β-Cell Control of Insulin Production During Starvation-Refeeding in Male Rats

Abstract

Mammalian metabolism has evolved to adapt to changes in nutrient status. Insulin, the key anabolic hormone, facilitates intracellular storage of nutrient fuels and plays a pivotal role in the transition away from catabolism upon refeeding. Although circulating insulin relative to nutrient levels has been well characterized during fasting and refeeding, how pancreatic β-cell biology caters to acute changes in insulin demand has not been sufficiently addressed. Here, we examined the dynamics of (pro)insulin production and associated changes in β-cell ultrastructure during refeeding after a 72-hour fast in male rats. We found that fasted β-cells had marked degranulation, which inversely coordinated with the upregulation of autophagolysomal and lysosomal organelles. There was also expanded Golgi that correlated with enhanced (pro)insulin biosynthetic capacity but, conversely, blunted in vivo insulin secretion. Within 4 to 6 hours of refeeding, proinsulin biosynthesis, cellular ultrastructure, in vivo insulin secretion, and glucose tolerance normalized to levels near those of fed control animals, indicating a rapid replenishment of normal insulin secretory capacity. Thus, during a prolonged fast, the β-cell protects against hypoglycemia by markedly reducing insulin secretory capacity in vivo but is simultaneously poised to efficiently increase (pro)insulin production upon refeeding to effectively return normal insulin secretory capacity within hours.

Rapid β-cell ultrastructural changes parallel increased (pro)insulin biosynthesis to meet glycemic demand after fasting-refeeding in rats.

The hunter-gatherer lifestyle during the late Paleolithic era (bc 100,000 to 50,000) drove the selection of the modern human genome (1). In contrast to the current availability of cheap macronutrient-rich foods in the Western world, early humans thrived in an environment characterized by periods of fasting and feasting. Critical to our survival as a species was the selection of regulatory metabolic genes that prioritize nutrient intake and storage to protect against starvation (2). The endocrine system remains wired to quickly adapt to homeostatic perturbations and, even at a cellular level, the pancreatic β-cell retains this design.

For instance, inappropriate insulin secretion during starvation could lead to deleterious hypoglycemia. It is thus not surprising that pancreatic insulin content (3) and circulating insulin levels are reduced during starvation but can quickly rebound after refeeding (4). After a prolonged fast, islet preproinsulin messenger RNA (mRNA) (5) and insulin secretion (6, 7) have been observed to decrease but rapidly recover after refeeding. However, the cellular adaptations underlying this acute transition have not been studied. How β-cells might adapt during starvation-refeeding was hinted at by early ultrastructural studies. Rats subjected to prolonged starvation of 12 days did not display β-cell injury, but insulin stores were markedly reduced. However, after ad libitum refeeding, they were able to fully replenish their insulin stores (8). Guinea pigs that were subjected to intermittent overnight fasts followed by ad libitum refeeding displayed a larger β-granule population compared with free-fed controls (9). Although these qualitative ultrastructural studies provided important clues about the underlying cause of the glucose intolerance observed during refeeding, they do not explain the rapid ability of the β-cell to manage refeeding-induced hyperglycemia, particularly at the level of proinsulin biosynthesis.

Another example of β-cell insulin production adapting to acute metabolic change can be found in insulinoma transplant experiments in rats, where chronic hypoglycemia is accompanied by almost complete degranulation of endogenous β-cells. Upon insulinoma excision, β-cells regranulate and “reappear” at a rate too rapid to be explained by atrophy and subsequent regeneration (10). This flexibility for insulin production is further emphasized by ex vivo studies of diabetic mouse islets. KS^db/db mice exhibit profound fasting hyperinsulinemia, and their islets are markedly degranulated due to persistently high (pro)insulin production and secretion. However, upon overnight recovery in euglyemic (5.6 mM) media, these isolated islets recover normal granularity and proinsulin production (11). It is unclear how the β-cell manages the glycemic shift at the fasting-refeeding transition so ably because neither proinsulin biosynthesis nor secretory function has been directly measured in parallel in islets from fasted-refed animals.

