Dr S. L. Miller, Department of Physiology, Monash University, Building 13F, Victoria, Australia.
Objective To determine whether activin A levels reflect oxygen availability in basal and hypoxic conditions in the late pregnant fetus and newborn lamb.
Design In vivo animal experimental study.
Setting Department of Physiology, Monash University.
Population Chronically catheterised fetal sheep in late gestation.
Methods Fetal hypoxia was induced at 125 (n= 4), 135 (n= 4) or 145 days (‘term’; n= 3) gestational age by maternal nitrogen exposure, for 4 hours, during which maternal and fetal arterial, and amniotic fluid samples were collected. Lambs (age one, five and eight days; n= 3) were exposed to 1 hour of hypoxia via nitrogen exposure.
Main outcome measures Activin A, prostaglandin E2 (PGE2) and cortisol were analysed in plasma and amniotic fluid, and whole blood was used to determine Pao2, Paco2, %O2, lactate and pH.
Results Basal activin A concentrations in the fetal arterial circulation remained unchanged between 125 days (0.230 [0.10] ng/mL) and term (0.28 [0.10] ng/mL), as did fetal oxygen saturation (59.11% [4.74%] to 52.25% [4.84%]) and pH (7.35 [0.02] to 7.37 [0.02]). Moderate fetal hypoxia (50% fall in fetal arterial %O2) produced a significant increase in circulating activin A (2.05 [0.67] ng/mL) and a significant decrease in pH (7.27 [0.03]) at 125 days of gestation, however, at 135 and 145 days, activin A and pH remained unchanged. Fetal activin A concentration was significantly correlated with pH (P= 0.036) but not %O2 (P= 0.072). Hypoxia in the lambs did not alter circulating activin A.
Conclusions In response to hypoxia, activin A is increased in the circulation of 125-day-old fetuses, but not in older fetuses. Fetal arterial activin A levels sensitively reflect pH but not oxygen saturation, with increasing activin A in conditions of metabolic acidosis.
Activin A is a dimeric protein belonging to the transforming growth factor-β (TGF-β) superfamily of growth and differentiation factors. In pregnancy, the predominant sources of circulating activin A are the placenta, fetal membranes and the fetus.1 High concentrations of activin are present during human pregnancy in both maternal and fetal serum and in amniotic fluid.2,3 Activin A levels are elevated in maternal serum in a variety of pregnancy disorders including pregnancy hypertension, gestational diabetes and intrauterine fetal growth restriction (IUGR).4–8 The observation that maternal serum levels of activin A are also significantly increased prior to the onset of pre-eclampsia,9 suggests that this protein may be useful as a predictive marker for subsequent pregnancy complications.10
We have previously demonstrated that activin A is increased in the circulation and amniotic fluid of ovine fetuses in late gestation during an experimental reduction in uteroplacental blood flow.11 This supports the hypothesis that activin A could be a marker of fetoplacental compromise and, more specifically, that circulating activin levels are related to fetoplacental oxygenation. While the precise role of activin A during pregnancy is yet to be determined, there is evidence to suggest that it is neuroprotective during hypoxia.12 Moreover, in vitro studies have shown that activin A induces prostaglandin E2 (PGE2) release from an amnion-derived cell line13 and in vivo the increase of activin A levels during induced hypoxia is followed by increasing PGE2.11 These observations are also consistent with activin A playing a protective role by releasing vasodilatory prostaglandins. Accordingly, we hypothesise that activin A is increased in response to decreasing fetal oxygen availability and that this response reflects the protective functions of activin A during pregnancy.
Maternal serum activin A levels increase in the third trimester in normal pregnancy,14,15 at a time that fetal venous pH is decreasing.16 In a small study of fetal activin levels at delivery, umbilical artery activin was inversely correlated with pH, consistent with existing experimental data linking activin secretion and hypoxia.2 In light of these data, the current experiment was designed to investigate basal activin A concentrations, fetal oxygen saturation and pH over the last trimester of ovine pregnancy and to more closely examine the relationship between fetoplacental oxygenation and activin secretion. We also wished to examine the response to fetal hypoxaemia at three different ages during this period: 125, 135 and 145 days of gestation, to determine whether the magnitude of the response changes across gestation. Finally, we wanted to further investigate whether a similar response was evident neonatally by exposing newborn through to eight-day old lambs to hypoxaemia. We hypothesised that basal activin A levels in lambs will be very low, reflecting removal from the uterus and the change from low to high arterial oxygen saturation.
