Goreisan alleviates cerebral edema: Possibility of its involvement in inhibiting aquaporin‐4 function

In Japan, goreisan (GRS) is used to alleviate cerebral edema and relieve headaches. Although improvement of water maldistribution in the brain may be one of the mechanisms of action of GRS for cerebral edema and headache, scientific evidence is limited. Here, we aimed to investigate the action mechanism of GRS against cerebral edema and headaches, focusing on water dynamics in the brain.


INTRODUCTION
Cerebral edema post-traumatic brain injury, ischemic infarcts, and intracerebral hemorrhages may result in increased intracranial pressure, leading to headache and neurological deterioration [1][2][3][4]. In particular, cytotoxic edema and ionic edema result from water entry into the brain without disrupting the blood-brain barrier (BBB) [5].
AQP4, a water channel expressed in astrocytes, is involved in water transfer across the BBB, regulation of cerebrospinal fluid volume, and secretion of hormones [6,7]. AQP4-mediated water transfer is accelerated during cerebral edema formation [8][9][10]. Cerebral edema formation in AQP4-knockout mice is mild [11]. Hence, suppressing AQP4 function could inhibit cytotoxic and ionic cerebral edema formation.
Goreisan (GRS), a Kampo medicine approved in Japan, is used to treat edema and headaches [12][13][14]. Traditionally, GRS has been thought to improve water maldistribution (called suidoku in Japanese Kampo medicine). Therefore, improvement in water maldistribution in the brain may be one of the action mechanisms of GRS against cerebral edema and headache. Although GRS suppresses chronic cerebral edema and AQP4 expression in a rodent model of acute cerebral ischemia [15,16], the effect of GRS on water dynamics and water permeability through AQP4 remains unclear.
Here, we aimed to investigate the effects of GRS on brain edema and water dynamics using a wellestablished mouse model of acute cerebral edema, and to evaluate the effects of GRS on water permeability via AQP4.

Animal experiments
Male C57BL/6N mice (six weeks old, 18-23 g), purchased from Charles River Japan Inc. (Kanagawa, Japan), were housed under a 12/12-h light:dark cycle at 23 ± 3.0 C and 50% ± 20% humidity. The mice had free access to water and standard laboratory food. The mice were divided into sham-intervention (when necessary), control, and GRS groups. Ethical approvals for animal experiments in this study are described in the Disclosure section. In all experiments, fasting duration was set for less than 24 h before GRS or vehicle administration.

Water intoxication and GRS administration
Mouse models of water intoxication were established as previously reported. [11] GRS dissolved in distilled water (1 or 3 g/10 mL/kg) or the vehicle was orally administered. The mice were intraperitoneally injected with distilled water equal to 20% of body weight with DDAVP (0.4 μg/kg) 30 min following the oral administration.

Effect of GRS on different phenotypes after water intoxication
The water-intoxicated mice exhibited tremors, loss of righting reflexes, and decreased respiratory rate, finally leading to death [11]. The onset time of these symptoms was recorded. Tremor was defined as the oscillation of a body part. Righting reflex loss was defined as the loss of an animal's ability to right itself when placed in a supine position. When a decrease in respiratory rate was observed as a clinical sign, the respiratory rate was counted for 10 s at one-min intervals; a decrease in respiratory rate was defined as a drop in the respiratory rate below 10 breaths. The mice with reduced respiratory rate were euthanized under deep sevoflurane anesthesia (3%-4%) through cervical dislocation.

Brain sampling
At 30 min after water intoxication, the mice were euthanized through exsanguination under deep sevoflurane anesthesia (3%-4%). The entire brain or cerebral cortex was removed and used in the subsequent experiments.
In this experiment, the number of animals in each group differed (i.e., sham: n = 12, control: n = 16, GRS 1 g/kg: n = 21, GRS 3 g/kg: n = 20) for the following reasons: the number of animals in the sham group was planned to be smaller than that in the control group based on the estimation by power analysis. Although the same number of animals was planned for the control and GRS groups, injury due to infighting resulted in the exclusion of some mice in the control and GRS groups.
Effect of GRS on brain water content The brain samples were dried at 105 C for 24 h. The brain water content was calculated using the following equation: Brain water content (%) = (wet weight À dry weight)/wet weight Â 100.

