PaperlayerView exact revision on arXiv

Source-linked AI summary

The multilayer connectome of Caenorhabditis elegans

Barry Bentley, Robyn Branicky, Christopher L. Barnes, Edward T. Bullmore, Petra E. Vértes, William R. Schafer

arXiv:1608.08793v1q-bio.NC

TL;DR

Connectomics has focused mainly on synaptic links, although extrasynaptic signaling also matters to brain function. This paper constructs draft monoamine and partial neuropeptide connectomes for C. elegans from gene-expression data, finding distinct network structures that can be represented in a multiplex connectome.

  • Problem

    Connectomics has focused mainly on synaptic connections, while extrasynaptic signaling also contributes to neural function.

    Source: Author summary · p. 3, Introduction · p. 6

  • Method

    The authors construct draft monoamine and partial neuropeptide connectomes for C. elegans using new and published gene-expression data.

    Source: Introduction · p. 6

  • Results

    The extrasynaptic connectomes have topological properties distinct from one another and from the wired connectome.

    Source: Introduction · p. 6

  • Takeaways & Limitations

    The neuronal connectome can be modeled as a multiplex network with structurally distinct synaptic, gap-junction, and neuromodulatory layers.

    Source: Introduction · p. 6

Abstract

from arXiv · show

Connectomics has focused primarily on the mapping of synaptic links in the brain; yet it is well established that extrasynaptic volume transmission, especially via monoamines and neuropeptides, is also critical to brain function. Here we present a draft monoamine connectome, along with a partial neuropeptide connectome, for the nematode C. elegans, based on new and published expression data for biosynthetic genes and receptors. Thus, the neuronal connectome can be represented as a multiplex network, with synaptic, gap junction, and neuromodulatory layers representing alternative modes of interneuronal interaction and with distinct network structures. In particular, the monoamine network exhibits novel topological properties, with a highly disassortative star-like structure and a rich-club of interconnected broadcasting hubs. Despite the low degree of overlap between layers, we find highly significant modes of interaction, pinpointing network locations and multilink motifs where aminergic and neuropeptide signalling modulate synaptic activity. The multilayer connectome of C. elegans represents a clear exemplar of a biological multiplex network and provides a prototype for understanding how extrasynaptic signalling can be integrated into the wired circuitry in larger brains.

Author summary · Introduction

The study adds draft monoamine and partial neuropeptide maps to C. elegans connectomics, showing that its neuronal connectome comprises structurally distinct interaction layers with communication points between them.

Author summary

  • The summary presents the monoamine and partial neuropeptide maps as wireless networks whose structural properties and interactions with wired connectomes can inform study of larger nervous systems.Source: Author summary · p. 3

Introduction

  • Neuromodulators can alter neural activity and behaviour through local or long-range communication between neurons unconnected by synapses, so full connectivity requires mapping extrasynaptic pathways.Source: Introduction · p. 5
  • Many aminergic-receptor neurons are not postsynaptic to releasing neurons, indicating monoamine signalling occurs outside the wired connectome and must be mapped alongside synapses and gap junctions.Source: Introduction · pp. 5–6
  • Neuropeptide signalling is critical to C. elegans nervous-system function and often uses hormonal or other extrasynaptic mechanisms, motivating its partial network map.Source: Introduction · p. 6
  • The study builds a draft extrasynaptic monoamine connectome and partial neuropeptide network for C. elegans from new and published gene-expression data.Source: Introduction · p. 6
  • The neuronal connectome can be modelled as a multiplex of structurally distinct synaptic, gap-junction, and neuromodulatory layers, with critical points for communication between layers.The extrasynaptic connectomes also have topological properties distinct from one another and from the wired connectome.Source: Introduction · p. 6

Results · A network of extrasynaptic monoamine signalling · The C. elegans connectome forms a multiplex network with nonredundant

Monoamine signalling forms a largely extrasynaptic and structurally distinct layer of the C. elegans connectome: most monoamine connections are not wired synapses, and monoamine hubs differ from wired-network hubs.

