Prolactin-sensitive olfactory sensory neurons regulate male preference in female mice by modulating responses to chemosensory cues

Abstract

Chemosensory cues detected in the nose need to be integrated with the hormonal status to trigger appropriate behaviors, but the neural circuits linking the olfactory and the endocrine system are insufficiently understood. Here, we characterize olfactory sensory neurons in the murine nose that respond to the pituitary hormone prolactin. Deletion of prolactin receptor in these cells results in impaired detection of social odors and blunts male preference in females. The prolactin-responsive olfactory sensory neurons exhibit a distinctive projection pattern to the brain that is similar across different individuals and express a limited subset of chemosensory receptors. Prolactin modulates the responses within these neurons to discrete chemosensory cues contained in male urine, providing a mechanism by which the hormonal status can be directly linked with distinct olfactory cues to generate appropriate behavioral responses.

Fig. 1.. Loss of social preference in conditional olfactory Prlr-knockout females.

(A) Two-choice preference tests were performed in a three-compartment cage. After 10-min habituation, pieces of filter paper with male and female urine were placed in the left and right compartment, respectively. i.p., intraperitoneally. (B) Representative movement traces of Prl-injected control and Prlr-cKO females in the three-compartment cage during test sessions. (C and D) Time that control and Prlr-cKO females with saline (C) or prolactin (D) injection spent in the left compartment during test sessions. (E and F) Preference toward the left compartment (male urine) over the right compartment (female urine) during test sessions. n = 6 for controls and n = 7 for cKOs. (G) Two-choice preference tests performed with a male (left) and a female (right) in mesh boxes. (H) Time that control and Prlr-cKO females at proestrus spent in the left compartment. n = 5 for controls and cKOs. (I) Preference toward the left compartment (a male) over the right compartment (a female). n = 5 for controls and cKOs. (J and K) Social preference of females in a paced-mating chamber separated into two compartments by a 10-cm divider (J) or a divider with small holes (K). (L to O) Time that control and Prlr-cKO females at diestrus spent in the male compartment after saline (L and N) or prolactin (M and O) injection in a paced-mating chamber with a 10-cm divider (L and M) or a divider with holes (N and O). (P to S) Time that females at diestrus (P and R) or proestrus (Q and S) spent in the male compartment of a paced-mating chamber with a 10-cm divider (P and Q) or a divider with holes (R and S). n = 10 for controls and n = 7 for cKOs. Asterisks show significant differences [*P < 0.05 and **P < 0.01; P = 0.0090 (D), P = 0.0039 (F), P = 0.0083 (H), P = 0.0219 (I), P = 0.0059 (M), P = 0.0309 (O), P = 0.0131 (P), P = 0.0016 (Q), and P = 0.0324 (S)] based on the two-tailed t test (C to F, H, I, and L to S). Error bars show average ± SEM.

Fig. 2.. Female Prlr-OSNs respond to male urine.

(A) Representative calcium traces of Prlr-OSNs using calcium imaging in Prlr-GCaMP3 females (top). Still frames at t = 0 (5 min before ACSF application), t = 5.5 (ACSF application), t = 11 (first male urine application), and t = 16.5 (second male urine application) are shown (bottom). (B) Responsiveness of Prlr⁺ OSNs to male urine application. n = 144 (seven females). (C) Responsiveness of Prlr⁺ VSNs to male urine application. n = 92 (three females).

Fig. 3.. A map of Prlr-glomeruli in the MOB.

