioural therapy together with aripiprazole but remained significantly impaired.

He was then prescribed sirolimus 1 mg daily, with blood concentrations maintained between 5 and 10 ng/mL.

Within approximately two weeks, improvements in social interaction, eye contact, hyperactivity and stereotypic behaviour were reported.

After roughly two months he was attending a mainstream school successfully.

Treatment reportedly continued for more than two years.

The principal adverse effect consisted of recurrent mild-to-moderate mouth ulcers (stomatitis), while routine blood tests remained largely normal during specialist monitoring.

This report is scientifically interesting.

It is not scientifically conclusive. Other parents have tried to repeat this effect.

Nevertheless, the report illustrates an important principle.

Clinicians did not need to wait three months before recognising that something biologically significant appeared to be happening.

The reported improvement was already evident within two weeks.

The subsequent months were important for determining whether the benefit persisted and whether side effects emerged—not for detecting the initial response.

That distinction is exactly the one highlighted by the UCLA study.

 

What can we learn from N-of-1 medicine?

One of the strengths of the autism community is its willingness to observe carefully.

Over the years, many readers of Epiphany ASD have shared their experiences after trialling therapies discussed on this blog under medical supervision.

Some families have reported remarkable improvements.

Others have seen no obvious benefit.

Some have stopped treatment because of unacceptable side effects.

These experiences cannot replace randomized controlled trials.

However, they do remind us of an important fact.

Autism is biologically heterogeneous.

A treatment that transforms one child may do absolutely nothing for another.

Rather than dismissing carefully documented individual experiences as "just anecdotes," they can serve as valuable hypotheses that identify biological subgroups deserving formal scientific investigation.

 

mTOR Beyond the Brain – Why It Builds Muscle and Bone, Where Precision Medicine Is Heading, and the Lasting Legacy of the UCLA Study

Up to this point we have focused almost entirely on the brain.

However, mTOR is not a "brain pathway."

It is one of the body's master regulators of growth, repair and adaptation.

Understanding this broader role helps explain both the promise and the limitations of rapamycin as a potential autism therapy.

 

mTOR: one pathway, many jobs

One reason mTOR has attracted so much scientific attention is that it sits at the centre of an enormous biological network.

It integrates information about:

• nutrients
• energy availability
• growth factors
• hormones
• inflammation
• mechanical loading

before deciding whether cells should:

• grow,
• divide,
• manufacture proteins,
• recycle damaged components through autophagy,
• or conserve energy.

This makes perfect evolutionary sense.

Growth is expensive.

The body should only build new tissue when sufficient energy and nutrients are available.

The same signalling pathway therefore helps regulate:

• brain development,
• immune function,
• muscle growth,
• bone formation,
• wound healing,
• ageing,
• metabolism.

That explains why manipulating mTOR can have such widespread effects throughout the body.

 

 

 

The figure illustrates an important concept.

Although the same signalling pathway is involved, its role differs dramatically depending on the tissue.

In the brain

Pathological activation may produce:

• neuronal hyperexcitability
• abnormal sensory processing
• disrupted functional brain networks
• repetitive behaviours
• increased seizure susceptibility

The goal is normalization.

In muscle

Physiological activation stimulates:

• muscle protein synthesis
• repair following exercise
• adaptation to resistance training
• increased strength

The goal is preservation.

In bone

Mechanical loading activates mTOR within osteoblasts, stimulating:

• bone formation
• collagen synthesis
• skeletal adaptation
• increased bone strength

Again, the goal is preservation, not inhibition.

 

Why exercise activates mTOR—and why that is a good thing

Readers of Epiphany ASD know that I frequently discuss the benefits of exercise.

At first glance, this might appear contradictory.

If excessive mTOR activity can contribute to autism, why encourage activities that activate mTOR?

The answer lies in one of the most important concepts in physiology.

It is essential to distinguish between:

physiological mTOR activation

and

pathological mTOR dysregulation.

When we perform resistance exercise, climb a steep hill carrying a backpack or lift weights, mTOR activation is:

• temporary,
• tightly regulated,
• largely confined to the muscles and bones being exercised.

This is exactly what we want.

It allows the body to adapt to increased mechanical loading by becoming stronger.

Similarly, consuming high-quality protein—particularly protein rich in the amino acid leucine—temporarily activates mTOR within skeletal muscle, stimulating muscle protein synthesis and recovery after exercise.

These are normal physiological responses.

