Chronically Increased Amino Acids Improve Insulin Secretion, Pancreatic Vascularity, and Islet Size in Growth-Restricted Fetal Sheep

Abstract

Placental insufficiency is associated with reduced supply of amino acids to the fetus and leads to intrauterine growth restriction (IUGR). IUGR fetuses are characterized by lower glucose-stimulated insulin secretion, smaller pancreatic islets with less β-cells, and impaired pancreatic vascularity. To test whether supplemental amino acids infused into the IUGR fetus could improve these complications of IUGR we used acute (hours) and chronic (11 d) direct fetal amino acid infusions into a sheep model of placental insufficiency and IUGR near the end of gestation. IUGR fetuses had attenuated acute amino acid-stimulated insulin secretion compared with control fetuses. These results were confirmed in isolated IUGR pancreatic islets. After the chronic fetal amino acid infusion, fetal glucose-stimulated insulin secretion and islet size were restored to control values. These changes were associated with normalization of fetal pancreatic vascularity and higher fetal pancreatic vascular endothelial growth factor A protein concentrations. These results demonstrate that decreased fetal amino acid supply contributes to the pathogenesis of pancreatic islet defects in IUGR. Moreover, the results show that pancreatic islets in IUGR fetuses retain their ability to respond to increased amino acids near the end of gestation after chronic fetal growth restriction.

Placental insufficiency disrupts development of the fetal endocrine pancreas and results in islets that are smaller and contain fewer β-cells than normally developed islets (1). These structural defects are associated with functional deficits, including decreased fetal glucose-stimulated insulin secretion (GSIS) (2). Placental insufficiency is characterized by decreased nutrient and O₂ transfer to the fetus (3). These nutrient transfer defects produce characteristic responses in intrauterine growth restriction (IUGR) fetuses, which include not only lower insulin concentrations and GSIS, but also lower concentrations of plasma glucose and IGF-1, lower blood O₂ content, and higher plasma concentrations of norepinephrine, cortisol, and glucagon (2, 4–7). One of the most fundamental deficiencies in placental insufficiency is decreased placental-to-fetal transfer of amino acids, especially the essential and branched chain (leucine, isoleucine, and valine) amino acids (8–10). However, it is unknown to what extent decreased amino acid transfer from the placenta to the fetus, relative to other placental functional defects, is responsible for fetal pancreatic islet structural and functional defects seen in late gestation IUGR fetuses. This question is critical as insulin is a dominant fetal growth factor (11). Furthermore, fetal pancreatic defects may underlie the long term increase in risk that former IUGR individuals have of developing β-cell failure in the face of insulin resistance leading to type 2 diabetes (12).

Offspring from animal models of placental insufficiency and IUGR are characterized by lower insulin, GSIS, islet size, and β-cell numbers (13). In some models, pancreatic and islet vascularity also has been found to be lower, similar to human fetuses with severe IUGR (1, 14–16). In this study, we used direct fetal amino acid infusions into a sheep model of placental insufficiency and IUGR to test the role of amino acids in these pancreatic defects. These sheep develop progressive and severe placental insufficiency and share many characteristics with severe human placental insufficiency including lower rates of amino acid transfer to the fetus and lower fetal insulin, GSIS, islet size, and β-cell numbers (3, 13, 17, 18). Furthermore, defects in fetal islet structure and function are associated with lower pancreatic vascularity and impaired islet hepatocyte growth factor (HGF) and vascular endothelial growth factor (VEGF)A paracrine signaling between the β-cell and vascular endothelial cell (16, 19).

Acute amino acid infusions (1.5 h) were used to test the responsiveness of amino acid-stimulated insulin secretion (AASIS) in IUGR fetuses in vivo. Confirmatory tests also were performed in isolated fetal pancreatic islets incubated in increasing amino acid concentrations in vitro. Chronic (∼11 d) amino acid infusions were used to determine if increasing the fetal amino acid supply, especially the supply of essential and branched chain amino acids (BCAAs), over the latter part of gestation could reverse structural and functional fetal islet defects in IUGR fetuses. At the end of the infusion we measured in vivo fetal GSIS and collected the pancreas for histological analysis and measurement of mRNA and proteins associated with insulin secretion, islet growth, vascularity, and paracrine islet signaling. Finally, we used isolated fetal pancreatic islets and fetal endothelial cells to test the relationship between amino acid supplementation and the paracrine regulation of islet function by HGF and vascular endothelial growth factor A (VEGFA). Our major findings show that, similar to acute glucose stimulation, acute amino acid stimulation of insulin secretion and insulin production is attenuated in IUGR fetuses and fetal islets. However, chronically increasing the amino acid supply for approximately 11 days to IUGR fetuses normalizes fetal pancreatic islet size and insulin secretion.

Research Design and Methods

Studies were conducted in pregnant Columbia-Rambouillet ewes carrying a singleton fetus in compliance with the Institutional Animal Care and Use Committee, University of Colorado Denver, National Institutes of Health, and United States Department of Agriculture. The laboratory is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. Placental insufficiency was induced by exposing pregnant sheep to elevated ambient temperatures for 72.1 ± 0.8 days beginning on 39.4 ± 0.4 days gestation age (16, 20). Control (CON) ewes were maintained at normal ambient temperatures and fed to treated ewe average intake. IUGR sheep were placed in normal ambient temperatures before surgery for placement of maternal and fetal vascular catheters (21).

