Direct Orthotopic Transplantation of Fresh Surgical Specimen Preserves CD133+ Tumor Cells in Clinically Relevant Mouse Models of Medulloblastoma and Glioma


  • Qin Shu,

    1. Laboratory of Molecular Neuro-Oncology, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    2. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Kwong Kwok Wong,

    1. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    2. Department of Gynecologic Oncology, University of Texas MD Anderson Cancer Center, Houston, Texas, USA
    Search for more papers by this author
  • Jack M. Su,

    1. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Adekunle M. Adesina,

    1. Department of Pathology, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Li Tian Yu,

    1. Laboratory of Molecular Neuro-Oncology, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    2. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Yvonne T. M. Tsang,

    1. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    2. Department of Gynecologic Oncology, University of Texas MD Anderson Cancer Center, Houston, Texas, USA
    Search for more papers by this author
  • Barbara C. Antalffy,

    1. Department of Pathology, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Patricia Baxter,

    1. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Laszlo Perlaky,

    1. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Jianhua Yang,

    1. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Robert C. Dauser,

    1. Neurosurgery, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Murali Chintagumpala,

    1. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Susan M. Blaney,

    1. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Ching C. Lau,

    1. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    Search for more papers by this author
  • Xiao-Nan Li M.D., Ph.D.

    Corresponding author
    1. Laboratory of Molecular Neuro-Oncology, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    2. Texas Children's Cancer Center, Texas Children's Hospital, Baylor College of Medicine, Houston, Texas, USA
    • Laboratory of Molecular Neuro-Oncology, Texas Children's Cancer Center, Texas Children's Hospital, 6621 Fannin St, MC 3-3320, Houston, Texas 77030, USA. Telephone: 832-824-4580; Fax: 832-825-4038
    Search for more papers by this author


Recent identification of cancer stem cells in medulloblastoma (MB) and high-grade glioma has stimulated an urgent need for animal models that will not only replicate the biology of these tumors, but also preserve their cancer stem cell pool. We hypothesize that direct injection of fresh surgical specimen of MB and high-grade glioma tissues into anatomically equivalent locations in immune-deficient mouse brains will facilitate the formation of clinically accurate xenograft tumors by allowing brain tumor stem cells, together with their non-stem tumor and stromal cells, to grow in a microenvironment that is the closest to human brains. Eight of the 14 MBs (57.1%) and two of the three high-grade gliomas (66.7%) in this study developed transplantable (up to 12 passages) xenografts in mouse cerebellum and cerebrum, respectively. These xenografts are patient specific, replicating the histopathologic, immunophenotypic, invasive/metastatic, and major genetic (analyzed with 10K single nucleotide polymorphism array) abnormalities of the original tumors. The xenograft tumor cells have also been successfully cryopreserved for long-term preservation of tumorigenicity, ensuring a sustained supply of the animal models. More importantly, the CD133+ tumor cells, ranging from 0.2%–10.4%, were preserved in all the xenograft models following repeated orthotopic subtransplantations in vivo. The isolated CD133+ tumor cells formed neurospheres and displayed multi-lineage differentiation capabilities in vitro. In summary, our study demonstrates that direct orthotopic transplantation of fresh primary tumor cells is a powerful approach in developing novel clinical relevant animal models that can reliably preserve CD133+ tumor cell pools even during serial in vivo subtransplantations.

Disclosure of potential conflicts of interest is found at the end of this article.


Author contributions: Q.S.: collection and/or assembly of data, data analysis and interpretation, manuscript writing; K.K.W.: collection and/or assembly of data, data analysis and interpretation; J.M.S.: provision of study material or patients, data analysis and interpretation; A.M.A.: provision of study material or patients, collection and/or assembly of data, data analysis and interpretation; L.T.Y.: collection and/or assembly of data, data analysis and interpretation, manuscript writing; Y.T.M.T.: collection and/or assembly of data, data analysis and interpretation; B.C.A.: collection and/or assembly of data, data analysis and interpretation; P.B.: collection and/or assembly of data, data analysis and interpretation, manuscript writing; L.P.: provision of study material or patients; J.Y.: data analysis and interpretation, manuscript writing; R.C.D.: provision of study material or patients; M.C.: provision of study material or patients, data analysis and interpretation, manuscript writing; S.M.B.: provision of study material or patients, data analysis and interpretation, manuscript writing; C.C.L.: provision of study material or patients, data analysis and interpretation, manuscript writing; X.-N.L.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript.

Medulloblastoma (MB) and high-grade glioma, the most common malignant brain tumors in children, are characterized by rapid proliferation, diffuse infiltration into neighboring brain, and frequent leptomeningeal spread. The prognosis for children with high-grade gliomas, including anaplastic astrocytoma (AA), glioblastoma multiforme (GBM), and recurrent MB remains dismal. Moreover, survivors of these tumors are often disabled by long-term neurocognitive and endocrine sequelae [1, [2], [3], [4]–5]. The recent identification and functional studies of cancer stem cells in multiple human cancers, including MBs and GBMs, have provided strong evidence to support their critical roles in the initiation and propagation of most, if not all, human cancers [6, [7], [8], [9], [10], [11], [12], [13], [14], [15]–16]. Cancer stem cells, shown to be resistant to conventional therapies, are also hypothesized to be the true causes of tumor recurrence and treatment failure [17, [18], [19], [20], [21]–22]. Clinically relevant animal models that replicate the biologic phenotypes of these tumors including preservation of their cancer stem cell pools are urgently needed in order to develop effective tumor-specific therapies for malignant central nervous system (CNS) tumors of childhood [23].

