IB1 Reduces Cytokine-induced Apoptosis of Insulin-secreting Cells*

IB1/JIP-1 is a scaffold protein that interacts with upstream components of the c-Jun N-terminal kinase (JNK) signaling pathway. IB1 is expressed at high levels in pancreatic b cells and may therefore exert a tight control on signaling events mediated by JNK in these cells. Activation of JNK by interleukin 1 (IL-1 b ) or by the upstream JNK constitutive activator D MEKK1 promoted apoptosis in two pancreatic b cell lines and decreased IB1 content by 50–60%. To study the functional consequences of the reduced IB1 content in b cell lines, we used an insulin-secreting cell line expressing an inducible IB1 antisense RNA that lead to a 38% IB1 decrease. Reducing IB1 levels in these cells increased phosphorylation of c-Jun and increased the apoptotic rate in presence of IL-1 b . Nitric oxide production was not stimulated by expression of the IB1 antisense RNA. Complementary experiments indicated that overexpression of IB1 in insulin-producing cells prevented JNK-mediated activation of the transcription factors c-Jun, ATF2, and Elk1 and decreased IL-1 b - and D MEKK1-in-duced apoptosis. These data indicate that IB1 plays an anti-apoptotic function in insulin-producing cells prob-ably by controlling the activity of the JNK signaling pathway. that maximal JNK activity occurred between 1 and 3 h of treatment with IL-1 b (10 ng/ml). Therefore, we prepared extracts after time points at 1, 3, and 6 h following IL-1 b treatment. Nitric Oxide Secretion— NO released in the cell medium was measured using the Griess reagent (Alexis). INS-1 CTR1 and AS7 cells were incubated for 16 h with doxycycline (200 ng/ml) and then cultured for two days in the presence of IL-1 b , TNF- a , and IFN- g before NO release was evaluated. 500 m l of medium was then combined with 500 m l of Griess reagent, and absorbance values (550 nm) were read 10 min later. Statistics— Results are presented as mean 6 S.E. or 6 S.D. (for n 5 2). Data were analyzed with a Wilcoxon’s test.

IB1/JIP-1 is a scaffold protein that interacts with upstream components of the c-Jun N-terminal kinase (JNK) signaling pathway. IB1 is expressed at high levels in pancreatic ␤ cells and may therefore exert a tight control on signaling events mediated by JNK in these cells. Activation of JNK by interleukin 1 (IL-1␤) or by the upstream JNK constitutive activator ⌬MEKK1 promoted apoptosis in two pancreatic ␤ cell lines and decreased IB1 content by 50 -60%. To study the functional consequences of the reduced IB1 content in ␤ cell lines, we used an insulin-secreting cell line expressing an inducible IB1 antisense RNA that lead to a 38% IB1 decrease. Reducing IB1 levels in these cells increased phosphorylation of c-Jun and increased the apoptotic rate in presence of IL-1␤. Nitric oxide production was not stimulated by expression of the IB1 antisense RNA. Complementary experiments indicated that overexpression of IB1 in insulin-producing cells prevented JNKmediated activation of the transcription factors c-Jun, ATF2, and Elk1 and decreased IL-1␤-and ⌬MEKK1-induced apoptosis. These data indicate that IB1 plays an anti-apoptotic function in insulin-producing cells probably by controlling the activity of the JNK signaling pathway.
IB1/JIP-1 are recently characterized mammalian scaffold proteins involved in the regulation of the JNK 1 signaling pathway (1)(2)(3). These two isoforms bind to and associate in a single transduction complex three kinases, MLK3, MKK7, and JNK, which together constitute an ordered unit of sequential signaling molecules transducing a variety of stress signals (2,3). To date, five different isoforms of the protein have been cloned, which are mainly N-terminal splice variants that arise from expression of one single gene on human chromosome 11p11.2-p12 (1,4,5). A S59N mutation close to the JNK binding domain of IB1 has recently been associated with a late onset type 2 diabetes (6). Functionally, this mutation led to an increased susceptibility of JNK-mediated apoptosis in different cell systems, implying a presumably important functional role of IB1 in controlling the cell response to proapoptotic stimuli (6).
