Identification of Insulin Signaling Elements in Human β-Cells: Autocrine Regulation of Insulin Gene Expression

Fetal bovine serum, l-glutamine, penicillin/streptomycin, mouse monoclonal anti–α-tubulin antibody, and trypsin-EDTA were obtained from Sigma Chemical. Lipofectamine 2000, CMRL-1066 medium, and Dulbecco’s minimal essential medium were obtained from Invitrogen. Rabbit polyclonal antibodies directed against the α-subunit of the IR and β-subunit of the IGF-1 receptor (IGF-1R) were from Autogen Bioclear. Horseradish peroxidase–coupled IgG was from Pierce. RiboGreen I and SYBR Green I fluorescent dyes were from Molecular Probes and Roche Diagnostics, respectively. TaqDNA polymerase, RNase A, rRNasin, and Moloney murine leukemia virus reverse transcriptase were obtained from Promega. The siRNA duplexes designed to knock down IR mRNA and protein expression levels were obtained from Dharmacon (Smart-Pool Technology). Primers used in this study were from Operon Biotechnology and are listed in Table 1.

Human islets were obtained from the King’s College Hospital Islet Transplantation Unit (King’s College Hospital, London, U.K.). Briefly, pancreata were removed (with appropriate ethical approval) from nondiabetic cadaver organ donors and islets were isolated under aseptic conditions as described (13).

After isolation, human islets were maintained (5% CO₂) in CMRL-1066 medium (5.5 mmol/l glucose) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 0.1 mg/ml streptomycin for up to 72 h before experimental use. They were then maintained for another 24 h in 2.5 mmol/l glucose-containing Dulbecco’s minimal essential medium for insulin gene expression experiments or in CMRL-1066 medium (5.5 mmol/l glucose) without serum for the transfection work. For measurements of insulin gene expression, after the initial 24-h incubation at 2.5 mmol/l glucose, islets were exposed for another 48 h to 2.5–25 mmol/l glucose in the absence or presence of 100 nmol/l exogenous insulin, and preproinsulin (PPI) mRNA levels were assessed by quantitative RT-PCR. When the role of the IR in the regulation of glucose-induced insulin gene expression was studied, islets were transfected either with four small siRNA duplexes that were designed to specifically knock down IR protein expression or with a nonsilencing RNA (control), preincubated for 24 h at 2.5 mmol/l glucose, and then treated with 2.5–12 mmol/l glucose for another 48 h. The characterization of IR and IGF-1R expression levels (mRNA and protein) in the siRNA-transfected group of islets was performed by Western blotting with protein extracts that were prepared just after transfection (control group) and 36–84 h after transfection.

Between 48 and 72 h after islet isolation, 10–20 handpicked islet equivalents were dispersed into cell suspensions by trypsinization and RT-PCR amplification of mRNAs isolated from single human islet cells was performed essentially as previously described (14). The first round of PCR amplification was performed for 45 cycles using 3–5 μl of the RT reaction by the addition of 40 μl of PCR mixture containing a final concentration of 0.5 μmol/l of the specific outer primers, 0.15 μmol/l of each dNTP, and 2.5 units of TaqDNA polymerase. Then, 2 μl of the first-round PCR product was added to 28 μl of a second PCR mixture containing 0.75 μmol/l of the specific inner primers, 0.15 μmol/l of each dNTP, and 1.5 units of TaqDNA polymerase. The second PCR amplification was performed for 38–45 cycles, depending on the gene being studied. The final PCR products were separated on 1.8% agarose gels. Inner and outer primer sequences, MgCl₂ concentrations, and the corresponding annealing temperatures used for single-cell RT-PCR are shown in Table 1.

Groups of ∼300 (for quantitative PCR [qPCR]) or ∼1,000 (for Western blot immunodetection) islet equivalents were handpicked under a dissecting microscope and transferred to 1% (wt/vol) gelatin-coated wells of a 24-well plate containing 400 μl of antibiotic-free CMRL-1066 medium (5.5 mmol/l glucose) before transfection with Lipofectamine 2000. The transfection mixture contained 500 nmol/l of four different siRNA duplexes that were designed to specifically knock down IR protein expression. Transfection was performed for 6 h using a rocker in a humidified atmosphere (37°C, 5% CO₂) after the addition of 100 μl of the transfection mixture to each well such that the final concentration of each siRNA was 100 nmol/l per well. Human islets that had been transfected with a fluorescein-labeled but nonsilencing siRNA duplex were used as controls and to evaluate the efficiency of transfection.

