Introduction

Major depressive disorder, affecting more than 120 million people worldwide every year, is a heterogeneous illness influenced by a variety of factors, including environmental stress-related factors (Duman et al, 2016). Subjects with repeated exposure to stressors have an elevated risk of depressive disorders (Wohleb et al, 2016). Although currently available antidepressant medications are the mainstay of treatment for patients with major depressive disorder, up to 60% of patients do not achieve adequate response following antidepressant therapy (Fava, 2003). This poor outcome is because of multiple subtypes and complex causes of major depressive disorders (Gerhard et al, 2016). Although neurochemical theory-based mechanisms have been intensively explored during the past several decades, neuroplasticity theory-oriented neurobiological mechanisms of depressive disorders remain elusive (Gerhard et al, 2016; Krystal et al, 2013).

The midbrain periaqueductal gray (PAG) is believed to be an essential part of the circuitry that participates in stress-associated behavior, such as defensive (Penzo et al, 2014; Tovote et al, 2016) and antinociceptive (Ho et al, 2013, 2015; Ho et al, 2011) behavior in response to threat. It integrates several inputs from various brain regions, such as the amygdala (Penzo et al, 2014), and sends excitatory outputs to the premotor targets in the medulla. The ventrolateral periaqueductal gray (vlPAG), a subdivision of the PAG, has been shown to be involved in not only defensive behavior such as freezing, flight, and analgesia (Ho et al, 2013, 2015; Ho et al, 2011; Tovote et al, 2016), but also stress-induced depression-like behavior (Johnson et al, 2016). A previous study demonstrated that activation of the vlPAG by electric stimulation initiates the expression of freezing behavior (Vianna et al, 2001). Recent reports indicated that the vlPAG receives direct anatomical and functional inputs originating from the central nucleus of the amygdala to control stress-induced maladaptive responses (Penzo et al, 2014). The vlPAG is, therefore, a critical core for stress-associated maladaptive behavior, including neuroplastic changes observed in chronic stress-induced psychiatric disorders, and for studying their underlying cellular mechanisms.

Evidence for the dysregulation of glutamatergic neurotransmission in patients with depression indicates that glutamate-mediated excitatory system is crucially involved in depressive disorders (Auer et al, 2000). It has been considered that dysfunction in the glutamatergic system in depression is related to the loss of postsynaptic-associated α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors (Wei et al, 2016; Yuen et al, 2012) as well as reduction in presynaptic glutamate release in the prefrontal cortex (PFC) (Yuen et al, 2012). A previous study reported that repeated stress diminishes not only presynaptic glutamate release but also postsynaptic AMPA receptor expression in the PFC (Yuen et al, 2012), resulting in chronic stress triggered-cognitive impairment. Moreover, another recent study demonstrated that chronic stress specifically decreases AMPA receptor-mediated neuroplasticity in the hippocampus, thereby contributing to depression (Kallarackal et al, 2013). However, it is still largely unclear whether the vlPAG is involved in chronic stress-elicited depression-like helplessness responses. In the current study, we sought to explore the functional alterations and specific targets in the vlPAG and to reveal its potential role in chronic stress-induced depressive behavior.

Materials and methods

Animal Preparations

Adult male Sprague-Dawley rats weighing 200–250 g were used. They were housed at room temperature (23±1 °C) with a 12 h light/dark cycle (lights on from 0800 to 2000 h) and were provided food and water ad libitum. All experimental procedures were performed in accordance with the policies and recommendations of the International Association for the Study of Pain and approved by the institutional animal care and use committee of Taipei Medical University, Taipei, Taiwan. All efforts were made to minimize the number of animals used.

Establishment of the Learned Helplessness Model

A well-established model of learned helplessness (LH) paradigm was used in the present study as described in previous studies (Hajszan et al, 2009) and are detailed in Supplementary Information.

Sucrose Preference Test (SPT)

Anhedonic behavior in rats was evaluated using the sucrose preference test and are detailed in Supplementary Information.

Tail Suspension Test (TST) and Forced Swim Test (FST)

Behavioral despair in rats was assayed using the tail suspension test and forced swim test and are detailed in Supplementary Information.

Locomotor Activity and Nociceptive Behavioral Test

Locomotor activity in rats was measured using the open field test. Nociceptive behavioral test in rats was assayed using the von Frey filament test and plantar test and are detailed in Supplementary Information.

Whole-Cell Patch-Clamp Recordings

Electrophysiological recordings were described as we published previously (Ho et al, 2011, 2013, 2015) and are detailed in Supplementary Information. All neurons were recorded from the ventrolateral subregion of PAG. Recordings were conducted blind to the rat groups.