In this study, Wistar rats were subjected to a prolonged fast (72 hours) and subsequently refed. We found that, in accordance with previous studies, circulating insulin levels during the fast were low and remained diminished after 6 hours of refeeding despite hyperglycemia. After glucose challenge, fasted rats displayed diminished circulating insulin levels coupled with significantly elevated glucose excursions. In contrast, the ex vivo secretory response was reduced by only 15% in freshly isolated islets from fasted vs refed animals, even though quantitative electron microscopy revealed profound insulin degranulation in fasted islets in parallel to an expanded Golgi apparatus. The rate of glucose-stimulated proinsulin biosynthesis was over twofold greater in fasted vs ad libitum fed control rat islets. Thus, it appears that the β-cells of fasted rats are poised to rapidly produce insulin upon refeeding. Our findings highlight the ability of the β-cell to promptly fine-tune insulin production and secretion to transition from a protective state during fasting to one that meets the acute physiological metabolic demand presented by refeeding.

Materials and Methods

Animals and study design

Male Wistar rats were purchased from Charles River Laboratories (Wilmington, MA) and housed two per cage in controlled conditions (12 hours light, 12 hours dark) with ad libitum access to chow and water until 11 weeks of age. At 12 weeks of age, rats either continued to have ad libitum access to chow (control) or were fasted for 72 hours and then refed for up to 60 hours (fasted-refed). Animal care and use and experimental protocols were approved by the Institutional Animal and Use Committee of the University of Chicago.

Analysis of circulating factors

The blood glucose levels of rats during the experimental period were assessed via tail vein blood using a Freestyle Lite glucometer (Abbott, Alameda, CA). Tail vein aliquots of blood (∼200 μL) were added to capillary blood collection tubes for serum collection (Microvette; Sarstedt, Nümbrecht, Germany) and then centrifuged for 5 minutes at 10,000g at 4°C to obtain serum, which was immediately frozen at −80°C until further analysis. Serum insulin and proinsulin levels were determined with rat ultrasensitive insulin and rat proinsulin enzyme-linked immunosorbent assay (ELISA) kits, respectively (ALPCO, Salem, NH).

Glucose tolerance test

An intraperitoneal injection of 1 g/kg of glucose was performed as described (12). Briefly, the blood glucose levels of rats were determined from tail vain blood with a FreeStyle Lite glucometer at time 0. Rats were then injected with 1 g/kg of glucose in sterile saline, and blood glucose was determined at 15, 30, 60, 120, and 180 minutes. At the same time points, blood samples were collected, and serum insulin levels were measured by ELISA as outlined previously.

Primary islet isolation

Rat pancreatic islets were isolated from rats by collagenase digestion as previously described (12). Experiments were performed in fresh islets immediately after isolation.

Analysis of in vitro islet proinsulin biosynthesis and total protein synthesis

We directly assessed proinsulin biosynthesis in freshly isolated islets by pulse-radiolabeling followed by (pro)insulin immunoprecipitation with guinea pig anti-insulin antibody [Research Resource Identifier (RRID): AB_2126544; Millipore, Billerica, MA] and alkaline-urea polyacrylamide gel electrophoresis (PAGE) autoradiograph analysis as previously described (12, 13). All samples were derived at the same time and processed in parallel. Islet lysate aliquots were collected to analyze the total protein content (Pierce BCA, Thermo Scientific, Rockford, IL) and total protein synthesis by trichloroacetic acid precipitation, as described (13).

