Hepatocellular carcinoma (HCC) is regarded as a suitable target for antiangiogenic strategies. However, antiangiogenic agents aimed at single targets can be neutralized by upregulation of other proangiogenic factors. Therefore, combined approaches addressing at least two angiogenic targets should be more effective. Employing an appropriate rat hepatoma model, we examined the effects of sFlt-1 (soluble vascular endothelial growth factor [VEGF] receptor 1 as an indirect inhibitor of angiogenesis) and endostatin (a direct inhibitor of angiogenesis) in both single-agent as well as combined approaches under in vitro and in vivo conditions. Similar to human HCC, rat Morris hepatoma (MH) cells secreted high levels of VEGF, but no endogenous sFlt-1. Parental MH or MHES(r) cells, stably expressing rat endostatin, were adenovirally transduced either with AdsFlt-1 (encoding sFlt-1) or control vector Adnull (containing no transgene), followed by subcutaneous inoculation into syngeneic ACI rats. Compared with MH/Adnull cells, expressing no antiangiogenic factors at all, tumor weights were reduced fourfold in the MHES(r)/Adnull group, 19-fold in the MH/AdsFlt-1-group, and 77-fold in the MHES(r)/AdsFlt-1 combination therapy group. Analysis of variance did not show a significant interaction between the effects of the two factors ES(r) and sFlt-1; their effects multiplied. In conclusion, combined expression of sFlt-1 and endostatin effectively suppresses HCC growth under in vivo conditions. Supplementary material for this article can be found on the HEPATOLOGYwebsite (http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). (HEPATOLOGY 2005.)
Antiangiogenic approaches aim at the suppression of tumor growth via the inhibition of tumor vascularization. Since introduction of this concept in 1971,1 numerous pro- and antiangiogenic pathways and interventions have been described (for review, see Carmeliet2).
Recently, a first phase III clinical study showed that the antiangiogenic drug bevacizumab, a recombinant humanized monoclonal antibody directed against vascular endothelial growth factor (VEGF), in combination with conventional chemotherapy, significantly improved survival of patients with metastatic colorectal cancer.3, 4
However, the multitude of proangiogenic and antiangiogenic factors that can be produced by tumors suggests that inhibition of angiogenesis may require the combined action of more than just 1 antiangiogenic factor.5 In this context, the initial hypothesis, that resistance against antiangiogenic therapies would not develop,6 has been recently challenged by observations that therapeutical attempts to antagonize only one of the players in the process of tumor angiogenesis can be offset by upregulation of other proangiogenic factors.7, 8 Therefore, combination therapies simultaneously addressing 2 or more angiogenic targets should prove useful.
In most clinical situations, hepatocellular carcinoma (HCC) presents as an incurable disease. The high degree of neovascularization observed in HCC9 makes this tumor type an interesting target for antiangiogenic approaches. As in many other tumor entities, VEGF is also considered to be the most important proangiogenic factor in HCC.10, 11 Therefore, inhibition of this key regulator of HCC angiogenesis is the aim of approaches such as the use of small molecule inhibitors, monoclonal antibodies, or the so-called VEGF trap.12 Expression of a soluble form of VEGF receptor 1 (sFlt-1) as a competitive VEGF inhibitor has been shown to inhibit tumor growth in different tumor models.13, 14
Endostatin, a C-terminal 20-kd peptide cleaved from the carboxyterminus of collagen XVIII, is produced by normal and overproduced by malignant hepatocytes in HCC.15, 16 Interestingly, HCC progression was found to be associated with a decrease in the expression of collagen XVIII in HCC.17
Our findings suggest that combined use of sFlt-1 together with endostatin could be an effective antiangiogenic approach for unresectable HCC.
Morris hepatoma 3924A (MH) is a poorly differentiated and rapidly growing nodular HCC18 that can be implanted only in syngeneic ACI rats. MH cells were purchased from the German Cancer Research Center Tumor Collection, Heidelberg, Germany, and maintained in DMEM supplemented with 10% fetal calf serum in a humidified incubator at 37°C in a 5% CO2 atmosphere. For animal experiments, male ACI rats weighing 200 to 220 g (Harlan-Winkelmann, Borchen, Germany) were used. The animals were kept under conventional conditions (temperature 22° ± 2°C, relative humidity 55% ± 10%, dark–light rhythm of 12 hours) and had free access to laboratory chow and tap water. All animal experiments were performed in agreement with the laws of the German government concerning the conduct of animal experimentation. The protocol was approved by the local ethics committee for animal experimentation.
