Prolonged Elimination of Negative Feedback Control Mechanisms Along the Insulin Signaling Pathway Impairs β-Cell Function In Vivo

Insulin and IGF-I actions are mediated by their receptors that function as Tyr-kinases. Key substrates for these receptors are the insulin receptor substrate (IRS) proteins IRS1 and IRS2, which integrate many of the pleiotropic effects of insulin and IGF-I. IRS proteins, mainly IRS2, play critical roles in pancreatic β-cells (1). Decreased IRS2 expression causes β-cell apoptosis (1,2) and mice lacking IRS2 develop diabetes 8–10 weeks after birth due to reduced β-cell mass and impaired β-cell function (1). Conversely, increased IRS2 expression promotes β-cell survival (3) and prevents diabetes in Irs2^−/− mice (4).

IRS proteins have a conserved amino terminus, which contains a pleckstrin homology domain flanked by a P-Tyr binding domain that interacts with the juxtamembrane domains of the insulin and IGF-I receptors (5,6). IRS proteins undergo phosphorylation on multiple Tyr residues at their COOH-terminal region, which serves as a docking site for SH2-containing proteins that further propagate insulin and IGF-I signals (7).

IRS proteins contain >70 potential Ser/Thr phosphorylation sites for kinases such as cAMP-dependent protein kinase, protein kinase C, and Akt (reviewed by Shanik et al. [8] and Boura-Halfon and Zick [9]). Insulin-induced Ser/Thr phosphorylation of IRS proteins dissociates them from the insulin receptor (IR), prevents their Tyr phosphorylation, and inhibits their interactions with downstream effectors (10). This serves as a physiological negative-feedback control mechanism, which is used by insulin and IGF-I to turn off their own signaling cascades. However, proinflammatory cytokines and other inducers of insulin resistance take advantage of this mechanism. By activating a number of IRS kinases, they uncouple the IRS proteins from the IR or IGF-I receptor and inhibit their biological activities (11). Accordingly, the mutation of selected inhibitory Ser sites of IRS1, located in close proximity to its P-Tyr binding domain, renders it less prone to the action of IRS kinases. As a result, the mutated IRS1 better propagates insulin and IGF-I actions (12,13). Similarly, the introduction of IRS2 proteins mutated at five potential inhibitory Ser sites (IRS2^5A) into pancreatic β-cells for a short period of time better promoted insulin signaling and improved β-cell survival and function in culture. Furthermore, when islets expressing IRS2^5A were transplanted into diabetic mice, they supported glucose homeostasis better than those expressing IRS2^WT (14). These findings suggest that short-term elimination of negative-feedback control mechanisms along the insulin/IGF-I signaling pathway improves β-cell function under stress.

To further assess the physiological significance of permanently eliminating “inhibitory” Ser phosphorylation sites, we have generated transgenic (Tg) mice that constitutively overexpress either IRS2^WT or IRS2^5A selectively in pancreatic β-cells. Our findings indicate that although the overexpressed IRS2 proteins, mainly IRS2^5A, exert beneficial effects on the animals, mainly at a very young age (5–8 days), a long-term expression of IRS2^5A impairs β-cell function. These findings indicate that the beneficial effects of IRS2^5A (14) are limited in duration and, in the long run, are overridden by its negative effects, leading eventually to β-cell exhaustion and demise.

Male C57BL/6J mice were housed under standard light/dark conditions and were given access to food and water ad libitum; experiments were approved by the Animal Care and Use Committee of the Weizmann Institute of Science (Rehovot, Israel).

