Available online at www.sciencedirect.com

www.elsevier.com/locate/brainres

b r a i n r e s e a r c h 1 6 2 5 ( 2 0 1 5 ) 6 4 – 7 2

http://dx.doi.org/10.0006-8993/& 2015 El

Abbreviations: OS

i, intracellular calc

response element b

monophosphate resnCorrespondence

Suzhou 215004, ChiE-mail address:1These authors c

Research Report

CIH-induced neurocognitive impairmentsare associated with hippocampal Ca2þ overload,apoptosis, and dephosphorylation of ERK1/2and CREB that are mediated by overactivationof NMDARs

Jing Wanga,1, Hong Minga,1, Rui Chena,n, Jing-mei Jua, Wan-da Penga,Guo-Xing Zhangb, Chun-feng Liuc

aDepartment of Respiratory Medicine, Sleeping Center, Second Affiliated Hospital of Soochow University, Suzhou,Jiangsu, ChinabDepartment of Physiology, Medical College of Soochow University, Suzhou, Jiangsu, ChinacLaboratory of Aging and Nervous Diseases, Institute of Neuroscience, Soochow University, Suzhou, Jiangsu, China

a r t i c l e i n f o

Article history:

Accepted 13 August 2015

Chronic intermittent hypoxia (CIH) is commonly seen in patients with obstructive sleep

apnea, and has been hypothesized to underlie the neurocognitive dysfunction in these

Available online 21 August 2015

Keywords:

Chronic intermittent hypoxia

Neurocognitive impairments

Calcium overload

NMDA

Receptors

Memantine

1016/j.brainres.2015.08.01sevier B.V. All rights rese

A, obstructive sleep ap

ium concentration; ERK1

inding protein; p-ERK 1/2

ponse element binding pto: Department of Respna.chenruigood@126.com (Rontributed equally to th

a b s t r a c t

patients. However, its cellular and molecular mechanisms remain to be defined. The

present study aimed to investigate, in a mouse CIH model, the role of NMDA receptor

(NMDAR) activation in mediating the CIH-induced neurocognitive impairments, caspase

expression and dysregulated Ca2þ signaling pathways in hippocampus. Male ICR mice

(n¼45) were exposed to CIH (8 h/day) or room air (control) for 4 weeks. After 4-week

treatment, neurobehavioral assessments were performed by Morris water maze test,

hippocampal [Ca2þ]i was evaluated by flow cytometry; and protein expressions of

caspase-3, caspase-9, PARP, p-ERK1/2 and p-CREB in hippocampus were measured by

Western blotting. Our results showed that, compared to control animals, 4-week exposure

to CIH produced significant spatial learning and memory impairments in CIH mice.

Increased caspase expression in hippocampus was observed in CIH mice associated with

significant elevation of [Ca2þ]i and dephosphorylation of ERK and CREB expression. When

the NMDAR antagonist memantine was administered by intraperitoneal injection prior to

daily exposure to CIH, at a sub-therapeutic dose of 5 mg/kg/day not shown to impact the

2rved.

nea; CIH, chronic intermittent hypoxia; NMDAR, N-methyl-D-aspartate receptor; [Ca2þ]

/2, extracellular-signal-regulated kinase 1/2; CREB, cyclic adenosine monophosphate

, phosphorylated extracellular-signal-regulated kinase 1/2; p-CREB, cyclic adenosine

roteiniratory Medicine, Second Affiliated Hospital of Soochow University, 1055 Sanxiang Road,

. Chen).is work.

b r a i n r e s e a r c h 1 6 2 5 ( 2 0 1 5 ) 6 4 – 7 2 65

neurobehavioral performance in control animals, the neurocognitive impairments as well

as the neurobiochemical changes were abolished or normalized in the CIH mice. Our study

suggests that overactivation of NMDARs and the Ca2þ overload-dependent ERK/CREB

dysregulation is one of the important mechanisms in mediating the CIH-induced

neurocognitive impairments.

& 2015 Elsevier B.V. All rights reserved.

1. Introduction

Obstructive sleep apnea (OSA) is a common type of sleepdisorder in which the upper airway collapses repeatedly duringsleep, resulting in chronic intermittent hypoxia (CIH) andassociated sleep fragmentation (Young et al., 1993). In additionto cardiovascular and metabolic morbidities, OSA can cause abroad range of neurocognitive deficits from attention andvigilance impairment to memory and executive dysfunctions(Chen et al., 2011; Jackson et al., 2011; Kheirandish-Gozal et al.,2013), which can profoundly impact the individual work andquality of life. In rodents, exposure to CIH during sleep thatproduced the similar nocturnal hypoxia/re-oxygenation cyclesin OSA have been shown to manifest not only metabolicdisorders (Messenger et al., 2013) but also certain features ofneurocognitive dysfunction seen in OSA patients, such aslearning and memory deficits and impaired vigilance (Gozalet al., 2001; Row et al., 2002; Nair et al., 2011). Multiplepathophysiological processes have been proposed to contributeto the cognition dysfunction associated with CIH, and theseinclude increased oxidative stress (Row et al., 2003; Xu et al.,2004) and inflammation (Ryan et al., 2007), altered gene regula-tion (Kheirandish et al., 2005a), and dysregulation of the cellularand molecular substrates of synaptic plasticity (Xie et al., 2010).However, the exact cellular and molecular mechanisms under-lying the CIH-induced neurocognitive impairments remain to bedefined.

Glutamatergic neurotransmission plays a vital role in

mediating synaptic plasticity and receptor function in theregulation of learning and memory. N-methyl-D-aspartate

receptors (NMDARs) are cation channels gated by the neuro-

transmitter glutamate and the essential regulators of neuro-

nal development, synaptic transmission and plasticity (Bliss

and Collingridge, 1993; Aamodt and Constantine-Paton, 1999).

