Placental Dysfunction Underlies Increased Risk of Fetal Growth Restriction and Stillbirth in Advanced Maternal Age Women

Pregnancies in women of advanced maternal age (AMA) are susceptible to fetal growth restriction (FGR) and stillbirth. We hypothesised that maternal ageing is associated with utero-placental dysfunction, predisposing to adverse fetal outcomes. Women of AMA (≥35 years) and young controls (20–30 years) with uncomplicated pregnancies were studied. Placentas from AMA women exhibited increased syncytial nuclear aggregates and decreased proliferation, and had increased amino acid transporter activity. Chorionic plate and myometrial artery relaxation was increased compared to controls. AMA was associated with lower maternal serum PAPP-A and sFlt and a higher PlGF:sFlt ratio. AMA mice (38–41 weeks) at E17.5 had fewer pups, more late fetal deaths, reduced fetal weight, increased placental weight and reduced fetal:placental weight ratio compared to 8–12 week controls. Maternofetal clearance of 14C-MeAIB and 3H-taurine was reduced and uterine arteries showed increased relaxation. These studies identify reduced placental efficiency and altered placental function with AMA in women, with evidence of placental adaptations in normal pregnancies. The AMA mouse model complements the human studies, demonstrating high rates of adverse fetal outcomes and commonalities in placental phenotype. These findings highlight placental dysfunction as a potential mechanism for susceptibility to FGR and stillbirth with AMA.

factor (PlGF) 29 . Clinical measures of placental dysfunction, including abnormal Doppler ultrasound resistance indices in both umbilical and uterine arteries (indicating increased vascular resistance and impaired spiral artery adaptations) are more prevalent in FGR and stillbirth 30,31 . Consistent with these clinical observations, chorionic plate resistance arteries in FGR exhibit reduced constriction and exacerbated relaxation 32 as assessed by wire myography. Conversely, myometrial arteries show attenuated relaxation in FGR 33 . Whether any of these aspects of placental development, structure or function are altered in mothers of AMA and whether they underpin the susceptibility to FGR and stillbirth is unknown.
Animal models are frequently used to dissect the role of placental dysfunction in adverse pregnancy outcomes. A previous study found decreased pup viability in ageing (40 week old) mice compared to younger controls, coupled with histological evidence of stunted uterine artery remodelling 34 . Although a recent study using aged pregnant rats found altered maternal uterine vascular function 35 , placental function analyses were not conducted in either of these rodent models.
The current study addressed the hypothesis that AMA is associated with placental and utero-placental dysfunction. Using a broad array of investigations in both human pregnancies and a murine model of AMA we identified multiple features of placental and utero-placental dysfunction, providing a potential mechanistic link for the increased FGR and stillbirth rates with AMA.

Results
Participant Demographics. Table 1 shows the demographics and obstetric outcome for the total study population. Demographic and biophysical characteristics were matched between maternal age groups wherever possible. However, paternal age was significantly higher for mothers of AMA compared to controls (p < 0.0001). More women of AMA were home owners (p < 0.05) and a greater proportion were parous (p < 0.05) but not grandparous. No differences were seen in IMD (a proxy for socioeconomic status), although there may potentially be a negative linear relationship if a continuum of maternal ages had been included. Although mothers ≥40 years had higher rates of previous miscarriage (p < 0.05), there were no differences in incidence of previous adverse pregnancy outcome (stillbirth or SGA infant <10 th individualised birthweight centile (IBC)) between age groups. No participants in this cohort had stated use of assistive reproductive techniques.
By study design, all women had uncomplicated pregnancies defined as: live birth, delivery between 37-42 weeks gestation, birthweight >2500 g, IBC between 10-95 th centile and absence of pre-existing or pregnancy-related maternal disease. Mothers of AMA delivered on average 5-7 days earlier than 20-30 year olds (p < 0.01), but no differences were seen in birthweight or IBC. Rates of induction of labour and normal vaginal delivery (NVD) were not different between groups. However, AMA groups had higher emergency Caesarean section and lower instrumental vaginal delivery rates (p < 0.05).

Human Placental Analyses.