All procedures were conducted with the approval of the Physiology Animal Ethics Committee, Monash University. Surgery was performed on 11 pregnant ewes between 116 and 125 days of gestation. Anaesthesia was induced with 5% thiopental sodium in water, intravenously, and maintained after tracheal intubation with 1–2% halothane (oxygen/nitrous oxide, 50:50 vol/vol). Catheters were inserted aseptically into the fetal femoral artery and vein (1.52 mm OD, 0.86 mm ID), the maternal carotid artery and jugular vein (2.70 mm OD, 1.50 mm ID) and the amniotic cavity (2.70 mm OD, 1.50 mm ID). A catheter was also inserted into the trachea of the ewe. The fetal and amniotic catheters were exteriorised via an incision in the ewe's flank. All catheters, except the tracheal catheter, were filled with sterile heparinised saline (0.9% NaCl; 25,000 IU heparin L−1). After surgery, the ewes were housed in metabolic cages and fed once daily. Water was provided ad libitum.
At 125, 135 or 145 days, fetal gestational age (term ∼147 days), maternal isocapnic hypoxaemia was induced via one of two methods. Originally, 10 hours of hypoxia was undertaken, based on previous findings,11 using insufflation of a gas mixture of approximately 12 L min−1 N2 and 0.2 L min−1 CO2 via the non-occluding tracheal cannula such that the fetal femoral artery %O2 saturation was reduced by 40–50%. We experienced difficulty in sufficiently reducing fetal oxygen saturation and maintaining the reduction over this prolonged period. Therefore, in subsequent animals, isocapnic hypoxaemia was induced for 4 hours, by placing a large clear plastic bag over the ewe's head, connected to a tube through which a gas mixture of approximately 22 L min−1 N2 and 0.2 L min−1 CO2, in room air was infused. This reduced fetal femoral artery O2 saturation by 40–50%. To provide a comparable level of hypoxaemia with our previous study,11 fetal %O2 saturation was closely monitored and the gas mixture was altered when required to achieve a 40–50% decrease in fetal femoral %O2 saturation.
Maternal and fetal blood and amniotic fluid samples were collected for up to 2 hours preceding induction of hypoxaemia, during and for 2 hours after the hypoxaemic insult. Fetal arterial blood gas parameters (pH, Pao2 and Paco2) were determined throughout the experiment to assess fetal wellbeing using an ABL5 acid–base blood gas analyser (Radiometer, Copenhagen, Denmark). Fetal % oxygen saturation (Sao2) was determined using an OSM2 hemoximeter (Radiometer) or an AVL912 Co-oxylite co-oximeter (Roche Diagnostics, Australia). Fetal arterial lactate was determined using a YSI Glucose and l-Lactate Analyser (YSI, Yellow Springs Instrument, Yellow Springs, Ohio, USA).
Blood and amniotic fluid samples were collected into EDTA tubes, containing 10 μL of 0.001 M indomethacin per mL, and centrifuged at 2000 ×g (Beckman GS-6R) for 10 minutes at 4°C. Plasma aliquots for activin and cortisol were immediately frozen and stored at −20°C until assay. Plasma aliquots for PGE2 were diluted 1:1 with 0.12 M methoxyamine hydrochloride (Sigma, St Louis, Missouri) in sodium acetate buffer containing 10% ethanol and incubated overnight at room temperature prior to freezing and storage at −20°C.
After birth, three lambs underwent surgery for placement of catheters aseptically into the carotid artery and jugular vein. At one, five and eight days after birth, hypoxaemia was induced by placement of a plastic bag over the lamb's head, as previously described. The lambs were placed in a sling in a quiet room to settle comfortably prior to being subjected to 1 hour of room air (6 L min−1), 1 hour of 8 L min−1 N2 and 0.2 L min−1 CO2, in room air, and a 1 hour recovery period of 6 L min−1 room air. Blood samples were collected every 30 minutes for analysis of hormone concentrations and blood gas parameters as above.
Activin A concentrations were determined by a commercial two-site enzyme-linked immunosorbent assay (ELISA) (Oxford Bio-innovations, Oxford, England). Ovine PGE2 and cortisol were measured by radio-immunoassay.17,18
Data are shown as mean (SEM). Where appropriate, data were normalised by square root or log transformation. Analysis was performed using two-factor ANOVA for repeated measures (SPSS 11.0). Due to the difference in the length of induced hypoxaemia, blood gas data (%O2 sat, Po2, Pco2 and pH) were divided into three groups for analysis. Pre (mean of −0.5 and −1 or −2 hours preceding hypoxaemia), during (mean data for 4 hours of induced hypoxaemia) and post (mean data up to 2 hours following cessation of hypoxaemia). Individual time points were used for hormone data analysis. Where significant differences were observed, post hoc testing was performed using Least Significant Difference analysis. Linear Regression analysis was used to examine the relationship between activin A and fetal oxygen saturation, pH and lactate. Results were considered significant when P≤ 0.05.
