Potential conflict of interest: Nothing to report.
Primary liver cancer, hepatocellular carcinoma (HCC), is the fifth most common cancer and the third leading cancer killer in the world. There is no effective therapeutic option for most HCC patients. A new therapeutic strategy is essential. Granulin-epithelin precursor (GEP, also called progranulin, acrogranin, or PC-derived growth factor) was identified as a potential therapeutic target for HCC from our earlier genome-wide expression profiles. We aimed to conduct a detailed investigation with in vitro and animal experiments. We developed the anti-GEP monoclonal antibody (mAb), and examined its effect on hepatoma cells and normal liver cells in vitro. A nude mice model transplanted with human HCC was used to investigate if anti-GEP mAb can inhibit tumor growth in vivo. We demonstrated that anti-GEP mAb inhibited the growth of hepatoma cells but revealed no significant effect on normal liver cells. In the nude mice model transplanted with human HCC, anti-GEP mAb decreased the serum GEP level and inhibited the growth of established tumors in a dose-dependent manner. The anti-GEP mAb reduced tumor cell proliferation via the p44/42 MAPK and Akt pathways, and reduced tumor angiogenesis to deprive the nutrient supply with reduced microvessel density and tumor vascular endothelial growth factor level. Conclusion: We have shown that anti-GEP antibody can inhibit HCC growth, providing evidence that GEP is a therapeutic target for HCC treatment. (HEPATOLOGY 2008.)
About half a million individuals die from hepatocellular carcinoma (HCC) annually1, 2 because the disease is frequently diagnosed at an advanced stage and thus precludes curative surgical treatment. Surgical resection in the form of partial hepatectomy or liver transplantation is the mainstay for curative treatment but is confined to highly selected patients.3 Most HCC patients have advanced cirrhosis that limits treatment options, and chemotherapy has shown only marginal efficacy and is accompanied by severe side effects.4, 5 Clearly, there exists an urgent need for more effective therapy to treat HCC.6 Targeted cancer therapy can limit nonspecific toxicity and improve therapeutic efficiency compared with chemotherapeutic agents.7 With the advances in hybridoma technology, targeted cancer therapy can be achieved by the use of monoclonal antibody (mAb) therapy,8 such as anti-HER2/neu (trastuzumab/Herceptin) for metastatic breast cancer and anti-EGFR (cetuximab/Erbitux) for colorectal cancer,8 which has proven efficacy in clinical cancer treatment. However, the development of targeted therapy for HCC is limited.6
We and others9–12 have reported the genome-wide expression profiles of HCC and their clinical implications. The microarray approach enables the study of genomewide expression patterns and identification of novel targets for disease treatment.13 We explored differentially expressed genes and demonstrated the important clinical relevance of a number of genes.14–18 We have performed extensive validation using a new independent clinical sample set and functional analyses to delineate their biological roles in HCC carcinogenesis. To identify potential molecular targets of cancer therapy, we hypothesized that they should (1) have clinically relevant biological functions; (2) differentially express in a significant portion of tumors such that it would have wide application; and (3) preferably have a secretory factor that allows identification of target (susceptible) patients and monitoring of treatment response by a simple blood test.
Recently, we demonstrated that granulin-epithelin precursor (GEP) fulfills the aforementioned criteria as a molecular target for cancer therapy.14 GEP is an autocrine growth factor involved in cancer progression,19 development,20 and wound healing,21 whereas mutation of GEP causes frontotemporal dementia.22, 23 We showed that GEP controls HCC proliferation, invasion, and tumorigenicity.14 These biological roles correspond to the clinical findings that expression of GEP is associated with aggressive HCC features including large tumors, venous infiltration (micrometastasis), and early recurrence after curative surgery.14 In addition, our earlier microarray examined GEP on the mRNA level,9–10 and subsequent validation on GEP protein expression in more than 200 HCC and liver tissue adjacent to tumor samples confirmed that more than 70% of HCCs revealed GEP overexpression.14 Thus, we hypothesized that GEP is a potential therapeutic target for HCC. In the present study, we examined the therapeutic potential of anti-GEP antibody on HCC treatment.
New Zealand white rabbits were immunized with Keyhole Limpet Hemocyanin (KLH)–conjugated custom-made specific peptide against the GEP central region (Zymed Laboratories Inc., San Francisco, CA). The rabbit antisera were affinity-purified using the immobilized antigen column.
Anti-GEP Monoclonal Antibody.
