Lipid Droplets’ Role in the Regulation of β-Cell Function and β-Cell Demise in Type 2 Diabetes

Abstract

During development of type 2 diabetes (T2D), excessive nutritional load is thought to expose pancreatic islets to toxic effects of lipids and reduce β-cell function and mass. However, lipids also play a positive role in cellular metabolism and function. Thus, proper trafficking of lipids is critical for β cells to maximize the beneficial effects of these molecules while preventing their toxic effects. Lipid droplets (LDs) are organelles that play an important role in the storage and trafficking of lipids. In this review, we summarize the discovery of LDs in pancreatic β cells, LD lifecycle, and the effect of LD catabolism on β-cell insulin secretion. We discuss factors affecting LD formation such as age, cell type, species, and nutrient availability. We then outline published studies targeting critical LD regulators, primarily in rat and human β-cell models, to understand the molecular effect of LD formation and degradation on β-cell function and health. Furthermore, based on the abnormal LD accumulation observed in human T2D islets, we discuss the possible role of LDs during the development of β-cell failure in T2D. Current knowledge indicates that proper formation and clearance of LDs are critical to normal insulin secretion, endoplasmic reticulum homeostasis, and mitochondrial integrity in β cells. However, it remains unclear whether LDs positively or negatively affect human β-cell demise in T2D. Thus, we discuss possible research directions to address the knowledge gap regarding the role of LDs in β-cell failure.

The rising incidence of type 2 diabetes (T2D) is attributed to increased exposure to nutritional load that cause insulin resistance, in part by impairing the integrity of insulin target tissues. While pancreatic β cells initially compensate for insulin resistance by increasing insulin secretion and cell mass, T2D develops when β cells fail to secret sufficient insulin to maintain glucose homeostasis (1). Primary islet tissue affected by T2D shows multiple functional and structural changes in glucose sensing (2), hormone secretion (3), islet transcription factor expression (4), secretory machinery protein production (5), β-cell mass, and islet amyloid deposition, among other changes (6). A possible trigger for the changes seen in T2D-affected islets is excessive nutritional exposure, which may come from circulating toxic lipids and increased local pancreatic fat accumulation (7, 8). However, some lipids including free fatty acids (FFAs), also serve as energy sources, building blocks for a biological membrane, and potentiators of glucose-stimulated insulin secretion (GSIS) (9). Thus, regulated trafficking of lipids into and out of lipid droplets (LDs) may aid β cells in maximizing the beneficial effects of lipids while preventing their toxicity. In this review, we focus on recent findings from rodent and human β-cell models to summarize how LD-associated FFA trafficking affects β-cell function. Furthermore, we discuss the possible role LDs play during the development of islet cell failure in T2D.

Lipid Droplets Are Formed in Pancreatic β Cells

What Are Lipid Droplets?

LDs are intracellular organelles present both in prokaryotic and eukaryotic cells, first identified microscopically in the 1880s as discussed in a recent review of LDs (10). LDs consist of a monolayer phospholipid membrane and a neutral lipid core composed of triglycerides (TGs), cholesterol ester, and retinol ester (Fig. 1A) (10, 11). LDs were viewed as an inert fat depot until the late 1980s, when investigations began to reveal the dynamic effect of LDs on cellular metabolism through work initially focused on the adipocytes (10). LDs are now recognized to play an integral part in lipid metabolism in a wide range of cells by allowing the flexible usage of FFAs (12, 13). The lipids stored in LDs can be released and used for fuel, signaling, and the synthesis of structural components at a time and place that fulfills cellular demands (14). LDs can also defend cells by sequestrating potentially toxic forms of lipid molecules (15). Moreover, LDs dynamically interact with the endoplasmic reticulum (ER), mitochondria, autophagosomes, lysosomes, and nuclei to coordinate lipid metabolism together (14). Thus, LDs are now recognized as important regulatory structures in adipocytes, myocytes, hepatocytes, and others.

