Abstract

Objective

   Ependymin-Related Protein 1 (EPDR1) was recently identified as a
   secreted human batokine regulating mitochondrial respiration linked to
   thermogenesis in brown fat. Despite that EPDR1 is expressed in human
   pancreatic β-cells and that glucose-stimulated mitochondrial metabolism
   is critical for stimulus-secretion coupling in β-cells, the role of
   EPDR1 in β-cell metabolism and function has not been investigated.

Methods

   EPDR1 mRNA levels in human pancreatic islets from non-diabetic (ND) and
   type 2 diabetes (T2D) subjects were assessed. Human islets, EndoC-βH1
   and INS1 832/13 cells were transfected with scramble (control) and
   EPDR1 siRNAs (EPDR1-KD) or treated with human EPDR1 protein, and
   glucose-stimulated insulin secretion (GSIS) assessed by ELISA.
   Mitochondrial metabolism was investigated by extracellular flux
   analyzer, confocal microscopy and mass spectrometry-based metabolomics
   analysis.

Results

   EPDR1 mRNA expression was upregulated in human islets from T2D and
   obese donors and positively correlated to BMI of donors. In T2D donors,
   EPDR1 mRNA levels negatively correlated with HbA1c and positively
   correlated with GSIS. EPDR1 silencing in human islets and β-cell lines
   reduced GSIS whereas treatment with human EPDR1 protein increased GSIS.
   Epdr1 silencing in INS1 832/13 cells reduced glucose- and pyruvate- but
   not K^+-stimulated insulin secretion. Metabolomics analysis in Epdr1-KD
   INS1 832/13 cells suggests diversion of glucose-derived pyruvate to
   lactate production and decreased malate-aspartate shuttle and the
   tricarboxylic acid (TCA) cycle activity. The glucose-stimulated rise in
   mitochondrial respiration and ATP/ADP-ratio was impaired in
   Epdr1-deficient cells.

Conclusion

   These results suggests that to maintain glucose homeostasis in obese
   people, upregulation of EPDR1 may improve β-cell function via
   channelling glycolysis-derived pyruvate to the mitochondrial TCA cycle.

   Keywords: Mitochondrial metabolism, Lactate, TCA cycle, Beta cells,
   Insulin secretion, Type 2 diabetes

Highlights

     * •
       EPDR1 expression is upregulated in pancreatic islets from obese and
       T2D donors.
     * •
       EPDR1 positively regulates insulin secretion in human pancreatic
       islets and β-cell lines.
     * •
       EPDR1 promotes the coupling of glycolisis to mitochondrial TCA
       cycle.
     * •
       EPDR1 is required for the glucose-stimulated increases of cytosolic
       ATP/ADP-ratio.

1. Introduction

   Type 2 Diabetes (T2D) results from an incapacity of β-cells to secrete
   sufficient amounts of insulin to compensate for peripheral insulin
   resistance. Glucose-induced mitochondrial metabolism is central for the
   stimulus-secretion coupling mechanism [[49]1]. Thus, alteration of
   mitochondrial metabolism in pancreatic β-cells impairs the
   glucose-stimulated insulin secretion (GSIS) [[50]2] and is a hallmark
   of T2D [[51]3]. Therefore, a comprehensive understanding of
   glucose-induced β-cell mitochondrial metabolism is required to fully
   clarify the pathophysiology of T2D.

   EPDR1 was first reported as the “mammalian homolog of ependymins”
   (MERP1) in 2001 [[52]4]. Ependymins are glycoproteins synthetized by
   fibroblasts and secreted into cerebrospinal fluids in fish [[53]4].
   Genetically, EPDR1 has been implicated in disease, including several
   types of cancer [[54][5], [55][6], [56][7], [57][8]], osteoporosis
   [[58]9], Dupuytren’s disease [[59]10,[60]11] and primary angle closure
   glaucoma [[61]12,[62]13]. Despite this notion, the role of EPDR1 in
   human physiology remain elusive.

   The crystal structure of human EPDR1 revealed hydrophobic binding
   grooves that can interact with lipids, suggesting a potential role of
   EPDR1 in lipid transport or degradation [[63]14]. Proteomics studies
   of mammalian mannose-6-phosphate (M6P) glycoproteins have identified
   EPDR1 as a lysosomal protein with unknown function [[64][15], [65][16],
   [66][17]]. The lysosomal localization of EPDR1 was further demonstrated
   in mouse brain homogenates by subcellular fractionation [[67]18].
   Recently, another proteomics-based study, identified EPDR1 as a protein
   secreted by hBAT, suggesting that EPDR1 is a novel human batokine
   [[68]19]. The protein was detected in human plasma samples suggesting
   it is involved in metabolic crosstalk between tissues [[69]17].
   EPDR1-silencing in human brown adipocytes reduces norepinephrine
   (NE)-induced mitochondrial proton-leak respiration and affects the
   expression of mitochondrial proteins [[70]19]. Collectively, these data
   suggest that EPDR1 regulates whole-body energy metabolism by modulating
   mitochondrial metabolism in target cells.

   RNAseq analysis of purified human fetal and adult pancreatic islet
   cells, shows low EPDR1 mRNA levels in fetal α and β-cells. In contrast,
   in adult β-cells, its expression increases whereas it decreases in
   adult α-cells [[71]20]. These data suggest that EPDR1 may play a role
   in mature β-cell function and may contribute to maintenance of glucose
   homeostasis in vivo.

   In the present study, we have studied the role of EPDR1 in regulating
   β-cell metabolism and function in human pancreatic islets and β-cell
   models. Our data indicate that EPDR1 is required for adequate GSIS,
   participating in a mechanism to channel glucose metabolism via the
   mitochondrial TCA cycle. We propose that EPDR1 is critical in
   maintaining β-cell glucose-responsiveness under increased metabolic
   demands imposed by obesity.

2. Research design and methods

2.1. Human pancreatic Islet cohort

   Human pancreatic islets (n = 188) were obtained from the EXODIAB Human
   Tissue Laboratory, which receives islets from the Nordic Network for
   Clinical Islet Transplantation ([72]http://www.nordicislets.org). The
   Clinical characteristics of the donors of islets and all methodological
   details related to RNA sequencing and genome-wide genotyping in islets
   from this cohort (188 donors) have recently been published [[73]21].
   For the correlational analysis of EPDR1 mRNA levels with different
   metabolic/functional parameters separated by disease status and with
   the purpose of ensure high quality data, a subgroup of donors was
   filtered under the following criteria: islets with purity≥70% and with
   days in culture≤5. The characteristics of donors for this specific
   subgroup is shown in [74]Table S1.

2.2. Cell culture

   INS1 832/13 and EndoC-βH1 cells were cultured as previously described
   [[75]22,[76]23]. EndoC-βH1 cells were cultured in
   Matrigel/fibronectin-coated (100/2 mg/mL, Sigma-Aldrich) flasks with
   DMEM containing 5.6 mmol/L glucose, 2% BSA, 10 mmol/L nicotinamide,
   50 mmol/L β-mercaptoethanol, 5.5 mg/mL transferrin, 6.7 ng/mL sodium
   selenite, 100 IU/mL penicillin, and 100 mg/mL streptomycin at 37 °C in
   a humidified atmosphere with 5% CO[2].

