Animals
All experiments were performed in accordance with the Science Council of Japan’s Guidelines for Proper Conduct of Animal Experiments. Experimental protocols were approved by the Institutional Animal Care and Use Committee at the University of Tsukuba. C57BL/6 mice (Jackson Laboratory) were bred in our colony at the University of Tsukuba and maintained on a 12-h light/dark cycle (lights on 9 am-9 pm) with ad libitum access to food and water. All mice were group-housed with 2–5 mice/cage, and only male mice were used in the experiments.
Surgical procedures and recording of EEG and EMG
Implantation of EEG/electro-myogram (EMG) sockets for recording was performed as previously described [65]. Briefly, the EEG/EMG recording electrode was composed of a six-pin header, six stainless steel wires (four 1.5-mm length and two 2.0-mm length), and four stainless steel screws (1.0-mm diameter). Mice were anesthetized with isoflurane (3–5%) during surgery. Using a carbide cutter (drill size: 0.8-mm diameter), four holes were made in the skull: two over the frontal cortical area (1.5 mm anterior to bregma, 1.5 mm lateral to midline) and two over the parietal area (3 mm posterior to bregma, 1.7 mm lateral to midline). Using a jeweler’s screwdriver, stainless steel EEG recording screws were placed into the holes. EMG recording wires were inserted into the trapezius (neck) muscle. After surgery, mice were handled for 2 min three times per day for 4 days and habituated to the recording chamber. The duration between electrode implantation surgery and subsequent behavioral testing was 7 days. Before behavioral testing, mice underwent baseline EEG/EMG recording in their home cage equipped with a data acquisition system (LabChart, AD Instruments, New Zealand; Vital recorder, KISSEI COMTEC, Japan). EEG/EMG signals were recorded during BL and AL periods for each mouse. EEG/EMG data were collected at a sampling rate of 128 Hz.
Fear conditioning paradigm
The context used for fear conditioning consisted of a stainless steel chamber (31 × 24 × 21 cm; MED Associates) with a stainless steel shock grid floor (Additional file 4: Fig. S3A). The grid floor was composed of bars (3.2-mm diameter) spaced 7.9 mm apart that allowed the delivery of electric foot shocks. Under the grid floor was a stainless steel drop pan that was lightly cleaned with 75% ethanol, which also provided background odor. The front, top, and back of the chamber were made of clear acrylic, and the two sides were made of aluminum panels.
For the tone test, a different context was used that consisted of a white plastic floor covering the grid floor and a grey plastic triangular insert placed inside the chamber to create artificial left and right sides (Additional file 4: Fig. S3B). The front side consisted of a piece of cardboard with a blue and white rectangular pattern in the center. The context was cleaned with water instead of ethanol.
Mice in the DS group were placed in the conditioning context for 360 s, and three tones (30 s each, 2800 Hz, 85 dB) were played at 120, 210, and 300 s, with each tone co-terminating with a 2-s foot shock (0.75 s for Fig. 2). Post-shock freezing [47, 48] (Fig. 5c) was analyzed during the last 30 s of conditioning starting immediately after termination of the third tone, which was a tone- and shock-free period. Freezing during the third tone (Fig. 5l–p) was analyzed during the 28-s shock-free period of the third tone. We employed two additional control groups involving no context-shock association; mice in the IS group received three 2-s foot shocks at 2-s intervals immediately after being placed in the context, and mice in the CO group were exposed to the context for 360 s without receiving foot shocks. In these two control groups, tones were played at the same times as in the DS group. Mice were then returned to their home cage. Except for mice from which the DG was sampled, mice were later returned to the same context for 5 min without the delivery of foot shocks (i.e., context test). Three h later, mice were placed in a different context for 360 s, and the tone was played during the last 180 s (i.e., tone test). As we initially expected that allantoin administration would increase freezing levels [62], we weakened the intensity of the foot shocks (0.4 mA) in the oral administration experiments (Fig. 5) to clearly observe a potential enhancement of freezing. A discrimination index [3, 49] (Fig. 5f) was calculated as (freezingcontextA – freezingcontextB first 3 min) /max(freezingcontextA, freezingcontextB first 3 min). A shock reactivity index during conditioning [66] (Fig. 5b) was calculated as (movement in the 2 s immediately before the first shock)/(movement in the 2 s of the first shock), and the nominator was also used to calculate relative movements (Additional file 4: Fig. S3C).
Real-time sleep analysis and tissue preparation
Real-time EEG/EMG recordings to identify sleep–wake states were performed 1–3 h after conditioning. Mice were sacrificed when they exhibited 5 min of continued wakefulness, 5 min of continued NREM sleep, or 1 min of continued REM sleep. These thresholds of time spent in various sleep states were determined based on the average durations of states observed in mice. Mice were sacrificed by overdose of isoflurane within the sleep recording chamber by covering their nose with a 50-ml plastic tube containing paper soaked with liquid isoflurane and immediate decapitation. Their brains were removed and frozen at − 80 °C (< 3 min after completion of the necessary duration of each sleep state). All mice were briefly awake (< 5 s) during isoflurane inhalation before tissue sampling, which may have hindered the full discovery of small changes between sleep–wake states. Brains were sliced coronally (500 µm; from − 1.75 to − 2.25 mm relative to bregma) using a microtome. Slices were viewed under a microscope to dissect the DG in a frozen state.
