Abstract
Functional genetics has identified drug targets for metabolic disorders. Opioid use impacts metabolic homeostasis, although mechanisms remain elusive. Here, we explore the OPRD1 gene (encoding delta opioid receptor, DOP) to understand its impact on type 2 diabetes. Large-scale sequencing of OPRD1 and in vitro analysis reveal that loss-of-function variants are associated with higher adiposity and lower hyperglycemia risk, whereas gain-of-function variants are associated with lower adiposity and higher type 2 diabetes risk. These findings align with studies of opium addicts. OPRD1 is expressed in human islets and beta cells, with decreased expression under type 2 diabetes conditions. DOP inhibition by an antagonist enhances insulin secretion from human beta cells and islets. RNA-sequencing identifies pathways regulated by DOP antagonism, including nerve growth factor, circadian clock, and nuclear receptor pathways. Our study highlights DOP as a key player between opioids and metabolic homeostasis, suggesting its potential as a therapeutic target for type 2 diabetes.
Similar content being viewed by others
Introduction
The failure to elucidate the complex pathophysiology of type 2 diabetes and obesity has frustrated current efforts to develop improved therapeutic and preventive strategies. The health consequences are important given the rapid rise in diabetes prevalence (currently >500 million patients worldwide) and the failure of 40% of patients to achieve glucose control targets with current medications1. Type 2 diabetes and obesity are multifactorial genetic disorders with an estimated heritability ranging between 40 and 70%2,3. So far, genome-wide association studies have identified several hundred susceptibility genes for type 2 diabetes and body mass index (BMI), which were respectively found to predominantly impact pancreatic islets, especially pancreatic beta cells, and the central nervous system4,5.
The identification and careful characterization of rare variants causing metabolic disorders has been proven to be a fantastic tool for the identification of new drug targets in the field of diabetes and other metabolic diseases6,7,8. Among the privileged drug targets, G-protein coupled receptors (GPCRs) represent the largest family of proteins targeted by approved drugs9.
Our present study stemmed from the observed correlation between the opioid use and metabolic trait abnormalities. There are various sources of opioids: natural (e.g., opium), semi-synthetic (e.g., heroin), fully synthetic (e.g., fentanyl), and endogenous (e.g., enkephalins, endorphins)10. In humans, a preponderance of studies has reported lower BMI among opioid users11,12. In the same line, most studies have shown that opioids decrease food intake in mice13. Paradoxically, in the late 70s, a first study reported that enkephalins (i.e., endogenous opioids) might be a cause of type 2 diabetes in humans14. Since then, several studies have shown a significant association between opium and heroin addictions on altered metabolic traits including glycated hemoglobin A1c (HbA1c)15 and plasma glucose16,17, although these results were not always consistent11.
Delta opioid receptor (DOP), encoded by OPRD1, is one of the three classical opioid receptors that are GPCRs coupled to inhibitory G proteins (Gαi). Opioid receptors are mostly expressed in the central nervous system18. In contrast with the other opioid receptor genes (i.e., OPRM1 and OPRK1), OPRD1 is also expressed in metabolic tissues, especially in pancreas18.
Here, we perform functional genetics of OPRD1 in cases with type 2 diabetes and/or obesity and controls to decipher the putative role of DOP in type 2 diabetes and obesity. Our study highlights DOP as a major link between opioids and metabolic homeostasis in humans, playing a direct role in the physiology of insulin secretion, with a potential impact for novel class of treatment against type 2 diabetes.
Results
Detection of 34 coding OPRD1 variants in RaDiO
Among 6,971 participants from the RaDiO study (more than 97% of European descent)19, we detected 31 rare coding OPRD1 variants, including one nonsense variant (encoding p.E355*), in a total of 36 carriers, and three coding OPRD1 variants that were common (i.e., with a minor allele frequency [MAF] above 1%) in at least one population from gnomAD (v2.1.1) (Fig. 1 and Table S1). Among the 31 rare variants, 14 were not reported in gnomAD (v2.1.1) (Table S1). Before functional investigations, the burden of rare OPRD1 variants was not associated with any metabolic traits (Table S2).
Black variants are neutral variants according to our in vitro assays; red variants are loss-of-function variants that are genetically null and/or deleterious according to our in vitro assays; green variants are gain-of-function variants according to our in vitro assays. Transmembrane domains are highlighted in yellow.
Identification of loss- and gain-of-function OPRD1 variants via functional genetics
Each mutant or wild-type OPRD1 was expressed in human embryonic kidney 293 (HEK293) cells, and the Gαi–dependent serum response element (SRE) luciferase activity was analyzed in response to increasing concentrations of [D-Ala2]-Deltorphin II (DII) and [D-Pen2,D-Pen5]-Enkephalin (DPDPE), which are two specific agonists for DOP20. We used two different agonists to enhance the specificity of our assay. When compared to wild-type OPRD1, we found that 12 OPRD1 variants encoding p.P14R, p.G36E, p.K81N, p.D95N, p.A96V, p.N131D, p.M132I, p.R160G, p.R192Q, p.A221S, p.G248V, and p.D322E significantly decreased SRE luciferase activity in response to both DOP agonists (loss-of-function variants; Fig. S1). In contrast, seven OPRD1 variants encoding p.P24S, p.A32G, p.I52V, p.I183V, p.K250R, p.T361N and p.P366L (gain-of-function variants) significantly increased the SRE luciferase activity in response to both DOP agonists when compared to wild-type OPRD1 (gain-of-function variants; Fig. S1). Immunofluorescence assays showed that all mutants were effectively expressed and localized at the plasma membrane of HEK293 cells, except for the two loss-of-function mutants p.P14R and p.G36E (Fig. S2). The variants encoding these two mutants were both located in the extracellular terminus region (Fig. 1). Western blots showed that these two mutants tended to have a lower expression than wild-type DOP, while the other DOP mutants were expressed at the same magnitude (Fig. S3).
The amino acids affected by missense loss-of-function OPRD1 variants were significantly more conserved than the amino acids affected by gain-of-function variants (83% versus 29%, P = 0.027; Fig. S4). In this regard, half of the loss-of-function variants and only one gain-of-function variant were not reported in gnomAD (Table S1). The burden of gain-of-function OPRD1 variants tended to be higher in extracellular and intracellular terminus regions (P = 0.085; Fig. 1), where amino acids were significantly less conserved than in transmembrane domains and loops (16% versus 79%, P = 5.7 × 10−22; Fig. S4).
Mirror effect of loss- and gain-of-function OPRD1 variants on metabolic homeostasis
We then analyzed the effect of the burden of rare, gain- or loss-of-function variants on metabolic homeostasis. Loss-of-function variants were null variants and/or deleterious variants according to the luciferase assay. In the RaDiO study, all carriers of rare, loss- or gain-of-function OPRD1 variants were heterozygote and unrelated, and they were of European ancestry except for one individual.
All carriers of a rare, loss-of-function variant were overweight or obesity, except for one individual (92% overweight/obesity in carriers versus 58% in non-carriers; Table S3). Using the mixed-effects score test (MiST) adjusted for age, sex and ancestry, the burden of rare, loss-of-function OPRD1 variants was indeed significantly higher among individuals with overweight or obesity (0.31% of carriers among cases versus 0.038% of carriers among controls; P = 0.0054 with an odds ratio [OR] of 11; 95% confidence interval [CI], 2.1–194; Table 1) and its effect was significant on a higher BMI (P = 0.044 with an effect [π] of 0.35 ± 0.17; Table 1). In up to 187,242 participants from UK Biobank, we detected 12 additional rare, null variants (encoding p.M1?, p.E10*, p.S44*, p.V59Cfs*22, p.S106Vfs*5, p. P115Afs*31, p.W114*, p.C232Vfs*80, p.Y233*, p.I277Hfs*33, p.F320Sfs*158 and p.R356Pfs*119) that we considered loss-of-function, as well as 7 rare, loss-of-function variants previously detected in RaDiO (Table S1). All carriers of these variants were unrelated. We confirmed that the burden of rare, loss-of-function OPRD1 variants in UK Biobank was significantly higher among individuals with overweight or obesity (P = 0.0047 with an OR of 2.6; 95% confidence interval [CI], 1.4–5.5; Table 1) and its effect was significant on a higher BMI (P = 0.0028 with a π of 0.081 ± 0.027; Table 1) using MiST adjusted for age, sex and ancestry.
