Identification of the leukocyte cell-derived chemotaxin 2 as a direct target gene of β-catenin in the liver

Authors


Abstract

To clarify molecular mechanisms underlying liver carcinogenesis induced by aberrant activation of Wnt pathway, we isolated the target genes of β-catenin from mice exhibiting constitutive activated β-catenin in the liver. Adenovirus-mediated expression of oncogenic β-catenin was used to isolate early targets of β-catenin in the liver. Suppression subtractive hybridization was used to identify the leukocyte cell-derived chemotaxin 2 (LECT2) gene as a direct target of β-catenin. Northern blot and immunohistochemical analyses demonstrated that LECT2 expression is specifically induced in different mouse models that express activated β-catenin in the liver. LECT2 expression was not activated in livers in which hepatocyte proliferation was induced by a β-catenin–independent signal. We characterized by mutagenesis the LEF/TCF site, which is crucial for LECT2 activation by β-catenin. We further characterized the chemotactic property of LECT2 for human neutrophils. Finally, we have shown an up-regulation of LECT2 in human liver tumors that expressed aberrant activation of β-catenin signaling; these tumors constituted a subset of hepatocellular carcinomas (HCC) and most of the hepatoblastomas that were studied. In conclusion, our results show that LECT2, which encodes a protein with chemotactic properties for human neutrophils, is a direct target gene of Wnt/β-catenin signaling in the liver. Since HCC develops mainly in patients with chronic hepatitis or cirrhosis induced by viral or inflammatory factors, understanding the role of LECT2 in liver carcinogenesis is of interest and may lead to new therapeutic perspectives. (HEPATOLOGY 2004;40:167–176.)

Hepatocellular carcinoma (HCC), the major primary liver cancer, is becoming increasingly common worldwide.1 The prognosis for patients with HCC is rather poor. The molecular changes underlying HCC remain largely unknown despite the fact that major risk factors, such as chronic hepatitis B or C infection and exposure to hepatocarcinogens like aflatoxin B1, are well recognized. Several genetic changes have been implicated in at least 3 pathways of carcinogenesis, specifically, the p53, RB and Wnt/β-catenin signaling pathways.2 Deregulation of the Wnt pathway appears to be most frequent of these changes in human HCC; it occurs in about 30% to 40% of patients.2, 3 It also occurs in more than 90% of hepatoblastomas, which are rare embryonal liver tumors.4 Mutations affecting 2 partners of the Wnt pathway have been found in liver cancers. One is a mutation that activates the β-catenin gene. Such mutations occur mainly in hepatitis B-negative HCC5 and in more than 50% of hepatoblastomas.6, 7 The other is a mutation that inactivates the axin 1, and, less commonly, the axin 2 gene.5, 8, 9 Mutations that activate the Wnt pathway result in β-catenin accumulation in the nucleus. This process, in association with LEF/TCF transcription factors, modulates the transcription of target genes.10, 11 It is now clear that the genetic program triggered by activation of β-catenin signaling depends on the cellular context. The β-catenin target genes c-myc and cyclin D1 are well recognized12, 13; neither c-myc nor cyclin D1 were induced in the liver of transgenic mice that express an oncogenic form of β-catenin, although such mice exhibit hepatomegaly and marked hepatocellular proliferation.14 We used several mouse models in which β-catenin signaling in the liver is activated to identify liver-specific target genes of the Wnt pathway that may be implicated in the development of liver cancer. We have identified 3 components of the metabolic pathway of glutamine and have demonstrated that the glutamine synthetase (GS) gene, which is frequently overexpressed in HCC, is a target of β-catenin signaling.15

To search for early genes that are sensitive to deregulation of the Wnt/β-catenin pathway in the liver, we used mice infected with an adenovirus that encodes an oncogenic form of β-catenin. We have previously shown that activation of the Wnt pathway using this approach may be achieved as early as 48 hours postinfection.15 This report describes the identification of a new β-catenin target gene, leukocyte cell-derived chemotaxin 2 (LECT2), which is expressed in the liver.

Abbreviations

HCC, hepatocellular carcinoma; GS, glutamine synthetase; LECT2, leukocyte cell-derived chemotaxin 2; SSH, suppression subtractive hybridization; cDNA, complementary DNA; PMN, polymorphonuclear leukocytes; PCR, polymerase chain reaction; HBSS, Hanks' Balanced Salt Solution; ISH, in situ hybridization; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; fMLP, N-formyl-methionyl-leucyl-phenylalanine; RT-PCR, reverse-transcriptase polymerase chain reaction; HA-tagged, influenza hemagglutinin-tagged.

Materials and Methods

Animals.