Magnetic resonance image acquisition
Magnetic resonance imaging (MRI) data were recorded using a 7.0 Tesla MRI system (BioSpec 70/16; Bruker Biospin, Ettlingen, Germany). After GRS (3 g/kg) or vehicle administration, the mice were anesthetized with isoflurane (1%-3%) and placed on a magnet in a dedicated animal bed. At 30 min post oral administration, the mice were intraperitoneally injected with water containing 10% H 2 17 O, which is equal to 20% of body weight with DDAVP (0.4 μg/kg). Respiratory cycle and body temperature were monitored (model 1025; SA Instruments, NY, USA) during MRI scanning for 50 min (from 5 min before H 2 17 O administration). SPM12 software (Wellcome Trust Centre for Neuroimaging, UK) was used for region-specific signal analysis. Regions of interest (ROIs) in one cross-section (bregma at À0.1 mm), including the motor and sensory cortexes or lateral ventricle (LV) (Figure 1), were selected. Signal intensity changes (ΔSI, %) are expressed as the difference from the mean signal value 5 min before H 2 17 O administration. To summarize water transfer to the brain, the area under the curve (AUC) was calculated from 0 to 30 min (i.e., before respiratory arrest). Details of the MRI measurements and analysis are described in the Supplementary Methods. F I G U R E 2 Effect of goreisan (GRS) on different phenotypes following water intoxication. At 30 min post oral administration of GRS or vehicle (control), C57BL/6N mice were intraperitoneally injected with desmopressin (0.4 μg/kg) in distilled water equal to 20% of their body weight. Following water intoxication, the time points at which tremor (a), loss of righting reflexes (b), and decreased respiratory rate (c, humanitarian endpoint) occurred were recorded. Differences between the control and GRS (1 g/kg) groups or control and GRS (3 g/kg) groups were evaluated using the log-rank test (*, P < 0.025; ***, P < 0.0005; Bonferroni correction).

Evaluation of water permeability in Xenopus laevis oocytes expressing AQP4
Water permeability in X. laevis oocytes expressing AQP4 was examined [17]. The oocytes, defolliculated by collagenase, were injected with human AQP4 complementary RNA (25 ng) or vehicle (mock injection, negative control) and incubated for 48 h at 18 C in Barth's medium. An osmotic swelling test was performed following the transfer of the oocytes from Barth's medium to distilled water with or without GRS. The oocytes were photographed every 2 s immediately after the transfer for 20 s. Osmotic water permeability (Pf, μm/s) was calculated as previously reported [18].

Statistical analyses
All data are expressed as Kaplan-Meier curves, mean ± standard error of the mean. Normality was checked using the D'Agostino & Pearson test (when sample size was eight or more) or Shapiro-Wilk test (when sample size was less than eight). Differences between groups were evaluated using the log-rank test, unpaired t-test with Welch's correction, Mann-Whitney U-test, or one-way analysis of variance, followed by post-hoc Dunnett's multiple comparisons test. All P-values were calculated by two-tailed test. The statistical significance level α was set at 0.025 (log-rank test, Bonferroni correction) or 0.05 (unpaired t-test with Welch's correction, Mann-Whitney U-test, and Dunnett's test). Statistical analyses were conducted using Prism 7 (GraphPad Software, Boston, MA, USA).

Effect of GRS on different phenotypes after water intoxication
The water-intoxicated mice exhibited tremors, loss of righting reflexes, and decreased respiratory rates. No difference in tremor was observed between the control and GRS groups (Figure 2a). In contrast, GRS inhibited the loss of righting reflexes and improved survival (P < 0.025; Figure 2b,c).

Effect of GRS on brain water content
The brain water content in the control group was higher than that in the sham-intervention group (P < 0.001; Figure 3). GRS 1 and 3 g/kg inhibited the increase in brain water content caused by water intoxication (P = 0.0249 and P < 0.001, respectively).

O dynamics in the brain
In the cortex and LV, the ΔSI was suppressed more in the GRS (3 g/kg) group than in the control group F I G U R E 3 Effect of goreisan (GRS) on brain water content. GRS (1 or 3 g/kg) or the vehicle (sham or control) was orally administered to C57BL/6N mice. At 30 min post oral administration, the mice in all groups, except those in the sham group, were intraperitoneally injected with desmopressin (0.4 μg/kg) in distilled water equal to 20% of body weight. At 30 min post water intoxication, the entire brain was collected. The brains of the sham-group mice were collected 30 min post vehicle administration. The water content was calculated using the wet and dry weights of the entire brain. Data in each column represent mean ± SEM (sham: n = 12, control: n = 16, GRS 1 g/kg: n = 21, GRS 3 g/kg: n = 20). The differences between the sham and control groups were evaluated using the Mann-Whitney U-test ( † , P < 0.05). The differences between the control and GRS groups were evaluated using a one-way analysis of variance (ANOVA) followed by a post-hoc Dunnett's test (*, P < 0.05; ***, P < 0.001).
( Figure 4a,b). A comparison of H 2 17 O kinetic parameters (Tables 1 and 2) showed that the maximum signal intensity change (C max ) of the cortex was lower in the GRS (3 g/kg) group than in the control group (P < 0.05). The AUCs of the cortex and LV were lower in the GRS (3 g/kg) group than in the control group (P < 0.05).
Evaluation of water permeability in X. laevis oocytes expressing AQP4 Pf was significantly higher in the control group than in the mock group (P < 0.05; Figure 5). This variation in Pf indicated water permeability via AQP4. Pf was considerably lower in the GRS (500 μg/mL) group than in the control group (P < 0.05), suggesting that GRS inhibits AQP4-mediated water permeability.

