Main

Inadequate functional pancreatic β cell mass is now recognized as a crucial component in the pathophysiology of type 2 diabetes. Cell lineage tracing and the generation of genetically modified mice have shown the requirement of a large number of transcription factors controlling the development of the endocrine pancreas (1,2) even though their chronology of action is not fully understood yet. The transcription factors Pdx-1/IPF1 (insulin promoting factor 1) and P48/PTF1A (pancreas transcription factor 1A) play determinant roles on pancreas development in mice and are expressed early in pancreatic precursor cells. Later, Pdx-1/IPF1 is mostly restricted to mature β cells and P48/PTF1A is a prerequisite to drive exocrine cell differentiation. The inactivation of Pdx-1/IPF1 or P48/PTF1A leads to pancreatic agenesis in mice (35) and in humans, mutations in IPF1 or PTF1A are also associated with pancreatic agenesis (68), MODY or type 2 diabetes (911) or neonatal diabetes mellitus (12). Studies on human pancreas development are scarce and little is known about the early expression of these markers in the human fetal pancreas.

There is growing evidence that glucocorticoid (GC) signaling may play a major role in pancreas development and thereby be involved in abnormalities of glucose homeostasis later in life. Hence, prenatal GC exposure has been proposed as one mechanism to explain the link between intra-uterine growth retardation and the increased risk of glucose intolerance, type 2 diabetes and other cardiovascular risks in adults (1316). In humans, Reinisch et al. (17) have described that prednisone treatment for infertility and subsequent pregnancy maintenance results in a significant decrease in the birth weight of full-term babies. In rodents, fetal overexposure to GC either by maternal dexamethasone treatment, low protein diet or food restriction leads to low birth weight, decrease in β cell mass and later impairment of glucose tolerance and/or hypertension (1820). In this situation Pdx-1/IPF1 mRNA levels were decreased (21,22) and P48/PTF1A mRNA levels increased (21). Additionally, inactivation of the glucocorticoid receptor gene in pancreatic precursor cells suppressing all GC effects in the pancreas is associated with increased β-cell mass (21).

The key role of GC on pancreas development in rodents by the modulation of specific transcription factors as well as the known effect of GC to favor organ maturation (23) prompted us to study if these hormones could play a role in human pancreas development. This question is of clinical importance in the context of antenatal GC therapy used for prenatal lung maturation or treatment of prenatal congenital adrenal hyperplasia (24,25).

We hypothesized that GC would be acting on pancreas development at a particular time window when glucocorticoid receptor (GR) is highly expressed, a period likely to be more sensitive to excess GC. In humans, GC bind to the active GRα isoform on target cells by triggering phosphorylation, dimerization, and translocation of GR into the nucleus. The GRα binds to specific response elements of DNA thereby enhancing or repressing the transcription of genes responsible for the hormonal effect (26). A second isoform, GRβ, truncated in the C-terminal hormone-binding domain is unable to bind the hormone (27).

The present work characterizes in the developing human pancreas the GC-sensitive cells expressing GR together with those expressing the transcription factors IPF1 and PTF1A and/or the pancreatic hormones, using immunohistochemistry on human pancreatic sections from very early (6 wk) to late (23 wk) stages of development. The mRNA expression profiles of the GR isoforms and various pancreatic transcription factors and hormones were studied by RT-PCR at the same developmental stages.

MATERIALS AND METHODS

Tissues.

The collection and use of human embryonic and fetal material were carried out following the French bioethical law and recommendation, informed consent were obtained and the study was approved by the Institut National de la Santé et de la Recherche Médicale (INSERM). Early embryological stage human embryos were collected following voluntary surgical termination of pregnancy by aspiration. Second trimester pancreata were obtained following infant death in utero after sudden medical problem of the mother. Known fetal pathologies, such as chromosomal aberrations, malformations, or genetic disease that could alter pancreas morphology were excluded. Estimation of fetal age (as weeks postconception or weeks of development) was given by echographic occipito-parietal measurements and hand and foot length measurement by the fetopathologist.

Immunohistochemical studies.