Placental analyses were performed on matched sub-cohorts, the participants of which did not differ in demographic or obstetric outcomes from the total study population unless stated otherwise.
Placentas used for histology were matched for gestational age, maternal BMI and were all from non-smokers (Supplementary Table 1). Mothers ≥40 years were less likely to have delivered by NVD (20.0% vs 62.5%; p = 0.05). There were higher numbers of SNAs in placentas from 35-39 and ≥40 year olds compared to 20-30 years olds (p < 0.05; Fig. 2a). There was no difference in number of capillaries per terminal villi (Fig. 2b) or number of avascular villi (not shown) with maternal age. Although apoptotic index was unaltered (Fig. 2d), reduced proliferative index in placentas from ≥40 year olds compared to controls was detected (p < 0.05; Fig. 2c).
Placental Function. Placental amino acid uptake via system A and TauT. The demographics for participants whose placentas were used for nutrient transport studies were matched to those in Table 1 (Supplementary  Table 2) with a few variances. There was no difference in gestation at delivery or parity with maternal age. Although birthweight was not different between the maternal age groups, IBC was significantly higher in mothers ≥40 years compared to controls (p < 0.05). System A activity (Na + dependent 14 C-MeAIB uptake) was increased in placentas from mothers aged ≥40 years compared to those aged 35-39 years (p < 0.005, Fig. 3a). TauT activity (Na + dependent 3 H-Taurine uptake) was increased in placentas from both mothers aged 35-39 and ≥40 years compared to 20-30 year old counterparts (p < 0.05, Fig. 3b).
Myometrial Artery Wire Myography. Constriction of myometrial arteries to U46619 was variable but not affected by maternal age (n = 15/group; Fig. 3e). However, pre-constricted myometrial arteries from women ≥40 years old showed greater relaxation response to bradykinin than controls (p < 0.05; Fig. 3f

Fetal and Placental Measurements
Litters were compared between control and AMA mice at E17.5. Data are displayed as litter averages (N) and total number of pups (n). AMA mice had 54% fewer total and viable pup numbers (p < 0.0001), twice as many resorptions (p < 0.0001) and more late fetal deaths (p < 0.01; control N = 34 (n = 156), AMA N = 23 (n = 99); Table 2). All the late fetal losses (non-viable pups) were in AMA mice, affecting 31% of AMA litters. The total number of conceptuses evident at E17.5 (viable/non-viable pups and reabsorptions) was lower in AMA than in young mice (p < 0.05; Table 2). Fetal and placental measurement data from AMA mice were analysed for viable and non-viable pups separately (control N = 22 (n = 165), AMA viable N = 13 (n = 86), AMA non-viable N = 12 (n = 15)). AMA viable fetal weights were lower than controls and AMA non-viable pups weighed less than both control and AMA viable pups (by 21% and 49% respectively p < 0.0001; Fig. 5a).
When analysed as a frequency distribution, the fetal weight curve from AMA pregnancies was shifted to the left compared with the control group, with 56% of AMA fetuses weighing < 5 th centile of the control population ( Fig. 5b). AMA viable-pups had higher placental weights compared to control (p < 0.005). However, placentas of AMA non-viable pups weighed significantly less than the viable AMA group, but were comparable to controls (p < 0.005; Table 2, Fig. 5c). AMA placental weights showed a rightward shift in weight frequency distribution compared to control, with 48% of AMA placentas measuring >95 th centile of the control population (Fig. 5d). FW:PW ratios were lower in AMA viable-pups compared to controls and lower still in AMA non-viable pups compared to both control and AMA viable (p < 0.0001; Fig. 5e). Furthermore, there was a leftward shift in frequency distribution, with 71% and 92% of AMA viable and non-viable pups respectively falling more than 2.5 standard deviations from the mean of the control population ( Fig. 5f).
All anthropometric measures (crown rump, abdominal and head circumference) were reduced in AMA mice and this was exacerbated if the pup was non-viable at E17.5 (p < 0.0001; Table 2). Head:abdominal circumference ratio was decreased between non-viable AMA pups and controls, indicating head-sparing in these offspring (p < 0.05; Table 2).

Placental Function.
