WikiAdultCognitiveDisease.com

The Map of Collapse

An interactive map of Alzheimer's disease as one homeostatic system failing in three successive cell populations — the three-phase Temporal Architecture, its two named bridges, the Groundwork / Convergence / Synthesis layers, the seven convergence nodes, and the genes each phase implicates. Click any card to expand.

Phase I — Bioenergetic Ignition
Phase II — Microglial Bridgehead
Phase III — Synaptic Disintegration
Groundwork
Convergence
Synthesis
⚡
Phase I — Bioenergetic Ignition
Locus coeruleus & brainstem aminergic nuclei · decades 3–5 (age 20–50), clinically silent
A decades-long, clinically silent erosion of mitochondrial and autophagy–lysosomal quality control — driven by PARP-1 hyperactivation, NAD⁺ depletion, and obstruction of mitochondrial protein import — pushes the LC past the point where damaged organelles can be cleared.
PARP-1 → NAD⁺ depletion Aβ ⊻ TOM40 import block PINK1/Parkin mitophagy failure PANTHOS

Mechanisms

PARP-1 hyperactivation → NAD⁺ depletion; mitochondrial protein-import obstruction (Aβ blocking TOM40); PINK1/Parkin mitophagy failure; autophagy–lysosomal / autolysosomal-acidification failure (BIN1 at the formation end, presenilin at the degradation end); PANTHOS (terminal autolysosomal swelling — the plaque as "gravestone"); catecholaminergic oxidative load plus autonomous-pacemaking metabolic demand with no reserve; ROS output priming the NLRP3 effector arm; a silent noradrenergic/serotonergic prodrome (sleep, mood, arousal).

The Silent Locus (spine, step 1)

The universal, age-dependent erosion of cellular quality control crosses from compensated to decompensated first in the locus coeruleus, because that neuron — autonomous pacemaking, vast arborization, catecholamine oxidative load — has no metabolic reserve. The result is undegraded damaged mitochondria and intraneuronal Aβ, ROS that primes NLRP3, and a prodrome that stays clinically silent for two to three decades.

Why it is "first"

Anatomically the earliest station of the Braak rostral order (brainstem → limbic → cortex); substrate is metabolic — the first of the architecture's three shifting substrates.
🛡️
Phase II — The Homeostatic Microglial Bridgehead
Hippocampus (resident microglia) · decades 6–7 (age 50–70)
The TGF-β/SMAD-maintained homeostatic microglial signature collapses, converting the brain's resident immune cells into post-homeostatic states in which protective withdrawal and destructive attack become a single, inseparable act.
TGF-β/SMAD collapse DAM / LDAM transition TREM2 pivot Complement tagging

Mechanisms

TGF-β/SMAD homeostatic-signature collapse (loss of P2RY12/TMEM119/CX3CR1/SALL1/HEXB; SMAD7 rises); the DAM transition (TREM2-dependent: APOE, CST7, cathepsins, LPL, SPP1, ITGAX); LDAM (lipid-droplet-accumulating microglia); dystrophic (iron/ferritin-driven, senescent) microglia; TREM2 pivot signaling (DAP12 → SYK → PI3K–AKT–mTOR); complement tagging & engulfment (C1q/C3 → CR3; C4d–LilrB2); NLRP3 inflammasome assembly; "attack = failure" at the perineuronal net.

The Bridgehead (spine, step 3)

Under chronic stress the homeostatic microglial program collapses; microglia enter DAM/LDAM/dystrophic post-homeostatic states pivoting through TREM2. Attack and failure become one event, and the hippocampus consolidates the first self-sustaining lesion — the bridgehead. Resilience (per de Vries) = preserved homeostatic microglia plus an intact PNN: pathology present, but not dementing.