In neurodegenerative diseases such as Huntington’s disease

and Alzheimer’s disease (AD), ischemic stroke and hypoxic

stress, excessive release of glutamate results in an over-

activation of NMDARs and the downstream effectors that

are implicated in neuronal damage and cell death. This

process is known as synaptic excitotoxicity (Olney et al.,1971; Huo et al., 2014). Choi (1987), Choi et al. (1987, 1988)

has reported that the overactivation of NMDARs and subse-

quent neuronal Ca2þ overload trigger a cascade of down-

stream events leading to excitotoxicity. The Ca2þ dependent

signaling pathways are the key component of the molecular

and cellular mechanisms underlying learning and memory,

and these include Ca2þ/calmodulin-dependent protein (CaM)

kinase, Ras-extracellular-signal regulated kinase (ERK) 1/2

signaling pathways and the transcription factor cyclic-AMP

response element binding protein (CREB). Dysregulation ofneuronal Ca2þ homeostasis resulting in abnormal Ca2þ-dependent signaling transduction, mitochondrial dysfunctionand programmed cell death has been hypothesized to con-tribute to the neurocognitive impairments in a number ofCNS disease states (Kalia et al., 2008; Wroge et al., 2012).

Recently, Huo and his colleagues reported that rat exposedto chronic intermittent hypoxia-hypercapnia (CIHH) displayedlearning and memory deficits, which may be caused byexcitotoxicity with increased NR2B subunit expression anddysregulated downstream signaling cascade. CIHH is an ani-mal model of chronic obstructive pulmonary disease (COPD),and it differs from CIH in that the CO2 level in CIHH model ismuch higher compared to CIH model (Huo et al., 2014). Hence,the role of synaptic excitotoxicity in CIH-induced neurocogni-tive impairments warrants further evaluations.

In the present research, we aimed to examine for the firsttime the role of NMDARs overactivation and the abnormal-ities of Ca2þ-dependent signaling pathways in neurocognitivedysfunction observed in a mouse model of CIH. The effect of4-week CIH (8 h/day) on the neurobehavioral performancewas assessed by the Morris water maze test, as well as theassociated changes in intracellular calcium concentration([Ca2þ]i), downstream protein expression of ERK1/2, CREB,and apoptosis in the hippocampus. In addition, we examinedthe effect of memantine, a noncompetitive NMDAR antago-nist, on the CIH-induced neurocognitive deficits and dysre-gulation of [Ca2þ]i and ERK/CREB protein expression todetermine the importance of NMDARs overactivation inmediating these changes and evaluate the underlyingmechanisms.

2. Results

2.1. Effect of CIH and memantine pretreatment on MWMtest

In the invention phase of the present study as described inSection 4, a total of 45 male ICR mice (4-week old, 15 animalsper group) exposed to CIH or room air (RA) for 4 weeks wastested for learning and memory ability using the Morris watermaze (MWM) test. In the place navigation test, CIHþVEHgroup showed significantly longer escape latencies (Day 3:40.9172.95 s vs. 31.5572.78 s, po0.05; Day 4: 37.4973.20 s vs.22.4872.77 s, po0.01) and longer distance travelled (Day 3:1045.07784.40 cm vs. 751.26767.87 cm, po0.05; Day 4:922.39785.46 cm vs. 579.65772.70 cm, po0.05) comparedwith RA group. Pre-treatment with memantine (CIHþMEM)

b r a i n r e s e a r c h 1 6 2 5 ( 2 0 1 5 ) 6 4 – 7 266

at a low dose of 5 mg/kg/day without a significant effect innormal control mice (see below) resulted in significantlyshorter escape latencies and distance travelled compared tothe CIHþVEH group on Day 4 (latency: 26.3772.85 s vs.37.4973.20 s, po0.01; distance travelled: 647.31772.69 cmvs. 922.39785.46 cm, po0.05) (Fig. 1A and B). The swim speedshown over the course of 4 training days during the MWMtest was overall consistent without clear pattern of majordifferences among all 3 animal groups examined, albeit asmall significant difference was noted only on Day 4 (Fig. 1C).

During the spatial probe test, percent time in the targetquadrant was significantly less in CIHþVEH animals comparedwith RA group (21.0671.37% vs. 38.8773.92%, po0.01), withfewer number of platform crossings than RA animals(1.1370.22 vs. 2.4770.31, po0.01). Memantine pretreatment alsosignificantly reduced the impairment of CIH on the performancein spatial memory ability by increasing both percent time in thetarget quadrant (30.8572.02% vs. 21.0671.37%, po0.05) and

Fig. 1 – Learning and memory ability in RA, CIHþVEH and CIHþescape latencies to locate the target platform during place trainilonger escape latencies compared with RA group. However, CIHcompared with CIHþVEH group. (B) and (C) Noted mean swim sgroup displayed longer distance and slower speed compared wiin target quadrant and mean number of platform crossings durindisplayed fewer percent time in target quadrant and number ofCIHþMEM group had significantly increased percent time in tarcompared with CIHþVEH group. Each bar represented mean7SCIHþVEH group.

number of platform crossings (2.0770.25 vs. 1.1370.22, po0.05)compared with CIHþVEH group (Fig. 1D and E).

