Advertisement

Characterization of the Secretome, Transcriptome, and Proteome of Human β Cell Line EndoC-βH1

Open AccessPublished:April 01, 2022DOI:https://doi.org/10.1016/j.mcpro.2022.100229

      Highlights

      • We validate EndoC-βH1 as a translational human β cell model using omics.
      • We present the first unbiased proteomics composition of human β cell line secretome.
      • The secretome of human β cells is more extensive than previously thought.
      • Untreated cells secreted 783 proteins, and stressed cells secreted 3135 proteins.
      • Secreted extracellular vesicles and bioactive peptides are subject of future work.

      Abstract

      Early diabetes research is hampered by limited availability, variable quality, and instability of human pancreatic islets in culture. Little is known about the human β cell secretome, and recent studies question translatability of rodent β cell secretory profiles. Here, we verify representativeness of EndoC-βH1, one of the most widely used human β cell lines, as a translational human β cell model based on omics and characterize the EndoC-βH1 secretome. We profiled EndoC-βH1 cells using RNA-seq, data-independent acquisition, and tandem mass tag proteomics of cell lysate. Omics profiles of EndoC-βH1 cells were compared to human β cells and insulinomas. Secretome composition was assessed by data-independent acquisition proteomics. Agreement between EndoC-βH1 cells and primary adult human β cells was ∼90% for global omics profiles as well as for β cell markers, transcription factors, and enzymes. Discrepancies in expression were due to elevated proliferation rate of EndoC-βH1 cells compared to adult β cells. Consistently, similarity was slightly higher with benign nonmetastatic insulinomas. EndoC-βH1 secreted 783 proteins in untreated baseline state and 3135 proteins when stressed with nontargeting control siRNA, including known β cell hormones INS, IAPP, and IGF2. Further, EndoC-βH1 secreted proteins known to generate bioactive peptides such as granins and enzymes required for production of bioactive peptides. EndoC-βH1 secretome contained an unexpectedly high proportion of predicted extracellular vesicle proteins. We believe that secretion of extracellular vesicles and bioactive peptides warrant further investigation with specialized proteomics workflows in future studies.

      Graphical Abstract

      Keywords

      Abbreviations:

      ACN (acetonitrile), BSA (bovine serum albumin), DIA (Data-Independent Acquisition), EV (Extracellular Vesicle), FA (formic acid), FACS (Fluorescence-activated cell sorting), FDR (False Discovery Rate), GLP1R (glucagon-like peptide 1 receptor), GPCR (G-protein–coupled receptor), TMT (Tandem Mass Tag), TPM (transcripts per million)
      Close to half a billion people are living with diabetes worldwide, and this number is expected to increase by 25% in 2030 (
      • Saeedi P.
      • Petersohn I.
      • Salpea P.
      • Malanda B.
      • Karuranga S.
      • Unwin N.
      • Colagiuri S.
      • Guariguata L.
      • Motala A.A.
      • Ogurtsova K.
      • Shaw J.E.
      • Bright D.
      • Williams R.
      IDF Diabetes Atlas Committee
      Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: Results from the International Diabetes Federation Diabetes Atlas, 9(th) edition.
      ). Despite diabetes being a global health burden, the mechanisms underlying normal β cell function and failure are not yet fully elucidated. This is partly due to limited availability of high-quality human pancreatic islets for research purposes combined with variable quality and variation in relative cell type composition of islet preparations (
      • Nasteska D.
      • Hodson D.J.
      The role of beta cell heterogeneity in islet function and insulin release.
      ,
      • Saponaro C.
      • Muhlemann M.
      • Acosta-Montalvo A.
      • Piron A.
      • Gmyr V.
      • Delalleau N.
      • Moerman E.
      • Thevenet J.
      • Pasquetti G.
      • Coddeville A.
      • Cnop M.
      • Kerr-Conte J.
      • Staels B.
      • Pattou F.
      • Bonner C.
      Interindividual heterogeneity of SGLT2 expression and function in human pancreatic islets.
      ). Moreover, primary human islets rapidly lose functional expression pattern in culture (
      • Negi S.
      • Jetha A.
      • Aikin R.
      • Hasilo C.
      • Sladek R.
      • Paraskevas S.
      Analysis of beta-cell gene expression reveals inflammatory signaling and evidence of dedifferentiation following human islet isolation and culture.
      ). To overcome these limitations, pancreatic islets from rodents have been used to study β cell biology for decades leading to significant advances in our understanding of the disrupted pathways causing impaired β cell function. However, the cellular architecture and secretory profile of rodent islets is substantially different from that of human islets (
      • Kim A.
      • Miller K.
      • Jo J.
      • Kilimnik G.
      • Wojcik P.
      • Hara M.
      Islet architecture: A comparative study.
      ,
      • Campbell S.A.
      • Golec D.P.
      • Hubert M.
      • Johnson J.
      • Salamon N.
      • Barr A.
      • MacDonald P.E.
      • Philippaert K.
      • Light P.E.
      Human islets contain a subpopulation of glucagon-like peptide-1 secreting alpha cells that is increased in type 2 diabetes.
      ), manifesting the need for human β cell lines, which mirror the expression and secretory pattern of primary human β cells.
      Several human β cell lines are currently available—1.1B4, 1.4E7 and 1.1E7 (
      • McCluskey J.T.
      • Hamid M.
      • Guo-Parke H.
      • McClenaghan N.H.
      • Gomis R.
      • Flatt P.R.
      Development and functional characterization of insulin-releasing human pancreatic beta cell lines produced by electrofusion.
      ), T6PNE (
      • Kiselyuk A.
      • Farber-Katz S.
      • Cohen T.
      • Lee S.H.
      • Geron I.
      • Azimi B.
      • Heynen-Genel S.
      • Singer O.
      • Price J.
      • Mercola M.
      • Itkin-Ansari P.
      • Levine F.
      Phenothiazine neuroleptics signal to the human insulin promoter as revealed by a novel high-throughput screen.
      ), EndoC-βH1 (
      • Ravassard P.
      • Hazhouz Y.
      • Pechberty S.
      • Bricout-Neveu E.
      • Armanet M.
      • Czernichow P.
      • Scharfmann R.
      A genetically engineered human pancreatic beta cell line exhibiting glucose-inducible insulin secretion.
      ), EndoC-βH2 (
      • Scharfmann R.
      • Pechberty S.
      • Hazhouz Y.
      • von Bulow M.
      • Bricout-Neveu E.
      • Grenier-Godard M.
      • Guez F.
      • Rachdi L.
      • Lohmann M.
      • Czernichow P.
      • Ravassard P.
      Development of a conditionally immortalized human pancreatic beta cell line.
      ), EndoC-βH3 (
      • Benazra M.
      • Lecomte M.J.
      • Colace C.
      • Muller A.
      • Machado C.
      • Pechberty S.
      • Bricout-Neveu E.
      • Grenier-Godard M.
      • Solimena M.
      • Scharfmann R.
      • Czernichow P.
      • Ravassard P.
      A human beta cell line with drug inducible excision of immortalizing transgenes.
      ) and ECN90 (
      • Rachdi L.
      • Maugein A.
      • Pechberty S.
      • Armanet M.
      • Hamroune J.
      • Ravassard P.
      • Marullo S.
      • Albagli O.
      • Scharfmann R.
      Regulated expression and function of the GABAB receptor in human pancreatic beta cell line and islets.
      ). The corresponding full-text searches for these cells lines in Google Scholar retrieved 774 (22.5%), 993 (28.8%), 947 (27.5%), 13 (0.4%), 598 (17.4%), 67 (1.9%), 41(1.2%), and 11(0.3%) entries as of March 2021, respectively. Thus, EndoC-βH1 cells remain one of the most commonly used human β cell lines despite availability of next-generation cell lines EndoC-βH2 (
      • Scharfmann R.
      • Pechberty S.
      • Hazhouz Y.
      • von Bulow M.
      • Bricout-Neveu E.
      • Grenier-Godard M.
      • Guez F.
      • Rachdi L.
      • Lohmann M.
      • Czernichow P.
      • Ravassard P.
      Development of a conditionally immortalized human pancreatic beta cell line.
      ) and EndoC-βH3 (
      • Benazra M.
      • Lecomte M.J.
      • Colace C.
      • Muller A.
      • Machado C.
      • Pechberty S.
      • Bricout-Neveu E.
      • Grenier-Godard M.
      • Solimena M.
      • Scharfmann R.
      • Czernichow P.
      • Ravassard P.
      A human beta cell line with drug inducible excision of immortalizing transgenes.
      ) with inducible growth arrest. Taken together, these three generations of EndoC-βH cells are used by over 150 research labs world-wide (https://www.humanbetacelllines.com/) (
      • Scharfmann R.
      • Staels W.
      • Albagli O.
      The supply chain of human pancreatic beta cell lines.
      ).
      Validity of EndoC-βH1 cells as a human β cell model has been previously established based on functional experiments such as glucose-stimulated insulin secretion (
      • Andersson L.E.
      • Valtat B.
      • Bagge A.
      • Sharoyko V.V.
      • Nicholls D.G.
      • Ravassard P.
      • Scharfmann R.
      • Spegel P.
      • Mulder H.
      Characterization of stimulus-secretion coupling in the human pancreatic EndoC-betaH1 beta cell line.
      ,
      • Hastoy B.
      • Godazgar M.
      • Clark A.
      • Nylander V.
      • Spiliotis I.
      • van de Bunt M.
      • Chibalina M.V.
      • Barrett A.
      • Burrows C.
      • Tarasov A.I.
      • Scharfmann R.
      • Gloyn A.L.
      • Rorsman P.
      Electrophysiological properties of human beta-cell lines EndoC-betaH1 and -betaH2 conform with human beta-cells.
      ,
      • Tsonkova V.G.
      • Sand F.W.
      • Wolf X.A.
      • Grunnet L.G.
      • Kirstine Ringgaard A.
      • Ingvorsen C.
      • Winkel L.
      • Kalisz M.
      • Dalgaard K.
      • Bruun C.
      • Fels J.J.
      • Helgstrand C.
      • Hastrup S.
      • Oberg F.K.
      • Vernet E.
      • et al.
      The EndoC-betaH1 cell line is a valid model of human beta cells and applicable for screenings to identify novel drug target candidates.
      ) and marker expression profiling by qPCR (
      • Ravassard P.
      • Hazhouz Y.
      • Pechberty S.
      • Bricout-Neveu E.
      • Armanet M.
      • Czernichow P.
      • Scharfmann R.
      A genetically engineered human pancreatic beta cell line exhibiting glucose-inducible insulin secretion.
      ,
      • Tsonkova V.G.
      • Sand F.W.
      • Wolf X.A.
      • Grunnet L.G.
      • Kirstine Ringgaard A.
      • Ingvorsen C.
      • Winkel L.
      • Kalisz M.
      • Dalgaard K.
      • Bruun C.
      • Fels J.J.
      • Helgstrand C.
      • Hastrup S.
      • Oberg F.K.
      • Vernet E.
      • et al.
      The EndoC-betaH1 cell line is a valid model of human beta cells and applicable for screenings to identify novel drug target candidates.
      ). Recently, EndoC-βH1 open chromatin, transcriptomics, and miRNA landscapes were reported to be most similar to adult human β cells or islets compared to α cells, adipocytes, and muscle and peripheral blood cells (
      • Lawlor N.
      • Marquez E.J.
      • Orchard P.
      • Narisu N.
      • Shamim M.S.
      • Thibodeau A.
      • Varshney A.
      • Kursawe R.
      • Erdos M.R.
      • Kanke M.
      • Gu H.
      • Pak E.
      • Dutra A.
      • Russell S.
      • Li X.
      • et al.
      Multiomic profiling identifies cis-regulatory networks underlying human pancreatic beta cell identity and function.
      ). Two proteomics studies investigated the response to interferons in EndoC-βH1 cells (
      • Colli M.L.
      • Ramos-Rodriguez M.
      • Nakayasu E.S.
      • Alvelos M.I.
      • Lopes M.
      • Hill J.L.E.
      • Turatsinze J.V.
      • Coomans de Brachene A.
      • Russell M.A.
      • Raurell-Vila H.
      • Castela A.
      • Juan-Mateu J.
      • Webb-Robertson B.M.
      • Krogvold L.
      • Dahl-Jorgensen K.
      • et al.
      An integrated multi-omics approach identifies the landscape of interferon-alpha-mediated responses of human pancreatic beta cells.
      ,
      • Ramos-Rodriguez M.
      • Raurell-Vila H.
      • Colli M.L.
      • Alvelos M.I.
      • Subirana-Granes M.
      • Juan-Mateu J.
      • Norris R.
      • Turatsinze J.V.
      • Nakayasu E.S.
      • Webb-Robertson B.M.
      • Inshaw J.R.J.
      • Marchetti P.
      • Piemonti L.
      • Esteller M.
      • Todd J.A.
      • et al.
      The impact of proinflammatory cytokines on the beta-cell regulatory landscape provides insights into the genetics of type 1 diabetes.
      ). Yet, a thorough systematic comparison of EndoC-βH1 omics profiles to primary β cells and evaluation of potential ectopic expression patterns have not been performed. It remains unclear what are the molecular discrepancies of EndoC-βH1 with adult β cells that could further inform the scope/applicability of this cell line.
      The aim of this study is to complement the prior functional studies (
      • Scharfmann R.
      • Staels W.
      • Albagli O.
      The supply chain of human pancreatic beta cell lines.
      ,
      • Andersson L.E.
      • Valtat B.
      • Bagge A.
      • Sharoyko V.V.
      • Nicholls D.G.
      • Ravassard P.
      • Scharfmann R.
      • Spegel P.
      • Mulder H.
      Characterization of stimulus-secretion coupling in the human pancreatic EndoC-betaH1 beta cell line.
      ,
      • Hastoy B.
      • Godazgar M.
      • Clark A.
      • Nylander V.
      • Spiliotis I.
      • van de Bunt M.
      • Chibalina M.V.
      • Barrett A.
      • Burrows C.
      • Tarasov A.I.
      • Scharfmann R.
      • Gloyn A.L.
      • Rorsman P.
      Electrophysiological properties of human beta-cell lines EndoC-betaH1 and -betaH2 conform with human beta-cells.
      ,
      • Tsonkova V.G.
      • Sand F.W.
      • Wolf X.A.
      • Grunnet L.G.
      • Kirstine Ringgaard A.
      • Ingvorsen C.
      • Winkel L.
      • Kalisz M.
      • Dalgaard K.
      • Bruun C.
      • Fels J.J.
      • Helgstrand C.
      • Hastrup S.
      • Oberg F.K.
      • Vernet E.
      • et al.
      The EndoC-betaH1 cell line is a valid model of human beta cells and applicable for screenings to identify novel drug target candidates.
      ) and validate EndoC-βH1 as an in vitro human β cell model based on omics. We use control samples from a larger study investigating effects of PCSK9 knockdown (
      • Saitoski K.
      • Ryaboshapkina M.
      • Hamza G.M.
      • Jarnuczak A.F.
      • Berthault C.
      • Carlotti F.
      • Armanet M.
      • Sengupta K.
      • Rye Underwood C.
      • Andersson S.
      • Guillas I.
      • Le Goff W.
      • Scharfmann R.
      PCSK9 affects expression of key surface proteins in human pancreatic beta cells through intra- and extracellular regulatory circuits.
      ) to evaluate the EndoC-βH1 cell transcriptome and proteome against corresponding omics profiles of primary β cells and insulinoma samples. Due to their proliferative nature, EndoC-βH1 cells are more similar to human primary insulinomas than to adult human β cells which have an extremely low proliferation rate. We show that EndoC-βH1 cells belong to the β cell lineage and recapitulate features of adult primary β cells well. We observe that EndoC-βH1 cells exhibit ∼90% similarity with reference adult human β cell transcriptome and proteome. Similarity exceeding 90% was noted for β cell markers, transcription factors, and enzymes. We also extensively studied negative control markers of other cell and tissue lineages and confirm lack of major expression abnormalities.
      Further, we address a major knowledge gap: β cells are a crucial endocrine cell type, yet little is known about the β cell secretome, especially in humans. Few studies reported secretome composition of rodent insulinoma cell lines (
      • Tattikota S.G.
      • Sury M.D.
      • Rathjen T.
      • Wessels H.H.
      • Pandey A.K.
      • You X.
      • Becker C.
      • Chen W.
      • Selbach M.
      • Poy M.N.
      Argonaute2 regulates the pancreatic beta-cell secretome.
      ,
      • Stutzer I.
      • Selevsek N.
      • Esterhazy D.
      • Schmidt A.
      • Aebersold R.
      • Stoffel M.
      Systematic proteomic analysis identifies beta-site amyloid precursor protein cleaving enzyme 2 and 1 (BACE2 and BACE1) substrates in pancreatic beta-cells.
      ,
      • Pepaj M.
      • Bredahl M.K.
      • Gjerlaugsen N.
      • Thorsby P.M.
      Proteomic analysis of the INS-1E secretome identify novel vitamin D-regulated proteins.
      ), while the secretome of human immune cell–depleted islet tissue has been profiled using the SOMAscan 1300 proteomics assay only in 2021 (
      • Midha A.
      • Pan H.
      • Abarca C.
      • Andle J.
      • Carapeto P.
      • Bonner-Weir S.
      • Aguayo-Mazzucato C.
      Unique human and mouse beta-cell senescence-associated secretory phenotype (SASP) reveal conserved signaling pathways and heterogeneous factors.
      ). The SOMAscan study suggests that the human and murine secretomes may be quite distinct (
      • Midha A.
      • Pan H.
      • Abarca C.
      • Andle J.
      • Carapeto P.
      • Bonner-Weir S.
      • Aguayo-Mazzucato C.
      Unique human and mouse beta-cell senescence-associated secretory phenotype (SASP) reveal conserved signaling pathways and heterogeneous factors.
      ), making it unclear to what extent findings of secretome studies in murine β cells can be extrapolated to human. In this study, we present the first characterization of human β cell line EndoC-βH1 secretome by data-independent acquisition (DIA) proteomics. These novel data indicate that the secretome of β cells may be more extensive than previously thought and highlights surprising abundance of proteins predicted to be secreted via extracellular vesicles (EVs). We subsequently refer to these proteins as EV-associated proteins.

