This study was carried out in accordance with the recommendations of Regulations for Animal Experiments and Related Activities at Tohoku University, Center for Laboratory Animal Research, Tohoku University. The protocol was approved by the Center for Laboratory Animal Research, Tohoku University. Virgin female C57BL/6N mice (CLEA Japan, Inc., Tokyo, Japan) about 5 weeks old were maintained under controlled lighting (12-h light cycle) and temperature (24°C). These were divided into two diet groups: Normal (N) and Low Protein (LP) by feeding them AIN-93G and modified AIN-93G (Oriental Yeast Co., Ltd., Tokyo, Japan), respectively. They were allowed free access to food and water ad libitum. The LP diet consisted of only 48% of the protein and 94% of the calorie contents of the N diet as suggested in previous studies (Reeves et al., 1993; Ito et al., 2011a,b). After a 2-week acclimatization period, the female mice were time mated and inspected for vaginal plugs the following morning. Plug-positive females were then transferred to single cages and fed ad libitum. Food consumption and body weight were recorded every day. There was a total of 25 pregnant mice per diet group.

We modified Magal's method (Magal et al., 1988) as follows: on day 17.5 of gestation, about 1–2 days before birth, pregnant mice were anesthetized by subcutaneous injection with Ketamine (Ketalar intramuscular 500 mg, Daiichi Sankyo Propharma Co., Ltd., Tokyo, Japan; 10 mg/kg) and Xylazine (Selactar, Bayer Yakuhin, Ltd., Osaka, Japan; 5 mg/kg). Maintenance of anesthesia was achieved using inhalational 0.5% isoflurane at a 200 ml/min air flow rate (Forane, Abbott Japan Co., Ltd., Tokyo, Japan). Each pregnant mouse was placed on a 38°C temperature controlled stage. Maternal abdomen was depilated with commercial hair removal cream (Veet, Reckitt Benckiser Group plc, Slough, England, UK), and this was followed by an abdominal midline incision. The maternal uterine horns containing 2–8 fetuses were exposed, and warmed with ultrasound gel (Parker Laboratories, Inc., Fairfield, NJ, USA) which also enabled ultrasound transmission described later. One fetus was chosen for ischemia/reperfusion treatment. Interval occlusion and opening of maternal uterine and ovarian arteries every 5 min induced ischemia and reperfusion (Figure 1). The ischemia/reperfusion sequence was repeated three times and the whole process lasted a total of 30 min. The number of fetuses treated in each diet group was n = 25.

Using a Vevo 2100 high-frequency, high-resolution ultrasound device equipped with a linear array transducer MS-550D (VisualSonics, Inc., Toronto, Canada), ultrasound B-mode imaging of a cross section of a fetus brain was performed before and all throughout ischemia/reperfusion intervals (Figure 1). The central frequency of the ultrasound was 40 MHz resulting in the axial resolution of about 40 μm. The field of view was 7 mm in width by 10 mm in depth, and three different focal depths, 4, 6, and 8 mm, were set, simultaneously. One hundred B-mode images were stored at 50 frame/s, wherein each captured cineloop duration was 2 s. Moreover, before and after the ischemia/reperfusion treatment, three-dimensional (3D) ultrasound power Doppler imaging was performed to visualize concomitant blood flow in the area. The scanning length for the 3D imaging was set as 4.98 mm at the interval of 0.102 mm.

To evaluate the presence or absence of intracranial hemorrhage after ischemia/reperfusion treatments, the fetuses were delivered and whole brain samples were collected. Brain tissues were supercooled in dry ice for 30 min and stored at −80°C. These were subsequently mounted using an optimal cutting temperature (OCT) compound (Tissue-Tek 4583, Sakura Finetek Japan Co., Ltd., Tokyo, Japan) and cut on a cryostat (Leica Cryostat CM3050 S, Leica Microsystems GmbH, Wetzlar, Germany) to obtain cross sections of 8 μm thickness. Images of brain sections corresponding to ultrasound B-mode images were captured using Leica CTR 5000 and DM 5000B microscope system (Leica Microsystems GmbH, Wetzlar, Germany).

