Human pancreatic islets were kindly provided by Prof. Olle Korsgren (Dept. of Radiology, Oncology and Clinical Immunology, Uppsala University Hospital, Uppsala, Sweden) through the Nordic Network for Clinical Islet Transplantation. Only organ donors who had agreed to donate for scientific purposes were included. Informed written consent to donate organs for medical and research purposes was obtained from donors, or relatives of donors, by the National Board of Health and Welfare (Socialstyrelsen), Sweden. Permission to obtain pancreatic islet tissue from the Nordic Network for Clinical Islet Transplantation was reviewed and approved by the local ethics committee (Regionala etikprövningsnämnden, Uppsala) in Uppsala, Sweden. The human pancreatic islets were isolated from the pancreas of brain-dead organ donors using collagenase digestion and Biocoll gradient centrifugation ^[31].

After isolation, the islets were cultured free-floating in Sterilin dishes in CMRL 1066 medium containing 5.6 mM glucose, 10% fetal calf serum (FCS), and 2 mM L-glutamine for 1–5 days.

The islet beta-cell percentages determined using Newport Green staining followed by fluorescence microscopy were routinely 30–60%. To evaluate islet functional quality, batches of twenty islets were perfused with Krebs-Ringer bicarbonate HEPES buffer (KRBH) containing 2 mg/ml human albumin and 1.67 mmol/l glucose at a flow rate of 0.3 ml/min. The glucose concentration was increased to 16.7 mM after a 30-min period. Fractions for insulin measurement were collected every 4 min. Only islet preparations that responded to a high glucose challenge with a more than doubled insulin release were used for further experimentation.

Following the initial culture period, islets were cultured for an additional 6–24 hours in CMRL 1066 containing antibiotics, 2 mM glutamine and one of the following supplements: 0.5 mM sodium palmitate solubilized in 0.5% (weight/volume) fatty acid and lipopolysaccaride free bovine serum albumin (BSA) (Sigma); recombinant human Interleukin 1beta (IL-1ß (50 units/ml) (PeproTech, London, U.K.) and Interferon-gamma (IFN-γ) (1,000 units/ml) (PeproTech); 100 mM hydrogen peroxide; 2 mM DETA/NO (Cayman Chemicals) or 10 mM streptozotocin (STZ) (Sigma). To some of the groups 10 µM of Imatinib was added at different time points prior to the addition of test substances given above. To controls equal amounts of vehicle (DMSO) were supplemented.

Total RNA was isolated from human pancreatic islet using the Ultraspec™ RNA Isolation System reagent (Biotecx Laboratories, Houston, TX, USA). Two µg of RNA was used for cDNA synthesis. cDNA was synthesized using the M-MuLV reverse transcriptase (Finnzymes, Espoo, Finland) and oligo-dT primers according to the manufacturer's instructions.

PCR amplification was performed using the Lightcycler instrument (Roche Diagnostics, Lewes, UK) and the Lightcycler FastStart DNA Master SYBR Green I kit (Roche Diagnostics). The following primers were used: GAPDH (human) (glyceraldehyde-3-phosphate dehydrogenase): [gene: 5′-ACCACAGTCCATGCCATCAC] (forward) [gene: 5′-TCCACCACCCTGTTGCTGTA] (reverse), IL-8 (human) [gene: 5′-GGCAACCCTACAACAGACCCACA] (forward), [gene: TGCTAACAGGCTTAGGACTTGC] (reverse). Semi-quantitative data for IL-8 expression were calculated using the formula: 2^{crossing point GAPDH - crossing point IL-8}.