Evaluation of water permeability in HEK293 cells expressing AQP4
Calcein is a fluorescent dye that is quenched in a concentration-dependent manner. Therefore, its  Note: Data in each column represent mean ± SEM (n = 6). Abbreviations: C max , maximum signal intensity change (ΔSI, %); AUC, area under the curve of ΔSI from 0 to 1800 s; T max , time of C max ; T half , half time of ΔSI but could not be calculated (NA). The difference between the control and goreisan (GRS) 3 g/kg groups was evaluated using the unpaired t-test with Welch's correction (***, P < 0.001). NA, not applicable.
concentration-dependent fluorescence can be used as a probe of cell volume and transport of water out of cells via AQP4. In the control group, the ratio of fluorescence between each time point and 0 s (F t /F 0 ) decreased rapidly after hypertonic shock, whereas in the GRS (100-1000 μg/mL) and TGN (an AQP4 inhibitor, 500 μM) groups, the F t /F 0 decreased more slowly (Figure 6a). The membrane water permeability, calculated from exponential fitting of the F t /F 0 , was lower in the GRS (1000 μg/mL, P < 0.001) and TGN (500 μM, P < 0.001) groups than in the control group.

DISCUSSION
To our knowledge, this is the first study to evaluate the effects of GRS on water dynamics in the brain and AQP4 function. The brain water content in the mouse model was reduced by GRS, and the symptoms related to cerebral edema improved. Furthermore, GRS reduced the AUC of H 2 17 O in the LV and cortex. These findings indicate that GRS inhibits cerebral edema formation by reducing water influx into the brain. As an action mechanism, GRS suppressed water permeability via AQP4 in X. laevis oocytes and HEK293 cells expressing AQP4. AQP4 is involved in cerebral edema formation. Thus, the inhibitory effect on AQP4 function may be a mechanism underlying the anti-cerebral edema effect of GRS.
AQP4 is involved in acceleration of cerebral edema, including hypoxia-induced cerebral edema and photothrombotic cerebral infarction cerebral edema [19,20], via water transfer from blood vessels to the brain [9]. Therefore, inhibiting AQP4 function may suppress cerebral edema formation and its associated symptoms.

