- Top of page
- Materials and methods
Objective To determine whether activin A concentrations are altered in chronic fetal hypoxemia and intrauterine fetal growth restriction (IUGR).
Design In vivo animal experimental model.
Setting Department of Physiology, Monash University.
Population Chronically catherised fetal sheep in late pregnancy.
Methods Chronic fetal hypoxia and IUGR were experimentally induced by single umbilical artery ligation (SUAL) in catheterised fetal sheep. Maternal and fetal blood samples and amniotic fluid (AF) samples were collected during surgery and thereafter on alternate days, until the time of delivery for analyte measurement. Fetal blood gas parameters were measured daily.
Main outcome measures Plasma and AF was used to analyse activin A, prostaglandin E2 (PGE2) and cortisol and fetal blood gas analysis was undertaken in whole blood.
Results SUAL produced asymmetric IUGR and non-acidaemic chronic fetal hypoxia and resulted in preterm labour (129  days). AF activin A concentrations were 10-fold higher in the SUAL group than in controls whereas levels in the fetal and maternal circulations were similar between groups.
Conclusions SUAL-induced IUGR and fetal hypoxaemia increases AF activin A. This may be an important adaptive or protective response to IUGR.
- Top of page
- Materials and methods
Intrauterine fetal growth restriction (IUGR) is a concept defining the fetus that fails to attain its genetically programmed growth potential. IUGR is clinically important because it is associated with an increased risk of perinatal mortality1 and both short2,3 and longer term morbidity.2,4 While IUGR may arise from a variety of aetiologies,5 impaired placentation is the single most common and clinically important cause.6 In such pregnancies complicated by placental IUGR, chronic and progressive fetal hypoxia is thought to be a key end—mechanism underlying the increased perinatal morbidity and mortality. However, timely detection and active management of these ‘at risk’ fetuses would be expected to improve outcome including decreasing mortality,7 if such detection was possible. While Doppler assessment of the umbilical artery flow velocity waveform, an index of fetoplacental vascular resistance,8 and fetal biophysical assessment7 are useful predictors of fetal compromise, they are both time consuming and require highly skilled personnel. A cheaper and more simple approach to fetal assessment would therefore be useful. The measurement of maternal serum levels of activin A may be such an approach.
Activin A9,10 is a dimeric glycoprotein member of the TGF-b superfamily.11 Maternal serum levels of activin A, mainly derived from the placenta,12 progressively increase across human pregnancy13,14 and are further increased in pregnancies complicated by placental insufficiency, such as those in which pre-eclampsia15–17 and/or IUGR.18,19 While increased placental production of activin A has been shown to be the main source of the higher circulating levels in association with pre-eclampsia,17 the mechanism(s) underlying this increase remains ill defined. However, it has been demonstrated, using an ovine model, that acute fetoplacental hypoxia increases fetal and amniotic fluid (AF) activin A levels20–22 and that this increase in activin A is associated with increased PGE2 and cortisol levels.21,22 These observations, in the humans and the sheep, suggest that activin A may be a useful marker of fetoplacental compromise and further lead to the intriguing possibility that activin A may have a role in fetoplacental adaptations to hypoxaemia. Whether the chronic fetal hypoxia that is characteristic of IUGR also results in a significant and sustained elevation in activin remains unexplored and was the purpose of this study.
