Contrasting roles of neuronal Msk1 and Rsk2 in Bad phosphorylation and feedback regulation of Erk signalling.

Activated extracellular-signal-regulated kinase (Erk) phosphorylates and activates downstream kinases including ribosomal S6 kinase 2 (Rsk2/RPS6KA3) and mitogen- and stress-activated kinase 1 (Msk1, RPS6KA5). Rsk2 plays an important role in neuronal plasticity, as patients with Coffin-Lowry syndrome, where Rsk2 is dysfunctional, have impaired cognitive function. However, the relative role of neuronal Rsk2 and Msk1 in activating proteins downstream of Erk is unclear. In PC12 cells and in cortical neurones, the calcium ionophore A23187-induced phosphorylation of Erk, Msk1, Rsk2 and also the Bcl-2-associated death protein (Bad), which protects against neurotoxicity. Specific knockdown of Msk1 with small interfering RNA reduced the ability of A23187 to induce Bad phosphorylation in both PC12 cells and cortical neurones. Conversely, specific knockdown of Rsk2 potentiated Bad phosphorylation following A23187 treatment, and also elevated Erk phosphorylation in both cell types. This indicates that Msk1 rather than Rsk2 mediates neuronal Bad phosphorylation following Ca(2+) influx and implicates Rsk2 in a negative-feedback regulation of Erk activity.

Properties of neurons in the trigeminal nucleus caudalis responding to noxious dural and facial stimulation.

Extracellular single unit recordings were made in the rat trigeminal nucleus caudalis (Vc) from cells with Adelta and C-fibre latency responding to electrical stimulation of the thinned cranium overlying the middle meningeal artery (MMA). The neurons had an ipsilateral facial receptive field (FRF) that mainly extended over areas innervated by the first and second division of the trigeminal nerve but in some cases also included areas innervated by the third division of the trigeminal nerve. No wind-up of either long latency C-fibre or short latency Adelta responses was seen during trains of electrical stimulation. Sensitisation of mechanical stimulation of the FRF could also not be observed at any time during dural stimulation. In contrast, extracellular single unit recordings in the Vc activated by electrical stimulation of the facial skin resulted in a significant wind-up response of long latency response in six of ten cells studied. The facial-elicited wind-up response was significantly enhanced, 18 min after the electrical stimulation protocol was started, indicating that the process of wind-up had generated central excitability. The findings in this study demonstrate a clear difference between the effects of electrical stimulation of cutaneous and non-cutaneous inputs. In the trigeminal system, this has implications for the study of pathways such as those involved in headache, where it is believed that an enhanced dural input to the Vc may generate central sensitisation and partly explain the hyperalgesia and allodynia reported by patients.

N-methyl-D-aspartate and brain-derived neurotrophic factor induce distinct profiles of extracellular signal-regulated kinase, mitogen- and stress-activated kinase, and ribosomal s6 kinase phosphorylation in cortical neurons.

Stimulation of N-methyl-D-aspartate (NMDA) receptors is believed to underlie long-term memory formation, and excessive NMDA receptor activation has been linked to several neuropathological conditions. Phosphorylation and activation of p42/44 mitogen-activated protein kinase (ERK) is believed to mediate many of these effects, but the downstream targets of ERK in response to NMDA activation have not been determined. In primary cultures of rat cortical neurons, we found that NMDA was able to elevate phosphorylation of mitogen- and stress-activated kinase 1 (MSK1) as well as ERK. Likewise, brain-derived neurotrophic factor (BDNF) treatment increased phosphorylation of MSK1 and ERKs. The NMDA-induced MSK1 phosphorylation was sensitive to the MEK inhibitor 2'-amino-3'-methoxyflavone (PD98059) and the p38 inhibitor 4-(4-fluorophenyl)-2-(4-methylsulfinylphenyl)-5-(4-pyridyl)1H-imidazole (SB203580). MSK1 activation by NMDA was transient, although ERK remained phosphorylated within the neuronal cytoplasm for several hours. Although BDNF increased ribosomal S6 kinase (RSK) phosphorylation, NMDA had no discernable effect on the phosphorylation of RSKs. Thus, phosphorylation and activation of MSK1 but not RSK could be an important step in the pathway linking NMDA-induced ERK phosphorylation to the activation of transcription factors required for the formation of long-term memory.