Human fetal pancreatic specimen were promptly fixed in 4% formaldehyde, embedded in paraffin, and sectioned at 5 μm thickness. After analysis on hematoxylin and eosin-stained sections, 14 pancreatic specimens showed sufficient tissue quality to be included in this study. Antigen retrieval was performed by microwave treatment in citrate solution (Biogenex). Nonspecific sites were blocked in 1× Tris Buffered Saline (TBS) containing 0.1%/Tween and 3%BSA after a pretreatment with 0.3% Triton before overnight incubation at 4°C with primary antibodies. They were: rabbit anti-IPF1 (kind gift from R. Scharfmann), rabbit anti-human GR polyclonal antibody (ABR, PA1-511, hGR 346-367), rabbit anti-human GRα (PA1-516, ABR hGRα 755-771), mouse anti-insulin (Sigma Chemical Co., St. Louis, MO), mouse anti-glucagon (Sigma Chemical Co. St. Louis, MO). The rabbit antiserum against PTF1A was raised by us against a synthetic amino acid peptide (C-KSFDNIENEPPFEFVS) corresponding to the carboxyl-terminal 16 amino acids of mouse and rat PTF1A. Secondary antibodies were biotin-conjugated anti-rabbit (Jackson Immuno Research Laboratories, West Grove, PA) and peroxidase-conjugated anti-mouse (Amersham Biosciences). DAB (3,3-diaminobenzidine tetrahydrochloride, Dako) and Fast Red (Dako) were used as substrates for peroxidase and alkaline phosphatase, respectively.

The specificity of the anti-GR antibody was assessed by clearing the signal upon incubation with the synthetic peptide, the absence of signal in GRnull/null mouse embryonic pancreas tissues and a positive signal in GR+/+ mouse pancreas (data not shown).

RNA extraction and cDNA synthesis.

Quickly dissected human fetal pancreatic specimens were frozen in liquid nitrogen and stored at −80°C. RNA was prepared with the RNeasy minikit (Qiagen) with on-column DNase treatment. Total RNA was reverse transcribed into cDNA using oligo(dT) and SuperScript II RNase H reverse transcriptase (In Vitrogen) following manufacturer's instruction. The contamination of pancreatic tissues by duodenum, liver, or spleen was excluded by the absence of Sonic Hedgehog amplification, which was detected in liver and lung.

RT-PCR.

M-fold software (http://www.bioinfo.rpi.edu/applications/mfold/old/dna/form1.ogi) allowed the analysis of the secondary structures of published sequence of cDNA encoding for human GRα, GRβ, IPF1, insulin, and NEUROG3. The fragments of sequences lacking any stable secondary structure at 58°C were imported into Oligo6 software (Molecular Biology Insights, Cascade, CO) to design highly stringent primer sets (Table 1). The primers were blasted in cDNA database to ensure their specificity. The nature of the amplified DNA for GRα and GRβ primer products was ultimately confirmed by sequencing. Human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control. Primers were intron spanning with the exception of IPF1.

Table 1 Primers used for regular PCR (Sonic Hedgehog) and RT-PCR (all others)

cDNAs were amplified using SYBR® Greensupermix from Bio-Rad (Hercules, CA) and 300 nM of each specific primer for 45 cycles. Amplification specificity was visualized by melting curve analyses. Each sample was assessed in duplicate. Expression levels were calculated as the ratio between the gene of interest and GAPDH (internal control).

RESULTS

Development of pancreatic structures in the human fetus.

At 6 wd, the human pancreatic epithelium was composed of tubular structures surrounded by a dense mesenchyme next to the duodenal structure. At this stage, the transcription factor IPF1 was present in duodenal structures and the majority of pancreatic epithelial cell nuclei (Fig. 1, 6 wd). At 9 wd, epithelial cell tubules were branched, few isolated insulin cells were observed within the epithelium, and small insulin cell clusters start to form (Fig. 1, 9 wd). The mesenchymal tissue was abundant. The number of IPF1 positive cells increased with further branching between 9 and 13 wd (Fig. 1). At 13 wd, insulin cell clusters were larger and next to the epithelial structures. At this stage, trefoiled epithelial cell clusters not belonging to islets but also expressing IPF1 became visible at distance of the duct-like structures (Figs. 1 and 4C–D). At 22–23 wd, mesenchymal tissue had decreased while epithelial cell clusters were well distinguished from islet cells. First typical acinar structures were recognized. At that stage and adulthood, IPF1 was exclusively expressed in β cells containing insulin and few ductal cells (Fig. 2) and not detected anymore in the exocrine tissue when acinar formation was completed. At adulthood, mesenchymal tissue had mostly disappeared and the islets were surrounded by abundant acinar structures and ductal cells (Fig. 2). As expected for a transcription factor, the IPF1 signal in our study was nuclear at all stages examined, in contrast to nuclear and cytoplasmatic IPF1 staining observed in another study (28).