Acute in vivo amino acid stimulated insulin secretion

AASIS was measured with a fetal hyperaminoacidemic clamp using a 10% TrophAmine (wt/vol) bolus (100-mg amino acids per kg estimated fetal weight) at time 0 followed by a constant infusion (350-mg amino acids per kg estimated fetal weight per hour) (22). A continuous transfusion of maternal whole blood into the fetus compensated for sampled fetal blood. Plasma glucose, amino acids, insulin, glucagon, and IGF-1 concentrations were determined before and throughout the clamp.

Chronic in vivo amino acid infusion with GSIS

Some IUGR fetuses were used for a chronic fetal amino acid infusion group with 10% TrophAmine (wt/vol) (IUGR+amino acids [AA], n = 10) and a chronic saline infusion group (n = 8). Some CON fetuses also received a chronic saline infusion (n = 7). The TrophAmine infusion was adjusted daily to achieve a 25%–50% increase in BCAA concentrations. IUGR and CON fetuses received a saline infusion to match the sodium delivery from TrophAmine (21, 23). The duration of infusions was similar among groups (11.0 ± 0.9 d, CON; 10.1 ± 1.0, IUGR; and 10.6 ± 0.7, IUGR+AA).

On the final day of the chronic infusions, a fetal hyperglycemic clamp followed by a fetal arginine infusion was used to determine fetal GSIS and glucose-potentiated arginine-stimulated insulin secretion (GPASIS) (21, 24). We calculated GPASIS and arginine stimulated glucagon secretion as the difference between the hormone concentrations immediately before and 5 minutes after the arginine infusion. The chronic infusions continued through the insulin secretion studies until tissue collection the next day.

Biochemical analyses

Biochemical analysis was performed as previously described (21, 23, 25). Carbon dioxide, O₂, pH, and hematocrit were measured with the ABL 520 analyzer (Radiometer). Plasma glucose and lactate concentrations were measured with the Yellow Springs Instruments model 2700 biochemistry analyzer. Plasma BCAA concentrations were measured with a spectrophotometric assay. Insulin, cortisol, and IGF-1 concentrations were measured by ELISA (ALPCO Immunoassays), glucagon by RIA (Millipore), and norepinephrine by HPLC. Amino acids were measured with a Dionex 300 model 4500 amino acid analyzer.

Organ isolation

At necropsy, some fetal pancreases from all groups were weighed and the splenic portion fixed in 4% paraformaldehyde (wt/vol) and frozen (16). The hepatic portion was snap frozen in liquid nitrogen and transferred to −80°C. Other pancreases from CON and IUGR fetuses were used for islet isolation (20). The placenta was dissected and weighed.

Pancreatic histology

The proportion of the fetal pancreas comprised of β-cells and α-cells, islet size, pancreatic and islet vascularity, β-cell size, and β-cell mitosis (β-cell costaining with insulin and phosphorylated histone 3 [pHH3]) were measured by individuals blinded to treatment groups (16, 26, 27).

Pancreatic islet incubation

Pancreatic islets were isolated from 10 CON and 16 IUGR fetuses and incubated overnight in 95% O₂ and 5% CO₂ in RPMI 1640 with 2.8 mmol/L glucose and 1% (vol/vol) fetal bovine serum (20). Experimental additions included recombinant human HGF (100-ng/mL ProSpec) or amino acids to a final concentration of 2× or 6× based on calculations for leucine (MEM Amino Acids 50× solution, catalog number 25–030; Corning Cellgro). The next day islets were washed, and islet insulin secretion and content were measured (20). For each animal, insulin secretion was tested in 3–6 batches of islets (10 islets per batch) per condition.

Endothelial cell culture

Large vessel (pulmonary) endothelial cells were isolated from normally growing fetal sheep, propagated and purity verified as previously described (28). After cells (passage 4–6) reached 70% confluence, they were serum starved overnight and then incubated for 72 hours in DMEM with 10% fetal bovine serum and experimental additions, including amino acids (MEM Amino Acids 50× solution), leucine (2.4 mmol/L; Sigma), and recombinant human VEGFA (50 ng/mL; R&D Systems). Media were changed daily and mRNA was collected (29).

mRNA and protein analysis

Total RNA was extracted from pulverized pancreas or endothelial cells and reverse transcribed into complimentary DNA (21). INS (insulin), GCG (glucagon), SST (somatostatin), PPY (pancreatic polypepetide), GCK (glucokinase), IGF1 (IGF-1), and IGF2 (IGF-2) mRNA were measured as previously described (21, 30, 31). Real-time quantitative PCR assays were developed for VEGFA (for-TTGCCTTGCTGCTCTACCTT, rev-GGGCACACACTCCAGACTTT), HGF (for-AGGAGACGAGAAACGCAAACAGGT, rev-ATAGCAGGCCTAGCAAGCTTCAGT), MET (HGF receptor [cMET]; for-GAATGGCCTGGGCTGTGAACATTT, rev-TGTTGGCAGACAGGTCTCTTGAGT), and FLT1 (VEGF receptor 1 [VEGFR1]; for-TTGTCATGCTAATGGCGTCCCAGA, rev-GGTTGCTGGCTTTGCAGTGATACA). Pancreatic samples were analyzed in triplicate and endothelial cell samples in duplicate. The standard curve method of relative quantification was used (32). Genes of interest were normalized to reference genes RPS15 (ribosomal protein s15), ACTA1 (actin), and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (21). Results are presented as fold change relative to the CON group for pancreatic mRNA and relative to no experimental additions for endothelial cell mRNA.

