Abstract Protein imbalance during pregnancy affects women in underdeveloped and developing countries and is associated with compromised offspring growth and an increased risk of metabolic diseases in later life. We studied in a porcine model the glucose and urea metabolism, and circulatory hormone and metabolite profile of offspring exposed during gestation, to maternal isoenergetic low–high (LP-HC), high–low (HP-LC) or adequate (AP) protein–carbohydrate ratio diets. At birth, LP-HC were lighter and the plasma acetylcarnitine to free carnitine ratios at 1 day of life was lower compared to AP offspring. Plasma urea concentrations were lower in 1 day old LP-HC offspring than HP-LC. In the juvenile period, increased insulin concentrations were observed in LP-HC and HP-LC offspring compared to AP, as was body weight from HP-LC compared to LP-HC. Plasma triglyceride concentrations were lower in 80 than 1 day old HP-LC offspring, and glucagon concentrations lower in 80 than 1 day old AP and HP-LC offspring. Plasma urea and the ratio of glucagon to insulin were lower in all 80 than 1 day old offspring. Aminoacyl-tRNA, arginine and phenylalanine, tyrosine and tryptophan metabolism, histidine and beta-alanine metabolism differed between 1 and 80 day old AP and HP-LC offspring. Maternal protein imbalance throughout pregnancy did not result in significant consequences in offspring metabolism compared to AP, indicating enormous plasticity by the placenta and developing offspring. Keywords: maternal protein restriction, offspring, body weight, carbohydrate and urea metabolism, metabolomics, porcine model 1. Background The Developmental Origins of Health and Disease hypothesis proposes that exposure of the developing fetus to a hostile uterine environment during critical phases of growth and development can significantly impact the offspring’s short- and long-term health [[38]1,[39]2]. Inadequate nutrition during pregnancy, both in composition and amount, is a significant stressor on the developing embryo/fetus as the mother is the sole source of nutrients. In underdeveloped and developing countries, access to sufficient amounts/quality protein sources is difficult and has become a major public health problem [[40]3,[41]4]. Studies have shown that exposure to maternal protein undernutrition during fetal development can affect the structural and functional features of offspring and lead to an increased propensity for chronic disease [[42]2]. In addition, human and animal studies have shown that it is not just maternal protein undernutrition that negatively impacts fetal development, but protein excess too [[43]5,[44]6,[45]7,[46]8,[47]9]. The majority of animal studies examining the effect maternal protein imbalance on offspring development, have been performed in rodents, which, at birth, are developmentally less mature than human infants [[48]10]. In many of these rodent studies the effect of maternal diet composition on offspring development and health, semi-synthetic diets were used [[49]11,[50]12,[51]13]. However, it remains controversial whether these models adequately reflect the transgenerational effects of diet. Additionally, exposure has been conducted during both gestation and lactation [[52]13,[53]14,[54]15,[55]16], complicating analysis of how in utero exposure is linked to the onset of metabolic diseases later in life. The pig is a well-established animal model for studying the effect of maternal nutrition on offspring metabolism in humans [[56]17,[57]18,[58]19], whose post-natal development is closely related to that of humans [[59]20,[60]21]. Thus, we developed a model of intrauterine growth restriction using a modest dietary protein imbalance, in first-pregnancy sows, by feeding them isoenergetic [[61]22] low protein to high carbohydrate (LP-HC; 1:10.4), high protein to low carbohydrate (HP-LC; 1:1.3) or adequate (AP; 1:5) diets, throughout pregnancy. Offspring from all three diet groups were then cross-fostered to sows fed an AP diet throughout, to remove any influence the pregnancy diet may have on post-natal development. Using