Sample preparation for brain metabolomic analysis
DG tissue was rapidly frozen in liquid nitrogen and stored at − 80 °C. Frozen tissue was homogenized using a manual homogenizer (BioMasherII, Nippi) in 150 µl ice-cold buffer (80% methanol containing 1% formic acid) with 6 µl of 0.1 mg/ml 2-isopropylmalic acid (Sigma-Aldrich), which was utilized as an internal standard. Samples with < 20 μg total protein were omitted from further analysis. After centrifugation at 16,000g for 30 min at 4 °C, 50 µl supernatant was dried in a centrifugal evaporator for GCMS analysis, and the remaining supernatant was used for LCMS analysis. Metabolites were quantified using GCMS (GCMS-QP2010 Ultra, Shimadzu) or LCMS (LCMS-8060, Shimadzu) as described previously [67]. Briefly, dried extract was subjected to methoximation followed by silylation, and 1 µl derivatized sample was injected into the GCMS system. For LCMS, polar metabolites were separated on a Shodex RSpak DE-213 column and measured with electrospray ionization in positive ion and multiple reaction monitoring mode. All ion transitions and collision energies for multiple reaction monitoring were optimized experimentally using authentic standards for each metabolite. Supernatant was directly injected into the LCMS system for lipid analysis and analyzed with hydrophilic interaction chromatography coupled with the electrospray ionization method [67].
Drag administration
Allantoin (130 μmol/kg/day, Sigma-Aldrich, 05,670), xanthine (91 μmol/kg/day, Sigma-Aldrich, X7375), hypoxanthine (130 μmol/kg/day, Sigma-Aldrich, H9377), or urea (130 μmol/kg/day, Kanto Chemical Co. Inc., JAPAN, 43009-00) was dissolved in drinking water and administered to mice for 7 days before fear conditioning. Mice in the control group were given normal drinking water.
Quantification and statistical analysis
Significant changes in metabolite levels were identified by PLS-DA using MetaboAnalyst (http://www.metaboanalyst.ca) [68, 69]. The quality of the PLS-DA models was assessed for R2, Q2, and accuracy based on a VIP score ≥ 1.0. Pathway analyses were performed using MetaboAnalyst. Pathways were considered affected if they were significantly enriched (p < 0.05) for all significantly altered metabolites. Three outliers were identified and excluded by PCA using MetaboAnalyst 4.0 as previously described [70]; these could have been due to sample degradation, instrumental error, changes in measurement conditions, or faulty measurements resulting from human error. To compare metabolite ratios among subgroups, bootstrapping to obtain 95% CIs and permutation analysis to obtain p-values were carried out by randomly replacing the measured value 10,000 times.
Statistical analysis was performed using GraphPad Prism version 8.4.0 for Windows (GraphPad Software, USA) or Igor Pro version 8.01B01 (Wave Metrics, USA). Type I error was set at 0.05. Shapiro–Wilk tests were performed to assess the normality of data. Brown-Forsythe tests were performed to assess homogeneity of variance. Other details of statistical analyses are described in the figure legends and below.
Figure 1b. REM, n = 10 mice; NREM, n = 10; Wake, n = 11, one-way ANOVA, F(2, 28) = 2.6, p = 0.095.
Figure 1k. See Additional file 1 for details of statistical analysis.
Figure 2b. DS (n = 8 mice) vs. IS (n = 9). Left (context test), two-way ANOVA, group × time, F(4, 60) = 2.67, p < 0.05; time: F(2.26, 33.9) = 1.17, p = 0.32; group, F(1, 15) = 24.5, p < 0.01; Sidak’s multiple comparisons tests, DS vs. IS, 1st min, p < 0.05, 2nd min, p < 0.01, 3rd to 5th min, p < 0.05. Right (tone test), two-way ANOVA, group × time, F(5, 75) = 15.0, p < 0.01; time: F(5, 75) = 24.7, p < 0.01; group: F(1, 15) = 8.85, p < 0.01; Sidak’s multiple comparisons tests, DS vs. IS, 1st to 3rd min, p > 0.5, 4th to 5th min, p < 0.01, 6th min, p = 0.16.
Figure 2c. Top (REM), DS, n = 11 mice; IS, n = 7; CO, n = 9; one-way ANOVA, F(2, 24) = 0.38, p = 0.69. Middle (NREM), DS, n = 10 mice; IS, n = 8; CO, n = 9; one-way ANOVA, F(2, 24) = 2.16, p = 0.14. Bottom (Wake), DS, n = 10 mice; IS, n = 9; CO, n = 10; one-way ANOVA, F(2, 26) = 0.30, p = 0.74.
