Background
Acute myeloid leukemia (AML) is one of the most common types of leukemia and accounts for ~ 42% of all leukemic deaths [
1]. This has warranted a huge focus on its pathogenesis and disease management, as compared to other types of leukemia. Acute myeloid leukemia, marked by abnormal proliferation and differentiation of myeloid leukemic stem cells, is genetically and biologically heterogeneous [
2]. Uncontrolled proliferation of leukemic stem cells forms leukemic blasts in the BM and peripheral blood circulation that eventually results in BM failure and deaths [
2]. Despite a better understanding of the genetic aberrations [
3,
4] that contribute to AML and recent therapeutic advances [
5], the overall five-year survival rate remains low at 30–40% in patients younger than 60 years and less than 20% for patients above 60 years [
6].
Therefore, there is a need to develop relevant AML animal models for the purpose of novel targets discovery and assessment of new therapies. Since the early 1900s, murine models have been extensively used to study AML, using approaches such as carcinogen-induced transplantable models, transgenic, xenograft, and mosaic models [
7]. In particular, xenograft of patient-derived AML cells into immunodeficient mice such as severe combined immunodeficient (SCID) [
8], non-obese diabetic (NOD)/SCID [
9], and NOD-
scid Il2rγ
null
(NSG) [
2] mice was instrumental in defining leukemic stem cells [
8] and their chemotherapy-resistant properties [
2,
10]. Due to their longer life span (> 90 weeks) and greater engraftment capacity, NSG mice are the most widely used animal model [
9,
11,
12].
While xenograft AML model can provide novel insights in understanding human AML biology, a vast improvement in existing models is desired. Often, construction of xenograft models relies on technically challenging methods such as neonatal craniofacial intravenous injection in neonatal mice [
2] and intratibial or intrafemoral injections in adult mice [
13‐
15]. In addition, the use of adult mice resulted in significantly lower engraftment capacity compared to newborn pups, hence, hindering long-term evaluation [
2]. Importantly, existing AML models that utilize adult mice exhibit limited peripheral blood engraftment [
11], a hallmark feature of human AML. Therefore, there is a need for an AML xenograft model that is easier to construct, adequately recapitulates human AML, and allows for long-term evaluation in vivo.
In this study, we sought to establish an improved pre-clinical AML xenograft model that is robust and easier to construct as compared to existing models. Using BM mononuclear cells obtained from seven AML patients, T cell-depleted AML cells were injected into sublethal irradiated NSG newborn pups via the intrahepatic route, a method routinely used in the humanization of NSG mice [
16]. Three (Leu 14, BMI 1690, and BMI 1808) out of the seven AML patients exhibited AML leukemic blasts-associated phenotype and successfully engrafted in NSG recipient mice. Cytometric and histological analysis revealed high level of AML engraftment in the peripheral blood, spleen, and BM of recipient NSG mice. Serial transplantation, up to tertiary transplantation, was performed to further characterize our model. We demonstrated that CD34
+ cells have significantly greater engraftment capacity than CD34
− cells. Furthermore, CD117 expression on CD34
+ cells enhanced engraftment level. When compared to the existing model constructed using NSG adult mice and intravenous injection, our method showed more efficient AML engraftment. Lastly, the therapeutic potential of multi-kinase inhibitors Sorafenib and Regorafenib against AML was evaluated in our model. The favorable outcome of Sorafenib and Regorafenib was recapitulated in our model, with AML cells in the periphery and spleen sensitive to treatments, while those in BM remained unaffected. Collectively, our model serves as a robust, easy-to-construct and reliable pre-clinical tool for AML that will facilitate the discovery of new targets and assessment of new therapeutics.
Methods
Cell preparation
Bone marrow cells were obtained from patients with acute leukemia who had marrow study done at the time of diagnosis. Patients gave informed consent for additional aliquot of the marrow aspirate to be used for research purposes in accordance with the ethical guidelines of Singapore General Hospital. Patients with AML were diagnosed using the French-American-British (FAB) classification system; subtype M1 (patients Leu 32 and BMI 1786), M4 (patient BMI 1808), M5 (patient BMI 1690), and M5a (patients Leu 14, Leu 29, and Leu 33). Bone marrow cells were processed using ficoll density gradient centrifugation to isolate mononuclear cells. Cells were frozen and stored in liquid nitrogen until use.