this model, we have reported that the modulated protein to carbohydrate ratios result in altered fetal cortisol regulation (LP-HC) [[62]6], birth weight, body composition, mitochondrial biogenesis and gene methylation (LP-HC), adipocyte energy processes and lipid transport, hepatic cell cycle and proliferation (LP-HC, HP-LC), and muscle filament and cell cycle progression in the offspring (HP-LC) [[63]22,[64]23,[65]24,[66]25,[67]26,[68]27,[69]28,[70]29]. Several of these companion studies assessed the impact of offspring age on the hepatic and muscle transcriptome and found that age had a stronger effect than maternal diet, implying considerable offspring resilience towards exposure to a mildly imbalanced maternal protein diet in utero [[71]26,[72]27,[73]28,[74]29]. Metabolomics is a tool that can provide a snapshot in time of biological processes underpinning experimentally observed phenotypic, proteomic and transcriptomic changes [[75]30]. The circulatory metabolite and hormone profiles can be most informative as they reflect systemic changes (interactions from all the organs affected) by, for example, the nutritional insult used in a dietary model of protein imbalance. Metabolomics has been applied to a variety of studies in pigs [[76]31,[77]32,[78]33], but only few exploring the plasma metabolome of neonatal piglets [[79]34,[80]35]. Therefore, the objectives of this study were twofold: (1) First, we were interested to know whether exposure to a moderate dietary protein imbalance in utero, known to moderately reduce neonatal birth weight, would also affect the metabolism of offspring during the juvenile stage. (2) Second, whether the investigation of plasma metabolite profiles and metabolic function in porcine offspring exposed to maternal protein imbalance during pregnancy could shed light on the previously reported proteomic, transcriptomic and phenotypic changes in this model [[81]22,[82]24,[83]25,[84]26,[85]27,[86]34,[87]35]. Therefore, we characterized the glucose and urea metabolism, and circulatory hormone and metabolite profile of porcine offspring exposed to a maternal LP-HC and HP-LC diets during gestation, and explored the associations with the different phenotypic, proteomic and transcriptomic data previously reported with this porcine model of mild dietary protein imbalance. 2. Methods 2.1. Experimental Pregnancy Diets Experimental pregnancy diets were formulated to have an adequate (AP; 12.1% crude protein, 3.7% crude fat, 7.4% crude fiber), low (LP-HC; 6.5% crude protein, 2.9% crude fat, 8.9% crude fiber) or high (HP-LC; 30% crude protein, 3.7% crude fat, 10.3% crude fiber) protein concentration [[88]22]. Diets were isoenergetic (13.7 MJ of metabolizable energy/kg) with a protein to carbohydrate ratio of 1:5, 1:10.4 and 1:1.3, respectively. This required the addition of crystalline L-amino acids (AA) to the LP-HC and HP-LC diets to achieve AA proportions similar to the AP diet. In order to test and compare practical diets under real-world conditions we refrained from semi-synthetic diets and extremely low or high protein levels as previously reported [[89]36,[90]37,[91]38]. However, it should be noted that the low and high protein levels used in the present study are clearly below and in excess of the required dietary protein level, respectively, as recommended for primiparous sows [[92]39]. In the LP-HC group, the intake of indispensable AA was 50% less than recommendations, whereas in the HP-LC group, gilts consumed ~250% of indispensable AA, as compared with the AP group [[93]22]. As the experimental diets were isoenergetic, the different dietary protein levels were balanced by different levels of carbohydrates and fat. Thus, the content of digestible and non-digestible carbohydrates and fat in LP-HC/HP-LC differed from the (AP) control diet. Gilts were fed the experimental diets twice daily (07:00 and 15:00 h) throughout pregnancy ([94]Figure 1) and had free access to water. Figure 1. [95]Figure 1 [96]Open in a new tab Experimental design. Primiparous sows were fed low (LP-HC; 1:10.4, n = 18), high (HP-LC; 1:1.3, n = 16) or adequate (AP; 1:5, n = 17) protein to carbohydrate ratio diets from 1 day prior to insemination until birth. 