Leukocyte cell‐derived chemotaxin 2 antagonizes MET receptor activation to suppress hepatocellular carcinoma vascular invasion by protein tyrosine phosphatase 1B recruitment

Leukocyte cell‐derived chemotoxin 2 (LECT2) has been shown to act as a tumor suppressor in hepatocellular carcinoma (HCC). However, the underlying mechanism has not yet been completely defined. Here, we employ a LECT2‐affinity column plus liquid chromatography coupled with tandem mass spectrometry to identify LECT2‐binding proteins and found that MET receptor strongly interacted with LECT2 protein. Despite the presence of hepatocyte growth factor, the LECT2 binding causes an antagonistic effect to MET receptor activation through recruitment of protein tyrosine phosphatase 1B. The antagonistic effect of LECT2 on MET activation also mainly contributes to the blockage of vascular invasion and metastasis of HCC. Furthermore, serial deletions and mutations of LECT2 showed that the HxGxD motif is primarily responsible for MET receptor binding and its antagonistic effects. Conclusion: These findings reveal a novel, specific inhibitory function of LECT2 in HCC by the direct binding and inactivation of MET, opening a potential avenue for treating MET‐related liver cancer. (Hepatology 2014;59:974‐985)

largely unsuccessful as a result of a high frequency of tumor recurrence and metastasis. Studies of the molecular pathophysiology of HCC revealed that growth factors and their corresponding receptors are commonly overexpressed and/or dysregulated in HCC. Such receptor axes include hepatocyte growth factor (HGF)/MET, insulin growth factor (IGF)/ IGF-1 receptor (IGF-1R), vascular endothelial growth factor (VEGF)/VEGF receptor (VEGFR), and ErbB family receptor tyrosine kinases. 1 Activation of these receptors and their corresponding downstream signaling cascades can lead to angiogenesis, cell proliferation, and metastasis in HCC. Hence, identifying novel therapeutic targets and effective treatments, particularly against receptor tyrosine kinase pathways, is urgent for this fatal disease.
Leukocyte cell-derived chemotaxin 2 (LECT2) was originally identified as a chemotactic factor for neutrophils and stimulates the growth of chondrocytes and osteoblasts. 2,3 The subsequent isolation of LECT2coding complementary DNA suggested that it is predominantly expressed in the liver. 4 In a previous study, LECT2 was identified as a direct target gene of bcatenin in the liver. 5 b-catenin-induced LECT2 expression suppresses tumor progression through its inflammation-suppressive effects. This balancing force of LECT2 to nuclear b-catenin was proposed to affect tumorigenesis of HCC. 6 However, the LECT2 expression did not correlate with the status of hepatitis B and C virus infection or cirrhosis, 7 suggesting the tumor-suppressive effects of LECT2 may be beyond inflammation. Furthermore, whether LECT2 is a clinically downstream molecule of b-catenin remains illusive, because LECT2 expression did not correlate with CTTNB mutations. 5,8 Recently, LECT2 expression was also found to suppress HCC cell invasion in vitro and negatively correlated with vascular invasion in HCC patients. 7 Although accumulating evidence supports LECT2 as an important tumor suppressor in HCC, the membrane-binding receptor as well as downstream-acting mechanisms of LECT2 on HCC progressions remain largely unclear.
In this study, we have identified LECT2 as a novel endogenous MET antagonist that reduces phosphorylation by direct interaction with the MET receptor and recruitment of protein tyrosine phosphatase 1B (PTP-1B).

Materials and Methods
Cell Culture, Transfection, and Established Stable Clone. Hepatoma cell lines (i.e., SK-Hep1, HepG2, and Huh7) were grown in Dulbecco's modified Eagle's medium medium containing 10% fetal bovine serum (Life Technologies, Inc., Grand Island, NY) at 37 C in a humidified atmosphere of 5% CO 2 /95% air. SK-Hep1, HepG2, and 293T cells were from American Type Culture Collection (Manassas, VA). Huh7 cells were from Joint Conference of Restoration Branches. For establishment of stable cell lines, both empty and pSecTag2A-LECT2-hygromycin B or pLKO-shLECT2-puromycine and shLuciferase (purchased from the National RNAi Core Facility) vectors were transfected into hepatoma cells using Lipofectin reagent (Invitrogen Life Technology, Carlsbad, CA) or infected with lentivirus. Target sequences for shLECT2 were CTATTGCCCTTGCAGAAAGTT (shLECT2-1) and GCATACAATCGCATGTGC-ACA (shLECT2-2). Stable cell populations were selected, and single clones were confirmed to have prominent LECT2 expression by western blotting analysis.