Double-Tg heterozygous mice that express Myc-Irs2 proteins (wild type [WT] or 5A) selectively in pancreatic β-cells were generated by crossing mice that express Myc-Irs2 proteins (WT or 5A) driven by the tetracycline operator–based promoter with mice that express the reverse tetracycline-regulated trans activator (rtTA) driven by the rat insulin 2 promoter fragment (15). In brief, pTet-Splice1 plasmids containing the tetracycline response element and Myc-tagged Irs2^WT or Irs2^5A were generated as described previously (14). The plasmids were linearized by digestion with Xho1 and Not1, gel purified, and microinjected into the pronuclei of CB6F1 zygotes. The mice that were born (F generation) were scanned for the presence of Myc-Irs2^5A or Myc-Irs2^WT by PCR (vide infra), and positive mice were outbred with WT CB6F1 mice to generate heterozygote mice (F1 generation). Heterozygote mice were crossed with ICR mice expressing the rtTA under the control of the rat insulin II promoter, consisting of the 9.5-kb 5′ flanking region of the gene (Rip7-rtTA) (16). Crossbreeding of the two Tg lines yielded a double-Tg heterozygote mice (Irs2^5A-β; Irs2^WT-β) that activates their transgene after the administration of tetracyclin to their drinking water. Since the original Tg mice were offspring of a crossbreeding of CB6F1 and ICR mice, the mice were backcrossed with C57BL/6 mice to achieve a homogeneous genetic background. After six backcrosses, the new litters expressed at least ∼98% of the genetic background of C57BL/6 mice.

The intensity of bands in autoradiograms was determined by densitometry carried out on exposures within the linear range. Graphic analysis was performed with National Institutes of Health image software. Results were analyzed using nonparametric comparisons based upon Mann-Whitney U tests. A P value of <0.05 was considered significant. In those studies for which the n values were low, the experiments may be underpowered. To quantify the difference between glucose tolerance tests (GTTs), the area under the curve (AUC) was calculated by the trapezoid rule. The mean AUC ± SEM was determined from at least three individual curves for each experimental condition reported.

The sources of materials, antibodies, and detailed methods concerning mouse genotyping; metabolic studies; mouse diet; preparation of cell extracts; isolation and treatment of murine islets; islet morphology; RNA analysis; glucose-stimulated insulin secretion (GSIS); immunofluorescence microscopy, and immunostaining analysis are provided in the Supplementary Data.

Transient expression in pancreatic islets of IRS2 proteins mutated at five potential inhibitory Ser sites (IRS2^5A; S303A, S343A, S362A, S381A, S480A) improved insulin signaling as well as β-cell survival and function in culture (14). Furthermore, when islets expressing Irs2^5A were transplanted into diabetic mice, they supported glucose homeostasis better than those expressing Irs2^WT (14). Our goal was to follow on these findings and determine the sustained effects of Irs2^5A expression under in vivo settings. To this end, double-heterozygous Tg mice that overexpress both Myc-IRS2 proteins (WT or 5A), driven by the tetracycline operator–based promoter, and the rtTA, driven by the rat insulin-2 promoter fragment, were generated. These mice were expected to express Myc-IRS2 (WT or 5A) selectively in pancreatic β-cells in an inducible manner.

To examine expression of the IRS2 proteins in pancreatic β-cells, doxycycline was administered into the drinking water of Irs2^5A-β and Irs2^WT-β mice. After 6 days, islets were isolated and analyzed for Myc expression. As shown in Fig. 1, there was no expression of Myc in the control Non-Tg animals both at the mRNA and protein levels; however, there was a significant increase in Myc-Irs2 mRNA in islets isolated both from Irs2^5A-β and Irs2^WT-β mice (Fig. 1A). Furthermore, the transgene was expressed at comparable levels in the two mouse lines (Fig. 1A). A marked comparable increase in mRNA levels of total Irs2 (exogenous and endogenous; either WT or 5A) was observed in the Tg animals (Fig. 1B) that was accompanied by increased IRS2 expression in islets of the Tg animals, compared with Non-Tg ones, as revealed by immunoblotting using Myc or IRS2 antibodies (Fig. 1C–E). It revealed a 6- to 10-fold increase in total IRS2 protein levels in the Tg animals (Fig. 1E) that is similar to those previously reported (4) for Irs2 transgenes. Immunohistochemistry of total IRS2 yielded similar, albeit lower, fold increases (vide infra). The mRNA levels of Myc-Irs2 and total Irs2 were also tested in other organs such as the liver, lung, spleen, thymus, testicles, and kidney, all of which were found negative with regard to Myc-Irs2 mRNA expression. Weak mRNA expression (2–4% that of islets) was detected in the brain (Supplementary Fig. 1A and B).

