Characterization of Antibodies to Products of Proinsulin Processing Using Immunofluorescence Staining of Pancreas in Multiple Species

Abstract

The efficient processing of proinsulin into mature insulin and C-peptide is often compromised under conditions of beta cell stress, including diabetes. Impaired proinsulin processing has been challenging to examine by immunofluorescence staining in pancreas tissue because the characterization of antibodies specific for proinsulin, proinsulin intermediates, processed insulin and C-peptide has been limited. This study aimed to identify and characterize antibodies that can be used to detect products of proinsulin processing by immunofluorescence staining in pancreata from different species (mice, rats, dog, pig and human). We took advantage of several knockout mouse lines that lack either an enzyme involved in proinsulin processing or an insulin gene. Briefly, we report antibodies that are specific for several proinsulin processing products, including: a) insulin or proinsulin that has been appropriately processed at the B-C junction; b) proinsulin with a non-processed B-C junction; c) proinsulin with a non-processed A-C junction; d) rodent-specific C-peptide 1; e) rodent-specific C-peptide 2; and f) human-specific C-peptide or proinsulin. In addition, we also describe two ‘pan-insulin’ antibodies that react with all forms of insulin and proinsulin intermediates, regardless of the species. These antibodies are valuable tools for studying proinsulin processing by immunofluorescence staining and distinguishing between proinsulin products in different species.

Introduction

Insulin is synthesized in pancreatic beta cells as preproinsulin within the rough endoplasmic reticulum (RER) and subsequently converted to proinsulin (Fig. 1). Once in the trans-Golgi network, proinsulin is then packaged into immature secretory granules, which undergo a series of processing steps that have been deduced largely based on mouse studies (Fig. 2). First, acidification of the granule lumen produces an environment conducive for the prohormone convertase (PC) enzymes, which are required to convert proinsulin into mature insulin and C-peptide (Goodge and Hutton 2000). Proinsulin undergoes post-translational processing by two endoproteinases, PC1/3 and PC2, as well as an exoproteinase, carboxypeptidase E (CPE). Although the proinsulin sequence differs between species (Fig. 1C), the peptide hormone segments are generally allocated as A-chain, B-chain, and C-peptide (Fig. 1A and 1B: blue, green and white boxes, respectively). PC1/3 cleaves proinsulin at the B-C junction, generating the ‘split-32,33 proinsulin’ intermediate, which is then further processed by CPE to remove two basic amino acids and generate the ‘des-31,32 proinsulin’ intermediate (Fig. 2). PC2 cleaves at the A-C junction to generate the ‘split-65,66 proinsulin’ intermediate (or ‘split-63,64 proinsulin’ in the case of mouse proinsulin1) and the di-basic residues are removed by CPE to produce ‘des-64,65 proinsulin’ (or ‘des-62,63 proinsulin’ in the case of mouse proinsulin1) (Fig. 2). Processing by PC1/3 and PC2 is thought to be sequential, based on observations that PC2 processed des-31,32 proinsulin more than 19 times faster than intact proinsulin, whereas PC1/3 had a similar processing rate for both proinsulin and des-64,65 proinsulin (Rhodes et al. 1992). Therefore, proinsulin is likely cleaved first at the B-C junction by PC1/3 and subsequently at the A-C junction by PC2, meaning that des-31,32 proinsulin would be the predominant intermediate product of proinsulin processing [(Goodge and Hutton 2000); illustrated in Fig. 2]. After removal of the basic residues left by PC1/3 and PC2, the final products of proinsulin processing include mature insulin (composed of the A-chain and B-chain) and C-peptide (Fig. 2), both of which are co-secreted in equimolar amounts within dense-core insulin secretory granules.

Figure 1.

Proinsulin amino acid sequence. (A) Schematic of the complete human proinsulin sequence. Blue boxes represent A-chain amino acids; green boxes represent B-chain amino acids, and white boxes represent C-peptide. One and three letter abbreviations are provided for each amino acid. Variations in the amino acid sequence are also illustrated for mouse proinsulin 1 (purple line) and mouse proinsulin 2 (pink line). (B) A general proinsulin sequence is shown to illustrate the site of cleavage for proinsulin processing enzymes. Yellow scissors indicate cleavage sites for prohormone convertase 1/3 (PC1/3) and PC2. Red triangles indicate cleavage sites for carboxypeptidase E (CPE). Yellow prisms represent di-sulphide bonds (S-S). (C) Proinsulin amino acid sequence of five different species, including human, dog, pig, mouse and rat. Grey boxes indicate regions with significant sequence variability among the species (amino acids that differ from the human proinsulin sequence are shown in purple within these regions).

Figure 2.

