Evidence for a Role of the Ubiquitin-Proteasome Pathway in Pancreatic Islets

Male Sprague-Dawley rats (Taconic Farms, Germantown, NY) weighing 90–100 g were submitted to 90 or 60% pancreatectomy as previously described (20). In sham-operated rats, the pancreas was manipulated with no removal of tissue. Islets were isolated 2 or 4 weeks after surgery. All animal procedures were done according to the guidelines of the Joslin Diabetes Center Animal Care Committee.

Rat islets were isolated according to a previously described method (21). Pancreatic tissue was digested with collagenase (Liberase RI; Roche Diagnostics, Indianapolis, IN). Islets were separated by centrifugation through Histopaque-1077 gradient (Sigma Chemical, St. Louis, MO) and further purified by hand picking. Islets from pancreatectomy experiments were immediately used for RNA isolation (22). Islets used for in vitro experiments were kept overnight in RPMI-1640 medium (Life Technologies, Grand Island, NY), 10% FCS (Mediatech, Herndon, VA), glutamine (2 mmol/l; Life Technologies), penicillin (100 IU/ml), and streptomycin (100 μg/ml).

The next day, islets were washed with RPMI-1640 containing 2.8 mmol/l glucose and then aliquoted (125–200 islets) and cultured in low (2.8 mmol/l) or high (16.7 mmol/l) glucose for 2 or 24 h. Four experiments with parallel low- and high-glucose incubations were performed, including in each experiment three to four replicates of each culture condition.

For glucose-stimulated insulin secretion experiments, the islets were preincubated in low-glucose RPMI for 2 h, adding the proteasome inhibitors 10 μmol/l lactacystin or 20 μmol/l MG-132 (Calbiochem, La Jolla, CA). Islets were then transferred to low- or high-glucose RPMI, with the mentioned proteasome inhibitors, for incubation during 2 or 24 h. Several experiments with different rat islet isolations were performed, with three to six replicates for each experiment. Incubations were done in 24-well culture plates, with 125–200 islets in 1 ml media.

Following the same scheme, three experiments were performed, incubating the islets in the presence or absence of 10 μmol/l lactacystin and 1 μmol/l Gö6976 (protein kinase C [PKC] inhibitor; Calbiochem), 2 mmol/l EGTA, or 10 mmol/l arginine (Sigma). Three repetitions of each culture condition were performed in each experiment, using 50 islets/0.5 ml with glucose stimulation lasting 2 h.

After the glucose stimulation, media were collected for insulin quantification. Islets were used for either RNA isolation or insulin extraction, the latter with acidic ethanol. Insulin levels were quantified by radioimmunoassay.

Total RNA was isolated from samples of 125–200 rat islets using the Trizol reagent as indicated by the manufacturer (Life Technologies). To obtain cDNA, reverse transcription was done in a 25-μl reaction containing 0.5 μg total RNA, 50 mmol/l Tris (pH 8.3), 75 mmol/l KCl, 3 mmol/l MgCl₂, 10 mmol/l dithiotreitol, 40 units RNasin (Promega), 1 mmol/l each dNTPs, 50 ng random hexamers, and 200 units Superscript-II RNaseH⁻ reverse transcriptase (Life Technologies). The reverse transcription reaction consisted of 10 min at 25°C, 42 min at 60°C, and 10 min at 95°C. At the end, 50 μl water was added to obtain 75 μl of a cDNA solution containing 10 μg RNA equivalents in a 1.5-μl volume.

A semiquantitative multiplex radioactive PCR was used for evaluation of the levels of several transcripts (Table 1). In each case, PCR conditions were optimized in terms of primer concentration, number of cycles, and annealing temperature in order for the reaction to work within a linear range of amplification (22). Several genes along with an internal control (α-tubulin or cyclophilin) were grouped for amplification in the same tube. Primers sequences, multiplex groups, and other PCR details are described in Table 1.