Western Blotting

Western blot analyses of glutamate receptor 1 (GluR1) and phosphorylated GluR1 (pGluR1) were conducted on protein samples from the plasma membrane and subcellular fractions of rat vlPAG and are detailed in Supplementary Information.

Immunofluorescence

Immunofluorescence analyses of GluR1, pGluR1, and NeuN expression were performed on the vlPAG-containing slices and are detailed in Supplementary Information.

Intra-vlPAG Microinjection

Intra-vlPAG microinjection was performed as reported previously (Ho et al, 2015) and are detailed in Supplementary Information.

Reagents and Vehicles

See Supplementary Information.

Statistical Analyses

Data are expressed as mean±SEM. One-way analyses of variance (ANOVAs) were used to compare the results of electrophysiological recording data from three groups followed by post hoc Tukey’s multiple comparison tests. A two-way ANOVA was used to analyze the current–voltage curve from different holding potential and firing rate, with the post hoc Bonferroni’s multiple comparison test. Unpaired Student’s t-tests were used for analyses between two groups. Differences were considered significant if p<0.05 was observed.

Results

Chronic Stress Elicits LH Depression-Like Behavior

To investigate the role of the vlPAG in mediating behavioral stress responses, we used rats that were subjected repeatedly to inescapable and unpredictable foot shock stress (Figures 1a and b) (Hajszan et al, 2009). After 6 days of the training procedure, rats exhibited high failure rate (Figure 1c; unpaired t-test, p<0.01) and prolonged escape latency (Figure 1d; unpaired t-test, p<0.01) from escapable foot shocks, a well-established depression-like phenotype. This suggests LH was successfully established by chronic inescapable and unpredictable stress (Li et al, 2011; Maier, 1984). In addition to the reduction in escape responses, LH rats also displayed an increase in immobile time during the TST (Figure 1e; unpaired t-test, p<0.01) and a reduction in sucrose consumption (Figure 1f; unpaired t-test, p<0.01), an indication of dispirited behavior and anhedonia behavior, respectively. Furthermore, LH rats exhibited reductions in paw withdrawal threshold during the von Frey test (Supplementary Figure S1a; unpaired t-test, p<0.01) and in paw withdrawal latency during plantar test (Supplementary Figure S1b; unpaired t-test, p<0.01). The above-mentioned results show that rats that received LH training successfully exhibit depression-like behavior and behavioral hypersensitivity.

Figure 1
figure 1

Inescapable unpredictable foot shock stress elicits helplessness behavior and depression-like behavior. (a) Schematic experimental procedure of learned helplessness (LH) model. (b) Training and testing protocol of LH model. (c, d) Vertical scatterplots depicting the distribution of helpless behaviors, number of failure (c) and escape latency (d), from control (n=9) and LH groups (n=16). (e) Vertical scatterplot depicting the distribution of dispirited behavior, immobility time of tail suspension test, from control (n=11) and LH (n=10) groups. (f) Vertical scatterplot depicting the distribution of anhedonic behavior, sucrose consumption of sucrose preference test, from control (n=9) and LH (n=8) groups. Data are expressed as mean±SEM; **p<0.01 vs Control.

PowerPoint slide

Full size image

Chronic Stress Leads to Hypofunction of Glutamatergic Transmission onto PAG Neurons through Pre- and Postsynaptic Mechanisms

The vlPAG has been reported as a crucial midbrain nucleus controlling various forms of stress-related behavior including freezing (Penzo et al, 2014; Tovote et al, 2016) and antinociceptive (Ho et al, 2011, 2013, 2015) responses. Activation of glutamatergic transmission in the vlPAG leads to stress-related antinociception by sending inhibitory outputs to the spinal dorsal horn (Heinricher et al, 2009; Lee et al, 2016; Ossipov et al, 2010). Therefore, we first examined whether inescapable foot shock stress alters glutamatergic transmission onto vlPAG neurons, resulting in depression-like behavior, by conducting acute brain slices electrophysiological recordings. We assessed the current–voltage relationship of excitatory postsynaptic AMPA currents (EPSCAMPAs) in the vlPAG neurons evoked by electrical stimulation at various holding potentials. The amplitudes of EPSCAMPAs of vlPAG neurons in control rats were reduced at positive membrane potentials. In contrast, the EPSCAMPAs recorded from LH rats displayed a near-linear current–voltage relationship (Figure 2a; two-way ANOVA, F7, 63=2.632, p<0.05). This suggests that calcium-permeable AMPA receptor-mediated responses in the vlPAG neurons might participate in chronic stress-induced depression-like behavior. We also found that synaptic strengths in the LH rats were drastically reduced compared with the control rats, revealed by reduction in the EPSCAMPA/NMDA ratio (NMDA: N-methyl-D-aspartate; Supplementary Figure S2; unpaired t-test, p<0.01). Next, to determine whether the decreased AMPA receptor-mediated responses originate from presynaptic, postsynaptic, or both loci, we conducted the paired-pulse ratio (PPR) experiments in both groups. Interestingly, we found that vlPAG neurons from LH rats exhibited higher PPR than did vIPAG neurons from the control group (Figure 2b; unpaired t-test, p<0.01), suggesting that a presynaptic change might modulate glutamatergic transmission induced by LH procedure. Moreover, we further investigated whether postsynaptic AMPA receptors were altered after the LH paradigm. Bath application of AMPA, an agonist of AMPA receptor, produced significantly smaller inward currents in the LH group compared with the control group (Figure 2c; unpaired t-test, p<0.01), implying that downregulation of postsynaptic AMPA receptors might occur in chronic stress-induced depression-like behavior.

