Functional Importance of a Proteoglycan Coreceptor in Pathologic LymphangiogenesisNovelty and Significance
Rationale: Lymphatic vessel growth is mediated by major prolymphangiogenic factors, such as vascular endothelial growth factor (VEGF-C) and VEGF-D, among other endothelial effectors. Heparan sulfate is a linear polysaccharide expressed on proteoglycan core proteins on cell membranes and matrix, playing roles in angiogenesis, although little is known about any function(s) in lymphatic remodeling in vivo.
Objective: To explore the genetic basis and mechanisms, whereby heparan sulfate proteoglycans mediate pathological lymphatic remodeling.
Methods and Results: Lymphatic endothelial deficiency in the major heparan sulfate biosynthetic enzyme N-deacetylase/N-sulfotransferase-1 (Ndst1; involved in glycan-chain sulfation) was associated with reduced lymphangiogenesis in pathological models, including spontaneous neoplasia. Mouse mutants demonstrated tumor-associated lymphatic vessels with apoptotic nuclei. Mutant lymphatic endothelia demonstrated impaired mitogen (Erk) and survival (Akt) pathway signaling and reduced VEGF-C–mediated protection from starvation-induced apoptosis. Lymphatic endothelial-specific Ndst1 deficiency (in Ndst1f/fProx1+/CreERT2 mice) was sufficient to inhibit VEGF-C–dependent lymphangiogenesis. Lymphatic heparan sulfate deficiency reduced phosphorylation of the major lymphatic growth receptor VEGF receptor-3 in response to multiple VEGF-C species. Syndecan-4 was the dominantly expressed heparan sulfate proteoglycan in mouse lymphatic endothelia, and pathological lymphangiogenesis was impaired in Sdc4(−/−) mice. On the lymphatic cell surface, VEGF-C induced robust association between syndecan-4 and VEGF receptor-3, which was sensitive to glycan disruption. Moreover, VEGF receptor-3 mitogen and survival signaling was reduced in the setting of Ndst1 or Sdc4 deficiency.
Conclusions: These findings demonstrate the genetic importance of heparan sulfate and the major lymphatic proteoglycan syndecan-4 in pathological lymphatic remodeling. This may introduce novel future strategies to alter pathological lymphatic-vascular remodeling.
Growth and remodeling of lymphatic vasculature in the tumor microenvironment may be supported by a variety of growth factors that stimulate cognate receptors on host lymphatic vessels.1,2 The process has been shown to contribute to lymph node metastasis.3–5 It is also known that a dominant prolymphangiogenic growth factor, vascular endothelial growth factor (VEGF-C), is frequently overexpressed in primary carcinomas. Along with a unique role in driving developmental lymphangiogenesis,2,6 VEGF-C also plays critical roles in tumor lymphangiogenesis along with the stimulatory actions of other tumor vascular growth effectors, such as VEGF-A and fibroblast growth factor-2, as well as cytokines.6,7
Heparan sulfate is a linear glycan polymer expressed on a variety of proteoglycans, which plays important roles in endothelial growth factor binding in unique pathological contexts, including tumor angiogenesis.8–10 Heparan sulfate proteoglycans (HSPGs) secreted into tumor matrix may release growth factors on the action of tumor heparinase, mobilizing banks of proangiogenic factors bound to sulfated domains on heparan sulfate in tumor matrix.11,12 Endothelial-surface proteoglycans may act in cis in a cell-autonomous manner or in trans to promote endothelial proliferation in response to growth factors.10,13 Although less is known with respect to lymphatic biology, preliminary work points to a role for heparan sulfate in VEGF-C–dependent proliferation of lymphatic endothelial cells (LECs) in culture.14 However, the genetic importance, mechanisms, and proteins involved in vivo remain poorly understood.
We generated lymphangiogenesis models in mice bearing a lymphatic deficiency in the heparan sulfate biosynthetic enzyme N-deacetylase/N-sulfotransferase-1 (Ndst1), involved in initiating sulfate modifications of nascent heparan sulfate chains.8 We demonstrate that lymphangiogenesis is inhibited in models of oil granuloma–induced lymphangiogenesis, wound inflammation, and carcinomas on the Ndst1 mutant background, including VEGF-C–dependent lymphangiogenesis on a stringent lymphatic-specific Ndst1-deficient background. The mutation is associated with defects in lymphatic mitogen and survival signaling, and VEGF receptor-3 (VEGFR-3) phosphorylation in response to VEGF-C. Using proteoglycan expression analyses, gene-targeted mice and primary-cell mechanistic analyses, we further highlight syndecan-4 (Sdc4) as a major HSPG coreceptor required for VEGF-C mitogen and survival signaling, which complexes with VEGFR-3 in a glycan-dependent manner on VEGF-C exposure.
Cells and Cell Lines
Primary LECs were isolated from mouse mesenteric oil granuloma/lymphangiomas, as previously described,15 and tested for LYVE-1/podoplanin expression.14 For some studies, LECs were purified from lungs of Ndst1f/fTekCre+ mutants.16 Primary human lung LEC (hLEC; Lonza; previously shown >99% pure at third passage by Prox1 staining) were also used. For Lewis lung carcinoma (LLC) models, retroviruses expressing pLTR-mVEGFC-GFP were used to transduce LLCs (kindly provided by G. Thurston; Regeneron) with full-length VEGF-C (LLC-VC cells),17 or GFP-expressing empty vector (LLC-ev) cells as controls. The mouse transformed mesenteric LEC line (svLEC)18 was kindly obtained from Dr Alexander (LSU Health Sciences Center).
