High glucose augments angiotensinogen in human renal proximal tubular cells through hepatocyte nuclear factor-5

High glucose has been demonstrated to induce angiotensinogen (AGT) synthesis in the renal proximal tubular cells (RPTCs) of rats, which may further activate the intrarenal renin-angiotensin system (RAS) and contribute to diabetic nephropathy. This study aimed to investigate the effects of high glucose on AGT in the RPTCs of human origin and identify the glucose-responsive transcriptional factor(s) that bind(s) to the DNA sequences of AGT promoter in human RPTCs. Human kidney (HK)-2 cells were treated with normal glucose (5.5 mM) and high glucose (15.0 mM), respectively. Levels of AGT mRNA and AGT secretion of HK-2 cells were measured by real-time polymerase chain reaction (PCR) and enzyme-linked immunosorbent assay (ELISA), respectively. Consecutive 5’-end deletion mutant constructs and different site-directed mutagenesis products of human AGT promoter sequences were respectively transfected into HK-2 cells, followed by AGT promoter activity measurement through dual luciferase assay. High glucose significantly augmented the levels of AGT mRNA and AGT secretion of HK-2 cells, compared with normal glucose treatment. High glucose also significantly augmented AGT promoter activity in HK-2 cells transfected with the constructs of human AGT promoter sequences, compared with normal glucose treatment. Hepatocyte nuclear factor (HNF)-5 was found to be one of the glucose-responsive transcriptional factors of AGT in human RPTCs, since the mutation of its binding sites within AGT promoter sequences abolished the above effects of high glucose on AGT promoter activity as well as levels of AGT mRNA and its secretion. The present study has demonstrated, for the first time, that high glucose augments AGT in human RPTCs through HNF-5, which provides a potential therapeutic target for diabetic nephropathy.


ELISA
Secreted AGT in the culture medium of HK-2 cells was measured using the human angiotensinogen assay kit (Immuno-Biological Laboratories, Gunma, Japan, catalog#27412) according to the protocol provided by the manufacturer as well as the previous description [15]. AGT levels were normalized by total cellular protein in the dish.

Plasmid constructs
Seven consecutive 5'-end deletion mutant plasmid constructs of the region from −4358 to +122 (relative to the transcription start site of human AGT gene) around human AGT promoter sequences (AGT_−4,358/+122), which has been proved sufficient for the promoter activity in human RPTCs [12], were used as described in a previous study [12]. Mutations in binding sites of possible candidates of glucose-responsive transcriptional factor(s) were generated from intact human AGT_−4,358/+122 with the specific primer sets (Table 1) and the QuickChange site-directed mutagenesis kit (Agilent Technologies, California, USA, cata-log#200519-5) according to manufacturer's instructions. Whole sequences of all plasmid constructs were analyzed using the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems, catalog#4336917), BigDye Terminator purification kit (Applied Biosystems, cata-log#4376484) and genetic analyzer (Applied Biosystems, catalog#3130x1) based on the manuals provided by the manufacturers. Accuracy of the DNA sequences was confirmed as expected. The specific primers used for sequencing are listed in Table 1.

Plasmid transfection
HK-2 cells were initially seeded into 24-well plates. After cell confluence reached to about 60%, cells were serum-starved for 24 h before transfection. For plasmid transfection, cells in each well were transfected with 250 ng of plasmid construct and 20 ng pRL-TK renilla luciferase reporter vector plasmid (Promega, Wisconsin, USA, catalog#E2241), as an internal control

Dual luciferase assay
To measure human AGT promoter activity, dual luciferase assay was performed as described previously [16] using the Dual-luciferase reporter assay system kit (Promega, catalog#E1910) and microplate reader (Corona Electric, Wisconsin, USA, catalog#SH-9000), following the manuals from manufacturers. Data were normalized based on the value shown by the pRL-TK renilla luciferase activity.

Chromatin immunoprecipitation
To measure the binding level of a transcriptional factor to human AGT promoter sequences, chromatin immunoprecipitation was performed, as described previously [17], using the EZ-CHIP chromatin immunoprecipitation kit (Merck Millipore, Darmstadt, Germany, cata-log#17-371), and either HNF-5 (also called HNF-3 [18,19]) monoclonal antibody (Santa Cruz, California, USA, catalog#sc-377033) or mouse IgG (Santa Cruz, catalog#sc-516176) as the negative control, according to the manufacturers' instructions. Precipitated DNA was analyzed by PCR using the specific primer sets listed in Table 1.

