The FASEB Journal • Research Communication
Amelioration of diabetes by imatinib mesylate (Gleevec威): role of -cell NF-B activation and anti-apoptotic preconditioning Robert Ha¨gerkvist, Stellan Sandler, Dariush Mokhtari, and Nils Welsh Department of Medical Cell Biology, Uppsala University, Biomedicum, Uppsala, Sweden It was recently reported that tyrosine kinase inhibitor imatinib mesylate (Gleevec威) improves Type 2 diabetes, possibly by decreasing insulin resistance. However, as both Type 2 and Type 1 diabetes are characterized by -cell dysfunction and death, we investigated whether imatinib counteracts diabetes by maintaining -cell function. We observed that imatinib counteracted diabetes in two animal models, the streptozotocin-injected mouse and the nonobese diabetes mouse, and that this was paralleled by a partial preservation of the -cell mass. In addition, imatinib decreased the death of human -cells in vitro when exposed to NO, cytokines, and streptozotocin. The imatinib effect was mimicked by siRNA-mediated knockdown of c-Abl mRNA. Imatinib enhanced -cell survival by promoting a state similar to ischemic preconditioning, as evidenced by NF-B activation, increased NO and reactive oxygen species production, and depolarization of the inner mitochondrial membrane. Imatinib did not suppress islet cell death in the presence of an NF-B inhibitor, suggesting that NF-B activation is a necessary step in the antiapoptotic action of imatinib. We conclude that imatinib mediates -cell survival and that this could contribute to the beneficial effects observed in diabetes.—Ha¨gerkvist, R., Sandler, S., Mokhtari, D., Welsh, N. Amelioration of diabetes by imatinib mesylate (Gleevec威): role of -cell NF-B activation and anti-apoptotic preconditioning. FASEB J. 21, 618 – 628 (2007)
ABSTRACT
Key Words: pancreatic islet 䡠 c-Abl 䡠 NO Imatinib mesylate (also known as Gleevec威 or Glivec) is a selective tyrosine kinase inhibitor that specifically inhibits the cellular Abelson tyrosine kinase (c-Abl), the PDGFR, the transmembrane receptor tyrosine kinase (c-Kit), and the Abl-related gene (Arg) (1, 2). In the clinic, imatinib is successfully used to treat chronic myeloid leukemia, a disease caused by the Bcr-Abl or v-Abl oncogenes, and gastrointestinal stromal tumors caused by c-Kit mutations (3, 4). It was recently observed that a modest number of patients suffering from both chronic myeloid leukemia and Type 2 diabetes were successfully treated not only for leukemia, but also for diabetes, when given imatinib (5, 6). The molecular 618
mechanisms underlying the beneficial effects of imatinib in these cases are unknown, but may be related to the propensity of imatinib to inhibit the nonreceptor tyrosine kinase c-Abl. In addition to the kinase domain, the c-Abl protein contains domains and motifs that allow interactions with signaling and adaptor proteins, nucleo-cytoplasmic shuttling, as well as DNA binding and actin binding (7). This infers that this protein can sense and integrate information from multiple signaling pathways in different cellular compartments, then interact with downstream effector proteins. The outcome of c-Abl activation can be very diverse depending on initial stimuli, cell type, and cellular location of c-Abl. When cells are exposed to different forms of damage, c-Abl becomes highly activated, which leads to cell cycle arrest and apoptosis (7). Genotoxin-induced apoptosis seems to require nuclear c-Abl, whereas apoptosis in response to oxidative stress and endoplasmatic reticulum (ER) stress is mediated by cytosolic c-Abl (8). c-Abl phosphorylation is known to promote activation of the downstream effectors such as the stress-activated protein kinases (JNK and p38 MAP-kinases) (9), the tumor suppressor p73 (10), and caspase 9 (11). Considering the proapoptotic actions of active c-Abl and that both noninsulin dependent (Type 2) and insulin-dependent (Type 1) diabetes mellitus are characterized by an augmented destruction of insulin-producing -cells (12, 13), we investigated whether two unrelated animal models for diabetes and -cell destruction, the nonobese diabetes (NOD) mouse and the single-dose streptozotocin (STZ) -injected diabetic mouse, are influenced by treatment with imatinib. The NOD mouse is a genetic model for Type 1 diabetes in which -cells are destroyed by islet-infiltrating immune cells (14). In the STZ mouse, STZ specifically kills -cells, which results in overt hyperglycemia (15). We report here that imatinib protected against diabetes and to some extent against the diabetes-associated decrease in -cell mass. Imatinib also protected against proinflammatory cytokine-, NO-, and STZ-induced -cell death in vitro. The antiapoptotic effect exerted by 1
Correspondence: Department of Medical Cell Biology, Uppsala University, Biomedicum, P.O. Box 571, SE-75123 Uppsala, Sweden. E-mail:
[email protected] doi: 10.1096/fj.06-6910com 0892-6638/07/0021-0618 © FASEB
imatinib might result from activation of NF-B, leading to an apoptosis-resistant state resembling that observed in ischemic or chemical preconditioning (16). In view of these findings, we hypothesize that inhibition of c-Abl might prove beneficial in treating diabetes in the future. MATERIALS AND METHODS Cell lines -TC-6 cells at passage numbers 20 –30 were maintained in Dulbecco’s modified Eagle medium (DMEM) ⫹ 10% FCS. INS-1 832/13 cells (kind gift from Dr. Hindrik Mulder, Lund, Sweden) were grown in RPMI 1640 ⫹ 10% FCS.
