Gene Signature of the Human Pancreatic ε Cell

Abstract

The ghrelin-producing ε cell represents the fifth endocrine cell type in human pancreatic islets. The abundance of ε cells in adult pancreas is extremely low, which has hampered the investigation on the molecular pathways regulating the development and the function of this cell type. In this study, we explored the molecular features defining the function of pancreatic ε cells isolated from adult nondiabetic donors using single-cell RNA sequencing technology. We focus on transcription factors, cell surface receptors, and genes involved in metabolic pathways that contribute to regulation of cellular function. Furthermore, the genes that separate ε cells from the other islet endocrine cell types are presented. This study expands prior knowledge about the genes important for ε cell functioning during development and provides a resource to interrogate the transcriptome of this rare human islet cell type.

Ghrelin is a hormone secreted by specialized cells in the stomach and intestine as well as from rare ε cells in the islets of Langerhans of the endocrine pancreas (1–5). As a ligand for the GH secretagogue receptor (GHSR), ghrelin is an important stimulator of GH synthesis and secretion (6, 7). Ghrelin also regulates energy homeostasis, body weight and composition, appetite, glucose control, and gastrointestinal motility (8–11). During development, the pancreatic ε cells account for ∼10% of the human islet cells and likely contribute to circulating ghrelin levels in the body (4, 5). In adult humans, the number of ε cells is reduced to <1% (4). Because of their scarcity in the adult state, it is unlikely that pancreatic ε cells contribute to the circulating ghrelin pool; however, it could be speculated that the locally produced ghrelin regulates islet function. In line with this, it has been shown that human δ cells and partly β cells express GHSR (12–14). Furthermore, ghrelin stimulates somatostatin secretion and suppresses glucose-stimulated insulin secretion (15, 16). The study of pancreatic ε cells in animal models is difficult because of extreme scarcity (e.g., in mice 2 weeks of age, few ε cells can be detected) (17, 18).

Ultrastructural studies have shed light on the features of the pancreatic ε cells. They have a round or ovoid shape with occasional cytoplasmic extensions. Their small and spherical granules vary in electron density and show some similarity to α-cell granules (7, 19). Single-cell studies reported a small number of ε cells (13, 20), but detailed information regarding the human pancreatic ε-cell transcriptome is not available. Instead, comparable knowledge has originated from purified gastric ghrelin cells or immortalized cell lines derived from ghrelinomas in the stomach [stomach-derived ghrelinoma (SG-1), mouse ghrelinoma 3 (MGN3-1)] and pancreas [pancreas-derived ghrelinoma (PG-1)] (21–27). In this study, we have obtained the transcriptome of adult human ε cells from 12 nondiabetic donors through a large-scale RNA sequencing effort of single human islet cells. We report on the genes enriched in ε cells when compared with the other endocrine islet cell types as well as the expression of transcription factors, cell surface receptors, and genes pertaining to metabolic pathways.

Materials and Methods

Human islets

Human islets were obtained (Prodo Laboratories) from 12 nondiabetic donors. Donor information and procedure for islet processing have been described previously (28, 29). Briefly, islet samples from each donor were processed independently as they became available. After receipt, islets were cultured overnight followed by enzymatic dissociation and filtering. Afterward, cells were immediately loaded on a Chromium Single Cell Instrument (10X Genomics) and sequenced. Their viability was assessed using trypan blue staining (91.2 ± 3.3% cell viability; n = 12). An online repository contains relevant donor information (30).

RNA fluorescence in situ hybridization of dissociated islet cells

Cytospin was used to place dissociated islet cells into slides. Afterward, cells were fixed in 10% formalin for 35 minutes, followed by permeabilization and hybridization with mRNA probes (GHRL, INS, GCG, PPY, and SST) per the manufacturer’s instructions (Advanced-Cell Diagnostics). Signal was amplified using a fluorescent kit and detected using a microscope slide scanner (Zeiss Axio-Scan.ZI). An RNA fluorescence in situ hybridization analysis module from HALO software (Indica Laboratories) was performed to determine the intensity of the signal followed by quantification analysis.