Statistical analysis
Data are presented as mean ± SEM. Unpaired t test was used to compare values between two groups and one-way ANOVA followed by the Tukey's test was used to compare values among more than two groups. P-value <0.05 was considered statistically significant. All statistical analyses were performed using Prism 5 software (GraphPad, California, USA).

Results
High glucose augmented levels of AGT mRNA and AGT secretion of HK-2 cells  1B). In the additional experiment, we also found that 8 mM, the marginal level of postprandial blood glucose in healthy subjects, is the threshold glucose concentration that can stimulate increased AGT (S1A Fig). To exclude the influence of osmotic stress, mannitol was used to equalize the total osmotic stress, which could be induced by both glucose and mannitol, in each group. As expected, mannitol treatment for 48 h did not influence AGT mRNA levels in HK-2 cells ( Fig  1C). Therefore, the effect of high glucose on AGT mRNA level was not due to osmotic stress. Taken together, these data suggested that high glucose directly augmented AGT mRNA levels of HK-2 cells in a time-and dose-dependent manner. Previous in vivo studies conducted in our lab have shown that the augmentation of intrarenal AGT was inhibited by treatment with an Ang II receptor blocker (ARB) in diabetes [4,8]. Additional experiments were performed to examine the effect of an ARB (valsartan) on high glucose-induced augmentation of AGT in HK2 cells. We used valsartan at a concentration of 10 μM, as previously described [20,21]. However, high glucose-induced augmentation of AGT was not affected by treatment with valsartan (S1B Fig). We further investigated the effects of high glucose on AGT secretion into the cell culture medium. Compared with normal glucose, high glucose (15.0 mM) treatment for 48 h significantly augmented secreted AGT levels (40.81±2.69 vs. 67.35±2.49 ng/mg total protein; P<0.0001), however, secretion of AGT was not affected by mannitol-induced osmotic stress (Fig 2).
Possible glucose-responsive transcriptional factor candidates GENETYX software was used to screen all the corresponding transcriptional factors of the glucose-responsive region (-22 to -1,896) in human AGT promoter sequences. Among the transcriptional factors that might bind to the glucose-responsive region (score ≧20) (Tables 2-4), three ones were found to be both glucose-responsive and closely associated with human AGT regulation as described below. The hepatocyte nuclear factor (HNF) family including HNF-5 (site position: -634 to -639; score >50, Table 2) is extensively involved in AGT regulation and/ or high glucose effects [22,23]. A previous study reported that glucosamine can stimulate AGT mRNA expression by activating cAMP-responsive element-binding protein (CREB) (site position: -837 to -841; score>40, Table 3) in rat RPTCs [24]. Additionally, Feng et al. [25] reported that AGT mRNA expression was significantly higher in the hearts of diabetic rats compared with healthy rats, which is in association with increased myocyte enhancer factor (MEF)2 (site position: -32 to -41; score>20, Table 4). Based on above information, we determined HNF-5, CREB and MEF2 as glucose-responsive transcriptional factor candidates of the AGT promoter in human RPTCs.
Mutation in HNF-5 binding sites reduced the effects of high glucose on human AGT promoter activity as well as AGT mRNA and its secretion levels Constructs with respective mutations of HNF-5 (Fig 4A), CREB (Fig 4B) or MEF2 (Fig 4C) binding sites in human AGT promoter sequences (AGT_-4,358/+122) were prepared. Equal amounts of empty vector, vector harboring either intact human AGT_-4,358/+122 or binding sites mutated human AGT_-4,358/+122 were respectively transfected into HK-2 cells, followed by AGT promoter activity measurement. Compared with normal glucose (5.5 mM), high glucose (15 mM) treatment for 48 h significantly augmented AGT promoter activity of HK-2 cells with intact human AGT_-4,358/+122 transfection (77±1 vs. 109±11, relative ratios to empty vector transfection group under the same glucose concentration; P<0.05) (Fig 4D). However, similar effect was not observed in HK-2 cells with mutated HNF-5 binding sites (66±17 vs. 56 ±14, relative ratios to empty vector transfection group under the same glucose concentration; P = 0.68) (Fig 4D). Different from the mutation of HNF-5 binding sites, mutations in the binding sites of CREB (58±8 vs. 106±6, relative ratios to empty vector transfection group under the same glucose concentration; P<0.01) and MEF-2 (28±3 vs. 73±17, relative ratios to empty vector transfection group under the same glucose concentration; P<0.05) had no effect on the response of AGT promoter activity to high glucose treatment (Fig 4E and 4F). Next, we investigated whether mutation in HNF-5 binding sites could attenuate AGT mRNA expression and its secretion under high glucose condition. In general, compared with groups transfected with the empty vector, those transfected with intact human AGT_-4,358/+122 showed much higher AGT mRNA and protein levels, especially under the high glucose condition (Fig 4G and 4H). These results suggest that the plasmid DNA could be integrated into the genome of HK-2 cells after transfection, moreover, the integrated promoter sequences even occupied a considerable proportion and were able to control AGT mRNA and protein expression together with the native AGT promoter sequences in HK-2 cells. Compared with normal glucose (5.5 mM), high glucose (15 mM) treatment for 48 h significantly augmented AGT mRNA levels of HK-2 cells transfected with either empty vector (1±0.01 vs. 2.74±0.27, relative ratios to the empty vector transfection group with normal glucose; P<0.001) or intact human AGT_-4,358/+122 transfection (1.93±0.13 vs. 5.02±0.35, relative ratios to the empty vector transfection group with normal glucose; P<0.001) (Fig 4G). However, this effect was not observed in HK-2 cells transfected with HNF-5 binding sites  mg total protein; P<0.0001) (Fig 4H). However, these effects were not observed in HK-2 cells transfected with HNF-5 binding sites mutated AGT_-4,358/+122 (59.12±5.49 vs. 43.72 ±5.65 ng/mg total protein; P = 0.12) (Fig 4H).