Total protein was extracted from pulverized pancreas, separated by gel electrophoresis, and transferred to a membrane for western blotting (25). The following primary antibodies were diluted in phosphate buffered saline with tween with 5% bovine serum albumin (wt/vol), except as noted: glucokinase (1:1000; Abcam), glucose transporter 2 (GLUT2) (1:1000; Santa Cruz Biotechnology), HGF (1:1000; diluted in 2% enhanced chemiluminescence Advance; R&D Systems), VEGFA (1:200 diluted in 2% enhanced chemiluminescence Advance; Santa Cruz Biotechnology), and actin (1:20,000; MP Biomedicals). Horseradish peroxidase-conjugated secondary antibodies were applied and detected. Densitometry results were normalized to β-actin and expressed relative to the CON group.

Statistical analysis

Analysis was performed using SAS 9.1 or GraphPad Prism 5 with a one-way or mixed models ANOVA which included terms for group (CON, IUGR, and IUGR+AA), time, and incubation condition. After determining that fetal sex did not impact outcomes in a preliminary mixed models ANOVA, we pooled data from male and female fetuses. Repeated measurements made in the same animal were accounted for. Individual means were compared using Fisher's least significant difference for mixed models ANOVA or Tukey's method for one-way ANOVA. Data were log transformed due to increased variability as indicated. P ≤ .05 was considered significant.

Results

Acute in vivo amino acid infusion

Acute AASIS was measured in CON (n = 5) and IUGR (n = 5) fetuses. The infusion rate was similar (377 ± 40 mg kg ⁻¹ per h⁻¹ CON, 332 ± 38 mg kg⁻¹ per h⁻¹ IUGR). Despite higher BCAA concentrations (P < .05) (Figure 1A), IUGR fetuses had attenuated AASIS (P < .0001) (Figure 1B). Glucose concentrations increased by about 10% during the hyperaminoacidemic clamp in both groups (P < .0001) (Figure 1C). Concentrations of glucagon and almost all amino acids increased in both groups during the hyperaminoacidemic clamp (P < .05) (Supplemental Table 1). IGF-1 concentrations did not change during the clamp in either group (Supplemental Table 1).

Figure 1.

IUGR fetuses have attenuated acute in vivo AASIS. CON (open circles, n = 5) and IUGR (closed circles, n = 5) underwent a primed, continuous, constant-rate hyperaminoacidemic clamp with a direct fetal infusion of 10% TrophAmine (wt/vol) starting at time 0. A, BCAA concentrations increased in both groups and were higher in the IUGR fetuses compared with CONs (*, P < .05). B, IUGR fetuses had significantly lower fetal arterial plasma insulin concentrations during the clamp (***, P < .0001). C, Fetal arterial plasma glucose concentrations increased slightly in both groups (P < .005) but were consistently lower in the IUGR-SAL fetuses (**, P < .0001). Mean ± SE fetal arterial plasma concentrations are plotted and statistics are by mixed models ANOVA.

Chronic in vivo amino acid infusion

Maternal and fetal metabolites, blood gases, and hormones

IUGR+AA fetuses were infused with increasing rates of TrophAmine (Supplemental Figure 1) and compared with IUGR and CON. The final weight adjusted infusion rate was 2.40 ± 0.17 g kg⁻¹ per d⁻¹. There were no maternal biochemical differences among the groups (Table 1). Most fetal essential amino acid concentrations were higher in IUGR+AA compared with IUGR (Table 1). Fetal lactate in IUGR+AA was greater than the other 2 groups at the end of the infusion (P < .01) (Table 1). There were no differences among the groups for fetal hematocrit, pCO₂, or pH (Table 1). Fetal pO₂, hemoglobin-O₂ saturation, total O₂ content, and glucose were lower in both IUGR groups compared with CON (P < .005), but the IUGR groups were not different from each other (Table 1). Norepinephrine and cortisol concentrations were higher and insulin and IGF-1 concentrations lower in the both IUGR groups compared with CONs (P < .05), but the IUGR groups were not different from each other (Table 1). The amino acid infusion increased glucagon concentrations in the IUGR+AA group (P < .0001) (Table 1).

Table 1.

