Improved Glucose Tolerance and Acinar Dysmorphogenesis by Targeted Expression of Transcription Factor PDX-1 to the Exocrine Pancreas

RESEARCH DESIGN AND METHODS

Creation of transgenic mice.

A 0.5-kb fragment of the rat elastase-1 promoter (19) was ligated into the plasmid PDX-1 (R1m) RGBpBS-KS containing the entire coding sequence of rat PDX-1. The strategy of the vector construction has been described in detail elsewhere (18). The recombinant plasmid also contains the 0.64-kb rabbit β-globin splicing signal cassette between the PDX-1 cDNA and the elastase promoter and the 0.52-kb rabbit β-globin polyadenylation signal cassette 3′ to the PDX-1 cDNA (20). DNA sequencing and polymerase chain reaction (PCR) analyses confirmed the proper orientation of the PDX-1 cDNA.

Transgenic (TG) mice were produced at the Beth Israel Transgenic Mouse Facility (Boston, MA) by standard transgenic protocols. DNA was isolated from tail snips obtained from mice age 4 weeks. Tails were digested with proteinase K (Boehringer Mannheim, Indianapolis, IN) and the DNA was precipitated with ethanol. Genomic DNA was analyzed by PCR using primers specific for the detection of the transgene (rabbit β-globin primer RBG 5′ CTCTGCTAACCATGTTCATGCCT 3′ and the PDX-1−specific reverse primer 5′ GCAGGCCAGCCAGGCTACAAAAAT 3′). The endogenous gene was amplified with the BZH05 forward primer 5′ AGATGTCCGGGGGACTTC 3′ and the same PDX-1 reverse primer. A total of 12 founders were identified, and a subset of these was characterized morphologically. Several lines were bred to homozygosity, as confirmed by backcrossing to FVB wild-type mice. After the initial characterization, subsequent analysis was performed in the representative homozygous elastase (El)-16 line.

Animal maintenance, handling, and experimentation were conducted according to guidelines laid down by the Policy on Use of Animals by the Massachusetts General Hospital and by Novo Nordisk A/S, and according to the regulations as specified under the Protection of Animals Act by the Authority in Denmark. The Massachusetts General Hospital Committee for Animal Care approved all animal protocols.

Immunocytochemistry.

Tissues were fixed overnight in formalin and embedded in paraffin. Slides were stained with rabbit anti−PDX-1 253 (1:3,000), guinea pig anti-insulin (1:5,000), mouse anti-glucagon (1:300), rabbit anti-human amylase (1:1,000), and rabbit anti-NKX 6.1 (1:2,000). The 1858 is a rabbit anti-PDX-1 antiserum (1:2,000) that specifically recognizes rat but not mouse PDX-1, and therefore allows for the specific detection of rat PDX-1 transgene expression. Staining was developed using Histostain SP kits containing streptavidin-peroxidase−conjugated secondary antibodies (Zymed Laboratories, San Francisco, CA), and visualized using diaminobenzidine, α-ethyl carbazole, or Fast Red as the chromagen or using immunofluorescence with Cy2-, Cy3-, and Texas Red−coupled secondary antisera (Jackson Immunoresearch, West Grove, PA).

Cellular replication and apoptosis.

Cellular replication was assessed by measuring by bromo-deoxy-uridine (BrdU) incorporation. Pregnant mice or individual postnatal offspring were injected with BrdU (2 mg/ml) 1.5 h before being killed to label the pool of mitotically active cells. BrdU incorporation into DNA was determined by immunocytochemistry with a mouse monoclonal antibody (Dako, Glostrup, Denmark). Anti−mitotic protein monoclonal (MPM)-2 (1:100; Dako) was also used to label mitotic cells in tissues from animals that had not received BrdU before being killed. Apoptosis was assessed by TUNEL assay (Oncor) according to manufacturer’s recommendations.

Electron microscopy.

Small pieces of pancreases were fixed in 2.5% gluteraldehyde in cacodylate buffer (pH 6.8) overnight at 4°C, postfixed in 1% osmium tetroxide, dehydrated, and embedded in Epon resin. Ultrathin sections were cut and examined using a Phillips electron microscope.

Stereological investigations.

