Reversal of behavioral deficits and synaptic dysfunction in mice overexpressing neuregulin 1

Abstract

Neuregulin 1 (Nrg1) is a susceptibility gene of schizophrenia, a disabling mental illness that affects 1% of the general population. Here, we show that ctoNrg1 mice, which mimic high levels of NRG1 observed in forebrain regions of schizophrenic patients, exhibit behavioral deficits and hypofunction of glutamatergic and GABAergic pathways. Intriguingly, these deficits were diminished when NRG1 expression returned to normal in adult mice, suggesting that damage which occurred during development is recoverable. Conversely, increase of NRG1 in adulthood was sufficient to cause glutamatergic impairment and behavioral deficits. We found that the glutamatergic impairment by NRG1 overexpression required LIM domain kinase 1 (LIMK1), which was activated in mutant mice, identifying a pathological mechanism. These observations demonstrate that synaptic dysfunction and behavioral deficits in ctoNrg1 mice require continuous NRG1 abnormality in adulthood, suggesting that relevant schizophrenia may benefit from therapeutic intervention to restore NRG1 signaling.

Figure 1. Temporal control of NRG1 expression in forebrains of ctoNrg1 mice

(A) Transgene structure. Full length NRG1 type I β1a was cloned in pMM400 between the promoter complex of TRE and CMV (cytomegalus virus minimal promoter) and SV40 polyadenylation signal. (B) Genotypes and NRG1 expression. Unless otherwise indicated, blank, red and blue histograms/curves in the paper represent data from control, ctoNrg1, and aDox-treated ctoNrg1 mice, respectively. aDox, Dox treatment in adulthood; dDox, Dox treatment during development (see Figure 7); o/e, overexpression. (C) Expression of NRG1 transgene in PFC, hippocampus (HPF), and striatum (STR), but not midbrain (MB) and cerebellum (CB). ctoNrg1 mice, 8 weeks of age, were treated with Dox for two weeks. Homogenates of various brain regions of indicated mice were subjected to western blot analysis with indicated antibodies. (D) Quantification of NRG1 expression in different brain regions. Expression in ctoNrg1 samples were normalized by those from control littermates. n = 3 per genotype; *** p < 0.001 for PFC, HPF and STR, one-way ANOVA. (E) HA staining in ctoNrg1 PFC. Sections were stained with anti-HA antibody and DAPI. Bar, 50 μm. (F) HA staining in ctoNrg1 hippocampus. Sections were stained with anti-HA antibody. Bar, 100 μm. (G, H) NRG1 expression of developing forebrains. Top, representative western blots; bottom, quantitative analysis. n = 3 per genotype at each time point. Data were normalized by NRG1 levels of E18 control mice (in G) or of P7 control mice (in H). In G, ** Genotype F (1, 20) = 184, p < 0.01, two-way ANOVA. In H, ** Genotype F (1, 16) = 33.03, p < 0.01, two-way ANOVA.

Figure 2. Behavioral deficits in ctoNrg1 mice were attenuated after aDox treatment