Fetuses that were determined to be non-viable at E17.5 and their placentas were excluded from functional analyses. System A and TauT Maternofetal Clearance. The rate of uni-directional maternofetal clearance of 14 C-MeAIB per gram of placenta was reduced in AMA mice (p < 0.005; control N = 9 (n = 66), AMA N = 8 (n = 37); Fig. 6a) compared to controls. Total 14 C-MeAIB transfer, independent of placental size, was also reduced (Fig. 6b). Uni-directional maternofetal clearance of 3 H-Taurine was also reduced in AMA mice (p < 0.01; control N = 8 (n = 63), AMA N = 6 (n = 26); Fig. 6c) compared to controls with correction for placental weight. Total maternofetal transfer of 3 H-Taurine, independent of placental size, showed no differences between groups (Fig. 6d).

Uterine and Umbilical Artery Myography.
Umbilical arteries exhibited no differences in constriction responses to U46619 or relaxation to SNP between maternal age groups (control N = 6 (n = 24), AMA N = 4 (n = 13), Fig. 6e,f). Uterine arteries from AMA dams showed no differences in constriction in response to PE compared to controls (Fig. 6g), but pre-constricted arteries exhibited increased relaxation to ACh when compared to controls (p < 0.001; control N = 7, AMA N = 5, Fig. 6h).

Discussion
This study addressed the hypothesis that utero-placental dysfunction is an underlying mechanism for susceptibility to FGR and stillbirth in AMA pregnancies. Detailed characterisation studies revealed increased placental weight, but reduced placental efficiency, in pregnancies from women of AMA, together with alterations in placental morphology and function resembling the placental phenotype in FGR and stillborn pregnancies. This phenotype was evident despite apparently normal pregnancy outcomes, indicating adverse effects of maternal ageing on the placenta and potential adaptive responses to achieve a normal birthweight. The abnormal placental phenotype was more marked in women ≥40 years of age, consistent with a higher risk of adverse pregnancy outcomes than women aged <35 years 36 . To exclude confounding effects in clinical studies, we developed a mouse model of AMA. AMA mice experienced a more severe pregnancy phenotype than in the human pregnancies, with reduced fecundity, increased resorptions and late fetal deaths, and over half of the offspring were classed as FGR (fetal weight <5 th centile). Many similarities were detected in the placental phenotype in the mouse model and human pregnancies, particularly increased placental weight, decreased placental efficiency and altered uteroplacental vascular function, although many of the changes were exaggerated consistent with the more severe fetal outcomes detected. These complementary data from human and murine studies support the hypothesis that maternal ageing is associated with placental dysfunction, which may contribute to the higher rates of stillbirth and FGR in AMA pregnancies.
Morphological differences were apparent in the placentas from women of AMA compared to younger counterparts from gestationally matched, uncomplicated pregnancies, including increased numbers of SNAs and reduced rates of trophoblast proliferation. Both are features of placental dysfunction in pathological pregnancies, including FGR, stillbirth and pre-eclampsia 37,38 , but also of post-term pregnancies signifying placental ageing 39 . Placental ageing is a probable cause of the dramatic increased risk of stillbirth in post term pregnancies (>41 weeks gestation) 3, 6, 40 . Accelerated placental ageing is a potential explanation for a comparable risk of stillbirth in women ≥40 years at 39 weeks gestation 40 . Further mechanistic studies are required to investigate this hypothesis further.
Reduced placental efficiency, indicated by a lower FW:PW ratio 41 , occurs in FGR pregnancies 42 . In women ≥40 years with uncomplicated pregnancies, placental weight was higher than in controls. This was accompanied by changes in placental function -namely elevated amino acid transport and augmented relaxation of CPA and myometrial arteries in response to vasodilators -which together would be proposed to be beneficial for fetal growth and survival. The absence of a parallel increase in birthweight is indicative of either deficiencies in other important aspects of placental function, or an adverse maternal environment that is unfavourable to fetal growth. In the context of this AMA environment, there are placental changes in terms of increased size and altered function. Whether these changes are an adaptation by the placenta in order to maintain appropriate growth of the fetus, or simply a consequence of the altered maternal environment, requires further study. These studies provide supportive evidence for the concept of placental adaptations in human pregnancies, as is well established in murine pregnancies 43 (e.g. Igf2, placental specific Igf2 and 11-βHSD2 knock out models), where placental amino acid transport is enhanced to compensate for reduced placental size. These support normal fetal growth and failure of the adaptations results in FGR 44 . Such functional adaptions may be driven by fetal nutrient demand, as in the mouse 43,45 , or in response to an adverse environment, for example elevated oxidative stress, as reported in neurones 46,47 , or as a consequence of the ageing maternal environment.