Analysis of in vitro insulin secretion

Insulin secretion was assessed in freshly isolated islets by static incubation as described (12, 14). Freshly isolated islets were used for all assays to best mirror the in vivo physiology and to avoid potential confounding culture effects. Batches of 50 islets were preincubated at 37°C for 90 minutes at 3 mM glucose in Krebs-Ringer bicarbonate HEPES buffer and then incubated for 1 hour at either basal 3 mM or stimulatory 17 mM glucose. The incubation medium was then collected to determine insulin secretion, and the islets were lysed to assay for insulin content. The insulin concentration of the incubation media and islet lysates was measured by ELISA.

Analysis of preprosinsulin mRNA

Total RNA was extracted from ∼100 isolated rat islets using the RNAeasy mini kit (Qiagen, Hilden, Germany). Quantitative reverse transcription polymerase chain reaction analysis of preproinsulin-1 and preproinsulin-2 mRNA transcript levels relative to control β-actin mRNA transcript levels was implemented in 40 ng of islet RNA using the one-step Power SYBR-Green RNA-to CT kit (Applied Biosystems, Foster City, CA) using an ABI Prism 7700 Sequence Detector System (Thermo Fisher Scientific, Waltham, MA) as described (15). The primers used were from IDT (Coralville, IA) and are as follows: rat preproinsulin-1 forward 5′-GTGGGGAACGTGGTTTCTT-3′, reverse 5′-GCAGTAGTTCTCCAGTTGGTAGAGG-3′; rat preproinsulin-2 forward 5′-TCATCCTCTGGGAGCCCCGC-3′, reverse 5′-GTTGCAGTAGTTCTCCAGTTGGT-3′; and β-actin forward 5′-TGTCACCAACTGGGACGATA-3′, reverse 5′-GGGGTGTTGAAGGTCTCAAA-3′. Results are shown as the preproinsulin mRNA to β-actin mRNA ratio and are normalized to the transcript levels of islets from ad libitum fed control rat islets using the 2^−ΔΔCt method.

Immunohistochemical analysis

Pancreata were fixed for 4 hours in 4% paraformaldehyde in phosphate-buffered saline, paraffin embedded, and cut into 5-μm sections. Sections were then deparaffinized, rehydrated, and stained on the Leica Bond RX Autostainer (Leica Microsystems, Inc., Buffalo Grove, IL) with rabbit anti–v-maf musculoaponeurotic fibrosarcoma oncogene homolog A (MafA) (RRID: AB_1279486; Bethyl Laboratories, Montgomery, TX), 30 minutes of antigen retrieval in ER2 solution (EDTA, pH 9; Leica Microsystems), and Leica Polymer Refine horseradish peroxidase followed by Leica 3,3′-diaminobenzidine tetrahydrochloride. Samples were removed from the Leica Bond, rinsed with Ventana reaction buffer, and placed on the Ventana Discovery Ultra (Ventana Medical Systems, Tucson, AZ) for glucagon and insulin detection. The MafA 3,3′-diaminobenzidine tetrahydrochloride–stained slides were heat inactivated and neutralized with Ventana reaction buffer and Discovery Inhibitor, respectively, prior to rabbit antiglucagon detection (RRID: AB_2716760; Ventana Medical Systems) and visualization with Ventana OmniMap anti-rabbit HRP and the Discovery Purple Kit. Rabbit monoclonal to insulin (RRID: AB_2716761; Abcam, Cambridge, UK) was detected with anti-rabbit nitropyrazole and anti–nitropyrazole-alkaline phosphatase and Discovery Yellow substrate (Ventana Medical Systems). The slides were counterstained with hematoxylin and coverslipped with permanent mounting media.

Electron microscopy analysis

Primary isolated islets were high-pressure fix-frozen, resin embedded, sectioned, and stained as previously described (11). Samples were imaged at ×2500 using the FEI Tecnai G2 SPIRIT electron microscope (FEI, Hillsboro, OR) equipped with a charge-coupled device camera (Pleasanton, CA) at 120,000 V. Images were acquired using digital micrograph software (GATAN, Pleasanton, CA). Electron micrographs of islet β-cells were viewed using 3Dmod software (16) on a Cintiq 22HD art tablet (Wacom, Vancouver, WA). The outlines of relevant organelles were traced, and their volumes or number per total cytoplasmic cell (nucleus excluded) area were calculated based on image magnification and micrograph thickness.