(A) Diagram of aDox treatment and behavioral test. Dox was added in the drinking water when mice were 8 weeks old and during entire behavioral tests, which started two weeks after treatment. (B) Travel distance in open field was increased in ctoNrg1 mice, compared to controls, but became similar between the genotypes after aDox treatment. n = 12 per group; * p < 0.05, # p > 0.05, one-way ANOVA. Blank, red, grey and blue histograms/curves represent data from control, ctoNrg1, aDox-treated control and aDox-treated ctoNrg1 mice. (C) Elevated response to 70dB background noise in ctoNrg1 mice, compared to controls, and the elevation was ameliorated by aDox treatment. AU, arbitrary units. n = 9 per group; * p < 0.05, # p > 0.05, one-way ANOVA. (D) PPI was impaired in ctoNrg1 mice, compared to controls, and the impairment was diminished after aDox treatment. n = 9 per group; * Genotype F (3, 84) = 10.2, p < 0.05, two-way ANOVA. (E) ctoNrg1 mice spent less time with social cylinder with stimulus mouse, compared to controls, and the interaction time was similar between two genotypes after aDox treatment. n = 12 per group; * Genotype F (3, 172) = 7.01, p < 0.05, two-way ANOVA. (F) Social interaction time with stimulus mice without cylinder was reduced in ctoNrg1 mice, and the reduction was diminished after aDox treatment. n = 12 for each group; * p < 0.05, # p > 0.05, one-way ANOVA. (G) Decreased correct entries in ctoNrg1 mice, compared to controls, and the difference was diminished after aDox treatment. n = 12 per group; ** Genotype F (3, 424) = 16.14, p < 0.01, two-way ANOVA. (H) Increased latency for ctoNrg1 mice to reach the hidden platform in MWM, compared to controls, and the two genotypes showed similar latency after aDox treatment. n = 9 for control, n = 8 for ctoNrg1, n = 8 for both groups of control + aDox and ctoNrg1 + aDox; * Genotype F (3, 87) = 6.4, p <0.05, two-way ANOVA. (I) Reduced platform crossing by ctoNrg1 mice in MWM probe test, compared to controls, and the two genotypes showed similar platform crossing after aDox treatment. n = 9 for control, n = 8 for ctoNrg1, n = 8 for both groups of control + aDox and ctoNrg1 + aDox; * p < 0.05, # p > 0.05, one-way ANOVA.

Figure 3. Glutamatergic and GABAergic impairment in ctoNrg1 mice was reduced after aDox treatment

(A) Diagram of aDox treatment and behavioral test. Mice were fed with Dox-containing drinking water at 8 weeks old for 2 weeks before recording. Dox was kept in the slice perfusion solution. (B) Depressed I/O curves in ctoNrg1 mice, compared to controls, and I/O curves were similar between the genotypes after aDox treatment. Left, representative traces of fEPSPs at SC-CA1 synapses at different stimulus intensities; right, I/O curves of four groups. n = 10 slices from 6 control mice, n = 10 slices from 6 ctoNrg1 mice, n = 9 slices from 4 aDox-treated control mice, n = 8 slices from 4 aDox-treated ctoNrg1 mice; ** Genotype F (3, 429) = 19.87, p < 0.01, two-way ANOVA. (C) Reduced mEPSC frequency in ctoNrg1 CA1 pyramidal neurons, compared to controls and mEPSC frequency was similar between the two genotypes after aDox treatment. Left, representative mEPSC traces; right, quantitative data. n = 10 cells from 4 mice per group; ** p < 0.01, # p > 0.05, one-way ANOVA. (D) Elevated PPF in ctoNrg1 SC-CA1 synapses, compared to controls and the PPF was similar between the two genotypes after aDox treatment. Top, representative traces; bottom, quantitative data. n = 10 slices from 6 control mice, n = 10 slices from 6 ctoNrg1 mice, n = 8 slices from 4 aDox-treated control mice, n = 8 slices from 4 aDox-treated ctoNrg1 mice; *** Genotype F (3, 201) = 29.76, p < 0.001, two-way ANOVA. (E) Attenuated mIPSC amplitudes in ctoNrg1 CA1 pyramidal neurons, compared to controls and mIPSC amplitudes were similar between the two genotypes after aDox treatment. Left, representative mIPSC traces; right, quantitative data. n = 11 cells from 5 control mice, n = 10 cells from 4 ctoNrg1 mice, n = 11 cells from 5 aDox-treated control mice, n = 10 cells from 4 aDox-treated ctoNrg1 mice; * p < 0.05, # p > 0.05, one-way ANOVA. (F) Reduced GABA_A receptor α1 subunit (GABA_ARα1) in ctoNrg1 forebrain and the reduction was diminished after aDox treatment. Top, representative blots; bottom, quantitative data. Data from other three groups were normalized by control. n = 3 per group; * p < 0.05, # p > 0.05, one-way ANOVA.

