Exploring the insulin secretory properties of the PGD2-GPR44/DP2 axis in vitro and in a randomized phase-1 trial of type 2 diabetes patients

Abstract

Aims/Hypothesis

GPR44 (DP2, PTGDR2, CRTh2) is the receptor for the pro-inflammatory mediator prostaglandin D2 (PGD2) and it is enriched in human islets. In rodent islets, PGD2 is produced in response to glucose, suggesting that the PGD2-GPR44/DP2 axis may play a role in human islet function during hyperglycemia. Consequently, the aim of this work was to elucidate the insulinotropic role of GPR44 antagonism in vitro in human beta-cells and in type 2 diabetes (T2DM) patients.

Methods

We determined the drive on PGD2 secretion by glucose and IL-1beta, as well as, the impact on insulin secretion by pharmacological GPR44/DP2 antagonism (AZD1981) in human islets and beta-cells in vitro. To test if metabolic control would be improved by antagonizing a hyperglycemia-driven increased PGD2 tone, we performed a proof-of-mechanism study in 20 T2DM patients (average 54 years, HbA1c 9.4%, BMI 31.6 kg/m2). The randomized, double-blind, placebo-controlled cross-over study consisted of two three-day treatment periods (AZD1981 or placebo) separated by a three-day wash-out period. Mixed meal tolerance test (MMTT) and intravenous graded glucose infusion (GGI) was performed at start and end of each treatment period. Assessment of AZD1981 pharmacokinetics, glucose, insulin, C-peptide, glucagon, GLP-1, and PGD2 pathway biomarkers were performed.

Results

We found (1) that PGD2 is produced in human islet in response to high glucose or IL-1beta, but likely by stellate cells rather than endocrine cells; (2) that PGD2 suppresses both glucose and GLP-1 induced insulin secretion in vitro; and (3) that the GPR44/DP2 antagonist (AZD1981) in human beta-cells normalizes insulin secretion. However, AZD1981 had no impact on neither glucose nor incretin dependent insulin secretion in humans (GGI AUC C-peptide 1-2h and MMTT AUC Glucose 0-4h LS mean ratios vs placebo of 0.94 (80% CI of 0.90–0.98, p = 0.12) and 0.99 (90% CI of 0.94–1.05, p = 0.45), despite reaching the expected antagonist exposure.

Conclusion/Interpretation

Pharmacological inhibition of the PGD2-GPR44/DP2 axis has no major impact on the modulation of acute insulin secretion in T2DM patients.

Trial registration

ClinicalTrials.gov NCT02367066.

About the Authors:

Stanko Skrtic

Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliations Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden, Department of Endocrinology, Sahlgrenska University Hospital and Institute of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden

ORCID logo http://orcid.org/0000-0002-1950-5418

Björn Tyrberg

Roles Conceptualization, Data curation, Formal analysis, Methodology, Writing – original draft, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

ORCID logo http://orcid.org/0000-0003-0976-1220

Malin Broberg

Roles Conceptualization, Methodology, Project administration, Supervision, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Hans Ericsson

Roles Conceptualization, Data curation, Formal analysis, Methodology, Visualization, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Volker Schnecke

Roles Data curation, Formal analysis, Software, Visualization, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Magnus Kjaer

Roles Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Marcus Hompesch

Roles Conceptualization, Investigation, Methodology, Supervision, Writing – review & editing

Affiliation: ProSciento Inc, Chula Vista, CA, United States of America

Eva-Marie Andersson

Roles Data curation, Investigation, Methodology, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Erik Ryberg

Roles Data curation, Investigation, Methodology, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Alexander Aivazidis

Roles Data curation, Investigation, Methodology, Visualization, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Charlotte Wennberg Huldt

Roles Data curation, Investigation, Methodology, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Lars Löfgren

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Linda Morrow

Roles Conceptualization, Investigation, Methodology, Project administration, Writing – review & editing

Affiliation: ProSciento Inc, Chula Vista, CA, United States of America

Joanna Parkinson

Roles Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Tina Rydén-Bergsten

Roles Conceptualization, Funding acquisition, Project administration, Resources, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

Elaine Watkins

Roles Data curation, Investigation, Methodology, Project administration

Affiliation: ProSciento Inc, Chula Vista, CA, United States of America

ORCID logo http://orcid.org/0000-0002-6901-6567

Maria Sörhede Winzell

Roles Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Visualization, Writing – review & editing

Affiliation: Cardiovascular, Renal and Metabolism, IMED Biotech Unit, AstraZeneca Gothenburg, Mölndal, Sweden

ORCID logo http://orcid.org/0000-0003-3125-4858

Introduction

Type 2 diabetes mellitus (T2DM) is characterized by a relative loss of functional beta-cell mass leading to insufficient insulin secretion during insulin resistance. It is believed that beta-cell insufficiency is caused by a combination of beta-cell death (apoptosis) and loss-of-function. Beta-cell functional loss may to some degree be explained by dedifferentiation to a cell without the capacity to produce or secrete insulin [1]. Therefore, disease modification in T2DM requires functional beta-cell mass restoration [2] through mechanisms preventing apoptosis or restoring loss-of-function. Although the beta-cell insulin secretory machinery is well understood and pharmacological treatments are available that increase insulin secretion, such as sulphonylureas and GLP-1 receptor agonists, there is a lack of drugs on the market that retain efficiency over time and truly modulate disease mechanisms coupled to insulin insufficiency. Prostaglandin D2 (PGD2) is a pro-inflammatory mediator where one of its receptors, GPR44 (DP2, PTGDR2, CRTh2) so far has been of particular interest to respiratory medicine [3]. Several selective GPR44/DP2 receptor antagonists were taken forward into clinical development as potential anti-asthmatic therapeutics, including AZD1981 developed by AstraZeneca [4–6]. Interestingly, in the human pancreas GPR44/DP2 is enriched in the islets versus the exocrine tissue [7, 8]. In fact, it is among the more highly expressed islet GPCRs according to a recent GPCRome survey [9]. Nevertheless, the function of this receptor in the endocrine pancreas is completely unknown. GPR44/DP2 is a Gαi/o-coupled GPCR with a negative impact on cAMP production [10]. Rodent islets produce prostaglandins (including PGD2) locally [11] that have a negative impact on beta-cell function, including insulin secretion [11–13]. Consequently, locally produced PGD2 via GPR44/DP2 may act as a cAMP dependent break on insulin secretion mediated by glucose metabolic stimulation or other GPCRs, such as the GLP-1 receptor. Therefore, antagonizing the GPR44/DP2 pathway in T2DM to restore insulin secretion via a potentially disease driving mechanism may be an opportunity to develop a disease modifying drug in diabetes.

