Acceleration of β-cell aging determines diabetes and senolysis improves disease outcomes (original) (raw)

• Journal List
• HHS Author Manuscripts
• PMC6610720

Cell Metab. Author manuscript; available in PMC 2020 Jul 2.

Published in final edited form as:

PMCID: PMC6610720

NIHMSID: NIHMS1528619

Cristina Aguayo-Mazzucato,1 Joshua Andle,1 Terrence B Lee, Jr,1 Ayush Midha,1 Lindsay Talemal,1 Vaja Chipashvili,1 Jennifer Hollister-Lock,1 Jan van Deursen,2 Gordon Weir,1 and Susan Bonner-Weir1

Cristina Aguayo-Mazzucato

1Joslin Diabetes Center, Harvard Medical School, Boston;

Joshua Andle

1Joslin Diabetes Center, Harvard Medical School, Boston;

Terrence B Lee, Jr

1Joslin Diabetes Center, Harvard Medical School, Boston;

Ayush Midha

1Joslin Diabetes Center, Harvard Medical School, Boston;

Lindsay Talemal

1Joslin Diabetes Center, Harvard Medical School, Boston;

Vaja Chipashvili

1Joslin Diabetes Center, Harvard Medical School, Boston;

Jennifer Hollister-Lock

1Joslin Diabetes Center, Harvard Medical School, Boston;

Jan van Deursen

2Department of Biochemistry and Molecular Biology, Mayo Clinic, Rochester, MN

Gordon Weir

1Joslin Diabetes Center, Harvard Medical School, Boston;

Susan Bonner-Weir

1Joslin Diabetes Center, Harvard Medical School, Boston;

1Joslin Diabetes Center, Harvard Medical School, Boston;

2Department of Biochemistry and Molecular Biology, Mayo Clinic, Rochester, MN

AUTHORS’ CONTRIBUTIONS

CAM and SBW conceived the project and wrote the manuscript; CAM, JA, TJL, AM, LT, VC, JHL researched data; GCW provided critical discussions and edited the manuscript. JVD provided critical discussion, animal model and important reagents. All authors reviewed the manuscript.

Supplementary Materials

Supplemental Table 3.

GUID: 4BA91BCE-CF86-4436-88D5-B2E73F78C89E

Supplemental figures and Tables.

GUID: 5918F4F6-A77C-4E97-9C10-A31BD677A97E

Data Availability Statement

The accession number for the genomic data reported in this paper is GSE121539.

SUMMARY

Type 2 diabetes (T2D) is an age-related disease. Although changes in function and proliferation of aged β-cells resemble those preceding the development of diabetes, the contribution of β-cell aging and senescence remains unclear. We generated a β-cell senescence signature and found that insulin resistance accelerates β-cell senescence leading to loss of function, cellular identity and worsening metabolic profile. Senolysis (removal of senescent cells), using either a transgenic INK-ATTAC model or oral ABT263, improved glucose metabolism and β-cell function while decreasing expression of markers of aging, senescence and senescence-associated secretory profile (SASP). Beneficial effects of senolysis were observed in an aging model as well as with insulin resistance induced both pharmacologically (S961) and physiologically (high fat diet). Human senescent β-cells also responded to senolysis, establishing the foundation for translation. These novel findings lay the framework to pursue senolysis of β-cells as a preventive and alleviating strategy for T2D.

Keywords: Beta-cells, senescence, senolytic therapies, type 2 diabetes, insulin secretion, glucose metabolism, senescence signature

eTOC Blurb

Aguayo-Mazzucato et al identify the signature of senescent pancreatic β-cells and show that the population of senescent β-cells is increased by insulin resistance but is partially reversible. Removing senescent cells improves insulin secretion, genetic identity and glucose homeostasis. These findings provide insight into how β-cell senescence contributes to type 2 diabetes, opening new therapeutic targets.

Graphical Abstract

INTRODUCTION

Type 2 diabetes (T2D) is an age-related disease characterized by a decrease of β-cell mass and function representing a failure to compensate for the high insulin demand of insulin resistant states (Leahy, 2005; Poitout and Robertson, 2002; Porte, 2001; Prentki et al., 2002; Weir and Bonner-Weir, 2004, 2013). Yet, the role of aging as it pertains to pancreatic β-cells is poorly understood and therapies that target the aging aspect of the disease are virtually non-existent. For many years β-cells can compensate for increased metabolic demands with increased insulin secretion, keeping hyperglycemia at bay. This compensation may be limited by the age-related decline in β-cell proliferation seen in rodents (Fan et al., 2011; Rankin and Kushner, 2009; Scaglia et al., 1997; Stolovich-Rain et al., 2012; Teta et al., 2005) and humans (Gregg et al., 2012). This deficiency in proliferative response to increased demand may arise partly from the accumulation of senescent β-cells.

Cellular senescence is a state in which cells cease to divide but remain metabolically active with an altered phenotype (Hayflick, 1965; Tchkonia et al., 2013). There are no universal markers of senescence and the markers that exist are not consistent in every senescent tissue (Campisi and d’Adda di Fagagna, 2007). p16Ink4a, a cyclin-dependent kinase inhibitor encoded by the Cdkn2a locus, has been identified as both marker and effector of β-cell senescence (Krishnamurthy et al., 2006; Krishnamurthy et al., 2004). An increase in p21Cis1, another effector of cellular senescence, is thought to mark the entry into early senescence leading to increased p16Ink4a expression, which then maintains senescence resulting in the expression of the senescence-associated secretory profile (SASP) (Rodier and Campisi, 2011; Stein et al., 1999). SASP profiles differ with tissue type and can include soluble and insoluble factors (chemokines, cytokines, ECM) that affect surrounding cells and contribute to multiple pathologies (Coppe et al., 2010).

With age, accumulation of dysfunctional senescent β-cells likely contributes to impaired glucose tolerance and diabetes. Yet the specific contribution of β-cell aging and senescence to diabetes has received little attention and the specific SASP profile of β-cells remains to be determined.

We have previously found (Aguayo-Mazzucato et al., 2017) that even in young (3–4 m old) mice, a population of β-cells express known aging markers (senescence-associated acidic β-galactosidase activity (β-gal), p16 Ink4a and p53BP1) and that this population increased with age. Aged β-cells had impaired function, characterized by higher basal insulin secretion and a lower recruitment to glucose challenges. Moreover, acute insulin resistance, induced by the insulin receptor antagonist S961, resulted in expression of aging markers p16 Ink4 and Bambi, suggesting that insulin resistance was a driver of accelerated β-cell aging. In the present study, we address the relationship between β-cell aging, the development of diabetes and whether strategies aimed at decreasing the load of aged β-cells can improve cellular identity, function and overall metabolic parameters.

To this end we developed a β-cell senescence signature, which characterizes senescent β-cells that actively secrete SASP factors. As hypothesized, metabolic stressors, such as S961-induced insulin resistance and high fat diet (HFD), accelerated the appearance of aging and senescence markers in β-cells and led to their loss of function and impaired glucose tolerance. Clearance of p16Ink4a+ cells, using the INKATTAC mouse, ameliorated glucose metabolism, improved insulin secretion and decreased expression of aging, senescence and SASP genes in islets from models of aging and insulin resistance. Additionally, an oral senolytic compound, ABT263, ameliorated hyperglycemia and improved the β-cell gene expression profile in animals challenged with insulin receptor antagonist S961. Human β-cells share the same biology: the load of senescent cells increased with age and diabetes and they overexpress p16 Ink4a. Our work provides the biological and cellular framework to pursue senolysis of β-cells as a potential therapy for inhibiting the progression of T2D.

RESULTS

Senescence and altered intercellular communication are main components of β-cell aging.

Pathway analysis of our previously published microarray data (Aguayo-Mazzucato et al., 2017) comparing β-cells from old and young mice, revealed that of the 9 described pathways of aging (Lopez-Otin et al., 2013), the old β-cells were enriched for genes related to cellular senescence (positive regulation of cell cycle, negative regulation of cell cycle, cell cycle arrest, regulation of mitotic cell cycle) and altered intercellular communication or SASP (cytokine and chemokine mediated signaling pathway)(Suppl Table 1). Therefore, our focus to understand β-cell aging was mostly centered on senescence and its SASP aspect.

A β-cell senescence signature revealed downregulation of hallmark β-cell genes and expression of “disallowed” genes.

