Endocytosis and Trafficking of Heparan Sulfate Proteoglycans in Triple-Negative Breast Cancer Cells Unraveled with a Polycationic Peptide

The process of heparan sulfate proteoglycan (HSPG) internalization has been described as following different pathways. The tumor-specific branched NT4 peptide has been demonstrated to bind HSPGs on the plasma membrane and to be internalized in tumor cell lines. The polycationic peptide has been also shown to impair migration of different cancer cell lines in 2D and 3D models. Our hypothesis was that HSPG endocytosis could affect two important phenomena of cancer development: cell migration and nourishment. Using NT4 as an experimental tool mimicking heparin-binding ligands, we studied endocytosis and trafficking of HSPGs in a triple-negative human breast cancer cell line, MDA-MB-231. The peptide entered cells employing caveolin- or clathrin-dependent endocytosis and macropinocytosis, in line with what is already known about HSPGs. NT4 then localized in early and late endosomes in a time-dependent manner. The peptide had a negative effect on CDC42-activation triggered by EGF. The effect can be explained if we consider NT4 a competitive inhibitor of EGF on HS that impairs the co-receptor activity of the proteoglycan, reducing EGFR activation. Reduction of the invasive migratory phenotype of MDA-MB-231 induced by NT4 can be ascribed to this effect. RhoA activation was damped by EGF in MDA-MB-231. Indeed, EGF reduced RhoA-GTP and NT4 did not interfere with this receptor-mediated signaling. On the other hand, the peptide alone determined a small but solid reduction in active RhoA in breast cancer cells. This result supports the observation of few other studies, showing direct activation of the GTPase through HSPG, not mediated by EGF/EGFR.


Introduction
Heparan sulfate proteoglycans (HSPGs) are a group of glycosylated proteins, classified on the basis of their cell localization or according to the nature of the glycosaminoglycan group attached to the core protein. Glycosaminoglycans (GAGs) are attached to serine residues and are linear acidic polymers designated as heparan sulfate (HS), chondroitin sulfate (CS) and dermatan sulfate (DS) proteoglycans [1].
Membrane HSPGs are syndecans and glypicans and act as receptors or co-receptors for a variety of ligands, particularly growth factors, and are therefore involved in various cell signaling pathways. HSPGs have long HS chains composed of 40-300 sugar residues, many of which are acidic sugars modified by sulfate groups and are, therefore, highly negatively charged [1].

NT4 Peptide Binds to HS and Is Internalized with HS
NT4 peptide previously proved to bind heparan sulfate chains in surface plasmon resonance experiments [13] and also to bind to MDA-MB231 cells in flow cytometry experiments [19]. Detailed mapping of NT4 binding on heparan sulfate chains showed a preference for hypersulfated patches of glypicans and syndecans [19]. The peptide bound to the plasma membrane ( Figure 1A) [13], most was internalized in cells within 15-30 min ( Figure 1B,C) and it was completely endocytosed within 1 h ( Figure 1D). The antibody anti-HS clone 10E4 [20], the epitope of which includes an N-sulfated glucosamine residue, was used in a colocalization test with the peptide. The peptide and the antibody colocalized in clusters on the plasma membrane ( Figure 1E) and inside the cell, proving that the bound pair is internalized unresolved ( Figure 1E). The overlap rate measured with the specific microscope software [21] was 71% (±4.41%; n = 9), with a mean Pearson correlation of 0.6561 (±0.03) on three experiments and three randomly taken fields.

NT4 Peptide Internalization Follows More than One Pathway
An anti-caveolin antibody was used in a colocalization experiment with NT4 ( Figure 2A). MDA-MB-231 cells were incubated with the peptide at 37 °C for different times: 5, 15 and 30 min and 1 h. The cells were then stained with the caveolin antibody. The colocalization rate in the first 30 min was 38% (±12.03%; n = 12) and the Pearson coefficient was 0.4976 (±0.08) (Figure 2A), obtained from three experiments and different randomly selected fields. After 1 h incubation, colocalization with caveolin fell below 30% ( Figure 2B).
Partial localization of the NT4 bound to HS in caveolae is consistent with previous studies which found that HSPGs and chondroitin sulfate proteoglycans (CSPGs) colocalized intracellularly with internalized caveolin-1 [22,23].

