Wnt/β-Catenin Signaling Regulates the Expression of the Ammonium Permease Gene RHBG in Human Cancer Cells

Ammonium is a metabolic waste product mainly detoxified by the liver. Hepatic dysfunction can lead to cytotoxic accumulation of circulating ammonium and to subsequent encephalopathy. Transmembrane ammonium transport is a widely spread process ensured by the highly conserved proteins of the Mep-Amt-Rh superfamily, including the mammalian Rhesus (Rh) factors. The regulatory mechanisms involved in the control of RH genes expression remain poorly studied. Here we addressed the expression regulation of one of these factors, RHBG. We identify HepG2 hepatocellular carcinoma cells and SW480 colon adenocarcinoma cells as expressing RHBG and show that its expression relies on β-catenin signaling. siRNA-mediated β-catenin knockdown resulted in significant reduction of RHBG mRNA in both cell lines. Pharmaceutical inhibition of the TCF4/β-catenin interaction or knockdown of the transcription factor TCF4 also downregulated RHBG expression. We identify a minimal RHBG regulatory sequence displaying a promoter activity and show that β-catenin and TCF4 bind to this fragment in vivo. We finally characterize the role of potential TCF4 binding sites in RHBG regulation. Taken together, our results indicate RHBG expression as a direct target of β-catenin regulation, a pathway frequently deregulated in many cancers and associated with tumorigenesis.


Introduction
Ammonium, hereafter referring to the sum of the NH 3 and NH 4 + molecular species, serves as principal nitrogen source for micro-organisms and plants [1]. It is however mostly described for the cytotoxic consequences of its accumulation in animals [2]. Hepatic metabolism of ammonium towards urea and glutamine synthesis is critical to maintain a low plasmatic level of the ammonium emerging from the catabolism of proteins and the activity of the intestinal flora. The impairment of ammonium detoxification occurring in case of liver dysfunction can lead to the development of hepatic encephalopathy and, in acute cases, to lethal cerebral paralysis. In parallel to these toxic effects, renal ammonium production from glutaminolysis and its subsequent urinary excretion is a crucial process to ensure blood pH homeostasis [3]. The view incubator with humidified air (5% CO2) at 37°C. PFK118-310 (#K4394) was purchased from Sigma and used at a 0,2 or 0,4μM concentration from a DMSO solution.

Plasmids construction
DNA fragment (2492 bp) corresponding to the potential RHBG promoter was amplified using polymerase chain reaction with the F-Pr-BG-and R-Pr-BG (Table 1) primers and the human genomic DNA of HEK293T cells as a template. The PCR product was digested with SacI and BglII restriction enzymes and cloned into pGL3-Basic (Promega) that has been linearized with the same restriction enzymes. Deletion mutants were then constructed using the pGL3-RHBG plasmid as a template and the indicated primers ( Table 1). The PCR products were digested with SacI and BglII restriction enzymes and cloned into pGL3-Basic. All constructs were verified by sequencing.

RNA extraction and qRT-PCR
Total cellular RNA was extracted using TRIzol reagent (Invitrogen) according to manufacturer instructions. DNase treatment was done using a DNA Removal Kit (Invitrogen, #AM1906). One μg of total RNA was reverse-transcribed to cDNA using the SuperScriptIII First-Strand Synthesis SuperMix (Invitrogen) according to manufacturer instructions. Realtime RT-PCR were performed on a StepOnePlus Real-Time PCR System (Applied Biosystems) using GoTaq qPCR Master Mix (Promega) using the indicated primers (Table 2) and normalized to β-actin mRNA level measured in parallel.

Transfection and luciferase assay
Cells were transiently co-transfected with the pTK-Renilla luciferase reporter vector (Promega) and empty plasmid (pGL3-Basic, Promega) or the plasmid containing the indicated RHBG promoter constructs for 48h using the Viafect transfection reagent (Promega). Luciferase activity in total cell lysates was measured using the Dual-Glo luciferase reporter assay (Promega)

Chromatin immunoprecipitation (CHIP) assay
CHIP assay was done using SimpleChIP Enzymatic Chromatin IP Kit (Magnetic Beads, Cell signaling) according to manufacturer instructions. Briefly, cells were fixed with 1% formaldehyde solution to cross-link histone and non-histone proteins to DNA. Nuclear chromatin was digested with Micrococcal Nuclease for 20 min at 37°C and then incubated overnight at 4°C with either anti-β-catenin (Cell signaling, #8480, 1:50), anti-TCF4 (#2569, 1:50) Normal Rabbit IgG (Cell signaling, #2729). Following washing with low and high salt ChIP buffers, the protein-DNA complexes were eluted and cross-links were then reversed. After proteinase K digestion, DNA is purified and quantified by Real time-PCR as described earlier using primers listed in Table 2 and designed to amplify the indicated promoter regions of the target genes.

