Glucagon Is Required for Early Insulin-Positive Differentiation in the Developing Mouse Pancreas

Time-dated pregnant mice were obtained by breeding CD-1 mice and checking for vaginal plugs the next morning. Pregnancy was defined as 0.5 day of gestation at noon of the day the plug was found. The mice were killed on days 11.5, 13.5, and 15.5 of gestation. Embryos were harvested and preserved on ice in Dulbecco’s modified Eagle’s medium (DMEM). Microdissection was performed on the individual embryos to isolate and remove embryonic pancreata as previously described (12). All procedures were approved by the University of Missouri Medical Center, Kansas City, Animal Care and Use Committee in accordance with the Guide for the Care and Use of Laboratory Animals (Department of Health and Human Services, National Institute of Health publication no. 86-23, revised 1985).

Gestational day 11.5, 13.5, and 15.5 pancreata with surrounding mesenchyme were grown in a Collagen I gel (Vitrogen, Celtrix Pharmaceuticals) on Millipore filter inserts by placing the inserts in standard 24-well plates and adding either 20–40 μmol/l Morpholino (Gene Tools), standard missense, or antisense-laden serum–containing media (E15 pancreata were cut into small pieces, as they are too large to be grown whole in a collagen gel culture). Media was prepared by adding 25 μl of antisense or standard missense oligonucleotides to 475 μl of filter-sterilized media containing 99% DMEM/F12K and 1% antibiotic/antimycotic solution (penicillin G 10,000 μg/μl, streptomycin sulfate 10,000 μg/ml, and Amphotericin B 200 μg/ml; Gibco). A 20-bp standard control oligo sequence was designed using random oligonucleotides (5′-CCTCTTACCTCAGTTACAATTTATA-3′) with a similar G-C content. An antisense sequence to preproglucagon was designed using a 20-bp sequence complimentary to a sequence spanning the preproglucagon adenine, thymine, guanine (ATG) start sequence (5′-GGTCTTCATTTTTTATTCTGCCTTG-3′). Standard or antisense treatment was continued for 5–6 days where media was replaced after 3 days or every other day on a few occasions with fresh standard or antisense-containing media. Some pancreata were also grown in normal media without any treatment as a control for the standard missense control oligo. Tissues were cultured in 5% CO₂ at 37°C for 5 days. In a few cases, we removed the antisense laden media after 3 days of treatment and then cultured for an additional 3–6 days with normal media without any treatment. Pancreata were then fixed and processed for histology and immunohistochemistry. For each experiment at least seven pancreata were used per treatment group, and most of the experiments were repeated on an average five times.

E11 gestational age pancreata were cultured in a collagen gel with a neutralizing antibody against glucagon (1:200 diluted mouse monoclonal antibody against glucagon; Sigma). As a control, E11 gestational age pancreata were cultured in a collagen gel with either normal mouse or normal rabbit antibody. Media was replaced after 3 days of culture, and pancreata were harvested, fixed, and processed for histology and immunohistochemistry after 6 days.

E11 gestational age pancreata were cultured in a collagen gel with 20 μmol/l standard missense or antisense oligonucleotide against preproglucagon with different doses (50, 100, and 200 nmol/l) of added exogenous glucagon peptide. Media was replaced after 3 days of culture. Pancreata were then fixed and processed for histology and immunohistochemistry. To see if exogenous glucagon would have any effect on insulin positive differentiation, E11 pancreata were also grown in media containing different doses (50, 100, and 200 nmol/l) of glucagon peptide.