Mice and Pathological Lymphatic Proliferation Models
Details on mouse models targeting lymphangiogenesis in Ndst1 and Sdc4 mutants, including relevant references, are presented in the expanded Online Data Supplement.
Pathological Tissue Processing and Analysis
Tumor/tissue specimens were formalin-fixed, paraffin-embedded, and hematoxylin and eosin stained, with immunostaining details outlined in the Online Data Supplement.
Lung digests were filtered through 100-μm strainers (Fisher), subjected to red-cell lysis (eBioscience), and stained with PE-labeled antimouse podoplanin (eBioscience) and APC-labeled antimouse LYVE-1 (R&D), with Aqua (Biolegend) viability marker for dead-cell exclusion. Dual PE/APC+ live cells were analyzed by an LSRII (BD) cytometer. The quantity of dual-positive cells as a percentage of total cells was analyzed/plotted and used in statistical analyses comparing lungs from mutant versus control mice.
Quantitative Polymerase Chain Reaction Analyses
RNA was isolated from primary LECs, reverse transcribed (Superscript III, Invitrogen), amplified using gene-specific primers to each core protein, and quantified (triplicate assays) using the 2−ΔΔCt method relative to β-actin. Primers included those for mouse HSPGs (Online Table I). For Ndst, the same method was used with primers for mouse Ndst1–Ndst4 isoenzymes.14
Primary hLEC at near confluence were transfected with siRNA targeting heparan sulfate biosynthetic enzymes xylosyltransferase 2 (XylT2; siXylT2) or Ndst1 (siNdst1), the HSPG core protein Sdc-4 (siSdc4), or receptors VEGFR-2 (siVEGFR-2) or VEGFR-3 (siVEGFR-3); with scrambled-duplex RNA mock transfectants (siDS) as controls. Transfections (20-nmol/L siRNA) were carried out using Lipofectamine (Invitrogen) after manufacturer recommendations. Transfection complex was added in Opti-Mem (Gibco), and incubated for 6 hours, with cell recovery overnight in normal growth medium.
Human recombinant mature VEGF-C was purchased (R&D). Untagged pro–VEGF-C was expressed from full-length cDNA using a chinese hamster ovary dhfr gene-amplification system.19 This was predominantly a mixture of unprocessed and partially processed propeptide forms of VEGF-C. Highly expressed clones were identified by the ability of the culture supernatant to sustain growth of VEGFR-3/EpoR Ba/F3 cells.20 The affinity of pro–VEGF-C for heparin was used in the capture step from serum-free culture supernatant (salt elution from heparin-sepharose column at 0.48 M NaCl, with minor peak at 0.53 M). Cation exchange chromatography (pH 6.6) and gel filtration were used to increase prep homogeneity. Identity was confirmed by Western blotting. A short form of VEGF-C consisting of minimum-binding domain residues A112-L215 was prepared as a strep-II–tagged protein in drosophila S2 cells and purified using streptactin resin. It did not bind to heparin. A mutant form of human VEGF-C (VEGF-CCys156Ser R&D) that binds exclusively to VEGFR-3 was used in some studies.
Detailed methods for Western blotting of lysates after VEGF-C stimulation of LECs are reported in the Online Data Supplement.
Receptor Tyrosine Kinase Phospho Arrays
Serum-starved hLEC were stimulated ± pro–VEGF-C (1 μg/mL) for 15 minutes. Lysates from 6-well plates were collected in 500-μL Lysis Buffer (R&D Phospho-RTK Array–assay instructions) and cleared with supernatant used in receptor tyrosine kinase-assay after manufacturer protocol. After blocking, diluted lysates were incubated with slide arrays (4°C overnight) and washed, and antiphosphotyrosine horseradish peroxidase–tagged antibody was added (2 hours at room temperature), followed by wash, chemiluminescent development, and digital-imaging densitometry. Further array details, including incorporation of mitogen-activated protein kinase array after VEGF-C stimulation, are described in the Online Data Supplement.
VEGFR-3 Phosphorylation Assays
Serum-starved cells were treated with VEGF-C species (1 µg/mL) and assayed using a human phospho-VEGFR3 Elisa Kit (R&D). Treated cells were lysed (R&D lysis buffer; 30 minutes, 4°C), spun-down, diluted, added to a precoated anti–VEGFR-3 plate overnight (4°C), followed by antiphosphotyrosine horseradish peroxidase (included in Kit). Biotin anti–VEGFR-3 (Reliatech) labeled with streptavidin-horseradish peroxidase (Vector) was added to detect total VEGFR3. After incubating in substrate solution (R&D), reactions were stopped with 2N sulfuric acid. Plates were assayed at absorbance A450 nm, with values corrected against total VEGFR-3.
Proximity Ligation Assays
Chamber slides (Lab-Tek) coated with 50-μg/mL PurCol (Advanced Biomatrix), layered with serum-starved hLEC were stimulated with human mature VEGF-C (R&D) for 5 minutes ± pre treatment with heparinase in some experiments. Cells were fixed in ice-cold methanol for 10 minutes and blocked (Olink PLA-blocking reagent; 30 minutes at 37°C). Rabbit antihuman VEGFR-3 (Reliatech; or for some studies, anti–VEGFR-2; Cell Signaling) and goat antihuman Sdc-4 (R&D) were then added (2 µg/mL) together (in blocking solution overnight; 4°C). After wash, Rabbit(−) and Goat(+) PLA Probes (Duolink assay; Olink Bioscience) were added (1:10 dilution). In other experiments, rabbit anti–VEGFR-3 was paired with either mouse antihuman Sdc-1 (Abcam) or mouse antihuman Sdc-2 (kind gift from G. David) antibodies. Antibodies were directed against extracellular domains, as per manufacturer protocol (imaging: 40× objective, room temperature).