2.3. GSIS in β-cell lines and human islets

   INS1 832/13 cells (1.7 × 10^5 cells/cm^2) and EndoC-βH1 cells
   (1.5 × 10^5 cells/cm^2) were seeded in 24 and 48 well-plates,
   respectively, 96 h before experiments were performed. Seeded cells and
   human pancreatic islets (10 islets/well, 24-WP, 4–6
   replicates/condition) were starved in secretion assay buffer (SAB)
   buffer (final SAB 1X composition was (in mM): 114 NaCl; 4.7 KCl; 1.2
   KH2PO4; 1.16 MgSO4; 20 HEPES; 2.5 CaCl2; 25.5 NaHCO3; 0.2% bovine serum
   albumin (BSA), pH 7.2–7.4) with low glucose (LG, 1–2.8 mM). Then, cells
   and islets were simultaneously (in parallel wells) stimulated for 1 h
   with SAB with LG, high glucose (HG, 16.7–20 mM) or
   HG+3-isobutyl-1-methylxanthine (IBMX) (100 μM). The insulin
   concentration in the supernatants (accumulated for 1 h of stimulation)
   was measured using the insulin ELISA: for samples from INS1 832/13
   cells (non-diluted) High Range Rat Insulin ELISA (Mercodia, catalogue
   number (#): 10-1145-01), for samples from EndoC-βH1 cells (dilution
   factor: 12.5) and from human islets (dilution factor: 5) the Human
   Insulin Elisa (Mercodia, #10-1113-10). In EndoC-βH1 cells and human
   islets, the total insulin content was measured by extracting total
   cells/islet protein with RIPA buffer (composition was: 150 mM NaCl, 1%
   NP40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM TRIS–HCl, pH 8.0) and
   insulin secretion was normalized to the total insulin content and
   expressed as % secreted insulin.

2.4. EPDR1 protein treatment and silencing of β-cell lines and human islets

   INS1 832/13 cells (1.7 × 10^5 cells/cm^2) and EndoC-βH1 cells
   (1.5 × 10^5 cells/cm^2) were seeded in 24 and 48 well-plates whereas
   400–500 human pancreatic islets by condition were seeded in 35 mm petri
   dish and treated with vehicle (PBS) or with increasing concentrations
   (1.25–10 μg/mL) of human recombinant EPDR1 protein (UNIPROT entry
   [77]Q9UM22, isoform 1, amino acids 39–224 [[78]19]) for 48 h prior to
   GSIS assays. Details of donors of pancreatic islets treated with EPDR1
   protein are shown in [79]Table S2. For Epdr1 silencing, INS1 832/13
   cells (1.7 × 10^5 cells/cm^2) were seeded in 24 well plates and
   transfected after 24 h with 10 nM of scramble negative control (NC)
   (Thermofisher, custom select siRNA: 5′- GAGACCCUAUCCGUGAUUAUU-3′) or
   rat Epdr1 siRNA (Thermofisher, #4390771; s145532). EndoC-βH1 cells
   (1.5 × 10^5 cells/cm^2) were seeded in 48 well plates and transfected
   24 and 48 h after (two successive days, double shot) with 40 nM of
   Silencer Select Negative Control No. 1 siRNA (Thermofisher, #4390843)
   or human EPDR1 siRNA (Thermofisher, # 4392420; s29379). Human
   pancreatic islets (400–500) were seeded in 35 mm petri dish containing
   2 mL of RPMI medium (5 mM glucose, 10% FBS (Sigma #7524), 200 mM
   l-Glutamine) and transfected on two successive days with 0.5 mL of
   50 nM of a Silencer Select Negative Control No. 1 siRNA (Thermofisher,
   #4390843) and human EPDR1 siRNA (Thermofisher, #4390771; s29379), using
   Lipofectamine RNAiMAX Transfection Reagent (Thermofisher, #13778075).
   Details of donors of pancreatic islets used for experiments of EPDR1
   silencing are given in [80]Table S3. Human pancreatic islets were
   cultured as previously described [[81]24]. All procedures were approved
   by appropriate Swedish Ethical Committees.

2.5. mRNA analysis

   Total RNA was extracted using the RNeasy Mini Kit (Qiagen) and cDNA was
   generated with the RevertAid First-Strand cDNA synthesis kit. 10 ng
   cDNA/well (in duplicates) were used for qPCR (Epdr1 rat (Rn01525240_m1)
   and human (Hs01556067_m1) taqman assay were obtained from
   Thermofisher). qPCR assays were performed using the Applied Biosystems
   QuantStudio7 Flex Real-Time PCR System (Thermofisher). Relative gene
   abundance was calculated using the ΔΔCt method with β-actin as
   housekeeping reference gene and expressed as fold change to control.

2.6. Secreted lactate measurement

   INS1 832/13 cells (1.7 × 10^5 cells/cm^2) were seeded in 24 well
   plates, transfected with scramble (negative control, NC) or Epdr1 siRNA
   (10 nM) (as described in section [82]2.3) and GSIS determined (as
   described in section [83]2.2). The concentration of the extracellular
   lactate accumulated during 1 h in response to stimulation with SAB
   buffer with LG and HG glucose was measured by use of the Lactate
   Colorimetric/Fluorometric Assay Kit (Biovision, #607-100) following the
   manufacturer’s instructions. Briefly, cell conditioned samples and
   lactate standards were mixed with an enzyme plus fluorescent probe mix
   in a ratio 1:1, incubated for 30 min (at room temperature) and
   Fluorescence emission read (Ex/Em = 535/590 nm) in a microplate reader.
   Lactate concentrations were normalized by total cell mass in respective
   wells (protein mass measured by Pierce BCA Protein Assay Kit).

2.7. Metabolomics analysis

   The cellular metabolome was analyzed by combined gas
   chromatography-mass spectrometry (GC-MS) as previously described
   [[84]25]. Briefly, INS1 832/13 cells were stimulated with SAB LG, HG or
   HG + IBMX for 1 h (as described for GSIS above) and total cell
   metabolites extracted with methanol/water (9/1, v/v) containing a
   cocktail of stable isotope labelled internal standards
   (^2H[4]-succinate (0.22 mg/L), ^2H[8]-valine (4.41 mg/L),
   ^13C[4]-3-hydroxybutyric acid (2.21 mg/L), ^13C[5]^15N-glutamic acid
   (22.06 mg/L), ^2H[19]-decanoic acid (4.41 mg/L) and ^2H[31]-palmitic
   acid (22.06 mg/L). Proteins were precipitated by three repeated rounds
   of shaking/snap freezing on liquid nitrogen, incubation in ice for 1 h
   and centrifugation. Seven independent experiments were performed.

   The generated raw data were pre-processed using an in-house script
   developed at Swedish Metabolomics Centre, Umeå, Sweden. The detected
   peaks were identified by comparison of mass spectra and retention
   indexes using NIST MS Search v.2.0, using in-house and NIST98 spectral
   databases. After pre-processing and filtering of the metabolites of
   high coefficient of variation in quality control pool samples (>25%),
   levels of 21 metabolites were used for further analysis.

2.8. Mitochondrial oxygen consumption (OCR) measurements

   OCR was evaluated with an XFe24 extracellular flux analyzer. INS1
   832/13 cells were transfected with scramble (negative control, NC) or
   Epdr1 siRNA (10 nM) and cultured for 72 h. Cells were starved in SAB LG
   (2.8 mM) for 2 h and OCR was measured in SAB (without
   bicarbonate/HEPES) every 3 min for 90 min. OCR was measured at basal
   glucose (2.8 mM glucose) and after adding HG (16.7 mM), 5 μM
   oligomycin, 4 μM Carbonyl cyanide p-trifluoromethoxyphenylhydrazone
   (FCCP), and 1 μM rotenone/antimycin A. Wave Seahorse Software and an
   online tool ([85]seahorseanalytics.agilent.com) were used to analyze
   data. Non-mitochondrial respiration was subtracted, and OCR data
   normalized to total protein content, determined by the colorimetric
   Biuret method (Pierce-BCA Protein Assay Kit, Thermofisher).

