PNAS Plus
Fetal endocannabinoids orchestrate the organization of pancreatic islet microarchitecture
SIGNIFICANCE
Glucagon and insulin are produced in distinct cell populations within pancreatic Langerhans islets, where intercellular interactions control their production and release. Modifications to the microstructure of pancreatic islets are implicated in disease pathogenesis, but the developmental rules underlying cell commitment and segregation are incompletely understood. We show that endocannabinoids (ω-6) via CB1 cannabinoid receptors and endovanilloid ligands via transient receptor potential cation channel subfamily V member 1, as well as dietary ω-3 polyunsaturated fatty acids, affect the cellular organization of pancreatic islets during organ development. Thus, lipid signaling emerges as a key determinant of tissue organization and can program hormone secretion for life.
ABSTRACT
Endocannabinoids are implicated in the control of glucose utilization and energy homeostasis by orchestrating pancreatic hormone release. Moreover, in some cell niches, endocannabinoids regulate cell proliferation, fate determination, and migration. Nevertheless, endocannabinoid contributions to the development of the endocrine pancreas remain unknown. Here, we show that α cells produce the endocannabinoid 2-arachidonoylglycerol (2-AG) in mouse fetuses and human pancreatic islets, which primes the recruitment of β cells by CB1 cannabinoid receptor (CB1R) engagement. Using subtractive pharmacology, we extend these findings to anandamide, a promiscuous endocannabinoid/endovanilloid ligand, which impacts both the determination of islet size by cell proliferation and α/β cell sorting by differential activation of transient receptor potential cation channel subfamily V member 1 (TRPV1) and CB1Rs. Accordingly, genetic disruption of TRPV1 channels increases islet size whereas CB1R knockout augments cellular heterogeneity and favors insulin over glucagon release. Dietary enrichment in ω-3 fatty acids during pregnancy and lactation in mice, which permanently reduces endocannabinoid levels in the offspring, phenocopies CB1R−/− islet microstructure and improves coordinated hormone secretion. Overall, our data mechanistically link endocannabinoids to cell proliferation and sorting during pancreatic islet formation, as well as to life-long programming of hormonal determinants of glucose homeostasis.
Anandamide (AEA) and 2-arachydonoylglycerol (2-AG), major endocannabinoids (eCBs), are involved in the regulation of energy homeostasis through coordinated actions in peripheral organs (adipose tissue, liver, and pancreas) and brain (hypothalamus, ventral striatum) (1). eCB signals are particularly significant to coordinate the regulated release of insulin and glucagon from mature pancreatic islets (2–6). Genetic evidence from CB1 cannabinoid receptor−/− (CB1R−/−) mice supports these findings because CB1R−/− mice are lean, resistant to high fat diet-induced obesity and diabetes (4, 7–9). Whether eCBs impact the formation of the endocrine pancreas and predispose it to long-lasting changes in hormone release postnatally remains unknown.
Because eCBs broadly affect cell proliferation, fate, motility, and differentiation (e.g., in sperm, hematopoietic and T cells, and neurons) (10–13), it is likely that they play a role in the cellular organization of developing pancreatic islets, possibly by affecting the spatial segregation of α and β cells. A contribution of eCBs to cell diversification and positioning in the developing pancreas is supported by the temporal control of their levels in fetal tissues (14) and circulation (15). Moreover, α and β cells in mature pancreatic islets express the molecular machinery for eCB metabolism together with CB1Rs and transient receptor potential cation channels, particularly subfamily V member 1 (TRPV1) (2, 16, 17). Understanding these developmental processes is also relevant to postnatal life because pancreatic α- and β-cell placement can be reconfigured upon metabolic demands in both rodents (18) and humans (19), altering the efficacy of endocrine responsiveness.
Differential ligand and receptor recruitment within pleiotropic eCB signaling networks might facilitate a cellular context- and stage-dependent diversification of eCB signals. Thus, the coordinated availability of 2-AG and AEA and their varied action on CB1Rs (20) and TRPV1 channels (21) are well-positioned to orchestrate progenitor proliferation and the survival, migration, and ability of hormone secretion from ensuing differentiated cell lineages. Here, we demonstrate that paracrine 2-AG signaling determines cell segregation via CB1R-mediated adhesion signaling in the fetal mouse pancreas. In turn, chemical or genetic inactivation of TRPV1s on β cells increases cell proliferation both in vitro and in vivo, typically affecting the size of islets formed. Reducing eCB precursor bioavailability during pregnancy by ω-3 polyunsaturated fatty acid (PUFA) intake increases cellular heterogeneity and improves the temporal coordination of glucagon/insulin release, phenocopying CB1R−/− mice, as well as human islets (19). Cumulatively, our results outline a candidate mechanism for life-long cellular adaptation to metabolic challenges.
RESULTS
Paracrine 2-AG Axis in the Fetal Pancreas.
It is critical to document the cellular sites of eCB metabolism and receptor-mediated signaling to appreciate the abundance, cellular specificity, and modes of communication for these signaling lipids during the formation of the endocrine pancreas, even more so because eCB signaling so far has been documented only in the fetal brain and musculature (10, 22) and has been shown to progress from a predominantly autocrine (prenatal) toward a paracrine (postnatal) mode of action (23). Similar changes in the molecular configuration of eCB signaling in the fetal pancreas would have a notable impact on both reconstructing signaling events and interpreting functional outcome and disease relevance.
We used multiple immunofluorescence histochemistry to simultaneously detect CB1Rs and TRPV1s as eCB-sensing receptors (20, 21), as well as sn-1-diacylglycerol lipase α (DAGLα), monoacylglycerol lipase (MAGL), and α/β hydrolase domain-containing 6 serine hydrolase (ABHD6) (24) (defining minimal machinery for 2-AG metabolism), in mice at embryonic day (E) 16.5 (Fig. 1 A–D1), at birth, and in adulthood (Fig. S1 and Tables S1 and andS2).S2). We refrained from dissecting the anatomy of AEA signaling because its biosynthetic machinery is incompletely understood (25), precluding reliable microscale reconstruction. β, but not α, cells were CB1R-positive (+) at all developmental stages (P < 0.003 for all ranked data β vs. α cells) (Fig. 1 A–A2 and E and Fig. S1 A–B2). In contrast, TRPV1 was present in both β and α cells at similar levels during pancreas development, as well as at maturity (Fig. 1 B–B2 and F and Fig. S1 C–E2).
We found DAGLα preferentially expressed in α cells of the pancreatic primordium at E16.5 (P < 0.05 vs. β cells) (Fig. 1 C–C2 and G), with its levels up-regulated in β cells at birth and throughout adulthood (Fig. S1 F–G2). MAGL expression was localized to α cells at all ages tested (Fig. 1 D and H and Fig. S1 H, H1, and J–J3). Conversely, ABHD6 immunoreactivity in β cells exceeded that seen in α cells at all ages (Fig. 1 D1 and I and Fig. S1 I, I1, and K–K3). These data suggest that 2-AG signaling is paracrine first, with α cells being the source and β cells the sensors of 2-AG action (Fig. S1 L and M). These results are compatible with the hypothesis that intercellular signaling modulates cell identity, mass, and migration during islet formation. In turn, mixed autocrine/paracrine signaling might be favored for cell survival.