Mean values (±SD) were obtained for lymphatic vessel density (LVD) or apoptotic body index for each genotype. For some analyses, means were compared using Student t test, with normalization to wild-type (or control) baseline values. Paired t tests were applied for comparing means of paired values (eg, for multiple experiments examining western phosphorylation responses pre- versus post–VEGF-C species in siRNA versus control-transfected cells; comparisons of change in caspase signal in response to VEGF-C in siRNA versus control-transfected cells). For some experiments in which binary-type response data were examined (eg, terminal deoxynucleotidyl transferase dUTP nick-end labeling [TUNEL] positivity versus negativity of lymphatic vessels or the presence versus the absence of LYVE-1/podoplanin dual-positive cells in flow cytometry analyses of tumor LECs) the Wald χ2 statistic was used. Two-way ANOVA was used in the analyses for experiments in which the significance of any interaction between genotype and biological response to growth factor (eg, VEGF-C–dependent lymphangiogenesis) was examined.
A 2-way ANOVA was also applied to assess for phospho–VEGFR-3 responses to VEGF-C as they depend on siRNA status and lymphatic endothelial PLA responses to VEGF-C as they depend on treatment ± heparinase. SPSS version 19, general linear model function, was used to compute these ANOVAs. A P value of ≤0.05 was considered significant for all analyses.
Endothelial Mutation Resulting in Lymphatic Ndst1 Deficiency Is Associated With Reduced Pathological Lymphangiogenesis and Altered Lymphatic Signaling
We assessed lymphangiogenesis in models of granulomatous inflammation, wound inflammation, and tumor lymphatic remodeling on the Ndst1f/fTekCre+ mutant background. We first used an established model of oil granuloma/lymphangioma induction in the mouse abdomen,15 wherein plaque-like lesions develop intense proliferation of LYVE-1+ lymphatic endothelium expressing VEGFR-3/Flt-4.15,21 In this model, although the Ndst1 mutation exists in all endothelia, lesion-associated LYVE-1+ vessel density in Ndst1f/fTekCre+ mutants was reduced (Figure 1A), suggesting that lymphatic Ndst1 deficiency affected pathological lymphangiogenesis. Wound lymphangiogenesis associated with early skin-wound remodeling was also reduced on the mutant background (Figure 1B). To examine tumor lymphangiogenesis, we crossed mutants with an MMTV-PyMT spontaneous mammary tumor strain. Tumors demonstrated reduced LVD in Ndst1f/fTekCre+ mutants (Figure 1C, top left). Tumor expression of VEGF-C was confirmed (Figure 1C, top right). TekCre transgene expression is pan-endothelial, and studies quantifying blood-vascular angiogenesis in the mutants (not shown) revealed a 67% reduction in tumor blood-vascular density relative to wild-type by CD105 staining (P<0.001) and 50% reduction by CD31 (P<0.01). The unique effect of mutation on LYVE-1+ vessel-density coupled with marked tumor VEGF-C production prompted us to further explore the effect of altered lymphatic heparan sulfate on VEGF-C–mediated lymphangiogenesis. As a developmental baseline, LVD in the ear bud of newborn mice, which is uniquely VEGF-C dependent,2 was modestly reduced in Ndst1f/fTekCre+ mutants (Online Figure IA). Although the LVD reduction was significant, this did not result in obvious lymphatic developmental defects, such as limb edema or chylous ascites. Knockdown of Ndst1 in primary lung LECs from nonchallenged mutants was confirmed by quantitative polymerase chain reaction (Online Figure IB).
Lymphatic vessels in tumors from MMTV-PyMT Ndst1f/fTekCre+ mutants were characterized by a greater percentage of lymphatic-associated TUNEL+ apoptotic bodies (Figure 1C, bottom). In separate primary human cell–based studies, we questioned whether altered heparan sulfate biosynthesis might impair VEGF-C–mediated protection of primary hLECs from apoptotic stress, as measured by cellular–cleaved caspase levels. In pilot studies, VEGF-C consistently lowered the generation of starvation-induced cleaved caspase by hLEC under several media conditions (Online Figure II). XylT2-deficient LECs (siXylT2; characterized by impaired glycan-chain initiation) were insensitive to VEGF-C during starvation (Figure 1D, right bar), whereas control-transfected hLEC consistently showed reduced apoptosis on starvation in the presence of VEGF-C (Figure 1D, siDS transfection control, left bar). Consistent with this, VEGF-C–dependent Akt phosphorylation in siXylT2-targeted hLEC was reduced relative to control cells (Figure 1E). When stimulated with a VEGF-C ligand that binds exclusively to VEGFR-3 (VEGF-CCys156Ser),22 Akt phosphorylation in siXylT2-transfected cells was also blunted (Figure 1E, inset graph). We next examined how siNdst1 targeting might affect Akt phosphorylation in response to VEGF-CCys156Ser: similar results were found (Figure 1F), implying that VEGFR-3–specific Akt signaling is sensitive to altered glycan sulfation. Moreover, mitogen-activated pathway signaling (phospho-Erk1/2) in response to VEGF-CCys156Ser was also sensitive to hLEC Ndst1 deficiency (Figure 1G; it is noteworthy that in preliminary collagen-attachment studies, hLEC attachment and spreading were somewhat slowed in Ndst1-deficient cells; data not shown).