The autism models discussed earlier are fundamentally different.

They involve:

• chronic dysregulation
• abnormal signalling
• specific neuronal populations
• altered brain circuit function

These are not equivalent biological states.

Healthy activation of mTOR after exercise should not be confused with pathological overactivation within particular brain circuits.

 

Leucine: a natural activator of mTOR

Leucine deserves particular mention because it illustrates how precisely the body regulates mTOR.

Leucine is one of the three branched-chain amino acids and is a powerful nutritional signal that activates mTORC1.

After a protein-rich meal—particularly one containing dairy products or whey protein—blood leucine concentrations rise.

Muscle cells interpret this as evidence that sufficient amino acids are available to build new proteins.

The result is increased muscle protein synthesis.

This explains why resistance exercise combined with adequate dietary protein is so effective for maintaining muscle mass during ageing.

It also illustrates an important point.

Simply activating mTOR is neither inherently good nor inherently bad.

Everything depends on:

• which tissue
• which cells
• how long
• and how much

The body uses exactly the same signalling pathway for healthy adaptation and for disease.

Context is everything.

 

Why chronic rapamycin has drawbacks

Unfortunately, rapamycin cannot distinguish between tissues.

It inhibits mTOR wherever it reaches sufficient concentrations.

That creates an obvious problem.

While reducing excessive mTOR activity in overactive neuronal circuits may be beneficial, suppressing normal mTOR activity elsewhere in the body may produce unwanted effects.

Long-term treatment can potentially reduce:

• muscle protein synthesis
• exercise adaptation
• bone formation
• wound healing
• immune function

These considerations are particularly important in children and adolescents, whose brains, muscles and skeletons are still developing.

This is another reason why rapamycin itself is unlikely to become a universal autism treatment.

The challenge is to normalize abnormal brain signalling without interfering with the many beneficial roles of mTOR elsewhere in the body.

 

The future lies in selective regulation

Ideally, future therapies will behave much more like thermostats than on-off switches.

Instead of suppressing mTOR throughout the entire body, they would:

• reduce excessive signalling where it is pathological,
• preserve normal signalling where it is physiological,
• restore deficient signalling where it is inadequate.

Achieving that level of precision remains a major scientific challenge.

However, many areas of medicine are already moving in exactly this direction.

Cancer therapy has evolved from broadly toxic chemotherapy towards molecularly targeted treatments.

Immunology has progressed from general immunosuppression to selective biological therapies.

Autism may eventually undergo a similar transformation.

 

Precision medicine will probably use combinations, not miracle drugs

One of the recurring themes throughout this article has been convergence.

Different biological abnormalities often disturb the same neuronal circuits.

Likewise, different therapies often move those circuits back towards normal through different mechanisms.

Future treatment may therefore consist of carefully selected combinations of relatively modest interventions.

For example:

• a statin reducing excessive Ras signalling;
• metformin moderating mTOR activity through AMPK;
• bumetanide restoring GABAergic inhibition;
• a calcium channel blocker correcting abnormal calcium influx;
• behavioural interventions strengthening newly balanced circuits.

Each intervention may contribute only part of the solution.

Together they may restore healthier neuronal physiology.

This approach resembles the management of hypertension, diabetes and heart failure, where combination therapy has become routine because complex biological systems rarely respond optimally to a single intervention.

 

Conclusions

Fifteen years ago researchers believed rapamycin might become a treatment for tuberous sclerosis.

Today, the UCLA study suggests something far more important.

Rapamycin has shown us that adult autistic brain circuits remain remarkably plastic.

Its greatest contribution may therefore not be as a medicine.

It may be as the experimental tool that demonstrated a new principle of autism biology.

The future is unlikely to belong to rapamycin itself.

Instead, it belongs to precision medicine.

Rather than treating "autism" as a single disorder, clinicians will increasingly identify the disrupted biological pathway in each individual and select therapies designed to restore that pathway to its normal physiological range.

Some individuals may require reduced mTOR signalling.

Others may require increased trophic support.

Still others may benefit from correcting calcium signalling, chloride transport, mitochondrial function or neuroinflammation.

The diagnosis may remain the same.

The biology will not.

Perhaps that is the greatest lesson of the UCLA study.

It reminds us that the question is no longer:

"Can we rebuild the autistic brain?"

Instead, we can begin asking:

"Can we restore healthier function to autistic brain circuits?"

For the first time, that question appears scientifically realistic.