Exacerbation of hepatic damage in endothelial aquaporin 1 transgenic mice after experimental heatstroke.

Background: Heatstroke is a life-threatening disease that causes body uid loss and electrolyte abnormalities due to exposure to high ambient temperature (AT) and relative humidity (RH). Recently, global warming has increased the number of heatstroke patients worldwide. Aquaporin 1 (AQP1) is a water-selective protein that plays an important role in water homeostasis. However, the role of AQP1 in heatstroke has not been elucidated yet. Therefore, in this study, we examined the role of endothelial AQP1 using Tie2-Cre/LNL-AQP1 double transgenic (dTG) mice with an upregulation of aqp1 on endothelial cells. Methods: Tie2-Cre/LNL-AQP1 dTG mice were generated by breeding CAG-LNL(STOP)-AQP1 Tg and Tie2-Cre Tg mice. The transgenes were evaluated in the brain because brain vessels do not express aqp1. For production of experimental heatstroke, the mice were kept in a heatstroke chamber that was pre-heated at 41 °C AT and >99% RH for 1 h, and the blood, brain, kidney, and liver were collected 24 h later. Blood samples were analyzed for biochemical parameters such as electrolytes and tissue damage markers, and the organs were examined via morphological and immunohistological (3-nitrotyrosine (3-NT), AQP1, and Iba-1) staining. Results: Although no difference was observed in aqp1 expression in the whole brain, aqp1 was detected only in Tie2-Cre/LNL-AQP1 dTG mice after capillary deprivation. Moreover, AQP1 immunostaining also revealed immunoreaction in blood vessels. After heat exposure, electrolyte abnormalities were observed in both the wild and Tie2-Cre/LNL-AQP1 dTG mice compared to non-heat stroke wild mice. Hepatic damage markers, aspartate aminotransferase and alanine aminotransferase, were increased, and serum lactate dehydrogenase levels were signicantly higher in Tie2-Cre/LNL-AQP1 dTG mice than in wild mice. Hematoxylin-eosin staining and 3-NT immunoreactivity in the liver also supported hepatic damage. Moreover, the number of hepatic vasculature adherent Iba-1 positive cells was signicantly higher in Tie2-Cre/LNL-AQP1 dTG mice than in wild mice. Conclusions: In the present study, for the rst time, we suggested that endothelial AQP1 contributes to hepatic damage after heatstroke. + , K + , and Cl - ), glucose, lactate (LA), and osmotic pressure were analyzed. TP was analyzed using the biuret method; ALB was analyzed using the BCG method; BUN was analyzed using the urease-GLDH method; T-bil, Cre, and LA were analyzed using the enzymatic method; AST, ALT, ALP, LDH, and CK were analyzed using the MDH-UV method; electrolytes were analyzed using the ion selective electrode method; and osmotic pressure was analyzed using the freezing point depression method. All assays were performed using a HITACHI 7180 (Hitachi, Tokyo, Japan). and Tie2-Cre/LNL-AQP1 dTG mice exhibited electrolyte abnormalities, Tie2-Cre/LNL-AQP1 dTG mice showed a signicantly increased liver injury, 3-NT levels, and hepatic vasculature adherent monocyte/macrophages. These results suggest that an increase in endothelial AQP1 plays an important role in hepatic inammation.