2.9. Immunoblot analysis

   INS1 832/13 cells (1.7 × 10^5 cells/cm^2) were seeded in 24 well
   plates, transfected with scramble (negative control, NC) or Epdr1 siRNA
   (10 nM) (as described in section [86]2.3) and 72 h later cells were
   washed with cold PBS and cells lysed with RIPA buffer (containing
   complete protease inhibitor cocktail). Protein concentration was
   measured by Pierce BCA Protein Assay Kit and 30 μg of total protein
   electrophoresed on 4–15% mini-Protean TGX precast gel (Bio-Rad),
   transferred onto a PVDF membrane (Bio-Rad) and blocked with BSA 5.0% in
   TBST (Tris-buffered saline+0.1% Tween 20 for 1 h at room temperature.
   Next, the membranes were incubated overnight with Total OXPHOS Rodent
   WB Antibody Cocktail (abcam, #ab110413) diluted 1:250 in TBST-BSA 2%.
   Horseradish-peroxidase-linked goat anti-mouse IgG (1:3000 dilution
   (Biorad, #1706516) was used as a secondary antibody. Blots were
   developed with enhanced chemiluminescence (ECL). Densitometry analysis
   was performed using BioRad ImageLab software and total protein load
   (stain free) was used as a loading control.

2.10. Cytosolic ATP/ADP-ratio

   Single cell cytosolic ATP/ADP-ratio measurements were performed using
   the genetically encoded biosensor PercevalHR [[87]26] and pHrodo Red AM
   intracellular pH indicator (c: p35372, Life Technologies). Cells were
   co-transfected with NC or Epdr1 siRNA (10 nM) and 1 μg PercevalHR
   plasmid (c:49083, Addgene). Cells were starved for 2 h in SAB LG and
   preincubated for 1 h with pHrodo Red AM (diluted 1/1,000). PercevalHR
   fluorescence was recorded by laser excitation/emission at 490/530 nm
   and normalized to the pHrodo Red AM fluorescence emission recorded at
   560/585 nm on a Zeiss LSM510 confocal microscope for 30 min/well (scan
   time of 7.86 s).

2.11. Quantification of whole-cell ATP and ADP

   Cellular ATP and ADP levels were quantified by ultra-high performance
   liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS). Prior to
   analysis, the cell extracts were diluted 10 times with MeOH/water (1/1,
   v/v) with final concentration of the labelled internal standards
   (AMP-^13C[10]^15N[5], ADP-^15N[5], ATP-^13C[10]) of 1 μM. Separation of
   the nucleotides was achieved with a 15-minute gradient using a
   iHILIC-(P) Classic column (PEEK, 50 × 2.1 mm, 5 μm, HILICON, Umeå,
   Sweden) with mobile phases composed of (A) 10 mM ammonium acetate in
   water at pH 9.4 and (B) 10 mM ammonium acetate at pH 9.4 in 90%
   acetonitrile, both mobile phases were supplemented with 5 μM medronic
   acid. The flow rate was 0.35 mL/min, and the gradient elution program
   was set as follows: 0.0 min (85% B), 5 min (60% B), 7 min (30% B),
   8 min (30% B), 9 min (85% B), and 15 min (85% B). The UHPLC-MS/MS
   system consisted of an Agilent 1290 UPLC connected to an Agilent 6490
   triple quadrupole tandem mass spectrometer (Agilent, CA, USA). Analytes
   were ionized using electrospray ionization operated in positive
   ionization mode. The source and gas parameters were set as follows: ion
   spray voltage 4.0 kV, gas temperature 150 °C, drying gas flow 11 L/min,
   nebulizer pressure 20 psi, sheath gas temperature 325 °C, sheath gas
   flow rate 12 L/min, fragmentor 380 V. Multiple reaction monitoring
   (MRM) transitions for AMP, ADP, ATP, and their respective labelled
   internal standards were optimized by flow injection analysis.
   Quantification of AMP, ADP, and ATP was conducted based on internal
   standard calibration. Calibration curves were linear from 5 nM to
   50 μM. The accuracy, determined through spiking experiments of the cell
   extracts (n = 4) with 10 μM standards, was within the acceptable range
   of 100 ± 15%.

2.12. Statistical analysis

   Data are shown as mean ± SEM of at least three independent experiments.
   The Student's t test was used to compare two groups and ANOVA for
   multiple groups. Statistical significance was set at p < 0.05. SPSS 28
   and Prism 9.0 (GraphPad) were used for statistical analysis and graph
   generation. Batch effects were removed from metabolomics data using
   ComBat [[88]27] and data expressed as fold change of each metabolite in
   HG and HG + IBMX relative to LG simulated cells. ANOVA was applied for
   identifying differentiating metabolites between two genotypes and p
   values were corrected for multiple testing with the Benjamini–Hochberg
   procedure. Prior to multivariate analysis, data was mean-centered and
   scaled to unit variance to ensure equal weight of all metabolites in
   analysis. Unsupervised principal component analysis (PCA) was first
   used to recognize underlying patterns and to detect outliers.
   Supervised orthogonal partial least-squares discriminant analysis
   (OPLS-DA) was then performed to find discriminating metabolites based
   on the variable importance on projection (VIP) score. The VIP measures
   the influence of metabolites on the predictive component in the model,
   in other words; how much influence the metabolite has on discriminating
   the two different genotypes (Epdr1-KD vs NC control). Metabolites with
   VIP score >1 were considered metabolites responsible for the separation
   between the two genotypes. The OPLS-DA model was validated with 3-fold
   cross-validation and permutation test (n = 200) ([89]Figure S1). All
   the metabolomics analysis was carried out in R 4.1.2. Pathway
   enrichment analysis was performed with MetaboAnalyst based on the Kyoto
   Encyclopedia of Genes and Genomes (KEGG) library using hypergeometric
   test.

3. Results

3.1. EPDR1 mRNA levels in human islets are upregulated in T2D and obese
subjects and correlated to GSIS ex vivo

   We first investigated whether EPDR1 mRNA expression in human pancreatic
   islets was affected by metabolic (obese vs normo-weight), glycaemic
   (HbA1c) and disease status (T2D vs. non-diabetic (ND)) ([90]Figure 1).
   We observed that EPDR1 mRNA levels were significantly elevated in
   islets from T2D vs ND ([91]Figure 1A) and from overweight/obese (BMI
   >27 kg/m^2) vs normo-weight donors (BMI <27 kg/m^2) ([92]Figure 1B).
   Similarly, a suggestive increase of EPDR1 mRNA levels in islets from
   T2D vs ND donors and their positive correlation with donors BMI was
   found when the expression was tested by qPCR in a small subset of
   donors ([93]Figure S2A and B). To gain insight into a potential role of
   EPDR1 in human islet function and glucose homeostasis, we evaluated the
   association between islet EPDR1 mRNA levels and glycemic control
   (HbA1c) and GSIS ex vivo; the data were analysed separated by disease
   status ([94]Figure 1C–G). In pancreatic islets from ND donors, we found
   that EPDR1 mRNA levels were positively correlated with BMI
   ([95]Figure 1C), HbA1c ([96]Figure 1D), and GSIS ([97]Figure 1E). In
   contrast, in pancreatic islets from T2D donors, we observed a negative
   correlation between EPDR1 mRNA levels and HbA1c ([98]Figure 1F) but a
   strong positive correlation with GSIS ([99]Figure 1G). These data
   indicate that EPDR1 mRNA levels are upregulated under metabolic stress
   conditions (hyperglycaemia/obesity) and associated with increased GSIS.
   This could be viewed as a compensatory mechanism to improve β-cell
   function, resulting in insulin hypersecretion to maintain glucose
   homeostasis in the context of insulin resistance of obese subjects.

