Insights into the role of sulfated glycans in cancer cell adhesion and migration through use of branched peptide probe

The tetra-branched peptide NT4 selectively binds to different human cancer cells and tissues. NT4 specifically binds to sulfated glycosaminoglycans on cancer cell membranes. Since sulfated glycosaminoglycans are involved in cancer cell interaction with the extracellular matrix, we evaluated the effect of NT4 on cancer cell adhesion and migration. We demonstrated here that the branched peptide NT4 binds sulfated glycosaminoglycans with high affinity and with preferential binding to heparan sulfate. NT4 inhibits cancer cell adhesion and migration on different proteins, without modifying cancer cell morphology or their ability to produce protrusions, but dramatically affecting the directionality and polarity of cell movement. Results obtained by taking advantage of the selective targeting of glycosaminoglycans chains by NT4, provide insights into the role of heparan sulfate proteoglycans in cancer cell adhesion and migration and suggest a determinant role of sulfated glycosaminoglycans in the control of cancer cell directional migration.


Inhibition of NT4 binding to PANC-1 cancer cells by heparin, HS and CS reflects GAG-NT4 binding affinity.
Since NT4 peptide bound heparin, HS and CS with different affinity, we measured the inhibition activity of the three GAGs on NT4 binding to membrane receptors on the human pancreas adenocarcinoma cell line PANC-1 by flow cytometry (Fig. 2). Using identical concentrations of heparin, HS and CS, complete inhibition of NT4 binding was obtained with heparin, about 80% inhibition was obtained with HS and no inhibition was detected with the same concentration of CS. This result confirms the specificity of NT4 binding to membrane GAGs on cancer cells, with preferential binding to HS.
Binding of NT4 to xylosyltransferase-I-deficient PgsA-745 cell line. The PgsA-745 cell line is derived from CHO-K1 cells treated with mutagens and screened for GAG synthesis defective mutants. PgsA-745 cells have a defect in xylosyltransferase, the enzyme responsible for coupling the first sugar in GAG synthesis, and do not produce GAGs.
In order to further verify specific binding to GAGs, we compared NT4 binding to these GAG-defective cells and to native CHO-K1 cells by confocal microscopy. As expected, NT4 binding to PgsA-745 cells was much lower (Fig. 3), confirming that GAGs are specific NT4 targets on the cell membrane. Nonetheless, flow cytometry analysis demonstrated that binding of NT4 to PgsA-745 cells, though much lower than to CHO-K1, was still inhibited by heparin. Since we had already demonstrated that NT4 binds HSPG as well as endocytic receptors like sortilin, Scientific RepoRts | 6:27174 | DOI: 10.1038/srep27174 LRP1 and LRP6 7 , the residual binding to PgsA-745 cells might be due to membrane receptors belonging to these families. We had already demonstrated that heparin and endocytic receptors compete for NT4 binding.
Inhibition of cancer cell adhesion by NT4. The results of Biacore experiments, confocal microscopy and flow cytometry described above indicate that sulfated GAGs are specifically recognized by NT4 on the cancer cell membrane. Since sulfated GAGs have an important role in cancer cell adhesion and migration, we tested the effect of NT4 on cancer cell adhesion and motility on different supports. Adhesion of PANC-1 human pancreas adenocarcinoma cells was tested on cell culture plates coated with human collagen IV, human fibronectin and on uncoated plastic wells.  Fibronectin exists in different isoforms, obtained by alternative splicing. Soluble plasma fibronectin, which is produced by hepatocytes and is abundant in plasma, lacks EDA and EDB domains 36 , whereas cellular fibronectin produced by fibroblasts, epithelial cells and other cell types is a major component of the ECM and contains the EDA and/or EDB segments, whose selective expression in cancer tissues has long been known [37][38][39] . In view of the different roles of plasma and cellular fibronectin, we tested the effect of NT4 peptide on cancer cell adhesion on both isoforms of human fibronectin.
NT4 inhibited adhesion of PANC-1 cancer cells to plastic, collagen IV and cellular oncofetal fibronectin with overlapping efficiency (EC50 of 2.8e-6 M), whereas adhesion of cancer cells to plasma fibronectin was scarcely affected and only at the highest concentration of NT4 (Fig. 4). This result suggests major involvement of HSPGs in PANC-1 cancer cell adhesion to cellular fibronectin of the ECM, compared to plasma fibronectin, and a possible role of the EDA and/or EDB domain in fibronectin binding to sulfated GAGs.
