Chronic restraint stress induces hippocampal memory deficits by impairing insulin signaling

All procedures that involved laboratory animals were approved by the Institutional Animal Care and Use Committee at the DGIST.

Eight-week-old male C57BL/6 N mice were housed with a 12:12 h light-dark cycle (lights on 07:00 to 19:00) with 22–24°C temperature, 40–60% humidity, and food and water were supplied ad libitum. Four male mice were housed in each cage until six-weeks old, and individually housed for a week and handled daily for acclimation for another week before experiments. Mice were randomly divided into the control (Ctrl) and stressed groups. Mice of the stressed group were horizontally immobilized for 6 h/day (from 10:00 to 16:00) in the acrylic cylindrical flat-bottom head-first restrainer (Φ25 × {H}90 mm, Jeungdo Bio & Plant Co., Seoul, Korea) for 2 weeks in their home cages. Restrainer has several slots to restrain the mouse firmly according to the size of each mouse, and suppresses the physical movement of the limbs without causing pain. After being restrained, mice were released back into their home cages immediately. Non-restraint mice (Ctrl) remained in their home cages without CRS procedure, and both Ctrl and CRS mice could not access food and water during the period of CRS exposure. The weight of the mice was measured every week before the experiment.

Corticosterone (Sigma-Aldrich, 27840), staurosporine (STS; Cell Signaling Technology, 9953), bafilomycin A1 (Sigma-Aldrich, B1793), insulin-FITC (Sigma-Aldrich, I3661), methyl cellulose (Sigma-Aldrich, M0512), sodium borohydride (Sigma-Aldrich, 213462), and insulin (Roche Diagnostics, 11 376 497 001) were purchased from the indicated companies. Antibodies against IR subunit β (3025), p-Akt-S473 (9271), Akt (9272), p-mTOR-S2448 (2971), mTOR (2972), and cleaved caspase-3 (C.Casp-3; 9664) were purchased form Cell Signaling Technology. Antibodies against p-IRS1-Y612 (Invitrogen, 44-816G), IRS1 (Invitrogen, PA1–1057), GR (Santa Cruz Biotechnology, SC-56851), LC3 (Novus, NB100–2220), p62 (Sigma-Aldrich, P0067) and β-actin-HRP (Santa Cruz Biotechnology, SC-47778) were purchased from the indicated companies.

Primary hippocampal neurons were obtained from C57BL/6 N mice on embryonic day 17. Briefly, hippocampi were dissociated in Hank’s Balanced Salt Solution (Invitrogen, 14175–095) and maintained in a Neurobasal Medium (Gibco, 21103–049) supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin (HyClone, SV30010), 2 mM L-glutamine (Invitrogen, 25030–081), and 2% B27 supplement (Invitrogen, 17504–044). After 3 days, one-third of the medium was replaced with medium containing cytosine β-D-arabinofuranoside (Sigma-Aldrich, C1768; final concentration, 3 μM). After 4 days, one-third of the medium was replaced with fresh medium without cytosine β-D-arabinofuranoside and cultures were used for experiments on day 11 or 12 days.

Cell death was assessed by dual staining of cells with the membrane-permeable dye Hoechst 33342 (Invitrogen, H1399) to stain all cells and with the membrane-impermeable dye propidium iodide (PI; Sigma-Aldrich, P4170) to stain dead cells. The percentage of cell death (%) was calculated by dividing the number of red-positive (PI) cells by that of blue-positive (Hoechst 33342) cells.

Insulin-FITC reconstituted in 0.01 N HCl was diluted in 0.9% saline with 0.001% methyl cellulose and was delivered twice into both nares: the first dose 6 h and the second dose at 1 h before sacrifice (2.5 μg/24 μl each). Control animals received 0.9% saline with 0.001% methyl cellulose.

Mice were hand-restrained in a supine position with the neck in extension and were not anesthetized. A total of 100 μg of insulin in 20 μl of vehicle (0.9% saline with 0.001% methyl cellulose) or the same volume of vehicle alone was delivered daily into both nares simultaneously 30 min before restraint stress for 7 days.

