Lipidomics of allantoic fluid collected by allantocentesis indicates a possible mechanism underlying partial litter loss in multifetal pregnancies of ewes

Abstract

Fetal mortality significantly impacts reproductive efficiency in sheep, with approximately 30% of fetuses failing to survive until lambing. In multifetal gestation, partial litter loss (PLL) occurs in 28% to 37% of cases. The biochemical composition of allantoic fluid (ALF), which plays a key role in metabolic regulation and fetal waste clearance, has been shown in other species to reflect fetal development and maturity. However, ALF metabolomics in sheep, especially in relation to PLL, remains unexplored. This study investigated late-gestation fetal mortality in multifetal sheep pregnancies by analyzing maternal metabolic status, placental gene expression, and ALF composition to identify potential metabolic compounds associated with PLL. The study was conducted on 25 ewes, of which 14 experienced PLL, and 11 had vital litters (VL). Plasma analysis revealed that VL ewes have higher β-hydroxybutyrate concentrations than PLL ewes, whereas plasma insulin and cortisol levels are higher in PLL ewes. Analysis of gene expression in the placental cotyledon demonstrated increased androgen receptor expression in PLL pregnancies and trends toward elevated solute carrier family 2 member 1, and reduced nuclear receptor subfamily 3 group C member 1 expression. Insulin-like growth factor 2 expression tended to be higher in VL placentas. Allantocentesis was performed on 25 ewes with multifetal pregnancies (11 VL and 14 PLL). ALF samples were collected from 74 fetuses at 140.4 ± 2.0 days in pregnancy, and 41 samples were selected for metabolomic profiling. Using genomic analysis, 25 ALF samples were confidently matched to individual lambs. Untargeted lipidomics of ALF from 14 lambs in PLL (7 born alive vs. 7 stillborn) revealed 24 differentially abundant lipid compounds. Stillborn lambs exhibited lower phosphatidylglycerol (PG) 15:0_20:1, but higher levels of sphingomyelin (SM), lysophosphatidylcholine (LPC), ether-linked phosphatidylethanolamines (PEs), lysophosphatidylethanolamines (LPEs), and monogalactosyldiacylglycerols (MGDGs). Collectively, PLL in multifetal ewe pregnancies is linked to maternal metabolic imbalance, elevated cortisol levels, dysregulated placental gene expression, and altered ALF lipid profiles in stillborn fetuses. Elevated SM and LPC, and lower PG, suggest immature lungs. These molecular and metabolic signatures may provide new insights into the mechanisms underlying PLL in late-gestation multifetal pregnancies.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-44636-9.

Introduction

Reproductive inefficiency due to fetal mortality is a major challenge in sheep production systems, with approximately 30% of fetuses failing to survive until lambing¹. An early study in multiparous Romney Marsh ewes reported embryonic and fetal loss rates of 27.0%, 18.1%, and 20.0% at 18, 30, and 140 days of gestation, respectively, whereas nulliparous ewes showed corresponding losses of 20.9%, 15.9%, and 22.0%². More recent data from mixed-breed flocks (primarily Dorset and Suffolk) indicate that 19.6% of fetuses were lost between day 25 of gestation and term, with 37% of multifetal pregnancies undergoing partial litter loss (PLL), 17.7% of these during the final trimester³. In our previous retrospective study, we recorded 675 lambings and reported PLL in 28% of multifetal pregnancies⁴.

Multifetal gestation in sheep is strongly associated with intrauterine growth restriction (IUGR), which compromises nutrient supply and fetal development. The average birth weight of lambs born from litters of two, three, or four is approximately 83%, 70%, and 63% that of singleton lambs, respectively⁵. This reduction is largely attributed to limited placental capacity. As the number of fetuses increases, the number of placentomes per conceptus declines. Although compensatory enlargement of placentomes occurs, it cannot match the total placental mass of a singleton pregnancy⁶. The resulting reduction in placental efficiency likely contributes to IUGR and may increase the risk of fetal demise late in gestation.

In humans, twins account for 98% of multifetal pregnancies and are linked to elevated risks of perinatal morbidity and mortality, with single intrauterine fetal demise occurring in 3.7–6.8% of twin pregnancies⁷. Afec-Assaf sheep, which carry the Booroola fecundity gene (FecB), exhibit a high incidence of multifetal pregnancies, with twin gestations comprising ~ 34% of cases⁸. As in humans, the mechanisms underlying late-gestation fetal demise or PLL in sheep remain unclear, particularly in fetuses that appear morphologically normal at birth and fall within the expected body weight (BW) range.

Allantoic fluid (ALF) is crucial for fetal waste removal and metabolism during gestation. In equine pregnancies (270–295 days), nuclear magnetic resonance-based profiling identified 28 metabolites in the ALF, with significant enrichment of betaine, creatine, creatinine, citrate, histidine, tryptophan, and π-methylhistidine compared to amniotic fluid (AF)⁹. In avian models, ALF metabolomics on days 10, 14, and 17 of embryogenesis revealed stage and line-specific differences in several metabolites, reflecting divergent metabolic trajectories¹⁰. In canine fetuses, fetal growth categories are associated with dynamic alterations in ALF biochemical markers¹¹. These findings demonstrate that ALF composition not only changes over time but also mirrors fetal development, metabolism, and maturity. However, despite its relevance to fetal mortality and PLL, sheep-specific ALF metabolomics remained largely unexplored.

In this study, we investigated the PLL phenomenon in multifetal pregnancies in sheep, focusing specifically on late-gestation fetal death in morphologically normal lambs. We hypothesized that late-gestation fetal death is a consequence of maternal metabolic failure that downstream affects the fetus’s nutrient supply and survival. We further evaluated the relationship between the maternal metabolic state, placental gene expression, and the biochemical composition of the ALF to reveal new insights into the etiology of PLL in sheep. To this aim, we performed allantocentesis in 25 pregnant ewes in late gestation [~ 140 days in pregnancy (DIP)] and collected blood and placental cotyledons.

