Glycosaminoglycan modification of NRP1 exon 4-skipping variant drives colorectal cancer metastasis via endosomal-exosomal trafficking

Abstract

Neuropilin-1 (NRP1) is a transmembrane glycoprotein that functions as a co-receptor with various cellular functions. Our previous studies identified the NRP1 exon 4-skipping (NRP1-ΔE4) splice variant as an aggressive metastasis driver by activating endosomal signals. Here, we demonstrate the critical role of glycosaminoglycan (GAG) modification in regulating NRP1-ΔE4’s cellular trafficking and oncogenic activity. NRP1-ΔE4 undergoes constitutive internalization into endosomes and subsequent exosomal release from colorectal cancer (CRC) cells. Exosomal NRP1-ΔE4 enhances the migration and invasion of both donor and recipient CRC cells. Genetic or pharmacological inhibition of exosome secretion, or immunodepletion of exosomal NRP1-ΔE4, markedly reduces its metastatic potential. Notably, GAG modification at the O-glycosylation site Ser612 is essential for NRP1-ΔE4’s endosomal trafficking and exosomal release. This modification also promotes the formation of a trimeric complex with Met and β1-integrin, leading to their co-internalization and accumulation in endosomes, which activates FAK signaling and drives CRC metastasis. These findings reveal GAG modification as a key regulatory process that governs the endosomal-exosomal trafficking of NRP1-ΔE4 to facilitate CRC cell dissemination.

1. Introduction

Neuropilin-1 (NRP1) is a cell surface glycoprotein that acts as a co-receptor for various extracellular ligands, including semaphorins, VEGF, TGF-β, HGF, FGF, and PDGF. Initially implicated in axon guidance and angiogenesis regulation, NRP1 has since been linked to cancer progression and metastasis. Its multifunctionality is attributed to its four ligand-binding domains (a1, a2, b1, b2), a membrane-proximal MAM domain (c), and a cytoplasmic C-terminal domain [1, 2]. NRP1 can activate diverse intracellular signaling cascades triggered by specific ligands through interactions with several receptor tyrosine kinases (VEGFR, Met, EGFR) and other transmembrane proteins (integrins, plexins/semaphorins) [1-5]. Following ligand binding, these receptors undergo internalization via endocytosis and are transported to early and late endosomes before either recycling back to the plasma membrane or being targeted for degradation [6]. Endosomal signaling, initiated by the endocytosis of receptor tyrosine kinases such as Met and EGFR, plays a pivotal role in various cellular functions, including organismal development and cancer progression [7-9].

The human NRP1 gene is located on chromosome 10p12 and consists of 17 exons [10, 11]. Several human NRP1 isoforms generated through alternative RNA splicing have been reported [11-14]. Splice variants lacking the transmembrane domain are known to antagonize VEGF₁₆₅ and NRP1-mediated cellular activities, exerting anti-angiogenic and anti-tumorigenic effects [11, 12, 15]. The NRP1-Δ7 splice variant, produced via an alternative 5’ splice site in exon 11, results in the deletion of a 7-amino-acid sequence (positions 615-621) near the O-glycosylation site at Ser612, which is required for Ser612-mediated glycosaminoglycan (GAG) modification, and also exhibits anti-tumorigenic effects [13]. In contrast, the NRP1-ΔE16 variant, which skips exon 16, shows no functional differences compared to wild-type (WT) NRP1 [14]. Recently, we identified two human NRP1 splice variants, NRP1-ΔE4 (exon 4 skipping) and NRP1-ΔE5 (exon 5 skipping), in colorectal cancer (CRC) cell lines and tissue specimens [16]. Interestingly, compared to NRP1-WT, both variants exhibit increased endocytosis and recycling activity with reduced degradation, leading to their accumulation in endosomes. Their enhanced endocytic trafficking, triggered by HGF but not VEGF₁₆₅ stimulation, is attributed to the partial loss of the a2 domain of NRP1, resulting in defective N-glycosylation at Asn150 or Asn261, respectively. Furthermore, these variants enhance interactions with Met and β1-integrin receptors, promoting their co-internalization and co-accumulation as a trimeric complex in endosomes. This complex generates persistent endosomal signals that activate FAK signaling, driving CRC cell migration, invasion, and metastasis. Notably, NRP1-ΔE4 is expressed in 71% of CRC tissues and significantly correlates with CRC progression across stages I–IV, whereas NRP1-ΔE5 is less frequent (25%) and not significantly associated to CRC progression [16]. These findings highlight NRP1-ΔE4 as a critical oncogenic splice variant that promotes CRC metastasis.

In this study, we demonstrate that NRP1-ΔE4 is secreted via exosomes and undergoes continuous internalization, maintaining the active NRP1-ΔE4/Met/β1-integrin trimeric complex within endosomes to sustain FAK signaling and promote CRC cell dissemination. Furthermore, our findings identify GAG side chain addition at the O-glycosylation site Ser612 as a critical post-translational modification that regulates the endosomal-exosomal trafficking of NRP1-ΔE4, ultimately enhancing CRC metastasis.

2. Materials and methods

2.1. Cell lines and culture

Human HCT116 and HT29 CRC cell lines were obtained from ATCC. Primary human CRC cell lines Pt93 and Pt130 were established from patient primary tumors at the Markey Cancer Center, University of Kentucky [16, 17]. HCT116 and HT29 cells were cultured in McCoy's medium (Sigma), while primary CRC cells were cultured in DMEM medium (Sigma), both supplemented with 10% FBS (Sigma), 100 μg/ml streptomycin, and 100 units/ml penicillin. All cells were maintained at 37 °C in 5% CO₂. HCT116 and HT29 cells with stable expression of NRP1-WT, NRP1-ΔE4, or vector control were generated in our previous study [16]. All cell lines were routinely tested for mycoplasma contamination by PCR using the e-Myco Plus kit (iNtRON Biotechnology) and authenticated by short tandem repeat (STR) profiling (Genetica).

2.2. Growth factors and chemicals

VEGF₁₆₅ and HGF were obtained from R&D Systems. LysoTracker Deep Red was purchased from Thermo Fisher Scientific. Cycloheximide, sulfisoxazole, chondroitinase ABC, and heparinase were purchased from Sigma.

2.3. Plasmids and virus infection

The human NRP1-WT and NRP1-ΔE4 genes were cloned into the pLenti6.3 or pCMV6-entry vector, as previously described [16]. The NRP1 mutants ΔE4/Δ7 and ΔE4/S612A were generated using the QuikChange XLII mutagenesis kit (Stratagene). Lentiviral shRNAs targeting human Hrs and Rab27a were cloned into the pLKO.1 vector (Sigma), while the Non-Target Control shRNA (SHC002) was obtained from Sigma. All sequences were verified by automated DNA sequencing. Primers used to generate NRP1 mutants ΔE4/Δ7 and ΔE4/S612A, along with shRNA sequences, are listed in Table S1 and Table S2, respectively. Lentiviruses were produced by co-transfecting the indicated pLenti6.3 or pLKO.1 constructs with packaging (psPAX2) and envelope (pMD2.G) plasmids into HEK293T cells using Lipofectamine 3000 (Thermo Fisher Scientific), as described previously [16, 18]. The medium was changed the next day, and virus-containing supernatants were collected at 48 h after transfection, followed by filtration through a 0.45 μM filter. HCT116 or HT29 cells were infected with the filtered viral supernatants three times within a 36 h period in the presence of 8 μg/ml polybrene. After infection, cells were selected with puromycin (2 μg/ml) or hygromycin (250 μg/ml) for 7–10 days [16, 18] to establish stable transfectants with specific protein expression or knockdown.

