Maternal intermittent fasting during pregnancy induces fetal growth restriction and downregulated placental system A amino acid transport in the rat.

During Ramadan, many pregnant Muslim women fast between dawn and sunset. Although the impacts of prolonged maternal intermittent fasting (IF) on fetal growth and placental function are under-researched, reported effects include reduced placental weight and birth weight. In this study, pregnant Wistar rats were used to model repeated cycles of IF on fetal development and placental function and to examine sex-specific effects. In the IF group, food was withdrawn daily from 17:00 to 09:00 over 21 days of gestation, while the control group received food ad libitum. Both groups had free water access. IF dams consumed less food, had significantly reduced weight compared to controls, with reduced plasma glucose and amino acids. Both fetal sexes were significantly lighter in the IF group, with reduced fetal plasma amino acids. Placental weights and morphology were unchanged. The profile of placental metabolites was altered in the IF group with sex-specific responses evident. Transplacental flux of 14C-methylaminoisobutyric acid (14C-MeAIB), a system A amino acid transporter substrate, was significantly reduced in both fetal sexes in the IF group. Sodium-dependent 14C-MeAIB uptake into isolated placental plasma membrane vesicles was unchanged. The gene expression of system A transporter Slc38a1, Slc38a2 and Slc38a4 was upregulated in IF male placentas only. No changes were observed in placental SNAT1 and SNAT2 protein expression. Maternal IF results in detrimental impacts on maternal physiology and fetal development, with changes in the placental and fetal metabolite profiles. Reduced placental system A transporter activity may be responsible for fetal growth restriction in both sexes.


Introduction
Fasting during the month of Ramadan is one of the five pillars of Islam. During this period, it is a religious requirement that healthy adult Muslims abstain from consuming food and drink from dawn to sunset for the duration of this holy month. From sunset to dawn, there are no restrictions on food and fluid intake [1]. Although this religious mandate excludes pregnant women who are allowed to postpone their Ramadan fast until after delivery, many still elect to partake in the fast with their families for spiritual and cultural reasons [2,3], as well as other motivational considerations [4,5]. It has been estimated that 50-90 % of pregnant Muslim women fast for at least part of the month of Ramadan [2, 4,5], with some choosing to extend their fast over the whole month [4,5]. By observing this fasting practice, pregnant women experience repeated cycles of intermittent fasting (IF). As the Islamic calendar follows the lunar cycle, fasting may last between 8-18 h a day depending on the season of the year in which Ramadan falls as well as geographical location [6]. Ramadan fasting is therefore associated with an alteration in the timing of food consumption as well as a change commonly in the quantity and quality of foods consumed [3,7]. Hence fasting pregnant women potentially expose their developing babies in utero to an altered nutrient and metabolic environment [6,8].
The impacts of sustained maternal IF on fetal development and the health of adult offspring are not understood fully and observations associated with Ramadan fasting are conflicting [9].
Differences in reported pregnancy outcomes may be related to the gestational timing when fasting occurred [9,10]: women are reported to participate in Ramadan fasting across each trimester of pregnancy [3,5,10]. Studies on the impact of Ramadan fasting on maternal and fetal health have, however, reported detrimental impacts for the mother and fetus, as well as effects on placental development. Adverse maternal outcomes reported include hyperemesis 5 gravidarum, dizziness, fatigue and nausea [3,11]. Fasting pregnant women have higher cortisol concentrations [12], hypoglycaemia [1,8] and diminished weight gain [13]. With regard to fetal health, studies have shown reduced fetal breathing movements [15], altered heart rate accelerations [15], and reductions in amniotic fluid index [16], fetal anthropometric measurements [16] and birthweight [12]. However, in a recent systematic review and metaanalysis, we did not find sufficient evidence to suggest that Ramadan fasting had an adverse effect on birthweight [9]. Other impacts on perinatal outcomes were not reported generally, so firm conclusions about the broader potential risks of fasting could not be made [9].
Notwithstanding this, exposure in utero to Ramadan fasting has been associated with lower cognitive test scores, learning difficulties and poorer academic performance in children and fewer employment hours worked in adulthood [3,10,17]. There is also an increased propensity to long-term health issues such as coronary heart problems and type 2 diabetes [2]. Impacts on placental growth and development have also been reported with placental weight and placental weight:birthweight ratio significantly reduced for both sexes in mothers who had fasted during Ramadan in the second and third trimesters [18].
The potential global impacts of Ramadan fasting whilst pregnant are considerable, with well in excess of 0.5 billion babies in each generation likely to be exposed in utero to a repeated cyclical pattern of maternal IF and ensuing associated maternal metabolic perturbations [9]. Hence, it has been estimated that more than 1 billion living Muslim adults could have been exposed to maternal fasting during Ramadan during their in utero development [5,10]. It is now well established that forms of maternal nutrient restriction or a sub-optimal maternal diet during pregnancy can have detrimental effects on fetal growth trajectory with enduring impacts on physiological function in the later life of the offspring [19]. Such developmental programming 6 phenomena have been recapitulated in several animal models in which restriction of protein or caloric intake during pregnancy has resulted in fetal growth restriction (FGR) with alterations in placental development, morphology and function [20,21]. Offspring from these models are predisposed to diseases in adulthood such as hypertension, diabetes and impaired cognitive function [20,21]. Further, sub-optimal maternal nutrition and restricted fetal growth are known to be associated with altered expression and activity of placental transporters [22,23], supporting the concept that the placenta acts as a nutrient sensor to modulate nutrient transfer to the fetus and regulate fetal growth [23].
Normal fetal growth is supported by the collective activity of several placental amino acid transporters [24]. The activity of system A, a sodium-dependent co-transporter that transports both essential and non-essential amino acids [24], is known to be crucial for optimal fetal growth [25][26][27]. Three isoforms of system A, sodium-dependent neutral amino acid transporter (SNAT) 1, 2 and 4, are expressed in the placenta of humans and rodents encoded by Slc38a1, Slc38a2 and Slc38a4 gene family subtypes respectively [28,29]. Inhibition of placental system A activity during rat pregnancy leads to diminished fetal weight [30], arguing for its integral involvement in supporting normal fetal growth.
It is particularly complex to replicate human maternal Ramadan fasting behaviour fully in an animal model, bearing in mind differences related to the duration of the fast, the variable adherence of women to the fasting period, the gestational timing of the fast and maternal food choices when the fast is broken. Furthermore, the molecular impact of regular and repeated episodes of maternal IF on fetal development across gestation, and the effects on placental function remain poorly defined. The investigative approach taken here was to simulate the repeated cycles of food abstinence, analogous to those experienced during Ramadan fasting, in Therefore, the aims of the study were to assess the effects of repeated cycles of maternal IF during pregnancy on maternal physiology, fetal growth and maternal and fetal metabolic profiles, as a prelude to investigating later enduring effects in the offspring. Changes in placental morphology were examined, as well as placental metabolism and placental system A amino acid transport function. As a number of studies have shown that placental responsiveness may differ between sexes [31], and that developmental diseases exhibit sex-specific differences [32], the sex-dependent specificity of fetal responses to maternal IF was also examined.

