Neuronal deterioration associated with hyperexcitability under mild chronic cerebral hypoperfusion

Abstract

Chronic cerebral hypoperfusion (CCH) has been indicated to impair cognitive and diverse brain functions. However, the neural mechanisms linking these cerebrovascular and phenotypic alterations remain unclear. Here, we investigated the effect of CCH on neuronal activity in male mice with unilateral common carotid artery occlusion using optical imaging and MRI. Our examinations revealed enhanced neuronal activity in concurrence with increased glutamate and tissue acidosis up to seven days after occlusion. At 21–28 days after occlusion, neuronal activity decreased below baseline, while the acidotic but not the hyperglutamatergic state persisted. Notably, pharmacological blockade of the N-methyl-D-aspartate-type glutamate receptor, initiated at an early stage of CCH, suppressed the onset of neuronal hyperexcitation and subsequent deficits in neuronal activity. Altogether, we provide experimental evidence that CCH induces a glutamate surge and results in neuronal hyperexcitation at an early phase, which thereafter gives rise to a non-lethal but progressive deterioration of neuronal functions.

Introduction

Chronic cerebral hypoperfusion (CCH) without cerebral infarction increases the risk of stroke. CCH alone can lead to cognitive and other functional impairments of the brain in humans.^1–4 However, the mechanism linking CCH to functional deterioration of the brain remains unclear. To identify the key pathological events downstream of CCH, unilateral common carotid artery occlusion (UCCAO) has been used to generate a mouse model of mild unilateral chronic cerebral ischemia without infarct lesions. This mouse model has been reported to exhibit white matter lesions and functional deficits, including impaired object recognition and short-term memory, 28 days after surgical manipulation.^5–8 UCCAO chronically reduces ipsilateral cerebral blood flow (CBF) by 20–30%,^9,10 leading to misery perfusion and reduced cerebrovascular reserves. ¹¹ This condition is defined as mild hypoperfusion. Correspondingly, immunohistochemical analysis has demonstrated a marked increase in activated microglia and astrocytes at 28 days,^7,8 but this was not accompanied by noticeable neuronal loss in the cortex.^8,11,12 In addition, diffusion tensor imaging studies have reported changes in water diffusion in the cortical gray and white matter of UCCAO model mice. ¹⁰ Rat models of chronic hypoperfusion have also shown the activation of astrocytes and microglia in the cortex. ¹³

In our previous study using in vivo two-photon microscopy imaging, dilatation of cerebral microvessels (i.e., arterioles, capillaries, and venules) in the somatosensory cortex and diminished responsiveness of arterioles to hypercapnia were observed 1–28 days after UCCAO. ¹¹ We also reported that cerebral blood flow changes coupled with neuronal activation in response to sensory stimulation decreased 7–28 days after UCCAO in the somatosensory cortex. ⁹ In these studies, the impairment of the cerebrovascular response to hypercapnia was significantly alleviated 28 days after UCCAO, ¹¹ while there was no recovery of the cerebrovascular response to sensory stimulation during this period. ⁹ It is likely that the deficit in neuronal functions and consequent vasodilatation persists over 28 days under CCH conditions, unlike transient disturbances of cerebrovascular responsiveness uncoupled from neuronal activities. However, it remains unclear whether the attenuated vascular response to sensory stimulation originates from neuronal dysfunction or disrupted neurovascular coupling, because no direct measurement or evaluation of neuronal activity changes due to CCH has been performed.

In cerebral infarction induced by ischemia, neuronal hyperexcitation, observed as cortical spreading depression and glutamate (Glu) release from astrocytes, occurs immediately after occlusion surgery,^14–16 and these alterations at non-lethal levels may also occur in mild hypoperfusion. It has been reported that Glu overflow, even at concentrations that cause no neuronal death, results in synaptic loss ¹⁷ and that Glu-dependent synaptic insults may critically contribute to brain dysfunction in CCH.

Based on these findings, we hypothesized that mild hypoperfusion after UCCAO initially provokes neuronal hyperexcitability, leading to neuronal deterioration at a later stage. Therefore, we presumed that the suppression of early neuronal hyperexcitability prevented the delayed diminution of neuronal activity. In this study, we aimed to examine these notions. To this end, we directly and longitudinally measured neuronal activity in UCCAO model mice using mesoscopic and microscopic optical imaging techniques. Cortex-wide oxygen metabolism, associated with neuronal activation in response to sensory stimuli, was quantified using flavoprotein autofluorescence imaging (FAI). ¹⁸ Spontaneous and sensory stimulation-evoked activity in individual neurons was assayed by intravital two-photon laser fluorescence microscopy of calcium signals. ¹⁹ In addition, Glu concentration and pH, which are mechanistically associated with neuronal hyperexcitability, were measured using magnetic resonance spectroscopy (MRS) ²⁰ and pH-weighted imaging based on chemical exchange saturation transfer (CEST)-MRI, ²¹ respectively. Finally, the causal relationship between early hyperexcitability and late neuronal deterioration was investigated by pharmacologically suppressing overactive glutamatergic signaling with memantine.

