It has been previously shown that BMP2 increases Akt serine/threonine kinase activity in serum-deprived 2T3 cells ^[10], and Akt/PKB signaling is known to activate mTOR pathway. Importantly, we have recently demonstrated that increasing oxygen tension induces activation of endogenous BMP pathway, through SMAD1/5/8 activation, in GBM-derived cells ^[3]. We sought to investigate if a progressive time dependent exposure of GBM cells, that have been constantly maintained under hypoxia (2% oxygen), to an acute 20% oxygen tension, was promoting Akt/mTOR signaling pathway activation. We observed activation of Akt at the level of threonine 308, but not at the level of serine 473 (data not shown), in a time dependent fashion following high oxygen exposure (Figure 1A,C). Also, mTOR phosphorylation at the level of serine 2448 was induced by high oxygen exposure (Figure 1A,D). Under constant hypoxia these activations were inhibited. mTOR is known to regulate several biological cell responses, amongst them: i) translation, through activation of p70S6-Kinase (p70S6K) and inhibition of the inhibitor 4eBP1, ii) cell proliferation, by acting as a cell cycle regulator, iii) cell survival and cell differentiation, by Stat3 activation, iv) angiogenesis, through activation of HIF-1? and Vascular Endothelial Growth Factor (VEGF). We found that all mTOR downstream targets were activated by acute high oxygen exposure in a time dependent manner. Total Akt/mTOR proteins analyses indicated a homogenous expression among conditions both in GBM and normal cells progressively exposed to high oxygen tension (Figure 1A,B). Particularly, Stat3 (Ser727) was activated after 30 min of high oxygen exposure (Figure 1A,E), this indicating gliogenesis ^[8] and/or activation of pro-survival response ^[11]. Moreover, while 4eBP1 was only modestly inhibited by oxygen (data not shown), which may be due to alternative regulatory pathways involved in 4eBP1 regulation, p70S6K (Thr389) was highly up-regulated especially after 120 min (Figure 1A,F). These results indicate that activation of Akt/mTOR dependent pro-translational pathways (p70S6K) and gliogenic and pro-survival mechanisms (Stat3) occur in response to acute high oxygen exposure. Moreover, this increase in translation seems to be directed toward cell differentiation, as indicated by increased p21^cip1 and increased endogenous BMP dependent astroglial commitment as we reported earlier ^[3]. In normal SVZ-derived cells Akt was only modestly and not significantly activated by oxygen exposure, while mTOR activation occurred after 1 hr; although mTOR downstream targets Stat3 and p70S6k were not activated following mTOR phosphorylation (Figure 1B?F). These results indicate differences in normal and tumour cell response to high oxygen tension exposure.

Another protein controlled by mTOR is HIF-1? ^[6]. We found that HIF-1? was rapidly degraded by acute high oxygen exposure, in both GBM and normal SVZ cells (Figure 2A?C). Although, a modest recovery of HIF-1? protein was visible after 2 hr of high oxygen exposure in GBM cells, but not in normal cells (Figure 2A?C), indicating that tumour cells may re-establish HIF-1? level through a hypoxic independent mechanism, possibly controlled by mTOR progressive activation.

We also analyzed REDD1 (RTP801), which has been shown to be strongly induced under hypoxic conditions in a HIF-1? dependent manner ^[14] and recent studies also suggest that REDD1 plays a role in the TSC1/TSC2-mediated inhibition of mTOR ^[15]. Accordingly, we found REDD1 downregulated in GBM cells, less intensively in normal SVZ cells, following acute high oxygen exposure (Figure 2A,B,D). Thus, high oxygen dependent Akt/mTOR signaling activation may occur after HIF-1?/REDD1 downregulation.

