Animals were placed from Day 5 of pregnancy until delivery in computer-controlled environmental chambers, in which the moment-to-moment desired oxygen concentration of the chamber was programmed and adjusted automatically. The IH profile consisted of alternating room air (RA) and 10% oxygen every 90 seconds from Day 5 of pregnancy until delivery during the light phase. Matched control animals were exposed to circulating normoxic gas in one of the chambers. Within less than 6 hours after delivery, litters were removed from the chambers, culled to eight pups, and maintained in normoxic conditions until measurements (see online supplement).

Cardiorespiratory measures were continuously acquired in the freely behaving animals, using the barometric method (Buxco Electronics, Troy, NY) (15, 16) with appropriate corrections for inspiratory and expiratory barometric asymmetries (17). Environmental temperature was maintained slightly below the thermoneutral range (26–31°C dependent on postnatal age as previously described [18]). Animals were allowed to acclimate to the chamber, and pressure changes due to the inspiratory and expiratory temperature changes were measured with a high-gain differential pressure transducer (model MP45-1; Validyne, Northridge, CA) (19). Analog signals were continuously digitized, and analyzed on-line by a microcomputer software program (Buxco Electronics). A rejection algorithm was included in the breath-by-breath analysis routine and allowed for accurate rejection of motion-induced artifacts. Tidal volume (Vt), respiratory frequency (f), minute ventilation (V̇e), mean inspiratory flow (Vt/Ti), and inspiratory duty cycle (Ti/Ttot) were computed and stored for subsequent off-line analysis (see online supplement).

Rat pups were studied on postnatal Days 5, 10, 15, and 30. The animals were weighed and placed in the barometric recording chamber. After stable baseline normoxic values were obtained for at least 5 minutes, rats were switched to 10% O₂ (balance N₂) using a premixed gas mixture. The hypoxic challenge lasted for 20 minutes, after which room air was reintroduced in the recording chamber, and recovery was recorded for 10 minutes.

In an additional eight IHRA and eight normoxia (RARA) 5-day-old pups, the Dejours test was conducted to assess the overall sensitivity of peripheral chemoreceptors. To this effect, after a stable normoxic baseline was established, 100% O₂ was flushed into the whole body plethysmograph, and the overall change in ventilation over the initial 15 seconds from gas transition was determined as peripheral chemoreceptor gain (20).

Ventilatory measures were averaged at 1-minute intervals and plotted. Analyses of variance (ANOVAs; one-way or two-way) were employed to compare normoxic ventilation, peak ventilatory response to hypoxia (pHVR), and hypoxic ventilatory depression (HVD). Significant comparisons were followed by Newman–Keuls tests. A p value < 0.05 was considered to achieve statistical significance.

Metabolic rate was measured with a computer-controlled indirect calorimetry system similar to that described by Jensen and colleagues (21). Air taken from a common air source was pulled through the plethysmographic chambers, a blank chamber, and a reference line connected via separate but identical air-sampling pathways. The blank chamber was used as reference to monitor ambient O₂ and CO₂ concentrations periodically. V̇o₂ and V̇co₂ values (corrected for body weight), and corresponding to 10 consecutive samples were calculated and averaged at time points representing normoxic baseline, pHVR, and HVD (see online supplement).

In a subset of 15-day-old rats, PE-10 catheters were introduced into the abdominal aorta under general anesthesia (pentobarbital [Nembutal], 30 mg/kg, intraperitoneal), and exteriorized in the dorsal aspect of the neck as previously described (22). Arterial blood samples were obtained from the implanted arterial catheter in the rats. After withdrawal of 75–100 μl of blood in the dead space of the catheter, another 150 μl was sampled for immediate analysis of Po₂, Pco₂, and pH with a blood gas analyzer (ABL150; Radiometer, Copenhagen, Denmark).