Maternal and Fetal Arterial Biochemistry

	CON	IUGR	IUGR+AA
Maternal arterial biochemistry
Glucose (plasma, mmol/L)	3.83 ± 0.13	3.91 ± 0.06	3.98 ± 0.20
Lactate (plasma, mmol/L)	0.9 ± 0.1	0.8 ± 0.1	0.7 ± 0.0
pH	7.44 ± 0.00	7.44 ± 0.01	7.44 ± 0.01
pCO2 (mm Hg)	34.3 ± 0.6	35.6 ± 0.8	35.3 ± 0.7
pO2 (mm Hg)	87.1 ± 1.7	87.7 ± 1.5	90.6 ± 2.4
Hematocrit (%)	32.1 ± 1.1	32.3 ± 1.0	30.7 ± 0.6
sO2 (%)	95.4 ± 0.6	95.8 ± 0.6	96.5 ± 0.5
O2 content (mmol/L)	5.9 ± 0.2	6.2 ± 0.2	5.7 ± 0.1
Fetal arterial biochemistry
Glucose (plasma, mmol/L)	1.04 ± 0.03	0.74 ± 0.04a	0.72 ± 0.06a
Lactate (plasma, mmol/L)	2.3 ± 0.1	2.6 ± 0.2	5.0 ± 1.4a,b
pH	7.37 ± 0.00	7.36 ± 0.00	7.35 ± 0.01
pCO2 (mm Hg)	49.9 ± 0.9	51.4 ± 0.5	52.0 ± 1.2
pO2 (mm Hg)	19.6 ± 0.6	15.6 ± 0.5a	14.2 ± 1.2a
Hematocrit (%)	37.8 ± 1.4	37.5 ± 0.8	40.9 ± 1.7
Hemoglobin-O2 saturation (%)	49.3 ± 2.4	35.2 ± 2.0a	30.5 ± 4.6a
O2 content (mmol/L)	3.6 ± 0.2	2.6 ± 0.2a	2.5 ± 0.4a
Insulin (plasma, ng/mL)	0.37 ± 0.04	0.19 ± 0.02a	0.24 ± 0.03a
Glucagon (plasma, pg/mL)	37.8 ± 3.0	49.5 ± 5.1	216.5 ± 88.9a,b,c
Norepinephrine (plasma, pg/mL)	641 ± 110	1901 ± 532a	1694 ± 571a,c
Cortisol (plasma, ng/mL)	8.1 ± 1.0	29.9 ± 10.0a	40.7 ± 17.7a,c
IGF-1 (plasma, ng/mL)	132.0 ± 7.9	56.4 ± 6.9a	57.2 ± 13.2a,c
Fetal arterial plasma amino acid concentrations (μmol/L)
Alanine	295.9 ± 13.3	330.2 ± 15.1	391.1 ± 91.2
Arginine	106.2 ± 11.2	63.9 ± 5.9a	93.2 ± 10.1b
Asparagine	44.0 ± 3.3	55.4 ± 6.3	47.4 ± 14.1
Aspartate	19.5 ± 1.9	16.4 ± 1.2	16.4 ± 1.5
Citrulline	182.4 ± 11.8	198.6 ± 13.6	204.4 ± 23.2
Cystine	12.6 ± 0.8	13.7 ± 1.4	11.1 ± 1.7
Glutamate	44.7 ± 4.1	38.5 ± 3.3	24.4 ± 3.5
Glutamine	394.8 ± 14.7	384.9 ± 12.8	360.6 ± 45.9
Glycine	341.4 ± 17.8	350.9 ± 17.5	387.9 ± 53.7
Histidine	60.7 ± 4.5	63.1 ± 3.5	91.6 ± 12.4a,b
Isoleucine	80.8 ± 4.7	86.1 ± 5.5	129.3 ± 21.8a,b[r
Leucine	126.0 ± 7.9	146.2 ± 7.4	242.4 ± 39.5a,b
Lysine	87.6 ± 7.8	85.7 ± 5.2	192.2 ± 53.7a,b
Methionine	86.8 ± 4.6	71.2 ± 4.2	86.0 ± 10.8
Ornithine	81.3 ± 6.3	76.0 ± 5.9	92.2 ± 9.5
Phenylalanine	95.7 ± 5.7	94.1 ± 4.3	129.3 ± 24.3a,b
Proline	142.5 ± 8.2	171.7 ± 10.7	385.0 ± 95.2a,b
Serine	569.9 ± 46.3	481.3 ± 46.9	317.3 ± 46.2
Taurine	91.9 ± 12.7	164.2 ± 23.4a	252.5 ± 36.2a,b
Threonine	339.9 ± 27.4	272.7 ± 18.0a	258.8 ± 40.6a
Tryptophan	41.6 ± 2.3	39.4 ± 2.6	58.4 ± 6.6b
Tyrosine	105.9 ± 5.8	106.1 ± 7.8	125.6 ± 28.0
Valine	311.2 ± 18.3	392.5 ± 20.3b	449.8 ± 48.9a,b

Mean ± SE from measurements made immediately before the final acute physiological studies are presented in CON (n = 17), IUGR (n = 25), and IUGR-AA (n = 10). Groups were compared by one-way ANOVA.

^aSignificant difference compared with CON.

^bSignificant difference compared with IUGR (P < .05) with Tukey's correction for multiple comparisons.

^cData were log transformed for analysis.

Fetal insulin secretion

GSIS and GPASIS were measured in CON (n = 6), IUGR (n = 6), and IUGR+AA (n = 5) fetuses. After initiation of the hyperglycemic clamp, glucose concentrations were similar among all groups (Figure 2A). GSIS was restored to CON levels in IUGR+AA, both of which had more robust insulin secretion than IUGR (P < .05) (Figure 2B). There was no difference between the groups for GPASIS (1.87 ± 0.39-ng/mL CON, 1.32 ± 0.56-ng/mL IUGR, 1.28 ± 0.69-ng/mL IUGR+AA). Arginine stimulated glucagon secretion was greater in IUGR+AA (678 ± 408 pg/mL, P < .05, after log transformation) compared with the other 2 groups (82 ± 31-pg/mL CON, 89 ± 40-pg/mL IUGR).