Mice
NOD-scid Il2rγ
null
(NSG) mice were purchased from The Jackson Laboratory. All mice were bred and kept under specific pathogen-free conditions in Biological Resource Centre, Agency for Science, Technology and Research, Singapore. All experiments and procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the Agency for Science, Technology and Research, Singapore, in accordance with the guidelines of the Agri-Food and Veterinary Authority and the National Advisory Committee for Laboratory Animal Research of Singapore.
Primary and serial xenotransplantation of AML cells
For primary xenotransplantation, BM mononuclear cells were depleted of CD3
+ cells using PE selection kit (STEMCELL Technologies) upon labeling with PE-conjugated mouse anti-human CD3 antibody (Biolegend) according to manufacturer’s instructions. One to 3-day-old NSG pups were sub-lethally irradiated at 1Gy and engrafted with 8.7 × 10
4–7.9 × 10
5 of CD3-depleted AML mononuclear cells from seven AML patients (Leu 14, Leu 29, Leu 32, Leu 33, BMI 1690, BMI 1786, and BMI 1808) via intrahepatic injection route. As described previously [
17], intrahepatic inoculation was performed by maintaining the irradiated NSG pup in posterior position (face-up) between the thumb and index finger to expose the abdomen and the liver, which is visible on the right flank. An insulin syringe loaded with 50 μl of cell mixture was then held perpendicular to the pup body by the other hand and inserted straight into the pup liver with bezel facing upwards to release the contents. Mice were bled submandibularly to evaluate the engraftment of AML cells in peripheral blood at 2–4-week intervals after week 6 post-engraftment using flow cytometry. Engraftment of AML cells in BM and spleen was evaluated at endpoint (week 12–20 post-engraftment) using flow cytometry. Cells from BM and spleen were pooled and used for serial xenotransplantation.
Secondary and tertiary xenotransplantation were evaluated on pooled BM and spleen cells from Leu 14. For serial xenotransplantation, CD34+ cells from pooled BM and spleen cells were purified with either fluorescence-activated cell sorting (FACS) using FACSAria (BD Biosciences) after labeling with fluorochrome-conjugated mouse anti-human CD45 (Biolegend) and anti-human CD34 (BD Biosciences) monoclonal antibodies or by magnetic-sorting using CD34 positive selection kit (STEMCELL Technologies) according to manufacturer’s instructions. The purity of human CD34+ cells was > 95% after FACs or magnetic-sorting. Cell number range from 1 × 104–5 × 105 cells were injected into irradiated NSG recipients. Xenotransplantation in NSG adult mice (6-week-old) was performed via tail-vein intravenous injection after sublethal irradiation at 2.5Gy.
Immune cell isolation from peripheral blood, spleen, and BM
Peripheral blood was collected submandibularly from mice in EDTA tube (Greiner Bio-One). Red blood cells (RBCs) were lyzed using RBC lysis buffer (Life Technologies) prior to flow cytometry analysis. For spleen and BM, tissues were meshed and cell contents from femur and tibia were flushed using a syringe, respectively. Cell debris was removed by passing contents through 70 μm cell strainer (Thermo Fisher Scientific). RBCs were further lyzed and contents passed through 70 μm cell strainer prior to flow cytometry analysis and storage.
Flow cytometry analysis of peripheral blood, spleen, and BM
Live immune cells from peripheral blood, spleen, and BM were determined by staining with live/dead fixable blue dead cell stain kit (Life Technologies) for 30 min prior to cell-specific marker labeling. Cells were labeled with anti-human CD34 (clone 581; BD Biosciences), anti-human CD3 (UCHT1; Biolegend), anti-human CD56 (MEM-188, Biolegend), anti-human CD14 (63D3; Biolegend), anti-human CD19 (SJ25C1; BD Biosciences), anti-human CD117 (104D2; Biolegend), anti-human CD38 (HB-7; Biolegend), anti-human CD33 (WM53; BD Biosciences), mouse CD45.1 (A20; BD Biosciences), anti-human CD8 (SK1; Biolegend), anti-human CD4 (SK3; BD Biosciences), and anti-human CD45 (HI30; Biolegend) monoclonal antibodies for 30 min at room temperature. After incubation, cells were washed and resuspended in FACs buffer containing phosphate buffered saline (PBS), 0.2% bovine serum albumin (GE Healthcare Life Sciences), and 0.05% sodium azide (Merck) for flow cytometry data acquisition. Data was acquired using a LSR II flow cytometer (BD Biosciences) with FACSDiva software, and analysis was performed using FlowJo software (version 10; Tree Star Inc). Absolute count of cells in peripheral blood was determined using CountBright™ Absolute Counting Beads (Thermo Fisher Scientific).