(A) GFP⁺ cells (green) in Prlr-GFP mice MOE. Nuclei counterstained with bisbenzimide (blue). White dots demarcate GFP⁺ cell positions. Scale bars, 200 (overview) and 20 μm (inset). (B) Normalized quantification of GFP⁺ cells from female Prlr-GFP MOE. GFP⁺ cells at embryonic day 12.5 (E12.5), E14.5, postnatal day 21 (P21), and P84 normalized to MOE area. Shown as means ± SEM. Asterisks depict significance (*P < 0.05, **P < 0.01, and ***P < 0.001) based on one-way analysis of variance (ANOVA) and Tukey’s multiple comparison test. ns, not significant. n = 3 per group. (C) RT-PCR on cDNA from female Prlr-GFP FACS-sorted GFP⁺ and GFP⁻ cells. (D) Omp probe in situ hybridization on female Prlr-GFP MOE sections. Black arrowheads, white arrowheads, and asterisks indicate dendrite, cell body, and axon, respectively. Scale bar, 10 μm. (E) Dual immunolabeling for GFP (green) and OSN markers; OMP, OMACS, and Nrp2 (magenta) on adjacent MOB sections. Nuclei counterstained with bisbenzimide (blue). Scale bars, 200 (overview) and 50 μm (inset). (F) Granule cell (Gr), mitral/tufted cell (M/T), external plexiform (EP), and glomerular (G) layers are devoid of GFP⁺ cells. (G) Whole-mount X-gal staining showing LacZ expression in female Prlr-τlacZ lateral MOB (top). Glomeruli positions are summarized for three females (bottom). Scale bar, 500 μm.

Fig. 4.. A distinct set of chemosensory receptors is expressed in Prlr-OSNs.

(A) FPKM values of RNA-seq data from three different Prlr-GFP females. FPKM values >1 × 10⁻² are shown. Several key olfactory signal transduction genes [Cnga2 (cyclic nucleotide–gated channel alpha 2), Gnal (guanine nucleotide–binding protein G(olf) subunit alpha), Adcy3 (adenylate cyclase type 3), and Ano2 (anoctamin 2)] displayed high FPKM values. (B) FPKMs of Ors and Taars found in the Prlr-OSNs were plotted from the most abundant to the least. Mean of FPKMs from three replicates (line) and the SEM (filled) are shown. Ors and Taars with FPKMs higher than 40 were shown in the inset. (C) FPKMs of Ors and Taars found in all OSNs from an external bulk RNA-seq (GSE112352) were plotted from the most abundant to the least. FPKMs for top 71 genes were shown in inset. (D) In situ hybridization with probes for Ors on MOE sections from Prlr-GFP females. Olfr1348, Olfr1428, and Olfr672 (magenta) were coexpressed with Prlr/GFP (green) (white arrowheads). Scale bar, 10 μm.

Fig. 5.. Prolactin modulation of OSN responses.

(A) Representative traces of normalized GCaMP3 responses to 100 nM IPA, prolactin (PRL; 500 ng/ml), and combination IPA + PRL. ACSF as control application at the beginning of measurements. (B) Heatmaps of normalized fluorescence intensities from 192 neurons. Vertical lines show corresponding substance application. Bottom: Time periods taken for Ca²⁺ response analysis (15 s before and after substance application). Time frame number indicates each different substance application. Neuron presentation order in the heatmaps was changed according to analysis of Ca²⁺ responses from (C). (C) Ca²⁺ response analysis. Neurons were clustered by their correlation of responses to IPA, PRL, and PRL + IPA. (C) Top: 3D plot of the ΔFluorescent z scores from 192 neurons (calculated with the central formula; applied substances indicated on the 3D plot axes). One circle represents each neuron, collected into clusters (color coded). For every cluster, the corresponding line and coefficient of determination (R²) represents the best linear fit (first principal component) between responses. Clustering based on the correlation matrix [heatmap; (C), middle and right] is presented as a color-coded bar. All the ΔFluorescent z scores for all application cases; time frames for ACSF (1), IPA (2), PRL (3), PRL + IPA (4) are seen as a heatmap [(C), left], where 192 neurons were clustered by their responses (color-coded bar). Histogram [(C), bottom and right] shows normalized ΔFluorescent z scores to ACSF and all responses (IPA, PRL, and PRL + IPA values collected together). The 95% quantile (red line, q_0.975, for ACSF distribution) was taken as a cutoff for “significant” responses. (D) Response variability. Pie chart (inner circle) shows percentage of neurons responding to IPA, PRL, and IPA + PRL [types of responses are listed on the right; color-coding matches clustering from (C)]. Pie chart outer sectors show percentage of neurons with responses above the cutoff [compared to control (ACSF) responses, determined in (C)].