Exercise-induced oxidative-nitrosative stress is associated withimpaired dynamic cerebral autoregulation and blood-brain barrierleakage.Bailey, D. M., Evans, K. A., McEneny, J., Young, I., Hullin, D. A., James, P. E., Ogoh, S., Ainslie, P. N.,Lucchesi, C., Rockenbauer, A., Culcasi, M., & Pietri, S. (2011). Exercise-induced oxidative-nitrosative stress isassociated with impaired dynamic cerebral autoregulation and blood-brain barrier leakage. ExperimentalPhysiology, 96(11)(11), 1196-1207. https://doi.org/10.1113/expphysiol.2011.060178

Published in:Experimental Physiology

Queen's University Belfast - Research Portal:Link to publication record in Queen's University Belfast Research Portal

General rightsCopyright for the publications made accessible via the Queen's University Belfast Research Portal is retained by the author(s) and / or othercopyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associatedwith these rights.

Take down policyThe Research Portal is Queen's institutional repository that provides access to Queen's research output. Every effort has been made toensure that content in the Research Portal does not infringe any person's rights, or applicable UK laws. If you discover content in theResearch Portal that you believe breaches copyright or violates any law, please contact openaccess@qub.ac.uk.

Download date:31. Jan. 2021

Exp

erim

enta

lPhy

siol

ogy

1196 Exp Physiol 96.11 pp 1196–1207

Research PaperResearch Paper

Exercise-induced oxidative–nitrosative stress is associatedwith impaired dynamic cerebral autoregulationand blood–brain barrier leakage

Damian M. Bailey1,2, Kevin A. Evans1, Jane McEneny3, Ian S. Young3, David A. Hullin4, Philip E. James5,Shigehiko Ogoh6, Philip N. Ainslie7, Celine Lucchesi2, Antal Rockenbauer8, Marcel Culcasi2

and Sylvia Pietri2

1Neurovascular Research Laboratory, Faculty of Health, Science and Sport, University of Glamorgan, UK2Sondes Moleculaires en Biologie, Laboratoire Chimie Provence UMR 6264, CNRS-Universite de Provence, Marseille, France3Centre for Clinical and Population Sciences, Queen’s University Belfast, Belfast, UK4Department of Medical Biochemistry, Royal Glamorgan Hospital, Llantrisant, UK5Wales Heart Research Institute, School of Medicine, Cardiff University, UK6 Department of Biomedical Engineering, Toyo University, Saitama, Japan7 Department of Human Kinetics, University of British Columbia Okanagan, Kelowna, BC, Canada8Chemical Research Center, Institute of Chemistry, Hungarian Academy of Sciences, Budapest, Hungary

The present study examined whether dynamic cerebral autoregulation and blood–brain barrierfunction would become compromised as a result of exercise-induced oxidative–nitrosative stress.Eight healthy men were examined at rest and after an incremental bout of semi-recumbent cyclingexercise to exhaustion. Changes in a dynamic cerebral autoregulation index were determinedduring recovery from continuous recordings of blood flow velocity in the middle cerebralartery (MCAv) and mean arterial pressure during transiently induced hypotension. Electronparamagnetic resonance spectroscopy and ozone-based chemiluminescence were employed fordirect detection of spin-trapped free radicals and nitric oxide metabolites in venous blood.Neuron-specific enolase, S100β and 3-nitrotyrosine were determined by ELISA. While exercisedid not alter MCAv, it caused a mild reduction in the autoregulation index (from 6.9 ± 0.6to 5.5 ± 0.9 a.u., P < 0.05) that correlated directly against the exercise-induced increase in theascorbate radical, 5-(diethoxyphosphoryl)-5-methyl-1-pyrroline N -oxide and N -tert-butyl-α-phenylnitrone adducts, 3-nitrotyrosine and S100β (r = –0.66 to –0.76, P < 0.05). In contrast,no changes in neuron-specific enolase were observed. In conclusion, our findings suggest thatintense exercise has the potential to increase blood–brain barrier permeability without causingstructural brain damage subsequent to a free radical-mediated impairment in dynamic cerebralautoregulation.

(Received 21 June 2011; accepted after revision 9 August 2011; first published online 12 August 2011)Corresponding author D. M. Bailey: Neurovascular Research Laboratory, Faculty of Health, Science and Sport,University of Glamorgan, Mid-Glamorgan, South Wales CF37 4AT, UK. Email: dbailey1@glam.ac.uk

Cerebral autoregulation (CA) is a homeostatic mechanismthat serves to maintain cerebral blood flow (CBF) constantover a wide range of perfusion pressures subject tomyogenic, neurogenic and metabolic control (Paulsonet al. 1990). Effective CA is important given the relianceof the human brain on oxygen and glucose to supportthe metabolic demands of neuronal activity and the needto protect brain tissue from the potentially damaging

effects of hypo/hyperperfusion. Disordered CA has beendocumented in diverse models of cerebral vascular diseaseand injury, many of which have been linked to vascularcognitive impairment (Panerai, 2009) and ischaemicstroke (Aries et al. 2010).

Acute exercise has the potential to compromise CAeven in healthy humans, given the marked increase inpulse pressures, which often exceed values considered to

DOI: 10.1113/expphysiol.2011.060178 C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

Exp Physiol 96.11 pp 1196–1207 Redox regulation of blood–brain barrier function 1197

represent the upper limit of CA (Ogoh et al. 2005b; Secheret al. 2008). Furthermore, owing to the latency of CA(∼3–5 s; Aaslid et al. 1989), the rapid fluctuations in meanarterial pressure (MAP), such as those encountered duringrowing and resistance exercise, have been shown to be toobrisk to be dampened effectively by the cerebrovasculature(Pott et al. 1997; Edwards et al. 2002). In the settingof disordered CA, it is tempting to speculate that the‘exercising brain’ may be less capable of buffering rapidsurges in MAP and may potentially be more vulnerable tomild overperfusion and extracellular (vasogenic) oedemasubsequent to mechanical disruption of the blood–brainbarrier (BBB).

The impact of exercise on CA in humans remainsequivocal, however, due in part to differences in intensity,duration and mode. Brys et al. (2003) were the first toapply transfer function analysis as an index of dynamicCA (dCA) during exercise; however, they failed to showany effect of moderate-intensity semi-recumbent cyclingexercise [3 min increments ranging from 50 to 150 W,equivalent to heart rates (HRs) of 108–154 beats min−1]on the low-frequency (LF) gain between arterial bloodpressure and middle cerebral artery velocity (MCAv),implying that dCA remained intact. In contrast, Ogohet al. (2005a) later identified that dCA became impairedduring more prolonged, intense semi-recumbent cyclingexercise (12–15 min at a fixed power output of 168 W,equivalent to a HR of 174 beats min−1 and after ∼27 minof continued exercise to the point of volitional exhaustionat 185 beats min−1) as indicated by an increase in LF gainand reduction in phase.

Emerging evidence suggests that free radicals may beinvolved in the metabolic regulation of dCA. In rodents,subdural perfusion with the superoxide anion (O•−

2 )was shown to impair dCA subsequent to activation ofpotassium-sensitive calcium (KCa) channels in cerebralvascular smooth muscle cells (Zagorac et al. 2005). Asexercise is an established stimulus for oxidative–nitrosative(OX-NOX) stress (Bailey et al. 2007), especially duringthe transition from moderate to severe intensity (Fogartyet al. 2010), it is reasonable to suggest that the impaireddCA previously observed in humans (Ogoh et al. 2005a)may prove the consequence of increased free radicalformation. This is consistent with a prior suggestion thatimpaired dCA may arise subsequent to an efflux of (asyet unknown) metabolites into the cerebrovascular tissuecaused by an increase in brain metabolism (Dalsgaard et al.2004).

