Promotion of mammalian angiogenesis by neolignans derived from soybean extracellular fluids

Excessive or insufficient angiogenesis is associated with major classes of chronic disease. Although less studied, small molecules which can promote angiogenesis are being sought as potential therapeutics for cardiovascular and peripheral arterial disease and stroke. Here we describe a bioassay-directed discovery approach utilising size exclusion and liquid chromatography to purify components of soybean xylem sap that have pro-angiogenic activity. Using high resolution accurate mass spectrometry and nuclear magnetic resonance spectroscopy, the structure of two pro-angiogenic molecules (FK1 and FK2) were identified as erythro-guaiacylglycerol-8-O-4'-(coniferyl alcohol) ether (eGGCE), and threo-guaiacylglycerol-8-O-4'-(coniferyl alcohol) ether (tGGCE). These two molecules, which are coniferyl neolignan stereoisomers, promoted in vitro angiogenesis in the μM to nM range. Independently sourced samples of eGGCE and tGGCE exhibited comparable pro-angiogenic activity to the soybean derived molecules. The cellular mode of action of these molecules was investigated by studying their effect on endothelial cell proliferation, migration, tube formation and adhesion to the extracellular matrix (ECM) components, fibronectin and vitronectin. They were found to enhance endothelial cell proliferation and endothelial cell tube formation on Matrigel, but did not affect endothelial cell migration or adhesion to fibronectin and vitronectin. Thus, this study has identified two coniferyl neolignan stereoisomers, eGGCE and tGGCE, as pro-angiogenic molecules, with eGGCE being less active than tGGCE.


Introduction
The formation of new capillaries from pre-existing vascular networks (angiogenesis) is a tightly controlled process in adult mammals. Excessive or insufficient angiogenesis is associated with major classes of chronic disease. Excessive angiogenesis is associated with cancer, psoriasis, age-related macular degeneration and arthritis [1][2][3] and this has fostered the a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 purchased from BOC Sciences, New York, NY and tGGCE (isolated from Bretschneidera sinensis) was kindly provided by Prof. Dr. Wei-Dong Zhang and Dr. Shan Lei from the Second Military Medical University, Shanghai, China.

Soybean growth and xylem sap isolation
Soybeans were grown in square pots (100 × 100 × 300 mm) containing vermiculite (Ausperl, Banksmeadow, NSW, Australia) under glasshouse conditions of 14 hr light (28˚C) and 10 hr dark (25˚C). Pots were watered every day with~200 mL of urban water which was replaced with modified Herridge's nutrients [37] containing 10 mM KNO 3 every second day from the second week of planting.
Xylem sap, a clear solution lacking any visible signs of chlorophyll contamination, was collected from decapitated plants, 4-6 weeks post-germination [38,39]. Sap was collected between 0-7 hr from the wound site by attaching 5 mL syringes via a short length (3-4 cm) of rubber tubing (i.d., 3 mm) fitted over the cut stump. Exudation was estimated at 0.34 mL/hr per plant over 24 hr. Sap from 50-300 plants was pooled and lyophilized overnight.

Size fractionation and HPLC separation
Molecular weight spin filters (10 kDa and 3 kDa; Merck Millipore, Billerica, MA) were utilised in succession to remove the high molecular weight sap proteins [38] and enrich the low molecular weight sap molecules. The 3 kDa filtrate was freeze dried, redissolved in 100 μL of water (equivalent to 40 mL of fresh xylem sap) and analysed by HPLC (LC10-VP series, Shimadzu, Kyoto, Japan) using a reverse phase Alltech Platinum C18 column (5 μm, 250 × 4.6 mm) and guard cartridge (Grace Discovery Sciences, Columbia, MD). The column was held at 35˚C and eluted over 30 min with a linear gradient of 5% to 85% acetonitrile (hold 5 min) at 1 mL/min. Sample components were detected with a photodiode array (PDA) detector and peaks collected by monitoring UV absorbance in real time at 254 nm. Fractions were collected from 2 to 30 min at 30 sec intervals, dried under nitrogen, diluted to 200 μL with water and either screened for activity in the pro-angiogenic bioassay or stored at -70˚C.