In contrast, all carriers from RaDiO of rare, gain-of-function OPRD1 variants did not have overweight or obesity, except one individual, but the effect of the burden on the decreased risk of overweight or obesity was not significant (0.083% of carriers among cases versus 0.19% of carriers among controls; Table 2). The gain-of-function OPRD1 variant encoding p.I52V (rs111622802; that is common in Africans only) was independently analyzed in the Accelerating Medicines Partnership (AMP) Type 2 Diabetes knowledge portal which enabled the assessment of its metabolic effect in African individuals including cases with type 2 diabetes21; the MAF of rs111622802 was too low in RaDiO and UK Biobank. We found that this variant was significantly associated with a lower BMI (P = 0.0038 with an effect [β] of −0.027 ± 0.0093; Table 2), when adjusting for age, sex, and ancestry.
In RaDiO, the burden of rare, loss-of-function variants tended to be lower among individuals with hyperglycemia or type 2 diabetes (0.063% of carriers among cases versus 0.25% of carriers among controls; P = 0.054 with an OR of 0.23; 95% CI, 0.033–0.95; Table 1), using MiST adjusted for BMI, age, sex, and ancestry. In the same line, none of the carriers of a rare, loss-of-function OPRD1 variant in UK Biobank had type 2 diabetes (Table 1). In contrast, in the AMP Type 2 Diabetes knowledge portal, we found that the gain-of-function OPRD1 variant encoding p.I52V was significantly associated with a higher risk of type 2 diabetes in 34,812 participants (P = 6.1 × 10−5 with an OR of 1.7; 95% CI, 1.3–2.2; Table 2), when adjusting for age, sex, BMI, and ancestry. This association signal was most probably driven by the effect in 4,966 African participants (P = 3.6 × 10−6 with an OR of 2.0; 95% CI, 1.5–2.6; Table 2).
In RaDiO, the burden of rare, gain-of-function variants improved lipid homeostasis with higher high-density lipoprotein (HDL) cholesterol and lower triglyceride levels (P = 0.049 with a π of 0.33 ± 0.17 and P = 0.046 with a π of −0.37 ± 0.18, respectively; Table 2), using MiST methods adjusted for BMI, age, sex, and ancestry. In the same line, the gain-of-function variant encoding p.I52V was associated with improved lipid homeostasis, with significantly lower triglyceride levels and low-density lipoprotein (LDL) cholesterol levels (P = 0.018 with a β of −0.13 ± 0.054, P = 0.0026 with a β of −0.37 ± 0.12, respectively; Table 2) and a nominal effect on increased HDL levels in African participants from the AMP Type 2 Diabetes knowledge portal (P = 0.055 with a β of 0.095 ± 0.049; Table 2), when adjusting for age, sex, BMI and ancestry.
OPRD1 expression in human pancreatic islets is inhibited under various conditions related to type 2 diabetes
Higher adiposity is usually strongly associated with higher risk of hyperglycemia and type 2 diabetes22. However, according to the present data, OPRD1 loss of function associates with higher adiposity and lower risk of hyperglycemia, while OPRD1 gain of function has opposite effects, in line with long-term consumption of opium11. We hypothesized that this discrepancy might be due to independent effects of DOP in the central nervous system versus the pancreas. Based on RNA-sequencing (RNA-seq) data from human and mouse pancreatic islets23,24,25,26,27, OPRD1 was the only opioid receptor gene expressed in human pancreatic islets (Fig. 2A). Neither Oprd1, Oprm1 nor Oprk1 was expressed in mouse pancreatic islets (Fig. 2A). Furthermore, OPRD1 was expressed in human purified beta cells, and its expression was significantly enriched in these purified beta cells when compared to beta-cell depleted islet preparations (P = 0.0075, accounting for islet purity and adjusted for multiple testing using the Benjamini-Hochberg method) (Fig. 2B)28. OPRD1 expression was significantly decreased in human pancreatic islets from donors with type 2 diabetes (P = 5.0 × 10−12; Fig. 2C)24,29,30,31, as well as under several in vitro exposures related to type 2 diabetes such as brefeldin A that perturbs endoplasmic reticulum (ER)-to-Golgi trafficking (P = 1.2 × 10−7; Fig. 2C) or a glucolipotoxic combination of palmitate and high glucose (P = 4.7 × 10−5; Fig. 2C). These results suggest that OPRD1 expression of in human pancreatic islets might be inhibited by conditions linked with type 2 diabetes pathophysiology including ER stress and glucolipotoxicity that are known to impair insulin secretion.
A Expression of OPRD1, OPRK1, and OPRM1 in human pancreatic islets (n = 11)24,25, and expression of Oprd1, Oprk1, and Oprm1 in mouse pancreatic islets (n = 14)26,27, assessed via RNA-seq. Data are mean ± SEM of TPM. B Expression of OPRD1, OPRK1, and OPRM1 was assessed in human purified pancreatic beta cells (n = 11) and beta-cell depleted islet preparations (i.e., non-beta cells; n = 5)28. Data are mean ± SEM of RPKM. C OPRD1 expression in human islets under different conditions and in type 2 diabetes. Brefeldin A (0.1 μg/mL; orange) vs control (green), 24 hours treatment (n = 4); a combination of palmitate (0.5 mM) and high glucose (22.2 mM) (orange) vs control (green), 48 hours treatment (n = 3); high glucose (22.2 mM; orange) vs control (green), 48 hours treatment (n = 4); palmitate (0.5 mM; orange) vs control (green), 48 hours treatment (n = 5); donors with type 2 diabetes (n = 47; orange) vs non-diabetic controls (n = 228; green). Data are box plots (showing the minimum, the median, the maximum, as well as all data). Adjusted P-values are shown relative to control and were calculated according to DESeq2 (v1.28.1; two-sided). ns not significant, RPKM reads per kilobase million, TPM transcripts per million.
DOP regulates insulin secretion from human pancreatic beta cells and islets
In human EndoCβH5 cells, i.e. post-mitotic mature pancreatic beta cells that endogenously express OPRD1 (transcripts per million [TPM] of 11 according to our RNA-seq data)32, a specific antagonist for DOP (i.e. naltrindole [NTI]) significantly increased insulin secretion in low and high glucose conditions (P = 0.0036 and P = 0.0017, respectively; Fig. 3A, B), in the presence of DII DOP agonist. Surprisingly, this stimulating effect of NTI remained when DOP was not activated with DII DOP agonist in low and high glucose conditions (P = 0.011 and P = 0.0015, respectively; Fig. 3A, B). Indeed, NTI sits in an exposed binding pocket of DOP, which inhibits the action of DII to suppress the signaling and biological response33. This result may suggest a constitutive or basal activity of DOP. Insulin content was similar between all conditions (Fig. S5). We confirmed the effect of NTI DOP antagonist on enhanced insulin secretion in human pancreatic islets. Indeed, NTI alone or the combination of NTI and DII significantly increased insulin secretion in both low (P = 0.022 and P = 0.0079, respectively; Fig. 3C, D) and high (P = 0.0040 and P = 0.00022, respectively; Fig. 3C, D) glucose conditions. Of note, we did not see any effect of DII on insulin secretion in either EndoCβH5 cells or human islets. This might be due to a lower affinity of DII for DOP when compared to those of NTI34.