BALB/c mice were immunized with KLH-conjugated custom-made specific peptide against the GEP carboxyl-terminus (Zymed Laboratories). Fusion of the spleen cells with a nonproducer myeloma line, NS0, was performed. Isotypes of the monoclonal antibody were determined using the Mouse MonoAB ID Kit (Zymed Laboratories) and determined as IgG1κ. The hybridoma cells were injected into the mice intraperitoneally to allow the formation of ascites, which were then collected and the antibodies purified using Protein G sepharose (AP Biotech, Chalfont St. Giles, UK).
Western Blot and Immunoprecipitation.
Protein extracts were separated on SDS-PAGE gels followed by Western blots.14 Monoclonal mouse anti-GEP (0.5 μg/mL) and polyclonal rabbit anti-GEP (2 μg/mL) were used. Polyclonal goat anti-β-actin antibody was used at 1:1,000 dilution (DAKO, Glostrup, Denmark). Akt, phospho-Akt (Ser473), p44/p42 mitogen-activated protein kinase (MAPK), and phospho-p44/42 MAPK (Thr202/Tyr204) antibodies were used according to the manufacturer's instructions (Cell Signaling Technology Inc., Danvers, MA). Vascular endothelial growth factor (VEGF) protein was detected with antihuman VEGF antibody (R&D Systems, Minneapolis, MN). Secondary antimouse, antirabbit, and antigoat horseradish peroxidase–conjugated antibodies (AP Biotech) were used at 1:3,000 dilutions. Immunoprecipitation was performed with 500 μg of cell lysate and incubated with 1 μg of anti-GEP monoclonal antibody A23. The immunocomplexes were separated on an SDS-PAGE gel and immunoblotted with the anti-GEP polyclonal antibody. Enhanced chemiluminescence (AP Biotech) was performed according to the manufacturer's instructions.
ELISA to Determinate GEP Level.
ELISA plates (Nalge Nunc International, Rochester, NY) were coated with anti-GEP monoclonal antibody A23 (10 μg/mL). The plates were then blocked (1× PBS, 1% BSA, 5% sucrose, 0.05% NaN3) and incubated with samples. Bound GEP was detected using the anti-GEP rabbit polyclonal antibody (0.5 μg/mL). Detection was performed by incubation with horseradish peroxidase–conjugated goat antirabbit IgG (Zymed Laboratories), followed with TMB (Pierce Biotechnology Inc., Rockford, IL) as substrates. To quantify the GEP level in the serum, a calibration curve of purified recombinant GEP protein in serial dilution (from 30 ng/mL to 469 pg/mL) was performed in parallel. Each sample was measured 3 times, each in duplicate.
Cell proliferation was measured using the standard MTT assay. Each data point represented results from 3 independent experiments, each performed in triplicate. For assays of secretory GEP level in the culture supernatants and the status of intracellular MAPK and Akt, the cells were serum-starved for 24 hours and then incubated with or without A23 at 100 μg/mL in DMEM supplemented with 1% FBS for 72 hours.
Anti-GEP Antibody Treatment in Nude Mice.
The study protocol was approved by the Committee on the Use of Live Animals in Teaching and Research of the University of Hong Kong. Hep3B cells (2 × 106cells/mouse) were injected subcutaneously in 5-week-old to 6-week-old male athymic nude mice. Treatments were started when the tumor size reached approximately 0.3 cm3, and the mice were then randomized into control and treatment groups. In the control treatment groups, the mice were injected with either purified mouse IgG 100 μg (n = 3; Sigma-Aldrich, Saint Louis, MO) or saline (n = 5). In the A23 treatment groups, the mice were injected with either 50 μg (n = 5) or 100 μg (n = 5) doses of A23 antibody. Tumor sizes were determined by Vernier caliper measurements, and the tumor volume was calculated according to the formula (ab2)/2, where a and b are the largest and smallest diameters, respectively.
Immunohistochemistry was performed on formalin-fixed paraffin-embedded sections.14 Ki-67 was stained using the anti-Ki-67 monoclonal antibody (BD Biosciences Pharmingen, San Diego, CA). The mitotic index was assessed by light microscopy on hematoxylin-eosin-stained sections, and the average number of mitotic figures in 10 high-power field (magnification ×400) was determined. The TUNEL assay was performed using an In Situ Cell Death Detection kit (Roche Diagnostics GmbH, Mannheim, Germany). Tissue sections from internal organs were stained with hematoxylin-eosin for histology assessment. For assessment of liver fibrosis, sections were stained with Masson's Trichrome, which would highlight the collagen in green. Paired sections stained with hematoxylin-eosin and Masson's Trichrome were compared if there were increased portal, perivenular, and septal fibrosis. All the histological assessments were performed by a clinical pathologist.