Here we present the first translational study to investigate the potential gluco-metabolic role of the PGD2-GPR44/DP2 pathway in T2DM, spanning from in vitro studies in human islets and beta-cells to a clinical proof-of-mechanism study in T2DM subjects.

Material and methods

Compound and in vitro pharmacology

The GPR44/DP2 antagonist, AZD1981, 4-(acetylamino)-3-[(4-chlorophenyl)thio]-2-methyl-1H-indole-1-acetic acid [6], previously a clinical candidate for treatment of asthma [5, 14] was investigated for treatment of T2DM. The biochemical and pharmacological properties of AZD1981 was recently described [14]. Here, the in vitro pharmacological properties of AZD1981 was further investigated in the human beta-cell line EndoC-betaH1, which has high expression of GPR44/DP2 (confirmed by qPCR, Fig Ba in S1 File). Briefly, EndoC-betaH1 cells [15] were plated in fibronectin- and extracellular matrix-coated 96-well plates for the in vitro pharmacology analyses and functional assays. Experimental details of the methods are described in the supporting information S1 File.

Insulin and PGD2 secretion in vitro and gene expression analyses

The EC50 for PGD2 inhibition of glucose stimulated insulin secretion (GSIS) was determined using EndoC-betaH1 cells. The cells were incubated with varying concentrations of the stable PGD2 analogue 15(R)-15-methyl PGD2 (Cayman Chemical, USA) at 11.1 mM glucose and insulin secretion was determined. The EC50 value for PGD2 was calculated from the dose response curve. Since EndoC-betaH1 cells has no expression of L-PGDS (confirmed by qPCR, data not shown) and thus cannot produce PGD2, the stable PGD2 analogue was added at EC80 concentration and the effect of the specific GPR44/DP2 antagonist AZD1981 was determined at 11.1 mM glucose or in the presence of a GLP-1 receptor agonist (100 nM exendin-4; Bachem AG, Switzerland).

Human islets (Prodo Laboratories Inc. USA) were used to study PGD2 secretion, GSIS and for qPCR analysis of the gene expression of enzymes in the PGD2 synthesis pathway. The acute effect of PGD2 on GSIS was measured by incubating human islets for 1h in low (2.8mM) or high (11.1mM) glucose, with or without addition of 1 nM 15(R)-15-methyl PGD2. In other experiments, islets were incubated in high glucose (22.2 mM) with and without addition of 20 ng/ml interleukin 1-beta (IL-1beta) and PGD2 secretion into the culture media was determined after 24 h (PGD2 EIA, Cayman Chemical). The expression of genes in the PGD2 synthesis pathway and GPR44/DP2 were further studied by reanalyzing published human islet single cell sequencing data (MTAB-5061, EBI accession number) [16]. Experimental details are available in the supporting information S1 File.

Clinical study design

This study (http://clinicaltrials.gov NCT02367066) was a phase 1, randomised, double-blind, placebo-controlled, multiple-dose, cross-over study conducted at a single centre. The aim was to explore the acute effects of GPR44/DP2 antagonism on both glucose and incretin dependent insulin secretion in T2DM. Hence a cross-over design was chosen with three-day treatment periods upon which steady state of the oral GPR44/DP2 antagonist AZD1981 was to be reached. The study consisted of a run-in, two three-day treatment periods separated by a wash-out period, and follow-up. During each treatment period, the pharmacokinetic (PK) and pharmacodynamic (PD) effects of AZD1981 or placebo were characterized in a mixed meal tolerance test (MMTT) [17] or by a graded glucose infusion (GGI) [18], respectively. A paracetamol test was conducted simultaneously with the MMTT to determine the effects of AZD1981 on gastric emptying.

The primary objectives were defined as change from baseline (day -1 to day 3) for endpoints; a) mixed meal tolerance test (MMTT) area under curve (AUC) (0-4h) for plasma glucose, b) MMTT Cmax and c) graded glucose infusion AUC (1-2h) for plasma C-peptide. The study was powered to detect therapeutically relevant changes for these primary variables. A number of secondary and exploratory variables all linked to insulin secretion or PK/PD were also assessed.

The study was conducted at ProSciento Inc in Chula Vista, CA, USA and the study protocol was approved by the Institutional Review Board Schulman IRB in Research Triangle Park, NC, USA. All subjects gave their written voluntary informed consent before participation. Patients were enrolled between March 2015 and April 2015. See Fig 1 for study flow chart. More study details are available in the supporting information S1 File.

[Figure omitted. See PDF.]

Fig 1. Study flow chart according to CONSORT 2010 guidelines.

The design of a human Proof of Mechanism study to test the Mechanism of Action hypothesis. The phase 1 study was designed as a randomised, double-blind, placebo-controlled, multiple-dose, cross-over study, consisting of a run-in, two treatment periods separated by a wash-out period and a follow-up for each subject. The subjects were T2DM patients on metformin (n = 20).

https://doi.org/10.1371/journal.pone.0208998.g001

The enrolled patients were diagnosed with T2DM and had inadequate glycaemic control on a metformin regimen. Patients were adult males or females, had glycosylated haemoglobin (HbA1c) levels between ≥58.5mmol/mol (7.5%) and ≤97 mmol/mol (11%) at enrolment, were treated with metformin alone for at least the last three months, and had a fasting glucose level from 3–14 mmol/l (54–252 mg/dl). See Table 1 for subject demographics at screening.

[Figure omitted. See PDF.]

Table 1. Demographic characteristics.

Data are presented as mean ± SD.

https://doi.org/10.1371/journal.pone.0208998.t001

Pharmacokinetic and pharmacodynamics sampling in clinical study

Patients were fasting overnight and all oral medications including metformin were dispensed 75 min prior to the first PD assessment (MMTT). During the MMTT (Ensure Plus drink, 473 ml, 100g carbohydrates), blood sampling was conducted at baseline, 15, 30, 60, 90 and 120 min post the mixed meal intake. During the GGI, by 20% glucose intravenous infusion over 2h with a constant rate GLP-1 infusion the last hour, blood sampling was conducted at 10 min intervals. Blood samples were obtained for assessments of AZD1981 PK on days 2, 3, 8, and 9 concurrent with the MMTT and GGI, and for paracetamol PK on days 3 and 9 during the MMTT. More details are available in the supporting information S1 File.

