The catecholamine norepinephrine is required for fetal survival, but its essential function is unknown. When catecholamine-deficient [tyrosine hydroxylase (Th) null] mouse fetuses die at embryonic day (E)13.5–14.5, they resemble wild-type (wt) fetuses exposed to hypoxia. They exhibit bradycardia (28% reduction in heart rate), thin ventricular myocardium (20% reduction in tissue), epicardial detachment, and death with vascular congestion, hemorrhage, and edema. At E12.5, before the appearance of morphological deficits, catecholamine-deficient fetuses are preferentially killed by experimentally induced hypoxia and have lower tissue Po2 levels than wt siblings. By microarray analysis (http://www.ncbi.nlm.nih.gov/geo; accession no. GSE10341), hypoxia-inducible factor-1 target genes are induced to a greater extent in null fetuses than in wt siblings, supporting the notion that mutants experience lower oxygen tension or have an enhanced response to hypoxia. Hypoxia induces a 13-fold increase in plasma norepinephrine levels, which would be expected to increase heart rate, thereby improving oxygen delivery in wt mice. Surprisingly, increasing maternal oxygen (inspired O2 33 or 63%) prevents the effects of catecholamine deficiency, restoring heart rate, myocardial tissue, and survival of Th null fetuses to wt levels. We suggest that norepinephrine mediates fetal survival by maintaining oxygen homeostasis.
- fetal stress
- thin myocardium
catecholamines, particularly norepinephrine (NE) and epinephrine, mediate the “fight or flight” response of the sympathetic nervous system to all stressors. The postnatal actions of catecholamines are well documented, but their role in the fetus is largely unexplored. Hypoxia is thought to be the most physiologically relevant stressor in utero, since the fetus experiences constitutively low oxygen tension and undergoes frequent bouts of increased hypoxia (18). Low oxygen tension is necessary for proper development (13). However, excessively low oxygen (hypoxia) is a major contributor to prenatal and perinatal death, as well as postnatal morbidity, with behavioral and cardiovascular effects that persist into adulthood (1, 4, 5, 59). Hence, the ability to regulate oxygen homeostasis is critical. The fetus can sense and respond to hypoxic stress, but little work has been done to elucidate these processes in early mammalian development. In this study we examine the role of catecholamines in fetal survival and in attenuating fetal hypoxia.
NE is the only catecholamine essential for fetal survival. Catecholamines are synthesized by the sequential action of four enzymes: tyrosine hydroxylase (TH), which converts tyrosine to dopa (3,4-dihydroxy-l-phenylalanine); dopa decarboxylase (also known as l-aromatic amino acid decarboxylase), a ubiquitous enzyme that decarboxylates dopa to dopamine, dopamine β-hydroxylase (DBH), which hydroxylates dopamine to form NE; and finally, phenylethanolamine N-methyltransferase (PNMT), which converts NE to epinephrine. Deletions of genes required for NE synthesis, [Th (23, 45, 61) and Dbh (53)], are lethal in midgestation [embryonic days (E)12.5–14.5]. Th nulls lack all catecholamines, whereas Dbh nulls lack NE and epinephrine but have dopamine. Using an elegant genetic strategy, Zhou and Palmiter (60) showed that dopamine-deficient fetuses that can make NE are viable. In contrast to Th and Dbh deletions, disruption of Pnmt is not lethal (10). These data demonstrate that NE is essential for fetal survival.
In NE-deficient (Th−/− or Dbh−/−) mice, genetic replacement of NE or agonists that activate NE receptors (adrenergic receptors) rescue survival (6, 41, 53, 60, 61). Treatment of the dam with a non-subtype-specific β-adrenergic receptor (β-AR) agonist rescues 90% of Th null fetuses to birth, whereas a β1-specific agonist rescues 74%, suggesting that survival is largely mediated by β1-ARs (6, 41). β1-ARs are detected only in the heart at E12.5, suggesting that the heart may be the primary target of NE in the early fetus (6).
Fetal hypoxia has been linked to intrauterine growth restriction, preterm delivery, and postnatal complications such as sudden infant death syndrome, behavioral disorders, and adult cardiovascular disease (1, 4, 5, 59). Because low oxygen tension is also required for normal placentation, vasculogenesis/angiogenesis, hematopoiesis, and chondrogenesis, oxygen regulation is critical to normal prenatal development (46, 50, 55).
In utero, normal human and ovine fetuses experience constitutively low oxygen tension with a maximum vascular Po2 of ∼25–30 mmHg compared with an arterial Po2 of 90–100 mmHg in adults (16, 48). Before placentation, oxygenation is even lower. At the time of blastocyst formation, the intrauterine oxygen tension in rabbits is 3.5% O2 (11). Moreover, after placentation, mammals experience transient periods of increased hypoxia throughout gestation as a result of spontaneous uterine contractions. On average, these contractions occur once an hour, last for 6–8 min, and can reduce fetal vascular Po2 by 10–25% (18). Therefore, both constitutive and transient low oxygen availability are normal occurrences in placental mammals.
