Human fetuses with severe intrauterine growth restriction (IUGR) have less pancreatic endocrine tissue and exhibit β-cell dysfunction, which may limit β-cell function in later life and contribute to their increased incidence of noninsulin-dependent diabetes mellitus. Three factors, replication, apoptosis, and neoformation, contribute to fetal β-cell mass. We studied an ovine model of IUGR to understand whether nutrient deficits lead to decreased rates of fetal pancreatic β-cell replication, increased rates of apoptosis, or lower rates of differentiation. At 90% of term gestation, IUGR fetal and pancreatic weights were 58% and 59% less than pair-fed control, respectively. We identified a selective impairment of β-cell mass compared with other pancreatic cell types in IUGR fetuses. Insulin and insulin mRNA contents were less than other pancreatic endocrine hormones in IUGR fetuses, as were pancreatic insulin positive area (42%) and β-cell mass (76%). Pancreatic β-cell apoptosis was not different between treatments. β-cell capacity for cell cycling, determined by proliferating cell nuclear antigen (PCNA) immunostaining, was not different between treatment groups. However, the percentage of β-cells actually undergoing mitosis was 72% lower in IUGR fetuses. These results indicate that in utero nutrient deficits decrease the population of pancreatic β-cells by lengthening G1, S, and G2 stages of interphase and decreasing mitosis near term. Diminished β-cell mass in IUGR infants at birth, if not adequately compensated for after birth, may contribute to insufficient insulin production in later life and, thus, a predisposition to noninsulin-dependent diabetes.
- fetal endocrine pancreas
- cell cycle
intrauterine growth restriction (IUGR) not only increases the risk of perinatal mortality and morbidity, but also has been linked to diseases of later life (4, 13). A summary of human epidemiological studies has shown that infants with impaired fetal growth are more likely to have glucose intolerance as adults (36). Although studies have not shown a clear relationship between birth weight and insulin secretion (36), infants who were small for gestational age because of IUGR have a greater risk of developing noninsulin-dependent diabetes mellitus in later life (24). This risk indicates specific β-cell deficiencies, because both β-cell dysfunction and insulin resistance occur in noninsulin-dependent diabetic patients (43, 49, 61).
β-cell dysfunction occurs in human IUGR fetuses with hypoglycemia and hypoxia (37), as well as in glucose-deprived fetal sheep (31) and protein-restricted fetal rat islets (14). Nutrient deficits at critical developmental periods might limit endocrine pancreas formation, resulting in abnormal insulin secretion (4). Reduction of pancreatic endocrine tissue was reported for a group of human fetuses with severe IUGR (<1.5 kg at term gestation) (59); however, in a second cohort of neonates with a mean birth weight of 2.3 kg, no change in the β-cell fraction was found (6). In both reports, weights were used to distinguish small-for-gestational age infants as IUGR, but no in utero diagnosis was presented to determine inappropriate fetal growth rates from constitutionally small fetuses.
Previous animal models of IUGR have allowed investigators to evaluate the impact of maternal or fetal malnutrition with respect to severity and type of nutrient restriction on pancreatic endocrine formation and function. For example, rodent models of IUGR have shown functional and structural defects in the fetal pancreas. Rat models of IUGR generated by maternal dietary protein or caloric restriction produce offspring with decreased β-cell mass, which can be augmented by continuing maternal dietary restriction during lactation (20, 21, 27, 53). Neonatal rat pups from mothers on a low-protein diet have decreased β-cell proliferation and increased islet cell apoptosis (42, 53). As fetuses, these pups do not have reduced total essential and nonessential amino acid, glucose, or insulin concentrations (47), although impaired pancreatic endocrine function appears to be dependent on normal taurine concentrations (12). In contrast, pups from malnourished mothers have increased glucocorticoid levels, which have been shown to reduce β-cell differentiation (8, 22, 29). These reports show that maternal nutrient restriction not only limits fetal growth, but also diminishes the amount of fetal, neonatal, and adult pancreatic endocrine tissue by affecting replication, apoptosis, and neoformation via a discrete set of nutrients or hormones (e.g., taurine or corticosterone).