2.4. Purification of exosomes

Exosomes were purified following the method described by Chen et al [19]. Cells were cultured in media supplemented with 10% exosome-depleted FBS (System Biosciences). Culture supernatants were collected after 48 h, and exosomes were isolated using a standard differential centrifugation protocol [19, 20]. Briefly, culture supernatants were centrifuged at 2,000 × g for 20 min to remove cell debris and dead cells. Microvesicles were eliminated by centrifugation at 16,500 × g for 45 min. The remaining supernatants were then centrifuged at 100,000 × g for 2 h at 4 °C (WX 100+ Ultracentrifuge, Thermo Fisher Scientific). The exosome pellet was resuspended in PBS and subjected to a final ultracentrifugation at 100,000 × g for 2 h to ensure purity.

2.5. Characterization of purified exosomes

The size and concentration of exosomes purified from cell culture supernatants were measured using the Nanoparticle Tracking Analyzer ZetaView PMX 110 (Particle Metrix, Munich, Germany), equipped with fast video capture and particle-tracking software (ZetaView 8.05.12). Exosomal protein concentration was determined using the Micro BCA Protein Assay Kit (#23235, Thermo Fisher Scientific). To verify the purified exosomes, transmission electron microscopy (TEM) analysis was performed as previously described [19]. Briefly, purified exosomes suspended in PBS were deposited onto formvar carbon coated-nickel grids. After staining with 2% uranyl acetate, the grids were air-dried and visualized using a Talos F200X G2 transmission electron microscope. For immunogold labeling, purified exosomes suspended in PBS were placed on formvar carbon-coated nickel grids, blocked, and incubated with rabbit anti-NRP1 monoclonal antibody (#3725, Cell Signaling Technology), followed by incubation with an anti-rabbit secondary antibody conjugated with protein A-gold particles (5 nm) (G3779, Sigma). After each staining step, grids were washed five times with PBS and ten times with double-distilled water, followed by contrast staining with 2% uranyl acetate.

2.6. Immunoprecipitation and immunoblot analysis

Cells were lysed in RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, 10% glycerol) supplemented with protease and phosphatase inhibitor cocktails. Protein concentrations were measured using the BCA protein assay reagent (Thermo Fisher Scientific). For immunoprecipitation, 500 μg of total protein was incubated with 2 μg of the indicated antibody overnight at 4 °C, followed by a 3 h incubation with a 50% slurry of protein G sepharose beads. Beads were washed three times with lysis buffer, and immunoprecipitated protein complexes were resuspended in 2× Laemmli sample buffer for immunoblot analysis. For immunoblotting, equal amounts of total protein were resolved by SDS-PAGE and transferred to PVDF membranes. Membranes were blocked with 5% non-fat milk in Tris-buffered saline with 0.1% Tween-20 and incubated overnight at 4 °C with the primary antibody, followed by incubation with a horseradish peroxidase (HRP)-conjugated secondary antibody. Bands were detected using a chemiluminescence substrate (Thermo Fisher Scientific). A list of all antibodies used in this study, along with their dilutions, is provided in Table S3.

2.7. Immunofluorescent staining

Cells (5×10⁴) were grown on collagen-precoated coverslips in regular growth medium containing 10% FBS or serum-starved overnight, followed by stimulation with VEFG₁₆₅ (50 ng/ml) or HGF (50 ng/ml) for 30 min. Cells were then fixed with 4% paraformaldehyde in PBS for 15 min, permeabilized with 0.2% Triton X-100 and 0.5% BSA in PBS for 5 min, and blocked with 4% BSA in PBS for 10 min. Next, cells were incubated with the indicated primary antibody overnight at 4 °C, followed by three washes with 0.05% Triton X-100 in PBS and incubation with the indicated secondary antibody for 1 h. After another three washes, cells were mounted with DAPI-containing mounting medium (H-1200, Vector Laboratories) and visualized under a Nikon A1⁺-Ti2 confocal microscope. For experiments involving exosome treatment, cells were incubated with exosomes at a concentration of 25 μg/ml for 24 h. For image quantifications, fields were randomly selected based on DAPI staining. In co-localization experiments, ten images were captured, and co-localization percentages were analyzed using ImageJ. A list of primary and secondary antibodies used in this assay is provided in Table S3

2.8. Cell migration and invasion assays

Cell migration and invasion assays were performed using Boyden chambers coated with collagen or Matrigel, respectively, following the manufacturer's instructions (BD Biosciences) [16]. Briefly, cells were added to the upper chamber of the transwell insert, while complete medium containing 10% FBS was added to the bottom chamber as a chemoattractant. For assays involving exosome treatment, cells were co-cultured with exosomes at a concentration of 25 μg/ml. To assess the impact of exosomal NRP1-ΔE4 on cell migration and invasion, purified exosomes were preincubated with the NRP1 antibody vesencumab or an IgG control for 1 h at 4°C. The Boyden chambers were then incubated at 37 °C with 5% CO₂ for the indicated time periods. After incubation, cells remaining in the upper compartment were removed with a cotton swab, While migrated or invaded cells on the filter surface facing the bottom chamber were fixed in 4% paraformaldehyde and stained with 0.2% crystal violet. The number of migrated or invaded cells was counted in at least five areas at × 20 magnification using an inverted microscope.

2.9. Wound healing assay

Cells were seeded in 6-well plates and grown to confluence overnight. The next day, a scratch wound was created in the cell monolayer using a 10 μl pipette tip, and the medium was replaced to remove cell debris. Images of the wound were captured at 0, 24, and 48 h post-scratching. Wound healing was assessed by measuring the wound width between the edges using ImageJ. Experiments were performed at least three times.

2.10. Cycloheximide chase assay

Cells were treated with cycloheximide (50 μg/ml) and harvested at the indicated time points. Cell lysates were prepared using RIPA buffer, and equal amounts of total protein were analyzed by immunoblotting.

2.11. Detection of GAG-modified NRP1-ΔE4

Chondroitinase ABC (C2905, Sigma) and heparinase III (H8891, Sigma) were added to the culture medium of HCT116 cells expressing NRP1-ΔE4 or its mutants at a final concentration of 1 unit/ml. The cells were then incubated for 2 h at 37°C in 5% CO₂. After incubation, the cells were lysed in RIPA buffer, and the lysates were analyzed by immunoblotting using an anti-NRP1 antibody (#3725, Cell Signaling Technology).

2.12. Cell viability assay

Cells (5 × 10⁴/ well) were seeded in triplicate in 6-well plates. The number of viable cells was counted daily for 3 days using the Vi-CELL XR 2.03 (Beckman Coulter). Each experiment was performed in triplicate and repeated at least three times.