The composition of the 25-μl PCR was 1.5 mmol/l MgCl₂, 200 μmol/l each dNTP, 40–400 nmol/l primers (Table 1), 1.5 μl cDNA, 2.5 units AmpliTaq-Gold DNA polymerase (Applied Biosystems, Foster City, CA), and 1.25 μCi (3,000 Ci/mmol) α³²P-dCTP (New England Nuclear, Boston, MA) in the GeneAmp PCR buffer provided with the enzyme.

The PCR started with activation of the polymerase (10 min at 94°C). Cycles (Table 1) consisted of denaturing (1 min at 94°C), annealing (1 min at 58°C), and extension (1 min at 72°C). Five microliters of loading buffer were added, and a 10-μl aliquot run in 6% polyacrylamide gel in Tris–boric acid–EDTA buffer. Dried gels were exposed to a Phosphorimager screen (Molecular Dynamics, Sunnyvale, CA) and the intensity of the bands quantified using the Image Quant software (Molecular Dynamics). The quantification of a particular band was expressed as a ratio with the corresponding internal control amplified in the same tube.

Rat pancreata were fixed in 10% neutral formalin and embedded in paraffin. Sections, 5 μm each, were double stained using rabbit anti–protein gene product 9.5 (PGP9.5) (the isoform Uch-L1) 1:2,000 (Chemicon International, Temecula, CA) and guinea pig anti-insulin 1:200 (Linco Research, St. Charles, MO). PBS (pH 7.4) was used for all washes and antibody dilutions. Antigen retrieval was performed by microwaving the sections (1,000 W, 100% power, three times for 5 min) with 0.01 mol/l sodium citrate (pH 6.0). Nonspecific binding was blocked with 1.5% donkey serum. Primary antibodies were incubated overnight at 4°C. Anti–guinea pig IgG Texas-Red and anti-rabbit IgG fluorescein isothiocyanate diluted 1:200 (Jackson Immunoresearch, West Grove, PA) were incubated for 1 h at room temperature. Stained sections were mounted with 90% glycerol and 0.1 mol/l Dabco (Sigma). Images were captured with a Zeiss LSM 410 confocal microscope.

For protein extraction, rat tissues were homogenized in 80 mmol/l Tris-HCl (pH 6.8), 5% SDS, and the Complete cocktail of protease inhibitors (Roche, Mannheim, Germany). Protein concentration was estimated with bicinchoninic acid (Pierce, Rockford, IL). Twenty micrograms of protein aliquots were run in 12% SDS-PAGE and transferred onto polyvinylidine fluoride membrane (BioRad, Hercules, CA). The membrane was blocked for 2 h with 2% nonfat milk in TTBS (Tris-buffered saline with Tween: 0.9% NaCl, 100 mmol/l Tris [pH 7.5], 0.05% Tween 20), and anti-PGP9.5 1:2,000 (Chemicon International) was incubated for 1 h at room temperature. After washes with TTBS, anti-rabbit IgG–horseradish peroxidase 1:8,000 (Promega; Madison, WI) was incubated for 1 h at room temperature. The chemilunescent reaction was done with Reinassance ECL (Perkin Elmer, Boston, MA).

Data are expressed as means ± SE. The unpaired Student’s t test was used to compare gene expression levels in cultured islets exposed to low or high glucose and in islets isolated from pancreatectomized or sham animals. Statistical significance for glucose-stimulated insulin secretion experiments was determined by ANOVA.

Islet cells of the endocrine pancreas and neurons share many features, such as positive immunostaining with the neuroendocrine marker PGP9.5 (23–26), which has been identified as the isoform Uch-L1 (10). We studied the distribution of the Uch-L1 protein in paraffin sections of rat pancreas with an anti-PGP9.5 antiserum. Strong anti-PGP9.5 staining was found only in endocrine islet cells, with no staining of the surrounding acinar tissue (Fig. 1). The staining was mostly cytoplasmic, although nuclear staining was also evident in some cells. Insulin-negative cells in the islet mantle were even more strongly stained than β-cells. To evaluate the level of expression of Uch-L1 in islets compared with other tissues, we performed Western blot and RT-PCR. The amount of the Uch-L1 protein in isolated islet extracts was similar to that in brain extracts (Fig. 2A). Testis extracts showed a less prominent band. No signal was detected in protein extracts from liver, kidney, spleen, heart, intestine, or pancreas. It is worth mentioning the absence of signal in total pancreas extracts compared with the strong band detected in isolated islets, which may result from dilution of endocrine-specific proteins within the whole organ extract.

