Glyoxalase 1 knockdown induces age‐related β‐cell dysfunction and glucose intolerance in mice

Abstract

Tight control of glycemia is a major treatment goal for type 2 diabetes mellitus (T2DM). Clinical studies indicated that factors other than poor glycemic control may be important in fostering T2DM progression. Increased levels of methylglyoxal (MGO) associate with complications development, but its role in the early steps of T2DM pathogenesis has not been defined. Here, we show that MGO accumulation induces an age‐dependent impairment of glucose tolerance and glucose‐stimulated insulin secretion in mice knockdown for glyoxalase 1 (Glo1KD). This metabolic alteration associates with the presence of insular inflammatory infiltration (F4/80‐positive staining), the islet expression of senescence markers, and higher levels of cytokines (MCP‐1 and TNF‐α), part of the senescence‐activated secretory profile, in the pancreas from 10‐month‐old Glo1KD mice, compared with their WT littermates. In vitro exposure of INS832/13 β‐cells to MGO confirms its casual role on β‐cell dysfunction, which can be reverted by senolytic treatment. These data indicate that MGO is capable to induce early phenotypes typical of T2D progression, paving the way for novel prevention approaches to T2DM.

Methylglyoxal accumulation promotes a pro‐inflammatory microenvironment in pancreatic islets of mice, leading to an age‐dependent impairment of glucose‐stimulated insulin secretion (GSIS) and glucose tolerance in mice.

Introduction

Tight control of blood glucose levels is still used as a major goal for treatment of diabetes mellitus (DM). Nonetheless, while most of the literature claims an indisputable role of good glycemic control in the prevention of vascular complications, clinical studies have indicated that lowering glycated hemoglobin (HbA1c) only partially reduces the risk of cardiovascular events in diabetic patients (Action to Control Cardiovascular Risk in Diabetes Study et al, ²⁰⁰⁸; Currie et al, ²⁰¹⁰; Lind et al, ^{2010, 2014}; DCCT/EDIC Research Group, ²⁰¹⁴), suggesting that factors other than poor glycemic control are important in inducing vascular complications (Fleming & Nawroth, 2014; Bots et al, ²⁰¹⁶).

Among the biochemical mechanisms by which hyperglycemia affects cellular function, the increase in reactive dicarbonyls, including methylglyoxal (MGO), has been identified as an important player (Nigro et al, ²⁰¹⁷; Schalkwijk & Stehouwer, ²⁰¹⁹). MGO is the major source of intra‐ and extracellular advanced glycation end products (AGEs; Brownlee, ²⁰⁰¹), which lead to cellular dysfunction and the progression of tissue damage (Rabbani & Thornalley, ^2014a; Nigro et al, ²⁰¹⁹). MGO is mainly formed by the non‐enzymatic dephosphorylation of triose phosphates, as a byproduct of the glycolytic flux (Sousa Silva et al, ²⁰¹³). However, its increase may be additionally raised by the oxidation of ketone bodies, protein catabolism, glyceroneogenesis, or by impaired detoxification by the glyoxalase system or depletion of available glutathione (GSH) (Rabbani et al, ²⁰¹⁶). Indeed, physiological levels of MGO are guaranteed by efficient detoxification in D‐lactate by the glyoxalase 1 and 2, requiring GSH as cofactor (Sousa Silva et al, ²⁰¹³). A decrease in Glo1 activity and transcription has been described to occur physiologically with aging (Xue et al, ²⁰¹¹; Fleming et al, ²⁰¹³), and in other pathological conditions, such as oxidative stress, inflammation, and senescence (Bierhaus et al, ²⁰⁰¹; Ikeda et al, ²⁰¹¹; Rabbani et al, ²⁰¹⁶).

Methylglyoxal levels have been estimated to be twofold to fivefold higher in type 2 and type 1 diabetic patients, respectively, compared with healthy individuals (Lapolla et al, ²⁰⁰³; Schalkwijk & Stehouwer, ²⁰¹⁹). However, it has been shown that MGO formation remains increased in diabetic patients even after normalization to glucose concentrations (Fleming et al, ²⁰¹²). A recent investigation performed in newly diagnosed type 2 diabetes (T2D) patients demonstrated a significant increase in MGO plasma levels already at an early stage in diabetes and in the absence of diabetic complications (Kong et al, ²⁰¹⁴). Furthermore, an increase in MGO adducts, that is, methylglyoxal‐derived hydroimidazolone (MG‐H1), has been observed in the pre‐diabetic state, irrespective of unaltered fasting blood glucose, in spontaneously diabetic rats (Chen et al, ²⁰¹⁵). This evidence supports the hypothesis that alterations in MGO metabolism may precede the onset of DM. We further hypothesize that this metabolic unbalance is crucial to the pathogenesis of the disease, acting not only as a result of hyperglycemia but more importantly as an upstream event in the alteration of glucose homeostasis. Indeed, previous studies have proved a damaging effect of MGO and its derived AGEs on micro and macrovascular function (Dobler et al, ²⁰⁰⁶; Dhar et al, ²⁰¹⁰; Nigro et al, ²⁰¹⁷), on insulin sensitivity (Jia & Wu, 2007; Engelbrecht et al, ²⁰¹⁴; Mey & Haus, ²⁰¹⁸) and β‐cell function (Dhar et al, ²⁰¹¹; Fiory et al, ²⁰¹¹; Elmhiri et al, ²⁰¹⁴). The majority of these studies have been carried out in models of hyperglycemia where MGO scavenging demonstrated to be effective in reducing damage or under exogenous administration of MGO. In more recent studies, the modulation of the Glo1 gene in non‐mammal models has demonstrated that MGO elevations induce metabolic defects, namely insulin resistance, obesity, and hyperglycemia in Glo1 knockout (Glo1KO) Drosophila (Moraru et al, ²⁰¹⁸), and the progression toward hyperglycemia and blood vessel alterations in Glo1KO Zebrafish under overfeeding conditions (Lodd et al, ²⁰¹⁹). In rodents, the silencing of Glo1 is sufficient to cause hyperalgesia and alterations in kidney morphology, similar to those induced by diabetes, in mice knockdown for Glo1 (Glo1KD; Bierhaus et al, ²⁰¹²; Giacco et al, ²⁰¹⁴).

However, the impact of the endogenous increase in MGO on glucose metabolism in mammals is still uncertain. The aim of this study is to prove the specific role of MGO accumulation, in the absence of other metabolic hits, on glucose homeostasis in Glo1KD mice.

Results

Glo1KD mice show reduced Glo1 expression and increased MGO plasma levels

Glo1KD mice and their WT littermates (WT mice) have been used to evaluate the effect of MGO accumulation on glucose homeostasis. As expected, Glo1 expression was lower in tissues from Glo1KD mice than WT mice. In detail, mRNA levels were reduced by 90%, 40%, and 62% in skeletal muscle, white adipose tissue, and liver, respectively, of Glo1KD mice compared with WT mice at 10 months of age (Fig 1A–C). The expression levels of the alternative enzymes for MGO detoxification, aldo‐keto reductase family 1 member B3 (Akr1b3) and aldehyde dehydrogenase family 1 member A3 (Aldh1a3), were not different in skeletal muscle, white adipose tissue, and liver from WT and Glo1KD mice (Fig 1D–I). The partial deletion of Glo1 expression is sufficient to induce with age the endogenous accumulation of MGO in this model. Indeed, MGO levels became increased by ca. 50% in the serum from Glo1KD mice compared with WT mice at 10 months of age, but not earlier at 5 months of age (Table 1).

Figure 1. mRNA tissue expression of MGO‐detoxifying enzymes and MGO serum levels.

• A–I
  Data distribution of the absolute quantification of Glo1 (A–C), Akr1b3 (D–F), and Aldh1a3 (G–I) gene expression measured by qRT–PCR in skeletal muscle (quadriceps) (A, D, and G), white adipose tissue (WAT, epididymal fat) (B, E, and H) and liver (C, F, and I) of 10‐month‐old Glo1KD (n = 5 in D; n = 6 in F and I; n = 7 in C and G; n = 8 in A and H; n = 9 in B and E) and WT mice (n = 6 in E, F and I; n = 7 in B–D, G and H; n = 8 in A). Cyclophilin A was used as reference gene. Gene expression is shown as the ratio of Glo1, Akr1b3, or Aldh1a3 to Cyclophilin A copy number (arbitrary units, AU).

Data information: Bars in the scatter plots show the mean ± SEM. ***P ≤ 0.05 (Mann–Whitney test in A, Student’s t‐test in B–I).

Table 1.

Metabolic parameters in WT and Glo1KD mice.

