Regeneration of Pancreatic Islets After Partial Pancreatectomy in Mice Does Not Involve the Reactivation of Neurogenin-3

Eight- to 10-week-old male BALB/c mice were purchased from Charles River Laboratories and housed under standard conditions. Female Ngn3^EGFP/+ mice (age 8 weeks) were generated as previously described (16) and maintained on a mixed Sv129/C57Bl6 genetic background. All procedures were approved by the University of Pennsylvania Institutional Animal Care and Use Committee.

Mice were anesthetized by administration of Nembutal Sodium solution (50 mg/kg i.p.). The abdomen was opened through a left lateral incision. The entire splenic portion of the pancreas was surgically removed, resulting in an ∼50% pancreatectomy, confirmed by weighing the removed and remnant portions during a pilot study. Sham operation was performed by opening the abdomen while leaving the pancreas intact. The incision was closed using 4-0 silk for the peritoneum and clips for the skin.

For islet replication measurements, BrdU (200 mg/kg) was administered 6 h before sacrifice. BrdU incorporation was assessed using a BrdU-specific mouse monoclonal antiserum (Sigma, St. Louis, MO). At least 1,200 islet endocrine cell nuclei were counted per pancreas. The rate of proliferation is expressed as the percentage of islet cell nuclei that are also BrdU positive. BrdU incorporation was determined at 0, 2, 4, 8, and 10 days after pancreatectomy.

β-Cell mass was determined 5 weeks after partial pancreatectomy by point-counting morphometry, as previously described (17). A section through the maximal footprint region of the pancreas was stained with guinea pig anti-human insulin (1:5,000; Linco Research, St. Charles, MO). The secondary antiserum was biotinylated donkey anti–guinea pig (1:200; Vector Laboratories, Burlingame, CA), and color development was with the ABC kit and diamino benzidine (Vector Laboratories, Burlingame, CA). Sections were evaluated using a Cool Snap digital camera (Photometrics, Huntington Beach, CA) connected to an Eclipse E600 microscope (Nikon). Using a 9 × 10 grid of points, the percentage of points overlying β-cells was determined and then multiplied by the weight of the pancreas to determine the mass of β-cells. At least 200 random fields were assessed for each pancreas.

E15.5 embryos were dissected from timed pregnancies. Digestive tracts, including liver, stomach, pancreas, and intestine, were removed and placed in 1× PBS. Enhanced green fluorescent protein (EGFP) expression was visualized and images were captured using a Leica dissecting microscope (Leica MZF III).

After vibratome sectioning, tissue slices were placed in 1× PBS in 24 wells and rinsed three times with 1× PBS. Tissues were washed in 1× PBS/0.1% TritonX for 40 min at room temperature and then blocked in 1× PBS/0.1% TritonX/5% BSA on a rotator at room temperature for 1 h. Guinea pig anti-insulin antibody (Linco) was diluted with 1× PBS/0.1% TritonX/5% BSA at 1:1,000 and incubated with the sections overnight at 4°C. Tissues were washed three times in 1× PBS/0.1% TritonX/1% BSA for 1.5 h, and anti–guinea pig antibody conjugated to rhodamine was added at 1:50 in 1× PBS/0.1% TritonX/5% BSA for 2 h at room temperature. Tissues were washed three times in 1× PBS/0.1% TritonX/1% BSA for 1.5 h, rinsed once in 1× PBS, mounted in 50% glycerol/PBS, and examined using confocal microscopy.

Ngn3 immunostaining was carried out as we described previously (18). Ngn3-EGFP expression was assessed in pancreatic and intestinal tissues harvested 4 days, 2 days, and 2 h postpancreatectomy. Tissues were fixed in 4% paraformaldehyde for 40 min, rinsed three times in 1× PBS, and then embedded in 4% agarose/PBS for vibratome sectioning. Tissue sections (150 μm) were cut, placed on microscope slides, and mounted in 50% glycerol/PBS. Slides were examined using confocal microscopy (Leica).