Results

In this study, we followed the metabolic and hormonal state of 26 multifetal ewes from 80 DIP until lambing and determined the relative expression of genes in the placenta of 14 multifetal ewes. In addition, we analyzed ALF metabolomics at 140.4 ± 2.0 DIP. One dam (out of 26) died of infection after the allantocentesis; all its fetuses were stillborn and were excluded from the analysis. In total, 82 lambs were born from 25 ewes, and allantocentesis was performed on 74 fetuses. The ewes and litters were divided postpartum into two groups according to their litter fate: (1) vital litter (VL), in which all lambs were born alive (n = 11); (2) PLL, in which at least one lamb was born dead (n = 14).

Forty-one ALF samples (20 VL and 21 PLL) were randomly selected for further analysis. The 41 ALF samples and 38 ear tissue samples were subjected to DNA analysis to match the ALF samples and the newborn lambs. From those samples, 25 lambs were confidently identified and matched with their ALF. The metabolomics analysis in the current paper focuses only on the PLL litters (n = 14), which were divided into two subgroups: (I) liveborn lambs (LB; n = 7), and (II) stillborn lambs (SB; n = 7).

We also analyzed the lambing data and blood plasma metabolites by the vitality status of the litter (VL vs. PLL), as defined above (11 VL vs. 14 PLL), or the placenta data (7 VL vs. 7 PLL). The study scheme is summarized in Fig. 1.

Fig. 1.

Study Scheme. Allantois fluid (ALF), Born alive (BL), Days in pregnancy (DIP), Partial litter loss (PLL), Stillborn (SB), Vital litter (VL). Created in BioRender. Daddam, J. (2025) https://BioRender.com/mdcaj4y.

Lambing outcomes

The lambing data for the 25 lambings are presented in Table 1. We did not observe differences in pregnancy outcome parameters, but the dams’ BCS at the allantocentesis tended to be lower by 0.2 units for the PLL. The male rate in litters tended to be higher in the PLL than in the VL group (67.9% and 50.7%, respectively; P = 0.11).

Table 1.

Pregnancy outcomes by litter vitality.

Parameter	VL	PLL	SEM	P-value
N	11	14
Pregnancy duration, days	144.4	143.4	0.45	0.14
Dam BW, kg	102.8	101.3	3.19	0.74
Dam BCS, units	3.0	2.8	0.08	0.08
Dam BW postpartum, kg	77.6	81.5	3.44	0.43
Litter size	3.1	3.5	0.17	0.12
Litter male rate, %	50.7	67.9	7.42	0.11
Litter survival rate	1.00	0.54	0.05	< 0.001
Total litter weight, kg	11.8	11.8	0.50	0.94
Male lamb BW (kg)	4.1	3.7	0.20	0.14
Female lamb BW (kg)	3.5	3.1	0.20	0.23
Male lamb CRL (cm)	52.2	51.1	1.02	0.45
Female lamb CRL (cm)	49.9	48.4	1.11	0.35

VL – vital litter; PLL – partial litter loss; BW – body weight; BCS – body condition score; CRL, crown-to-rump length.

Dam plasma metabolite and hormone concentrations

The dams’ plasma metabolite and hormone concentrations are presented in Table 2. The BHB concentration was 2.9-fold higher in the VL vs. PLL ewes. We also observed numerically 17% higher plasma glucose and 31% lower plasma NEFA concentrations in the PLL compared to the VL ewes. Insulin concentration was 1.9-fold higher in the PLL vs. VL ewes, and cortisol concentration was 1.6-fold higher in the PLL vs. VL ewes (Fig. 2). We did not observe any differences in lactate, TG, or testosterone concentration between groups.

Table 2.

Dam’s plasma metabolites and hormones at 140.4 ± 2.0 days in pregnancy by litter vitality.

Parameter	VL	PLL	SEM	P-value
Glucose, mg/dL	41.1	48.1	3.1	0.17
BHB, mg/dL	11.9	4.1	2.2	0.05
NEFA, µEq/L	768.0	531.9	138.0	0.29
Insulin, µIU/mL	25.7	47.9	5.6	0.03
Lactate, mmol/L	1.4	1.9	0.5	0.45
TG, mg/dL	20.4	24.5	2.5	0.32

VL – vital litter; PLL – partial litter loss; BHB – β-hydroxybutyrate; NEFA – nonesterified fatty acid; TG – triglyceride.

Fig. 2.

Dam plasma concentrations of cortisol and testosterone at 140.4 ± 2.0 days in pregnancy by litter vitality: vital litter (VL; n = 11) or partial litter loss (PLL; n = 14). The error bars represent the SEM. Asterisks above the bars indicate a significant effect of treatment: **P < 0.01.

Placental cotyledon gene expression

Relative mRNA expression levels were quantified in placental cotyledon tissue derived from 7 VL and 7 PLL pregnancies to investigate the differential expression of candidate genes involved in placental function. The six target genes analyzed were CYP19, SLC2A1, NR3C1, IGF2, CRHR1, and AR. The results are presented in Fig. 3. AR expression was markedly elevated in PLL (2.60 ± 0.63) compared to VL (0.35 ± 0.69; P = 0.05) samples. The SLC2A1 expression tended to be higher in PLL vs. VL placentas (5.74 ± 1.30 and 1.38 ± 1.55, respectively; P = 0.07). Expression of NR3C1, encoding the glucocorticoid receptor, showed a tendency to be lower in PLL (0.156 ± 0.08) vs. VL (0.389 ± 0.09; P = 0.10) placentas. A tendency toward higher IGF2 expression was observed in VL (2.25 ± 0.75) vs. PLL (0.34 ± 0.69; P = 0.10) placentas.

Fig. 3.

Relative expression of placental genes by litter vitality: vital litter (VL; n = 7) or partial litter loss (PLL; n = 7). The error bars represent the SEM. Asterisks above the bars indicate a significant effect of treatment: *P < 0.05, † 0.05 < P ≤ 0.10.