Figure 2
figure 2

Chronic stress leads to depression of excitatory synaptic transmission and pGluR1 protein expression in the ventrolateral periaqueductal gray (vlPAG). (a) The current–voltage (I–V) curves of EPSCAMPAs of vlPAG neurons recorded at holding potentials ranging from −70 to +70 mV in control and LH groups. Left inset: a diagram illustrating the positions of the stimulating (S) and recording (R) electrodes, respectively, in the vlPAG slice. Right inset: represented traces of EPSCAMPAs obtained from control (n=5) and LH (n=6) groups. Calibration: 50 pA, 50 ms. (b) Vertical scatterplot depicting the distribution of paired-pulse ratio (PPR) evoked by two identical electric stimuli with 50 ms interpulse interval. Inset: represented traces of PPR obtained from control (n=7) and LH (n=8) groups. Calibration: 50 pA, 50 ms. (c) Vertical scatterplot depicting the distribution of AMPA (1 μM)-induced inward currents in control (n=7) and LH (n=6) groups. Horizontal bars denote the periods that AMPA were applied. The dashed line indicates the baseline. Calibration: 50 pA, 1 min. (d–f) Vertical scatterplots depicting the distribution of total GluR1 (d), surface and cytoplasm (e), and pGluR1 (f) expressions in the vlPAG. Inset: represented western blot of protein expression from control (n=5) and LH (n=5) groups. (g, h) Immunohistochemistry images (g) and quantitative analysis (h) of pGluR1 and NeuN in the vlPAG slices dissected from control (n=5) and LH (n=5) groups. Calibration: 50 μm. Data are expressed as mean±SEM; *p<0.01, **p<0.01 vs Control.

PowerPoint slide

Full size image

Calcium-permeable AMPA receptors are a determinant factor in the generation of some forms of neuroplastic changes in the central nervous system (Park et al, 2016). Dysfunctions of postsynaptic calcium-permeable AMPA receptors, such as GluR1 AMPA receptor, have been evidenced as involved in stress-induced neuropsychiatric disorders (Wei et al, 2016; Yuen et al, 2012). Hence, here we estimated the protein expression of GluR1, the main subunit of calcium-permeable AMPA receptors for excitatory transmission, in the brain. Unexpectedly, we found no significant difference in GluR1 protein expression level (Figure 2d; p=0.306) and number of neurons expressing GluR1 (Supplementary Figure S3; unpaired t-test, p=0.78) in the vlPAG after the LH paradigm. Based on a previous study, this may result from AMPA receptor mobilization between the cell surface and cytoplasm (Shi et al, 1999). Therefore, we isolated the membrane and cytosolic fractions to further investigate the distribution of GluR1 protein expression. We found that LH caused decreased GluR1 expression on the cell surface and increased GluR1 expression in the cytoplasm as compared with the controls (Figure 2e; unpaired t-test, p<0.01). This suggests that redistribution, but not overall increase in the quantity, of GluR1 expression may be caused by LH. A previous report indicated that phosphorylation of GluR1 on the serine 845 site is essential for AMPA receptor mobilization (Lee et al, 2003). Given that, we further identified whether LH reduced surface GluR1 expression by quantification of pGluR1 level. Anticipatedly, reduced pGluR1 protein expression was found in the vlPAG after the LH paradigm (Figure 2f; unpaired t-test, p<0.01). Furthermore, neurons expressing pGluR1 in the vlPAG were observed using immunohistochemical analyses in both groups (Figure 2g). However, reduced pGluR1 expression (Figure 2h; unpaired t-test, p<0.01), but not neuronal numbers (Figure 2h; unpaired t-test, p=0.858), was observed in the LH group, demonstrating that chronic stress may reduce phosphorylation-dependent expression of surface GluR1 in the vlPAG.