Statistical analysis
Data are expressed as means ± S.E.M. Statistical comparisons were assessed by Student's t-tests using Graph Pad Prism version 5.00 software (Graph Pad Software). Differences were considered significant when the p value is below 0.05 ( Ã p < 0.05, ÃÃ p < 0.01, ÃÃÃ p < 0.001), n = 3, except for Chip experiments where n = 2.

RHBG is highly expressed in HepG2 hepatoma cells
The human RHBG gene was found to be overexpressed in a subset of hepatocellular carcinoma [18], however, the regulatory mechanisms involved remain unknown. In order to identify a cancer cell line that could be used to address the regulation of RHBG, we evaluated its expression levels in the HepG2 and Hep3B hepatoma cells. HepG2 cells bear a heterozygous deletion in the exon3 of the β-catenin CTNNB-1 gene, producing a truncated β-catenin protein lacking key residues for its phosphorylation by the destruction complex and thus resulting in its cellular accumulation [26][27][28]. Hep3B cells are derived from a hepatitis B-infected liver tumor, and do not contain mutations in the β-catenin gene [29,30]. The RHBG gene appeared highly expressed in HepG2 compared to Hep3B cells, the latter showing slightly higher expression level compared to that of HEK293T cells taken as a normal cell model ( Fig 1A). Expression level of GLUL, the gene of glutamine synthetase (GS) was also upregulated in HepG2 cells compared to Hep3B ( Fig 1B). As the GS gene is a reported target of β-catenin [31,32], this could be consistent with a more effective β-catenin signaling in HepG2 cells compared to Hep3B. The GLUL expression levels were corroborated by the resulting GS protein levels which appeared higher in HepG2 compared to Hep3B cells ( Fig 1C). Upon checking the expression of the second non-erythroid RH gene, RHCG, very low levels were found in HepG2, Hep3B and HEK293T cells ( Fig 1D). These data thus point to HepG2 cells as suitable for the study of RHBG expression.
We next checked the β-catenin protein levels in HepG2, Hep3B and HEK293T cells. Consistent with previous reports [33,34], HepG2 cells harbored two β-catenin species likely corresponding to the wild-type and the truncated forms ( Fig 1E). A β-catenin protein with the size of the wild-type form was detected in both HEK293T and Hep3B, with no obvious accumulation in the latter cell line. It was previously shown that β-catenin is present in the nucleus of HepG2 cells in contrast to Hep3B cells [34]. These data suggest a correlation between RHBG expression in HepG2 cells and nuclear localization of β-catenin.

Silencing of β-catenin correlates with RHBG down-regulation in HepG2 cells
We next used β-catenin siRNA to test whether the RHBG gene expression observed in HepG2 cells is related to β-catenin function. Transfection of HepG2 cells with β-catenin siRNA for 72 hours led to a decrease in β-catenin mRNA level compared to cells transfected with non-targeting siRNA (Fig 2A). The reduction of β-catenin mRNA resulted in a reduction of β-catenin protein levels (Fig 2B). Of note, the RHBG gene expression was largely reduced upon β-catenin silencing (Fig 2C). Similarly, expression levels of Axin2 and Cyclin D1, two targets of β-catenin regulation [35][36][37][38], were also decreased with β-catenin silencing (Fig 2D and 2E). In contrast, the low expression level of RHCG observed in HepG2 cells was not affected by β-catenin β-Catenin-Dependent Expression of the Ammonium Permease Gene RHBG silencing ( Fig 2F). Hence, inhibition of β-catenin is accompanied by the down-regulation of RHBG gene expression in HepG2 cells.