Acutely, hypoxia causes bradycardia (i.e., slowed heart rate) and release of NE in late-term mammalian fetuses (14, 20, 41). If β-ARs, especially β1-ARs, are blocked during hypoxia in late-term fetal sheep, heart rate is dramatically slowed and death accompanied by cardiovascular collapse ensues (8, 14, 36). In the early fetus, hypoxia slows heart rate to a greater extent in catecholamine-deficient fetuses than in wild-type (wt) siblings. This reduction is completely reversed by β-AR agonists (6, 41), consistent with the view that released NE increases heart rate during hypoxia in wt animals. Because cardiac output in the fetus is primarily dependent on heart rate (reviewed in Ref. 22), NE deficiency would be expected to compromise fetal oxygenation. These data suggest that NE helps maintain fetal Po2 by reversing hypoxia-induced bradycardia.
If NE attenuates hypoxia, NE-deficient fetuses would be expected to be more vulnerable to hypoxia than wt siblings. Using Th−/− fetuses as a model of NE deficiency, we have shown that null fetuses share many characteristics with hypoxic wt fetuses (44). Th−/− fetuses are less tolerant to hypoxia induced by reducing maternal oxygenation [inspired O2 (FiO2) 8%], and demonstrate an enhanced transcriptional response during hypoxia. Most surprisingly, increasing maternal oxygenation rescues Th−/− survival to birth and reverses hypoxic phenotypes associated with catecholamine deficiency. The current work describes the response of Th−/− fetuses to hypoxia and proposes that NE mediates survival by increasing hypoxia tolerance.
MATERIAL AND METHODS
Fetuses were obtained by mating albino CD-1 Th+/− mice in which the Th null allele had been backcrossed for seven or eight generations onto the CD-1 background (45). Albino mice were used to avoid the small amount of catecholamines synthesized by tyrosinase present in pigmented Th−/− fetuses (45). Day E0.5 was determined by the presence of a vaginal plug. Animals were genotyped using a PCR protocol that simultaneously identified the Th wt and null alleles. This protocol used a common forward primer in Th exon 6 (5′-GGGTGAGCCAATTCCCCAC-3′), with a reverse primer in Th exon 8 (5′-TGAGTGCATAGGTGAGGAGG-3′) and a reverse primer in the neomycin gene from the null allele (5′-CATAGCCGAATAGCCTCTCC-3′). All experiments were conducted with an approved protocol from the Duke University Institutional Animal Care and Use Committee. At the time of dissection, fetuses were determined to be alive or dead by the presence of heart beat. Survival data were calculated by χ2 statistical analysis using JMP software (SAS, Cary, NC).
In vivo hypoxia and hyperoxia.
Cages of pregnant dams were placed into an acrylic environmental chamber (Plas Labs, Lansing, MI) or a plastic bag with premixed gas flowing at a rate of 0.2 l/min. Oxygen in the hypoxic environment was measured with an oxygen analyzer (Engineered Systems & Designs, Newark, DE). Hypoxia was induced with 8% O2-balance N2. For experiments with continuous hypoxia, hypoxia was initiated such that the time in hypoxia was complete at E12.5. For discontinuous hypoxia experiments, dams were placed in 8% O2 for 1 or 3 h and returned to normoxia at E12.0. Fetuses were dissected 12 h later at E12.5. For survival and cardiac morphology experiments, hyperoxia was induced from E11.5 to either E14.5 or birth (E19/P0). Oxygen content in hyperoxia experiments was maintained at either 33 or 63% by adjusting inflowing oxygen (100%). For heart rate measurements, premixed 95% O2-balance N2 was administered acutely to the anesthetized dam at 0.2 l/min by using a small breathing cone placed over the dam's head.
Ventricular tissue area.
Ventricular tissue area at E14.5 was determined as described (44). Briefly, fetuses were prepared as described for 2-(2-nitro-1H-imidazol-1-yl)-N-(2,2,3,3,3-pentafluoropropyl)acetamide (EF5) measurements (see EF5 immunohistochemistry). Sections (10 μm) were stained with hematoxylin and eosin according to standard protocol and digitally photographed. With the use of NIH ImageJ software (http://rsb.info.nih.gov/ij/), a region of interest (ROI) was drawn around both ventricles on every third slide (images were 90 μm apart). The image threshold was set such that the area representing ventricular compact myocardium, trabeculae, interventricular septum, and atrioventricular cushion was summed throughout the length of the ventricles. Staining representing blood in the ventricles was digitally removed before image processing. Tissue area was determined for one Th+/+ and one Th−/− fetus from each litter. Values are expressed as mean pixel area or mean fraction of Th+/+ ventricular area SE.