Single-cell RNA sequencing, read mapping, and ERCC Spike-In

Single cells suspended in PBS with 0.04% BSA were loaded on a Chromium Single Cell Instrument (10X Genomics). RNA sequencing libraries were prepared using Chromium Single Cell 3′ Library, Gel Beads & Multiplex Kit (10X Genomics). Paired-end sequencing was performed on Illumina NextSeq500 [Read 1 for unique molecular identifier (UMI) and cell barcode; Read 2 55-bp transcript read]. Cell Ranger Single-Cell Software Suite (10X Genomics) was used to perform sample demultiplexing, alignment, filtering, and UMI counting. The Human B37.3 Genome assembly and UCSC gene model were used for the alignment.

ERCC Exfold Spike-In Mix 2 from Ambion (ThermoFisher, catalog no. 4456739) was used at 1:10 dilution. Diluted ERCC Mix 2 (2.3 μL) was combined with Chromium Single Cell 3′ Library, Gel Beads & Multiplex Kit (10X Genomics) and loaded to Chromium Single Cell Instrument (10X Genomics). Gel beads in emulsion (GEM) were subsampled (1.3 μL) and then processed with 10X Genomics GEM-RT cleanup, amplification, and library construction protocols (10X Genomics).

Single-cell data analysis

Cell clustering analysis and cell type identification were conducted as described previously (28, 29). Briefly, cells were excluded if they fell in one of the following groups: (i) detected with >1 hormone gene; (ii) high mitochondrial gene ratio (>0.2) (31); (iii) detected <500 genes; or (iv) <3000 total UMI. Seurat package was used to cluster single islet cells and identify enriched genes in each cell type. The 11 ε cells were compared with α, β, δ, and PP cells separately using FindMarkers in Seurat. Enriched genes in ε cells compared with another cell type were identified by Bonferroni-adjusted P < 0.05, log-scale fold change >0.25, and detection rate in 11 ε cells >0.25.

Pathway analysis

Gene pathway annotated datasets were obtained for citrate cycle tricarboxylic acid (KEGG-hsa00020), fatty acid metabolism (KEGG-hsa01212), glycolysis and gluconeogenesis (KEGG-hsa00010), pentose phosphate pathway (KEGG-hsa00030), G protein–coupled receptors (GPCRs) (Hugo Gene Nomenclature Committee), and transcription factors (31).

Accession number

Single-cell sequencing data described in this study have been made available previously (28, 29) and can be found within the Gene Expression Omnibus database (accession no. GSE114297).

Results

Frequency of ε cells in adult human islets

Single-cell RNA sequencing analysis of 14,779 human islet endocrine cells revealed a population of 11 ε cells (frequency, 0.08%). The cells were identified by high expression of GHRL. Independent detection of GHRL⁺ cells using RNA fluorescence in situ hybridization revealed a comparable frequency in populations of single human islet cells (average frequency, 0.14%; 160 of 113,109 endocrine islet cells) (30).

We detected a total of 5445 unique genes within the 11 GHRL⁺ cells at ≥1 UMI, with an average of 1267 genes per cell (30). The total number of genes within the ε-cell population was comparable to the α- and β-cell populations that were also sequenced within this effort (28, 29). Down sampling to 11 cells in each of these populations resulted in detection of 5033 genes in α cells and 4655 genes in β cells. However, when the cell number is increased to 6000 to 7000 α or β cells, we are capable of detecting 17,000 to 18,000 genes, which is comparable to about 18,000 genes detected in whole islets (reads per kilobase per million mapped reads >0.1). The explanation for these differences relies in the low number of cells and low sensitivity of the single-cell RNA sequencing technology. The latter is exemplified in spike-in experiments (see “Materials and Methods”) revealing a gene detection rate of 10% (90% dropout). Consequently, the detection was biased toward higher abundance genes (30). These limitations should be considered when interpreting the dataset. Important implications are that (i) the data may not serve to deduce absolute abundance of a gene and (ii) the absence of expression of a gene should not be considered as a true negative result, especially for low abundant genes. Despite this limitation, the number of detected genes in our ε-cell population was still of considerable measure. This, in addition to our quality control measures, gave us confidence to move forward with the present analysis.