We also conducted a series of pre-experiment to assess theeffects of low dose memantine (MEM) on neurocognitivefunction in normal animals. Male ICR mice (n¼42) in roomair were randomized into memantine (5 mg/kg/day) treatmentgroup or vehicle group, and neurobehavioral assessments wereperformed by Morris water maze test after 4-week treatment.The results supported that intraperitoneal injection of mem-antine at a sub-therapeutic dose of 5 mg/kg/day selected forthe intervention phase of CIH experiment did not produce ameasurable or meaningful effect on the neurocognitive perfor-mance in normal control mice (Fig. S1, see online supplement).

2.2. Effect of CIH and memantine pretreatment on the levels ofcaspase-3, caspase-9 and PARP in hippocampusThe elevated expression of caspase-3, caspase-9 and PARP isrecognized as a universal marker of apoptosis. As described

MEM groups as tested by MWM test (n¼15/group). (A) Meanng in three groups. Mice in CIHþVEH group had significantlyþMEM group had significantly shorter escape latenciespeed and distance traveled in each group. Mice in CIHþVEHth the other two groups. (D) and (E) Noted mean percent timeg spatial probe test in three groups. Mice in CIHþVEH groupplatform crossings compared with RA group, howeverget quadrant and mean number of platform crossingsEM, npo0.05, nnpo0.01 vs. RA group; #po0.05, ##po0.01 vs.

Fig. 2 – The expression of caspase-3, caspase-9 and PARP in hippocampus in RA, CIHþVEH and CIHþMEM groups as measuredby Western blotting (n¼4/group). The relative density of caspases expression was normalized by GAPDH level. (A) Sampleimmunoblots probed for caspase-3, caspase-9, PARP and GAPDH are showed above. (B)–(D) Noted CIH up-regulated theexpression of caspase-3, caspase-9, PARP compared with RA group, however pretreatment memantine inhibited theirexcessive expression in the hippocampus. Each bar represented mean7SEM, npo0.05, nnpo0.01 vs. RA group; #po0.05,##po0.01 vs. CIHþVEH group.

b r a i n r e s e a r c h 1 6 2 5 ( 2 0 1 5 ) 6 4 – 7 2 67

in Section 4, the relative density of hippocampal caspaseexpression was normalized by GAPDH level in Western blotexperiments. The expression of caspase-3 was significantlyincreased in the hippocampus of CIHþVEH animals com-pared to the RA animals (0.6670.04 vs. 0.2370.04, po0.01). Inaddition, data showed similar significantly increased expres-sion of caspase-9 (0.6170.06 vs. 0.3070.08, po0.01) and PARP(0.7170.09 vs. 0.2170.01, po0.05) in CIH group. In contrast,pretreated with memantine abolished the elevated hippo-campal expression of caspase-3, caspase-9 and PARP com-pared with CIHþVEH mice (caspase-3: 0.2970.10 vs.0.6670.04, po0.01; caspase-9: 0.3070.06 vs. 0.6170.06,po0.01; PARP: 0.3270.05 vs. 0.7170.09; po0.05) (Fig. 2).

2.3. Effect of CIH and memantine pretreatment on theintracellular calcium concentration ([Ca2þ]i) of hippocampalcellsExposure to 4-week CIH induced a marked elevation in [Ca2þ]iin hippocampus, as assessed by the fluorescence densityof [Ca2þ]i compared to the RA group (155.0076.12 vs.91.9074.14, po0.01). Pre-treatment with memantine pre-vented the [Ca2þ]i elevation induced by CIH exposure(90.1373.78 vs. 155.0076.12, po0.01 vs. CIHþVEH group)(Fig. 3).

2.4. Effect of CIH and memantine pretreatment on the level ofp-ERK1/2 and p-CREB in hippocampusThe relative density of p-ERK1/2 and p-CREB expression wasnormalized by the total ERK1/2 and CREB protein levels,respectively, in Western blot experiments. As shown inFig. 4, the exposure to 4-week CIH significantly decreased p-ERK1/2 levels (0.6870.06 vs. 1.3970.10, po0.05) and p-CREBlevels (0.2970.04 vs. 0.7670.11, po0.05) compared with RAgroup, whereas such CIH-induced reduction was abolished inCIHþMEM mice pretreated with memantine (p-ERK1/2:1.2870.04 vs. 0.6870.06, po0.05; p-CREB: 0.6470.10 vs.0.2970.04; po0.05 vs. CIHþVEH group) (Fig. 4).

3. Discussion

Neurocognitive impairment in OSA patients has been increas-ingly recognized, and a better understanding of its underlyingmechanisms and potential intervention is clearly warranted as aresult of its profound impact on individual work and quality oflife (Rada, 2005; Gottlieb et al., 2010; Jackson et al., 2011). In thepresent study, exposure to 4-week CIH resulted in significantspatial learning and memory deficits, as evidenced by impairedneurobehavioral performance of CIH mice in the Morris water

Fig. 3 – The concentration of intracellular calcium concentration ([Ca2þ]i) in hippocampus in RA, CIHþVEH and CIHþMEMgroups as measured by flow cytometry (n¼4/group). (A) Representative photographs showed differences in the [Ca2þ]iconcentration in hippocampal cells in all three groups. X-axis represents the intensity of fluo-3 fluorescence (arbitrary unit),while Y-axis specifies the number of Hippocampal cells. (B) The value presented was the mean fluorescence intensity inducedby the calcium-sensitive fluorochome, Fluo-3. CIH induced a remarkable elevation in the fluorescence density of [Ca2þ]i inhippocampal cells, pre-treatment with MEM turned the [Ca2þ]i back to the control levels. Each bar represented mean7SEM,nnpo0.01 vs. RA group; ##po0.01 vs. CIHþVEH group.

