COSMC Is Overexpressed in Proliferating Infantile Hemangioma and Enhances Endothelial Cell Growth via VEGFR2

Infantile hemangiomas are localized lesions comprised primarily of aberrant endothelial cells. COSMC plays a crucial role in blood vessel formation and is characterized as a molecular chaperone of T-synthase which catalyzes the synthesis of T antigen (Galβ1,3GalNAc). T antigen expression is associated with tumor malignancy in many cancers. However, roles of COSMC in infantile hemangioma are still unclear. In this study, immunohistochemistry showed that COSMC was upregulated in proliferating hemangiomas compared with involuted hemangiomas. Higher levels of T antigen expression were also observed in the proliferating hemangioma. Overexpression of COSMC significantly enhanced cell growth and phosphorylation of AKT and ERK in human umbilical vein endothelial cells (HUVECs). Conversely, knockdown of COSMC with siRNA inhibited endothelial cell growth. Mechanistic investigation showed that O-glycans were present on VEGFR2 and these structures were modulated by COSMC. Furthermore, VEGFR2 degradation was delayed by COSMC overexpression and facilitated by COSMC knockdown. We also showed that COSMC was able to regulate VEGF-triggered phosphorylation of VEGFR2. Our results suggest that COSMC is a novel regulator for VEGFR2 signaling in endothelial cells and dysregulation of COSMC expression may contribute to the pathogenesis of hemangioma.


Introduction
Infantile hemangiomas are characterized by an aberrant growth of endothelial cells in the infant age. They are the most common tumors in infants and children. The incidence is ranged from 1.1 to 2.6%, with the highest estimates approaching 10 to 12% [1]. The life cycle of hemangiomas differs from that of most tumors in that it progresses from a phase of rapid proliferation followed by spontaneous involution [2]. During the proliferative phase, hemangiomas are composed of densely packed endothelial cells that form small capillaries [2]. While in the involuting phase, the capillaries have larger lumens with thickened capillary basement membranes, less-plump endothelial cells, and fibrofatty replacements of the lesion [3].
Angiogenesis is crucial for blood vessel formation and embryonic development. Vascular endothelial growth factor (VEGF) is the most critical growth factor expressed at the sites of angiogenesis and its expression levels are closely correlate with the spatial and temporal events of blood vessel development [4]. VEGF receptor 2 (VEGFR2, KDR/Flk-1) is the receptor that mediates endothelial cell responses to VEGF. Interestingly, VEGFR2-dependent signaling has been found to be upregulated in hemangioma endothelial cells [5], however, the mechanisms accounting for the increased VEGFR2 activity in hemangiomas remain unclear.
Glycosylation is the most common post-translational modification of proteins. Changes in carbohydrates, such as Tn and T antigens, are often correlated with tumor progression and prognosis in human cancers [6,7]. Two major types of protein glycosylation in mammalian cells exist: N-linked and O-linked. The most frequently occurring O-glycosylation is the mucin-type, initiated by the transfer of N-acetylgalactosamine (GalNAc) to serine or threonine residues [8]. Tn antigen has a relatively simple structure composed of GalNAc with a glycosidic linkage to serine/ threonine residues on glycoproteins (GalNAc-Ser/Thr). The Tn antigen is further modified to become T antigen (Galbeta1,3Gal-NAc) by core 1 ß1,3-galactosyltransferase (C1GALT1 or Tsynthase) in the Golgi apparatus of cells [9]. Expression of an active T-synthase is uniquely dependent on the molecular chaperone COSMC [10].
Mice deficient in T synthase or COSMC are both embryonically lethal [11,12]. Mosaic mice with dominant deletion of COSMC develop hemorrhages in the lung and GI tract, bloody chylous ascites, and growth retardation, which resembles the observations in mice with a conditional deletion of T-synthase in endothelial and hematopoietic cells [12]. These findings suggest that deficiency in endothelial cell O-glycans leads to blood/ lymphatic misconnections and that T synthase and its chaperone COSMC are required for correct development of blood vessels [13]. Although several lines of evidence suggest critical roles of COSMC in endothelial cells the expression and functions of COSMC in hemangiomas are still unknown.
In the present study, we showed that COSMC and T antigen were overexpressed in human proliferating hemangiomas. We also demonstrated that COSMC overexpression increased cell growth in HUVECs and PI3K inhibitor, LY294002 significantly blocked the enhanced cell growth. Our results are the first to show that Oglycans are present on VEGFR2. Furthermore, COSMC overexpression modulated O-glycosylation of VEGFR2 and increased VEGF-triggered phosphorylation of VEGFR2. Conversely, COSMC knockdown inhibited endothelial cell growth and suppressed VEGF-mediated signaling. These findings suggest that COSMC is a novel regulator for VEGFR2 signaling and endothelial cell proliferation, which provides new insights into the mechanism of hemangioma development.