Isolation of splenocytes and intracellular cytokine staining Splenocytes were suspended in 0.19 M NH4Cl for 10 min to lyse the erythrocytes. Macrophages were removed by incubation of the cells in attachment culture dishes for 1 h. The nonattached splenocytes were either incubated with 1 g/ml brefeldin A or stimulated with a combination of brefeldin A ⫹ 5 ng/ml PMA ⫹ 500 ng/ml ionomycin for 4 h. Splenocytes were then fixed in 4% paraformaldehyde for 5 min and permeabilized in PBS ⫹ 2% FCS ⫹ 0.1% BSA and 0.1% Triton-X for 10 min, followed by incubation with FITCconjugated anti-mouse-IFN-gamma or IL-10 antibodies (Becton Dickinson, Temse, Belgium) for 30 min on ice. The splenocytes were analyzed for FL-1 fluorescence using the FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA, USA). Evaluation of islet viability
Animals, human islets, islet isolation, and tissue culture Male NMRI (Bomholt Gaard, Ry, Denmark), Sprague-Dawley rats (B&K Universal, Sollentuna, Sweden), and a local colony of NOD mice were kept under standard pathogen-free condition, with free access to tap water and pelleted food. Isolated human pancreatic islets were kindly provided by Professor Olle Korsgren at the Department of Radiology, Oncology and Clinical Immunology at Uppsala University Hospital (Uppsala, Sweden). Islets from male NMRI mice and Sprague-Dawley rats were isolated by a collagenase digestion procedure and cultured as described (17). STZ treatment in vivo Male NMRI mice were injected with STZ (Sigma, St. Louis, MO, USA) (160 mg/kg body wt) into the tail vein. The mice received either gavage with 0.9% NaCl or imatinib (Novartis, Basel, Switzerland) (200 mg/kg body wt) in the morning for three consecutive days starting 1 day before STZ injection. Imatinib was administered 2 h before STZ when given the same day. Blood glucose levels were determined by blood samples from the tail tip using ExacTech blood glucose meter (Baxter Travenol, Deerfield, IL, USA). Development of insulitis and diabetes in NOD mice NOD mice were gavaged with either saline or imatinib (100 mg/kg body wt) daily from 3 to 9 wk of age to study insulitis. To study the development of diabetes, female NOD mice were treated as above between 9 and 35 wk of age. Blood glucose levels were measured every other week until the first sign of hyperglycemia, then weekly. Animals were considered diabetic and sacrificed on two consecutive days of blood glucose ⬎ 12 mM. Pancreases were fixed in 10% formalin, dehydrated, and embedded in paraffin. Sections were stained with hematoxylin and eosin or stained for insulin. For insulin staining, an antibovine insulin antibody was used (1:200; ICN Pharmaceuticals, Costa Mesa, CA, USA), followed by a peroxidase-anti peroxidase/diaminobenzidine detection system (DAKO, Glostrup, Denmark, and Kem-En-Tec, Copenhagen, Denmark, respectively). Insulitis was scored on sections from two different pancreatic locations according to an arbitrary scale: A denotes normal islet structure; B denotes mononuclear cell infiltration in the islet peri-insular area; C denotes heavy mononuclear cell infiltration into a majority of islets (i.e., insulitis); D denotes only a few residual islets present, often showing an altered architecture. The examiner of the sections was unaware of the origin of the sections. IMATINIB MESYLATE-MEDIATED PROTECTION AGAINST DIABETES
Islets or islet cells were vital stained with propidium iodide (20 g/ml) and bisbenzimide (Sigma) (5 g/ml) as described (18). Total number of cells as well as necrotic and apoptotic nuclei were counted with NIH Image 1.63. Analysis of medium nitrite content and determination of insulin release A cytokine combination of 25 or 50 U/ml human IL-1 ⫹ 1000 U/ml of murine IFN-␥ ⫹ murine TNF-␣ (1000 U/ml) was used to induce NO formation and nitrite was analyzed after 24 h (17). Islets in groups of 5 were precultured overnight with 10 M imatinib, then incubated for 60 min in Krebs-Ringer bicarbonate buffer (KRBH) containing 1.67 mM glucose and 60 min 16.7 mM glucose. Insulin medium concentration was determined using a high range rat insulin ELISA (Mercodia, Uppsala, Sweden). Extraction of nuclear proteins and electromobility shift assay (EMSA) INS-1 832/13 cells were treated with IL-1 (50 U/ml) for 20 min or 10 M imatinib for 180 min, after which the nuclear proteins were extracted and used for gel shift analysis as described previously (19). Immunoblotting Anti-NOS-2 or ATF-2 antibodies were used for immunoblot analysis of INS-1 832/13 cells and anti-IB␣ or p65 NF-B antibodies for analysis of human islets. All antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Filter-associated antibodies were visualized using enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ, USA) and quantified by densitometry. Measurements of reactive oxygen species (ROS) and mitochondrial membrane potential INS-1 832/13 were treated with 10 M imatinib in culture medium for 6 h and subsequently labeled with 10 M of 5,6-carboxy-2⬘,7⬘-dichloroflourescein-diacetate (DHCF-DA) for 30 min. Cells were trypsinized and the fluorescent form of DHCF-DA, DCF, was measured using a flow cytometer before and every 5 min up to 1 h after the addition of STZ (2 mM). For the analysis of mitochondrial membrane potential, human islets pretreated with imatinib were labeled with 5 M 619
of the fluorescent probe JC-1 (5,5⬘,6,6⬘-tetrachloro-1,1⬘,3,3⬘ tetraethylbenzimidazolylcarbocyanine iodide/chloride; ref. 20) for 30 min at 37°C. Cells were then trypsinized, followed by flow cytometric analysis. The ratio between the 585 and 530 nm signals was calculated to monitor changes in mitochondrial membrane potential. siRNA treatment of mouse islet cells or TC-6 cells TC-6 cells or dispersed islet cells from male NMRI mice were transfected with siRNA directed against c-Abl, c-Kit, and GL3 lucifierase as described previously (21). c-Abl and GL3 luciferase siRNA was kindly provided by Dr. Jason W. Myers at the Department of Biochemistry, Stanford University School of Medicine (Stanford, CA, USA). Murine c-Kit siRNA and primers for real-time polymerase chain reaction (PCR) verification of siRNA-mediated down-regulation were obtained from Sigma. Real-time PCR or RT-polymerase chain reaction (RT-PCR) was performed using the Lightcycler instrument (Roche Diagnostics, Mannheim, Germany) and the SYBR Green Jumpstart Taq Readymix reagent (Sigma). Results were normalized to -actin (21).