Figure 1.

   [100]Figure 1
   [101]Open in a new tab

   EPDR1 mRNA levels in human islets are upregulated in T2D and obese
   donors and are associated with ex vivo GSIS. (A) EPDR1 mRNA in ND
   (n = 160) vs T2D (n = 35) islets and (B) in normo-weight (BMI<27;
   n = 67) and overweigh/obese (BMI>27; n = 33) donors, ∗p < 0.05,
   ∗∗p < 0.01 by unpaired student’s t-test. Rank–Spearman correlation
   analysis between EPDR1 mRNA levels and (C) Donor BMI, (D) HbA1c, and
   (E) GSIS in ND donor islets. Rank–Spearman correlation analysis between
   EPDR1 mRNA levels and (F) HbA1c and (G) GSIS in T2D donor islets.

3.2. EPDR1 eQTLs are associated with functional outcomes of human islets

   To explore whether genetic variance could affect EPDR1 gene expression
   in human pancreatic islets, we first looked up genome-wide significant
   expression quantitative trait loci (eQTL) for EPDR1 in the Inspire
   cohort with data from 420 donors [[102]28]. We found loci with
   genome-wide significant associations to EPDR1 expression in human
   islets (rs6965395 and rs1524054) and others that were borderline
   significant (rs4720265, rs3734952, rs10488617 and rs3213975 with
   nominal p-values = 2.5 × 10^−21, 7.14 × 10^−21, 4.5 × 10^−22,
   4.5 × 10^−22, respectively). Next, we assessed whether these SNPs were
   also eQTLs in sorted human β-cells in the Inspire cohort. We found that
   the SNP rs6965395 was significantly associated with EPDR1 expression in
   sorted β-cells (nominal p-value = 1.69 × 10^−17).

   Next, we evaluated whether the EPDR1 eQTLs found in the Inspire cohort
   could be replicated in our human pancreatic islet cohort (188
   donors-EXODIAB [[103]21]) and whether it could translate into β-cell
   functional outcomes. Linear models were built to assess the association
   between EPDR1 genotypes in these SNPs (rs4720265, rs3734952, rs10488617
   and rs3213975) and gene expression ([104]Figure 2A–D). Age, BMI, purity
   and days of culture were set as covariates. We confirmed that carriers
   of the alternative allele for all these SNPs had reduced EPDR1 gene
   expression in pancreatic islets ([105]Figure 2A–D). Interestingly,
   genotypes of these eQTLs had suggestive association with parameters of
   β-cell function, with carriers of the minor allele and lower EPDR1 mRNA
   levels, showing reduced GSIS stimulatory index ([106]Figure 2E–H) and
   GLP1-promoted GSIS ([107]Figure 2I–L). These data indicate that EPDR1
   eQTLs in human islets could potentially affect human β-cell function,
   via changes in EPDR1 expression.

Figure 2.

   [108]Figure 2
   [109]Open in a new tab

   EPDR1 eQTLs are associated with functional outcomes of human islets.
   The alleles (GG-GA-AA) of SNPs; rs4720265, rs3734952, rs10488617 and
   rs3213975 are associated with: (A–D) EPDR1 mRNA levels, (E–H)
   GSIS-Stimulatory Index and, (I–L) GLP1-promoted GSIS in human
   pancreatic islets. Linear regression analysis for the SNPs alleles
   (eQTLs) vs. EPDR1 mRNA levels was performed adjusting by the next
   covariates; age, BMI, T2D, purity (%) and days in culture of isolated
   islets. For the relation between the EPDR1 SNPs alleles and the
   functional variables one-way ANOVA was performed. ∗p < 0.05,
   ∗∗∗p < 0.001.

3.3. EPDR1 positively regulates GSIS in human β-cells

   To evaluate the potential regulation of human β-cell function by EPDR1,
   the effect of silencing EPDR1 expression and of treatment with the
   recombinant human EPDR1 protein [[110]19] on GSIS was tested in
   pancreatic islets obtained from ND donors and in the human EndoC-βH1
   β-cell line ([111]Figures 3 and [112]S2). Reduced expression of EPDR1
   in human islets transfected with EPDR1 vs scramble (control) siRNA was
   confirmed by qPCR analysis ([113]Figure 3A). GLP1R is a
   Gαs-protein-coupled receptor and acts by increasing cAMP accumulation
   and activation of signaling pathways mediated by PKA and Epac
   [[114]29]. Thus, to mimic GLP1R activation, IBMX was used to increase
   the accumulation of cAMP.

Figure 3.

   [115]Figure 3
   [116]Open in a new tab

   EPDR1 increases GSIS in human islets. Pancreatic islets from ND donors
   were transfected with a scramble (negative control-NC) siRNA or EPDR1
   siRNA (50 nM - EPDR1-KD) in two successive days, 72 h after the first
   transfection; (A) EPDR1 mRNA levels in EPDR1-KD and NC control human
   islets were analysed by qPCR (n = 4), (B) Insulin secretion in response
   to stimulation with SAB low (2.8 mM), high glucose (16.7) or high
   glucose + IBMX (100 μM) for 1 h in EPDR1-KD vs NC control human islets,
   data are expressed as % of insulin secreted from total insulin content
   (n = 4). (C) Pancreatic islets from ND donors were treated for 48 h
   with recombinant human EPDR1 protein in various doses (1.25–10 μg/ml)
   and insulin secretion in response to stimulation with SAB low (2.8 mM),
   high glucose (16.7) or high glucose + IBMX (100 μM) for 1 h measured
   (n = 3). (D) EndoC-βH1 cells were treated for 48 h with recombinant
   human EPDR1 protein in various doses (1.25–10 μg/mL) and the insulin
   secretion in response to stimulation with SAB low (2.8 mM) and high
   glucose (16.7) for 1 h measured (n = 3). For Figures C and D, the
   statistical analysis compared the levels of insulin secretion in
   response to glucose in cells treated with each dose of EPDR1 vs
   vehicle. Graph bars represent mean ± SEM. ∗p < 0.05, ∗∗p < 0.01 by
   paired student’s t-test.

   As expected, a clear increase in insulin secretion in response to
   glucose and glucose + IBMX was observed in the control human islets
   ([117]Figure 3B). However, insulin secretion in response to glucose was
   significantly reduced upon EPDR1 silencing in human islets
   ([118]Figure 3B). The effect of EPDR1 silencing on GSIS was also tested
   in EndoC-βH1 β-cells; no significant changes were observed
   ([119]Figure S3), which may be explained by the low endogenous EPDR1
   mRNA expression level in EndoC-βH1 β-cells ([120]Figure S2A), a
   potential consequence of their fetal genomic signature [[121]30].

