Protein malnutrition potentiates the amplifying pathway of insulin secretion in adult obese mice

Pancreatic beta cell (β) dysfunction is an outcome of malnutrition. We assessed the role of the amplifying pathway (AMP PATH) in β cells in malnourished obese mice. C57Bl-6 mice were fed a control (C) or a low-protein diet (R). The groups were then fed a high-fat diet (CH and RH). AMP PATH contribution to insulin secretion was assessed upon incubating islets with diazoxide and KCl. CH and RH displayed increased glucose intolerance, insulin resistance and glucose-stimulated insulin secretion. Only RH showed a higher contribution of the AMP PATH. The mitochondrial membrane potential of RH was decreased, and ATP flux was unaltered. In RH islets, glutamate dehydrogenase (GDH) protein content and activity increased, and the AMP PATH contribution was reestablished when GDH was blunted. Thus, protein malnutrition induces mitochondrial dysfunction in β cells, leading to an increased contribution of the AMP PATH to insulin secretion through the enhancement of GDH content and activity.

reported in obese malnourished mice 18 ; however, the impact of obesity and malnourishment on mitochondrial function in β cells and the AMP PATH of insulin secretion is still poorly understood.
Considering the hypothesis that β cell dysfunction is at the center of malnutrition-induced T2D development 19,20 and mitochondria are the main intracellular targets of this programming, investigating the molecular mechanisms by which protein malnutrition induces a metabolic program might provide clues to the establishment of therapeutic targets to avoid the undesirable outcomes. Thus, we investigated the effects of protein malnutrition on the AMP PATH of GSIS in obese mice and the underlying molecular mechanisms.

Malnourished obese mice present an increased contribution of the AMP PATH of insulin secretion.
Restricted mice displayed decreased body weight (BW) (Fig. S1A, Table S1), increased glucose tolerance, and insulin sensitivity (Fig. 1A-C) with a decrease in plasma insulin levels (Table S1). HFD intake increased BW, fat depots, the Lee index, glucose intolerance, and insulin resistance ( Fig. 1A-C) in the CH and RH groups (Table S1). Both of the groups displayed increased total fasting plasma insulin levels (Table S1). The RH group presented diminished energy expenditure. R secreted less insulin in the presence of 22.2 mmol/L glucose compared with C, whereas the mice treated with a HFD displayed higher insulin secretion (Fig. 1D), despite no reported changes in the total content of insulin in islets from all groups (Fig. 1E).
The AMP PATH was studied while holding the K + ATP channels open with diazoxide and depolarizing the islet cells with high concentrations of KCl in the presence of low or high glucose. In such experimental conditions, the increase in [Ca 2+ ]i was similar in the presence of 2.8 and 22.2 mmol/L glucose (Fig. 2B-E), and the AMP PATH contribution was higher only in the islets from RH mice ( Fig. 2A).
Islets from malnourished obese mice present an altered mitochondrial profile and GDH interferes in the AMP PATH of insulin secretion. PKAα and PKCα content did not change in the islets from RH mice (Fig. 2F,G). RH displayed decreased mitochondrial fluorescence compared to R (Fig. 3B). In addition, ATP flow was similar between R and RH (Fig. 3J,K). Despite it was not reported changes on the protein levels of the electron transport chain complexes (Fig. 3L), we observed a significant decrease in the mitochondrial membrane potential of RH islets (Fig. 3F-I), which was followed by increased levels of hydrogen peroxide (Fig. 3D).
Although no differences in GLUD1 gene expression were detected (Fig. 4A), GDH protein content and activity increased in the islets from RH (Fig. 4B-D). In agreement, the GSIS AMP PATH was almost inhibited in the presence of the GDH inhibitor, epigallocatechin-monogallate (EGCG) (Fig. 4H). RH group also displayed lower  levels of ubiquitin proteasome pathway proteins, Murf1 and ubiquitin (Fig. 4E,G). GDH in CH did not change (Fig. S3G). The apoptosis levels were not different in the islets from the R, RH (Fig. S2A,B) and CH groups (Fig. S3F).