NT4 Peptide Internalization Follows More than One Pathway
An anti-caveolin antibody was used in a colocalization experiment with NT4 ( Figure 2A). MDA-MB-231 cells were incubated with the peptide at 37 • C for different times: 5, 15 and 30 min and 1 h. The cells were then stained with the caveolin antibody. The colocalization rate in the first 30 min was 38% (±12.03%; n = 12) and the Pearson coefficient was 0.4976 (±0.08) (Figure 2A), obtained from three experiments and different randomly selected fields. After 1 h incubation, colocalization with caveolin fell below 30% ( Figure 2B).
Partial localization of the NT4 bound to HS in caveolae is consistent with previous studies which found that HSPGs and chondroitin sulfate proteoglycans (CSPGs) colocalized intracellularly with internalized caveolin-1 [22,23].
To confirm the result, nystatin was used as inhibitor of caveolin-mediated endocytosis. Nystatin distorts the structure and functions of cholesterol-rich membrane domains, including caveolae, and is considered a selective inhibitor of the lipid raft and caveolae pathway [24]. MDA-MB-231 cells were incubated for 15 min with nystatin and NT4. Endocytosis was inhibited by 60% ( Figure 2D), in line with the fact that the caveolin-mediated pathway is not the only pathway of endocytosis of NT4.
Clathrin-mediated endocytosis was analyzed using an anti-clathrin antibody and chlorpromazine, a cationic amphipathic drug that prevents formation of new clathrin-coated vesicles [25]. The colocalization rate was 28%, with a Pearson coefficient of 0.4051 (±0.1; n = 9). The effect of chlorpromazine was evident and dose-dependent, showing reductions of 57% (±13%) and 27% (±13%) with 10 and 1 µg/mL of the inhibitor, respectively. The result is in agreement with previous findings showing that NT4 is also a ligand of LRP6, a co-receptor of low density lipoprotein (LDL) that is internalized via clathrin endosomes [13,26,27]. To confirm the result, nystatin was used as inhibitor of caveolin-mediated endocytosis. Nystatin distorts the structure and functions of cholesterol-rich membrane domains, including caveolae, and is considered a selective inhibitor of the lipid raft and caveolae pathway [24]. MDA-MB-231 cells were incubated for 15 min with nystatin and NT4. Endocytosis was inhibited by 60% ( Figure 2D), in line with the fact that the caveolin-mediated pathway is not the only pathway of endocytosis of NT4.
Clathrin-mediated endocytosis was analyzed using an anti-clathrin antibody and chlorpromazine, a cationic amphipathic drug that prevents formation of new clathrin-coated vesicles [25]. The colocalization rate was 28%, with a Pearson coefficient of 0.4051 (±0.1; n = 9). The effect of chlorpromazine was evident and dose-dependent, showing reductions of 57% (±13%) and 27% (±13%) with 10 and 1 g/mL of the inhibitor, respectively. The result is in agreement with previous findings showing that NT4 is also a ligand of LRP6, a co-receptor of low density lipoprotein (LDL) that is internalized via clathrin endosomes [13,26,27].
Dynasore is a specific inhibitor of dynamin [28], a protein that severs vesicles from the plasma membrane during clathrin-and caveolin-dependent endocytic processes [29] and also during macropinocytosis [30]. When treated with dynasore, MDA-MB-231 cells no longer internalized NT4 ( Figure 2E). The impairment of internalization was greater than with chlorpromazine and nystatin, consistent with the fact that dynasore hampered both clathrin-and caveolin-mediated endocytosis.
Macropinocytosis is a cell process by which much extracellular material is engulfed. It is driven by actin that induces membrane ruffling and formation of large endosomes, i.e., macropinosomes [31]. We investigated macropinocytosis as a further possible endocytic pathway for HSPG-mediated NT4 internalization. Rhodamine-labelled dextran was used in localization experiments. Dynasore is a specific inhibitor of dynamin [28], a protein that severs vesicles from the plasma membrane during clathrin-and caveolin-dependent endocytic processes [29] and also during macropinocytosis [30]. When treated with dynasore, MDA-MB-231 cells no longer internalized NT4 ( Figure 2E). The impairment of internalization was greater than with chlorpromazine and nystatin, consistent with the fact that dynasore hampered both clathrin-and caveolin-mediated endocytosis.
Macropinocytosis is a cell process by which much extracellular material is engulfed. It is driven by actin that induces membrane ruffling and formation of large endosomes, i.e., macropinosomes [31]. We investigated macropinocytosis as a further possible endocytic pathway for HSPG-mediated NT4 internalization. Rhodamine-labelled dextran was used in localization experiments.