Lymphatic-Specific Deficiency in the Sulfation of Heparan Sulfate Results in Altered VEGF-C–Driven Tumor Lymphangiogenesis
To examine the effect of a lymphatic-exclusive mutation in heparan sulfate, we used mice bearing a conditional mutation in Ndst1 driven by tamoxifen-inducible Cre under the control of the lymphatic-specific promoter Prox1 (Ndst1f/fProx1+/CreERT2 mutants). In Prox1+/CreERT2Rosa26R reporter studies, inguinal and mediastinal lymph nodes showed a relatively high degree of Cre− LYVE-1 colocalization (Figure 2A, left), noted also in ear dermal lymphatics, albeit in a more patchy distribution (Figure 2A, right). With this in mind, we established subcutaneous VEGF-C overexpressing Lewis lung carcinomas (LLC-VC) in the right flank of tamoxifen-induced Ndst1f/fProx1+/CreERT2 mutant and wild-type (Ndst1f/fProx1−/CreERT2) littermates. LYVE-1 staining showed robust lymphangiogenesis only at the tumor periphery in this model (F4/80 macrophage staining revealed a diffuse-tumor pattern, which decreased toward the tumor periphery, with nearly complete non-overlap of F4/80 with LYVE-1 staining; Online Figure III). The tumors showed no significant difference in size between mutant and wild-type groups (data not shown). Empty-vector control tumors (LLC-ev) were established in the opposite (left) flank of each mouse. In Cre− wild-type mice, LLC-VC tumors showed a significantly higher mean LVD than that of LLC-ev tumors (Figure 2B, graph, left bars), indicating a VEGF-C–dependent boost in LVD caused by tumor-associated VEGF-C expression in wild-type mice (compare representative photomicrographs of LYVE-1 immunofluorescence on left of the panel set, for Cre− animals). However, among Cre+ mutants, LVD was not greater in LLC-VC tumors when compared with that of LLC-ev tumors (Figure 2B right panels and graph, right bars). The findings suggest that the inhibitory effect of lymphatic-targeted Ndst1 mutation on LVD in this model was specifically associated with VEGF-C–mediated lymphatic-vessel growth.
To examine lymphatic proliferation in an orthotopic-tumor setting, LLC-VC cells were intravenously injected into Ndst1f/fProx1+/CreERT2 mutants and Prox1−/CreERT2 controls. The mean quantity of LYVE-1/podoplanin double-positive LECs (as a percentage of total LECs) from lung digests 7 days post injection was reduced in mutants (Figure 2C), indicating an inhibitory effect of mutation on total LECs purified from LLC-VC tumor-harboring lungs.
Phosphorylation of VEGFR-3 Is Sensitive to Altered Biosynthesis of Heparan Sulfate in Cultured Human LECs
Preliminary assessments of VEGF-C produced by cultured LLC-VC cells revealed unprocessed VEGF-C in the supernatants and lysates. The quantity of this species relative to post-translationally processed, including mature, VEGF-C species produced by the tumors in vivo (which we confirmed by Western blotting) is unknown. Because unprocessed VEGF-C contains heparin-binding propeptide extensions,2,23 and because nonmature forms of VEGF-C are variably secreted from tumors,24,25 we first screened the degree to which pro–VEGF-C (a mixture of unprocessed and partially processed VEGF-C propeptides) is able to phosphorylate receptor tyrosine kinase receptors, including VEGFR-3, in heparan sulfate–deficient versus control hLECs (Online Figure IV highlights the composition of pro–VEGF-C separated on a silver-stained gel). Initially, in a highly sensitive receptor tyrosine kinase phospho array, inhibition of hLEC heparan sulfate biosynthesis robustly blocked pro–VEGF-C–mediated VEGFR-3 phosphorylation (Figure 3A), as well as VEGFR-2 phosphorylation. Although phosphorylation of both receptors seemed to be sensitive to the glycan alteration, the baseline phosphorylation of VEGFR-3 on ligand stimulation seemed to be markedly greater in this primary cell line, consistent with its lymphatic endothelial identity. Nevertheless, the marked sensitivity of VEGFR-2 phosphorylation to glycan targeting points to an additional role for heparan sulfate in facilitating VEGFR-2 activation in response to VEGF-C, reminiscent of its importance in VEGF-A signaling.26 The effect of glycan targeting on VEGFR-3 phosphorylation in response to pro–VEGF-C was also tested in a specific (albeit less sensitive) ELISA-based phospho–VEGFR-3 assay (Figure 3B), with phosphorylation blockade that resulted from siXylT2 targeting. VEGFR-3 phosphorylation by mature VEGF-C was also significantly reduced in siXylT2-targeted hLEC (Figure 3C). Interestingly, siXylT2-inhibition of glycan-chain biosynthesis also inhibited receptor phosphorylation by a minimum receptor-binding species of VEGF-C (Figure 3D) that excludes basic amino acids in the C terminus (L216–R227; which may contribute to a weak interaction of mature VEGF-C with heparan sulfate14), implying an important cell-autonomous role for the glycan in receptor activation.
