|Home | About | Journals | Submit | Contact Us | Français|
Infantile hemangioma (IH) is a common childhood vascular tumor. Although benign, some hemangiomas cause deformation and destruction of features or endanger life. The current treatments, corticosteroid or propranolol, are administered for several months and can have adverse effects for the infant. We designed a high-throughput screen to identify FDA-approved drugs that could be used to treat this tumor. Rapamycin, an mTOR inhibitor, was identified based on its ability to inhibit proliferation of a hemangioma-derived stem cell population, human vasculogenic cells we had previously discovered. In vitro and in vivo studies show that Rapamycin reduces the self-renewal capacity of the hemangioma stem cells, diminishes differentiation potential, and inhibits the vasculogenic activity of these cells in vivo. Longitudinal in vivo imaging of blood flow through vessels formed with hemangioma stem cells shows that Rapamycin also leads to regression of hemangioma blood vessels, consistent with its known anti-angiogenic activity. Finally, we demonstrate that Rapamycin-induced loss of stemness can work in concert with corticosteroid, the current standard therapy for problematic hemangioma, to block hemangioma formation in vivo. Our studies reveal that Rapamycin targets the self-renewal and vascular differentiation potential in patient-derived hemangioma stem cells and suggests a novel therapeutic strategy to prevent formation of this disfiguring and endangering childhood tumor.
Infantile Hemangioma (IH) is a common childhood tumor composed of disorganized blood vessels and immature cells (Mulliken, 1988) (Drolet et al., 1999). IH represents a disruption of neonatal vasculogenesis, the de novo formation of vessels from progenitor cells, and of angiogenesis, the sprouting of new vessels from pre-existing vasculature. Herein we show that Rapamycin, an inhibitor of the mTOR pathway, diminishes the self-renewal capacity of IH-derived stem cells (HemSCs), initiates the differentiation of HemSCs towards a perivascular cell phenotype and inhibits de novo formation of vessels from HemSCs. This anti-vasculogenic activity of Rapamycin, which is distinct from its known anti-angiogenic effects, is a novel discovery, with implications for the mTOR pathway in human neonatal vasculogenesis as well as for potentially new strategies to treat the most endangering IH.
Although a benign and usually harmless tumor, some IH deform or destroy facial features and/or obstruct vision and breathing. 40–80% result in permanent cutaneous residua, which can be disfiguring. Corticosteroids are traditional first-line therapy, however the adverse effects are numerous (Boon et al., 1999) and approximately 16% of hemangiomas do not respond (Bennett et al., 2001). Propranolol has recently been introduced as a safe and effective treatment for IH (Leaute-Labreze et al., 2008) (Siegfried et al., 2008). However, its use is not without risks, and not all tumors respond (Frieden and Drolet, 2009). Thus, there is a need for additional therapies that will shorten treatment duration or, ultimately, prevent problematic hemangiomas from ever forming.
We recently isolated stem cells (HemSCs) and endothelial cells (HemECs) from specimens of proliferating IH (Boye et al., 2001; Khan et al., 2008). HemSCs have a mesenchymal morphology, robust proliferation, undergo multi-lineage differentiation and form human blood vessels with features of IH when injected subcutaneously into nude mice (Khan et al., 2008). Corticosteroids inhibit this vessel formation by suppressing VEGF-A in HemSCs, while having no effect on HemECs or normal human cord blood-derived endothelial progenitor cells (cbEPCs)(Greenberger et al., 2010). However, corticosteroids have no effect on the proliferation of HemSCs. Therefore, we hypothesized a drug that inhibits proliferation of HemSCs might prevent the growth of proliferating IH. To pursue this, we screened chemical libraries to identify drugs that preferentially inhibit the proliferation of HemSCs.
We designed a proliferation assay to test in parallel HemSCs and bone marrow-derived stem cells (BM-MSCs). BM-MSCs share several characteristics with HemSCs: expression of mesenchymal markers, spindle-shaped morphology, and the potential to differentiate into bone, fat and cartilage. In contrast to HemSCs, BM-MSCs do not form vessels when injected as a single cell type in Matrigel into nude mice (Khan et al., 2008). The methods and results of the screen are provided in the Supplement (Supp. Fig 1, Supp. Table 1).