Variable	WT mice	Glo1KD mice	WT mice	Glo1KD mice
	5 months		10 months
Body weight (g)	25.4 ± 0.8	24.6 ± 0.8	32.6 ± 1.5###	35.9 ± 2.0###
Food intake (g/day)	2.88 ± 0.06	3 ± 0.12	3 ± 0.12	2.8 ± 0.24
Fasting glycemia (mg/dl)	105.9 ± 5.3	98.1 ± 3.5	114.6 ± 8.4	110.4 ± 7.4
Fed glycemia (mg/dl)	129.7 ± 6.4	134.8 ± 6.4	149.3 ± 8.5	153.1 ± 6.1
Serum Insulin (ng/ml)	0.75 ± 0.14	0.88 ± 0.14	0.68 ± 0.12	0.89 ± 0.05
Serum Glucagon (pg/ml)	285.4 ± 20	275.1 ± 42.7	285.4 ± 35.8	290.1 ± 32.2
Plasma MGO (nM)	197.6 ± 6.4	185.4 ± 13.7	182.6 ± 15.6	267.2 ± 5.7***,###

Body weight, food intake, fasting and fed glycemia, serum insulin, glucagon, and plasma MGO were measured in Glo1KD mice and their WT littermates at 5 and 10 months of age (body weight, food intake, and fasting glycemia: 5 months n = 19, 10 months n = 15; fed glycemia, plasma insulin, glucagon and MGO: 5 months n = 6, 10 months n = 8). Data are reported as mean ± SEM values. Statistical analysis was performed by Student’s t‐test for data with a normal distribution, or with Mann–Whitney test for data with a not normal distribution, where P ≤ 0.05 was considered statistically significant (***P < 0.001, age‐matched Glo1KD vs. WT mice; ^### P < 0.001, 10 months vs. 5 months).

Glo1KD mice develop an impairment of glucose tolerance with aging

Metabolic characteristics of Glo1KD mice and their WT littermates have been monitored during lifespan. As reported in Table 1, there was no significant difference in body weight, food intake, fed and fasting glycemia, serum insulin, and glucagon levels in age‐matched WT and Glo1KD mice. As expected, body weight increased at 10 months of age compared with 5 months in both genotypes (Table 1). Differently, glucose tolerance was impaired with age, as demonstrated by higher glycemic levels observed during an intraperitoneal glucose tolerance test (ipGTT) in Glo1KD mice at 10 months (Fig 2B) but not at 5 months of age (Fig 2A), as compared to WT mice. Insulin sensitivity was evaluated by an ITT and appeared not to be significantly compromised in Glo1KD mice either at 5 (Fig 2C) or at 10 months of age (Fig 2D), as shown in Fig 2 by both the glycemic trend (line graphs) and the inverse area under the curve (iAUC) calculation (bar graphs). These results indicate that the impairment of glucose tolerance developed by Glo1KD mice with age was not associated with peripheral insulin resistance.

Figure 2. Evaluation of glucose tolerance and insulin sensitivity.

• A, B
  Evaluation of glucose tolerance in 5‐month‐old (A, WT n = 19 and Glo1KD n = 20) and 10‐month‐old mice (B, n = 15) by ipGTT. Line graphs show the trend of blood glucose levels within 120 min after intraperitoneal administration of a glucose bolus (2 g/kg). Bar graphs on the right side of panels A and B show the GTT IAUC of Glo1KD and WT mice.
• C, D
  Evaluation of insulin sensitivity in 5‐month‐old (C, WT n = 8 and Glo1KD n = 7) and 10‐month‐old mice (D, WT n = 5 and Glo1KD n = 7) by ITT. Line graphs show the trend of blood glucose levels within 120 min after intraperitoneal administration of an insulin bolus (0.75 U/kg). Blood glucose was measured by the use of a portable blood glucometer at the indicated time points (X axes). Bar graphs on the right side of panels C and D show the ITT iAUC of Glo1KD and WT mice.

Data information: Data are presented as mean ± SEM. **P ≤ 0.01 (Mann–Whitney test for line graph in B, Student’s t‐test for scatter plots in B); *P ≤ 0.05 (Student’s t‐test).

Interestingly, while insulin secretion was similar in Glo1KD mice and WT mice at 5 months of age (Fig 3A), it was strongly impaired in Glo1KD mice at 10 months (Fig 3B). In detail, 10‐month‐old WT mice showed a 2.7‐fold increase in insulinemia in the early phase of secretion, which was significantly reduced to 1.3‐fold increase in Glo1KD mice. The age‐dependent impairment of glucose‐stimulated insulin secretion in Glo1KD mice was also evident by analyzing the incremental area under the curve (IAUC, bar graphs in Fig 3).

Figure 3. Evaluation of glucose‐stimulated insulin secretion (GSIS).

• A, B
  Line graphs show the serum insulin levels at 0, 3, and 30 min following the administration of an intraperitoneal glucose bolus (3 g/kg) in mice at 5 months (A, WT n = 6 and Glo1KD n = 4) and 10 months of age (B, n = 6). Bar graphs on the right side of panels show the IAUC monitored during the insulin secretion test in Glo1KD and WT mice.

Data information: Data are presented as mean ± SEM. *P ≤ 0.05, **P ≤ 0.01: 3 and 30 min vs. time 0 of WT mice; ^# P ≤ 0.05, ^## P ≤ 0.01: 3 and 30 min vs. time 0 of Glo1KD mice; ^$ P ≤ 0.05: Glo1KD vs. WT mice (Student’s t‐test).

Pancreas from Glo1KD mice reveals macrophage infiltration and impaired GSIS of isolated islets

Hematoxylin and eosin staining and immunohistochemistry analysis performed in pancreatic sections from 10‐month‐old mice (Fig 4A) showed no differences in the number (Fig 4D), the morphometry (Fig 4E–G), and the insulin (Fig 4H) and glucagon staining (Fig 4I) of the islets from WT and Glo1KD mice. The measurement of insulin and glucagon content in the isolated islets confirmed no differences in the two genotypes (Fig 4 K and L). As shown in panel 4b, the presence of insular inflammatory infiltration was visible with an incidence mean value of 6% at different severity degree (Fig 4C) in the pancreatic sections from Glo1KD mice but not in the sections from WT mice (Fig 4B and C, and Table EV1). More in detail, insular macrophage marker F4/80 was present with a median value of 5% in the pancreatic sections from Glo1KD mice, while no F4/80‐positive cells were found in the sections from WT mice (Fig 4 J and A, right panels). The tissue levels of MGO were 2.5‐fold increased in the pancreas from 10‐month‐old Glo1KD mice compared with WT mice (Fig 4M).

Figure 4. Immunohistochemistry evaluations and hormones measurement in pancreatic islets of 10‐month‐old WT and Glo1KD mice.

• A
  Pancreatic sections (20× magnification) from 10‐month‐old WT and Glo1KD mice stained with hematoxylin and eosin (H&E), anti‐insulin, anti‐glucagon, and anti‐F4/80 antibodies are representative of sections from 7 mice each group.
• B, C
  Insular inflammatory infiltration indicated as incidence (% of infiltrated islets) (B) and severity score (C). Severity of infiltration has been graded as follows: absent (0), low (1), moderate (2), and high (3) (n = 7 mice for each genotype).
• D–G
  Morphometric analysis of islets indicated as the number of islets per sections (D) and the size of islets, calculated as the area (E), the major diameter (F), and the minor diameter (G) of the islets per mouse (n = 7 mice for each genotype).
• H–J
  Insulin and glucagon IS, calculated as the intensity score multiplied with percentage of positive cells (H, I) and the percentage of islets area stained with F4/80 (J) (n = 7 mice for each genotype).
• K, L
  Quantification of insulin (K) and glucagon (L) content in islet extracts by immunoenzymatic assay (n = 7 WT mice vs. n = 10 Glo1KD mice).
• M
  MGO content in pancreatic lysates by LC‐MS/MS (n = 6 mice for each genotype).

Data information: In (B–M), box plots show the minimum and maximum values (ends of the whiskers), interquartile range (length of the box), median (line through the box), and mean (+) values. *P ≤ 0.05, **P ≤ 0.01 (Mann–Whitney test).

Source data are available online for this figure.

mRNA levels of F4/80 and CD68 have been used as macrophage markers. Pancreatic expression levels of both F4/80 encoding gene, adhesion G protein‐coupled receptor E1 (Adgre1), and Cd68 were not different in the two genotypes at 5 months of age (Fig 5A and B), while pancreatic Glo1 expression was reduced by about 50% in Glo1KD mice compared with WT mice (Fig 5C). Ten‐month‐old Glo1KD mice showed increased expression in both Adgre1 and Cd68 in the pancreas, compared with age‐matched WT mice (Fig 5A and B), consistently with the immunohistochemistry analysis. The reduction in Glo1 expression found in 5‐month‐old Glo1KD mice was maintained in these mice at 10 months of age, compared with age‐matched WT mice (Fig 5C). As shown in some insulin target tissues in Fig 1, the expression of both Aldh1a3 (Fig EV1A) and Akr1b3 (Fig EV1B) was unaltered in the pancreas from WT and Glo1KD mice.