Total RNA was isolated from groups of ∼300 human islet equivalents according to the manufacturer’s instructions (Qiagen animal cells protocol) and quantified by fluorescence using RiboGreen fluorescent dye. Immediately after RNA isolation, cDNAs were synthesized and qPCR amplifications were performed using synthetic oligonucleotide primers by monitoring in real time the increase in fluorescence of SYBR Green I dye using a LightCycler rapid thermal cycler system (Roche) as described previously (15). Each sample value was normalized to actin copy numbers, which were determined in a similar manner. In all qPCR experiments, the presence of possible contaminants was checked by control reactions in which amplification was performed in reaction mixtures without cDNA templates. The specificity of each primer pair was confirmed by melting curve analysis and agarose gel electrophoresis of PCR products. Primer sequences, MgCl₂ concentrations, and the corresponding annealing temperatures used for qPCR are given in Table 1 and are marked with an asterisk.

Total protein (50 μg) from transfected islets was separated on 10% polyacrylamide gels and transferred to a nitrocellulose membrane that was incubated for 16 h with a polyclonal anti-IRα antibody (1:1,500 dilution), a polyclonal anti–IGF-1Rβ antibody (1:500 dilution), or a monoclonal anti–α-tubulin antibody (1:2,000 dilution) and for 1 h with a horseradish peroxidase–coupled anti-rabbit or anti-mouse IgG (1:5,000 dilution). Secondary antibodies were revealed by chemiluminescence (GE Healthcare).

Statistical comparisons were performed using ANOVA or one-tailed Student’s t tests, as appropriate. Differences were considered significant at P < 0.05.

Two IR mRNA transcripts, IR_A and IR_B, resulting from alternative splicing of exon 11 in the receptor gene exist naturally (16–18) and have been proposed to play distinct roles in β-cells (19). The expression of both isoforms by human islets was therefore assessed using exon 11–flanking primers to generate two products of 333 bp (IR_B) and 297 bp (IR_A) by PCR amplification. As shown in Fig. 1, both subtypes of the IR were expressed in human islets with IR_B (exon 11⁺) mRNA expression being more abundant. PCR amplifications using primers for IRS-1 and IRS-2, which have been demonstrated to be key mediators in the insulin signaling cascade, produced two single products of 324 and 311 bp, respectively. All members of class I and II PI3Ks were also found to be expressed in human islets but with different expression levels. Whereas the mRNAs encoding the catalytic subunits of p110α, p110β, and PI3KC2α were readily detectable, mRNA levels coding for the PI3K catalytic domains p110γ, p110δ, PI3KC2β, and PI3KC2γ were low and barely detectable (Fig. 1). Because PDK-1 exists as two different variants, its expression in human islets was therefore determined using a single pair of primers that amplify a consensus sequence of both variants. As shown in Fig. 1, PDK-1 mRNA was readily detectable. PKB exists as three different isoforms (termed PKBα, PKBβ, and PKBγ) that possess >85% sequence identity and are widely expressed in human tissues. Messenger RNAs for all of these PKB isoforms were also detectable in human islets (Fig. 1). No products were formed when mRNA rather than cDNA was used as a template for PCR amplifications (Fig. 1, lower panel).

To determine whether the two IR subtypes are expressed in human islet endocrine cells and whether there is a specific expression of IR_A and/or IR_B in α-, β-, and δ-cells, we further characterized the expression of IR in individual human islet cells by single-cell RT-PCR. Expression of preproglucagon (PPG), PPI, and preprosomatostatin (PPS) was determined in single islet cells after a two-round PCR (Fig. 2A) using nested primers, and then the remaining cDNA was used in a second PCR to amplify one or both cDNAs encoding for IR_A and IR_B. Human islet cells that did not express mRNAs for PPG, PPI, or PPS were designated nonendocrine cells. As shown in Fig. 2B, whereas both IR_A and IR_B mRNAs were expressed in β- and δ-cells, only the mRNA coding for IR_A was amplified in α-cells. In all, IR_A mRNA was amplified from 3 of 12 α-cells (25%), 2 of 11 β-cells (18%), and 2 of 9 δ-cells (22%). In contrast, IR_B mRNA was amplified from 0 of 12 α-cells (0%), 7 of 11 β-cells (63%), and 1 of 9 δ-cells (11%).

