Jump to
Abstract
Background—Arteriogenesis and collateral formation are complex processes requiring integration of multiple inputs to coordinate vessel branching, growth, maturation, and network size. Factors regulating these processes have not been determined.
Methods and Results—We used an inhibitor of NFκB activation (IκBαSR) under control of an endothelial-specific inducible promoter to selectively suppress endothelial nuclear factor-κB activation during development, in the adult vasculature, or in vitro. Inhibition of nuclear factor-κB activation resulted in formation of an excessively branched arterial network that was composed of immature vessels and provided poor distal tissue perfusion. Molecular analysis demonstrated reduced adhesion molecule expression leading to decreased monocyte influx, reduced hypoxia-inducible factor-1α levels, and a marked decrease in δ-like ligand 4 expression with a consequent decrease in Notch signaling. The latter was the principal cause of increased vascular branching as treatment with Jagged-1 peptide reduced the size of the arterial network to baseline levels.
Conclusions—These findings identify nuclear factor-κB as a key regulator of adult and developmental arteriogenesis and collateral formation. Nuclear factor-κB achieves this by regulating hypoxia-inducible factor-1α–dependent expression of vascular endothelial growth factor-A and platelet-derived growth factor-BB, which are necessary for the development and maturation of the arterial collateral network, and by regulating δ-like ligand 4 expression, which in turn determines the size and complexity of the network.
Introduction
Development of arterial circulation (arteriogenesis) in general and arterial collateral circulation in particular is poorly understood. During embryogenesis, formation of arteries is driven by little-known molecular processes that specify arterial fate and determine the physical extent and branching pattern of the newly forming vasculature as well as by complex spatial guidance clues.1–4 In adult settings, arteriogenesis occurs in a limited set of circumstances, with shear stress5,6 and inflammation7,8 thought to be key drivers. Collaterals represent a special case of arterial circulation.9 By definition, these are arterial vessels connecting 2 arteries or artery branches, and for this reason they have been thought to play a protective role, allowing continuous arterial blood supply in cases of individual arterial trunk occlusions.10 Indeed, there is a substantial body of literature connecting extensive collateral circulation to improved clinical and functional outcomes.11,12
Clinical Perspective on p 2600
Similar to arteriogenesis, drivers of collateral development and factors specifying their location and extent are largely unknown. The extent of arterial branching during development is one of the key factors determining overall arterial density and the frequency of arterial connections, thereby directly affecting collateral formation. Notch signaling plays a central role in regulation of vascular branching morphogenesis.2 Alterations in expression of Notch receptors (Notch-1 and Notch-4) and their ligands δ-like ligand 1 (Dll1) and δ-like ligand 4 (Dll4) have been linked to alterations in collateral extent. In particular, Dll4 is required for normal vascular development. Loss of even a single Dll4 allele results in early embryonic lethality as a result of a lack of well-defined arteries and an increased number of vessel branches and sprouts.13,14 A similar phenotype is observed in Notch-1 knockout mice.15 In adult tissues, Dll4 expression is observed in newly forming arteries and capillary sprouts after ischemia.16
Dll1 is also essential for arterial development, with Dll1-Notch signaling required for arterial expression of vascular endothelial growth factor (VEGF) receptor-2 (VEGFR2) and neuropilin-1, with the absence of Dll1 resulting in increased expression of a vein-specific transcription factor COUP-TFII.17 Mice heterozygous for Dll1 have fewer tip cells during angiogenic sprouting in the developing retinal circulation and impaired vessel branching.18 Furthermore, adult heterozygous Dll1 mice demonstrate impaired collateral growth and reduced endothelial Notch activation in a hindlimb ischemia model, leading to impaired blood flow recovery.19 Whereas Delta/Notch signaling determines arterial vasculature patterning, VEGF-A plays a central role in directing developmental and adult arteriogenesis,20 and it is equally critical to collateral formation.21,22 A decrease in VEGF signaling impairs arterial developmental and adult arterial branching morphogenesis.23,24
Whereas VEGF plays a central role in development and postnatal arteriogenesis, the source of the growth factor and the stimulus for its production in arteriogenesis in adult tissues are subject to debate. Principal possibilities include ischemic tissues, infiltrating blood-derived monocytes/macrophages, and the blood vessels themselves. The key factor regulating VEGF production in ischemic tissues is hypoxia-inducible factor-1α (HIF-1α). HIF-1α levels in turn are controlled in a highly complex manner by several regulators, including VHL and PHD proteins.25 To add to the complexity, recent studies have suggested that HIF levels are also controlled by the transcriptional factor nuclear factor-κB (NF-κB),26 with NF-κB subunits RelA (p65) and p50 directly interacting with HIF-1α promoter.27 Importantly, unlike other HIF-1α regulators, NF-κB can stabilize HIF-1α levels not just under hypoxic conditions but also under normoxic conditions.28 The latter is critically important to arteriogenesis because growth of collateral arteries occurs in tissues demonstrating minimal ischemia.29
Monocyte-derived macrophages are another source of VEGF during adult arteriogenesis. Indeed, monocytes have long been considered to be crucial to arteriogenesis7,30 in part because of their production of VEGF-A and fibroblast growth factor-2.29 A number of signals have been proposed to trigger monocyte recruitment at arteriogenesis sites, including shear stress–induced activation of adhesion molecules and local production of VEGF, which in, turn, trigger synthesis of stromal cell–derived factor-1.31
In settings of adult arteriogenesis, in which hypoxia is not a major driver, NF-κB in the endothelium can be activated by increased shear stress caused by an arterial occlusion. This activation of the endothelial NF-κB cascade can lead not only to local production of VEGF but also to accumulation of blood-derived monocytes/macrophages due to increased expression of adhesion molecules, leading to further increase in local accumulation of VEGF. The NF-κB cascade therefore appears central to 2 important events in arteriogenesis: production of VEGF and accumulation of monocytes/macrophages. Despite the seeming importance of this role, NF-κB function in arteriogenesis has not been fully addressed, and there is no known link between NF-κB and the Delta/Notch signaling pathway.