Complications in human IUGR pregnancies include decreased placental mass, rates of oxygen, amino acid, and glucose uptake by the fetus, and umbilical vein blood flow (17, 25, 34, 39–41, 48). Additionally, amino acid, glucose, and insulin concentrations are lower in human IUGR fetuses, likely contributing to the growth restriction (23, 28, 37). Clinical severity of human IUGR fetuses is determined by abnormalities in umbilical artery Doppler velocimetry (40), which is associated with an increased incidence of fetal hypoglycemia and hypoxia (34, 41). Defects of insulin secretion have been found in other mammalian models of IUGR (14, 31). None of these models adequately replicates the complications observed in human IUGR and may only reflect limitations in one or a subset of nutrients. In contrast, our ovine model of placental insufficiency and IUGR, established by exposing pregnant ewes to a warm environment, replicates all of the complications found in human pregnancies with severe fetal growth restriction (2, 5, 19, 45, 45, 46, 50, 56). Therefore, the sheep fetus with placental insufficiency provides a useful mammalian model to evaluate adaptations in pancreatic development and growth to determine compensatory mechanisms to global nutrient deprivation that might lead to metabolic programming. In this study, we have used the placental insufficiency model of fetal growth restriction to determine whether decreased pancreatic endocrine development is limited by increased rates of apoptosis, decreased rates of β-cell replication, or lower rates of β-cell neoformation. We show in this model that β-cell area and mass are reduced to a greater extent than other mature pancreatic endocrine cell types, not by increased rates of apoptosis, but by decreased rates of β-cell mitosis and potentially β-cell differentiation.
MATERIALS AND METHODS
Ovine model of intrauterine growth restriction.
Columbia-Rambouillet crossbred ewes carrying singleton pregnancies were purchased from Nebeker Ranch, Santa Monica, CA, and managed in compliance with the Institutional Animal Care and Use Committee, University of Colorado Health Sciences Center (UCHSC). All animal care and use were conducted at the UCHSC Perinatal Research Center, in Aurora, CO, which is accredited by the National Institutes of Health, the U.S. Department of Agriculture, and the American Association for Accreditation of Laboratory Animal Care. Animal preparations included pair-fed gestational age-matched control fetuses (n = 6) from normal healthy pregnant ewes maintained at 25°C and intrauterine growth-restricted fetuses (n = 6) generated by exposing pregnant ewes to elevated ambient temperatures (40°C for 12 h; 35°C for 12 h) from 37 ± 3 days gestation age (dGA) until 120 ± 3 dGA as previously described (32, 46, 56).
The ewes and fetuses were anesthetized with ketamine (4.4 mg/kg) and diazepam (0.11 mg/kg) between 131 dGA and 138 dGA with term gestation at 147 days. After a hysterectomy, the fetus was removed, blotted dried, and weighed. The ewe was then killed with intravenous concentrated pentobarbital sodium (10 ml, Sleepaway, Fort Dodge Animal Health). The fetal pancreas was dissected free, weighed, and divided from the common bile duct to the anatomic left side of the portal vein (pancreatic notch, when visible); the hepatic portion was frozen in liquid nitrogen, and the splenic portion was fixed in 4% paraformaldehyde overnight and then embedded in paraffin. Additional pancreases from control fetuses at 132 dGA were collected for histological comparison of the hepatic and splenic regions.
Fetal pancreas tissue from the hepatic portion was pulverized in liquid nitrogen. Pancreatic hormones were acid-ethanol extracted from 35 mg of tissue in 1 mL of 1 M HCl, 70% ethanol at −20°C. Three independent pancreas samples per fetus were evaluated for insulin and glucagon content. The concentration of insulin was determined using the ovine insulin ELISA (Alpco, Windham, NH). Glucagon concentrations were measured with a Glucagon RIA kit (Linco Research, St. Charles, MO) that cross-reacts with most mammalian pancreatic glucagon, including sheep, which shares 100% amino acid identity to human glucagon. The detection efficiency of glucagon in sheep plasma was 98.6 ± 2.2%. Data are presented as means ± SE of micrograms of pancreatic hormone per gram of pancreas.
Construction and preparation of ovine cDNAs.
Total RNA from fetal pancreas at 135 dGA was extracted with Tri Reagent (Molecular Research Center, Cincinnati, OH). Polyadenylated RNA was reverse-transcribed with oligo(dT)12–18 and Superscript II reverse transcriptase (Invitrogen Life Technologies, Carlsbad, CA). Insulin, glucagon, somatostatin, and pancreatic polypeptide cDNAs were amplified by a PCR using sense and antisense synthetic oligonucleotides (Invitrogen Life Technologies). Oligonucleotide primer sequences were 5′-TCAGCAAACAGGTCCTCGCAAG-3′ and 5′-GGGCCAGGTCTAGTTACAGTAG-3′ for insulin (U00659), 5′-TCACTCTCTCTTCACCTGCTCTGT-3′ and 5′-GACACACTTACTTCCTGTCAG-3′ for glucagon (AF529185), 5′-TCTCCATCGTCCTGGCTCTTG-3′ and 5′-CTCCAGCCTCATTTCATCCTG-3′ for somatostatin (AF031488), and 5′-TGCTCCTTCTGTCCACGTG-3′ and ACCTGGGGACTGCTGAG-3′ for pancreatic polypeptide (AY427976). Taq polymerase-amplified PCR products were cloned into the pCR II-TOPO plasmid vector; plasmids were transformed into Top 10 F′ competent Escherichia coli. (Invitrogen Life Technologies). Plasmids were purified using QIAprep Spin Miniprep Kit (Qiagen, Valencia, CA) and nucleotide sequenced to confirm identity of the cDNA.