Beta-catenin drives RHBG expression in SW480 colon cancer cells
Mutations inducing β-catenin activation have been identified in various types of tumors, including melanoma, prostate, breast, and colon cancers [21,33,39,40]. We next tested whether the β-catenin signaling could be correlated with RHBG expression in another cancer cell line harboring β-catenin signaling activating mutations. The SW480 colon cancer cells bear a truncating mutation in the APC gene, resulting in stabilization and nuclear accumulation of βcatenin and leading to constitutive activation of β-catenin signaling [41][42][43][44]. Consistently, immunofluorescence experiments revealed a strong nuclear localization of β-catenin in these cells (Fig 3A). Transfection of SW480 cells with β-catenin siRNA for 72 hours led to a decrease in both mRNA and protein levels of β-catenin (Fig 3B and 3C). Similarly to HepG2 cells, βcatenin silencing was accompanied by a large decrease in RHBG expression in SW480 cells  Fig 3D). The expression levels of Axin2, and CylcinD1 (Fig 3E and 3F) were also decreased upon β-catenin silencing, consistent with β-catenin signaling inhibition. RHCG expression was not significantly affected upon β-catenin silencing (Fig 3G).
These results indicate that the correlation between β-catenin signaling and RHBG expression can be extended to SW480 colon cancer cells.

Activation of β-catenin correlates with RHBG gene induction in HEK293T cells
We next addressed whether activation of β-catenin in a cell line with no major nuclear activity of β-catenin could be sufficient to induce RHBG expression. LiCl is reported to inhibit GSK3 kinase [45,46], leading to β-catenin stabilization and nuclear accumulation [47,48]. We therefore treated HEK293T cells with LiCl (10 and 20 mM) to activate the Wnt/β-catenin pathway. In keeping with previous observations, immunofluorescence and western blot experiments revealed that treatment of HEK293T cells with LiCl for 24 hours induced a nuclear accumulation and stabilization of β-catenin (Fig 4A and 4B). Of note, LiCl treatment concomitantly induced an increase in the mRNA level of RHBG in a dose-dependent manner (Fig 4C). This result indicates that RHBG expression can be upregulated by artificial activation of β-catenin signaling