(A) Insulin secretion was assessed in EndoCβH5 cells treated with no glucose (green; LG) either with 10 μM NTI (n = 10 independent cell preparations), 5 μM DII (n = 12), a combination of both (n = 12) or no treatment (n = 12), for 40 min. (B) Insulin secretion was assessed in EndoCβH5 cells treated with 20 mM glucose (violet; HG) supplemented with 10 μM NTI (n = 10), 5 μM DII (n = 12), a combination of both (n = 12) or no treatment (n = 11), for 40 min. (C) Insulin secretion was assessed in human islets treated with 2.8 mM glucose (green; LG) supplemented with 10 μM NTI (n = 24 single isolated islets), 5 μM DII (n = 24), a combination of both (n = 29), or no treatment (n = 28), for 120 min. (D) Insulin secretion was assessed in human islets treated with 16.7 mM glucose (violet; HG) supplemented with 10 μM NTI (n = 23), 5 μM DII (n = 24), a combination of both (n = 29) or no treatment (n = 28), for 120 min. Data are box plots (showing the minimum, the median, the maximum, as well as all data). P-values are shown relative to control by unpaired two-tailed Mann-Whitney test. DII, [D-Ala2]-Deltorphin II; HG, high glucose; LG, low glucose; ns, not significant; NTI, naltrindole.
In EndoCβH5 cells cultured at high glucose, RNA-seq analyses revealed that NTI treatment upregulated 904 genes and downregulated 795 genes (Fig. S6) when compared to the control condition. We next found enrichment of significantly upregulated genes in several pathways in response to NTI (Table S4): i/ nerve growth factor-stimulated transcription (P = 3.3×10−18) that is known to stimulate insulin secretion35, which is in line with the outcomes of insulin secretion assays in EndoCβH5 cells in response to NTI; ii/ negative regulation of intracellular signal transduction (P = 5.4 × 10−10) including negative regulation of ERK1 and ERK2 cascade (P = 5.6 × 10−5), which is in line with the canonical effect of NTI on increased cAMP production and decreased ERK pathway via DOP; iii/ regulation of protein transport (P = 2.6 × 10−9) including positive regulation of insulin secretion (P = 5.5 × 10−4), which is in line with our insulin secretion assays in EndoCβH5 cells in response to NTI; iv/ nuclear receptors meta-pathway (P = 4.0 × 10−8) and circadian clock (P = 6.4 × 10−8); and v/ apoptosis pathways which have already been reported in other cells36 (Table S4).
We verified the specificity of the agonistic and antagonistic effects of DII and NTI, respectively, on DOP in HEK293 and EndoCβH1 cells (i.e. a human pancreatic β-cell line) that do not endogenously express OPRD1 (or at least at a very low level). For this purpose, wild-type OPRD1 plasmid was overexpressed or not in these cells (Fig. S7 and Fig. S8). In HEK293 control cells, we found a slight effect of DII DOP agonist on decreased cAMP production (Fig. S9A). This could be due to the low expression of OPRD1 in HEK293 cells (Fig. S7). When overexpressing OPRD1, DII DOP agonist led to a reduction in cAMP production by more than a tenfold decrease (P < 7.4 × 10−7; Fig. S9B), while the addition of NTI DOP antagonist restored cAMP production by a threefold increase (P = 5.2 × 10−6; Fig. S9B). In EndoCβH1 control cells, we did not observe any significant effect of DII or NTI on Gαi–dependent SRE luciferase activity (Fig. S10A). When overexpressing OPRD1, DII DOP agonist led to a strong increase in SRE luciferase activity (P < 7.4 × 10−7; Fig. S10B), while the addition of NTI DOP antagonist inhibited this stimulation (P < 7.4 × 10−7; Fig. S10B). Therefore, DII and NTI only target DOP in these two cell models.
Discussion
Based on functional genetics, we found that OPRD1 loss of function led to higher adiposity but tended to decrease the risk of hyperglycemia. In contrast, OPRD1 gain of function was associated with lower adiposity and improved lipid metabolism but led to markedly higher risk of type 2 diabetes (independently of BMI).
The present results related to OPRD1 gain of function are in line with observational studies of people addicted to opium and its derivatives that reported an association between opioid use and lower adiposity but higher risk of hyperglycemia11,12,13,15,16,17. The pathophysiological mechanisms explaining the effects of opioids on type 2 diabetes and related traits in human addicts have been surprisingly unexplored in the past three decades13. Morphine that is the most prevalent and pharmacologically active alkaloid in opium has a high affinity for DOP34,37. However, multiple studies based on knock-out mice firmly established MOP as the primary opioid receptor driving effects of opium and its derivatives including morphine38. The lack of data on the link between opioids and diabetes might in part be due to the fact that the expression of mouse opioid receptor genes including Oprd1 is centralized in the nervous system and none mouse opioid receptor genes are expressed in pancreatic islets (Fig. 2A). Therefore, the putative effect of opium and its derivatives in mouse peripheral organs was poorly investigated. In humans, OPRD1 exhibited high expression levels in human pancreatic islets, surpassing those observed in several human brain regions30. Furthermore, early studies showed that met-enkephalin and β-endorphin (i.e., high-affinity, endogenous ligands for opioid receptors including DOP34) inhibited insulin secretion, and consequently increased plasma glucose concentrations39,40,41. Here, we also found that DOP inhibition through a specific antagonist strongly increased insulin secretion from human pancreatic beta cells and islets. Based on these results, we suggest therefore that the contribution of opioids to increased plasma glucose concentrations or even increased type 2 diabetes risk might be due to their stimulating effect on DOP activity in pancreatic islets. Of note, the present RNA-seq analyses showed that POMC encoding the precursor of β-endorphin was expressed in human pancreatic islets and β cells while the two other opioid precursor genes (i.e., PDYN and PENK) were not expressed. In addition, the genes that encode enzymes involved in POMC processing (i.e., PCSK1 encoding the proprotein convertase subtilisin/kexin type 1 [PC1/3], PCSK2 encoding proprotein convertase subtilisin/kexin type 2 [PC2], and CPE encoding carboxypeptidase E) were all expressed in human islets and β cells. Therefore, POMC might be processed into β-endorphin in human pancreatic islets and β cells, potentially exerting an autocrine or paracrine effect on decreased insulin secretion from β cells via DOP. This could explain the basal activity of DOP that we observed in human islets and EndoCβH5 cells.
Intriguingly, we found that OPRD1 expression in human islets was decreased in type 2 diabetes and under diabetogenic conditions. This points to a second layer of DOP regulation, at the level of gene expression. We speculate that OPRD1 expression might be inhibited in response to impaired insulin secretion.
Regarding adiposity, mice deleted for Oprd1 were hyperphagic, which was in line with the fact that opioids decreased food intake in mice13, but gained less weight when compared to wild-type mice fed an high‐energy diet42. This was due to an enhancement of the thermogenic program in brown adipose tissue42. Although not reaching genome-wide significance, genetic associations exist between common single nucleotide polymorphisms in OPRD1 and anorexia nervosa43,44. The anorexia nervosa-associated risk allele of rs569356 that was located in the promoter of OPRD1 increased OPRD1 expression45.
In conclusion, our present study identifies DOP as a key GPCR involved in metabolic disorders – in particular type 2 diabetes. A peripherally acting DOP antagonist unable to cross the blood-brain barrier might represent a drug to restore insulin secretion in response to glucose, and constitute an original class of drugs in the armamentarium for type 2 diabetes.
The present study is limited by the fact that we did not investigate the other classical opioid receptor genes (i.e., OPRM1 and OPRK1 encoding mu and kappa opioid receptors, respectively). Although they are not expressed in pancreatic islets, we cannot rule out a putative involvement in the link between opioids and diabetes, in particular through central effects on adiposity or even on metabolism.
In addition, we only analyzed one signaling pathway of DOP in response to two agonists for the definition of gain- and loss-of-function OPRD1 variants, while GPCRs trigger several pathways when activated. Gain- or loss-of-function variants in a GPCR gene can differently affect these pathways, as we previously found for deleterious variants of MTNR1B encoding melatonin receptor 1B or GLP1R encoding glucagon-like peptide-1 receptor46,47. In this regard, our observed genetic association between gain- or loss-of-function OPRD1 variants and type 2 diabetes is primarily supported by the variant effects on the Gαi–dependent SRE pathway only, and we cannot exclude the additional functional impact on other GPCR-related pathways. Furthermore, we did not differentiate between variant responses among putative subtypes of DOP (δ1 versus δ2), which do not respond with the same affinity to DPDPE and DII agonists48, while we observed that some variants responded differently to DPDPE and DII (Fig. S1). A larger study would be needed to distinguish the effect of distinct variants in response to DPDPE and DII, conduct appropriate association analyses, and analyze the putative effect of this distinct variants on insulin secretion.