Northern blot hybridization analysis.
Total pancreatic RNA extracted from the hepatic portion was electrophoresed through a 1.5% agarose gel, as previously described (30). Northern blot hybridization was performed with the ovine cDNAs for insulin, glucagon, somatostatin, and pancreatic polypeptide. Purified cDNAs were radiolabeled with [α-32P]-dCTP using a Multi-Prime Kit (Amersham Pharmacia Biotech, NJ), denatured at 95°C for 5 min, and added to ULTRAhyb (Ambion, Austin, TX) at a final concentration of 1 × 106 cpm/ml. After hybridization the membrane was washed twice for 5 min in 2× SSC (in mM) (1× 150 NaCl, 15 Na3Citrate, pH 7.0) with 0.1% SDS at 42°C followed by two stringent washes in 0.1× SSC with 0.1% SDS at 60°C for 20 min each. The membrane was exposed on a phosphorimager screen (Molecular Dynamics, Sunnyvale, CA) and quantified with ImageQuant (Molecular Dynamics). To remove hybridized radiolabeled cDNAs, the membrane was washed three times in 0.5% SDS at 65°C for 30 min. The cDNAs were hybridized a second time with a radiolabel ovine β-actin cDNA (Accession #U39357; corresponding to 634–1135 bp) to normalize expression. Data were expressed as specific hormone mRNA intensity divided by β-actin mRNA intensity.
Histology of fetal pancreatic endocrine cells.
Six tissue sections of 5 μm were cut from a paraffin-embedded control and IUGR pancreases at 100-μm intervals for histological and morphometric evaluation. Pancreatic sections were dewaxed with two washes in xylene (5 min) and hydrated with a series of descending ethanol washes to water. Pancreatic sections were microwaved twice for 5 min at 60% power in 10 mM citric acid buffer, pH 6.0, cooled for 20 min, and washed three times in PBS for 10 min. Pancreatic sections were blocked with 1.5% normal donkey serum in PBS for 30 min. Mature pancreatic endocrine hormones were identified in the fetal sheep pancreas with guinea pig anti-porcine insulin (Dako, Carpinteria CA, 1:500) or mouse anti-human insulin (Abcam, Cambridge UK, 1:1,000), mouse anti-porcine glucagon (Sigma-Aldrich, St. Louis, MO, 1:500), rabbit anti-human somatostatin (Dako, 1:500), and rabbit anti-human pancreatic polypeptide (Dako, 1:500). Dual immunostaining with the two insulin antisera showed an identical immunostaining pattern, confirming their ability to determine β-cells in sheep pancreas as reported for the mouse monoclonal anti-human insulin (Abcam), which was only used for determining β-cell size. Primary antisera were diluted in blocking buffer and incubated at 4°C overnight; negative controls were included for which the primary antiserum was omitted. After this incubation, the pancreatic sections were washed three times for 10 min with PBS; immunocomplexes were detected with affinity-purified secondary antiserum conjugated to Cy2, Rhodamine Red, Texas Red, or 7-amino-4-methylcoumarin-3-acetic acid (AMCA; Jackson ImmunoResearch Laboratories, West Grove, PA) or AlexaFluor 488 (Molecular Probes, Eugene OR) diluted 1:500 in block buffer for 60 min at 22°C. The pancreatic sections were washed three times for 10 min each with PBS and mounted in 50% glycerol and 10 mM Tris-HCl, pH 8.
β-cell apoptosis (programmed cell death).