The total number of β-cells and mean and total β-cell volume were estimated by the physical fractionator principle, as previously described (21). Briefly, whole pancreases were fixed in acidic formalin, subfractionated, and embedded in Technovit 8100 (Kulzer, Germany). The blocks were sectioned by ultramicrotome (Leica, Denmark), and 6–10 pairs of consecutive 1.5-μm sections per pancreas were immunostained for insulin and sampled by systematically uniformly random sampling (SURS) with known sampling fractions. The section pairs were investigated by two identical microscopes (Leica DMLB) equipped with projection arms to project the images onto the table at final magnification of 1260×. A point-counting grid was randomly superimposed to the sampled areas, and the points hitting β-cells were counted. The β-cells were counted by the dissector-counting principle (22) in the physical fractionator setup (23). Sampling fractions were chosen so that ∼100 cells were counted per pancreas. Because all sampling fractions were known, unbiased estimates of the total number of β-cells and the total and mean β-cell volume could be calculated. The total pancreatic volume was likewise estimated by the fractionator principle, except that the final magnification was 84×.

From mice age 8 months, the pancreases were removed and weighed before fixation and paraffin embedding. A set of sections (n = 6–9) was sampled by SURS from each pancreas. Unbiased estimates of volume fractions of β-cells, pancreas, and fat tissue were determined by the Delesse principle (24) using an Olympus BH-2 microscope equipped with a video camera and an automated x-y stepper and connected to a computer with the C.A.S.T.-grid software (Olympus). Sampling and grid density were chosen so that ∼150 points hitting β-cells and approximately the same number hitting pancreas were counted.

Intraperitoneal glucose tolerance test.

Age-matched female FVB control or TG mice, age 6 weeks, were housed on a 12-h light/dark cycle. Food was removed at 2:30 p.m. before the glucose challenge. At 9:00 a.m., mice were administered intraperitoneally 1.5 g of glucose per kilogram of body weight. Blood was withdrawn at 0, 15, 30, 60, and 120 min by orbital eye bleeding, and blood glucose was measured using a Hypocount MX B glucose monitor (Hermedico A/S, København, Denmark).

Insulin.

For six control and six TG mice, the pancreases were weighed on a Mettler analytical balance and immediately frozen on dry ice. The tissue was lyophilized, extracted in 3 mol/l acetic acid three times, lyophilized again, and reconstituted in 0.1 mol/l acetic acid. Protein was measured using the Bradford Assay (Biorad), and insulin was measured with an enzyme-linked immunosorbent assay.

Multiplex reverse transcriptase-polymerase chain reaction.

The multiplex reverse transcriptase (RT)−polymerase chain reaction (PCR) method described previously in detail (25) allows the coamplification of several cDNA products from total RNA preparations in a single tube. Briefly, RNA was prepared from pancreases using RNAzol (Cinna Biotecx), according to manufacturer’s instructions. cDNA was synthesized from 0.2 μg/μl of RNA, and all samples were diluted to the same final cDNA concentration. PCR was performed for 18–28 cycles, depending on the primer set. Internal standards for every PCR were included, and all data were normalized to the internal standard. Data were quantitated from the phosphorimage using the ImageQuant program. The sequences of the PCR primers used in these studies are available upon request.

Oil Red O staining.

Next, 5-μm sections of fresh frozen pancreases were cut, rinsed in 70% ethanol, and immersed in a 1% solution of Oil Red O (Sigma, St. Louis, MO) prepared in 70% ethanol for 15 min. Slides were rinsed, counterstained with hematoxylin, and mounted. Negative control slides were pretreated with acetone for 10 min to extract the lipids before staining.

Triglyceride measurements.

Pancreases were removed from mice ages 10, 20, 30, 42, and 56 days; frozen in liquid nitrogen; homogenized; and sonicated in ice-cold phosphate-buffered saline. The protein concentration was measured using the Bradford Assay (Bio-Rad). Triglycerides were measured with a triglyceride kit (Sigma), according to the manufacturer’s instructions. Data are expressed as triglyceride equivalents per milligram of protein.

Organ weight measurements.

Founder mice and complete litters of genotyped offspring of various ages were killed after an overnight fast. Their liver, spleen, kidney, and pancreas were removed and weighed.

Data analysis.

Data are expressed as means ± SE. Mean changes were compared using the Student’s t test. P ≤ 0.05 was considered to be statistically significant.

RESULTS

Transgene expression.