E11 gestational age pancreata were cultured in a collagen gel with 20 μmol/l Morpholino antisense or standard oligo as described above, with 0.001 nmol/l exendin-4 (a GLP-1 agonist). In a separate culture we blocked the exendin-4 effect by also culturing the pancreata with 20–40 μmol/l Morpholino antisense or standard oligo with 0.001 nmol/l exendin-4 and 0.01 nmol/l of exendin-(9-39) (a GLP-1 antagonist). We also studied the effect of different doses of exendin-(9-39) (0.01, 0.1, 1, 10, and 50 nmol/l) alone on E11 and E15 embryonic pancreas culture to determine whether endogenous GLP-1 may play a role in insulin cell differentiation. Culture media was replenished after 3 days (except every 2 days in the exendin-[9-39] groups). Both exendin-4 and exendin-(9-39) have been shown to have a half-life of >24 h. Stoffers et al. (25) have shown that a single intraperitoneal injection of exendin-4 daily can stimulate insulin secretion and lower blood glucose levels in diabetic mice. An even longer functional half-life has been reported for exendin-(9-39). Exendin 9-39 coinjected with exendin-4 once a day into mice either intraperitoneally or intravenously effectively blocked exendin-4–mediated insulin secretion (25). The pancreata were harvested after 5 days of total culture and fixed and processed for histology and immunohistochemistry.

Tissues were immediately fixed in 4% paraformaldehyde (in PBS) solution for 1 h. They were then dehydrated in 70% ethanol for 15 min and transferred to 30% sucrose (in PBS) for 4–12 h. The pancreata were pooled according to their specific gestational age and treatment conditions and embedded in optimal cutting temperature compound for subsequent cryostat sectioning.

The frozen blocks of pooled embryonic pancreata were cut into 6-μm sections using a Bright OTF (open top freezer) model cryostat (Hacker Instruments). Except when indicated, all incubations were performed at 25°C. Antibodies against insulin (sheep anti-human; The Binding Site), glucagon (mouse monoclonal; Sigma); and amylase (rabbit anti-human; Sigma) had been quality tested and do not cross-react with other proteins. The antibodies were diluted in PBS, and their optimal concentrations for immunofluorescence were determined to be 1:500 for all of the antibodies used (insulin, glucagon, and amylase). Controls for the primary antibodies were performed at each time point using normal serum rather than the individual antibodies. For immunofluorescent double staining, the slides were also subjected to treatment with 10 mmol/l sodium citrate and Tween-20 (0.5% in PBS). Incubation with the blocking serum (4% normal donkey serum in PBS) took place overnight at 4°C. Both primary antibodies were applied together for 2 h at 25°C after another exposure to Tween-20 (0.5% in PBS). Next, the secondary antibodies (donkey anti-sheep tetramethylrhodamine isothyocyanate [TRITC] 1:200, donkey anti-mouse fluorescein isothiocyanate [FITC], and donkey anti-rabbit FITC 1:200; Jackson Laboratories) were applied together after one more Tween-20 (0.5% in PBS) wash.

Serial sections of pancreas were mounted on slides. On average, 15–20 sections per pancreas were analyzed for each expressed gene that was quantified. Each group had a minimum of five pancreata. The experiment was repeated in triplicate. Fluorescent images of sections were scanned at the same magnification and exposure conditions. Quantification was done by Image Pro-Plus software (Media Cybernetics). First, total area of the pancreas was measured, and then insulin (Rhodamine)- and glucagon (FITC)-specific areas were determined on the same section. The percentage of insulin- and glucagon-specific area was then determined. Percentage of amylase positive area (FITC) was also quantified in the same manner on adjacent slides of serial section. Statistical analysis was done by paired t test.

Total RNA was isolated from whole pancreas using the RNeasy kit (Qiagen). Isolated RNA was treated with DNase for 1 h at 37°C to remove any contaminating genomic DNA. Oligo (dt) primed reverse transcriptase of RNA was done for 2 h at 37°C using superscript II (Gibco). First-strand cDNA was used as a template for PCR. Optimum reaction conditions for PCR were obtained with 5 μl 10× PCR buffer (200 mmol/l Tris-HCl [pH 8.4] and 500 mmol/l KCl), 3 mmol/l MgCl₂, 0.2 mmol/l dNTP, 0.025 units AmpliTaq Gold DNA polymerase (Applied Biosystems), and 2 μmol/l primers. Amplifications were performed starting with a 10-min denaturation step at 95°C, followed by 40 cycles of denaturation at 95°C for 30 s, 58°C for 40 s for primer annealing, and 72°C for 45 s for extension. Gene-specific primers for glucagon receptor (5′-TTGCCTTTGTGACTGACGAG-3′ and 5′-TAGAGAACAGCCACCAGCAG-3′) and GLP-1 receptor (5′-CTTGATGGTGGCTATCCTGT-3′ and 5′- TTCATGCTGCAGTCTCTCTG-3′) were designed using Primer-3 (http://genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi) to determine the relative expression. As internal control, primers for β-tubulin were designed (5′-TGGGACTATGGACTCCGTTC-3′ and 5′-GCACCACATCCAAGACAGAG-3′).

