Connectotype-Vulnerability Mapping: testing whether mesoscale connectivity motifs predict cell-type-specific transcriptomic stress in the 5xFAD mouse model

arXiv Preprint NeurIPS Nature Methods PLOS ONE

---

🌐 Open in Overleaf →

Proposed Allen Showcase Experiment

Working title

Connectotype-Vulnerability Mapping: testing whether mesoscale connectivity motifs predict cell-type-specific transcriptomic stress in the 5xFAD mouse model

Core question

Do the wiring motifs captured in the Allen Mouse Brain Connectivity Atlas predict which cell types will show the earliest transcriptomic stress signatures in a neurodegenerative mouse model? Can we pre-identify vulnerable circuits from anatomy alone?

Why this fits Hongkui Zeng and Brain Science at the Allen Institute

• The experiment directly leverages the Allen Mouse Brain Connectivity Atlas — the mesoscale projection matrix that Zeng's group built (Oh et al., Nature 2014, PMID: 25470075). Every connection in the matrix is a testable hypothesis about which circuits carry vulnerability.
• It integrates the Mouse Whole Brain Cell Type Atlas (Yao et al., Nature 2023, PMID: 37656950), which provides the transcriptomic cell-type taxonomy that the experiment will cross-reference against connectivity-derived predictions.
• It uses Allen Cre driver lines — the transgenic tools Zeng's lab developed for targeting specific cell populations — to sample predicted-vulnerable and predicted-resilient circuits in parallel.
• The Common Coordinate Framework (CCFv3; Wang et al., Cell 2020, PMID: 32386544) anchors all measurements to a shared anatomical reference.

Proposed biological system

Mouse model

5xFAD (C57BL/6J background) — chosen because:

• Amyloid pathology begins at ~2 months, allowing a 3-month pilot to capture pre-plaque and early-plaque transcriptomic stress.
• Single-cell RNA-seq studies (e.g., Mathys et al., Nature 2019, PMID: 31227861; and subsequent 5xFAD-specific scRNA-seq) show that excitatory neuron stress and microglial DAM activation emerge before overt plaque deposition.
• The model is widely available, well-characterized, and compatible with Allen Cre lines on the C57BL/6J background.

Connectivity-based hypothesis

The Allen Connectivity Atlas provides a normalized projection-strength matrix between ~700 CCF-parcellated brain regions. We hypothesize that:

Regions receiving high convergent input (many strong projections converging) will show earlier transcriptomic stress than regions receiving divergent or sparse input.

Cell types within those high-convergence zones that are postsynaptic to the strongest-weighted projections will be the first to show stress signatures.

This relationship holds even after controlling for distance from the nearest amyloid plaque.

Test design

Select 6 circuits from the connectivity matrix:

• 3 predicted-vulnerable: circuits with high convergent-input scores (e.g., retrosplenial cortex receiving hippocampal + thalamic input; subiculum receiving CA1 + entorhinal input; anterior cingulate receiving hippocampal + amygdalar input).
• 3 predicted-resilient: circuits with low convergent-input scores matched for region size and cell-type composition (e.g., primary visual cortex, primary motor cortex, somatosensory barrel cortex).

Cre lines for cell-type targeting

Use Allen Cre lines to isolate specific cell types at each circuit:

• Rbp4-Cre: Layer 5 IT/subcerebral projection neurons
• Tlx3-Cre: Cortical excitatory neurons (layers 2-5)
• Pvalb-IRES-Cre: Parvalbumin+ fast-spiking interneurons
• Sst-IRES-Cre: Somatostatin+ interneurons
• Sim1-Cre: Thalamic projection neurons (for thalamocortical afferents)

Experimental approach

Derive connectivity priors: From the Allen projection matrix, compute convergent-input scores for each CCF region. Rank regions. Select top-3 vulnerable and matched bottom-3 resilient.

Collect tissue: At 2, 3, and 4 months of age, sacrifice 5xFAD mice and wild-type littermates (n=4 per genotype per timepoint). Microdissect the 6 target regions using CCF-aligned coordinates.

Cell-type-specific snRNA-seq: Use Cre-dependent nuclear tagging (e.g., Cre-dependent Sun1-sfGFP for INTACT) to isolate nuclei from target cell types, then perform snRNA-seq on sorted populations.