2.13. Animal studies

Male and female athymic nude mice (5-6 weeks old) were purchased from Taconic, maintained, and treated under specific pathogen-free conditions. Experiments were conducted under a protocol (#2021-3839) approved by the University of Kentucky Institutional Animal Care and Use Committee. For the experimental lung metastasis assay, cells co-expressing firefly luciferase and GFP were injected into the tail vein (1 × 10⁶/ mouse) of athymic nude mice as previously described [21]. For the orthotopic mouse model of CRC liver metastasis, cells (4 × 10⁶/ mouse) co-expressing firefly luciferase and GFP were injected into the cecal submucosa of athymic nude mice, as described [22]. To monitor metastasis, mice, as well as lung and liver tissues, were imaged for luciferase signals using the IVIS Spectrum system, and the results were analyzed using Living Image 3.0 software (Caliper Life Science).

2.14. Statistical analysis

Statistical analysis for each experiment was performed as described in the corresponding figure legends. Data between groups were compared using a two-tailed unpaired Student’s t test, one-way ANOVA, and the Mann–Whitney test, as appropriate. All quantitative data are presented as mean ± SEM, and p < 0.05 was considered statistically significant. GraphPad Prism 9 software was used for these analyses.

3. Results

3.1. NRP1-ΔE4 is released from CRC cells via secreted exosomes

Exosomes are vesicles derived from late endosomes, also known as multivesicular bodies, and are secreted into the extracellular environment by most eukaryotic cells [23-28]. Our previous work demonstrated that NRP1-ΔE4 undergoes continuous shuttling between the cell surface and endosomes without being translocated to lysosomes for degradation [16]. Moreover, NRP1-ΔE4 is rarely detected on the plasma membrane [16]. These findings suggest that NRP1-ΔE4 is rapidly internalized after recycling to the cell surface and/or transported to the extracellular space via secreted exosomes. Indeed, we found that NRP1-ΔE4 co-localized with several common exosome marker proteins [27], including CD63, CD81, and Alix (Fig. 1A and B). To further investigate this, we purified exosomes from the culture supernatants of HCT116 cells expressing either NRP1-ΔE4 or NRP1-WT using differential centrifugation [19, 20, 29] and verified the isolated exosomes through nanoparticle tracking analysis, transmission electron microscopy, and immunoblot analysis of positive and negative exosome marker proteins. Our analysis confirmed that the purified exosomes were spherical particles with an average diameter of approximately 90 nm (Fig. 1C and D), consistent with the typical morphology and size of exosomes [23-28]. Additionally, they expressed three positive exosome marker proteins (CD63, CD81, Alix) but lacked the negative exosome marker GM130 (Fig. 1E). Notably, immunoblot analysis revealed a higher level of NRP1-ΔE4 in exosomes compared to NRP1-WT (Fig. 1E and Fig. S1A), while the overall exosome secretion levels remained comparable between cells expressing the two isoforms (Fig. 1F). Using immunoelectron microscopy, we further observed a substantial presence of NRP1-ΔE4 on the surface of exosomes (Fig. 1G). Exosomes are generated and secreted through a well-defined intracellular trafficking pathway [23, 24, 26]. Hrs, a subunit of the Endosomal Sorting Complex Required for Transport (ESCRT), mediates cargo recognition and sorting into exosomes [30, 31]. Through immunofluorescence and co-immunoprecipitation analyses, we found that NRP1-ΔE4 co-localized with Hrs and exhibited a stronger interaction with Hrs than NRP1-WT (Fig. 1H and I and Figs. S1B and C). Genetic knockdown of Hrs in NRP1-ΔE4-expressing HCT116 cells using shRNAs reduced exosome production, as indicated by decreased levels of CD63, CD81 and Alix, and led to a corresponding decrease in NRP1-ΔE4 levels in exosomes (Fig. 1J). Furthermore, knockdown of Rab27a, a key mediator of exosome release [32], also inhibited NRP1-ΔE4 secretion via exosomes (Fig. 1K). Collectively, these findings strongly indicate that NRP1-ΔE4 is secreted through exosomes via the Hrs/Rab27a-mediated pathway.

Fig. 1.

NRP1-ΔE4 is secreted in exosomes via the Hrs/Rab27a-mediated pathway. (A) Representative confocal images of HT29 cells expressing NRP1-WT or NRP1-ΔE4, immunostained for NRP1 (green), exosome markers (red), including CD63, CD81, and Alix, and DAPI (blue). Scale bar, 10 μm. (B) Co-localization analysis of the indicated NRP1 isoforms with exosome markers as shown in (A), using ImageJ software. (C) Characterization of exosomes purified from HCT116 cells expressing NRP1-WT or NRP1-ΔE4, assessed using the ZetaView nanoparticle tracking analyzer. (D) Representative TEM image of exosomes purified from HCT116 cells expressing NRP1-ΔE4. Scale bar, 100 nm. (E) Immunoblot analysis of the indicated proteins in whole cell lysates (WCL) and exosomes from HCT116 cells expressing NRP1-WT or NRP1-ΔE4. (F) The number of secreted exosomes from HCT116 cells expressing NRP1-WT or NRP1-ΔE4 was measured using a ZetaView nanoparticle tracking analyzer. (G) Representative TEM image of NRP1-ΔE4-expressing HCT116 cell-derived exosomes, immunogold-labelled with anti-NRP1 antibodies. Arrowheads indicate 5-nm gold particles. Scale bar, 50 nm. (H and I) HCT116 cells expressing NRP1-WT or NRP1-ΔE4 were assessed by confocal microscopy (H) for immunofluorescent staining of NRP1 (green), Hrs (red), and DAPI (blue), followed by quantification of co-localization between Hrs and the indicated NRP1 isoforms using ImageJ software (I). Scale bar, 10 μm. (J and K) Immunoblot analysis of the indicated proteins in WCL and exosomes from NRP1-ΔE4-expressing HCT116 cells with stable expression of two different sets of Hrs (J) or Rab27a (K) shRNA or control shRNA. Graphical data in (B) and (H) are presented as mean ± SEM (n = 3 independent experiments). **p < 0.01; ***p < 0.001, using the Student’s t test.

