Biome of Overstimulation: Sensory Sensitivity as a Unifying Mechanism

We outline connections between the gut microbiome, autoimmune conditions, neuropathic pain, eye pain, chemical intolerance, and a specific set of “overactive” mental illnesses. All seem to be connected to a sensory processing disorder. This is joint work with Luca Estinto.

The question we want to answer is whether a single underlying state explains why these conditions tend to show up in the same people, or whether there are several.

We propose there is collection of disorders characterized by sensory processing issues. This collection of symptoms affects many and is rarely viewed holistically. We propose to view them holistically and approach their treatment as such – keeping in mind that autoimmune disorders usually are best detected genetically, and medication-wise sensory processing is best treated using tricylic antidepressants, GABA analogues, and glasses prescribed by a behavioral optometrist.

⚠ A caveat

This is a set of hypotheses, and we have not tested them. Most of the mechanisms below were shown in mice, the human studies are correlational, and as far as we know no study has measured all of these conditions in the same group of people. The last section describes what a test would look like and what result would prove us wrong. None of this is medical advice.

Sensory processing disorders and connectivity

People who react strongly to perfumes and strong smells (chemical intolerance) are such that they perceive a smell at a constant level of intensity, the signal is processed at the same intensity over time. Normal people perceive the signal intensity to decrease over time, they “get used to it”. Signal intensity is monitored via brain blood flow (measured using EEG and fMRI). The same blood flow pattern as those with chemical intolerance is found in those with chronic pain.^1,2

Sensory processing disorders and white matter

Children with sensory processing disorders have disconnected white matter (measured using DTI)³.

⚠ Contested images

The three images below are from the Amen Clinics, which interprets the “ring of fire” blood flow pattern on SPECT scans as a sign of bipolar disorder, OCD and autism. This interpretation is contested. SPECT measures blood flow, so it can’t tell you anything about glutamate directly, it is not an accepted diagnostic test for any psychiatric condition, and the clinic’s findings have not been independently replicated. We include them because they are often cited, but the argument here doesn’t depend on them.

Amen Clinics: “healthy” scan next to one labelled bipolar disorder.

Amen Clinics: “healthy” scan next to one labelled OCD.

Amen Clinics: “healthy” scan next to one labelled autism (ASD).

GABA, the Brake on Glutamate

GABA is the main inhibitory neurotransmitter in the brain, so it does roughly the opposite job to glutamate. The body makes GABA out of glutamate, using an enzyme called glutamic acid decarboxylase (GAD). GAD is also one of the main targets of the immune system in type 1 diabetes, and antibodies against it (anti-GAD65) are one of the standard autoantibodies used to diagnose and predict the disease.⁴ We find it interesting that the enzyme which makes the brain’s inhibitory signal is attacked in the same autoimmune disease that is linked to high sensitivity and to celiac.

GABA can be measured directly in a living brain with magnetic resonance spectroscopy (MRS). One study compared 37 autistic children with 35 typically developing children and found less GABA in the autistic children’s sensorimotor cortex, which processes touch, and normal levels in the visual cortex. The children with lower GABA also did differently on touch tasks.⁵ A study of autistic adults found no difference, so this is not settled.⁶ Both studies are testing an older idea, that in autism there is too much excitation relative to inhibition in sensory and social systems.⁷

Some gut bacteria also make GABA. Of 91 strains isolated from the human intestine, one Lactobacillus and four Bifidobacterium strains converted glutamate to GABA, the best one converting up to 90% of it.⁸ Bacteroides produce a lot of GABA, and in a study of people with major depression, higher levels of Bacteroides in stool were associated with weaker depression-related signatures on brain scans.⁹ Several of these groups of bacteria are the ones that are reduced in autism or under stress.