Schematic diagram representing the sequence of events required for processing proinsulin into mature insulin and C-peptide. The pathway on the right (indicated by solid arrows) is thought to be the predominant route of proinsulin processing, with cleavage first at the B-C junction by prohormone convertase 1/3 (PC1/3) to generate split-32,33 proinsulin, followed by removal of the two basic amino acids to generate des-31,32 proinsulin. Finally, PC2 and CPE act on the A-C junction to generate the mature C-peptide and insulin. The pathway on the left (indicated by dashed arrows), in which the A-C junction is processed first and the B-C junction second, is less dominant and thus the split-65,66 proinsulin, des-64,65 proinsulin, and diarginyl proinsulin intermediates are not typical products of proinsulin processing.

In healthy beta cells, proinsulin processing is highly efficient, with approximately 95% of proinsulin being converted into mature insulin and C-peptide (Steiner 2004). However, under conditions of beta cell stress, such as in patients with type 1 or 2 diabetes, there are disproportionately high levels of proinsulin intermediates in the pancreas and in the circulation (Roder et al. 1994; Saad et al. 1990; Snorgaard et al. 1990; Truyen et al. 2005; Ward et al. 1987; Yoshioka et al. 1988). Impaired proinsulin processing has also been described in various rodent models with genetic deletions or mutations, illustrating that a wide range of genes play an important role in this process (Furuta et al. 1998; Gosmain et al. 2012; Hatanaka et al. 2011; Naggert et al. 1995; Wijesekara et al. 2010; Zhu et al. 2002a). These studies generally used radioimmunoassays (RIA) or enzyme-linked immunosorbent assays (ELISA) to measure the ratio of circulating proinsulin to insulin, combined with an examination of insulin granule morphology by electron microscopy. Very few studies have used immunohistochemical or immunofluorescence techniques as a complimentary tool for studying proinsulin processing under conditions of beta cell stress, likely due to the lack of thorough validation of available commercial antibodies. Evaluating proinsulin processing by immunofluorescence is further complicated by the significant differences in proinsulin sequences between species (Fig. 1C) and the fact that rodents possess two proinsulins, each with different amino acid sequences spanning around A-C and B-C junction cleavage sites (Fig. 1A, 1C). Thus, antibodies that recognize proinsulin intermediates in one species may not be specific for the same intermediates in another.

The aim of this study was to characterize a panel of antibodies that would recognize each of the intermediate forms of processed proinsulin by immunofluorescence staining. To validate antibody specificity, we utilized pancreas tissue from various knockout mouse strains lacking either an enzyme involved in proinsulin processing (PC1/3, PC2 or CPE) or an insulin gene product (INS1 and/or INS2). We also examined the specificity of these antibodies in various species, including human, mouse, dog, pig and rat.

Materials & Methods

Immunofluorescence Staining

Immunofluorescence was performed on paraffin-embedded pancreas sections. Pancreas tissue was fixed overnight at 4°C in 4% paraformaldehyde (PFA; 1:10 vol/vol ratio of tissue to PFA), transferred to 70% ethanol for long-term storage, then paraffin-embedded and sectioned (5-μm thickness) by Wax-it Histology Services (Vancouver, Canada).

Pancreas sections were deparaffinized and rehydrated using a series of 5-min washes in xylene (×3) and graded alcohol solutions (100% ethanol ×2, 95% ethanol ×1, and 70% ethanol ×1), followed by washing in 1× PBS for 10 min on a shaker. Heat-induced epitope retrieval (HIER) was performed on slides using an EZ Retriever microwave oven (BioGenex; Fremont, CA) for 10 min at 95°C in 10 mM sodium citrate buffer (0.5% Tween 20, pH 6.0). Following HIER, slides were placed under running tap water for 5 min and then washed for 5 min in de-ionized water on a shaker, followed by PBS. Pancreas sections were then circumscribed with an ImmEdge Hydrophobic Barrier Pen^TM (Vector Laboratories, Burlingame, CA) and incubated for a minimum of 10 min with DAKO^® Protein Block, Serum Free (Dako; Burlington, Canada) at room temperature in a humid chamber. Sections were then incubated overnight in primary antibodies (Table 1) at 4°C in a humid chamber. On the following day, sections were gently rinsed with PBS and then washed 3 times in PBS (10 min per wash on a shaker), followed by 1 hr in secondary antibodies (Table 1) at room temperature in a dark humid chamber. Finally, sections were counterstained with a nuclear staining DAPI and mounted with a coverslip using VECTASHIELD^® Hard Set Mounting Medium with DAPI (Vector Laboratories, Burlingame, CA). The following negative control conditions were also included: no primary antibody, no secondary antibody, or a species-specific isotype control.

Table 1.

List of Antibodies, Sources, Catalog Numbers and Dilutions Used for Immunofluorescence Staining Experiments.

