Characterization of β-cell plasticity mechanisms induced in mice by a transient source of exogenous insulin

Abstract

β-Cell plasticity governs the adjustment of β-cell mass and function to ensure normoglycemia. The study of how β-cell mass is controlled and the identification of alternative sources of β-cells are active fields of research. β-Cell plasticity has been implicated in numerous physiological and pathological conditions. We developed a mice model in which we induced major β-cell mass atrophy by implanting insulin pellets (IPI) for 7 or 10 days. The implants were then removed (IPR) to observe the timing and characteristics of β-cell regeneration in parallel to changes in glycemia. Following IPR, the endocrine mass was reduced by 60% at day 7 and by 75% at day 10, and transient hyperglycemia was observed, which resolved within 1 wk. Five days after IPR, enhanced β-cell proliferation and an increased frequency of small islets were observed in 7-day IPI mice. β-Cell mass was fully restored after an additional 2 days. For the 10-day IPI group, β-cell and endocrine mass were no longer significantly different from those of the control group at 2 wk post-IPR. Furthermore, real-time quantitative PCR analysis of endocrine structures isolated by laser capture microdissection indicated sequentially enhanced expression of the pancreatic transcription factors β₂/NeuroD and Pdx-1 post-IPR. Thus, our data suggest this mouse model of β-cell plasticity not only relies on replication but also involves enhanced cell differentiation plasticity.

glucose homeostasis is achieved via the adaptation of β-cell function, β-cell mass, or both (1, 17, 19, 20). Studies aimed at understanding how β-cell mass is regulated and at identifying alternative sources of β-cells have shown that, at least under normal conditions during postnatal life, the control of β-cell mass relies mainly on β-cell proliferation (11). However, the possibility that neogenesis and transdifferentiation occur under particular experimental conditions (in rodents) or pathological conditions (in humans) remains a matter of debate and investigation (25, 43).

Glucose homeostasis is disrupted in patients with type 2 diabetes. Obesity and insulin resistance, which are often present in patients who subsequently develop diabetes, are generally accompanied by an increase in β-cell mass in response to the increased demand for insulin (6, 23, 29, 47). However, this compensatory adaptation subsequently becomes inadequate and diabetes ensues (27, 33). Although β-cell mass is only slightly reduced at the time of disease onset, the functional β-cell mass is insufficient at this time (33). Changes in β-cell mass are also observed in conditions other than diabetes. It has been shown in rodents that β-cell mass increases as an adaptation to the progressive insulin resistance that develops during pregnancy (21, 42), and this process involves both β-cell hypertrophy and increased β-cell proliferation. After parturition, β-cell mass decreases and returns to a normal level via increased levels of apoptosis, decreased proliferation, and reduced β-cell size (22, 37). In humans, there is also an adaptative increase of β-cell number during pregnancy although not as large as in rodents (5); moreover, in a morphometric analysis conducted on human pancreas obtained at autopsies from women who died while pregnant, Butler and colleagues concluded that neogenesis was the major source of increase of β-cell mass in pregnant women, whereas β-cell replication was predominant in rodents (21).

In the last several years, our group has shown that β-cell plasticity occurs in several human diseases in addition to diabetes. So, in the focal form of persistent hyperinsulinemic hypoglycemia of infancy (FoPHHI), area of adenomatous hyperplasia with large islets bearing mutations in genes with well-characterized functions in β-cells and showing signs of insulin overproduction and hypersecretion have been observed (38, 41). In the same pancreata, shrunken islets with hypotrophic β-cells exhibiting low levels of proinsulin production are also observed, and these are considered to be in a resting state (39). When the pancreatic lobules bearing abnormal islets, i.e., area of adenomatous hyperplasia with hypersecreting islets, are surgically resected, transient hyperglycemia occurs for several days but is followed by normoglycemia, which is most likely achieved by the reactivation of resting islets (10). This finding suggests that the plasticity of normal β-cells is modulated by the pathological source of insulin. Except for the insulinoma implantations in rats, none of the animal models previously developed to study β-cell plasticity have addressed this issue. To gain insight into the mechanisms controlling β-cell plasticity in this situation, we developed a model based on the subcutaneous implantation of insulin pellets, which release insulin steadily and independently of the glycemic state. Major hypoglycemia occurred rapidly after implantation and was accompanied by progressive islet atrophy. When the source of exogenous insulin was removed, there was a short, transient period of hyperglycemia, after which β-cells regenerated and glycemia returned to normal levels.

At different time points following the withdrawal of the insulin pellets, we quantified the expression of several mRNAs encoding transcription factors using real-time PCR following laser capture microdissection (LCM) to specifically examine selected clusters of endocrine cells. These findings were then correlated with the cellular localization of the proteins using immunofluorescence.

These morphological, morphometrical, and molecular data were used to address whether, in this model of insular plasticity, the replication of differentiated β-cells is the sole factor involved in the regeneration of β-cell mass.

MATERIALS AND METHODS

This study was approved by the ethics committee of the University of Louvain and was conducted in accordance with the guidelines issued by the National Belgian Animal Care Committee.

Experiments were carried out on 3-mo-old female NMRI mice (25–35 g) that were provided free access to water and a standard laboratory diet.

For comparisons with the mouse model under investigation, human pancreatic samples from six cases of FoPHHI were chosen from a large series of previously reported cases based on the availability of the materials (35).

Mouse Insulin Implants

A schematic of the experimental design is shown in Fig. 1. Mice (n = 112) were anesthetized with 50 mg/kg ketamine and 5 mg/kg xylazine. Test animals (n = 71) were subcutaneously implanted with five insulin pellets (LinBit; LinShin Canada, Ontario, Canada), each of which released 0.1 IU bovine insulin/day. Controls (n = 41) received sham operations. The effects of continuous insulin release were studied in animals killed after 3, 7, and 10 days postinsulin pellet implantation (IPI) (n = 23).

Fig. 1.

Experimental design for the implantation of insulin pellets at day 0 and their removal at day 7 or day 10 postimplantation. Bold nos. along the timeline correspond to days of death for morphometry analysis. Italic nos. along the timeline correspond to days of death for molecular analysis.