Pancreatic islets isolated from C57Bl6/J male retired breeders (7–8 m) were dispersed into single cells and FACS sorted (Suppl. Fig. 1) based on β-galactosidase (β-gal) activity as previously described (Aguayo-Mazzucato et al., 2017)(Fig. 1A, Suppl Fig. 1D) and gated for an enriched β-cell subpopulation (Fig. 1B). β-cell enrichment was confirmed using zinc selective indicator FluoZin-3AM that specifically labels β-cells (Suppl Fig. 1E). To evaluate the potential presence of immune cells in our FACS sorted populations, a sort for CD45+ revealed that about 0.5% of our population were immune cells mainly represented by resident macrophages F4/80+CD11b+ (Suppl Fig. 1F,G) and were distributed within β-gal+ and β-gal- fractions (Suppl Fig. 1H,I). When quantified, our FACS sorted population was composed of a 90% of pure β-cells and 0.5% of resident macrophages (Fig. 1C). As is characteristic for senescent cells, β-gal+ cells were significantly larger than β-gal- cells (Fig. 1D), and their diameter did not overlap with that of resident macrophages (>20 um). RNASeq analysis showed that 3732 genes out of 15,500 were differentially regulated between β-gal+ and β-gal- β-cell samples from the same animals. Importantly, we observed a downregulation of key hallmark β-cell identity genes (GSEA)(Fig. 1E) including Insulin 1, Mafa, Nkx6.1 and Pdx1. Simultaneously, there was an upregulation of genes whose expression are usually repressed in β-cells “disallowed genes” (Pullen et al., 2010; Thorrez et al., 2011), such as Ldha and catalase (Fig. 1F). Specifically curated molecular signature databases (GSEA) for aging (Fig. 1G), and senescence (Fig. 1H) genes showed an upregulation of these genes in β-gal+ as compared to β-gal- cells. Using these data we generated indices for assessment of β-cell identity, aging and senescence (see Methodsfor specific genes).

Generation of a β-cell senescence signature and cellular characteristics.

Islets isolated from 7–8 m old C57Bl/6J male retired breeders were FACS sorted into non-senescent (β-gal negative) and senescent (β-gal positive) subpopulations(A) for RNASeq after gating for enrichment of β-cells (B). Cellular identity of FACS sorted cells was 90% β-cells and 0.5% of immune cells (C). D. As described for senescent cells, β-gal+ β-cell were significantly larger than β-gal-cells. At least 100 cells counted/condition from 3 different fields. For RNA seq data there were 7 sets of paired samples, each set from islets pooled of 30 mice. Senescent β-cells were characterized by a downregulation of hallmark β-cell identity genes (E) and upregulation of “disallowed” genes (F). Senescent β-cells also showed upregulation of specific aging (G) and senescent genes (H).

Senescent β-cells actively produce and secrete SASP factors.

Analysis revealed an upregulation of SASP genes in the β-gal+ subpopulation (Fig. 2A) and, as part of their SASP profile, primary β-cells secreted more IL6, TNF and CXCL1 than non-senescent cells (Fig. 2B). Conditioned media (CM) generated by collecting media from cultured sorted β-gal+ and β-gal- β-cells populations was used to culture dispersed isolated islets. Cells exposed to CM from β-gal+ cells increased expression of p16Ink4a with no change in p21 Cis1 (Fig. 2C) compared with those cultured with CM from β-gal- cells, suggesting that SASP from senescent primary β-cells was functional. Given that β-cells are not an inflammatory cell type and there was potential participation of resident macrophages (<1%), we tested the effect of CM from senescence-induced MIN6 cells, a mouse β-cell immortalized cell line. Senescence was induced by treating these cells with 200 nM doxorubicin or 450uM H2O2 for 24h and then collecting conditioned media (CM) after 24–48h to measure SASP protein secretion by a β-cell derived line without contamination by other cell types. Induction of senescence was confirmed by increased expression of p16 Ink4a and p21 Cis1 mRNA (Fig. 2D,​F); SASP factor proteins were measured in CM showing greater secretion from senescent cells (Fig. 2E,​G). Viability of MIN6 cells was not affected by doxorubicin but was greatly diminished by H2O2 (Suppl. Fig. 2).

SASP factors are produced and secreted by senescent β-cells.

Senescent β-cells were characterized by an upregulation of SASP factors (A), some of which are significantly increased in conditioned media from β-gal+cells (B). Each point represents a FACS sorting experiment. Mean concentrations of proteins were IL1a 6pg/ml; TNFa 23 pg/ml, IL6 83 pg/ml, CCL5 5 pg/ml. CSCL1 22 pg/ml, CCL3 6 pg/ml, CCL4 4 pg/ml, CXCL10 2 pg/ml. C. Isolated mouse islets were cultured 4d in the presence of CM from β-gal+β-cells had regulation of p16 _Ink4a_compared to those in CM form exposed to CM from β-gal-β-cells, indicating a functional β-cell SASP. CM from 5 FACS sorts with each used on 2–3 separate islet isolations. D. In immortalized β-cell line MIN6 cells senescence was induced by 24 hr exposure to 200 nM doxorubicin; senescence was confirmed by upregulation of p16 Ink4a and p21 Cis1 mRNA. E. Conditioned media from doxorubicin-treated cells had detectable protein levels of several SASP factors compared to non-senescent MIN6 cells. Each point represents an experiment; *p<0.05. F. MIN6 cell senescence was induced by exposing them to 450 μM H2O2 for 24h; senescence was confirmed by upregulation of p16 Ink4a and p21 Cis1 mRNA. G. Conditioned media from H2O2-treated cells cells had detectable protein levels of several SASP factors compared to non-senescent MIN6 cells. Each point represents an experiment; *p<0.05. H. MIN6 cells were treated with p16 Ink4a siRNA, decreasing its expression by 50%. Several SASP factor mRNAs were significantly changed (red bars) compared to cells treated with Scr siRNA, suggesting their regulation in β-cells is downstream of p16 Ink4a. n=4 experiments in triplicate. Means plotted with each point representing a single sample. p<0.05.

It is worth noting that the SASP profile differed between primary β-cells (Fig 2B) and MIN6 cells (Fig. 2E,​G). This suggests that β-cell senescence is a complex stepwise process (as recently suggested by (De Cecco et al., 2019)) and that more research regarding the timeline and molecular mechanisms behind β-cell SASP is needed. To evaluate the correlation between senescence and SASP in a pure β-cell model, we knocked down p16Ink4a in MIN6 cells using siRNA. A 50% decreased p16Ink4a expression resulted in significantly decreased Il6, Il1a, Igfbp5, Lamb1 and Lamc1 mRNA and increased Cxcl2, Cxcr4 and Ccl2 mRNA (Fig. 2H). These data suggest that these factors are downstream of p16Ink4a, although the mechanism of action behind this relationship remains to be determined.

Insulin resistance accelerated the appearance of senescent β-cells.

To study the effects of insulin resistance we used both the insulin receptor antagonist S961 and high fat diet (HFD). Both approaches increased the proportion of β-Gal+ cells as well as increasing the aging and SASP indices of gene expression (Fig. 3, Suppl Fig. 3A–C). First, an acute and severe model of insulin resistance was induced in mice using the insulin receptor antagonist S961 (osmotic minipump administration) that induced marked hyperglycemia and hyperinsulinemia (Fig. 3A,​B,​H,​I). By qPCR and immunostaining, p21Cis1 was significantly increased at both the mRNA and protein level (Fig. 3F,​G, Suppl Fig. 3). Blood glucose and insulin levels were completely reversed two weeks after removal of the minipumps (Fig. 3H,​I). Interestingly, normalization of hyperglycemia also reversed the aging and SASP indices (Fig. 3J,​K, Suppl Fig. 3D) suggesting that cellular senescence is at least partially reversible. After the recovery, β-cell specific genes (comprising the β-cell index) were also upregulated suggesting improved cellular health of β-cells.

Insulin resistance accelerates β-cell senescence.

Osmotic minipump administration of the insulin receptor antagonist S961 for 2 wk induced marked hyperglycemia (A) and hyperinsulinemia (B) in 7–m old male C57Bl6/J mice. Islets from treated mice had greater proportion of β-gal+β-cells (C) (12,089–100,000 events/data point) and increased expression of aging (D) and SASP-related (E) genes. A,B,D,E: 7m male C57Bl6/J; C 4m male INK ATTAC. n=5 per group. *p<0.05. F. Image of pancreas showing heterogenous colocalization of insulin and P21CIS1 in S961 treated animals compared to untreated controls. Magnification bar= 50 μm. G. Quantification of cellular P21CIS1 staining intensity presented as mean per β-cells of each islet; at least 30 islets counted per pancreas, n=3 control, 6 S961 animals. Two wk after minipump excision and normalization of blood glucose (H) and insulin (I) levels, some changes induced by S961 were reversed: both the aging (J) and SASP (K) indices decreased with respect to S961 islets, with no change on b cell-related genes (L) (See Suppl Fig 3 for individual values). 7m C57Bl6/J male; n=5 per group. Eight wk of high fat diet (HFD) starting at 8 wk increased body weight (M), fasting glucose (N) and induced glucose intolerance (O). Islets from treated animals showed higher proportion of β-gal+ cells (P)(20,856–40,933 events/data point) and increased aging (Q) and SASP (R) indices. n= 12 C57Bl6/J mice per group. *p<0.05 respect to control diet.

The second model of insulin resistance used HFD. After 8 wk HFD started at 8 wk of age, body weight and fasting blood glucose levels increased, glucose tolerance deteriorated as judged by IPGTT (Fig.3 M–O), and peripheral insulin resistance was modestly increased as determined by insulin tolerance test (ITT) (Suppl Fig. 3E); all consistent with a higher demand on the mass and function of pancreatic β-cells. Under these conditions, the percentage of β-gal+ cells in dispersed islets increased from 2% to 8% (Fig. 3P); additionally, there were significant increases in the aging and SASP indices (Fig. 3Q,​R, Suppl Fig. 3B), indicating increased β-cell senescence.