Relative tissue Po2 was determined using EF5, which forms protein adducts in hypoxic cells (24, 30). EF5 (a gift from Dr. Cameron Koch, University of Pennsylvania) was dissolved in sterile lactated Ringer solution at 10 mM and injected at two subcutaneous dorsal thoracic sites for a total dose of 30 mg/kg (MW = 302.5 g/mole). At E12.5, the dam was then placed in either ambient or 8% O2 for 3 h, after which time the dam was anesthetized with isoflurane (Aerane; Baxter, Deerfield, IL) and euthanized by cervical dislocation. The fetuses were dissected, genotyped, and immersion-fixed in 10% buffered neutral formalin (VWR, West Chester, PA) overnight, dehydrated through graded ethanol, and embedded in paraffin by the Duke Immunohistology Laboratory (Durham, NC). Tissue was cut in 10-μm sections onto Superfrost Plus slides (Fisher Scientific, Hampton, NH), deparaffinized in xylene (Mallinckrodt Chemicals, Phillipsburg, NJ), and rehydrated through an ethanol series to deionized H2O. Nonspecific binding sites were blocked for 3 h at room temperature with 2% dry milk, 1.5% bovine serum albumin, 5% goat serum, and 0.3% Tween 30 in phosphate-buffered saline (PBS). EF5 protein adducts were detected using monoclonal Cy3-conjugated primary antibody Elk3-51 at 75 μg/ml (gift of Cameron Koch). Nonspecific antibody staining was determined using antibody preabsorbed with 0.5 mM EF5. Primary or preabsorbed antibody was applied to sections overnight at 4°C in the dark. Slides were then washed in PBS, dehydrated through ethanol to xylene, and coverslipped using Krystalon mounting medium (EMD Chemicals, Darmstadt, Germany). All tissue sections were photographed (Axioskop Zeiss, Oberkochen, Germany) using a black and white charge-coupled device camera (RTE CCD-1317-E11; Princeton Instruments, Trenton, NJ), and image acquisition was by IPLab 3.5 software (Scanalytics, Rockville, MD) at 500-ms exposure with fixed white and black points. The 8-bit image was imported into the NIH ImageJ program, which was used to draw a ROI around the dorsal root ganglia and interventricular septum from comparable sections. The image threshold was set to include tissue but not extracellular space. The average pixel intensity in the ROI from sections exposed to preabsorbed antibody was subtracted from primary antibody staining in adjacent sections to remove the effect of nonspecific binding and background fluorescence. Three to six slides were stained per fetus, and the mean pixel intensity in the ROI was averaged for each tissue area for each fetus, using one Th+/+ and one Th−/− fetus per litter. Data are expressed as either mean pixel intensity or mean ratio of hypoxic to normoxic binding SE.
Western blot analysis.
Protein samples were obtained from flash-frozen Th+/+ and Th−/− E12.5 fetuses from dams exposed to either normoxia or hypoxia (24 h of 8% O2 from E11.5 to E12.5). Extracts were homogenized in 8 M urea buffer as described previously (25). Protein was estimated using Bradford reagent (Bio-Rad, Hercules, CA) with γ-globulin as the standard. Protein samples (50 μg/lane) were electrophoresed in duplicate through 3.5–17% exponential gradient Fairbanks gels and transferred to a nitrocellulose membrane. The blot was cut in half and probed with either an anti-hypoxia-inducible factor-1α (HIF-1α) rabbit polyclonal antibody (1:1,000 dilution; catalog no. 100-479; Novus Biologicals, Littleton, CO) or a β-actin monoclonal antibody (1:10,000 dilution; catalog no. A5441; Sigma, St. Louis, MO). Horseradish peroxidase-conjugated goat anti-rabbit or goat anti-mouse IgG were used as secondary antibodies (catalog nos. 1858415 and 1858413, respectively; Pierce), and the signal was developed using an ECL-Plus kit (Amersham Biosciences, Piscataway, NJ). Protein samples were also prepared from Hepa 1-6 cells grown under ambient (21%) or low oxygen (1%) for 4 h. Hypoxia induced the same HIF-1α band in cultured cells as in the hypoxic fetuses (see Supplemental Fig. S1). (Supplemental data for this article is available online at the American Journal of Physiology-Regulatory, Integrative and Comparative Physiology website.)
Heart rate measurement.
At E13.5, pregnant females were anesthetized with a subcutaneous injection of 2 μl/g 50% ketamine/25% xylazine in normal saline. The uterus was exposed through an incision in the abdominal wall. Body temperature was maintained with a heating pad and frequent application of prewarmed PBS to the exposed uterus. With the uterus intact, echocardiography was performed with a HDI 5000 echocardiograph (Philips, Andover, MA) in pulse mode, fitted with a 10.5-MHz pediatric transducer probe (3 × 1 cm) that was wrapped with tape to increase the depth of gel between the probe and tissue (∼1–2 cm). For each fetus, heart rate was measured using at least three consecutive RR intervals (from the beginning of ventricular depolarization of 1 beat to the ventricular depolarization of the next beat), as represented by images of ventricular wall movement. Uteri were then removed and individual fetuses genotyped. These data are expressed as means ± SE for each genotype.
In vitro hypoxia and blood collection.
E12.5 wt fetuses were freed from yolk sac membranes according to our fetal culture protocol (41). Fetuses were maintained for 15 min in 37°C W3 buffer [120 mM NaCl, 5 mM KCl, 1 mM NaH2PO4, 20 mM HEPES, and 20 mM glucose (pH 7.3)] equilibrated either with 95% O2-balance N2 bubbling into the buffer or with atmospheric O2. In culture, tissue Po2 is determined by diffusion such that 95% O2 in the buffer maintains a tissue Po2 of 29.5 mmHg (6), which is similar to in vivo Po2 under normoxia. Fetal culture equilibrated with ambient oxygen (21%) represents a hypoxic in vitro condition. A preincubation period of 15 min allowed for the reuptake of catecholamines released during dissection [circulating NE half-time = 1–4 min (2, 20)]. After 15 min in culture, fetuses were placed on a warmed platform and blotted dry, and blood was collected (1–10 μl/fetus) through a microcapillary tube inserted into the thorax. The blood was expelled into 200 μl of cold PBS containing 3.5 mM EDTA, 10 units of heparin, and 1 pmol of dihydroxybenzylamine (DHBA) as an internal standard and maintained on ice until all blood was collected. Litters were divided equally between oxygen conditions, and blood from one-half the litter (usually 3–5 fetuses) was pooled to represent one sample. Samples were centrifuged to remove cells (5,000 g, 10 min, 4°C), and the plasma was centrifuged again and frozen at −80°C until catecholamine determination by high-performance liquid chromatography (HPLC). Pelleted cells were lysed by resuspension in 2 ml of H2O and centrifuged to remove debris. The volume of blood was determined by absorbance at A414 using a standard curve prepared from known volumes of fetal blood.