Inhibition of cancer cell motility by NT4. The effect of NT4 peptide on cancer cell migration was tested in vitro in wound healing experiments. PANC-1 cells were grown to confluence on 24-well cell culture plates coated with collagen IV, cellular fibronectin or plasma fibronectin and on uncoated wells. A cell-free zone was created by inserting a silicone spacer, which was removed once cells had reached confluence. Cells were then incubated with different concentrations of NT4 and the void area was monitored. In order to distinguish cell division from cell motility, which might both interfere with filling of the void area, the effect of NT4 on cell division was checked by MTT assay. NT4 clearly inhibited filling of the cell-free area on uncoated plastic wells as well as on plates coated with collagen IV, cellular fibronectin and plasma fibronectin and the effect was dose-dependent (Fig. 5). No effect of NT4 peptides on cancer cell division was detected in identical experimental conditions (not shown), indicating that the effect on cancer cell re-colonization of the void area was entirely due to inhibition of cell migration.
Cancer cell migration was monitored by time lapse microscopy in order to visualize modifications of cancer cell morphology or motility produced by the NT4 peptide. Time lapse imaging of cancer cell movements indicated that NT4 does not affect cancer cell morphology or ability to produce protrusions (Fig. 6A, supplementary material Figs S1 and S2) but rather the directionality and polarity of cell movement. In wound healing experiments on plates coated with collagen IV, fibronectins or uncoated wells, cancer cells incubated without NT4 clearly moved towards the cell-free zone, as indicated by single cell migration tracking (Fig. 7). On the other hand, when incubated with NT4, cancer cells seemed to slip on the plate and lose their direction of migration. They produced protrusions like in control wells, however these did not produce effective forward migration but rather random circular movement with little progression toward the cell-free zone (Figs 6A and 7, supplementary material Figs S1 and S2).
NT4 ability to inhibit cell migration of PANC-1 cell was also confirmed in wound healing experiments where cells were covered with Matrigel. As expected, in these experiments cell migration and speed were reduced compared to 2D wound healing (supplementary material Figs S3-S5).
The effect of NT4 on cancer cell morphology was also checked by scanning electron microscopy. PANC-1 cells were incubated with 10 μ M NT4 for 3 hours and then analyzed (Fig. 6B). No modification of general cell shape or cell protrusions was observed.

Inhibition of cancer cell trans-well migration by NT4.
In order to evaluate whether NT4 peptide inhibited cancer cell trans-well migration, a classical migration assay on Boyden chambers was performed with PANC-1 cells (Fig. 8). Cells were incubated for 24 h with different concentration of NT4 (20, 10 and 5 μ M) in uncoated or collagen IV-coated upper wells and migrated cells were then counted. Results in Fig. 8 show that 20 μ M NT4 caused about 85% and 68% reduction of PANC-1 cells trans-well migration in uncoated and in collagen IV-coated Boyden chambers, respectively.

Discussion
In previous papers we reported that NT4 tetra-branched peptides are promising cancer-selective theranostic agents by virtue of their ability to selectively deliver different functional units to cancer cells for imaging or therapy. NT4 contains four copies of the human neurotensin sequence, which many years ago had been reported as a potential selective cancer-targeting agent because neurotensin receptors seemed to be over-expressed in different human cancers. However, we demonstrated that the high cancer selectivity of NT4, not shared by monomeric native neurotensin, is actually generated by its multimericity, which switches NT4 selectivity towards membrane receptors different from the canonical NTR1 and NTR2 G protein-coupled neurotensin receptors. NT4 was found to bind heparin, as well as LRP1, LRP6 and sortilin, by means of a multimeric positively-charged motif that might interact with negatively-charged motives in sulfated GAGs and LRP receptors, already known to share several different heparin-binding ligands 7 .