At all ages of gestation (125, 135 and 145 days), maternal arterial blood Pao2 and %O2 saturation decreased significantly from 88.93 (6.95) to 44.95 (6.72) mmHg (P < 0.001) and from 99.65 (0.60) %O2 saturation to 68.55 (5.28) %O2 saturation (P < 0.005), respectively, during the hypoxaemic episode. Values had returned to pre-induction values within 2 hours of return to normoxia. The maternal arterial blood pH (7.46 [0.03]) and Paco2 (34.39 [3.49] mmHg) remained unchanged in all three age groups throughout the treatment period.
There was no change in activin A concentrations in the maternal circulation throughout the period of the study in all three gestational age groups (0.10 [0.03] ng/mL).
Results have been combined for the two methods of inducing maternal hypoxia. Seven animals were exposed to hypoxia at one gestational age only, and four fetuses were exposed to hypoxia at two gestational ages. Only those animals that reached the required level of hypoxia were included in the final analysis, which included five animals at one gestational age (either 125, 135 or 145 days) and two animals at two gestational ages (125 and 135 days; 135 and 145 days). Hormone and blood gas data from the fetuses receiving a second exposure to hypoxia were not different to results from other fetuses within the same gestational age group.
There was no difference in basal fetal arterial %O2 saturation of the three age groups (59.11% [4.74%], 55.88% [4.58%] and 52.25% [4.84%], respectively). In response to hypoxia, fetal arterial blood Pao2 and %O2 saturation at all ages studied decreased significantly from 24.73 (2.92) mmHg to 15.68 (1.95) mmHg (P < 0.01; Fig. 1) and from 55.76 (2.58) %O2 to 28.09 (2.93) %O2 (P < 0.001), respectively. Both Pao2 and %O2 saturation returned to control values within 2 hours of cessation of the insult. Arterial blood Paco2 did not change during the hypoxic period but was significantly higher at each time point at day 125 compared with days 135 or 145 (P < 0.005). There was no difference in basal fetal arterial pH of the three age groups (7.35 [0.02], 7.34 [0.01], 7.37 [0.02], respectively). At 125 days, fetal arterial blood pH was significantly decreased from basal levels at the end of the recovery period (7.27 [0.03]; P < 0.05). Fetal arterial pH remained statistically unchanged in response to hypoxia at 135 and 145 days (Fig. 1). Basal fetal lactate concentrations (1.6 [0.5], 1.5 [0.3] and 1.9 [0.4] mmol/L, respectively) were significantly increased at all ages in response to hypoxia at 4 hours (10.3 [2.8], 12.8 [1.7] and 10.4 [3.9] mmol/L, respectively; P < 0.001).
At 125 days, fetal circulating activin A increased from 0.23 (0.10) to 2.05 (0.67) ng/mL during the treatment period and remained significantly elevated above control values following treatment (P < 0.001; Fig. 2). There was no change in fetal activin A concentrations throughout the treatment period at 135 and 145 days of gestation, although basal activin A concentrations were significantly higher at 135 days (1.18 [0.30] ng/mL) than at 145 days (0.28 [0.10] ng/mL) (P < 0.05). Linear regression analysis showed that there was no significant correlation between fetal activin A and %O2 saturation for all gestational ages (r2= 0.288, P= 0.072), however, activin A and fetal pH were significantly correlated (r2= 0.371, P= 0.036) as were activin A and fetal lactate (r2= 0.872, P= 0.002). In amniotic fluid, basal activin A levels were not significantly different at any of the gestational ages studied (5.69 [2.93] ng/mL at 125 days, 5.44 [2.63] ng/mL at 135 days and 6.78 [4.31] ng/mL at 145 days) and did not change throughout the insult.
Figure 3 shows PGE2 concentrations in fetuses throughout the treatment period at each gestational age studied. There was no change in PGE2 concentrations in the fetal circulation throughout the period of the study in all three gestational age groups. Basal PGE2 concentrations were not significantly different between 125 and 145 days of gestation.