Bioassay for angiogenesis modulating activity using the in vitro rat aorta assay
The rat aorta angiogenesis bioassay [33] was modified to enhance the ability to screen for proangiogenic activity in the HPLC-derived fractions by reducing the concentration of heat inactivated foetal calf serum (HIFCS) from 20% to 5% and incubating for 7 days in 48 well culture plates (Costar, Corning, Lowell, MA) at 37˚C in 5% CO 2 [36]. Six technical and three biological replicates were used and the medium was changed on day 4 of culture. Vessel outgrowth was quantified as percent growth under a Nikon TMS inverted phase contrast microscope (Nikon Instrument Inc., Tokyo, Japan) at 40 × magnification on days 5, 6 and 7. Control cultures either received medium with the diluent alone or with the known anti-angiogenic agent, PI-88 [33] added at 100 μg/mL. Vessel outgrowth was determined as the percentage of a microscope field, outside the aorta ring, that was occupied by blood vessels using ImageJ software as previously described [40] or measured manually, particularly when large numbers of column fractions needed to be scored.

Liquid chromatography mass spectrometry (LC-MS)
Mass spectrometry of the purified pro-angiogenic components was carried out using an Agilent 6530 Accurate Mass LC-MS Q-TOF (Santa Clara, CA). Samples were subjected to electrospray ionisation (ESI) in the Jetstream interface in positive and negative modes under the following conditions: gas temperature 300˚C, drying gas 4 L/min, nebulizer 35 psig, sheath gas temperature 350˚C and flow rate of 11 L/min, capillary voltage 3500 V, fragmentor 175 V, and nozzle voltage 1000 V. Unfractionated xylem sap or HPLC purified pro-angiogenic components (3 μL) were injected onto an Agilent Eclipse XDB-C18 (2.1 × 50 mm column; 1.8 μm) and eluted with a linear gradient from 10 to 50% of mobile phase B in 8 min, then to 70% in 4 min (hold for 8 min) at a flow rate of 200 μL/min. Mobile phase A was water containing 0.1% formic acid; mobile phase B was acetonitrile/water (9:1 v/v) containing 0.1% formic acid. The instrument was run in extended dynamic range mode from m/z 100-3000 and data acquired by targeted collision induced dissociation (CID; N 2 collision gas supplied at 18 psi) MS/MS (2 spectra/s). Data were acquired and analysed with Agilent's MassHunter software.

Nuclear magnetic resonance spectroscopy (NMR)
Nuclear magnetic resonance (NMR) spectroscopy experiments were recorded on either a Bruker AVANCE 800 or 600 MHz NMR spectrometer with TCI cryoprobe (Bruker, Billerica, MA) using D 2 O as the solvent at 298 K. Spectra were analysed using Bruker TopSpin 2.1 software. 1 H NMR chemical shifts in parts per million (ppm) are reported using the HOD signal as an internal chemical shift reference (4.72 ppm at 298 K). 13 C chemical shifts in ppm are referenced indirectly to the proton shift. 1 H: 1D proton NMR spectra were performed using a standard Bruker pulse program, zgpr, which included solvent suppression via pre-saturation. A total experiment time of 5 min was used with a t 1max of 1.25 sec. 1 H-1 H DQF-COSY: A phase-sensitive DQF-COSY spectrum was also measured using the standard Bruker sequence cosydfphpr, which includes a double-quantum filter and pre-saturation. A total experiment time of 38 min was used with a t 1max of 26 msec and a t 2max of 104 msec. 1 H-1 H TOCSY: A TOCSY spectrum was measured using the Bruker pulse program mlevphpp that was modified to include pre-saturation. A total experiment time of 19 min was used with a t 1max of 13 msec, a t 2max of 250 msec and a TOCSY mixing time of 60 msec. 1 H-1 H NOESY: A NOESY spectrum was recorded using the Bruker pulse program noesygpphpp modified to include pre-saturation. A total experiment time of 14.2 hr was used with a t 1max of 39 msec, a t 2max of 1.62 sec and a 500 msec NOE mixing time. 13 C-1 H HSQC: A 13 C-HSQC spectrum was recorded using the standard Bruker pulse sequence hsqcedetgpsisp2.2. A t 1max of 15.3 msec, a t 2max of 152 msec and a total experiment time of 2.3 hr was used. 13 C-1 H HMBC: A 13 C-HMBC spectrum was recorded using the standard Bruker pulse sequence hmbcetgpl2nd, which includes a two-fold low-pass J-filter to suppress one-bond correlations. A t 1max of 10.1 msec, a t 2max of 304 msec and a total experiment time of 8 hr was used.

