The application of antitumor drug-targeting models on liver cancer

Abstract

Hepatocarcinoma animal models, such as the induced tumor model, transplanted tumor model, gene animal model, are significant experimental tools for the evaluation of targeting drug delivery system as well as the pre-clinical studies of liver cancer. The application of antitumor drug-targeting models not only furnishes similar biological characteristics to human liver cancer but also offers guarantee of pharmacokinetic indicators of the liver-targeting preparations. In this article, we have reviewed some kinds of antitumor drug-targeting models of hepatoma and speculated that the research on this field would be capable of attaining a deeper level and expecting a superior achievement in the future.

• Animal models
• antitumor drugs
• gene knock-out
• liver cancer
• targeting agents

Introduction

With the development of bioscience, biomaterials and molecular biology, the investigations of targeting drug delivery system (TDDS) have been turned into research focus. Among them, great importance should be attached to the research of TDDS in the application of anti-cancer area (Nasongkla et al., 2006; Liu et al., 2007; Peer et al., 2007; Brannon-Peppas et al., 2012). In the traditional treatment of cancer by chemotherapy drugs, there are various adverse effects such as the tendency of in-vivo drug degradation and diffusion after regional administration, which could cause a relatively average tissue distribution and a certain amount of damage on the entire body (Murali et al., 2015). Furthermore, free drug could not form an effective antitumor drug concentration in the tumor region. On the contrary, the TDDS refers to the drug carrier that selectively concentrates on the medicine target area (Sudimack & Lee, 2000), which could effectively deliver antitumor drugs to the target site and simultaneously reduce the dosage as well as the possibility of side effects. Especially in liver cancer therapy, preliminary studies have suggested that the use of targeting agents possess many desirable properties, such as good specificity and selectivity, the capacity of reducing the drug dosage and frequency, as well as decreasing the cell toxicity and improving the drug efficacy (Seymour et al., 2002; Brillet et al., 2005).

In recent decades, valid approaches have been exploiting for treating cancer and the rapid development of targeting agents has brought a breeze of fresh air in the tumor therapy (Semenza, 2003; Thorpe, 2004; Weigelt et al., 2010; Gao et al., 2014). During the study of liver-targeting agents, animal models supply significant methods on the evaluation of the anti-hepatoma TDDS (Ko et al., 2001; Qian et al., 2007; Kota et al., 2009). The study of animal models of liver cancer provides a powerful means of the pharmaco-dynamics and pharmacokinetics experiments of targeting agents (Ho & Chien, 2014).

Development of anti-tumor models

About 90% amount of cancer pre-clinical studies are by the assistance of a variety of animal models (Kapadia et al., 1998; Kaplan-Lefko et al., 2003; Van der Spoel et al., 2011), including the spontaneous tumor models (Rosol et al., 2003), induced tumor models (Talmadge et al., 2007), transplanted tumor models (Peterson et al., 1994), gene animal models, and so on (Van Dyke & Jacks, 2002; Khanna & Hunter, 2005).

Tumor animal models originally appeared in the 18th century, when researchers established homologous transplantation models by transplanted homograft tumor cells into animals and first performed the artificial manipulation of tumor (Waxman & Anderson, 2001). This appearance marked the onset of the experimental oncology and made up the defects of spontaneous tumor models, namely, it was difficult to collect abundant research materials on oncology in a very short time due to different incidences of tumors (Babbs, 1990). The study of induced tumor models made it possible to better understand the etiology of tumor. In 1915, Japanese scientist Katsusaburo Yamagiwa continuously painted crude coal tar on the inner surface of 137 rabbits’ ears in every 2 days; 1 year later, seven rabbits appeared invasive tumors on the tar-covered region (Fujiki, 2014). This is the first time that human had ever produced cancer successfully metastasis was rarely happened even been born the subcutaneous tumor for a long period. Subsequently, various sorts of chemical substances were used to establish induced tumor models, meanwhile, lay the foundation of chemical carcinogenesis theory (Mahfoozur et al., 2015).

In 1969, Rygaard & Povlsen successfully transplanted xenograft human malignant tumor cells into nude mice, discovering that the biological characteristics of the created animal models accorded better with the human body. Until today, using human cells or primary tumor tissues to produce xenograft models have become the main method of current in-vivo tumor research (Rygaard & Povlsen, 1969).