Hematoxylin & Eosin stain and immunohistochemistry
Multiple organs including brain, heart, lungs, liver, kidneys, forelimbs/hind limbs, and spleen were removed from sacrificed mice at endpoint. The organs were fixed in 10% formalin, embedded in paraffin wax, processed to obtain 5 μm sections, and subjected to Hematoxylin & Eosin (H&E) (Thermo Fisher Scientific) or immunohistochemistry staining following established protocols. Primary antibodies including anti-human CD45 (cat# ab781), anti-human MPO (cat# ab134132), and anti-human c-kit (cat# ab32363) monoclonal antibodies were purchased from Abcam and used for immunohistochemistry. The primary antibody was detected using Rabbit specific IHC polymer detection kit HRP/DAB (AbCam) or Mouse on Mouse Polymer IHC Kit (AbCam) following manufacturer’s instructions. Histopathological images were acquired using Axio Scan. Z1 slide scanner (Zeiss) and analyzed using Zen 2 (blue edition; Zeiss) software.
Regorafenib and Sorafenib treatment
Sorafenib tosylate (Nexavar®; Bayer Healthcare Pharmaceuticals Inc) and Regorafenib (Stivarga®; Bayer HealthCare Pharmaceuticals Inc.) tablets were crushed and dissolved in isotonic saline water (B. Braun Medical Inc). Successfully engrafted mice with more than 30 human CD45 cells per microliter of blood (between week 12 and 16 post-engraftment) were randomly assigned to either untreated Regorafenib or Sorafenib treatment groups. Mice were given a daily dose of Regorafenib (5 mg/kg body weight) or Sorafenib (10 mg/kg body weight) via oral gavage and monitored for 1 month.
Statistical analysis
Statistical analysis was performed using GraphPad Prism 5.0 software (GraphPad Software Inc). Pairwise comparison was performed using two-tailed Mann Whitney U test. P value less than 0.05 is considered statistically significant. All data are represented as mean ± standard error of mean (SEM).
Discussion
The availability of an in vivo model for human AML is attractive for the understanding of AML biology and for the development of new therapeutic strategies. Due to its prolonged lifespan and enhanced engraftment capacity, xenotransplantation of patient-derived AML cells in NSG mice has long been considered the gold standard for evaluating human hematologic malignancies. However, existing models are not without limitations. As shown by others and by us in this study, engraftment of AML cells is less efficient and unpredictable when transplanted in NSG adult recipient mice [
2,
11]. The influence of age in engraftment efficiency was addressed by Ishikawa et al. in 2007, where engraftment efficiency was shown to be enhanced when AML cells were transplanted into newborn NSG pups via craniofacial intravenous injection [
2]. Despite the improvement and usefulness, this model is not commonly adopted by the scientific community in the past decade as it is technically difficult to construct.
In contrast, our approach which uses intrahepatic injection is widely-accepted and routinely used in the humanization of NSG mice [
16,
22]. Technically, intrahepatic injection of AML cells in newborn NSG pups is relatively easy as compared to neonatal intravenous injection or intrahepatic injection in adult NSG mice, as the liver is visually obvious through the skin of newborn pups and has a large surface area for injection. In this model, we consistently observed high levels of engraftment (> 10%) in the peripheral blood, spleen, and BM in all samples exhibiting CD45
loCD33
+ AML leukemic blast phenotypes (3/3). The engrafted AML cells retained the phenotype of primary cells. The engraftment levels and phenotypes persisted in secondary and tertiary recipients and were not altered by multiple passages in mice. Pathological features of human AML including myeloid sarcoma, spleen enlargement, and infiltration of leukemic cells into circulation and tissues were recapitulated in our model. Furthermore, transplantation dose as low as 1 × 10
4 CD34
+ cells is sufficient for the construction of this AML model. Mice engrafted with 1 × 10
5 CD34
+ cells can expand in vivo to yield large number of AML cells (10
6–10
8) cells after 16 weeks. Through serial transplantation, the recipient mice can potentially provide an unlimited source of AML cells repetitively for direct experimental use, downstream molecular analysis, and ex vivo genetic manipulations. Taken together, these results demonstrate the robustness and specificity of our model with potential for long-term characterization of engrafted patient cells.
