Isolation, characterization, and in vitro propagation of infantile hemangioma stem cells and an in vivo mouse model

Infantile hemangioma tissues were obtained with approval of Yale University Institutional Review Board and all participants gave informed written consent.

Staining was performed according to standard techniques as previously reported [18].

Post-operative IH tissue samples were washed using PBS and transferred into a sterile Petri dish. The tissue was minced into a fine paste (1 × 1 mm²) and washed with PBS again. IH tissues were then digested with 2 mg/ml collagenase (Invitrogen, CA) for at least 2 hours shaking at 37°C. The cells were dispersed by repeat passage through a pipette tip and filtration through a 100 μM nylon cell strainer. Cells were plated on low attachment Petri dish at a density of 2 × 10⁵ cells/ml using a stem cell culture (SCC) media (all ingredients from Invitrogen, CA) consisting of Knockout DMEM, 15% Knockout Serum Substitute, 1x non-essential amino acids (NEAA), and 20 ng/ml of both basic fibroblast growth factor (bFGF) and human endothelial growth factor (hEGF). Tumor spheres were passaged approximately every seven days (roughly 200 cells/sphere); spheres were washed once with PBS and then single cell suspension made by incubation with 2 mg/ml collagenase IV at 37°C for 10 minutes.

Tumor spheres were transferred to gelatin coated dishes to attach and were differentiated using IH differentiation medium containing Knockout DMEM, 15% Knockout Serum Substitute, 1x non-essential amino acids (NEAA), 20 ng/ml each of basic fibroblast growth factor (bFGF), 20 ng/ml human endothelial growth factor (hEGF), and 50 ng/ml vascular endothelial growth factor (VEGF) (R&D Minneapolis MN).

Cells were fixed with 4% freshly prepared paraformaldehyde for 15 minutes. Following three washes, the cells were blocked with 5% bovine serum albumin (BSA) in 1x phosphate buffered saline-Triton X-100 (TPBS) for 1 h at room temperature. Primary antibodies [mouse anti-GLUT1 (1:80 Abcam, Cambridge, MA), rabbit anti- GLUT1 (1:80, Abcam, Cambridge, MA), rabbit anti-mouse FDR (1:50, Alpha Diagnostic International, San Antonio, TX), rat-anti-mouse CD31 (1:25 BD Biosciences, San Jose, CA), rabbit anti-CD133 (1:25 Biocare Medical, Concord, CA) or rabbit anti-SALL4 (1:400; developed by this laboratory)] in TPBS/1% BSA were added to the tissue sections and the slides were incubated at 4°C overnight. Following a wash with TPBS, an appropriate secondary antibody [goat anti-rabbit-FITC (SouthernBiotech, Birmingham, AL), goat anti-rabbit-RPE (SouthernBiotech, Birmingham, AL), donkey anti-mouse-Rhodamine (Chemicon, Temecula, CA), or donkey anti-mouse-FITC (Sigma, Louis, MO)] was added and the slides were incubated at room temperature for 2 h in the dark. After another three washes with TPBS, the cells were counterstained with 50 ng/ml 4', 6-diamidino-2-phenylindole (DAPI) for 10 minutes.

Tumor spheres were cultured in stem cell culture media for 15 days. The conditioned media, which had been exposed to the tumor cells for 5, 10 and 15 days, was collected and analyzed for the levels of the growth factors (VEGF, EGF, FGF basic and G-CSF) using the human growth factor (GF)4-Plex kit (Invitrogen, Carlsbad, CA) and following manufacture's recommended protocol.

Tumor spheres were seeded onto Petri dishes and exposed to 1 nM, 2 nM, and 4 nM Rapamycin in stem cell culture media (see Methods, IH Tumor Sphere Culture) for 10 days. Approximately 30 tumor spheres were collected at each time point (0, 3, 6 and 10 days), dissociated by collagenase IV to single cells and counted for cell numbers. Each sample was assayed in triplicate.

IH tumor spheres cultured for one week were dissociated by collagenase to single cells and resuspended in Matrigel at a concentration of 5 × 10⁶ cells per ml. 200 μl (10⁶ cells) was injected subcutaneously in both sides of flanks of immunodeficient NOD/SCID mice. Matrigel with PBS or 0.9% NaCl (without tumor cells) was injected into two additional subcutaneous sites as a control. After 10, 20 and 30 days, the mice were sacrificed, the tumor tissues excised, and fixed in formalin. Some tissue sections were stained with standard hematoxylin and eosin (H & E). Other IH tissue sections were analyzed using anti-Glut1 antibody and anti-CD31, a human endothelial cell marker, for their expression levels.

Until now, there has been no adequate in vivo model to evaluate novel therapeutic modalities for IH. The ability to culture IH specimens would be an extremely valuable translational research tool. Recently, Khan and colleagues described an hemangioma stem cell population that expressed CD133 and, when transplanted into nude mice, produced tumors that shared many characteristics of hemangiomas but lacked an obvious proliferative stage[12].

Our group and others [19, 20] have previously shown that mouse SALL4 plays an essential role in maintaining the self-renewal and pluripotent properties of embryonic stem (ES) cells and in governing the fate of the inner cell mass through transcriptional modulation of Oct4 (also known as Pou5f1) and Nanog [19‐24]. We have also observed expression of SALL4 in hematopoietic stem cells suggesting a role in stem cells of various organ systems [25, 26]. Very recently, we demonstrate that SALL4 acts as a robust stimulator for both human and mouse hematopoietic stem/progenitor cell ex vivo proliferation [27‐29]. We have hypothesized that IHs are a stem cell disease and sought to determine if IHs express either SALL4 and/or CD133. By flow cytometry assay, it was shown that the proliferative IHs expressed both markers either alone or together (data not shown). Indeed, using immunohistochemistry, we observed significantly more cells expressing SALL4 and CD133 in proliferative IHs than in involuting phase IHs (Figure 1). This supports the hypothesis that IHs may involute because of depletion of stem cells or endothelial progenitor cell populations.