The effects of insulin pellet removal (IPR) were studied in 7-days and in 10-days IPI mice. For the 7-day IPI group, animals were killed at days 2, 3, 5, and 7 post-IPR for histological evaluation (n = 16) or at 10 h and 1, 2, 3, and 5 days post-IPR for gene expression studies (n = 16). The mice used for histological evaluations received intraperitoneal injections of 5-bromo-2′-deoxyuridine (BrdU) at a dose of 60 mg/kg (Sigma Aldrich, St. Louis, MO) 6 h before death. The 10-day IPI group was killed at days 2, 3, 7, and 14 post-IPR for morphological evaluation (n = 16).

Blood glucose concentrations were measured throughout the experiment using a glucose analyzer (One Touch-R Ultra TM; LifeScan, Milpitas, CA).

Mice Monitoring

Mice were weighed daily, and their behavior was observed.

At the time of death, blood was collected through heart punction; serum was obtained by centrifugation and stored at −20°C.

Hormonal Assays

Determination of C-peptide and glucagon levels was made using commercial kits [C-Peptide Elisa kit (Alere, Waltham, MA) and Glucagon RIA Kit (EMD Millipore, Billerica, MA)] according to the manufacturer's recommendations. Analyses were realized on serum stored from blood collected at death.

Tissue Processing and Ultrastructural Analysis

After dissection, the pancreata were rapidly stripped of fat and weighed. For histology and morphometry, the whole pancreas was fixed for 24 h in 4% buffered paraformaldehyde and routinely embedded in paraffin. For ultrastructural studies, fragments of the splenic portion of the pancreas from mice killed 3 or 10 days post-IPI were fixed in 2.5% glutaraldehyde, postfixed in 1% osmium tetroxide, embedded in Epon 812, and examined on a Zeiss EM109 electron microscope. Finally, pancreata embedded in TissueTek OCT medium (Sakura Finetek, Torrance, CA) were frozen in chilled isopentane, stored at −80°C, and used for nucleic acid studies and double-immunofluorescence staining.

Immunohistochemical Staining

The characteristics of the primary and secondary antibodies and the detection systems are listed in Table 1.

Table 1.

The primary and secondary antibodies used

Primary Antibody and Host Species	Reference and Final Dilution	Antigen Retrieval	Detection System Reference and Dilution	Color of Visualization
Anti-Insulin
Mouse	Chemicon, Mab 107, 1/100	No	EnVision peroxidase complex (Dako, Carpintera) and DAB revelation (15a)	Brown
Guinea pig	Dako, A0564, 1/500	Yes*	Alexa 594, anti-guinea pig (A11076; Invitrogen), 1/1,000	Red
			Alexa 647 anti-guinea pig (706-605-148, Jackson ImmunoResearch) 1/1,000	Blue (transformed in red color)
Anti-Proinsulin
Mouse	Novo Biolabs, HPI-005, 1/1,000	Yes*	Alexa 594, anti-mouse (A21203, Invitrogen) 1/1,000	Red
Anti-Glucagon
Rabbit	Novocastra, NCL-GLUCp, 1/1,000	Yes*	EnVision G/2 System/AP, Rabbit/Mouse (Permanent Red) (Dako, Carpintera)	Red
Mouse	Abcam, ab10988, 1/1,000	Yes*	Alexa 488, anti-mouse (A21202; Invitrogen)	Green
Anti-BrdU
Mouse	Dako, M744, 1/50	No	EnVision peroxidase complex (Dako, Carpintera) and DAB revelation	Brown
Anti-Somatostatin
Rabbit	Neomarkers, Ab1, 1/150	Yes*	EnVision G/2 System/AP, Rabbit/Mouse (Permanent Red) (Dako, Carpintera)	Red
Anti-PDX-1
Rabbit	Abcam, ab47267 Gift from P. Jacquemin (UCL), 1/10,000	No	Alexa 594, anti-rabbit (A-11072; Invitrogen, Life Technologies, Carlsbad, CA), 1/400	Red
Goat	Abcam, ab47383, 1/1,000	Yes*	Alexa 488, anti-goat (A11055; Invitrogen), 1/1,000	Green
Anti-Nkx6.1
Rabbit	Gift from P. Jacquemin (UCL), 1/1,000	No	Alexa 594, anti-rabbit, (A-11072; Invitrogen), 1/200	Red
Anti-CK19
Rat	University of Iowa*, TromaIII, 1/50	Yes*	Alexa 488, anti-rat, A-11006 (Invitrogen), 1/200	Green
Anti-Nkx2.2
Mouse	University of Iowa*, 74.5A5, 1/50	Yes*	TSA kit with HRP-goat anti-mouse IgG and Alexa fluor 594 tyramide, T20915 (Invitrogen)	Red
Anti-Glut-2	Chemicon-Millipore, 07-1402
Rabbit	1/400	No	Alexa 594, anti-rabbit, A-11072 (Invitrogen) if combined with CK19, 1/200	Red
Rabbit	1/4,000	Yes*	Alexa 488, anti-rabbit, A-11008 (Invitrogen) with other antibodies, 1/200	Green
	Gift from Bernard Thorens, 1/100	Yes*	Alexa 594, anti-rabbit, A-11072 (Invitrogen), 1/1,000	Green

BrdU, 5-bromo-2′-deoxyuridine; CK19, cytokeratin 19; Glut-2, glucose transporter-2; DAB, diaminobenzidine; TSA, tyramide signaling amplification detection system; HRP, horseradish peroxidase.

^*Antigen retrieval: 0.01 M citrate buffer, pH 5.8, for 75 min at 95°C. The monoclonal anti-Nkx2.2 antibody developed by T. M. Jessel and S. Brenner-Morton and the monoclonal anti-CK19 antibody developed by R. Kemler were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by The University of Iowa, Department of Biology, Iowa City, IA.

Paraffin sections (5 μm thick) were immunostained as follows: with an anti-insulin antibody alone [diaminobenzidine (DAB), brown]; with anti-insulin (brown)/anti-glucagon and somatostatin cocktail (Fast Red); with anti-BrdU (brown)/anti-glucagon and somatostatin cocktail (Fast Red); or with anti-glucose transporter-2 (GLUT-2, brown)/anti-insulin (FITC).