Statistical analysis

Results are presented as the mean ± standard error of the mean of at least three independent experiments. Statistically significant differences between groups were analyzed using one-way analysis of variance (ANOVA) with Fisher Least Significant Difference post hoc test for parametric data or Kruskal-Wallis test with Dunn post hoc test for nonparametric data. A P value ≤ 0.05 was considered statistically significant.

Results

In vivo parameters of fasted-refed rats

Male Wistar rats fasted for 72 hours displayed an ∼20% reduction in body weight (Fig. 1A) but rapidly recovered after only 12 hours of ad libitum refeeding. After fasting, blood glucose levels significantly decreased by ∼33%, remained elevated for 12 hours after refeeding, and then returned to normoglycemic levels (Fig. 1B). Changes in serum insulin and proinsulin levels tended to inversely correlate with glucose levels and after the fast were significantly reduced by threefold (Fig. 1C) and twofold (Fig. 1D), respectively. One hour after refeeding, serum insulin levels spiked back to baseline levels but were then depressed at 2 hours before gradually increasing throughout the refeeding period (Fig. 1C). Serum proinsulin levels were significantly diminished for 1 hour after refeeding before gradually increasing toward baseline levels (Fig. 1D).

Figure 1.

Circulating in vivo parameters of study animals. Analysis of (A) body weight, (B) plasma glucose, (C) plasma insulin, and (D) plasma proinsulin in 12-week-old rats undergoing a 72-hour fast followed by ad libitum refeeding. Statistical significance was determined by one-way ANOVA followed by a Fisher least significant difference post hoc test. Results are presented as the mean ± standard error (n ≥ 4). *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001 compared with control rats fed ad libitum.

Glucose intolerance and hypoinsulinemia in fasted rats

To better assess the in vivo glucose homeostasis and insulin response in fasted-refed animals, intraperitoneal glucose tolerance tests (GTTs) on ad libitum fed, 72-hour fasted, and 72-hour fasted, and 2-, 4-, 6-, 24-, and 60-hour refed rats were performed. Fasted rats had significantly reduced time 0 glucose and insulin levels compared with ad libitum fed controls (Fig. 2A and 2B and Fig. 2H and 2I, respectively). However, the excursion of glucose was relatively normal (Fig. 2B and 2O) despite extremely low excursions in circulating insulin during the GTT (Fig. 2I and 2P) implicating additional control of glucose homeostasis during starvation (17). At 2 and 4 hours after refeeding, rats displayed significant glucose intolerance (Fig. 2C, 2D, and 2O) relative to controls (Fig. 2A). This was mostly caused by significantly insufficient parallel in vivo insulin secretion (Fig. 2J, 2K, and 2P). At 6 hours after refeeding and beyond, the glucose excursions returned to normal (Fig. 2E and 2F and Fig. 2O). However, the excursion in insulin during the GTT at 6 hours of refeeding remained low (Fig. 2L and 2O), although it was sufficient to control glycemia, and did not return to normal until 24 hours of refeeding and beyond (Fig. 2M and 2N and Fig. 2P). Using time 0 glucose and insulin levels from the GTT data, the ratio of insulin (ng/mL) to glucose (mg/dL) was calculated to provide a rough metric of insulin sensitivity in these animals. Remarkably, 72-hour fasted animals had a >20-fold reduction in their insulin/glucose ratio compared with ad libitum fed controls (Fig. 2Q), whereas 2-, 4-, and 6-hour refed animals retained significantly depressed insulin/glucose ratios compared with ad libitum control animals (Fig. 2Q). By 24 hours, this insulin/glucose ratio had normalized relative to that of the control animals (Fig. 2Q). These findings further suggest that fasted animals, despite being remarkably insulin sensitive, are unable to immediately manage glycemic demand after refeeding.