Why it is "second"

The disease has shifted substrate from metabolic to immunological, and location from brainstem to limbic system.
🧱
Phase III — Synaptic Disintegration
Cortical & hippocampal PV interneurons + perineuronal nets · decade 8+ (age 70+), where the disease becomes visible
The perineuronal net is digested, the parvalbumin interneuron loses its protective coat and inhibitory competence, excitatory–inhibitory balance collapses, and the self-sustaining circuit failure experienced as dementia ensues.
PNN digestion (aggrecan/brevican) PV disinhibition E/I balance collapse Gamma-rhythm degradation

Mechanisms

Perineuronal-net digestion (aggrecan/brevican/tenascin-R loss); parvalbumin-interneuron disinhibition (loss of perisomatic inhibition); excitatory–inhibitory balance collapse; gamma-rhythm degradation; retromer/endosomal-trafficking failure (VPS35/VPS26/VPS29, SORLA) — the Phase-I quality-control lesion recurring in cortical neurons; ApoER2–Dab1 recycling arrest by reactive lipid aldehydes (APOE4); intraneuronal Aβ in multivesicular/late endosomes; glutamatergic/NMDA insufficiency with tau & Aβ as maladaptive compensatory scar; the crossing from the intrinsic to the extrinsic pole of cell death.

The Visible Disease (spine, step 5)

Net digestion strips the PV interneuron of its structural scaffold and antioxidant buffer → disinhibition, E/I collapse, gamma-rhythm degradation, and a compounding feed-forward loop (disinhibition → excitotoxic/oxidative load → more microglial activation → more matrix digestion). The crossing from the intrinsic (slow, ageing, survivable) to the extrinsic (excitotoxic/ferroptotic/phagoptotic/necroptotic) pole is why dementia, once it arrives, arrives comparatively fast.

Why it is "third"

Substrate is now structural (the aggrecan–brevican matrix), location cortical — the terminal station of the rostral order.

The Temporal Architecture

One disease read as a sequence in time, not a single lesion

Brainstem → limbic → cortex (anatomically directional, the Braak rostral order); metabolic → immunological → structural (substrate-shifting — why no single-substrate theory fits the whole disease); and bridged — the transitions between phases are named mechanisms, not gaps. The Locus Coeruleus Bridge carries Phase I into Phase II; the Proteolytic Turn carries Phase II into Phase III.

The Unifying Claim

Not three diseases but one homeostatic system failing in three successive cell populations — joined by one shared substrate (the PV interneuron / its perineuronal net), one shared upstream signal (TGF-β/SMAD), and one shared molecular pivot (TREM2). Resilience therefore requires joint preservation of all three layers; the MCI→dementia transition behaves as a threshold, not a slope.

🧬
Groundwork
The substrate beneath every phase
The substrate layers that underlie and condition all three phases rather than sitting in one — the inherited and cellular groundwork on which the temporal sequence plays out.
Genetics (GWAS / APOE) Mitochondria Vasculature / BBB Lipids / MAM Senescence

What it holds

Mitochondria, genetics, vasculature, lipids/MAM, the cellular census, senescence, epigenetic erosion, the protective genome, Down syndrome, BDNF, exercise, and the glial-consortium and Khachaturian-calcium first-principles papers.

How it conditions the phases

Inherited risk architecture is the "loaded dice" set before any pathology (APOE, the LOAD GWAS loci, chr21 dosage). Baseline metabolic vulnerability sets how early the locus coeruleus decompensates; vascular risk and lipid/MAM biology and cellular senescence raise or lower the threshold at every station. None of these is a phase — each is a layer the phases run on.
🔀
Convergence
Autophagic collapse, and the upstream triggers that feed it
The shared downstream funnel — autophagy-lysosomal / autolysosomal-acidification failure and PANTHOS — together with the upstream etiologies that feed it.
Autophagy–lysosomal failure PANTHOS Viral / glymphatic triggers Oxidative stress

The shared funnel

Autophagy-lysosomal and autolysosomal-acidification failure is the compartment where independent pathways jam — the de-acidified lysosome, the PANTHOS neuron, the "inside-out" plaque. It recurs at Phase I (endosomal/retromer) and again at Phase III (lysosomal).

Upstream triggers feeding it

The viral trajectory, glymphatic and clearance collapse, oxidative stress, and the vascular contributions — the multiple entry lanes that load the same downstream failure. Note this is one high-value final common bottleneck, not the sole cause.
🗺️
Synthesis
The cross-axis architecture
The whole-architecture, cross-axis integrations that read the phases, bridges, and layers as one falsifiable structure.
Spectrum of Collapse Unified Architecture (graded) HMS model Amyloid steelman

What it holds

Spectrum of Collapse, the Biomarker Cascade, the Unified Architecture (graded), the HMS model, Terminal Collapse, the Depression Continuum, the Ketamine Paradox, and the amyloid steelman.