2.2. Animals and Postnatal Offspring Diets The animal study was conducted from 2005 to 2008, with all animal care and experimental procedures conducted in accordance with the German Animal Welfare Act approved by the State Office for Agriculture, Food Safety and Fishing Mecklenburg-Western Pomerania, Germany (7221.3–1.1–006/04; 7221.3–1.2–05/06; 7221.3–1.2–013/06). German Landrace sows were sourced from the experimental pig facility of the Research Institute for Farm Animal Biology. One day prior to insemination, gilts (n = 93) were randomly assigned to one of three pregnancy diets: AP, LP-HC or HP-LC. At 115 days post insemination labor was induced, as previously described [[97]22]. Litter characteristics were recorded immediately after birth and detailed results reporting gilt development, feed intake and colostrum composition of the entire cohort have been previously published [[98]22]. At farrowing, experimental male and female (AP = 80, LP-HC = 74, HP-LC = 85) offspring from 51 litters (AP = 17, LP-HC = 18, HP-LC = 16) spread across 8 experimental blocks were selected for the current study ([99]Figure 1). For the first 24–36 h of life, all experimental piglets remained with their mother to ensure sufficient uptake of colostrum. Within 36 h after birth, age class 1 day piglets (AP, n = 53; LP-HC, n = 49; HP-LC, n = 58) were euthanized with T61 (Tetracaine hydrochloride 5 mg/mL, mebezonium iodide 50 mg/mL, Embutramid 200 mg/mL, Intervet Deutschland GmbH, Unterschleißheim, Germany). While age class 80 day piglets (AP = 27, LP-HC = 25, HP-LC = 27) were cross-fostered to standardized litters (11 piglets/sow) from sows (parity 2–4) fed AP pregnancy and standard lactation diets ([100]Figure 1). This was done to remove potential postnatal effects of gilt experimental diet and litter size on measurements taken after the colostrum intake period (24–36 h post birth). At day 6 of life, all experimental male piglets were castrated. Post-weaning (day 28 of life) piglets had ad libitum access to the same standard commercial diets [[101]40]. From day 28 of life, piglets were provided with a starter diet (Turbostart, 15 MJ ME/kg; Trede & von Pein, Dammfleth, Germany) for 4 days, before being introduced to a conventional pig starter diet (Porcistart 14 MJ ME/kg; Trede & von Pein, Dammfleth, Germany). At day 48, piglets were switched to a grower diet (Porcibig; 13.8 MJ ME/kg; Trede & von Pein, Dammfleth, Germany) until day 76 when they were provided with Vormast Cafo Top (13.6 MJ ME/kg; Trede & von Pein, Dammfleth, Germany). At day 79, offspring were fasted overnight for 12 h and then euthanized via exsanguination after electro-stunning at day 80 ([102]Figure 1). Offspring body weight (BW) for both age classes was determined at birth and at euthanasia. For age class 80 day, body weight was also recorded at days 28, 35, 42, 49 and 56, and on day 68 (intravenous glucose tolerance test (IVGTT) and glucose turnover test), day 71 (intravenous insulin challenge (IIC) and day 76 (urea turnover). 2.3. Blood Sampling At euthanasia, blood samples were collected into Li–Heparin tubes (Sarstedt, Germany) for both age classes and centrifuged for 20 min at 1576× g and 4 °C to obtain plasma. A subset (age class 1 day: AP, n = 23; LP-HC, n = 28; HP-LC, n = 18; age class 80 day: AP, n = 13; LP-HC, n = 14; HP-LC, n = 7) of blood samples from blocks 1–4 was collected into Serum Z tubes (Sarstedt, Germany), and left at room temperature for 2 h to coagulate and obtain serum. Plasma and serum samples were stored at −80 °C for subsequent clinical chemistry and targeted metabolite analysis. Clinical chemistry samples were analyzed within three months after collection, while targeted metabolite samples were analyzed eight years after collection. We have previously shown that the overall stability of the metabolites stored at −80 °C for five years is very good [[103]41], and our own data (unpublished) shows that 7 years has the same low or