Analytical techniques in clinical study

Bedside plasma glucose during GGI was measured by an YSI-2300 analyzer (YSI Life Science, USA). For MMTT & GGI plasma insulin, C-peptide, glucagon and total GLP-1 were analyzed by the respective ELISA (Mercodia AB, Sweden) and glucose by a colorimetric method. Plasma paracetamol and AZD1981 concentrations were measured by LC-MS (Covance Bioanalytical services, IN, USA). EDTA plasma and urine was collected on day -1, 3, 6 and 9 for analysis of exploratory PGD2 biomarkers performed at AstraZeneca. More details are available in the supporting information S1 File.

Sample size calculations & statistical analysis

The clinical study was powered to detect therapeutically relevant changes for the primary variables. The sample size for the MMTT was based on internal studies that estimated the intra-individual standard deviation (SD) for ln(MMTT AUCGlc 0-4h) to be 0.1. With this estimated SD and an alpha of 0.1, one-sided test and at least 8% difference versus control in MMTT AUCGlc 0-4h, a sample size of 20 evaluable patients was determined to provide at least 80% power. The 8% difference was chosen based on the rational that it is a clinically relevant magnitude as this is the effect observed by DPPIV inhibitors [19]. Similarly, the planned sample size for the GGI was determined based on an estimated SD of ln(GGI AUCC-pep 1-2h) as 0.17. Based on similar power calculations, a sample size of 20 evaluable patients was determined to provide at least 80% power to detect at least a 20% difference versus control for GGI AUCC-pep 1-2h.

Summary statistics are provided for all of the PK and PD variables assessed. The PK evaluable population was used to assess the PK endpoints. PK parameters were derived using actual sampling times and standard non-compartmental methods, and summarized descriptively for both AZD1981 and paracetamol. The three primary PD variables were the MMTT AUCGlc 0-4h, MMTTGlc-Cmax, and the GGI AUCC-pep 1-2h (PD evaluable population). The difference in primary variables between AZD1981 and placebo were analyzed using a mixed-effects model that includes treatment (2 levels), treatment sequence (2 levels), and period (2 levels) as fixed effects, baseline values of the dependent variable as a covariate, and subject-within-sequence as a random effect, were used to analyze the difference in MMTT AUCGlc 0-4h, MMTTGlc-Cmax, and the GGI AUCC-pep 1-2h between AZD1981 and placebo. We assumed that there was no carryover effect in the crossover study as the 3 day wash-out period should have appropriately removed any impact from carryover between treatment periods. The primary variables were log-transformed prior to analysis and then transformed back to linear scale. Significance was established 1-sided at 10%. Treatment estimates were provided as estimated least square mean ratios with a 80% 2-sided CI. All secondary endpoints of the PD parameters and the pre-hepatic ISR (fasting β-cell responsiveness) was analyzed similarly as those described for the primary PD endpoints. No tests for multiplicity testing were performed for the secondary endpoints. The SAS MIXED procedure was used for the statistical analysis.

Pre-hepatic insulin secretion rate (ISR) over time was calculated using deconvolution of peripheral C-peptide concentrations during MMTT using the method of Hovorka et al [20].

Results

Expression of GPR44/DP2 and PGD2 synthesis pathway genes

Since the expression of GPR44/DP2 (PTGDR2) in the pancreas is restricted to the islets with similar protein expression in both healthy and T2DM [8] we decided to further characterize the expression of GPR44/DP2 and genes related to PGD2 synthesis in all islet cell types by mining a single cell sequencing data set [16]. We found that GPR44/DP2 is prominently expressed in beta-cells (Fig 2A), whereas the related receptor DP1 (PTGDR1) and the genes in the PGD2 synthesis pathway mainly are found in pancreatic stellate cells from both healthy and T2DM donors (Fig 2B). GPR44/DP2 mRNA was marginally upregulated in beta-cells from T2DM donors, but several genes in the PGD2 synthesis pathway were quite substantially upregulated in the stellate cells (Fig 2A and 2B). Interestingly, the total number of immune-activated stellate cells (according to classification by Baron et al [21]) was higher in T2DM donors and the expression of the PGD2-related genes were particularly prominent in this subpopulation of stellate cells (Fig 2C–2E). Other endocrine cells types do not express these genes, except GPR44/DP2 that is expressed also in alpha-, delta-, and PP-cells (Fig A in S1 File). Together, the gene expression pattern suggests that PGD2 is produced in the islet by activated stellate cells, particularly in T2DM, and may impact beta-cell function through activation of GPR44/DP2.

[Figure omitted. See PDF.]

Fig 2. Tissue expression of GPR44/DP2 and PGD2 related genes and characterization of pancreatic stellate cells.

PTGDR2 (GPR44/DP2) mRNA expression in human beta-cells (Healthy, n = 197; T2DM, n = 112) (a) and pancreatic stellate cells (Healthy, n = 23; T2DM, n = 31) (b). Heat map of stellate cell signature gene expression sub-classified as quiescent, immune activated (imm) and standard activated (stand) stellate cells according to the description in the main text (c). Quantification of stellate cell subpopulations in healthy and T2D donor human islets (d). Expression of GPR44/DP2 and PGD2 related genes in subclasses of stellate cells (e). Data are presented as mean ± SEM (a-b), total number of cells (d), and individual cells (e). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

https://doi.org/10.1371/journal.pone.0208998.g002

PGD2 secretion by human islets

To confirm activation of the GPR44/DP2 pathway during metabolic stress, human islets were treated with high glucose with or without IL-1beta. GPR44/DP2 was highly expressed in intact human islets (Fig 3A, Fig Ba in S1 File), while DP1 was expressed at very low levels (Fig 3B, Fig Bb in S1 File), as expected based on the findings that only a small pool of cells in islets express DP1 (stellate cells, see above). Interestingly, a significant reduction in GPR44/DP2 expression was observed after incubation with high glucose and IL-1beta, while high glucose in itself had no impact (Fig 3A). PGD2 is produced from arachidonic acid via enzymatic activity of phospholipase A, the cyclooxygenases (PTGS) 1 and 2, and lipocalin-prostaglandin D2 synthase (PTGDS). PTGDS and PTGS1 gene expression were not significantly affected by either high glucose or IL-1beta (Fig 3C and 3D), while PTGS2 was significantly increased by high glucose (35-fold compared to 5.6 mM glucose, p<0.01) and even more prominent by IL-1beta (165-fold compared to control, p<0.0001; Fig 3E).

