The effects in vivo of hypoxia on brain injury

BRAIN RESEARCH ELSEVIER

Brain Research 725 (1996) 184-191

Research report

The effects in vivo of hypoxia on brain injury Paul Pearigen, Ryder Gwinn, Roger P. Simon * Department of Neurology, Unit'ersity c)[ Cali~brnia, San Francisco, USA Accepted 7 February 1996

Abstract To separately analyze the hypoxic component of hypoxic-ischemic encephalopathy, rats were prepared such that their paO2 was maintained at 20 mmHg while maintaining systemic arterial pressures. During the 20-rain experiment, brain oxygen concentration and extracellular amino acid concentrations were monitored. At sacrifice, the brains were studied for morphologic evidence of injury by immunocytochemical staining for the non-constituitive stress protein HSP-72 or neuronal death by acid fuchsin staining. Oxygenated rats subjected to global ischemia were prepared for comparison. In these experiments, hypoxia resulted in no increase in extracellular glutamate concentration, and no morphologic injury was detected. Thus, hypoxia without ischemia is well tolerated by brain. Keywords: Hypoxia; Heat shock protein; Hypoxia-ischemia; Glutamate

1. Introduction

It is a clinical axiom that removal of oxygen from the brain will result in the death of that organ. However, the mechanism by which such brain injury occurs is not quite as clear in the experimental literature. In vivo, hypoxic injury to the brain almost always occurs in the setting of ischemia; what portion of the injury results from oxygen deprivation as opposed to loss of blood flow may be difficult to establish in the intact organism. Clinically, this distinction is usually ignored and the terms hypoxic encephalopathy and ischemic encephalopathy are generally used interchangeably. To best order interventions during critical care situations, the relative roles of hypoxia and ischemia in 'hypoxic-ischemic encephalopathy' should be defined. The pathogenesis of hypoxic-ischemic injury to the brain has been clarified over the last decade [6]. Injury begins with elevation of glutamate concentrations in the extracellular compartment, which opens voltage-gated and receptor-operated calcium channels. The resultant intracellular calcium toxicity induces catabolic processes within the cell, ultimately resulting in cell death. Early and subtle injury can be detected immunocytochemically by the use of antibodies to nonconstitutive heatshock proteins (HSP)

* Corresponding author. Present address: Department of Neurology, University of Pittsburgh Medical School 325 Scaife Hall, Pittsburgh, PA 15213, USA. Fax: (1) (412) 648-1239.

induced in nervous system cells by a wide range of stress, including seizures and ischemia [12,16,17]. HSP induction represents a response to the presence of denatured protein in the cell [1]. The presence of denatured proteins activate heatshock factors [18] which bind to heat shock elements, resulting in the transcription of HSP RNA [19,20]. Neurons which imunocytochemically stain with HSP antibody therefore contain denatured protein as evidence of injury. We used the techniques of in vivo microdialysis [10] and heat shock protein immunocytochemistry to examine the role of pure hypoxia, in the absence of ischemia, in brain injury. Acid fuchsin staining was used to identify dead cells.

2. Materials and m e t h o d s

2.1. Animal preparation 2.1.1. Ventilation and physiologic monitoring Adult male Sprague-Dawley rats (325-375 g) were intubated, placed on a homeothermic blanket to maintain rectal temperature at 36.5-37.5°C, and mechanically ventilated using a Harvard small animal ventilator with 2% halothane in 24% oxygen with a balance of nitrogen gas. A femoral artery catheter was placed for continuous blood pressure recording and for arterial blood gas sampling. A femoral vein was canalized for administration of neuromuscular paralyzing agents and normal saline. Cutaneous

0006-8993/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved. Pll S 0 0 0 6 - 8 9 9 3 ( 9 6 ) 0 0 2 1 5 - 6