Materials and methods

Viral vector production

Full-length cDNA encoding a fluorescent calcium indicator, GCaMP6s ¹⁹ (Addgene plasmid #40753), was amplified using PCR and subcloned into the customized multi-cloning site of an adeno-associated virus (AAV) transfer plasmid containing a pan-neuronal rat synapsin promoter, woodchuck post-transcriptional regulatory element, and poly A signal flanked by inverted terminal repeats. For large-scale preparation of recombinant AAV, a transfer plasmid and helper-free DJ packaging plasmids (Cell Biolabs, CA, USA) were introduced into HEK293T cells by polyethyleneimine transfection. At 48 h after transfection, the cells were harvested and lysed, and the viral particles were purified using a HiTrap heparin column (GE Healthcare Life Sciences, UK). The titer of the virus stock was determined using the AAVpro® Titration kit ver2 (TaKaRa, Shiga, Japan) and adjusted to 3.7 × 10¹¹ to 9.9 × 10¹¹ viral genomes per mL.

Animal preparation

Fifty male C57BL/6J mice weighing 22–28 g and 7–11 weeks of age (Japan SLC Inc., Hamamatsu, Japan) were used for three separate in vivo measurements: (1) simultaneous FAI and intrinsic optical signal imaging (IOSI) using a charge-coupled device (CCD), (2) calcium imaging using two-photon laser microscopy, and (3) MRS measurements (Fig. S1). The mice were housed with food and water provided ad libitum in their cages at 25°C under a 12-hour light/dark cycle. All experimental procedures involving mice were performed in accordance with the institutional guidelines on humane care and use of laboratory animals (Regulations on the Implementation of Animal Experiments) and were approved by the Institutional Animal Care and Use Committee of the National Institutes for Quantum Science. This manuscript was written in accordance with the Animal Research Reporting In vivo Experimental (ARRIVE) guidelines.

Two weeks before in FAI, a chronic cranial window was attached to the left hemisphere of the mouse brain. Inflammation due to the cranial window surgery was conceived to be resolved within this period. ²² For cranial window attachment, animals were anesthetized with a mixture of air, oxygen, and isoflurane (3–5% for induction and 2% for surgery; Mylan Japan, Tokyo, Japan) via a facemask. The cranial window, measuring 4–5 mm in diameter, was centered over the left somatosensory cortex at coordinates 1.8 mm caudal and 2.5 mm lateral to the bregma, following the ‘Seylaz-Tomita method'. ²³

To perform calcium imaging with two-photon microscopy, a chronic cranial window was attached to the left hemisphere of the mouse brain, and AAV vectors carrying the GCaMP6s gene (AAV-rSyn-GCaMP6s, 3.5 × 10^11 vg/mL) were injected into the same brain cortex using glass needles. Measurements were conducted at least four weeks post-injection to ensure stable gene expression and resolution of inflammation.

The methods used for UCCAO surgical procedures have been described in our previous studies.^9–11 Briefly, a mixture of air, oxygen, and isoflurane (3–5% for induction and 1.5–2% for surgery) anesthesia was provided via a facemask. A midline cervical incision was made, and the right common carotid artery was isolated from the adjacent vagus nerve and double-ligated by 6–0 silk sutures.

Memantine administration

MEM (lot M9292, Sigma-Aldrich, St. Louis, MO, USA) was administered orally via drinking water at a dose of 30 mg/kg/day for two or six weeks. This method of drug administration has been validated in previous studies, producing a steady-state plasma concentration of 1 µM, which is comparable to the therapeutic level in humans. ²⁴

Employment of awake animals for optical imaging

Two-photon imaging and FAI were performed at the somatosensory cortex ipsilateral to the UCCAO via a chronic cranial window in awake mice, as described elsewhere.^9,11 During the optical imaging sessions, the animal's head was fixed by screwing bars at the front and rear ends of the skull into a custom-made stereotactic apparatus. The animal was then placed on a styrofoam ball floated using a stream of air. This allowed the animal to exercise freely and stress-free on the ball and reduced motion artifacts, as the force produced by the animal was absorbed by the floating ball.^25,26

Sensory stimulation was applied to awake mice to evoke neuronal activation, according to a previously documented protocol. ²⁵ Briefly, an air puff was delivered to all the right whiskers at a pressure of approximately 15 psi using an air compressor. Rectangular pulse stimuli (50-ms pulse width and 100-ms onset-to-onset interval producing 10 Hz frequency) were generated using Master-8 (A.M.P.I., Jerusalem, Israel).

FAI and image corrections

FAI data were acquired as previous study.^26,27 Briefly, FAI and IOSI were simultaneously performed using two CCD cameras (MiCAM02, Brainvision, Tokyo, Japan) with temporal and spatial resolutions of 10 Hz for 25 s (250 frames/trial) and 192 × 128 pixels (each pixel size was 15 × 15 µm). Twenty-five trials were conducted with an inter-trial interval of 30 s. The images were averaged over trials to improve the signal-to-noise ratio.

