Revisiting Proinsulin Processing: Evidence That Human β-Cells Process Proinsulin With Prohormone Convertase (PC) 1/3 but Not PC2

In 1967, Steiner et al. (1) demonstrated with pulse-chase experiments that insulin is generated from a larger precursor they named “proinsulin.” After the general structure of insulin was identified as containing an NH₂-terminal B-chain and COOH-terminal A-chain with a connecting C-peptide, many studies have attempted to clarify how β-cells excise the C-peptide to liberate mature insulin. Current theory posits that the B-chain–C-peptide junction is cleaved by prohormone convertase 1/3 (PC1/3; gene PCSK1) before cleavage at the C-peptide–A-chain junction by prohormone convertase 2 (PC2; gene PCSK2). PC2 knockout mice have impaired processing at the C-A junction, resulting in a buildup of des-31,32 proinsulin (2), and PC1/3 knockout mice have severely impaired processing at the B-C junction, resulting in a buildup of des-64,65 proinsulin (3). Based on relative processing rates of intact human proinsulin versus des-31,32 proinsulin or des-64,65 proinsulin by rat PC1/3 and PC2, processing at the B-C junction by PC1/3 likely occurs before processing by PC2 at the C-A junction (4), but some data have countered this hypothesis by showing more buildup of des-64,65 proinsulin during the processing of rat insulin 2 in islets (5).

While past work supports the theory that primary mouse β-cells process proinsulin sequentially by PC1/3 and then PC2, there is some indication that PC2 may not be as important in human β-cells as it is in mouse β-cells. Humans with mutant PCSK1 have circulating hyperproinsulinemia (6), but there are no associations between PCSK2 polymorphisms and circulating proinsulin (7). Though a human β-cell line (EndoC-βH2) has abundant PCSK1 and PCSK2 (8), RNA-sequencing experiments on sorted primary α-cells and β-cells indicate higher expression of PCSK2 than PCSK1 in mouse β-cells (9), whereas human β-cells expressed ∼20 times more PCSK1 than PCSK2 (10) (Supplementary Fig. 1). Additionally, Davalli et al. (11) reported a deficiency of immunoreactive PC2 in human islet β-cells transplanted into mice and some human insulinomas are not immunoreactive for PC2 (12). Collectively, there is no definitive evidence that proinsulin is processed by both PC1/3 and PC2 in human β-cells, and there is some indication that PC2, while critical for proglucagon processing in α-cells (13), is not abundantly expressed in human β-cells.

In this study, we use validated antibodies and oligonucleotide probes to determine that primary human β-cells have no detectable PC2 and little detectable PCSK2. We also performed pulse-chase experiments and suppressed the function of PC2 by temperature blockade (14), brefeldin A (15,16), the weak base chloroquine (14), and multiple inhibitors (17,18). In the absence of full PC2 function, human β-cells can produce mature insulin, but neighboring α-cells produce little mature glucagon. Moreover, we provide evidence that PC1/3 is responsible for processing human proinsulin by inhibiting the function of PC1/3 using two inhibitors (18,19) to impair formation of mature insulin with no significant effect on proglucagon processing. These findings provide a more advanced understanding of the processing of proinsulin and suggest the need to reconsider the widely accepted thought that human proinsulin is processed sequentially by PC1/3 and then PC2. Our findings suggest that in human β-cells, PC1/3 is responsible for processing human proinsulin without PC2.

Human pancreas tissue biopsies were collected by the Ike Barber Human Islet Transplant Laboratory (Vancouver, British Columbia, Canada). Human pancreas tissue biopsies and cadaveric human islets were provided by the Alberta Diabetes Institute IsletCore (Edmonton, Alberta, Canada) after isolation by standardized protocol (20). Sample collection and islet isolation were approved by the Human Research Ethics Board at the University of Alberta (Pro00013094). All donors’ families gave informed consent for the use of pancreatic tissue in research, and all work with human tissues was approved by the Research Ethics Board (H14-02949), The University of British Columbia (Vancouver, British Columbia, Canada). Basic donor demographics are detailed in Supplementary Tables 2 and 3. All experiments with animals were approved by the University of British Columbia Animal Care Committee and carried out in accordance with the Canadian Council on Animal Care Guidelines.