In the present study, we sought to examine the role of endothelial NF-κB in adult arteriogenesis by expressing an inhibitor of NFκB activation (IκBαSR) that fully blocks NF-κB activation. This resulted in decreased activation of adhesion molecules, a marked reduction in monocyte influx into tissues, and decreased HIF-1α activation. Surprisingly, this was also associated with a massive increase in arterial collaterals and a very poor restoration of flow that was due to decreased expression of Dll4. Furthermore, the newly formed vasculature was very immature, with poor mural cell coverage. Thus, in addition to controlling HIF-dependent activation of VEGF production and monocyte accumulation, NF-κB also regulates Dll4/Notch-dependent regulation of branching morphogenesis and the extent of collateral formation. These data place NF-κB at the center of arteriogenic response.
Methods
Conditional Transgenic Mice: IκBα Super Repressor
To generate conditional transgenic mice, IκBα super repressor (IκBαSR)–HA cDNA (kindly provided by Dr J.A. DiDonato)32 was subcloned in pBI-G vector (Clonetch) that is sensitive to tetracycline transactivator and allows simultaneous expression of LacZ and IκBαSR-HA under the control of a tetracycline response promoter. Endothelial-specific expression of IκBαSR was achieved by crossing tetracycline response promoter–IκBαSR with Tie2–tetracycline transactivator33 or VE-cadherin (VEC)–tetracycline transactivator34 transgenic mouse lines. For the hindlimb ischemia model, the transgene expression of IκBαSR, repressed during embryonic and postnatal development by doxycycline in diet at 200 mg/kg (Bio-Serv), was induced on a normal chow diet at 10 to 12 weeks of age in mice 3 days before femoral artery ligation. In the case of retinal vasculature assessment, the transgene expression of VEC-IκBαSR was continuously induced during embryonic and postnatal development. Littermates inheriting only IκBαSR or tetracycline transactivator transgene were used as controls. All animal experiments were performed under a protocol approved by the Institutional Animal Care and Use Committee of Yale University.
The hindlimb ischemia model, laser Doppler imaging, and high-resolution micro–computed tomography (micro-CT) were performed as described previously.35 Flow images were analyzed with Moor laser Doppler imaging processing software and reported as the ratio of flow in the right (ischemic) to left (nonischemic) hindlimb. Quantitative micro-CT data were expressed as a vascular segment number, representing total number of vessels, of specified diameter, counted in 500 Z-stack sections for thigh or 250 Z-stack sections for calf images.
Immunohistochemistry
Frozen sections of adductor or gastrocnemius muscle samples were immunostained with antibodies specific for CD45 (eBioscience), CD31 (Chemicon), HA-tag (Roche), and NG2 chondroitin sulfate proteoglycan (Millipore). Nuclei were visualized with 4′,6-diamidino-2-phenylindole staining. Cell coverage area was determined in captured images with a Zeiss confocal microscope and analyzed with the use of Image J software.
Flow cytometry (forward scatter detector/side scatter detector) of mechanically dissociated muscle specimens was performed on a FACSCanto (BD Biosciences, San Jose, CA) device, with gating on CD45+ leukocytes. Antibodies were specific for CD45 (30-F11) and CD11c (HL3) (from BD Biosciences) and for F4/80 (BM8), CD11b (M1/70), Gr1 (RB6–8C5), B220 (RA3–6B2), and CD3 (145–2C11) (from eBioscence).
Cell Culture Experiments
Confluent human umbilical vein endothelial cells (HUVECs) (Lonza), passage 3, cultured in complete medium (Lonza) at 37°C, 5% CO2, were transduced with Ad-Null, Ad- green fluorescent protein (GFP), or Ad-IκBαSR (UNC Gene Therapy Vector Core) (100 multiplicity of infection) for 24 to 48 hours. The treatment with tumor necrosis factor (TNF)-α (10 ng/mL) or VEGF-A (50 ng/mL) was performed for 24 hours in starvation medium (endothelial basal medium-2; Lonza) supplemented with 0.5% fetal bovine serum, 0.25% bovine serum albumin, and antibiotics. Cells were starved overnight before treatment.
Hypoxia Experiments
HUVECs were transduced with Ad-GFP or Ad-IκBαSR and cultured under either normoxic (room air) or hypoxic (0.4% O2) conditions. After 24 hours, endothelial cells were harvested, and nuclear extract was isolated with the use of NE-PER nuclear and cytoplasmic extraction reagents (Thermo Scientific).