The Morris water maze was configured to test performance that reflects types of learning and memory traditionally defined as spatial reference, spatial working, and nonspatial reference (23, 24). A Plexiglas escape platform (12 cm in diameter) was positioned 2 cm below the water surface and placed at various locations throughout the pool. The platform was retracted manually during probe trials. A circular black curtain surrounded the pool and extended from the ceiling to the floor. For spatial tasks that depended on allocentric strategies, white extramaze cues were hung from the curtain at fixed locations. Maze performances were recorded by a video camera. For water maze experiments, only male rats were tested to prevent potential variability introduced by the estrous cycle on maze learning (25). Animals were tested at 30 days, an age at which no sex differences occur (26), and subsequently at 120 days of age (an age that clearly exhibits sex-dependent effects [25]). One day before place learning, male rats were habituated to the water maze during a free swim. Place learning was assessed on four daily training trials separated by 180 seconds over 4 days. On a daily session, each littermate was placed into the pool from four quasi-random start points and allowed a maximum of 90 seconds to escape to the platform, where he was allowed to remain for 15 seconds. Rats that failed to escape were led to the platform. The position of the platform remained constant across trials. After the final training trial of each day, the platform was retracted for a 30-second probe trial during which the time spent in each of the four quadrants of the tank and the number of target crossings over the previous location of the platform were recorded. Probe trials thus provide a measure of spatial bias developed during learning. To assess performance during place training, mean escape latencies, swim distances, and swim speeds were analyzed by two-way ANOVA (hypoxia condition, four-trial block) with repeated measures on block, followed by Newman–Keuls tests when appropriate. In addition, the occurrence of specific search strategies used during place training was noted and analyzed. To assess probe performance, mean quadrant times and target crossings were compared by two-way ANOVA (hypoxia condition and trial) with repeated measures on trial followed by Newman–Keuls tests when appropriate (see online supplement).

Rat pups born to IH-exposed mothers (IHRA) weighed significantly less than control animals (RARA; Figure 1)

Figure 1. Mean body weights at various postnatal ages of pups born to rats exposed to either intermittent hypoxia (solid circles) or room air (open squares) during gestation. n = 18/group; *p < 0.01, intermittent hypoxia and room air (IHRA) versus normoxia (RARA).

Download Figure | Download Powerpoint

[More] [Minimize]

At all postnatal ages, normoxic ventilation was significantly greater in IH-exposed rat pups when compared with controls (Figure 2

Figure 2. Normoxic minute ventilation at various postnatal ages of pups born to rats exposed to either intermittent hypoxia (solid circles) or room air (open squares) during gestation. n = 18/group; *p < 0.01, IHRA versus RARA. Minute ventilation is expressed as milliliters per minute per gram body weight.

Download Figure | Download Powerpoint

[More] [Minimize]

In 5-day-old pups, 100% oxygen resulted in a mean V̇e reduction of 23.2 ± 4.7% in RARA pups and of 7.2 ± 3.5% in IHRA pups (p < 0.03). Thus, prenatal exposures to intermittent hypoxia attenuate peripheral chemoreceptor sensitivity.

Twelve IH-exposed and 12 control male rats were studied at 30 and 120 days postnatally. No differences emerged in the ability to acquire the spatial task in the two treatment groups, such that the mean latencies and distances that the rats swam in the pool were similar across trials (Figure 5)

Figure 5. Mean latency to access the target platform (seconds; top left), swim distance (cm; top right), and swim speed (cm/second; bottom right) to locate the target platform during place training in 30-day-old pups born to rats exposed to either intermittent hypoxia (solid circles) or room air (open squares) during gestation. Bottom left: Relative proximity to the target platform location during probe trial in these animals. No significant differences emerged during either spatial task acquisition or retention (probe trial). n = 12/group.

Download Figure | Download Powerpoint

[More] [Minimize]

The effects of hypoxia on somatic growth have been relatively well studied. Most studies support the concept that low birthweight is a frequent result of stunted fetal growth due to placental hypoxia elicited by a sustained hypoxic environment such as in high altitude (27–29). In animal models, gestational exposure to hypoxia elicited stunted fetal growth (30, 31) and, in fact, even short (1–2 hours) exposures during the last third of pregnancy in rats will result in significant reductions in fetal body weight and selectively affected organ growth (32). However, it is possible that catecholamines released via the stress response to IH, rather than hypoxia per se, may underlie the reduced fetal growth associated with gestational IH (33).

Additional evidence concerning human pregnancies suggests fetal growth retardation occurs in offspring of snoring women (11, 34). Snoring was postulated to impact on maternal cardiorespiratory reserve and indirectly on the fetus, and recordable changes in fetal heart rate and in the acid–base status of the fetus were identified (35, 36). It is possible that the altered placental function could ensue as a result of vascular changes induced by obstructive sleep apnea and the accompanying intermittent hypoxia, the latter possibly inducing or aggravating preexisting preeclampsia (10, 37). In support of this possibility, Edwards and colleagues have shown the favorable effect of continuous positive airway pressure on systemic blood pressure in pregnant women with preeclampsia (38).

