Alternations of blood ammonia concentration and astrocytes in HBV transgenic mice
HBV transgenic mice overproduce the hepatitis B virus (HBV) large envelope polypeptide and accumulate toxic quantities of hepatitis B surface antigen (HBsAg) in hepatocyte that develop a severe and persistent hepatocellular injury, leading to precancerous proliferation reaction and unrestrained cell growth [17]. To identify the progress of hepatitis B virus-related liver diseases, alternations of blood ammonia concentration and astrocytes in HBV transgenic mice, we performed HBV transgenic mice C57BL/6J-TG(ALB1HBV)44BRI/J and negative control of C57BL/6 mice.
The H&E staining of mice liver tissues showed that the livers of control mice aged 4 to 18 months were normal, while hepatocellular overexpression of HBV protein leaded to severe and persistent hepatocellular injury in HBV transgenic mice (Fig. 1a). With the HBV accumulation, the liver of HBV transgenic mice aged 4 to14 months gradually progressed fatty liver; Local liver fibrosis in 16 months old HBV transgenic mice; The liver of 18 months old HBV transgenic mice developed extensive liver fibrosis and cirrhosis (Fig. 1a).
Ammonia is a neurotoxin related to the pathogenesis of HE and plays an important role in astrocyte swelling, so we compared the serum ammonia of HBV transgenic mice and control mice. The serum ammonia concentration of HBV transgenic mice from 4 to 18 months old was significantly increased and was significantly higher than control mice at 8 to 18 months old (Fig. 1b).
In order to observe the distribution of cells in the brain of mice, the H&E staining of mice heads tissues showed that granulosa cells were mainly found in the olfactory bulb and cerebellum (Additional file 1: Fig. S1), and astrocytes were mainly observed in cerebral hemisphere and midbrain (Fig. 1c). The intracellular accumulation of glutamine leads to astrocyte swelling, eventually leads to neurological dysfunction [18]. The results in Fig. 1c only indicated that the distribution of cells in HBV transgenic mice and control mice was consistent, but whether the astrocyte swelling could not be determined directly from the images. To further examine the swelling of astrocytes, we measured the diameter of astrocytes at 1000 times magnification. The diameter of astrocytes in the left and right hemispheres of HBV transgenic mice aged 14/10 to 18 months was significantly larger than the control mice (Fig. 1d–e), and the diameter of astrocytes in the midbrain of HBV transgenic mice aged 10 and 18 months was significantly larger than the control mice (Fig. 1f). These results demonstrated that the serum ammonia concentration was up-regulated, and astrocytes were swollen during the development of hepatic disease in HBV transgenic mice.
Alterations of tissue glycopatterns in the brain of HBV transgenic mice
Although abnormal glycosylation has been reported in various diseases, these studies have focused on tumor to exclude brain complications of chronic hepatic disease. Therefore, we sought to determine the alternations of glycosylation in the brain of HBV transgenic mice. The protein of olfactory bulb, left hemisphere, right hemisphere, midbrain and cerebellum from HBV transgenic mice (n = 6, at the same age) and control mice (n = 6, at the same age) were measured by lectin microarrays, which included 37 lectin probes, providing specific information for the glycan repertoire of tissue glycoproteins [19]. The layout of the lectin microarrays is shown in Additional file 1: Fig. S2.
The glycopatterns of proteins in olfactory bulb from HBV transgenic mice and control mice bound to the lectin microarrays and their normalized fluorescent intensities (NFIs) for each lectin are shown in Additional file 1: Fig. S3a. The generated data from each mouse was imported into EXPANDER 6.0 to perform a hierarchical clustering analysis (Additional file 1: Fig. S3b). The results showed the (GlcNAc)2–4 binder LEL, the α-D-Man, Fucα-1,6GlcNAc, α-D-Glc binder PSA, Fucα-1,6GlcNAc, α-D-Man, α-D-Glc binder LCA and the High-Mannose, Manα1-6Man binder NPA, exhibited significantly altered NFIs in the olfactory bulb of HBV transgenic mice compared with control mice (Additional file 1: Fig. S3c).
