Editing Iowa Mutation Modeling Program (section)

== Iowa Mutation Modeling Program ==
'''Mutation:''' Aβ Asp23Asn  
'''Objective:''' Determine whether mutation-specific changes in aggregation kinetics are sufficient to explain altered amyloid deposition patterns and clinical onset timing.

----

=== 1. Biological Background ===

==== 1.1 Genetic and Molecular Context ====
* Description of the Asp23Asn substitution
* Location within Aβ sequence
* Known biochemical effects (e.g., altered charge, isomerization propensity)

==== 1.2 Reported Phenotypic Features ====
* Age of onset distribution
* Neuropathological characteristics
* Reported fibril properties
* Relevant clinical observations

----

=== 2. Working Mechanistic Hypothesis ===

=== 2. Working Mechanistic Hypothesis ===

==== 2.1 Core claim ====
The Iowa mutation (Aβ D23N) increases the ''in vivo'' burden of neurotoxic soluble oligomers by driving the formation of unusually stable, rapidly forming fibrils that deposit predominantly in vascular-associated, diffuse fibril “clouds” (CAA / perivascular deposits), thereby increasing catalytic fibril surface area available for oligomer generation.

==== 2.2 Empirical anchors from Tomidokoro et al. (2010) ====
The working hypothesis is grounded in the following observations/arguments:

* A substantial fraction of deposited Aβ in Iowa brain tissue is post-translationally modified at residue 23: approximately 10–30% contains isoAsp23. <ref name="Tomidokoro2010_isoAsp">Tomidokoro Y. ''et al.'' (2010). Iowa Variant of Familial Alzheimer’s Disease: Accumulation of Posttranslationally Modified AβD23N in Parenchymal and Cerebrovascular Amyloid Deposits. (See discussion text reporting ∼10–30% isoAsp23.)</ref>
* D23N may act as a “fast track” to isoAsp formation because Asn deamidation typically produces isoAsp much faster than Asp isomerization (order-of-magnitude statement in the discussion). <ref name="Tomidokoro2010_fasttrack">Tomidokoro Y. ''et al.'' (2010). Discussion: Asn deamidation typically yields ∼30× faster isoAsp formation than Asp isomerization.</ref>
* isoAsp-bearing molecules exhibit enhanced ''in vitro'' fibrillization kinetics relative to wild-type, and D23N strongly accelerates fibrillization (shortened lag phase; much faster ThT kinetics). <ref name="Tomidokoro2010_ThT">Tomidokoro Y. ''et al.'' (2010). Discussion + Thioflavin-T kinetics section/figure: D23N drives dramatic acceleration; isoAsp contributes modestly but measurably.</ref>
* isoAsp at other Aβ sites (e.g., positions 1 and 7) is associated with decreased proteolytic sensitivity, suggesting isoAsp23 may also contribute to impaired clearance once aggregated. <ref name="Tomidokoro2010_clearance">Tomidokoro Y. ''et al.'' (2010). Discussion: isoAsp at positions 1 and 7 decreases proteolytic sensitivity; potential relevance to Iowa aggressiveness.</ref>

==== 2.3 Structural/mechanistic support from Warmack et al. (2019) ====
Warmack et al. provide atomic-resolution structural evidence that isoAsp23 alters protofilament interfaces in a way that can plausibly increase both aggregation rate and fibril stability, and explicitly discuss that the Iowa mutation may operate similarly and/or via faster production of isoAsp23. They also report that PCMT1 does not effectively repair isoAsp23 once in aggregated forms. <ref name="Warmack2019">Warmack RA. ''et al.'' (2019). Structure of amyloid-β (20–34) with Alzheimer’s-associated isomerization at Asp23 reveals a distinct protofilament interface. ''Nature Communications''.</ref>

==== 2.4 Quantitative hypothesis: “surface-area amplification” ====
We translate the above into a quantitative causal chain:

(1) D23N increases the rate of isoAsp23 formation and/or stabilizes the fibril state.  
(2) Faster fibrillization and higher stability favor vascular/perivascular diffuse fibril deposition (CAA-associated “clouds”).  
(3) Diffuse fibrils yield higher total exposed fibril surface area per deposited mass than compact neuritic plaques.  
(4) Fibril surfaces catalyze oligomer generation (secondary nucleation / surface-catalyzed pathways).  
(5) If clearance is impaired (proteolysis resistance; limited repair; immune sink saturation), soluble oligomer concentration increases sufficiently to drive toxicity.

