Pancreatic β cell derived extracellular vesicles containing surface preproinsulin are involved in glucose stimulated insulin secretion

Abstract

Aims:

Extracellular vesicles (EVs) are involved in intercellular communication and are a topic of increasing interest due to their therapeutic potential. The aim of this study was to determine whether human islet-derived EVs contain insulin, and if so, what role do they play in glucose stimulated insulin secretion.

Main methods:

We isolated EVs from human islets culture and plasma to probe for insulin. Plasma from hyperglycemic glucose clamp experiments were also used to isolate and measure EV insulin content in response to a secretory stimulus. We performed immunogold electron microscopy for insulin presence in EVs. Co-culture experiments of isolated EVs with fresh islets were performed to examine the effect of EV cargo on insulin receptor signaling.

Key findings:

EVs isolated from culture medium contained insulin. Glucose treatment of islets increased the level of EV insulin. Hyperglycemic glucose clamp experiments in humans also lead to increased levels of insulin in plasma-derived EVs. Immunogold electron microscopy and proteinase K-digestion experiments demonstrated that insulin in EVs predominantly associated with the exterior surface of EVs while western blot analyses uncovered the presence of only preproinsulin in EVs. Membrane-bound preproinsulin in EVs was capable of activating insulin signaling pathway in an insulin receptor-dependent manner. The physiological relevance of this finding was observed in priming of human naïve islets by EVs during glucose stimulated insulin secretion.

Significance:

Our data suggest that (1) human islets secret insulin via an alternate pathway (EV-mediated) other than conventional granule-mediated insulin secretion, and (2) EV membrane bound preproinsulin is biologically active.

1. Introduction

Extracellular vesicles (EVs) are lipid bound vesicles secreted by cells into the extracellular milieu and they therefore carry contents and specific hallmarks of the cells from which they originated. Since their discovery, these submicron-sized vesicles have been shown to play roles in both physiological and pathological conditions. EVs are classified based solely on morphology and site of origin: Exosomes are EVs that are approximately 50 to 150 nM in size and originate from endosomal compartments; microvesicles are approximately 200 to 1000 nM in size and are generated from plasma membranes. These vesicles are cell-secreted phospholipid bilayer-bound structures that are generated by an evolutionary conserved process. EVs contents also include damage-associated molecular patterns (DAMPs), cytokines, RNAs, DNAs, and mitochondrial DNAs that are not unique to a specific cell of origin [1].

Glucose stimulated insulin secretion (GSIS) is integral to maintaining normal blood glucose. Insulin is first synthesized as preproinsulin which consists of signal peptide, insulin B-chain, C-peptide, and insulin A-chain. Newly synthesized preproinsulin is then goes through a co-translation/translocation process to the endoplasmic reticulum (ER), where the signal peptide (the pre portion) gets cleaved by signal peptidase. The proper folding of the remaining proinsulin happens in ER with the formation of three disulfide bonds. Proinsulin then migrates to insulin granules via Golgi complex, where mature insulin and C-peptide are formed. It has been reported that 5 to 10% of newly synthesized preproinsulin avoids the co-translation/translocation process and some is present in circulation (REF 16). The significance of free preproinsulin release from beta cells is unknown. In the present study we have shown that EVs secreted from human pancreatic beta cells carry preproinsulin mainly on their surface. We have demonstrated that the EV surface preproinsulin is functional with regards to the activation of insulin signaling pathway in an insulin receptor-dependent manner. The physiological significance of this is demonstrated by the priming of naïve human islets by the EVs in glucose stimulated insulin secretion. In both mice and humans, pancreatic islet-derived EVs are involved in both normal and pathophysiological islet biology [2, 3].

2. Materials and Methods

2.1. Human pancreatic islets

Human primary islets obtained from deceased individuals were obtained from Integrated Islets Distribution Program (IIDP). Donor information is provided in Table 1. The islets were delivered in PIM(S) medium from Prodo Laboratories (Aliso Viejo, CA).

Table 1.

Information of the human islet donors from which EVs were prepared.