Figure 1
figure 1

Ontogenesis of IPF1 and GR in the human pancreasTissue organization of normal human pancreas at different stages of development was visualized on paraffin sections by hematoxy-eosin staining (left column, scale bars 50 μmm). Dual immunohistochemistry shows the timing of appearance of IPF1 (brown, middle column) and GR (brown, right column), in relationship with that of insulin in red (middle and right columns); scale bars 25 mμmm, except GR/Insulin at 6wd, 20 μmm. At 6 wk of development Pancreas transcription (wd) IPF1 is detected in all epithelial cells (e) whereas GR is only expressed in mesenchymal cells (m). IPF1 expression increases with branching of the pancreatic epithelium at 9–13 wd while nuclear GR expression is now observed in clusters of insulin-containing cells. d, duct; du, duodenum; e, epithelial cell; m, mesenchymal cell; i, islet.

Figure 4
figure 2

Exocrine pancreas development. At 8 wd, PTF1A (brown, B) is expressed in a large subset of IPF1-positive epithelial progenitor cells (brown, A). At 13 wd, IPF1 (brown, C) is expressed in insulin cells forming islets (red, ic), as well as in trefoiled (see Fig. 1) or tubule-like epithelial cells (diamonds) also expressing the exocrine transcription factor PTF1A (D, asterisks), likely representing the future acini. A–B and C–D are consecutive sections, respectively. PTF1A-positive structures shown by asterisks are also observed at 16 wd (E) and 22 wd (F). Scale bars 25 μm.

Figure 2
figure 3

IPF1 and GR expression at later stages. At 22–23 wd, IPF1 and GR are detected in β cells and few ductal cells (A–D) and GR is also expressed in forming acini (arrows, B and D). At adult age, IPF1 and GR are exclusively expressed in β cells (E–H). a, acini; d, duct; m, mesenchymal cell; i, islet; ic, insulin cells grouped in clusters. Scale bars 25 m μmm (A, B), 20 μmm (C, D, G, H) or 50 m μmm (E, F).

Ontogenesis of the GR in the developing human pancreas during islet differentiation.

At 6 wd, the first GR-positive cells were the mesenchymal cells and the first IPF1-positive epithelial cells were GR-negative (Fig. 1). The first insulin cells at 9 wd, which were isolated and located within the wall of the epithelial tubules (Fig. 3, ii) were GR-and-IPF1-negative (Fig. 3). During islet formation, insulin-expressing cells appear as clusters (Fig. 3, ic), which were first attached to the duct-like structures at 10 wd and located at distance of the ducts at 13 wd (Fig. 4). Insulin cells grouped in clusters expressed both GR and IPF1 (Fig. 3, ic). The colocalization of IPF1 and GR in human epithelial cells thus coincided with the onset of islet formation. In formed islets at 22–23 wd, β cells continued to express GR and IPF1 and were the only pancreatic cells to do so until adulthood (Fig. 2). Glucagon cells did not show any GR immunostaining at early (Fig. 3) or later stages (not shown).

Figure 3
figure 4

Expression of IPF1 and GR during human islet formation. At 9 and 10 wd, IPF1 (brown, left column) is expressed in epithelial cells but not in isolated insulin cells (red, ii). From 10 wd, IPF1 is expressed in insulin cells grouped into clusters (ic) and so does GR (brown, middle column). Note that glucagon cells (red, right column) express neither IPF1 nor GR. g, glucagon cell; e, epithelial cell; m, mesenchymal cell; i, islet; ii, isolated insulin cell; ic, insulin cells grouped in clusters. Scale bars 20 μmm.