Fetal cortisol concentrations increased in all three age groups during the period of hypoxaemia (P < 0.001; Fig. 4). At 125 days, cortisol increased from 9.59 (2.07) to 18.81 (3.73) ng/mL and remained elevated above control values following cessation of the insult. At 135 days, cortisol increased from 12.16 (3.17) to 18.69 (3.63) ng/mL and had returned to control values within 30 minutes of cessation of the insult. At 145 days, cortisol had increased significantly within 30 minutes and peaked at 1 hour (36.76 [0.41] ng/mL), after which concentrations began to return to control values.
In response to acute hypoxaemia in lambs, arterial blood Sao2 decreased by 60% from 62.09% (2.26%) to 23.96% (2.48%) (P < 0.001). Arterial blood Pao2 was significantly lower in one-day-old lambs than lambs at five or eight days of age (P < 0.05). On day one, Pao2 decreased from 28.22 (4.08) to 13.0 (1.26) mmHg (P < 0.005); at day five, Pao2 decreased from 41.0 (2.22) to 18.33 (1.42) mmHg (P < 0.005); and at day eight, Pao2 decreased from 36.89 (1.13) to 17.5 (1.16) mmHg (P < 0.01). Arterial Pao2 and Sao2 returned to pre-induction values within 1 hour after hypoxaemia. At all three ages, pH increased significantly from 7.38 (0.02) to 7.41 (0.02) during hypoxaemia (P < 0.001) and remained elevated above pre-induction values for at least 1 hour following cessation of the insult (Fig. 5).
There was no change in circulating activin A concentrations in the lambs during the period of induced hypoxaemia although levels were higher in eight-day-old lambs (1.19 [0.54] ng/mL) than lambs at one or five days (0.20 [0.02] and 0.74 [0.54] ng/mL, respectively) (P < 0.05; Fig. 6). There was no correlation between circulating activin A concentration and pH in the lambs for combined age groups (r2= 0.198, P= 0.317).
In the current study, developmental changes in circulating activin A levels of the late gestation sheep fetus and newborn lamb in response to hypoxia were investigated. At gestational age 125, 135 or 145 days, fetal hypoxaemia was achieved by maternal insufflation of a nitrogen/air mixture. Similarly, hypoxaemia in the lamb at one, five and eight days after birth was achieved by exposure to nitrogen and air. In fetuses and lambs at all ages studied, there was a 50–60% decrease in arterial oxygen saturation under hypoxic conditions. Results showed that there was no change in maternal circulating activin A concentrations in response to materno-fetal hypoxaemia, however, there was a fourfold increase in fetal arterial activin A levels in the 125 day fetuses, consistent with a previous study from our laboratory.11 Of the three late-gestation age groups studied, only the 125 day fetuses responded to hypoxia with increased levels of circulating activin A, which corresponds to the only age group that demonstrated significant acidosis resulting from the hypoxia. Linear correlation analysis showed that fetal circulating activin A and pH were significantly correlated, as was activin A and lactate, across all age groups. Moreover, it can be seen that the highest fetal activin A levels achieved at 125 days were after the period of hypoxia, concomitant with the lowest fetal pH. It therefore appears that activin A is a sensitive marker of fetal metabolic acidosis and may be playing a role in the restoration of homeostasis. After birth, there was no change in circulating activin A levels in lambs during hypoxia, despite a small but significant increase in pH that was attenuated with CO2 administration.