Cell proliferation (mitogenic) assay
Human umbilical vein endothelial cell (HUVEC) proliferation was assessed by determining 3 Hthymidine incorporation [41]. HUVECs were cultured in M199 medium supplemented with 20% HIFCS, 0.24 mg/mL gentamycin, 2 mM L-glutamine, 0.04 mg/mL endothelial cell growth supplement (ECGS) (Sigma-Aldrich, St. Louis, MO) and 0.135 mg/mL heparin (Sigma-Aldrich, St. Louis, MO) in gelatin-coated 96-well plates and incubated for 4 days at 37˚C in 5% CO 2 to reach confluence. The confluent cells were then serum starved for 24 hr. Subsequently, 100 μL/ well of test compound in serum-free medium with or without basic fibroblast growth factor (bFGF) (R&D Systems, Minneapolis, MN) was added to each well for another 24 hr before adding 3 H-thymidine. 3 H-thymidine (0.5 μCi/well; MP Biomedicals, Solon, OH) was added and incubated for the final 24 hr. Incubation was stopped by adding 100 μL/well of trypsin/EDTA to lift the cells from the gelatin layer with 100 μL/well of M199 medium supplemented with 20% HIFCS added to neutralize the trypsin enzymatic activity. Finally, plates were frozen and thawed 3 times and cells were harvested using a 96-well cell Filtermate 196 harvester, with EasyTab TM -C selfaligning filters and the incorporated radioactivity counted with a Topcount 1 NXT™ Microplate Scintillation and Luminescence Counter (Packard Bioscience, Meriden, CT).

Cell migration assay (wound healing assay)
Label-free, kinetic assays for cell migration used the IncuCyte TM live-cell imaging system (Essen BioSciencen, Ann Arbor, MI) following the techniques described previously [42]. Briefly, human microvascular endothelial cells (HMEC) were added to 96-well ImageLock Essen plates (Essen BioSciences, Ann Arbor, MI) at 2.5 × 10 4 cells/well in 100 μL/well of MCDB 131 medium (Invitrogen) supplemented with 10% HIFCS, 0.1% PSN (0.03 g/L penicillin G, 0.05 g/L streptomycin sulfate, 0.05 g/L neomycin sulfate), 2 mM L-glutamine, 0.01 μg/ mL endothelial cell growth factor (ECGF), (Gibco BRL, Grand Island, NY) and 1 μg/mL hydrocortisone (Sigma-Aldrich, St. Louis, MO) and incubated for 2 days at 37˚C in 5% CO 2 to reach confluence. After removing the medium from each well, a wound was made in the HMEC monolayer using the 96-pin WoundMaker device. The use of Essen ImageLock plates ensured that the wounds were automatically located and registered by the IncuCyte software for imaging and data recording.
Each well was then rinsed twice with 100 μL/well of culture medium to prevent dislodged cells from setting and re-attaching. Medium (100 μL/well), with or without test compound, was added to each well. Wound images were then captured and saved at 2 hr intervals until control wounds had recovered completely. The data was analysed either by the IncuCyte software package or extracted by three integrated metrics (wound width, wound confluence, relative wound density) and analysed by Prism statistical software.