The measurement of third trimester placental hormones in maternal circulation has been previously studied to assess placental function and risk of adverse pregnancy outcome 28,48 . Lower maternal serum PAPP-A in early gestation is associated with PE, prematurity, FGR and stillbirth 27,49 , and third trimester concentrations of hPL and PlGF are decreased in FGR 28,29,50 . Lower levels of placental hormones are presumed to infer reduced placental size and/or function in FGR 28,49,51 . Although PlGF concentrations were not altered, the higher PlGF:SFlt ratio indicates higher free bioavailable PlGF, which is consistent with the increased placental size detected. The absence of alterations in hCG, hPL or progesterone concentrations, and the significantly lower PAPP-A, in AMA pregnancies are thus prominent findings given the higher placental weight and instead imply reduced placental endocrine function.
The human studies were complemented by studies of a mouse model of AMA, which showed similarities in pregnancy outcomes to those reported by large epidemiological studies of human AMA pregnancies, including reduced fecundity 52 , increased early fetal loss 53 , FGR 10 and increased late stillbirths 54 . These were detected at higher frequency than in human pregnancies, indicating a more extreme model, which is valuable in terms of delineating the mechanisms causing stillbirth with AMA. 40 week old mice were chosen as this best represents the age range of women in the human studies 34,55 . Similarities were detected in the placental phenotype in both mouse and human studies, most notably increased placental weight and reduced PW:FW. The increase in placental weight was only apparent in viable pups at E17.5, supporting the hypothesis from the human studies for increased placental growth as an adaptation to promote fetal growth and survival. The significant overlap in FW:PW ratios between the maternal age groups in the human pregnancies is likely reflective of the diversity when studying human pregnancy, emphasising the need for an animal model. The biological significance of a moderate change in placental efficiency in humans (12.2% between controls and mothers ≥40 years) is uncertain; however this is consistent with all women having normal outcomes. The much greater reduction in FW:PW in our mouse model (37.4% in AMA from controls) is likely to reflect the more severe phenotype in the mouse model which presents with high rates of FGR and stillbirth.
Enhanced endothelial-dependent relaxation of uterine arteries was also detected in the AMA mouse model, suggesting that the observations in human pregnancies are not an artefact relating to increased parity and prior  irreversible pregnancy-related adaptations in older women 56 . A rat model of AMA found raised systolic blood pressure but reduced uterine artery maximal constriction and no differences in endothelial-dependent vascular relaxation in aged pregnant rats compared to young controls 35 . Comparisons between these models is limited however as their aged dams were calorie constricted and endothelial-independent relaxation of uterine arteries was not reported 35 . A major difference was the reduced amino acid transport detected in AMA mice, in contrast to the enhanced transport in human pregnancies. This is probably attributable to the key difference in the models. Placentas from human studies were selected to only include those with birthweights in the normal range to exclude any overt effects of pathology on the placenta, whilst mice experienced severe adverse fetal outcomes. The reduced transport in mice corresponded to 56% of pups being growth restricted. Potential mechanisms for the altered transport include down-regulation of placental mTOR or insulin/IGF-I signalling, both of which are known modulators of ageing 57 and underpin the reduced amino acid transport in a baboon model of FGR 58 . More detailed temporal  A limitation of the mouse model is the fact that AMA mice were significantly heavier than control mice. Although this was highly variable, the effect on pregnancy outcome could not be separated from those of ageing. Although obesity is associated with increased adverse outcomes, including macrosomia and stillbirth 59, 60 , it is not associated with reduced litter sizes, severe FGR, or increased placental weight 59,60 . Furthermore, mouse models of obesity in pregnancy are generated by providing a high fat, obesogenic diet 61, 62 which may cause adverse outcomes rather than the increased fat mass per se 63 . The higher pre-pregnancy maternal weight in our AMA mice is not due to dietary alterations and restricting their calorie intake would add further confounders 64,65 . Of note, the pregnancy outcomes for AMA mice with greater maternal weight did not differ from AMA mice without significant age-related maternal weight gain. A further limitation of utilising a mouse model of pregnancy is their polytocous nature, introducing potential confounders, including uterine positioning, litter size and litter-mate competition that do not exist in human pregnancies. However, a carefully chosen animal model can complement human studies providing the limitations are identified and understood.