Two nestlets were added in each cage at 16:00 on the 13th day after restraint stress and scores were measured on the 14th day (09:50) before the start of restraint stress.

Y-maze test was used to evaluate spatial working memory of the mice. The test was performed in a Y-shaped maze with three white opaque plastic arms. Mice were placed into the center of the maze and were allowed to explore the three arms freely for 6 min. Each experiment was video recorded with EthoVision software (Noldus, EthoVision XT 11.5). After exploration, alternation (%) was calculated as the number of three consecutive arm entries divided by the number of possible alternations (total arm entries minus two).

Mouse blood serum samples were acquired on the 14th day 30 min after the end of stress from the submandibular vein using animal lancets. Blood CORT levels were measured using an ELISA kit (Enzo Life Sciences, ADI-900-097).

Mouse hippocampal neurons were fixed with 4% paraformaldehyde (PFA) and permeabilized with 0.1% Triton X-100 in phosphate buffered saline (PBS) containing 0.5% bovine serum albumin (BSA; Affymetrix/USB, 10857). Following blocking with 0.5% BSA, the samples were incubated overnight with cleaved caspase-3 antibody at 1:1000 dilutions and then with an anti-rabbit Alexa 488 secondary antibody (Jackson ImmunoResearch, 711–545-152). Samples were stained with Hoechst 33342 and visualized under a confocal laser scanning microscope (Carl Zeiss, LSM 700); images were analyzed using ZEN12 software (Carl Zeiss).

Mice were anesthetized by Zoletil and Rompun injection and perfused with PBS, followed by 4% PFA. Brains were then removed, post-fixed in 4% PFA for 12 h and cryoprotected in 30% sucrose until they sank to the bottom of the tube. For fluorescence microscopy, brains were frozen in optimal cutting temperature compound and cryosectioned coronally with a 30 μm thickness. Brain section coordinates were based on the mouse brain atlas [55]. The sections were free-floated in PBS, blocked with 0.1% sodium borohydride in PBS to reduce background autofluorescence, counter stained with Hoechst 33342, and mounted. FITC-tagged insulin was visualized by a slide scanner (Zeiss, Axio Scan.Z1) and images were analyzed using ZEN12 software.

Chronic psychological stress or long-term application of CORT results in time-dependent hippocampal neuronal damage [17‐19, 38, 42]. However, the underlying mechanisms of stress-induced hippocampal neuronal damage are still not well understood. To examine the neurotoxicity of CORT, we treated mouse hippocampal neurons in vitro with a wide range of CORT doses and found that 100 or 200 μM CORT significantly increased hippocampal neuronal cell death (Fig. 1a). Because there were reports that glucocorticoids induced neuronal apoptosis [17, 42], we examined whether the CORT-induced neurotoxicity was apoptotic or not. We used STS, a widely used apoptosis inducer, as a positive control of apoptosis [56]. C.Casp-3 is active form of caspase-3 and is the best-known marker of apoptosis [57]. Of interest, STS but not 200 μM CORT induced robust C.Casp-3 immunoreactivity (Fig. 1b). Moreover, nuclear condensation, another well-known characteristic of apoptosis, was observed in STS-treated but not in CORT-treated cells (Fig. 1b, arrow). Immunoblot analysis showed that treatment with STS but not with 200 μM CORT induced C.Casp-3 (Fig. 1c). Since apoptotic markers were not detected after CORT treatment, we wondered whether another mode of programmed cell death, such as autophagic cell death [44, 48, 58], is involved. CORT increased the level of LC3-II (Fig. 1d, e). Bafilomycin A1 (Baf.A1) is a well-known inhibitor of the late phase of autophagy and is used to prevent maturation of autophagic vacuoles by inhibiting fusion between autophagosomes and lysosomes [59]. Although an increase in LC3-II level is typically regarded as an indicator of increased autophagy flux, impaired autophagy flux and blockage of autophagic degradation can also result in LC3-II accumulation level, because LC3-II itself is degraded by autophagy. These possibilities can be distinguished by blocking the late phase of autophagy flux. Further accumulation of LC3-II after treatment Baf.A1 would indicates an increase in autophagic flux, not the suppression of autophagic degradation [52, 56]. On the other hand, if the LC3-II level is not significantly affected by Baf.A1, it may suggest impaired autophagy flux. Since p62 is also degraded in autolysosomes, blocking of autophagosome fusion with lysosomes or inhibition of lysosomal hydrolases can cause p62 accumulation under the conditions of high autophagy flux. Baf.A1 treatment of control cells for 2 h led to accumulation of p62 and LC3-II (Fig. 1d, e). CORT treatment increased LC3-II and decreased p62, and the CORT+Baf.A1 group accumulated p62 together with a higher increase in LC3-II than in the CORT-alone group (Fig. 1d, e). These data suggest that CORT does not induce apoptosis, but enhances autophagy flux in mouse hippocampal neurons, suggesting the involvement of autophagic cell death in CORT-induced neurotoxicity.