Deletion of p16Ink4a-expressing cells improved β-cell function and identity.

To test whether the deletion of senescent cells had beneficial effects, we tested the 3 models (aging, S961 treatment and HFD) in INK-ATTAC mice, a whole body FLAG-tagged transgenic that allows deletion of cells expressing p16Ink4a upon administration of B/B homodimerizer (Baker et al., 2016; Baker et al., 2011). First, we verified that senescent islet cells were deleted after administration of two 3-day courses of B/B homodimerizer (10mg/kg) with 14 d between courses (Xu et al., 2015) in 1–1.9 y old animals by measuring eGFP and Caspase 8 levels by qPCR (Fig. 4A). At the protein level, this deletion in pancreatic islets was confirmed by staining pancreas for the transgenic FLAG and insulin (Fig. 4B,​C). Quantification of FLAG staining (Fig. 4B) showed significantly increased proportion of FLAG negative islets and decreased proportion of those with medium/high FLAG staining after B/B treatment. Then, in aged (1.3–1.6 y old female) INK-ATTAC mice treatment with B/B homodimerizer improved β-cell aging and SASP indices (Fig. 4E,​F, Suppl Fig. 4A) without significant changes in glucose tolerance (Fig. 4D). However, β-cell function was improved as seen by in vivo GSIS (Fig. 4G), which showed decreased basal insulin levels after an overnight fast followed by a significant increase of insulin levels 15 min after a glucose load.

Specific deletion of p16Ink4a expressing cells improved metabolic profile, β-cell function and gene expression.

A. Evaluation of deletion protocol with B/B Homodimerizer (10mg/kg) revealed decreased eGFP and Casp8 mRNA in INKATTAC islets. Treatment was two courses of 3 d with 14 d between courses to activate the transgene caspase-8 moiety and lead to cell deletion of p16Ink4a expressing cells in the INKATTAC mice. Each dot represents the islets from an individual animal. B,C. Effects of B/B homodimerizer treatment on deletion of p16Ink4a β-cells were evaluated by quantification of FLAG and insulin co-staining. Pancreas from 5 m old animals (n=6 per group) were stained in parallel and confocal pictures were taken under the same settings such that differences in intensity reflect differences in protein concentration. Magnification bar= 50mm *p<0.05. Old INK-ATTAC mice partially improved glucose metabolism (D) and recovered β-cell function (G) after treatment with B/B homodimerizer. Isolated islets had decreased expression of genes of the aging (E) and SASP (F) indices. 1.3–1.6 y INK-ATTAC female mice. n=7–8 per group. *p<0.05. INKATTAC mice treated with insulin receptor antagonist S961 and B/B homodimerizer had improved metabolic profile (H) as shown by the AUC of their fed glucose levels (I). Improvement of aging (J) and SASP (K) indices was also observed. 9–14 m INK-ATTAC male S961 (20nM/wk) treatment for 2 wk. n=3–4 per group. S961 *p<0.05. (See Suppl Fig 4 for individual values).

With the second model, acute insulin resistance was induced by S961 over two wk in 9–14 m INK ATTAC male mice. Two 3-d courses of B/B homodimerizer treatment separated by 7 d, significantly decreased fed blood glucose levels (Fig. 4H,​I), islet indices for aging and SASP expression (Fig. 4J,​K, Suppl Fig 4B,C).

Finally, HFD administered to 9-m old INK ATTAC mice induced significantly increased body weight and fed glucose levels (Fig. 5A,​B) after 8 wk, which were blunted by B/B homodimerizer administration in courses of 3 d followed by 14 d in between. After 12 wk HFD, B/B-treated female INK ATTAC mice had improved glucose tolerance (Fig. 5C, ​D), improved β-cell function as reflected by in vivo GSIS (Fig. 5E) and improved aging and β-cell indices, yet with an increase in the SASP index (Fig. 5F–H, Suppl Fig. 5A, B), most likely due to the length of the metabolic stress and induction of irreversible late senescence. The improved genetic changes in β-cells induced by B/B senolysis likely explains their improved function.

Beneficial effects of p16Ink4a deletion in HFD INK-ATTAC model.

Changes induced with HFD such as increased body weight (A), fed hyperglycemia (B) and glucose intolerance (C,D) were blunted after B/B homodimerizer. In vivo insulin secretion parameters improved after B/B treatment (E) and although no beneficial changes were seen in aging (F) and SASP (G) indices, islets had enhanced β-cell index (H). (See Suppl Fig 5 for individual values). INK-ATTAC female mice 9m, n=6–7 animals/group at start of the study; at isolation 4–7 animals/group remained. *p<0.05 respect to control; +p<0.05 respect to HFD.

INK-ATTAC animals are whole body transgenics, and B/B treatment will have effects in all tissues in which p16Ink4a is expressed, raising the possibility that the improvements in glucose metabolism were due to improved insulin action in fat, liver and/or muscle. However, the ITTs (which evaluate peripheral insulin resistance) of the treated and untreated groups did not differ but tended to normalization (Suppl Fig. 5C). Additionally, in peripheral tissues important for glucose homeostasis (fat, liver, white and red muscle), B/B treatment resulted in no significant changes in p16 Ink4a and p21 Cis1 mRNA (Suppl Fig. 5D,E). These results underline the importance of β-cell senescence in glucose homeostasis and suggest that targeting this cell population is a strategy to consider in diabetes.

Senolytic drug, ABT263, improved glucose metabolism and β-cell identity.

Senescent cells have upregulated anti-apoptotic pathways that conserve their presence in otherwise healthy tissues (reviewed in (Kirkland et al., 2017). Their deleterious functional effects are then amplified by their secreted SASP that can lead to impairment of neighboring cells. At least 5 senescent cell anti-apoptotic pathways have been identified in different tissues. Senolytic therapies that specifically target these pathways are a promising approach to alleviate some of the conditions associated with an increased load of senescent cells. Based on pathway analysis of our RNASeq data, β-gal+ β-cells had upregulation of two of these pathways: the HIF1α pathway (FDR= < 0.0001) that can be targeted with quercetin and dasatinib, and anti-apoptotic members of the BCL2 pathway, such as A1/Bfl1 (FDR < 0.0001) or Mcl1 (FDR <0.001) that can be targeted with ABT263 (navitoclax). We tested the in vitro effects of ABT 263 on sorted β-gal+ and β-gal- β-cells and found that ABT263 killed a significant portion of β-gal+ subpopulation at a dose of 5uM after 4 days of treatment (Fig. 6A). Based on these results, we selected ABT263 for in vivo treatment of S961 treated (6–9 m old) INKATTAC male mice. S961-induced insulin resistant mice that were simultaneously treated with ABT263 had only a 3-fold increase in blood glucose levels as compared to a 5-fold increase of S961 only treated mice (Fig. 6B,​C). The load of senescent β-cells decreased by 25% as revealed by β-gal+ FACS sorted population analysis (Fig. 6D). When analyzing gene expression, the aging index of islets from ABT263-treated mice did not change compared to the untreated insulin resistant mice, although their p16 Ink4a levels decreased (Suppl Fig. 6B,E). ABT263 treatment decreased the SASP index of islets but had little effect on aging or β-cell indices (Fig. 6E–G, Suppl Fig. 6A, B). The main caveat of senolytic therapies is that they can act broadly across all cells and tissues. To evaluate which of the main tissues that contribute to glucose metabolism were targeted by the ABT263 treatment, we measured p16 Ink4_a and p21 Cis1 transcripts in islets, liver, white fat, red and white muscle of treated versus untreated animals. p16 Ink4a expression decreased only in islets and liver (Suppl Fig. 6E), while p21 Cis1 was unchanged in all tissues (Suppl Fig. 6F). ABT263 administration was also tested in a HFD model. 5–6 m old INK-ATTAC female mice were placed on a HFD for 12 weeks and oral ABT administered during 5 d courses every 3 wk. Although there were no effects on fed blood glucose (Fig. 6H), the percentage of β-gal+ cells, aging and SASP indices decreased with respect to the HFD group with no change in the β-cell index (Fig. 6I–K; Suppl Fig. 6C,D). Their peripheral tissues had significantly decreased p16 Ink4a (Suppl Fig. 6G) with only red muscle having significant decreased p21 Cis1 (Suppl Fig. 6H). We also tested i_n vitro the effects of quercetin and quercetin+dasatinib on sorted β-gal+ and β-gal-β-cells. While quercetin alone did not have effects on senescent β-cell mortality (Suppl Fig. 7A), the combination of quercetin and dasatinib decreased the senescent cell number by 40% (Suppl Fig. 7B). In vivo, acute insulin resistance was induced in mice using S961 and combined quercetin (50mg/kg) and dasatinib (5mg/kg) was given orally by gavage once per week. By the end of the two weeks, blood glucose levels decreased in the treated group (Suppl Fig. 7C,D), underlying the physiological value of senolytic therapies in the treatment of diabetes. There were few if any CD45+ immune cells in the islets; this percentage did not change from the 0.5% control values with S961 or senolytic treatment (Suppl Fig. 7E). These results show that oral administration of a senolytic agent during either acute or chronic insulin resistance was able to partially reverse the adverse metabolic effects as well as provide some restoration of β-cell identity.

