The effect of pregnancy on nitrogen retention, maternal insulin sensitivity, and mRNA abundance of genes involved in energy and amino acid metabolism in gilts

Abstract

Twenty-one of each pregnant (P) and nonserviced, nonpregnant (NP) sister-pairs of gilts were selected to investigate the effect of pregnancy on protein deposition (Pd; whole body and maternal), insulin sensitivity, and mRNA abundance of genes involved in energy and AA metabolism. Between breeding (study day 0) and day 111, P and NP gilts received 2.16 kg of the experimental diet (3.34 Mcal ME/kg, 17.6% crude protein, 0.78% standardized ileal digestible lysine) that was formulated to meet the estimated ME requirements of pregnant gilts (and meet or exceed AA requirements). Nitrogen balances were conducted on day 63 and 102 ± 0.2 of the study during 4-d periods. Blood samples were collected on day 43, 56, 71, 85, 98, and 108 ± 0.3 of the study to determine plasma concentrations of fasted IGF-1, estradiol (E2), and estrone sulfate (E1S). Frequently sampled intravenous glucose tolerance tests (FSIGTT) were conducted on day 75 ± 0.7 in 6 P and 5 NP gilts and on day 107 ± 0.4 in 17 P and 17 NP gilts and the MINMOD approach was applied to evaluate whole body insulin sensitivity and pancreatic responsiveness. Longissimus muscle (LM) and s.c. adipose tissue (AD) samples were excised from 12 P and 12 NP gilts at day 111 ± 0.4 of the study after euthanasia to determine mRNA abundance of key genes. Whole body Pd was greater (P < 0.001) at day 102 and maternal Pd was lower (P < 0.002) at day 63 and 102 for P compared to NP gilts. Plasma concentrations of E1S and E2 increased (P < 0.05) with study day for P gilts and remained constant for NP gilts, which coincided with reduced plasma concentrations of IGF-1 and increased estrogen receptor alpha (ESR1) mRNA abundance in LM of P gilts. Glucose effectiveness was not different between P and NP gilts, but whole body insulin sensitivity was lower (P = 0.004) in P compared to NP gilts on day 75 and 107, which corresponded with reduced mRNA abundances of SLC2A4, HK2, SREBF1, and FASN, and increased abundances of PDK4 and PPARGC1A in LM and AD. When fed identically, P gilts had greater whole body Pd at day 102, which reflects Pd in the pregnancy-associated tissues (at the expense of maternal Pd), likely driven by estrogen-stimulated insulin resistance in peripheral tissue and subsequent modulation of gene expression relating to glucose metabolism.

Introduction

During gestation in the sow, the nutrient requirements of the developing conceptus are relatively low for the first 2 trimesters of pregnancy, but increase exponentially during the final trimester (NRC, 2012). In order to meet these increased nutrient demands in late gestation, maternal metabolism is modified to favor nutrient partitioning towards the fetus. For example, a decrease in N retention in maternal tissues occurs toward the end of gestation in gilts (Close et al., 1985; Miller et al., 2016).

One of the major modifications of maternal metabolism during gestation is a progressive reduction in insulin sensitivity (Barbour et al., 2007). In both maternal skeletal muscle and adipose tissue, insulin sensitivity is reduced during late pregnancy in rats (Leturque et al., 1984), rabbits (Hauguel et al., 1987), and humans (Barbour et al., 2007), and is characterized by reduced glucose uptake into these tissues (Barbour et al., 2007). Based on a delayed peak in plasma insulin concentration and prolonged elevation of both plasma insulin and glucose concentrations during glucose tolerance tests, it was proposed that sows in late gestation experience impaired insulin sensitivity, regardless of feeding level (George et al., 1978; Schaefer et al., 1991; Père et al., 2000). Reduced insulin sensitivity during (late) gestation is a mechanism by which the partitioning of glucose (and AA) to the fetus is increased (Père et al., 2000), though studies combining whole-body and tissue-level measurements of insulin sensitivity for pregnant sows are not available. Therefore, it was hypothesized that from mid to late gestation in gilts, energy and AA metabolism in maternal tissues shift to favor nutrient partitioning toward the developing piglets. The objectives of this study were to determine 1) the effect of pregnancy on whole body and maternal Pd, 2) minimal model parameters of glucose homeostasis from mid to late gestation, and 3) mRNA abundance of genes involved in maternal energy and AA metabolism in late gestation.

MATERIALS AND METHODS

The experimental protocol was approved by the University of Guelph Animal Care Committee and followed Canadian Council of Animal Care guidelines (CCAC, 2009).

Animals and Feeding

The study was conducted at the Arkell Swine Research Station (Ontario Ministry of Agriculture, Food and Rural Affairs, Guelph, ON; University of Guelph, Guelph, Ontario, Canada). Sixteen Yorkshire and 26 Yorkshire × Landrace gilts were sourced in 3 blocks of 14 gilts. To investigate the effect of pregnancy, littermate pairs of gilts (from 17 different litters; 8 gilts were unpaired) with similar BW and ultrasound back fat thickness (BF; 6.5 cm from the midline over the last rib) were divided between pregnant (P) and nonserviced, nonpregnant (NP) treatment groups. Sample size was based on Miller et al. (2016). Gilts were housed in groups of 10 to 12 until P gilts were bred, after which both P and NP gilts were housed individually throughout the study in conventional gestation stalls (0.64 × 2.13 m).