Enriched human ε-cell genes among endocrine islet cell types

We investigated what sets apart ε cells from the other pancreatic endocrine cell types. In this regard, we found 36 genes to be significantly enriched in ε cells (see “Materials and Methods”) when compared with the other endocrine cell populations (α, β, δ, and PP cells) (Fig. 1; detailed information regarding endocrine cell comparisons can be found in an online repository) (30). On average, each enriched gene showed significantly higher expression in ε cells than one or two other endocrine cell types. Among the genes with a more specific pattern of enrichment in ε cells was acyl-coenzyme A synthetase long chain family member 1 (ACSL1) (ε vs PP cell-adjusted P < 6.12E-04), which has been suggested to have a potential role in the processing of ghrelin. Ghrelin consists of 28 amino acids and can become modified with the addition of an n-octanoyl at its third serine residue. This modification is essential for its binding to the GHSR (6). Acsl1, an isoenzyme that converts long-chain fatty acids (LCFAs) into acyl-coenzyme A esters, was previously found to be expressed at high levels in MGN3-1 (32). Furthermore, results from this study suggested that incorporation of LCFAs, as part of the supply of octanoic acid for ghrelin acylation, was mediated in part by ACSL1 (32).

Figure 1.

Differentially expressed genes in human ε cells vs other endocrine cells (α, β, δ, and PP cells). Enriched genes in ε cells were determined using a Bonferroni-adjusted P value <0.05, log-scale fold change >0.25, and detection rate in 11 ε cells >0.25. Each row represents a gene; each column represents a distinct endocrine cell population. Gene expression is presented in each row as a cumulative percent of the population expressing the gene. The level of expression of each gene is presented as scaled expression.

Surprisingly, we found two ε cell–enriched genes known for their proteinase inhibitory capacity: serine peptidase inhibitor Kazal type 1 (SPINK1) (ε vs α cell-adjusted P < 2.67E-11; ε vs β cell-adjusted P < 3.97E-47; ε vs δ cell-adjusted P < 4.58E-18) and serpin family A member 1 (ε vs β cell-adjusted P < 4E-24). The first is a serine protease inhibitor that binds to erroneously activated trypsin, forming a stable inactive complex. In the islet, SPINK1 is expressed and packaged into acinar cell secretory granules. Its major role in the pancreas relates to the protection from premature trypsinogen activation, which can lead to damage of acinar cells and ultimately development of pancreatitis (33, 34). Serpin family A member 1, a gene encoding the protein α-1 antitrypsin not only inhibits trypsin and elastase proteases, but also has recently been found to inhibit caspase-3 activity (35). It is believed to be involved in tissue repair after injury, with proinflammatory cytokines enhancing its production and release. Administration of α-1 antitrypsin has resulted in islet allograft protection and survival, as well as in prevention of streptozotocin-induced diabetes in mice (36, 37). Similar to our data, immunohistochemistry analysis in human islets has shown this factor to be expressed in α and δ cells, but not in β cells. Unfortunately, its expression in ε cells, where we currently observe its highest enrichment, was, to our knowledge, not assessed (37).