Western Blotting and Immunoprecipitation. Western blotting and immunoprecipition (IP) analysis were performed, as previously described, 9 by indicating antibodies (Abs; listed in the Supporting Materials). For western blotting, the indicated cells were harvested with radioimmunoprecipitation assay buffer (Merck Millipore, Taiwan) containing protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO) and phosphatase inhibitor cocktail (Merck Millipore). Protein lysates were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene difluoride membrane and immunoblotted with the indicated Ab. Blottings were developed with enhanced chemiluminescence western blotting reagents (Merck Millipore). For IP, cells were lysed on ice with IP lysis buffer (50 mM of Tris [pH 7.4], 150 mM of NaCl, 1% Triton X-100, and 1% Nonidet P-40) and immunoprecipitated with the indicated Abs combined with protein A agarose beads (Sigma-Aldrich). The collected protein complex was washed four times with IP buffer and eluted by boiling with protein sample buffer under reducing conditions. Then, proteins were resolved on SDS-PAGE and analyzed by western blotting.
Full additional materials and methods are described in the Supporting Information.

Results
MET Is a Direct Binding Target of LECT2. To identify receptor(s) that are targeted by the LECT2 protein in regulating HCC progression, we utilized a LECT2-Fc immobilized affinity column to isolate candidate receptors from the SK-Hep1 cell membrane proteins and analyzed by nanoLC-MS/MS (liquid chromatography coupled with tandem mass spectrometry). Multiple proteins were detected and assumed to be bound by LECT2, including receptor tyrosine kinases (RTKs) and their associated factors (Supporting Table 1). RTKs play an important role in regulating cell motility and intravasation of multiple cancer cells. 10,11 Next, we used a human phospho-RTK array to detect alterations in phosphorylated RTKs after LECT2 treatment. Interestingly, we found that the phosphorylation of hepatocyte growth factor receptor (HGFR/MET), but not epidermal growth factor receptor, was strongly inhibited after treatment with LECT2 recombinant protein (Supporting Fig. 1). MS analysis also revealed that MET was a LECT2associated protein (Fig. 1A). To further examine the interaction between LECT2 and MET proteins, we performed coimmunoprecipitation (Co-IP) experiments and confirmed that interactions between exogenous expressed LECT2 and MET in 293T cells and endogenous expressed LECT2 and MET in Huh7 cells (Fig. 1B,C). An in vitro binding assay also revealed that the rLECT2-Fc protein directly binds to the extracellular domain (1-932 amino acids) of recombinant MET protein (Fig. 1D). In addition, confocal microscopy analysis demonstrated that endogenous LECT2 is highly expressed and located in the cytoplasm or colocalized with the MET receptor at the Huh7 cell membrane (Fig. 1E, lower panel). We also observed the same colocalization of MET and recombinant LECT2 protein in SK-Hep1 cells (Fig. 1E, upper panel). Together, these results demonstrated that LECT2 directly binds to the MET receptor at the cell membrane.
LECT2 Inhibits MET RTL Activity Without Competing With HGF Binding. We next examined whether the invasion-suppressive activity of LECT2 is mediated through MET binding. We first observed that MET phosphorylation (Tyr 1234/1235) levels were significantly decreased in the LECT2overexpressing SK-Hep1 cells and elevated in the LECT2-knockdown Huh7 cells (Supporting Fig. 2A). Next, we tested whether LECT2 could antagonize HGF, the native MET ligand-induced 12 MET phosphorylation. Interestingly, recombinant Fc-tagged LECT2, but not Fc, demonstrated a concentrationdependent inhibition in the HGF-induced MET phosphorylation (Supporting Fig. 2B) and invasion of SK-Hep1 cells (Supporting Fig. 2C). SU11274, a MET kinase inhibitor, also suppressed MET phosphorylation in LECT2-knockdown Huh7 cells and returned the invasive ability to the control level (Supporting Fig.  2D). Furthermore, overexpression of the wild-type (WT) MET dogmatically increased phosphorylation and restored the invasiveness of SK-Hep1/LECT2 cells, whereas overexpression of the kinase-dead MET (K1110A) mutant did not (Supporting Fig. 2E).