The mechanisms underpinning the abnormal placental phenotype in AMA pregnancies remain unknown, but may be related to the ageing oocyte accumulating genetic (and epigenetic) damage 66,67 (Fig. 7). Interestingly, our AMA mouse model shares striking similarities to a mouse model of IVF used to investigate placental function and development 68 including reduced fetal viability and weight and increased fetal loss and placental size, paired with reduced expression and activity of nutrient transport systems. Epigenetic modulation of imprinted genes, namely H19 and PHLDA2, governing early embryo development was cited as the underlying mechanism 68 . The effect of AMA on the epigenetic modulation of imprinted genes is currently unknown 69 . Alternatively, the ageing maternal environment may also contribute to placental dysfunction (Fig. 7). Ageing is associated with increased levels of inflammatory and oxidative stress markers 70 , many of which have been associated with poor pregnancy outcome independent of maternal age 71,72 . Placental ageing is largely attributed to the placenta being a major source of reactive oxygen species (ROS) which leads to oxidative stress, cellular senescence and loss of tissue function 73,74 . In in vitro models, ROS exposure reproduces some of the placental phenotype seen in AMA pregnancies, such as increased SNAs 22 , altered amino acid transport 46 and decreased proliferation 75 . A recent study reported increased oxidative stress in the utero-placental bed of aged pregnant mice, and partially improved pregnancy outcome with antioxidant treatment 76 . Further studies are required to determine whether these and other factors in the ageing maternal environment are altered in human pregnancies and are directly related to placental function.
In summary, AMA is associated with multiple aspects of placental dysfunction and altered utero-placental vascular function that are detectable even in pregnancies with apparently normal outcomes. Both human and mouse studies provide evidence for placental and vascular adaptations to support fetal growth, which may be necessary to produce a healthy, adequately grown fetus in an adverse environment. Our human model suggests that placentas from women ≥40 years showed the most severe functional changes and this could guide future studies of AMA. Our mouse model of AMA provides evidence that failure of adaptation leads to severely compromised fetal growth and demise and is a valuable model for delineating the mechanisms causing increased rates of FGR and stillbirth with AMA.
We believe that the evidence presented here is sufficient proof of principal that advancing maternal age affects placental function. The human studies provide evidence of placental dysfunction even in pregnancies that end in apparently normal outcomes, implying underlying and currently unidentified stressors in pregnancy. The mouse model of AMA shows severe placental dysfunction consistent with other studies in FGR and enables analyses of adverse outcomes without the confounders present in human studies. We recognise comparisons between the mouse and human studies should be made cautiously, but emphasise that despite differences in the models, many of the same key markers of placental function are altered. These studies provide the first evidence for a link between advancing maternal age and placental dysfunction and identify important avenues for future study.

Methods: Human Studies
Tissue Collection. Women with uncomplicated pregnancies aged 20-30 years (control), 35-39 years, and ≥40 of age years delivering at St Mary's Hospital, Manchester were recruited to Manchester Advanced Maternal Age Study (MAMAS). Exclusion criteria were BMI <18.5 or ≥30, multiple pregnancy, fetal anomalies and pre-existing (e.g. cardiovascular disease, type I or II diabetes) or current maternal disease that may affect pregnancy (e.g. gestational diabetes mellitus, pre-eclampsia). Data regarding the outcome of pregnancy were recorded from maternal and infant case notes. Only IBCs 10-95 (calculated using GROW software (version 6.7.5.1, UK)) were included. Placentas (from NVDs or caesarean sections; n = 139) and/or myometrium (from caesarean sections; not from placental bed; n = 23) samples were collected (≤1 hour of delivery) after uncomplicated pregnancy. Placentas were weighed and sampled by standard protocols 77 . Placental centiles were calculated based on untrimmed placental weight, gestation and infant sex 78 . Maternal blood was collected following venepuncture (BD Vacutainer CAT) at 28 and 36 weeks gestation and centrifuged to extract serum fractions. Serum samples were stored at −80 °C until analysis.