Clinical tissue collection
Human hemangioma tissues were obtained from the Department of Surgery, Cathay General Hospital, Taipei, Taiwan. The local hospital ethics committee approved the use of human tissues for this study, and written consent was obtained from patients before the collection of samples. For immunohistochemistry, specimens were fixed in 4% (w/v) paraformaldehyde/PBS at 4uC overnight. For RNA extraction, specimens were soaked in RNAlater (Qiagen, Tokyo, Japan) at 4uC overnight and then stored at 220uC.

Immunohistochemistry
Human hemangioma tissue sections were deparaffinized in xylene and rehydrated in a series of graded alcohols. After quenching the activity of endogenous peroxidase with 1% H 2 O 2 in phosphate-buffered saline (PBS) for 10 min, the sections were rinsed three times with PBS and then incubated with 5% non-fat milk/PBS for 30 min to reduce non-specific bindings. Sections were incubated with an anti-COSMC polyclonal antibody (1:30, Sigma, St. Louis, MO), biotinylated peanut agglutinin (PNA) (1:250, Vector Laboratories, Burlingame, CA), or biotinylated Vicia villosa agglutinin (VVA) (1:500, Vector Laboratories, Burlingame, CA) diluted with 1% bovine serum albumin (BSA)/ PBS for 16 h at 4uC. After rinsing twice with PBS, Super Sensitive TM Link-Label immunohistochemistry Detection System (BioGenex, San Ramon, CA) was used and the specific immunostaining was visualized with 3,3-diaminobenzidine liquid substrate system (Sigma, St. Louis, MO). All sections were counterstained with hematoxylin for 1 min and mounted with UltraKitt (Mallinckrodt Baker, Inc., Phillipsburg, NJ). Negative controls were performed by replacing the primary antibody with a control IgG at the same concentration.

Real-time RT-PCR
Total cellular RNA was isolated from hemangioma tissues or cells grown to 70% confluence by use of the Trizol reagent (Invitrogen, Carlsbad, CA) according to manufacturer protocols as previously described [14]. For cDNA synthesis, 2 mg of total RNA were used as templates in a 25 ml reverse transcription reaction. For GAPDH detection, sense and anti-sense primers were 59-ACAGTCAGCCGCATCTTCTT -39 and 59-GA-CAAGCTTCCCGTTCTCAG -39, respectively. For detection of COSMC, sense and anti-sense primers were 59-TTTGAAGGGTGTGATGCTTG -39 and 59-ATGCGCT-CATCCTCTGAAAT -39, respectively. PCR products were separated by 1% agarose gels. For real-time PCR reactions, quantitative PCR System Mx3000P (Stratagene, La Jolla, CA) was used according to manufacturer's protocol. Briefly, reaction was performed in a 25-ml volume with 2 ml cDNA, 400 nM each of sense and anti-sense primers, and 12.5 ml Brilliant R SYBR R Green QPCR Master Mix (Stratagene, La Jolla, CA). PCRs were incubated for 15 min at 95uC followed by 40 amplification cycles with 30-s denaturation at 95uC, 50-s annealing at 54uC, 30-s extension at 72uC. Samples were analyzed in triplicate, and product purity was checked through dissociation curves at the end of real-time PCR cycles. Relative quantity of gene expression normalized to b-actin was analyzed with MxPro Software (Stratagene, La Jolla, CA).

Trypan blue exclusion assay
Cells (4610 4 ) were seeded in 6-well plates with RPMI 1640 containing 10% FBS (PAA Laboratories, New Bedford, MA). Viable cells in triplicate wells were determined at 24 h intervals for 72 h using hemocytometer with trypan blue exclusion staining (Sigma, St. Louis, MO).

Lectin pull down assay
For lectin pull down assays, 200 mg of cell or tissue lysates were incubated with PNA or VVA agarose beads (Vector Laboratories, Burlingame, CA) overnight at 4uC. The pulled down proteins were then subjected to Western blotting.

Chemical inhibition
The MEK inhibitor PD98059 and PI3K inhibitor LY294002 (Calbiochem, San Diego, CA), and VEGF receptor tyrosine kinase inhibitor SU1498 (Santa Cruz Biotechnology, Santa Cruz, CA) were dissolved in dimethyl sulfoxide (DMSO) to prepare a 10 mM of stock solution. For the inhibition of cell growth, 10 mM PD98059, 5 mM PY294002, or 10 mM SU1498 were used. DMSO was used for a solvent control.

