ErbB4 in the OB is critical for odour discrimination and dishabituation.

(A) Behavioural paradigm for olfactory associative learning without odour discrimination (go/go task). Mice inserted their snout into the sampling port to trigger odours. The schematic describes the timeline of a single trial. Mice learned to lick the metal tube to receive water in response to either of the odours in the pair (reward, hit). (B) Timeline for a single trial in the go/no-go odour discrimination task. Mice learned to avoid licking the metal tube for the unrewarded odour (correct rejection, CR). Licking when presented with the unrewarded odour (false alarm, FA) led to no water reward and a timeout of up to 10 s. (C and D) ErbB4 activity in the OB was elevated after training on the reinforced go/no-go odour discrimination task. OB cell lysates were subjected to the IP or IB using the indicated antibodies. Relative p-ErbB4/ErbB4 levels were normalized to their respective control groups in the western blot analysis. (n = 4 mice per group, t(6) = 5.75, P = 0.0012, unpaired t test). (E) Schema indicating virus injection sites. To specifically delete ErbB4 protein in the OB, AAV-Cre-GFP was injected into bilateral OB of neonatal loxP-flanked Erbb4 mice. (F and G) Reduced ErbB4 expression in the OB of a mouse injected with AAV-Cre-GFP (n = 5 mice per group, t(8) = 24.7, P < 0.0001, unpaired t test). (H) Odour discrimination performance under the reinforced go/no-go task. The accuracy for simple odour pairs was similar for the control and experimental groups (n = 8 mice per group, F(1, 14) = 2.83, P = 0.1148, two-way ANOVA). However, the accuracy for difficult odour mixtures (6/4 V 4/6) was reduced in AAV-Cre-GFP mice (F(1, 14) = 16.14, P = 0.0013, two-way ANOVA). (I) Olfactory behavioural performance under a spontaneous habituation/dishabituation task. Both animal groups showed a decline in investigation time to isoamyl acetate over the habituation period (n = 9 AAV-GFP mice, P < 0.0001; n = 10 AAV-Cre-GFP mice, P < 0.0001, F(3, 51) = 21.83, two-way ANOVA). However, AAV-GFP (F(1, 17) = 3.52, P = 0.0065), but not AAV-Cre-GFP mice (P = 0.6296, two-way ANOVA), showed an increase in investigation time toward limonene in the dishabituation period. (J) Odour detection threshold to isoamyl acetate. For AAV-GFP mice, the sniffing time toward isoamyl acetate was significantly higher than that for mineral oil at concentrations of 10-5 and 10-4, but not 10-6. (n = 12 mice, F(1, 21) = 0.19, 4.18 and 14.88, P = 0.5689, 0.0111, and 0.0090, two-way ANOVA). These results show that AAV-GFP mice were able to detect isoamyl acetate at a concentration of 10−5. For AAV-Cre-GFP mice, the sniffing time toward isoamyl acetate was significantly higher than that for mineral oil at a concentration of 10-4, but not at concentrations of 10-5 and 10-6 (n = 11 mice, P = 0.9672, 0.8713, and 0.0172, two-way ANOVA). These results show that AAV-Cre-GFP mice only detect isoamyl acetate at a concentration of 10−4. (K) Similarly, AAV-GFP mice were able to detect limonene at a concentration of 10−5 (n = 12 mice, F(1, 21) = 0.03, 4.77 and 12.96, P = 0.9603, 0.0011, and 0.0069, two-way ANOVA), whereas AAV-Cre-GFP mice only detected limonene at a concentration of 10−4 (n = 11 mice, P = 0.7671, 0.5490, and 0.0463, two-way ANOVA). Data are presented as means ± s.e.m. * P < 0.05, ** P < 0.01, **** P < 0.0001, n.s. = not significant. EPL, external plexiform layer; GCL, granule cell layer; GL, glomerular layer; IPL, internal plexiform layer; MCL, mitral cell layer.

ErbB4 proteins are largely expressed in PV interneurons in the OB.