Donor #	Gender	Age	BMI	Cause of death (COD)
1	F	28	24.7	Cerebrovascular/stroke
2	M	25	26.8	Head trauma
3	M	58	31.7	Anoxia
4	M	40	27.4	Head trauma
5	M	42	37.9	Cerebrovascular/stroke

2.2. EV isolation

Islets were washed three times with PIM(S) without human AB serum and resuspended in the same medium for further treatments. Islet-derived EVs were isolated from condition medium collected after 24 hrs. The conditioned medium was centrifuged at 300 x g for 10 mins, followed by another centrifugation at 2500 x g for 10 mins. The supernatant was then subjected to ultracentrifugation at 4°C for 2 hrs at 120,000 x g. Pellet was washed with 3 ml cold filtered PBS and was centrifuged at 4°C for 1.5 hrs at 110,000 x g. EV pellet was resuspended in filtered PBS and stored at −80°C. Human plasma EVs were isolated using ExoQuick reagent (System Biosciences) according to the manufacturer’s procedure. For co-culture experiments with naïve islets, islets were washed three times with PIM(S) without AB serum and resuspended in the same medium for treatment with either medium alone or with 16 mM glucose for 24 hrs. EVs were isolated from conditioned medium. An aliquot of 50 naïve islets were mixed with 35 x 10⁶ EV particles in a 24-well plate and incubated for 24 hrs. Another aliquot of 35 x 10⁶ EV particle was added to the same islet culture and incubated for further 24 hrs. After incubation, islets were washed twice with PIM(S) without serum and treated with 16 mM glucose for 18 hrs. Supernatants were used to measure the insulin levels by ELISA. Insulin levels were normalized by protein concentrations of total islets.

2.3. EV characterization

Particle concentration and the size distribution were determined by nano-flow cytometer (nanoFCM). Levels of EV cargo insulin was measured using the EV lysates by ELISA kits from Mercodia (Cat# 10-1131-10) following manufacturer’s instructions. For measuring human plasma EV insulin, Ultrasensitive Insulin ELISA kit from Mercodia (Cat# 10-1132-01) was used.

2.4. Immunogold electron microscopy studies (IEM)

IEM studies were performed by Microscope Facility in Johns Hopkins School of Medicine according to the procedure established by the facility [4, 5].

2.5. Proteinase K treatment of EVs

This procedure was followed as described in the manuscript by Javeed et al. [3]. In brief, EVs (6 x 10⁷ particles) were treated with either buffer alone or with buffer containing proteinase K (20 μg/ml) in the presence or absence of 1% Triton X-100 at 37°C for 1hr with occasional stirring. EVs were lysed and the levels of insulin were measured by ELISA.

2.6. Western blot analysis

Cells/islet tissues were washed with cold PBS before being lysed with ice-cold RIPA buffer, homogenized, and centrifuged at 14,000 rpm for 10 mins at 4°C, and then the supernatant was analyzed by western blot as described before [6]. Recombinant human insulin was obtained from Eli Lilly (Humulin R; Cat# NDC002-8215-01), and recombinant proinsulin was obtained from Amidebio (Cat# AA-015-01).

2.7. Antibodies

Anti-Alix (#92880), anti-Flotilin-1 (#18634), Anti-β actin (#3700), anti-IRβ (#23413), anti-phospho-AKT (S473) (#4060), ant-AKT (#9272) antibodies were from Cell Signaling. Antu-insulin (#ab181547) and rabbit IgG (#ab172730) were from Abcam. Anti-insulin antibody (#bs-0862R) for neutralization was from Bioss Antibodies. Anti-phospho-IRβ (MA515148), anti-rabbit IgG-HRP conjugated (#A27036, and anti-mouse IgG-HRP conjugated were from Thermo Fisher Scientific.

2.8. Co-culture experiments

Chinese hamster ovary (CHO)-K1 cells or CHO cells carrying human insulin receptors (CHO-IR) cells [7] were serum starved for 3 hrs at 37°C. 1ml cell suspension containing 0.32 x 10⁶ cells was distributed in a 6-well plate. An aliquot of EVs (total 5 x 10⁸ particles) isolated from human islets cultured overnight in medium containing 16 mM glucose was added to both CHO-K1 and CHO-IR cells and were incubated overnight again at 37°C. As a positive control both cell types were treated with insulin (1 nM) for 5 minutes at 37°C. Total cell lysates were analyzed by western blot analyses. For neutralization experiment, EVs (total 5 x 10⁸ particles) were rotated with either anti-insulin antibody or rabbit IgG for 4 hrs before addition to CHO-IR cells and were incubated overnight at 37°C. Total cell lysates were analyzed by western blot analyses.