VEGFR2 degradation assays
Cells were cultured on 12-well plates and serum starved for 4 h. Cycloheximide (10 mg/ml) (Sigma, St. Louis, MO) and 20 ng/ml of VEGF were added to the cells for different time points. Cells were lysed in lysis buffer containing Tris 20 mM, pH 8.0, NaCl 137 mM, 1% NP-40, 10% glycerol, Na 3 VO 4 2 mM, b-glycerophosphate 2 mM, PMSF 2 mM, and 1% protease inhibitor cocktail (Sigma, St. Louis, MO). VEGFR2 protein levels were determined by Western blotting. b-actin was used as an internal control. The 0 h-control was set to 100% and the detected level of VEGFR2 was calculated as the percentage of 0 h-control for each time point.

Statistical analyses
Student's t-test was used for statistical analyses. Data are presented as means 6 S.D. Chi-square tests were used to test associations between COSMC and T antigen expression. P,0.05 was considered significant.

COSMC is overexpressed in proliferating hemangiomas
To investigate the expression of COSMC in hemangiomas, immunohistochemistry was performed. Our results showed that COSMC expression level was the highest in proliferating hemangiomas, followed by involuting hemangiomas, and the lowest in involuted hemangiomas (Fig. 1). We also observed that COSMC expression in the surrounding normal blood vessels was not detectable or relatively weak, similar to that found in involuted hemangiomas (Fig. S1)(if accepted, production will need this reference to link the reader to the figure). These findings suggest that COSMC is overexpressed in proliferating hemangiomas compared with involuted hemangiomas which entails that COSMC may play a role in hemangioma development.
Since COSMC is a chaperone for T synthase to synthesize T antigen which is significant in tumor malignancy [16], we analyzed whether T antigen was also overexpressed in proliferating hemangiomas. Immunohistochemical analyses with PNA indicated that T antigen was also overexpressed in proliferating hemangiomas compared with involuted hemangiomas (Fig. 1). Furthermore, our data showed that proliferating hemangiomas expressed Tn antigen, as revealed by Vicia villosa agglutinin (VVA) staining ( Fig. S2) (if accepted, production will need this reference to link the reader to the figure). The staining was augmented after neuraminidase treatment, indicating the presence of sialyl Tn structures in proliferating hemangiomas. These results suggest that both COSMC is overexpressed in the proliferating phase of infantile hemangiomas compared with the involuted phase, which is associated with increased T antigen expression.

COSMC overexpression enhances T synthase and surface T antigen expression in HUVECs
To investigate the roles of COSMC in endothelial cells, we first analyzed COSMC expression in HUVECs and proliferating hemangiomas. Real-time RT-PCR showed that HUVECs expressed lower levels of COSMC compared with proliferating infantile hemangiomas ( Fig. 2A). COSMC overexpression or knockdown in HUVECs (Fig. 2B) and human endothelial cell line EA.hy926 (Fig. 2C) were achieved by transfection with COSMC/ pcDNA3.1 or COSMC-specific siRNA. . We found that COSMC overexpression enhanced T synthase expression at protein levels, which in turn increased cell surface T antigen expression as reflected by an increased PNA binding (Fig. 2D, left panel). In contrast, COSMC knockdown decreased T synthase expression but increased Tn expression, as revealed by VVA binding (Fig. 2D,  right panel). These findings suggest that COSMC can increase protein levels of T synthase and enhance T antigen expression in endothelial cells.

COSMC overexpression enhances cell growth of HUVECs
Since increased cell proliferation is a prominent phenotype for proliferating hemangiomas, we analyzed the effects of COSMC on cell proliferation in endothelial cells. Our results showed that overexpression of COSMC significantly enhanced cell proliferation in HUVECs by trypan blue exclusion assay (Fig. 3A). Although the increased levels of cell growth were dependent on passage times of HUVECs we consistently found that COSMC significantly increased cell growth at day 2 after transfection. Furthermore, knockdown of COSMC suppressed cell growth of HUVECs (Fig. 3B). Similar results were obtained by MTT assays (Fig. S3) (if accepted, production will need this reference to link the reader to the figure). To confirm the role of COSMC in cell growth of endothelial cells similar assays were performed in EA.hy926 cells. Comparable results were obtained in that COSMC overexpression enhanced cell growth (Fig. 3C), whereas COSMC knockdown suppressed cell growth (Fig. 3D). These results suggest that COSMC plays a role in regulating endothelial cell proliferation.