To test this hypothesis, we examined changes indCA during the recovery period following exercise,using transcranial Doppler ultrasound and subsequentderivation of a previously validated autoregulation index(ARI; Tiecks et al. 1995). Electron paramagnetic resonance(EPR) spectroscopy and ozone-based chemiluminescencewere employed for direct detection of spin-trapped free

radicals and nitric oxide, respectively. We hypothesisedthat intense exercise would impair dCA, as indicated by areduction in ARI, and that this would correlate inverselyagainst the increase in OX-NOX stress biomarkers.We specifically employed the β-phosphorylated nitrone,5-(diethoxyphosphoryl)-5-methyl-1-pyrroline N-oxide(DEPMPO), in a first attempt to spin trap O•−

2

in exercising humans to confirm previous findingsestablished in the rodent model (Zagorac et al.2005). We further hypothesised that in the setting ofdisordered CA, the haemodynamic stress of exercisewould increase BBB permeability, as indicated by systemicaccumulation of the astrocytic BBB-specific proteinS100β.

Methods

Subjects

Following ethics approval, eight physically active, healthymen aged 35 ± 7 years old (mean ± SD) provided writteninformed consent according to the Declaration of Helsinki.Subjects with a history of vascular disease, migraine,head injury or those taking medication that may haveinfluenced the autonomic nervous system were excludedfrom participation. They were encouraged to follow a lownitrate/nitrite diet for 4 days, with specific instructions toavoid fruits, salads and cured meats (Wang et al. 1997).Subjects were also asked to refrain from caffeine andphysical activity for 2 days prior to the study.

Design

Subjects attended the laboratory following a 12 hovernight fast and rested for 20 min in a semi-recumbent position. Blood samples and haemodynamicmeasurements were performed at baseline (rest) andfollowing completion of a cycling test to volitionalexhaustion (exercise).

Peak exercise test

Following two familiarisation sessions with theexperimental set-up, each subject was seated on anelectronically braked, semi-recumbent cycle ergometer(Corival; Lode BV, Groningen, The Netherlands) with thebackrest maintained at 70 deg. The initial workload was setat 35 W for 5 min (70 r.p.m.) and increased by 35 W min−1

until volitional exhaustion. Each subject was instructed tosignal clearly to the investigators when they consideredthey could continue at the specified power output forno longer than 60 s, as previously described (Bailey et al.2000). Cardiorespiratory and cerebrovascular recordings

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

1198 D. M. Bailey and others Exp Physiol 96.11 pp 1196–1207

were averaged during the last 60 s of the resting period andduring the last 60 s of peak exercise.

Cardiorespiratory data

Mean arterial pressure. Beat-to-beat arterial pressurewaveforms were recorded continuously by fingerphotoplethysmography (Finometer R© PRO; FinapresMedical Systems, Amsterdam, The Netherlands).Correction for vertical displacement of the finger cuffrelative to heart level was achieved using a reference probeplaced on the chest at the fourth intercostal space in themid-clavicular line.

Gas exchange. Subjects wore a leak-free mask, and gasexchange was measured using a semi-automated Douglasbag system. Expired gas fractions were measured using fastresponding paramagnetic O2 and infrared CO2 analysers(Servomex 1400 Series Analyser, Sussex, UK). The volumeof expired gas was measured using a dry gas meter(Harvard Ltd, Kent, UK), and oxygen uptake (VO2 ) andcarbon dioxide output (VCO2 ) were calculated via theHaldane equation.

Haemodynamic data

Middle cerebral artery velocity. The right middle cerebralartery was insonated using established search techniquesthrough the temporal window ∼1 cm above the zygomaticarch at a depth of 45–58 mm with 2 MHz pulsedtranscranial Doppler ultrasound (Multi-Dop X4; DWLElektroniche Systeme GmbH, Sipplingen, Germany).Following signal optimisation, the probe was attached tothe skull at a fixed angle and secured with an adjustableheadset.

Cerebrovascular resistance (CVR) was calculated asMAP/MCAv and cerebrovascular conductance (CVC) asMCAv/MAP.

Dynamic cerebral autoregulation. An ARI wasdetermined using the thigh-cuff inflation–deflationtechnique (Aaslid et al. 1989) prior to and during therecovery period immediately following exercise. Briefly,bilateral thigh cuffs were connected to a Hokanson E20(Bellevue, WA, USA) and inflated to 30 mmHg above therecorded systolic blood pressure for 3 min. Lower limbischaemia was confirmed using Doppler ultrasound by alack of blood flow in the dorsalis pedis artery. Cuffs weresubsequently deflated (<0.1 s) and the process repeatedthree times with a 3 min inter-trial recovery period.An ARI was assigned to each of the trials (mean valuerecorded) following computation of a second-order linear

differential equation (Tiecks et al. 1995) given by:

dPn = MAP − MAPbase

MAPbase − CCP

x2n = x2n−1 + (x1n − 2D · x2n−1)

f · T

x1n = x1n−1 + (dPn − x2n−1)

f · T

mVn = MCAvbase · (1 + dPn − k · x2n)

where dPn is the normalised change in MAP relative to thecontrol value (MAPbase) adjusted for the estimated criticalclosing pressure (CCP); x2n and x1n are state variables(0 at baseline); mVn is modelled mean velocity; MCAvbase

is baseline MCAv; f is the sampling frequency (100 Hz)and n is the sample number. The mVn generated from 10predefined combinations of parameters T (time constant),D (dampening factor) and k (dynamic autoregulatorygain) that best fit the recorded MCAv was taken as anindex of dCA and ranged between 0 (entirely passiveautoregulation) to 9 (most brisk autoregulation).

Metabolic data

Blood samples were obtained from a catheter located ina forearm antecubital vein following 20 min of rest in thesemi-recumbent position and timed to coincide with thetermination of exercise. Blood was centrifuged at 600g(4◦C) for 10 min and the supernatant immediately snap-frozen and stored under liquid nitrogen (Cryopak CP100;Taylor-Wharton, Theodore, AL, USA) prior to analysis.

Chemicals

All chemicals, including the spin trap N-tert-butyl-α-phenylnitrone (PBN), 2-hydroxypropyl-β-cyclodextrin(HBC), diethylenetriaminepentaacetic acid (DTPA), ferricchloride (FeCl3), ferrous sulfate (FeSO4) and hydrogenperoxide (H2O2) were of the highest available purityfrom Sigma-Aldrich R© (UK). The spin trap DEPMPO wassynthesised and purified according to published methods(Frejaville et al. 1995; Barbati et al. 1999).

Free radical detection and oxidative stress

General. Electron paramagnetic resonance spectra wererecorded at X-band (∼9.8 GHz) using either a Bruker(Karlsruhe, Germany) ESP 300 (DEPMPO studies) orEMX (Karlsruhe, Germany) spectrometer (other studies)using 100 kHz modulation frequency, 10 mW microwavepower and standard TM110 cavities unless otherwisenoted.

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

Exp Physiol 96.11 pp 1196–1207 Redox regulation of blood–brain barrier function 1199

In vitro EPR studies. Preliminary in vitro experimentswere performed to investigate whether HBC may interferewith formation of the DEPMPO/hydroxyl radical (HO.)spin adduct (DEPMPO-OH). Thus, a mixture of 0.1M DEPMPO and 50 mM HBC was incubated at roomtemperature with either a Fenton reagent [consisting of1 mM FeSO4 and 1 mM H2O2 in 20 mM phosphate buffersolution (PBS)] or 1.5 mM FeCl3 in water, and the sampleswere added to 50 μl glass capillary tubes (HirschmannLaborgerate GmbH & Co. KG, Eberstadt, Germany) andsealed with Critoseal R© (Cardinal Health, Dublin, OH,USA). Single-scan EPR spectra were recorded 1–20 minfollowing addition of iron salt and the signals comparedto corresponding controls (no HBC). Instrument settingswere as follows: magnetic field resolution (R), 4096 points;modulation amplitude (MA), 0.313 Gauss (G); receivergain (RG), 6.3 × 104; time constant (TC), 81.92 ms; sweeprate (SR), 3.34 G s−1 for a sweep width (SW) of 140 G.