Figure 2.

Insulin and glucose values during intraperitoneal GTT. Intraperitoneal injection of glucose (1 g/kg body weight) to rats and subsequent analysis of circulating glucose and insulin levels over a 3-hour period was determined. The glucose excursions of rats fed (A) ad libitum, (B) fasted for 72 hours, or refed for (C) 2, (D) 4, (E) 6, (F) 24, or (G) 60 hours after a 72-hour fast are displayed. The corresponding insulin excursions during this test are indicated for rats that were fed (H) ad libitum, (I) fasted for 72 hours, or refed for (J) 2, (K) 4, (L) 6, (M) 24, or (N) 60 hours after a 72-hour fast. The corresponding glucose excursions as gross areas under the curve (AUC_gross) during the intraperitoneal GTT for (O) glucose and (P) insulin. (Q) Insulin (ng/mL) to glucose (mg/dL) ratio at time 0 of the intraperitoneal GTT. Statistical significance was determined by one-way ANOVA followed by a Fisher least significant difference post hoc test. Results are presented as the mean ± standard error (n ≥ 4). *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001 compared with ad libitum fed controls. AUC, area under the curve.

In vitro proinsulin biosynthesis and secretion in fasted-refed islets

To better understand the in vivo response to refeeding after a prolonged fast, we isolated primary islets from rats during the experimental period to assess in vitro proinsulin biosynthesis, protein biosynthesis, insulin secretion, and preprosinsulin transcript levels. Freshly isolated islets from rats were incubated for 90 minutes at basal 3 mM or stimulatory 17 mM glucose, and proinsulin biosynthesis was determined (Fig. 3A). We found a twofold increase in glucose-stimulated proinsulin biosynthesis in islets isolated from 72-hour fasted animals (Fig. 3B) compared with control rats fed ad libitum despite no changes in total protein synthesis during the experimental period (Fig. 3C). Transcript levels of preproinsulin-1 were significantly depressed in 72-hour fasted animals and remained significantly depressed throughout the refeeding period (Fig. 3D). Transcript levels of preproinsulin-2 were also depressed in islets from 2- and 6-hour refed animals (Fig. 3D). This reinforces that the regulation of proinsulin biosynthesis is predominantly mediated at the translational level. In vitro insulin secretion contrasted with that observed in vivo. After 72-hour fasting and up to 6 hours of refeeding, basal and glucose-induced insulin secretion rates were significantly inhibited in vivo (Fig. 1). However, in isolated islets, in vitro basal and glucose-regulated insulin secretion rates were similar, albeit slightly reduced after fasting (Fig. 3E). This suggests that intracellular insulin content in fasted animals, although diminished, is sufficient to mount a normal glucose response. However, there are additional regulatory constraints on insulin secretion in vivo (17) to restrain circulating insulin levels when not needed physiologically, such as during starvation and the heightened insulin sensitivity upon refeeding after a prolonged fast.

Figure 3.

Isolated islet protein and specific (pro)insulin biosynthesis, insulin secretion, and insulin transcript analysis. (A) Representative alkaline-urea PAGE autoradiograph images of immunoprecipitated [³H]proinsulin-1 and [³H]proinsulin-2 biosynthesis from freshly isolated islets treated with glucose (3 or 17 mM for 90 min) and pulse-radiolabeled with [³H]leucine. (B) Densitometric analysis of [³H]proinsulin alkaline-urea PAGE autoradiography of 17 mM glucose–treated islets normalized to 3 mM glucose-stimulated biosynthesis. (C) Total protein synthesis analysis by trichloroacetic acid precipitation of isolated islets from fasted-refed animals normalized to 3 mM glucose-stimulated biosynthesis. (D) Ratio of islet preproinsulin-1 and preproinsulin-2 mRNA levels (Ins1 and Ins2) to β-actin mRNA determined by quantitative reverse transcription polymerase chain reaction and normalized to the transcript levels of islets from ad libitum fed control rat islets using the 2^−ΔΔCt method. (E) One-hour total insulin secretion from freshly isolated islets normalized to total islet insulin content. Statistical significance was determined by one-way ANOVA followed by a Fisher least significant difference post hoc test. Results are presented as the mean ± standard error (n ≥ 3). *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001 compared with ad libitum fed controls.