We used GLUT-2 as a β-cell marker to localize the expression of certain transcription factors. For this purpose, double-immunofluorescence labeling was performed on 5-μm-thick frozen sections that had been fixed in acetone with 5% buffered paraformaldehyde 4% for 2 min at 4°C. Cytokeratin 19 (CK19) was used as an alternative ductal marker.

Sections were incubated with anti-Nkx2.2, anti-Pdx-1, or anti-CK19 antibodies overnight at room temperature, which was followed by incubation with the appropriate fluorescent secondary antibody (at room temperature for 1 h; Table 1). This step was followed by incubation with the anti-GLUT-2 or CK19 antibody overnight at 4°C in addition to a second specific fluorescent secondary antibody (Table 1). Slides were mounted with a fluorophore-preserving mounting medium (Vectashield 2/3: Vectashield with DAPI 1/3; Vector Laboratories, Burlingame, CA).

Similar double-immunofluorescence labeling methods were used to stain 5-μm-thick paraffin sections from FoPHHI pancreas samples.

Detection of Apoptosis

To label apoptotic cells, we had recourse to the “In situ Apoptosis Detection Kit” (Takara). The kit is designed to detect fragmented DNA histochemically by terminal labeling and is based on the TdT-dUTP nick end-labeling method, which uses terminal deoxynucleotidyl transferase to label 3′-OH ends of DNA fragments that are generated during the process of apoptosis. Incorporated fluorescein-labeled nucleotides were detected at light microscopy, using a peroxidase-labeled anti-fluorescein antibody, according to the manufacturer's recommendations.

Morphometric Analysis

β-Cell volume density and mass.

The relative area of endocrine tissue [β-cells and non-β-cells labeled with DAB (brown) and Fast Red deposits, respectively] vs. total pancreatic tissue was determined by planimetry using a Leitz microscope (magnification ×400) connected to a semiautomatic image analyzer (Videoplan; Kontron Bildanalyse, Eching, Germany), as previously reported (34). Two histological sections involving the entire pancreas and taken at intervals of 320 μm were measured in all mice. The endocrine mass was calculated by multiplying the endocrine cell relative area by the weight of the pancreas. The relative proportion of β-cells and non-β-cells was evaluated using the point-counting method described by Chalkley (7) with a 42-point ocular grid. β-Cell mass was calculated by multiplying this proportion by the absolute mass of endocrine cells, as previously reported (15).

Crowding.

β-Cell crowding (number of β-cell nuclei/1,000 μm² of β-cell area) (40) was used to evaluate changes in β-cell size. Crowding was evaluated for at least 10 islets/section for each mouse. An increase in crowding reflects a decrease in the size of the cells, and vice versa.

β-Cell replication.

The percentage of BrdU-labeled β-cell nuclei was assessed by counting the number of positive and negative nuclei. An average of 1,000 β-cell nuclei was evaluated for each group (one section/pancreas, n ≥ 3/group).

Distribution of islet profiles.

Islet profiles were measured by planimetry and sorted into the following three categories: <400, 401–50,000, and >50,000 μm². Quantification was performed on double-immunostained sections (anti-insulin and a cocktail of anti-glucagon and anti-somatostatin antisera, two entire sections/pancreas, n ≥ 5/group).

LCM and Real-Time Quantitative PCR

All microdissection procedures were performed within 15 min of collection to preserve RNA quality (28). These procedures included the staining of 7-μm-thick frozen sections (Histogene LCM Frozen Section Staining Kit; Arcturus) and islet microdissection using a 337-nm PALM MicroLaser System (Bernied). The samples were collected in 20 μl RNAqueous Micro Isolation solution (Ambion, Huntingdon, UK) and were stored at −80°C. A total of ∼180,000 μm² endocrine tissue (corresponding to 15–30 islet profiles) was dissected. Moreover, two entire 12-μm sections from each pancreas (without any staining and microdissection) were immediately collected in lysis solution (Ambion) and then frozen.

Total RNA was extracted using an RNAqueous Micro Isolation kit (Ambion), which involved treatment with DNase I (20 min at 37°C) and an RNase inhibitor (RNaseOUT, 20 U; Invitrogen, Life Technologies, Carlsbad, CA). For microdissected samples, the RT reaction was performed using 7 μl total RNA, Sensiscript reverse transcription (Qiagen, Benelux, Venlo, The Netherlands), oligo(dT)_12–18 (1 μM), and random primers (10 μM) (Invitrogen, Life Technologies). A 5-μl sample of the reverse transcription product was used for real-time quantitative PCR (RT qPCR).

For entire sections, the relative abundance and integrity of mRNA was determined using a NanoDrop spectrophotometer (Wilmington, DE) to measure the 260/280 nm optical density ratio (range: 2.0 and 2.1). Total RNA (500 ng) was reverse transcribed using oligo(dT) primers (250 ng), random primers (200 ng), and the Superscript II reverse transcription kit (Invitrogen, Life Technologies). A 12.5-ng sample of the reverse transcription product was used for RT qPCR.

All RT qPCR reactions were performed in duplicate with SYBR Green master mix (Bio-Rad, Hercules, CA) and 0.4 μM of each specific primer in an Opticon 2 Real-Time cycler system (Bio-Rad). Primers were designed in-house (Primer 3 software; Broad Institute) for mouse cyclophilin, proinsulin, glucagon, Nkx6.1, Nkx2.2, Pdx-1, β₂/NeuroD, and CK19 (Table 2). Primers for Ngn3 were provided by P. Jacquemin. The cycling conditions included a polymerase activation step (5 min at 95°C) and 40 cycles of amplification with temperature parameters depending on the primers (Table 2) and a single fluorescence measurement.

Table 2.