Considering the fundamental role of sulfated GAGs in many aspects of cancer cell biology, we investigated NT4 selectivity towards GAGs, as well as the effect produced by NT4 on cancer cell adhesion and migration. We found that NT4 binds heparin, HS and CS, which are all sulfated GAGs, but does not bind (non-sulfated) hyaluronic acid. Binding affinity of NT4 to sulfated GAGs, as measured by Biacore, was higher for heparin, and progressively lower for HS and CS. Inhibition of NT4 binding to human pancreas adenocarcinoma PANC-1 cells by the same sulfated GAGs was essentially in line with their binding affinity, with complete inhibition of NT4 binding obtained with heparin, less inhibition obtained with HS and no inhibition in the presence of the same concentration of CS. Since neither binding to NT4 or inhibition of NT4 binding to cancer cell membranes was detected with the non-sulfated GAG, hyaluronic acid, and heparin has a greater content of sulfated groups than HS and CS 12,13,40 , the results from affinity and inhibition activity indicate that NT4 specifically binds sulfated GAGs on the cancer cell surface, with preferential binding to HS. This result was further confirmed by the very low binding of NT4 peptides to the xylosyltransferase-I-deficient PgsA-745 cell line, which does not synthesize GAGs. Indeed, binding of NT4 to this GAG-defective cell line was much lower than to the native CHO-K1 cell line from which PgsA-745 is derived. Nonetheless, binding of NT4 to PgsA-745 cells, while low, was still completely inhibited by heparin. This may be explained by NT4 binding to LRP receptors on PgsA-745. Synthesis of LRP receptors is not affected by lack of xylosyltransferase I, which is responsible for transfer of the first xylose  residue of any GAG chain to the protein core of HSPG and in fact wild type CHO-K1 and PgsA-745 cells were reported to have comparable amount of membrane LRP receptors 41 . Actually, we previously demonstrated that NT4 binds to heparin and LRP receptors via the same multimeric positively-charged motif and that binding to endocytic receptors is completely inhibited by heparin 7 .
Sulfated GAGs and LRP receptors both have an important role in many aspects of cancer cell biology, particularly aspects related to contact between cancer cells and ECM enabling cell adhesion and migration. We then tested the effect of NT4 in adhesion and migration of PANC-1 human pancreas adenocarcinoma cells on different ECM proteins. Cell adhesion and migration were tested on non-coated culture plates and plates coated with collagen IV or fibronectin. We tested the two main forms of human fibronectin: soluble plasma fibronectin, produced by hepatocytes, abundant in plasma, and lacking the EDA and EDB domains, and oncofetal cellular fibronectin, which has EDA and/or EDB domains, is produced by fibroblasts, epithelial cells and other cell types  was almost unaffected by NT4. This confirms that sulfated GAGs are involved in cancer cell adhesion to different supports, including oncofetal fibronectin. On the other hand, the lack of inhibition of cancer cell adhesion to soluble fibronectin by NT4 suggests that EDA and/or EDB are essential for cell adhesion mediated by sulfated GAGs. This is in line with recent data indicating that cellular fibronectin and not plasma fibronectin is involved in epithelial mesenchymal transition 42 , cell adhesion and migration 43 in different cancers.
In the wound healing assay, NT4 inhibited cancer cell migration on all supports, including serum and cellular fibronectin, in contrast to what we observed for cancer cell adhesion, where inhibition by NT4 only affected cancer cell adhesion to cellular and not to plasma fibronectin. PANC-1 cancer cell migration was also inhibited in wound healing experiments where cells were embedded in Matrigel, although cell migration and speed were reduced in this condition compared to the 2D wound healing experiments. Comparison of cancer cell migration by time lapse analysis, with or without NT4, showed that inhibition of cell migration was not accompanied by any dramatic modification of cell morphology and NT4 did not affect cell ability to produce protrusions. Moreover, NT4 did not modify cell morphology even under static conditions, as verified by scanning electron microscopy of cells incubated with NT4. What appeared to be dramatically affected by NT4 was cell ability to move towards the void space by oriented migration. In the presence of NT4, cancer cells lost their direction of migration and their average speed was therefore significantly reduced. This was evident in wound healing experiments in uncoated wells, as well as on plates coated with collagen, with either cellular or plasma fibronectin, as well as in Matrigel. Moreover NT4 dramatically reduced cancer cell migration in trans-well experiments.