Vesicle Trafficking in Triple-Negative Breast Cancer Cells
NT4 was internalized in a few minutes after membrane-binding, and once in the cytosol, it could be observed in vesicles with HSPG ( Figure 1). Hence, trafficking of NT4 is linked to that of HSPGs. Rab proteins are small GTPases that regulate intracellular membrane trafficking with active roles in formation, fusion and movement of vesicles [9,33]. Rab5 controls and regulates the intracellular transport of various receptors [10] and its presence on vesicles defines them as early endosomes. NT4 showed a colocalization rate with Rab5 of 39% (n = 11, SD ± 13) and a Pearson coefficient of 0.51 (n = 11, SD ± 0.06) in immunofluorescence experiments using a Rab5A antibody ( Figure 4A-D). NT4 and dextran partially colocalized, 26.44% (±8.31%; n = 10) ( Figure 3A,B), with a Pearson coefficient of 0.5339 (±0.11) ( Figure 3C). Amiloride is an inhibitor of macropinocytosis [32] and can impair NT4 internalization by more than 50% ( Figure 3D,E).

Vesicle Trafficking in Triple-Negative Breast Cancer Cells
NT4 was internalized in a few minutes after membrane-binding, and once in the cytosol, it could be observed in vesicles with HSPG ( Figure 1). Hence, trafficking of NT4 is linked to that of HSPGs. Rab proteins are small GTPases that regulate intracellular membrane trafficking with active roles in formation, fusion and movement of vesicles [9,33]. Rab5 controls and regulates the intracellular transport of various receptors [10] and its presence on vesicles defines them as early endosomes. NT4 showed a colocalization rate with Rab5 of 39% (n = 11, SD ± 13) and a Pearson coefficient of 0.51 (n = 11, SD ± 0.06) in immunofluorescence experiments using a Rab5A antibody ( Figure 4A-D).

Effects of NT4 on Cancer Cell Migration
The polycationic peptide NT4 has already been shown to interfere with cancer cell migration [34], impairing 2D and 3D migration of pancreas cancer cells and HUVEC endothelial cells and hampering invasive migration of MDA-MB-231 cells through collagen [18]. In a standard 2D Rab11, a small GTPase associated with endosome-recycling vesicles, gave a colocalization rate of 24% (n = 6; SD ± 5.89) and a low Pearson coefficient (0.38; n = 6, SD ± 0.039) ( Figure 5C), indicating scarce presence in the recycling endosomes.

Effects of NT4 on Cancer Cell Migration
The polycationic peptide NT4 has already been shown to interfere with cancer cell migration [34], impairing 2D and 3D migration of pancreas cancer cells and HUVEC endothelial cells and hampering invasive migration of MDA-MB-231 cells through collagen [18]. In a standard 2D migration experiment, NT4 slowed MDA-MB-231 migration in a wound-scratch experiment (Figure 6a). Cell migration is closely linked to cytoskeletal rearrangements and vesicle trafficking, cell events regulated by Rho GTPases [35].
We hypothesized that NT4 could interfere with Rho-GTPases, particularly CDC42, Rac-1 and RhoA, which have well-established roles in regulating actin polymerization and, thus, migration and endocytosis. Activation of CDC42, Rac1 and RhoA was stimulated by EGF in starved MDA-MB-231 cells.
G-LISA experiments [7] showed that CDC42-GTP is increased by EGF and that NT4 mitigates activation of CDC42 induced by the growth factor [36]; on the other hand, NT4 without EGF has no effect on CDC42 activation (Figure 6b).
Rac1-GTP was transiently activated by EGF in G-LISA experiments, as already described by other authors [37]. The activation was not significantly changed by NT4. Besides, basal levels of Rac-1-GTP, which are very low in MDA-MB-231 [38], were not affected by NT4 in a pulldown experiment (Supplementary Materials).
RhoA-GTP was decreased by EGF, as already observed by other authors [38,39], and NT4 did not significantly interfere with this effect. RhoA was reduced by EGF, NT4 and EGF with NT4 of 26%, 20% and 21%, respectively; the three groups showed no statistical differences-in other words, NT4 and EGF do not compete or synergize in RhoA activation. Besides, NT4 itself reduced the basal level of activated RhoA, even without EGF, thus with no interference with the growth factor function. migration experiment, NT4 slowed MDA-MB-231 migration in a wound-scratch experiment ( Figure  6a). Cell migration is closely linked to cytoskeletal rearrangements and vesicle trafficking, cell events regulated by Rho GTPases [35]. We hypothesized that NT4 could interfere with Rho-GTPases, particularly CDC42, Rac-1 and RhoA, which have well-established roles in regulating actin polymerization and, thus, migration and endocytosis. Activation of CDC42, Rac1 and RhoA was stimulated by EGF in starved MDA-MB-231 cells.
G-LISA experiments [7] showed that CDC42-GTP is increased by EGF and that NT4 mitigates activation of CDC42 induced by the growth factor [36]; on the other hand, NT4 without EGF has no effect on CDC42 activation (Figure 6b).
Rac1-GTP was transiently activated by EGF in G-LISA experiments, as already described by other authors [37]. The activation was not significantly changed by NT4. Besides, basal levels of Rac-1-GTP, which are very low in MDA-MB-231 [38], were not affected by NT4 in a pulldown experiment (Supplementary Materials).
RhoA-GTP was decreased by EGF, as already observed by other authors [38,39], and NT4 did not significantly interfere with this effect. RhoA was reduced by EGF, NT4 and EGF with NT4 of 26%, 20% and 21%, respectively; the three groups showed no statistical differences-in other words, NT4 and EGF do not compete or synergize in RhoA activation. Besides, NT4 itself reduced the basal level of activated RhoA, even without EGF, thus with no interference with the growth factor function.