A network of extrasynaptic monoamine signalling

  • Additional profiling found dop-5, dop-6, and lgc-55 expression in small, largely distinct neuronal subsets, enabling a largely complete map of hermaphrodite monoamine receptor expression.The three receptors had missing or incomplete expression data before reporter lines were examined.Source: A network of extrasynaptic monoamine… · p. 7, Figure 1
  • About 94% of tyramine-responsive neurons, 100% of octopamine-receptor neurons, 78% of dopamine-receptor neurons, and 76% of serotonin-receptor neurons receive no synaptic input from neurons releasing the cognate monoamine.These results indicate that most neuronal monoamine signalling occurs outside the wired synaptic connectome.Source: A network of extrasynaptic monoamine… · p. 8

layers

  • Among 2,036 monoamine connections, only 89 overlap with chemical or electrical synapses, meaning 96% of monoamine connections are unique to the monoamine layer.Source: layers · p. 11
  • Adding monoamine signalling yields a multiplex connectome of synaptic, gap-junction, and monoamine layers, with wired, electrical, and wireless broadcasting links differing in direction and speed.Monoamine connections are described as unidirectional and slow, while chemical synapses are wired and gap junctions are reciprocal electrical connections.Source: layers · p. 10
  • Reducibility analysis places monoamine networks in a distinct cluster, showing less overlap with wired networks than synaptic and gap-junction networks show with each other.Figure 3 presents the reducibility dendrograms supporting this separation.Source: layers · p. 11
  • Synaptic and gap-junction degrees correlate positively (R=.594), but wired and monoamine layers show no significant degree-degree correlation, indicating distinct hubs.The monoamine network therefore has a different hub structure from the wired layers.Source: layers · p. 12
  • The monoamine layer’s dissimilarity from wired networks may reflect either an independent network with little relation to them or a complementary function coupled to them.The passages present these as two possible interpretations of the observed dissimilarity.Source: layers · p. 12

Analysis of monoamine network topology

This section examines the topology of the monoamine and neuropeptide layers, including their network structures, overlap with other connectome layers, and relationships to one another.

Analysis of monoamine network topology

  • In the adult hermaphrodite, 248 neurons (82%) were found to express one or more monoamine receptors; the network has star-like topology, a heavy-tailed degree distribution, and high disassortativity.Source: Analysis of monoamine network topology · pp. 12–13
  • Adding monoamine connections greatly reduces aggregate path length and provides a direct sensory-to-motor route that bypasses premotor interneurons.These connections link more segregated wired-network subgraphs and may provide efficient global connections through highly connected hubs.Source: Analysis of monoamine network topology · p. 13
  • Monoamine connections reduce aggregate modularity, but the network remains more modular than degree-preserving random networks.The monoamine layer also has greater-than-random modularity under null models that preserve degree distribution.Source: Analysis of monoamine network topology · p. 14, Figure 4
  • The hub-and-spoke monoamine layer has lower transitivity than the wired network, while receptors on releasing neurons create a central cluster of hubs.Multiple receptors on a receiving neuron can form triangles with two transmitting neurons, enabling dual lateral inhibition in which one releasing neuron inhibits antagonistic hub signals and their downstream effects; dopamine hubs may also mutually suppress dopamine release.Source: Analysis of monoamine network topology · p. 14
  • The monoamine network contains a rich club of dopamine-, serotonin-, and tyramine-releasing neurons whose cross-expression of receptors suggests coordination among the amines.Most serotonergic neurons have tyramine and dopamine receptors, while dopaminergic and tyraminergic neurons express receptors for the other aminergic transmitters.Source: Analysis of monoamine network topology · p. 15, Figure 5
  • Neuropeptide hubs are distinct from those in other layers, and neuropeptide edges show low overlap with monoamine, synaptic, and gap-junction edges.Overall low overlap masks variation among individual neuropeptide systems.Source: Analysis of monoamine network topology · p. 17
  • CKR-2 overlaps significantly with monoamine networks, whereas NPR-1/2/5/11 show little overlap with wired or other extrasynaptic networks.The overlap patterns differ among individual neuropeptide systems.Source: Analysis of monoamine network topology · p. 17
  • Both neuropeptide and monoamine networks have shorter characteristic path lengths and lower modularity than wired networks.These are topological properties the two extrasynaptic networks share.Source: Analysis of monoamine network topology · pp. 17–18
  • Neuropeptide networks have higher clustering than any other connectome layer, and this remains significant against null models with the same edge density.Their larger number of connections partly explains the clustering, but does not account for the density-matched result.Source: Analysis of monoamine network topology · p. 18
  • Compared with monoamine networks, neuropeptide networks are significantly less disassortative and have higher reciprocity; individual neuropeptide systems generally lack the monoamines’ star-like topology.These are among the key topological differences between the two network types.Source: Analysis of monoamine network topology · p. 18

Modes of interaction between wired and extrasynaptic layers

Despite distinct connectome-layer topologies and little overlap, synaptic, gap-junction, monoamine, and neuropeptide connections show significant cross-layer relationships. RIM and DVA link layers, and multilink motifs identify recurring combinations of wired and extrasynaptic connections.