Tracing experiments in mice have shown that PTF1A is necessary for the maintenance of pancreatic progenitors and later for the differentiation of exocrine cells (3,4). In the human fetal pancreas, PTF1A was expressed as early as 8 wd in a large subset of IPF1-positive epithelial cells (Fig. 4A and B), indicating that PTF1A is also a marker of human pancreatic progenitors. At 12–13 wd, trefoiled IPF1-positive-and-insulin-negative epithelial cell clusters located at distance of the islets also expressed PTF1A in their nucleus (Fig. 4C and D), likely representing the future acinar cell population. A cytoplasmic signal for carboxypeptidase A was observed in these cells (results not shown). At 16 and 22–23 wd, PTF1A positive cells formed more typical acinar structures (Fig. 4) that expressed GR (Fig. 3, 22–23 wd) but mature acini did not (Fig. 3, adult).

mRNA expression profiles during human pancreas development.

The profile of mRNA expression of the active GRα and the inactive (or suppressor) GRβ forms were analyzed by RT-PCR and compared with that of insulin and pancreatic transcription factor mRNAs (Fig. 5). The active GRα form was found as soon as 6 wd with a higher expression around 8–9 wd. The GRβ was never detected at any developmental stage examined (GRβ/GAPDH ratio <0.00001). Insulin mRNA was weakly expressed until 9 wd (insulin/GAPDH ratio = 0.018) with a 12-fold increase between 8–9 and 10–11 wd (insulin/ GAPDH ratio = 0.23) and a further 10-fold increase between 10–11 and 22 wd (insulin/ GAPDH ratio = 2.06). IPF1 mRNA was detected at all stages examined with minimal variations, consistent with the expression of this factor in pancreatic cells at various stages of differentiation, including differentiated β cells. NEUROG3 mRNA was observed throughout pancreas development, consistent with the continuous endocrine cell differentiation in the pancreas during fetal life in human (2931). PTF1A mRNA was detectable at all stages with low levels until 10 wd and maximal expression levels at late stages, underlying its importance in early pancreatic progenitors and consistent with its implication in exocrine cell differentiation.

Figure 5
figure 5

Quantitative expression of GR and transcription factor mRNA during human fetal pancreas development. Three pancreata were analyzed at 6–7 wd, 8–9 wd or 10–11 wd and two at 22 wd for mRNA levels of (A) GRα, (B) GRβ, (C) Insulin, (D) Pdx-1, (E) NEUROG3 and (F) PTF1A-. Results are obtained by RT-PCR and expressed as relative ratio to GAPDH mRNA value. The mean ratio value of duplicates for each specimen at each time-point is shown as a single dot.

DISCUSSION

Our results show the presence of the active form of the glucocorticoid receptor (GR) in the human developing pancreas as soon as 6 wd in the mesenchymal cells but not in the first immature endocrine cells dispersed in the epithelial tubules. As differentiation progresses, the GR is observed in a subfraction of IPF1-positive epithelial cells at the time they express insulin and organize into islet-like clusters. These findings further document how human pancreas develops and support the novel idea that glucocorticoids (GC) could possibly modulate this development.

The human pancreas develops as ventral and dorsal outgrowths of the foregut endoderm, initiating at about 3–4 wd with the dorsal derivative. The fusion of the ventral and dorsal bud takes place at 8 wd (28). At 6 wd, IPF1 was present in the majority of epithelial cell nuclei from the tubular structures budding from the duodenum into a dense mesenchyme. IPF1 therefore represents an early marker of the human pancreatic epithelium. At this stage, the active form of the GR is present in the mesenchymal cells. As epithelio-mesenchymal interaction is important for pancreas morphogenesis in rodents (32,33), the presence of GR in mesenchymal cells suggests such a role for glucocorticoids during human pancreas development before endocrine and exocrine differentiation take place.

GR and human β-Cell development.

