ErbB4 in parvalbumin-positive interneurons is critical for neuregulin 1 regulation of long-term potentiation

Abstract

Neuregulin 1 (NRG1) is a trophic factor that acts by stimulating ErbB receptor tyrosine kinases and has been implicated in neural development and synaptic plasticity. In this study, we investigated mechanisms of its suppression of long-term potentiation (LTP) in the hippocampus. We found that NRG1 did not alter glutamatergic transmission at SC-CA1 synapses but increased the GABA(A) receptor-mediated synaptic currents in CA1 pyramidal cells via a presynaptic mechanism. Inhibition of GABA(A) receptors blocked the suppressing effect of NRG1 on LTP and prevented ecto-ErbB4 from enhancing LTP, implicating a role of GABAergic transmission. To test this hypothesis further, we generated parvalbumin (PV)-Cre;ErbB4(-/-) mice in which ErbB4, an NRG1 receptor in the brain, is ablated specifically in PV-positive interneurons. NRG1 was no longer able to increase inhibitory postsynaptic currents and to suppress LTP in PV-Cre;ErbB4(-/-) hippocampus. Accordingly, contextual fear conditioning, a hippocampus-dependent test, was impaired in PV-Cre;ErbB4(-/-) mice. In contrast, ablation of ErbB4 in pyramidal neurons had no effect on NRG1 regulation of hippocampal LTP or contextual fear conditioning. These results demonstrate a critical role of ErbB4 in PV-positive interneurons but not in pyramidal neurons in synaptic plasticity and support a working model that NRG1 suppresses LTP by enhancing GABA release. Considering that NRG1 and ErbB4 are susceptibility genes of schizophrenia, these observations contribute to a better understanding of how abnormal NRG1/ErbB4 signaling may be involved in the pathogenesis of schizophrenia.

Fig. 1.

AMPAR- or NMDAR-mediated currents were not altered in NRG1-treated hippocampal slices. BMI (20 μM) was included in artificial cerebrospinal fluid (ACSF) for experiments in this figure. AMPA and NMDA currents were evoked by holding membrane potentials at −70 mV and +40 mV, respectively. NRG1 has no effect on peak amplitudes of AMPA currents (A; n = 9; P > 0.05) or NMDA currents (B; n = 6; P > 0.05). Notice that 20 μM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 100 μM D(−)-2-amino-5-phosphonovaleric acid (AP5) were included in ACSF in A and B, respectively. Con, control. (C) NRG1 did not change PPRs of evoked EPSCs at SC-CA1 synapses (n = 6; P > 0.05). (Inset) Representative recordings. (D–F) There was no effect of NRG1 on mEPSCs on CA1 pyramidal neurons. Shown are representative traces (D) and cumulative plots (E) of mEPSC frequencies and amplitudes before and after NRG1 treatment. (F) Summary of results (n = 9; P > 0.05).

Fig. 2.

NRG1 increases GABA_A receptor-mediated synaptic response. (A) Increased eIPSCs in NRG1-treated slices. Shown are representative eIPSC traces (Upper) and quantitative analysis (Lower) (n = 11). (B) AG1478 blocks NRG1 enhancement of eIPSCs. Slices were pretreated without or with 5 μM AG1478 for 10 min and then with 1 nM NRG1, boiled NRG1, or vehicle (n = 8 for control, AG1478, and AG1478 plus NRG1; n = 7 for denatured NRG1; n = 5 for vehicle) (C–E) NRG1 increases mIPSC frequency but not amplitudes in CA1 pyramidal cells. Shown are representative traces (C) and cumulative plots (D) of mIPSC frequencies and amplitudes. (E) Summarized data of NRG1 (n = 8), denatured NRG1 (n = 4), and vehicle (n = 6). (F) NRG1 reduced PPRs of eIPSCs in the hippocampus (n = 5). *P < 0.05; **P < 0.01.

Fig. 3.

NRG1-induced suppression of LTP is blocked by GABA_A antagonists. (A) NRG1 suppresses LTP at SC-CA1 synapses in the hippocampus. Normalized fEPSP slopes were plotted every 1 min for control (○) or slices treated with 1 nM NRG1 (●), which was applied during the period indicated by the bar. (Right) Averaged traces taken before (1), 20 min after starting NRG1 application (2), and 50 min after tetanus stimuluation (3). (B) BMI prevented NRG1 from inhibiting LTP. Normalized fEPSP slopes were plotted as in A. BMI (20 μM) was bath-applied 20 min before the application of NRG1. (Right) Representative traces. (C) PTX prevented NRG1 from inhibiting LTP. (D) Dose-dependent inhibition by BMI. (E) Ecto-ErbB4 increases LTP, and this effect is blocked by BMI. Brain slices were treated with ecto-ErbB4 30 min prior to LTP induction. (Left) Normalized fEPSP slopes of control or slices treated with 1 μg/mL ecto-ErbB4, 20 μM BMI, ecto-ErbB4 plus BMI, or denatured ecto-ErbB4. (Right) Quantitative analysis of data (n = 5–8 slices). *P < 0.05; **P < 0.01.

Fig. 4.

Ablation of ErbB4 in PV-positive neurons prevents NRG1 from increasing GABAergic transmission and suppressing LTP. (A) Representative eIPSC traces from control and PV-Cre;ErbB4^−/− mice. (Right) Quantitative analysis of data with eIPSCs before application of 100% NRG1 (n = 9 for control; n = 8 for PV-Cre;ErbB4^−/− mice). (B) Normalized fEPSP slopes from control and NRG1-treated hippocampal slices of indicated mice. (Right) Quantitative data (n = 4–6 slices). (C) NRG1 regulation of eIPSCs in CaMKII-Cre;ErbB4^−/− mice. eIPSCs were recorded from CaMKII-Cre;ErbB4^−/− mice (n = 9) or control littermates (n = 5). (D) NRG1 suppression of LTP induction in CaMKII-Cre;ErbB4^−/− mice. Shown was % potentiation of fEPSP slopes that were recorded as Fig. 1 (n = 4–5 slices). *P < 0.05; **P < 0.01.

Fig. 5.

Impaired contextual fear conditioning in PV-Cre;ErbB4^−/− but not CaMKII-Cre;ErbB4^−/− mice. (A) Memory for contextual training measured 24 h after training was reduced in PV-Cre;ErbB4^−/− mice (n = 6) compared with control littermates (n = 7). (B) Memory for contextual training measured 24 h after training was not changed in CaMKII-Cre;ErbB4^−/− mice (n = 6) compared with control littermates (n = 6). *P < 0.05.