How to Establish Acute Myeloid Leukemia Xenograft Models Using Immunodeficient Mice

Acute myeloid leukemia (AML) is the most common hematological malignancy in adults, characterized by anomalous proliferation and differentiation of malignant myeloid progenitors. The resultant progeny, or blasts exhibit early differential blocks and proliferate uncontrollably, resulting in a relentless accumulation of immature, abnormal hematopoietic cells in the bone marrow and peripheral blood (Li et al., 2012), and frequently lead to fatal infection, bleeding, or organ infiltration, with or without leukocytosis (Lowenberg et al., 1999; Estey and Döhner, 2006; Fritsche-Polanz et al., 2010). The annual incidence of AML was approximately 1.8 per 100 000 and progressively increased with age to a maximum prevalence of 17.2 per 100 000 adults of 65+ years of age (Ries et al., 2008). Despite modern advances in supportive care and therapeutics (Appelbaum et al., 2001; Trail et al., 2003) and improvement in the diagnosis of AML subtypes (Bennett et al., 1985; Bennett et al., 1989; Cheson et al., 1990; Casasnovas et al., 1998; Lowenberg et al., 1999), the rate of long-term survivors was greatly low. Our understanding of leukemogenic disease pathogenesis has to a large extent been derived from several decades of research on human subjects (Appelbaum et al., 2001). Primary human AML cells could be isolated and studied in vitro, but many experimental questions could only be addressed using in vivo models. Thus the use of mice models of leukemia served the function of making the disease accessible to experimentation impossible in human


Introduction
Acute myeloid leukemia (AML) is the most common hematological malignancy in adults, characterized by anomalous proliferation and differentiation of malignant myeloid progenitors. The resultant progeny, or blasts exhibit early differential blocks and proliferate uncontrollably, resulting in a relentless accumulation of immature, abnormal hematopoietic cells in the bone marrow and peripheral blood (Li et al., 2012), and frequently lead to fatal infection, bleeding, or organ infiltration, with or without leukocytosis (Lowenberg et al., 1999;Estey and Döhner, 2006;Fritsche-Polanz et al., 2010). The annual incidence of AML was approximately 1.8 per 100 000 and progressively increased with age to a maximum prevalence of 17.2 per 100 000 adults of 65+ years of age (Ries et al., 2008). Despite modern advances in supportive care and therapeutics (Appelbaum et al., 2001;Trail et al., 2003) and improvement in the diagnosis of AML subtypes (Bennett et al., 1985;Bennett et al., 1989;Cheson et al., 1990;Casasnovas et al., 1998;Lowenberg et al., 1999), the rate of long-term survivors was greatly low.
Our understanding of leukemogenic disease pathogenesis has to a large extent been derived from several decades of research on human subjects (Appelbaum et al., 2001). Primary human AML cells could be isolated and studied in vitro, but many experimental questions could only be addressed using in vivo models. Thus the use of mice models of leukemia served the function of making the disease accessible to experimentation impossible in human In retrospect, nude mice or SCID mice were the first immunodeficient strains. Subsequently, their strains were generated to improve engraftment capacities. NOD/SCID mice established in 1995 have been a milestone in this field. In early 2000s, NSB, NOG and NSG mice were established. Due to the complete loss of murine immune systems, these strains showed extremely high engraftment rates