RESULTS Imatinib protects against development of diabetes in NOD mice The precipitation of overt diabetes in female NOD mice usually occurs from ⬃ 15 wk of age. To study whether imatinib affects the onset of disease, we gavaged female NOD mice from 9 to 35 wk of age with saline or 100 mg/kg body wt of imatinib daily. At 35 wk of age, 4 of 10 saline-treated mice had developed diabetes whereas all 9 imatinib-treated mice remained nondiabetic (Fig. 1A). A diabetes prevalence of 40% is somewhat lower than the normal 60 –70%, which is typical for our local NOD mouse colony, but it is possible that the extra manipulation of the mice when gavaged daily slightly decreased the incidence of diabetes. Blood glucose levels of the imatinib-treated mice were similar to those of the six nondiabetic salinetreated mice (Fig. 1A, insert). We also scored the severity of insulitis and observed that imatinib did not affect the extent of peri-insulitis or insulitis (results not shown), indicating unaffected immune cell activation in the imatinib-treated mice. The -cell area was similar in the 9 imatinib-treated mice compared with the 10 control mice (Fig. 1B). However, when comparing the nine imatinib mice with the four diabetic saline-treated control mice, there was a significant difference in -cell area (Fig. 1B). This suggests that the bulk of the -cell destruction, which occurs as the NOD mice progress into overt diabetes, is counteracted by imatinib. To further analyze whether imatinib affects the early immune events preceding -cell destruction and diabetes, NOD mice were gavaged with imatinib (100 mg/ kg) or with vehicle alone daily from 3 to 9 wk of age. After sacrifice, the extent of infiltration of immune cells into the pancreatic islets was scored according to an arbitrary scale on histological sections of the pancreases. We found no apparent difference in immune cell 620
Vol. 21
February 2007
Figure 1. Imatinib prevents diabetes development in the NOD mouse. A) To study the development of diabetes, female NOD mice were gavaged daily from 9 to 35 wk of age with 100 mg/kg of imatinib or saline. Blood samples were taken at the time points indicated to assess hyperglycemia. At 35 wk of age, 4 of 10 control mice had developed diabetes whereas all 9 imatinib-treated mice remained nondiabetic (P⬍0.05 using the 2 test). The insert shows blood glucose levels of the 6 nondiabetic control mice, 9 imatinib-treated mice, and the 4 diabetic control mice on sacrifice. Results are means ⫾sem. B) Imatinib partially preserves the -cell area in NOD mice. Formalin-fixed sections from NOD mice were subjected to immunohistochemical staining for insulin. The insulin-positive area was determined in two nonconsecutive sections of the pancreas by morphometric analysis and was expressed as a percentage of the total pancreatic section area. Results are means ⫾sem. n ⫽ 10 for control, 9 for imatinib, and 4 for diabetic. *P ⬍ 0.05 using Mann-Whitney rank sum test when comparing to the imatinib-treated mice.
infiltration score between saline- and imatinib-treated mice (Table 1). We also analyzed splenocyte production of IFN-␥ and IL-10, both in unstimulated cells and in cells stimulated
The FASEB Journal
HA¨GERKVIST ET AL.
for 4 h in vitro with PMA and ionomycin. The percentages IFN-␥ and IL-10 positive splenocytes isolated from imatinib-treated mice were slightly higher in the imatinib mice (Table 2), but no statistically significant differences were observed. These data indicate that imatinib may protect against diabetes in the NOD mouse without affecting Th1/Th2 activities or insulitis.
pidium iodide positive cells, indicating improved islet cell survival (Fig. 3). The protective effect of imatinib was statistically significant at 10 M but not at 1 M. However, a 3 h pretreatment with imatinib did not significantly protect against DETA/NONOate. These results indicate that a 24 h pretreatment with imatinib promotes islet cell survival.
Imatinib prevents -cell death and protects against diabetes in STZ-injected mice
Imatinib protects against STZ-induced islet cell death in vitro
We next investigated whether imatinib protects against STZ-induced -cell damage and diabetes in vivo. NMRI mice were treated with 200 mg/kg imatinib by gavage the day before, the same day 2 h before, and the day after an i.v. STZ injection. Imatinib treatment protected partially against the rise in blood glucose after the STZ injection (Fig. 2A). The -cell area of mice treated with the high dose of STZ was markedly lowered compared with control mice (Fig. 2B). This decrease was partially counteracted by imatinib, which indicates that imatinib protected against diabetes by preventing -cell death.
The classic -cell toxin STZ kills insulin-producing cells by alkylating macromolecules such as DNA, decreasing NAD levels and promoting production of free radicals. To determine whether imatinib protects against STZ in vitro, we preincubated rat, mouse, and human islets overnight with 10 M imatinib, then added different concentrations of STZ. We observed that imatinib protects rat islets potently against a low (0.4 mM) concentration of STZ, but only weakly against the high (0.75 mM) concentration of STZ (Fig. 4A). This indicates that very high concentrations of STZ promote islet cell death by mechanisms noninhibitable by imatinib, possibly necrosis. Imatinib also protected against cell death induced by STZ in isolated human and mouse islets (Fig. 4B, C).
Imatinib protects human islets against NO-induced cell death We next investigated whether imatinib, at different concentrations and preincubation periods, protects against -cell death in vitro. For this purpose, we exposed human pancreatic islets to the NO donor DETA/NONOate. A 24 h exposure to DETA/NONOate induced massive islet cell death resulting in ⬎ 70% propidium iodide positive cells (Fig. 3). However, a 24 h pretreatment with imatinib partially prevented the DETA/NONOate-induced increase in the fraction proTABLE 1. Pancreatic islet morphology of imatinib- or salinetreated NOD micea Morphology rank A
B
C
D
Age group
Treatment
w3-w9 w3-w9 w3-w9 w3-w9
NaCl imatinib NaCl imatinib
1 2 1 0
5 4 1 3
0 0 1 0
0 0 0 0
NaCl imatinib
0 0
2 3
6 6
2 0
(female) (female) (male) (male)
w9-w35 (female) w9-w35 (female)
(Number of animals)
a NOD mice were gavaged with either saline or imatinib (100 mg/kg body wt) daily either from 3 wk of age to 9 wk of age (w3-w9) or between 9and 35 wk of age (w9-w35). Upon sacrifice after 9 or 35 wk of age, the pancreatic islet morphology was ranked according to an arbitrary scale. Rank A denotes normal islet structure; rank B denotes mononuclear cell infiltration in the islet peri-insular area; rank C denotes heavy mononuclear cell infiltration into a majority of islets, (i.e., insulitis); rank D denotes only a few residual islets present, often showing an altered architecture. Insulitis was not significantly affected by imatinib at any age (P⬎0.05 using the Chi-square test).