Effect of IL-1beta microinjection into the posterior hypothalamic area on trigeminal nociception in the rat.

The hypothalamus has been implicated in the pathophysiology of the most disabling forms of primary headache, namely migraine and cluster headache. Interleukin-1beta (IL-1beta) is highly expressed in the hypothalamus. We recorded from the trigeminal nucleus caudalis of rats using extracellular electrophysiological methods from neurons responding to electrical stimulation of the peri-middle meningeal artery dura mater and having receptive fields in the ophthalmic division of the trigeminal nerve. Data were collected from fifteen clusters of wide-dynamic range neurons with stable baseline firing responses between 97 and 101% ( n=3 for each unit) to stimulation. Microinjection of IL-1beta into the posterior hypothalamus of 9 animals resulted in a modest inhibition of evoked trigeminal responses in three units, no effect in six and no overall effect for the entire cohort studied. The mean maximum response was a non-significant reduction in firing to 83+/-7% ( n=9) at 30 minutes post-injection of IL-1beta. There was some variation of effect dependent on site of injection with central posterior hypothalamus being the predominant area that resulted in inhibition. There was no inhibition in the six animals injected with vehicle (saline). If there is an important effect for IL-1beta in the posterior hypothalamus it is likely to be highly somatotopically restricted.

CB1 and CB2 cannabinoid receptors are implicated in inflammatory pain.

The cannabinoid agonist, HU210 has been evaluated in vivo in nociceptive and inflammatory pain models in the rat. The ED50 for the anti-nociceptive (increasing mechanical withdrawal threshold) effect was 0.1 mg/kg-1 i.p., and for anti-hypersensitivity and anti-inflammatory activity was 5 g/kg-1 i.p. (in the carrageenan model). The selective CB1 antagonist, AM281 (0.5 microg/kg-1 i.p.) reversed effects of HU210 (10 and 30 microg/kg-1 i.p.) in both nociceptive and inflammatory models of hypersensitivity. The selective CB2 antagonist, SR144528 (1 mg/kg-1 i.p.) antagonised effects of HU210 (30 microg/kg-1 i.p.) in the carrageenan induced inflammatory hypersensitivity. The CB2 agonist, 1-(2,3-Dichlorobenzoyl)-5-methoxy-2-methyl-(2-(morpholin-4-yl)ethyl)-1H-indole (GW405833) inhibited the hypersensitivity and was anti-inflammatory in vivo. These effects were blocked by SR144528. These findings suggest that CB1 receptors are involved in nociceptive pain and that both CB1 and CB2 receptors are involved in inflammatory hypersensitivity. Future studies will investigate effects on identified inflammatory cells within the inflamed tissue to further elucidate the role of cannabinoid receptors.

Increased expression of dendritic mRNA following the induction of long-term potentiation.

A small number of mRNAs, including Ca2+/calmodulin-dependent protein kinase II alpha-subunit (CamKIIalpha) mRNA and microtubule-associated protein 2 (MAP2) mRNA, are present in the dendrites of neurones as well as in the cell bodies. We show here that the induction of long-term potentiation (LTP) in the hippocampal perforant path/granule cell synapses in anaesthetised rats is associated with increased levels of CamKIIalpha mRNA and MAP2 mRNA in the granule cell dendrites after 2 h. Similarly, induction of LTP in the Schaffer collateral/CA1 pyramidal cell synapses in hippocampal slices maintained in vitro also results in elevated dendritic levels of CamKIIalpha mRNA and MAP2 mRNA 2 h later. In both models, the levels of various other mRNA species restricted to the cell body region were unaffected by the induction of LTP. Increased expression of dendritic CamKIIalpha mRNA and MAP2 mRNA appears to be a general feature of hippocampal plasticity, since it occurs following LTP induction in both the dentate gyrus and the CA1 region. The elevation of mRNA levels in a restricted region close to the afferent synapses would allow a highly-localised enhancement of the synthesis of the corresponding proteins, providing an elegant mechanism for protein-synthesis-dependent synaptic plasticity to maintain a high degree of anatomical specificity.

Involvement of two isoforms of SNAP-25 in the expression of long-term potentiation in the rat hippocampus.

