|Home | About | Journals | Submit | Contact Us | Français|
The mannose 6-phosphate/insulin-like growth factor II receptor (M6P/IGF2R), a multifunctional protein, plays a central role in intracellular targeting of lysosomal enzymes and control of insulin-like growth factor II (IGF-II) bioactivity. Importantly, the gene encoding this receptor is frequently inactivated in a wide range of malignant tumors including hepatocellular carcinomas. Thus, M6P/IGF2R is considered a putative liver tumor suppressor. The aim of this study was to establish the impact of the receptor on the invasive properties of liver cells.
Reconstitution experiments were performed by expression of wild type and mutant M6P/IGF2R in receptor-deficient FRL14 fetal rat liver cells. RNA interference was used to induce M6P/IGF2R downregulation in receptor-positive MIM-1–4 mouse hepatocytes.
We show that the M6P/IGF2R status exerts a strong impact on the invasiveness of tumorigenic rodent liver cells. M6P/IGF2R-deficient fetal rat liver cells hypersecrete lysosomal cathepsins and penetrate extracellular matrix barriers in a cathepsin-dependent manner. Forced expression of M6P/IGF2R restores intracellular transport of cathepsins to lysosomes and concomitantly reduces the tumorigenicity and invasive potential of these cells. Conversely, M6P/IGF2R knock-down in receptor-positive mouse hepatocytes causes increased cathepsin secretion as well as enhanced cell motility and invasiveness. We also demonstrate that functional M6P-binding sites are important for the anti-invasive properties of M6P/IGF2R, whereas the capacity to bind IGF-II is dispensable for the anti-invasive activity of the receptor in liver cells.
M6P/IGF2R restricts liver cell invasion by preventing the pericellular action of M6P-modified proteins.
Liver cancer represents worldwide one of the most frequent malignancies . Hepatocellular carcinomas (HCCs) are the most common primary liver tumors. Although many improvements have been made in terms of diagnosis and treatment, HCCs are associated with poor clinical prognosis . Aggressive HCCs have the capacity to penetrate extracellular matrix (ECM) barriers and spread into the surrounding parenchyma, leading to intrahepatic metastasis and portal venous invasion .
Different proteinases are involved in the breakdown of ECM components during tumor invasion and metastasis, including plasminogen activators, matrix metalloproteinases, and cathepsins [4–6]. Hepatocytes are known to produce substantial amounts of the lysosomal proteinases cathepsin B, cathepsin D, and cathepsin L . As typical for lysosomal enzymes, the N-glycan moieties of cathepsins are modified during their biosynthesis with mannose 6-phosphate (M6P) residues which permit interaction with the main lysosomal sorting receptors, the 300-kDa mannose 6-phosphate/insulin-like growth factor II receptor (M6P/IGF2R) and the 46-kDa mannose 6-phosphate receptor (MPR46) .
M6P/IGF2R is a multifunctional receptor involved in (a) transport of newly synthesized M6P-tagged lysosomal proteins from the Golgi network to lysosomal compartments, (b) endocytosis of extracellular M6P-tagged lysosomal enzymes, (c) proteolytic activation of transforming growth factor β, and (d) regulation of the bioavailability of IGF-II [8–10]. To perform all these important functions, the receptor depends on multiple ligand-binding sites. For some M6P/IGF2R ligands, the receptor domains accommodating the respective binding sites have been identified. Furthermore, point mutations have been described which specifically interfere with the interaction of the receptor with IGF-II or M6P-tagged proteins [11–14].
Several lines of evidence support the hypothesis that loss of M6P/IGF2R function is associated with liver tumor progression. The gene encoding M6P/IGF2R has been shown to undergo frequent loss of heterozygosity in human HCCs and adenomas, with concomitant inactivating mutations in the remaining allele [15,16]. It has been demonstrated that the receptor gene is lost early in liver tumorigenesis, which suggests that loss of M6P/IGF2R may represent an initiation event . Given this clinical significance, it comes as a surprise that the impact of the M6P/IGF2R status on the properties of hepatocytes and HCC cells has not yet been firmly established. However, studies in choriocarcinoma and breast cancer cells have demonstrated that a decrease in M6P/IGF2R expression enhances tumor cell growth [18,19], whereas overexpression of the receptor causes the opposite effect [20,21].