Ethical Approval
All experiments involving animals were conducted under the authority of a project licence (PPL 40/3646) issued in accordance with the UK Animals (Scientific Procedures) Act 1986. Local ethical approval was granted by the University of Manchester Animal Welfare and Ethical Review Body. All animal work was conducted at the University of Manchester.

Animals
Virgin female Wistar rats (Charles River, UK) weighing 250-275 g were mated with a Wistar male; gestational day 1 of pregnancy (GD1) was defined as the day that a vaginal plug was found (term is GD23). Females were housed singly and maintained under a constant 12 h:12 h dark/light cycle (19:00 dark cycle/07:00 light cycle) at 21-23 °C. Animals were allocated randomly to one of two dietary groups. To mimic the pattern of repeated fasting-feeding cycles adopted by pregnant women who fast over Ramadan, the maternal IF group (IF, N = 36) was subjected to food restriction for 16 h overnight, predominantly over the active dark cycle phase, with food (standard rat chow diet, (BK001 (E) SDS Rodent Breeder and Grower, LBS Biotech, Redhill, UK) removed at 17:00 and returned at 09:00 daily, from GD1 onwards until the end of the experiment on GD21. The control group (C, N = 36) was fed standard rat chow ad libitum throughout pregnancy. Both groups had free access to water at all times. Food and water intakes and the dam's body weight were recorded daily. Dams were excluded from the study when a vaginal plug was absent preventing the accurate determination of the start of pregnancy (N=6), or when the IF regimen resulted in a lack of pregnancy weight gain with greater than 15 % maternal weight loss (N=3).

Collection of blood and placenta
At GD21 between 09:00 and 10:00, the pregnant dams were anaesthetised with isoflurane by inhalation (4% in oxygen at 2 L/min), and maternal blood was collected by cardiac puncture.
Dams were killed by cervical dislocation after which placentas and fetuses were harvested rapidly. Individual fetal blood samples were collected from an axillary vessel incision using heparinised capillary tubes. Following centrifugation, maternal and fetal plasma were stored at -80 °C until further analysis. Placentas were stored in RNAlater (Sigma Aldrich, UK) for gene expression analysis, snap-frozen for biochemical analysis including the metabolomic assays and stored at -80 °C, or fixed in 10% neutral buffered formalin for histological analyses. In some instances, the junctional (JZ) and labyrinth zones (LZ) of the placenta were dissected separately and each frozen rapidly. Placentas and fetuses were selected randomly for subsequent analysis, dependent on fetal sex, and analyses were not performed in a blinded manner with respect to dietary intervention.

Fetal anthropometric measurements
In a set of dams (N = 5 control and N = 7 IF litters), fetal anthropometric measurements of crown-rump length, abdominal circumference and head circumference were performed as described previously [33].

