Oligomer β-amyloid Induces Hyperactivation of Ras to Impede NMDA Receptor-Dependent Long-Term Potentiation in Hippocampal CA1 of Mice

The activity of Ras, a small GTPase protein, is increased in brains with Alzheimer’s disease. The objective of this study was to determine the influence of oligomeric Aβ1-42 on the activation of Ras, and the involvement of the Ras hyperactivity in Aβ1-42-induced deficits in spatial cognition and hippocampal synaptic plasticity. Herein, we show that intracerebroventricular injection of Aβ1-42 in mice (Aβ-mice) enhanced hippocampal Ras activation and expression, while 60 min incubation of hippocampal slices in Aβ1-42 (Aβ-slices) only elevated Ras activity. Aβ-mice showed deficits in spatial cognition and NMDA receptor (NMDAR)-dependent long-term potentiation (LTP) in hippocampal CA1, but basal synaptic transmission was enhanced. The above effects of Aβ1-42 were corrected by the Ras inhibitor farnesylthiosalicylic acid (FTS). ERK2 phosphorylation increased, and Src phosphorylation decreased in Aβ-mice and Aβ1-42-slices. Both were corrected by FTS. In CA1 pyramidal cells of Aβ1-42-slices, the response of AMPA receptor and phosphorylation of GluR1 were enhanced with dependence on Ras activation rather than ERK signaling. In contrast, NMDA receptor (NMDAR) function and GluN2A/2B phosphorylation were downregulated in Aβ1-42-slices, which was recovered by application of FTS or the Src activator ouabain, and mimicked in control slices treated with the Src inhibitor PP2. The administration of PP2 impaired the spatial cognition and LTP induction in control mice and FTS-treated Aβ-mice. The treatment of Aβ-mice with ouabain rescued Aβ-impaired spatial cognition and LTP. Overall, the results indicate that the oligomeric Aβ1-42 hyperactivates Ras and thereby causes the downregulation of Src which impedes NMDAR-dependent LTP induction resulting in cognitive deficits.


HIGHLIGHTS
(1) Ras activity is increased by in vivo or in vitro Aβ  application.

INTRODUCTION
AD is the most frequent cause of dementia in the elderly, featured by progressive loss of memory and amyloid β (Aβ) peptides accumulation in the brain. The injection of synthetic Aβ 1-42 into the brain or overexpression of Aβ-generating fragments causes learning and memory deficits in various models, indicating a connection between Aβ accumulation and dementia. However, it remains to be clarified how these actions of oligomeric Aβ 1-42 impair learning and memory, since the memory deficits at early stages of Aβ accumulation are not always associated with massive neuronal degeneration and loss. High level of Ras, a small GTPase superfamily protein, is found in brains with AD (McShea et al., 2007). Isoprenylated GTPases have been reported to involve in the pathogenesis of AD (Ostrowski et al., 2007;Scheper et al., 2007). Ras interacts with its guanine nucleotide exchange factor to facilitate the conversion of inactive Ras-GDP to active Ras-GTP (Prior and Hancock 2012). Elevated Ras levels were found in early stages of AD (Gartner et al., 1999;Kirouac et al., 2017). The hyperactivation of Ras leads to the disorder of LTP induction (Stornetta and Zhu 2011). The localization of Ras and phosphorylated ERK were increased in plaques and tangles of AD brains (Ferrer et al., 2001;Pei et al., 2002). The Ras-MAPK signaling pathway has been found prior to the formation of plaques and tangles (Gartner et al., 1999) and to influence LTP formation (Ohno et al., 2001). Patients with neurofibromatosis type I (NF1) caused by loss-of-function mutations in the NF1 oncogene have hyperactive Ras, and most NF1 children have cognitive deficits (North et al., 2002). The pharmacological suppression of Ras signaling reverses the deficits in learning in mouse models of NF1 (Li et al., 2005;Cui et al., 2008). However, whether or how hyperactivation of Ras is involved in Aβ 1-42 -induced cognitive deficits remains to be elucidated.
An earlier study reported that tyrosine phosphorylation of NMDA receptors (NMDARs) was enhanced in mice lacking H-Ras (Manabe et al., 2000). Src is a novel H-Ras binding partner. The activation of Ras has been found to modify the downstream effector Src (Thornton et al., 2003). Hippocampal Src kinase activity was increased by H-Ras deficiency, and subsequently enhanced NMDAR function and facilitated LTP induction (Thornton et al., 2003). The administration of farnesyl transferase inhibitor in mice can enhance NMDAR GluN2A/ GluN2B phosphorylation through enhanced Src signaling . Farnesylthiosalicylic acid (FTS), a synthetic Ras inhibitor (Kloog et al., 1999), can dislodge Ras from its anchorage domains to prevent Ras-membrane interactions (Niv et al., 2002). The administration of FTS in mice through enhancing Src activity increases GluN2A/2B phosphorylation (Wang et al., 2018). The inhibition of Ras activity restored normal spine structural plasticity and presynaptic glutamate release in Aβ-treated neurons (Ye and Carew 2010). Therefore, it is speculated that oligomeric Aβ 1-42 impedes NMDAR-dependent LTP induction via Ras hyperactivity, resulting in cognitive deficits.
The objective of this study was to determine whether oligomeric Aβ 1-42 affects the activation of hippocampal Ras, and then to investigate how the hyperactivity of Ras is involved in Aβ 1-42 -induced damages in spatial cognition and hippocampal CA1 LTP induction, and to explore the underlying molecular mechanisms. Our results indicate that the oligomeric Aβ 1-42 causes hyperactivation of Ras, which in turn leads to the downregulation of Src to impede hippocampal NMDAR-dependent LTP induction, resulting in cognitive deficits.