2.9. Statistical analyses

GraphPad Prism v9 software was used for statistical analysis, and data are presented as means ± SEM.

3. Results

3.1. Characterization of human islet EVs

Before investigating the insulin-containing EVs in circulation we first looked for their presence in media in which human islets were cultured. Pancreatic islets from healthy human donors (2,000-5,000) were cultured for 24 hrs in PIM(S) medium containing 5.8 mM glucose in the absence of human AB serum before isolation of medium-derived EVs (Donor demographics are in Table 1) in the presence or absence of stimulatory levels of glucose (final 16 mM). Glucose treatment led to increased amounts of insulin in EVs (Fig. 1A). The percentage of total insulin secreted via EVs upon glucose stimulation varied from 0.9- 18%, as compared to basal level (5.8 mM glucose) of 0.06 to 5.9%. Particle concentration and size distribution profiles of EVs as determined by nanoFCM from a representative donor is shown in Fig. 1B. The sizes (mean: 66.49 ± 24.74 nm and median: 63.75 ± 21.74 nm for control; and mean: 54.39 ± 17.25 nm and median: 52.25 ± 18.33 nm for glucose treatment) were in the exosomes range (50 – 150 nm) [8]. Western blot analyses showed the characteristic markers for EVs (Fig. 1C). These data are in agreement with previously reported studies where human islets-derived EVs have been shown to contain insulin [9-11].

Fig. 1.

Characterization of human pancreatic islets EVs. (A) Glucose treatment of human islets increased EV insulin levels. Pancreatic islets were cultured in the absence or presence of glucose (16 mM) for 24 hrs. Isolated EVs were used to measure insulin content by ELISA. Insulin levels were normalized by EV counts. Individual results from five donors are shown here. (B) Particle concentration and size distribution profiles of EVs from a representative donor. (C) Western blot analyses of islet EV lysate. (D) Hyperglycemic glucose clamp experiment was performed. Plasmas were collected at time 0 and 100 mins post glucose infusion. EVs were used to determine insulin levels by ELISA. The levels of insulin were normalized by particle concentrations.

To investigate if EVs in human plasma also contain islet-derived EVs, we performed hyperglycemic continuous infusion clamp experiments (fasting blood glucose levels ± 90 mg/dl) for 2 hrs in normal healthy individuals (n=10) [12] and then isolated EVs from plasma. Donor information is provided in Table 2. As shown in Fig. 1D, 9 out of 10 individuals had higher levels of insulin in EVs at 100 minutes after continuous glucose infusion compared to 0 time points. Collectively our data suggest that EV-mediated insulin secretion is a physiologically relevant alternative pathway of insulin secretion as opposed to conventional granule-mediated secretion.

Table 2.

Information of the human donors used in Hyperglycemic continuous glucose clamp experiment.

Samples	Gender	Age	BMI
Donor 1	M	42	30.6
Donor 2	M	38	29.7
Donor 3	M	48	25.5
Donor 4	M	30	28.1
Donor 5	M	25	27.6
Donor 6	M	46	29.2
Donor 7	M	26	23.2
Donor 8	M	40	27.4
Donor 9	M	42	21.5
Donor 10	M	49	28

3.2. Characterization of EV insulin

To investigate the location of insulin in islet-derived EVs, we performed proteinase K digestion experiments as described in Materials and Methods. Proteinase K treatment alone without Triton X-100 significantly decreased EV insulin level (76.7% compared to untreated level), whereas in the presence of Triton X-100, proteinase K completely eliminated insulin (99.8%) (Fig. 2A). These data suggest that EV-associated insulin is predominantly present at the exterior surface of EVs but some is also within EVs. To confirm the presence of insulin on the surface of EVs, we took advantage of Immunogold Electron Microscopy technique (IEM) as described in Materials and Methods. Electron micrographs of sectioned EVs isolated from cultured human islets (Fig. 2B) showed immunogold-labelled EVs, where arrows indicate the gold particles. Isotype matched antibody was used as a negative control (Fig. 2C). Similar to islet-derived EVs, insulin was present both on the surface and inside of plasma-derived EVs (Fig. 2D), whereas no gold particle was observed in a negative control group (Fig. 2E). These data demonstrate that EVs from cultured human islets and human plasma contain insulin both inside and on their surface.