We sought to determine whether β-cell regeneration in adult mice recapitulates the embryonic pathway involving the proendocrine transcription factor Ngn3. The 50% partial pancreatectomy (Ppx) model was chosen because β-cell mass fully regenerates in this paradigm. Islet replication after 50% Ppx in BALB/c mice increases from a basal rate of 1.7% to ∼5–6% by day 4 after surgery and remains elevated until at least day 10 (Fig. 1A–D). This correlates well with the nearly complete recovery of β-cell mass by 4 weeks after Ppx (Fig. 1E). In contrast, pancreatic organ weight only partially recovers to 65 ± 0.98% of sham-operated pancreas weight (P = 0.003), suggesting that the endocrine compartment has greater regenerative capacity than the exocrine compartment in this surgical model and genetic background.

To determine whether β-cell regeneration involves a reactivation of Ngn3 expression, we first examined Ngn3 immunoreactivity at 0, 0.5, 1, 2, 3, 4, 6, 8, and 10 days after Ppx. Although our conditions permitted detection of Ngn3 during fetal development (Fig. 2A and B), we failed to detect specific Ngn3 staining in the pancreas at any post-Ppx time points (Fig. 2C and D). In contrast, strong nuclear and cytoplasmic Ngn3 staining of a subset of intestinal crypt epithelial cells was detected in both sham and Ppx mice (Fig. 2E–H), confirming the efficacy of the staining protocol in adults.

Administration of the glucagon-like peptide-1 receptor agonist exendin-4 stimulates pancreatic Pdx-1 expression in mice (19) and augments the regeneration of β-cell mass after 90% Ppx in rats and mice (20 and D.D.L., D.A.S., unpublished observations). Therefore, we next examined Ngn3 expression in mice that received exendin-4 (1 nmol/kg i.p. once daily) for 10 days after surgery. Tissues were harvested at the same time points as described above (0–10 days). Once again, we detected no pancreatic Ngn3 immunoreactivity at any time point (data not shown).

We considered the possibility that Ngn3 reactivation in this model occurs during a narrow temporal window that escaped detection due to the short half-life of Ngn3 protein and the spacing of the time points chosen for analysis. To address this issue, we used Ngn3^EGFP/+ mice in which one Ngn3 allele is replaced by EGFP (16). Due to the long half-life of EGFP in vivo, even transient activation of the Ngn3 locus would result in EGFP expression that is easily detectable for at least 48 h after gene activation. Thus, we can trace the fate of Ngn3-expressing cells regardless of the duration of Ngn3 gene transcription. Examination for EGFP expression both early (2 h) and later (2 and 4 days) after Ppx failed to detect EGFP signal in the pancreas (Fig. 3C–E), despite obvious expression during embryonic development (Fig. 3A and B) and in the intestine of adult Ppx animals (Fig. 3F and G).

Taken together, these data indicate that Ngn3 is not expressed in adult mouse pancreatic tissue and that its expression is not reactivated during islet regeneration after partial pancreatic resection. These findings complement recent lineage-tracing studies implicating replication from existing β-cells as the primary mechanism of β-cell renewal in adult mice (12). Although ectopic expression of Ngn3 can promote an endocrine transcriptional program in adult cells and may thus have therapeutic potential in the development of a surrogate β-cell source for transplantation, our studies indicate that a reactivation of endogenous Ngn3 expression is not required for adult β-cell regeneration in vivo.

This work was supported by a type 1 diabetes pilot grant from the Penn Diabetes Center (P30-DK-19525 to D.A.S. and K.H.K.), NIDDK (R01 DK55342 to K.H.K.), an American Diabetes Association Career Development Award (7-01-CD-15 to D.A.S.), and the Morphology Core of the Center for Molecular Studies of Liver and Digestive Diseases (P30 DK50306).

The authors acknowledge Biva M. Desai and Cyrus Farzad for outstanding technical assistance.