The Hippo kinase LATS2 impairs pancreatic β-cell survival in diabetes through the mTORC1-autophagy axis

Diabetes results from a decline in functional pancreatic β-cells, but the molecular mechanisms underlying the pathological β-cell failure are poorly understood. Here we report that large-tumor suppressor 2 (LATS2), a core component of the Hippo signaling pathway, is activated under diabetic conditions and induces β-cell apoptosis and impaired function. LATS2 deficiency in β-cells and primary isolated human islets as well as β-cell specific LATS2 ablation in mice improves β-cell viability, insulin secretion and β-cell mass and ameliorates diabetes development. LATS2 activates mechanistic target of rapamycin complex 1 (mTORC1), a physiological suppressor of autophagy, in β-cells and genetic and pharmacological inhibition of mTORC1 counteracts the pro-apoptotic action of activated LATS2. We further show a direct interplay between Hippo and autophagy, in which LATS2 is an autophagy substrate. On the other hand, LATS2 regulates β-cell apoptosis triggered by impaired autophagy suggesting an existence of a stress-sensitive multicomponent cellular loop coordinating β-cell compensation and survival. Our data reveal an important role for LATS2 in pancreatic β-cell turnover and suggest LATS2 as a potential therapeutic target to improve pancreatic β-cell survival and function in diabetes.

B oth type 1 diabetes (T1D) and type 2 diabetes (T2D) result from an absolute or relative decline in pancreatic β-cell function and/or mass 1,2 , and β-cell apoptosis is its hallmark [3][4][5][6] . T1D is an autoimmune disease resulting from selective destruction of pancreatic islet β-cells 7 . T2D is a complex metabolic disorder characterized by insulin resistance as well as decreased insulin secretory function and ultimately reduced β-cell mass, resulting in the development of chronic β-cell dysfunction and relative insulin deficiency 8 . Given these complex causes of βcell failure, inhibition of apoptosis and/or β-cell dysfunction represents a potential therapeutic intervention to the treatment of diabetes. Understanding the varied β-cell responses to metabolic assaults and how diabetes-related signals impair β-cell function and survival is important for the understanding of pathogenesis as well as target identification towards a β-cell-directed therapy of diabetes.
Hippo signaling-first discovered using genetic screens in Drosophila-is an evolutionarily conserved pathway that critically regulates development, growth, and homeostasis of various tissues in response to a wide range of extra-and intracellular signals 9 . The mammalian Hippo pathway constitutes core kinases (MST1/2 and LATS1/2), adaptor proteins (SAV1 for MST1/2 and MOB1 for LATS1/2), downstream terminal effectors (YAP and TAZ), and transcription factors (TEAD1-4). Mammalian Sterile 20-like kinases (MST1/2) and Large-tumor suppressors (LATS1/2) represent core kinases of the Hippo pathway. MST1/2, in complex with a regulatory protein Salvador (Sav1), phosphorylates and activates LATS1/2 kinases, which also form a complex with a regulatory protein Mps-one binder 1 (MOB1). The function of the effector transcriptional coactivator Yes-associated protein (YAP) is therefore mainly regulated by phosphorylation-dependent mechanisms. The kinases LATS1/2 inactivate YAP by direct phosphorylation on multiple serine residues including at S127, enhancing YAP binding to 14-3-3 proteins, its cytoplasmic sequestration and subsequent proteasomal degradation. In association with TEA domain (TEAD) family transcription factors, YAP fosters the expression of target genes, with pro-proliferative and anti-apoptotic outcomes 9-12 . Hippo's dysregulation has been implicated in many human disorders such as cancer and metabolic diseases 9,13 . Hippo components such as MST1, Merlin/NF2, YAP, and TEAD control various aspects of β-cell life including development, β-cell function, survival, and proliferation [13][14][15][16][17][18][19][20] . For example, MST1 is an important regulator of pancreatic β-cell death and dysfunction in human and rodent β-cells. MST1 inhibition restores normoglycemia, β-cell function, and survival under diabetic conditions in vitro and in vivo 14,21 . LATS2, an MST1 downstream substrate, is a ubiquitously expressed serine/threonine kinase and involved in multiple cellular processes such as morphogenesis, proliferation, stress responses, apoptosis, and differentiation [22][23][24][25][26][27] . LATS2 promotes cell death through regulation of multiple downstream targets such as P53, FOXO1, c-Abl, and YAP 24,[28][29][30][31] . In this regard, the MST1-LATS2 axis is an important regulator of apoptosis in the heart: knockdown or genetic deletion of MST1 or LATS2 in cardiomyocytes provides protection against ischemic injury 26,30,32 . As YAP is excluded from mature β-cells during development [13][14][15][16][17][18][19]33 , MST1 and LATS2 core kinases can also act independently of their classical terminal effector YAP. In this study, we investigated whether LATS2 hyper-activation would trigger β-cell death and impaired insulin secretion, whether its deficiency would promote β-cell survival under diabetic conditions in vitro and in vivo as well as the molecular mechanism of pathogenic action of LATS2 in the β-cells.

Results
LATS2 was activated under diabetogenic conditions and induced β-cell death and dysfunction. In order to identify whether LATS is activated in response to diabetogenic stimuli, a recently established bioluminescence-based biosensor (LATS-BS) that monitors the activity of LATS kinase in cells in real-time with accurate quantification, high sensitivity, and excellent reproducibility has been used 34 (Fig. 1a). LATS1/2 kinases phosphorylate their own established target YAP on S127, which exposes the docking site for binding of 14-3-3 proteins leading to YAP cytoplasmic sequestration. LATS-BS consists of a minimal YAP fragment that interacts with 14-3-3 in a phosphorylationdependent manner. Therefore, a LATS-BS construct has been generated with fusion of YAP fragment and 14-3-3 with N-terminal and C-terminal Firefly luciferase fragments (N-luc and C-luc), respectively that measures LATS kinase activity by assessing the complementation between pS127-YAP and 14-3-3 in a LATS-phosphorylation-dependent manner 34 (Fig. 1a). The β-cell line INS-1E was transfected with LATS-BS and control Renilla constructs and then exposed chronically to increased glucose concentrations (glucotoxicity), or its combination with the free fatty acid palmitate (glucolipotoxicity). Both diabetogenic treatments increased bioluminescence signals indicating enhanced exogenously expressed YAP phosphorylation at S127 (LATS specific site) and thus hyper-activated Hippo kinases LATS1/2 (Fig. 1b, c). LATS1/2 hyper-activation under diabetic conditions was further confirmed by immunoblot analysis of increased exogenously expressed YAP-S127 phosphorylation levels triggered by high glucose as well as high glucose/palmitate (Fig. 1d, e). Since the Hippo pathway adaptor protein MOB1 interacts with and activates LATS1/2 kinase activity 23,35 , we analyzed the protein expression of MOB1 in β-cells. Prolonged culture of INS-1E cells under high glucose or high glucose/palmitate up-regulated MOB1 protein levels ( Supplementary  Fig. 1a). Also, overexpression of LATS2 itself increased MOB1 levels in INS-1E β-cells and human islets ( Supplementary Fig. 1b, c). LATS activation (as represented by exogenously expressed pYAP levels) and MOB1 were also increased in islets of hyperglycemic high fat/ high sucrose (HFD) fed mice for 16 weeks ( Fig. 1f and Supplementary Fig. 1d). Similarly, exogenously expressed pYAP and MOB1 protein levels were robustly elevated in islets of another model of T2D, the obese diabetic leptin receptor-deficient db/db mice ( Fig. 1g and Supplementary  Fig. 1e). These data show that LATS activity and its associated protein MOB1 is markedly elevated by pro-diabetic conditions in metabolically stressed β-cells and islets isolated from mouse models of diabetes.