      Experimental Procedures

      Experimental Design and Statistical Rationale

      The experiment series was designed to investigate the effects of PCSK9 knockdown on β cell transcriptome, proteome, and secretome. We used nongrowth-arrested EndoC-βH1 cell line because it is one of the most commonly used human β cell lines, and it enables generation of a large number of cells needed for secretome proteomics (in total, >100 million cells for our samples). In total, one RNA-seq and five proteomics experiments were conducted with at least three biological replicates per condition (supplemental Table S1). Nontargeting control siRNA and untransfected samples were used to characterize EndoC-βH1 and are presented in this manuscript. Effects of PCSK9 knockdown are presented in a follow-up study (
      • Saitoski K.
      • Ryaboshapkina M.
      • Hamza G.M.
      • Jarnuczak A.F.
      • Berthault C.
      • Carlotti F.
      • Armanet M.
      • Sengupta K.
      • Rye Underwood C.
      • Andersson S.
      • Guillas I.
      • Le Goff W.
      • Scharfmann R.
      PCSK9 affects expression of key surface proteins in human pancreatic beta cells through intra- and extracellular regulatory circuits.
      ).
      For all experiments, the PCSK9 knockdown was confirmed at the mRNA level by RT-qPCR and at the protein level by Western Blot in both cell extracts and conditioned media prior to omics data acquisition. However, PCSK9 knockdown in tandem mass tag (TMT) experiment PXD027898 could not be detected due to small effect size versus plex-effect. We used PXD027898 only to verify lack of major impact of control siRNA treatment on proteome composition. All other experiments were used to support both manuscripts.

      Sample Size Calculation

      GSE182016, PXD027921, PXD027898, and PXD027920 were unpowered pilot experiments with three biological replicates per condition. Sample size for PXD027911 and PXD027913 was calculated based on PCSK9 knockdown in the pilot cell lysate TMT experiment PXD027921 as 12 nontargeting control siRNA and 12 siPCSK9 samples (power 80%, alpha 0.05, N proteins expected to be detected 9,000, alpha corrected for multiple testing 5.6e-6, Cohen d for PCSK9 knockdown -2.93, distribution of normalized protein intensities assumed to be approximately normal, two-tailed t test).

      EndoC-βH1 Cell Culture

      EndoC-βH1 cells were cultivated in low glucose Dulbecco's modified Eagle's medium (5,6 mmol/l; Sigma-Aldrich) supplemented with 2% bovine serum albumin (BSA) fraction V (Roche Diagnostics), 50 μmol/L 2-β-mercaptoethanol (Sigma-Aldrich), 10 mmol/L nicotinamide (Calbiochem), 5.5 μg/ml transferrin (Sigma-Aldrich), 6.7 ng/ml sodium selenite (Sigma-Aldrich), 100 units/ml penicillin, and 100 μg/ml streptomycin (ThermoFisher Scientific). The cells were seeded on plates coated with 1.2% Matrigel containing-3 μg/ml fibronectin (both from Sigma-Aldrich) and cultured at 37 °C and 5% CO2 as previously described (
      • Ravassard P.
      • Hazhouz Y.
      • Pechberty S.
      • Bricout-Neveu E.
      • Armanet M.
      • Czernichow P.
      • Scharfmann R.
      A genetically engineered human pancreatic beta cell line exhibiting glucose-inducible insulin secretion.
      ).
      For siRNA transfection, EndoC-βH1 cells were passaged and seeded at 105 cells/cm2. Twenty-four hours later, cells were transfected in OptiMEM using Lipofectamine RNAiMAX (Life Technologies) with siRNA SMARTpools (Horizon Discovery LTD): nontargeting control (siCTRL, D-001810–01–20) or siRNA PCSK9 (M-005989–01–0005) at the final concentration of 80 nM as described (
      • Oshima M.
      • Knoch K.P.
      • Diedisheim M.
      • Petzold A.
      • Cattan P.
      • Bugliani M.
      • Marchetti P.
      • Choudhary P.
      • Huang G.C.
      • Bornstein S.R.
      • Solimena M.
      • Albagli-Curiel O.
      • Scharfmann R.
      Virus-like infection induces human beta cell dedifferentiation.
      ). Medium was replaced 2.5 h later with fresh EndoC-βH1 culture medium.
      For proteome experiments, EndoC-βH1 cells were preincubated for 2 h in BSA/phenol red-free culture medium before transfection. After transfection, the cells were grown in fresh culture medium without BSA and without phenol red for 3 days. Cell pellets were collected by trypsinization and washed three times in PBS. Conditioned media were collected, treated with protease and phosphatase inhibitors (Roche Diagnostics, 05892791001; 04906837001), and the debris was eliminated by centrifugation at 1500 rpm at 4 °C for 5 min and filtered through 0.22 μm-diameter filters. All the samples were stored at −80 °C.

      Fluorescence-Activated Cell Sorting

      EndoC-βH1 cells were trypsinized and washed three times in PBS. Cell death was assessed by incubating the cells with annexin-V antibody (APC, #350520, Biolegend) for 15 min at room temperature in the dark in annexin-V buffer (#422201, Biolegend). Propidium iodide was added before fluorescence-activated cell sorting (FACS) analysis. For proliferation studies, cells were fixed and permeabilized with a transcription factor staining buffer set (#00–5523–00, Thermo Fisher Scientific) according to the manufacturer’s instructions. Cells were then incubated with Ki67 antibody (PerCP/Cy5.5, #B238642, Biolegend) for 30 min at room temperature in the dark in permeabilization buffer and then with DAPI (BD Biosciences) for 10 min at room temperature in the dark. FACS analysis was carried out using a FACS Aria III (BD Biosciences). Data were analyzed using FlowJo 10.7 software (RRID:SCR_008520).

      RT-qPCR

      A RNeasy Micro Kit (Qiagen) was used to extract total RNA from EndoC-βH1 cells (
      • Oshima M.
      • Knoch K.P.
      • Diedisheim M.
      • Petzold A.
      • Cattan P.
      • Bugliani M.
      • Marchetti P.
      • Choudhary P.
      • Huang G.C.
      • Bornstein S.R.
      • Solimena M.
      • Albagli-Curiel O.
      • Scharfmann R.
      Virus-like infection induces human beta cell dedifferentiation.
      ). Genomic DNA was removed by DNAse treatment following the RNeasy Micro Kit protocol. RNAs were reverse transcribed by using the Maxima First Strand cDNA kit (Thermo Fisher Scientific). RT-qPCR was performed using Power SYBR Green mix (Applied Biosystems) with a QuantStudio 3 analyzer (Thermo Fisher Scientific). Custom primers (supplemental Table S2) were designed with Primer-Blast online, and their efficiency and specificity were determined for each pair by RT-qPCR on a serial dilution of cDNA samples. Relative quantification (2ˆ-dCt method) was used to calculate the expression levels of each target gene, normalized to CYCLOPHILIN-A transcripts.

      Insulin Content

      EndoC-βH1 cells were transfected with control siRNA and cultured with 2% BSA or without BSA. Three days later, cells were washed three times with PBS and collected following trypsin treatment. Intracellular insulin content was measured by ELISA (Mercodia AB) as previously described (
      • Ravassard P.
      • Hazhouz Y.
      • Pechberty S.
      • Bricout-Neveu E.
      • Armanet M.
      • Czernichow P.
      • Scharfmann R.
      A genetically engineered human pancreatic beta cell line exhibiting glucose-inducible insulin secretion.
      ).

      RNA Sequencing

      RNeasy Micro Kit (Qiagen) was used to extract total RNA from EndoC-βH1 cells. Genomic DNA was removed by DNAse treatment following the RNeasy Micro Kit protocol. RNA quality (RNA integrity number) was determined by electrophoresis using the Agilent 2100 Bioanalyzer (Agilent Technologies) as per the manufacturer’s instructions. To construct the libraries, 600 ng of high-quality total RNA sample (RIN >8.5) was processed using TruSeq Stranded mRNA kit (Illumina) according to manufacturer’s instructions. Briefly, after purification of poly-A containing mRNA molecules, mRNA molecules were fragmented and reverse-transcribed using random primers. Replacement of dTTP by dUTP during the second strand synthesis enabled strand specificity. The addition of a single A base to the cDNA was followed by ligation of Illumina adapters. Then, the libraries were quantified with Qubit fluorometer DNA HS kit (Invitrogen). The library profiles were assessed using the DNA High Sensitivity LabChip kit on an Agilent Bioanalyzer. Libraries were sequenced on an Illumina Nextseq 500 instrument using 75 base-length reads V2.5 chemistry in a paired-end mode. After sequencing, a primary analysis based on AOZAN software (Ecole Normale Supérieure) was applied to demultiplex and control the quality of the raw data (based on FastQC modules/version 0.11.9). Raw reads were aligned to human genome GRCh38 release 96 using STAR, version 2.7.1a (
      • Dobin A.
      • Davis C.A.
      • Schlesinger F.
      • Drenkow J.
      • Zaleski C.
      • Jha S.
      • Batut P.
      • Chaisson M.
      • Gingeras T.R.
      STAR: Ultrafast universal RNA-seq aligner.
      ). Transcript quantification was performed with RSEM (
      • Li B.
      • Dewey C.N.
      RSEM: Accurate transcript quantification from RNA-seq data with or without a reference genome.
      ). Quality control metrics were generated with Picard [“Picard Toolkit.” 2019. Broad Institute, GitHub Repository. http://broadinstitute.github.io/picard/; Broad Institute] and STAR (
      • Dobin A.
      • Davis C.A.
      • Schlesinger F.
      • Drenkow J.
      • Zaleski C.
      • Jha S.
      • Batut P.
      • Chaisson M.
      • Gingeras T.R.
      STAR: Ultrafast universal RNA-seq aligner.
      ) and examined visually in addition to clustering and principal component analysis. All samples passed quality control. TPM (transcript per million) normalization was applied for descriptive analysis in this study.

      Cell Proteome Sample Preparation

      Cell pellets were thawed on ice and subjected to sample preparation with the PreOmics iST NHS or iST kit (PreOmics) according to manufacturer's protocols. Briefly, iST or iST-NHS lysis buffer was added directly to the cell pellet and incubated at 95 °C for 10 min for cell lysis, reduction, and alkylation of proteins. Cell lysate was normalized using BCA assay (Thermo Fisher Scientific), and 50 μg from each condition was subjected to enzymatic cleavage for 3 h by adding equal amounts of endoproteinase Lys-C and trypsin (ThermoFisher Scientific, # A40009) in a 1:50 (wt/wt) enzyme:protein ratio. For experiments utilizing TMT, TMT reagents were reconstituted in anhydrous acetonitrile (ACN) and added to peptides at a 4:1 reagent:peptide. De-salting and purification were performed according to the PreOmics iST or PreOmics iST-NHS protocol on a styrene divinylbenzene reversed-phase sulfonate sorbent. IST Purified peptides were vacuum-centrifuged to dryness and reconstituted in double-distilled water with 2 vol% ACN and 0.1 vol% formic acid (FA) for single-run LC-MS analysis. IST-NHS TMT-labeled peptides were reconstituted in pH 10.0, 20 mM ammonium hydroxide, and combined evenly to create a single sample.

      High pH Reversed-Phase Fractionation

      TMT-labeled peptides were fractionated into 96 fractions across a 96-well plate, using high pH reverse phase chromatography on an Agilent 1100 system and Phenomenex Gemini 5 μm C18 250 mm × 2 mm column. The linear gradient profile consisted of 2% to 90% mobile phase B in 70 min at flow rate of 0.20 ml/min. Mobile phase A consisted of pH 10.0, 20 mM ammonium hydroxide in water, and mobile phase B consisted of 20 mM ammonium hydroxide in ACN. The 96 fractions were combined within each column orthogonally into a total of 12 concatenated analytical samples. Samples were vacuum-centrifuged to dryness and were then reconstituted with double-distilled water with 2 vol% ACN and 0.1 vol% FA for LC-MS analysis.

      Secretome Sample Preparation

      Cell conditioned media were thawed on ice and were concentrated down to ∼100 μl using Amicon Ultra-15 3k molecular weight cut off (MWCO) centrifugal filter units (Millipore Sigma). Buffer exchange into PreOmics iST lysis buffer was performed, followed by incubation at 95 °C for 10 min for reduction and alkylation of proteins. Protein amounts were normalized using BCA assay (Thermo Fisher Scientific), and equal amounts from each condition were subjected to enzymatic cleavage for 3 h by adding equal amounts of endoproteinase Lys-C and trypsin (ThermoFisher Scientific, # A40009) in a 1:50 (wt/wt) enzyme:protein ratio. De-salting and purification were performed according to the PreOmics iST protocol on a styrene divinylbenzene reversed-phase sulfonate sorbent. Purified peptides were vacuum-centrifuged to dryness and reconstituted in double-distilled water with 2 vol% ACN and 0.1 vol% FA for single-run LC-MS analysis.

      LC-MS/MS Measurement

      Orbitrap Exploris 480 Mass Spectrometer (PXD027911 and PXD027920)

      Peptides were loaded onto a 25 cm IonOpticks Aurora Series emitter column (25 cm × 75u ID, 1.6um C18; IonOpticks) performed by a Dionex Ultimate 3000 coupled online to an Exploris 480 Mass Spectrometer equipped with a Nanospray Flex Ion Source, integrated with a column oven (PRSO-V1, Sonation) maintained at 50 °C. Peptides were separated using a nonlinear gradient. Mobile phase A was 0.1 vol% FA and 3 vol% ACN in water, while Mobile phase B was 90 vol% can and 0.1 vol % FA. The gradient was operated at 400 nl/min flowing 3 vol% B for 25 min, 5 to 17 vol% B over 72 min, 17 to 24 vol % B over 18 min, 24 to 30 vol % B over 10 min, 30 to 85 vol % B over 3 min, hold at 85 vol% B for 7 min, 85 to 3 vol% B over 0.1 min, and hold at 3 vol% B for 15 min. Orbitrap Exploris 480 was operated in BoxCar DIA-positive mode where spray voltage was set to 1600V, funnel RF at 40%, and heated capillary temperature at 275 °C. Method timeline experiment consisted of one MS1 scan, one tSIM scan, and one tMS2 scan. MS1 scan was operated at 120k resolution, 400 to 1200 m/z scan range, 40% RF lens, 300% AGC target, and 54 ms IT. MS1 tSIM was operated with multiplexed ions enabled (12 ions), at 120k resolution, 300% AGC target, 20 ms IT with a set loop control of 2N (number of spectra). BoxCar windows spanned 400 to 1200 m/z space. A total of 48 MS2 DIA variable windows were operated at 15k resolution with normalized collision energy of 28%, 1000% AGC target with 22 ms IT spanning 400 to 1200 m/z range taken into account heavily dense peptide regions.