Figure 5. Macrophage markers gene expression in pancreatic tissue, isolated islets, and ex vivo GSIS.

• A–C
  Data distribution of the absolute quantification of Adgre1 (A), Cd68 (B), and Glo1 (C) gene expression measured by qPCR in pancreatic tissue from WT and Glo1KD mice at 5 (upper panels, WT n = 9, Glo1KD n = 7 in A, C and n = 8 in B) and 10 months of age (lower panels, WT n = 7, Glo1KD n = 9 in A, B and n = 10 in C).
• D, E
  Data distribution of the absolute quantification of Adgre1 (D) and Cd68 (E) gene expression measured by qPCR in the isolated islets from WT and Glo1KD mice at 5 (upper panels, WT n = 7 in D and 6 in E, Glo1KD n = 7) and 10 months of age (lower panels, WT n = 7, Glo1KD n = 7 in D and 6 in E). Cyclophilin A was used as reference gene. Gene expression is shown as the ratio of Adgre1, Cd68, or Glo1 to Cyclophilin A copy number (AU).
• F, G
  Ex vivo GSIS showed as fold of increase (F) and as insulin concentration released into the secretion media (G) in response to the basal condition at 2.8 mM glucose and to the stimulated condition at 16.7 mM glucose, of pancreatic islets from WT (n = 8) and Glo1KD mice (n = 9) at 10 months of age. Insulin concentration was normalized on DNA islets content.

Data information: Bars in the scatter plots show the mean ± SEM. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001: WT vs. Glo1KD mice; ##P ≤ 0.01, ###P ≤ 0.001: 16.7 mM glucose vs. 2.8 mM glucose (Mann–Whitney test).

Figure EV1. Expression levels of alternative detoxification enzymes of MGO in the pancreatic tissue.

• A, B
  Data distribution of the absolute quantification of Aldh1a3 (A) and Akr1b3 (B) gene expression measured by qRT–PCR in the pancreas of WT and Glo1KD mice (n = 5). Cyclophilin A was used as reference gene. Gene expression is shown as the ratio of Aldh1a3 or Akr1b3 to Cyclophilin A copy number (AU).

Data information: Bars in the scatter plots show the mean ± SEM. Statistical analysis was performed by Mann–Whitney test (A) or Student’s t‐test (B), where P ≤ 0.05 was considered statistically significant.

Increased Adgre1 and Cd68 expression was confirmed in the isolated islets from 10‐month‐old WT and Glo1KD mice, while no difference was revealed in the islets from 5‐month‐old mice (Fig 5D and E). The expression of integrin subunit alpha x (Itgax) has been evaluated in the isolated islets as further marker of M1 macrophages. Its expression was increased in the islets from Glo1KD vs. WT mice at 10 months of age (Fig EV2B), but not at 5 months of age (Fig EV2A).

Figure EV2. Gene expression of the macrophage marker Itgax in isolated islets.

• A, B
  Data distribution of the absolute quantification of Itgax gene expression measured by qPCR in the isolated islets from WT and Glo1KD mice at 5 (A, WT n = 6 and Glo1KD n = 8) and 10 months of age (B, WT n = 7 and Glo1KD n = 6).

Data information: Cyclophilin A was used as reference gene. Gene expression is shown as the ratio of Itgax to Cyclophilin A copy number (AU). Bars in the scatter plots show the mean ± SEM. **P ≤ 0.01 (Mann–Whitney test).

Differently from the pancreas, no differences in Adgre1 expression were detected in other tissues involved in maintaining of glucose homeostasis, namely liver (Fig EV3A), white adipose tissue (Fig EV3B), and skeletal muscle (Fig EV3C).

Figure EV3. Adgre1 gene expression in other tissues.

• A–C
  Data distribution of the absolute quantification of Adgre1 gene expression measured by qRT–PCR in liver (A), white adipose tissue (WAT, perigonadal fat) (B), and skeletal muscle (C) of WT (n = 7 in A, C and n = 6 in B) and Glo1KD mice (n = 7) at 10 months of age. Cyclophilin A was used as reference gene.

Data information: Gene expression is shown as the ratio of Adgre1 to Cyclophilin A copy number (arbitrary units, AU). Bars in the scatter plots show the mean ± SEM. Statistical analysis was performed by Mann–Whitney test (A, B) or Student’s t‐test (C), where P ≤ 0.05 was considered statistically significant.

In addition to the cellular components, pancreatic inflammation has also been tested by the local presence of inflammatory soluble factors, which are known to be involved in the cross talk between β‐cells and macrophages and to be associated with β‐cell dysfunction (Donath et al, ²⁰¹³; Butcher et al, ²⁰¹⁴). Multiplex immunoassays revealed a 40% increase in both MCP‐1 and TNF‐α concentration in the pancreas lysate of 10‐month‐old Glo1KD mice, compared with WT mice. IL‐1β pancreatic levels were not found to be significantly higher in Glo1KD mice than WT mice (Table 2). The expression levels of MCP‐1, TNF‐α, and IL‐1β mRNA were significantly increased in the pancreatic islets isolated from 10‐month‐old Glo1KD mice, compared with WT mice (Fig EV4).

Table 2.

Cytokine levels in pancreatic tissue.

	WT mice	Glo1KD mice
IL‐1β (pg/ml × 1/mg of proteins)	80.39 ± 3.68	90.47 ± 6.08
MCP‐1 (pg/ml × 1/mg of proteins)	845.48 ± 55.64	1212.01 ± 137.88*
TNF‐α (pg/ml × 1/mg of proteins)	678.22 ± 46.54	955.57 ± 94.84*

Mean ± SEM values of IL‐1β, MCP‐1, and TNF‐α in pancreatic lysate from 10‐month‐old WT (n = 7) and Glo1KD (n = 10) mice are reported in table. Statistical analysis was performed by Student’s t‐test for data with a normal distribution, or with Mann–Whitney test for data with a not normal distribution (*P < 0.05, 10‐month‐old Glo1KD vs. WT mice).

Figure EV4. Expression levels of IL‐1β, MCP‐1, and TNF‐α in the isolated islets.

• A–C
  Data distribution of the absolute quantification of IL‐1β (A), MCP‐1 (B), and TNF‐α gene expression measured by qRT–PCR in the islets isolated from WT and Glo1KD mice (n = 6).

Data information: Cyclophilin A was used as reference gene. Gene expression is shown as the ratio IL‐1β, MCP‐1, or TNF‐α to Cyclophilin A copy number (AU). Bars in the scatter plots show the mean ± SEM. *P ≤ 0.05 (Student’s t‐test).

The observed pro‐inflammatory microenvironment associates with an impaired GSIS ex vivo of the isolated islets (Fig 5F and G). A 2.5‐fold increase in insulin secretion in response to glucose stimulation, observed in the islets from WT mice, was decreased to 1.2‐fold increase in the islets from 10‐month‐old Glo1KD mice.

Glo1KD isolated islets express senescence markers

Unveiled by aging, the phenotype observed in Glo1KD mice recalls typical traits described by Aguayo‐Mazzucato et al (2017, 2019) in senescent islets induced by aging and metabolic stress. To test whether senescence may be the mechanism sustained by MGO accumulation in our model, we tested the expression of the senescence markers transformation‐related protein 53 (Trp53) and cyclin‐dependent kinase inhibitor 1A (Cdkn1a) (Rufini et al, ²⁰¹³), besides two novel markers of age‐related β‐cell functional decline, insulin‐like growth factor 1 receptor (Igf1r) and fibroblast growth factor receptor 1 (Fgfr1) (Aguayo‐Mazzucato et al, ²⁰¹⁷). mRNA levels of Trp53, Cdkn1a, and Igf1r were not significantly different in pancreatic islets isolated from both 5‐month‐old WT and Glo1KD mice (Fig 6A, C and E). Conversely, the expression of Trp53, Cdkn1a, and Igf1r was increased by 48%, 88%, and 39%, respectively, in the pancreatic islets isolated from 10‐month‐old Glo1KD mice, compared with WT islets (Fig 6B, D and F). The expression of Fgfr1 was reduced by 37% in the islets from both 5‐ and 10‐month‐old Glo1KD mice, compared with the islets from age‐matched WT mice (Fig 6G and H).

Figure 6. Trp53, Cdkn1a, Igf1r, and Fgfr1 gene expression in isolated islets.

• A–H
  Data distribution of the absolute quantification of Trp53 (A, B), Cdkn1a (C, D), Igf1r (E, F), and Fgfr1 (G, H) gene expression measured by qPCR in pancreatic islets of WT and Glo1KD mice at 5 months (A, C, E, and G; WT n = 7, Glo1KD n = 7 in C and n = 8 in A, E and G) and 10 months of age (B, D, F, and H; WT n = 6 in D and n = 8 in B, F, and H, Glo1KD n = 6). β‐Actin was used as reference gene. Gene expression is shown as the ratio of Trp53, Cdkn1a, Igf1r, or Fgfr1 to β‐Actin copy number (AU).