Structural analysis of fasted-refed islets and pancreatic β-cells

Ultrastructural morphological analysis of pancreatic islet β-cells in fasted-refed animals was assessed by immunohistochemistry and transmission electron microscopy (Fig. 4). MafA is an exclusive β-cell transcription factor that, in the context of pancreatic islets, controls numerous genes necessary for β-cell differentiation (18) and can be used to assess β-cell populations independent of insulin (11). MafA positivity was unchanged between fasted and refed animals (Fig. 4A–4F), indicating that β-cell numbers were unaltered in response to fasting. However, fasting caused a noticeable reduction in insulin content in islets of fasted rats (Fig. 4B), consistent with insulin degradation during prolonged nutrient deprivation (9). Electron microscopy analysis (Fig. 4G–4L) indicated insulin degranulation in 72-hour fasted rat islets compared with ad libitum fed control islets (Fig. 4H vs 4G). However, mature β-granules rapidly returned after refeeding (Fig. 1L). Moreover, 72-hour fasted islets displayed an abundance of Golgi and autophagolysosomes. Quantification of electron micrographs (>10 electron micrograph images from ≥3 distinct islet preparations from each timepoint; ∼30 β-cells per group) indicated that 72-hour fasted islet β-cells were significantly degranulated (>fivefold; Fig. 5A) compared with ad libitum fed controls, but then the β-granule population significantly increased after only 2 hours of refeeding and slowly recovered to normal levels by 60 hours (Fig. 5A). In contrast, a significant increase in immature granules after 2 hours of refeeding was observed (Fig. 5B) compared with ad libitum fed controls. The presence of autophagolysosomes was nearly 10-fold greater in 72-hour fasted islet β-cells compared with controls but was rapidly reduced upon refeeding to reach normal levels by 4 hours (Fig. 5C), inverse to the increase in mature β-granules (Fig. 5A). The presence of lysosomes was increased in the 72-hour fasted islets and in islets that had been fasted and refed compared with controls (Fig. 5D). The β-cell area occupied by the Golgi apparatus was also significantly increased in 72-hour fasted islets but returned to control levels after refeeding (Fig. 5E). However, the β-cell area occupied by the endoplasmic reticulum was slightly increased at 2 hours but was essentially unchanged during fasting or during the period of refeeding (Fig. 5F). In addition, unique membranous structures in immediate proximity to the cis-Golgi were observed only in 72-hour fasted animals that were termed “multi-membraned organelles” (MMOs) (Fig. 4, asterisks; see Fig. 6 for multiple examples) and may represent membrane recycling centers (11).

Figure 4.

Representative immunohistochemical staining of fixed pancreata sections and electron micrographs of freshly isolated islet β-cells. Immunohistochemical staining of insulin (yellow), glucagon (purple), and MafA (brown) of fixed pancreata from rats that were (A) ad libitum fed, (B) 72-hour fasted, or 72-hour fasted and ad libitum refed for (C) 2, (D) 4, (E) 24, or (F) 60 hours. Representative transmission electron micrographs of high-pressure, fix-frozen pancreatic islet β-cells isolated from rats that were (G) ad libitum fed, (H) 72-hour fasted, or 72-hour fasted and ad libitum refed for (I) 2, (J) 4, (K) 24, or 60 (L) hours. Multimembraned organelles localized exclusively near the Golgi are annotated with yellow asterisks in (H). These structures were not observed in ad libitum fed or fasted-refed animals. Scale bars, (A–F) 1 mm and (G–L) 1 µm.