3.2. Exosomal NRP1-ΔE4 enhances CRC cell migration and invasion

Exosomes are increasingly recognized as key mediators of intercellular communication. They can be transferred locally or systemically upon uptake by recipient cells and play a crucial role in regulating tumor progression and metastasis [24, 28, 33]. To investigate the functional significance of exosomal NRP1-ΔE4 in CRC cell dissemination, HCT116 and HT29 cells were treated with exosomes isolated from HCT116 cells expressing either NRP1-WT or NRP1-ΔE4. Consistent with our previous observations on the localization of NRP1-WT and NRP1-ΔE4 in CRC cells [16], exosomal NRP1-ΔE4 was predominantly internalized by both HCT116 and HT29 CRC cells (Fig. 2A and Fig. S2A). In contrast, exosomal NRP1-WT primarily remained on the cell membrane, as confirmed by the membrane protein α6-integrin, which was used as a positive control (Fig. 2A and Fig. S2A). Notably, exosomal NRP1-ΔE4 promoted HCT116 and HT29 cell migration and invasion by 4-6 fold more than exosomal NRP1-WT, as assessed by Boyden chamber assays (Fig. 2B-D and Figs. S2B and C). Furthermore, the migratory and invasive capabilities of NRP1-WT-expressing HCT116 cells were further enhanced upon treatment with exosomes from NRP1-ΔE4-expressing CRC cells (Figs. S2D-F). Moreover, exosomes derived from Pt93 primary CRC cells, which predominantly express endogenous NRP1-ΔE4 [16], significantly increased HCT116 cell migration and invasion compared to exosomes from Pt130 primary CRC cells, which express only endogenous NRP1-WT[16] (Figs. S2G-I). Conversely, genetic inhibition of exosomal NRP1-ΔE4 secretion by silencing Hrs or Rab27a profoundly suppressed NRP1-ΔE4-driven cell invasion in both donor and recipient HCT116 cells (Fig. 2E and F and Figs. S2J and K). Additionally, immunodepletion of exosomal NRP1-ΔE4 using the clinically relevant NRP1 antibody vesencumab [34] substantially reduced cell migration and invasion in recipient HCT116 cells (Fig. 2G-K). Sulfisoxazole (SFX), an FDA-approved antibiotic recently identified as an inhibitor of small extracellular vesicle biogenesis and secretion [35], significantly reduced exosome secretion from NRP1-ΔE4-expressing HCT116 cells, lowered exosomal NRP1-ΔE4 levels, and suppressed NRP1-ΔE4-driven CRC cell migration and invasion without causing detectable cytotoxic effects (Fig. 2L-Q). Collectively, these findings demonstrate that exosome-mediated NRP1-ΔE4 secretion not only promotes migration and invasion in recipient CRC cells, including those expressing NRP1-WT, but is also essential for NRP1-ΔE4-expressing CRC cells to maintain their aggressive migratory and invasive phenotypes.

Fig. 2.

Exosomal NRP1-ΔE4 enhances the migratory and invasive capabilities of CRC cells. (A) HCT116 cells were treated with exosomes isolated from NRP1-WT- or NRP1-ΔE4-expressing HCT116 cells for 24 h, followed by immunofluorescent staining for α6-integrin (green), NRP1 (red), and DAPI (blue). Scale bar, 10 μm. (B-D) Transwell migration and invasion assays were performed on HCT116 cells treated with exosomes from NRP1-WT- or NRP1-ΔE4-expressing HCT116 cells for 24 h and 48 h, respectively. Scale bar, 100 μm. The number of migrated and invaded cells per field (n =5) was counted. Results are expressed as fold changes in the number of migrated (C) or invaded (D) cells relative to those treated with exosomes from NRP1-WT cells. (E and F) Transwell invasion assays of HCT116 cells treated with exosomes from NRP1-ΔE4-expressing HCT116 cells with stable knockdown of Hrs or Rab27a using two different sets of shRNA, or control shRNA, for 48 h. Scale bar, 100 μm. Results are expressed as a percentage of invasion relative to cells treated with exosomes from shCtrl-expressing cells. (G) Exosomes derived from NRP1-ΔE4-expressing HCT116 cells were immunodepleted of exosomal NRP1-ΔE4 via immunoprecipitation (IP) using vesencumab (ves) or an IgG control, followed by immunoblot analysis with NRP1 antibodies. (H) The relative levels of exosomal NRP1-ΔE4 following IP with ves were quantified using ImageJ software by comparison with the IgG control, as shown in (G). (I-K) Transwell migration and invasion analyses of HCT116 cells treated with exosomes immunodepleted of NRP1-ΔE4 using ves or an IgG control, as shown in (G and H). Results are expressed as a percentage of migration (J) or invasion (K) relative to IgG control exosomes. (L and M) NRP1-ΔE4-expressing HCT116 cells were treated with the indicated concentrations of SFX for 24 h, followed by measuring the number of secreted exosomes (L) and immunoblot analysis (M) of the indicated proteins in exosomes. (N-P) Transwell migration and invasion assays were performed on NRP1-ΔE4-expressing HCT116 cells treated with the indicated concentrations of SFX for 24 h and 48 h, respectively. Scale bar, 100 μm. Results are expressed as percentages of cell migration (O) or invasion (P) relative to DMSO-treated controls. (Q) NRP1-ΔE4-expressing HCT116 cells were treated with the indicated concentrations of SFX for the specified times, following by counting the number of viable cells. All graphical data are presented as mean ± SEM (n = 3 independent experiments). *p < 0.05; **p < 0.01; ***p < 0.001, using the Student’s t test. ns, not significant, using a one-way ANOVA test.

3.3. GAG modification modulates the endocytosis and exosomal release of NRP1-ΔE4

Our recent work demonstrates that the loss of N150-linked glycosylation in NRP1-ΔE4 increases its sensitivity to HGF-stimulated endocytosis, rather than VEGF₁₆₅-stimulated endocytosis, under basal cell culture conditions [16]. In contrast, VEGF₁₆₅ primarily induces full internalization of NRP1-WT from the plasma membrane, whereas NRP1-ΔE4 shows minimal internalization in response to VEGF₁₆₅ stimulation [16]. GAG modification of NRP1 at the O-glycosylation site Ser612 has been shown to play diverse roles in modulating VEGF signaling, migration, and cancer invasion in a cell type-dependent manner [36-38]. To investigate the role of GAG modification and its potential coordination with the loss of N150-linked glycosylation in regulating NRP1-ΔE4 trafficking and oncogenic effects in CRC, we generated non-GAG-modified NRP1-ΔE4 mutants by either deleting 7 amino acids (positions 615-521) [13] (NRP1-ΔE4/Δ7) or replacing the O-glycosylation site serine 612 with alanine [36] (NRP1-ΔE4/S612A) (Fig. S3A). Expression of either NRP1-ΔE4/Δ7 or NRP1-ΔE4/S612A in HCT116 cells resulted in the loss of high molecular weight products recognized by the NRP1 antibody, compared to those observed in NRP1-ΔE4-expressing cells (Fig. 3A). GAG addition at Ser612 in NRP1 involves chondroitin sulfate (CS) or heparin sulfate (HS) [36-38]. Treating NRP1-ΔE4-expressing HCT116 cells with chondroitinase or heparinase, enzymes that digest CS and HS, respectively, showed that chondroitinase treatment alone completely removed the high molecular weight products, whereas heparinase had no effect (Fig. S3B). These findings indicate that NRP1-ΔE4 in HCT116 cells primarily exists in a chondroitin sulfate GAG-modified form.

Fig. 3.