In Vitro Transduction of MH Cells With Adenoviral sFlt-1 Vector.
A replication-deficient adenoviral vector coding for the soluble form of human Flt-1 (AdsFlt-1; serotype 5 derived E1a-, partially E1b-, and partially E3-deficient vector) was constructed according to the methods described recently.13 (Further details published online at http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html.) For in vitro transduction with AdsFlt-1, MH cells were first plated in 6-well dishes. Culture medium was replaced with serum free OptiMEM1 medium (Gibco BRL, Karlsruhe, Germany) and supplemented with 0.5 mL adenovirus vector solution at 37°C for 4 hours. An extensive washing procedure was performed before cell injection into animals.
Quantification of sFlt-1 and VEGF Expression.
Expression of sFlt-1 was quantified by using the Total Soluble Vascular Endothelial Growth Factor Receptor-1 ELISA (enzyme-linked immunosorbent assay; Research Diagnostics, Flanders, NJ). For quantification of VEGF expression, the Quantikine M mouse VEGF immunoassay was employed (R&D Systems, Minneapolis, MN); this sandwich ELISA shows cross-reactivity with rat-derived VEGF.
Reverse Transcription Polymerase Chain Reaction for Detection of sFlt-1 mRNA.
RNA was isolated from tissue samples or cultured cells with the RNeasy Kit (Quiagen, Hilden, Germany). Semiquantitative reverse transcription polymerase chain reaction (RT-PCR) analysis for the detection of sFlt-1 RNA was performed by using the Qiagen OneStep RT-PCR Kit (forward primer CCAAAGCAATTCCCATGC; reverse primer GTAGCCACGAGTCAAATAGCG; annealing temperature 60°C; length of product 316 bp). For semiquantitative evaluation, GAPDH (rat) RNA was coamplified (forward primer AAAGGGTCATCATCTCCGC; reverse primer GGATGACCTTGCCCACAG; annealing temperature 60°C; length of products 315 bp).
Generation of Stably Transduced MH Cells Expressing Rat Endostatin.
The rat endostatin (ES(r)) cDNA (GenBank AC:AJ23687315) was introduced into the adenoviral recombination plasmid pTG6600 (a generous gift from Transgene S.A., Strasbourg, France), which then served as a template for the addition of the PCR-amplified E3-19 kd secretion signal of adenovirus 2 together with a modified Kozak sequence. Both primers (forward primer: GGGGCTCGAGCCACCATGAGGTACATGATTTTAGGCTTGCTCGCCCTTGCGGCAGTCTGCAGCGCGGCCCATACTCATCAGGACTTTCAG; reverse primer: GGGTCTAGACTATTTGGAGAAAGAGGTCAT) were designed to contain additional XhoI or XbaI restriction sites (underlined), which were used for transfer of the modified ES(r) coding sequence into plasmid pAdShuttleCMV (AdEasy1 System, Quantum Biotechnologies, Newington, NH). From the resulting plasmid pAdShuttleCMV-E3/ES(r) the rat endostatin gene was PCR-cloned using the forward primer TACGAATTCCCACCATGAGGTACATGATT (containing an EcoRI restriction site [underlined], the modified Kozak sequence, and the first 15 bp of the secretion signal) and the reverse primer TACGGATCCCTATTTGGAGAAAGAGGTCA (containing a BamHI restriction site [underlined]). Using the additional EcoRI and BamHI restriction sites, the transgene was then cloned into retroviral plasmid pLXGreen (Clontech, Palo Alto, CA); XL1-Blue bacteria were transformed with the resulting retroviral plasmid pLXGreen-E3/ES(r). After plasmid purification, confirmatory sequence analysis was performed. Finally, Phoenix-Eco cells were transfected with pLXGreen-E3/ES(r) via lipofection for the production of recombinant retroviral RpLXGreen-E3/ES(r) particles. Then, MH cells were transduced with RpLXGreen-E3/ES(r)-containing supernatant and polyprene containing medium (8 μg/mL) for 2 hours. Finally, G418-selected clones were checked for retroviral-mediated ES(r) expression and secretion by Western blot analysis following standard protocols. For detection of ES(r), a rabbit anti-mouse endostatin polyclonal antibody (Chemicon, Temecula, CA) showing cross-reactivity with rat endostatin was used (dilution 1:100). As secondary antibody, a horseradish peroxidase–conjugated goat anti-rabbit immunoglobulin G (H+L) (Bio Rad, Hercules, CA) was used (dilution 1:4.000). One stably rat ES expressing cell line was selected and termed MHES(r).