Fig. 2.

Characterization of EV insulin. (A) EVs (6 x 107) were treated with either PBS alone, or with proteinase K (20 ug/ml) in the presence or absence of 1% Triton X-100 for 1 hr. Total EV lysates were used to measure insulin by ELISA. (B) Electron micrographs of sectioned EVs from islets showed immunogold-labelled EVs, where arrows indicate the gold particles. (C) showed the negative control of immune staining using isotype matched antibody. (D) Electron micrographs of sectioned EVs isolated from human plasma, and the negative control (E). (F) Left panel: western blot analyses using lysates from human islets (20 μg) and human islet EVs (25 μg); Right panel: western blot analysis using recombinant human insulin, recombinant human proinsulin, and islet lysates from two different donors. The membranes were blotted with anti-insulin antibody.

3.3. Protein characterization of insulin in EVs

Next, we analyzed the EV insulin protein in lysates from EVs and corresponding islets by western blotting. Three bands corresponding to 12, 8, and 5 kd were present in islet lysates, whereas a single band corresponding to 12 kd only was in EV lysates (Fig. 2F, left panel). To identify these bands, we performed western blot analysis using recombinant insulin and proinsulin proteins along with islet lysates (Fig. 2F, right panel). The bands corresponding to 5 and 8 kd were identified as insulin and proinsulin, respectively. It is reasonable to conclude that the 12 kd band must be preproinsulin. Therefore, EVs contained only preproinsulin without any detectable amount of either proinsulin or mature insulin.

3.4. Functional significance of EV membrane-bound preproinsulin

To investigate whether membrane-bound preproinsulin can activate insulin receptors, CHO-K1 and CHO-IR cells were incubated with EVs (5x10⁸ EVs/0.32x10⁶ cells) as described in Materials and Methods. The phosphorylation status of insulin receptor β (IRβ Y1162/1163), was increased upon EV treatment, whereas this increase was not observed in CHO-K1 cells (Fig. 3A); similar results were also observed in insulin-treated CHO-IR cells compared to CHO-K1 cells (Fig. 3A). Additionally, the levels of phospho-AKT (Serine p473) were also increased only in CHO-IR cells in comparison to CHO-K1 cells upon EV and insulin treatment (Fig. 3A). The difference in the phosphorylation status was not due to the difference in protein loading as shown by the presence of equal amounts of β actin (Fig. 3A). Interestingly, the levels of total IRβ and AKT were increased after EV treatment. These increases were not observed with insulin treatment because of the short stimulation time. Therefore, the phosphorylation status of both IRβ and AKT was increased by the activation of kinases in case of insulin treatment, whereas in case of EVs, the increase in basal phosphorylation of both IRβ and AKT was due to the increase in protein levels. This indicates that there is a qualitative difference between mature insulin signaling and the signaling mediated by the EVs. These interesting observations could be due to the fact that the kinetics of IR signaling via membrane-bound preproinsulin is different from mature insulin.

Fig 3.

Functional significance of EV membrane-bound preproinsulin. CHO-K1 or CHO-IR cells were treated with either medium alone, EVs, or insulin as described in Materials and Methods. Total cell lysates were analyzed by western blot analyses (A). (B) Neutralization experiment with EVs. (C) Co-culture experiments with naïve human islets. Data shown are from representative experiments. This experiment was done at least three times with similar results. (D) Plasmas were collected from normal (n=6) and type 2 diabetic participants (n=10). Isolated EVs were used to determine insulin levels by ELISA. The levels of insulin were normalized by particle concentrations.

To determine if the membrane-bound preproinsulin of EVs was involved in activating CHO-IR cells, we used anti-insulin antibody to block the interaction between the preproinsulin and IR during co-culture experiment as described in Materials and Methods. As shown in Fig. 3B, the phosphorylation of AKT induced by EVs during co-culturing was inhibited significantly by the anti-insulin antibody, whereas normal rabbit IgG did not have any effect. These data suggest that the membrane-bound preproinsulin was involved in activating IR during coculture.