Fig. 4 – The expression of p-ERK1/2 and p-CREB in hippocampus in RA, CIHþVEH and CIHþMEM groups as measured byWestern blotting (n¼3/group). Sample immunoblots probed for p-ERK1/2, ERK1/2, p-CREB and CREB are showed above. Therelative density of p-ERK1/2 and p-CREB expression was normalized by ERK1/2 and CREB level, respectively. CIH significantlydecreased the expression of p-ERK1/2 and p-CREB compared with RA group, but pretreatment with memantine increases theexpression of p-ERK1/2 and p-CREB compared with CIHþVEH group. Each bar represented mean7SEM, npo0.05 vs. RA group;#po0.05 vs. CIHþVEH group.

b r a i n r e s e a r c h 1 6 2 5 ( 2 0 1 5 ) 6 4 – 7 268

maze compared to control animals. These CIH-induced neuro-cognitive dysfunction was associated with abnormal elevationof [Ca2þ]i, increased expression of caspases, and dephosphor-ylation of ERK1/2 and CREB in hippocampus. Pretreatment of

CIH mice with memantine, a noncompetitive NMDAR antago-nist, however, effectively prevented or abolished these CIH-induced changes in neurobehavioral performance and theassociated neurobiochemical changes in hippocampus. Our

b r a i n r e s e a r c h 1 6 2 5 ( 2 0 1 5 ) 6 4 – 7 2 69

findings demonstrated that NMDAR overactivation is one of theimportant mechanisms underlying neurocognitive impairmentsin a mouse model of CIH, which triggers hippocampal calciumoverload, dephosphorylation of ERK1/2 and CREB, and increasedexpression of caspases.

CIH as the hallmark of OSA has been postulated to triggerand/or precipitate a variety of morbidities experienced inthese patients (Neubauer, 2001; Prabhakar et al., 2001; Row,2007). In rodent models, animals exposed to CIH manifest anumber of neurocognitive deficits characteristic of OSA. Forexample, it was reported that rodent exposed to intermittenthypoxia during sleep cycle was associated with poor neuro-behavioral performance involved in learning and memory,attention, and locomotor activity (Gozal et al., 2001; Rowet al., 2002; Nair et al., 2011). Subsequent studies haveconfirmed that CIH exposure could impair memory of rodentsto different degrees, as measured by the conventional watermaze tests (Gozal et al., 2003; Kheirandish et al., 2005b). Inthis study, a CIH mouse model was established that success-fully produced neurocognitive impairments. After 4-weekexposure to CIH, these animals manifested significantlylonger escape latencies, less percent time in target quadrantand fewer crossings compared to the control animals inMWM tests, demonstrating spatial learning and memorydeficits. These changes in MWM test characteristics of neu-rocognitive dysfunction were detected without markedchanges in swim speed, suggesting no contribution of loco-motor effect to the observed neurocognitive performancedeficits. The experimental findings from our CIH mousemodel are consistent with prior reports and support its useto further investigate the mechanisms underlying CIH-induced neurocognitive impairments.

There are a number of hypotheses related to the patho-physiology of CIH contributing to neurocognitive dysfunction,with multiple complex mechanisms, including increasedoxidative stress, inflammation and/or impaired synapticplasticity (Row et al., 2003; Ryan et al., 2007; Xie et al., 2010;Xu et al., 2004). The NMDARs play a key role in neuronalplasticity, learning and memory. It has been reported thatNMDAR overactivation triggers excitotoxicity leading to neu-ronal death or degeneration that are documented in anumber of neurological disorders, including chronic neuro-degenerative diseases, hypoxic and/or ischaemic brain injury(Rothman and Olney, 1986; Choi, 1988; Lipton and Rosenberg,1994). Choi (1987), Choi et al. (1987, 1988) demonstrated thatCa2þ was a key mediator of glutamate excitotoxicity, and thatthe NMDAR overactivation was the primary source of toxicCa2þ influx. The NMDAR-mediated excessive Ca2þ influxtargets mitochondria with activation of programmed deathsignals and mitochondria dysfunction. However, the role ofglutamate excitotoxicity, in particular changes in the down-stream of Ca2þ-dependent signaling pathways, in the CIH-induced neurocognitive deficits has not been well examinedin the OSA-related CIH animal model.

Our study generated several lines of consistent experi-mental evidence demonstrating that excitotoxicity mediatedby NMDAR overactivation is one of the important mechan-isms in the CIH-induced neurocognitive dysfunction. First,the exposure to 4-week CIH produced a significant impair-ment of neurobehavioral performance; and compared to the

control group, these neurocognitive changes were associatedwith elevated hippocampal [Ca2þ]i and significantly increasedexpression of caspase-3, caspase-9 and PARP indicative ofhippocampal cell apoptosis. Second, pretreatment withmemantine provided pharmacological rescuing by improvinglearning and memory of CIH mice in both the acquisition testand probe trial of the Morris water maze test; in addition,administration of memantine to CIH animals abolished thehippocampal calcium overload and normalized the expres-sion of caspase-3, caspase-9 and PARP. Taken together, theseobservations suggested that the hippocampal calcium over-load and increased expression of caspases resulting from theNMDAR overactivation may contribute to the CIH-inducedneurocognitive impairment. Our findings provide novelexperimental evidence that OSA and its associated CIH sharethe common pathway with the other disease states (e.g.,neurodegenerative disorders and brain hypoxia, etc.) in thatthe neurocognitive impairment reported among thesepatients could be attributed to NMDAR-mediated excitotoxi-city. The importance of this mechanism in relation to otherhypothesized pathways for CIH-induced neurocognitiveimpairments warrants further investigation.

