Reversing anxiety by targeting a stress-responsive signaling pathway

Abstract

Current treatments of anxiety and depressive disorders are plagued by considerable side effects and limited efficacies, underscoring the need for additional molecular targets that can be leveraged to improve medications. Here, we have identified a molecular cascade triggered by chronic stress that exacerbates anxiety- and depressive-like behaviors. Specifically, chronic stress enhances Src kinase activity and tyrosine phosphorylation of calmodulin, which diminishes MyosinVa (MyoVa) interaction with Neuroligin2 (NL2), resulting in decreased inhibitory transmission and heightened anxiety-like behaviors. Importantly, pharmacological inhibition of Src reinstates inhibitory synaptic deficits and effectively reverses heightened anxiety-like behaviors in chronically stressed mice, a process requiring the MyoVa-NL2 interaction. These data demonstrate the reversibility of anxiety- and depressive-like phenotypes at both molecular and behavioral levels and uncover a therapeutic target for anxiety and depressive disorders.

Keywords: GABAergic synapses; Neuroligin2; Src kinase; anxiety; chronic stress.

Fig. 1.

Chronic stress (RS) increases anxiety and strongly suppresses GABAergic transmission. (A) Schematic representation of the RS protocol in mice. (B) RS mice spent significantly less time and made fewer entries in the open arm of the EPM than the Control (n = 12 for both Control and RS; Student's t test). (C) RS mice buried a significantly high number of marbles with strongly reduced latency to bury the first marble in the MBT compared to the Control (n = 12 for both Control and RS; Student's t test). (D) RS mice showed significantly higher immobility and lower mobility in the last 4 min of the FST than Control (n = 12 for both Control and RS; Student's t test). (E) RS did not change the total expression of indicated excitatory and inhibitory synaptic proteins in the hippocampal lysates of the RS mice compared to the Control (n = 3; two-way ANOVA test). (F) Expression of indicated inhibitory but not the excitatory synaptic proteins was significantly reduced in the synaptosomal fractions of the RS mice as compared to the Control (n = 3; two-way ANOVA test). (G) mIPSC frequency, but not amplitude, was significantly reduced in CA1 pyramidal neurons in acute hippocampal slices prepared from RS mice as compared to the Control (n = 14 for both Control and RS; Student's t test). For cumulative distributions, the Kolmogorov–Smirnov test was used. (Scale bar, 20 pA and 1 s.) Error bars indicate SEM. ****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05.

Fig. 2.

Chronic stress disrupts the NL2 proteome in the mouse hippocampus. (A) Representative proteins that were associated with NL2 in mass spectroscopy analysis. NL2 was not detected in IgG Control samples, and the ratio of 100 is the default number for maximum fold change allowed in Proteome Discoverer software. (B) Indicated proteins were coimmunoprecipitated with NL2 from the hippocampal lysates of 7- to 9-wk-old WT mice. IP, coimmunoprecipitation; IB, immunoblot; IgG, immunoglobulin G. (n = 3). (C) Schematic representation of the RS protocol in mice. (D) In the hippocampal lysates of the RS mice, coimmunoprecipitation with NL2 revealed a notable increase in the association between NL2 and Src, along with a marked reduction in the interactions between NL2 and MyoVa or CaM, in comparison to the Control. (n = 3; two-way ANOVA test). (E) Active Src and tyrosine phosphorylated CaM (pCaM) but not the total Src or total CaM expressions were significantly increased in the hippocampal lysates of RS mice as compared to the Control (n = 3; two-way ANOVA test). (F) Surface but not total NL2 expression was significantly decreased in hippocampal neurons expressing CA Src (constitutively active Src); however, WT Src or KD Src (kinase-dead Src) expression has no significant effect on the surface and total NL2 expression compared to the Control [Control, n = 14; GFP, n = 15; WT Src, n = 14; CA Src, n = 16; KD Src, n = 18; one-way ANOVA test (Scale bar, 5 μm.)] (G) mIPSC frequency but not amplitude was significantly reduced in hippocampal neurons expressing CA Src but not in cells expressing either WT Src or KD Src constructs as compared to the Control (Control, n = 16; GFP, n = 14; WT Src, n = 16; CA Src, n = 15; KD Src, n = 16; one-way ANOVA test). For cumulative distributions, the Kolmogorov–Smirnov test was used. (Scale bar, 20 pA and 1 s.) Error bars indicate SEM. ****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05.

Fig. 3.