GAG modification is essential for the endosomal trafficking and exosomal release of NRP1-ΔE4. (A) Cell lysates from HCT116 cells expressing Flag-tagged NRP1-ΔE4 or its mutants (ΔE4/Δ7, ΔE4/S612A) were immunoprecipitated with anti-Flag antibodies, followed by immunoblot analysis with anti-NRP1 antibodies. The smear of higher molecular weight products corresponds to the glycosaminoglycan (GAG)-modified form of NRP1-ΔE4. (B) Immunoblot analysis of HT29 and HCT116 cells expressing NRP1-WT, NRP1-ΔE4, NRP1-ΔE4/Δ7, or vector control for the indicated proteins. (C and D) Representative confocal images of HT29 (C) and HCT116 (D) cells expressing NRP1-WT, NRP1-ΔE4 or NRP1-ΔE4/Δ7, immunostained for α6-integrin (green), NRP1 (red), and DAPI (blue). Scale bars, 10 μm. (E-I) HT29 cells expressing the indicated NRP1 isoforms were assessed by confocal microscopy (E, F and H) for immunofluorescent staining of NRP1 (green), endosome markers or lysotracker (red), and DAPI (blue), followed by quantification of co-localization between the indicated NRP1 isoforms and EEA1, Rab7, Rab11 (Fig. S3D), or lysotracker using ImageJ software (G and I). Scale bars, 10 μm. (J) HCT116 cells expressing the indicated NRP1 isoforms were treated with 50 μg/ml cycloheximide (CHX) for the indicated times, followed by immunoblot analysis. (K) The immunoblots of NRP1 isoforms shown in (J) were quantified using ImageJ software. The levels of the indicated NRP1 isoforms were normalized to β-actin at each time point. Results are expressed as percentages of remaining NRP1 isoforms relative to time 0 controls. (L) HCT116 cells expressing NRP1-ΔE4 or NRP1-ΔE4/Δ7 were lysed and immunoprecipitated with NRP1 antibodies or IgG control, followed by immunoblot analysis of the indicated proteins. (M) The relative levels of Hrs were quantified using ImageJ software by normalizing its binding to the indicated NRP1 isoforms as shown in (L). (N) Representative confocal images of HT29 cells expressing NRP1-ΔE4 or NRP1-ΔE4/Δ7, immunostained for NRP1 (green), exosome markers (red), including CD63, CD81, and Alix, and DAPI (blue). Scale bar, 10 μm. (O) Quantification of co-localization between the indicated NRP1 isoforms and exosome markers (CD63, CD81, and Alix) using ImageJ software. (P and Q) Exosomes isolated from HCT116 cells expressing the indicated NRP1 isoforms were analyzed by immunoblot (P) and quantified for exosome secretion (Q). All graphical data are presented as mean ± SEM (n = 3 independent experiments). **p < 0.01; ***p < 0.001; ns, not significant, using the Student’s t test.

Our previous study showed that while NRP1-WT and NRP1-ΔE4 mRNA expression was detected in the HCT116 CRC cell line, their protein expression levels were barely detectable in this cell line [16]. Additionally, NRP1-ΔE4 was not expressed in the HT29 CRC cell line, and NRP1-WT expression was significantly lower [16]. To characterize the specific function of GAG-modified NRP1-ΔE4, we first stably expressed NRP1-WT, ΔE4, and ΔE4/Δ7 at comparable levels in both HCT116 and HT29 CRC cells (Fig. 3B). Expression of NRP1-WT, ΔE4, or ΔE4/Δ7 in these cells had no discernible impact on the protein levels of the Met and β1-integrin receptors (Fig. 3B). Interestingly, NRP1-ΔE4/Δ7 exhibited reduced internalization compared to the predominantly intracellular localization of NRP1-ΔE4 in both HCT116 and HT29 cells under standard culture conditions with 10% fetal bovine serum (FBS). In contrast, NRP1-WT was localized on the plasma membrane, as indicated by the membrane protein α6-integrin, which was used as a positive control (Figs. 3C and D). Under FBS-free culture conditions, NRP1-ΔE4/Δ7 also showed reduced internalization compared to NRP1-ΔE4 upon HGF stimulation (Fig. S3C). Both NRP1-ΔE4 and NRP1-ΔE4/Δ7 exhibited less internalization than the complete internalization observed with NRP1-WT following VEGF₁₆₅ stimulation (Fig. S3C). Furthermore, co-immunofluorescence staining of NRP1 with various intracellular vesicle trafficking markers revealed significantly reduced co-localization of NRP1-ΔE4/Δ7 with the early endosomal marker EEA1, the late endosomal marker Rab7, and the recycling endosomal marker Rab11 compared to NRP1-ΔE4 in both HT29 and HCT116 cells (Fig. 3E-G and Figs. S3D-F). In contrast, NRP1-ΔE4/Δ7 exhibited notably high co-localization with the lysosomal marker Lysotracker compared to NRP1-ΔE4 (Fig. 3H and I and Fig. S3G). Additionally, cycloheximide chase analysis demonstrated rapid degradation of NRP1-WT and substantial degradation of NRP1-ΔE4/Δ7 between 0 and 8 h (Fig. 3J and K). In contrast, NRP1-ΔE4 showed no degradation up to 8 h after cycloheximide treatment (Fig. 3J and K).

The intracellular cargo in late endosomes can either be directed to lysosomes for degradation or released as exosomes [27]. Given that NRP1-ΔE4/Δ7 primarily translocated to lysosomes and less to late endosomes compared to NRP1-ΔE4 (Fig. 3F-I), we reasoned that a defect in GAG modification might reduce the exosomal release of NRP1-ΔE4. Indeed, we found that NRP1-ΔE4/Δ7 exhibited a reduced interaction with the exosome transporter Hrs compared to NRP1-ΔE4 (Fig. 3L and M). Furthermore, NRP1-ΔE4/Δ7 displayed decreased co-localization with the exosome markers CD63, CD81, and Alix compared to NRP1-ΔE4 (Fig. 3N and O). Similar to NRP1-WT, NRP1-ΔE4/Δ7 was secreted via exosomes at much lower levels than NRP1-ΔE4, while the overall exosome secretion levels remained comparable among cells expressing the three isoforms (Fig. 3P and Q and Fig. S3H).

The impact of GAG modification on NRP1-ΔE4 trafficking, observed through the deletion of amino acids 615-621 (Δ7), was further corroborated by examining the Ser612 O-linked glycosylation-defective NRP1-ΔE4 mutant, NRP1-ΔE4/S612A. Similar to NRP1-ΔE4/Δ7, NRP1-ΔE4/S612A exhibited decreased internalization into endosomes but increased translocation to lysosomes, leading to its degradation compared to NRP1-ΔE4 (Fig. 4A-E and Figs. S4A and B). Moreover, NRP1-ΔE4/S612A showed reduced internalization compared to NRP1-ΔE4 upon HGF stimulation, while both NRP1-ΔE4 and NRP1-ΔE4/S612A displayed minimal internalization upon VEGF₁₆₅ stimulation under FBS-free culture conditions (Fig. S4C). Additionally, NRP1-ΔE4/S612A demonstrated reduced interaction with Hrs, as well as decreased co-localization with Hrs and the exosome markers CD63 and CD81 (Fig. 4F-I and Figs.S4D and E). This resulted in a marked reduction in its exosomal levels compared to NRP1-ΔE4 (Fig. 4J).

Fig. 4.