Gene-specific primers for insulin (5′CAGAAACCATCAGCAAGCAGG-3′ and 5′-TTGACAAAAGCCTGGGTGGG-3′) and glucagon (5′-AGAAGAAGTCGCCATTGCTG-3′ and 5′-CCGCAGAGATGTTGTGAAGA3′) were designed using Primer-3 (http://genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi) to determine the relative expression. As an internal control, primers for β-tubulin were designed (5′-TGGGACTATGGACTCCGTTC-3′ and 5′-GCACCACATCCAAGACAGAG-3′). Real-time PCR was carried out using the iCycler iQ real-time PCR detection system (Bio-Rad). Total RNA was extracted from E11 pancreas cultured with preproglucagon antisense or missense standard control oligo using the Rneasy kit (Qiagen). Isolated RNA was treated with DNase for 1 h at 37°C to remove any contaminating genomic DNA. Oligo (dt)-primed reverse transcription of mRNA was done for 2 h at 37°C using superscript II (Gibco). The first-strand cDNA was used as a template for semiquantitative PCR. Optimum reaction conditions for semiquantitative PCR were obtained with 5 μl 10× PCR buffer (200 mmol/l Tris-HCl [pH 8.4] and 500 mmol/l KCl), 3 mmol/l MgCl₂, 0.2 mmol/l dNTP, 0.025 units AmpliTaq Gold DNA polymerase (Applied Biosystems), and 2 μmol/l primers, and SYBER Green PCR reagents (S9340; Sigma) were used in 1:20,000. A total of 10 μl of 1:10 dilution of cDNA template was used. Amplifications were performed starting with a 10-min denaturation step at 95°C, followed by 40 cycles of denaturation at 95°C for 15 s, 58°C for 20 s for primer annealing, and 72°C for 30 s for extension. The fluorescence increase was automatically measured during PCR. The quantitative standards for insulin and β-tubulin gene as internal control were used to determine the level of gene expression for insulin and β-tubulin in the treated pancreas sample. The iCycler software was used for instrument control, data acquisition, and analysis. Statistical analysis was done using Student’s t test.

To begin studying how glucagon may affect β-cell differentiation and insulin gene expression, we blocked glucagon gene expression in cultured embryonic pancreas using Morpholino antisense oligonucleotides against preproglucagon mRNA. We found that in E11 whole developing pancreata (n = 35) (with mesenchyme intact), cultured for 5 days in the presence of preproglucagon antisense oligonucleotides, insulin expression as measured by immunohistochemistry was almost completely suppressed (Figs. 1B and 2).

The amount of glucagon-positive staining was significantly reduced under the antisense treatment (Fig. 1C and 2) as measured by positive area on immunohistochemistry.

We attempted direct protein quantification of glucagon through enzyme-linked immunosorbent assay to confirm the effects of antisense, but given the small amount of tissue available, the assay was not adequately sensitive. Amylase expression (Fig. 1D), however, as a marker of exocrine development, appeared to be normal in the pancreata treated with preproglucagon antisense oligos. Insulin, glucagon, and amylase expression all showed a normal quantity and expression in the presence of a standard missense control oligonucleotide (Fig. 1F–H). Semiquantitative RT-PCR analysis of E11 embryonic pancreas treated with preproglucagon antisense showed a significant decrease (P < 0.002) in insulin gene expression as compared with missense standard control oligo-treated pancreas (Fig. 3).