Experimental Animals
All animal experiments complied with the ARRIVE guidelines, and were performed according to the National Institutes of Health guide for the care and use of Laboratory animals (NIH Publications No. 8023, revised 1978). 16-week-old (39.8 ± 1.2 g) and 4-week-old (17.8 ± 1.1 g) male mice (ICR) were in a regulated environment with a 12 h light/dark cycle (lights on at 9:00 A.M.).
FTS (Cayman chemical, Ann Arbor, USA) was injected (i.p.) at a dose of 3 mg/kg, because this dose of FTS was effective and could enter the brain within 20 to 30 min (Shohami et al., 2003). In in vitro experiments, FTS (4.5 μM) could inhibit Ras activity (Kloog et al., 1999), thus this study used five μΜ FTS to treat hippocampal slices.
NMDAR antagonist AP-V, MEK inhibitor U0126, Src-family kinases (SFKs) inhibitor PP2 and the Src activator ouabain (OU) were purchased from Sigma (St. Louis, USA). The slices were treated with PP2 (20 μN) (Karni et al., 2003), AP-V (20 μN), U0126 (10 μN) and ouabain (5 μN) (Tian et al., 2006). PP2 is reported to pass through the blood-brain barrier in some special animal models (Schumann et al., 2008). However, the injection (i.p.) of PP2 at the dose (0.03 mg/kg) did not alter the levels of NMDAR GluN2A/2B phosphorylation in hippocampus . In this study, the mice were treated daily with the injection (i.c.v.) of PP2 (1.2 nmol/3 μl per mouse) for eight consecutive days. For repeated injection (i.c.v.), mice were anesthetized with an injection (i.p.) of ketamine (100 mg/kg)/ xylazine (10 mg/kg), and placed into a stereotaxic frame (Stoelting). A small hole (2 mm diameter) was drilled in the skull using a dental drill. A stainless steel guide cannula (26-G, Plastics One, Inc., Roanoke, VA) was implanted in the right lateral ventricle (0.22 mm caudal to bregma, 1 mm lateral to the midline, and 1 mm ventral to the pial surface) and anchored to the skull with stainless steel screws and dental cement. On day three after surgery, PP2 was injected using an infusion cannula (30-G) coupled to a motorized injector (Stoelting, Wood Dale, IL, USA) at a rate of 0.5 μL/min. The infusion cannula was left in place for 5 min after injection. Control mice were treated with the injection (i.c.v.) of vehicle (same volume). Ouabain, was administered intraperitoneally at the dose of 1 μg/kg to mice subjected to traumatic brain injury, could improve the neurological function (Dvela-Levitt et al., 2014). In this study, the mice were injected (i.p.) daily with ouabain (1 μg/kg).
Hippocampal slices (approximately four to five slices per mouse) obtained from 4-week-old control mice (n 136) were treated with Aβ 1-42 and various drugs or vehicle, and then divided into two experimental groups ( Figure 1B): The first group was used to examine Ras activities, ERK/Src phosphorylation, expression and phosphorylation of GluR1/2 or GluN2A/2B; the second group was used to examine AMPAR/NMDAR function.