Tie2-Cre/LNL-AQP1 dTG mice on the B6 background were designed for aqp1 expression in ECs. Wild-type mice were obtained from littermates.
Capillary depletion (CD) in brain and mRNA isolation To con rm dTG mice, we checked the gene expression of aqp1 in the brain because brain ECs generally do not express aqp1 (9). CD was performed as previously described with minor modi cations (16,17). After cervical dislocation in male Tie2-Cre/LNL-AQP1 dTG mice and wild littermate mice, the telencephalon was freshly obtained, and the pia maters were carefully removed with a paper towel. The brain was then homogenized with 4.2 mL of capillary buffer (10 mM HEPES, 141 mM NaCl, 4 mM KCl, 2.8 mM CaCl 2 , 1 mM NaH 2 PO 4 , 1 mM MgSO 4 , 10 mM D-glucose, pH 7.4) using a loose glass dounce tissue grinder (10 strokes, Wheaton, Millville, NJ), 10.2 mL of 26 % dextran (Sigma, St Louis, MO) was added, and the mixture was gently mixed. The homogenate was centrifuged at 4,300 × g in a swing-bucket rotor for 30 min, and the supernatant was removed. The pellet was resuspended in 10 mL capillary buffer, and it was centrifuged at 8,000 × g for 30 min. The pellet was mixed with TRIzol Reagent (Invitrogen, Carlsbad, CA), and mRNA was isolated (18). The purity and concentration of extracted RNA were determined using a spectrophotometer (NanoDrop, Wilmington, DE, USA). A High-Capacity RNA-to-cDNA Kit (Applied Biosystems, Foster City, CA, USA) was used to synthesize cDNA using 2 μg total RNA, according to manufacturer's instructions. The experiment was repeated twice to check reproducibility. PCR PCR was performed using TaKaRa Ex Taq (TaKaRa, Shiga, Japan). The reaction mixture was prepared with an appropriate volume of cDNA mixture containing the following: 0.25 μL of forward and reverse primers (50 nmol/mL), 2.0 μL of dNTP mixture (0.25 mM each), 0.1 μL of TaKaRa Ex Taq (5 units/μL), and 2.0 μL of 10 × Ex Taq Buffer in a total volume of 20 μL. Thermal cycling parameters were set as follows: 95 °C for 1 min for initial denaturation, 30-40 cycles at 95 °C for 45 s, 55 °C or 60 °C for 30 s, and 72 °C for 45 s. At the end of the nal cycle, an additional 7-min extension step was performed at 72 °C. Ten microliters of each reaction mixture was electrophoresed on a 2.0 % agarose gel, and the bands were visualized using ethidium bromide. Mouse ribosomal protein S18 (rps18) was used as the housekeeping gene. The primers are shown in Table 1. Foot note, All primers were obtained from Euro ns Genomics (Tokyo, Japan).

Multiple-staining for AQP1 localization
To visualize AQP1 on ECs in the brain, we examined the in situ labeling of ECs with tomato-lectin and multiple immunostaining. Following an overdose of sodium pentobarbital (100 mg/kg, i.p.) anesthesia, the male Tie2-Cre/LNL-AQP1 dTG mice and wild littermate mice (n=3) aged 11 month olds were transcardially perfused with 0.9% NaCl to remove blood, and subsequently, 1% paraformaldehyde was injected. The mice were then perfused with 10 mL of 5 μg/mL DyLight 488-labeled Lycopersicon Esculentum (Tomato) lectin (Vector Lab, Burlingame, CA), and the brain was collected. The brain was immersed in 20% sucrose in 0.1 M phosphate buffer (pH 7.2; PB) for two nights, and it was embedded in liquid nitrogen-cooled isopentane using an embedding solution (20% sucrose in 0.1 M PB: O.C.T. compound (Sakura Finetech, Tokyo, Japan, 2:1). Coronal sections (thickness, 8 μm) were obtained using a cryostat (Hyrax50, Carl Zeiss, Oberkochen, Germany), and they were immunostained.
After washing with PBS, the sections were subjected to heat-mediated antigen retrieval with 10 mM sodium citrate (pH 6.0) at 95°C for 25 min. The sections were washed with phosphate-buffered saline (PBS) containing 0.05% Tween 20 (PBST), and they were incubated with MOM mouse IgG blocking (Vector Lab). The sections were blocked with 5% normal horse serum (NHS, Vector) for 60 min, and subsequently, they were incubated with mouse anti-rat AQP1 (1/22) (1:200, Santa Cruz Biotechnology, Santa Cruz, CA) and/or rabbit anti-GFAP (1:10, DAKO, Glostrup, Denmark) antibodies at 4 °C overnight. The sections were incubated with appropriate Alexa labeled secondary antibodies at room temperature for 2 h. The cell nuclei were stained with 4,6diamidine-2-phenylindole dihydrochloride (1:10,000; Roche, Manheim, Germany) for some cases, and they were incubated in 1.0 mM CuSO 4 in 50 mM ammonium acetate buffer (pH 5.0) to diminish auto uorescence (19). Fluorescence was detected using an Axio Imager optical sectioning microscope with ApoTome (Carl Zeiss).