(A) In vitro imaging of ErbB4 mRNA in sections from PV-Erbb4+/+ mice. Double single-molecule fluorescence in situ hybridization of ErbB4 (green) and PV (red) in the OB. Scale bar, 200 μm. (B) Magnified view of the EPL box from A. Scale bar, 50 μm. (C) Magnified view of the IPL box from A. Scale bar, 50 μm. (D) Summarized data showing the proportion of ErbB4-expressing neurons in different layers (n = 9 slices from 4 mice). (E) Summarized data showing the proportion of the ErbB4/PV double-positive neurons relative to the total number of PV interneurons in the EPL and IPL (n = 9 slices from 4 mice). DAPI staining was used to determine the total number of cells. (F) In vitro electrophysiology experiments performed in slices from PV-Erbb4+/+ or PV-Erbb4−/− mice. Representative examples of action potentials (APs) elicited by positive current injection (500 ms, 300 pA), recorded from an MC (left) and a fast-spiking PV interneuron (right). (G and H) Corresponding single-cell RT-PCR analyses showing that ErbB4 mRNA is detected only in PV interneurons (PVN) from PV-Erbb4+/+ OB (n = 6 and 4 neurons from 4 mice, P = 0.0048, unpaired t test). Dl1000 was used as the size reference (M, 300, 200 and 100 base-pair fragments are indicated). (I and J) Specific deletion of ErbB4 in EPL PV interneurons of the OB (n = 11 slices from 4 mice per group, t(20) = 16.91, P < 0.0001, unpaired t test). OB sections from PV-Erbb4+/+ and PV-Erbb4−/− mice (P28) were stained with anti-PV, ErbB4 and DAPI antibody. Scale bars represent 20 μm. (K and L) Western blots showing that ErbB4 in PV-Erbb4−/− OB was largely reduced from P7 onward, whereas ErbB4 in the PFC and hippocampus began to decrease only at P21. Relative levels were normalized to their respective P7 groups of control littermates (n = 3 mice per group, OB: F(1, 8) = 245.70, P < 0.0001, P = 0.0002, 0.0006, and 0.0010; PFC: F(1, 8) = 61.20, P = 0.0532, 0.1791, 0.0075, and 0.0021; Hi: F(1, 8) = 38.25, P = 0.2585, 0.1005, 0.0139 and 0.0382, two-way ANOVA). Data are presented as means ± s.e.m. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.001, n.s. = not significant. EPL, external plexiform layer; GCL, granule cell layer; GL, glomerular layer; Hi, hippocampus; MCL, mitral cell layer; PFC, prefrontal cortex.

ErbB4 in PV interneurons is critical for olfactory behaviours.

(A) Odour discrimination under the reinforced go/no-go task in PV-Erbb4−/− mice. The accuracy for simple odour pairs was indistinguishable (n = 5 and 6 mice, F(1, 9) = 0.70, P = 0.4260, two-way ANOVA). However, the accuracy for difficult odour pairs was significantly lower in PV-Erbb4−/−mice (F(1, 9) = 9.12, P = 0.0144, two-way ANOVA). (B) Olfactory behavioural performance under the spontaneous habituation/dishabituation task. Both animal groups habituated to isoamyl acetate (n = 12 mice per group, F(3, 66) = 6.68, P = 0.0349 and 0.0164, two-way ANOVA). However, PV-Erbb4+/+mice (F(1, 22) = 8.93, P = 0.0025), but not PV-Erbb4−/−mice (P = 0.8451, two-way ANOVA), dishabituated to limonene. (C) Olfactory behavioral performance under the reversed habituation/dishabituation task. Both animal groups habituated to limonene (F(3, 54) = 16.33, P < 0.0001 and 0.0003, two-way ANOVA). However, PV-Erbb4+/+ mice (F(1, 18) = 5.22, P = 0.0023), but not PV-Erbb4−/−mice (P = 0.7890, two-way ANOVA), dishabituated to isoamyl acetate. (D) PV-Erbb4+/+mice were able to detect isoamyl acetate at a concentration of 10−5 (n = 10 mice, F(1, 18) = 0.60, 3.74 and 16.69, P = 0.6069, 0.0498, and 0.0096 for 10-6, 10-5, and 10-4, two-way ANOVA). PV-Erbb4−/− mice only detected isoamyl acetate at a concentration of 10−4 (n = 10 mice, P = 0.5764, 0.5353, and 0.0100 for 10-6, 10-5, and 10-4, two-way ANOVA). (E) PV-Erbb4+/+ mice could detect limonene at a concentration of 10−5 (n = 10 mice, F(1, 18) = 1.75, 7.95 and 32.51, P = 0.4540, 0.0106, and 0.0064 for 10-6, 10-5, and 10-4, two-way ANOVA). PV-Erbb4−/− mice only detected limonene at a concentration of 10−4 (n = 10 mice, P = 0.2842, 0.2709, and < 0.0001 for 10-6, 10-5, and 10-4, two-way ANOVA). (F) The latency for mice to locate the buried and visible food pellets did not differ between the groups (n = 14 and 11 mice, for the buried food pellet, F(1, 23) = 0.21, P = 0.9786, 0.9279, and 0.8873; for the visible food pellet, P > 0.9999, two-way ANOVA). Data are presented as means ± s.e.m. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001, n.s. = not significant.