Antibody	Source	Catalog Number	Working Dilution
Rabbit anti-β-Galactosidase; pAb	Life Technologies (Carlsbad, CA)	A-11132	1:100
Rabbit anti-Insulin(C27C9); mAb	Cell Signaling Technology, (Beverly, MA)	3014	1:200
Guinea pig anti- Insulin(I8510); pAb	Sigma-Aldrich (St. Louis, MO)	I8510	1:1000
Mouse anti-Insulin(MAb1); mAb	EMD Millipore (Billerica, MA)	05-1108	1:200
Mouse anti-Proinsulin(GS9A8); mAb	Beta Cell Biology Consortium (Nashville, TN)	GS-9A8	1:50
Mouse anti-Proinsulin(82-PIN); mAb	Alpco (Salem, NH)	82-PIN-AC-mAb	1:1000
Rabbit anti-C-Peptide1 Ab657; pAb	Beta Cell Biology Consortium	AB1044	1:1000
Rabbit anti-C-Peptide2 Ab660; pAb	Beta Cell Biology Consortium	AB1042	1:1000
Rat anti-Proinsulin/C-Pep(GN-ID4); mAb	Developmental Studies Hybridoma Bank (DSHB) (Iowa City, IA)	GN-ID4	1:100
Guinea pig anti-C-Peptide(Abcam); pAb	Abcam (Cambridge, MA)	ab30477	1:100
Goat anti-Rabbit AF488	Life Technologies	A-11034	1:1000
Goat anti-Rabbit AF594	Life Technologies	A-11037	1:1000
Goat anti-Guinea Pig AF594	Life Technologies	A-11076	1:1000
Goat anti-Mouse AF488	Life Technologies	A-11029	1:1000
Donkey anti-Rat AF488	Life Technologies	A-21208	1:1000

All antibodies were diluted in Dako Antibody Diluent; specific dilutions and details about the primary and secondary antibodies are listed in Table 1. All insulin/proinsulin/C-peptide antibodies were co-stained with a general “pan-insulin” antibody, either rabbit anti-insulin^(C27C9) (Cell Signaling Technology; Table 1) or guinea pig anti-insulin^(I8510) (Sigma-Aldrich; Table 1), both of which were demonstrated to be immunoreactive for (pro)insulin in all species examined and also to cross-react with all intermediate forms of the processed proinsulin molecule (data not shown for anti-insulin^(I8510)). As a negative control for all insulin-related antibodies, we examined immunoreactivity in pancreas tissue from mice lacking both insulin1 and insulin2 (Ins1^-/-;Ins2^-/-).

Images were captured and analyzed using the ImageXpress® Micro XLS System, controlled by MetaXpress® High-Content Image Acquisition & Analysis Software (Molecular Devices Corporation; Sunnyvale, CA) with a scientific CMOS camera, a Nikon 20× Plan Apo objective (NA=0.75, 1-6300-0196; Nikon, Tokyo, Japan) and DAPI (DAPI-5060B), FITC (FITC-3540B) and Texas Red (TXRED-4040B) filter cubes.

Tissue Sources

All animal experiments were approved by the UBC Animal Care Committee. Pancreas tissues from the following species were used for validation of the proinsulin processing antibodies: human (adult tissue provided by the Ike Barber Human Islet Transplant Laboratory; Vancouver, BC), dog (adult male canine slides provided by Dr. Alan Cherrington; Vanderbilt University Medical Center, Nashville, TN), pig (4-year-old Yucatan^TM Miniature Swine; Sinclair Bio Resources, Columbia, MO), rat (8-week-old male Wistar rat; University of British Columbia Center for Disease Modeling, Vancouver, BC), and wild type mice (male C57BL/6, 12-week-old mice; The Jackson Laboratory, Bar Harbor, Maine). In addition, pancreas tissue was obtained from mice bred and maintained in our facility with the following genetic deletions: PC1/3^-/- (neonatal mice; Pcsk1tm1Dfs/J; Stock #006327; The Jackson Laboratory), PC2^-/- (18-week-old mice; B6;129-Pcsk2tm1Dfs/J; The Jackson Laboratory), CPE^-/- (B6.HRS(BKS)-Cpe^fat/J; The Jackson Laboratory), Ins1^-/-;Ins2^+/+ and Ins1^+/+;Ins2^-/- [12-week-old mice provided by Dr. James Johnson (University of British Columbia, Vancouver, BC); original source described in detail elsewhere (Mehran et al. 2012)]. Finally, pancreas tissue was obtained from neonatal mice with genetic deletion of both insulin genes, Ins1^-/-;Ins2^-/- (generated by breeding of Ins1^-/-;Ins2^+/- mice, which were provided by Dr. James Johnson).

