CoChat

Synaptic counts approximate synaptic contact area in Drosophila

Open access
Barnes, Christopher L., Bonnéry, Daniel, Cardona, Albert · 2022 · PLOS ONE

Published version of a preprint

The DOI matches its Crossref record; 55 of the 56 references it cites were found in a registry or an index; 13 of the 35 sources its claims cite were checked for support.

Abstract

The pattern of synaptic connections among neurons defines the circuit structure, which constrains the computations that a circuit can perform. The strength of synaptic connections is costly to measure yet important for accurate circuit modeling. Synaptic surface area has been shown to correlate with synaptic strength, yet in the emerging field of connectomics, most studies rely instead on the counts of synaptic contacts between two neurons. Here we quantified the relationship between synaptic count and synaptic area as measured from volume electron microscopy of the larval Drosophila central nervous system. We found that the total synaptic surface area, summed across all synaptic contacts from one presynaptic neuron to a postsynaptic one, can be accurately predicted solely from the number of synaptic contacts, for a variety of neurotransmitters. Our findings support the use of synaptic counts for approximating synaptic strength when modeling neural circuits.

Overview

CoChat Summary

TL;DR

The paper quantifies the relationship between synaptic contact number and total synaptic surface area in the larval *Drosophila* central nervous system using volume electron microscopy. It finds that total synaptic surface area per connection can be accurately predicted from the number of synaptic contacts, supporting the use of synaptic counts as edge weights in circuit models.

Key findings

  • Across 540 synaptic contacts representing 80 graph edges and 4 edge types, a weighted linear regression of total synaptic surface area per edge against contact number gave R² = 0.905, with y = 0.014x + 0.002 (gradient in μm² count⁻¹, intercept in μm²).
  • Independent regressions by edge type also showed similarly high R² values, and Bayesian hierarchical modeling indicated that adding edge-type information did not meaningfully improve confidence in predicting total edge contact area from contact number.
  • Individual synapse areas varied widely: the α = 0.1 confidence interval spanned approximately an order of magnitude within each edge type, and differences in means between edge types had small effect size relative to intra-type variation, even where statistically significant.
  • The authors conclude that synaptic counts can approximate synaptic strength when modeling neural circuits, because prior work shows synaptic area correlates linearly with strength and their data show counts predict area.

Methods

  • Volume EM data came from the whole CNS of a 6-hour-old female L1 *Drosophila melanogaster*: 4,850 sections, each 50 nm thick, imaged at 3.8 × 3.8 nm resolution with an FEI Spirit TEM; neurons and synapses were reconstructed in CATMAID from previously published datasets.
  • Postsynaptic membrane area was measured for 540 contacts in 3D using BigCAT; annotations were skeletonized, smoothed with a Gaussian kernel (σ = 3), and converted to area by multiplying 2D length by xy resolution (3.8 nm) and z resolution (50 nm), then summed across sections per contact and across contacts per edge.
  • Analysis used weighted least-squares linear regression of total synaptic area per edge versus contact count, weighted by inverse contact count, plus a Bayesian hierarchical gamma model in R/JAGS (100,000 iterations, thinned by 2, 1,000-iteration burn-in) to assess edge-type predictive power.

Limitations

  • Stated: full validation would require comprehensive measurement of all synaptic surface areas of all neurons of a central nervous system; whether the correlation holds through changes in neuronal arbor dimensions is an open question; and the study is limited to specific edge types in the *Drosophila* larval CNS at the same life stage.
  • Stated/evident: the data come from one 6-hour-old female L1 larva and four edge types (80 edges, 540 contacts), so generalizability to other developmental stages, species, or brain regions is not directly tested.
  • Evident: synaptic area is approximated from postsynaptic membrane annotations in ssTEM, with manual reconstruction and annotation, and the dataset does not include paired electrophysiological validation of synaptic strength.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1371/journal.pone.0266064

