Biology of Autism — Molecular Pathways Map

Upstream Triggers

SIRT1 Hub & Functions

Astrocyte / Glial Layer

Synapse Architecture

ASD Features

Somatostatin (SST) Pathway

Self-Sustaining Loop

Upstream
Triggers

MIA

L1A — Maternal Immune Activation

Infections, autoimmunity, or stress during pregnancy elevate IL-6 and IL-17a in fetal circulation. Directly alters cortical neuron migration and programs long-term glial reactivity. Animal MIA models produce autism-relevant behavioral profiles. Developmental priming: MIA causes epigenetic silencing of SIRT1 and PGC-1α in offspring microglia and specifically impairs PV/SST interneuron development — meaning the SIRT1 deficiency and SST vulnerability begin before birth. MIA also alters the fetal gut environment, so affected children may be born with an already-compromised GI barrier.

IL-6 ↑ IL-17a ↑ → cortical dysplasia

Gut

L1B — Gut Dysbiosis + LPS Translocation

Disrupted microbiome → impaired intestinal barrier → LPS enters systemic circulation → TLR4 activation on macrophages and microglia → perpetual NF-κB-driven inflammation. Metabolic endotoxemia without active infection.

LPS → TLR4 → NF-κB → sustained cytokines

Mitochondria

L1C — Mitochondrial Dysfunction

Impaired mitochondria release mtDNA fragments and reactive oxygen species → NLRP3 inflammasome activation → IL-1β + IL-18. Simultaneously reduces ATP available for hevin/glypican secretion and synaptic transmission.

ROS + mtDNA → NLRP3 → IL-1β ↑ IL-18 ↑

Toxins

L1D — Environmental Toxin Load

Heavy metals, glyphosate, and organochlorine pesticides are associated with increased ASD risk in epidemiological studies. Proposed mechanisms include activation of innate immune pathways, glutathione depletion, impaired cellular detoxification, and microglial priming. Evidence is primarily association-level in humans; mechanistic data from animal and cell models.

GSH ↓ → oxidative burden ↑ → immune activation (association-level evidence)

chronic inflammatory signal → IDO1 induction → tryptophan diverted · stress → SST release

◈ Developmental Priming (dashed): MIA epigenetically silences SIRT1/PGC-1α in fetal microglia and impairs SST interneuron development in utero — the spiral can be initiated before birth, independent of postnatal triggers.
    

Somato-
statin
Node

Co-Equal Central Node — Stress + Inflammation Activated

L2B — Somatostatin (SST) Elevated

Somatostatin is released in response to the same upstream stressors — chronic stress, inflammation, metabolic strain, and excess neuronal activity — that drive the kynurenine/functional NAD⁺ insufficiency cascade. Acting via SSTR2 and SSTR5 receptors on astrocytes, microglia, hippocampal neurons, and thalamic relay cells, elevated SST functions as a system-wide brake on plasticity, synaptogenesis, and gut function. SST and SIRT1 are antagonistic co-regulators: SST pushes the system toward conservation and inhibition; SIRT1 toward resilience and growth. Both converge on CREB suppression from different directions.

Stress + Inflammation → SST ↑ → SSTR2/5 → AC ↓ → cAMP ↓ → PKA ↓ → CREB ↓ → BDNF ↓

SST on Glia

L2C — Astrocyte + Microglial Suppression via SST

SSTR2/5 on astrocytes: suppresses hevin (SPARCL1) and glypican secretion, reduces glutamate clearance, increases SPARC — directly worsening the synapse protein imbalance at Level 5. SSTR2/5 on microglia: promotes low-plasticity, high-surveillance reactive state, lowering the threshold for inflammatory activation and amplifying M1 microglial shift.

SST → astrocytes: Hevin ↓ Glypicans ↓ SPARC ↑ · microglia: reactive priming ↑

SST on Gut

L2D — GI Brake: Malabsorption + Dysbiosis

SST inhibits gastric acid secretion, pancreatic enzyme output, bile release, and intestinal motility. Result: malabsorption of amino acids (tryptophan, tyrosine, phenylalanine), reduced nutrient availability for mitochondrial support, altered microbiome composition, and fluctuating neurotransmitter precursors. SST links the stress response directly to gut–brain disruption.