Interleukin 1 (IL-1␤), which activates JNK essentially through the MKK7 pathway in several cells and tissues (7)(8)(9)(10), is believed to play a key role in the process of selective ␤ cell destruction observed in type 1 diabetes. Chronic exposure of pancreatic islets or of ␤-derived cell lines to IL-1␤ had been shown to lead to the selective death of the ␤ cells, whereas non-␤ cells such as glucagon-producing cells appeared more resistant to the action of the cytokine (reviewed in Refs. [11][12][13][14][15]. The molecular basis for the preferential killing of pancreatic ␤ versus ␣ cells by IL-1␤ is not fully understood. One important player in this phenomenon is the inducible nitric-oxide synthase gene iNOS, which is specifically expressed in the ␤ cells upon IL-1␤ treatment (16 -20). A number of reports have indeed clearly shown that ␤ cell apoptosis is NO-dependent (see for example two recent reports (21,22)). In line with this, pancreatic islets from iNOS KO mice show a better resistance to IL-1␤ cytotoxicity (23).
This NO-dependent killing of ␤ cells has been, however, challenged by several reports pointing to the existence of NOindependent death signaling pathways (24,25). For example, there is no direct correlation between expression of iNOS and sensitivity to IL-1␤ between ␤ cells at different stages of differentiation (26). Importantly, the iNOS inhibitor L-NMMA does not prevent IL-1␤-induced ␤ cell death in rat or human islets (24,25). It is also possible to block NO synthesis by blocking the extracellular signal-regulated kinase and p38 MAP kinase pathways, however, without any positive effect on cell survival (17).
To better understand the molecular mechanisms that specifically sensitize ␤ cells to IL-1␤-induced death, we recently used two different subclones of the pluripotent pancreatic endocrine stem cell clone (MSL). The MSL AN697C1 subclone gave rise to two derived cell lines, namely the glucagon-secreting AN-glu, and after stable transfection with the transcription factor pancreatic duodenal homeobox factor-1, the insulin secreting ANins (27). Despite having similar rates of NO synthesis, we found that the AN-ins cells were more susceptible to apoptosis elicited by IL-1␤. The AN-ins cells show a markedly increased activation of JNK in response to IL-1␤, thus providing a molecular basis for the observed difference in their IL-1␤ sensitivity. In these cell systems, we demonstrated that the two MAP kinases p38 and the extracellular signal-regulated kinase were fully dispensable to promote the apoptotic response. In contrast, JNK activation is essential as blocking JNK with the use of the JNK binding domain (JBD) of JIP-1/IB1 (2) prevented apoptosis by more than 90%. 2 Not only did JBD prevent apoptosis, but also a significant fraction of cells exposed to the cytokine were able to retain their ability to divide in culture. 2 * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Taken together, these data indicate that activation of the JNK pathway certainly plays an important role in IL-1␤-mediated apoptosis. Because IL-1␤ is known to activate JNK through MKK7 in different cell lines and tissues (7-10) and because both kinases interact with the scaffold protein IB1, these data suggest that the control of the MKK7-JNK signaling pathway by IB1 may interfere with the response of cells activated by IL-1␤. In this report, we specifically examined the role of IB1 in IL-1␤-induced apoptosis in pancreatic ␤ cell lines.

MATERIALS AND METHODS
Plasmids and Antibodies-The IB1, FLAG-IB1, JBD 1-280 (amino acids 1-280 of IB1 (1, 2)), and FLAG-JBD expressing vectors in the plasmid pBK (Stratagene) have been described previously (1). Expression of the constructs was monitored by Western blotting with anti-FLAG (Sigma) and anti-IB1 antibodies (1). The pEGFP-N1 vector encoding the green fluorescent protein (GFP) was from CLONTECH. This plasmid was modified by inserting a FLAG sequence (Kodak) into the NheI site of the polylinker in frame with the GFP. This FLAG sequence was designed to start with an ATG embedded in a Kozack's consensus to allow for efficient translation of the FLAG-GFP fusion protein (pEGFP-FLAG construct). The constitutively active kinase domain of MEKK1, ⌬MEKK1, was cloned into the expression vector pCDNA3 (Invitrogen) (28).