The surface of mammalian LDs is typically decorated by members of the perilipin (PLIN) protein family (ie, PLIN1-5, Fig. 1A-B). PLIN composition varies between different cell types and can profoundly affect the behavior of LDs by altering LD size, lipid cargo composition (eg, diacylglycerol, TGs, cholesterol ester), and LD interacting proteins (eg, lipases, colipases, chaperons, transcription factors). PLINs may also directly mediate the tethering of LDs to other organelles (Table 1) (16). In addition to the PLIN family of proteins, the LD membrane houses a variety of proteins that support the diverse functions of LDs. Proteins that regulate lipid metabolism, vesicular trafficking, cell signaling, transcription, and interorganelle communication were demonstrated in a recent proteomic study of LDs in which a proximity labeling strategy was applied to detect LD-associated proteins (17).

Lipid Droplets Are Detected in Pancreatic β Cells

In a 1998 seminal work, Shimabukuro et al (18) proposed that lipid accumulation directly contributes to β-cell demise. Unger and colleagues (19) demonstrated that islet TG content of the diabetic ZDF rat increased 100-fold compared to the lean control. Since then, volumes of work have examined the regulation of lipid homeostasis in β cells under normal and pathological conditions. However, the role of LDs in islets has been untapped for many years, likely because of an inability to visualize LDs in mouse β cells by imaging technique, including electron microscopy (EM) (20-22). TG synthesis in mouse islets is nutritionally regulated, as a high-fat diet (HFD) and fasting in vivo have been shown to increase TGs along with the expression of 2 PLINs (PLIN2 and PLIN5) that are known to associate with TG-rich LDs (23-25). However, this increase in TGs has not been proven to cause visible LD accumulation in mouse islets despite the increase in PLIN expression (20, 23).

Islet LD research began to capture more interest recently when LDs were clearly demonstrated in rat and human adult islet β cells (20, 24, 26). Although the lipid composition of LDs in β cells has not been directly assessed, β-cell LDs appear to be enriched with TGs. BODIPY 558/568 C12 (BODIPY C12) is a fluorescent long-chain fatty acid (LCFA) analogue preferentially incorporated into TG-rich LDs (25, 27). Trevino et al showed that LDs in human β cells are enriched with BODIPY C12 and coated by 2 PLINs (PLIN2 and PLIN5) known to associate with TG-rich LDs (24, 25) (Fig. 1C). In addition, knockdown of adipose triglyceride lipase (ATGL), a key TG catabolic enzyme, has been shown to expand the size and number of LDs in human β cells, further supporting that β-cell LDs are enriched with TGs (21).

The size of LDs in mammalian cells varies widely, from 20 nm to more than 100 µm in diameter, depending on cell type and differentiation status (28). LDs in human β cells are small, with the majority being smaller than 1.2 µm (21). In comparison, mature white adipocytes that specialize in TG storage possess an extremely large unilocular LD occupying most of the cell volume. This unilocular LD is unique to mature adipocytes as immature adipocytes and nonadipocytes usually possess multiple smaller LDs per cell (29). The diameter of the human adipocyte is reported to be 80 to 120 µm on average and may reach up to 300 µm (30). Among nonadipocytes, hepatocytes have a high capacity for TG synthesis and are well studied for pathological accumulation of LDs, known as hepatosteatosis. LD size in hepatocytes is between human β cells and adipocytes, ranging from 60 nm to 5 µm (31). The composition of PLINs expressed in each cell is one likely factor that affects LD size. The expression of PLIN1 is highly specific to mature adipocytes and considered to play a key role in the formation of a large unilocular LD, while PLIN2 is the dominant PLIN in nonadipocytes, including hepatocytes and human β cells (7, 22).

Factors Associated With Lipid Droplet Accumulation in Human β Cells

The number and size of LDs in a given cell also change in response to extrinsic factors, such as stress and the availability of nutrients (12, 15, 29). Indeed, LD accumulation in rat islets and human β cell lines has been shown to increase because of exposure to FFAs and glucose (26, 32). Increased LD accumulation was also a prominent feature of nutritionally stressed and dysfunctional human β cells in vivo (20). In a study by Dai et al (20), human and mouse islets were transplanted into immunodeficient 3-month-old NSG SCID mice to compare their response to chronic nutritional stress. When the recipient mice were placed on a 60% HFD for 3 months after transplantation, transplanted mouse islet β cells showed the expected compensatory increase in proliferation and glucose-stimulated insulin secretion (GSIS). However, β cells in the transplanted human islets did not show a compensatory increase in proliferation or GSIS. Moreover, the expression levels of the key islet transcription factors (MAFB and NKX6.1) were compromised in transplanted human islets but not in mouse islets. Amyloid plaque is a marker associated with dysfunctional human T2D islets and is observed in some young adults with type 1 diabetes (6, 33). Increased deposition of amyloid plaques was seen in human islet transplant on HFD exposure but not in the mouse islet transplant (20). Another unique feature of transplanted human islets was the presence of LDs whose numbers increased on HFD exposure.