The glycopatterns of proteins in left hemisphere from HBV transgenic mice and control mice bound to the lectin microarrays, as shown in Fig. 2a. Their NFIs for each lectin are showed in Additional file 1: Fig. S4a. The generated data from each mouse were imported into EXPANDER 6.0 to perform a hierarchical clustering analysis (Fig. 2b). The results showed the β-D-GlcNA, (GlcNAcβ1-4)n, Galβ1-4GlcNAc binder DSA, the α-D-Man, the High-Mannose, Manα1-6Man binder NPA, the Fucα1-2Galβ1-4GlcNAc, Fucα1-3(Galβ1-4)GlcNAc, anti-H blood group specificity binder LTL and the Fucα-1,6GlcNAc, α-D-Glc binder PSA, exhibited significantly altered NFIs in left hemisphere of HBV transgenic mice compared with control mice (Fig. 2c).
The glycopatterns of proteins in right hemisphere from HBV transgenic mice and control mice bound to the lectin microarrays, as shown in Fig. 2d. Their NFIs for each lectin are summarized in Additional file 1: Fig. S4b. The generated data from each mouse was imported into EXPANDER 6.0 to perform a hierarchical clustering analysis (Fig. 2e). The results showed the High-Mannose, Manα1-3Man, Manα1-6Man, Man5-GlcNAc2-Asn binder HHL, the α-/β-linked terminal GalNAc, (GalNAc)n, GalNAcα1-3Gal, blood-group A binder SBA and the terminating in GalNAcα/β1-3/6Gal binder WFA, exhibited significantly altered NFIs in right hemisphere of HBV transgenic mice compared with control mice (Fig. 2f).
The glycopatterns of proteins in midbrain from HBV transgenic mice and control mice bound to the lectin microarrays, as shown in Fig. 2g. Their NFIs for each lectin are summarized in Additional file 1: Fig. S4c. The generated data from each mouse was imported into EXPANDER 6.0 to perform a hierarchical clustering analysis (Fig. 2h). The results showed the Fucα1-2Galβ1-4GlcNAc, Fucα1-3(Galβ1-4)GlcNAc, anti-H blood group specificity binder LTL, the Bisecting GlcNAc, biantennary complex-type N-glycan binder PHA-E, the High-Mannose, Manα1-3Man, Manα1-6Man, Man5-GlcNAc2-Asn binder HHL, the α-/β-linked terminal GalNAc, (GalNAc)n, GalNAcα1-3Gal, blood-group A binder SBA, the Terminal in GalNAc and Gal, anti-A and anti-B human blood group binder SJA, the terminating in GalNAcα/β1-3/6Gal binder WFA and the Galβ1-3GalNAc binder MPL, exhibited significantly altered NFIs in the midbrain of HBV transgenic mice compared with control mice (Fig. 2i).
The glycopatterns of proteins in cerebellum from HBV transgenic mice and control mice bound to the lectin microarrays and their NFIs for each lectin are shown in Additional file 1: Fig. S3d. The generated data from each mouse were imported into EXPANDER 6.0 to perform a hierarchical clustering analysis (Additional file 1: Fig. S3e). The results showed that the NFIs did not exhibit significantly in the midbrain of HBV transgenic mice compared with control mice.
Next, protein microarrays and lectin blotting analyses were performed using LEL selected randomly to confirm the different abundances of glycopatterns in pooled tissues from olfactory bulb. The results of protein microarrays showed that the glycopatterns recognized by LEL from HBV transgenic mice exhibited higher expression than control mice (Additional file 1: Fig. S5a–b). The results of SDS-PAGE demonstrated that the protein bands from HBV transgenic mice and control mice were similar (Additional file 1: Fig. S5c). The intensity of LEL bond to protein of olfactory bulb was consistent with NFIs trend of lectin microarrays in HBV transgenic mice and control mice (Additional file 1: Fig. S5d–e).