Basal concentrations of activin A were not increased between 125 and 145 days in the fetal and maternal arterial samples, or in the amniotic fluid. This finding is at odds with observations in the human, where maternal serum activin A levels increase significantly over the last trimester of pregnancy.14,15 In the current ovine study, the earliest samples were collected one to two days after surgery, corresponding to 117–118 days of gestation (data not shown). In a separate study, samples have been obtained as early as 106 days, from pregnant ewes and their fetuses, and activin A concentrations are not different to basal values obtained in this study (S. Miller, unpublished observations). These results suggest that in ovine pregnancy there is no increase in activin A levels towards term, although, it may be that circulating basal activin A levels rise at the beginning of the third trimester, corresponding to approximately 95 to 100 days of gestation in the sheep. Over the course of this ovine study, basal fetal oxygen saturation and pH also remained stable. Concentrations of maternal circulating activin A are significantly elevated in compromised human pregnancy, above those seen in healthy pregnancies.6,7,19 However, maternal activin A concentrations do not increase in sheep in response to hypoxia, which may be due to the difference in placentation and vascular arrangement of these species, and may support the observation that the placenta and membranes are the predominant source. Placentation in the human is invasive and haemochorial, while the sheep has non-invasive cotyledonary epithelio-chorial placentation.20
At gestational age 125 days, activin A concentrations were elevated approximately fourfold in the arterial circulation of fetuses within 1 hour of the start of the hypoxic period, consistent with earlier observations from our laboratory in which hypoxia was induced via constriction of the uteroplacental blood flow.11 In the present study, activin A concentrations continued to be raised above basal levels in the 2-hour sampling period following the cessation of hypoxia, which is likely to be due to the sustained depression of pH and oxygen saturation. In the previous study by Jenkin et al.,11 amniotic fluid activin A concentrations were also increased in response to fetal hypoxia, whereas amniotic fluid activin A remained unchanged in the current study. In the current investigation, fetal hypoxia was induced by altering maternal gas concentrations, thus inducing hypoxaemia in both the mother and the fetus, whereas in the earlier study, uterine blood flow was directly manipulated by mechanical occlusion of the common internal iliac artery. It would therefore appear that the mode of hypoxia induction affects the production and release of activin A. This is not unexpected, as other studies have shown that both maternal and fetal endocrine and haemodynamic responses differ depending on the mode of hypoxia induction.21 Umbilical blood flow is not decreased in response to maternal nitrogen inhalation,22 which suggests that a disruption of uteroplacental blood flow may be the stimulus for release of activin A into the amniotic fluid. Combined, these observations indicate that multiple sources of activin A contribute to fetal circulating and amniotic fluid levels. Fetal circulating concentrations may be due to adrenal, pituitary and/or brain production, as seen in adult animals,23 while in pregnancy, the placenta and membranes also produce activin A,2,7,24 this also being the source for PGE2.25
It has been hypothesised that, during hypoxia, uteroplacental PGE2 release may be induced by the increase in activin A.11,13 In the present study, there was no change in basal PGE2 concentrations between 125 and 145 days, and hypoxia did not produce a rise in PGE2. PGE2 has previously been shown to increase in response to fetal hypoxia induced by reduced uterine blood flow.11,26 However, in these studies, elevated levels are not observed until 3 to 4 hours from the beginning of the insult. Therefore, in the current experiment, the duration of hypoxia may have been too short to observe altered prostaglandin levels. Alternatively, because amniotic fluid activin A was not increased in this study, the membranes and placenta may not have been sufficiently affected by the insult to release either activin or PGE2.
It does not appear that circulating activin A levels are related to basal circulating cortisol concentration, as basal activin A levels are unchanged between 125 and 145 days of gestation in late pregnant ovine fetuses, whereas cortisol concentrations are elevated over this period, as shown previously,27 reflecting its accepted role in the processes leading to parturition.28 Previously, Carmichael et al.29 showed that the threshold for cortisol release was a fetal oxygen saturation of 30%, which is supported in the current study, where saturation fell to approximately 28%. Cortisol was increased in response to hypoxia at all three ages, however, the cortisol response was greater at 145 days, peaked at 1 hour during hypoxia and began to return to basal levels during the insult. This reflects increasing sensitivity of the hypothalamic–pituitary–adrenal axis to elevated basal cortisol, and maturation of the negative feedback action of cortisol on adrenocorticotrophin release.30
In summary, this study has demonstrated that fetuses responded to hypoxia at 125 days of gestation with an increase in circulating activin A levels, but did not respond later in gestation. That activin A was not increased following hypoxia in the 135 and 145 day age groups does not necessarily reflect a decrease in activin A response, but is more likely due to fetal metabolic status during and after hypoxia. Fetal arterial pH was significantly decreased following the hypoxic insult at 125 days of gestation, with a clear relationship with elevated fetal activin A concentrations. These results support the hypothesis that activin A may play a role in the restoration of fetal metabolic homeostasis in compromised conditions although how activin would achieve this remains to be defined. Neither fetal activin A nor pH was significantly altered in response to hypoxia at 135 and 145 days. Basal activin A levels remained unchanged over the 125 to 145 day gestation period, as did fetal pH and oxygen saturation. Amniotic fluid levels of activin A were unchanged at all gestational ages during hypoxia in this study, which suggests that the source of activin A in the fetus and amniotic fluid is different and that the mode of hypoxia induction produces activin release from different compartments. Further experiments are required to examine the source of activin A contributing to fetal, amniotic and maternal fluids and to further elucidate the possible protective role of activin A in response to hypoxia.
This research was supported by the NHMRC Australia. The authors would like to thank Professor Nigel Groome for provision of activin A assay reagents and Alex Satragno for assistance with the preparation of animals.