Shear Stress Experiments
Flow experiments were performed with the use of a parallel plate flow chamber. HUVECs transduced with Ad-GFP or Ad-IκBαSR were plated on 10-μg/mL fibronectin–coated slides. After overnight starvation in media containing 2% fetal bovine serum and 0.1% endothelial cell growth supplement, cells were subjected to oscillatory flow for 18 hours as described.36
Endothelial Cell Sprouting Assay
HUVECs were transduced with 2×106 plaque-forming units Ad-IκBαSR or Ad-Null. Twenty-four hours after transduction, cells were harvested and resuspended in 300 μL fibrinogen solution (2.5 mg/mL fibrinogen [Sigma] in endothelial basal medium-2 supplemented with 2% fetal bovine serum and 50 μg/mL aprotinin [Sigma]) and plated on top of a precoated fibrin layer (400 μL fibrinogen solution clotted with 1 U thrombin [Sigma] for 20 minutes at 37°C). The second layer of fibrin was added and allowed to clot for 1 hour at 37°C. Human fibroblasts, WI-38 cells (250 000 cells per well) in endothelial basal medium-2 supplemented with 2% fetal bovine serum and 25 ng/mL VEGF-A, were then plated on top of the second fibrin layer. Cultures were then incubated at 37°C, 5% CO2. After 48 hours, the 17-mer Jag-1 peptide (CDDYYYGFGCNKFCRPR) or control scrambled peptide (RCGPDCFDNYGRYKYCF)37 was added to the cultures. After 3 to 5 days, the cultures were labeled with 4 μg/mL Calcein AM for 1 hour and imaged by fluorescence with the use of a standard fluorescein isothiocyanate filter.
Whole Mount Retina Staining
Eyeballs were removed from neonates at postnatal day 4 and prefixed in 4% paraformaldehyde for 15 min at room temperature. The dissected retinas were blocked overnight at 4° C in TNB (0.1M Tris-HCl, 150 mM NaCl, 0.2% Blocking reagent [PerkinElmer]) supplemented with 0.5% Triton-X. Then, retinas were stained with IsolectinB4 and immunostained with antibodies specific for NG2 chondroitin sulfate proteoglycan (Millipore) or Jagged-1 or Dll4 (R&D) in TNB with 0.5% Triton-X. Quantification of retinal vasculature was done with the use of Biological CMM Analyzer software.38 The data are presented as number of vessel branch points, vessel length, and number of segments per ×10 magnification fields.
Whole Mount Brain Staining
Brains were harvested at postnatal day 4, fixed in 1:4 dimethyl sulfoxide/methanol overnight at 4°C, washed 3 times in methanol, and stored in methanol at −20°. Brains were then rehydrated, washed in phosphate-buffered saline, and blocked overnight in TNB (0.1Tris-HCl, 150 mM NaCl, 0.2% Blocking reagent [PerkinElmer]) with 0.5% Triton-X. After overnight incubation with Cy3-SMA (Sigma), brains were washed and imaged with the use of a fluorescent dissecting microscope.
Jagged Peptide Treatment in Neonates
Jag-1 peptide or scrambled peptide was resuspended in 50% dimethyl sulfoxide and 50% H2O at a concentration of 10 mg/mL. The peptides were injected (25 μL per injection) subcutaneously 3 times every 12 hours starting at postnatal day 2. Mice were euthanized at postnatal day 4, and retinas were harvested.
Western Blotting
Immunoblotting
Proteins, tissue, or cell homogenates were extracted in radioimmunoprecipitation assay lysis buffer supplemented with protease inhibitor cocktail (Roche). The lysates were clarified by ultracentrifugation (14 000 rpm, 15 minutes at 4°C), and 30 μg protein extract, determined by bicinchoninic acid assay, was subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis. After transfer to Immobilon membranes (Millipore), the membranes were blocked with 5% nonfat dried milk in Tris-buffered saline, pH 7.4, containing 0.1% Tween 20, and immunoblotted with the use of specific antibodies against the following: IκBα, VEGF-A, VEC, platelet-derived growth factor (PDGF)-B, intercellular adhesion molecule-1 (ICAM-1), actin (Santa Cruz Biotechnology); HA tag (Roche); Jagged-1, Dll4, Notch1, Cleaved Notch (Val 1744), VEGFR2, PDGF receptor-β, fibroblast growth factor receptor-1, insulin-like growth factor-1 receptor-β, GFP (Cell Signaling); HIF-1α (BD Transduction Laboratories); HIF-2α (Novus Biological); and -TBP: TATA binding protein (Abcam). Immunoreactive bands were visualized with the use of the horseradish peroxidase–conjugated secondary antibody and enhanced chemiluminescent substrate (Pierce). Images were captured with G:Box gel imaging system (Syngene).
Statistical Analysis
Data are presented as mean±SEM. Differences between multiple groups were assessed with 1-way ANOVA followed by the Bonferroni post hoc test for multiple comparisons. Comparisons between 2 independent groups were performed with a 2-sample t test. Flow measurements were analyzed with a linear mixed effect model for repeated measurements as well as t tests. All P values were calculated with 2-tailed statistical tests. Differences were considered significant when P<0.05. Data were analyzed with the use of IBM SPSS Statistics 19.00.