Sdc-4 Is a Dominant Proteoglycan Core Protein on Lymphatic Endothelium With Functional Significance in Pathological Lymphangiogenesis
We explored whether a dominant proteoglycan might present heparan sulfate on the lymphatic cell surface. Although LECs were not easily purified from tumors, pathological primary LECs could be isolated from mesenteric oil granuloma/lymphangioma lesions,14,15 and they were examined for the repertoire of HSPG core proteins quantitative polymerase chain reaction. Sdc-4 was the dominantly expressed lymphatic cell-surface HSPG (Figure 4A, left graph). Cells did not express CD44v3 proteoglycan, known to be expressed by blood-vascular endothelia,27,28 although they did express perlecan, which is secreted into basement membranes, and which has been detected around proliferating and collecting lymphatics.29 The HSPGs expressed by svLEC, an immortalized mouse mesenteric-lymphatic cell line, showed a similar profile (Online Figure V). It should be noted that we were able to measure a moderate increase in Sdc4 expression on Ndst1 silencing in this cell line (Figure 4A, upper-right graph); however, the expression of the other syndecans did not change in that setting, with Sdc4 remaining the dominantly expressed lymphatic HSPG.
With this in mind, we generated oil granuloma/lymphangioma lesions in Sdc-4 null (Sdc4−/−) mice, and noted reduced lesion LVD (Figure 4B). Lymphangiogenesis in Sdc4(−/−) Ndst1f/fTekCre+ double mutants examined using this pathological model was not significantly reduced in comparison with that in Sdc4(−/−)Ndst1f/fTekCre− littermates (data not shown), suggesting that Sdc-4 likely serves as a quantitatively dominant functional scaffold for lymphatic cell-surface heparan sulfate, playing a critical role in pathological lymphatic mitogen responses. We also measured HSPG core protein expression in nonpathological LECs purified from the lung (Figure 4C) or lymph nodes (Figure 4D) of wild-type and Ndst1f/fTekCre+ mutants. Sdc-4 was not only the dominant HSPG but was also disproportionately upregulated in the setting of Ndst1 deficiency, indicating that targeting the sulfation of lymphatic heparan sulfate upregulates major HSPG core protein expression.
Sdc-4 Specifically Associates With VEGFR-3 in Response to VEGF-C in Human LECs in a Heparan Sulfate–Dependent Manner and Mediates VEGF-C–Dependent Signaling
We asked whether Sdc-4 might associate with VEGFR-3 (as a possible ternary complex) in hLECs on VEGF-C stimulation. In human LECs, we previously found that expression of Sdc2 was somewhat greater than that of Sdc4.16 (as a reference, for primary human dermal microvascular blood-endothelial cells, Sdc4 seems to be the dominantly expressed transmembrane HSPG; Online Figure VI). Core protein studies also revealed that the dominant cell-surface HSPGs on hLECs were Sdc-2 and Sdc-4 (as assessed by HSPG core protein blotting; Online Figure VII). Nevertheless, proximity ligation analysis (PLA) revealed that resting starved hLEC (ie, pre-VEGF-C stimulation) are characterized by a significant degree of Sdc-4–VEGFR3 association at baseline. Exposure to mature VEGF-C strikingly increased Sdc-4–VEGFR-3 association, whereas Sdc-2–VEGFR-3 association under identical conditions was minimal (Figures 5A and 5B, left side of graph; hLEC). The magnitude of Sdc-1–VEGFR-3 association (not shown) was comparable with that of Sdc-2–VEGFR-3, with signals remaining <10% that of baseline Sdc-4–VEGFR-3 signal. Examination of Sdc-4–VEGFR-3 PLA in the mouse oil granuloma–derived svLEC line also revealed a marked rise in Sdc-4–VEGFR-3 association on VEGF-C exposure (quantified in Figure 5B, right graph; svLEC). In separate experiments examining VEGFR-2, stimulation of hLECs with mature VEGF-C did not lead to engagement of Sdc-4 with VEGFR-2 (Figure 5C; representative panels), suggesting that Sdc-4 serves as a specific coreceptor for VEGFR-3. To further explore mechanism, we found that formation of Sdc-4–VEGFR-3 complexes on VEGF-C treatment was sensitive to hLEC pretreatment with heparanase (Figure 5D), suggesting that lymphatic heparan sulfate is required for stabilizing the proteoglycan-receptor complex on ligand exposure.
With these findings in mind, we asked whether Sdc-4 deficiency might affect signaling by mature VEGF-C: phosphorylation of the mitogen-pathway intermediate Erk1/2 was sensitive (Figure 5E; representative immunoblot shown to right). As a receptor-signaling control, Erk1/2 phosphorylation in response to VEGF-C was comparatively sensitive to VEGFR-3 deficiency, and somewhat less sensitive to VEGFR-2 deficiency (Figure 5E, upper-right representative histogram). These Erk1/2 signaling findings further prompted us to use a commercial phosphosignaling array to examine patterns in VEGF-C–dependent activation of other lymphatic endothelial mitogen-activated protein kinase–associated intermediates in the setting of Sdc-4 silencing. In addition to replicating the pattern we found in Erk phosphorylation in the array (with predominantly Erk1 showing a blunted response to VEGF-C stimulation in the setting of Sdc-4 deficiency), we also noted inhibition of a second mitogen-activated protein kinase intermediate (p38β) along with associated inhibition of HSP27 phosphorylation in the same setting (Figure 5F, with quantified responses below). The array also demonstrated concomitant reduction in VEGF-C–dependent Akt2 and TOR activation, which corroborates the original findings in Figure 1, showing altered lymphatic endothelial survival signaling as a result of targeting the glycan chain.