We tested the effect of blocking glucagon expression in older embryonic pancreas at gestational ages E13 and E15 using Morpholino antisense oligos against preproglucagon. On average, we used seven pancreata per treatment condition, and pancreata were cultured for 5 days with and without antisense. In E13 pancreas, preproglucagon antisense effectively blocked insulin expression (Fig. 4A) and, as with the E11 pancreas, glucagon expression was greatly reduced (Fig. 4B), as determined by immunohistochemistry. Missense standard control oligo-treated E-13 pancreas, on the other hand, showed normal levels of insulin (Fig. 4C) and glucagon (Fig. 4D) expression.

Amylase expression appeared normal in both antisense- and missense-treated pancreas (data not shown). Missense control oligo-treated E11 and E13 pancreata showed similar levels of insulin expression compared with untreated cultured pancreas for the same duration, suggesting that control oligos did not have an effect. Interestingly, E15 embryonic pancreas treated with preproglucagon antisense oligos showed normal expression of insulin (Fig. 4E) and glucagon (Fig. 4F), suggesting normal differentiation of α- and β-cells. Lack of effect of preproglucagon antisense on insulin expression in E15 embryonic pancreas could be due to high expression of glucagon at this gestational age. Higher doses (up to 40 μmol/l) of preproglucagon antisense treatment on E15 pancreas for 5 days also did not have any effect on insulin or glucagon expression. Similarly, as with younger embryonic pancreata, preproglucagon antisense treatment did not affect amylase expression in E15 pancreas (data not shown). Based on the data, we are unable to directly determine whether glucagon is necessary for the second wave of insulin differentiation.

Glucagon ontogeny was determined by semiquantitative RT-PCR. Preproglucagon expression in embryonic pancreas increased with age. The level of expression increased significantly by E13 and E15, as compared with E11 or E12 (Fig. 5A). Using RT-PCR, we showed presence of the glucagon receptor in the embryonic pancreas. We then performed semiquantitative RT-PCR to determine the ontogeny of the glucagon receptor and found a profile similar to that of preproglucagon in that its expression was low at E12 and then increased significantly by E13 and E15 (Fig. 5B).

To test whether the insulin-blocking effect is due to the absence of glucagon, or of other proteolytic products of preproglucagon, we used a neutralizing antibody against glucagon. This antibody does not cross-react with GLP-1, GLP-2, or glicentin. E11 pancreas cultured with neutralizing antibody against glucagon again showed minimal expression of insulin (Fig. 6A). To check if blocking early insulin differentiation by glucagon antisense may have an effect on subsequent development of the embryonic pancreas, we treated E11 embryonic pancreas with preproglucagon antisense for 3 days, changed the media to fresh media without antisense, and then cultured for an additional 3–6 days (n = 25). Here insulin-positive differentiation was recovered (Fig. 6 D–G).

We then tested the ability of exogenous glucagon peptide to rescue a normal phenotype in cultured pancreas treated with either preproglucagon antisense or glucagon-neutralizing antibody. Excess glucagon peptide (100 nmol/l) plus either antisense against preproglucagon or neutralizing antibody against glucagon was able to rescue insulin gene expression (Fig. 6B and C). Exogenous glucagon peptide alone, on the other hand, did not have any effect on insulin-positive differentiation in embryonic pancreas (Fig. 6H).

GLP-1 can stimulate insulin differentiation in AR42J cells and insulin secretion by β-cells. First, we used real-time RT-PCR to study the ontogeny of GLP-1 receptor in mouse embryonic pancreas. We found a significant increase in expression at E15 (Fig. 5C). We then confirmed this expression using immunohistochemistry for GLP-1 receptor in E11 pancreas up to adult and found staining for GLP1 receptor at all ages (Fig. 6I–K). We stained E11 pancreas, with or without treatment with preproglucagon antisense, for GLP-1 to determine whether the preproglucagon antisense effect may be due to decreased levels of GLP-1. As expected, immunofluorescence staining for GLP-1 was either significantly reduced or completely abolished in pancreata treated with preproglucagon antisense (Fig. 6L and M). Exendin 4 is a GLP-1 analogue that is ∼100-fold more potent than GLP-1 in its insulinotropic action. We used exendin-4 in our pancreas culture treated with preproglucagon antisense to determine whether a GLP-1 agonist could rescue insulin expression. Exendin 4 at 0.001 nmol/l rescued insulin expression in E11 embryonic pancreas cocultured with preproglucagon antisense (n = 30, P < 0.02) (Fig. 6N).