The ability to detect circulating AML cells in the peripheral blood allows examination of AML cells in a single recipient mouse across multiple time points. To take advantage of this, attempts were made to identify an immunophenotype that characterized leukemic stem cells which are believed to influence engraftment potential in NSG recipient mice and are responsible for disease resistance or relapse in patients [
34]. Consistent with a recent report [
15], we have demonstrated, through both in vitro and in vivo studies, that leukemic stem cells are enriched in the CD34
+ population and in particular CD34
+CD117
+ fraction, while CD34
+CD117
− fraction is more mature and less potent in proliferation. Interestingly, CD34
+CD117
− engrafted mice gave rise to both CD117
− and CD117
+ cells; therefore, it is not clear if leukemogenesis and pathological outcomes observed in CD34
+CD117
− engrafted mice are driven by CD117
− or CD117
+ cells. Given the heterogeneity and plasticity of leukemic stem cells, these results raise the possibility that the engrafted CD34
+CD117
− cells can de-differentiate to give rise to CD34
+CD117
+ cells and acquire an immature, stem cell-like property to drive the disease progression [
12,
35,
36].
Although it is generally accepted that leukemic stem cells are phenotypically characterized as CD34
+CD38
− [
2,
8,
34], increasing evidence from various groups have challenged this notion and have demonstrated that leukemic stem cells also exist in the CD34
+CD38
+ fraction and CD34
− subpopulation [
12,
14,
37‐
39]. It is possible that the leukemic stem cell activity is mediated by the CD34
+CD38
+ population as majority of the CD34
+ cells from Leu 14, BMI 1690, and BMI 1808 patients express high levels of CD38.
Global gene expression profiling of AML cells before and after transplantation in mice might be useful to ascertain if the observed differential phenotype is due to an outgrowth of a subclone or because AML cells acquired a different differentiation pattern in mice as opposed to patients [
40]. This does not imply that xenotransplantation in NSG mice is unstable, but rather it underscores the plasticity of leukemic stem cells, such that their commitment to certain fate(s) can be multidirectional or reversible, depending on the intrinsic and extrinsic signals [
36]. Clonal evolution of AML has been observed previously in patients during relapse and xenograft models during serial transplantations [
40‐
42]; however, the mechanism underlying clonal evolution is not known. With a larger patient cohort, these dimensions can be further investigated using our model.
In addition, the discrepancies in engraftment potential as shown by the “low engrafters” (Leu 32 and Leu 1786) despite the presence of CD45
lo leukemic blasts can possibly be a reflection of their in vivo proliferative ability, or alternatively, an indication of prognosis. It was reported in previous studies that AML cells from patients with poor prognosis features such as the presence of FLT3 mutations, high white blood cell count at diagnosis, or chromosomal rearrangements would tend to engraft more efficiently in mice than AML cells isolated from patients with good prognostic features [
9,
40,
43]. Intrahepatic delivery of these “low engrafters” into neonatal NSGS mouse strain can also be explored in future studies, as it was demonstrated that constitutive expression of human cytokines (SCF, GM-CSF, and IL-3) in NSGS mice improved engraftment efficiency of “low engrafters” [
15].
Chemotherapy drug resistance mediated by BM microenvironment is increasingly recognized as a major obstacle to the treatment of AML [
2,
44]. The BM microenvironment, which is rich in growth factors, cytokines, and stromal cells, provides a permissive environment for leukemogenesis and also contributes to chemotherapy resistance through mechanisms involving growth factors and cell-cell interaction [
45]. Adhesion of leukemic blasts to marrow stromal cells and fibronectin via molecules, such as CD117 and CXCR4, or to osteoblast-rich areas has been shown to protect AML leukemic blasts from drug-induced apoptosis [
2,
46,
47]. Thus, it is not surprising to observe Sorafenib- and Regorafenib-induced apoptotic, anti-proliferative, and anti-angiogenic effects on leukemic blasts in the periphery and spleen but not BM [
48,
49]. Our work suggests that future therapeutic strategy should consider drug design that directly targets the leukemic blasts in the BM. Alternatively, Sorafenib/Regorafenib can be combined with small molecule inhibitor (e.g., AMD3100; CXCR4 inhibitor or CD117 inhibitor) that disrupts the leukemic blasts-BM interactions and mobilizes leukemic blasts to the periphery, thereby sensitizing them to the cytotoxic effects induced by Sorafenib and Regorafenib [
10]. This new model can provide a pre-clinical platform for the testing of the combined therapies.