Data information: Bars in the scatter plots show the mean ± SEM. *P ≤ 0.05, **P ≤ 0.01 (Student’s t‐test).

MGO impairs GSIS in vitro in a β‐cell line through a senescence‐mediated mechanism

To confirm that the β‐cell dysfunction found in the Glo1KD model was an effect mediated by: (i) the MGO increase and (ii) a consequent induction of senescence, we exposed INS‐1 832/13 cells to MGO in vitro and looked at some functional and senescence markers. Forty‐eight hours of exposure to MGO induced an increase in both the mRNA (Fig 7A and B) and protein levels (Fig 7C) of two canonical inducers of cellular senescence: p53 and p21. In detail, an increase of 35% of Trp53 expression resulted in a twofold increase of p53 protein levels, and an increase of 30% of Cdkn1a expression resulted in a 2.5‐fold increase of p21 protein levels in INS‐1 832/13 cells exposed to MGO, as compared to not‐treated control cells. Further 24 h of cell treatment with dasatinib (31 nM) and quercetin (6 μM), used in combination as senolytic agents (Aguayo‐Mazzucato et al, ²⁰¹⁹), were able to reduce both the mRNA and protein expression of p53 and p21 at levels comparable to controls (Fig 7A–C). Several cells entering senescence lost nuclear high mobility group box 1 (HMGB1), depending on p53 activity (Davalos et al, ²⁰¹³). The exposure to MGO also induced the nuclear exclusion of HMGB1 in INS‐1 832/13, while the nuclear protein levels were rescued by the senolytics treatment (Fig 7D). Furthermore, the senescence‐activated β‐galactosidase (β‐Gal) activity was tested as marker of senescence in our experimental model. Under MGO treatment, the percentage of β‐Gal⁺ cells increased from 1.9% to 5.4% and was reversed by senolytics (Fig 7E).

Figure 7. Senolytics effect on MGO‐treated INS‐1 832/13 β‐cell line.

• A, B
  mRNA levels of Trp53 and Cdkn1a in INS‐1 832/13 cells treated with MGO 0.05 mM for 48 h or with MGO plus 24 h of dasatinib 31 nM and quercetin 6 μM (D/Q) (n = 6 biological replicates for each group). Gene expression is measured by qPCR as relative quantification to Cyclophilin A, used as reference gene.
• C
  Whole cell protein lysates of INS‐1 832/13 β‐cells analyzed by Western Blot with anti‐p53 and anti‐p21 antibodies; anti‐vinculin antibody was used as loading control (n = 9 biological replicates for each treatment group).
• D
  Nuclear protein lysates of INS‐1 832/13 β‐cells analyzed by Western Blot with anti‐HMGB1 antibody; anti‐histone H3 antibody was used as loading control (n = 5 biological replicates for each treatment group).
• E
  Proportion of β‐Gal⁺ β‐cells analyzed by a fluorescence‐based assay using flow cytometry (n = 4 biological replicates for each treatment group).
• F
  Cytokine levels measured in the β‐cell culture medium by a multiplex immunoassay are shown as fold increase over not‐treated control cells (n = 6 biological replicates for each treatment group).
• G
  GSIS measured in the β‐cell culture medium by an ELISA assay. Insulin concentration is reported at basal condition (2.5 mM glucose) and in response to glucose stimulation (16.7 mM) (n = 4 biological replicates for each treatment group).

Data information: Data in bar graphs are presented as mean ± SEM. In C and D, blots show representative image of replicate experiments; protein levels were quantified by the densitometric analysis of independent experiments and showed in the bar graphs as the percentage over not‐treated control cells. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001: MGO vs. CTR; #P ≤ 0.05, ##P ≤ 0.01, ###P ≤ 0.001: D/Q vs. MGO (Mann–Whitney test in A–D and F, Student’s t‐test in E and G).

INS‐1 832/13 cells treated with MGO showed an increased secretion of the senescence‐associated secretory profile (SASP) cytokines: IL‐1β, MCP‐1, and TNF‐α, by 40%, 35%, and 30%, respectively (Fig 7F). This increase was blunted in MGO‐exposed cells by dasatinib and quercetin combined treatment (Fig 7F).

The observed effects associate with the modulation of GSIS. Indeed, while not‐treated INS‐1 832/13 cells showed a sevenfold increase in insulin secretion in response to glucose, MGO‐treated cells showed a fourfold increase, which was significantly lower than the insulin secretion observed in not‐treated control cells (Fig 7G). Interestingly, treatment with dasatinib and quercetin was able to rescue the GSIS in INS‐1 832/13 cells exposed to MGO (Fig 7G).

A decline in mtDNA has been reported in human pancreatic islets with aging (Cree et al, ²⁰⁰⁸) and its depletion in β‐cells directly decreases GSIS (Nile et al, ²⁰¹⁴). Levels of mitochondrial transcription factor A (mtTFA), which plays a key role in regulating of mtDNA replication, are proportional to mtDNA (Choi et al, ²⁰⁰¹). Here, we found a 30% reduction in both mtDNA content and mtTFA expression in INS832/13 exposed to MGO (Fig EV5).

Figure EV5. MGO effect on mtDNA content and mtTFA expression.

1. Quantitative RT–PCR detection of mtDNA content relative to nuclear DNA (nucDNA) in INS832/13 β‐cells exposed or not to MGO (n = 10 biological replicates for each group).
2. Quantitative RT–PCR detection of mtTFA expression levels in INS832/13 β‐cells exposed or not to MGO. Gene expression data are shown as relative to Cyclophilin A, used as reference gene (n = 7 biological replicates for each group).

Data information: Data are presented as mean ± SEM. In A, **P ≤ 0.01 (Student’s t‐test). In B, ***P ≤ 0.001 (Mann–Whitney test).

These data provide the proof of concept that the β‐cell dysfunction induced by MGO is mediated by a senescence‐associated phenotype.

Discussion

Our work demonstrates that a metabolic unbalance promoting MGO accumulation is able to alter glucose homeostasis. This study shows for the first time, indeed, that the knockdown of Glo1 induces the development of glucose intolerance and β‐cell dysfunction in C57BL/6 mice. We also demonstrate that the β‐cell dysfunction associates with the presence of a senescence‐related pro‐inflammatory microenvironment both in the pancreas of Glo1KD mice and in MGO‐treated β‐cells in vitro.

Methylglyoxal levels are elevated in diabetic patients and animal models of diabetes (Rabbani & Thornalley, 2015; Huang et al, ²⁰¹⁶; Maessen et al, ²⁰¹⁶). Most studies analyzing the consequences of dicarbonyl stress have been carried out in their complex metabolic milieu. The advantage of using the Glo1KD mouse is the opportunity to evaluate whether and which alterations result from the MGO increase by itself, excluding the presence of confounding factors deriving from hyperglycemia and obesity. Previous works on the Glo1KD mice have proved the onset of diabetes‐like kidney morphological alterations (Giacco et al, ²⁰¹⁴) and both thermal and mechanical hyperalgesia (Bierhaus et al, ²⁰¹²) as a consequence of protein modifications by MGO. However, no evidence on the role of Glo1 knockdown in mouse glucose metabolism is presently available.

Differently from the Glo1KO model (Schumacher et al, ²⁰¹⁸), partial deletion of Glo1 causes the accumulation of endogenous MGO with age but not the increase in the alternative enzymes for MGO detoxification, ALDH and AKR. These data suggest that when Glo1 is decreased but not completely abolished, it fails to keep down MGO levels with age, and compensatory mechanisms of MGO detoxification are not activated.

In the current study, we demonstrate that the knockdown of Glo1 does not have an effect either on feeding behavior or on body weight, confirming previous findings by Wortmann et al (2016). In their work, no difference in body composition (i.e., total body weight, relative fat and lean mass) and energy expenditure in 5‐month‐old Glo1KD mice compared with WT mice was reported (Wortmann et al, ²⁰¹⁶). Here, we have further expanded the monitoring of glucose metabolism until 10 months. While no significant difference occurs in both fasting and fed glycemia, in serum insulin and glucagon levels of Glo1KD mice compared with WT mice, either at 5 or 10 months of age, Glo1KD mice develop an impairment of glucose tolerance with age.

This study has been performed in age‐matched male mice. While the use of both sexes would give additional insight in the potential impact of sex differences on the observed phenotype, we chose to use male mice to reduce the variability due to the known sex‐dependent biological difference in glucose metabolism (Mauvais‐Jarvis et al, ²⁰¹⁷; Dockman et al, ²⁰²²). Our analyses are based on the comparison of age‐matched WT and Glo1KD mice; therefore, the age‐related impairment of glucose tolerance in the mouse strain C57bl/6 does not impact on genotype‐related phenotype.

