Astragaloside IV enhances memory and modulates hippocampal synaptic plasticity by decreasing GABAergic inhibition through EGR-1 mediated BDNF/TrkB signaling pathway in mice

Background Astragaloside IV (ASIV) is one of the saponins isolated from Astragalus membranaceus, a widely used traditional Chinese medicine and a health product sold all over the world. However, so far, the effect of ASIV on GABAergic synaptic transmission has not been elucidated yet. In the present study, the effect of ASIV on memory and hippocampal synaptic plasticity was investigated in mice and down-regulated early growth response protein 1 (EGR-1) knockout mice. Methods Behavior tests including radial-arm maze test and shuttle-box test, liquid chromatography-tandem mass spectrometry, western blotting analysis, quantitative PCR, electrophysiological recording, and electron microscopy were used in this study. The difference of data was detected by unpaired student t-test or two-factor analysis of variance (ANOVA) or Mann-Whitney U test.


Introduction
Hippocampal synaptic plasticity is vital to memory formation and retention. Long-term potentiation (LTP) and long-term depression (LTD) are in-depth studies of the process to understand the synaptic basis of learning and memory [1]. LTD is modulated by the gamma-aminobutyric acid (GABA) [2], which is the main inhibitory neurotransmitter in mammals and is widely distributed in the central nervous system (CNS) of mammals. GABA receptors mediate inhibitory nerve transfer, preventing neurons from overexciting in the adult brain (Suzdak & Jansen, 1995). LTD seems to be a supplement to the candidate memory, which is independent of GABA type A receptor (GABA A R) activation [2]. Inhibitory GABA A R activity regulates GABAergic synaptic plasticity through extracellular signal-regulated kinases (ERK) and brain-derived neurotrophic factor (BDNF) signaling [3]. BDNF is a member of the neurotrophic family, which is found mainly in the hippocampus, amygdala, cortex and cerebellum. It is synthesized by neurons and glial cells and is involved in the survival, differentiation and regeneration of neurons by binding to high a nity receptor, (TrkB) [4,5].
Early growth response protein 1 (EGR-1), also known as Zif268, Zenk, Krox-24, NGF1-A, TIS8, and Krox-24, belongs to the early growth response (EGR) gene family and is also described as induced transcription factor [6]. It plays a vital role in the maintenance of synaptic plasticity including LTP [7,8] and LTD [9], which is well known to be essential for many cognitive functions. And tremendous evidence demonstrated that EGR-1 actively participates in different memory forms as well as different memory processes, from learning memory consolidation [10,11] and system consolidation [12,13] to reconsolidation [14,15].
Astragaloside IV (ASIV) is a saponin molecule found in Astragalus membranaceus (Fisch.) Bge, an herbal medicine proverbially used in China and a health product widely sold in Europe. Multiple pharmacological activities of ASIV have been disclosed, such as anti-oxidation [16] anti-apoptosis [17], anti-in ammation [18], and immuno-regulation [19,20]. And ASIV also shows prominent neuroprotective effects in multifarious CNS injuries, including cerebral ischemic-reperfusion injury [21][22][23], Parkinson's disease [24] and Alzheimer's disease [25,26]. However, so far, the study of the effect of ASIV on synaptic plasticity and memory is lacking. In this study, we rstly reported that ASIV could enhance memory and improve hippocampal synaptic plasticity in mice. Furthermore, our study disclosed that the effect of ASIV was achieved probably by decreasing GABAergic inhibition via EGR-1 mediated BDNF/TrkB signaling pathway. The novel ndings can contribute the potential application of ASIV in neurological diseases with impaired memory.

Materials And Methods
Animal and Drug Administration C57BL/6 mice (Male, 14 ~ 18 g, 4 ~ 5 weeks old) were provided by the Laboratory Animal Center of Shanghai University of Traditional Chinese Medicine (SHUTCM, Shanghai). All experiments on animals were performed according to the protocol approved by Animal Care and Use Committee of SHUTCM and all animals received humane care (Ethical approval no. SZY201610005). Animals were acclimatized for 2 weeks before the formal behavior experiments.
Fourteen mice were injected intraperitoneally (i.p.) with ASIV (25 mg/kg, 40% 1,2-Propanediol + 5% Ethanol + 1% Polyethylene glycol in phosphate buffer saline solution) for ve weeks, while fourteen mice were administered with solvent served as the control. All of the twenty-eight mice were subjected into behavioral tests. Another forty-four C57BL/6 mice of the same source were with only injection for further experiments; twenty-two mice received an i.p. injection of ASIV (25 mg/kg) daily for two weeks, while twenty-two mice were administered with solvent served as the control.
For LTD induction in vitro experiment, eight C57BL/6 mice (male, 12 ~ 17 g, 3 ~ 4 weeks old) were provided by SPF biotechnology Co., Ltd (Beijing, China). Four of them were i.p. administered with ASIV (25 mg/kg) for two weeks, while four mice were administered with solvent served as the control.
All mice were housed at room temperature (25 ± 1 °C) under a 12 h light / 12 h dark cycle, and fed with food and drank with water ad libitum freely. All behavioral tests were conducted in the light phase between 12:00 a.m. and 18:00 p.m. In order to avoid experimental deviation, all behavioral observers were blinded to the administration of the experimental mice. All animal experiments were conducted in accordance with the guidelines for Laboratory Animal care and Use Committee of SHUTCM and National Institutes of Health guide for the care and use of Laboratory animals (NIH Publications No. 8023, revised 1978).