Laboratory rodents present core-mantle islet morphology with an α cell mantle enwrapping its β cell-dense core (18). In contrast, the organization of α and β cells in adult human pancreatic islets is remarkably different (Fig. 1J and SI Discussion) because these cell types remain predominantly interspersed (19). This cellular configuration might be favored upon metabolic challenge because it could allow the improved coordination of insulin and glucagon release (26). Nevertheless, CB1R and DAGLα distribution seem evolutionarily conserved and unrelated to cell positioning, inasmuch as CB1R and DAGLα strictly preferred β and α cells (Fig. 1 K–L2), respectively, even if at variable levels in the human biopsies available to us (SI Discussion).
Genetic Disruption of 2-AG Signaling Impairs Cell Segregation in Pancreatic Islets.
If 2-AG signaling is significant in modulating pancreatic islet microstructure (that is, either the size of α and β cell pools or their spatial positions), then genetic disruption of 2-AG metabolism might impose permanent phenotypes. We have previously found that MAGL-mediated degradation is rate-limiting for fetal 2-AG signaling in brain (23). Therefore, we first tested the analogous hypothesis by determining pancreatic islet morphology in adult MAGL−/− mice. Although the size of pancreatic islets remained unchanged (Fig. 2 A–B), MAGL−/− mice showed significantly increased α cell numbers per islet [47 ± 14 (MAGL−/−) vs. 22 ± 12 (MAGL+/+) cells per section; P< 0.01] (Fig. 2C). Because the number of β cells was unchanged (Fig. 2C1), the calculated α/β cell ratio increased [0.78 ± 0.14 (MAGL−/−) vs. 0.29 ± 0.10 (MAGL+/+), P < 0.001] (Fig. 2C2). Significantly, α cells were scattered within islet cores in MAGL−/− mice [0.34 ± 0.07 (MGL−/−) vs. 0.14 ± 0.13 (MAGL+/+), P < 0.01, α cells in core per all α cells] (Fig. 2D).
MAGL ablation-induced increased α cell mass and displacement can be due to the transdifferentiation, altered migration, survival, or proliferation of α cell precursors. Therefore, we mapped, in MAGL−/− islets, the expression of Pdx1 and MafA transcription factors that define β cell identity (27) and excluded their presence in glucagon+ α cells (Fig. 2 A and A1, Insets and Fig. S2 A–B2). This observation argues against transdifferentiation as a mechanism eliciting altered islet architecture in MAGL−/− mice. Because eCBs modulate cell migration and survival in the nervous system, adipose tissue, muscle, and immune system (13, 22, 28–32), we instead favor that altered cell migration could, at least in part, contribute to incomplete cell segregation in MAGL−/−mice (Fig. S3). Loss of CB1R function due to receptor desensitization, characterized as CB1R pools being present intracellularly in MAGL−/− pancreas (Fig. S2 C and D), supports this hypothesis. Furthermore, the presence of the small GTPase Rac1 (Fig. S4 A–C1) and the microtubule-associated protein doublecortin (Fig. S4 D–I1) in β and α cells (33) presents downstream targets linked to cytoskeletal reorganization during cell migration (34, 35). These findings are significant because both Rac1 and doublecortin expressions remain cell type-specific (Fig. S4 D–I) and endure into adulthood.
2-AG Signaling Modulates β Cell Survival in Pancreatic Pseudobodies in Vitro.
Next, we tested 2-AG’s effects on cell aggregation and survival in a model of ordered two-cell clustering (36) amenable to the in vitro chemical probing of signal transduction. αTC1-6 (“α-like”) and INS-1E (“β-like”) cells were mixed in suspension, producing ordered pseudobodies (“pseudoislets”), which resemble the murine pancreatic islet with an αTC1-6 cell mantle encapsulating a core made up of INS-1E cells (Movie S1). Although αTC1-6 and INS-1E cells are of mouse and rat origins, respectively, and their immortalized nature cautions data interpretation, they are still amenable for our purposes because (i) αTC1-6 cells express DAGLα but not CB1Rs, thus resembling α cells in vivo (Fig. S5); (ii) αTC1-6 cells contain and release significantly more 2-AG and AEA than INS-1E, recapitulating in vivo data on α cells being the primary source of eCBs (Fig. S5 E and E1); (iii) neither cell type transdifferentiates when mixed in vitro (Fig. 2 E–E3); (iv) αTC1-6 cells form spatially ordered 3D-aggregates with INS-1E cells restricted to the core of the pseudoislets (Fig. S6 A and B); (v) both cell lines express vinculin, suggestive of adhesion signaling (Fig. S6C) (37); and (vi) both retain either glucagon+ or insulin+ and differential responsiveness to glucose when aggregated (Fig. S6 D and E).
Initially, we assayed whether pharmacological enhancement of 2-AG bioavailability (MAGL inhibition by JZL184, 200 nM) (38) or reduced 2-AG biosynthesis [DAGL inhibition by OMDM188 (100 nM)] (39) affected αTC1-6 and INS-1E interactions. JZL184 moderately, yet significantly, reduced pseudoislet size [6,335 ± 493 (JZL184) vs. 8,731 ± 1,997 µm2 (control); F(3,815) = 6.56, P < 0.05]. In contrast, OMDM188 robustly reduced cell clustering (3,976 ± 2,556 µm2, P < 0.001 vs. control) (Fig. 2 E2 and F). JZL184 significantly rescued the OMDM188-induced pseudoislet phenotype [5,697 ± 365 µm2 (JZL184 + OMDM188) vs. (OMDM188), P < 0.01] (Fig. 2 E3 and F).
We then determined whether either treatment modulates α or β cell numbers in individual pseudoislets in vitro. JZL184 did not affect the total number of αTC1-6 cells per pseudoislet (Fig. 2G). Nevertheless, JZL184 significantly increased αTC1-6 “mixing,” measured as αTC1-6 cells translocated to >10 μm depth from the surface in an OMDM188-sensitive manner [153 ± 35% (JZL184) vs. 100 ± 44% (control); F(3,36) = 3.25, P < 0.05] (Fig. 2G1).
Subsequently, we showed that JZL184 increased αTC1-6 cell proliferation, as measured by Ki67 cytochemistry (40) (Fig. 2I and Fig. S7 A–B3). For INS-1E cells, DAGL inhibition was detrimental and reduced their numbers [F(3,36) = 11.82, e.g., 21 ± 17 (OMDM188) vs. 99 ± 62 cells per cluster (control), P < 0.01] (Fig. 2H) by promoting apoptosis (Fig. 2J). These data suggest that 2-AG signaling is required for pseudoislet assembly and affects their size by limiting the survival of INS-1E cells. Moreover, augmentation of 2-AG signaling for ∼48 h impairs α/β-like cell segregation by misplacement of αTC1-6 cells, corroborating the MAGL−/− phenotype in vivo.
Pseudoislets under control conditions preserved glucose-stimulated insulin (high glucose) and glucagon (low glucose) secretion (Fig. S6 D and E). As such, JZL184 pretreatment improved both insulin and glucagon release, and significantly increased the insulin-to-glucagon ratio [0.30 ± 0.26 (JZL184) vs. 0.12 ± 0.05 ng/pg (control), F(3,8) = 4.12, P < 0.05 under peak insulin-permissive conditions] (Fig. S6F). This increase was due to the continuously elevated release of insulin upon JZL184 application. In contrast, OMDM188 inhibited glucagon release in 2.75 mM glucose. In sum, 2-AG–induced incomplete cell segregation correlated with improved hormone release.
Altered Adhesion Signaling in MAGL−/− Mice.