Endothelial cell tube formation assay
HMEC and HUVEC cultures were used for in vitro endothelial tube formation. Matrigel (BD Biosciences, Bedford, MA) was thawed overnight at 4˚C and plated into ice-cold 96-well ImageLock Essen plates at 50 μL/well using pre-cooled pipets, tips and tubes. The plates were incubated at 37˚C in 5% CO 2 for 1 hr to allow the Matrigel to form a stable gel. HUVECs at 4×10 4 cells/well in 100 μL of M199 medium supplemented with 20% HIFCS, 0.24 mg/mL gentamycin, 2 mM L-glutamine, 0.04 mg/mL ECGS, and 0.135 mg/mL heparin with or without test compounds were added and Matrigel cultures placed inside the IncuCyte and incubated at 37˚C for 24 hr. HMEC (100 μL; 5×10 4 cells/well) in MCDB 131 medium supplemented with 10% HIFCS, 0.1% PSN, 2 mM L-glutamine, 0.01 μg/mL ECGF, and 1 μg/mL hydrocortisone ± test compounds were added to the Matrigel. Tube formation was imaged at 2 hr intervals using the IncuCyte TM live-cell imaging system with the phase-contrast ImageLock scan type. The images were analysed by measuring percentage of denuded area and number of sprouting cells at early stages of tube formation (4 hr) and the total number and length of tubes formed at later stages of tube formation (6 hr) in each well using IncuCyte and NIH ImageJ software with the Angiogenesis Analyzer plugin [43] for quantification of tube networks.

Rose Bengal cell adhesion assay
The Rose Bengal cell adhesion assay is based on a previously described method [44] used to measure the interaction of antibodies with cell surface antigens using an automated colorimetric assay. A 96 round bottom (U-well) culture plate (Costar, Corning, Lowell, MA) was first coated either with bovine plasma fibronectin (Invitrogen, Carlsbad, CA) or purified human vitronectin (Gibco BRL, Grand Island, NY) at 50 μL/well at selected concentrations ranging from 0.313 to 10 μg/mL and incubated overnight at 4˚C. The liquid layer was then removed and the cells washed by submerging twice in a PBS bath (removing the PBS between washes). The non-specific binding sites were then blocked by incubation with 200 μL/well of 1% BSA in Hank's balanced salt solution (HBSS) for 1 hr at 37˚C before removing the supernatant. HMEC (5 × 10 4 cells/well) were added in 100 μL/well of serum free culture medium supplemented with 0.1% BSA with or without the test compounds. The plate was then incubated for (2.5-60 min) at 37˚C. Unbound cells were removed and 100 μL/well of 0.25% (w/v) Rose Bengal dye (Koch-Light Laboratories Ltd., Colnbrook, Berkshire, England) in PBS was added to stain the bound cells, for 3 min at room temperature. The non-absorbed dye was then removed and the plate washed twice by submerging in fresh PBS baths.
After draining the plate, 200 μL/well of 50% ethanol in PBS was added and mixed to liberate dye from the cells. Non-specific binding of the dye to fibronectin/vitronectin-coated and uncoated wells in the presence and absence of cells was also determined and subtracted. The relative number of bound-cells in each well was quantified by determining each well's optical density (OD) at λ 1 = 540 nm and λ 2 = 650 nm, on a Thermomax microplate reader (Molecular Devices, Sunnyvale, CA). The data was then analysed by GraphPad prism 5.04 software (GraphPad Software, San Diego, CA).

Statistical analyses
Data are reported as mean ± SEM. Statistical significance was measured using a two-tailed unpaired t-test between the sample treated and untreated groups and one-way and two-way ANOVA, comparing each group to control and other groups using the GraphPad prism 5.04 software. P values less than 0.05 were considered statistically significant.