F I G U R E 5 Effect of goreisan (GRS) on water permeability in
Xenopus laevis oocytes expressing aquaporin-4 (AQP4). The oocytes were injected with AQP4 complementary RNA (25 ng) or vehicle (mock injection) and incubated for 48 h at 18 C in Barth's medium. An osmotic swelling test was performed following the transfer of the oocytes from Barth's medium to distilled water with or without GRS. The oocytes were photographed every 2 s immediately after transfer for 20 s. The osmotic water permeability (Pf, μm/s) was calculated. Data in each column represent mean ± SEM (n = 5). The difference between the mock and control groups was evaluated using a Mann-Whitney U-test ( † , P < 0.05). Differences between the control and GRS groups were evaluated using a one-way analysis of variance (ANOVA) followed by a post-hoc Dunnett's test (*, P < 0.05).
F I G U R E 6 Effect of goreisan (GRS) on water permeability in HEK293 cells expressing AQP4. HEK293 cells were transfected with AQP4 complementary DNA, after which water permeability of the plasma membrane was measured using the calcein fluorescence quenching method. The cells were treated with GRS or N-(1,3,4thiadiazol-2-yl)-3-pyridinecarboxamide (TGN-020 [TGN], 500 μM) for 15 min, after which they were stimulated with hyperosmotic solution. (a) Representative calcein signal (the ratio of fluorescence between each time point and 0 s [F t /F 0 ] after hyperosmotic stimulation in control, GRS-treated, and TGN-treated cells. (b) The membrane water permeability calculated from exponential fitting of the data. The differences between the control and each group were evaluated using a one-way analysis of variance (ANOVA) followed by a post-hoc Dunnett's test (***, P < 0.001). Note: Data in each column represent mean ± SEM (n = 6). Abbreviations: C max , maximum signal intensity change (ΔSI, %); AUC, area under the curve of ΔSI from 0 to 1800 s; T max , time of C max ; T half , half time of ΔSI. The difference between the control and goreisan (GRS) 3 g/kg groups was evaluated using the unpaired t-test with Welch's correction (*, P < 0.05; ***, P < 0.001).
Although AQPs have been validated as important drug targets, no available drugs can successfully target AQPs [21]. MRI has been used to analyze brain water dynamics in humans and animals [22][23][24]. Igarashi et al. [22] found that intravenous administration of H 2 17 O in AQP1-and AQP4-knockout mice induces signal changes. AQP1-knockout mice showed no signal change compared to wild-type mice, whereas AQP4knockout mice demonstrated reduced signal changes in the LV. These results indicate that AQP4 is involved in water transfer to the brain. Here, induced signal changes were observed in the LV and cortex, whereas the C max and AUC of the signal changes were reduced following GRS administration in the model. The MRI analysis supported that GRS inhibits water entry to the brain, and the involvement of AQP4 was also predicted.
The MRI analysis also suggested an association between survival and the effect of GRS on water influx into the brain. The H 2 17 O signals of surviving mice (C max of LV: 3.24%, C max of cortex: 2.15%, AUC of LV: 3.26 Â 10 4 %Ás, AUC of cortex: 1.91 Â 10 4 %Ás) were lower than those of deceased mice (C max of LV: 3.73%, C max of cortex: 2.33%, AUC of LV: 4.61 Â 10 4 %Ás, AUC of cortex: 2.82 Â 10 4 %Ás) in the GRS 3 g/kg group. Although no statistical comparisons were made between surviving and deceased mice owing to the small sample size, the results suggest that mice in the GRS group with less cerebral edema survived.
Although GRS has been reported to affect AQP4 expression [15,16], whether GRS directly inhibits the water permeability of AQP4 remains unclear. Here, we examined the effects of GRS on brain water content and symptom development in a model of acute water intoxication, in which AQP4 expression was not expected to fluctuate. Quantitative reverse transcription-polymerase chain reaction and western blotting were performed to measure the mRNA and protein expression of AQP4, respectively ( Figures S2  and S3). No significant changes in the mRNA and protein expression of AQP4 were observed between the control and GRS groups. However, GRS ameliorated the increase in brain water content and inhibited symptoms, suggesting that GRS improved cerebral edema through other mechanisms aside from modulating AQP4 expression. Additionally, GRS inhibited AQP4 function in vitro, which may be one of its mechanisms of action.
This study has a potential limitation; the complete mechanism of GRS in a mouse model of water intoxication remains unclear. As AQP4 is reportedly involved in the pathophysiology of this model, we focused on the effect of GRS on AQP4 function and did not examine other mechanisms. Headache caused by humidity [25][26][27] and rainfall [25] may be related to water maldistribution (called suidoku in Japanese Kampo medicine) and is suggested to involve abnormality in the clearance system of cerebrospinal fluid (i.e., the glymphatic system) [28]. Recently, the perivascular spaces (PVS) have been proposed as the exit route of cerebrospinal fluid, including excitatory and inflammatory substances [29]. AQP4-mediated swelling of astrocyte end-feet is associated with closure of the PVS, accumulation of inflammatory substances, and cortical spreading depression [29]. Therefore, the mechanisms of action of GRS on headache related to suidoku may involve not only the inhibition of water influx into the brain, but also the normalization of the glymphatic system to promote the clearance of cerebral spinal function (CSF) and inflammatory substances. In support of this hypothesis, cerebral blood flow (CBF) and AQP4 protein expression were increased during lower atmospheric pressure in a female meteoropathy mouse model [30], and GRS prevented the increase in CBF [31]. Moreover, GRS is expected to have a potential anti-inflammatory effect by inhibiting AQPs [28]. In our previous study, we found that GRS inhibits the phosphorylation of extracellular signal-regulated kinase (ERK) in cultured cells, such as intestinal cells, and that it may also affect various signal transductions in astrocytes [32]. The relationship between headache and the mechanism of GRS from the perspective of normalizing the glymphatic system and its antiinflammatory effects warrants further investigation.
To our knowledge, this is the first study to report the inhibitory effect of GRS on the AQP4 function, not AQP4 expression. Considering that GRS is used in treating headaches, the results of this study have important clinical implications.

ACKNOWLEDGMENTS
We thank our colleagues in the laboratory for their technical assistance and helpful comments. We thank the Central Institute for Experimental Animals (Kanagawa, Japan) for the MRI acquisitions. We thank the Kitagawa Scientific General Research Institute (Oita, Japan) for evaluating water permeability in X. laevis oocytes expressing AQP4. We thank Editage (www.editage. com) for the English language editing. This study was supported by Tsumura & Co. (Ami-machi, Ibaraki, Japan).

CONFLICT OF INTEREST STATEMENT
Author Tomofumi Shimizu, Chinami Matsumoto, and Toshitaka Kido are employed by Tsumura & Co. (Amimachi, Ibaraki, Japan). This study received funding from Tsumura & Co. The funder was involved in the following aspects of the study: study design, data collection and analysis, publication decision, and manuscript preparation. The authors declare no other competing interests.
DATA AVAILABILITY STATEMENT All data generated or analyzed during this study are included in this published article and its supplementary information files.