Senolytic therapies selectively killed senescent cells and improved glucose metabolism, SASP and β-cell indices.

A. ABT263 specifically killed β-gal+ FACS sorted islet cells in vitro. Islets from male retired breeders C57Bl/6CRL. n= 2–3 experiments with up to 4 replicates/condition; 4 days. When ABT263 was administered in vivo by daily gavage to INK:ATTAC mice treated with S961, circulating blood glucose levels significantly improved (B,C), the percentage of β-gal+cells decreased (D)(31,684–64,940 events per data point) with no change in aging index, SASP and β-cell indices (E, F, G) (See Suppl Fig 6 for individual values). INKATTAC male mice 6–9 m old. n=3–5 animals/group. Administration of ABT263 during 12 wk HFD had no effects on fed glucose levels (H) but decreased the percentage of β-gal+cells (I) as well as the aging and SASP indices (J,K) and unchanged β-cell index (L) compared to those animals receiving HFD without ABT263 treatment (See Suppl Fig 6 for individual values). INK ATTAC 6–9 m old female mice, n= 3–4 each group.

Translation into humans.

To further our understanding about how these findings translate into human β-cell biology, aging and type 2 diabetes (T2D), we analyzed islets isolated from donors of different ages with and without T2D. As with rodents, the percentage of β-gal+ islet cells increased in islets isolated from older donors compared to younger ones (Fig. 7A), and this proportion seems to be increased further in islets from T2D donors, suggesting that there is a component of β-cell senescence in this disease. Furthermore, we confirmed increased expression of P16 INK4A (Fig. 7B) and SASP factors CCL4 and IL6 mRNA (Fig 7C) in the β-gal+ human subpopulation. When human islets were sorted into β-gal+ and β-gal-cells and treated with ABT263, the β-Gal+ subpopulation had a significantly higher cell mortality as reflected by propidium iodide incorporation (Fig. 7D). To further characterize the correlation between β-cell aging, senescence and diabetes, pancreatic sections from different aged donors with and without diabetes were stained for β-cell aging marker IGF1R and for DNA damage/senescence marker nuclear P53BP1. In donors younger than 40 years of age, the presence of T2D was associated with a higher intensity of IGF1R in β-cells (Fig. 7E, ​F) suggesting an early aging phenomena associated with diabetes. For P53BP1 (Fig. 7G), in non-diabetic donors, there was a direct correlation between nuclear P53BP1 and BMI (Fig. 7H), suggesting that states of higher insulin resistance, such as associated with obesity, correlated with higher expression of P53BP1. Finally, when donors with high BMI (>33) were excluded, an increase in P653BP1 intensity was observed in islets within pancreas from patients with T2D compared with non-diabetics (Fig. 7I). These results are consistent with our animal models and suggest that human β-cells are a potential target for the senolytic drug therapies in the context of insulin resistance and early diagnosed diabetes.

Increased load of senescent human β-cells with donor age associated with enrichment of p16 INK4A.

A. The percentage of β-gal+cells in human islets increased as the age of the donor increased. Samples from T2D were enriched for senescent cells compared to age-matched non-diabetic donors. B. β-gal+β-cell subpopulation expressed higher levels of P16 INK4A mRNA than β-gal- population of the same donor. C. mRNA expression levels of SASP factors from human β-gal+β-cells compared to β-gal-cells. D. In vitro treatment with ABT263 (5 μM) revealed higher cell mortality of β-gal+ human β-cells when compared to β-gal-cells from the same donor. E. Representative pictures of IGF1R stained islets in sections of human pancreas. Sections were stained in parallel and pictures taken in the confocal microscope under the same setting such that differences in intensity reflect differences in protein concentration. F. Quantification of IGF1R intensity in islets from donors of different ages, with and without T2D. G. Representative pictures of nuclear P53BP1 of human pancreas. H. Linear correlation between P53BP1 intensity and BMI in donors without diabetes older than 39 y. I. Increased P53BP1 intensity in islets from donors older than 39 y, excluding those with BMI>33. Each data point is separate donor. Magnification bars = 50μm.

DISCUSSION

It is widely accepted that T2D has an important aging component, however no therapies to date have been directed at this aspect. Our work demonstrates markers of β-cell aging and senescence are increased by metabolic stressors of insulin resistance, suggesting that targeting the senescent β-cell population may have therapeutic benefit.

By generating a novel β-cell senescence signature, we showed that as they senesce, β-cells downregulate expression of key genes vital to their function and identity, such as i_nsulin, Mafa, Pdx1, Neurod1_. At the same time, the “disallowed”or usually suppressed genes, such as Ldha, become expressed as are genes directly related to both aging and senescence like p21 Cis1 and Igf1r. This loss of identity in senescent β-cells resembles the changes found with glucose toxicity (Jonas et al., 1999).

We show how β-cell senescence is a dynamic process that can be accelerated by insulin resistance and be partially reversed. As shown by our S961 experiment, once the induction of insulin resistance is removed and islets allowed to recover for 2 weeks, many of the gene expression changes associated with early senescence reverted towards normal. Given that senescence is a multistep process characterized by the initial upregulation of p21 Cis1 followed by that of p16 Ink4a (De Cecco et al., 2019; Stein et al., 1999) and then SASP factors, our results suggest that some of these initial steps (eg. p21 Cis1 increase), might be reversible. However, if the stressor is continued, the senescence program might become irreversible and late senescence SASP factors turned on as suggested by (De Cecco et al., 2019). Interestingly, some of our models (mainly the acute insulin resistance with S961) showed greater upregulation of p21 Cis1 than of p16 Ink4a, suggesting that we might still be within a critical window within which senescence could be reversed if the metabolic stressor were removed. Yet, many cases of insulin resistance and type 2 diabetes are longstanding, therefore interventions aimed at decreasing the load of senescent β-cells could be beneficial at various time points as shown by our aged INK ATTAC model. When these mice were treated with B/B homodimerizer, their glucose metabolism improved, as did their β-cell function and gene profile. Understanding from an aging point of view the changes in gene expression of β-cells with insulin resistance and detecting a window of reversibility open up an exciting new therapeutic opportunity to inhibit progression of diabetes.

T2D typically develops in response to over-nutrition and lack of physical activity in subjects with underlying predisposition to insulin resistance and β-cell dysfunction. Over time, β-cell compensation for the insulin resistance fails, resulting in β-cell dysfunction (Leahy, 2005; Porte, 2001; Prentki et al., 2002; Weir and Bonner-Weir, 2004, 2013). As demonstrated in rodents, the mechanisms of compensation for increasing insulin resistance include increased secretion (Weir and Bonner-Weir, 2013) and β-cell mass (Bruning et al., 1997). The chronological age of the animals seems to limit the proliferative capacity of β-cells (Kushner, 2013) most likely through cellular senescence (Krishnamurthy et al., 2006; Krishnamurthy et al., 2004).

Although cellular senescence has been implicated in the pathophysiology of T2D (Cordisco et al., 2017; Goldstein et al., 1969; Goldstein et al., 1978, 1979), many of these studies focused on the replicative capacity of fibroblasts or senescence markers in endothelial cells. Both may contribute to diabetes complications, however they are not central to the development of hyperglycemia.

Our results suggest that an acceleration of β-cell senescence, representing cellular aging, contributes to the progression of declining glucose metabolism induced by insulin resistance. This was characterized by a decline in islet function as well as an increased expression of senescence, aging genes and SASP genes. These findings contrast to a study in which 10 d induction of p16Ink4a under the Ins or Pdx1 promoters in β-cells of 3–4-wk old transgenic mice resulted in both markers of senescence and improved β-cell function (Helman et al., 2016), leading to their conclusion of a novel functional benefit of senescence in β-cells. However, β-cells of 5–6-wk old mice are not yet fully mature; islets from 3-wk old rats are glucose-responsive but without the robustness found in 2–3-m old rats (Bliss and Sharp, 1992). Moreover, the fact that p16Ink4a was driven by the insulin or Pdx1 promoter means that the resulting p16Ink4a levels were supraphysiological, which does not necessarily translate into a model of aging or senescence.

Although the population of senescent β-cells (β-gal+) was only 8–10% in our models, SASP secretion, part of the senescent phenotype, can contribute to multiple pathologies associated with diabetes (Coppe et al., 2010; Thompson et al., 2019). Here, we have characterized a specific SASP profile of β-cells and show that some of these factors are detectable in higher concentrations in the conditioned media (CM) obtained from cultured β-gal+ cells than from β-gal- ones. We also show, that these secreted factors were able to upregulate p16 Ink4a in islet cells, meaning that the SASP from β-cells is functional. Importantly, we saw different SASP profiles between primary rodent and human β-cells and with β-cell derived MIN6 cells. This should not be surprising since as recently shown by (De Cecco et al., 2019), senescence has a progressive development and our experiments document cross sectionally the stage of senescence at a given point. In addition, differences may also be due to distinct experimental approaches in our two in vitro models. In primary β-cells, we observed a correlation between β-gal activity and SASP expression without any experimental intervention. In MIN6 cells, we sought to identify a causal relationship between senescence-inducing chemicals (H2O2, doxorubicin) and SASP expression. These chemicals may have also induced other cellular changes that modified SASP expression. Further experiments should be performed to understand the whole senescence process in different models of β-cells and only then will comparisons between models and species be valid.