Cites

55 of 55 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig 1 The anatomy and connectivity of the 4 circuits of interest. Presynaptic partners are shown in red, and postsynaptic in blue: synaptic sites are shown in cyan. PN-containing circuits are anterior XY projections. Basin-containing circuits are dorsal XZ projections. The connectivity matrices are have presynaptic partners on the Y axis, and postsynaptic partners on the X axis. Ambiguous pairs of neurons (lch5–2/4 and vchA/B, which are indistinguishable as their cell bodies lie outside the VNC) are distinguished by a truncated reconstruction ID. Squares in the connectivity matrix are coloured by what fraction of the target’s dendritic input, by contact number, is represented by that edge. The absolute number of contacts is also included. In total, there are 4 edge types (the 4 rows above), 80 edges (one each for each non-zero cell across all 4 matrices), and 540 contacts (each being a morphological synapse).
Fig 1 The anatomy and connectivity of the 4 circuits of interest. Presynaptic partners are shown in red, and postsynaptic in blue: synaptic sites are shown in cyan. PN-containing circuits are anterior XY projections. Basin-containing circuits are dorsal XZ projections. The connectivity matrices are have presynaptic partners on the Y axis, and postsynaptic partners on the X axis. Ambiguous pairs of neurons (lch5–2/4 and vchA/B, which are indistinguishable as their cell bodies lie outside the VNC) are distinguished by a truncated reconstruction ID. Squares in the connectivity matrix are coloured by what fraction of the target’s dendritic input, by contact number, is represented by that edge. The absolute number of contacts is also included. In total, there are 4 edge types (the 4 rows above), 80 edges (one each for each non-zero cell across all 4 matrices), and 540 contacts (each being a morphological synapse).CC BY · PubMed Central
Fig 2 Results of synaptic area labelling. Ai) The ssTEM images upon which the analysis is based. Aii) A schematic showing the plane of a neurite sectioned in the top image, and the pre- and post-synaptic sites of that neurite and one of its partners, in orange and red respectively. Note the T-bar, cleft and and postsynaptic membrane specialisation. Bi) For each of the four circuits, the distribution of synaptic areas on a log10 scale. The number of synapses targeting a left-sided neuron are shown in blue; right-sided in orange. Each is overlaid with the best-fitting normal distribution (black dashed line) and 90% confidence interval (green dotted line). Bii) Table of normal distribution parameters, in log10 nm2 to 3 decimal places. Biii) Raw p-values (colouring) and FWER corrected [34] significance levels for pairwise ranksum comparisons of circuit synapse area distributions.
Fig 2 Results of synaptic area labelling. Ai) The ssTEM images upon which the analysis is based. Aii) A schematic showing the plane of a neurite sectioned in the top image, and the pre- and post-synaptic sites of that neurite and one of its partners, in orange and red respectively. Note the T-bar, cleft and and postsynaptic membrane specialisation. Bi) For each of the four circuits, the distribution of synaptic areas on a log10 scale. The number of synapses targeting a left-sided neuron are shown in blue; right-sided in orange. Each is overlaid with the best-fitting normal distribution (black dashed line) and 90% confidence interval (green dotted line). Bii) Table of normal distribution parameters, in log10 nm2 to 3 decimal places. Biii) Raw p-values (colouring) and FWER corrected [34] significance levels for pairwise ranksum comparisons of circuit synapse area distributions.CC BY · PubMed Central
Fig 3 Least-squares linear regressions of contact number vs area for each edge. Weighted by the reciprocal of the contact number to reduce leverage by high-n edges. A) Joint regression line for all edges (black dashed line), y = 0.014x + 0.002, R2 = 0.905. B) For each circuit, a zoomed-in region of A, showing the joint regression (grey dotted line) and the circuit-specific regression line (black dashed line). Left-right pairs, when unambiguous, are shown in the same colour and joined with a dashed line of that colour. †shows outliers beyond the limits of the ORN-PN plot. C) Table of regression line gradient (μm2count−1 to 3 decimal places), y-intercept (μm2 to 3 decimal places) coefficient of determination R2 (to 3 decimal places).
Fig 3 Least-squares linear regressions of contact number vs area for each edge. Weighted by the reciprocal of the contact number to reduce leverage by high-n edges. A) Joint regression line for all edges (black dashed line), y = 0.014x + 0.002, R2 = 0.905. B) For each circuit, a zoomed-in region of A, showing the joint regression (grey dotted line) and the circuit-specific regression line (black dashed line). Left-right pairs, when unambiguous, are shown in the same colour and joined with a dashed line of that colour. †shows outliers beyond the limits of the ORN-PN plot. C) Table of regression line gradient (μm2count−1 to 3 decimal places), y-intercept (μm2 to 3 decimal places) coefficient of determination R2 (to 3 decimal places).CC BY · PubMed Central
Fig 4 Bayesian hierarchical modelling of distributions of synapse areas across graph edges and edge types. A) Model dependency graph, showing how each level parametrises the levels below it, and the number of samples available at each level. B) Results of sampling from the generated model, including the bounds in which 90%, 95%, and 99% of samples fall, demonstrating that our data is in line with the model’s expectation. C) Comparing the results of sampling sub-models with a fixed edge type with the full joint model (red lines): the small narrowing of the result intervals suggests that the edge type does not have predictive power of the relationship between contact number and contact area for a graph edge, and therefore that the sizes of individual synapses do not differ much between edge types.
Fig 4 Bayesian hierarchical modelling of distributions of synapse areas across graph edges and edge types. A) Model dependency graph, showing how each level parametrises the levels below it, and the number of samples available at each level. B) Results of sampling from the generated model, including the bounds in which 90%, 95%, and 99% of samples fall, demonstrating that our data is in line with the model’s expectation. C) Comparing the results of sampling sub-models with a fixed edge type with the full joint model (red lines): the small narrowing of the result intervals suggests that the edge type does not have predictive power of the relationship between contact number and contact area for a graph edge, and therefore that the sizes of individual synapses do not differ much between edge types.CC BY · PubMed Central