Results
NF-κB Suppression in Endothelial Cells Impairs Effective Arteriogenesis
To study the role of endothelial NF-κB in angiogenesis and arteriogenesis, we generated a transgenic mouse expressing a mutant form of IκBα under control of ischemia-sensitive Tie2 or VEC promoter using a “tet-off” approach. The double transgenic mice (Tie2-IκBαSR, VEC-IκBαSR) were kept on doxycycline throughout development and postnatal life, with the drug being discontinued 3 days before the initiation of experiments.
Serial measurements of blood flow in the distal hindlimb after ligation of the common femoral artery with the use of deep penetrating laser Doppler demonstrated nearly complete recovery in control mice, whereas perfusion in Tie2-IκBαSR mice never recovered (Figure 1A). Surprisingly, despite this profound reduction in blood flow recovery, micro-CT imaging of the hindlimb vasculature demonstrated extensive development of the above-the-knee arterial collateral circulation in both control and Tie2- IκBαSR mice groups (Figure 1B). Quantitative analysis demonstrated higher numbers of smaller arteries in Tie2-IκBαSR transgenics (Figure 1C). Furthermore, below-the-knee micro-CT images of arterial vasculature showed extensive neovascularization in Tie2-IκBαSR mice, with a markedly abnormal vascular branching pattern reminiscent of a tumor circulation (Figure 1D and 1E).
Impaired blood flow recovery and increased arteriolar branches in Tie2–IκBαSR and VE-cadherin (VEC)–IκBαSR transgenic mice after ligation of femoral artery. A, Blood flow perfusion in Tie2-IκBαSR, shown as ratio of right to left hindlimb (R/L) at various time points after femoral artery ligation. B, Representative micro–computed tomography (micro-CT) angiograms of entire hindlimb 21 days after femoral artery ligation. C, Quantitative analysis of micro-CT angiograms presented as total number of vascular structures of specified diameters counted in thigh. D, Distribution of vascular structures below the knee. Representative 2-dimensional micro-CT images at 4 comparable levels in Tie2-IκBαSR transgenic mice compared with controls are shown. E, Quantitative analysis of micro-CT angiograms presented as total number of vascular structures of specified diameter counted in calf. F, Blood flow perfusion in VEC-IκBαSR transgenic mice as the ratio R/L hindlimb at various time points after femoral artery ligation. G, Representative micro-CT angiograms of entire hindlimb 21 days after femoral artery ligation. *P<0.05, ** P<0.01; n=3 to 8 mice per group. Levene's test for equality of variance was applied; it was found that measurements at days 7 and 14 had heterogeneous variances, and therefore the unequal-variance version of the t test was used (A).
To exclude the possibility that Tie-2 expression in cells other than endothelium was responsible for the observed phenotype, we used VEC promoter as a driver for IκBαSR expression. Similar to Tie2- IκBαSR mice, VEC-IκBαSR mice demonstrated a significant decline in blood flow recovery after common femoral artery ligation (Figure 1F), which was associated with an increase in the number of small arteries in the ischemic hindlimb (Figure 1G).
Arteriogenesis is thought to be driven to a large extent by accumulation of monocyte-derived macrophages around the remodeling vasculature. To explore the effect of shutdown of NF-κB signaling on this process, we examined accumulation of various mononuclear cell subsets in above- and below-the-knee tissues of Tie2-IκBαSR and control mice after induction of ischemia. Immunocytochemical analysis demonstrated extensive transgene expression in endothelial cells in tissue sections from ischemic gastrocnemius muscle, whereas its expression in the relatively nonischemic adductor muscle was quite low (Figure 2A and 2B). In agreement with the shutdown in endothelial NF-κB activation, there was a marked decrease in the number of CD45+ mononuclear cells in gastrocnemius of Tie2-IκBαSR compared with control mice (Figure 2C), whereas there were no differences in the presence of these cells in the adductor region (Figure 2D). To confirm that ischemic activation of Tie2-driven IκBαSR construct expression reduces equally the influx of various subtypes of blood-derived mononuclear cells, fluorescence-activated cell sorting was used to determine the presence of Cd11b, CD11c, and F4/80 macrophages. Although absolute numbers were markedly lower in tissues from Tie2-IκBαSR, the relative composition of the cellular subtypes was the same in both groups (Figure 2E).
Decreased accumulation of blood-derived mononuclear cells in Tie2–IκBαSR transgenic mice. A, Representative immunostaining of gastrocnemius muscle sections with antibodies against CD45, CD31, and HA-tag at day 3. DAPI indicates 4′,6-diamidino-2-phenylindole. B, Quantification of CD45+ cells in gastrocnemius muscle in Tie2-IκBαSR transgenic mice compared with controls. C, Representative immunostaining of adductor muscle sections with antibodies against CD45, CD31, and HA-tag at day 3. D, Quantification of CD45+ cells in adductor muscle in Tie2-IκBαSR transgenic mice compared with controls. E, Representative flow cytometry (forward scatter detector/side scatter detector) analysis of monocytes (CD11b+F4/80−), macrophages (CD11b+F4/80+), and dendritic cells (CD11b+CD11c+) in adductor and gastrocnemius muscles 3 days after femoral artery ligation.
