Dysregulated Fatty Acid Metabolism in Preeclampsia Among Highland Andeans: Insights Into Adaptive and Maladaptive Placental Metabolic Phenotypes

Abstract

High-altitude pregnancy presents the complex physiological challenge of fulfilling maternal, placental, and fetal metabolic demands under chronic ambient hypoxia. Highland Andeans exhibit signs of adaptation to high-altitude hypoxia, showing relative protection against altitude-associated fetal growth restriction (FGR) and the positive selection of metabolic genes linked to placental mitochondrial capacity. Not all infants are protected, with both FGR and preeclampsia occurring among highland-resident Andeans. In Andeans, placental metabolic dysfunction is evident. By integrating metabolomic studies of maternal-placental-fetal triads with adaptive genetic signals in the fetal genome, we sought to identify adaptive and maladaptive placental metabolic phenotypes in highland Andeans (La Paz, Bolivia; 3850 m), including normotensive and preeclamptic pregnancies. Widespread differences in metabolite abundance were evident between normotensive and preeclamptic pregnancy across maternal, placental, and fetal compartments. Preeclampsia was characterized by a pronounced accumulation of fatty acid derivatives, specifically medium and long-chain acylcarnitines; these were also associated with low birth weight. Genotype-phenotype association analyses revealed novel links between putatively adaptive fetal haplotypes and placental metabolite abundance. Carriers of specific adaptive fetal haplotypes comprising genes linked to lipid metabolism had a greater abundance of placental short-chain acetyl-carnitine alongside decreased levels of linolenic acid (CPT2/LRP8), lower levels of the medium-chain octanoylcarnitine (EXOC4), and greater abundance of free carnitine (LIPG). Collectively, our study reveals a distinct metabolic phenotype in Andean preeclampsia characterized by incomplete fatty acid oxidation and highlights novel links between putatively adaptive fetal haplotypes and healthy placental metabolic phenotypes.

Keywords: adaptation; hypertensive disorders of pregnancy; hypoxia; metabolome.

FIGURE 1

Differential metabolomic profiles in preeclampsia (PE) across maternal and fetal compartments. (A) Analysis of metabolomic profiles across cord plasma (left), placenta (center), and maternal plasma (right) using orthogonal partial least squares discriminant analysis (OPLS‐DA) in control (CON, white) and preeclampsia (PE, blue). Each OPLS‐DA model was validated using a permutation (100) test and presented as permR² and permQ² values (B). Before analysis, values were normalized using pareto scaling and log transformation. Discriminant metabolites were identified from an S plot as ±1.5 SD from the mean. (C) The control vs. preeclampsia metabolite values for top model discriminants were tested using a T‐test followed by a false discovery rate correction two‐stage linear step‐up procedure of Benjamini, Krieger, and Yekutieli, Q = 1%. Significant values from this are presented in the heatmap as normalized values. N = 14 control, 14 preeclampsia (PE).

FIGURE 2

Abundance of acylcarnitines (AC) and fatty acids (FA) in umbilical cord plasma, placenta, and maternal plasma in normotensive (control) and preeclamptic (PE) pregnancy. (A) Short‐chain AC, carbon chain lengths (C) 2–5 were raised in maternal plasma in PE. (B) Medium‐chain AC (C6‐12), were raised across cord plasma, placenta, and maternal plasma in PE. (C) Long‐chain AC (C13‐20) were raised in cord plasma and placenta in PE. (D) Long‐chain fatty FAs (C14‐22) were raised in cord plasma in PE. Peak intensities were normalized using Pareto scaling and log transformation. Control and preeclamptic values were compared using an unpaired Student's t‐test. Significant differences in metabolite abundance between control and PE pregnancies are indicated by an asterisk, *p < 0.05. n = 14 control (white), 14 preeclampsia (PE, blue).

FIGURE 3

Relationship of medium‐chain acylcarnitine abundance across compartments within individuals. Significant within‐subjects associations were observed between maternal plasma and cord plasma (A), maternal plasma and placenta (B), and cord plasma and placenta (C). Statistical analysis was conducted using simple linear regression adjusted to gestational age. n = 14 controls (white), n = 13 preeclamptic (PE; blue).

FIGURE 4

Relationship between medium‐chain acylcarnitine (AC) abundance and birth weight. Birthweight was inversely associated with medium chain AC levels in maternal plasma (A), placenta (B), and cord plasma (C). Statistical analysis was conducted using simple linear regression adjusted to gestational age. n = 13 controls (white), n = 14 PE (blue).

FIGURE 5

Genotype–phenotype association analyses reveal novel links between adaptive fetal haplotypes and the abundance of placental metabolites. Scatterplots illustrate the relationship between putatively adaptive haplotype copy number (0, 1, 2) and placental metabolite abundance. The associated placental metabolites included: Fatty acids (FA) carbon chain lengths 18:3 (A, B), acylcarnitines (AC) carbon chain lengths 2:0 and 8:0 (C, D), L‐carnitine (E). Linear regression analyses identified associations; p value < 0.01 was considered significant. n = 28 control (white), n = 22 preeclampsia (PE, blue). (F) Table detailing the fetal haplotype location, including chromosome (Chr) and top marker position (hg19 reference genome), the gene within the 200 kb haplotype region most functionally relevant to lipid metabolism (prioritized gene), and the iHS score. Abbreviations: Carnitine palmitoyl transferase 2 (CPT2), low density lipoprotein receptor‐related protein 8 (LRP8), exocyst complex component 4 (EXOC4), lipase G endothelial type (LIPG), nudix hydrolase 19 (NUDT19). An asterisk distinguishes haplotypes containing the same prioritized genes (CPT2/LRP8) *.

FIGURE 6

Adaptive placental metabolic signals associate with the expression of mitochondrial unfolded protein response proteins. Acylcarnitine (AC) 2:0 Abundance was negatively associated with (A) heat shock protein 60 (HSP60) and positively associated with (B) activating transcription factor 5 (ATF‐5) and (C) caseinolytic peptidase P (CLPP). Placental protein expression was normalized to β‐Actin. Statistical analysis was performed using simple linear regression. A‐C, n = 5 control (white), 7 preeclampsia (PE, blue).