We designed a glucagon fusion construct [Fc-glucagon –pcDNA3.1(+)] as follows: the amino acid sequence of glucagon derived from human proglucagon (GenScript, USA; http://www.genscript.com) was fused to the 3′ end of cDNA encoding the CH2/CH3 domain of mouse IgG-2b (Fc), preceded by a 28 amino acid signal peptide as described previously (19). As a negative control for all transfections, proteomics, immunofluorescence microscopy, and co-immunoprecipitation experiments, we also designed a Fc-pcDNA3.1(+) construct. DNA sequences were confirmed at the London Regional Genomics Facility, Western University.

Wild type α-TC1-6 cells (a kind gift from C. Bruce Verchere, Vancouver, BC) were cultured in DMEM containing 25 mM glucose, L-glutamine, 15% horse serum, and 2.5% fetal bovine serum, as described previously (17, 23). Based on the ATCC product sheet, the base cell culture medium for α-TC1-6 cells is low glucose (5.5 mM) Dulbecco's Modified Eagle's Medium (DMEM); however, for glucagon secretion (glucagon hypersecretion) studies, high glucose DMEM (16.7 or 25 mM) has been traditionally used to prepare α-TC-6 cells for downstream experiments (17, 24). Cells were grown to 90% confluency and transfected with Fc alone or Fc-glucagon using Lipofectamine 2000. To determine changes in granule size, mass, and proteome, cells were incubated with or without GABA (25 μM), insulin (100 pM), or GABA (25 μM) plus insulin (100 pM) in either 25 or 5.5 mM glucose prior to the granule enrichment procedure. To account for all potential modulators of glucagon secretion, including the possibility of autocrine regulation of glucagon secretion (25, 26) we chose long-term cumulative incubation that has previously been used by our team (17) and other investigators (23) for secretion studies in α-TC1-6 cells.

At the end of the incubation period, granules were extracted as previously published (27) with some modifications. Briefly, cells were detached using 5 mM EDTA in PBS (pH 7.4) containing cOmplete Mini Protease Inhibitor Cocktail (Supplementary Table 1) on ice, centrifuged and resuspended in ice-cold homogenization buffer (20 mM Tris-HCl pH7.4, 0.5 mM EDTA, 0.5 mM EGTA, 250 mM sucrose, 1 mM DTT, cOmplete Mini Protease Inhibitor Cocktail, and 5 μg/mL Aprotinin). The cells were passed 10 times through a 21G needle and again 10 times through a 25G needle. The resulting lysates were centrifuged to obtain a post-nuclear supernatant (PNS). The nuclear fraction was washed seven times in ice-cold homogenization buffer and stored at −80°C. The post-nuclear supernatant (PNS) was centrifuged at 5,400 × g for 15 min at 4°C to obtain a post-mitochondrial supernatant, which then was spun at 25,000 × g for 20 min, and the resultant pellet was washed five times at 4°C. Enrichment was confirmed through immunoblotting for organelle-specific markers as described below.

The enriched preparations of secretory granules from α-TC1-6 cells were lysed using non-ionic lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1% Triton X-100 plus cOmplete Mini Protease Inhibitor Cocktail, and 5 μg/mL Aprotinin). Proteins were resolved by 4–12% NuPAGE, transferred to a PVDF membrane and probed with the following antibodies (Supplementary Table 1): vesicle-associated membrane protein 2 (VAMP2) for mature secretory granules; calreticulin for the endoplasmic reticulum; TGN46 for the trans-Golgi network; and Lamin B1 for the nuclear envelope. Immunoreactive bands were visualized using HRP-conjugated goat anti-rabbit secondary antibody and Clarity Western ECL substrate. Images were acquired on a BioRad ChemiDoc Imaging System. Total cell extracts were used as positive controls.

To validate the interaction of glucagon with either GRP78 or histone H4 within secretory granules we first purified Fc-glucagon or Fc (as control) from the secretory granule preparation by incubating the secretory granule lysate with Protein A-Sepharose beads overnight at 4°C with rotation. The Fc or Fc-glucagon complex was eluted from the beads with 0.1 M glycine buffer (pH 2.8). The eluate was concentrated 50 times using a speed vac, run on a 10% Bis-Tris NuPAGE gel (Supplementary Table 1) and proteins were transferred onto a PVDF membrane. After an overnight incubation with primary antibodies against GRP78 or histone H4, bands were visualized with HRP-conjugated goat anti-rabbit secondary antibody and Clarity Western ECL substrate (Supplementary Table 1). Images were acquired on a BioRad ChemiDoc Imaging System.