To understand if HIF-1? mediates the repressive effect of low oxygen on Akt/mTOR signaling, we silenced HIF-1? using a lentiviral vector containing siHIF-1? along with enhanced green fluorescent protein as indicator of efficiency of transduction (siHIF-1?-EGFP) (Figure 3A,D), which was compared to a siLuciferase-EGFP vector (siLUC-EGFP) as a negative control ^[3] (not shown). By silencing HIF-1? in GBM cells a strong differentiation and eventually cell death occurred after 1 week, as already reported in our previous work ^[3] and these effects were not observed with siLUC-EGFP. Importantly, Akt, mTOR, Stat3 and p70S6K were activated even in HIF-1? silenced cells cultured at 2% oxygen compared to control group and to GBM cells transduced with siLUC-EGFP vector (Figure 3B,C). In normal SVZ cells HIF-1? silencing did not elicit any significant effect and eventually Akt and mTOR inhibition occurred compared to control (Figure 3E,F). HIF-1? silencing correlates with a stronger BMP pathway activity in GBM cells, as shown by SMAD1/5/8 phosphorylation ^[3], and BMP pathway is known to activate Akt ^[10] and to be correlated to Stat3 regulation in promoting astrogliogenesis ^[12]. These results indicate that HIF-1? in hypoxic tumor cells may be required to repress convergent signals (SMAD1/5/8 and Stat3) directed to promote an astroglial fate. To prove that HIF-1? downregulation is required in evoking high oxygen dependent Akt/mTOR activation we stabilized HIF-1? by using cobalt chloride (CoCl₂, 100 ?M), which mimics the effect of hypoxia on HIF-1?, and 12 hr later we exposed cells to high oxygen tension. Performing these experiments directly on HIF-1? silenced cells was not feasible given the scarce viability following HIF-1? silencing. We found that by chemically stabilizing HIF-1?, REDD1 was upregulated and Akt/mTOR pathway was maintained inhibited in GBM derived cells exposed to high oxygen tension (Figure 3G), only p70S6K activation did occur and this may depend on effectors of p70S6K activation alternative to mTOR. Importantly, normal SVZ cells responded in a different way; indeed, by chemically stabilizing HIF-1?, REDD1 was transiently upregulated and Akt/mTOR pathway was not inhibited, unlike in tumor cells, following exposure to high oxygen tension (Figure 3H), and Akt and Stat3 were eventually upregulated by CoCl₂. These results indicate that HIF-1? stabilization, probably through REDD1, prevents high oxygen induced Akt/mTOR activation, and this seems to occur more specifically in GBM derived cells.

We sought to investigate if exogenous BMP2, alike acute high oxygen exposure, could affect Akt/mTOR pathway activation. While total Akt/mTOR signaling proteins resulted to be homogenously expressed both in GBM and normal cells after BMP2 treatment (Figure 4A and Figure S1A), we found that activation of Akt (Thr308) and also mTOR (Ser2448) occurred with time (Figure 4A,C,D), and, importantly, this activation was accelerated and improved under acute high oxygen exposure, remaining highly activated even after a long term BMP2 exposure (72 hr) (Figure 4B). Importantly, under hypoxia Akt and mTOR were inhibited after short time treatment, resulting activated after 72 hr of BMP2 stimulus. In normal SVZ cells we found that Akt and mTOR were only transiently modulated by BMP2 addition (Figure S1A,C,D) and after 72 hr these activations resulted down-regulated (data not shown). Analyses of mTOR downstream targets, revealed a time dependent increase of Stat3 (Ser727) and p70S6K (Thr389) activation following BMP2 treatment, and these effects were decelerated and decreased by maintaining cells under hypoxia (Figure 4A,E,F), and this occurred transiently also in normal cells (Figure S1A,E,F). Also p21^cip1, involved in cell cycle arrest and induction of differentiation, was found up-regulated by BMP2, but hypoxia inhibited this effects (data not shown).

HIF-1? expression is known to be controlled by mTOR ^[6] and REDD1, mTOR inhibitor, is transcriptionally activated by HIF-1? ^[14]. We analyzed if HIF-1? protein was affected following a progressive BMP2 treatment and we found that HIF-1? was strongly down-regulated even under hypoxia following 15 min of BMP2 stimulation in GBM cells (Figure 5A,B). Importantly, HIF-1? level was recovered after 120 min, however not when GBM cells were also acutely exposed to high oxygen. This later stabilization may be consequential to the BMP2 mediated mTOR signaling activation (Figure 4A,B). According to our recent work ^[3], HIF-1? level was reduced after a longer BMP2 exposure (72 hrs) (not shown), which is likely correlated to BMP2 dependent glial differentiation in GBM cells. Oppositely, maintenance of HIF-1? expression seems to correlate to tumour cells de-differentiation and normal cells primitiveness ^[5], ^[13]. In normal SVZ-derived cells HIF-1? was transiently down-regulated following BMP2 stimulation and with time it was recovered (Figure S1A,G), but its expression did not change after long term (72 hr) treatment (data not shown). Importantly, also REDD1 was down-regulated by BMP2 under hypoxia (Figure 5A,C), and this was only transiently occurring in normal cells (Figure S1A). This suggests that BMP2, analogously to an acute high oxygen exposure, may promote mTOR activation by down-modulating inhibitory REDD1.