Islets in groups of 50 were washed once with PBS and then lysed with SDS sample buffer; 2% SDS, 5% β-mercaptoethanol, 100 mM Tris-HCL pH 6.8, 10% glycerol and bromophenolblue. The protein samples were boiled for 5 min before separation on a 9% SDS-PAGE. The separated proteins were electrophoretically transferred to Hybond™-P membranes (Amersham Bioscience), which were then pre-blocked in 5% BSA. The membranes were hybridized with monoclonal mouse or polyclonal rabbit antibodies directed against IκB-α (C-21, Santa Cruz), Phospho-Ser32- IκB-α (Cell Signaling), Acetylated-Lys310-p65 (Abcam), Phospho-Ser276-p65 (Santa Cruz), Phospho-Ser536-p65 (Cell Signaling), p65 (F-6, Santa Cruz), Methyl-Lys221-p65 ^[32] or Phospho-Tyr701-STAT1 (Cell Signaling) over night at 4°C. The horseradish peroxidase conjugated anti-mouse or anti-rabbit antibodies were used as secondary antibodies and luminescence detected with the ECL system (Amersham Bioscience). The resulting bands were then digitalized using the Kodak 4000MM system and the optical densities were measured using the Kodak Molecular Imaging Software.

INS-1 832/13 cells were treated with Imatinib (10 µM) for 1 or 24 hours and IL-1ß (50 U/ml)+IFN-γ (1000 U/ml) for 30 min, after which nuclear proteins were extracted for EMSA, which was performed as previously described ^[32]. The κB-binding activity of the nuclear extracts was determined using the LightShift Chemiluminescent EMSA Kit (Pierce Biotechnology) according to the instructions of the supplier. As labeled probe, a double-stranded biotin-labeled oligonucleotide with the sequence [gene: 5′-AGTTGAGGGGACTTTCCCAGGC] was used. The resulting bands were digitalized using the Kodak 4000MM system and the optical densities were measured using the Kodak Molecular Imaging Software.

Human islets in groups of 1000 were either treated with DMSO or with 10 µM Imatinib for 4 hours. Following the treatments, total RNA was isolated using the Ultraspec™ RNA Isolation System reagent (Biotecx Laboratories, Houston, TX, USA). The RNA was then converted to biotin-labelled cRNA using the TrueLabeling-AMP™ Linear RNA Amplification Kit (SABiosciences) and hybridized overnight to the the Oligo GEArray Human Signal Transduction Pathway Finder Microarray filters (SABiosciences). Signals from the filters were detected by luminescence and quantified by densitometry using the Kodak 4000MM system and Kodak Molecular Imaging Software, all according to the to the instructions of the supplier (SABiosciences).

Human islets in groups of 50 were cultured in 0,5 ml CMRL 1066. The islets were pre-cultured for 6 hours with our without 10 µM of Imatinib and then cultured for another 24 hours with either 100 µM hydrogen peroxide, 10 mM streptozotocin, 1 mM DETA/NO or cytokines (50 U/ml IL-1ß+1000 U/ml IFN-γ). Islet culture medium was then recovered and analyzed for interleukin-8 (CXCL8/IL-8), RANTES (CCL5/RANTES), monokine induced by interferon-γ (CXCL9/MIG), monocyte chemoattractant protein-1 (CCL2/MCP-1), and interferon-γ–induced protein-10 (CXCL10/IP-10) levels using the BD Cytometric Bead Array Human Chemokine Kit according to the instructions of the supplier (Becton Dickinson).

We have previously reported that Imatinib promoted IκB-α degradation in human islets ^[25]. To better understand the effects of Imatinib on human islet NF-κB activation, we first performed time course studies of IκB-α phosphorylation, Ser276-p65 phosphorylation, Ser536-p65 phosphorylation, Lys310-p65 acetylation and Lys221-p65 methylation. These are all NF-κB-associated post-translational modifications that have been reported to stimulate NF-κB activity ^[33]–^[36]. We observed that IκB-α phosphorylation was significantly increased at 20 min of Imatinib (10 µM) exposure (Figure 1). At later time points there was a trend to increased IκB-α phosphorylation, but it did not reach statistical significance, probably due to the low number of observations. Ser276-phosphorylated p65 migrated slightly slower (72 kDa) than non-ser276-phosphorylated p65 (68 kDa), and the intensity of the P-ser276-p65 band was weak. This indicates that P-ser276-p65 constitutes a minor proportion of all p65 and this fraction could represent the transcriptionally active form of p65. The phosphorylation of p65 at position Ser276 was increased by Imatinib at 20 min (Figure 1). At 60–360 min, however, the ser276-p65 phosphorylation was no longer increased, indicating that Imatinib-induced ser276-p65 phosphorylation is transient. P-Ser536-p65 immunoblotting signals were strong and P-ser536-p65 migrated to the same position as total p65. This finding is consistent with the view that p65 is constitutively phosphorylated at position ser536 in human islets, and that this post-translational modification has limited regulatory importance. Indeed, Imatinib did not stimulate ser536-p65 phosphorylation; instead there was a trend towards decreased ser536-p65 phosphorylation with increasing exposure time to Imatinib (Figure 1). Methylation of p65 as position lys221 was not affected by Imatinib at 20 min. However, at both 60 and 180 min Imatinib promoted a modest increase in p65 methylation (Figure 1), which was less pronounced than that of cytokines (results not shown). Finally, we probed for effects of Imatinib on p65 Lys310 acetylation. However, we could not detect specific immunreactivity for Ac-lys310-p65 in human islets, neither under conditions of Imatinib exposure or when stimulated by cytokines (results not shown). It is not clear whether the lack of Ac-p65 signal is due to a low sensitivity of the Abcam anti-Ac-p65 antibody, or to a constitutively low p65 acetylation activity in human islets.