All procedures involving animals reported in this paper were carried out in accordance with French government regulations (Services Vétérinaires de la Santé et de la Production Animale, Ministère de l'Agriculture). L-PK/c-myc,3 ΔN131β-catenin,14 and ASV16 transgenic mice have been previously described.

Adenoviral Gene Transfer.

The adenoviruses AdGFP, AdLacZ, and AdβcatS37A have been described previously.17 B6D2/F1 mice were injected intravenously with 5 × 109 plaque-forming units of AdGFP, AdLacZ, or AdβcatS37A, and were sacrificed 48 hours later.

Suppression Subtractive Hybridization (SSH).

We generated complementary DNA (cDNA) from 1 μg poly(A)+ RNA isolated from the livers of 3 mice injected intravenously with AdβcatS37A (tester) and 6 mice injected intravenously with AdLacZ or AdGFP (driver) using the SMART PCR cDNA Synthesis Kit (BD Biosciences, Paris, France). SSH was undertaken using the PCR-Select cDNA Subtraction Kit (BD Biosciences, Clontech, Palo Alto, CA) according to the manufacturer's protocol. The subtracted cDNA library was subcloned into T/A cloning vector pT-Adv (Clontech, Paris, France) and transformed to ElectroMAX DH10B cells (Invitrogen, Carlsbad, CA). The library was plated on LB-ampicillin plates and incubated at 37°C overnight. Individual clones of the library were plated on LB-ampicillin 96 plates and 2 replicas were made using Hybond N+ (Amersham Biosciences, UK) nylon membranes. For differential screening, replica filters were hybridized with 32P-labeled subtracted tester end driver probes. Blast search was used to analyze sequence homologies in the gene database.

Human Tumor Samples and RNA Sources.

All tumor samples were obtained from surgical liver resections. RNA samples were extracted from frozen liver sections. HCC RNA samples were obtained; 29 RNA samples were kindly provided by Dr. Marie-Annick Buendia (Institut Pasteur, France) and 22 RNA samples were extracted from tumor samples obtained at Cochin Hospital (Paris, France). These tumor samples were evaluated for the presence of mutations that activate the β-catenin gene as previously described.3 Liver specimens were obtained from patients with hepatoblastoma managed at the Bicêtre Hospital (Le Kremlin-Bicêtre, France). All of these patients had received preoperative chemotherapy. Samples were fixed in 10% neutral buffered formalin and were embedded in paraffin. Hepatoblastoma RNA samples were kindly provided by Dr Marie-Annick Buendia (Institut Pasteur, France). According to French law and ethical guidelines, no informed consent is required before analysis of RNA samples from specimens of resected tissue that would otherwise be discarded.

Preparation of Polymorphonuclear Leukocytes.

Heparinized human venous blood was obtained from healthy volunteers. Polymorphonuclear leukocytes (PMN) were isolated using a 2-step sedimentation. Whole heparinized (10 units/mL) blood on 2% Dextran T500 in saline was centrifuged, and the granulocyte-rich supernatant was then centrifuged on a Ficoll-Hypaque cushion (Eurobio, Paris, France), as previously described.18 Purified PMN (95%-97 %) were subjected to hypotonic lysis for 30 seconds, washed, and suspended in Hanks' Balanced Salt Solution (HBSS) containing 1.2 mmol calcium at pH 7.4.

Cell Transfection Studies.

Huh7 and HepG2 cells were maintained in DMEM containing 10% (vol/vol) calf serum. Transient transfections were undertaken when cells were 60% to 70% confluent in 12-well plates, using Lipofectamine Plus Reagent (Invitrogen). A TK-Renilla plasmid (10 ng) was included in each transfection as a reference for monitoring transfection efficiency. Cells were lysed 24 hours after transfection and the luciferase and Renilla activities were assayed using Dual Luciferase Reporter Assay (Promega, Madison, WI). All experiments were undertaken in duplicate and were repeated at least 3 times. The total amount of transfected DNA was kept constant by adding the empty expresssion vector pCAN. The ΔN89β-catenin-pCAN expression vector was kindly provided by P. Polakis (San Francisco, CA), and the expression plasmid encoding ΔNTcf4 was kindly provided by H. Clevers (Utrecht, The Netherlands).

Plasmid Constructs.

The sequences of the human and mouse LECT2 genes have already been described.19, 20 The human and mouse promoters were synthesized from human and mouse genomic DNA by polymerase chain reaction (PCR) cloning.19In situ hybridization (ISH) probes were prepared from full-length human LECT2 cDNA (gi: 4504976) cloned in the SmaI site of pSP72 (Promega). Riboprobes were synthesized from linearized BglII (sense) and XhoI (antisense) plasmid. GS riboprobes were generated as previously described.21

Northern Blotting.