Figure 2.

In vivo GSIS is normalized in IUGR+AA fetuses at the end of the chronic amino acid infusion period. Mean ± SE fetal arterial plasma glucose (A) and insulin (B) concentrations are shown relative to the start of a primed continuous variable-rate fetal hyperglycemic clamp (time = 0 min) in CONs (open circles, n = 6), IUGR (closed circles, n = 6), and IUGR+AA (gray squares, n = 5); *, significantly lower insulin concentrations in IUGR compared with both CON and IUGR+AA (P < .05) by mixed models ANOVA.

Characteristics of the fetal pancreas (Table 2)

Table 2.

Characteristics of the Fetal Pancreas

	CON	IUGR	IUGR+AA
Pancreatic morphology
Islet size (μm2)	2876 ± 295	1706 ± 230a	2878 ± 280a
Insulin+ area (%)	3.52 ± 0.21	2.07 ± 0.07a	2.98 ± 0.25a,d
β-Cell mass (mg)	108.1 ± 7.5	43.3 ± 5.5a	77.6 ± 12.4a,d
β-Cell size (μm2)	73.4 ± 2.9	79.8 ± 5.3	82.6 ± 5.8
β-Cell mitosis (%)	0.39 ± 0.02	0.24 ± 0.04a	0.27 ± 0.03a
Glucagon+ area (%)	0.90 ± 0.06	0.68 ± 0.10	0.78 ± 0.06
α-Cell mass (mg)	27.6 ± 2.2	15.4 ± 3.0a	20.4 ± 2.8
Islet vessel density (%)	15.03 ± 1.63	16.61 ± 1.64	14.11 ± 0.88
Pancreatic vessel density (%)	3.29 ± 0.32	2.25 ± 0.19c	3.63 ± 0.43b,d
Pancreatic protein and mRNA expression
Endocrine hormones
Insulin (mRNA)	1.00 ± 0.11	0.49 ± 0.08c	0.96 ± 0.28d
Gucagon (mRNA)	1.00 ± 0.05	0.60 ± 0.09	1.70 ± 0.46b,d
Somatostatin (mRNA)	1.00 ± 0.12	0.67 ± 0.13	0.96 ± 0.19
Pancreatic polypeptide (mRNA)	1.00 ± 0.34	0.81 ± 0.54	1.63 ± 0.61
Glucose transport and metabolism
GLUT2 (protein)	1.00 ± 0.28	4.01 ± 0.84a	2.13 ± 0.49d
Glucokinase (protein)	1.00 ± 0.07	0.90 ± 0.18	1.00 ± 0.06
Glucokinase (mRNA)	1.00 ± 0.22	0.91 ± 0.15	0.77 ± 0.22
Paracrine hormones and receptors
VEGFA (protein)	1.00 ± 0.25	2.41 ± 0.37	4.14 ± 0.54a,b
VEGFA (mRNA)	1.00 ± 0.23	1.41 ± 0.20	1.63 ± 0.59d
HGF (protein)	1.00 ± 0.13	1.35 ± 0.27	0.91 ± 0.22
HGF (mRNA)	1.00 ± 0.26	1.05 ± 0.27	0.78 ± 0.24
cMET (mRNA)	1.00 ± 0.41	1.12 ± 0.42	1.09 ± 0.40
VEGFR1 (mRNA)	1.00 ± 0.29	1.22 ± 0.32	1.51 ± 0.63d
IGF-1 (mRNA)	1.00 ± 0.25	0.67 ± 0.26	0.55 ± 0.21
IGF-2 (mRNA)	1.00 ± 0.07	0.80 ± 0.25	0.92 ± 0.09

Data are presented as mean ± SE for CON (n = 7), IUGR (n = 6–8), and IUGR-AA (n = 9).

^aSignificant difference compare with CON.

^bSignificant difference compared with IUGR (P < .05) by one-way ANOVA with Tukey's correction for multiple comparisons.

^cSignificant difference between CON and IUGR on post hoc Student's t test (P < .05).

^dData were log transformed for analysis.

Histology of the fetal pancreas

Islet size, the percentage of the pancreatic sections staining for insulin (β-cell area), and β-cell mass were all greater in IUGR+AA (n = 9) than IUGR (n = 6) (P < .05). They were equivalent in IUGR+AA to CON (n = 7). β-Cell size was not different among the groups. The percentage of β-cells staining for pHH3 was lower in both IUGR groups compared with CON. There was a positive correlation between β-cell pHH3 staining and the pancreatic β-cell area (P < .01) (Figure 3).

Figure 3.

Pancreatic insulin⁺ area is correlated with the percentage of β-cells in mitosis. Fetal pancreatic sections from CONs (open circles, n = 7), IUGR (closed circles, n = 6), and IUGR+AA (gray squares, n = 9) were immunostained for insulin to define β-cells and pHH3 to define cells in mitosis. A linear association was found between the insulin⁺ area (y-axis) and the percentage of β-cells in mitosis (x-axis) by linear regression for all fetuses combined (P < .01).