Experiments in transgenic mice have indicated that the anti-tumorigenic properties of M6P/IGF2R could be linked to its capacity to downregulate the biological activities of IGF-II . However, other studies indicate that failure to express M6P/IGF2R may result in an increase of the proteolytic load in the pericellular environment and thus enhance the invasive capacity of tumor cells [23–25]. Thus, it seems that multiple M6P/IGF2R ligands play a role in tumor formation and metastasis, possibly in a tissue-specific manner. In this study, we have now assessed the impact of M6P/IGF2R on the growth, motility and invasiveness of liver cells.
A detailed account of the methodology used in this study can be found in the Supplementary Materials and methods.
There is now compelling evidence to suggest that in some liver pathologies HCCs derive from liver stem/progenitor cells or immature hepatoblasts. Hence, HCC cells often more closely resemble fetal than adult hepatocytes in terms of their gene expression pattern . To permit detailed studies on a cellular level, substantial efforts have been undertaken to establish transformed hepatocyte-like cell lines from fetal rat liver of different gestational stages. One such cell line, FRL19, has been previously derived from a prenatal rat liver at day 18.5 of gestation and shown to express several hepatocyte markers, including the late-onset enzymes tyrosine aminotransferase and alpha-glutathione S-transferase [27,28]. We have now generated a cell line derived from day 13.5 fetal rat hepatocytes. The morphology of these FRL14 cells reflects their origin from the hepatocyte lineage. Furthermore, FRL14 cells express early-stage hepatocyte markers such as transferrin, while late-onset genes are not expressed (Supplementary Fig. 1).
FRL14 cells are capable of migrating across ECM barriers as typical for malignant cancer cells. Since such invasive properties frequently depend on ECM proteolysis, the contribution of matrix-degrading proteinases to the invasive potential of FRL14 cells was assessed by invasion assays performed in the presence of different synthetic and endogenous proteinase inhibitors (Supplementary Fig. 2). The strongest effect was caused by the general lysosomal cysteine proteinase inhibitor E-64 (60% reduction). Inhibition of FRL14 invasion by the more selective E-64 derivative CA-074 (40%) and the matrix-metalloproteinase inhibitor GM6001 (46%) was less pronounced. When the efficacy of physiological proteinase inhibitors was tested, it was found that the potent serine proteinase inhibitor aprotinin had a much weaker effect on cell invasion (22% reduction) than the cysteine proteinase inhibitor cystatin C (49% reduction). We have earlier observed a similar proteinase inhibition profile for the invasive properties of M6P/IGF2R-deficient murine squamous cell carcinoma cells . When FRL14 extracts were immunoblotted with antibodies recognizing rat M6P/IGF2R, it became evident that these cells lack expression of the receptor (Fig. 1).
Receptor-deficient FRL14 cells were stably transfected with human wild type M6P/IGF2R cDNA to assess the impact of the M6P/IGF2R status on their cellular properties. Two clones were selected for further studies, FRL14/IGF2R wt-1 and FRL14/IGF2R wt-2. By comparison with receptor-positive HeLa human cervical carcinoma cells, the M6P/IGF2R content of FRL14/IGF2R wt-1 and FRL14/IGF2R wt-2 cells was estimated to be 2.5 and 2.1 pmol/mg total cell protein, respectively (Supplementary Table 1). Hence, the receptor level of the selected clones was within the physiological range . The subcellular localization of M6P/IGF2R was assessed by immunofluorescence microscopy. As expected, the ectopically expressed receptor was found to reside in the Golgi apparatus (Supplementary Fig. 3).