Memantine, a drug recently approved for treatment ofAlzheimer’s disease, has been characterized as a uniqueNMDAR antagonist that confers protection against excito-toxic neurodegeneration without the serious side effects thatare known to be associated with other NMDAR antagonists. Itis a noncompetitive NMDAR antagonist with a fast off-ratebinding kinetics, which could effectively block NMDAR activ-ity that is triggered by elevated levels of glutamate (Chen andLipton, 2006). Several studies show that memantine pos-sesses neuroprotective efficacy in hypoxia/ischemia (Chenet al., 1998), global ischemia (Block and Schwarz, 1996), andsubarachnoid hemorrhage (Huang et al., 2015), which arecurrently thought to be mediated in part by excessive NMDARactivity. In our present study, daily intraperitoneal injectionof memantine inhibited over-activation of NMDAR under CIH,deceased influx of extracellular Ca2þ mainly through NMDAreceptors and the subsequent Ca2þ-dependent events includ-ing increasing of caspases expression and dysregulation ofERK signaling pathway, rescued neuron excitotoxicity inhippocampus in CIH mice. Moreover, we showed that dailyintraperitoneal injection of memantine at a sub-therapeuticdose of 5 mg/kg/day for 4 weeks in normal control mice didnot produce significant changes in the neurocognitive per-formance measured by the Morris water maze test (Fig. S1).This is further supported by the findings from the otherpublished studies showing no overt effects on the neurocog-nitive variables in MWM test in cognitively normal micefollowing chronic oral memantine treatment at 30 mg/kg/day for 3–4 weeks up to 3 months that produced thetherapeutic plasma level of memantine in mice (Martinez-Coria et al., 2010; Minkeviciene et al., 2004)

Previous studies have demonstrated that ERK activationoccurs as a Ca2þ-dependent NMDAR-mediated response,transferring extracellular stimuli to the nucleus and control-ling the synaptic plasticity and learning (Sweatt, 2004; Thomasand Huganir, 2004). The phosphorylation of CREB, as a down-stream target of ERKs, is also a critical component of learningand memory formation processes (Yin et al., 1994). In this

b r a i n r e s e a r c h 1 6 2 5 ( 2 0 1 5 ) 6 4 – 7 270

study, protein expression of phosphorylated CREB and ERK1/2in hippocampus was quantified and we demonstrated that 4-week CIH exposure produced significantly reduced expressionof phosphorylated ERK1/2 and CREB without measurablechanges in the total protein levels of both markers. Thesechanges were abolished by pretreatment withmemantine. Ourresults indicate dephosphorylation (i.e., inactivation) of ERK1/2and CREB occurred as a result of NMDAR overactivation duringexposure to CIH. These findings reveal the downstreamsignaling pathways as a part of NMDAR-mediated excitotoxi-city which may contribute to caspases expression, and conse-quently neurocognitive impairment observed in CIH animals.

There are several limitations of the present study. First,the separate experiments performed for the pre-experimentand intervention phase of our study resulted in a lack of“concurrent” control group that is desirable for neurobeha-vioral examinations and makes cross-comparison betweenthe two separate series of experiments difficult to interpret.Animals exposed to CIHþVEH for 4 weeks still showedstatistically significant neurobehavioral changes when com-pared to the combined RA and RAþVEH control animals fromboth phases of the study with longer escape latencies (Day 4:37.4973.20 s vs. 28.3872.09 s, po0.05) and distance travelled(Day 4: 922.39485.46 cm vs. 665.88753.38 cm, po0.05). Recog-nizing the limitation of such analysis as described above, thedata however does further support the CIH-induced neuro-cognitive impairment in our mouse model. Second, thefindings of increased hippocampal protein expression ofcaspases in CIH mice detected by Western blotting and itsreversion by memantine pre-treatment can be substantiatedby immunohistochemical staining of the hippocampal sec-tions to compliment the western data further. This aspectwarrants future experimental investigations.

In summary, the present study using a mouse CIH modelprovides the experimental evidence that NMDAR overactiva-tion with intracellular Ca2þ overload and dysregulation ofERK/CREB signaling pathway play a role in mediating the CIH-induced neurocognitive impairments, and that memantinetreatment is effective in preventing these adverse changes.These results offer new mechanistic insights for both basicresearchers and clinicians in our combat against the neuro-cognitive deficits in OSA patients.

4. Experimental procedures

4.1. Animals and grouping

Four-week old male ICR mice (20–22 g) were randomlyassigned to three groups, room air (RA) exposure (controlgroup, n¼15), CIH exposure with pretreatments of vehicle(CIHþVEH group, n¼15) or memantine (CIHþMEM group,n¼15), housed in a 12-h light/dark cycle (lights on from 7:00am to 7:00 pm) at a constant temperature (26 1C) and allowedaccess to food and water ad libitum. This study was carriedout in strict accordance with the recommendations in theGuide for the Care and Use of Laboratory Animals of China.The animal study protocol was reviewed and approved by theInstitutional Animal Care/User Ethical Committee of Soo-chow University (approval no. 20110121). All procedures were

performed in accordance with the Regulations for the Admin-istration of Affairs Concerning Experimental Animalsapproved by the State Council of People’s Republic of China.

4.2. Chronic intermittent hypoxia exposures

Mice were maintained in two identical commerciallydesigned chambers (HOPE Industry & Trade CO., LTD. Tianjin,China.) and exposed to intermittent hypoxia (IH) or room airfor 4 weeks. The IH chamber was circulated with N2 (30 s) andO2 (30 s), consists of cycles of oxygen levels under the controlof an O2 analyzer (Oxygen Sensor 1-01, Wismar, Germany)between 6–8% and 19–21% every 60 s, i.e. 60 cycles/h. Inter-mittent hypoxia was given during the daytime for 8 h from 9am to 5 pm. In addition, age-matched male mice wereexposed to room air and included as a control group; theremaining conditions were the same.

