Touch Over-reactivity in Autism Models

Deleting a gene from peripheral sensory neurons, while sparing the brain and spinal cord, was enough to change how a mouse reacted to a gentle puff of air, and to shift its scores in anxiety-like and social assays. The brain in those animals carried the gene throughout. That experiment is the reason this lesson exists, and most of the work of reading it is keeping straight what it does and does not claim.

Touch and pain gave you the afferent classes and the dorsal horn; dorsal horn touch gave you presynaptic inhibition of afferent terminals and the interneurons that deliver it. The genetics of Rett syndrome, fragile X and the autism-associated synapse genes belong to synapse disorders. Start where those three meet: a tactile phenotype in mouse models whose cause sits in a neuron outside the brain.

What “over-reactive to touch” means in people, and what it does not

Sensory features are in the diagnostic criteria. Criterion B4 of the DSM-5 covers hyperreactivity, hyporeactivity and unusual interest in sensory aspects of the environment, and it is one of four restricted and repetitive behaviour criteria of which at least two must be met. Two things follow. Sensory difference runs in both directions, and a sensory feature is not required for a diagnosis. Not every autistic person is over-reactive to touch.

The laboratory version of the question is narrower still. Puts and colleagues (2014) tested 32 autistic and 67 typically developing children aged 8 to 12, with vibration delivered to the second and third digits of the left hand. They measured static detection thresholds, a dynamic ramp task at 25 Hz rising at 2 µm per second, and amplitude and frequency discrimination.

Work the numbers. Table 1 of that paper gives a mean static detection threshold of 6.59 ± 2.81 µm in the typically developing cohort and 9.13 ± 4.81 µm in the autistic cohort (mean ± SD, human children, in vivo psychophysics).

difference = 9.13 µm − 6.59 µm = 2.54 µm ratio = 9.13 / 6.59 = 1.385 relative increase = (2.54 / 6.59) × 100 = 38.5 per cent

The autistic cohort needed a stimulus about 38.5 per cent larger to detect it. A higher threshold is poorer detection, so whatever everyday over-reactivity is, it is not a lower detection threshold in this task. Baseline amplitude discrimination was also worse in that cohort, while frequency discrimination showed no significant diagnosis effect. Those standard deviations, 2.81 and 4.81 µm, are large next to a 2.54 µm difference: the distributions overlap heavily, and a group mean predicts nothing about one child.

Hold that beside the mouse readout you are about to meet. The mouse measure is the size of a startle-like movement to a fixed stimulus, a 0.9 psi puff lasting 50 ms on back hairy skin. That is reactivity at one stimulus level, not a threshold, and not a report of what the animal perceives. Human psychophysics and mouse reactivity are two different quantities, and the lesson keeps them apart.

A peripheral locus: conditional deletion and selective restoration

Orefice and colleagues (2016, Cell, Ginty laboratory) began with four mouse lines carrying mutations in genes associated with autism in people: Mecp2-null, Mecp2R306C, Shank3B heterozygous and Fmr1-null. All four reacted more strongly than controls to the air puff, tested from 6 to 8 weeks of age.

Then came the restriction. Advillin-Cre supplies Cre recombinase to primary somatosensory neurons, so crossing it to a floxed allele removes that gene from those cells and leaves the brain’s copy intact. (A Cre driver labels the cells that expressed it at some time, which here is the whole population of dorsal root and trigeminal neurons, plus sympathetic ganglion neurons, not one subtype.) Deleting Mecp2 or Gabrb3 this way was sufficient for the tactile phenotype: stronger air-puff responses, altered tactile prepulse inhibition, and impaired texture discrimination in a textured novel-object task. It was also sufficient for anxiety-like and social-interaction differences in the assays tested.

Two controls make that result readable rather than merely striking. First, timing. An inducible Advillin-CreERT2 version let the same deletion begin at P28 instead of during development. The tactile deficits still appeared. The anxiety-like phenotype did not. The paper’s word for that condition is “adult deletion”, but the induction age is P28, and that is the number to keep. Second, direction. Restoring Mecp2 only in the Advillin-targeted neurons (mostly sensory, with the sympathetic coverage noted above), on an otherwise null background, rescued tactile sensitivity and selected anxiety-like and social measures. Lethality, memory deficits and motor deficits were not rescued. A peripheral contribution to part of the phenotype, then, and an explicit failure to account for the rest.

What is wrong in the mutant terminals? GABRB3-containing GABA-A receptors localise to the presynaptic compartments of Aβ- and Aδ-LTMR central terminals in mouse dorsal horn, that is, to the endings of peripheral neurons inside the spinal cord. (An LTMR is a low-threshold mechanoreceptor afferent.) In sensory-neuron Mecp2 mutants, GABRB3 puncta associated with vGlut1-positive terminals were reduced by about 70 per cent. That figure counts immunolabelled puncta in fixed tissue; it is not a measured conductance. Two physiological results point the same way: in P14 to P19 slices, dorsal-root-evoked quantal EPSC frequency onto postsynaptic dorsal column neurons went up, and in isolated cords, evoked dorsal root potentials were greatly diminished after sensory-neuron deletion of Mecp2 or Gabrb3. Dorsal root potentials are the back-propagating signature of primary afferent depolarisation. The authors attributed the touch phenotype to deficient GABA-A-mediated presynaptic inhibition.