As REDD1 is transcriptionally activated by HIF-1? ^[14], we investigated if BMP2 besides modulating HIF-1? protein, promoted also inhibition of HIF-1? transcriptional activity, by using a hypoxia responsive element (HRE)-luciferase reporter construct. Despite a physiologic tumour samples variability, we recorded a 40% reduction of HIF-1? transcriptional activity under hypoxia following 8 hr of BMP2 treatment compared to untreated cells. Conversely, in normal SVZ cells BMP2 was operating in an opposite way, promoting a nearly 10% increase of HIF-1? activity (Figure 5D), pointing to differences between normal and tumor cells in BMP2 responsiveness.

We also tested if rapamycin treatment alone or in combination with BMP2 to evaluate if blocking mTOR downstream signaling affected also HIF-1? stability. We found that 72 hr of rapamycin treatment, which did not affect HIF-1? transcriptional activition (data not shown), induced a less pronounced HIF-1? protein reduction, in accordance with reported results ^[16], whereas in combination with BMP2 these effects were not further improved (Figure 5E). Normal SVZ cells responded in a similar fashion (data not shown).

To prove that Akt/mTOR activation was due to inhibition of HIF-1?/REDD1 by BMP2, we performed BMP2 treatment on CoCl₂ pretreated GBM cells. We found that by stabilizing HIF-1? and consequentially also REDD1, Akt/mTOR signaling was not activated following BMP2 treatment (Figure 6A). p70S6K activation lately occurred, probably due to effectors of p70S6K activation alternative to mTOR, as commented above, and Stat3 (S727) was eventually more stably expressed in CoCl₂ treated cells, although it was not modulated by BMP2 during time (Figure 6A). Importantly, in normal SVZ cells, chemically HIF-1? stabilization was only transiently inducing REDD1 upregulation and Akt/mTOR pathway was not inhibited, unlike in tumor cells, following BMP2 treatment (Figure S1B), and Akt and Stat3 were eventually upregulated. Alternatively to the use of CoCl₂, we used the proteasome inhibitor Z-LLF-CHO (Z-LLF, 30 ?M) added 30 min prior to 8 or 24 hr of BMP2 time treatment under hypoxia. By preventing proteasomal degradation, BMP2 did not affect HIF-1? protein stability (Figure 6B). Moreover, REDD1 was upregulated and Akt/mTOR signaling was not activated following BMP2 treatment under hypoxia (Figure 6B). These results confirm that HIF-1? degradation and consequentially also REDD1 downregulation are required in BMP2 dependent Akt/mTOR activation, and this preferentially occurs in GBM cells.

As we saw that HIF-1? protein stability was affected following BMP2 treatment, we investigated if proline hydroxylases (PHDs), involved in HIF-1? proline hydroxylation and consequential proteasomal degradation, were modulated by BMP2. PHD2 in particular has been described as the critical oxygen sensor setting the low steady-state levels of HIF-1? in normoxia ^[17]. In the same work PHD2 was found up-regulated by hypoxia, providing a HIF-1-dependent auto-regulatory mechanism driven by oxygen tension itself. We found that PHD2 protein was rapidly up-regulated by BMP2, and this occurred more slowly under hypoxia, in both GBM (Figure 7A,C) and normal SVZ cells (Figure S1A,H). Also after a prolonged BMP2 treatment PHD2 protein resulted up-regulated in GBM cells (Figure 7B), which is in accordance to our previous work ^[3]. QRT-PCR revealed that while PHD2 transcript was up-regulated under hypoxia, according to literature ^[17], 72 hr of BMP2 treatment induced a modest but not significant PHD2 mRNA decrease (Figure 7D). PHD2 transcript reduction was recorded also in BMP2 treated normal SVZ cells (Figure S1I). Notably, PHD2 promoter is characterized by the presence of HRE consensus sequences directly controlled by HIF-1? ^[18]. Thus, we hypothesize that the observed PHD2 mRNA reduction may depend on HIF-1? transcriptional inhibition induced by BMP2, as previously described (Figure 5D).