These results suggest that although sulfated GAGs appear to be crucial for cancer cell adhesion and migration, different mechanisms of cancer cell contact with ECM proteins, particularly with fibronectin, seem to be involved in cell migration in contrast with the static conditions of cancer cell adhesion experiments. Our results indicate that fibronectin EDA and/or EDB domains could be determinant for interaction with sulfated GAGs for cancer cell adhesion under static conditions, but not for cancer cell migration.
A fundamental role of HSPGs has already been reported in endothelial cell migration and mechanotransduction, where, in addition to regulating cell adhesion and migration speed, HSPGs seem to be important in sensing the direction of shear stress and transmitting the mechanical signal into the intracellular space. This signaling might be able to regulate the direction of cell migration 44 . Interestingly, in the same paper a different role of HSPG in cancer cell adhesion and migration was reported, which is in line with our results. Moreover, a monoclonal antibody targeting the HS chain of glypican 3 was recently reported to inhibit migration of hepatocellular carcinoma cells 45 .
Since NT4 peptides bind to sulfated GAGs and also to LRP receptors via the same multimeric ligand motif, we cannot distinguish which of the two receptor families is more involved in the effect produced by NT4 in cancer cell adhesion and migration. Nonetheless, results obtained with the GAG defective PgsA-745 cell line strongly suggest that the effect of NT4 on cancer cell migration is mostly related to its interference with the function of sulfated GAGs. Although it has long been recognized that HSPG are crucial for cell adhesion and migration, their role in respect of the better characterized function of integrins has not been fully characterized. HSPG were assigned a sort of supporting role in the organization of cell contacts with ECM and were generally considered as integrin co-receptors. In the last few years results were reported suggesting that beside the integrin-ECM contacts and related intracellular responses, HSPG are essential for the correct organization of focal adhesion, stress fibers and intracellular signaling leading to oriented cell migration [46][47][48] .
By using a selective ligand of HSPG sulfated GAG chains, we have here provided strong indications of a primary role of HSPG in cell adhesion and migration. Moreover, our results confirm the hypothesis of a synergic role of HSPG and LRP receptors, which besides sharing many morphogenic ligands and collaborating in modulating their cell signals, may also share binding sites on different ECM macromolecules and therefore collaborate in maintaining and modulating contacts and relationships between cancer cells and ECM, as well as in controlling the directionality of cancer cell migration. In conclusion, our results suggest that besides their possible use as selective cancer theranostics, NT4 peptides may interfere with cancer cell migration and adhesion, thus reducing tumor aggressiveness and metastatic potential. The results also provide insights into the role of HSPG and LRP receptors in cancer cell motility and adhesion to different ECM proteins.

Experimental Procedures
Peptide Synthesis. Peptides were synthesized on an automated multiple synthesizer (MultiSynTech, ethanolamine pH 8.5 was used to block any activated carboxyl groups. NT4 peptide conjugated with biotin, diluted in HBS-EP+ (Hepes 10 mM, NaCl 150 mM, 3.4 mM EDTA, 0.05% p20, pH7.4) at 30 μ g/ml, was injected for 2 min at a flow rate of 10 μ l/min. GAGs kinetics. GAGs were diluted at concentrations ranging in HBS-EP+ and then injected over immobilized NT4 peptides. The sensor chip surface was regenerated with a short pulse of 10 mM NaOH / 0.5 M NaCl 5 minutes after the end of the injections.
Kinetics were analyzed with "Biacore T100 evaluation 1.1.1" software using the Langmuir model 1:1 for fitting curves.
Cell lines. PANC-1 human pancreas adenocarcinoma, CHO-K1 Chinese hamster ovary cell and PgsA-745 Chinese hamster ovary cell mutant deficient in xylosyltransferase (UDP-D-xylose: serine-1,3-D-xylosyltransferase) were grown in DMEM Medium (for PANC-1 and CHO-K1) and F 12 K Medium supplemented with 10% fetal bovine serum, 200 μ g/ml glutamine, 100 μ g/ml streptomycin and 60 μ g/ml penicillin. Cell lines were purchased from ATCC (The Global Bioresource Center). Wound healing. Cell migration was measured using an in vitro wound healing assay. Briefly, PANC-1 cells (2 × 10 5 cells/well) were seeded on each side of a culture insert for live cell analysis (Ibidi, Munich, Germany). Inserts were placed in wells of a pre-coated 24-well plate (20 μ g/ml human collagen IV and 10 μ g/ml human cellular fibronectin and human plasma fibronectin (Sigma Aldrich) for 2 h at 37 °C) and incubated at 37 °C and 5% CO 2 to allow cells grow to confluence. Afterwards, inserts were removed with sterile tweezers to create a cell-free area of approximately 500 μ m and cells were treated with NT4 peptide at different concentrations (20, 10, 5 and 1 μ M) in complete medium. Cells were allowed to migrate in an appropriate incubator. CytoSMART Lux 10x System (Lonza) was used to take a picture at time point zero and every 10 min for a total of 22 h. At the beginning and end of the time intervals the wound area was visualized under an inverted microscope (Zeiss Axiovert 200 microscopy) at 5x magnification and photographed with a Nikon ACT-1 Version 2.63 camera. The percentage of void area with respect to time 0 was determined using Tscratch software after 16 h, when control wells had completely filled the gap.