We have used an experimental fetal ovine model—single umbilical artery ligation (SUAL)—to induce IUGR and to examine the effect of chronic fetal hypoxia and growth restriction, in late-pregnant sheep, on activin A, cortisol and PGE2 levels in the maternal and fetal circulations and AF. SUAL has been used previously to induce chronic fetal hypoxia and fetal growth restriction,23 inducing placental insufficiency by producing infarction of approximately half of the placental bed23 and subsequently impaired placental substrate transfer.24
Materials and methods
- Top of page
- Materials and methods
Surgical and experimental procedures were approved by the Animal Ethics Committee, Department of Physiology, Monash University. Surgery was performed on 11 pregnant ewes carrying singleton fetuses, at 105–110 days of gestational age (term mean [SEM] = 147 [2.5] days25). Prior to the induction of the anaesthesia, 1 g ampicillin (iv) was administered to the ewe. Anaesthesia was induced with 5% thiopental sodium in water (iv) and maintained using 1.5–2% halothane via an endotracheal tube. Catheters were inserted into the maternal carotid artery and jugular vein, fetal femoral artery and jugular vein and the AF cavity. Fetal and maternal arterial blood samples and an AF sample were collected during surgery. Seven of these fetuses were subjected to SUAL, where the fetus was exteriorised, exposing the umbilical cord allowing the umbilical artery to be identified through a small incision in the cord sheath. Two size 0 silk ligatures were tied around the umbilical artery, 2.5–3 cm from the fetal abdominal wall. In the control group (n= 4), the umbilical artery was exposed but not ligated. A three-wire electromyography (EMG) lead was sutured to the external wall of the uterus. The fetal and AF catheters along with the EMG lead were exteriorised via a flank incision in the ewe. The catheters were flushed with heparinised saline (0.9% NaCl; 25 IU heparin/mL).
Fetal and maternal arterial blood samples and AF samples were taken at the time of surgery, after which samples were taken on alternate days for the duration of the experiment. Samples were collected into ethylene diamine tetra-acetic acid (EDTA) tubes containing 10 μl of indomethacin per 1 mL blood volume (0.001 M). All blood and AF samples were centrifuged at 1800 ×g for 5 minutes, at 4°C. Fetal and maternal plasma and AF samples were aliquoted into an equal volume of methoximating reagent for PGE2 analysis. Plasma and AF were aliquoted for activin A and cortisol analysis and stored at −20°C. Fetal arterial plasma samples were also stored for glucose and lactate measurements, which were determined using a YSI Glucose and L-Lactate Analyser (YSI, Yellow Springs Instrument, Yellow Springs, Ohio, USA). Fetal arterial blood samples were taken daily for blood gas analysis [PaO2, PaCO2, % oxygen saturation (SaO2), pH and % total haemoglobin (THb)] using an ABL 510 blood gas analyser (Radiometer, Copenhagen, Denmark).
Fetal heart rate (FHR) and mean arterial pressure (MAP), corrected for AF pressure, was recorded twice a week for 4 hours via the arterial and AF catheters connected to pressure transducers (AD Instruments, Australia) and an online recording system (PowerLab 800, AD Instruments, Australia). Blood and AF samples were taken from the animal at the start and end of the 4 hours. In the SUAL group, EMG activity was also recorded online. A complete postmortem was performed when the ewe was deemed to be in labour, as indicated by significantly increased EMG activity.26 Postmortem was performed on animals in the control group at 132–135 days of gestational age. The first animal from the SUAL group commenced labour at 135 days of gestational age; therefore, control animals were killed at this age to be equivalent and results for individual animals were compared with a growth chart.27 Fetal dimensions as well as fetal body weight (BW), brain weight and organ weights were measured at postmortem.
Plasma samples and AF samples were analysed for activin A using an enzyme-linked immunosorbent assay (ELISA)28 and PGE229,30 and cortisol31 concentrations by radio-immunoassay. The intra- and inter-assay coefficients of variation were 10% and 20% for activin A, 8% and 25% for PGE2 and 8% and 22% for cortisol. Fetal blood gas data, MAP and heart rate (HR) as well as the endocrine profiles, for both the SUAL and control group, were collated into five gestational age groups [surgery–115 days GA, 116–120 days GA, 121–125 days GA, 2–4 days prior to labour or prior to postmortem and day of labour (SUAL) or immediately prior to postmortem (control)].
All data were expressed as mean [SEM]. The effect of SUAL on fetal blood gas data, MAP and HR was analysed using a repeated measures analysis of variance (ANOVA) and this was also used to analyse fetal, maternal and AF activin A, PGE2 and cortisol values. The post hoc test of least significant difference (LSD) was used to identify significant differences within gestational age groups. One-factor ANOVA was used for fetal BW and organ weight analysis.
- Top of page
- Materials and methods
All of the animals in the SUAL group commenced labour prior to postmortem, at a mean [SEM] gestation of 129  days (range: 119–141 days). None of the control animals had commenced labour at the time of postmortem, which was performed at a mean [SEM] gestation of 134  days (range: 132–135 days).
