Melanoma Cell Adhesion and Migration Is Modulated by the Uronyl 2-O Sulfotransferase

Although the vast majority of melanomas are characterized by a high metastatic potential, if detected early, melanoma can have a good prognostic outcome. However, once metastasised, the prognosis is bleak. We showed previously that uronyl-2-O sulfotransferase (Ust) and 2-O sulfation of chondroitin/dermatan sulfate (CS/DS) are involved in cell migration. To demonstrate an impact of 2-O sulfation in metastasis we knocked-down Ust in mouse melanoma cells. This significantly reduced the amount of Ust protein and enzyme activity. Furthermore, in vitro cell motility and adhesion were significantly reduced correlating with the decrease of cellular Ust protein. Single cell migration of B16VshUst(16) cells showed a decreased cell movement phenotype. The adhesion of B16V cells to fibronectin depended on α5β1 but not αvβ3 integrin. Inhibition of glycosaminoglycan sulfation or blocking fibroblast growth factor receptor (FgfR) reduced α5 integrin in B16V cell lines. Interestingly, FgfR1 expression and activation was reduced in Ust knock-down cells. In vivo, pulmonary metastasis of B16VshUst cells was prevented due to a reduction of α5 integrin. As a proof of concept UST knock-down in human melanoma cells also showed a reduction in ITGa5 and adhesion. This is the first study showing that Ust, and consequently 2-O sulfation of the low affinity receptor for FgfR CS/DS, reduces Itga5 and leads to an impaired adhesion and migration of melanoma cells.


Introduction
A critical event in tumorigenesis of melanoma is the conversion from a primary tumor into an aggressive, metastasizing tumor. Tumor metastasis is a complex process involving its stroma, cell migration and invasion. Cell surface glycans, especially proteoglycans are involved in different stages of metastasis [1,2]. Proteoglycans are proteins covalently modified by a linear glycosaminoglycan (GAG) chain composed of repeating disaccharide units of an amino sugar and uronic acid [3]. Physiologically, GAGs are involved in multiple cellular functions, such as a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Cell culture
Murine melanoma (B16V) cells [29] were grown to confluence in bicarbonate buffered RPMI 1640 (Sigma) supplemented with 10% (v/v) bovine serum (FBS) at 37˚C in a humidified atmosphere of 5% CO 2 . Of note, B16V cells display a black color due to their melanin. All experiments were performed at passages where cells contained melanin. Human HT168-M1, HT199 [30] and MV3 [31] melanoma cells were grown in RPMI 1640 with 10% (v/v) FBS and cultured as described before.

Knock-down of Ust in melanoma cells
B16V cells were stably transfected with shRNA-Ust(m) plasmid as a pool of 3 target-specific lentiviral vector plasmids, each encoding 19-25 nt (plus hairpin) shRNAs designed to knockdown Ust gene expression (Santa Cruz), following the manufacturer's protocol. Control cells were mock transfected with shRNA plasmid-A. Cells were selected with 10 μg/ml puromycin (Santa Cruz) for 2 weeks and further subcloned by single cell limiting-dilution. For human MV3 melanoma cells, UST siRNA and the respective scrambled siRNA were used according to the manufacturer (Santa Cruz) and the cells were analyzed 48 h after transfection. 0.2 μm filter, 20-40 μg of protein lysates were analyzed for Ust and α5 integrin. They were visualized with enhanced chemiluminescence (Perkin-Elmer Life Sciences, USA) and monitored with Fusion-SL 4.2 MP (PeqLab, Germany). Intensities were quantified as described previously [23,33]. Of note, immune blots of the lysates before and after filtration led to the same results. The influence of the cell surface sulfation was evaluated after 6h of cell treatment with 30 mM NaClO 3 [5]. For blocking FgfR, cells were incubated for 6h with PD173074 (20 mM) inhibitor as determined based on titration curves.

