Small-molecule discovery in the pancreatic beta cell

Abstract

The pancreatic beta cell is the only cell type in the body responsible for insulin secretion, and thus plays a unique role in the control of glucose homeostasis. The loss of beta-cell mass and function plays an important role in both type 1 and type 2 diabetes. Thus, using chemical biology to identify small molecules targeting the beta cell could be an important component to developing future therapeutics for diabetes. This strategy provides an attractive path toward increasing beta-cell numbers in vivo. A regenerative strategy involves enhancing proliferation, differentiation, or neogenesis. On the other hand, protecting beta cells from cell death, or improving maturity and function, could preserve beta-cell mass. Here, we discuss the current state of chemical matter available to study beta-cell regeneration, and how they were discovered.

Induction of beta-cell proliferation

An attractive strategy to regenerating beta-cell mass has been to stimulate these cells to divide. Unlike, say, the liver, where hepatocytes can repopulate the organ, cells resident in pancreatic islets appear to be fixed, leading to the hypothesis that increasing beta-cell division with small molecules could help treat, for example, type 1 diabetes. In recent years, a substantial portion of the efforts to apply chemical biology to beta cells has been in the area of beta-cell proliferation. The question of how beta cells propagate in the pancreas—whether through self-duplication, differentiation from a stem cell population, or some other means—was under debate for many years. A seminal study in 2004 showed that beta cells in mice can themselves replicate [1]. A few years later, an evaluation of islets from juvenile humans demonstrated that replication is the primary means by which beta-cell mass increases from birth to adulthood [2]. The dramatic decrease in Ki67-positive beta cells observed after adolescence, however, suggested that proliferation was not accessible as a natural mechanism of regeneration in adult humans. For example, pregnancy causes a dramatic increase in beta-cell mass in mice [3,4], but in the only such study ever undertaken in humans, beta-cell area only increased 1.4-fold during pregnancy [5]. Further, data suggested that this increase was due to neogenesis from ductal cells in the pancreas, and not proliferation.

Thus, for many years, the prevailing wisdom remained that adult human beta cells could not be induced to divide. However, three studies in 2015 and 2016 pointed to inhibition of the kinase DYRK1A as a new mechanism capable of promoting human beta-cell proliferation (Figure 1) [6–8••]. Interestingly, different approaches to discovery were used in each case. Wang et al. performed a high-throughput screen for inducers of MYC expression in HepG2 cells; the natural product harmine emerged as a strong hit and was found to promote beta-cell proliferation through its inhibition of DYRK1A [8••]. Another effort took a more targeted approach, evaluating an aminopyrazine dual inhibitor of DYRK1A and GSK3β in beta cells [7••]. Finally, a third approach examined the effects of 5-iodotubercidin (5-IT), originally thought to result in phenotypic activity due to inhibiting adenosine kinase [9]. Instead, target identification efforts in human islets showed that, remarkably, 5-ITalso inhibits DYRK1A to promote beta-cell proliferation [6••]. Subsequent work to evaluate combinations of compounds showed that this effect of DYRK1A inhibitors (e.g., harmine, 5-IT, leucettine-41, INDY) in human islets can be enhanced by TGFβ inhibitors [10] or by GLP-1 agonists [11]. In fact, rather than just acting on DYRK1A alone, the activity of this class of compounds was shown to involve inhibition of both DYRK1A and DYRK1B, two closely related isoforms [12]. Importantly, these compounds have been demonstrated to restore glucose homeostasis in several mouse models of diabetes, including partial pancreatectomy and streptozotocin-induced diabetes. The fact that nearly all kinase inhibitors in this area inhibit both kinases allows for a more robust induction of beta-cell proliferation, an effect which could not be achieved by genetic means alone, emphasizing the power of leveraging polypharmacology to promote phenotypes of interest [13].

Figure 1.

Representative small molecules reported to increase beta-cell proliferation. The mechanism of action for each compound is indicated below the name, and details about the biology of these compounds are included in the text.