In mice, feeding a strain of Lactobacillus changed GABA receptor levels in the brain and reduced anxiety- and depression-like behaviour. When the vagus nerve was cut, the effect disappeared.¹⁰ When the same strain was tried in healthy men it had no measurable effect on stress or cognition.¹¹

Remark: As far as we know GABA made in the gut does not reach the brain directly, and the mouse results haven’t carried over to people so far. So this shows the gut can make GABA and respond to it, but not that it controls the balance in the brain.

SPD causing Visual Tunneling and Autoimmune conditions affecting Eyesight

Visual tunneling is a particularly pervasive issue that affects all senses, it results from an overload of visual sensation in the brain. Tunneling is a subconscious spatial adaptation to reduce the information the individual handles. The individual is either overwhelmed with informatio or has trouble sorting out the significant area for attention.¹²

“Cognitively speaking, for these individuals to succeed, they subconsciously fragment space or tunnel, i.e. reduce the amount of information processed in order not to be overwhelmed in dealing with abstract or conceptual activities. In the physiological aspect, the complications of this adaptation for the hyperopic patient include moving the pelvis backward and the forehead forward, and the eyes feel like they are turned out.” This visual tunnelling affects their ability to control their attention, as well as motor and postural changes. For example, one might walk with smaller steps, round their shoulders, lock their knees when standing, and develop upper back, shoulder, or neck tension.

Nutrient deficiences such as low vitamin D as well as being sensorily overwhelmed cause eyesight issues (autoimmune diseases of the eye and visual tunneling respectively). For vitamin D what has been shown is an association: people with low vitamin D were about three times as likely to have noninfectious anterior uveitis.¹³

Sensory processing disorders and Autoimmune Disease

People who are highly sensitive (a genetic trait) are more likely to have autoimmune diseases such as diabetes, and I would postulate, celiac.¹⁴ Celiac and diabetes are often comorbid: about 6% of people with type 1 diabetes have biopsy-proven celiac disease.¹⁵

This study does not show causalty, it shows that 50% of children with autism also suffer gastrointestinal symptoms, such as celiac disease or constipation, comorbidly. Children with ASD have altered gut microbial composition, with decreased gut bacterial variety and underestimation of potentially helpful bacteria such as Bifidobacterium, Prevotella, and Desulfovibrio, as well as a decrease in short-chain fatty acids.¹⁶

It runs the other way too: in 3.56 million Danes, a prior autoimmune disease raised the risk of a later mood disorder by 45%.¹⁷

The Enteric Nervous System is a Sensory System

We usually think of the gut as something our senses act on. We would argue that the gut is itself a sense organ, and that it is built from the same parts as touch and taste.

Your gut has its own sensory neurons (intrinsic primary afferent neurons, or IPANs), and they can run a reflex from start to finish without asking the brain. This was only firmly shown in the mid-1990s. Something chemical or mechanical sets them off, they tell the interneurons, the interneurons tell the motor neurons.¹⁸ There is also work suggesting these neurons are not specialists, each one does several jobs and they act together as a sensory network, and this holds across different parts of the gut and across species.¹⁹

The lining of the gut has enterochromaffin cells, which carry chemosensory receptors, are electrically excitable, and talk through synaptic contacts to serotonin-sensitive nerve fibres²⁰. They also respond to touch: when pressed they release serotonin, and they express Piezo2, the same channel you use for light touch and for sensing where your limbs are²¹. Beyder’s group compares them to the Merkel cells in skin²¹.

Neuropod cells synapse directly onto the vagus nerve and use glutamate to pass along information about what is in the gut within milliseconds²², which means the inside of the gut is one synapse away from the brainstem.

Many of the same genes are involved in both. Over 90% of the 62 top-ranked autism risk genes (in SFARI) are expressed in the gut as well as in the brain²³. Cntnap2, a synaptic adhesion molecule which matters for sensory processing, is expressed mainly in enteric sensory neurons²⁴. SHANK3 is present in the myenteric plexus, and mice without it have guts that are structured and function differently²⁵.