Moreover, we studied the passive membrane properties of vlPAG neurons in both groups. Neither membrane capacitance nor resistance was significantly different in the control and LH groups (Supplementary Figure S4). Furthermore, there were no significant differences in the resting membrane potentials in both the control and LH groups (Supplementary Figure S4). These results suggested that LH does not alter the passive membrane properties of vlPAG neurons. In addition, we also examined the neuronal excitability of vlPAG and dlPAG neurons by current injection experiment. We found that vlPAG neurons exhibited fewer spike numbers in the LH group compared with the control group (Supplementary Figures S5a and c; two-way ANOVA, F10, 230=4.353, p<0.01), suggesting that chronic stress reduced vlPAG neuronal excitability. In contrast to vlPAG neurons, dlPAG neurons displayed higher firing frequency in the LH group compared with the control group (Supplementary Figures S5b and d; two-way ANOVA, F10, 160=6.51, p<0.01), demonstrating that chronic stress enhanced dlPAG neuronal excitability. These results demonstrated that chronic stress reduced neuronal excitability in the vlPAG but not dlPAG neurons.

Inhibition of Calcium-Permeable GluR1 AMPA Receptor in the vlPAG Contributes to Depression-Like Behavior

Our results showed that neurons from the control group showed inward rectification, implying that calcium-permeable AMPA receptors, GluR1, might be involved in synaptic EPSCAMPAs (Figure 2a). In contrast, EPSCAMPAs recorded from the LH rats have a characteristic near-linear current–voltage relationship. This suggested that synaptic EPSCAMPAs might be mediated predominantly by GluR1 AMPA receptors. Therefore, to determine whether reduced GluR1 AMPA receptor function in the vlPAG contributes to chronic stress-induced diminished glutamatergic transmission, we used NASPM (Martinez-Rivera et al, 2017; Nishitani et al, 2014), a selective antagonist of calcium-permeable AMPA receptors. Inhibition of glutamatergic transmission by NASPM was higher in neurons from the control group than in neurons from the LH group (Figures 3a and b; unpaired t-test, p<0.01). This suggested that chronic stress decreased the prevalence of GluR1 calcium-permeable AMPA receptors in the vlPAG. Moreover, intra-vlPAG microinjection of NASPM increased TST and decreased SPT in control group (Figure 3c; one-way ANOVA, F3, 20=8.318, p<0.01 and Figure 3d; one-way ANOVA, F3, 20=6.065, p<0.01). This suggested that inhibition of GluR1 calcium-permeable AMPA receptors in the vlPAG contributed to depression-like behavior.

Figure 3
figure 3

Inhibition of calcium-permeable GluR1 AMPA receptor in the vlPAG causes depression-like behavior. (a) Effect of bath application of NASPM (100 μM) on the amplitude of EPSCs in vlPAG neurons dissected from control and LH groups. Horizontal bar denotes the periods that NASPM was applied. Inset: representative traces recorded at the time points denoted by the numbers on the time-course graph. Calibration: 50 pA, 20 ms. (b) Vertical scatterplot depicting the distribution of NASPM-induced EPSC inhibition recorded from vlPAG neurons obtained from control (n=6) and LH (n=6) groups. (c) Vertical scatterplot depicting the distribution of immobility time of tail suspension test, from control (n=6) and LH (n=6) groups before (circle) and after (triangle) intra-vlPAG microinjection of NASPM. (d) Vertical scatterplot depicting the distribution of sucrose consumption of sucrose preference test, from control (n=6) and LH (n=6) groups before (circle) and after (triangle) intra-vlPAG microinjection of NASPM. Inset: cannula tip placements for rats infused with NASPM in control (open circle) and LH (filled circle) groups. Data are expressed as mean±SEM; **p<0.01 vs Control.

PowerPoint slide

Full size image

Chronic Stress Causes Hypofunction of GluR1 in the vlPAG via Glucocorticoid Receptor-Dependent Mechanism

To determine how chronic stress diminishes the functions of GluR1 in repeatedly stressed animals, we examined the involvement of the glucocorticoid receptor (GR) as GR activation initiates the reduction of AMPA receptor function contributing to cognitive impairment (Wei et al, 2016). To block the effects of GR, rats received daily subcutaneous injection of RU486 (10 mg/kg), a GR antagonist, 30 min before the LH paradigm (Figure 4a). Daily GR antagonist injection prevented chronic stress-induced decreased rectification index (Figure 4b; one-way ANOVA, F2, 25=13.78, p<0.01), frequency (Figure 4d; one-way ANOVA, F2, 25=8.22, p<0.01), and amplitude (Figure 4e; one-way ANOVA, F2, 25=22.89, p<0.01) of miniature EPSCs (mEPSCs) in vlPAG neurons. Compared with control group (variance (σ2)=0.15), LH+RU486 (variance (σ2)=0.79) accompanies with much greater variance. Moreover, increased burst-like events were found in the vlPAG neurons after daily GR antagonist injection (baseline frequency: 1.28±0.28 Hz vs burst-like event frequency: 47.24±4.87 Hz, unpaired t-test, p<0.01; Figure 4c, lower panel). In addition, not only surface GluR1 but also pGluR1 expression was enhanced by daily subcutaneous RU486 injection (Figures 4f and g; unpaired t-test, p<0.01). The reversal effects of RU486 were also revealed by immunohistochemical results in vitro (Figures 4h and i; one-way ANOVA, F2, 12=34.08, p<0.01) and behavioral results in vivo including TST, SPT, and FST (Figure 4j; one-way ANOVA, F2, 19=9.906, p<0.01, 4K; one-way ANOVA, F2, 19=25.20, p<0.01, and Supplementary Figure S6; one-way ANOVA, F2, 15=18.64, p<0.01), respectively. These results demonstrated that chronic stress increased endocytosis of postsynaptic GluR1 through GR-dependent mechanism.