Behavior Analysis
Morris water maze task (MWM): A pool (diameter 120 cm) was prepared with black-colored plastic. The water temperature (22 ± 1°C) of pool was regulated using a bath heater. For the hiddenplatform test, a cylindrical dark-colored platform (diameter 7 cm) was placed in 0.5 cm below the water surface. The trial was stopped when the mouse did not reach the platform within 90 s. Each mouse started pseudorandomly from one of four quadrants. Four trials with an interval of 30 min were conducted each day. The probe trial was performed after the hidden-platform test at 24 h. A computer with a video camera system was used to record the swimming paths and time. In the hidden-platform test, latency to reach the platform and swimming speed were measured.
Y-maze task: The mouse could move freely within 8 min. When the hind paws were completely placed in the arm, arm entry was counted. Alternation was determined as successive Frontiers in Pharmacology | www.frontiersin.org December 2020 | Volume 11 | Article 595360 entries into the three arms on overlapping triplet sets. The percentage alternation was calculated as the ratio of actual to possible alternations (defined as the total number of arm entries minus two).

Histological Examination and Analysis
After anesthetized with ketamine (100 mg/kg)/xylazine (10 mg/kg, i. p.), the mice were perfused transcardially with 4% paraformaldehyde. The removed brains were postfixed overnight in the 4% paraformaldehyde. The paraffin embedding was processed, and then the coronal sections (5 μm) were cut. The hippocampal sections were stained by toluidine blue. We used a conventional light microscope (Olympus DP70, Japan, × 40) to observe the hippocampal pyramidal cells. For the analysis of cell quantification, every fourth section was stained by toluidine blue. The healthy pyramidal cells was counted throughout hippocampal CA1 regions (n 10 sections per brain) using the manual tag function of Image Pro-Plus 6 (Media Cybernetics). The density of cells was expressed as cell number per square millimeter (Dawson et al., 2003).

Electrophysiological Analysis
Slice preparations: Hippocampal slices were obtained from 16week-old mice for field potential recording or 4-week-old mice for whole cell patch-clamp recording. The brains were kept in icecold and oxygenated ACSF (in mM: NaCl 126, CaCl 2 1, KCl 2.5, MgCl 2 1, NaHCO 3 26, KH 2 PO 4 1.25, and D-glucose 20, pH: 7.4) for 10 min. The coronal slices (400 μm) of hippocampus were cut using a vibrating microtome. The slices were recovered for 1 h, and then transferred to a recording chamber of 30 ± 1°C.
Cell patch-clamp recording: AMPA-evoked currents (I AMPA ) and NMDA-evoked currents (I NMDA ) were induced in hippocampal CA1 pyramidal cells using a picospritzer (rapid drug delivery system) and recorded using an EPC-10 amplifier (HEKA Elektronik, Germany) as described (Zhou et al., 2016;Wang et al., 2018). A glass pipette was filled with an internal solution (pH 7.2) (in mM: Cs-gluconate 120, NaCl 2, MgCl 2 4, Na 2 -ATP 4, HEPES 10, EGTA 10). The holding potential was kept at -60 mV. I AMPA was induced by AMPA (1-300 μM) in the same neuron to produce dose-response curve. To record I NMDA , the slices were perfused with the oxygenated magnesium-free ACSF. NMDA (1-1,000 μM) was applied to I NMDA . The AMPAR antagonist CNQX (5 μM) or the NMDAR antagonist AP-V (20 μM) was used to verify the obtained I AMPA and I NMDA .

