Chronic constant hypoxia (CCH) and chronic intermittent hypoxia (CIH) are known to have deleterious effects on the central nervous system. Because of the difference in the pattern of hypoxic exposure, it is possible that the pathological outcome would vary. The N-acetyl aspartate/creatine (NAA/Cr) ratio is a reliable marker of neuronal integrity, and this can be noninvasively measured by proton nuclear magnetic resonance spectroscopy. P2 CD1 mouse pups with their dams were exposed to either CCH, where the FiO2 was maintained at 11% continuously or to CIH, where the FiO2 was varied between 21 and 11% every 4 min. P30 mice exposed to intermittent hypoxia for 4 wk demonstrated a significant decrease in the NAA/Cr ratio in the hippocampus and thalamus, which was reversed by a subsequent exposure to 4 wk of normoxia. Meanwhile, mice exposed to 4 wk of constant hypoxia did not demonstrate any differences in their NAA/Cr ratios from controls in these brain regions. These results indicate that an intermittent pattern of hypoxic exposure may have a more adverse effect on neuronal function and integrity than a continuous one. The reversal of NAA/Cr levels to baseline during the return to normoxia indicates that therapeutic strategies targeted at alleviating the intermittent hypoxic stress in diseases, such as obstructive sleep apnea, have the potential for inducing significant neurocognitive recovery in these patients.
- magnetic resonance imaging
- energy metabolism
- oxygen deprivation
hypoxia often results in deleterious effects on central nervous system (CNS) function and viability (29, 34). The length and extent of hypoxic exposure, as well as the age at the time of exposure, determine the severity of the clinical outcome (4). Two paradigms of hypoxia have been investigated extensively in the recent past: chronic constant hypoxia (CCH) and chronic intermittent hypoxia (CIH). However, it is not clear whether structure or function is altered to the same degree in various tissues in response to these two patterns of stress. We have started to study this aspect of hypoxia, and we have previously demonstrated that right ventricular hypertrophy and cardiac enlargement occur in the CCH but not in the CIH-exposed mouse heart (20).
Our laboratory has long been interested in hypoxia and its effect on brain development and function. Hypoxia can be observed in clinical situations such as chronic obstructive pulmonary disease, hypoventilation in newborns, congenital heart disease with right-to-left shunts and asthma, as well as life at high altitude. Intermittent hypoxia occurs during apnea of prematurity, obstructive sleep apnea (OSA) and sickle cell disease. OSA is characterized by periodic breathing, episodic hypoxemia, and repeated arousal from sleep (44, 66). OSA and other sleep-disordered breathing states are associated in children with neurocognitive abnormalities, including attention deficit and poor school performance (4, 25). Various regions in the CNS, such as the subcortex and especially the hippocampus, have been shown to be more vulnerable than others to hypoxic exposure (47). For example, it is known that hypoxia/ischemia can result in significant damage to the hippocampal formation and that intermittent hypoxic exposure can lead to apoptosis in the hippocampal CA1 region (26, 27).
Our laboratory has been interested in developing tools for the noninvasive assessment of CNS function (32). In vivo proton nuclear magnetic resonance spectroscopy (1H-MRS) measurements of cerebral N-acetyl aspartate (NAA) content have been widely used as a surrogate measure for neuronal injury in a variety of pathologies, including epilepsy (31), Alzheimer's disease (38), multiple sclerosis (48), and stroke (13, 65). Here, we utilize 1H-MRS to noninvasively study the effect of hypoxia on CNS function in animals exposed to either CCH or CIH.
MATERIALS AND METHODS
Pregnant CD-1 mice were obtained from Charles River Laboratories (Wilmington, DE). Care was exercised in the handling of these animals, and the minimal number of animals that was absolutely required was used in this study. This study was conducted in conformity with the Guiding Principles for Research Involving Animals and Human Beings and was approved by the Albert Einstein Animal Care and Use Committee. Animal experiments were carried out in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals (NIH Publications No. 80–23), revised 1996.