We have previously demonstrated that MGO administration impairs insulin sensitivity in C57BL/6 mice (Nigro et al, ²⁰¹⁴). However, Glo1KD mice do not show systemic insulin resistance, either at 5 or at 10 months of age. The difference in the insulin sensitivity between mice exposed to MGO and the genetically modified Glo1KD mice may depend on the intrinsic differences of the two models. Mice exposed to chronic exogenous administration of MGO showed a large 2.5‐fold increase in serum MGO (Nigro et al, ²⁰¹⁴). A smaller increase has been measured in the serum of Glo1KD mice, better reflecting the in vivo conditions established when the endogenous accumulation of MGO occurs (Skapare et al, ²⁰¹³; Giacco et al, ²⁰¹⁴; Masania et al, ²⁰¹⁶; Nigro et al, ²⁰¹⁷).

Studies in experimental models have demonstrated that MGO causes pancreatic β‐cell dysfunction (Dhar et al, ²⁰¹¹; Fiory et al, ²⁰¹¹; Elmhiri et al, ²⁰¹⁴; Bo et al, ²⁰¹⁶) and, therefore, may contribute to the pathogenesis of diabetes. The Glo1KD model gave us the possibility to investigate the glucose homeostasis in a nondiabetic condition where the endogenous increase in MGO is the only trigger of metabolic imbalance. As expected, insulin secretion in response to glucose physiologically decreases with age in WT C57BL/6 mice. However, consistent with the impairment of glucose tolerance, in this study, we demonstrate that Glo1KD mice develop a significant defect in glucose‐stimulated insulin secretion, tested both in vivo and ex vivo in isolated islets at 10 months of age. Both glucose tolerance and insulin secretion were not impaired in younger animals at 5 months of age. These results, together with the unaltered insulin sensitivity, support the hypothesis that glucose intolerance is a result of age‐related impairment of β‐cell function developed by Glo1KD mice.

Although ipGTT and GSIS are impaired in 10‐month‐old Glo1KD mice, but not in 5‐month‐old mice, neither fast nor fed blood glucose levels are significantly different in Glo1KD compared with WT mice. One can suggest that Glo1KD mice metabolic phenotype recapitulates with age some endophenotypes of diabetes; however, they do not feature overt diabetes, as shown by normal levels of fed and fast glycemia.

Unveiled by aging, this phenotype suggests a potential role for a senescence‐associated mechanism triggered by MGO.

Immunohistochemistry analysis has revealed an association between the impaired insulin secretion and the presence of inflammatory infiltration in the islets from Glo1KD mice. The immune system plays a significant role in the regulation of metabolism (Osborn & Olefsky, 2012). A large number of immune cells, predominantly macrophages, have been detected in the islets of patients with T2D and in animal models in conjunction with increased levels of cytokines and chemokines (Ehses et al, ²⁰⁰⁷; Donath et al, ²⁰¹³). The contribution of other immune cells besides macrophages to islet cell infiltration in T2D is less clear (Marchetti, 2016). We found the presence of macrophages in the islets of Glo1KD mice. As described in T2D patients, macrophages are dispersed through and around the pancreatic islets of Glo1KD mice (Donath et al, ²⁰¹³). Besides the histological staining for the macrophage marker F4/80, the mRNA expression levels of F4/80 (Adgre1) and the other macrophage markers, CD68 and Itgax, are higher in the isolated islets from Glo1KD mice compared with WT mice. Although these are extensively used as macrophage markers (Marchetti, 2016; Chan et al, ²⁰¹⁹; He et al, ²⁰¹⁹; Mukhuty et al, ²⁰²¹), CD68 has been found to be increased in aged and β‐Gal⁺ β‐cells (Aguayo‐Mazzucato et al, ^{2017, 2019}; Ebrahimi et al, ²⁰²⁰). In line with this, CD68 gene expression increases with aging also in the islets of our control WT mice (Fig EV6B). Conversely, neither Adgre1 (Fig EV6A) nor Itgax (Fig EV6C) expression is increased in the islets of older (10‐month‐old) versus younger (5‐month‐old) WT mice, as also reported in a previous work by Aguayo‐Mazzucato et al (2017), where Adgre1 and Itgax expression is even reduced in old β‐cells. This evidence suggests that, in our experimental condition, while CD68 could be expressed by both the senescent cells and macrophages in the islets of aging mice, the source of Adgre1 and Itgax is derived from the macrophagic but not the senescent component. The increased expression of both Adgre1 and Itgax confirms the macrophage infiltration in the islets from Glo1KD compared with WT mice at 10 months of age. These data indicate that, similarly to the impairment of insulin secretion, this inflammatory phenotype is fostered by Glo1 knockdown during aging.

Figure EV6. Gene expression of Adgre1, CD68, and Itgax in isolated islets from old versus young WT mice.

• A–C
  Data distribution of the absolute quantification of Adgre1 (A), CD68 (B), and Itgax (C) gene expression measured by qPCR in the isolated islets from WT mice at 5 (n = 7 in A and n = 6 in B, C) and 10 months (n = 7) of age.

Data information: Cyclophilin A was used as reference gene. Gene expression is shown as the ratio of Adgre1, CD68, or Itgax to Cyclophilin A copy number (AU). Bars in the scatter plots show the mean ± SEM. *P ≤ 0.05 (Student’s t‐test).

Data obtained in other tissues crucial to the regulation of glucose tolerance suggest that macrophage infiltration is specific to pancreas. Indeed, we found no difference in the expression of the F4/80 coding gene, Adgre1, in the liver, adipose tissue, and skeletal muscle of Glo1KD mice compared with WT mice.

Cytokine and chemokine levels were previously identified as molecular signature of islet inflammation in T2D (Marchetti, 2016). Among these, we focused on those associated with β‐cell failure. It is known that IL‐1β is produced by β‐cells exposed to high glucose, contributing to islet glucotoxicity and impaired β‐cell function (Maedler et al, ²⁰⁰²; Mahdi et al, ²⁰¹²). Furthermore, IL‐1β sustains the autocrine and paracrine activation of both β‐cells and macrophages, exacerbating the chronic inflammatory responses in the islets (Donath et al, ²⁰¹³). Butcher et al (2014) demonstrated that TNF‐α and MCP‐1 expression is higher in leucocyte‐infiltrated islets of T2D subjects and negatively correlates with first‐phase glucose‐stimulated insulin secretion of the isolated islets.

In this study, we show that TNF‐α and MCP‐1 levels are higher in the pancreas from Glo1KD mice compared with WT mice, confirming an association between the local increase in these cytokines and the impaired insulin secretion ability by β‐cells in response to glucose. Low levels of IL‐1β were detected in the pancreatic lysates, and the increase is not statistically different in the Glo1KD compared with WT tissues. However, the mRNA expression of IL‐1β, TNF‐α, and MCP‐1 was specifically measured in the isolated islets and resulted to be higher in the islets from Glo1KD mice compared with WT mice. In accordance with the study by Butcher M.J. et al. in the islets from T2D patients (Butcher et al, ²⁰¹⁴), we provide evidence that the accumulation of MGO is sufficient to increase TNF‐α and MCP‐1, suggesting a causal role of the pro‐inflammatory microenvironment in the β‐cell dysfunction.

TNF‐α and IL‐1β are known to be primary cytokines regulating the function of several other cytokines and chemokines, including MCP‐1, thereby contributing to the recruitment of immune cells. Furthermore, both β‐cells and macrophages are potential sources of pro‐inflammatory cytokines within the islets of patients with T2D (Maedler et al, ²⁰⁰²; Morris, ²⁰¹⁵).

The experiments performed in vitro in INS‐1 832/13 cells confirm that MGO is able to induce the β‐cell secretion of all the three cytokines analyzed, TNF‐α, MCP‐1, and IL‐1β. In light of these in vitro data, it is conceivable that in the analysis performed ex vivo from the whole pancreas of mice, the IL‐1β increase could be hidden by the contribution of other tissue components.

The aging‐dependent onset of β‐cell dysfunction in Glo1KD mice highlights a potential role of senescence as mechanism induced by MGO. In support of our evidence, different islet phenotypes have been identified in T2D and are characterized by the differential expression of novel senescence markers. The latter can also be induced by both aging and metabolic stress (Aguayo‐Mazzucato et al, ²⁰¹⁷). These findings were followed by the description of a SASP actively produced by senescent β‐cells, which plays not only an autocrine but also a paracrine effect on surrounding cells. It is worth noting that the cytokines analyzed in our models: TNF‐α, MCP‐1, and IL‐1β are part of the SASP secreted by senescent β‐cells (Aguayo‐Mazzucato et al, ²⁰¹⁹).