Although several transgenic and xenograft mouse models have been developed for pediatric high-grade gliomas and MBs [23, 24], their clinical relevance remains controversial and the existence of cancer stem cells undetermined [23]. An ongoing concern about genetically engineered mouse models is that the clinical and biological heterogeneities of human CNS tumors are not fully replicated through targeting selected genes. The resultant rodent-derived tumors may potentially have drastically different growth characteristics and variable responses to therapy when compared to human brain tumors [23]. In xenograft models, orthotopic transplantation recreates the original tumor microenvironment more accurately than subcutaneous xenografts [25]. However, nearly all the existing orthotopic xenograft models of malignant gliomas and MBs were initiated from cell lines, and such cell line-based models often yield well-circumscribed intracranial tumors versus the infiltrative or metastatic phenotypes that are characteristic of GBM and MB, respectively [26, [27], [28], [29], [30]–31]. In addition, recent studies have shown that the traditional, serum-based cell culture techniques can cause major genetic/phenotypic changes in genetically engineered mouse models of MB [32], and deplete the cancer stem cell pool in human GBM [33]. Thus, these inherent shortcomings of transgenic and cell line-based models limit their usefulness for cancer stem cell studies and pre-clinical drug screenings.

Direct orthotopic engraftment of fresh primary tumors in immunodeficient animals has been shown to better replicate the cellular, molecular, and clinical phenotypes of multiple human cancers [23, 34]. For brain tumors, however, orthotopic injection into mouse brains, particularly into the cerebellum where MBs originate, remains a surgically challenging procedure. This, combined with additional obstacles such as limited tumor tissue availability, difficulties of preserving tumor cell viability, and low-tumor take-rate of fresh tumors even when injected into subcutaneous sites [35, 36], might have been responsible for the lack of primary-tumor-based orthotopic xenograft models of pediatric MBs and gliomas. Although recent successes of MB and GBM xenograft formation in mouse brains from freshly isolated cancer stem cells support their tumorigenicity [9], the rarity of cancer stem cells in primary brain tumors preclude the feasibility of establishing xenografts routinely with pure cancer stem cells. Furthermore, biological behaviors of cancer stem cells are highly dependent on their microenvironment [6, 37, [38]–39], which is usually composed of stromal and non-stem tumor cells. We, therefore, hypothesize that orthotopic injections of brain tumor stem cells, with their accompanying non-stem and stromal components from primary pediatric GBMs or MBs, should replicate the tumor microenvironment and biologic behaviors, preserve cancer stem cells, and maximize successful growth of transplanted xenografts in immunodeficient mice.

In this study, we set out to explore whether clinically relevant orthotopic xenograft mouse models can be established from fresh surgical specimens of pediatric brain tumors, and if the developed xenograft tumors can stably maintain the relevant cancer stem cell pools in vivo. The feasibility and reproducibility of this approach were examined by injecting a total of 17 pediatric brain tumors into the corresponding locations in the brains of severe combined immunodeficiency (SCID) mice. Our results, that is, the high success rate (58.8%) of model development, faithful recapitulation of the biological phenotypes of original patient tumors, and stable preservation of cancer stem cell pools even during serial in vivo subtransplantations, demonstrated that direct orthotopic transplantation of fresh tumor tissues is a useful tool in developing clinically relevant animal models and in preserving CD133+ tumor cells in vivo.

Materials and Methods

Pediatric Brain Tumor Specimens

Freshly resected brain tumor specimens from 17 pediatric patients undergoing surgery at Texas Children's Hospital were obtained for this study. Signed informed consent was obtained from the patient or their legal guardian prior to sample acquisition in accordance with local Institutional Review Board policy. The patients' demographic and clinical information is described in Table 1. There were 14 MBs and three malignant gliomas (one AA and two GBMs). Tumor tissues were washed and minced with fine scissors into small fragments. Single cells and small clumps (3–5 cells per clump) of tumor cells were collected with a 35 μ cell strainer, then resuspended in Dulbecco's modified Eagle's medium (DMEM) to achieve a final concentration of 1 × 108 live cells per ml, as assessed by trypan blue staining, and transferred to animal facility on ice.

Table Table 1.. Clinical and pathological information of the 17 pediatric brain tumors
original image

Direct Heterotransplantation of Primary Tumor Cells into Mouse Brain

The Rag2 SCID mice were bred and housed in a specific pathogen-free animal facility at Texas Children's Hospital. All the experiments were conducted using an Institutional Animal Care and Use Committee approved protocol. Surgical transplantation of tumor cells into mouse cerebellum and cerebrum, usually completed within 60 minutes of tumor removal, was performed as described previously [40]. Both male and female mice, aged 6–8 weeks, were anesthetized with sodium pentobarbital (50 mg/kg, i.p. injections). Tumor cells (1 × 105) were suspended in 2 μl of culture medium and injected into the right cerebellum (1 mm to the right of the midline, 1 mm posterior to the lamboidal suture, and 3 mm deep), or right cerebral hemisphere (1 mm to the right of the midline, 1.5 mm anterior to the lamboidal suture and 3 mm deep) via a 10 μl 26-gauge Hamilton Gastight 1,701 syringe needle (Hamilton Company, Reno, NV, The animals were then monitored daily for development of neurological deficits, at which time they were euthanized and their brains removed for histopathologic examination. Those mice without any neurological deficit after 12 months were euthanized and examined for tumor development.

