TNF-α Dilates Cerebral Arteries via NAD(P)H Oxidase-Dependent Ca²⁺ Spark Activation

Abstract

Expression of Tumor Necrosis Factor alpha (TNF-α), a pleiotropic cytokine, is elevated during stroke and cerebral ischemia. TNF-α regulates arterial diameter, although mechanisms mediating this effect are unclear. Here, we tested the hypothesis that TNF-α regulates the diameter of resistance-size (∼150 μm diameter) cerebral arteries by modulating local and global intracellular calcium (Ca²⁺) signals in smooth muscle cells. Laser-scanning confocal imaging revealed that TNF-α increased Ca²⁺ spark and Ca²⁺ wave frequency, but reduced global Ca²⁺ concentration ([Ca²⁺]_i) in smooth muscle cells of intact arteries. TNF-α elevated reactive oxygen species (ROS) in smooth muscle cells of intact arteries and this was prevented by apocynin or diphenyleneiodonium (DPI), NAD(P)H oxidase blockers, but was unaffected by inhibitors of other ROS generating enzymes. In voltage-clamped (-40 mV) cells, TNF-α increased the frequency and amplitude of Ca²⁺ spark-induced large-conductance Ca²⁺-activated potassium (K_Ca) channel transients ∼1.7-fold and ∼1.4-fold, respectively. TNF-α-induced transient K_Ca current activation was reversed by apocynin or MnTMPyP, a membrane-permeant antioxidant, and prevented by intracellular dialysis of catalase. TNF-α induced reversible and similar amplitude dilations in either endothelium-intact or-denuded pressurized (60 mm Hg) cerebral arteries. MnTMPyP, thapsigargin, a sarcoplasmic reticulum Ca²⁺ ATP-ase blocker that inhibits Ca²⁺ sparks, and iberiotoxin, a K_Ca channel blocker, reduced TNF-α-induced vasodilations to between 15 and 33% of control. In summary, data indicate that TNF-α activates NAD(P)H oxidase, resulting in an increase in intracellular H₂O₂ that stimulates Ca²⁺ sparks and transient K_Ca currents, leading to a reduction in global [Ca²⁺]_i, and vasodilation.

INTRODUCTION

Stroke and transient brain ischemia are vascular-born brain pathologies accompanied by inflammatory changes in blood vessels and parenchyma tissues (15; 46). Tumor Necrosis Factor-alpha (TNF-α), a pleiotropic cytokine, is a pro-inflammatory agent generated by glial cells, neurons, macrophages and vascular endothelium during brain injury. Clinical data demonstrate a correlation between TNF-α levels and the severity of ischemic brain damage, but controversy exists regarding the functional role of TNF-α, with both beneficial and detrimental roles proposed (3; 8; 9; 15; 36; 46). A chronic TNF-α elevation is associated with risk factors for other vascular pathologies, including coronary artery disease and atherosclerosis (26). During stroke, TNF-α may cause neuronal damage directly and by accelerating ischemia-induced local inflammatory responses. In contrast, studies also indicate that TNF-α may be cytoprotective (8; 15; 36; 46).

TNF-α acts in a paracrine manner by binding to two membrane receptors, p55 and p75. Besides its actions on neurons, astrocytes and microglia, TNF-α also regulates arterial contractility (15), an effect that may modulate injury by altering local blood supply. However, the regulation of arterial diameter and blood flow by TNF-α is complex. In vivo, intracranial injection of TNF-α constricted pial arterioles and reduced cerebral blood flow (29; 38; 44). TNF-α also induced pro-contractile effects in coronary arteries (23; 51). In contrast, TNF-α dilated cerebral, mesenteric and cremaster muscle arterioles, and relaxed endothelium-denuded aorta (4; 7; 16; 22; 32; 37). In bronchial arteries, TNF-α initially induced dilation followed by constriction 2 hours later (45). TNF-α-mediated vasoregulation can also occur through both endothelium-dependent and endothelium-independent mechanisms (4; 7; 16; 22; 23; 29; 32; 37; 38; 44; 45; 51). Collectively, data suggest TNF-α can induce both vasoconstriction and vasodilation, and effects may be time-dependent. Regardless, mechanisms mediating vasoregulation by TNF-α are poorly understood.