non-impact. At day 60, the age class 80 d offspring were surgically fitted with an indwelling jugular catheter, to allow frequent blood sampling as previously described [[104]42]. All offspring were fasted for 16 h overnight and a basal blood sample was collected 15 and 5 min prior to the start of each experimental protocol. At day 68, an IVGTT and glucose turnover measurements were conducted. Blood samples were collected 2, 3, 5, 8, 10, 15, 20, 30, 40, 60 and 240 min post injection and treated according to the work in [[105]42]. The resulting plasma samples were stored at −20 °C, and subsequent analyses performed within 1 year of collection. At day 71, an IIC was performed, and blood samples were taken 5, 10, 15, 30, 45, 60, 75, 90, 120, 150, and 180 min relative to dosing. At day 76, the same offspring were utilized for urea turnover measurements and blood samples were collected 5, 10, 15, 20, 30, 60, 90, 120, 150, 240 and 360 min post injection. Blood samples were treated according to [[106]42] and the resulting plasma samples were stored at −80 °C for subsequent analyses. Intravenous insulin challenge samples were analyzed within 3 months of collection, while urea turnover samples were analyzed within one year of sample collection. 2.4. Plasma and Serum Biochemical Parameters Plasma glucose, non-esterified fatty acids (NEFA), triglycerides, total cholesterol and urea concentrations were measured at the University of Veterinary Medicine Hannover, Germany [[107]42]. Serum high-density lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterol concentrations were measured using HDL (650207) and LDL (969706) cholesterol kits from Beckman Coulter GmbH (Krefeld), at the Institute for Clinical Chemistry and Laboratory Medicine, University of Rostock. Plasma insulin (PI–12K) and glucagon (GL–32K) were measured by radioimmunoassay using commercially available porcine kits (Biotrend Chemikalien GmbH, Köln, Germany). The minimum detectable concentration as reported by the kit manufacturer was 1.611 μU/mL and 18.453 pg/mL (±2 SD), respectively, while the intra- and inter-assay coefficients of variance were for insulin (8.4% and 9%) and glucagon (4% and 8.1%). The glucagon to insulin ratio was calculated as the glucagon concentration in nmol/L divided by the insulin concentration in nmol/L. The glucose to insulin ratio was calculated as the glucose concentration in mmol/L divided by the insulin concentration in pmol/L. 2.5. Intravenous Glucose Tolerance Test/Insulin Challenge and Glucose Turnover Measurements At day 68, offspring were given a mixed bolus of 0.2 g glucose (Glucose 50, Braun, Melsungen AG, Melsungen, Germany) and 1.5 mg of ^13C[6]-glucose (99 atom% ^13C; Berlin–Chemie AG, Berlin, Germany) per kg BW through the jugular catheter, for IVGTT and glucose turnover measurements. Plasma samples were used to determine glucagon, glucose and insulin concentrations for the IVGTT and plasma glucose turnover was calculated according to [[108]43]. Glucose concentrations were measured using a microplate absorbance reader (Sunrise^TM, Tecan Austria GmbH, Grödig/Salzburg, Austria), while insulin and glucagon concentrations were measured as described in the previous section. The enrichment of ^13C[6]-glucose was measured by GCMS in positive chemical ionization mode after conversion to aldonitrile pentaacetate derivatives using m/z 328 (m + 0) and m/z 334 (m + 6) as diagnostic ions, and the glucose turnover was calculated according to [[109]44]. At day 71, an intravenous insulin challenge was conducted by administering a 0.5 I.U./kg BW bolus of porcine insulin (27 I.U./mg; Sigma-Aldrich, Taufkirchen, Germany), as reported in a companion study [[110]45]. Glucose was determined as described above. 2.6. Urea Turnover Measurements At day 76, pigs were given a bolus injection of 3.5 mg ^15N[2]–urea (99 At.-% ^15N, Chemotrade, Leipzig) per kg BW through the jugular catheter. Plasma ^15N urea enrichment was measured as tert–butyldimethylsilyl derivatives by GCMS with positive chemical ionization in selected ion monitoring mode for m/z 231 to 233 quantification and urea turnover rate was determined [[111]46]. 