Tissue architecture relies on how neighboring cells adhere to one another, which is mediated by anchoring systems that link in transelements on partner cells and the extracellular matrix to the cell’s cytoskeleton (41). CB1R inhibition or desensitization disrupts adhesion signaling (2). Here, we tested whether the subcellular distribution of E-cadherin (42) is changed in pancreatic islets of MAGL−/− mice. By quantitative immunofluorescence microscopy, we show that E-cadherin immunoreactivity significantly decreased in adult MAGL−/− β cells [0.75 ± 0.19 (MAGL−/−) vs. 1.00 ± 0.29 fold change (MAGL+/+), P < 0.05] (Fig. 3 A–D), which was due to the loss of cytoplasmic E-cadherin [6.81 ± 5.44 (MAGL−/−) vs. 32.45 ± 21.14 a.u. (MAGL+/+), P < 0.01] (Fig. 3E). In contrast, E-cadherin levels were significantly increased in MAGL−/− α cells [1.65 ± 0.18 (MAGL−/−) vs. 1.00 ± 0.18 fold change (MAGL+/+), P < 0.05] (Fig. 3 C–D), particularly of membranous E-cadherin [91.28 ± 28.16 (MAGL−/−) vs. 43.93 ± 22.31 a.u. (MAGL+/+), P < 0.001] (Fig. 3F). These results suggest altered cell–cell contacts as a molecular correlate of incomplete cell segregation.
CB1Rs Control Cell Sorting in Developing Pancreatic Islets.
We hypothesize that eCBs act as positional signals for α and β cells (Fig. S3). However, receptor heterogeneity exists because CB1R can be coexpressed with TRPV1 receptors in β cells. Therefore, we sought to address the specific contribution of CB1Rs and TRPV1s to the determination of islet size and cell segregation.
Our above data suggest that 2-AG, which acts at CB1R but not TRPV1 receptors (43), possibly controls cell sorting in immature pancreatic islets. This notion prompted us to determine whether CB1R loss of function alters the microarchitecture of mouse pancreatic islets in vivo. The size of pancreatic islets from CB1R−/− mice remained unchanged (Fig. 4 A–B and Fig. S2 E–F2). However, we observed an increased number of α cells [38 ± 16 (CB1R−/−) vs. 25 ± 12 (CB1R+/+) cells per section; P < 0.01] (Fig. 4C). The number of β cells (Fig. 4C1) was unaltered, thus positively skewing the α/β cell ratio [0.48 ± 0.16 (CB1R−/−) vs. 0.33 ± 0.13 (CB1R+/+), P < 0.001] (Fig. 4C2). Moreover, α cells were scattered within the islet core [0.21 ± 0.07 (CB1R−/−) vs. 0.12 ± 0.08 (CB1R+/+), P < 0.001, α cells in core per all α cells] (Fig. 4D). These results, and our data from MAGL−/− mice, raise the possibility that disrupted 2-AG signaling impairs cell segregation either due to dysfunctional (CB1R−/−) or desensitized CB1Rs (MAGL−/− produce chronically and congenitally supraphysiological tissue 2-AG content, which internalizes CB1Rs in β cells) (Fig. S2 C and D) (7, 44).
Next, we took advantage of pseudoislets as an in vitro model and used O-2050, a neutral CB1R antagonist (100 nM) (2), to pharmacologically disrupt CB1R involvement in α/β cell sorting. O-2050 did not affect pseudoislet size (Fig. 4E and Fig. S8 A and A1), corroborating our in vivo results. Instead, O-2050 significantly increased the number of αTC1-6 cells in pseudoislets [103 ± 43 (O-2050) vs. 17 ± 10 cells per pseudoislet (control), F(3,37) = 27.04, P < 0.001] (Fig. 4F) and their mixing with β cells in pseudoislet cores (Fig. 4F1). Subsequently, we exposed pseudoislets to AEA, a mixed eCB/endovanilloid ligand that acts as an agonist at both CB1Rs and TRPV1s (43). AEA significantly increased pseudoislet diameter [13,789 ± 3,739 μm2 (AEA) vs. 8,731 ± 1,997 μm2 (control)] (Fig. 4E). AEA not only increased the number of αTC1-6 cells per pseudoislet (32 ± 19) (Fig. 4F) but also led to incomplete cell sorting (Fig. 4F1 and Fig. S8A2). Likewise, α and β cell mixing occurred when endogenously produced AEA was tested (URB597; 100 nM) (Fig. S8 B–B3). However, AEA did not affect INS-1E cell recruitment to pseudoislets. If dual AEA effects are due to simultaneous signaling at CB1R and TRPV1 receptors, then coapplied O-2050 can be expected to reveal TRPV1-selective changes. Accordingly, AEA+O-2050 led to the segregation of α- and β-like cells into separate pseudoislets (>40% of pseudoislets contained one predominant cell type) (Fig. S8A3). This AEA effect was robust enough to positively bias our size measurements, leading to reduced sizes of single cell type-containing pseudoislets [5,308 ± 360 (AEA + O-2050), F(3,669) = 6.17, P < 0.05]. Combining AEA and O-2050 reinstated control-equivalent αTC1-6 cell numbers per pseudoislet [18 ± 16 (AEA+O-2050), P < 0.01]. Lastly, we observed significant β cell loss upon coapplied AEA and O-2050 [141 ± 57 (AEA) vs. 58 ± 56 (AEA+O-2050), F(3,37) = 2.84, P < 0.05] (Fig. 4G).
αTC1-6 cells do not express appreciable levels of CB1Rs (Fig. S5). Consequently, O-2050 did not affect either the rate of their proliferation or apoptosis (Fig. 4H and Fig. S7 C–D3). AEA did not affect αTC1-6 cell proliferation or apoptosis either. In contrast, CB1R inhibition alone provoked the apoptosis of INS-1E cells [254 ± 26% (O-2050), F(3,10) = 23.12, P < 0.001], which was not antagonized by AEA coapplication (Fig. 4I and Fig. S7 D–D3). AEA alone also triggered the apoptosis of INS-1E cells. Neither O-2050 nor AEA affected INS-1E cell proliferation significantly. These data suggest mutually exclusive roles for CB1R and TRPV1 in determining cellular heterogeneity in pancreatic islets.
TRPV1 Controls Pseudoislet Size.
Both native α and β cells express TRPV1 receptors. Our data on AEA suggest that TRPV1 agonism can affect pancreatic islet size. This hypothesis is plausible also because Ca2+ signaling (partly through TRPV1) is involved in regulating cell motility and adhesion (42).
To dissect TRPV1 involvement in size determination, we analyzed the morphology of pancreatic islets from TRPV1−/− mice (Fig. 5 A and A1). Genetic ablation of TRPV1 significantly increased islet size [148 ± 43 (TRPV1−/−) vs. 118 ± 44 (TRPV1+/+) cells per section, P < 0.05] (Fig. 5B). The number of α cells did not change significantly. However, we found more β cells in islets from adult TRPV1−/− mice [113 ± 35 (TRPV1−/−) vs. 94 ± 30 (TRPV1+/+) cells per section, P < 0.05] (Fig. 5 C–C2). Notably, the relative peripheral position of α cells within the islets was not affected (Fig. 5D). Based on these results, we concluded that TRPV1 signaling could regulate the size of, but not cell segregation in pancreatic islets.