Modes of interaction between wired and extrasynaptic layers

  • Although wired and monoamine rich clubs do not overlap, significant links connect them.RIM neurons are among the connections linking these layers.Source: Modes of interaction between wired and… · p. 18
  • RIM and DVA rank among the five highest-participation neurons across synaptic, gap-junction, and monoamine layers; DVA receives serotonin, dopamine, and tyramine, while RIMs connect strongly to wired rich-club premotor neurons.The five highest-ranked neurons include RIML and RIMR from the monoamine rich club and RIBL and DVA from the wired rich club.Source: Modes of interaction between wired and… · p. 18
  • Of 20 possible multilink motifs, seven were overrepresented and four underrepresented relative to networks with randomized layers.Motif 1, which has no connections, was overrepresented, indicating that some link combinations and empty neuron pairs occur more often than expected by chance.Source: Modes of interaction between wired and… · p. 20, Figure 7
  • Synapses and gap junctions frequently overlap, consistent with overrepresentation of reciprocal chemical synapses (motif 3).Source: Modes of interaction between wired and… · p. 21
  • Monoamine motif 10—a unidirectional monoamine link coincident with reciprocal synapses—is overrepresented and may support positive or negative feedback in response to experience.The authors suggest this motif may be a functionally important aspect of monoamine activity in the wider network.Source: Modes of interaction between wired and… · p. 21
  • Motif 10 connections have been linked to pathogen-avoidance learning in ADF–AIY, escape head-movement suppression in RIM–RMD [Pirri et al., 2009], and neuropeptide release in PDE–DVA.Source: Modes of interaction between wired and… · p. 21
  • Despite little structural overlap between neuropeptide and monoamine networks, RIM and DVA remain central across four layers, and neuropeptide motif 10 is significantly overrepresented.Neuropeptide motif 10 combines a unidirectional neuromodulatory connection with a reciprocal synaptic connection.Source: Modes of interaction between wired and… · p. 22
  • Neuropeptide motif 20—reciprocal neuropeptide and synaptic connections coincident with a gap junction—was even more overrepresented; examples occur in the RMG hub-and-spoke network implicated in locomotion, aggregation, and pheromone response [Jang et al., 2012; Macosko et al., 2009].Source: Modes of interaction between wired and… · pp. 22–23

Discussion

The expanded C. elegans connectome adds extrasynaptic monoamine and neuropeptide networks with distinct topologies, which may coordinate neural activity independently and through interactions with the wired network.

  • The expanded connectome incorporates newly compiled monoamine and neuropeptide networks inferred from gene-expression data, adding many links largely nonoverlapping with the wired connectome.Source: Discussion · pp. 23–24
  • The neuropeptide network is highly clustered and substantially more reciprocal, suggesting a role in nervous-system module cohesion and functional segregation.Source: Discussion · pp. 23–24
  • Monoamine signalling forms a star-like network with a few high-degree broadcasting hubs interconnected as a rich-club, suited to coordinating global neural and behavioural states.The discussion describes this topology as highly disassociative.Source: Discussion · p. 23
  • Although extrasynaptic networks do not overlap with the wired connectome, their hubs are interconnected, with multilink motifs marking specific interaction points.Source: Discussion · p. 24
  • Extrasynaptic networks may act independently, for example through the monoamine rich-club, and in concert with synaptic circuitry through RIM and DVA hubs and overrepresented multilink motifs.Source: Discussion · p. 24
  • The inferred neuromodulatory links have largely unknown valence and strength, but the networks provide a more complete picture of potential communication pathways in the C. elegans nervous system.Source: Discussion · p. 24

Topological properties of monoamine and neuropeptide networks

Monoamine and neuropeptide networks have distinct topologies, and multilink analysis identifies overrepresented combinations linking extrasynaptic signalling with synaptic and gap-junction structure. The patterns suggest roles in collective dynamics, functional-module cohesion, and feedback.