Ex vivo spin trapping in blood (DEPMPO). Heparinisedwhole blood samples (1 ml) drawn at the end of rest andexercise periods (defined in ‘Metabolic data’) were placedinto 2 ml cryovial tubes prefilled with a mixture of (finalconcentration) 0.2 mM DTPA, 80 mM HBC, 40 mM PBSand 0.5 ml of aqueous DEPMPO (0.1 M) and immediatelyfrozen in liquid N2. The EPR acquisition was initiated1 min after complete thawing in the conditions describedabove except that RG was 1.6 × 105.

Ex vivo spin trapping in blood (PBN). Aliquots (4.5 ml) ofwhole blood were added to a 6 ml glass vacutainer serumseparation tube primed with 1.5 ml of PBN dissolvedin physiological saline (50 mM final concentration). Thevacutainer was gently mixed then placed in the dark to clotfor 10 min. Following centrifugation, 1 ml of the serumadduct was added to a borosilicate glass tube containing1 ml of spectroscopic-grade toluene and vortex mixed for10 s. The sample was centrifuged at 600g for a further10 min and 200 μl of the organic supernatant added toa (N2-flushed) precision-bore quartz EPR tube, vacuumdegassed to remove O2, and blocks of 10 incremental EPRscans were recorded using the following parameters: R,2048 points; MP, 20 mW; MA, 0.50 G; RG, 1 × 105; TC,82 ms; SR, 0.4 G s−1 for SW, 50 G (Bailey et al. 2009c).

Ascorbate radical (A•−). Plasma (1 ml) from bloodsamples collected as described above was injected intoa high-sensitivity multiple-bore sample cell (AquaX;Bruker Instruments Inc., Billerica, MA, USA), andthe characteristic A•− EPR doublet (hydrogen couplingconstant, aH

β = 1.76 G, see structure in Fig. 2) wasrecorded 12 min after the end of plasma recovery bysignal averaging three scans with the following instrument

parameters: R, 1024 points: MP, 20 mW; MA, 0.65 G; RG,2 × 105; TC, 40.96 ms; SR, 0.25 G s−1 for SW, 15 G.

Electron paramagnetic resonance signal quantification.Electron paramagnetic resonance spectra were usedas is (DEPMPO adducts) or filtered identically usingBruker WinEPR 2.11 and then simulated with Bruker(BioSpin, Bruker UK Limited, Coventry, UK), SimFonia,WinSim (Duling, 1994), ROKI (Rockenbauer & Korecz,1996) or SimEPR32 software (Adamski et al. 2003)for A•−, DEPMPO and PBN adducts, respectively.Relative free radical concentrations were calculated bydouble integration of simulated (DEPMPO adducts) orfiltered spectra using the ROKI or Origin 5.0 software,respectively. The intra- and interassay coefficients ofvariation for all assays were both <10%.

Oxidation of low-density lipoprotein (LDL). Low-densitylipoprotein was isolated by rapid ultracentrifugation andpurified by size-exclusion chromatography (McDowellet al. 1995). The protein concentration was standardised to50 mg L−1, and oxidation was initiated following additionof copper II chloride (2 μmol final concentration) at37◦C. Conjugated diene formation was monitoredspectrophotometrically in triplicate by the change inabsorbance at 234 nm using a 96-well microplate reader(Spectromax 190; Molecular Devices Corp., Sunnyvale,CA, USA). The time at half-maximal absorbance of thepropagation phase (time 1/2max, in minutes) was takenas a marker of the susceptibility of LDL to oxidation. Theintra- and interassay coefficients of variation were <3 and<5%, respectively.

Antioxidants

For ascorbic acid measurements, plasma was stabilised anddeproteinated by adding 900 μl of 5% metaphosphoricacid (Sigma-Aldrich, Gillingham, UK) to 100 μl K-EDTA plasma. Ascorbic acid was subsequently assayed byfluorimetry based on the condensation of dehydroascorbicacid with 1,2-phenylenediamine (Vuilleumier & Keck,1993). Concentrations of α/γ-tocopherol, α/β-carotene,retinol, lycopene, zexanthin, β-cryptoxanthin and luteinwere determined using an HPLC method (Catignani& Bieri, 1983; Thurnham et al. 1988). The intra andinterassay coefficients of variation were both <5%.

Nitrosative stress

Nitric oxide. Plasma NO metabolites were measured byozone-based chemiluminescence (Bailey et al. 2009b,c).Samples (20 μl) were analysed for the total concentrationof NO [nitrate (NO3

−) + nitrite (NO2−) + S-nitrosothiols

(RSNO)] by vanadium (III) reduction (Ewing & Janero,

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

1200 D. M. Bailey and others Exp Physiol 96.11 pp 1196–1207

Table 1. Haemodynamic data

Rest Exercise

CardiovascularHeart rate (beats min−1) 69 ± 8 185 ± 7∗Mean arterial pressure (mmHg) 87 ± 11 107 ± 8∗

RespiratoryVO2 (L min−1) 0.28 ± 0.08 3.25 ± 0.73∗VCO2 (L min−1) 0.25 ± 0.12 3.60 ± 0.74∗Respiratory exchange ratio (a.u.) 0.90 ± 0.37 1.16 ± 0.07∗PvO2 (mmHg) 40 ± 7 55 ± 18∗PvCO2 (mmHg) 45 ± 3 40 ± 5∗

CerebrovascularMCAv (cm s−1) 49 ± 6 47 ± 8CVR (mmHg cm−1 s−1) 1.73 ± 0.27 2.27 ± 0.33∗CVC (cm s−1 mmHg−1) 0.59 ± 0.09 0.45 ± 0.06∗

Values are means ± SD, based on the last 60 s of a 20 minresting (semi-recumbent) baseline and during the last 60 s ofpeak exercise. Abbreviations: CVC, cerebral vascular conductance;CVR, cerebral vascular resistance; MCAv, middle cerebral arteryvelocity; PvCO2 , venous partial pressure of carbon dioxide; PvO2 ,venous partial pressure of oxygen; VCO2 , carbon dioxide output;and VO2 , oxygen uptake. ∗P < 0.05 different from rest.

1998). A separate sample (200 μl) was injected into tri-iodide reagent (Hausladen et al. 2007; Rogers et al. 2007)for the measurement of NO2

− and RSNO, and 5% acidifiedsulphanilamide was added and left to incubate in the darkat 21◦C for 15 min to remove NO2

− for the measurementof RSNO in a third parallel sample. Plasma NO3

−

was calculated as total NO minus (NO2− + RSNO). All

calculations were performed using Origin/Peak Analysissoftware (Northampton, MA, USA). The intra- andinterassay interassay coefficients of variation for all NOmetabolites were 7 and 10%, respectively.

3-Nitrotyrosine (3-NT). Plasma 3-NT was measuredby ELISA (Hycult Biotechnology, b.v., Uden, TheNetherlands), with a lower detection limit of 2 nM andintra- and interassay interassay coefficients of variation of<2 and <5%, respectively.