We confirmed that TRPV1 activation is adverse for pseudoislet formation by using capsaicin (300 nM), a TRPV1 agonist [3,128 ± 654 (capsaicin) vs. 8,731 ± 1,997 μm2 (control), F(3,184) = 26.67, P < 0.001] (Fig. 5E and Fig. S8 C and C1). Conversely, capsazepine (10 μM), a TRPV1 antagonist (45), significantly increased the size of the pseudoislets (38,423 ± 4,370, P < 0.001) (Fig. 5E and Fig. S8 C and C2). Next, we confirmed these data using AMG 9810, an alternative TRPV1 antagonist (Fig. S8 D–D3), which also occluded the effect of capsaicin. Moreover, AEA at the concentration tested did not affect capsazepine (or AMG 9810) effects on islet size [50,751 ± 9,943 μm2 (AEA + capsazepine), P < 0.001 vs. control] (Fig. 5E and Fig. S8 C and C3) but induced significant recruitment [61 ± 53 (AEA + capsazepine) vs. 17 ± 10 cells (control), F(3,36) = 3.61, P< 0.05] (Fig. 5F) and mixing of αTC1-6 cells [185 ± 91% (AEA + capsazepine) vs. 100 ± 44% (control), F(3,36) = 3.75, P < 0.05] (Fig. 5F1) in the pseudoislets. These data clearly dissociate CB1R vs. TRPV1 outcome in the determination of islet microarchitecture.
Because TRPV1 antagonism alone did not affect the number of αTC1-6 cells in enlarged pseudoislets, we reasoned that an increased contingent of β cells might be recruited (or their survival affected). As Fig. 5G shows, quantitative morphometry confirmed this hypothesis [26 ± 16 (capsaicin), 171 ± 45 (capsazepine) and 230 ± 113 (AEA + capsazepine) vs. both P < 0.01 vs. 99 ± 62 (control) cell number per section, all F(3,36) = 16.42, P < 0.01 vs. control].
Lastly, we tested whether changes in αTC1-6 and INS-1E cell numbers in the pseudoislets were related to their altered rate of proliferation and/or apoptosis. Capsaicin remained ineffective in both cell types (Fig. 5 H and I). Inhibition of TRPV1 signaling by capsazepine, however, significantly reduced the number of Ki67+ (proliferating) αTC1-6 cells [59 ± 40% (capsazepine) vs. 100 ± 15% (control), F(3,9) = 4.77, P < 0.05] (Fig. 5H and Fig. S7 E–E3) while being ineffective in altering the rate of αTC1-6 apoptosis (Fig. 5H and Fig. S7 F–F3). Notably, capsazepine also decreased INS-1E cell turnover [that is, a simultaneous decrease of the histological indices of apoptosis F(3,9)= 31.13 and proliferation F(3,9) = 21.43, both P < 0.05] (Fig. 5I and Fig. S7 E–F3). These data cumulatively suggest that TRPV1 is a key signaling node controlling pancreatic islet size.
Mixed Pancreatic Islets Favor Insulin over Glucagon Secretion.
Increased dietary intake of ω-3 fatty acids is generally accepted to promote leanness by increasing adaptive hormone release from the endocrine pancreas (46, 47). eCBs are derived from arachidonic acid (20:4), an ω-6 PUFA. Correspondingly, ω-3 PUFA-enriched diet during pregnancy lowers 2-AG and AEA levels in the fetus (48). ω-3 PUFA intake is also beneficial for insulin secretion and sensitivity (49–51). Here, we hypothesized that lowering eCB levels during pregnancy and lactation might be reflected in a mixed cell-pancreatic phenotype and improved hormonal responses to glucose in the offspring. We administered an ω3-PUFA–enriched diet to dams starting 3 months before pregnancy and through pregnancy and lactation (Fig. S9A). Offspring were weaned onto normal laboratory chow and analyzed when reaching adulthood.
Maternal feeding with an ω3-PUFA–enriched diet resulted in reduced AEA levels in the blood of offspring weaned from ω3-fed mothers [5.20 ± 2.96 (ω3-PUFAs) vs. 24.70 ± 5.06 pmol/mg of lipids (control), P < 0.01] (Fig. 6A). Dietary intake of ω3-PUFAs did not change the size of pancreatic islets (Fig. 6 B–C) or the absolute numbers of α or β cells (Fig. 6 C1 and C2). Interestingly, ω3-PUFAs significantly increased the number of α cells scattered in the islet core [0.28 ± 0.13 (ω3-PUFAs) vs. 0.11 ± 0.09 (control), P < 0.05] (Fig. 6C3). We then confirmed ω3-PUFA involvement by applying docosahexaenoic acid (DHA) (10 µM) to pseudoislets in vitro. DHA increased the size of pseudoislets [180 ± 55 (DHA) vs. 126 ± 29 (control), P < 0.05] (Fig. 6 D–D2), as well as increased the number of αTC1-6 cells within pseudoislet cores [163 ± 56% (DHA) vs. 100 ± 46% (control), P < 0.05] (Fig. 6D3), recapitulating eCB depletion.
By performing a glucose tolerance test (GTT), we show that blood glucose levels were lower in mice that were characterized by a “mixed” islet phenotype [185.7 ± 28.7 (ω3-PUFA) vs. 231.0 ± 37.5 mg/dL (control), P < 0.05, 30 min after glucose injection; area-under-curve, 85 ± 7% of controls] (Fig. 6E). In pancreatic islets isolated from these mice, insulin release in response to high glucose was unchanged (Fig. S9B). In contrast, glucagon secretion was significantly inhibited in ω3-PUFA–fed mice in response to 2.75 mM glucose (Fig. 6F). These changes, when analyzed as an insulin/glucagon ratio, led to a significant shift toward insulin signaling upon stimulation with 16.5 mM glucose [15 min: 0.15 ± 0.03 (ω3-PUFA) vs. 0.08 ± 0.01 ng/pg (control); 20 min: 0.29 ± 0.04 (ω3-PUFA) vs. 0.20 ± 0.05 ng/pg (control), both P < 0.05] (Fig. 6F1). These data suggest that restricting eCB signaling in utero through dietary modulation of precursor availability might be beneficial for pancreatic functions.
Lastly, we assessed whether morphological phenocopy of pancreatic islets from CB1R−/− mice produces similar functional outcomes. GTT showed slower decrease in blood glucose in CB1R−/−mice [218.9 ± 79.3 (CB1R−/−) vs. 145.9 ± 42.0 mg/dL (CB1R+/+), and 174.8 ± 70.1 (CB1R−/−) vs. 116.2 ± 21.1 mg/dL (CB1R+/+), 60 and 90 min after glucose injection respectively, both P < 0.05] (Fig. S9C). However, CB1R−/− mice were lean [17.7 ± 2.2 g (CB1R−/−) vs. 20.7 ± 2.3 g (CB1R+/+); P < 0.05] (8, 52), expressing lower adipose tissue and muscle to body weight ratio (Fig. S9 D and D1). These data, supported by an adequate response to insulin [50 ± 8% (CB1R−/−) vs. 54 ± 4% (CB1R+/+) of baseline glucose level 1 h upon insulin challenge], make insulin resistance in CB1R−/− mice unlikely. Moreover, we observed significant increase in insulin secretion from pancreatic islets isolated from CB1R−/− in response to 16.5 mM glucose [20 min, 2.36 ± 0.29 (CB1R−/−) vs. 1.37 ± 0.13 ng/mL per islet (CB1R+/+), P < 0.05] (Fig. S9E) with no change in glucagon secretion (Fig. S9E1). This difference manifested as an increased insulin/glucagon ratio upon stimulation with 16.5 mM glucose [15 min, 0.88 ± 0.07 (CB1R−/−) vs. 0.54 ± 0.11 ng/pg (CB1R+/+), P < 0.05] (Fig. S9E2). Acute CB1R antagonism reduces insulin release in adult (53). Perinatal CB1R (in-)activity might instead be beneficial by repositioning cell contingents to improve hormonal coupling. Together, these data suggest that the secretory responsiveness of α and β cells is significantly enhanced by the architectural heterogeneity of pancreatic islets.