To further validate and assess the distribution and localization of specific glycans, including galactose type, fucose type (Fuc), N-acetylglucosamine (GlcNAc) and mannose type glycans in HBV transgenic mice and control mice, the fluorescence-based lectin histochemistry was performed using three lectins selected randomly (NPA, PSA and SBA), according to our previous protocol [20]. The negative controls showed no positive signal (Additional file 1: Fig. S6a), and the selected lectins showed various binding patterns (Additional file 1: Fig. S6b–g). The High-Mannose, Manα1-6Man recognized by NPA showed strong binding to the nuclear and cytoplasmic regions of granulosa cells in olfactory bulb (Additional file 1: Fig. S6b), while NPA mainly bond to the cytoplasmic regions in the left hemisphere (Additional file 1: Fig. S6c). The α-D-Man, Fucα-1,6GlcNAc, α-D-Glc recognized by PSA showed strong binding to the cytoplasmic and membrane areas in the olfactory bulb (Additional file 1: Fig. S6d) and left hemisphere (Additional file 1: Fig. S6e). Galactose type recognized by SBA showed strong binding to the cell membrane, and little binding to cytoplasmic regions in the left hemisphere (Additional file 1: Fig. S6f) and midbrain (Additional file 1: Fig. S6g).
These results demonstrated that the protein glycopatterns identified by 12 lectins (MPL, PSA, SJA, LCA, PHA-E, DSA, LEL, SBA, LTL, NPA, HHL and WFA) exhibited significantly changed NFIs in the brain of HBV transgenic mice compared with control mice. Notably, a total of 10 lectins (MPL, PSA, SJA, PHA-E, DSA, SBA, LTL, NPA, HHL and WFA) showed significant differences in the cerebral hemispheres and midbrain, where astrocytes were widely expressed. The sugar-binding specificities of the 10 altered lectins in HBV transgenic mice are shown in Additional file 1: Table S1.
Effects of NH4Cl on protein glycosylation in astrocytes
Ammonia is the most typical neurotoxin in the pathogenesis of HE, which could be synthesized into glutamine in astrocytes [21]. However, the relevance between ammonia and glycosylation changes in astrocytes remains unknown. We hypothesized that the abnormal glycosylation patterns observed in the brain of HBV transgenic mice could contribute to the ammonia concentration. To test this hypothesis, we utilized lectin microarrays to measure the effect of NH4Cl on protein glycosylation in astrocytes (SVG p12, SW1088 and CCF-STTG1). After serum-free for 24 h, astrocytes were left untreated for 72 h or treated with 0.5, 2, 5 or 10 mmol/L NH4Cl for 72 h. The medium was changed daily to ensure the concentration of NH4Cl. The NFIs for 37 lectins in astrocytes are summarized in Additional file 1: Fig. S7. The NFIs of 10 lectins, which altered in the brain of HBV transgenic mice, in the NH4Cl treated astrocytes and untreated astrocytes are shown in Fig. 3a. The preferred specificity of Maclura pomifera lectin (MPL) is Galβ1-3GalNAc structure, so MPL is called binder of Galβ1-3GalNAc structure. The binding amount of MPL with Galβ1-3GalNAc structure could reflected by the fluorescence intensity to indicate its expression level [19]. The results showed that the Galβ1-3GalNAc binder MPL exhibited significantly increased NFIs in NH4Cl treated compared with untreated SVG p12, SW1088 and CCF-STTG1 cells (Fig. 3a), which was consistent with the changing trend in the brain of HBV transgenic mice. This result suggested that ammonia affected the expression of Galβ1-3GalNAc.
Fluorescence-based lectin cytochemistry is an experimental technique similar to lectin microarrays, which could also reflect the binding amount of MPL with Galβ1-3GalNAc structure via the fluorescence intensity to indicate its expression level. Next, we validated the distribution of Galβ1-3GalNAc in NH4Cl treated astrocytes by fluorescence-based lectin cytochemistry. It was suggested that astrocytes were treated with 5 mmol/L NH4Cl, a concentration of ammonia was found in the brains of experimental hepatic failure [22]. The negative controls showed no positive signal (Additional file 1: Fig. S8a). Consistent with our observations in lectin microarrays, the expression levels of Galβ1-3GalNAc recognized by MPL increased significantly in 5 mmol/L NH4Cl treated in comparison with untreated SVG p12, SW1088 and CCF-STTG1 cells (Fig. 3b).