Odour-evoked responses in M/TCs are decreased in PV-Erbb4−/− mice.

(A) Schematic of in vivo odour-evoked electrophysiological recordings in awake, head-fixed mice with ErbB4 knocked out in PV interneurons. (B) Representative raw traces of spike activity before (spontaneous), during (odour-evoked), and after 2-s odour stimulation in PV-Erbb4+/+ and PV-Erbb4−/− mice. (C) Examples of raster plots (top) and PSTHs of the firing rate (bottom) for odour-evoked excitatory (left) and inhibitory (right) responses in PV-Erbb4+/+and PV-Erbb4−/− mice. PSTHs were smoothed with a Gaussian filter with a standard deviation of 1500 ms. (D) Heat maps of the odour-evoked changes in mean firing rate (ΔMFR) across all unit–odour pairs in PV-Erbb4+/+mice (n = 240 unit–odour pairs from 4 mice) and PV-Erbb4−/−mice (n = 232 unit–odour pairs from 4 mice). (E) Quantitative analysis of the spontaneous firing rate (t(470) = 0.27, P = 0.7899), odour-evoked MFR (t(470) = 3.30, P = 0.0010), absolute value of odour-evoked changes (t(470) = 5.05, P < 0.0001), and SNR (t(470) = 9.29, P < 0.0001, unpaired t test) across all unit–odour pairs. (F) Odour-evoked excitatory changes in MFR (ΔMFR) for M/TCs recorded from PV-Erbb4+/+ mice (n = 66 unit–odour pairs from 4 mice) and PV-Erbb4−/− mice (n = 16 unit–odour pairs from 4 mice). (G) Quantitative analysis of the spontaneous firing rate (t(80) = 7.67, P < 0.0001), odour-evoked MFR (t(80) = 1.43, P = 0.1578), odour-evoked ΔMFR (t(80) = 3.10, P = 0.0027), and SNR (t(80) = 5.91, P < 0.0001, unpaired t test) across excitatory unit–odour pairs. (H) Odour-evoked inhibitory ΔMFR for M/TCs recorded from PV-Erbb4+/+mice (n = 81 unit–odour pairs from 4 mice) and PV-Erbb4−/−mice (n = 148 unit–odour pairs from 4 mice). (I) Quantitative analysis of spontaneous firing rate (t(227) = 7.52, P < 0.0001), odour-evoked MFR (t(227) = 7.23, P < 0.0001), odour-evoked ΔMFR (t(227) = 2.72, P = 0.0070), and SNR (t(227) = 6.91, P < 0.0001, unpaired t test) across inhibitory unit–odour pairs. (J) ΔMFR for units with no response to odour in PV-Erbb4+/+ mice (n = 93 unit–odour pairs from 4 mice) and PV-Erbb4−/− mice (n = 68 unit–odour pairs from 4 mice). (K) Quantitative analysis of spontaneous firing rate (t(159) = 3.13, P = 0.0021), odour-evoked MFR (t(159) = 3.38, P = 0.0009), absolute value of ΔMFR (t(159) = 1.63, P = 0.1057), and SNR (t(159) = 0.43, P = 0.6687, unpaired t test) across “no response” unit–odour pairs. (L) Distribution of excitatory, inhibitory, and “no response” units in PV-Erbb4+/+ and PV-Erbb4−/− mice (χ2(2) =53.85, P < 0.0001, Chi-Square tests). ** P < 0.01, *** P < 0.001, **** P < 0.0001, n.s. = not significant.

The ongoing LFP in the OB is increased in PV-Erbb4−/− mice.