Of the nine compounds identified, we selected Rapamycin for further study because its activity was maintained at low concentrations and its favorable safety profile has been established in pediatric clinical trials (Sindhi et al., 2005). Rapamycin led to ~40% inhibition of HemSCs proliferation at concentrations of 1 ng/ml and higher, while BM-MSCs or normal human dermal fibroblasts (NHDF) were less sensitive (Fig 1A). Proliferating cell nuclear antigen (PCNA) served as an additional measure of proliferation (Prelich et al., 1987) (Hall and Levison, 1990). Rapamycin led to dose-dependent decrease in the percentage of PCNA-positive HemSCs with 80% suppression at 10ng/ml (Fig 1B). Viability was not affected by Rapamycin as seen by normal cell morphology visualized by phalloidin staining (Supp. Fig 2).
Rapamycin is an inhibitor of mTOR (mammalian target of Rapamycin). To verify that rapamycin inhibited mTOR in the HemSCs, we tested the phosphorylation status of two targets of mTOR signaling - serine/threonine kinase p70 S6 kinase and eIF-4EBP1 (Hara et al., 1997). Rapamycin caused a dose-dependent decrease in the constitutive and serum/growth factor stimulated levels of phosphorylated 4EBP1 (Fig 1C, D). To verify that mTOR is a relevant target in IH, we analyzed its upstream regulator AKT. Phosphorylated AKT was detected in proliferating IH tumor specimens. In contrast, very low expression was observed in involuting IH specimens (Fig 1E).
Next, we tested the effect of systemic Rapamycin in a two cell model of IH we reported previously(Greenberger et al., 2010). HemSCs and cbEPCs were mixed with Matrigel and injected subcutaneously in nude mice. This two-cell model has advantages over injecting HemSC alone: first it is more robust and thus more sensitive for detecting dose-response activity. Second, it allows one to test separately the effect of rapamycin on the HemSC versus cbEPCs, which are primary ECs isolated from cord blood that serve as prototypic young ECs. Rapamycin was administered daily by intra-peritoneal injection. Compared with the vehicle-injected mice, Rapamycin caused dose-dependent inhibition of vascularization of the Matrigel implants (Fig 2A–C). Anti-human CD31 staining verified that Rapamycin inhibited the de novo formation of human CD31-positive blood vessels: CD31-positive cells were detected, but without lumen formation (Fig 2D). Rapamycin also inhibited vessel formation when HemSC were implanted alone (Supp. Fig. 3A, B). Two other mTOR inhibitors, Everolimus (Novartis) and Temsirolimus (Wyeth/Pfizer), injected systemically as described for Rapamycin, also significantly inhibited formation of blood vessels (Fig 2E).
To test whether or not HemSCs are a target of Rapamycin, versus the cbEPCs or a murine host cell, we pre-treated HemSCs with Rapamycin for three days in vitro, washed out the drug, resuspended the Rapamycin-treated or control HemSCs with cbEPCs in Matrigel, and injected into mice. Pre-treatment of HemSCs with Rapamycin led to 80% inhibition of blood vessel formation (Fig 2F, G). Pre-treatment of cbEPCs had no effect (data not shown). These results demonstrate that HemSCs are directly targeted by Rapamycin.