COSMC overexpression modulates O-glycans on VEGFR2 and enhances phosphorylation of VEGFR2 in HUVECs
Since AKT and ERK are two major signaling pathways known to regulate HUVEC proliferation we analyzed whether they were modulated by COSMC. HUVECs transfected with COSMC/ pcDNA3.1 or COSMC siRNA, and their controls were serum starved for 4 h and then treated with 10% FBS for 10 min. Our results showed that phosphorylation of both AKT and ERK were increased by COSMC overexpression (Fig. 4A). COSMC knockdown suppressed phosphorylation of AKT and ERK (Fig. 4B). To evaluate the role of AKT and ERK in cell proliferation, HUVECs were treated with chemical inhibitors LY294002 and PD98059, respectively. MTT assays showed that LY294002 dramatically inhibited COSMC-enhanced cell growth (Fig. 4C, left panel). PD98059 inhibited cell proliferation to a lesser extent (Fig. 4C, right panel). These results suggest that COSMC enhances phosphorylation of AKT and ERK and the PI3K-AKT signaling pathway plays a predominant role in HUVEC proliferation.

VEGFR2 is identified as an O-glycosylated protein and its O-glycosylation can be modified by COSMC in HUVECs
To further investigate the molecular mechanism by which COSMC increased cell proliferation we attempted in identifying glycoproteins with increased T antigens by COSMC overexpression. Since VEGFR2 is a crucial endothelial growth factor receptor with a molecular mass around 220 kDa (data not shown) and a protein with increased T antigen was coincidentally found at 220 kDa (Fig. 2D, left panel) , we tested whether VEGFR2 was the mediator to enhance endothelial cell growth by COSMC overexpression. Consistent with previous observations, we found that VEGFR2 inhibitor SU1498 dramatically inhibited HUVEC growth (data not shown). Interestingly, our results showed that VEGFR2 could be easily pulled down by O-glycan-binding lectins VVA and PNA after neuraminidase treatment ( Fig. 5A and 5B). Moreover, binding of PNA to VEGFR2 in COSMC-overexpressing HUVECs was increased, whereas binding of VVA to VEGFR2 was decreased (Fig. 5A). In contrast, knockdown of COSMC decreased PNA binding to VEGFR2 (Fig. 5B). In addition, we found that VEGFR2 in primary hemangioma tissues could be pulled down by PNA after removal of sialic acids (Fig. 5C) indicating that VEGFR2 carries O-glycans in vivo. These results suggest that VEGFR2 is O-glycosylated and its O-glycosylation is modulated by COSMC in HUVECs.
We next investigated whether COMSC could regulate VEGFR2 and its downstream signaling. Western blotting showed that VEGF-triggered phosphorylation of VEGFR2 was increased in COSMC-overexpressing HUVECs (Fig. 5D). We also observed that phosphorylation of AKT and ERK was enhanced. In contrast, knockdown of COSMC inhibited VEGF-mediated signaling, as revealed by decreased phosphorylation of VEGFR2 and its downstream signaling (Fig. 5E). These findings suggest that COSMC overexpression can enhance VEGF-induced phosphorylation of VEGFR2.

COSMC regulates VEGFR2 degradation in endothelial cells
To assess whether O-glycosylation could modulate VEGFR2 properties, we performed protein degradation assays in HUVECs. HUVECs were treated with cycloheximide to block protein synthesis and 20 ng/ml of VEGF was added at different time points to induce VEGFR2 internalization and degradation. Our results showed that VEGF was able to induce VEGFR2 protein degradation and COSMC overexpression delayed the degradation of VEGFR2 (Fig. 6A). Conversely, knockdown of COSMC facilitated degradation of VEGFR2 (Fig. 6B). These results suggest that COSMC can regulate VEGF-triggered degradation of VEGFR2.