RHBG expression in HepG2 cells is dependent on TCF4
The Wnt/β-catenin pathway drives Wnt-specific transcriptional programs via the interaction with DNA-binding factors of the TCF/LEF family [21,22]. However, it is reported that β-catenin can also activate gene expression in a TCF4-independent manner [49][50][51]. To address a role of the TCF4/β-catenin complex in RHBG expression, we evaluated the effect of inhibiting the β-catenin activity in HepG2 cells by using an antagonist of the TCF4/β-catenin complex, PKF118-310. This compound disrupts the TCF4/ β-catenin complex and inhibits expression of TCF4-dependent genes [52]. Treatment of HepG2 cells with PKF118-310 was accompanied by a decrease in RHBG expression (Fig 5A). The GLUL expression level was also decreased by the treatment, consistent with a likely reduction of TCF4/β-catenin mediated transcription in these conditions (Fig 5B).
To further assert a role of TCF4 in RHBG expression, we tested the impact of TCF4 knockdown in HepG2 cells. Transfection of the latter cells with TCF4 siRNA for 72 hours decreased both the mRNA and protein levels of TCF4 compared to cells transfected with non-targeting siRNA (Fig 5C and 5D). Importantly, the RHBG mRNA level was reduced upon TCF4 silencing (Fig 5E) Similarly, the Axin2 and Cyclin D1 mRNA levels were also decreased (Fig 5F and  5G), in keeping with previous observations describing the corresponding genes as targets of TCF4 [38,40]. Moreover, similar TCF4 silencing experiments performed in SW480 colon adenocarcinoma cells also decreased RHBG, Axin2 and Cyclin D1 mRNA levels (Fig 6A-6E).  These results together indicate that β-catenin-mediated expression of RHBG is at least partially TCF4-dependent in both HepG2 and SW480 cells.
The RHBG promoter is activated by β-catenin/TCF4 To further study RHBG expression and identify potential regulators, a genomic fragment (Fig 7) containing 2349 bp upstream and 142 bp downstream of the RHBG predicted transcriptional start site (TSS) was directionally subcloned into the pGL3-basic firefly luciferase reporter vector. To test whether this fragment possesses a promoter activity, the RHBG promoter construct (pGL3-RHBG) and the native pGL3-basic vector were used for transient co-transfection  of HepG2 cells together with the pTK-Renilla luciferase reporter vector as transfection control. 48 hours after transfection, the luciferase activity in pGL3-RHBG transfected cells was about 30 fold higher than with the pGL3-basic plasmid indicating that the cloned RHBG sequence contains an active promoter (Fig 8A).
Sequence analysis of the RHBG regulatory sequence did not reveal a potential TATA box, while the region proximal to the predicted TSS was enriched in G/C content, indicating that the RHBG promoter likely corresponds to a TATA-less GC-rich promoter. Two potential GC boxes were depicted, embedded in potential Sp-1 binding sites (Fig 7). To further dissect the regions important for the activity of the cloned RHBG promoter, a series of constructs were generated bearing progressive deletions in this DNA fragment (Figs 7 and 8).
Though fluctuations were noted according to the considered fragment, all the constructs containing the -60/+142 region produced a luciferase activity very close or higher than the HepG2 cells were transfected with the empty plasmid (pGL3) or RHBG promoter (pGL3-RHBG) together with Renilla plasmid. 48 hours after transfection, RHBG promoter activity in total cell lysates was determined by luciferase assay. B) HepG2 cells were transfected with RHBG promoter (pGL3-RHBG) or the indicated construct together with Renilla plasmid. 48 hours after transfection, RHBG promoter activity was determined by measuring luciferase activity in total cell lysates. Data are expressed as mean of triplicate determinations ± S.E.M of the pGL3-RHBG construct relative to pGL3-Basic. C) HepG2 cells were transfected with the indicated construct together with Renilla plasmid. 48 hours after transfection, RHBG promoter activity was determined by measuring luciferase activity in total cell lysates.
doi:10.1371/journal.pone.0128683.g008 β-Catenin-Dependent Expression of the Ammonium Permease Gene RHBG full-length pGL3-RHBG construct (Fig 8B). Of note the -60/+142 region of fragment H, comprising only one of both GC boxes, retained high luciferase activity. The constructs bearing further 5' truncation into this region led to a major decrease of the RHBG promoter function, the -22/+142 region showing a very low luciferase activity (Fig 8B). This underlines the importance of the DNA segment between fragment H and I, and indicates that the expression impairment is most likely due to the loss of the second GC box. Additionally, analysis of the -60/+142 functional segment revealed the presence of three CTTTG/CAAAG motifs which could serve as TCF4 binding sites (Fig 7). These motifs are either juxtaposed to or downstream of the putative TSS. To evaluate a potential contribution of these motifs to the regulation of RHBG gene expression, promoter constructs bearing the -60/+142 region of fragment H with deletion of potential TCF4 binding motif 2, or motifs 2 and 3, were generated. Both constructs showed a promoter activity (Fig 8C). However, deletion of motif 2 reduced the promoter activity to the half of fragment H, and simultaneous deletion of motifs 2 and 3 further decreased the promoter activity, suggesting a contribution of these motifs to the functionality of the RHBG promoter.
We finally performed chromatin immunoprecipitation (ChIP) assays to determine whether TCF4 and β-catenin are capable of binding the -60/+142 segment of RHBG promoter in vivo. Nuclear extracts obtained from the HepG2 cells were subjected to protein/DNA complex crosslinking and immunoprecipitation was performed using antibodies targeting either β-catenin, TCF4 or IgG, as a control. qPCR using primers within the H region reveal that TCF4 and βcatenin bind to this fragment of the RHBG promoter (Fig 9). Consistently, TCF4 and β-catenin did also bind to the Axin2 promoter, taken as control, as previously reported [37,53].
These results indicate that TCF4/β-catenin specifically binds to the -60/+142 region of the RHBG promoter, and likely enhances RHBG expression in HepG2 cells.