The sexes, gestational ages at delivery and BWs for individual fetuses are shown in 1Table 1. The control animals were appropriately grown as defined by the standard growth chart reproduced by Cloete 1939,27 using the following algorithm: ln (BW) = [9.2167 ln (age)] − [0.4652 (ln (age)2− 25.8778)], whereas the SUAL animals were growth restricted with a mean difference between actual and expected birthweight of 22% (range: 0–45%). The difference between actual, and expected birthweight increased with duration of pregnancy (Table 1). After correcting for differences in gestational age at delivery, there were no differences in fetal crown-rump length, head length, forelimb and hindlimb lengths between the two groups (data not shown). The fetal organ weight to BW ratios for the two groups are shown in 2Table 2. The brain–BW, combined adrenal–BW and liver–BW ratios, was significantly elevated whereas the spleen–BW ratio was significantly reduced in the SUAL animals compared with controls.
Table 1. Fetal gestational age and BW and change in fetal BWs relative to standard growth chart27 at the time of postmortem. Postmortem was performed on SUAL group at the time of labour and control group at 132–135 days of gestational age
|Animal number and (treatment)||Fetal gestational age at postmortem (days)||Sex of fetus||Fetal BW (kg) and (difference in BW relative to expected BW at that gestational age taken from growth charts of Cloete27)|
|1 (SUAL)||135||Female||2.00 (45%)|
|2 (SUAL)||127||Female||1.89 (27%)|
|3 (SUAL)||130||Male||2.20 (21%)|
|4 (SUAL)||119||Female||2.01 (no change)|
|5 (SUAL)||129||Female||2.56 (24%)|
|6 (SUAL)||121||Male||2.00 (11%)|
|7 (SUAL)||141||Female||2.48 (42%)|
|8 (control)||135||Male||3.41 (no change)|
|9 (control)||134||Male||3.87 (no change)|
|10 (control)||132||Female||3.46 (no change)|
|11 (control)||135||Female||3.50 (no change)|
Table 2. Mean [SEM] fetal organ weight (g) to BW (kg) ratios in seven SUAL and four control animals
|SUAL||18.07 [0.42]||12.0 × 10−2 [13.42] × 10−3||4.79 [0.83]||32.98 [1.79]||8.195 [4.40] × 10−1||35.17 [4.24]|
|Control||13.45 [1.21] × 10−1||4.93 × 10−2 [4.10] × 10−3||3.23 [2.82] × 10−1||30.90 [2.20]||7.23 [6.60] × 10−1||21.92 [1.48]|
|Significance||P= 0.0001||P= 0.010||P= 0.056||P= 0.491||P= 0.475||P= 0.049|
In the SUAL group, the overall mean FHR (160  beats/minute) and MAP (41.70 [0.60] mm Hg) were not significantly different to those in the control group (FHR: 166  beats/minute and MAP: 40.40 [0.81] mm Hg).
Fetal arterial blood gas data are shown in 3Table 3. In the control animals, the fetal arterial PaO2, pH and THb did not change for the duration of the experiment (P > 0.05). PaCO2 was significantly higher (P= 0.02) at 121–125 days GA, 2–4 days prior to postmortem and on day of postmortem, compared with at surgery–115 days GA (1Fig. 1b). Fetal arterial SaO2 was significantly lower (P= 0.04) at 2–4 days prior of postmortem and on day of postmortem compared with at surgery–115 days GA.
