Changes in the Metabolic Footprint of Placental Explant-Conditioned Medium Cultured in Different Oxygen Tensions from Placentas of Small for Gestational Age and Normal Pregnancies

Article Outline

• Abstract
• 1. Introduction
• 2. Methods and materials
  • 2.1. Culture of placental villous explants
  • 2.2. Ultra Performance Liquid Chromatography–Mass Spectrometry (UPLC–LTQ/Orbitrap) analysis
  • 2.3. Statistical analysis
  • 2.4. Metabolite identification
• 3. Results
  • 3.1. Global changes in the metabolic footprint of conditioned culture medium from different O2 tensions
  • 3.2. Characterising the differences between the metabolic footprint of conditioned culture medium from SGA and control pregnancies at each O2 tension
  • 3.3. Relative changes in the normoxic metabolic footprint due to culturing at 1% and 20% O2 tensions
    • 3.3.1. Changes in relative metabolite levels when culture conditions were changed from 6% to 1% O2
    • 3.3.2. Changes in relative metabolite level when culture conditions were changed from 6% to 20% O2
• 4. Discussion
  • 4.1. The global differences in the metabolic footprint of the conditioned culture media were primarily dependent on the O2 tension the samples were cultured under
  • 4.2. The metabolic footprints of SGA and control conditioned culture media were different when compared at each O2 tension in isolation
  • 4.3. Characterisation of the metabolic changes in the conditioned culture media suggests that changing from normoxic to hypoxic culturing conditions significantly increases disruption in SGA metabolism when compared with normal placenta
  • 4.4. Characterisation of the metabolic changes in the conditioned culture media suggests that changing from normoxic to hyperoxic culturing conditions has only a small effect on the metabolic differences between SGA and normal placenta
• Acknowledgements
• Appendix. Supplementary data
• References
• Copyright

Abstract 

Being born small for gestational age (SGA) confers significantly increased risks of perinatal morbidity and mortality. Accumulating evidence suggests that an SGA fetus results from a poorly perfused and abnormally developed placenta. Some of the placental features seen in SGA, such as abnormal cell turnover and impaired nutrient transport, can be reproduced by culture of placental explants in hypoxic conditions. Metabolic footprinting offers a hypothesis-generating strategy to investigate factors absorbed by and released from this tissue in vitro. Previously, metabolic footprinting of the conditioned culture media has identified differences in placental explants cultured under normoxic and hypoxic conditions and between normal pregnancies and those complicated by pre-eclampsia. In this study we aimed to examine the differences in the metabolic footprint of placental villous explants cultured at different oxygen (O₂) tensions between women who deliver an SGA baby (n = 9) and those from normal controls (n = 8). Placental villous explants from cases and controls were cultured for 96 h in 1% (hypoxic), 6% (normoxic) and 20% (hyperoxic) O₂. Metabolic footprints were analysed by Ultra Performance Liquid Chromatography coupled to an electrospray hybrid LTQ-Orbitrap Mass Spectrometry (UPLC-MS). 574 metabolite features showed significant difference between SGA and normal at one or more of the oxygen tensions. SGA explant media cultured under hypoxic conditions was observed, on a univariate level, to exhibit the same metabolic signature as controls cultured under normoxic conditions in 49% of the metabolites of interest, suggesting that SGA tissue is acclimatised to hypoxic conditions in vivo. No such behaviour was observed under hyperoxic culture conditions. Glycerophospholipid and tryptophan metabolism were highlighted as areas of particular interest.

Keywords: Metabolomics, Metabolic footprint, Placental villous explants, Oxygen tension, Small for gestational age, Systems biology

Back to Article Outline

1. Introduction 

Being born small for gestational age (SGA) confers significantly increased risks of perinatal morbidity and mortality and increases the possibility of hypertension and metabolic disorders in later life [1], [2], [3], [4]. Depending on the definition used, up to 10% of all live-born neonates are designated as SGA. Although some SGA infants are constitutionally small without adverse risk, a significant proportion of SGA infants have fetal growth restriction (FGR), a condition characterised by a failure of the fetus to achieve its genetically determined size [5], [6]. As failure to achieve optimum fetal growth potential is complex to define and measure, the term SGA is used as an alternative. Use of individualised birthweight centiles, which are used in this study, excludes more babies that are constitutionally small, focussing on infants who have “pathological SGA” who at risk of perinatal morbidity and mortality [7].

SGA has a diverse aetiology, which makes elucidation of its pathophysiology challenging. Most research has focused on the role of uteroplacental insufficiency which results in a compromised supply of nutrients and oxygen to the fetus. Accumulating evidence suggests that SGA may result from complex biochemical interactions within a poorly perfused placenta [8], [9] or reduced placental transport of nutrients [10], [11]. Compared to those from pregnancies resulting in normal birthweight infants, placentas from women who deliver SGA infants have macroscopic evidence of infarction and microscopic changes including increased formation of syncytial knots, reduced cytotrophoblast proliferation and increased apoptosis [12], [13].

Changes in normal oxygen tensions in the placenta have been implicated in the pathophysiology of FGR. Oxygen (O₂) tension is a key regulator of cytotrophoblast proliferation and differentiation [14], [15] and therefore on placental growth. Placental hypoxia is hypothesized to be involved in the pathophysiology of SGA as abnormal uteroplacental blood flow and/or umbilical-placental blood flow may result in a reduction of O₂ (and nutrients) available for fetal use [16]. In support of this, pregnancies in women living at high altitude, and therefore relative placenta hypoxia, have an increased incidence of SGA accompanied by expected histological features [17]. Several authors have shown that reduced oxygenation in vitro replicates some features of SGA placentas including increased apoptosis and necrosis, reduced cytotrophoblast proliferation and increased syncytial knots [14], [15], [18], [19], [20]. Increasing O₂ tension also alters placental cell turnover, with exaggerated cytotrophoblast proliferation and syncytium formation and increased formation of syncytial knots in placental villous explants [18], [19].