As it was not possible to use the same placentas for all methodologies, placental function analyses were performed on sub-cohorts of placentas or maternal bloods based on power calculations for each methodology. Demographics were matched where possible to those of the overall study cohort.

Analysis of Placental Morphology by Histological Techniques.
Formalin fixed and paraffin embedded samples from four regions of each placenta (centre, two mid and edge) were used for studies of placental histology (n = 15/age group). Haematoxylin and Eosin staining was performed to assess presence of syncytial nuclear aggregates (SNAs) 79 . Immunohistochemistry using standard colorimetric detection was performed using antibodies for CD31 (Dako, Ely, UK), Ki67 (Dako) and Active Caspase 3 (Abcam, Cambridge UK) as previously described 77 . Ten random images (10x magnification) from each section were taken on an Olympus microscope. SNAs and villous vascularity (based on endothelial CD31 immunostaining) were quantified by manual counting, with normalisation for total villous area using ImagePro (Media Cybernetics). Proliferative and apoptotic indices were calculated following quantification of immunopositive Ki67 and Caspase 3 using HistoQuest (TissueGnostics, Vienna, Austria) for high throughput and unbiased analysis 80 . The number of DAB positive stained nuclei/cells per total number of nuclei detected was determined. All results are displayed as average values of the four regions per placenta.
Placental Function. Placental Amino Acid Transport Activity. Placental amino acid transporter activity was determined by sodium (Na + ) dependent uptake of 14 C-methylaminoisbutyric acid (meAIB; System A) and 3 H-Taurine (TauT) by freshly sampled placental fragments (n = 15/group) as described previously 25,81 and analysed as Na + -dependent uptake per milligram of villous tissue.
Circulating Placental Hormones. Concentrations of placental hormones (progesterone, PAPP-A, hPL and hCG (DRG Instruments GmbH, Germany), PlGF and sFlt (VEGF R1/Flt-1) (DuoSet ELISA, R&D Systems Inc. UK) were measured in a subset of maternal serum samples (n = 40/age group, nested case control study design, matched for demographic and biophysical characteristics) collected at 28 and 36 weeks gestation by ELISA by manufacturer's standard protocol.
Placental Chorionic Plate Artery Wire Myography. Placental chorionic plate artery (CPA) vascular reactivity was assessed by wire myography as previously described 82 . Arteries (gassed at 5% O 2 ) underwent a dose response curve of the thromboxane mimetic U46619 (10 −9 to 10 −6 M) to determine maximal constriction of each vessel. From this, an EC 80 dose of U46619 was applied followed by a dose response of the endothelium independent vasorelaxant sodium nitroprusside (SNP) (10 −11 to 10 −6 M). Vascular responses were plotted as averages of the four vessel fragments per placenta and expressed as either change in active effective pressure from baseline (KPa; for constriction) or as a percentage of maximal constriction (for relaxation) (n = 15/age group).

Myometrial Artery Function by Wire Myography.
Myometrial arteries were dissected from myometrial biopsies obtained during Caesarean deliveries (n = 7-14/age group). Arteries (gassed at 18-20% O 2 ) were tested for to vasoagonists were determined as above, with the additional assessment of responses to the endothelium dependent vasorelaxant bradykinin (BK; 10 −10 -10 −5 M). Data were presented as described above.