To determine whether the expression of Myc-IRS2 proteins is doxycycline dependent, mice were administered doxycycline for 2 weeks. However, the transcription levels of the exogenous Myc-Irs2 mRNA as well as total Irs2 mRNA were increased to comparable levels regardless of the administration of doxycycline (Fig. 1F and G). This effect was also observed in two other mouse lines (data not shown), suggesting that the TetOn system in all founders was leaky, possibly due to a strong positional effects on the TetOn minimal promoter (17), resulting in the expression of the transgenes from the time the insulin promoter was activated in β-cells, at approximately the 20-somite stage (18).

To further assess the specificity of Myc-Irs2 expression, Myc staining was performed on pancreas sections of Irs2^5A-β and Irs2^WT-β mice (without doxycycline administration). As shown in Fig. 2A, IRS2^WT and IRS2^5A were detected in most insulin-expressing cells; however, Myc staining was not detected in non–β-cells. Similarly, more intense (twofold to threefold) staining of IRS2 itself was evident selectively in β-cells of both young and adult mice (Fig. 2B and C). Of note, although the expression of Myc-IRS2^WT was observed mainly in the cytoplasm, the expression of Myc-IRS2^5A was observed mostly in the nucleus (Fig. 2A). The reason for this difference and its biological significance are currently unknown.

The differences in phenotypes between Irs2^5A-β and Irs2^WT-β male mice (7–8 weeks old), fed a regular diet, were evaluated next. As shown in Supplementary Fig. 2A, there was no significant change in body weight between the mouse lines, and fasting blood glucose levels of Irs2^WT-β and Irs2^5A-β mice were indistinguishable compared with Non-Tg mice (Supplementary Fig. 2B).

It has been previously shown that the overexpression of IRS2 in β-cells of Tg mice does not change β-cell content per islet but leads to a twofold increase in islet area, mainly due to an increased number of normal-sized islets (4). Consistent with these observations, the mean islet area in Irs2^WT-β mice was increased by ∼25% compared with their Non-Tg littermates, whereas the overexpression of IRS2^5A increased islets area by ∼60% compared with Non-Tg mice (Fig. 3A). This effect was mainly due to a greater proportion of large islets (>30 μm²) in Irs2^5A-β mice (Fig. 3B), although the average number of β-cells/islet-area remained unaltered (Supplementary Fig. 3A). The overexpression of Irs2^WT in β-cells also resulted in an ∼1.5-fold increase in the number of islets per pancreas section compared with Non-Tg mice. A further ∼1.5-fold increase was observed in Irs2^5A-β mice compared with Irs2^WT-β mice (Fig. 3C). Thus, our data support previous findings regarding the effects of IRS2^WT on islets morphology and density, whereas the overexpression of IRS2^5A further enhances these effects. Yet, despite the hyperplastic nature of the islets derived from 10-week-old Irs2^5A-β mice, we failed to detect an increase in the cell proliferation marker Ki67 in these islets (Supplementary Fig. 3B and C), although a trend of increased cell proliferation, determined by a twofold increase in proliferating cell nuclear antigen (PCNA)-positive cells, was observed in islets derived from Irs2^WT-β mice, and an even greater increase was observed in islets derived from Irs2^5A-β mice (Fig. 3D and Supplementary Fig. 3D). The increase in PCNA expression could, however, be accounted for, at least in part, by the fact that PCNA marks not only actively dividing cells but also those in the process of DNA repair (19).

β-Cells are highly sensitive to reactive oxygen species that induce β-cell dysfunction (20). We therefore studied the mRNA levels of catalase (21) and superoxide dismutase (SOD) (22) in islets derived from wt and Irs2 Tg mice. As shown in Fig. 4, the mRNA levels of catalase, sod1, and sod2 were significantly increased in islets derived from Irs2^WT-β mice, and an even greater elevation was observed in islets derived from Irs2^5A-β mice. These effects were evident in islets derived from mice fed regular chow (Fig. 4A) or high-fat diet (HFD) (Fig. 4B). Conversely, the mRNA levels of nitric oxide synthase-2 (nos2) were reduced 25% in islets derived from Irs2^5A-β mice compared with Non-Tg animals (Fig. 4C). These findings might be considered as supportive of the concept that IRS proteins can reduce the oxygen burden of pancreatic islets through the activation of Akt and its downstream target FOXO1 (21) and exert beneficial effects on β-cell function.