Image correction methods for removing the interference of CBF during FAI were implemented, as stated previously. ²⁷ An uncorrected FAI signal contains not only flavoprotein autofluorescence but also absorption and scattering effects of CBF. The IOSI signal is closely correlated with cerebral blood volume if hematocrit remains constant and is proportional to CBF. Therefore, the absorption and scattering effects of CBF were estimated with the IOSI signal and were used for correcting the FAI signal.

To quantify FAI and IOSI, a circle with a diameter of 1 mm around the peak of the sensory-evoked signal change was defined as the region of interest (ROI). The time–response curves of the FAI were recorded by estimating the mean signal intensity in the ROI. The baseline value was calculated by averaging signals over a 10-s resting-state period, and the mean stimulus-induced signal increase was determined as a percentage change from the baseline.

Two-photon laser microscopy

Mice expressing GCaMP6s were used for two-photon calcium imaging. To visualize the cerebral blood vessels, 10 mM sulforhodamine 101 (SR101; MP Biomedicals, Irvine, CA) dissolved in saline was injected intraperitoneally (8 µL/g body weight) just before the initiation of the imaging experiments. An awake animal was placed on a custom-made apparatus and real-time imaging was conducted using a two-photon laser microscope (TCS-SP5 MP; Leica Microsystems GmbH, Wetzlar, Germany) at an excitation wavelength of 900 nm. The emission signals were separated using a beam splitter (560/10 nm) and simultaneously detected using bandpass filters for SR101 (610/75 nm) and GCaMP6s (525/50 nm). A single image plane consisted of 512 × 512 pixels and the in-plane pixel size was 0.9 µm. Images were acquired at a depth of approximately 150 µm (layer 2/3) from the brain surface in the same region throughout the longitudinal measurements using the cerebral vascular image as a spatial landmark. The activity of the vast majority of neurons in the field of view can be monitored over time, apart from when there is a change in the presence or absence of firing. The number of neurons measured in each mouse before UCCAO was 50–150.

For sensory stimulation experiments, data were acquired at a temporal resolution of 5 Hz for 30 s (150 frames/trial). Three trials were successively performed with an inter-trial interval of 60 s. Ten seconds after the initiation of the scan, neuronal activation was induced by sensory stimulation with an air puff for 2 s. For measurements of spontaneous neuronal activity, data acquisition was conducted with a temporal resolution of 5 Hz for 960 s (4,800 frames).

Images were analyzed using MATLAB software (MathWorks, MA, USA). For the sensory stimulation experiments, motion correction was first performed using a non-rigid algorithm. ²⁸ A maximum intensity projection (MIP) process for the time axis was applied to the images, and a circular ROI with a diameter of 18 µm was drawn on the neuron using the MIP image as a template. The rate of intensity change in GCaMP6s signals in response to sensory stimulation was evaluated for each neuron. The signals from all frames were divided by the average baseline value before stimulation and were expressed as the signal change ratio (ΔF/F0). The average value during the stimulation period was determined, and neurons whose value exceeded 1.2 was defined as those that responded to the stimulation. The signal changes of these responding neurons were averaged and peak values were obtained.

For the analysis of spontaneous neuronal activity, motion correction and ROI definition were performed as described above. A neuronal fluorescence time series F(t) was constructed by averaging the pixel values in the ROI for each frame. To correct baseline fluctuations, a baseline (F0) was calculated as the minimum of F(t) in the preceding 5 s, and a time series of a relative fluorescence signal change (ΔF/F0) was generated following the procedure described by Jia et al. ²⁹ When ΔF/F0 exceeded 50%, this change was considered to be caused by neuronal firing. The number of neurons that fired at least once in the measured area (460.8 × 460.8 µm) per minute was counted and averaged for 16 minutes. The frequency of neuronal firing per minute was determined for the activated neurons.

MRI and MRS measurements

In vivo magnetic resonance (MR) measurements were performed on a 7.0 T MRI scanner with a 20-cm bore (Biospec, Avance-III system; Bruker Biospin, Ettlingen, Germany) in conjunction with a volume coil for transmission and a two-channel phased-array cooled surface coil for reception (cryoprobe for mouse brain; Bruker Biospin). Mice were anesthetized with 3.0% isoflurane and then with 1.5% to 2.0% isoflurane for induction, and with a 1:5 oxygen/room air mixture for maintenance during the MR experiments. Rectal temperature was continuously monitored using an optical fiber thermometer (FOT-M; FISO, Quebec, QC, Canada) and the body temperature was maintained at 37.0 ± 0.5°C using a heating pad (Temperature control unit; Rapid Biomedical, Rimpar, Germany). Throughout the experiments, warm air was provided by a custom-made automatic heating system regulated by an electric temperature controller (E5CN; Omron, Kyoto, Japan). During MR scanning, the mice were placed in a prone position on an MRI-compatible cradle and held in place using custom-made ear bars.