To evaluate whether LECT2 competed with HGF binding to MET, a competition enzyme-linked immunosorbent assay (ELISA) was used. The results demonstrated that the binding of HGF (0-100 nM) with MET was not affected with the addition of excess LECT2 (100 nM) and vice versa (Supporting Fig. 3A,B). Furthermore, flow cytometry (FCM) also revealed that HGF binding with cell-surface MET was not altered by coincubation with 0.5-10.0 nM of LECT2 protein (Supporting Fig. 3C). Similarly, LECT2 binding with cell-surface MET also remained consistent in the presence of up to 5 nM of HGF (Supporting Fig. 3D), confirming that LECT2 and HGF do not interfere with the binding of each other with the MET receptor. Furthermore, HGF protein can only be coimmunoprecipitated with LECT2 Ab in the presence of MET receptor (Supporting Fig. 3E).
LECT2 Functionally Binds on a-chain of MET Extracellular Domain. HGF has previously been shown to bind to the MET terminal Ig3-4 12 and, with lower affinity, the semaphorin domain. 13 To identify the LECT2-binding site on the MET receptor, a series of expression constructs containing different portions of the MET receptor extracellular domain were generated ( Fig. 2A). LECT2 had strong interaction with the a-chain and relatively weak interaction with the SEMA and IPT repeat domains ( Fig. 2A). We subsequently dissected the a-chain into four fragments to identify the minimal MET ectodomain (Fig. 2B). Co-IP results revealed that LECT2 had strong interaction with achain fragment 2 (residues 92-175) and a weaker interaction with a-chain fragment 3 (residues 159-242; Fig.  2B). An a-chain fragment with residues 159 to 175 deletion (a-D159-175) lacked most of the binding capacity to LECT2 (Fig. 2B, right panel). Invasion assays also demonstrated that a-D159-175 competed out the LECT2 inhibitory effect on invasion ability and MET receptor phosphorylation in SK-Hep1 cells (Fig. 2C).
The HxGxD Motif Is Critical for LECT2's Inhibitory Activity. To map the MET-interacting LECT2 region, we divided LECT2 into three fragments and performed co-IPs (Fig. 3A)). Co-IP results demonstrated that fragments D1 (residues 19-77) and D2 (residues 53-117) sufficiently bound MET and significantly inhibited its phosphorylation (Fig. 3A). Conditional medium of fragments D1and D2 also retained invasion-suppressive abilities to SK-Hep1 cells, such as LECT2 (Fig. 3B). Taken together, these results indicated that the 53-77 amino acid (aa) region of LECT2 binds to the 159-175 aa region of the MET a-chain and mediates the inhibition of MET phosphorylation.
To structurally characterize the LECT2 protein, we compared its amino acid sequence with known protein domains. We identified a M23 peptidase domain located between aa 51 and 147. The M23 peptidase domain contained a highly conserved HxGxD motif at aa 53-57 (Supporting Fig. 4A,B), a putative enzymeactive site previously also known to be involved in protein-protein interactions. 14 To further evaluate the function of the HxGxD motif, MET mutants harboring substitutions in this domain were generated to determine the importance of this domain to its activity (Supporting Fig. 4C). Significantly, we found that mutations at aa 53-57 (i.e., mLECT2-a and -b), but no other LECT2 aa (i.e., mLECT2-c and -d), disrupted MET-binding capacity (Fig. 3C) and signifi-cantly decreased LECT2-inhibitory effects on MET phosphorylation and invasion ability (Fig. 3D).