Comparison of Growth Characteristics of MH, sFlt-1 Transduced MH, and MHES(r) Cell Lines In Vitro.
For the exclusion of a direct negative influence of the sFlt-1 transgene and the ES(r) transgene on the growth of MH cells, the in vitro growth kinetics were analyzed. To compare the growth of cell lines MH (parental) and MHES(r), 105 cells per well were seeded in 6-well plates. Medium was changed daily. From day 1 to day 5, cell numbers from 6 wells per cell line were counted in a Neubauer chamber. To rule out a negative influence of sFlt-1 on the MH cell growth, a similar experiment was done with two sets of MH (parental) cells. One day after seeding, 1 set was transduced with AdsFlt-1 (MOI 50), whereas the other set was exposed to the control vector Adnull (MOI 50) for 4 hours. Again, cell numbers from 6 wells per set were counted in a Neubauer chamber from day 1 to day 5.
The in vitro antiangiogenic efficacy of sFlt-1 derived from our expression systems was determined by using a spheroid assay.19 Confluent monolayers of human umbilical vein endothelial cells (HUVECs; PromoCell, Heidelberg, Germany) were harvested, resuspended in a 20% methocel cellulose solution (Sigma-Aldrich, München, Germany), and transferred into 96-well plates. After incubation for 24 hours at 36°C, 1 spheroid per well had formed. A concentration of 30 μg/mL VEGF was used as a standard proangiogenic stimulus. After incubation with the cell culture supernatants of interest, the number and length of the sprouts were counted microscopically (phase contrast, ×100).
For animal experiments, cell lines MH and MHES(r) were transduced at MOI 50 either with AdsFlt-1 or a control vector containing no transgene (Adnull). After 4 hours, cells were washed and 6 × 106 cells in 70 μL per ACI rat (n = 6 per group) injected subcutaneously on the back. At day 11, animals were sacrificed and tumors explanted and weighed. Subsequently, tumor samples were processed for (1) cryoconservation, (2) semiquantitative RT-PCR analysis, and (3) recultivation of tumor-passaged MH/MHES(r) cells.
Endothelial cells were stained for expression of alkaline phosphatase (AP).20, 21 After embedding in Tissue Tek (O.C.T. Compound, Sakura Finetek Europe B.V., Zoeterwoude, the Netherlands), 6-μm cryosections were cut, fixed in acetone, stained with fast red and naphtolphosphate (both Sigma-Aldrich), and counterstained with hematoxylin.
The spheroid assay was analyzed by Dunnett's multiple comparison of the treated group with three control groups, to adjust the significance level in a manner that kept it at .05 for all three comparisons together. Square roots of cumulative sprout lengths were used. Results were transformed back for plotting medians and their 95% confidence intervals (CI). The in vitro cell growth data are presented as exponential growth curves of MH and MHES(r) cell lines, which were compared in an analysis of covariance of logarithms. Cell growth for MH after transduction with Adnull and AdsFlt-1 was modeled as 3-parameter logistic growth curves with different times of half-maximal cell numbers (ED50) and growth rates (g) and common maximal cell number and was fitted by minimizing the squared differences of logarithms. In vivo tumor growth was investigated by analysis of variance with factors (1) cell line, (2) treatment, and (3) their interaction. Natural logarithms were analyzed as they were the optimal Box-Cox transformation. Results were transformed back to yield geometric means, their CIs, and size ratios. In all 3 analyses, the assumptions of normal-distributed residuals with constant variance were verified by residual-by-predicted plots and normal-quantile plots. When P values were less than .05, tests results were called significant.