There is a wide variation in the delivery of O₂ to the intervillous space in normal pregnancy and the effect of O₂ is dependent on the gestational age, the type of trophoblast and the type of tissue culture under investigation [14], [15], [21]. At term 6–8% O₂ represents normoxic and 1–2% hypoxic conditions for villous tissue at term [22]. In this study 20% O₂ was used as hyperoxic conditions as in previous studies [14], [18], [19], [23], [24].

Metabolic footprinting [25], [26], [27] is a powerful hypothesis-generating strategy for investigating the low molecular weight (bio)chemical factors (metabolites) absorbed by and released from a cell culture or tissue in vitro [28], [29]. Hence, the conditioned culture medium is studied. The position of the metabolome as the final downstream product of gene expression provides a high-resolution multi-factorial phenotypic signature of disease aetiology, manifestation or pathophysiology [30], [31].

Metabolomic strategies have been applied to study hypoxia and anoxia in placental tissue from normal and pre-eclamptic pregnancies and those from high altitude. Tissot van Patot et al. demonstrated metabolic differences in placental tissue from pregnancies at low and high altitude using nuclear magnetic resonance (NMR) technology [32]. Studies employing metabolic footprinting using gas chromatography–mass spectrometry showed changes in the concentration of redox metabolites at different oxygen tensions (1, 6 and 20%) [23]. A separate study has shown metabolic differences between placental explants cultured at 1% and 6% O₂ in normal pregnancies and women with pre-eclampsia using Ultra Performance Liquid Chromatography–Mass Spectrometry (UPLC–MS) [33].

We proposed to use this state of the art metabolomics technology (UPLC–MS) to examine changes in the metabolic footprints of placental villous explants cultured in different O₂ tensions (1, 6 and 20%) from placentas of small for gestational age and those from normal pregnancies. We hypothesized that samples from normal pregnancies would show their normal in vivo metabolic signature at 6% O₂ whilst the SGA samples would reflect their normal in vivo metabolic signature at something less than 6%. In addition, reduction of O₂ tension to 1% should render the control samples more like the SGA, whilst 6% O₂ for the SGA samples might result in changes consistent with an increased O₂ tension (20%) in the control.

Back to Article Outline

2. Methods and materials 

2.1. Culture of placental villous explants 

Placentas were obtained with written consent in compliance with local ethics committee approval (Central Manchester Research Ethics Committee, 08/H1008/22). Placentas were collected from women with uncomplicated term pregnancies resulting in delivery of a healthy singleton fetus (n = 8) and from women with suspected SGA which was subsequently confirmed after delivery based upon individualised birthweight centile scores less than the 5th percentile (n = 9) [34] (www.gestation.net). Pregnancies complicated by any other maternal or fetal factor, including pre-eclampsia, diabetes mellitus and congenital anomalies were excluded.

Placentas (all received within 20 min of delivery) were sampled at three central placental lobules. Central sampling was chosen as it was deemed the most representative and reproducible location to reflect optimal placental function [35]. Nine explants (3 sets of 3) were cut from three separate placental lobules (27 explants from each placenta), and subsequently stratified so that 3 explants sets (one from each lobule) were cultured at each O₂ tension (1%, 6% and 20%). In total, 153 metabolic footprint samples were collected.

Placental villous explants weighing approximately 5 mg per fragment were prepared as previously described [18] and cultured at the gas–medium interface of fetal bovine serum supplemented with CMRL-1066 medium derived from a single batch for 48 h at 37 °C in an atmospheric O₂ tension of either 1%, 6% or 20%. After 48 h, culture medium was replaced with fresh medium equilibrated to the appropriate O₂ level for 24 h. The explants cultured at 1% O₂ tension remained within the hypoxic chamber when cultured medium was changed. Following a further 48 h conditioned culture medium was collected and immediately frozen at −80 °C.

2.2. Ultra Performance Liquid Chromatography–Mass Spectrometry (UPLC–LTQ/Orbitrap) analysis 

Metabolic footprint samples were prepared as previously described by thawing on ice of the conditioned culture medium followed by lyophilisation of a 100 μl aliquot [33]. All media samples were analysed in a random order after reconstitution in 200 μl of water using Ultra Performance Liquid Chromatography–Mass Spectrometry (Waters ACQUITY UPLC system coupled to a ThermoFisher Scientific electrospray hybrid LTQ-Orbitrap mass spectrometer). Instrument methods applied here have been previously described in Ref. [33]. Raw centroid data was deconvolved into a peak table using XCMS software v4.1 to report 1676 features (described by a retention time and accurate mass). The data is not quantitative as calibration curves for each detected feature have not been experimentally acquired. The data is reported as responses, calculated as the chromatographic peak area. The term metabolic signature is used to reflect the resulting changes elicited by specific conditions. Full details of all methods pertaining to sample preparation and UPLC–MS analysis are described in the attached Supplementary methodology file.

2.3. Statistical analysis 

A combination of univariate and multivariate statistical analysis were performed. Supervised multivariate analysis was employed in the form of Canonical Variates Analysis (CVA) [36] combined with Principal Components Analysis (PC-CVA), in order to look at system-wide multivariate changes in the metabolic footprint. The number of Principal Components used in the discriminant model was optimized by employing 0.632 + Bootstrap substitution [37], using 100 bootstrap samples to produce an unbiased estimate of prediction. All data were normalised to zero mean and unit variance before performing any multivariate statistics in order to give equal weighting to all metabolite features.

Univariate statistical analysis was used to mine the data further once global changes were characterised. Metabolite features were tested using either the Mann–Whitney test or Student’s t-test, depending on data normality. The critical p-value for significance was set to 0.05. No correction for multiple comparisons was performed as the univariate analysis mainly consisted in a comparison between the different experimental conditions. However, False Discovery Rate (FDR) [38] analysis was performed on all the statistical comparisons, and FDRs quoted where appropriate. All identified metabolites are quoted with both p-values and FDR corrected q-values.