Next, we examined whether CORT affects insulin signaling, since insulin signaling is closely related to autophagy and is also important for hippocampal development and function, and for survival of hippocampal neural stem cells and neurons [47, 48, 60]. Treatment of mouse hippocampal neurons with 200 μM CORT increased the expression of GR, a positive marker of CORT action (Fig. 2a, b). CORT treatment downregulated phosphorylation of IRS1 on Y608 (Y612 in humans) without changing the total IRS1 level (Fig. 2a, b). The phosphorylation of Akt on S473 and mTOR on S2448 was also decreased after CORT treatment (Fig. 2a, b). Intriguingly, the expression of IR subunit β (IRβ) was increased by CORT treatment (Fig. 2a, b), which might be a compensation to overcome the impaired insulin signaling [48]. Taken together, these results suggest that CORT impairs insulin signaling (the IRS1–Akt–mTOR pathway) in mouse hippocampal neurons.

Next, we wondered whether psychological stress, such as CRS, which involves high levels of corticoid hormones, affects insulin signaling in vivo. To generate CRS, we used a flat bottom restrainer for 6 h/day for 2 weeks (Fig. 3a). First, we measured stress responses during and after CRS. Whereas the initial body weight did not differ between the CRS and Ctrl groups, CRS significantly inhibited body weight gain compared to Ctrl (Fig. 3b) and significantly increased the circulating blood CORT level after 2 weeks (Fig. 3c). We used nesting behavior as another index of neuronal stress. Nesting behavior depends on normal hippocampal function and is impaired by hippocampal lesions [61]. Nesting behavior was measured by scoring the quality of the nest with a 5-point rating scales. Rating scale 1 means that nestlets are not touched, and 5 means a nearly perfect nest with a crater and higher walls. Unlike the Ctrl group, the CRS group had difficulties in building a nest with nestlets (Fig. 3d, e).

To determine whether CRS influences insulin signaling in the hippocampus, we measured the expression and phosphorylation of relevant signaling proteins by immunoblotting (Fig. 3f, g). The expression of GR was increased in the hippocampi of the CRS group, which is likely a response to an increased blood CORT level. Similar to hippocampal cultures treated with CORT, CRS increased the expression of IRβ. Interestingly, the levels of p-IRS1-Y608, p-Akt-S473 and p-mTOR-S2448 were notably decreased in the CRS group. In summary, these data demonstrate successful induction of stress and subsequent impairment of insulin signaling in the mouse hippocampus following 2 weeks of CRS.