Sulfotransferase activity of B16 cell lines
Sulfotransferase reaction was carried out according to the manufacturer's instructions in a 96-well plate using the universal sulfotransferase assay (R&D). Briefly, protein lysates  μg) of B16V cell lines were incubated with 10 mM chondroitin 6-sulfate as substrate, PAPS (R&D), and a coupling phosphatase as control. The color was developed with a Malachite reagent for 20 min at room temperature and monitored at 620 nm with an ELISA reader. A phosphate standard curve was used to determine the activity (OD/pmol). The specific activity was determined with the following equation: Specific activity (pmol/min)/μg) = S(OD/μg) x CF(OD/pmol) / Time(min), where S is the slope of the line with the OD values of the sulfotransferase assay and CF the phosphate conversion factor (taken from the phosphate standard) [5].

Characterization of cell surface chondroitin/dermatan sulfate and heparan sulfates
GAGs were extracted from~2x10 7 cells and highly-sulfated cell surface CS/DS were released by β-elimination and purified as described previously [23]. The HexUA content was determined using an m-hydroxydiphenyl reaction. Uronic acid was hydrolyzed in 80% sulfuric acid containing tetraborate at 80˚C, incubated with m-hydroxydiphenyl (Sigma Aldrich) at room temperature and measured at 540nm using heparin as standard [5].
10 μg CS/DS were digested with 10 mU of chondroitin ABC for 2h. The unsaturated disaccharides were labeled with 5 μl of 0.1 M 2-Aminoacridon (AMAC) in 15% CH 3 COOH/DMSO solution. After 10 min incubation at RT, 1 M NaBH 3 CN was added and the mixture was incubated 16 h at 37˚C followed by fluorophore assisted carbohydrate electrophoresis (FACE). AMAC-labeled disaccharides were separated on 30% Borate-polyacrylamid gel [35]. HS were analyzed as described before. In order to analyze HS composition, cell pellets of B16V cell lines were prepared as described previously. After enzymatic removal of CS/DS, the heparin lyase I-, II-and III-digested GAGs were fractioned by RPIP-HPLC. The peaks were identified by co-elution with standard HS disaccharides [5].
Proliferation of melanoma cells 3×10 4 B16V cells/cm 2 cells were seeded and cultured for 24h and starved for 16h prior to the experiment. Experiments were performed in serum-free RPMI and proliferation was determined by BrdU incorporation for 16h (Cell Proliferation ELISA, Roche).

Cell adhesion assay
Static adhesion assays were performed with 1×10 6 cells of the different B16V cell lines or MV3 cells in the presence of the fluorescent marker 2 0 7 0 -bis-(2 carboxyethyl)-5 carboxyfluorescein acetoxy-methyl ester (Molecular Probes, USA) dissolved in DMSO as described previously [32,36]. Labeled cells were seeded in non-or fibronectin-coated (10 μg/ml) 96-well plates and incubated for 30 to 360 min at 37˚C. Cell adhesion to fibronectin was quantified after 1h with an ELISA reader (Epoch, Bioteck) as previously described [32]. For further adhesion experiments cells were preincubated with i) the different LEAF™ integrin blocking antibodies (5 μg/ml) for 3h at 37˚C [33], ii) 6h pre-treatment with 30(mM NaClO 3 or iii) enzymatic digestion of cell surface HS and CS/DS with 4 mU heparitinase II/III and/or chondroitin ABC lyase for 1h at 37˚C.
Wound scratch assay and migration on 3D collagen-rich matrices 1×10 5 cells/cm 2 of B16V cell lines were seeded in 12-well plates in RPMI medium and starved overnight. An artificial wound was generated and cells were incubated with serum-free RPMI medium (control) or RPMI supplemented with 10 ng/ml Fgf2 for 20h at 37˚C. Images were captured at time points 0 and 20h, using a Zeiss Axiovert 100 microscope with AxioCam ICc1 camera. Cell migration was evaluated as described [5]. For each well 2-4 pictures were acquired (n = 3 independent experiments).
Primary C57BL/6 skin fibroblasts were cultured for 10 days in 35 mm petri dishes with 1 mM L-ascorbate-2-phosphate (Sigma) to obtain a 3D ECM [37]. Confluent B16V cell lines were detached from the culture dish with trypsin/EDTA, and B16V cell suspension in serumfree RPMI1640 was added to the 10 day old and 24h serum-starved C57BL/6 fibroblast cultures. Migration of cells was monitored, evaluated and calculated as described before [38].