   Next, human islets from ND donors were treated with vehicle or
   increasing concentrations of human EPDR1 protein for 48 h and insulin
   secreted in response to stimulation with glucose or glucose + IBMX was
   measured. No significant changes were observed in insulin secretion at
   basal, high glucose and glucose + IBMX conditions in response to EPDR1
   vs vehicle treatment ([122]Figure 3C). To overcome the potential
   obstacle of EPDR1 protein to reach β-cells within the pancreatic islet
   core or any confounding effects of islet non-β-cells (secretion of
   other islet hormones), human EndoC-βH1 β-cells were also treated with
   same human EPDR1 protein concentrations and insulin secretion in
   response to glucose stimulation measured. Treatment of EndoC-βH1
   β-cells with human EPDR1 protein resulted in significantly increased
   GSIS ([123]Figure 3D). Overall, the data described here indicate that
   human EPDR1 protein is positively regulating human β-cell function by
   increasing glucose-responsiveness and insulin secretion.

3.4. EPDR1 is required for glucose- and pyruvate- but not K^+-stimulated
insulin secretion in β-cells

   To evaluate whether the role of EPDR1 in β-cells is conserved among
   different species and to deepen our knowledge about the molecular
   mechanisms of EPDR1 in β-cell function, we examined the EPDR1 actions
   on INS1 832/13 cells, a well characterized glucose-responsive clonal
   rat β-cell line [[124]22]. Epdr1 was silenced by transfection of Epdr1
   siRNA and reduced Epdr1 expression confirmed by qPCR ([125]Figure 4A).
   Then, the effects of silencing Epdr1 on insulin secretion in response
   to different metabolic and pharmacologic stimuli were assessed
   ([126]Figure 4B–E).

Figure 4.

   [127]Figure 4
   [128]Open in a new tab

   EPDR1 is required for glucose- and pyruvate- but not KCl-stimulated
   insulin secretion in β-cells. A-D) INS1 832/13 cells were transfected
   with NC scramble or Epdr1 siRNA (10 nM) for 72 h. (A) mRNA levels in
   Epdr1-KD vs NC control cells. (B) Insulin secretion in response to low
   (2.8 mM), high glucose (16.7), Pyruvate (10 mM) or KCl (30 mM) in
   Epdr1-KD vs NC control cells (n = 4). (C) Insulin secretion in response
   to low (2.8 mM), high glucose (16.7), high glucose + IBMX (100 μM) or
   high glucose + FSK (10 μM) in Epdr1-KD vs NC control cells (n = 6). (D)
   Insulin secretion in response to low (2.8 mM) or high glucose (16.7) in
   Epdr1-KD vs NC INS1 832/13 cells treated with high glucose (20 mM),
   high palmitic acid (0.5 mM) or Thapsigargin (1 μM) for 24 h (n = 4).
   (E) Insulin secretion in response to low (2.8 mM) or high glucose
   (16.7) INS1 832/13 cells treated with human EPDR1 protein
   (0.63–10 μg/ml) (n = 4). Graph bars represent mean ± SEM. ∗p < 0.05,
   ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 by paired student’s t-test and
   two-way ANOVA accordingly.

   Silencing of Epdr1 expression resulted in significantly reduced insulin
   secretion in response to stimulation with glucose and pyruvate but not
   with the plasma membrane depolarizing agent KCl ([129]Figure 4B). These
   data suggest that EPDR1-increased GSIS is mediated by changes on
   mitochondrial metabolism of glycolysis-derived pyruvate but may leave
   other processes such as insulin production, β-cell mass or the
   exocytosis machinery unaffected.

   INS1 832/13 cells are poorly responsive to GLP1 or GLP1R agonist such
   as Exendin-4 [[130]22]. Thus, to mimic GLP1R activation, IBMX and
   forskolin (FSK) were used to trigger cAMP/PKA/Epac signalling. As
   expected, IBMX and FSK amplified GSIS in control cells and this effect
   was significantly reduced in the Epdr1-KD cells ([131]Figure 4C).

   The harmful effects that chronic high glucose and free fatty acids
   exert on β-cell metabolism and function are well described
   [[132]3,[133]31]. To further explore the metabolic mechanisms of
   EPDR1-regulated GSIS, Epdr1-KD and control INS1 832/13 cells were
   exposed to high glucose (20 mM) or palmitate (0.5 mM) for 24 h
   (mimicking mild glucotoxic or lipotoxic conditions). The deleterious
   effects of high glucose and high palmitate in vitro in control cells
   were confirmed as they reduced the average stimulatory index (fold
   change of insulin secretion in response to high vs. low glucose
   stimulation) from 10.3 to 3.5 and 5.0, respectively ([134]Figure 4D).
   Interestingly, Epdr1 silencing resulted in reduced GSIS in the control
   (normo-glucose) and high palmitate conditions but not at high glucose
   ([135]Figure 4D). Considering that the glucotoxic effects on β-cells
   are driven by chronically elevated mitochondrial glucose metabolism
   [[136]3], our results support the idea that EPDR1 is positively
   regulating GSIS by promoting mitochondrial glucose metabolism in
   β-cells.

   Finally, to confirm the positive effects of EPDR1 on β-cell function
   observed in human islets and EndoC-βH1 β-cells ([137]Figure 3C and D),
   INS1 832/13 cells were long-term treated with human EPDR1 protein and
   GSIS measured. GSIS was significantly increased in response to
   treatment with the EPDR1 protein ([138]Figure 4E), indicating that
   EPDR1 protein positively regulates β-cell function.

3.5. EPDR1 is not required for glucose promoted cytosolic calcium oscillation
in β-cell

   To explore potential molecular mechanisms underlying the positive
   effects of EPDR1 on GSIS, we first performed experiments using the
   calcium sensor dye (Fluo4-AM) and confocal microscopy to evaluate
   single cell cytosolic calcium response to glucose or pyruvate
   stimulation in control (NC) and Epdr1-KD INS1 832/13 cells
   ([139]Figure S4). In control cells, cytosolic calcium increased in
   response to glucose ([140]Figure S4A–D) and pyruvate
   ([141]Figure S4E–H) stimulation and acutely spiked in response to KCl
   (plasma membrane depolarization). This confirms that INS1 832/13 cells
   were responsive to metabolic and electrochemical signals that converge
   into plasma membrane depolarization, opening of VDCC and calcium
   uptake. However, there was no differences in the cytosolic calcium
   responses to glucose or pyruvate in NC vs Epdr1-KD cells when the
   average AUC ([142]Figure S4B and F) or the average oscillation
   frequency (Fig. S4D and H) was compared. These data indicate that the
   EPDR1 mode of action is not mediated by changes in calcium uptake or
   buffering cytosolic calcium (into ER or mitochondria) in response to
   glucose or pyruvate. Instead, the mode of action is likely downstream
   of glucose-induced cytosolic calcium influx.

3.6. Metabolomics analyses suggest that EPDR1 is required for coupling
glycolytic to mitochondrial metabolism

   To further investigate the molecular mechanism by which EPDR1
   positively regulates GSIS, mass-spectrometry-based metabolomics
   analysis in response to glucose stimulation in NC and Epdr1-KD INS1
   832/13 cells was performed ([143]Figure 5). PCA and OPLS-DA were used
   to find the set of metabolite changes in response to glucose (fold
   change HG/LG after 60 min of stimulation) that best explained the
   variance between NC and Epdr1-KD cells ([144]Figure 5A and B). Based on
   the VIP for metabolites, namely, a score of the power each metabolite
   has to influence the discrimination of genotypes (Epdr1-KD vs NC
   control), metabolic pathway analysis (over-representation analysis
   tests) was performed ([145]Figure 5C). Remarkably, metabolic pathways
   critical for GSIS such as the Warburg effect, the malate-aspartate
   shuttle and the TCA cycle were significantly over-represented
   ([146]Figure 5C) and may explain the molecular mechanism by which EPDR1
   regulate GSIS in β-cells.