P. Pearigen et al. / Brain Research 725 (1996) 184-191

electroencephalography (EEG) leads were placed for monitoring. Ventilator settings were adjusted to maintain normal blood gas while the animals were further prepared for study. Rats were placed on a stereotaxic frame and, via craniectomy, the dura was exposed above the caudal portion of the right cortex. Following probe placement as described below, animals were paralyzed by intravenous injection of pancuronium bromide (0.4 m g / k g ) ; the stability of ventilation and physiologic parameters was assessed by sampling arterial blood gases. Halothane was discontinued as hypoxia was induced, because hypoxia alone to this degree is a profound anesthetic and continuing the halothane as well could have a systemic hypotensive effect. The effectiveness of the hypoxic anesthesia was attested to by the fact that the animals evinced no physiologic response to stress as monitored by EEG and blood pressure. After the hypoxic period (as described below) halothane anesthesia was reinstituted during the recovery phase and continued until surgical manipulations were reversed. When animals recovered adequate spontaneous respirations, the endotracheal tube was removed, and the animals were returned to their cages with food and water ad lib. After 72 h, the animals were reanesthesized with chloral hydrate (400 m g / k g i.p.) and sacrificed by perfusion with 4% paraformaldehyde through the left ventricle. Brains were removed immediately and immersed in 4% paraformaldehyde overnight for later immunocytochemical staining.

2.2. Brain oxygen tension In three rats, a thin film oxygen probe (Ottosensors #01152-11) was placed 3.0 mm below the surface of the dura, 5.0 mm caudal to bregma, and 3.0 mm lateral to midline. This reproducibly located the sampling portion of the probe within the hippocampus. Continuous voltage output was recorded and a 30 s average output was calculated before the insult began and every 4 min during the hypoxic period and during recovery periods. Voltage output from the probe was converted to oxygen tension in keeping with the manufacturer's recommendations by creating a standard curve from probe measurements of saline with two known oxygen tensions (6% O 2 and 21% 02) taken immediately prior to probe placement in the animal.

rate, 2.0 txl/min) was collected before hypoxia, then sequential ten minute samples were collected throughout the insult and recovery periods for a total of 60 min. Dialysates were analyzed for glutamate concentrations by high-pressure liquid chromatography (HPLC) using a method previously described [20].

2.4. Hypoxia For the hypoxic insult, the previously delivered ventilation gas was changed to a premixed tank of 6% oxygen with a balance of nitrogen. This oxygen content was assured by an inline oxygen sensor monitoring the delivered F i O 2 . Mean arterial blood pressure was maintained above 60 mmHg, using intravenous boluses of normal saline at 10 m g / k g when necessary. After 20 min of 6% hypoxia, an F~O2 of 24-30% was delivered for a 40 min recovery period during which sampling of brain oxygen tension or microdialysate continued. For comparison, oxygenated animals (FiO 2 20%) were prepared with global ischemia via 4-vessel occlusion for 4, 8, and 20 min (n = 3 per group), with the duration of global ischemia confirmed by electroencephalographic isoelectricity [16]. Perfusion was then re-established, the animals were permitted to recover, and sacrificed at 70 h. Details of halothane anesthesia were identical to the hypoxia animals above.

2.5. Immunocytochemical staining All brains were cut in 50 Ixm sections on a vibratome, and placed in tris buffer for immunocytochemical staining or dried on gelatinized slides for acid fuchsin staining. A mouse monoclonal antibody raised against the 72 kDa non-constituitive heat shock protein (HSP-72) was used. HSP immunocytochemistry and acid fuchsin staining were performed as previously reported [20].

tr° t 140

! E

2.3. Microdialysis In five rats, a microdialysis probe (CMA-12, BAS/Carnegie Medicine, 2.0 mm tip) was placed 3.5 mm below the surface of the dura, 3.8 mm caudal to bregma, and 2.0 mm lateral to midline. Because of the difference in size between the microdialysis and oxygen tension probe tips, this placement approximated the same sampling location in hippocampus as that of the oxygen probe noted above. One 10-min sample of extracellular dialysate (flow

185

100

g

80 60 40 20 0 Pre-insult

5 min

10 min

15 min

Post-insult

Time

Fig. 1. M e a n arterial pressures ( M A P ) during time o f hypoxia.