SST → gastric acid ↓ · enzymes ↓ · motility ↓ → amino acid absorption ↓ → microbiome shift

Metabolic
Pivot

Kynurenine Pathway

L3A — Tryptophan Hijack: Functional NAD⁺ Insufficiency

Inflammation induces IDO1, shunting tryptophan away from serotonin and into the kynurenine pathway. Under chronic inflammatory conditions, quinolinic acid — a neurotoxic NMDA agonist — accumulates faster than it can be converted to NAD⁺ by the terminal enzyme QPRT. The result is not classical NAD⁺ depletion but functional NAD⁺ insufficiency: NAD⁺ regeneration rate lags the metabolic demand created by ongoing inflammation. Triple consequence: serotonin depleted, quinolinic acid elevated, NAD⁺ regeneration insufficient to sustain SIRT1 activity.

Tryptophan → (IDO1) → Kynurenine → Quinolinic Acid ↑ NAD⁺ ↓ Serotonin ↓

NAD⁺ Insufficiency

L3B — SIRT1 Fuel Depleted

SIRT1 requires NAD⁺ as its catalytic cofactor. Without adequate NAD⁺, SIRT1 activity falls regardless of gene expression levels. All four downstream regulatory systems — inflammation, mitochondria, antioxidants, synaptic plasticity — begin to fail simultaneously.

NAD⁺ ↓ → SIRT1 activity ↓ → 4-system regulatory failure

metabolic stress + NAD⁺ insufficiency → AMPK/mTOR dysregulation → cleanup failure

Metabolic
Reset
Failure

AMPK / mTOR / Autophagy — Dysregulated

L4A — Cellular Cleanup and Reset Failure

SIRT1 and AMPK are co-activators of the cellular energy-sensing and cleanup axis. Under metabolic stress and NAD⁺ insufficiency, AMPK is under-supported while mTOR — the growth-and-suppress-autophagy signal — becomes relatively overactive. Result: the cell's recycling and reset system fails. Damaged mitochondria are not cleared (impaired mitophagy). Misfolded proteins accumulate. Synaptic pruning is dysregulated. Inflammatory signaling persists because its substrate — dysfunctional organelles and debris — is not removed. This is the mechanism by which acute metabolic stress becomes chronic structural disorder. The PTEN/TSC/Fragile X mTOR literature documents this in genetically defined ASD subtypes; the same axis is engaged by inflammatory-metabolic stress in idiopathic ASD.

SIRT1 ↓ + metabolic stress → AMPK ↓ / mTOR ↑ → autophagy ↓ / mitophagy ↓ → debris accumulates → inflammation persists · synaptic remodeling impaired

Cleanup Failure Consequences

L4B — Why Damage Becomes Permanent

Impaired autophagy explains three features the rest of the cascade does not fully account for: (1) why damaged mitochondria keep accumulating despite upstream trigger reduction; (2) why excess dendritic spines persist — pruning requires autophagy machinery; (3) why the inflammatory loop is self-sustaining — dysfunctional organelles continuously re-activate NLRP3 and NF-κB. Restoring autophagic flux is a prerequisite for structural recovery, not just symptom management.

autophagy ↓ → damaged mito persist → NLRP3 re-activated · spine excess → local over-connectivity · debris → NF-κB re-activated

NAD⁺ insufficiency + cleanup failure → SIRT1 catalytic activity silenced

Central
Hub

Master Regulator — NAD⁺ Dependent Deacetylase

L5A — SIRT1 Deficient

SIRT1 (Sirtuin-1) is expressed in neurons, astrocytes, microglia, and endothelial cells. Its developmental expression peaks during synaptogenesis and synaptic pruning — precisely the windows disrupted in autism. When NAD⁺ is functionally insufficient, SIRT1 can no longer coordinate the four systems it normally regulates.