4.3. Drug administration

Memantine (Sigma-Aldrich) was dissolved in 0.9% saline at aconcentration of 1 mg/ml. For the intervention phase of theexperiments, memantine (5 mg/kg) or vehicle (0.9% saline)was administered through intraperitoneal injection inapproximately 15 min prior to start of daily CIH exposure inthe morning and continued for 4 weeks. The time and dailydosing regimen of memantine administration was based onthe findings from previous studies (Costa et al., 2008;Romberg et al., 2012).

A series of pre-experiment was also performed separatelyin age-matched male ICR mice (n¼42) to evaluate the effect of4-week intraperitoneal injection of memantine at 5 mg/kg/day or vehicle on neurocognitive performance in controlanimals as measured by Morris water maze test (data wereshow in supplemental material).

4.4. MWM test

One day after 4 weeks exposure, three groups of 15 mice pergroup were tested for learning and memory ability using theMorris water maze (MWM) test (Jiliang Software TechnologyCo., Ltd. Shanghai, China) by investigators blinded to thegroup conditions. In place navigation test, mice received fourtraining trials daily and each daily training session consistedof two training trials. During each trial, mice were gentlyplaced in the water facing the wall of the maze at one of thefour equally spaced start positions (north, south, east, orwest). The time it took to locate the submerged platform(defined by the escape latency cut-off time of 60 s) wererecorded. On day 5, probe trials test was conducted, wherebythe platform was removed. The number of platform passesand the percent time in the target quadrant in a 60 s periodwas determined.

4.5. The intracellular calcium ([Ca2þ]i) concentrationmeasured by flow cytometry

After MWM test, hippocampal tissue samples were success-fully processed from four mice per group for intracellular

b r a i n r e s e a r c h 1 6 2 5 ( 2 0 1 5 ) 6 4 – 7 2 71

calcium concentration measurement. Hippocampus tissueswere quickly separated and digested with 0.125% trypsin(Sigma, USA) terminated with Dulbecco’s modified eaglemedium (DMEM, Corning, China) containing 20% horse serum(Gibco, USA) at 37 1C for 30 min. Then samples were trituratedwith Pasteur pipettes and dispersed into mono-cell suspen-sion. The suspension was collected and centrifuged at1000 rpm for 5 min. The precipitated cells were resuspendedin 500 μl D-Hank’s solution, adjusted to the density with1�106 cells/ml, and then added fluo-3/AM (Terminal concen-tration 5 μmol/l, Beyotime, China) to incubate for 45 min at37 1C away from light, washed with D-Hank’s solution, andcentrifuged at 1000 rpm for 5 min. The pellet (containingcells) was resuspended in D-Hank’s solution and measuredby flow cytometry (Beckman Coulter FC500, USA), excitationat a wavelength of 488 nm, and emission at 525 nm. The[Ca2þ]i in hippocampus was determined by fluorescenceintensity of Fluo-3. Fluo-3 is one of the most suitable Ca2þ

indicators for flow cytometry, and has been shown to be ableto detect modest (e.g., o1-fold) increase in Ca2þ concentra-tion in hippocampus and other tissue or cell samples (Alzoubiet al., 2015; Kucherenko and Lang, 2014; Zhen et al., 2014).

4.6. Western blotting

After MWM test, hippocampal tissue samples were success-fully processed from three to four mice per group for Westernblotting. Total protein in the hippocampus was extracted, andconcentrations were measured. Equal amount of protein(50 μg) from each homogenate was resolved by 10% or 12%SDS-PAGE. Proteins were transferred to PVDF membranes(Hybond TM-ECL; Amersham Pharmacia Biotech, Inc.). Thesemembranes were blocked for 2 h at room temperature with5% skimmed milk in PBS and 0.1% Tween 20. The membraneswere incubated with polyclonal antibodies (anti-caspase-3,–9,PRAP, p-ERK1/2, ERK1/2, p-CREB, CREB, 1:1000 dilution, CellSignaling, USA; anti-GAPDH, 1:1000 dilution, Santa Cruz, USA)at 4 1C overnight, and then washed three times in TBST andincubated with anti-rabbit or -mice IgG conjugated to HRP(1:4000; beyotime, China) at room temperature for 2 h. Thereaction products were visualized by enhanced chemilumi-nescence (ECL; beyotime, China), and the membranes wereexposed to X-ray film. Densitometry analysis of proteinexpression was performed by a microcomputer imaginganalysis device (Gel Doc EQ, Bio-Rad, USA). The relative levelof caspase-3,–9 and PRAP expression were normalized toGAPDH, p-ERK1/2 and p-CREB were normalized to ERK1/2and CREB, respectively. Integrated density value was used asthe parameter representing the level of protein expression.

4.7. Statistical analysis

All data were presented as mean7SEM. Escape latencies wereanalyzed by two-way repeated measures ANOVA (with treat-ments and training days as factors). Other data were ana-lyzed by one-way ANOVA, followed by Least-significantdifference test to compare values of different groups.po0.05 were considered to be statistically significant.

Acknowledgments

The authors are indebted to the financial support of theNational Natural Science Foundation of China (Grant nos.81170070, 81270147) and subject to the second affiliatedhospital of Soochow university preponderant clinic disciplinegroup project funding (Project nos. XKQ2015002).

Appendix A. Supporting information

Supplementary data associated with this article can be foundin the online version at http://dx.doi.org/10.1016/j.brainres.2015.08.012.

r e f e r e n c e s

Aamodt, S.M., Constantine-Paton, M., 1999. The role of neuralactivity in synaptic development and its implications for adultbrain function. Adv. Neurol. 79, 133–144.