P. Pearigen et al./Brain Research 725 (1996) 184-191

186

100 90 80 ~:::[:~i::.~:'~E:::'~h-"-?<:!E£EE~EEE

70

A

1-" E E v

60 50

~:-.~: :<+:~+:.: :+>b:.:.>: :.F:

0

,~. v-~z

Q.

..~-- .x<

•

.:~.

40 .t-

30

t: <

20 ~:$-hz:i~.:.~,:::::[:i::f.

:~'?]:i

10 0 Pre-insult

I

!

I

I

min 5

min 10

min 15

Post-insult

Time lit

p < 0.01 for difference from Pre-insult mean p < 0.001 for difference from Pre-insult mean

Fig. 2. Arterial pO 2 during time of hypoxia ('period of insult').

180 160

T

140 tO

120 100

iiiiiiiiiiiiiiiiiiiiiii!;i!iiii 8

x

..~

¢"

v

0

oo-.t

i~i~i~i~:!!~!!~ii~]~i~i

iiiiiiiii!ii!i~iiiiiii!iiii!!ii

T

v

I

::::::::::::::::::::::::::::::: :::::::::::::::::::::::::::::::

*

....,....

*',It

.........

!::i]i~i::iiiiiii::i::i::!::i:: ii~ii~i~i~!!~?:':':': 60

if)

::::::::::::::::::::::: ]]]]]]]]]]]]]]]]]]]]]][]i][i]]] •:.:.:.:.:+:.:+:+ :.:+:,

4o 20 0

i]i]i]i]i]i]i]i]i]i]!]i]i]i]i]i

[]][]]]]][]]]]][]]]]]]]]]]]]]]:: :~?]-'---:U:]:]:]:]: :::::::::::::::::::::::::::::::

:::' .........................-....., :::::::::::::::::::::::::::::::::

"

.................. ........ ...............................4 .......................,

i!i!ili!iiiiiiiiiii!i!i! il iii

.....................,

::::::::::::::::::::::::::

. ........................,

ii!ililiiiiiiiiiiililiiiiii!i! :::::::::::::::::::::::::::::::

:i:~:~:!:!:!:~:!:!:!?!:]:i:i:ii ......................=

............i .............. Pre-insult

ii!ii!i i!!iii!iii!i!iii!i !i

"k

:i:i:i:i:ii!:~:!:i(!:;:!:!:i:i: ~: :::::::::::::2::

, ......................,

I

I

4 min

8 min

!

12 min

16 min

I

Time ',k

p < 0.05 for difference from Pre-insult mean "A-*

I

20 min Post 4 min

p < 0.01 for difference from Pre-insult mean

Fig. 3. Brain oxygen tension (expressed as % change from baseline) during 20 rain of FiO 2 6%.

P. Pearigen et a l . / Brain Research 725 (1996) 184-191

187

Table l

pH pCO 2 (mmHg) pO 2 (mmHg) MAP (mmHg)

Baseline

5 min

10 min

15 min

7 . 4 2 0 4- 0 . 0 2 9 43.2 _ 4.2 87.6 4- 12.8 103.3 + 4.0

7.338 4- 0.057 30.9 + 3.5 * 20.4 4- 2.3 + 106.7 5 : 2 3 . 5

7.234 5:0.062 + 27.0 4- 3.6 * 22.7 5 : 3 . 0 + 105.7 5 : 3 5 . 6

7.152 5 : 0 . 0 7 8 + 25.1 5 : 3 . 5 * 23.7 5 : 2 . 9 + 99.0 5 : 1 5 . 7

Values expressed as m e a n + S.D. * P < 0 . 0 l for difference f r o m baseline. + P < 0.001 for difference f r o m baseline.