Allantois metabolomics

Metabolomics analysis of the hydrolyzed ALF samples revealed 338 polar compounds, 388 lipids, and 18 steroids. In a general comparison of VL to PLL, we found differences in 41 polar, 12 lipid, and 4 steroid compounds (P < 0.05). In this paper, we focus on the lipids in lambs born to PLL pregnancies: 7 LB vs. 7 SB. Untargeted lipidomic profiling of the ALF collected from multifetal pregnancies identified significant differences in 24 lipid compounds between LB and SB fetuses (P < 0.05; Table 3; Fig. 4). These differentially expressed lipids spanned several classes, including phosphatidylglycerol (PG), sphingomyelin (SM), ether-linked phosphatidylethanolamines (PEs), lysophosphatidylcholine (LPC), lysophosphatidylethanolamines (LPEs), monogalactosyldiacylglycerols (MGDGs), and monogalactosylmonoacylglycerols (MGMGs) (Table 3). The PG (15:0_20:1) was lower (P = 0.04) in the ALF of the SB vs. LB fetuses, with a negative estimate. Four SM compounds: SM (t42:1; P = 0.02), SM (t40:1; P = 0.03), SM (d39:1; P = 0.04), and SM (d40:1; P = 0.05) were elevated in the SB group fetuses. Moreover, 11 SM compounds tended to be higher in the SB vs. LB ALF (P < 0.1). Seven PEs: (16:0p_18:1), (16:0p_17:1), (20:1e_20:4), (16:0p_16:1), (16:1e_18:1), (16:1e_17:1), and (18:1e_18:1) were significantly higher (P = 0.03, P = 0.04, P = 0.04, P = 0.04, P = 0.04, P = 0.05, P = 0.05, respectively) in the SB vs. LB fetuses (Table 3). One LPC (18:2) was elevated (P = 0.05) in the SB fetuses compared to the LB fetuses. Five LPE-type lipids were higher in the ALF of the SB fetuses compared to the LB group, with a positive estimate: (18:1) (2), (20:4) (2), (16:1), (20:3) (1), and (18:1) (1) (P = 0.001, P = 0.003, P = 0.02, P = 0.03, P = 0.04, respectively). MGDG (45:8; P = 0.002) and MGDG (44:8; P = 0.03) showed the largest absolute effect sizes, indicating markedly higher levels in SB vs. LB lambs. MGMG (31:0; P = 0.03) was also elevated in the SB vs. LB fetuses (Table 3).

Table 3.

Comparison of lipid compounds between SB and LB fetuses.

Compound common name	Abbreviation	KEGG ID	Estimate	P-value
Ether-linked phosphatidylethanolamines (16:0p_16:1)	PE (16:0p_16:1)	C00350	1.417	0.040
Ether-linked phosphatidylethanolamines (16:0p_17:1)	PE (16:0p_17:1)	C00350	1.481	0.035
Ether-linked phosphatidylethanolamines (16:0p_18:1)	PE (16:0p_18:1)	C00350	1.221	0.031
Ether-linked phosphatidylethanolamines (16:1e_17:1)	PE (16:1e_17:1)	C00350	1.513	0.045
Ether-linked phosphatidylethanolamines (16:1e_18:1)	PE (16:1e_18:1)	C00350	1.173	0.044
Ether-linked phosphatidylethanolamines (18:1e_18:1)	PE (18:1e_18:1)	C00350	1.156	0.049
Ether-linked phosphatidylethanolamines (20:1e_20:4)	PE (20:1e_20:4)	C00350	7.439	0.037
Lysophosphatidylethanolamine (16:1)	LPE (16:1)	C04438	2.019	0.020
Lysophosphatidylethanolamine (18:1) (1)	LPE (18:1) (1)	C04438	1.228	0.038
Lysophosphatidylethanolamine (18:1) (2)	LPE (18:1) (2)	C04438	1.726	0.001
Lysophosphatidylcholine (18:2)	LPC (18:2)	C04230	8.228	0.046
Lysophosphatidylethanolamine (20:3) (1)	LPE (20:3) (1)	C04438	1.253	0.029
Lysophosphatidylethanolamine (20:4) (2)	LPE (20:4) (2)	C04438	2.177	0.003
Monogalactosyldiacylglycerol (44:8)	MGDG (44:8)	C03692	9.713	0.030
Monogalactosyldiacylglycerol (45:8)	MGDG(45:8)	C03692	12.174	0.002
Monogalactosylmonoacylglycerol (31:0)	MGMG (31:0)	C04315	7.286	0.031
Octadecanoic acid	FA (18:0)	C01530	−0.522	0.025
Phosphatidylglycerol (15:0_20:1)	PG (15:0_20:1)	C00344	−2.741	0.043
Phosphatidylserine (33:0)	PS (33:0)	C02737	1.364	0.037
Triglyceride (18:1_18:2_20:4)	TG (18:1_18:2_20:4)	C00422	−0.427	0.009
Sphingomyelin (d39:1)	SM (d39:1)	C00550	1.310	0.037
Sphingomyelin (d40:1)	SM (d40:1)	C00550	1.308	0.049
Sphingomyelin (t40:1)	SM (t40:1)	C00550	0.574	0.031
Sphingomyelin (t42:1)	SM (t42:1)	C00550	1.447	0.021

Fig. 4.

Forest plot for allantoic fluid (ALF) lipids. Fold change for lipids in the stillborn (SB; n = 7) vs. liveborn (LB; n = 7) ALF. Asterisks indicate a significant effect of treatment: *P < 0.05, ** P < 0.01, *** P < 0.001. Lipid abbreviations are defined in Table 3.

Discussion

Although PLL has been observed in up to 37% of multifetal pregnancies in sheep³, the underlying pathophysiology remains incompletely understood. In this study, we demonstrate that PLL is associated with altered maternal lipid metabolism, increased maternal cortisol, dysregulated placental gene expression, and differential fetal ALF lipid signatures, highlighting a complex interplay of endocrine and metabolic factors in fetal demise. This study provides novel insights into the physiological, biochemical, and molecular profiles associated with PLL in multifetal sheep pregnancies.