      Orbitrap Fusion Lumos Mass Spectrometer (PXD027913)

      Peptides were loaded onto a ReproSil-Pur 120 C18AQ 1.9um in-house packed to a 5um tip 75u ID × 360u × 50 cm column using a Thermo EASY-nLC1200 connected through a Nanospray Flex Ion Source, integrated with a Column Oven (PRSO-V1, Sonation) maintained at 50 °C. Mobile phase A was 0.1 vol% FA and 3 vol% ACN in water, while Mobile phase B was 90 vol% can and 0.1 vol% FA. The gradient which had a flow of 300 nl/min consisted of a nonlinear ramp of 3 to 24 vol% B in 85 min, 24 to 30 vol% B for 15 min, 30 to 95 vol% B in 5 min, 95 vol% B hold for 5 min, and reequilibration to 3 vol% B for 10 min. The LC was connected to an Orbitrap Fusion Lumos Mass Spectrometer operated in DIA-positive mode. Briefly, spray voltage was set to 2500V, ion transfer tube set to 300 °C, MS1 resolution at 120k, MS2 resolution at 30k, and scan range of 350 to 1650 m/z for MS1 and 350 to 1650 m/z for MS2. RF lens was set to 30%, MS1 IT to 20 ms, and MS2 IT to 60 ms. MS1 AGC target set to 3e6, MS2 AGC target set to 1.5e6 with 28% higher-energy C-trap dissociation collision energy. DIA variable windows covered 400 to 1650 m/z space taken into account heavily dense peptide regions.

      Orbitrap Fusion Lumos Mass Spectrometer (PXD027898 and PXD027921)

      Peptides were loaded onto a ReproSil-Pur 120 C18AQ 1.9um in-house packed to a 5um tip 75u ID × 360u × 50 cm bed volume column using a Thermo EASY-nLC1200 connected through a Nanospray Flex Ion Source, integrated with a Column Oven (PRSO-V1, Sonation) maintained at 50 °C. Mobile phase A was 0.1 vol% FA, 3 vol% ACN in water, while Mobile phase B was 90 vol% ACN and 0.1 vol% FA. The gradient which had a flow of 250 nl/min consisted of nonlinear ramp of 3 to 5 vol% B in 5 min, 5 to 20 vol% B for 80 min, 20 to 32 vol% B in 30 min, 32 to 95 vol% B in 1 min, and hold at 95 vol% B for 14 min. The LC was connected to an Orbitrap Fusion Lumos Mass Spectrometer operated in data-dependent acquisition–positive mode. Briefly, spray voltage was set to 2200V, ion transfer tube set to 275 °C, MS1 resolution at 60k, MS2 resolution at 15k, scan range was set to 375 to 2000 m/z for MS1, first mass of 100 m/z for MS2, MS1 IT to 50 ms, MS2 IT to 22 ms, MS1 AGC target set to 1e6, and MS2 AGC target set to 1e6 with 38% higher-energy C-trap dissociation collision energy. Quadrupole isolation window was set to 1 m/z with 30 s exclusion duration. Monoisotopic precursor selection filter was set to peptide, filter intensity threshold was activated with maximum intensity set to 1E20, minimum intensity set to 4E4, and intensity filter type set to intensity threshold.

      Mass Spectrometry Analysis

      DIA experiments were analyzed using Spectronaut V15.5 (Biognosys AG) direct DIA analysis using a combined human, mouse (UniProt downloaded 22.02.2021) and MaxQuant contaminants (
      • Tyanova S.
      • Temu T.
      • Cox J.
      The MaxQuant computational platform for mass spectrometry-based shotgun proteomics.
      ) database utilizing the Pulsar search engine. Analysis settings were maintained to factory settings where identification was set to 1% false discovery rate (FDR) for precursor and protein level. FDR was calculated with Spectronaut algorithm based on fraction of negative control/synthetic decoy peptides in the samples (
      • Rosenberger G.
      • Bludau I.
      • Schmitt U.
      • Heusel M.
      • Hunter C.L.
      • Liu Y.
      • MacCoss M.J.
      • MacLean B.X.
      • Nesvizhskii A.I.
      • Pedrioli P.G.A.
      • Reiter L.
      • Rost H.L.
      • Tate S.
      • Ting Y.S.
      • Collins B.C.
      • et al.
      Statistical control of peptide and protein error rates in large-scale targeted data-independent acquisition analyses.
      ). Quantification was conducted on MS2 level for specific digest type of Trypsin/P. Static modifications of carbamidomethyl (+57.021 Da) and dynamic modification N-terminal acetylation (+42.011 Da) were used.
      TMT experiments were analyzed using Thermo Proteome Discoverer 2.4. Briefly, the raw files were searched through Sequest HT using a combined human, mouse (UniProt downloaded 22.02.2021) and MaxQuant contaminants (
      • Tyanova S.
      • Temu T.
      • Cox J.
      The MaxQuant computational platform for mass spectrometry-based shotgun proteomics.
      ) database with the following parameters: trypsin (full), max missed cleavage of 2, minimum peptide length of 6, max peptide length of 144, precursor mass tolerance of 10 ppm, fragment mass tolerance of 0.02 Da, dynamic modification: oxidation (+15.995 Da) of methionine, deamidated asparagine and glutamine (+0.984 Da), N-terminal acetylation (+42.011 Da), N-terminal Met-loss (−131.040 Da), N-terminal Met-loss + Acetyl (−89.030), static modification: TMT (+229.163 Da) on peptide N-terminus and lysine, PreOmics cystine modification (+113.084 Da). FDR was estimated based on target/decoy principle (
      • Elias J.E.
      • Gygi S.P.
      Target-decoy search strategy for mass spectrometry-based proteomics.
      ). FDR on peptide and protein level was 1%. TMT reporter ion quantification was carried out using 50% co-isolation threshold and average reporter S/N threshold of 10.

      Downstream Analysis

      TMT proteomics data were normalized using scaling factor normalization and log2 transformed. DIA proteomics data were normalized using quantile normalization as implemented in limma (
      • Ritchie M.E.
      • Phipson B.
      • Wu D.
      • Hu Y.
      • Law C.W.
      • Shi W.
      • Smyth G.K.
      Limma powers differential expression analyses for RNA-sequencing and microarray studies.
      ) and log2-transformed. DIA proteomics data contained small number of missing values: 0.8 (0.5–1.7)% missing protein intensity values per sample in cell lysate PXD027911 and 1.6 (1.1–2.2)% in secretome PXD027913 experiments. Missing values were imputed using sequential imputation method (
      • Verboven S.
      • Branden K.V.
      • Goos P.
      Sequential imputation for missing values.
      ,
      • Wang S.
      • Li W.
      • Hu L.
      • Cheng J.
      • Yang H.
      • Liu Y.
      NAguideR: Performing and prioritizing missing value imputations for consistent bottom-up proteomic analyses.
      ). Approximately normal distribution of normalized protein intensities were confirmed by visual examination of quantile–quantile plots. Ambiguously annotated proteins (e.g., tryptic peptides aligning to several immunoglobulin kappa chain proteins) were excluded prior to statistical analysis. Differential protein expression analysis of siRNA control versus untreated cells was performed with mixed-effect linear model with condition as fixed effect term and TMT plex as random intercept. Functional classification of the proteins (
      Gene Ontology Consortium
      The gene ontology resource: Enriching a GOld mine.
      ) was evaluated using AmiGO2 (
      • Carbon S.
      • Ireland A.
      • Mungall C.J.
      • Shu S.
      • Marshall B.
      • Lewis S.
      AmiGO HubWeb Presence Working Group
      AmiGO: Online access to ontology and annotation data.
      ) and PANTHER (
      • Mi H.
      • Ebert D.
      • Muruganujan A.
      • Mills C.
      • Albou L.P.
      • Mushayamaha T.
      • Thomas P.D.
      PANTHER version 16: A revised family classification, tree-based classification tool, enhancer regions and extensive API.
      ). Proteins may have more than one functional annotation (e.g., be involved in several pathways); therefore, the sum of percentages of proteins with different annotations may exceed 100%. Figures were made in R version 4.0.2 (
      R Core Team
      R: A Language and Environment for Statistical Computing.
      ) with ggplot2 (
      • Wickham H.
      ggplot2: Elegant Graphics for Data Analysis.
      ) and ggVennDiagram https://github.com/gaospecial/ggVennDiagram. EV-associated proteins were annotated based on ExoPred (
      • Ras-Carmona A.
      • Gomez-Perosanz M.
      • Reche P.A.
      Prediction of unconventional protein secretion by exosomes.
      ), Gene Ontology annotation (
      Gene Ontology Consortium
      The gene ontology resource: Enriching a GOld mine.
      ) and known markers (
      • Pang H.
      • Luo S.
      • Xiao Y.
      • Xia Y.
      • Li X.
      • Huang G.
      • Xie Z.
      • Zhou Z.
      Emerging roles of exosomes in T1DM.
      ).

      Results

      Quality Control of the Biological Material

      One of the biggest challenges in secretome studies is determining whether the proteins are truly secreted by viable cells or are observed as a result of cell death or leakage. Therefore, we verified that treatment with nontargeting control siRNA did not drastically affect cell viability, cell type identity, or functional state. We confirmed that treatment with control siRNA did not affect cell proliferation (Fig. 1A) or viability (Fig 1, B and C, gating supplemental Fig. S1), which are important for secretome assessment as these factors could affect membrane permeability. We assessed the differences in protein expression between untransfected day 0 cells and control nontargeting siRNA–treated EndoC-βH1 cells in PXD027898. Treatment with control siRNA altered expression of 1.9% of proteins at Benjamini–Hochberg FDR <0.05. The cells treated with control siRNA displayed signs of altered lipid metabolism and stress (Fig. 1D). However, qualitative protein composition did not differ between untransfected day 0 and control siRNA–treated cells, and the differentially expressed proteins did not overlap with β cell markers. Thus, treatment with control siRNA did not change cell type identity.
      Figure thumbnail gr1
      Fig. 1Quality control of the cells. Panels A–D, show lack of effect of transfection with nontargeting control siRNA on proliferation, apoptosis and cell identity. Ki67/Dapi staining for cell proliferation (A) and Annexin-V/PI staining for apoptosis (B), and live cells (C) were assessed by FACS in untransfected day 0 cells compared to 72 h post transfection with control siRNA or siPCSK9. D, cell lysate protein expression in EndoC-βH1 cells treated with control siRNA compared to untransfected day 0 cells in PXD027898. Each dot corresponds to a protein. Proteins in orange are differentially expressed. Proteins in black are not differentially expressed, among them β cell markers are highlighted in gray. Panels E–G, show lack of effect of BSA withdrawal on top of transfection with control nontargeting siRNA. EndoC-βH1 cells were transfected with control siRNA and then cultured in medium containing BSA (+BSA) or without BSA (-BSA). Analyses were performed 3 days later in biological triplicate per condition. E, expression of β cell markers by RT-qPCR. F, expression of proliferation marker KI67 by RT-qPCR. G, cells insulin content quantified by ELISA in cell pellets. BSA, bovine serum albumin; FACS, fluorescence-activated cell sorting.
      We maintained EndoC-βH1 cells in the absence of serum, which is standard practice for this cell line (
      • Ravassard P.
      • Hazhouz Y.
      • Pechberty S.
      • Bricout-Neveu E.
      • Armanet M.
      • Czernichow P.
      • Scharfmann R.
      A genetically engineered human pancreatic beta cell line exhibiting glucose-inducible insulin secretion.
      ), and did not glucose-starve the cells. 2% BSA was initially present in culture media but was removed 3 days prior to collection of conditioned media to avoid contaminants in the secretome experiments. We assessed key markers (Fig. 1, EG) to test if withdrawal of BSA impacted the functional state of EndoC-βH1 cells. We observed no differences between cells cultured with versus without BSA.