Table 3. Fetal arterial partial pressure of oxygen (mm Hg), SaO2 (%), partial pressure of carbon dioxide (mm Hg), pH (d) and THb content (%) for SUAL and control groups
| ||Treatment||Surgery–115 days||116–120 days||121–125 days||2–4 days prior to labour/PM||Labour/PM|
|PaO2 (mm Hg)||SUAL||16.73 [1.67]*||16.75 [2.27]*||16.63 [0.47]*||15.44 [0.84]*||9.22 [1.90]*|
| ||Control||21.34 [0.74]||22.83 [0.77]||21.48 [0.87]||20.93 [0.44]||21.50 [1.05]|
|SaO2 (%)||SUAL||50.79 [3.19]*||50.44 [5.19]*||48.45 [3.58]||45.38 [2.63]*||33.95 [8.70]*|
| ||Control||68.21 [1.00]||67.60 [3.21]||62.87 [5.57]||60.25 [3.24]#||58.04 [4.68]#|
|PaCO2 (mm Hg)||SUAL||48.52 [0.97]*||51.33 [0.52]*||48.00 [3.91]||53.74 [1.02]*||54.98 [2.56]*|
| ||Control||45.24 [0.85]||46.57 [1.10]||47.20 [1.36]#||48.09 [0.72]#||48.50 [2.45]#|
|pH||SUAL||7.33 [0.01]||7.34 [0.01]||7.34 [0.02]||7.34 [0.01]||7.26 [0.05]|
| ||Control||7.35 [0.01]||7.35 [0.01]||7.35 [0.01]||7.35 [0.01]||7.31 [0.02]|
|THb (%)||SUAL||9.30 [0.62]||7.92 [0.56]||7.57 [0.44]||8.49 [0.62]||7.98 [0.25]|
| ||Control||8.32 [0.43]||9.12 [0.86]||9.30 [1.19]||9.45 [0.91]||8.50 [0.69]|
Figure 1. Fetal arterial activin A (ng/mL) (a), AF activin A (ng/mL) (b), maternal arterial activin A (ng/mL) (c) for SUAL (▪) and control (□) groups.
*Significant difference to control values (P < 0.05).
Download figure to PowerPoint
In the SUAL group, the fetal arterial PaO2 (Fig. 1a) and %SaO2 (Fig. 1b) were significantly lower at surgery–115 days GA (P= 0.04), 116–120 days GA (P= 0.03), 121–125 days GA (P < 0.01), 2–4 days prior to labour (P= 0.001) and at labour (P= 0.001) for PaO2 and at surgery–115 days GA (P= 0.001), 116–120 days GA (P= 0.02), 2–4 days prior to labour (P= 0.02) and at labour (P= 0.048) for %SaO2, compared with the corresponding values in the control group whereas PaCO2 was significantly higher at surgery–115 days GA (P= 0.04), 116–120 days GA (P= 0.01) and 2–4 days prior to labour (P= 0.01). In SUAL fetuses, the arterial pH and THb did not change over the duration of the experiment (P > 0.05).
Activin A (Fig. 1): In both the control and the SUAL group, the activin A concentrations were higher in the AF than in the fetal and maternal circulations. In the control animals, activin A levels in all compartments remained unaltered throughout. In the SUAL animals, activin A levels in the fetal circulation were significantly elevated at 121–125 days GA compared with controls (P= 0.04). Similarly, in the SUAL animals AF activin A levels were increased significantly at all time points (surgery–115 days GA, P= 0.01; 116–120 days GA, P= 0.001; 121–125 days GA, P= 0.004; 2–4 days prior to labour, P= 0.002; P= 0.0001) compared with controls (Fig. 1b). In the maternal circulation, the overall mean activin A concentration in the SUAL group was not different to that of the control group (Fig. 1c).
PGE2: In the SUAL group, the overall mean fetal PGE2 concentration was 0.60 [0.13] nM for three animals and the fetal PGE2 for the remaining three animals in this group was less than or equal to the minimal detectable value (0.4 nM). In all of the control animals, fetal PGE2 was less than the minimum detectable value. Although not significantly different, the overall mean PGE2 in the AF of the SUAL group was higher (21.41 [7.18] nM) than the overall mean for three of the control animals, which was 3.72 [1.47] nM (P > 0.05). The AF PGE2 concentration of one of the remaining control animals was less than or equal to the minimal detectable value.
Cortisol: In the control animals, fetal arterial cortisol concentrations at 116–120 days GA (2.77 [0.63] ng/mL) and 121–125 days GA (2.35 [0.66] ng/mL) were significantly decreased compared with at surgery–115 days GA (4.47 [0.70] ng/mL) (P= 0.01). The SUAL fetal cortisol concentrations were significantly elevated at labour (41.23 [16.40] ng/mL) (P= 0.02; n= 6) compared with the control group immediately prior to postmortem (4.52 [0.87] ng/mL). The AF and maternal serum cortisol concentrations in the SUAL group were not significantly different to those in the control group (P > 0.05) at any time point.