Heatstroke model
The acrylic semi-enclosed heatstroke chamber (200 mm × 340 mm × 300 mm) was originally produced by placing one animal cage on top of another, similar to a greenhouse (20). An ultrasonic humidi er (USB-68, Sanwa, Okayama, Japan) and a digital thermo-hygrometer (AD-5696, A&D Company, Tokyo, Japan) were used for humidi cation and monitoring of AT, RH, and wet bulb globe temperature (WBGT). WBGT is an environmental index that accounts for AT and RT in a room (21,22), and it recommends the use of International Labour Organization to evaluate the environment of workers (https://www.ilo.org). The heatstroke chamber was placed in an incubator (Bio-chamber, BCP-120F, TITEC, Aichi, Japan), which was pre-heated at 41 °C AT and >99% RH for 3 h or more.
Three hours prior to heat exposure, the male mice aged 11 month olds were weighed and dehydrated. Following this, the mice were placed in a heatstroke chamber for 60 min, and they were returned to room temperature (23 ± 1 °C) with free access to food and water. Body weights (BWs) were measured four times before dehydration, immediately before/after heat exposure, and after 24 h of heat exposure. Nine mice (four wild and ve Tie2-Cre/LNL-AQP1 dTG mice) were subjected to heat exposure.
After twenty-four hours of 1 h heat exposure, the mice were anesthetized with an overdose of sodium pentobarbital (100 mg/kg, i.p.). Blood samples were collected from the right ventricle of the heart, and the mice were transcardially perfused with 0.9% NaCl and 10% neutralized formalin. Blood samples were centrifuged at 1,500 × g for 10 min to collect the serum. Serum samples were examined for biochemical parameters. Fixed organs, such as the liver, kidney, and brain, were collected, para n-embedded 4 μmthick sections were prepared, and morphohistological changes were evaluated.

Serum biochemical parameters
Serum levels of total protein (TP), albumin (ALB), total bilirubin (T-bil), aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), lactate dehydrogenase (LDH), creatinine kinase (CK), blood urea nitrogen (BUN), creatinine (Cre), electrolytes (Na + , K + , and Cl -), glucose, lactate (LA), and osmotic pressure were analyzed. TP was analyzed using the biuret method; ALB was analyzed using the BCG method; BUN was analyzed using the urease-GLDH method; T-bil, Cre, and LA were analyzed using the enzymatic method; AST, ALT, ALP, LDH, and CK were analyzed using the MDH-UV method; electrolytes were analyzed using the ion selective electrode method; and osmotic pressure was analyzed using the freezing point depression method. All assays were performed using a HITACHI 7180 (Hitachi, Tokyo, Japan).

Immunohistochemistry and cell counts
Para n-embedded sections were depara nized using a series of concentrations of xylene and alcohol, and they were subjected to hematoxylin-eosin and immunohistochemical staining.
After washing with PBS, the sections were subjected to heat-mediated antigen retrieval with 10 mM sodium citrate buffer (pH 6.0) at 95 °C for 25 min, and subsequently, they were immersed in 0.3 % H 2 O 2 /MeOH for 30 min. For AQP1 immunostaining, the sections were incubated with MOM mouse IgG blocking. The sections were blocked with non-speci c binding with 5% NHS, and they were incubated with rabbit anti-Iba1 (1:500, Wako), rabbit anti-3-nitrotyrosine (1:500, Upstate Biotechnology, Lake Placid, NY), or mouse anti-rat AQP1 (1/22) (1:200) antibodies at 4 °C overnight. The sections were incubated with either biotinylated goat anti-rabbit (Invitrogen) or anti-mouse IgG (DAKO) secondary antibodies for 2 h. They were then incubated in an avidin-biotin complex solution (Vector) and diaminobenzidine (Sigma), a chromogen. Sections were observed under a microscope (Olympus BX53; Olympus, Tokyo, Japan), and images were captured with an Olympus cellSens standard 2.1 (Olympus). Staining was conducted on sections obtained from 4 and 5 wild-type and Tie2-Cre/LNL-AQP1 dTG mice, respectively.
To determine the in uence of AQP1 on monocyte / macrophage in ltration, we manually counted the number of monocytes / macrophages attached to the vasculature after Iba1-immunostaining in the brain, liver, and kidney using the process mode in Olympus cellSens standard 2.1. The number of Iba1-positive immunoreactions was counted in 30 vessels of the telencephalons and livers and 20-24 vessels in the kidneys from 3-4 to sections in each mouse. The boundary length of counted vessels was measured, and the estimated diameter of each vessel was calculated. The cells were counted in small-and medium-sized vessels, and they were not counted in capillaries, including sinusoids of the liver. An investigator (M.N. and H.Y.) who was blinded to the mouse genotype performed the counting.