The percentage of the pancreatic sections staining for glucagon (α-cell area) was the same among groups. α-cell mass was lower in IUGR compared with CON (P < .01). The α-cell mass in IUGR+AA fetuses was intermediate. IUGR+AA pancreatic vessel density was higher than the IUGR group (P < .05) and equivalent to CON. Islet vessel density was not different among the groups.

Protein and mRNA expression (Table 2)

Expression of specific pancreatic protein and mRNAs were measured (7 CON, 8 IUGR, and 9 IUGR+AA). Pancreatic insulin mRNA expression was similar between IUGR+AA and CON. Pancreatic glucagon mRNA was higher in IUGR+AA compared with IUGR (P < .05). There were no differences among the groups for pancreatic somatostatin or pancreatic polypeptide mRNA. Pancreatic GLUT2 protein expression was higher in IUGR compared with CON (P < .05). Expression in IUGR+AA was intermediate. Pancreatic glucokinase protein and mRNA were not different among groups. Pancreatic VEGFA protein expression was higher in IUGR compared with the CON (P ≤ .05) and was even higher in IUGR+AA compared with IUGR (P < .05). Pancreatic HGF protein and mRNA, and cMET, VEGFR1, IGF-1, and IGF-2 mRNA were not different among groups.

Necropsy measurements

Placental weight, fetal weight, and fetal lengths were lower in both IUGR groups compared with CON. Relative to fetal weight, pancreases in IUGR+AA were heavier than the pancreases from IUGR or CON (P < .05) (Table 3).

Table 3.

Necropsy Measurements

	CON	IUGR	IUGR+AA
Gestational age (d)	133.6 ± 0.7	134.4 ± 0.5	134.6 ± 0.5
Sex (% male)	59	68	60
Fetal weight (gm)	3127 ± 141	2059 ± 120a	2128 ± 225a
Crown rump length (cm)	46.9 ± 0.8	42.5 ± 0.9a	41.2 ± 1.8a
Lower limb length (cm)	34.1 ± 0.8	29.9 ± 0.7a	29.1 ± 1.2a
Placentome number	74.2 ± 2.9	72.4 ± 2.7	76.6 ± 3.4
Placental weight (gm)	345.7 ± 22.0	190.8 ± 13.3a	183.1 ± 23.4a
Pancreas to fetal weight ratio (%)	0.095 ± 0.005	0.100 ± 0.006	0.127 ± 0.016a,b

Data are presented as mean ± SE for CON (n = 17), IUGR (n = 25), and IUGR-AA (n = 10). For the pancreas, n = 7, 6, and 9 in CON, IUGR, and IUGR-AA, respectively.

^aSignificant difference compared with CON.

^bSignificant difference compared with IUGR (P < .05) by one-way ANOVA with Tukey's correction for multiple comparisons.

In vitro islet insulin secretion

Pancreatic islets were isolated from CON (n = 10) and IUGR (n = 16) fetuses. Islets were not obtained from IUGR+AA fetuses. To determine if the attenuated response to acute AASIS in the IUGR fetus in vivo was independent of other in vivo abnormalities, we incubated isolated islets in increasing amino acid concentrations overnight and then measured islet insulin concentration and secretion. Insulin content was lower in IUGR islets compared with CON regardless of supplemental amino acids (P < .0005). A dose-dependent increase in islet insulin content and glucose- and potassium-stimulated insulin secretion after incubation with supplemental amino acids was observed in CON islets but not in IUGR islets (P < .05) (Figure 4A). As we have previously reported, the fractional insulin release was higher in IUGR islets compared with CON after overnight incubations in normal media (P < .05) (20). However, there were no differences between IUGR and CON islet fractional insulin release after overnight incubation in 6× amino acids (Figure 4B).

Figure 4.

In vitro pancreatic islet insulin secretion. Pancreatic islets were isolated from CON (white bars, n = 10) and IUGR (black bars, n = 16) fetuses and incubated overnight with or without additional amino acids (2× or 6×) or HGF (100 ng/mL). A, Insulin content was lower in IUGR islets than CON islets (P < .0005). CON islet insulin content increased with increasing overnight amino acid supplementation (*, relative to basal conditions, P < .005), an effect not seen in IUGR islets. B, Overnight amino acid supplementation had no effect on IUGR islet glucose and potassium-stimulated insulin secretion. However, CON islets had a dose-response increase in glucose and potassium-stimulated insulin secretion after overnight incubation with supplemental amino acids (P < .05); *, significantly more fractional insulin release from IUGR islets compared with CON islets incubated in the same conditions (P < .05). C, Overnight incubation with HGF resulted in higher insulin contents of both CON and IUGR islets (P < .05). D, Overnight incubation in HGF had no effect on CON or IUGR islet glucose or potassium-stimulated insulin secretion. Data are presented as mean ± SE, and statistics are by mixed models ANOVA.