[Figure omitted. See PDF.]

Fig 3. Human islet gene expression and regulation by high glucose with and without IL-1beta.

Expression of PTGDR2 (GPR44/DP2) (a), PTGDR1 (DP1) (b), PTGDS (L-PGDS) (c), PTGS1 (COX-1) (d) and PTGS2 (COX-2) (e). In vitro human islet PGD2 secretion (f). Data are presented as mean ± SEM from 5 different human islet donors, where each condition was evaluated in quadruplicates. Glc = glucose. *p<0.05, **p<0.01, ****p<0.0001.

https://doi.org/10.1371/journal.pone.0208998.g003

After incubation of human islet for 24 h, degraded PGD2 accumulated in the cell culture media at low glucose (5.6 mM) (Fig 3F). A significant increase in PGD2 production was observed at 22.2 mM glucose with a 2-fold increase (p<0.01), which was further potentiated by IL1-beta with a 10-fold increase in accumulated PGD2 over 24h (p<0.0001; Fig 3F).

Effect of PGD2 and the GPR44/DP2 antagonist AZD1981 in human beta cells

Since the endogenous ligand for GPR44/DP2, PGD2, is unstable, a stable analogue, 15(R)-15-methyl-PGD2, was used to determine the effects on insulin secretion in human beta-cells (EndoC-betaH1). Activation of GPR44/DP2 by 15(R)-15-methyl-PGD2 resulted in a dose dependent inhibition of GSIS (Fig 4A). EC50 was calculated using three methods; DMR- dynamic mass distribution [22], cAMP and GSIS (Fig C in S1 File). The potency of AZD1981 on GPR44/DP2 signaling in EndoC cells were established at a constant concentration (150 pM, EC80) of the stable PGD2 analogue using the same three assays (DMR, cAMP and GSIS). All three assays correlated well and resulted in EC50 values for AZD1981 in the same range (Fig 4B–4D). Additional pharmacological profiling of AZD1981 and GPR44/DP2 in EndoC cells are presented in Figures D and E in S1 File.

[Figure omitted. See PDF.]

Fig 4. Effect of PGD2 on insulin secretion and in vitro pharmacology of AZD1981 in the human beta-cell line EndoC- betaH1.

Dose response curve for the PGD2 analogue (15(R)-15-methyl-PGD2) on insulin secretion, from which the EC50 was determined (a). The in vitro potency of the GPR44/DP2 antagonist AZD1981 was determined using DMR (IC50 = 4.9±1nM) (b), cAMP (IC50 = 21±8nM) (c) and by glucose-stimulated insulin secretion (IC50 = 13±2nM) (d). The assays were performed in the human beta cell line EndoC-betaH1 with addition of 150 pM 15(R)-15-methyl-PGD2. Insulin secretion in EndoC-betaH1 cells was determined with and without addition of 150pM 15(R)-15-methyl-PGD2 at 11.1 mM glucose and 100nM exendin-4 (GLP-1 receptor agonist). 100nM of AZD1981 was added to restore GSIS. The effect of PGD2 on human islet insulin secretion was studied in islets from three donors where each incubation condition was run with 5 islets in 6 replicates (f). Data are presented as mean ± SEM.*p<0.05, **p<0.01***p<0.001.

https://doi.org/10.1371/journal.pone.0208998.g004

Addition of 150 pM 15(R)-15-methyl-PGD2 to EndoC cells potently inhibited both GSIS as well as exendin-4-potentiated insulin secretion (Fig 3E). PGD2 inhibition of insulin secretion induced by both glucose and exendin-4, was completely restored by addition of 100 nM AZD1981 (Fig 4E). The inhibitory effect of 15(R)-15-methyl-PGD2 on GSIS was confirmed in human islets (Fig 4F). The GPR44/DP2 antagonist used in this experiment was however different from AZD1981but with similar potency in all three of the in vitro potency assays described for AZD1981 (data not shown).

We demonstrate that GPR44/DP2 is activated in human beta cells by stress induced either by high glucose or high glucose in combination with IL-1beta, resulting in dysregulated glucose and GLP-1 stimulated insulin secretion in vitro, at least partly through production of PGD2 in stellate cells. Based on that we formulated a hypothesis that stellate cells are activated by glucose and cytokines, producing PGD2, which in turn acts on GPR44/DP2 on the beta-cells, reducing intracellular cAMP levels leading to inhibition of GSIS (Fig 5). To test the hypothesis that the PGD2-GPR44/DP2 axis is activated in T2DM and can be a therapeutically relevant target in treatment of T2DM, a mechanistic clinical study was designed and executed using the oral GPR44/DP2 antagonist AZD1981. The study’s aim was to explore the acute effects of GPR44/DP2 antagonism on both glucose and incretin dependent insulin secretion in T2DM subjects. Hence, to investigate the insulinotropic action in a physiological context an oral challenge was given through a 4h MMTT [17] and also an intravenous glucose challenge, i.e. a graded glucose (GGI) ranging from a normoglycemic to hyperglycaemic levels [18].

[Figure omitted. See PDF.]

Fig 5. Hypothesis for the endogenous PGD2-GPR44/DP2 axis mechanism of action in islets.

Glucose and IL-1beta induced PTGS2 (COX-2) expression in stellate cells which stimulates the production of PGD2 synthesis. PGD2 is secreted and may act in a paracrine fashion on GPR44/DP2 on beta-cells, inhibiting cAMP and thereby reducing GSIS.

https://doi.org/10.1371/journal.pone.0208998.g005

The study outline is described in more detail both the Materials and Method section and illustrated in Fig 1 as well as in the supporting information S1 and S3 Files.