      Computational Quality Control

      We cultured the cells on mouse cell culture matrix Matrigel, so it was important to remove potential contaminants. We excluded mouse proteins and proteins from the contaminant library from further analysis. Some proteins could be either human or mouse due to shared sequence similarity. We estimated that 31% of unique tryptic peptide sequences could belong to either human or mouse using in silico digestion of our combined human, mouse, and contaminant database with Skyline (
      • MacLean B.
      • Tomazela D.M.
      • Shulman N.
      • Chambers M.
      • Finney G.L.
      • Frewen B.
      • Kern R.
      • Tabb D.L.
      • Liebler D.C.
      • MacCoss M.J.
      Skyline: An open source document editor for creating and analyzing targeted proteomics experiments.
      ). The fractions of ambiguous human/mouse protein species (e.g., 26.1% proteins in PXD027913 and 15.0% PXD027911) were comparable or below the Skyline estimate. The ambiguous proteins had ∼20% overlap with known Matrigel composition, but the distributions of normalized intensities and missing values for ambiguous proteins were similar to human-unique proteins and not mouse-unique proteins or contaminants (supplemental Fig. S2). Therefore, we considered ambiguous proteins as human and retained them in subsequent analysis.
      Next, we cross-referenced coordinates of peptides detected in the secretome experiments and mapped to membrane proteins with topology of proteins in the human surfaceome (
      • Bausch-Fluck D.
      • Goldmann U.
      • Muller S.
      • van Oostrum M.
      • Muller M.
      • Schubert O.T.
      • Wollscheid B.
      The in silico human surfaceome.
      ). If a secretome preparation is high-quality, peptides should map to extracellular regions that can be cleaved-off or shed. By contrast, contamination with peptides that map to transmembrane or intracellular regions indicates lysis or mechanical cell damage. The ratio of combined lengths of extracellular versus intracellular regions in the reference database was 2.4:1. This expected ratio was preserved in the cell lysate DIA experiment PXD027911 with 66.9% peptides mapped to extracellular regions and 29.8 % peptides mapped to intracellular regions. By contrast, 94.8% peptides mapped to extracellular regions in PXD027913 and 94.2% in PXD027920 secretome experiments. Cell lysate proteome in PXD027911 and secretome in PXD027913 were generated based on the same EndoC-βH1 cell cultures. Expression abundances of proteins with cytoplasmic subcellular location according to UniProt (
      UniProt Consortium
      UniProt: The universal protein knowledgebase in 2021.
      ) did not correlate between cell lysate and the corresponding secretome samples—median IQR Spearman correlation 0.02 (−0.12–0.19)—indicating that leakage was unlikely. Proglucagon was not detected in the secretome experiments. Somatostatin was secreted by cells treated with control siRNA but not by untransfected EndoC-βH1 cells.
      Secretome time course experiment PXD027920 contained both untransfected day 0 samples and samples treated with siRNA. We followed the approach by Villarreal et al. (
      • Villarreal L.
      • Mendez O.
      • Salvans C.
      • Gregori J.
      • Baselga J.
      • Villanueva J.
      Unconventional secretion is a major contributor of cancer cell line secretomes.
      ) to assess if transfection had a negative impact on the quality of the secretome preparation. Untransfected day 0 samples clustered separately from siRNA-transfected samples (Fig. 2A), indicating a substantial difference of secretomes collected from cells stressed with a siRNA. The fraction of signal corresponding to secreted apoptosis markers (
      • Villarreal L.
      • Mendez O.
      • Salvans C.
      • Gregori J.
      • Baselga J.
      • Villanueva J.
      Unconventional secretion is a major contributor of cancer cell line secretomes.
      ) increased in siRNA-treated cells (Fig. 2B) but did not increase with time and, importantly, stayed under 2% (
      • Villarreal L.
      • Mendez O.
      • Salvans C.
      • Gregori J.
      • Baselga J.
      • Villanueva J.
      Unconventional secretion is a major contributor of cancer cell line secretomes.
      ) in agreement with FACS data showing no increase in apoptosis (Fig. 1, B and C). We selected several negative control marker sets corresponding to different intracellular structures based on HGNC annotation (
      • Tweedie S.
      • Braschi B.
      • Gray K.
      • Jones T.E.M.
      • Seal R.L.
      • Yates B.
      • Bruford E.A.
      Genenames.org: The HGNC and VGNC resources in 2021.
      ). The fraction of signal corresponding to these negative control markers was within 0 to 2% (Fig. 2, CG). Thus, some of the negative control markers were detected albeit at low level. This led us to test if noncanonical secretion via EVs was a plausible explanation. We assessed proteins detected in ≥ 2 untransfected day 0 samples and ≥ 2 siControl samples at 72 h with ExoPred (
      • Ras-Carmona A.
      • Gomez-Perosanz M.
      • Reche P.A.
      Prediction of unconventional protein secretion by exosomes.
      ). Positive control was the proteome of EVs isolated by ultracentrifugation and with confirmed enrichment of CD63, CD81 and CD9 in the isolate by Western blot PXD013419 (
      • Qiao Z.
      • Zhang Y.
      • Ge M.
      • Liu S.
      • Jiang X.
      • Shang Z.
      • Liu H.
      • Cao C.
      • Xiao H.
      Cancer cell derived small extracellular vesicles contribute to recipient cell metastasis through promoting HGF/c-Met pathway.
      ). Background distribution in the human proteome based on UniProt (
      UniProt Consortium
      UniProt: The universal protein knowledgebase in 2021.
      ) and ExoCarta (
      • Keerthikumar S.
      • Chisanga D.
      • Ariyaratne D.
      • Al Saffar H.
      • Anand S.
      • Zhao K.
      • Samuel M.
      • Pathan M.
      • Jois M.
      • Chilamkurti N.
      • Gangoda L.
      • Mathivanan S.
      ExoCarta: A Web-based compendium of exosomal cargo.
      ) was used as negative control. The percentages of proteins with predicted signal peptide (conventional secretion) and predicted EV-associated proteins (nonconventional secretion) were higher in our samples than in the negative control (Fig. 2H). The proportion of predicted EV-associated proteins were similar in untransfected day 0 and samples treated with control siRNA for 72h, and only slightly below the positive control.
      Figure thumbnail gr2
      Fig. 2Computational quality control of the secretome time course experiment . A, hierarchical clustering. BG, relative abundance of selected negative control marker sets. Relative abundance was calculated as sum of raw protein group intensities corresponding to the markers versus total in each sample. H, percentages of proteins containing signal peptide and proteins that can be secreted via extracellular vesicles (EV) predicted with ExoPred. Overall percentages of these protein categories in the human proteome were used as negative control and indicated based on UniProt and ExoCarta (gray bars).
      In theory, secretomes containing EVs should contain substantial proportion of tryptic peptides corresponding to intracellular or transmembrane domains of membrane proteins. Therefore, we investigated if the predicted percentage of EV-associated proteins contradicts that tryptic peptides in our secretome samples aligned predominantly to the extracellular domains of transmembrane proteins. Positive control experiment PXD013419 had 75.7% of the tryptic peptides aligning to extracellular domains of transmembrane proteins (
      • Qiao Z.
      • Zhang Y.
      • Ge M.
      • Liu S.
      • Jiang X.
      • Shang Z.
      • Liu H.
      • Cao C.
      • Xiao H.
      Cancer cell derived small extracellular vesicles contribute to recipient cell metastasis through promoting HGF/c-Met pathway.
      ). We analyzed an additional study (
      • Askeland A.
      • Borup A.
      • Ostergaard O.
      • Olsen J.V.
      • Lund S.M.
      • Christiansen G.
      • Kristensen S.R.
      • Heegaard N.H.H.
      • Pedersen S.
      Mass-spectrometry based proteome comparison of extracellular vesicle isolation methods: Comparison of ME-kit, size-exclusion chromatography, and high-speed centrifugation.
      ), in which EV were isolated with different methods leading to different particle concentration, to obtain more context. High speed centrifugation reference samples had ∼e9 particles/ml and 77.8 ± 1.9% tryptic peptides aligning to extracellular domains, peptide-affinity precipitation samples had ∼e10 particles/ml and 91.5 ± 2.2% tryptic peptides aligning to extracellular domains, and size exclusion chromatography samples had ∼e11 particles/ml and 86.0 ± 1.6% of tryptic peptides aligning to extracellular domains of transmembrane proteins. Interestingly, tryptic peptides were representing CD9, CD81, and CD63 in our secretome samples and the reference EV studies aligned to the extracellular domains, i.e., it was not possible to unambiguously determine from tryptic peptides if these proteins were EV markers or part of the sheddome. We still considered CD9/63/81 as EV markers because they were detected not in isolation but together with other predicted EV-associated proteins. In summary, high percentage of peptides representing extracellular domains did not contradict the prediction of EV-associated proteins. Also, based on external reference data, this metric did not correlate with EV particle concentration.

      Validation of EndoC-βH1 as Human β Cell Model Based On Omics

      We compared the transcriptome and cell lysate proteome of EndoC-βH1 both globally and across selected positive control and negative control marker categories to published human primary β cells and insulinomas. The reference human β cell preparations were selected based on purity of the tissue: RNA-seq of adult primary human β cells from seven nondiabetic donors, isolated by FACS sorting with at least 97% β cell purity in GSE67543 (
      • Blodgett D.M.
      • Nowosielska A.
      • Afik S.
      • Pechhold S.
      • Cura A.J.
      • Kennedy N.J.
      • Kim S.
      • Kucukural A.
      • Davis R.J.
      • Kent S.C.
      • Greiner D.L.
      • Garber M.G.
      • Harlan D.M.
      • diIorio P.
      Novel observations from next-generation RNA sequencing of highly purified human adult and fetal islet cell subsets.
      ) and proteomics of four nondiabetic human islet cultures on nanostructured zirconia with 80% ± 10% purity of β cells in PXD007569 (
      • Galli A.
      • Maffioli E.
      • Sogne E.
      • Moretti S.
      • Di Cairano E.S.
      • Negri A.
      • Nonnis S.
      • Norata G.D.
      • Bonacina F.
      • Borghi F.
      • Podesta A.
      • Bertuzzi F.
      • Milani P.
      • Lenardi C.
      • Tedeschi G.
      • et al.
      Cluster-assembled zirconia substrates promote long-term differentiation and functioning of human islets of Langerhans.
      ). The reference insulinoma samples were selected from tumors with typical presentation and without metastases: RNA-seq samples WT_MK27 and WT_MK8 in GSE118014 (
      • Chan C.S.
      • Laddha S.V.
      • Lewis P.W.
      • Koletsky M.S.
      • Robzyk K.
      • Da Silva E.
      • Torres P.J.
      • Untch B.R.
      • Li J.
      • Bose P.
      • Chan T.A.
      • Klimstra D.S.
      • Allis C.D.
      • Tang L.H.
      ATRX, DAXX or MEN1 mutant pancreatic neuroendocrine tumors are a distinct alpha-cell signature subgroup.
      ) and proteomics by Song et al. (
      • Song Y.L.
      • Yu R.
      • Qiao X.W.
      • Bai C.M.
      • Lu C.M.
      • Xiao Y.
      • Zhong D.R.
      • Chen J.
      • Zhao Y.P.
      • Zhang T.P.
      • Song T.T.
      • Gao H.L.
      • Wan Y.H.
      • Shen L.
      • Chen J.
      • et al.
      Prognostic relevance of UCH-L1 and alpha-internexin in pancreatic neuroendocrine tumors.
      ). The reference transcriptome and proteome contained less molecular species than the EndoC-βH1 transcriptome and proteome acquired in this study. Hence, we calculated percentage of the reference transcriptome/proteome that was recapitulated by EndoC-βH1 cells.
      First, we compared EndoC-βH1 cells to primary human adult β cells. Overall, transcriptome and proteome of EndoC-βH1 cells resembled adult human β cells (Fig. 3, A and B). EndoC-βH1 recapitulated 90.7% of β cell protein-coding transcriptome and 90% of the β cell-enriched islet proteome. The reference human β cell proteomics data originated from an older study and contained limited number of protein species. Therefore, the agreement between marker gene sets was calculated based on transcriptomics. Good agreement between 69% and 91% versus adult β cells was observed across individual gene/protein categories such as β cell markers, genes associated with diabetes, transcription factors, enzymes, ion channels, and lncRNA (supplemental Table S3, supplemental Figs. S3–S8 and S10). The lowest agreement was obtained for G-protein–coupled receptors (GPCRs) with 52% of β cell GPCRs detected on mRNA level and only a few species confirmed by proteomics (supplemental Fig. S9). In particular, median glucagon-like peptide 1 receptor (GLP1R) expression in control siRNA-treated EndoC-βH1 cells was 0.8 TPM, whereas median expression in primary adult β cells was 61.7 TPM. We also verified low GLP1R expression with median 5 TPM in an external experiment with untransfected EndoC-βH1 cells deep-sequenced at 200 million reads per sample in GSE133218 (
      • Colli M.L.
      • Ramos-Rodriguez M.
      • Nakayasu E.S.
      • Alvelos M.I.
      • Lopes M.
      • Hill J.L.E.
      • Turatsinze J.V.
      • Coomans de Brachene A.
      • Russell M.A.
      • Raurell-Vila H.
      • Castela A.
      • Juan-Mateu J.
      • Webb-Robertson B.M.
      • Krogvold L.
      • Dahl-Jorgensen K.
      • et al.
      An integrated multi-omics approach identifies the landscape of interferon-alpha-mediated responses of human pancreatic beta cells.
      ).
      Figure thumbnail gr3
      Fig. 3Transcriptome and proteome of EndoC-βH1 cells compared to adult human β cells and insulinomas. A, Venn diagram showing the number of quantified protein-coding genes and proteins in EndoC-βH1 cells versus adult nondiabetic β cells (GSE67543). B, comparison to human β cell–enriched islet proteomics (PXD007569). C, comparison to protein-coding genes with expression >1 TPM in human low-grade insulinomas without lymph node and distant metastases (GSE118014). D, comparison to proteome of human insulinomas with typical tumor presentation and without lymph node and distant metastases (
      • Song Y.L.
      • Yu R.
      • Qiao X.W.
      • Bai C.M.
      • Lu C.M.
      • Xiao Y.
      • Zhong D.R.
      • Chen J.
      • Zhao Y.P.
      • Zhang T.P.
      • Song T.T.
      • Gao H.L.
      • Wan Y.H.
      • Shen L.
      • Chen J.
      • et al.
      Prognostic relevance of UCH-L1 and alpha-internexin in pancreatic neuroendocrine tumors.
      ). EndoC-βH1 cells were treated with control nontargeting siRNA and harvested at 72 h. Expression cut-off for genes: median expression ≥ 1 TPM. Expression cut-off for proteins: quantified in ≥ 3 samples and unambiguously annotated. Sample sizes in EndoC-βH1 experiments: RNA-seq N = 3, TMT proteomics = 3, cell lysate DIA proteomics = 12 replicates. Blue area indicates protein-coding genes/proteins recapitulated in EndoC-βH1 cells. Blue + gray area indicates total number of protein-coding genes/proteins in reference β cell or insulinoma samples. DIA, data-independent acquisition; TMT, tandem mass tag, TPM, transcripts per million.
      Interestingly, the EndoC-βH1 transcriptome and proteome were more similar to insulinomas than to primary human β cells (Fig. 3, C and D). EndoC-βH1 cells recapitulated 96% of the reference insulinoma transcriptome and 91.4% of the reference insulinoma proteome. Improved concordance was also observed for individual protein categories (Fig. 4, A and B). Overall, the genes and proteins quantified in EndoC-βH1 cells but not in primary β cells were annotated with cell cycle, DNA repair, proliferation, ATP generation, and mitochondrial function Gene Ontology terms. This ectopic expression pattern was consistent with proliferative phenotype. EndoC-βH1 cells double in number every 174 h (
      • Andersson L.E.
      • Valtat B.
      • Bagge A.
      • Sharoyko V.V.
      • Nicholls D.G.
      • Ravassard P.
      • Scharfmann R.
      • Spegel P.
      • Mulder H.
      Characterization of stimulus-secretion coupling in the human pancreatic EndoC-betaH1 beta cell line.
      ) whereas proliferation of adult human β cells is an extremely rare event (
      • Rutti S.
      • Sauter N.S.
      • Bouzakri K.
      • Prazak R.
      • Halban P.A.
      • Donath M.Y.
      In vitro proliferation of adult human beta-cells.
      ). EndoC-βH1 expressed β cell hormones INS and IAPP at gene and protein levels and ADCYAP1 (also known as PACAP) at mRNA level (Fig. 4C). EndoC-βH1 expressed mRNA of other islet endocrine hormones (Fig. 4C) which can also be observed in primary β cells and insulinomas (Figs. 4C and S11). However, detection of proglucagon and somatostatin on protein level (Fig. 4D) is atypical for adult human β cells. Somatostatin expression in EndoC-βH1 cells has been noted previously (
      • Tsonkova V.G.
      • Sand F.W.
      • Wolf X.A.
      • Grunnet L.G.
      • Kirstine Ringgaard A.
      • Ingvorsen C.
      • Winkel L.
      • Kalisz M.
      • Dalgaard K.
      • Bruun C.
      • Fels J.J.
      • Helgstrand C.
      • Hastrup S.
      • Oberg F.K.
      • Vernet E.
      • et al.
      The EndoC-betaH1 cell line is a valid model of human beta cells and applicable for screenings to identify novel drug target candidates.
      ) and is observed in insulinomas (
      • Song Y.L.
      • Yu R.
      • Qiao X.W.
      • Bai C.M.
      • Lu C.M.
      • Xiao Y.
      • Zhong D.R.
      • Chen J.
      • Zhao Y.P.
      • Zhang T.P.
      • Song T.T.
      • Gao H.L.
      • Wan Y.H.
      • Shen L.
      • Chen J.
      • et al.
      Prognostic relevance of UCH-L1 and alpha-internexin in pancreatic neuroendocrine tumors.
      ). Proglucagon protein was detected based on a single peptide DFINWLIQTK aligning to amino acid positions 166 to 175 and corresponding to glucagon-like peptide 2. Proglucagon co-expression with insulin is a feature of insulinomas (
      • Song Y.L.
      • Yu R.
      • Qiao X.W.
      • Bai C.M.
      • Lu C.M.
      • Xiao Y.
      • Zhong D.R.
      • Chen J.
      • Zhao Y.P.
      • Zhang T.P.
      • Song T.T.
      • Gao H.L.
      • Wan Y.H.
      • Shen L.
      • Chen J.
      • et al.
      Prognostic relevance of UCH-L1 and alpha-internexin in pancreatic neuroendocrine tumors.
      ,
      • Wang Z.
      • You J.
      • Xu S.
      • Hua Z.
      • Zhang W.
      • Deng T.
      • Fang N.
      • Fang Q.
      • Liu H.
      • Peng L.
      • Wang P.
      • Lou J.
      Colocalization of insulin and glucagon in insulinoma cells and developing pancreatic endocrine cells.
      ). EndoC-βH1 cells expressed β cell transcription factors on both gene and protein level (Fig. 4, C and D). By contrast, α cell transcription factor ARX was detected only on mRNA level and with low expression abundance (Fig. 4C). More in-depth evaluation of negative control marker sets indicated no ectopic expression of other α cell markers (supplemental Fig. S12). EndoC-βH1 cells were derived from fetal islet cells at 9 to 11 weeks of gestation (
      • Ravassard P.
      • Hazhouz Y.
      • Pechberty S.
      • Bricout-Neveu E.
      • Armanet M.
      • Czernichow P.
      • Scharfmann R.
      A genetically engineered human pancreatic beta cell line exhibiting glucose-inducible insulin secretion.
      ). Interestingly, expression of endocrine progenitor markers was considerably reduced in EndoC-βH1 cells compared to human fetal pancreas at 9 weeks gestation (supplemental Fig. S13). Expression of NEUROG3 was lost. Markers of fetal β cell at 12 to 18 weeks gestation were not expressed (supplemental Fig. S14). We also observed no ectopic expression of markers of exocrine pancreas acinar (supplemental Fig. S15) and ductal cells (supplemental Fig. S16) and markers of other tissues (supplemental Fig. S17).
      Figure thumbnail gr4
      Fig. 4Selected gene and protein expression in EndoC-βH1 cells compared to β cells and insulinomas. A, top 15 PANTHER pathways. B, top 15 PANTHER Protein classes. The percentages indicate the number of proteins annotated with a given PANTHER term versus the total number of proteins quantified in each experiment. C, mRNA expression of endocrine hormones and selected transcription factors. D, protein expression of endocrine hormones and selected transcription factors. β cell proteomics from PXD007569 and RNA-seq from GSE67543. Insulinoma proteomics from (
      • Song Y.L.
      • Yu R.
      • Qiao X.W.
      • Bai C.M.
      • Lu C.M.
      • Xiao Y.
      • Zhong D.R.
      • Chen J.
      • Zhao Y.P.
      • Zhang T.P.
      • Song T.T.
      • Gao H.L.
      • Wan Y.H.
      • Shen L.
      • Chen J.
      • et al.
      Prognostic relevance of UCH-L1 and alpha-internexin in pancreatic neuroendocrine tumors.
      ) and RNA-seq from GSE118014. EndoC-βH1 cells were treated with control nontargeting siRNA and harvested at 72 h Expression cutoff for proteins: quantified in ≥ 3 samples and unambiguously annotated. Sample sizes in EndoC-βH1 experiments: RNA-seq N = 3, TMT proteomics = 3, cell lysate DIA proteomics = 12 replicates. DIA, data-independent acquisition; TMT, tandem mass tag.