Phospho-FGFR1 cell-based phosphorylation ELISA
Mouse/Human/Rat Phospho-FGFR1/FGF Receptor 1 (Y654) Cell-Based Phosphorylation ELISA Kit was used to determine the activation state of FgfR1 according to manufacturer instructions (LifeSpan Biosciences). Briefly, 20.000 cells/well were seeded in a 96-well plate and incubated overnight. Cells were starved overnight followed by treatment with PD173074 (20 nM) for 6 h to inhibit FgfR1. Cells were fixed with 4% PFA for 20 min, washed and blocked for 1 h prior to the incubation with the first antibodies i) anti-FGFR1-Phospho-Y654, ii) anti-FGFR1 or iii) anti-GAPDH overnight at 4˚C. HRP-conjugated secondary antibody was incubated for 30 min and developed with a ready to use substrate. The enzyme activity was measured at OD450 nm (Epoch, Bioteck). GAPDH served as an internal positive control to normalize the values. Following the colorimetric measurement the crystal violet whole-cell staining method was used to determine cell density. After staining, the results were analyzed by normalizing the absorbance values to cell amounts. pY654-FgfR1 was normalized to FgfR1 and GAPDH. The same protocol was applied for α5 integrin and normalized to GAPDH and cell number.
FACS analysis of cell surface α5 integrin 1 x 10 6 of B16V cell lines were seeded in a 6-well plate for 24 h. Cells were washed with cold PBS, scraped from the plates and aliquoted to 1 x 10 6 cells in 10 μl 2% FBS/PBS. To detect α5 integrin cells were incubated with Alexa Fluor 1 647 anti-mouse CD49e antibody (0.5 μg/100μl) for 30 min at 4˚C. After 3 x washing with cold PBS, cells were resuspended in 1 ml 2% FBS/PBS and analyzed with FACSAria IIu. Non-stained and isotope controls were analyzed simultaneously.
Animal experiments and B16 syngenic tumors 10 weeks old female C57BL/6 mice (Charles River, Germany) were grouped into 5 and kept for one week prior to the experiments and cared according to the standards of the German Council on Animal Care and Institutional Animal Care and Use Committee. Animals were housed in the animal facility of the Medical Faculty, University of Münster, Germany. Standard rodent chow and water were available ad libitum throughout the study and shredded paper was available for nest building. Mice were housed using a 14:10 light:dark cycle starting at 06:00 a.m. This study was carried out in strict accordance to the German Council on Animal Care under a specifically approved protocol by the ethics committee LANUV, NRW, Germany (protocol #84-02.04.2013.A007). All surgery was performed under isoflurane anesthesia, and all efforts were made to minimize suffering. 10 6 cells parental control and B16V shUst (16) cells in 70 μl PBS were injected s.c. into the right flanks of the mice. Mice and primary tumors were monitored every other day. Tumors were categorized as + < 0.5 cm 3 , ++ = 0.5-1 cm 3 , +++ > 1 cm 3 . Animals found with clinical signs, like weight loss or respiratory difficulty, were subjected to euthanasia. Euthanasia was carried out with an overdose of inhalant anesthetic followed by cervical dislocation. Primary tumors were removed after 15-21 days because of the size of the tumor (tumor size: B16V from 0.09 to 1.78 mg and B16V shUst (16) from 0.1 to 3.9 mg) and weighed. Metastasis was monitored over a 6-7 week period followed by autopsies of the sacrificed animals [38]. Animals found with clinical signs were subjected to euthanasia. Pulmonary metastasis was evaluated macroscopically.