The primer sequences, amplicon lengths, and cycling conditions used

Gene mRNA	Sense Primer	Reverse Primer	Length of Amplicon, bp	Cycling Conditions
Cyclophilin	TTG GGA AGG TGA AAG AAG G	GAA GGA ATG GTT TGA TGG G	154	95°C for 30 s; 60°C for 1 min; 72°C for 1 min
Proinsulin	TCT TCT ACA CAC CCA TGT C	GGT GCA GCA CTG ATC TAC	149	95°C for 10 s; 56°C for 1 min; 84°C for 15 s
Glucagon	CAG AGG AGA ACC CCA GAT CA	GAC GTT TGG CAA TGT TGT TCC	199	95°C for 15 s; 60°C for 1 min
Nkx2.2	CGA GTC ACC GGA CAA TGA C	GTA GGT CTG CGC TTT GGA GA	97	95°C for 15 s; 60°C for 1 min
PDX-1	GAG GAC CCG TAC TGC CTA CA	TTC AAC ATC ACT GCC AGC TC	111	95°C for 15 s; 60°C for 1 min
B2/NeuroD	CTC GGA CTT TCT TGC CTG AG	TCA AAG AAG GGC TCC AGA GC	196	95°C for 15 s; 60°C for 1 min
CK19	TGT CGA CCT AGC CAA GAT CC	GAG CAA GGT AGG TGG CTT CA	95	95°C for 15 s; 60°C for 1 min
Ngn3	TCA CTG ACT GAC CTG CTG CT	TTG GAA CTG AGC ACT TCG TG	148	95°C for 30 s; 60°C for 1 min

Amplicon specificity was confirmed by melting-curve analysis as well as the molecular weights of the products, as evaluated by agarose gel electrophoresis. For all amplicons, the RT qPCR efficiency was between 1.95 and 2.04 (36). Following normalization of target gene expression to that of the housekeeping gene (cyclophilin), the relative changes in expression levels at each time point were calculated using the 2 − (δδC_t) method, where δδC_t = (C_t target − C_t cyclophilin)time x − (C_t target − C_t cyclophilin)time 0; time x corresponds to the time of death, i.e., 10 h or 1, 2, 3, or 5 days after the removal of the implants, whereas time 0 corresponds to the control sample. The results are shown as the means of 2 − (δδC_t) values ± SD (26).

Statistical Analysis

Numerical results are expressed as means ± SD, and these data were compared between time points using a one-way ANOVA. In cases of statistically significant heterogeneity, the ANOVA was followed by a Dunnett test to compare the variable parameter with the control value on each day. The substantial variability in the number of β-cell clusters in mice implanted with insulin pellets for 7 days (maximum-to-minimum ratio of 140) required the use of the Kruskal-Wallis nonparametric ANOVA followed by the Wilcoxon rank sum test. All tests were two-tailed and were performed using SPSS 15.0 (SPSS, Chicago, IL). The intraobserver variations for morphometrical parameters were ∼10%.

RESULTS

Severe hypoglycemia occurred during IPI, whereas apoptosis and atrophy significantly altered the islets. Mice behavior, however, did not change substantially.

Following the first day of IPI, blood glucose levels decreased for 7 days (27.8 ± 9.4 mg/dl) and then plateaued (29.7 ± 12.3 mg/dl at day 10) (Fig. 2). There was no significant difference between the weight of animals, despite the fact that the weight of the mice with implants was slightly higher, probably reflecting a greater food intake related to hypoglycemia [day 7 post-IPI: 31.5 ± 2.6 vs. 29.2 ± 2.3 g for control, not significant (NS); day 10 post-IPI: 32.2 ± 2.4 vs. 28.9 ± 2.6 g for control, NS; day 7 post-IPR in 7-days IPI mice: 29.3 ± 2.6 vs. 30.2 ± 2.7 g for control; day 14 post-IPR in 10-days IPI mice: 31.2 ± 2.8 vs. 31.2 ± 2.6 g for control, NS]. Even highly hypoglycemic mice were not apathetic and did move in a way similar to that of controls. Hormonal dosages (C-peptide and glucagon) did not reveal significant differences between groups in our particular groups of mice, even if at day 3 post-IPI a trend toward a decrease in C-peptide was observed. Values remained low, whatever the group. For C-peptide dosages, mean values and SD were, respectively: 186.4 ± 31.9 pM for controls (n = 11); 83.7 pM for day 3 post-IPI (n = 1); 150.7 ± 42.8 pM for day 7 post-IPI (n = 11); 176.6 ± 51.7 pM for day 10 post-IPI (n = 7); 184.9 ± 93.6 pM for day 1 post-IPR in 7-day IPI mice (n = 3); 182.8 ± 20.6 pM for day 2 post-IPR in 7-day IPI mice (n = 4); 132.4 ± 41.9 pM for day 3 post-IPR in 7-day IPI mice (n = 3); 175.9 ± 42 pM for day 5 post-IPR in 7-day IPI mice (n = 4); 181.9 ± 60 pM for day 7 post-IPR in 7-day IPI mice (n = 6); 203.4 ± 91.6 pM for day 2 post-IPR in 10-day IPI mice (n = 3); 222.3 ± 34.6 pM for day 3 post-IPR in 10-day IPI mice (n = 3); and 195.2 ± 52.3 pM for day 7 post-IPR in 10-day IPI mice (n = 7). For glucagon, results (mean) were, respectively: 92.3 ± 28.1 pg/ml for controls (n = 11); 98.1 pg/ml for day 3 post-IPI (n = 1); 70.1 ± 19 pg/ml for day 7 post-IPI (n = 9); 77.1 ± 22 pg/ml for day 10 post-IPI (n = 7); 90.3 pg/ml for day 1 post-IPR in 7-day IPI mice (n = 1); 69.2 ± 15.4 pg/ml for day 2 post-IPR in 7-day IPI mice (n = 4); 72.4 ± 8.1 pg/ml for day 3 post-IPR in 7-day IPI mice (n = 2); 82.4 ± 17.8 pg/ml for day 5 post-IPR in 7-day IPI mice (n = 4); 72.5 ± 15 pg/ml for day 7 post-IPR in 7-day IPI mice (n = 6); 64.7 ± 11.6 pg/ml for day 2 post-IPR in 10-day IPI mice (n = 4); 95.2 ± 53.1 pg/ml for day 3 post-IPR in 10-day IPI mice (n = 3); and 85 ± 25 pg/ml for day 7 post-IPR in 10-day IPI mice (n = 6).

Fig. 2.

Evolution of glycemia during the experiment. IPI, insulin pellet implant; D, day.