Total RNA was extracted from frozen liver using the guanidium thiocyanate single-step procedure; an aliquot (15 μg) was electrophoresed through 1% agarose-6% formaldehyde gel. The RNA samples were transferred to Hybond N+ (Amersham Pharmacia) membranes and hybridized with the corresponding 32P-labeled probes.

Western Blotting.

Tissues were homogenized in Laemmli buffer (1:10 wt/vol). Total proteins (10 μg) were separated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and were transferred to a nitrocellulose membrane. LECT2 was detected using polyclonal anti-mouse LECT2.22 The signals were visualized using the chemiluminescence detection system.

Radioactive In Situ Hybridization.

The radioactive ISH detection of messenger RNA sequences has been extensively described previously.21 The LECT2 probes (both orientations) were double-labeled. Slides were exposed for 7 days for GS and 21 days for LECT2.

Chemotactic Assay.

N-formyl-methionyl-leucyl-phenylalanine (fMLP) was purchased from Sigma (St Louis, MO). Stock solutions were stored in DMSO at −20°C. Recombinant mouse and human LECT2 proteins were produced from CHO cells as previously described.22 Random and directed migration of PMN was studied using the Boyden chamber technique.23 Suspensions of 0.5 × 106 PMN in 100 μL HBSS containing 1% BSA were incubated in the upper compartment of a chamber separated from the lower compartment by a cellulose filter (3 μm diameter pores; Millipore, Bedford, MA). The lower compartment contained various concentrations of LECT2 or fMLP. In some experiments, the chemoattractant was placed in the PMN chamber to destroy the chemotactic gradient. Chambers were incubated at 37°C for 40 minutes in 95% humidified air; filters were treated with ethanol and stained with hemalum. Unstimulated and directed migrations of PMN were measured as the migration front (at least 10 PMN) under a microscope. Five fields were analyzed for each filter. Results, obtained from 3 experiments, are expressed in micrometers.

Immunohistochemistry.

Immunohistochemistry was undertaken using 5 μm sections of formalin-fixed, paraffin-embedded liver. Sections were incubated with specific antibodies for 1 hour. The primary antibodies used were polyclonal anti-mouse LECT2,20 monoclonal anti-GS (1/200; BD Transduction Laboratories, Lexington, KY), monoclonal anti–β-catenin (1/500; BD Transduction Laboratories) polyclonal anti-GFP (1/50; BD Biosciences), and polyclonal antihemagglutinin (1/400; Clontech: Roche, Basel, Switzerland). LECT2, GS, β-catenin, GFP, and hemagglutinin (HA) were visualized using an immunoperoxidase protocol (Vectastain ABC Kit; Vector, Burlingame, CA). The proliferation marker Ki67 was detected immunohistochemically as previously described.14

Real-Time Reverse-Transcriptase (RT)-PCR Analysis.

Two μg of total RNA was reverse transcribed in a final volume of 40 μL as previously described.14

PCR reactions were undertaken using the LightCycler Instrument (Roche Molecular Biochemicals) and the LightCycler FastStart DNA Master Sybr Green I Reagent Kit (Roche Molecular Biochemicals), according to the manufacturer's protocol. PCR was run in 10 μL total volume (2 μL of diluted cDNA corresponding to 5 ng of total RNA and 8 μL of reaction mix containing 1 μL of 10 μmol of each primer) for 45 cycles. For each tumor sample, 2 reactions, 1 for the target gene and 1 for the reference 18S ribosome RNA, were undertaken in separate capillaries.

Quantification was undertaken using the calibrator-normalized Relative Quantification Software. The relative target gene expression was normalized on the basis of its 18S ribosomal content and to a calibrator which was normal human liver. The calibrator was included in each run and its normalized gene expression was set at a value 1. For each tumoral liver sample, the relative gene expression was expressed as x-fold the realtive expression of the calibrator. Primer sequences were: human LECT2 gene lectF: 5′-GGCAAGTCTTCCAATGA-3′, LectR: 5′CACATGCGATTGTATGC-3′, human GS gene GSF: 5′AAGTGTGTGGAAGAGTTGCC-3′, GSR: 5′-TGC TCACCATGTCCATTATC-3′, 18S gene: 18SF: 5′- GTAACCCGTTGAACCCCATT-3′, 18SR: 5′-CCA TCCAATCGGTAGTAGCG.

Results

Identification of LECT2 Gene by SSH.