Behavioral Tests
Radial-arm Maze Test (RAMT): The apparatus for radial-arm maze (RAM) (Mobiledatum Inc, Shanghai, China) consists of eight equally spaced arms (length 30 cm, height 15 cm, width 6 cm) radiating from a central maze hub (diameter 12 cm). It was made of opaque plexiglass with manually operated doors lead from the hub of the central maze to each arm. RAMT was conducted in accordance with the procedure described previously, with minor modi cation [27]. As shown in Fig. 1A, the mice were acclimatized to the RAM for 4 days before administrated with ASIV on day 1. Then, they were trained to explore the RAM with four arms placed with 50 mg bait (sugar: regular chow = 1:1) from day 2 to day 6. At the time of a week's administration, the mice were placed in the octagonal arena at the beginning of the experiment. The experiment ended when the mice explored the maze for 5 min or visited four baited arms. At the time of two weeks' administration, the test was conducted again. The maze was wiped down with 10% ethanol between each run to reduce olfactory cues. During these days, mice were given semi-food diet feeding. In these two tests, reference and working memory errors and the time required to complete the tasks were recorded and analyzed.

Shuttle-box test (SBT)
SBT was performed following RAMT on day 22. The chamber (Mobiledatum Inc, Shanghai, China) is divided into two equal compartments connected by a gate. A light is switched on alternately in the two compartments for conditioned stimulus. The test was conducted following a procedure described by Cheng et al [28]. Brie y, each mouse was allowed to adapt to the chambers for 4 min before the formal test. At the beginning of the experiment, the mice were placed in a compartment of the shuttle-box and back to the gate. Each mouse was given 30 consecutive trials at intervals of 20 s (light 5 s; interval, 3 s; 0.2 mA electric shock, 5 s; interval, 7 s) for ve consecutive days. The active avoidance response was recorded automatically if the mouse moved to another compartment during conditioned stimulus. At the day 35 of continuous ASIV administration, all animals received the same experimental protocol to assess memory consolidation.

Neurotransmitter Analysis
After all behavioral tests were completed; the mice as well as those administered with ASIV only for two weeks were sacri ced after anesthetized with 1.5% pentobarbitalum. Immediately, the hippocampus of mice was dissected on ice, frozen rapidly in liquid nitrogen and stored at − 80 ºC until analysis. The content of GABA in mouse hippocampus was determined by liquid chromatography-tandem mass spectrometry (LC-MS/MS) method as reported previously [29].