Figure 5.

   [147]Figure 5
   [148]Open in a new tab

   Metabolomics analysis in cells stimulated by glucose indicates that
   EPDR1 tightens the glycolysis-mitochondrial metabolism coupling.
   Metabolomics analysis after 1 h of glucose (HG) stimulation in NC and
   Epdr1-KD INS1 832/13 cells (n = 7). (A) PCA for metabolites levels at
   HG/LG, (B) OPLS-DA loading scatter plot; it describes the relationship
   between a metabolite and the specific component, the absolute value of
   the loading indicates how much the metabolite influences the specific
   component whereas the sign of the loading indicates whether the
   influence is positive or negative on this component. Horizontal axis
   shows X-loadings p and the Y-loadings q of the predictive component
   (pq1). The vertical axis showing the X-loading p(o) and the Y-loadings
   s(o) for the Y-orthogonal component (poso1). Inserted in the Figure is
   the Supervised orthogonal partial least-squares discriminant analysis
   (OPLS-DA) score plot, showing the predictive (t1) and 1st orthogonal
   component (to1). (C) Metabolic Pathway Analysis using the predictive
   component (VIP) of the OPLS-DA model, showing the metabolic pathways
   that explain the variance between Epdr1-KD vs. NC control cells. (D)
   Mass spectrometry measured cellular metabolites abundance in response
   to glucose (fold change HG/LG) of pyruvate and lactate. (E) Ratio of
   cellular abundance of Lactate/Pyruvate. (F) Secreted extracellular
   lactate concentration during GSIS (1 h) in NC and Epdr1-KD INS1 832/13
   cells (n = 4). (G) TCA cycle scheme. (H) Mass spectrometry measured
   cellular metabolites abundance in response to glucose (fold change
   HG/LG) showing differences of TCA cycle and malate/aspartate shuttle
   metabolites such as; reduced citrate and alpha-ketoglutarate (AKGA) and
   increased aspartate in Epdr1-KD vs NC control cells. Data are
   mean ± SEM of n = 7 independent experiments. ∗p < 0.05; ∗∗p < 0.01;
   ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001 with Combat batch correction method
   followed by ANOVA on genotype.

   Relative changes in cell metabolite levels in response to glucose for
   each of these metabolic pathways were further analysed
   ([149]Figure 5D,E, G, and H). The data indicated that Epdr1 silencing
   results in decreased levels of pyruvate, but increased levels of
   lactate ([150]Figure 5D) and the lactate/pyruvate-ratio
   ([151]Figure 5E); this suggests diversion of glycolysis-derived
   pyruvate from mitochondrial metabolism to lactate production (Warburg
   effect). Consequently, we quantified (via an enzymatic method) the
   secreted lactate accumulated (in SAB conditioned medium) during 1 h of
   glucose stimulation in the control (NC) and Epdr1-KD cells and found a
   38% higher concentration of lactate in the Epdr1-KD compared to the NC
   control cells ([152]Figure 5F); this may reflect a deviation of
   pyruvate from the TCA cycle to lactate production in the
   Epdr1-deficient β-cells. In addition, in the Epdr1-KD cells, the
   glucose responsiveness of several TCA cycle metabolites, such as
   citrate, isocitrate, α-ketoglutarate (AKGA), succinate and malate were
   reduced while the glucose-provoked reduction in aspartate levels was
   also decreased ([153]Figure 5G andH). Likewise, the citrate/pyruvate-
   and AKGA/pyruvate-ratios were strongly reduced in the Epdr1-KD vs NC
   control cells, indicating reduced entrance of pyruvate into the
   mitochondrial TCA cycle or, alternatively, reduced TCA cycle
   anaplerosis/increased cataplerosis ([154]Figure 5G and H). In line with
   the metabolic pathway analysis, increased aspartate (less reduced) and
   reduced malate levels in response to glucose in the Epdr1-KD vs NC
   control cells ([155]Figure 5G and H) suggested altered activity of the
   malate-aspartate-shuttle. This may lead to reduced anaplerosis and
   reduced replenishment of cytosolic NAD^+, which may also contribute to
   uncoupling glycolysis and mitochondrial metabolism. In line with this,
   a compensatory increase in lactate production is observed, which may
   serve to maintain glycolytic flux (upon replenishing the cytosolic
   NAD^+). Overall, the metabolic changes observed in Epdr1-KD vs NC
   control cells were compatible with a model where the glycolysis-derived
   pyruvate is diverted from mitochondria to production of lactate.

   An essential role of the malate-aspartate shuttle for the
   incretin-induced amplification of GSIS has been reported [[156]32].
   Thus, similar metabolomics analysis were also applied in response to
   glucose + IBMX stimulation in Epdr1-KD vs NC INS1 832/13 cells
   ([157]Figure S5). In line with the notion that Epdr1 silencing results
   in defective mitochondrial metabolism, the Warburg effect, the
   malate-aspartate shuttle and the TCA cycle also came up as significant
   metabolic pathways explaining the differences between Epdr1-KD vs NC
   control INS1 832/13 cells in response to glucose + IBMX stimulation
   ([158]Figure S5D). Univariate analyses confirmed an increase of the
   lactate and lactate/pyruvate ratio (Warburg effect) and reduced TCA
   cycle metabolite levels in the Epdr1-KD vs NC control cells in response
   to glucose + IBMX stimulation ([159]Figure S5E–G). Therefore, our data
   suggested that EPDR1 is increasing GSIS by directing the metabolism of
   glycolysis-derived pyruvate through the mitochondrial TCA cycle.

3.7. EPDR1 is required for the glucose-stimulated increases of the cytosolic
ATP/ADP-ratio

   To deepen our understanding of the mechanisms of EPDR1 in regulating
   glucose-stimulated mitochondrial metabolism and insulin secretion, we
   measured mitochondrial oxygen consumption rate (OCR) in response to
   glucose in the Epdr1-KD and NC control INS1 832/13 cells
   ([160]Figure 6). We observed a slight but significant decrease in the
   acute OCR response to glucose ([161]Figure 6B) in response to glucose
   in the Epdr1-KD and NC control INS1 832/13 cells. No significant
   changes in mitochondrial ATP-linked OCR (p = 0.090, [162]Figure 6C),
   the mitochondrial proton-leak or maximal OCR were observed
   ([163]Figure 6D and E). We also tested the expression level of
   representative protein of mitochondrial Electron Transport Chain (ETC)
   Complexes (I–V). Interestingly, significantly increased levels of SDHB
   protein of complex II was found in Epdr1-KD vs NC control INS1 (832/13)
   cells ([164]Figure S6).

Figure 6.

