Distribution of aquaporins and sodium transporters in the gastrointestinal tract of a desert hare, Lepus yarkandensis

Lepus yarkandensis is a desert hare of the Tarim Basin in western China, and it has strong adaptability to arid environments. Aquaporins (AQPs) are a family of water channel proteins that facilitate transmembrane water transport. Gastrointestinal tract AQPs are involved in fluid absorption in the small intestine and colon. This study aimed to determine the distribution of AQPs and sodium transporters in the gastrointestinal tract of L. yarkandensis and to compare the expression of these proteins with that in Oryctolagus cuniculus. Immunohistochemistry was performed to analyse the cellular distribution of these proteins, and the acquired images were analysed with IpWin32 software. Our results revealed that AQP1 was located in the colonic epithelium, central lacteal cells, fundic gland parietal cells, and capillary endothelial cells; AQP3 was located in the colonic epithelium, small intestinal villus epithelium, gastric pit and fundic gland; AQP4 was located in the fundic gland, small intestinal gland and colonic epithelium; and epithelial sodium channel (ENaC) and Na+-K+-ATPase were located in the epithelial cells, respectively. The higher expression levels of AQP1, AQP3, ENaC and Na+-K+-ATPase in the colon of L. yarkandensis compared to those in O. cuniculus suggested that L. yarkandensis has a higher capacity for faecal dehydration.


Results
Histology of the stomach, small intestine and large intestine of L. yarkandensis. Since the histological structure of the L. yarkandensis gastrointestinal tract has not been reported, we used haematoxylin and eosin staining to observe this structure. After the stomach, small intestine and large intestine of L. yarkandensis were fixed with 4% paraformaldehyde, paraffin sections of these tissues were stained with haematoxylin and eosin. The gastric mucosal epithelium of L. yarkandensis is mainly composed of surface mucous cells (SMCs), and some parts of the epithelium are depressed to form many gastric pits (GPs) (Fig. 1A-C). The fundic gland of L. yarkandensis can be divided into the neck, body and bottom. The neck is connected to the gastric pits, the body is relatively long, and the bottom extends to the mucosal muscle. The fundic glands are mainly composed of parietal cells and chief cells (Fig. 1D-F). Parietal cells (PC) show a pink colour when stained with haematoxylin and eosin; they have a large volume and their nuclei are round and located at the middle of the cell. Chief cells (CC) are blue when stained with haematoxylin and eosin, and their nuclei are round and located at the base of the cell.
The wall of the small intestine of L. yarkandensis is divided into the mucosa, submucosa, muscular layer and serosa, progressing from the inside to the outside. There are many plicas and intestinal villi on the small intestinal Localization of AQP1, AQP3 and AQP4 in the O. cuniculus and L. yarkandensis gastrointestinal tracts. King and colleagues 4 demonstrated that AQPs are a family of highly conserved water-specific membrane-channel proteins. AQP1, AQP3 and AQP4 mRNA and amino acid sequences were available in GenBank (Table 1). Furthermore, amino acid sequence alignment showed that the AQP1, AQP3 and AQP4 amino acid sequences were 99%, 98% and 99% identical, respectively, between O. cuniculus and L. yarkandensis. Immunohistochemistry was performed to analyse the localization of AQP1, AQP3 and AQP4 in the gastrointestinal tract of O. cuniculus and L. yarkandensis. AQP1 staining was localized in endothelial cells of capillaries in the surrounding gastric pit (Fig. 4A,E), parietal cells of the fundic gland (Fig. 4B,F), central lacteal cells of the small intestinal villus (Fig. 4C,G), and surface-absorptive cells of the colonic epithelium (Fig. 4D,H). Densitometric analysis of the immunohistochemical results revealed higher expression levels of AQP1 in the gastric pit, small intestinal villus, and colonic epithelium of L. yarkandensis-166 ± 19%, and 202 ± 14%, 168 ± 12% of those in O. cuniculus, respectively (n = 6 animals per group) (Fig. 4I,K,L)-and a decreased level of AQP1 in the fundic gland ( Fig. 4J) of L. yarkandensis, 97 ± 2% of that in O. cuniculus (n = 6 animals per group).