At 9 wd, insulin cells and glucagon cells were dispersed in the epithelial cell tubules, in agreement with other observations of the first endocrine cells around 7–8 wd in the human pancreas (2931). These first insulin cells do not express IPF1 or GR, suggesting that their development is not directly GR-dependent and not IPF1-dependent, as shown in mice (34). This study shows that islet formation starts from 10 wd onwards with insulin cell clusters forming within the epithelial cell tubules. Interestingly, the isolated insulin expressing cells still observed at 10 wd express neither GR nor IPF1 while insulin cells grouped in clusters express both markers. The first period when IPF1-positive epithelial cells start expressing insulin thus coincides with endocrine cell clustering. The presence of the GR at this moment in the same epithelial structures points out the potential role of GC on the endocrine differentiation of IPF1-positive cells, possibly by modulating IPF1 expression, as already shown in rodents in vivo and in vitro (21). Interestingly, prenatal dexamethasone treatment of pregnant monkeys resulted in reduced β-cell number in the offspring, highlighting the negative role of GC on β-cell development also in nonhuman primates (35). From the period of islet formation until adult age, GR and IPF1 colocalize in insulin expressing cells. The fact that mature human β cells express a persistent GR signal suggests a possible role of GC on β cell function, consistent with the direct inhibition of GC on insulin secretion demonstrated in vivo in humans (36), as well as in vitro and in vivo (37) in mice.

GR and exocrine pancreas development.

The development of the exocrine pancreas started 2–3 wk later than that of the endocrine and was observed between 13 and 23 wd, characterized at early stages by the nuclear expression of the PTF1A transcription factor and later by the expression of carboxypeptidase A (not shown). In mature human acinar cells, GR immunoreactivity was never observed, contrasting with its persistent expression in rodent adult acinar cells (21). The presence of GR may no longer be necessary once acinar differentiation is resumed, suggesting that GC would be more likely implicated in the exocrine differentiation process rather than in acinar cell expansion.

The active form of the GR is expressed in the developing human pancreas.

The actions of GC are mediated by the GR, which binds GC hormones and regulates gene expression, cell signaling, and homeostasis (38). In recent years, increasing numbers of human GR isoforms generated from one gene have been reported. The GRα form binds the hormone and induces the expression of a GC responsive reporter gene in a hormone-dependent manner. The GRβ isoform acts as a dominant negative inhibitor for GRα transcriptional regulation (39,40). Increased GRβ expression has been correlated with several diseases related to glucocorticoid resistance and inflammatory conditions but is minimally expressed in physiologic conditions (41,42). In line with these observations, the GRβ isoform mRNA was never detected in the developing human pancreas between 6 and 23 wd and the GRα was the only human GR isoform detected. The profile of the active GRα expression by RT-PCR could define a GC-vulnerable period corresponding to the beginning of islet formation. These results closely match our immunohistochemical observations and are consistent with those obtained after in situ hybridization in mice showing a peak expression of the GR around E14.5–16.5, just before the second wave of β-cell differentiation (43). Alternatively, it cannot be excluded that the apparent decrease of GRα mRNA observed at late stages reflects variations in nonendocrine cell numbers, such as the continuous expansion of the acinar cells not expressing the GR and the progressive decrease of mesenchymal cells initially expressing the GR.

The expression profiles of insulin, NEUROG3 and PTF1A were studied in the developing human pancreas. The insulin mRNA profile showing weak levels until 8–9 wd followed by an increase around 11 wd follows our immunohistological findings where insulin cells are present from 8–9 wk and expand thereafter. The pro-endocrine transcription factor NEUROG3 (2) was already present around 6–7 wd just before islet formation and its expression persists until 22 wk, contrasting with the down-regulation of this gene at late fetal stages in rodents, but in line with the ongoing islet neogenesis described in the human developing pancreas (2931). This increase could also suggest that NEUROG3 has functions other than islet differentiation, such as islet innervation. PTF1A mRNA was detectable at all stages examined, consistent with its immunohistological observation as early as 8 wd, suggesting that PTF1A is also an early marker of human pancreatic progenitors.

Taken together our results show the presence of the active form of the GR as soon as 6 wd and its colocalization with insulin cells from the beginning of islet formation. These data suggest for the first time a potential regulation of human fetal β-cell development by GC. Our data show that the mRNA and protein machinery are present in the human fetus for such a regulation to take place. The GC-sensitive period around islet formation points out the potential harmful action of early prenatal GC treatment or stress on human fetal pancreas development with the risk of developing β-cell insufficiency later in life.