The microvessels were highlighted by CD34 immunostaining, and microvessel density was evaluated according to Gasparini's criteria24 by 2 independent observers who were blinded to the treatment group. The mean microvessel count of 5 fields at a magnification ×200 was taken as the microvessel density, which was expressed as the absolute number of microvessels per 0.74 mm2 (the area at a magnification ×200).
All statistical analyses were performed by SPSS version 12.0 for Windows (SPSS Inc., Chicago, IL). Continuous variables were assessed by the Spearman correlation and compared between groups by the Student's t test. A P value less than 0.05 was considered statistically significant.
The anti-GEP monoclonal antibody A23 (raised against the carboxyl-terminal of GEP) can specifically recognize the glycosylated GEP at about 90 kDa (Fig. 1A). The specificity of the antibody was confirmed by immunoprecipitation, and the immunocomplexes pulled down by the monoclonal antibody A23 were detected by the polyclonal antibody (raised against the central region of GEP; Fig. 1B). In agreement with our earlier findings using polyclonal antibody,14 GEP was revealed in the cytoplasm of the cancerous liver cells. Notably, GEP was not detected in the endothelial cells or fibroblastic components of the cancerous tissues or in the liver cells adjacent to the tumors.
To examine if GEP is a potential therapeutic target for HCC, we measured the level of GEP secreted into the culture medium. We showed that incubation of anti-GEP monoclonal antibody A23 with hepatoma cells Hep3B and HepG2 reduced the GEP level in the medium (Fig. 1C), reduced the intracellular GEP mRNA level (Fig. 1D), and inhibited MAPK activation (Fig. 1E). Notably, Akt activation was not detectable in vitro (Supplementary Fig. 1) and thus would be examined later in the mouse model. The cell proliferation rates of both types of hepatoma cells were significantly inhibited in a dose-dependent manner with the addition of A23 but had no significant effect on the normal human liver cells MIHA (Fig. 2A). Thus, anti-GEP suppressed the growth of HCC cells but not normal liver cells.
We then investigated whether anti-GEP treatment can inhibit the growth of subcutaneous Hep3B xenografts in nude mice (both HepG2 and MIHA expressed a lower level of GEP and were nontumorigenic). To simulate the clinical situation of intervention on large unresectable tumor, treatment was started at a tumor size of 0.3 cm3. Intraperitoneal injection of anti-GEP monoclonal antibody A23 inhibited the growth of HCC xenografts in a dose-dependent manner compared with that in the controls (Fig. 2B). A23 treatment decreased the serum GEP level, and the reduction in the GEP level correlated with the extent of tumor growth inhibition (Fig. 2C). No mouse showed signs of disability, behavior abnormalities, or significant changes in body weight. Comparing the control and A23-treated mice, no additional tissue damage was observed in the hearts, lungs, spleens, pancreas, kidneys, and livers. Liver fibrosis status was examined on sections stained with Masson's Trichrome, which highlighted the collagen in green. No increase in portal, perivenular, and septal fibrosis was observed in the livers (Supplementary Fig. 2). This indicated that the A23 treatment did not affect vital organs or induce tissue damage such as fibrosis in normal livers. To further investigate possible treatment toxicity, liver function (aspartate aminotransferase, alanine aminotransferase, and total bilirubin levels) and blood pictures (white blood cell count, red blood cell count, and hemoglobin level) were analyzed. There were no differences between the treatment and the control groups (Fig. 3). This further consolidated the evidence that side effects were absent after A23 treatment.