We examine herein the genetic importance of lymphatic heparan sulfate and that of a key proteoglycan core protein in pathological lymphangiogenesis. Genetic targeting of the glycan impairs pathological lymphangiogenesis in vivo and lymphatic mitogen and survival signaling and phosphorylation of VEGFR-3 in response to VEGF-C. We also demonstrate the genetic importance of Sdc-4 as a key HSPG that scaffolds heparan sulfate on the lymphatic surface, and propose that it functions as a major coreceptor in VEGF-C–mediated pathological lymphangiogenesis.
In Ndst1f/fTekCre+ mutants, Ndst1 inactivation under the Tek-promoter generates a pan-endothelial mutation, and Ndst1 expression in LECs purified from mutants was markedly reduced. Lymphatic signaling via VEGF-C/VEGFR-3 seemed to contribute to lymphatic proliferation in both oil granuloma/lymphangioma models21 (Figure 1A) as well as transgenic carcinoma (Figure 1C) models. Although altering lymphatic Ndst1 inhibits VEGF-C–dependent sprouting and growth signaling in vivo, other heparin-binding growth factors such as VEGF-A, fibroblast growth factor-2, or PDGF may also contribute to lymphatic growth and remodeling. Nevertheless, deficiency in the glycan not only altered VEGF-C–mediated protection of primary LECs from apoptotic stress (Figure 1C, bottom) but consistent with this, Erk- and Akt-mediated signaling in primary hLECs was also inhibited in mutants (Figure 1E through 1G). Pathological blood-vascular angiogenesis seemed to be altered in such pan-endothelial Ndst1 mutants, consistent with previous work.10 Although an indirect effect of the blood-vascular mutation on lymphangiogenesis is possible in the setting of pathological angiogenesis, findings using high-specificity lymphatic gene targeting in vivo (discussed below) together with ex-vivo and cell-based work herein point to an important and direct role for lymphatic-specific heparan sulfate in VEGF-C–mediated VEGFR-3 activation.
To stringently target heparan sulfate in VEGF-C–mediated lymphatic-specific remodeling, we used a VEGF-C expressing lung carcinoma model on a genetic background wherein Ndst1 is specifically inactivated in lymphatic endothelium through the Prox1+/CreERT2 transgene. The findings point to the genetic importance of appropriately sulfated heparan sulfate in mediating the action of VEGF-C on lymphatic endothelium in vivo (Figure 2B). It is possible that nonmature forms of VEGF-C produced by this or other neoplastic cell lines24 may be more sensitive to the effects of tumor-lymphatic Ndst1 mutation on VEGFR-3 activation because such species have a greater affinity for heparan sulfate than shorter (eg, mature) forms of VEGF-C. This may have pathophysiological importance in neoplasia, where nonmature forms of VEGF-C may play important roles in tumor-lymphatic remodeling, with possibly additional regulation through binding to HSPGs secreted into matrix (eg, perlecan). It is noteworthy that the predominant species present in pro–VEGF-C used in Figure 3A and 3B (ie, 29/31-kD propeptide; Online Figure IV) may compete with other species for VEGFR-3 binding,30 and thus contribute to negative regulation. This may explain the relatively weak stimulation by pro–VEGF-C in Figure 3B. Nevertheless, siXylT2 targeting resulted in complete inhibition of VEGFR-3 phosphorylation in this setting, possibly as a result of greater heparan sulfate binding by unprocessed and propeptide VEGF-C species. In tumors, this binding may allow for greater presence of propeptide VEGF-C on the lymphatic cell surface, where proteases (eg, ADAMTS3 tethered to endothelium30) may yield local release of mature VEGF-C. Interestingly, VEGFR-3 activation by a nonheparin-binding short form of VEGF-C remained sensitive to altered heparan sulfate biosynthesis (Figure 3D), pointing to the importance of intact lymphatic cell-surface heparan sulfate in VEGFR-3 activation/function. This is reminiscent of altered VEGFR-2 responses to a key nonheparin-binding form of VEGF-A (ie, VEGF121) when endothelial heparan sulfate is genetically altered.26
A variety of proteoglycans may tether heparan sulfate to the lymphatic cell surface or pericellular matrix.8 Although we found abundant expression of Sdc-4 on lymphatic endothelium (Figure 4), a model limitation is that Sdc4−/− mutation is not tissue specific. However, pairing the genetic importance of appropriate glycan sulfation in lymphangiogenesis with the finding that Sdc-4 forms a specific and robust association with VEGFR-3 in response to VEGF-C (Figure 5A through 5C) suggests that Sdc-4 plays a critical role in VEGFR-3–mediated lymphatic growth. Importantly, the VEGFR-3–specific mitogen response to VEGF-C seems to be glycan dependent (Figure 5D) and reduced in the setting of Sdc-4 deficiency. In light of the altered signaling with this mutation (Figure 5E), the collective findings suggest that lymphatic-endothelial Sdc-4 deficiency (or Ndst1 deficiency, which would affect glycans on all HSPGs, including the dominant membrane-bound pool of Sdc-4) results in both altered mitogen-pathway as well as altered survival/Akt signaling (Figure 1F and 1G as well as Figure 5E). Although other HSPGs could theoretically partially compensate through the collective actions of their glycans to partially support mitogen-pathway responses in Sdc-4–deficient lymphatic endothelium, the silencing of this unique proteoglycan seems to critically alter lymphatic VEGFR-3–dependent growth/survival signaling and possibly pathways that affect cytoskeletal rearrangement during pathological lymphangiogenesis. Interestingly, the cooperative activation of Erk1/2 along with p38 mitogen-activated protein kinase and HSP27 (which we found altered in VEGF-C–treated Sdc-4–deficient primary LECs; Figure 5F) has been reported to play an important role in endothelial actin cytoskeletal reorganization in response to VEGF-A, with known inhibition of HSP27 phosphorylation in response to several angiogenesis-pathway inhibitors.31–33 The findings suggest that VEGF-C–mediated activation of this pathway during lymphatic endothelial cytoskeletal remodeling may also be important, and sensitive to alterations in Sdc-4 as a coreceptor. More generally, the findings point to a critical coreceptor role for the dominant lymphatic HSPG Sdc-4 in pathological lymphangiogenesis.