In subsequent experiments we investigated whether LATS2 overexpression is directly detrimental for β-cell survival. LATS2 was overexpressed through adenoviral transduction, which efficiently up-regulated LATS2 (Fig. 2a, b). Overexpression of LATS2 itself was sufficient to induce β-cell apoptosis in both INS-1E cells and human islets, as determined by cleavage of caspase-3 and poly-(ADP-ribose) polymerase (PARP), a downstream substrate of caspase 3 (Fig. 2a, b). In line with these findings, LATS2 overexpression increased the number of TUNEL-positive apoptotic β-cells in human islets confirming β-cell-specific induction of apoptosis by LATS2 hyper-activation (Fig. 2c, d). Furthermore, in isolated human islets, LATS2 overexpression led to impairment of glucose-stimulated insulin secretion (GSIS; Fig. 2e, f). Together, our data show that LATS2 impairs β-cell survival and function.
Loss of the LATS2-MOB1 axis improved β-cell survival in vitro. Further analyses aimed to investigate whether LATS2 inhibition can rescue β-cells from apoptosis under diabetogenic conditions. β-cells were transfected with small interfering RNA (siRNA) against LATS1 and/or LATS2 and then exposed to is activated under diabetogenic conditions. a Schematic structure and mechanism of action for LATS-BS. b-e INS-1E cells transfected with the firefly luciferase reporter plasmids N-luc-YAP15-S127, C-luc-14-3-3, and pRL-Renilla luciferase were treated with 22.2 mM glucose alone or in combination with 0.5 mM palmitate for 24 h. b, c Downstream phosphorylation of the exogenously expressed YAP-S127 fragment was determined by firefly luciferase activity and normalized to the Renilla signal (n = 3 biologically independent samples). d, e Representative western blots and pooled quantitative densitometry analysis for exogenously expressed YAP-127 by a phospho-specific antibody are shown (n = 6 biologically independent samples). Isolated islets from f HFD-treated C57BL/6 J mice for 16 weeks or g the obese diabetic leptin receptor-deficient db/db mice and their corresponding controls cultured overnight and transfected with the N-luc-YAP15-S127 plasmid for 24 h. Representative western blots and pooled quantitative densitometry are shown (n = 3 independent experiments) and results were normalized to the respective control conditions and ratios, in which a normal distribution of results cannot be proven, were analyzed. Data are expressed as means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001; all by two-tailed Student's t-tests. glucotoxic and glucolipotoxic conditions as well as the mixture of pro-inflammatory cytokines interleukin-1beta (IL-1β) and interferon-gamma (IFN-γ). Depletion of endogenous LATS2 but not LATS1 activity protected β-cells from glucose-, glucose/palmitate-and cytokine-induced apoptosis as demonstrated by decreased caspase-3-and PARP-cleavage (Fig. 3a-c and Supplementary Fig. 2a, b) showing a potential functional diversity between LATS1 and LATS2 in the regulation of β-cell apoptosis. The pro-survival function of loss of LATS2 in the β-cells was further validated by a second siRNA pool to the LATS2 gene with comparable gene silencing efficiency ( Supplementary Fig. 2c). LATS kinases redundantly exert phosphorylation-induced inactivation of the transcription factors YAP/TAZ 36 . We thus assessed whether LATS2's canonical kinase activity was necessary for its pro-apoptotic function by overexpressing LATS2 kinase dead (KD) mutant (K697R), where the ATP-binding site K697 was mutated to another positively charged amino acid arginine (R) 37 . LATS2-KD overexpression profoundly reduced the levels of β-cell apoptosis under diabetogenic conditions (Fig. 3d, e). Also, β-cell specific LATS2 deletion in mice markedly suppressed the number of TUNEL-positive β-cells under a pro-diabetic milieu in vitro (Fig. 3f, g). In human islets, LATS2 was silenced by siRNAmediated transfection as well as adenoviral-mediated infection either with Ad-shLATS2 or control shScr viruses. In line with rodent β-cells, apoptosis triggered by pro-inflammatory cytokines as well as by the mixture of high glucose/palmitate was diminished by LATS2 knockdown in isolated primary human islets (Fig. 3h, i). Consistently, repression of LATS2 by shRNAmediated depletion of LATS2 expression efficiently prevented human β-cell death induced by diabetogenic conditions in human islets (Fig. 3j, k). Our data show that loss of LATS2 expression markedly protected both rodents and human β-cells from apoptosis under several diabetic conditions in vitro.
Mechanistically, silencing of endogenous LATS2 abolished MOB1 levels induced by prolonged treatment with high glucose suggesting a LATS2-dependent regulation of MOB1 by prodiabetic stimuli (Supplementary Fig. 2d). To further prove such regulatory LATS2-MOB1 axis, MOB1 was silenced in order to directly assess its pro-apoptotic function. Knockdown of MOB1 antagonized the apoptotic effect of high glucose as well as high glucose/palmitate in INS-1E β-cells (Fig. 4a, b). Consistently, MOB1 knockdown antagonized LATS2-induced caspase-3 cleavage ( Fig. 4c) indicating an indispensable role of MOB1 in the mechanism of LATS2-induced β-cell apoptosis. These data suggest the LATS2-MOB1 axis as a determinant for β-cell apoptosis under a diabetic milieu in β-cells.