We next sought to establish an in vitro culture system to study the proliferation of IHs. Primary surgical specimens were dissociated and plated on Petri dishes in serum-free media. Mature endothelial cells and fibroblasts were unable to survive in these conditions. However, structures resembling the tumor spheres of other tumor types were able to grow as shown in Figure 1, 1. Tumor spheres derived from IH samples could be cultured for prolonged periods of time (currently up to passage 30). In addition, IH tumor spheres could also be expanded into high density cultures (Figure 1) - an advantageous feature for therapeutic uses. The IH tumor spheres also resembled induced pluripotent (iPS) stem cells that have been generated from fibroblasts in our laboratory (Figure 1, 1), and also express SALL4. This suggests that the tumor spheres may contain stem cells.

Next, we wanted to ensure that spheres formed in our assay were enriched for stem/progenitor cell markers and that IHs were the cell of origin. As shown in Figure 2 and 2, these spheres were positive when immunostained with antibodies against human stem cell markers, SALL4 and CD133, and FDR (an early marker for endothelial cell precursors) as well as an IH signature marker, GLUT1. These data suggest that IH tumor spheres are enriched for stem cells or progenitor cells and that they express markers for early endothelial differentiation and are derived from IH tissue.

Since IH tumor spheres express markers for immature cell lineages and they are the IH cell of origin, we sought to examine if tumor spheres bear the stem cell properties of self renewal and differentiation. Following the preparation of single cell suspensions, replating of single cells from tumor spheres gave rise to secondary tumor spheres within 3 to 4 days. The process of dissociation to single cells was repeated and the formation of tertiary tumor spheres was observed (Figure 3 and 3) and tumor spheres could be passaged for prolonged periods of time (currently up to passage 30). This suggests that tumor spheres are able to self renew in the absence of differentiation signals. When tumor spheres were cultured in media favoring endothelial differentiation using VEGF, they would adhere and differentiate toward an endothelium- like morphology (Figure 3 and 3). The differentiating tumor spheres continually expressed GLUT1 (Figure 3). Expression of an endothelial marker, CD31, increased from day 3 to day 12 and expression of CD133 decreased over the same time period (Figure 3). This data illustrates that tumor spheres derived from IH tissues have stem cell properties and supports the hypothesis that IH is a stem cell disease.

Studies have indicated that tumor stem cells often contribute to or are associated with a microenvironmental niche. Understanding this process may allow development of a tumor stem cell therapy. We measured the concentration of growth factors, VEGF, EGF, FGF basic and G-CSF, in tumor sphere culture media at different time points (Figure 4). The level of the growth factors in the media was set as a background (blank). These values were automatically deducted by the software from the concentrations of the growth factors which were measured in the samples of interest. The concentrations were derived from the measured level of fluorescence in the samples by comparison with the fluorescence of the standard curve of known concentrations. The values for EGF, bFGF and G-CSF were below the range of the standard curve and considered as insignificant. Thus they are not presented on the graph. Media was collected on day 0, 5, 10 and 15 and analyzed for the level of growth factors by using the Human Growth Factor (GF) 4-Plex Kit (Invitrogen, Carlsbad, CA). The concentration of VEGF was undetectable at day 0 but became detectable at day 5 and was significantly increased during the 15-day measurement period (Figure 4). However, none of the other growth factors including EGF were detectable during the 15 days.

As shown above, IH tumor stem cells were able to specifically secrete VEGF and this study indicates that VEGF may play a significant role in tumor stem cell growth. Should this hypothesis be correct, inhibition of VEGF functions should inhibit IH stem cell growth [30‐32]. In fact, treatment of IH tumor stem cells with Rapamycin leads to inhibition of the growth of IH stem cells in our tumor sphere culture system. As shown in Figure 4, three different concentrations of Rapamycin, 1 nM, 2 nM and 4 nM, all in a biologically significant range, inhibited IH stem cell growth in a dose dependent manner. At days 0, 3, 6, 10, tumor sphere samples were collected, dissociated with collagenase IV, and the number of cells counted. Treatment with Rapamycin dramatically inhibited IH tumor stem cell growth.

In order to generate an in vivo hemangioma mouse model, and to determine whether IH stem cells have the ability to generate tumors in immunodeficient mice, dissociated IH tumor sphere cells were resuspended in Matrigel and injected subcutaneously into 6-week-old NOD/SCID mice (10⁶ cells/200 μl Matrigel/animal). Tumors were produced and as shown in Figure 5, all the tumors were GLUT1 positive indicating an IH cell of origin. Tumors exhibiting a proliferative cell phase were positive for CD31, an endothelial cell marker. These results demonstrated that IH stem cells can form hemangiomas in vivo. Tumors visible on day 10 exhibited mostly proliferative endothelial cells after H & E staining. Twenty days after injection, both blood vessels and fatty tissue became visible in H & E sections although proliferating cells were still present. By day 30, most of the cells injected into the animal had differentiated, and tumor involution occurred, with a high percentage of fatty tissue present (See Figure 5). This 30 day cycle of in vivo tumorigenis in mice represents the 5-9 year process seen in human cases of IH. The apparent faithful replication of hemangioma growth in this IH stem cell-induced in vivo hemangioma mouse model suggests that this system can be used as a tool to discover novel therapeutics specifically targeting IHs.