Results

Validation of Insulin, Proinsulin and C-peptide Antibodies in Insulin Knockout Mice

The specificity of each antibody for a proinsulin-derived product was first validated in pancreas tissue from mice lacking both insulin genes (Ins1^-/-;Ins2^-/-). Since β-galactosidase (β-Gal) was knocked into the endogenous Ins2 locus in these mice, islets could be located by the presence of β-Gal protein (Fig. 3). We observed no detectable immunoreactivity with the insulin, proinsulin or C-peptide antibodies in pancreas tissue from mice lacking endogenous insulin production (Fig. 3), implying that these antibodies are specific for products of the insulin gene and do not cross-react with an unrelated protein in mouse pancreatic beta cells. Note that the proinsulin^(GNID4) and C-peptide^(Abcam) antibodies were not included in this experiment because they are human-specific antibodies that do not cross-react with mouse proinsulin/C-peptide (see later Figs. 9 and 10).

Figure 3.

Double immunofluorescence staining for β-Galactosidase (β-Gal; red) and insulin^(I8510), insulin^(MAb1), proinsulin^(GS9A8), or proinsulin^(82-PIN) (all green) in pancreas sections of mice lacking both insulin genes (Ins1^-/-;Ins2^-/-). Single immunofluorescence staining for insulin^(C27C9), C-peptide1, or C-peptide2 (green) was performed on adjacent serial sections. Two different islets are shown for β-Gal as a positive control; lack of immunoreactivity for each insulin-related antibody is shown from one of these two islet regions. All images are merged with DAPI nuclear staining (grey). Scale, 100 μm.

Figure 9.

Double immunofluorescence staining for (A) insulin^(C27C9) (pan-insulin antibody; red) and proinsulin^(GNID4) (green) in pancreas sections of wild type C57/BL6 mouse, rat, human, pig and dog. Individual channels, as well as merged overlay images are shown with DAPI nuclear staining (grey). (B) Predicted epitope for the proinsulin^(GNID4) antibody is shown in orange. Scale, 100 μm.

Figure 10.

(A) Double immunofluorescence staining for insulin^(C27C9) (pan-insulin antibody; red) and C-peptide^(Abcam) (green) in pancreas sections of wild type C57/BL6 mouse, rat, human, pig and dog. Individual channels, as well as merged overlay images are shown with DAPI nuclear staining (grey). (B) Predicted epitope for the C-peptide^(Abcam) antibody is shown in orange. Scale, 100 μm.

Insulin^(MAb1) Antibody (Mouse Anti-Insulin; Millipore)

The ‘insulin^(MAb1)’ antibody is reported to bind the free C-terminal end of the B-chain, including the B30 amino acid at the B-C junction of human insulin. Therefore, we predicted that this antibody would only bind to mature insulin or the ‘des-31,32 proinsulin’ intermediate that has been processed by both PC1/3 and CPE, since the epitope would be inaccessible if linked to C-peptide (proinsulin), or if the two arginine amino acids remained intact (split-32,33 proinsulin).

In our studies, wild type mouse islets showed strong insulin^(MAb1) immunoreactivity, which was notably diminished, but not abolished, in mice lacking either PC1/3 or CPE (Fig. 4A). The presence of weak insulin^(MAb1) immunoreactivity in these mice suggests that other processing enzymes may cleave the B-C junction to compensate for lack of PC1/3 or CPE, although with less efficiency. In PC2^-/- mice, the B-C junction would theoretically be cleaved normally by functional PC1/3 and CPE, meaning that the insulin^(MAb1) epitope would be accessible. Indeed, insulin^(MAb1) immunoreactivity was similar in PC2^-/- pancreas relative to wild type mice (Fig. 4A). Ins1^-/-;Ins2^+/+ mice had no insulin^(MAb1) immunoreactivity, whereas Ins1^+/+;Ins2^-/- mice had bright staining, indicating that this antibody is specific for mouse insulin1 and does not cross-react with insulin2 (Fig. 4A). These data indicate that replacing lysine (K) with methionine (M) at B29 is sufficient to disrupt the insulin^(MAb1) epitope (Fig. 4C).

Figure 4.

Double immunofluorescence staining for insulin^(C27C9) (pan-insulin antibody; red) and insulin^(MAb1) (green) in pancreas sections of (A) various mouse strains, including wild type C57/BL6, PC1/3^-/-, PC2^-/-, CPE^-/-, Ins1^-/-;Ins2^+/+ and Ins1^+/+;Ins2^-/-, and (B) various species, including rat, human, pig and dog. Individual channels, as well as merged overlay images are shown with DAPI nuclear staining (grey). (C) Predicted epitope for the insulin^(MAb1) antibody is shown in orange. Scale, 100 μm.