In the SUAL group the overall mean fetal arterial glucose (0.82 [0.06] mmol/L) and lactate (2.13 [0.31] mmol/L) concentrations were not significantly different to those in the control group (glucose = 0.84 [0.85] mmol/L; lactate = 1.32 [0.08] mmol/L) at any time point.
There was a negative correlation between fetal arterial %SaO2 and AF activin A (r2= 0.087; P= 0.001). AF activin A and AF PGE2 showed a positive correlation (r2= 0.071; P= 0.02), as did AF activin A and fetal arterial cortisol (r2= 0.067; P= 0.01). There was a negative correlation between AF activin A and fetal arterial pH (r2= 0.013; P= 0.047) as well as a negative correlation between AF activin A and fetal arterial PaO2 (r2= 0.14; P= 0.0001) and AF activin A fetal arterial lactate (r2= 0.277; P= 0.0001).
- Top of page
- Materials and methods
Previously, it has been shown that experimentally induced acute hypoxia, via reduction in fetoplacental blood flow, increases ovine fetal and AF activin A concentrations.20,22 However, such acute hypoxia does not reflect the chronic fetoplacental compromise that occurs in human pregnancies complicated by placental insufficiency and IUGR. Therefore, in this current study, we aimed to utilise an animal model of chronic hypoxia and fetal growth restriction to more closely mimic severe human IUGR allowing us to explore the physical endocrine response, and in particular activin A response, in this setting of chronic fetoplacental compromise.
As reported over three decades ago,24 the experimental model of ovine SUAL successfully produced chronic fetal hypoxia, as demonstrated by a significant reduction in fetal SaO2 over the course of the experiment, and significant fetal growth restriction. In addition, the growth restriction observed in the SUAL group was asymmetric in nature, as defined by reduced BW with normal fetal anthropometric measurements and significantly increased fetal brain–BW ratio, consistent with brain sparing secondary to the decreased cerebral vascular resistance that occurs during hypoxic stress.32 This asymmetric pattern and increased cerebral blood flow response are similar to that observed in human IUGR of placental origin6 and validate the SUAL model as an appropriate one for studying human IUGR. The two-fold increase in the SUAL fetal adrenal–BW ratio compared with that in the controls is also consistent with previous reports of increased adrenal activity during prolonged fetal hypoxaemia.33 SUAL also produced partial infarction of the placental bed, therefore potentially reducing the full functional capacity of the placenta.
Consistent with the effect of acute fetal hypoxia, chronic hypoxia induced by SUAL was associated with significantly increased concentrations of activin A in the AF. However, there are important differences in the results of the current study and those previously reported in acute hypoxia models. We believe that these differences may offer some insights into the potential sources and mechanisms underlying the altered activin secretion in fetal hypoxia. Firstly, the AF activin A levels observed in the current chronic model were at least 12-fold higher than those previously reported during acute uteroplacental hypoxia.20,21,34 While this may simply be a result of a sustained insult, compared with an acute insult, in the acute studies the increased activin A levels attained a plateau after 30 minutes and did not increase further, despite continued hypoxia. This would suggest that the chronic hypoxia observed in this study may be inducing additional or different sources of activin release from those in the acute study. Furthermore, in the current study, fetal activin A concentrations in the SUAL animals were similar to those in controls whereas acute hypoxia induced by either reduced uterine blood flow (RUBF)20 or via maternal isocapnic hypoxemia21 increased fetal activin A levels. This suggests that either the fetus adapts to the chronic hypoxia stimulus or the source of activin A in the AF and fetal circulation may be of different origin and also differentially regulated, depending on the method of hypoxia induction. In this regard, maternal isocapnic hypoxemia, induced via maternal nitrogen insufflation,34 renders both the fetus and the mother hypoxic, whereas RUBF22 produces hypoxia only in the fetus, as a result of disruption to the uteroplacental blood flow. In the current study of SUAL, the fetus was made chronically hypoxic as a result of disruption to the fetoplacental circulation, inducing infarction of approximately half of the placental mass. In SUAL-induced IUGR, placental oxygenation from the mother is not perturbed and thus we hypothesise that placental activin output is not increased and therefore fetal levels remain unchanged.