IMATINIB MESYLATE-MEDIATED PROTECTION AGAINST DIABETES
Down-regulation of c-Abl mRNA, but not c-Kit mRNA, protects against STZ and cytokines Imatinib is known to inhibit c-Abl, c-Kit, the plateletderived growth factor receptor (PDGFR) and the c-Abl homologue ARG (2). Even though it is unlikely that c-Kit and the PDGFR participate in -cell death, it is necessary to demonstrate that a specific down-regulation of c-Abl results in effects similar to those observed with imatinib. We previously developed a model to silence c-Abl expression in islet cells using siRNA produced by recombinant Dicer enzyme in vitro (21). Using this approach, we observed protection against STZ-induced mouse islet cell death. Dispersed islets from NMRI mice were transfected with D-siRNA against c-Abl or GL3 luciferase in one set of experiments, and siRNA against c-Kit and GL3 luciferase in another set of experiments. Aliquots of the transfected cells were exposed 30 h later to 1.5 mM STZ, which was followed after another 18 h by vital staining with propidium iodide and bisbenzimide. The stained cells were washed and analyzed by fluorescence microscopy. Percentages of dead cells were 5.6 ⫾ 0.6 and 58.4 ⫾ 6.3 for control (GL3) and STZ-treated islets, respectively (n⫽3– 4). Corresponding percentages for c-Abl siRNA-treated islets were 3.4 ⫾ 0.6 and 24.2 ⫾ 2.2 (P⬍0.05 when comparing the STZ-treated groups using Student’s t test). Corresponding results in the series of experiments using siRNA against c-Kit were 4.4 ⫾ 1.6 (GL3), 67.3 ⫾ 13.8 (GL3 ⫹ STZ), 5.9 ⫾ 2.0 (c-Kit), and 61.6 ⫾ 19.8 (c-Kit ⫹ STZ) (P⬎0.05, n⫽3). These findings strengthen the notion that imatinib acts by inhibiting c-Abl and not one of its other targets c-Kit. 621
TABLE 2. IL-10 and IFN-␥ production in splenocytes isolated from control and imatinib-treated NOD micea IL-10-positive cells (%) PMA ⫹
Unstimulated
PMA ⫹ inonomycin
ionomycin
0.28 ⫾ 0.079 0.38 ⫾ 0.093
0.21 ⫾ 0.05 0.38 ⫾ 0.093
0.087 ⫾ 0.012 0.072 ⫾ 0.027
0.61 ⫾ 0.09 0.96 ⫾ 0.09
Unstimulated
Control imatinib
IFN-␥-positive cells (%)
a
Splenocyte production of IFN-␥ and IL-10 was assessed by flow cytometry in unstimulated cells and in cells stimulated for 4 h with PMA and ionomycin. Cells from 6 control mice and 6 imatinib-treated mice were analyzed. No significant differences in response to imatinib were observed.
Since we could prevent human islet cell death after NO exposure (Fig. 3), we wanted to examine whether inhibition of c-Abl has an influence on cell death
Figure 2. Imatinib partially protects against STZ-induced hyperglycemia in vivo. A) NMRI mice were either gavaged with saline or imatinib (200 mg/kg) once daily on days ⫺1, 0, and 1. On day 0 the mice were injected with either saline or 160 mg/kg STZ i.v., and blood glucose was determined on the days indicated. Points are means ⫾sem for 5–10 animals per group. *P ⬍ 0.05, **P ⬍ 0.01, and ***P ⬍ 0.001 vs. corresponding STZ⫹imatinib group, using unpaired Student’s t test. B) Imatinib partially preserves the -cell area STZ mice. Formalin-fixed sections from NMRI mice were subjected to immunohistochemical staining for insulin. The insulin-positive area was determined in two nonconsecutive sections of the pancreas by morphometric analysis and was expressed as percentage of the total pancreatic section area. Results are means ⫾sem. *P ⬍ 0.05 using Mann-Whitney rank sum test when comparing imatinib-treated mice. 622
Vol. 21
February 2007
induced by the proinflammatory cytokines IL-1, TNF-␣, and IFN-␥, known inducers of NO production. Therefore, we treated ⌻C-6 cells with either scrambled siRNA or siRNA specific for c-Abl and analyzed c-Abl mRNA levels by RT-PCR. We observed a markedly weaker c-Abl band in cells treated with c-Abl-specific siRNA (Fig. 5A). We next investigated cell death of ⌻C-6 cells deficient in c-Abl mRNA and exposed to a combination of IL-1 ⫹ IFN-␥ ⫹ TFN-␣. We observed that cytokine-induced ⌻C-6 cell death was potently counteracted by c-Abl-specific siRNA 2 and 3 days after treatment (Fig. 5B). This indicates that a presumably
Figure 3. Pretreatment with imatinib prevents NO-induced cell death in human pancreatic islets. Human islets were incubated with imatinib at a low or high concentration (1 or 10 M) for 3 h or 24 h. This was followed by exposure to DETA/NONOate (2 mM) for another 18 h. Islets were stained for 15 min with 5 g/ml of bisbenzimide and 20 g/ml PI, carefully washed, placed on coverslips, and examined by fluorescence microscopy using the Openlab 3.0.4 software. Number of dead islet cells was quantified by fluorescence microscopy and NIH 1.63 data analysis. Results are means ⫾sem for 4 observations. *P ⬍ 0.05 vs. DETA/ NONOate-treated islets that were not exposed to imatinib using two-way ANOVA and Tukeys post-ANOVA test.
The FASEB Journal
HA¨GERKVIST ET AL.