To examine the effect of misdirected overexpression of PDX-1 on the exocrine pancreas, 0.5 kb of the 5′ flanking region of the rat elastase-1 gene containing the promoter was fused to the coding sequences of the rat PDX-1 gene (20). The elastase 0.5-kb promoter fragment has previously been shown to drive the expression of transgenes exclusively in the exocrine pancreas (19). A total of 12 lines of TG mice were generated. The expression of PDX-1 was enhanced in all lines examined, assessed initially by Western blot analysis of total pancreatic protein (data not shown). These lines were examined histologically and bred to homozygosity. One highly expressing line, El-16, was used for the subsequent studies described herein.

Expression of PDX-1 in the exocrine pancreas was examined by immunostaining at e15 and e18 and postnatal days 10, 20, 30, 42, and 56. Enhanced expression of PDX-1 was observed in all epithelial cells at e15 (Fig. 1B). This contrasts with FVB control mice, which at e15 showed PDX-1 staining in a small subset of epithelial cells (Fig. 1A). By e18, all exocrine epithelial cells were strongly PDX-1 positive (Fig. 1C); by this time, PDX-1 is no longer expressed to a significant degree in control acinar cells. Staining for transgene-derived PDX-1 in the exocrine pancreas was similar in intensity to that for endogenous PDX-1 in islets. The transgene was not expressed in ducts or islets, as determined by a rat-specific PDX-1 antiserum that does not recognize mouse PDX-1 (Fig. 1D). Transgenic PDX-1 expression was maintained at all ages examined (up to age 8 months).

Alteration in pancreas size.

The size of the pancreases in the founder mice increased markedly as the mice aged. Five of the original founder mice were killed around age 1 year, and their body and pancreas weights were determined. The pancreases of the founder mice averaged 550 ± 45 mg wet wt, whereas the pancreases of the control animals of the same age weighed significantly less at 265 ± 35 mg (P < 0.002). Although there was a trend toward a larger body weight (data not shown), the differences were not significant (P < 0.10). No differences were found between TG mice and control non-TG littermates in liver, spleen, or kidney weight at any of the ages examined (data not shown). Analysis of the organ weights in the F1 generation showed a trend toward a smaller pancreatic weight, with no differences detected for the other organs. The homozygous El-16 animals age 8 months had a significantly smaller pancreas compared with age-matched control animals (158 ± 10 vs. 259 ± 11 mg wet wt, respectively; n = 6; P < 0.001).

Enhanced β-cell neogenesis but not β-cell mass.

Although single insulin-positive cells can be found in the exocrine pancreas of normal FVB mice, these cells are found in greater numbers in the exocrine pancreas of TG mice. These cells can be doublestained for insulin and NKX 6.1, a marker of mature β-cells (Figs. 1F and G), suggesting increased neogenesis of β-cells in TG mice. We also observed amylase/insulin double-positive cells at e18 in the TG mice only, suggesting that transdifferentiation could be taking place (Figs. 1H and I).

To determine whether the enhanced neogenesis of β-cells resulted in expansion of β-cell mass, we estimated the total number of β-cells and mean and total volume of β-cells. Table 1 shows the results of the stereological investigations. As a consequence of a selective decrease in the volume of the exocrine compartment (see below), the fraction of β-cell tissue per volume of pancreatic tissue was significantly increased at all ages examined; however, there was no difference in the total amount of β-cell tissue between TG and control mice at any age. In mice age 6 weeks, the total number of β-cells and the mean β-cell volume were also quantitated, and were similar between TG and control mice. The frequency of large islets was much greater in TG mice than in control mice (data not shown). Large islets were also observed in the El-1 and -3 lines (data not shown).

Increased glucose tolerance and pancreatic insulin content.

The in vivo response to glucose was evaluated in TG and FVB mice age 6 weeks. No body weight differences were observed in the two groups of age-matched mice (data not shown). Fasting blood glucose was ∼1 mmol/l lower in the TG mice compared with control animals (Fig. 2A). At 15 min after the injection of glucose, blood glucose values were not different between the control and TG mice. After 30, 60, and 120 min, blood glucose values declined more rapidly in the TG mice compared with the controls (P < 0.05).