Discussion
Tn is the simplest mucin-type O-glycan, and could be further modified to generate sialyl Tn or T-related antigens, which are commonly found in tumorous tissues [7]. These tumor-associated carbohydrate antigens have been reported to be expressed in a variety of epithelial cancers, including breast, colon, lung, bladder, cervical, ovarian, stomach, and prostate, and they are used for development of cancer vaccines [7]. Expression of Tn/T-related carbohydrates often correlates with tumor progression and poor prognosis of diseases [16][17][18]. In consistent with previous reports, we found that T, Tn, and sialyl Tn were highly expressed in proliferating hemangiomas, whereas no or relatively low expression was observed in the endothelium of involuted hemangiomas or normal liver. The abundance of Tn and T carbohydrate structures may be of promising targets for developing diagnostic and therapeutic reagents for infantile hemangiomas.
In our study, we found that COSMC was overexpressed in proliferating hemangiomas compared with involuted hemangio-mas. COSMC functions as a chaperone for folding and stabilizing T synthase, which is essential for its enzymatic activity to generate T antigen [15]. In addition, COSMC overexpression in HUVECs enhanced expression levels of T synthase protein, which in turn increased T antigen expression on cell surfaces. At molecular levels, we also demonstrated that COSMC overexpression increased sialyl T antigen expression on VEGFR2. Therefore, it is reasonable to conclude that COSMC exerts its biological functions mainly via increased protein levels of T synthase.
Our data suggested that the major O-glycans of VEGFR2 in HUVECs could be sialyl Tn and sialyl T since VEGFR2 could be easily pulled down by VVA and PNA after removal of sialic acids. Moreover, COSMC overexpression enhanced PNA binding, while VVA binding was decreased. We also found that PNA was able to bind to neuraminidase-treated VEGFR2 of primary hemangiomas, suggesting that VEGFR2 expresses sialyl T antigens in vivo. However, we did not observe enhanced sialyl T antigen expression in proliferating hemangiomas compared with involuted hemangiomas. One of the explanations could be that several factors, in addition to COSMC, collaboratively modulate the expression of T antigen on VEGFR2 in vivo. Interestingly, prediction of Oglycosylation sites using NetOGlyc 3.1 indicates that there are four potential O-glycosylation sites in the extracellular domain of VEGFR2. These results strongly suggest the presence of O-glycans on VEGFR2. To our knowledge, we are the first to report that VEGFR2 expresses O-glycans and changes in the O-glycans can modulate activity, protein stability, and signaling of VEGFR2. It will be of great interest to further investigate the exact structures and sites of O-glycans on VEGFR2 to understand how Oglycosylation modulates VEGFR2 activity.
VEGFR2 is an important surface receptor to regulate cellular properties of endothelial cells, including cell proliferation and angiogenesis [19]. The significance of VEGFR2 in infantile hemangioma cells has been documented, in which higher VEGFR2 activity, but not protein levels, has been suggested to be a key determinant for aberrant growth of hemangioma cells [5,20,21]. Mutations have been found in the kinase domain of VEGFR2, resulting in an increased kinase activity of VEGFR2 [5,22]. However, these mutations are found only in rare cases of infantile hemangiomas. One of the other possibilities for the increased VEGFR2 activity may result from post-translational modification of VEGFR2. Our data showed that O-glycosylation modification of VEGFR2 could enhance VEGFR2 phosphorylation and increased its downstream signaling. Therefore, this study provides novel insights into the significant role of O-glycosylation in VEGFR2 activity and the molecular pathogenesis of infantile hemangiomas.
In this study, we showed that COSMC was overexpressed in proliferating hemangiomas, which was associated with increased T antigen expression. COSMC overexpression increased AKT and ERK1/2 signaling and enhanced endothelial cell growth in HUVECs. Mechanistic investigation showed that COSMC changed O-glycosylation and degradation of VEGFR2, and enhanced VEGF-mediated phosphorylation of VEGFR2 and its downstream signaling. These results indicate that the COSMCenhanced endothelial growth is at least, in part, mediated through activation of VEGFR2. Since there are several potential acceptor substrates for T synthase in endothelial cells, it has to be noted here that other pathways may act in concert with VEGFR2 to regulate endothelial cell proliferation. In conclusion, our results suggest that COSMC is a novel regulator for VEGFR2 signaling in endothelial cells and may contribute to the aberrant growth of human infantile hemangioma. This study opens up avenues for treating VEGFR2-dependent diseases by targeting not only the receptors themselves but their glycosylation regulators.

Supporting Information
Figure S1 Immunohistochemistry of COSMC in normal blood vessels. The arrows show relatively weak COSMC staining of the surrounding normal blood vessels in proliferating hemangioma, using tissue section obtained from the same tissue sample as used in Fig. 1A. Stars indicate sweat glands with strong COSMC staining. Scale bar, 50 mm. (TIF) Figure S2 Immunohistochemistry of VVA in proliferating hemangioma and liver. The proliferating infantile hemangioma, which is the same as in Fig. 1A, and human liver tissue with or without neuraminidase treatment were stained with VVA. Scale bar, 50 mm.