C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) were sacrificed at 12 weeks of age. Following euthanasia, the pancreas was quickly dissected out of mice, washed in PBS, and fixed in 4% paraformaldehyde overnight before being transferred to 70% ethanol for storage prior to paraffin embedding and sectioning (5-µm thickness) (Wax-it Histology Services Inc., Vancouver, British Columbia, Canada). A similar protocol was used to collect pancreas from rats, pigs, and dogs. Human pancreas tissue biopsies were collected using the same protocol (Ike Barber Human Islet Tranplant Laboratory).

Immunofluorescent staining was performed as previously described (21). Briefly, sections were deparaffinized in xylene (three times for 5 min) and rehydrated in graded ethanol (100%, two times for 5 min, 95% for 5 min, 70% for 5 min, and PBS for 10 min) before heat-induced epitope retrieval in an EZ-Retriever microwave oven (BioGenex, Fremont, CA) for 15 min at 95°C in 10 mmol/L citrate buffer (0.5% Tween 20, pH 6.0) (Thermo Fisher Scientific, Waltham, MA). Samples were blocked in Dako Protein Block, Serum Free (Dako Canada, Burlington, Ontario Canada), and incubated overnight in primary antibody diluted in Dako Antibody Diluent. The following day, slides were washed and incubated in secondary antibody (Alexa Fluor–conjugated secondary antibodies; Life Technologies) for 1 h at room temperature before mounting and counterstaining with VECTASHIELD HardSet Mounting Medium with nuclear stain DAPI (Vector Laboratories, Burlingame, CA). All images were captured and analyzed with an ImageXpress Micro XLS System (Molecular Devices, LLC, San Jose, CA) with a scientific CMOS camera, a Nikon 20X Plan Apo objective (numerical aperture of 0.75, 1-6300-0196; Nikon, Tokyo, Japan), and DAPI (DAPI-5060B), FITC (FITC-3540B), Cy3 (Cy3–4040B), Texas Red (TXRED-4040B), and Cy5 (Cy5–4040A) filter cubes. Image analysis was performed on MetaXpress software (version 6.2.3.733; Molecular Devices, LLC).

A total of 10,000–20,000 human islet equivalents were dispersed in 0.05% trypsin before magnetic bead purification according to a published protocol (22).

Groups of 250 mouse islets, 1,000 human islet equivalents, or ∼1 million EndoC-βH1 cells were lysed in 200 μL lysis buffer (50 mmol/L Tris, pH 8.0, 150 mmol/L NaCl, 0.02% Na azide, 0.1% SDS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mmol/L phenylmethyl sulfonyl fluoride, and protease inhibitor cocktail [Sigma-Aldrich, St. Louis, MO]) as per a published protocol (15). Levels of proteins of interest were assessed by fluorescent Western blotting methods using two PC2 antibodies (1:1,000, MAB6018, R&D Systems; and 1:1,000, PA5-14594, Thermo Fisher Scientific), anti-PC1/3 antibody (1:2,500; gift from Lakshmi Devi), and an anti–α-tubulin antibody (1:1,000; Sigma-Aldrich) and detected on a LI-COR Odyssey 9120 Imaging system (LI-COR Biosciences, Lincoln, NE).

For all in situ hybridization experiments, we used tissue samples collected and sectioned in all RNase-free solutions (including paraformaldehyde, 70% ethanol, and water for sectioning) and cleaned all equipment with RNase AWAY (Thermo Fisher Scientific). We performed a standard protocol using a modified version of a commercially available in situ hybridization kit (BioChain Institute Inc., Newark, CA). We used three probes specific for PCSK2 in regions highly conserved between human and mouse and used a primary mouse anti-digoxigenin antibody and secondary goat anti-mouse secondary antibody conjugated to alkaline phosphatase (Supplementary Key Resources Table).

Quantitative PCR was performed using a standard protocol. mRNA expression level is represented as 2^−ΔCt (where Ct is threshold cycle) using the average of two internal control genes (GAPDH and ACTB).