In 1984, Brinster & Palmiter first transduced SV40’s T antigen (Tag) oncogene into mice fertilized eggs, resulting the transgenic mice be enabled to form brain choroid plexus papilloma, which turned to be the first genetically engineered mouse tumor model (Brinster & Palmiter, 1984). Thereafter, the technology of gene knock-out and gene insertion was also popularly applied to cancer researches. Genetically engineered animal models not only display a revolutionary position in fundamental researches, such as the function of genes, the synergies between genes, the mechanism of tumor formation and the law of tumor development, but also play an innovation role in practical researches like the prevention and treatment of cancer (Hansen & Khanna, 2004; Sharpless & DePinho, 2006). At the end of 1990s, the successful construction of bcr-abl gene-deficient mice prompted the birth of the first molecularly targeting drug, imatinib (Zheng et al., 2008; De Carvalho et al., 2010); and then, an increasing number of molecular-targeting agents was approved and available after the year of 2000 (Tsuruo et al., 2003; Huang et al., 2004; Ferrari et al., 2005; Weinstein & Joe, 2006).

At present, the studies of tumor animal models have been improved and progressed constantly. As can be seen from the development law of tumor animal models during the past years, the purpose of the studies focus more on the clinical research, the content has been changed from mechanism to application, the methodology has been turned from passive observation into active manipulation, and the practical applications were grown stronger.

Targeting models of liver cancer

Induced animal models of liver cancer

Induced liver cancer model refers to that, under experimental conditions, using chemical, physical or biological carcinogens onto animals to induce liver canceration (Qian et al., 2007), which is a frequently used method in experimental oncology research (Kim et al., 1991; Dawson et al., 2002). Of which, chemical carcinogens are the most common, such as nitrosamines (DEN) (Williams, 1980), aflatoxins (AFB1) (Ghoshal & Farber, 1984), amino azo fuels (3'-Me-DAB), aromatic amines (2-AAF) and so on (Naugler et al., 2007). The induction can be completed by a single sort of chemical drugs or by combinations of multiple drugs with different tumor promoters and liver resections in a certain sequence to form diverse comprehensive-induced liver cancer schemes (Fisher et al., 1978; Sun, 1990).

As far as the present state of study, this sort of model is mainly applied for the research of the etiology, embryology, pathogenesis and the genetic and biological characteristics of liver cancer (Toyokuni, 1999). Silveira et al. (2001) constantly gave DEN (200 mg/kg) to the male Wistar rats (50–60 g) by gavage for 2 weeks, and then switched to 2-AFF dissolved in dimethyl sulfoxide and corn oil at a dose of 20 mg/kg for 4 days. After 24 h of the last feed of 2-AFF, 70% liver was resected, then continuously gave 2-AFF 20 mg/kg for 2–4 days. 10 months later, the animal model was established successfully.

Current studies of this kind of animal model remain inconsistent perspectives of the choice of mice gender (Antonio et al., 2015). It is generally believed that, compared to female mice, male mice have higher success rate in cancer induction (Nakatani et al., 2001). However, others insist that the choice should be female mice (Ornstein et al., 2000). The features of induced liver cancer models denominate the similarities in terms of characteristics and processes of human liver cancer (Gerin, 1990; Yang et al., 2014). Generally, the whole carcinogenesis process is divided into four stages: the activation of liver cells, the proliferation of liver cells, the hyperplasia nodules of liver cells and the hepatocellular carcinoma (Sell & Leffert, 1982). In addition, such model also showed its defects in practice including long inducing period (3–5 months, even 1–2 years) and individual differences in aspects of time, location, lesions number of the occurrence of hepatoma.

The induced animal models of liver cancer have been used in the research of targeting agents. Uehara et al. (2002) allowed Wistar rats fed with 0.01% DEN for 12 weeks and then started to give liposomes OK-432 (OK-lipo) and non-liposomal (Non-OK) at the 5th and 9th week, respectively. At the 13th week, five rats were treated per group to compare the immune parameters (NK and IFN-γ), while the remaining rats were given water to observe their survival time. Results indicated that the two groups given OK-lipo were superior to the Non-OK group in life span and the immune biochemistry aspects (Figure 1).