We used SSH to identify early target genes regulated by Wnt/β-catenin signaling. The livers of mice injected with an adenovirus that encodes for an oncogenic form of β-catenin (AdβcatS37A, HA-tagged) or with control adenoviruses that encode for LacZ or GFP (AdLacZ and AdGFP) were removed 48 hours postinjection and the RNA extracted. The resulting subtracted cDNA library was screened. We isolated four genes that are strongly expressed: the GS gene that we had previously isolated as a hepatic β-catenin target gene,15 CYP2E1, RNase A family 4, and the LECT2 gene (data not shown). We focused on the LECT2 gene, because our recent results of studies of transgenic mice that lack the LECT2 gene suggested that LECT2 may regulate the homeostasis of natural killer T cells in the liver and may be involved in the pathogenesis of hepatitis LECT2 expression was substantially increased in the livers of mice that had been injected with AdβcatS37A (Fig. 1A). The early expression of LECT2 in response to activated β-catenin was also confirmed in an independent experiment. Livers were obtained from mice that had been sacrificed after injection of AdβcatS37A or control AdGFP. The LECT2 signal began to increase 24 hours after injection and reached a peak at 48 hours; it then remained almost stable for up to 120 hours in AdβcatS37A-infected livers (Fig. 1B).

Figure 1.

Identification of LECT2 as a potential β-catenin target gene in infected liver of mice. (A) LECT2 gene expression in adenovirus-infected livers used for the SSH library. RNA samples from livers of mice used for the SSH library, infected with recombinant adenoviruses that encoded for GFP, Lac Z (Ad.Cont), or an activated β-catenin (Ad.βcatS37A) were evaluated by Northern blotting. Blots were standardized with ribosomal probe R45. (B) LECT2 gene expression in adenovirus-infected livers. Mice were injected with the indicated adenoviruses and sacrificed 12, 24, 48, or 120 hours after injection (2 mice for each time point). RNA from infected livers was evaluated by Northern blotting. (C) Distribution of LECT2 in adenovirus-infected livers. (a and d) Sections of liver infected with adenoviruses for 120 hours were assessed for distribution of LECT2. (b and e) transgene expression, and (c and f) proliferation by Ki-67 immunohistochemical staining. Transgene expression was assessed using (b) GFP immunohistochemical staining and (e) HA immunohistochemical staining that visualized HA-tagged β-catenin S37A. Activation of β-catenin signaling was revealed by cytosolic and nuclear HA immunohistochemical staining. (a, b, and c) Livers infected with adenoviruses encoding GFP (Ad.GFP) for 120 hours. (Magnification, ×200). (d, e, and f) Livers infected by AdβcatS37A for 120 hours.

The distribution of LECT2 protein in mouse liver has not been published previously. Immunohistochemical studies showed that LECT2 expression is restricted to the perivenous hepatocytes in normal mouse liver (Fig. 2C). However, LECT2 expression was found throughout the liver lobules of AdβcatS37A-infected mice; the distribution of oncogenic β-catenin, as revealed by HA immunohistochemistry, was similar. Activation of the Wnt pathway was revealed by HA cytosolic and nuclear staining (Fig. 1C). Thus, the targeted activation of the Wnt pathway in hepatocytes induces LECT2 expression in these cells. In contrast, no induction of LECT2 expression was observed in AdGFP-infected livers, even though the livers were infected to about the same extent, as judged by GFP and HA staining (Fig. 1C).

Figure 2.

LECT2 and GS gene expression in the livers of ΔN131β-catenin and PK/c-myc transgenic mice. (A) LECT2 gene expression in ΔN131β-catenin transgenic mice. RNA samples from the liver of wild type (WT) and ΔN131β-catenin transgenic (ΔN131) mice were evaluated by Northern blotting. Blots were standardized using ribosomal probe R45. (B) Western blot analysis of livers from ΔN131β-catenin transgenic mice. Total cell extracts were analyzed using SDS-PAGE and immunoblotted with anti-LECT2 antibody. (C) Sections of the livers of WT and ΔN131β-catenin (ΔN131) mice were stained with an anti-LECT2 antibody. (Magnification, ×200.) (D) Northern blot analysis of RNA from liver tumor tissue (T, T1, T2, T3) from PK/c-myc transgenic mice and nontumor liver (NT). The asterisk (*) marks tumors with activated β-catenin. (E) Immunohistochemical detection of β-catenin, LECT2, and GS in liver tumors from PK/c-myc mice. (Left) a representative tumor nodule (T) showing cytosolic and nuclear staining for β-catenin that was not present in nontumor tissue (NT). (Middle) immunohistochemical detection of LECT2 showing uniform staining for LECT2 in the tumor (T). The adjacent nontumoral tissue (NT) exhibits perivenous staining for LECT2. (Right) Section with uniform GS staining throughout the tumor (T) and perivenous GS staining in the adjacent nontumor tissue (NT). (Magnification, ×200.) This illustration is representative of 15 liver tumors analyzed.

