**3.2.2 Rodent liver progenitor cell induction models**

The majority of commonly used LPC induction models was originally developed to study the process of hepatocarcinogenesis. They generally combine an injuring mitotic stimulus, usually in the form of functional liver mass loss (chemically or otherwise-induced), with a manipulation that chronically damages hepatocytes or blocks their ability to divide and prevents them from contributing to the liver regeneration process. Described below are four examples of the most commonly used regimens.

#### **3.2.2.1 D-galactosamine**

This model is mainly used to induce liver injury in the rat. Administration of Dgalactosamine inhibits RNA and protein synthesis in centrilobular hepatocytes by trapping and depleting uridine-nucleotides and UDP-glucose (Decker & Keppler, 1972), leading to acute necrosis. Hepatocyte replication is not fully blocked in this model; the response is only delayed. LPCs are resistant to the chemical as they do not metabolise D-galactosamine and are induced to proliferate within 48 hours after injury. They migrate into the parenchyma, where they generate both ductular cells and small hepatocytes (Lemire *et al.*, 1991; Dabeva & Shafritz, 1993).

Liver Progenitor Cells, Cancer Stem Cells and Hepatocellular Carcinoma 25

hepatocytes and LPCs. It offers an alternate model to investigate mechanisms that regulate LPC proliferation and differentiation. In the context of liver cancer, the DDC model has been used extensively to demonstrate a link between LPCs and HCC. LPCs isolated from p53 null mice subjected to a DDC diet are able to generate both hepatocarcinomas and cholangiocarcinomas following transplantation into immunodeficient mice (Suzuki *et al.*, 2008). By placing a Hepatitis B Virus X transgenic mouse on a DDC diet, Wang and colleagues were able to show that LPCs overexpressing HBx were tumorigenic (Wang *et al.*, 2012). Interestingly, over the same period of seven months, DDC treatment did not induce tumours in wild type mice. In another study, the importance of the Hippo-Salvador pathway, working through inhibition of the yes-associated protein YAP, was shown by subjecting mice with liver-specific ablation of WW45 (drosophila homolog of Salvador and adaptor for the Hippo kinase) to a DDC diet. These mice displayed liver tissue overgrowth, an enhanced LPC response and they developed liver tumours with HCC as well as

cholangiocarcinoma characteristics that appeared to be LPC-derived (Lee *et al.*, 2010).

Fig. 3. Histology of normal and chronically injured liver. Adult mice on a control diet display normal liver architecture with orderly cords of hepatocytes and sinusoidal structures in-between the plates (A). On day 21 of the CDE diet, the liver architecture is highly disrupted by steatosis, scattered aggregates of infiltrated inflammatory cells and

LPCs have been identified in a variety of human liver pathologies and are activated like their rodent counterparts to regenerate chronically injured liver (Haque *et al.*, 1996; Theise *et al.*, 1999). Like oval cells in rodents, human LPCs are usually associated with prolonged fibrosis, hepatocellular necrosis, cirrhosis and chronic inflammatory liver diseases. Hence, their proliferation is frequently seen in patients with hereditary haemochromatosis, chronic hepatitis B or C infection, alcoholic liver disease (ALD) and non-alcoholic fatty liver disease (NAFLD) when hepatocytes are inhibited by DNA-damaging oxidative stress (Lowes *et al.*, 1999; Roskams *et al.*, 2003a; Clouston *et al*., 2005). The degree of stem cell activation and the number of proliferating LPCs in these pathologies was demonstrated to correlate with the progression and severity of the underlying liver disease (Lowes *et al.*, 1999). The activation of human LPCs is characterised by the appearance of reactive ductules, also referred to as Ductular Reaction. Cirrhotic livers have been shown to contain nodules that are usually in close contact with reactive ductules and consist entirely of intermediate hepatocytes, which

proliferating LPCs (B).

**3.2.3 Liver progenitor cells in human pathologies** 

#### **3.2.2.2 Solt-Farber model and the modified 2-AAF/PHx regime**

In this model, which is commonly used in rats and only rarely in mice, injection of the ethylating hepatocarcinogen diethylnitrosamine (DEN) is followed two weeks later by a two-week treatment with 2-AAF and PHx one week into 2-AAF feeding (Solt & Farber, 1976). The most commonly used regimen is a modification to the original Solt-Farber protocol, in which the "initiation" step of DEN injection is omitted and 2-AAF is administered four days before and after PHx, the 2-AAF/PHx regime (Tatematsu *et al.*, 1984). Both models induce proliferation of ductular or periductular LPCs, which accelerates when 2- AAF feeding is terminated, indicating that not only hepatocytes are growth-inhibited by 2- AAF but also LPCs, although to a lesser extent. LPCs differentiate more efficiently into hepatocytes at low doses of 2-AAF, whereas they tend to undergo apoptosis at higher dosages (Alison *et al.*, 1997). As a consequence, the rate at which LPCs differentiate into hepatocytes can easily be controlled through variation of the 2-AAF dose (Paku *et al.*, 2004).