DISCUSSION
The present study suggests that paracrine eCB signals are present early in pancreas development in vivo. Even though we are aware of potential limitations of constitutive (vs. inducible) knock-out models, their combination with in vitro pharmacology can sufficiently support the differential engagement of CB1R and/or TPRV1 receptors to determine the pool size and microtopology of α and β cells in pancreatic islets (Fig. S3). CB1R and TRPV1s are expressed during postnatal life, and the reconfiguration of pancreatic islets is an “on-demand” mechanism driven by metabolic challenges. Thus, tissue-derived and circulating 2-AG and AEA might bring about critically distinct islet phenotypes associated with or predisposing to metabolic hindrances or disease conditions.
Our secretion assays suggest that the microarchitecture of pancreatic islets is a primary determinant of coordinated insulin and glucagon secretion, with mixed islet phenotypes in rodents being superior to the regular “core-mantle” arrangements. This observation is significant because pancreatic islet morphology is evolutionarily varied (18), reflective of the lifestyle, energy expenditure, and body mass of vertebrate species. As such, mixed pancreatic islets are characteristic of humans and nonhuman primates (18, 19) and suggest an evolutionary selection toward an anatomical microstructure that supports the increased dynamics of hormonal responses, especially in the presence of nutrient abundance. In rodents, reorganization of the core-mantle morphology of pancreatic islets, often interpreted as inadequate, might in fact confer adaptation to metabolic or pathogenic challenges. Accordingly, mixed pancreatic islet phenotypes have been associated with both physiological (i.e., pregnancy) (19) and pathophysiological (i.e., obesity, diabetes) (54, 55) conditions impacting glucose sensing and hormone secretion. Thus, we identify eCBs as a signaling network whose ligand diversity in conjunction with the receptor repertoire expressed by α and β cells is poised to tune hormone responsiveness.
eCB signaling has been linked to the molecular control of insulin and glucagon release (2–4). Considering that both CB1R−/− and MAGL−/− mice are lean (7, 8), we suggest that their mixed pancreatic islets allow for improved short-range insulin signaling between α and β cells (“histoarchitectural gain of function”), which can serve as a feedback mechanism increasing the ratio of insulin to glucagon secretion under high glucose availability. This hypothesis is also supported by genetic data upon conditional insulin receptor knockout in α cells (56), which induces hyperglucagonemia and glucose intolerance. Nevertheless, islet morphologies of adult CB1R−/−and MAGL−/− mice might represent either the outcome of developmental mechanisms or use-dependent and transient postnatal reorganization. The latter arrangement is particularly relevant in MAGL−/− mice because MAGL deletion affects not only 2-AG contents but also arachidonate precursor pools for prostaglandin synthesis (57), diversifying phenotypic outcome and cellular composition (e.g., macrophage infiltration) (58). Our ω3-PUFA enrichment during embryonic and neonatal development and the ensuing incomplete α/β cell sorting phenotype suggest that reduced eCB bioavailability and signaling in utero (48) generate pancreatic islets mimicking CB1R−/− or MAGL−/− phenotypes. Implicating eCBs in these processes is appealing because mouse β cells lack GPR120, a “ω3-PUFA receptor” (59) (SI Discussion).
AEA and 2-AG exhibit remarkably different receptor specificities: Whereas both 2-AG and AEA bind CB1/CB2 cannabinoid receptors, only AEA activates TRPV1 channels (43, 60). We took advantage of this receptor selectivity when combining an in vitro model of ordered α- and β-cell aggregation with mouse genetics. This approach allowed us to dissect the ligand and receptor specificity of eCB signaling in relation to cell sorting and the control of pancreatic pseudoislet size. Thus, CB1R antagonism by O-2050 or indirectly by inhibiting DAGL-dependent 2-AG biosynthesis increased pseudoislet heterogeneity, identifying CB1Rs as a critical signaling node for regulating the spatial organization of α and β cells. Notably, OMDM188 eliminated β-cell recruitment to pseudoislets in vitro and enhanced β-cell death whereas both O-2050 and OMDM188 remained ineffective on α-cell proliferation or survival. These results are concordant with our receptor profiling of αTC1-6 and INS-1E cells (2) that suggested the lack of CB1R expression in αTC1-6 cells. Moreover, pancreatic islet morphology in MAGL−/− and CB1R−/−mice, where α cells venture extensively into the islet core, supports that both genetic deletion (CB1R−/−) and desensitization (MAGL-/) of CB1Rs (7, 44, 61) inhibit the acquiring of prototypic cellular topography in the murine endocrine pancreas, which we consider as gain of function.
Another key finding of the present report is that TRPV1 agonism decreases whereas antagonism and genetic ablation increase the size of (pseudo)islets (62) without affecting spatial arrangements of α and β cells. Our developmental profiling and in vitro models suggest that TRPV1 signaling in pancreatic islets is independent of pancreatic innervation and relies on AEA as endogenous ligand. Chemical probing of TRPV1s selectively affected β-cell recruitment, and TRPV1 antagonism slowed cell proliferation. Therefore, we hypothesize that TRPV1 signaling is particularly efficacious to regulate the formation of cell–cell contacts underpinning cell aggregation. This finding is not entirely unexpected because previous data implicate signaling at TRPV1s in the proliferation of neural precursors, adipocytes, smooth muscle cells, and keratinocytes (63–66). In addition, the mobilization and aggregation of α and β cells is Ca2+-dependent (42), consistent with a TRPV1-mediated Ca2+ influx in pancreatic β cells (45). Therefore, it is tempting to speculate that activation of ligand- or voltage-gated Ca2+-permeable channels other than TRPV1 would recapitulate this prototypic TRPV1 response and could represent druggable targets to ameliorate β-cell dysfunction in diabetes.
Reorganization of pancreatic islet architecture can result from cell transdifferentiation, altered proliferation, survival, or migration related to remodeled cell–cell contacts (27, 67). Transdifferentiation is unlikely to confound our experiments because both α and β cells retained their lineage identities as suggested by the selective presence of Pdx1 and MafA transcription factors in β cells, and glucagon expression restricted to α cells (27, 68). This observation suggests that an eCB-mediated multicomponent mechanism operates during islet development: (i) 2-AG–induced CB1R activity is a survival signal for β but not α cells and supports the increased α/β cell ratio in (pseudo)islets when CB1R are inhibited; (ii) eCBs control cell turnover of both cell types; this suggestion is not surprising because eCBs regulate the proliferation of neural progenitors, adipocytes, and myoblasts (22, 69, 70); and (iii) eCBs affect cell motility and adhesion signaling.
We suggest that 2-AG signals can reorganize subcellular E-cadherin localization, with opposite outcomes in α and β cells. We propose that increased E-cadherin expression in α cells dispersed in islet cores is for the ectopic anchoring of resident β cells (67), and allows in trans signaling for coordinated hormone secretion. Our E-cadherin data also support the hypothesis that active cell migration participates in mixed islet configuration (67). As such, CB1R activation by both eCBs and synthetic agonists promotes the long-distance migration of neuroblasts and endothelial cells (29, 30). Moreover, CB1Rs can signal through small GTPases, including RhoA and Rac1 in neurons (10), thus critically tuning cytoskeletal instability during neuronal morphogenesis and polarization. Our finding that Rac1 is expressed by both α and β cells and that Rac1 is crucial for pancreatic morphogenesis (35) identifies cell migration as a candidate mechanism for core-mantle cell sorting. This notion is further underscored by high doublecortin expression in α cells scattered in islet cores because doublecortin is a ubiquitous marker of cell motility during fetal organogenesis.