Up-regulation of LECT2 Expression Is Linked to an Activation of β-Catenin Signaling in the Liver.

Northern and Western blot analyses demonstrated that LECT2 gene expression is more strongly increased in the liver of ΔN131β-catenin transgenic mice than in the liver of a nontransgenic mouse (wild type; Figs. 2A and B). As expected, LECT2 expression was found throughout the liver lobule where oncogenic β-catenin was present (Fig. 2C).

We also studied L-PK/c-myc mice with liver tumors: more than 50% of the tumors had activating mutations in the β-catenin gene.3 Our earlier studies of such mice showed that activation of β-catenin signaling in liver tumors was associated with the overexpression of the gene that encodes for GS.15 Accordingly, we assayed LECT2 and GS gene expressions in isolated hepatic tumor nodules—with or without activated β-catenin—and compared their expression to that in adjacent nontumor tissue. The expression of LECT2 gene was up-regulated in the tumors associated with activated β-catenin and GS expression. The liver tumors that were negative for GS were also negative for LECT2 (Fig. 2D). Immunohistochemistry revealed that expression of LECT2 was restricted to the perivenous region of nontumor tissue. Both LECT2 and GS were widely distributed throughout tumor tissue in which there was activation of β-catenin signaling, as indicated by positive immunohistochemical staining for β-catenin in the cytosol and, occasionally, in the nucleus (Fig. 2E).

To confirm that the up-regulation of LECT2 expression was associated with activation of β-catenin signaling in the liver, we analyzed the expression of LECT2 in liver tumors that developed in ASV transgenic mice following the expression of SV40 early sequences.16 The Wnt pathway is not activated in the tumors of such mice24. No up-regulation of LECT2 was observed in liver tumors of ASV transgenic mice, although these tumors were highly proliferative (Figs. 3B, C, and D). Absence of induction of LECT2 expression in this model was confirmed by Northern blot analysis (data not shown). We also found no increased LECT2 expression in AdGFP-infected mice, although there was substantial hepatocellular proliferation in this model (Fig. 1C).

Figure 3.

Distribution of LECT2 and GS in liver tumors from SV40 large T-antigen transgenic mice. Immunohistochemical detection of Ki67, LECT2, and GS in liver tumor of SV40 large T-antigen transgenic mice (ASV). (A) Histology, hematoxylin-eosin staining. (B) Ki67 staining showing hepatocyte proliferation in tumor tissue (T) and no staining in the adjacent nontumor tissue (NT). (C) Immunohistochemical detection of LECT2. There is perivenous LECT2 staining in the nontumor tissue (NT), no staining for LECT2 in the tumor (T). (D) Immunohistochemical detection of GS. GS staining was similar to that for LECT2: it was restricted to the perivenous region of the nontumoral tissue. (Magnification, ×200.) This illustration is representative of the results obtained from 3 different ASV mice.

LECT2 Is a Downstream Target of the β-Catenin/TCF-4 Signaling Pathway.

The identification of several LEF/TCF binding sequences in the mouse and human LECT2 promoters (Fig. 4A) led us to determine whether the LECT2 promoter was a direct transcriptional target for activation by β-catenin. To this end, we used Huh7 and HepG2 hepatoma cell lines which differ in their β-catenin status; Huh7 and HepG2 are wild type and activated mutant, respectively.3

Figure 4.

The mouse and human LECT2 promoters are activated by β-catenin. (A) Diagram showing the mouse LECT2 promoter (−2076LECT2) and the human LECT2 promoter (−2351LECT2). Putative LEF/TCF binding sites are located from the ATG site; the proximal LEF/TCF binding site matched to the consensus LEF/TCF site. It is the same in both the mouse and human LECT2 promoter. (B) Huh7 cells were transfected with 0.1 μg of mouse LECT2 promoter (−2076LECT2-LUC) together with increasing concentrations of activated β-catenin: 0, 0.5, 1, 2, and 3 μg of ΔN89β-catenin–encoding plasmid (ΔN89β-cat). Results are presented as x-fold induction relative to transfection of the reporter by an empty expression vector. The bars represent the mean values of 3 experiments, each undertaken in duplicate. Error bars represent SEMs. *P < .05. (C) Huh7 cells were transfected with 1 μg of human promoter (−2351hLECT2) together with 0, 0.5, 1, 2, and 3 μg of ΔN89β-cat. Results are presented as x-fold induction relative to transfection of the reporter by an empty expression vector. The bars represent the mean values of 3 experiments, each undertaken in duplicate. Error bars represent SEMs. *P < .05.