Fetal sex determination
For fetal sex determination, genomic DNA was isolated from fetal tail tips using a DNeasy kit (Qiagen, Crawley, UK) according to the manufacturer's instructions. PCR was performed using primer sequences specific for rat Sry gene encoding the sex-determining region Y protein (sequences 5'→3'; forward primer: CATCGAAGGGTTAAAGTGCCA; reverse primer: Downloaded from http://portlandpress.com/clinsci/article-pdf/doi/10.1042/CS20210137/911701/cs-2021-0137.pdf by guest on 21 May 2021 10 ATAGTGTGTAGGTTGTTGTCC), as described previously [34]. The presence or lack of a Sry gene amplification product, as visualised by agarose gel electrophoresis, was designated as male (M) or female (F) respectively.

Plasma glucose and hormone analyses
Glucose concentration in maternal and fetal plasma was determined using a commercial kit (GAHK-20; Sigma Aldrich, UK). Insulin and glucagon concentrations in maternal and fetal plasma were measured by the Ultrasensitive Rat Insulin and Rat Glucagon ELISA kits respectively (10-1251-01 and 10-1271-01 respectively, Mercodia, Uppsala, Sweden). All assays were performed according to the manufacturer's instructions.

Plasma amino acid concentration
The concentration of amino acids in maternal and fetal plasma was measured by reversed-phase high pressure liquid chromatography (HPLC). Maternal (50 μL) or fetal plasma (10 μL) was deproteinised with 10 % trichloroacetic acid, and amino acid concentrations measured using an Agilent 1100 HPLC (Agilent Technologies, Stockport, UK) fitted with a Gemini 50 x 4.6 mm column (Phenomenex, Macclesfield, UK) and an Agilent fluorescence detector, as described previously [35].

Placental morphology
Sections (5 μm) of formalin-fixed, paraffin-embedded placentas were stained with haematoxylin and eosin and scanned by the Bio-imaging Facility (University of Manchester) and the images processed by Panoramic Viewer software (3DHISTECH Ltd, Budapest, Hungary). JZ and LZ areas were quantified using ImageJ software analysis from two representative placental sections (6 placentas per sex from 3 litters per dietary group).

Tissue glycogen content
Glycogen content of the placental JZ and fetal liver was measured using a glycogen assay kit (MAK016; Sigma Aldrich, UK), according to the manufacturer's instructions.

Placental metabolomics
Placentas from control and IF dams at GD21 (n = 6 placentas per fetal sex from N = 6 litters per dietary group) were harvested and a placental-quarter was flash-frozen immediately and stored at -80 °C prior to metabolomic analysis at the University of Birmingham. 12 microcentrifuge tubes and dried under a nitrogen gas stream (room temperature) prior to storage at -80 C until analysis. An additional homogenisation tube, without placental tissue, was employed to generate a blank control sample. Quality control (QC) samples were prepared by pooling all remaining extraction supernatants followed by transfer of 450 L aliquots to 2 mL microcentrifuge tubes followed by drying under a nitrogen gas stream (room temperature) prior to storage at -80 C until analysis. Analysis order was composed of 10 QC sample injections for system conditioning followed by a QC sample injection every 6th injection with two QC sample injections at the end of the analytical run. Tissue extracts were analysed in a random order. Blank samples were analysed as injection 6 and as the last injection of the analysis batch.

Data processing and statistical analysis
UHPLC-MS raw data profiles were first converted into a NetCDF format within the Xcalibur software's File Converter program. Each NetCDF based three-dimensional data matrix (intensity x m/z x time -one per sample) was converted (or deconvolved) into a vector of peak responses, where a peak response is defined as the sum of intensities over a window of specified mass and time range (e.g. m/z = 102.1 and time = 130 s). In this experiment the deconvolution was performed using the freely available XCMS software [36]. Data were exported from XCMS as a .csv file for further data analysis. The quality of data was assessed applying QC data with all metabolite features with a relative standard deviation (RSD) of > 20% for QC samples being removed from the dataset prior to data analysis [36]. The data for each sample were normalised (as a percentage) to the total peak area for all metabolites in the sample. Metabolite annotation was performed applying the PUTMEDID_LCMS workflow [37] at level 2 or 3 (putatively annotated compounds) according to MSI reporting standards [38]. The processed data were analysed in R applying the unsupervised multivariate principal components analysis (PCA) and Gene expression was normalised to Ywhaz (tyrosine 3-monooxygenase/tryptophan 5monooxygenase activation protein, zeta polypeptide) as a reference gene; Ywhaz mRNA expression was stable across dietary groups and fetal sex (data not shown).

System A protein expression
Rat placental membrane vesicles, prepared from the maternal-facing plasma membrane of syncytiotrophoblast layer II were isolated from placentas at GD21 using one cycle of Mg 2+ precipitation, as described previously [39,40]. Vesicle membrane purity was measured as enrichment of alkaline phosphatase activity in the vesicle suspension relative to that in placental homogenate [39,40]. Protein expression of SNAT1 and SNAT2 in rat placental membrane vesicles was determined by immunoblotting. For SNAT1, samples were heated at 60 °C for 5 min in reducing buffer (4 % SDS, 0.004 % bromophenol blue, 20 % glycerol, 10 % β-mercaptoethanol in 0.125 M Tris-HCl pH 6.8); for SNAT2, samples were added to a non-boil reducing buffer (8 M Blots were blocked using Odyssey Blocking Buffer (LiCOR Biotechnology, UK) for 1 h and then probed with either rabbit affinity-purified anti-SNAT1 (1.5 µg/mL) or anti-SNAT2 (1.58 µg/mL) antibodies [41] diluted in Odyssey Blocking Buffer with 0.2 % Tween 20 overnight at 4 °C, followed by a further 2 h antibody incubation at room temperature. A blocking peptide was included as negative control. Following washes with PBS/0.1 % Tween, blots were incubated with 1:20,000 anti-rabbit secondary antibody (IRDye ® 800CW, green, Li-COR Biotechnology, UK).
System A activity in isolated rat placental vesicles was measured as the Na + -dependent uptake of 14 C-methylaminoisobutyric acid (MeAIB), a non-metabolisable specific substrate for the system A amino acid transporter, as described previously [40].