Figure 5.

Quantification of conventional electron micrograph analysis of freshly isolated islet β-cells. Islets isolated from fasted-refed rats were high-pressure fix-frozen immediately after isolation. Then, micrographs of β-cells were collected, and the number of (A) mature granules, (B) immature granules, (C) autolysosomes, and (D) lysosomes or the area occupied by the (E) Golgi or (F) endoplasmic reticulum was quantified as described in Materials and Methods. Statistical significance was determined by Kruskal-Wallis test followed by Dunn post hoc test. The quantification results are presented mean ± standard error (n = 4 islet isolations; n ≥ 10 micrographs). *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001 compared with ad libitum fed controls. Representative electron micrographs of quantified structures are included in the inset of each graph.

Figure 6.

Representative examples of MMOs. (A–G) MMOs were exclusively observed near the cis-Golgi of β-cells from 72-hour fasted rats.

Discussion

The pancreatic β-cell continually monitors metabolic homeostasis and can rapidly adapt to meet the metabolic demand for insulin (19). Here, we demonstrate β-cell adaptation in response to fasting and subsequent refeeding in rats, which likely has relevance to humans. In humans, 6 days of fasting significantly reduced glucose tolerance to an oral glucose challenge that was accompanied by a delayed elevation in insulin (20). The decreased glucose tolerance was likely associated with impaired glucose utilization; euglycemic clamp studies of 48-hour fasted humans found a 50% reduction in glucose disposal during fasting and a fourfold reduction in skeletal muscle glucose uptake after exogenous insulin administration (21, 22). Humans starved for 2 weeks maintained some sensitivity to exogenous insulin but also displayed glucose intolerance in the immediate refeeding period, which was presumed to be due to insufficient (pro)insulin synthesis and secretion (23). However, the coordinated relationship between glucose homeostasis, (pro)insulin production, and insulin secretion during a prolonged fast and subsequent refeeding has not been assessed.

In this study, after a prolonged fast in rats (72 hours), plasma insulin levels, like in humans, were significantly decreased but returned to baseline levels only 4 hours after refeeding, paralleling recovery of glucose homeostasis. Fasted rats maintained reasonably good glucose tolerance despite markedly depressed insulin secretion in vivo. This did not appear to be due to insufficient β-cell secretory function per se because, although there was diminished insulin secretory capacity due to increased degradation of β-granule stores in islet β-cells of fasted rats, adequate normal glucose-induced insulin secretion was found from these isolated islets in vitro. This suggests additional negative control of insulin secretion in vivo during starvation that overrides glucoregulatory control of β-cell secretory dysfunction. Indeed, in addition to direct neuronal inhibition, a host of hormones associated with negative energy balance has been indicated to suppress in vivo insulin secretion, including epinephrine, corticosterone, and ghrelin (17, 24–29). However, despite this repression of insulin secretion in vivo, glucose tolerance was relatively normal, which could imply a degree of increased insulin sensitivity and/or insulin-independent glucose uptake mechanisms in peripheral tissues during fasting. However, in contrast, some previous work has suggested that prolonged starvation is an insulin-resistant state, mostly at the level of skeletal muscle (30, 31), but calorie restriction in humans has been shown to improve insulin sensitivity and glucose homeostasis (32, 33). This rat study supports the latter scenario, as indicated by the 20-fold reduction in the 6-hour fasting insulin/glucose ratio in 72-hour fasted rats, suggesting that it is suppression of in vivo insulin secretion, rather than insulin resistance, that is a primary defense mechanism against anabolic activity during starvation.