The induction of both mouse and human LECT2 promoter activities by β-catenin was demonstrated by transient cotransfection studies in Huh7 cells using promoter-luciferase reporter constructs containing about 2 kbp of 5′-flanking regulatory region of the mouse and human LECT2 genes and an oncogenic ββ-catenin construct (ΔN89β-catenin). The result was a significant dose-dependent increase in the mouse and human promoter activities in response to activated β-catenin. The mouse promoter was activated up to 10-fold (Figs. 4B and C). The effect was strongly inhibited by cotransfection with a dominant-negative TCF-4 expression construct (ΔNTCF4) that prevented transactivation through the LEF/TCF recognition motifs (Fig. 5A). The cotransfection of the ΔNTCF4 expression vector in HepG2 cells, which express an endogenous activated β-catenin, reduced substanially the activity of the LECT2 promoter induced by β-catenin (Fig. 5B). Thus, β-catenin/TCF transactivation controls the activity of the LECT2 promoter. Since the coactivators CBPand p300 have been shown to activate β-catenin/TCF transcription of some Wnt-responsive genes,25 they could cooperate with β-catenin to activate LECT2 promoter. Transfection with CBP/p300 more than doubled the LECT2 promoter activity induced by β-catenin, whereas the CBP/p300 expression construct alone had little effect on the LECT2 promoter in the absence of β-catenin (Fig. 5C).

Figure 5.

Control of LECT2 promoter activity by β-catenin: involvement of LEF/TCF sites and the CBP/p300 coactivator. (A)Huh7 cells were transfected with 0.1 μg of mouse LECT2 promoter and 1 μg of β-catenin or empty vector, and 1 μg of ΔNTCF4 expression vector or empty vector. Data are expressed as normalized luciferase activity relative to Renilla control. (B) HepG2 cells were transfected with 0.1 μg of mouse LECT2 promoter and 1 μg of ΔNTCF4 or empty expression vector. Data are expressed as normalized luciferase activity relative to Renilla control. (C) LECT2 promoter transactivation by the CBP/p300 coactivator. Huh7 cells were transfected with 0.1 μg of mouse LECT2 promoter and 0.5 μg of ΔN89β-cat or empty vector, plus 1.5 μg of CBP/p300 or empty vector. Results are presented as x-fold induction relative to transfection of the reporter by empty expression vector. Bars represent the means of 3 independent duplicates, each undertaken in duplicates. Error bars represent the SD. *P < .05.

Analysis of the mouse promoter shows that it contains 3 potential LEF/TCF binding sequences. To identify the sequences in the LECT2 promoter that can confer transactivation by β-catenin, 2 deletion constructs were generated (Fig. 6A). Analysis of these deletion constructs revealed a progressive decrease of the transactivation induced by β-catenin (Fig. 6B). The smallest construct, which contained only the proximal LEF/TCF site, still mediated a 2- to 3-fold induction of LECT2 promoter activity. This finding indicates that the proximal site could be important in cooperation with distal sites for the transactivation of mouse LECT2 promoter by β-catenin. The proximal LEF/TCF site matched perfectly to the consensus sequence and is conserved in the human LECT2 promoter (Fig. 4A). Thus, this site appears to be a good candidate to support the transcriptional activity of LECT2 promoter by β-catenin. We tested this hypothesis by introducing point mutations in the LEF/TCF proximal site in the full-length LECT2 promoter (−2076mtLECT2-LUC; Fig. 6C). Mutation of the proximal LEF/TCF binding site completely abolished transactivation of the LECT2 promoter by β-catenin (Fig. 6D).

Figure 6.

Role of the proximal LEF/TCF binding sequence of the LECT2 promoter in β-catenin–mediated transactivation (A) Deletion constructs of the mouse LECT2 promoter (−579LECT2-LUC and −299LECT2-LUC) in the pGL3 luciferase reporter plasmid. (B) The deletion constructs of the mouse LECT2 promoter (0.1 μg) were transfected into Huh7 cells as shown in Fig. 5C. (C) The mutated mouse LECT2 reporter construct showing the AT to GC change at nucleotides −209 and −208 (−2076mtLECT2-LUC). D) Huh7 cells were transfected with the wild type or mutated mouse LECT2 reporter construct (0.1 μg) as described in Fig. 5C. Transfections were carried out in triplicate, each undertaken in duplicate. Data are means ± SD. *P < .05.

LECT2 is therefore a direct transcriptional target of β-catenin.

Chemotactic Properties of LECT2.