cDNA was prepared from total RNA extracted from different rat tissues to perform semiquantitative RT-PCR. Uch-L1 band was quantified and expressed relative to an internal control (α-tubulin) and amplified in the same tube. The levels of Uch-L1 mRNA in brain, testis, and isolated islets were much higher than in liver, kidney, spleen, muscle, and pancreas (Fig. 2B), consistent with the protein levels determined by Western blot.

Knowing the abundance of Uch-L1 in islets, attention was focused on the possible role of Uch-L1 on β-cell function. We determined the effect of glucose on Uch-L1 expression by RT-PCR in isolated cultured islets exposed to low (2.8 mmol/l) and high (16.7 mmol/l) glucose. Results of expression at basal 2.8 mmol/l glucose were designated as 100%. Uch-L1 expression did not change after 2-h exposure to high glucose (91 ± 5%), but there was a significant decrease (54 ± 6%) after 24-h exposure (Fig. 3A).

To investigate the effect of chronic hyperglycemia in an in vivo system, we measured Uch-L1 mRNA levels in islets isolated from 90% pancreatectomized rats or from sham-operated animals. Plasma glucose measured in the fed state confirmed hyperglycemia in the pancreatectomy animals (251 ± 24 vs. sham 78 ± 2 mg/dl). Two weeks after surgery, the expression of Uch-L1 in islets from pancreatectomized rats was significantly reduced (60 ± 9 vs. sham 100 ± 8%) (Fig. 3B). Islets from animals 4 weeks after pancreatectomy still showed a reduced Uch-L1 expression (74 ± 3 vs. sham 100 ± 6%). These data indicate a downregulation of the Uch-L1 gene after long-term high-glucose exposure.

With the aim of dissecting the effect of hyperglycemia from that of the islet growth triggered by partial pancreatectomy, we used the 60% pancreatectomy model, in which the reduction of islet mass does not lead to hyperglycemia (20,27). In contrast to the reduction of Uch-L1 expression seen in islets from 90% hyperglycemic pancreatectomized rats, islets obtained from 60% pancreatectomized rats 4 weeks after surgery showed a slight but significant increase in Uch-L1 expression (130 ± 4 vs. sham 100 ± 3%) (Fig. 3C).

The numerous proteins involved in the ubiquitin-proteasome pathway are widely expressed and play important roles in a variety of cellular processes. To examine whether the glucose effect on Uch-L1 extended to other proteins in this pathway, we selected an array of genes to evaluate the possible influence of glucose on their expression levels in islets. So far, there are no studies of ubiquitin-proteasome pathway genes in pancreatic islets. A list of selected genes with a brief description of their functions is presented in Table 2.

Isolated rat islets were used for total RNA isolation and semiquantitative RT-PCR analyses. Expression levels of each gene are presented relative to a housekeeping control gene amplified in the same tube (Table 1). Results are expressed as a percentage of the expression of the gene in islets incubated in basal (2.8 mmol/l) glucose. We did not detect any major change in the mRNA levels of any of the genes studied after 2-h incubation, although some (parkin, E6-associated protein [E6-AP], and human ubiquitin-conjugating enzyme 5 [UbcH5]) showed a slight tendency to decrease (Fig. 4A). A longer incubation (24 h) revealed that high glucose produced a significant decrease in the mRNA levels of parkin (49 ± 7%), UbcH5 (62 ± 4%), and β-TRCP (transducin repeat–containing protein; 73 ± 4%); the decrease in E6-AP did not reach statistical significance. The levels of the transcripts for c-Cbl, Nedd4, and the polyubiquitin genes A and B remained unchanged (Fig. 4B).