Serial Subtransplantation of Xenografts In Vivo in Mouse Brains

Whole brains of donor mice were aseptically removed, coronally cut into halves, and transferred back to the tissue culture laboratory. Tumors were then dissected under the microscope, mechanically dissociated into cell suspensions, and injected into the brains of recipient SCID mice as described above.

Long-Term Cryopreservation of Xenograft Cells in Liquid Nitrogen

Freshly prepared xenograft tumor cells were re-suspended with DMEM medium supplemented with 10% fetal bovine serum and 10% dimethyl sulfoxide. After overnight storage at −80°C, the cryovials were transferred into liquid nitrogen. For re-transplantation into SCID mice, tumor cells were quickly thawed at 37°C, washed, counted with trypan blue, and injected intracranially (five mice per tumor; 1 × 105 cells per injection) as described above.

IHC Staining

Immunohistochemical (IHC) staining was performed using a Vectastain Elite kit (Vector Laboratories, Burlingame, CA, as described previously [40]. In brief, antigen retrieval was performed in a microwave oven in 0.03 M sodium citrate acid buffer. Endogenous peroxidase was quenched using hydrogen peroxide before sections were blocked in avidin D and biotin blocking reagent. Primary antibodies included the following: mouse anti-synaptophysin (SYN) (1:200), glial fibrillary acidic protein (GFAP) (1:200), vimentin (VMT) (1:200) (Dako, Carpinteria, CA,, MAP-2 (1:200), Ki67 (1:20) (Abcam Inc, Cambridge, CA,, CD31 (1:50) (Chemicon, Temecula, CA,, CD34 (Lab Vision, Fremont, CA,, and rabbit anti-Von Willebrand Factor (vWF) (1:200) (Chemicon). After slides were incubated with primary antibodies for 90 minutes at room temperature, the appropriate biotinylated secondary antibodies (1:200) were applied and incubated for 30 minutes, and the final signal was developed using the 3,3′-diaminobenzidine substrate kit for peroxidase. The IHC staining was assessed by combining the intensity, scored as negative (−), low (+), medium (++) and strongly positive (+++), and extent of immunopositivity as 0= negative; 1= 1–25%; 2= 26 – 50%; 3= 51 – 75%; 4= >75% positive cells. Histochemical staining of reticulin in MB tumors and xenografts was performed following standard protocol.

SNP Array and Data Analysis

Genomic DNA was extracted from patient tumor samples and xenograft tumors using the Trizol reagent (Invitrogen, Carlsbad, CA, DNA from patients' blood was prepared by QiaAmp DNA blood mini kit (Qiagen, Valencia, CA, Array experiments were performed according to the manufacturer's (Affymetrix, Santa Clara, CA, recommendations [41]. Briefly, genomic DNA (250 ng) was digested with XbaI (New England Biolabs, Ipswich, MA,, ligated to XbaI adaptors, and subsequently amplified by polymerase chain reaction (PCR). Fragmented PCR products were then labeled, denatured, and hybridized to the arrays. After hybridization, the arrays were stained on the fluidics station 450 and scanned using a high-resolution microarray scanner 3,000. GeneChip Human Mapping 10K arrays (Affymetrix), cover 11,555 single nucleotide polymorphism (SNP) loci in the human genome with an average resolution of one SNP per every 210 kb. SNP calls were determined by GDAS version 2.0.

FACS and In Vitro Characterization of CD133+ Tumor Cells

Xenograft tumor cells were labeled with phycoerythin (PE) conjugated monoclonal antibodies against human CD133 (CD133/2-PE), (Milteny Bio, Inc., Auburn, CA, at 4°C for 10 minutes per manufacturer's instructions. Cells were then washed, resuspended in stem cell growth medium, consisted of Neurobasal media (Invitrogen), N2 and B27 supplements (0.5X each) (Invitrogen), human recombinant basic fibroblast growth factor (bFGF) and epidermal growth factor (EGF), (50 ng/ml each) (R&D Systems) [33], penicillin G, and streptomycin sulfate (1:100) (GIBCO-Invitrogen). Cells were then flow-sorted with Cytomation MoFlo (Dako). Dead cells were excluded by propidium iodide (PI) staining.

The isolated CD133+ cells were incubated in the aforementioned stem cell growth medium to allow for the formation of neurospheres [42]. To determine the multi-lineage differentiation capabilities of CD133+ cells, the neurospheres were transferred into poly-D-lysine coated chamber slides and incubated with DMEM growth medium supplemented with 10% fetal bovine serum (FBS). For immunofluorescent staining, cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. The slides were then probed with mouse anti-GFAP, MAP-2 and rabbit anti-O-4 antibodies. The secondary antibodies were either FITC- or Texas Red-conjugated anti-mouse or anti-rabbit IgG (Vector Laboratories). Cells were counterstained with 4′, 6′-diamidino-2-phenylindole (DAPI) (Vector Laboratories), examined with a Nikon microscope, and their images captured with a color CCD at specific magnifications.