Induction of long-term potentiation (LTP) in the hippocampus is associated with changes in expression of a variety of different proteins and is thought to be the mechanism which underlies synaptic plasticity. The 25 kDa synaptosomal-associated protein (SNAP-25) is a presynaptic protein which is involved in neurotransmitter exocytosis at the nerve terminal. Two isoforms of SNAP-25 have so far been identified (a and b) which differ in their distribution and developmental regulation. Using in situ hybridization, we demonstrated that the mRNA levels of the two isoforms of this protein are increased 2 h after the induction of LTP in granule cells of the dentate gyrus following high frequency stimulation of the perforant path in vivo. These observations further demonstrate the involvement of both isoforms of SNAP-25 in functional synaptic plasticity, although their exact roles have yet to be fully determined.

Long-term potentiation in perforant path/granule cell synapses is associated with a post-synaptic induction of proenkephalin gene expression.

Enkephalin peptides released from hippocampal mossy fibres lower the threshold for the generation of long-term potentiation (LTP) at the mossy fibre synapses. High frequency stimulation of the hippocampal dentate gyrus, sufficient to induce mossy fibre LTP, is associated with increased expression of the proenkephalin gene in the granule cells. We show here that a similar elevation in proenkephalin mRNA levels is observed, in anaesthetised rats, following stimulation of the perforant path sufficient to induce LTP in the perforant path/granule cell synapses. This strengthens the evidence implicating granule cell enkephalins as mediators of functional plasticity in the hippocampus. Furthermore. the results hint at a form of 'domino plasticity', where potentiation of transmission at the perforant path/granule cell synapses is subsequently followed by an enkephalin-mediated potentiation of transmission at the mossy fibre synapses.

Changes in hippocampal gene expression associated with the induction of long-term potentiation.

The expression of four genes: zif/268, c-fos, tubulin and alpha Ca2+/calmodulin-dependent protein kinase II (alpha CAMKII) was studied following the induction of LTP in Schaffer collateral CA1 neurone synapses in rat hippocampal slices maintained in vitro. Levels of c-fos mRNA and tubulin (T26) mRNA in area CA1 were unchanged after induction of LTP, however, zif/268 and alpha CAMKII mRNA levels showed a significant increase compared to non-potentiated controls. It is possible, therefore, to measure changes in gene expression using in situ hybridisation following induction of LTP in vitro and these results strengthen the theory that zif/268 and alpha CAMKII are involved in some aspect of the induction or maintenance of hippocampal LTP.

Characterisation of the specific binding of the histamine H3 receptor antagonist radioligand [3H]GR168320.

We have examined the specific binding of the tritiated derivative of the potent histamine H3 receptor antagonist, [3,4-3H2]-cyclohex-yl-¿[4-(3H-imidazol-4-yl)-piperidin-l-yl] iminomethyl¿- amine ([3H]GR168320), to homogenates of rat cerebral cortex. Specific binding of [3H]GR168320 at 37 degrees C associated and dissociated rapidly. Binding was saturable (Bmax 412 +/- 89 fmol/mg protein) and of high affinity (Kd 0.12 +/- 0.11 nM). Saturation studies suggested the involvement of a single site. Histamine H3 receptor agonists and antagonists inhibited [3H]GR168320 binding with high affinity. Agonist and antagonist affinities correlated when compared with affinities obtained using the tritiated histamine H3 agonist radioligand N alpha-methylhistamine.

Investigation of the role of tachykinin NK1, NK2 receptors and CGRP receptors in neurogenic plasma protein extravasation in dura mater.

The aim of this study was to investigate the role of tachykinin NK1 and NK2 receptors, and calcitonin gene-related peptide (CGRP) receptors in neurogenic plasma protein extravasation, induced by electrical stimulation of the trigeminal ganglion, in dura mater of anaesthetised rats and guinea-pigs. This response was significantly (P < 0.05) blocked in both species by the selective peptide tachykinin NK1 receptor antagonist, GR82334 ([D-Pro9[spiro-gamma-lactam]Leu10,Trp11]physalaemin-(1-11)) (0.02-0.2 mg/kg i.v.) whilst the selective tachykinin NK2 receptor antagonist, (+/-)-SR 48968 ((S)-N-methyl-N-[4-(4-acetylamino-4-phenylpiperidino)-2-(3,4- dichlorophenyl) butyl] benzamide) (1 mg/kg i.v.) was without effect. The CGRP receptor antagonist, alpha-CGRP-(8-37) (0.1 mg/kg i.v.) significantly (P < 0.05) blocked this response in dura mater of guinea-pigs but not rats. These results suggest that substance P, acting via tachykinin NK1 rather than NK2 receptors, mediates neurogenic plasma protein extravasation in dura mater and that CGRP may have an involvement in this response in guinea-pigs.