Between study day 0 (breeding) and 111, gilts were fed a corn and soybean meal-based diet, formulated to meet the estimated ME and AA requirements of gestating gilts at day 90 of gestation (NRC, 2012; Table 1). In each block, feed allowance was based on mean BW at breeding (136.0, 124.5, and 129.5 kg for blocks 1, 2, and 3, respectively), parity (1), gestation length (114 d), anticipated total BW gain (65 kg), anticipated litter size (12.5 piglets), and anticipated mean piglet birth weight (1.40 kg) according to the NRC (2012). Feed allowances were 2.20, 2.12, and 2.16 kg/d as-fed for blocks 1, 2, and 3, respectively. Rations were given in one meal at 0800 h between day 1 and 35 of the study, and divided over 2 meals per day thereafter at 0800 and 1500 h to achieve an overnight fast before blood sampling; no feed refusals were observed. From 7 d prior to and during each N balance period, 0.2% titanium dioxide was added to the feed, at the expense of corn, as an indigestible marker; weekly feed samples were collected from each batch of feed for nutrient analyses. Water was provided ad libitum. Gilt BW was measured approximately every 2 wk during the study before the morning meal and BF was measured on day −6, 59, and 101 ± 0.6.

Table 1.

Ingredient composition and nutrient content of experimental diets

Item	Experimental1	Titanium dioxide experimental2
Ingredient composition, % (as-fed)
Corn, 8.3% CP	69.50	69.30
Soybean meal, 47.5% CP	24.85	24.85
Animal and vegetable fat blend	2.00	2.00
Calcium carbonate	1.45	1.45
Monocalcium phosphate	1.30	1.30
Salt	0.40	0.40
Vitamin and mineral mix3	0.50	0.50
Titanium dioxide	―	0.20
Calculated nutrient content
ME, Mcal/kg	3.34	3.34
CP, %	17.60	17.57
Total lysine, %	0.91	0.91
Standardized ileal digestible lysine, %	0.78	0.78
Calcium, %	0.84	0.84
Phosphorus, %	0.64	0.64
Analyzed nutrient content, %
CP	17.3	17.2
Total lysine	0.96	0.97
Calcium	0.83	0.89
Phosphorus	0.65	0.64

¹Values reflect the mean of 7 batches.

²Values reflect the mean of 3 batches.

³As supplied per kilogram of complete diet: vitamin A, 12,000 IU as retinyl acetate (3.0 mg) and retinyl palmitate (2.04 mg); vitamin D3, 1,200 IU as cholecalciferol; vitamin E, 67 IU as dl-α-tocopherol acetate (52.8 mg); vitamin K, 3.0 mg as menadione; choline, 600 mg; pantothenic acid, 18 mg; riboflavin, 6 mg; folic acid, 2.4 mg; niacin, 30 mg; thiamine, 1.8 mg; vitamin B6, 1.8 mg; biotin, 0.24 mg; vitamin B12, 0.03 mg; Se, 0.36 mg from Na₂SeO₃; Cu, 18 mg from CuSO₄^.5H₂O; Zn, 124.8 mg from ZnO; Fe, 120 mg from FeSO₄; Mn, 22.8 mg from MnO₂; and I, 0.36 mg from KI (DSM Nutritional Products Canada Inc., Ayr, ON, Canada).

Nitrogen Balance and Blood Sampling

Two, 4-d nitrogen balances were conducted starting on day 63 and 102 ± 0.2 of the study as described by Miller et al. (2016). Blood samples were collected on day 43, 56, 71, 85, 98, and 108 ± 0.3 of the study (Fig. 1) after a 17-h fast via orbital sinus puncture to determine fasted serum concentrations of insulin, glucose, estradiol (E2), and plasma concentrations of IGF-1, and estrone sulfate (E1S). Blood was collected into one 10-mL serum and two 6-mL EDTA plasma tubes (BD Vacutainers, Mississauga, ON, Canada). Serum and EDTA tubes were centrifuged at 3,400 × g and 1,500 × g, respectively, at 4 °C for 15 min. Serum and plasma were divided into 5 aliquots and stored at −20 °C until further analysis.

Figure 1.

Timeline of measurements throughout the study for P and NP gilts.

Frequently Sampled Intravenous Glucose Tolerance Tests (FSIGTT)

On day 74 ± 0.7 for 6 P and 5 NP gilts and on day 106 ± 0.4 for 17 P and 17 NP gilts, indwelling micro-renathane catheters (1.78 mm outer diameter, 1.02 mm inner diameter; TYGON, Saint-Gobain Performance Plastics Corp., Cleveland, OH) were inserted through an external ear vein to the jugular vein according to the methods of de Ridder et al. (2014). The catheter was then secured to the ear with a suture and tape and blocked with heparinized saline (100 IU/mL). The number of gilts selected was based on Huber et al. (2018).

On the following day, FSIGTT were performed after a 17-h fast. A 1.665 M glucose solution was infused (Watson Marlow SciQ 323S; 2.79 mm manifold tubing; 66 mL/min) into the catheter to deliver 0.5 g glucose/kg BW, after which 20 mL of saline solution were manually infused to clear the catheter. The infusion lasted for 4 min 52 s in P and NP gilts at day 75 and 5 min 54 s and 5 min 29 s for P and NP gilts at day 106, respectively. Blood was sampled via the catheter at 60 and 15 min before the infusion and at 0, 3, 6, 9, 12, 15, 18, 21, 25, 30, 35, 40, 45, 50, 60, 70, 80, and 90 min after the infusion stopped. For each blood sample, 6 mL of saline-diluted blood was removed before collecting 4 to 6 mL blood into a clean syringe. The saline-diluted blood was returned to the catheter, followed by 6 mL of heparinized saline (15 IU/mL). Blood samples were transferred into serum separator tubes with a clot activator (BD Vacutainers) and centrifuged for 10 min at 3,400 × g at room temperature within 20 min of collection. Serum was divided into 3 aliquots and stored at −20 °C until further analysis. Ear vein catheters were removed after the last blood sample was taken, with the exception of 12 P and 12 NP gilts, which were sacrificed for tissue sampling.