As the increase in LDs appears to be a feature in the dysfunctional, transplanted human islets, Tong et al (22) addressed whether LD accumulation is also increased in native T2D islets. When LDs were assessed by neutral lipid staining (BODIPY 493/503 [BODIPY 493]) and PLIN immunostaining in human pancreas, both BODIPY 493 and PLIN2 signals were significantly enriched in α and β cells of T2D islets in relation to surrounding exocrine acinar cells. The increase of β-cell PLIN2 immunoreactivity in T2D islets was also reported independently (34).

In addition to nutritional stress and T2D status, β-cell age and maturation levels appear to be factors associated with LD accumulation in human β cells. Tong et al (22) reported that the number of LDs detected by EM or BODIPY 493 within β cells was low in young human pancreatic donors (age < 10 y) and steadily increased to greater than 6% (LD area/islet area) in older adult donors. LDs have been found to accumulate in human stem cell derived insulin⁺ β-like cells (35) that were molecularly mature and glucose-responsive when assessed during culture in vitro and after exposure to an in vivo environment by transplantation into NSG mice. In contrast, LDs were absent in less mature, glucose nonresponsive insulin⁺ β-like cells at an earlier developmental stage (22).

Collectively, LD accumulation in human islets is associated with multiple conditions that also affect β-cell health and function, highlighting the importance of addressing a potential role of LDs in human islets. It also raises the key question of whether abnormal LD accumulation and/or function contribute to the dysfunction of transplanted human islets exposed to an HFD or islets from individuals with T2D.

Distinguishing Lipid Droplets From Lipofuscin

More than 60% of lipid-rich structures visualized by BODIPY493 in human β cells colocalized with PLIN2⁺ and PLIN3⁺ immunoreactivity, indicating that they are LDs (Fig. 1D) (22). The remaining neutral lipid staining that lacks PLINs likely represents lysosomal-derived lipofuscins (Fig. 2) (22, 34). First characterized by Cnop et al (36, 37) in human islets, lipofuscins are lipid-laden lysosome-derived structures that show immunoreactivity for cathepsin D and accumulate in postmitotic cells. Lipofuscins react to neutral lipid staining (eg, oil red O and BODIPY 493) and are identified by EM as a heavily pigmented crown structure surrounding a single- or multiple-lipid core. Lipofuscins are found in many different cell types, with abundance positively correlating with cell age (37, 38). In human islets, the number of lipofuscins is substantially lower in α cells compared to β cells (37). Theoretically, BODIPY 493-positive structures that contain lysosomal LAMP1, and not PLIN2 or PLIN3, are expected to be lipofuscins in human β cells (Fig. 2). However, it is plausible that lipofuscins contain lipid cores originating from LDs and delivered to lysosome through autophagy (lipophagy) (39). Unfortunately, detailed information regarding the contribution of LDs to the formation of lipofuscins is lacking and awaits future study.

What Is the Role of β-Cell Lipid Droplets in the Regulation of Insulin Secretion?

Glycerolipid/Free Fatty Acid Cycle and Insulin Secretion

As elegantly summarized by Prentki et al (9, 40), lipolysis of TGs has long been considered to generate metabolites that augment insulin secretion through a process termed the glycerolipid/free fatty acid (GL/FFA) cycle. The GL/FFA cycle starts from the incorporation of FFAs to GLs represented by TGs and is followed by the release of FFAs from TGs through lipolysis. LDs take the central stage of the GL/FFA cycle; newly synthesized TGs are primarily transferred to LDs and lipolysis occurs on LDs. While LD formation has not been studied in depth in β cells, the processes and key proteins involved in LD formation and lipolysis (GL/FFA cycle) are well conserved among mammalian cells and likely shared with β cells (41). Fig. 3 summarizes the current understanding of the process of LD formation and lipolysis, along with the known effects of key lipolytic metabolites on insulin secretory machinery.