Changes of blood flow in fetal brains were quantified based on image intensity of ultrasound B-mode images. First, in order to suppress noise in the B-mode images, 100 images obtained at each measurement time point (see Figure 1) were averaged, with adjusting for image shifting due to maternal breathing and heartbeat as well as fetal movement. A customized image analysis program calculated the cross-correlation function between the first image and each subsequent image in the measurement period. Each image was shifted so as to maximize the function, and the time-averaged B-mode image at each time point was then created by averaging image intensities at each pixel. Subsequently, the summation of image intensity, I, in a circular region of interest (ROI) with 15-pixel diameters of 0.35 mm was obtained with a separate image-processor (ImageJ, NIH, Bethesda, MD, USA). The ROIs were set at four different locations: cingulate cortex, area 2 (Cg2), basal forebrain (BF), and bilateral caudate putamen (Cpu). Finally, for each ROI, the summation of image intensities, I, was normalized with that before the treatment, I₀, and the variation of the relative value of I/I₀ was evaluated. A total of five N fetal brains without hemorrhage and five LP fetal brains with hemorrhage, all previously confirmed by above-mentioned histology, were analyzed.

The number of hemorrhage-positive fetuses on histology for each diet treatment group were counted, and the difference between the two groups were evaluated by Fisher's exact test. The differences in intensity of ultrasound B-mode images between the groups were analyzed using a t-test. Statistical significance for each analysis was set at p < 0.05.

A snapshot of fetuses with or without intracranial hemorrhage after ischemia/reperfusion treatments is shown in Figure 2A. A dark region, indicated by a white arrow, in the head of the right-hand side fetus indicates a hemorrhage. The number of fetuses being positive/negative for intracranial hemorrhage determined by histological observation in each group is summarized in Figure 2B. LP fetuses had a 5-fold higher frequency of intracranial hemorrhage than N fetuses; intracranial hemorrhage was noted in 16% and 76% of N and LP groups, respectively. Microscopic observation revealed that bleeding-prone areas were mostly located in the outer peripheral potion of the lateral ventricles, namely Cpu, as observed in the dotted circles in Figure 2C.

Sequential ultrasound B-mode images of the same cross section in Figure 2C were successfully obtained during the ischemia/reperfusion treatments as shown in Figure 3A. This achievement enabled us to quantify changes in blood flow by intensity analysis of the sequential images described later with Figures 3B–D. At the Cpu beside the lateral ventricles where intracranial hemorrhage observed in the microscopic image, increases of the intensity were recognized in the sequential ultrasound B-mode image.

Blood flow in N and LP fetal brains before and after the ischemia/reperfusion treatments are demonstrated in 3D ultrasound power Doppler images of Figure 4. No significant change was observed in the N fetal brain before and after the treatment. On the other hand, several main blood flows disappeared in the LP fetus as indicated by triangles in Figure 4.

Changes in blood flow within the fetal brain were evaluated by intensity quantification of ultrasound B-mode images with each five N and LP mice. Sequential ultrasound B-mode images of five N fetal brains without hemorrhage and five LP fetal ones with hemorrhage, confirmed by histological observation using microscopy, were analyzed. The variations of the relative intensities in ultrasound B-mode images in the four ROIs are shown in Figures 3B–D. Note that the results of Cpu were obtained by combining 10 data sets of bilateral Cpu in five fetuses. Image intensities in all the ROIs of N fetal brains without hemorrhage were slightly increased with fluctuations due to ischemia/reperfusion treatments. In contrast, the image intensities of LP fetal brains gradually decreased in the Cg2 and BF ROIs, but increased in Cpu ROI, showing a large standard deviation. The relative intensities in the Cg2 and BF ROIs of LP fetal brains were significantly lower than those in N fetal brains at some time points after the second ischemic period. No significant difference was detected between the relative intensities in the Cpu ROI between N and LP fetuses, but the fluctuation of relative intensity in the Cpu ROI during the ischemic/reperfusion treatments were opposite between the two groups.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.