Statistics
Data are presented either as box (with median) and whisker plots (whiskers represent 5 th and 95 th centiles) or as litter mean ± SEM as appropriate, with N = number of dams/litters and n = number of individual fetuses. The Shapiro-Wilk test was used to evaluate the normal distribution of the data. Data were analysed by two-way ANOVA (with repeated measures where appropriate) and Tukey tests or Kruskal-Wallis and Dunn's multiple comparison tests, or unpaired t-tests. Litter size was analysed by Fisher's exact test. Fetal weight frequency distribution curves were compared by non-linear regression (Gaussian distribution), with the 5 th percentile weight calculated as described previously [42]. Data were analysed using SPSS

Maternal food intake and weight gain
Dams in the IF group consumed 30 ± 1 % less food at all stages of pregnancy compared with control dams, despite having free access to food daily over a 8 h period (from 09:00 to 17:00) (P < 0.001, Figure 1A). Pregnant dams of both control and IF groups exhibited an incremental weight gain as pregnancy progressed (P < 0.001, Figure 1C), sustained by a daily food intake that remained relatively stable throughout gestation when normalised to maternal body weight ( Figure 1A). Daily water consumption was comparable between diet groups ( Figure 1B).
Although pregnant dams in both groups gained weight steadily throughout pregnancy, daily weight gain in IF dams was significantly lower compared to control dams from GD8 onwards (P < 0.001, Figure 1C). All dams were weighed routinely at 17:00 before food was removed from the IF group. However on GD21, the day of harvest, dams were weighed at 09:00 prior to sacrifice, accounting for the apparent fall of 19 ± 1 g in IF dam body weight at GD21 following the overnight fast ( Figure 1C). This also implies that there was some recovery of the dams' overnight weight loss achieved by the opportunity to feed again following the re-introduction of food in the morning to the IF group.
Normalisation of maternal organ weights to 100 g dam body weight (after subtraction of uterine content weight) showed that variable effects were associated with maternal IF. Maternal liver weight in IF dams was significantly reduced (P < 0.01), while kidney weight was significantly increased, compared with the control dams (P < 0.05); heart weight was not affected (Supplementary Table 2).

Glucose concentration and regulatory hormones in maternal and fetal plasma
Maternal IF had differing effects on circulating glucose and associated regulatory hormone concentrations in dams and fetuses. Plasma glucose concentration was reduced by 33 ± 3 % in IF dams compared with controls (P < 0.001, Table 1). This was associated with a significant reduction in plasma glucagon concentration (P < 0.01, Table 1); insulin concentration was unaffected in IF dams. In contrast, plasma glucose and glucagon concentrations did not differ between IF and control fetuses, whereas the plasma insulin concentration was significantly lower in IF fetuses (P < 0.05, Table 1).

Fetal glycogen content
To investigate whether the maintained fetal glucose concentration in the IF group was underpinned by catabolism of fetal glycogen as a potential source of mobilisable glucose, the glycogen content of the placental JZ, which harbours clusters of glycogen cells, and fetal liver was measured. Placental glycogen content was unaltered in IF fetuses compared with controls (Table 2). However, fetal liver glycogen content was reduced significantly in both sexes of IF fetuses compared with control fetuses (P Male < 0.01, P Female < 0.05, Table 2).

Fetal and placental weights
There were no differences in litter size or fetal sex ratio between the IF and control groups (Table 3). Neither was the number of fetal resorptions (as % of total number of implantations) different between groups (2.50 ± 1.06 and 1.26 ± 0.43 % for IF and control respectively).
However, the mean fetal weight per litter was significantly reduced by 15 ± 1 % in the IF group compared with the controls (P < 0.001, Figure 2A). Male fetuses were heavier than females in  Figure 2A). Plotting fetal weight frequency distribution curves revealed that 61 -67 % fetuses in the IF group fell below the 5 th centile of the control population ( Figure 2B, C), taken as a clinical indicator of FGR [46]. In contrast, placental weight (as a proxy of placental growth) was unaffected by exposure to maternal IF ( Table 3). The fetal:placental weight ratio (as an index of placental transport efficiency [43]), was significantly reduced in both sexes of the IF group (P Male < 0.01, P Female < 0.05, Table 3), suggesting an adverse effect of maternal IF on placental transport function.
The growth restriction in IF fetuses was accompanied in both sexes by a significant reduction in crown-rump length (P < 0.01) and head circumference (P < 0.05), while abdominal circumference was unaltered (Table 3). Fetal organ weights relative to body weight did not differ between groups (Table 3). However, the fetal brain:liver weight ratio was significantly increased in both sexes of IF fetuses compared with controls (P < 0.01, Table 3).