Plasma catecholamine assay.
Frozen plasma was thawed on ice. Two-hundred microliters of cold 2 M Tris (pH 8.6) and 5 mg of activated alumina (ICN Alumina N; ICN Biomedicals, Costa Mesa, CA) were added to the 200 μl of plasma. The sample was vortexed and then centrifuged for 1 min at 1,000 rpm at 4°C. The supernatant was discarded, and alumina was transferred to a 0.22-μm filter (Ultrafree MC filter; Millipore, Billerica, MA) and washed twice with 100 μl of distilled H2O. Catecholamines were eluted with 50 μl of cold 0.2 N perchloric acid, with 10 min allowed for elution before the eluant was centrifuged through a new 0.22-μm Ultrafree filter at 5,000 rpm for 10 min at 4°C. Catecholamines were separated by HPLC on a 3 × 150-mm (3 μm) C-18 column (catalog no. MD-150; ESA, Chelmsford, MA) at a flow rate of 0.5 ml/min at room temperature with MD-TM mobile phase (ESA) and quantified by electrochemical detection (EC) using an ESA model 5020 guard cell set to +350 mV, followed by an ESA model 5014B analytical cell (electrode 1 at −150 mV and electrode 2 at +220 mV). The data are expressed as mean NE concentration SE corrected for DHBA recovery.
Pregnant mice were placed in 8% O2 for 6 h (beginning at E12.25) and euthanized at E12.5. Each fetus was homogenized (model 10/35; Brinkmann, Newbury, NY) in 1 ml of RNA STAT-60 reagent (Tel-Test, Friendswood, TX). RNA was extracted with 0.2 ml of chloroform, precipitated with 0.5 ml of isopropanol, washed with ethanol, and resuspended in 200 μl of RNase-free water. Total RNA from each of three fetuses of the same genotype and oxygen condition from different litters was pooled. RNA was further purified using the RNeasy Mini Kit (Qiagen, Valencia, CA). One-hundred micrograms of total RNA from the pooled sample were used for both the microarray hybridizations and quantitative RT-PCR (qRT-PCR). For microarray hybridization, three pooled samples representing nine individual fetuses for each condition were used. RNA quality was determined by gel electrophoresis with an Agilent RNA quality analysis procedure (Duke Microarray Facility, Durham, NC). Ten micrograms of each sample were used to prepare Cy3-labeled cDNA using an oligo(dT) primer. This was mixed with cDNA prepared from 10 μg of Cy5-labeled reference RNA (Stratagene, La Jolla, CA). The mixed Cy-labeled cDNA probe was hybridized to an oligo-spotted slide (Operon Mouse Oligo Set v. 3 containing 31,867 oligonucleotides representing ∼29,000 unique genes) at 42°C for 16–20 h as described (http://mgm.duke.edu/genome/dna_micro/core/spotted.htm).
Microarray data were analyzed using GeneSpring software 7.3.1 (Agilent Technologies, Palo Alto, CA). Raw data were filtered on expression level using the average base/proportional value on a per spot per chip basis, requiring values to appear in 6 of 12 samples (3 each of wt normoxic, wt hypoxic, null normoxic, and null hypoxic), resulting in interrogation of 19,308 oligonucleotides, and normalized to the cohybridized internal RNA standard. Statistical analysis using ANOVA with P < 0.05, without correction for multiple tests, was performed for comparisons between each condition. Fold ratios were derived by comparing the normalized data between conditions. Microarray data are available at the Geo website (http://www.ncbi.nlm.nih.gov/geo/; accession no. GSE10341).
qRT-PCR analysis was performed using RNA from the same samples used for microarray hybridization. RNA was treated with RQ-1 RNase-free DNase (Promega, Madison, WI) to remove DNA contamination and reverse transcribed using the Ambion Retroscript kit (Ambion, Austin, TX) with 1 μg of total RNA in a 20-μl reaction volume. PCR reactions were run in triplicate with a 20-μl reaction volume per well. Each reaction contained 1 μl of a 1:2-, 1:10-, or 1:50-fold dilution of the RT product, 10 μl of SYBR green PCR master mix (Applied Biosystems, Foster City, CA), 0.25 μl each of 20 μM sense and antisense primers, and 8.5 μl of molecular biology grade water (GIBCO 1614; Invitrogen, Carlsbad, CA). With the use of an Applied Biosystems 7300 Real-Time PCR system, the reaction was heated to 95°C for 10 min and then subjected to 40 cycles of 95°C for 15 s, annealing for 30 s at a primer-dependent temperature (Table 1), 72°C for 30 s, followed by a dissociation stage consisting of 95°C for 15 s, 60°C for 30 s, and 95°C for 15 s. Data were collected during the 72°C step of each cycle and analyzed using relative quantity determined by a standard curve of 3 serial dilutions of a standard template made from all 12 samples. Expression was normalized to that of Rpl6, a ribosomal gene that did not change in the microarray or qRT-PCR analyses and was abundantly expressed at the same level in all samples. The PCR product size was confirmed by agarose gel electrophoresis. Data are expressed relative to the wt 21% condition (ambient oxygen).