A major regulator of arterial diameter is the smooth muscle cell global intracellular Ca²⁺ concentration ([Ca²⁺]_i). In smooth muscle cells, global [Ca²⁺]_i is regulated by extracellular Ca²⁺ influx and intracellular Ca²⁺ release (19). An elevation in global [Ca²⁺]_i induces constriction, whereas a decrease in global [Ca²⁺]_i leads to vasodilation. Another Ca²⁺ signal, termed a “Ca²⁺ spark ”, is a rapid, localized intracellular Ca²⁺ transient that occurs due the activation of several ryanodine-sensitive Ca²⁺ release (RyR) channels on the sarcoplasmic reticulum (SR) (19; 33). In arterial smooth muscle cells, Ca²⁺ sparks occur close to the plasma membrane and activate several large-conductance Ca²⁺-activated K⁺ (K_Ca) channels, resulting in a transient K_Ca current. The resulting membrane hyperpolarization reduces voltage-dependent Ca²⁺ channel activity, leading to a decrease in Ca²⁺ entry, a reduction in global [Ca²⁺]_i and vasodilation (19; 33). In contrast, Ca²⁺ spark inhibition leads to vasoconstriction due to depolarization-induced voltage-dependent Ca²⁺ channel activation (19). Alterations in Ca²⁺ sparks and effective coupling to K_Ca channels have been implicated in hypertension and diabetes (1; 6; 19; 31). Another intracellular Ca²⁺ signal, termed a “Ca²⁺ wave ”, is a propagating [Ca²⁺]_i elevation that occurs due to the activation of SR RyR channels and/or inositol-trisphosphate-gated Ca²⁺ channels. Although the physiological functions of Ca²⁺ waves are less well described, when stimulated, Ca²⁺ waves contribute Ca²⁺ for contraction (35).

The goal of this study was to investigate whether TNF-α regulates cerebral artery diameter by modulating smooth muscle Ca²⁺ signaling and if so, to determine the underlying mechanisms. Such information would improve our understanding of cytokine-mediated changes in arterial diameter during vascular pathologies, including stroke.

METHODS

Tissue preparation and single cell isolation

Sprague-Dawley rats (∼250 g) of either sex were euthanized by peritoneal injection of a sodium pentobarbital overdose (150 mg kg^-1) in accordance with the Animal Care and Use Committee policies and procedures at the University of Tennessee. The brain was removed and placed into ice-cold (4 °C), oxygenated (95 % O₂-5 % CO₂), physiological saline solution (PSS) containing (mM): 119 NaCl, 4.7 KCl, 24 NaHCO₃, 1.2 KH₂PO₄, 1.6 CaCl₂, 1.2 MgSO₄, 0.023 EDTA and 11 glucose. Posterior cerebral, middle-cerebral and cerebellar arteries (50-200 μm diameter) were removed, cleaned of connective tissue and maintained in ice-cold PSS. Where appropriate, the endothelium was removed by introducing an air bubble into the artery lumen for 2 minutes followed by a 30 s wash with H₂O. Individual smooth muscle cells were dissociated from cerebral arteries using enzymes and an isolation solution of the following composition (in mM): 55 NaCl, 80 sodium glutamate, 5.6 KCl, 2 MgCl₂, 10 HEPES, and 10 glucose (pH 7.3 with NaOH), a previously described (18). Cells were stored on ice and used for experiments between 1 and 6 hours after isolation.

Laser scanning confocal Ca²⁺ imaging

Artery segments (1-2 mm in length) were placed into HEPES-buffered PSS of composition (mM): 134 NaCl, 6 KCl, 2 CaCl₂, 1 MgCl₂, 10 HEPES and 10 glucose (pH 7.4, NaOH) containing fluo-4 AM (10 μM) and pluronic F-127 (0.05 %) for 60 min at 22 °C. To allow indicator de-esterification, arteries were subsequently placed into HEPES-buffered PSS for 30 min at 22 °C. Smooth muscle cells within the artery wall were imaged using a Noran Oz laser scanning confocal microscope (Noran Instruments, Middleton, WI, USA) and a ×60 water immersion objective (NA = 1.2) attached to a Nikon TE300 microscope. Cells were illuminated with a krypton-argon laser at 488 nm and emitted light greater than 500 nm was collected. Planar images (56.3 μm × 52.8 μm) were recorded every 16.7 ms, i.e. 60 images s^-1. To compare confocal Ca²⁺ imaging data with electrophysiological recordings performed in this study, Ca²⁺ sparks in smooth muscle cells of artery segments were measured in an extracellular solution containing 30 mM K⁺, which depolarizes smooth muscle cells to ∼-40 mV. This membrane potential is similar to that of cerebral arteries pressurized to 60 mmHg (see (20) for similar procedure). The 30 mM K⁺ bath solution contained (mM): NaCl 110; KCl 30; HEPES 10; CaCl₂ 2; MgCl₂ 1; and glucose 10 (pH 7.4, NaOH). At least two different representative areas of the each artery were scanned for at least 10 s under each condition. The same area of artery was scanned only once to avoid any laser-induced changes in Ca²⁺ signaling and the effects of drugs were measured in paired experiments. Ca²⁺ sparks were detected in smooth muscle cells using custom analysis software written using IDL 5.2 (Research Systems Inc., Boulder, CO, USA) kindly provided by Drs M. T. Nelson and A. D. Bonev (University of Vermont, VT, USA). Automated and manual detection of Ca²⁺ sparks were performed by dividing an area 1.54 μm (7 pixels) × 1.54 μm (7 pixels) (i.e. 2.37 μm²) in each image (F) by a baseline (F₀) that was determined by averaging 10 images without Ca²⁺ spark activity. The entire area of each image was analyzed to detect Ca²⁺ sparks. A Ca²⁺ spark was defined as a localized increase in F/F₀ that was greater than 1.2 (10). Artery Ca²⁺ spark frequency (Hz) was calculated by averaging values from at least two different areas. Ca²⁺ waves were defined as an elevation in F/F₀ greater than 1.2 that propagated for at least 20 μm. Global Ca²⁺ fluorescence was calculated from the same images used for Ca²⁺ spark analysis and was the mean pixel value of 100 different images acquired over 10 s. To calculate whether TNF-α altered global Ca²⁺, global Ca²⁺ fluorescence in TNF-α was divided by the corresponding control value.