Dam metabolic alterations in PLL pregnancies

Ewes that experienced PLL had significantly lower plasma BHB concentrations, suggesting a shift from ketogenesis and possibly impaired lipid oxidation during late gestation. Ketone bodies, such as BHB, are critical for sustaining fetal energy demands, particularly in ruminants with relatively limited glucose availability¹².

Insulin concentrations were 1.9-fold higher in the PLL ewes compared to the VL ewes. Previous research¹³ reported a marked decrease in insulin concentrations as litter size increased, with ewes carrying four fetuses exhibiting insulin levels that were five times lower than those carrying a single fetus. This reduction occurred several weeks prepartum, suggesting that declining insulin may function as homeorhetic regulation to preferentially spare glucose for the fetoplacental units rather than maternal tissues¹³. In the context of our findings, PLL ewes, having lost at least one fetus, would experience a reduced fetoplacental glucose requirement, potentially resulting in higher circulating insulin than in VL ewes that are still supporting all fetuses. This finding is in accordance with the low BHB concentration and numerically higher glucose and lower NEFA levels observed in the plasma of the PLL vs. VL ewes; this might be associated with the lower glucose requirement, resulting in reduced fat mobilization into the liver, and most likely lower BHB production. The higher insulin concentrations in PLL litter dams may reflect a protective mechanism that favors the dam’s survival over the fetuses in cases of severe negative energy balance, as demonstrated by the tendency toward lower BCS in PLL ewes.

Notably, maternal plasma cortisol levels were 1.6-fold higher in PLL vs. VL ewes. Elevated maternal glucocorticoids during late pregnancy are well-documented to impair placental function, reduce utero-placental blood flow, and increase fetal exposure to bioactive cortisol^14,15. In sheep, prenatal glucocorticoid exposure has been shown to alter fetal cardiovascular and metabolic development, increase oxidative load, and contribute to stillbirth or IUGR^16,17.

Placental gene dysregulation and fetal survival

Gene-expression analysis in placental cotyledon tissues revealed markedly increased AR expression in PLL pregnancies. This observation is consistent with previous studies, which have linked elevated placental androgen signaling to inflammation and compromised angiogenesis¹⁸.

SLC2A1 [encoding glucose transporter 1 (GLUT1)] expression tended to be elevated in PLL placentas. While GLUT1 is essential for basal glucose transport to the fetus¹⁹, its overexpression may reflect compensatory upregulation in response to nutrient or oxygen insufficiency. However, its expression is not often altered in IUGR contexts, indicating that compensatory upregulation of GLUT1 may not be a reliable mechanism for adapting to nutrient or oxygen insufficiency^20,21. Concomitantly, NR3C1 and IGF2 tended to be downregulated in PLL placentas. The NR3C1 encodes the glucocorticoid receptor, which mediates cortisol responses; its reduced expression may impair placental adaptation to stress²². It is plausible that sustained high cortisol levels detected in these pregnant dams’ plasma induce a downregulation of its receptor in the placenta as a compensatory mechanism to attenuate excessive glucocorticoid signaling and protect placental and fetal tissues. The IGF2, a key regulator of placental growth and vascularization, is often reduced in compromised pregnancies and correlates with fetal growth restriction and reduced survival^23,24. Collectively, these alterations in gene expression suggest that PLL is characterized by a maladaptive placental phenotype marked by excessive androgen signaling, compensatory glucose transport, and impaired growth and stress pathways, which could underlie reduced fetal survival.

ALF lipids and fetal outcome

The biochemical composition of the ALF and amniotic fluid (AF) has been investigated in sheep during the last trimester of pregnancy^25,26. In a previous study in sheep, the concentrations of TG and cholesterol were found to be similar in the ALF and AF between 121 and 145 DIP²⁷. In another study in sheep, no differences were reported between ALF and AF in the concentrations of glucose, creatinine, total protein, urea, cholesterol, or enzymes between 121 and 140 DIP²⁸, and similar results were reported in goats in late gestation²⁹. Although there are similarities between ALF and AF biochemical compositions^25–27, there are also differences; however, to our knowledge, there is no report on differences in the lipid profiles of ALF and AF in the final stages of pregnancy in sheep.

Metabolomics profiling of ALF revealed lipid differences between SB and LB lambs in PLL pregnancies. The PG (15:0_20:1) was 2.7-fold lower in the SB vs. LB ALF in this study. One well-documented essential function of PG in mammals is in the developing fetal lung³⁰. PG is a critical component of pulmonary surfactant, the lipid-protein mixture that lines the alveoli and enables breathing at birth. It is the second-most abundant phospholipid in mature lung surfactant (comprising ~ 10% of surfactant lipids)^30–32, a proportion far higher than its content in typical cell membranes^33–35. During late gestation, the fetal lung begins secreting surfactant rich in PC and PG. PG appears in the AF only in the final weeks of pregnancy and rises sharply in the last 10–14 days before term³⁶. The emergence of PG in AF is a hallmark of lung maturity, and its absence is a known risk factor for neonatal respiratory distress syndrome^37,38. Clinically, a lack of PG in late-gestation AF correlates with the development of respiratory distress syndrome in preterm infants, underscoring PG’s crucial role in perinatal lung function^39,40. In addition, PG in the pulmonary surfactant can suppress inflammation and even directly inhibit certain respiratory viruses⁴¹. Taken together, these findings demonstrate that PG, as a class, provides indispensable support for fetal lung maturation. Therefore, we suggest that the lower abundance of ALF PG (15:0_20:1) in the PLL indicates surfactant deficiency and impaired function, which might cause intrauterine death at term.