(A) Examples of ongoing LFP signals recorded in the OB from PV-Erbb4+/+and PV-Erbb4−/− mice. The five rows show the raw traces and the filtered theta, beta, low-gamma, and high-gamma signals. (B-E) Quantitative analysis of the averaged power spectra in the theta, beta, low-gamma, and high-gamma bands for the two groups. (F-I) Comparisons of power in the theta (n = 5 mice per group, t(8) = 2.80, P = 0.0233, unpaired t test), beta (t(8) = 2.80, P = 0.0232, unpaired t test), low-gamma (t(8) = 3.80, P = 0.0053, unpaired t test), and high-gamma (t(8) = 3.73, P = 0.0058, unpaired t test) oscillations in the OB in the two groups. * P < 0.05, ** P < 0.01.

PV-Erbb4−/− mice have impaired information output from the OB.

(A and B) Increased frequency of MC sAPs and olfactory nerve-evoked APs (eAPs) in PV-Erbb4−/− mice, but decreased ratio of eAPs to sAPs (n = 8-9 from 3 mice per group, sAPs: t(15) = 4.84, P = 0.0002, eAPs: t(15) = 3.87, P = 0.0015; ratio: t(15) = 3.90, P = 0.0014, unpaired t test). (C and D) The eAP-to-sAP ratio was reduced in PV-Erbb4−/− mice as the intensity of the stimulus increased (n = 8 from 3 mice per group, F(1, 70) = 32.39, P < 0.0001, two-way ANOVA). (E and F) AP frequency elicited by injection of positive currents was higher in PV-Erbb4−/− mice than PV-Erbb4+/+ mice (n = 9 from 4 and 3 mice per group, F(1, 80) = 78.9, P < 0.0001, two-way ANOVA). Data are presented as means ± s.e.m. ** P < 0.01, *** P < 0.001, **** P < 0.0001.

Decreased GABAergic transmission mediates the OB output impairments in PV-Erbb4−/− mice.

(A and B) The increased sAP frequency and decreased ratio of eAPs to sAPs in MCs could not be further enhanced by bicuculline in PV-Erbb4−/− mice (n = 7 from 3 PV-Erbb4+/+ mice, n = 9 from 3 PV-Erbb4−/− mice; for sAP, F(1, 14) = 17.08, P = 0.0008 and 0.5531; for ratio of eAPs to sAPs, F(1, 14) = 13.99, P = 0.0007 and 0.9360, two-way ANOVA). (C and D) The frequency but not the amplitude of MC mIPSCs was lower in PV-Erbb4−/− mice (n = 9 from 7 PV-Erbb4+/+and 6 PV-Erbb4−/−mice; for frequency, t(16) = 2.45, P = 0.0263; for amplitude, t(16) = 0.53, P = 0.6038, unpaired t test). (E and F) Neither frequency nor amplitude of MC mEPSCs was different in PV-Erbb4−/− mice versus PV-Erbb4+/+mice (n = 10 from 5 PV-Erbb4+/+ mice, n = 9 from 7 PV-Erbb4−/− mice; for frequency, t(17) = 0.14, P = 0.8899; for amplitude, t(17) = 0.89, P = 0.3863, unpaired t test). Data are presented as means ± s.e.m. * P < 0.05, *** P < 0.001, **** P < 0.0001, n.s. = not significant.

Reduced recurrent inhibition and EPL lateral inhibition of MCs in PV-Erbb4−/− mice.