Discussion
Protein malnutrition early in life leads to β cells dysfunction and a derangement of glucose metabolism during adulthood 11 . In addition, it worsens the deleterious effects of obesity on β cells, compromising the efficacy of therapeutic approaches to reestablish the triggering mechanisms of insulin secretion 18 . Insulin resistance alters the expression and/or activity of various proteins that form the K + ATP and voltage dependent Ca 2+ channels 21,22 . Because the triggering and AMP PATH work together to regulate insulin secretion 7 , the loss of one of these components may overload the other, leading to β cell failure; however, the impact of impaired triggering mechanisms on the AMP PATH of obese mice has not yet been addressed.
A higher fold change in body mass and the Lee index from the protein-restricted group shows a metabolic program outcome and might result from the increased energy intake supplied by the HFD, associated with lower energy expenditure. These results are in agreement with impaired hypothalamic insulin signaling previously reported in obese malnourished mice 23 .
Obesity reduced glucose tolerance and insulin sensitivity in normal-and low-protein-fed groups. In response to an increase in insulin resistance, insulinemia promptly increased in both groups. Thus, we proceeded to investigate pancreatic β cell function and observed that GSIS was also enhanced in the HFD-treated groups. Although the degree of insulin resistance was similar between CH and RH as judged by blood glucose uptake during the euglycemic-hyperinsulinemic clamp, the absolute levels of GSIS were lower in RH for a given glucose concentration, suggesting a possible signal of early pancreatic β cell failure.
While investigating the contribution of the AMP PATH to the enhanced GSIS in obese mice and the possible contribution brought by protein malnutrition, we observed that protein malnutrition (R) or HFD (CH) alone did not interfere in the AMP PATH; however, when mice were subjected to a low-protein diet followed by a HFD, an increased participation of the AMP PATH was detected. This outcome could be due to an acute response to a higher amplitude of intracellular Ca 2+ influx 24 in RH islets; however, no differences in the Ca 2+ amplitude were reported in these groups 18 . Therefore, the increase in the AMP PATH might be a compensatory mechanism to maintain adequate levels of GSIS. Indeed, pancreatic islets and β cell lines chronically exposed to high levels of nutrients undergo changes in glucose metabolism, pyruvate shuttling, mitochondrial function and oxidative stress associated with altered GSIS 25 . In addition, islets from genetically obese (ob/ob) mice display higher firing frequency, intracellular Ca 2+ mobilization, and oscillatory pattern, even at low glucose concentrations 26,27 . All of these changes point to compensatory adaptations to adequately adjust GSIS demand. Protein kinases A and C (PKAα and PKCα ) activate the AMP PATH, potentiating insulin secretion 6 . In addition, GSIS amplification also depends on mitochondrial metabolism through the production of the so-called metabolic coupling factors. Thus, we attempted to clarify which of these components of the AMP PATH are modulated by protein malnutrition.
To establish which mechanisms are responsible for the increased response of the AMP PATH, we first analyzed PKAα and PKCα protein content. In contrast to other studies 28,29 , our model displayed no alterations in the content of these kinases; however, the mitochondrial profile was modified. Chronic exposure to high levels of nutrients increases β cell mitochondrial membrane potential, enhancing the triggering and AMP PATH in non-protein-restricted mice 26,30 . In our RH mice, the mitochondrial content and the inner membrane potential were reduced, even at non-stimulatory glucose concentrations. Mitochondrial protein expression and metabolism are altered in response to protein malnutrition or HFD 31,32 . A low-protein diet reduces the content of several mitochondrial proteins including respiratory chain proteins, leading to reduced ATP production 33 . Thus, protein restriction seems to compromise mitochondrial function elicited by a HFD in β cells from RH, leading to reduced membrane potential despite no differences reported on mitochondrial complexes content and ATP flow. These data suggest dysfunctional activity of mitochondrial complexes in obese malnourished mice, since the protein content was not altered. Considering that the mitochondrial inner membrane potential was reduced in RH, we investigated the shuttling cycles that have been proposed as the source of other metabolic coupling factors for insulin secretion. GDH has a close relationship with GSIS [34][35][36] . In addition, GDH was already reported to modulate the AMP PATH in lean mice, increasing glutamate production 37 . Thus, this enzyme may be involved with the increased participation of the AMP PATH on GSIS in obese malnourished mice. Whereas the expression of GDH mRNA was similar between the groups, the protein content and activity were increased in RH islets, correlating positively with the GSIS AMP PATH, indicating that GDH protein may undergo post-translational modulation. Considering the reduced amino acids supply during protein malnourishment, we hypothesized that cells would reduce protein degradation rather than increase protein synthesis. Thus, we measured proteins from the ubiquitin proteasome, and reported lower content in two of them in obese malnourished mice. These results led us to propose that protein degradation should be lower in RH, and this modulation might explain the higher protein content of GDH found in this group.
Indeed, the participation of GDH on the enhanced AMP PATH was confirmed when insulin secretion was restored in RH islets incubated in presence of Dz, KCl and EGCG. In agreement, higher mitochondrial shuttling cycles were reported in obese Zucker rats, providing evidence that obesity enhances the AMP PATH to support their increased GSIS 32 . These findings support that mitochondria are intracellular targets for malnutrition-induced β cell programming 15 . Mitochondrial dysfunction reported in RH (i.e., reduced membrane potential) is associated with higher ROS production 38 , which may trigger signaling pathways involved with β cell death 39 . In an attempt to establish a possible relationship between β cell dysfunction and death in this model, we measured ROS and cell viability. ROS content was increased in RH whereas in CH it was not altered; however, we did not detect evidence of increased apoptosis or cell death in any group, indicating that at this stage of treatment, there was only impairment of β cell function. Nevertheless, the protein malnutrition-induced earlier exposure to higher levels of ROS might reduce β cell viability and predispose its failure. In addition to the classical DOHaD concept, that establishes the origins of metabolic diseases in early life periods, our group and others have shown that the protein deprivation, during the adolescence also alters glucose homeostasis, showing a new window for the metabolic programing. Moreover, the association of protein malnutrition followed by a high fat diet impairs metabolic control, leading to glucose intolerance and insulin resistance in adulthood 18,40,41 . The increase in ROS production in RH islets provides evidence for the malnutrition-induced metabolic reprograming, favoring degenerative processes, such as oxidative stress, which might contribute to the reduced viability of β cells.
In conclusion, protein malnutrition programming of β cells induces mitochondrial dysfunction, changing the response to a HFD challenge. Whereas the signaling of the triggering pathway through ATP production is impaired, the AMP PATH is enhanced due to increased GDH. Further studies are needed to establish the mechanisms of malnutrition-induced mitochondrial impairment in response to nutrient overload and its role on β cell failure and T2D onset.