β-cell specific LATS2 ablation protected from STZ induced diabetes in vivo. As LATS2 depletion protected from β-cell apoptosis under multiple diabetic conditions in vitro, we hypothesized that LATS2 deficiency may protect from diabetes development in vivo. To test this hypothesis, we generated β-cellspecific LATS2 knockout mice (β-LATS2 −/− ) by crossing LATS2 floxed (LATS2 fl/fl ) mice with the β-cell-specific Cre transgenic line driven by the rat insulin-2 promoter (Rip-Cre). Cre recombinase expression, as well as Rip-Cre-mediated specific deletion of LATS2 gene in pancreatic β-cell, was confirmed in isolated islets from β-LATS2 −/− and LATS2 fl/fl mice ( Supplementary Fig. 3a). As Rip-Cre has been reported to delete genes in β-cells but also in hypothalamic neurons 38 , the specificity of LATS2 gene deletion was tested in genomic DNA from isolated liver, spleen, kidney, heart, hypothalamus, and pancreatic islets of β-LATS2 −/− mice. PCR analysis demonstrated that Cre-mediated LATS2 deletion was islet specific with no leakage in the hypothalamus or any other tested tissues ( Supplementary Fig. 3b). β-LATS2 −/− mice were viable, fertile and showed no difference in food intake and body weight neither under a normal (ND) nor under a high fat diet (HFD) compared to LATS2 fl/fl mice, or to LoxP-negative mice ( Supplementary Fig. 3c, d). To assess whether β-LATS2 −/− might protect against β-cell injury and diabetes, we induced diabetes by multiple-low dose streptozotocin (MLD-STZ) injection in β-LATS2 −/− , LATS2 fl/fl , and Rip-Cre mice. While MLD-STZ injection induced progressive hyperglycemia and severely impaired glucose tolerance in LATS2 fl/fl and Rip-Cre mice,  . f, g Isolated islets from β-LATS2KO and Rip-Cre control mice were recovered after isolation overnight and exposed to diabetogenic conditions (IL-1β/IFNγ or the mixture of 22.2 mM glucose and 0.5 mM palmitate (HG/ Pal)) for 72 h. β-cell apoptosis was analyzed by double staining of TUNEL (black nuclei), and insulin (green). Representative images (f) and quantitative percentage of TUNEL positive β-cells (g) are shown (n = 8, 8, 9, 10, 10, 8 mice, respectively for WT cont, WT HG/Pal, WT IL/IF, β-LATS2-KO cont, β-LATS2-KO HG/Pal, and β-LATS2-KO IL/IF conditions). h, i Representative western blots of human islets transfected with siLATS2 or transduced with Ad-hShLATS2 and treated with h 22.2 mM glucose plus 0.5 mM palmitate or i mixture of IL/IF for 72 h (n = 2 different human islets isolations). j, k Human islets transduced with Ad-hShLATS2 or Ad-shScr control were exposed to diabetogenic conditions (IL-1β/IFNγ or the mixture of 22.2 mM glucose and 0.5 mM palmitate (HG/Pal)) for 72 h. β-cell apoptosis was analyzed by double staining of TUNEL (black nuclei) and insulin (green). Representative images (j) and quantitative percentage of TUNEL positive β-cells (k) (n = 4 different human islets isolations) are shown. All lanes were run on the same gel but were noncontiguous (b, d, e, h, i). Data are expressed as means ± SEM. Pooled quantitative densitometry of western blots were normalized to the respective control conditions and ratios (except c), in which a normal distribution of results cannot be proven, were analyzed. *p < 0.05, **p < 0.01, ***p < 0.001; all by two-tailed Student's t-tests. scale bar depicts 10 μm. random blood glucose levels were significantly reduced and glucose tolerance improved in β-LATS2 −/− mice (Fig. 5a, b). Also, glucose-induced insulin secretion was fully blunted in MLD-STZtreated LATS2 fl/fl and Rip-Cre mice, but significantly restored in β-LATS2 −/− mice, together with an increased insulin-to-glucose ratio, compared to both LATS2 fl/fl and Rip-Cre controls ( Fig. 5c-e). Histological examination of the pancreas and quantification of β-cell mass revealed an increased β-cell mass in STZinjected β-LATS2 −/− mice, compared to control groups (Fig. 5f). To identify whether the restoration in β-cell mass was a result of increased β-cell numbers due to enhanced β-cell proliferation and/or decreased β-cell apoptosis, we next assessed β-cell proliferation and apoptosis in response to MLD-STZ treatment. A significant increase in double-labeled proliferation marker Ki67/ insulin-positive β-cells was observed in MLD-STZ-treated β-LATS2 −/− relative to control from LATS2 fl/fl mice (Fig. 5g, h). Additionally, TUNEL-positive β-cells were markedly reduced in β-LATS2 −/− mice compared to the control group (Fig. 5i, j). This suggests a combined additive impact on augmented proliferation as well as reduced apoptosis as mechanism of β-cell mass restoration in β-LATS2 −/− mice. Altogether, our data show that β-cell-specific ablation of LATS2 diminished progressive hyperglycemia and improved glucose tolerance, insulin secretion, and β-cell mass in the MLD-STZ mouse model of β-cell destruction and diabetes.
β-cell specific LATS2 ablation protected from HFD induced diabetes in vivo. In a second model of diet induced diabetes, we checked whether LATS2 is critical for the long-term β-cell compensatory response by subjecting 8-week-old control Rip-Cre and β-LATS2 −/− mice to normal or diabetogenic high-fat/high sucrose diets (ND or HFD) for 17 weeks, which led to chronic hyperglycemia, insulin resistance as well as β-cell failure in wildtype mice 14,39 . HFD treatment revealed impaired glucose tolerance in Rip-Cre control mice compared to the ND-treated group.