      Secretome Composition of EndoC- βH1 Cells

      Higher number of proteins were detected in the secretome of EndoC-βH1 cells treated with control nontargeting siRNA than in untransfected day 0 cells (Fig. 5A). This observation was true both within the secretome PXD027920 (triplicates per condition) and in comparison to the 12 control secretome samples from PXD027913, i.e., not linked to discrepancies in sample size. As previously shown in Figure 2, these differences were related to transfection with siRNA. Nonetheless, both secretomes of siRNA-stressed and untransfected cells contained four main categories of proteins: (1) known β cell hormones and granins, (2) predicted EV-associated proteins, (3) proteins with signal peptide, and (4) sheddome.
      Figure thumbnail gr5
      Fig. 5Composition of the EndoC-βH1 cell secretome. A, secretome of untransfected day 0 cells (PXD027920, N = 3) compared to secretome of cells treated with control nontargeting siRNA for 72 h (PXD027913, N = 12). B, known positive control proteins that were quantified in EndoC-βH1 cell secretome. C, number of proteins by mode of secretion. Proteins can belong to more than one category (e.g., secreted based on both UniProt and HPA annotation). D, top 25 PANTHER protein families. Expression cutoff for proteins: quantified in min two out of three untransfected day 0 samples, quantified in min three out of 12 samples at 72 h.
      EndoC-βH1 cells secreted known β cell hormones INS and IAPP and autocrine regulator IGF2 (
      • Modi H.
      • Cornu M.
      • Thorens B.
      Glutamine stimulates biosynthesis and secretion of insulin-like growth factor 2 (IGF2), an autocrine regulator of beta cell mass and function.
      ) (Fig. 5B). The secretome contained proteins known to generate bioactive peptides such as all eight proteins of the granin family (Fig. 5B), among which chromogranins A and B, secretogranins 2 and 3, and VGF have been previously reported in insulinomas (
      • Bartolomucci A.
      • Possenti R.
      • Mahata S.K.
      • Fischer-Colbrie R.
      • Loh Y.P.
      • Salton S.R.
      The extended granin family: Structure, function, and biomedical implications.
      ). Secretome proteomics indicated potential presence of several bioactive peptides (supplemental Figs. S18–S52) including chromogranin A peptides pancreastatin and vasostatins, chromogranin B peptides GAWK and CCB, and secretogranin-2 peptide manserin and VGF peptides. Some peptides were established β cell secretion products: IGF2 peptide preptin is co-secreted with insulin (
      • Cornish J.
      • Callon K.E.
      • Bava U.
      • Watson M.
      • Xu X.
      • Lin J.M.
      • Chan V.A.
      • Grey A.B.
      • Naot D.
      • Buchanan C.M.
      • Cooper G.J.
      • Reid I.R.
      Preptin, another peptide product of the pancreatic beta-cell, is osteogenic in vitro and in vivo.
      ) while VGF peptides neuroendocrine regulatory peptide-2 (
      • Moin A.S.
      • Yamaguchi H.
      • Rhee M.
      • Kim J.W.
      • Toshinai K.
      • Waise T.M.
      • Naznin F.
      • Matsuo T.
      • Sasaki K.
      • Minamino N.
      • Yoon K.H.
      • Nakazato M.
      Neuroendocrine regulatory peptide-2 stimulates glucose-induced insulin secretion in vivo and in vitro.
      ) and TLQP-62 (
      • Petrocchi-Passeri P.
      • Cero C.
      • Cutarelli A.
      • Frank C.
      • Severini C.
      • Bartolomucci A.
      • Possenti R.
      The VGF-derived peptide TLQP-62 modulates insulin secretion and glucose homeostasis.
      ) regulate glucose-induced insulin secretion. The secretome also contained CPE, PAM, and proprotein convertases PCSK1 and PCSK2—the enzymes involved in generation of bioactive peptides from prohormones. Interestingly, EndoC-βH1 cells also secreted PCSK9, a member of proprotein convertase family without known enzymatic function but involved in LDL receptor clearance in hepatocytes. We further investigated the function of PCSK9 in EndoC-βH1 cells in a separate follow-up study (
      • Saitoski K.
      • Ryaboshapkina M.
      • Hamza G.M.
      • Jarnuczak A.F.
      • Berthault C.
      • Carlotti F.
      • Armanet M.
      • Sengupta K.
      • Rye Underwood C.
      • Andersson S.
      • Guillas I.
      • Le Goff W.
      • Scharfmann R.
      PCSK9 affects expression of key surface proteins in human pancreatic beta cells through intra- and extracellular regulatory circuits.
      ).
      We quantified EV-associated markers (
      • Pang H.
      • Luo S.
      • Xiao Y.
      • Xia Y.
      • Li X.
      • Huang G.
      • Xie Z.
      • Zhou Z.
      Emerging roles of exosomes in T1DM.
      ) in the secretome samples—CD63, CD81, heat shock 70 kDa proteins, and TSG101 (Fig. 5B) and members of the Rab family (Fig. 5D). EV-associated proteins constituted the majority of unique protein species of the secretome, both in untransfected day 0 cells and in cells transfected with control siRNA (Fig. 5C). This analysis (Fig. 5C) was based on an orthogonal method and different sample subset to previously presented in Figure 2H.
      The second most common category of proteins in the secretome was classically secreted proteins containing signal peptide (Fig. 5C). For example, both untreated and control siRNA-treated EndoC-βH1 cells secreted factors regulating cell differentiation and survival NENF, MANF, MYDGF, and CREG1. EndoC-βH1 cells also secreted proteins that are measurable in circulation (
      • Uhlen M.
      • Karlsson M.J.
      • Hober A.
      • Svensson A.S.
      • Scheffel J.
      • Kotol D.
      • Zhong W.
      • Tebani A.
      • Strandberg L.
      • Edfors F.
      • Sjostedt E.
      • Mulder J.
      • Mardinoglu A.
      • Berling A.
      • Ekblad S.
      • et al.
      The human secretome.
      ) such as annexins, apolipoproteins, serpins, IGF-binding proteins, and VEGFA.
      In addition, we captured known proteins with shed extracellular domain in islets or rodent cell lines (
      • Stutzer I.
      • Selevsek N.
      • Esterhazy D.
      • Schmidt A.
      • Aebersold R.
      • Stoffel M.
      Systematic proteomic analysis identifies beta-site amyloid precursor protein cleaving enzyme 2 and 1 (BACE2 and BACE1) substrates in pancreatic beta-cells.
      ,
      • Yesildag B.
      • Bock T.
      • Herrmanns K.
      • Wollscheid B.
      • Stoffel M.
      Kin of IRRE-like protein 2 is a phosphorylated glycoprotein that regulates basal insulin secretion.
      ,
      • Inoue T.
      • Hagiyama M.
      • Yoneshige A.
      • Kato T.
      • Enoki E.
      • Maenishi O.
      • Chikugo T.
      • Kimura M.
      • Satou T.
      • Ito A.
      Increased ectodomain shedding of cell adhesion molecule 1 from pancreatic islets in type 2 diabetic pancreata: Correlation with hemoglobin A1c levels.
      ) in the EndoC-βH1 cell secretome (Fig. 5B). Sheddases ADAM10, ADAM17, ADAM22, and ADAM9 were quantified in the EndoC-βH1 secretome. Sheddases BACE1 and BACE2 were not quantified in the secretome samples but were quantified in the DIA and TMT cell lysate experiments.
      Finally, we investigated the most abundantly secreted proteins in the untransfected day 0 cells (supplemental Table S4) and in control siRNA-treated samples (Table 1). β cell hormones were among the most abundant secreted proteins confirming β cell line identity.
      Table 1Top 150 secreted proteins
      RankUniProtSymbolRankUniProtSymbolRankUniProtSymbol
      1Q5TEC6H3-251P62851RPS25101P07237P4HB
      2Q9H161ALX452P13533MYH6102O43405COCH
      3Q6ZNA1ZNF83653Q6ZVC0NYAP1103P62249RPS16
      4Q96JB3HIC254P30049ATP5F1D104P11142HSPA8
      5P25398RPS1255P60660MYL6105Q9Y2S6TMA7
      6P61769B2M56Q92820GGH106Q16769QPCT
      7O75037KIF21B57Q06830PRDX1107P60900PSMA6
      8Q8N4Y2CRACR2B58P06733ENO1108Q8IWT0ZBTB8OS
      9P01308INS59P28072PSMB6109Q15084PDIA6
      10P10997IAPP60Q96M02C10ORF90110P30040ERP29
      11P05386RPLP161P01857IGHG1111O43617TRAPPC3
      12P04080CSTB62Q9Y2V2CARHSP1112P49458SRP9
      13Q99797MIPEP63Q9UHG2PCSK1N113P40926MDH2
      14P00441SOD164Q16555DPYSL2114Q05519SRSF11
      15P13473LAMP265P61088UBE2N115P09651HNRNPA1
      16P06753TPM366P62937PPIA116P06703S100A6
      17Q8NGI1OR56B2P67Q13522PPP1R1A117Q8WXD2SCG3
      18Q14011CIRBP68Q14103HNRNPD118O00264PGRMC1
      19Q13795ARFRP169Q9BVA1TUBB2B119Q9Y5S9RBM8A
      20P68133ACTA170P23528CFL1120P26641EEF1G
      21P63279UBE2I71P23284PPIB121P61956SUMO2
      22P02790HPX72P55145MANF122P29401TKT
      23Q92526CCT6B73P30046DDT123P30044PRDX5
      24P14854COX6B174Q8NBP7PCSK9124Q9NXJ5PGPEP1
      25P00352ALDH1A175P16870CPE125Q9Y3C8UFC1
      26P02766TTR76P41567EIF1126P62304SNRPE
      27Q96FJ2DYNLL277P01859IGHG2127O43598DNPH1
      28P10599TXN78P01034CST3128Q00688FKBP3
      29Q9H3M0KCNF179P63261ACTG1129Q99729HNRNPAB
      30P20674COX5A80P23526AHCY130P16401H1-5
      31P07737PFN181P35030PRSS3131P31946YWHAB
      32P62318SNRPD382Q9H299SH3BGRL3132P07305H1-0
      33P62805H4C183Q9BUJ0ABHD14A133Q9NQG5RPRD1B
      34O75093SLIT184O95336PGLS134Q9NPF5DMAP1
      35P07108DBI85Q9GZX9TWSG1135P62857RPS28
      36P32119PRDX286P07900HSP90AA1136P35527KRT9
      37P61457PCBD187Q15631TSN137Q9BRA2TXNDC17
      38P26583HMGB288Q13526PIN1138P68036UBE2L3
      39Q04760GLO189Q13404UBE2V1139P25789PSMA4
      40P60866RPS2090P62316SNRPD2140P63173RPL38
      41P00492HPRT191P62861FAU141P04075ALDOA
      42P24666ACP192Q9Y5Q9GTF3C3142O94985CLSTN1
      43P60174TPI193Q15819UBE2V2143O75367MACROH2A1
      44P25391LAMA194P30101PDIA3144O76038SCGN
      45P09936UCHL195Q15413RYR3145P43251BTD
      46P10645CHGA96Q92520FAM3C146P07205PGK2
      47P62942FKBP1A97Q08380LGALS3BP147O00299CLIC1
      48P62258YWHAE98Q8IUH3RBM45148P55327TPD52
      49Q86TJ2TADA2B99Q9UJ72ANXA10149Q14061COX17
      50Q92688ANP32B100P63104YWHAZ150P30041PRDX6
      Proteins are ordered by decreasing median normalized expression abundance in EndoC-βH1 cells treated with control nontargeting siRNA at 72 h in the DIA secretome experiment (PXD027913). N = 12.