Pancreatic sections were dewaxed and rehydrated. Terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick translation end labeling (TUNEL) was performed with the In Situ Cell Death Detection, POD Kit (Roche Molecular Biochemicals, Mannhiem, Germany). Pretreatment of the paraffin-embedded pancreatic sections included a 10-min incubation in proteinase K (50 μg/ml) in 10 mM Tris, pH 8.0 at 22°C, three PBS washes (5 min), and a 2-min incubation in 0.1% Triton X100 0.1% sodium citrate on ice. After three PBS washes (5 min), the pancreatic sections were incubated with 50 μl of TUNEL reaction mixture or 50 μl of Label Solution (negative control with no terminal deoxynucleotidyl transferase) for 45 min at 37°C. Pancreatic sections were blocked with 1% BSA-PBS for 30 min; guinea pig anti-porcine insulin (1:500) in 1% BSA-PBS was added, and the sections were incubated at 4°C overnight. Pancreatic sections then were washed in PBS (10 min) three times, and immunocomplexes were detected with affinity-purified secondary antiserum conjugated to Texas Red (1:500, Jackson ImmunoResearch Laboratories, West Grove, PA) in block buffer for 60 min. Pancreatic sections were washed three times with PBS (10 min) and mounted in VECTASHIELD mounting medium with DAPI (4′,6 diamidino-2-phenylindole; Vector Laboratories, Burlingame, CA) to identify cell nuclei.
Pancreatic endocrine cell proliferation.
Fetal pancreatic sections were dewaxed and hydrated. Antigen retrieval was accomplished with a microwave treatment in 10 mM citric acid buffer, pH 6.0, for the rabbit polyclonal anti-human proliferating cell nuclear antigen (PCNA; Santa Cruz Biotechnology, 4 μg/ml). Antigen retrieval for rabbit polyclonal anti-phospho-Histone H3 (pHH3; Upstate, Lake Placid, NY, 7.5 μg/ml) was accomplished with an incubation in 0.2% Triton X100 PBS for 15 min, two 10 min washes with water, a 10-min Proteinase K digestion (50 μg/ml in 10 mM Tris, pH 8.0), two 10-min water washes, and a citric acid microwave treatment. Fetal pancreatic sections were washed 3 times for 10 min in PBS, and nonspecific binding blocked with 1.5% normal donkey serum in PBS (anti-PCNA) or 1% bovine serum albumin in PBS (anti-pHH3) for 1 h. Primary antisera cocktails, including guinea pig anti-insulin, mouse anti-glucagon, and anti-PCNA or guinea pig anti-insulin and anti-pHH3, were diluted in blocking buffer and incubated with pancreatic sections at 4°C overnight. After three 10-min PBS washes, immunocomplexes were detected with affinity-purified secondary antiserum conjugated to AlexaFluor 488 anti-rabbit IgG, Rhodamine Red anti-mouse IgG, Texas Red anti-guinea pig IgG, or AMCA anti-guinea pig diluted 1:500 in block buffer for 60 min at 22°C. Pancreatic sections were washed three times with PBS and mounted in 50% glycerol and 10 mM Tris-HCl, pH 8 or VECTASHIELD mounting medium with DAPI.
Fluorescent images were visualized on an Olympus BX51 systems microscope and digitally captured with a Pixera 600CL camera. Morphometric analysis was performed with Image Pro 4.5 software (Media Cybernetics, Silver Spring, MD). Hepatic and splenic pancreas portions were evaluated at every 200–250 μm (n ≥ 9 sections/portion) for insulin, glucagon, and the combination of somatostatin and pancreatic polypeptide. Data are expressed as a percentage of total pancreas area. For control and IUGR pancreases, insulin+ and glucagon+ areas were determined for >20 fields of view (FOV = 0.31 mm2) on six or more pancreatic sections per animal separated by ≥100-μm intervals and expressed as a percentage of total pancreas area. β-cell (insulin+ cells), and α-cell (glucagon+ cells) mass was determined by multiplying the pancreas weight by the percent-positive area. Fetal pancreatic islet size and endocrine cellular content were examined by triple immunofluorescence detection with insulin, glucagon, and somatostatin in control and IUGR fetuses. Fetal islets were distinguished as endocrine cell clusters greater than 500 μm2 containing at least two endocrine cell types. The percent of fluorescent islet area for each of the mature endocrine hormones, insulin, glucagon, or somatostatin, is presented relative to the total fluorescent islet area. Islet endocrine areas were calculated by multiplying the percent area by total islet area. The percentage of apoptotic β-cells (TUNEL+/insulin+) was determined by evaluating >3,000 nuclei (DAPI+) of insulin+ cells within each fetal pancreas on ≥3 pancreatic sections per fetus separated by 100 μm. Insulin+ cells (>1,400 per fetus) and glucagon+ cells (>800 per fetus) were classified PCNA positive or negative for ≥10 FOV on ≥3 pancreatic sections separated by 100 μm. β-cells undergoing mitosis (pHH3+) were determined for ≥20 FOV per pancreatic section (n = 3), which represented between 4,000 and 8,000 insulin+ cells evaluated per fetal pancreas on ≥3 pancreatic sections.