Methods: Animal Studies
Animal Husbandry. C57Bl/6J female mice (Envigo Ltd, United Kingdom) were housed on a 12 hour lightdark cycle at 21-23 °C with free access to food (Beekay Rat and Mouse Diet, Bantin & Kingman, Hull, UK) and water. Our AMA mouse model was based on established murine ageing profiles 55 and previous studies conducted by Van de Heijden 34 . Control mice (8-16 weeks old) and ageing (38-42 weeks old) mice were mated with C57Bl/6J males (3-6 months). Embryonic day (E) 0.5 of pregnancy was classified at the appearance of a copulation plug. At E17.5, females were sacrificed and fetal and placental tissues were harvested. Late fetal losses were classified as fully formed pups that fail to show signs of life at E17.5 (no detectable movement with one or combinations of: lack of perfusion, early signs of tissue deterioration or negligible detection of radiolabelled ligand post transport studies).
Fetal and Placental Measurements. Placentas and fetuses were harvested and weights were taken following laparotomy on the dam and expressed as wet weights and as fetal:placental weight ratios as a marker of placental efficiency. Fetal anthropometric measurements were taken by a single observer using thread and a ruler to obtain crown-rump length, abdominal circumference and head circumference measurements as previously described 83 . Fetal and placental weight frequency distribution curves were created as previously described 41 . 2.5 th , 5 th and 95 th percentile weights were calculated using the formula: (Z critical value x standard deviation)+ mean with Z critical values of -1.959, -1.649 and 1.649 respectively.
Unidirectional Maternofetal Transfer of 14 C-MeAIB or 3 H-TauT. At E17.5, a subset of dams were anaesthetised with an i.p injection of 1:1:2 mixture of fentanyl/fluanisone (Hypnorm, VetPharma Ltd., Leeds, UK), Midazolam (Roche, UK) and sterile H 2 0 (Braun medical Inc., Pennsylvania, USA) and the tail vein cannulated. The activity of system A transporter was assessed by injection of 140 µl (3.5 µCi) of 14 C-MeAIB as previously described 84 . This method was optimised for assessment of TauT activity by injection of 3 H-Taurine (3.5 µCi). Fetuses and placentas were collected within 3 minutes for quantification of radioactivity accumulation in each tissue. Data were presented as K mf (μl/min/g placenta) based on maternal plasma disappearance curves 84 . Uterine and Umbilical Artery Function. In a further cohort, culled dams were laparotomised and uterine arteries (UtA) and umbilical arteries (UbA) were dissected. Using wire myography, arteries (gassed at 20% O 2 for UtA and 5% O 2 for UbA) were assessed for vasconstriction by exposure to 10 −5 M phenylephrine (PE, UtA) or U46619 (UbA) for 20 minutes. Vasorelaxation was assessed by dose response of acetylcholine (Ach, 10 −10 -10 −5 mol/L) or SNP (10 −11 to 10 −6 M) to vessels pre-constricted with PE or U46610 (10 −5 M) respectively. Vascular responses were plotted as averages of the four artery fragments per litter and as a percentage of maximal constriction.
Statistics. All statistical analyses were performed using GraphPad Prism (Version 6.04, ©1995-2014 GraphPad Software, Inc.). Data was considered significant with p values less or equal to 0.05 (95% CI). Statistics performed on continuous data were Mann-Whitney or Kruskal-Wallis if data were not normally distributed or one-way ANOVA and two-way ANOVA as appropriate for normally distributed data with Dunn's multiple comparisons or Tukey's post hoc tests where appropriate. Categorical data were analysed by Fishers exact test. Human placental amino acid uptake over time was compared using linear regression. Placental hormones data were log transformed and presented as geometric means with 95% CI and analysed by one-way ANOVA and Tukey's post hoc test. Mouse data were displayed and statistically analysed as litter averages (N (number of litters), n (number of pups)). Study Approval. Written and informed consent was received from all human participants prior to inclusion in this study. All procedures have been approved by NRES Committee North West, (12/NW/0015) and North West REC (08/H1010/55). All methods involving human participants were carried out in accordance with the Human Tissue Act 2004 (HT Act) and accosiated regulations. All animal husbandry and experimental practices were carried out in accordance with the UK Animals (Scientific Procedures) Act 1986 under Home Office Licence 40/3385 and 70/8504. The Local Ethical Review Process of the University of Manchester approved all protocols. Data Availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.