Senolytic therapies, which specifically target senescent cells, have recently been shown to be beneficial to an array of age-related conditions such as hepatic steatosis, stem cell biology and longevity (Chang et al., 2016; Xu et al., 2018). One of the hallmarks of senescent cells is their resistance to apoptosis and coupled with their secretion of SASP, they represent a cell population that can induce dysfunction and senescence in their neighboring cells. Using the INK-ATTAC transgenic mouse, we were able to specifically delete p16 _Ink4a_+ cells in models of aging and insulin resistance (S961 and HFD) through the administration of B/B homodimerizer. In all three models, this treatment improved glucose homeostasis, β-cell function and β-cell gene expression profile.

One way to target senescent cells is through drugs that focus on upregulated antiapoptotic pathways. At least 5 anti-apoptotic pathways have been described in senescent cells (reviewed in (Kirkland et al., 2017). Our RNASeq data indicated that at least two of these pathways HIF1α and Bcl2 upregulated. We tested drugs specific for each of these two pathways in vitro: quercetin and quercetin + dasatinib for HIF1α and ABT263 for Bcl2 pathway. In line with previously published work regarding beneficial effects of quercetin + dasatinib in hepatic steatosis (Ogrodnik et al., 2017), physical function and lifespan (Xu et al., 2018), quercetin + dasatinib selectively cleared βGal+ β-cells and improved blood glucose levels in animals treated with S961 without affecting the proportion of immune cells. ABT 263, which has been shown to rejuvenate aged hematopoietic stem cells in mice (Chang et al., 2016), was effective in selectively reducing the percentage of β-gal+ β-cells. When administered in vivo in mice with S961 and HFD, ABT263 improved their glucose metabolic profile and β-cell genetic identity and specifically reduced p16 Ink4a in islets of treated animals. Even though parallel effects were shown by clearing cells with high p16 Ink4a expression from INK-ATTAC mice and using senolytics to target an antiapoptotic pathway upregulated in senescence cells, it should be kept in mind that the mechanisms behind these approaches are not the same.

To further the translation potential of our studies into humans, we obtained human islets from donors of different age and found that β-cell senescence load is age-dependent and T2D seems to accelerate this process. Moreover, β-gal+ human β-cells express higher levels of P16 INK4A than the β-gal- sorted cells. In mice, ABT263 specifically reduced the load of p16 Ink4a transcript in islets of treated animals, which opens up the possibility that β-gal+ human cells may respond to senolytic therapies.

The main caveat of senolytic therapies is that they target cells and tissues indiscriminately and broadly. Our model shows that a short-term treatment with ABT263 decreased p16 Ink4a mRNA only in β-cells and liver, while p21 Cis1 was unchanged in all tested tissues. This should not be surprising since, as previously mentioned, several different antiapoptotic pathways can be upregulated in senescent cells, and which ones are upregulated vary from one tissue to another. Moreover, it has been shown that ABT263 causes apoptosis of senescent endothelial cells but has little effect on senescent fat cell precursors (Zhu et al., 2015), therefore, it might not be expected to have as big an effect on fat tissue as other senolytics. However, in our long-term HFD model, ABT263 decreased levels of p16 Ink4a in most peripheral tissues, suggesting that tissue specificity of senolytics might also be achieved by the duration of treatment, with some tissues being more sensitive than others. Identifying cell-specific pathways may render some tissue specificity to senolytic therapies. Moreover, the deletion of senescent cells has been shown to be beneficial for different pathologies including hepatic steatosis, autoimmune diabetes, hematopoietic stem cells and longevity, therefore even if not specific, off-target effects may not represent a clinical problem.

Considering specifically ABT263 as a senolytic agent, it was originally tested as a chemotherapeutic agent for its anti-neoplastic effects, however its oncologic uses are limited by thrombocytopenia (Gandhi et al., 2011; Wilson et al., 2010). While its use as a senolytic agent would necessitate lower and less frequent doses to limit some of its side effects, evaluation of other senolytic compounds with greater potency, specificity and less side effects is necessary. Even so, as a proof of concept this approach is a potential new therapeutic avenue that should be further explored in diabetes. We believe that β-cell senolysis might also be applicable for new onset of autoimmune type 1 diabetes (T1D) where rising blood glucose levels are likely to induce glucotoxicity and lead to loss of β-cells (Eisenbarth, 1986), a concept supported by (Thompson et al., 2019). That paper, based on a very aggressive model of autoimmune Type 1 diabetes, the NOD mouse, reveals the presence of a senescent β-cell population. Although the pathophysiology of type 1 and type 2 diabetes are very different, both studies support the presence of a β-cell senescent subpopulation capable of secreting SASP.

An important consideration in assessing β-cell senescence is the potential role of immune cells infiltrating the islets and contributing to the changes in SASP factors, β-cell dysfunction, and even whether a reduction of these immune cells could be responsible for the beneficial effects of ABT263, which is known to cause neutropenia. To evaluate the potential participation of the immune system, we characterized the presence of immune cells in the islets from the mice we were working with: they represent only 0.5% of the total cell population while 90% are β-cells. Also, the diameters of the β-gal+ (14 μm) and β-gal-(12 μm) cells did not overlap with that of resident macrophages (20–80 μm). Therefore, we believe that the effects we saw were not due to immune cells but were specific to changes in the number and phenotype of the β-cells.

In summary, we have established a β-cell senescence signature and shown that β-cell senescence plays a role in the loss of function and identity, and this process is accelerated by insulin resistance. Using both transgenic and pharmacological senolytic models, we showed that these changes can be delayed and even reversed leading to a recovery of β-cell function and identity. These pathways are preserved in human β-cells opening up a new and exciting approach to address the decline of β-cell compensation in T2D.

Limitations of Study

Senescence is a progressive, multistep process that involves the recruitment of different pathways. At this point, the senescence progression in β-cells is not fully understood, neither is it clear which steps along the pathway are reversible if the metabolic stress were removed versus those that are irreversible and can only be targeted through senolytic therapies. As we move forward into understanding the specifics of β-cell senescence mechanisms, these points will become clear as will the most effective interventions at each step. The main caveat of senolytic therapies is that they target cells and tissues indiscriminately and broadly since they are directed at pathways that can be upregulated in all cell types. However, it would be desirable to obtain senolytic drugs that have a higher potency when administered orally as the one we observed with the transgenic INK-ATTAC model. Hopefully, as companies and groups are actively working in identifying potent senolytic compounds with tolerable side effects, the availability of such a drug is feasible in the near future.

STAR METHODS TEXT

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Cristina Aguayo-Mazzucato (ude.dravrah.nilsoj@otacuzzamoyauga.anitsirc).

EXPERIMENTAL MODELS AND SUBJECT DETAILS

Animals

All experiments were conducted at Joslin Diabetes Center with approval of its Animal Care and Use Committee; mice were kept on a conventional facility in a 12-hour light/dark cycle with water and food ad libitum, with a temperature between 22.2–22.7°C. When specified, DIO very high fat diet (VHFD) 60kcal% fat (Fisher Scientific) was used for the specified amount of time. C57Bl6/J mice were acquired from Jackson Labs. Breeding pairs of INK-ATTAC mice (C57Bl/6) were the gift of Dr. Jan van Deursen (Baker et al., 2011) and all the animals used came from our colony. Both male and female mice were used except when noted.

Cell lines.

For induction of senescence models and transfection experiments murine β-cell line MIN6 cells were used and maintained in high-glucose Dulbecco’s modified Eagle’s medium (DMEM-H) supplemented with 15% FBS and 0.05% b-Mercaptoethanol (99% Cell culture tested) at 37°C, 5% CO2. These were originally received from Dr. Junichi Miyazaki, Osaka University Medical School and the sex of the cell line is not available.

Human tissue

Pancreases from adult brain-dead donors were provided by the NIDDK-funded Integrated Islet Distribution Program (IIDP) at City of Hope. Upon arrival, islets were cultured in CMRL media 5.5 mM glucose complemented with 10% FBS, 1% Glutamax, 1% Pen/Strep. After overnight culture, islets were dispersed into single cells using 0.25% Trypsin-EDTA for 15 mins at 37°C. Single cells were then FACS-sorted based on acidic β-gal activity (Aguayo-Mazzucato et al., 2017). Additionally paraformaldehyde fixed paraffin pancreatic sections of donor human pancreas were obtained through nPOD and from the Joslin Clinical Islet Isolation core for immunostaining. All tissues were approved for research and had IRB exempt status and we used the samples of suitable donors (in terms of age and diabetes status) that became available from IIDP, Joslin and NPOD. Details of donors are given in Supplemental Table 2.

METHOD DETAILS

Assessment of glucose homeostasis in animals

Body weight and morning fed glucose levels were monitored longitudinally. Blood glucose values were measured using a Contour glucometer (Bayer) on blood from tail snip. For intraperitoneal glucose tolerance tests, blood samples from mice fasted overnight (15 h) were collected at 0, 15, 30, 60, 90, and 120 min after an intraperitoneal injection of glucose (2 g/kg body weight). For in vivo GSIS, the insulin was measured from serum collected at the 0 and 15 min timepoints of the IPGTT. Concentrations were determined using the Alpco Stellux rodent insulin ELISA kit (NH). For insulin tolerance tests, mice were fasted for 4 h, insulin (Humulin R; Eli Lilly, Indianapolis, IN; 0.5 units/kg body weight) injected intraperitoneally, and blood glucose measured at 0, 15, 30, and 60 min.