Data Analysis
SPSS software was used to perform Statistical analyses. The data were expressed as the means ± standard error (SE). Repeated measures ANOVA were used to analyze the behavioral and electrophysiological data. The actual values of the ANOVAs (one-way ANOVA, two-way ANOVA, repeated measures ANOVA) followed by Bonferroni post hoc analysis are shown in the section of Results. Differences at p < 0.05 were considered statistically significant.

Aβ 1-42 Impairs Spatial Cognition Through Enhanced Ras Activation
Spatial cognitive behaviors were examined in mice injected (i.c.v.) with Aβ 1-42 (Aβ-mice) (n 8 per group; Figure 1A). The repeated measures ANOVA revealed a progressive decline in the escape latency to reach the hidden platform in the MWM with training days in all groups (F (4, 112) 41.460, p < 0.001; Figure 2A). In comparison with control mice, the escape latency was significantly increased in Aβ-mice (F (1, 14) 13.095, p 0.003) without changes in swimming speed (p > 0.05). A probe trial was carried out to measure the swimming time spent in four quadrants. Notably, the swimming time in the target quadrant was reduced in Aβ-mice compared with control mice (p 0.042; Figure 2B). In Y-maze, the alternation ratio in Aβ-mice was lower compared with control mice (p 0.038; Figure 2C).
The Ras-GTP pull-down assay using a Ras effector protein that recognizes the active state of Ras was performed to examine the Frontiers in Pharmacology | www.frontiersin.org December 2020 | Volume 11 | Article 595360 influence of Aβ 1-42 on the expression and activity of hippocampal Ras (n 6 per group). Compared with control mice, the level of Ras protein in Aβ 1-42 -mice was increased by approximately 30% (p 0.049; Figure 2D) and the level of Ras-GTP was elevated at least 2-fold (p < 0.001). FTS has been demonstrated to selectively disrupt the interactions of active Ras proteins with the plasma Treatment of Aβ-mice with FTS ( Figure 1A) corrected the increase in escape latency in the MWM (p 0.009) and the decline in swimming time in the target quadrant of the probe trial (p 0.032). Subsequently, the alternation ratio in the Y-maze was restored in Aβ 1-42 /FTS-mice (p 0.027). By contrast, FTS-mice