The effects of IRS2 overexpression on GTT results and insulin secretion were evaluated next. Unlike expectations, Irs2^WT-β mice responded like Non-Tg animals, whereas Irs2^5A-β male mice were glucose intolerant (Fig. 5A). Accordingly, the AUC of Irs2^5A-β mice increased by ∼27% compared with Non-Tg mice (Fig. 5B). The same trend was also observed in IRS2^5A-β female mice (data not shown). Peripheral insulin resistance and/or β-cell dysfunction are the main reasons for impaired GTT results, therefore GSIS was analyzed in islets isolated from these mice. The GSIS of Irs2^WT-β mice was comparable to that of the Non-Tg mice (Fig. 5C); however, the GSIS of Irs2^5A-β mice was reduced by ∼67%. Although this effect could not be explained by significant changes in total insulin content/islet (Fig. 5D), a more sensitive immunostaining of individual β-cells revealed a significant increase in the mean insulin staining only in islets derived from Irs2^WT-β mice (Fig. 5E). These results suggest that sustained overexpression of IRS2^5A in β-cells significantly inhibits GSIS, which may account for the impaired GTT results despite islet hyperplasia.

To better understand the molecular basis for the impaired GTT results and GSIS in Irs2^5A-β mice, we examined the mRNA levels of a number of β-cell transcription factors. A significant reduction of ∼30% in MafA mRNA was observed in islets isolated from Irs2^WT-β, and a much greater reduction (∼60%) was observed in Irs2^5A-β mice, when compared with Non-Tg mice (Fig. 6A). Reduced mRNA and protein levels of Nkx6.1 (∼40–50%) were observed only in islets isolated from Irs2^5A-β mice (Fig. 6B–D and Supplementary Fig. 4A). A similar ∼35% reduction was observed in Glut2 mRNA (Fig. 6E), a downstream target of Nkx6.1 (23). No change in Pdx1 mRNA was observed in the Tg animals (Fig. 6A).

The effects of IRS2 overexpression on animals placed on an HFD (60% kcal fat) were next evaluated. GTT experiments revealed that 2 weeks of eating an HFD significantly impaired, as expected, the glucose tolerance of Non-Tg mice. Further impairment was observed in Irs2^WT-β mice, whereas Irs2^5A-β mice were the most glucose intolerant (Fig. 7A), having the largest AUC (Fig. 7B). This was accompanied by a marked reduction (∼40%) in insulin content in islets isolated from Irs2^5A-β mice, whereas islets isolated from Irs2^WT-β mice showed comparable or even slightly higher insulin content (Fig. 7C and D). These results suggest that when placed on an HFD, the overexpression of IRS2 in β-cells, either WT or even more so 5A, impairs the clearance of blood glucose and reduces insulin content in Irs2^5A-β mice.

Notable changes in mRNA levels of a number of β-cell transcription factors were observed as well. A slight reduction in MafA mRNA as well as significant reductions in Pdx1 and Nkx6.1 mRNA levels was observed only in islets isolated from IRS2^5A-β mice (Fig. 7E). A reduction at the protein levels of Nkx6.1 was detected as well (Fig. 7F and Supplementary Fig. 4B and C). Similarly, reductions (yet not significant) in NeuroD and Ngn3 mRNA levels were observed with no changes in mRNA levels of Nkx2.2 and Kir6.2 (Fig. 7G). A marked reduction in mRNA levels of Top2a (24), a marker of cell proliferation, was observed as well (Fig. 7H).

The efficacy of the unfolded protein response (UPR) influences the ability of islets to meet increased insulin demand under metabolic stressors such as chronic hyperglycemia or insulin resistance (25). To examine the effects of sustained overexpression of IRS2 on the UPR, we measured the gene expression of three key players in this process (XBP1, ATF4, and CHOP). Although no changes were observed in spliced Xbp1 mRNA, we could show a significant reduction in mRNA levels of Atf4 and Chop in islets from Irs2^5A-β mice that had been placed on an HFD, whereas no such effect was observed in islets from Irs2^WT-β mice (Fig. 7I). These findings suggest that the unaltered and even slightly reduced UPR observed in islets derived from Irs2^5A-β mice could be a natural downstream response to the reduced insulin production, insulin content, and GSIS in these cells (26,27).