All data are means ± SE. Data were analyzed using one-way ANOVA, except for survival data, which were calculated by χ2 statistical analysis using JMP software (SAS, Cary, NC).
Th−/− fetuses resemble hypoxic wt fetuses.
Because low levels of catecholamines are synthesized in pigmented Th−/− fetuses via tyrosinase (45), we bred the Th null allele onto an albino CD-1 background. Survival of CD-1 Th−/− fetuses was similar to that of Th nulls on other genetic backgrounds (61). Under ambient oxygen, CD-1 Th−/− fetuses began to die between E12.5 and E13.5 (Fig. 1A). By E14.5, 19% of Th−/− fetuses were alive and only 8.4% survived to term (E19.5/P0).
E14.5 Th−/− fetuses died with signs of congestive heart failure, with loss of yolk sac circulation (Fig. 1, B–D; arrows in B point to yolk sac vessels in wt fetus), vascular congestion (arrowhead in Fig. 1C), edema, and hemorrhage (blood in the amniotic fluid in Fig. 1D). Surviving E14.5 Th null fetuses had a 20% reduction in ventricular myocardial mass (Fig. 1G) and thinner myocardial walls (red lines in Fig. 1, E and F), as well as loosely compacted and thin interventricular septa (yellow arrowheads in Fig. 1, E and F) and mild epicardial detachment (black arrow in Fig. 1F). Th−/− fetuses also exhibited bradycardia at E13.5. When measured by ultrasound in utero, Th nulls showed a 28% lower heart rate than Th+/+ fetuses under ambient oxygen (Fig. 1H). Myocardial thinning and bradycardia also have been reported in other NE-deficient fetuses (23, 53, 61).
The phenotype exhibited by Th−/− fetuses resembles that of wt fetuses exposed to 18–24 h of maternal hypoxia. Hypoxic wt fetuses are bradycardic (6, 41) and show signs of congestive heart failure, myocardial thinning, and epicardial detachment (44).
Th−/− fetuses experience greater reduction in tissue oxygenation than wt siblings after induced hypoxia.
Because Th null fetuses resembled hypoxic wt fetuses, we examined differences in tissue Po2 between wt and Th−/− fetuses using the hypoxia-sensitive dye EF5. We initially chose E14.0 fetuses, because we expected the greatest difference between wt and Th null animals at this age, just before the majority of Th nulls die. However, at E14.0, we found only low levels of EF5 binding in Th+/+ and Th−/− fetuses, and there was no statistical difference between genotypes (data not shown). EF5 and related nitroimidazole compounds bind to cellular proteins when tissue Po2 falls below 10 mmHg (42). Therefore, mild to moderate hypoxia between 10 and 25 mmHg (normal fetal Po2) would not be detected by EF5 binding.
To explore the possibility that a role for catecholamines is unmasked during hypoxia, we examined EF5 binding under conditions of reduced oxygen supply (Fig. 2). To decrease fetal Po2, pregnant dams carrying E12.5 fetuses were housed in 8% O2 for 3 h. At E12.5, wt mice exhibit a robust response to hypoxia (44) and Th nulls are fully viable (Fig. 1A) and show no morphological deficits. Hence, at E12.5, hypoxia-induced changes in survival, morphology, or physiology can be attributed to lack of catecholamines rather than to preexisting morphological deficits. As expected, EF5 immunofluorescence increased with fetal hypoxia (Fig. 2A). Although many tissues showed increased dye binding after hypoxia, we quantified signal intensity in the dorsal root ganglia and interventricular septum of the heart, because these tissues are clearly delineated at E12.5 and are relatively homogenous at the cellular level. Under ambient oxygen, we observed no difference in dye binding between genotypes, suggesting that at E12.5, like at E14.0, both wt and null fetuses have tissue Po2 above 10 mmHg (Fig. 2B). Both wt and null fetuses exhibited increased dye binding after 3 h of hypoxia (Fig. 2B). Hypoxic Th nulls had 7.7- and 19.4-fold increases in dye binding (Fig. 2C) in the interventricular septum and dorsal root ganglia, respectively, compared with sibling wt fetuses that showed 4.6- and 5.5-fold increases, respectively. The fold increase in the dorsal root ganglia reached statistical significance, whereas that in the interventricular septum did not. This finding suggests that catecholamines help attenuate the degree of hypoxia experienced by the fetus, at least in some tissues.
Null fetuses have increased HIF-1 accumulation.
