1. Introduction
Diabetes is one of the most abundantly spread metabolic diseases worldwide and one of the leading causes of death, remaining without effective treatment methods. The antidiabetic monotherapies are still deficient in maintaining long-term glycemic regulation and are coupled with side effects [
1]. Therefore, there is a constant need for the development of alternative medication [
2]. On the other hand, selecting appropriate cell models for in vitro development of pancreatic islet model which would resemble natural islets of Langerhans more closely, is also challenging. The majority of research completed to date has used rodent pancreatic β cell lines due to their immortality and stimulus-induced insulin-secretion [
3]. Murine (e.g., MIN6, NIT-1, βTC, and βHC clones) and rat (e.g., RIN, INS-1) insulinoma cells are most often applied [
3,
4,
5,
6,
7]. However, several investigated β cell lines present a limited ability to produce insulin (e.g., some subclones of βTC and βHC) regardless of the applied method of cell line generation [
8,
9,
10]. For example, the oncogenic SV40 large T antigen was utilized to generate βTC clones demonstrating little glucose responsiveness [
8] as well as MIN6 β cells displaying near-normal glucose-stimulated insulin secretion (GSIS) ratio corresponding to properties of healthy functioning β cells [
11]. MIN6 is one of the most commonly used models of pancreatic β cells, as indicated by estimated publication numbers (over 1600 published research results). Nonetheless, the βTC subclones have served as research material in projects published around 140 times so far.
Although culturing β cells as monolayers have been the most frequently used, βTC clones and MIN6 cell lines can spontaneously form three dimensional aggregates (pseudoislets, PIs) that resemble primary islets of Langerhans in size and appearance. The main advantage of PIs is the maintenance of some degree of the cell to cell interactions present in vivo, which allows for improved basal insulin production, as well as GSIS compared to corresponding monolayers [
12,
13,
14,
15]. Spelios et al. conducted several projects co-culturing the βTC3 cell model of impaired insulin secretion and islet-derived endothelial cells. Free-floating PIs showed improved insulin production and enhanced glucose responsiveness [
7,
16].
Pseudoislets have been shown to release an enhanced insulin amount in response to a variety of stimuli, including physical or pharmacological depolarizing agents and nutrients [
12,
17,
18,
19]. However, to date, no attempts have been made to exploit the effect of PI formation on G protein-coupled receptors (GPCRs) expression and/or activation, although modulating the activity of GPCRs is an essential approach in modern drug discovery. An estimated 30–40% of FDA-approved drugs target or signal GPCRs [
20]. To the best of our knowledge, the only example of GPCRs studied so far in pseudoislet models is the glucagon-like peptide-1 receptor (GLP1R). Green et al. showed that the GLP1R gene was significantly upregulated in pseudoislets formed by human 1.1B4 β cells compared to monolayers [
21]. Similarly, higher levels of GLP1R accompanied with better responsiveness to exendin-4 were demonstrated in PIs formed by co-culture of the human EndoC-βH1 β-cell line and murine MS1 endothelial cells [
22]. Among pancreatic GPCRs, Free Fatty Acid Receptor 1 (FFAR1, also referred to as GPR40) and GPR119 are popular examples of receptor targets that have received recent attention in the field of diabetes therapeutics [
23,
24,
25,
26,
27,
28,
29]. In turn, McKillop et al. were one of the precursors to show that cannabinoid receptor–GPR55 also plays a role in direct modulation of insulin secretion [
30,
31]. Apart from cannabinoids and lysophosphatidylinositol, lysophosphatidylcholine (LPC) has also been evidenced as a GPR55 agonist [
32]. Very recent papers define GPR55 as a new anti-diabetic target [
33].
Our group has recently shown that in the MIN6 cell line all three GPCRs (GPR40, GPR55, and GPR119) can be efficiently activated by LPC, which is the most abundant lysophospholipid in human plasma [
28]. LPCs appeared as insulin secretagogues also in βTC3 cells [
29]. The limitation of natural LPCs is their instability when administered in vivo. We have overcome this barrier with the development of phosphorothioate LPC analogues significantly resistant towards enzymatic degradation due to two incorporated modifications: A methoxy group in
sn-2 position and a hydrophilic phosphate head modified by a sulfur atom [
34]. When natural and synthetic LPCs were used to stimulate insulin production in different cell models (MIN6 and βTC3), it was found that different β cells preferred LPCs bearing fatty acid chains of varying lengths [
28,
29]. Considering these differences, the work presented hereby attempts to examine the ability of various LPC agonists of GPR40, GPR55, and GPR119 receptors to potentiate GSIS in co-cultures of MIN6 and βTC3 with the MS1 pancreatic endothelial cell line. Several forms of LPC containing lauroyl (12:0), myristoyl (14:0), palmitoyl (16:0), stearoyl (18:0), and oleoyl (18:1) acyl chains were assessed, including phosphorothioate stabilized analogues. We also used the monolayer and PI models to investigate the expression pattern of GPR40, GPR55, and GPR119 receptors in these experimental systems.
