Distinct projections of PVN oxytocin neurons regulate the divergent defensive behaviors in response to social threat

LH (Lateral habenula) MBA:186 functionally connects to PVH (Paraventricular hypothalamic nucleus) MBA:38 · Mouse

Optical activation of LHb axons in the PVN produced TTX-sensitive, 4-AP-restorable EPSCs in oxytocin neurons, consistent with a monosynaptic excitatory connection, and also triggered local oxytocin release detected by the OXT1.7 sensor.

fig. S4, E to L

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

MPO (Medial preoptic area) MBA:523 synapses onto PVH (Paraventricular hypothalamic nucleus) MBA:38 · Mouse

Rabies-labelled presynaptic cells were plentiful in the medial preoptic nucleus for both NAc- and CeA-projecting PVN oxytocin neurons, with similar amounts for the two populations.

Fig. 5G

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LH (Lateral habenula) MBA:186 projects to SO (Supraoptic nucleus) MBA:390 · Mouse

Anterograde tracing from the lateral habenula also labelled fibres in the supraoptic nucleus.

fig. S4, A to D

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LHA (Lateral hypothalamic area) MBA:194 projects to PVH (Paraventricular hypothalamic nucleus) MBA:38 · Mouse

Anterograde virus in the lateral hypothalamic area combined with oxytocin immunostaining revealed fibres in the PVN, and the PVN received more LHA than LHb input.

fig. S4, A to D

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

PVH (Paraventricular hypothalamic nucleus) MBA:38 projects to CEA (Central amygdalar nucleus) MBA:536 · Mouse

Axon-GCaMP6m driven in PVN oxytocin neurons labelled terminals in the central amygdala, allowing terminal calcium recordings there.

Fig. 2I

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LHA (Lateral hypothalamic area) MBA:194 functionally connects to PVH (Paraventricular hypothalamic nucleus) MBA:38 · Mouse

Photostimulating channelrhodopsin-expressing LHA terminals evoked excitatory postsynaptic currents in PVN oxytocin neurons that were blocked by TTX and rescued by 4-AP, indicating a monosynaptic excitatory input; the same stimulation in vivo also raised the PVN OXT1.7 sensor signal.

fig. S4, E to L

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LH (Lateral habenula) MBA:186 synapses onto PVH (Paraventricular hypothalamic nucleus) MBA:38 · Mouse

Rabies tracing from PVN oxytocin starter cells labelled lateral habenula neurons, with markedly more LHb input to CeA-projecting than to NAc-projecting oxytocin neurons.

Fig. 5, G and H

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LHA (Lateral hypothalamic area) MBA:194 projects to MPO (Medial preoptic area) MBA:523 · Mouse

The medial preoptic area was among the oxytocin-containing regions screened for LHA input, but only the PVN and supraoptic nucleus were reported as receiving projections, leaving the preoptic result unresolved.

fig. S4, A to D

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

PVH (Paraventricular hypothalamic nucleus) MBA:38 functionally connects to ACB (Nucleus accumbens) MBA:56 · Mouse

Light stimulation of ChrimsonR-expressing PVN oxytocin axons in the accumbens produced a rapid rise of the OXT1.7 sensor signal there, indicating oxytocin release from this pathway.

fig. S3, F and G

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

ZI (Zona incerta) MBA:797 synapses onto PVH (Paraventricular hypothalamic nucleus) MBA:38 · Mouse

The zona incerta provided abundant and comparable rabies-traced input to both NAc-projecting and CeA-projecting PVN oxytocin neurons.

Fig. 5G

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

PVH (Paraventricular hypothalamic nucleus) MBA:38 functionally connects to CEA (Central amygdalar nucleus) MBA:536 · Mouse

Optogenetic activation of PVN oxytocin terminals in the central amygdala increased the local OXT1.7 fluorescence, showing oxytocin release onto CeA.

fig. S3, H and I

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

PVH (Paraventricular hypothalamic nucleus) MBA:38 projects to ACB (Nucleus accumbens) MBA:56 · Mouse

A retrograde virus placed in the nucleus accumbens labelled oxytocin-immunoreactive neurons in the paraventricular hypothalamic nucleus, and these NAc-projecting OXT cells occupied a territory separate from the CeA-projecting ones.

fig. S3, A to E

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

PVH (Paraventricular hypothalamic nucleus) MBA:38 projects to CEA (Central amygdalar nucleus) MBA:536 · Mouse

Retrograde virus injected into the central amygdala labelled oxytocin-positive neurons in the PVN, which were spatially segregated from the NAc-projecting population.

fig. S3, A to E

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LHA (Lateral hypothalamic area) MBA:194 projects to BST (Bed nuclei of the stria terminalis) MBA:351 · Mouse

The bed nucleus of the stria terminalis was examined for LHA fibres near oxytocin cells, but the paper reports positive findings only for PVN and SON, so the BNST outcome is not stated.

fig. S4, A to D

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

PVH (Paraventricular hypothalamic nucleus) MBA:38 projects to ACB (Nucleus accumbens) MBA:56 · Mouse

Cre-dependent axon-targeted GCaMP6m expressed in PVN oxytocin neurons labelled axonal terminals within the nucleus accumbens, where fibre photometry signals were recorded.

Fig. 2C

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LH (Lateral habenula) MBA:186 projects to PVH (Paraventricular hypothalamic nucleus) MBA:38 · Mouse

Anterograde tracing from the lateral habenula showed labelled projections reaching the PVN, though fewer than those from the LHA.

fig. S4, A to D

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LH (Lateral habenula) MBA:186 projects to BST (Bed nuclei of the stria terminalis) MBA:351 · Mouse

The bed nucleus of the stria terminalis was screened for LHb input to oxytocin neurons, but the reported positive targets were PVN and SON only.

fig. S4, A to D

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LH (Lateral habenula) MBA:186 projects to MPO (Medial preoptic area) MBA:523 · Mouse

LHb anterograde tracing included the medial preoptic area among the surveyed oxytocin regions, but only PVN and SON were reported as targets, leaving this connection unresolved.

fig. S4, A to D

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LHA (Lateral hypothalamic area) MBA:194 synapses onto PVH (Paraventricular hypothalamic nucleus) MBA:38 · Mouse

Monosynaptic rabies tracing from PVN oxytocin starter cells labelled presynaptic neurons in the lateral hypothalamic area; input to NAc-projecting oxytocin neurons tended to be greater than to CeA-projecting ones.

Fig. 5, G and H

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.

LHA (Lateral hypothalamic area) MBA:194 projects to SO (Supraoptic nucleus) MBA:390 · Mouse

Anterograde viral tracing showed that the supraoptic nucleus, like the PVN, receives projections from the lateral hypothalamic area.

fig. S4, A to D

Made by an AI model reading the paper (claude-opus-5, extract@0.3.0); a second AI model (claude-opus-5) agrees.