Treatment means for all experiments were analyzed by ANOVA using general linear model procedure (ProcGLM; Ref. 51). Differences between treatment means were determined with a Fisher's protected least significant difference test (51), with a significance level at P values ≤ 0.05.
Fetal and pancreatic characteristics of control and IUGR animals.
At the same gestational age at autopsy (Table 1), mean fetal weight in the IUGR group was 58% (P < 0.01) less than the mean fetal weight in the control group (Table 1). The percentage of males was 67% for both the control and IUGR treatment groups. The mean pancreas weight was 59% lower (P < 0.01) in the IUGR group than the control group (Table 1). Pancreas weight among all fetuses in both groups correlated significantly with fetal weight (r2=0.86). The percentage of pancreas weight (g) per body weight (g) was not different between control (0.11 ± 0.01%) and IUGR (0.11 ± 0.01%) fetuses. Thus pancreas weight was reduced in proportion to fetal weight in IUGR fetuses.
Pancreatic insulin content per pancreas weight was 47% lower in IUGR than control pancreases (Table 1). The IUGR group had 76% less total insulin content than the control group (Table 1). Pancreatic glucagon content was not different between the control and IUGR groups (Table 1), indicating a β-cell-specific defect in the IUGR fetuses.
Ovine β-actin mRNA concentration was not different between treatment conditions; mean integrated pixel intensity per 20 μg total pancreas RNA was 1.01×106 ± 0.09×106 in control fetuses and 1.14×106 ± 0.10×106 in IUGR fetuses. Insulin mRNA relative to β-actin mRNA was significantly lower (P < 0.01) in the IUGR pancreas and was 66% less than control pancreas (Fig. 1, Table 2). Glucagon, somatostatin, and pancreatic polypeptide mRNA normalized by β-actin mRNA were not different from control levels (Table 2).
Endocrine cells in the ovine fetal pancreases.
All four endocrine cell types (insulin, glucagon, somatostatin, and pancreatic polypeptide) were present in control and IUGR fetal pancreases at 90% of term gestation (Fig. 2,A–D). The majority of cells expressing mature endocrine hormones were located in islets or endocrine cell clusters, but single cells and small endocrine clusters (2–4 cells) were observed for each endocrine cell type (Fig. 2, A, B, D). Large endocrine cell clusters, consisting primarily of insulin-positive cells, were centrally located, surrounded by interlobular connective tissue, and, in some cases, displayed hemorrhages or sinusoids filled with red blood cells (Fig. 2, E, I). Large islets were found in all fetal pancreases from control or IUGR treatment groups. However, the majority of endocrine clusters were on average 2,365 ± 178 μm2 in control fetuses, distributed throughout the parenchyma, and composed of two or more endocrine cell types, with insulin-positive cells (β-cells) in the center surrounded by glucagon (α-cells), somatostatin (δ-cell), or pancreatic polypeptide (PP cells)-positive cells (Fig. 2, A–D).
Insulin, glucagon, and the combination of somatostatin and pancreatic polypeptide fractions of pancreas area were determined for the hepatic and splenic portions of a single fetal sheep pancreas (132 dGA). We identified a slight but significant difference (P < 0.05) for the insulin-positive area between the hepatic and splenic portions, 3.7% and 4.1%, respectively. No difference was detected for glucagon-positive areas between the two portions (1.1% vs. 1.2%). Immunopositive area for the combination of somatostatin and pancreatic polypeptide was greater (P < 0.05) in the hepatic portion (1.3%) than the splenic portion (0.8%). δ-cells and PP cells were evenly distributed throughout the splenic portion with β- and α-cells (Fig. 2, A–D). In the hepatic portion, some interlobular areas contained β-, δ-, and PP cells with no α-cells, and other interlobular areas had an intermingling of all cell types. Total endocrine area between the hepatic and splenic pancreatic portions was not different, at 6.1%. No trend or change in endocrine areas was observed in either portion of the pancreas from the consecutive tissue sections evaluated throughout the organ, thereby allowing random sections within a defined portion of the fetal sheep pancreas to provide an adequate representation of the endocrine mass for that pancreatic portion. Identical results for β-cell, α-cell, and the combination of δ- and PP cell areas between the hepatic and splenic portions were obtained from a second fetal sheep pancreas collected at 132 dGA.
β-cell mass in control and IUGR fetal pancreases and islets.
The mean pancreatic insulin+ area was 42% lower (P < 0.001) in the IUGR group compared with the control group (Table 3). The estimated β-cell mass for the total fetal pancreas was 76% lower in IUGR fetuses (Table 3). β-cell mass was reduced to a greater extent than pancreas mass and was the same as the reduction in total pancreatic insulin estimated from the hepatic portion, thereby partially explaining the lower insulin content and insulin mRNA concentrations.