4. Discussion
Native islets of Langerhans seem to be ‘the gold standard’ model for the investigation of β cell physiology and the development of new therapeutics. Unfortunately, fresh islets culture techniques can rarely be used for more than a few days, limiting their usefulness. As a consequence, rodent insulin-secreting cell lines are usually preferred and most widely used [
3]. Importantly, pseudoislet cellular configuration may provide an especially good alternative for isolated islets. The remarkable augmentation in insulin secretion compared with correspondent monolayers in response to secretagogues following homotypic PIs formation was observed with MIN6 [
12,
14,
15,
44,
45], 1.1B4 [
21], or EndoC-βH1 [
46,
47] cell lines. The enhanced cell-cell contact within PIs was connected with increased calcium signaling, increased E-cadherin levels, and connexin 36 regulating cellular communication and islet architecture [
12,
45,
48]. In addition, proteomics analysis revealed eleven highly enriched pathways in PIs, including those controlling glucose metabolism, cell interaction, and translational regulation. Moreover, during protein profiling, the expression level of Pdx1, a homeobox-containing transcription factor that plays a key role in pancreatic development, and Glut2, the major glucose transporter in β cells, was found to be very similar in MIN6 monolayer cultures and MIN6 homotypic PIs [
14]. PIs created with heterotypic co-cultures of insulinoma cell lines and accessory cells such as endothelial [
7,
16,
22], stellate [
49], neuroblastoma [
50], glucagon- and somatostatin secreting [
51] as well as GLP-1 releasing [
18] or mesenchymal stem cells [
52] have also been explored to provide the added benefit of restoring transformed insulinomas to resemble primary islets.
Hereby, we used PIs generated with MS1 endothelial cells. Such heterotypic aggregates were shown to be superior to monolayer cells and homotypic PIs in terms of improved
de novo deposition of key extracellular matrix components such as laminin and collagen IV. Additionally, higher expression and altered glycosylation patterns of integrin β1 characteristics for native pancreas were also detected [
7,
16]. Coculturing of MS1 endothelial cells with the EndoC-βH1 human β cell line affected the expression of key genes involved in the transport of glucose, β cell differentiation, glucose sensing, and insulin processing. Compared with monolayer cells, the MS1-EndoC-βH1 PI formation led to a decreased expression of glucagon mRNA, whereas somatostatin mRNA levels were unchanged. A slight increase in
PDX1 and
MAFA, a key regulator of GSIS, was also detected in PIs. Additionally, mRNA levels of proinsulin processing proteins (such as proprotein convertase subtilisin/Kexin types 1 and 2 or carboxypeptidase E) were similar between PIs and monolayer cultures and the expression of glucose transporter
GLUT1,
GLUT2, and
GLUT3, as well as the
ABCC8 gene, a component of the K-ATP channel, was increased in PIs. Higher levels of GLP1R were also demonstrated in PIs [
22].
However, scarce reports on the expression of G protein-coupled receptors involved in insulin release in PIs exist in the literature. Looking for new strategies for the prevention and treatment of diabetes, GPCRs have attracted attention as potential pharmacological targets, as they regulate pancreatic cell physiology, and have accessible druggable sites at the cell surface [
20]. GPR40, GPR55, and GPR119 are crucial targets responsible for the regulation of insulin secretion from β cells in a glucose-dependent manner [
6]. The GPR40 mRNA expression was shown to be around 17 times greater in the rat pancreatic islets than in the pancreas as a whole [
53]. A very similar ratio was observed in human pancreatic islets and adjacent pancreatic tissue [
54]. Importantly, chronic hyperglycaemia causes the abrogated expression of GPR40 and the downregulated release of insulin.
Gpr40 mRNA significantly decreased to 36.8% in islets of hyperglycemic
db/db mice, which correlates with impaired glucose sensing [
55]. Almost total abolishment of
Gpr40 expression in all endocrine cells of the pancreas was observed in diabetic Goto-Kakizaki rats compared with Wistar control islets. In addition, the culture of normal islets isolated from Wistar rats depleted Gpr40 protein expression in β cells being associated with almost total suppression of palmitate-stimulated insulin release [
56].