Compute transcriptomic stress: Quantify early stress signatures — ISR/UPR activation, synaptic gene downregulation, inflammatory milieu — per cell type per region.

Test prediction: Correlate convergent-input score with transcriptomic stress magnitude across the 6 circuits. The primary endpoint is a significant positive correlation (Spearman rho > 0.5, p < 0.05) between connectivity-predicted vulnerability and observed stress.

Readouts

• Per-cell-type DEG counts and stress-pathway scores (GSEA for ISR, UPR, oxidative stress)
• Plaque distance (from immunohistochemistry on adjacent sections)
• Cell-type composition shifts per region
• Connectivity-stress correlation across the 6 circuits

Success criterion

This is a success if the Allen connectivity matrix, used without any pathological information, produces ranked predictions of cell-type vulnerability that significantly correlate with measured transcriptomic stress in 5xFAD mice. This would establish that anatomy contains predictive information about neurodegenerative vulnerability — a result that would directly validate the utility of the Connectivity Atlas for disease research.

Cost estimate (3-month pilot)

• Mice: 5xFAD × 3 timepoints × 4 per genotype × 2 genotypes = 24 mice, plus Cre-line crosses ≈ 72 mice total (~$3,600)
• snRNA-seq: 6 regions × 5 cell types × 3 timepoints × 2 genotypes = 180 libraries × ~$500/lib = ~$90,000
• IHC and imaging: ~$5,000
• Bioinformatics: ~$10,000
• Total: ~$110K

IIG estimate

If connectivity predicts vulnerability: IIG = 0.7 (high impact — opens a new avenue for using the Connectivity Atlas in disease research; directly connects two major Allen resources; publishable in a high-impact journal).

Hongkui Zeng Review

Initial reaction

Greenlight: yes, with important refinements to make this maximally leverage our infrastructure.

This proposal asks exactly the right kind of question for the Allen Institute: it takes two major community resources we built — the Connectivity Atlas and the Cell Type Atlas — and tests whether they have predictive power for disease biology. That is what these resources are for: not just description, but prediction and mechanism.

What I like

• The core hypothesis is strong and testable: connectivity motifs carry information about selective vulnerability. This is what Seeley's group showed at the macroscale for human FTD/AD (Seeley et al., Neuron 2009), but nobody has tested it at cell-type resolution using a systematic connectivity matrix.
• The choice of 5xFAD is pragmatic: early onset, fast enough for a 3-month pilot, and extensive existing scRNA-seq data to benchmark against.
• The proposal uses our Cre lines exactly as they were designed — to access specific cell populations in defined anatomical locations.
• Using the CCF as the common coordinate system for alignment between connectivity data, dissection, and sequencing is correct and is how we do things here.

What needs tightening

1. Define the connectivity prior more precisely

"Convergent input" is a reasonable starting metric, but the Allen projection matrix contains richer information. I would compute three connectivity features and test each:

• Convergent input strength: sum of normalized projection densities into a region (your proposed metric).
• Input diversity: Shannon diversity of the input distribution — a region receiving equally strong input from many sources may differ from one receiving very strong input from few.
• Reciprocity: whether a region has strong reciprocal connections with its inputs, which could indicate feedback regulation that buffers stress.

This enriches the analysis and tests whether the relationship is driven by a specific aspect of connectivity architecture.

2. Narrow the Cre-line panel for the pilot

Five Cre lines × 6 regions × 3 timepoints × 2 genotypes = 180 libraries is a lot for a pilot. For a 3-month proof of concept, I would start with two cell types:

• Tlx3-Cre (excitatory glutamatergic neurons — the primary vulnerable population in early AD)
• Pvalb-IRES-Cre (fast-spiking interneurons — implicated in network dysfunction)

That cuts libraries to 72, which is feasible and still tests whether the connectivity-stress relationship is cell-type-specific.

3. Use INTACT for nuclear isolation

The Cre-dependent Sun1-sfGFP (INTACT) approach is the right method. But specify the crossing strategy explicitly: use the B6;129S-Intact2 line (Mo et al., Cell Reports 2024) crossed with Cre lines to achieve cell-type-specific nuclear tagging. This is a well-validated pipeline in our lab.

4. Add MERFISH validation

After the snRNA-seq screen identifies predicted-vulnerable vs. predicted-resilient cell types, a targeted MERFISH panel in the 6 regions would validate that the stress signatures are spatially coherent and not artifacts of dissociation. The Allen Institute has the Vizgen MERSCOPE platform and standardized MERFISH panels aligned to CCF.