The focus of this study was to examine the rapid changes in cellular ultrastructure that parallel the changes in β-cell (pro)insulin biosynthesis and insulin secretory capacity observed during a prolonged fast and a subsequent refeeding period. β-Cells from study animals indicated a dynamic ability to respond to these metabolic changes. Although there was no apparent change in β-cell mass, islet β-cells had decreased intracellular insulin stores mediated by increased microautophagic degradation of β-granules (34, 35). Under normal circumstances, β-granules have a half-life of ∼3 to 5 days, and if they do not undergo exocytosis they are degraded by autophagy (35). Because in vivo insulin secretion was shut down in the 72-hour fasted rats with a prevailing circulating glucose level of <50 to 60 mg/dL, (pro)insulin biosynthesis would likely be negligible in vivo (35, 36). Thus, increased autophagic activity was likely due to the increased degradation of aged mature β-granules (35).

The predominant regulation of proinsulin biosynthesis is at the translational level (35–37). Here, when directly measured ex vivo in freshly isolated islets, proinsulin biosynthesis was significantly increased in 72-hour fasted rats. However, this was unlikely to reflect in vivo rates of proinsulin biosynthesis in β-cells of fasted animals where hypoglycemia would ensure that (pro)insulin production was markedly reduced (35, 36). The analysis of proinsulin biosynthesis in this study reflects an increased capacity for proinsulin biosynthesis in the β-cells of 72-hour fasted rats. An expanded Golgi apparatus in the β-cells of fasted animals also reflects such a potential increased capacity for (pro)insulin production. Likewise, the increased frequency of MMOs, in proximity to the Golgi network, could also be related to the β-cells of fasted rats being poised for increased proinsulin production. These MMO structures were unlikely to be multilamellar bodies, which have been associated with impaired autophagy (38), or multivesicular bodies, which are associated with late endosomes (39), increased endocytosis (40), or cellular remodeling during mitosis (41). Rather, MMOs are most likely membrane recycling centers (11), indicative of the anticipated enhanced membrane redistribution necessary for increased proinsulin production and β-granule biogenesis to restore β-cell insulin secretory capacity. Thus, fasting ideally poises β-cells for the effective and efficient upregulation of proinsulin biosynthesis upon refeeding while protecting the organism from unnecessary insulin secretion during starvation. There was no indication of β-cell dedifferentiation during fasting or refeeding. Indeed, considering the number of MafA-positive β-cells [a key transcription factor necessary for β-cell differentiation and a marker of “mature” β-cells (18)] was unchanged during fasting or refeeding, their transcriptional identity was likewise unaffected.

β-Cells from fasted animals rapidly regranulated and returned to normal secretory capacity within 4 to 6 hours of refeeding to obtain normal glucose homeostasis. During this time, there was some transient glucose intolerance as the in vivo insulin secretory response to increased circulating glucose levels recovered, but by 6 hours, glucose tolerance and in vivo insulin secretory capacity returned to normal. The cell biology of fasted pancreatic β-cells reflected this recovery to normal secretory capacity during this short time frame with a marked expansion in the mature β-granule population that inversely paralleled a rapid decline in autophagolysomes. Therefore, as β-granule biogenesis increased, β-granule degradation was decreased upon refeeding, as previously predicted (42). Specific glucose-induced translational control of proinsulin biosynthesis applies not only to proinsulin but also to the proinsulin processing endopeptidases and the vast majority of β-granule proteins (35, 43) and can increase by as much as 10-fold within an hour (36). Future studies that quantify changes in β-cell proteins during the fasting-refeeding transition, particularly those associated with the β-granule membrane, would provide key information to better characterize this β-cell flexibility for insulin production. This mechanism is designed for a rapid increase in β-granule biogenesis upon refeeding after a fast and represents an important component of the ability of the pancreatic β-cell to effectively and efficiently respond to changing metabolic demand (19). Therefore, therapeutic strategies that support the innate ability of the β-cell to rapidly regain secretory function should be emphasized as first-line treatments for obesity-linked type 2 diabetes.