2.6. Statistical analysis Values are reported as mean _+S.D. Mean arterial blood pressure, arterial pH, pCO2, and pO2, as well as brain oxygen tension and brain dialysate glutamate concentrations were compared using repeated-measures analysis of variance with application of the Hynyh-Feldt test for sphericity due to correlation of measurements in a repeated-measure design. Log transformation of the brain oxygen tension and the dialysate glutamate concentrations was performed. Post-hoc comparison of changes from baseline were performed using Dunnett's method. The alpha error rate was set at 0.05 throughout.

3. Results

drop compared to baseline ( P = 0.001) during the 6% FiO 2 hypoxia period ( P < 0.001 for different from baseline for each time point during insult and P < 0.01 for post-insult time point) (Fig. 2). The nadir of pO e was 20.4 _+ 2.3 mmHg at 5 rain after the onset of hypoxia. Changes in pH and pCO 2 were statistically significant and are also shown in Table 1. 3.2. Brain oxygen tension During the 20-min period, the animals experienced a rapid and statistically significant decrease in brain oxygen tension compared to baseline (Fig. 3). The brain oxygen tension, after falling to 32 _+ 16% of baseline by the end of the hypoxic period, recovered rapidly when the FiO 2 was returned to 30%.

3.1. Physiologic parameters 3.3. MicrodiaIysis Baseline mean arterial blood pressure, arterial pH, arterial pCO2, and arterial p O 2 values are shown in Table 1. Mean arterial blood pressure was not significantly different from baseline at any time point during or after hypoxia ( P = 0.91) (Fig. 1). Arterial p O 2 underwent a significant

The animals experienced no significant change from baseline in brain dialysate glutamate concentration (expressed as a percentage of baseline) during or following the 20 min hypoxic insult ( P = 0.45) (Fig. 4).

20O !80 160 ~.,

140 ~-

c ~

120

~!:]:i:!:iS~:~:!:i:i:i:

100

!il]iii!i~]i!~ii~i]iiiii

L oO

80

©

60 40 20 iiii!ii !

Baseline m i n l 0

min20

min30

min40

min50

mm60

ix:+:, iiiiiii!iiiii!!!! mmTO

Time

Fig. 4. G l u t a m a t e concentration in brain microdialysate prior to (baseline), during (10- a n d 2 0 - m i n samples), a n d following a 2 0 - m i n period o f FiO 2 o f 6%.

I

P. Pearigen et al./Brain Research 725 (1996) 184-191

189

Fig. 5. Heat shock protein immunocytochemistryof rat brain: A: subjectedto 20 min of pO2 20 mmHg, with maintainedblood pressure. B: normalpO2, 4 min of global ischemia. C: normal PO2, 8 min of global ischemia. D: normal pO2, 20 min of global ischemia. E: cortical detail of D: at 100 X animals sacrificed at 24 h.

3.4. Immunocytochemistry / microscopy Fig. 5a shows a typical section from an animal exposed to 20 rain of hypoxia. It demonstrates the absence of HSP-72 expression as a result of hypoxia. For comparison purposes, sections from normoxic animals who experienced global cerebral ischemia by 4-vessel occlusion for 4, 8, or 20 min are shown (Fig. 5 b - e ) . No acid fuchsin staining was seen in any brain section examined from the hypoxic animals (data not shown). In the ischemic animals, acid fuchsin stained neurons were found throughout all regions expressing HSP. With 4 min of ischemia, acid

fuchsin staining was seen in the dentate illius and in the CA 1 sector of the hippocampus. With 20 min acid fuchsin staining was apparant throughout all hipppocampal regions, thalamus and cortex.