In order to expand the chemical matter available in islet biology, several groups developed screening platforms for unbiased identification of compounds and new targets for beta-cell proliferation [14–20]. While some screening efforts focused on rat islets, which are more readily available [15], the challenge of this approach is the attrition of many compounds when translating findings to human islets, due to the additional brakes on the cell cycle applied in the latter tissue [21]. As a result, more attention has been paid to screening dissociated human islet cells, either using chemical libraries [14,19,20] or RNAi reagents [17•]. The use of EdU labeling can provide a larger signal-to-noise ratio over time in culture than staining for Ki67, which provides a snapshot of cell division. The use of zebrafish in whole-organism screens for beta-cell proliferation, similar to the cases of developmental and neurobiological screening [22,23], has also yielded good results. Using a FUCCI-expressing [24] zebrafish model for screening revealed compounds affecting the serotonin and retinoic acid pathways, as well as well-known glucocorticoids [18]. More recently, inhibition of salt-inducible kinase (SIK) by the pan-SIK inhibitor HG-9-91-01 resulted in a strong induction of proliferation in zebrafish beta cells, with a present (but milder) increase in human beta cells [16]. Together, these approaches will enhance our ability to expand beta cells as well as our understanding of the mechanisms required.

Open questions remain in this field. Key among them: first, can we find new targets to promote beta-cell division? The TGFβ pathway has been especially scrutinized, as small-molecule inhibition of Alk5 reduces p16^INK4A expression and Smad3 transcriptional activity [25]. Modulation of the E-prostanoid receptors EP3 and EP4, which are regulated by the proliferation-associated FoxM1, with small molecules has also been shown to promote human beta-cell proliferation [26]. Further targets may emerge from screening, but most efforts have converged on these common threads of biology. A second major question is the specificity of these compounds to beta-cell proliferation; in addition to concerns about inducing oncogenesis, recent work questioning the effects on alpha cells and other cell types have raised specificity concerns [27]. Thus, selective delivery of compounds to beta cells, mostly focused on leveraging the very high levels of zinc in these cells, has been proposed as a safer therapeutic approach for the future. A third key question in the field regards whether beta-cell maturity is maintained in these proliferating cells. A mouse model in which insulin expression was reduced by w50% demonstrated a two-fold increase in beta-cell proliferation, accompanied by downregulation of master regulators of beta-cell identity and, somewhat surprisingly, a coordinate increase in glucagon expression [28]. Similarly, an evaluation of Myc overexpression in beta cells showed that the proliferating cells have lost some of their maturity. Further study of these effects, including whether this effect can be reversed after beta-cell expansion, are required to determine the ultimately translational impact of this approach.

Promoting beta-cell survival

Because beta cells are lost in both type 1 and type 2 diabetes, much energy has been put into finding small molecules that can protect beta-cell mass in the first place. Given the relative simplicity of a viability phenotype for high-throughput screening, small-molecule discovery in an effort to promote beta-cell survival has a slightly longer history. In a fashion similar to cancer biology, assay readouts detecting cell numbers (e.g., CellTiter-Glo, MTT) or induction of apoptosis (e.g., caspase activation) have been used. However, the opposite direction is desired here: rather than identify compounds that induce apoptosis in cancer cell lines, screening here aims to detect compounds that enable beta cells to remain impervious to the perturbation. Three primary models have been used for small-molecule suppressor screening. For compounds relevant to type 1 diabetes, inflammatory cytokines (usually a combination of IL-1β + IFN-γ or IL-β + IFN-γ + TNFα) are used. For compounds relevant to type 2 diabetes, free fatty acids, such as sodium palmitate, in the presence of high glucose (typically called “glucolipotoxicity”, or GLT) provide a model of overnutrition and hyperglycemia. Finally, several efforts have aimed to more specifically suppress ER stress in the beta cell, using perturbations such as tunicamycin or thapsigargin; theoretically, this modality could be applicable to either T1D or T2D, as ER stress appears to play an important role in cell death in both cases.

Suppression of inflammatory cytokines

In T1D, the immune system recognizes the beta cell as foreign, resulting in beta-cell death and thus the need for insulin therapy. IL-1β and TNF-α induce NFκB expression and nitric oxide (NO) signaling, increasing ER stress pathways [29]. IFN-γ-induced STAT1 signaling [29,30] works together with NFκB activation to abolish insulin secretion and to induce beta-cell apoptosis. Protein-based receptor antagonists had previously progressed to clinical trials, but have not yet led to approved therapies [31]. Cell replacement therapies, involving the implantation of stem cell-derived beta cells, are showing great promise, but routine clinical implementation is still a ways off, and many approaches will also require immunosuppression. On the other hand, discovery of small-molecule suppressors of cytokine-induced beta-cell apoptosis could lead to the identification of new cellular targets as well as candidate therapeutics (Figure 2a).

Figure 2.

Representative small molecules that protect beta cells from death induced by (a) inflammatory cytokines or (b) glucolipotoxicity and ER stress.

