Fig 1.
Schematics of molecular detail and structure of multi-scale computational model.
(A) Whole-body compartmental model structure and mass flow. VEGF and PlGF are secreted from parenchymal cells, and sR1 is secreted by endothelial cells into the tissue interstitial space. Ligands and sR1 can then bind to EC receptors (leading to internalization and degradation), and can be transported between the tissue and blood via bi-directional vascular permeability or lymphatic draining of tissues into the circulation. Soluble species in the blood can be directly cleared from the blood. (B) Molecular interactions in tissue interstitial space between VEGF121, VEGF165, VEGF189, PlGF1, PlGF2, NRP1, sR1, and extracellular HSPGs/GAGs (M). It is assumed that, similar to NRP1-VEGFR1 complexes, VEGF121 and PlGF1 can bind to sR1-M. ECM-bound VEGF165, VEGF189, and PlGF2 can also bind to sR1. (C) Trafficking processes simulated in endothelial cells. (D) Site-specific phosphorylation and dephosphorylation of VEGFR2. (E) Abluminal (tissue-side) endothelial cell-surface molecular interactions between VEGF121, VEGF165, VEGF189, PlGF1, PlGF2, VEGFR1, VEGFR2, NRP1, sR1, and extracellular HSPGs/GAGs in the endothelial basement membrane (EBM).
Table 1.
Binding/Unbinding reactions: KD.
Table 2.
Binding/Unbinding reactions: kon.
Table 3.
Binding/Unbinding reactions: koff.
Table 4.
Targets & secretion/production rates at steady-state.
Fig 2.
Nonlinearity of ligand & sR1 secretion and EC receptor production rates in the model.
(A) One at a time, each baseline ligand secretion or receptor production rate (inputs- listed across the top), was increased by 2%, then decreased by 2%. For each perturbation, the change in plasma ligand and EC surface receptor levels (outputs- listed on the left) in in both the main body mass (“Body”) and calf muscle (“Calf”) were obtained. The average change in output from baseline levels was calculated, and divided by the change in input (+/-2%) to give the relative change in output per % change in input. (B) Schematic of positive feedback in VEGF gene and protein levels in the model. An increase in VEGF expression increases local VEGF protein, increasing VEGF binding to VEGFR2, and subsequent internalization and degradation. This decreases total VEGFR2 protein levels, leading to reduced VEGF-VEGFR2 complex formation, which reduces net endothelial consumption of VEGF protein. To accommodate, in the model, VEGFR2 expression was increased until target baseline levels were achieved for all ligands and receptors. A similar positive feedback loop exists for changes in VEGFR2 expression.
Fig 3.
Pharmacokinetics of VEGF, PlGF, and sR1 at steady-state.
(A) Predicted free and sR1-bound ligands, and free and ligand-bound sR1 in plasma. (B) Predicted VEGF, PlGF, and sR1 distribution in healthy tissue in “Main Body Mass” compartment, shown in pM of tissue. (C) Extracellular (not bound to or inside ECs) VEGF, PlGF, and sR1 in “Main Body Mass” compartment, in pM of tissue. (D) Steady-state net flow profiles for VEGF, PlGF, sR1, and sR1-ligand complexes between the calf muscle, blood, and main body mass. All VEGF isoforms are aggregated, as are both PlGF isoforms. Green arrows represent production, red arrows EC consumption, black arrows bi-directional vascular permeability, gray arrows lymphatic drainage, and pink arrows with red outlines direct clearance from blood. The white arrows show the net association or dissociation of VEGF-sR1 and PlGF-sR1 complexes in each compartment. Displayed concentrations are free ligand, sR1, or complex in interstitial fluid or plasma. The numbers under each compartment are the respective compartment volumes. Flows are given in pmoles/day. (E) Comparison of VEGF and PlGF isoform distribution with relative isoform production rates demonstrates locations and complexes where each isoform is under- or over-represented relative to the fraction of total VEGF or PlGF production. (F) Matrix site occupancy in the EBM, ECM, and PBM.
Fig 4.
Pharmacodynamics of ligand binding to VEGFR1 and VEGFR2.