Isolation and fractionation of low molecular weight pro-angiogenic molecules from soybean xylem sap
Concentrated soybean xylem sap was fractionated using reversed-phase C18 HPLC chromatography and screened for the presence of hydrophobic, low molecular weight angiogenesis-modulating molecules using the rat aorta ring bioassay. Although yield varied depending upon plant growth, fractions eluting at 7-7.5, 13.5-14 and 18-18.5 min gave highly consistent and significant pro-angiogenic activity when measured at days 5, 6 and 7 using the rat aorta assay ( Fig  1A). We focussed on the more hydrophobic fractions at 13.5-14 and 18-18.5 min because activity strongly correlated with the presence of UV absorbing peaks at these elution times (S1 Fig). The UV absorbing and biologically active fractions were separated using an optimised HPLC gradient (Fig 1B). Two peaks eluting between 13.5-14 min gave identical UV absorption spectra each of which were distinguishable from the peak at 18-18.5 min (S1 Fig). The three purified fractions (designated FK1, FK 2 and P6) showed significant pro-angiogenic activity (Fig 1C-1E). Upon long term storage the activity of P6 diminished, suggesting instability, but FK1 and FK2 were stable. Due to instability and insufficient material, P6 was not analysed further.

NMR analysis of purified FK1 and FK2
A scaled-up extraction of 300 plants yielded 2000 mL of xylem sap. Optimization of the HPLC separation of FK1 and FK2 enabled the isolation of 458 μg of FK1 and 387 μg of FK2 of sufficient purity to enable NMR and MS analysis. The pro-angiogenic activity of the purified FK1 and FK2 was confirmed by bioassay before nuclear magnetic resonance (NMR) and mass spectrometry (MS) analysis (see below) which validated that the isolated FK1 and FK2 were pure, UV absorbing and biologically active.  Initial 1D 1 H NMR spectra were recorded for both FK1 and FK2 products. FK1 was determined to be more pure and of higher concentration, so initial structure determination was performed on this product. 13  Assignments could easily be made for FK2 due to the similarity of the 13 C-HSQC crosspeaks to those of FK1 and 1 H-1 H couplings constants (S2 Fig). It was then proposed that FK1 and FK2 were diastereomers, with undefined chirality of C-7 and C-8. The compound identified (Fig 2) was proposed to be the previously described natural product guaiacylglycerol-8-O-4'(coniferyl alcohol) ether (GGCE), coniferyl neolignans of which there are four possible stereoisomers. The NMR spectra were found to be in agreement with previously reported chemical shifts and couplings [45].

LC-MS/MS analysis of purified FK1 and FK2
Initial QTOF MS data implied a chemical formula of C 20 H 22 O 6 , for both FK1 and FK2. However, through the NMR structure elucidation process, an extra proton shift was identified correlated to the spin system of FK1. It was proposed that the mass initially observed in the QTOF MS was in fact the [M-H 2 O]peak, implying a formula of C 20 H 24 O 7 . This was confirmed by reducing the nebulizer spray temperature to reveal the predicted molecular ions for FK1 and FK2 by both -ve ESI ([M-H] -, m/z 375) (Fig 3 and S1 Table) and +ve ESI (M + , m/z 376; and [M+Na] + , m/z 399) (S3 Fig and S2 Table).
The deprotonated  (Fig 3) [46,47]. A comparison of the observed and calculated masses for the two deprotonated isomers and their associated fragments showed them to be in good agreement (Δppm between -3.19 and 4.85) (S2 Table).  Fig). A comparison of the observed and calculated masses for these ions showed them to also be in good agreement (Δppm between -1.32 and 0.58) (S2 Table). It is worth noting the presence of the odd electron ion [M] + (S3 Fig). Although it was present in low abundance, this was still sufficient to calculate an elemental formula from the measured accurate mass. Odd electron molecular ions are occasionally observed in ESI MS, particularly when compounds with low redox potential are analysed at low flow rates so that the neutral species is in contact with the capillary/solution interface for a longer time [48]. The NMR and MS data are all consistent with FK1 and FK2 being identified as diastereomers of GGCE, namely erythro-guaiacylglycerol-8-O-4 0 -(coniferyl alcohol) ether (eGGCE) and threo-guaiacylglycerol-8-O-4 0 -(coniferyl alcohol) ether (tGGCE) (Fig 2), which are classified as neolignans.
In addition, a commercially purchased sample of eGGCE and an independently derived natural product isolate of tGGCE (isolated from Bretschneidera sinensis) were sourced and found to have near identical spectroscopic and chromatographic properties as the isolated FK1 and FK2, respectively (data not shown).