M6P/IGF2R transports newly synthesized lysosomal enzymes from the Golgi network to lysosomal compartments [30,31]. To test the sorting function of the recombinant receptor, we analyzed the intra- and extracellular activity of the lysosomal marker β-N-acetylhexosaminidase. Due to the lack of M6P/IGF2R, parental and mock-transfected FRL14 cells secrete large amounts of β-N-acetylhexosaminidase (39 ± 2% and 51 ± 3%, respectively). Ectopic expression of human M6P/IGF2R reduced β-N-acetylhexosaminidase secretion considerably (FRL14/IGF2R wt-1: 21 ± 2%, FRL14/IGF2R wt-2: 21 ± 1%). The addition of NH4Cl, a lysosomotropic base interfering with M6P/IGF2R function [23,32], strongly increased secretion of β-N-acetylhexosaminidase by FRL14/IGF2R wt cells. The fact that also both parental and mock-transfected FRL14 cells showed an increase in β-N-acetylhexosaminidase secretion upon NH4Cl treatment is probably due to the presence of MPR46, which also transports lysosomal enzymes in an NH4Cl-sensitive manner (Fig. 2A). Additionally, the intracellular transport of cathepsin B and cathepsin D was analyzed. Substantially more procathepsin B was found in the medium of parental FRL14 cells (38% secretion) as compared to M6P/IGF2R transfectants (16% secretion). For cathepsin D, the difference in secretion between FRL14/IGF2R wt (21% secretion) and parental FRL14 cells (27% secretion) was not as pronounced (Fig. 2B). When FRL14/IGF2R wt cells were subjected to subcellular fractionation by Percoll density-gradient centrifugation, 21% of the total β-N-acetylhexosaminidase activity was found in the dense (lysosomal) fraction, whereas in parental FRL14 cells only 7% of this enzyme resides in lysosomes (Supplementary Fig. 4). These results demonstrate that ectopic expression of M6P/IGF2R in FRL14 cells leads to improved intracellular retention of lysosomal enzymes and restoration of dense lysosome formation, as previously shown for receptor-deficient squamous cell carcinoma cells .
It has been established that the metastatic properties of tumor cell lines closely correlate with the ability of the cells to migrate across Matrigel barriers in vitro . The invasiveness of M6P/IGF2R-expressing FRL14 cells was found to be reduced by 76% compared to mock-transfected FRL14 cells, and by 86% compared to parental FRL14 cells (Fig. 2C). This indicates that M6P/IGF2R is indeed a potent anti-invasive factor in liver cells.
The motility of parental and receptor-expressing FRL14 cells was also tested in wound healing assays. Parental FRL14 monolayers were highly active in wound closure. After 7 h, all wounds were fully closed, reflecting a mean covered distance of 218 μm. Mock-transfected FRL14 cells behaved similar to parental cells (186 μm). In contrast, the wounds of FRL14/IGF2R wt cells (118 μm) were only about half-closed after this incubation period. These results show that the presence of M6P/IGF2R strongly reduces the motility of FRL14 cells in vitro (Fig. 2D and E).
Anchorage-independent growth is believed to be a hallmark of cellular transformation . Soft-agar assays were performed to test whether FRL14 cells are able to grow in an anchorage-independent manner. The colony-formation efficiency of FRL14/IGF2R wt cells (81 ± 6 colonies) was considerably lower than that of parental FRL14 cells (466 ± 23 colonies; p <0.001). In addition, colonies of FRL14/IGF2R wt cells (0.013 ± 0.002 mm2) were much smaller than those of FRL14 parental cells (0.037 ± 0.005 mm2; p = 0.005).
Furthermore, we tested the ability of FRL14 cells to form tumors in immunodeficient mice. Contrary to parental FRL14 cells (tumors/injections: 9/9; tumor weight: 13–219 mg; median: 48 mg), tumor formation by FRL14/IGF2R wt cells was reduced and the tumors were smaller (tumors/injections: 5/9; tumor weight: 0–100 mg; median: 18 mg). Collectively, our results thus indicate that reconstitution of functional M6P/IGF2R expression interferes with FRL14 tumor progression.
M6P/IGF2R is a multifunctional protein with separate binding sites for IGF-II (domain 11) and M6P-modified ligands (domains 3 and 9). These binding sites were individually mutated, thus creating M6P/IGF2R variants either impaired in their interaction with IGF-II (M6P/IGF2R dom11mut) or M6P (M6P/IGF2R dom3/9mut). Parental FRL14 cells were then stably transfected with M6P/IGF2R dom11mut and M6P/IGF2R dom3/9mut cDNAs. The receptor content of the resulting cell lines was found to be similar to or slightly above that of FRL14/IGF2R wt cells (Supplementary Table 1). Indirect immunofluorescence studies revealed that the subcellular distribution of M6P/IGF2R dom11mut and M6P/IGF2R dom3/9mut is identical to that of the wild type receptor. This confirms that these mutations do not compromise folding and/or intracellular targeting of M6P/IGF2R in FRL14 cells.
Wild type and mutant forms of M6P/IGF2R were subjected to IGF-II and phosphomannan binding assays. As expected, the wild type receptor was able to bind both ligands. The interaction of M6P/IGF2R dom11mut with IGF-II was much weaker, while this mutation did not impede binding of M6P. The M6P binding site mutant M6P/IGF2R dom3/9mut showed no detectable binding of M6P, but the ability to interact with IGF-II was preserved (Fig. 3A and Supplementary Fig. 5).