Tissue Sampling

After a 17-h fast, 12 P and 12 NP gilts were weighed and then euthanized at day 111 ± 0.4 using an intravenous infusion of sodium pentobarbital via the jugular catheter (50 mg/kg BW; Intervet Canada Corp. Kirkland, QC, Canada) and exsanguination via severing of the carotid artery. Samples of LM (between third and fourth rib) and s.c. adipose tissue (AD; between third and fourth rib) were immediately excised, rinsed with saline, and plunged into liquid N₂.

Nutrient Analysis

Weekly feed samples were pooled within each batch and fecal samples from N balances were pooled within each gilt and N balance period. Samples of diet and freeze-dried feces were sent to a commercial laboratory (SGS Agri-Food Laboratories, Guelph, ON, Canada) for analyses by AOAC (1997) methods for determination of DM (method 930.15), and Ca and P (method 985.01; feed only). Feed, freeze-dried feces, and 24-h subsamples of liquid urine were analyzed for N (FOSS Kjeltec 8200; FOSS Analytical, Hillerød Denmark) according to AOAC (1997; method 978.02). Titanium dioxide concentrations in feces and each batch of diet were quantified according to Myers et al. (2004), with minor adaptations (digestion for 24 h at 120 °C in 10 mL tubes and addition of H₂O₂ after precipitate settled in 100 mL volumetric flasks). Absorbance of standards and samples were measured by spectrophotometry (Beckman DU-7400; Beckman Instruments Inc., Fullerton, CA) at 408 nm.

Blood Analysis

Serum insulin concentrations were determined using a commercial ELISA kit according to manufacturer’s instructions (R&D Systems, Minnesota; DINS00; interassay CV of 6.5% and intra-assay CV of 5.8%). Serum glucose concentrations were determined using a glucose oxidase colorimetric assay (Sigma-Aldrich Canada Co., Oakville, ON, Canada; GAGO-20; interassay CV of 2.3% and intra-assay CV of 1.9%). Serum E2 concentrations were determined using a commercial ELISA kit (Arbor Assays, Michigan; KB30-H1; interassay CV of 4.6% and intra-assay CV of 6.9%). Plasma IGF-1 concentrations were determined with a commercial ELISA kit (R&D Systems; DG100; interassay CV of 3.3% and intra-assay CV of 1.3%) and plasma E1S concentrations were determined via radioimmunoassay (interassay CV of 7.0% and intra-assay CV of 5.7%), as described by Raeside and Rosskopf (1980).

RNA Isolation and Quantitative RT-PCR

Tissue samples were ground under liquid N₂ using a mortar and pestle prechilled to −80 °C. The RNA were isolated from between 50 and 100 mg ground tissue using TRIzol reagent (Invitrogen, Life Technologies, Burlington, ON, Canada) in singlicate for muscle and triplicate for adipose. The manufacturer’s protocol was modified such that adipose samples were centrifuged for 20 min at 13,000 rpm after TRIzol addition to separate a lipid layer. Liquid underlying the lipid layer was moved to a separate tube for chloroform addition. Additionally, RNA pellets from all tissue types underwent an additional wash in 75% ethanol, and the wash fluid was twice aspirated from the pellet after each wash. The RNA were resuspended in 50-µL for muscle and in 20-µL aliquots fro adipose, combining the triplicate adipose sample in DEPC-treated water (Fisher Scientific, Hampton, NH), and stored at −80 °C. The RNA concentration was determined using a NanoDrop 8000 spectrophotometer (Thermo Scientific, Waltham, MA) and RNA quality was determined using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA), where a RNA integrity number of greater than 5 was considered acceptable. Two micrograms of isolated RNA were treated with DNase I (Fisher Scientific, Hampton, NH) following the manufacturer’s protocol. The cDNA was synthesized from 500 ng of extracted total RNA with random hexamers using High Capacity cDNA Reverse Transcription Kit (Fisher Scientific, Cat. Number 4368814), following the manufacturer’s instructions and was stored at −20 °C until RT-PCR was performed using PerfeCTa SYBR Green FastMix (Quanta BioSciences, Gaithersburg, MD) with Applied Biosystems StepOnePlus Real Time PCR instrument. Primers (Integrated DNA Technologies, Coralville, IA) were designed to yield PCR amplification of 100 to 300 bp with efficiency of 90% or greater (Table 2).

Table 2.