2.4. Metabolite identification 

Metabolites were putatively identified by the matching of the measured accurate mass to accurate mass(es) present in the Manchester Metabolomics Database (MMD) [39]. Where further retention time data was available in mass spectral libraries developed in Manchester, a definitive identification was performed. Definitive identifications are described with an asterisk (*) in Table 2. Using UPLC–MS, a single metabolite is often detected as multiple chromatographic peaks of the same retention time and different accurate mass. This is due to chemical adduction, dimerization, multiple-charging, isotope peaks and fragmentation. In this paper the term metabolite feature refers to any detected chromatographic peak and metabolite refers to a ‘unique’ identified metabolite. For further details please see the attached Supplementary methodology file.

Back to Article Outline

3. Results 

Maternal characteristics and pregnancy outcomes of the cases and controls are described in Table 1. Age, BMI, parity, smoking and baby sex were carefully matched across cases and controls. All SGA babies had an individualised birthweight centile ≤5th centile, six were <3rd centile. Eight of the nine SGA babies had oligohydramnios. All control babies were term deliveries with normal liquor volume. One case of SGA was delivered prior to 37 weeks’ gestation.

Table 1. Maternal characteristics and pregnancy outcomes.

Parameter	Control (range)	SGA (range)	t-test
Age	29 (18–35)	28 (20–35)	NS
Parity –nullip	6/8	7/9	NS
BMI Kg/m2	22.9 (19.5–25.5)	23.2 (20.2–25.9)	NS
Smoker –yes	2/8	2/9	NS
Sex –male	5/8	5/9	NS
Birthweight (g)	3376 (2810–4655)	2253 (769–2840)	0.002
Mode of Delivery –vaginal –C/S	4 4	3 6	NS NS
Gestation	39 + 1 (37+2 − 40 + 4)	38 + 3 (29+5 − 41 + 4)	NS

The table shows the maternal characteristics and pregnancy outcomes of the SGA (n = 9) and control (n = 8) babies. BMI = body mass index; C/S = caesarean section.

Table 2. The table shows the 79 metabolites that showed a significant increase in SGA sample concentration but no significant change in the controls when culture conditions were changed from 6% to 1% O₂. The table shows the significance value of each metabolite at each oxygen tension and those with p < 0.05 are shaded. The q-value represents the False Discovery Rate for each metabolite. These 79 metabolites represent the triangles in quadrant A and B in Fig. 3(a). MG = monoacylglycerol; CE = cholesterol ester; PC = phosphocholine; PE = phosphoethanolamine; PI = phosphoinositol; PGE2 = prostaglandin E2; PGD2 = prostaglandin D2; TG = triglyceride/triacylglycerol. Those metabolites marked with an asterisk are definitely identified. Where PC or PE or PI is listed without a fatty acid configuration, it was not possible to give a more specific identification but they are classified as PC or PE or PI.

3.1. Global changes in the metabolic footprint of conditioned culture medium from different O₂ tensions 

UPLC–MS analysis reproducibly detected a total of 1676 metabolite features. A cross-validated multivariate discriminant model (PC-CVA) was built using all the detected features. The resulting scores plot (Fig. 1) demonstrated differences between the six defined sample groups (control_1%; SGA_1%; control_6%; SGA_6%; control_20%; SGA_20%). The primary change in metabolic profile (CV1) reflected the change in O₂ tension. A secondary difference between SGA and control classes at each O₂ tension was also detected (CV2). The 90% confidence circles show that there was a significant difference in the multivariate means between the SGA and control classes at 6% and 20% O₂; however, not at 1%, although there was a slight difference in populations this difference was not significant.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1 

  The figure shows the PC-CVA scores plot (using all the detected metabolite features) for a model built and cross-validated using the first 20 PC scores to discriminate between the six medium classes of interest. Circles represent 1% oxygen samples, squares represent 6% oxygen samples and triangles represent 20% oxygen samples. Open shapes represent control samples and closed shapes represent SGA samples at each oxygen tension. For clarity the placental biological replicates have been averaged. The major change in metabolic footprint is due to oxygen tension (CV1). CV2 reflects a change in footprint reflecting SGA phenotype across all O₂ tensions. The units of the x- and y-axis are arbitrary due to normalisation of the data. CV = canonical variate; PC-CVA = Principal Components-Canonical Variates Analysis; SGA = small for gestational age; inner circle = 90% CI of the population mean; outer circle = 90% CI of the population itself.

Further investigation showed that the discrimination seen in CV1 was due to a systemic increase in relative (to 6% O₂ tension) metabolite concentration as the O₂ tension was reduced and a systemic decrease in relative metabolite concentration as the O₂ tension was increased.

3.2. Characterising the differences between the metabolic footprint of conditioned culture medium from SGA and control pregnancies at each O₂ tension 

Univariate hypothesis testing was performed across the 1676 detected metabolite features. A total of 574 metabolite features showed significant difference between SGA and normal at one or more of the oxygen tensions. Fig. 2 shows a Venn diagram showing the distribution of the significant metabolite features at each oxygen tension. Of the 574 features that were significantly different at one or more of the oxygen tensions, 221 were identified as ‘unique’ endogenous metabolites, after removing the multiple matches of, for example, chemical adducts and isotope peaks [39]. These metabolites are listed in full in Supplementary file. 16 metabolites were significant (p < 0.05) at all oxygen tensions, 12 metabolites at both 1% and 6% O₂, but not 20%, and 45 metabolites at both 6% and 20% O₂, but not 1%.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 2 

  Venn diagram showing the distribution of the significant metabolite features at each oxygen tension. A total of 574 metabolite features showed significant difference between SGA and normal controls at one or more of the oxygen tensions (p < 0.05). At 6% O₂ 427 metabolite features were significantly different with a False Discovery Rate of 0.1. At 1% O₂, 160 metabolite features were significantly different (FDR = 0.4). At 20% O₂, 254 metabolite features were significantly different (FDR = 0.2). Of the 574 metabolite features, 52 metabolite features were significantly different at all oxygen tensions (9% of the 574 metabolites); 32 metabolite features were significant at both 1% and 6% O₂, but not 20% (6% of the 574); 124 were significant at both 6% and 20% O₂, but not 1% (22%); 7 were significant at both 1% and 20% O₂, but not 6% (1%).