The time lapse stacks of images were also analyzed using ImageJ and the two plug-ins: Manual Tracking, and Chemotaxis and Migration Tool. Individual cells were randomly selected and tracked throughout the 10 h time period.

Scanning electron microscopy. The morphology of cancer cells was investigated by Scanning Electron
Microscopy (SEM). PANC-1 cells were plated at a density of 3 × 10 4 per well in 24-well plates with cover glass slides and incubated with 10 μ M NT4 peptide in complete medium for 3 h at 37 °C. Cells were fixed in 2.5% glutaraldehyde solution in phosphate buffer 0.1 M pH 7.2 (PB) for 2 h at 4 °C, washed in PB, postfixed in 1% OsO 4 in PB for 30 min at 4 °C, dehydrated in ascending alcohol series, incubated for 10 min in tert-butanol, and finally freeze dried.
Afterwards, the coverslip was mounted on an aluminum stub, coated with 20 nm gold in a Balzers MED010 sputtering device, and observed in a Philips XL20 scanning electron microscope operating at an electron accelerating voltage of 10 kV. Trans-well cell migration. Standard trans-well inserts for 24-well plates (transparent PET membrane, 8.0 μ m pore size, Corning) were equilibrated with serum free tissue culture medium for 2 h at 37 °C. In a set of experiment the upper compartment was coated with 20 μ g/ml of human collagen IV. 35000 PANC-1 cells diluted in serum free medium containing NT4 (20, 10 or 5 μ M), were placed in uncoated or collagen IV-coated upper chambers and the lower chambers were filled with medium containing 10% serum. After 24 h incubation at 37 °C in a humidified atmosphere with 5% CO 2 , cells were fixed with 4% PFA in PBS and stained with Crystal Violet. Non-invading cells were then removed from the top of the membrane using a cotton swab dipped in PBS, according to the manufacturer's protocol. Ten field images for each Boyden chamber were taken with a 10x objective in a Zeiss Axiovert 200 microscope. Cells were counted using the ImageJ software, NIH.
Cancer cell migration in Matrigel. Cell migration was measured using an in vitro wound healing assay in the presence of Matrigel (Corning Matrigel Basement membrane Matrix). Briefly, PANC-1 cells (2 × 10 5 cells/well) were seeded on each side of a culture insert for live cell analysis (Ibidi, Munich, Germany). Inserts were placed in 24-well plate and incubated at 37 °C and 5% CO 2 to allow cells grow to confluence. Inserts were removed with sterile tweezers to create a cell-free area of approximately 500 μ m. Wells were covered with Matrigel for 30 min at 37 °C and then 10 μ M of NT4 peptide in complete medium was added to each well. Cells were allowed to migrate in an appropriate incubator. CytoSMART Lux 10x System (Lonza) was used to take a picture at time point zero and every 7.5 min for a total of 42 h. Statistical analysis. All our data are parametric data calculated with Kolmogorov-Smirnov (KS) test using Graph Pad Prism 5.0 software. P values were calculated using: two-way ANOVA for adhesion assay (Fig. 4), one-way ANOVA with Dunnett post-test for wound healing assay (Fig. 5E) and for invasion assay (Fig. 8) and one-tailed Student's t-test for directionality of cancer cell migration analyzed by time lapse microscopy ( Fig. 7) using Graph Pad Prism 5.03 software.
All p values were reported in figure legends.