Although the animals became hypoglycemic at day 3 post-IPI, their islets retained normal features (Fig. 3A), such as the absence of apoptosis. However, at day 7 post-IPI, islets had a disturbed appearance with reduced size and more prominent α- and δ-cells (Fig. 3B). Moreover, apoptosis was detected in most islets (Fig. 3C) with the In situ Apoptosis Detection Kit. At day 10 post-IPI, most islets had become very small, were less numerous, and essentially constituted by α- and δ-cells (Fig. 3D), and few apoptotic bodies were present (data not shown). The central portions of the islets consisted of shrunken β-cells, and the β-cell nuclear crowding indicated an average 50% reduction in β-cell size (P = 0.005 vs. control values) (Table 3). Insular lymphocytic infiltration was not observed at any time point. At days 3, 7, and 10 following pellet implantation, the endocrine mass had decreased by 50, 60 (P = 0.048), and 75% (P = 0.008), respectively (Fig. 4A). During this period, the β-cell mass decreased in parallel (P = 0.015 and P = 0.002 at day 7 and day 10, respectively), whereas the mass of non-β-cells remained unchanged (P > 0.86) (Fig. 4, D and G).

Fig. 3.

Morphology of the islets of Langerhans. On day 3 post-IPI (A), islet morphology was normal, whereas at 7 days post-IPI (B), the proportion of β-cells was reduced. In Situ Apoptosis Detection Kit revealed apoptosis in day-7 IPI mice (brown deposit, C). At day 10, islets of Langerhans appeared almost devoid of insulin-producing cells (D). β-Cell proliferation peaked 5 days following insulin pellet removal (IPR) in 7-day IPI mice (E), and normal proportions of endocrine cells were restored (F). A, B, D, and F: immunohistochemical staining with insulin is shown in brown, and staining with the glucagon-somatostatin cocktail is shown in red. C: In Situ Apoptosis Detection Kit (Takara) to detect apoptotic cells. E: nuclear immunohistochemical staining with 5-bromo-2′-deoxyuridine (BrdU) is shown in brown, and insulin staining is shown in red. Scale bars = 50 μm.

Table 3.

Crowding measurements

	Mice Implanted with Insulin Pellets				Days after the Removal of Insulin Pellets in Mice That Had Kept Implants for 7 Days				Days after the Removal of Insulin Pellets in Mice That Had Kept Implants for 10 Days
	Control	Day 2	Day 3	Day 7	Day 14	Day 3	Day 5	Day 7	Day 2	Day 3	Day 7	Day 14
Mean, %	100	147.6	125.4	103.0	109.5	122.2	117.5	90.5	147.6	125.4	103.0	109.5
SD	20.6	19	17.5	14.3	14	17.5	12.7	19	19	17.5	14.3	14

The crowding was expressed as a percentage of the control value (controls were normalized to 100%). The implantation of insulin pellets induced a drastic reduction in β-cell size, although this size returned to normal 7 days after insulin pellet removal.

Fig. 4.

Endocrine (A–C), β-cell (D–E), and non-β-cell (G–I) masses. A: the endocrine mass had began to decrease at day 3 and was reduced by 60% at day 7 and 75% at day 10. Similarly, a profound drop in β-cell mass (D) was observed at day 7 and day 10 in animals implanted with insulin pellets. Seven days following the withdrawal of the implants, both endocrine mass (B) and β-cell mass (E) were restored. In mice that had received insulin for 10 days, the endocrine mass (C) and the β-cell mass (F) were not restored until 14 days post-IPR. The non-β-cell mass remained unchanged throughout the experimental procedure (G–I). ^✱P = 0.048 (A), ^●P = 0.015 (D), ^✱✱P = 0.008 (A) and ^●●P = 0.002 (D).

IPR Induced Marked but Transient Hyperglycemia before a Complete Recovery of β-Cell Mass was Achieved

The subsequent removal of the insulin pellets (IPR) induced marked hyperglycemia that began at day 1 post-IPR. This was followed by a progressive normalization of glycemia (Fig. 2). The delay in the return to normal values was 2 days longer in mice implanted for 10 days compared with those implanted for 7 days. At 4 days post-IPR, the islets remained sparse and contained cellular debris. At day 5 post-IPR, the general features of the islets and their endocrine cell proportions had returned to normal (Fig. 3F). When the insulin pellets were removed, β-cell size returned to normal, regardless of the duration of implantation (Table 3). Both total endocrine and β-cell mass progressively returned to normal values, and this rebound was more rapid in 7-day-implanted than 10-day-implanted mice (7 days post-IPR in 7-day IPI mice: P = 1.000 both for endocrine and β-cell mass vs. control; 14 days post-IPR in 10-day IPI mice: P = 0.894 for endocrine mass vs. control and P = 0.887 for β-cell mass vs. control) (Fig. 4, B, C, E, and F). Similar to IPI of any duration, IPR had no effect on non-β-cell mass (Fig. 4, H and I).

After IPR, There was a Peak in β-Cell Proliferation, and Islet Size Distribution Showed a Higher Frequency of Small Endocrine Cell Clusters

Following IPR, the replication rate of β-cells increased progressively from 0.5% at day 3 post-IPR to 3.0% at day 5 post-IPR (P = 0.03; Figs. 3E and 5). At day 5 post-IPR, the islet size distribution of 7-days IPI mice corresponded to a higher frequency of small endocrine cell clusters (<400 μm²) (P = 0.041; Fig. 6B) that were dispersed throughout the parenchyma, in the vicinity of the ducts or among the acini. For the 10-days IPI mice, the number of small endocrine cell clusters also increased 14 days post-IPR, but the value did not reach significance at that time (Fig. 6C). Proliferation rate was identical whatever the size of the islet (data not shown).

Fig. 5.

A peak in β-cell proliferation occurred 5 days after IPR (P = 0.030) before values returned to normal 2 days later. The straight bars indicate the means.

Fig. 6.

Distribution of islet profiles. Small clusters (A–C): a drastic increase in small clusters was observed 5 days after IPR in 7-day IPI mice (P = 0.041; B). For the 10-days IPI mice, the no. of small endocrine cell clusters also increased 14 days post-IPR, but the value did not reach significance at that time. Large islets (D–F): in the initial days after IPR, the proportion of these islets was lower than that observed in controls (D). In 7-day IPI mice, the proportion of these very large islets had returned to normal by 7 days post-IPR (E), whereas in 10-day IPI mice, the number of large islets remained fewer at 7 and 14 days after IPR than that observed in the control group (D) (P = 0.015; F). ^✱P = 0.041 and ^✱✱P = 0.003.