There was no difference in glucagon+ area in the splenic region of the pancreas, similar to the glucagon mRNA and content data measured on the hepatic portion of the pancreas; thus the lower α-cell mass in the IUGR group (Table 3) was due to the decrease in IUGR pancreas weight (Table 1).
Reductions in β-cell area and mass can be attributed to declines in two β-cell compartments, intra- and extra-islet β-cells. In control sheep, fetuses 90.0 ± 1.0% of the insulin-positive cells were located in endocrine clusters of more than 4 cells with sections immunostained for insulin, glucagon, and somatostatin (Fig. 2, A and B). The percentage of β-cells within islet clusters was slightly but significantly greater (P = 0.05) for the IUGR fetuses at 92.3 ± 0.3%.
IUGR fetal islets were significantly smaller than control fetal islets (P < 0.05; Table 4). The density of islet or endocrine clusters (>4 cells) within the fetal pancreas was not different between control (37.3 ± 0.4 islet clusters/mm2) and IUGR fetuses (39.3 ± 1.6 islet clusters/mm2). The insulin+ area within the IUGR fetal islets was lower (P < 0.05) than the controls. The relative areas for glucagon and somatostatin were increased as a result of the decreased insulin+ area. However, α-cell (glucagon+) or δ-cell (somatostatin+) islet area was not different between treatment groups. Islet insulin+, β-cell area was 52% lower in IUGR fetal islets, 994 ± 111 μm2, compared with control islets, 1,907 ± 158 μm2.
β-cell apoptotic and mitotic rates.
Mean β-cell size was determined by evaluating nuclear density of insulin+ area, similar to insulin immunostaining with DAPI (Fig. 2), in greater than 400 β-cells per fetal pancreas. The average β-cell area for control fetuses was 105.7 ± 5.0 μm2, not different from the IUGR fetal β-cell size of 115.3 ± 6.8 μm2, indicating that the lower β-cell area and mass were primarily due to a lower number of β-cells.
Rates of β-cell apoptosis and replication were examined between control and IUGR fetuses to determine the effects of nutrient deprivation on β-cell area and mass. β-cell apoptosis, measured by TUNEL histochemistry and insulin immunostaining, was not different (P = 0.88) between control fetuses (1.96 ± 0.46% TUNEL+/insulin+ cells) and IUGR fetuses (1.87 ± 0.42% TUNEL+/insulin+ cells; Fig. 2, E and F, and Fig. 3A). Localized areas displayed greater numbers of TUNEL+ cells in control and IUGR fetal pancreases, indicating regions undergoing tissue remodeling.
β-cell replication was examined with PCNA, a marker of cells in G1, S, and G2 phases (10, 33, 63), and phosphorylated-Histone H3 (pHH3), a marker of mitosis (26, 60). The mean percent of PCNA+ β-cells in control fetal pancreases (n = 6) was 10.2 ± 0.8%, not significantly different (P = 0.25) from the mean percentage of PCNA+ β-cells in IUGR pancreases, 12.0 ± 1.1% (n = 6, Fig. 2, G and H, and Fig. 3B). However, the percentage of PCNA+-α-cells, those expressing glucagon, was significantly greater (P < 0.05) in IUGR than controls, 13.5 ± 1.7% and 8.5 ± 0.5%, respectively. Because PCNA expression distinguishes cells in progress through the cell cycle, we also evaluated a more discrete period of the cell cycle, mitosis, with pHH3 (Fig. 2, I–K). The percent of immunopositive pHH3-β-cells was much less relative to that of PCNA+-β-cells, consistent with their time spent in M phase when histone H3 is phosphorylated. Strikingly, the percentage of pHH3+ β-cells was significantly less (P < 0.001) in IUGR fetuses (0.15 ± 0.02%) compared with control fetuses (0.54 ± 0.04%; Fig. 3C), showing that β-cell mitosis was significantly decreased in IUGR fetuses.