Loss of Ser612-linked GAG modification reduces endosomal trafficking and exosomal secretion of NRP1-ΔE4. (A) Representative confocal images of HCT116 cells expressing NRP1-ΔE4 or NRP1-ΔE4/S612A, immunostained for α6-integrin (green), NRP1 (red), and DAPI (blue). Scale bar, 10 μm. (B-E) HCT116 cells expressing NRP1-ΔE4 or NRP1-ΔE4/S612A were assessed by confocal microscopy (B and D) for immunofluorescent staining of NRP1 (green), endosome markers (red), lysotracker (red), and DAPI (blue), followed by quantification of co-localization between the indicated NRP1 mutants and EEA1, Rab7, or lysotracker using ImageJ software (C and E). Scale bars, 10 μm. (F) HCT116 cells expressing NRP1-ΔE4 or NRP1-ΔE4/S612A were lysed and immunoprecipitated with NRP1 antibodies or IgG control, followed by immunoblot analysis of the indicated proteins. (G) The relative levels of Hrs binding to the indicated NRP1 mutants shown in (F) were quantified and normalized using ImageJ software. (H and I) HCT116 cells expressing NRP1-ΔE4 or NRP1-ΔE4/S612A were assessed by confocal microscopy (H) for immunofluorescent staining of NRP1 (green), CD63 or CD81 (red), and DAPI (blue), followed by quantification of co-localization between the indicated NRP1 mutants and CD63 or CD81 using ImageJ software (I). Scale bars, 10 μm. (J) Immunoblot analysis was performed for the indicated proteins in whole cell lysates (WCL) and exosomes from NRP1-ΔE4- or NRP1-ΔE4/S612A-expressing HCT116 cells. All graphical data are presented as mean ± SEM (n = 3 independent experiments). *p < 0.05; **p < 0.01; ***p < 0.001, using the Student’s t test.

Taken together, our findings indicate that the loss of GAG modification at the O-glycosylation site Ser612 disrupts the endosomal trafficking and exosomal release of NRP1-ΔE4. These results underscore the functional importance of GAG modification in regulating both the endosomal and exosomal trafficking of NRP1-ΔE4.

3.4. GAG modification is essential for NRP1-ΔE4 to promote CRC metastasis

Since our recent study identified NRP1-ΔE4 as an aggressive splice variant of NRP1 that promotes metastasis [16], we next investigated whether disrupting GAG modification impacts NRP1-ΔE4-driven metastasis. Analysis of cell growth rates revealed no significant differences among HCT116 cells with stable expression of NRP1-WT, ΔE4, or ΔE4/Δ7 (Fig. S5A). In contrast, cells expressing NRP1-ΔE4/Δ7 exhibited a marked reduction in both migratory and invasive capabilities compared to NRP1-ΔE4, with levels even lower than those observed in NRP1-WT-expressing cells (Fig. 5A-C and Figs. S5B and C). Similarly, the non-GAG-modified NRP1-ΔE4/S612A exhibited profound suppression of migration and invasion in HCT116 cells compared to NRP1-ΔE4 (Fig. 5D-F). Additionally, both HCT116 and HT29 cells treated with exosomes derived from HCT116 cells expressing either NRP1-ΔE4/Δ7 or NRP1-ΔE4/S612A showed a marked decrease in migration and invasion compared to cells treated with exosomes from NRP1-ΔE4-expressing HCT116 cells (Fig. 5G-L and Figs. S5D and E). To assess the role of GAG modification in NRP1-ΔE4-driven metastasis in vivo, we utilized an experimental lung metastasis model and an orthotopic CRC liver metastasis model. Luciferase-labeled HCT116 cells stably expressing NRP1-ΔE4 or NRP1-ΔE4/Δ7 were injected intravenously or into the cecal submucosa of athymic nude mice, and lung or liver metastasis formation was evaluated through bioluminescent imaging. Compared to NRP1-ΔE4, expression of NRP1-ΔE4/Δ7 remarkably reduced lung and liver metastases in mice (Fig. 5M and N and Figs. S5F and G). Thus, these results reveal GAG modification as a crucial regulatory mechanism in NRP1-ΔE4-driven CRC cell dissemination.

Fig. 5.

GAG modification is crucial for NRP1-ΔE4-driven CRC metastasis. (A-F) Transwell migration and invasion assays were performed on HCT116 cells expressing the indicated NRP1 isoforms or mutants for 12 h and 36 h, respectively. Scale bars, 100 μm. The number of migrated and invaded cells per field (n =5) was counted. Results are expressed as fold changes in migrated (B and E) or invaded (C and F) cells relative to NRP1-WT cells. (G-L) Transwell migration and invasion assays were performed on HCT116 cells treated with exosomes derived from HCT116 cells expressing the indicated NRP1 isoforms or mutants for 24 h and 48 h, respectively. Scale bars, 100 μm. Results are expressed as fold changes in migrated (H) or invaded (I) cells relative to those treated with exosomes from NRP1-WT cells, or as a percentage of migration (K) or invasion (L) relative to cells treated with exosomes from NRP1-ΔE4 cells. (M) Bioluminescence imaging of liver metastases in athymic nude mice injected with HCT116-luciferase/GFP cells expressing either NRP1-ΔE4 or NRP1-ΔE4/Δ7 into the cecal submucosa at week 6 post-injection. (N) Quantification of bioluminescence signal in liver metastases from (M). Dara are presented as mean ± SEM (n = 6 mice/group). **p < 0.01, using the Mann–Whitney test. Graphical data in (B, C, E, F, H, I, K, and L) are presented as mean ± SEM (n = 3 independent experiments). **p < 0.01; ***p < 0.001, using the Student’s t test.

3.5. GAG-modified NRP1-ΔE4 regulates co-internalization of Met and β1-integrin into endosomes

Building upon our previous work, which demonstrated that NRP1-ΔE4 requires both Met and β1-integrin receptors as partners to form a trimeric complex in endosomes to promote CRC metastasis [16], we examined whether GAG modification of NRP1-ΔE4 also modulates its interaction with Met and β1-integrin, thereby influencing their endosomal trafficking. Co-immunoprecipitation analysis showed that, compared to NRP1-WT, NRP1-ΔE4 exhibited greater interaction with both Met and β1-integrin (Fig. 6A and B). However, GAG-deficient NRP1-ΔE4 mutants, NRP1-ΔE4/Δ7 and NRP1-ΔE4/S612A, displayed significantly reduced interactions with β1-integrin, while their interactions with Met remained unchanged (Fig. 6A-D). Consistent with our previous report [16], Met and β1-integrin were co-internalized and co-localized with NRP1-ΔE4 in endosomes in both HCT116 and HT29 cells, in contrast to their co-localization with NRP1-WT on the plasma membrane (Fig. 6E-I and Figs. S6A and B). Notably, with the reduced internalization of NRP1-ΔE4/Δ7 and NRP1-ΔE4/S612A, both Met and β1-integrin also showed decreased internalization and largely remained on the plasma membrane with NRP1-ΔE4/Δ7 or NRP1-ΔE4/S612A (Fig. 6E and J and Fig. S6A). Moreover, β1-integrin significantly reduced its co-localization with NRP1-ΔE4/Δ7 or NRP1-ΔE4/S612A (Fig. 6F and K and Fig. S6B), and both Met and β1-integrin dissociated from the late endosomes (Fig. 6G-I, L and M). Likewise, in recipient CRC cells, both Met and β1-integrin primarily co-internalized and co-localized with exosomal NRP1-ΔE4 in endosomes, while they largely co-localized with exosomal NRP1-WT or NRP1-ΔE4/Δ7 on the plasma membrane (Figs. S6C-F). Collectively, these findings demonstrate that GAG modification enhances the interaction of NRP1-ΔE4 with β1-integrin and facilitates the co-internalization of Met and β1-integrin with NRP1-ΔE4 into endosomes.