Aβ 1-42 -Activated Ras Inhibits NMDAR Function via Downregulation of Src
Subsequently, we examined NMDAR activity in hippocampal CA1 pyramidal cells (n 6 cells/6 slices/4 mice per group). In the presence of AMPAR antagonist CNQX, the application of NMDA evoked an inward current (I NMDA ). The densities of I NMDA showed a dose-dependent difference (F (5,75) 73.185, p < 0.001; Figure 6A). The densities of I NMDA in Aβ 1-42 -slices were less than those in control slices (F (1,10) 31.472, p < 0.001), which was rescued by the addition of FTS (p 0.004; Figure 6B) but not U0126 (p > 0.05). The protective effect of FTS on the densities of I NMDA in Aβ 1-42 -slices was sensitive to the use of PP2 (p 0.017). In addition, the density of I NMDA in FTS-slices was higher than in the control slices (p 0.026), which was corrected by the use of PP2 (p < 0.001). Haas et al. (2002) reported previously that ouabain binding to the Na+/K + -ATPase activated Src kinase in several different cell lines. The exposure to 5 μM ouabain for 30 min failed to increase the density of I NMDA in control slices (p > 0.05). However, the treatment with ouabain could correct the density of I NMDA in Aβ 1-42 -slices (p 0.043), which was sensitive to PP2 (p 0.028).
The administration of PP2 in Aβ 1-42 /FTS-mice (n 8 mice per group) prevented the protective effects of FTS on the escape latency of the MWM (p 0.038; Figure 7D), the swimming time in the target quadrant of the probe trial (p 0.026; Figure 7E) and the alternation ratio in the Y-maze (p 0.032; Figure 7F). In addition, PP2-treated control mice showed deficits in spatial cognition, such as increased escape latency in the MWM (p 0.037) and reduced swimming time in the target quadrant (p 0.031) or alternation Frontiers in Pharmacology | www.frontiersin.org December 2020 | Volume 11 | Article 595360 ratio (p 0.045). The treatment with ouabain in Aβ 1-42 -mice recovered the escape latency of the MWM (p 0.034), the swimming time in the target quadrant of the probe trial (p 0.041) and the alternation ratio in the Y-maze (p 0.048), whereas it did not affect the spatial cognition in control mice (p > 0.05).
Frontiers in Pharmacology | www.frontiersin.org December 2020 | Volume 11 | Article 595360 hippocampal LTP induction, which were improved by the administration of FTS. These results manifest that the oligomeric Aβ 1-42 impairs spatial cognition via the hyperactivation of Ras. In the hippocampus of Aβ 1-42 -mice, the level of Ras protein was increased by approximately 30%, while the level of Ras-GTP was elevated at least 2-fold. In hippocampal slices, the 60 min bath application of Aβ 1-42 could enhance the activation of Ras, although it failed to alter the level of Ras protein. The results provide a clear indication that oligomeric Aβ 1-42 not only enhances Ras expression but also stimulates Ras activation. Because Aβ binds to specific neuronal membrane APP, it has been suggested that APP is a possible receptor for Aβ (Lorenzo et al., 2000). Aβ can induce phosphorylation on APP to enhance APP proteolysis (Kirouac et al., 2017). The cytoplasmic YENPTY motif of APP is known to be a docking site for the adaptor proteins Shc and Grb2, which recruit the GEF SOS2 for activation and expression of Ras (Kirouac et al., 2017). Andras and Toborek (2013) reported that treatment with Aβ (1 μM) for 3 min caused redistribution of lipid rafts and caveola, which elevated the level of Ras-GTP. Acute treatment with Aβ 1-42 can cascade ERK signaling in hippocampal neurons (Dineley et al., 2001). APP can positively regulate the Ras-ERK signal pathways in neuronal cell lines (Chaput et al., 2012). York et al. (1998) reported that ERK can be persistently activated by the formation of a stable upstream complex between small G proteins. In this study, either in vivo or in vitro (60 min) treatment with Aβ 1-42 caused an increase in the levels of hippocampal ERK2 phosphorylation without changes in the expression levels, which were sensitive to Ras inhibition. The Ras isoform H-Ras may contribute to the long-lasting activation of ERK2 (Krapivinsky et al., 2003). It is conceivable that Aβ 1-42, through the hyperactivation of Ras, causes a rapid and sustained increase in the ERK signaling. The activation of NMDAR-mediated CaMKII can cascade the Ras-ERK signaling pathway (Carlisle et al., 2008). However, NMDAR function was downregulated in Aβ 1-42 -slices, and the blockade of NMDAR did not affect ERK2 phosphorylation. Dineley et al. (Dineley et al., 2001) reported, in the hippocampus, that chronic exposure to Aβ 1-42 increased α7 nicotinic acetylcholine receptor (nAChR) protein, which led to chronic stimulation of ERK2. Therefore, whether Aβ 1-42 -induced hyperactivation of Ras increases ERK2 phosphorylation through enhanced α7nAChR function should be an interesting topic for future work. The basal transmission of hippocampal CA3-CA1 synapses was enhanced in Aβ 1-42 -mice with no change in the capability of presynaptic glutamate release