Activation of sensory nerves in guinea-pig isolated basilar artery by nicotine: evidence for inhibition of trigeminal sensory neurotransmission by sumatriptan.

Nicotine (100 microM), but not electrical field stimulation or potassium chloride (0.1-3 microM), caused capsaicin (1 microM)- and tetrodotoxin (1 microM)-sensitive relaxations of guinea-pig isolated basilar artery precontracted with prostaglandin F2 alpha. Nicotine-induced responses were blocked by the neurokinin NK1 receptor antagonist, GR82334 (10 microM), but were unaffected by the calcitonin gene-related peptide (CGRP) receptor antagonist, CGRP-(8-37) (1 microM). This suggests that nicotine activates capsaicin-sensitive sensory nerves in guinea-pig basilar artery to cause relaxation predominantly via substance P release. The vascular 5-HT1 receptor agonist, sumatriptan (0.3 and 3 microM), inhibited nicotine-induced relaxation (by 50 and 80% respectively); the inhibitory effect of sumatriptan (0.3 microM) was attenuated in the presence of the non-selective 5-HT1 receptor antagonist, methiothepin (0.1 microM). These data suggest that sumatriptan can inhibit sensory neurotransmission in guinea-pig basilar artery via activation of inhibitory prejunctional 5-HT1 receptors on sensory nerve terminals.

Pharmacological activity of VUF 9153, an isothiourea histamine H3 receptor antagonist.

The pharmacological activity of the histamine H3 receptor antagonist VUF 9153 (S-[3-(4(5)-imidazolyl)]propyl-N-(4-chlorobenzyl)isothiourea) has been investigated in vitro and in vivo. VUF 9153 displaced [3H]N alpha-methylhistamine binding to rat cortex/hippocampal membranes (pKi = 9.77 +/- 0.03) and antagonised the inhibitory responses to (R)-alpha-methylhistamine against electrical field stimulation in the isolated longitudinal smooth muscle preparation of guinea-pig ileum (pKB = 9.95 +/- 0.07). In these assays, VUF 9153 was 10-50-fold more potent than the prototype H3 receptor antagonist thioperamide. VUF 9153 showed no or very weak activity in in vitro functional assays for histamine H1 or H2 receptors. Systemic administration of VUF 9153 (s.c. or p.o.) dose-dependently inhibited the ex vivo binding of [3H]N alpha-methylhistamine to rat cortex/hippocampal membranes and dipsogenic responses induced by (R)-alpha-methylhistamine. Calculation of ED50 values, at the 1 h pretreatment time used, revealed that VUF 9153 administered s.c. or p.o., was approximately 2-fold weaker than thioperamide. These data indicate that, like thioperamide, VUF 9153 is a potent and selective antagonist for histamine H3 receptors in vitro, possesses the ability to penetrate the blood-brain barrier to access central H3 receptors and can inhibit H3 receptor-mediated functional responses in vivo.

Extracellular recordings of membrane potential from guinea-pig isolated trigeminal ganglion: lack of effect of sumatriptan.

The aim of this study was to investigate the effect of the anti-migraine drug and selective 5-HT1 receptor agonist, sumatriptan, on membrane potential of guinea-pig isolated trigeminal ganglion. Ganglia were divided into three longitudinally, placed in two-compartment baths and the d.c. potential between compartments was recorded extracellularly. Drugs were applied to the Krebs superfusion fluid of one compartment. KCl (3 mmol/l) and GABA (0.1 mmol/l) caused depolarization (0.30 +/- 0.05 and 0.55 +/- 0.08 mV respectively, n = 11-19). 5-HT (1-10 mumol/l) caused small depolarizations (0.06 +/- 0.02 mV, n = 8) but sumatriptan (0.1-10 mumol/l) had no effect on trigeminal ganglion membrane potential. Collagenase pretreatment, to enhance desheathing, or modification of the composition of the Krebs solution failed to reveal any effect of sumatriptan. These data provide no evidence to suggest that sumatriptan inhibits neurotransmission in trigeminal ganglion. However, 5-HT1 receptors may be present in insufficient numbers in the trigeminal ganglion to elicit a change in membrane potential. Further studies are required to investigate the effect of sumatriptan at the level of the sensory nerve terminals within the intracranial vasculature, where 5-HT1 receptors may be concentrated.