Exendin 9-39 is a specific antagonist of the GLP-1 receptor and has been shown both in vivo and in vitro to inhibit the insulinotropic action of GLP-1 in a dose-dependent manner. In E11 pancreas cocultured with 0.001 nmol/l, exendin-4, and 20 μmol/l preproglucagon antisense oligo, 0.01 nmol/l exendin-(9-39) blocked the rescue of insulin-positive differentiation by exendin-4 (n = 20, P < 0.05) (Fig. 6O), thus confirming that the exendin-(9-39) was active and also implying that the exendin-4 rescue effect is mediated through the GLP-1 receptor. In older gestational pancreas (E15) exendin-(9-39) did not have any effect in blocking insulin-positive differentiation, even at higher concentrations of exendin-(9-39) (Fig. 6P).

To determine whether GLP-1, like glucagon, may be acting endogenously in the induction of insulin-positive differentiation in the early embryonic pancreas, we treated cultured pancreata with the GLP-1 inhibitor, exendin-(9-39) alone. Pancreata of ages E11, E13, and E15 were treated with different concentrations of exendin-(9-39) alone (at 0.01, 0.1, 1, 10, and 50 nmol/l), and we saw no effect of exendin-(9-39) on insulin expression. Consistent with these results, we described above that GLP-1 receptor expression was lower at earlier ages of pancreatic development (up to age E13), but by E15 GLP-1 receptor expression increased significantly (Fig. 5C).

Immunohistochemistry for the GLP-1 receptor at ages E11, E15, and adult shows a pattern similar to the semiquantitative RT-PCR data. The staining for GLP-1 receptor was very weak at age E11 but increased by age E15 and by adult (Fig. 6I–K). Based on this low level of early GLP-1 receptor expression, the lack of effect of the GLP-1 antagonist is not surprising.

The development of the pancreas appears to require the orchestration of a complex set of events involving transcription factors and growth factors. Many studies in cell-cell signaling in pancreatic development have focused on epithelial-mesenchymal interactions (13–15), with little or no studies of the interactions between the pancreatic epithelial cells themselves. Glucagon seems to be the earliest peptide hormone to be expressed in appreciable amounts in the developing pancreas. Early studies suggested that other endocrine cells of the pancreas might originate from glucagon-expressing α-cells (2,4,5,16). However, lineage-tagged transgenic mice showed that insulin and glucagon cells might originate independently in the adult mouse (17). Pax6 null mutant mice developed without α-cells or glucagon expression, and these mice failed to express insulin during the early embryonic stages of development, although as adults they did develop β-cells and expressed insulin (8). These latter results suggest that insulin-positive differentiation may be differentially regulated in early versus late gestation embryos. In our study, embryonic pancreas at gestational ages 11–13, grown in culture in the presence of antisense oligonucleotides against preproglucagon, resulted in marked inhibition of insulin gene expression, as detected by immunohistochemistry. However, gestational age E15 pancreas grown with preproglucagon antisense showed a normal level and pattern of insulin gene expression. While we cannot directly show here that glucagon expression is necessary for the second wave of insulin differentiation, our results are consistent with those of Dohrmann et al. (8), wherein pax-6 mutant embryos that had no glucagon expression showed no insulin-positive cells early in gestation but did have normal insulin-positive differentiation during the second wave later in gestation. During development, the pancreas appears to have two phases of insulin-positive differentiation. The first wave of insulin differentiation begins around day 9.5–11.5 of gestation. The second wave of insulin differentiation is a larger proliferation of endocrine-specific cells, particularly β-cells, that begins around day 15 of gestation. Consistent with the results in the pax-6 null mutant mice, our results using antisense against preproglucagon suggest that glucagon exerts its effects on insulin gene expression during the first wave of insulin-positive differentiation at embryonic day 11. Earlier ages could not be studied because the embryonic mouse pancreas does not differentiate well in culture if harvested before day 11 of gestation. On the other hand, the second wave of endocrine cell proliferation does not appear to be affected by inhibition of preproglucagon gene expression. An alternative explanation for the lack of effect at older ages is that higher glucagon levels are not suppressed as well by the antisense or neutralizing antibodies, and thus higher glucagon expression allows some leaky stimulation of insulin-positive differentiation. This possibility seems unlikely because we could block insulin differentiation at E13 where the glucagon levels were similar to that of E15, in which we could not block.