95% of the 574 metabolite features that showed significant difference between SGA and normal at any O₂ tension showed a lower mean metabolite level in the SGA samples when compared to the controls.

3.3. Relative changes in the normoxic metabolic footprint due to culturing at 1% and 20% O₂ tensions 

The data were reduced to the 427 metabolite features that showed significant difference between SGA and control samples cultured at 6% O₂ (normoxic conditions). 161 ‘unique’ endogenous metabolites have been putatively identified from these 427 features. For clarity, the footprint for this group of metabolites will be identified as the ‘normoxic biomarker signature of SGA’ for the remainder of the paper.

In order to characterise exactly how the metabolic footprint for each sample type differed when cultured at different O₂ tensions, a set of paired two-sample t-tests was performed. This was possible as explants from the same placenta were cultured at the different O₂ tensions allowing pair-wise matching. For each metabolite of interest, the following four comparison groups were tested: SGA_6%v1% (paired t-test between the SGA explants cultured at 6% O₂ and the SGA explants cultured at 1% O₂) control_6%v1%, SGA_6%v20% and control_6%v20%. Fig. 3(a) compares the significant changes in footprint due to changing oxygen culture conditions from 6% to 1% O₂ (SGA_6%v1% vs. control_6%v1%), while Fig. 3b compares the significant changes due to changing from 6% to 20% O₂ (SGA_6%v20% vs. control_6%v20%).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 3 

  (a) and (b) are biplots. Each point in the biplot represents one of the observed metabolite features. Each point’s co-ordinate location (x, y), is defined by the significance of the relative change in metabolite concentration combined with the mean direction of that change (x = control; y = SGA). In these biplots, points lying in zone A show a mean increase in metabolite level for SGA samples and a mean decrease in control samples; zone B show a mean increase in metabolite level for both SGA and control samples; zone C show a decrease in mean metabolite level for both SGA and control samples; zone D show a decrease in mean metabolite level for SGA samples and an increase for control samples. (a) Biplot of change in metabolites, significant at 6% O₂, when explants were cultured at 1% O₂. The triangles (n = 81, 50%) show significant change in metabolite level for SGA samples but no significant change for control samples and the squares (n = 16, 10%) show significant change in metabolite level for control samples but no significant change for SGA samples (P = 0.05). (b) Biplot of change in metabolites, significant at 6% O₂, when explants were cultured at 20% O₂. The triangles (n = 6, 4%) represent significant change in metabolite level for SGA samples but no significant change for control samples and the squares (n = 31, 19%) represent significant change in metabolite level for control samples but no significant change for SGA samples (P = 0.05). A circle indicates a metabolite which significantly changes in both the SGA and control groups. The triangles indicate metabolites that are significantly changed in SGA but not significantly changed in control, and the squares indicate metabolites that are significantly changed in control but not significantly changed in SGA. The crosses indicate no significant change in either the SGA or control samples.

When comparing the change in metabolite levels after decreasing the O₂ tension from 6% to 1% O₂ (normoxia to hypoxia) four general trends were observed (Fig. 3a):

When comparing the change in metabolite levels due to increasing the oxygen tension from 6% to 20% O₂ three trends were observed (Fig. 3b):

Only 3 metabolites showed a significant increase in SGA sample concentration but no significant change in the controls (triangles; 2%).

Back to Article Outline

4. Discussion 

4.1. The global differences in the metabolic footprint of the conditioned culture media were primarily dependent on the O₂ tension the samples were cultured under 

The PC-CVA model built using all the detected metabolite features showed that the major differences in the metabolic footprint were due to the differing O₂ tensions (Fig. 1). This was to be expected, given the extreme differences in oxygen tensions in which the villous explants were cultured, and such metabolic changes have been demonstrated previously using this model [23], [33] and tissue damage (including apoptosis) has been shown by exposing to hypoxic conditions [19].

4.2. The metabolic footprints of SGA and control conditioned culture media were different when compared at each O₂ tension in isolation 

There was a significant difference between SGA and control classes when both were cultured at either 6% or 20% O₂ tensions. However, at 1% O₂ there was no significant difference between SGA and controls (Fig. 1). These multivariate findings were corroborated by the univariate statistical analysis. There were significant differences between the SGA and control classes at each individual O₂ tension, and a subset of metabolite features was significant across every oxygen tension (Fig. 2). However, this subset was relatively small (9% (52/574) of the total number of metabolite features that were significantly different at any O₂ tension).

4.3. Characterisation of the metabolic changes in the conditioned culture media suggests that changing from normoxic to hypoxic culturing conditions significantly increases disruption in SGA metabolism when compared with normal placenta 

By performing pair-wise univariate analysis it was possible to characterise the nature of the changes in the medium metabolic signature due to decreasing the oxygen tension from 6% to 1%. Of the 161 metabolites that made up the normoxic biomarker signature of SGA, 49% exhibited a signature of SGA at 1% similar to that of the controls at 6% (Table 2). This suggests that, if placental hypoxia is the ‘normal’ in vivo condition for SGA as suggested by increased expression of hypoxia inducible factors, then exposure to normoxic oxygen levels significantly increases metabolic dysfunction in SGA placenta.

In addition, a small but significant number of metabolites (16; 10%) changed such that the control signature at 1% O₂ showed the same signature as SGA at 1% and 6%, but not the same as controls at 6% (where relative metabolite concentrations were significantly decreased) (Table 3). In this case, if placental hypoxia is the ‘normal’ in vivo conditions for SGA then, for these metabolites only, culturing control samples under hypoxic conditions induces the SGA metabolic signature. The metabolite numbers are small so this cannot be interpreted as a widespread trend.