Electrophysiological Recording LTD induction in vitro
After continous i.p. injection of ASIV (25 mg/kg) for two weeks from 4 weeks old, the mice's brain were removed, and three brain slices (350 µm thick) were obtained from each mouse, covering the transverse ventral hippocampus. Slices were incubated at room temperature for at least 90 min in arti cial cerebrospinal uid (ACSF; containing 24 mmol/L NaCl, 26 mmol/L NaHCO 3 , 2 mmol/L KCl, 2 mmol/L CaCl 2 , 2 mmol/L MgSO 4 , 1.25 mmol/L NaH 2 PO 4 and 10 mmol/L D-glucose) and ventilated with a mixture of 95% oxygen and 5% carbon dioxide. Subsequently, the slices were transferred to the recording chamber and in ated ACSF at a ow of 2 mL/min rate at (22 ± 2) °C. After the temperature (33 ± 1) °C and perfusion rate, eld excitatory post synaptic potentials (fEPSPs) were recorded. A bipolar stimulus electrode was placed on Schaffer's collateral/commissural bundle in CA3 hippocampus, and a recording glass micropipette (1-3 MΩ, lled with 3 mol/L NaCl) was placed on the radiator in CA1 hippocampus. AxoClamp 700B ampli er lter was set at 3 kHz, while the sampling frequency was set at 20 kHz (Digidata 1440A, GE MDS, NY, USA). Meanwhile, the data were collected and stored with Clampex 10.3 (GE MDS company). Current intensity adjusted to 60-70% of maximal fEPSP (base lines stimulation).
Low frequency paired pulse stimulations (LFPS; 25 millisecond paired pulse interval, 1 Hz, monophasic quare pulses same as the baseline) induced LTD for 15 min. Immediately after each single stimulus for 60 s, the fEPSPs reaction was recorded for 60 min. Throughout the course the stimulation intensity is maintained constant. And throughout the recording period bicuculline (10 µmol/L) was included in the continuous perfusion (ACSF). The slope of fEPSPs was analyzed using Clamp t (GE MDS Company) and standardized with the average baseline response. The average changes of fEPSPs in 5 min after prestimulus baseline conditions were used to compare with the average response time of the rst 30-35 min after stimulation conditions for statistical purpose.
Input-output (I/O) function in vivo I/O curve re ects the relationship between the amplitude of fEPSP and the intensity of stimulation, which is used to evaluate the synaptic potency [30]. After continous i.p. injection of ASIV (25 mg/kg) for two weeks from 6 weeks old, the mice were anesthetized with 25% urahtane (1 mL/100 g) and xed on the stereotaxic device (Narishige Ins, Tokyo, Japan). The location parameters of granular cell layer in DG area were 2.0 mm after anterior fontanel, 1.4 mm near midline, 1.5 mm deep under dura mater; and 3.8 mm after anterior fontanel, 3.0 mm near midline, 1.5 mm deep under dura mater. In the front fontanel, recording electrode and stimulating electrode (Nihon Kohden, Tokyo, Japan) were almost in a straight line. According to the above positioning parameters, drill holes in the skull with a skull drill (Minimo, Tokyo, Japan), then insert recording electrodes into the granular cell layer of DG area, and insert stimulating electrodes into performant pathway. The reference electrodes were clamped on the scalp. With 0.1 mA (0.1-2.0 mA) as a step, the stimulation current was changed systematically to produce I/O curve. Three responses at each current level were averaged, and the population spike (PS) amplitude was examined. The signals were transferred through the ampli er (Axon, CA, USA), and ltered by PCLab202 software (Microsignalstar, Beijing, China).

Electron Microscopy
After deeply anesthetized with 20% urethane, mice were perfused with 2.5% glutaraldehyde in 0.1 M phosphate (pH 7.4) transcardially. The brains were dissected and post-xed in 2.5% glutaraldehyde for 24 h. Sequentially, they were incubated with 2% osmium tetroxide, dehydrated with series acetone and embedded with epoxy resin. The 60 nm-thick ultrathin sections were obtained on an ultra-microtome (ultracut UCT, Leica, Wetzlar, Germany) and observed by transmission electron microscope (Tecnai G2 Spirit BioTWIN, Hongkong, China).

Statistical analysis
All data were performed through Graphpad Prism 6 (Graphpad, La Jolla, CA). The difference of measurement data and numeration data were detected by unpaired student t-test or two-factor analysis of variance (ANOVA) and Mann-Whitney U test, respectively. Error bars represented the standard error of the mean (S.E.M.). P-value less than 0.05 was regarded as a signi cant difference.