Patch-clamp electrophysiology

K⁺ currents were measured using either the conventional whole-cell or perforated-configuration of the patch-clamp technique with an Axopatch 200B amplifier (Axon Instruments, Union City, CA, USA). Bath solution was 6 mM K⁺ HEPES-buffered PSS (composition described above). For perforated-patch experiments, the pipette solution contained (mM): 110 potassium aspartate, 30 KCl, 10 NaCl, 1 MgCl₂, 10 HEPES, 0.05 EGTA (pH 7.2 with KOH). For conventional whole-cell experiments, the pipette solution contained (mM): 140 KCl, 1.9 MgCl₂, 0.037 CaCl₂, 0.1 EGTA, 10 HEPES, 2 Na₂ATP (pH 7.2 with KOH); the calculated free Ca²⁺ and free Mg²⁺ concentrations of this solution are 100 nM and 1 mM, respectively (WEBMAXC, Stanford University, CA, USA). In some experiments, catalase (200 U/ml) was included in the whole-cell pipette solution. Where appropriate, catalase was heat-inactivated by incubation at 92 °C for 25 min. TNF-α was applied to the bath solution 10 minutes after formation of the whole-cell configuration, to ensure exchange of the pipette solution with the cytosol. All experiments were performed using a holding potential of-40 mV. Membrane currents were recorded with a sample rate of 2.5 kHz and filtered at 1 kHz. Transient K_Ca current analysis was performed off-line. A transient K_Ca current was defined as the simultaneous opening of three or more K_Ca channels.

Reactive Oxygen Species (ROS) measurements

1) Dichlorofluorescein (DCF) imaging

ROS were measured in isolated intact artery segments using H₂DCFDA (2`7`-dichlorodihydrofluorescein diacetate). Isolated endothelium-denuded cerebral artery segments (1-2 mm) were incubated for 45 min at room temperature in 10μM H₂DCFDA. H₂DCFDA is hydrolyzed by intracellular esterases to the nonfluorescent derivative H₂DCF. ROS oxidize H₂DCF to fluorescent DCF. DCF fluorescence was excited with 488 nm light and emitted light greater than 525 nm was collected. Planar images were acquired using a Zeiss LSM 5 laser-scanning confocal microscope. Images were recorded 10 minutes after TNF-α or H₂O₂ application. Time controls were also performed, wherein agents were not applied. Since DCF fluorescence increases spontaneously upon exposure to excitation light (12), artery regions were imaged only once. Images were background corrected. ImageJ software (NIH) was used for analysis.

2) Dihydroethidium (DHE) imaging

Cerebral artery segments were incubated for 20 min in HEPES-buffered PSS without (time-matched control) or with TNF-α. Arteries were then embedded in OTC and snap frozen in liquid N₂. 30 μm thick transverse sections were cut, mounted on glass slides, and incubated with DHE (10 μM) in a 100 % N₂ atmosphere in the dark for 30 min at 37°C. Arterial sections were excited with 488 nm light and emitted light >590 was collected using a Noran Oz laser-scanning confocal microscope. Custom written software was used to measure DHE fluorescence intensity. Fluorescence intensity in diazoxide was divided by that in time-matched controls and multiplied by 100 to determine effects of diazoxide on DHE intensity. Tissue processing did not elevate ROS, because DHE fluorescence intensity of arteries pre-incubated with MnTMPyP were 1.04 ± 0.06 of those that were untreated (P>0.05, n=5).

Pressurized artery diameter measurements

Cerebral artery segments (100-200 μm diameter) were placed in a temperature-controlled perfusion chamber (Living Systems, Vermont) and cannulated with glass pipettes at each end. The chamber was continuously perfused with physiological saline solution equilibrated with 5% CO₂, 21% O₂ and 74% N₂ to pH 7.4 and maintained at 37°C. Arteries were observed with a charge-coupled device (CCD) camera attached to an inverted microscope (Nikon TS 100-F). Artery diameter was monitored using edge-detection software (IonOptix; Milton, MA). Artery intraluminal pressure was elevated to 60 mmHg, which resulted in the development of myogenic tone. After diameter stabilization, tested agents were applied by chamber perfusion. Artery diameter was digitized at 1 Hz using the edge-detection function of Ionwizard (Ionoptix). Endothelium-denudation was confirmed by the absence of a dilatory response to carbachol (10^-5 M), an endothelium-dependent cerebral vasodilator. At the end of each experiment, passive artery diameter was determined using a Ca²⁺ -free bath solution containing 1 mM EGTA.