Alzoubi, K., Calabro, S., Faggio, C., Lang, F., 2015. Stimulation ofsuicidal erythrocyte death by sulforaphane. Basic Clin.Pharmacol. Toxicol. 116, 229–235.

Bliss, T.V., Collingridge, G.L., 1993. A synaptic model of memory:long-term potentiation in the hippocampus. Nature 361,31–39.

Block, F, Schwarz, M, 1996. Memantine reduces functional andmorphological consequences induced by global ischemia inrats. Neurosci. Lett. 208, 41–44.

Chen, HS, Wang, YF, Rayudu, PV, Edgecomb, P, Neill, JC, Segal,MM, Lipton, SA, Jensen, FE, 1998. Neuroprotectiveconcentrations of the N-methyl-D-aspartate open-channelblocker memantine are effective without cytoplasmicvacuolation following post-ischemic administration and donot block maze learning or long-term potentiation.Neuroscience 86, 1121–1132.

Chen, H.S., Lipton, S.A., 2006. The chemical biology of clinicallytolerated NMDA receptor antagonists. J. Neurochem. 97,1611–1626.

Chen, R., Xiong, K.P., Huang, J.Y., Lian, Y.X., Jin, F., Li, Z.H., Zhao,M.Y., Liu, C.F., 2011. Neurocognitive impairment in Chinesepatients with obstructive sleep apnoea hypopnoea syndrome.Respirology 16, 842–848.

Choi, D.W., 1987. Ionic dependence of glutamate neurotoxicity.J. Neurosci. 7, 369–379.

Choi, D.W., 1988. Glutamate neurotoxicity and diseases of thenervous system. Neuron 1, 623–634.

Choi, D.W., Koh, J.Y., Peters, S., 1988. Pharmacology of glutamateneurotoxicity in cortical cell culture: attenuation by NMDAantagonists. J. Neurosci. 8, 185–196.

Choi, D.W., Maulucci-Gedde, M., Kriegstein, A.R., 1987. Glutamateneurotoxicity in cortical cell culture. J. Neurosci. 7, 357–368.

Costa, A.C., Scott-McKean, J.J., Stasko, M.R., 2008. Acute injectionsof the NMDA receptor antagonist memantine rescueperformance deficits of the Ts65Dn mouse model of Downsyndrome on a fear conditioning test.Neuropsychopharmacology 33, 1624–1632.

Gottlieb, D.J., Yenokyan, G., Newman, A.B., O’Connor, G.T.,Punjabi, N.M., Quan, S.F., Redline, S., Resnick, H.E., Tong, E.K.,Diener-West, M., Shahar, E., 2010. Prospective study ofobstructive sleep apnea and incident coronary heart diseaseand heart failure: the sleep heart health study. Circulation 122,352–360.

b r a i n r e s e a r c h 1 6 2 5 ( 2 0 1 5 ) 6 4 – 7 272

Gozal, D., Daniel, J.M., Dohanich, G.P., 2001. Behavioral andanatomical correlates of chronic episodic hypoxia duringsleep in the rat. J. Neurosci. 21, 2442–2450.

Gozal, D., Row, B.W., Gozal, E., Kheirandish, L., Neville, J.J.,Brittian, K.R., Sachleben Jr, L.R., Guo, S.Z., 2003. Temporalaspects of spatial task performance during intermittenthypoxia in the rat: evidence for neurogenesis. Eur. J. Neurosci.18, 2335–2342.

Huo, X.L., Min, J.J., Pan, C.Y., Zhao, C.C., Pan, L.L., Gui, F.F., Jin, L.,Wang, X.T., 2014. Efficacy of lovastatin on learning andmemory deficits caused by chronic intermittent hypoxia-hypercapnia: through regulation of NR2B-containing NMDAreceptor-ERK pathway. PLoS One 9, e94278.

Huang, CY, Wang, LC, Wang, HK, Pan, CH, Cheng, YY, Shan, YS,Chio, CC, Tsai, KJ, 2015. Memantine alleviates brain injury andneurobehavioral deficits after experimental subarachnoidhemorrhage. Mol. Neurobiol. 51, 1038–1052.

Jackson, M.L., Howard, M.E., Barnes, M., 2011. Cognition anddaytime functioning in sleep-related breathing disorders.Prog. Brain Res. 190, 53–68.

Kalia, L.V., Kalia, S.K., Salter, M.W., 2008. NMDA receptors inclinical neurology: excitatory times ahead. Lancet Neurol. 7,742–755.

Kheirandish, L., Gozal, D., Pequignot, J.M., Pequignot, J., Row, B.W.,2005a. Intermittent hypoxia during development induceslong-term alterations in spatial working memory,monoamines, and dendritic branching in rat frontal cortex.Pediatr. Res. 58, 594–599.

Kheirandish, L., Row, B.W., Li, R.C., Brittian, K.R., Gozal, D., 2005b.Apolipoprotein E-deficient mice exhibit increasedvulnerability to intermittent hypoxia-induced spatial learningdeficits. Sleep 28, 1412–1417.

Kheirandish-Gozal, L., McManus, C.J., Kellermann, G.H., Samiei,A., Gozal, D., 2013. Urinary neurotransmitters are selectivelyaltered in children with obstructive sleep apnea and predictcognitive morbidity. Chest 143, 1576–1583.

Kucherenko, Y.V., Lang, F., 2014. Niflumic acid affects store-operated Ca(2þ)-permeable (SOC) and Ca(2þ)-dependent K(þ)and Cl(� ) ion channels and induces apoptosis in K562 cells.J. Membr. Biol. 247, 627–638.