      Discussion

      All β cell models have advantages and limitations (reviewed in detail in (
      • Scharfmann R.
      • Staels W.
      • Albagli O.
      The supply chain of human pancreatic beta cell lines.
      )). EndoC-βH1 cells have the advantage of being a pure cell population of human origin, which can be proliferated to yield large numbers of cells needed for in vitro and omics experiments. Unsurprisingly, EndoC-βH1 cells are among the most commonly used human β cell lines for early-stage diabetes research. Yet, despite extensive functional, morphological, and electrophysiological characterization (
      • Scharfmann R.
      • Staels W.
      • Albagli O.
      The supply chain of human pancreatic beta cell lines.
      ), EndoC-βH1 cells have not been validated as an in vitro human β cell model based on omics.
      Here, we confirm that EndoC-βH1 are cells of β cell lineage using transcriptomics and proteomics data. The observed discrepancies with adult human β cells and increased similarity with insulinomas were due to elevated proliferation rate (
      • Andersson L.E.
      • Valtat B.
      • Bagge A.
      • Sharoyko V.V.
      • Nicholls D.G.
      • Ravassard P.
      • Scharfmann R.
      • Spegel P.
      • Mulder H.
      Characterization of stimulus-secretion coupling in the human pancreatic EndoC-betaH1 beta cell line.
      ). Therefore, inducible growth arrest will likely ameliorate the observed discrepancies in gene and protein expression pattern. In fact, next-generation EndoC-βH2 and EndoC-βH3 cells display a dramatic increase in insulin content and secretion capacity after the immortalization cassette is excised (
      • Scharfmann R.
      • Pechberty S.
      • Hazhouz Y.
      • von Bulow M.
      • Bricout-Neveu E.
      • Grenier-Godard M.
      • Guez F.
      • Rachdi L.
      • Lohmann M.
      • Czernichow P.
      • Ravassard P.
      Development of a conditionally immortalized human pancreatic beta cell line.
      ,
      • Benazra M.
      • Lecomte M.J.
      • Colace C.
      • Muller A.
      • Machado C.
      • Pechberty S.
      • Bricout-Neveu E.
      • Grenier-Godard M.
      • Solimena M.
      • Scharfmann R.
      • Czernichow P.
      • Ravassard P.
      A human beta cell line with drug inducible excision of immortalizing transgenes.
      ). Increased expression of β cell markers postexcision has also been shown in EndoC-βH2 cells (
      • Scharfmann R.
      • Pechberty S.
      • Hazhouz Y.
      • von Bulow M.
      • Bricout-Neveu E.
      • Grenier-Godard M.
      • Guez F.
      • Rachdi L.
      • Lohmann M.
      • Czernichow P.
      • Ravassard P.
      Development of a conditionally immortalized human pancreatic beta cell line.
      ).
      The limited detection of GPCRs in our proteomics experiments might be of technical nature. In general, membrane proteins are hydrophobic, but the lysis buffers used in these proteomics experiments are likely not harsh enough to solubilize GPCRs. In addition, the hydrophobic nature of GPCRS leads to low frequency of polar amino acids (i.e., very few lysin and arginine residues), which affects tryptic digestion and subsequent detection by mass spectrometry (
      • Helbig A.O.
      • Heck A.J.
      • Slijper M.
      Exploring the membrane proteome--challenges and analytical strategies.
      ,
      • Zvonok N.
      • Xu W.
      • Williams J.
      • Janero D.R.
      • Krishnan S.C.
      • Makriyannis A.
      Mass spectrometry-based GPCR proteomics: Comprehensive characterization of the human cannabinoid 1 receptor.
      ). By contrast, low GLP1R expression appeared to be a true biological finding based on our data and validation in GSE133218. GLP1R expression may also be remedied by using next-generation cell lines. For example, improved response to GLP-1 mimetics has been reported by vendors in EndoC-βH5 cells (https://www.humancelldesign.com/human-beta-cells-endoc-bh5/).
      In summary, growth-arrested next-generation EndoC cells may be more suitable for some research questions such as studies of factors influencing proliferation of adult β cells, GLP1R-mediated drug delivery, or effects of GLP-1 mimetics. We believe that systematic comparison of growth-arrested and nongrowth–arrested next generation EndoC-βH2 to EndoC-βH5 cells to primary human β cells constitutes an important direction of future research.
      As EndoC-βH1 cells proved to be a reasonable translational human β cell model based on our omics analysis and prior functional studies (
      • Ravassard P.
      • Hazhouz Y.
      • Pechberty S.
      • Bricout-Neveu E.
      • Armanet M.
      • Czernichow P.
      • Scharfmann R.
      A genetically engineered human pancreatic beta cell line exhibiting glucose-inducible insulin secretion.
      ,
      • Scharfmann R.
      • Staels W.
      • Albagli O.
      The supply chain of human pancreatic beta cell lines.
      ,
      • Andersson L.E.
      • Valtat B.
      • Bagge A.
      • Sharoyko V.V.
      • Nicholls D.G.
      • Ravassard P.
      • Scharfmann R.
      • Spegel P.
      • Mulder H.
      Characterization of stimulus-secretion coupling in the human pancreatic EndoC-betaH1 beta cell line.
      ,
      • Hastoy B.
      • Godazgar M.
      • Clark A.
      • Nylander V.
      • Spiliotis I.
      • van de Bunt M.
      • Chibalina M.V.
      • Barrett A.
      • Burrows C.
      • Tarasov A.I.
      • Scharfmann R.
      • Gloyn A.L.
      • Rorsman P.
      Electrophysiological properties of human beta-cell lines EndoC-betaH1 and -betaH2 conform with human beta-cells.
      ,
      • Tsonkova V.G.
      • Sand F.W.
      • Wolf X.A.
      • Grunnet L.G.
      • Kirstine Ringgaard A.
      • Ingvorsen C.
      • Winkel L.
      • Kalisz M.
      • Dalgaard K.
      • Bruun C.
      • Fels J.J.
      • Helgstrand C.
      • Hastrup S.
      • Oberg F.K.
      • Vernet E.
      • et al.
      The EndoC-betaH1 cell line is a valid model of human beta cells and applicable for screenings to identify novel drug target candidates.
      ), we used EndoC-βH1 cells to uncover the β cell secretome. EndoC-βH1 cells secreted β cell hormones INS and IAPP. Both adult human β cells and insulinoma cells secrete low amounts of insulin under basal conditions and not only after glucose or KCl challenge (
      • Ramchandani N.
      • Ellis M.K.
      • Jain S.
      • Bhandari S.
      • Anhalt H.
      • Maclaren N.K.
      • Ten S.
      Basal insulin requirements on continuous subcutaneous insulin infusion during the first 12 months after diagnosis of type 1 diabetes mellitus.
      ,
      • Guettier J.M.
      • Gorden P.
      Insulin secretion and insulin-producing tumors.
      ). Hence, presence of insulin and co-secreted products IAPP and IGF2/preptin in the secretome of unstimulated EndoC-βH1 is not abnormal.
      Our secretome proteomics workflow was aimed at identification of secreted proteins and contained a tryptic digestion step. Thus, we could not establish full-length sequences of endogenously cleaved peptides. However, we could make inference based on the spectra and prior knowledge of protein families generating bioactive peptides. Our secretome data indicated that β cells may be capable of producing and secreting bioactive peptides. In fact, we confirmed the presence of preptin, neuroendocrine regulatory peptide-2, and TLQP-62 (
      • Cornish J.
      • Callon K.E.
      • Bava U.
      • Watson M.
      • Xu X.
      • Lin J.M.
      • Chan V.A.
      • Grey A.B.
      • Naot D.
      • Buchanan C.M.
      • Cooper G.J.
      • Reid I.R.
      Preptin, another peptide product of the pancreatic beta-cell, is osteogenic in vitro and in vivo.
      ,
      • Moin A.S.
      • Yamaguchi H.
      • Rhee M.
      • Kim J.W.
      • Toshinai K.
      • Waise T.M.
      • Naznin F.
      • Matsuo T.
      • Sasaki K.
      • Minamino N.
      • Yoon K.H.
      • Nakazato M.
      Neuroendocrine regulatory peptide-2 stimulates glucose-induced insulin secretion in vivo and in vitro.
      ,
      • Petrocchi-Passeri P.
      • Cero C.
      • Cutarelli A.
      • Frank C.
      • Severini C.
      • Bartolomucci A.
      • Possenti R.
      The VGF-derived peptide TLQP-62 modulates insulin secretion and glucose homeostasis.
      ) in conditioned medium from EndoC-βH1 cells. We also found that EndoC-βH1 cells secrete proteins of the granin family and possess the enzymatic machinery necessary to produce bioactive peptides, which suggests that the endogenous bioactive peptide repertoire of β cells might be richer than previously anticipated. Bioactive peptides may play an important role in both autocrine and endocrine signaling as well as crosstalk between β cells and other islets cell types. We believe that presence of full-length endogenous peptides and their secretion dynamics and function warrant further investigation in subsequent studies with specialized proteomics methods.
      The secretome of EndoC-βH1 encompassed classically secreted proteins with signal peptide. However, we speculate that these secreted proteins may play a limited role as biomedically relevant biomarkers. Only a few proteins in the secretome had enriched expression in β cells (INS, IAPP) or nervous system and β cells (e.g. PTPRN, VGF) (
      GTEx Consortium
      Human genomics. The genotype-tissue expression (GTEx) pilot analysis: Multitissue gene regulation in humans.
      ). In addition, the relative contribution of β cells to the plasma pool of broadly expressed proteins in human is unknown, and deconvoluting tissue-of-origin for such blood-borne biomarkers can be extremely challenging. By contrast, classically secreted proteins may play an autocrine or paracrine function. For example, the functions of NENF, MANF, MYDGF, and CREG1 as regulators of cell survival and differentiation were described in a different tissue context (
      UniProt Consortium
      UniProt: The universal protein knowledgebase in 2021.
      ), but they may play similar roles in islets. In fact, MANF has been recently characterized as a β cell protecting factor (
      • Hakonen E.
      • Chandra V.
      • Fogarty C.L.
      • Yu N.Y.
      • Ustinov J.
      • Katayama S.
      • Galli E.
      • Danilova T.
      • Lindholm P.
      • Vartiainen A.
      • Einarsdottir E.
      • Krjutskov K.
      • Kere J.
      • Saarma M.
      • Lindahl M.
      • et al.
      MANF protects human pancreatic beta cells against stress-induced cell death.
      ).
      Proteins, that are typically annotated as intracellular, can be found in conditioned media either due to unconventional secretion via EV or because of leakage. Computational predictions with two methods indicated high proportion of EV-associated protein species in both untransfected day 0 cells (not stressed) and cells treated with control siRNA (a nonspecific stressor). It was an unexpected finding because we did not specifically enrich for EV in our secretome proteomics workflow. However, we confirmed lack of effect of siRNA transfection on proliferation, apoptosis, and β cell markers defining the cell identity. Secreted apoptosis-associated proteins accounted for 0% to <2% of the protein intensities in the secretome of untransfected cells and cells treated with control siRNA, respectively. Furthermore, abundances of proteins annotated as cytoplasmic in UniProt did not correlate between paired cell lysate and secretome samples (PXD027911 versus PXD027913). Therefore, leakage was unlikely.
      EVs are released from the plasma membrane via exocytosis. β cells store insulin in intracellular vesicles called insulin granules and secrete insulin by transporting insulin granules to the plasma membrane and subsequent exocytosis (
      • Hou J.C.
      • Min L.
      • Pessin J.E.
      Insulin granule biogenesis, trafficking and exocytosis.
      ). However, it was not clear from current literature if the content of insulin granules is spilled into the extracellular space or if the granules retain membrane and persist as EVs after they are secreted. The proteomics studies on insulin granules isolated from whole-cell rodent β cell lines report only 50 to 140 proteins (
      • Norris N.
      • Yau B.
      • Kebede M.A.
      Isolation and proteomics of the insulin secretory granule.
      ) and do not correlate with the repertoire of EV-associated proteins in our secretome samples. Therefore, we speculate that our secretome preparations represented the total EndoC-βH1 secretome including a diluted EV fraction that is not identical to insulin granules.
      EV may offer additional opportunities to discover and noninvasively monitor biomarkers reflecting β cell state. Therefore, future studies need to further elucidate composition of β cell EV with dedicated EV isolation and proteomics workflows, to establish EV secretion dynamics and if it is possible to enrich β cell derived EV from plasma, serum, or fecal samples based on additional biomarkers.

      Data Availability

      RNA-seq data have been deposited to NCBI GEO (
      • Barrett T.
      • Wilhite S.E.
      • Ledoux P.
      • Evangelista C.
      • Kim I.F.
      • Tomashevsky M.
      • Marshall K.A.
      • Phillippy K.H.
      • Sherman P.M.
      • Holko M.
      • Yefanov A.
      • Lee H.
      • Zhang N.
      • Robertson C.L.
      • Serova N.
      • et al.
      NCBI GEO: Archive for functional genomics data sets--update.
      ) with accession GSE182016.
      The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (
      • Perez-Riverol Y.
      • Csordas A.
      • Bai J.
      • Bernal-Llinares M.
      • Hewapathirana S.
      • Kundu D.J.
      • Inuganti A.
      • Griss J.
      • Mayer G.
      • Eisenacher M.
      • Perez E.
      • Uszkoreit J.
      • Pfeuffer J.
      • Sachsenberg T.
      • Yilmaz S.
      • et al.
      The PRIDE database and related tools and resources in 2019: Improving support for quantification data.
      ) partner repository with the following identifiers: PXD027898, PXD027921, PXD027920, PXD027911, and PXD027913. Protein quantification tables for each experiment and a summary table are provided as Supplemental Data in this article.
      The previously published datasets used as reference primary human biological material were GTEx (
      GTEx Consortium
      Human genomics. The genotype-tissue expression (GTEx) pilot analysis: Multitissue gene regulation in humans.
      ), Human Protein Atlas (HPA) (
      • Uhlen M.
      • Karlsson M.J.
      • Hober A.
      • Svensson A.S.
      • Scheffel J.
      • Kotol D.
      • Zhong W.
      • Tebani A.
      • Strandberg L.
      • Edfors F.
      • Sjostedt E.
      • Mulder J.
      • Mardinoglu A.
      • Berling A.
      • Ekblad S.
      • et al.
      The human secretome.
      ), GSE67543 (
      • Blodgett D.M.
      • Nowosielska A.
      • Afik S.
      • Pechhold S.
      • Cura A.J.
      • Kennedy N.J.
      • Kim S.
      • Kucukural A.
      • Davis R.J.
      • Kent S.C.
      • Greiner D.L.
      • Garber M.G.
      • Harlan D.M.
      • diIorio P.
      Novel observations from next-generation RNA sequencing of highly purified human adult and fetal islet cell subsets.
      ), GSE57973 (
      • Arda H.E.
      • Li L.
      • Tsai J.
      • Torre E.A.
      • Rosli Y.
      • Peiris H.
      • Spitale R.C.
      • Dai C.
      • Gu X.
      • Qu K.
      • Wang P.
      • Wang J.
      • Grompe M.
      • Scharfmann R.
      • Snyder M.S.
      • et al.
      Age-dependent pancreatic gene regulation reveals mechanisms governing human beta cell function.
      ), GSE118014 (
      • Chan C.S.
      • Laddha S.V.
      • Lewis P.W.
      • Koletsky M.S.
      • Robzyk K.
      • Da Silva E.
      • Torres P.J.
      • Untch B.R.
      • Li J.
      • Bose P.
      • Chan T.A.
      • Klimstra D.S.
      • Allis C.D.
      • Tang L.H.
      ATRX, DAXX or MEN1 mutant pancreatic neuroendocrine tumors are a distinct alpha-cell signature subgroup.
      ), GSE84133 (
      • Baron M.
      • Veres A.
      • Wolock S.L.
      • Faust A.L.
      • Gaujoux R.
      • Vetere A.
      • Ryu J.H.
      • Wagner B.K.
      • Shen-Orr S.S.
      • Klein A.M.
      • Melton D.A.
      • Yanai I.
      A single-cell transcriptomic map of the human and mouse pancreas reveals inter- and intra-cell population structure.
      ), GSE81608 (
      • Xin Y.
      • Kim J.
      • Okamoto H.
      • Ni M.
      • Wei Y.
      • Adler C.
      • Murphy A.J.
      • Yancopoulos G.D.
      • Lin C.
      • Gromada J.
      RNA sequencing of single human islet cells reveals type 2 diabetes genes.
      ), E−MTAB−5061 (
      • Segerstolpe A.
      • Palasantza A.
      • Eliasson P.
      • Andersson E.M.
      • Andreasson A.C.
      • Sun X.
      • Picelli S.
      • Sabirsh A.
      • Clausen M.
      • Bjursell M.K.
      • Smith D.M.
      • Kasper M.
      • Ammala C.
      • Sandberg R.
      Single-cell transcriptome profiling of human pancreatic islets in health and type 2 diabetes.
      ), GSE86469 (
      • Lawlor N.
      • George J.
      • Bolisetty M.
      • Kursawe R.
      • Sun L.
      • Sivakamasundari V.
      • Kycia I.
      • Robson P.
      • Stitzel M.L.
      Single-cell transcriptomes identify human islet cell signatures and reveal cell-type-specific expression changes in type 2 diabetes.
      ), PXD007569 (
      • Galli A.
      • Maffioli E.
      • Sogne E.
      • Moretti S.
      • Di Cairano E.S.
      • Negri A.
      • Nonnis S.
      • Norata G.D.
      • Bonacina F.
      • Borghi F.
      • Podesta A.
      • Bertuzzi F.
      • Milani P.
      • Lenardi C.
      • Tedeschi G.
      • et al.
      Cluster-assembled zirconia substrates promote long-term differentiation and functioning of human islets of Langerhans.
      ), and PXD013419 (
      • Qiao Z.
      • Zhang Y.
      • Ge M.
      • Liu S.
      • Jiang X.
      • Shang Z.
      • Liu H.
      • Cao C.
      • Xiao H.
      Cancer cell derived small extracellular vesicles contribute to recipient cell metastasis through promoting HGF/c-Met pathway.
      ). Marker lists were obtained from HGNC (
      • Tweedie S.
      • Braschi B.
      • Gray K.
      • Jones T.E.M.
      • Seal R.L.
      • Yates B.
      • Bruford E.A.
      Genenames.org: The HGNC and VGNC resources in 2021.
      ), OMIM (
      • Amberger J.S.
      • Bocchini C.A.
      • Scott A.F.
      • Hamosh A.
      OMIM.org: Leveraging knowledge across phenotype-gene relationships.
      ), GENECODE v36 (
      • Frankish A.
      • Diekhans M.
      • Ferreira A.M.
      • Johnson R.
      • Jungreis I.
      • Loveland J.
      • Mudge J.M.
      • Sisu C.
      • Wright J.
      • Armstrong J.
      • Barnes I.
      • Berry A.
      • Bignell A.
      • Carbonell Sala S.
      • Chrast J.
      • et al.
      GENCODE reference annotation for the human and mouse genomes.
      ) and supplementary datasets in (
      • Villarreal L.
      • Mendez O.
      • Salvans C.
      • Gregori J.
      • Baselga J.
      • Villanueva J.
      Unconventional secretion is a major contributor of cancer cell line secretomes.
      ,
      • Song Y.L.
      • Yu R.
      • Qiao X.W.
      • Bai C.M.
      • Lu C.M.
      • Xiao Y.
      • Zhong D.R.
      • Chen J.
      • Zhao Y.P.
      • Zhang T.P.
      • Song T.T.
      • Gao H.L.
      • Wan Y.H.
      • Shen L.
      • Chen J.
      • et al.
      Prognostic relevance of UCH-L1 and alpha-internexin in pancreatic neuroendocrine tumors.
      ,
      • Ryaboshapkina M.
      • Hammar M.
      Tissue-specific genes as an underutilized resource in drug discovery.
      ,
      • Ramond C.
      • Beydag-Tasoz B.S.
      • Azad A.
      • van de Bunt M.
      • Petersen M.B.K.
      • Beer N.L.
      • Glaser N.
      • Berthault C.
      • Gloyn A.L.
      • Hansson M.
      • McCarthy M.I.
      • Honore C.
      • Grapin-Botton A.
      • Scharfmann R.
      Understanding human fetal pancreas development using subpopulation sorting, RNA sequencing and single-cell profiling.
      ,
      • Lambert S.A.
      • Jolma A.
      • Campitelli L.F.
      • Das P.K.
      • Yin Y.
      • Albu M.
      • Chen X.
      • Taipale J.
      • Hughes T.R.
      • Weirauch M.T.
      The human transcription factors.
      ,
      • Vinuela A.
      • Varshney A.
      • van de Bunt M.
      • Prasad R.B.
      • Asplund O.
      • Bennett A.
      • Boehnke M.
      • Brown A.A.
      • Erdos M.R.
      • Fadista J.
      • Hansson O.
      • Hatem G.
      • Howald C.
      • Iyengar A.K.
      • Johnson P.
      • et al.
      Genetic variant effects on gene expression in human pancreatic islets and their implications for T2D.
      ).