Statistical Analysis

The differences in tumorigenicity among MB subgroups were compared using chi-square analysis. Animal survival times were evaluated by log-rank analysis using SigmaStat and plotted with SigmaPlot. A p value less than 0.05 was considered statistically significant.


Tumorigenicity and Subtransplantability of Primary MBs and High-Grade Gliomas in SCID Mice

We prepared tumor cell suspensions through mechanical dispersion and injected 1 × 105 cells per mouse within 60 minutes of tumor resection. All 14 MBs and one AA that originated in human cerebellum were injected into mouse cerebellum, and the two cerebral GBMs were injected into mouse cerebrum. Eight of the 14 MBs (57.1% success rate) and two of the three malignant gliomas (one AA and one GBM) (66.7%) formed intra-cerebellar (ICb) or intra-cerebral (IC) tumors in the mice (Fig. 1A), while the remaining six MBs and one GBM failed to yield tumors after 12 months of monitoring (Table 1). Four out of five (80%) anaplastic MBs formed intra-cerebellar tumors, compared to three out of six (50%) nodular MBs and one out of three (33.3%) classic MBs. These differences, however, were not statistically significant (p= .38). Success of MB xenograft growth also did not appear to be associated with the clinical M-stage of the patient, as five out of eight (62.5%) M0 MBs resulted in tumor formation, versus three out of six (50%) xenografts in M+MBs (p= .64). Among MB patients who had poor clinical outcomes (recurrent tumor, alive with disease, or died of disease), five out of seven (71.4%) primary tumors formed intra-cerebellar xenografts, compared with only three xenografts out of seven primary tumors (42.9%) from patients who did well clinically (no evidence of tumor after treatment) (p= .13) (Table 1).

Figure Figure 1..

Tumor formation in Rag2/SCID mice. (A): Gross appearance of mouse brain showing the enlarged cerebellum and abnormally formed blood vessels (arrowhead) caused by the growth of intracerebellar anaplastic MB xenograft (ICb-984MB). (B): Time-course in vivo growth of an invasive anaplastic MB xenograft tumor (ICb-984MB) (Ba–Bf) as compared with a non-invasive classic MB xenograft tumor (ICb-1192MB) (Bg–Bl). (C): Examples of leptomeningeal spread as evidenced by micro-satellite nodules formed in areas away from tumor mass (Δ) in anaplastic (ICb-984 MB) and desmoplastic (ICB-1338 MB) xenografts. (D): Diffuse infiltration of GBM xenograft tumors grown in mouse cerebrum (IC-1128GBM) and cerebellum (ICb-1227GBM) and dilation of lateral ventricle (*) caused by obstructive hydrocephalus. In addition to H&E staining, invasive tumor cells (arrow) were detected with immunohistochemical staining using human-specific antibodies against mitochondria (Dp–Dr, Dv–Dx). Abbreviations: GBM, glioblastoma multiforme; MB, medulloblastoma; SCID, severe combined immunodeficiency.

We tested the subtransplantability of established xenograft tumors by repeated injections of 1 × 105 cells into the brains of recipient SCID mice. We succeeded in sustaining serial in vivo passage of xenograft tumors, with nearly 100% tumorigenicity for up to 12 passages (Table 1). Xenograft tumors typically yields 2 × 107 to 2 × 108 live cells per tumor, thereby enabling further subtransplantations in 200–2000 additional mice.

Histopathologic Examination of Xenograft Tumors

Both MB and malignant glioma xenografts typically cause dramatic enlargement of the cerebrum and cerebellum, respectively (Fig. 1A). Although some tumors extended outward to the brain surface, we did not find any discreet extra-axial tumors without brain parenchymal invasion.

Neuropathology review of the H&E stained sections of xenograft tumors from the initial injections and all subsequent subtransplantations demonstrated that all eight MB and two malignant glioma xenograft models replicated the major histologic characteristics of the original tumors. The MB xenografts maintained the key cellular features of the original anaplastic, classic or nodular subtypes (Fig. 1B and 2A), and displayed characteristic reticulin staining patterns, particularly in the nodular MBs (Fig. 2B). The GBM and the AA xenografts showed diffuse infiltrating growth patterns with distinctive features of palisading or geographic necrosis, microvascular proliferation, and marked cellular pleomorphism (Fig. 1D).

Figure Figure 2..