To investigate the influence of glucose in islets exposed for longer periods of time, we moved to the 90% pancreatectomy model of chronic hyperglycemia. For pancreatectomy islets isolated 4 weeks after surgery, polyubiquitin-B and c-Cbl mRNA levels were significantly increased (133 ± 9 and 131 ± 9%, respectively), while E6-AP expression was decreased (59 ± 9%) (Fig. 5). As occurred with cultured islets exposed to high glucose, parkin and UbcH5 showed a tendency to be downregulated in pancreatectomy islets (54 ± 18 and 71 ± 8% respectively), although the changes were not significant.

The results shown above indicate the presence in islets of various ubiquitin pathway components and suggest that there could be an interaction between the glucose metabolism and the ubiquitin-proteasome pathway. We then questioned whether this pathway could have some role in glucose-stimulated insulin secretion. For this purpose, we used two inhibitors of the proteasome, MG-132 and lactacystin. MG-132 belongs to the peptide aldehyde group of proteasome inhibitors. Its inhibition is reversible but also affects other proteolytic activities such as lysosomal cysteine proteases and calpains. Lactacystin, an antibiotic isolated from actinomycetes, is much more specific (although it can also inhibit cathepsin-A) and inhibits proteasomal degradation binding covalently and irreversibly to the proteasome complex (7).

Rat islets were incubated in RPMI with 2.8 or 16.7 mmol/l glucose during 2 or 24 h. When required by the experimental condition, lactacystin (10 μmol/l) or MG-132 (20 μmol/l) were present during these time periods. After 2-h incubation (Fig. 6A), lactacystin significantly increased the insulin released at high glucose (812 ± 44 pg · islet^–1 · ml^–1 · h^–1) compared with glucose-stimulated islets without lactacystin (547 ± 53 pg · islet^–1 · ml^–1 · h^–1). Similar increases were found in separate sets of experiments (Fig. 7). At basal glucose, insulin secretion in the presence of lactacystin (259 ± 56 pg · islet^–1 · ml^–1 · h^–1) was not significantly different from that found with control islets (191 ± 35 pg · islet^–1 · ml^–1 · h^–1). In contrast, MG-132 did not increase the glucose-stimulated secretion (411 ± 96 pg · islet^–1 · ml^–1 · h^–1) (Fig. 6A). We checked the viability of the islets after the incubation and found that MG-132 was causing some cell death, while lactacystin exerted no visible adverse effect on islet viability (results not shown). Therefore, we decided to continue the study using only the more specific, less damaging agent lactacystin. When we prolonged the incubation period to 24 h (Fig. 6C), the basal insulin secretion in the absence of lactacystin was 118 ± 11 pg · islet^–1 · ml^–1 · h^–1, and glucose stimulated secretion up to 733 ± 147 pg · islet^–1 · ml^–1 · h^–1. In the presence of lactacystin, the basal insulin secretion was unaffected (96 ± 12 pg · islet^–1 · ml^–1 · h^–1) but glucose-stimulated secretion decreased to about half (360 ± 59 pg · islet^–1 · ml^–1 · h^–1) compared with that of control (733 ± 147 pg · islet^–1 · ml^–1 · h^–1).

To check the possibility that lactacystin was causing depletion of insulin stores due to a toxic effect on β-cells, we measured the insulin content of the islets and found no difference between control (15.2 ± 2.3 ng/islet) and lactacystin-treated (13.8 ± 1.0 ng/islet) islets (Fig. 6B).

To determine which element of the insulin secretory machinery was affected by lactacystin, we performed glucose stimulation in the presence of the PKC inhibitor Gö6976, the Ca²⁺ chelator EGTA, and the secretagogue amino acid arginine (Fig. 7). As previously observed, lactacystin increased the insulin secretion in glucose-stimulated islets (741 ± 88 vs. control 405 ± 126 pg · islet^–1 · ml^–1 · h^–1), but there was no significant difference when the stimulation was performed in the presence of Gö6976 (891 ± 90 pg · islet^–1 · ml^–1 · h^–1). Likewise, EGTA (2 mmol/l) abolished glucose-stimulated insulin secretion in both lactacystin-treated (149 ± 37 pg · islet^–1 · ml^–1 · h^–1) and control (190 ± 105 pg · islet^–1 · ml^–1 · h^–1) islets. Moreover, using arginine (10 mmol/l) instead of glucose as stimulus for insulin release (Fig. 7B), lactacystin did not increase secretion (315 ± 57 pg · islet^–1 · ml^–1 · h^–1) compared with control islets without lactacystin (316 ± 61 pg · islet^–1 · ml^–1 · h^–1).