Effects on glucose, insulin and C-peptide after MMTT and GGI in T2DM patients

Treatment with the GPR44/DP2 antagonist AZD1981 at steady state exposure (AZD1981 at 100 mg BID for three days) did not exert any clinically significant insulinotropic effect in T2DM patients on metformin (Fig 6 and Table 2) for any of the co-primary variables in the study. Neither following the insulinotropic challenge in a physiological context during an oral challenge eg a 4h MMTT (MMTT AUCGlc 0-4h, p = 0.12 and MMTTGlc-Cmax, p = 0.06) nor after the intravenous glucose challenge i.e. the graded glucose infusion from a normoglycemic to hyperglycaemic range (GGI AUCC-pep 1-2h p = 0.41) In addition a number of linked secondary variables were also analysed in a similar manner with a mixed effect model to determine any treatment effect of AZD1981 but without any multiplicity test correction. In order to exclude confounding effects by hepatic insulin clearance, C-peptide secretion was evaluated with three parameters, Cmax, AUC(0-4h), and deconvoluted prehepatic insulin secretion rate (ISR) according to Hovorka [20]. Notably, the hypothesized insulinotropic effect was not captured by ISR in any subject or time point after neither the MMTT nor the GGI tests. Further none of the secondary variables did show any significant differences from placebo apart from MMTT AUC C-peptide 0-4h and GGI AUC Glucose0-1h which showed mechanistically contradictory borderline significant effects of p = 0.04 and 0.045 respectively (Fig F in S1 File and Table 2).

[Figure omitted. See PDF.]

Fig 6. Mixed Meal Tolerance Test in T2DM patients.

Glucose assessments at baseline (a), and after 3 days treatment (b) or Insulin Secretion Rate (ISR) at baseline (c) and after 3 days treatment (d). Circles connected with dotted lines represent placebo and squares connected with full lines represent AZD1981. Data are presented as mean ± SD.

https://doi.org/10.1371/journal.pone.0208998.g006

[Figure omitted. See PDF.]

Table 2. Results of mixed effects model for treatment differences in pharmacodynamic parameters before and after GPR44/DP2 antagonism treatment.

https://doi.org/10.1371/journal.pone.0208998.t002

Effects on glucagon, GLP1, paracetamol, PK/PD, PGD2 biomarkers and safety in T2DM patients

To further delineate the mode of action and impact on the other major hormonal islet axes, insulin, glucagon and total GLP-1 were assessed during the MMTT and GGI challenges. However, as outlined in Table 2, AZD1981 had no significant effect on glucagon secretion or total GLP-1. To fully understand the impact on oral glucose disposal the change in oral paracetamol PK was used to assess the impact on gastric emptying concomitantly with the MMTT. In the paracetamol test, the 90% confidence intervals (CI) of the geometric least squares (LS) mean ratios (AZD1981 to placebo) for Cmax and AUC(0-t) were fully contained within the range of 0.80–1.25 (0.90–1.17 for Cmax, 0.91–1.20 for AUC(0-t)) and thus the results demonstrated that AZD1981 had no significant effect on gastric emptying in T2DM patients (Fig G in S1 File).

A PK/PD analysis was conducted between AZD1981 exposure (AUC) during either the MMTT or the GGI tests and all of the efficacy endpoints listed in Table 2. However, the PK/PD analysis indicated that no relationship existed between AZD1981 exposure and the MMTT or GGI efficacy variables (Fig H in S1 File).

We explored if we could identify responder sub-populations with high PGD2 tone by determining circulating surrogate biomarkers of local PGD2 and to correlate that with PD response. However, it was not possible to establish a correlation between efficacy and such biomarkers (Fig I in S1 File).

Finally, there were no AZD1981 safety and tolerability findings that could have impacted the PD effects (Tables A and B in S1 File). All of the adverse events (AEs) reported were mild, none resulted in withdrawal from the study, and there were no serious adverse events (SAEs). The incidence of AEs was similar in the AZD1981 (n = 5, 25%) and placebo (n = 6, 30%) treatments. Two (2) of the AEs reported following AZD1981 treatment (diarrhoea) were considered related to the investigational product. There were no unusual or clinically significant vital sign changes observed, and no significant differences between treatments. No clinically significant abnormal findings were detected by ECG or physical examination.

Discussion

To our knowledge, this is the first study elucidating the effect of PGD2 on human islet function, although it has been known for almost 30 years that PGE2 inhibits glucose-stimulated insulin secretion [12]. Since then, PGE2 control of both beta-cell function and mass has been extensively studied and its receptor EP3 is a potential drug target for treatment of diabetes (recently reviewed by Carboneau et al. [23]). Taking that into account, the unknown impact of PGD2 on GSIS and the finding that GPR44/DP2 is highly expressed in human beta-cells [7, 8], motivated us to explore the role of PGD2 in human beta-cell function.

We hypothesized that GPR44/DP2 modulates beta-cell function through an impact on the insulin secretory machinery, based on tissue expression of the GPR44/DP2 pathway genes and robust functional in vitro human islet/beta-cell data. Surprisingly, we demonstrate that GPR44/DP2 antagonism do not acutely improve insulin secretion in T2DM patients irrespective of the endogenous incretin response, i.e. neither during MMTT nor GGI. Furthermore, the GPR44/DP2 antagonist AZD1981 had no significant effect on the paracetamol profile and thus gastric emptying in T2DM patients, which could have confounded the interpretation of MMTT data. At a dose of 200 mg per day for three days, AZD1981 was safe and well tolerated in T2DM patients treated with metformin. The AZD1981 PK profile was adequate and in line with previous reports, and thus, allowed for a clinically and therapeutically relevant testing of the hypothesis. Thereby, we cannot conclude that the PGD2-GPR44/DP2 axis is a major therapeutically relevant negative modulator of insulin secretion in T2DM patients.

The patient population recruited to the study was adequately representative of a therapeutically relevant target population e.g. patients with T2DM who have failed to achieve glycemic control on first line standard of care (oral metformin with a balanced gender, ethnic, BMI, and HbA1c distribution. However, it is noteworthy that the study population presented with comparatively long-standing disease as indicated by 9.5 years disease duration and HbA1c of 79 mmol/mol. In line with that, there are several typical pathophysiological features that accompany extended T2DM duration such as the observed glycemic profile during MMTT and the comparatively delayed gastric emptying as demonstrated by the delayed Cmax observed in our study (See Fig F in S1 File and, for comparison, see Fig 2 in [24]). A potential limitation in the clinical study might be the chosen insulinotropic detection level of 8% in MMTT AUCGlc 0-4h for which the study was powered (N = 20 evaluable subjects for 80% power at the 5% level). The 8%change in MMTT AUCGlc 0-4h is the observed GSIS effect by DPPIV inhibitors [19]. It can be debated if this detection level was set too high given that we did not observe any effects. However, given that DPPIV inhibitors have comparatively modest chronic glycemic efficacy in T2DM, averaging 0.3–0.5% HbA1c reduction, when detecting and validating a novel insulin secretory regulatory pathway of broad clinical relevance and interest we found it hard to argue for a lower efficacy threshold.