      Supplemental data

      This article contains supplemental data.

      Conflict of interest

      S. A., K. Se., M. R., A. J., C. R. U., and G. H. are employed by AstraZeneca. R. S. is a shareholder in and a consultant for Univercell-Biosolutions.

      Funding and additional information

      The R. S. laboratory received funding from the Laboratoire d’Excellence consortium Revive, Fondation pour la Recherche Médicale (EQU201903007793), the Agence Nationale de la Recherche (ANR-15-CE17–0018–01), the Dutch Diabetes Research Foundation, the DON Foundation, and Fondation Francophone pour la Recherche sur le Diabetes (FFRD).

      Author contributions

      R. S., S. A., K. Se., and G. H. methodology; K. S., C. B., and S. P. investigation; G. H., M. R., and A. J., formal analysis; M. R. writing-original draft. R. S., S. A., K. Se., K. S., C. B., S. P., M. R., A. J., C. R. U., and G. H. data curation; R. S., C. R. U., S. A., K. Se., K. S., C. B., S. P., M. R., A. J., and G. H writing-review and editing.

      Supplemental Data

      References

        • Saeedi P.
        • Petersohn I.
        • Salpea P.
        • Malanda B.
        • Karuranga S.
        • Unwin N.
        • Colagiuri S.
        • Guariguata L.
        • Motala A.A.
        • Ogurtsova K.
        • Shaw J.E.
        • Bright D.
        • Williams R.
        • IDF Diabetes Atlas Committee
        Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: Results from the International Diabetes Federation Diabetes Atlas, 9(th) edition.
        Diabetes Res. Clin. Pract. 2019; 157: 107843
        • Nasteska D.
        • Hodson D.J.
        The role of beta cell heterogeneity in islet function and insulin release.
        J. Mol. Endocrinol. 2018; 61: R43-R60
        • Saponaro C.
        • Muhlemann M.
        • Acosta-Montalvo A.
        • Piron A.
        • Gmyr V.
        • Delalleau N.
        • Moerman E.
        • Thevenet J.
        • Pasquetti G.
        • Coddeville A.
        • Cnop M.
        • Kerr-Conte J.
        • Staels B.
        • Pattou F.
        • Bonner C.
        Interindividual heterogeneity of SGLT2 expression and function in human pancreatic islets.
        Diabetes. 2020; 69: 902-914
        • Negi S.
        • Jetha A.
        • Aikin R.
        • Hasilo C.
        • Sladek R.
        • Paraskevas S.
        Analysis of beta-cell gene expression reveals inflammatory signaling and evidence of dedifferentiation following human islet isolation and culture.
        PLoS One. 2012; 7e30415
        • Kim A.
        • Miller K.
        • Jo J.
        • Kilimnik G.
        • Wojcik P.
        • Hara M.
        Islet architecture: A comparative study.
        Islets. 2009; 1: 129-136
        • Campbell S.A.
        • Golec D.P.
        • Hubert M.
        • Johnson J.
        • Salamon N.
        • Barr A.
        • MacDonald P.E.
        • Philippaert K.
        • Light P.E.
        Human islets contain a subpopulation of glucagon-like peptide-1 secreting alpha cells that is increased in type 2 diabetes.
        Mol. Metab. 2020; 39: 101014
        • McCluskey J.T.
        • Hamid M.
        • Guo-Parke H.
        • McClenaghan N.H.
        • Gomis R.
        • Flatt P.R.
        Development and functional characterization of insulin-releasing human pancreatic beta cell lines produced by electrofusion.
        J. Biol. Chem. 2011; 286: 21982-21992
        • Kiselyuk A.
        • Farber-Katz S.
        • Cohen T.
        • Lee S.H.
        • Geron I.
        • Azimi B.
        • Heynen-Genel S.
        • Singer O.
        • Price J.
        • Mercola M.
        • Itkin-Ansari P.
        • Levine F.
        Phenothiazine neuroleptics signal to the human insulin promoter as revealed by a novel high-throughput screen.
        J. Biomol. Screen. 2010; 15: 663-670
        • Ravassard P.
        • Hazhouz Y.
        • Pechberty S.
        • Bricout-Neveu E.
        • Armanet M.
        • Czernichow P.
        • Scharfmann R.
        A genetically engineered human pancreatic beta cell line exhibiting glucose-inducible insulin secretion.
        J. Clin. Invest. 2011; 121: 3589-3597
        • Scharfmann R.
        • Pechberty S.
        • Hazhouz Y.
        • von Bulow M.
        • Bricout-Neveu E.
        • Grenier-Godard M.
        • Guez F.
        • Rachdi L.
        • Lohmann M.
        • Czernichow P.
        • Ravassard P.
        Development of a conditionally immortalized human pancreatic beta cell line.
        J. Clin. Invest. 2014; 124: 2087-2098
        • Benazra M.
        • Lecomte M.J.
        • Colace C.
        • Muller A.
        • Machado C.
        • Pechberty S.
        • Bricout-Neveu E.
        • Grenier-Godard M.
        • Solimena M.
        • Scharfmann R.
        • Czernichow P.
        • Ravassard P.
        A human beta cell line with drug inducible excision of immortalizing transgenes.
        Mol. Metab. 2015; 4: 916-925
        • Rachdi L.
        • Maugein A.
        • Pechberty S.
        • Armanet M.
        • Hamroune J.
        • Ravassard P.
        • Marullo S.
        • Albagli O.
        • Scharfmann R.
        Regulated expression and function of the GABAB receptor in human pancreatic beta cell line and islets.
        Sci. Rep. 2020; 10: 13469
        • Scharfmann R.
        • Staels W.
        • Albagli O.
        The supply chain of human pancreatic beta cell lines.
        J. Clin. Invest. 2019; 129: 3511-3520
        • Andersson L.E.
        • Valtat B.
        • Bagge A.
        • Sharoyko V.V.
        • Nicholls D.G.
        • Ravassard P.
        • Scharfmann R.
        • Spegel P.
        • Mulder H.
        Characterization of stimulus-secretion coupling in the human pancreatic EndoC-betaH1 beta cell line.
        PLoS One. 2015; 10e0120879
        • Hastoy B.
        • Godazgar M.
        • Clark A.
        • Nylander V.
        • Spiliotis I.
        • van de Bunt M.
        • Chibalina M.V.
        • Barrett A.
        • Burrows C.
        • Tarasov A.I.
        • Scharfmann R.
        • Gloyn A.L.
        • Rorsman P.
        Electrophysiological properties of human beta-cell lines EndoC-betaH1 and -betaH2 conform with human beta-cells.
        Sci. Rep. 2018; 8: 16994
        • Tsonkova V.G.
        • Sand F.W.
        • Wolf X.A.
        • Grunnet L.G.
        • Kirstine Ringgaard A.
        • Ingvorsen C.
        • Winkel L.
        • Kalisz M.
        • Dalgaard K.
        • Bruun C.
        • Fels J.J.
        • Helgstrand C.
        • Hastrup S.
        • Oberg F.K.
        • Vernet E.
        • et al.
        The EndoC-betaH1 cell line is a valid model of human beta cells and applicable for screenings to identify novel drug target candidates.
        Mol. Metab. 2018; 8: 144-157
        • Lawlor N.
        • Marquez E.J.
        • Orchard P.
        • Narisu N.
        • Shamim M.S.
        • Thibodeau A.
        • Varshney A.
        • Kursawe R.
        • Erdos M.R.
        • Kanke M.
        • Gu H.
        • Pak E.
        • Dutra A.
        • Russell S.
        • Li X.
        • et al.
        Multiomic profiling identifies cis-regulatory networks underlying human pancreatic beta cell identity and function.
        Cell Rep. 2019; 26: 788-801.e6
        • Colli M.L.
        • Ramos-Rodriguez M.
        • Nakayasu E.S.
        • Alvelos M.I.
        • Lopes M.
        • Hill J.L.E.
        • Turatsinze J.V.
        • Coomans de Brachene A.
        • Russell M.A.
        • Raurell-Vila H.
        • Castela A.
        • Juan-Mateu J.
        • Webb-Robertson B.M.
        • Krogvold L.
        • Dahl-Jorgensen K.
        • et al.
        An integrated multi-omics approach identifies the landscape of interferon-alpha-mediated responses of human pancreatic beta cells.
        Nat. Commun. 2020; 11: 2584
        • Ramos-Rodriguez M.
        • Raurell-Vila H.
        • Colli M.L.
        • Alvelos M.I.
        • Subirana-Granes M.
        • Juan-Mateu J.
        • Norris R.
        • Turatsinze J.V.
        • Nakayasu E.S.
        • Webb-Robertson B.M.
        • Inshaw J.R.J.
        • Marchetti P.
        • Piemonti L.
        • Esteller M.
        • Todd J.A.
        • et al.
        The impact of proinflammatory cytokines on the beta-cell regulatory landscape provides insights into the genetics of type 1 diabetes.
        Nat. Genet. 2019; 51: 1588-1595
        • Saitoski K.
        • Ryaboshapkina M.
        • Hamza G.M.
        • Jarnuczak A.F.
        • Berthault C.
        • Carlotti F.
        • Armanet M.
        • Sengupta K.
        • Rye Underwood C.
        • Andersson S.
        • Guillas I.
        • Le Goff W.
        • Scharfmann R.
        PCSK9 affects expression of key surface proteins in human pancreatic beta cells through intra- and extracellular regulatory circuits.
        bioRxiv. 2021; ([preprint])https://doi.org/10.1101/2021.09.28.461975
        • Tattikota S.G.
        • Sury M.D.
        • Rathjen T.
        • Wessels H.H.
        • Pandey A.K.
        • You X.
        • Becker C.
        • Chen W.
        • Selbach M.
        • Poy M.N.
        Argonaute2 regulates the pancreatic beta-cell secretome.
        Mol. Cell Proteomics. 2013; 12: 1214-1225
        • Stutzer I.
        • Selevsek N.
        • Esterhazy D.
        • Schmidt A.
        • Aebersold R.
        • Stoffel M.
        Systematic proteomic analysis identifies beta-site amyloid precursor protein cleaving enzyme 2 and 1 (BACE2 and BACE1) substrates in pancreatic beta-cells.
        J. Biol. Chem. 2013; 288: 10536-10547
        • Pepaj M.
        • Bredahl M.K.
        • Gjerlaugsen N.
        • Thorsby P.M.
        Proteomic analysis of the INS-1E secretome identify novel vitamin D-regulated proteins.
        Diabetes Metab. Res. Rev. 2016; 32: 514-521
        • Midha A.
        • Pan H.
        • Abarca C.
        • Andle J.
        • Carapeto P.
        • Bonner-Weir S.
        • Aguayo-Mazzucato C.
        Unique human and mouse beta-cell senescence-associated secretory phenotype (SASP) reveal conserved signaling pathways and heterogeneous factors.
        Diabetes. 2021; 70: 1098-1116
        • Oshima M.
        • Knoch K.P.
        • Diedisheim M.
        • Petzold A.
        • Cattan P.
        • Bugliani M.
        • Marchetti P.
        • Choudhary P.
        • Huang G.C.
        • Bornstein S.R.
        • Solimena M.
        • Albagli-Curiel O.
        • Scharfmann R.
        Virus-like infection induces human beta cell dedifferentiation.
        JCI Insight. 2018; 3e97732
        • Dobin A.
        • Davis C.A.
        • Schlesinger F.
        • Drenkow J.
        • Zaleski C.
        • Jha S.
        • Batut P.
        • Chaisson M.
        • Gingeras T.R.
        STAR: Ultrafast universal RNA-seq aligner.
        Bioinformatics. 2013; 29: 15-21
        • Li B.
        • Dewey C.N.
        RSEM: Accurate transcript quantification from RNA-seq data with or without a reference genome.
        BMC Bioinformatics. 2011; 12: 323
        • Tyanova S.
        • Temu T.
        • Cox J.
        The MaxQuant computational platform for mass spectrometry-based shotgun proteomics.
        Nat. Protoc. 2016; 11: 2301-2319
        • Rosenberger G.
        • Bludau I.
        • Schmitt U.
        • Heusel M.
        • Hunter C.L.
        • Liu Y.
        • MacCoss M.J.
        • MacLean B.X.
        • Nesvizhskii A.I.
        • Pedrioli P.G.A.
        • Reiter L.
        • Rost H.L.
        • Tate S.
        • Ting Y.S.
        • Collins B.C.
        • et al.
        Statistical control of peptide and protein error rates in large-scale targeted data-independent acquisition analyses.
        Nat. Methods. 2017; 14: 921-927
        • Elias J.E.
        • Gygi S.P.
        Target-decoy search strategy for mass spectrometry-based proteomics.
        Methods Mol. Biol. 2010; 604: 55-71
        • Ritchie M.E.
        • Phipson B.
        • Wu D.
        • Hu Y.
        • Law C.W.
        • Shi W.
        • Smyth G.K.
        Limma powers differential expression analyses for RNA-sequencing and microarray studies.
        Nucleic Acids Res. 2015; 43: e47
        • Verboven S.
        • Branden K.V.
        • Goos P.
        Sequential imputation for missing values.
        Comput. Biol. Chem. 2007; 31: 320-327
        • Wang S.
        • Li W.
        • Hu L.
        • Cheng J.
        • Yang H.
        • Liu Y.
        NAguideR: Performing and prioritizing missing value imputations for consistent bottom-up proteomic analyses.
        Nucleic Acids Res. 2020; 48: e83
        • Gene Ontology Consortium
        The gene ontology resource: Enriching a GOld mine.
        Nucleic Acids Res. 2021; 49: D325-D334
        • Carbon S.
        • Ireland A.
        • Mungall C.J.
        • Shu S.
        • Marshall B.
        • Lewis S.
        • AmiGO Hub
        • Web Presence Working Group
        AmiGO: Online access to ontology and annotation data.
        Bioinformatics. 2009; 25: 288-289
        • Mi H.
        • Ebert D.
        • Muruganujan A.
        • Mills C.
        • Albou L.P.
        • Mushayamaha T.
        • Thomas P.D.
        PANTHER version 16: A revised family classification, tree-based classification tool, enhancer regions and extensive API.
        Nucleic Acids Res. 2021; 49: D394-D403
        • R Core Team
        R: A Language and Environment for Statistical Computing.
        R Foundation for Statistical Computing, Vienna, Austria2020
        • Wickham H.
        ggplot2: Elegant Graphics for Data Analysis.
        Springer, New York, NY2016
        • Ras-Carmona A.
        • Gomez-Perosanz M.
        • Reche P.A.
        Prediction of unconventional protein secretion by exosomes.
        BMC Bioinformatics. 2021; 22: 333
        • Pang H.
        • Luo S.
        • Xiao Y.
        • Xia Y.
        • Li X.
        • Huang G.
        • Xie Z.
        • Zhou Z.
        Emerging roles of exosomes in T1DM.
        Front. Immunol. 2020; 11: 593348
        • MacLean B.
        • Tomazela D.M.
        • Shulman N.
        • Chambers M.
        • Finney G.L.
        • Frewen B.
        • Kern R.
        • Tabb D.L.
        • Liebler D.C.
        • MacCoss M.J.
        Skyline: An open source document editor for creating and analyzing targeted proteomics experiments.