Figure 5. Inhibition of c-Abl mediates the protective effect of imatinib against proinflammatory cytokines in vitro. ⌻C-6 cells were treated with scrambled siRNA or c-Abl-specific siRNA for 3 h. The next day c-Abl mRNA levels were analyzed by RT-PCR (A). The siRNA transfection was also followed by a 24 h exposure to the cytokines IL-1 (25 U/ml) ⫹ IFN-␥ (1000 U/ml) ⫹ TNF-␣ (1000 U/ml) at 24, 48, and 72 h post-transfection. Cell death was quantified by vital staining and flow cytometry (B). Results are means ⫾sem for 3– 4 observations. *P ⬍ 0.05 vs. cytokine-treated cells using Student’s t test.
slow turnover of the c-Abl protein results in a delay in the effect of the siRNA treatment. More important, however, the data suggest that inhibition of c-Abl results in protection against a combination of cytokines. Imatinib stimulates NO production and activates NF-B
Figure 4. Imatinib protects against STZ-induced cell death in rat (A), human (B), and mouse (C) islets. Imatinib (10 M) was added 24 h before STZ to islets in groups of 10. STZ (10 mM) was added to the human islets in the absence of glucose, whereas mouse islets were treated with 1.5 mM STZ in the presence of 5.6 mM glucose for 30 min. Rat islets were incubated in the absence of glucose when treated with different concentrations of STZ for 30 min. Islets were carefully washed, placed on coverslips, and examined by fluorescence microscopy using Openlab 3.0.4 software. Total number of cells as well as necrotic and apoptotic nuclei were counted using the NIH 1.63 software. Bars are means ⫾sem from three separate observations. *P ⬍ 0.05 and ***P ⬍ 0.001 vs. corresponding STZ⫹imatinib group using Student’s t test.
IMATINIB MESYLATE-MEDIATED PROTECTION AGAINST DIABETES
Our finding that a 24 h preincubation with imatinib was necessary to promote protection against STZ, NO, and cytokines, prompted us to investigate whether imatinib induces a state similar to that of ischemic preconditioning, as protection is often observed the day after the initial ischemia (16). To investigate this, we determined levels of nitrite from human islets or INS-1 832/13 cells incubated with imatinib and/or cytokines for 24 h. As anticipated, NO production was markedly stimulated by cytokines, both in islets and INS-1 cells (Fig. 6A, B). What is more interesting, imatinib increased nitrite levels in the presence of cytokines. Imatinib did not directly interfere with the Griess reagent (results not shown). To provide an explanation for the imatinib-induced increase in cytokine-induced NO formation, we also investigated cytokine-induced iNOS expression. Immu623
Figure 6. Potentiation of cytokine-induced NO production, NF-B activation, and iNOS expression by imatinib. Human pancreatic islets or INS-1 832/13 cells were preincubated overnight with 10 M imatinib, then further incubated with a combination of IL-1 (50 U/ml) and IFN-␥ (1000 U/ml) for 24 h. A) Nitrite content in culture medium from human islets. Bars are means ⫾sem for three separate observations. *P ⬍ 0.05, using Student’s paired t test. B) Nitrite content in culture medium from INS-1 832/13 cells. Bars are means ⫾sem for two separate experiments performed in triplicate. *P ⬍ 0.05 vs. cytokine-treated cells, using Student’s paired t test (n⫽6). C) Immunoblotting of iNOS and ATF-2 in whole cell lysates from INS-1 832/13 cells treated as shown in Fig. 7B. The transcription factor ATF-2 is used as loading control. D) Immunoblotting of NF-B and IB in human islet lysates pretreated with 1 or 10 M of imatinib overnight. E) Effect of imatinib and IL-1 on the ratio IB/p65 NF-B. Results were obtained by densitometry of immunoblots shown in panel D and are expressed are means ⫾sem. *P ⬍ 0.05 vs. control using Student’s t test (n⫽4 –5). F) EMSA for NF-B performed on INS-1 832/13 nuclear fractions. The figure is representative of three separate observations.
noblot analysis revealed that iNOS expression in cytokine-stimulated cells was more pronounced when coincubated with imatinib (Fig. 6C). On the other hand, imatinib did not induce iNOS expression in the absence of cytokines (Fig. 6C). As expression of the iNOS gene is controlled by the transcription factor NF-B, and as c-Abl has been demonstrated to inhibit NF-B activation (22), we next investigated whether imatinib affects IB levels in insulin-producing cells. Using human islets we observed an imatinib-mediated decrease in IB protein levels (Fig. 6D) but unaltered NF-B p65 levels. Furthermore, EMSA revealed that imatinib promoted an increase in nuclear NF-B activity in INS-1 823/13 cells, which was similar to that induced by IL-1 (Fig. 6D). Taken together, these data indicate that imatinib stimulates NF-B signaling by enhancing the degradation of IB. Inhibition of NF-B attenuates imatinib-induced protection against DETA/NONOate To establish the functional role of imatinib-induced NF-B activation, we blocked NF-B activity using a 624
Vol. 21
February 2007
synthetic NF-B activation inhibitor (23) in experiments using DETA/NONOate as an islet cell deathpromoting agent. As expected, DETA/NONOate induced marked islet cell death, which was partially counteracted by a 24 h preincubation with imatinib (Fig. 7). However, the protective effect of imatinib could not be observed in islets treated with the NF-B inhibitor (Fig. 7). This finding suggests that imatinibmediated survival requires NF-B activity. Imatinib augments ROS production and decreases mitochondrial membrane potential To continue to investigate the possibility that imatinib induces preconditioning, we analyzed JC-1 585/530 fluorescence in human islet cells pretreated with 10 M of imatinib overnight. JC-1 585/530 fluorescence is considered a valid estimate of mitochondrial membrane potential, and opening of mitochondrial potassium channels results in a lowered mitochondrial membrane potential (24). We observed a dose-dependent decrease in the 585/530 ratio (Fig. 8), indicating that
The FASEB Journal
HA¨GERKVIST ET AL.
Figure 7. Inhibition of NF-B prevents imatinib-mediated protection of human islets. Human islets were preincubated for 24 h with 10 M of imatinib alone or in combination with 10 nM NF-B activation inhibitor (6-amino-4-(4-phenoxiphenylethylamino)quinazoline; Calbiochem, La Jolla, CA, USA), followed by exposure to 1 mM DETA/NONOate for 18 h. Cell viability was assessed using vital staining and fluorescence microscopy. Results are means ⫾sem for islets from 2 donors, each analyzed in triplicate.
inhibition of c-Abl might result at least in part in islet cell survival through mitochondrial potassium channel opening. In additional support for a preconditioning effect of imatinib, we analyzed ROS production in imatinibtreated cells in response to STZ. STZ is known to generate oxygen free radicals and NO on intracellular decomposition (25). We labeled cells with DHCFDA, then analyzed DCF fluorescence by flow cytometry. Our results show that STZ-induced ROS production was not decreased by 6 h preincubation with imatinib (Fig. 9). Instead, ROS levels remained elevated for a longer time in imatinib-treated cells than in control cells.