Briefly, LD formation starts as TG accumulation within the ER membrane bilayer at the area marked by the highly conserved, multifunctional protein seipin (41). Diacylglycerol acyltransferase 1 (DGAT1) is an ER-resident protein that catalyzes the last step of TG formation in the ER (12). One function of seipin is to aid the induction of membrane curvature to accommodate the accumulation of synthesized TGs for the formation of a lens structure (see Fig. 3) (42). Seipin and proteins recruited by seipin, such as lipid droplet-associated factor 1 (LDAF1), facilitate further accumulation of TGs to form a lipid core. Then, a lipid core covered by a phospholipid monolayer eventually buds off as a nascent LD (41). Normally, LD formation at the ER membrane is a highly efficient process, with FFAs incorporated into TGs within seconds after entering the cells (14, 43). Glucose drives TG synthesis in β cells by blocking fatty acid oxidation (FAO) and directing FFAs from an energy substrate to TG synthesis (7). PLIN proteins are inserted into LDs soon after budding from the ER and play a key role in regulating lipolysis rate by modulating the access of neutral lipases to the LD surface (see Fig 1A, Fig. 3, and Table 1). After emerging from the ER, LDs can continue to grow by adding TGs using LD-associated enzymes including DGAT2 (12). Even though both DGAT1 and DGAT2 catalyze the last step of TG synthesis, the structural differences lead to their unique localization, with DGAT2 principally residing on the phospholipid monolayer of LD and DGAT1 mostly in the ER (12). Interestingly, the expression of DGAT2 but not DGAT1 was found to be upregulated in human pseudoislets after ATGL downregulation, implicating the role of DGAT2 in the expansion of the LD pool that occurs in ATGL-deficient human β cells (21). It currently remains unknown whether DGAT1 and/or 2 regulate LD levels in an age- or nutrition-dependent manner in human β cells.

Insulin granule biogenesis is one of the most important bilayer vesicle-trafficking events in β cells. It differs from LD biogenesis in the synthesis site of contents, trafficking of vesicles, and the structure of surface membranes. Proinsulin processing starts at the ER lumen followed by transport of proinsulin aggregates to the Golgi apparatus in a COPII (and likely also COPI) vesicle-dependent manner (44, 45). The trans-Golgi network plays a key role in the formation of insulin granules that eventually bud off as vesicles encased by a phospholipid bilayer. Insulin granules continue maturation before delivery to the plasma membrane for exocytosis in response to secretagogue stimulation (46).

How does LD degradation (ie, lipolysis) regulate insulin secretion? It is unlikely that FFAs released from LDs support GSIS through FAO, as FAO is suppressed by glucose and negatively correlates with insulin secretion (9). FFAs released from LDs are thought to augment insulin secretion by activating peroxisome proliferator-activated receptors-δ (PPAR δ) and promoting adenosine 5′-triphosphate production in the mitochondria (47). FFAs secreted locally may activate the FFAR1 cell surface receptor and potentiate GSIS (9). 1-monoacylglyceride (1-MAG) is also generated during lipolysis and activates MUNC13-1 in the SNARE complex to stimulate insulin granule exocytosis (48). Protein modification by FFAs such as palmitoylation (49) is another mechanism by which FFA might affect insulin secretion as discussed in later (21).

Adipose Triglyceride Lipase Regulates Glucose-Stimulated Lipolysis and Lipid Droplet Formation in β Cells

For the GL/FFA cycle to increase insulin secretion effectively, lipolysis must be active in glucose-stimulated β cells. Liu et al (21) showed that ATGL is the enzyme responsible for glucose-stimulated lipolysis in human β cells. ATGL downregulation impaired glucose- and KCl-stimulated insulin secretion in human pseudoislets and rat INS-1 cells. The reduction of insulin secretion in these models was accompanied by a decrease in the syntaxin1a (STX1a) vesicle docking protein without changes in MUNC18 and SNAP25 protein levels. Palmitoylation of STX1a by FFAs was reduced in ATGL-deficient cells, likely accelerating STX1a degradation. In addition to the activation of MUNC13-1 and PPARδ mentioned earlier, STX1a stability regulated by palmitoylation represents another example of how lipolytic metabolites support insulin secretion (see Fig. 3) (47, 48). Unlike human islets and INS-1cells, mouse islets fail to increase lipolysis in response to glucose, demonstrating the differences in the regulation of lipolysis between human, rat, and mouse islets (21). In general, lipolysis of LDs is regulated by multiple factors, including phosphorylation of PLIN1/5, recruitment of lipases to LDs, and interaction between lipases and lipase activators and inhibitors (eg, ABHD5, G0S2, HILPDA) (16, 39, 50). Currently, a mechanism by which glucose activates lipolysis in β cells has not been determined. Interestingly, glucose-stimulated lipolysis is impaired in human T2D islets (21), a defect that could disrupt the GL/FFA cycle, increase LD accumulation, and impair insulin secretion.