To test whether down-regulated insulin signaling can be recovered by insulin delivery to the hippocampus, we delivered insulin intranasally. First, to confirm that insulin arrived into the hippocampus, we tagged insulin with FITC and tracked it by histology image analysis after intranasal delivery (Fig. 4a). Administration of insulin-FITC but not saline resulted in green fluorescence in the dentate gyrus areas of the hippocampus 6 h after administration (Fig. 4b). Administration of insulin-FITC in CRS mice showed similar green fluorescence distribution (Fig. 4c). These data validated successful delivery of insulin into the hippocampus in both Ctrl and CRS mice. Next, we divided mice into 4 groups: Ctrl with saline (Ctrl/Saline), CRS with saline (CRS/Saline), Ctrl with insulin (Ctrl/Insulin), and CRS with insulin (CRS/Insulin). Insulin or saline was intranasally delivered for 7 days during the 2nd week of CRS (Fig. 5a). Both Ctrl/Saline and Ctrl/Insulin groups showed an increased in body weight, while the CRS/Saline group showed significantly reduced body weight (Fig. 5b). Interestingly, the CRS/Insulin group also showed reduced body weight after 1-week-CRS, but the reduction was reversed after intranasal administration of insulin (Fig. 5b). This shows a potential beneficial effect of intranasal insulin delivery for the functional and metabolic recovery of stressed hippocampus. Blood CORT level and nest-building behavior were also measured after experiments. Interestingly, intranasal administration of insulin did not alter the level of CORT increased by CRS (Fig. 5c). This data suggest that the neuroprotective activity of insulin is by suppressing autophagy in the hippocampus against the detrimental effects of CORT, but not by alteration of the Hypothalamic-Pituitary-Adrenal axis. The CRS/Insulin group also regained the ability to build a nest compared to the CRS/Saline group (Fig. 5d, e).

We next measured the function of hippocampal spatial working memory by using the Y-maze test. Rodents typically prefer to explore a new arm of a maze instead of going to the previously visited arm. The Y-maze test measures the willingness of mice to explore new environments by counting the number of arm entries and calculating the percentage of alternation, which is related to the hippocampal spatial working memory. Normal mice show a high percentage of alternation. Thus, this test can easily quantify the cognitive deficits in mice [62, 63]. The number of total arm entries was similar in all four groups of mice, showing intact general locomotor activity (Fig. 5f, top). However, the CRS/Saline group showed a significantly reduced percentage of alternation compared to the Ctrl/Saline group, while the CRS/Insulin group recovered it to a level similar to those of Ctrl groups (Fig. 5f, bottom). To confirm biochemically whether intranasal insulin delivery recovered insulin signaling impaired by CRS, we measured the expression and phosphorylation levels of the relevant signaling proteins in the hippocampus by immunoblotting analysis (Fig. 6a). The expression levels of GR and IRβ were increased only in the CRS/Saline group (Fig. 6b, left). Interestingly, the levels of p-IRS1-Y608, p-Akt-S473, and p-mTOR-S2448 were reduced in the CRS/Saline groups but recovered in the CRS/Insulin groups (Fig. 6b, right). Altogether, these data suggest that intranasal insulin delivery relieved stress in the hippocampus.

Previous study reveals that insulin withdrawal causes a caspase-independent autophagic cell death in adult hippocampal neural stem cells [48], and we confirmed CORT impairs insulin signaling and induces autophagy features in mouse hippocampal neurons in Figs. 1 and 2. We next examined whether CRS-induced impairment of insulin signaling induced the feature of apoptosis or autophagy. LC3-II levels in the hippocampal lysates were increased by CRS, which were suppressed by intranasal insulin administration (Fig. 6c). However, the level of C.Casp-3 was not significantly different between Ctrl and CRS mice hippocampus regardless of intranasal administration of insulin (Fig. 6d). STS-treated in vitro hippocampal neuronal lysate was loaded as a positive marker of C.Casp-3 (Fig. 6a, last lane).

In summary, our data demonstrate that, although psychological stress impairs hippocampal functions and insulin signaling, intranasal insulin delivery can efficiently prevent CRS-induced insulin signaling impairment and rescue deficits in hippocampus-dependent neurobehavior.