Statistical analysis
Data are expressed as the mean ± standard error of the mean. Statistical comparisons were conducted using one-way analyses of variance following non-parametric multiple comparisons, as indicated in each gure legend. A value of p<0.05 was considered statistically signi cant. Analyses were performed using Bell Curve for Excel (Bell Curve, Tokyo, Japan).

Results
Evaluation of Tie2-Cre/LNL-AQP1 dTG mice AQP1 was rst identi ed in erythrocyte membranes (23) and is expressed in the kidneys, lungs, heart, liver, cornea, blood vessels, cornea, and choroid plexus (6). However, it is not usually expressed in brain blood vessels (9). Therefore, we examined aqp1 expression in the capillaries of telencephalon in the brain. We evaluated the speci city of Tie2-Cre/LNL-AQP1 dTG mice. Aqp1 expression was similarly detected in whole brain homogenates of both wild-type and Tie2-Cre/LNL-AQP1 dTG mice ( Figure. 1a). However, after collecting the cerebral vasculature (CD fraction), aqp1 expression could accurately detect signals in Tie2-Cre/LNL-AQP1 dTG mice but not in wild mice. Since the vasculature transcriptomes, pecam (two sets of primers) and vwf, were similarly expressed in both animals, the brain in Tie2-Cre/LNL-AQP1 dTG mice suggested an increase in aqp1 expression. Moreover, CD fraction showed a lower expression of neuronal or glial transcriptomes, such as eno2, gfap, mbp, and aif1, than the whole brain fraction. It also indicated that the brain in Tie2-Cre/LNL-AQP1 dTG mice showed an increased vasculature aqp1 expression. We then observed the stained brain sections for AQP1 antibody ( Figures. 1b and c). AQP1-immunoreactions (-ir) in wild mice were observed in the neuronal layer of piriform cortex. However, AQP1-ir in Tie2-Cre/LNL-AQP1 dTG mice was observed in vasculaturelike structures as well as in the neuronal layer of cortex. Following this, we examined the in situ vasculature with tomato-lectin, and we performed multiple staining of AQP1. Although AQP1-ir in the wild mice was weakly observed in the same eld with higher magni cation, AQP1-ir did not merge with tomato-lectin signals and it mostly merged with GFAP-ir. On the other hand, in Tie2-Cre/LNL-AQP1 dTG mice, AQP1-ir was merged with tomato-lectin signals, indicating its expression on ECs ( Figure. 1c).
Changes in AT, RH, and WBGT during heat exposure We exposed the animals to experimental heatstroke, which were subjected to high AT and RH for 60 min. Figure. 2a shows trends in AT, RH, and WBGT in the semi-enclosed heatstroke chamber. During 3 h of incubation and before heat exposure, AT and RH values inside the chamber were 38.1 °C and >99.9 %, respectively, and WBGT was calculated at 42.0 °C. Although these conditions temporally decreased when the animals were dropped into the chamber, AT and WBGT gradually increased and nally reached 39.9 °C and 43.9 °C after 60 min. The animals exhibited jumping at approximately 40 min after exposure and/or gathering at the corner of the chamber. However, no animals died for 60 min during heat exposure and 1 day following recovery.
Changes in BW during the experimental period are shown in Figure. 2b. Compared to prior hydration, BW decreased to approximately 96.5 % after heat exposure, and it returned to basal level 1 day after heat exposure. No signi cant difference was observed between the groups.
Serum biochemical parameters after 1 day of heat exposure The electrolytes, such as Na + , K + , and Cl -, and osmotic pressures are shown in Figure. 3. To con rm electrolyte abnormalities after heat exposure, Na + and K + were decreased and Clwas increased in both experimental groups compared to that in non-heatexposed wild animals. Signi cant differences were observed for all electrolytes in wild mice and for Clin Tie2-Cre/LNL-AQP1 dTG mice after heat exposure compared to non-heat-exposed wild animals; however, no difference was observed between the heat expose groups.
Next, we determined tissue damage markers in the serum after heat exposure ( Figure. 4). Most of them were not signi cantly different between the wild and Tie2-Cre/LNL-AQP1 dTG mice after heat exposure. However, LDH, a tissue damage marker, was signi cantly higher in Tie2-Cre/LNL-AQP1 dTG mice than in wild-type mice (p<0.05). Moreover, hepatic damage markers, ALT and AST, increased in Tie2-Cre/LNL-AQP1 dTG mice; however, the difference was not signi cant. These results suggest that Tie2-Cre/LNL-AQP1 dTG mice could increase tissue damage, probably hepatic damage, as compared to wild mice after heat exposure.
Increase in in ammation in the liver of Tie2-Cre/LNL-AQP1 dTG mice We compared hepatic morphological alterations between the wild and Tie2-Cre/LNL-AQP1 dTG mice after heat exposure (Figure. 5). Hepatocytes in the non-heat exposure wild mice exhibited large and round nuclei, with distinct outline. The sinusoid radiated outward from the central vein. After heat exposure, the nuclei in the hepatocytes of wild mice somewhat shrank, and they were deeply stained. The sinusoids were still recognized radially, and few leukocytes were also observed. After heat exposure, the hepatic lobules in Tie2-Cre/LNL-AQP1 dTG mice had a fuzzy sinusoid arrangement, and many leukocytes were observed. The nuclei in the hepatocytes shrank, and they were deeply stained. Morphological observations suggested that the hepatic tissues in Tie2-Cre/LNL-AQP1 dTG mice were in amed.
Increase in oxidative stress in Tie2-Cre/LNL-AQP1 dTG mice To con rm hepatic damage, a proteinic oxidative metabolite, 3-NT, was immunostained in the liver ( Figure. 6). Minimal or no staining was observed in the primary antibody-free negative control staining in the livers of Tie2-Cre/LNL-AQP1 dTG mice. 3-NT-ir was slightly detected in wild mice, and it was prominent in the vessels, including sinusoids of the livers of Tie2-Cre/LNL-AQP1 dTG mice. This result also supported the increased hepatic damage in Tie2-Cre/LNL-AQP1 dTG mice.