After IPR, the number of large islets (50,001–400,000 μm²) was ∼50% lower in treated mice than in controls. However, these numbers returned to normal within 7 days post-IPR in the 7-days IPI mice (Fig. 6E) but remained reduced until the end of the experiment in the 10-days IPI mice (P = 0.015) (Fig. 6F).

In 10-days IPI mice, cells remained unstained, either with anti-insulin antibody or glucagon or somatostatin antibodies within the islets (Fig. 7A). This suggested the existence of β-cells that had lost insulin expression. By coimmunofluorescent labeling with insulin and GLUT-2 taken as a β-cell marker, we confirmed this hypothesis: some islet cells showed weak expression of GLUT-2 and absence of insulin expression (Fig. 7B). Total or near-total absence of endocrine granules in these cells was confirmed by electron microscopy (Fig. 7E). Such cells were never observed in control mice: in these mice, insulin granules were easily found (Fig. 7D). Before pellet removal, proinsulin expression was found in a weak proportion of islet cells. Surprisingly, when colabeling for proinsulin and GLUT-2 was performed, we observed that cells expressing proinsulin did not express GLUT-2 and vice versa (Fig. 7C). GLUT-2 expression was disturbed, both before and after pellet removal, contrary to the situation observed in control islets where GLUT-2 expression was more homogenous (Fig. 8A). At days 7 and 10 post-IPI, the proportion of β-cells expressing GLUT-2 decreased. Several β-cells still expressed proinsulin while they became GLUT-2 negative (Fig. 8B). At day 2 post-IPR, this situation persisted, with a particularly weak expression of GLUT-2 (Fig. 9, B and E). Moreover, in 10-days IPI mice, we also observed, 2 days after IPR, cells expressing proinsulin, in the absence of GLUT-2; these cells were situated in the close vicinity of ducts (Fig. 8C). Later on, gradual increased expression of GLUT-2 was progressively restricted to the cytoplasmic membranes, similarly to controls (Fig. 8A).

Fig. 7.

Islet cells without any immunostaining in 10-days IPI mice. In day 10 IPI mice, cells remained unstained, either with anti-insulin antibody or glucagon or somatostatin antibodies (arrow) in islets (A). Coimmunofluorescent labeling with insulin (red) and glucose transporter-2 (GLUT-2, green) showed the absence of insulin expression in islet cells with weak expression of GLUT-2 (arrows) (B). Colabeling for proinsulin (red) and GLUT-2 (green) in mice before pellet removal revealed cells expressing proinsulin but not GLUT-2 and inversely (C). Electron microscopy revealed a drastic decrease in the no. of insulin-secreting granules in some cells (E). Inset: characteristic insulin granules (dense core and clear halo). In control mice, β-cells contain numerous insulin granules (D). Scale bar for A and B = 50 μm; for C = 10 μm; and for D and E = 2.6 μm.

Fig. 8.

Immunolocalization of proinsulin and GLUT-2 (A–C); immunolocalization of insulin and glucagon (D). Control mice show a characteristic coexpression of GLUT-2 (green) and proinsulin (red) (A), whereas, in 10-days IPI mice, GLUT-2 expression is heterogenous (B): some cells expressing proinsulin are devoid of GLUT-2 staining (GLUT-2 revealed in green; proinsulin revealed in red). Two days after IPR, in 10-days IPI mice, cells expressing proinsulin (red), in the absence of GLUT-2, were observed in close vicinity of a duct (C). Colocalization of glucagon (green) and insulin (red) was never observed here, 5 days after IPR, in 7-days IPI mice (D). Scale bar for A, B, and D: 20 μm; and for C: 10 μm. Solid arrows indicate cells expressing proinsulin (red) in the absence of GLUT-2 located in the close vicinity of a duct. Open arrows indicate isolated cells expressing proinsulin (red) in the absence of GLUT-2 within the exocrine parenchyma.

Fig. 9.

Cellular localization of transcription factors. In control pancreas, GLUT-2 expression was restricted to β-cell with an almost exclusive membranous pattern (green staining; A and D). No significant modification in the expression pattern of Nkx2.2 (nuclear red staining) was observed after IPR (B and C). At day 3 post-IPR in 7-days IPI mice, nuclear expression of Pdx-1 (nuclear red staining) (F) in GLUT-2-positive cells (green) became evident, whereas almost no staining could be detected the days before (E) in cells expressing GLUT-2. In the focal form of persistent hyperinsulinemic hypoglycemia of infancy, expression of Pdx-1 (nuclear green) is more pronounced in adenomatous islets (I) than in islets outside of tumor (H) (insulin in red; H and I). In control pancreas as in IPR mice, cytokeratin 19 expression (green) was only observed in ducts, without any colocalization with GLUT-2 (red)-positive islet cells (G). Scale bar for A–F and H–I: 20 μm; scale bar for G: 40 μm.

A Sequential Increase in the Expression of Specific Pancreatic Transcription Factors Occurred Post-IPR

The use of LCM allowed us to dissect selected islets and endocrine clusters while avoiding contamination from other nonendocrine cell types. Such experiments were performed after IPR. To validate our dissections, we first measured insulin and glucagon mRNA levels in the dissected islets. The levels of insulin mRNA expression remained very low post-IPR (98% of decrease at 10 h and 1 day post-IPR compared with controls, P < 0.001) (Fig. 10A). This result confirmed our immunofluorescent data described above. Conversely, at day 3 post-IPR, the expression of the insulin gene tended to be higher than that measured in controls (2.3×) but subsequently returned to normal levels at day 5 post-IPR (Fig. 10A). The expression of glucagon mRNA remained stable throughout the experiment (Fig. 10B). These results confirmed that the endocrine clusters were appropriately dissected.

Fig. 10.