Under anesthesia, pancreas was excised for either fixation in 4% (para)-formaldehyde (PFA) for 2h and embedded in paraffin for histology or islets isolated by collagenase digestion (Gotoh et al., 1987) and handpicked for RNA or FACS sorting. Briefly, isolation of pancreatic islets was done after a pancreatic ductal injection of collagenase followed by a stationary in vitro digestion for 18 minutes in a 37oC waterbath. After the reaction was stopped, islets were isolated using a Ficoll gradient followed by centrifugation. The tissue in the interface was collected, washed, sedimented and handpicked for purity.

Senolytic treatments.

The deletion protocol of p16Ink4a-expressing cells for INK ATTAC mice consisted in the administration of two 3-day courses of B/B homodimerizer (10mg/kg) or with 7–14 days in between courses to activate the caspase-8 moiety. In vitro. A dose response kill curve for ABT263 and for quercetin, quercetin (50 uM) + dasatinib (250 nM) were used on FACS sorted β-gal+ and β-gal- cells for 4 days. Cell viability was evaluated after trypsinizing the cells and quantifying propidium iodine negative cells on MACSQuant. In vivo. Mice were treated with vehicle (DMSO:polyethylene glycol 400:Phosal 50 PG At 10:30:60) or ABT263 (Selleck Chemicals, in ethanol:polyethylene glycol 400:Phosal 50 PG). ABT263 was administered to mice by gavage at 50 mg/ kg body weight per day (mg/kg/d) for 4–5 d per cycle, with a week between the cycles. DQ was administered orally by gavage once per week at a dose 5mg/kg dasatinib, 50mg/kg quercetin.

Induction of Senescence.

MIN6 cells were treated with 200 nM of doxorubicin or 450 μM of H2O2 in DMEM media for 24h. Doxorubicin or H2O2 was removed and media was replaced with DMEM-H, 15% charcoal-stripped FBS, 0.05% β-Mercaptoethanol (99% Cell culture tested), and 1% Penicillin-Streptomycin for 24h to generate conditioned media (CM).

FACS

Isolated rodent or human islets were dispersed using 0.25% Trypsin-EDTA for 15 mins at 37°C and were resuspended in FACS buffer (2% fetal bovine serum (FBS) (Cellgro, Manassas, VA) in PBS). Using a DakoCytomation MoFlo Cytometer (Dako, Ft. Collins, CO) or Aria (BD FACS Aria IIu and BD FACS Aria Special Order Research Product), cells were gated according to forward scatter and the percentage of β-cells by insulin staining ranged from 80–90%. Propidium iodide was used to exclude dead cells. For sorting based on acidic β-gal activity, a fluorescent substrate was used (Enzo Life Sciences enz-kit 130–0010) following the manufacturer’s instructions; incubation time with the substrate was optimized at 37°C for 1h in rodents and 2h in humans. β-gal+ and β-gal- subpopulations were either collected for RNA extraction or plated in previously 804g treated wells and used for staining, analysis of SASP, or evaluation of response to senolytic therapies. For evaluation of cellular composition, zinc -selective indicator FluoZin-3 AM was used as specified by the manufacturer to evaluate the β-cell subpopulations. To detect immune cells, dispersed islets were blocked during 45 mins with a Blocking solution of PBS-CMF, 2% FBS, 1% rat normal serum, 1% goat normal serum followed by an incubation of 30 mins at 4°C in the presence of anti-CD45, anti-F4/80 and anti-CD11b antibodies (Suppl. Table 4) (Calderon et al., 2015). Cells were then washed (PBS-CMF 2% FBS) and analyzed in FACS buffer. Our full FACS gating strategy can be seen in Suppl. Fig. 1.

RNASeq.

β-cells from dispersed islets of 7–8 month old male C57BL/6J mice were FACS sorted based on acidic β-galactosidase activity into β-gal + and β-gal -populations, and RNA extracted using PicoPure Arcturus kit. Seven sets of paired samples each from islets pooled from 30 mice were used for RNAseq. Gene expression profiles using HiSeq v4 SE50 250 mil reads by Hudson Alpha Institute for Biotechnology (Huntsville, AL). Reads were aligned to the mouse genome (GRCm38) using Subread aligner and counted with featureCounts v 1.5.2 (Liao et al., 2014). Read counts were transformed to log2-counts per million (LogCPM), their mean-variance relationship was estimated, their weights were computed with voom (Law et al., 2014), and their differential expression was assessed using linear modeling with the R package limma (Ritchie et al., 2015). P-values were corrected using the Benjamini-Hochberg false discovery rate (FDR). Gene sets based on canonical pathways, ChIP Enrichment Analysis, and Targetscan conserved miRNA targets were tested using the limma Roast method (Wu et al., 2010).

Quantification of secreted SASP .

Mouse β-cells FACS-sorted for β-gal activity were plated in 100 μL of media (RPMI + 10% FBS + 1% Penicillin-Streptomycin). After 24 hours, the media was replaced with 100 μL of charcoal-stripped media (RPMI + 10% Charcoal-Stripped FBS + 1% Penicillin-Streptomycin) and cultured for another 24 hours to produce conditioned media (CM) for analysis of secreted factors. CM for each well of cultured β cells was collected and analyzed using LEGENDPlex (Biolegend) selected panels for Mouse Inflammatory Factors and Mouse Proinflammatory Chemokines. LEGENDPlex panels measure the concentration of candidate proteins using a bead-based immunoassay. In the assay, capture beads are mixed into the CM, and specific antibodies on the beads bind to target proteins. The addition of detection antibodies and streptavidin-phycoerythrin produces unique fluorescent signal for each target protein with signal intensities proportional to the protein concentration. Those fluorescent signals were measured using MACSQuant and FACSAria machines. For transfer experiments, β-gal+ and β-gal- cells were plated in a 96-plate and cultured in the presence of charcoal-stripped media for 48 h. CM was then collected and transferred to dispersed and plated dispersed islet cells during 4 days. At the end, cells were picked up, RNA extracted and used for qPCR quantification.

Quantitative real-time PCR (QPCR)

Total RNA isolated with PicoRNA extraction kit (Arcturus) or RNEasy Plus Mini Kit (QIAGEN) was reverse transcribed (SuperScript reverse transcriptase, Invitrogen). QPCR used SYBR green detection and specific primers (Suppl. Table 3). Samples were normalized to b-actin for mouse and TBP for human, and the comparative CT (threshold cycle) method used to calculate gene expression levels. We generated indices for presentation of the PCR data : Aging index: mean of normalized values of p16 Ink4a , p21 Cis1 , Bambi and Igf1r. SASP Index: Mean of normalized values _Il6, Il1a, Ctsb, Plau, Cd68, Serpine 1, Igfbp3, Igfbp5, Fgf2, Cxcl2, Cxcr4, Lamb1, Lamc1, Hgf, Ccl2, TNF_α. β-cell Index: Mean of normalized values of Ins1, Mafa, Pdx1, Neurod1, Nkx6.1, Gck.

S961 treatment.

S961 was a generous gift from Dr. Lauge Schaffer (Novo Nordisk) (Schaffer et al., 2008). Vehicle (PBS or NaCl 0.9%) or 20 nmol S961 was loaded into Alzet osmotic pump 2001 and implanted subcutaneously on the back of mice (Dai et al., 2016) and changed weekly for a total of two weeks.

Knockdown Experiments.

As previously published (Aguayo-Mazzucato et al., 2017) siRNA against mouse p16 Ink4a and RNA interference-negative control were purchased from Thermo Scientific/Dharmacon (Lafayette, CO). MIN6 cells were transfected using DharmaFECT following manufacturers’ instructions. After 48 h transfection, the cells were harvested for RNA and qPCR analysis. Results are presented as fold change to MIN6 cells treated with nonspecific siRNA (siScr).

Immunostaining and morphometric evaluation

Paraffin sections were deparaffinized with ethanol gradients, washed with PBS and antigen retrieval with citric acid was done for P21, P53BP1 and IGF1R. For FLAG, permeabilization was done with a Triton X 0.3% solution. After washing with PBS + 1% NDS, slides were incubated overnight with primary antibody (Suppl. Table 4). This was followed by subsequent washes and incubations for 1h with secondary antibodies that were coupled to Cy3, FITC, Cy5, Texas Red, and DAPI for nuclear staining. For P21 staining, TSA amplification was used. For quantification, islet images were captured systematically covering the whole section in confocal mode on a Zeiss LSM 710 microscope. Every cluster of insulin-stained cells (3–7 cells) or islet (8 or more cells)/section was evaluated; sections were coded and read blindly. For each age, 3–4 animals were evaluated. For human samples, sections from one block from the body of the pancreas from donors as listed in Suppl Table 2. Immunostaining processing and imaging were done in parallel and using the same confocal settings for each antigen such that differences in intensity reflect the differences in protein.