Consequently, to test for the reversibility of NH4Cl-induced Galβ1-3GalNAc, SVG p12 cells were left untreated or exposed to NH4Cl (5 mmol/L) for 72 h followed by another incubation for 72 h in an NH4Cl-free culture medium. The results showed that Galβ1-3GalNAc returned to basal level when NH4Cl containing medium was removed after 72 h (Additional file 1: Fig. S8b–c). Our finding provided evidence that the increase of Galβ1,3-GalNA related to the upgrade of ammonia concentration.
Effects of glutamine, nicotinamide adenine dinucleotide phosphate (NADPH) and pH changes on Galβ1-3GalNAc in astrocytes
Ammonia accumulated in the blood can cross the blood–brain barrier into astrocytes, metabolized into glutamine by glutamine synthetase [23]. Ammonia-induced glutamine accumulation in astrocytes is a HE marker in patients with liver cirrhosis. To test the role of NH4Cl (5 mmol/L, 72 h) induced glutamine formation on the expression of Galβ1-3GalNAc, astrocytes were treated with NH4Cl and the glutamine synthetase inhibitor (L-Methionine sulfoximine, MSO, 3 mmol/L). In liver cirrhosis, HE is accompanied by oxidative stress due to the activation of NADPH oxidase [24]. The synergistic relationship between oxidative stress and hyperammonemia is essential for brain edema in HE [25]. To test for the role of NH4Cl (5 mmol/L, 72 h) induced NADPH on Galβ1-3GalNAc, astrocytes were exposed to NH4Cl and NADPH oxidase inhibitor (apocynin, 0.3 mmol/L). To determine whether NH4Cl-induced intracellular pH changes trigger increased Galβ1-3GalNAc, astrocytes were treated with CH3NH3Cl (5 mmol/L) for 72 h, which is a compound that produces the same intracellular pH changes as NH4Cl [26, 27]. As shown in Fig. 3c, inhibition of glutamine synthetase and NADPH oxidase significantly prevented the NH4Cl mediated upgrade of Galβ1-3GalNAc, and exposure to CH3NH3Cl did not affect the expression of Galβ1-3GalNAc.
To further validate and assess the expression of Galβ1-3GalNAc, fluorescence-based lectin cytochemistry was performed. Consistent with our observations in lectin microarrays, inhibition of glutamine synthetase and NADPH oxidase completely blocked the upgrade levels of Galβ1-3GalNAc in NH4Cl treated SVG p12, SW1088 and CCF-STTG1 cells (Fig. 3b and d), while astrocytes were exposed to CH3NH3Cl did not affect Galβ1-3GalNAc (Additional file 1: Fig. S9). These findings suggest that ammonia up-regulated Galβ1-3GalNAc in astrocytes in a glutamine synthesis- and NADPH oxidase-dependent manner but was independent of ammonia-induced intracellular pH changes.
Ammonia stimulated C1GALT1 expression and intracellular calcium activity in astrocytes
Glycosyltransferases have known to be involved in the synthesis of glycan epitopes. Glycosyltransferases of C1GALT1 and it’ s molecular chaperone (C1GALT1C1) are responsible for the synthesis of Galβ1-3GalNAc [28]. To gain insight into the biosynthetic underpinnings of the up-regulated Galβ1-3GalNAc, the expression of C1GALT1 and C1GALT1C1 were assessed. Consistent with our observed glycan changes, C1GALT1 showed higher mRNA expression levels in NH4Cl treated astrocytes for 72 h than the untreated astrocytes, whereas the mRNA expression of C1GALT1C1 was not affected (Fig. 4a). Inhibition of glutamine synthetase completely blocked the upregulation of C1GALT1 mRNA expression in NH4Cl treated astrocytes; However, inhibition of NADPH oxidase could not block the upregulation of C1GALT1 mRNA expression in all astrocytes treated with NH4Cl (Fig. 4b). The protein expression levels of C1GALT1 shown by western blotting were consistent with their mRNA expression (Fig. 4c). Altered expression of C1GALT1 would explain the changes in glycan structures associated with ammonia-induced, which is in a glutamine synthesis-dependent manner.