Primer sequences for RT-PCR analysis in porcine LM and AD¹

Gene	Protein	Primer Sequence	NCBI Reference
ESR1	ER-α	5′-AGG GTG CCA GGA TTT TTG GA 3′-GCG AGA TGA TGT AGC CAG CA	NM_214220.1
FASN	FASN	5′-CGTTGGGTCGACTCACTGAA 3′-GAGACAGTTCACCATGCCCA	NM_001099930.1
FBXO32	Atrogin-1	5′-CAG CTC ACA TCC CTG AGT GG 3′-GAC TTG CCG ACT CTC TGG AC	NM_001044588.1
HK2	HK2	5′-AGA TGA TCG CCT CGC ATC TG 3′-GCT CCA AGC CCT TTC TCC AT	NM_001122987.1
HMBS	HMBS	5′-AGCTTGATCCCTGTGTGTGC 3′-TATTTTCTTCCGCCGTTGCG	NM_001097412.1
HPRT1	HPRT1	5′-GGG AGG CCA TCA CAT CGT AG 3′-CGC CCG TTG ACT GGT CAT TA	NM_001032376.2
IGF1R	IGF1R	5′-GGA ACT GTA TGG TGG CCG AA 3′-GTG GCA ATC TCC CAG AGG AC	NM_214172.1
INSR	INSR	5′-ATT TTC GCT GGA CTC TGC CA 3′-CAA AGG ACA GCA GCG GTA GA	XM_005654749.2
IRS1	IRS1	5′-CCCTTCTGGGAGCAGCTATG 3′-CTTCGGGGCCATAGTAGCAG	NM_001244489.1
PDK4	PDK4	5′-AAG CCA CAT TGG CAG CAT TG 3′-GGT GTT CGA CTG TAG CCC TC	NM_001159306.1
PPARG2	PPAR-γ2	5′-CAAACATTTCACAAGAGGTGACCA 3′-ATAATAAGGCGGGGACACAGG	NM_214379.1
PPARGC1A	PGC-1α	5′-CAC CAG CCA ACA CTC AGC TA 3′-TTG AGA AGC TCG GAG CAT GG	NM_213963.2
SREBF1	SREBP1	5′-GCG AGT CAA GAC CAG TCT CC 3′-TCC CCA TCC ACG AAG AAA CG	NM_214157.1
SLC2A4	GLUT4	5′-GGCCATCGTCATTGGCATTC 3′-GTCAGGCGCTTCAGACTCTT	NM_001128433.1
TBP2	TBP2	5′-AGGGTTTCAGGAAGACGACG 3′-GCTGCAACCTAAACCGAACG	XM_013991786.1
TRIM63	TRI63	5′-CAT GTG CAA GGA GCA CGA AG 3′-CCA GCA TGG AGA TGC GGT TA	NM_001184756.1

¹ESR-α, estrogen receptor alpha; FASN, fatty acid synthase; FBXO32, F-box protein 32; HK2, hexokinase 2; HMBS, hydroxymethylbilane synthase; HPRT1, hypoxanthine phosphoribosyltransferase 1; IGF1R, insulin-like growth factor 1 receptor; INSR, insulin receptor; IRS1, insulin receptor substrate 1; PDK4, pyruvate dehydrogenase kinase 4; PPARG2, peroxisome proliferator activated receptor gamma isoform-2; PPARGC1-α, peroxisome proliferator activated receptor coactivator 1 alpha; SREBF1, sterol regulatory element binding transcription factor 1; SLC2A4, solute carrier family 2 member 4; TBP2, TATA box binding protein 2; TRIM63, tripartite motif containing 63.

Calculations and Statistical Analysis

Nitrogen retention was calculated as N intake – (fecal N + urine N), where N intake was calculated from feed intake and analyzed N concentration of the associated batch of feed, and N excretion was calculated from fecal and urinary outputs. Fecal N output (g/d) was calculated from N intake and apparent fecal N digestibility, with N digestibility estimated using titanium dioxide as an indigestible marker.

The NRC (2012) gestating sow model was used to calculate pregnancy-associated Pd in each pool (fetus, mammary gland, uterus, and placenta and fluids) based on actual litter size and mean piglet birth weight (or weight at day 111 ± 0.4 of gestation for 12 P gilts that were euthanized) for individual P gilts. Pregnancy-associated Pd was subtracted from whole body Pd to arrive at maternal Pd. Whole body and maternal Pd were considered synonymous for NP gilts.

The apparent postabsorptive efficiency of using dietary Lys for whole body Pd (above maintenance) was calculated on an individual basis for each N balance period using estimated Lys contents of maternal body, fetus, uterus, mammary gland, placenta, and fluids and accounting for maternal maintenance requirements (NRC, 2012; Miller et al., 2016); standardized ileal digestible Lys intake was calculated based on estimated dietary Lys concentration and estimated SID of Lys-containing ingredients according to NRC (2012).

The minimal model (MINMOD) of glucose homeostasis developed by Toffolo et al. (1980) was fit to serum glucose and insulin curves during a FSIGTT to quantify the degree of whole body insulin sensitivity and pancreatic responsiveness. For analysis of the FSIGTT, areas under insulin and glucose curves (AUC_I and AUC_G, respectively) above the fasting level for the entire time course (90 min) were estimated using the trapezoidal rule between each time point (Cardoso et al., 2011). Parameters of glucose–insulin dynamics were estimated by fitting the minimal models of glucose disappearance and insulin kinetics described in the MINMOD computer program (Pacini and Bergman, 1986) to observed glucose and insulin concentrations following each intravenous glucose infusion. Model equations were written in ACSLX (Aegis Technologies Group, Inc., Orlando) and solved with a fourth-order Runge Kutta algorithm using an integration step size of 0.001 min. Parameters were estimated with a differential evolutionary algorithm (Storn and Price, 1997) to minimize the residual sums of squares between predicted and observed glucose and insulin concentrations. For model fitting purposes, glucose and insulin concentrations measured 60 and 15 min before the infusion were averaged to obtain basal concentrations, and residuals prior to 6 min of the FSIGTT were zero-weighted. The evolutionary algorithm was run for 900 generations with 80 sets of parameter values in each generation. Best-fit parameter estimates were used to calculate the insulin sensitivity (S_I) describing the ability of insulin to promote glucose disappearance, glucose effectiveness (S_G) describing the ability of glucose to promote its own disappearance independent of insulin, and first- (ϕ₁) and second-(ϕ₂) phase pancreatic responses according to Pacini and Bergman (1986).