Experimental Procedures
Animals and dietary interventions. All animal experiments were carried out in accordance with the protocols approved by the Animal Care and Use Committee of the Campinas University (UNICAMP) (number: 3057-1). C57Bl/6 mice were obtained from UNICAMP and maintained at 22 ± 1 °C in a 12-h light-dark cycle. Thirty-day-old mice were assigned into the following groups: normal protein diet (14% protein) (Control: C) or protein-restricted diet (6% protein) (Restricted: R). After 6 weeks, some of the C and R mice were fed either a normal or high fat diet (HFD) (35% fat) for 8 weeks (CH and RH).
Scientific RepoRts | 6:33464 | DOI: 10.1038/srep33464 Physiological Measurements. The Lee Index was calculated 42 . The fat pads were weighed. Plasma glucose and insulin were measured using a glucometer (Accu-Chek Performa) and by radioimmunoassay (RIA), respectively.
For indirect calorimetry, the mice were allowed an 8 h acclimation to the apparatus. The mice remained at rest during a light/dark period for 24 h. O 2 and CO 2 were measured by an Oxylet system (Pan Lab/Harvard Apparatus). The respiratory quotient (RQ) was calculated from these data. Locomotor activity analysis was performed in multitake cages LE 001 PH (Pan Lab/Harvard Apparatus).