Petraglia et al.35 have shown an increased expression of activin βA subunit and activin receptor mRNA in human amnion and chorion as a result of premature labour and Casagrandi et al.36 have shown the up-regulation of the activin βA subunit gene in pre-eclamptic placental tissue. Evaluation of activin protein and mRNA levels in the placenta in acute and chronic hypoxia may help to clarify the source(s) of circulating fetal activin A in these models. Whatever the source of the activin in the fetal circulation, we propose that the AF activin observed in both acute and chronic studies is largely derived from the fetal membranes, as has been shown in the human.37,38 Although the source of the AF activin A is not known in this study, we propose that fetal but not placental hypoxaemia increased fetal membrane activin A production, as supported by the negative correlation between AF activin A and fetal arterial SaO2, PaO2 and pH in this and previous studies.34 In support of this, Keelan et al.39 showed that under conditions of 20% O2, the production of activin A in vitro by human amnion was increased by pro-inflammatory cytokines, the effect in turn being mediated by hypoxia.40 Importantly, in human clinical studies of IUGR, maternal serum levels of activin A are significantly elevated.18 The maternal concentration of activin A was not altered in the current study of IUGR or in previous acute hypoxia studies, despite large increases in AF activin A. Although ovine maternal activin A concentrations were not changed in response to fetal hypoxia, we propose that the source of activin A in both humans and sheep is likely to be the same. That activin A was increased in AF but not in the maternal circulation in ovine pregnancy may be due to the structural differences in placentation between humans and sheep, thus limiting activin A transfer across the fetal–maternal interface in the sheep.41
The elevation in AF PGE2 observed in the SUAL animals is twofold higher than that reported during periods of acute hypoxia.21 In this study we demonstrated a positive correlation between AF activin A and PGE2. Hence, it is possible that the high concentration of activin A in the AF of the SUAL fetus could be involved in the release of PGE2 from the cotyledons and fetal membranes42 that are in close proximity to the AF. In keeping with this, in acute fetal hypoxia, the increase in activin A precedes the increase in PGE2.20 Moreover, in vitro studies have shown that activin A induces release of PGE2 from a human amnion derived cell line.43 Such an effect of activin is of potential relevance to the health of the fetus, as PGE2 may be involved in maintaining fetal homeostasis through vasodilation or via the antagonism of catecholamine action44,45 during periods of chronic hypoxia.
In the healthy ovine fetus, the adrenal glands are capable of producing cortisol from approximately 130 days of gestation.46 All IUGR fetuses in this study entered preterm labour, suggesting a high degree of stress and in keeping with this the fetal cortisol concentrations were significantly elevated. The positive correlation between AF activin A and fetal arterial cortisol may reflect a pathophysiological relationship between activin and cortisol. In support of this, activin A induces the release of cortisol from human fetal adrenal zone cells in vitro47 and activin βA subunit is localised in the zona fasciculata of the fetal adrenal cortex,22 the site of cortisol production. However, in the present study, fetal circulating activin A concentrations were unchanged during SUAL. This could argue against activin A being a key mediator of early cortisol release from the fetal adrenal. Whatever the stimulus for the increased cortisol, this early activation of the fetal hypothalamo–pituitary–adrenal axis likely represents the initiation of an escape mechanism induced by the severely compromised SUAL fetus resulting in preterm labour thereby circumventing death in utero.25
In summary, the current study has demonstrated that SUAL induces chronic fetal hypoxia, asymmetric fetal growth restriction and preterm labour, which mimic the clinical picture often observed in human IUGR. In SUAL-induced IUGR fetuses, there was a highly significant increase in AF activin A, but no change in fetal or maternal circulating levels of this hormone. AF PGE2 concentration was raised, as was fetal circulating cortisol. AF activin A concentration was negatively correlated with fetal SaO2 and pH and positively correlated with the changes observed in PGE2 and cortisol. We propose that activin A may, either directly or indirectly, regulate the release of PGE2 and cortisol thus playing an integral part in maintaining fetal homeostasis as well as preparing the fetus for preterm birth if the compromise becomes intolerable for intrauterine survival. Together, these results suggest that activin A may be essential for the adaptive changes that occur during periods of chronic placental insufficiency and accompanying fetal hypoxia.