QUANTIFICATION AND STATISTICAL ANALYSIS

Data are shown as mean ± SEM. For statistical analysis, unpaired Student’s t-tests were used to compare two groups, and one-way ANOVA followed by post-hoc test for more than two groups. Normality analysis was performed and non parametric statistics were run when samples did not meet the criteria for normal distribution. A p value <0.05 was considered significant. Prism software was used for graphs and statistical analysis (significance and distribution). Animals were assigned to either control, intervention or treatment groups depending on their age and gender to have equal distribution among all groups. In the case of INK ATTAC animals, their genotype was used to determine which animals would be in the B/B homodimerizer group, trying to have an equal representation in the vehicle treated groups.

Processing of samples for qPCR, slides and image quantification were done in a blind manner. The sample size was a minimum of 3 per group and was determined by the number of animals in the colony of a determined age and gender.

Animals were excluded from the anlysis if they became sick or developed physical anomalies. Data outliers were determined using Grubbs outlier test or deviating more than 2 standard deviations from the mean (unless they were determined to not follow a normal distribution).

DATA AND SOFTWARE AVAILABILITY

The accession number for the genomic data reported in this paper is GSE121539.

​

KEY RESOURCES TABLE

Highlights

• β-cell senescence signature reveals loss of identity and upregulation of SASP factors.
• Insulin resistance accelerates the appearance of senescent β-cells.
• Clearance of senescent cells improves glucose levels, β-cell function and identity.
• In humans, β-cell senescence increases with type 2 diabetes, age and BMI

Context and significance

Type 2 diabetes is a disease that increases with age and defective function of the insulin-producing pancreatic β-cells has a decisive role in its development. However, treatments that target the aging component are currently lacking. This works sheds light on the role of β-cell aging or “senescence” in the development of diabetes by identifying gene expression changes associated with β-cell aging and demonstrating that insulin resistance directly increases the proportion of senescent β-cells. Decreasing the number of aged β-cells is an effective strategy to restore β-cell function, identity and improve glucose metabolism. These results open novel therapeutic approaches against type 2 diabetes.

Supplementary Material

Supplemental Table 3

Supplemental figures and Tables

ACKNOWLEDGEMENTS.

We thank Dr. C.R. Kahn for helpful and insightful discussion. Human islets were provided by NIDDK-funded Integrated Islet Distribution Program (IIDP) at City of Hope, NIH Grant # 2UC4DK098085. Human pancreatic sections were provided by the Joslin Clinical Islet Isolation Core and the Network for Pancreatic Organ Donors with Diabetes (nPOD), a collaborative research project sponsored by JDRF. Organ Procurement Organizations (OPO) partnering with nPOD to provide research resources are listed at http://www.jdrfnpod.org/for-partners/npod-partners. We are grateful to Brooke Sullivan and Aref Ebrahimi for technical and bioinformatic support. This study was supported by grants from the NIH (R01 DK093909 and R01 DK110390, SBW), P30 DK036836 Joslin Diabetes Research Center (DRC Cores: Advanced Microscopy (Chris Cahill), Bioinformatics (Jonathan Dreyfuss and Hui Pan), Flow Cytometry (Angela Wood and Alison Marotta), and Animal Facilities (John Stockton) and P&F to CAM), and P30 DK057521 (BADRC P&F to CAM), the Diabetes Research and Wellness Foundation, and an important group of private donors. AM was supported through the Harvard College Research Program, Office of Undergraduate Research.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

DECLARATION OF INTERESTS

The authors declare no competing interests.