Topological properties of monoamine and neuropeptide networks

  • The monoamine network has a star-like architecture, reflected in high disassortativity and few recurrent connections.Its highly interconnected high-degree releasing neurons form a rich club distinct from, though linked to, the wired connectome’s rich club.Source: Topological properties of monoamine… · p. 25
  • Mammalian monoamine systems share features with C. elegans, including sparse releasing neurons with broad targets and substantial extrasynaptic volume transmission (De-Miguel and Fuxe, 2012; Fuxe et al., 2012).Mammalian monoamine-releasing neurons also appear to act as high-degree broadcasting hubs with functionally and spatially diverse targets (Trueta and De-Miguel, 2012).Source: Topological properties of monoamine… · p. 25
  • The cross-species parallel is not exact: most, if not all, C. elegans aminergic neurons appear capable of long-distance signalling, whereas glial diffusion barriers can restrict monoamines in larger nervous systems.Understanding these hubs within the fully mapped C. elegans wired circuitry may inform currently unknown multilayer-network structures.Source: Topological properties of monoamine… · p. 26
  • Although only partially characterized, the neuropeptide layer differs from other connectome layers in its strikingly high clustering and much higher reciprocity than monoamine signalling.Its clustering remains strikingly high even accounting for its high connection density.Source: Topological properties of monoamine… · p. 26
  • The neuropeptide network’s properties suggest a role in segregation and cohesiveness of functional modules within the nervous system.The monoamine network’s star-like structure is described as well suited to collective network dynamics.Source: Topological properties of monoamine… · pp. 25–26
  • In both extrasynaptic networks, motif 10—an unidirectional extrasynaptic link coincident with a recurrent synaptic connection—is overrepresented and suited to feedback between linked nodes.It occurs in several microcircuits implicated in learning and memory.Source: Topological properties of monoamine… · p. 26
  • For neuropeptides, motif 20—reciprocal neuromodulatory and synaptic links coincident with a gap junction—is even more highly overrepresented.It occurs in several locations in the RMG-centred hub-and-spoke circuit, which controls aggregation and arousal; further examples may emerge as more neuropeptide systems are characterized.Source: Topological properties of monoamine… · pp. 26–27

A prototype for multiplex network analysis

The C. elegans multiplex connectome could become a gold standard for multilayer network study, with synaptic, gap-junction, and monoamine layers providing relatively reliable, complete maps of three connection types.

A prototype for multiplex network analysis

  • Multiplex datasets are often limited in size or in the interaction types they can include because of data availability.The connectome’s potential significance comes amid these data-collection challenges.Source: A prototype for multiplex network… · pp. 27–28
  • The synaptic, gap-junction, and monoamine layers are relatively reliable and complete maps of three distinct connection types, positioning the C. elegans multiplex connectome as a potential gold standard.The wired connectome has served as a reference for monoplex network study over the past 15 years.Source: A prototype for multiplex network… · p. 28
  • The lack of degree-degree correlation between some layers suggests they are distinct communication channels, likely coupled in higher-order motif structures.Rather than facets of one underlying network, the layers may interact through multilayer motifs.Source: A prototype for multiplex network… · p. 28
  • Different operating timescales across layers may allow interesting dynamical phenomena to emerge.The passage presents this as a possibility arising from the layers’ different timescales.Source: A prototype for multiplex network… · p. 28
  • The many distinct extrasynaptic interactions offer scope for a more refined dataset aligned to the same complete set of 302 nodes.Source: A prototype for multiplex network… · p. 28

Prospects for complete mapping of multilayer connectomes · How feasible is it to obtain a complete multiplex neuronal connectome?

The monoamine network is a reasonable draft, but incomplete receptor identification likely leaves its non-synaptic edges and the importance of broadcasting hubs understated. A fuller multilayer connectome is feasible in C. elegans, though completing neuropeptide and other signalling layers remains challenging.

How feasible is it to obtain a complete multiplex neuronal connectome?