Figure 6 shows a model to illustrate the functional importance of these molecules in lymphatic receptor functions: in 1 mode, glycans on the lymphatic surface may serve in a cell-autonomous role as a depot for VEGF-C that, in turn, may affect the availability for VEGFR-3 interactions (Figure 6, pathway B to right). The degree of this may depend on the species of VEGF-C available and the heparin-binding affinity of that species; with signaling by the different species variably sensitive to genetic absence of the glycan on the cell surface (Figure 3B through 3D). However, the glycan seems to play an essential role in stabilizing a ternary complex that mediates proximity of the proteoglycan core protein to the growth receptor (Figure 5), with the HSPG thus serving as a coreceptor (Figure 6, bottom). This may occur either simultaneously when ligand becomes available (Figure 6, pathway A), or step-wise via initial concentration and availability of the ligand for receptor-binding events before complex stabilization (Figure 6, 2-step pathway B). Conceptually, in considering distinct vascular beds, this leaves the possibility of proteoglycan core proteins other than Sdc-4 that could depend on the common glycan (heparan sulfate) in mediating receptor activation in response to VEGF-C. However, in the proof-of-concept studies shown here, we have demonstrated important roles for heparan sulfate expressed on lymphatic endothelial Sdc-4.
These findings may have translational potential. In tumors, although alterations in LVD might not primarily affect primary tumor size, the alteration in lymphatic conduit may reduce the potential for lymphatic metastasis; and in many invasive tumors, upregulation of VEGF-C as well as the expression of other heparin-binding growth factors and cytokines contributes to tumor lymphangiogenesis.2,6,34 Considering also the roles of both Ndst1 (appropriate glycan chain sulfation) and Sdc-4 in the proliferation of lymphatic vessels as well as the mechanistic importance of Sdc-4 as a coreceptor for lymphatic signaling (Figure 6), approaches that target the expression of these molecules in the lymphatic microenvironment may serve as new selective strategies to modulate critical lymphatic remodeling in disease.
We appreciate advice and review by Dr Jeffrey Esko, and histology assistance by Dr Nissi Varki as well as members of her core-histology team. We acknowledge Dr Goetinck for Sdc4 homozygous-null mice as well as Dr Guillermo Oliver for original provision of Prox1Cre/ERT2 transgenic mice. We thank Dr Steven Alexander for providing the SV-LEC cell line; and Dr Guido David for 3G10 and Syndecan-2 antibodies. We thank Dr Gavin Thurston (of Regeneron) for providing LLC-VC and LLC-ev cell lines. Finally, we acknowledge Dr Paul Clopton of the VA San Diego Healthcare System and James Proudfoot of the UCSD Clinical and Translational Research Institute for assistance with statistical analyses. We thank Dr Philip Gordts for valuable insights as well.
Sources of Funding
We acknowledge funding from National Institutes of Health (NIH)/National Heart, Lung, and Blood Institute (NHLBI; R01-HL107652 to M.M. Fuster and P01-HL57345), the VA Career Development and Merit Review Programs (VA-CDTA Award and VA-Merit I01BX000987-01 to M.M. Fuster), the American Cancer Society (RSG-08-153-01-CSM to M.M. Fuster). We also acknowledge NIH/NHLBI (T32 HL098062 to X. Yin and R. El Ghazal), and appreciate support from the Veterans Medical Research Foundation (M.M. Fuster) as well as the Academy of Finland (decisions 265982 and 273612 to M. Jeltsch) and the K. Albin Johansson Foundation (to M. Jeltsch).
In April 2016, the average time from submission to first decision for all original research papers submitted to Circulation Research was 15.28 days.
↵* S.C. Johns and X.Yin are joint first authors.
The online-only Data Supplement is available with this article at http://circres.ahajournals.org/lookup/suppl/doi:10.1161/CIRCRESAHA.116.308504/-/DC1.
- Nonstandard Abbreviations and Acronyms
- heparan sulfate proteoglycan
- human lung LEC
- lymphatic endothelial cells
- Lewis lung carcinoma
- lymphatic vessel density
- proximity ligation assay
- siRNA duplex scrambled control
- vascular endothelial growth factor
- VEGF receptor-2
- VEGF receptor-3
- xylosyltransferase 2
- Received February 8, 2016.
- Revision received May 23, 2016.
- Accepted May 25, 2016.