Lipton, S.A., Rosenberg, P.A., 1994. Excitatory amino acids as afinal common pathway for neurologic disorders. N. Engl. J.Med. 330, 613–622.

Martinez-Coria, H., Green, K.N., Billings, L.M., Kitazawa, M.,Albrecht, M., Rammes, G., Parsons, C.G., Gupta, S., Benerjee, P.,LaFerla, F.M, 2010. Memantine improves cognition andreduces Alzheimer’s-like neuropathology in transgenic mice.Am. J. Pathol. 176, 870–880.

Messenger, S.A., Moreau, J.M., Ciriello, J., 2013. Effect of chronicintermittent hypoxia on leptin and leptin receptor proteinexpression in the carotid body. Brain Res. 1513, 51–60.

Minkeviciene, R., Banerjee, P., Tanila, H., 2004. Memantineimproves spatial learning in a transgenic mouse model ofAlzheimer’s disease. J. Pharmacol. Exp. Ther. 311, 677–682.

Nair, D., Dayyat, E.A., Zhang, S.X., Wang, Y., Gozal, D., 2011.Intermittent hypoxia-induced cognitive deficits are mediatedby NADPH oxidase activity in a murine model of sleep apnea.PLoS One 6, e19847.

Neubauer, J.A., 2001. Invited review: physiological andpathophysiological responses to intermittent hypoxia. J. Appl.Physiol. 1985 (90), 1593–1599.

Olney, J.W., Ho, O.L., Rhee, V., 1971. Cytotoxic effects of acidic andsulphur containing amino acids on the infant mouse centralnervous system. Exp. Brain Res. 14, 61–76.

Prabhakar, N.R., Fields, R.D., Baker, T., Fletcher, E.C., 2001.Intermittent hypoxia: cell to system. Am. J. Physiol. Lung Cell.Mol. Physiol. 281, L524–528.

Rada, R., 2005. Obstructive sleep apnea and head and neckneoplasms. Otolaryngol. Head Neck Surg. 132, 794–799.

Romberg, C., McTighe, S.M., Heath, C.J., Whitcomb, D.J., Cho, K.,Bussey, T.J., Saksida, L.M., 2012. False recognition in a mousemodel of Alzheimer’s disease: rescue with sensory restrictionand memantine. Brain 135, 2103–2114.

Rothman, S.M., Olney, J.W., 1986. Glutamate and thepathophysiology of hypoxic—ischemic brain damage. Ann.Neurol. 19, 105–111.

Row, B.W., 2007. Intermittent hypoxia and cognitive function:implications from chronic animal models. Adv. Exp. Med. Biol.618, 51–67.

Row, B.W., Kheirandish, L., Neville, J.J., Gozal, D., 2002. Impairedspatial learning and hyperactivity in developing rats exposedto intermittent hypoxia. Pediatr. Res. 52, 449–453.

Row, B.W., Liu, R., Xu, W., Kheirandish, L., Gozal, D., 2003.Intermittent hypoxia is associated with oxidative stress andspatial learning deficits in the rat. Am. J. Respir. Crit. CareMed. 167, 1548–1553.

Ryan, S., McNicholas, W.T., Taylor, C.T., 2007. A critical role for p38map kinase in NF-kappaB signaling during intermittenthypoxia/reoxygenation. Biochem. Biophys. Res. Commun. 355,728–733.

Sweatt, J.D., 2004. Mitogen-activated protein kinases in synapticplasticity and memory. Curr. Opin. Neurobiol. 14, 311–317.

Thomas, G.M., Huganir, R.L., 2004. MAPK cascade signalling andsynaptic plasticity. Nat. Rev. Neurosci. 5, 173–183.

Wroge, C.M., Hogins, J., Eisenman, L., Mennerick, S., 2012.Synaptic NMDA receptors mediate hypoxic excitotoxic death.J. Neurosci. 32, 6732–6742.

Xie, H., Leung, K.L., Chen, L., Chan, Y.S., Ng, P.C., Fok, T.F., Wing,Y.K., Ke, Y., Li, A.M., Yung, W.H., 2010. Brain-derivedneurotrophic factor rescues and prevents chronic intermittenthypoxia-induced impairment of hippocampal long-termsynaptic plasticity. Neurobiol. Dis. 40, 155–162.

Xu, W., Chi, L., Row, B.W., Xu, R., Ke, Y., Xu, B., Luo, C.,Kheirandish, L., Gozal, D., Liu, R., 2004. Increased oxidativestress is associated with chronic intermittent hypoxia-mediated brain cortical neuronal cell apoptosis in a mousemodel of sleep apnea. Neuroscience 126, 313–323.

Yin, J.C., Wallach, J.S., Del, V.M., Wilder, E.L., Zhou, H., Quinn,W.G., Tully, T., 1994. Induction of a dominant negative CREBtransgene specifically blocks long-term memory inDrosophila. Cell 79, 49–58.

Young, T., Palta, M., Dempsey, J., Skatrud, J., Weber, S., Badr, S.,1993. The occurrence of sleep-disordered breathing amongmiddle-aged adults. N. Engl. J. Med. 328, 1230–1235.

Zhen, J.L., Wang, W.P., Zhou, J.J., Qu, Z.Z., Fang, H.B., Zhao, R.R.,Lu, Y., Wang, H.C., Zang, H.M., 2014. Chronic intermittenthypoxic preconditioning suppresses pilocarpine-inducedseizures and associated hippocampal neurodegeneration.Brain Res. 1563, 122–130.