Ca2+ signaling is an important regulator of many functions of astrocytes [29]. We doubted whether Ca2+ signaling might be subjected to ammonia regulation, so we used Fluo-4 AM to detect the Ca2+ in SVG p12 and SW1088 cells. The results demonstrated that after NH4Cl treatment, the intensity of Ca2+ was significantly stronger (Fig. 4d). Ca2+ homeostasis in astrocytes is regulated by key components of Ca2+, including mGluR5 and IP3R1 [16]. To determine whether the up-regulated Ca2+ was related to the expression of mGluR5 and IP3R1, their expression levels in astrocytes were examined by real-time PCR and western blotting. The mRNA expression of IP3R1 and mGluR5 was up-regulated in SVG p12 and SW1088 cells after NH4Cl was treated for 72 h (Fig. 4e). Only the protein expression level of IP3R1 was up-regulated in SVG p12 and SW1088 cells after NH4Cl treatment, while the protein expression level of mGluR5 was not up-regulated (Fig. 4e). These data strongly suggest that upregulation of ammonia regulated calcium homeostasis in astrocytes according to the expression of IP3R1.
C1GALT1 silencing decreased ammonia-induced calcium activity in astrocytes
To determine whether ammonia-induced C1GALT1 is involved in regulating Ca2+ activity, we silenced C1GALT1 in the SVG p12 and SW1088 cells. In brief, cell lines were stably transduced with two independents short siRNAs targeting C1GALT1 (siC1GALT1-1 or siC1GALT1-2) or a non-targeting control (siNTC). The transfection efficiency was determined by negative control FAM (Additional file 1: Fig. S10). As expected, C1GALT1 knockdown, confirmed by both real-time PCR and western blotting (Fig. 4f). Then, we tested whether C1GALT1 was required for the expression of IP3R1. We found that both siC1GALT1-1 and siC1GALT1-2, but not siNTC, were able to restrain the expression of IP3R1 (Fig. 4g).
The Ca2+ activity was significantly decreased by siC1GALT1-1 and siC1GALT1-2, but not siNTC, indicating that C1GALT1 may contribute to the Ca2+ activity in astrocyte (Fig. 5a). We next examined whether C1GALT1 knockdown could reverse ammonia-induced up-regulation of Ca2+ activity. We knocked down C1GALT1 and added NH4Cl in astrocytes simultaneously. The results showed that both siC1GALT1-1 and siC1GALT1-2, but not siNTC, could suppress ammonia-induced up-regulation of Ca2+ activity (Fig. 5b). Overall, ammonia-induced C1GALT1 regulated calcium activity.
C1GALT1 decreased calcium activity by regulating IP3R1 expression in astrocytes
The previous results showed that C1GALT1 affected the expression of IP3R1. To determine whether IP3R1 was also involved in the regulation of Ca2+, we took the most potent known agonists of IP3R1 (Adenophostin A, AdA) to analyze the roles of IP3R1in Ca2+ dysbiosis induced by C1GALT1. This result suggested that AdA significantly increased intracellular Ca2+ in astrocytes, and AdA plus NH4Cl induced a slight increase in Ca2+ compared with AdA alone (Fig. 5c). In another similarly designed experiment, we found that AdA restored the down-regulation of Ca2+ induced by siC1GALT1, and AdA plus NH4Cl induced a more significant recovery in Ca2+ compared with AdA alone (Fig. 5d). In summary, C1GALT1 controlled Ca2+ homeostasis by regulating IP3R1.
Calcium activity in the brain of HBV transgenic mice
Finally, in order to further explore whether C1GALT1, IP3R1 and related calcium expressions in the brain of HBV transgenic mice were consistent with the change trends in vitro experiments, we pooled brain tissue samples from six mice in each group to explore the calcium activity and the expression of related molecules in the brains of HBV transgenic mice. We noted that the protein expression levels of C1GALT1 and IP3R1 in the brains of 18 months old HBV transgenic mice were higher than those of control mice (Fig. 6a). In addition, Ca2+ concentration in the brains of 18 months old HBV transgenic mice were also higher than those of control mice (Fig. 6b). These results were consistent with what we observed in astrocytes.