Fig. 6.

Impaired GAG modification of NRP1-ΔE4 disrupts the co-internalization of Met and β1-integrin into endosomes. (A-D) HCT116 cells expressing the indicated NRP1 isoforms or mutants were lysed and immunoprecipitated with NRP1 antibodies or IgG control, followed by immunoblot analysis of the indicated proteins (A and C). The relative levels of Met or β1-integrin were normalized to their binding to the indicated NRP1 isoforms or mutants (B and D). (E and F) HCT116 cells expressing the indicated NRP1 isoforms were analyzed by confocal microscopy for immunofluorescent staining of NRP1 (green), Met (red), β1-integrin (purple), and DAPI (blue), followed by quantification of co-localization between the indicated NRP1 isoforms and Met or β1-integrin using ImageJ software. Scale bar, 10 μm. (G-I) HCT116 cells expressing NRP1-ΔE4 or NRP1-ΔE4/Δ7 were analyzed by confocal microscopy (G and H) for immunofluorescent staining of NRP1 (green), Met or β1-integrin (red), Rab7 (purple), and DAPI (blue), followed by quantification of Rab7 co-localization with Met or β1-integrin using ImageJ software (I). Scale bars, 10 μm. (J and K) HCT116 cells expressing NRP1-ΔE4 or NRP1-ΔE4/S612A were analyzed by confocal microscopy for immunofluorescent staining of NRP1 (green), Met (red), β1-integrin (purple), and DAPI (blue), followed by quantification of co-localization between the indicated NRP1 isoforms and Met or β1-integrin using ImageJ software. Scale bar, 10 μm. (L and M) HCT116 cells expressing NRP1-ΔE4 or NRP1-ΔE4/S612A were analyzed by confocal microscopy (L) for immunofluorescent staining of NRP1 (green), Met or β1-integrin (red), Rab7 (purple), and DAPI (blue), followed by quantification of Rab7 co-localization with Met or β1-integrin using ImageJ software (M). Scale bar, 10 μm. All graphic data are presented as mean ± SEM (n = 3 independent experiments). *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant, using the Student’s t test.

3.6. Impaired GAG modification of NRP1-ΔE4 inhibits FAK signaling in endosomes

We previously demonstrated that the formation of the NRP1-ΔE4/Met/β1-integrin trimeric complex on endosomes provides sustained signaling to activate the FAK/p130Cas pathway, thereby promoting CRC cell migration, invasion, and metastasis [16]. Given that non-GAG-modified NRP1-ΔE4 significantly decreases its association with Met and β1-integrin on endosomes (Fig. 6), we hypothesize that GAG modification of NRP1-ΔE4 is essential for activating FAK signaling in endosomes. Immunoblot analysis revealed that HCT116 cells expressing NRP1-ΔE4/Δ7 or NRP1-ΔE4/S612A showed a marked decrease in phosphorylation levels of FAK at Y397 (active form) and its interacting protein p130Cas at Y249, compared to the elevated levels seen with NRP1-ΔE4 (Fig. 7A and B). These effects were not due to a decrease in total levels of FAK and p130Cas in cells expressing NRP1-ΔE4/Δ7 or NRP1-ΔE4/S612A (Fig. 7A and B). By contrast, total protein levels and phosphorylation of AKT and ERK were similar among cells expressing NRP1-WT, NRP1-ΔE4, NRP1-ΔE4/Δ7, NRP1-ΔE4/S612A, or vector control (Fig. 7A and B). Immunofluorescence staining further showed a significant dissociation of phosphorylated FAK from NRP1-ΔE4 in endosomes upon disrupting GAG modification by expressing NRP1-ΔE4/Δ7 or ΔE4/S612A in HCT116 cells (Fig. 7C-F). Similar findings were observed in recipient HCT116 or HT29 cells treated with exosomes carrying NRP1-ΔE4/Δ7, compared to those with NRP1-ΔE4 or NRP1-WT (Fig. 7G-I and Fig. S7A). Additionally, blocking exosome secretion by silencing Hrs or Rab27 profoundly inhibited FAK and p130Cas phosphorylation in NRP1-ΔE4 expressing cells (Figs. S7B and C). Together, these findings strongly support that the endosomal activation of the FAK/p130Cas signaling pathway in both donor and recipient CRC cells depends on GAG modification and exosomal release of NRP1-ΔE4 (Fig. 7J).

Fig. 7.

GAG modification of NRP1-ΔE4 is essential for the endosomal activation of FAK signaling. (A and B) HCT116 cells expressing the indicated NRP1 isoforms, mutants, or vector control were analyzed by immunoblot for the indicated proteins. (C-F) HCT116 cells expressing NRP1-ΔE4 or its mutants were analyzed by confocal microscopy (C and E) for immunofluorescent staining of NRP1 (green), p-FAK (red), Rab7 (purple), and DAPI (blue), followed by quantification of co-localization between p-FAK and the indicated NRP1 mutants or Rab7 (D and F) using ImageJ software. Scale bars, 10 μm. (G) HCT116 cells were treated with exosomes isolated from HCT116 cells expressing NRP1-WT, NRP1-ΔE4, or NRP1-ΔE4/Δ7 for 24 h, followed by immunoblot analysis of the indicated proteins. (H and I) HT29 cells were treated with exosomes isolated from HCT116 cells expressing NRP1-WT, NRP1-ΔE4, or NRP1-ΔE4/Δ7 for 24 h, followed by immunofluorescent staining (H) for NRP1 (green), p-FAK (red), Rab7 (purple), and DAPI (blue). Co-localization between p-FAK and NRP1-WT, NRP1-ΔE4, NRP1-ΔE4/Δ7, or Rab7 was quantified using ImageJ software (I). Scale bar, 10 μm. (J) A schematic model illustrating the role of GAG modification in regulating the intracellular and intercellular trafficking of NRP1-ΔE4 to promote CRC metastasis. All graphical data are presented as mean ± SEM (n = 3 independent experiments). *p < 0.05; **p < 0.01; ***p < 0.001, using the Student’s t test.