Immunodeficient mice
The history of the development on immunodeficient mice for establishing AML model was shown in Figure  1. The discoveries of nude and severe combined immunodeficiency (SCID) mice were key advances in the development of immunodeficient mice for xenotransplantation (Bosma et al., 1983;Pearce et al., 2006). SCID mice provided an attractive humanized model for research in AML. However, engraftment rates were not high. The development of non-obese diabetic NOD/SCID (NS) mice also contributed to the generation of humanized mice Koyanagi et al., 1997). NOD/Cg -Prkdc scid B2m tmIUnc /J (NSB) was subsequently derived from NS mice (Christianson et al., 1997). Compared to NS, absence of β2 microglobulin in NSB mice lead to a lack of MHC class I expression and decreased NK-cell function (Feuring-Buske et al., 2003;Kawano et al., 2005;Shultz et al., 2007). Since the early 2000s, a series of immunodeficient mice were developed by combining the IL-2Rcnull gene with conventional SCID mice. These strains showed extremely high engraftment rates, resulting in remarkable advances in the development of AML mice models. These strains included the NOD/Shi-Prkdc scid Il2rγ tm1Sug /Jic (NOG) mice reported in 2002 (Ninomiya et al., 2006); and NOD/ LtSz-Prkdc scid Il2rγ tm1Wjl /J (NSG) mice reported in 2005 (Shultz et al., 2005). Compared to NS, the IL-2Rg-chain deficiency in NSG mice impaired the signaling through multiple cytokine receptors blocking NK development and resulted in additional defects in innate immunity (Shultz et al., 2005;Shultz et al., 2007). Data from Agliano et al suggested that NOG and NSG mice were the best recipients for cell engraftment (Agliano et al., 2008;Brehm et al., 2010). Today, SCID and NOD/SCID strains have been commonly used for establishing AML models, especially the NOD/SCID strains. Nude mice were rarely used in recent years.
Despite the obvious advantages of NOD/SCID model over all other immunodeficient models, there were still some inherent obstacles. The disadvantages of NOD/ SCID mice included the frequent occurrence of thymic lymphoma and discernable NK cell activity , which hindered execution of enduring engraftment. While various strategies had been evaluated co-transplantation of growth, only 70% of all AML samples exhibited detectible engraftment in NOD/SCID mice (Terpstra et al., 1997;Ailles et al., 1999;Bonnet et al., 1999;Rombouts et al., 2000b;Nitsche et al., 2003). The shortened life span of this strain (mean 8.5 months) also limited the duration of tumor watches (Ishikawa et al., 2007). However, NOG and NSG mice showed no leakiness or spontaneous thymic lymphoma, but prolonged the life span of mice (Ishikawa et al., 2007;Kato et al., 2009;Katano et al., 2011;Woiterski et al., 2013). This might be attributed to inactivation of IL-2Rc. Some researchs (Agliano et al., 2008;Brehm et al., 2010;Woiterski et al., 2013) indicated that NOG and NSG mice were better recipients of AML cells.

Xenogeneic Transplantation and Assessment of Engraftment
All animals used for xenogeneic transplantation had some impairment in their immune system and might succumb to infections not affecting normal mice. Thus mice should be kept in a specific pathogen-free environment within barrier systems to protect them from intercurrent infections.
Before cells were inoculated, SCID and NOD/SCID strains were usually pretreated by cyclophosphamide (CTX) or irradiationg to improve engraftment rates. Conditioning with CTX had often been used in doses 100mg/kg (Rombouts et al., 2000b;Tan et al., 2012), and it had also been used in doses 20mg/kg (Zhang et al., 2012a). The dose of total body irradiation (TBI) depended on the mouse strains and also on the source and irradiator. It usually varied from 2 Gy to 4Gy (Chelstrom et al., 1994;Tavor et al., 2004;Liu et al., 2007;Sanchez et al., 2009). Mice should be kept in a specific pathogenfree environment with acidified water at least one week before receiving pretreatment. Transplantation of AML cells was done at lest 24h later after the pretreatment (Liu et al., 2007;Zhang et al., 2012b). In a typical transplant, 5-40×10 6 cells of AML were often injected into mice (Fortier and Graubert, 2010). Woiterski et al reported that the engraftment levels itself did not correlate with the number of injected blasts (Woiterski et al., 2013). Diverse routes of administration of cells had been employed, including subcutaneous (s.c.), intraperitoneal (i.p.) and intravenous (i.v.) routes (Kiser et al., 2001;Zhang et al., 2012a;Zhang et al., 2012b). In recent years, Wang et al (Wang et al., 2012) reported a successful engraftment of human acute lymphoblastic leukemia cells in NOD/SCID mice via intrasplenic inoculation. In addition, Chen et al (Chen et al., 2012) highlighted the potential of a novel in vivo model of human BM microenvironment that could be genetically modified.
Mice should be monitored daily after transplantation of the AML cells, and we could observe the changes of body weight and the number of WBC and leukemia cells among peripheral blood every week. Mice might be culled if they showed signs of ill health, such as piloerection, hunched posture, inactivity or inappetence for a period of 48 h. In addition, any animal that lost 20% body weight or that developed more serious clinical signs, such as diarrhoea or dyspnoea should also be sacrificed (Bonnet, 2009). Peripheral blood and tissues including bone marrow, solid tumors, spleen, liver and kidney were collected from the mice. Peripheral blood samples were then smeared onto glass microscope slides and stained using Wright's stain or analysed by flow cytometry to monitor the percentages of immature granular leukocytes and monocytes (Na et al., 2000;Henschler et al., 2005). Tissues were fixed in 4% formaldehyde, embedded in paraffin, sectioned, and stained with hematoxylin and eosin or using immunohistochemistry to check for the presence or absence of tumor cells under the microscope (Liu et al., 2007;Agliano et al., 2008;Sanchez et al., 2009). To determine the quantitative level of engraftment, tissues could be also flushed in PBS and made into single cell suspensions for flow cytometry analysis (Sanchez et al., 2009). The percentage of engraftment was determined as the number of AML cells over the total number of viable cells harvested per mouse (Sanchez et al., 2009). However, flow cytometry could only detect human leukemia cells as 0.1-0.5% of the cells. Thus we could further confirm human engraftment by PCR (Agliano et al., 2008). Furthermore, we could select fluorescent in situ hybridization (FISH) to detect the presence of cytogenetic abnormalities characteristic of the leukemic cells (Ailles et al., 1997).