Furthermore, SK-Hep1 cells expressing WT LECT2 or LECT2 containing mutations outside of the presumed MET-binding site (mLECT2-d) grew small tumors, compared with HxGxD motif-mutated LECT2 (mLECT2-b) cells. Control and mLECT2-b cells had higher frequency of intrahepatic metastasis, as confirmed by histopathology (Fig. 4A). In addition, HxGxD mutant cells (mLECT2-b) exhibited strong extravascularization and lung metastasis activities, as determined by survival colonies from mouse inferior vena cava (IVC) blood (Fig. 4B), minced lung tissue (Fig. 4C), and histopathological analysis (Fig. 4D). On the contrary, orthotopically xenografted models revealed that LECT2-knockdown Huh7 cells developed larger tumor masses, more circulating cancer cells, more intrahepatic metastases, and more lung metastasis than control Huh7 cells (Supporting Fig. 5A-D). Based on these data, we suggest that the HxGxD motif is important for the LECT2 function in vascular invasion and metastasis.
LECT2 Inhibits MET Activation by Recruiting PTP1B. Phosphorylation of MET can be regulated by phosphotyrosine phosphatases (PTPs), such as densityenhanced protein-tyrosine phosphatase-1 (DEP-1), PTP-1B, and T-cell PTP (TC-PTP). [15][16][17] Because LECT2 reduced phosphorylation of MET without affecting HGF binding or MET protein levels, we examined whether LECT2 could affect interaction between MET and PTPs. We found an increased association between PTP-1B, but not DEP-1 or TC-PTP, and MET receptor in LECT2-overexpressing SK-Hep1 cells (Fig. 5A). In contrast, the association between PTP-1B and MET was decreased in LECT2-knockdown cells (Fig. 5A). Furthermore, the association between MET and PTP-1B was elevated in response to treatment with 2.5 nM of recombinant LECT2, even in the presence of 40 ng/mL of HGF (Fig. 5B).
After depletion of PTP-1B, but not TC-PTP, by specific small interfering RNA (siRNA), the recombinant LECT2 protein had no effect on HGF-induced MET phosphorylation (Fig. 5C). When the HxGxD motif was mutated, PTP-1B failed to be recruited to MET, accompanied with the retaining of MET phosphorylation and cell invasion abilities (Fig. 3D). The associations of MET downstream adaptor proteins, such as growth factor receptor-bound protein 2 (Grb2), Src, GRB2-associated binding protein 1 (Gab1), and p85, were impaired by LECT2 protein with or without HGF administration (Fig. 5D). Downstream signaling of the HGF/MET axis, as with Raf-1 and extracellular signal-regulated kinase (Erk) phosphorylation levels were significantly decreased in LECT2-overexpressing SK-Hep1 cells and increased in LECT2-knockdown Huh7 cells (Fig. 5E). Despite a slight decrease in recruitment of p85 to MET, phosphorylation of protein kinase B (Akt) was not affected in response to LECT2 treatment (Fig. 5E). These results support the role of LECT2 as a negative regulator of MET signaling in HCC.
LECT2 Was Negatively Correlated With the Phospho-MET Level in Patients With HCC. MET expression is an important prognostic factor for vascular invasion and HCC progression. 18,19 To determine the clinical significance of our findings, we analyzed LECT2 and phospho-MET levels in primary tumors from HCC patients (n 5 73). As expected, there was a significant inverse correlation between LECT2 and MET phosphorylation level in these specimens (P 5 0.0004; Fig.  6A,B). Furthermore, LECT2 and MET were coimmuniprecipitated with each other in all four nonvascular invasive tumor specimens tested. MET phosphorylation levels were almost undetectable in these nonvascular invasive specimens. In contrast, MET was precipitated without LECT2 and had high phosphorylation levels in the vascular-invasive specimens (Fig. 6C). We also observed that patients with high LECT2 expression and low p-MET/MET level had less vascular-invasive tumors with longer survival time, whereas patients with low LECT2 expression and high p-MET/MET level had more vascular-invassive tumors and shorter survival time (Fig.  6D,E). Moreover, five HGF/MET regulated genes, including tubulin, beta 6 class V, FYVE, RhoGEF and PH domain-containing 6, NCK adaptor protein 2, actinrelated protein 2/3 complex, subunit 1B, and highmobility group AT-hook 1, which play important roles in metastasis formation or are related to cell motility and invasiveness in HCC, 18 were significantly inversely correlated with LECT2 expression levels in a Gene Expression Omnibus database (GSE9843; Supporting Fig. 6A-E). These results indicate that the role of the LECT2/MET axis is associated with vascular invasion of HCC.