Evidentiary honesty

These integrations grade every claim: Tier I (established), Tier II (bridging inference), Tier III (conjectural), consolidated in a Validity Ledger. Amyloid is conceded as a validated, causally-active node and the initiating trigger in autosomal-dominant AD — but "not sole author of the outcome." Anti-amyloid antibodies are reframed as right target, wrong phase.

The Seven Convergence Nodes

Compartments where independent pathways jam. Click any node to see what converges and the phases it spans. The color dots mark which phases the node reaches.

1 · Endosomal Nexus
The early-endosome / endo-lysosomal-autophagy network as the compartment where independent pathways jam. Spans the Phase-I quality-control lesion and its Phase-III recurrence in cortical neurons.
What converges: lipid-peroxidation disruption of receptor trafficking (ApoE–ApoER2–Dab1); retromer dysfunction / "endosomal traffic jams" (SORL1, VPS26b, VPS35); intraneuronal Aβ42 in MVBs / synaptic endosomes years before plaques (the "inside-out" reading); autophagy-lysosomal failure → de-acidified lysosome → PANTHOS neuron (APP-βCTF → v-ATPase inhibition); APOE4 compounds every arm.

Key molecules: ApoER2, Dab1, SORL1, VPS26b/35, BACE1, APP-βCTF, v-ATPase, Rab5, PSEN1, cathepsins.

Phases spanned: Phase I · Phase III · Groundwork.
2 · Cytoskeletal Collapse
The actin + microtubule cytoskeleton attacked from multiple directions → dendritic-spine loss and synaptic failure. The executional endpoint of Phase III.
What converges: LIMK1 hyperactivation + cofilin inactivation (the GSK3β cascade); Ephexin5–RhoA–ROCK actin-myosin contraction; competitive synaptic-plasticity dysregulation; complement-mediated spine destruction.

Key molecules: GSK3β, LIMK1, cofilin, Ephexin5, RhoA, ROCK, PSD95.

Phases spanned: Phase II · Phase III.
3 · Compensatory Paradigm Nexus
Amyloid + tau reinterpreted as initially protective responses that become pathologically reinforced. This node itself encodes the cause-vs-consequence tension.
What converges: Aβ as an antioxidant response to lipid peroxidation; Aβ/tau as compensation for excitatory insufficiency / NMDA hypofunction; lipid-raft restructuring as allostatic adaptation; Aβ-monomer depletion driving compensatory overproduction.

The tension: the corpus concedes amyloid is a validated, causally-active node and the initiating trigger in autosomal-dominant AD — but "not sole author of the outcome" (COLBOS/Christchurch carriers resisted despite amyloid).

Phases spanned: Phase I · Phase III · Convergence.
4 · Neuroimmune Interface
The microglial-complement pruning cascade that accelerates synaptic loss — the executional arm carried across the Proteolytic Turn.
What converges: C4d deposition on synapses → LilrB2 signaling; TREM2-mediated microglial phagocytosis of synapses; pattern-recognition of DAMPs (oxidized proteins, Aβ, pTau); trans-synaptic immune signaling → cytoskeletal collapse.

Key molecules: C1q, C3, C4d, LilrB2, TREM2, CR3.

Phases spanned: Phase II · Phase III.
5 · APOE4 Hub
A single polymorphism activating 8+ pathogenic mechanisms, both loss- and gain-of-function; it feeds nearly every other node and reaches every phase.
What converges: loss of the disulfide bridge → unprotected PUFA cargo → lipid peroxidation; impaired retromer/endosomal trafficking; reduced excitatory tone; altered lipid-raft organization + complement activation; Ephexin5–RhoA activation → spine collapse; degradation of reelin signaling at shared ApoER2/VLDLR receptors (apoE vs reelin competition).

Where it meets the metal: the TOMM40 import gate at the organelle (rs10524523 in LD with APOE4).

Phases spanned: Groundwork · Phase I · Phase II · Phase III.
6 · Transcriptional-Epigenetic Dysregulation
Progressive gene-expression changes that lock in the pathological state, making the disease self-sustaining independent of the original triggers.
What converges: CREB-mediated changes from NMDA hypofunction; HDAC6 upregulation + chromatin remodeling; miRNA dysregulation (miR-134, let-7) on spine stability; LINE-1 transposable-element reactivation.