Because we detected IR expression in human β-cells, we further determined whether the multiple insulin signal transduction cascade isoforms that were expressed in the whole islets (Fig. 1) were also present and might therefore participate in the insulin signaling cascade in human β-cells. As shown in Fig. 3, mRNAs coding for IRS-1, IRS-2, p110α, p110β, PI3KC2α, PI3KC2γ, PDK-1, PKBα, PKBβ, and PKBγ were identified in human β-cells. In all, 50% of the tested β-cells expressed mRNAs for IRS-1 (6 of 12), 33% for IRS-2 (3 for 9), 28% for p110α (2 of 7), 50% for p110β (3 of 6), 78% for PI3KC2α (7 of 9), 75% for PI3KC2γ (3 of 4), 25% for PDK-1 (2 of 8), 66% for PKBα (4 of 6), 100% for PKBβ (5 of 5), and 100% for PKBγ (5 of 5). In contrast, mRNAs encoding p110γ (0 of 6), p110δ (0 of 6), and PI3KC2β (0 of 5) were not amplified in human β-cells.

As expected, increasing the glucose concentration from 2.5 to 25 mmol/l (48 h) significantly stimulated human islet insulin gene expression (Fig. 4A). However, a shorter exposure (30 min to 24 h) to 25 mmol/l glucose did not result in a significant increase in PPI mRNA levels. The effect of glucose on insulin gene expression was also concentration dependent, with a maximal effect observed at 12 mmol/l glucose (Fig. 4B). Indeed, although insulin gene expression was stimulated ∼1.6- and 1.9-fold when the glucose concentration was increased from 2.5 to 8 and 12 mmol/l, respectively, further increases to 17 and 25 mmol/l glucose did not further stimulate PPI mRNA levels. In addition to stimulating insulin gene expression, exposure of islets to 12 mmol/l glucose for 48 h after preincubation at 2.5 mmol/l glucose for 24 h also caused a significant increase in insulin secretion (2.5 mmol/l glucose: 430.0 ± 28.3 ng/ml; 12 mmol/l glucose: 1085.0 ± 219.2 ng/ml).

To determine whether an autocrine effect of insulin might be involved in glucose-stimulated insulin gene expression in human islets, we assessed the effect of 100 nmol/l exogenous insulin at various glucose concentrations. As shown in Fig. 5, insulin gene expression was significantly stimulated by the addition of exogenous insulin at 2.5, 8, and 12 mmol/l glucose after 48 h. This stimulatory effect of insulin was additive to the glucose stimulatory effect: even when the maximal stimulatory concentration of glucose (12 mmol/l) was used in this study (Fig. 4B), addition of 100 nmol/l exogenous insulin still significantly increased PPI mRNA levels.

To determine whether the insulin signaling cascade plays a key role in the glucose stimulation of insulin gene expression, groups of ∼300–1000 islet equivalents were transfected with siRNA duplexes designed to specifically knock down expression of both subtypes of the insulin receptor, and glucose-induced insulin gene expression was analyzed at 2.5, 5.5, and 12 mmol/l glucose. The efficiency of transfection was ∼58% and, as shown in Fig. 6A, islets maintained in culture with these specific IR siRNAs showed a rapid depletion in IR mRNA levels such that they were decreased by ∼40% 36 h after transfection. This level of depletion was sustained for 60 h, and levels remained reduced by ∼15% 84 h after transfection. IR protein expression levels were also reduced as shown in Fig. 6B, and a significant reduction in IR expression lasted for at least 72 h. In contrast, neither the mRNA (Fig. 6A) nor the protein (Fig. 6B) expression levels of IGF-1R, which belongs to the same family and shares a 63.8% sequence identity with IR, were significantly affected in the siRNA-treated islet group, showing specificity of the siRNA duplexes for IR. Exposure of nonsilencing RNA-transfected islets to 5.5 and 12 mmol/l glucose resulted in a significant concentration-dependent stimulation of PPI mRNA levels compared with 2.5 mmol/l glucose (Fig. 7). However, the stimulatory effect of glucose on insulin gene expression was completely suppressed when IR expression was reduced in siRNA-transfected islets.