As expected, the insulin^(MAb1) antibody was strongly immunoreactive for insulin in adult human pancreas (Fig. 4B), where a threonine (T, bearing an accessible hydroxyl group) is located at the B30 position. The strong immunoreactivity observed in wild type mouse pancreas (Fig. 4A) indicates that the insulin^(MAb1) epitope was not affected by loss of the B30 methyl group, as a result of replacing threonine with serine. However, loss of the B30 hydroxyl group, caused by replacing threonine with alanine (A), abolished insulin^(MAb1) immunoreactivity in pig pancreas but did not affect insulin^(MAb1) staining in the dog pancreas (Fig. 4B). These data cannot simply be explained by differences in the amino acid sequence of the epitopes and thus are likely attributed to differences in the 3-dimensional conformation of insulin in dog and pig.

Taken together, our results suggest that the insulin^(MAb1) epitope spans the B29-30 amino acids in human, dog, mouse and rat insulin (rodent insulin 1, but not insulin 2) and is not accessible unless processed by PC1/3 and CPE (Fig. 4C). These data are summarized in Table 2.

Table 2.

Summary of Observed Immunoreactivity in Pancreas Tissue from Various Species and Knockout Mouse Models with Each of the Tested Antibodies.

Antibody	Fig(s).	Mouse (Wild type)	PC1/3-/-	PC2-/-	CPE-/-	Ins1-/- ; Ins2+/+	Ins1+/+ ; Ins2-/-	Ins1-/- ; Ins2-/-	Rat	Human	Pig	Dog
Insulin(C27C9)	3-6, 9, 10	+++	+++	+++	+++	+++	+++	-	+++	+++	+++	+++
Insulin(I8510)	7, 8	+++	+++	+++	+++	+++	+++	-	+++	+++	+++	+++
Insulin(MAb1)	3, 4	+++	+	+++	+	-	+++	-	+++	+++	-	++
Proinsulin(GS-9A8)	3, 5	+++ (PN)	+++ (CP)	N/A	+++ (CP)	+ (PN)	+ (PN)	-	+ (PN)	+++ (PN)	-	++ (PN)
Proinsulin(82-PIN)	3, 6	++ (PN)	N/A	+++ (CP)	N/A	++ (PN)	++ (PN)	-	++ (PN)	+++ (PN)	-	+/- (PN)
C-Peptide1	3, 7	+++	N/A	N/A	N/A	-	+++	-	+/-	-	-	-
C-Peptide2	3, 8	+++	N/A	N/A	N/A	+++	-	-	++	-	-	-
Proinsulin(GN-ID4)	9	-	N/A	N/A	N/A	N/A	N/A	N/A	-	+++	-	-
C-Peptide(Abcam)	10	-	N/A	N/A	N/A	N/A	N/A	N/A	-	+++	-	-

Level of immunoreactivity is indicates as follows: +++ Strong; ++ Moderate; + Weak; +/- Very Weak; - Negative; N/A Not Applicable. PN indicates perinuclear immunoreactivity and CP indicates cytoplasmic immunoreactivity.

Proinsulin^(GS-9A8) (Mouse Anti-Proinsulin; Beta Cell Biology Consortium)

We next examined the specificity and cross-reactivity of a putative proinsulin antibody that was reported to bind to the B-C junction of human proinsulin. Thus, it was predicted that this antibody would only react with unprocessed proinsulin that has not yet been cleaved by PC1/3 and/or CPE.

This antibody showed perinuclear and punctate staining in wild type mouse beta cells (Fig. 5A), which suggests localization of non-processed proinsulin within immature secretory granules in the trans-Golgi network or ER. In contrast, mice lacking PC1/3 or CPE showed strong cytoplasmic immunoreactivity (Fig. 5A), consistent with an accumulation of proinsulin with an intact B-C junction or the split-32,33 proinsulin intermediate. Proinsulin^(GS-9A8) immunoreactivity was significantly reduced, but not abolished in the absence of either insulin 1 or insulin 2 (Fig. 5A).

Figure 5.

Double immunofluorescence staining for insulin^(C27C9) (pan-insulin antibody; red) and proinsulin^(GS9A8) (green) in pancreas sections of (A) various mouse strains, including wild type C57/BL6, PC1/3^-/-, CPE^-/-, Ins1^-/-;Ins2^+/+ and Ins1^+/+;Ins2^-/-, and (B) various species, including rat, human, pig and dog. Individual channels, as well as merged overlay images, are shown with DAPI nuclear staining (grey). (C) Predicted epitope for the proinsulin^(GS9A8) antibody is shown in orange. Scale, 100 μm.