Methods
Ethical approval
In RaDiO, all population studies followed ethical principles defined in the Helsinki Declaration (revised in 1996), and they were approved by local ethical committees from Corbeil-Essonnes Hospital (France), Comité Consultatif de Protection des Personnes se prêtant à des Recherches Biomédicales (CCPPRB) of Lille - Lille Hospital (Lille, France), Hotel-Dieu hospital (France), Bicêtre hospital (France). All participants signed an informed consent form.
UK Biobank has approval from the North West Multi-centre Research Ethics Committee (MREC) as a Research Tissue Bank (RTB) approval. This approval means that researchers do not require separate ethical clearance and can operate under the RTB approval. UK Biobank individual-level, phenotypic data, and whole exome sequencing data used in the present work were obtained under application #67575.
Study participants
In RaDiO, we analyzed up to 6971 blood DNA samples (that were successfully sequenced) from several population studies19: 1/ the D.E.S.I.R. 9-year prospective study including middle-aged men and women from western France49; 2/ participants who were recruited and followed-up either by the UMR 1283/8199 unit (Lille, France), by the Department of Nutrition of Hotel-Dieu Hospital (Paris, France), or by the Centre d’Etude du Polymorphisme Humain (CEPH, Saint-Louis Hospital, Paris, France)50,51; 3/ participants who were recruited from the Department of Endocrinology of the Corbeil-Essonnes Hospital (Corbeil-Essonnes, France)50; 4/ the French Fleurbaix-Laventie Ville Santé study52. Normal weight was defined as BMI < 25 kg m−², overweight as 25 ≤ BMI < 30 kg m−² and obesity as BMI ≥ 30 kg m−². Hyperglycemia was defined as fasting plasma glucose levels ≥ 5.6 mmol l−1 and/or use of drug therapy for hyperglycemia53. HDL cholesterol and triglycerides were only assessed in participants who were not under lipid-lowering agents. Systolic blood pressure and diastolic blood pressure were only assessed in participants who were not under blood pressure lowering medications.
In UK Biobank, we analyzed up to 187,242 samples, with available exome sequencing data. The participants were recruited by UK Biobank from across the United Kingdom between 13 March 2006 and 1 October 2010. Normal weight was defined as BMI < 25 kg m−², overweight as 25 ≤ BMI < 30 kg m−² and obesity as BMI ≥ 30 kg m−². People with type 2 diabetes in UK Biobank were defined as follows. Inclusion criteria: ‘HbA1c ≥ 48 mmol mol−1’ (field #30750) or ‘use of insulin’ (fields #6153 or #6177, code [3]) or ‘diabetes diagnosed by doctor’ (field #2443); Exclusion criteria: ‘at least one International Classification of Diseases (ICD)−10 code starting with E10’ (type 1 diabetes), ‘at least one ICD-10 code starting with O24’ (diabetes mellitus in pregnancy). Controls for the type 2 diabetes case-control study in UK Biobank were defined as follows. Inclusion criteria: ‘HbA1c < 42 mmol mol−1’ (field #30750) and ‘age ≥ 40 years’. Exclusion criteria: ‘at least one ICD-10 code starting with E10’ (type 1 diabetes), ‘at least one ICD-10 code starting with E11’ (non-insulin-dependent diabetes mellitus), ‘at least one ICD-10 code starting with E13’ (other specified diabetes mellitus), ‘at least one ICD-10 code starting with R73’ (elevated blood glucose level), ‘at least one ICD-10 code starting with O24’ (diabetes mellitus in pregnancy), ‘gestational diabetes only’ (field #4041), ‘diabetes diagnosed by doctor’ (field #2443), ‘use of insulin’ (fields #6153 or #6177, code [3]).
OPRD1 sequencing and detection of variants
In the RaDiO study, DNA sequencing of OPRD1 (NM_000911.4) was performed by next-generation sequencing as previously described54. Briefly, NimbleGen SeqCap EZ Choice XL target enrichment (Roche, Pleasanton, USA) was performed according to the manufacturer’s protocol for next-generation sequencing on the HiSeq 4000 system (Illumina, San Diego, USA), using a paired-end 2 × 150 bp protocol. The demultiplexing of sequence data was performed using bcl2fastq Conversion Software (Illumina; v2.17). Sequence reads were then mapped to the human genome (hg19/GRCh37) using Burrows-Wheeler Aligner (v0.7.13). The variant calling was performed using Genome Analysis ToolKit (GATK; v3.3). Only variants with a coverage higher than 8 reads were kept for further analyses. The annotation of variants was performed using the Ensembl Perl Application Program Interfaces (version 75) and custom Perl scripts to include data from both dbSNP (version 135) and dbNSFP (v3.0) databases. All rare coding variants (with a minor allele frequency [MAF] <1%) that were detected in OPRD1 had a QUAL score higher than 50. Furthermore, no variant had more than 5% missing genotype (with coverage below 8 reads or a QUAL score below 50) across the participants. We removed from the analyses 289 participants who had more than 5% missing genotypes (with coverage below 8 reads or a QUAL score below 50) across OPRD1, leading to 6971 participants who were successfully sequenced and analyzed in the present study.
In the UK Biobank, we used exome data from pVCF format (field #23156). Only variants with coverage higher than 10 reads and quality GQ score higher than 20 were kept for further analyses. Annotation of variants in OPRD1 (NM_000911.4) was done using the Ensembl Variant Effect Predictor (VEP) tool version 103 (RefSeq). Subsequently, the analysis was focused on loss-of-function variants only, namely the variants that were genetically null (i.e. nonsense, frameshift, canonical ±1 or 2 splice sites, start lost) or that were loss-of-function according to our functional analyses. No loss-of-function variant had more than 5% missing genotype (i.e. with coverage below 10 reads or a GQ score below 20) across the participants. All loss-of-function variants that were detected in OPRD1 actually had a mean GQ score higher than 48.
Ancestry assessment
In the RaDiO study, the ancestry of participants was assessed using the first five genotypic principal components (PC1 to PC5), as previously described19. Briefly, PC1 to PC5 were computed using a principal component analysis (PCA) with the R package flashpcaR, on 15,020 single nucleotide polymorphisms (SNPs; with a MAF > 5%) available in both RaDiO study and in the 1,000 Genomes project. In the UK Biobank, we also used the first five genetic principal components available in the field #22009.
Relatedness assessment
In RaDiO, pairwise identity by descent was computed using the ‘--genome’ command in PLINK, based on genotype data from Illumina MetaboChip arrays (genotyping success rate for each analyzed participant was above 99%). Prior to the analysis, variants in strong linkage disequilibrium (i.e. no variants with MAF < 1% and no pair of variants within 50 kb have an r² > 0.2) were excluded. The analyzed participants were considered ‘not related’ if the expected pairwise (‘PI_HAT’) was below 0.2.
In UK Biobank, we used ‘gfetch’ command (resource #668), with the ‘rel’ parameter. The analyzed participants were considered ‘not related’ if the kinship coefficients for all pairs of samples was below 0.2.
Statistical analyses for genetic association studies
In RaDiO and UK Biobank, the rare variants were analyzed as single clusters (e.g. clusters of loss-of-function OPRD1 variants) using the MiST method55. MiST provides a score statistic S(π) for the mean effect (π) of the cluster, and a score statistic S(τ) for the heterogeneous effect (τ) of the cluster. Let the equation of the model be: \(Y=aX+\pi GZ\) where Y is the trait of interest, X is the matrix of covariates (i.e. age, sex, BMI, PC1 to PC5 for assessing hyperglycemia or type 2 diabetes, cholesterol levels, and blood pressure, and age, sex, PC1 to PC5 for assessing adiposity), G is the matrix of OPRD1 variants and Z is a vector of ones repeated n times, with n the number of rare OPRD1 variants, leading to: \(\pi GZ={\pi }_{(i=1)}^{n}\pi {G}_{i}\) BMI and TG data were log-transformed before analysis. Loss-of-function OPRD1 variants included null variants or deleterious variants according to our in vitro analyses based on luciferase assays. Gain-of-function OPRD1 variants were defined according to our in vitro analyses based on luciferase assays. As none of the association studies of loss-of-function, gain-of-function, or neutral OPRD1 variants had significant heterogeneity, we only showed the P-values associated with the mean effect (π) of the cluster. These statistical analyses were performed using R software (v4.0.2).