To verify that the observed effects of Imatinib on human islet IκB-α and p65 phosphorylation/methylation events result in increased NF-κB target gene transcription, we performed a microarray analysis. Human islets were exposed to vehicle or Imatinib for 4 hours and then analyzed using the Oligo GEArray Human Signal Transduction Pathway Finder Microarray. By this approach we observed that 10 genes were increased more than 30% in response to the Imatinib treatment (Table 1). Out of these 10 genes 7 are typical inflammatory or stress genes (IL-4R, TCF5, DR5, I-TRAF, I-CAM, HSP27 and IL-8), indicating that Imatinib promoted transcription of NF-κB target genes.

Because IκB-α and ser276-p65 phosphorylation were stimulated by Imatinib (Figure 1), we next investigated whether cytokine-induced IκB-α and ser276-p65 phosphorylation was affected by Imatinib treatment. We also studied the effects of sodium palmitate, since it has been reported that this saturated free fatty acid promotes NF-κB activation in human beta-cells ^[37]. We observed that a 20 min exposure to the combination of IL-1ß (50 U/ml) and IFN-γ (1000 U/ml) increased IκB-α phosphorylation (Figure 2). Somewhat surprisingly, sodium palmitate was without effect (Figure 2). Similar to the results of Figure 1, a 6-hour Imatinib pre-exposure promoted an increased IκB-α phosphorylation. Interestingly, cytokine-induced IκB-α phosphorylation in the presence of Imatinib was not further increased as compared to Imatinib only (Figure 2). Using the same experimental setup as in Figure 2, we observed that cytokines also promote phosphorylation of ser276-p65 (Figure 3). Cytokine-induced ser276-p65 phosphorylation was not affected by the 6-hour Imatinib pretreatment (Figure 3). Thus, at the 6-hour time point, cytokine-induced IκB-α and ser276-p65 phosphorylation is not further potentiated by Imatinib.

To clarify whether a prolonged (over-night) Imatinib-exposure period affects cytokine-induced signaling, we pre-exposed human islets to Imatinib for 6 hours and then continued the culture period for another 24 hours with and without both Imatinib and the combination of IL-1ß and IFN-γ. In this situation human islet IκB-α phosphorylation was enhanced in cells exposed to cytokines, but not in cells exposed to both cytokines and Imatinib (Figure 4A). Interestingly, also cytokine-induced STAT1 phosphorylation was weakened by Imatinib (Figure 4B). Thus, a prolonged Imatinib exposure resulted in attenuation of cytokine-induced signaling events.