Lauren Orefice’s lab reports that in mouse models of autism, the nerves which run from the gut to the spinal cord (dorsal root ganglion neurons) send heightened signals²⁶. The same lab found that mutations in the peripheral neurons alone were enough to make a mouse over-react to touch.²⁷

In humans, a study of 2,973 autistic children found that every type of chronic GI problem came with higher rates of sensory over-responsivity and of anxiety, and each of those independently predicted GI problems²⁸. A smaller study found the same combined sensory and GI profile in ADHD as well as in autism²⁹. These studies do not show causality, they are correlations from parent reports.

The microbiome affects how excitable these neurons are. Mice raised germ-free have less excitable enteric sensory neurons, and need a microbiome for their nerves to work normally^30,31. In the other direction, serine proteases made by commensal bacteria calm down dorsal root ganglion neurons³².

If enterochromaffin cells are kept activated for long enough the result is lasting visceral hypersensitivity, even with no inflammation to start it³³. I would postulate that this is the gut’s version of never getting used to the perfume.

Remark: Whether glutamate does anything inside the enteric nervous system itself is contested. One review says all of the glutamatergic machinery is there³⁴, but Wood’s lab found that neither glutamate nor its receptor agonists excited enteric neurons³⁵. The result we trust is the synapse from the neuropod cells to the vagus.

Remark: Nearly all of the mechanism above is from mice. We have not found a study which tests sensory processing disorder on its own against the enteric nervous system.

Hypersensitivity that Outlasts the Stimulus

The most relevant study for us is Bayrer et al. (2023).³³ They increased and decreased the activity of enterochromaffin cells in living mice. Increasing it was enough to make the gut hypersensitive to stretching, and the cells were needed for isovalerate, a short-chain fatty acid made by bacteria, to sensitize the gut. When the cells were kept active for long enough, the hypersensitivity lasted after the activation stopped, without any inflammation.

This connects to the first section, on people with chemical intolerance who don’t habituate to a smell. In both cases a sensitivity persists after the stimulus is gone. If there is one state linking these conditions, I would postulate that it is this kind of persistence, which has the advantage that it can be measured over time rather than asked about once in a questionnaire.

The same paper found that changing enterochromaffin activity produced anxiety-like behaviour, which went back to normal when serotonin signalling was blocked. The circuit also behaved differently in females, where it seemed to be engaged all the time. This would fit with irritable bowel syndrome affecting about three times as many women as men.³³

Remark: These are mice, and the genetic tools used would not be used in people. The study shows that this kind of persistence exists and identifies a circuit for it, but it doesn’t show that the same thing happens in people with sensory processing disorder.

Enteric Nervous System and Excitable Mental Illness

Our guts produce 90% of serotonin, if you are low on folic acid or any other basic vitamins/nutrients you literally cannot create serotonin, dopamine, and other neurotransmitters³⁶. Would a severe serotonin deficiency show up in a SPECT scan? SPECT cannot see serotonin itself, but it can see the serotonin transporter, and in a small study people with major depression had less transporter binding in the midbrain, the more depressed the less³⁷.

Gut bacteria help set the level: spore-forming bacteria push the enterochromaffin cells to make serotonin (measured in mice, and with human microbiota).³⁸

There is ongoing work to directly detect and monitor this for diagnostic and treatment purposes.³⁹

The overall changing of gut bacterial community in terms of β-diversity was consistently observed in ASD patients compared with HCs. Furthermore, Bifidobacterium, Blautia, Dialister, Prevotella, Veillonella, and Turicibacter were consistently decreased, while Lactobacillus, Bacteroides, Desulfovibrio, and Clostridium were increased in patients with ASD relative to HCs in certain studies.^40,41

Stress also changes the bacteria in the gut. A single two-hour social stressor is enough to shift the bacterial community in mice, and Lactobacillus decreases under stress in rodents, primates and humans.⁴²

There has been some recent research on fecal transplants curing some forms of mental illness. The inverse is also true “with the transmission of depressive and anxiety-like symptoms and behaviors resulting from the transplantation of microbiota from psychiatrically ill donors to healthy recipients”. We collect some examples here:^{43,44,45,46,47}

Enteric Nervous system/SPD causing Lowered Immune System causing Autoimmune Disease

If you have a lower immune system (low igA or igG or igM) you are more likely to have an autoimmune disease.