In vitro activity of independently sourced eGGCE (FK1) and tGGCE (FK2) in the rat aorta ring bioassay
The biological activity of the independently sourced eGGCE and tGGCE were investigated over a range of concentrations (5 × 10 −6 M to 5 × 10 −9 M) to confirm that they had the same activity as those derived from soybeanThe results, expressed as percent growth, showed significant enhancement of vessel outgrowth by both eGGCE and tGGCE compared to the control after 5, 6 and 7 days of culture at the concentrations examined (Fig 4).

Effects of eGGCE (FK1) and tGGCE (FK2) on endothelial cell proliferation
eGGCE and tGGCE were tested at concentrations ranging from 5 × 10 −6 M to 5 × 10 −12 M for their effect on the proliferation of serum-starved confluent HUVEC with or without the mitogen bFGF (at 12.5 ng/mL). In the presence of bFGF, these molecules significantly induced HUVEC proliferation at 5 × 10 −6 M to 5 × 10 −8 M and 5 × 10 −6 M to 5 × 10 −12 M concentrations, respectively (Fig 5). The peak of activity occurred at 5 × 10 −8 M in both cases and tapered off at lower and higher concentrations, with tGGCE being more active than eGGCE at the lower concentrations (i.e., 5 × 10 −9 M to 5 × 10 −12 M). In addition, the results showed that in the absence of bFGF, tGGCE (but not eGGCE) significantly enhanced HUVEC proliferation between 5 × 10 −8 M to 5 × 10 −10 M compared to the control. Based on these findings it can be concluded that both neolignans eGGCE and tGGCE can enhance endothelial proliferation, but tGGCE is more effective than eGGCE.

Effect of eGGCE (FK1) on endothelial cell differentiation (tube formation assay)
HUVEC and HMEC tube formation assays were conducted on Matrigel in the presence and absence of eGGCE and the number of tubes, the number of branch points, the length of tubes, and the percentage of area covered by tubes versus total area were assessed (Aranda and Owen, 2009) using NIH ImageJ [43] and IncuCyte software. Initially, different concentrations of HUVEC (1-4 × 10 4 /well) were cultured for 6 hr with or without 5 × 10 −6 M eGGCE, with enhanced tube formation by HUVECs being observed at all three cell concentrations tested (Fig 6A). Subsequent studies were undertaken at the highest HUVEC concentration (4 × 10 4 / well) and eGGCE was shown to enhance HUVEC tube formation above control levels with all four measured parameters, with the enhancing activity tapering down with increasing dilution Promotion of mammalian angiogenesis by neolignans derived from soybean extracellular fluids (Fig 6B-6E). Denuded areas and total tube length were enhanced only at 5 × 10 −6 M, the highest concentration tested, this enhancement being significant for denuded area, whereas the total number of sprouting cells and the number of tubes was significantly higher with eGGCE addition at the three and two highest concentrations tested, respectively (5 × 10 −6 M to 5 × 10 -8 M) (Fig 6B-6E). With HMEC (4 × 10 4 cells/well), tube formation was best measured by circle formation and was evident after 2 hr incubation, with an enhanced response being observed at the two highest eGGCE concentrations tested (5 × 10 −6 M and 5 × 10 −7 M) (S4 Fig). Based on these data it can be concluded that the neolignan, eGGCE, can enhance tube formation by human endothelial cells from two different sources, namely the umbilical vein and the microvasculature.