For RT-PCR, fold changes in mRNA abundance were calculated by the 2-ΔΔCt method (Livak and Schmittgen, 2001) after normalizing to the geometric mean of cycle threshold values for HPRT2, TBP2, and HMBS3 as reference genes for both muscle and adipose tissue. Reference genes were chosen based on their suitability for the tissue types analyzed, according to Nygard et al. (2007).

Statistical analyses were conducted using the mixed model procedure of SAS (v9.4) with repeated measures (when necessary) and gilt as the experimental unit (SAS Inst. Inc., Cary, NC). For measures taken over multiple time points, the model included the fixed effect of physiological state (P vs. NP), repeated effect of day of study or sampling time, and their interactions. Random effects of block and gilt were included. Preplanned contrasts were constructed to compare physiological states within each time point. Linear regression analyses were performed for BW during gestation using individual means to generate the regression equations. Degrees of freedom were computed with the Kenward–Roger adjustment for repeated measures. A P < 0.05 was considered significant.

RESULTS

Growth Performance and Nitrogen Balance

One NP gilt died on day 73 of the study and her data were removed before analysis. Around breeding (day 0 of the study), BW (day −5 ± 0.6; 127.3± 2.0 kg) and BF (day −7 ± 0.6; 12.1 ± 0.3 mm) did not differ between P and NP gilts. There was an interaction between physiological state and day of study (P < 0.001), where P gilts weighed more than NP gilts at each time point between day 56 and 110 (Fig. 2). Back fat was not influenced by the interaction between study day and physiological state, but was greater (P < 0.05) for P gilts compared to NP gilts and increased (P < 0.001) with study day (data not shown). At the final measurement (day 101), BF tended to be greater for P compared to NP gilts (14.9 vs. 13.6 ± 0.5 mm; P = 0.08).

Figure 2.

Body weight (± SEM) of P and NP gilts from day 15 to 110 ± 0.6 of gestation (or equivalent day in NP). The linear regression describing BW in P gilts (coefficients ± SEM) is BW = 0.548 ± 0.03 × (day) + 122.65 ± 2.66; slope and intercept different from zero (P < 0.001); R² = 0.810. The linear regression describing BW in NP gilts (coefficients ± SEM) is BW = 0.33 ± 0.01 × (day) + 125.97 ± 1.85; slope and intercept different from zero (P < 0.001); R² = 0.653. P_PS, P_D, and P_{PS × D} represent P-values for physiological state (PS), study day (D), and the interaction between physiological state and study day (PS × D).

An interaction between physiological state and day of study (P < 0.001) was detected for urinary N excretion, whole body Pd, and estimated efficiency of Lys retention, where urinary N excretion decreased (P < 0.01) and whole body Pd and efficiency of Lys retention increased (P < 0.01) with study day for P gilts, but did not change for NP gilts (Table 3). Model-derived pregnancy-associated Pd increased (P < 0.001) between day 63 and 102 for P gilts. Maternal Pd and N digestibility were lower (P < 0.05) in P compared to NP gilts, but were not influenced by study day or the interaction between physiological state and study day.

Table 3.

Nitrogen balance of P and NP gilts at study day 63 and 102 ± 0.2

	P		NP			P-value1
Item	Day 63	Day 102	Day 63	Day 102	SEM2	PS	D	PS × D
No.3	21	21	21	20
Feed allowance,4 kg/d	2.16	2.16	2.16	2.16	―	―	―	―
N digestibility, %	87.2	86.9	87.9	88.2	0.4	0.014	0.992	0.529
Urinary N excretion, g/d	38.1a	31.2b	39.2	40.2	1.0	<0.001	<0.001	<0.001
Fecal N excretion, g/d	7.2	7.3	6.8	6.6	0.2	0.014	0.931	0.496
Whole body Pd,5 g/d	69.7a	109.7b	66.7	55.1	4.5	<0.001	0.002	<0.001
Pregnancy-associated Pd,6 g/d	26.9	65.5	―	―	2.3	―	<0.001	―
Maternal Pd,7 g/d	42.7	45.9	66.8	55.0	4.5	0.002	0.298	0.074
Lys utilization efficiency, %	29.8a	45.4b	29.4	25.5	2.2	<0.001	0.007	<0.001

¹Probability values for the main effects of physiological state (PS), study day (D), and the interactive effect of physiological state and day (PS × D).

²Maximum value of the standard error of the means.

³Data from 1 NP gilt was eliminated before analyses due to mortality at d 73.

⁴Mean of all 3 blocks. Feed allowance based on actual mean BW for each block, mean N intake = 56.1 g/d.

⁵Pd = N retention × 6.25.

⁶Represents protein deposition (Pd) attributed to pregnancy-associated tissues (fetus, mammary gland, uterus, and placenta and fluids), calculated using the NRC (2012) gestating sow model, based on actual litter size and mean piglet birth weight.

⁷For P gilts, maternal Pd = whole body Pd – maternal Pd; for NP gilts, maternal Pd = whole body Pd.