We identified Rapamycin based on its preferential inhibition of HemSCs proliferation. We then tested whether a similar anti-proliferative effect, achieved with another drug, would be anti-vasculogenic. To do this, we used the anti-proliferative drug Roscovitine, a purine analogue and cyclin-dependent kinase inhibitor (De Azevedo et al., 1997) (Meijer et al., 1997). First, we titrated the effect of Roscovitine on HemSCs proliferation in vitro. Roscovitine, at 1.1 µg/ml (3.125µM) led to 55% inhibition of HemSCs proliferation (Fig 3A), a similar effect to that of Rapamycin (20 ng/ml) assayed in the same experiment (not shown). Next, we pre-treated HemSCs in vitro for three days with either Roscovitine or Rapamycin, washed out the drugs, and then injected the cells with untreated cbEPC into mice. Rapamycin inhibited microvessel density by 60%. In contrast, Roscovitine did not (Fig 3B, C), suggesting that pre-treatment of HemSC in vitro with an anti-proliferative drug was not sufficient to reduce the vasculogenic potential.
mTOR activity has been shown to be essential for maintaining stem cell self-renewal (Zhou et al., 2009). To test self-renewal capacity of HemSCs, we performed a clonogenic assay(Ingram et al., 2004). Pre-treatment of HemSCs with Rapamycin resulted in a 60% decrease in the number of HemSC colonies formed at 9 days (Fig 4A). Moreover, stratifying the colonies into groups with increasing cell number demonstrated that Rapamycin pre-treatment abolished the appearance of colonies with >500 cells, reduced the percentage of colonies with 100–250 or 250–500 cells/colony and increased the percentage of colonies with 10 or fewer (Fig 4 B). These results demonstrate that pre-treatment of cells with Rapamycin for four days resulted in a diminished number of highly proliferative clones.
Clonally-derived populations of HemSCs have the ability to differentiate into ECs and adipocytes in vitro and in vivo (Khan et al., 2008). We assayed whether the multi-lineage differentiation of HemSCs could be altered by Rapamycin. Rapamycin blocked expression of the adipogenic markers lipoprotein lipase (LPL), CCAAT enhancer binding protein α (C/EBP-α) (Darlington et al., 1998) and the accumulation of lipid droplets (Oil Red O) when HemSC were subjected to an adipogenic differentiation protocol (Fig 4C, D). No effect was found on the ability of HemSCs to differentiate, under empirically-defined in vitro conditions, towards an endothelial phenotype (data not shown). These observations indicated that Rapamycin inhibits the adipogenic differentiative activity of HemSCs.
The reduction in self-renewal capacity and inhibition of adipogenesis suggested that rapamycin “pushes” the HemSCs to maturate past the state of multipotency. In accord with this hypothesis, HemSCs incubated with Rapamycin have an elongated morphology compared to the control cells (Supp. Fig 2). Therefore, we tested whether or not Rapamycin-treated HemSCs differentiate towards a smooth muscle cell (SMC) or pericytic phenotype. As a positive control, we used endothelial co-culture to stimulate HemSCs towards SMC/pericytic differentiation. Similar to co-cultures of mesenchymal progenitor cells and ECs (Hirschi et al., 1998) (Melero-Martin et al., 2008), when clonal HemSCs are co-cultured with ECs, an up-regulation of SMC/pericyte markers in the HemSCs occurs starting on day three (Boscolo et al., 2011). Incubation of HemSCs with Rapamycin for four days led to upregulation of smooth muscle myosin heavy chain (smMHC) and SM22-alpha transcript levels (Li et al., 1996) (Fig 4E). Similarly, SM22-alpha and α-smooth muscle actin (αSMA) were increased at the protein level (Fig 4F). Expression levels of calponin, NG2 and PDGF-Rβ were unchanged (Supp. Figure 4A). To test for functional features of SMCs, we analyzed calcium influx in HemSC treated with Rapamycin or DMSO for six days. The results indicate Rapamycin-treated HemSCs express L-type voltage gated channels (Supp. Fig 4B and Supp. Movies 1,2). To verify that this mesenchymal maturation has a negative effect on vasculogenic capability, we plated HemSCs alone or in co-culture with cbEPCs for four days. After four days, cbEPCs were separated from the HemSCs using anti-CD31-conjugated magnetic beads. The differentiated HemSCs were combined with fresh cbEPCs, suspended in Matrigel and injected into mice (schematic in Fig 4 G). This prior co-culture step led to a 75% decrease in the microvessel density (Fig 4 H, I). These findings indicate that, independent of the means by which mesenchymal differentiation was induced, either rapamycin pre-treatment or endothelial co-culture, the vasculogenic potential of HemSCs was diminished.