Why it matters: once the programs shift, the disease becomes self-reinforcing — the reversible information layer that the Fading Score cross-section maps.

Phases spanned: Groundwork · Phase II · Convergence.
7 · Metabolic-Homeostatic Axis
The richest node: systemic insulin resistance + mitochondrial decline + brain bioenergetic failure converge on the very signaling that maintains microglial homeostatic identity and neuronal metabolic fitness.
What converges: TGF-β/SMAD failure → loss of the Butovsky homeostatic microglial signature; TREM2 → DAP12 → SYK → PI3K/AKT/mTOR gating microglial metabolic/phagocytic fitness; the mitochondrial cascade setting an age-dependent bioenergetic threshold; the "selfish brain + allostatic load" loop; PV+ interneuron vulnerability (PNN-ensheathed, highest-firing) as the convergence substrate.

Core claim: the attack/failure dichotomy resolves into one upstream event — collapse of the metabolically-maintained homeostatic state. Named levers: ketogenic therapy, GLP-1 agonism, metformin, pioglitazone.

Phases spanned: Phase I · Phase II · Phase III · Groundwork.

The Two Bridges & Cross-Phase Themes

The named mechanisms that carry the disease from one phase to the next, and the shared threads that run across phases. Chips mark which phases each theme touches.

Bridge 1 — The Locus Coeruleus Bridge
Phase I Phase II Withdrawal of the noradrenergic brake (β2-AR→cAMP→PKA→NF-κB) Trans-synaptic tau seeding (LRP1 / HSPG)
Bridge 2 — The Proteolytic Turn
Phase II Phase III NLRP3→IL-1β→MMP-9/ADAMTS switch Iron + Fenton catalysis C1q→C4d complement priming
TGF-β/SMAD homeostatic signal (shared upstream)
Groundwork Phase II Phase III
TREM2 molecular pivot (shared)
Groundwork Phase II
Perineuronal net / PV interneuron (shared substrate)
Phase II Phase III
Autophagy–lysosomal failure / PANTHOS
Phase I Convergence Phase III
NMDA / excitatory insufficiency & the compensatory scar
Phase III Convergence
Endosomal / retromer trafficking failure
Phase I Phase III Groundwork
Iron / ferroptosis / oxidative execution
Phase II Phase III
Inherited risk architecture (the loaded dice)
Groundwork Phase I Phase II

Genes by Phase

Where each well-known AD gene acts along the temporal arc — from the inherited Groundwork through the three phases to the shared Convergence funnel. Hover any cell for the phase-specific mechanism. Click any gene card below for details.

Relevance:
Not implicated
Peripheral
Significant
Major role
Central mechanism
Gene / Protein Groundwork Phase I Phase II Phase III Convergence
APOEApolipoprotein E · 19q13.32 ★ ● ● ● ○
APPAmyloid Precursor Protein · 21q21.3 ● ★ ● ● ●
PSEN1Presenilin 1 · 14q24.2 ★ ● — ● ●
PSEN2Presenilin 2 · 1q42.13 ★ ● — ○ ●
MAPT (Tau)Microtubule-Associated Protein Tau · 17q21.31 ○ ★ ● ● —
TREM2Triggering Receptor on Myeloid Cells 2 · 6p21.1 ★ — ★ ● ●
SORL1Sortilin-Related Receptor 1 · 11q24.1 ● ★ — ● ●
BIN1Bridging Integrator 1 · 2q14.3 ★ ● ○ — ●
ABCA7ATP-Binding Cassette A7 · 19p13.3 ● — ● — ○
CLUClusterin (ApoJ) · 8p21.1 ● — ● ○ —
PICALMPhosphatidylinositol Clathrin Assembly · 11q14.2 ● ● — — ●
CR1Complement Receptor 1 · 1q32.2 ● — ● ● —
CD33Siglec-3 · 19q13.41 ● — ● ○ —
PLCG2Phospholipase C-γ2 · 16q23.3 ● — ★ — ○
TYROBP (DAP12)TYRO Protein Tyrosine Kinase-Binding Protein · 19q13.12 ● — ★ — ●
SPI1 (PU.1)Spleen Focus Forming Virus Integration 1 · 11p11.2 ● — ● — ●
MS4A (4A/6A)Membrane-Spanning 4A cluster · 11q12.2 ● — ● — —
INPP5D (SHIP1)Inositol Polyphosphate-5-Phosphatase D · 2q37.1 ● — ● — —
ABI3ABI Family Member 3 · 17q21.32 ● — ● — —
TOMM40Translocase of Outer Mito Membrane 40 · 19q13.32 ● ★ — ○ —
PARP1Poly(ADP-ribose) Polymerase 1 · 1q42.12 ○ ★ — ● —
NLRP3NLR Family Pyrin Domain 3 · 1q44 — ○ ★ ● —
MMP9Matrix Metalloproteinase 9 · 20q13.12 — — ● ★ —
RELN (Reelin)Reelin · 7q22.1 ○ — ○ ★ ●
ACAN (Aggrecan)Aggrecan · 15q26.1 — — ○ ★ —
GPX4Glutathione Peroxidase 4 · 19p13.3 — ○ ● ★ ○
C1QAComplement C1q A-chain · 1p36.12 — — ● ● —
GSK3βGlycogen Synthase Kinase 3 Beta · 3q13.33 — ● ○ ★ —