To investigate the physiological relevance of the interaction between membrane-bound preproinsulin and IR, we designed an experiment where naïve islets (50 islets) were incubated with either medium alone or with EVs as described in Materials and Methods. As shown in Fig. 3C, human islets pre-treated with EVs, secreted much higher levels of insulin after glucose treatment compared to glucose-treated naïve islets. These data suggest that EVs were able to entrain naïve islets responses to glucose. It has long been known that giving a glucose tolerant test (GTT) 2 days in a row will lead to increased insulin secretion on day 3 [13].

3.5. Functional significance of EV membrane-bound preproinsulin in type 2 diabetes

Next, we examined the insulin levels in plasma EVs isolated from normal and type 2 diabetic participants of the Baltimore Longitudinal Study of Aging (BLSA) [14]. As shown in Fig. 3D, higher levels of EV insulin were observed in diabetic participants compared to normal donors. Donor information is provided in Table 3. Among the diabetic donors having BMI 30 or higher (total 5 out of 10), 3 donors had higher EV insulin levels whereas all 5 donors had 100 mg/dL or higher basal plasma glucose levels. Total number of diabetic donors that had 100 mg/dL or higher basal plasma glucose level was 8 out of total 10 participants. 5 out of those 8 donors had higher EV insulin levels. To ascertain if there is any causal relation between EV insulin levels, BMI, and basal plasma glucose levels, we need to analyze more diabetic donors. Collectively, our data suggest that insulin carrying EVs may play a role in the development of diabetes.

Table 3.

Information of the normal and Type 2 diabetic donors.

Donors#	Gender	Age	BMI	Basal plasma glucose (mg/dL)
Normal donor 1	M	94	27.3	83
Normal donor 2	M	80	26.7	83
Normal donor 3	F	81	23.4	86
Normal donor 4	M	84	29.6	90
Normal donor 5	F	81	25.1	91
Normal donor 6	M	74	22.4	91
Diabetic donor 1	F	77	25.9	94
Diabetic donor 2	M	80	27.9	103
Diabetic donor 3	M	39	25.3	104
Diabetic donor 4	F	90	33.2	100
Diabetic donor 5	F	84	20.8	91
Diabetic donor 6	M	74	30.9	117
Diabetic donor 7	F	75	31	103
Diabetic donor 8	F	73	30.5	117
Diabetic donor 9	F	74	30.6	115
Diabetic donor 10	M	85	28.3	100

4. Discussion

In this report we have shown for the first time that human pancreatic islets secrete preproinsulin via EVs. Secreted preproinsulin is mainly on the surface of EVs indicating possible involvement of direct insulin receptor signaling. Indeed, we have demonstrated that EV surface preproinsulin activated insulin signaling pathway in an insulin receptor-dependent manner in a model system. More importantly, we demonstrated that EV-associated preproinsulin primed naïve islets to glucose stimulated insulin secretion. This autocrine mechanism may be involved in the development of type 2 diabetes where at the initial stage it helps beta cells maintaining normal physiology, whereas at the later stage it causes beta cell demise, as it has been proposed recently regarding the autocrine insulin signaling [15]. The exact mechanism of preproinsulin loading onto EVs is under investigation. It has been reported that 5 to 10% of newly synthesized preproinsulin avoids co-translation/translocation process involved in insulin granule synthesis [16]. It is possible that some of these newly synthesized preproinsulin shuttles to multivesicular bodies (MVB) where it becomes incorporated into intraluminal vesicles (ILV) and secreted as EVs [17]. The increased secretion of preproinsulin via EVs during the hyperglycemic glucose clamp experiment support our hypothesis that continuous glucose transport into beta cells diverts the migration of newly synthesized preproinsulin towards MVB and away from co-translation/translocation process.

The signal peptide of preproinsulin has been shown to be an autoantigen in type 1 diabetes [18-20]. The authors found that the epitopes present in the signal peptide are targeted by HLA-A restricted CD8 T cells resulting in the destruction of β cells in type 1 diabetes. It is tempting to speculate that EVs carrying preproinsulin is involved in activating immune system via uptake by dendritic cells. Similar phenomena have been reported in the case of EV containing proinsulin in the possible pathogenesis of type 1 diabetes [11].

Conclusions

In conclusion, our finding regarding the presence of functional preproinsulin on the surface of EVs may shed a light on the physiological role of islet-derived EVs in both normal and pathological conditions.

Data availability

All the data will be made available upon request to the senior author.