The LECT2 protein was initially purified from the culture fluid of phytohemagglutinin-activated human T-cell leukemia SKW-3 cells as a possible chemotactic factor for human neutrophils.26

To further characterize the chemotactic property of LECT2, we used a recombinant LECT2 protein produced from CHO cells22 and tested its chemotactic properties using human PMN. The chemotactic activity of LECT2 was compared to that of the tripeptide fMLP, which is a potent neutrophil chemoattractant.18, 27 LECT2 induced a bell-shaped migration pattern as a function of its concentration; this finding is characteristic of a chemotactic ligand (Fig. 7). Significant directed migration was observed in the presence of 10 to 25 nM LECT2, whereas at a high concentration (50 nmol) LECT2 was inactive, presumably because of chemotactic desensitization of PMN. fMLP induced more directed migration over a wider active concentration range than LECT2. To determine whether the stimulated migration induced by LECT2 may be mediated by chemokinetic activity of PMN, we abolished the chemotactic gradient by titrating LECT2 or fMLP into the cell compartment of the Boyden chamber. Under these conditions, the directed PMN migration was completely prevented (data not shown). This finding confirms that LECT2 is a chemoattractant for human neutrophils.

Figure 7.

PMN chemotactic activity of LECT2 and fMLP. The migration of PMN was measured in the presence or absence of various concentrations of (A) LECT2 or (B) fMLP placed in the lower compartment of the Boyden chamber. Results represent the migration front expressed in micrometers (mean ± SEM of 5 individual fields for each concentration, from a representative experiment). The significance of differences between data obtained from 3 independent experiments with random and stimulated migrations were determined using Student's t test. *P < .05.

Expression of the LECT2 Gene in Human Liver Tumors.

To investigate if LECT2 could be implicated in human liver cancer, we analyzed LECT2 gene expression using real-time RT-PCR of liver tumors in which the Wnt pathway is frequently altered. We first quantified the relative degree of RNA expression of LECT2 and GS in 18 human HCC samples, in which there were activated mutations of the β-catenin gene. As expected, all these tumors exhibited a high degree of GS expression.15 In contrast, LECT2 expression was up-regulated in only 5 of these tumors (Fig. 8). Analysis of 33 HCC samples that did not display β-catenin mutations indicated up-regulation of LECT2 in only 6 samples. Interestingly, all of the 6 LECT2-positive HCC samples were also positive for GS, suggesting that these samples may have mutations of another partner of the Wnt pathway (Fig. 8). Altogether, these results indicated that even though LECT2 was generally down-regulated in human HCC samples, LECT2 was up-regulated only in a subset of human HCC, a subset in which the Wnt pathway was altered. We also analyzed hepatoblastoma samples; deregulation of the Wnt pathway occurs in more than 90% of these tumors4 We analyszed LECT2 expression in 14 samples using real-time RT-PCR. A high level of LECT2 expression was observed in 13 samples. Overexpression of GS was substantial in all of the samples tested (Fig. 8). ISH, used to localize LECT2 expression, confirmed that LECT2 was up-regulated in hepatoblastomas. LECT2 was detected in the tumor areas that exhibited activation of β-catenin signaling associated with GS expression (Fig. 9).

Figure 8.

Up-regulation of LECT2 in human liver tumors Histograms on a logarithmic scale showing the relative expression of LECT2 and GS genes measured in (A) a panel of hepatocellular carcinomas (HCC) with β-catenin mutations and a panel of HCC that have no activating mutations of β-catenin, and (B) in a panel of hepatoblastomas. Relative gene expression was determined using real-time RT-PCR, as described in Materials and Methods, and was expressed as fold difference relative to normal liver, after normalization of each measurement to 18S ribosomal RNA.

Figure 9.

Expression of LECT2, GS, and β-catenin in hepatoblastoma by in situ hybridization. Representative in situ hybridization of [35S]-labeled probes for (A) GS antisense, (C) LECT2 antisense, and (D)LECT2 sense in serial sections of a single hepatoblastoma. This illustration is representative of 11 cases analyzed. (Magnification, ×40.) (B) Immunohistochemical staining of β-catenin. Note the tumor area displayed activated β-catenin signaling, as revealed by cytosolic and nuclear accumulation of β-catenin, that parallels the signals of LECT2 and GS. (Magnification, ×40.) (Insert) β-catenin cytosolic and nuclear accumulation. (Magnification, ×400.)