In the AMP Type 2 Diabetes knowledge portal (accessed in September 2020), the association analyses for assessing the effect of the OPRD1 variant encoding p.I52V were performed using an additive regression model adjusted for age, sex, BMI, and PC1 to PC5 (for assessing type 2 diabetes, cholesterol levels, and blood pressure) or adjusted for age, sex, and PC1 to PC5 (for assessing BMI). For the analyses of cholesterol levels and blood pressure, we only included participants without type 2 diabetes. For the analyses of type 2 diabetes, we only included controls older than 50 years.
Plasmids
The plasmid containing wild-type OPRD1, the ‘STOP’ plasmid containing OPRD1 mutated for an early nonsense variant (c.3_4insTAA/p.E2*) which was used as a negative control, and three plasmids containing respectively the rare OPRD1 variants encoding p.F220L, p.F280L, and p.G365D were purchased from e-Zyvec (Loos, France). The plasmids containing each of the remaining 31 coding variants were generated from the wild-type OPRD1 plasmid using the QuickChange site-directed mutagenesis kit (Stratagene, San Diego, USA). The sequence of each plasmid was verified by Sanger sequencing.
Culture of human embryonic kidney 293 (HEK293) cells
HEK293 cells were cultured in Dulbecco’s Modified Eagles’ Medium (Gibco/Thermo Fisher Scientific, Waltham, USA) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin (Gibco/Thermo Fisher Scientific) at 37 °C, 5% CO2.
Luciferase assays
Using FuGENE HD (Promega, Madison, USA), HEK293 cells were transfected in suspension with 375 ng ml−1 OPRD1 plasmid (wild-type or with a variant), 225 ng ml−1 plasmid expressing the firefly luciferase gene under the control of Gαi–dependent SRE, and 150 ng ml−1 plasmid expressing the β-galactosidase gene. 100 µl of these cells (1 × 106 cells ml−1) were then seeded in a poly-L-lysine coated 96-well plate. The day after, cells were starved overnight in a medium including 0.1% fetal bovine serum. 48 h after the transfection, cells were treated for 5 h with increasing concentrations (1, 10, and 100 nM) of either DII (Tocris Bioscience, Bristol, UK) or DPDPE (Tocris Bioscience), both of which are DOP agonists. Then, the cells were lysed in 100 µl luciferase cell culture lysis buffer (Promega) for 30 minutes. Luciferase and β-galactosidase activity were measured as previously described54. Briefly, the luminescence was assessed by adding 25 μl of the luciferase assay system reagent (Promega) to 40 μl of lysate and was read using a Glomax luminometer (Promega). β-galactosidase activity was measured after incubation of 40 μl of lysate for 5 min with 100 μl of buffer and was read at 450 nm. Beta-galactosidase activity was very stable between all experiments. The experiments were performed in triplicate and were repeated at least four times. Luciferase measurements in relative luminescence units were normalized to β-galactosidase measures. Fold change (FC) was computed by dividing the normalized luciferase activity by the mean of the baseline luciferase activity (i.e. under dimethyl sulfoxide [DMSO] 0.1%). The effect of each OPRD1 variant (or wild-type) was analyzed using the following ANOVA model: FC = β0 + β1C + δ1V + δ2C×V + β2E + ϵ, where C was the concentration of DII or DPDPE (as a categorical variable), V was the OPRD1 variant (or wild-type) and E was the experiment ID.
Immunofluorescence assays
HEK293 cells were transfected in suspension using FuGENE HD (Promega) with 375 ng ml−1 OPRD1 plasmid (wild-type, STOP or with a variant). 2 ml of these cells (5×105 cells ml−1) were then seeded in a 6-well plate containing poly-L-lysine coated coverslips. 48 h after the transfection, the cells were fixed in paraformaldehyde 4% for 20 minutes and permeabilized in PBS – Triton X-100 0.1% for 20 minutes. The unoccupied binding surfaces were blocked in a blocking buffer (PBS, donkey serum 5% [Millipore/Sigma-Aldrich, St. Louis, USA], Triton X-100 0.1%) for 1 h. The cells were then incubated with the anti-DOP primary antibody (ab176324 [Abcam, Cambridge, United Kingdom]; diluted at 1/100 in blocking buffer) overnight at 4 °C, followed by a fluorescent secondary antibody (A-21206 [Thermo Fisher Scientific]; diluted at 1/500 in blocking buffer) for 2 h at 4 °C. F-actin was labelled with fluorescent phalloidin (#21836 [Thermo Fisher Scientific]; diluted at 1/300 in blocking buffer), and nuclei were labelled with DAPI (R-37606 [Thermo Fisher Scientific]) for 30 min at room temperature. The slides were finally mounted with Prolong Gold Antifade Mountant (Thermo Fisher Scientific). Cells were observed with a Cell Observer Spinning Disk microscope (Zeiss, Oberkochen, Germany).
Western blot assays
HEK293 cells were transfected in suspension using FuGENE HD (Promega), with 375 ng ml−1 OPRD1 plasmid (wild-type, STOP, or with a variant). 2 ml of these cells (1 × 106 cells ml−1) were then seeded in a poly-L-lysine coated 6-well plate. Two days after the transfection, the proteins were extracted in Pierce RIPA buffer (Thermo Fisher Scientific), supplemented with dithiotreitol 1 mM (Invitrogen/Thermo Fisher Scientific), protease, and phosphatase inhibitors (Roche). Protein concentration was measured via the Pierce BCA protein assay kit (Thermo Fisher Scientific), according to manufacturer’s instructions. 40 µg proteins were then loaded on a 12% polyacrylamide gel, separated by electrophoresis, and then transferred on a 0.2 µm nitrocellulose membrane (GE Healthcare, Chicago, USA). The unoccupied binding surfaces on the membranes were blocked in blocking buffer (Tris buffered saline [TBS] 1×, Tween 20 0.1%, and skimmed milk 5%) for 1 h. DOP was then labelled using the primary anti-DOP antibody (ab176324 [Abcam]; diluted at 1/1000 in blocking buffer), followed by a fluorescent anti-rabbit secondary antibody (SA5-35571 [Invitrogen/Thermo Fisher Scientific]; diluted at 1/5000 in blocking buffer), and revealed via the Odyssey CLx imaging system at an 800 nm wavelength. To normalize our data, β-actin, used as a loading control, was labelled using the primary anti-β-actin antibody (#3700 [Cell Signaling Technology, Leiden, The Netherlands]; diluted at 1/5000 in blocking buffer), followed by a fluorescent anti-mouse secondary antibody (#35518 [Thermo Fisher Scientific]; diluted at 1/5000 in blocking buffer) or a HRP anti-mouse secondary antibody (#NA931 [Merck/Sigma-Aldrich]; diluted at 1/5000 in blocking buffer). β-actin was then revealed via the Odyssey CLx imaging system at a 700 nm wavelength or via the Chemidoc imaging system (Bio-Rad, Hercules, California, USA), respectively. At least, three independent experiments were performed for each mutant. Two-by-two expression comparisons of each DOP mutant versus wild-type DOP were performed using a Wilcoxon rank sum test.
RNA-seq analysis in mouse and human pancreatic islets
We analyzed RNA-seq data from 11 human and 14 mouse pancreatic islet preparations, which were available through the GEO data repository24,25,26,27. Fastq files were downloaded from GEO using fastq-dump (v2.9.6). The quality control of Fastq files was performed using FastQC (version 0.11.9). Sequencing reads were aligned to the human genome (hg38) or the mouse genome (mm10) using STAR aligner (v2.7.1a). On average, 51.6 million reads and 52.0 million reads are accurately mapped against the genomes of human and mouse pancreatic islets, respectively. Processed count tables and transcripts per million (TPM) were then obtained per gene using RSEM (v1.3.0) and Ensembl gene annotation (Ensembl Archive Release 100) using biomaRT package (v2.45) with R (v3.6.3).