Data analysis

Values are expressed as mean ± standard error of the mean (SEM). Students t-test was used for comparing paired and unpaired data from two populations, and ANOVA and Student-Newman-Keuls test were used for multiple group comparisons. Evaluation of whether distributions were Gaussian was by the method of Kolmogorov and Smirnov. P<0.05 was considered significant.

Chemicals

Unless stated otherwise, all chemicals used in this study were obtained from Sigma or Calbiochem (USA). Papain was purchased from Worthington Biochemical Co. (Lakewood, NJ), fluo-4 AM and H₂DCFDA from Molecular Probes (Eugene, OR) and rat recombinant TNF-α from BD Pharmingen (CA).

RESULTS

TNF-α activates Ca²⁺ sparks and Ca²⁺ waves and reduces global [Ca²⁺]_i in smooth muscle cells of intact cerebral arteries

The regulation of local and global Ca²⁺ signals in smooth muscle cells by TNF-α was measured in cerebral artery segments. TNF-α (50 ng/ml) increased mean Ca²⁺ spark frequency from 1.53 ± 0.16 Hz to 2.6 ± 0.51 Hz, or ∼1.7 -fold (Fig. 1A, B). TNF-α did not change Ca²⁺ spark amplitude (F/F₀; control, 1.36 ± 0.01, n=184 sparks; TNF-α, 1.34 ± 0.01, n=329 sparks; P>0.05). In the same cells, TNF-α also elevated mean Ca²⁺ wave frequency from 0.32 ± 0.02 Hz to 0.48 ± 0.03 Hz, or ∼1.5 -fold (Fig. 1C). In contrast, TNF-α decreased mean global [Ca²⁺]_i to 0.85 ± 0.02 of control (n=6 arteries, p<0.05). These experiments suggest TNF-α elevates Ca²⁺ spark and wave frequency and reduces global [Ca²⁺]_i in cerebral artery smooth muscle cells.

Fig. 1.

TNF-α activates Ca²⁺ sparks and Ca²⁺ waves in smooth muscle cells of intact cerebral arteries. A, average fluorescence (100 of 600 images) over 10 s of two different areas of the same cerebral artery in control and TNF-α. The locations of Ca²⁺ sparks that occurred during 10 s are indicated by white boxes (1.54 μm×1.54 μm) and representative localized F/F₀ changes with time are illustrated below respective images and labeled accordingly. In control, 8 sparks were identified in traces a and b. In TNF-α, 11 sparks and 2 waves were detected in traces a-d. B, Mean effects of TNF-α (50 ng/ml) on Ca²⁺ spark frequency (n=6 arteries). C, Mean effects on Ca²⁺ wave frequency. ^* P<0.05, when compared with control.

TNF-α activates NAD(P)H oxidase leading to a ROS elevation in smooth muscle cells of intact cerebral arteries

TNF-α activates ROS generation in smooth muscle cells (13). Since RyR channels are redox-sensitive (42), we tested the hypothesis that TNF-α regulates local Ca²⁺ signaling by elevating intracellular ROS. Several enzymes generate ROS including NAD(P)H oxidase, xanthine oxidase, cytochrome P450, and mitochondrial electron transport chain complexes (13; 28; 41; 49). The DCF fluorescence method was used to determine pathways by which TNF-α elevates ROS in smooth muscle cells of small cerebral arteries. TNF-α increased DCF fluorescence intensity 2.7 ± 0.32-fold, and DHE intensity 1.51 ± 0.03-fold, in smooth muscle cells of intact arteries (Fig. 2A-D). Over the same time period, H₂O₂ elevated DCF fluorescence 4.7 ± 0.68-fold, whereas in control (absence of applied agents) DCF fluorescence did not change (Fig. 2B). In the presence of apocynin or diphenyleneiodonium (DPI), NAD(P)H oxidase blockers, TNF-α did not elevate DCF fluorescence (Fig. 2B). MnTMPyP, a membrane-permeant superoxide and catalase mimetic, also prevented the TNF-α-induced DCF fluorescence elevation (Fig. 2B). Apocynin (1.07 ± 0.23 of control, n=5), DPI (0.85 ± 0.1, n=4), and MnTMPyP (1.14 ± 0.17, n=4) did not change ROS when applied alone (P>0.05 for each). Oxypurinol, a xanthine oxidase inhibitor, 17-octadecanoic acid, a cytochrome P-450 blocker, or lonidamine, a mitochondrial permeability transition pore (PTP) opener did not reduce TNF-α-induced DCF fluorescence elevations (Fig. 2B). Oxypurinol (0.98 ± 0.13 of control, n=4), 17-octadecanoic acid (1.09 ± 0.12, n=5), and lonidamine (0.86 ± 0.13, n=4) did not change DCF fluorescence when applied alone (P>0.05 for each). Data indicate that TNF-α activates NAD(P)H oxidase, leading to a ROS elevation in cerebral artery smooth muscle cells.