Conversely, HFD fed β-LATS2 −/− mice exhibited marked improvement in glucose tolerance relative to the HFD Rip-Cre control (Fig. 6a). To further examine β-cell function, an in vivo GSIS was performed. A significant increase in basal (0 min) and glucose-stimulated insulin levels (15 and 30 min) was found in HFD fed β-LATS2 −/− mice compared to Rip-Cre counterparts ( Fig. 6b) indicating higher insulin secretion in LATS2 deleted βcells. Consistent with the improved metabolic phenotype in the HFD model, a marked compensatory β-cell mass expansion in response to high-fat/high sucrose feeding was observed in the β-LATS2 −/− mice. In contrast, Rip-Cre control mice failed to compensatively increase β-cell mass in response to HFD feeding ( Fig. 6c). We next analyzed β-cell proliferation and death in response to HFD. As expected, an increase in Ki-67 positive proliferating β-cells in the β-LATS2 −/− mice vs. their Rip-Cre controls was observed when HFD-treated animals were compared (Fig. 6d, e). While β-cell apoptosis remained unchanged in β-LATS2 −/− mice under the ND, HFD-treated β-LATS2 −/− mice showed a highly attenuated response; the number of TUNELpositive β-cells was ∼70% reduced, compared to HFD-treated Rip-Cre control mice (Fig. 6f, g). This is consistent with the antiapoptotic effect of LATS2 depletion observed in vitro and in the STZ in vivo model. The glucose-lowering effect of LATS2 deletion seems to be fully attributable to the β-cell, as β-LATS2 −/− HFD mice exhibited comparable insulin sensitivity to that of agematched HFD-treated Rip-Cre control mice, irrespective of the diet ( Supplementary Fig. 3e).
Taken together, these data suggest that the HFD induced LATS2 hyper-activation has a detrimental impact on β-cell viability, β-cell compensatory response and insulin secretion in the diet induced HFD model of β-cell decompensation and diabetes.
LATS2 induced β-cell apoptosis by activating the mTORC1 pathway. Previous studies revealed β-cell-mTORC1 hyper- activation under chronic diabetogenic conditions; in islets isolated from T2D patients and from animal models of T2D as well as in β-cells cultured under diabetes-associated glucotoxic conditions [40][41][42][43] . It is still not well-understood which signals regulate such mTORC1 hyperactivation and downstream β-cell apoptosis. Both, Hippo and mTORC1 are major growth/viability regulating pathways in our cells and it is not surprising that they mutually control each other [44][45][46][47] . mTORC1 activity was profoundly elevated by LATS2 overexpression in INS-1E β-cells ( Fig. 7a) and isolated human islets (Fig. 7b). Hyper-activation of mTORC1 was demonstrated by increased phosphorylation of its downstream target S6K1 at Thr 389 (pS6K), and the direct S6K substrate ribosomal protein S6 at Ser 235/236 (pS6) (Fig. 7a, b). Also, inhibition of mTORC1 by genetic and pharmacological tools restores insulin secretion in human T2D islets as well as in islets of diabetic mice 41,48 and corrects metabolic derangement in metabolically stressed β-cells 49 . Therefore, we further analyzed the LATS2-mTORC1 crosstalk and whether mTORC1 mediates the pro-apoptotic function of LATS2 in the context of diabetes. In order to define whether LATS2 regulate mTORC1 activity under diabetic conditions, LATS2 was first silenced and then, islets/βcells were exposed to elevated glucose or its combination with palmitate. LATS2 knockdown resulted in decreased levels of pS6K1, pS6, and p4EBP1 (mTORC1 readouts) which were upregulated upon exposure to a diabetic milieu in INS-1E β-cells and human islets providing a direct evidence for mTORC1 regulation by LATS2 (Fig. 7c, d). To establish whether inhibition of mTORC1-S6K1 signaling is sufficient to block pro-apoptotic function of LATS2 in β-cells, we blocked mTORC1 by the use of selective inhibitors against mTORC1 (rapamycin) and S6K1 (PF-4708671; S6K1i) 50 (Fig. 7e, f). Rapamycin and S6K1i fully blocked LATS2-induced mTORC1 (represented by pS6) in INS-1E β-cells and human islets; together with counteracting apoptosis, as observed by reduced caspase-3 or PARP cleavage (Fig. 7e, f). In line with this observation, selective inhibition of endogenous mTORC1 by siRNA-mediated silencing of Raptor, mTORC1's critical subunit, efficiently reduced mTORC1 signaling and substantially protected INS-1E β-cells from LATS2-induced apoptosis (Fig. 7g), further corroborating hyper-activated mTORC1 as downstream player of LATS2 in the context of β-cell apoptosis.
Of note, glucose-dependent mTORC1 activation also appears to be partially regulated by Rag GTPases, making Rag a "multi-input nutrient sensor", which signals nutrients to the downstream mTORC1 activation 40,53 . To examine the effect of LATS2 on key components of the growth factor dependent branch, we overexpressed LATS2 in INS-1E β-cells, which activated mTORC1. LATS2 hyperactivation did not change the level of AKT-Ser473 phosphorylation (important event downstream of PI3K-IRS activation triggered by growth factors) as well as TSC2-Thr1462 phosphorylation (AKT specific site; Supplementary Fig. 4a) suggesting that the AKT-TSC1/2 axis is dispensable for LATS2induced mTORC1 activation. In contrast, loss-and gain-offunctional experiments using exogenously introduced RagA mutants 54 showed that LATS2-induced mTORC1 activation is controlled by the amino acid branch. In particular, overexpression of dominant negative GDP bound RagA (HA-RagA-GDP; RagA-T21L) diminished LATS2-induced mTORC1 activation (Supplementary Fig. 4b; HA-metap2 was used as negative control 55 ). Conversely, constitutively active GTP bound RagA (HA-RagA-GTP; RagA-Q66L) restored mTORC1 activation in LATS2 depleted INS-1E β-cells treated with high glucose (Supplementary Fig. 4c). Altogether, these data suggest that LATS2-induced β-cell apoptosis is mediated by Rag-mTORC1 activation.
It has recently been shown that LATS1 phosphorylates Raptor at Ser606, leading to mTORC1 inhibition in HEK293 kidney cells 47 . We took advantage of the phospho-null (S606A) Raptor mutant 47 to investigate the potential regulation of Ser606phosphorylation in pancreatic β-cells. As presented in Supplementary Fig. 5a, overexpression of WT-Raptor as well as S606A mutant in INS-1E β-cells similarly activated mTORC1 signaling represented by higher pS6, pS6K, and p4EBP1, compared to control transfected cells. Also, neither WT-Raptor nor S606A mutant altered high glucose induced S6-phosphorylation or apoptosis in INS-1E cells, compared to the high glucose treated control ( Supplementary Fig. 5b). All this suggests that Raptor-Ser606 phosphorylation is dispensable for basal mTORC1 activation as well as mTORC1's deleterious role under glucotoxic conditions in β-cells.