RNA-seq analysis in human pancreatic islets from patients with type 2 diabetes or under various treatments related to type 2 diabetes
Bulk RNA-seq raw datasets were downloaded from GEO under accession number GSE152615 for brefeldin A-exposed human islets (0.1 mg/ml for 24 h, n = 4)24 and under GSE159984 for human islets exposed to palmitate (0.5 mM) and/or high glucose (22.2 mM, 48 h, n = 3-5 per condition)29. Brefeldin A is a stressor of endoplasmic reticulum which was found to significantly decrease insulin secretion from human pancreatic islets56. Data on human islets from donors with type 2 diabetes were integrated from three cohorts, one from T2DSystems’ TIGER, downloaded from the European Genome-Phenome Archive under accession number EGAS00001005535, and the other two from GEO under accession numbers GSE159984 and GSE50244, resulting in a total of 47 donors with type 2 diabetes and 228 donors without type 2 diabetes29,30,31. None of the donors carried rare coding variants in OPRD1. For each dataset, raw sequencing reads were processed with fastp (v0.19.6) using default parameters for adaptor and low-quality base trimming to obtain clean reads that were longer than half of the sequencing read length. Transcript and gene expression was quantified by Salmon (v1.4.0) with additional parameters “--seqBias --gcBias --validateMapings” using indexed GENCODE V36 (GRCh38.p13) as transcriptome reference and normalized as TPM. Gene expression was visualized via R package ggplot2. Differential gene expression was assessed with R package DESeq2 (v1.28.1) using the raw count matrix generated by Salmon. For the integrated cohort with type 2 diabetes, we applied batch correction in the general linear model, taking batch as a confounding factor. Fold change and Benjamini-Hochberg corrected P-values were computed via the Wald test of the DESeq2 algorithm.
Culture of EndoCβH5 cells
EndoCβH5 cells (Human Cell Design) were cultured in Ultiβ1 medium (Human Cell Design) for one week following manufacturer’s instructions.
Insulin secretion assays in EndoCβH5 cells
Static insulin secretion assays on EndoCβH5 cells were performed following manufacturer’s instructions, using four lots. Briefly, one day before insulin secretion experiments, EndoCβH5 cells seeded in a 96-well plate coated with βcoat were cultured with ULTI-ST medium (Human Cell Design). These cells were then washed twice with Krebs-BSA buffer and starved with Krebs-BSA buffer for 1 h, supplemented or not with 10 μM of the DOP antagonist naltrindole (naltrindole hydrochloride, Sigma-Aldrich) for the conditions that would later be treated with naltrindole or the combination of naltrindole and DII. The cells were then incubated with Krebs-BSA buffer including 0 mM (LG) or 20 mM (HG) glucose, supplemented with 10 μM naltrindole (n = 10 independent cell preparations for LG and HG conditions), 5 μM DII (n = 12 for LG and HG conditions), a combination of both (n = 12 for LG and HG conditions) or no treatment (n = 12 for LG and n = 11 for HG) for 40 minutes. The supernatant was collected, and the cells were lysed in TETG buffer for 2 minutes at 4 °C. Supernatants and cell lysates were then centrifuged at 500 g for 5 minutes. Insulin levels in the supernatants and in cell lysates were assessed using the human insulin ELISA kit (Mercodia), as described by the manufacturer. For each condition, the concentration of insulin measured in the supernatants at 0 and 20 mM glucose was normalized to the insulin content of the cell lysate.
Insulin secretion assays in human pancreatic islets
Insulin secretion assays were performed on single isolated human pancreatic islets purchased from InSphero (Schlieren, Switzerland) as described by the manufacturer, from four donors. Briefly, single isolated islets were washed twice with Krebs-Ringer HEPES buffer (KRHB) including 2.8 mM glucose, and then starved for 1 h in KRHB including 2.8 mM glucose combined or not with 10 µM naltrindole for the conditions that would later be treated with naltrindole or the combination of naltrindole and DII. Then, the islets were washed and incubated with KRHB including 2.8 mM glucose, either with 5 µM DII (n = 24 single isolated islets), 10 µM naltrindole (n = 24), a combination of both (n = 29), or no treatment (n = 28) for 2 h. The supernatant was collected and replaced by KRHB including 16.7 mM glucose, either with 5 µM DII (n = 24), 10 µM naltrindole (n = 23), a combination of both (n = 29) or no treatment (n = 28) for 2 h. This supernatant was then collected. Finally, the islets were lysed and their viability assessed using the CellTiter-Glo 3D reagent (Promega) supplemented with protease inhibitors. ATP content, corresponding to the viability of the islets, was evaluated by luminescence reading using a Glomax device. Insulin levels in the supernatants and in the cell lysates were assessed using the STELLUX Chemi Human Insulin ELISA purchased from Alpco (Windham, NH, USA), following manufacturer’s instructions. For each condition, the concentration of insulin secreted in the supernatant was normalized to insulin content of the cell lysate.
RNA-seq in EndoCβH5 cells
For this experiment, we used the high glucose condition (16.7 mM glucose) supplemented or not with naltrindole (please see Insulin secretion assays in EndoCβH5 cells for further details). RNA was extracted from EndoCβH5 cells using the NucleoSpin RNA kit (Macherey-Nagel, Düren, Germany), following manufacturer’s recommendations. 100 ng total RNA was used to perform library preparation using the NextFlex Poly(A) Bead 2.0 (v21.01; PerkinElmer Waltham, USA). Sequencing was done on the NovaSeq6000 system (flowcell SP, Illumina) using a paired-end 2 × 100 bp protocol. Raw data were demultiplexed using bcl2fastq v2.20.0.422 (Illumina). An adapter trimming step was done using trimmomatic version 0.39 (MINLEN:35 AVGQUAL:20). Mapping of reads on the human genome (Hg38) was performed using STAR (v2.7.3a). On average, 49 million reads are accurately mapped against the human genome. Raw and normalized counting steps were done using RSEM (v1.3.1) using a GTF from Encode (version 39), and Ensembl (version 105) for gene name annotations. Differential analysis was performed using DESeq2 (v1.24.0). Gene-set enrichment analysis was done using Metascape (accessed in April 2022)57.
Specificity assessment of DII and NTI on DOP in HEK293 cells
Using FuGENE HD, HEK293 cells were transfected in suspension with 100 ng ml−1 OPRD1 plasmid or its backbone that does not contain OPRD1 (called empty plasmid). 500 µl of these cells (4 × 105 cells ml−1) were then seeded in a poly-L-lysine coated 24-well plate. The day after, cells were starved overnight in a medium containing 0.1% fetal bovine serum. 48 h after the transfection, cells were treated for 45 minutes with 10 µM forskolin (Sigma-Aldrich), in combination with 10 nM of DOP agonist DII with or without 5 nM of DOP antagonist naltrindole. These cells were then lysed in 400 µL of sample diluent 1X for 10 minutes, and AMPc production was measured for each condition using the Cyclic AMP Direct EIA Kit (Arbor Assays) following manufacturers’ instructions. This experiment was performed in four replicates and was repeated three times. Transfection efficiency was verified for each experiment through quantitative PCR. For this purpose, RNA from cells previously seeded in 24-well plate and transfected with OPRD1 plasmid or the empty plasmid was extracted 48 h after the transfection using the NucleoSpin RNA kit (Macherey-Nagel) following manufacturer’s recommendations, and retrotranscribed using the High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). Quantitative PCR was performed using the Power SYBR Green PCR Master Mix (Applied Biosystems byThermo Fisher Scientific), on the Quantstudio 7 Pro device (Thermo Fisher Scientific). RPLP0 was used as a reference gene.
Culture of EndoCβH1 cells
EndoCβH1 cells (Human Cell Design, Toulouse, France) were cultured at a density of 2.5 × 106 in T75 flasks primarily coated with βcoat (Human Cell Design) at 37 °C, 5% CO2, according to manufacturer’s instructions58.