Because of the known importance of the ubiquitin-proteasome pathway, it is not surprising that this pathway could play a central role in β-cell physiology. Uch-L1 expression is highly restricted to the central and peripheral nervous system and dispersed neuroendocrine cells (23–25) but is present in other locations as well, including testis (24,28). It has been used as a marker for neuroendocrine tumors (29). In a mouse model of tumorigenesis, Power et al. (30) showed an increase in PGP9.5 staining in hyperplastic and neoplastic β-cells but no staining in normal islets, while other authors (26) report PGP9.5 in normal islets and duct cells during development and regeneration. Here we demonstrate the expression of Uch-L1 in islets by immnostaining, Western blot, and RT-PCR and show that the level of expression is as high as in brain. Furthermore, immunocytochemistry served to locate the protein in both β- and non–β-cells, with seemingly higher expression in islet non–β-cells.

The decrease observed in Uch-L1 expression in islets exposed to high glucose either in tissue culture (short-term exposure) or in 90% pancreatectomized rats (chronic exposure) indicates a regulation depending on the physiologic status of the endocrine cells. In the 90% pancreatectomy model, a similar decrease of other key β-cell genes has been previously reported (22). In contrast, in 60% pancreatectomized rats, which do not become hyperglycemic, we observed a slight increase in Uch-L1 expression. The 60% pancreatectomy is a model of β-cell adaptation and growth (27); Uch-L1 overexpression could be related to such processes, as its expression has been reported in pancreas development and regeneration (26) and in neoplastic β-cells (30).

Several human neurodegenerative diseases are characterized by the formation of inclusion bodies in the brain, which frequently contain both ubiquitin and Uch-L1 (31). It has been suggested that under a continuous stress, the proteasome pathway in neurons would be saturated and unable to dispose of defective proteins, resulting in their accumulation and formation of plaques (32). Similarly, in type 2 diabetes, some islets contain amyloid deposits that may contribute to β-cell destruction (33). Interestingly, in insulinomas, the amyloid deposits are also ubiquitin positive (34). The gad mouse, which carries a mutation in the Uch-L1 gene, suffers from axonal dystrophy (35), but no description of its glycemic status has been reported. Furthermore, a few cases of familial Parkinson’s disease are due to mutations in the Uch-L1 gene (36).

The results obtained with Uch-L1 prompted us to investigate other ubiquitin-proteasome pathway genes in islets. Although none of these genes were previously reported to be expressed in islets, some of them are quite ubiquitous (c-Cbl and β-TRCP), while others are more tissue specific (parkin is mostly expressed in certain brain areas) (37). The polyubiquitin genes are rapidly activated in stress situations. In our experiments, their expression levels did not change in cultured islets, a situation reported not to activate antioxidant genes (38), but they were upregulated in islets from 90% pancreatectomized animals, a hyperglycemic situation associated with an islet stress response (19,39).

Similarly to Uch-L1, parkin was initially discovered as a gene linked to some forms of Parkinson’s disease (40); it is an E3 ubiquitin ligase (41) mostly expressed in the nervous system. E6-AP is also an E3 ubiquitin ligase that is mutated in another neurological disorder, the Angleman syndrome (42). The ubiquitin-proteasome pathway is crucial in regulating the activation of the transcription factor complex nuclear factor-κB (NF-κB). The ubiquitinating enzymes UbcH5 and β-TRCP are directly involved in NF-κB activation (6,43,44). NF-κB is activated in pancreatic islets during chronic hyperglycemia after 90% pancreatectomy (19). We observed no change of UbcH5 and β-TRCP levels in 90% pancreatectomy; the relationship between this finding and the NF-κB activation is unclear.