The application of intermittent hypoxia during daylight, the rest–sleep period component of the circadian phase in our rat model, obviously does not completely mimic all of the alterations associated with sleep-disordered breathing, because neither hypercapnia nor sleep fragmentation is present (12). However, IH has been clearly identified as the major contributor to vascular dysfunction (39) and hypertension in sleep apnea (37, 40). Thus, it is possible that the fetal effects of intermittent hypoxia during sleep will be further exacerbated when underlying vascular dysfunction, for example, preeclampsia or diabetes mellitus, is already present.

Normoxic ventilation was increased after fetal intermittent hypoxia, and this effect persisted well into adulthood. However, attenuation of the ventilatory equivalent responses to acute hypoxia was also present during the first postnatal month. These long-lasting changes in respiratory control are not only indicative of the induction of respiratory plasticity, but also suggest that the latter can occur through application of conditioning stimuli in utero. Indeed, small for gestational age newborn infants who developed intrauterine growth retardation exhibit subtle, albeit significant differences in respiratory rhythm during sleep, and such differences have been attributed to alterations in brainstem development (41). Furthermore, chronic gestational hypoxia is associated with increased methionine-enkephalin and altered substance P concentrations in the pontomedullary brainstem regions (42, 43). The functional ventilatory outcome of gestational hypoxia, as evidenced from a study by Gleed and Mortola and the present study, suggests the emergence of relative hypoplasticity of lung tissue and increased basal brainstem respiratory rhythm activity, resulting in increased ventilatory output even when adjusted for underlying metabolism (44).

The mechanisms underlying the neural plasticity associated with manipulations of oxygen concentration in breathed air have not been extensively investigated, and in fact little is currently known about the effect of prenatal hypoxia on respiratory control. Before we address this particular issue however, a methodological problem pertaining to the analysis of ventilatory responses to hypoxia needs to be examined. In the context of increased normoxic ventilation, actual ventilatory measurements may be similar at pHVR, for example, yet the relative change will be smaller (e.g., 5-day-old pups; Figures 2 and 3). These ventilatory responses could be then construed as unaffected (based on actual ventilatory measures) or attenuated (based on relative change from normoxic baseline). This problem is not apparent during HVD, whereby both absolute and relative ventilation in 5-day-old pups was decreased compared with control animals (Figure 3). The excitatory respiratory plasticity that occurs after intermittent hypoxic stimulation and manifests as increased normoxic ventilation has been ascribed to the sustained induction of serotoninergic neurotransmission systems with opposing effects from downregulation of noradrenergic-dependent neural transmission (45, 46). This is in contrast with the absence of such long-term facilitation when the stimulus is continuous hypoxia (45, 47). It has been further postulated that these changes in serotonin and noradrenergic transmission will attenuate fluctuations in ventilatory drive, especially when stimuli change rapidly (48). As such, application of episodic hypoxia to the adult rodent for 7 days resulted in increased overall phrenic nerve output responses to hypoxia and to carotid sinus nerve stimulation that were dependent on serotonin receptors (49). However, episodic hypoxia may not always induce enhancements of subsequent acute hypoxic ventilatory responses. For example, exposure of 2- to 3-day-old rat pups to a series of eight cycles of IH consisting of 5 minutes of hypoxia (i_O2, 0.10) and 10 minutes of normoxia revealed that the late phase of HVR but not the peak HVR was modified by the intermittent hypoxia paradigm when compared with normoxia (50). The major effects consisted of reductions in HVD that were attributed at least in part to an increased expression of the neuronal isoform of nitric oxide synthase within the caudal brainstem (50). Similarly, Trippenbach reported that repeated and successive hypoxic runs will elicit enhancements of the HVR in rabbit pups during a subsequent hypoxic challenge (51). These findings concur with the postulated increase in peripheral chemoreceptor sensitivity attributed to episodic hypoxia exposures (52). In contrast with such findings, however, Waters and Tinworth showed that in piglets repeated daily exposure to hypoxia (30 minutes/day) induced attenuated respiratory responses to a subsequent acute hypoxic stimulus (53). Our present results as derived from both hypoxic and hyperoxic exposures would suggest that peripheral chemosensitivity to hypoxia is reduced and not enhanced after long-lasting cyclical hypoxic exposures. Furthermore, such repeated hypoxic exposures also increased the interstitial concentrations of substance P and methionine-enkephalin within the nucleus of the solitary tract (42, 43, 54, 55), suggesting that in addition to serotoninergic and adrenergic systems, other respiratory-relevant neurotransmitter systems may also be affected. Because N-methyl-d-aspartate (NMDA) glutamatergic transmission is critical to HVR in the adult rat (56), the substantial developmental changes in expression and function of the NMDA glutamate receptor that occur within critical brainstem respiratory sites (57) are responsible for maturation and plasticity of neuronal systems in multiple other brain regions (58, 59). Thus, it is possible that altered NMDA receptor transmission is elicited by IH and accounts for the findings reported herein. Indeed, in young piglets, NMDA receptor-mediated Ca²⁺ influx and receptor nitration will increase with hypoxic severity, and such changes in NMDA receptor function are manifest within 1 hour of a single exposure to hypoxia (60). Exposure to intermittent hypercapnic hypoxia over a period of 4 days was associated with differential changes in the expression of the NR1 NMDA receptor mRNA (increased) and NR1 protein (decreased) (61). Furthermore, in the adult rat dorsocaudal brainstem, long-lasting intermittent hypoxia (30 days) was associated with significant changes in NMDA NR2 receptor subtype expression (62). Taken together, these findings suggest that intermittent hypoxia will induce substantial alterations in multiple downstream neurotransmitters or signaling molecules (e.g., nitric oxide, protein kinase C) that modulate respiratory control, and that such changes exhibit significant dependency on the developmental stage at which they occur (63).