4.4. Characterisation of the metabolic changes in the conditioned culture media suggests that changing from normoxic to hyperoxic culturing conditions has only a small effect on the metabolic differences between SGA and normal placenta 

When looking at the changes in the medium metabolic signature due to increasing the O₂ tension from 6% to 20%, hyperoxia largely reduced the metabolite levels of both sample types while keeping the metabolic signature difference between SGA and control intact (48%), suggesting a general increase in catabolism of nutrients under 20% O₂ tension conditions. Overall, three quarters (75%) of metabolites showed no changes or concordant change implying that the SGA samples do not differentially respond to this increase in oxygen tension and that control samples differentially change only minimally.

The metabolites that have been identified (Table 2 and Supplementary file) may provide further understanding of the pathophysiology of SGA. Although due to the novelty of these studies, the significance of many of these metabolites in relation to SGA is uncertain as they have not been detected previously. A wide range of metabolism is disrupted including a number of metabolites (16 identified putatively) at all oxygen tensions. The most interesting results were found when culture conditions were changed from 6% to 1% O₂ tension as described in Section 4.3 (see Table 2, Table 3). An in depth discussion of all these metabolites is not feasible in this article but some metabolites will be considered further here.

A metabolite class that showed much disruption (upregulation) in the culture medium of SGA samples when conditions were changed from 6% to 1% was the phospholipids. Phospholipids are the major lipid constituents of cell membranes. Glycerophosphocholines and glycerophosphoethanolamines are two of the common phospholipids as has been observed for the placenta [40]. Changes in normal oxygen tensions can cause changes to glycerophospholipids resulting in many different products which have many different proposed biological properties [41]. Phosphatidylcholine metabolism is affected by hypoxia particularly in the heart in animal studies [42] but there is a lack of evidence on the effect of hypoxia on glycerophospholipids in the placenta. The phospholipid changes observed in the described study are most likely a result of cell membrane damage leading to the subsequent release of phospholipids. However, there is recent evidence of anti-phospholipid antibodies (and complement activation) co-operating in triggering a local inflammatory process, eventually leading to placental thrombosis, hypoxia, and neutrophil infiltration [43].

A number of essential amino acids were also increased in SGA samples when culture conditions were changed from 6% to 1% including tryptophan, methionine and phenylalanine. Tryptophan functions as a biochemical precursor for serotonin and niacin. Of particular interest to SGA is that tryptophan is a precursor of serotonin. Elevated tryptophan and serotonin metabolism has been reported in plasma from newborn SGA infants [44] and in the brain of malnourished rat fetuses [45] and serotonin and its receptors have been localized in the placenta implying a role in placental development [46]. Kynurenine is a metabolite of tryptophan and kynurenine metabolism is significantly altered in fetal sheep when placental function is chronically compromised in late gestation [47]. Kynurenine was noted to be significantly decreased in SGA at these culture conditions but was also decreased in controls. This may be explained by indoleamine 2,3-dioxygenase activity which is involved in kynurenine metabolism. This has been localized in the placenta and detected from 14 weeks gestation and its placental activity was found to be lower in FGR [48]. Kynurenine was also noted to change by Dunn et al. in their study investigating metabolic disturbances in placental explants in pregnancies complicated by pre-eclampsia and in response to hypoxia [33].

Methionine is an essential amino acid required for normal growth and is a precursor of homocysteine. Hyperhomocysteinemia in particular is associated with a number of complications including SGA and cardiovascular disease and can result from a variety of acquired factors (deficiency of vitamins B6, B12 and folic acid, high meat diet, smoking and others) or genetic factors (abnormalities of methionine – homocysteine metabolism) [49]. The exact mechanism by which homocysteine is associated with vasculopathy is still unclear. Accumulating evidence suggests that hyperhomocysteinaemia alters cellular structure and metabolism, mediated by free radicals generated during the oxidation of homocysteine [50]. In vitro studies show that exposure of endothelial cells to homocysteine results in oxidative effects, including generation of superoxide anion radicals and hydrogen peroxide, which lead to inactivation of nitric oxide-mediated pathways and endothelial-cell damage [51].

Phenylalanine is an essential amino acid and the precursor for the amino acid tyrosine. Phenylalanine has been found to be lower in umbilical venous samples from growth restricted pregnancies [52] and also in umbilical venous samples from growth restricted twins where there was discordant growth among monochorionic twins [53]. Another study showed significantly higher concentrations of phenylalanine (P < 0.05), in the SGA compared to the term normotensive and pre-eclamptic groups in placental villi homogenates [54]. Two of these studies used high performance liquid chromatography for the analysis [53], [54]. These support our findings and suggest that there are abnormalities in placental metabolism, amino acid transfer into the fetal circulation, and also fetoplacental perfusion in pregnancies complicated by SGA. Changing O₂ tension has an effect on this placental metabolism but it is uncertain as to the extent of this effect.

Interestingly, of the metabolites that changed in the control samples when culture conditions were changed from 6% to 1% a number of cholesterol esters were identified. Cholesterol esters are cholesterol molecules with long-chain fatty acids linked to the hydroxyl group. Cholesterol esters are major constituents of the lipoprotein particles carried in blood. They also accumulate in the fatty lesions of atherosclerotic plaques. Low cholesterol ester concentrations and low high density lipoprotein levels have been found in SGA newborns and in the umbilical venous blood of SGA babies [55], [56]. Cholesterol ester transfer has also been found to be increased in SGA neonates [57]. This increased cholesteryl ester transfer may in part, explain the increased risk of coronary heart disease (CHD) of small for gestational age neonates in later life.

We acknowledge several limitations of our study. SGA has been defined here as birth weight less than the 10th centile. A proportion of SGA infants defined as less than the 10th centile will be constitutionally small and thus there are limitations in the use of the birth weight percentile as a surrogate marker of FGR. However, in this study we used individualised centiles and all the cases had an IBR ≤the 5th centile, significantly increasing the probability that these cases resulted from placental dysfunction rather than being constitutionally small.