Thus, we find that four genes (Uch-L1, parkin, E6-AP, and UbcH5) of the ubiquitin-proteasome pathway, with regulatory functions of different cellular processes, are expressed in islets and downregulated after high-glucose exposure, suggesting a role of this pathway in islet physiology and more specifically in the hyperglycemia-associated stress response.

The c-Cbl protein (E3 ligase) (45) promotes ubiquitination and internalization of activated receptor tyrosine kinases, thus terminating signaling (46). We found a modest increase in the expression of c-Cbl in islets exposed to high glucose, indicating that it could influence insulin signaling pathways in β-cells.

Apart from our gene expression studies, we also investigated the effect of proteasome inhibitors on glucose-stimulated insulin secretion. In short-term experiments (2 h), the proteasome inhibitor lactacystin enhanced glucose-stimulated insulin secretion. The opposite effect was obtained in long-term experiments (24 h), which we attribute to the toxicity of the agent, as the proteasome is implicated in a huge variety of processes.

Some recent studies using several protease inhibitors report a similar enhancement of insulin release from islets in short-term experiments (14) but a suppression of insulin release after 48-h exposure of islets to inhibitors (15). The authors concluded that calpains are responsible for these effects, playing a role in insulin secretion and action and participating in the pathophysiology of type 2 diabetes. Lactacystin is one of the most specific proteasome inhibitors (although it has also some cathepsin activity) (7). Therefore, we suggest that the increased insulin secretion observed in our experiments is related to the proteasome inactivation. Kalbe et al. (47) also observed an increase in insulin secretion from fetal islets exposed to lactacystin. However, another recent report (48) demonstrates that lactacystin decreases insulin release from islets exposed to high glucose without affecting the total insulin content. The same work shows that lactacystin decreases insulin synthesis and favors its degradation. Despite the opposite results regarding insulin release at high glucose found by these authors versus ours, all these data clearly point toward a role of the proteasome pathway in the regulation of insulin turnover and release by islets.

In Alzheimer’s disease, an increase in the maturation and secretion of the β-amyloid precursor was found in cells exposed to lactacystin, suggesting that the proteasome pathway is responsible for maintaining levels of presenilins 1 and 2, which are involved in the maturation of the β-amyloid precursor protein (49). Interestingly, the ubiquitin-proteasome pathway also has a regulatory role in the endocytic pathway and in vesicle biogenesis and trafficking (4,50).

In an attempt to determine which element of the insulin secretion was affected by lactacystin, we used a PKC inhibitor, a Ca²⁺ chelator, and the secretagogue arginine. PKC modulates the glucose-stimulated insulin secretion by islets (51). Interestingly, activators of PKC trigger its ubiquitination and degradation by the proteasome, thus serving as a downregulation mechanism (52). In spite of these potential relationships, our results indicate that the lactacystin enhancement of insulin release is not directly related to PKC. Moreover, the lactacystin effect is Ca²⁺ dependent, as it was abolished in the presence of EGTA, and is linked to glucose stimulation but not arginine stimulation of secretion.

This work presents novel evidence for a potentially important role of the ubiquitin-proteasome pathway in islets. We find that blockade of the proteasome degradation machinery enhances glucose-stimulated insulin secretion. As this and other studies show, the proteasome seems to participate in insulin secretion and turnover. The effect of glucose on the expression of certain genes of this pathway raises questions of their involvement in the β-cell adaptation and decompensation found in different stages of diabetes. We hope these results lead to future studies aimed to elucidate the regulatory function of the ubiquitin-proteasome pathway in pancreatic islet cells.

This study was supported by grants from the National Institutes of Health (DK35449 to G.C.W.), the Juvenile Diabetes Research Foundation, the Diabetes Research and Wellness Foundation, and an important group of private donors. Help was also provided by the Islet Core of the Juvenile Diabetes Research Foundation Center for Islet Transplantation at Harvard Medical School and the Joslin Diabetes and Endocrinology Research Center, supported by the National Institutes of Health (P30 DK36836-16). M.D.L.-A. was a recipient of a Mentor-Based Fellowship from The American Diabetes Association. V.F.D.-K. was supported by a research fellowship from the Juvenile Diabetes Research Foundation.