Activation of the insulin signaling pathway involves the successive stimulation of the IR, IRS proteins, PI3K, PDK-1, and PKB in a cellular cascade that is well established and accepted. However, this sequence of events is deceptively simple because it does not take into account the relative participation of multiple isoforms of elements of the cascade, and two isoforms of the same family may have overlapping but also distinct functions. The best example comes from knockout mice for either IRS-1 or IRS-2, which display entirely different features such as an increased β-cell mass when IRS-1 is depleted but a decreased β-cell mass when IRS-2 expression is suppressed (9). Thus, to better define the insulin signaling cascade in human β-cells, we first demonstrated by RT-PCR the expression of the key docking protein and kinase isoforms in human islets, then we profiled the gene expression of these isoforms in single isolated β-cells.

The insulin receptor exists as two mRNA transcripts that are expressed in a highly regulated tissue-specific fashion (20), but their expression, abundance, and function(s) in human β-cells are unknown. We have now used exon 11–flanking primers to demonstrate that both subtypes of the receptor are present in human islets, with the Β subtype (exon 11⁺) being predominantly expressed, consistent with a previous report (21). Interestingly, our single-cell PCR amplification of the two transcripts showed for the first time the existence of a high heterogeneity within the islets. Whereas only the Α subtype of the receptor (exon 11⁻) was detected in glucagon-secreting cells, both the Α and Β subtypes were expressed in insulin- and somatostatin-producing cells. Heterogeneity in expression of the insulin receptor within the islets could also be observed at the level of the frequency of expression of both subtypes: 18–25% of the three cell types expressed IR_A mRNA, suggesting that this subtype of the receptor might mediate similar signaling cascades in α-, β-, and δ-cells, whereas the IR_B mRNA was detected in only 11% of δ-cells but in 63% of β-cells, suggesting that the insulin signaling cascades activated through IR_B stimulation may play key roles in regulating the β-cell function.

To determine whether the whole insulin signaling cascade is expressed in human β-cells and, more specifically, which isoforms might be involved in transducing the signal triggered by insulin, we profiled the gene expression of the IRS proteins, the catalytic subunits of the PI3K enzymes, PDK-1, and the PKBs in single human β-cells. Our results showed that the mRNAs coding for IRS-1 and IRS-2, which are proposed to be the major insulin-stimulated tyrosine phosphoproteins (22), were expressed in only 50 and 33% of the β-cell population. In a similar manner, the mRNA coding for PDK-1 was amplified from only 33% of the tested β-cells. In contrast with these relatively low levels of expression, the frequency of expression of PKBα, PKBβ, and PKBγ, which represent the end point of the insulin signaling cascade and mediate the majority of the effects triggered by insulin, was high. Indeed, 66% of the tested β-cells were found to express PKBα and all β-cells tested expressed PKBβ and PKBγ. Interestingly, although it is generally accepted that members of the first class of the PI3K enzymes mediate the majority of the insulin effects in their target tissues, the mRNAs coding for PI3KC2α and PI3KC2γ, two members of the class II PI3K family, were expressed with the highest frequency in the β-cell population (≥75%). In contrast, only 28 and 50% of the tested human β-cells expressed the PI3K class I catalytic subunits p110α and p110β, and none of them expressed p110γ and p110δ. Thus, in accordance with the finding that the α isoform of the class II PI3K family is activated by the IR (23,24), our results suggest that the α and also the γ isoforms of the class II PI3Ks might play important roles in the regulation by insulin of β-cell function. The discovery that the catalytic subunit p110δ of the class I PI3Ks is not expressed in human β-cells is consistent with the report that this isoform is mainly expressed in leukocytes (25). Finally, the fact that some mRNAs (p110γ, p110δ, and PI3KC2β) can be amplified from human islets but not from single β-cells demonstrates the fact that detecting the expression of a specific mRNA within whole islets does not prove that it is expressed in β-cells and therefore might have no direct involvement in the regulation of the β-cell function and further emphasizes the importance of β-cell–specific strategies, as described here.