Figure 1. Survival curve of rats with chemically induced hepatocellular carcinoma with or without intravenous administration of OK-Lipo 10 KE. Survival time of the OK-5w group (n = 7) was significantly longer than any other group (p < 0.01; log-rank test).

Homograft animal models of liver cancer

Transplanted hepatic cancer model implies that this type of tumor strains is transplantable, serially grow and passage in inbred animals and the same species, possesses more stable histological types and growth characteristics. Besides, the invasion, metastasis and the sensitivity to chemotherapeutic drugs of such model have not been identified yet (Feray et al., 1990). The features of transplanted liver cancer models refer to the simple model reproduction, the conform tumor’s growth rate, the tiny individual differences, the similar effects on the host and the high-survival rate of inoculation, and it is easier to estimate the curative efficacy of targeting agents objectively; in addition, tumor cell lines could be saved in liquid nitrogen for long-term experimental usage (Freise et al., 1999). Nevertheless, the shortcomings of transplanted models mainly represent the differences with human tumors on growth characteristics, such as the higher growth speed and proliferation rate, but shorter volume doubling time than human cancer cells. In addition, there is no single tumor strain that is sensitive to all anticancer drugs (Sakai et al., 2005). Generally, the transplanted liver cancer models are irreplaceable in the research of applications of targeting preparations.

At present, the H22 tumor model is the most commonly used mice transplanted liver model (Yao et al., 2006; Bishayee et al., 2010). This model has been generally utilized in the screening of liver-targeting agents, and also has been a well-acknowledged model in the study of the curative effect and mechanism of anti-tumor traditional Chinese medicine (Wiesner et al., 2001). The transplantation of H22 hepatoma cells could be operated in three ways: the ascites passage, hypodermic inoculation and the orthotopic liver transplantation. These three methods have been applied to the screening of the antitumor-targeting agents of hepatoma (Tian et al., 2012). According to previous reports (Tardi et al., 2000; Peer & Margalit, 2004), Yuan et al. adjusted the concentration of H22 cell to 1 × 10⁷/mL, given each BALB/c mouse 0.1 mL of the cell suspension by intraperitoneal injection to establish ascites model, then administrated respectively and observed the changes of survival time, body weight and peritoneal permeability of mice. Results showed that 100 mg/kg of quercetin nanoliposomes could prolonged the survival time of liver cancer mice, decreased the peritoneal permeability, inhibited the carcinomatous ascites and accumulated in tumor tissues effectively; hence, it was expected to be a kind of promising antitumor drug (Ren et al., 2003).

As for the sites where liver tumor transplanted, there are still no unified comprehensions. The intrahepatic tumor reflects the malignant behaviors and the growing status of human primary infiltration (Yang et al., 1992), and the transplanted tumor obtained shows much closer biological characteristics to the clinical liver carcinoma (Yao et al., 2001). Moreover, neither the subcutaneous transplanted tumor nor the ascites hepatoma could possess the above-mentioned properties (Makino, 1956).

The reported experimental results have demonstrated that the metastasis was rarely happened even been born the subcutaneous tumor for a long period (more than 70 days), mainly due to the invasion and low-metastasis rate caused by the poor blood supply and the lymphatic drainage inside the tumor (Wu et al., 2013). The blood supply condition is a key factor that affects the biological behavior of the tumor. Therefore, the orthotopic model of liver cancer is essential for verifying the pharmacodynamics of antitumor-targeting agents (Blakey et al., 2002). Lu et al. (2006) administrated in Kunming mice with 2.88 × 10⁸/mL ascitic hepatoma H22 cells injected under the liver capsule at a dose of 0.05 mL per mouse, then closed abdominal cavity. Mice were previously abdominal anesthetized by pentobarbital natrium. After grouping, mice were respectively delivered DHAQ-PBCA-NS and DHAQ for three times at the 1st, 5th and 9th day after inoculation. The result of the experiment indicated that the chemotherapeutic index of DHAQ-PBCA-NS was 3.42 times higher than the level of DHAQ (Table 1). From this pharmacodynamics experiment, it can be drawn that the agents presented obvious hepatic-targeting and tumor inhibiting efficiency.

Table 1. Relationship of dose-effect of DHAQ and DHAAQ-PBCA-NS on liver tumor H22.