Figure 1. Exercise-induced reduction in the autoregulationindex (ARI)Values are means + SD. Dotted lines refer to fastest (A), typical (B) andslowest (C) rates of autoregulation; ∗P < 0.05 different from rest.

Structural stress

S100β and neuron-specific enolase (NSE). The serumconcentrations of S100β and NSE were measured byELISA (S100β, BioVendor Candler, NC, USA; NSE, CanAgFujirebio Diagnostics Inc, Malvern, PA, USA). The lowerdetection limits were 5 ng l−1 and <1 μg l−1, respectively,with intra- and interassay coefficients of variation both<5% for S100β and NSE.

Blood gases

Venous blood (2–3 ml) was presented anaerobically toa blood gas analyser (ABL 5; Radiometer, Copenhagen,Denmark) for determination of PO2 and PCO2 . The intra-and interassay coefficients of variation were both <5%.

Statistical analysis

Following confirmation of distribution normality usingShapiro–Wilk W tests, data were analysed using Student’spaired t tests and relationships determined withPearson product–moment correlations. Significance wasestablished at P < 0.05, and data are expressed asmeans ± SD.

Results

Cardiorespiratory and cerebrovascular data

Exhaustion occurred within 8.5 ± 1.3 min during the peakexercise test. Exercise did not increase MCAv, which wasprobably due to hyperventilation-induced hypocapnia asindicated by the reduction in venous PCO2 (Table 1). Incombination with an exercise-induced increase in MAP,this translated into an increased CVR and decreased CVC.A mild reduction in ARI was observed during the recoveryperiod following exercise, indicative of impaired dCA(Fig. 1).

Metabolic data

Effect of HBC on the in vitro formation of DEPMPO-OH.Excess HBC was originally included in the DEPMPO-supplemented spin-trapping milieu in an attempt toprotect nitroxide spin adducts from reduction intoEPR-silent species due to the action of blood and/orplasma constituents such as ascorbate, glutathione orα-tocopherol. Although HBC has been shown to forminclusion complexes with the DEPMPO/O•−

2 adduct(DEPMPO-OOH), leading to a significant increase in itshalf-life in the presence of ascorbate in vitro (Karoui et al.2002), no data were previously available related to its effecton DEPMPO-OH.

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

Exp Physiol 96.11 pp 1196–1207 Redox regulation of blood–brain barrier function 1201

In a DEPMPO-supplemented biological system,DEPMPO-OH may be formed by the following routes:(i) reduction of DEPMPO-OOH (e.g. by thiols orenzymes such as glutathione peroxidase); (ii) genuinetrapping of HO. (e.g. formed by a Fenton system); or(iii) nucleophilic synthesis (NS) via metal ion-catalysedaddition of water to the nitronyl function (Frejaville et al.1995). In control experiments, both a Fenton reaction(Fe2+ + H2O2) or Fe3+-assisted nucleophilic synthesisin PBS or water, respectively, yielded DEPMPO-OH,the EPR signal of which consists of eight lines, asshown in Fig. 2A. These EPR spectra were satisfactorily(r2 > 0.99) simulated assuming superimposition of the cis(hyperfine coupling constants: aN (Nitrogen) = 13.95 G;aP (Phosphorous) = 47.45 G; and aH

β (Hydrogen)= 14.47 G) and trans diastereoisomers (aN = 14.05 G;aP = 47.30 G; and aH

β = 12.77 G; Culcasi et al. 2006b).Over a 20 min observation period at room temperature,the mean cis- versus trans-DEPMPO-OH percentagewas constantly higher in the NS (48%) than in theFenton system (34%). Addition of excess HBC in themilieu before DEPMPO-OH formation was triggereddid not modify the difference in the cis versus transisomer ratios (i.e. 33 and 41% in the Fenton and NS,respectively) but caused a ∼4.4-fold decrease in the totalFenton-, but not in NS-derived EPR signal intensity.Moreover, in the (Fenton + HBC) experiment, additionalsignals consistent with the scavenging of two carbon-centred radicals [DEPMPO-R; aN = 14.39 G (14.71%);aP = 46.45 G (48.98%); and aH

β = 20.76 G (21.36%)]were observed, accounting for 34–40% of the overalltrapped species (Fig. 2A). Though this implies a stronginteraction of HO. with HBC, preformed DEPMPO-OHin PBS vanished instantaneously following addition of afew drops of whole human blood (not shown), confirmingthat the initial presence of HBC is a prerequisite for signaldetection (Karoui et al. 2002).

Electron paramagnetic resonance and biochemicalevidence of exercise-induced oxidative stress (Table 2).Exercise resulted in a ∼1.5-fold increase in A•− (Fig. 2B).In DEPMPO + HBC experiments, blood samples thatwere paramagnetic (i.e. six of the eight subjects) containedeither a single component consistent with a DEPMPO-Radduct (in two of six subjects), the calculated couplings ofwhich (aN = 14.80 G; aP = 47.90 G; and aH

β = 21.40 G)were different from that of the in vitro experimentsdescribed above, or mixtures of DEPMPO-OH and16–38% of a signal attributed to the reduced formof DEPMPO [DEPMPO-H; aN = 14.5 G; aP = 55.5 G;and aH

β = 16.0 G]. Figure 2C shows typical DEPMPO-derived signals for subjects 5 and 7. In DEPMPO-OH-containing blood samples, the cis-DEPMPO-OHproportion ranged between 30 and 36%, compatiblewith a Fenton-like mechanism (see above). Despite

one of the six subjects showing DEPMPO adductsin his blood sample at rest, the overall mean totalspin adduct levels increased markedly following exercise(P < 0.05; n = 8). In the ex vivo experiments withPBN, EPR sextets from toluene-extracted blood samples,satisfactorily simulated assuming alkoxyl spin adducts(PBN-RO′: aN = 13.6 G; and aH

β = 1.9 G), were observedin all blood samples (Fig. 2D), with a mean approximatelytwofold increase postexercise. Exercise also decreased LDLlag time, and the exercise-induced increase in oxidantbiomarkers correlated directly against the reduction in ARI(Fig. 3A–C).

Antioxidants. Exercise did not alter the venousconcentration of ascorbate or any of the lipid solubleantioxidants (data not shown).

Nitrosative stress. While exercise did not alter total NO,NO3

− or NO2−, an increase in RSNO and 3-NT was

observed, the latter correlating against the reduction inARI (r = –0.81, P < 0.05).

Structural stress. Exercise did not affect NSE, whereasS100β increased, correlating against the reduction in ARI(Fig. 3D).

Discussion

Consistent with our original hypotheses, the presentstudy has revealed two major findings. First, exerciseincreased OX-NOX stress, which correlated directlyagainst the degree of impairment in dCA during thepostexercise recovery period. Second, impaired dCA wasassociated with an increased formation of S100β that,when combined with a lack of change in NSE, wastaken to reflect disruption of the BBB independent ofstructural tissue damage. In combination, these findingssuggest that intense exercise has the potential to increaseBBB permeability subsequent to a free radical-mediatedimpairment in dCA.