To establish the molecular mechanism responsible for the marked increased in vascular branching in Tie2-IκBαSR mice, we next sought to examine key potential regulators. Western blotting of gastrocnemius tissues confirmed expression of IκBαSR construct in Tie2-IκBαSR mice (Figure 3A). This endothelial shutdown of NF-κB activation was associated with a marked reduction in tissue expression of HIF-1α-dependent genes VEGF-A and PDGF-BB and a moderate reduction in VEGFR2. Examination of the Notch signaling cascade demonstrated somewhat increased Jagged 1 and markedly decreased Dll4 and Notch intracellular domain levels, suggesting reduced Notch activation (Figure 3A). Because PDGF-BB is a key mediator of vascular smooth muscle coverage, we examined newly formed blood vessels from the ischemic gastrocnemius region in control and Tie2-IκBαSR mice. In agreement with the reduction in PDGF-BB expression, there was a significant decrease in pericyte coverage of the neovasculature of Tie2-IκBαSR mice (Figure 3B and 3C).
Molecular signature of increased vascular branching in Tie2–IκBαSR transgenic mice. A, Western blot analysis of IκBαSR, vascular endothelial growth factor-A (VEGFA), vascular endothelial growth factor receptor-2 (VEGFR2), VE-cadherin, platelet-derived growth factor-BB (PDGFBB), PDGF receptor-β (PDGFRb), Jagged-1, δ-like ligand 4 (DLL4), and Notch-1 intracellular domain (ICD) in gastrocnemius tissue lysates in Tie2-IκBαSR transgenic mice compared with controls 7 days after femoral artery ligation. B, Representative immunostaining of gastrocnemius muscle section with antibodies specific for CD31 and NG2 chondroitin sulfate proteoglycan at day 7. C, Mural cell coverage determined as ratio of NG2 to CD31 staining. *P<0.05.
Endothelial NF-κB Suppression Affects Delta-Notch Signaling
Reduction in Notch signaling suggested by decreased tissue Dll4 and Notch intracellular domain levels may well account for the increased vascular branching seen in Tie2-IκBαSR mice. To confirm the effect of a shutdown in NF-κB activity on Notch/Delta signaling in endothelial cells in vitro, Ad-IκBαSR or control viruses (Ad-Null) were expressed in HUVECs. Western blotting demonstrated a reduction in baseline -Dll4, Jagged-1, and Notch intracellular domain levels as well as a decrease in VEGFR2 expression (Figure 4A). To determine whether IκBαSR expression in HUVECs affects other angiogenic signaling, we assessed the expression of fibroblast growth factor receptor-1 and insulin-like growth factor-1 receptor in HUVECs transduced with Ad-IκBαSR. Western blotting demonstrated no changes in expression of either receptor (Figure 4B) after transduction with Ad-IκBαSR compared with Ad-GFP virus.
Nuclear factor-κB suppression in endothelial cells affects Delta-Notch signaling. A, Western blot analysis of inhibitor of κBα super repressor (IκBαSR), vascular cell adhesion molecule (VCAM), Jagged-1, δ-like ligand 4 (DLL4), cleaved Notch (Val 1744), vascular endothelial growth factor receptor-2 (VEGFR2), and platelet-derived growth factor-BB (PDGFBB) in human umbilical vein endothelial cells (HUVECs) transduced with Ad-IκBαSR or Ad-Null compared with control (untreated cells). B, Western blot analysis of fibroblast growth factor receptor-1 (FGFR1) and insulin-like growth factor-1 receptor-β (IGF-IRβ) in HUVECs transduced with Ad-IκBαSR or Ad-green fluorescent protein (GFP). C, Control and Ad-IκBαSR–transduced HUVECs were treated with vascular endothelial growth factor-A (VEGFA) (50 ng/mL) or tumor necrosis factor-α (TNFα) (10 ng/mL) for 24 hours. Western blot analysis of VCAM, Jagged-1, DLL4, and cleaved Notch-1 (Val 1744) is shown. D, Intercellular adhesion molecule-1 (ICAM-1) expression in HUVECs transduced with Ad-IκBαSR or Ad-GFP subjected to oscillatory shear stress for 18 hours. E, Western blot analysis of hypoxia-inducible factor-1α (HIF1α) and hypoxia-inducible factor-2α (HIF2α) expression in Ad-IκBαSR– or Ad-GFP–transduced cells subjected to 0.4% O2 (hypoxia) or 21% O2 (normoxia) for 24 hours.
Induction of arteriogenesis after arterial ligation in vivo is accompanied by an influx of mononuclear cells secreting, among other factors, VEGF and TNF-α. To evaluate the effect of NF-κB suppression on endothelial cell responses to these growth factors, control and Ad-IκBαSR–transduced HUVECs were treated with either VEGF-A or TNF-α. As expected, TNF-α induced vascular cell adhesion molecule-1, which was suppressed by IκBαSR expression. TNF-α also stimulated expression of Jagged-1 while suppressing Dll4. Both of these actions were suppressed by IκBαSR expression (Figure 4C). In agreement with previous observations, VEGF-A induced expression of Dll4, leading to Notch-1 cleavage. Remarkably, this action of VEGF turned out to be NF-κB dependent because IκBαSR fully blocked both an increase in Dll4 expression and Notch activation (Figure 4C).