Drug	Dose (mg/kg)	Tumor weight (g)	RTI (%)
Normal saline	1.80	1.002 ± 0.318	0
DHAQ-PBCA-NS	1.80	0.094 ± 0.038	90.62
	1.08	0.210 ± 0.140	79.04
	0.65	0.447 ± 0.357	55.39
	0.39	0.486 ± 0.142	53.49
	0.23	0.555 ± 0.178	44.61
DHAQ	1.80	0.368 ± 0.111	63.27
	1.08	0.495 ± 0.178	50.60
	0.65	0.623 ± 0.360	37.82
	0.39	0.719 ± 0.293	28.24
	0.23	0.194 ± 0.152	8.78

Xenograft models of human hepatoma

The ultimate purpose of the establishment of xenotransplanted tumor models is to provide an in-vivo method in order to directly explore the biological characteristics and pathogenesis of human tumor (Goldstein et al., 1995). The orthotopic transplantation tumor models of human hepatoma not only maintain the structure of the primary tumor, but also reserve the most biological characteristics of human tumors, especially the property of metastasis, which enables tumors to present similar malignant behavior in the host body to that in patient’s body (Kubota, 1994). This feature is particularly supportive of the targeting agents’ study (Furukawa et al., 1993). At present, nude mice are the best models of human liver cancer (Flatmark et al., 2004). The complete human hepatoma tissues of orthotopically implanted tumors in nude mice retain the structural characteristics of human hepatoma tissues, as well as the potential of invasion and metastasis. Thus, this modeling approach is most close to the biological evolution and characteristics of human liver cancer let alone it takes less time to establish, but with a higher success rate. Therefore, it is a kind of desirable animal model of the study on anti-hepatoma-targeting agents (Kukowska-Latallo et al., 2005). On the other side, xenograft models of human liver cancer have to face following disadvantage factors: nude mice are incapable of performing the immune-related anti-tumor metastasis research for the lack of T-cell immunity function. In addition, nude mice are expensive, difficult to rear and uncommonly used (Cai et al., 2005). As a result, using orthotopic nude models to validate targeting agents is rarely reported (Leenders et al., 2008).

Nude mice models are typically divided into subcutaneous transplantation model and orthotopic transplantation model. Hasegawa et al. (2001) and Hirano et al. (2001) adjusted the concentration of HuH7 (human hepatocellular carcinoma) to 5 × 10⁶/mL and subcutaneously injected it into male SCID mice, and 3 weeks later, consecutively injected mice with HVJ liposomes for 5 days (Figure 2). As a consequence, HVJ liposomes acted pretty solid improvement on the body weight and pathological indicators of nude mice. Moreover, researchers subcutaneously inoculated human hepatoma cells into BALB/c nude mice to investigate targeted ultrasound contrast agents, and the outcome demonstrated that the agents are able to specifically enhance the targeting efficiency of hepatocellular carcinoma (Hamilton et al., 2002; Leong-Poi et al., 2003).

Figure 2. Scheme of the in-vivo study design. (A) Single-injection model. (B) Repeat injection model. About 3 weeks after the tumor cells injection, HVJ liposomes solution was injected once a week. GCV was injected i.p. once a day for 5 days from the day after each injection of HVJ liposome solution.

Gene animal models

With the increasingly deep acknowledgement of the cell biology characteristics and the progression of genetic engineering, the research of animal methods had improved greatly (Ellinwood et al., 2005). However, the above-mentioned tumor models cannot meet the demand of the comprehensive study on liver cancer. As a consequence, the gene model emerged at the right moment (Lampson et al., 2001). Currently used gene models are mainly in the two patterns, one is transgenic tumor model and the other is knock-out tumor model.

Transgenic tumor models

Transgenic tumor animal model is established by virtue of the genetic engineering techniques, transferring certain tumor exogenous genes into the host chromosomes through germ cells or early embryonic stem cells, stably integrating the exogenous tumor genes in genomes and being passed onto future generations (Guy et al., 1992; Gingrich et al., 1996). The essential reason that leaded normal cells to display malignant lesions is the gene mutation (Stephen et al., 2015). Exploring the formation, development and evolution of tumors at the gene level has created a new era of cancer research (Adams & Cory, 1991). The emergence of some hi-techs, like molecular cloning technology and micro-injection technique, paved the way for the establishment of transgenic tumor models. Mouse is the most commonly used animal species in transgenic experiment (Sharpless & DePinho, 2006). Since the location, structure and function of oncogene and proto-oncogene are still unclear by far and require continuous exploration, thus, the success rate of established transgenic animals is remaining to be low (Hogan & Lyons, 1988).