Mind the sign. GABA acting on an afferent terminal depolarises it, because those terminals hold chloride high (synaptic integration sets out why the sign follows the gradient, not the transmitter). The 2016 study interprets that depolarisation as reducing action-potential-evoked release. Depolarising is not the same as exciting.

Not one mechanism: a second model and a drug that stays outside

The 2019 follow-up (Orefice and colleagues, Cell) broke the single story. Large-diameter dorsal root ganglion neurons cultured one to two days from 6- to 8-week-old Shank3 mutants had reduced Ih, the hyperpolarisation-activated current, and increased excitability. No altered GABRB3 expression was detected at Shank3 LTMR central terminals. Same behavioural phenotype, different cellular defect.

The same paper asked whether a drug acting only outside the brain could change the behaviour. Isoguvacine is a GABA-A agonist; its EC50 was 19.7 µM at recombinant human α1β3γ2 and 18.7 µM at α2β3γ2 receptors expressed in HEK293T cells, an in vitro receptor measurement and not a tissue exposure target. After 20 mg/kg intraperitoneally in male mice aged 8 to 12 weeks, LC-MS detected the compound in blood and liver but not in brain or cerebrospinal fluid, with detection limits of 640 pM in fluids and 16 nM in tissue. Acute 2 mg/kg dosing reduced tactile over-reactivity in five genetic models and one maternal-immune-activation model, without the sedation reported for the brain-penetrating comparator midazolam in those assays.

The control that makes this a mechanism claim rather than a drug effect is genetic. In anaesthetised mice, isoguvacine reduced stimulus-evoked Aβ-LTMR firing recorded from the L4 ganglion. In mice with homozygous sensory-neuron Gabrb3 deletion, it did not. Remove the receptor subunit from the neurons and the drug stops working on them.

Chronic dosing drew the limits again. Mecp2R306C and Shank3B heterozygous litters received 2 mg/kg daily from P1 to P42, with testing from P42. Anxiety-like behaviours and some social impairments improved. Texture discrimination deficits persisted, as did Mecp2 motor impairment and Shank3 overgrooming. This is a selective effect on some mouse endpoints. Isoguvacine is not an approved treatment for anything, I found no registered human study of it on that date, and the 2019 authors present peripheral GABA-A intervention as a strategy that would require testing in patients.

Timing: early afferent receptors, later spinal inhibition

Tasnim and colleagues (2024, Nature Neuroscience, Ginty corresponding author, Orefice a coauthor) asked when each model’s phenotype starts. Gabrb3 and Mecp2 loss-of-function pups were already more reactive to an air puff at P4, measured as maximal body displacement within 500 ms. Nlgn2 and Rorb mutants were not, despite tactile abnormalities as adults. Sampling two ages does not pin the onset day; it does separate early-onset from later-onset models.

The machinery matches that split. GABRB3 immunoreactivity was detectable near VGLUT1-positive sensory terminals at P0 to P1, and Gabrb3 deletion virtually abolished GABA-evoked depolarisations in medium- and large-diameter dorsal root ganglion neurons at both P4 and P18 to P30 (recorded with a high-chloride internal solution, which sets the driving force the experimenter wanted to see). Deep dorsal-horn miniature inhibitory currents, by contrast, were smaller and rarer at P4 than at P19 to P21, and feedforward inhibition in the LTMR-recipient zone was predominantly glycinergic by P19. Nlgn2 sits in that later system: Lbx1-Cre deletion from dorsal-horn neurons was sufficient for adult tactile alterations, and P19 to P21 Cdx2-Cre mutants (deleted in all cells below cervical level 2) had reduced dorsal-horn mIPSC frequency and amplitude. In E18.5 exteriorised embryos, optogenetic activation of forepaw Ret-positive Aβ-LTMRs evoked greater movement in Gabrb3 heterozygotes than in controls (11 control and 9 heterozygous embryos). That is evoked movement in an embryo preparation, not a prenatal test of anything.

Across the models tested, neonatal over-reactivity was associated with adult anxiety-like and social differences. Association across genotypes is not a demonstrated causal sequence, and mouse developmental windows have no validated human equivalent. Why timing should matter at all is the subject of sensitive periods in behaviour and critical period control.

Design problem

Test whether early tactile over-reactivity causes later anxiety-like behaviour in mice, or merely predicts it.

What this does and does not mean for people

The mouse work establishes that in these models part of a tactile phenotype, and some behavioural measures that go with it, can begin in neurons outside the brain. It does not make autism a peripheral condition. The genetic manipulations are sufficiency tests in one species, the behaviours are operational assays rather than experiences, and the rescue experiments left major features untouched. In people, the measurable differences are task-specific, go in both directions, and vary between individuals far more than any group mean conveys. No therapy targeting peripheral GABA-A receptors is approved, and the registry searches that back that statement were queries run on one date, which cannot exclude every jurisdiction or unregistered study.

What would settle the next question is a human measurement with the specificity of the mouse genetics behind it. Chronic pain and spinal pain circuits face the same gap from the other side.