Statistical analysis
Statistical evaluation was performed with GraphPad Prism4 using Student's t-test. P<0.05 was considered as significant.

Silencing Ust in B16V cells
To demonstrate that human cancer cells express UST we analyzed the human melanoma cell lines HT168-M1, HT199 and MV3 [29, 31] by qRT-PCR. All 3 cell lines express human UST (S1 Fig) with ΔCT ranging from~1.67 to~3.69. We previously published the Ust expression of CHO-K1 cells which showed a ΔCT value of~3.4 [5].

Functional characterization of the B16V shUst cells in vitro
Previously, the importance of cell surface DS [20] and 2-O sulfation [5,23] for cell migration was reported. In the present study, scratch assays showed that B16V melanoma cells close the gap within 20h (Fig 2A). The reduction of Ust led to a significantly slower migration of B16V shUst (6) and B16V shUst(16) cells on plastic. Fgf2 is a critical regulator of melanoma progression and it is expressed by melanoma cells [40]. However, Fgf2 addition had no impact on cell migration (Fig 2A and 2B). To exclude an overlap of migration and proliferation, BrdU incorporation was assessed. Within 24h under serum-free conditions B16V and B16V shUst(6) cells showed similar proliferation rates, only B16V shUst (16) cells displayed a significant increase in proliferation (Fig 2C). Therefore, we can conclude, that proliferation does not interfere with migration. Since molecular analysis of clone 6 showed no significant but detectable changes, we mostly show clone 16 in the following.
To get closer to an in vivo situation we used skin fibroblasts which were cultured for 10 days to deposit their own 3D ECM. Migration of B16V cell lines was monitored by timelapse microscopy (movies S3 and S4 Figs). Tracking single cells on the ECM showed that B16V shUst (16) cells migrated significantly more slowly (0.30±0.12 μm/min) than controls (0.35±0.09 μm/min) (Fig 2D and 2E). Furthermore, controls covered significantly longer distances than B16V shUst (16) cells (Fig 2F).
These results show that Ust and consequently CS/DS 2-O sulfation affect melanoma cell migration. To clarify the altered migration of the B16V shUst cell lines we analyzed cell adhesion.