A protocol deviation with GLP1 entrapment due to a microfilter incompatibility led to a less than planned GLP1 exposure during the GGI PD test and affected the ability to assess potential effects of GPR44/DP2 antagonism on GLP1 responsiveness during the GGI test. However, it was still possible to draw overall conclusions on GLP1 responsiveness due to the complete lack of effect on ISR after both the MMTT and GGI. The rationale for this is that we observed the expected increase in endogenous GLP1 during the MMTT but it did not result in increased insulinotropic action. This conclusion is further supported by the observation that the endogenous total GLP1 concentration after the MMTT was only about one third lower than the expected GLP1 exposure during the GGI/GLP1 PD test, and furthermore, by the knowledge that the GLP1-insulin secretion dose response relationship in T2DM is linear [25].

The underlying hypothesis for the clinical study, i.e. the insulinotropic action of GPR44/DP2 antagonism in vitro, was based on our data demonstrating that AZD1981 restores PGD2-inhibited GSIS by potentiation of intracellular cAMP. That is, we observed that a PGD2-analogue acutely reduces GSIS in the human beta-cell line EndoC-betaH1 through inhibition of cAMP. Moreover, high glucose with or without IL-1beta induced significant PGD2 production in human islets, as previously seen in rodent islets [11]. Therefore, we hypothesized that the PGD2 tone in islets is high in diabetics and that GPR44/DP2 antagonism consequently would improve the insulin response to glucose in humans with T2DM. However, irrespective of endogenous GLP1 secretion after both MMTT and GGI, AZD1981 for 3 days did not have any significant acute insulinotropic efficacy, and we concluded that the in vitro data did not translate to a therapeutically relevant in vivo disease setting. The possible reasons underlying the lack of translatability are speculative, but the primary hypotheses are [1] an absence of a PGD2 tone in T2DM islets in vivo and/or [2] few/absent GPR44/DP2 receptors in T2DM islets. We analyzed metabolites for PGD2 both in plasma and in urine but did not find any correlation between insulinotropic effects and PGD2 metabolite levels. Unfortunately, PGD2 has a very short half-life [26] and it is produced in multiple cell types and tissues [27]. Since islets comprise only a very small portion of the total cell pool that produce PGD2, it is likely that global PGD2 metabolite concentrations do not accurately reflect the PGD2 production in islets. The lack of specific islet PDG2 metabolites makes it impossible to conclude if a low islet PGD2 tone was the reason for lack of efficacy. Obviously, it was not possible to measure GPR44/DP2 expression in the islets of the trial subjects. However, gene expression data from isolated islets do not support downregulation of GPR44/DP2 in diabetes (see Fig 2) and previously published data comparing GPR44/DP2 protein expression demonstrate no distinct difference between healthy and T2DM islets [8]. GPCR activity is, however, dependent on protein expression on the cell surface, and it is possible that long term diabetes leads to reduced GPR44/DP2 activity via receptor internalization or other mechanisms, which could have contributed to the poor in vitro-in vivo translation. In contrast to traditional insulin secretagogues, which have all demonstrated excellent in vitro-in vivo translation, the GPR44/DP2 mechanism depends on antagonism of a presumed negative feed-back loop. If this feed-back loop is not present in diabetics the antagonistic action of AZD1981 would have no acute effect, which seems to be the case.

This study demonstrates the complexity of islet function in vivo, where many different GPCRs work in concert to modulate insulin secretion to match the demand and achieve glucose control. The role of prostaglandins for beta-cell function was recently reviewed [23], further illustrating the complexity of prostaglandin signaling through different pathways that modulate beta-cell function, both stimulating and inhibiting insulin secretion. It has long been known that the enzymes needed for prostaglandin synthesis are expressed in rodent islets and that prostaglandins are produced in response to high glucose and IL-1beta [12, 13]. Also in human islets, activation of the prostaglandin synthesis pathway has been observed by increased expression of COX-2 after treatment with cytokines and high glucose [28, 29]. We confirm this mechanism in intact human islet preparations. However, by mining a human islet single cell sequencing data set [16] we found, to our surprise, that the prostaglandin machinery in human islets is not expressed in endocrine cells but in the stellate cells. This is a novel finding and is not in line with the current consensus for prostaglandin synthesis in islets [28]. To our knowledge the expression pattern has so far only been studied in whole islet preparations and this is the first-time single cell sequencing of human islet has been utilized, allowing analysis of the expression of genes at the cell level. The pancreatic stellate cells are few and quiescent under normal conditions and they surround the acinar, ductal and peri-islet areas [30]. The single cell sequencing data clearly demonstrate that stellate cells are present in human islet preparations [16]. During development of pancreatitis and pancreatic cancer, the stellate cells activate into a myofibroblast phenotype, deteriorating the tissue by rapid growth and secretion of cytokines [31, 32]. Recent studies in animal models have demonstrated activation of stellate cells during development of islet fibrosis and beta-cell dysfunction in T2DM [33, 34], which we confirm, albeit with a small number of cells. The role of stellate cells in the development of human T2DM is currently not known and it could only be speculated that inhibition of PGD2 signaling through GPR44/DP2 may require longer treatment periods than performed in the present study to alter beta-cell function and have an impact on glycemic control.

Supporting information

[Figure omitted. See PDF.]

S1 File.