Postnatal day 2 (P2) CD-1 mouse pups were housed with their dams in isobaric chambers. Litters were culled to eight to ensure adequate nutrition of the pups. Food and water were provided ad libitum within the chamber. The protocols used to expose mice to hypoxia have been previously described (18, 37). In brief, a combination of nitrogen (N2) and oxygen (O2) was injected into the chambers through a network of tubing to achieve selected concentrations of O2. This was controlled by the Oxycycler hydraulic system (model A44x0, BioSpherix, Redfield, NY) and ANA-Win2 Software (Version 2.4.17, Watlow Anafaze, CA). For CCH, the O2 level was maintained at 11 ± 0.1% continuously. For CIH, every cycle consisted of a 4-min period during which O2 concentration was maintained at 11 ± 0.1% followed by a 4-min period at 21 ± 0.1%. The ramp time between the two levels took <1–2 min. This cycle was repeated throughout the hypoxia exposure experiments. Normal control mice were kept in the same room and were exposed to the same level of noise and light during the duration of each experiment. Two different litters of mice, with their respective controls, were exposed to CCH and CIH for 4 wk, starting from P2. We selected a duration of 4 wk on the basis of previous studies showing a significant increase and decrease in capillary density and myelination, respectively, in the CNS, as well as significant changes in acid-base protein expression in response to hypoxia (18, 37). After the period of hypoxic exposure, P30 mice exposed to 4 wk of CCH (n = 5) and CIH (n = 4), as well as age-matched controls (n = 5), were assayed by NMR spectroscopy for metabolite levels. The mice that had been exposed to CIH and tested for metabolite concentrations by 1H-MRS were returned to their littermates and maintained in normoxia for another 4 wk. CIH-exposed mice, including those studied at 4 wk, which were returned to normoxia for 4 wk (n = 6), were then again assayed for metabolite levels by 1H-MRS.
Measurement of Brain Metabolites by NMR Spectroscopy
Control and hypoxia-exposed CD-1 mice (20–25 g) were used in this study. Within 2–6 h after removal from the environmental chambers, the mice were anesthetized with a face mask using 1.5% isoflurane (AErrane, Baxter; Deerfield, IL) and pure oxygen. The mice were maintained at a constant temperature, 37°C, throughout the experiment using a circulating water blanket controlled via rectal temperature monitoring. All experiments were carried out on a 9.4-T, 21-cm horizontal bore imaging system (Varian INOVA, Palo Alto, CA) using an actively detunable birdcage coil for transmission and a 8 × 12-mm elliptical surface coil for reception. Scout images were acquired using an inversion recovery gradient echo imaging sequence (128 × 128 resolution, 0.5-mm slice thickness, 24 × 24-mm field of view, 0.9-s inversion time).
All first- and second-order shim terms were automatically adjusted for each slice using a B0 map-based shim calculation method. The detail of this method is described elsewhere (3). Spectroscopic images of the mouse brain were acquired using a modified LASER (localization by adiabatic selective refocusing) sequence (2) with 1-mm slice thickness. Water suppression was obtained by use of a semiselective excitation pulse and a selective 90° water suppression pulse. Two dimensions of 24 × 24 phase encodes were acquired over a field of view of 24 mm × 24 mm, resulting in a nominal voxel size of 1 μl. The data were filtered in the spatial domain with a hanning filter. The data were acquired with a repetition time of 2,000 ms, and an echo time of 50 ms using two averages for a total acquisition time of 38 min.
To compare regional differences in metabolites, N-acetyl aspartate/creatine (NAA/Cr) ratios were measured from the spectra obtained from the hippocampus and thalamus. These CNS sites were chosen on the basis of their sensitivity to hypoxic insult (39, 58). To improve the spatial accuracy for the selection of each brain region, the voxels across each brain region were selected using an image-guided single voxel reconstruction algorithm using voxel-shifting methods (49, 62). Briefly, the hippocampi (Fig. 1) were outlined manually (red and green lines), and a midline was calculated (blue line). For the thalamus, a single volume was identified, including the bilateral thalami (red line), and a midline was calculated (blue line). Voxels were reconstructed along the midline (yellow circles), starting from the most medial position automatically, yielding six voxels for the hippocampi and five for the thalamus. The voxels were reconstructed by translating along the calculated midline with a voxel-to-voxel spacing of 1.13 mm so as to eliminate spatial overlap and provide spatially independent measures. For the thalamus and hippocampus, a mean value was calculated by averaging over the reconstructed voxels (49, 62).