AQP4 staining was localized in surface mucous cells of the gastric pit (Fig. 6A,E), parietal cells of the fundic gland (Fig. 6B,F), small intestinal gland cells (Fig. 6C,G) and the colonic epithelium (Fig. 6D,H). Densitometric analysis of the immunohistochemical results revealed higher expression levels of AQP4 in the gastric pit and colonic epithelium of L. yarkandensis-137 ± 2% and 123 ± 11% of those in O. cuniculus, respectively (n = 6 animals per group) (Fig. 6I,L)-and lower levels of AQP4 in the fundic gland and small intestinal gland of L. yarkandensis-82 ± 2% and 82 ± 5% of those in O. cuniculus, respectively (n = 6 animals per group) (Fig. 6J,K). Together, these results suggested that the expression levels of AQP1, AQP3 and AQP4 in the gastric pit, small intestinal villus and colonic epithelium, especially AQP1 and AQP3 in the colonic epithelium, are higher in L. yarkandensis than in O. cuniculus. Localization of the epithelial sodium channel and na + -K + -Atpase in the O. cuniculus and L. yarkandensis gastrointestinal tracts. Water absorption through AQPs is driven by an osmotic gradient that is generated by transcellular Na + transport. Na + entry is conductive and mediated by the apically located epithelial sodium channel (ENaC), and Na + exit is mediated through the basolateral Na + -K + -ATPase. Therefore, we investigated the localization of ENaC and Na + -K + -ATPase in the O. cuniculus and L. yarkandensis gastrointestinal tracts. ENaC and Na + -K + -ATPase mRNA and amino acid sequences were available in GenBank (Table 1). Amino acid sequence alignment showed 98% sequence identity in the ENaC and Na + -K + -ATPase amino acid sequences between O. cuniculus and L. yarkandensis. ENaC staining was localized in surface mucous cells of the gastric pit (Fig. 7A,E), parietal cells of the fundic gland (Fig. 7B,F), surface-absorptive cells of the small intestinal villus epithelium (Fig. 7C,G) and the colonic epithelium (Fig. 7D,H). Densitometric analysis of the immunohistochemical results revealed higher expression levels of ENaC in the gastric pit, small intestinal villus, and colonic epithelium of L. yarkandensis-203 ± 8%, 156 ± 14%, 155 ± 5% of those in O. cuniculus, respectively (n = 6 animals per group) (Fig. 7I,K,L)-and lower levels of ENaC in the fundic gland of L. yarkandensis, 79 ± 5% of that in O. cuniculus (n = 6 animals per group) (Fig. 7J).
Na + -K + -ATPase staining was localized in surface mucous cells of the gastric pit (Fig. 8A,E), parietal cells of the fundic gland (Fig. 8B,F), and surface-absorptive cells in the small intestinal villus epithelium (Fig. 8C,G) and colonic epithelium (Fig. 8D,H). Densitometric analysis of the immunohistochemical results revealed higher expression levels of Na + -K + -ATPase in the small intestinal villus and colonic epithelium of L. yarkandensis-170 ± 6% and 282 ± 10% of those in O. cuniculus, respectively (n = 6 animals per group) (Fig. 8K,L)-and lower  www.nature.com/scientificreports www.nature.com/scientificreports/ levels of Na + -K + -ATPase in the fundic gland of L. yarkandensis, 71 ± 3% of that in O. cuniculus (n = 6 animals per group) (Fig. 8J). Together, these results indicated that the levels of epithelial sodium channel and Na + -K + -ATPase expression in the small intestinal villus and colon epithelium were higher in L. yarkandensis than in O. cuniculus. mRNA expression levels of AQP1, AQP3, epithelial sodium channel and Na + -K + -Atpase in the O. cuniculus and L. yarkandensis colon. Immunohistochemistry data concluded that AQP1, AQP3, ENaC and Na + -K + -ATPase proteins were higher expression levels in the colon of L. yarkandensis than those of O. cuniculus. It is necessary to investigate whether the higher expression of AQP1, AQP3, ENaC and Na + -K + -ATPase protein abundance was in parallel with their mRNA in the colon of L. yarkandensis. The nucleotide sequence alignment showed that the identity of AQP1 nucleotide sequence among O. cuniculus and L. yarkandensis was 99%, and primer-BLAST showed AQP1 primer that was specific to O. cuniculus and L. yarkandensis AQP1. The alignment results of AQP3, ENaC and Na + -K + -ATPase were similar to AQP1. So, we performed quantitative RT-PCR to determine the levels of AQP1, AQP3, ENaC and Na + -K + -ATPase mRNA expression in the colon of www.nature.com/scientificreports www.nature.com/scientificreports/ O. cuniculus and L. yarkandensis. Quantitative RT-PCR suggested that higher expression levels of AQP1, AQP3, ENaC and Na + -K + -ATPase mRNA in the colon of L. yarkandensis -232 ± 18%, 250 ± 16%, 229 ± 13% and 277 ± 20% of those in O. cuniculus, respectively (n = 6 animals per group) (Fig. 9A-D). Thus, these results were in concordance with immunohistochemistry results.