For single-voxel ¹H-MRS of regional brain metabolites, a volume of interest (VOI) with a dimension of 1.1 × 2.0 × 3.5 mm was defined predominantly in the neocortex. Outer volume suppression combined with a point-resolved spectroscopy sequence was used for signal acquisition with the following parameters: number of repetitions = 296; repetition time = 3000 ms; echo time = 9.572 ms; spectral bandwidth = 4 kHz; and number of data points = 2048. The LCModel was used to quantify the ¹H-MR spectra. Unsuppressed water signals from identical VOIs were used as internal references for absolute concentrations. ³⁰ The concentrations of total N-acetylaspartate (tNAA; N-acetylaspartylglutamate [NAAG]), total creatine (tCr; creatine + phosphocreatine), Glu, glutamine (Gln), γ-aminobutyric acid (GABA), lactate (Lac), and taurine (Tau) were considered to show acceptable reliability when the estimated standard deviations (Cramér-Rao lower bounds, CRLBs) were less than 20.^31,32

A pH-sensitive CEST-MRI contrast was generated by exploiting the exchange between the hydrogen atoms of free water and amide hydrogen atoms of endogenous mobile cellular proteins and peptides. The intensity of MRI signals depended on the rate of this exchange, which was altered as a function of local pH. For pH-weighted imaging, endogenous CEST-MRI measurements were performed using a CEST-Fast Imaging with Steady-State Precession ³³ (FISP) MRI acquisition protocol with the following parameters: TR = 6.32 ms; TE = 3.16 msec; averages = 1; excitation pulse angle = 20°; slice thickness = 1.0 mm; FOV = 1.92 × 1.92 cm²; in-plane spatial resolution = 0.15 × 0.15 mm²; matrix = 128 × 128; and acquisition time = 8 min. A saturation period was applied before each FISP acquisition, consisting of one continuous-wave radiofrequency pulse of 5.0 s at a saturation power of 1.0 µT, with no additional spoiling and fat saturation pulses. It has been reported that a low saturation power (<1.5 µT) and a long saturation time (>2.5 s) are required to achieve the maximum CEST effect in the case of a relatively slow exchanging protons such as amide protons. ³⁴ Selective saturation was applied at 81 frequencies ranging from −10 to 10 ppm in increments of 0.25 ppm. A control image with a saturation offset of 300 ppm was acquired. Data processing was performed using customized scripts in MATLAB. First, B0 inhomogeneity was corrected using WASSR. ³⁵ Z-spectra were calculated from the mean of the ROI placed over the ipsilateral and contralateral sides of the somatosensory cortex. The CEST signal was quantified using magnetization transfer ratio with asymmetric analysis (MTR_asym) at particular offsets of interest (i.e., Δω = +3.5 ppm) using the following definition: MTR_asym = (S^–Δω–S^+Δω)/S(+300 ppm), where and S^{[−Δω, +Δω]} is the water signal intensity in the presence of saturation pulses at offset ±Δω. Thus, the resulting pH-weighted image represents the distribution of percent signal difference between the images with presaturation pulse at +3.5 ppm and −3.5 ppm relative to reference image that was obtained with the saturation pulse at an extreme far offset (300 ppm) and expected to exclude any saturation effects. ³⁶ The MTR_asym value for each animal was calculated from the mean of the same ROI as that of the Z-spectra.

Statistical analysis

All data obtained from two-photon calcium imaging and FAI were normalized to the values obtained before UCCAO for each animal. Statistical analyses were performed using the Statistics and Machine Learning Toolbox in MATLAB. The normality of each dataset was confirmed using The Kolmogorov-Smirnov test. All values are presented as mean ± standard deviation in the Results section and as bar graphs. Statistical analyses of the FAI data were performed using ANOVA with a post-hoc Tukey-Kramer test. Statistical analyses of two-photon calcium imaging and MRS data were performed using one-way repeated-measures ANOVA (rm-ANOVA) with the Bonferroni post-hoc test. Statistical analyses of CEST-MRI experiments were performed using Student’s two-tailed paired t-test. The null hypothesis was rejected when the P-value was <0.05.

Results

Cerebral oxygen metabolism evoked by sensory stimulation

Oxygen metabolism associated with neuronal activity in UCCAO mice was measured using FAI. FAI, using a CCD camera system, has a wide field of view that covers the entire somatosensory cortex. Increases in FAI signals evoked by sensory stimulation of the right whiskers were observed in the contralateral somatosensory cortex (Figure 1(a)), but the location of intensified flavin signals was not markedly altered during the observation period. The size of the active region in the FAI increased two days after UCCAO and gradually decreased from 14 to 28 days (Figure 1(a)). The average time–response curves of FAI and peak fluorescence values during sensory stimulation are shown in Figure 1(b) and (c), respectively. Peak values significantly increased two days after UCCAO (P < 0.05) and then significantly decreased at 21 (P < 0.05) and 28 (P < 0.05) days after UCCAO, compared to the pre-occlusion condition (Pre) (Figure 1(c)).

Figure 1.