Lack of Ust in B16V affects adhesion to fibronectin
Cell surface GAGs influence adhesion and migration of cancer cells [41]. Assessment of B16V shUst cell adhesion to plastic showed a significant reduction when compared to control (S5A Fig). Integrins, such as α5β1, are involved in adhesion to fibronectin [25]. After 1h and 6h, B16V cell adhesion to fibronectin was significantly higher compared to the B16V shUst cell lines (Fig 3A). To demonstrate the influence of sulfation, cells were pre-treated with 30 mM NaClO 3 . Chlorate treatment leads to an inhibition of the 3'-phosphadenosine-5-phosphosulfate synthesis and reduction of the sulfate content of cell surface GAG [42]. Of note, after chlorate treatment B16V shUst (16) cell adhesion monitored at the time point 1h was significantly lower To narrow down the type of GAGs involved in adhesion, cell surface CS/DS was digested with chondroitin ABC lyase and HS with heparitinase [5]. Enzymatic digestion of CS/DS or HS led to a significantly reduced adhesion of B16V cells compared to the non-treated cells after 1h. Upon digestion of cell surface CS/DS (ABCase), but not HS, we observed a significant difference between control and B16V shUst (16) cells. Digestion of both CS/DS (ABCase) and HS (Hep-Mix) reduced adhesion of both cell types (Fig 3B). To determine which integrin dimer is responsible for the impaired adhesion of the B16V shUst(16) cells, we used specific blocking antibodies for αvβ3 and α5β1 integrin and the respective isotype controls. Blocking of αv or β3 integrin had no impact on the adhesion of either B16V cells or B16V shUst (16) cells to fibronectin when cells were compared to their isotype treated controls or the non-treated cells (Fig 3C). In contrast, blocking of α5 or β1 integrin significantly reduced adhesion of B16V to fibronectin compared to isotype treated controls or the non-treated cells (Fig 3D). Interestingly, B16V shUst (16) cells showed still basal adhesion compared to isotopic treated control or the non-treated cells (Fig 3D). The results indicate that α5β1 integrin, but not αvβ3 integrin, mediates adhesion of B16V cells to fibronectin.
Control and B16V shUst (16) cells were seeded on fibronectin and analyzed by confocal microscopy. F-actin and the distribution of α5 integrin were evaluated after 3h (Fig 4B). In contrast to control, B16V shUst (16) cells showed less α5 integrin and altered F-actin distribution which might explain the impaired adhesion of the Ust knock-down cells (Fig 3B). The immunofluorescence results were supported by immunoblots for α5 integrin. Controls showed similar amounts of α5 integrin whereas for B16V shUst(16) cells α5 integrin was significantly reduced (Fig 4C). Again, B16V shUst (6) showed also a reduced amount of α5 integrin similar to the qRT-PCR data. Next, we determined the amount of α5 integrin on the cell surface by FACS analysis and observed no differences for all tested cell lines (S6A, S6B and S6C Fig). To link GAG sulfation to integrins the cells were treated with chlorate. Chlorate treatment depleted α5 integrin in both cell lines (Fig 4D), indicating a link between sulfation and α5 integrin. Itga5 is known to be regulated by Fgf2 [43,44]. Therefore, we used the specific inhibitor PD173074 to target FgfR1. Blocking FgfR1 for 6h led to a reduction of α5 integrin of control, mock, B16V shUst (6) and B16V shUst(16) cells (Fig 4E and 4F). Next, we analyzed FgfRs transcripts and only FgfR1 and 3 showed higher expression levels. Interestingly, the expression levels of FgfR1 in the B16V cell lines were significantly reduced correlating with the decrease of Ust. FgfR3 was significantly increased only in the B16V shUst (16) cells (Fig 4G).
Next we determined the amount of FgfR1 and its activation in the B16 cell lines. An ELISA type assay showed a significant reduction of FgfR1 protein in the knock-down cells correlating with the amount of Ust (Fig 4H). Moreover, the activation of FgfR1-Y654 showed that both knock-down cells lines displayed significantly less phosphorylation compared to control and mock transfected B16V cells (Fig 4I). Therefore, we can conclude that, depending on Ust levels, ITGa5 and FgfR1 expression is affected as well as the activation of FgfR1. In addition, 2-O sulfated CS/DS proteoglycans influence the function of α5β1 integrin and consequently cell adhesion to fibronectin.

Melanoma cell lung metastasis is influenced by Ust expression
The metastatic potential is greatly affected by the expression levels of integrins. Therefore, we investigated the behavior of B16V shUst (16) cells in vivo. When control and B16V shUst (16) cells were inoculated into female syngenic C57BL/6 mice no difference in tumor growth was observed ( Fig 5A; Table 2; n = 12 for control and 11 for B16V shUst (16) ). 15-21 days after the resection of the primary tumor, mice were sacrificed and further analyzed. There was no difference in weight of the mice (control: 21.36±2.1 g, B16V shUst(16) 22.54±1.0 g). Macroscopic evaluation of lungs showed metastases in 6 out of 9 mice inoculated with control cells, whereas the 11 mice with B16V shUst (16) tumors showed no macroscopic lung metastasis (Table 2 and   Fig 5B). Notably, detection of lung metastasis was independent of the tumor size. Western blot analysis revealed a reduction of Ust and α5 integrin in the B16V shUst (16) tumors (Fig 5C). The amount of β1 integrin in the tumor lysates was not altered (S7 Fig). These in vivo results support the in vitro observation of impaired B16V shUst(16) adhesion due to a reduced amount of α5β1 integrin induced by the lack of Ust.