Is Lipid Droplet Accumulation in Aging and Diabetic β Cells a Defender or a Contributor of β-Cell Demise?

LD accumulation in adult human β cells is potentially important for insulin secretion by providing metabolic intermediates through regulated lipolytic processes (21). In contrast, the LD accumulation seen in diabetic and possibly aging β cells may reflect an imbalance in LD formation and catabolism (22). Here, we discuss how dysregulated LD accumulation may affect β cells.

Lipid Droplets Appear to Defend β Cells Under Glucolipotoxic Conditions

Exposure to saturated LCFA (C13-21) in the presence of high glucose levels causes cytotoxic stress in β-cell lines and cultured islets, a phenomenon termed glucolipotoxicity (51). Palmitic acid (PA, C16:0) is the most abundant LCFA in the human body (52), and its level is significantly elevated in the blood of individuals with obesity or metabolic syndromes (53, 54). PA impairs rodent β-cell viability and function by inducing oxidative, ER, and inflammatory stress pathways (7, 51). In addition, PA compromises insulin secretion and causes apoptosis in human islets (55, 56). In contrast, monounsaturated FAs (MUFAs), represented by oleic acid (OA), are well tolerated by β cells. Interestingly, the toxicity of FFAs in rat INS-1 cells and rat islets inversely correlates with their ability to be incorporated into TGs. OA promotes TG synthesis and LD formation, whereas PA is much less efficiently incorporated into TGs (57). The cytotoxic effects of PA are greatly alleviated by coincubation with OA, which might be due to improved sequestration of PA into LDs (57, 58). Stearoyl-CoA desaturase (SCD) is an enzyme that catalyzes double bond formation, and its activation increases the OA to PA ratio favoring TG synthesis and LD biogenesis (59). In fact, overexpression of SCD2 reduces, while downregulation/inhibition of SCD1 exacerbates, PA toxicity in mouse MIN6 cells and rat INS-1 cells (59-61).

Interestingly, the expression of SCD1 is high in the human EndoC-βH1 cell line, a cell line that shows efficient LD formation and resistance to PA-induced stress (62, 63). Downregulation of SCD1 has been shown to make EndoC-βH1 cells susceptible to inflammation and ER stress after PA exposure (64). Collectively, these results imply that glucolipotoxicity in β cells can be prevented/reduced by increasing TG synthesis and expanding the LD pool size. However, results obtained in vitro using a single FA species may not reproduce changes that islets undergo in vivo, as FFA milieu in vivo is composed of a complex mixture of FFAs. Also, FFA toxicity is not always diminished by the efficiency of FFAs to induce LD biogenesis; monounsaturated erucic acid (C22:1) is highly cytotoxic to EndoC-βH1/2 cells despite efficient promotion of LD formation (32, 65).

Possible Function of Stress-induced Lipid Droplets in β Cells

Multiple stress pathways are activated in β cells undergoing aging- or T2D-related cell demise (1, 66), appreciating that age alone increases the risk of T2D (67-69). Interestingly, the same stress conditions are reported to promote LD accumulation in many different cellular contexts (11). While a detailed description of the relationship between LDs and each stress pathway is beyond the scope of this review, LD buildup correlates with oxidative stress in cancer cells and hepatocytes (11). Moreover, ER stress induces LD accumulation in hepatocytes, which is attributed both to increased lipogenesis and reduced FAO activity (70, 71).