A significant challenge of metabolomic studies is metabolite identification. UPLC–MS is a very sensitive analytical technique. Further studies will be performed to establish the identity of features with multiple possible identifications.

This study has used metabolomic technologies to show differences between SGA and normal pregnancies under different oxygen conditions. To our knowledge this is the first time that metabolomics has been used to investigate SGA. The results from in vitro culturing of placental villous explants suggest that placental hypoxia is the likely in vivo condition for SGA but that simply restoring ‘normoxia’ may be deleterious to the tissue. Oxygen delivery to the placenta of an SGA fetus is likely to be a changing variable and the placenta of an SGA fetus is possibly more tolerant to low oxygen tensions. This may be similar to placental adaptation to chronic hypoxia at high altitude [32]. There are also a number of metabolites that are significantly different between SGA and controls at all oxygen levels, suggesting that other factors such as nutrient deficiency may play an important role in the pathophysiology of SGA. This may provide a clear differentiation between SGA and PE, where hypoxia was observed to be the dominant determinant of the phenotype [33]. There is also some overlap in disrupted metabolic classes between our study and the study by Dunn et al. [33]. It requires further clarification whether these changes are the result of a general disruption in SGA or PE-complicated pregnancies or are specific to SGA. We also wish to see how our findings relate to early pregnancy in women who go on to deliver an SGA baby.

This study further adds to the growing data from pregnancy metabolomics research. In summary, we have shown that the metabolic footprint of placental tissue exhibits different phenotypes when cultured at different O₂ tensions and moreover the metabolic profile of SGA placental explants can be distinguished from controls irrespective of O₂ tension. These results may provide some insight into the pathophysiology of this complex condition and will provide direction for future research to aid the pathophysiological understanding of SGA.

Back to Article Outline

Acknowledgements 

The authors would like to thank the midwives on Central Delivery Unit, St Mary’s Hospital, Manchester for their assistance in acquiring placental tissue and the patients for donating their placentas. RPH is funded by the Health Research Board and Health Service Executive of Ireland (CRT/2007/5). The funding body did not have any involvement in the preparation of this article. WD would like to thank BBSRC and EPSRC for financial contributions to The Manchester Centre for Integrative Systems Biology (grant number BB/C008219/1). The research was partially supported by The Manchester Biomedical Research Centre.