Oxidative–nitrosative stress

We specifically employed EPR spectroscopy becauseit represents the most direct, specific and sensitiveanalytical technique for the molecular detection andsubsequent characterisation of free radicals sine quanon. Given that the concentration of ascorbate inhuman plasma is orders of magnitude greater than anyoxidising free radical, combined with the low one-electronreduction potential (Eo′ = 282 mV) associated with theA•−/ascorbate monanion (AH−) couple (Williams &Yandell, 1982), any radical generated within the systemiccirculation has the potential to react endogenously with

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

1202 D. M. Bailey and others Exp Physiol 96.11 pp 1196–1207

A B

Simulated

C

Simulated

DEPMPO-OH

NS

Fenton

Exercise

Rest

Simulated

D

Rest Exercise

240 25

Rest

Exercise

Simulated

O O

OO-

HO OH

A-

HN Me

O

P(O)(OEt)2+ HO

DEPMPO cis-DEPMPO-OH trans-DEPMPO-OH

+

-N

O

HO

H Me

P(O)(OEt)2+

N

O

H

HO Me

P(O)(OEt)2

NO

tBu

+ -N

O

tBu

OR

H+ RO

RO-NBPNBP

Figure 2. Electron paramagnetic resonance evidence [via conventional or cyclodextrin (HBC)-assistedspin trapping] for exercise-induced free radical formationA, general principle of the formation of the EPR-distinguishable cis- and trans- diastereoisomers of the 5-(diethoxyphosphoryl)-5-methyl-1-pyrolline N-oxide (DEPMPO) hydroxyl radical spin adduct (DEPMPO-OH), typicalsignals with their respective intensities in aqueous medium recorded 1 min after running a nucleophilic synthesis(NS) or HBC-assisted Fenton reaction, and their computer simulations demonstrating a relative cis- versus trans-DEPMPO-OH proportion of 47 and 34% in NS and Fenton conditions, respectively, and the additional presencein the Fenton experiment of two HBC-derived DEPMPO alkyl adducts (some non-composite lines indicated byarrows) accounting for 34% of the total signal. B, ascorbate radical (A•−) structure and signal directly recorded inblood samples from subject 1. C, HBC-assisted DEPMPO spin adducts observed in the blood from two subjects,showing major formation of either DEPMPO-OH (84% of the signal, with 30.2% cis isomer proportion; othercomponent DEPMPO-H; subject 7; outer lines indicated by arrows) or only DEPMPO-R (subject 5); magnification ofexercise versus corresponding rest signals is also indicated. D, general principle of the formation of N-tert-butyl-α-phenylnitrone (PBN)/alkoxyl radical (RO.) spin adduct and signal observed in blood samples from subject 1. Detailson sampling protocol, spectral recording conditions and splitting parameters used for simulations are indicated inthe Methods and Results.

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

Exp Physiol 96.11 pp 1196–1207 Redox regulation of blood–brain barrier function 1203

this terminal small-molecule antioxidant to form thedistinctive A•− EPR doublet [(R• + AH− → A•− + R-H] (Buettner, 1993). Thus, the elevation in A•− providesdirect evidence for an increased systemic accumulation offree radicals in response to intense exercise, which is inbroad agreement with the literature (Powers & Jackson,2008).

We followed this up with spin trapping in an attemptto characterise the individual free radical species formedduring exercise. The synthetic spin trap DEPMPO isrecognised as one of the most effective O•−

2 EPR reportersbased on the increased stability of the correspondingDEPMPO-OOH adduct in buffers compared with thetraditional spin trap of choice, 5,5-dimethyl-1-pyrrolineN-oxide (Frejaville et al. 1995). However, despiteoptimising our potential to trap O•−

2 in human bloodby adding HBC to the sampling mixture, we consistentlyfailed to detect DEPMPO-OOH, which yields distinctive,highly distorted EPR lines due to its partial insertionwithin the cyclodextrin cavity (Karoui et al. 2002).

This does not exclude the possibility of exercise-induced O•−

2 formation; rather, it highlights the needto improve detection sensitivity further, given theunavoidable limitations posed by human spin trapping,which ultimately relies on the ex vivo detection, in sucha highly reductive milieu as blood, of species formedclearly downstream of the primary reaction pathwaythat we assume reflects spontaneous events that occurin vivo. In support, the increased formation of 3-NT, a surrogate biomarker of peroxynitrite (ONOO−;Pacher et al. 2007) and tendency towards a reductionin the plasma NO pool, provides tentative evidence forO•−

2 formation and its corresponding reaction with NO

(O•−2 + NO

k=16−20×109Ms−1→ ONOO−; Nauser & Koppenol,2002).

In contrast, most HBC-supplemented samples obtainedpostexercise contained DEPMPO-OH signals, the cis/transdiastereoisomeric analysis (Culcasi et al. 2006b) of whichindicated a Fenton pathway rather than a free radical-independent degradation of the spin trap. In our samplingprotocol for EPR analysis (i.e. blood was in contact withthe DEPMPO + HBC solution for less than 1 min beforefreezing), the possibility that DEPMPO-OH was formedby conversion of primary DEPMPO-OOH by glutathioneperoxidase is unlikely. First, HBC encapsulation of theO•−

2 adduct offers almost complete protection againstenzymatic reduction, with visible DEPMPO-OH evolvingonly after a ∼40 min delay (Karoui et al. 2002). Second, thereported mean cis-DEPMPO-OH ratio for this reactionin vitro (Culcasi et al. 2006b) is much lower than thevalue reported in our study. Additional control studiesemploying competitive inhibitors (e.g. DMSO, ethanol,catalase and a mixture of nitrones that react at differentrates with HO•) are warranted to confirm unequivocallythe trapping of authentic HO•.

Table 2. Metabolic data

Rest Exercise

Oxidative stressA•− (a.u.) 2, 922 ± 385 4, 306 ± 722∗Total DEPMPO adducts(a.u.)

0.8 ± 2.3 10.9 ± 8.7∗

Total PBN adducts (a.u.) 29, 311 ± 13, 086 61, 186 ± 24, 070∗LDL lag time at 1/2max(min)

80.1 ± 9.9 76.5 ± 10.5∗

Nitrosative stressTotal nitric oxide(μmol L−1)

40.6 ± 15.8 38.1 ± 17.3

Nitrate (μmol L−1) 40.2 ± 15.8 37.7 ± 17.2Nitrite (nmol L−1) 316 ± 47 371 ± 113S-Nitrosothiols(nmol L−1)

59 ± 22 81 ± 23∗

3-Nitrotyrosine(nmol L−1)

47.2 ± 51.5 105.8 ± 59.2∗

Structural stressS100β (ng L−1) 58.3 ± 13.1 104.9 ± 44.5∗Neuron-specific enolase(μg L−1)

2.99 ± 0.59 2.85 ± 0.66

Values are means ± SD. Abbreviations: A•−, ascorbate radical;DEPMPO, 5-(diethoxyphosphoryl)-5-methyl-1-pyrolline N-oxide[the spin-adduct stabiliser, HBC (80 mM), was added tothe DEPMPO-containing sampling medium]; LDL, low-densitylipoprotein; and PBN, N-tert-butyl-α-phenylnitrone. ∗P < 0.05different from rest.

The finding of minor EPR signals in exercise samplesattributable to DEPMPO-H is intriguing, given that thisadduct typically arises following reaction of DEPMPOwith either inorganic one-electron donors or mildoxidants (Barbati et al. 1999; Gosset et al. 2011). A previousstudy reported this particular DEPMPO adduct to beformed together with DEPMPO-OH upon activation ofNADPH oxidase (a known cellular source of O•−

2 ) ofsodium sulfite-treated rat lymphocytes, or in a peroxidise–H2O2–sulfite pseudo-Fenton system (Chamulitrat, 1999).

The above hypothesis of exercise-induced HO•

formation is mirrored by the detection of increasedLDL oxidation and PBN-RO′ formation. For these latterspin adducts, calculated EPR parameters matched thoseassigned to lipid-derived alkoxyl radicals (LO•) found intoluene extracts of PBN added to blood during open-heartsurgery (Tortolani et al. 1993) and during exercise (Baileyet al. 2004).