We hypothesized that the dicarbonyl stress induced by the knockdown of Glo1 acts as the metabolic hit leading to the senescence of pancreatic islets, which foster the onset of a pro‐inflammatory condition, increasing macrophage infiltration and thereby resulting in impaired GSIS. Indeed, aged islets have been described to have impaired GSIS, but still maintain their insulin‐producing capability and β‐cell number. As a matter of fact, no difference in the number and size of islets nor in the islet insulin storage was found in Glo1KD and WT mice. To prove this hypothesis, we have demonstrated that isolated islets from 10‐month‐old Glo1KD mice show a higher expression of the senescence markers Trp53 and Cdkn1a, together with the modulation of the novel β‐cell age markers Igf1r and Fgfr1 (Rufini et al, ²⁰¹³; Aguayo‐Mazzucato et al, ²⁰¹⁷). A lower expression of Fgfr1 is already evident at 5 months of age in the islets of Glo1KD mice compared with the WT littermates.

It has been described that the β‐cell expression of Igf1r associates with senescence and dysfunctional insulin secretion (Aguayo‐Mazzucato et al, ²⁰¹⁷). As the reduced expression of Fgfr1 has been associated with β‐cell age but not with impaired insulin secretion, it is conceivable that Fgfr1 modulation may precede the functional alteration found later at 10 months of age. In mesenchymal stem cells, the FGFR1 activation prevents cellular senescence acting upstream MDM2 and its mediated degradation of p53 (Coutu et al, ²⁰¹¹).

Upregulation of p53 by MGO and Glo1 depletion has been extensively documented in vitro (Zhou et al, ²⁰¹⁵; Chen et al, ²⁰¹⁷; Braun et al, ²⁰¹⁹; Lai et al, ²⁰²⁰). Our data in INS‐1 832/13 cells demonstrate that MGO itself leads to the β‐cell dysfunction developed by Glo1KD mice. Indeed, β‐cells exposed to MGO show an impairment of GSIS and higher levels of the senescence inducers p53 and the downstream effector p21, together with the nuclear exclusion of HMGB1 and the increase in the SA‐β‐Gal‐positive cells. Although the increase in SA‐β‐Gal‐positive INS‐1 cells may look to be modest in response to MGO, previous evidence from our group and others demonstrated how a similar increase (threefold or less) in the percentage of senescent cells is sufficient to impair cell function and interfere with tissue homeostasis (Chen et al, ²⁰⁰²; Xu et al, ²⁰¹⁸; Spinelli et al, ²⁰²²). Interestingly, β‐cell treatment with a combination of senolytic agents (dasatinib and quercetin) (Aguayo‐Mazzucato et al, ²⁰¹⁹) demonstrates that the clearance of senescent cells is able not only to dampen pro‐inflammatory cytokines secretion but also to rescue the MGO‐impaired GSIS. We and others have previously identified GSK3 and mitochondrial dysfunction as mediators of MGO‐induced effect on GSIS impairment in INS‐1 cells (Fiory et al, ²⁰¹¹; Bo et al, ²⁰¹⁶). Both of them can drive the senescent phenotype (Souder & Anderson, 2019; Vasileiou et al, ²⁰¹⁹). Besides being reduced by aging, a decrease in mtDNA copy number has been shown to precede the T2D development, suggesting that the decrease in mtDNA content might be related to the disease pathogenesis (Lee et al, ¹⁹⁹⁸; Cree et al, ²⁰⁰⁸). We gained data on the effect of MGO on the reduction of both mtDNA content and its transcription regulator, mtTFA, expression in INS‐1 832/13.

Our results demonstrate for the first time that the MGO harmful effect on insulin secretion is mediated, at least in part, by cellular senescence. Further studies involving the use of senolytic agents in vivo will be useful to draw the clinical implications of the phenotype described here for the Glo1KD mouse. The clearance of senescent cells would reduce the SASP factors release, which contributes to the related inflammation and metabolic dysregulation, acting at paracrine level, thus affecting tissue function (Kirkland & Tchkonia, 2017). Therefore, alleviating the senescent cells burden in vivo will result in a beneficial effect on the senescent β‐cell microenvironment and the whole tissue function.

Conclusions

We have used a unique model in which the increase in MGO levels is due to the partial deletion of Glo1 to demonstrate that MGO accumulation induces a pancreatic damage similar to that observed in T2D. While Glo1KD mice are not hyperglycemic, they develop an age‐dependent glucose intolerance due to a senescence‐associated pro‐inflammatory phenotype of islets resulting in GSIS impairment. These data provide new evidence for an active role of MGO in the etiology of T2D, paving the way for novel prevention approaches to T2D progression.

Materials and Methods

Mice

The transgenic Glo1KD were a kind gift of the laboratory of Prof. Peter Nawroth (Department of Medicine I and Clinical Chemistry, Heidelberg University Hospital, Heidelberg, Germany) and were generated on a C57BL/6 background, as previously described (El‐Osta et al, ²⁰⁰⁸). Mice were housed in a specific pathogen free facility, in a temperature‐controlled (22°C) room with a 12‐h light/dark cycle. All the experiments were conducted in age‐matched male mice with the approval of the local ethics committee of the Ministry of Education, University and Research (MIUR) [approval n° 252/2018‐PR]. Three mice were allocated per cage to minimize distress and potential confounders in the evaluation of metabolic parameters. This research has been performed in adherence of ARRIVE guidelines 2.0.

Cell culture and reagents

INS‐1 832/13 rat insulinoma cells (cod. SCC207, Sigma‐Aldrich, St Luis, MO, USA) were cultured in T75 tissue culturing flasks at 37°C in a humidified atmosphere containing 5% CO2 in RPMI 1640 with sodium bicarbonate, supplemented with 10% (vol/vol) fetal calf serum (Gibco, Thermo Fisher Scientific, Waltham, MA USA), 10 mm HEPES (pH 7.4), 2 mm l‐glutamine, 1 mm sodium pyruvate, and 50 μm β‐mercaptoethanol (Sigma‐Aldrich, St Luis, MO, USA). Cells were authenticated and tested for mycoplasma contamination. Where indicated, cells were exposed to MGO (Sigma‐Aldrich, St Luis, MO, USA) 0.05 mM as previously described (Fiory et al, ²⁰¹¹) for 48 h and treated or not with dasatinib (31 nM) and quercetin (6 μM) in combination (Sigma‐Aldrich, St Luis, MO, USA), for additional 24 h.

Measurement of MGO

The MGO content in the pancreas tissue and plasma was determined by isotope dilution and tandem mass spectroscopy, following derivatization with 1,2‐diaminobenzene (Rabbani & Thornalley, ^2014b). Briefly, 20 µl of serum were precipitated by the addition of 10 µl of ice‐cold 20% (wt/vol) trichloroacetic acid in 0.9% (wt/vol) sodium chloride (20 µl) and water (40 µl). For the pancreas tissue, pre‐weighted amounts (ca.10 mg) were homogenized in ice‐cold 20% (wt/vol) trichloroacetic acid in 0.9% (wt/vol) sodium chloride (20 µl) and water (80 µl). An aliquot (5 µl) of the internal standard ([¹³C₃]‐methylglyoxal) was then added and the samples vortexed mixed. Following centrifugation (14,000 rpm; 5 min @ 4°C), 35 µl of the supernatant was transferred to HPLC vials containing a 200 µl glass interest. An aliquot (5 µl) of 3% sodium azide (wt/vol) was then added to each sample followed by 10 µl of 0.5 mM DB in 200 mM HCl containing 0.5 mM diethylenetriaminepentaacetic acid (DETAPAC) in water. The samples were then incubated for 4 h at room temperature, protected from the light. Samples were then analyzed by LC‐MS/MS using an ACQUITY™ ultra‐high‐performance liquid chromatography system with a Xevo‐TQS LC‐MS/MS mass spectrometer (Waters, Manchester, UK). The column was a Waters BEH C18 (100 × 2.1 mm) and guard column (5 × 2.1 mm). The mobile phase was 0.1% formic acid in water with a linear gradient of 0–100% 0.1% formic acid in 50% Acetonitrile:water over 0–10 min; the flow rate was 0.2 ml/min, and column temperature was 5°C. The capillary voltage was 0.5 kV, the cone voltage 20 V, the interscan delay time 100 ms, the source and desolvation gas temperatures 150 and 350°C, respectively, and the cone gas and desolvation gas flows were 150 and 800 l/h, respectively. Mass transitions (parent ion > fragment ion; collision energy), retention time, limit of detection, and recoveries were as follows: 145.0 > 77.1; 24 eV, 5.93 min, 0.52 pmol and 98%. Acquisition and quantification were completed with MassLynx 4.1 and TargetLynx 2.7 (Waters^®).