Discussion
NT4 binds HSPGs on the plasma membrane and is internalized with HSPG in endosomes. Internalization occurred by different pathways; caveolin-and clathrin-dependent endocytosis and macropinocytosis were all employed by the peptide, in line with what we already know about HSPGs [2][3][4][5][6][7][8]. NT4 then localized in early and late endosomes in a time-dependent manner.
The polycationic peptide had previously been shown to impair migration of MDA-MB-231 and other cancer cell lines in 2D and 3D models. Our hypothesis was that HSPG endocytosis could affect actin organization, influencing two important phenomena of cancer development: movement and

NT4 binds HSPGs on the plasma membrane and is internalized with HSPG in endosomes.
Internalization occurred by different pathways; caveolin-and clathrin-dependent endocytosis and macropinocytosis were all employed by the peptide, in line with what we already know about HSPGs [2][3][4][5][6][7][8]. NT4 then localized in early and late endosomes in a time-dependent manner.
The polycationic peptide had previously been shown to impair migration of MDA-MB-231 and other cancer cell lines in 2D and 3D models. Our hypothesis was that HSPG endocytosis could affect actin organization, influencing two important phenomena of cancer development: movement and nourishment. Using NT4 to mimic heparin-binding ligands, we studied the effect of Rho-GTPases, specifically CDC42, Rac1 and RhoA.
CDC42 is a small GTPase, the multiple roles of which have been abundantly described. Its downstream effectors are various kinases, such as PAKs, MLKs, and MRCKs [40], which initiate processes such as cell polarity [41], cytoskeletal remodeling, migration, adhesion, membrane trafficking, proliferation and transcription [40]. Activated by EGF, CDC42 promotes actin branching in invadopodia, using its effector N-WASP and, in turn, Arp2/3 [37]. NT4 shared localization with HSPGs and had a negative effect on CDC42-activation triggered by EGF. The effect is presumably due to competitive inhibition of EGF on HS by NT4, which impairs the co-receptor activity of the proteoglycan and reduces EGFR activation (Figure 7). The reduction of the invasive and migratory phenotype of MDA-MB-231, induced by NT4, can be ascribed to this effect.
RhoA is a member of the Rho GTPase family and, as with Rac1 and CDC42, is often described as having a role downstream of EGF activation and also as promoting carcinogenesis and metastasis [37]. In fact, RhoA activation is damped by EGF in MDA-MB-231 [42], as also confirmed by our findings. Indeed, EGF reduced RhoA-GTP and the HS-binding peptide did not interfere with this receptor-mediated signaling. On the other hand, the peptide itself determined a small but solid reduction in active RhoA in breast cancer cells (Figures 6 and 7). Recently, a role of syndecans, not involving growth factor activation but rather the cytoplasmic domains of transmembrane syndecans that can act as signaling receptors, was observed in RhoA activation [43]. Our result, showing that NT4 dampens RhoA activation, a key step for cell movement, correlates well with the peptide anti-migratory activity. The observation that the HSPG-binding peptide NT4 down-regulates active RhoA without antagonizing EGF, is a reasonable confirmation that RhoA is directly activated through HSPG [43].
to understand possible roles of HSPGs in vesicle transport, not only by virtue of their ability to capture exosomes [44], which is well described, but also through regulating endosome traffic, which is vital for cell movement and nourishment. Besides, the results support the observation of few other studies, showing that HSPGs are not simple co-receptors that promote or prevent ligand binding to their receptors, as they were considered for many years, but have a direct role in signaling. Figure 7. EGF binding to the receptor, assisted by heparan sulfate proteoglycans (HSPGs), triggers CDC42 activation. NT4 binding to HSPG disrupts the EGF-EGFR-HSPG triad and reduces CDC42 activation. Besides, NT4 dampens RhoA activation, without interference with EGF-EGFR systems, confirming the ability of HSPGs to directly modulate this GTPase. Figure 7. EGF binding to the receptor, assisted by heparan sulfate proteoglycans (HSPGs), triggers CDC42 activation. NT4 binding to HSPG disrupts the EGF-EGFR-HSPG triad and reduces CDC42 activation. Besides, NT4 dampens RhoA activation, without interference with EGF-EGFR systems, confirming the ability of HSPGs to directly modulate this GTPase.
In conclusion, our data suggest that HSPGs are, as other authors observed, internalized following different pathways and are localized in specific endosomes. Further studies are necessary to understand possible roles of HSPGs in vesicle transport, not only by virtue of their ability to capture exosomes [44], which is well described, but also through regulating endosome traffic, which is vital for cell movement and nourishment. Besides, the results support the observation of few other studies, showing that HSPGs are not simple co-receptors that promote or prevent ligand binding to their receptors, as they were considered for many years, but have a direct role in signaling.
HPLC purification was performed on a C18 Jupiter column (Phenomenex, Torrance, CA, USA). Water with 0.1% TFA (A) and acetonitrile (B) were used as eluents. Linear gradients over 30 min were run at flow rates of 0.8 and 4 mL/min for analytical and preparatory procedures, respectively.