Figure 4. No effect of ErbB4 mutation on glutamatergic impairment in ctoNrg1 mice

(A) ErbB4 heterozygous mutation attenuated NRG1-mediated PPR reduction in both control and ctoNrg1 hippocampal slices. Slices were subjected to a pair of stimulation in the absence or presence of NRG1. Left, representative eIPSC traces in ctoNrg1 and ctoNrg1;ErbB4 +/- slices; right, quantitative data. n = 10 cells from 3 mice for each group. ** p < 0.01, one-way ANOVA. (B) No effect of ErbB4 heterozygous mutation on maximal fEPSP slope in ctoNrg1 hippocampal slices. Left, representative I/O curves in control, ErbB4 +/-, ctoNrg1 and ctoNrg1;ErbB4 +/- slices; right, quantitative data. n = 12 slices from 3 mice for each group. ** p < 0.01, * p < 0.05, # p > 0.05, one-way ANOVA. (C) Reduced mEPSC frequency in ctoNrg1 CA1 pyramidal neurons was not altered by ErbB4 heterozygous mutation. Left, representative mEPSC traces in control, ErbB4 +/-, ctoNrg1 and ctoNrg1; ErbB4 +/- slices; right, quantitative data. n = 10 cells from 3 mice for each group. *** p < 0.001, # p > 0.05, one-way ANOVA. (D) ErbB4 heterozygous mutation had no effect on elevated PPF at ctoNrg1 SC-CA1 synapses. Top, representative traces in control, ErbB4 +/-, ctoNrg1 and ctoNrg1; ErbB4 +/- slices; right, quantitative data at 20 ms interval, n = 12 slices from 3 mice for each group. ** p < 0.01, # p > 0.05, one-way ANOVA.

Figure 5. Synaptic recruitment and activation of LIMK1 by overexpressed NRG1

(A) NRG enrichment in P2 and synaptosomes in ctoNrg1 forebrains. Subcellular fractions were subjected to western blotting with indicated antibodies. Left, representative blots; right, quantitative data where NRG1 levels were normalized by synaptagmin1 (Syt1), and those in controls were taken as 1. n = 3 per genotype; *** Genotype F (1, 8) = 124.48, p < 0.001, two-way ANOVA. N, nucleus; C, cytoplasm; Syn, synaptosomes. (B) Similar levels of LIMK1 in control and ctoNrg1 forebrains. Homogenates of individual control (lanes 1-4) and ctoNrg1 (5-8) mice were subjected to western blotting with antibodies against LIMK1 and α-tubulin. Left, representative blots; right, quantitative data where LIMK1 was normalized by α-tubulin. n = 4 per genotype; p > 0.05, t-test. (C) Increased LIMK1 in P2 and synaptosomes of ctoNrg1 forebrains. Left, representative blots; right, quantitative data where LIMK1 in cytoplasm and P2 or synaptosomes were normalized by tubulin and Syt1, respectively. Levels in control mice were taken as 1. n = 3 per genotype; ** Genotype F (1, 12) = 17.73, p < 0.01, two-way ANOVA. (D) Increased p-cofilin in P2 and synaptosomes of ctoNrg1 mice. Left, representative blots; right, quantitative data where p-cofilin was normalized by total cofilin. Data from controls were taken as 1. n = 3 per genotype; *** Genotype F (1, 8) = 26.68, p < 0.001, two-way ANOVA. (E) Similar levels of LIMK1 and p-cofilin in P2 and Syn in aDox-treated control and ctoNrg1 forebrains. Left, representative blots; right, quantitative data where LIMK1 in cytoplasm was normalized by tubulin and that in P2 and synaptosomes by Syt1, p-cofilin was normalized by total cofilin; and those in controls were taken as 1; n = 3 per genotype; For LIMK1, Genotype F (1, 8) = 0.26, p > 0.05, two-way ANOVA; for p-cofilin, Genotype F (1, 8) = 0.00, p > 0.05, two-way ANOVA. (F) HA-NRG1 was not detectable in forebrains of aDox-treated mice. Left, representative blots; right, quantitative data where NRG1 levels were normalized by Syt1, and those in controls were taken as 1; n = 3 per genotype; Genotype F (1, 8) = 1.57, p > 0.05, two-way ANOVA.