Knock-out tumor models

Based on the principle of homologous recombination, the knock-out technology transfers the exogenous DNA into cells for producing a recombination with the specific homologous sequence on endogenous chromosomes, in order to modify the genome (Jackson et al., 1995; Serra et al., 2004). Although the models’ incidence is a bit low and there are obstacles remained to be solved in the practical operation (Zhang et al., 2004), this kind of model still shows us the promising future (Matalon et al., 2000; Asahi et al., 2001).

The establishment of transgenic and knock-out models is on the basis of the combination of in-vitro and in-vivo studies (Ristevski, 2005). While operating on the molecular and cellular levels, and generating effects at the overall animal level, enable us to more completely investigate the mechanism of tumor development, the immunological relationship between tumor cells and the organism, as well as the evolution and apoptosis of tumor (Keffer et al., 1991).

In current studies, there are mainly two ways to produce knock-out mice: gene targeting and gene trapping (Sundaresan et al., 1995; Leighton et al., 2008). The gene-targeting technology, including gene knock-out and knock-in (Kuhn et al., 1995), is on account of the success of embryonic stem cell culture and the development of in-vitro homologous recombination, so as to discover the function deficits of mouse genomes caused by gene mutation (Capecchi, 1989). In 1987, the world’s first mice gene-targeting experiment was successfully completed (Thomas & Capecchi, 1987; Mansour, 1988). In the past two decades, gene-targeting technology has accelerated the research of gene function, leading to three Nobel Prize winners in 2007 for contributing this method (Mak, 2007) (Figure 3). Gene trapping technology was formed as an alternative plan of gene targeting to induce the mutation (Abuin et al., 2007), with the properties of high-throughput and random mutation (Kothary et al., 1988; Gossler et al., 1989). This method is not as specific as gene targeting. However, it takes less time to knock out larger amounts of genes (Takeuchi, 1997; Zambrowicz & Friedrich, 1998). The combination of gene targeting and gene trapping is expected to knockout all of the genes in mice.

Figure 3. A flowchart for the large-scale acquisition storage and cataloguing of gene traps in mouse ES cell lines.

Morris et al. (2014) assessed the effect of the TGF-β signaling pathway on liver tumors induced by phosphatase and tensin homolog (Pten) loss. Interestingly, deletion of both Pten and Tgfbr2 (Pten^LKO;Tgfbr2^LKO) in the mouse liver resulted in a dramatic shift in tumor type to predominantly CC. However, phosphorylation of p70 S6 kinase was observed in the liver of all three phenotypes indicating that the loss of Tgfbr2 and/or Pten leads to an increase in this signaling pathway. Furthermore, Scf and EpCam expressions were also increased in the double knock-out mice. These results suggested that the alteration in tumor types between the Pten^LKO mice and Pten^LKO;Tgfbr2^LKO mice is secondary to the altered regulation of stem-cell features induced by the loss of TGF-β signaling.

Zheng et al. (1996) determined whether the mRNA expression of the proto-oncogene c-jun as well as the tumor suppressor gene p53 is increased by Cd in the target organ for Cd toxicity, namely, the liver. The effect of CdCl₂ on the mRNA levels of c-jun and p53 was studied in livers of C57BL/6J (control) and MT-null mice by northern- and slot-blot analyses. The experimental data demonstrate that Cd induces the mRNA levels of c-jun and p53 in liver of mice, and MT-null mice are more sensitive to this effect; the Cd-induced increases in proto-oncogene/tumor suppressor gene expression may be mechanistically important for its carcinogenicity and hepatotoxicity (Figure 4).

Figure 4. Liver Cd concentration (top) and serum alanine amino transferase levels (bottom) 12 h after Cd administration in C57BL/6J control and MT-null mice.