2.7. Targeted Metabolomics Targeted metabolite analysis was conducted using a subset of plasma samples (n = 8/diet/age class) which were derived from 36 different litters and balanced for offspring sex. Analysis was based on LC-ESI-MS/MS and FIA-ESI-MS/MS measurements using the AbsoluteIDQ^TM p180 Assay Kit (Biocrates Life Sciences AG, Innsbruck, Austria). The assay utilized 10 µL plasma and enabled 188 metabolites to be quantified; 39 acylcarnitines + free carnitine, 21 amino acids (AA; 19 proteinogenic + citrulline + ornithine), 21 biogenic amines, 90 glycerophospholipids (14 lysophosphatidylcholines (lysoPC) and 76 phosphatidylcholines (PC)), 15 sphingolipids and hexoses (sum of hexoses—about 90–95% glucose) simultaneously. The abbreviation Cx:y is used to describe the total number of carbons (x) and double bonds (y) of all chains, respectively [[112]47]. For LC-ESI-MS/MS and FIA-ESI-MS/MS, compound identification and quantification were based on scheduled multiple reaction monitoring measurements. The method of AbsoluteIDQ^TM p180 Kit conforms with the EMEA-Guidelines “Guideline on bioanalytical method validation” (21 July 2011) [[113]48], which implies proof of reproducibility within a given error range. Measurements were performed as described by the manufacturer (manual UM-P180). The limit of detection (LOD) was set to three times the value of the zero samples (PBS). The AbsoluteIDQ^TM p180 Kit assay procedures, metabolite nomenclature, sample handling and mass spectrometric analyses have been all previously described in detail [[114]47]. 2.8. Data Evaluation and Statistical Analysis Offspring body weight data were analyzed using a mixed model that contained the fixed effects of maternal diet, experimental block, offspring sex, litter size class (litters <13 and ≥13 piglets according to 13 as the median of litter size), the random factor gilt nested in maternal diet × experimental block (1–8) × litter size class and the interactions of maternal diet × offspring sex and maternal diet × litter size class. Clinical blood parameters, glucose and urea turnover data and fasted analyte values prior to IVGTT were analyzed with a mixed model using the fixed effects of maternal diet, offspring age class and sex, with the additional fixed effects of experimental block and litter size class, and the random factor gilt nested in maternal diet × experimental block × litter size class × offspring age class, and the interactions of maternal diet × offspring sex × offspring age class, maternal diet × offspring age class and offspring sex × offspring age class. The model for IVGTT and IIC were analyzed with a mixed model containing the fixed effects of maternal diet, offspring sex, litter size class, the random factor gilt nested in maternal diet × experimental block × litter size class × time, the interactions of maternal diet × offspring sex, maternal diet × litter size class, maternal diet × time, maternal diet × offspring sex × time and maternal diet × litter size class × time and an exponential spatial model (SP(EXP)) to account for unequally spaced repeated measurements. Metabolite ratio and pool data were analyzed using a model that contained the fixed effects of maternal diet, offspring age class, sex and the interactions of maternal diet × offspring sex × offspring age class, maternal diet × offspring age class and offspring sex × offspring age class. All statistical analyses were conducted in SAS 9.4 (Copyright, SAS Institute Inc., Cary, NC, USA) and for all models, least squares means (LSmeans) and their standard error (SE) were computed for each fixed effect and analyzed with Tukey’s multiple comparison post hoc tests. Concentrations are reported as LSmean values ± SE. Significance is reported as p ≤ 0.05. Targeted metabolite data evaluation for quantification and quality assessment was performed with the software packages MultiQuant 3.0.1 (Sciex Deutschland GmbH, Darmstadt, Germany) and MetIDQ™ (Biocrates Life Sciences AG). Internal standards served as references for the