   [165]Figure 6
   [166]Open in a new tab

   EPDR1 is required for glucose-stimulated increases of ATP/ADP-ratio.
   (A–E) Mitochondrial Oxygen Consumption Rate (OCR) was measured by
   Seahorse (XFe24) in INS1 832/13 transfected with scramble or Epdr1
   siRNA (10 nM). (A) Average OCR traces in basal glucose (2.8 mM),
   glucose (16.7 mM), Oligomycin (4 μM), FCCP (4 μM) and Antimycin
   A + Rotenone (1 μM) conditions (traces represent average OCR). (B–E)
   Mitochondrial respiratory parameters; (B) Acute Response OCR, (C)
   ATP-linked OCR, (D) Proton-Leak OCR and (E) Maximal OCR. Data are
   mean ± SEM of n = 6 independent experiments. (F–G) Single-cell
   cytosolic ATP/ADP-ratio (PercevalHR) in response to glucose in Epdr1-KD
   vs NC INS1 832/13 cells. (F) Average traces and (G) Average AUC of
   traces Data are mean ± SEM of n = 2 independent experiments. (H–J)
   Whole-cell adenine nucleotides measured by LC-MS in Epdr1-KD vs NC INS1
   832/13 cells during GSIS (60 min). (H) Whole-cell ADP concentration,
   (I) Whole-cell ATP concentration and (J) Whole-cell ATP/ADP-ratio. Data
   are mean ± SEM of n = 4 independent experiments, ∗p < 0.05 with one-way
   ANOVA followed by Kruskal Wallis post hoc test.

   To further investigate the bioenergetic alteration caused by
   Epdr1-deficiency in INS1 832/13 cells, single cell cytosolic
   ATP/ADP-ratio was measured using the genetically encoded PercevalHR
   biosensor [[167]26] ([168]Figure 6F and G). As expected for
   glucose-responsive β-cells, the cytosolic ATP/ADP-ratio in control
   cells was increased in response to glucose but this was blunted in the
   Epdr1-KD cells ([169]Figure 6F and G).

   To confirm this finding, whole-cell adenine nucleotide content in
   response to glucose stimulation (1 h) was measured by mass spectrometry
   and the whole-cell ATP/ADP-ratio calculated in the Epdr1-KD vs NC
   control cells ([170]Figure 6H–J). As expected, levels of ADP decreased
   in control cells ([171]Figure 6H) while those of ATP slightly increased
   ([172]Figure 6I). Consequently, the ATP/ADP-ratio increased about 15
   times (fold change in HG vs LG. [173]Figure 6J) after 1 h of glucose
   stimulation. Remarkably, in the Epdr1-KD cells, the glucose-elicited
   rise in the ATP/ADP-ratio was blunted ([174]Figure 6J). Thus, these
   data support the idea that EPDR1 is required for coupling of glycolysis
   and TCA cycle metabolism in β-cells.

4. Discussion

   Previous reports have described a relationship between EPDR1 expression
   and tumor progression/prognosis in different types of cancers
   [[175]7,[176]33,[177]34]. Although it is known that metabolic
   reprograming occurs in tumor cells, no obvious clues about the
   biological role of EPDR1 have been inferred.

   Progress regarding this was recently reached by two studies: one
   revealing the crystal structure of human EPDR1 [[178]35], and the other
   identifying EDPR1 as a secreted human batokine [[179]19]. Thus, EPDR1
   influences the expression of mitochondrial proteins and the
   NE-stimulated mitochondrial proton-leak respiration associated with
   thermogenesis in hBAT [[180]19]. Moreover, it was recently shown that
   EPDR1 regulates osteoblast metabolism and maturation [[181]36]. Thus,
   EPDR1 may also influence the cross-talk between bone and glucose
   homeostasis [[182]37]. Altogether, these studies suggest that EPDR1
   potentially acts as a circulating signal to communicate between
   endocrinologically active tissues such as pancreatic islets, adipose
   tissue and bone to regulate their cell metabolism and maintain
   whole-body energy and glucose homeostasis.

   In this study, we demonstrated that EPDR1 is expressed in human
   pancreatic islets and β-cell models and positively regulates
   glucose-induced mitochondrial metabolism and GSIS. A previous study
   using RNAseq analysis to compare gene expression in highly purified
   cells from fetal vs. adult human pancreatic islets showed that EPDR1
   mRNA levels decrease in adult α-cells but increase in adult β-cells
   [[183]20]. Moreover, single cell RNAseq analysis of human islets shows
   a variable degree of EPDR1 mRNA expression in β-cells but almost absent
   mRNA expression in other islet cell types [[184]38]. Moreover, a recent
   study using single-cell transcriptomic analysis of mouse β and α cells
   sorted from fetal to adult stages found that Epdr1 was not among the
   genes differentially expressed between β- and α-lineages during the
   maturation process of mouse pancreatic islet cells [[185]39]. However,
   they found that Epdr1 was among the genes variably expressed in adult
   (P60) β-cells. Pancreatic islets from P60 mice represent the more
   mature and the less proliferative stage of β-cells; thus, a significant
   variation of Epdr1 expression in β-cells may support a role of EPDR1 in
   β-cell function. These data support the idea that EPDR1 may be
   selectively expreessed in mature β-cell to positively regulate their
   function.

   In this study we show that EPDR1 expression was upregulated in
   pancreatic islets from T2D and obese donors and that its expression was
   positively correlated with BMI of donors. A linear positive correlation
   between BMI of ND subjects with both the basal (fasting) and stimulated
   insulin secretion during oral glucose tolerance test (OGTT) has
   previously been established [[186]40,[187]41]. This relationship has
   been further demonstrated by ex vivo experiments, assessing dynamics of
   insulin secretion from pancreatic islets isolated from ND donors,
   confirming a positive correlation between donor BMI [[188][19],
   [189][20], [190][21], [191][22], [192][23], [193][24], [194][25],
   [195][26], [196][27], [197][28], [198][29], [199][30], [200][31],
   [201][32], [202][33]] and islet insulin content and GSIS [[203]42].
   Thus, our observation of an upregulation of EPDR1 expression in human
   islets from obese people leads us to propose that EPDR1 upregulation in
   human islets from obese people is part of an adaptive program
   attempting to re-establish glucose homeostasis by increasing β-cell
   glucose-responsiveness to cope with an increased metabolic demand
   imposed by insulin resistance. In support of this hypothesis, we also
   found a strong positive correlation between EPDR1 mRNA levels and
   insulin secretion whereas a negative correlation with Hb1Ac was
   evident. Accordingly, treatment of human pancreatic islets and β-cell
   models with human EPDR1 protein resulted in increased GSIS whereas
   silencing of EPDR1 expression in human islets and INS1 (832/13) cells
   decreased GSIS.

   Interestingly, EPDR1 is also upregulated in BAT of obese subjects
   [[204]19] and it has been shown that obese vs lean healthy subjects
   have elevated fat accumulation in BAT-depots [[205]43] and pancreases
   [[206]44]. Thus, a potential common signal driving the rise of EPDR1
   expression in pancreatic islet and hBAT could be increased ectopic
   cellular fat accumulation in subjects with elevated BMI. Notably, Epdr1
   has been recently identified as a target gene of a non-canonical
   insulin signaling in endothelial cells [[207]45]. Thus, another
   potential explanation for the observed positive correlation between
   EPDR1 expression and donors BMI maybe the obesity-associated
   hyperinsulinemia.