Figure 6. Inhibition of LIMK1 ameliorated glutamatergic impairment in ctoNrg1 slices

(A) Amino acid sequences of LIMK1 control (Rev-11R) and inhibitory peptide (S3-11R). (B) Penetration of FITC-labeled S3-11R into neurons of hippocampal slices. Bar, 20 μm. (C) Treatment with S3-11R, but not Rev-11R, attenuated mEPSC frequency reduction in ctoNrg1 CA1 pyramidal neurons. Hippocampal slices were treated with 1 μM S3-11R and Rev-11R, respectively. Left, representative traces; right, quantitative data. Rev-11R: n = 8 cells from 3 mice per genotype, S3-11R: n = 8 cells from 3 control mice, n = 9 cells from 3 ctoNrg1 mice; * p < 0.05, # p > 0.05, one-way ANOVA. (D) S3-11R, but not Rev-11R, reduced PPF at SC-CA1 synapses in ctoNrg1 mice. Slices were treated as in C. Top, representative traces; bottom, quantitative data. Rev-11R: n = 9 slices from 3 control mice, n = 10 slices from 3 ctoNrg1 mice, S3-11R: n = 8 slices from 3 mice per genotype; ** Genotype F (3, 155) = 88.07, p < 0.01, two-way ANOVA. (E) Partial recovery of I/O curves at ctoNrg1 SC-CA1 synapses by S3-11R. Top, representative traces; bottom, quantitative data. Rev-11R: n = 9 slices from 3 control mice, n = 10 slices from 3 ctoNrg1 mice, S3-11R: n = 8 slices from 3 control mice, n = 9 slices from 3 ctoNrg1 mice; ** Genotype F (3, 416) = 31.32, p < 0.01, two-way ANOVA.

Figure 7. Adult NRG1 overexpression is sufficient to cause glutamatergic impairment and behavioral deficits

(A) Diagram of dDox treatment and tests. Mice were fed with Dox-containing drinking water from P0 till 8 weeks of age, when Dox was omitted. Three weeks later, mice were subjected to analysis. Such treatment was designated as dDox for Dox treatment during development. (B) Expression of the HA-NRG1 transgene in forebrains was not detectable after P3 in mice being treated by dDox. (C) HA-NRG1 was not detectable in forebrains of dDox-treated mice at 8 weeks of age, but at 11 weeks of age (i.e., 3 weeks after dDox withdrawal). (D) Depressed I/O curves in dDox-treated ctoNrg1 mice, compared to dDox-treated controls. Top, representative traces; bottom, quantitative data; n = 12 slices from 4 dDox-treated control mice, n = 11 slices from 5 dDox-treated ctoNrg1 mice; ** Genotype F (1, 273) = 40.77, p < 0.01, two-way ANOVA. Blank and yellow histograms/curves represent data from dDox-treated control and ctoNrg1 mice, respectively. (E) Decreased mEPSC frequency, but not amplitude, in CA1 pyramidal neurons of dDox-treated ctoNrg1 mice, compared to controls. Top, representative traces; bottom, quantitative data; n = 9 cells from 3 dDox-treated control mice; and n = 10 cells from 4 dDox-treated ctoNrg1 mice; ** p < 0.01, t-test. (F) Elevated PPF at SC-CA1 synapses in dDox-treated ctoNrg1 mice, compared to controls. Left, representative traces; right, quantitative data; n = 10 slices from 4 dDox-treated control mice; n = 10 slices from 5 dDox-treated ctoNrg1 mice; ** Genotype F (1, 90) = 70.21, p < 0.01, two-way ANOVA. (G) Normal mIPSC in dDox-treated ctoNrg1 mice, compared to controls. Left, representative traces; right, quantitative data; n = 10 cells from 4 dDox-treated control mice; n = 12 cells from 4 dDox-treated ctoNrg1 mice; p > 0.05, t-test. (H) Increased travel distance by dDox-treated ctoNrg1 mice in open field test for 30 min, compared to controls. n = 10 per genotype; * p < 0.05, t-test. (I) Reduced PPI in dDox-treated ctoNrg1 mice, compared to controls. n = 10 per genotype; * Genotype F (1, 51) = 16.06, p < 0.05, two-way ANOVA. (J) Increased latency of dDox-treated ctoNrg1 mice to reach the hidden platform in MWM, compared to controls. n = 10 per genotype; * Genotype F (1, 68) = 13, p < 0.05, two-way ANOVA. (K) Reduced platform crossing by dDox-treated ctoNrg1 mice in MWM probe test, compared to controls. n = 10 per genotype; * p < 0.05, t-test.