Figure 4
figure 4

Chronic stress diminishes excitatory synaptic transmission through both pre- and postsynaptic mechanisms in a glucocorticoid receptor (GR)-dependent manner. (a) Schematic experimental procedure of the effect of glucocorticoid receptor antagonist, RU486, in LH. (b) Rectification index (EPSC−70 mV/EPSC+70 mV) of AMPA receptor obtained from control (n=10), LH (n=11), and LH+RU486 (n=7) groups. Inset: represented traces of rectification index obtained from control and LH groups. Calibration: 50 pA, 20 ms. (c) Representative traces of mEPSCs recorded from vlPAG neurons dissected from control, LH, and LH+RU486 groups. Calibration: 25 pA, 600 ms. (d, e) Vertical scatterplots depicting the distribution of frequency (d) and (e) amplitude of mEPSCs recorded from vlPAG neurons obtained from control (n=9), LH (n=9), and LH+RU486 (n=10) groups. (f, g) Vertical scatterplot depicting the distribution of surface and cytoplasm (f) and pGluR1 (g) expressions in the vlPAG. Inset: represented western blot of protein expression from LH (n=5) and LH+RU486 (n=5) groups. (h, i) Immunohistochemistry images (h) and quantitative analysis (i) of pGluR1 and NeuN in the vlPAG slices dissected from control (n=5), LH (n=5), and LH+RU486 (n=5) groups. Calibration: 50 μm. (j, k) Vertical scatterplot depicting the distribution of immobility time of tail suspension test (j) and sucrose consumption of sucrose preference test (k) from control (n=8), LH (n=8), and LH+RU486 (n=6) groups. Data are expressed as mean±SEM; **p<0.01 vs Control, ##p<0.01 vs LH.

PowerPoint slide

Full size image

To clarify the role of GR in the chronic stress-induced hypofunction of glutamatergic transmission and depression-like behavior, we generated naive rats that received subcutaneous injection of dexamethasone (DEX) (2 mg/kg) daily (Figure 5a), a selective glucocorticoid receptor agonist, to activate GR exogenously and mimic chronic stress-induced depression-like behavior. Rats that received subcutaneous injection of DEX had reduced body weight gain per se (Supplementary Figure S7a; unpaired t-test, p<0.01). As expected, DEX-treated rats exhibited decreased rectification index (Figure 5b; unpaired t-test, p<0.01), frequency (Figure 5d; unpaired t-test, p<0.01), and amplitude (Figure 5e; unpaired t-test, p<0.01) of mEPSCs in the vlPAG neurons compared with the control group. Moreover, DEX also diminished both surface GluR1 and pGluR1 protein expression in the vlPAG (Figures 5f and g; unpaired t-test, p<0.01). Immunohistochemical evidence also demonstrated that pGluR1, but not NeuN, expression was reduced by daily DEX injection (Figures 5h and i; unpaired t-test, p<0.01). Moreover, rats that received daily subcutaneous injection of DEX had increased TST and decreased SPT (Figures 5j and k; unpaired t-test, p<0.01). In addition, DEX also suppressed locomotor activity (Supplementary Figures S7b and c; unpaired t-test, p<0.01) and time in central zone (Supplementary Figures S7b and d; unpaired t-test, p<0.01) of open field test. The above-mentioned results confirmed that GR activation is required for both the downregulation of GluR1 on the cell surface of vlPAG neurons and depression-like behavior after chronic stress.