Supplementary material vs 11.docx: Fig A. Expression of PGD2-related genes in all human islet endocrine cell types. GPR44 (a), DP1 (b), L-PGDS (c), COX1 (d), and COX2 (e). Alpha-cell, n = 886. Beta-cell, n = 309. Delta-cells, n = 114. PP-cells, n = 197. Fig B. Gene expression of PGD2 receptors in rodent and human islets and beta cells. (a) The mRNA expression of GPR44 in rodent islet and beta cell lines were significantly lower compared to human islet and beta cells. (b) mRNA expression of GPR44 and DP1 in rat and human primary islets shows almost undetectable levels of DP1 in human islets. The data were normalized against the expression of the housekeeping gene HPRT for each sample. Fig C. PGD2 potently activates signalling in human beta cells. The natural ligand for GPR44 is PGD2. Since PGD2 is unstable in water solution, 15(R)-15-methyl PGD2 was used in all experiments. The potency for 15(R)-15-methyl PGD2 at the GPR44 receptor was determined using DMR, cAMP and insulin assays. The agonist showed strong potency and good correlation between the assays. A) The DMR signal (Epic) for the PGD2 effect on human beta cells (Endo-βH1) (EC50 = 0.8x10-10M), B) the intracellular cAMP levels (EC50 = 0.6x10-10M) and C) effects on glucose-stimulated insulin secretion (EC50 = 1.2x10-10M) were determined. The curves illustrate one representative experiment for each method. Number of data points/assay were for cAMP and insulin assays 6-plicates and 3-plicates for DMR (Epic). Values shown are mean±SEM. Fig D. Signaling pathway for GPR44 in human beta cells. GPCRs couple to different G-proteins and it is known that GPR44 couple to the Gαi pathway, inhibiting cAMP levels. To establish the signaling pathway in human beta cells, pertussis toxin (PTX) was added to inhibit signaling through Gαi. Using the DMR assay, PGD2 induced a potent effect on the human EndoC-beta cells, while addition of PTX completely blocked the total cell response. This indicate that the major signaling pathway for GPR44 in human beta cells occur through the Gαi pathway. Fig E. AZD1981 dose-response in human beta cell-line (Endo-βH1). AZD1981 in different concentrations induced no significant agonist response in human beta cells while 15(R)-15-methyl-PGD2, which was used as positive control, produced a dose response curve, demonstrating the presence of active GPR44 receptors on the cells. Fig F. Graded Glucose Infusion in T2DM patients. C-peptide assessments at baseline (A), after 3 days treatment (B) or Insulin Secretion Rate (ISR) at baseline (C) and after 3 days treatment. Data are presented as mean ± SD. Fig G. Mixed meal test in T2DM patients with plasma paracetamol assessments after 3 days treatment of AZD1981 or placebo. Data are presented as mean ± SD. Fig H. PK/PD analysis of AZD1981 exposure vs selected PD variable after 3 days treatment. Mixed Meal Tolerance Test and AUC Glucose (0-4h) (A) and for Graded Glucose Infusion and AUC C-peptide (0-1h) (B) in T2DM patients. Data are presented as linear regression. Fig I. Analysis of exploratory surrogate biomarkers of PGD2 vs AZD1981 PD response eg MMTT AUC Glc (0-4h) after 3 days treatment. Plasma 11-beta-PGF2-alpha (A) Urine 11-beta-PGF2-alpha /creatinine (B) Urine tetranor-PGDM / creatinine (C) and plasma L-PGDS (D) in T2DM patients. Data are presented as linear regression. Table A. Safety evaluation: Adverse events / Serious Adverse events Table. B. Safety evaluation: Clinical chemistry. Table C. Maximum Plasma AZD1981 Concentration at Steady-State, Css,Max

https://doi.org/10.1371/journal.pone.0208998.s001

(DOCX)

S2 File. CONSORT 2010 checklist.

https://doi.org/10.1371/journal.pone.0208998.s002

(DOC)

S3 File. D6420C00001 clinical study protocol final.

https://doi.org/10.1371/journal.pone.0208998.s003

(PDF)

Acknowledgments

The authors would like to thank Anne-Christine Carlsson, Bengt Kull and Gun-Britt Forsberg for excellent work on the exploratory biomarkers.

Citation: Skrtic S, Tyrberg B, Broberg M, Ericsson H, Schnecke V, Kjaer M, et al. (2018) Exploring the insulin secretory properties of the PGD2-GPR44/DP2 axis in vitro and in a randomized phase-1 trial of type 2 diabetes patients. PLoS ONE 13(12): e0208998. https://doi.org/10.1371/journal.pone.0208998

References

1. White MG, Shaw JA, Taylor R. Type 2 Diabetes: The Pathologic Basis of Reversible beta-Cell Dysfunction. Diabetes Care. 2016;39(11):2080–8. pmid:27926891

2. Song I, Muller C, Louw J, Bouwens L. Regulating the beta cell mass as a strategy for type-2 diabetes treatment. Current drug targets. 2015;16(5):516–24. pmid:25654737

3. O'Byrne PM, Naji N, Gauvreau GM. Severe asthma: future treatments. Clinical and experimental allergy: journal of the British Society for Allergy and Clinical Immunology. 2012;42(5):706–11.

4. Norman P. Update on the status of DP2 receptor antagonists; from proof of concept through clinical failures to promising new drugs. Expert Opin Investig Drugs. 2013;23(1):55–66. pmid:24073896

5. Kuna P, Bjermer L, Tornling G. Two Phase II randomized trials on the CRTh2 antagonist AZD1981 in adults with asthma. Drug design, development and therapy. 2016;10:2759–70. pmid:27621597

6. Luker T, Bonnert R, Brough S, Cook AR, Dickinson MR, Dougall I, et al. Substituted indole-1-acetic acids as potent and selective CRTh2 antagonists-discovery of AZD1981. Bioorg Med Chem Lett. 2011;21(21):6288–92. pmid:21944852

7. Hellström-Lindahl E, Danielsson A, Ponten F, Czernichow P, Korsgren O, Johansson L, et al. GPR44 is a pancreatic protein restricted to the human beta cell. Acta Diabetol. 2016;53(3):413–21. pmid:26467464

8. Lindskog C, Korsgren O, Ponten F, Eriksson JW, Johansson L, Danielsson A. Novel pancreatic beta cell-specific proteins: antibody-based proteomics for identification of new biomarker candidates. Journal of proteomics. 2012;75(9):2611–20. pmid:22465717

9. Amisten S, Salehi A, Rorsman P, Jones PM, Persaud SJ. An atlas and functional analysis of G-protein coupled receptors in human islets of Langerhans. Pharmacology & therapeutics. 2013;139(3):359–91.