Anti-IB1 antibodies raised against amino acids 1-280 of the protein have been described (1). Anti-FLAG and anti-GST antibodies were from Sigma and Upstate Biotechnology, respectively.
Cell Lines-The insulin-secreting cell line ␤TC-3 (29) was cultured in RPMI 1640 medium (11.1 mM glucose) supplemented with 10% fetal calf serum, 100 g/ml streptomycin, 100 units/ml penicillin, and 2 mM glutamine. The INS-1 CTR1 and AS7 cell clones (6,30) were cultured in the same medium supplemented with 50 mM ␤-mercaptoethanol (31). The INS-1 AS7 cells express IB1 antisense RNA under the control of a tetracycline regulatable promoter. The addition of doxycycline (200 ng/ml) leads to full expression of the antisense RNA and to a 38% decrease in IB1 protein content (6). INS-1 CTR1 do not express a transgene construct. The cell line INS-1 appeared more resistant to the cytotoxic activity of IL-1␤ than the ␤TC-3 cells. Therefore, we potentiated the action of IL-1␤ in INS-1 by the addition of TNF-␣ and IFN-␥ (32). IL-1␤ (2 ϫ 10 5 units/g, Alexis) was used at a concentration of 10 ng/ml (TNF-␣ (10 5 units/g, Alexis) at a concentration of 10 ng/ml and IFN-␥ (Alexis) at a concentration of 100 units/ml).
Transfections-3 ϫ 10 5 cells were transfected with plasmids in 3-cm dishes using DOTAP (Roche Molecular Biochemicals) following instructions from the manufacturer. For experiments involving GAL-Jun (amino acids 1-89), GAL-ATF2 (amino acids 1-96), and GAL-Elk1 (amino acids 307-428) (all from Stratagene), 20 ng of each of these plasmids were transfected with 1 g of the reporter plasmid pFR-Luc (containing five repeats of the GAL4 DNA binding site, Stratagene) and 0.5 g of ⌬MEKK1 for 24 h. The luciferase activities were measured using the "Dual Reporter System" from Promega. To evaluate the number of apoptotic cells, the pEGFP-FLAG vector (0.5 g) was added to each transfection to identify transfected cells. Cells were then observed under an inverted fluorescence microscope (Zeiss, Axiovert 25). Apoptotic cells were discriminated from normal cells by the characteristic "blebbing" of the cytoplasm, easily determined from the fluorescence emitted by the GFP. For experiments involving the INS-1 CTR1 and AS7 clones (6), cells were incubated with Hoechst 33342 and propidium iodide (33) for 7 min before visualization under the inverted fluorescent microscope. A minimum of 1000 cells in duplicate was counted for each experiment.
Whole Cell Lysate and Solid Phase JNK Assay-Cellular extracts were prepared by scraping cells in lysis buffer (20 mM Tris acetate, 1 mM EGTA, 1% Triton X-100, 10 mM p-nitrophenyl-phosphate, 5 mM sodium pyrophosphate, 10 mM ␤-glycerophosphate, 1 mM dithiothreitol). Debris were removed by centrifugation for 5 min at 15,000 rpm in a SS-34 rotor (Beckman). One g of GST-Jun (amino acids 1-89), GST-ATF2 (amino acids 1-96), or GST-Elk1 (amino acids 307-428) was then added to 100 g of cellular extracts supplemented with 10 mM MgCl 2 and 5 Ci of [␥-33 P]ATP. Following incubation at 30°C for 20 min, reaction products were separated by SDS-polyacrylamide gel electrophoresis on a denaturing 10% polyacrylamide gel. The gels were dried, stained with Coomassie Blue to check for equal loading of the samples, and subsequently exposed to x-ray film (Kodak).