Expression of hepatic AQP1 on vessels
We then determined the localization of AQP1 in the liver after 24 h of heat exposure ( Figure. 7). AQP1-ir was ubiquitously expressed in the hepatic vasculature, including the sinusoids of non-heat-exposed wild mice, and AQP1-ir showed no staining in the hepatocytes and small bile ducts of portal triads. AQP1-ir staining was not drastically different between the wild and Tie2-Cre/LNL-AQP1 dTG mice after heat exposure.
Increase in Iba1-positive cells on vessels in the liver of Tie2-Cre/LNL-AQP1 dTG mice Next, we stained the hepatic tissues with an anti-Iba1 antibody ( Figure. 8). Iba1 is a marker of monocytes/macrophages. Iba1-ir did not drastically differ in the sinusoids between the wild and Tie2-Cre/LNL-AQP1 dTG mice, suggesting the presence of Kupffer cells ( Figures. 8a and b). However, Iba1-ir was highly noticeable on the small-and medium-sized vessels in Tie2-Cre/LNL-AQP1 dTG mice ( Figures. 8c-f). The number of vasculatures that adhered to Iba1 + in the liver was counted among the groups.
Moreover, to compare the livers, Iba1 + numbers were counted in the telencephalic vessels in the brain and interlobal to interlobular vessels in the kidneys (Fig. 9). Generally, the brain does not express AQP1 in wild mice, as shown in Fig. 1, and the kidney is a critical tissue for water homeostasis. Iba1 + numbers on the hepatic vessels of wild mice with/without heat exposure were counted, and they were 1.76 ± 0.16 and 2.55 ± 0.53/mm, respectively. Iba1 + number on the hepatic vessels of Tie2-Cre/LNL-AQP1 dTG mice was 5.36 ± 1.45 (p<0.01), and it was signi cantly greater than that of the wild mice both with/without heat exposure ( Figure. 9a). In contrast, the numbers in the brains and kidneys were less than 2.2/mm in all groups, and they were not signi cantly different ( Figures. 9b and c).
To determine the role of AQP1 in hepatic vessels, the relationship between Iba1 + cells and vessel diameter was determined between the wild and Tie2-Cre/LNL-AQP1 dTG mice ( Figure. 9d). While Iba1 + numbers showed a greater trend in all sizes of vessels in Tie2-Cre/LNL-AQP1 dTG mice than in wild mice, Iba1 + numbers were signi cantly greater in Tie2-Cre/LNL-AQP1 dTG mice, in particular, in relatively small vessels (0 -150 and 150 -300 μm), as compared to that in heat-exposed wild mice.

