Contents

What feeds the moonwalker descending neuron?

MDN's strongest presynaptic partners, cross-checked across two independent connectomes

Asked at the VFB table, NeuroFly 2026 (Cologne), 7 September 2026.

The question

“What neuron does the moonwalker descending neuron take input from?”

The moonwalker descending neuron (MDN) drives backward walking in Drosophila — a well-known circuit — and the visitor wanted its presynaptic partners.

How we answered it

  1. search_terms for “moonwalker descending neuron” — resolves to the adult class (there’s a separate larval “mooncrawler” homonym; the class label disambiguates them).
  2. run_query (UpstreamClassConnectivity) on the class — this only returns umbrella classes (“adult neuron”, “interneuron”…), not useful biology, because MDN sits under many overlapping ontology parents.
  3. run_query (ListAllAvailableImages) — MDN has 20 registered instances in VFB, across male CNS, MANC, hemibrain, FlyWire and BANC.
  4. run_query (NeuronInputsTo) on one MDN cell from two independent volumes — male CNS v0.9 MDN_R and FlyWire LAL.GNG.12 — to get real per-cell presynaptic partner lists (891 and 359 rows) and see which partners show up in both.

A live cross-dataset query_connectivity call was also tried but timed out (it runs uncached, on demand); the per-cell partner lists above already answer the question.

What we found

The strongest input in both independently reconstructed volumes is CB0095, a GABAergic (inhibitory) neuron in the gnathal ganglion. The next tier is a cholinergic DNpe023 (descending neuron of the posterior brain) and several lateral-accessory-lobe local neurons (LAL162, LAL144). A glutamatergic route comes in via lobula tangential cells (LT51 / VPNl&d1 lineage) — a visual input. In the whole-CNS male CNS volume, a substantial ascending GABAergic input from the VNC (AN06B012, hemilineage 6/NB5-2) also appears — absent from FlyWire because that volume is brain-only.

MDN and its strongest presynaptic partners
Open full page

Worth being honest about: these are single-cell partner lists from one MDN per dataset, not a class-level aggregate across all 20 registered MDN instances — weights differ between volumes partly for reconstruction-completeness reasons, not just biology. FlyWire’s neurotransmitter calls are predictions, not measurements.

Sources