ASIV enhanced the memory of mice in both RAMT and SBT
To explore the effect of ASIV on the memory of mice, the mice were subjected to the RAMT and SBT after ASIV administration. For RAMT, the working memory and reference memory that refer to short-term memory and long-term memory respectively were analyzed. As shown in Fig. 1B and Fig. 1C, ASIV treated mice performed better in the RAMT as their working memory error times (Fig. 1B, for 1-week, U = 62.00, P = 0.0782; for 2-week, U = 54.50, P = 0.0358 ) and reference memory error times were reduced signi cantly (Fig. 1C, for 1-week, U = 46.00, P = 0.1276; for 2-week, U = 33.00, P = 0.0174), compared with the control mice. Moreover, ASIV markedly increased the active avoidance times of mice in SBT (Fig. 1D, U = 30.50, P = 0.0012). These results showed that ASIV could enhance the memory of mice.
To determine the effect of ASIV on the GABAergic system, the concentration of GABA and the expression of GAD65 in hippocampi of mice were examined, respectively. As shown in Fig. 2A, hippocampal GABA level in ASIV treated mice was reduced when compared with that of the control mice (for 2-week, t (20) = 5.003, P = 0.0000; for 5-week, t (25) = 2.507, P = 0.0190). Accordingly, the expression of GAD65 in hippocampus of ASIV treated mice was signi cantly lower than that of the control mice ( Fig. 2B and C, for 2-week, t (4) = 3.349, P = 0.0286; for 5-week, t (4) = 2.779, P = 0.0499). To further investigate if ASIV could affect the strength of inhibitory synaptic transmission in hippocampus, untrathin sections obtained from blocks (1 mm × 1 mm × 1 mm) of CA3 region were subjected to electron microscopy observation. According to the reports, there are two major morphologic types of synapses, i.e. asymmetric and symmetric synapses [31,32]. Excitatory synapses are asymmetrical synapses with signi cant postsynaptic density, while inhibitory synapses are symmetrical synapses with thinner postsynaptic density. And there is also another type of synapse with oblique synaptic cleft and associated membrane density that is considered to be uncharacterized synapses [33]. As displayed in Fig. 2D, excitory and inhibitory synapses could be clearly identi ed in the electron microscopy image. ASIV treatment for 2 weeks and 5 weeks decreased the ratio of inhibitory synapse remarkably in CA3 region (Fig. 2E, for 2week, t (5) = 2.632, P = 0.0464; for 5-week, t (5) = 2.769, P = 0.0394). Since GABAergic system mainly participates in the synaptic inhibition, we next investigated the impact of ASIV on the LTD. As shown in Fig. 3, the average fEPSP slope of the control mice was decreased from (100.4670 ± 2.0450) to (74.4116 ± 1.8078), while that of the ASIV treated mice was descended from (103.0129 ± 1.8071) to (69.0670 ± 1.6496). Therefore, the LTD after LFS of ASIV treated mice was signi cantly lower than that of the control mice (t (148) = 2.430, P = 0.0163). The results suggested that ASIV could enervate hippocampal GABAergic neurotransmission.

EGR-1 KO abrogated the memory bene cial effect of ASIV on mice in both RAMT and SBT
To reveal the role of EGR-1 in the memory bene cial effect of ASIV on mice, EGR-1 KO mice were treated with ASIV and subjected to the RAMT and SBT using the same time line as illustrated in Fig. 1A. As shown in Fig. 6A and B, ASIV treatment did not reduce the working memory (for 1-week, U = 19.00, P = 0.2228; for 2-week, U = 43.00, P = 0.6158) and reference memory errors (for 1-week, U = 23.00, P = 0.4815; for 2-week, U = 35.00, P = 0.2775) of EGR-1 KO mice in RAMT. And ASIV treatment also did not change the active avoidance times of EGR-1 KO mice in SBT (Fig. 6C, U = 44.00, P = 0.6684). These results con rmed that EGR-1 played an indispensable role in the memory bene cial effect of ASIV.

EGR-1 KO abolished the potentiation of ASIV on LTD in mice
To further corroborate the role of EGR-1 in hippocampal GABAergic system of ASIV treated mice, the LTD response of EGR-1 KO mice treated with ASIV for 2 weeks was recorded. As shown in Fig. 7A-B, the average fEPSP slope of EGR-1 KO control mice was decreased from (100.397 ± 2.414) to (58.474 ± 1.753). By contrast, the average fEPSP slope of ASIV treated EGR-1 KO mice were descended from (100.349 ± 1.477) to (75.653 ± 1.708). Therefore, the LTD after LFS of ASIV treated EGR-1 KO mice was signi cantly higher than that of the vehicle treated EGR-1 KO mice (Fig. 7C, t (148 These results demonstrated that EGR-1 KO weakened the potentiation of ASIV on LTD in mice.

ASIV increased basic synaptic transmission and enhanced mRNA expression of EGR-1 in response to external stimuli in mice
To test the effects of ASIV on basic synaptic transmission in the CA1 region, the I/O function of ASIV treated mice was evaluated. As illustrated in Fig. 9A, the PS amplitude of ASIV-treated mice showed the increasing tendency, especially when the stimulus current was at 1.2 mA, 1.3 mA, 1.5 mA and 1.6 mA (P < 0.1), suggesting that ASIV might enhance hippocampal synaptic transmission in mice.
To further understand the physiological signi cance of the inhibitory effect of ASIV on EGR-1, the micetreated with ASIV for 2 weeks were subjected to a single-trial SBT together with the control mice. Consequently, the hippocampal mRNA expression of EGR-1 was analyzed. In Fig. 9B, mRNA expression level of EGR-1 in both control and ASIV-treated mice was elevated. However, the increase amplitude of the mRNA expression level of EGR-1 in ASIV treated mice was greater than that in the vehicle-treated control mice (t (11) = 2.327, P = 0.0400). These results suggested ASIV could increase the response of EGR-1 to external stimuli.