UST in human melanoma MV3 cells
Our concept was supported by the results obtained with siRNA-mediated UST knock-down in human MV melanoma cells. UST siRNA transfection of MV cells caused an 80% reduction of UST mRNA (Fig 6A). In these cells ITGa5 was also reduced by 70% (Fig 6B), and consequently adhesion to fibronectin (Fig 6C). These results show that CS/DS 2-O sulfation mediates α5 integrin expression via FgfR at least after Ust knock-down. Our results show that Ust and 2-O sulfation levels of CS/DS affect the synthesis of Itga5 and FgfR1 and, in addition, the function of α5β1 integrin which leads to impaired melanoma cell adhesion (Fig 7).

Discussion
Migration of tumor cells is an important step during metastasis. The motility of melanoma cells contributes to their highly invasive and metastatic potential. Melanoma cells display a shift from HS and DS to CS [13]. However, the function of the fine structure of the 2-O sulfation of the CS/DS chains and the respective enzymes involved, especially Ust, are not known. So far the function of Ust has been studied during brain development or under physiological conditions [5,19]. CHO-K1 cells express Ust which subsequently leads to CS/DS 2-O sulfation, later involved in Fgf-2-induced migration [5]. Recently, a patient with a microdeletion on chromosome 6q25.1 was described with among other symptoms an Ehlers-Danlos syndrome in skin [45]. The microdeletion included the lack of human UST gene indicating that the minor sulfation of CS/DS affects also the organization of the extracellular matrix, similar to the DS [46]. Understanding the impact of Ust and 2-O sulfation might identify potential therapeutic targets in melanoma metastasis.
We used an experimental metastatic model of B16 cells which has the advantage of forming a primary orthologue tumor followed by metastasis from the skin to the lung [39]. Our in vivo  experiments show a marked inhibition of pulmonary metastasis after the knock-down of Ust in B16V cells. The importance of the CS/DS fine structure has been shown in the LLC pulmonary metastasis model, were the knock-down of Chst15 reduced ΔDi4,6S units and consequently the transmigration of cells from the blood into the lung tissue [17]. A reduced proliferation was observed for the LCC Chst15 knock-down cells [17] or when DS was removed from melanoma cells [14]. In contrast, in vitro Ust knock-down increased proliferation in CHO-K1 [5] and B16V cells. In vivo we observed a similar size of the tumors for the B16V model which could be explained by similar amounts of ΔDi4,6S units. Of note, B16 cells contain 1.5 times more DS than LCC cells [18] indicating that adhesion and migration of melanoma cells could be influenced by 2-O sulfated CS/DS. Under physiological conditions, Ust has been shown to be important for in vivo cell migration and possibly development [19,45]. The structures of GAGs influence migration, too. Aortic smooth muscle cells with reduced DS show a decrease in directional migration, although the velocity and the total distance are increased [20]. Of note, the reduction of DS in mice reduced also the 2-O sulfation of CS/DS [47]. Knock-down of DS-epimerase 1 and consequently DS in cancer cells also reduced migration [7]. Vice versa, CHO-K1 cells migrate faster when they overexpress Ust. B16V melanoma cells are highly metastatic [39] and display an increased velocity compared to CHO-K1 cells [5]. Ust knock-down in B16V cells consequently reduced migration, like Chst15 knock-down in LCC cells [17]. Our observation that the amount of ΔDi4S-units was not affected and migration was opposed to smooth muscle cells with reduced DS might indicate that the amount of DS on the cell surface of B16V cell lines is not affected by the Ust knock-down.
