Stress-induced β cell early senescence confers protection against type 1 diabetes

Highlights

• •ATF6 and IRE1α deletion in NOD β cells before insulitis triggers early senescence
• •p21-mediated secretome induces the recruitment of M2 macrophages to the islets
• •ATF6 and IRE1α deficiency reduces terminal β cell senescence and diabetes incidence
• •Early senescence signature is conserved in residual β cells of T1D patients

Summary

During the progression of type 1 diabetes (T1D), β cells are exposed to significant stress and, therefore, require adaptive responses to survive. The adaptive mechanisms that can preserve β cell function and survival in the face of autoimmunity remain unclear. Here, we show that the deletion of the unfolded protein response (UPR) genes Atf6α or Ire1α in β cells of non-obese diabetic (NOD) mice prior to insulitis generates a p21-driven early senescence phenotype and alters the β cell secretome that significantly enhances the leukemia inhibitory factor-mediated recruitment of M2 macrophages to islets. Consequently, M2 macrophages promote anti-inflammatory responses and immune surveillance that cause the resolution of islet inflammation, the removal of terminally senesced β cells, the reduction of β cell apoptosis, and protection against T1D. We further demonstrate that the p21-mediated early senescence signature is conserved in the residual β cells of T1D patients. Our findings reveal a previously unrecognized link between β cell UPR and senescence that, if leveraged, may represent a novel preventive strategy for T1D.

Introduction

Maintenance of cellular and organismal homeostasis under stress conditions is achieved by the activation of highly conserved cellular stress responses. The adaptive stress responses engage a network of signaling mechanisms to resolve the stress and re-establish cellular homeostasis, but persistent and/or excessive stress responses can be detrimental and lead to apoptosis. Cellular stress often affects multiple organelles simultaneously, triggering multiple stress responses. The crosstalk among these stress responses, their coordinated regulation, and non-cell-autonomous effects likely play a critical role in determining adaptive versus maladaptive outcomes, but knowledge of the mechanisms involved in this crosstalk is limited.

One type of cellular stress is endoplasmic reticulum (ER) stress, which is caused by the accumulation of unfolded proteins inside the ER, viral infections, toxins, and chronic inflammation. ER stress triggers the unfolded protein response (UPR) to mitigate the stress and restore cellular homeostasis. The canonical UPR is mediated by the activation of ER-membrane localized proteins inositol-requiring protein-1 (IRE1), protein kinase RNA-like ER kinase (PERK), and activating transcription factor-6 (ATF6). PERK attenuates translation of most mRNAs but also specifically induces translation of the transcription factor Atf4, which directs transcription of genes involved in amino acid metabolism and oxidative stress reduction.¹ IRE1α activation leads to a highly specific splicing reaction of XBP1 mRNA generating spliced XBP1 (sXBP1), a transcription factor that regulates the expression of the UPR target genes, including ER chaperones, ER-associated degradation (ERAD) components, and lipid biosynthetic enzymes. ATF6, upon undergoing proteolytic cleavage in the Golgi, functions as an active transcription factor upregulating target genes encoding ER chaperones, ERAD components, and XBP1. Although the UPR initially attempts to promote adaptation of the cells to ER stress, in the presence of prolonged or severe stress, it induces a pro-apoptotic response that eliminates terminally stressed cells.^2,3

Type 1 diabetes (T1D) results from insulin insufficiency owing to the near-complete destruction of insulin-producing pancreatic β cells by an autoimmune process. Over the last decade, the active participation of pancreatic β cells in their own autoimmune destruction and the impact of aberrant stress responses to T1D disease progression have gained considerable attention.^4,5,6 Early studies demonstrated that β cells exposed to proinflammatory cytokines have significantly increased ER stress and UPR.^7,8 Preclinical and clinical studies provided further evidence for the presence of β cell ER stress and dysregulated UPR long before the initiation of T1D.^9,10,11,12 Pharmacological mitigation of ER stress and inhibition of IRE1α activity prevented T1D in preclinical models.^9,13,14 However, the β cell-specific functions of the other UPR sensors (ATF6 and PERK), the crosstalk between β cells and immune cells, and the intricate relationship between ER stress and other cellular stress responses during autoimmune progression have remained largely unknown.¹⁵

Senescence is a stress response program of stable growth arrest mediated by cyclin-dependent kinase inhibitors such as p21^Cip1 and p16^Ink4a and involves a variety of cellular changes, such as a prosurvival phenotype and a complex and dynamic secretome consisting of growth factors, cytokines, chemokines, and other factors known as the senescence-associated secretory phenotype (SASP).^16,17 The SASP elicits immune surveillance, leading to removal of senescent cells from tissues and restoration of homeostasis,^18,19,20 but when the immune system is compromised, such as during aging, senescent cells accumulate, leading to tissue dysfunction.²¹ Notably, senescent β cells accumulate during the natural history of T1D in non-obese diabetic (NOD) mice and humans and contribute to disease progression, as small molecule targeting of senescence prevents T1D in NOD mice.²² The accumulation of senescent β cells with SASP in T1D suggests a defect or lack of immune surveillance, but whether senescent β cells elicit surveillance by the immune system has not been determined. Moreover, how senescence relates to other β cell stress responses, such as the UPR, is not clear.

Here, we show that deletion of key UPR mediators, Atf6α or Ire1α, in β cells leads to an early senescence program driven by p21, resulting in a unique secretome that recruits M2 macrophages to islets. Consequently, M2-macrophage-mediated anti-inflammatory, immunosuppressive responses and immune surveillance result in markedly diminished terminally senesced β cells, resolution of islet inflammation, reduction of β cell apoptosis, and increased β cell survival, leading to protection from T1D in NOD mice. Analysis of single-cell transcriptomics data from human T1D islets and studies inhibiting ATF6 in human EndoC-βH1 cells and donor islets show that the p21-mediated early senescence signature is also present in residual β cells of T1D patients. Our work demonstrates a novel link between β cell UPR and senescence in the restoration of islet homeostasis and attenuation of T1D.

Results

Atf6 deletion in NOD β cells protects against T1D

Previous reports show that although whole-body deletion of α and β isoforms of Atf6 in mice leads to embryonic lethality,²³ β cell-specific deletion of Atf6α on C57BL/6J background leads to mild glucose intolerance and a decrease in glucose-stimulated insulin secretion.⁹ When crossed with a virally induced diabetes mouse model, Atf6α deficiency in β cells does not alter diabetes incidence.⁹ To get insight into the role of β cell ATF6 in T1D development, we generated β cell-specific Atf6α knockout mice (Atf6^β−/−) on NOD background. To this end, we mated NOD Atf6^fl/fl mice with NOD Ins2Cre^ERT mice.¹³ The resulting Atf6^fl/fl; Ins2Cre^ERT mice were further mated with Atf6^fl/fl mice. Of note, we had previously characterized the NOD Ins2Cre^ERT mice,¹³ which gave rise to the non-littermate controls in this study. To achieve deletion of Atf6 in β cells prior to insulitis, we administered tamoxifen to all pups during postnatal day (P)1–P3 (Figure 1A). We confirmed the β cell-specific deletion of Atf6 by co-staining pancreatic sections with anti-ATF6 and anti-insulin antibodies (Figure 1B). The qPCR analysis showed an ∼60% reduction of Atf6 mRNA in the islets of Atf6^β−/− mice (Figure 1C).