Reverse transcription and quantitative Real‐Time PCR (qPCR)

RNA was isolated from mouse tissues (skeletal muscle, perigonadal adipose tissue, liver, brain, pancreas, and pancreatic islets) or INS‐1 832/13 cells using miRNeasy mini kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions. After quantification with NanoDrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA, USA), reverse transcription of 1 μg of total RNA was performed using Super Script III (Life Technologies, Carlsbad, CA, USA), according to the manufacturer's instructions. RNA samples which did not meet the quality cut‐off (e.g., 260/280 ratio < 1.8) were excluded from the analysis of gene expression. qPCR was performed in triplicate by using iQ SYBR Green Super mix on iCycler Real‐Time Detection System (Bio‐Rad Laboratories, Hercules, CA, USA). Cyclophilin A and β‐Actin were used as reference genes. Gene expression data from cells are shown as relative quantification by the use of the 2^{−ΔΔ C t} method. Gene expression data from mice are shown as the absolute quantification normalized on reference gene expression (ratio of gene of interest copy number to reference gene copy number; Schmittgen & Livak, ²⁰⁰⁸).

Primer‐Blast (http://www.ncbi.nlm.nih.gov/tools/primer‐blast/) was used to design primers specific, which were then purchased from Sigma‐Aldrich (St Louis, MO, USA). Primers used for qPCR are as follows:

β‐Actin mouse	Forward: 5’‐ACCGTGAAAAGATGACCCAG‐3’
	Reverse: 5’‐GTACGACCAGAGGCATACAG‐3’
Adgre1 mouse	Forward: 5’‐ACAGTACGATGTGGGGCTTT‐3’
	Reverse: 5’‐GTGTGGTCATCCCCCATCTG‐3’
Akr1b3 mouse	Forward: 5’‐GGCCGTGAAAGTTGCTATTG‐3’
	Reverse: 5’‐ATGCTCTTGTCATGGAACGTG −3’
Aldh1a3 mouse	Forward:: 5’‐GGGTCACACTGGAGCTAGGA‐3’
	Reverse:: 5’‐CTGGCCTCTTCTTGGCGAA‐3’
Cdkn1a mouse	Forward: 5’‐ GCAGACCAGCCTGACAGATTT‐3’
	Reverse: 5’‐ CTGACCCACAGCAGAAGAGG‐3’
Cdkn1a rat	Forward: 5’‐TCGTACCCCGATACAGGTGA‐3’
	Reverse: 5’‐ TGTCTAGGAAGCCCTCCCG‐3’
Cd68 mouse	Forward:: 5’‐ GCTAGGACCGCTTATAGCCC‐3’
	Reverse:: 5’‐ GGATGGCAGGAGAGTAACGG‐3’
Cyclophilin A mouse/rat	Forward: 5'‐GCAGACAAAGTTCCAAAGACAG‐ 3'
	Reverse: 5'‐CACCCTGGCACATGAATCC‐ 3'
Glyoxalase 1 mouse	Forward: 5’‐CCCTCGTGGATTTGGTCACA‐3′
	Reverse: 5′‐AGCCGTCAGGGTCTTGAATG‐3′
Fgfr1 mouse	Forward: 5’‐CCTGAACAAGATGCACTCCCA‐3’
	Reverse: 5′‐TCTGGGGATGTCCAGTAGGG‐3’
Igf1r mouse	Forward: 5’‐ATTCTGATGTCTGGTCCTTCG‐3’
	Reverse: 5′‐AGCATATCAGGGCAGTTGTC‐3’
Il‐1β mouse	Forward: 5’‐CCTCATCCTGGAAGGTCCAC‐3’
	Reverse: 5′‐AACCTTTGACCTGGGCTGTC‐3’
Itgax mouse	Forward: 5’‐ GCAGAGCCAGAACTTCCCAA −3’
	Reverse: 5’‐ TGCTACCCGAGCCATCAATC −3’
Mcp‐1 mouse	Forward: 5’‐CTGTAGTTTTTGTCACCAAGCTCA‐3’
	Reverse: 5′‐GTGCTGAAGACCTTAGGGCA‐3’
mtTFA rat	Forward: 5’‐GATGAGTCAGCTCAGGGGAA‐3’
	Reverse: 5’‐CACACTGCGACGGATGAGAT‐3’
Tnf‐α mouse	Forward: 5’‐AGCCCCCAGTCTGTATCCTT‐3’
	Reverse: 5′‐CTCCCTTTGCAGAACTCAGG‐3’
Trp53 mouse	Forward: 5’‐CCTCTCCCCCGCAAAAGAAA‐3’
	Reverse: 5’‐GACTCCTCTGTAGCATGGGC‐3’
Trp53 rat	Forward: 5’‐ GTGATATGTACCAGCCACAGG‐3’
	Reverse: 5’‐ CGAACAGACGACGGCATACT‐3’

Glucose tolerance test (GTT)

Mice were overnight‐fasted, and glucose levels were determined at 0, 15, 30, 45, 60, 90, and 120 min after the administration of a 20% w/v d‐glucose solution (2 g/kg body weight) by intraperitoneal injection. Whole blood was obtained from the tail vein and glucose levels assessed by the use of a portable glucometer One Touch^® Verio (Lifescan, MI, Italy).

Insulin tolerance test (ITT)

Mice were 4 h‐fasted, and then insulin (Humulin R, Eli Lilly Italia S.p.A., Italy) was administrated by an intraperitoneal injection (0.75 U/kg body weight). Whole blood samples were obtained from mice tail at 0, 15, 30, 45, 60, 90, and 120 min following insulin injection and were used for glucose levels measurement by the use of a portable glucometer One Touch^® Ultra (Lifescan MI, Italy). Mice whose blood glucose levels dropped below 45 mg/dl were excluded from the analysis.

Pancreatic islets isolation

Islets of Langerhans were isolated by collagenase digestion (Li et al, ²⁰⁰⁹). Mice were killed by cervical dislocation, and an incision around the upper abdomen was made to expose the liver and the intestine. Once located the ampulla, it was clamped on the duodenum wall and the pancreas was perfused by injecting 3 ml of Collagenase P (Roche Applied Sciences, Penzberg, Germany) solution (1,000 U ml⁻¹) in Krebs (NaCl 120 mM, KCl 4.8 mM, CaCl₂ 2.5 mM, MgCl₂·6H₂O 1.2 mM, NaHCO₃ 24 mM, Hepes 5 mM, penicillin/streptomycin 50 U/ml, pH 7.35). Digestion was completed in a shaking water bath (37°C) for 8–10 min. The digested pancreas was treated with DNase I (QIAGEN, Hilden, Germany). The islets were hand‐picked under a stereomicroscope using a syringe with a 25‐gauge needle and then pelleted for RNA extraction, homogenization, or incubated overnight in RPMI for ex vivo glucose‐stimulated insulin secretion (GSIS). Independent islets samples were isolated and processed for each different use. One mouse each group was in parallel processed to minimize potential confounders.

Glucose‐stimulated insulin secretion (GSIS)

Overnight‐fasted mice were injected with a 30% w/v D‐glucose solution (3 g/kg body weight), and whole blood glucose was collected from mandibular vein at 0, 3, and 30 min after glucose injection. Blood samples were allowed to clot for 30 min and then centrifuged at 4,500 rpm for 5 min at room temperature for serum collection. One mouse each group a time was in parallel analyzed to minimize potential confounders. Ex vivo GSIS was performed as previously described (Lombardi et al, ²⁰¹²). Briefly, batches of 20 size‐matched islets were preincubated at 37°C for 30 min in glucose‐free Krebs–Ringer bicarbonate HEPES buffer (KRBH) and then incubated for 1 h in a thermomixer at 37°C; incubation was with 1 ml KRBH medium containing 2.8 mmol/l glucose or 16.7 mmol/l glucose. Islets were then pelleted, supernatant fractions were collected to measure insulin secretion, and the genomic DNA was extracted from pellets to normalize insulin secretion. DNA extraction was performed using Wizard Genomic DNA purification kit (Promega, Milan, Italy), according to the manufacturer’s protocol. In vitro GSIS was analyzed in INS‐1 832/13 as previously described (Fiory et al, ²⁰¹¹). Briefly, at the end of treatments, cells were maintained for 1 h in KRBH with 0.25% (w/v) BSA and then incubated in KRBH with 2.8 mM (basal condition) or 16.7 mM glucose (stimulated condition). Incubation was stopped by putting the plates on ice, the supernatants were collected for insulin measurement, and cells solubilized for protein extraction as previously reported (Mirra et al, ²⁰¹⁷) for insulin normalization. Insulin levels in serum, islets, or cell supernatants were measured by the use of an ELISA assay (EMD Millipore Corporation, Billerica, MA, USA), according to the manufacturer’s protocol.