After receipt of islets, they were incubated overnight in complete media (CMRL) (Corning insulin-transferrin-sodium selenite [Corning, NY], GlutaMAX [Thermo Fisher Scientific], BSA [Roche Diagnostics, Laval, Quebec, Canada], and penicillin/streptomycin). The next day, islets were preincubated for 60 min in Krebs Ringer bicarbonate buffer, HEPES, and 16.7 mmol/L glucose (KRBH-16.7) at 37°C, pulsed for 40 min in 1 mL KRBH-16.7 with 800 μCi/mL [³H]Leu (PerkinElmer, Waltham, MA), and chased for 120 min. Islets were split into groups of 1,000 and chased in KRBH-2.8 in eight conditions: 37°C, 20°C, 37°C with 10 μg/mL brefeldin A (Sigma-Aldrich), 37°C with 30 μmol/L chloroquine, 37°C with one of two PC1/3 inhibitors (537076, EMD Millipore; or 166811, MedChem ShortCut LLC) (18,19), or 37°C with one of two PC2 inhibitors (166830, MedChem ShortCut LLC; or 5408-0471, ChemDiv) (17,18). Islets were lysed in 150 μL lysis buffer (15).

Islet lysates were incubated overnight at 4°C with 5 μg carrier-free anti-insulin antibody (C27C9; Cell Signaling Technology, Danvers, MA) followed by 6-h incubation at 4°C with 50 μL protein-A/G magnetic beads (50% slurry in lysis buffer) (Thermo Fisher Scientific). After washing three times, immunoprecipitated insulin was eluted by heating to 95°C in Tris-tricine sample buffer (Bio-Rad Laboratories). We immunoprecipitated glucagon with the same protocol using an anti-glucagon antibody (EP3070; Abcam). Samples were then separated using a Tris-tricine PAGE approach to separate small proteins (23). After separation, gels were stained by colloidal Coomassie (24), and visualized mature and prohormone bands were excised manually. Gel slices were then dissolved in 0.3 mL of 30% hydrogen peroxide for 3 days at 50°C. Scintillation counting was performed on an LS6500 scintillation system (Beckman Coulter, Brea, CA) with quench curve correction for final disintegrations per minute counts in Ultima Gold LLT scintillation fluid (PerkinElmer). Counts were adjusted for background and number of leucines in prohormones compared with processed mature hormones (25).

Statistical analyses were done in GraphPad Prism V7, and statistical tests performed are noted in figure legends. We used P < 0.05 as the cutoff for significance.

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, T.J.K. The data sets generated and/or analyzed during the current study are available from the corresponding author on reasonable request. Critical antibodies and probes used in the current study are also available from the corresponding author on reasonable request.

To minimize risk of nonspecific findings, we performed antibody validation experiments. There were identical immunostaining patterns for PC2 using multiple antibodies in human and mouse pancreas (Fig. 1A). To negatively validate with a knockout model, we used an antibody that binds to the deleted NH₂-terminal prodomain of PC2 (26). This antibody produced no immunoreactivity in pancreas from PC2 knockout mice (Fig. 1B). By Western blot, two antibodies (MAB6018, red, R&D Systems; and PA5-14594, green, Thermo Fisher Scientific) immunoreacted with recombinant human PC2 and produced a blot with the expected bands for pro-PC2 (75 kDa) and mature PC2 (64 kDa) from both human and mouse islet extracts (Fig. 1C). Using four antibodies, we detected robust immunoreactivity for PC2 in mouse β-cells and neighboring islet cells but failed to detect immunoreactivity for PC2 in human β-cells (Fig. 1D and see Supplementary Fig. 2A for single channels). Notably, though not detectable in human β-cells, we observed the expected robust PC2 immunoreactivity in α-cells. These findings were replicated in four additional organ donors (Supplementary Fig. 3). We determined whether there is a distinct distribution of PC2 in the head, body, and tail regions of human pancreas and whole mouse pancreas by immunostaining for insulin, glucagon, and PC2 (Fig. 1E). Unlike in the mouse, in human pancreas the intensity of immunoreactivity for PC2 in β-cells was dramatically less than in α-cells and there was almost no colocalization of insulin and PC2 in all regions of human pancreas. The rate of colocalization of PC2 and insulin was not significantly different than the rate of colocalization of insulin and glucagon in the head, body, or tail of human pancreas (P > 0.05; data not shown). In all regions of human and mouse pancreas, there was high colocalization of PC2 with glucagon. Unlike in the mouse, PC1/3 and PC2 did not colocalize in human pancreas (Supplementary Fig. 2B). We immunostained for the essential PC2 cofactor neuroendocrine protein 7B2 (gene SCG5), and though there was immunoreactivity for 7B2 in human α-cells and β-cells, signal intensity was lower in β-cells, unlike in mice, in which there was similar immunoreactivity in both cell types (Supplementary Fig. 4).