Despite an increasing body of evidence that the pleiotropic effects of glucose on β-cell function may involve an autocrine activation of the β-cell insulin signaling cascade, the importance of this regulatory loop still remains controversial, and no consensus has yet been achieved. The second part of this study was therefore designed to determine whether insulin secreted in response to glucose regulates, at least in part, the expression of its gene in isolated human islets.

As expected, PPI mRNA levels were significantly increased in a concentration-dependent manner when human islets were exposed to stimulatory concentrations of glucose. However, although it had been previously documented that short-term exposure (15 min) of rat islets to high glucose levels stimulates PPI promoter activity and gene expression (26–28), we did not observe any glucose-stimulated increase in PPI mRNA levels before 48 h exposure in our study. Consistent with our findings, it has been reported that glucose-stimulated PPI gene expression was required for adequate insulin production during chronic, but not acute, exposure to high glucose levels (29), and high glucose concentrations have been demonstrated to initially stabilize preexisting PPI mRNA, and no real increase in its level can be detected over 24 h (30).

The possible autocrine effect of insulin in glucose-induced insulin gene expression was first proposed by two different groups that both used transfected insulin-secreting cell lines (31,32). However, it has been suggested that the use of artificially transfected systems that overexpress insulin signaling elements may result in overestimation of the physiological role of the insulin signaling cascade in insulin production (33). Therefore, in the current study, the autocrine effect of insulin in the regulation of its gene expression was first studied in normal human islets and then in siRNA-treated human islets that showed lower IR expression levels. In addition to the stimulatory effect of glucose on insulin gene expression, exogenous insulin caused a twofold increase in PPI mRNA at 2.5 mmol/l glucose. Interestingly, insulin further stimulated insulin gene expression even when a maximal stimulatory glucose concentration was used (12 mmol/l), suggesting that the effect of insulin was additive to that of glucose. Thus, although these results do not prove that glucose-stimulated insulin gene expression is mediated through an autocrine pathway, they suggest an autocrine role for insulin and the insulin signaling cascade in its regulation.

The hypothesis that insulin plays a key role in the regulation of glucose-stimulated insulin gene expression is supported by our data showing that siRNA-induced downregulation of the insulin receptor in human islets was associated with a complete loss of the stimulatory effects of 5.5 and 12 mmol/l glucose. Decreased IR expression did not modify basal PPI mRNA levels at 2.5 mmol/l glucose, which is also consistent with an autocrine role for insulin in glucose-stimulated PPI gene expression. In addition, because newly synthesized mRNA encoding PPI is a requirement for β-cells to produce and secrete sufficient amounts of insulin during chronic exposure to high glucose levels (29), our observations that PPI mRNA levels were decreased in a glucose concentration–dependent manner in siRNA-transfected islets are not surprising. These results demonstrate rather that reduction in IR expression completely abolished the stimulatory effect of glucose on insulin gene expression, resulting in no replacement of the degraded PPI mRNAs. Interestingly, whereas the maximal reductions in the insulin receptor mRNA and protein levels were ∼50–55%, the stimulatory effect of 12 mmol/l glucose on insulin gene expression was completely suppressed. One possible explanation for this result is that insulin regulates the expression of its gene through binding and activation of IR_A as previously suggested (19,21). Thus, as we observed that only 18% of the β-cell population express IR_A, in contrast to the 63% that express IR_B, a 50% decrease in the expression of the insulin receptor would be expected to have a stronger impact on the signaling pathway mediated by IR_A, in this case, regulation of the insulin gene expression.

In conclusion, this study provides, for the first time, direct evidence for expression of the insulin signaling pathway elements in human β-cells, and, because of a high frequency of expression, our findings suggest that IR_B, PI3KC2α, and PI3KC2γ may play important roles in the regulation of human β-cell function. In addition, our results provide strong evidence that the insulin signaling cascade regulates the effect of glucose on the expression of the insulin gene in human β-cells. Taken together, our data suggest that the well-documented insulin resistance in type 2 diabetes may also adversely affect pancreatic β-cells, resulting in decreased insulin biosynthesis and exacerbation of the metabolic dysfunction.

We gratefully acknowledge the Eli Lilly International Foundation, the European Foundation for the Study of Diabetes, and the Juvenile Diabetes Research Foundation for grant support.