Factors influencing Engraftment of Human AML in Immunodeficient Mice
There were several factors influencing engraftment of human AML in the immunodeficient mice (Table  1). Firstly, engraftment of human AML was linked with the characteristic of immunodeficient mice. Data accumulated to date suggested that NOG and NSG mice were the best recipients for humanized tissue and human cell engraftment occurred in the following order: NSG or NOG>NSB>NS>SCID (Greiner et al., 1995;Lapidot et al., 1997;Feuring-Buske et al., 2003;Agliano et al., 2008;Snchez et al., 2009). Although NOD/SCID strains had been widely used in recent years, we also considered the SCID mice to be a valuable system for the evaluation of new therapies for AML (Tan et al., 2012;Zhang et al., 2012a). The second factor determining the engraftment potential of human AML was FAB subtypes. Cells with FAB subtypes M3 and, to a lesser extent, M2, engrafted more poorly than those from other subtypes (Ailles et al., 1999). Rombouts et al reported that a significantly higher engraftment was observed in the FAB class M0 than in FAB classes M2, M4 and M5 (Rombouts et al., 2000b). However, there was also a research reporting that engraftment did not correlate with FAB subtype (Sanchez et al., 2009). Thirdly, engraftment potential of AML was associated with routes of administration of the cells. The i.v. route generally resulted in higher levels of engraftment than the i.p. route (Terpstra et al., 1995;Lumkul et al., 2002). Another factor influencing engraftment potential of AML was a poorer prognosis of AML patients. Samples from patients with a poorer prognosis tended to engraft better (Ailles et al., 1999;Rombouts et al., 2000b;Yan et al., 2009). Adverse prognostic features, such as high WBC, FLT3-ITD positive, high CD34 expression, relapse (Giavazzi et al., 1995;Rombouts et al., 2000b;Lumkul et al., 2002;Sanchez et al., 2009), were also associated with an enhanced engraftment potential in AML model. However, high WBC, relapse and FLT3-ITD were not predictors of engraftment success in other studies (Ailles et al., 1999;Pearce et al., 2006). Pearce et al reported that the ability of AML to engraft in the NOD/SCID assay seemed to be an inherent property of AML cells, independent on homing, conditioning, or cell frequency/ source (Pearce et al., 2006). Further studies were needed to investigate whether these adverse prognostic features were linked with enhanced engraftment. Additionally, CXCR4 also influenced engraftment of human AML in the immunodeficient mice. CXCR4 played a important role in the control of human AML cell trafficking and disease progression, and CXCR4 antibodies or CXCR4 inhibitors blocked AML cells homing into the BM and spleen of transplanted mice (Tavor et al., 2004;Zhang et al., 2012b). Another study demonstrated that CXCR4 was not critical for the engraftment of AML CD34+ cells in NOD/SCID mice (Monaco et al., 2004). The last factor influencing engraftment potential of AML was the sex of mice. There were two studies showing that male recipients conferred significantly higher engraftment levels than female recipients (Martin-Padura et al., 2010;Woiterski et al., 2013).

Conclusions
There have been significant advances in the understanding of AML models. Contemporary xenographic murine model systems have been critical to our comprehension and treatment of AML. Over the last decades, remarkable progress has been achieved in AML mouse models. Various mouse models have been established and enabled direct research of human AML, which was previously impossible in immunocompetent mice. These models have certainly been fundamental to our current understanding of leukemogenesis. However, they may not be sufficient for the further development necessary in the evolution of more optimal AML models (Nicolini et al., 2003;Ahmed et al., 2004;Avigdor et al., 2004;Perez et al., 2004;Spiegel et al., 2004). With many recent failures of new therapeutic strategies, we must question the current application and validity of these model systems. Furthermore, these in vitro models could not precisely recapitulate the in vivo setting of human AML.
It is essential that current murine AML models are further developed by revolutions of molecular biological and genomic eras to exploit more specific, targeted therapeutics. Due to the failure of significant numbers of new drugs in late clinical trials, it is necessary to use of more salient, higher throughput, prognostic preclinical animal models. Combining the use of such models should also offer more pertinent preclinical data, thus reducing current expenditure on drug development and afford greater clinical success.