Enriched secretory granule fractions were prepared, resuspended in 0.2 N HCl, passed 10 times through a 30G needle, and kept at 4°C overnight. The reaction was stopped by addition of 0.2 volumes of 1N NaOH. The supernatant was collected after centrifugation at 6,500 × g at 4°C for 10 min. Protein levels were determined by BCA assay, and 100 ng of protein was used for measuring total histone H4 (Histone H4 Modification Multiplex ELISA-like format Kit, Supplementary Table 1), as per the manufacturer's instructions. The nuclear fraction was also assayed for histone H4 as a positive control.

To validate the presence of GRP78 and histone H4 in glucagon-positive secretory granules, α-TC1-6 cells were cultured on collagen1-coated coverslips (three per experiment), and processed for immunofluorescence microscopy as described previously (18). Briefly, cells were fixed in 4% paraformaldehyde and permeabilized in 0.1% saponin in 0.5% BSA for 1 h. After blocking in 10% goat serum, cells were incubated with primary antibodies (mouse anti-glucagon and rabbit anti-GRP78 or rabbit anti-histone H4) overnight. Coverslips were washed in PBS and incubated with goat anti-mouse Alexa Fluor IgG 488 and goat anti-rabbit Alexa Fluor 594 (Supplementary Table 1) for 3 h in the dark at room temperature, then mounted using ProLong Gold Antifade Mountant. Images were acquired on a Nikon A1R Confocal microscope with a 60x Nikon Plan-Apochromat oil differential interference contrast objective lens using NIS-Elements, software. To show secretory granule co-localization, images were post-processed by 2D deconvolution. To measure the degree of co-localization, regions of interest were manually drawn around distinct single or multicell bodies, positive for Fc-glucagon and either GRP78 or histone H4 and cropped for analysis. Co-localization of the pixels from each pseudo-colored image were used to calculate Pearson's correlation coefficient (PCC), as we described previously (19).

After treatment of α-TC1-6 cells with GABA and/or insulin in media containing 25 mM glucose as described above, the proteomes were tabulated, and Venn diagram analysis revealed 27 metabolic/regulatory/secretory proteins and 36 histone/cytoskeletal/ribosomal proteins that were common between the groups treated with GABA and insulin. We selected 11 of these proteins (based on availability of the pre-designed siRNA) for siRNA-mediated depletion: Peroxiredoxin-2 (PRDX2), Malate dehydrogenase 1 (MDH1), Aconitate hydratase, mitochondrial (ACO2), 14-3-3 protein zeta/delta (KCIP-1), ELKS/Rab6-interacting/CAST family member 1 (ERC1), Alpha-tubulin 2 (AT2), ATP synthase F1 subunit alpha (ATP5F1A), Histone H4, GRP78, FXYD domain-containing ion transport regulator 2 (FXYD2), and Protein disulfide-isomerase (PDI), (Silencer siRNA, Thermo Fisher Scientific Inc. MA, USA).

Gene silencing was based on a published protocol (28). Briefly, α-TC1-6 cells were cultured to 60% confluency and transfected with final concentrations of 50 nM of pooled siRNAs (three siRNAs for each target) or control scrambled siRNA using Lipofectamine2000. Cells were incubated for 48 h, after which media were removed and replaced. After 24 h, expression levels of the targeted proteins were evaluated by immunoblotting using primary antibodies against each protein (Supplementary Table 1). Meanwhile, siRNA mediated knockdown of the proglucagon gene was shown as a positive control using real-time PCR (Quant Studio Design and Analysis Real-Time PCR Detection System) (Supplementary Figure 4).

To measure cellular and secreted glucagon levels after siRNA-mediated gene silencing, cell lysates or media were acidified in HCl-ethanol (92:2 v/v) in a 1:3 ratio, kept at −20°C overnight, then centrifuged at 13,000 × g for 15 min at 4°C. The supernatant was then mixed 1:1 with 20 mM Tris, pH 7.5 26 and glucagon levels were measured by ELISA (Thermo Fisher Scientific, Supplementary Table 1) according to the manufacturer's instructions. To measure Fc-glucagon, samples were diluted to reach an OD at the linear part of the standard curve.