To study the function of the mutant receptors, the intra- and extracellular activity of β-N-acetylhexosaminidase was measured. FRL14/IGF2R dom11mut cells secreted 19 ± 1% of their β-N-acetylhexosaminidase activity, which is similar to cells expressing the wild type receptor (21% secretion). Both FRL14/IGF2R dom3/9mut clones secreted a larger proportion of this enzyme (31 ± 1% in either case), but β-N-acetylhexosaminidase secretion is not as pronounced as in mock-transfected cells (51%). The addition of NH4Cl strongly increased the secretion of β-N-acetylhexosaminidase by all cell lines tested, including those expressing mutant receptors (Fig. 3B).
As already observed for FRL14/IGF2R wt cells, the invasive potential of FRL14/IGF2R dom11mut cells was found to be strongly diminished compared to parental FRL14 cells (80% reduction). In contrast, the invasive capacity of FRL14/IGF2R dom3/9mut cells (14% reduction) was only slightly lower than that of parental FRL14 cells. This suggests that the ability to bind IGF-II is dispensable for the anti-invasive activity of M6P/IGF2R in liver cells. In contrast, the anti-invasive potential of the receptor appears to be largely based on its ability to interact with M6P-modified ligands (Fig. 3C).
RNAi-mediated gene silencing was used to knock-down M6P/IGF2R expression in receptor-positive MIM-1–4 mouse hepatocytes. For transient knock-down studies, MIM-1–4 cells were transfected with siRNA oligonucleotides targeting Igf2r mRNA. Densitometric analysis of immunoblots revealed that siRNA treatment reduced the receptor content of the cells to <5%. This almost quantitative depletion of endogenous M6P/IGF2R provoked a strong increase in the secretion of β-N-acetylhexosaminidase (37 ± 1%) as compared to mock-transfected cells (11 ± 1%). The invasive potential of siRNA-treated MIM-1–4 cells was also markedly higher than that of the respective controls, using either HGF/FBS (1.9-fold increase) or FBS alone (3.8-fold increase) as chemoattractant (Supplementary Fig. 6).
For permanent receptor knock-down, MIM-1–4 cells were stably transfected with an M6P/IGF2R-specific shRNA construct. Densitometric analysis of immunoblots revealed that the residual receptor content of the two MIM-1–4/IGF2R shRNA clones chosen for further studies was <1% as compared with cells transfected with a control shRNA sequence (Fig. 4A). This goes in hand with enhanced β-N-acetylhexosaminidase secretion, with MIM-1–4/IGF2R shRNA cells (line 1: 55%; line 2: 47%) secreting far more of this lysosomal marker enzyme than parental (7%) and control cells (16%). Similar observations were made for cathepsin D. MIM-1–4/IGF2R shRNA cells secreted substantially more of this lysosomal proteinase (11%) than parental MIM-1–4 cells (1%). The difference in cathepsin L secretion between MIM-1–4/IGF2R shRNA and parental cells was less pronounced (87% and 81%, respectively). NH4Cl treatment resulted in strongly enhanced β-N-acetylhexosaminidase secretion by parental and control MIM-1–4 cells. Conversely, MIM-1–4/IGF2R shRNA cells displayed no further increase in β-N-acetylhexosaminidase secretion upon addition of NH4Cl (Fig. 4B and C).
Wound healing assays were performed to test the effect of M6P/IGF2R knock-down on the motility of MIM-1–4 cells. MIM-1–4/IGF2R shRNA-1 (mean covered distance: 152 μm) and MIM-1–4/IGF2R shRNA-2 cells (106 μm) migrated substantially faster than parental (56 μm) and control cells (78 μm; Fig. 4D and E). In Matrigel invasion assays performed with HGF/FBS as chemoattractant, MIM-1–4/IGF2R shRNA-1 cells proved 2.3-fold more invasive than control cells. MIM-1–4/IGF2R shRNA-1 cells also displayed considerable invasiveness in the absence of exogenous HGF, while FBS alone was not sufficient to induce an appreciable invasive response by control cells (Fig. 4F).