Key Gene Profiles

Expression changes, chromosomal location, and the specific role each gene plays across the three phases. Click to expand.

APOE 19q13.32 Lipid Transport
Apolipoprotein E (ε2 / ε3 / ε4 alleles)
The strongest common genetic risk factor for AD. Set in the Groundwork, it then reaches every phase and anchors the APOE4 convergence hub.
Groundwork Phase I Phase II Phase III Convergence

Position in the architecture

↑ Upregulated in reactive astrocytes and microglia around lesions; ApoE4 shows Trapped in endolysosomal compartments. Inherited groundwork risk that conditions how early and how hard each phase fails.

Allele-specific effect

ε4: heterozygote OR ≈ 3–4, homozygote OR ≈ 12–15 (vs ε3/ε3); cannot form the lipid-protecting disulfide bridge. ε2: protective, OR ≈ 0.6 — but raises CAA/hemorrhage risk (a double-edged defense).

Across the phases

Phase I: unprotected PUFA cargo raises the lipid-peroxidation load in the LC. Phase II: a core DAM-signature gene driving the microglial switch. Phase III: arrests ApoER2–Dab1 recycling and competes with reelin. Convergence: feeds nearly every node and meets the TOMM40 import gate.

APP 21q21.3 Transmembrane
Amyloid Precursor Protein
Causal in autosomal-dominant EOAD and triplicated in trisomy 21. Its βCTF/Aβ products act at the organelle across Phase I and feed the Convergence funnel.
Groundwork Phase I Phase II Phase III Convergence

Position in the architecture

↑ Elevated processing under oxidative stress; triplicated on chr21 (the Compressed Architecture). The A673T variant is protective (OR ≈ 0.2), proving the axis runs through APP itself.

Phase I — the import block

C99/Aβ obstructs the TOM40 mitochondrial-import channel, stopping import-based repair; intraneuronal Aβ accumulates in synaptic endosomes years before plaques.

Convergence — PANTHOS

APP-βCTF inhibits the v-ATPase, de-acidifying the lysosome → the PANTHOS neuron whose "gravestone" is the plaque. Read here as a modification and amplifier, not the sole author of the outcome.

MAPT 17q21.31 Microtubule
Microtubule-Associated Protein Tau
Pretangle tau in the locus coeruleus is the Tier-I earliest lesion. Seeding-competent tau then rides the LC bridge into the limbic system.
Phase I Phase II Phase III Groundwork

Phase I — kindled in the LC

↑ Hyperphosphorylated pretangle tau accumulates in LC neurons from early adulthood (Tier I — Established: pretangle tau in LC before cortex). AEP/DOPEGAL cleavage at N368 helps kindle it (the First Ember).

Bridge 1 — the seeding cargo

Release- and seeding-competent tau is internalised via LRP1 and heparan-sulfate-proteoglycan endocytosis, templating host tau synapse-by-synapse along the ascending LC projection.

Phase III — the compensatory scar

Redistributed to the somatodendritic compartment as a maladaptive scar. The H1 haplotype is a weak/inconsistent AD signal (strong in PSP/FTD) — groundwork, not driver.