Discussion
Memory formation is a complex process. This is partly due to the information is assumed to be transiently stored in working memory (short-term memory) while it is considered to integrate long-term memory permanently [34]. At present, a lot of knowledge about memory comes from the study of memory impairment, especially amnesia. [35][36][37]. Other neurological diseases, such as Alzheimer's disease [38], Parkinson's disease [39,40] and Korsakoff's syndrome [41] also accompany with impaired memory. In addition, a common temporary failure of memory retrieval is often found under some speci c situations like emotion [42], aging [43], sleep [44] or other unrealized factors [45,46]. Therefore, drugs facilitating information process, consolidation, store or retrieval may contribute to the improvement of memory de cits under the aforementioned conditions. In our study, ASIV was found to facilitate the memory formation of mice as it promoted reference memory (long-term memory) and working memory (short-term memory) in RAMT, as well as procedural memory (long-term memory) in SBT, suggesting that ASIV might bene t memory loss occurred in neurological disorders.
LTD is the process that re ects the weakening of speci c synaptic transmission in order to constructively utilize synaptic reinforcement. It is a necessary process because, if allowed to continue increasing in strength, synapses would ultimately reach a ceiling level of e ciency, which would inhibit the encoding of new information. (Purves D (2008). Neuroscience (4th ed.). Sunderland, Mass: Sinauer. pp. 197-200. ISBN 0-87893-697-1). In this study, ASIV administration was displayed to enhance LTD which was induced by adding GABA A receptor blockers bicuculline. Meanwhile, ASIV decreased the hippocampal GABA level as well as the expression of GAD65, suggesting the weakening of GABAergic synaptic transmission, which was further con rmed by the reduced inhibitory synapse ratio. Previous study reported that blockade of GABAergic synaptic transmission enhances memory consolidation [47]. Therefore, our ndings indicated that ASIV enhanced memory of mice probably through modulating GABAergic inhibition, thus indirectly enhanced synaptic plasticity.
Expression of EGR-1 is indicated to be associated with LTD of synaptic e cacy. In recent years, the most remarkable conclusion is that EGR-1 not only plays a key role in different forms of memory, but also plays a key role in different processes operating on memory, from post learning memory consolidation and system consolidation to reconsolidation, updating, and extinction [48]. Chemical LTD induced in hippocampal slice cultures was shown to induce the expression of EGR-1 [9]. However, in vivo electrophysiological studies reported no evidence that the expression of EGR-1 was regulated after LTD induction in CA1 region [49]. In the study, EGR-1 was found to be decreased in mouse hippocampus after ASIV administration. However, a single-trail of SBT could boost the expression of EGR-1 rapidly. Meanwhile, ASIV administration for two weeks reduced LTD in hippocampal slices. Together, these results implicated that ASIV could decrease the basal level of EGR-1, which in turn increases its activitydependent capacity in response to external stimuli.
Previous studies have shown that expression of GAD65 is likely modulated by BDNF and TrkB at mRNA or protein level [50][51][52]. It can be imagined that the activity-dependent release of BDNF may affect the inhibition circuit by triggering GAD65 transcription, thereby increasing the synaptic level of GABA, thereby regulating the activity of neural networks [53]. Here we described that BDNF/TrkB was decreased together with the GABAergic inhibition of ASIV and this effect was blocked by EGR-1 knockout. Therefore, it might be possible that ASIV treatment decreased the expression levels of BDNF/TrkB rather favor a reduction for the GAD65 probably by inhibition of the EGR-1 activity.
Together, ASIV could enhance the memory of mice and improve hippocampal synaptic plasticity, which might be achieved through modulating hippocampal GABAergic inhibition mediated EGR-1. Our current work bene ts for the clinical application of ASIV in the prevention and treatment of memory-related diseases.

Declarations
Ethics approval and consent to participate All experiments on animals were performed according to the protocol approved by Animal Care and Use Committee of SHUTCM and all animals received humane care (Ethical approval no. SZY201610005). The animal-related procedures we studied are in accordance with the 1975 Declaration of Helsinki on animal rights (as revised in 2008).

Consent for publication
Not applicable.

Availability of data and material
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Competing of interests
The authors declare that they have no competing interests.   The ratio of inhibitory synapse in total synapse, ASIV administration decreased the ratio of inhibitory synapse/total synapse in mice hippocampus both for 2-week and 5-week, n = 4/group. *P < 0.05, **P < 0.01, the data are shown as the mean ± S.E.M.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. OnlineFig1020190603.Png