- © 2016 The Authors.
Circulation Research is published on behalf of the American Heart Association, Inc., by Wolters Kluwer. This is an open access article under the terms of the Creative Commons Attribution Non-Commercial-NoDervis License, which permits use, distribution, and reproduction in any medium, provided that the original work is properly cited, the use is noncommercial, and no modifications or adaptations are made.
- Karpanen T,
- Egeblad M,
- Karkkainen MJ,
- Kubo H,
- Ylä-Herttuala S,
- Jäättelä M,
- Alitalo K
- Fuster MM,
- Wang L,
- Castagnola J,
- Sikora L,
- Reddi K,
- Lee PH,
- Radek KA,
- Schuksz M,
- Bishop JR,
- Gallo RL,
- Sriramarao P,
- Esko JD
- Yang Y,
- Macleod V,
- Miao HQ,
- Theus A,
- Zhan F,
- Shaughnessy JD Jr.,
- Sawyer J,
- Li JP,
- Zcharia E,
- Vlodavsky I,
- Sanderson RD
- Yin X,
- Johns SC,
- Lawrence R,
- Xu D,
- Reddi K,
- Bishop JR,
- Varner JA,
- Fuster MM
- Gale NW,
- Prevo R,
- Espinosa J,
- Ferguson DJ,
- Dominguez MG,
- Yancopoulos GD,
- Thurston G,
- Jackson DG
- Achen MG,
- Roufail S,
- Domagala T,
- Catimel B,
- Nice EC,
- Geleick DM,
- Murphy R,
- Scott AM,
- Caesar C,
- Makinen T,
- Alitalo K,
- Stacker SA
- Joukov V,
- Kumar V,
- Sorsa T,
- Arighi E,
- Weich H,
- Saksela O,
- Alitalo K
- McColl BK,
- Baldwin ME,
- Roufail S,
- Freeman C,
- Moritz RL,
- Simpson RJ,
- Alitalo K,
- Stacker SA,
- Achen MG
- Lopez de Cicco R,
- Watson JC,
- Bassi DE,
- Litwin S,
- Klein-Szanto AJ
- Xu D,
- Fuster MM,
- Lawrence R,
- Esko JD
- Rutkowski JM,
- Boardman KC,
- Swartz MA
- Jeltsch M,
- Jha SK,
- Tvorogov D,
- Anisimov A,
- Leppänen VM,
- Holopainen T,
- Kivelä R,
- Ortega S,
- Kärpanen T,
- Alitalo K
- Keezer SM,
- Ivie SE,
- Krutzsch HC,
- Tandle A,
- Libutti SK,
- Roberts DD
Novelty and Significance
What Is Known?
Lymphatic-vascular remodeling in disease states such as neoplasia or inflammation may facilitate important downstream pathophysiological events, such as lymphatic metastasis or organ fibrosis, among other consequences depending on the tissue and disease process.
Although the major lymphatic vascular mitogen vascular endothelial growth factor-C (VEGF-C) is critical in driving pathological lymphangiogenesis primarily through the activation of the major lymphatic receptor VEGFR-3, coreceptors that critically regulate ligand and receptor activation are poorly understood.
Cell-surface proteoglycans displaying sulfated carbohydrate chains known as heparan sulfate are known to play key roles in endothelial growth factor binding and receptor signaling, although the genetic importance and function(s) of these molecules in lymphatic remodeling in vivo remain unknown.
What New Information Does This Article Contribute?
The sulfation of heparan sulfate on lymphatic endothelium is important in mediating the actions of VEGF-C on lymphatic endothelial growth in pathological models in vivo.
Syndecan-4 is a major heparan sulfate proteoglycan expressed on lymphatic endothelium, and it serves as a novel coreceptor for lymphatic VEGFR3 signaling, forming a glycan-dependent complex with ligand and receptor on VEGF-C stimulation.
Syndecan-4 deficiency results in reduced pathological lymphangiogenesis and reduced VEGFR-3 mitogen signaling in primary lymphatic endothelial cells, introducing a novel mode of biological modulation and possibly therapeutic targeting.
The process of lymphangiogenesis plays critical roles in the pathological progression of several important diseases. These include metastasis-promoting lymphatic remodeling in cancer, lymphatic proliferation associated with lymphagioleiomyomatosis, or fibrotic progression in idiopathic fibrosis and renal tubulointerstitial disease, among others. Overexpression of VEGF-C in the lymphatic microenvironment of these disorders is a central requirement, and growth signaling primarily through the cognate lymphatic VEGFR-3 receptor plays a critical molecular role. We report the genetic importance of a novel glycan coreceptor for pathological VEGF-C–dependent lymphangiogenesis in vivo. Mechanistic work points to important roles for appropriately sulfated lymphatic heparan sulfate in mediating Akt- and Erk-dependent lymphatic signaling as well as activation of lymphatic endothelial VEGFR-3 by multiple VEGF-C species. We also discovered that syndecan-4 is the major lymphatic heparan sulfate proteoglycan involved in mediating lymphatic VEGF-C–VEGFR-3 complex formation and signaling in a manner that critically depends on its glycan chains, and we propose a novel role for heparan sulfate proteoglycans in mediating this biological process. The findings may guide development of novel glycan-biosynthesis inhibitors or proteoglycan-targeting strategies to inhibit carcinoma spread, fibrosis in a variety of inflammatory states, or conditions where lymphangiogenesis contributes to pathological progression.