α-TC1-6 cells cultured and kept under chronic exposure to 25 mM glucose and at confluency rate of ~70% were plated out into six-well plates. After 24 h, two sets of experiments were designed. In one set, medium was replaced by fresh 25 mM glucose-containing medium and in the other set medium was replaced by fresh medium containing 5.5 mM glucose. In both sets, cells were treated by GABA (25 μM), insulin (100 pM), or GABA (25 μM) + insulin (100 pM), and incubated for 24 h. At the end of incubation, plates were placed on ice and media were collected, centrifuged at 16,000 × g for 5 min, and supernatant was removed for glucagon measurement. The cells were washed three times with ice-cold PBS and scraped in Glycine-BSA buffer (100 mM glycine, 0.25% BSA, cOmplete Mini Protease Inhibitor Cocktail, 5 μg/mL Aprotinin, pH 8.8). The scraped cells were lysed by sonication (12 s at 30% amplitude on ice), and centrifuged at 16,000 × g for 45 min, from which the supernatant was collected for analysis. The protein concentration of the cell lysate was measured using BCA assay. To measure glucagon levels, the cell lysate or medium was mixed in an ethanol-acid solution (96% ethanol containing 0.18 M HCl) in a 1:3 ratio, kept at −20°C overnight, then centrifuged at 16,000 × g for 15 min at 4°C. The supernatant was then mixed with 20 mM Tris buffer, pH 7.5, and glucagon measurements were conducted by ELISA.

Experiments were done in three biological replicates, each of which had two technical replicates. Values were compared among treatment groups by one-way ANOVA using Sigma Stat 3.5 software (α = 0.05). For image analysis, co-localization of channels in the merged images was calculated by PCC using NIS-Elements software (Nikon, Canada).

Immunoblotting for organelle-specific markers confirmed enrichment of secretory granules (Supplementary Figures 1A–D). The final granule fraction was positive for the secretory granule marker, VAMP2. In contrast, the granule fraction did not contain the trans-Golgi marker TGN46, the nuclear envelope marker LaminB1, or the endoplasmic reticulum marker Calreticulin. As a positive control, the general cell lysate contained all four markers.

Secretory granules in alpha cells have been previously studied using transmission electron microscopy and their average sizes have been reported to be in the range of 180–240 nm (29–31). Accordingly, we confirmed the presence of secretory granules using nano-scale flow cytometry with Fc-glucagon as an exclusive marker for alpha cell secretory granules (32, 33). We used beads in the range of 110–880 nm for calibration in the range of the reported sizes for secretory granules (Supplementary Figure 2A). Fc-glucagon⁺ secretory granules distributed mostly to the gated regions of 179 and 235 nm (Supplementary Figures 2B,C), confirming enrichment of secretory granules from α-TC1-6 cells.

Fc or Fc-glucagon was purified from the granule lysate by affinity purification, and proteins that interact with either Fc alone or Fc-glucagon were identified with LC-MS/MS. Proteins that were pulled down by Fc alone in both 25 mM glucose (Supplementary Table 2) and 5.5 mM glucose (Supplementary Table 3) conditions were subtracted from the list of proteins identified using Fc-glucagon, thus identifying proteins that specifically interact with glucagon, which we term the glucagon interactome. Proteins were assigned the following categories: metabolic-secretory-regulatory, histones, cytoskeletal, and ribosomal. We identified 42 and 96 glucagon-interacting proteins within the category of metabolic-regulatory-secretory proteins when the cells were cultured in media containing 25 mM (Figure 1A) and 5.5 mM glucose (Figure 1B), respectively.

In media containing 25 mM glucose, there was a predicted direct interaction of glucagon with glucose regulated protein 78 kDa (GRP78 or Hspa5), and ATPase copper transporting alpha polypeptide (Atp7a) (Figure 1A), while in media containing 5.5 mM glucose, GRP78, Stathmin1 (Stmn1), and Heat shock protein 90- alpha (Hsp90aa1) were predicted to directly interact with glucagon (Figure 1B). Under conditions of either 25 or 5.5 mM glucose, one common predicted interaction was that between glucagon and GRP78.

Affinity purification of Fc-glucagon or Fc alone from the secretory granule lysate was followed by immunoblotting for GRP78. The presence of GRP78 immunoreactivity with Fc-glucagon, and not Fc alone, demonstrates a direct interaction with glucagon in the enriched secretory granules (Figure 2A). Immunofluorescence microscopy showed co-localization of GRP78 and endogenous glucagon within the secretory granules in α-TC1-6 cells (Figure 2B). There was a strong positive correlation between glucagon and GRP78 immunoreactivities (PCC = 0.85 ± 0.08), indicating significant co-localization of GRP78 and glucagon.