Characterization of the chronic stress-sensitive signaling pathway. (A) Surface but not total NL2 expression was significantly reduced in hippocampal neurons expressing CaM YY/EE, while neurons expressing CaM WT or CaM YY/FF showed no significant difference in surface and total NL2 expression compared to the Control. [Control, n = 29; CaM WT, n = 20; CaM YY/EE, n = 30; and CaM YY/FF, n = 25; one-way ANOVA test (Scale bar, 5 μm.)] (B) mIPSC frequency but not amplitude was significantly reduced in neurons expressing CaM YY/EE, while CaM WT or CaM YY/FF overexpression showed no significant alterations in mIPSC frequency or amplitude compared to the Control. (Control, n = 28; CaM WT, n = 22; CaM YY/EE, n = 33; CaM YY/FF, n = 26; one-way ANOVA test, and the Kolmogorov–Smirnov test was used for cumulative distributions). (Scale bar, 20 pA and 1 s.) (C) Schematic of the CaM gRNA construct (CaM triple gRNA) used to knock out endogenous CaM 1, 2, and 3 from mouse hippocampal neurons. mIPSC frequency but not amplitude was significantly reduced in hippocampal neurons expressing CaM triple gRNA. However, coexpression of CaM WT* or CaM YY/FF* with the CaM triple gRNA led to a recovery in mIPSC frequency, whereas coexpression with CaM YY/EE* did not result in a recovery of mIPSC frequency. (Control, n = 15; CaM Triple gRNA, n = 15; CaM Triple gRNA + CaM WT*, n = 16; CaM Triple gRNA + CaM YY/EE*, n = 15; CaM Triple gRNA + CaM YY/FF*, n = 15; one-way ANOVA test, and the Kolmogorov–Smirnov test was applied for cumulative distributions). *Indicates gRNA-resistant CaM construct. (Scale bar, 20 pA, and 1 s.) (D) Surface but not total NL2 expression was significantly reduced in neurons expressing the DN MyoVa construct, as compared to the Control [Control, n = 27; DN MyoVa, n = 28; Student's t test (Scale bar, 5 μm.)] DN, dominant negative. (E) mIPSC frequency but not amplitude was significantly reduced in hippocampal neurons expressing DN MyoVa compared to the Control (Control, n = 14; DN MyoVa, n = 15; Student's t test; for cumulative distributions, the Kolmogorov–Smirnov test was used). (Scale bar, 20 pA, and 1 s.) Error bars indicate SEM. ****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05.

Fig. 4.

Pharmacological manipulation of the stress-sensitive pathway reverses elevated anxiety phenotypes in RS mice. (A) 1 and 2: Schematic experimental procedures of the RS and vehicle (Veh.), PP2, or PP3 administration in WT mice to perform experiments. (B) RS mice spent significantly less time and made very fewer entries in the open arm of the EPM, as compared to the Control; however, after PP2 injections, RS mice showed a comparable number of entries and time spent in the open arms as relative to the Control (Control + Veh., n = 18; RS + Veh., n = 20; Control + PP2, n = 18; RS + PP2, n = 18; two-way ANOVA test). (C) RS mice buried significantly more marbles and had strongly reduced latency to bury the first marble in the MBT, and importantly, upon PP2 administration, the marble burying and latencies were ameliorated in PP2 treated RS mice as compared to the Control (Control + Veh., n = 18; RS + Veh., n = 20; Control + PP2, n = 18; RS + PP2, n = 18; two-way ANOVA test). (D) RS mice showed significantly higher immobility and lower mobility times in the last 4 min of the FST as compared to the Control, and the phenotype was rescued upon PP2 administrations in RS mice (Control + Veh., n = 18; RS + Veh., n = 20; Control + PP2, n = 18; RS + PP2, n = 18; two-way ANOVA test). (E) Expression of active Src and tyrosine pCaM was significantly increased in hippocampal lysates of RS mice as compared to the Control, and upon PP2 treatment, the increased expression was significantly reduced in RS mice (n = 3; two-way ANOVA test). (F) Expression of inhibitory synaptic proteins (NL2, gephyrin, vGAT, and α2) but not PSD95 was strongly reduced in hippocampal synaptosomal fractions of RS mice as compared to the Control, and importantly, PP2 administration rescued expression of inhibitory synaptic proteins in RS mice. (n = 3; two-way ANOVA test). (G) mIPSCs frequency but not amplitude was strongly reduced in CA1 neurons in acute hippocampal slices prepared from RS mice as compared to the Control, and PP2 administration rescued the deficits in mIPSC frequency in RS mice [Control + Veh., n = 22; RS + Veh., n = 30; Control + PP2, n = 21; RS + PP2, n = 28; two-way ANOVA test, and the Kolmogorov–Smirnov test was used for cumulative distributions (Scale bar; 20 pA, and 1 s.)] Error bars indicate SEM. ****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05.