Histological examination of the tumor xenografts revealed that with A23 treatment there was a significant increase in tumor necrosis when compared with that in the control group (13% in the control group versus 42% and 20% in the 50- and 100-μg A23 treatment groups, respectively; Fig. 4A). To further delineate how A23 inhibited tumor growth, proliferation status was evaluated using Ki-67 immunostaining and the mitotic index. After A23 treatment, significant reduced number of Ki-67-positive cells denoted a decreased number of proliferating cells (173 in the control group versus 130 and 128 in the 50- and 100-μg A23 treatment groups, respectively; Fig. 4B). The data were coherent with the decreased mitotic index (72 in the control group versus 64 and 52 in 50- and 100-μg A23 treatment groups, respectively). The p44/42 MAPK in the extracellular-regulated kinase (ERK) pathway and Akt in the phosphatidyl-inositol-3-kinase (PI-3 kinase) pathway have been implicated previously in GEP-induced cell proliferation in some cancer cell types19 and in survival of neuronal cells.22, 23 Significant suppression of p44/42 MAPK and Akt phosphorylation was observed in the A23-treated HCC tumors (Fig. 5A), indicating that both the ERK and PI-3 kinase signaling pathways were essential for anti-GEP-mediated growth inhibition. Tumor growth is a balance of cell proliferation and cell death; thus, the effect of anti-GEP on apoptosis (programmed cell death) was examined by the TUNEL assay. The number of apoptotic cells between the treatment and control groups was not significantly different (Supplementary Fig. 3). Tumor angiogenesis, in the form of microvessels in the tumor, is a key determinant of tumor growth. GEP increased the number of capillaries and small blood vessels during the early wound response.21 We showed that A23 treatment significantly reduced tumor angiogenesis with decreased microvessel density highlighted by CD34 immunostaining (19 in the control group versus 6 and 5 in the 50- and 100-μg A23 treatment groups, respectively; Fig. 4C). GEP stimulates VEGF expression in breast cancer cells,25 and VEGF is an important angiogenic factor for HCC.26 Treatment with A23 significantly reduced tumor VEGF level (Fig. 5B). Thus, anti-GEP antibody has antiproliferative and antiangiogenic functions.
GEP overexpression has been reported in many types of human tumor tissue including HCC.9–10, 14 Serum GEP was detectable in HCC patients (Ho et al., manuscript submitted). Functionally, GEP was shown to stimulate proliferation of cancer cells in vitro and to alter tumorigenicity in nude mice models.14, 19 Notably, down-regulating GEP using the antisense approach could significantly reduced the tumorigenicity and tumor growth rate of HCC in a athymic nude mice model, as shown in our earlier study.14 These observations suggested that GEP is an attractive target for cancer therapy. However, the mode of gene delivery and infection/transfection efficiency of the antisense approach has limited its use as successful cancer-targeted therapy. Unlike the antisense approach, targeted therapy by antibody7, 8 such as Herceptin and anti-VEGF has higher efficacy and lower toxicity, making antibody-targeted therapy more appropriate in cancer patients. Because GEP is important in HCC progression,14 we therefore hypothesized that neutralizing the secreted GEP by a specific mAb is an effective means of HCC treatment. To our knowledge, this is the first report of use of an mAb against GEP in in vivo therapeutic study.
In this study, an established tumor of relatively large size was employed to mimic the clinical situation in which patients were diagnosed at a late stage with large tumor burden and curative treatment such as resection or liver transplantation was no longer applicable. Because anti-GEP antibody has significantly and effectively inhibited tumor growth even in established tumors, the use of A23 provides an alternative for large inoperable HCCs by inhibition of tumor growth and stabilization of the disease. A23 inhibited in a dose-dependent manner the growth of 2 HCC cell lines regardless of their endogenous GEP level as shown with the in vitro MTT assay. Nonetheless, the tumor response in the animal model was only examined in the Hep3B cells with high GEP expression, which were able to form xenografts in nude mice. Because HepG2 cells had low GEP expression, being nontumorigenic in nude mice, the tumor response after A23 treatment could not be examined. A panel of HCC cell lines were examined, and SNU475 cells revealed a low GEP level similar to that of HepG2 (unpublished data). However, the SNU475 cells were also nontumorigenic. Thus, the A23 effect in vivo could not be examined in HCC cells with low GEP levels. Nonetheless, we screened other cancer types to search for cell lines with low expression of GEP that were able to form tumors in nude mice. We observed that a nasopharyngeal carcinoma cell line, C666-1, a cancer type prevalent in our locality in the southern part of China but rare in most parts of the world, showed a low GEP level but was able to form tumors in nude mice.27 We demonstrated that the growth of C666-1 xenografts in nude mice was inhibited by A23 intraperitoneal injection (unpublished data). Therefore, A23 treatment is also effective in vivo in cancer cells with low GEP levels. This highlighted the feasibility of anti-GEP therapy in the treatment of tumors that express different endogenous GEP levels.