5. Be explicit about the plaque-distance control

The hardest control is distinguishing "connectivity predicts vulnerability" from "vulnerable regions simply have more plaques nearby." You need to:

• Quantify plaque load in each of the 6 regions at each timepoint by IHC (anti-Aβ 6E10).
• Include plaque density as a covariate in the connectivity-stress regression.
• If the correlation holds after controlling for plaque density, that is the strongest result.

6. Consider the 4-month timepoint carefully

At 4 months, 5xFAD mice have substantial amyloid. The interesting biological signal is at 2-3 months — the pre-plaque and early-plaque window where connectivity-predicted stress would be most informative. I would allocate more mice to 2 and 3 months.

My requested changes before finalization

• Compute three connectivity metrics, not just convergent input.
• Reduce to Tlx3-Cre + Pvalb-IRES-Cre for the pilot.
• Specify INTACT2 nuclear tagging.
• Include MERFISH validation on a subset.
• Control for plaque density explicitly in the statistical model.
• Emphasize the 2-3 month pre-symptomatic window.

With those changes, this is exactly the kind of experiment I would want to see proposed: it takes our atlas resources and uses them to answer a disease question that cannot be asked without them.

Karel Svoboda — Neural Dynamics Review

Overall assessment

Greenlight: maybe — the question is important, but I want to flag a dynamics gap that could weaken the result if unaddressed.

The proposal asks whether static anatomical connectivity predicts degenerative vulnerability. That is an important question, and the Allen Connectivity Atlas is the right tool to anchor it. But as someone who thinks about neural dynamics, I want to push on whether anatomy alone is sufficient, or whether you need to know something about activity patterns.

The dynamics concern

Here is the core issue: connectivity is a structural scaffold, but vulnerability may be driven by activity, not anatomy. A circuit with high convergent input that is chronically active may be vulnerable because of metabolic load, calcium stress, or synaptic fatigue — not because of the wiring per se. A circuit with the same convergent input that is mostly silent may be fine.

This means:

If you find a connectivity-stress correlation, it could be confounded by activity differences.

If you don't find it, it could be because connectivity alone doesn't capture the relevant variable (activity).

My suggested mitigation

You don't need to solve this in the pilot, but you should measure basal activity in at least a subset of the predicted-vulnerable vs. predicted-resilient circuits. Two practical options:

• c-Fos immunohistochemistry at baseline (no stimulation) in 5xFAD vs. wild-type at 2 and 3 months. This gives you a coarse but scalable readout of which circuits are hyperactive.
• Calcium imaging in a small cohort using head-mounted miniscopes (e.g., Inscopix) in retrosplenial cortex (predicted vulnerable) vs. primary visual cortex (predicted resilient) during free behavior.

If the connectivity-stress correlation holds after controlling for basal activity, the result is much stronger. If it disappears, you've learned something equally important: that activity, not anatomy, drives vulnerability.

What I like

• The CCF alignment is correct and essential — I agree with Hongkui on this.
• The INTACT approach for nuclear isolation is well-validated and compatible with the Cre lines.
• The proposal to use three connectivity metrics (convergent input, input diversity, reciprocity) is smart because it decomposes what aspect of connectivity matters.
• The plaque-distance control is critical and well-designed.

Practical feasibility notes

• Timeline: 3 months is tight for 72 snRNA-seq libraries + MERFISH. I would suggest 4 months or dropping MERFISH to a planned follow-on.
• Cell-type resolution: The two-Cre-line panel (Tlx3 + Pvalb) is sensible. If you add more lines, do it in a second phase.
• Sample size: n=4 per genotype per timepoint is the minimum for detecting moderate effects. Consider n=5 to guard against attrition in the 5xFAD cohort.

Feasibility score

I would score this 0.55 to 0.6 feasible for a 3-month pilot — achievable but ambitious. The main risk is not the biology but the throughput: snRNA-seq library prep and sequencing for 72 samples in 3 months requires dedicated technician time and core facility scheduling.

Bottom line

The question is worth asking, and the Allen resources make it uniquely possible to ask it. But I want to see basal activity measurements added so that the connectivity-prediction claim can be distinguished from an activity-prediction claim. If you add c-Fos at minimum, I'd move to a full greenlight.