TREM2 6p21.1 Neuroimmune
Triggering Receptor Expressed on Myeloid Cells 2
The shared molecular pivot of the whole architecture: a rare loss-of-function risk gene that gates the microglial DAM transition.
Groundwork Phase II Phase III Convergence

Groundwork — the risk allele

Rare coding LoF: R47H OR ≈ 2.5–4, R62H OR ≈ 1.5. Ranked among the dominant drivers of microglial dysfunction alongside aging and APOE.

Phase II — the pivot

TREM2 → DAP12 → SYK → PI3K–AKT–mTOR gates microglial metabolic and phagocytic fitness and the DAM transition (APOE, CST7, cathepsins, LPL, SPP1, ITGAX).

Convergence

The single pivot named across the Metabolic-Homeostatic Axis node — where the attack/failure dichotomy resolves into one collapse of the metabolically-maintained homeostatic state.

NLRP3 1q44 Inflammasome
NLR Family Pyrin Domain-Containing 3
The innate-immune sensor that converts the disease. Assembled in Phase II, it fires the proteolytic switch of the Proteolytic Turn.
Phase I Phase II Phase III

Primed early, fired late

↑ Assembled from lipid droplets in the post-homeostatic microglion. Phase-I ROS output primes the effector arm; Phase-II assembly turns the cell.

Bridge 2 — the proteolytic switch

NLRP3 → caspase-1 → mature IL-1β → (NF-κB/AP-1) transcription of MMP-9, MMP-3, ADAMTS-4/5: a cytokine-secreting cell becomes a matrix-digesting one.

Verdict & lever

Graded as an amplifier, not initiator — best-evidenced is not most-important. Levers: MCC950/Inzomelid (NLRP3), canakinumab (anti-IL-1β).

MMP9 20q13.12 Protease
Matrix Metalloproteinase 9 (Gelatinase B)
The Phase-III load-bearing target: the enzyme that digests the perineuronal net and unstages the parvalbumin interneuron.
Phase II Phase III

How it is unleashed

↑ Hyperactivated downstream of IL-1β in the turning microglion, and by norepinephrine via β-AR (the Coerulean Shears). It digests aggrecan/brevican — the three offices of the sulfated surface.

Therapeutic window & knife-edge

Load-bearing class for Phase III: minocycline/doxycycline (broad), SB-3CT and JNJ0966 (selective), AAV-TIMP3 restoration. But MMP-9 also serves LTP, neurogenesis and the BBB — the marimastat/batimastat toxicity lesson demands selectivity and pulsed dosing.

Evidentiary honesty

NE→MMP-9 is periphery-proven; that MMP-9 directly cleaves reelin is not established (Tier II–III).

RELN 7q22.1 ECM / Reelin
Reelin
The resilience axis: a secreted architect that brakes tau via ApoER2/VLDLR→Dab1, keystoned on the RELN-COLBOS resistance case.
Groundwork Phase II Phase III Convergence

The resilience keystone

The RELN-COLBOS variant (Lopera 2023) delayed dementia despite heavy pathology — twinned with APOE3-Christchurch. Both converge on the heparan-sulfate co-receptor (Pan 2025).

Staged in the net, gating tau

Reelin is secreted into the perineuronal net that stages its signal; ApoER2/VLDLR → Dab1 → Fyn → GSK-3β brakes tau. The "rising-reelin paradox": expression ↑ while signal ↓ (reelin resistance).

The therapeutic knife-edge

An HS-blocking anti-tau drug could silence protective reelin signalling at the same sugar — tune the clasp, do not break it († double-edged).

SORL1 11q24.1 Trafficking
Sortilin-Related Receptor 1 (SORLA)
The retromer-associated sorting receptor whose loss is the Phase-I endosomal traffic jam — the quality-control lesion that recurs in cortical neurons at Phase III.
Groundwork Phase I Phase III Convergence

Groundwork — a strong risk gene

↓ Reduced in sporadic AD. Common-variant OR ≈ 1.1, but rare loss-of-function variants are semi-Mendelian (OR ≈ 2–12) — the strongest LOAD locus after APOE.

Phase I — the traffic jam

SORLA directs APP away from the late endosomes where secretases wait; its loss, with retromer subunits VPS35 / VPS26, produces the enlarged-endosome lesion that is AD's earliest cellular hallmark.

Recurrence & node

The same clearance lesion returns in cortical neurons at Phase III, and SORL1 anchors the Endosomal Nexus convergence node. Levers: AAV9-SORL1, retromer chaperones (R55).