The endocrine pancreas is indispensable for adequately orchestrated insulin and glucagon release to maintain the body’s energy homeostasis (1, 71) and to protect it from noxious metabolic stress (72–74). Our knockout analysis focused on adult animals because both CB1R−/− and MAGL−/−mice are lean (7, 8, 52) and resistant to high-fat diet-induced obesity or diabetes (4, 7–9). Although unchanged basal blood glucose and insulin levels of CB1R−/− and MAGL−/− mice were reported under standard feeding conditions (7, 52), glucagon regulation—and particularly insulin/glucagon balance—in these mice remain unknown. We demonstrate slowed glucose clearance in young CB1R−/− mice, which is corroborated by recent data (9). Our mechanistic analysis suggests that this profile rather reflects the decreased glucose need of peripheral tissues than any metabolic impairment because (i) both fasting (baseline) and terminal glucose levels are unchanged, (ii) CB1R−/− mice exhibit reduced muscle and adipose tissue mass (8), (iii) CB1R−/− mice respond properly to insulin challenge, and (iv) in islets isolated from CB1R−/− mice when islet dissection precludes the influence of tissue-derived or environmental confounds on glucose utilization, an improved relationship of insulin and glucagon secretion was observed. Likewise, we found improved insulin vs. glucagon secretion from morphologically similar pancreatic islets isolated from mice subjected to ω3-PUFA enrichment during fetal development. These data cumulatively link α cells infiltrating the core of pancreatic islets to enhanced α/β cell interplay and hormonal responsiveness. Therefore, we formulate the hypothesis that the increased number of α and β cell contacts drives paracrine signaling in the endocrine pancreas (19, 73), possibly with insulin and glucagon secretion serving as a dual feedback mechanism for α and β cells, respectively (56, 75, 76).
In conclusion, our report identifies fundamental roles for eCBs acting at CB1R and TRPVs in determining cellular diversity, structural complexity, and life-long plasticity of the endocrine pancreas. We also highlight that maternal dietary choices during pregnancy can program fetal pancreas development by altering eCB bioavailability (48, 77), which prospectively determines the offspring’s sensitivity to metabolic stressors. This observation has clinicopathological significance by pointing to an increased risk of diabetes in either malnourished or obese mothers and underpins the clinical potential of tissue-selective regulation of eCB levels.
EXPERIMENTAL PROCEDURES
Formation of α/β Cell Clusters, “Pseudoislets,” in Vitro.
INS-1E and αTC1-6 were cocultured at a 2:1 ratio in hybrid medium (10 mL) in 125-mL Erlenmeyer flasks under continuous agitation on gyrating shakers (70 rpm) at 37 °C for 48 h (36). The microarchitecture of pseudoislets was analyzed by capturing serial orthogonal z-image stacks (2.5 μm optical thickness) by laser-scanning microscopy. The density and positions of αTC1-6 and INS1-E cells were calculated using ImageJ v1.45 with appropriate plug-ins. The core of each pseudoislet was defined as its volume >10 μm from the outer surface of the spherical structure.
Prenatal Exposure to ω3-Polyunsaturated Fatty Acid-Enriched Diet.
Female C57Bl6/N mice (6 wk of age) were continuously fed with an experimental diet enriched in ω3-polyunsaturated fatty acids (21.42 kJ/g; 35% fat including 23.9% ω3 PUFA in the total diet, 25% crude protein, 40% carbohydrate; Special Diets Services) for 3 months before mating and during pregnancy and lactation. Pups were weaned on P21 and reared on a standard laboratory formulation. Control animals were exposed to a standard chow (energy, 15.38 kJ/g; 10% fat, 20% crude protein, 70% carbohydrate; Special Diets Services). Bodyweight of the offspring was measured every other day. At 6 wk of age, animals (n = 6 per group) underwent a glucose tolerance test (GTT) (SI Experimental Procedures) followed by the isolation of their pancreatic islets.
Statistics.
Experiments were performed in triplicate unless stated otherwise. The Wilcoxon rank-sum test was used to analyze data on expression of the eCB system in pancreas during pre- and postnatal development after semiquantitative assessment with data presented as medians. Data from pharmacology experiments were analyzed using one-way ANOVA followed by pairwise comparisons. Student’s t test (independent group design) was used to statistically evaluate data on pancreatic islet morphology and architecture in transgenic mice and after ω3-PUFA feeding. Data are expressed as means ± SD unless stated otherwise. Fold changes represent the percentage change from the untreated (control) value in individual experiments. A P < 0.05 value was considered statistically significant. A 3D rendering was used to qualitatively highlight cumulative changes in “eCB tone” during pancreas development.
Experiments on live animals conformed to the 86/609/EEC directive and were approved by the regional authority (Stockholm Norra Djuretiska Nämnd; N512/12). Ethical approval for use of human samples (Dnr: 00–128, Dnr 2010–279) was obtained from the Committee of Ethics, Faculty of Medicine, Uppsala University with informed consent from the individuals.
SI DISCUSSION
Pancreas Morphology.
Cellular organization of the endocrine pancreas shows phylogenetic variations, as well as adaptation to extreme environments (e.g., in naked mole rats) (90). As such, the localization of α and β cells is significantly different between laboratory rodents and humans (see also Fig. 1). Instead of a typical core-mantle islet morphology with an α-cell mantle and β-cell dense core (18), β cells in human islets do not cluster tightly, allowing wider association and direct functional interplay with other endocrine cells. Primarily, α-to-β cell (or reverse) signaling might be more efficacious in humans. Thus, a more rapid and dynamic response pattern for interspersed α and β cells (also in CB1R−/− and MAGL−/− mice) might be viewed as gain of function, allowing better coordination of hormone secretion through signaling mechanisms affecting transcription, translation, or the mobilization of readily available vesicles. Our histochemical data on CB1R and DAGLα localized in the adult human pancreas suggest more widespread expression in both α and β cells. These data allude to translational restrictions with regard to our experimental data. However, and to the best of our knowledge, data on eCB signaling in the developing human endocrine pancreas are not available. Therefore, future studies on human-specific cellular arrangements, their developmental control, and overall impact on organ function are warranted. Yet the significance of eCB-mediated mechanisms regulating islet size, progenitor proliferation, and cell sorting/migration through CB1Rs and TRPV1s might be conserved, particularly because eCBs constitute a family of evolutionarily conserved signaling systems (91, 92). Even if phylogenetic variations occur as to how ligand gradients are produced or if paracrine or autocrine signaling pathways coexist or instead dominate, the molecular principles we show remain unchanged. Clearly, adult contexts can recruit additional cell types, such as macrophages, which can regulate (at least) β-cell fate and survival through CB1R-dependent cytokine release (93). Thus, many cellular modalities to the eCB rules we propose might exist with ultimate functional relevance to determining pancreatic cell heterogeneity and functional competence.
Docosahexaenoic Acid Metabolism.