Remark: As a note, you are also more likely than average to have an autoimmune disease with an abnormally strong immune system. This is not the case we are discussing here.

The enteric nervous system generates igA and igG in our body. There is a link between igA and levels with a person’s perception of stress, implying the nervous system has the ability to control igA levels.⁴⁸

People with IgA deficiency have a different gut microbiota from everyone else, even though their bodies secrete IgM to compensate.⁴⁹

People with IgA deficiency should be tested for celiac disease because they are 10 to 20 times likely to develop an autoimmune response to gluten than the general population.⁵⁰

Using tests of raised igA and igG to detect for celiac disease is harmful and ineffective, as having naturally lowered igA and igG is usually comorbid with celiac disease. Thus, the rasied igA and igG register as normal levels on tests, giving many patients false negatives. The most effective test is a DNA test for the celiac markers.⁵¹

Immunosenescence or the aging of the immune system is linked to the increase of autoantibody frequency while the antibodies decline⁵².  One way to frame the connection between SPD and being immunocompromised is that higher sensitivity leads to greater stress / perceived stress, which leads to a decrease in immunity which then naturally causes an increase in autoantibodies that would otherwise have occured during someones natural immunosenescence.⁵³

In 106,464 Swedes with a stress-related disorder, the risk of a later autoimmune disease was 36% higher, and 46% higher with PTSD.⁵⁴

Mechanisms of Functional Dyspepsia and Treatment with Tricyclic Antidepressent

A proposed mechanism is that there is original inflammation, and then continued faulty pain signaling when input is gone. That is, a central sensitization model.^55,56,57,58

Autoimmune Disease is Under Direct Enteric Neural Control

The following are quotes from this paper:⁵⁹

“As in the salivary gland, sIgA is rapidly secreted in mucosal explants from porcine distal colon in response to both cholinergic and adrenergic receptor agonists, but remains unaltered by acute exposure to an enteroadherent microbial pathogen. The enteric neurotransmitters acetylcholine and NE appear to act on muscarinic cholinergic and alpha-adrenergic receptors expressed by crypt epithelial cells to promote transcytosis of the polymeric Ig receptor and enhance vectorial secretion of both sIgA and free SC toward the mucosal surface. In its early stages, this action appears to be temporally coincident with, but mechanistically independent of active ion transport. The precise receptor types and associated signal transduction mechanisms mediating sIgA secretion by these and possibly other enteric neurotransmitters at this mucosal immune effector site remain to be characterized.”

“In a more complex, vascularly-perfused porcine ileum preparation however, electrical stimulation of extrinsic nerves innervating the intestine increases sIgA secretion only after atropine and phentolamine pretreatment; this effect is inhibited by hexamethonium, a blocker of ganglionic neurotransmission. This result suggests that there may exist important inhibitory reflex circuits containing muscarinic cholinergic and alpha-adrenergic receptors which modulate the enteric neural control of sIgA output. Additional investigations will be necessary to establish how basal sIgA secretion onto the mucosal surface in vivo is modulated by the release of acetylcholine from enteric nerves.”

Virome Mediates Stress and Memory Responses

This will be a topic of a future post, it is critical to note that in fact the virome controls the gut biome entirely. Bacteriophages maintain biodiversity of a healthy gut.⁶⁰

Chronic stress changes the phages in a mouse’s gut, and giving a stressed mouse the virome it had before the stress protects it from the behavioural and immune effects.⁶¹

In children at genetic risk of type 1 diabetes, the gut virome became less diverse before the autoantibodies appeared.⁶²

How to Tell One Cause from Several

The evidence so far is pairwise

Each link in this essay connects two things at a time, and the pairs come from different populations. The links between autoimmune and psychiatric conditions come from national registries of diagnoses in adults: autoimmune disease before mood disorders in 3.56 million Danes,¹⁷ and stress-related disorders before autoimmune disease in 106,464 Swedes.⁵⁴ These registries have no information on sensory processing. The links between sensory, gut and anxiety symptoms come from autistic children, reported by their parents: 2,973 children in the Autism Treatment Network,²⁸ and a smaller cross-disorder study.²⁹ These cohorts have no information on autoimmune disease.