Discussions
AQP1 is a water-selective transport protein that plays an important role in uid production, osmotic regulation, and electrolyte balance (6). Heatstroke causes body uid loss and electrolyte abnormalities due to exposure to high AT and RH (1), and appropriate rehydration to maintain osmotic and electrolyte balance is an important strategy to decrease the risk of heatstroke (20). However, the role of AQP1 in heatstroke has not been determined yet. In the present study, we generated Tie2-Cre/LNL-AQP1 dTG mice using Cre/LoxP systems with an upregulation of aqp1 on ECs under the control of Tie2 promoter. The mice were then subjected to experimental heatstroke by exposing them to high AT and RH to evaluate the role of vasculature AQP1. To con rm the genetic modi cation of Tie2-Cre/LNL-AQP1 dTG mice, we rst determined the telencephalic AQP1 level because the ECs in the brain do not express AQP1 (9).
Aqp1 expression in the whole brain of both wild-type and Tie2-Cre/LNL-AQP1 dTG mice was similar. However, after CD, aqp1 expression was detected only in Tie2-Cre/LNL-AQP1 dTG mice while the levels of endothelial marker genes, pecam and vwf, were similar in both the wild and Tie2-Cre/LNL-AQP1 dTG mice. Moreover, immunostaining for AQP1 also supported these results. In the brain, AQP1 is known to localize strongly in the apical membrane of choroid plexus (24), neurons, and astrocytes (6, 25). Immunostaining revealed that the neuronal layers of piriform cortex and astrocytes, which show very low intensity, were also recognized as AQP1-ir. The brains of Tie2-Cre/LNL-AQP1 dTG mice showed a clearly labeled vasculature-like structure that merged with ECs.
In the present study, initially, the mice were exposed to AT and RH of 38.1 °C and >99.9%, respectively, and after 60 min, AT and RH became 39.9 °C and >99.9 %, respectively. WBGT was calculated at 42.0 °C and 43.9 °C. In Japan, a WBGT of more than 31°C is considered dangerous for human health (https://www.wbgt.env.go.jp/en/wbgt.php). Therefore, heat stress is considered to be extremely severe and sub-lethal. After heat exposure, the wild and Tie2-Cre/LNL-AQP1 dTG mice showed a 3 -4 % decrease in body weight, with no signi cant difference. Serum Na + and K + levels decreased and Clincreased in Tie2-Cre/LNL-AQP1 dTG mice than in non-heat-exposed wild mice, and no difference was observed between the wild and Tie2-Cre/LNL-AQP1 dTG mice after heat exposure. Although the osmotic pressures were different after heat exposure due to large deviations, the heat-exposed animals showed early symptoms of heatstroke (1).
We compared the serum biochemical parameters, and we observed that the cellular damage marker LDH was signi cantly higher in Tie2-Cre/LNL-AQP1 dTG mice than in wild mice. Moreover, hepatic damage markers, AST and ALT, also increased in Tie2-Cre/LNL-AQP1 dTG mice; however, the increase was not signi cant. Therefore, we further analyzed hepatic morphology. The hepatic morphology and oxidative stress detected by 3-NT-ir also suggested hepatic impairment in Tie2-Cre/LNL-AQP1 dTG mice.
No study has reported the role of AQP1 in heatstroke. However, some studies have suggested its role in hepatic damage. AQP1 expression is increased in liver ECs in cirrhosis via bile duct ligation. AQP1 gene de cient (KO) mice with cirrhosis showed decreased angiogenesis, brosis, and portal hypertension, which depends on the osmotically sensitive microRNA to AQP1 pathway (26-28). These results are consistent with our results, suggesting that an increased expression of AQP1 in ECs deteriorates hepatic injury. The caspase-3 inhibitor prevented pulmonary injury induced by common bile duct ligation, which is an experimental model of hepatopulmonary syndrome, and it decreased apoptosis and endothelial AQP1 levels (29). Fibroblasts obtained from gene-de cient mice with N-glycanase 1 and human gene-silencing broblasts decreased aqp1, and they were resistant to hypotonic lysis (30).
In the present study, we generated mice that highly expressed AQP1 on ECs, and AQP1 immunostaining was also expressed in the intralobular vessels and sinusoids but not in the hepatocytes and intralobular bile ducts of portal triads. Together these results suggest that endothelial AQP1 increment might contribute to hepatic injury, depending on electrolyte abnormalities and osmotic changes.
The role of AQP1 in human heatstroke patients needs further investigation because the localization of AQP1 in the liver can exist in a species difference. The localization of AQP1 in rodents has mainly been reported in ECs, but the levels are minimal or low in the bile ducts and hepatocytes (26-28), similar to our results. However, AQP1 localization in pigs and humans is not only observed in the vessels, also in the bile ducts and hepatocytes (36, 37) and AQP1 is more dominant in bile ducts than in vessels (38, 39).
Finally, we observed that vasculature-adherent Iba-1 + monocytes/macrophages increased in the small and medium sized intralobular vessels of Tie2-Cre/LNL-AQP1 dTG mice, but they were not prominent in sinusoidal Iba-1 + cells, Kupffer cells. An increase in the number of Iba-1 + cells in Tie2-Cre/LNL-AQP1 dTG mice was not observed in the telencephalic and renal vessels.
These results also suggest that an increased hepatic damage in Tie2-Cre/LNL-AQP1 dTG mice can induce the migration of monocytes/macrophages. In particular, Iba-1 + monocytes/macrophages in Tie2-Cre/LNL-AQP1 dTG mice were signi cantly greater in small vessels, probably due to an impairment in hepatic circulation because the sinusoids in Tie2-Cre/LNL-AQP1 dTG mice were unclear and partially obstructed. Expression of endothelial AQP1 might also correlate with the induction of endothelial adhesion molecules and acquisition of circulating leukocytes. Myocardin-related transcription factor-A KO mice showed a signi cantly attenuated neointima formation and decreased expression of aqp1. Moreover, the expression of icam1, mmp9, and itgb1 decreases in KO mice (40). Single nucleotide polymorphisms in the priapism in patients with sickle cell disease are associated with TGFBR3, AQP1, and integrin-αv (41). Further studies are needed to clarify the properties of hepatic vessels and AQP1 after heatstroke.