The biosynthesis of eCBs and eCB-like bioactive small signal lipids involves multiple metabolic pathways (94), which often operate in parallel in individual cell lineages. Most recently, N-docsahexaenoylethanolamide (DEA; also termed “synaptamide”) was identified as a derivative of docosahexaenoic acid (DHA), an essential ω-3 fatty acid (95). Synaptamide is thus a structural analog of anandamide. It promotes neurite outgrowth, synapse formation, and synaptic transmission in cultured neurons (95, 96) in a cannabinoid receptor-independent fashion (97). Nevertheless, and at least in neural stem cells, synaptamide induces protein kinase A/cAMP response element binding protein phosphorylation (98). Notably, brain synaptamide contents were shown to closely follow the amount of dietary DHA intake. Therefore, it is tempting to speculate that a DHA-to-DEA conversion might also occur during pancreas development and promote cell proliferation and differentiation. Unfortunately, neither the organ system-wide distribution nor the molecular identity/developmental regulation of DEA biosynthesis machinery has been elucidated to date. Moreover, ω-3 fatty acids act through GPR120 (99). Even if GPR120 is not expressed by murine β cells (but δ cells) (59, 100), existing evidence suggests GPR120 expression in human β cells (100). Therefore, GPR120 alone might confer significant differences to the ω-3 fatty acid sensing and responsiveness of rodent vs. human islets. Overall, this line of research is appealing because it might lead to the identification of novel determinants of hormone secretion that can be exploited as druggable targets in obesity and diabetes.
CB1R Antagonism as a Therapeutic Opportunity to Modulate Hormone Secretion.
During the past 20 y, a series of CB1R antagonists have been developed and indicated to a broad spectrum of disorders. Most prominently, SR141716A (rimonabant) (101) was temporarily marketed for weight management and type 2 diabetes until its withdrawal due to serious neurological side effects. Ever since, pharmaceutical interest in developing CB1R antagonists for clinical use has diminished. As of today, peripherally restricted CB1R antagonists might carry significant appeal for further development (102) to modulate hormonal output from the endocrine pancreas. Such interventions might prove successful to treat obesity and its comorbidities, particularly cardiovascular diseases (58).
SI EXPERIMENTAL PROCEDURES
Cell Lines.
INS-1E (“β cell-like”) cells (2) were cultured in RPMI-1640 medium also containing glutamine (2 mM), glucose (11 mM), Hepes (10 mM), heat-inactivated FBS [5% (vol/vol)], sodium pyruvate (1 mM), β-mercaptoethanol (50 μM), penicillin (50 μg/mL), and streptomycin (100 μg/mL) at 37 °C. αTC1-6 (“α cell-like”) cells were cultured in DMEM containing glucose (16.6 mM), Hepes (15 mM), FBS [10% (vol/vol)], BSA (0.02%), nonessential amino acids (0.1 mM), and sodium bicarbonate (1.5 g/L) (all from Sigma). Cells were routinely subcultured in 24-well plates up to passage 120 and allowed to reach ∼80% confluence. For coculture experiments, INS-1E and αTC1-6 cells were grown in a 1:1 hybrid medium made up of RPMI-1640 and DMEM including the above supplements.
Isolation of Pancreatic Islets.
Islets were obtained after perfusion of the pancreas with HBSS (Invitrogen/Gibco) containing collagenase (type I, 0.33 mg/mL; Sigma) and Hepes (25 mM), followed by purification on Histopaque 1077 gradients (Sigma) (78). After repeated washes in HBSS containing 10% (vol/vol) FBS, isolated islets were maintained in RPMI-1640 medium supplemented as above in humidified atmosphere (5% CO2) at 37 °C overnight before measuring hormone release.
Assessment of Pancreas Morphology in Vivo.
CB1R, MAGL, DAGL, and ABHD6 distribution at E16.5, birth (P0), and adulthood (2–3 mo of age) in C57Bl6/N mice of both sexes (n≥ 3 animals per time point) was assessed by means of semiquantitative histochemistry using affinity-purified antibodies whose epitope specificity had been validated (2) (Table S2). Expression levels of target proteins in 10 pancreatic islets per animal were scored as follows: 0, negative; 1, weak expression; and 2, strong expression.
Next, we assessed the morphology of pancreatic islets in young adult CB1R−/− (79), MAGL−/−(44), and TRPV1−/− mice (80) (in ≥10 islets per animal; n = 3–4 animals per genotype, 2–3 mo of age) and WT littermates, including (i) the size (diameter, μm2), (ii) total cell number, (iii) α- and β-cell density (as a fraction of the total cell number), and (iv) α-cell misplacement {expressed as [α cells localized in islet core (>10 μm from the outer surface)]/[total number of α cells]} after simultaneous localization of insulin+ and glucagon+ cell contingents. Intensity of E-cadherin immunosignals was calculated using ImageJ v1.45 with appropriate plug-ins. Experiments on live animals conformed to the 86/609/EEC directive and were approved by the regional authority (Stockholm Norra Djuretiska Nämnd; N512/12).
Liquid Chromatography-Atmospheric Pressure Chemical Ionization-Mass Spectrometry.
Extraction, purification, and quantification of AEA from the blood of dams on standard chow and ω3-PUFA diet and their offspring, as well as αTC1-6 and INS-1E cells, followed published protocols (6, 81). After lipid extraction and prepurification on silica gel columns, AEA and 2-AG levels from three animals per condition or triplicates per group (in vitro) were analyzed by isotope dilution using liquid chromatography-atmospheric pressure chemical ionization-mass spectrometry (Shimadzu LCMS-2020).
Molecular Pharmacology.
The effect of the following drugs, alone or in combination, on pseudoislet morphology, cell proliferation, and apoptosis was assessed: N-arachidonoylethanolamine (AEA, mixed CB1R/TRPV1 agonist; 10 µM), (2E)-N-(2,3-dihydro-1,4-benzodioxin-6-yl)-3-[4-(1,1-dimethylethyl)phenyl]-2-propenamide (AMG 9810, TRPV1 antagonist, 1 µM) (82), (E)-N-[(4-Hydroxy-3-methoxyphenyl)methyl]-8-methyl-6-nonenamide (capsaicin, TRPV1 agonist; 300 nM), N-[2-(4-Chlorophenyl)ethyl]-1,3,4,5-tetrahydro-7,8-dihydroxy-2H-2-benzazepine-2-carbothioamide (capsazepine, TRPV1 antagonist; 10 µM), 4-[Bis(1,3-benzodioxol-5-yl)hydroxymethyl]-1-piperidinecarboxylic acid 4-nitrophenyl ester (JZL 184, MAGL inhibitor; 200 nM), (6aR,10aR)-1-Hydroxy-3-(1-Methanesulfonylamino-4-hexyn-6-yl)-6a,7,10,10a-tetrahydro-6,6,9-trimethyl-6H-dibenzo[b,d] pyran (O-2050, CB1R antagonist; 100 nM), OMDM-188 (DAGL inhibitor, 100 nM) (39), and cyclohexylcarbamic acid 3′-(aminocarbonyl)-[1,1′-biphenyl]-3-yl ester [URB 597, fatty acid amide hydrolase (FAAH) inhibitor, 100 nM] (83).
RNA Isolation and Gene Expression Analysis in αTC1-6 Cells.
We have previously reported the combined mRNA and protein profiles of cannabinoid receptors and molecular components of eCB metabolism in INS-1E cells (2) to show the predominance of CB1Rs in this cell line. Here, total RNA was isolated from αTC1-6 cells, mouse spleen, cortex, and cerebellum (as tissue-specific positive controls) (79, 84, 85) using the RNeasy Mini Kit (Qiagen) followed by DNase digestion, and verifying RNA integrity on 2% (wt/vol) agarose gels (500 ng of RNA). cDNA was prepared by reverse transcription with random primers using the High-capacity cDNA Reverse Transcription Kit (Applied Biosystems) and PCR amplified by mouse-specific primer pairs (Table S1). PCR products were resolved on 2% (wt/vol) agarose gels and imaged on a ChemiDoc XRS+ system (Bio-Rad).