Back to Article Outline

Appendix. Supplementary data 

Back to Article Outline

References 

1. McIntire DD, Bloom SL, Casey BM, Leveno KJ. Birth weight in relation to morbidity and mortality among newborn infants. N Eng J Med. 1999;340(16):1234–1238
   • View In Article
2. Clausson B, Gardosi J, Francis A, Cnattingius S. Perinatal outcome in SGA births defined by customised versus population-based birthweight standards. Br J Obstet Gynaecol. 2001;108(8):830–834
   • View In Article
3. Law CM, Shiell AW, Newsome CA, Syddall HE, Shinebourne EA, Fayers PM, et al. Fetal, infant, and childhood growth and adult blood pressure – a longitudinal study from birth to 22 years of age. Circulation. 2002;105(9):1088–1092
   • View In Article
   • CrossRef
4. Barker DJP. The fetal origins of type 2 diabetes mellitus. Ann Intern Med. 1999;130(4):322–324
   • View In Article
   • MEDLINE
5. Goldenberg RL, Cliver SP. Small for gestational age and intrauterine growth restriction: definitions and standards. Clin Obstet Gynecol. 1997 Dec;40(4):704–714
   • View In Article
   • MEDLINE
   • CrossRef
6. Gardosi J. New definition of small for gestational age based on fetal growth potential. Horm Res. 2006;65(Suppl. 3):15–18
   • View In Article
   • CrossRef
7. McCowan LME, Harding JE, Stewart AW. Customised birthweight centiles predict SGA pregnancies with perinatal morbidity. BJOG. 2005;112(8):1026–1033
   • View In Article
   • MEDLINE
   • CrossRef
8. Trudinger BJ, Giles WB. Elaboration of stem villous vessels in growth restricted pregnancies with abnormal umbilical artery Doppler waveforms. Br J Obstet Gynaecol. 1996;103(5):487–488
   • View In Article
   • MEDLINE
9. Jackson MR, Walsh AJ, Morrow RJ, Mullen JBM, Lye SJ, Ritchie JW. Reduced placental villous tree elaboration in small-for-gestational-age pregnancies – relationship with umbilical artery Doppler wave-forms. Am J Obstet Gynecol. 1995;172(2):518–525
   • View In Article
   • Full-Text PDF (788 KB)
   • CrossRef
10. Weedon MN, Frayling TM, Shields B, Knight B, Turner T, Metcalf BS, et al. Genetic regulation of birth weight and fasting glucose by a common polymorphism in the islet cell promoter of the glucokinase gene. Diabetes. 2005;54(2):576–581
    • View In Article
    • MEDLINE
    • CrossRef
11. Gagnon R. Placental insufficiency and its consequences. Eur J Obstet Gynecol Reprod Biol. 2003;110:S99–S107
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (149 KB)
    • CrossRef
12. Smith SC, Baker PN, Symonds EM. Increased placental apoptosis in intrauterine growth restriction. Am J Obstet Gynecol. 1997;177(6):1395–1401
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (2776 KB)
    • CrossRef
13. Chen CP, Bajoria R, Aplin JD. Decreased vascularization and cell proliferation in placentas of intrauterine growth-restricted fetuses with abnormal umbilical artery flow velocity waveforms. Am J Obstet Gynecol. 2002;187(3):764–769
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (154 KB)
    • CrossRef
14. Genbacev O, Zhou Y, Ludlow JW, Fisher SJ. Regulation of human placental development by oxygen tension. Science. 1997 Sep 12;277(5332):1669–1672
    • View In Article
    • MEDLINE
    • CrossRef
15. Crocker IP, Tansinda DM, Jones CJ, Baker PN. The influence of oxygen and tumor necrosis factor-alpha on the cellular kinetics of term placental villous explants in culture. J Histochem Cytochem. 2004 Jun;52(6):749–757
    • View In Article
    • MEDLINE
    • CrossRef
16. Kingdom JC, Kaufmann P. Oxygen and placental villous development: origins of fetal hypoxia. Placenta. 1997 Nov;18(8):613–621[discussion 23–6]
    • View In Article
    • Abstract
    • Full-Text PDF (1155 KB)
    • CrossRef
17. Khalid ME, Ali ME, Ali KZ. Full-term birth weight and placental morphology at high and low altitude. Int J Gynaecol Obstet. 1997 Jun;57(3):259–265
    • View In Article
    • Abstract
    • Full-Text PDF (765 KB)
    • CrossRef
18. Heazell AE, Lacey HA, Jones CJ, Huppertz B, Baker PN, Crocker IP. Effects of oxygen on cell turnover and expression of regulators of apoptosis in human placental trophoblast. Placenta. 2008 Feb;29(2):175–186
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (2961 KB)
    • CrossRef
19. Heazell AE, Moll SJ, Jones CJ, Baker PN, Crocker IP. Formation of syncytial knots is increased by hyperoxia, hypoxia and reactive oxygen species. Placenta. 2007 Apr;28(Suppl. A):S33–S40
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (1138 KB)
    • CrossRef
20. Huppertz B, Kingdom J, Caniggia I, Desoye G, Black S, Korr H, et al. Hypoxia favours necrotic versus apoptotic shedding of placental syncytiotrophoblast into the maternal circulation. Placenta. 2003 Feb–Mar;24(2–3):181–190
    • View In Article
    • Abstract
    • Full-Text PDF (1332 KB)
    • CrossRef
21. Kilani RT, Mackova M, Davidge ST, Guilbert LJ. Effect of oxygen levels in villous trophoblast apoptosis. Placenta. 2003 Sep–Oct;24(8–9):826–834
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (322 KB)
    • CrossRef
22. Miller RK, Genbacev O, Turner MA, Aplin JD, Caniggia I, Huppertz B. Human placental explants in culture: approaches and assessments. Placenta. 2005;26(6):439–448
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (237 KB)
    • CrossRef
23. Heazell AEP, Brown M, Dunn WB, Worton SA, Crocker IP, Baker PN, et al. Analysis of the metabolic footprint and tissue metabolome of placental villous explants cultured at different oxygen tensions reveals novel redox biomarkers. Placenta. 2008;29(8):691–698
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (259 KB)
    • CrossRef
24. Kudo Y, Boyd CA, Sargent IL, Redman CW. Hypoxia alters expression and function of syncytin and its receptor during trophoblast cell fusion of human placental BeWo cells: implications for impaired trophoblast syncytialisation in pre-eclampsia. Biochim Biophys Acta. 2003 May 20;1638(1):63–71
    • View In Article
    • MEDLINE
25. Allen J, Davey HM, Broadhurst D, Heald JK, Rowland JJ, Oliver SG, et al. High-throughput classification of yeast mutants for functional genomics using metabolic footprinting. Nat Biotechnol. 2003;21(6):692–696
    • View In Article
    • MEDLINE
    • CrossRef
26. O’Hagan S, Dunn WB, Brown M, Knowles JD, Kell DB. Closed-loop, multiobjective optimization of analytical instrumentation: gas chromatography/time-of-flight mass spectrometry of the metabolomes of human serum and of yeast fermentations. Anal Chem. 2005 Jan 1;77(1):290–303
    • View In Article
    • MEDLINE
    • CrossRef
27. Kell DB, Brown M, Davey HM, Dunn WB, Spasic I, Oliver SG. Metabolic footprinting and systems biology: the medium is the message. Nat Rev Microbiol. 2005;3(7):557–565
    • View In Article
    • MEDLINE
    • CrossRef