Water content of O. cuniculus and L. yarkandensis faeces. To quantitatively assess the water content of O. cuniculus and L. yarkandensis faeces, whole-faeces wet/dry weight ratios were determined. The ratio of wet/ dry weights for whole faeces was lower for L. yarkandensis (136 ± 13%, n = 6 animals per group, P <0.01) than for O. cuniculus (246 ± 20%, n = 6 animals per group) (Fig. 10).  www.nature.com/scientificreports www.nature.com/scientificreports/ parietal cells, and capillary endothelial cells of O. cuniculus and L. yarkandensis. The distribution of AQP1 in the colonic epithelium indicated that it is involved in transepithelial water transport, as has been revealed for the proximal tubules and descending thin limb segments of the mammalian nephron 35,36 . A role for AQP1 as a water channel in fluid absorption in the colon corresponds to the significant decrease in the fluid absorption rate in the presence of p-chloromercuribenzenesulfonic acid (a mercurial agent) 37 . AQP1 was located in intestinal lacteals and capillaries, implying a role in the water permeability of lymphatics and capillary beds 38 . The present experiments do not clearly indicate that AQP1 is expressed on parietal cells; however, in this study, AQP1 was found in fundic gland parietal cells, which may imply a role in gastric acid secretion. Our results revealed that the expression levels of AQP1 in the colonic epithelium and central lacteal were higher in L. yarkandensis than in O. cuniculus, suggesting that the water permeability of the colonic epithelium, lymphatics and capillary beds is higher in L. yarkandensis. www.nature.com/scientificreports www.nature.com/scientificreports/ We found the locations of AQP4 in the fundic gland, small intestinal gland and colonic epithelium of O. cuniculus and L. yarkandensis. To date, AQP4 has been detected in the stomach. Experiments with rats demonstrated that AQP4 was localized to the basolateral membrane of fundic gland parietal cells 49 . Subsequently, experiments with humans suggested that AQP4 was cloned from the stomach and localized to both parietal cells and chief cells of the fundic gland 50 . It has been postulated that AQP4 participates in gastric fluid secretion. However, experiments with AQP4 knockout mice revealed no effect of AQP4 deletion on gastric fluid secretion, and the results provided direct evidence against a role of AQP4 in gastric fluid secretion 51 . Our results showed that weak labelling was detected in the colonic epithelium of O. cuniculus and L. yarkandensis, in agreement with previous  www.nature.com/scientificreports www.nature.com/scientificreports/ findings in transgenic null mice, in which there was little or no effect of AQP4 deletion on colonic fluid transport or faecal dehydration 30 .
Water absorption through AQPs is driven by an osmotic gradient that is generated by transcellular Na + transport. Apical Na + entry in surface-absorptive cells of the colonic epithelium is mediated by the ENaC, and basolateral Na + exit is mediated through the Na + -K + -ATPase, sodium transporters that were identified in the mammalian colon and lung at the mRNA and protein levels 31,52 . We found the ENaC and Na + -K + -ATPase distribution in the colonic epithelium, small intestinal villus epithelium, gastric pit and fundic gland, implying the roles of these transporters in Na + absorption in the colon and small intestinal villus epithelium 31 . This Na + absorption may provide the osmotic gradient for water absorption across both membranes of epithelial cells through apical AQP1 and basolateral AQP3 and AQP4.
We found a lower water content in L. yarkandensis faeces than in O. cuniculus faeces, suggesting that L. yarkandensis had a high capacity for faecal dehydration. This finding is consistent with the observation that animals that inhabit the desert exhibit physiological and morphological adaptations to arid environments, for example, high-concentration urine production and faecal dehydration 53 . Experiments with Octodon degus, a desert rodent, demonstrated that the colon of O. degus had a higher capacity for faecal dehydration than the rat colon 37 . Experiments with rats administered HgCl 2 (an AQP3 functional inhibitor) demonstrated that the faecal water content in the HgCl 2 administration group markedly increased to approximately 4-fold that in the control group 54 . Our results revealed that the expression levels of AQP3 in the colon were higher in L. yarkandensis than in O. cuniculus indicating that the colon of L. yarkandensis has a higher capacity for faecal dehydration.
We found that the expression levels of AQPs in the stomach, small intestine and colon were different. The low AQP expression in the stomach and small intestine was consistent with functional data in vesicles derived from these tissues, suggesting low plasma membrane water permeability 40 . The colonic epithelium is a tight epithelium with substantially higher electrical resistance and probably a much lower paracellular water permeability than the small intestinal epithelium 7 . L. yarkandensis had higher expression levels of AQP1 and AQP3 in colonic epithelium than O. cuniculus, which could contribute to the extraction of water from faeces to produce dehydrated faecal matter. The higher expression levels of AQP1 and AQP3 in colonic epithelium was more likely due to L. yarkandensis living in an arid desert environment for a long time. Ambient pressure can accelerate the rate of evolution of specific stress-sensitive proteins, produce new functions for specific environments or enhance existing functions, and improve animal fitness for this stressful environment 55 . For example, many studies showed that chronic cold exposure caused endotherms increased intestinal nutrients intake to meet increased energy demand for maintaining thermal homeostasis [56][57][58] . And the expression of digestive features that approximately match digestive capacities with dietary loads 59,60 . Furthermore, we also found higher levels of AQP1 and AQP3 in the kidneys of L. yarkandensis 61 . This may be related to the strategy of the L. yarkandensis to conserve body water.