Western Blotting of αTC1-6 Cells.
αTC1-6 cells were washed, scraped, and extracted using a radioimmunoprecipitation assay buffer (86) containing 50 mM Tris (pH 7.4), 150 mM NaCl, 10 mM NaF, 5 mM EDTA (pH 8.0), 1 mM Na3VO4, 1 mM PMSF, 1% Triton X-100, 5 μg/μL pepstatin A, 10 μg/μL leupeptin, and 2 μg/μL aprotinine. Protein lysates were centrifuged at 12,000 × g at 4 °C for 10 min. Samples were denatured in Laemmli buffer and boiled at 95 °C for 5 min. Proteins were probed by loading 20-μg aliquots (8–10% gels) under denaturing conditions on SDS/PAGE, followed by wet transfer onto PVDF membranes (Millipore). Primary antibodies are listed in Table S2. After exposure to horseradish (HRP)-conjugated secondary antibodies (1:10,000, 2 h; Bio-Rad), target proteins were visualized by enhanced chemiluminescence (Pierce).
Tissue Preparation, Cytochemistry, and Histochemistry.
For immunocytochemistry, cells were plated on 12-mm coverslips coated with poly-d-lysine (0.001%), separately or in coculture (INS-1E and αTC1-6; 2:1 ratio), washed in 0.1 M phosphate buffer (PB), and immersion-fixed in 4% (wt/vol) paraformaldehyde (PFA) for 20 min. Pseudoislets were treated similarly, except that immersion fixation lasted 1 h.
Mice were transcardially perfused with ice-cold 0.1 M PB (pH 7.4), followed by 4% (wt/vol) PFA in PB (2 mL/min flow speed). Pancreata were rapidly dissected and postfixed in 4% (wt/vol) PFA overnight. E16.5 and P0 torsos were immersion-fixed in 4% (wt/vol) PFA overnight. After equilibrating in 30% (wt/vol) sucrose for 48–72 h, tissues were cryosectioned at a thickness of 14 μm (Leica CM1850) and thaw-mounted onto SuperFrost+ glass slides.
Human pancreatic tissues were obtained at surgery, fixed in buffered formalin for 6 h, immersed in 10% (wt/vol) sucrose in phosphate buffer, and cut at a thickness of 14 µm on a cryostat microtome (Microm). Ethical approval (Dnr: 00–128, Dnr 2010–279) was obtained from the Committee of Ethics, Faculty of Medicine, Uppsala University with informed consent from the individuals.
After rinsing in 0.1 M PB, specimens were exposed to a blocking solution composed of 0.1 M PB, 10% (vol/vol) normal donkey serum, 5% (wt/vol) BSA, and 0.3% Triton X-100 for 3 h followed by overnight incubation with select combinations of primary antibodies (Table S2). Carbocyanine 2-, 3-, or 5-conjugated secondary antibodies (1:300; Jackson) were applied in 0.1 M PB supplemented with 2% (wt/vol) BSA (2 h, 20–22 °C). Nuclei were routinely counterstained by Hoechst 33342 (1:20,000; Sigma).
Cells, pseudoislets, and pancreas sections were imaged on a Zeiss LSM700 laser-scanning microscope equipped with a 40×/1.4 water immersion objective and 1.0× to 2.5× optical zoom. Images were acquired in the ZEN2010 software package. Multipanel images were assembled in CorelDraw ×5 (Corel Corp.). Alternatively, 3D images of pseudoislets made up of αTC1-6 and INSE-1E cells were captured on a Zeiss Lightsheet Z.1 microscope (Zeiss) at 20× primary magnification with appropriate excitation (ex.) and emission (em.) settings (ex. 405 nm/em. band-pass 460–500 nm; ex. 488 nm/em. band-pass 505–545 nm; ex. 561 nm/em. long-pass 585 nm) for maximal signal separation. Image rendering and reconstruction were aided by Imaris (Bitplane).
Insulin and Glucagon Secretion ELISA.
Pseudoislets or pancreatic islets were incubated in drug-free modified Krebs–Ringer bicarbonate Hepes buffer (KRBHB) containing 135 mM NaCl, 3.6 mM KCl, 1.5 mM CaCl2, 0.5 mM MgCl2, 1.5 mM NaH2PO4, 5 mM NaHCO3, 10 mM Hepes (pH 7.4), and 0.1% BSA supplemented with 2.75 mM glucose for 1 h. Subsequently, samples were transferred to Eppendorf tubes at a density of 10 (pseudo)islets per tube. Next, pseudoislets or islets were incubated in modified KRBHB supplemented sequentially with 2.75 mM, 16.5 mM, and 0.1 mM glucose (each for 10 min). Samples were taken at 5-min intervals and stored at −20 °C until processing. Insulin levels were measured by a rat/mouse insulin ELISA Kit whereas glucagon was assessed by a chemiluminescent glucagon ELISA Kit (both from Millipore) (87, 88). Insulin and glucagon concentrations were normalized to the number of (pseudo)islets.
Glucose and Insulin Tolerance Tests.
Male mice (n = 6 per group) were fasted overnight (∼15 h) with water available ad libitum. Blood glucose level was measured from tail blood using a FreeStyle Lite glucose meter (Abbott Diabetes Care) before (baseline, 0 min), as well as 15, 30, 60, 90, and 120 min after i.p. glucose challenge [2 g/kg bodyweight (b.w.)]. Alternatively, male mice (n > 6 per group) were fasted for 6 h with water available ad libitum, and blood glucose was measured from tail blood before (baseline, 0 min) and 30 and 60 min after i.p. insulin injection (0.75 U/kg b.w.). Moreover, WT and CB1R−/− mice were dissected as described to determine visceral fat and hind limb muscle volumes (22). Data were expressed as mg/dL blood glucose (89).
ACKNOWLEDGMENTS
We thank Dr. Benjamin F. Cravatt, Dr. Z. Lele, and the National Institute on Drug Abuse for monoacylglycerol lipase (MAGL), TRPV1, and CB1R knockout mice, and Dr. C. Wollheim and Dr. H. Mulder for INS-1E and αTC1-6 cells, respectively. We thank the CLICK Imaging Facility at Karolinska Institutet for assistance with image processing. This work was supported by Foundation for Polish Science Grants TEAM/2010–5/2 (to A.D.) and MPD/2009/4 (to A.D.), National Science Center Grant UMO-2011/03/B/NZ4/03055 (to A.D.), National Science Centre Grant UMO-2013/10/E/NZ3/00670 (to A.D.), National Institutes of Health Grants DA023214 (to T.H.), DA011322 (to K. Mackie), and DA021696 (to K. Mackie), the Swedish Research Council (T.H. and T.G.M.H.), Hjärnfonden (T.H.), the Petrus and Augusta Hedlunds Foundation (T.H. and T.G.M.H.), and the Novo Nordisk Foundation (T.H. and T.G.M.H.). The Visby Scholarship Program supported K. Malenczyk’s studies in Sweden. D.C. is the recipient of a Karolinska Institutet-National Institutes of Health “Training Program in Neuroscience” grant.
FOOTNOTES
The authors declare no conflict of interest.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1519040112/-/DCSupplemental.