#### **3.2.2.3 Choline-deficient, ethionine supplemented diet (CDE diet)**

A dietary deficiency of the lipotrope choline is known to induce hepatic steatosis (Lombardi *et al.*, 1966; Lombardi *et al.*, 1968). This pathology reflects an impaired release of triglycerides in the form of very low-density lipoprotein (VLDL) from hepatocytes, leading to intracytoplasmic deposition of fat vacuoles within a few hours of choline withdrawal. Choline-deficiency has also been reported to induce hepatocarcinogenesis (Ghoshal & Farber, 1984; Yokoyama *et al.*, 1985; Locker *et al.*, 1986). Similar effects were shown for another well-known carcinogen, DL-ethionine. Administered alone, ethionine is an antagonist of methionine and as such an inhibitor of *de novo* choline-biosynthesis thus induces fatty liver (Farber, 1967) and also leads to HCC (Farber, 1956a). When tested in combination with choline-deficiency, ethionine enhances the formation of liver tumours (Shinozuka *et al.*, 1978b), yet surprisingly diminishes the formation of fatty liver during choline-deficiency (Sidransky & Verney, 1969).

An interesting observation during early choline-deficient, ethionine-supplemented (CDE) diet-induced hepatocarcinogenesis studies in rats was the massive proliferation of αfetoprotein-positive LPCs in the liver (Shinozuka *et al.*, 1978a). Numerous studies using this model to provoke an LPC response in rats were subsequently described. Due to the extensive availability of genetically engineered mouse strains, it became desirable to apply this regimen to mice. The conventional CDE diet used in rats however caused high mortality in mice and was therefore modified to a CD diet with separate administration of 0.165% DLethionine in the drinking water. This customised CDE diet (Akhurst *et al.*, 2001) reliably induces the proliferation of LPCs (Fig. 3, Tirnitz-Parker *et al.*, 2007; Tirnitz-Parker *et al.*, 2010) as well as inflammatory cells (Knight *et al*., 2005) and serves as a murine model of hepatic fibrogenesis (Ruddell *et al.*, 2009; Van Hul *et al.*, 2009) and tumorigenesis following prolonged CDE diet exposure (Knight *et al.*, 2000; Knight *et al.*, 2008).

#### **3.2.2.4 3,5-diethoxycarbonyl-1,4-dihydro-collidine diet (DDC diet)**

The hepatotoxin 3,5-diethoxycarbonyl-1,4-dihydro-collidine is also an effective inducer of LPCs as it causes extensive and prolonged liver damage while the diet is administered (Jakubowski *et al.*, 2005). However, in contrast to the CDE diet (see above), a fraction of hepatocytes continue to proliferate for the duration of diet administration (Wang *et al.*, 2003a). Thus the model is unusual in that liver regeneration is accomplished by both

In this model, which is commonly used in rats and only rarely in mice, injection of the ethylating hepatocarcinogen diethylnitrosamine (DEN) is followed two weeks later by a two-week treatment with 2-AAF and PHx one week into 2-AAF feeding (Solt & Farber, 1976). The most commonly used regimen is a modification to the original Solt-Farber protocol, in which the "initiation" step of DEN injection is omitted and 2-AAF is administered four days before and after PHx, the 2-AAF/PHx regime (Tatematsu *et al.*, 1984). Both models induce proliferation of ductular or periductular LPCs, which accelerates when 2- AAF feeding is terminated, indicating that not only hepatocytes are growth-inhibited by 2- AAF but also LPCs, although to a lesser extent. LPCs differentiate more efficiently into hepatocytes at low doses of 2-AAF, whereas they tend to undergo apoptosis at higher dosages (Alison *et al.*, 1997). As a consequence, the rate at which LPCs differentiate into hepatocytes