Angiogenic Features of Rat Morris Hepatoma Cells.
Naïve MH cells were found to secrete high levels of VEGF (up to 1,800 pg/mL in the supernatant); in contrast, only baseline levels were measured for cell culture medium before MH cell culturing. Hence, MH rat cells show a high innate expression of VEGF comparable to human hepatoma cells.
Next, MH cells were checked for endogenous expression of soluble forms of the VEGF receptor 1 (sFlt-1), which functions as a potent soluble interceptor for VEGF and therefore might interfere with sFlt-1-mediated therapeutic approaches. Employing a commercially available ELISA, MH cells did not exhibit any innate expression (e.g., generated by differential splicing) of the rat Flt-1 gene, whereas transduction of MH cells with AdsFlt-1 resulted in high sFlt-1 expression levels in the supernatant (up to 75 ng/mL). Because a maximum of MH-based sFlt-1 expression could be achieved at MOIs ranging between 50 and 100, all further AdsFlt-1-mediated transductions of MH cells were carried out at these optimized MOIs.
We also analyzed AdsFlt-1-mediated interception of VEGF intrinsically produced by MH rat hepatoma cells by ELISA. Complete trapping of endogenously expressed rat VEGF by adenovirally produced soluble VEGF receptors was found. For in vitro determination of adenoviral mediated sFlt-1 antiangiogenic activity, a spheroid assay was performed. Spheroids of HUVEC were stimulated with 30 μg/mL VEGF. Whereas addition of supernatant of MH cells transduced with control virus Adnull did not impair sprouting (Fig. 1A, left panel), supernatant from MH cells transduced with AdsFlt-1 caused complete inhibition of sprouting (Fig. 1A, right panel). Likewise, quantitative analysis of the mean cumulative length of sprouts demonstrated sprouts shortened to baseline lengths when spheroids were treated with supernatant from MH cells transduced with AdsFlt-1 (Fig. 1B). In contrast, supernatants from MH cells transduced with control vector Adnull or from naïve MH cells without virus treatment did not influence sprouting achieved under VEGF stimulation (Fig. 1B). All three comparisons were significant at the multiple level .05. Therefore, sFlt expressed by MH hepatoma cells efficiently intercepted VEGF, leading to complete abrogation of endothelial cell sprouting in vitro.
For high-level expression of endostatin, MH cells were transduced with rat endostatin encoding retroviral vector RpLXGreen-E3/ES(r), leading to the generation of the stable cell line MHES(r). As shown by Western blot analysis, this cell line expresses high levels of rat endostatin (Fig. 2B, lane 1), whereas no ES(r) expression could be detected for parental MH cells (Fig. 2B, lane 2). To exclude a negative influence of the ES(r) transgene on the cellular growth of the retrovirally generated MHES(r) cells, in vitro growth of both cell lines was compared. No significant growth differences between MH and MHES(r) cells were found (Fig. 2A). To exclude a negative influence of the sFlt-1 transgene, cellular growth of MH cells being transduced with control vector Adnull or AdsFlt-1 was compared. No significant growth differences between Adnull and AdsFlt-1 transduced cells were found (Fig. 2C).
In Vivo Antiangiogenic Effects of sFlt-1 and Endostatin.