Following-up on our observation that BMP2 mediated PHD2 protein increase was not dependent on de novo protein synthesis (Figure 7D), we sought to investigate if BMP2 was somehow increasing PHD2 protein stability. Peptidyl prolyl cis/trans isomerase FK506-binding protein 38 (FKBP38) has been described as a PHD2 protein regulator by targeting PHD2 to proteasome degradation ^[19]. We found that FKBP38 was progressively down-regulated by BMP2 under hypoxia (Figure 7A,E), and this effect was observed also after a prolonged BMP2 treatment (Figure 7B). Conversely to what has been described ^[19], we also found that FKBP38 was up-regulated in hypoxic GBM cells, suggesting that it may be involved in hypoxic HIF-1? stability by promoting PHD2 degradation. QRT-PCR analysis confirmed that FKBP38 was down-regulated by acute exposure to high oxygen and by BMP2 under hypoxia also at the transcriptional level (Figure 7F). Opposite, this was not occurring in normal SVZ-derived cells in which BMP2 treatment combined with high oxygen exposure was inducing FKBP38 up-regulation (Figure S1J), also at the protein level (Figure S1A,K). Thus, we hypothesize that BMP2 may promote PHD2 stabilization by down-modulating FKBP38 expression in GBM cells.

We finally sought to investigate if BMP2 treatment increased PHD2 activity thus causing HIF-1? modulation. It has been previously reported that impaired SDH activity in several cancer types is associated to HIF-1? stabilization, through a mechanism involving intra-cytoplasmic succinate accumulation and consequential PHD2 inhibition ^[20]. Indeed, SDH is the enzyme complex II bound to the inner mitochondrial membrane that converts succinate to fumarate via FAD reduction to FADH2. Importantly, it has also been shown in other cell models that induction of differentiation by BMPs increases mitochondrial oxidative phosphorylation, as seen by higher SDH activity ^[21]. We found that BMP2 treatment induced increase of SDH activity in GBM cells (Figure 8A,B). This suggests that pro-differentiating agents, such as BMPs, may promote a metabolic shift toward oxidative phosphorylation in tumor cells and that BMP2, by decreasing intracellular succinate through induction of SHD activation, may increase PHD2 activity leading to HIF-1? modulation. Notably, in normal SVZ cells SDH activation was not changed following BMP2 treatment (Figure S2A), this indicating differences in metabolic response to BMP2 treatment between normal and tumor cells. Further addition of exogenous esterificated diethyl-succinate (5 mM, Sigma) either in presence or absence of BMP2 did not induce SDH activation (not shown).

We also tested if additional succinate in combination with BMP2, was promoting a recovery of HIF-1? protein level and/or determining an inhibition of Akt/mTOR signaling in GBM cells. We found that HIF-1?, REDD1 and FKBP38 proteins were upregulated in BMP2/succinate treated cells, even when exposed to 20% oxygen. Moreover, in particular Akt and p70S6K activation were modestly decreased in presence of combined BMP2 and succinate, as compared to BMP2 only treated cells (Figure 8C). This indicates that additional succinate, by upregulating HIF-1? and REDD1, partially inhibits Akt/mTOR activation following BMP2 treatment.

Normal precursor cells were derived from brain subventricular zone (SVZ) tissue of 3 premature neonates who died shortly after birth (HuSC23, HuSC27 and HuSC30), previously characterized ^[34] and hereafter termed normal SVZ precursors (see Table 1).

Ethics Statement: Written informed consent for the donation of neonatal brain SVZ tissues was obtained from parents prior to tissue acquisition, under the auspices of the protocol for the National Human Neural Stem Cell Resource (NHNSCR) approved by the Children's Hospital of Orange County Institutional Review Board. All tissues were acquired in compliance with NIH and institutional guidelines.

GBM precursors were derived from 7 tumors taken at surgery (see Table 1); initial pathological review was followed by secondary neuropathological review to reconfirm diagnosis. Ethics Statement: Written informed consent for the donation of adult tumor brain tissues was obtained from patients prior to tissue acquisition, under the auspices of the protocol for the acquisition of human brain tissues obtained from the Ethical Committee board of the University of Padova and Padova Academic Hospital. All tissues were acquired following the tenets of the Declaration of Helsinki.