^a,bMeans without a common superscript letter within a row and physiological state differ (P < 0.05).

Blood Sampling and Frequently Sampled Intravenous Glucose Tolerance Tests

Changes in fasted serum insulin, glucose, E2, and plasma IGF-1 and E1S concentrations are shown in Table 4 and Fig. 3. Fasted serum insulin and glucose concentrations were not influenced by the interaction between physiological state and study day, were not different between P and NP gilts, but increased (P < 0.001) with study day for both P and NP gilts (Table 4). For plasma IGF-1, there was an interactive effect between physiological state and study day (P < 0.01), where IGF-1 concentration decreased with study day for P gilts and did not change for NP gilts. There was an interactive effect between physiological state and study day (P < 0.001) for E2 and E1S, where concentrations were not different for P and NP on day 43, 56, and 71, but were greater (P < 0.001) for P compared to NP gilts for the remaining sampling time points (Fig. 3).

Table 4.

Fasted serum insulin and glucose, and plasma insulin-like growth factor 1 (IGF-1) concentrations of P and NP gilts

	Day 43		Day 56		Day 71		Day 85		Day 99		Day 108			P-value1
Item	P	NP	P	NP	P	NP	P	NP	P	NP	P	NP	SEM2	PS	D	PS × D
No.3	21	20	21	20	20	20	21	20	19	20	21	20
Insulin,  pM	30.9	36.0	34.8	34.7	36.0	44.9	38.7	37.5	37.3	35.2	48.3	55.64	4.1	0.354	<0.001	0.470
Glucose,  mM	3.83	4.03	3.60	3.64	3.58	3.82	3.59	3.85	3.81	3.89	4.26	4.26	0.11	0.098	<0.001	0.659
IGF-1,  ng/mL	81.0	84.6	77.3a	83.7b	77.5x	83.4y	77.2a	83.5b	73.7a	86.1b	73.1a	85.0b	2.2	0.005	0.104	0.009

¹Probability values for the main effects of physiological state (PS), study day (D), and the interactive effect of physiological status and day (PS × D).

²Maximum value of the standard error of the means.

³Data from 1 NP gilt was eliminated before analyses due to mortality at day 73. Missing observations for P gilts occurred when previous afternoon meal was not eaten.

^a,bMeans without a common superscript letter within a time point differ (P < 0.05).

^x,yMeans without a common superscript letter within a time point tended to differ (0.05 > P < 0.10).

Figure 3.

Fasted serum E2 and plasma E1S concentrations of P and NP gilts throughout gestation (or equivalent day in NP). P_PS, P_D, and P_{PS × D} represent P-values for physiological state (PS), study day (D), and the interaction between physiological state and study day (PS × D), respectively for E2 and E1S.

At the day 75 FSIGTT, fasting serum glucose and insulin concentrations were not different between P and NP gilts before the infusion (Table 5). Serum glucose and insulin concentrations peaked at time 0 for both P and NP gilts (27.8 ± 0.4 mM and 491 ± 22 pM, respectively; Fig. 4A and B). Serum glucose was greater (P < 0.05) for P gilts compared to NP gilts between 3 and 45 min (Fig. 4A, panel A). Serum insulin was greater (P < 0.023) for P gilts compared to NP between 21 and 35 min after the glucose infusion (Fig. 4B).

Table 5.

Measured and fitted parameters (minimal model) FSIGTT for P and NP gilts on study day 75 ± 0.7 and day 107 ± 0.4

	Day 75		Day 107			P-value1
Item	P	NP	P	NP	SEM2	PS	D	PS × D
No.	6	5	17	17
Measured parameters
Fasting glucose, mM	4.38	4.14	4.62	4.31	0.27	0.349	0.247	0.833
Total AUCG, mM · min3	264.2	118.4	287.9	194.4	40.6	<0.001	0.129	0.421
Fasting insulin, pM	57.3	49.7	48.9	62.0	10.8	0.743	0.817	0.218
Total AUCI, nM · min3	10.8	8.6	10.3	6.8	1.5	0.017	0.321	0.596
Fitted parameters
SI, pM ⋅ min−1 × 104	0.78	2.73	1.11	1.91	0.62	0.004	0.595	0.209
SG, min−1	0.044	0.025	0.034	0.037	0.013	0.420	0.947	0.259
ϕ 1, pM⋅mM−1⋅min	71.1	105.2	62.4	70.0	41.6	0.530	0.479	0.669
ϕ 2, pM⋅mM−1⋅min−2	18.1	3.6	282.6	65.6	156.3	0.398	0.224	0.449
rMSPEG, mM4	0.22	0.19	0.22	0.12	0.03	0.736	0.908	0.265
rMSPEI, pM5	0.12	0.10	0.11	0.12	0.02	0.859	0.789	0.166

¹Probability values for the main effects of physiological state (PS), study day (D), and the interactive effect of physiological status and day (PS × D).

²Maximum value of the standard error of the means.

³Area under the curve (AUC) for glucose (G) or insulin (I) during the 90 min FSIGTT.

⁴Residual mean squared prediction error for glucose; lower value signifies a better fit of the minimal model to observed data.

⁵Residual mean squared prediction error for insulin; lower value signifies a better fit of the minimal model to observed data.

Figure 4.

Serum glucose (panel A) and insulin (panel B) concentrations for P and NP gilts at day 75 ± 0.7 of gestation (or equivalent day in NP) following a glucose infusion (0.5 g glucose/kg BW). P_PS, P_T, and P_{PS × T} represent P-values for physiological state, sampling time, and the interaction between physiological state and sampling time, respectively.