We also tested whether human islet IκB-α phosphorylation was affected by the beta-cell toxins hydrogen peroxide, which induces oxidative stress, streptozotocin, which induces alkylation of macromolecules and NAD+ depletion, and DETA/NO, which induces nitrosative stress. As we have previously observed that beta-cell death induced by these toxins is counteracted by Imatinib ^[24], ^[25], it could be hypothesized that stressed beta-cells activate NF-κB and that this event is modulated by Imatinib. Although these substances tended to increase IκB-α phosphorylation, there were no significant effects, either induced by themselves or in combination with Imatinib (Figure 4A). Moreover, STAT1 phosphorylation was undetectable in all groups other than the cytokine-treated islets (Figure 4B). The limited access of human islet material precludes a more sensitive and careful analysis of these parameters, something that is probably necessary to successfully detect modest alterations in NF-κB and STAT1 signaling.

To obtain a functional read-out of Imatinib-modulated NF-κB signaling events, we analyzed human islet chemokine release using a flow cytometric bead assay for the chemokines IP10, IL-8, MCP-1, RANTES and MIG. The experimental setup was the same as for Figure 4, i.e. human islets were pre-exposed to Imatinib for 6 hours and then cultured for another 24 hours with or without Imatinib and hydrogen peroxide, streptozotocin and DETA/NO (Figure 5), or with or without Imatinib and cytokines (Figure 6). We observed that islet release of the chemokine IL-8 to the culture medium was increased by Imatinib (Figure 5A). This observation corroborates the Imatinib-induced expression of IL-8 mRNA assessed by microarray analysis (Table 1). The beta-cell cytotoxic agents hydrogen peroxide, streptozotocin and DETA/NO did not increase human islet release of IL-8 (Figure 5A), which is in agreement with our observation that IκB-α and STAT1 phosphorylation were not affected by these toxins (Figure 4). Imatinib increased the release of IL-8 also in islet cells exposed to hydrogen peroxide and DETA/NO (Figure 5A), supporting the notion that Imatinib enhances basal NF-κB activity and chemokine release. Due to large error bars, there were no significant effects of Imatinib and the cytotoxic agents on IP10 (Figure 5B) and MCP-1 (Figure 5C) release. Nevertheless, there was a trend to higher IP10 and MCP-1 release in response to Imatinib, also supporting an NF-κB stimulatory effect of Imatinib. The levels of MIG and RANTES were barely detectable and we could not observe any effects on the release of these chemokines (results not shown).

The release of IL-8, IP10, MIG and MCP-1, but not of RANTES, was dramatically (100–1000-fold) increased by the cytokines IL-1ß and IFN-γ (Figure 6). Interestingly, Imatinib decreased the cytokine-induced release of IP10 and MIG, and tended to decrease MCP-1 and IL-8 release (Figure 6). Thus, a prolonged Imatinib exposure decreased both cytokine-induced transcription factor phosphorylation (Figure 4) and chemokine release. To validate the results obtained from the microarray and the chemokine release kit, we analyzed IL-8 mRNA levels in human islets using real time PCR. Imatinib exposure for 30 hours resulted in a 8-fold induction of IL-8 mRNA levels when compared to control. (Figure 6 B). Cytokine (IL-1ß and IFN-γ) treatment for 30 hours led to a strong induction (over a 35-fold) of IL-8 mRNA levels. (Figure 6 B). However when pretreated with Imatinib, the cytokine-induced IL-8 mRNA expression was significantly inhibited (Figure 6 B). The result from real time PCR are in concert with the chemokine release and microarray data, further strengthening the notion that a prolonged Imatinib exposure dampens the cytokine-induced chemokine release in human islets.

Human islets consist of 30–60% beta-cells. However they also contain other endocrine cells such as alpha- and delta-cells. To confirm that the observed effects of Imatinib on NF-κB activation in human islets are specific for beta-cells we performed electromobility shift assay (EMSA) on nuclear extracts from the INS-1 832/13 beta-cell line. Cytokine (IL-1β and IFN-γ) exposure for 20 min resulted in a significant increase in κB-binding activity in INS-1 832/13 nuclear extracts (Figure 7). A short time Imatinib exposure (1 hour) tended to increase the κB-binding activity (p = 0.06) whereas a long exposure (24 hours) of INS-1 cells to Imatinib resulted in a significant decrease of κB-binding activity (Figure 7). These results strengthen the idea that a short-term exposure to Imatinib results in a modest activation of NF-κB, whereas a prolonged Imatinib exposure dampens NF-κB activation in beta-cells.