Fold change in transcription factor mRNA expression following IPR. mRNA levels of insulin, glucagon, β₂/NeuroD, Pdx-1, and Nkx2.2 were measured by real-time PCR. The relative changes in expression levels at each time point relative to the controls were calculated using the 2 − (δδC_t) method. The level of insulin mRNA expression (A) remained low during the initial days post-IPR (98% lower than in controls; P < 0.001 at 10 h and 1 day post-IPR), but, at 3 days post-IPR, the expression of the insulin gene tended to be elevated (2.3× the control value) but subsequently returned to baseline. The expression of glucagon mRNA (B) remained stable throughout the experiment. Precociously enhanced expression of β₂NeuroD (C) was observed 10 h after retrieval, compared with the controls, and its expression remained elevated afterward. Pdx-1 mRNA expression (D) was increased 2.3-fold compared with the controls at 2 days post-IPR. The level of Nkx2.2 mRNA expression (E) was increased by almost 2-fold compared with the controls at 5 days post-IPR, without statistical significance. The data are presented as the median logarithms of numerical results. °°°P < 0.001 (A), °P = 0.013 (C), and °°P = 0.012 (D) at day 3.

At 10 h post-IPR, at a time where mice were already hyperglycemic (191.75 ± 24.44 mg/dl measured 7 h after IPR), the expression level of β₂/NeuroD mRNA was increased by 3.66-fold compared with controls (P = 0.013) (Fig. 10C), and this expression level remained elevated after IPR with large individual variations. However, at subsequent time points, the expression level of β₂/NeuroD mRNA was not different between treated and control animals.

At day 2 post-IPR, the Pdx-1 mRNA expression level was increased by 2.3-fold compared with controls (P = 0.012) (Fig. 10D) and remained upregulated at 3 (1.7-fold increased compared with control; NS) and 5 (1.8-fold increased compared with controls: P = 0.052) days after IPR. Pdx-1 mRNA expression was not detected in exocrine tissue.

At day 5 post-IPR, the Nkx2.2 mRNA expression level was increased by almost twofold compared with controls (Fig. 10E), but this difference did not reach statistical significance. The expression of Ngn3 mRNA was never detected during the regenerative process in this experiment (data not shown). Whereas virtually no expression of Pdx-1 was detected in islets during the first 2 days post-IPR, in particular in cells expressing GLUT-2 (Fig. 9E), a significant increase of its expression occurred at day 3 post-IPR. The nuclear expression of Pdx-1 was evident in cells characterized by a membranous expression of GLUT-2 (Fig. 9F). Pdx-1 expression was never detected outside islets. At the times of death chosen, no significant difference in the expression of Nkx2.2 was observed in immunocytochemistry (Fig. 9, A–C). During the restoration of the β-cell mass, after IPR, colocalization of insulin and glucagon in the same cell was never detected (Fig. 8D). After pellet removal, few β-cells expressing proinsulin as small dots were observed close to the duct wall, evoking β-cell neogenesis during embryonic stages. This observation suggests that these cells, which did not express GLUT-2, could be neogenic β-cells generated from cells located in ducts (Fig. 8C).

In human segmental resected pancreata from FoPHHI, the β-cells in the abnormal hypersecreting area had abundant cytoplasm, and some β-cell nuclei were abnormally large as previously reported (39). Outside the abnormal focus, the islets were small, β-cells had shrunken cytoplasm, and the β-cell nuclear crowding was high (as in mouse islets during IPI). The expression of Pdx-1 was more intense in the hyperactive β-cells from the focal lesion (as in hyperactive mouse islets after IPR) than in the resting β-cells outside the lesion (Fig. 9, H and I).

DISCUSSION

In the current study, we developed and characterized an original mouse model of β-cell plasticity that enabled the study of β-cell adaptation in the face of metabolic perturbation. Our experimental device induced profound β-cell atrophy due to insulin implantation, and, following the withdrawal of the implants, there was a rapid regeneration characterized by the restoration of functional capacities, i.e., insulin synthesis and secretion, the normalization of glycemia, and the restitution of the β-cell mass. The initial decrease in β-cell mass closely paralleled that of glucose, and, from day 7 onward, the β-cell mass continued to decrease although the degree of hypoglycemia remained unaltered. Thus, in addition to the level of glycemia, the persistence of hypoglycemia determined the degree of β-cell mass reduction. This reduction was caused by both apoptosis and β-cell atrophy. Interestingly, during this period of β-cell atrophy and disrupted islet architecture, Pdx-1 expression was low; this observation is consistent with the link established between Pdx-1 and the maintenance of β-cell mass in adulthood, notably through the regulation of apoptosis (18). The occurrence of transient hyperglycemia following IPR is not surprising because the residual β-cell mass was still very low at this time. However, it is interesting to note that β-cell function was restored 2 days before β-cell mass, which suggests that functional plasticity arises more rapidly than anatomical plasticity. The regeneration process is initiated in the context of hyperglycemia, which is quite similar to that observed following experimental subtotal (90%) pancreatectomy; in the latter, nevertheless, β-cell mass was not fully restored (3). Moreover, the time to regenerate and to reach to control values, in our model, was short, which is, from a practical point of view, an advantage in the course of experiments.

In normal rats, hyperglycemia has been shown to increase β-cell mass (2), regardless of the blood insulin level (31). In these studies, however, this increase relied on an intact β-cell mass, whereas, in the present study, β-cell mass was restored during a period of marked hyperglycemia despite a very low number of residual β-cells. Thus, neither severe hyperglycemia nor a small number of residual β-cells prevented β-cell regeneration. The peak of β-cell regeneration typically occurred at day 5 and was concomitant with the normalization of glycemia, as previously shown by others (16).

The IPI model reproduces some features of β-cell plasticity that we previously described in focal forms of congenital hyperinsulinism (38, 39). In this rare disease, the excessive and uncontrolled release of insulin by abnormal hypersecreting islets (with β-cells bearing mutations in the ABCC8 and KCNJ11 genes) induces the resting and atrophy of normal islets throughout the pancreas. Surgical segmental resection restricted to the focus with unregulated hyperplasic islets cures the patient after a transient period of hyperglycemia. Similarly, the excessive release of insulin in our insulin implant model induces resting of the islets, increases β-cell crowding, and decreases islet size, which lead to a decreased β-cell mass. Similar to infants who have received segmental surgery, the suppression of exogenous insulin in these IPI mice induces a transient diabetic state that precedes a complete recovery.