Being highly reactive in vivo (lifetime of only 1 ns;von Sonntag, 1987), HO• is the most oxidising radicallikely to be formed within the exercising humancirculation (Eo’ = 2,310 mV; Koppenol & Butler, 1985).The mechanisms that promote HO• formation remainunclear, although they may involve catalytic ferrousiron (Fe2+) liberated during exercise-induced activationof polymorphonuclear leukocytes, metabolic acidosis orstructural damage to membrane proteins (Bailey et al.2004). The reductive cleavage of H2O2 (the product of

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

1204 D. M. Bailey and others Exp Physiol 96.11 pp 1196–1207

O•−2 dismutation) can yield HO• via the ‘classical’ Fenton

pathway (Fe2+ + H2O2k=103−105Ms−1→ Fe3+ + OH− + HO•),

which is plausible because our in vitro Fenton systemyielded identical DEPMPO-OH signals. In support,exercise-induced oxidative stress has been associated withan increase in bleomycin-detectable iron (Jenkins et al.1993; Childs et al. 2001), and iron chelation therapyhas been shown to provide effective prophylaxis (Smithet al. 1989). Alternatively, decomposition of protonatedONOO− (Pou et al. 1995) subsequent to the oxidativeinactivation of NO cannot be excluded as an alternativesource of DEPMPO-OH, given the exercise-inducedincrease in 3-NT. It should be noted, however, that whenDEPMPO-OH results from NO-based chemistry the cisadduct ratio is far lower (i.e. ∼10%; Culcasi et al. 2006a)than that reported here.

Instead of DEPMPO-OH, postexercise samples fromtwo subjects contained only weak DEPMPO-R signalswhere the trapped species is probably not an HBC radicalfragment, but may represent a lipid carbon-centredradical (LC•). This may be a short-chain radical, suchas the ethyl or pentyl species formed during β-scissionof LO• (Gardner, 1989), the primary species trapped

by PBN. This is consistent with our previous exercisestudies, where low levels of PBN-LC• were detected, andwe have suggested that these radicals may have evolvedduring the metal-catalysed reductive decomposition oflipid hydroperoxides (LOOH) formed subsequent to freeradical-mediated damage to membrane phospholipids

(Fe2+ + LOOH → Fe3+ + OH− + LO• β - scission→ LC•;Bailey et al. 2004, 2007, 2010). Though clearly downstreamof the primary reaction pathway, these second-generationradicals are still eminently capable of initiating lipidperoxidation (�Eo’ ≈ +1000–1300 mV versus PUFA-H/PUFA-OOH; Buettner, 1993). Considering the verylow stability of PBN-HO• adduct versus DEPMPO-OHin aqueous media (Janzen et al. 1992) combined with thegreater ability of PBN to scavenge free radicals in a lipidicenvironment (PBN is ∼90 times more lipophilic thanDEPMPO; Gosset et al. 2011), our study illustrates thatthe combination of DEPMPO (+HBC) and PBN spintraps can be utilised to provide unique complementaryinsight into the mechanisms underlying exercise-inducedoxidative stress.

We recently documented a positive relationshipbetween the systemic accumulation of A•− and increase in

Figure 3Relationships observed between the exercise-induced increase in the ascorbate radical (A•−; A), 5-(diethoxyphosphoryl)-5-methyl-1-pyrolline N-oxide (DEPMPO; B), N-tert-butyl-α-phenylnitrone (PBN) adducts (C),S100β (D) and the decrease in the autoregulation index (ARI).

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

Exp Physiol 96.11 pp 1196–1207 Redox regulation of blood–brain barrier function 1205

low frequency transfer gain (impaired dCA) in volunteersexposed to hypoxia (Bailey et al. 2009b). The presentstudy extends these findings to the exercising humans,but we acknowledge the limitations associated withARI (as is arguably the case with other measures)as an index of dCA (Willie et al. 2011) and theunavoidable time delay because measurements had tobe performed during the postexercise recovery period.Precisely why OX-NOX stress impaired dCA remainsto be established, given that the combined effects of(presumed) sympathetic activation and hyperventilation-induced hypocapnia would have been expected to improvedCA (Levine et al. 1994; Panerai et al. 1999), and the lackof change in MCAv, despite its acknowledged limitations(Willie et al. 2011), argues against a corresponding increasein CBF.

Redox regulation of dCA and BBB function

While not disassociating cause from effect, therelationships observed between the various spin adductsand ARI tentatively suggest a direct role for free radicals.In support of this, Zagorac et al. (2005) identified thatsuperfusion of the rodent brain with O•−

2 impaired dCAsubsequent to activation of KCa channels in cerebralvascular smooth muscle cells. The conflicting reports ofBrys et al. (2003; intact dCA during moderate exercise)and Ogoh et al. (2005a; impaired dCA during intenseexercise) may therefore be explained by differences inthe magnitude of the OX-NOX stress response, which isknown to increase with exercise intensity (Fogarty et al.2010), implying that an as of yet undefined ‘threshold’may exist before dCA becomes impaired. It would seemunlikely that the oxidative inactivation and subsequentreduction in the vascular bioavailability of NO (confirmedby the increase in 3-NT) played a significant role as haspreviously been suggested (White et al. 2000), given thatthe systemic NO metabolite pool was largely preservedand the recent observation that NO synthase blockade didnot affect dCA in humans (Zhang et al. 2004).

Release of the astrocytic protein S100β to the systemiccirculation correlates directly with the extent and temporalsequence of BBB opening, as confirmed by contrast-enhanced magnetic resonance imaging, and is thusconsidered a reliable blood-borne biomarker of barrierintegrity (Kanner et al. 2003). We combined this with themeasurement of NSE, which is found almost exclusivelyin neuronal cell bodies, axons and neuroendocrine cells(Schmechel et al. 1978), to distinguish potential BBB‘leak’ from neuronal–parenchymal ‘damage’ (Marchi et al.2003). The observed rise in S100β and lack of changein NSE in the present study therefore suggest thatexercise increased BBB permeability without causingstructural brain tissue injury, though we cannot dismisspotential contamination from ‘extracerebral’ sources, such

as adipose and skeletal tissue (Zimmer et al. 1995).However, the inverse relationship observed between S100βand ARI suggests that the increase in S100β is authentic(i.e. BBB related), in that impaired dCA may havepreceded BBB disruption, which is conceivable, giventhat intense exercise can increase systolic blood pressurebeyond the autoregulatory range in the brain. Giventhe complications associated with the interpretation ofS100β during exercise (Hasselblatt et al. 2004), follow-up studies employing gadolinium-enhanced diffusion-weighted magnetic resonance imaging are warranted toquantify the extent of barrier disruption more accuratelyand to determine whether this is accompanied byvasogenic oedema. Such oedema would be expected topromote an O2 diffusion limitation, which may pose acentral limitation to maximal exercise performance byinhibiting cortical activation of efferent motor neurons(Amann et al. 2006).

In conclusion, the present findings suggest that anacute bout of maximal exercise has the potential toincrease BBB permeability subsequent to a free radical-mediated impairment in dCA, which persists into therecovery period. What health implications this poses inthe long term remain unclear, though recent evidencesuggests that the cerebral formation of free radicals andBBB disruption may represent adaptive stress responses(Bailey et al. 2009a, 2011). Future studies incorporatingthe targeted delivery of antioxidants to the human braincombined with transfer function analysis to assess dCA‘real time’ during the exercise challenge will ultimatelyconfirm whether these events are indeed subject to redoxregulation.

References

Aaslid R, Lindegaard K, Sorteberg W & Nornes H (1989).Cerebral autoregulation dynamics in humans. Stroke 20,45–52.

Adamski A, Spalek T & Sojka A (2003). Application of EPRspectroscopy for elucidation of vanadium speciation inVOx/ZrO2 catalysts subject to redox treatment. Research onChemical Intermediates 29, 793–804.