28. Goodacre R, Vaidyanathan S, Dunn WB, Harrigan GG, Kell DB. Metabolomics by numbers: acquiring and understanding global metabolite data. Trends Biotechnol. 2004 May;22(5):245–252
    • View In Article
    • MEDLINE
    • CrossRef
29. Nicholson JK, Wilson ID. Opinion: understanding ‘global’ systems biology: metabonomics and the continuum of metabolism. Nat Rev Drug Discov. 2003 Aug;2(8):668–676
    • View In Article
    • MEDLINE
30. Sreekumar A, Poisson LM, Rajendiran TM, Khan AP, Cao Q, Yu J, et al. Metabolomic profiles delineate potential role for sarcosine in prostate cancer progression. Nature. 2009 Feb 12;457(7231):910–914
    • View In Article
    • CrossRef
31. Oresic M, Simell S, Sysi-Aho M, Nanto-Salonen K, Seppanen-Laakso T, Parikka V, et al. Dysregulation of lipid and amino acid metabolism precedes islet autoimmunity in children who later progress to type 1 diabetes. J Exp Med. 2008 Dec 22;205(13):2975–2984
    • View In Article
    • CrossRef
32. van Patot MCT, Murray AJ, Beckey V, Cindrova-Davies T, Johns J, Zwerdlinger L, et al. Human placental metabolic adaptation to chronic hypoxia, high altitude: hypoxic preconditioning. American Journal of Physiology-Regulatory Integrative and Comparative Physiology. 2010;298(1):R166–R172
    • View In Article
33. Dunn WB, Brown M, Worton SA, Crocker IP, Broadhurst D, Horgan R, et al. Changes in the metabolic footprint of placental explant-conditioned culture medium identifies metabolic disturbances related to hypoxia and pre-eclampsia. Placenta. 2009 Nov;30(11):974–980
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (189 KB)
    • CrossRef
34. Gardosi J, Chang A, Kalyan B, Sahota D, Symonds EM. Customized antenatal growth charts. Lancet. 1992;339(8788):283–287
    • View In Article
    • Abstract
    • CrossRef
35. Horgan R, Brown M, Dunn W, Heazell A, Kell D, Kenny L, et al. The difference in the metabolic footprint of placental villous explants cultured in different oxygen tensions. BJOG. 2009;116(10):1409–1410
    • View In Article
36. Krzanowski W. Principles of multivariate analysis: a user’s perspective. Oxford: Oxford Univeristy Press; 1988;
    • View In Article
37. Efron B, Tibshirani R. Improvements on cross-validation: the .632+ bootstrap method. J Am Stat Assoc. 1997;92(438):548–560
    • View In Article
38. Storey JD. A direct approach to false discovery rates. J R Stat Soc Ser B Stat Methodol. 2002;64:479–498
    • View In Article
39. Brown M, Dunn WB, Dobson P, Patel Y, Winder CL, Francis-McIntyre S, et al. Mass spectrometry tools and metabolite-specific databases for molecular identification in metabolomics. Analyst. 2009 Jul;134(7):1322–1332
    • View In Article
    • CrossRef
40. Bayon Y, Croset M, Chirouze V, Tayot JL, Lagarde M. Phospholipid molecular species from human placenta lipids. Lipids. 1993 Jul;28(7):631–636
    • View In Article
    • MEDLINE
    • CrossRef
41. Fruhwirth GO, Loidl A, Hermetter A. Oxidized phospholipids: from molecular properties to disease. Biochim Biophys Acta. 2007 Jul;1772(7):718–736
    • View In Article
    • MEDLINE
42. Choy PC, Chan M, Hatch G, Man RY. Phosphatidylcholine metabolism in ischemic and hypoxic hearts. Mol Cell Biochem. 1992 Oct 21;116(1–2):53–58
    • View In Article
    • MEDLINE
    • CrossRef
43. Tincani A, Cavazzana I, Ziglioli T, Lojacono A, De Angelis V, Meroni P. Complement activation and pregnancy failure. Clin Rev Allergy Immunol. 2009 Nov 11;
    • View In Article
44. Hernandez-Rodriguez J, Meneses L, Herrera R, Manjarrez G. Another abnormal trait in the serotonin metabolism path in intrauterine growth-restricted infants. Neonatology. 2009;95(2):125–131
    • View In Article
    • CrossRef
45. Hernandez J, Manjarrez GG, Chagoya G. Newborn humans and rats malnourished in utero: free plasma l-tryptophan, neutral amino acids and brain serotonin synthesis. Brain Res. 1989 May 29;488(1–2):1–13
    • View In Article
    • MEDLINE
    • CrossRef
46. Huang WQ, Zhang CL, Di XY, Zhang RQ. Studies on the localization of 5-hydroxytryptamine and its receptors in human placenta. Placenta. 1998;19(8):655–661
    • View In Article
    • Abstract
    • Full-Text PDF (4696 KB)
    • CrossRef
47. Nicholls T, Nitsos I, Smythe G, Walker DW. Kynurenine production and catabolism in fetal sheep with embolized or nonembolized placentas. Am J Obstet Gynecol. 2001 Oct;185(4):988–995
    • View In Article
48. Kamimura S, Eguchi K, Yonezawa M, Sekiba K. Localization and developmental-change of indoleamine 2,3-dioxygenase activity in the human placenta. Acta Med Okayama. 1991;45(3):135–139
    • View In Article
49. Aubard Y, Darodes N, Cantaloube M. Hyperhomocysteinemia and pregnancy – review of our present understanding and therapeutic implications. Eur J Obstet Gynecol Reprod Biol. 2000 Dec;93(2):157–165
    • View In Article
50. Domagala TB, Undas A, Libura M, Szczeklik A. Pathogenesis of vascular disease in hyperhomocysteinaemia. J Cardiovasc Risk. 1998 Aug;5(4):239–247
    • View In Article
    • MEDLINE
    • CrossRef
51. Loscalzo J. The oxidant stress of hyperhomocyst(e)inemia. J Clin Invest. 1996 Jul 1;98(1):5–7
    • View In Article
    • MEDLINE
    • CrossRef
52. Paolini CL, Marconi AM, Ronzoni S, Di Noio M, Fennessey PV, Pardi G, et al. Placental transport of leucine, phenylalanine, glycine, and proline in intrauterine growth-restricted pregnancies. J Clin Endocrinol Metab. 2001 Nov;86(11):5427–5432
    • View In Article
    • CrossRef
53. Bajoria R, Sooranna SR, Ward S, D’Souza S, Hancock M. Placental transport rather than maternal concentration of amino acids regulates fetal growth in monochorionic twins: implications for fetal origin hypothesis. Am J Obstet Gynecol. 2001 Nov;185(5):1239–1246
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (101 KB)
    • CrossRef
54. Morris NH, Burston D, Ramsay B, Sooranna SR. Free amino acid concentrations in normal and abnormal third trimester placental villi. Eur J Clin Invest. 1995 Oct;25(10):796–798
    • View In Article
    • MEDLINE
    • CrossRef
55. Merzouk H, Lamri MY, Meghelli-Bouchenak M, Korso N, Prost J, Belleville J. Serum lecithin: cholesterol acyltransferase activity and HDL2 and HDL3 composition in small for gestational age newborns. Acta Paediatr. 1997 May;86(5):528–532
    • View In Article
    • MEDLINE
    • CrossRef
56. Spencer JA, Chang TC, Crook D, Proudler A, Felton CV, Robson SC, et al. Third trimester fetal growth and measures of carbohydrate and lipid metabolism in umbilical venous blood at term. Arch Dis Child Fetal Neonatal Ed. 1997 Jan;76(1):F21–F25
    • View In Article
    • MEDLINE
    • CrossRef
57. Kaser S, Ebenbichler CF, Wolf HJ, Sandhofer A, Stanzl U, Ritsch A, et al. Lipoprotein profile and cholesteryl ester transfer protein in neonates. Metabolism. 2001 Jun;50(6):723–728
    • View In Article
    • Abstract
    • Full-Text PDF (114 KB)
    • CrossRef