Islets isolated from TG mice showed a 50% higher insulin release in response to 10 and 20 mmol/l glucose compared with FVB control mice (data not shown). To explain the enhanced in vivo and in vitro response to glucose without an increase in β-cell mass or number, we postulated that the TG mouse islets might have an increased insulin content. Indeed, despite the 60% decrease in overall pancreas weight of TG mice at this age (Fig. 2B), total insulin content per pancreas was 3.7-fold increased in TG versus control mice (6,203 ± 820.4 vs. 1,679.8 ± 748 pmol) (Fig. 2C).

Dysmorphogenesis of the exocrine pancreas.

As the mice aged, the well-formed exocrine acini seen in the early pancreas became less organized and showed increasing amounts of fatty infiltration. This was observed in multiple heterozygous founders as well as in several lines that were bred to homozygosity. This dysmorphogenesis was especially noticeable at 42 and 56 days after birth (Fig. 1E). Although the amount of pancreatic tissue was already reduced at postnatal day 8, the morphologic appearance of the exocrine tissue was unaffected in day 8 mice. Beginning at 20–30 days after birth, gaps were observed within the exocrine pancreas (Fig. 1D). By age 6 weeks, and especially evident by age 8 weeks, adipose tissue appeared throughout the exocrine pancreas (Fig. 3B). By age 8 weeks, ∼30–40% of the pancreatic mass had been replaced by adipose tissue. To confirm the presence of adipocytes, frozen sections were stained with Oil Red O. No Oil Red O staining was observed in the sections of the TG mouse in which the lipids had been extracted with acetone (Fig. 3B) or in the pancreas of an FVB control mouse (Fig. 3A). In contrast, there was marked staining throughout the exocrine tissue of the pancreas in TG mice age 56 days (Fig. 3C). As an independent marker of adipose tissue, pancreatic triglyceride content was measured in older control and TG mice ages 10, 20, 30, 42, and 56 days. As seen in Fig. 3D, the triglyceride content of the pancreatic extracts was greatly increased at 42 and 56 days in the TG animals, but was just above the level of detection in the control FVB animals. These observations corresponded in time to the appearance of the adipose tissue in the exocrine pancreas. Over a similar time course, multiplex RT-PCR analyses showed a dramatic increase in the expression of adipose-specific mRNAs encoding adipocyte protein-2, hormone-sensitive lipase, fatty acid synthase, and leptin in TG mice (data not shown).

We hypothesized that the observed dysmorphogenesis of acinar cells might be caused by an alteration in rates of proliferation and/or apoptosis in the TG mice. At e18, TG mice demonstrated a nearly twofold increase in BrdU incorporation in the exocrine pancreas compared with age-matched controls (Fig. 4A). The rates of replication during early postnatal days were consistently at least 20% greater in TG than in control animals. At the postweaning ages of 20, 30, and 42 days, a marked difference in replication was still apparent, as measured by MPM-2 immunoreactivity (Fig. 4B). It should be noted that BrdU labels more cells than the more specific MPM-2 antiserum. No differences in the labeling index were observed in the islets of TG versus control animals at any time point (data not shown). Many visible mitotic figures were observed throughout the exocrine tissue (data not shown). We considered the possibility that the altered architecture of the exocrine pancreas was attributable to increased apoptosis of acinar cells. By TUNEL assay, we observed a significant number of apoptotic acinar cells in the TG mice at ages 30 and 42 days (Fig. 5D), whereas few or no apoptotic cells were observed in control animals of the same age.

To further examine the ultrastructural changes in the exocrine pancreas, electron microscopy was performed on pancreases from animals ages 6 weeks and 6 months. Widespread acinar cell apoptosis was again observed in TG mice at age 6 weeks (Fig. 5C). At that age, the ultrastructure of the acinar cell in TG mice appeared, with classic exocrine granule morphology. TG mice had a larger number of degranulated cells compared with control mice (Figs. 5A and B). Colocalization of insulin and exocrine zymogen granules was not observed. At age 6 months, acinar cell size was greatly reduced. No exocrine granules were observed in the abundant adipocytes at this age.

Changes in gene expression.

To examine changes in gene expression in the pancreases of TG mice, multiplex PCR was performed on RNA from e15 to postnatal day 10. Expression of cholesterol esterase (Fig. 6A), chymotrypsin (Fig. 6B), and amylase (Fig. 6C) was greatly decreased in the TG mice. No differences were observed in levels of the exocrine gene markers lipase, RNase, and trypsin (data not shown). Expression of duct marker genes carbonic anhydrase and cytokeratin 19 was unchanged in TG mice compared with control animals (data not shown). Insulin gene expression was already increased at e15 and e17 (Fig. 6D), as well as at postnatal days 1 and 10.