A dietary deficiency of the lipotrope choline is known to induce hepatic steatosis (Lombardi *et al.*, 1966; Lombardi *et al.*, 1968). This pathology reflects an impaired release of triglycerides in the form of very low-density lipoprotein (VLDL) from hepatocytes, leading to intracytoplasmic deposition of fat vacuoles within a few hours of choline withdrawal. Choline-deficiency has also been reported to induce hepatocarcinogenesis (Ghoshal & Farber, 1984; Yokoyama *et al.*, 1985; Locker *et al.*, 1986). Similar effects were shown for another well-known carcinogen, DL-ethionine. Administered alone, ethionine is an antagonist of methionine and as such an inhibitor of *de novo* choline-biosynthesis thus induces fatty liver (Farber, 1967) and also leads to HCC (Farber, 1956a). When tested in combination with choline-deficiency, ethionine enhances the formation of liver tumours (Shinozuka *et al.*, 1978b), yet surprisingly diminishes the formation of fatty liver during

An interesting observation during early choline-deficient, ethionine-supplemented (CDE) diet-induced hepatocarcinogenesis studies in rats was the massive proliferation of αfetoprotein-positive LPCs in the liver (Shinozuka *et al.*, 1978a). Numerous studies using this model to provoke an LPC response in rats were subsequently described. Due to the extensive availability of genetically engineered mouse strains, it became desirable to apply this regimen to mice. The conventional CDE diet used in rats however caused high mortality in mice and was therefore modified to a CD diet with separate administration of 0.165% DLethionine in the drinking water. This customised CDE diet (Akhurst *et al.*, 2001) reliably induces the proliferation of LPCs (Fig. 3, Tirnitz-Parker *et al.*, 2007; Tirnitz-Parker *et al.*, 2010) as well as inflammatory cells (Knight *et al*., 2005) and serves as a murine model of hepatic fibrogenesis (Ruddell *et al.*, 2009; Van Hul *et al.*, 2009) and tumorigenesis following

The hepatotoxin 3,5-diethoxycarbonyl-1,4-dihydro-collidine is also an effective inducer of LPCs as it causes extensive and prolonged liver damage while the diet is administered (Jakubowski *et al.*, 2005). However, in contrast to the CDE diet (see above), a fraction of hepatocytes continue to proliferate for the duration of diet administration (Wang *et al.*, 2003a). Thus the model is unusual in that liver regeneration is accomplished by both

**3.2.2.2 Solt-Farber model and the modified 2-AAF/PHx regime** 

can easily be controlled through variation of the 2-AAF dose (Paku *et al.*, 2004).

**3.2.2.3 Choline-deficient, ethionine supplemented diet (CDE diet)** 

prolonged CDE diet exposure (Knight *et al.*, 2000; Knight *et al.*, 2008). **3.2.2.4 3,5-diethoxycarbonyl-1,4-dihydro-collidine diet (DDC diet)** 

choline-deficiency (Sidransky & Verney, 1969).

hepatocytes and LPCs. It offers an alternate model to investigate mechanisms that regulate LPC proliferation and differentiation. In the context of liver cancer, the DDC model has been used extensively to demonstrate a link between LPCs and HCC. LPCs isolated from p53 null mice subjected to a DDC diet are able to generate both hepatocarcinomas and cholangiocarcinomas following transplantation into immunodeficient mice (Suzuki *et al.*, 2008). By placing a Hepatitis B Virus X transgenic mouse on a DDC diet, Wang and colleagues were able to show that LPCs overexpressing HBx were tumorigenic (Wang *et al.*, 2012). Interestingly, over the same period of seven months, DDC treatment did not induce tumours in wild type mice. In another study, the importance of the Hippo-Salvador pathway, working through inhibition of the yes-associated protein YAP, was shown by subjecting mice with liver-specific ablation of WW45 (drosophila homolog of Salvador and adaptor for the Hippo kinase) to a DDC diet. These mice displayed liver tissue overgrowth, an enhanced LPC response and they developed liver tumours with HCC as well as cholangiocarcinoma characteristics that appeared to be LPC-derived (Lee *et al.*, 2010).

Fig. 3. Histology of normal and chronically injured liver. Adult mice on a control diet display normal liver architecture with orderly cords of hepatocytes and sinusoidal structures in-between the plates (A). On day 21 of the CDE diet, the liver architecture is highly disrupted by steatosis, scattered aggregates of infiltrated inflammatory cells and proliferating LPCs (B).