Because NF-κB plays a central role in activation of adhesion molecule expression in response to shear stress,39,40 we next determined the effect of Ad- IκBαSR transduction on the expression of ICAM-1 in response to shear stress. As expected, the induction of ICAM-1 expression in response to oscillatory shear stress was impaired in HUVECs expressing IκBαSR (Figure 4D).
To investigate whether NF-κB suppression results in decreased hypoxia-driven induction of HIF levels, we assessed HIF-1α/-2α expression in Ad-IκBαSR-transduced HUVECs. Twenty-four hours of culture under hypoxic (0.4% O2) conditions resulted in a substantial increase in HIF-1α and HIF-2α levels in GFP-transduced HUVECs. This increase was substantially attenuated in Ad-IκBαSR–expressing cells (Figure 4E).
To further study the significance of shutdown of NF-κB–dependent Delta-Notch signaling, we examined retinal vasculature in newborn VEC-IκBαSR mice. Whole mounts of retinal vasculature on postnatal day 4 demonstrated excessive branching in VEC-IκBαSR mice, with an overall increase in vessel length and the number of vessel segments (Figure 5A and 5B). Lectin staining of postnatal day 4 retinal vasculature confirmed increased vessel sprouting (Figure 5B and 5C). Similar to the hindlimb vasculature, retinal vessels in VEC-IκBαSR mice demonstrated decreased pericyte coverage, as demonstrated by NG2 staining (Figure 5D). Staining with anti-Dll4 antibody demonstrated reduced Dll4 expression in tip cells in VEC-IκBαSR compared with control mice (Figure 5E), whereas there was no change in Jagged-1 expression (Figure 5F). To evaluate whether suppression of NF-κB signaling increases the number of preexisting collaterals, we examined the arteries and pial arteriolar-arteriolar collateral anastomoses between the middle cerebral artery and the anterior cerebral artery in control and VEC- IκBαSR mice 4 days after birth. Immunofluorescent staining demonstrated a significant increase in artery-to-artery connections in the pial circulation in VEC-IκBαSR mice (Figure 5G).
Retinal vasculature and pial collateral in VE-cadherin (VEC)–IκBαSR transgenic mice at postnatal day 4. A, Representative isolectin B4 (IB4) staining of whole-mounted retina. Retinal vasculature (B) and quantitative analysis (C) of vessel branch points, length, and segments in VEC-IκBαSR transgenic mice compared with controls are shown. D, Representative IB4 staining and immunostaining with anti-NG2 antibody of vascular front and mural cell coverage determined as ratio of NG2/-IB4 staining in VEC-IκBαSR transgenic mice compared with controls. E, Representative immunostaining of retinal vasculature with anti-δ-like ligand 4 (Dll4) antibody in VEC-IκBαSR transgenic mice compared with control mice. Note Dll4 expression in tip cells in controls (arrows) and reduced Dll4 expression in tip cells in VEC-IκBαSR mice. F, Representative immunostaining of retinal vasculature with anti-Jagged-1 antibody in VEC-IκBαSR transgenic mice compared with control mice. Note no difference in Jagged-1 expression in stalk cells between VEC-IκBαSR mice and controls. G, Representative immunostaining with anti-smooth muscle actin antibody and pial collateral quantification in VEC-IκBαSR transgenic mice compared with control mice. Note increased collateral formation in VEC-IκBαSR transgenic mice (arrows). *P<0.05, ** P<0.01; n=9 mice per group.
Taken together, these observations suggest that decreased Notch-1 activation due to reduced Dll4 expression is the cause of abnormally increased vascular density in these mice. To confirm this hypothesis, VEC-IκBαSR mice were injected with Jagged-1 or scrambled peptide shortly after birth, and the retinal vasculature was examined at postnatal day 4. Treatment with Jagged-1 but not scrambled peptide reduced retinal vessel branching to control levels (Figure 6A and 6B). As an additional confirmation, we examined endothelial cell sprouting in vitro. Transduction of HUVECs with an Ad-IκBαSR but not Ad-Null construct increased sprouting, which was reduced by Jagged-1 peptide treatment to the baseline level (Figure 6C and 6D).
Treatment with Jagged-1 reduces vessel branching in VE-cadherin (VEC)-IκBαSR transgenic mice and regulates vessel sprouting in human umbilical vein endothelial cells (HUVECs) expressing IκBαSR. Retinal vasculature (A) and quantitative analysis (B) of vessel branch points, length, and segments in VEC-IκBαSR transgenic mice treated with Jagged-1 or scrambled peptide compared with controls are shown. C, Representative Calcein AM staining of HUVEC sprouting. HUVECs were transduced with Ad-IκBαSR or Ad-Null and treated with Jagged-1 or scrambled peptide. D, Quantitative analysis of tube area. *P<0.05, **P<0.01; n=3 mice per group, n=5 sprouting assays per group.
Discussion
The present study demonstrates that suppression of endothelial NF-κB activation results in decreased HIF-1α expression, which is accompanied by decreased VEGF-A and PDGF-BB levels, and a reduction in Dll4 expression, which, in turn, leads to decreased Notch activation. In vitro, this results in increased sprouting, and in vivo, this results in markedly increased arterial density and a larger number of artery-to-artery connections both during development (pial collaterals and retinal vasculature) and in adult arteriogenesis settings (hindlimb ischemia model) despite markedly reduced levels of blood-derived tissue macrophages.