Figure 5
figure 5

Glucocorticoid receptor activation mimics chronic stress-induced depression of excitatory synaptic transmission via both pre- and postsynaptic mechanisms. (a) Schematic experimental procedure of the effect of glucocorticoid receptor agonist, dexamethasone (DEX), in naive rats. (b) Rectification index of AMPA receptor obtained from control and DEX groups. Inset: represented traces of rectification index obtained from control (n=8) and DEX (n=10) groups. Calibration: 50 pA, 20 ms. (c) Representative traces of mEPSCs recorded from vlPAG neurons dissected from control and DEX groups. Calibration: 25 pA, 600 ms. (d, e) Vertical scatterplots depicting the distribution of frequency (d) and amplitude (e) of mEPSCs recorded from vlPAG neurons obtained from control (n=7) and DEX (n=12) groups. (f, g) Vertical scatterplot depicting the distribution of surface and cytoplasm (f) and pGluR1 (g) expressions in the vlPAG. Inset: represented western blot of protein expression from control (n=5) and DEX (n=5) groups. (h, i) Immunohistochemistry images (h) and quantitative analysis (i) of pGluR1 and NeuN in the vlPAG slices dissected from control (n=6) and DEX (n=6) groups. Calibration: 50 μm. (j, k) Vertical scatterplot depicting the distribution of immobility time of tail suspension test (j) and sucrose consumption of sucrose preference test (k) from control (n=7) and DEX (n=7) groups. Data are expressed as mean±SEM; **p<0.01 vs Control.

PowerPoint slide

Full size image

Chronic Stress Does Not Affect GABAergic Transmission in the vlPAG

Alteration in GABAergic transmission leads to the pathogenesis of major depressive disorder (Luscher et al, 2011). In addition to glutamatergic excitatory transmission in the vlPAG, GABAergic-mediated inhibitory transmission also plays a critical role in controlling the excitability of vlPAG neurons (Aubrey et al, 2017; Lau and Vaughan, 2014). Hence, we investigated whether chronic stress altered GABAergic transmission in the vlPAG causing depression-like behavior. We evaluated the PPR of GABAergic transmission and miniature IPSC (mIPSC) recordings to identify whether chronic stress alters GABAergic-mediated transmission in the vlPAG. We observed no significant differences in the PPR between the control and LH groups (Supplementary Figure S8b; unpaired t-test, p=0.742). Moreover, we found that neither the frequency (Supplementary Figures S8c and d; unpaired t-test, p=0.517) nor the amplitude (Supplementary Figures S8c and e; unpaired t-test, p=0.092) of mIPSCs recorded in vlPAG slices from the LH group were significantly different from that in the control group. These results suggest that chronic stress does not affect GABAergic transmission in the vlPAG.

Discussion

To the best of our knowledge, this is the first report to show that chronic stress suppresses glutamatergic transmission in the vlPAG and leads to depression-like maladaptive behavior through reduced glutamate release presynaptically and diminished GluR1 expression on the surface postsynaptically. Decreased excitatory transmission in the vlPAG contributes to the development of chronic stress-induced maladaptive behavior.

In contrast to defensive behaviors such as flight, the ventrolateral periaqueductal gray has been evidenced recently as a center controlling passive coping and depression-like behaviors (Berton et al, 2007; Johnson et al, 2016; Landgraf et al, 2016; Wang et al, 2016). Previously, Berton et al (2007) found that local overexpression of deltaFosB in the vlPAG dramatically reduces inescapable shock-induced depression-like behavior. This hints that the vlPAG actually controls stress-induced despair behavior. A recent report notes that a loss of circadian rhythms in the PAG is associated with LH-induced depression-like behavior, demonstrating that the PAG is a brain region directly involved in the development of behaviors associated with helplessness (Landgraf et al, 2016). Another study demonstrates that knockdown of galanin receptor 1 specifically in the vlPAG rescues chronic mild stress-induced depression-like behavior (Wang et al, 2016). This implies that the vlPAG might represent a novel brain region for controlling chronic stress-induced depression-like behavior. Moreover, inactivation of the bed nuclei of the stria terminalis–vlPAG pathway increases immobility time in tail suspension and forced swim tests, suggesting that the vlPAG serves as a core that modulates passive coping behavior (Johnson et al, 2016). Taken together, all of the above evidence demonstrates that the vlPAG is a critical brain nucleus governing stress-induced depression-like behavior. In the current study, we found that the expression level of GluR1 in the vlPAG correlated negatively with LH-induced depression-like behavior. Local pharmacological inhibition of GluR1 in the vlPAG leads to depression-like behavior. Interestingly, intradorsal PAG microinjections of glutamatergic agonists have been shown to enhance conditioned freezing and promote proaversive effects (Reimer et al, 2012). Another study demonstrates that activation of glutamatergic transmission in the dorsolateral PAG enhances the flight reaction (Moreira et al, 2004). This body of evidence suggests that glutamatergic transmission in the dorsal PAG is crucial for the regulation of freezing, flight, and aversive behaviors. However, although it is still unclear whether alteration of GluR1 regulates freezing, flight, and aversion, these behaviors are known to be controlled by the vlPAG too. Therefore, our results that the vlPAG governs chronic stress-induced depression-like behavior might provide a new insight into the pathophysiology of depression and an unexplored aspect of study about psychiatric-related behaviors.