We treat steps (2)–(5) as a model to be constrained by data.

----

==== 2.5 Minimal “brain-sized” production–clearance model (first-pass) ====

Let:
* <math>M(t)</math> = local monomer (or soluble Aβ) concentration (mol·L<sup>−1</sup>)
* <math>F(t)</math> = fibril mass concentration (or total fibril mass in a compartment)
* <math>A(t)</math> = total exposed fibril surface area (m<sup>2</sup>) in the relevant compartment (e.g., vascular wall + perivascular region)
* <math>O(t)</math> = soluble oligomer concentration (mol·L<sup>−1</sup>)
* <math>f_{\mathrm{iso}}(t)</math> = fraction of deposited molecules containing isoAsp23

'''IsoAsp formation (phenomenological):'''
<math>
\frac{df_{\mathrm{iso}}}{dt} = k_{\mathrm{iso}}(1-f_{\mathrm{iso}}) - k_{\mathrm{repair}} f_{\mathrm{iso}}
</math>

'''Aggregation / fibril growth (minimal):'''
<math>
\frac{dF}{dt} = k_n M^n + k_2 M^m A - k_{\mathrm{clear},F} F
</math>

Here <math>k_2</math> captures surface-catalyzed addition / secondary nucleation terms; <math>k_{\mathrm{clear},F}</math> is effective clearance of fibrillar material.

'''Surface area mapping (“morphology parameter”):'''
<math>
A(t) = \alpha_{\mathrm{morph}} \, F(t)
</math>

where <math>\alpha_{\mathrm{morph}}</math> has units of (surface area)/(fibril mass) and encodes the difference between compact plaques vs diffuse CAA fibril clouds. The key claim is:
<math>
\alpha_{\mathrm{CAA}} \gg \alpha_{\mathrm{plaque}}
</math>

'''Oligomer production on fibril surfaces:'''
<math>
\frac{dO}{dt} = k_{\mathrm{cat}} M^p A - k_{\mathrm{clear},O} O
</math>

At quasi–steady state (fast oligomer dynamics relative to years-scale deposition), this gives:
<math>
O^{*} \approx \frac{k_{\mathrm{cat}}}{k_{\mathrm{clear},O}} M^p A
</math>

Thus the “surface-area amplification” prediction is:
<math>
\frac{O^{*}_{\mathrm{CAA}}}{O^{*}_{\mathrm{plaque}}} \approx
\frac{\alpha_{\mathrm{CAA}}}{\alpha_{\mathrm{plaque}}}
\cdot
\frac{F_{\mathrm{CAA}}}{F_{\mathrm{plaque}}}
\cdot
\left(\frac{M_{\mathrm{CAA}}}{M_{\mathrm{plaque}}}\right)^p
\end{math}

'''Toxicity / neuronal loss (minimal hazard model):'''
Let <math>N(t)</math> denote viable neurons (or synapses) in a region. A minimal coupling is:
<math>
\frac{dN}{dt} = -k_{\mathrm{tox}} \, O^q \, N
</math>

which implies an exponential survival curve with an oligomer-dependent hazard.