(A and B) Recurrent inhibition did not increase alongside MC hyperactivity in PV-Erbb4−/− mice (n = 9 from 5 and 6 mice; for peak amplitude, t(16) = 1.57, P = 0.1351; for decay time constant, t(16) = 0.28, P = 0.7821, unpaired t test). (C and D) Recurrent IPSPs elicited by the same number of MC APs were smaller in PV-Erbb4−/− mice. In PV-Erbb4−/− mice, an 80 pA current induced ten APs, whereas in PV-Erbb4+/+ mice, a 100 pA current was needed to induce ten APs. Both the peak amplitude (n = 10 from 6 and 5 mice, t(18) = 2.46, P = 0.0242, unpaired t test) and the decay time constant (t(18) = 2.56, P = 0.0198, unpaired t test) of IPSPs evoked by the same number of APs were reduced in PV-Erbb4−/− mice. (E) Schematic of the EPL lateral inhibition experimental configuration. Whole-cell recording of an MC (right side of the schematic) upon stimulus of an adjacent glomerulus (left side of the schematic). A cut was made through the GL and GCL between the sites of conditioning stimulation and target glomerulus to isolate EPL lateral inhibition. (F and G) EPL lateral inhibition was observed in the OB of PV-Erbb4+/+ mice but not PV-Erbb4−/− mice (n = 10 from 7 PV-Erbb4+/+mice, F(1, 18) = 26.48, P < 0.0001; n = 10 from 6 PV-Erbb4−/− mice, P = 0.3426, two-way ANOVA). (H-J) Cutting through the EPL as well abolished the EPL lateral inhibition in PV-Erbb4+/+ mice. (H) Schematic of the experimental configuration. (I and J) Representative and quantitative analysis of spike frequency in PV-Erbb4+/+mice (n = 9 from 5 mice, t(8) = 0.80, P = 0.4468, paired t test). EPL, external plexiform layer; GL, glomerular layer; GCL, granule cell layer; MCL, mitral cell layer; ONL, olfactory nerve layer; PVN, PV interneuron. Data are presented as means ± s.e.m. * P < 0.05, **** P < 0.0001, n.s. = not significant.

ErbB4 in PV interneurons of the OB is critical for odour discrimination and sensitivity.

(A) Schema indicating virus-injection sites. To specifically delete ErbB4 protein in the PV interneurons of the OB, AAV-PV-Cre was injected into the bilateral OB of neonatal loxP-flanked Erbb4 mice. (B and C) Specific deletion of ErbB4 in PV interneurons (white arrow) but not in non-PV cells (purple arrow) of the OB. OB sections from AAV-PV-GFP injected and AAV-PV-Cre injected mice were stained with DAPI, anti-PV and ErbB4 antibody. Scale bars represent 20 μm. n = 14 slices from 4 mice per group, t(26) = 9.30, P < 0.0001 for PV interneurons, t(26) = 0.41, P = 0.6833 for non-PV cells, unpaired t test. (D) The accuracy in discriminating simple odour pairs was similar for the two groups (n = 6 and 7 mice, F(1, 11) = 0.06, P = 0.8147). The accuracy in discriminating difficult odour pairs was significantly lower in AAV-PV-Cre mice (F(1, 11) = 5.74, P = 0.0355, two-way ANOVA). (E) Both animal groups habituated to isoamyl acetate (n = 10 mice per group, F(3, 54) = 20.94, P < 0.0001 and = 0.0010, two-way ANOVA). However, the AAV-PV-GFP (F(1, 18) = 3.95, P = 0.0045), but not the AAV-PV-Cre mice (P = 0.7107, two-way ANOVA), dishabituated to limonene. (F) Both animal groups habituated to carvone+ (n = 10 mice per group, F(3, 54) = 9.92, P = 0.0019 and 0.0035, two-way ANOVA). However, AAV-PV-GFP (F(1, 18) = 6.72, P = 0.0025), but not AAV-PV-Cre mice (P = 0.9845, two-way ANOVA), dishabituated to carvone-. (G) AAV-PV-GFP mice were able to detect isoamyl acetate at a concentration of 10−5 (n = 10 mice, F(1, 18) = 0.04, 2.63 and 12.42, P = 0.8309, 0.0258, and 0.0164 for 10-6, 10-5, and 10-4, two-way ANOVA), whereas AAV-PV-Cre mice only detected isoamyl acetate at a concentration of 10−4 (n = 10 mice, P = 0.9484, 0.8930, and 0.0312, two-way ANOVA). (H) AAV-PV-GFP mice detected limonene at 10−5 (n = 10 mice, F(1, 18) = 0.28, 3.81 and 11.79, P = 0.7894, 0.0154, and 0.0131 for 10-6, 10-5, and 10-4, two-way ANOVA) but AAV-PV-Cre mice only detected limonene at a higher concentration of 10−4 (n = 10 mice, P = 0.6408, 0.9349, and 0.0496 for 10-6, 10-5, and 10-4, two-way ANOVA). Data are presented as means ± s.e.m. * P < 0.05, ** P < 0.01, **** P < 0.0001, n.s. = not significant. EPL, external plexiform layer; GCL, granule cell layer; GL, glomerular layer; IPL, internal plexiform layer; MCL, mitral cell layer.