At the day 107 FSIGTT, serum glucose concentration peaked at time 0 (27.9 ± 0.3 mM; Fig. 5A) and serum insulin peaked at 12 min after the glucose infusion (392 ± 12.0 pM; Fig. 5B) for both P and NP gilts. Serum glucose was greater (P < 0.05) for P gilts compared to NP gilts between 9 and 50 min after the glucose infusion (Fig. 5A). Serum insulin was greater (P < 0.05) for P gilts compared to NP gilts between 21 and 45 min after the glucose infusion (Fig. 5B). Overall, the AUC_G, AUC_I, and S_I were not influenced by the interaction between physiological state and study day, or the main effect of study day, but were greater (AUC_G and AUC_I) and less (S_I) for P compared to NP gilts (P < 0.05; Table 5). There were no differences in S_G, ϕ ₁, or ϕ ₂ between the P and NP gilts at either stage of gestation.

Figure 5.

Serum glucose (panel A) and insulin (panel B) concentrations for P and NP gilts at day 107 ± 0.4 of gestation (or equivalent day in NP) following a glucose infusion (0.5 g glucose/kg BW). P_PS, P_T, and P_{PS × T} represent P-values for physiological state, sampling time, and the interaction between physiological state and sampling time, respectively.

RT-PCR

The influence of physiological state on the mRNA abundance of candidate genes in the LM and AD are shown in Table 6. In LM, mRNA abundance of ESR1, SLC2A4, and FASN were lower (P < 0.05) and PDK4 mRNA abundance was greater (P < 0.05) for P gilts compared to NP gilts. All other candidate genes were not different in the LM between P and NP gilts. In AD, mRNA abundance of SLC2A4 and SREBF1 were lower (P < 0.05) and IGF1R was greater (P < 0.05) in P gilts compared to NP gilts. There was a tendency for mRNA abundance of FASN and HK2 to be lower (P = 0.057 and 0.081, respectively) and PPARGC1A to be greater (P = 0.062) for P compared to NP gilts. All other candidate genes were not different in the AD between P and NP gilts.

Table 6.

mRNA abundance of 13 genes involved in energy and AA metabolism for P and NP gilts in LM and s.c. adipose tissue (AD) at study day day 111 ± 0.4

	LM				AD
Gene	P	NP	SEM1	P-value2	P	NP	SEM1	P-value2
No.	12	11			6	6
ESR1	0.565	0.957	0.081	0.002	1.203	1.948	0.602	0.402
INSR	1.061	1.004	0.072	0.567	1.230	1.008	0.106	0.168
IGF1R	1.142	1.031	0.109	0.476	1.371	1.025	0.110	0.050
IRS1	1.402	1.254	0.248	0.672	1.240	1.007	0.110	0.166
SLC2A4	0.768	1.151	0.126	0.040	0.443	1.266	0.261	0.050
FBXO32	0.917	1.038	0.109	0.441	1.558	1.073	0.212	0.126
TRIM63	0.944	1.045	0.146	0.632	2.321	1.277	0.441	0.114
SREBF1	0.962	1.226	0.149	0.214	0.505	1.131	0.199	0.050
FASN	0.361	0.845	0.114	0.005	0.514	1.093	0.191	0.057
HK2	0.832	1.066	0.098	0.107	0.549	1.128	0.211	0.081
PDK4	15.712	1.696	3.930	0.018	3.200	1.577	0.865	0.199
PPARG2	1.071	1.149	0.245	0.820	1.059	1.132	0.292	0.864
PPARGC1A	0.894	0.869	0.085	0.832	1.849	1.092	0.254	0.062

¹Maximum value of the standard error of the means.

²Probability values for the main effects of physiological state (PS).

Discussion

The aim of the current study was to investigate the effect of pregnancy on protein deposition, insulin sensitivity, and mRNA abundance of genes involved in energy and AA metabolism in gilts. By mid gestation (i.e., at day 63), maternal Pd was lower for P gilts, while whole body Pd was not different, which reflects preferential AA partitioning toward the pregnancy-associated tissues, even during a time of relatively low AA requirements for these tissues (NRC, 2012). This partitioning of AA (and energy) toward pregnancy-associated tissues is likely driven by the reduction in plasma IGF-1 (measured on day 56) and whole body insulin sensitivity (measured on day 75) and increase in plasma and serum estrogens (E1S and E2; measured on day 85) for P gilts. The E2 is known to bind to receptors on the liver, the main production site of IGF-1 (Murphy and Friesen, 1988), which likely reduces the hepatic production of IGF-1 and thus slows the drive for maternal growth during pregnancy; it is unknown if similar results would be observed for (mature) sows. In the second N balance (starting at day 102; late gestation), whole body Pd was greater and maternal Pd tended to be lower for P gilts, reflecting both increased N (Lys) utilization efficiency and even greater partitioning toward pregnancy-associated tissues when exponential fetal growth is occurring (NRC, 2012).