Hypoxia results in stabilization of the transcription factor HIF-1. To determine whether the genotype affected HIF-1α levels, we performed Western blot analysis on proteins extracted from E12.5 fetuses from normoxic and hypoxic litters (8% O2 for 24 h) (Fig. 3). In normoxia, wt and null fetuses had similar levels of HIF-1α. Hypoxia increased HIF-1α levels in both genotypes, but the increase was greater in null than in wt fetuses. These results indicate that E12.5 fetuses increase expression of HIF-1α protein with hypoxia and that Th−/− fetuses have a more robust response than wt littermates.
Th−/− fetuses are more sensitive to episodes of hypoxia than wt siblings.
If Th−/− fetuses experience greater hypoxia than wt siblings, they might be less able to survive hypoxia in utero. To assess hypoxia tolerance, we exposed pregnant dams to hypoxia for 18 h, ending at E12.5. Hypoxia preferentially killed null fetuses in a dose-dependent manner (Fig. 4A). Thirty percent of Th−/− fetuses survived in 10% O2, and only 5% survived in 8% O2, compared with ∼92 and 55% of Th+/+ fetuses, respectively (Fig. 4A). At E13.5, no nulls survived 8% O2 (data not shown), suggesting that there is increased vulnerability as the fetus grows larger. Eight percent maternal O2 was chosen for further experiments because it gave the greatest difference between null and wt fetuses. Th nulls died sooner than wt siblings during hypoxia (Fig. 4B). When dams were housed in 8% O2 for varying lengths of time, Th−/− fetuses began dying after 6 h, and only 8% survived for 12 h. In contrast, the majority of Th+/+ and Th+/− fetuses were alive even after 18 h of hypoxia.
To determine whether hypoxic death requires continuous hypoxia or if a short period of hypoxia could initiate events that lead to death, we placed pregnant females in 8% O2 for 1 or 3 h at E12.0, returned them to 21% O2 for 12 h, and assessed survival at E12.5. There was no fetal death with 1 h of hypoxia followed by 12 h of normoxia (Fig. 4C). However, after 3 h of hypoxia followed by 12 h of normoxia, 54% of Th−/− fetuses died, whereas only 6.4 and 1.5% of wt and heterozygous mice died, respectively (Fig. 4C). When fetuses were examined immediately following 3 h of hypoxia at E12.0, no wt fetuses had died and only 5% (1/19) of null fetuses were dead (not statistically different from wt fetuses), suggesting that death required some period of time following the 3-h hypoxic episode to occur. Dead fetuses at E12.5 looked like previously reported wt fetuses that died after continuous hypoxia (44), with loss of yolk sac circulation, vascular congestion, edema, and hemorrhage into the yolk sac (Fig. 4D, compare live Th+/− fetus at left with dead Th−/− sibling at right). These experiments indicate that catecholamines enhance hypoxia tolerance at E12.5 and that a period as short as 3 h selectively kills Th nulls, but lethality requires time to ensue.
Hypoxia induces NE release.
If NE increases hypoxia tolerance, its levels might be regulated by oxygen. To determine whether hypoxia alters circulating NE levels, we measured plasma NE by HPLC in the blood of E12.5 wt fetuses cultured for 15 min under normoxia or hypoxia. Hypoxia increased NE levels 13-fold from 0.2 to 2.8 nM (Fig. 5). We conclude that hypoxia induces a large acute release of NE from sympathoadrenal precursor cells at midgestation.
Th−/− fetuses superinduce HIF-1 target genes during hypoxia.
To assess whether the increased sensitivity to hypoxia experienced by null fetuses results in transcriptional differences, we performed expression profiling on E12.5 wt and null fetuses in normoxia and after 6 h of hypoxia (see Supplemental Table S1).
Hypoxia induced large transcriptional changes in both wt and null fetuses. Approximately 21% (4,140/19,308) of transcripts changed with hypoxia in wt fetuses, and 26% (5,097/19,308) changed in null fetuses. Notably, 3.8% of transcripts (743) were changed by twofold or more in wt animals (listed in Supplemental Table S2) compared with 3.2% (612) in null fetuses. Changes were typical of those induced by hypoxia in other tissues and in culture (43) and are listed by functional category (glycolysis, transporters, stress induced, growth retardation, vascularization, chondrocyte formation, and oxygen sensing) in Supplemental Table S3. Changes in five genes from different categories were confirmed by qRT-PCR (Table 2). Of the 743 genes that changed by twofold or more in wt fetuses, 21% (155) did not change in nulls with hypoxia (Supplemental Table S4).
To determine how catecholamines affect known hypoxia-responsive genes, we interrogated HIF-1 targets. Of the ∼70 HIF-1 target genes compiled by Wenger et al. (56), 38 were expressed in E12.5 mouse fetuses and are listed in Table 2. Of the expressed HIF-1 targets, 39% (15 genes) and 45% (17 genes) were significantly induced in wt fetuses (Th+/+) and null (Th−/−) fetuses, respectively, after 6 h of hypoxia. These include vascular endothelial growth factor (VEGFα), glucose transporters 1 and 3 (GLUT-1, GLUT-3), glycolytic enzymes (hexokinase-2 and 6-phosphofructo-2-kinase), a prolyl hydroxylase that regulates HIF-1α stability (Egln3), IGFBP1, which induces growth restriction, and Bnip3, a proapoptotic gene induced by hypoxia in the heart. Most interestingly, the degree of induction of HIF targets was significantly greater in null fetuses (3.8-fold, P = 0.01) than in wt fetuses (2.5 times). This superinduction supports the notion that nulls experience greater tissue hypoxia than wt siblings under conditions of maternal hypoxia.