Specificity assessment of DII and NTI on DOP in EndoCβH1 cells
EndoCβH1 cells were seeded in a 96-well plate coated with βcoat (confluency of 75,000 cells/well). The day after, these cells were transfected using NeuroMag (OZ Biosciences, Marseille, France) with three plasmids: the one containing wild-type OPRD1, the one containing the luciferase gene under the control of SRE, and the one containing the β-galactosidase gene. For the control cells, the plasmid containing wild-type OPRD1 was replaced by its backbone, in which OPRD1 is missing. Three days after the transfection, after being starved overnight, these cells were treated with DMSO 0.2% (basal condition), 100 nM DII (Tocris Bioscience, Bristol, UK), or a combination of 100 nM DII and 100 nM NTI (naltrindole hydrochloride, Sigma-Aldrich) for six hours. Then, the cells were lysed in 100 µl luciferase cell culture lysis buffer (Promega) for 30 minutes. This experiment was performed in six replicates and was repeated two times. Luciferase and β-galactosidase activity were measured and analysed as mentioned above. Transfection efficiency was verified for each experiment through quantitative PCR as mentioned above.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
All relevant data have been included in the manuscript and/or in its supplementary tables and figures. Given the sensitivity and risk of re-identification, all clinical data linked with DNA-seq data for this study are available only upon request from Amélie Bonnefond (amelie.bonnefond@inserm.fr; timeframe for response to requests: one month). One limitation of the request will be its focus on the OPRD1 gene, and the transfer of data under a material transfer agreement. RNA-seq raw data from EndoCβH5 are available in Gene Expression Omnibus (GEO accession: GSE271922; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE271922). RNA-seq data from the 11 human and 14 mouse pancreatic islet preparations (control conditions) were available through GEO (GSE35296 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE35296], GSE102371, GSE53949, GSE156903, GSE152576, GSE149488). For human islets treated under different type 2 diabetes-related conditions, RNA-seq data were downloaded from GEO under accession number GSE152615 for brefeldin A-exposed human islets and under GSE159984 for human islets exposed to palmitate and/or high glucose. Data on human islets from donors with type 2 diabetes were integrated from three cohorts, one from T2DSystems’ TIGER, downloaded from the European Genome-Phenome Archive under accession number EGAS00001005535, and the other two from GEO under accession numbers GSE159984 and GSE50244. We also used the following web links for publicly available datasets: (1) AMP T2D Knowledge Portal (September 2020). http://www.type2diabetesgenetics.org/gene/geneInfo/OPRD1; (2) Genome Aggregation Database (gnomAD). https://gnomad.broadinstitute.org/; (3) dbNSFP. https://sites.google.com/site/jpopgen/dbNSFP; (4) dbSNP. https://www.ncbi.nlm.nih.gov/snp/. Source data are provided with this paper.
Code availability
Code to perform analyses related to genetics (bioinformatics and biostatistics) in this manuscript are available at https://github.com/umr1283/OPRD1 (https://doi.org/10.5281/zenodo.12666882).
References
Gourgari, E., Wilhelm, E. E., Hassanzadeh, H., Aroda, V. R. & Shoulson, I. A comprehensive review of the FDA-approved labels of diabetes drugs: Indications, safety, and emerging cardiovascular safety data. J. Diab. Complications 31, 1719–1727 (2017).
Bonnefond, A. & Froguel, P. Rare and common genetic events in type 2 diabetes: what should biologists know? Cell Metab. 21, 357–368 (2015).
El-Sayed Moustafa, J. S. & Froguel, P. From obesity genetics to the future of personalized obesity therapy. Nat. Rev. Endocrinol. 9, 402–413 (2013).
Locke, A. E. et al. Genetic studies of body mass index yield new insights for obesity biology. Nature 518, 197–206 (2015).
Ndiaye, F. K. et al. Expression and functional assessment of candidate type 2 diabetes susceptibility genes identify four new genes contributing to human insulin secretion. Mol. Metab. 6, 459–470 (2017).
Babenko, A. P. et al. Activating mutations in the ABCC8 gene in neonatal diabetes mellitus. N. Engl. J. Med. 355, 456–466 (2006).
Kühnen, P. et al. Proopiomelanocortin deficiency treated with a Melanocortin-4 receptor agonist. N. Engl. J. Med. 375, 240–246 (2016).
Fitzgerald, K. et al. A highly durable RNAi therapeutic inhibitor of PCSK9. N. Engl. J. Med. 376, 41–51 (2017).
Sriram, K. & Insel, P. A. G protein-coupled receptors as targets for approved drugs: how many targets and how many drugs? Mol. Pharm. 93, 251–258 (2018).
Pathan, H. & Williams, J. Basic opioid pharmacology: an update. Br. J. Pain. 6, 11–16 (2012).
Najafipour, H. & Beik, A. The impact of opium consumption on blood glucose, serum lipids and blood pressure, and related mechanisms. Front Physiol. 7, 436 (2016).
Elman, I. et al. Metabolic and addiction Indices in patients on opioid agonist medication-assisted treatment: a comparison of buprenorphine and methadone. Sci. Rep. 10, 5617 (2020).
Vuong, C., Van Uum, S. H. M., O’Dell, L. E., Lutfy, K. & Friedman, T. C. The effects of opioids and opioid analogs on animal and human endocrine systems. Endocr. Rev. 31, 98–132 (2010).
Leslie, R. D., Pyke, D. A. & Stubbs, W. A. Sensitivity to enkephalin as a cause of non-insulin dependent diabetes. Lancet 1, 341–343 (1979).
Karam, G. A. et al. Effects of opium addiction on some serum factors in addicts with non-insulin-dependent diabetes mellitus. Addict. Biol. 9, 53–58 (2004).
Gozashti, M. H., Yazdi, F., Salajegheh, P., Dehesh, M. M. & Divsalar, K. Fasting blood glucose and insulin level in opium addict versus non-addict individuals. Addict. Health 7, 54–59 (2015).
Ceriello, A. et al. Impaired glucose metabolism in heroin and methadone users. Horm. Metab. Res 19, 430–433 (1987).
Peng, J., Sarkar, S. & Chang, S. L. Opioid receptor expression in human brain and peripheral tissues using absolute quantitative real-time RT-PCR. Drug Alcohol Depend. 124, 223–228 (2012).
Bonnefond, A. et al. Pathogenic variants in actionable MODY genes are associated with type 2 diabetes. Nat. Metab. 2, 1126–1134 (2020).
Raynor, K. et al. Pharmacological characterization of the cloned kappa-, delta-, and mu-opioid receptors. Mol. Pharm. 45, 330–334 (1994).
Flannick, J. et al. Exome sequencing of 20,791 cases of type 2 diabetes and 24,440 controls. Nature 570, 71–76 (2019).
American Diabetes Association. 8. Obesity Management for the Treatment of Type 2 Diabetes: Standards of Medical Care in Diabetes-2020. Diabetes Care 43, S89–S97 (2020).
Rachdi, L. et al. Regulated expression and function of the GABAB receptor in human pancreatic beta cell line and islets. Sci. Rep. 10, 13469 (2020).
Bone, R. N. et al. A computational approach for defining a signature of β-Cell Golgi stress in diabetes. Diabetes 69, 2364–2376 (2020).
Kolic, J. et al. Differential effects of Voclosporin and Tacrolimus on insulin secretion from human islets. Endocrinology 161, bqaa162 (2020).
Gharavy, S. N. M. et al. Sexually Dimorphic Roles for the Type 2 Diabetes-Associated C2cd4b Gene in Murine Glucose Homeostasis. http://biorxiv.org/lookup/doi/10.1101/2020.05.18.099200 (2020) https://doi.org/10.1101/2020.05.18.099200.
Berthault, C., Staels, W. & Scharfmann, R. Purification of pancreatic endocrine subsets reveals increased iron metabolism in beta cells. Mol. Metab. 42, 101060 (2020).