Rapamycin is known to have an anti-angiogenic effect on ECs in pathological settings (Guba et al., 2002) (Geerts et al., 2008) (Phung et al., 2007). To study this effect, we tested Rapamycin on cbEPCs. Rapamycin suppressed growth of cbEPCs, which was not rescued by exogenous VEGFA or placental growth factor (PLGF)-1. This observation suggests, as expected, that proliferation and response to VEGF and PLGF-1 mitogenic signals are mTOR-dependent. (Supp Fig 5A, B). In addition, Rapamycin caused a dose-dependent suppression of endogenous PLGF-1 in human ECs, including HemECs. (Supp Fig 5C, D). Thus, we hypothesized that Rapamycin, able to target both the stem cells and ECs, could cause regression of pre-existing vessels formed from HemSCs and cbEPCs. To test this, we used contrast-enhanced micro-ultrasonic imaging to measure blood flow in the Matrigel implants. First, we verified the sensitivity and specificity of the micro-ultrasonic imaging in our model (Supp. Fig 5). For the experiment, mice were injected with cells and imaged at day 7 to record baseline blood flow in the implants. Mice were then randomized into two groups: Rapamycin, 2mg/kg by i.p. injection or vehicle alone by i.p. injection. At day 17, follow-up imaging was performed. Mice injected with vehicle had either a stable or increased blood flow in the implant (Fig 5A, 5B, red lines). Mice injected with Rapamycin had reduced blood flow (Fig 5B, grey lines). The average change in contrast intensity change between 7 and 17 days is shown in Fig 5C. (Movies of the contrast agent (blood) flow through the implants are provided in the Supplement, Movies 3–6). In accord with the imaging results, Matrigel implants of Rapamycin-treated mice were less vascularized macroscopically and by histological analysis (Fig 5D, E).
To investigate whether corticosteroids and Rapamycin share common mechanisms for suppressing HemSCs-mediated vascularization, we evaluated the effect of Rapamycin on expression of a panel of 43 pro-angiogenic proteins. In contrast to corticosteroid treatment, Rapamycin did not block the expression of VEGF-A by HemSCs at the mRNA or protein level (Figure 6 A, B). Indeed, we found no overlap between the effects of rapamcyin and dexamethasone on expression of the pro-angiogenic cytokines. Dexamethasone suppressed VEGF-A, interleukin-6 (IL-6), matrix metalloproteinase-1 (MMP-1), monocyte chemoattractant protein-1 (MCP-1) and urokinase plasminogen activator receptor (UPAR)(Greenberger et al., 2010). In contrast, Rapamycin downregulated angiogenin and PLGF-1 in HemSCs and cbEPCs (Fig 6C, verification of the array results in Supp. Fig 7).
Because corticosteroid and Rapamycin appeared to have non-overlapping effects on HemSCs and ECs, we hypothesized that the two drugs could “collaborate”, i.e., vascularization would be inhibited at low doses of each drug, comparable to either drug used alone at high dose. To test this concept, mice were given i.p. injections of low dose of dexamethasone (0.25 mg/kg), low dose of Rapamycin (0.25 mg/kg) or the combination of both. Combining Rapamycin and dexamethasone enhanced their potency, as evidenced by the inhibition of vasculogenesis that was comparable to high dose of either drug alone (1 and 2 mg/kg of Rapamycin or dexamethasone, respectively, Figure 6D, E). The efficacy of the two drugs at low doses suggests targeting the HemSCs with dexamethasone and HemSCs and HemECs with Rapamycin may be an effective therapy for endangering IH.