Intraperitoneal glucose test and hyperinsulinemic-euglycemic clamp. For the intraperitoneal glu-
cose test (ipGTT), mice were fasted overnight (12 h). An intraperitoneal glucose load (2 g/kg) was administered, and blood glucose measurements recorded at 0, 15, 30, 60, and 120 min via tail snip using a handheld glucometer. A hyperinsulinemic-euglycemic clamp (120 min) was carried out with a prime continuous insulin infusion (30 mU kg −1 . min −1 ). Blood glucose was measured at 5 min intervals, and glucose (5% wt/vol) was infused at a variable rate to maintain blood glucose at fasting levels.
Islet isolation, GSIS, and participation of AMP PATH of insulin secretion. Islets were isolated by collagenase digestion of the pancreas. Islets were incubated for 30 min at 37 °C in Krebs-bicarbonate buffer (KBB). This medium was then replaced with fresh buffer, and the islets were incubated for 1 h with 2.8, 11.1 and 22.2 mmol/L glucose. To investigate the AMP PATH of insulin secretion, 30 mmol/L K + and 0.25 mmol/L diazoxide (Dz) were added to the medium containing 2.8 or 22.2 mmol/L glucose. The insulin content of the medium was measured by RIA.

Measurement of oscillations of cytoplasmic Ca 2+ .
Pancreatic islets were incubated with fura-2 acetoxymethyl ester (5 μ mol/L) for 1 h at 37 °C in KBB. The medium was replaced, and the islets were plated in a chamber at 37 °C on the stage of an inverted microscope (Nikon UK). The islets were then perfused with albumin-free KBB, which contained 2.8 mmol/L or 22.2 mmol/L glucose with 0.25 mmol/L Dz and 30 mmol/L K + , as indicated in the figures. A ratio image was acquired every 5 seconds with an inverted epifluorescence microscope (Nikon Eclipse TE200, Tokyo, Japan). The data were obtained using ImageMaster 3 software (Photon Technology International) 18 .
Fluorescent staining of mitochondria. MitoTracker Green and Red CMXRos (100 nmol/L) was loaded into intact pancreatic islets at 37 °C in a humidified incubator for 20 min and observed using a FLoid Cell Image Station (Life Technologies).

Measurements of ATP flow and mitochondrial membrane potential (Ψ). Islets were loaded with
MgG acetoxymethylester (5 mmol/L, Molecular Probes) for 60 min or Rhod-123 (10 μ g/mL, Sigma) for 20 min at 37 °C in a humidified incubator and then were perfused with KBB containing varying concentrations of glucose (G2.8 and G22.2), and the fluorescence intensity was measured using a SpectraMax M3 (Molecular Devices).
Hydrogen peroxide production (Amplex Ultra Red). Islets were transferred to 96 wells culture plates containing 100 μ l of KBB solution with G22.2 and Amplex Ultra Red (50 μ M) for 60 min at 37 °C, fluorescence intensity was measured using a SpectraMax M3 (Molecular Devices). Catalase (300 U/L) was used as a negative control.
Quantification of HO-PI fluorescence. The percentage of dead and viable cells was assessed by quantifying the fluorescence of the DNA-binding dyes PI and HO 33342 (Sigma-Aldrich), respectively. The islets were incubated with HO-PI for 20 minutes (5 mg/ml) and observed using a FLoid Cell Image Station (Life Technologies).
Assay of GDH activity. GDH activity was determined using a specific kit (Abcam 102527). Measurement of GDH activity was conduct using the SpectraMax M3 plate reader (Molecular Devices).
Statistical Analysis. The data are presented as the means ± SEM, and the differences were considered significant when p < 0.05. Comparisons were performed using 2-tailed unpaired Student's t test or ANOVA followed with Newman Keuls correction when necessary.