Conventional chemotherapy, either a single agent or in combination, showed only a limited response rate (<20%) even at high doses4, 5 and was always accompanied by severe side effects such as weight loss and cytotoxicity to noncancer cells. In contrast, we showed that A23 did not significantly affect the growth of normal liver cells in vitro and that the anti-GEP mAb treatment did not damage the normal liver as shown in the animal study. In the mouse model, anti-GEP mAb A23 exhibited no side effects affecting general health status or tissue damage to vital organs including the liver. In an independent study, A23 was shown to recognize mouse GEP (unpublished data). Thus, the lack of toxicity to the mouse should be a result of the specificity of A23 to HCC cells but should have no effect on normal liver cells. No significant weight loss or signs of physical abnormality were observed in the treatment groups. A23 would specifically bind with serum GEP and formed an immune complex (Ag-Ab complex) that would be rapidly removed from the circulation by the effective B cells in the nude mice. Severe deposition of the immune complex, which commonly would lead to glomerulonephritis,28 was not observed in the kidneys of the treated mice in the present study. In addition, the mice that received A23 treatment revealed no abnormality in the livers and spleens, which were the important organs for phagocytosis of the immune complex.29 The dose of A23 used could effectively inhibit cancer cell growth while retaining the integrity of the liver organ and was well tolerated by the animals. The current study demonstrated that anti-GEP mAb could be a safe alternative for cancer therapy. Nonetheless, the current dose of A23 employed could only delay tumor growth but did not lead to tumor regression. As the current A23 regimen was well tolerated, increases in the treatment dose and frequency and extent of the treatment period may further enhance treatment response. In a pilot study, we demonstrated that further increases in A23 dosing can further enhance the inhibitory response. Moreover, combination of anti-GEP with other targeted therapy or chemotherapy may further enhance treatment efficacy. We observed that A23 can sensitize HCC cells to chemodrugs (unpublished data). We have initiated the next phase of study to design the optimal protocol with increasing dose of A23 in combination with chemodrugs to further improve treatment response.
In addition to the use in HCC therapy, serum GEP levels provide an economical and convenient method for assessment of treatment response. Because the tumor burden was positively correlated with serum GEP level, the quantity of GEP in circulation could be determined by a simple and cost-effective ELISA with the patient serum sample. The serum profile of GEP level can therefore reflect tumor burden and thus provides an alternative for assessment of treatment response. This could greatly help to relieve the concerns of high-cost and radiological exposure resulting from repeated imaging needs.
Consistent with the in vitro results, the delayed tumor growth of Hep3B tumor by A23 treatment was caused by inhibition of cell proliferation. GEP is a growth factor that stimulates the growth of tumor cells through the MAPK pathway, and GEP also activates the Akt pathway in a number of cell types.19 To delineate the mechanism of A23-induced growth inhibition, we investigated the possible involvement of these 2 kinases. Phosphorylation of p44/42 MAPK and Akt were reduced after A23 treatment in a dose-dependent manner. The data suggested that the antiproliferative effect of anti-GEP mAb treatment resulted from A23 suppression of p44/42 MAPK and Akt phosphorylation. Furthermore, A23 treatment led to increased tumor necrosis but no significant effect on apoptosis, thus hinting at its possible effect on tumor angiogenesis. VEGF was an important factor for HCC angiogenesis, whereas histological evaluation of microvessel density provided information on the angiogenesis activity of the tumor. We showed that A23 treatment could reduce tumor VEGF level and microvessel density. The present study indicated that anti-GEP mAb reduced tumor growth through inhibition of phosphorylation of MAPK and Akt and tumor angiogenesis.
In summary, GEP is a promising therapeutic target for HCC. In a separate study, we demonstrated that serum GEP level was elevated in about 60% of HCC patients in comparison with healthy individuals (Ho et al., manuscript submitted). Thus, a blood test can identify target patients susceptible to anti-GEP treatment. In the current mouse model, the reduction in serum GEP level correlated with the reduction in tumor burden. Therefore, the same blood test can also monitor treatment response, which relieves the need to do repeat imaging with its concerns about high cost and radiological exposure. Anti-GEP inhibited tumor growth with massive necrosis through antiproliferative and antiangiogenic functions. The current study demonstrates that A23 neutralizes GEP, but whether A23 induces complement-dependent cytotoxicity will need further investigation. Notably, overexpression of GEP has also been shown in breast,30 ovarian,31 and prostate32 cancers; therefore, anti-GEP therapy may also be applicable to a broad spectrum of human cancer types.