We dissociated ^[35] and cultured cells on fibronectin-coated dishes in DMEM/F12 (Irvine Scientific, Irvine, CA) supplemented with BIT9500 (1% bovine serum albumin, 10 ?g/ml insulin, 200 ?g/ml transferrin; Stem Cell Technologies, CA), 20 ng/ml basic fibroblast growth factor, (bFGF) and 20 ng/ml epidermal growth factor (EGF, both human and from R&D Systems, Minneapolis, MN) as previously described ^[36]. In vitro works are generally not fully representative of a reliable physiological condition. Although, a way to mimic tumor cell microenvironment is providing a lowered oxygen tension, which is characteristic of the solid tumor mass ^[9]; for this reason we cultured cells in an atmosphere of 2%, 5% carbon dioxide and the balance nitrogen, as previously described ^[36], ^[37]. Analyses of some metabolic parameters (i.e. lactate release and cytochrome c oxidase (COX) activity) indicate that, despite the use of high glucose concentration in our cell culture conditions (25 mM glucose), metabolic differences between tumor and normal cells are clearly detectable (Figure S2B,C). For continuous expansion, one-half of the medium was replaced every day and cultures were passaged every 7?10 days using TrypLE (Invitrogen, Carlsbad, CA). Cells were not cultured for more than 8 passages in vitro in order to avoid long term culture related effects. In some experiments, cells were supplemented with bone morphogenetic protein 2 (BMP2, 10 ng/ml or 50 ng/ml, R&D Systems) or rapamycin (100 ?M, Sigma). Z-LLF-CHO (Z-LLF, Calbiochem, San Diego, CA), proteasome inhibitor, was used at the concentration of 30 ?M added 30 min prior to BMP2 treatment. Esterificated diethyl-succinate (5 mM, Sigma) was added either alone or in combination with BMP2 (10 ng/ml) for 72 hr. To stabilize HIF-1? cells have been pre- treated for 12 hr with cobalt chloride (CoCl₂, 100 ?M, Sigma) prior to starting progressive high oxygen exposure and/or time course treatment with BMP2 the day after.

Total protein extracts were isolated in lysis buffer as described ^[37]. Equal amounts of protein (10?20 ?g) were resolved using a SDS-PAGE gels and transferred to PVDF Hybond-p membrane (GE Healthcare), along with purified protein standard (Full Range Rainbow? Molecular Weight Markers, from 10 to 250 kDa, GE Healthcare) to insure qualitative accuracy of analyzed proteins. Membranes were blocked with ECL Advance Blocking (Amersham Pharmacia, 2%) or I-block? Blocking (Tropix, Sigma) for at least 1 hour or overnight, under rotation at RT or 4?C respectively. Membranes were then incubated overnight at 4?C under constant shaking with the following primary antibodies: Akt (rabbit, 1?1000, Cell Signaling), Akt (T308) (rabbit, 1?1000, Cell Signaling), HIF-1? (mouse, 1?250, BD), mTOR (rabbit, 1?1000, Cell Signaling), mTOR (S2448) (rabbit, 1?1000, Cell Signaling), p70S6k (rabbit, 1?1000, Cell Signaling), p70S6K (T389) (rabbit, 1?1000, Cell Signaling), Stat3 (rabbit, 1?1000, Cell Signaling), Stat3 (S727) (rabbit, 1?1000, Cell Signaling), PHD2 (goat, 1?300, Santa Cruz), REDD1 (rabbit, 1?1000, Cell Signaling), FKBP38 (rabbit, 1?200, Acris Antibodies GmbH) and ?-actin (mouse, 1?10000, Sigma). Membranes were next incubated with peroxidase-labeled goat anti-rabbit IgG (Sigma, 1?100.000), peroxidase-labeled goat anti-murine IgG (Sigma, 1?100.000) or donkey anti-goat IgG (Santa Cruz, 1?150.000) for 60 min. All membranes were visualized using ECL Advance (GE Healthcare) and exposed to Hyperfilm MP (GE Healthcare). Densitometric analysis of the films was performed using Image J densitometer software. Values were normalized to ?-actin and calibrated to control at 2% oxygen.

RNA was isolated from cells using Trizol (Invitrogen) and 1 ?g of total RNA reverse-transcribed using SuperScript RNAse H-Reverse Transcriptase (Invitrogen). Quantitative RT-PCR reactions were run in duplicate using Brilliant? SYBR? Green QPCR Core Reagent Kit (Stratagene, La Jolla, CA). Fluorescent emission was recorded in real-time (Sequence Detection System 7900HT, Applied Biosystems, Foster City, CA, USA). Gene expression profiling was completed using the comparative Ct method of relative quantification. Relative RNA quantities were normalized to beta-glucuronidase (GUSB) as a control and 2% oxygen, untreated group, was used as the calibrating condition. PCR amplification conditions consisted of 35 cycles with primers annealing at 56?C. The specificity of primers (see Table 2) was confirmed for every PCR run by dissociation curve analysis.