DISCUSSION

In these studies, we examined the effects of targeted overexpression of the transcription factor PDX-1 to the exocrine pancreas lineage using the rat elastase-1 promoter. PDX-1 is a critical regulator of pancreas development and is highly expressed in the insulin-producing β-cells of the endocrine pancreas (1,2,3,4). We postulated that PDX-1 might be sufficient to drive a program of increased β-cell development if expression was targeted to the cells destined to become the exocrine pancreas. In keeping with the initial postulate, we observed a marked increase in the growth of the pancreas in the founder animals. In contrast, the same lines of TG mice showed decreased pancreatic size when bred to homozygosity, possibly due to a doubling of the transgene dosage. The possibility of an insertional artifact exists but is unlikely, as multiple homozygous lines analyzed showed small pancreas size and fat deposition in the pancreas.

The TG mice had an increased pancreatic insulin mRNA and peptide content and an enhanced response to glucose when compared with control mice. PDX-1 overexpression in the exocrine pancreas led to increased numbers of amylase-insulin double-positive cells (Fig. 1H), suggestive of a low level of transdifferentiation, as well as increased numbers of isolated β-cells within the acinar tissue (Figs. 1F and G), indicative of increased β-cell neogenesis. Surprisingly, islet size was increased, but overall β-cell mass was not. One explanation for the increased insulin biosynthesis and secretion in this setting is that upregulation of PDX-1 in exocrine tissue leads to expression of a signal or growth factor that stimulates insulin gene expression, biosynthesis, and secretion in the islets. Alternatively, the signal could be related to acinar cell death, as increased β-cell neogenesis has been observed in a number of pancreatic injury models (26,27).

The exocrine phenotype of these mice suggests that PDX-1 is a strong inducer of the cell cycle in the exocrine pancreas. The increase in proliferation is accompanied by a marked increase in apoptosis, leading to decreased pancreas volume and the replacement of exocrine tissue by adipose tissue. Interestingly, double TG mice that overexpress both neuroD and PDX-1 in the exocrine tissue have lower rates of proliferation, minimal apoptosis, and less adipose deposition, although glucose tolerance remains enhanced (R.S.H., unpublished observations). The expression of several exocrine-specific genes (amylase, chymotrypsin, and cholesterol esterase) was severely reduced when PDX-1 was ectopically overexpressed in the acinar cells. Because the expression of some exocrine genes was unaffected, a global loss of exocrine gene expression is not the explanation for this result. Whether this circumstance is attributable to a direct effect of PDX-1 on the promoters of these genes or a secondary effect on acinar cell differentiation is not yet clear. A recent report suggests that PDX-1 together with PBX-1b and MRG-1 (MEIS2) can act to positively regulate the exocrine elastase gene (28). Overall, the data suggest that appropriate downregulation of PDX-1 in the exocrine pancreas is required for normal acinar cell function, and that the regulation of PDX-1 in the context of the entire transcriptional milieu plays a critical role in the life cycle and differentiation state of the acinar cell.

The infiltration of the pancreas by adipose tissue has been observed in the pancreas of cystic fibrosis patients, in patients with long-standing chronic pancreatitis (G. Klöppel, personal communication), and in several other TG mouse models, including the overexpression of a dominant-negative transforming growth factor-β type II receptor in the pancreas (29). The mechanism by which cells within the exocrine pancreas differentiate into adipose tissue is unknown. One explanation is that the PDX-1−induced proliferation of the exocrine tissue results in the expression of a factor(s) that directs fibroblasts in the pancreas to develop into adipocytes. A less likely explanation is that PDX-1−expressing exocrine cells may transdifferentiate into adipocytes.

A recent study of heterotopic expression of PDX-1 in liver in vivo demonstrated the expression of insulin in a subpopulation of hepatocytes (30). We also found a low level of acinar cells appearing to transdifferentiate to insulin-producing β-cells. The misdirected overexpression of PDX-1 in the exocrine pancreas of mice led to accelerated apoptosis of acinar cells, which were ultimately replaced by adipose tissue, suggesting that proper downregulation of PDX-1 is critical for acinar cell survival.