GPR40 mRNA expression was demonstrated to be significantly reduced in human diabetic islets with respect to non-diabetic islets [
57]. GPR40 was also found to be expressed in a large variety of pancreatic β cell lines [
53,
58]. Consistent with a previous report [
53], we found its higher level in MIN6 than in βTC3 cells. On the other hand,
GPR119 and
GPR40 mRNA levels were similar in isolated human and murine pancreatic islets [
54]. According to Ekberg et al., both these receptors GPR40 and GPR119 can act in synergy [
59]. Likewise, we detected that the level of
Gpr119 in MIN6 PI and MIN6-MS1 PI is comparable to that of
Gpr40. βTC3-MS1 PI expressed the highest level of
Gpr40 transcript. Taking into consideration
GPR55 gene expression in the heterotypic βTC3-MS1 and MIN6-MS1 pseudoislets, our results show for the first time a high level of GPR55 mRNA level comparable to those of GPR40 and GPR119. Previously, Romero-Zerbo et al. reported a high level of the GPR55 mRNA and protein in rat islets and β cells. However, they did not analyze the
GPR40 and
GPR119 gene expression in these tissues and cells [
60]. Similarly, Liu et al. have shown that MIN6 cells, mouse, and human islets express GPR55, again without any comparison to levels of GPR40 and GPR119 receptors [
61].
To ascertain the mechanisms responsible for significant changes in the PIs GPCR pattern, we decided to analyze the expression of receptors in MS1 endothelial cells. In 3D structures formed only by endothelial cells, we observed augmented expression of all transcripts with
Gpr40 mRNA reaching the highest abundance. However, expression of all GPCRs studied was not as high as in heterotypic βTC3-MS1 and MIN6-MS1 PIs. On the other hand, homotypic PIs formed solely by βTC3 cells, failed to increase
Gpr40,
Gpr55, and
Gpr119 to the level comparable to βTC3-MS1 PIs. Unfortunately, at this moment, we cannot answer whether the changes observed in the βTC3-MS1 co-culture result from altered expression in the β cells or the endothelial cells. However, this observation indicates that MS1 endothelial cells are crucial to restore the functionality of βTC3-MS1 PIs. Islet endothelial cells produce many molecules that impinge on the function and survival of β cells. It was shown that proliferating islet endothelial cells could produce substances that stimulate β cell proliferation, such as hepatocyte growth factor (HGF). This secretion could be induced by soluble signals from the islets, such as vascular endothelial growth factor-A and insulin [
62]. Other islet endothelial-derived factors that modulate β cell expansion include endothelin-1 [
63], connective tissue growth factor [
64], or thrombospondin-1 [
65]. Spelios et al., who conducted several projects co-culturing β cell models and islet-derived endothelial cells, showed that MS1, but not βTC3 cells, can produce laminin and collagen IV in vitro. Both proteins were found in and around the PIs, suggesting the continuous deposition of extracellular matrix (ECM) proteins during PI formation [
16]. Exposure of β cells to various laminin isoforms has been shown to increase insulin gene transcription and insulin release [
62]. In turn, integrin-laminin interactions were shown to affect insulin secretion [
66]. Whereas, expression of integrin β1 was detected in many β cell types, including βTC3, βTC3-MS1 PI formation leads to the maturation (glycosylation) of integrin β1 similar to that seen in native pancreas and mouse islets [
7]. In turn, integrins, composed of noncovalently linked α and β subunits, form clusters upon which cytoplasmic proteins gather to form a link between ECM and the cell cytoskeleton. The integrin affinity for ECM ligand influences its clustering. For instance, laminin 1 induced clustering of integrins and cytoskeletal-associated molecules into thick, short aggregates, while on laminin 5 these elements were clustered into thin, discontinuous-line structures [
67]. In light of this data, the studies shown by Waldeck-Weiermair et al. [
68] seems to be extremely interesting. They found that in human endothelial cells upon anandamide/O1602 stimulation, GPR55 clusters with αvβ3 and α5β1 integrins, which is a prerequisite to transmitting signaling towards intracellular downstream targets such as phospholipase PLCγ that, in turn, instigates inositol 1,4,5-triphosphate-triggered intracellular Ca
2+ mobilization from the endoplasmic reticulum. Interestingly, under conditions of inactive (unclustered) integrins, anandamide binds to the cannabinoid 1 receptor (CB1R), resulting in G
i protein-mediated activation of spleen tyrosine kinase (Syk) that inhibits phosphoinositide 3-kinase (PI3K) that represents a key signaling protein in the transduction of GPR55-originated signaling. However, once integrins are clustered, Syk does not further inhibit GPR55-triggered signaling. Therefore, the presence of islet endothelial cells in co-culture with β cells supports not only the proper growth of the latter but may also influence GPCR signaling.