Figure 2d
Top (REM), DS, n = 11 mice; IS, n = 7; CO, n = 9; one-way ANOVA, F(2, 24) = 0.046, p = 0.95.
Middle (NREM), DS, n = 9 mice; IS, n = 7; CO, n = 7; one-way ANOVA, F(2, 20) = 2.11, p = 0.15.
Bottom (Wake), DS, n = 9 mice; IS, n = 8; CO, n = 9; one-way ANOVA, F(2, 23) = 2.71, p = 0.087.
Figure 2e
Top (REM), DS, n = 10 mice; IS, n = 6; CO, n = 8; Kruscal–Wallies test, Wake, p = 0.19; NREM, p = 0.08; REM, p = 0.16; total, p = 0.28.
Middle (NREM), DS, n = 9 mice; IS, n = 6; CO, n = 7; Kruscal–Wallies test, Wake, p = 0.35; NREM, p = 0.56; REM, p = 0.19; total, p = 0.42.
Bottom (Wake), DS, n = 9 mice; IS, n = 7; CO, n = 8; Kruscal-Wallies test, Wake, p = 0.39; NREM, p = 0.49; REM, p = 0.53; total, p = 0.85. Figure 2O. See Additional file 2 for details of statistical analysis. Figure 3. See Additional file 1 and 2 for details of statistical analysis. Figure 4b-e, l. See Additional file 3 for details of statistical analysis. Figure 5b. Control, n = 16 mice; hypoxanthine, n = 12; xanthine, n = 13; allantoin, n = 19; urea, n = 16, one-way ANOVA, F(4, 71) = 1.36, p = 0.26. Figure 5c. Control, n = 16 mice; hypoxanthine, n = 12; xanthine, n = 13; allantoin, n = 19; urea, n = 16, Kruskal–Wallis test, p = 0.057. Figure 5d. Control, n = 16 mice; hypoxanthine, n = 12; xanthine, n = 13; allantoin, n = 19; urea, n = 16, two-way ANOVA, group × time, F(16, 284) = 0.88, p = 0.71; time, F(3.5, 246.6) = 5.9, p < 0.01; group, F(4, 71) = 0.90, p = 0.47. Figure 5e. Control, n = 16 mice; hypoxanthine, n = 12; xanthine, n = 13; allantoin, n = 19; urea, n = 16, two-way ANOVA, group × time, F(20, 355) = 0.75, p = 0.77; time, F(2.2, 156.6) = 65.5, p < 0.01; group, F(4, 71) = 0.84, p = 0.51.
Figure 5f. Control, n = 16 mice; hypoxanthine, n = 12; xanthine, n = 13; allantoin, n = 19; urea, n = 16, Kruskal–Wallis test, p = 0.88.
Figure 5g. n = 16 mice/group, two-tailed Pearson’s correlation, r = 0.58, p < 0.05.
Figure 5h. n = 12 mice/group, two-tailed Pearson’s correlation, r = 0.22, p = 0.48.
Figure 5i. n = 13 mice/group, two-tailed Pearson’s correlation, r = 0.62, p < 0.05.
Figure 5j. n = 19 mice/group, two-tailed Pearson’s correlation, r = 0.62, p < 0.01.
Figure 5k. n = 16 mice/group, two-tailed Pearson’s correlation, r = 0.27, p = 0.32.
Figure 5l. n = 16 mice/group, two-tailed Pearson’s correlation, r = 0.51, p < 0.05.
Figure 5m. n = 12 mice/group, two-tailed Pearson’s correlation, r = -0.26, p = 0.41.
Figure 5n. n = 13 mice/group, two-tailed Pearson’s correlation, r = 0.39, p = 0.19.
Figure 5o. n = 19 mice/group, two-tailed Pearson’s correlation, r = 0.16, p = 0.51. Figure 5p. n = 16 mice/group, two-tailed Pearson’s correlation, r = 0.28, p = 0.29.
Figure 5q, n = 14 mice/group, two tailed Peason’s correlation, r = 0.54, p < 0.05.
Figure 5r. n = 12 mice/group, two tailed Peason’s correlation, r = -0.10, p = 0.76.
Figure 5s. n = 8 mice/group, two tailed Peason’s correlation, r = -0.11, p = 0.78.
Figure 5t. n = 17 mice/group, two tailed Peason’s correlation, r = 0.38, p = 0.14.
Figure 5u. n = 10 mice/group, two tailed Peason’s correlation, r = 0.30, p = 0.41. Fig. S3C, Control, n = 16 mice; hypoxanthine, n = 12; xanthine, n = 13; allantoin, n = 19; urea, n = 16, Kruskal–Wallis test, p = 0.015; Dunn’s multiple comparisons test, no difference between control vs. hypothantine (p > 0.96), xanthine (p > 0.99), allantoin (p > 0.99), or urea (p > 0.085).