Binding level of HNF-5 to AGT promoter sequences was reduced by mutation in its binding sites and elevated by high glucose treatment
Equal amounts of empty vector, vector harboring intact human AGT_-4,358/+122 and vector harboring HNF-5 binding sites mutated human AGT_-4,358/+122 was transfected into HK-2 cells, respectively. Then the binding levels of HNF-5 protein to human AGT promoter sequences were detected by chromatin immunoprecipitation. As shown in at least 3 independent experiments, compared with the intact human AGT_-4,358/+122 transfection groups, HK-2 cells with HNF-5 binding sites mutation showed significantly attenuated HNF-5 binding levels. This was detected through both band intensities after PCR ( Fig 5A) and quantitative real-time PCR (1402±287 vs. 240±23, relative ratios to the empty vector transfection group; P<0.05) (Fig 5B). These data confirm that mutation in HNF-5 binding sites reduced the binding level of HNF-5 to AGT promoter sequences. Additionally, we also measured the binding level of HNF-5 to the native genome of HK-2 cells without plasmid transfection. HNF-5 binding level was very low under normal glucose

Discussion
Three novel findings were uncovered in the present study. First, high glucose augments the levels of AGT mRNA and AGT secretion of human RPTCs. Second, high glucose augments AGT promoter activity in human RPTCs and the glucose-responsive region within human AGT promoter sequences is from -22 to -1,896. Third, HNF-5 is one of the glucose-responsive HNF-5 and high glucose-induced augmentation of human proximal tubular angiotensinogen transcriptional factors which bind to the glucose-responsive region of AGT promoter sequences and mutation of its binding sites may abolish the high glucose effects on AGT in human RPTCs Intrarenal AGT is primarily generated in RPTCs and secreted into tubular fluid [10,26]. Visavadiya et al. [27] have shown that high glucose can increase the expression levels of the RAS components including renin, angiotensin-converting enzyme as well as Ang II levels in HK-2 cells. In the present study, we aim to determine the detailed molecular mechanism responsible for high glucose-induced proximal tubular augmentation of AGT levels. We found that high glucose significantly augmented AGT mRNA levels in human RPTCs with the most obvious effects by 48 h-treatment with 15 or 20 mM glucose, which is similar to the blood glucose level of diabetic patients. Consistent findings were previously discovered in the case of rats, it is reported that high glucose stimulated AGT gene expression in rat RPTCs, which subsequently resulted in cell hypertrophy [13,14]; On the other hand, stable transfection of antisense rat AGT cDNA into rat RPTCs effectively inhibited AGT expression and prevented high glucose-induced cell hypertrophy [28]. However, high glucose-induced augmentation of AGT in HK-2 cells was not affected by treatment with valsartan. We have no good explanation as to why ARB does not inhibit high glucose-induced augmentation of AGT in vitro, however, it is probably due to the low renin activity and AGT utilizing efficiency in HK-2 cells.
Additionally, we also found that high glucose significantly augmented AGT secretion from human RPTCs. AGT secreted from proximal tubules is regarded as the main source of urinary AGT [29]. Moreover, urinary AGT level reflects the activity of intrarenal RAS [30,31] and tubular injury [32] in patients with type 2 diabetes. Elevation of urinary AGT has also been confirmed to be related to not only development but also progression of DN based on studies in patients with DN [31,33], as well as mouse [34,35] and rat [6,36] models of diabetes. Therefore, finding in the present study suggests that high glucose-induced AGT secretion from RPTCs may play an important role in the pathogenesis of DN.