  • The monoamine network is a reasonable draft, but additional receptor-expressing cells mean non-synaptic edges are undercounted and the importance of monoamine-releasing hubs is understated.The neuropeptide network described is only a sample.Source: How feasible is it to obtain a… · pp. 28–29
  • False positives in identifying monoamine-receptor-expressing neurons are probably very low, whereas false negatives are likely higher because reporters can underreport expression or identify only subsets of expressing cells.Recently developed marker strains could help revisit cell identification and fill some gaps.Source: How feasible is it to obtain a… · pp. 28–29
  • Other monoamines and currently uncharacterized orphan receptors may add signalling targets, while reanalysis using reference strains could yield a largely complete C. elegans monoamine network.Melatonin is one possible additional monoamine neuromodulator.Source: How feasible is it to obtain a… · pp. 29–30
  • Despite these challenges, comprehensive multilayer mapping is uniquely feasible in C. elegans because its nervous system is small and precisely characterized, and could prototype how signalling modes interact.A complete monoamine network appears more attainable than a complete neuropeptide network.Source: How feasible is it to obtain a… · pp. 29–30
  • Mapping extrasynaptic connectomes in larger brains, especially mammals, will likely be much more complicated because of both increased size and additional structural and dynamical properties that alter extracellular diffusion.The cited examples include glial barriers, cellular swelling, and arterial pulsations (Sykova and Nicholson, 2008; Taber and Hurley, 2014).Source: How feasible is it to obtain a… · p. 29
  • A complete neuropeptide network would require comprehensive de-orphanization of neuropeptide GPCRs and expression patterns for hundreds of receptor and peptide genes; extrasynaptic signalling by insulin-like peptides, purines, and acetylcholine also remains unmapped.Source: How feasible is it to obtain a… · p. 30

Materials & Methods … Neuron identification & microscopy

The methods combined published synaptic and gap-junction connectomes with expression-based monoamine and neuropeptide networks, and used reporter strains and microscopy to identify receptor-expressing neurons.

Synaptic & gap junction networks

Monoamine network construction

  • Monoamine-associated genes were identified through literature searches for receptors, transporters and synthetic enzymes, followed by searches for cell-level expression data, including curated WormBase data (Howe et al., 2016).Source: Monoamine network construction · p. 31
  • In the monoamine layer, multiple receptors for one monoamine on a neuron yield one edge per sender, while reciprocal links are represented as separate directed connections.Source: Monoamine network construction · p. 31

Neuropeptide network construction

Neuron identification & microscopy

  • Dopamine receptor expression was localized using reporter strains and cell-body position and shape, usually with marker colabeling; confocal image stacks were acquired and processed with ImageJ.Source: Neuron identification & microscopy · pp. 32, 34

Topological network measures … Assortativity

This methods section specifies network conventions and defines the topological measures and null-model procedures used to compare layers.

Topological network measures

Clustering coefficient

Characteristic path length

  • Characteristic path length is based on the minimum, or geodesic, distance between each pair of nodes.Source: Characteristic path length · p. 35
  • Modularity partitions a network into non-overlapping modules to maximize within-module connectivity and minimize between-module connectivity, measuring the proportion of edges within modules.Source: Characteristic path length · p. 36

Assortativity

  • Assortativity correlates endpoint degrees, while structural reducibility compares relative Von Neumann entropies and Ward clustering visualizes layer similarity.Source: Assortativity · pp. 36–37

Reciprocity … Supplemental Information Supplemental Tables

The section defines network reciprocity and the rich-club coefficient, describes multilink motif analysis across synaptic, gap-junction, and monoamine layers, and provides supplemental expression tables.

Reciprocity

  • Reciprocity is the fraction of network edges that are reciprocal, with M denoting all edges and E↔ the reciprocal edges.Source: Reciprocity · p. 37

Rich-‐club coefficient

  • For each degree k, nodes with degree ≤k are removed, and the rich-club coefficient Φ_k is the remaining connections divided by the maximum possible connections.Source: Rich-‐club coefficient · p. 37
  • A rich club is defined using the prior wired-network threshold: normalized Φ_k ≥ 1 + 1σ, where σ is the standard deviation of normalized Φ_k.Source: Rich-‐club coefficient · p. 38

Multilink motifs

  • Motif instances were counted across the three layers and compared with 100 randomized three-layer networks to calculate motif z-scores and p-values.Source: Multilink motifs · p. 38

Supplemental Information Supplemental Tables

ntr-1 Expr11372 ASH, RIC, ADL, ADF, PVW, PVR, · Supplemental References

The supplied material lists neuronal expression annotations and supplemental references, and notes a binding-assay caveat for NPR-11/NLP-1. It provides no EC50 value for that receptor-ligand pair, though strong biological activity was observed in the micromolar range.

ntr-1 Expr11372 ASH, RIC, ADL, ADF, PVW, PVR,

Supplemental References

ntr-1 Expr11372 ASH, RIC, ADL, ADF, PVW, PVR,

  • NPR-11/NLP-1 has no reported EC50 value, although strong biological activity was seen in the micromolar range.This caveat appears in Table S10 on neuropeptide receptor-ligand binding.Source: Table S10 · p. 58

Related papers

Loading 1608.08793v1…