Numerous studies suggest that chronic stress induces divergent synaptic plasticity in different brain regions (Hajszan et al, 2009; Kallarackal et al, 2013; Li et al, 2011; Penzo et al, 2014; Wei et al, 2016; Yuen et al, 2012). Repeated stress leads to the loss of glutamatergic transmission including reduction in presynaptic glutamate release and postsynaptic GluR1 in the PFC, hence resulting in object recognition memory impairment (Yuen et al, 2012). Moreover, chronic unpredictable stress elicits depression-like affective state because of diminishing GluR1-mediated synaptic transmission in the temporoammonic-CA1 synapse (Kallarackal et al, 2013). The PAG, a crucial midbrain region controlling susceptibility to stress and generation of fear-related responses, is involved in stress-related psychiatric disorders (Norman et al, 2010; Wang et al, 2016). A previous study indicated that chronic restrain stress decreases excitatory amino acid transporter 2 protein level in the vlPAG; this implies that suppression of excitatory neurotransmission and neuronal dysfunction is involved in stress-induced neuropsychiatric disorders (Imbe et al, 2012). Therefore, in the current study, we first investigated the cellular mechanisms of neurotransmission in the vlPAG underlying chronic stress-induced depression-like maladaptive behavior. We found that chronic unpredictable foot shock stress elicits hypofunction of glutamatergic transmission through both pre- and postsynaptic mechanisms. Presynaptically, the projection fibers originating from other brain regions show decreased glutamate release onto vlPAG neurons. Postsynaptically, GluR1 on the cell surface endocytosed into cytoplasm, causing reduction in glutamatergic transmission in the vlPAG. The activity of glutamatergic transmission in the vlPAG not only affects nociceptive modulation (Ho et al, 2013, 2015), but also stress-associated defensive reaction and autonomic regulation (Tovote et al, 2016). Thus, the vlPAG plays an essential role in stress-induced maladaptive behavior and might be a new target brain region for the development of therapeutic strategy.

In addition to stress-induced depression-like maladaptive behavior, stress also affects pain sensation bidirectionally. Stress may either suppress or exacerbate pain sensation, termed stress-induced analgesia and stress-induced hyperalgesia, respectively, depending on the duration and intensity of the stressor. A previous report demonstrates that acute footshock stress increases paw withdrawal latency; this implies that acute stress may result in stress-induced analgesia (Donello et al, 2011). Another study reports that chronic footshock leads to visceral hyperalgesia, suggesting that chronic stressors may cause stress-induced hyperalgesia (Robbins and Ness, 2008). In the present study, we found that chronic stress reduced both paw withdrawal threshold and latency, that is, stress-induced hyperalgesia. These results are consistent with our previous reports showing that spinal nerve ligation induces mechanical allodynia (Ho et al, 2013, 2015). Actually, it is difficult to disentangle the relationship between pain and depression, as both pain and depression usually co-occur and share some similar symptoms, implying overlapping neurobiological underpinnings. Therefore, the comorbidity of pain and depression should be further investigated to reveal this issue. Furthermore, we could not exclude the possibility that acute footshock stress affects the cellular changes in the vlPAG and behavioral maladaptation. The neuroplastic and behavioral changes after acute footshock stress should also be further examined.

Activation of GR signaling in the brain in response to stress has been implicated in the pathogenesis of stress-associated psychiatric disorders (Wei et al, 2016; Yuen et al, 2012). Several lines of evidence demonstrate that stress induces neuroplastic changes in different brain regions through glucocorticoid-related cascades and downstream signaling affecting synaptic transmission and maladaptive behavior (Di et al, 2016; Licznerski et al, 2015; Yuen et al, 2012). A previous study indicated that repeated stress decreases glutamatergic transmission in the PFC contributing to the impairment of object recognition memory. The impairments of cognitive processes are blocked and mimicked by GR antagonist and corticosterone, respectively, indicating that stress downregulates glutamatergic transmission through GR-dependent mechanisms (Yuen et al, 2012). A recent report demonstrated that LH paradigm significantly reduces serum and glucocorticoid-regulated kinase 1, a GR-associated kinase, protein expression in the PFC, associated with posttraumatic stress disorder, implying GR-mediated signaling plays a critical role in learned helplessness-induced depression-like behavior (Licznerski et al, 2015). In the current study, systemic injection of GR antagonist abolished chronic stress-induced suppression of glutamatergic transmission. Exogenous administration of GR agonist elicited similarly chronic stress-induced neuroplastic changes in the vlPAG of naive rats. Although high-dose GR agonist decreased body weight and locomotor activity, we could not exclude the possibility that the systemic effects of high-dose GR agonist suppress glutamatergic transmission in the vlPAG. These results suggest chronic stress-induced neuroplastic changes and maladaptive behavior occur through GR-mediated mechanisms.