Fig. 2.

TNF-α elevates ROS in smooth muscle cells of intact cerebral arteries via NAD(P)H oxidase activation. A, representative confocal images of DCF fluorescence in two different areas of the same cerebral artery in control (left) and after TNF-α (50 ng/ml) (right). B, mean effects of TNF-α on DCF fluorescence. TNF-α (50 ng/ml, n= 10 arteries), H₂O₂ (100 μM, n=6), TNF-α + apocynin (Apo, 25 μM, n=10), TNF-α + DPI (10 μM, n=7), TNF-α + MnTMPyP, (MnT, 10 μM, n=10), TNF-α + oxypurinol (Oxy, 10 μM, n=6), TNF-α + 17-octadecanoic acid (Oda, 10 μM, n=10), TNF-α + lonidamine (Lon, 100 μM, n=5), Time control (Cntrl, n=5). C, TNF-α (50 ng/ml) elevated DHE fluorescence in intact cerebral arteries. D, Average effects of TNF-α (50 ng/ml) on DHE fluorescence (n=5). ^* P<0.05 when compared with control. # P<0.05 when compared with TNF-α.

TNF-α activates transient K_Ca currents in isolated arterial smooth muscle cells

In arterial smooth muscle cells, a Ca²⁺ spark activates several K_Ca channels, resulting in a transient K_Ca current that induces membrane hyperpolarization (19). To investigate the effects of TNF-α-induced changes in Ca²⁺ signaling, transient K_Ca current regulation by TNF-α was measured in isolated, voltage-clamped (-40 mV) arterial smooth muscle cells. When using the perforated-patch clamp configuration, TNF-α evoked a gradual increase in transient K_Ca current frequency and amplitude, that reached a steady-state elevation of ∼1.7-fold and ∼1.4-fold, respectively in ∼10 minutes (Fig. 3A, C, D). Similar observations were made when using the conventional whole-cell configuration (Fig. 3B-D). Apocynin did not alter transient K_Ca current frequency (0.89 ± 0.12 of control, n=6) or amplitude (1.0 ± 0.06, n=6) when applied alone (P>0.05 for each), but reversed the TNF-α-induced increase in transient K_Ca current frequency and amplitude (Fig. 3B-D). Data suggest that TNF-α stimulates transient K_Ca currents by activating NAD(P)H oxidase.

Fig. 3.

TNF-α stimulates transient K_Ca currents in isolated cerebral artery smooth muscle cells due to NAD(P)H oxidase activation. A, B, original traces illustrating transient K_Ca currents in cells voltage-clamped at -40 mV. A, TNF-α activated transient K_Ca currents when using the perforated-patch clamp configuration. B, TNF-α activated transient K_Ca currents when using the conventional-whole cell configurations, and this was reversed by apocynin. C, D, Mean effects on transient K_Ca current frequency and amplitude of: TNF-α (50 ng/ml) perforated-patch (p-p, n=6); TNF-α (50 ng/ml) whole-cell (w-c, n=6); TNF-α + apocynin (Apo, 25 μM, n=6), TNF-α washout (Wash, n=6); ceramide (Cer, 10 μM, n=6), atractyloside (Atr, 100 μM, n=4). ^* P<0.05 when compared with control. # P<0.05 when compared with TNF-α.

TNF-α-induced intracellular signaling may involve mitochondrial PTP activation and ceramide generation (14; 22; 28; 30; 39; 41; 52). However, ceramide did not alter transient K_Ca currents, indicating that it is unlikely to mediate TNF-α-induced activation of these events (Fig. 3C, D). Lonidamine decreased mean Ca²⁺ spark frequency from 1.45 ± 0.16 Hz to 0.73 ± 0.18 Hz, or to ∼0.5 of control (n=4, P<0.05), and inhibited transient K_Ca currents (10). Similarly, atractyloside, another mitochondrial PTP opener, decreased transient K_Ca current frequency and amplitude to ∼0.4 -fold and ∼0.6 of control, respectively (Fig. 3C, D). Data demonstrate that PTP openers inhibit transient K_Ca currents, indicating that such an effect cannot mediate TNF-α-induced transient K_Ca current activation.