Bidirectional regulation of LATS2 and autophagy in β-cells. The nutrient-sensing pathway mTORC1 is the most characterized negative regulator of autophagy and its sustained activation compromises autophagic flux and induces apoptosis in pancreatic β-cells 40,56 . As LATS2 activates mTORC1 and the latter inhibits a protective-autophagy response in β-cells, we determined the direct impact of blocked autophagy on reduced β-cell viability in the context of LATS2 signaling. We first tested whether gain-/ loss-of-LATS2 function could modulate the enhanced cell death under diminished autophagy. INS-1E β-cells and isolated human islets were treated with two different well-established late-stage autophagy inhibitors: BafilomycinA1 (Baf) and Chloroquine (CQ). As expected, suppression of autophagic flux by Baf as well as CQ triggered apoptosis, which was further exacerbated by LATS2 overexpression in INS-1E β-cells (Fig. 8a and Supplementary Fig. 6a). Consistently, also LATS2-overexpressing human islets exhibited increased caspase-3 cleavage (Supplementary Fig. 6b). Conversely, Baf-or CQ-induced β-cell apoptosis was greatly decreased by LATS2 silencing in INS-1E β-cells as well as in human islets (Fig. 8b, c and Supplementary Fig. 6c). Together with exacerbated apoptosis, LATS2 impaired the autophagic flux. Both, the autophagic flux markers microtubuleassociated protein 1A/1B-light chain 3 (LC3-II), a key component of the autophagosome membrane, and p62 (also known as SQSTM1), an adapter protein that recruits the cargo proteins into the autophagosome, showed a strong accumulation upon LATS2 overexpression indicating that the forced expression of LATS2 further impaired autophagic flux in human islets ( Supplementary  Fig. 6b). In contrast, loss of LATS2 attenuated LC3-BII and p62 accumulation induced by autophagy blockers in human islets (Fig. 8c) suggesting a restrictive function of LATS2 in autophagic flux. In line with chemical inhibition of autophagy, knockdown of autophagy-related gene 7 (ATG7), a key component of macroautophagy, exacerbated high glucose-induced apoptosis as represented by increased cleavage of caspase 3 and PARP which is reversed by LATS2 silencing further confirming that LATS2 mediates defective autophagy-induced β-cell apoptosis (Fig. 8d). These results show LATS2 as mediator of defective autophagyinduced apoptosis and autophagic flux in β-cells.
To further assess the effect of LATS2 deficiency on mTORC1 and autophagy, immunohistochemical analyses for pS6 and p62 were performed on pancreatic sections isolated from the HFD-fed mice. HFD-treated control mice (Rip-Cre) showed a markedly higher pS6 in pancreatic islet β-cells compared to the ND mice. This is consistent with the previously reported mTORC1 hyperactivation in HFD diabetic islets by us 41 and others 57 . In contrast, pS6-expression was normalized in HFD-treated β-LATS2 −/− mice ( Supplementary Fig. 7a).
In line with our in vitro results in β-cells and human islets, where LATS2 depletion correlated with reduced p62 levels, protein expression of p62 was clearly seen in β-cells in diabetic HFD-fed mice, but not in β-LATS2 −/− mice ( Supplementary  Fig. 7b). These findings show that LATS2 deletion inhibits HFD induced mTORC1 activation as well as p62 accumulation and further support the interplay between LATS2 and the mTORC1autophagy axis.
We also noticed that endogenous LATS2 protein was increased by autophagy inhibition, suggesting LATS2 as potential substrate for autophagy-mediated degradation (Fig. 8b, c and Supplementary Figs. 6c and 8a). The culminating step for both major autophagy pathways macroautophagy and chaperone-mediated autophagy (CMA) is the degradation of substrate proteins in the lysosomes. Therefore, the presence of a non-lysosomal protein in isolated lysosomes corroborates that the protein is an autophagy substrate. To provide direct evidence for LATS2 not only regulating autophagy, but also vc.vs. for autophagy as regulator of LATS2 protein levels in β-cells, we demonstrated the lysosomal localization of LATS2. Because of the relatively low sensitivity of available specific LATS2 antibodies, immunoprecipitation of endogenous LATS2 was not feasible and therefore, lysosomes of LATS2-overexpressing INS-1E β-cells were isolated by immunepurification, the "Lyso-IP method" 58 . Lysosomes were labeled with the HA-expressing lysosomal membrane protein TMEM192 (TMEM192-3xHA) and anti-HA magnetic beads used for immunoprecipitation of intact lysosomes (Fig. 8e). As a negative control for precipitation, Flag-expressing TMEM192 (TMEM192-2xFLAG) was used. HA-TMEM192-isolated lysosomes represented a highly pure fraction as determined by the absence of markers for other cellular compartments such as mitochondria, ER, Golgi, and endosomes ( Supplementary Fig. 8b); 3xHA tagged organelles were successfully isolated seen by the enrichment of HA (TMEM192-3xHA), as compared to the FLAG tagged control (Fig. 8f and Supplementary Fig. 8b). The pulldown efficiency was equal for samples treated with or without CQ as similar levels of HA/TMEM192 as well as LAMP2, the other lysosomal membrane marker, were enriched in both test samples, compared to the Flag precipitated negative control (Fig. 8f). Both the lysosome enriched samples displayed an accumulation of LATS2 protein. LATS2 was considerably higher in the presence of CQ which is unequivocally in line with our previous observation of increased LATS2 by autophagy inhibition. These data suggest LATS2 as substrate for autophagy, as blocking lysosomal degradation by lysosomotropic agents like CQ resulted in the accumulation of LATS2 in the lysosomes. Also, immunofluorescence microscopy of cultured βcells showed colocalization of LATS2 with LAMP1-containing compartments in CQ-treated cells indicating the lysosomal localization of LATS2 (Supplementary Fig. 8c).
In order to identify the specific type of autophagy pathway that may be involved in the lysosomal degradation of LATS2, we selectively targeted ATG7 and lysosome-associated membrane  Fig. 8d) induced substantial upregulation of endogenous LATS2 again confirming LATS2 upregulation by autophagy inhibition and supporting macroautophagy as major autophagic mechanism for LATS2 destruction in pancreatic β-cells. In order to corroborate that macroautophagy is involved in the autophagic regulation of LATS2 in β-cells, we isolated autophagosomes from stable GFP-LC3 expressing INS-1E β-cells, which have a GFP tagged to the N-terminus of the autophagosomal membrane protein LC3. The GFP tag on the cytosolic side of the membrane can then be exploited to isolate autophagosomes 59,60 . Autophagosomes were immunoprecipitated in LATS2-overexpressing INS-1E β-cells using anti-GFP coated magnetic beads (Fig. 8h) in the presence or absence of CQ. LATS2 was enriched in isolated autophagosomes as represented by successful pulldown of GFP (GFP-LC3), compared to IgG control, where LAT2 was absent, indicating the selective LATS2 accumulation in autophagosomes as a direct evidence for LATS2 regulation by macroautophagy (Fig. 8i). While no significant difference was observed in the autophagosomal accumulation of LATS2 by CQ treatment, the level of autophagosomal p62 was higher than without the inhibitor suggesting a cargo-associated accumulation. In addition, the colocalization of LATS2 and LC3 was confirmed by immunofluorescence microscopy in CQ-treated INS-1E cells (Supplementary Fig. 8e).