Consistent with the wild type mouse pancreas, bright peri-nuclear immunoreactivity was observed for proinsulin^(GS-9A8) in human pancreata and, to a slightly lesser extent, in dog and rat beta cells (Fig. 5B). Proinsulin^(GS-9A8) was virtually undetectable in pig pancreas (Fig. 5B) even though the predicted epitope is similar to that of dog (Fig. 5C). These data suggest that either the epitope is not accessible for this antibody in pig beta cells or porcine proinsulin is processed differently than dog proinsulin, resulting in undetectable proinsulin antigen in porcine beta cells.

Taken together, our results confirm that the proinsulin^(GS-9A8) epitope likely spans amino acids B28-30, plus the two arginines that link the B-C junction. Moreover, the sequence differences among the species did not affect the epitope availability, with the exception of pig pancreas (Fig. 5C). These data are summarized in Table 2.

Proinsulin^(82-PIN) (Mouse Anti-Proinsulin; Alpco)

The Alpco proinsulin^(82-PIN) antibody was reported to bind at the intact A-C junction of proinsulin. Thus, we predicted that this antibody would be useful for detecting a defect in processing by PC2.

As with the proinsulin^(GS-9A8) antibody (Fig. 5), we observed peri-nuclear and punctate staining with the proinsulin^(82-PIN) antibody in wild type mouse beta cells and in cells from both Ins1^-/-;Ins2^+/+ and Ins1^+/+;Ins2^-/- mice (Fig. 6A). These data suggest binding of proinsulin^(82-PIN) to non-processed proinsulin in immature secretory granules within the trans-Golgi network or ER and no specificity for either insulin 1 or insulin 2. Strong cytoplasmic immunoreactivity was observed in PC2^-/- beta cells (Fig. 6A), confirming that the predicted epitope likely spans the PC2 cleavage site (Fig. 6C). This antibody showed similar peri-nuclear immunoreactivity for proinsulin in mouse, rat and human pancreata, but the intensity was significantly reduced in dog beta cells and undetectable in pig beta cells (Fig. 6B and 6C). This is not surprising given the sequence differences at the A-C junction between pig and human proinsulin. These data are summarized in Table 2.

Figure 6.

Double immunofluorescence staining for insulin^(C27C9) (pan-insulin antibody; red) and proinsulin^(82-PIN) (green) in pancreas sections of (A) various mouse strains, including wild type C57/BL6, PC2^-/-, Ins1^-/-;Ins2^+/+ and Ins1^+/+;Ins2^-/-, and (B) various species, including rat, human, pig and dog. Individual channels, as well as merged overlay images are shown with DAPI nuclear staining (grey). (C) Predicted epitope for the proinsulin^(82-PIN) antibody is shown in orange. Scale, 100 μm.

C-Peptide 1 Antibody (Rabbit Anti-C-Peptide1; Beta Cell Biology Consortium)

We next examined an antibody that was predicted to be specific for mouse C-peptide 1 (amino acids C5-19) and to not cross-react with the rat C-peptide 1 or rat/mouse C-peptide 2. Our experiments confirmed that this antibody was highly immunoreactive in wild type mouse pancreas and did not cross react with C-peptide 2, as demonstrated by the absence of immunoreactivity in Ins1^-/-;Ins2^+/+ pancreas, compared with the strong immunoreactivity in Ins1^+/+;Ins2^-/- mice (Fig. 7A). This antibody produced no immunoreactivity in dog and pig pancreata, but did cross-react weakly with rat C-peptide 1 (Fig. 7B). In human pancreata, there was no immunoreactivity within beta cells, but the antibody cross-reacted with an unknown protein in a non-beta cell within human islets (Fig. 7B). Taken together, our results suggest that the antibody epitope likely spans amino acids C15-19 in mouse C-peptide 1, given that this region differs significantly from that on mouse C-peptide 2 (Fig. 7C). These data are summarized in Table 2.

Figure 7.

Double immunofluorescence staining for insulin^(I8510) (pan-insulin antibody; red) and C-peptide1 (green) in pancreas sections of (A) various mouse strains, including wild type C57/BL6, Ins1^-/-;Ins2^+/+ and Ins1^+/+;Ins2^-/-, and (B) various species, including rat, human, pig and dog. Individual channels, as well as merged overlay images are shown with DAPI nuclear staining (grey). (C) Predicted epitope for the C-peptide1 antibody is shown in orange. Scale, 100 μm.

C-Peptide 2 Antibody (Rabbit Anti-C-Peptide2; Beta Cell Biology Consortium)

In contrast to the C-peptide 1 antibody, the C-peptide 2 antibody was predicted to be specific for mouse/rat C-peptide 2 (amino acids C5-21) and not cross-react with C-peptide 1. As expected, strong immunoreactivity was detected in wild type and Ins1^-/-;Ins2^+/+ mouse pancreata, but no detectable staining was observed in Ins1^+/+;Ins2^-/- beta cells (Fig. 8A). This antibody was also immunoreactive in rat pancreas and cross-reacted with an unknown protein in non-beta cells from human islets, but produced no immunoreactivity in dog or pig pancreata (Fig. 8B). Therefore, our data suggest that the C-peptide 2 antibody epitope likely spans C13-21 amino acids in mouse and rat C-peptide 2 (Fig. 8C). These data are summarized in Table 2.