Numerous natural products with anti-inflammatory properties have been shown to protect beta cells from cytokine-induced apoptosis, including resveratrol [32], silymarin [33], and wedelolactone in zebrafish [34]. To systematically identify novel synthetic suppressors of cytokine-based beta-cell death, we performed a high-throughput screen of nearly 400,000 compounds, using the rat INS-1E insulinoma cell line, leading to the discovery of BRD0476 [35–37•], a molecule derived from diversity-oriented synthesis (DOS) [38]. BRD0476 suppressed the activity of the triple cytokine cocktail in cell lines as well as human islets, and mechanism-of-action studies led to the finding that BRD0476 binds the deubiquitinase (DUB) USP9X, leading to a decrease in IFN-γ-induced STAT1 signaling through JAK2 [36]. Study of the interaction between USP9X and JAK2 is ongoing, and highlights the power of a phenotypic screening approach, as this DUB had never been associated with beta cells or even with diabetes in the past. In a more target-focused approach, we also found that inhibition of HDAC3 is able to suppress beta-cell death in the presence of both cytokine treatment and GLT conditions [39,40], likely due to more general anti-inflammatory properties. These inhibitors can also reduce the development of diabetes in nonobese diabetic (NOD) mice, a model of autoimmune disease [41]. The more general involvement of chromatin-modifying enzymes in ER stress has been explored in great depth by the Mandrup–Poulsen lab, who first identified the role of histone deacetylases [42••] and, more recently, lysine demethylases [43] in beta-cell apoptosis induced by cytokines.

Other groups have taken high-content screening approaches to discover small molecules that can protect beta cells [44,45]. For example, Yang et al. used a four-parameter live-cell imaging assay to measure several aspects of apoptosis in the mouse MIN6 insulinoma cell line [44]. Screening the Prestwick Chemical Library, they focused on the hit carbamazepine, a sodium channel inhibitor used to treat epilepsy, and found that the compound’s use-dependent inhibition was responsible for activity. Carbamazepine also inhibited ER signaling, so it is not clear whether this mechanism is specific to cytokine signaling. In other areas, most recently, kinase signaling downstream of IFN-α signaling has been explored, with inhibitors of JAK1 and the JAK-related kinase TYK2 undergoing preclinical evaluation [46,47•].

Suppression of glucolipotoxicity

Of the phenotypes typically explored in beta cells, it seems the largest number of screens have been performed to suppress beta-cell apoptosis induced by ER stress, including that induced by palmitate (Figure 2b). For example, 17,600 compounds were screened to find suppressors of tunicamycin in the mouse βTC6 cell line [48], leading to a series of 2,4-diaminoquinazolines [49]. Importantly, these compounds were also evaluated for their effects on proinsulin misfolding, a vulnerability in beta cells due to the massive insulin protein production needed to have a ready pool for secretion [49]. Other screening efforts identified GSK3β inhibitors and endocannabinoids such as anandamide [50]. And in a direct effort to counteract ER stress pathways in the beta cell, the Maly and Papa labs have developed reagents inhibiting IRE1α kinase/RNase activity (KIRAs) or partial antagonists of its RNase activity (PAIRs) [51–53•]. These approaches have shown promise in vitro and in vivo. Oxidative stress in the beta cell also plays an important role in cell death due to glucolipotoxicity [54,55]. For example, N-acetyl cysteine [56] and other antioxidants such as rosmarinic acid [57] can protect the beta cell from oxidative stress due to glucose and palmitate treatment.

An examination of the literature describing suppressors of ER stress, oxidative stress, and glucolipotoxicity reveals a common theme that inhibition of calcium influx can protect beta cells. Shalev et al. first described that the cardiovascular drug verapamil inhibits TXNIP expression (thioredoxin-interacting protein, important to cellular redox control) and protects beta cells both in culture and in ob/ob mice [58]. Indeed, a later analysis of patients taking verapamil revealed a mild reduction in type 2 diabetes [59], and a randomized double-blind placebo-controlled phase 2 trial of verapamil in new-onset type 1 diabetes found that patients had an improved C-peptide response to a mixed meal and reduced insulin requirements [60]. The same lab later screened ~300,000 compounds, focused on reducing TXNIP expression in islets, and found SRI-37330, which also prevented fatty liver and inhibited gluconeogenesis in mice, though perhaps through a different mechanism [61••]. Related to calcium signaling, NMDA receptors also conduct inward calcium currents, and dextromethorphan, a centrally acting cough suppressant and noncompetitive NMDAR antagonist, was shown to improve islet insulin content, insulin secretion, and viability in mouse and human islets [62,63]•. In another effort to protect beta cells in the presence of palmitate, a group composed of researchers across industry screened 312,000 compounds and identified a chemical series that reduced cytosolic calcium overload by reducing calcium influx [64•]. However, unlike nifedipine, a specific L-type calcium channel inhibitor, this series, including “compound D,” resulted in maintenance of optimal calcium levels in the beta cell. Although the target remains unknown, this work provides further evidence that calcium levels in the beta cell are so important for viability. This realization is leading to more focused efforts to screen directly for, for example, compounds that inhibit ER calcium depletion to stabilize ER-resident proteins [65]. Together, these results suggest that modulating calcium dynamics in the beta cell may result in greater viability and function.