Histopathological and genetic profiles of orthotopic xenograft tumors as compared with the original patient tumors. (A): Major histologic features of anaplastic, classic, nodular and large-cell subtypes of MBs were replicated in the xenograft tumors. (B): Reticulin staining showing the differences of staining patterns between nodular and anaplastic MB. The typical biphasic pattern with highly desmoplastic areas and reticulin-free islands was only observed in nodular MBs. (C): Representative images showing the similarity of immunohistochemical profiles between the primary tumor and the xenograft of the anaplastic MB (984MB). (D): Immunohistochemical characterization of species identity of blood vessels in xenograft tumors. The antibodies against CD31 and CD34 are human-specific, while the antibodies against vWF can recognize both human and mouse endothelial cells. Compared with the patient tumor in which blood vessels were positively stained with all three antibodies, the xenograft tumor was only positive for vWF but negative for CD31 and CD34 antibodies. (E): Cell proliferation and angiogenesis in the xenograft tumors were determined through immunohistochemical staining using antibodies against Ki-67 (left panel) and endothelial marker vWF (right panel). (F): DNA genotyping with 10K SNP array of orthotopic MB and GBM xenografts during serial in vivo subtransplantation from passage one (P-I) up to passage five (P-V) as compared with the original patient tumor (Patient Tumor). The whole genome profiles covering all 23 pairs of chromosomes show regions with LOH in blue while the yellow regions show no LOH. These data were generated using the matched blood DNA as controls. For IC-1128 GBM, which was originated from a recurrent tumor GBM11282nd, the LOH profile from the primary tumor of the same patient (GBM11281st) was also included. Abbreviations: GBM, glioblastoma multiforme; LOH, loss of heterozygosity; MB, medulloblastoma.

To evaluate whether the xenograft tumors maintained the immunophenotypes of the original tumor, protein expression of neuronal and glial markers was examined with IHC staining. In MB xenografts, the distribution and overall heterogeneous staining intensities of neuronal marker SYN, MAP2, and intermediate filament VMT closely resembled those in the corresponding patient tumors (Table 2, Fig. 2C). Most of the GFAP positive cells were stellate shaped and clustered near the neighboring normal brain, consistent with reactive astrocytes (Fig. 2C). Both glioma models showed diffuse GFAP positivity and strong vimentin expression, consistent with the patients' original tumors.

Table Table 2.. Summary of immunohistochemical characteristics of xenograft tumors during serial subtransplantation
original image

The proliferative index was estimated with Ki-67 immunostaining (Fig. 2E). In five of the eight MBs and both of the malignant gliomas, elevated cellular proliferation (60–80%) observed in patients' tumors was reproduced in subsequent xenografts from serial subtransplantations (50–80%). In the single classic MB (1197MB), a similar proliferative index of 30% was seen in both the originating tumor and subsequent xenografts. An increased proliferation index from 35–40% to 70–90% was observed in the two remaining MBs (Table 2).

Tumor Angiogenesis and Microvessel Density

The growth of solid tumors is dependent on an adequate blood supply through angiogenesis [43]. Since some human cancers can mimic the activities of endothelial cells by forming vascular-like networks or be directly involved in tumor vessel formation in xenografts [43, [44]–45], we utilized human specific antibodies against endothelial markers (CD31 and CD34) and mitochondria to identify potential blood vessels of human origin in the xenografts. Although strong staining was obtained in the original human tumors, no immunoreactivity was detected in any of the xenograft tumors (Fig. 2D). Since all blood vessels were positively stained with vWF antibodies (Fig. 2E), which recognize both human and mouse endothelial cells, our findings indicate that the neovasculatures in the xenografts were of mouse origin, probably through endothelial sprouting from pre-existing vessels. Microvessel density (MVD) was counted at a magnification of ×400 in at least five intense tumor neovascularization spots on each section. Although there was variability among the xenograft models, ranging from 2.5 up to 27, many of the xenograft tumors exhibited similar MVDs to their corresponding patient tumors (Table 2). Except for the two MB models (ICb-984MB and ICb-1323MB), in which the MVDs increased (>2 folds), eight of the 10 models displayed stable MVDs during serial in vivo subtransplantations (Table 2). These results suggest that the xenograft tumor cells play an important role in dictating the xenograft angiogenesis in mouse brains, and no major changes of tumor angiogenesis were induced during repeated passaging.

Invasive Growth Patterns of Xenograft Tumors in Mouse Brain

To determine the time-course and in vivo growth patterns of xenograft tumors, we randomly sacrificed three mice every 2 weeks post tumor cell injection and performed serial sections on mouse brains, as previously described [40]. Four MB (ICb-984MB, ICb-1192MB, ICb-1299MB and ICb-1338MB) and two glioma models (IC-1128GBM and ICb-1227AA) were selected because they have been passaged in vivo more than three times and exhibited reproducible animal survival times. The classic MB model ICb-1192MB was the only tumor that maintained sharply-demarcated tumor-brain margins, although occasional cerebrospinal fluid (CSF) metastatic foci were present. In the anaplastic and nodular MB models, extensive invasion into normal brain tissues, early spread (2 weeks post tumor injection) into subarachnoid spaces and micro-satellite metastasis resulted from leptomeningeal spread are frequently observed (Fig. 1B and 1C). Growth of xenografts invariably led to progressive destruction of normal brain and obstruction of CSF circulation, resulting in hydrocephalus (Fig. 1B).