We immunostained pancreas from pig, rat, and dog (Fig. 2A). Unexpectedly, β-cells were not immunoreactive for PC2 in rat pancreas. β-Cells in pig and dog pancreas were immunoreactive for PC2 using two antibodies, but one antibody failed to produce immunoreactivity in any islet cells of these species, likely due to incompatible epitopes (Supplementary Table 1). There is an R-to-G and N-to-H substitution in pig PC2 in the immunizing peptide sequence used to generate the PC2 COOH-terminal antibody and a D-to-N substitution in dog PC2 in the likely epitope of the PC2 NH₂-terminal antibody (Supplementary Table 1).

We investigated the presence of PC2 in the human β-cell line EndoC-βH1 (27). We used the PC2 knockout αTC1ΔPC2 mouse α-cell line, α-TC1 mouse α-cell line that has PC2 nearly exclusively, and β-TC3 mouse β-cell line with both PC2 and PC1/3 as controls. EndoC-βH1 cells are immunoreactive for both PC1/3 and PC2 (Fig. 2B). Transcript levels of PCSK1 and PCSK2 are comparable in EndoC-βH1 cells, similar to human islets (Fig. 2C). We investigated protein levels of PC1/3 and PC2 by Western blots on EndoC-βH1 cells and human islets and confirm histological findings of comparable levels of PC1/3 and PC2 in EndoC-βH1 cells (Fig. 2D). The major PC1/3 band was at the expected size of mature PC1/3 (66 kDa; labeled with pink arrowhead in Fig. 2D), and minor bands were at the expected sizes of the PC1/3-processing precursors (74 kDa and 87 kDa) (28). Additionally, we analyzed sorted human β-cells and generated four to eight times enrichment of β-cells (based on Ins ΔCt) and >99% removal of α-cells (based on Gcg ΔCt; data not shown). Using highly limited sample sizes of protein extracts for Western blot, there was no visible PC2 band in purified β-cells, whereas whole islets from the same donors yielded clear PC2 bands (Supplementary Fig. 5).

To localize PCKS2 mRNA within pancreatic islets, we performed in situ hybridization on sections of human and mouse pancreas (Fig. 3). With overlaying immunofluorescence for insulin and glucagon, intense PCSK2 signal within human islets was localized to α-cells, with a marginally detectable signal in β-cells (Fig. 3A). This contrasts to the expected pan-islet signal in mouse islets (Fig. 3B). We prioritized generating robust, sensitive signal in human pancreas and note that this led to substantial background signal in mouse pancreas. By assessing the pattern of chromogenic deposition, positive cytoplasmic signal (with discernable nuclear holes) in mouse pancreatic islets can be clearly separated from background signal on the edges of acinar lobules.

Given that PC2 is ∼80–100 times more catalytically active than PC1/3 (29), we sought to validate histological findings with sensitive biochemical approaches to assess whether human β-cells have low yet catalytically relevant levels of PC2. Given that neighboring α-cells within human islets require PC2 to produce glucagon from proglucagon (13), we assessed both proinsulin processing and proglucagon processing in pulse-chase experiments. Importantly, we chose to study primary human islets in order to specifically answer the question as to how human β-cells process proinsulin. It seems that PC2 can access and cleave human proinsulin (30), but studying the ability of PC2 to process human proinsulin in models like EndoC-βH1 cells or other cell lines would not clarify the important question of how proinsulin is processed in bona fide human β-cells. In human islets, there was a near-total blockage of glucagon production by all PC2-inhibiting chase conditions (Fig. 4A) but no significant difference in proinsulin processing (Fig. 4B). To determine if PC1/3 is predominantly responsible for proinsulin processing in human β-cells, we used two PC1/3 inhibitors. Neither inhibitor significantly impaired the processing of proglucagon, but both led to a significant blockade of proinsulin processing.