Based on the above results, and previous findings (14), our working hypothesis predicted that the overexpression of IRS2 might initially be beneficial, yet it leads to overuse of the β-cells, resulting in β-cell demise, similar to the effects observed upon transition from an insulin-resistant state to overt diabetes (28). To further challenge this hypothesis, we assayed for the expression of a number of β-cell transcription factors using islets isolated from young 10-day-old mice. Indeed, the mRNA levels of Pdx1, MafA, and Nkx6.1 were significantly increased twofold to threefold in islets isolated from young Irs2^WT-β mice, but not Irs2^5A-β mice (Fig. 8A), suggesting that the transient beneficial effects of IRS2^5A, might have taken place at even earlier times. To further corroborate these findings, even younger mice (5–8 days old) were analyzed. Indeed, when compared with Non-Tg mice, higher β-cell proliferation (Ki67⁺ cells) was evident in Irs2^5A-β mice (Fig. 8B and Supplementary Fig. 5). Significantly a higher number of cyclinD1⁺ cells was also evident both in Irs2^WT-β and Irs2^5A-β mice (Fig. 8C) combined with the increased expression of PDX1 and NKX6.1 proteins (Fig. 8D and E and Supplementary Fig. 6), indicating that the beneficial effects of IRS2^5A are indeed evident only in very young animals.

This study was aimed to challenge in vivo the hypothesis that selective expression in β-cells of IRS2^5A interferes with negative-feedback control mechanisms along the insulin-signaling pathway and improve β-cell function under stress. For this purpose, Tg mice that overexpress IRS2, either WT or 5A selectively in β-cells, were generated, using the Tet on/off system (29). However, unexpectedly, the IRS proteins were expressed in a constitutive rather than an inducible manner. Thus, whereas in a previous study (14) IRS2^WT and IRS2^5A expression was induced in adult isolated islets by infection with adenovirus, in our in vivo system the IRS2 proteins, under the control of an insulin promoter, were expressed from an early embryonic stage (18).

We have previously shown that IRS2^5A was much more effective than IRS2^WT in protecting β-cells from apoptosis induced by proinflammatory cytokines (14). Furthermore, when expressed in isolated adult mouse islets transplanted into diabetic mice, IRS2^5A was more effective than IRS2^WT in restoring normoglycemia (14). In the current study, we could show that IRS2^5A functions in vivo in some aspects better than IRS2^WT. The expression of IRS2^5A increased the mRNA levels of catalase and sod while lowering those of nos2. Assuming that the protein levels of these enzymes are altered in the same way, this could reduce the oxidative burden inflicted upon β-cells and could exert beneficial effects on islet functionality and insulin secretion (30). The expression of IRS2 also induced islet hyperplasia, with an increased number of big islets (area >30 × 10³ μm²) in Irs2^WT-β mice and an even bigger increase in islets from Irs2^5A-β mice. These results were in accordance with previous findings (4) that reported an increase in the number of islets in Tg mice overexpressing Irs2^WT in β-cells.

Yet, despite certain beneficial effects, most striking is the observation that sustained overexpression of Irs2^5A in β-cells in vivo is rather detrimental to islet functionality, as is evident by the reduced insulin immunostaining per islet and the impaired GTT results and GSIS of Irs2^5A-β mice. These deleterious effects were even more pronounced in mice placed on an HFD. At the molecular levels, we observed a significant reduction in mRNA levels of a number of β-cell transcription factors that are key players in maintaining β-cell function. These included MafA, Nkx6.1, and Pdx1, which were markedly reduced in Irs2^5A-β mice, especially in those placed on an HFD. The loss of MafA is an early indicator of β-cell inactivity, and the subsequent deficit of the more impactful Nkx6.1 results in overt dysfunction associated with T2DM (31). Indeed, the reduced expression of NKx6.1 could account for the reduced transcription of its downstream target Glut2, observed in Irs2^5A-β mice, and could explain the impaired GSIS observed in these animals because Glut2 is the main glucose transporter in pancreatic islets (32). Reduced expression of some elements mediating the UPR, such as ATF4 and CHOP, that were observed selectively in islets derived from Irs2^5A-β mice placed on an HFD could also reflect a downstream response to the reduced insulin production and GSIS in these cells. Of note, the mRNA levels of other transcription factors (e.g., Ngn3, Nkx2.2, and Kir6.2) were unchanged, highlighting the complexity and selectivity of the inactivation of islet-enriched transcription factors (31).