In our study, PC (36:3e) tended to be lower (− 8.2 fold; P = 0.1) in the SB vs. LB group. Another indication of surfactant functionality and lung maturation is an abundance of lecithin (PC), SM, and the ratio of lecithin to SM in AF^42,43. In the last weeks of an uncomplicated pregnancy in ewes and other mammals, lecithin concentration increases in the AF, while the concentration of SM remains stable or decreases^37,44,45. The PC is the predominant phospholipid comprising pulmonary surfactant³¹. However, the differences between SM composition in SB and LB fetuses were more significant: out of 43 compounds recognized in the analysis, 4 of them - SM (t42:1), SM (t40:1), SM (d39:1), and SM (d40:1) - were significantly elevated, and 11 tended to be higher in the SB vs. LB ALF. These results might indicate impaired lung maturation in the SB fetuses.

Seven PEs were significantly higher in SB vs. LB ALF. The PEs constitute a small fraction of surfactant lipids, but research indicates their outsized functional importance. In the context of surfactant, they facilitate the formation of special surfactant superstructures, such as tubular myelin, a lattice-like lipid structure in the alveolar fluid that aids in surfactant-film dispersion⁴⁶. Experimental evidence shows that including plasminogen-PE in model surfactant mixtures enhances the formation of these structures and supports membrane-fusion events, which are critical during the recycling and spreading of surfactant over the alveolar surface³². These results indicate that plasmalogens dramatically improve the biophysical efficacy of surfactant films. However, in healthy ewe pregnancies, PE concentrations are reduced in the AF during pregnancy⁴⁴, which was not the case in our study, in which PEs were more abundant in the SB vs. LB fetuses’ ALF, which might be a result of paused lung maturation.

The LPC (18:2) in the SB fetus ALF was higher than in the LB fetus ALF. LPC also plays a pivotal role in fetal lung maturation by serving as a key intermediate in the biosynthesis of surfactant PC, particularly dipalmitoyl-PC, through the LPCAT1-mediated remodeling pathway^47,48. LPCAT1 activity increases in late gestation and is stimulated by antenatal corticosteroids, enhancing surfactant production^49,50. Disruptions in this pathway, such as those seen in diabetic pregnancies or genetic variants in LPCAT1, have been associated with impaired surfactant synthesis and respiratory distress syndrome^51,52. Beyond its biosynthetic role, LPC also reflects fetal metabolic and inflammatory status. Importantly, LPCAT1 mRNA in AF and protein in maternal plasma correlate with traditional markers of fetal lung maturity, positioning LPC-related pathways as promising biomarkers for assessing both fetal pulmonary readiness and vitality^53,54. These findings support our observation of elevated (8-fold) LPC in the SB ALF, which indicates impaired biosynthesis of PC, which is important for surfactant functionality.

Collectively, the findings on PG, SM, ether-linked PE, and LPC support our assumption of impaired lung development in SB fetuses. Moreover, the reduced expression of NR3C1 in the placenta of PLL litters can explain this, as this gene encodes glucocorticoid receptors that are part of the lung maturation and surfactant secretion-signaling mechanism. However, at this stage of the study, we cannot determine if the impaired lung development in SB fetuses is a cause or an effect of the PLL phenomenon in multifetal pregnancy.

Sex bias and vulnerability

Although this study did not primarily focus on fetal sex, PLL litters exhibited a numerically higher rate of males. Previous studies have consistently shown that male fetuses are more susceptible to hypoxia, oxidative stress, and placental insufficiency in both humans and animals^55,56. Male vulnerability may stem from reduced placental plasticity, differential endocrine signaling, or increased nutrient demands. In our previous study in sheep, male co-twins were shown to negatively impact the survival and growth of their female littermates, indicating competitive disadvantage or sexually antagonistic interactions⁴.

Verification of the study hypothesis

The central hypothesis of this study was that partial litter loss (PLL) in multifetal sheep pregnancies reflects unrecognized late-gestation intrauterine fetal demise in morphologically normal fetuses, driven by an unfavorable maternal-placental-fetal metabolic environment. The present findings provide converging evidence that largely supports this hypothesis. Specifically, PLL pregnancies were characterized by maternal metabolic and endocrine alterations, including reduced β-hydroxybutyrate and elevated insulin and cortisol concentrations, alongside dysregulated placental gene expression indicative of impaired stress adaptation and growth signaling. Importantly, stillborn fetuses within PLL litters exhibited a distinct ALF lipidomic signature, marked by reduced PG and increased SM, LPC, and PE patterns associated with delayed pulmonary maturation. Together, these results indicate that fetal demise in PLL pregnancies is not a stochastic event occurring at parturition, but rather the outcome of progressive intrauterine compromise developing during late gestation. However, the cross-sectional nature of the study does not allow definitive discrimination between causative and consequential mechanisms. Nevertheless, the alignment between maternal metabolic stress, placental maladaptation, and fetal ALF lipid signatures strongly supports the proposed hypothesis that PLL represents a biologically structured, late-gestation pathophysiological process rather than an incidental periparturient loss.

Strengths and limitations

This study provides a comprehensive analysis of PLL in sheep by integrating endocrine, transcriptomic, and metabolic data. However, the relatively small sample size and partial matching between ALF samples and lambs limit broader generalizability. Future studies should expand sample size and consider integrating single-cell RNA-sequencing of placental tissue and targeted metabolomics to better resolve fetal and maternal contributions to PLL. In addition, allantocentesis was performed at 140.4 ± 2.0 DIP, whereas lambing occurred at 144.6 ± 1.4 DIP; consequently, fetal vitality was not monitored during the interval between the procedure and lambing, making it impossible to determine accurately whether fetal death occurred antenatally due to IUGR or intrapartum because of dystocia.

Conclusions

PLL in multifetal sheep pregnancies is associated with maternal metabolic disruption, elevated cortisol, placental gene dysregulation, and specific lipidomic alterations in the ALF of compromised fetuses. These findings suggest that PLL reflects a cumulative effect of suboptimal maternal-fetal association. Metabolomics of ALF obtained using allantocentesis revealed lower PG and elevated SM, PE, and LPC in SB compared to LB fetuses, indicating immature lungs in the former. These molecular and metabolic signatures may provide new insights into the mechanisms underlying PLL in late-gestation multifetal pregnancies, and warrant further investigation of, among others, sex-specific and placental mechanisms driving fetal demise in multifetal gestations.