Given the known defects in proinsulin processing in patients with diabetes, we immunostained pancreata from donors with type 1 (n = 7) and type 2 (n = 22) diabetes and compared them to those of a new cohort of control subjects without diabetes (n = 10). We again observed minimal colocalization of PC2 and insulin in control subjects, but there was some heterogeneity, with 1 out of 10 donors having occasional colocalization of PC2 and insulin. This contrasts with observations that almost all donors with type 2 diabetes (20 out of 22) had obvious insulin and PC2 colocalization (see high-magnification insets in blue boxes in Fig. 5A). We then quantified the fluorescent intensity of PC1/3 or PC2 immunoreactivity in β-cells (using insulin immunoreactivity to designate β-cell area) and α-cells (using glucagon immunoreactivity to designate α-cell area). Pancreas from donors with type 2 diabetes had higher PC1/3 and PC2 immunoreactivity in both β-cells and α-cells (Fig. 5B). Furthermore, the intensity of PC1/3 immunoreactivity in α-cells relative to β-cells was significantly higher in donors with type 2 diabetes, and the intensity of PC2 immunoreactivity in β-cells relative to that in α-cells was also significantly higher in donors with type 2 diabetes compared with donors without diabetes (Fig. 5C). There was no predictive value of diabetes duration, HbA_1c, BMI, sex, use of metformin (n = 15) versus solely lifestyle intervention (n = 7), age, or cold ischemia time of organ collection to predict outcome variables of PC immunoreactivity (data not shown). We also immunostained pancreas from donors with type 1 diabetes (n = 7). Regrettably, we were unable to clearly identify insulin-immunoreactive β-cells in tissue sections. Abundantly identifiable glucagon-immunoreactive α-cells in pancreas from donors with type 1 diabetes had PC1/3 and PC2 immunoreactivity similar to that of control pancreas (data not shown). Furthermore, in sections from donors with type 1 diabetes, we observed frequent PC2-immunoreactive cells that were not immunoreactive for insulin or glucagon (Fig. 5A).

The prevailing theory that proinsulin processing requires both PC1/3 and PC2 is best supported by studies of islets from PC2 and PC1/3 knockout mice (2,3). To our knowledge, the roles of PC1/3 and PC2 for processing human proinsulin have never been rigorously examined. We performed immunohistofluorescence with well-validated antibodies and made the surprising observation of virtually absent PC2 immunoreactivity in both human and rat β-cells. Interestingly, others have observed minimal PC2 immunoreactivity in human β-cells (31) and, in one case, interpreted the finding as a sign of β-cell dysfunction after transplantation into immunodeficient rodents (11). Yet others have reported abundant PC2 immunoreactivity in human β-cells (32), but that work used a polyclonal antibody (33) with an immunizing peptide containing significant homology between human PC1/3 (RRDELEE) and PC2 (KKEELEE), raising the possibility of significant cross-reactivity with PC1/3. Likewise, prior detection of PC2 immunoreactivity in rat β-cells by immunohistochemistry (34) and immunogold electron microscopy (35) relied on polyclonal antibodies generated with large PC2 immunogens containing regions of high homology to PC1/3 (PC2: TNACEGKEN vs. PC1/3: TRACEGQEN [34] or identical sequences FALALEAN and LTWRDMQHL in both PC2 and PC1/3 [35]). By Western blot using an antisera to a large fragment of PC2 (D174-S384) containing substantial regions of homology to PC1/3, rat insulinoma granules have immunoreactivity for PC2 (36). However, using a specific antibody with an immunogen similar to that of one of the antibodies we use in the current study (PC2^C-term), there seems to be little PC2 in rat β-cells at ∼95% purity (37). Given that rat insulin 2 is processed first at the C-A junction (38), a site favorable for PC1/3 (RQKR) (39), the theory is supported that PC1/3 could be exclusively responsible for proinsulin processing in rat, though the possibility is not ruled out that M at the P4 position (with the system of Schechter and Berger [40] for denoting positions prior to [Px] or after [P′x] the scissile bond) is the cause for slower B-C junction processing. Additionally, only PC1/3 is regulated by high glucose in rat islets (41), suggesting that rat β-cells may not depend on the function of PC2. The processing of proinsulin in rat β-cells is worth studying in greater detail in the future, and rat islets may be an excellent model for studying human proinsulin processing.

Unlike in human and rat pancreas, abundant PC2 immunoreactivity was evident in β-cells from mouse, pig, and dog pancreas. In these three species, the C-A junction of proinsulin has PC1/3-unfavorable amino acids at the P4 position (mINS2, Q; pigINS, P; and dogINS, L) (42), while both rat insulins have a PC1/3-favorable R (39). Mouse pro–islet amyloid polypeptide (pro-IAPP) requires PC2 to process its NH₂-terminal site (43). Mouse, dog, and pig NH₂-terminal pro-IAPP all have the amino acid M in the P4 position. By contrast, both rat and human have a V at P4. We propose that the P4 M at the NH₂-terminal processing site of proIAPP is unfavorable for processing by PC1/3 and a contributor to the species differences in β-cell PC2 expression. This is supported by observations that rat INS2 with a P4 M at the B-C junction of proinsulin is unique among rodent insulins in that it is processed at the CA junction before the B-C junction (5), perhaps because rat β-cells lack PC2 and PC1/3 is slow to process the B-C junction. Additionally, in PC2^−/− mice that possess a human IAPP transgene, there is no increase in the ratio of circulating NH₂-proIAPP_1-48 to IAPP and no increased amyloid formation (44), suggesting no defect in human pro-IAPP processing in the absence of PC2. Taken together, our findings of the species differences of PC2 immunoreactivity in β-cells suggest that PC2 is essential for the processing of NH₂-terminal pro-IAPP when there is a P4 M.