Another highly enriched gene that encodes a defense peptide is defensin beta 1 (ε vs β cell-adjusted P < 2.30E-02; ε vs PP cell-adjusted P < 2.28E-02). This β-defensin protein has the ability to kill a wide variety of microorganisms including bacteria, virus, and yeast. Furthermore, it binds several receptors and enhances the immune activity. Interestingly, β-defensin has been shown to have anti- and proinflammatory actions, depending on the disease and pathogen exposure (38). Interestingly, we also detected enrichment of retinol-binding protein 4 (RBP4) (ε vs αcell-adjusted P < 1.82E-03; ε vs PP cell-adjusted P < 1.17E-03) as well as of transthyretin (TTR) (ε vs δ cell-adjusted P < 6.38E-04). Among the known functions of TTR are the transport of thyroid hormone and formation of a complex with retinol-binding protein that transports retinol in the bloodstream (39). Both RBP4 and TTR have been found to be highly expressed in enriched pools of gastric ghrelin cells, in the gastric cell line (SG-1), but more important in a ghrelinoma cell line of pancreatic origin (PG-1) (24). Last, transmembrane 4 L six family member 4 (TM4SF4), a member of the L6 domain tetraspanin family, was found to be enriched in ε cells (ε vs β cell-adjusted P < 1.87E-05), and the transmembrane 4 L six family member 5 (ε vs β cell-adjusted P < 8.87E-15; ε vs δ cell-adjusted P < 1.53E-02). Although to our knowledge no function is known regarding 4 L six family member 5 in ε cells, the function of TM4SF4 in pancreatic islets has been associated with cell migration and endocrine differentiation. In NK2 homeobox 2 (Nkx2.2)–null mice, in which there is an abundance of ghrelin cells, its expression was found to be significantly elevated. In this same study, knockdown of tm4sf4 in zebrafish resulted in a decrease in the ghrelin cell population, highlighting its potential importance for the ε-cell fate (40).

Thus, we found that genes differentially expressed in ε cells when compared with the other pancreatic endocrine populations span functions related but not limited to hormone processing, proteinase inhibition, cell migration, and antimicrobial activity. Although we have highlighted some of these genes, there are several others whose function in ε cells merit further investigation.

Pancreatic ε cells express relevant islet cell transcription factors

Prior studies in mice have given us insights into the regulatory networks influencing the development of ε cells (17, 18, 41–45). To complement this knowledge, we now report on 366 transcription factors expressed in human adult ε cells (30). Figure 2 shows the top 50 transcription factors ranked based on their average expression. Among them, we observed expression of known genes essential for the differentiation and maintenance of islet endocrine cells including aristaless-related homeobox, ISL LIM homeobox 1, neuronal differentiation 1, and NKX2.2 (46–50). Although not much is known regarding transcriptional regulation of human pancreatic ε cells, immunohistochemistry analysis of fetal samples is largely consistent with our findings because they have shown expression of ISL LIM homeobox 1 and NKX2.2 in this cell population (4). Interestingly, NKX2.2 has been found to activate the mouse and human ghrelin promoter that, together with our findings, points to its potential importance in regulating ghrelin expression in the mature state (51, 52). We also found members of the HES family to be expressed: hes family bHLH transcription factor (HES) 1, HES4, and HES6. To our surprise, HES4 was the only transcription factor found to be significantly enriched in ε cells (ε vs δ cell-adjusted P < 1.71E-02) (Fig. 1). Unfortunately, not much information exists on HES4 because it is expressed in humans only and not mice. However, a member of its family HES1 is a factor known to be involved in the regulation of pancreatic progenitor cell differentiation and maintenance (53). Last, we found aristaless-related homeobox to be detected in ε cells, a finding consistent with developmental mouse studies showing this factor to be expressed within this cell population (30, 54, 55). The function in ε cells of these and other transcription factors found in the current study is an area of biology that merits further investigation.

Figure 2.

Top 50 transcription factors expressed in human ε cells. Average gene expression was used to select the highest expressed transcription factors within the ε-cell population. Each row represents a gene; each column represents a distinct endocrine cell population. Gene expression is presented in each row as a cumulative percent of the population expressing the gene. The level of expression of each gene is presented as scaled expression.