For animal experiments, cell lines MH and MHES(r) were transduced at MOI 50 either with AdsFlt-1 or control vector Adnull. Next, 6 × 106 cells were injected subcutaneously in ACI rats (n = 6 per group). At day 11, animals were killed and tumors explanted and weighed. Macroscopic analysis (Fig. 3, upper panel) showed profound differences in tumor volumes, depending on the antiangiogenic factors being expressed (sFlt-1 and ES[r] as single agents or combined). Control tumors that expressed neither sFlt-1 nor ES(r) exhibited the biggest tumor sizes, whereas single-agent expression of ES(r) or sFlt-1 led to clearly visible reductions in the tumor burden of the animals (Fig. 3A-C). Combined treatment with sFlt-1 and ES(r) resulted in a nearly complete abrogation of tumor growth (Fig. 3D). Compared with MH naïve/Adnull transduced cells, expressing no antiangiogenic substances at all, the median tumor weights were reduced by a factor of 4.1 (CI 2.6–6.6) in the MHES(r)/Adnull group, by a factor of 18.7 (CI 11.7–29.8) in the MH naïve/AdsFlt-1 group, and by a factor of 77 (CI 40–149) in the MHES(r)/AdsFlt-1 combination therapy group (Fig. 3, lower panel). In an analysis of variance of logarithms of tumor weights, no significant interaction between the two factors sFlt-1 and ES(r) was found.
Histological examination showed maximal suppression of vascularization in the tumors treated with sFlt-1/ES(r) combination therapy (Fig. 4B; avascular tumors of pinhead size), whereas a highly vascularized tumor status was found in therapy-naïve MH tumors (Fig. 4A).
Finally, a detailed expression analysis of antiangiogenic transgenes sFlt-1 and ES(r) under in vivo MH tumor passage was performed (Fig. 5). In supernatants from cells recultivated from tumors explanted on day 11, expression and secretion of ES(r) could be demonstrated by Western blot analysis (Fig. 5A, lane 2). Similarly, by RT-PCR, expression of the sFlt-1 transgene could be detected before tumor implantation (day 0, Fig. 5B) and after tumor explantation (day 11, Fig. 5C). Therefore, the observed reductions in tumor sizes appear to be attributable to continuous expression of both sFlt-1 and ES(r) under MH in vivo cell passage.
Taken together, combined antiangiogenic gene therapy with sFlt-1 and ES(r) was found to be highly effective in our rat MH model of HCC.
Hepatocellular carcinoma is still an uncurable disease in most patients.22 New treatment strategies targeting the HCC tumor vasculature, thereby avoiding direct damage to the nontumorous liver parenchyma, are very attractive. The recent success of antiangiogenic therapy in the treatment of metastatic colorectal carcinoma4 stimulated clinical studies on HCC using the monoclonal antibody bevacizumab (Avastin) alone23 (http://www.clinicaltrials.gov/ct/show/NCT00055692) or in combination with transcatheter arterial chemoembolization (TACE) (http://www.clinicaltrials.gov/ct/show/NCT00049322). However, preclinical data concerning the efficacy of such approaches are still sparse.24 To avoid the pitfalls, potential inefficiencies, and side effects of repeated administrations of antiangiogenic substances, gene therapeutic approaches are potentially useful.25 However, such preclinical antiangiogenic gene therapeutic approach has to rely on a suitable animal model, matching as closely as possible with the angiogenic features present in human HCCs.
In our study, we therefore first characterized important angiogenic features of rat Morris hepatoma cells, which can be implanted both subcutaneously and orthotopically in syngeneic ACI rats,26 leading to hepatomas that display a constant tumor growth rate and absence of metastases. Similar to human HCC, MH cells were found to secrete high levels of VEGF, but not to express any endogenous sFlt-1. Furthermore, MH cells expressed high levels of sFlt-1 in the course of transduction with adenoviral vector AdsFlt-1. To investigate the biological effect of sFlt-1 in vitro, we chose the well-established spheroid assay. In this assay, endothelial cells of human origin (HUVEC) are used. A well-known heterogeneity of endothelial cells exists, which depends on the species and organ of origination.27, 28 Even in the same organ or tumor, endothelial cells of different phenotypes are found.29 Therapeutically expressed sFlt-1 was able to function as a highly potent interceptor of VEGF naturally produced by MH rat hepatoma cells and exerted a strong inhibitory effect in the HUVEC spheroid sprouting assay. The use of HUVEC for in vitro testing allowed only a first assessment of an antiangiogenic effect and therefore had to be confirmed by in vivo testing.