References

• Aguayo-Mazzucato C, van Haaren M, Mruk M, Lee TB Jr., Crawford C, Hollister-Lock J, Sullivan BA, Johnson JW, Ebrahimi A, Dreyfuss JM, et al. (2017). beta Cell Aging Markers Have Heterogeneous Distribution and Are Induced by Insulin Resistance. Cell Metab 25, 898–910 e895. [PMC free article] [PubMed] [Google Scholar]
• Baker DJ, Childs BG, Durik M, Wijers ME, Sieben CJ, Zhong J, Saltness RA, Jeganathan KB, Verzosa GC, Pezeshki A, et al. (2016). Naturally occurring p16(Ink4a)-positive cells shorten healthy lifespan. Nature 530, 184–189. [PMC free article] [PubMed] [Google Scholar]
• Baker DJ, Wijshake T, Tchkonia T, LeBrasseur NK, Childs BG, van de Sluis B, Kirkland JL, and van Deursen JM (2011). Clearance of p16Ink4a-positive senescent cells delays ageing-associated disorders. Nature 479, 232–236. [PMC free article] [PubMed] [Google Scholar]
• Bliss CR, and Sharp GW (1992). Glucose-induced insulin release in islets of young rats: time-dependent potentiation and effects of 2-bromostearate. Am J Physiol 263, E890–896. [PubMed] [Google Scholar]
• Bruning JC, Winnay J, Bonner-Weir S, Taylor SI, Accili D, and Kahn CR (1997). Development of a novel polygenic model of NIDDM in mice heterozygous for IR and IRS-1 null alleles. Cell 88, 561–572. [PubMed] [Google Scholar]
• Calderon B, Carrero JA, Ferris ST, Sojka DK, Moore L, Epelman S, Murphy KM, Yokoyama WM, Randolph GJ, and Unanue ER (2015). The pancreas anatomy conditions the origin and properties of resident macrophages. J Exp Med 212, 1497–1512. [PMC free article] [PubMed] [Google Scholar]
• Campisi J, and d’Adda di Fagagna F (2007). Cellular senescence: when bad things happen to good cells. Nat Rev Mol Cell Biol 8, 729–740. [PubMed] [Google Scholar]
• Chang J, Wang Y, Shao L, Laberge RM, Demaria M, Campisi J, Janakiraman K, Sharpless NE, Ding S, Feng W, et al. (2016). Clearance of senescent cells by ABT263 rejuvenates aged hematopoietic stem cells in mice. Nat Med 22, 78–83. [PMC free article] [PubMed] [Google Scholar]
• Coppe JP, Desprez PY, Krtolica A, and Campisi J (2010). The senescence-associated secretory phenotype: the dark side of tumor suppression. Annu Rev Pathol 5, 99–118. [PMC free article] [PubMed] [Google Scholar]
• Cordisco S, Magenta A, Briganti S, Teson M, Castiglia D, Dellambra E, and Guerra L (2017). Accelerated features of senescence in cultured type 2 diabetic skin fibroblasts. Eur J Dermatol 27, 408–410. [PubMed] [Google Scholar]
• Dai C, Kayton NS, Shostak A, Poffenberger G, Cyphert HA, Aramandla R, Thompson C, Papagiannis IG, Emfinger C, Shiota M, et al. (2016). Stress-impaired transcription factor expression and insulin secretion in transplanted human islets. J Clin Invest 126, 1857–1870. [PMC free article] [PubMed] [Google Scholar]
• De Cecco M, Ito T, Petrashen AP, Elias AE, Skvir NJ, Criscione SW, Caligiana A, Brocculi G, Adney EM, Boeke JD, et al. (2019). L1 drives IFN in senescent cells and promotes age-associated inflammation. Nature 566, 73–78. [PMC free article] [PubMed] [Google Scholar]
• Eisenbarth GS (1986). Type I diabetes mellitus. A chronic autoimmune disease. N Engl J Med 314, 1360–1368. [PubMed] [Google Scholar]
• Fan R, Kang Z, He L, Chan J, and Xu G (2011). Exendin-4 improves blood glucose control in both young and aging normal non-diabetic mice, possible contribution of beta cell independent effects. PLoS One 6, e20443. [PMC free article] [PubMed] [Google Scholar]
• Gandhi L, Camidge DR, Ribeiro de Oliveira M, Bonomi P, Gandara D, Khaira D, Hann CL, McKeegan EM, Litvinovich E, Hemken PM, et al. (2011). Phase I study of Navitoclax (ABT-263), a novel Bcl-2 family inhibitor, in patients with small-cell lung cancer and other solid tumors. J Clin Oncol 29, 909–916. [PMC free article] [PubMed] [Google Scholar]
• Goldstein S, Littlefield JW, and Soeldner JS (1969). Diabetes mellitus and aging: diminished planting efficiency of cultured human fibroblasts. Proc Natl Acad Sci U S A 64, 155–160. [PMC free article] [PubMed] [Google Scholar]
• Goldstein S, Moerman EJ, Soeldner JS, Gleason RE, and Barnett DM (1978). Chronologic and physiologic age affect replicative life-span of fibroblasts from diabetic, prediabetic, and normal donors. Science 199, 781–782. [PubMed] [Google Scholar]
• Goldstein S, Moerman EJ, Soeldner JS, Gleason RE, and Barnett DM (1979). Diabetes mellitus and genetic prediabetes. Decreased replicative capacity of cultured skin fibroblasts. J Clin Invest 63, 358–370. [PMC free article] [PubMed] [Google Scholar]
• Gotoh M, Maki T, Satomi S, Porter J, Bonner-Weir S, O’Hara CJ, and Monaco AP (1987). Reproducible high yield of rat islets by stationary in vitro digestion following pancreatic ductal or portal venous collagenase injection. Transplantation 43, 725–730. [PubMed] [Google Scholar]
• Gregg BE, Moore PC, Demozay D, Hall BA, Li M, Husain A, Wright AJ, Atkinson MA, and Rhodes CJ (2012). Formation of a human beta-cell population within pancreatic islets is set early in life. J Clin Endocrinol Metab 97, 3197–3206. [PMC free article] [PubMed] [Google Scholar]
• GSEA, B.I. Gene Set Enrichment Analysis. Molecular Signatures Database v6.2. (Broad Institute). [Google Scholar]
• Hayflick L (1965). The Limited in Vitro Lifetime of Human Diploid Cell Strains. Exp Cell Res 37, 614–636. [PubMed] [Google Scholar]
• Helman A, Klochendler A, Azazmeh N, Gabai Y, Horwitz E, Anzi S, Swisa A, Condiotti R, Granit RZ, Nevo Y, et al. (2016). p16(Ink4a)-induced senescence of pancreatic beta cells enhances insulin secretion. Nat Med 22, 412–420. [PMC free article] [PubMed] [Google Scholar]
• Jonas JC, Sharma A, Hasenkamp W, Ilkova H, Patane G, Laybutt R, Bonner-Weir S, and Weir GC (1999). Chronic hyperglycemia triggers loss of pancreatic beta cell differentiation in an animal model of diabetes. J Biol Chem 274, 14112–14121. [PubMed] [Google Scholar]
• Kirkland JL, Tchkonia T, Zhu Y, Niedernhofer LJ, and Robbins PD (2017). The Clinical Potential of Senolytic Drugs. J Am Geriatr Soc 65, 2297–2301. [PMC free article] [PubMed] [Google Scholar]
• Krishnamurthy J, Ramsey MR, Ligon KL, Torrice C, Koh A, Bonner-Weir S, and Sharpless NE (2006). p16INK4a induces an age-dependent decline in islet regenerative potential. Nature 443, 453–457. [PubMed] [Google Scholar]
• Krishnamurthy J, Torrice C, Ramsey MR, Kovalev GI, Al-Regaiey K, Su L, and Sharpless NE (2004). Ink4a/Arf expression is a biomarker of aging. J Clin Invest 114, 1299–1307. [PMC free article] [PubMed] [Google Scholar]
• Kushner JA (2013). The role of aging upon beta cell turnover. J Clin Invest 123, 990–995. [PMC free article] [PubMed] [Google Scholar]
• Law CW, Chen Y, Shi W, and Smyth GK (2014). voom: Precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biol 15, R29. [PMC free article] [PubMed] [Google Scholar]
• Leahy JL (2005). Pathogenesis of type 2 diabetes mellitus. Arch Med Res 36, 197–209. [PubMed] [Google Scholar]
• Liao Y, Smyth GK, and Shi W (2014). featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930. [PubMed] [Google Scholar]
• Lopez-Otin C, Blasco MA, Partridge L, Serrano M, and Kroemer G (2013). The hallmarks of aging. Cell 153, 1194–1217. [PMC free article] [PubMed] [Google Scholar]
• Ogrodnik M, Miwa S, Tchkonia T, Tiniakos D, Wilson CL, Lahat A, Day CP, Burt A, Palmer A, Anstee QM, et al. (2017). Cellular senescence drives age-dependent hepatic steatosis. Nat Commun 8, 15691. [PMC free article] [PubMed] [Google Scholar]
• Poitout V, and Robertson RP (2002). Minireview: Secondary beta-cell failure in type 2 diabetes--a convergence of glucotoxicity and lipotoxicity. Endocrinology 143, 339–342. [PubMed] [Google Scholar]
• Porte D Jr. (2001). Clinical importance of insulin secretion and its interaction with insulin resistance in the treatment of type 2 diabetes mellitus and its complications. Diabetes Metab Res Rev 17, 181–188. [PubMed] [Google Scholar]
• Prentki M, Joly E, El-Assaad W, and Roduit R (2002). Malonyl-CoA signaling, lipid partitioning, and glucolipotoxicity: role in beta-cell adaptation and failure in the etiology of diabetes. Diabetes 51 Suppl 3, S405–413. [PubMed] [Google Scholar]
• Pullen TJ, Khan AM, Barton G, Butcher SA, Sun G, and Rutter GA (2010). Identification of genes selectively disallowed in the pancreatic islet. Islets 2, 89–95. [PubMed] [Google Scholar]
• Rankin MM, and Kushner JA (2009). Adaptive beta-cell proliferation is severely restricted with advanced age. Diabetes 58, 1365–1372. [PMC free article] [PubMed] [Google Scholar]
• Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, and Smyth GK (2015). limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43, e47. [PMC free article] [PubMed] [Google Scholar]
• Rodier F, and Campisi J (2011). Four faces of cellular senescence. J Cell Biol 192, 547–556. [PMC free article] [PubMed] [Google Scholar]
• Scaglia L, Cahill CJ, Finegood DT, and Bonner-Weir S (1997). Apoptosis participates in the remodeling of the endocrine pancreas in the neonatal rat. Endocrinology 138, 1736–1741. [PubMed] [Google Scholar]
• Schaffer L, Brand CL, Hansen BF, Ribel U, Shaw AC, Slaaby R, and Sturis J (2008). A novel high-affinity peptide antagonist to the insulin receptor. Biochemical and biophysical research communications 376, 380–383. [PubMed] [Google Scholar]
• Stein GH, Drullinger LF, Soulard A, and Dulic V (1999). Differential roles for cyclin-dependent kinase inhibitors p21 and p16 in the mechanisms of senescence and differentiation in human fibroblasts. Mol Cell Biol 19, 2109–2117. [PMC free article] [PubMed] [Google Scholar]
• Stolovich-Rain M, Hija A, Grimsby J, Glaser B, and Dor Y (2012). Pancreatic beta cells in very old mice retain capacity for compensatory proliferation. J Biol Chem 287, 27407–27414. [PMC free article] [PubMed] [Google Scholar]
• Tchkonia T, Zhu Y, van Deursen J, Campisi J, and Kirkland JL (2013). Cellular senescence and the senescent secretory phenotype: therapeutic opportunities. J Clin Invest 123, 966–972. [PMC free article] [PubMed] [Google Scholar]
• Teta M, Long SY, Wartschow LM, Rankin MM, and Kushner JA (2005). Very slow turnover of beta-cells in aged adult mice. Diabetes 54, 2557–2567. [PubMed] [Google Scholar]
• Thompson PJ, Shah A, Ntranos V, Van Gool F, Atkinson M, and Bhushan A (2019). Targeted Elimination of Senescent Beta Cells Prevents Type 1 Diabetes. Cell Metab. [PubMed] [Google Scholar]
• Thorrez L, Laudadio I, Van Deun K, Quintens R, Hendrickx N, Granvik M, Lemaire K, Schraenen A, Van Lommel L, Lehnert S, et al. (2011). Tissue-specific disallowance of housekeeping genes: the other face of cell differentiation. Genome Res 21, 95–105. [PMC free article] [PubMed] [Google Scholar]
• Weir GC, and Bonner-Weir S (2004). Five stages of evolving beta-cell dysfunction during progression to diabetes. Diabetes 53 Suppl 3, S16–21. [PubMed] [Google Scholar]
• Weir GC, and Bonner-Weir S (2013). Islet beta cell mass in diabetes and how it relates to function, birth, and death. Ann N Y Acad Sci 1281, 92–105. [PMC free article] [PubMed] [Google Scholar]
• Wilson WH, O’Connor OA, Czuczman MS, LaCasce AS, Gerecitano JF, Leonard JP, Tulpule A, Dunleavy K, Xiong H, Chiu YL, et al. (2010). Navitoclax, a targeted high-affinity inhibitor of BCL-2, in lymphoid malignancies: a phase 1 dose-escalation study of safety, pharmacokinetics, pharmacodynamics, and antitumour activity. Lancet Oncol 11, 1149–1159. [PMC free article] [PubMed] [Google Scholar]
• Wu D, Lim E, Vaillant F, Asselin-Labat ML, Visvader JE, and Smyth GK (2010). ROAST: rotation gene set tests for complex microarray experiments. Bioinformatics 26, 2176–2182. [PMC free article] [PubMed] [Google Scholar]
• Xu M, Palmer AK, Ding H, Weivoda MM, Pirtskhalava T, White TA, Sepe A, Johnson KO, Stout MB, Giorgadze N, et al. (2015). Targeting senescent cells enhances adipogenesis and metabolic function in old age. Elife 4, e12997. [PMC free article] [PubMed] [Google Scholar]
• Xu M, Pirtskhalava T, Farr JN, Weigand BM, Palmer AK, Weivoda MM, Inman CL, Ogrodnik MB, Hachfeld CM, Fraser DG, et al. (2018). Senolytics improve physical function and increase lifespan in old age. Nat Med 24, 1246–1256. [PMC free article] [PubMed] [Google Scholar]
• Zhu Y, Tchkonia T, Pirtskhalava T, Gower AC, Ding H, Giorgadze N, Palmer AK, Ikeno Y, Hubbard GB, Lenburg M, et al. (2015). The Achilles’ heel of senescent cells: from transcriptome to senolytic drugs. Aging Cell 14, 644–658. [PMC free article] [PubMed] [Google Scholar]