In general, the accumulation of LDs is considered to be important in neutralizing the stress response (15, 72). During Drosophila development, glial cells generate polyunsaturated FAs (PUFAs) at the plasma membrane by oxidative stress. Sequestration of generated PUFA into LDs is shown to protect the glial cells and adjacent neuronal stem cells from reactive oxygen species generated from free PUFA peroxidation (73). Promotion of FFA storage into LDs also appears to mitigate ER stress in the mouse liver (11). TGs stored in LDs provide a precursor for eicosanoid production to modulate the cellular response to inflammation (43). Healthy neurons typically contain very few LDs as they transfer lipids to surrounding microglia cells. Microglia cells incorporate transferred lipids into LDs, but eventually reach an LD-saturation threshold. This coincides with the timing when adjacent neurons start to accumulate LDs and develop signs of dysfunction (74). Thus, the transfer of lipids from neurons to LDs forming microglia cells appears to be a protective mechanism for neurons. Currently, the specific role(s) of LDs in human β cells under stress is still unclear, although targeting of PLIN protein production in β-cell models has begun to provide some insight (Table 2).

Perilipins Control β-Cell Lipid Droplet Levels

Perilipin 5 Augments Insulin Secretion and Protects β Cells Under Stress

PLIN5 is highly expressed in oxidative tissues and possesses features that promote mitochondrial activity and FAO (16). Protein kinase A (PKA)-dependent phosphorylation of PLIN5 increases LD lipolysis (16, 75). PLIN5 also promotes the association of LDs to the mitochondria and facilitates the transfer of FFAs released from lipolysis to the mitochondria (76, 77). PLIN5 can bind FAs and translocate to the nuclei to allosterically activate peroxisome proliferator-activated receptor gamma coactivator 1-a (PGC1a) (78). While PLIN5 is expressed at moderate levels both in human and mouse islets, levels in human islets are higher than in mouse islets (24).

Currently, functional information about PLIN5 in human β cells is very limited (7). However, PLIN5 actions have been investigated in overexpression experiments in rodent β-cell models (24). Trevino and colleagues (24) found that the lipolysis rate was higher on PLIN5 overexpression than PLIN2 overexpression in MIN6 cells when TG synthesis was increased to a similar extent by overexpression of PLIN2 and PLIN5. The role of PLIN5 in promoting lipolysis was further confirmed, as a cell-permeable 3′,5′-cyclic adenosine 5′-monophosphate (cAMP) analogue bromoadenosine (8Br-cAMP) increased lipolysis in PLIN5 overexpressing MIN6 cells. The change in lipolysis correlated with insulin secretion in MIN6 cells and mouse islets overexpressing PLIN2 or PLIN5. GSIS under acute PA and 8Br-cAMP exposure was increased in β-cell models overexpressing PLIN5, but not in those overexpressing PLIN2. Furthermore, GSIS in vivo was augmented by the glucagon-like peptide-1 (GLP-1) receptor agonist in mice with β-cell–specific PLIN5 overexpression. A physiological role of islet PLIN5 in vivo was implicated from the induction of islet PLIN5 expression by fasting in mice. Adipocytes release FFAs during fasting that are redistributed to the heart, liver, and other tissues where FFAs are stored in LDs (79, 80). In the study by Trevino et al (24), fasting induced a marked increase of PLIN5 expression and TG accumulation in mouse islets, implicating that PLIN5 upregulation during fasting allows β cells to store TGs in PLIN5-coated LDs that can release lipolytic metabolites on refeeding to facilitate insulin secretion.

Overexpression of PLIN5 is also reported to improve GSIS in INS-1 cells chronically exposed to PA as well as in HFD-fed mice (81, 82). Moreover, PLIN5 overexpression was found to prevent PA-induced apoptosis and ER stress in INS-1 cells (81). Improvement in viability and reduction in apoptosis after chronic exposure to PA in PLIN5 overexpressed INS-1 cells was shown to be associated with activation of the nuclear erythroid 2-related factor (Nrf2)-antioxidant response element pathway, which defends cells against oxidative stress (82). Although data in human β cells are lacking, it appears that PLIN5 stimulates GSIS and protects β cells from nutritional stress, at least in rodent models. It is necessary to remember that studies performed in mice did not specify if visible LDs are formed in β cells overexpressing PLIN5 (24, 81).

Currently, little is known about PLIN1, PLIN3, or PLIN4 function in β cells. According to human islet single-cell RNA sequencing data, PLIN1 and PLIN4 are expressed at very low levels relative to PLIN2, PLIN3, or PLIN5 (83). Overexpressing PLIN1 in INS-1 cells has been shown to reduce GSIS under regular culture conditions but prevent chronic PA-induced GSIS reduction (84). This could be another example of an increase in LD formation mediated by PLIN overexpression being protective to β cells.