Materials and methods

Animals

The Volcani Center Animal Care Committee approved all procedures performed on the animals in the present study (approval no. 981/23 IL), and all methods were performed in accordance with the relevant guidelines and regulations, and with the ARRIVE guidelines. The experiment was conducted with the experimental flock at the Agricultural Research Organization Volcani Center in Rishon LeZion, Israel, comprising the Afec-Assaf breed. The prolific Afec-Assaf strain is considered a dual-purpose breed and was developed by introgression of the B allele of the FecB (Booroola) locus into the Assaf breed⁸. Flock reproductive management included three 40-day annual breeding periods per year when each ewe was hand-mated with a single ram following hormonally synchronized estrus. Lamb birth weights, sex, and vitality were determined and recorded within a few hours after lambing.

Study design

We followed 25 multiparous Afec-Assaf ewes in two clusters, each comprised of different animals. In the first cluster (n = 11), ewes were mated from the end of July through mid-August and lambed from late December through mid-January, i.e., the winter season in Israel. In the second cluster (n = 14), ewes were mated in mid-November through mid-December and lambed from April through mid-May, i.e., the spring season in Israel. The flock was kept in open sheds under non-dairy management. All ewes were housed in one pen and group-fed the same diet ad libitum. Pregnancy was determined in mated ewes on day 35 post-mating by transabdominal ultrasonography (3100v, Shenzhen Well.d Medical Electronics, China). We determined the litter size of pregnant sheep at ~ 60 DIP (Mindray Z6; Shenzhen Mindray Bio-Medical Electronics, China) and selected 26 ewes carrying 2 to 4 fetuses (average litter size 3.3).

Blood samples were taken from the dams at 0630 h on ~ 80 DIP, 138 DIP, and on the day of the allantocentesis procedure (140.4 ± 2.0 DIP). The samples were taken from the jugular vein into vacuum tubes containing lithium heparin (BD Vacutainer, Becton Dickinson, UK) and immediately placed on ice. The plasma was separated by centrifugation for 15 min at 1000g, divided into two tubes, and stored at −80 °C pending analysis.

ALF collection

We collected ALF by an allantocentesis procedure using ultrasound guidance at 140.4 ± 2.0 DIP (the mean pregnancy length was 144.6 ± 1.4 days). Anesthesia was induced with an intravenous injection of propofol (3 mg/kg; Propofol-lipuro, B. Braun Medical Supplies, Philippines) and diazepam (0.15 mg/kg; Assival, Teva Pharmaceutical Industries, Israel), followed by tracheal intubation and maintenance with 2.5% isoflurane (Forane, Abbott Laboratories, England). Sheep were ventilated, and heart rate, oxygenation, and depth of anesthesia were continuously monitored throughout the procedure. The sheep were restrained in a left lateral position, and the ventral/right abdominal skin was shaved and disinfected with iodine solution and 70% ethanol. The uterus, pregnancy sacs, and fetuses were examined using ultrasound (Mindray Z6), and an optimal site for ALF aspiration was selected to avoid damage to the fetus. The aspiration was performed under ultrasound guidance using a spinal needle (18G x 3.5 inches or 21G x 3.5 inches) and a 10-mL syringe. The needle was inserted through the abdominal wall at a 90° angle and advanced until it penetrated the uterine wall and fetal membranes. A sample of a volume of 10 mL was aspirated from the allantois sac of each fetus individually and immediately placed on ice.Afterwards, the sheep were restrained in a right lateral position, and the same procedure was performed on fetuses located on the left side of the uterus. The ALF was then separated by centrifugation for 15 min at 1000×g, divided into four tubes, and stored at −80 °C pending analysis. The sediment was separated and stored at −80 °C pending DNA analysis.

At the end of the aspiration, 10 mL of solution containing blue surgical marker (Indigo carmine, 4 mg/mL, S.A.L.F S.p.A. LABORATORIO FARMACOLOGICO, Italy) was diluted in saline at a final concentration of 0.64% and injected into the allantois sac to avoid double aspiration from the same fetus. After the needle was withdrawn, the sampling site was disinfected again with 70% ethanol.

At lambing, the number of lambs born, their vitality status, and sex were recorded. The lambs’ birth weights and crown-to-rump length were recorded within a few hours after lambing. For genetic matching between the ALF samples and the newborn lamb, ear tissue samples of each lamb were collected immediately after lambing and stored at −20 °C until DNA purification using a DNeasy Kit (QIAGEN, USA). The DNA samples were sent to an external laboratory (Neogene, Scotland) to perform a genetic array (GeneSeek^® Genomic Profiler™ ovine 50 K, Illumina, USA).

Samples from placental cotyledons were collected from 14 ewes, one cotyledon from each placenta, for a total of 14 cotyledons, and placed on ice immediately after lambing. The cotyledon samples were cleaned, frozen in liquid nitrogen, and stored at −80 °C pending analysis. Dams’ BW and body condition score (BCS) were recorded on the day of the allantocentesis and within 24 h after lambing. The gestation length for each ewe was calculated based on mating and lambing records.

Metabolomics analysis

Metabolite extraction

Extraction and analysis of 200 µL of ALF (including free steroids and thyroid hormones) and polar metabolites were performed as previously described^57,58. The following internal standards were used in 1 mL of extraction solution: 0.1 µg/mL phosphatidylcholine (PC) (17:0/17:0) (Avanti), 0.4 µg/mL phosphatidylethanolamine (PE) (17:0/17:0, 0.15 nmol/mL of Ceramide/Sphingoid Internal Standard Mixture II; Avanti, LM6005), 0.0267 µg/mL d5-TG Internal Standard Mixture I (Avanti, LM6000), 0.1 µg/mL palmitic acid-¹³C (Sigma, 605573) and 10 µL Steroids Internal Standards Mix, 1 µg/mL in 50:50 MeOH: DDW (d4_estrone, d5_estradiol, d3_estriol, d9-progesterone, d8-cortisone). During the extraction, the following internal standards were added for the polar phase: ¹³C- and ¹⁵N-Labeled Amino Acids Standard Mix (Sigma, 767964; 1:1500). For analysis, the dried lipid extracts were resuspended in 50 µL Acetonitrile: Isopropyl Alcohol (ACN: IPA) (7:3) for injection. Dried polar samples were resuspended in 120 µL methanol: DDW (1:1).