Body weight data are reported as means ± SE and were analyzed using the Student's unpaired t-test (GraphPad Prism ver. 4.00 for Windows, GraphPad Software, San Diego, CA). Differences in the means were considered statistically significant when P < 0.05. 1H-MRS data are reported as means ± SD. The data were analyzed using Sigma Stat 3.01 for Windows (SPSS, Systat Software, Point Richmond, CA). To evaluate the difference between NAA/Cr values for the hippocampus and thalamus, due to intrinsic metabolic heterogeneity, a two-tailed paired t-test was used in the control animals, with P < 0.05 for statistical significance. To test the hypothesis that CCH and/or CIH results in neuronal injury (decreased NAA/Cr values) in the thalamus and hippocampus, we compared NAA/Cr in the hippocampus and thalamus of CCH and CIH animals to the corresponding values (hippocampus or thalamus) in the control animals. Statistical significance for the hypothesis was assessed using an unpaired single tailed t-test. To correct for evaluating two experimental groups (i.e., CCH and CIH), statistical significance was set at P < 0.025. On the basis of the statistically significant alteration in the CIH group, we then evaluated the hypothesis that a return to normoxia would significantly improve the NAA/Cr value in the CIH animals. An unpaired single tailed t-test was used to assess for statistical significance, using P < 0.05, in comparing CIH animals following a return to normoxia for 4 wk (CIH-N) to their original controls.
CD1 mice that had been exposed to 4 wk of CCH and CIH demonstrated a decrease in body weight. Although there is variability, both CIH and CCH groups were compared with age-matched controls; with P2 litters being matched in weight between control and experimental, each time the experiment was done. We matched CIH or CCH litters at P2 with their respective controls because P2 weights can affect body weight gain, besides hypoxia (unpublished observations). The body weight of young (P30) mice exposed to CCH (22.6 ± 0.5 g; n = 16) for 4 wk was significantly less than that of their age-matched controls (25.6 ± 0.8g; n = 8; P < 0.03), as was the body weight of young (P30) mice exposed to CIH (19.94 ± 0.4 g; n = 15) for 4 wk, compared with their age-matched controls (21.38 ± 0.6g, n = 16; P = 0.001). In each case, body weights of control mice were greater than that of mice in the CIH and CCH groups. However, body weights of P60 control mice (37.9 ± 2.2 g, n = 2) and P60 mice that were returned to normoxia for 4 wk (after 4 wk of CIH) (37.7 ± 2.1 g; n = 6) were similar. Although n is relatively small for the P60 mice, these weights are similar to data that we have reported previously in a much larger study (37).
On the basis of the echo time used, 50 ms, three dominant resonances are seen in the 1H spectrum, NAA (2.0 ppm), creatine (Cr) (3.0 ppm), and choline (Ch) (3.2 ppm) (Fig. 2). Spectral overlap of NAA and Cr with resonances from amino acids (glutamate and glutamine) and macromolecules is eliminated by J-modulation at this echo time. Elimination of the other resonances enhances the accuracy of the measurement of the NAA and Cr resonances. Measurements of NAA/Cr from the hippocampus and thalamus were averaged over 6 and 5 pixels, respectively, for each animal. The thalamic NAA/Cr (0.93 ± 0.05) was significantly higher (P = 0.003, two-tailed paired t-test) than hippocampal NAA/Cr (0.82 ± 0.05). In all groups studied, the same pattern was seen, that is, significantly larger NAA/Cr in the thalamus compared with the hippocampus (P = 0.03, constant hypoxia; P = 0.008 intermittent hypoxia; and P = 0.011, animals returned to normoxia after exposure to intermittent hypoxia) (Fig. 3, A and B).
Displayed in Fig. 2 are spectra from a control mouse and a mouse exposed to 4 wk of intermittent hypoxia at P30. As seen in this figure, there is a substantial decline in NAA/Cr in the hypoxic mouse relative to the control mouse. As a group, compared with control mice, mice exposed to CIH for 4 wk showed a statistically significant decline in NAA/Cr in the hippocampus (0.71 ± 0.07 vs. 0.82 ± 0.05 in controls, P < 0.015, one-tailed t-test) and thalamus (0.85 ± 0.04 vs. 0.93 ± 0.05; P < 0.017, two-tailed t-test) (Fig. 3, A and B). Unlike the animals subjected to intermittent hypoxia, no statistically significant changes were found in the mice exposed to CCH for 4 wk compared with control mice in either the hippocampus (0.80 ± 0.04) or thalamus (1.00 ± 0.16).
Following CIH-N, mice exposed to intermittent hypoxia showed a statistically significant increase in NAA/Cr in both the hippocampus (0.83 ± 0.05, P = 0.006, one-tailed t-test) and thalamus (0.97 ± 0.09, P < 0.015, one-tailed t-test), returning to values very similar to the control mice, 0.82 and 0.93, respectively (Fig. 3, A and B). Thus the neuronal injury induced by intermittent hypoxia, that is, reduced NAA/Cr, was reversible.