Thus formed, the arterial circulation is disorganized with multiple interconnections and poor mural cell coverage. Remarkably, despite a massive increase in arterial collateral density after common femoral artery ligation both in Tie2-IκBαSR and VEC-IκBαSR mice, the distal blood flow was profoundly impaired, leading to a poor recovery and extensive tissue injury. This increase in collateral density can be traced to decreased expression of Dll4 and subsequent reduced activation of Notch signaling because activation of Notch with systemically administered Jagged-1 peptide restores the arterial circulation to normal. Thus, NF-κB–dependent activation of Delta-4/Notch signaling and HIF-1α/2α expression is critical to proper formation of arterial circulation and arterial collaterals.
Inhibition of endothelial NF-κB signaling was achieved with the use of an IκBα mutant construct (IκBαSR) that has been validated previously.41 For the hindlimb arteriogenesis studies, we chose a Tie2 promoter construct that is expressed predominantly under ischemic conditions and is not expressed during development.33 This has allowed us to compare the extent of arteriogenesis and collateral formation after common femoral artery ligation without worrying about different levels of preexisting collaterals. Because Tie2 is expressed to some extent in certain mononuclear cell lineages and to examine arteriogenesis and collateral formation during development, we also studied an IκBαSR mouse line under the control of VEC promoter that is much more endothelially restricted. In both cases, inhibition of endothelial NF-κB activation led to increased arterial branching (retina, hindlimb vasculature) and collateral formation (pial circulation, hindlimb vasculature) and similar effects on endothelial gene expression in vitro.
As expected, the dominant-negative IκBαSR construct fully suppressed NF-κB activation in the endothelium, as demonstrated, for example, by decreased expression of vascular cell adhesion molecule and ICAM-1. This decrease in adhesion receptor expression likely explains a marked decrease in the influx of monocytes into the tissue surrounding the ligated common femoral artery, as demonstrated by both histological analysis and fluorescence-activated cell sorting analysis of tissue extracts. Among the monocyte subtypes examined, all were decreased equally, suggesting that reduced endothelial adhesion was the primary cause. In agreement with prior studies, this suppression of NF-κB activation resulted in decreased stabilization of HIF-1α26 and HIF-2α42 and reduced expression of HIF-dependent genes VEGF-A and PDGF-BB.
The unexpected increase in arterial branching after suppression of endothelial NF-κB activation appears to be due to decreased Dll4 expression and a consequent reduction in Notch activation. Dll4 levels were dramatically reduced in the ischemic hindlimb of Tie2-IκBαSR mice as well as in the developing retinal vasculature of VEC-IκBαSR mice. This decrease in Dll4 levels is a direct consequence of suppressed NF-κB activation because transduction of HUVECs with Ad-IκBαSR in HUVECs results in a decrease in basal as well as VEGF-induced Dll4 expression. Finally, the increased arterial branching observed in these settings could be reversed by activation of Notch by Jag-1 peptide. These findings are similar to the recently reported endothelial deletion of HIF-2α, which also results in decreased expression of Dll4, poor recovery from common femoral artery ligation, and excessive arterial branching.42
Similar to Dll4, Jagged-1 expression is also directly NF-κB dependent both in terms of its baseline level and in terms of the TNF-α–dependent induction of expression. However, whereas in the case of Dll4, NF-κB was required for VEGF-driven induction of expression, in the case of -Jagged 1, TNF-α was the major driver. Ischemic tissues in Tie2-IκBαSR mice had a higher expression of -Jagged 1 than controls, likely reflecting more severe tissue damage. This excess of -Jagged 1 expression was not sufficient to compensate for reduced Dll4 expression in terms of the suppression of excessive endothelial Notch signaling.
In addition to excessive branching, newly formed vasculature in our study appeared immature, with poor pericyte coverage. The latter can be attributed to decreased HIF-dependent endothelial secretion of PDGF-BB, the main pericyte attractor. In addition, Notch signaling has been reported to regulate PDGF-β receptor expression in smooth muscle cells.43 Decreased Notch activation in these mice may have made developing vasculature less sensitive to PDGF stimulation because its levels were also reduced.
This link between NF-κB activation and Dll4 expression has not, to our knowledge, been reported. An indirect support for this concept is the observation of increased mammary duct branching in IκBα knockout mice.44 Little is known about regulation of Dll4 expression, and the mechanism of NF-κB–dependent regulation of its activity is not certain. The effect can be direct, given a RelA binding site in the Dll4 promoter.