Lipid extracts, identification, and quantification

Lipids and polar metabolites were analyzed as described in Bendikov-Bar et al. (2021)⁵⁹. Orbitrap data were analyzed using LipidSearch™ software (Thermo Fisher Scientific, MA, USA; version. 4.2)⁶⁰. The putative identification of lipids was validated by comparison to an in-house library, which contains lipids produced by various organisms, and retention time and carbon chain length and degree of unsaturation were correlated.

LC–MS polar metabolite analysis

Metabolic profiling of the polar phase was done as described by Zheng et al. (2015)⁵⁷ with minor modifications described below. Briefly, analysis was performed using Acquity I class UPLC System combined with a mass spectrometer Q Exactive Plus Orbitrap™ (Thermo Fisher Scientific), which was operated in a negative ionization mode. The LC separation was done using the SeQuant Zic-pHilic (150 mm × 2.1 mm) with the SeQuant guard column (20 mm × 2.1 mm) (Merck). The mobile phase B: acetonitrile and mobile phase A: 20 mM ammonium carbonate with 0.1% ammonia hydroxide in DDW: acetonitrile (80:20, v/v). The flow rate was kept at 200 µL/min, and the gradient was as follows: 0–2 min 75% of B, 14 min 25% of B, 18 min 25% of B, 19 min 75% of B, for 4 min, 23 min 75% of B.

Data processing was done using TraceFinder (Thermo Fisher Scientific; version 5.1) software⁶¹ when detected compounds were identified by retention time, and fragments were verified using an in-house-generated mass spectra library. Relative levels of lipids and polar metabolites were normalized to the internal standards and volume of the samples.

LC–MS steroid and thyroid hormone analysis

Hormones were measured by UPLC-ESI-MS/MS equipped with an Acquity UPLC I class system (Waters Corp., MA, USA). The MS detector (Waters Xevo TQ-XS) was equipped with an ESI source. The measurement was performed in positive and negative ionization modes using MRM. The capillary voltage was 2.5 kV. Chromatographic separation was conducted based on Waters Application Note on an ACQUITY UPLC HSS T3 column (1.0 × 100 mm, inner diameter, 1.8 μm) (Waters), with 0.3 mM ammonium fluoride in water as mobile phase A and methanol as mobile phase B. Free hormones were measured from the lipid phase. The sum of free and conjugated hormones was measured in the hydrolyzed fraction. Data were analyzed using Targetlynx software (Waters; version 4.2)⁶². Hormone data were normalized to internal standards (Supplemental Table S1).

Chemical analyses

Plasma β-hydroxybutyrate (BHB) concentration was determined with a RANBUT D-3-Hydroxybutyrate Kit (Randox, CruMLin, UK), in which a reaction of 3-hydroxybutyrate and dehydrogenase generates a UV emission correlated with the sample BHB concentration. The samples were examined at 340 nm with an optical density reader (Spectro V-11D, MRC, Israel); the results were calibrated against a known BHB concentration that was supplied by the manufacturer. The intra- and interassay coefficients of variation for the BHB assay were 1.3 and 1.8%, respectively.

Plasma nonesterified fatty acid concentration was determined with a NEFA C Test Kit (Wako Chemicals GmbH, Germany). The intra- and interassay coefficients of variation for the NEFA assay were 5.7 and 6.4%, respectively.

The blood glucose, lactate, and triglyceride (TG) were determined using the Cobas c111 autoanalyzer (Roche Holding GmbH, Germany). The calibration for all metabolites was performed using Roche Calibrator for Automated Systems (Calibrator f.a.s., Roche Diagnostics), which contains traceable, defined specific metabolite concentration following the manufacturer’s calibration protocol. Briefly, calibration was performed by the autoanalyzer by automatically sampling the manufacturer-provided calibrator solution before sample measurement, upon installation of each new reagent set.

Plasma glucose concentration was determined using a Cobas c111 autoanalyzer (Roche Holding GmbH, Germany) with the reagent set GLUC2 (REF: 04657527190), which uses two enzymatic reactions with hexokinase and glucose-6-phosphate dehydrogenase to generate a UV emission that is correlated with the glucose concentration in the sample; the results were calibrated against known glucose concentrations.

Plasma lactate concentration was examined in the Cobas c111 autoanalyzer with the reagent set LACT2 (Roche Holding GmbH; REF: 05401666190), which uses two enzymatic reactions with lactate oxidase and peroxidase to generate a UV emission that is correlated with the sample L-lactate concentration; the results were calibrated against known L-lactate concentrations.

Plasma TG concentration was determined in the Cobas c111 autoanalyzer with the reagent set TRIGL (Roche Holding GmbH; REF: 04657594190), which uses four enzymatic reactions with lipase, glycerol kinase, glycerol phosphate oxidase, and peroxidase to generate a UV emission that is correlated with the sample TG concentration; the results were calibrated against known TG concentrations.

Plasma insulin was determined by the Insulin IRMA Test Kit (Beckman Coulter, CA, USA; REF: IM3210). The intra- and interassay coefficients of variation for the insulin assay were 7.7 and 6.3%, respectively.

According to the manufacturer’s recommendations, cortisol and testosterone were quantified using ELISA kits (DRG Diagnostics, Germany, item no. EIA-1887 for cortisol and item no. EIA-1559 for testosterone). The sensitivity of the cortisol assay was 1.3 ng/mL, and the intra- and interassay coefficients of variation for the cortisol assay were 6.7 and 8.1%, respectively. The sensitivity of the testosterone assay was 0.083 ng/mL, and the intra- and interassay coefficients of variation for the testosterone assay were 9.9 and 4.2%, respectively.