Because multiple gene products are derived from the preproglucagon gene, we sought to determine whether the specific alternate preproglucagon processing products, glucagon or GLP-1, were necessary for insulin-positive differentiation. To our knowledge, specific antagonists or neutralizing antibodies are not available for GLP-2 or glicentin, the other known processing products of preproglucagon, and thus were not studied. GLP-1 is a peptide hormone primarily localized to the gut (18), which has been shown to have insulinotropic activity (11). AR42J cells grown in the presence of GLP-1 can develop into insulin-producing cells (11). Because glucagon and GLP-1 are both post-translational products of preproglucagon, morpholino antisense treatment (blockage at the translational level) will presumably decrease levels of both of these products. Exendin 4 is a potent and specific analog of GLP-1 (19–21). An intraperitoneal injection of exendin-4 once daily for 2 weeks has lowered blood glucose levels in wild-type and diabetic mice (25). In vitro treatment of AR42J cells with very low dose exendin-4 for 48 h resulted in those cells turning into insulin-producing cells (11). Coculturing pancreas with preproglucagon antisense and 0.001 nmol/l exendin-4 reversed the insulin inhibition effect of preproglucagon antisense in embryonic pancreas, suggesting that GLP-1 could be the endogenous molecule mediating the early differentiation of insulin-specific cells in the embryonic pancreas. Alternatively, GLP-1 could be capable of mimicking the effect of another preproglucagon processing product, but in that case would not actually be the key endogenous molecule. In that scenario, the GLP-1 receptor would have to be capable of mimicking the effect, since exendin-4 is not known to bind to receptors other than the GLP-1 receptor.

RT-PCR analysis of embryonic pancreas showed the presence of the GLP-1 receptor in embryonic pancreas. Immunohistochemistry for GLP1 receptor and semiquantitative RT-PCR for GLP-1 receptor confirmed the presence of GLP-1 receptor in embryonic pancreas. To determine whether GLP-1 was the key endogenous molecule regulating the differentiation of early insulin specific cells, we used a known GLP-1 antagonist, exendin 9-39, in normal cultured E11 pancreas. Exendin 9-39 is a smaller peptide derived from exendin-4 and is an effective antagonist of GLP-1 and exendin-4, specifically blocking both the binding sites and biological activities of GLP-1 and exendin-4 (19,20,22). Exendin 9-39 has also been shown to be a relevant physiological antagonist for assessing the biological importance of GLP-1 by blocking the binding and activity of GLP-1 in vivo (23,24). Exendin 9-39 coadministered with exendin-4 by intraperitoneal injection reversed the insulinotropic activity of exendin-4 in wild-type and diabetic mice but not in GLP-1 receptor knockout mice (25). In our whole pancreas culture, we demonstrate that exendin-(9-39) had an antagonistic activity on exogenous GLP-1. Exendin 9-39 inhibited the insulin rescue by exendin-4 of preproglucagon antisense treated E11 pancreas, thus both confirming that exendin-(9-39) was effective as an antagonist, and also confirming that the rescue of differentiation of insulin expressing cells here was due to exendin-4. However, exendin-(9-39) alone did not inhibit insulin-positive differentiation or insulin gene expression in the normal developing E11 pancreas, suggesting that GLP-1 may not be the critical endogenous molecule. Repeated intraperitoneal injection of very high concentrations of exendin-(9-39) every 4 h significantly reduced insulin level and blood glucose level in normal mice but not in GLP-1 receptor knockout mice (26). A single intraperitoneal coadministration of exendin-(9-39) and exendin-4 once a day for 2 weeks has effectively blocked exendin-4–mediated action in mice (25). We feel that the addition of a relatively high concentration (50 nmol/l) of exendin-(9-39) every other day to our pancreas culture was far in excess of the amount of antagonist needed in the culture to block any endogenous GLP-1 that may have an effect on β-cell differentiation. It is possible that endogenous peptide action may be difficult to antagonize compared with antagonizing coadministered exogenous peptides. We are not certain that repeated treatment of pancreas with exendin-(9-39) replenished in culture every few hours may have a different outcome other than what we have shown.