Neurokinin NK1 receptors mediate plasma protein extravasation in guinea-pig dura.

Selective neurokinin receptor agonists and calcitonin gene-related peptide (CGRP) were administered i.v. to anaesthetised guinea-pigs, and plasma protein extravasation was measured in dura and conjunctiva, intra- and extracranial tissues, respectively, innervated by the trigeminal nerve. The neurokinin NK1 receptor agonist, GR73632 enhanced plasma protein extravasation in guinea-pig dura (10 nmol/kg i.v.) and conjunctiva (3 and 10 nmol/kg i.v.). Pretreatment with the neurokinin NK1 receptor antagonist GR82334 (200 nmol/kg i.v.) blocked the response to GR73632 in both tissues. Neurokinin NK2 and NK3 receptor-selective agonists, GR64349 (10 nmol/kg i.v.) and senktide (30 nmol/kg i.v.) respectively, and also the neuropeptide CGRP (10 nmol/kg i.v.) had no significant effect on plasma protein extravasation in intra- or extracranial tissues. We conclude that the neurokinin NK1 receptor mediates plasma protein extravasation in tissues innervated by the trigeminal nerve in guinea-pigs; neurokinin NK2, NK3 and CGRP receptors do not directly mediate extravasation of plasma proteins in these tissues.

Cyclic AMP analogues increase excitability and enhance epileptiform activity in rat neocortex in vitro.

Effects of the cyclic AMP agonists 8-(4-chlorophenylthio)-adenosine 3':5' cyclic monophosphate (CPT-cAMP), dibutyryl cyclic AMP (dbcAMP) and forskolin were studied on extracellular field potentials in rat neocortex slices in vitro. CPT-cAMP and forskolin produced a prolonged enhancement of epileptiform activity resulting from removal of Mg2+ from the bathing medium. DbcAMP had no apparent effect except at high concentrations (1 mM), when it reduced bursting activity. Field potentials observed following electrical stimulation of the corpus callosum in the presence of Mg2+ were enhanced by CPT-cAMP and dbcAMP; however forskolin was without effect. Intracellular recording techniques demonstrated a transient excitatory influence of dbcAMP. The results indicate a role for cyclic AMP in seizure mechanisms.

Sensory nerve-mediated relaxation of guinea-pig isolated pulmonary artery: prejunctional modulation by alpha 2-adrenoceptor agonists but not sumatriptan.

1. Effects of the alpha 2-adrenoceptor agonists, UK14304 and clonidine, the 5-HT1 receptor agonist, sumatriptan and the kappa-opioid receptor agonist, GR103545, on sensory neurotransmission in histamine-contracted guinea-pig isolated pulmonary artery (GPPA) have been studied. 2. Electrical field stimulation (EFS) induced frequency-dependent relaxations of histamine-contracted GPPA, which were attenuated by tetrodotoxin and capsaicin pretreatment but not by the nitric oxide synthase inhibitor, N omega-nitro-L-arginine methyl ester (L-NAME). 3. Substance P (0.3 microM) induced relaxations which were subject to rapid tachyphylaxis. Neither the NK1 receptor antagonist, (+/-)-CP 96,345, nor desensitization to substance P had any effect of EFS-induced relaxations of histamine-contracted GPPA. 4. Calcitonin gene-related peptide (CGRP; 3 and 30 nM) induced concentration-dependent relaxations of histamine-contracted GPPA. The putative CGRP receptor antagonist, CGRP8-37 (1 microM), markedly attenuated EFS-induced relaxations as well as relaxations induced by a low concentration of CGRP. 5. Sumatriptan (0.1 and 1 microM) and the selective kappa-opioid receptor agonist, GR103545 (10 and 100 nM) had no effect on EFS-induced relaxations of histamine-contracted GPPA. In contrast, the alpha 2-adrenoceptor agonists UK14304 (1-100 nM) and clonidine (300 nM) attenuated responses to EFS, the attenuation of UK14304 (100 nM) being reversed by yohimbine (300 nM). 6. It is concluded that in GPPA, where a presynaptic inhibition of sensory neurotransmission by alpha 2-adrenoceptor activation could be shown, there was no evidence for such modulation by either sumatriptan-sensitive 5-HT1 receptors or kappa-opioid receptors.