Conclusions
In the present study, we genetically generated Tie2-Cre/LNL-AQP1 dTG mice with upregulated aqp1 on ECs under the control of Tie2 promoter. After a severe heat exposure, although both the wild and Tie2-Cre/LNL-AQP1 dTG mice exhibited electrolyte abnormalities, Tie2-Cre/LNL-AQP1 dTG mice showed a signi cantly increased liver injury, 3-NT levels, and hepatic vasculature adherent monocyte/macrophages. These results suggest that an increase in endothelial AQP1 plays an important role in hepatic in ammation.

Availability of data and materials
The datasets generated for this study are available on request from the corresponding author

Competing interests
The authors declare that the study was conducted in the absence of any commercial or nancial relationships that could be construed as a potential con ict of interest.  was not different. After capillary depletion (CD), the expression levels of endothelial marker genes (two different primer sets of pecam and vwf) and housekeeping gene (rsp18) were similar in both the wild and dTG mice. However, the expression of aqp1 was greater in dTG than in wild mice. Whole brain fractions were also detected in neuronal (eno2), astroglial (gfap), oligodendroglial (mpb), and microglial (iba1) sections, but they were decreased or not detected in CD fractions, suggesting that CD fractions are rich in ECs. (B) Immuno uorescence staining for AQP1 in the brain demonstrated that AQP1 immunoreactions in dTG were detected in the vessels, in addition to the neurons of piriform cortex. (C) Multiple-staining of AQP1, endothelial (Tometo-lectin), and astroglial (GFAP) markers were used to determine whether AQP1 immunoreactions were well merged with tomato-lectin in dTG.