which depended on the hyperactivation of Ras rather than ERK. Postsynaptic AMPAR plays an important role in hippocampal synaptic transmission (Schwenk et al., 2014). AMPAR is synthesized dendritically and inserted into the synaptic membrane (Ju et al., 2004). The 60 min application of Aβ 1-42 caused an obvious increase in AMPAR function and GluR1 phosphorylation, which were sensitive to Ras inhibition. The hyperactivation of Ras drives the synaptic insertion of AMPAR by triggering the phosphorylation of GluR1 at S845 and S831 (Zhu et al., 2002). Manabe et al. (2000) reported that the AMPAR synaptic responses were unchanged in CA1 pyramidal neurons of H-Ras knockout mice. Qin et al. (2005) found that the ERK1/2 phosphorylation and subsequent activation of CREB can increase AMPAR function. Enhanced ERK activity increases the expression of the GluR1 subunits (Schumann and Yaka 2009). The decline in CREB phosphorylation caused a specific downregulation of postsynaptic GluR1 (Borges and Dingledine 2001). However, the increases in AMPAR function and GluR1 phosphorylation in Aβ 1-42 -slices were insensitive to the inhibition of MEK. Furthermore, the levels of GluR1 proteins in Aβ 1-42 -slices were unchanged. A-type K+ channels have been reported to be a substrate of ERK1/2 (Schrader et al., 2006). The phosphorylation of K+ channels is known to be a primary mechanism for the ERK1/2-decreased A-type currents . We observed that the inhibition of MEK failed to alter the increased fEPSP slopes in Aβ 1-42 -mice (data not shown).
NMDARs are linked to cognitive impairments in AD (Goussakov et al., 2010). The expression of GluN2A and GluN2B was downregulated in AD patient brains (Hynd et al., 2004). Aβ can induce the dysfunction of hippocampal NMDAR via the inactivation of GluN2B (Yin et al., 2015). The NMDAR activation is negatively regulated by the Ras signaling (Li et al., 2005). H-Ras overexpression decreases tyrosine phosphorylation of NMDAR GluN2A (Thornton et al., 2003). By inhibiting RACK1, a Ras effector protein, NMDAR currents in hippocampal neurons were increased (Yaka et al., 2002). Ras activation induces a decline in the Src autophosphorylation (Thornton et al., 2003). Src tyrosine phosphorylation positively regulates NMDAR channel activity . We observed that the 60 min application of Aβ 1-42 suppressed the phosphorylation of Src and GluN2A/2B, leading to NMDAR dysfunction, which could be corrected by the inhibition of Ras or the activation of Src rather than the inactivation of MEK. The deficits in hippocampal NMDAR-dependent LTP and spatial cognition in Aβ 1-42 -mice could be rescued by the inhibition of Ras. The inhibition of Src not only reduced the phosphorylation of GluN2A/2B and NMDAR activity and depressed LTP induction and spatial cognition in control mice, but also blocked the NMDAR-dependent LTP and spatial cognition in Aβ 1-42 -mice treated with the Ras inhibitor. Moreover, the administration of Src activator at a low dose could rescue the Aβ-impaired spatial cognition and NMDAR-dependent LTP. In addition, treatment of rat brain slices with Tat-H-Ras decreased Src phosphorylation of GluN2A, and decreased the magnitude of hippocampal LTP (Thornton et al., 2003). The induction of hippocampal NMDAR-dependent LTP has been widely thought to be the basis for explicit memory formation and storage (McGaugh 2000). Thus, the results in this study give an indication that Aβ 1-42 through suppressed Src activation blocks NMDAR-dependent LTP induction leading to impairment of spatial cognition. Ras-ERK signaling is considered to be critical for LTP induction (Patterson et al., 2010) since the inhibition of Ras or ERK blocks LTP induction (Zhu et al., 2002). Ras-ERK signaling is reported to induce the activation of L voltagesensitive Ca2+ channel (L-VGCC) (Ekinci et al., 1999), which can enhance L-VGCC-dependent LTP (Moosmang et al., 2005).

CONCLUSION
The isoprenylated GTPases have been associated with the pathogenesis of AD (Ostrowski et al., 2007;Scheper et al., 2007). Ras levels are increased in brains with AD (McShea et al., 2007). In the current study, we provided the first in vivo and in vitro evidence that the Aβ 1-42 -induced abnormal increase in Ras activity may account for the impairment of NMDARdependent LTP induction and spatial cognition through the downregulation of Src, suggesting that targeting Ras signaling may be an effective strategy to protect hippocampal synaptic plasticity against Aβ-induced cognitive decline at early stages of AD.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The animal study was reviewed and approved by Ethical Committee of the Nanjing Medical University.

AUTHOR CONTRIBUTIONS
YW performed the electrophysiological experiments and the preparation of the manuscript. ZS performed all statistical analysis. YZ undertook the western blot analysis. JY carried out the animal care and the behavioral examinations. LC and WY carried out the experimental design and revision of the manuscript.