In normoxia, Th+/+ and Th−/− fetuses did not significantly differ in their gene expression patterns; that is, the number of genes that differed was less than that predicted by chance. Twenty-six transcripts differed by twofold or more (listed in Supplemental Table S5). However, three (Tcf7, Timp1, and Cspg3) that were among those most different as indicated by microarray analysis were not confirmed to be different by qRT-PCR analysis, supporting the view that there are very few transcriptional differences between the genotypes in normoxia. This suggests that catecholamines have little influence on gene expression under ambient conditions.
Increased maternal oxygen rescues the Th−/− phenotype.
To determine whether the lethality, thin myocardium, and bradycardia exhibited by null fetuses might be prevented by increasing fetal oxygenation, we increased maternal FiO2, which modestly increases fetal Po2 in large mammals (31) and raises tissue oxygenation in E12.5 mice (44). Maternal hyperoxia (33 or 63% O2) from E11.5 to term increased survival of Th−/− fetuses ninefold, rescuing almost 80% of nulls to birth, without affecting the viability of Th+/+ or Th+/− animals (Fig. 6A). Hyperoxia from E11.5 to E14.5 prevented the myocardial thinning that occurred under normoxia such that hyperoxic Th−/− hearts showed no apparent morphological deficits (Fig. 6, B and C). Acutely, hyperoxia restored Th−/− heart rate to that of wt siblings in utero at E14.0 (Fig. 6D). It increased the heart rate of Th−/− fetuses from 115 to 176 beats/min, showing that null fetuses are fully capable of matching the rate of wt siblings under hyperoxia, although they lack NE. Notably, hyperoxia did not cause leakage of maternal NE into the fetus as measured by HPLC (data not shown), suggesting that mutant fetuses rescued by maternal hyperoxia were still NE deficient.
Tyrosine hydroxylase deficiency resembles fetal hypoxia.
Our results provide a striking parallel between Th−/− fetuses and hypoxic wt fetuses. Both exhibit bradycardia, thin myocardium, epicardial detachment, and midgestational death with signs of heart failure. The phenotype seen in normoxic Th−/− fetuses is less extreme than that of wt fetuses subjected to hypoxia (8% O2 for 18–24 h) (44), which suggests that if Th−/− fetuses are hypoxic due to their catecholamine deficiency, they probably experience mild chronic hypoxia.
Th−/− as a thin myocardium syndrome mutant.
The Th null resembles other mutants that exhibit thin myocardium syndrome (TMS) (17). All TMS mutants resemble wt fetuses exposed to hypoxia, have myocardial hypoplasia, and die at midgestation with signs of congestive heart failure. These mutants primarily fall into four categories: 1) mutations in erythropoiesis such as erythropoietin and EpoR nulls (51, 57); 2) mutations affecting the epicardium or signals emanating from it, including α4 integrin (58), vascular cell adhesion molecule-1 (VCAM-1) (26), and the retinoic acid receptors [RXRα (9, 49), RARα/RXRβ (21)]; 3) mutations in myocardial transcription factors such as Fog-2 (52) and N-myc (34); and 4) mutations affecting norepinephrine availability, such as Th (23, 61), Dbh (53), Phox2b (38), ErbB3 (27, 32), HIF-2α (also known as EPAS-1) (39, 54), and Pax3 (also known as Splotched) (12).
Because myocardial hypoplasia is the hallmark of TMS, cardiac insufficiency is likely to be common to all TMS mutants. Ventricular ejection fraction has been measured in three mutants (Pax3, RXR, and β-ARK-1) and was reduced by >62% (7, 9, 17). Reduced ejection fraction lowers cardiac output (the product of heart rate and ejection fraction) and would be expected to result in congestion and edema due to venous leakage. Reduced cardiac output would also lead to reduced perfusion and hypoxia. Hence, all TMS mutants, including Th−/−, could experience a common pathway to death involving hypoxia.
Hypoxia reduces heart rate and increases NE release.
Acutely, hypoxia causes bradycardia and NE release. Bradycardia may be mediated by inhibition of oxygen-sensitive L-type Ca2+ channels (29). In fetuses with a mature autonomic nervous system, such as near-term sheep, hypoxia-induced bradycardia also has a parasympathetic component (15).
Hypoxia induces NE release by depolarizing sympathoadrenal cells via closure of oxygen-responsive K+ channels (33, 37). Released NE, in turn, acutely acts to reverse bradycardia by activating cardiac β1-AR (8, 19), increasing the rate and strength of contraction. This feedback loop is well documented in fetal sheep, where hypoxia induces rapid bradycardia that rebounds when NE is released. Rebound is abrogated by sympathectomy (47) and fails to occur when β-AR are blocked (19).
Our data show that hypoxia induces NE release in the early mouse fetus, increasing circulating NE concentrations by more than 10-fold, typical of hypoxia-induced changes in serum NE concentration in other systems (28, 35, 40, 47). These results suggest that the immature mouse fetus employs the same feedback loop as the late-term fetus. Overall, these studies demonstrate that hypoxic stress induces bradycardia that is reversed by hypoxia-induced NE release in immature fetuses.