4. Discussion These experiments demonstrate that profound hypoxia (pO e = 20) is well tolerated by the brain. We could not determine if further degrees of hypoxia would produce different results, as a lower pO 2 resulted in a fall in blood

190

P. Pearigen et al. / Brain Research 725 (1996) 184-191

pressure and superimposed ischemia. However, Auer and Kiyamota [2] have recently shown that such hypoxia associated hypotension in rats does not result in neuronal necrosis. Their pO 2 was 25 m m H g for 15 min with arterial pressure falling to 30 mmHg. Cerebral blood flow (CBF) was not measured. The experiments presented here confirm the interpretations of Brierley et al. [5] who studied lightly anesthesized primates subjected to a 3.2% oxygen in nitrogen atmosphere. Brain damage resulting from these experiments was rare, and when it occurred was found in arterial border zones: therefore, Brierly concluded that such damage was the result of superimposed ischemia rather than hypoxia. Similar findings were reported by DeCourten-Meyers et al. [8] in cats exposed to 25 min of a p~O 2 of approximately 17. In the absence of superimposed hypotension the hypoxic cats were normal clinically and neuropathologically. W e have extended thse experiments to show that hypoxia results in neither cellular injury (HSP staining) nor in elevation of extracellular glutamate concentration (Fig. 2). Thus, the major toxin in ischemic brain injury [3] is not found in hypoxia. Human data also support the benign nature of hypoxia for the central nervous system. In humans, hypoxia with pO 2 below 50 impairs judgement and performance; loss of consciousness ensues with pO 2 less than 20 [9]. Further information about survival and neuropathologic change in brain is available from patients with hypoxia. At the Yale New Haven Hospital, 22 patients were reported who had arterial pO 2 concentrations less than 20 m m H g (the lowest 7.5 mmHg). Neurologic signs included coma and decerebrate posturing. With resolution of the medical illness which produced the hypoxia, recovery to the premorbid condition occurred in 13 of the 22 patients; neurologic signs were reversed in most patients, including both patients with decerebrate rigidity [11]. Thus, clinical recovery from prolonged hypoxia is recognized. A neuropathologic correlate is provided by Rie and Bernad [14] who reported three young adults in whom acute respiratory hypoxia (pO 2 less than 45 mmHg) developed and persisted for 1 to 8 days. Normotension was preserved throughout and documented by continuous intrarterial blood pressure monitoring. Each patient died of sudden cardiac failure. At neuropathologic examination no evidence of ' h y p o x i c ' brain injury was found [14]. Further, chemical hypoxia using cyanide administration in vivo also causes little neuronal injury if systemic pressure is maintained [4,13]. The proposal of Brierley [4] of nearly 20 years ago is thus supported by the studies of the intervening years. The hypoxic brain has substantial protective mechanisms, particularly the increase of greater than 200% of control in CBF [15]. This may expalin the lack of neuronal necrosis in hypoxic rats whose blood pressure is permitted to fall to 30 m m H g [2]; the enhanced C B F continues to perfuse the brain. With progressive hypoxia, protective mechanisms will eventually be overcome. Systemic hypoxia results in an increase in myocardial contractility with pO2s lowered

to 15 m m H g [7]. However, hypoxia confined to the cerebral circulation results in hypotension and bradycardia at pO2s between 15 and 40 mmHg. In the context of the duration and degree of systemic hypoxia studied here, the brain is tolerant of hypoxia. Neither elevated glutamate concentrations nor cellular injury occur. The contrast of hypoxic injury to ischemia is of note.

Acknowledgements Paul Pearingen was supported in part by National Research Service Award 5 T32 GM07546 from the National Institutes of Health.