Cell surface receptors present in human pancreatic ε cells

Ghrelin cells respond to nutrients, hormones, and other signaling molecules (21, 56, 57). To gain understanding of how ε cells sense and respond to changes in their environment, we examined the expression of cell surface proteins, including GPCRs (30). We found 24 GPCRs, whose ligands include: neuropeptides, metabolites, hormones, and neurotransmitters, among others (Fig. 3). Of the presently detected GPCRs, we found six reported to be expressed in mouse ghrelin cells of gastric origin: free fatty acid receptor 3 (FFAR3), FFAR4, gastric inhibitory polypeptide receptor (GIPR), KISS1 receptor (KISS1R), somatostatin receptor 1 (SSTR1) and somatostatin receptor 2 (SSTR2). Of these, three of them have been shown to have ghrelin secretion inhibitory functions (SSTR1, SSTR2, and FFAR4) and one with stimulatory effect (GIPR) (21). FFAR4, which recognizes LCFAs, showed higher expression in ε cells as well as in δ and β cells, whereas it was minimally expressed in α and PP cells. Interestingly, whereas in ghrelin cells of gastric origin, FFAR2 was found to be the most enriched short-chain fatty acid receptor (21), our data show instead higher enrichment of FFAR3, pointing to potential tissue differences in fatty acid sensing. Along these lines, G protein–coupled bile acid receptor 1 encoding a bile acid–sensitive receptor, also showed distinctive enrichment in this cell population. In rats, activation of this receptor has been shown to have stimulatory effects on glucagon-like peptide-1, peptide YY, and neurotensin secretion (58). Its significance in human pancreatic ghrelin secretion remains to be evaluated.

Figure 3.

GPCRs expressed in human ε cells. GPCRs are categorized based on their respective classes. Each row represents a gene; each column represents a distinct endocrine cell population. Gene expression is presented in each row as a cumulative percent of the population expressing the gene. The level of expression of each gene is presented as scaled expression.

Apart from GPCRs we also found components of the receptors for both type I and type II interferons: interferon-α and interferon-β receptor subunit 1 and interferon-γ receptor 2 (59) as well as a receptor known for its regulation of immunoglobulin G uptake (Fc fragment of immunoglobulin G receptor and transporter, (30, 60). Two receptors known for their modulation of plasminogen activation were also detected: plasminogen receptor with a C-terminal lysine and S100 calcium-binding protein A10. Interestingly, activation of plasminogen plays an important role in cellular migration by degrading extracellular matrixes (61, 62). Last, CD320 molecule, which encodes a receptor responsible for binding and mediating the uptake of trans-cobalamin saturated with vitamin B12, was also observed (63). Collectively, we found a variety of receptors with distinct cellular functions whose ligands include but were not limited to hormones, nutrients, lipids, and cytokines.

Primary metabolic pathways in human ε cells

Last, we interrogated the expression of factors involved in cellular metabolic pathways. Using annotated gene pathway datasets, we detected expression of genes involved in glycolysis and gluconeogenesis (26/61 annotated genes, 43% detected), tricarboxylic acid cycle (20/30 annotated genes, 67% detected), pentose phosphate pathway (14/27 annotated genes, 52% detected), and fatty acid metabolism (22/42 annotated genes, 52% detected) (Fig. 4). Although not all genes were expressed or detected within the first three pathways, there was a balanced representation of some of the main factors involved in each of them, suggesting their likely participation in regulating ε-cell function. Interestingly, of the members that compose the fatty acid metabolism pathway, ACSL1 was observed to be significantly enriched in ε cells (Fig. 1). Its expression level in this population is in concordance with comparable high expression levels found in MGN3-1 (32). Its function in fatty acid metabolism along with its selective enrichment in this cell population provide support for its further exploration regarding its contribution in ghrelin acylation.

Figure 4.

Metabolic pathways present in human ε cells. Expression of genes pertaining to distinct metabolic pathways is presented accordingly. Each row represents a gene; each column represents a distinct endocrine cell population. Gene expression is presented in each row as a cumulative percent of the population expressing the gene. The level of expression of each gene is presented as scaled expression. TCA, tricarboxylic acid.