Perilipin 2 Is Crucial to the Health and Activity of Pancreatic β Cells

PLIN2 is the most abundant PLIN in nonadipocytes, including pancreatic β cells (see Table 1) and promotes LD pool expansion by preventing lipolysis and lipophagy (16). PLIN2 expression levels closely correlate with cellular LDs and neutral lipid content, while its downregulation is sufficient to reduce LD accumulation (7, 32). PLIN2 protein levels show the most robust increase in response to FFA load relative to other PLIN family members in human islet cells (23, 24). Both PLIN2 and PLIN3 expression markedly increase in human EndoC-βH2 cells on cessation of cell proliferation and induction of glucose-responsive insulin secretion, indicating that the proliferation and maturation status of the β cell may regulate PLIN gene/protein expression (32).

PLIN2 appears to be crucial for insulin secretion and the prevention of β-cell stress. Mishra et al (85) studied the effect of PLIN2 downregulation on INS1 cells, human pseudoislets, and β-cell–specific PLIN2 knockout mice. Downregulating PLIN2 in INS-1 cells impaired GSIS under regular culture conditions and after chronic incubation with PA. In addition, reduction of PLIN2 levels impaired GSIS in human pseudoislets. In β-cell–specific PLIN2 knockout mice placed on an HFD, GSIS was compromised in vivo and in isolated islets, in which islet TG content was reduced. All 3 models of PLIN2-deficient β cells/islets showed reduced levels of the oxidative phosphorylation (OXPHOS) complex components and impaired mitochondrial integrity. This included reduction in oxygen consumption rate, an alteration in the acylcarnitine profile, decreased FAO levels, and increased mitochondrial fragmentation in PLIN2-deficient INS-1 cells. Metabolic labeling with BODIPY C12 showed that PLIN2 deficiency increased trafficking of FFAs into the mitochondria, possibly contributing to mitochondrial dysfunction and impaired insulin secretion. However, ER stress markers were not altered in these 3 models, nor were islet-enriched PDX1, MAFA, and NKX6-1 transcription factor levels in PLIN2-deficient human pseudoislets.

More recently, the EndoC-βH2 cell line was used to study how LD regulation might influence human β cells (32). In nonproliferative EndoC-βH2-Cre cells (86), PLIN2 knockdown reduced LD levels and impaired GSIS, whereas overexpression of PLIN2 increased LD levels and improved GSIS. RNA sequencing revealed that more genes were differentially expressed after PLIN2 downregulation (1479 genes increased and 493 genes reduced) than overexpression (237 genes increased and 48 genes reduced), which included activated levels of ER stressors (inositol requiring enzyme 1 A [IRE1 A] and sXbp1), non–β-cell hormones (somatostatin and glucagon), and progenitor markers (SOX9, FEV) associated with T2D islets. In contrast, overexpression of PLIN2 increased GSIS and effectors of insulin secretion, including exocytosis complex component 5 (EXOC5), potassium inwardly rectifying channel, subfamily J, member 11 (KCNJ11), STX1a, and sarcoplasmic reticulum calcium-ATPase 2b (SERCA2b). Treatment with tauroursodeoxycholic acid (TUDCA), an ER stress inhibitor, restored GSIS on PLIN2 deficiency in EndoC-βH2-Cre cells, while PLIN2 overexpression protected against erucic acid-induced toxicity. Mitochondrial gene expression was also compromised on downregulation of PLIN2 in EndoC-βH2-Cre cells, in agreement with the reduction of the OXPHOS complex reported in PLIN2-deficient human pseudoislets (32, 85). However, it is presently unclear why β-cell identity regulators and ER stressors are altered by PLIN2 downregulation in a human β-cell line context but not in human pseudoislets (32, 85). These moderate discrepancies could be due to differences in cellular context (eg, cell line vs primary islet cells), the collective effect from other LD-containing islet cell types (eg, α cells) on global deletion of PLIN2 in pseudoislets, and/or the difference in the duration of PLIN2 downregulation.