   We observed that long-term treatment of both human and rodent β-cell
   lines with the human EPDR1 protein resulted in an inverted “U-Shaped”
   dose-effect on GSIS, namely, EPDR1 positively affect GSIS reaching a
   peak at intermediate tested doses (2.5–5.0 μg/ml) whereas losing its
   effect at the highest tested doses (10 μg/mL). Even though a plasma
   membrane receptor for EPDR1 has not been identified so far, this
   inverted “U-Shaped” dose-effect of EPDR1 on GSIS may suggest that EPDR1
   signaling through a plasma membrane receptor in β-cells is desensitized
   by reducing its expression or translocation to the plasma membrane
   ligand concentrations are saturated. An alternative explanation could
   be that long-term activation of EPDR1-dependent signaling in β-cells
   may result in increased production of oxidative stress at the
   mitochondrial ETC level; this may become detrimental to β-cell function
   as we have observed previously for a serotonin receptor, which acutely
   increases GSIS but conversely, after long-term stimulation decreases
   GSIS [[208]46].

   We identified eQTLs for EPDR1 in suggestive association with two
   ex vivo parameters of β-cell function: GSIS stimulatory index and
   GLP1-promoted GSIS. This is a relevant finding since linking genetic
   variation with EPDR1 expression in human islets and β-cell function may
   be useful to translate these in vitro results to human epidemiological
   studies. For example, it may reveal whether EPDR1 expression in human
   β-cells, based on subject’s eQTLs genotype, is associated with in vivo
   β-cell function and glucose homeostatic traits (derived from OGTT) and
   T2D incidence/prognosis.

   By use of a metabolomic approach, we found evidence for EPDR1
   contributing to a tightening of the coupling of glycolysis to
   mitochondrial TCA-cycle metabolism, a mechanism underlying its positive
   regulation of GSIS ([209]Figure 7). We have demonstrated that Epdr1
   silencing in β-cells resulted in a Warburg effect (i.e. diverting
   glycolysis-derived pyruvate from mitochondria to lactate production)
   and reduced activity of the malate-aspartate shuttle and reduction in
   the TCA cycle metabolite levels. Re-oxidation of the cytosolic NADH is
   required to maintain glycolytic and TCA cycle flux [[210]47,[211]48].
   In most cells NADH is recovered by the lactate dehydrogenase A
   (LDHA)-mediated reduction of pyruvate to lactate. However, LDHA is a
   disallowed/forbidden gene in β-cells [[212]49,[213]50]. In β-cells, the
   cytosolic NADH re-oxidation relies on two metabolic redox systems: the
   malate-aspartate shuttle and the glycerol-phosphate (GP) shuttle
   [[214]48]. The findings in our metabolomics analysis of Epdr1-KD cells
   are compatible with a model where defects in the transfer of reducing
   potential from glycolysis to TCA cycle by the malate-aspartate shuttle
   forces the activation of LDHA, resulting in a decreased supply of
   substrates to TCA-cycle. Increased metabolic flux of glucose-derived
   pyruvate into the TCA cycle increases GSIS [[215]51]. Conversely,
   impaired glucose-responsiveness and GSIS are often accompanied with
   disturbed coupling of glycolytic and TCA cycle mitochondrial metabolism
   and increased lactate production [[216]52]. Consistently, we have shown
   increased glucose-stimulated lactate cell content and secretion from
   Epdr1-KD cells, supporting an activation of the anaerobic glycolysis
   pathway.

Figure 7.

   [217]Figure 7
   [218]Open in a new tab

   Model of EPDR1 regulation of β-cell metabolism and function. EPDR1 can
   regulate β-cell metabolism by binding and signaling through a not yet
   identified plasma membrane receptor or be taken up to act
   intracellularly. EPDR1 positively regulates GSIS of β-cells by
   increasing the coupling of glucose metabolism to mitochondrial TCA
   cycle. EPDR1 is required to channel the glycolysis-derived pyruvate to
   the TCA cycle, avoiding production of lactate; this results in
   increased levels of cytosolic ATP/ADP-ratio, which blocks the K[ATP]
   channels, raises calcium influx and triggers the insulin granule
   exocytosis. Additionally, EPDR1 may positively regulate GSIS by
   increasing the mitochondrial production of TCA cycle-derived citrate,
   which would be exported to the cytosol and act as precursors of
   metabolic coupling factors such as NADPH, Malonyl-CoA, long chain
   acyl-CoAs or Glutamate; these are all metabolites that positively
   regulate insulin granule exocytosis. PDH: Pyruvate dehydrogenase, PC:
   Pyruvate Carboxylase, AKGA: Alpha-ketoglutaric acid.

   Ample evidence support that disturbance of mitochondrial metabolism is
   a cause of β-cell dysfunction in different models of diabetes [[219]3].
   In this study we showed that Epdr1 silencing reduced glucose-stimulated
   mitochondrial respiration and decreased the cytosolic and the
   whole-cell ATP/ADP-ratio. In addition, we observed that Epdr1 silencing
   increased the content of mitochondrial ETC proteins, which may reflect
   compensatory biogenesis of mitochondria replacing defective
   mitochondria [[220]53]. These data indicated defective
   glucose-stimulated mitochondrial activity upon Epdr1 silencing in
   β-cells ([221]Figure 7).

   Interestingly, a recent study has demonstrated that pyruvate kinase,
   rather than mitochondrial oxidative phosphorylation, is the main driver
   of the cytosolic ATP/ADP-ratio to close β-cell K[ATP] channels and
   initiate insulin secretion [[222]54]. Thus, an alternative explanation
   to such strong reduced cytosolic ATP/ADP-ratio we observed in the Epdr1
   deficient cells, could be a reduced anaplerotic capacity of
   mitochondria to produce phosphoenolpyruvate required for the pyruvate
   kinase-mediated synthesis of ATP. Together, our data indicated that
   EPDR1 is positively regulating GSIS by a metabolic mechanism tightening
   the coupling of glycolysis to mitochondrial TCA cycle; this contributes
   to an adequate glucose-stimulated rise in the cytosolic ATP/ADP-ratio
   ([223]Figure 7).

5. Conclusion

   EPDR1 expression in β-cells may increase under metabolic overload
   (obesity), channelling glycolysis-derived pyruvate to the mitochondrial
   TCA cycle and, consequently enhance stimulus-secretion coupling and
   increasing GSIS. Upregulation of EPDR1 expression in pancreatic islets
   from obese people may reduce their risk to progress from glucose
   intolerance to T2D.

Funding

   This work was supported by Novo Nordisk Foundation (NNF18CC0034900),
   Knut and Alice Wallenberg Foundation (KAW2018.0094), Swedish Research
   Council (2021-01777 and 2019-01406), The Royal Physiographic Society of
   Lund (41521-2020 and 42009-2021) and Bo & Kerstin Hjelt Diabetes
   Foundation. The Novo Nordisk Leif Groop Young-Scientist Scholarship.
   The laboratory is part of the LUDC-IRC and was supported by the Swedish
   Foundation for Strategic Research Dnr IRC15-0067.

Duality of interest

   No potential conflicts of interest relevant to this article were
   reported.

Authorship contributions

   L.R.C. conceptualized and designed this study. L.R.C., L.A., O.H.,
   E.C., S.H., S.G. performed experiments and analyzed data. R.P. and Q.G.
   performed statistical data análisis and assisted with experimental
   design. P.S., C.S., M.F., H.M. and T.M., L.E. contributed to the
   interpretation and discussion of data. L.R.C., TM and H.M. wrote the
   manuscript. All authors revised the manuscript and approved the final
   version to be published. L.R.C. is the guarantor of this work and, as
   such, had full access to all the data in the study and takes
   responsibility for the integrity of the data and the accuracy of the
   data analysis.

Prior Presentation

   Preliminary results of this study were presented at the 57th Annual
   Meeting of the EASD (poster #378), Sept 27-Oct 1-2021.

Acknowledgements