Changes in flavoprotein autofluorescence in response to sensory stimulation in the UCCAO mice. Oxygen metabolism associated with neuronal activity is measured using FAI with correction for the effects of CBF. (a) Baseline images (top) and maps of signal changes during the stimulation (bottom) in a representative animal. There are no noticeable changes in the morphology of cortical cerebral blood vessels across all experimental periods. (b) Time-response curves of FAI signals in the sensory stimulation session. The value indicates a ratio of the signals to the mean baseline measure at −10 to 0 sec (ΔF/F0) and (c) chronological changes in the peak ΔF/F0 value. The values are normalized to the measure at the pre-UCCAO stage (Pre) in the same animal. The plots and black bars correspond to the value of each animal and the mean over animals. There is a significant main effect of time in one-way ANOVA (F_5,58 = 33.35, P < 0.001). *, P < 0.05 for the difference from Pre by post-hoc Tukey-Kramer test. n = 12 for Pre, D7, D14, and D28, and n = 8 for D2 and D21. FAI, flavoprotein autofluorescence imaging; UCCAO, unilateral common carotid artery occlusion.

Evoked and spontaneous neuronal activity

Using two-photon calcium imaging, the evoked and spontaneous activities of individual neurons in the somatosensory cortex was measured as changes in GCAMP6s fluorescence (Figure 2(a)). The average fluorescence time–response curves and peak values during sensory stimulation are shown in Figure 2(b) and (c), respectively. The average peak fluorescence values in response to sensory stimuli significantly increased seven days after UCCAO (P < 0.01). The sensory-evoked response was significantly decreased compared to the Pre value 28 days after UCCAO (P < 0.01). For the evaluation of spontaneous neuronal activity, a 50% change from baseline was defined as neuronal firing (Figure 2(d)), and the number of firing neurons and frequency of firing per neuron were obtained from each ROI. The number of spontaneously firing cells per unit time significantly increased seven days after UCCAO relative to Pre (P < 0.05; Figure 2(e)) and subsequently decreased at 21 (P < 0.01) and 28 (P < 0.05) days after UCCAO compared with Pre. The frequency of firing per neuron was not significantly altered until 14 days (P > 0.05), but was significantly decreased at 21 (P < 0.05) and 28 (P < 0.05) days after UCCAO compared with Pre (Figure 2(f)).

Figure 2.

Changes in the intensity of GCaMP6s fluorescence induced by sensory-evoked and spontaneous neuronal firing in the UCCAO mice. (a) Two-photon microscopic images showing at baseline (top) and during the sensory stimulation (bottom). Signals from GCaMP6s expressed in neurons and SR101 labeling blood vessels are displayed in green and red, respectively. (b) Averaged time-response curves of the GCaMP6s fluorescence intensity in the sensory stimulation session. The value indicates a ratio of signals to the mean baseline measure at −10 to 0 sec (ΔF/F0). (c) Chronological changes in the averaged peak ΔF/F0 value normalized to the mean measure at the pre-UCCAO stage (Pre) (n = 7 for each time point). There is a significant main effect of time in rm-ANOVA (F_4,28 = 33.65, P < 0.001). (d) Changes in the GCaMP6s fluorescence intensity showing spontaneous firing of a representative neuron. A change up to 50% or larger from the baseline is judged as firing (asterisks). (e) Chronological changes in the averaged number of spontaneously activated neurons normalized to the measure at Pre (n = 8 for each time point). There is a significant main effect of time in rm-ANOVA (F_4,32 = 28.38, P < 0.001) and (f) chronological changes in the averaged frequency of firing per neuron normalized to the measure at Pre (n = 8 for each time point). There is a significant main effect of time in rm-ANOVA (F_4,32 = 16.07, P < 0.001). The plots and black bars correspond to the value of each animal and the mean over animals. *, P < 0.05; **, P < 0.01 for the difference from Pre by post-hoc Bonferroni test. UCCAO, unilateral common carotid artery occlusion.

Cerebral metabolites

Brain metabolite concentrations were assessed using MRS measurements. Figure 3 shows the concentrations of cerebral metabolites in the somatosensory cortex estimated by LCModel before and two and 28 days after UCCAO. The reliability of the metabolite concentrations was confirmed based on the estimated standard deviations. The concentrations of tNAA and Glu were significantly increased two days after UCCAO compared to Pre (P < 0.05), but did not differ from Pre at 28 days (P > 0.05). No significant differences in tCr, Gln, GABA, Lac, and Tau concentrations were detected at two or 28 days after UCCAO. The increase in Glu without alterations in GABA two days after UCCAO suggests the central role of Glu overflow in neuronal hyperexcitability, as demonstrated by FAI and calcium imaging.

Figure 3.

Temporal changes in the concentrations of cerebral metabolites following UCCAO. The concentrations of tNAA, total creatine (tCr), gluatamate (Glu), glutamine (Gln), GABA, lactate (Lac), and taurine (Tau) in the cerebral cortex are measured using MRS and quantified using the LC model. A T2-weighted image used as a template for defining the target area (blue box) is also displayed. Data were acquired from four mice, and values in individual animals at Pre (blue), Day 2 (orange), and Day 28 (green) are denoted by circles. Measures from the same individual are connected by gray lines. Horizontal bars represent the means of data from the four mice at Pre (blue), Day 2 (orange), and Day 28 (green). There is a significant main effect of time in rm-ANOVA for NAA (F_2,9 = 12.52, P < 0.01) and Glu (F_2,9 = 5.34, P < 0.05). *, P < 0.05 by post-hoc Bonferroni test. MRS, magnetic resonance spectroscopy.