4. Discussion

Alternative splicing, a process that produces different mRNA isoforms from a single gene, is frequently disrupted in cancer, generating multiple functionally altered protein isoforms that impact tumor progression and metastasis. While NRP1 plays an important role in angiogenesis and cancer progression, several previously known NRP1 splice variants exhibit anti-angiogenic and anti-tumorigenic effects or show no functional difference compared to NRP1-WT. Our recent work identified two additional NRP1 splice variants, establishing exon 4-deleted NRP1 (NRP1-ΔE4) as a crucial oncogenic isoform that aggressively promotes CRC metastasis and significantly correlates with CRC progression [16]. Compared to NRP1-WT, NRP1-ΔE4, which lacks N150-linked glycosylation, is more sensitive to HGF-induced endocytosis, leading to its accumulation in endosomes. Furthermore, NRP1-ΔE4 enhances interactions with Met and β1-integrin receptors, promoting their co-internalization and co-accumulation as a trimeric complex in endosomes. This complex provides sustained endosomal signals to activate FAK, thereby promoting CRC cell migration, invasion, and metastasis. In this study, we reveal exosomal release and GAG modification of NRP1-ΔE4 as additional critical regulatory processes that sustain its oncogenic activity and drive CRC metastatic progression.

Ligand-induced endocytosis of signaling receptors has traditionally been viewed as a process that attenuates signaling strength from the cell surface. This involves rapid internalization and transport to late endosomes, followed by fusion with lysosomes for degradation [6]. However, mounting evidence suggests that endosomal vesicles themselves can serve as vital platforms for receptors, such as EGFR and Met receptor tyrosine kinases (RTKs) as well as β1-integrin, enabling them to remain active and deliver spatially and temporally restricted signals that contribute to pathway-specific tumor progression [7, 8, 39-42]. Interestingly, we have shown that upon HGF stimulation, NRP1-ΔE4 promotes the co-internalization and accumulation of Met and β1-integrin in endosomes, while avoiding lysosomal degradation [16]. Here, we further discovered that internalized NRP1-ΔE4 can be recognized by the ESCRT subunit Hrs and subsequently released in exosomes under the mediation of Rab27a. This exosomal NRP1-ΔE4 can reinternalize with Met and β1-integrin receptors and co-accumulate in endosomes, maintaining persistent endosomal activation of FAK signaling to enhance metastatic potential. These findings suggest that the constitutive trafficking of NRP1-ΔE4 through the endosomal-exosomal pathway [43] effectively recycles NRP1-ΔE4, sustaining robust endosomal FAK signaling that promotes CRC metastasis. Additionally, our data indicate that exosomal NRP1-ΔE4 can be taken up by recipient CRC cells, including those expressing high levels of NRP1-WT, significantly enhancing their migration and invasion. Inhibition of exosome secretion pathways—whether by genetic or pharmacological means—or neutralization of exosomal NRP1-ΔE4 with NRP1-targeting antibodies markedly suppresses the metastatic potential of CRC cells expressing NRP1-ΔE4. This strongly indicates that the pro-metastatic effects of NRP1-ΔE4 are largely driven by its presence in exosomes. Intratumor genomic diversity has long been recognized as a contributor to cancer’s lethality, therapeutic failure, and drug resistance [44-46]. Emerging evidence underscores the importance of viewing cancer as an ecosystem of subclones that may cooperate in driving metastases [46]. It is therefore likely that exosomal NRP1-ΔE4 not only sustains the aggressive metastatic phenotype of CRC subclones expressing NRP1-ΔE4 but may also modify other CRC subclones or various types of cells, such as fibroblasts, endothelial cells, and immune cells, creating a supportive microenvironment for CRC metastasis and outgrowth.

Glycosylation is an important post-translational modification for both intracellular and secreted proteins, playing critical roles in maintaining protein integrity and directing proper intracellular trafficking [47]. Aberrant glycosylation is commonly observed in cancer and is associated with disrupted cancer signaling, tumor progression, metastasis, and poor clinical prognosis [48]. Structurally, NRP1-ΔE4 lacks part of the NRP1 a2 domain (amino acids 144-219), resulting in a defect in N-linked glycosylation at site N150 [16]. Our previous work demonstrated that this partial deletion, or impaired N150-linked glycosylation, makes NRP1-ΔE4 less responsive to VEGF₁₆₅-stimulated endocytosis but more susceptible to HGF-stimulated endocytosis. This alteration enhances NRP1-ΔE4’s recycling and stability, even in cells with depleted Met or β1-integrin under basal conditions [16]. Furthermore, we showed that N150-linked glycosylation-deficient NRP1-ΔE4 strengthens its interactions with Met and β1-integrin, relying on these partners within endosomes to activate FAK signaling and promote CRC metastasis [16]. In this study, we found that NRP1-ΔE4 also binds more effectively to the exosome transporter Hrs, facilitating its exosomal release and increasing its metastatic potential. Notably, removing chondroitin sulfate GAG modification at the O-glycosylation site S612, either by deleting 7 amino acids (positions 615-521) in the NRP1 b-c linker domain or using an S612A mutation, significantly reduced HGF-stimulated endocytosis, stability, and exosomal release of NRP1-ΔE4 due to decreased interaction with Hrs. Although the interaction between NRP1-ΔE4 and Met remained unaffected in the absence of O-linked GAG modification, O-glycosylation-deficient NRP1-ΔE4 mutants exhibited a marked reduction in binding to β1-integrin. These findings highlight that while a defect in N150-linked glycosylation sensitizes NRP1-ΔE4 to HGF stimulation over VEGF₁₆₅ for endocytic trafficking, O-linked GAG modification at S612 is also crucial for the endocytic trafficking, stability, and exosomal release of NRP1-ΔE4. This modification enables endosomal FAK activation, promoting CRC cell dissemination through enhanced interactions with Hrs and β1-integrin, and potentially strengthening binding to HGF. However, our prior study showed that, similar to NRP1-WT, the NRP1 S612A mutant lacking O-linked GAG modification remains localizes on the plasma membrane and is less responsive to HGF-stimulated endocytosis than the N-linked glycosylation-deficient mutant NRP1 N150Q or NRP1-ΔE4 [16]. Thus, impaired N150-linked glycosylation appears to coordinate with O-linked GAG modification at S612 to regulate the endosomal and exosomal trafficking of NRP1-ΔE4, thereby activating endosomal FAK signaling and facilitating CRC metastatic spread. This coordination likely induces a conformational change in NRP1-ΔE4, enhancing its interactions with Met, β1-integrin, Hrs, HGF, and possibly other unknown partners. Further detailed analysis is needed to fully elucidate the complexities of these interactions and their functional significance in the oncogenic activities driven by NRP1-ΔE4.

In summary, our study provides new insights into the role of GAG modification in NRP1-ΔE4-driven CRC metastatic progression. Our findings highlight that impaired N-linked glycosylation and O-linked GAG modification are critical co-regulatory processes contributing to the aggressive oncogenic properties of NRP1-ΔE4. Strategies aimed at blocking the endosomal and exosomal trafficking of NRP1-ΔE4, or developing drugs that target GAG modification, may open new therapeutic avenues for treating or preventing NRP1-ΔE4-driven CRC metastasis.

Highlights.

• NRP1-ΔE4 undergoes constitutive trafficking via the endosomal-exosomal pathway.

• Exosomal NRP1-ΔE4 enhances both donor and recipient CRC cell migration and invasion.

• Glycosaminoglycan modification governs endosomal-exosomal trafficking of NRP1-ΔE4.

• Glycosaminoglycan modification is essential for NRP1-ΔE4 to promote CRC metastasis.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.