Identification of invasive cells can be difficult on H&E stained paraffin sections. Therefore, to precisely identify tumor invasion at a single cell level, we performed IHC staining using human-specific mitochondrial antibodies. With the lone exception of the classic MB model (ICb-1192MB), we observed a characteristic “irregular edge” of local invasion in all the MB and GBM models. The MB xenografts infiltrated only relatively short distances, typically extending 20–50 cells deep into the surrounding neuropil. Xenografts from anaplastic MBs often migrated further than nodular MBs and formed more metastatic foci. In contrast, the GBM xenograft (IC-1128GBM) showed extensive and long-range infiltration into the normal brain parenchyma (Fig. 1D). In the ICb-1127AA model, the tumor cells, mostly as single cells, infiltrated the entire cerebellar parenchyma and extended to the subarachnoid spaces, while keeping the cerebellar architecture seemingly intact (Fig. 1D). The leading invasive edges of the MB and GBM xenografts were mostly composed of single tumor cells, migrating away from micro-tumors and the trailing tumor cores. These invasive cells showed infrequent association with blood vessels, even in micro-tumors that were approximately 20 cells in cross section. These results suggest that single-cell infiltration was the primary invasive pattern in our xenografts.

Maintenance of Genotypic Characteristics During Serial In Vivo Transplantations

To determine if serial subtransplantations of xenografts caused major changes in genetic profiles, we conducted whole genome allelotyping using 10K SNP array in three MB and one GBM model, for which patient DNA from the original tumor and peripheral blood was available. As shown in Figure 2F, the major regions with loss of heterozygosity (LOH) in the original primary tumor were preserved in ICb-984MB xenografts, whereas absence of LOH in the primary tumor was maintained in ICb-1299MB xenografts, after as many as five in vivo passages. The tumor specimen used to establish IC-1128GBM xenografts was a recurrent tumor removed from a second surgery GBM11282nd, therefore we also analyzed the LOH profile of the initial tumor, GBM11281st. Although GBM11281st showed several regions of LOH, GBM11282nd showed essentially normal genotyping, suggesting that this surgical specimen was probably contaminated with many normal cells. Nonetheless, the GBM11282nd surgical specimen yielded transplantable xenografts, of which the LOH profiles were essentially identical to that of the initial tumor (GBM11281st). These results suggest that, despite normal cell contamination in the GBM11282nd surgical specimen, direct heterotransplantation in SCID mice enriched tumorigenic cells, leading to xenograft engraftment and restoration of the initial genotyping profile. There were, however, some regions of LOH that were either new or lost in ICb-984MB and IC-1128GBM models, particularly after passage five (Fig. 2F), indicating that repeated in vivo subtransplantations can lead to some genetic drifts in these models.

Reproducibility of Animal Survival Times During Serial In Vivo Passages in Mouse Brains

To determine if serial transplantation could change the biological and growth characteristics of the xenografts, we injected an identical number of tumor cells (1 × 105) during serial in vivo passaging and monitored animal survival times (Fig. 3). With the exception of ICb-984MB, ICb-1299MB, and ICb-1338MB models, animal survival times were shortened in passage two and remained relatively stable with subsequent subtransplantations. One possible explanation for this phenomenon is that the original transplants contained significant amounts of normal stromal cells that do not grow following transplantation, and loss of these cells leads to the enrichment of tumor cells in the later passages. The majority of tumor-bearing mice also died within relatively narrow time intervals, indicating that reproducible in vivo tumor growths had been achieved beyond the second passages (Fig. 3A).

Figure Figure 3..

Log-rank analysis of survival times of SCID mice bearing ICb or IC xenograft tumors. (A): Changes of animal survival times during serial in vivo sub-transplantation. All mice received injection of 105 tumor cells in their cerebellum, either from fresh surgical specimen (P-I) or from established xenografts (P-II to P-IX). (B): Correlation between animal survival times and the injected tumor cell numbers. SCID mice were injected with 105, 104 or 103 of tumor cells, respectively. Inverse correlation between cell number and animal survival times was observed in the two MB xenografts (p < .01) but not in the two glioma xenograft models (p > .05). (C): Effect of long-term cryopreservation on the tumorigenicity (upper panel), and animal survival times (middle panel) and histology (lower panel) of a representative MB xenograft model (ICb-984 MB). All mice were injected with 105 xenograft tumor cells. Abbreviations: ICB, intra-cerebellar; IC, intra-cerebral; MB, medulloblastoma; SCID, severe combined immunodeficiency.

Impact of Cell Numbers on Tumorigenicity and Animal Survival Times

To determine the impact of tumor cell number on tumorigenicity and animal survival times, SCID mice (n= 5 per group) were injected with 105, 104 or 103 cells from two representative MB (anaplastic ICb-984MB and classic ICb-1192MB) and malignant glioma (IC-1128GBM and ICb-1227AA) models, respectively. Tumor formation was confirmed in all the animals, including those injected with 103 cells. Increasing the number of the injected tumor cells successively shortened survival in mice with MBs. However, the survival times in mice with AA and GBM remained relatively unchanged despite similar increases in tumor cells injected (Fig. 3B). These results indicate that, once a xenograft model is well established, injecting as few as 103 cells will lead to tumor formation, but the animal survival times are not solely determined by the injected tumor burden, but are most likely affected by the tumors' biological characteristics as well.

Preservation of Tumorigenicity After Long-Term Cryopreservation of Xenograft Tumors

To ensure a timely supply of xenograft models whenever needed and to minimize the genetic drift and phenotypic changes induced by serial passaging, we cryopreserved xenograft tumor cells in liquid nitrogen after every subtransplantation. Cells from two MB models (ICb-984MB and ICb-1299MB) and two glioma models (IC-1128GBM and ICb-1227AA) were retrieved and re-injected into SCID mice (five animals per tumor), and all mice developed ICb or IC xenograft tumors (Fig. 3C). Although mice receiving cryopreserved tumor cells survived slightly longer than those injected with fresh xenograft cells, the histological features of these tumors were nearly identical (Fig. 3C).