Another evidence for the macroautophagy-LATS2 interaction comes from direct protein co-immunoprecipitation experiments. Myc-LATS2 was overexpressed in stable GFP-LC3 expressing INS-1E β-cells and anti-GFP magnetic beads were used to immunoprecipitate GFP-LC3 and its potential interacting partners. Subsequent immunoblot analysis revealed that LATS2 co-immunoprecipitated with GFP-LC3 providing an experimental proof for the direct interaction of LC3 and LATS2 (Fig. 8j). We repeated this experiment with normal INS-1E cells, transiently transfected with GFP-LC3 and/or Myc-LATS2 constructs and included all appropriate negative controls (GFP-LC3 alone vs Myc-LATS2 alone). The specific LATS2-LC3 interaction was confirmed and none of the controls showed any unspecific signals ( Supplementary Fig. 8f).
It has been shown in multiple cases that the interaction between target proteins (receptors) and LC3-family proteins is mediated by an LC3-interacting region (LIR) motif [61][62][63] . Therefore, the presence of a LIR would be a potential molecular signature for LC3-interacting proteins. We took advantage of the freely available iLIR database (https://ilir.warwick.ac.uk) developed by Jacomin et al. 64 and searched for putative canonical LIR motifs in the human LATS2 protein. Based on the in silico analysis of experimentally verified functional LIR motifs, Jacomin et al. redefined the previously described LIR motif-WxxL (where x can be any amino acid) to the 6 amino acids consensus sequence-referred to as the xLIR motif: (ADEFGLPRSK)(DEGMSTV)(WFY) (DEILQTV)(ADEFHIKLMPS TV)(ILV), where the residues marked in bold (positions 3 and 6) correspond to the important residues for the interaction with LC3family proteins 64 . Our in silico analysis identified one complex xLIR motif as well as more than ten LIR motifs-WxxL ( Supplementary  Fig. 9a, b) supporting our co-IP data which showed interaction between LATS2 and LC3. Further experimental investigation using LIR-inactive LATS2 mutants is required to verify this. Altogether, our data suggest the existence of a mutual regulatory axis between LATS2 and autophagy to fine-tune the β-cell apoptosis program.

Discussion
The pancreatic β-cell's vast metabolic plasticity as well as its stress response to cope with high metabolic demands and subsequent pro-diabetic signals is directed by the structure and spatiotemporal dynamics of complex signal transduction networks. Diabetes-associated perturbations in these orchestrated networks occur at various levels, resulting in the dysregulation of physiological functional β-cell mass adaptation. Our work provides direct evidence that Hippo pathway's central kinase LATS2 is activated under diabetogenic conditions, which induced β-cell failure through increased β-cell apoptosis and impaired β-cell function, while LATS2's inactivation resulted in resistance to βcell apoptosis, improved glycemia, insulin secretion and β-cell mass in in vitro, ex vivo and in vivo experimental models of diabetes.
We identified LATS2 as a key upstream activator of mTORC1 in stressed β-cells in which mTORC1 inhibition blocked β-cell apoptosis, suggesting that the pro-apoptotic action of LATS2 is mTORC1-dependent. While LATS2 activated mTORC1 in βcells, LATS kinases can also do the contrary in a different cellular context, namely suppress mTORC1 by phosphorylating Raptor at Serine 606 and subsequently impairing the Raptor downstream interaction with Rheb 47 . There are several possible explanations for such distinct effect: (i) LATS1 and LATS2 kinases do not necessarily share redundant functions in all cell types. In the Gan et al. study, the majority of biochemical and functional experiments have been performed in the model HEK293 kidney cell line, and not in primary cells, or other cell types and thus, a unifying nature for the proposed mechanism was not provided. This is supported by data from the homozygous phosphomimetic (S606D; Raptor D/D ) knock-in mice, in which the LATS mediated Raptor-S606 phosphorylation site is constitutively active. Diminished mTORC1 signaling shown in the liver, heart and kidneys but not in the spleen and brain indeed suggests a tissue or cell type dependent regulation of mTORC1 by LATS kinases. (ii) Another explanation may arise from a different cellular context in our and Gan's et al. studies. While contact inhibition -as a result of high cell density-is a driving force and critical factor for the LATS1/2 mediated mTORC1 suppression, LATS2 induced mTORC1 hyperactivation under diabetogenic conditions of stress and nutrient overload in β-cells is likely to be independent of cell density. (iii) Additionally, LATS mediated phosphorylation of Raptor reduces growth factor induced mTORC1 stimulation through Rheb 47 , while we show here that LATS2-induced mTORC1 activation is regulated, at least in part, by the other amino acid branch of mTORC1 activation through its key component Rag-GTPase.
Downstream of mTORC1, autophagy is a key process for maintaining β-cell viability and functional homeostasis 40 . Defective autophagy is a hallmark of β-cell failure in T2D [65][66][67][68] . Our current study expands a previously unrecognized mechanistic link between LATS, mTORC1, and autophagy. Firstly, LATS2 controls the autophagic flux and autophagy-induced apoptosis through the regulation of mTORC1. Secondly, autophagy itself regulates LATS2's protein turnover by directly targeting LATS2. This suggests that LATS2, mTORC1 and autophagy may constitute a stress-sensitive survival pathway (Fig. 9). Under acute stress conditions, autophagy promotes β-cell survival by directly degrading LATS2 and consequently reinforcing protective-autophagy mechanism through a positivefeedback loop. However, prolonged stress activated LATS2 leading to mTORC1 hyper-activation, defective autophagy, ultimately further LATS2 accumulation and subsequent β-cell apoptosis. This antagonism between LATS2 and autophagy suggests that the outcome of the mutual regulation of both pathways under conditions of increased β-cell stress and demand in a diabetic microenvironment is probably determined by the extent and duration of activated LATS2. In this context, aberrant LATS2 activity triggered by diabetic stimuli may shift the balance towards chronic mTORC1 activation, resulting in defective autophagic flux and β-cell apoptosis. This highlights the existence of a functional bidirectional cross-communication between LATS2 and autophagy for the regulation of β-cell viability under physiological conditions and uncovers LATS2 to mediate a crosstalk between the Hippo pathway and autophagy.