Discussion

The Wnt pathway is frequently deregulated in carcinogenesis. Considerable effort is currently being made to identify the downstream β-catenin target genes; those that have been identified may be important for understanding the role of Wnt signaling in carcinogenesis because they are involved in cell proliferation, survival, differentiation, and migration.10, 11 However, it has become clear that the Wnt response is tissue-specific. Expression of activated β-catenin is sufficient to induce polyposis in the intestine28, 29 but is not sufficient to induce hepatocarcinogenesis in the liver. We and others have observed substantial hepatomegaly in response to activation of β-catenin signaling in the liver, but other genetic events are likely to be required for hepatic tumorigenesis.14, 30 Therefore, it is important to identify the target genes of β-catenin in the liver. We used an adenovirus-mediated expression of activated β-catenin to isolate early target genes. We identified the LECT2 gene as a direct target of β-catenin signaling in the liver.

We found that LECT2 promoter activity is increased considerably by activation of β-catenin, and is blocked by a dominant-negative form of TCF-4. We have also identified the LEF/TCF site as being crucial for a response to β-catenin. Our studies in mice with activated β-catenin showed that β-catenin signaling strongly induced LECT2 expression in hepatocytes. As LECT2 is mainly expressed in the liver, we assessed whether LECT2 could be a liver-specific response of the Wnt pathway. We analyzed LECT2 expression in another tumor in which activation of the Wnt pathway is frequent: colon cancer. We observed no induction of LECT2 expression in intestinal epithelial cells of polyps that develop in a mouse model of colon cancer that we recently developed (Colnot et al., unpublished manuscript). Such mice have a new germ-line mutation of the Apc gene and a high level of activation of β-catenin signaling, as indicated by cytosolic and nuclear staining of β-catenin in all epithelial cells of their polyps. So far, few β-catenin target genes have been isolated from liver.15, 31 LECT2 can be induced by β-catenin in the liver but not in the intestine, and it may be an important target relating to the tissue-specific response of the Wnt pathway in hepatic tissue.

Interestingly, LECT2 gene expression is confined to a well-defined pericentral compartment, consisting of 1 to 2 layers of hepatocytes in the normal mouse liver; this expression is similar to that of 3 other hepatic β-catenin-induced genes linked to glutamine metabolism that we identified previously.15 The zonation of the liver lobule is determined during development of the liver; the molecular mechanisms involved are not known. From our results, it is tempting to speculate that the Wnt pathway in the liver may participate in the establishment of zonation.

The relevance of LECT2 in human liver carcinogenesis has been addressed by examining the level of expression of LECT2 in human liver tumors. We found substantial induction of LECT2 in most of the hepatoblastomas studied; in this tumor, activation of the Wnt pathway is frequent.4 In contrast, an up-regulation of LECT2 was found only in a subset of HCCs associated with activation of the Wnt pathway. However, LECT2 expression was generally decreased in human HCCs in accordance with previously published findings.32, 33 Hepatoblastoma and HCC are 2 primary liver tumors that differ with respect to their histological manifestations and genetic changes. HCC is a heterogeneous tumor associated with multiple genetic changes; only a limited number of chromosomal changes have been found in hepatoblastomas.4 The loss of LECT2 expression in HCCs that have aberrant β-catenin signaling could result from “cross-talking” with one or more other signaling pathways induced by different genetic changes that may occurr during tumor progression. At present, the role of LECT2 in liver carcinogenesis is unknown. LECT2 was first isolated as a possible chemotactic factor for neutrophils.26 Using recombinant protein, we have further characterized the chemotactic property of LECT2 for human neutrophils in vitro. We searched for leukocyte infiltration in tumors that overexpressed LECT2, such as those associated with L-PK/c-myc, but we did not find any evidence of an inflammatory process in such tumors. Two hypotheses may be proposed: (1) leukocyte infiltration may be a transient event that cannot be observed in the tumor, and (2) the role of LECT2 in liver carcinogenesis may be linked to another function of LECT2. The study of transgenic mice lacking the LECT2 gene revealed that the role of LECT2 is related to the homeostasis of NK T cells in the liver; LECT2 may modulate the inflammatory and immune response induced by the development of the tumor. Since HCC develops most commonly in patients with chronic hepatitis or cirrhosis induced by viral or inflammatory factors, understanding the role of LECT2 in liver carcinogenesis is of interest and may lead to new therapeutic perspectives.

Acknowledgements

We thank Professor Axel Kahn for strong support and for critically reading the manuscript, and all the members of our team for helpful discussion. We also thank M. A. Buendia (Institut Pasteur, France) for providing HCC and hepatoblastoma RNA samples, Professor Benoit Terris (Hôpital Cochin, Service d'Anatomopathologie) for providing frozen HCC samples, J. Kitajewski (Columbia University, NY) for providing the recombinant adenoviruses, F. Letourneur and N. Lebrun (Institut Cochin) for sequencing, and Dr. Owen Parkes for editing the manuscript.

Ancillary