Preservation of CD133+ Tumor Cells in Xenograft Tumors

To examine whether cancer stem cells were present as well as preserved after serial in vivo subtransplantations, we used monoclonal antibodies against human CD133, a cell surface marker expressed in multiple normal and cancer stem cells [9, 46, 47], to detect and isolate CD133+ tumor cells with fluorescence activated cell sorting (FACS) (Fig. 4A). In all ten mouse models, a small fraction of CD133+ cells was identified (Fig. 4D). Subsequent microscopic detection of red fluorescence in only the CD133+ fraction but not in the CD133 fraction confirmed the presence of CD133+ cells in our xenografts (Fig. 4B).

Figure Figure 4..

Isolation and in vitro characterization of CD133+ tumor cells. (A): Representative profile showing the isolation of CD133+ cells through FACS using PE-conjugated CD133 antibodies. Cells incubated without CD133-PE antibody were used to set the baseline. (B): Microscopic confirmation of CD133+ (Ba–Bd) and CD133 (Be-Bh) cell population from freshly isolated live cells (Ba, Bb, Be, Bf) and in paraformaldehyde fixed cells (Bc, Bd, Bg, Bh). (C): In vitro self-renewal and multi-lineage differentiation of isolated CD133+ cells. When incubated in the stem cell medium containing EGF and bFGF, the CD133+ cells from IC-1128 GBM formed neurosphere (Ci), which attached upon withdraw of growth factors (Cii, Ciii). Immunofluorescent staining was performed to examine the expression of neuronal marker MAP2, glial marker GFAP and oligodendrocytic marker O4 in the cells that spread out. (D): Preservation of CD133+ tumor cell pools during in vivo subtransplantation as analyzed by FACS. Except ICb-1197MB and ICb-1323MB in which only the initial passage (P-I) was analyzed, the subtransplanted xenograft tumor cells from passage two (II) in ICb-1494MB up to passage 12 (XII) in ICb-984MB of the remaining eight models were presented. Abbreviations: EGF, epidermal growth factor; FACS, fluorescence activated cell sorting; IC, intra-cerebral; GFAP, glial fibrillary acidic protein; PE, propidium iodide.

To further determine if CD133+ tumor cells could be maintained or enhanced during serial subtransplantation, the percentage of CD133+ cells was estimated in eight of the 10 xenograft models with FACS by gating out dead cells, cell debris, and doublets. When compared with the first passage xenograft tumors, in which CD133+ cells ranged from 0.23% to 3.4%, fractions of CD133+ cells were increased in all of the eight models after in vivo subtransplantations (Fig. 4D). These results showed that CD133+ tumor cells were present in the initial xenograft tumors and were preserved and even enriched during repeated in vivo subtransplantations.

Self-Renewal and Multi-Lineage Differentiation Capabilities of CD133+ Cells Isolated from Xenografts

Cancer stem cells have been shown to possess exclusive capabilities of self-renewal and multi-lineage differentiation [7, 9, 10]. To examine in vitro self-renewal of the isolated CD133+ cells, we incubated them in serum-free neurobasal media supplemented with EGF and bFGF. Formation of neurospheres from single CD133+ cells was observed in all the 10 xenograft models, and they often reached the size of 20–40 cells in approximately 14 days (Fig. 4C). To determine the multi-lineage differentiation capacity of the CD133+ cells, the established neurospheres were transferred into normal growth medium. Upon withdrawal of growth factors, these spheres began to attach in ∼4 hours, and started to spread out forming a monolayer in 16–24 hours. Expressions of neuronal marker MAP2, glial marker GFAP, and oligodendrocyte marker O4 were detected with immunofluorescence in the attached cells, confirming the multi-lineage differentiation capabilities in these cells. In MB xenografts derived neurospheres, however, GFAP positive cells were fewer than those expressing neuronal marker MAP2.


The results of current study showed that direct orthotopic engraftment of freshly resected brain tumor tissues into brains of SCID mice is a powerful approach for development of clinically relevant animal models and for stable preservation of CD133+ tumor cell pool in vivo. The relatively high success rate (58.8%) of model generation, that is, 10 models from 17 patient tumors, provided experimental evidence to exhibit the reproducibility and reliability of this technique. The eight MB and two glioma models recapitulated key histopathologic, genetic, and biologic characteristics of pediatric MBs and gliomas. More importantly, we demonstrated that the tumor cells expressing cancer stem cell marker CD133 were preserved in these models even during serial in vivo subtransplantations. These xenograft models represent, to the best of our knowledge, the first and the largest panel of transplantable, primary tumor-based orthotopic xenograft models of pediatric brain tumors. It also represent the first report to demonstrate that CD133+ tumor cell pools can be stably maintained in vivo in an environment that closely resembles their natural habitat.