Interestingly, proteomic analysis also revealed the presence of histone proteins, along with structural proteins and ribosomal proteins, within the secretory granules in α-TC1-6 cells (Supplementary Table 4). Histone H4 was predicted to interact with glucagon in cells incubated in medium containing 5.5 mM glucose. Therefore, we reasoned that this interaction was responsive to external effectors. We treated α-TC1-6 cells with GABA, a well-known modulator of glucagon secretion (21) and examined the interaction between histone and glucagon. Co-immunoprecipitation of granule lysates, histone H4 ELISA of granule lysates, and immunofluorescence microscopy all validated the interaction of histone H4 with glucagon and presence of histone H4 in secretory granules of α-TC1-6 cells after treatment with GABA (Figure 3). Affinity purification of Fc-glucagon or Fc alone from the secretory granule lysate was followed by immunoblotting for histone H4 (Figure 3A). The presence of histone H4 immunoreactivity with Fc-glucagon, and not Fc alone, demonstrates a direct interaction with glucagon in the enriched secretory granules (Figure 3A). We then confirmed the presence of histone H4 in the enriched secretory granules of α-TC1-6 cells by ELISA (Figure 3B). In cells treated with GABA in 25 mM glucose, there was a detectable amount of histone H4 in the granules. That this result was not due to contamination from the nuclear fraction was shown by the finding that histone H4 levels were undetectable in the secretory granules of cells not treated with GABA. As a positive control, the nuclear fraction showed high levels of histone H4. Finally, immunofluorescence microscopy showed the presence of histone H4 in glucagon-containing secretory granules (Figure 3C), and there was significant co-localization with glucagon as assessed by Pearson's correlation coefficient (PCC = 0.78 ± 0.08).

Since the interaction between histone H4 and glucagon was dependent on glucose levels and GABA, we determined the effects of the major alpha cell paracrine effectors, GABA and insulin, on the glucagon interactome. The profiles of the metabolic-regulatory-secretory proteins that associate with glucagon within secretory granules were altered upon treatment with GABA, insulin or GABA + insulin, respectively, when α-TC1-6 cells were cultured in medium containing 25 mM glucose (Figure 4) and in 5.5 mM glucose (Figure 5). Additionally, we tabulated the profiles of histone, cytoskeletal, and ribosomal proteins in response to GABA, insulin and GABA + insulin in 25 mM glucose (Supplementary Tables 5A–C) or 5.5 mM glucose (Supplementary Tables 6A–C). The glucagon interactomes were functionally classified into the following groups: Binding, Structural molecule, Catalytic, Receptor, Translation regulator, Transporter, Signal transducer, Antioxidant. The proportion of proteins in each category is shown in the context of 25 mM glucose (Supplementary Table 7) and 5.5 mM glucose (Supplementary Table 8).

The protein networks that are predicted to interact with glucagon within the secretory granules under conditions of 25 mM glucose are illustrated in Figure 4. In cells treated with GABA, glucagon is predicted to directly interact with GRP78, HSP1B, HSP90, and vimentin (Figure 4A); however, in cells treated with insulin and GABA + insulin, glucagon interacts directly with only GRP78 (Figures 4B,C). The clusters of metabolic-secretory-regulatory proteins that make up the rest of the glucagon interactomes change in composition in response to the different treatments. The numbers of proteins categorized as “structural molecule activities” decreased in response to GABA (~45%) or insulin (~38%) and increased in the GABA + insulin group (~16%) compared to the control (Supplementary Table 7). The numbers of cytoskeletal proteins increased in the GABA (29%), insulin (12%), and GABA + insulin (35%) groups, while the numbers of ribosomal proteins decreased in those groups by 51, 14, and 66%, respectively (Table 1A).