Discussion
Through cell functional assays, animal studies, and biochemical and molecular dissections, we demonstrated that LECT2 has unique functions in HCC tumor suppression through a mechanism involving the interaction with the MET receptor and subsequent inhibition of its activation. Of note, our data demonstrate that LECT2 associates with MET without affecting HGF binding. LECT2 was able to suppress HGFinduced MET activation, even in the presence of excess levels of HGF. The absence of LECT2 strongly promoted HCC progression, presumably as a result of increased HGF/MET activation. Furthermore, our results also demonstrate that LECT2 directly binds to the a-chain of MET, leading to recruitment of PTP1B. The recruitment of PTP-1B leads to the MET dephosphorylation and dissociation of adaptor proteins, including Gab1, Grb2, p85, and Src (Fig. 7).
Despite clinical observations, the functional roles and molecular mechanisms of LECT2 in HCC have not been fully investigated. LECT2 was identified as a direct target gene of b-catenin in mouse liver. 5 Recently, b-catenin-activated LECT2 has been shown to antagonize the pro-inflammatory effects of nuclear factor of kappa light polypeptide gene enhancer in B-cells 1 and b-catenin. Furthermore, b-catenin-activated transgenic mice also developed highly malignant HCC with lung metastasis in a LECT2-depleted background (LECT2 2/2 ). 6 The HGF/MET-mediated activation of b-catenin has been well documented, [20][21][22] and it is possible that LECT2 expression may also be regulated by MET-activated b-catenin. However, given our current data showing an inverse correlation between LECT2 and MET, it is unlikely that LECT2 is activated by the MET-activated b-catenin pathway. Although a previous report suggested that LECT2 expression correlates with a deregulated Wnt-signaling pathway in hepatocytes and hepatoblastoma cells, LECT2 was not up-regulated in all HCC specimens harboring b-catenin-activated mutations, suggesting additional regulation machinery for LECT2 expression in HCC. 5 Nevertheless, the molecular interplay between MET, b-catenin, and LECT2 in HCC malignancy requires further investigation.
The HGF/MET axis plays a pleotropic role in cell proliferation, migration, invasion, angiogenesis, and survival. 23,24 The central role of MET activity in cancer progression and disparities in quiescent HGF/MET signaling in normal tissue and overexpression in tumors may provide a degree of tumor selectivity for therapeutic intervention, making HGF or MET inhibition an attractive strategy in oncology. A variety of HGF/MET axis inhibitors have been developed, including small-molecule compounds targeting MET kinase activity, neutralizing  anti-MET 25 and anti-HGF Abs, 26 decoy receptors, 27 and HGF-derived factors, such as NK4. 28 Remarkably, most of these inhibitors act as competitive antagonists to the HGF/MET axis, with the exception of RTK inhibitors. In our study, LECT2 predominately binds to aa 159-175 of the MET a-chain and does not interrupt HGF binding, as demonstrated by an ELISA competition assay and FCM analysis. Unlike known MET antagonists, LECT2 acts as a noncompetitive MET antagonist. Such endogenous allosteric regulation is frequently observed for metabolic enzymes, but is rarely found for RTKs. Allosteric antagonists may prevent ligand-induced conformational changes required for receptor activation. Furthermore, the association of PTP-1B and the dissociation of adaptor proteins, such as Gab1, Grb2, and Src, which are induced by LECT2 binding, may also result from conformational stabilization or changes in MET. Future structural basis analyses, such as X-ray cocrystallization or nuclear magnetic resonance spectroscopy, may provide a better understanding of how LECT2 alters MET activation. Taken together, our results suggest that LECT2 is a novel MET antagonist, and the epitope at aa 159-175 in the MET achain may be a potential target for developing MET inhibitors.
In conclusion, we have revealed a significant correlation between LECT2 expression and MET activation, which is extensively recognized as a key regulator to HCC aggressiveness. Our findings also uncover a novel mechanism of MET regulation by the LECT2 protein and display the potential for developing LECT2 as an HCC therapeutic agent.