Effect of eGGCE (FK1) on endothelial cell migration (wound healing assay) and endothelial cell adhesion (Rose Bengal assay)
An endothelial cell migration assay based on cell migration into a denuded area was employed using HMEC to determine the effect of a range of eGGCE concentrations (5 × 10 −6 -5 × 10 −12 M) on endothelial cell migration. The extent and speed of wound closure was monitored microscopically every 2 hr over 26 hr (S5 Fig). In six independent experiments (n = 6), eGGCE imparted no measurable effect on the speed and extent of wound closure (S6 Fig). Since eGGCE was able to significantly enhance HUVEC and HMEC tube formation on the artificial basement membrane, Matrigel, we examined the effect of this compound on cell adhesion to the ECM components, fibronectin and vitronectin, using a Rose Bengal adhesion assay with HMEC. The fibronectin and vitronectin concentrations required for cell adhesion were first optimised by coating the microplate plastic wells with a concentration range of fibronectin and vitronectin (0 to 10 μg/mL) to establish the concentration where sub-optimal cell binding was observed. Under these conditions it was reasoned that both enhancement and  (FK1) and tGGCE (FK2). Effect of eGGCE and tGGCE, at concentrations from 5 × 10 −6 M to 5 × 10 −12 M on confluent serum-starved HUVEC proliferation in serum free media (SFM) with and without the mitogen bFGF (12.5 ng/mL). All the treatments were undertaken in the same experiment and control cultures contained the same diluent dilution as the test compounds. Cell proliferation was measured as 3 H thymidine incorporation after 24 hr incubation. Data analysis was performed by Student-Newman-Keuls test after one-way ANOVA comparing each group to control in each treatment. Error bars represent SEM (n = 6). Ã , P 0.05, ÃÃ , 0.01, ÃÃÃ , 0.001. Also, a two-way ANOVA data analysis showed a significant difference for all tested concentrations comparing their effect in SFM to SFM+bFGF (12.5 ng/mL) treatments, P 0.0001. https://doi.org/10.1371/journal.pone.0196843.g005 Promotion of mammalian angiogenesis by neolignans derived from soybean extracellular fluids Promotion of mammalian angiogenesis by neolignans derived from soybean extracellular fluids inhibition of cell adhesion may be detected in the presence of eGGCE at different concentrations. The results showed the optimal concentrations of fibronectin and vitronectin for cell binding were 10 μg/mL and 5 μg/mL, respectively, and sub-optimal cell binding occurred at concentrations ranging from 0.313 to 2.5 μg/mL (S7 Fig). eGGCE did not significantly affect HMEC adhesion to fibronectin or vitronectin after 30 or 60 min adhesion nor over a time course from 5-30 min (Fig 7 and S8 Fig). Thus, it was concluded that eGGCE had no significant effect on either endothelial cell migration or endothelial cell adhesion to the ECM components fibronectin and vitronectin.