Although the aforementioned studies indicate that proper maintenance of PLIN2-coated LDs in human β cells supports cell health and function, results from some rodent β-cell models suggest that downregulation of PLIN2 has either neutral or protective effects against stress (see Table 2). One study reported that downregulation of PLIN1 or PLIN2 did not affect the viability of rat RIN-mM5F or INS-1 cells cultured with PA and OA (87). Another demonstrated that whole-body PLIN2 knockout prevented ER stress, preserved β-cell mass, and was protective against hyperglycemia in mice carrying the Akita insulin mutation, which causes severe ER stress and apoptosis in β cells (88). In the same study, ER stress induced by tunicamycin or PA was reduced in PLIN2-deficient mouse MIN6 cells and mouse islets. While it is presently difficult to explain the different results obtained on compromising PLIN2 levels in rodent models, all existing data strongly suggest that LD accumulation controls human β-cell function and integrity under both normal and pathological circumstances.

Conclusions and Future Directions

Our general understanding of the effect of LDs on multiple aspects of cellular activity has increased considerably in recent years. The importance of β-cell LDs has attracted little attention until recent studies reported their presence in human β cells and enrichment in T2D islets. Studies of ATGL and PLIN2 in human islet β-cell models strongly indicate that the formation and degradation of LDs are equally critical for β cells in maintaining normal insulin secretion, ER homeostasis, and mitochondrial integrity.

A number of critical questions regarding the regulation of α- and β-cell function by LDs remain. Why are LDs not visible in mouse islets even when PLIN protein levels and TG contents increase in response to lipid load? How does age regulate LD accumulation in human islets? How do LDs regulate adult human islet α- and β-cell activity? Which LD-associated proteins regulate the functions of LDs in human islet cells, and how? Does LD accumulation influence the functional and molecular heterogeneity of the human islet α- and β-cell populations? Why are LDs enriched in T2D islets; do they protect or accelerate β-cell demise?

We propose examining the effect of LDs on human islet cell health using models translatable to human pathophysiology. Embryonically derived islet cells and human pseudoislets represent relevant and tractable model systems, especially when combined with xenotransplantation, for investigating the effect of LDs on human islet cells in vivo under normal and HFD stress conditions.

Abbreviations

Abbreviations
 
• ATGL

  adipose triglyceride lipase

 
• cAMP

  3′,5′-cyclic adenosine 5′-monophosphate

 
• DGAT1

  diacylglycerol acyltransferase 1

 
• EM

  electron microscopy

 
• ER

  endoplasmic reticulum

 
• FAO

  fatty acid oxidation

 
• FFA

  free fatty acid

 
• GL

  glycerolipid

 
• GLP-1

  glucagon-like peptide 1

 
• GSIS

  glucose-stimulated insulin secretion

 
• HFD

  high-fat diet

 
• INS

  insulin

 
• LCFA

  long-chain fatty acid

 
• LD

  lipid droplet

 
• MUFA

  monounsaturated fatty acid

 
• Nrf2

  nuclear erythroid 2-related factor

 
• OA

  oleic acid

 
• OXPHOS

  oxidative phosphorylation

 
• PA

  palmitic acid

 
• PKA

  protein kinase A

 
• PLIN

  perilipin

 
• PUFA

  polyunsaturated fatty acid

 
• SCD

  stearoyl-CoA desaturase

 
• STX1a, syntaxin 1a; T2D

  type 2 diabetes

 
• TGs

  triglycerides

 
• TUDCA

  tauroursodeoxycholic acid

Acknowledgments

Human pancreatic islets provided by the National Institute of Diabetes and Digestive and Kidney Diseases–funded Integrated Islet Distribution Program (IIDP) at City of Hope (No. 2UC4DK098085) were used for this review. Fig. 1A and 3 were created with BioRender.com. Materials from (24) have been used for Fig. 1C and from (22) for Fig. 1D with the permission of the American Diabetes Association, which holds the copyright of these materials and all rights reserved.

Financial Support

This work was supported by the National Institutes of Health (grant Nos. R01-DK090490 to Y.I. and RO1 DK050203, RO1 DK090570, RO1 DK126482, RO1 DK065949 to R.S.), the Department of Veterans Affairs (grant No. I01 BX005107), the Fraternal Orders of Eagles Diabetes Research Center at the University of Iowa, the Mark Collie endowment for diabetes research, and a JDRF (No. 3-PDF-2019-738-A-N to X.T.).

Disclosures

The authors have nothing to disclose.

Data Availability

Some or all data sets generated during and/or analyzed during the present study are not publicly available but are available from the corresponding author on reasonable request.

References