The Hippo pathway and autophagy share a complex network with bidirectional connections tightly controlling the autophagic flux in response to extracellular cues, nutrients availability, and metabolic adaptations to direct cellular and systemic homeostasis 69 . For instance, LATS1/2 upstream kinases MST1/2 were shown to differentially regulate autophagic outcomes. While MST1 directly phosphorylates LC3 to foster autophagosomeslysosomes fusion and autophagic flux for intracellular cargo clearance (such as bacteria) in mouse-embryonic fibroblasts (MEFs) 70 , it phosphorylates the autophagy regulator Beclin1, subsequently inhibiting the Beclin1-Vps34 complex, which leads to autophagy inhibition in cardiomyocytes 71 suggesting contextual and perhaps cell-type dependent regulation of autophagy.
Downstream target of LATS1/2 kinases is the Hippo effector, the transcriptional co-activator YAP, which itself is also strongly linked to autophagy 69 . Like LATS2 (but with opposite functional outcomes), YAP not only functions as upstream regulator of autophagy by regulating the degradation of autophagosomes as well as the maturation of autophagosomes 72, 73 , but also acts as an autophagic substrate in a way that the autophagic flux triggers protein turnover of YAP and its targets 74,75 . During embryonic development, YAP is highly expressed throughout the pancreas in a multipotent progenitor stage 13,[16][17][18]33 . However, after differentiation, YAP remains expressed in ductal and acinar cells 18 , but is excluded from β-cells and other endocrine islet cells at the time when the key endocrine progenitor transcription factor and marker Ngn3 emerges 19 coinciding with their very low proliferative capacity. Re-expression of the "disallowed" YAP in human islets specifically fosters β-cell proliferation 19,76 . Single cell RNA-seq data (for example, ref. 77 ), as well as gene and protein functional expression analyses [17][18][19]76 , confirm that YAP is not expressed in terminally differentiated mature primary human and mouse islets and β-cell lines. While YAP signals are excluded as Hippo targets in mature islets and β-cells, functional MST1/2 and LATS1/2 kinases operate Hippo signaling in the absence of YAP directing alternative Hippo downstream effector (s). Thus, in our experimental setting, LATS2 maneuvers its downstream events, i.e., apoptosis, impaired insulin secretion by a crosstalk with autophagy signals through non-canonical YAP independent mechanisms in β-cells. Such YAP-independent actions of LATS have also been reported in other cellular contexts; e.g., the mitotic spindle orientation in Drosophila 78 or the regulation of mTORC1 signaling in mammals 47 in which YAP knockdown has insignificant effects on mTORC1 activity, indicating that the Hippo/LATS pathway primarily regulates mTORC1 independent of YAP. While we could show by multiple experimental settings that LATS-induced effects on β-cell apoptosis and autophagy is mediated by mTORC1 activation, it is equally possible that LATS2 activation and mTORC1 hyperactivation act in parallel under stress-and diabetogenic conditions to dysregulate the β-cell compensatory machinery and induce β-cell death, dysfunction and thus metabolic failure and diabetes.
The role of the Hippo pathway in human disease, especially in cancer has been substantiated by a robust line of research over the past 10 years 9, 10, 79 . Using a multi-model approach, we have identified LATS2 as pro-apoptotic kinase whose abnormal activation led to impaired β-cell survival and function, while its depletion restored functional β-cell mass and protected against diabetes progression. Blocking LATS2 could be a promising strategy to improve β-cell survival in diabetes.

Methods
Cell culture, treatment, and islet isolation. Human islets were isolated from pancreases of non-diabetic organ donors at the Universities of Illinois at Chicago, Wisconsin, Lille or ProdoLabs and cultured on extra cellular matrix (ECM)-coated dishes (Novamed, Jerusalem, Israel) 80 or on Biocoat Collagen I coated dishes (#356400, Corning, ME, USA). Human islets were cultured in complete CMRL-1066 (Invitrogen) medium at 5.5 mM glucose and mouse islets, the clonal rat β-cell line INS-1E, and the human insulinoma CM cell line in complete RPMI-1640 medium at 11.1 mM glucose 14,81 . Islets from β-cell specific LATS2 knockout (β-LATS2 −/− ) and control mice were isolated by pancreas perfusion with a Liberase TM (#05401119001, Roche, Mannheim, Germany) solution 80 according to the manufacturer's instructions and digested at 37°C, followed by washing and handpicking. Human and mouse islets and INS-1E cells were exposed to complex diabetogenic conditions: 22.2 mM glucose, 0.5 mM palmitic acid 82  Organ donors are not identifiable and anonymous. This work includes islet cells from adult brain-deceased donors insufficient in number for clinical transplantation with informed consent (USA) or when there has been no written refusal for organ transplantation after verification there has been no written refusal (national refusal registry) for research; verbal consent for research donation given is requested from next of kin (Europe; France). France has presumed consent legislation in place for deceased donors. All human islet experiments were performed in the islet biology laboratory, University of Bremen. Ethical approval for the use of human islets in this project had been granted by the Ethics Committee of the University of Bremen. The study complied with all relevant ethical regulations for work with human cells for research purposes. 2% calories from fat, carbohydrate, and protein, respectively) or a high fat/ high sucrose diet (HFD, "Surwit" Research Diets, New Brunswick, NJ, containing 58, 26, and 16% calories from fat, carbohydrate and protein, respectively 86 ) for 17 weeks. For both models, random blood was obtained from the tail vein of non-fasted mice and glucose was measured using a Glucometer (Freestyle; TheraSense Inc., Alameda, CA). Heterozygous leptin receptor-deficient mice on the C57BLKS/J background (Lepr db/+ , db/ + ) were purchased from Jackson Laboratory. By breeding of these mice, we obtained diabetic Lepr db/db (db/db) as well as non-diabetic heterozygous Lepr db/+ (db/+) mice. For data presented in Fig. 1 and Supplementary Fig. 1, islets were isolated after 16 weeks of HFD or at the age of 12 weeks (db/db). Mice were killed and pancreases isolated at the end of experiment. All mice used in this experiment were male and housed in a temperature-controlled room with a 12-h light-dark cycle and were allowed free access to food and water in agreement with NIH animal care guidelines, §8 German animal protection law, German animal welfare legislation and with the guidelines of the Society of Laboratory Animals (GV-SOLAS) and the Federation of Laboratory Animal Science Associations (FELASA). All protocols were approved by the Bremen Senate (Senator for Science, Health and consumer protection) and we have complied with all relevant ethical regulations for animal testing and research.