Sex-related differences in placental morphology
The gross morphology of the placenta was not affected by exposure to maternal IF ( Figure 3A), with organisation of the placental cellular layers appearing normal, with clear discrimination of the JZ and LZ. There were no differences observed in relative area of the JZ and LZ for either sex between the two dietary groups ( Figure 3B, C). However, sex-dependent differences were apparent independent of diet; placentas of female fetuses of both dietary groups had a proportionately thicker JZ, whilst male fetuses had a thicker LZ ( Figure 3A-C). Hence females of both dietary groups had a significantly higher relative ratio of JZ:placental total area compared to their respective male counterparts, whereas the converse was apparent as regards the relative ratio of LZ:placental total area ( Figure 3B, C).

Maternal and fetal plasma amino acid concentrations
The amino acid concentrations in maternal and fetal plasma on GD21 of pregnancy are shown in Table 4. Irrespective of dietary regimen, amino acid concentrations in fetal plasma were significantly higher compared with maternal plasma (P < 0.05), with a fetal:maternal concentration ratio of variable magnitude and greater than a value of 1 across all amino acids, reflective of amino acid concentrative capacity within the fetal compartment ( Supplementary   Figure 1). In the case of IF animals, the fetal:maternal amino acid concentration ratio was significantly increased for several amino acids including neutral aromatic (tyrosine, phenylalanine and tryptophan), neutral non-aromatic (asparagine, alanine, glutamine, serine and threonine) and cationic (lysine and arginine) amino acids (P < 0.05, Supplementary Figure 1). This increased fetal:maternal concentration ratio was driven in part by significant reductions in maternal tyrosine, asparagine, alanine, serine, threonine and arginine concentrations (P < 0.05, Table 4). Other amino acids that demonstrated a significantly reduced concentration in IF maternal plasma (histidine, methionine and valine, Table 4) were accompanied by a significant reduction in fetal plasma concentration (P < 0.05, Table 4), resulting in an unchanged fetal:maternal concentration ratio (Supplementary Figure 1). Interestingly, there were two amino acids, isoleucine and aspartate, that demonstrated a significant reduction in IF fetal plasma, yet were unaltered in maternal plasma (Table 4). A subset of amino acids (glutamine, glycine, leucine, tryptophan, phenylalanine, lysine and glutamate) revealed no significant changes in either maternal or fetal plasma concentration in response to maternal IF compared to control (Table 4). Therefore, there were differential effects of maternal IF on amino acid concentrations within both maternal and fetal plasma, with divergent trends across the same subclasses of amino acids. in control fetuses lysine, arginine and tyrosine were significantly lower in females compared with males (P < 0.05, Table 4) and in IF fetuses, the aromatic amino acids phenylalanine, tryptophan and tyrosine were significantly lower in females compared with males (P < 0.05, Table 4). Maternal IF diet was also associated with sex-specific changes in fetal amino acid concentration; threonine, arginine and valine were reduced significantly in IF male fetuses compared to control males (P < 0.05, Table 4), while serine was reduced significantly in IF female fetuses compared to control females (P < 0.05, Table 4).

Placental system A amino acid transport activity
The selective reduction in fetal plasma amino acid concentrations of the IF group, together with the particularly marked reduction in fetal alanine concentration (of 26 % in male and 23 % in female fetuses, respectively), led us to postulate that the IF dietary regimen caused a reduction in placental amino acid transporter activity and compromised placental nutrient transport capacity, leading to FGR. As alanine is a preferred amino acid substrate for the system A amino acid transporter [24], we hypothesized that placental system A activity would be reduced by the IF dietary regimen. To investigate this hypothesis in vivo, the unidirectional maternofetal clearance of 14 C-MeAIB ( MeAIB K mf ), a non-metabolisable specific substrate for system A [33], was measured across the placenta. As shown in Figure 4A, there was a significant reduction in MeAIB K mf across the placentas of both IF male and female fetuses by 37 ± 9 % (P < 0.05) and 41 ± 8 % (P < 0.01) respectively, compared with control litters. Measurement of system activity A in vitro, as the Na + -dependent uptake of 14 C-MeAIB into isolated placental membrane vesicles from  Figure 4C). However there were no significant differences in the rate of 14 C-MeAIB uptake between dietary groups ( Figure 4C), and at 60 s the amount of accumulated tracer was comparable between groups ( Figure 4B).