Discussion

Epsilon cells compose a rare population in the adult pancreatic islet. In fact, human is the only species known to still have ghrelin-expressing cells in the adult state (7). This makes the use of animal models impossible, leaving high-throughput RNA sequencing as one of few options to gain unbiased insights into the function and regulation of the islet ε cell. One limitation of the study is the low number of cells. This, coupled with a 90% dropout rate, limits the number of detected transcripts, especially less abundant ones. Within this sequencing effort, we found that analyzing >6000 α or β cells provided us with enough resolution to overcome some of these limitations (28, 29). Unfortunately, having a frequency of 0.08% for ε cells would entail sequencing >10 million islet cells; given the limited access to human tissue, even enriching for this population would be an infeasible approach. Therefore, even with these limitations, the current study provides a resource to study the transcriptome of this rare cell population.

The function and significance of human pancreatic ε cells in the adult stage is still unclear. During gestation, they develop to form an almost-continuous layer around the islet, likely secreting ghrelin into the circulation. Afterward, their numbers decrease, and local actions might be their main function. It could be speculated that ε cells in the adult stage are responsible for regulating intraislet hormone concentration, which would entail other pancreatic endocrine cells expressing Ghsr. In this regard, studies in mouse islets and cell lines have found expression of Ghsr in α, β, and δ cells (14, 64–66). In humans, recent islet single cell RNA sequencing studies have shown an enrichment of GHSR in δ cells (12, 13), a finding we confirm in our dataset. Because islet endocrine cell types contribute to the regulation of whole-body glycemic control, a response to ghrelin translating into such regulation would be expected. However, its direct effect in islet hormone secretion has been difficult to interpret, an issue likely emerging from the difference in study models and hormone treatment concentrations. A considerable number of studies have found ghrelin to inhibit insulin secretion in humans, mice, and rats (67–74); however, these findings have not been unanimously observed, because they have also been shown to have the opposite effect (68, 75, 76). With regard to other islet hormones, studies in humans have shown ghrelin to increase circulating levels of somatostatin and pancreatic polypeptide (72). In vitro studies show that ghrelin potentiates glucose-stimulated somatostatin secretion in mouse and human islets (15); it has the opposite effect in rats (70). Although further studies are needed to refine the effects of this hormone in islet cell types, the present evidence suggests that an effect of ghrelin on hormone secretion is likely. In our study, we found SSTR1 and SSTR2 detected in the ε-cell population and confirmed the enrichment of GHSR in human δ cells, providing supporting evidence for the previously suggested feedback loop between these cell populations in the regulation of insulin secretion.

Ghrelin acyl modification is particularly important for its activation of the GHSR (6). Although the source of medium-chain fatty acids used for this modification has been thought to be primarily derived from the diet, recent findings using PG-1 suggest instead that ε cells are able to obtain them through β-oxidation of LCFAs (77). In the current study, we found ACSL1 to be significantly enriched in ε cells. ACSL1 is an enzyme involved in the first step of fatty acid oxidation. Findings from an in vitro study using MGN3-1 suggested that elevated levels of Acsl1 contributed to the octanoic acid supply derived from LCFAs (32). It is therefore tempting to speculate that β-oxidation of LCFAs is also an important source of the medium-chain fatty acids needed for ghrelin acylation in human pancreatic ε cells.