To determine if TNF-α-induced transient K_Ca current activation occurs due to a ROS elevation, transient K_Ca current regulation by MnTMPyP, and by catalase, that converts H₂O₂ to H₂O, was measured. MnTMPyP reversed TNF-α-induced elevations in transient K_Ca current frequency and amplitude (Fig. 4A-C). In contrast, MnTMPyP did not change transient K_Ca current frequency (1.0 ± 0.1 of control, n=5) or amplitude (1.0 ± 0.07) when applied alone (P>0.05 for each). Similarly, when using the conventional whole-cell configuration, inclusion of catalase in the pipette solution (to allow intracellular access), prevented TNF-α-induced transient K_Ca current activation (Fig. 4B, C). In contrast, when supplementing the pipette solution with heat-inactivated catalase TNF-α increased transient K_Ca current frequency and amplitude ∼1.8 and ∼1.4-fold, respectively (Fig. 4B, C). In summary, data indicate that TNF-α activates Ca²⁺ sparks and transient K_Ca currents by inducing an NAD(P)H oxidase-mediated ROS elevation.

Fig. 4.

TNF-α induced transient K_Ca current activation occurs due a ROS elevation. A, TNF-α (50 ng/ml)-induced transient K_Ca current activation was reversed by MnTMPyP. B, C, Mean transient K_Ca current frequency and amplitude. TNF-α (50 ng/ml, n=12), TNF-α + MnTMPyP (10 μM, n=6), TNF-α with catalase (Cat, 200 U/ml, n=6), TNF-α with boiled catalase (Boil. catalase, 200 U/ml, n=6). ^* P<0.05 when compared with control. # P<0.05 when compared with TNF-α.

TNF-α dilates pressurized cerebral arteries by elevating ROS, SR Ca²⁺ release and K_Ca channel activity

TNF-α has been demonstrated to induce both vasoconstriction and vasodilation by endothelial-dependent and -independent pathways (4; 7; 16; 22; 23; 29; 32; 37; 38; 44; 51). Here, data suggest TNF-α activates a dilatory Ca²⁺ signaling pathway in smooth muscle cells of small, resistance size cerebral arteries. Therefore, vasoregulation by TNF-α was investigated using resistance-size pressurized (60 mm Hg) cerebral arteries. In endothelium-intact arteries, from a mean active diameter of 142 ± 3 μm (0.64 ± 0.01 of passive diameter) TNF-α induced a mean dilation of 22 ± 1 μm (Fig. 5A-C). In endothelium-denuded arteries, TNF-α increased mean diameter by 18 ± 3 μm from 130 ± 9 μm (0.57 ± 0.02 of passive diameter, Fig. 5C). TNF-α-induced dilations were not significantly different in endothelium-intact or endothelium-denuded arteries (P>0.05). TNF-α-induced dilations were fully reversible (Fig. 5A) and consecutive TNF-α applications resulted in dilations of similar magnitude (first application, 19 ± 2 μm; second application 21 ± 2 μm (n=2).

Fig. 5.

TNF-α dilates pressurized (60 mm Hg) cerebral arteries by elevating ROS and by activating SR Ca²⁺ release and K_Ca channels. A, TNF-α reversibly dilated a pressurized (60 mmHg) endothelium-intact cerebral artery. B, TNF-α (10 ng/ml)-induced vasodilation was inhibited by ibertiotoxin (IbTX, 100 nM). C, mean TNF-α (10 ng/ml)-induced dilations in endothelium-intact arteries in control (Cntrl, n=29), MnTMPyP (MnT, 10 μM, n=4), thapsigargin (+Tg., 100 nM, n=7), and iberiotoxin (+IbTX, 100 nM, n=9), and in endothelium-denuded arteries (Denud, n=6). MnTMPyP reduced arterial diameter by 4 ± 1 μm (n=4, P<0.05). Thapsigargin and iberiotoxin reduced mean arterial diameter by 9 ± 3 μm and 9 ± 1 μm, respectively (P<0.05 for each), consistent with Ca²⁺ spark and K_Ca channel inhibition (19). ^* P < 0.05 when compared with control. # P<0.05 when compared with TNF-α in endothelium-intact arteries.

To investigate the importance of ROS generation in TNF-α-induced vasodilation, responses to TNF-α were measured in control and then in the presence of MnTMPyP (10 μM) in the same artery. MnTMPyP reduced mean TNF-α-induced vasodilations to 7 ± 2 μm, or to 0.34 ± 0.03 of control dilations (Fig. 5C). The same protocol was also performed with thapsigargin (100 nM), a SR Ca²⁺ ATP-ase blocker that inhibits Ca²⁺ sparks, or iberiotoxin (100 nM), a selective K_Ca channel blocker (19). Thapsigargin reduced mean TNF-α-induced vasodilations to 4 ± 2 μm, or 0.15 ± 0.01 of those obtained in control in the same arteries (Fig. 5C). Similarly, iberiotoxin reduced mean TNF-α-induced vasodilations to 6 ± 1 μm or 0.29 ± 0.05 of control (Fig. 5B, C). Data indicate that TNF-α dilates cerebral arteries by generating ROS that stimulate Ca²⁺ sparks and K_Ca channels in smooth muscle cells.