Amann M, Eldridge MW, Lovering AT, Stickland MK, PegelowDF & Dempsey JA (2006). Arterial oxygenation influencescentral motor output and exercise performance via effects onperipheral locomotor muscle fatigue in humans. J Physiol575, 937–952.

Aries MJ, Elting JW, De Keyser J, Kremer BP & Vroomen PC(2010). Cerebral autoregulation in stroke: a review oftranscranial Doppler studies. Stroke 41, 2697–2704.

Bailey DM, Bartsch P, Knauth M & Baumgartner RW (2009a).Emerging concepts in acute mountain sickness andhigh-altitude cerebral edema: from the molecular to themorphological. Cell Mol Life Sci 66, 3583–3594.

Bailey DM, Davies B & Baker JS (2000). Training in hypoxia:modulation of metabolic and cardiovascular risk factors inmen. Med Sci Sports Exerc 32, 1058–1066.

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

1206 D. M. Bailey and others Exp Physiol 96.11 pp 1196–1207

Bailey DM, Evans KA, James PE, McEneny J, Young IS, Fall L,Gutowski M, Kewley E, McCord JM, Moller K & Ainslie PN(2009b). Altered free radical metabolism in acute mountainsickness: implications for dynamic cerebral autoregulationand blood–brain barrier function. J Physiol 587,73–85.

Bailey DM, Lawrenson L, McEneny J, Young IS, James PE,Jackson SK, Henry R, Mathieu-Costello O, McCord JM &Richardson RS (2007). Electron paramagnetic spectroscopicevidence of exercise-induced free radical accumulation inhuman skeletal muscle. Free Radic Res 41,182–190.

Bailey DM, McEneny J, Mathieu-Costello O, Henry RR, JamesPE, McCord JM, Pietri S, Young IS & Richardson RS (2010).Sedentary aging increases resting and exercise-inducedintramuscular free radical formation. J Appl Physiol 109,449–456.

Bailey DM, Taudorf S, Berg RMG, Lundby C, McEneny J,Young IS, Evans KA, James PE, Shore A, Hullin DA, McCordJM, Pedersen BK & Moller K (2009c). Increased cerebraloutput of free radicals during hypoxia: implications for acutemountain sickness? Am J Physiol Regul Integr Comp Physiol297, R1283–R1292.

Bailey DM, Taudorf S, Berg RM, Lundby C, Pedersen BK,Rasmussen P & Moller K (2011). Cerebral formation of freeradicals during hypoxia does not cause structural damageand is associated with a reduction in mitochondrial PO2;evidence of O2-sensing in humans? J Cereb Blood Flow Metab31, 1020–1026.

Bailey DM, Young IS, McEneny J, Lawrenson L, Kim J, Barden J& Richardson RS (2004). Regulation of free radical outflowfrom an isolated muscle bed in exercising humans. Am JPhysiol Heart Circ Physiol 287, H1689–H1699.

Barbati S, Clement JL, Frejaville C, Boutellier JC, Tordo P,Michel JC & Yadan JC (1999). 31P-labeled pyrrolineN-oxides: synthesis of 5-diethylphosphone-5-methyl-1-pyrroline N-oxide (DEPMPO) by oxidation ofdiethyl (2-methylpyrolidin-2-yl)phosphonate. Synthesis 12,2036–2040.

Brys M, Brown CM, Marthol H, Franta R & Hilz MJ (2003).Dynamic cerebral autoregulation remains stable duringphysical challenge in healthy persons. Am J Physiol Heart CircPhysiol 285, H1048–1054.

Buettner GR (1993). The pecking order of free radicals andantioxidants: lipid peroxidation, α-tocopherol, andascorbate. Arch Biochem Biophys 300,535–543.

Catignani GL & Bieri JG (1983). Simultaneous determinationof retinol and α-tocopherol in serum or plasma by liquidchromatography. Clin Chem 29, 708–712.

Chamulitrat W (1999). Activation of the superoxide-generatingNADPH oxidase of intestinal lymphocytes produces highlyreactive free radicals from sulfite. Free Radic Biol Med 27,411–421.

Childs A, Jacobs C, Kaminski T, Halliwell B & Leeuwenburgh C(2001). Supplementation with vitamin C andN-acetyl-cysteine increases oxidative stress in humans afteran acute muscle injury induced by eccentric exercise. FreeRadic Biol Med 31, 745–753.

Culcasi M, Muller A, Mercier A, Clement JL, Payet O,Rockenbauer A, Marchand V & Pietri S (2006a). Earlyspecific free radical-related cytotoxicity of gas phase cigarettesmoke and its paradoxical temporary inhibition by tar: anelectron paramagnetic resonance study with the spin trapDEPMPO. Chem Biol Interact 164, 215–231.

Culcasi M, Rockenbauer A, Mercier A, Clement JL & Pietri S(2006b). The line asymmetry of electron spin resonancespectra as a tool to determine the cis:trans ratio forspin-trapping adducts of chiral pyrrolines N-oxides: themechanism of formation of hydroxyl radical adducts ofEMPO, DEPMPO, and DIPPMPO in the ischemic-reperfused rat liver. Free Radic Biol Med 40,1524–1538.

Dalsgaard MK, Ogoh S, Dawson EA, Yoshiga CC, Quistorff B &Secher NH (2004). Cerebral carbohydrate cost of physicalexertion in humans. Am J Physiol Regul Integr Comp Physiol287, R534–R540.

Duling DR (1994). Simulation of multiple isotropic spin-trapEPR spectra. J Magn Reson B 104, 105–110.

Edwards MR, Martin DH & Hughson RL (2002). Cerebralhemodynamics and resistance exercise. Med Sci Sports Exerc34, 1207–1211.

Ewing JF & Janero DR (1998). Specific S-nitrosothiol(thionitrite) quantification as solution nitrite aftervanadium(III) reduction and ozone-chemiluminescentdetection. Free Radic Biol Med 25, 621–628.

Fogarty MC, Hughes CM, Burke G, Brown JC, Trinick TR,Duly E, Bailey DM & Davison GW (2010). Exercise-inducedlipid peroxidation: implications for deoxyribonucleic aciddamage and systemic free radical generation. Environ MolMutagen 52, 35–42.

Frejaville C, Karoui H, Tuccio B, Le Moigne F, Culcasi M, PietriS, Lauricella R & Tordo P (1995). 5-(Diethoxyphosphoryl)-5-methyl-1-pyrroline N-oxide: a new efficient phosphorylatednitrone for the in vitro and in vivo spin trapping ofoxygen-centered radicals. J Med Chem 38, 258–265.

Gardner HW (1989). Oxygen radical chemistry ofpolyunsaturated fatty acids. Free Radic Biol Med 7,65–86.

Gosset G, Clement JL, Culcasi M, Rockenbauer A & Pietri S(2011). CyDEPMPOs: a class of stable cyclic DEPMPOderivatives with improved properties as mechanistic markersof stereoselective hydroxyl radical adduct formation inbiological systems. Bioorg Med Chem 19, 2218–2230.

Hasselblatt M, Mooren FC, von Ahsen N, Keyvani K, FrommeA, Schwarze-Eicker K, Senner V & Paulus W (2004). SerumS100β increases in marathon runners reflect extracranialrelease rather than glial damage. Neurology 62, 1634–1636.

Hausladen A, Rafikov R, Angelo M, Singel DJ, Nudler E &Stamler JS (2007). Assessment of nitric oxide signals bytriiodide chemiluminescence. Proc Natl Acad Sci USA 104,2157–2162.