To identify novel drug candidates for the common childhood tumor IH, we screened chemical libraries for compounds that would inhibit the vasculogenic stem cell, HemSCs, isolated from IH tumor tissue. We identified Rapamycin, an mTOR inhibitor, based on its relative selectivity for HemSCs over BM-MSCs, a normal human stem cell with a similar phenotype. Rapamycin also suppressed self-renewal and modified the differentiative status of the HemSCs. Finally, Rapamycin prevented HemSCs, either alone or combined with cbEPCs, from forming blood vessels in vivo and increased regression of already formed vessels. Our data suggest that the primary action of Rapamycin is to cause the HemSCs to lose their stem cell properties; this represents an entirely unique mechanism for blocking human post-natal vasculogenesis. We base this on the experiments in which pre-treatment of HemSCs for defined periods of time diminished self-renewal in vitro and inhibited vasculogenesis in vivo. Furthermore, inhibiting proliferation per se with Roscovitine was not sufficient to block vasculogenesis. An alternative mechanism to Rapamycin's anti-vasculogenic effect in vivo might be the inhibition of the interaction between HemSC and cbEPCs. Evidence for the mTOR pathway was shown by the activation of the upstream, positive modulator AKT in proliferating IH tumor specimens. Finally, we show that Rapamycin exerts non-overlapping, inhibitory effects on HemSCs and ECs compared to corticosteroids (See schematic in Supplemental Fig. 8).
The role of phosphatase and tensin homologue (PTEN) in controlling the homeostasis and self renewal of stem/progenitor cells in multiple tissues is well established (Hill and Wu, 2009) (Dubrovska et al., 2009; Groszer et al., 2001; Inoue-Narita et al., 2008). However, the contribution of pathways downstream of PTEN, and especially the AKT/mTOR axis, is less defined. Both over-activation and suppression of mTOR activity has been shown to affect selfrenewal of stem cells. When human embryonic stem cells (hESCs) are cultivated under self-renewal conditions, mTOR inhibition suffices to disrupt pluripotency and trigger mesoderm and endoderm activities (Zhou et al., 2009). In hematopoeitic stem cells (HSC), over-activation of the mTOR pathway by conditional deletion of PTEN or tuberous sclerosis protein 1, or by constitutive activation of AKT, (Kharas et al., 2010) led to loss of HSC quiescence and reduced long-term HSC function. Rapamycin was shown to rescue HSC function (Yilmaz et al., 2006) (Chen et al., 2008; Zhang et al., 2006). The effects of rapamycin treatment on IH-derived stem cells suggest that tightly-controlled mTOR activity is also essential for post-natal, tissue-resident stem cells.
IH appears in the first weeks of life and with higher incidence in low birth weight pre-term infants (Amir et al., 1986). Following the proliferative phase, there is spontaneous regression leading to a fibrofatty residuum. This unique life cycle suggests that this tumor derives from immature progenitor cells, perhaps of neural crest origin (Haggstrom et al., 2006) (Wu et al., 2008), that have not accomplished their full differentiation plan. If that is the case, an intervention that will push these cells to lose their “stemness”, perhaps using Rapamycin or another modulator of mTOR activity, is an appealing strategy to prevent growth and/or induce early involution.
Taken together, our findings suggest two potential advantages of mTOR inhibitors for the treatment of non-responsive IH. First, our finding that corticosteroids and Rapamycin act by distinct mechanisms to suppress the vasculogenic potential of HemSCs suggests that mTOR inhibitors can be used as “steroid sparing” agents, reducing the required dose of corticosteroid. Moreover, due to the induction of HemSCs differentiation by Rapamycin, the treatment period might be shortened. Rapamycin has adverse effects, but in most cases, these effects are dose or concentration-dependent(Sindhi et al., 2005). Second, in some patients, the permanent fibrofatty residuum after tumor involution can be disfiguring. We found that Rapamycin blocks HemSCs adipogenic differentiation, in accordance with previous reports demonstrating a role for intact mTOR activity for adipogenic differentiation (Zhang et al., 2009) It might then be the case that treating proliferating IH with Rapamycin would prevent this sequela and be advantageous over a natural involution.