This study revealed that gestational IH was associated with significant attenuations of the peak hypoxic ventilatory response and enhanced late hypoxic depression in the younger neonatal rats. However, when ventilation was adjusted for oxygen consumption, IH-exposed animals displayed lower ventilatory recruitments during both phases of the hypoxic response. Interestingly, both control and IH-exposed pups displayed clear developmentally regulated patterns of ventilatory equivalent responses to hypoxia, in which late decreases in V̇e/V̇o₂ were particularly prominent during postnatal ages 10–15 days. However, such decreases were enhanced after gestational IH, suggesting the possibility of an age- and IH-dependent selective vulnerable period. Indeed, early exposure to intermittent hypoxia will lead to significant alterations in the gasping responses to asphyxia and in the ability to autoresuscitate (64). Qualitatively similar results were reported in the newborn piglet metabolic response to intermittent cycles of hypoxia, whereby age-dependent failure of ventilatory and metabolic adaptations emerged at 2 weeks of age and manifested as more severe lactic acidosis (65), and in rats exposed prenatally to sustained hypoxia (2). Thus, further exploration of the conceptual framework of fetal IH-induced postnatal selective hypoxic vulnerability may help to explain why face-down positioning and repetitive apnea appear to lead to sudden infant death syndrome in only selected infants.

The absence of any alterations in water maze performance in IH-exposed rats was somewhat unexpected despite the relatively mild magnitude of the hypoxic stimulus. Indeed, significant differences in neurobehavioral performance have been previously noted following hypoxia during gestation (3, 5). Most of these differences, however, manifest as subtle alterations in gait behaviors and sexual maturation, particularly testosterone regulation (4, 5), rather than as detectable changes in water maze task acquisition or retention. In addition, Nyakas and colleagues have also shown that abnormal open-field, social, learning, and emotional behaviors, as well as altered plasma corticosterone responses to stress, occur throughout adulthood after exposure to prenatal hypoxia (66, 67). Similarly, when the IH paradigm used in the present study was applied to 10-day-old postnatal rat pups for 15 days, marked decreases in the ability to acquire and retain a spatial task in the water maze emerged (26). Thus, either the gestational IH profile does not lead to detectable neurobehavioral deficits in the offspring, or these changes, if indeed present, are readily reversed during the first month of life (i.e., representing a form of compensatory neural plasticity), and do not impose long-lasting changes in the acquisition and retention of spatial function.

In summary, we have shown that the occurrence of gestational intermittent hypoxia is associated with prolonged and a priori irreversible alterations in respiratory control, but does not induce any detectable obvious neurocognitive deficits. We postulate that intermittent hypoxia during pregnancy may lead to maladaptive responses to postnatal hypoxia and potentially increase the vulnerability to fatal conditions such as sudden infant death syndrome in selected infants.