MIN6 pseudoislets were shown to possess increased expression of proteins involved in cell-to-cell communication, especially gap junction and tight junction such as E-cadherin and connexins [
14,
69]. Recent studies indicate an unprecedented role of GPR40 in facilitating tight junction assembly in airways epithelial cells via 5′ AMP-activated protein kinase (AMPK) activation [
70]. It was also reported that linoleic acid-activated GPR40 is responsible for increasing the connexin 43 level in the cell membrane of gastric epithelial cells via the Akt-dependent mechanism [
71]. On the other hand, GPR40 is indicated as the receptor influencing gene expression of almost one hundred genes. GPR40-overexpressing RIN-40 β cells treated with linoleic acid showed significantly altered expression of 93 genes associated with olfactory transduction, neuroactive ligand-receptor interaction, MAP kinase signaling pathway, cytokine-cytokine receptor interaction, and regulation of the actin cytoskeleton. More than 30% of the genes were associated with signal transduction and cell proliferation [
72]. Summing up the abovementioned observations, we hypothesize that GPR40 may possess the function of master receptor initiating not only FFA- and glucose-stimulated insulin secretion but also structural and functional changes in the cell membrane of pseudoislets-forming cells. It should be underlined that GPR40 is activated by a broad range of medium- to long-chain saturated and unsaturated fatty acids of chain lengths of more than 6 carbons. Miyamoto et al. [
64] have reported at least 16 free fatty acids activating the receptor. However, the list of GPR40 ligands is still increasing: Among them is a linoleic gut microbial metabolite (HYA or 10-hydroxy-
cis-12-octadecenoic acid), omega hydroxylated arachidonic acid metabolite (20-HETE or 20-hydroxyeicosatetraenoic acid), and lysophosphatidylcholines [
28,
73,
74].
Another important observation of the present study is the acyl chain dependent activity of native LPC and their phosphorothioate analogues as GPCR ligands influencing glucose-stimulated insulin secretion. We can hypothesize that different patterns of responses in βTC3 cells and βTC3-MS1 PI are connected with the expression levels of Gpr40, Gpr55, and Gpr119 since the formation of βTC3-MS1 PI shifts the expression pattern of GPCRs toward a MIN6 cell line model. Such interpretation, however, must be considered in the context of known limitations. Gpr40, Gpr55, and Gpr119 protein levels and localization would give a more appropriate approach to responsiveness in further studies. We also intend to compare the potential differential response of experimental models studied to agonists different from LPCs, namely selective ligands. Such research would enable us to perform a more detailed characterization of the functionality of the pseudoislets to accurately and undoubtedly disclose the mechanisms responsible for the observed differences between monolayer cultures and pseudoislets.
Additionally, it should be underlined that the native and modified LPC used in our studies have in their structure hydrophilic “head” and lipophilic “tail” which can interact with the cell membrane bilayer. Due to different lengths of fatty acid residues, LPC 12:0, LPC 14:0, and LPC 16:0 interact in a slightly different way with the cell membrane and, possibly, with the transmembrane region or even the intracellular face of the receptors. So far, few studies have been focused on the partitioning of lysophospholipids into membranes. The incorporation of LPCs into lipid bilayers is expected to organize with their polar PC-head groups close to the bilayer interface and their hydrocarbon chain buried into the hydrophobic core of the membrane. The longest acyl-chain LPC 16:0 (displaying the smallest acyl-chain mismatch to the model 1-palmitoyl-2-oleoyl-sn-glycerol-3-phosphatidylcholine (POPC)) gave the weakest perturbation per LPC molecule associated with the membrane. In contrast, the LPC 12:0, with the shortest acyl-chain (showing the most considerable acyl-chain mismatch), displayed the most substantial perturbation [
75]. However, membranes used in these studies consisted of only one phospholipid type and, in general, had a very simplified structure. Therefore, conclusions based on the results have some limitations. One must realize that the lysophospholipid interaction with cell membranes depends on their components, complex structure, and physiologic state. Our previous studies confirmed different ways of LPC incorporation into the lipid bilayer of βTC3 cells and even cytotoxic effects caused by LPC 16:0, while LPC 12:0 and LPC 14:0 with shorter acyl chains had no toxic effects [
76]. LPC concentration used in our studies with βTC3, MIN6, and psudoislets were lower than those showing cytotoxic effects. Therefore, we did not observe any toxic effect, but the only different level of GSIS in the presence of these LPCs, probably caused by a different way of their binding to membranes and the receptors.
In summary, we have compared some aspects of GPCR-related functional properties of βTC3 and MIN6 cell lines, grown either as adherent monolayers or forming pseudoislets with pancreatic MS1 endothelial cells. We show that the co-culture of MS1 with the chosen pancreatic β cell models brings about uniformization of Gpr40, Gpr55, and Gpr119 expression patterns and changes the sensitivity to treatment with their LPC ligands. It proves the imperfection of traditional monolayered β cell models and presents the possible GPCR-related mechanisms responsible for the impaired response. However, further studies taking into account other GPCRs, e.g., GLP1R, are necessary to demonstrate the improved functionality of PIs. Moreover, the observed varying efficiencies in GSIS potentiation between LPCs bearing various acyl chains could suggest their multiple applications.