DISCUSSION

The mechanisms by which TNF-α, a pleiotropic cytokine, regulates the diameter of small cerebral arteries was investigated. Data indicate that TNF-α activates smooth muscle cell NAD(P)H oxidase, leading to the generation of ROS that activate Ca²⁺ sparks, and thus, transient K_Ca currents. The ensuing decrease in global [Ca²⁺]_i leads to vasodilation. These findings show for the first time that an inflammatory mediator central in a variety of vascular diseases induces vasodilation through Ca²⁺ spark activation.

TNF-α is produced in the brain during ischemia and is implicated in the etiology of several cardiovascular diseases, including stroke, cardiac failure, coronary heart disease and atherosclerosis (15; 46). TNF-α is reported to induce both vasoconstriction and vasodilation. Topical application of TNF-α dilated cerebral arterioles (7; 37), whereas intracranial injection of TNF-α induced cerebrovascular constriction and reduced blood flow (29; 38; 44). TNF-α also dilated mesenteric and cremaster muscle arterioles (4; 32; 45). Earlier studies suggest TNF-α promotes vasodilation by elevating NO and prostaglandin production in the arterial wall and perivascular tissues (4; 7; 37). In contrast, in pulmonary artery rings, a hypoxia-induced increase in TNF-α mRNA was associated with contraction (43). TNF-α also reduced bradykinin- and acetylcholine (ACh)-induced vasodilation in rat mesenteric arteries (48), reduced bradykinin- and A23187-induced vasodilation in bovine coronary arteries (51), blunted ACh-induced vasodilation in cat carotid arteries (2) and evoked endothelium-dependent vasoconstriction in human and rat coronary arteries (17; 23). In humans, intrabrachial TNF-α injection impaired bradykinin- and ACh-induced vasodilation (11). Data also suggest that TNF-α inhibits vasodilation by triggering ROS that scavenge endothelial NO (51). It is unclear what underlies differential regulation of arterial contractility by TNF-α, but several possibilities exist.

Conceivably, vasoregulatory actions of TNF-α may be vascular bed specific. Different experimental protocols employed in studies may also explain the diversity of observations. In addition, the effects of TNF-α on arterial contractility may be time-dependent. For example, in sheep bronchial arteries, TNF-α initially induced dilation, followed by constriction two hours later (45). Data also suggest differential effects on contractility may occur when TNF-α acts primarily on either endothelial or smooth muscle cells. Conceivably, TNF-α-induced vasoconstriction may occur primarily due to an endothelium-dependent mechanism, whereas vasodilation may be mediated by a direct effect on smooth muscle cells. Here, we show that in isolated, pressurized, myogenic rat cerebral arteries, acute TNF-α application (<30 minutes) induced vasodilation that was endothelium-independent and was markedly attenuated by an antioxidant and inhibitors of SR Ca²⁺ release and K_Ca channels.

TNF-α activates several signaling pathways in vascular smooth muscle cells. TNF-α stimulated NAD(P)H oxidase leading to ROS production in rat aortic smooth muscle cells (13). TNF-α relaxed endothelium-denuded, phenylephrine-constricted aortic rings by PLA₂ activation, implicating ceramide as a mediator (22). In cardiac and endothelial cells, signaling mechanisms for TNF-α involve mitochondria ^{20, 23-25, 35, 36}. TNF-α disrupted the mitochondrial electron transport chain, caused release of cytochrome c, induced mitochondrial ROS generation and triggered apoptosis (14; 28; 30; 41; 52). Although the downstream intramitochondrial target for TNF-α is unclear, in endothelial cells PTP involvement was proposed (52). The PTP is also a downstream target for TNF-α-induced signaling in other cell types, and PTP opening may be triggered by electron transport chain inhibition and mitochondrial depolarization (5; 10; 39). Ceramide had no effect on transient K_Ca currents and PTP openers inhibited these events, consistent with a previous report in arterial smooth muscle cells (10). Therefore, TNF-α is unlikely to activate Ca²⁺ sparks and transient K_Ca currents by elevating ceramide or by inducing PTP opening. In some cell types, including neutrophils, pre-incubation with apocynin is required to block association of NAD(P)H oxidase subunits (40). Here, we show that in cerebral artery smooth muscle cells, apocynin reverses TNF-α-induced transient K_Ca current activation within ∼10 minutes. This effect may reflect rapid subunit turnover in the presence of TNF-α in these cells. Data indicate that TNF-α stimulates NAD(P)H oxidase, generating ROS that activate Ca²⁺ sparks and transient K_Ca currents in cerebral artery smooth muscle cells. These findings are consistent with a recent report that mitochondria-derived ROS activate Ca²⁺ sparks and transient K_Ca currents in arterial smooth muscle cells (50). Thus, ROS derived from different sources induce vasodilation through a common pathway, namely Ca²⁺ spark activation.