In addition to examining protein levels of PC2, we assessed mRNA levels of PCSK2 in human pancreas. Our detection of low-level PCSK2 in human β-cells by in situ hybridization aligns with single-cell RNA-sequencing experiments (45,46) reporting detectable PCSK2 in human β-cells, albeit at much lower levels than in α-cells, unlike in mice in which PCSK2 is abundant in both α-cells and β-cells (47). Occasional intense PCSK2 signals in non–α-cells within human islets are likely δ-cells, given the known role for PC2 in excising somatostatin-14 (26). Detectable PCSK2 mRNA in conjunction with a lack of PC2 immunoreactivity suggests that there is posttranscriptional regulation of PCSK2 in human β-cells—a possibility warranting follow-up. We also detected immunoreactivity for the essential PC2 cofactor 7B2 (gene SCG5) in human β-cells, albeit at lower intensity than in human α-cells. These results are not surprising, given high SCG5 mRNA levels in sorted human β-cells (10). Though all brain PCSK2⁺ cells are also SCG5⁺, there are many SCG5⁺ cells that lack PCSK2 (48). 7B2 has been shown to suppress aggregation of β-amyloid in cell lines and synuclein in cell-free experiments (49). Further, 7B2 has been shown to suppress cytotoxicity and aggregation of IAPP in culture (50) and could have other roles in human β-cells.

It is notable that convertase expression can be regulated. In rodent models of hyperglycemia, there can be a compensatory upregulation of PC1/3 in α-cells (51,52), which can increase intraislet production of GLP-1 and reduce rates of apoptosis in β-cells (53,54). PC2-deficient mouse α-cells used to generate the αTC1ΔPC2 cells spontaneously produce PC1/3 and begin excising GLP-1 from proglucagon (55). Intestinal L cells of PC1/3^−/− mice abnormally produce mature glucagon, likely attributable to PC2 action (56). Potentially, a similar initiation of PC2 expression occurs in β-cells of humans with PCSK1 mutations who present with hyperglycemia and elevated circulating des-64,65 proinsulin (57). Additionally, expression of PC2 in the human EndoC-βH1 cell line could reflect immaturity of the cells, as PC2 is expressed during fetal development of early endocrine cells (58). Further investigation into the dynamic regulation of PC1/3 and PC2 in not only rodent models but also human β-cells under normal and pathological conditions is warranted.

By pulse chase, human β-cells did not have significant impairment of proinsulin processing in conditions capable of fully blocking the PC2-dependent processing of proglucagon to glucagon. Based on our immunostaining experiments, if β-cells do contain any PC2, it is far less than the PC2 content of α-cells, thus making blockade of proglucagon processing a stringent positive control of PC2 inhibition. It is unsurprising that temperature blockade at 20°C and brefeldin A had variable blockade in the processing of proinsulin because both block shuttling of proinsulin into secretory granules and temperature blockade have been demonstrated to partially supress proneuropeptide Y processing by PC1/3, albeit to a lesser degree than by PC2 (14). Additionally, given that we can only confirm robust PC2 inhibition based on blockade of proglucagon processing but cannot confirm a lack of PC1/3 suppression, nonsignificant yet numerical decreases in proinsulin processing in PC2-inhibited conditions may be best attributed to modest reductions in PC1/3 activity. Both PC2 inhibitors used in pulse-chase experiments function as allosteric inhibitors and have been shown to inhibit PC1/3 by 10–20% when used in cell-free experiments at lower concentrations (10 μmol/L [18] and 25 μmol/L [17]). Notably, β-cells are sensitive to PC1/3 inhibition, as PC1/3 heterozygous mouse β-cells with a likely ≤50% loss of PC1/3 (and abundant PC2) have modestly impaired proinsulin processing (3). Additionally, PC2^−/− mouse β-cells have a substantially worse impaired proinsulin processing phenotype than human islets exposed to robust PC2 inhibitors. In contrast to PC2-inhibited conditions, PC1/3-inhibited conditions that did not impair proglucagon processing significantly impaired proinsulin processing. We cannot confirm the extent of PC1/3 inhibition, and incomplete PC1/3 inhibition is a potential explanation for the incomplete blockade of proinsulin processing. Regardless, these findings provide validation for our experimental technique and clearly show that PC1/3 is important to the processing of proinsulin.