pH-weighted imaging

We examined pH in the brains of mice before (Figures 4(a) and (b)), 2 (Figures 4(c) and (d)), and 28 days (Figures 4(e) and (f)) after UCCAO using CEST MRI.^21,33 MTR_asym was positively correlated with pH ²¹ (see Methods). The MTR_asym in extensive brain areas on the UCCAO side was lower than that on the contralateral side at two and 28 days after surgery (Figures 4(d) and (f)). Quantitative assays showed that MTR_asym in the ipsilateral somatosensory cortex was significantly lower than in the contralateral somatosensory cortex at both two (P < 0.01) and 28 (P < 0.01) days after UCCAO (Figure 4(g)). These data suggest that a persistent acidotic state two days after UCCAO underlies the pathological alterations in neuronal activity.

Figure 4.

In vivo CEST-MRI detection of relative pH at 2 and 28 days after UCCAO. Z-spectra averaged over animals (n = 7, respectively) (a, c, and e), and representative CEST (MTRasym maps at 3.5 ppm) (b, d, and f) images at pre (a and b), 2 (c and d), and 28 days (e and f) after UCCAO and (g) average MTR_asym (3.5 ppm) values in the ipsi and contra ROIs at Pre, 2 and 28 days after UCCAO. The plots and bars correspond to the value of each animal and the mean over animals. *, P < 0.01 for the difference between Day 2 and Day 28 data by paired t-test. CEST, chemical exchange saturation transfer; MTR_asym, magnetization transfer ratio with asymmetric analysis; ROI, region of interest; UCCAO, unilateral common carotid artery occlusion.

Effects of memantine on neuronal activity

To investigate the mechanistic link between mild neuronal hyperexcitation at an early stage of CCH and the subsequent attenuation of neuronal activity (Figures 1 and 2), we inhibited the initial hyperactivity of neurons by systemic administration of memantine, a partial antagonist of the N-methyl-D-aspartate receptor (NMDAR), and investigated longitudinal changes in neuronal activity using intravital two-photon imaging of calcium signals. Baseline neuronal activity was measured twice before UCCAO, with Pre1 and Pre2 corresponding to the periods before and seven days after the initiation of memantine administration. The treatment was terminated after the activity measurement seven days after UCCAO (Figure 5(a)). In contrast to untreated mice, the neuronal response to sensory stimulation and the number of spontaneously firing neurons in memantine-administered mice showed no significant changes seven days after UCCAO compared with Pre2 (P > 0.05; Figure 5(b) and (c)). Termination of memantine treatment provoked hyperexcitation of neurons because neuronal responses to sensory stimulation were significantly increased relative to Pre2 (P < 0.01) 14 days after UCCAO (i.e., seven days after the end of memantine administration) (Figure 5(b)). The number of spontaneously firing neurons was also significantly increased 21 days after UCCAO (i.e., 14 days after the end of memantine administration) compared to Pre2 (P < 0.05; Figure 5(c)). These hyperactive phases were followed by a notable decrease in the response to sensory stimulation (P < 0.05) and spontaneous firing frequency (P < 0.05) in neurons 35 days after UCCAO (i.e., 28 days after the end of memantine administration) compared to Pre2 (Figure 5(b) and (d)). Unlike in the CCH model mice, short-term administration of memantine had no effect on neuronal activity in mice without UCCAO (Fig. S2). We also investigated the effects of long-term memantine administration (Figure 5(e) to (h)) and found that sensory-evoked and spontaneous neuronal activity remained unaltered from Pre2 (P > 0.05) after drug treatment until 35 days after UCCAO. These observations indicate that the pathway triggering neuronal hyperexcitability is mediated by Glu, which is long-lasting in CCH, and that a hyperactive state is required for the delayed deterioration of neuronal activity.

Figure 5.

Effects of memantine administered at an early phase or throughout a long period following UCCAO. (a–d) Memantine is administered to mice from 7 days before UCCAO until 7 days after UCCAO, and neuronal activity is measured using two-photon calcium imaging before (Pre1) and after (Pre2) the initiation of the treatment prior to UCCAO and weekly at 7–35 days after UCCAO (n = 7 for each time point; a). (b) Chronological changes in the averaged peak ΔF/F0 value in the sensory-evoked neuronal activation normalized to the mean measure at Pre2. There is a significant main effect of time in rm-ANOVA (F_6,42 = 24.62, P < 0.001). (c) Chronological changes in the averaged number of spontaneously activated neurons normalized to the measure at Pre2. There is a significant main effect of time in rm-ANOVA (F_6,42 = 28.38, P < 0.001). (d) Chronological changes in the averaged frequency of firing per neuron normalized to the measure at Pre2. There is a significant main effect of time in rm-ANOVA (F_6,42 = 7.56, P < 0.001). (e–h) Memantine is administered to mice from 7 days before UCCAO until 35 days after UCCAO, and neuronal activity is measured as in the protocol shown in a (n = 4 for each time point; e). (f) Chronological changes in the averaged peak ΔF/F0 value in the sensory-evoked neuronal activation normalized to Pre2. There is no significant main effect of time in rm-ANOVA (F_6,21 = 0.40, P = 0.87). (g) Chronological changes in the averaged number of spontaneously activated neurons normalized to Pre2. There is no significant main Continued.effect of time in rm-ANOVA (F_6,21 = 2.18, P = 0.09). (h) Chronological changes in the averaged frequency of firing per neuron normalized to Pre2. There is no significant main effect of time in rm-ANOVA (F_6,21 = 0.69, P = 0.66). Outlined x-axis labels indicate assays under a memantine-treated condition. Measures obtained from the memantine-treated mice are displayed as black plots and black bars, and data from untreated mice (identical to those in Figure 2(c), (e) and (f)) are also presented as white plots and gray bars for comparison. The error bars indicate SD. **, P < 0.01; and *, P < 0.01 for the difference from Pre2 by post-hoc Bonferroni test. UCCAO, unilateral common carotid artery occlusion.