Other studies have shown that AdsFlt-1 presents the best antiangiogenic effect when applied locally.13 In contrast, intravenous application of AdsFlt-1 into ACI rats leads to only transient systemic expression of sFlt-1 (our unpublished data) and was shown to result in liver toxicity.30 To reach high local concentrations of sFlt-1 at an early stage of HCC development and angiogenesis, we therefore transduced MH hepatoma cells before implantation into the animals. Of interest, a comparable approach transfecting angiostatin DNA into mouse HCC cells before implantation recently has been reported to be successful.31 In our model, high-level adenoviral mediated expression of sFlt-1 was proven by semiquantitative RT-PCR analysis performed before subcutaneous implantation and after tumor explantation (day 11). Thus, long-lasting sFlt-1 expression was ensured.
We chose endostatin as a partner for our combined antiangiogenic approach. This strategy was supported by recent observations of a positive correlation between angiogenic scores found in hepatoma tissues and a combined positivity for VEGF and negativity for endostatin in serum of HCC patients.32 However, we have no clear picture concerning how endostatin exerts its antiangiogenic effect. A recent gene expression analysis study33 indicated a complex pattern of gene expression that likely switches the balance toward inhibition of angiogenesis.34 To make our model as consistent as possible concerning the effects of endostatin, we chose to express rat endostatin (ES[r]), which shows 77% and 89% homology with human and mouse endostatin, respectively. Thus, we hypothesized that interception of VEGF together with therapeutic overexpresssion of endostatin should result in a stronger downregulation of angiogenesis than overexpression of the single factors.
The overwhelming results achieved with the delivery of recombinant endostatin in different mouse tumor models by O′Reilly et al.35 raised much interest in this candidate molecule36 for antiangiogenic therapy, but they have been challenged by others.37 A few attempts to treat HCC with endostatin have been undertaken: Sun et al.38, 39 were able to inhibit growth and recurrence of HCC xenografts in mice by administration of recombinant endostatin. Schmitz et al.40 applied a murine endostatin expressing adenovirus in a human Huh7 hepatoma mouse xenotransplant model and found retardation of tumor growth, which, however, did not reach significance.40 Li et al.41 used a human endostatin expressing adenoviral vector, which was applied into a human BEL-7402 hepatoma mouse xenotransplant model, reaching a significant fourfold growth inhibition of tumors. RT-PCR for human endostatin RNA was positive 3 days after administration of the virus, but already negative at day 7,41 indicating cross-species difficulties in long-term efficacy of antiangiogenic transgenes.
The observation of Abdollahi et al.33 that endostatin interferes with VEGF signaling by downregulating and dephosphorylating VEGFR-2 indicates that a combined therapy directly addressing the VEGF system (by the expression of sFlt-1) and indirectly by the action of endostatin could act in a superadditive effect. We therefore performed a 2×2 analysis of variance on the data of the animal experiment. However, no evidence for an interaction of the two principles was found: their effects multiply. That means that both factors impair tumor growth independently and points to endostatin actions targeting proangiogenic pathways other than those elicited by VEGF. Moreover, the absence of an interaction between the two factors ES(r) and sFlt-1 indicates that there is no relevant upregulation of angiogenic escape pathways when only one of both antiangiogenic substances is expressed. Our ES(r)/sFlt-1 combined therapy led to minimal avascular residual tumors of nearly pinhead size, similar to the “dormancy state” described as the theoretical endpoint of antiangiogenic tumor therapy.42
In terms of clinical perspectives, the particular anatomical conditions of HCC that allow local access by means of interventional radiology are suited for regional application of antiangiogenic vectors with or without transcatheter arterial chemotherapy or chemoembolization protocols. Importantly, tumor hypoxia subsequent to interventions by TACE protocols is known to induce VEGF expression. High levels of VEGF, which correlate with a poor prognosis after TACE,43 therefore could be antagonized by a concomitant antiangiogenic (gene) therapy. Furthermore, antiangiogenic therapy also could be useful in the prevention of HCC recurrence after surgery.
In summary, taken together, our data provide evidence that angiostatic gene therapy may form a feasible strategy for the treatment of established hepatocellular carcinomas.
The authors thank T. Korff and H. Augustin, Tumor Biology Center Freiburg (TBCF), Freiburg, Germany, who introduced us in the performance of the spheroid assay.