Figure 8. Imatinib decreases JC-1 fluorescence in human pancreatic islet cells. Human pancreatic islets were pretreated with 10 M of imatinib for 24 h, followed by staining with JC-1 and flow cytometry. Data are expressed as the 585/530 nm fluorescence ratio normalized to the 0 time point reading of control cells. Results are means ⫾sem for islets from 3 donors. *P ⬍ 0.05 vs. corresponding nonimatinib-treated islets using Student’s t test.
DISCUSSION In this article we present a mode of action for imatinib that results in protection against diabetes. More specifically, imatinib partially or completely counteracted the Type 1 diabetes that develops spontaneously in NOD mice or in response to a single STZ injection, and this
Imatinib does not perturb normal -cell function Finally, we investigated whether imatinib affected normal -cell function, measured as the ability to release insulin in response to glucose stimulation. Performing an insulin release assay on cultured rat islets from three separate animals or human islets from six different donors preincubated with 10 M imatinib for 24 h, we could not detect any functional difference in imatinibtreated vs. untreated islets (Fig. 10). IMATINIB MESYLATE-MEDIATED PROTECTION AGAINST DIABETES
Figure 9. Imatinib augments the generation of ROS after STZ exposure. INS-1 832/13 cells were pretreated with 10 M imatinib for 24 h and labeled with DCFA-DA for 20 min. Cells were analyzed for DCF fluorescence before and after addition of STZ at the times indicated. Data are expressed as 530 nm fluorescence normalized to the 0 time point value. Results are means ⫾sem for 4 observations. *P ⬍ 0.05 and **P ⬍ 0.01, respectively, when comparing control and imatinib using Student’s t test. 625
Figure 10. glucose-stimulated insulin release from rat and human islets is not affected by imatinib. Insulin release assays on cultured rat islets from 3 separate animals (A) or human islets obtained from 6 donors (B) preincubated with 1 M or 10 M imatinib for 24 h. Islets, in groups of 5, were incubated for 60 min in KRBH containing 1.67 mM glucose, followed by another 60 min in KRBH containing 16.7 mM glucose. Insulin levels in the incubation buffers were measured using an insulin ELISA. Both imatinib-treated and control islets responded to a glucose challenge with a ⬎ 10-fold increase in insulin release. Results are means ⫾sem.
was paralleled by a partially preserved -cell mass. Although it cannot be excluded that imatinib affected signaling events pertinent to the control of peripheral insulin sensitivity and lipoprotein metabolism, it is generally agreed that the diabetes of the two models presently investigated essentially results from -cell destruction. Indeed, in the STZ-injected mouse the vast majority of the -cells are rapidly killed in response to the specific toxin; as imatinib was given only 1 day before, the same day, and 1 day after the STZ injection, effects on peripheral insulin sensitivity and lipoprotein metabolism were probably negligible during the subsequent 8 day imatinib-free period. Regarding the presently observed protection against diabetes of the NOD 626
Vol. 21
February 2007
mouse, imatinib did not seem to affect the immune system of these mice, as judged by the development of insulitis and the Th1/Th2 cytokine balance in resting or activated splenocytes. Moreover, we analyzed whether treatment of overtly diabetic NOD mice (⬎20 mM blood glucose) with imatinib for 7 days affected blood glucose levels, but failed to see a protective effect (results not shown). This further corroborates our suggested mode of action for imatinib, as there are very few -cells left to save in already diabetic NOD mice, and it argues against an imatinib-induced alteration in insulin sensitivity, as this should have manifested itself during the 7 day treatment regime. Thus, it is likely that imatinib acted in both diabetes models mainly by preventing -cell death and dysfunction. Although the results obtained here were from Type 1 diabetes models, it may be that imatinib-induced -cell protection contributes to the recently observed regression of Type 2 diabetes (5, 6). Indeed, Type 2 diabetes is also characterized by increased -cell death (12, 13), and assuming that imatinib counteracted the proapoptotic signals that are active in Type 2 diabetes patients, it is possible that restoration of normoglycemia was achieved at least in part by the imatinib effect demonstrated here. Additional studies will establish whether imatinib counteracts diabetes in animal models for Type 2 diabetes and, if so, by protecting against -cell death. Imatinib is known to protect against genotoxic agent-, death receptor activation-, and hydrogen peroxide-induced apoptosis in various cell types via inhibition of the c-Abl kinase (8, 9, 11, 26). Our in vivo and in vitro results extend such findings to the insulin-producing -cell and indicate that the proapoptotic impact of several stress-induced -cell signaling pathways is dampened by the inhibition of c-Abl, achieved either using imatinib or RNAi-induced knockdown of the c-Abl messenger. Thus, it is possible that different intracellular pathways leading to -cell death (e.g., oxidative stress activation of the ER stress pathway and/or sustained elevation of the cytosolic Ca2⫹ concentration) converge at c-Abl. Hypothetically, this could give us, employing only one approach, the ability to block a multitude of death signals, thereby achieving -cell survival in Type 1 and Type 2 diabetes. In cell types other than -cells, activation of cytosolic c-Abl has been demonstrated to result in phosphorylation of the MAP and ERK kinase-1 (MEKK-1), which in turn promotes MKK4 (mitogen-activated protein kinase kinase 4) and JNK1/2 activation (27). In response to ER stress, c-Abl has also been reported to translocate from the ER to the mitochondria and participate in the decrease of mitochondrial membrane potential (28). In -cells, chemical donors of NO or cytokine-induced NO production promote ER stress (29), loss of the mitochondrial membrane potential (30), and activation of the proapoptotic JNK1/2 kinases (31). As we observe a protective effect of imatinib against NOinduced JNK1/2 phosphorylation (32) and cell death (present study), it is likely that the events described
The FASEB Journal
HA¨GERKVIST ET AL.