Discussion

The current two-photon microscopy imaging and FAI assays in the UCCAO mouse model presented in this study demonstrated that CCH provoked non-epileptogenic, non-lethal hyperexcitability in neurons, which was sufficient to induce the gradual progression of neuronal deterioration at the functional level. The consistency of the findings in measurements with the two imaging modalities indicates that the calcium signal alterations reflected hypoperfusion-related alterations rather than the effects of AAV injections and calcium buffering by GCaMP. Moreover, MRS and CEST-MRI experiments revealed glutamatergic overflow and an acidic state as putative mediators of this augmented excitatory activity in the early stage of CCH. Finally, treatment of UCCAO mice with memantine supports the view that the diminution of glutamatergic hyperactivity protects neurons against delayed functional deficits (Figure 6).

Figure 6.

Schematic diagram illustrating the mechanism of neuronal impairment associated with mild chronic cerebral hypoperfusion. Changes in brain state observed in UCCAO mice and corresponding measurement techniques are presented. CMRO₂: cerebral metabolic rate of oxygen; UCCAO: unilateral common carotid artery occlusion; Glu: glutamate.

Reduced perfusion without stroke can be elicited by atherosclerosis, increasing the risk of dementia in Alzheimer’s disease and other types of cognitive deficits in Parkinson’s disease, according to epidemiological, transcranial Doppler, and neuroimaging reports.^37–39 However, clinical observations without mechanistic interventions may not provide clues as to whether hypoperfusion is the cause or result of neurodegenerative changes in these illnesses. In addition, neurochemical and molecular processes linking deficiencies in cerebral circulation and neuronal dysfunction in non-infarction ischemia are yet to be clarified. Here, we implemented in vivo imaging technologies in conjunction with memantine administration to investigate the UCCAO-triggered deleterious pathway involving a causative association of elevated Glu tone and non-bursting neuronal hyperexcitement with the perturbed functions of these cells in a slowly progressive fashion. To our knowledge, this is the first study to directly demonstrate impaired neuronal activity in an animal model of CCH. The persistence of lowered pH until 28 days after UCCAO, as assessed using CEST-MRI, along with the delayed onset of neuronal hyperexcitation emerging after the termination of memantine treatment, suggests that acidosis is the cause of the enhanced propensity of neurons to fire. However, this hypothesis needs to be tested by performing CEST-MRI in a memantine-treated UCCAO mouse model.

It is well known that transient epileptiform neuronal activity, which frequently manifests as symptomatic epilepsy, arises in cerebral infarction,^40–42 which is a more severe form of ischemic condition than CCH. In contrast, the augmented inclination of neurons for firing demonstrated in the UCCAO mice did not present an epileptogenic profile. A reduced oxygen supply, a condition characterized by cerebral ischemia, induces anaerobic glycolysis, leading to lactate overproduction.^43,44 Tissue pH after the onset of cerebral ischemia decreases because it exhibits an inverse linear relationship with tissue lactate concentration.^45,46 Because of acidosis, astrocytes release Glu into the brain parenchyma, leading to neuronal hyperexcitation and eventual death during cerebral infarction.^14,47 Our CEST-MRI and MRS data indicated a similar mechanism involving a decline in pH and excess Glu two days after UCCAO (Figures 3 and 4). Hence, the Glu overflow from astrocytes is likely to elicit neuronal hyperexcitability. On the other hand, the present MRS data did not show changes in lactate after UCCAO. The factors linking the decrease in CBF and the decrease in pH require further study. The mechanistic links between acidosis and excess Glu have not yet been examined; therefore, it is yet to be determined whether this Glu surge originates from neurons or astrocytes. This hyperactive state of neurons may drive a vicious cycle consisting of increased oxygen and energy consumption, and the resultant exacerbation of oxygen deficiency, inducing a further decrease in pH.