Kennedy et al. (2014) employed a triple knock-out (TKO) mouse model with null alleles at the loci encoding the three relevant receptors for tumor necrosis factors α and β and IL-1α and IL-1β (i.e. null alleles at the Tnfrsf1a, Tnfrsf1b and Il-1r1 loci). The observation that TKO mice were resistant to the tumor promoting effects of dioxin in liver suggests that inflammatory cytokines play an important step in dioxin-mediated liver tumor promotion in the mouse (Figure 5). The results support that the mechanism of dioxin acute hepatotoxicity and its activity as a promoter in a mouse two stage liver cancer model may be similar.

Figure 5. Decrease of dioxin-induced tumor multiplicity in the TNF/IL-1receptors triple null mice. (A) Dioxin-induced liver tumor multiplicity in the DEN-initiated WT and TKO mice. (B) Levels of 7-methylguanine in liver DNA of WT and triple-null (TKO) mice.

Chen et al. (2014) established PIWIL2-specific transcript mono-allele and bi-allele knockout HepG2 cell lines using transcription activator-like effector nuclease (TALEN) technology. It was found using the cell line models that specific transcript knockdown of full length PIWIL2 could suppress cell proliferation (Figure 6), while ectopic expression of PIWIL2 could enhance proliferation of HepG2 via suppressing the TGF-β pathway. Taken together, the study revealed critical negative regulation of TGF-β signaling by PIWIL2 in HepG2 tumor cells and provided an effective strategy to study specific gene transcript functions in cells.

Figure 6. Loss of PIWIL2-reduced HepG2 cell proliferation. Cell proliferation levels of normal HepG2, PIWIL2+/− and PIWIL2 −/− cell lines. Clones #7 and #8 represent the two PIWIL2 knockout cell lines mentioned earlier.

Prospects

Selecting and researching a preferable animal model are the crux in evaluating the targeting agents and the pharmacodynamics of preparations. Although subcutaneously transplanting the H22 cells into nude mice is one of the most widely used liver cancer models by far, the orthotopic transplantation mice model is more appropriate for judging the pharmacodynamics of antitumor-targeting agents since it is better to imitate the in-vivo blood supply and growth environment of human hepatocellular carcinoma with low cost and short period. However, orthotopic transplantation mice model still could not meet the requirement of fully simulating the human hepatoma. An ideal liver cancer model should possess the following characteristics: the tumor cells simply grow on liver and mainly present in solitary; the occurrence and development of hepatoma are similar to the natural process of human liver cancer; the pathological features in carcinoma cells resemble that of human hepatocytes; the detection index of tumor enzymology and the serum alpha-fetoprotein would be positive; and animal models should be stable as well as easy to copy and operate within short experiment cycle. Collectively, it is necessary to exploit more desirable liver cancer models for the sake of validating the anti-hepatoma-targeting agents.

With the appearance of gene models, especially knock-out model, the research has attained unprecedented development. However, this technology still exists some drawbacks, and in recent years, the key of contention seems to be focused on whether the phenotype of knock-out animal is caused by the mutation of target gene. Gerlai (2003) pointed out that the current knock-out model ignored the effect of background genotype, and due to the loss of knock-out gene, animal organism was likely to produce an avalanche of compensatory process and lead to the second alteration of genes. Therefore, it is reasonable to speculate that certain complex change of phenotype may not be on account of one particular causality. Furthermore, Fülöp et al. (2003) noted that, besides the genetic background, some other factors would also complicate the results of gene knock-out. Since the gene mutant would probably delete in every tissue’s development processes, the rest of organ defects are likely to involve in the phenotypic changes as well. Taken together, the investigation of animal models is an essential aspect of the study on antitumor-targeting agents (Arap et al., 1998; Kukowska-Latallo et al., 2005). With the further studies of pharmacology, pharmacodynamics, biopharmaceutics and pharmacokinetics (Callegari et al., 2014; Denzler et al., 2014), it is expected that the research level of the antitumor-targeting models would attain a deeper level and accomplish more superior achievements (Chen & Calvisi, 2014; Verhaegh et al., 2014).

Declaration of interest

This work was financially supported by China Postdoctoral Science Foundation Funded Project (2014M562429) and Fundamental Research Funds for the Central University (xjj2013054 and xj08142016). The authors have no conflict of interest. The authors alone are responsible for the content and writing of this article.