above also occur in insulin-producing cells and help protect against diabetes. The model presented above for imatinib’s mode of action, although straightforward and conceptually attractive, is confounded by the finding that overnight preincubation in the presence of imatinib was required for the antiapoptotic effect. This speaks in favor of a more complex chain of events involving altered gene expression rather than only a direct imatinib-induced inhibition of the stress-activated c-Abl kinase. The findings of this study support the notion that the alternative pathway, which might operate in parallel with the direct pathway, promotes a state resembling that of ischemic preconditioning originally observed in cardiomyocytes (16). Ischemic preconditioning is defined as an increased tolerance to ischemia and reperfusion induced by a previous sublethal ischemia. On the molecular level, ischemic preconditioning is characterized by events such as NF-B and PKC activation, opening of mitochondrial potassium channels, mitochondrial membrane depolarization and reduced ATP production, increased generation of ROS, and release of cytochrome c from the mitochondria (33, 34). In many situations these events are proapoptotic, but if the initial ischemia is only transient, the proapoptotic events will instead boost the resistance to apoptosis resulting in the late protective phase (33, 34). This phase lasts for 24 to 72 h, and chemical preconditioning, induced by the pharmacological opening of mitochondrial potassium channels, is known to mimic the late phase of ischemic preconditioning (33). The findings of a previous report demonstrating that KATP channel openers protect islet cells from death (35) indicate that a preconditioning pathway is also operational in -cells. Moreover, an overnight exposure to imatinib resulted in four characteristics for preconditioning: increased ROS production, decreased mitochondrial membrane potential, NF-B activation, and increased NO production. Among these, NF-B activation appears to be a necessary event as the pharmacological inhibitor of NF-B prevented the imatinibinduced protection against NO. The imatinib-induced NF-B activation may have resulted from inhibition of c-Abl-mediated phosphorylation of IB, an event that antagonizes NF-B activation (22). In summary, we report here that imatinib treatment prevents -cell apoptosis and thereby maintains the pancreas insulin secretion capacity in situations of -cell stress. Imatinib may act by counteracting the proapoptotic effects of c-Abl activation; perhaps more important, imatinib promotes NF-B activation and an ischemic preconditioning-like state. Furthermore, in view of recent reports it is likely that imatinib has an effect on the peripheral insulin action. Clinical findings showing that chronic myeloid leukemia patients on imatinib therapy improved both their Type 2 diabetes (5, 6) and lipid status (36) support the notion that imatinib might also act peripherally to enhance insulin sensitivity. Thus, further studies of imatinib and imatinibrelated compounds may pave the way for a conceptually IMATINIB MESYLATE-MEDIATED PROTECTION AGAINST DIABETES
new diabetes drug that improves not only -cell function, but also peripheral insulin signaling. This work was supported in part by the Swedish Research Council, the Swedish Diabetes Association, the Family Ernfors Fund, the Novo-Nordisk Fund, and the Eli/Lilly-European Foundation for the Study of Diabetes.
REFERENCES 1.
2. 3.
4.
5. 6.
7. 8.
9.
10. 11.
12. 13. 14. 15. 16. 17.
18. 19.
Buchdunger, E., Zimmermann, J., Mett, H., Meyer, T., Muller, M., Druker, B. J., and Lydon, N. B. (1996) Inhibition of the Abl protein-tyrosine kinase in vitro and in vivo by a 2-phenylaminopyrimidine derivative. Cancer Res. 56, 100 –104 Okuda, K., Weisberg, E., Gilliland, D. G., and Griffin, J. D. (2001) ARG tyrosine kinase activity is inhibited by STI571. Blood 97, 2440 –2448 O’Brien, S. G., Guilhot, F., Larson, R. A., Gathmann, I., Baccarani, M., Cervantes, F., Cornelissen, J. J., Fischer, T., Hochhaus, A., Hughes, T., et al. (2003) Imatinib compared with interferon and low-dose cytarabine for newly diagnosed chronic-phase chronic myeloid leukemia. N. Engl. J. Med. 348, 994 –1004 Demetri, G. D., von Mehren, M., Blanke, C. D., Van den Abbeele, A. D., Eisenberg, B., Roberts, P. J., Heinrich, M. C., Tuveson, D. A., Singer, S., Janicek, M., et al. (2002) Efficacy and safety of imatinib mesylate in advanced gastrointestinal stromal tumors. N. Engl. J. Med. 347, 472– 480 Veneri, D., Franchini, M., and Bonora, E. (2005) Imatinib and regression of type 2 diabetes. N. Engl. J. Med. 352, 1049 –1050 Breccia, M., Muscaritoli, M., Aversa, Z., Mandelli, F., and Alimena, G. (2004) Imatinib mesylate may improve fasting blood glucose in diabetic Ph⫹ chronic myelogenous leukemia patients responsive to treatment. J. Clin. Oncol. 22, 4653– 4655 Van Etten, R. A. (1999) Cycling, stressed-out and nervous: cellular functions of c-Abl. Trends Cell Biol. 9, 179 –186 Sun, X., Majumder, P., Shioya, H., Wu, F., Kumar, S., Weichselbaum, R., Kharbanda, S., and Kufe. D. (2000) Activation of the cytoplasmic c-Abl tyrosine kinase by reactive oxygen species. J. Biol. Chem. 275, 17237–17240 Raina, D., Mishra, N., Kumar, S., Kharbanda, S., Saxena, S., and Kufe, D. (2002) Inhibition of c-Abl with STI571 attenuates stress-activated protein kinase activation and apoptosis in the cellular response to 1--D-arabinofuranosylcytosine. Mol. Pharmacol. 61, 1489 –1495 Wang, J. Y., and Ki, S. W. (2001) Choosing between growth arrest and apoptosis through the retinoblastoma tumour suppressor protein, Abl and p73. Biochem. Soc. Trans. 29, 666 – 673 Raina, D., Pandey, P., Ahmad, R., Bharti, A., Ren, J., Kharbanda, S., Weichselbaum, R., and Kufe, D. (2005) c-Abl tyrosine kinase regulates caspase-9 autocleavage in the apoptotic response to DNA damage. J. Biol. Chem. 280, 11147–11151 Mathis, D., Vence, L., and Benoist, C. (2001) -Cell death during progression to diabetes. Nature 414, 792–798 Kolb, H., and Mandrup-Poulsen, T. (2005) An immune origin of type 2 diabetes? Diabetologia 48, 1038 –1050 Atkinson, M. A., and Leiter, E. H. (1999) The NOD mouse model of type 1 diabetes: as good as it gets? Nat. Med. 5, 601– 604 Rakieten, N., Rakieten, M. L., and Nadkarni, M. R. (1963) Studies on the diabetogenic action of streptozotocin (NSC37917). Cancer Chemother. Rep. 29, 91–98 Riksen, N. P., Smits, P., and Rongen, G. A. (2004) Ischaemic preconditioning: from molecular characterisation to clinical application–part I. Neth. J. Med. 62, 353–363 Welsh, N., Eizirik, D. L., Bendtzen, K., and Sandler, S. (1991) Interleukin-1b induced nitric oxide production in isolated rat pancreatic islets requires gene transcription and may lead to inhibition of the Krebs cycle enzyme aconitase. Endocrinology 129, 3167–3173 Saldeen, J. (2000) Cytokines induce both necrosis and apoptosis via a common Bcl-2-inhibitable pathway in rat insulin-producing cells. Endocrinology 141, 2003–2010 Saldeen, J., and Welsh, N. (1994) Interleukin-1 induced activation of NF-B in insulin producing RINm5F cells is prevented
627
20.