SIRT1 ↓ → NF-κB ↑ + PGC-1α ↓ + FOXO ↓ + CREB ↓

SIRT1
Outputs

Output 1 — Inflammation

L5B — NF-κB Unleashed

SIRT1 normally deacetylates the p65 subunit of NF-κB, suppressing cytokine gene transcription. Without SIRT1, NF-κB activity rises → sustained IL-6, TNF-α, IL-1β → microglial M1 polarization → astrocyte A1 conversion.

SIRT1 ↓ → NF-κB p65 hyperacetylated → cytokines ↑

Output 2 — Mitochondria

L5C — PGC-1α Falls

SIRT1 activates PGC-1α, the primary driver of mitochondrial biogenesis. Without SIRT1, PGC-1α is inactive → impaired mitochondrial renewal → reduced ATP + elevated ROS → NLRP3 feedforward loop.

SIRT1 ↓ → PGC-1α ↓ → ATP ↓ ROS ↑

Output 3 — Antioxidants

L5D — FOXO Defense Collapses

SIRT1 activates FOXO transcription factors → catalase, SOD, and peroxiredoxin production. Without SIRT1, FOXO is inactive → glutathione depleted → oxidative stress unopposed → astrocyte damage and hevin misfolding.

SIRT1 ↓ → FOXO ↓ → GSH ↓ → ROS damage

Output 4 — Learning · SST + SIRT1 Convergence · Lethal Loop

L5E — CREB/BDNF Suppressed from Both Directions

CREB suppression arrives via two independent routes that reinforce each other. Route 1 — SIRT1 deficiency: reduced SIRT1 → impaired CREB co-activation → BDNF falls → LTP impaired → circuits fail to strengthen through experience. Route 2 — SST elevation: SSTR2/5 inhibits adenylyl cyclase → cAMP falls → PKA underactivated → CREB phosphorylation drops → BDNF transcription suppressed. Both routes land on the same outcome: a chronically hypoplastic neural state during the critical developmental windows when CREB activity is required for learning, synaptic refinement, and behavioral flexibility. Lethal Loop: Low SIRT1 → low cAMP → low CREB → reduced BDNF → impaired plasticity → stress response maintained → SST remains elevated → AC further suppressed → SIRT1 further depleted. The two nodes lock each other in a self-reinforcing arrested state.

SIRT1 ↓ → CREB ↓ · SST ↑ → AC ↓ → cAMP ↓ → CREB ↓ → BDNF ↓ → LTP impaired

NF-κB activation → IL-1β + TNF-α + C1q triad → astrocyte A1 conversion

Glial
Layer

M1 Microglia

L6A — Microglial Pro-Inflammatory Shift

NF-κB-driven cytokines push microglia from homeostatic to M1 states. Activated M1 microglia release the IL-1β + TNF-α + C1q triad — the precise signal that converts astrocytes to A1 reactive phenotype. Also directly prune synapses via complement-mediated engulfment.

NF-κB → M1 microglia → IL-1β + TNF-α + C1q ↑

A1 Reactive Astrocytes

L6B — Astrocyte Phenotype Shift

The IL-1β + TNF-α + C1q triad from M1 microglia converts astrocytes from their normal supportive (A2) state to A1 reactive state. A1 astrocytes are characterized by elevated GFAP, S100β, complement C3, and a complete reversal of their synaptogenic protein profile.

IL-1β + TNF-α + C1q → A1 astrocytes → GFAP ↑ C3 ↑

A1 astrocytes invert synaptogenic protein balance

Synapse
Proteins

Hevin (SPARCL1) ↓ Suppressed

L7A — Synapse Builder Silenced

Hevin bridges Neurexin-1α to Neuroligin-1, initiating thalamocortical synapse assembly. In A1 astrocytes, hevin production falls. SPARCL1 mutations associated with ASD risk reduce hevin secretion and trigger ER stress. Fewer synapses form in sensory integration and social processing circuits.