Histological analysis

Sections of formalin‐fixed, paraffin‐embedded (FFPE) pancreas were cut into 4 μm sections, deparaffinized, rehydrated through graded alcohols and stained with hematoxylin and eosin for the evaluation of islets morphology. Seven mice each phenotype were analyzed, and 3 sections per mouse were evaluated to determine number, size, morphology, and infiltration of islets, by the use of Olympus DP21 software (Olympus, Center Valley, PA, USA). Immunohistochemical staining was carried out on whole sections from FFPE tissues to evaluate the expression of insulin, glucagon, and F4/80 markers. For insulin and glucagon, we performed a manual procedure. Primary antibodies were Rabbit anti‐glucagon (clone EP3070‐ 1:6,500) and Rabbit anti‐insulin (clone EPR17359‐ 1:8,000) from Abcam (Cambridge, UK). Secondary reagent was Dako EnVision⁺ System‐ HRP Labelled Polymer Anti‐Rabbit (Dako; Agilent Technologies, Santa Clara, CA, USA). Peroxidase reactivity was visualized using 3,3’‐diaminobenzidine (DAB). F4/80 immunohistochemical staining was carried out using the BOND III slide stainer (Leica Biosystems, Wetzlar, Germany) in accordance with the manufacturer’s instructions. Staining was achieved using NBP2‐12506 F4/80 antibody (Clone SP115, Novus Biologicals, Centennial, CO, USA) and visualized using the Bond™ Polymer Refine Detection kit (Leica Biosystems, Wetzlar, Germany) using 3,3′‐diaminobenzidine tetrahydrochloride chromogen as substrate.

Using a semi‐quantitative scoring system microscopically, an observer (P.I.) performed a blind evaluation of the intensity, extent, and subcellular distribution of the markers by Olympus BX43 light microscope and Olympus DP21 microscope camera and software (Olympus, Center Valley, PA, USA). For each sample, ten fields and at least > 500 cells were analyzed. There are no standardized criteria for insulin, glucagon, and F4‐80 markers staining evaluation. For insulin and glucagon, the staining intensity was graded as follows: 0, negative; 1, weak; 2, moderate; and 3, intense. The intensity score was multiplied with percentage of positive cells in order to generate an immunoreactive score (IS). For insulin, high expression is defined as an IS ≥ 150, for glucagon ≥ 93, which represent the average value of the IS.

F4/80 stain showed for all positive samples the same staining intensity. For this reason, we evaluated the percentage of immunostained cells and we scored it as follows: high when positivity was ≥ 1%.

Multiplex immunoassay

Whole pancreata or isolated islets from wild‐type (WT) and Glo1KD mice were treated with Bio‐Plex Cell Lysis kit (Bio‐Rad, Hercules, CA, Stati Uniti) and homogenized by TissueLyser LT (QIAGEN, Hilden, Germany) accordingly to the manufacturer’s instructions. At the end of treatments, INS‐1 832/13 were cultured in serum‐free RPMI medium with 0.25% (w/v) BSA for 16 h. Mouse pancreatic lysates and cell‐conditioned medium were used to measure, respectively, the tissue and secreted concentration of interleukin‐1 beta (IL‐1β), tumor necrosis factor‐alpha (TNF‐α) and monocyte chemoattractant protein‐1 (MCP‐1) by the use of a custom‐blended ELISA multiplex assay (Bio‐Plex ProTM, Bio‐Rad) according to manufacturer's protocol. Islets lysates and serum samples were used to measure insulin and glucagon content by the use of an immunoassay in multiplex (Bio‐Plex Pro mouse diabetes immunoassay, Bio‐Rad) according to manufacturer's protocol. Data were acquired using a Bio‐Plex 200 system equipped with Bio‐Plex Manager software v5.0 (Bio‐Rad). The standard curves optimization and the calculation of analyte concentrations were performed by using the Bio‐Plex Manager software.

Western blot analysis

For whole cell proteins extraction, cells were solubilized in lysis buffer (50 mmol/l HEPES, pH 7.5, 150 mmol/l NaCl, 10 mmol/l EDTA, 10 mmol/l Na₂P₂O₇, 2 mmol/l Na₃VO₄, 100 mmol/l NaF, 10% glycerol, 1% Triton X‐100) for 2 h at 4°C. Cell lysates were clarified by centrifugation at 16,000 g for 20 min. Nuclear proteins were isolated using the NE‐PER Nuclear and Cytoplasmic Extraction Kit purchased from Thermo Fisher (Waltham, Massachusetts, USA), according to the manufacturer's instructions. Protein samples were separated by SDS–PAGE and transferred into 0.45‐μm Immobilon‐P PVDF membranes (Merck, Darmstadt, Germany). Upon incubation with primary and secondary antibodies, immunoreactive bands were detected by electrochemiluminescence according to the manufacturer's instructions (Pierce, Thermo Scientific, Rockford, IL, USA) and densitometric analysis was performed using ImageJ software. The antibodies used were anti‐p53 (cod. 2524, Cell Signaling Technology, Danvers, MA, USA), anti‐p21 (cod. sc397), anti‐HMGB1 (cod. Sc 56698, Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti‐histone H3 (cod. ab 1791, Abcam, Cambridge, UK), and anti‐vinculin (cod. sc73614, Santa Cruz Biotechnology, Santa Cruz, CA, USA).

Detection of Senescence‐associated beta‐galactosidase (SA‐βgal) activity

SA‐βgal activity has been evaluated in vitro by a fluorescence‐based assay using flow cytometry, as described in (Debacq‐Chainiaux et al, ²⁰⁰⁹). Briefly, at the end of treatments, INS‐1 832/13 were exposed to 100 nM bafilomycin A1 for 1 h in fresh cell culture medium to induce lysosomal alkalinization. Subsequently, cells were incubated in 33 µM C₁₂FDG (5‐dodecanoylaminofluorescein di‐β‐D‐galactopyranoside) for 2 h and then washed twice with PBS, harvested by trypsinization, and resuspended in ice‐cold PBS. Cell suspension was run in a BD FACS and the acquisition and analysis performed by BD FACSDiva Software (BD Bioscience). Once selected the analysis region which excludes dead cells and cellular debris, the percentage of positive cells were estimated by dividing the number of events within the bright fluorescence compartment (set at fluorescence intensity > 10⁴) by the total number of events.

Determination of mitochondrial DNA (mtDNA) content

Quantification of the mtDNA content was obtained using real‐time PCR amplification on a Quant Studio 7 Flex instrument (Applied Biosystem), according to the previously published protocol (https://doi.org/10.1007/978‐1‐4939‐1875‐1_3). The mtDNA amount was carried out by qPCR amplification of a mtDNA fragment tRNA‐Leu (NC_012374.1) and comparison with an endogenous reference. The nuclear‐encoded Gapdh gene (Gene ID: 24383) was used as a nuclear reference (nucDNA). Each sample was amplified in duplicate. The mitochondrial DNA content, relative to nuclear DNA, was calculated using the following equations: ΔCT = (nucDNA CT – mtDNA CT); Relative mitochondrial DNA content = 2 × 2ΔCT.

Statistic procedures

Data are expressed as means ± SEM. Data distribution was tested for normality by Shapiro–Wilk test. Comparison between groups was performed using Student’s t‐test (unpaired) or Mann–Whitney U test for comparison of non‐parametric data. Fisher’s exact test was performed for categorical variables. A P‐value of < 0.05 was considered statistically significant. Statistical analysis was performed with GraphPad Prism (version 7.0; San Diego, CA, USA) or R studio software (version 3.5.3; R studio, Boston, MA, USA). Outlier values were calculated by performing Grubb’s test and excluded from the analysis. During the performance of the experiments, the authors were not aware of the group allocation.

Author contributions

Immacolata Prevenzano: Formal analysis; Investigation; Writing—original draft. Alessia Leone: Formal analysis; Investigation; Writing—original draft. Michele Longo: Data curation; Formal analysis; Investigation; Writing—original draft. Antonella Nicolò: Investigation; Methodology. Serena Cabaro: Investigation. Francesca Collina: Formal analysis; Investigation. Iacopo Panarese: Formal analysis; Investigation. Gerardo Botti: Formal analysis. Pietro Formisano: Writing—review and editing. Raffaele Napoli: Writing—review and editing. Francesco Beguinot: Funding acquisition; Writing—review and editing. Claudia Miele: Data curation; Project administration; Writing—review and editing. Cecilia Nigro: Conceptualization; Data curation; Supervision; Writing—original draft; Project administration.

In addition to the CRediT author contributions listed above, the contributions in detail are:

IPr, AL, and ML performed the experiments, analyzed the data, and wrote the manuscript. AN performed the in vitro experiments, the sample collection and prepared figures. SC performed the multiplex immunoassay. FC, IPa, and GB performed the histological examination, acquired microscopy images, and analyzed the data. CM, FB, PF, and RN provided intellectual input on the project and revised the manuscript. CN conceived the study, supervised the overall project, and revised the manuscript.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Source Data for Figure 4

EMBO reports (2022) 23: e52990.

Data availability

No data were deposited in a public database.