One of the most relevant questions in the study and understanding of ghrelin cells concerns the signals that regulate the appropriate secretion of this hormone. As an approach to explore this question, we looked into the expressed receptors with an emphasis on GPCRs. Consistently, we detected GPCRs previously shown to be expressed in mouse purified ghrelin cells of gastric origin (FFAR4, FFAR3, KISS1R, SSTR1, SSTR2, and GIPR) (21) as well as in the ghrelinoma cell line MGN3-1 (FFAR4, SSTR2, and BAI2) (22). Intriguingly, we did not detect expression of the adrenoreceptor 1 (ADRB1), which has been found to be highly expressed in purified ghrelin cells as well as in the ghrelinoma cell lines (PG-1, SG-1, and MGN3-1) (21–23). Although differences between mice and humans could explain the absence of expression of this and other potential receptors that were not detected, we believe it is likely that limitations in the sensitivity of the present technology are an important factor to be considered. Interestingly, we also detected expression of orphan receptors whose function in this cell type has not been explored (G protein–coupled receptors 157 GPR157, 137 GPR137, and GPR108). We believe that the variety of receptors identified whose ligands include fatty acids, hormones, neurotransmitters, and neuropeptides, among others, represent a resource to mine in the quest for understanding human islet ghrelin secretion.

Limited information exists with regard to the transcription factors necessary for ε-cell maturation and maintenance of function. In the current study, we inquired into the transcription factors expressed in adult ε cells and found 366 transcriptions factors to be detected, encompassing 22% of all annotated factors. Although the majority of these factors and their function in ε cells has not been studied, a few have been shown to influence the differentiation of this endocrine cell lineage. Developmental mouse studies have found that loss of Nkx2.2 and paired box 6 results in an overabundance of ghrelin-expressing cells (17, 18). Interestingly, we found mature ε cells to express both NKX2.2 and paired box 6. Although counterintuitive, these findings hint at the complexity of their function within ε cells, where, during development, they seem to antagonize the ε-cell lineage, whereas in the mature stage they may be important for the function of this cell type. Evidence supporting this concept has emerged for NKX2.2; it has been shown to be able to activate the mouse and human ghrelin promoter (51, 52). Consistent with mouse studies in which loss of Nkx6.1 did not have an effect on the ε-cell population (78) and human development immunohistochemistry studies that have shown ghrelin-expressing cells to be devoid of NKX6.1,4 we also did not detect expression of this factor within this cell type. Interestingly, we found HES4 to be uniquely enriched in ε cells. To our knowledge, the function of this factor in ε cells has not been described and, along with the other transcription factors, merits further investigation.

Although other studies have inquired into the gene expression signature of ghrelin-expressing cells, they have been possible only in tissues with higher enrichment of this cell type, such as that of gastric origin or through using previously described ghrelinoma cell lines (21–27). Of these, the most relevant for our study is PG-1. Using quantitative RT-PCR analysis, several studies have inquired into its transcriptome, providing useful information regarding genes with potential involvement in ghrelin hormone secretion (23–25). We confirmed the expression of genes composing membrane-trafficking proteins [synaptotagmin (SYT) 4, SYT5, SYT11, SYT14, and SYT17] (25), neuroendocrine markers [CHGA; secretogranin II (SCG2), SCG3, and SCG5; and proprotein convertase subtilisin/kexin type 1 and proprotein convertase subtilisin/kexin type 2] (23), and genes associated with retinol metabolism (RBP4 and TTR) (24).

In addition to complementing previous knowledge acquired in mice, the current study provides a repository containing the genes found in adult human ε cells. We review the potential function and implication of several of these factors in the ε-cell population and provide opportunities for future avenues of investigation.

Glossary

Abbreviations

ACSL1, acyl-coenzyme A synthetase long chain family member 1

FFAR

free fatty acid receptor

GEM

gel beads in emulsion

GHSR

GH secretagogue receptor

GIPR

gastric inhibitory polypeptide receptor

GPCR

G protein–coupled receptor

HES

hes family bHLH transcription factor

LCFA

long-chain fatty acid

MGN3-1

mouse ghrelinoma 3

NKX2.2

NK2 homeobox 2

PG-1

pancreas-derived ghrelinoma

RBP4

retinol-binding protein 4

SG-1

stomach-derived ghrelinoma

SPINK1

serine peptidase inhibitor Kazal type 1

SYT

synaptotagmin

TM4SF4

transmembrane 4 L six family member 4

TTR

transthyretin

UMI

unique molecular identifier