Putting these pairs together doesn’t answer our question. If sensitivity is associated with gut problems in one group and autoimmunity with mood disorders in another, that is consistent with a single shared cause and with several overlapping ones. To tell them apart all of the conditions have to be measured in the same people.

Predictions

If there is a single shared state, then once you know a person’s level of it, their gut symptoms shouldn’t tell you anything more about their sensory reactivity, anxiety or autoimmune risk. In other words the correlations between domains should go away after conditioning on the shared state. If there are several, some correlation will be left over after conditioning on the best single state, and it should cluster, for example sensory with anxiety and gut with autoimmune, so that a model with two or more factors fits noticeably better than a model with one.

We would consider the single-state hypothesis refuted if the leftover correlations between domains are about as large as the raw ones, if a two-factor model fits clearly better and separates the sensory side from the immune side, or if the shared state doesn’t predict persistence (the failure to habituate from the first section) when that is measured directly.

Methods

We would use a latent-variable model, where each person has one latent state (or several, in the competing model) and each observation is a noisy measurement of it. Different kinds of observations need different measurement models. Parent reports of sensory and gut symptoms are ordinal and depend on the parent as well as the child. Diagnostic codes are binary, miss cases, and depend on how often someone sees a doctor.

There are datasets this could be fit on. SPARK, the Simons Foundation’s cohort, was designed to recruit 50,000 autism families with genetic data.⁶³ The Simons Simplex Collection has about 2,600 families with one autistic child and their unaffected siblings, with detailed phenotyping.⁶⁴ Both collect parent-reported medical history along with genetics, although we still need to check which sensory, gut and immune items each one includes.

Inference is the difficult part. What we want to estimate is how strongly the domains are correlated. The standard fast method for models of this size, factorized (“mean-field”) variational inference, sets the posterior correlations between latent quantities to zero and underestimates their variances.⁶⁵ For our question this means it would be biased toward finding no correlation between domains, and overconfident about it. So the model would need full-rank or structured approximations, checked against exact sampling on subsets of the data.

Genetics

If the shared state exists it should have a genetic component. Over 90% of the 62 top-ranked autism risk genes are expressed in the gut as well as the brain.²³ The first large genome-wide study of autism also found five places in the genome where common variants raise the risk.⁶⁶ We would test whether the latent state tracks rare damaging variants in these genes, or polygenic risk for autism. With a single state, a gene’s effects on the senses, the gut and mood should all go through it, and a gene that changes gut symptoms without changing the state would be evidence for more than one. The siblings in the Simons Simplex Collection make within-family comparisons possible.

An animal readout

Bayrer’s result suggests an experiment in mice: activate the enterochromaffin cells and measure how long the effect lasts, which a single behavioural test at one time point can’t do. Ultrasonic vocalizations can be recorded in the home cage continuously, without handling the animal. I worked on software for this kind of monitoring at Vium (Learning the ultrasonic language of rodents). The prediction is that the calls change after activation and stay changed after it stops, on the same time course as the visceral hypersensitivity. How mouse calls relate to pain and anxiety is only partly understood, so this would have to be validated against the standard behavioural tests first.

Keyword summary:

• visual tunneling
• chemical intolerance
• chronic pain
• excitation/inhibition balance
• GABA
• high sensitivity
• autism/ocd/add/ptsd/bipolar
• functional dyspepsia
• sensory processing disorder.
• low autoimmune system
• likelihood to have autoimmune disease
• enteric nervous system
• lowered ability to create serotonin
• joint hyperflexibility