Nica, A. C. et al. Cell-type, allelic, and genetic signatures in the human pancreatic beta cell transcriptome. Genome Res 23, 1554–1562 (2013).
Marselli, L. et al. Persistent or transient human β cell dysfunction induced by metabolic stress: specific signatures and shared gene expression with Type 2 diabetes. Cell Rep. 33, 108466 (2020).
Alonso, L. et al. TIGER: The gene expression regulatory variation landscape of human pancreatic islets. Cell Rep. 37, 109807 (2021).
Fadista, J. et al. Global genomic and transcriptomic analysis of human pancreatic islets reveals novel genes influencing glucose metabolism. Proc. Natl Acad. Sci. 111, 13924–13929 (2014).
Szczerbinska, I. et al. Large-scale functional genomics screen to identify modulators of human β-cell insulin secretion. Biomedicines 10, 103 (2022).
Granier, S. et al. Structure of the δ-opioid receptor bound to naltrindole. Nature 485, 400–404 (2012).
Borsodi, A. et al. Opioid receptors (version 2019.4) in the IUPHAR/BPS Guide to Pharmacology Database. GtoPdb CITE 2019, https://journals.ed.ac.uk/gtopdb-cite/article/view/3202 (2019).
Rosenbaum, T., Sánchez-Soto, M. C. & Hiriart, M. Nerve growth factor increases insulin secretion and barium current in pancreatic β-Cells. Diabetes 50, 1755–1762 (2001).
Chen, Y. L., Law, P. Y. & Loh, H. H. Inhibition of akt/protein kinase B signaling by naltrindole in small cell lung cancer cells. Cancer Res 64, 8723–8730 (2004).
Carlin, M. G., Dean, J. R. & Ames, J. M. Opium alkaloids in harvested and thermally processed Poppy seeds. Front Chem. 8, 737 (2020).
Fairbanks, C. A. & Peterson, C. D. The opioid receptor: emergence through millennia of pharmaceutical sciences. Front Pain. Res 4, 960389 (2023).
Giugliano, D. et al. Inhibitory effect of enkephalin on insulin secretion in healthy subjects and in non insulin-dependent diabetic subjects. Metabolism 36, 286–289 (1987).
Giugliano, D. et al. Beta-endorphin and islet hormone release in humans: evidence for interference with cAMP. Am. J. Physiol. 257, E361–E366 (1989).
Feldman, M., Kiser, R. S., Unger, R. H. & Li, C. H. Beta-endorphin and the endocrine pancreas. Studies in healthy and diabetic human beings. N. Engl. J. Med. 308, 349–353 (1983).
Czyzyk, T. A. et al. Mice lacking δ-opioid receptors resist the development of diet-induced obesity. FASEB J. 26, 3483–3492 (2012).
Brown, K. M. O. et al. Further evidence of association of OPRD1 & HTR1D polymorphisms with susceptibility to anorexia nervosa. Biol. Psychiatry 61, 367–373 (2007).
Wang, K. et al. A genome-wide association study on common SNPs and rare CNVs in anorexia nervosa. Mol. Psychiatry 16, 949–959 (2011).
Zhang, H. et al. Functional impact of a single-nucleotide polymorphism in the OPRD1 promoter region. J. Hum. Genet. 55, 278–284 (2010).
Karamitri, A. et al. Type 2 diabetes-associated variants of the MT2 melatonin receptor affect distinct modes of signaling. Sci. Signal 11, eaan6622 (2018).
Gao, W. et al. Human GLP1R variants affecting GLP1R cell surface expression are associated with impaired glucose control and increased adiposity. Nat. Metab. 5, 1673–1684 (2023).
Dietis, N., Rowbotham, D. J. & Lambert, D. G. Opioid receptor subtypes: fact or artifact? Br. J. Anaesth. 107, 8–18 (2011).
Balkau, B. An epidemiologic survey from a network of French Health Examination Centres, (D.E.S.I.R.): epidemiologic data on the insulin resistance syndrome]. Rev. Epidemiol. Sante Publique 44, 373–375 (1996).
Sladek, R. et al. A genome-wide association study identifies novel risk loci for type 2 diabetes. Nature 445, 881–885 (2007).
Meyre, D. et al. Genome-wide association study for early-onset and morbid adult obesity identifies three new risk loci in European populations. Nat. Genet. 41, 157–159 (2009).
Romon, M. et al. Relationships between physical activity and plasma leptin levels in healthy children: the Fleurbaix-Laventie Ville Santé II Study. Int. J. Obes. Relat. Metab. Disord. 28, 1227–1232 (2004).
American Diabetes Association. 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes-2019. Diabetes Care 42, S13–S28 (2019).
Baron, M. et al. Loss-of-function mutations in MRAP2 are pathogenic in hyperphagic obesity with hyperglycemia and hypertension. Nat. Med. 25, 1733–1738 (2019).
Sun, J., Zheng, Y. & Hsu, L. A unified mixed-effects model for rare-variant association in sequencing studies. Genet. Epidemiol. 37, 334–344 (2013).
Bugliani, M. et al. Modulation of autophagy influences the function and survival of human pancreatic beta cells under endoplasmic reticulum stress conditions and in type 2 diabetes. Front Endocrinol. (Lausanne) 10, 52 (2019).
Zhou, Y. et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat. Commun. 10, 1523 (2019).
Ravassard, P. et al. A genetically engineered human pancreatic β cell line exhibiting glucose-inducible insulin secretion. J. Clin. Invest. 121, 3589–3597 (2011).
Acknowledgements
We are grateful to all individuals included in the different cohort studies. We thank Julie Maillet and Frédéric Allegaert for technical assistance, and Anthony Piron for the OPRD1 variant lookup in TIGER. We acknowledge the Genome Aggregation Database (gnomAD) and the groups that provided exome and genome variant data to this resource (the full list is available at https://gnomad.broadinstitute.org/about). We thank the AMP Type 2 Diabetes knowledge portal and the groups that provided data to this resource. This research has been conducted using the UK Biobank Application #67575. This study was funded by the French National Research Agency (ANR-10-LABX-46 [European Genomics Institute for Diabetes] to PF and AB), the French National Research Agency (ANR-10-EQPX-07-01 [LIGAN-PM] to PF and AB), the European Research Council (ERC Reg-Seq – 715575 and ERC OpiO – 101043671, to AB), the EFSD New Targets for Diabetes or Obesity-related Metabolic Diseases Programme supported by an educational research grant from MSD (to AB) and the National Center for Precision Diabetic Medicine – PreciDIAB, which is jointly supported by the French National Agency for Research (ANR-18-IBHU-0001), by the European Union (FEDER), by the Hauts-de-France Regional Council and by the European Metropolis of Lille (MEL). The study was also supported by “France Génomique” consortium (ANR-10-INBS-009). XY was supported by the Fondation ULB and the China Scholarship Council. MCnop acknowledges support by the Walloon Region SPW-EER (Win2Wal project BetaSource), the Fonds National de la Recherche Scientifique (FRS-FNRS) and the Francophone Foundation for Diabetes Research (FFRD, which is sponsored by the French Diabetes Federation, Abbott, Eli Lilly, Merck Sharp & Dohme and Novo Nordisk).
Author information
Authors and Affiliations
Contributions
Conceptualization: P.F., ABonnefond; Methodology/Investigation: S.M., G.Q., J.M., C.B., R.B., A.L., H.L., M.H., A.D., V.S., M.Baron, E.D., B.T., E.V., S.A., X.Y., M.D., A.B., MBoissel, M.C., AB.; Sample acquisition: B.B., G.C., S.F., M.M., R.R., P.M., M.C., R.S. and P.F.; Visualization: S.M., A.B.; Funding acquisition: P.F., A.B.; Supervision: PF, A.B.; Writing – original draft: S.M., A.B.; Writing – review & editing: all authors
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Meulebrouck, S., Merrheim, J., Queniat, G. et al. Functional genetics reveals the contribution of delta opioid receptor to type 2 diabetes and beta-cell function. Nat Commun 15, 6627 (2024). https://doi.org/10.1038/s41467-024-51004-6
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s41467-024-51004-6
