Introduction
The loss of functional beta cell mass is critical in diabetes pathogenesis [
1]. Research aimed at discovering beta cell survival factors has typically been conducted one at a time and has been limited by prior knowledge [
2‐
4]. Glucagon-like peptide 1 (GLP1) is considered a gold standard for beta cell protective factors [
5]. Although local GLP1 increases islet transplant success in animals [
6], clinical evidence to support its efficacy to durably increase human beta cell mass is lacking. Clearly, there is an unmet need to identify more and better beta cell survival factors.
Our efforts to identify novel factors with sustained anti-apoptotic effects led us to mine expression databases and characterise locally acting pro-survival factors in the islet secretome, which includes >200 expressed ligands [
7]. Our initial analyses of candidates revealed glucose-dependent protective roles for Notch, Netrin and Slit [
7‐
9]. However, without a side-by-side comparison, it is impossible to determine the relative merits of each candidate. We recently developed high-throughput, kinetic, live-cell imaging methods that allow the effects of hundreds of factors on multiple cell death parameters to be simultaneously evaluated in dispersed primary islet cells over relatively long periods of time in culture [
10].
Here, we surveyed 206 factors, rationally chosen based on previous bioinformatic analysis [
7], and ranked their effects on islet cell survival under five distinct stress conditions. We found many factors with previously unreported pro-survival effects. Remarkably, each stress condition was associated with a relatively distinct set of protective and deleterious factors, consistent with fundamental mechanistic differences in the cell death pathways. This first systems level analysis has important implications for the development of beta cell protective and regenerative therapies [
11].
Discussion
The goal of the present study was to simultaneously compare the effects of 206 putative beta cell survival factors, under multiple conditions, using a newly developed imaging platform. Our survey pointed to many previously unappreciated factors that may protect islet cells. A principal observation was that each cellular stressor examined appears to require its own unique set of protective factors, and that factors (i.e. SEMA4A) can switch from pro-death to pro-survival. These findings have significant implications for the understanding of the molecular mechanisms controlling beta cell fate and for the development of therapeutic approaches to prevent or treat type 1 or type 2 diabetes at various disease stages.
Factors were selected for testing based on their expression or the expression of their receptors in islets. Harnessing local islet autocrine and/or paracrine survival factor signalling may be an ideal scenario for diabetes prevention or treatment. Many local factors act on self-limiting signalling mechanisms that prevent over-stimulation. Insulin, for example, is a potent and self-limiting islet survival factor and physiological doses of insulin can increase beta cell proliferation [
15,
26,
27]. An unbiased search for other potent survival factors not involved in peripheral metabolism is needed. Another approach would be to exploit pancreatic development factors, including Notch, TGF-beta superfamily, FGFs and bone morphogenic proteins [
8,
28‐
31]. We also cannot overlook the potential importance of distally secreted factors, including adipokines such as adiponectin.
In type 1 diabetes, beta cells may be destroyed by a combination of toxic cytokines and other factors including granzyme, perforin and Fas [
32,
33]. Studies have also implicated ER stress in beta cell death associated with type 1 diabetes [
34]. Thus, it is possible that factors showing protection in both of these conditions may be therapeutic in type 1 diabetes. In type 2 diabetes, excessive fatty acids and ER stress act through partially common pathways to increase beta cell death following the initial compensation phase [
13,
35,
36]. These stresses had in common protection by OLFM1, neuroligin family members and members of the FGF family. Notably, the known beta cell anti-apoptotic factors, insulin and IGF2 [
14,
37], were protective in the context of lipotoxicity more consistently relative to the other stresses. Persistent hyperglycaemia, present in poorly controlled type 1 or type 2 diabetes, induces further beta cell apoptosis [
38‐
40] and factors that are protective under this condition may be candidates for adjunct or second-line treatments.
