Skip to content

Amanda Urke — SynapseSeq: sequencing-based synaptic connectomics

Speaker: Amanda Urke — MD/PhD student, Macosko Lab, Broad Institute / Harvard. Senior author: Evan Z. Macosko. Part of the Brain Connects Consortium. Slot: SCGD26, Jun 12, 2026 (12:20–12:40 PM) Talk title: "Protein-guided RNA barcoding links transcriptomes to synaptic architecture" Method: SynapseSeq / Synapse-seq — AAV-barcoded synaptic neuroanatomy linked to single-cell gene expression Recording: SCGD26 livestream — preprint: bioRxiv 10.64898/2026.02.26.705527 (posted Feb 27, 2026) Status: Captured from session transcript

Thesis

Pair synaptic neuroanatomy with the transcriptomically defined neuron it belongs to. Existing viral tracers (Sindbis/rabies — e.g. MAPseq/BARseq lineage) label connectivity but lose endogenous gene expression and aren't the field's standard for stable expression. AAVs (low immunogenicity, cell-type targeting, BBB-crossing engineerable) preserve endogenous expression — so SynapseSeq builds barcoded connectomics on AAV.

How it works

Two components co-delivered by AAV:

  1. Targeting protein — an interchangeable subcellular-localization domain fused to GFP + the RNA-binding protein PCP.
  2. Barcode transgene — mScarlet + a UTR with PP7 stem-loops, a unique 32-bp barcode, and poly(A).

PCP binds the PP7 stem-loops, so the barcoded mRNA is trafficked wherever the targeting protein goes: - Presynaptic: synaptophysin → axonal terminals. - Postsynaptic: a nanobody to endogenous PSD-95 → excitatory postsynapses.

Read-out: single-nucleus / spatial transcriptomics at the soma for source-cell identity; Slide-seq (10 µm beads) or bulk RNA-seq at projection/dendrite sites for the trafficked barcodes; then computationally match barcodes back to source cells. Removing the stem-loops is the negative control (barcode no longer rides along).

Key results

  • Presynaptic validation (mouse V1): ~80% of barcode colocalizes with GFP-synaptophysin at the thalamus (DLGN); ~7× barcode-density drop in the stem-loop-deleted control; specificity stable over weeks of incubation (relevant for primate timelines).
  • Brain-wide projections: ~900 layer-specific cells mapped to targets; reproduced known motifs (layer-5 ET widespread projections; layer-5 NP lacking long-range). Slide-seq resolution teased apart two layer-6 CT subtypes with different DLGN-vs-LP connectivity. Anterior-cortex injection revealed medial/lateral layer-5 ET topography and collaterals (one neuron → medulla + caudoputamen), plus a cadherin-defined layer-5 ET subtype lacking the striatal collateral.
  • Cross-species dopamine (marmoset): 10×10 mm barcoded bead arrays; >14,000 SynapseSeq barcodes across ~9,000 beads; highest density in ipsilateral caudate-putamen / nucleus accumbens (nigrostriatal) + mesocortical motor-cortex signal.
  • Postsynaptic (PSD-95) in hippocampus: Slide-seq on serial sections reconstructed 3D dendritic trees; ~250,000 matches / ~18,000 barcodes, ~8,000 pyramidal + ~2,000 dentate-granule neurons. Recovered stereotyped geometries (CA1 proximal-apical density; DG unidirectional) and a coordinated gene-expression × morphology comparison across CA subfields (CA1 uniquely enriched for postsynaptic GO terms + most pronounced superficial/deep apical-density difference).

Q&A highlights

  • Cell-type transduction/trafficking bias (e.g. CA2 high AAV tropism) is largely normalized by dividing barcode read-out by per-cell expression.
  • Main current challenge: scaling to large primate/human brains on much larger spatial arrays — riding the falling cost/rising scale of sequencing so any lab can run it without specialized equipment.

Connections to the corpus

  • Sequencing-based connectomics / barcode recording — sibling to GEMINI (in situ recording) and the MAPseq/BARseq lineage; spatial read-out overlaps AACR single-cell-spatial-omics.
  • Macosko-lab / Slide-seq spatial substrate connects to the spatial-transcriptomics thread in single-cell-genomics-2026/.

Sources

  • SCGD26 schedule (speaker, affiliation, talk title): https://satijalab.org/scgd26/ — listed as Amanda Urke (Harvard University), "Protein-guided RNA barcoding links transcriptomes to synaptic architecture", 12:20–12:40 PM.
  • Preprint: Urke A, Dolan M-J, Silverman J, Kim M, Pineda J, Garcia S, Luu J, Buckley A, Kumar V, Zhao B, Chan K, Nadaf N, Balderrama KS, Arnold DB, Stevens B, Deverman BE, Macosko EZ. "Protein-guided RNA barcoding links transcriptomes to synaptic architecture." bioRxiv, posted Feb 27, 2026. DOI 10.64898/2026.02.26.705527.

Verify / open questions

  • Open-resource links (code/data repos): not yet located — to be released with the preprint; search returned no GitHub/Zenodo link.