In conclusion, the locations of AQP1, AQP3, AQP4 and sodium transporters in the gastrointestinal tract of O. cuniculus and L. yarkandensis to sum up in Fig. 11. In the gastrointestinal tract of L. yarkandensis, AQP1 was located in the colonic epithelium, central lacteal cells, fundic gland parietal cells, and capillary endothelial cells; AQP3 was located in the colonic epithelium, small intestinal villus epithelium, gastric pit and fundic gland; AQP4 was located in the fundic gland, small intestinal gland and colonic epithelium; and ENaC and Na + -K + -ATPase were located in the colonic epithelium, small intestinal villus epithelium, gastric pit and fundic gland. The dramatically higher expression levels of AQP3 in the colon of L. yarkandensis than in the colon of O. cuniculus revealed that the colon of L. yarkandensis had higher water permeability, and the higher levels of ENaC and Na + -K + -ATPase expression in the colon of L. yarkandensis provided an osmotic gradient for water absorption through AQPs. From these data, it can be concluded that the colon of L. yarkandensis has a higher capacity for faecal dehydration.

Materials and Methods
Animals and tissues. This study was carried out in male adult O. cuniculus and L. yarkandensis (1.5-1.8 kg). All experiments were performed according to international regulations for animal care and were approved by the Animal Care and Use Committee of Xinjiang Uygur Autonomous Region of China. Adult O. cuniculus was provided by the animal laboratory station of Tarim University. L. yarkandensis was collected from Shaya County, Aksu Prefecture, northwest of the Tarim Basin. And animals were assessed to be adult based on a skull length of greater than 75.50 mm. These animals were maintained initially in individual cages and had free access to food and drinking water at all times. One week after feeding, we collected the faeces of these animals to measure their water content; the animals were anaesthetized with 3% pentobarbital sodium (0.9 ml/kg). The stomach, duodenum, jejunum, ileum, caecum, colon and rectum (the length of each segment was approximately 0.5 cm) were removed and placed in ice-cold 0.85% sodium chloride solution to remove the contents and were then inflated with 4% paraformaldehyde (Sigma-Aldrich, Shanghai, China) and fixed overnight for HE staining and immunohistochemical examination 61 . And some tissues were stored in RNA preservation solutions for analysis of RNA levels.
Wet/dry weight ratios of faeces. Faecal samples were collected from L. yarkandensis and O. cuniculus (n = 6 animals per group) and weighed to obtain the "wet" faecal weights. These faeces were then placed in a 60 °C oven with desiccant and weighed after 4 to 6 d. The "dry" faecal weights were recorded after the weights no longer changed on successive days. The ratio of the wet weight to the dry weight of the faeces was calculated as the wet weight obtained by weighing divided by the dry weight.
Haematoxylin and eosin staining. We used haematoxylin and eosin staining to observe the histological structure of the stomach, small intestine and large intestine. Fixed gastrointestinal tissues were dehydrated in a gradient alcohol series, cleared with xylene, and embedded in paraffin. Paraffin-embedded blocks were fixed on a Lycra paraffin slicer (Leica RM2125RTS, Shanghai, China) for serial sectioning (6 μm thick). Sliced slides were deparaffinized with xylene and hydrated with a gradient alcohol series. Sections were stained with haematoxylin (Sigma-Aldrich, Shanghai, China) for 1 min to 2 min, rinsed with running water for 20 min, stained with 0.5% eosin (Sigma-Aldrich, Shanghai, China) for 30 s, dehydrated in a gradient alcohol series, cleared with xylene, and used for histological observation and imaging (Motic BA600, Beijing, China) after sealing.
immunohistochemistry. Localization of the AQP1, AQP3, AQP4, epithelial sodium channel (ENaC) and Na + -K + -ATPase proteins was evaluated in fixed gastrointestinal tissues of L. yarkandensis and O. cuniculus by immunocytochemistry. Immunocytochemical studies were performed in Paraffin-embedded gastrointestinal tissue, previously fixed in 4% paraformaldehyde. The experimental steps of immunohistochemistry have been described elsewhere 61 . The primary antibodies used were as follows: anti-AQP1, anti-AQP3, and anti-AQP4 (diluted to 4.0 μg/ml; Proteintech) and anti-ENaC and anti-Na + -K + -ATPase (diluted to 3.0 μg/ml; Proteintech). The acquired images were analysed in IpWin32 software 61 .