21.
22. 23.
24. 25. 26.
27.
28.
628
by the protease inhibitor tosyl-lysine chloromethyl ketone. Biochem. Biophys. Res. Commun. 203, 149 –155 Svensson, C., Borg, L. A. H., and Welsh, N. (1999) Sodium palmitate induces partial mitochondrial uncoupling and reactive oxygen species in rat pancreatic islets in vitro. Endocrinology 140, 3422–3428 Ha¨gerkvist, R., Mokhtari, D., Myers, J. W., Tengholm, A., and Welsh, N. (2005) siRNA produced by recombinant dicer mediates efficient gene silencing in islet cells. Ann. N. Y. Acad. Sci. 1040, 114 –122 Kawai, H., Nie, L., and Yuan, Z. M. (2002) Inactivation of NF-kappaB-dependent cell survival, a novel mechanism for the proapoptotic function of c-Abl. Mol. Cell. Biol. 22, 6079 – 6088 Tobe, M., Isobe, Y., Tomizawa, H., Nagasaki, T., Takahashi, H., and Hayashi, H. (2003) A novel structural class of potent inhibitors of NF-kappa B activation: structure-activity relationships and biological effects of 6-aminoquinazoline derivatives. Bioorg. Med. Chem. 11, 3869 –3878 Ghosh, S., Standen, N. B., and Galinanes, M. (2000) Evidence for mitochondrial K ATP channels as effectors of human myocardial preconditioning. Cardiovasc. Res. 45, 934 –940 Kwon, N. S., Lee, S. H., Choi, C. S., Kho, T. and Lee, H. S. (1994) Nitric oxide generation from streptozotocin. FASEB J. 8, 529 –533 Kumar, S., Mishra, N., Raina, D., Saxena, S., and Kufe, D. (2003) Abrogation of the cell death response to oxidative stress by the c-Abl tyrosine kinase inhibitor STI571. Mol. Pharmacol. 63, 276 –282 Kharbanda, S., Pandey, P., Yamauchi, T., Kumar, S., Kaneki, M., Kumar, V., Bharti, A., Yuan, Z. M., Ghanem, L., Rana, A., et al. (2000) Activation of MEK kinase 1 by the c-Abl protein tyrosine kinase in response to DNA damage. Mol. Cell. Biol. 20, 4979 – 7989 Kumar, S., Bharti, A., Mishra, N. C., Raina, D., Kharbanda, S., Saxena, S., and Kufe, D. (2001) Targeting of the c-Abl tyrosine kinase to mitochondria in endoplasmic reticulum stress-induced apoptosis. Mol. Cell. Biol. 21, 6233– 6242
Vol. 21
February 2007
29.
30.
31.
32. 33.
34.
35.
36.
Oyadomari, S., Takeda, K., Takiguchi, M., Gotoh, T., Matsumoto, M., Wada, I., Akira, S., Araki, E., and Mori, M. (2001) Nitric oxide-induced apoptosis in pancreatic beta cells is mediated by the endoplasmic reticulum stress pathway. Proc. Natl. Acad. Sci. U. S. A. 98, 10845–10850 Barbu, A., Welsh, N., and Saldeen, J. (2002) Cytokine-induced apoptosis and necrosis are preceded by disruption of the mitochondrial membrane potential (Deltapsi(m)) in pancreatic RINm5F cells: prevention by Bcl-2. Mol. Cell. Endocrinol. 190, 75– 82 Welsh, N. (1996) Interleukin-1-induced ceramide and diacylglycerol generation may lead to activation of the c-Jun NH2terminal kinase and the transcription factor ATF2 in the insulinproducing cell line RINm5F. J. Biol. Chem. 271, 8307– 8312 Yellon, D. M., and Downey, J. M. (2003) Preconditioning the myocardium: from cellular physiology to clinical cardiology. Physiol. Rev. 83, 1113–1151 Baines, C. P., Cohen, M. V., and Downey, J. M. (1999) Signal transduction in ischemic preconditioning: the role of kinases and mitochondrial K(ATP) channels. J. Cardiovasc. Electrophysiol. 10, 741–754 Kullin, M., Li, Z., Bondo Hansen, J., Welsh, N., Karlsson, F. A., and Sandler, S. (2003) Protection of rat pancreatic islets by potassium channel openers against alloxan, sodium nitroprusside and interleukin-1 mediated suppression—possible involvement of the mitochondrial membrane potential. Diabetologia 46, 80 – 88 Ha¨gerkvist, R., Makeeva, N., Elliman S., and Welsh, N. (2006) Imatinib mesylate (Gleevec) protects against streptozotocininduced diabetes and islet cell death in vitro. Cell Biol. Int. In press Gottardi, M., Manzato, E., and Gherlinzoni, F. (2006) Gleevec and hyperlipidemia. N. Engl. J. Med. 353, 2722–2723
The FASEB Journal
Received for publication July 17, 2006. Accepted for publication September 14, 2006.
HA¨GERKVIST ET AL.