Specimens of IH were obtained under a protocol approved by the Committee on Clinical Investigation, Children’s Hospital Boston; the clinical diagnosis was confirmed in the Department of Pathology. Informed consent was obtained for use of IH specimens, according to the Declaration of Helsinki. The derivation, sources and culture conditions of HemSCs and other cells used in this study have been reported (Khan et al., 2008; Khan et al., 2006; Melero-Martin et al., 2007). For co-culture experiments, HemSCs and cbEPCs (104 cells per cm2) were plated on fibronectin-(1µg/cm2) coated dishes at a 1:1 ratio. At the indicated day, cells were trypsinized to prepare a single cell suspension. Removal of cbEPCs was performed by Dynabeads® CD31 Endothelial Cells (Invitrogen) according to the manufacturer’s instructions. Differentiation assays towards endothelial cells and adipocytes were done as reported (Khan et al., 2008).
Nude/Nude mice were purchased from Massachusetts General Hospital animal facility. Procedures were approved by the Animal Care and Use Committees of the Children’s Hospital, Boston. Anesthesia was induced with 4% isofluorane and maintenance in 1–3% oxygen. Mice were anesthetized once for up to 5 min. The one cell and two cell models of IH were performed as reported (Greenberger et al., 2010; Khan et al., 2008) and microvessel density (MVD) was assessed by counting red blood cell-filled lumens(Melero-Martin et al., 2007). Values reported for each experimental condition correspond to the average MVD value ± standard error of the mean (SEM) obtained from all the individual animals.
Rapamycin, Everolimus and Temsirolimus were purchased from LC Laboratories. Drugs were dissolved in DMSO followed by solvent solution (5% polyethylene glycol, 0.2% carboxymethylcellulose and 0.25% Tween-80 in H2O). Aliquots were kept in −20°C. Mice were given daily intra-peritoneal (i.p.) injections containing 200µl of drug or the vehicle alone.
Contrast-enhanced high resolution ultrasonic (US) imaging of the blood flow in Matrigel implants was performed using a Vevo 770 System with a 40-MHz RMV scanhead (Visual Sonics) (Loveless et al., 2009; Olive et al., 2009). MicroMarker non-targeting contrast agent (Visualsonics) was prepared according to manufacturer’s guidelines. For imaging, mice were anesthetized with 2% isofluorane, coupling gel was applied to cover the entire region of interest, and 2-D B-mode scout images established the central slice of the Matrigel. The transducer was placed such that the 6 mm focal spot converged at the center of the Matrigel. Baseline images were acquired in Contrast Mode, and a 50µL bolus of contrast agent suspension was administered via tail vein catheter during acquisition of a second contrast video. The baseline image was subtracted from the contrast image, and the difference was displayed with a contrast setting of 90 and a threshold setting of 0. For these experiments (Figure 5), mice were subjected to isofluorane-mediated anesthesia two times (days 7 and 15) for approximately 15 minutes.
HemSCs were incubated with Rapamycin 20ng/ml or DMSO, as a control, for 4 days. Following washout of the drug, cells were serially diluted to 8 cells per ml and plated in 384 well plates, 40µl/well. Medium was replaced every 48 hours. At day 9, cells were fixed and stained with Hoechst. Following image capture by automated fluorescence microscopy, colony formation and the number of cells/colony were determined by quantifying the number of wells with Hoechst-stained nuclei and the number of Hoechst-stained nuclei/well using MetaXpress Software.
Data was analyzed by Student’s two-tailed t test. One-way-ANOVA was used to test for differences among multiple groups. Differences were considered significant at P < 0.05
High-throughput screening capability was provided by the ICCB-Longwood (ICCB-L) facility at Harvard Medical School. We thank Caroline Shamu, Ph.D., Director, ICCB-L, for advice with the screen. We also wish to thank Camille L. Stewart for pilot experiments and helpful discussions. The work was supported by a Translational Research Program grant from Children’s Hospital Boston, R01 HL096384, P01 P01 AR48564 (JB), the Talpiot Medical Leadership Program, Sheba Medical Center, Israel (S.G.), Harvard Skin Diseases Pilot Study Grant (S.G.), and the John Butler Mulliken Foundation.
Conflict of Interest
The authors have no conflicts of interests to declare.