In agreement with ex utero data (23, 60), catecholamine-deficient fetuses are constitutively bradycardic (Fig. 1) in utero, suggesting that NE also may influence the nonstressed heart rate. Since fetal cardiac output is largely determined by heart rate (reviewed in Ref. 22), constitutive bradycardia would reduce cardiac output and induce chronic hypoxia.
Increased oxygenation rescues catecholamine-deficient fetuses.
Th−/− lethality can be accelerated by hypoxia (Fig. 4) and prevented by hyperoxia (Fig. 6). Hyperoxia also prevented the bradycardia and hypoplastic myocardium of null animals. Because null fetuses rescued by increased oxygen are still NE deficient, NE is likely to play a physiological role rather than act as a trophic molecule or morphogen. Our data suggest that NE may be needed primarily to maintain oxygen homeostasis and that without hypoxic stress, NE is dispensable.
How might a modest increase in oxygen (from 21 to 33%) rescue Th nulls? Hyperoxia would not significantly alter maternal hemoglobin saturation, which is already near 100% under ambient oxygen, where maternal Po2 is 90–100 mmHg, but would increase dissolved oxygen in the maternal circulation in direct relation to FiO2 levels. In contrast, fetal Po2 is normally 25–30 mmHg, where hemoglobin saturation is ∼50%. This level of fetal saturation lies on the steepest part of the hemoglobin-oxygen binding curve such that a slight increase in oxygen dissolved in maternal blood could result in a significant increase in fetal hemoglobin saturation. This result has been observed in human fetuses at term, where maternal FiO2 of 27% (which is only 6% above ambient) increases fetal hemoglobin saturation by 7.8% (31).
We attribute the fetal effects of maternal hyperoxia to events occurring in the fetus, independent of alterations in maternal circulating factors that may change with oxygen status. However, it is possible that increased maternal oxygen could circumvent the requirement for NE through a redundant mechanism that does not involve increasing fetal Po2.
NE controls gene expression during hypoxia.
Our data provide the first catalog of early fetal gene expression in response to catecholamine availability and hypoxia. The early fetus exhibits a robust transcriptional response to hypoxia such that 20–25% (4,000–5,000 transcripts) of the interrogated transcripts are altered by 6 h of maternal hypoxia (FiO2 = 8%) (Supplemental Tables S1–S3). About 3% of transcripts change by twofold or more. Since we have only interrogated a single time point, our data likely underestimate the total number of hypoxia-induced changes. However, even at a single time point, our data suggest that the fetus senses and responds to hypoxia with significant transcriptional changes.
Microarray analysis revealed 155 candidate genes for hypoxia-induced NE regulation (Supplemental Table S4) that were induced by hypoxia in wt but not null animals. These included embryonic (Hba-x) and adult (Hba-1) α-globin genes, which increased >3.5-fold in Th+/+ fetuses during hypoxia but did not change in Th−/− fetuses. Hemoglobin synthesis is induced by protein kinase A activation (3), an event downstream of NE activation of β-ARs, which suggests a mechanism by which NE can regulate α-globin transcription. These data suggest that NE may play a role in attenuating hypoxia by mediating transcription of certain adaptive genes such as hemoglobin.
Hypoxia induced greater expression of HIF-1 target genes in null fetuses compared with wt animals (Table 2). Four genes (Bnip3, Glut-1, Pfkfb3, and Vegf) were selected for qRT-PCR analysis to confirm the validity of the microarray data, and all four showed greater expression in nulls. The superinduction of known HIF-1 targets indirectly suggests that null fetuses may have an even lower oxygen tension or have a greater transcriptional response to hypoxia than wt sibs. Transcription of HIF-1α itself was not induced by hypoxia in our experiments, suggesting that in the early fetus, hypoxia stabilizes HIF-1α protein but does not induce its transcription, as is the case in most cells (43).
Perspectives and Significance
We propose that NE is acutely required to maintain heart rate and reverse hypoxia-induced bradycardia. To serve this function, NE is released during hypoxia (Fig. 5) and acts at β1-AR to increase heart rate and contractility (6, 41). Without NE, constitutive bradycardia ensues (Fig. 1), creating a chronic hypoxic condition, which causes myocardial hypoplasia (Fig. 1) (44). We propose that continued bradycardia and myocardial thinning further reduce cardiac output and worsen hypoxia, leading to death due to insufficient tissue perfusion. The present work follows from our previous results on the effects of hypoxia in wt mice (44). Although we have not directly demonstrated basal hypoxia in Th−/− fetuses, the facts that normoxic null mice phenocopy hypoxic wt mice and that hyperoxia prevents deficits associated with lack of catecholamines (bradycardia, lethality, and ventricular hypoplasia) provide insight into the essential role of catecholamines during fetal development.
This work was funded by National Institute of Neurological Disorders and Stroke Grant NS-22675 (to D. M. Chikaraishi).
We thank Marybeth Groelle, Kevin Fang, Miwako Nishio, and Kedar Kirtane for technical assistance.
Present address of H.-.G Kim: Dept. of Pharmacology, College of Medicine, Dankook Univ., San-29, Anseo-dong, Chenonan, Choong-nam 330-714, Korea.
Present address of R. Chandra: Dept. of Medicine, Duke Univ. Medical Center, Durham, NC.
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 © 2008 the American Physiological Society