Figure 8.

Double immunofluorescence staining for insulin^(I8510) (pan-insulin antibody; red) and C-peptide2 (green) in pancreas sections of (A) various mouse strains, including wildtype C57/BL6, Ins1^-/-;Ins2^+/+ and Ins1^+/+;Ins2^-/-, and (B) various species, including rat, human, pig and dog. Individual channels, as well as merged overlay images are shown with DAPI nuclear staining (grey). (C) Predicted epitope for the C-peptide2 antibody is shown in orange. Scale, 100 μm.

Human-specific Proinsulin/C-Peptide Antibodies

Finally, we tested two antibodies that claimed to be specific for human proinsulin/C-peptide (Proinsulin^(GN-ID4); rat anti-Proinsulin/C-Peptide; DSHB) or human C-peptide (Guinea pig anti-C-Peptide; Abcam), although the exact epitopes were not specified. Based on the product information, these antibodies were predicted to cross-react with either mature human C-peptide or C-peptide of non-processed human proinsulin.

As expected, the proinsulin^(GN-ID4) antibody was found to be specific for human beta cells and immunoreactivity was undetectable in pancreatic beta cells from any other species tested (Fig. 9A). Likewise, the Abcam human C-peptide antibody was specific for human beta cells and did not produce immunoreactivity in beta cells of mouse, rat, pig or dog pancreas (Fig. 10A). Based on these data, we predict that both antibody epitopes likely span C15-C30 amino acids, as this region differs significantly between human, rodent, pig and dog proinsulin sequences (Figs. 9B and 10B). However, we are unable to conclude from the present data whether these antibodies are specific for either proinsulin and/or C-peptide, since we were unable to test the antibodies in appropriate human control tissues lacking various proinsulin processing enzymes. These data are summarized in Table 2.

Discussion

Impaired proinsulin processing is a hallmark feature of both type 1 and type 2 diabetes (Roder et al. 1994; Saad et al. 1990; Snorgaard et al. 1990; Truyen et al. 2005; Ward et al. 1987; Yoshioka et al. 1988). However, detecting the presence of various proinsulin intermediates within pancreata samples from patients with diabetes or through the use of animal models has been challenging without a reliable panel of validated antibodies. Here, we characterized commercially available antibodies with the goal of providing a valuable tool for examining proinsulin processing by immunofluorescence staining. Overall, we were encouraged to find that the reported epitope information for most antibodies allowed for accurate prediction of their binding specificity. We validated the specificity of these antibodies in various species for mature insulin, various proinsulin intermediates (resulting from defects in PC1/3, PC2 or CPE processing enzymes), rodent C-peptide 1 or 2, and human C-peptide. In addition, two pan-insulin antibodies that reacted with all forms of (pro)insulin, regardless of species, were reported.

Our studies took advantage of several mouse lines containing a genetic deletion for either a proinsulin processing enzyme or insulin gene(s) to validate the specificity of commercially available antibodies. For instance, PC2^-/- mice were previously shown to have reduced beta cell mass, accompanied by a 35% increase in total pancreatic proinsulin content and accumulation of des-31,32 proinsulin compared with wild type mice (Furuta et al. 1998; Furuta et al. 1997). However, despite this clear defect in proinsulin processing, approximately two-thirds of proinsulin was still converted to mature insulin, suggesting that, in the absence of PC2, rodent proinsulins can be completely converted by PC1/3, although with reduced efficiency (Furuta et al. 1998). This was reflected in our studies with PC2^-/- pancreas, in which the insulin^(MAb1) antibody showed strong immunoreactivity for proinsulin cleaved by PC1/3 and CPE. Moreover, we observed the cytoplasmic accumulation of proinsulin that contained an intact A-C junction, as demonstrated by the proinsulin^(82-PIN) antibody, confirming the PC2 defect in these mice. In contrast, a PC1/3 deficiency has been associated with more severe defects in proinsulin processing, as demonstrated by severe hyperproinsulinemia in PC1/3^-/- mice [~90% of pancreatic and circulating insulin-related immunoreactivity was a proinsulin intermediate product, predominately in the form of des-64,65 proinsulin (Zhu et al. 2002a; Zhu et al. 2002b)]. In our studies, the accumulation of proinsulin was easily detectable in PC1/3^-/- mice with the proinsulin^(GS-9A8) antibody, along with a significant reduction in insulin^(MAb1) immunoreactivity.