GBM cells and normal SVZ-derived cells were transfected using a Promega Corporation (Madison WI) protocol for transient transfection of adherent cells using Effectene reagent. HRE-luciferase reporter construct used (wHRE) was kindly provided by Dr. Indraccolo. It consists of a trimerized 24-mer containing 18 bp of sequence from the PGK promoter including the HRE (5?-tgtcacgtcctgcacgactctagt, HRE) and an 8-bp linker sequence followed by a 50-bp minimal tyrosine kinase promoter in a pGL2-firefly luciferase basic Vector backbone (Promega) ^[38]. The mutant HRE (mHRE) construct used to evaluate aspecific effects, has the ACG of the HIF-1 binding site mutated to CAT, abolishing binding, as well as a point mutation that abolished a BsgI restriction site for diagnostic purposes. GBM cells and normal SVZ-derived cells were transfected either with a wHRE or with a mHRE. Along with these vectors, also a Renilla luciferase vector has been transfected in order to normalize luciferase detection (Promega). 12 hr after transfection, total medium change was done and cells were treated with 50 ng/ml BMP2 for 7 hrs. Finally, cells were processed and analysis of HRE-luciferase activity was performed as described (Dual-Luciferase Reporter Assay System, Promega) by using a plate-reading luminometer (Victor, Perkin Elmer).

The lentiviral plasmids containing HIF-1? siRNA and LUC siRNA target sequences, termed pLSLG-HIF-1?-siRNA and pLSLG-Luciferase-siRNA, respectively, were a kind gift of Dr. O.V. Razorenova (Department of Molecular Cardiology, Lerner Research Institute, Cleveland, OH) ^[39]. The lentiviral vectors were produced as previously described ^[40] and used to infect GBM cells and normal SVZ-derived cells. Target cells were incubated with stocks for at least 12 hr. Transduced cells were cultured for 5 days and analyzed by western blot.

To evaluate aerobic metabolic activity succinate dehydrogenase (SDH) activity has been measured, since enzyme activity is considered an indicator of tricarboxylic acid (TCA) cycle activity. To measure SDH activity, we used the methods described by Levine et al. with little modifications ^[41]. Briefly, frozen cell monolayers (not fixed) were incubated for 20 min in the dark at 37?C in a reaction buffer containing 12.3 mM succinate (Sigma), 0.2 mM 1-methoxy-5-methylphenazine methyl sulphate (Sigma), 1.2 mM nitro blue tetrazolium (NBT, Sigma) in 50 mM Tris-HCl, pH 7.6. We used 12.3 mM malonate (Sigma) to test aspecific reactions, and we also tested 3-nitropropionic acid (3-NPA) (500 ?M, Sigma), known to irreversibly inactivate SDH, as negative control. The enzyme activity is determined by measuring the formation of nitro blue diformazan (NBT-dfz) due to NBT reduction and the end product (i.e. blue cells) is visible by bright field microscopy (Olympus IX50) or can be measured at 570 nm.

Extracellular lactate measurement was performed as directed (BioVision) using a lactate assay kit kindly provided by Prof. Burlina (Department of Paediatrics, University of Padova).

Cell pellets were resuspended in PBS and lysed by two freeze/thaw cycles. COX activity was measured by the method of Salviati and co-workers ^[42]. Bovine cytochrome c was reduced with ascorbate and then desalted using a Sephadex G-25 column (Amersham). The concentration of the reduced substrate was determined spectrophotometrically at a wavelength of 550 nm. Spectrophotometrical measurements were performed using a Cary UV 100 spectrophotometer (Varian Inc., Walnut Creek, CA, U.S.A.).

Graphs and statistical analyses were prepared using Prism 3.03 (Graph Pad). All values are presented as mean?standard error of the mean (S.E.M.). Statistical significance was measured by one-way ANOVA with Newman-Keuls multiple comparison post Test, *p<0.05, **p<0.01, ***p<0.001. For all graphs, an asterisk directly above a bar indicates a significant difference with its 2% oxygen counterpart; an asterisk over a bracket indicates a significant difference with another variable as indicated.