Expression of GLP-1 receptor by semiquantitative RT-PCR in the developing pancreas was significantly lower at ages E13 and earlier. GLP-1 receptor expression increased dramatically at E15. Immunohistochemistry for GLP-1 receptor on embryonic pancreas sections showed the same profile as semiquantitative RT-PCR. GLP-1 receptor staining was very weak at E11, which increased in E15 and adult pancreas. These data also support that GLP-1 is not the key endogenous molecule responsible for early insulin differentiation, and that the rescue effect shown by exendin-4 on E11 pancreas treated with preproglucagon antisense may be due to exendin-4 acting pharmacologically through the lowly expressed GLP-1 receptor.

To determine whether glucagon itself could be the key endogenous molecule responsible for regulating early insulin-positive differentiation in E11 pancreas, we treated the pancreas with 100 nmol/l glucagon along with 20 μmol/l preproglucagon antisense. We were able to rescue the insulin-positive differentiation in preproglucagon antisense-treated E11 pancreas using exogenous glucagon. To confirm an endogenous role of glucagon, we treated normal cultured E11 pancreas with neutralizing antibody specific for glucagon and successfully blocked insulin-positive differentiation. This antibody blockage of insulin-positive differentiation was also rescued by the addition of excess exogenous glucagon, suggesting that glucagon could be the endogenous molecule mediating the early differentiation of insulin-specific cells in the embryonic pancreas, although exogenous glucagon added to the culture did not have any positive effect on insulin-positive differentiation in E11 pancreas. At the same time we were able to rescue the preproglucagon antisense-induced insulin inhibition by replacing the antisense-media with normal media. This result suggests that the cells that do not express insulin after antisense treatment did not loose the insulin-positive differentiation potential. The lack of effect of exogenous glucagon on insulin-positive differentiation may be due to a near-maximal number of precursor cells normally becoming insulin-positive, and thus it may be difficult to demonstrate insulin positive differentiation over and above the normal near-maximal level.

In summary, we conclude that preproglucagon-derived signals, particularly glucagon itself, within the endocrine compartment of the developing pancreas, are necessary for the differentiation of early insulin-producing cells. Such interactions make teleologic sense, since it would be preferable to have a mechanism for raising blood glucose (regulated by glucagon) in place before creating cells that can lower blood glucose. Specifically, glucagon cannot cross the placental barrier to reach the embryo or fetus (27–29), and thus it may be critical to the embryo to have local control of glucagon in order to maintain blood glucose before it develops insulin-positive cells that will decrease the blood glucose level, particularly early in gestation when maternal glucose may be less available due to an immature vascular system to bring the maternal glucose to tissues. These data represent a novel concept in endocrine differentiation, i.e., that peptide hormones of one type of endocrine cell can control, through paracrine pathways, differentiation of other types of endocrine cells in the embryo.

This work was supported by a grant from the Juvenile Diabetes Research Foundation (no. 2-199-636, the National Institutes of Health (no. R21 DK57224-03), and Children’s Mercy Hospital, Kansas City, MO.

We thank Susan Bonner-Weir for reviewing the manuscript.