Some factors promoted survival under only one condition. The most striking of these was the palmitate specific pro-survival effect of semaphorin 4A, which acts through plexin receptors expressed in beta cells [
41]. Clearly, this factor would not be an ideal therapeutic target owing the presence of multiple, concurrent beta cells stresses in vivo. While it might seem counter-intuitive to some that islets would respond to an endogenous factor with significant death, it is possible that the local concentrations of semaphorin 4A are lower than the toxic levels employed in our experiments and that this factor plays an important role in constraining excessive beta cell growth. We expect that functional beta cell mass is controlled in vivo by a robust balance of positive and negative factors [
31].
In addition to the discovery of stress-specific islet cell survival factors, our analyses also enabled the identification of factors with generalisable survival effects across different conditions. Using the simple average ranking analysis of both PI and cell loss data over the first two days, the most broadly effective protective factors appeared to be melanin-concentrating hormone, vasoactive intestinal peptide (VIP) and adiponectin. Melanin-concentrating hormone plays a role in obesity and has been implicated in islet growth [
42]. VIP has known effects on insulin secretion [
43]. Adiponectin is an insulin sensitising adipokine that protects beta cells against multiple stresses as found in our research and that of others [
16‐
19]. Anti-apoptotic effects of adiponectin may not extend to all cell types [
44], suggesting a degree of beta cell specificity. Using the rank product testing to assess the consistency of PI incorporation observations over 3 days, we identified several ‘pan-protective’ factors including neurotrophin 4, ACTH, FGF 12, somatostatin and OLFM1. We believe both the simple average ranking and the rank product testing have value, as they represent different aspects of the data and may be differentially influenced by multiple factors, including peptide stability in storage over the ~3 years these studies took place.
Our findings complement previous studies on the pro-survival signalling mediated by axon guidance factors, netrins and slits [
7,
9]. We observed effects of slits, neuroligins and semaphorins in our parallel comparisons, all factors known to modulate the actin cytoskeleton and play roles in pancreas morphogenesis [
30]. In the present study, we chose to pursue OLFM1, also known as noelin 1 or pancortin. OLFM1 is a modulator of Wnt signalling involved in neuronal development and axon elongation [
45], which interacts with the Nogo A receptor (NgR1) complex [
46] expressed in beta cells [
41]. Our studies demonstrated a dose-dependent effect of OLFM1 on mouse and human beta cell survival in a number of conditions. Thus, despite a lack of adequate statistical power within each condition, our survey/ranking approach identified a novel beta cell survival factor, with conserved effects in mouse and human cells. Although confirming the effects of other interesting factors in human islets is beyond the scope of this project, it is essential to take most of these observations beyond the exploratory stage. We expect our survey will be broadly applicable to human islets, because the factors we surveyed were selected based on both rodent and human expression studies [
7] and because our previous studies suggest broadly similar cell survival pathways in mouse and human islets [
7,
8,
13,
47]. Recent RNA sequencing analysis confirms remarkable similarity between species, with only ~1.5% of genes being unique to either mouse or human beta cells [
41,
48]. However, approximately 6% of genes show species enrichment, including 61 ligands or receptors (ESM Table
4), including the enrichment of
Prlr,
Ghr and
Cntfr in mouse beta cells compared with human beta cells [
41,
48]. Interestingly, SEMA4A is also enriched in mouse beta cells [
41,
48]. One caveat is that the mice used to generate these transcriptomic data sets were 3–6 months old (a time when mice are still growing), whereas the humans donors averaged ~55 years of age, meaning that some of the gene expression differences can be ascribed to relative age differences.
Collectively, our survey of endogenous soluble factors identified multiple hormones/cytokines/growth factors with robust islet cell survival effects under five stress conditions designed to model aspects of type 1 and type 2 diabetes. Perhaps the most important finding was that beta cells were best protected from each specific stress condition by a relatively distinct set of factors. This observation provides important insight into the complexity of beta cell survival signalling pathways and guides therapeutic efforts to protect beta cells.
Acknowledgements
We thank, from the University of British Columbia, X. Hu for assistance with islet isolations, and J. Lee and S. Rangan for assistance with insulin radioimmunoassays.