10. Hata AN, Zent R, Breyer MD, Breyer RM. Expression and molecular pharmacology of the mouse CRTH2 receptor. The Journal of pharmacology and experimental therapeutics. 2003;306(2):463–70. pmid:12721327

11. Evans MH, Pace CS, Clements RS Jr. Endogenous prostaglandin synthesis and glucose-induced insulin secretion from the adult rat pancreatic islet. Diabetes. 1983;32(6):509–15. pmid:6354779

12. Robertson RP. Eicosanoids as pluripotential modulators of pancreatic islet function. Diabetes. 1988;37(4):367–70. pmid:2897939

13. Kimple ME, Keller MP, Rabaglia MR, Pasker RL, Neuman JC, Truchan NA, et al. Prostaglandin E2 receptor, EP3, is induced in diabetic islets and negatively regulates glucose- and hormone-stimulated insulin secretion. Diabetes. 2013;62(6):1904–12. pmid:23349487

14. Schmidt JA, Bell FM, Akam E, Marshall C, Dainty IA, Heinemann A, et al. Biochemical and pharmacological characterization of AZD1981, an orally available selective DP2 antagonist in clinical development for asthma. Br J Pharmacol. 2013;168(7):1626–38. pmid:23146091

15. Ravassard P, Hazhouz Y, Pechberty S, Bricout-Neveu E, Armanet M, Czernichow P, et al. A genetically engineered human pancreatic β cell line exhibiting glucose-inducible insulin secretion. J Clin Invest. 2011;121(9):3589–97. pmid:21865645

16. Segerstolpe A, Palasantza A, Eliasson P, Andersson EM, Andreasson AC, Sun X, et al. Single-Cell Transcriptome Profiling of Human Pancreatic Islets in Health and Type 2 Diabetes. Cell Metab. 2016;24(4):593–607. pmid:27667667

17. Zaccaria M, De Palo E, Zago E, Sicolo N, Erle G, Federspil G. Metabolic and endocrine responses to a standard mixed meal. A physiologic study. Acta Diabetol Lat. 1979;16(1):45–53. pmid:452819

18. Byrne MM, Sturis J, Polonsky KS. Insulin secretion and clearance during low-dose graded glucose infusion. Am J Physiol. 1995;268(1 Pt 1):E21–7. pmid:7840177

19. Solis-Herrera C, Triplitt C, Garduno-Garcia JeJ, Adams J, DeFronzo RA, Cersosimo E. Mechanisms of glucose lowering of dipeptidyl peptidase-4 inhibitor sitagliptin when used alone or with metformin in type 2 diabetes: a double-tracer study. Diabetes Care. 2013;36(9):2756–62. pmid:23579178

20. Hovorka R, Chassin L, Luzio SD, Playle R, Owens DR. Pancreatic beta-cell responsiveness during meal tolerance test: model assessment in normal subjects and subjects with newly diagnosed noninsulin-dependent diabetes mellitus. J Clin Endocrinol Metab. 1998;83(3):744–50. pmid:9506719

21. Baron M, Veres A, Wolock SL, Faust AL, Gaujoux R, Vetere A, et al. A Single-Cell Transcriptomic Map of the Human and Mouse Pancreas Reveals Inter- and Intra-cell Population Structure. Cell systems. 2016;3(4):346–60 e4. pmid:27667365

22. Schröder R, Janssen N, Schmidt J, Kebig A, Merten N, Hennen S, et al. Deconvolution of complex G protein-coupled receptor signaling in live cells using dynamic mass redistribution measurements. Nat Biotechnol. 2010;28(9):943–9. pmid:20711173

23. Carboneau BA, Breyer RM, Gannon M. Regulation of pancreatic β-cell function and mass dynamics by prostaglandin signaling. J Cell Commun Signal. 2017.

24. Monnier L, Colette C, Dunseath GJ, Owens DR. The loss of postprandial glycemic control precedes stepwise deterioration of fasting with worsening diabetes. Diabetes Care. 2007;30(2):263–9. pmid:17259492

25. Kjems LL, Holst JJ, Vølund A, Madsbad S. The influence of GLP-1 on glucose-stimulated insulin secretion: effects on beta-cell sensitivity in type 2 and nondiabetic subjects. Diabetes. 2003;52(2):380–6. pmid:12540611

26. Schuligoi R, Schmidt R, Geisslinger G, Kollroser M, Peskar BA, Heinemann A. PGD2 metabolism in plasma: kinetics and relationship with bioactivity on DP1 and CRTH2 receptors. Biochem Pharmacol. 2007;74(1):107–17. pmid:17452035

27. Murata T, Maehara T. Discovery of anti-inflammatory role of prostaglandin D2. The Journal of veterinary medical science. 2016;78(11):1643–7. pmid:27498997

28. Parazzoli S, Harmon JS, Vallerie SN, Zhang T, Zhou H, Robertson RP. Cyclooxygenase-2, not microsomal prostaglandin E synthase-1, is the mechanism for interleukin-1beta-induced prostaglandin E2 production and inhibition of insulin secretion in pancreatic islets. J Biol Chem. 2012;287(38):32246–53. pmid:22822059

29. Shanmugam N, Todorov IT, Nair I, Omori K, Reddy MA, Natarajan R. Increased expression of cyclooxygenase-2 in human pancreatic islets treated with high glucose or ligands of the advanced glycation endproduct-specific receptor (AGER), and in islets from diabetic mice. Diabetologia. 2006;49(1):100–7. pmid:16341840

30. Omary MB, Lugea A, Lowe AW, Pandol SJ. The pancreatic stellate cell: a star on the rise in pancreatic diseases. J Clin Invest. 2007;117(1):50–9. pmid:17200706

31. Apte MV, Wilson JS. Dangerous liaisons: pancreatic stellate cells and pancreatic cancer cells. J Gastroenterol Hepatol. 2012;27 Suppl 2:69–74.

32. Pandol S, Gukovskaya A, Edderkaoui M, Edderkoui M, Dawson D, Eibl G, et al. Epidemiology, risk factors, and the promotion of pancreatic cancer: role of the stellate cell. J Gastroenterol Hepatol. 2012;27 Suppl 2:127–34.

33. Ryall CL, Viloria K, Lhaf F, Walker AJ, King A, Jones P, et al. Novel role for matricellular proteins in the regulation of islet β cell survival: the effect of SPARC on survival, proliferation, and signaling. J Biol Chem. 2014;289(44):30614–24. pmid:25204658

34. Li FF, Chen BJ, Li W, Li L, Zha M, Zhou S, et al. Islet Stellate Cells Isolated from Fibrotic Islet of Goto-Kakizaki Rats Affect Biological Behavior of Beta-Cell. J Diabetes Res. 2016;2016:6924593. pmid:26697502