Verification

Claims the paper makes

6 Valid · 2 Not valid · 13 Not checked · 27 Own results and interpretations

25 more claims

Create a free account to see all 48 claims and to verify the 22 references still open.

See all claims — free account
  • Probably not a substantial claim. 9
  • It may state two separate claims as one. 10

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Show all 55 references

Research works printed but not linked: 4

  1. Thompson D. On Growth and Form. 1942nd ed. Cambridge Univ. Press; 1942.
  2. Fielding RT. type [; 2000]Available from: https://www.ics.uci.edu/{~}fielding/pubs/dissertation/fielding{_}dissertation.pdfhttp://citeseer.ist.psu.edu/viewdoc/summary?doi=10.1.1.91.2433http://www.ics.uci.edu/{~}fielding/pubs/dissertation/top.htm.
  3. Plummer M. JAGS: A Program for Analysis of Bayesian Graphical Models Using Gibbs Sampling JAGS: Just Another Gibbs Sampler. International Workshop on Distributed Statistical Computing. 2003; 3(Dsc): 1– 10.
  4. Oliphant TE. A guide to NumPy. vol. 1. Trelgol Publishing USA; 2006.

Cited by

See all 34

Metadata

Authors
Barnes, C. L., Bonnéry, D., & Cardona, A.
Affiliations
  • University of Cambridge (GB)
  • MRC Laboratory of Molecular Biology (GB)
Corresponding author
Cardona, Albert
Venue
PLOS ONE
Published
2022-04-04
Volume
17
Issue
4
Pages
e0266064
ISSN
1932-6203 (ISSN-L)
OpenAlex
W4226069364
Cited by
29
References
56
Citations by year
2026: 3 · 2025: 5 · 2024: 10 · 2023: 7 · 2022: 1
FWCI
2.54
Type
journal article
Language
en
Open access
Locations
  • PLoS ONE (publishedVersion, cc-by)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Apollo (University of Cambridge) (submittedVersion, cc-by)
  • Apollo
  • Apollo (cc-by)
Funders
Howard Hughes Medical Institute
Topics
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Neurobiology and Insect Physiology Research
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Neuroscience and Neuropharmacology Research
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neural dynamics and brain function
Keywords
synaptic efficacy, connectomics, neurotransmitters, presynaptic neurons, postsynaptic neurons, neural circuit modeling, volume electron microscopy
MeSH terms
Animals, Drosophila, Larva, Neurons, Synapses, Neurotransmitter Agents, Connectome

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-07, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-07 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (qwen3.8-omni-flash and gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.