Many experimental models have been used to study the β-cell plasticity. For example, solid insulin-producing tumors have been transplanted in rats to induce severe hypoglycemia and insular atrophy (9). Next, removal of the tumor cells results in a regeneration process that is associated with the normalization of blood glucose levels. However, the present model of IPI has several major advantages compared with the “insulinoma model”; first, the current model is highly standardized, controlled, and reproducible; second, the delay in the induction of hypoglycemia is short; and third, there is no confounding effect due to factors potentially secreted by the tumor. Indeed, despite the presence of a capsule of connective tissue surrounding the insulinomas, direct tissue invasiveness or contamination of the area during resection cannot be excluded (13), and this could interfere with β-cell mass plasticity.

Partial pancreatectomy and duct ligation models have long been used to study β-cell plasticity, but these models do not specifically target the β-cell compartment, unlike the more selective IPI model. Moreover, the absence of pancreatitis and insulitis in this model reduces the potential interference of inflammatory factors. Streptozotocin injection is also a well-known alternative method for affecting the β-cell compartment, since this treatment induces diabetes secondary to β-cell necrosis (24) but can also cause liver and renal damage (4).

In contrast, the IPI model relies on physiological adaptations, such as hypotrophy and apoptosis. The factors responsible for this insular atrophy in the context of excessive insulin release have been studied previously (8, 31), and severe hypoglycemia, rather than an excess of insulin per se, has been shown to induce islet atrophy. Consistent with this concept, we observed that β-cell mass continued to decrease as long as hypoglycemia was present, whereas the levels of insulin remained constant.

The transcriptional profile of LCM-isolated endocrine structures revealed the sequentially enhanced expression of β₂-NeuroD followed by Pdx-1. The activation of these transcription factors, which are known to be associated with islet cell development (14, 32), may suggest that the dynamic modifications of β-cell mass observed in our model also rely on an enhanced plasticity of cell differentiation, as previously reported by Li et al. in a model of partial pancreatectomy in adult rats (25). Moreover, markers such as Pdx-1 or β₂-NeuroD can be detected in mature endocrine islet cells, and some authors have proposed that the upregulated expression of transcription factors at specific times of the regenerative process signifies neogenesis in the adult pancreas (44, 46). In our model, however, Ngn3 mRNA and Ngn3 protein expression were not observed in LCM products by RT qPCR or immunohistochemistry, respectively.

The origin of new β-cells for maintenance or regeneration remains a subject of active research and debate, and this occurrence seems widely dependent on experimental conditions (11, 46). The first potential source of new β-cells is via the replication of preexisting β-cells (11), which has been shown to be the predominant mechanism for adult β-cell mass expansion. The second source, which is typically referred to as neogenesis, involves the expansion and redifferentiation of less or dedifferentiated progenitor-like cells into mature endocrine cells, which mimics normal embryonic development, and this process has been observed both in vivo (16,30) and in vitro (10). Third, it is possible that stem cells could be bypassed by the direct conversion of differentiated cells into β-cells, which corresponds to a recently described mechanism called reprogramming or transdifferentiation (12, 43). In our model, at 5 days post-IPR in 7-day IPI mice, a peak of β-cell proliferation was measured within the islets, which represented the self-replication of β-cells. In addition to the increased rate of proliferation, the number of small endocrine clusters increased significantly throughout the parenchyma, at the vicinity of the ducts or within the acini. It is unlikely that these small clusters originated from islets that were disrupted during IPI because they were not present at day 7 post-IPI or during the initial days post-IPR. In addition, the observation of cells remaining unstained in the core of atrophic islets at day 10 post-IPI regardless of the sensitivity of the immunohistochemical detection system used is also of interest. The ultrastructural analyses confirmed the total or nearly total absence of secretory granules in these cells, which may have corresponded to the presence of less-differentiated, progenitor-like cells. This hypothesis was reinforced by the detection of cells expressing the GLUT-2 but not insulin, and these types of cells have been previously described as potential β-cell precursors (16, 30, 45). However, we also observed cells expressing faint proinsulin staining and no GLUT-2 staining in close proximity to ducts, and it is tempting to speculate that this could represent the transdifferentiation of duct cells into β-cells. However, this observation could also correspond to remodeling of the regenerating parenchyma, which leads to the juxtaposition of immature β-cells and ducts.

Altogether, the rapid β-cell mass recovery and the morphological particularities observed here suggest that, in addition to enhanced replication, an additional regeneration pathway could contribute to the complete recovery of the β-cell mass, and this mechanism seems to involve an enhanced plasticity of β-cell differentiation.

In conclusion, we developed an in vivo mouse model of insular plasticity that first generates a dramatic reduction in the β-cell population and then drives a rapid increase in the endocrine pancreatic β-cell mass and functional recovery, which enables metabolic demands to be met within 1 wk of treatment. Although mass and function are two closely related parameters, changes in mass and function do not always change in parallel over time. In our model, regeneration relied mostly on the enhanced replication of β-cells. However, intriguing morphological data do not exclude the adjunction of another repair mechanism involving the plasticity of β-cell differentiation, which has been termed neogenesis in the literature (25). The model proposed herein could be used to investigate further questions in the field of β-cell plasticity and may be useful for molecular studies aimed at improving our understanding of the processes that drive the regeneration of the β-cell mass in a well-defined in vivo model.

GRANTS

This work was supported by Grant 3.4.616.05.F F.R.S.M from the Fonds National de la Recherche Scientifique Médicale (Belgium).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: M.-C.N., J.R., and Y.G. conception and design of research; M.-C.N., J.M., P.T., S.G., G.V.d.S., and J.J. performed experiments; M.-C.N., J.R., J.J., and Y.G. analyzed data; M.-C.N., J.R., and Y.G. interpreted results of experiments; M.-C.N., P.J., and Y.G. prepared figures; M.-C.N. drafted manuscript; M.-C.N., J.R., J.M., P.T., C.S., P.J., and Y.G. edited and revised manuscript; M.-C.N., J.R., J.M., P.T., S.G., G.V.d.S., J.J., C.S., P.J., and Y.G. approved final version of manuscript.