The gene encoding M6P/IGF2R is frequently mutated during human and rodent hepatocarcinogenesis [35,36], and some of these mutations have been shown to inactivate individual receptor functions [12,13]. However, the role of aberrant M6P/IGF2R expression in HCC formation and progression is still unknown. In this study, we have assessed the impact of the M6P/IGF2R status on the tumorigenic and invasive properties of a receptor-negative transformed fetal rat liver cell line (FRL14). We have found that reconstitution of M6P/IGF2R expression in FRL14 cells suppresses their tumorigenicity and invasiveness. Furthermore, the reconstituted cells were less motile and displayed diminished growth under anchorage-independent conditions. Vice versa, RNAi-mediated M6P/IGF2R knock-down in receptor-positive mouse hepatocytes enhanced their migratory and invasive potential. These results clearly indicate a mechanistic link between dysfunctional M6P/IGF2R expression and HCC pathogenesis, in agreement with the tumor-suppressive activities of the receptor in other forms of cancer [22,37].
In squamous cell carcinoma cells, the anti-invasive activity of M6P/IGF2R appears to rely on restriction of the pericellular accumulation of M6P-tagged lysosomal proteinases [23–25]. In contrast, the anti-invasive effects of M6P/IGF2R in breast cancer cells have been attributed to decreased IGF-II bioavailability . Moreover, evidence has been provided that the anti-invasive properties of M6P/IGF2R in renal carcinoma cells are based on downregulation of cell-mediated plasminogen activation . In our further studies, we therefore sought to establish the importance of the individual receptor-ligand interactions for the anti-invasive activity of M6P/IGF2R in liver cells. We selected to focus on the capacity of M6P/IGF2R to bind IGF-II and M6P-containing ligands, since plasmin was found to play only a minor role in FRL14 cell invasion. Furthermore, IGF-II has been recently found to enhance the motility of human HCC cells . Structure-function studies have led to the identification of the M6P/IGF2R domains accounting for binding of IGF-II (domain 11) and M6P residues (domains 3 and 9). Point mutations within these domains have been described that selectively interfere with either interaction [11,14]. The introduction of a mutation known to reduce the affinity of M6P/IGF2R for IGF-II did not impede its anti-invasive capacity in FRL14 cells. Conversely, simultaneous mutation of both M6P-binding sites almost abolished this activity of the receptor. These results strongly support the notion that the anti-invasive properties of M6P/IGF2R in liver cells depend on its interaction with M6P-tagged lysosomal hydrolase(s). Various lysosomal enzymes have been implicated to participate in the process of tumor invasion and metastasis. So far, the lysosomal enzymes most extensively studied in this context are cysteine cathepsins .
Liver tumor cells are known to secrete large amounts of cathepsins into the surrounding tissue . Therefore, it can be envisaged that absence of M6P/IGF2R fosters HCC invasion by exacerbating the secretion of these potent proteinases. Indeed, we have observed that treatment with cysteine cathepsin inhibitors results in substantial reduction of the invasiveness of FRL14 cells. The main cellular targets of these inhibitors in liver cells are the two most prominent cysteine cathepsins, cathepsin B, and cathepsin L . For cathepsin B, it has been shown previously that this enzyme may exhibit extracellular functions under pathophysiological conditions . Compelling evidence for a major role of cathepsin B and related proteinases in tumor invasion and metastasis has been derived from pancreatic and breast cancer models in cathepsin knock-out mice [42–44]. Furthermore, transgenic mice overexpressing human cathepsin B have been found to develop breast cancer metastasis at a much higher rate than their control littermates . Intriguingly, treatment of rodents with a broad-spectrum cysteine cathepsin inhibitor has proven effective in halting breast cancer progression when combined with other chemotherapeutic agents, encouraging consideration of cysteine cathepsins as therapeutic targets in human cancers . However, it remains to be established whether pharmacological intervention with tumor-associated cysteine cathepsin activity holds promise for HCC treatment.
The authors who have taken part in this study declare that they do not have anything to disclose regarding conflict of interest with respect to this manuscript.
Austrian Science Fund (FWF): P16925-B11; Austrian Science Fund (FWF): P20918-B11; Austrian Academy of Sciences: H-7/2005.
We authors express gratitude to Vincent Keng for assisting in the generation of the FRL14 cell line. We also thank Ann Erickson, Regina Pohlmann, Magnus Abrahamson, Bernard Hoflack, Nobuhiko Katunuma, Vladimir Leksa and Hannes Stockinger for reagents. The expert technical assistance of Barbara Svoboda is gratefully acknowledged.
Appendix ASupplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jhep.2012.03.026.
Supplementary Materials and methods
M6P/IGF2R content of FRL14 cells expressing human M6P/IGF2R.
PCR primer sets for generation of mutant M6P/IGF2R cDNAs.