Ust knock-down in CHO-K1 cells and fibroblasts also results in an impaired migration as previously reported for neuronal outgrowth [5,19,23]. The residual migration of B16V shUst (6) cells, in contrast to B16V shUst (16) , can be explained by the presence of ΔDi4,6S units [17] and the impact of the level of 2-O sulfation. In vivo, for HS not only the amount of sulfation but also the HS structure affects hedgehog signaling during development [48]. Migration on a complex ECM generated by fibroblasts requires integrins that recognize several matrix ligands including fibronectin (e.g. α5β1, αvβ3, α4β1) or collagens (α1β1, α2β1). Previously, we showed that the lack of decorin and the reduced amount of 2-O sulfated CS/DS in fibroblasts lead to an increase in β1 integrin [23,33]. Melanoma cells express α5β1 and αvβ3 integrin which are involved in adhesion and migration [26] and are tightly regulated [24]. Overexpression of miR-148b in melanoma cells significantly inhibited metastasis by reducing ITGa5 [49]. In contrast to decorin-deficient fibroblasts, B16V Ust knock-down cells displayed a reduction in α5 integrin. B16V cell adhesion to fibronectin was only reduced by blocking α5β1 but not αvβ3. The cell surface amount of α5 integrin was not altered in B16V Ust knock-down cells indicating that 2-O CS/DS sulfation functions as a structural component involved in adhesion. Various studies showed that cell surface CS and DS and their structures have an impact on cancer cell adhesion [7,14,15,17,41]. One might speculate that the level of CS/DS 2-O sulfation on the cell surface modulates α5β1 integrin conformation and fibronectin binding. This speculation can be supported by a recent publication about different distinct global conformations of α5β1 integrin which determine adhesive and non-adhesive function to fibronectin. Under conditions in which the bent-closed conformation predominates, α5β1 integrin impairs adhesion to fibronectin in a K562 cell line [50].
As modulating signaling events, Ust knock-down and consequently, CS/DS 2-O sulfation are associated with the expression of Itga5 and FgfR1. The reduced expression of Itga5 and FgfR1 can be explained by the function of CS/DS as low affinity receptors for different growth factors [1,2,7] and therefore affecting signaling. This hypothesis is supported by a reduced Itga5 expression either after inhibition of cell surface sulfation or by blocking FgfR1. Moreover, also FgfR1 and its activation are reduced in B16 Ust knock-down cells. The link of α5 integrin, Fgf2 and FgfR has been demonstrated for angiogenesis [43] and in 3T3 fibroblasts [44]. In addition, the aggressiveness of melanoma is due to Fgf2 induced α5 integrin expression [28]. Under physiological conditions Fgf2 signaling requires GAGs [1] and we could show that 2-O sulfated CS/DS are involved in migration [5]. A possible downstream mechanism could be the transcription factor Twist-1 which has been recently shown to induce Itga5 expression and leads to epithelial-mesenchymal transition [51]. The link between reduced Twist-1 expression, lack of DS and adhesion to fibronectin has been shown for Xenopus neural crest cells [52]. Of note, adhesion of B16V shUst (16) cells was significantly reduced by either inhibiting CS/DS sulfation or by digesting cell surface GAGs.
To support a possible role of Ust in melanoma, we tested three metastasizing human melanoma cell lines, MV3 [31], HT199 and HT168M [30] that all express UST. The biological relevance of the data obtained with mouse melanoma cells is supported by the human melanoma cell line MV3 which expresses UST. siRNA-mediated UST knock-down in MV3 cells also showed a reduction in ITGa5 and adhesion.
Overall our data propose Ust and consequently 2-O sulfated CS/DS as a regulator of adhesion via the amount and activation of FgfR1 and the expression of Itga5 in melanoma cells (Fig 7). In addition, the amount of 2-O sulfated CS/DS influences α5β1 integrin function in melanoma cells indicating that Ust could be a potential marker for melanoma metastasis and a target for a therapeutic approach.  (6) and 70% less for B16VshUst (6). The FACE analysis supported the result obtained by the enzyme activity test (see Fig 1C). Immuno blots of three control and three B16V shUst (16) primary tumors lysates for β1 integrin and β-actin as loading control. The β1 integrin blot was used after stripping. Therefore, the loading control β-actin is the same as in Fig 5C. (TIF)