An increase in tumor vascularization due to increased sprouting and numerous vascular interconnections has been observed after suppression of Dll4 activity or expression. At the same time, tumors themselves appeared more ischemic and poorly perfused, and tumor growth was actually decreased.45 In addition, similar to findings in the present study, the excessive vasculature forming after the loss of a Dll4 allele or inhibition of its signaling showed decreased maturation and markedly decreased pericyte coverage.46 Furthermore, inhibition of Notch signaling with Ad-driven soluble Dll4 in the setting of hindlimb ischemia in mice decreased blood flow recovery and led to chaotic capillary sprouting, resulting in disorganized, low-perfusion vascular networks.16
These results place endothelial NF-κB squarely at the center of both developmental and adult arteriogenesis and provide new insights into its biology (Figure 7). Although numerous factors, including TNF-α, inflammatory cytokines, and lipopolysaccharide, among others, can activate NF-κB, the key activator involved in arteriogenic settings is probably shear stress. Increased shear stress would certainly be expected after common femoral artery ligation, and blood flow in the developing retina may be sufficient to activate NF-κB in that setting as well. Once activated, NF-κB directly regulates expression of adhesion receptors, thereby leading to tissue accumulation of monocytes and stabilization of HIF-1α and HIF-2α, which in turn activates expression of VEGF-A and PDGF-BB. Increased local VEGF levels further recruit blood-derived monocytes via stromal cell–derived factor-1,31 which in turn contribute to tissue VEGF levels, thereby setting up a positive feedback loop. This leads to active remodeling of capillaries into the new arterial circulation,47 whereas PDGF-BB–driven pericyte recruitment leads to its maturation. In parallel, NF-κB activates DLL4 expression, leading to activation of Delta/Notch signaling and effective control of branching of the forming network.
Schema of nuclear factor (NF)-κB role in arteriogenesis and collateral formation. NF-κB activation leads to stabilization of hypoxia-inducible factor (HIF)-1α expression with a subsequent induction of vascular endothelial growth factor (VEGF)-A and platelet-derived growth factor (PDGF)-BB expression, capillary proliferation, and maturation into a new arterial collateral network. The size of the network is controlled by NF-κB–dependent activation of δ-like ligand 4 (DLL4) expression and Notch signaling. In addition, NF-κB induces monocyte influx that further promotes collateral growth. TNFα indicates tumor necrosis factor-α.
All 3 process regulated by NF-κB are crucial to effective arteriogenesis, and all 3 need to operate in tandem to ensure formation of the functional vasculature. Lack of shear stress sensing, impaired monocyte recruitment, and lack of monocytes themselves have all been linked to impaired arteriogenesis.8,48 Similarly, HIF-dependent factors VEGF-A and PDGF-BB are critical to arteriogenesis21,22 and vascular maturation, respectively.49 Finally, Dll4 signaling input is required for proper trimming of the vascular network, thereby preventing excessive interconnections that impair effective distal perfusion.45
Another important finding is that excessive arteriogenesis leading to formation of excessive collateral network is not always a good thing. Our findings in this regard confirm observations in cancer studies and emphasize the need for balance in induction of collateral growth. They also suggest that manipulation of Delta/Notch signaling may not be the best way to induce therapeutic arteriogenesis.
In summary, we find that endothelial NF-κB signaling plays a key role in regulation of adult and developmental arteriogenesis and formation of collateral circulation by regulation of monocyte entry, production of VEGF and PDGF, and, via Dll4, determination of the size and complexity of the collateral network.
Sources of Funding
This study was supported in part by National Institutes of Health grant HL84619 (Dr Simons), Leducq Foundation ARTEMIS Transatlantic network grant (Drs Eichmann and Simons), and American Heart Association Scientist Development Grant 0635107N (Dr Tirziu).
Disclosures
None.
Acknowledgments
We thank Karen Moodie (Dartmouth Medical School) and Jiasheng Zhang (Yale University) for their expert surgical assistance and Jose R. Conejo-Garcia for his help with tissue macrophage fluorescence-activated cell sorting.
- Received May 21, 2012.
- Accepted October 10, 2012.
- © 2012 American Heart Association, Inc.
References
Clinical Perspective
Collateral circulation plays a protective role in case of an ischemic insult precipitated by the occlusion of an arterial vessel. Despite their presumed physiological and clinical importance, little is known about the manner in which collaterals develop. A thorough understanding of this process is important to any attempts at therapeutic arteriogenesis. Collaterals are thought to arise either by expansion from the preexisting vessels or by de novo growth. In either case, the key stimulus is probably mechanical in nature, such as shear stress, and not tissue ischemia per se. One well-understood signaling mechanism that controls the size and density of a vascular network is the Delta/Notch pathway. A number of studies have demonstrated that activation of Notch-1 by its ligand δ-like ligand 4 (Dll4) leads to a decrease in vascular branching. However, the link between Dll4/Notch signaling and shear stress has not been established. Endothelial shear stress activates the transcription factor nuclear factor-κB, which controls expression of a broad array of genes involved in cell adhesion and migration. In this study, we unexpectedly show that Dll4 expression is critically dependent on nuclear factor-κB, likely via nuclear factor-κB–dependent stabilization of hypoxia-inducible factor-1 and -2 genes. In the absence of endothelial nuclear factor-κB activation, reduced Dll4 levels lead to a much greater expansion in collateral growth than seen under normal circumstances. Paradoxically, that leads to a severe impairment in tissue blood flow. Thus, although increasing collateral density is a worthy therapeutic goal, the quality of the collateral network is just as important as its size.
This Issue
Jump to
Article Tools
Share this Article
- Endothelial Nuclear Factor-κB–Dependent Regulation of Arteriogenesis and BranchingClinical Perspective