In this study, we evaluated the fetal endocrine pancreas in an ovine model of placental insufficiency-induced IUGR. This model shares essentially all of the distinguishing growth and physiological and biochemical characteristics found in human IUGR fetuses (3). Pancreas weight was reduced in the IUGR fetuses and directly associated with the reduction in fetal weight. β-cell mass and pancreatic insulin content in the IUGR pancreases were reduced by 76% on average and to a greater extent than the weight of the pancreas or the mass of the other endocrine cell types. There also was less insulin mRNA content in the IUGR pancreases. The reduction in β-cell mass in the IUGR pancreas was due to decreased numbers of β-cells, because cell size was unaltered. Fewer β-cells in the IUGR pancreases was due to a decreased rate of β-cell mitosis, not to an increased rate of programmed cell death. In addition, there were fewer extra-islet β-cells, which might represent decreased rates of β-cell differentiation, but this will need to be confirmed. Therefore, fetal nutrient deprivation produced by placental insufficiency appears to reduce fetal β-cell replication rates to a greater extent than other pancreatic cell types. These observations indicate that long-term nutrient restriction in the fetus disproportionately reduces β-cell number and insulin transcription, setting the stage for potential pancreatic β-cell/insulin insufficiency and diabetes later in life.
In our histological evaluation of fetal sheep endocrine cells, we did observe single β-cells and small β-cell clusters of ≤4 cells in the pancreas parenchyma (Fig. 2, A–D). Quantification of these extra-islet β-cells showed that they account for ∼10% of the population in the control pancreas and for only 8% of the IUGR β-cell population, a significant reduction. These cells may represent new endocrine cell formation or even small islets (9, 44, 57), but it has been shown that the percentage of individual endocrine cells declines after midgestation in the sheep as they form small and large endocrine clusters (44, 62). Although reduction in this nonislet β-cell population contributes to the overall reduction in whole pancreatic β-cell mass, we show that a similar reduction of 52% occurs in the β-cell fraction localized to the small islets in the IUGR pancreas. On postnatal day 1, the endocrine portion is 7–8% of the newborn lamb pancreas with ∼50% of the endocrine tissue localized in large islets (57). Our data show that the endocrine area of normal fetal sheep at 90% of gestation is 5–7% of the total pancreatic area (Table 3). The 58% reduction in whole pancreatic β-cell mass in the IUGR fetuses was similar to the 52% reduction in islet β-cell mass and 47% reduction of single β-cells, indicating that near term, the decreases in number of β-cells in the IUGR fetuses occurs equally in all endocrine compartments.
The pancreatic endocrine compartment and the percent of β-cells per islet also are reduced by about half in human fetuses with severe IUGR (59), similar to the present data in our ovine model of placental insufficiency induced IUGR (Table 4). Reductions in β-cell mass also have been observed in three rat models with fetal growth restriction, maternal low protein diet (14, 42, 53), maternal caloric restriction (20, 21), and uterine artery ligation (15). Interestingly, the effect on the β-cell population in fetal rat pups appears to be dependent on the type of treatment or condition producing IUGR. For example, fetuses or newborn pups from pregnant rat dams fed a low-protein diet during pregnancy have decreased numbers of islet cells in S-phase (42, 53) and increased islet cells undergoing apoptosis (42). In contrast, newborn rat pups from mothers fed a 50% calorie-restricted diet between embryonic day 15 and 21 do not have reduced numbers of β-cells in S-phase; instead, these pups have β-cell hypotrophy, decreased β-cell aggregate density, and impaired β-cell development (20). Apoptosis has not been determined in this model.
We observed a 58% reduction in total pancreas β-cell area in the IUGR pancreases. This reduction was due to decreased number of β-cells and was equally distributed across all endocrine compartments, including small and large islets. Our data indicate that apoptosis was not involved in decreasing the β-cells mass (Fig. 3A). Instead, we found that the portion of β-cells positive for PCNA did not differ between control and IUGR pancreases (Fig. 3B), whereas the percentage of pHH3-positive β-cells was significantly reduced (Fig. 3C), indicating that the rate of mitosis was lower and the length of interphase, specifically G1, S, and G2, increased. Similar results were observed in fetuses from rat dams fed a low-protein diet where a higher percentage of islet cells or β-cells were PCNA positive, but the BrdU-labeling index was reduced (42). The discrepancies between the PCNA and BrdU-labeling indices were found to result in a longer G1 phase of the cell cycle (42). Furthermore, PCNA staining in the α-cells was increased 1.6 fold, suggesting that other endocrine cell types have a longer cell cycle during nutrient deficiencies, but their cell mass relative to the whole pancreas mass remains proportional. Therefore, for other pancreatic cell types, nutrient deficiencies reduce pancreas or fetal mass by slowing the cell cycle with respect to nutrient availability.