Morphometric analysis. For morphometric data, ten sections (spanning the width of the pancreas) per mouse were analyzed. Pancreatic tissue area and insulinpositive area were determined by computer-assisted measurements using a Nikon MEA53200 (Nikon GmbH, Dusseldorf, Germany) microscope and images were acquired using NIS-Elements software (Nikon). β-cell mass was obtained by multiplying the β-cell fraction by the weight of the pancreas 14 .
LATS-BS luciferase assay. INS1-E cells or isolated mouse islets were transfected with LATS-BS firefly luciferase reporter constructs using jetPRIME transfection reagent (PolyPlus, Illkirch, France). As internal transfection control, pRL-Renilla luciferase control reporter vector (Promega) was co-transfected into each sample. 24 h after transfection, INS-1E cells were treated with 22.2 mM glucose alone or plus palmitate for another 24 h. Thereafter, Western blot analysis (see above) and luciferase assay was performed using Dual-Luciferase Reporter Assay System (Promega) 89 in a parallel set of experiments. Luciferase signal was calculated based on the ratio of luciferase activity of LATS-BS to control reporter vector.
Lyso-IP. Approx. 30 million INS-1E cells were used for each condition. Each dish was transfected with Tmem192-2XFlag/Tmem192-3XHA and LATS2-Myc plasmids after an adapted previously well-established protocol for the isolation of lysosomes 58 . INS-1E cells were washed with ice-cold PBS and collected in 1 ml KPBS (136 mM KCl, 10 mM KH 2 PO 4 , [pH 7.25 adjusted with KOH]) supplemented with Protease and Phosphatase Inhibitors and centrifuged at 1000 × g for 2 min at 4°C. Pellet was resuspended in 950 μl of KPBS and 25 μl suspension was saved for the input. The remaining cell suspension was homogenized with 60 strokes of a dounce homogenizer (125 rpm, setting 1). Lysate was centrifuged at 1000 × g for 2 min at 4°C. The supernatant was incubated with 25 µl of KPBS prewashed anti-HA magnetic beads (ThermoFischer) on rotation at 4°C for 15 min. Beads were separated using a magnet and immuno-captured lysosomes were gently washed 4 times with 1 mL KPBS on rotation at 4°C for 4 min each. Proteins were extracted from bound lysosomes directly by adding 2× loading buffer followed by heating at 95°C for 10 min. Beads were separated, sample was spun shortly and collected for immunoblot analysis, for which half of the eluted sample was loaded on the gel (approximately 50-60× enrichment of the loaded input).
Isolation of autophagosomes. An adapted protocol from ref. 59 has been used for the isolation of autophagosomes. Approximately 20 million stable GFP-LC3 expressing INS-1E β-cells were used per condition. Cells were washed twice with PBS, scrapped in 2 ml ice-cold PBS and centrifuged at 1000 × g for 3 min at 4°C. Pellet was resuspend in 1 ml cold PBS and 50 µl was removed for input. The remaining suspension was centrifuged at 1000 × g for 3 min at 4°C and pellet was resuspend in 1 ml ice-cold resuspension buffer (0.25 M sucrose, 1 mM EDTA, and 10 mM HEPES-NaOH [pH 7.4]). Cell were lysed by a Dounce homogenizer (60 strokes, 125 rpm, 1st setting) followed by centrifugation at 1000 × g for 10 min at 4°C to remove cell debris and nucleus. The supernatant was then centrifuged at 20,000 × g for 20 min at 4°C to enrich autophagosomes. Pellet containing autophagosomes were resuspended in resuspension buffer and incubated with equilibrated anti-GFP-Trap ® Magnetic beads (Chromotek, Planegg, Germany) or rabbit IgG (CST) conjugated magnetic beads (25 µl) for 2 h at 4°C on rotation. The beads were separated using a magnet and washed four times 4 min each with wash buffer (resuspension buffer supplemented with 0.15 M NaCl) at 4°C. The beads were separated and boiled in 2× SDS sample buffer at 95°C for 10 min and spun down. Beads were separated and the sample was centrifuged at high speed for 5 min and collected for immunoblot analysis.
Co-immunoprecipitation. Protocol adapted from GFP-Trap Magnetic Agarose Kit (Chromotek, Planegg, Germany). For one immunoprecipitation reaction, approximately 10 million GFP-LC3 expressing INS-1E cells or normal INS-1E cells were used. INS-1E cells were washed twice with PBS, scrapped in 2 ml ice-cold PBS and centrifuged at 1000 × g for 3 min at 4°C. Pellet was resuspended in 200 µl icecold lysis buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 0.5 % Nonidet™ P40 Substitute, 0.09% sodium azide) supplemented with Protease and Phosphatase Inhibitors. The tube was placed on ice for 30 min with extensively pipetting every 10 min. Cell lysate was centrifuged at 15,000 × g for 15 min at 4°C and 300 μl dilution buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 0.018% sodium azide) supplemented with Protease and Phosphatase Inhibitors was added to the supernatant. 50 µl of the suspension was saved for input. The remaining suspension was incubated with equilibrated anti-GFP conjugated magnetic beads (25 µl) for 1 h at 4°C on rotation. The beads were separated using a magnet and washed four times 4 min each with dilution buffer at 4°C. The beads were separated and boiled in 2× SDS sample buffer at 95°C for 10 min and spun down. Beads were separated and the sample was collected for immunoblot analysis. Genomic PCR. Genomic DNA was extracted from liver, heart, spleen, kidney, hypothalamus, and isolated pancreatic islets according to the manufacturer's instruction (DNeasy ® Blood&Tissue Kit, QIAGEN). Genotyping was performed using the following primers: flox-F 5′ CCG GAG TCA TTG CTT GTT TT 3′, flox-R 5′ GGA GAT CCT GGG TAC TGC AC 3′, flox-F del 5′ ACA TGA CAC TAC GGG GCC TAG C 3′ 84 . A 300 bp band was amplified from floxed mice and 400 bp band was amplified if the LATS2 gene was deleted.
Statistical analyses. To perform statistical analysis, at least 3 independent experiments were performed for the human islets (3 different donors) and INS-1E cells, as reported in all figure legends. Data are presented as means ± SEM. Mean differences were determined by two-tailed Student's t-tests or one-way ANOVA for multiple group comparisons with Tukey's post hoc test. Densitometry analyses of western blot bands were examined by two-tailed Student's t-tests. Results were normalized to the respective control conditions and ratios analyzed, in which a normal distribution of results cannot be proven. P value < 0.05 was considered statistically significant.