References [1] Ananthan, J., Goldberg, A.L. and Voellmy, R., Abnormal proteins serve as eukaryotic stress signals and trigger the activation of heat shock genes, Science, 232 (1986) 522-524. [2] Auer, R.N. and Mivamoto, O., Effect of hypoxia on ischemic and non-ischemic brain., SocieO'for Neuroscience Abstracts, (1995). [3] Benveniste, H., Drejer, J., Schousboe, A. and Diemer, N.H., Elevation of the extracellular concentrations of glutamate and aspartate in rat hippocampus during transient cerebral ischemia monitored by intracerebral microdialysis, J. Neumchem., 43 (1984) 1369-1374. [4] Brierley, J.B., Prior, P.F., Calverley, J. and Brown, A.W., Cyanide intoxication in Macaca mulatta. Physiological and neuropathological aspects, J. Neurol. Sci., 31 (1977) 133-157. [5] Brierley, J.B., Prior, P.F., Calverley, J. and Brown, A.W., Profound hypoxia in Papio anubis and Macaca mulatta physiological and neuropathological effects. 1. Abrupt exposure following normoxia. II. Abrupt exposure following moderate hypoxia, J. Neurol. Sci., 37 (1978) 1-29. [6] Choi, D.W., Glutamate neurotoxicity and diseases of the nervous system, Neuron, 1 (1988)623-634. [7] Cross, C.E., Rieben, P.A., Barron, C.l. and Salisbury, P.F., Prolonged hypoxia in man without ciruclatory compromise fails to demostrate cerebral pathology, Am. J. Physiol., 205 (1963) 963 979. [8] de Courten-Myers, G.M., Yamaguchi, S., Wagner, K.R., Ting, P. and Myers, R.E., Brain injury from marked hypoxia in cats: role of hypotension and hyperglycemia, Stroke, 16 (1985) 1016-1021. [9] Ernsting, J., The effects of hypoxia upon human performance and the electroencephalogram, lnt. Anesthesiol. Clin., 4 (1966) 245-259. [10] Graham, S.H., Shiraishi, K., Panter, S.S., Simon, R.P. and Faden, A.I., Changes in extracellular amino acid neurotransmittersproduced by focal cerebral ischemia, Neurosci. Lett., 110 (1990) 124 130. [I 1] Gray, F.D., Jr. and Hurner, G.J., Survival following extreme hypoxemia, J. Am. Med. Assoc., 211 (1970) 1815-1817. [12] Lowenstein, D.H., Simon, R.P. and Sharp, F.R., The pattern of 72-kDa heat shock protein-like immunoreactivity in the rat brain following flurothyl-induced status epilepticus, Brain Res., 531 (1990) 173-182. [13] MacMillan, V.H., Cerebral energy metabolism in cyanide encephalopathy, J. Cereb. Blood Flow Metab., 9 (1989) 156-162. [14] Rie, M.A. and Bernad, P.G., Prolonged hypoxia in man without circulatory compromise fail to demonstrate cerebral pathology, Neurology (abstract), 30 (1980) 443. [15] Shapiro, H.M., Intracranial hypertension: therapeutic and anesthetic considerations, Anesthesiology, 43 (1975)445-471.

P. Pearigen et al./Brain Research 725 (1996) 184-191 [16] Simon, R.P., Cho, H., Gwinn, R. and Lowenstein, D.H., The temporal profile of 72-kDa heat-shock protein expression following global ischemia, J. Neurosci., 11 (1991) 881-889. [17] Vass, K., Welch, W.J. and Nowak, T.S., Jr., Localization of 70-kDa stress protein induction in gerbil brain after ischemia, Acta Neuropathol. (Berl.), 77 (1988) 128-135. [18] Wu, C., Wilson, S., Walker, B., Dawid, I., Paisley, T., Zimarino, V.

191

et al., Purification and properties of Drosophila heat shock activator protein, Science, 238 (1987) 1247-1253. [19] Xiao, H. and Lis, J.T., Germline transformation used to define key features of heat-shock response elements, Science, 239 (1988) 1139-1142.

[20] Zimarino, V., Tsai, C. and Wu, C., Complex modes of heat shock factor activation, Mol. Cell Biol., 10 (1990) 752-759.