The pulse-chase study design was limited by a lack of a molecular knockdown approach for PCSK2 (e.g., lentiviral delivery of shRNA) (59). Though the specificity of a molecular approach is appealing and may be a worthwhile follow-up study, the confounding effects of islet dispersion, cell reaggregation, and prolonged culture to allow time for mRNA knockdown and PC2 protein turnover are considerable. Another important limitation is that as we were unable to differentiate intact proinsulin from proinsulin processing intermediates using size-based proinsulin/insulin separation by SDS-PAGE, we were unable to rule out the potential that some single-site processing of proinsulin is occurring in PC1/3-inhibited conditions. Given relatively normal production of fully mature insulin (necessarily processed at both the B-C and C-A junctions to liberate C-peptide and reduce the molecular mass to ∼6 kDa) in PC2-inhibited conditions that were sufficient for an ∼100% blockade of proglucagon processing, this appears unlikely.

Impaired proinsulin processing has prognostic value for progression from autoantibody positivity to type 1 diabetes (60) and from impaired glucose tolerance to type 2 diabetes (61). Some patients with type 1 diabetes lose detectable circulating C-peptide but retain detectable proinsulin (62) and pro-IAPP (63). With new insight into the processing of prohormones in healthy β-cells, we immunostained pancreas from organ donors with diabetes to determine if there were defects in PC immunoreactivity as an explanation for defective proinsulin processing in diabetes. Interestingly, there was an increase in both PC1/3 and PC2 immunoreactivity in β-cells of donors with type 2 diabetes. Notwithstanding challenges of substantial heterogeneity and a lack of data on circulating proinsulin in donors studied, we hypothesize that increased PC1/3 immunoreactivity could function as a compensatory response to increased insulin-production demands during type 2 diabetes. Further, in conjunction with adaptive increased β-cell PC1/3, aberrant upregulation of PC2 may contribute to incomplete proinsulin processing. Additionally, we observed increased α-cell PC1/3 in donors with type 2 diabetes, confirming previous findings (64) and indirectly suggesting increased islet production of GLP-1 in patients with type 2 diabetes.

Clarifying the process of proinsulin processing is important to understand defective prohormone processing during diabetes. The current work challenges the prevailing assumption that human β-cells process proinsulin by PC1/3 and then PC2 sequentially and reveals that abnormal PC2 expression in human β-cells is associated with type 2 diabetes. Though evidence supports the dogma that PC1/3 and then PC2 process proinsulin in mouse β-cells, our findings suggest PC2 has little to no role in healthy human β-cells.

Acknowledgments. The authors express their gratitude to organ donors and their families. Human samples for research were provided by the Alberta Diabetes Institute IsletCore at the University of Alberta in Edmonton (http://www.bcell.org/human-islets.html) with the assistance of the Human Organ Procurement and Exchange (HOPE) program, Trillium Gift of Life Network (TGLN), and other Canadian organ procurement organizations. The authors also thank Drs. Allan Cherrington (Vanderbilt University), Gunilla Westermark (Uppsala University), and Lakshmi Devi (Mount Sinai School of Medicine) for sharing antibodies and Drs. Lucy Marzban and C. Bruce Verchere (The University of British Columbia [UBC]) for their helpful advice. A.R. gratefully acknowledges studentship support from the Canadian Institutes of Health Research (CIHR) (Vanier Canada Graduate Scholarship) and Vancouver Coastal Health (CIHR-UBC MD/PhD Studentship).

Funding. This work was supported by grants from the Canadian Diabetes Association and the CIHR (CIHR Foundation Scheme).

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

Author Contributions. A.R. and A.A. performed experiments. A.R. analyzed data and drafted the manuscript with contributions from A.A. and T.J.K. All authors approved the final draft. T.J.K. 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.