Placental expression of Slc38a1, Slc38a2 and Slc38a4 genes and SNAT protein expression
Based on our observation that transplacental system A activity was reduced when dams were exposed to a repeated pattern of maternal IF ( Figure 4A), we examined the placental expression of genes that encode the SNAT1, 2 and 4 protein isoforms which mediate system A activity.
Expression of Slc38a1, Slc38a2 and Slc38a4 was significantly increased in IF male placentas only compared to control, with no differences between groups in female fetuses, highlighting a sexspecific induction in response to IF diet in male fetuses only (P < 0.05, Figure 5A-C).
To assess whether the expression and functional changes observed in the IF group were accompanied by altered SNAT protein expression, we examined SNAT1 and SNAT2 expression in rat placental vesicles, applying the reasoning that these SNAT isoforms were most likely to be implicated functionally as SNAT1 and SNAT2 have shared functional transport characteristics and a much higher affinity for MeAIB as a substrate relative to SNAT4 [44]. For SNAT1, an intense immunoreactive species at 52 kDa was observed in all samples, whereas for SNAT2, two distinct bands at approximately 60 and 150 kDa were visualized (Supplementary Figure 2A, C).

Quantitation of immunoreactive species intensity revealed that expression of SNAT1 (52 kDa)
and SNAT2 (both 60 and 150 kDa species) were similar between the control and IF groups for both sexes (P > 0.05, Figure 6A-C). All immunoreactive signals were abolished in the absence of primary antibody (data not shown) or following pre-absorption of antibody with excess blocking  Figure 2B), allowing it to be used to normalise SNAT protein expression ( Figure 6).

Placental metabolomics
Bearing in mind the differences observed for glucose and amino acid concentrations in both the maternal and fetal compartments in response to a maternal IF diet (Tables 1 and 4), we postulated that this dietary regimen would be associated with an altered placental concentration of other metabolite classes not investigated so far. Table 5 shows that 22 metabolites, of different metabolite classes, were significantly altered in the placentas of both sexes of the IF group compared to controls, with the majority (16 metabolites) demonstrating an increased placental concentration, the most striking being ophthalmic acid. However, in contrast, five metabolites related to fatty acid and lipid metabolism were significantly lower in both fetal sexes of the IF group compared to control. There was a significant increase in aromatic amino acids such as tryptophan, tyrosine and phenylalanine in the IF male placentas compared to IF females (Table 6); these amino acids were also significantly higher in the plasma of IF males as compared to IF females (Table 4). In addition, the metabolite class of trisaccharides (which are catabolized to free hexose sugars) was also elevated in IF male placentas (Table 6). Collectively, the placental metabolites altered in association with the IF dietary regimen, as listed in Tables 5 and 6, suggest that divergent metabolic pathways are affected in the placenta, with evidence of amino acid, fatty acid, phospholipid, steroid and nucleotide metabolism being particularly perturbed.