Our results are somewhat at variance with earlier findings (4) that described the beneficial effects of selective overexpression of Irs2^WT in pancreatic β-cells. In these studies, the overexpression of Irs2 in WT mice (denoted rip13^->IRS2) improved pancreatic insulin content by 2.7-fold; however, glucose homeostasis remained unaltered (4). A possible explanation for the different results in the two studies could stem from subtle differences in the strains of mice being used, because different origins of mice (in this case, C57BL6 substrains maintained by different vendors) can cause major differences in how mice respond (33). Furthermore, given that after six backcrosses the Tg mice expressed ∼98% of the genetic background of C57BL/6 mice, the remaining 2% difference could be important. The site of insertion and copy number of the transgene or epigenetic differences of unknown origin could also be contributing factors.

The essential role of IRS2 in maintaining proper β-cell growth and functionality is very well established (1–4). This includes the beneficial effects of overexpressing IRS2^5A on insulin signaling, GTT results, and GSIS in cultured adult mouse and human pancreatic islets (14). Therefore, the phenotype of the Irs2^5A-β mice, which were glucose intolerant and had impaired GSIS, was counter to expectation. However, we could rationalize this phenomenon assuming that, indeed, the overexpression of Irs2^WT and Irs2^5A exerts initial beneficial effects on β-cell function and growth, which are exemplified by the islet hyperplasia observed in the Tg animals as well as by the increased mRNA levels of catalase and sod that are expected to improve cellular reducing power. However, the sustained expression of Irs2^WT and even more so of Irs2^5A might have generated stress signals in the β-cells due to the overuse of the islet growth and secretion machineries driven by the overexpressed IRS2 proteins. This could lead to β-cell dysfunction by targeting MafA, Nkx6.1, Pdx1, and Glut2. Support for this hypothesis is provided by the fact that, indeed, in young Irs2^WT-β mice at 10 days of age the mRNA levels of these transcription factors are still elevated, whereas in Irs2^5A-β mice their levels have already declined. The beneficial effects of IRS2^5A were evident only in even younger mice (5–8 days old). β-Cell proliferation (Ki67 and cyclinD1) and the protein levels of Nkx6.1 and Pdx1 were significantly increased in these young mice, which is in accordance with IRS2^5A being more efficacious than IRS2^WT in promoting insulin signaling in β-cells (9).

The ability of potential promoters of β-cell function to curtail β-cell activity was demonstrated in other model systems as well. For example, the overexpression in β-cells of a constitutively active form of glucokinase, which was expected to promote GSIS, was in fact deleterious to the cells, resulting in β-cell demise and apoptosis (34). In both model systems, β-cells initially responded by enhanced replication. However, at a later stage, β-cell failure caused hyperglycemia and impaired glucose tolerance (Tornovsky-Babeay et al. [34] and the present study). Also, in both models the key β-cell transcription factors MafA and Nkx6.1, but not others, were dramatically reduced. These observations strengthen the hypotheses that overexpression for a limited period of time of proteins expected to be beneficial to β-cell function, such as an active mutated glucokinase or IRS2, indeed promotes β-cell function (14,34). However, allowing the expression of such proteins for too long induces β-cell stress, leading to their demise. The translational implication of this study is the need to identify the proper “time window” when the expression of beneficial proteins is supportive for human β-cells without running the risk of expression that is too long and might end up being detrimental.

Acknowledgments. The authors thank Dr. Sanford Sampson (Bar Ilan University, Ramat Gan, Israel) for insightful discussions.

Funding. This work was supported by JDRF (grant 40-2009-720 to Y.Z.) and the Israel Science Foundation (grant 759/09 to Y.Z.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.I., Y.V., S.B.-H., L.F., and Z.K. conceived and designed the experiments, performed the experiments and statistical analysis, and contributed to the discussion. S.S. performed the experiments and statistical analysis. E.E. conceived and designed the experiments, contributed to the discussion, and reviewed and edited the manuscript. Y.Z. conceived and designed the experiments, contributed to the discussion, and wrote the manuscript. Y.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Keystone Symposium: Diabetes: New Insights into Molecular Mechanisms and Therapeutic Strategies, Kyoto, Japan, 25–29 October 2015.