Due to the increase in pancreatic VEGFA protein and pancreatic vessel density in IUGR+AA fetuses, we sought to further determine the relationship between amino acids, VEGFA, and the paracrine regulation of islet function and growth by endothelial cells. We incubated large vessel fetal sheep endothelial cells in supplemental amino acids and VEGFA. Amino acids dose dependently increased HGF mRNA (P < .0001) (Figure 5). Supplemental leucine and VEGFA also increased HGF mRNA (P < .0005) (Figure 5). Given these results and the role that HGF plays in endothelial cell stimulation of β-cell insulin production and secretion (16, 19), we incubated islets overnight in supplemental HGF. HGF increased the insulin content of both CON and IUGR islets (P < .05) (Figure 4C). Fractional insulin release, however, did not increase in either group (Figure 4D).

Figure 5.

Large vessel endothelial cell HGF mRNA is increased with in vitro supplemental amino acids, leucine, and VEGFA. Large vessel endothelial cells from normal fetal sheep (n = 3) were incubated for 72 hours in basal conditions or with indicated supplementation (amino acids [2× or 6×], leucine [2.4 mmol/L], VEGFA [50 ng/mL]). HGF mRNA was then measured and normalized to basal conditions. For each incubation condition, cells derived from all 3 animals were tested in replicates of 1–3; *, significant difference from basal conditions (P < .005). Data are presented as mean ± SE, and statistics are by mixed models ANOVA after log transformation.

Discussion

In this study, we have shown that acute AASIS is decreased in IUGR fetuses compared with CON, but chronically (∼11 d) increasing IUGR fetal amino acid supply and concentrations restores fetal GSIS, pancreatic vascularity, and islet size, as well as pancreatic β-cell mass. The specific ability of amino acids to restore GSIS, islet size, and β-cell mass is not a general property of all fetal nutrients, as we have previously demonstrated that chronically increasing the fetal glucose supply in IUGR has no impact on β-cell mass and actually lowers fetal GSIS (33). These results advance our understanding of the pathogenesis of β-cell dysfunction and impaired islet development in IUGR. Our novel findings point to amino acid insufficiency as fundamental in regulation of fetal islet development and insulin secretion in IUGR.

Previously published evidence of a role for decreased fetal amino acid supply in the pathogenesis of islet dysfunction in IUGR comes from pregnant rats fed a low-protein diet. The fetuses from these dams are characterized by smaller islets, less islet insulin secretion, and lower β-cell mass (14, 34–36). However, amino acid delivery has not been measured directly in the low-protein diet rat model of IUGR. This is unlike the situation for human IUGR and the sheep model of placental insufficiency and IUGR used in this study, in which decreased amino acid supply to the fetus from the placenta has been directly measured and shown to be decreased (9, 10, 18, 31). Therefore, we undertook a comprehensive investigation of the role of amino acids in the pathogenesis of islet dysfunction using this sheep model of IUGR. Similar to the low-protein diet model of IUGR and to severe human IUGR, the placental insufficiency model of IUGR used in the current study is characterized by decreased fetal insulin concentrations, decreased insulin secretion, decreased islet size, decreased β-cell mass, and decreased islet vascularity (13, 16, 20, 27).

We first measured acute fetal AASIS with frequent blood sampling during an amino acid clamp, which has never been specifically tested in a model of placental insufficiency and IUGR. The significant attenuation of the acute insulin secretory response to an infused amino acid mixture in the late gestation IUGR fetuses is consistent with previous reports of their β-cell dysfunction, as demonstrated by decreased in vivo GSIS and smaller islets with lower β-cell mass and less islet insulin content (20, 27). These functional and structural defects certainly point to a possible fixed β-cell deficit by the end of gestation. However, recent studies have shown that acute GSIS in IUGR fetuses also is regulated by catecholamines and acute adrenergic signaling, implicating a more complex situation than might be expected based on structural analysis of the IUGR pancreas alone (37, 38). Therefore, we confirmed the attenuated in vivo AASIS using isolated fetal sheep islets incubated in different concentrations of supplemental amino acids, which eliminated confounding by adrenergic signaling. Consistent with our in vivo results, only the islets isolated from CON fetuses showed an increase in islet insulin concentrations or fractional insulin release in response to acute (overnight) exposure to supplemental amino acids. In contrast, the islets isolated from IUGR fetuses were not responsive to amino acids.

Contrary to the lack of responsiveness to acute amino acid supplementation in IUGR fetuses or isolated IUGR islets, an 11-day amino acid infusion completely normalized in vivo GSIS in the IUGR fetus. This was independent of changes in glucose, O₂, IGF-1, cortisol, and norepinephrine, demonstrating specificity of the amino acid supplementation. Of the nutrients we measured, only lactate and amino acid concentrations increased. Glucagon concentrations also were notably increased in the IUGR+AA fetuses. This is consistent with previous studies in isolated α-cells, adult animals, and fetal animals that have demonstrated amino acid stimulated glucagon secretion (21, 39–41). Increased fetal plasma glucagon concentrations and fetal pancreatic glucagon mRNA in our IUGR+AA fetuses is especially noteworthy as glucagon can potentiate GSIS by binding to specific receptors on the β-cell and augmenting insulin secretion by increasing cAMP concentrations (21, 40, 42–46).