Discussion
An activity-guided rat aorta ring bioassay with a proven ability to identify anti-angiogenic compounds [33] was modified by reducing the concentration of HIFCS from 20% to 5% [36]. This slowed the rate of tube formation but enabled the screening of pro-and anti-angiogenic activities over three consecutive days. Using this assay in combination with size fractionation and HPLC, soybean xylem sap derived from plants grown under several conditions was repeatedly shown to have three fractions of pro-angiogenic activity. The complexity of the UV spectrum for the earliest eluting material combined with its hydrophilic nature and the instability of P6, led us to focus on fractions which contained the UV absorbing peaks FK1 and FK2. Sufficiently pure FK1 and FK2 enabled their identification using NMR and high resolution mass spectrometry as the coniferyl lignin precursors, erythro-guaiacylglycerol-8-O-4 0 -(coniferyl alcohol) ether (eGGCE) and threo-guaiacylglycerol-8-O-4 0 -(coniferyl alcohol) ether (tGGCE), respectively, and this was verified by comparison with independently sourced samples of each.
In plants, these molecules are termed neolignans, which are derived from the dimerisation of the lignin precursors called monolignols. Lignin is a heterogeneous high molecular weight polymer and a key structural component of secondary plant cell walls [49,50]. Three monolignols, namely ρ-coumaryl, coniferyl and synapyl alcohols, are synthesised intracellularly and their lignan dimers are secreted extracellularly. This is consistent with the presence of eGGCE and tGGCE in extracellular xylem sap fluid. Lignans have been found in a range of plants and shown to have a number of medically important biologically activities such as anti-tumor [51,52], anti-oxidant [53] and anti-inflammatory activities [54,55]. Of particular relevance here is the anti-tumor activity of neolignan metabolites, which has been proposed to be due to the anti-angiogenic activity of these molecules [56]. Collectively, these findings suggest that subtle changes in the structure of neolignans can dramatically change their effects on angiogenesis. In fact, we have previously reported that lipo-chitin oligosaccharides behave in a similar manner, the molecules varying dramatically in their pro-or anti-angiogenic activity depending on their structure [36]. eGGCE and tGGCE induced tube formation using the rat aorta ring assay at μM to nM concentrations. Although this bioassay provided initial valuable information about the overall biological activity, other bioassays portraying the major steps in angiogenesis, including endothelial cell proliferation, migration, tube formation and adhesion to extra-cellular matrix components were utilised to understand the cellular mode of action of eGGCE and tGGCE. A summary of the results obtained is shown in Table 1. A proliferation assay conducted using HUVEC in the presence of a sub-optimal concentration of bFGF (at 12.5 ng/mL) showed that eGGCE and tGGCE, in particular, were biologically active in enhancing proliferation over a large concentration range peaking at 5 × 10 −8 M. In addition, tGGCE partially stimulated HUVEC proliferation in the absence of bFGF. This suggests that tGGCE may be more active than eGGCE. Although eGGCE stimulated various aspects of HUVEC and HMEC tube formation in the low μM range it did not significantly affect endothelial cell migration or the binding of endothelial cells to the extracellular matrix components fibronectin or vitronectin. Therefore, it is most likely that eGGCE and tGGCE work primarily by enhancing vascular tube formation, and may be acting synergistically with bFGF.
In a parallel study the four stereoisomers of GGCE were synthesised and a paper describing the synthesis of these four compounds was recently published [57]. It was found that all four stereoisomers could enhance endothelial cell tube formation, but eGGCE was the least active. Furthermore, the enhanced endothelial cell tube formation induced by the four GGCEs was completely inhibited by PD98059, a flavone that inhibits mitogen-activated protein kinase 1/2 (MEK1/2) and bFGF signalling. These data suggest that the GGCEs enhance angiogenesis via direct activation of the FGFR signalling pathway upstream of MEK1/2.
Thus, this study has successfully identified and characterised two novel pro-angiogenic compounds, erythro-guaiacylglycerol-8-O-4'-coniferyl alcohol (eGGCE), and threo-guaiacylglycerol-8-O-4'-coniferyl alcohol (tGGCE) in soybean xylem sap using a rat aorta bioassay guided fractionation approach. Both eGGCE and tGGCE could significantly enhance in vitro endothelial cell proliferation and tube formation on an artificial ECM, possibly by potentiating the potent mitogen, bFGF, and its downstream signalling. Thus, the coniferyl neolignans eGGCE and tGGCE represent novel pro-angiogenic molecules with considerable clinical potential.