Serum insulin concentrations peaked at the same concentrations and times after glucose infusion for P and NP gilts during both FSIGTT sampling periods, suggesting that the pancreas of P gilts functioned similarly to that of NP gilts in both mid and late gestation in the current study. Furthermore, insulin-independent glucose metabolism (glucose effectiveness; S_G) was not affected by physiological state. Others noted a delay in insulin secretion and lower peak insulin concentration during FSIGTT of multiparous sows that arose between day 85 and 108 of gestation (Père et al., 2000), and late-pregnancy-induced delays in glucose clearance after intravenous glucose injections in sows and gilts (George et al., 1978; Père et al. 2000; Père et al. 2007). Pancreatic insulin release is described as biphasic, where ϕ ₁ (phase one) consists of a transient and rapid insulin release, while ϕ ₂ (phase 2) consists of a gradual insulin release due to a slow recruitment of insulin from a pancreatic storage or reserve pool (Rorsman et al., 2000). Neither phases of the pancreatic response to insulin were different between P and NP gilts, further verifying that the pancreas was functioning similarly in P gilts compared to NP gilts in the current study and that the impairment of glucose clearance must be downstream from pancreatic insulin release (i.e., at the level of peripheral tissues).

In general, once insulin is released from the pancreas, it binds to the insulin receptor (INSR) in various tissues, including skeletal muscle and adipose tissue, to facilitate glucose and AA uptake via activation of several phosphorylation cascades (Tremblay et al., 2005). Primary among these cascades is that induced by phosphorylation of insulin receptor substrate 1 (IRS-1), and downstream activation of protein kinase B (Akt). Activated Akt stimulates translocation of glucose transporter 4 [GLUT4; encoded by solute carrier family 2 member 4 (SLC2A4)] to the cell membrane, which is an insulin-dependent glucose transporter found in skeletal muscle and adipose tissues (Barbour et al., 2007). Activated Akt also stimulates mechanistic target of rapamycin complex 1-mediated activation of protein synthesis, and inhibits protein degradation by the ubiquitin-proteasome pathway (Wang et al., 2006; Barbour et al., 2007; Shimobayashi and Hall, 2014). Any impairment of insulin binding to its receptor or downstream signaling can result in an insulin-resistant phenotype.

While mRNA abundance does not necessarily equate to gene expression, and only 2 tissues were sampled in the current study, some inferences can be made from the differences in mRNA abundance for genes relating to glucose and AA metabolism between P and NP gilts. The INSR and IRS-1 mRNA abundance were not different between P and NP gilts, suggesting that the impairment in the insulin signaling cascade is downstream of INSR and IRS1 expression in LM and AD. The lower GLUT4 mRNA abundance observed in LM and AD tissue of P gilts compared to NP gilts in the current study agrees with previous studies of skeletal muscle in rats given doses of E2 to mimic pregnancy (compared to control; Barros et al., 2008) and adipose tissue in P humans (compared to NP; Okuno, 1995), suggesting that E2 plays a role in regulation of GLUT4 mRNA abundance during pregnancy. Once transported into cells, the first step of intracellular glucose metabolism is phosphorylation by hexokinase 2 (HK2), whose mRNA also tended to be reduced in abundance in AD of P versus NP gilts. Decreased expression of HK2 in the periphery is associated with insulin resistance and type II diabetes (Ducluzeau et al., 2001). Similarly, muscle expression of pyruvate dehydrogenase kinase 4 (PDK4) is elevated in type II diabetes and contributes to hyperglycemia (Kwon and Harris, 2004), and was found to be more abundant in LM of P compared to NP gilts. The PDK4 enzyme inhibits pyruvate oxidation to acetyl-CoA and thereby slows glucose utilization (Kwon and Harris, 2004). Moreover, the reduction of sterol regulatory element binding transcription factor 1 (SREBF1) and its target, fatty acid synthase (FASN), in AD (and LM for FASN) indicate a downregulation of lipid synthesis from glucose in P gilts compared to NP. In conjunction with repressed lipogenic gene expression, there was an increased abundance of peroxisome proliferator activated receptor coactivator 1 alpha (PPARGC1α) in AD of P gilts, which stimulates expression of genes of fatty acid oxidation (Puigserver and Spiegelman, 2003). Typically, insulin stimulates mRNA expression of SREBF1, GLUT4, and HK2 in peripheral tissues, and represses expression of PDK4 (Ducluzeau et al., 2001; Tsintzas et al., 2013). We observed that all of these insulin-induced effects were attenuated in P versus NP gilts. Together, these changes in mRNA abundance support a reduction in glucose utilization in maternal tissues in late gestation.

In the current study, the mRNA abundance of FBX032 and TRIM63 were not different in LM and AD tissue of P and NP gilts, suggesting that the lower maternal Pd in P gilts may be related to lower overall protein synthesis and not greater protein degradation in the maternal tissues. The concept of reduced (maternal) protein synthesis for P gilts in late gestation is also supported by reduced plasma IGF-1 concentrations. It is important to note, however, that even in late gestation, maternal protein deposition was still occurring for these P gilts. Therefore, for immature sows, maternal growth is not completely arrested, even in late gestation when the fetal demand for nutrients and energy is high and feed allowance is still restricted.

In summary, pregnancy triggers a multifactorial metabolic shift in the gestating gilt, which is apparent as early as day 75 of gestation, and is suspected to be influenced by increasing concentrations of plasma E2 and subsequent changes in mRNA abundance of key genes involved in glucose metabolism. One of the major changes in metabolism is due to insulin resistance, which was present at both day 75 and 107 of gestation compared to NP, and is characterized by lower peripheral tissue insulin sensitivity, but no change in pancreatic responsiveness. This finding supports the concept that the pancreas is functioning normally, but there is modification at the level of the peripheral tissues (skeletal muscle and adipose tissue) impeding normal insulin-dependent glucose uptake, which coincides with lower maternal Pd and mRNA abundance of genes involved in glucose metabolism.