Loss of postsynaptic density protein 95 (PSD-95), a major postsynaptic constituent, has been reported to be caused by high concentrations of Glu at a non-cell-lethal level. ¹⁷ Likewise, CCH sustained for one month due to bilateral carotid artery stenosis in mice was shown to result in a reduction of PSD-95, disrupting synaptic architectures. ⁴⁸ These alterations diminish the excitatory postsynaptic current^49,50 and disorganize the connectivity and plasticity of the neural circuits responsible for diverse brain functions, as exemplified by learning and memory. ⁵¹ Therefore, the CCH-related low-grade Glu surge and consequent enhancement of neuronal firing may diminish synaptic integrity in UCCAO animals.^17,48 It would also be of significance to prove reduced synaptic components in UCCAO mice using longitudinal positron emission tomographic assays, as conducted for other models ⁵² in addition to neuropathological examinations of excised brain tissues.

The present findings also support the therapeutic potential of pharmacological interventions for minute but deleterious synaptic hyperexcitability in CCH, as illustrated by the administration of memantine, a noncompetitive NMDAR antagonist. Our data indicate the need to suppress excessive glutamatergic tones at an early stage of CCH, and the use of in vivo examinations, such as MRS, to detect early CCH-dependent glutamate changes could provide critical information on whether treatment targeting neuronal hyperexcitability should be initiated. In addition to CCH unrelated to massive infarctions, pathological changes in brain tissues equivalent to CCH are likely to exist in association with incomplete reperfusion of microvessels after recanalization of large vessels as a treatment for ischemic stroke, as reported in clinical trials and animal studies.^53–55 Moreover, the penumbra, an area of hypoperfusion surrounding an irreversibly injured core of infarction, ⁵⁶ may be involved in a state resembling CCH. Neurons under these conditions can be functionally rescued by the inhibition of NMDA receptors early after the onset of vascular events. Memantine has been documented to be efficacious in patients with mild vascular dementia,^57–60 and its therapeutic effects are reinforced in individuals with elevated glutamate levels in the proximity of relatively new vascular lesions along the course of disease progression.

Although studies with simultaneous calcium imaging and electrophysiological recordings have reported that the calcium signal correlates with the number of action potentials, ⁶¹ adding electrophysiological measurements to our experimental procedure may help assess the faster frequency components of neuronal activity. In addition, an increased concentration of lactate in the brain has been reported to be closely associated with Glu release, driving neuronal hyperexcitation, ¹⁴ whereas changes in lactate levels were not detected in the present MRS measurements. One potential explanation for the inability to detect changes could be the application of isoflurane anesthesia during MRS. Indeed, isoflurane is known to alter lactate concentrations in the brain, ⁶² precluding the robust quantification of lactate in CCH model animals. While studies involving severe cerebral ischemia have detected lactate changes, even under isoflurane anesthesia,^63–64 we postulate that mild hypoperfusion may lead to less pronounced fluctuations in lactate levels than those observed in severe cerebral ischemia. Performing MRS without anesthesia will putatively offer the detection of relatively small lactate changes. ⁶⁵

A technical limitation of our study was the absence of two-photon microscopy and MRS data on the side contralateral to the UUCAO. We performed these assays only on the ipsilateral side to minimize physical stress on the animals by reducing the experimental time. However, our previous studies demonstrated that CBF^9,10 vasodilation induced by CO₂ inhalation ¹¹ and whisker stimulation ⁹ did not change throughout the month after occlusion. In addition, CEST-MRI analysis indicated no pH changes in the contralateral hemisphere (Figure 4). Considering these findings, it is unlikely that profound changes in neuronal activity and brain metabolite levels occurred on the contralateral side.

In addition, since only male mice were used in this study, the present results are limited to male mice. Sexual dimorphism has been recognized in neurovascular connections ⁶⁶ and mitochondrial function^67,68 in the healthy brain, as well as in cerebrovascular disease.^69–74 The time course and degree of neuronal activity changes in female CCH model mice may be significantly different from the results in male mice and will need to be examined in further analyses.

In conclusion, the current study using a UCCAO mouse model revealed a circular link between hypoperfusion, hypoxia, acidosis, Glu overload, hyperexcitation, and oxygen overconsumption early in CCH, perturbing synaptic transmission at a later stage. Therefore, pharmacological interference in this pernicious cycle, as exemplified by treatment with memantine, could offer a means of precluding the onset and aggravation of neuropsychiatric disorders with cerebrovascular abnormalities, irrespective of the presence or absence of infarcts.

Supplemental Material

sj-pdf-1-jcb-10.1177_0271678X251328971 - Supplemental material for Neuronal deterioration associated with hyperexcitability under mild chronic cerebral hypoperfusion

Supplemental material, sj-pdf-1-jcb-10.1177_0271678X251328971 for Neuronal deterioration associated with hyperexcitability under mild chronic cerebral hypoperfusion by Takuya Urushihata, Manami Takahashi, Masafumi Shimojo, Yuhei Takado, Nobuhiro Nitta, Yosuke Tajima, Kazuto Masamoto, Iwao Kanno, Yutaka Tomita, Naruhiko Sahara, Masaya Takahashi, Takayuki Obata, Hiroshi Ito, Tetsuro Yamashita, Tetsuya Suhara, Makoto Higuchi and Hiroyuki Takuwa in Journal of Cerebral Blood Flow & Metabolism