DISCUSSION
The primary aim of this study was to recapitulate in an animal model aspects of maternal IF that occur during Ramadan fasting, in order to elucidate the impact of repeated dietary 'fastingfeeding' cycles during pregnancy on placental function and fetal development. The model was not designed to mimic Ramadan fasting in humans fully, in so much as the pregnant rats were deprived of food for a period of 16 h each day throughout 21 days of gestation and had free access to water. This contrasts with religious practice during Ramadan, where daily repeated fasting lasts for one lunar month and involves abstaining from both food and water intake during daylight hours.
In this rat model, in order to mimic food abstinence adopted by humans during Ramadan fasting over daylight active hours, food was withdrawn during the rat's active phase over the dark cycle.
As already outlined, the duration of daily Ramadan fasting undertaken by humans can be variable within the range of 8-18 h dependent on geographical location and the season in which Ramadan falls. Our approach in this study was to examine a fasting interval that approached the maximal time of fasting in humans, to understand more fully the impact that daily prolonged maternal fasting may have. We elected not to simultaneously withhold water, as is the practice in humans during Ramadan fasting, as we sought to identify the effects of intermittent food abstinence alone without any confounding effects of dehydration. We acknowledge, however, that water deprivation may have further impacts upon fetal developmental and placental function.
As already mentioned, women may partake in the Ramadan fast at any stage of pregnancy.
However because little is known about the impact of IF on the fetus, we opted to fast rats 26 throughout pregnancy in order to gain a more detailed understanding of the consequences of fasting across all phases of fetal growth and development. Nonetheless, we acknowledge that the dietary challenge imposed on the rats in this study is greater than that experienced by a pregnant woman fasting during Ramadan. We further recognise that we were unable to model the changes in both the quantity and quality of foods consumed after the daily fast in broken in humans, not least because the type of food eaten varies between different communities.
Notwithstanding these limitations, the study allowed us to evaluate the impact of maternal food restriction alone during pregnancy on maternal physiology, fetal development, placental development and function. It also enabled us to explore how fetal sex might influence phenomena that associate with an altered maternal dietary intake and a changed in utero nutrient environment.
The dietary regimen of maternal IF imposed was tolerated well by the pregnant rat dams, with no observed impact on either fetal loss or litter size, in agreement with others using a comparable Ramadan-type fasting regimen in pregnant rats [45]. As expected, maternal food intake and maternal gestational weight gain were reduced significantly following the regular withdrawal of food overnight, as observed previously [45]. Although we did not measure lean body mass, it is interesting to note that there was no consistent pattern of change in the relative weight of different maternal organs in the IF dams. These observations are in broad agreement with previous studies where pregnant rats were exposed to food-restriction for 19 h per day [46]. Dams subjected to IF consumed ~30 % less food than controls. The failure to compensate when food was made freely available in the morning was accounted for by the dams' inclination to sleep following their morning feeding session and their observed relative inactivity thereafter.
Hence overall food consumption was insufficient to sustain normal weight gain during pregnancy, in common with other other food-restricted models in pregnant rats [45,47].
Although caloric deficiencies and diminished maternal weight gain in pregnant women undertaking Ramadan fasting have been reported previously [13], others emphasize that caloric intake may in fact increase during Ramadan due to the consumption of calorie-dense foods when the fast is broken [48].
Fetal weights of male rat fetuses were heavier than females in the control group at GD21, in agreement with others [49], and consistent with findings in human newborns [18]. However, in the IF group this sex-divergence in fetal weight was not apparent suggesting that normal fetal growth trajectory patterns had been disrupted. Indeed, fetal weight of both sexes in the IF group was reduced significantly at GD21 compared to the control group, with a substantial proportion of fetuses exhibiting FGR. Additional evidence of altered fetal growth dynamics is provided by the increase in the brain:liver weight ratio of IF fetuses. This is a marker of fetal 'brain-sparing' and an index of intrauterine growth restriction, reflective of a redirection of nutrients to vital organs such as the brain at the expense of other organs [50]. Whilst this action serves to try to protect the brain, it is associated with impaired neurocognitive function later in development [51]. This is of interest in light of the lower cognitive test scores, learning difficulties and poorer academic performance reported in children exposed in utero to Ramadan fasting [3,17].
The reduced maternal glucose concentration observed in IF dams accords with previous studies [45], yet a normal glucose concentration was maintained in the fetuses of the IF group. As control dams were not fasted prior to collecting blood samples we cannot rule out the possibility that the difference in plasma glucose concentration is due, in part, to prior food intake. However, it is established that fasting regimens of a similar pattern to that adopted in this study [45] or as short as 24 h [52] in pregnant rat dams result in significantly reduced maternal glucose concentrations, so our data are in agreement with these previous reports.
The observation that fetal liver glycogen content was reduced, whereas placental glycogen content was preserved in IF fetuses, is consistent with this being achieved primarily through catabolism of fetal liver glycogen, rather than a change in the contribution from placental glycogen stores. During normal rat pregnancy, fetal plasma glucose concentration is lower, and insulin concentration higher, than in the maternal circulation [52], as replicated here. This may reflect a high fetal glucose demand, with the higher fetal insulin concentration promoting glucose utilisation by fetal insulin-sensitive organs. It is therefore possible that the fetal hypoinsulinaemia observed in the IF group might also contribute to the fetal growth impairment.
Along with maternal undernutrition, placental insufficiency can contribute to impaired fetal growth and accompanying FGR [33]. Morphological differences in the placental exchange barrier, reduced placental blood flow arising from placental maldevelopment with impaired vascular development, together with dysregulation of key placental nutrient transporters, may all contribute to drive different patterns of fetal growth [53]. Maternal IF did not affect placental weight, and consistent with this, placental junctional and the labyrinth areas (as a percentage of placental total area) of IF fetuses were unaltered compared to controls, suggesting a lack of overt morphological change. Interestingly, male fetuses had a relatively larger labyrinth area compared to females, consistent with a greater exchange area to support fetal growth and compatible with a heavier fetal weight in male fetuses. However, in humans, the placental weight of babies in utero during Ramadan fasting is significantly reduced in both sexes [9], but only when maternal fasting occurred during the second and third trimesters of pregnancy [18]. By definition, the genesis of a FGR phenotype in the IF group must reflect a reduced net placental nutrient flux. Therefore the FGR phenotype seen is most likely to be attributable to dysregulated placental function. Indeed, the collective data are consistent with maternal IF eliciting effects both on placental nutrient transport and metabolic function. The fetal:placental weight ratio, a proxy of placental transport efficiency [43], was reduced significantly in both fetal sexes of the IF group. Additionally, the concentration of several fetal plasma amino acids (as a reflection of placental transport capacity) were altered, including histidine, isoleucine, methionine, threonine, valine, alanine, serine, arginine and aspartate; sometimes independent of a changed maternal concentration (e.g. isoleucine, aspartate). Conversely, fetal plasma asparagine was unaltered in the face of a profound reduction in maternal plasma concentration.