Lack of effect of sumatriptan and UK-14,304 on capsaicin-induced relaxation of guinea-pig isolated basilar artery.

1. The objectives of this study were to assess the effects of sensory neuropeptide antagonists and presynaptically acting receptor agonists on capsaicin-induced relaxations of guinea-pig isolated basilar artery (GPBA). 2. Capsaicin, human alpha-calcitonin gene-related peptide (CGRP) and substance P (SP) caused concentration-related relaxations of GPBA which had been pre-contracted with prostaglandin F2 alpha (PGF2 alpha). Responses to capsaicin were not modified by the peptidase inhibitors, phosphoramidon (1 microM) and bestatin (100 microM). 3. The relaxant responses to capsaicin were blocked in a selective manner by ruthenium red (3 microM) and by the CGRP antagonist, CGRP8-37 (1 microM). CGRP8-37 also selectively inhibited the relaxant effects of CGRP. 4. The selective NK1 receptor antagonist, GR82334 (10 microM), inhibited SP-induced relaxations but had little effect on capsaicin-induced relaxations. 5. The 5-HT1 receptor agonist, sumatriptan, produced small contractions of GPBA under conditions of resting tone. In the presence of PGF2 alpha, sumatriptan had no further contractile effect. Sumatriptan (0.3 and 3 microM) did not modify capsaicin-induced relaxations of GPBA. 6. The alpha 2-adrenoceptor agonist, UK-14,304 (0.1 microM), had no effect on basal or PGF2 alpha-induced tone. UK-14,304 did not modify capsaicin-induced relaxations. 7. These results suggest that capsaicin causes relaxation of GPBA via a release of CGRP. This process is amenable to blockade by CGRP8-37 and ruthenium red, but not to modulation by either sumatriptan or UK-14,304.

Lipocortin-1 is an endogenous inhibitor of ischemic damage in the rat brain.

Lipocortin-1 (annexin-1) is an endogenous peptide with antiinflammatory properties. We have previously demonstrated lipocortin immunoreactivity in certain glial cells and neurons in the rat brain (Strijbos, P.J.L.M., F.J.H. Tilders, F. Carey, R. Forder, and N.J. Rothwell. 1990. Brain Res. In press.), and have shown that an NH2-terminal fragment (1-188) of lipocortin-1 inhibits the central and peripheral actions of cytokines on fever and thermogenesis in the rat in vivo (Carey, F., R. Forder, M.D. Edge, A.R. Greene, M.A. Horan, P.J.L.M. Strijbos, and N.J. Rothwell. 1990. Am. J. Physiol. 259:R266; and Strijbos, P.J.L.M., J.L. Browning, M. Ward, R. Forder, F. Carey, M.A. Horan, and N.J. Rothwell. 1991. Br. J. Pharmacol. In press.). We now report that intracerebroventricular administration of lipocortin-1 fragment causes marked inhibition of infarct size (60%) and cerebral edema (46%) measured 2 h after cerebral ischemia (middle cerebral artery occlusion) in the rat in vivo. The lipocortin-1 fragment was effective when administered 10 min after induction of ischemia. Ischemia caused increased expression of lipocortin-1 around the area of infarction as demonstrated by immunocytochemistry. Intracerebroventricular injection of neutralizing antilipocortin-1 fragment antiserum increased the size of infarct (53%) and the development of edema (29%). These findings indicate that lipocortin-1 is an endogenous inhibitor of cerebral ischemia with considerable therapeutic potential.

Effects of hypoxia on fetal rat brain metabolism studied in utero by 31P-NMR spectroscopy.

An animal model of perinatal asphyxia, in which near-term fetal rats are subjected to-short periods of hypoxia, has been investigated by 31P-NMR spectroscopy. Changes in the high-energy phosphates and intracellular pH of the fetal rat brain were measured in utero following ligation of the placental blood vessels, and during reperfusion after a 20-min period of occlusion. The hypoxia-induced changes observed in the fetal brain were substantially slower than in the adult, and were completely reversible after 20 min of hypoxia.