A fundamental assumption in using human tumor xenografts for cancer research is that the xenografts preserve key features of the corresponding primary tumors [48]. By injecting fresh tumor tissues into mouse brain within 60 minutes of tumor resection, we minimized the chances of major phenotypic and/or genetic changes, which are frequently observed after long-term in vitro expansion [24]. Furthermore, previous studies have demonstrated that the tumor microenvironment can critically affect the biological behavior of xenograft tumors [31]. Recognizing the potential functional and cytostructural differences between the cerebrum and the cerebellum, we transplanted human brain tumors into anatomically matched locations in mouse brains to simulate the microenvironment of the original tumors. These key characteristics of our xenograft models are aimed to maximize tumorigenicity and to preserve relevant clinical and biological attributes of the original tumors.

A large cohort of animals from a well characterized and clinically relevant xenograft model is often required for both biological studies and pre-clinical drug screening. The limited availability of tumor tissues from pediatric patients, however, only allows direct transplantation into small numbers of animals. Without serial subtransplantations, such small numbers of animals are insufficient for any meaningful biological or pre-clinical studies. In this report, we showed that 2 × 107 to 2 × 108 live cells can be routinely harvested from a single xenograft tumor. These cells, when subtransplanted at 1 × 105 cells per mouse, will yield 200–2000 new tumor-bearing mice. We did, however, observe some phenotypic and genetic changes in the xenograft tumors after repeated subtransplantations, including altered animal survival times and changes in genotyping profiles. These observations indicate that adaptive and/or progressive changes are inevitable in xenografts after multiple subtransplantations. Although these changes did not significantly alter the xenografts' histopathologic features or invasive growth patterns, it is still highly desirable to minimize the number of in vivo passage. Our success in long-term cryopreservation of xenograft tumor cells in liquid nitrogen thus provides a strategy for preserving the inherent characteristics of the original tumors, while enabling timely re-establishment and expansion of large numbers of homogenous xenograft tumors on demand for biological and preclinical studies.

Tumor angiogenesis is often critical for sustaining in vivo growth. Although vessel cooption, in which tumors use host vessels to provide their blood supply, has been previously described in xenograft tumors [49, 50], other studies showed that cancer cells may be directly involved in neovascularization [43, [44]–45]. Using specific antibodies, we demonstrated that the new vessels in these xenografts' originated from mouse blood vessels, with no definitive evidence of new blood vessels of human origin. Given the current interest in anti-angiogenic therapies, the absence of neovascularization of human origin in our xenograft models may be a limitation for testing therapies targeting human angiogenesis. Unlike previous studies in which adult GBM xenografts were found to be increasingly dependent on angiogenesis [51], the majority of our models displayed insignificant changes of MVDs during subtransplantations. Use of different starting materials, that is, fresh patient tumors in our models, versus primary cultures of spheroids in FBS-based growth medium in the previous study, may have contributed to this discrepancy.

Diffuse infiltration into normal brain tissues and metastasis throughout the CSF and leptomeninges are major causes for tumor recurrence and treatment failure in patients with malignant brain tumors. It has long been recognized that malignant glioma cells invade the human brain through single cell migration, along myelinated axons or blood vessels [52]. Although infiltration along myelinated axons and blood vessels has been previously reported in traditional cell-line based orthotopic xenograft models [52, [53], [54]–55], these studies rarely duplicated the single-cell migration and invasion. In our study, we further reproduced single-cell invasion in both GBM models and also in anaplastic and nodular MB xenograft tumors, with leptomeningeal spread often observed in association with extensive local invasions. Our models, by replicating the invasive biological nature of pediatric MBs and GBMs, should enable more detailed molecular studies of brain tumor invasion and metastasis.

Cancer stem cells have recently been identified in various human cancers [7, [8], [9]–10, 12, 14, [15]–16]. In the current study, we confirmed the preservation of CD133+ tumor cells in our MB and GBM xenograft tumors, and showed their neurosphere formation and multi-lineage differentiation in vitro. Because CD133 is also present on normal brain stem cells and various non-stem cells in tumors and normal tissues [37], success in serial transplantations of xenografts in animal models, as we have demonstrated in our models, is now recognized as the gold standard for validating the self-renewal and tumor propagation potential of cancer stem cells [37]. Although we were unable to quantify the percentage of CD133+ cells in the primary tumors due to limited tissue samples, we demonstrated that CD133+ tumor cell pools were maintained or increased in our xenograft models during serial subtransplantations. Assuming the hypothesis that cancer stem cells are primarily responsible for treatment failures is correct, preservation of CD133+ tumor cells in our transplantable xenograft tumors would be paramount for maintaining the biological and clinical fidelities of these animal models.


In summary, we have demonstrated that direct orthotopic transplantation of freshly resected brain tumor tissues into brains of SCID mice is a feasible approach for the development of clinically relevant animal models and, at same time, for the preservation of CD133+ tumor cell pools even during repeated in vivo subtransplantations. This result combined with our capability of producing large cohorts of animal models, that is, through subtransplantations or from cryopreserved xenograft tumor cells, should facilitate future biological studies of cancer stem cells and pre-clinical testing of novel therapeutic agents.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


This work was supported by grants from: Childhood Brain Tumor Foundation (X.N. Li), Cancer Fighters of Houston (X.N. Li, C.C. Lau), National Brain Tumor Foundation (X.N. Li).