Neurexin-1α ←[Hevin]→ Neuroligin-1 ✗ blocked

SPARC ↑ Over-Expressed

L7B — Pruning Signal Pathological

SPARC binds hevin and physically blocks the Neurexin-Neuroligin bridge. In A1 astrocytes, SPARC is massively over-expressed — triggered by TNF-α, IL-1β, and microglial activation. Elevated SPARC removes synapses that developmental circuits required. SPARC as DAMP: Secreted SPARC also acts as a damage-associated molecular pattern (DAMP), activating TLR4 on microglia and feeding back into NF-κB — creating a direct synapse-to-inflammation feedforward loop.

SPARC ↑ → blocks Hevin bridge → synapse elimination ↑

Glypicans 4/6 ↓ Irregular

L7C — Synapse Maturers Impaired

Glypicans recruit AMPA receptors (GluA1) to postsynaptic sites, converting silent synapses into functional ones. In A1 astrocytes, glypican secretion becomes irregular. Existing circuits remain structurally present but functionally silent — incapable of efficient signal transmission.

Glypican 4/6 ↓ → AMPA ↓ → silent synapses persist

hevin ↓ + SPARC ↑ + glypicans ↓ → aberrant synaptic architecture

Neural
Architecture

Thalamocortical Circuit

L7D — Under-Connected Sensory Relay

Hevin is most critical for thalamocortical synapse formation. Reduced hevin → fewer, weaker connections between thalamic sensory relay nuclei and cortical processing regions → sensory information arrives at the cortex unfiltered and poorly integrated.

Thalamus ⟷ Cortex: hevin-dependent synapses ↓

Connectivity Signature

L7E — Hyper-Local + Reduced Long-Range Connectivity

The hevin/SPARC imbalance produces the connectivity pattern consistently documented in autism neuroimaging: dense local connectivity within cortical regions, with reduced long-range integration across brain networks. Both over- and under-connectivity are present simultaneously — in different circuits.

Local connectivity ↑ | Long-range integration ↓

aberrant connectivity → functional and behavioral expression

ASD
Features

Sensory

L8A — Sensory Over-Reactivity

Unfiltered thalamocortical input → sensory information not gated or integrated before reaching conscious awareness → sensory over-responsivity to sound, touch, light, texture. Common in 80%+ of autistic individuals.

Social

L8B — Social Cognition Difficulty

Complex social understanding requires long-range integration across medial prefrontal cortex, superior temporal sulcus, and amygdala. Reduced long-range connectivity → social circuit under-development → difficulty with theory of mind, pragmatic language, social reciprocity.

Cognitive

L8C — Rigidity + Perseveration

CREB/BDNF impairment (via SIRT1 deficiency) + silent synapses (via glypican deficit) → circuits fail to update through experience → cognitive rigidity, difficulty with transitions, restricted interests, perseverative patterns.

L8D — GI Disruption

Gut dysbiosis (upstream trigger) + serotonin depletion (kynurenine pathway) + vagal nerve inflammatory signaling → dysmotility, constipation, pain, leaky gut. GI symptoms and autistic traits share a common upstream origin — the inflammatory cascade.

Sleep + Mood

L8E — Sleep Dysregulation + Mood Instability

Serotonin depletion (kynurenine pathway) → impaired melatonin synthesis → circadian dysregulation. HPA axis dysregulation (cortisol blunting) → emotional dysregulation, anxiety, low frustration tolerance.

Theoretical framework — not clinical guidance. The Autism Spectrum Disorder (ASD) Cascade is a systems-biology model integrating peer-reviewed findings across immunology, metabolism, gut biology, and neuroscience into a proposed mechanistic map. Individual components are supported by published research; the full integrated cascade has not been validated as a unified model in large clinical trials. It is intended as a research-informed framework — not a diagnostic tool or treatment protocol. All intervention decisions require qualified clinical oversight. For the evidence base, see the ASD Cascade Citations document in this suite.