It is also interesting to note that, as in the PC1/3^-/- mice, a human subject with PC1/3 deficiency showed accumulation of plasma proinsulin and des-64,65 proinsulin (Jackson et al. 1997; O’Rahilly et al. 1995). However, the human patient developed early-onset obesity (Jackson et al. 1997), whereas the PC1/3^-/- mice displayed severe postnatal growth restriction (Zhu et al. 2002b). These contradicting reports suggest that the role of PC1/3 in processing proinsulin or non-insulin substrates may differ between mice and humans, thus warranting further studies in this area. Since both the proinsulin^(GS-9A8) and proinsulin^(82-PIN) antibodies displayed perinuclear immunoreactivity and lacked cytoplasmic staining in human beta cells (consistent with the pattern observed in mouse beta cells), we predict that these antibodies also bind to the B-C and A-C junction of human proinsulin, respectively. Therefore, these proinsulin antibodies may be useful for detecting the cytoplasmic accumulation of non-processed proinsulin or proinsulin intermediates (due to a PC1/3, PC2 and/or CPE defect) in clinical pancreas samples from human patients.

CPE^-/- mice were previously shown to exhibit obesity and hyperglycemia, associated with defective proinsulin processing and hyperproinsulinemia (Naggert et al. 1995). In our studies, CPE^-/- mice showed dramatically reduced immunoreactivity for mature insulin or des-31,32 proinsulin (using insulin^(MAb1) antibody) and an accumulation of cytoplasmic proinsulin (using proinsulin^(GS-9A8) antibody), thus confirming the previously reported proinsulin processing defects. The Ins1 and Ins2 knockout mice were ideal for validating the specificity of C-peptide 1 and C-peptide 2 antibodies. Although these antibodies will not be useful for detecting C-peptide in human, pig or dog tissues, they will be valuable for differentiating between the C-peptide molecules in rodent studies and for specifically detecting rodent C-peptide in islet xenograft studies. Likewise, the two human-specific C-peptide/proinsulin antibodies that do not cross-react with C-peptide/proinsulin in other species will be valuable for specifically detecting human insulin production in xenotransplant studies. However, the significant species differences in proinsulin peptide sequences proved to be challenging for detecting proinsulin products in pig and dog pancreata. In fact, with the exception of the pan-insulin antibodies, none of the commercially available antibodies tested here produced robust immunoreactivity in pig pancreas. Moreover, while our studies focused on immunofluorescence staining, we must also consider that the same challenges will apply to other antibody-based assays, including western blotting, RIA, ELISA.

Immunofluorescence staining provides a useful tool for detecting subtle heterogeneity that would otherwise be overlooked within a cell population by measuring the expression of proteins or genes in plasma/serum, or preparations from whole tissue or cell homogenates. A growing body of literature has revealed considerable heterogeneity within both the developing and adult beta cell populations (Bosco et al. 2007; Katsuta et al. 2012; Katsuta et al. 2010; Kiekens et al. 1992; Szabat et al. 2009; Van Schravendijk et al. 1992). With respect to proinsulin processing, a subset of beta cells within adult human pancreas tissue has been reported to lack PC2 expression (Portela-Gomes et al. 2008), which would certainly have implications for beta cell function; this also generates questions about the potential plasticity of human beta cells. Similarly, in the Goto-Kakizaki (GK) rat model of type 2 diabetes, immunofluorescence staining of pancreas tissue revealed a heterogeneous distribution of PC1/3, PC2, their respective inhibitors, 7B2 and ProSAAS, and CPE among insulin-positive cells, and this appeared to correlate with the level of insulin in each cell (Guest et al. 2002). Although there was no overall defect in proinsulin processing detected in the circulation of GK rats (Guest et al. 2002), we predict that heterogeneous processing defects would be observed in a subset of beta cells by immunofluorescence staining using the antibodies characterized here. Moreover, the heterogeneity of insulin-producing cells that arise during human pancreas development is not well understood, including the key differences between those that arise during the primary versus secondary transitions, and those that co-produce multiple pancreatic hormones versus monohormonal insulin-producing cells (Riedel et al. 2012). Our panel of proinsulin/insulin antibodies will be useful for studying proinsulin processing during pancreas development both in rodent models and human fetal samples. Moreover, proinsulin processing could be carefully assessed in insulin-producing cells generated by in vitro differentiation of human pluripotent stem cells to determine the degree of maturation and heterogeneity in this putative beta cell population as compared with both developing and mature insulin-producing cells from human tissues. In addition to the potential utility of these antibodies for investigating proinsulin processing, our studies reveal the importance of carefully characterizing antibody specificity using appropriate controls. This is particularly crucial for proinsulin antibodies, given that the predicted epitopes differ substantially between rodent proinsulin 1 and 2, and among different species.