TNF-α-induced ROS may activate Ca²⁺ sparks and transient K_Ca currents due to a direct or indirect effect on RyR and K_Ca channels. Cardiac and skeletal muscle RyR channels are redox-sensitive and are activated by oxidizing agents, including H₂O₂ (42). ROS also activate a number of other signal transduction pathways that may regulate RyR channel activity (42; 49). Although superoxide (O₂^-•) is the initial product of NAD(P)H oxidase, O₂^-• has a short half-life and rapidly dismutates to H₂O₂, a relatively stable oxidant. Superoxide is unlikely to mediate effects of TNF-α on Ca²⁺ sparks, particularly since catalase blocked TNF-α-induced transient K_Ca current activation. Data suggest H₂O₂, formed from NAD(P)H oxidase-generated O₂^-•, activates RyR channels, leading to the Ca²⁺ spark frequency elevation. Indeed, exogenous H₂O₂ activates transient K_Ca currents in cerebral artery smooth muscle cells (10). Another potential ROS mediating Ca²⁺ spark activation in this study is hydroxyl radical (OH•), a highly reactive species produced in a transition metal catalyzed reaction (Fenton reaction) between H₂O₂ and O₂^-•. Conventional whole-cell experiments used a pipette solution absent in transition metals, and dialysis of the pipette solution within the cell interior should wash out endogenous cytosolic transition metals. OH• generation in these experiments should have been minimal, arguing against a significant role for OH• in TNF-α-induced Ca²⁺ spark and K_Ca channel activation.

ROS production or inhibition of ROS degradation leads to vasoconstriction, high blood pressure, and vascular hypertrophy (24). However, ROS also stimulate vasodilation. For example, flow-mediated vasodilation occurs due to the generation of mitochondria-derived H₂O₂ (27). Furthermore, topical application of ROS dilates cerebral arteries in vivo (47). Thus, ROS induce both vasoconstriction and vasodilation. Conceivably, there may be a delicate balance between ROS concentrations that regulate physiological functions and higher ROS concentrations that lead to pathologies (24). Data here indicate that TNF-α regulates Ca²⁺ sparks by generating ROS from a single source, NAD(P)H oxidase. During disease, vasoconstriction may result from ROS produced in higher concentrations, from additional enzymes, from exogenous sources, and for prolonged periods. For example, oxyhemoglobin, that may contribute to vasoconstriction following subarachnoid hemorrhage, inhibits Ca²⁺ sparks in cerebral artery smooth muscle cells and this can be blocked by antioxidants (21). TNF-α receptors were recently proposed to spatially confine NAD(P)H oxidase-mediated redox signaling in human microvascular endothelial cells (25). TNF-α may also activate Ca²⁺ sparks by elevating ROS in the local vicinity of Ca²⁺ spark sites. Through restriction, global effects of ROS would be lessened and dilatory signaling could be maintained. Since Ca²⁺ spark sites are close to sarcolemmal K_Ca channels, local NAD(P)H oxidase-derived ROS should also regulate K_Ca channel activity. TNF-α did not alter Ca²⁺ spark amplitude, but increased transient K_Ca current amplitude. NAD(P)H oxidase inhibitors and antioxidants reduced the TNF-α-induced transient K_Ca current amplitude elevation, indicating the effect was mediated by NAD(P)H oxidase-derived ROS. Thus, TNF-α also enhances the effective coupling of Ca²⁺ sparks to K_Ca channels, presumably by elevating K_Ca channel Ca²⁺-sensitivity within the micromolar concentration range generated by these Ca²⁺ transients (34). The combined effects of elevated Ca²⁺ spark frequency and transient K_Ca current amplitude would significantly increase K_Ca channel activity, leading to vasodilation (19; 33).

TNF-α elevated Ca²⁺ wave frequency ∼ 1.5-fold, but reduced global [Ca²⁺]_i. In the absence of receptor agonists, Ca²⁺ wave frequency in arterial smooth muscle cells is low, and Ca²⁺ waves contribute little to global [Ca²⁺]_i (18). When stimulated to high frequencies, Ca²⁺ waves can contribute to global [Ca²⁺]_i and thus, induce smooth muscle contraction (35). Although TNF-α stimulated Ca²⁺ waves, the frequency elevation was insufficient to contribute significantly to global Ca²⁺. In contrast, TNF-α-induced Ca²⁺ spark activation significantly elevated K_Ca channel activity. The resulting membrane hyperpolarization would reduce voltage-dependent Ca²⁺ channel activity, leading to the observed decrease in global [Ca²⁺]_i (19). Therefore, the net effect of TNF-α-induced Ca²⁺ spark and Ca²⁺ wave activation is a reduction in global [Ca²⁺]_i, leading to vasodilation.

In summary, this study defines a novel vasodilatory signaling pathway for an inflammatory cytokine mediated by Ca²⁺ spark and K_Ca channel activation in smooth muscle cells. Following ischemia and stroke, TNF-α-induced Ca²⁺ spark activation and cerebrovascular dilation would be important in regulating local blood supply in the affected brain region.