        Bioinformatics. 2010; 26: 966-968
        • Bausch-Fluck D.
        • Goldmann U.
        • Muller S.
        • van Oostrum M.
        • Muller M.
        • Schubert O.T.
        • Wollscheid B.
        The in silico human surfaceome.
        Proc. Natl. Acad. Sci. U. S. A. 2018; 115: E10988-E10997
        • UniProt Consortium
        UniProt: The universal protein knowledgebase in 2021.
        Nucleic Acids Res. 2021; 49: D480-D489
        • Villarreal L.
        • Mendez O.
        • Salvans C.
        • Gregori J.
        • Baselga J.
        • Villanueva J.
        Unconventional secretion is a major contributor of cancer cell line secretomes.
        Mol. Cell Proteomics. 2013; 12: 1046-1060
        • Tweedie S.
        • Braschi B.
        • Gray K.
        • Jones T.E.M.
        • Seal R.L.
        • Yates B.
        • Bruford E.A.
        Genenames.org: The HGNC and VGNC resources in 2021.
        Nucleic Acids Res. 2021; 49: D939-D946
        • Qiao Z.
        • Zhang Y.
        • Ge M.
        • Liu S.
        • Jiang X.
        • Shang Z.
        • Liu H.
        • Cao C.
        • Xiao H.
        Cancer cell derived small extracellular vesicles contribute to recipient cell metastasis through promoting HGF/c-Met pathway.
        Mol. Cell Proteomics. 2019; 18: 1619-1629
        • Keerthikumar S.
        • Chisanga D.
        • Ariyaratne D.
        • Al Saffar H.
        • Anand S.
        • Zhao K.
        • Samuel M.
        • Pathan M.
        • Jois M.
        • Chilamkurti N.
        • Gangoda L.
        • Mathivanan S.
        ExoCarta: A Web-based compendium of exosomal cargo.
        J. Mol. Biol. 2016; 428: 688-692
        • Askeland A.
        • Borup A.
        • Ostergaard O.
        • Olsen J.V.
        • Lund S.M.
        • Christiansen G.
        • Kristensen S.R.
        • Heegaard N.H.H.
        • Pedersen S.
        Mass-spectrometry based proteome comparison of extracellular vesicle isolation methods: Comparison of ME-kit, size-exclusion chromatography, and high-speed centrifugation.
        Biomedicines. 2020; 8: 246
        • Blodgett D.M.
        • Nowosielska A.
        • Afik S.
        • Pechhold S.
        • Cura A.J.
        • Kennedy N.J.
        • Kim S.
        • Kucukural A.
        • Davis R.J.
        • Kent S.C.
        • Greiner D.L.
        • Garber M.G.
        • Harlan D.M.
        • diIorio P.
        Novel observations from next-generation RNA sequencing of highly purified human adult and fetal islet cell subsets.
        Diabetes. 2015; 64: 3172-3181
        • Galli A.
        • Maffioli E.
        • Sogne E.
        • Moretti S.
        • Di Cairano E.S.
        • Negri A.
        • Nonnis S.
        • Norata G.D.
        • Bonacina F.
        • Borghi F.
        • Podesta A.
        • Bertuzzi F.
        • Milani P.
        • Lenardi C.
        • Tedeschi G.
        • et al.
        Cluster-assembled zirconia substrates promote long-term differentiation and functioning of human islets of Langerhans.
        Sci. Rep. 2018; 8: 9979
        • Chan C.S.
        • Laddha S.V.
        • Lewis P.W.
        • Koletsky M.S.
        • Robzyk K.
        • Da Silva E.
        • Torres P.J.
        • Untch B.R.
        • Li J.
        • Bose P.
        • Chan T.A.
        • Klimstra D.S.
        • Allis C.D.
        • Tang L.H.
        ATRX, DAXX or MEN1 mutant pancreatic neuroendocrine tumors are a distinct alpha-cell signature subgroup.
        Nat. Commun. 2018; 9: 4158
        • Song Y.L.
        • Yu R.
        • Qiao X.W.
        • Bai C.M.
        • Lu C.M.
        • Xiao Y.
        • Zhong D.R.
        • Chen J.
        • Zhao Y.P.
        • Zhang T.P.
        • Song T.T.
        • Gao H.L.
        • Wan Y.H.
        • Shen L.
        • Chen J.
        • et al.
        Prognostic relevance of UCH-L1 and alpha-internexin in pancreatic neuroendocrine tumors.
        Sci. Rep. 2017; 7: 2205
        • Rutti S.
        • Sauter N.S.
        • Bouzakri K.
        • Prazak R.
        • Halban P.A.
        • Donath M.Y.
        In vitro proliferation of adult human beta-cells.
        PLoS One. 2012; 7e35801
        • Wang Z.
        • You J.
        • Xu S.
        • Hua Z.
        • Zhang W.
        • Deng T.
        • Fang N.
        • Fang Q.
        • Liu H.
        • Peng L.
        • Wang P.
        • Lou J.
        Colocalization of insulin and glucagon in insulinoma cells and developing pancreatic endocrine cells.
        Biochem. Biophys. Res. Commun. 2015; 461: 598-604
        • Modi H.
        • Cornu M.
        • Thorens B.
        Glutamine stimulates biosynthesis and secretion of insulin-like growth factor 2 (IGF2), an autocrine regulator of beta cell mass and function.
        J. Biol. Chem. 2014; 289: 31972-31982
        • Bartolomucci A.
        • Possenti R.
        • Mahata S.K.
        • Fischer-Colbrie R.
        • Loh Y.P.
        • Salton S.R.
        The extended granin family: Structure, function, and biomedical implications.
        Endocr. Rev. 2011; 32: 755-797
        • Cornish J.
        • Callon K.E.
        • Bava U.
        • Watson M.
        • Xu X.
        • Lin J.M.
        • Chan V.A.
        • Grey A.B.
        • Naot D.
        • Buchanan C.M.
        • Cooper G.J.
        • Reid I.R.
        Preptin, another peptide product of the pancreatic beta-cell, is osteogenic in vitro and in vivo.
        Am. J. Physiol. Endocrinol. Metab. 2007; 292: E117-E122
        • Moin A.S.
        • Yamaguchi H.
        • Rhee M.
        • Kim J.W.
        • Toshinai K.
        • Waise T.M.
        • Naznin F.
        • Matsuo T.
        • Sasaki K.
        • Minamino N.
        • Yoon K.H.
        • Nakazato M.
        Neuroendocrine regulatory peptide-2 stimulates glucose-induced insulin secretion in vivo and in vitro.
        Biochem. Biophys. Res. Commun. 2012; 428: 512-517
        • Petrocchi-Passeri P.
        • Cero C.
        • Cutarelli A.
        • Frank C.
        • Severini C.
        • Bartolomucci A.
        • Possenti R.
        The VGF-derived peptide TLQP-62 modulates insulin secretion and glucose homeostasis.
        J. Mol. Endocrinol. 2015; 54: 227-239
        • Uhlen M.
        • Karlsson M.J.
        • Hober A.
        • Svensson A.S.
        • Scheffel J.
        • Kotol D.
        • Zhong W.
        • Tebani A.
        • Strandberg L.
        • Edfors F.
        • Sjostedt E.
        • Mulder J.
        • Mardinoglu A.
        • Berling A.
        • Ekblad S.
        • et al.
        The human secretome.
        Sci. Signal. 2019; 12eaaz0274
        • Yesildag B.
        • Bock T.
        • Herrmanns K.
        • Wollscheid B.
        • Stoffel M.
        Kin of IRRE-like protein 2 is a phosphorylated glycoprotein that regulates basal insulin secretion.
        J. Biol. Chem. 2015; 290: 25891-25906
        • Inoue T.
        • Hagiyama M.
        • Yoneshige A.
        • Kato T.
        • Enoki E.
        • Maenishi O.
        • Chikugo T.
        • Kimura M.
        • Satou T.
        • Ito A.
        Increased ectodomain shedding of cell adhesion molecule 1 from pancreatic islets in type 2 diabetic pancreata: Correlation with hemoglobin A1c levels.
        PLoS One. 2014; 9e100988
        • Helbig A.O.
        • Heck A.J.
        • Slijper M.
        Exploring the membrane proteome--challenges and analytical strategies.
        J. Proteomics. 2010; 73: 868-878
        • Zvonok N.
        • Xu W.
        • Williams J.
        • Janero D.R.
        • Krishnan S.C.
        • Makriyannis A.
        Mass spectrometry-based GPCR proteomics: Comprehensive characterization of the human cannabinoid 1 receptor.
        J. Proteome Res. 2010; 9: 1746-1753
        • Ramchandani N.
        • Ellis M.K.
        • Jain S.
        • Bhandari S.
        • Anhalt H.
        • Maclaren N.K.
        • Ten S.
        Basal insulin requirements on continuous subcutaneous insulin infusion during the first 12 months after diagnosis of type 1 diabetes mellitus.
        J. Diabetes Sci. Technol. 2010; 4: 610-614
        • Guettier J.M.
        • Gorden P.
        Insulin secretion and insulin-producing tumors.
        Expert Rev. Endocrinol. Metab. 2010; 5: 217-227
        • GTEx Consortium
        Human genomics. The genotype-tissue expression (GTEx) pilot analysis: Multitissue gene regulation in humans.
        Science. 2015; 348: 648-660
        • Hakonen E.
        • Chandra V.
        • Fogarty C.L.
        • Yu N.Y.
        • Ustinov J.
        • Katayama S.
        • Galli E.
        • Danilova T.
        • Lindholm P.
        • Vartiainen A.
        • Einarsdottir E.
        • Krjutskov K.
        • Kere J.
        • Saarma M.
        • Lindahl M.
        • et al.
        MANF protects human pancreatic beta cells against stress-induced cell death.
        Diabetologia. 2018; 61: 2202-2214
        • Hou J.C.
        • Min L.
        • Pessin J.E.
        Insulin granule biogenesis, trafficking and exocytosis.
        Vitam Horm. 2009; 80: 473-506
        • Norris N.
        • Yau B.
        • Kebede M.A.
        Isolation and proteomics of the insulin secretory granule.
        Metabolites. 2021; 11: 288
        • Barrett T.
        • Wilhite S.E.
        • Ledoux P.
        • Evangelista C.
        • Kim I.F.
        • Tomashevsky M.
        • Marshall K.A.
        • Phillippy K.H.
        • Sherman P.M.
        • Holko M.
        • Yefanov A.
        • Lee H.
        • Zhang N.
        • Robertson C.L.
        • Serova N.
        • et al.
        NCBI GEO: Archive for functional genomics data sets--update.
        Nucleic Acids Res. 2013; 41: D991-D995
        • Perez-Riverol Y.
        • Csordas A.
        • Bai J.
        • Bernal-Llinares M.
        • Hewapathirana S.
        • Kundu D.J.
        • Inuganti A.
        • Griss J.
        • Mayer G.
        • Eisenacher M.
        • Perez E.
        • Uszkoreit J.
        • Pfeuffer J.
        • Sachsenberg T.
        • Yilmaz S.
        • et al.
        The PRIDE database and related tools and resources in 2019: Improving support for quantification data.
        Nucleic Acids Res. 2019; 47: D442-D450
        • Arda H.E.
        • Li L.
        • Tsai J.
        • Torre E.A.
        • Rosli Y.
        • Peiris H.
        • Spitale R.C.
        • Dai C.
        • Gu X.
        • Qu K.
        • Wang P.
        • Wang J.
        • Grompe M.
        • Scharfmann R.
        • Snyder M.S.
        • et al.
        Age-dependent pancreatic gene regulation reveals mechanisms governing human beta cell function.
        Cell Metab. 2016; 23: 909-920
        • Baron M.
        • Veres A.
        • Wolock S.L.
        • Faust A.L.
        • Gaujoux R.
        • Vetere A.
        • Ryu J.H.
        • Wagner B.K.
        • Shen-Orr S.S.
        • Klein A.M.
        • Melton D.A.
        • Yanai I.
        A single-cell transcriptomic map of the human and mouse pancreas reveals inter- and intra-cell population structure.
        Cell Syst. 2016; 3: 346-360.e4
        • Xin Y.
        • Kim J.
        • Okamoto H.
        • Ni M.
        • Wei Y.
        • Adler C.
        • Murphy A.J.
        • Yancopoulos G.D.
        • Lin C.
        • Gromada J.
        RNA sequencing of single human islet cells reveals type 2 diabetes genes.
        Cell Metab. 2016; 24: 608-615
        • Segerstolpe A.
        • Palasantza A.
        • Eliasson P.
        • Andersson E.M.
        • Andreasson A.C.
        • Sun X.
        • Picelli S.
        • Sabirsh A.
        • Clausen M.
        • Bjursell M.K.
        • Smith D.M.
        • Kasper M.
        • Ammala C.
        • Sandberg R.
        Single-cell transcriptome profiling of human pancreatic islets in health and type 2 diabetes.
        Cell Metab. 2016; 24: 593-607
        • Lawlor N.
        • George J.
        • Bolisetty M.
        • Kursawe R.
        • Sun L.
        • Sivakamasundari V.
        • Kycia I.
        • Robson P.
        • Stitzel M.L.
        Single-cell transcriptomes identify human islet cell signatures and reveal cell-type-specific expression changes in type 2 diabetes.
        Genome Res. 2017; 27: 208-222
        • Amberger J.S.
        • Bocchini C.A.
        • Scott A.F.
        • Hamosh A.
        OMIM.org: Leveraging knowledge across phenotype-gene relationships.
        Nucleic Acids Res. 2019; 47: D1038-D1043
        • Frankish A.
        • Diekhans M.
        • Ferreira A.M.
        • Johnson R.
        • Jungreis I.
        • Loveland J.
        • Mudge J.M.
        • Sisu C.
        • Wright J.
        • Armstrong J.
        • Barnes I.
        • Berry A.
        • Bignell A.
        • Carbonell Sala S.
        • Chrast J.
        • et al.
        GENCODE reference annotation for the human and mouse genomes.
        Nucleic Acids Res. 2019; 47: D766-D773
        • Ryaboshapkina M.
        • Hammar M.
        Tissue-specific genes as an underutilized resource in drug discovery.
        Sci. Rep. 2019; 9: 7233
        • Ramond C.
        • Beydag-Tasoz B.S.
        • Azad A.
        • van de Bunt M.
        • Petersen M.B.K.
        • Beer N.L.
        • Glaser N.
        • Berthault C.
        • Gloyn A.L.
        • Hansson M.
        • McCarthy M.I.
        • Honore C.
        • Grapin-Botton A.
        • Scharfmann R.
        Understanding human fetal pancreas development using subpopulation sorting, RNA sequencing and single-cell profiling.
        Development. 2018; 145: dev165480
        • Lambert S.A.
        • Jolma A.
        • Campitelli L.F.
        • Das P.K.
        • Yin Y.
        • Albu M.
        • Chen X.
        • Taipale J.
        • Hughes T.R.
        • Weirauch M.T.
        The human transcription factors.
        Cell. 2018; 172: 650-665
        • Vinuela A.
        • Varshney A.
        • van de Bunt M.
        • Prasad R.B.
        • Asplund O.
        • Bennett A.
        • Boehnke M.
        • Brown A.A.
        • Erdos M.R.
        • Fadista J.
        • Hansson O.
        • Hatem G.
        • Howald C.
        • Iyengar A.K.
        • Johnson P.
        • et al.
        Genetic variant effects on gene expression in human pancreatic islets and their implications for T2D.
        Nat. Commun. 2020; 11: 4912