RNA extraction and gene-expression analysis

We determined gene expression in a total of 14 placenta samples. RNA extraction of placental tissue was performed by Maxwell^® RSC simplyRNA Kit (#AS1390, Promega, WI, USA) in a Maxwell^® RSC Instrument (#AS4500, Promega). A total of 200 µL of each sample was transferred into a 1.5-mL microfuge tube along with 200 µL of 1-thioglycerol/homogenization solution. Next, the homogenate was vortexed and added to prefilled cartridges with 10 µL DNase I. RNA was eluted by the addition of 50 µL nuclease-free water. RNA purity and concentration were determined using a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific), and all samples had a 260/280 absorbance ratio above 2. First-strand cDNA was obtained using the qScript cDNA SuperMix (#95048-025, Quantabio, MA, USA) according to the manufacturer’s protocol, and the resulting cDNA was diluted to a concentration of 5 ng/µL in RT-PCR-grade water. Quantitative detection of specific mRNA transcripts was assayed in duplicate and carried out by real-time PCR with LightCycler 480 SYBR Green Master Mix (#04707516001, Roche Pharmaceuticals, Israel).

In placental tissues, we examined the transcription levels of several genes related to fetal growth and general metabolism: (1) Placental aromatase (CYP19) - fetal survival depends on placental capacity and endocrine support, and estrogens are critical regulators of uteroplacental blood flow, trophoblast differentiation, and placental function. The CYP19 may reflect impaired steroidogenic support and placental insufficiency⁶³. (2) Solute carrier family 2 member 1 (SLC2A1) - encodes GLUT1, the primary placental glucose transporter, and was included as a marker of placental nutrient transfer capacity. Reduced GLUT1 expression or activity has been associated with placental insufficiency, impaired fetal growth, and increased fetal loss⁶⁴. (3) Nuclear receptor subfamily 3 group C member 1 (NR3C1) - encodes the glucocorticoid receptor. Glucocorticoids regulate placental nutrient transport, lipid mobilization, and fetal growth by modulating placental endocrine and metabolic pathways, highly relevant to mechanisms underlying partial litter loss^22,65. (4) Insulin-like growth factor 2 (IGF2) - a key placental growth factor regulating trophoblast proliferation, placental size, and nutrient transfer capacity. Its expression is tightly linked to maternal nutrient availability and energy balance, integrating glucose and lipid-derived energy supply with fetal demand²⁴. (5) Corticotropin-releasing hormone receptor 1 (CRHR1) - CRH signaling is implicated in trophoblast invasion and placental function, and altered CRHR1 expression is reported in placental dysfunction phenotypes associated with poor fetal growth/survival⁶⁶. (6) Androgen receptor (AR) - mediates androgen signaling in the placenta, which influences placental development, vascularization, sex-specific fetal growth, modulates trophoblast function, and nutrient allocation⁶⁷. The primers are listed in Supplemental Table S2. The reference genes were: Ovis aries actin beta (ACTB) and Ovis aries tubulin alpha 1a (TUBA1A).

Statistical analysis

Lambing data were analyzed with the PROC GLM procedure, version 9.2⁶⁸. Means were estimated using the LSMEANS statement of SAS and tested for significance by the Tukey–Kramer procedure.

Plasma metabolites and hormone concentrations were analyzed as repeated measurements using the PROC MIXED procedure, version 9.2⁶⁸, with the model:

Y_ijkl = µ + T_i + L_j + C_k + H_l + T_iH_l + E(T*C)_ijklm +DIP_jklmn + E_ijklmno,

where µ = overall mean; T_i = fixed effect of group, i = VL or PLL; L_j = fixed effect of dams’ BCS; H_k = fixed effect of dams’ BW; E(T)_im = random effect of ewe_m nested in group; E_ijklmn = random residual.

For all metabolites and hormones, the values at 80 DIP were used as a covariate in the model. Autoregressive order 1 was used as a covariance structure in the model because it resulted in the lowest Bayesian information criterion for most of the tested variables.

Genes’ relative expression was analyzed with the PROC GLM procedure, version 9.2⁶⁸. The litter size (live and dead) was included in the model.

For the lipidomics analysis, data of the stillborn (SB) and liveborn (LB) fetuses were analyzed using linear mixed-effects models on log2-transformed values. This satisfied the assumptions of normality and homogeneity of variance within the models.

The SB and LB data were tested as fixed factors along with study number and ewe ID as random factors (plus lamb ID if there were lambs with multiple samples). All analyses were done using R, (version 4.5.2), and the packages `lme4` (version 1.1–38) and `lmerTest (version 3.2-0.2)⁶⁹.

Significance was set at P ≤ 0.05, and tendencies were reported at 0.05 < P ≤ 0.10 unless otherwise stated.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

**TA: ** Methodology, Conceptualization, Project administration, Ultrasound monitoring, Data curation, Visualization, Laboratory analysis, Formal analysis, Validation, and Writing. **MR** : Allantocentesis. **JS** : Data curation, Visualization **. PDSS** : Laboratory analysis, Formal analysis. **AR** : Data curation, Animal care, Formal analysis, Methodology. **LL** : Data curation, Laboratory analysis, Methodology, Validation. **GK** : Laboratory analysis, Formal analysis. **MG** : Genomic data processing. **SM** : Formal analysis, Metabolomics, Lipidomics. **MI** : Formal analysis, Metabolomics, Lipidomics **. RR** : Data processing, Statistical analysis. **UM: ** Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Visualization, Writing – review & editing. All authors have read and agreed to the published version of the manuscript.

Funding

This research was financially supported by the Chief Scientist of the Ministry of Agriculture, Rishon LeZion – grant number 362081023.

Data availability

All data are presented in the paper.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The Volcani Center Animal Care Committee approved all of the procedures performed on the animals in the present study (approval no. 981/23 IL). Consent to participate is not relevant.