Fig. 5.

Genetic disruption of the MyoVa–NL2 interaction reduces synaptic expression of NL2 and increases anxiety. (A) MyoVa was coimmunoprecipitated with NL2 in WT, but not NL2ΔPDDVP, hippocampal lysates. (n = 4, IP, immunoprecipitation; IB, immunoblot; IgG, immunoglobulin G). (B) Expression of inhibitory synaptic proteins (vGAT, gephyrin, NL2, and α2), but not excitatory synaptic protein PSD95, was strongly reduced in hippocampal synaptosomal fractions prepared from NL2ΔPDDVP KI mice as compared to Control mice (n = 3; two-way ANOVA test). (C) Surface but not total NL2 staining was strongly reduced in cultured hippocampal neurons prepared from NL2ΔPDDVP mice as compared to Control hippocampal neurons [WT Control, n = 12; NL2ΔPDDVP, n = 14; Student's t test (Scale bar, 5 μm.)] (D) Immunohistochemical images from hippocampal CA1 regions showed that the punctual densities of NL2, vGAT, and gephyrin were strongly reduced in NL2ΔPDDVP mice, as compared to the Control (n = 3; Student's t test). (E) mIPSC frequency and amplitude were significantly reduced in CA1 neurons in acute hippocampal slices prepared from NL2ΔPDDVP mice as compared to the Control [WT Control, n = 14; NL2ΔPDDVP, n = 15; Student's t test; the Kolmogorov–Smirnov test was used for the cumulative distributions (Scale bar, 20 pA, and 1 s.)] (F) NL2ΔPDDVP mice spent significantly less time and made fewer entries in the open arm of the EPM as compared to the Control (n = 12 for both WT control and NL2ΔPDDVP; Student's t test). (G) Schematic representation of the RS protocol performed in NL2ΔPDDVP mice. (H) RS did not exacerbate anxiety-like behavior in NL2ΔPDDVP mice, as evidenced by similar time spent and entries made in the open arms of the EPM as compared to the Control (n = 6 for both conditions; Student's t test). Error bars indicate SEM. ****P < 0.0001; ***P < 0.001; **P < 0.01.

Fig. 6.

PP2 administration in NL2ΔPDDVP mice did not lead to the recovery of the enhanced anxiety-like phenotypes observed in these animals. (A) Schematic of the experimental procedure. (B) NL2ΔPDDVP mice spent significantly less time and fewer entries in the open arm of the EPM compared to the Control, and importantly, PP2 administration did not rescue the phenotype in NL2ΔPDDVP mice (n = 12 for every condition; two-way ANOVA test). (C) NL2ΔPDDVP mice buried a significantly high number of marbles with strongly reduced latency to bury the first marble in the MBT as compared to the Control. However, PP2 administration did not affect this phenotype in NL2ΔPDDVP mice (n = 12 for each condition; two-way ANOVA test). (D) NL2ΔPDDVP mice were significantly less mobile and had significantly high mobility in the last 4 min of the FST, and PP2 administration did not improve this behavioral phenotype in these animals compared to the Control (n = 12 for every condition; two-way ANOVA test). Error bars indicate SEM. ****P < 0.0001; ***P < 0.001.

Fig. 7.

Schematic model illustrating the effect of RS and PP2 on the stress-sensitive Src-CaM-MyoVa-NL2 pathway in WT and NL2ΔPDDVP mice. (A) The schematic model shows that RS triggers a substantial increase in active Src activity in WT mice. This, in turn, leads to a significant increase of tyrosine phosphorylation of CaM, resulting in disruption of MyoVa-dependent NL2 surface trafficking. As a consequence, there is a reduction in GABAergic synaptic density and inhibitory synaptic transmission, leading to heightened anxiety-like behaviors in WT mice. Importantly, PP2 administration effectively ameliorates molecular and behavioral deficits induced by RS in WT mice. (B) Genetic ablation of the domain in the NL2 C terminus (PDDVP) that mediates the interaction with MyoVa in NL2ΔPDDVP mice reduces NL2 surface expression, resulting in decreased GABA_AR synaptic abundance and inhibitory synaptic transmission, leading to an enhanced anxiety-like phenotype. Importantly, RS does not exacerbate anxiety-like behaviors, and PP2 administration fails to rescue the deficits at the molecular and behavioral levels in these NL2ΔPDDVP mice.