#### **3.2.3 Liver progenitor cells in human pathologies**

LPCs have been identified in a variety of human liver pathologies and are activated like their rodent counterparts to regenerate chronically injured liver (Haque *et al.*, 1996; Theise *et al.*, 1999). Like oval cells in rodents, human LPCs are usually associated with prolonged fibrosis, hepatocellular necrosis, cirrhosis and chronic inflammatory liver diseases. Hence, their proliferation is frequently seen in patients with hereditary haemochromatosis, chronic hepatitis B or C infection, alcoholic liver disease (ALD) and non-alcoholic fatty liver disease (NAFLD) when hepatocytes are inhibited by DNA-damaging oxidative stress (Lowes *et al.*, 1999; Roskams *et al.*, 2003a; Clouston *et al*., 2005). The degree of stem cell activation and the number of proliferating LPCs in these pathologies was demonstrated to correlate with the progression and severity of the underlying liver disease (Lowes *et al.*, 1999). The activation of human LPCs is characterised by the appearance of reactive ductules, also referred to as Ductular Reaction. Cirrhotic livers have been shown to contain nodules that are usually in close contact with reactive ductules and consist entirely of intermediate hepatocytes, which

Liver Progenitor Cells, Cancer Stem Cells and Hepatocellular Carcinoma 27

Metastatic breast cancer was the first solid tumour in which CSCs were identified and prospectively isolated. The CD44+CD24-/lowLineage- cell population initiated tumours upon transplantation into mice with as few as 100 cells per injection. Importantly, they could be serially passaged and reliably reproduced the heterogeneous phenotype of the original breast cancer. In contrast, unsorted cells from the primary tumour or injection of a large number of alternate phenotypes, such as CD44+CD24+ cells, failed to form tumours (Al-Hajj *et al*., 2003). Furthermore, it was established that increased expression of the detoxifying enzyme aldehyde dehydrogenase (ALDH) identifies the tumorigenic breast stem cell fraction and high ALDH1 activity correlates with poorer prognosis (Ginestier *et al*., 2007).

The discovery of breast CSC was reported in the same year as the identification of tumourinitiating stem cells in the brain. Singh and colleagues identified and prospectively isolated a CD133+ population of cells from a range of human brain tumours including medulloblastomas, pilocytic astrocytoma, glioblastoma and anaplastic ependymoma that *in vitro* exhibited stem cell properties and gave rise to heterogeneous cell populations with the same phenotype as the original tumour cells. Upon transplantation of as few as 100 CD133+ glioma cells into the frontal lobes of NOD/SCID mice, serially transplantable tumours were initiated that mirrored the original tumour phenotype, whereas no tumours developed after

Only very recently have liver CSCs been described. However the mounting evidence is compelling and ever more markers are suggested to describe the population of cells that may be responsible for liver cancer initiation, maintenance and potentially tumour

The first evidence for the existence of liver CSCs came from Haraguchi and colleagues who performed Hoechst 33342 side population (SP) analyses of various human gastrointestinal cell lines and identified a subpopulation of cells with CSC properties. The SP approach is based on the finding that cells without stem cell characteristics accumulate the fluorescent nucleic acid-binding dye Hoechst 33342, whereas stem cells and CSCs do not as they are capable of effectively effluxing the dye through high activity of adenosine triphosphate (ATP)-binding cassette (ABC) transporters such as the multidrug resistance transporter 1 (MDR1) or breast cancer resistance protein (BCRP, also known as ABCG2). These ABC transporters employ ATP hydrolysis to facilitate substrate export across membranes against steep concentration gradients and thereby protect cells from cytotoxic agents and importantly from chemotherapeutic drugs such as cisplatin and doxorubicin. The authors report that the HCC lines HuH7 and Hep3B contained 0.9% to 1.8% SP cells with CSC properties, respectively, whereas no SP cells could be purified from the less aggressive hepatoma cell line HepG2 (Haraguchi *et al*., 2006). These results were confirmed shortly after by Chiba and colleagues who identified SP cells in some human liver cell lines, which

cells from the same tumour (Singh *et al*., 2003;

**4.2.1 Breast cancer stem cells** 

**4.2.2 Central nervous system cancer stem cells** 

injection of a much larger number of CD133-

recurrence after HCC resection, as described below. **4.2.3.1 Side population (Hoechst 33342 dye efflux)** 

Singh *et al*., 2004).

**4.2.3 Liver cancer stem cells** 

strongly suggests they originate from LPCs (Roskams *et al.*, 2003a; Roskams *et al.*, 2003b, Falkowski *et al.*, 2003). LPCs always emerge in pathologies with a predisposition to cancer and their proliferation in an environment rich in inflammatory mediators, growth factors or reactive oxygen species renders them likely targets for transformation. Furthermore, inhibition of the LPC response has been demonstrated to reduce the formation of cancerous lesions, strongly supporting a role for LPCs in hepatocarcinogenesis (Davies *et al*., 2006; Knight *et al*., 2005; Knight *et al*., 2008). Very recently LPCs have not only been discussed as cellular precursors for liver cancer but also as potential liver cancer stem cells, which could be responsible for tumour maintenance and recurrence (Marquardt *et al.*, 2011; Rountree *et al.*, 2012).