Previous studies have demonstrated that proximal promoter elements of human AGT genes from −4358 to +122 are sufficient to measure AGT promoter activity in human RPTCs [12,37]. Furthermore, the importance of proximal regulatory elements to human AGT regulation has been confirmed by analyses with deletion mutants of AGT promoter sequences [12,37]. Based on the previous study [12], we also employed the region from -4358 to +122 within human AGT promoter sequences and further identified its glucose-responsive region in human RPTCs. We found that high glucose treatment significantly augmented AGT promoter activity in HK-2 cells respectively transfected with the deletion mutant constructs with 5'-ends from -344 to -4,358. These data indicated that high glucose activates AGT by increasing AGT promoter activity in human RPTCs. Furthermore, within the high glucose treatment groups, consecutive increases of AGT promoter activity were observed in HK-2 cells harboring constructs with 5'-ends from -344 to -1,896, while followed by the marked decreases of AGT promoter activity in HK-2 cells harboring constructs with 5'-ends from -2,414 to -4,358. These data indicated that the glucose-responsive region, where glucose-responsive transcriptional factor(s) bind(s), should be from -22 to -1,896 of human AGT promoter sequences.
Among all the transcriptional factors that possibly bind to the glucose-responsive region of human AGT promoter sequences (score ≧ 20), regulatory effect on AGT promoter activity in human RPTCs was observed in HNF-5. Soon after HNF-5 was initially designated [38], Nitsch et al. [18] demonstrated that HNF-5 is identical to HNF-3, because the HNF-3 binding sites within gene promoter sequences, identified by them, coincides with those of HNF-5. In the following literatures, both names were kept and applied to refer to the same factor [19]. HK-2 cells transfected with the HNF-5 binding sites mutated human AGT_-4,358/+122 did not show any increase in AGT promoter activity as well as AGT mRNA and its secretion levels under high glucose treatment. In other words, mutation of HNF-5 binding sites in human AGT promoter sequences abolished the high glucose effects on AGT in human RPTCs. Moreover, we also confirmed that binding level of HNF-5 to AGT promoter sequences was reduced by mutation in its binding sites and elevated by high glucose treatment. All of the above results uncovered the molecular mechanism that in human RPTCs, through HNF-5, high glucose enhances the activity of the AGT gene promoter, thus increasing the levels of transcription and secretion of AGT. As a result, the levels of AGT mRNA and AGT secretion were augmented (Fig 6). It has been well known that the HNF family is extensively involved in human AGT regulation, for example, HNF-1α [39] and HNF-4 [23] can bind to the AGT promoter sequences and activate AGT expression in human hepatoma cells. In the present study, we demonstrated, for the first time, that HNF-5 is one of the glucose-responsive transcriptional factors that bind to the AGT promoter sequences in human RPTCs. As far as we know, HNF-5 also plays a role as a glucocorticoid-responsive transcriptional factor in hepatoma cells [40]. However, in hepatoma cells, glucose-induced AGT augmentation is mediated predominantly through HNF-1α-and HNF-4-dependent pathways [23,39]. Further studies are needed to determine the possible role of HNF-5 in AGT production in other organs.
As shown in Fig 4E and 4F, mutation of CREB or MEF2 binding sites within AGT promoter sequences did not abolish the high glucose-induced enhancement of AGT promoter activity in HK-2 cells. These data suggest that CREB and MEF2 does not contribute to augmenting AGT expression in RPTCs under high glucose condition. Using GENETYX software and published data, candidates for glucose-responsive transcriptional factors were selected based on their likelihood of binding to human AGT promoter sequences. There may be other glucose-responsive transcriptional factors that work on AGT promoter, but we omitted them due to the lack of published supporting evidence for their potential involvement in high glucose-induced effects and/or AGT regulation. HNF-5 is located from -634 to -639. However, as shown in Fig 3B, AGT promoter activity apparently stepped-up from AGT_-1,351/+122 to AGT -1,896/+122, suggesting that the sequences between -1,351 and -1,896 may play an important role in amplifying AGT induction under high glucose condition. Therefore, other glucose-responsive transcriptional factors that bind to this region should be investigated further.
In conclusion, the present study has demonstrated, for the first time, that high glucose augments AGT in human RPTCs through HNF-5, a glucose-responsive transcriptional factor that functions on the AGT gene promoter, which provides a potential therapeutic target for DN in humans.    Nishiyama.