It has become increasingly apparent that the pattern of hypoxic exposure or stimulus presentation has an impact on neuronal injury and long-term neurocognitive sequelae (46, 53). Therefore, the purpose of this study was to determine the extent of damage incurred in the mouse CNS due to exposure to two different paradigms of hypoxia exposure: CCH and CIH, using a noninvasive protocol. 1H-MRS is a noninvasive method for the in vivo assessment of pathological and metabolic changes in the CNS. Consequently, we used 1H-MRS to measure the relative concentrations of NAA and creatine during and after hypoxic exposure in mice. The reason for using NAA normalized to creatine is that it has been used as an indicator of neuronal loss or neuronal injury (61) and minimizes uncertainties in concentration estimates arising from variable inclusion of cerebrospinal fluid, which contains negligible amounts of both metabolites. NAA is found in high concentrations (8–11 mM) in mammalian brain (3, 6) and is readily detected by 1H-MRS. NAA is synthesized only in neuronal mitochondria (23) and is localized within neurons and neuronal processes (63) as NAA is synthesized from l-aspartic acid and acetyl-CoA by an enzyme, l-aspartate N-acetyltransferase (Enzyme Commission no. 22.214.171.124), which is only found in neuronal mitochondria (23, 60).
The Cr resonance is composed of resonances from Cr and phosphocreatine (PCr). The concentrations of Cr and PCr are relatively similar; however, in vivo NMR measurements show the total of both PCr and Cr concentrations as a prominent singlet at 3.0 ppm (24). Thus conversion of PCr to Cr and vice versa due to acute changes in energetic demand or status, do not alter the total Cr resonance therefore, the NAA/Cr ratio has been considered a reliable marker of neuronal integrity (10, 14).
CIH Causes a Decrease in NAA/Cr Levels
Although CCH did not have a major impact on metabolite ratios in the CNS, CIH significantly decreased the NAA/Cr ratio in the hippocampus and thalamus of young (P30) mice after 4 wk of exposure. That NAA/Cr levels were decreased in response to hypoxia may not be surprising. For example, in utero hypoxia and substrate depletion lead to decreased NAA/Cr levels in chick embryo brains (17). Additionally, Kamba et al. (36) have reported significant decreases in NAA/Ch ratios, but only modest decreases in NAA/Cr in cerebral white matter of patients with modest to severe OSA compared with mild OSA and controls. Decreased NAA has also been observed in piglets during hypoxia-ischemia (52) and during global ischemia in gerbils (54). The reduction of NAA in traumatic brain injury and birth asphyxia appears to be related to energetic impairment (59) via secondary energy failure (7).
The mechanism by which CIH leads to decreased NAA levels is unknown at present. However, it has been proposed that chronic intermittent hypoxia leads to a reduction in cerebral tissue oxygenation that is attributed to cerebral vascular dysregulation, an abnormality that is reversible (21). It is also possible that the oscillatory nature of intermittent hypoxia, which has periods of hypoxia and, importantly, periods of increased oxygenation, could induce mitochondrial dysfunction or injury that would then lead to decreased NAA/Cr levels. It has been reported that OSA, which is being mimicked by the oscillatory nature of CIH, may lead to increased reactive oxygen species production (19, 56), which impairs mitochondrial function and therefore NAA synthesis. Recently, Pan and Takahashi (51) have demonstrated that in vivo levels of NAA in the healthy human brain are strongly correlated with cerebral ADP levels, reinforcing the linkage between NAA and mitochondrial function, as initially described by Bates and colleagues (5, 30) in studies of neuronal mitochondria.
What is surprising is that NAA/Cr levels were not decreased in the brains of young (P30) mice exposed to CCH for 4 wk. Because hypoxia has been shown to decrease NAA levels in the CNS, it is possible that the long duration of continuous hypoxic exposure utilized in our paradigm led to a recovery of NAA levels. The implication is that adaptive mechanisms may have been invoked to compensate for the reduced O2 availability (12). Chavez et al. (12) and others have reported that CCH leads to increased hypoxia inducible factor-1α and vascular endothelial growth factor expression, capillary density and some forms of neuronal plasticity (9, 41, 43). Therefore, gradual or prolonged reduction in O2 availability may allow time for the development of multiple homeostatic mechanisms, which help to maintain normal brain function (22).