Maintaining beta-cell maturity and function

Coupled with efforts to protect beta-cell viability are studies to find compounds that help keep beta cells mature and functional (i.e., showing appropriate glucose-stimulated insulin secretion). Previous studies have called into question whether beta cells undergo cell death or dedifferentiation during the development of diabetes [66•], and new chemical probes could help shed light on this area (Figure 3a). Thus far, relatively few unbiased efforts have been reported. We screened human PANC-1 ductal cancer cells for compounds that could induce PDX1 expression, and reported the discovery of BRD7552 [67], while the Johnson lab found that carbamazepine, the sodium channel inhibitor that protected beta cells from ER stress as described above, also positive modulates Ins1 and Ins2 expression in mouse islets [68]. Further, an evaluation of the reversal of dedifferentiation, focused on the urocortin 3 gene as a biomarker of beta-cell identity, found that TGFβ inhibitors can restore beta-cell differentiation over extended time in cell culture [69•]. As in the case of beta-cell proliferation, high-content screens in zebrafish for beta-cell function have also yielded tool compounds for follow-up [70,71].

Figure 3.

(a) Small molecules and mechanisms that enhance beta-cell maturity and function. (b) Emerging strategies to target senescent beta cells (Venetoclax) or to enable selective delivery to beta cells, due to their high levels of zinc, either through compound chelation or prodrug strategies.

Ultimately, the primary role of the beta cell is to secrete insulin in response to glucose, so direct screening for insulin secretagogues is of great interest to the beta-cell field. Historically, scarce methods available for high-throughput detection of insulin secretion limited discovery. Nonetheless, a number of mechanisms have been pursued to develop small molecules that can promote insulin secretion, including activation of glucokinase [72], activation of the GPCR free fatty acid receptor 1 (GPR40) [73], and inhibitors of the ATP-dependent potassium channel [74,75]. However, many of these mechanisms remain glucose-independent, so finding new glucose-dependent activities is an important next step in the field. Our group reported a luciferase-based assay, in which Gaussia luciferase was inserted into the C-peptide portion of proinsulin, that is highly correlated with traditional ELISA methods and enables HTS for this phenotype [76••]. Several screens have been performed using this modality; so far inhibitors but no inducers of insulin secretion have been reported [77–79]. The development of small molecule-based GLP-1R agonists, for example, may represent the best chance at success in this area.

Conclusions and outlook

An increasing appreciation for the power of small molecules to provide new insights into beta-cell biology has led to the exploration of additional phenotypes. For example, beta-cell senescence is an emerging area of great interest; recent reports indicate that insulin resistance induces a senescent state [80], and senolysis with Navitoclax or Venetoclax results in improved glucose metabolism, beta-cell function, and even prevention of autoimmune diabetes (Figure 3b) [80,81••]. New compounds in this exciting area will improve our understanding of this process. From a cell-based standpoint, development of sophisticated three-dimensional culture systems, for example from stem cell-derived beta cells [82], will allow the discovery of new compounds with strong physiological relevance. Finally, the beta-cell field will benefit tremendously from integrating new and emerging chemical tools; an outstanding recent review provides a summary of the chemical toolbox available for studying beta-cell function [83••]. Therapeutically, delivery mechanisms that provide beta-cell specificity will improve the safety of mechanisms like inducing proliferation (Figure 3b). Due to the high levels of zinc in the beta cell, for example, zinc-based approaches to either sequester compounds [84] or develop a prodrug system [85,86•] have shown early promise. In conclusion, a greater integration of the principles of beta-cell biology with the tools available from chemical biology (Figure 4) will accelerate the study of this important cell type in diabetes.

Figure 4.

Summary of representative pathways affected in the beta cell. Two key pathways modulated by multiple compounds are highlighted: activation and inhibition of calcium influx into the beta cell, and inhibition of the kinase DYRK1A. VDCC, voltage-dependent calcium channel.