However, on the whole, a significant reduction in both essential and non-essential amino acids in pregnant IF dams was accompanied by reductions in fetal plasma (albeit with some sexspecific differences). Similar effects have been observed previously in other rodent fasting models of shorter duration [52]. Nonetheless, fetal amino acid concentrations were maintained at higher concentrations than their mother's, suggesting that fetal concentrative capacity was sufficient to maintain a positive fetal:maternal concentration ratio for all amino acids in both dietary groups.
During Ramadan fasting pregnant women show a significant reduction in fasting blood alanine [54], in agreement with the profound reduction in maternal (~54 %) and fetal (~25 %) alanine concentration observed in the IF group. Others have also shown that fasting in rodents causes a striking reduction in the concentration of alanine, a major gluconeogenic substrate in both maternal and fetal compartments [52], which could reflect increased gluconeogenesis, or alternatively, a greater flux of pyruvate into acetyl-CoA for fatty acid synthesis and/or ATP Downloaded from http://portlandpress.com/clinsci/article-pdf/doi/10.1042/CS20210137/911701/cs-2021-0137.pdf by guest on 21 May 2021 synthesis via the TCA cycle, which would be required for the brain 'sparing' process as lipids are a major component of the brain. Alanine is a high affinity substrate for system A amino acid transporter subtypes [24] and its activity is crucial for fetal growth; impairment of placental system A activity leads to FGR in humans and in rodent models [22,25,27,30]. Our observation that the maternofetal clearance of 14 C-MeAIB across the placentas of IF fetuses in vivo was reduced significantly in both sexes compared with that of control fetuses, is in keeping with the postulate that diminished placental system A activity is associated with FGR. Surprisingly, in contrast, when system A-mediated uptake of 14 C-MeAIB into isolated plasma membrane vesicles was measured in vitro, no differences were observed between dietary groups. This contrasts with other rodent models displaying a FGR phenotype, where these two experimental approaches to measure placental system A activity show good correspondence [33]. A possible explanation for the discordance observed in placental system A activity between these two experimental approaches is that the reduction in maternofetal clearance of 14 C-MeAIB arises downstream of transport across the maternal-facing plasma membrane of syncytiotrophoblast layer II from which the plasma membrane vesicles are derived, across another plasma membrane within the syncytial layers of the labyrinth villi. Alternatively, regulatory factors or prevailing driving forces in vivo may modulate transplacental amino acid flux mechanisms.
As pregnancy advances, fetal growth is supported by increased placental system A transporter expression and activity [55,56]. As in human placenta [28], rat placenta expresses all three genes, Slc38a1, Slc38a2 and Slc38a4, encoding the SNAT1, 2 and 4 isoforms of system A [29], also confirmed here. Interestingly, maternal IF elicited significant increases in Slc38a1, Slc38a2 and Slc38a4 gene expression solely in the placentas of male fetuses compared to controls; no changes were observed in IF females. Likewise, in other dietary models, placental Slc38a2 and Downloaded from http://portlandpress.com/clinsci/article-pdf/doi/10.1042/CS20210137/911701/cs-2021-0137.pdf by guest on 21 May 2021 an adaptive response elicited by male fetuses to a compromised maternal amino acid environment to try to stimulate fetal growth investment [31]. However, this up-regulation in mRNA in IF male placentas was not accompanied by increased SNAT1 and SNAT2 expression at the protein level at GD21. Nor was it effective in overcoming the reduction in SNAT-mediated transplacental system A activity observed in IF fetuses at this gestational stage. Divergent trends between SNAT gene and protein expression and system A transporter activity have been reported previously [58]. However, some dietary models show an allied reduction in SNAT protein expression and placental system A transport activity [59], highlighting complexity in the regulation of SNAT-mediated placental system A activity.
The demonstration of higher placental aromatic amino acid concentrations (phenylalanine, tyrosine and tryptophan) in IF males compared with IF females further suggests sexual dimorphism in placental transport or fetal metabolism. The increased placental ophthalmic acid concentration of IF fetuses suggests increased oxidative stress and altered glutathione metabolism [60], and is synthesized from gamma-glutamyl-gamma-aminobutyrate which is also increased in placentas of IF fetuses. N-(4-guanidinobutyl)-4-hydroxycinnamide and gammaglutamyl-gamma-aminobutyrate are both involved in arginine and proline metabolism and this is of interest in relation to the reduced plasma arginine concentrations observed in male fetuses.
In summary, although it is difficult to model all aspects of Ramadan fasting during pregnancy precisely, this study has demonstrated that a principal feature of Ramadan fasting, namely repeated cycles of maternal IF, has a negative impact on maternal physiology, as reflected by a concentrations. In contrast, IF fetuses were able to maintain normal glucose and glucagon concentrations while insulin concentrations were reduced. The normal fetal glucose concentration was coupled with a reduction in fetal liver glycogen content. Fetuses exposed in utero to IF exhibited growth restriction but the fetal brain seem to be 'spared', as evidenced by an increased fetal brain:liver weight ratio. Fetuses exposed to IF suffered from reductions in certain essential and non-essential amino acids in parallel with their mothers. Interestingly, sexspecific differences emerged in both groups, with females having a lower concentration of certain amino acids compared to males. Likewise, aromatic amino acids and trisaccharides were lower in the placentas of IF female fetuses compared to IF male fetuses. Sub-optimal maternal nutrition induced by IF was associated with a reduction in transplacental 14 C-MeAIB flux measured in vivo. This alteration in system A activity was not observed in the maternal-facing plasma membrane of syncytiotrophoblast layer II in vitro, suggesting that placental efflux may be rate-limiting. However, regulation of placental transport in vivo by circulating factors may also exert an influence. Placental weight near-term was unaltered by IF and this was associated with unchanged junctional and labyrinth area ratios, although sex-specific differences appeared in both dietary groups, with males having higher labyrinth and lower junctional relative ratios compared with females.
These data have provided evidence that maternal IF has physiological effects within the maternal, placental and fetal compartments. This raises the possibility that pregnant women who observe the practice of Ramadan fasting may experience adverse effects on maternal