(A) Total soluble growth factor (in available interstitial fluid) and immobilized growth factor (in innermost 25nm of EBM) accessible to ECs. Growth factor bound to EC receptors is not included in this plot. (B) Break-down of EC surface-bound ligand, by isoform. Note the difference in quantities of total ligated VEGFR2, VEGFR1, and NRP1 (panel C). (C) Occupancy of VEGFR2, VEGFR1, and NRP1 on ECs, broken down by ligand and NRP1-binding. VEGFR2 occupancy is shown on the cell surface, in early signaling endosomes (Rab4/5), and in recycling endosomes (Rab11), while VEGFR1 and NRP1 are shown only on the cell surface. Quantities are given in pM of total tissue in the “Main Body Mass” compartment. (D) VEGFR2, VEGFR1, and NRP1 ligation on ECs, excluding receptor not bound to ligand. Complexes not listed in the legend are present at levels too low to be seen in the figure. (E) Break-down of percentage of EC surface VEGFR1 and VEGFR2 ligation comprised by each isoform, compared to the relative production of each isoform. Production fractions are calculated separately for VEGF and PlGF, while for receptor binding the combined distribution is shown.
Fig 5.
VEGF isoform-specific trafficking and site-specific phosphorylation of VEGFR2 in vivo.
(A) VEGF isoform-specific NRP1-binding properties result in isoform-specific trafficking of VEGFR2. (B) Subcellular location-specific dephosphorylation rates for Y1175 and Y1214 (S9 Table) lead to preferential activation of tyrosine 1214 on the EC surface, compared to signaling in endosomes. (C) Isoform-specific trafficking and location-specific dephosphorylation combine to result in isoform-specific trends in relative activation of VEGFR2 on tyrosine 1175 and tyrosine 1214. (D) Total VEGFR2 phosphorylation, on at least one tyrosine (pR2) and specifically on Y1175 or Y1214, across all subcellular locations. (E-F) Distribution of pY1175 (E) and pY1214 (F), by VEGF isoform and location.
Fig 6.
Complex regulation of VEGF family signaling by PlGF, EBM binding sites, and sR1.
(A-C) Changes in free ligand levels in tissue interstitial fluid (A), EC surface VEGFR1 ligation and VEGFR2 phosphorylation (B), and the breakdown of VEGF and PlGF bound to EC surface VEGFR1 (C), in response to varying PlGF production. Quantities shown are normalized to baseline cases. (D-F) Effect of endothelial basement membrane (EBM) binding site density on EBM site occupancy (D), fraction of occupied EBM sites bound to different ligands and receptors (E), and VEGFR1 and VEGFR2 ligation by immobilized VEGF or PlGF (F). (G-I) Total activation of VEGFR1 and VEGFR2 (G), and break-down of relative ligation by each VEGF and PlGF isoform (H-I) with varying sR1 production.
Fig 7.
Immobilized ligand binding to sR1 alters tissue distribution, while immobilized ligand binding to EC receptors alters activation state.
Panels show percent change from baseline. Thus, the smallest bars indicate little impact of the removed reactions on a given output, while large bars indicate large change when the reactions are removed. Cell Only: Immobilized ligand allowed to bind to EC receptors, but not sR1. Binding of ligand to immobilized sR1 is also not allowed. sR1 Only: Immobilized ligand allowed to bind to sR1, and ligand to immobilized sR1, but binding of immobilized ligand to EC receptors is not included. No MLR: No matrix-ligand-receptor or matrix-ligand-sR1 complexes are allowed to form. Top: Changes in fit ligand secretion and receptor production rates to match plasma ligand and sR1 targets and tissue EC surface receptor targets. Middle: Distribution of free, total, and matrix-bound VEGF and PlGF. Bottom: EC receptor activation.
Fig 8.
Predicted signaling changes in the human body with expression of single VEGF isoforms mirror experimentally observed murine phenotypes.
(A) Levels of free VEGF, PlGF, and sR1 in tissue interstitial fluid, normalized to baseline, when all VEGF production is VEGF121, VEGF165, or VEGF189. (B) Endothelial cell surface ligation of VEGFR1 and phosphorylation of VEGFR2. Changes in pR2 and ligated VEGFR2 were very similar. (C) Ratio of total VEGFR2 phosphorylation on tyrosine Y1214 to phosphorylation of tyrosine Y1175. (D) Percent of ligated EC surface VEGFR1 and VEGFR2 bound to EBM-immobilized ligand.
Fig 9.
Summary of key model predictions.
(A) Overview of key predictions. (B) Due to differences in NRP1- and ECM-binding, VEGF isoform-VEGFR2 complexes are trafficked differently, leading to distinct downstream signaling, cellular behavior, and vascular network architecture. (C) Summary of predicted ligand binding to VEGFR1 and VEGFR2. All ligands in the respective boxes can bind to VEGFR1 or VEGFR2. The size of the ligands represents the predicted contribution to receptor binding in vivo. The model suggests that, for each receptor, a subset of the ligands dominate.