----

==== 2.6 What we would try to estimate (order-of-magnitude targets) ====
To turn the above into numbers, we need:

* <math>\alpha_{\mathrm{morph}}</math>: surface area per unit fibril mass for (i) diffuse CAA-associated fibrils vs (ii) compact plaques.
* <math>k_{\mathrm{cat}}</math>: oligomer generation flux per unit surface area (or inferred from ''in vitro'' secondary nucleation fits).
* <math>k_{\mathrm{clear},O}</math>: effective oligomer clearance (proteolysis + transport + immune uptake).
* Compartments/volumes: vascular wall / perivascular space volume to convert oligomer counts to concentration.
* A toxicity mapping <math>k_{\mathrm{tox}}, q</math> that can be anchored to synapse loss or neuronal death observables.

----

==== 2.7 Immediate, testable predictions ====
Even with bold assumptions, the model yields falsifiable consequences:

* If CAA morphology implies <math>\alpha_{\mathrm{CAA}} \gg \alpha_{\mathrm{plaque}}</math>, oligomer load should scale with vascular/perivascular amyloid burden more strongly than with parenchymal plaque load.
* Regions with high vascular amyloid should exhibit disproportionately high markers of synaptic injury relative to plaque-centric AD, after controlling for total deposited mass.
* Interventions that reduce fibril surface availability (without necessarily reducing total mass) should reduce oligomer-associated toxicity signals more than expected from plaque mass reduction alone.

(These predictions should be linked later to specific observables: PET patterns, CSF markers, region-specific atrophy, vascular dysfunction readouts.)

----

=== 3. Mathematical Formalization ===

==== 3.1 Aggregation Kinetics Model ====
* Governing equations (e.g., nucleation–elongation framework)
* Definition of state variables
* Rate constants
* Conservation constraints

==== 3.2 Mutation-Specific Modifications ====
* Which parameters differ from wild-type?
* Justification for parameter changes
* Sensitivity to parameter variation

==== 3.3 Assumptions and Simplifications ====
* Spatial homogeneity vs heterogeneity
* Deterministic vs stochastic treatment
* Boundary conditions

----

=== 4. Parameterization ===

Structured parameter table:

{| class="wikitable"
! Parameter !! Symbol !! Value !! Source !! Experimental Context
|-
| Primary nucleation rate || k_n || ... || ... || in vitro aggregation assay
|-
| Elongation rate || k_+ || ... || ... || ...
|-
| Secondary nucleation rate || k_2 || ... || ... || ...
|}

Include:
* Units
* Measurement uncertainty
* Scaling assumptions (in vitro → in vivo)

----

=== 5. Scaling Strategy ===

How molecular kinetics propagate upward.

==== 5.1 From Aggregates to Plaque Load ====
* Mapping monomer/polymer concentration to plaque mass
* Spatial assumptions

==== 5.2 From Plaque Load to Imaging Signal ====
* Relationship to PET SUVR
* Assumed conversion functions

==== 5.3 From Plaque Accumulation to Clinical Onset ====
* Threshold hypotheses
* Network vulnerability assumptions

Explicitly state which links are empirical vs theoretical.

----

=== 6. Quantitative Predictions ===

Clearly enumerated, falsifiable predictions:

* Predicted shift in age of detectable amyloid PET signal
* Predicted plaque growth rate differences vs wild-type
* Predicted regional deposition differences
* Sensitivity of onset timing to kinetic parameters

Avoid vague language. Use numerical ranges where possible.

----

=== 7. Comparison with Existing Data ===

* PET studies in Iowa mutation carriers
* Reported onset ages
* Neuropathological findings

Explicitly state:
* Agreement
* Discrepancies
* Unexplained features

----

=== 8. Sensitivity and Identifiability Analysis ===

* Which parameters dominate model behavior?
* Are multiple parameter combinations indistinguishable?
* Which measurements would most constrain uncertainty?

This section signals modeling seriousness.

----

=== 9. Open Problems ===

* Missing kinetic measurements
* In vivo vs in vitro scaling uncertainties
* Spatial heterogeneity
* Interaction with tau pathology

----

=== 10. Next Steps ===

* Required data
* Computational refinement
* Proposed experimental collaborations