Figure 2
Condition of heat stroke model and body weight (BW) during the experiment (A) Transition of ambient temperature (AT), relative humidity (RH), and wet bulb globe temperature (WBGT) before and during heat exposure. AT and RH inside the chamber were 38.1 °C and >99.9%, and WBGT was calculated at 42 °C. AT and WBGT temporally decreased when the animals were dropped into the chamber, and after 60 min, AT and WBGT gradually increased and nally reached 39.9 and 43.9 °C, respectively. (B) Compared to the BWs before hydration, which was 100 %, after 1 h of heat exposure they decreased and returned mostly to basal levels after one day of heat exposure. No signi cant difference was observed in BW during the experimental periods. Serum electrolytes and osmotic pressures one day after heat exposure Levels of Na+ (A), K+ (B), Cl-(C), and osmotic pressures (D) determined in wild (wild; n=4) and Tie2-Cre/LNL-AQP1 dTG (AQP1 dTG; n=5) mice. Na+ and K+ were decreased and Clincreased in both heat-exposed groups as compared to those in non-heat-exposed (no-HS) wild animals (n=3). Data are expressed as the mean ± SE. *p<0.05, **p<0.01 (Tukey's post hoc test).

Figure 4
Comparison of serum biochemical parameters, including tissue damage markers Serum biochemical parameters were determined after one day of heat exposure. Lactate dehydrogenase (LDH), a tissue damage marker, was signi cantly higher in Tie2-Cre/LNL-AQP1 dTG mice (n=5) than in wild-type mice (n=4; p<0.05). Data are expressed as the mean ± SE. *p<0.05 (Student's t-test).

Figure 5
Morphological comparison of the liver Representative images of the livers after hematoxylin-eosin staining in the non-heatexposed and heat-exposed wild and Tie2-Cre/LNL-AQP1 dTG (AQP1 dTG) mice. High-magni cation images of each rectangle are shown at the bottom of each image. Leukocytes are observed after heat exposure (arrows). CV: central vein of hepatic lobules. Figure 6 3-nitrotyrosine (3-NT) immunostaining in the liver A proteinic oxidative metabolite, 3-NT, was immunostained in the liver after one day of heat exposure. Minimal or no staining was observed in the primary antibody-free negative control (1st AB-free NC; inset of bottom) in the liver of Tie2-Cre/LNL-AQP1 dTG mice. 3-NT-ir was slightly detected in wild mice (top), and it was prominent in the vessels, including sinusoids of the livers in Tie2-Cre/LNL-AQP1 dTG (AQP1 dTG) mice (bottom).

Figure 7
Page 17/18 AQP1 immunostaining in the liver AQP1 immunostaining was performed in the livers of wild mice with/without heat exposure and Tie2-Cre/LNL-AQP1 dTG (AQP1 dTG) mice. AQP1-ir was ubiquitously observed in the hepatic vasculatures, including the sinusoids of non-heat-exposed wild mice (Nor), and AQP1-ir showed no staining in the hepatocytes and small bile ducts (asterisk) of portal triads (inset). No or few staining was observed in the primary antibody free negative control (1st AB free NC) in the liver of wild mice (Nor). The stainabilities of AQP1-ir were not drastically different between the wild (HS) and Tie2-Cre/LNL-AQP1 dTG (HS) mice after heat exposure. C: central vein of hepatic lobules; P: portal triad.

Figure 8
Iba1 immunostaining in the liver The red-brown colored Iba1-ir was observed in the sinusoids, suggesting the presence of Kupffer cells, and it did not drastically differ between the wild (A) and Tie2-Cre/LNL-AQP1 dTG (AQP1 dTG, B) mice. Iba1-ir (arrows) was also recognized in the small-and medium-sized vessels, such as the interlobular vessels (IV), and it was more prominent in Tie2-Cre/LNL-AQP1 dTG (D and F) mice than in wild mice (C and E).