Immunofluorescence
MDA-MB-231 cells were plated at a density of 5 × 10 4 per well in 24-well plates with glass cover slides and maintained overnight at 37 • C, 5% CO 2 .
4.3.1. Detection of NT4 Binding to HS and Internalization with HS Cells were incubated with 2 µM NT4-biotin and 0.5 µg/mL Streptavidin-Atto 488 in PBS-1% BSA at room temperature for 15 min, then washed and grown in medium for 5, 15 and 30 min at 37 • C to allow peptide internalization. They were then fixed with PBS-4% paraformaldehyde (PFA) and saturated with PBS-5% BSA 0.3%-Triton X-100. HS was stained using 1 µg/mL anti-heparan sulfate (10E4 epitope) (Amsbio, Abingdon, UK) in PBS-1% BSA for 1 h, followed by incubation with 1 µg/mL anti-mouse IgG Rhodamine Red-X (Thermo Fisher Scientific, Waltham, MA, USA). Nuclei were stained with DAPI (0.5 µg/mL in PBS-1% BSA). Each step was followed by three washes in PBS. Peptide binding was analyzed by confocal laser microscope (Leica TCS SP5) with 380 λ ex and 450-470 λ em, 501 λ ex and 523 λ em and 633 λ ex and 660-680 λ em for DAPI, Atto 488 and Atto 647, respectively. The experiment was repeated three times and three different fields were acquired for each sample.

Detection of NT4 Internalization in Presence of Pharmacological Agents
Cells were pretreated for 15 min with 1-10 µg/mL Chlorpromazine (Merck, Darmstadt, Germany), 25 µg/mL Nystatin (Merck), 80 µM Dynasore (Merck) and 50 µM Amiloride (Merck) at 37 • C and then incubated with 2 µM NT4-biotin and 0.5 µg/mL Streptavidin-Atto 488 in PBS-1% BSA at room temperature for 15 min. To study macropinocytosis in the presence of Amiloride, 5 mg/mL Rhodamine B isothiocyanate-Dextran in PBS-1% BSA (Merck) was incubated at room temperature for 15 min. Cells were then washed and grown in medium, with and without pharmacological agents, for 15 and 30 min at 37 • C to allow peptide internalization. After internalization, they were fixed with PBS-4% PFA and plasma membranes were stained with 2.5 µg/mL wheat germ agglutinin, Alexa Fluor 647 (Thermo Fisher Scientific) conjugate in PBS-1% BSA, incubated for 10 min at room temperature. Nuclei were stained with DAPI (0.5 µg/mL in PBS-1% BSA). Each step was followed by three washes in PBS.
Peptide binding was analyzed by confocal laser microscope (Leica TCS SP5) with 501 λ ex and 523 λ em for Atto 488, 633 λ ex and 660-680 λ em for Atto 647 and 364 λ ex and 458 λ em for DAPI. Images were single Z-planes for all channels used. All images were processed and quantified using the Leica Application Suite X program (LAS X) of the confocal laser microscope (Leica TCS SP5). The experiment was repeated at least three times and at least three different fields were acquired for each sample.