In addition to the AMPA receptor, accumulating evidence has shown that the NMDA receptor and metabotropic glutamate receptors (mGluRs) are highly implicated in learned helplessness-induced psychopathology (Burgdorf et al, 2013; Pignatelli et al, 2013). Burgdorf et al (2013) demonstrate that an NMDA receptor glycine-site functional partial agonist rescues LH-induced depression-like behavior, implying NMDA receptor involvement in LH-induced depression-like behavior. In the present study, we found a reduced EPSCAMPA/NMDA ratio in the vlPAG of LH-induced rats, implying that chronic stress might cause an enhanced NMDA receptor-mediated current in the vlPAG. However, the detailed mechanisms of LH-induced upregulation of NMDA receptors in the vlPAG should be examined. Furthermore, mGluRs have also been reported to participate in LH-induced depression-like behavior, as enhanced expression and function of mGluR5 in the hippocampus is associated with the depression-like phenotype of LH rats (Pignatelli et al, 2013). In the LH paradigm, we could not exclude the possibility that LH-induced depression-like behavior is due to the alteration of mGluR in the vlPAG.

Multiple cell types in the vlPAG could be associated with a distinct behavioral phenotype (Bobeck et al, 2014; Tovote et al, 2016). A previous report demonstrates that inhibition of the GABAergic neurons in the vlPAG leads to muscle atonia, implying that vlPAG suppresses rapid eye movement sleep by providing GABAergic inhibition to rapid eye movement-promoting neurons (Weber et al, 2015). A recent study reports that activation of glutamatergic neurons in the vlPAG specifically triggers freezing behavior, suggesting that some of the glutamatergic vlPAG neurons are positively correlated with freezing behavior (Tovote et al, 2016). In the current study, however, whether or not the cell type specificity in the vlPAG governs chronic stress-induced depression-like maladaptive behavior is still uncertain and should be further investigated.

The imbalances of excitatory and inhibitory neurotransmission in the brain regions have been evidenced as possible pathological mechanisms of neuropsychiatric disorders (Ghosal et al, 2017). Defects in GABAergic transmission have been reported to contribute to the etiology of major depressive disorder (Luscher and Fuchs, 2015). Mice with γ2 subunit-lacking GABAA receptors in the forebrain exhibit core symptoms of behavioral despair and hypothalamic–pituitary–adrenal axis hypersensitivity, demonstrating that GABAergic deficits contribute to depression-like behavior (Shen et al, 2010). A recent study also revealed that downregulation of glutamatergic transmission owing to stress-induced GABAergic deficits is involved in depression-like brain states, providing novel mechanisms of etiology of major depressive disorder (Ren et al, 2016). Mounting evidence has demonstrated that deficits in the GABAergic-mediated system play a pivotal role in major depressive disorder. However, we found no significant changes in GABAergic transmission in the vlPAG after chronic stress paradigm in the current study. This discrepancy may be because of the following possibilities. First, insufficient GABAergic-mediated responses might be recruited originating from other brain nuclei. Neurons in the vlPAG receive GABAergic inputs from both local interneurons (Aubrey et al, 2017; Lau and Vaughan, 2014) and the central amygdala projection (CeL) (Penzo et al, 2014). A previous study indicated that foot shock potentiates excitatory synapses onto somatostatin-positive neurons in the CeL, sending direct long-range GABAergic projections onto vlPAG neurons (Penzo et al, 2014). However, in the present study, we did not find hyperactivity of GABAergic transmission in the vlPAG after the foot shock paradigm. This may be because the long-range GABAergic projections from the CeL might not be intact and sufficient for recruitment; hence, no deficits of the GABAergic system were observed in our results. Second, different stress paradigms may cause distinct effects on the glutamatergic and GABAergic systems. Previous reports demonstrate that diverse chronic pain models elicit neuroplastic changes selectively not only in the glutamatergic (Ho et al, 2015) but also GABAergic system (Tonsfeldt et al, 2016), implying various types of stress may affect different systems.

In summary, our study provides evidence that chronic foot shock exposure leads to depression-associated behavioral despair and anhedonia. Because of the complex nature of depression with altered neurotransmission across several brain regions, deficits in excitatory glutamatergic transmission in the vlPAG might be an underlying mechanism of chronic stress-elicited neuropsychiatric disorder-associated maladaptive behavior. Understanding the pathophysiological mechanisms involved in the expression of maladaptive behavior could be highly beneficial in fostering the development of novel therapeutic strategies for neuropsychiatric disorders.

Funding and disclosure

The authors declare no conflict of interest.