Discussion
This study identifies the hepatoma HepG2 cells as expressing the RHBG gene encoding an ammonium transport protein. We show that in these cells RHBG expression is largely dependent on β-catenin function. We perform a functional analysis of the RHBG upstream regulatory sequence, revealing a minimal region bearing a promoter activity. Our data indicate that the RHBG regulatory sequence is a TATA-less GC-rich promoter. We show that β-catenin and TCF4 are both able to bind the minimal promoter region in vivo and characterize potential TCF4 binding motifs important for the promoter activity. Our data support a direct role of βcatenin/TCF4 in the regulation of RHBG expression in this cell line, and further indicate that RHBG could serve as a direct reporter of the Wnt/β-catenin pathway in specific cancer cell contexts.
Hepatocellular carcinoma is the most common adult liver malignancy and many lines of evidence associate hyperactivation of the Wnt/β-catenin pathway to its initiation and development [56]. Abnormal activation of Wnt/β-catenin signaling, due to loss-of-function mutations in APC or activating mutations in β-catenin has been linked to various human malignancies including melanoma, breast, and colon carcinomas [22]. For instance, more than 80% of colon cancers bear truncations in APC, resulting in active β-catenin accumulation in the nucleus, the initial stage of transformation [56][57][58]. We show that RHBG is expressed in the colon cancer SW480 cells bearing an APC mutation and that its expression is also dependent on TCF4/ β-catenin. In contrast, the expression of the gene encoding the second non erythroid ammonium transport protein, RhCG, is independent of β-catenin signaling in either HepG2 or SW480 cells.
Our data are consistent with the RHBG overexpression observed in hepatocarcinoma obtained from surgical resections and bearing activating mutations in the β-catenin gene CTNNB-1 [18]. RHBG was up-regulated in 9 over 10 HCC of the latter resections compared to normal liver, while it was slightly overexpressed in only one over 15 HCC showing wildtype CTNNB-1. RHBG overexpression was strongly correlated with upregulation of GLUL, but also of SLC13-A3 encoding a sodium-dicarboxylate (including α-ketoglutarate and succinate) transporter and GPR49, also known as LGR5, a coreceptor of Wnt signaling. A correlation between HCC with activated β-catenin pathway and upregulation of GS, of ornithin amino transferase (involved in glutamate synthesis), and of the Glt1 glutamate transporter, were also reported [31]. The upregulation of GS suggests that HepG2 cells, and possibly specific HCC, could be able to adapt their metabolism to favor glutamine synthesis from glutamate and ammonium, a function restricted to perivenous hepatocytes in normal liver. For instance, in mouse, the Wnt/β-catenin pathway has been shown to play a key role in liver zonation [19,20]. This process ensures a functional specialization of hepatocytes along the porto-central axis of the liver lobule and determines the fate of periportal hepatocytes, active in urea synthesis, or perivenous hepatocytes, active in glutamine synthesis for instance. TCF4/β-catenin binds to the RHBG promoter. HepG2 cells were cross-linked with formaldehyde followed by chromatin digestion. Chromatin immunoprecipitations were performed using antibodies targeting either β-catenin, TCF4 or IgG, as a control. Purified DNA was analyzed by qPCR using the indicated primers. The amount of immunoprecipitated DNA with each antibody is represented as signal relative to IgG (equivalent to 1) (n = 2). doi:10.1371/journal.pone.0128683.g009 β-Catenin-Dependent Expression of the Ammonium Permease Gene RHBG Mouse Rhbg is specifically present at the cell surface of the latter hepatocytes and co-localizes with GS [59]. HepG2 cells were shown to have a reduced activity of the urea cycle [60]. However, it should be kept in mind that these cells show important plasticity of the metabolic networks according to the availability of key metabolite in the surrounding medium as glucose and insulin [61]. Of note, it was recently shown that HepG2 cells have a glutamine-addiction phenotype [34]. Addiction to glutamine is a metabolic particularity of many cancer cells showing concomitant high rates of glutamine transport and metabolism [62]. Proliferation of HepG2 was importantly reduced upon withdrawal of exogenous glutamine, and simultaneous drug-mediated inhibition of GS activity further hampered proliferation [34]. In conditions where glutamine synthesis would be favourable, it is tempting to hypothesize that correlated upregulation of RhBG could help to scavenge ammonium, providing one of the substrates of GS. Whether RhBG actively participates to cancer cell metabolism will require further investigation. Rh factors were shown to act as bidirectional ammonium transport proteins [8,63]. Rhcg is expressed at the apical membrane of specific epithelial kidney cells, together with the H + V-ATPase which is supposed to drive NH 3 efflux by favouring urinary trapping of NH 4 + [12,13,17]. The GS activity could serve as a trapping system, driving ammonium influx via co-expressed Rh factors such as RhBG by consuming ammonium for glutamine generation. Interestingly, a corresponding mechanism exists in E. coli where the GS activity is strictly required to drive substrate uptake via the Rh orthologue AmtB [64]. However, to date, a role of Rhbg in the process of ammonium detoxification via glutamine synthesis has not been highlighted in vivo, as plasma levels of glutamine and urea appear normal in mice lacking Rhbg [15]. GS is the sole enzyme catalyzing glutamine synthesis. In addition to its presence in perivenous hepatocytes, a detailed analysis in mouse revealed that it is also highly expressed and active in the epididymis epithelial cells and in Leydig cells, the testosterone-producing cells in the testis [65]. Although the physiological role of GS in these cells is unknown, it should be noted that Rh factors are also co-expressed [17,66].
Ammonium was recently proposed to play a particular role in a tumoral context [67]. Upregulated glutaminolysis in glutamine-addicted cancer cells results in NH 3 production. The latter molecule was shown to act as an autocrine and paracrine diffusible signal that triggers a specific autophagic program, in turn enabling survival and proliferation of cancer cells deep in a tumour mass [67,68]. Whether Rh factors could play a role in these processes by participating to trans-cellular ammonium movements remains to be evaluated.