Compared to cells incubated in medium containing 25 mM glucose, there were dramatic increases in the numbers of metabolic-regulatory-secretory proteins associated with glucagon after treatment with GABA, insulin or GABA + insulin in cells incubated in media containing 5.5 mM glucose (Figure 5). In cells treated with GABA, glucagon is predicted to directly interact with the following proteins: GRP78 (Hspa5), HSP 90alpha (Hsp90aa1), proprotein convertase subtilisin/kexin type 2 (PCSK2), heat shock 70 kDa protein 1B (Hsp1b), calmodulin 1(Calm1), and guanine nucleotide-binding protein G(I)/G(S)/G(O) subunit gamma-7 (Gng7) (Figure 5A). Under insulin treatment, the following proteins were predicted to directly interact with glucagon: GRP78, HSP 90-alpha, annexin A5 (Anxa5), stathmin1 (Stmn1), fatty acid synthase (Fasn), and chromogranin A (Chga) (Figure 5B); and only two proteins, GRP78 and PCSK2, were predicted to directly interact with glucagon after treatment with GABA + insulin (Figure 5C).

In the context of 5.5 mM glucose, the number of cytoskeletal proteins decreased, and the number of ribosomal proteins increased compared to cells treated with GABA, insulin and GABA + insulin in 25 mM glucose (Table 1B). Interestingly, the total numbers of proteins classified as “structural molecule activities” did not change appreciably across treatments (Supplementary Table 8). However, differences became apparent when cytoskeletal and ribosomal proteins were compared separately. When compared to 5.5 mM glucose alone, there were decreases of ~24 and ~35%, respectively, in the numbers of cytoskeletal proteins when cells were treated with GABA or insulin alone, but a ~71% increase in response to GABA + Insulin. Conversely, the numbers of ribosomal proteins increased by ~26 and ~43% in response to GABA and insulin, respectively, and decreased by ~69% in response to GABA + Insulin (Table 1B).

From our glucagon interactomes, we identified 11 proteins that interact with glucagon after treatment of α-TC1-6 cells with either GABA or insulin in media containing 25 mM glucose. To determine their effects on glucagon secretion, these proteins were depleted with siRNAs (Supplementary Figure 3) and glucagon secretion and cell content were measured. Of these 11 proteins, knockdown of ELKS/Rab6-interacting/CAST family member 1 (ERC1) increased glucagon secretion (p < 0.001), while gene silencing of 14-3-3 zeta/delta (KCIP-1), cytosolic malate dehydrogenase (MDH1), FXYD domain-containing ion transport regulator 2 (FXYD2) and protein disulfide-isomerase (PDI) reduced glucagon secretion to the same statistically significant level (p < 0.001). As well, knockdown of peroxiredoxin-2 (PRDX2), ATP synthase F1 subunit alpha (ATP5F1A), histone H4, and aconitate hydratase mitochondrial (ACO2) reduced glucagon secretion (p < 0.01), as did knockdown of alpha-tubulin 2 (AT2) (p < 0.05) (Figure 6A). Gene silencing of MDH1, PRDX2, ATP5F1A, and FXYD2 reduced cellular glucagon content to a significance level of p < 0.001. Gene silencing of KCIP-1, ACO2, Histone H4 and PDI all reduced the levels of cellular glucagon content to a significance level of p < 0.01 and that for ERC1 at p < 0.05 (Figure 6B). Gene silencing of GRP78 had no effect on glucagon secretion, and reduced cellular glucagon content (p < 0.05).

In the context of 5.5 mM glucose, significant reduction of glucagon secretion occurred by depletion of MDH1 (p < 0.05), PDI(p < 0.05), ERC1(p < 0.01), and ACO2 (p < 0.01). However, silencing of the other abovementioned genes did not significantly alter glucagon secretion (Figure 6C). Cellular glucagon content was significantly decreased by silencing of ATP5F1A (p < 0.05), AT2, PDI, ERC1, FXYD2, KCIP-1, histone H4, GRP78 (p < 0.01), ACO2, and PRDX2 (p < 0.001) (Figure 6D).

α-TC1-6 cells were cultured under high glucose conditions (25 mM) and then treated with paracrine effectors (GABA, insulin or GABA + insulin). The profiles of cumulative glucagon secretion and cellular glucagon content in 25 mM glucose was different from that in 5.5 mM glucose. While neither GABA nor insulin affected glucagon secretion in 5.5 mM glucose, they suppressed glucagon secretion in 25 mM glucose (Supplementary Figure 5A). In the context of 25 mM glucose, GABA reduced cellular glucagon content, while insulin increased cellular glucagon content (Supplementary Figure 5B); in contrast, neither GABA nor insulin alone affected cellular glucagon content, but in combination, they decreased cellular glucagon content.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.