The specific reduction in β-cell proliferation in contrast to other pancreatic cells, including the other islet cell types, is puzzling. It is tempting to speculate that β-cell mass adjustment mechanisms (as yet uncharacterized) are functional in the embryonic β-cells. If so, the β-cell, in contrast to other embryonic cells, may be capable of specifically sensing the metabolic load, and adjusting its proliferation rate accordingly, thereby modulating the fetal anabolic hormone, insulin, to coordinate fetal growth rate with nutrient availability. This behavior is well established as a mechanism for β-cell mass adjustments during pregnancy, in rat models of insulinoma-bearing animals and quite possibly during the β-cell mass adaptation that develops in response to obesity and insulin resistance (7, 11, 54, 58). Notwithstanding, our observations of a selective impairment of β-cell replication during IUGR is important, as it places this particular cell type in focus as a vulnerable point during fetal growth restriction that might manifest into an adult disease, diabetes.
The relatively lower insulin mRNA content (to 34% of controls; Fig. 1) relative to insulin protein content (to 53% of controls; Table 1) in the IUGR pancreas was somewhat surprising. Usually, there is a parallel relationship between transcription of the insulin gene and insulin mRNA translation (16, 35, 38). The disproportionately low-insulin mRNA in the IUGR β-cells may indicate an underlying reduction in transcription of the insulin gene. In this regard, Pdx-1 has been identified as a glucose-responsive transacting factor for the insulin gene (35), as well as a key factor in β-cell development and function (1, 55). Ovine Pdx-1 mRNA in IUGR pancreas samples, measured by quantitative PCR, is reduced (Limesand and Hay, unpublished observation). Insulin content also might be better preserved in the IUGR pancreas as a result of slower turnover of the granule pool, that is, decreased glucose-stimulated insulin release from the β-cells because of the lower plasma glucose concentrations. Both of these mechanisms might be acting together, that is, decreased insulin secretion because of lower glucose concentrations and decreased transcription of the insulin message despite persistent (or even possibly enhanced) insulin mRNA translation to insulin protein to increase the difference between insulin mRNA and protein content in IUGR fetuses.
Human epidemiological studies in most cases have identified an inverse relationship between birth weight and the incidence of noninsulin-dependent diabetes in later life (36). In part, this might be the result of excessive demands for insulin production in later life on the smaller starting population of β-cells in IUGR offspring. Also, it is interesting to consider that β-cell dysfunction might contribute to the deficiency in insulin secretory response in IUGR offspring. In rat models of fetal growth restriction, deficits in insulin secretion capacity, as well as decreased β-cell mass, have been observed in the adult animal (14, 52), although such long-term consequences also were dependent on early postnatal nutritional experiences (18). Similar to the rat models, chronically hypoglycemic fetal sheep (31) and the ovine model of maternal hyperthermia-induced placental insufficiency used in the present study (Limesand and Hay, unpublished observation) both have decreased insulin secretory response to glucose, similar to that observed in human IUGR fetuses (37). And at least in the hypoglycemic model, the in utero glucose deficiency appears to continue to limit β-cell function even after glycemic correction, as shown by our observations that even after 5 days of glycemic recovery, the rate of increase in insulin secretion in response to glucose stimulation remains blunted (31). Therefore, it is plausible that both β-cell number and function are reduced in utero in response to nutritional deficits. Although some aspects of endocrine function might return shortly after birth in response to enhanced nutrient supply, over longer periods β-cell function might deteriorate again in response to hyperglycemic stress.
In conclusion, in an ovine model of placental insufficiency-induced fetal nutrient restriction and IUGR we have shown that the fetal β-cell mass is reduced to a greater extent than other pancreas cell types. The reduction of the fetal pancreatic β-cell population is at least partially due to a decrease in β-cell proliferation and not from increased apoptosis. Reduction in β-cell mass and insulin availability during fetal growth, therefore, might define a mechanism by which the fetus adapts its somatic growth rate to nutrient availability. If this pancreatic endocrine adaptation that appears necessary to preserve fetal life is not corrected postnatally, it might contribute to inappropriate β-cell mass and inadequate insulin secretion in response to glucose stimulation and the development of noninsulin-dependent diabetes in later life.
This work was supported by grants from the National Institutes of Health (HD-42815, W. W. Hay, Jr., principal investigator) and The Children's Hospital Research Institute, in Denver, CO, (W. W. Hay, Jr., principal investigator). Sean W. Limesand was supported by National Institutes of Health National Research Service Award DK-60300 and DK-067393 (S. W. Limesand, principal investigator). Fetal pancreatic tissues also were collected from animals provided by National Institutes of Health grant HD-20761 (W.W. Hay, Jr., principal investigator). Molecular Biology Core Services were provided by the UCHSC Barbara Davis Center for Childhood Diabetes, which is supported by National Institutes of Health Diabetes and Endocrinology Research Center grant P30-DK-57516.
We thank Dr. T. R. H. Regnault, Director of the Environmental Core at the Perinatal Research Center, for providing fetal pancreases from control and IUGR fetuses.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society