Janzen EG, Kotake Y & Hinton RD (1992). Stabilities ofhydroxyl radical spin adducts of PBN-type spin traps. FreeRadic Biol Med 12, 169–173.

Jenkins R, Krause K & Schofield L (1993). Influence of exerciseon clearance of oxidant stress products and loosely boundiron. Med Sci Sports Exerc 25, 213–217.

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society

Exp Physiol 96.11 pp 1196–1207 Redox regulation of blood–brain barrier function 1207

Kanner AA, Marchi N, Fazio V, Mayberg MR, Koltz MT,Siomin V, Stevens GH, Masaryk T, Ayumar B, VogelbaumMA, Barnett GH & Janigro D (2003). Serum S100β: anoninvasive marker of blood-brain barrier function andbrain lesions. Cancer 97, 2806–2813.

Karoui H, Rockenbauer A, Pietri S & Tordo P (2002). Spintrapping of superoxide in the presence of beta-cyclodextrins.Chem Commun (Camb) 21, 3030–3031.

Koppenol WH & Butler J (1985). Energetics of interconversionreactions of oxyradicals. Advances in Free Radical Biology andMedicine 1, 91–131.

Levine BD, Giller CA, Lane LD, Buckey JC & Blomqvist CG(1994). Cerebral versus systemic hemodynamics duringgraded orthostatic stress in humans. Circulation 90, 298–306.

McDowell I, McEneny J & Trimble E (1995). A rapid methodfor measurement of the susceptibility to oxidation oflow-density lipoprotein. Ann Clin Biochem 32, 167–174.

Marchi N, Rasmussen P, Kapural M, Fazio V, Kight K, MaybergM, Kanner A, Ayumar B, Albensi B, Cavaglia M & Janigro D(2003). Peripheral markers of brain damage and blood-brainbarrier dysfunction. Restor Neurol Neurosci 21, 109–121.

Nauser T & Koppenol WH (2002). The rate constant of thereaction of superoxide with nitrogen monoxide: approachingthe diffusion limit. Journal of Physical Chemistry A 106,4084–4086.

Ogoh S, Dalsgaard MK, Yoshiga CC, Dawson EA, Keller DM,Raven PB & Secher NH (2005a). Dynamic cerebralautoregulation during exhaustive exercise in humans. Am JPhysiol Heart Circ Physiol 288, H1461–H1467.

Ogoh S, Fadel PJ, Zhang R, Selmer C, Jans O, Secher NH &Raven PB (2005b). Middle cerebral artery flow velocity andpulse pressure during dynamic exercise in humans. Am JPhysiol Heart Circ Physiol 288, H1526–H1531.

Pacher P, Beckman JS & Liaudet L (2007). Nitric oxide andperoxynitrite in health and disease. Physiol Rev 87, 315–424.

Panerai RB (2009). Transcranial Doppler for evaluation ofcerebral autoregulation. Clin Auton Res 19, 197–211.

Panerai RB, Deverson ST, Mahony P, Hayes P & Evans DH(1999). Effects of CO2 on dynamic cerebral autoregulationmeasurement. Physiol Meas 20, 265–275.

Paulson OB, Strandgaard S & Edvinsson L (1990). Cerebralautoregulation. Cerebrovasc Brain Metab Rev 2, 161–192.

Pott F, Knudsen L, Nowak M, Nielsen HB, Hanel B & SecherNH (1997). Middle cerebral artery blood velocity duringrowing. Acta Physiol Scand 160, 251–255.

Pou S, Nguyen SY, Gladwell T & Rosen GM (1995). Doesperoxynitrite generate hydroxyl radical? Biochim BiophysActa 1244, 62–68.

Powers SK & Jackson MJ (2008). Exercise-induced oxidativestress: cellular mechanisms and impact on muscle forceproduction. Physiol Rev 88, 1243–1276.

Rockenbauer A & Korecz L (1996). Automatic computersimulation of ESR spectra. Applied Magnetic Resonance 10,29–43.

Rogers SC, Khalatbari A, Datta BN, Ellery S, Paul V, FrenneauxMP & James PE (2007). NO metabolite flux across thehuman coronary circulation. Cardiovasc Res 75, 434–441.

Schmechel D, Marangos PJ & Brightman M (1978).Neurone-specific enolase is a molecular marker forperipheral and central neuroendocrine cells. Nature 276,834–836.

Secher NH, Seifert T & Van Lieshout JJ (2008). Cerebral bloodflow and metabolism during exercise: implications forfatigue. J Appl Physiol 104, 306–314.

Smith J, Carden D, Grisham M, Granger D & Korthuis R(1989). Role of iron in postischemic microvascular injury.Am J Physiol Heart Circ Physiol 256, H1472–H1477.

Thurnham DI, Smith E & Flora PS (1988). Concurrentliquid-chromatographic assay of retinol, α-tocopherol,β-carotene, α-carotene, lycopene, and β-cryptoxanthin inplasma, with tocopherol acetate as internal standard. ClinChem 34, 377–381.

Tiecks FP, Lam AM, Aaslid R & Newell DW (1995).Comparison of static and dynamic cerebral autoregulationmeasurements. Stroke 26, 1014–1019.

Tortolani AJ, Powell SR, Misik V, Weglicki WB, Pogo GJ &Kramer JH (1993). Detection of alkoxyl and carbon-centeredfree radicals in coronary sinus blood from patientsundergoing elective cardioplegia. Free Radic Biol Med 14,421–426.

von Sonntag C (1987). The Chemical Basis of Radiation Biology.Taylor & Francis, London.

Vuilleumier JP & Keck E (1993). Fluorimetric assay ofvitamin C in biological materials using a centrifugal analyserwith fluorescence attachment. Journal of MicronutritionalAnalysis 5, 25–34.

Wang J, Brown MA, Tam SH, Chan MC & Whitworth JA(1997). Effects of diet on measurement of nitric oxidemetabolites. Clin Exp Pharmacol Physiol 24, 418–420.

White RP, Vallance P & Markus HS (2000). Effect of inhibitionof nitric oxide synthase on dynamic cerebral autoregulationin humans. Clin Sci (Lond) 99, 555–560.

Williams NH & Yandell JK (1982). Outer-sphereelectron-transfer reactions of ascorbic anions. AustralianJournal of Chemistry 35, 1133–1144.

Willie CK, Colino FL, Bailey DM, Tzeng YC, Binsted G, JonesLW, Haykowsky MJ, Bellapart J, Ogoh S, Smith KJ, Smirl JD,Day TA, Lucas SJ, Eller LK & Ainslie PN (2011). Utility oftranscranial Doppler ultrasound for the integrativeassessment of cerebrovascular function. J Neurosci Methods196, 221–237.

Zagorac D, Yamaura K, Zhang C, Roman RJ & Harder DR(2005). The effect of superoxide anion on autoregulation ofcerebral blood flow. Stroke 36, 2589–2594.

Zhang R, Wilson TE, Witkowski S, Cui J, Crandall CG & LevineBD (2004). Inhibition of nitric oxide synthase does not alterdynamic cerebral autoregulation in humans. Am J PhysiolHeart Circ Physiol 286, H863–H869.

Zimmer DB, Cornwall EH, Landar A & Song W (1995). TheS100 protein family: history, function, and expression. BrainRes Bull 37, 417–429.

Acknowledgements

We are grateful to the subjects for their commitment to this studyand for informative discussions with Professor Damir Janigro ofthe Cleveland Clinic Lerner College of Medicine (Cleveland, OH,USA). D.M.B. was supported by a grant from the University ofProvence (Laboratoire Chimie Provence, UMR CNRS 6264).

C© 2011 The Authors. Journal compilation C© 2011 The Physiological Society