Recovery of NAA/Cr Levels During the Return to Normoxia After CIH
Interestingly, the decrease in NAA/Cr induced by CIH in both the hippocampus and thalamus was alleviated by a return to normoxia over a period of 4 wk. The present investigation and those of others would suggest that hypoxia-induced decrements in NAA levels was without apparent neuronal cell death since the levels of NAA/Cr are reversible after CIH. Indeed, although NAA was believed to be a measure of neuronal loss, longitudinal studies in patients with epilepsy and studies correlating neuronal loss and NAA have demonstrated a lack of correlation between decreases in NAA/Cr and neuronal loss (10, 14, 35, 40, 57, 61). In fact, quantitative studies of NAA content have shown a strong correlation with measures of neuronal function in Alzheimer's disease (64) and multiple sclerosis (50). These strong correlations are most likely reflective of the high cost of cortical function (2) and therefore the brain's dependence on intact bioenergetics. Therefore, it is suggested that mitochondria decrease energy consumption and utilization during chronic hypoxia.
Similarly, while body weight was decreased in response to either CCH or CIH, there was no difference in body weight among previously hypoxic mice and their normoxic counterparts subsequent to 4 wk of normoxia (37). This indicates that if the hypoxic stress is removed, there is the potential for significant recovery of certain physiological processes in the rodent. This recovery may reflect either neurogenesis or restoration of function in existing neurons. For example, there are several reports that indicate that NAA levels can recover if corrective measures are taken. Recovery of NAA has been demonstrated in patients after remission of early MS lesions (1, 15), in the contralateral temporal lobes after successful resection of the epileptic focus (11, 35) and after incomplete reversible ischemia (8) and brain injury (16). This allows us to postulate that cellular dysfunction occurring during CIH leads to a decrease in NAA that can be reversed during a return to normoxia. However, there are biological systems in mammals that are not responsive to amelioration of the etiological stress, and therefore there are long-term deficits that are accrued during such stresses. For example, while capillary density recovers to control levels during a return to normoxia, demyelination does not (37).
A question that has been debated recently in the literature is whether CIH is more deleterious than CCH. Our data and those of others may suggest that CIH can have a more adverse outcome. Gozal et al. (27) have reported that CIH induces apoptosis and significant neurocognitive deficits to a greater extent than does CCH of similar duration. Patients with OSA have been reported to suffer grey matter loss in the hippocampus (44) and in the CNS, in general (42). Also, repetitive, intermittent hypoxia-ischemia rather than continuous hypoxia-ischemia leads to pronounced damage in the immature rat brain (45). If OSA is treated, it is possible that pathological sequelae induced by sublethal hypoxia can be reversed in the CNS.
The underlying causes of differences in the CNS response to CCH and CIH may lie in the cellular response to the two different paradigms. Evidence is accumulating supporting the notion that there are distinct differences between CIH and CCH, as well as their effects on the CNS and disease outcome. For example, Gozal et al. (28) have recently demonstrated significant differences in the apoptotic profile of PC12 cells exposed to mild CIH and CCH, i.e., 2 and 4 days of 5% O2. Although sustained hypoxia only induced apoptosis without caspase activation at 4 days in PC-12 cells, mild IH produced significant apoptosis at 2 days and 4 days associated with enhanced caspase activity. It has been suggested that, in contrast to CCH (9, 33), CIH may fail to induce adaptive responses in neurons (28). Because decreased NAA levels have been correlated with clinical and cognitive disability (55), it is possible that patients who experience intermittent hypoxia, such as occurs in OSA, may be compromised to a greater extent than those suffering from continuous hypoxic exposure. This hypothesis needs further study to be verified.
The present study indicates that NAA/Cr levels can recover if the intermittent hypoxic stress is removed. We believe that this is the first report demonstrating reversibility of NAA/Cr after CIH, indicating that a return to normoxia can permit neurons to recover normal function. This reversible neuronal function in response to a return to normoxia may be important clinically such as in the correction of sleep apnea with continuous positive airway pressure or other modalities which improve oxygenation during sleep.
This study was supported by National Institutes of Health Grants EB001743, HD32573, and NS042202.
We would like to thank Orit Gavrialov and Vera Desprat for excellent technical assistance.
Present addresses: R. M. Douglas, Department of Pediatrics, University of California, San Diego, School of Medicine, 9500 Gilman Dr., 0735, La Jolla, CA 92093; and G. G. Haddad, Departments of Pediatrics and Neuroscience, University of California, San Diego, School of Medicine, 9500 Gilman Dr., 0735, La Jolla, CA 92093.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society