A major and unique observation in the current studies is that islet size in the IUGR+AA fetuses was equivalent to CON fetuses, providing another possible explanation for restored GSIS. The increase in islet size represented an expansion of all cells measured but with a greater expansion of β-cells compared with the other cell types. β-Cell mass in the IUGR+AA fetuses was almost completely restored to levels found in CON fetuses. We could not explain the higher proportion of β-cells by identifying larger β-cells or increased rates of β-cell mitosis based on costaining of insulin and pHH3. However, the proportion of β-cells in mitosis was directly associated with the percent of the pancreas staining for insulin, indicating that increased rates of β-cell mitosis might have played a role in the improved structure and function of the IUGR+AA fetuses.

The increase in islet size with a proportional increase in islet vascularity, as well as increased overall pancreatic vascularity in the IUGR+AA fetuses, was associated with more pancreatic VEGFA protein compared with saline infused IUGR fetuses. We did not identify any associated changes in pancreatic HGF protein or mRNA. Given the stimulatory effect that VEGFA has on islet endothelial cell HGF production and the ability of endothelial cell-produced HGF to stimulate islet growth and insulin secretion (47–49); however, we tested the effect of these growth factors and amino acids in vitro on HGF production and insulin secretion using isolated fetal islets and large vessel fetal sheep endothelial cells. Both amino acids and VEGFA significantly increased HGF mRNA in fetal sheep endothelial cells. These results are consistent with experiments in isolated adult rodent islet endothelial cells and other cell types (49, 50). Together, they show the potential for either amino acids directly or via increased pancreatic concentrations of VEGFA to increase local islet concentrations of HGF. Given these results, we tested the impact of HGF on in vitro function in isolated fetal pancreatic islets and found that HGF could increase the insulin concentrations of both CON and IUGR islets after an overnight incubation, thereby further supporting a link between amino acids, VEGFA, and the paracrine regulation of islet insulin production in both normal and IUGR islets by HGF (16, 19).

Our results pose questions for future studies. Here, we tested GSIS while the amino acids were being infused. It remains to be determined if increased GSIS will persist after the amino acid infusion is discontinued. Future studies also will be required to determine which amino acid, or combination of amino acids, was most critical for our outcomes. Given the major defects in IUGR islet morphology (16, 27), we chose to focus our studies here on the islet structure of the IUGR+AA group. However, this precluded isolation of pancreatic islets and pancreatic islet endothelial cells from these same fetuses. Thus, a new cohort of animals will be required to measure islet- and islet endothelial cell- specific proteins and mRNA as opposed to pancreatic proteins and mRNAs measured in the current study. We also will need to replicate our findings from isolated fetal sheep large vessel endothelial cells in fetal islet-derived endothelial cells as there may be differences in the response of different types of endothelial cells, again requiring a new cohort of animals.

Overall, our studies identified a novel role for AA to improve GSIS in the placental insufficiency induced IUGR fetus. An 11-day chronic infusion of a complete amino acid mixture restored fetal GSIS and islet size in IUGR fetuses. However, acute in vivo AASIS and in vitro responsiveness to supplemental amino acids were attenuated in the IUGR fetal sheep. These results demonstrate that chronically decreased fetal amino acid supply contributes directly to the pathogenesis of islet dysfunction in fetuses that are growth restricted from placental insufficiency. The results also point to the need for testing the exciting potential role of glucagon for promoting islet size and β-cell function, as well as the roles that VEGFA and HGF may play in this process. Late gestation IUGR fetuses respond to chronically increased amino acid supply with increased fetal insulin secretion. Furthermore, chronically increased amino acid supply to late gestation IUGR fetuses not only increases insulin secretion, but also increases islet size. This has important implications when considering the timing of interventions that might be used to prevent or even reverse the complications of IUGR.

Appendix

See Table 4.

Table 4.

Antibody Table

Peptide/Protein Target	Antigen Sequence (if Known)	Name of Antibody	Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody		Species Raised in; Monoclonal or Polyclonal	Dilution Used
Glucagon		Glucagon	Sigma-Aldrich	G2654	Mouse; monoclonal	1:500
Insulin		Insulin	Dako	A0564	Guinea pig; polyclonal	1:250
Pancreatic polypeptide		Pancreatic polypeptide	Millipore	AB939	Rabbit; polyclonal	1:500
Somatostatin		Somatostatin	Dako	A0566	Rabbit; polyclonal	1:500
Phosphorylated histone 3		pHH3	Upstate (now Millipore)	06-775	Rabbit; polyclonal	7.5 μg/mL
Mouse IgG		Mouse IgG	Jackson ImmunoResearch	715-545-150	Donkey; polyclonal	1:500
Rabbit IgG		Rabbit IgG	Jackson ImmunoResearch	711-585-152	Donkey; polyclonal	1:500
Guinea pig IgG		Guinea pig IgG	Jackson ImmunoResearch	706-155-148	Donkey; polyclonal	1:500
Glucokinase		Glucokinase	Abcam	ab88056	Rabbit; polyclonal	1:1000
GLUT2		GLUT2	Santa Cruz Biotechnology	sc-7580	Goat; polyclonal	1:1000
VEGFA		VEGFA	Santa Cruz Biotechnology	sc-507	Rabbit; polyclonal	1 μg/mL
Actin		Actin	MP Biomedicals	869 133	Mouse; monoclonal	1:20 000