For the JNK solid phase assays, extracts were incubated for 1 h at room temperature with 1 g of GST-Jun and 10 l of glutathione-agarose beads (Sigma). Following four washes with the scraping buffer, the beads were resuspended in the same buffer supplemented with 10 mM MgCl 2 and 5 Ci of [␥-33 P]ATP. Following incubation at 30°C for 30 min, extracts were processed as before. In INS-1 and ␤TC-3 cells, we observed that maximal JNK activity occurred between 1 and 3 h of treatment with IL-1␤ (10 ng/ml). Therefore, we prepared extracts after time points at 1, 3, and 6 h following IL-1␤ treatment.
Nitric Oxide Secretion-NO released in the cell medium was measured using the Griess reagent (Alexis). INS-1 CTR1 and AS7 cells were incubated for 16 h with doxycycline (200 ng/ml) and then cultured for two days in the presence of IL-1␤, TNF-␣, and IFN-␥ before NO release was evaluated. 500 l of medium was then combined with 500 l of Griess reagent, and absorbance values (550 nm) were read 10 min later.
Statistics-Results are presented as mean Ϯ S.E. or Ϯ S.D. (for n ϭ 2). Data were analyzed with a Wilcoxon's test.

RESULTS
Cytokines, Low Glucose, UV Light, and ⌬MEKK1 Decreased IB1 Content-JNK regulates the stability of its associated transcription factors by targeting them to ubiquitination and proteolytic degradation (34,35). To determine whether the JNK activator IL-1␤ could modulate IB1 content, we performed Western blot analyses of ␤TC-3 and INS-1 cellular extracts treated with IL-1␤ for two days. Compared with control ␤TC-3 cells, IL-1␤ induced a 55% decreased IB1 content normalized to ␤-tubulin. Similarly, a combination of IL-1␤, TNF-␣, and IFN-␥ induced a 52% decrease in normalized IB1 content in INS-1 cells (Fig. 1A). IL-1␤ and TNF-␣ only weakly decreased IB1 content in this cell line (Fig. 1B). The combination of all three cytokines was chosen because we found that neither cytokine alone induced significant apoptosis in this cell line (32).
To determine whether other JNK activators could decrease IB1 levels, we treated cells with UV light or incubated them in a low glucose medium (0.5 mM glucose) (36,37). We also cotransfected cells with expression vectors encoding the activated form of the upstream JNK activator ⌬MEKK1 (28), together with the pEGFP-FLAG vector used as an internal standard to normalize for transfection efficiencies. Compared with cells in control conditions, UV light decreased IB1 levels by 68%, low glucose treatments lowered IB1 by 73%, and ⌬MEKK1 decreased IB1 by 67% (Fig. 1C).
IB1 Prevented Cytokine and ⌬MEKK1-induced Apoptosis-We evaluated the number of apoptotic ␤ cells with a combination of propidium iodide and Hoechst 33342 nuclear staining as described previously (33). This combination of dyes allows for the differential staining of necrotic, apoptotic, and live nuclei (33). Two days of culture in the presence of IL-1␤ (10 ng/ml) induced a 3.5-fold increased apoptotic rate in ␤TC-3 cells (Fig. 2). Transfection with ⌬MEKK1, a strong activator of the MAP kinase kinase MKK4 (also termed JNK kinase 1), promoted a strong apoptotic response from 3.3 Ϯ 1.1% to 23.1 Ϯ 4.8% in control and ⌬MEKK1 transfected cells, respectively (Fig. 2). This rate of apoptosis appeared similar to the one previously reported for other non-␤ cell lines (38).
To determine whether overexpression of IB1 could prevent IL-1␤-or ⌬MEKK1-induced apoptosis, we transfected ␤TC-3 cells with vectors expressing the IB1 protein (1). In these experiments, cells were co-transfected with an expression vector encoding the green fluorescent protein (pEGFP-FLAG). Following transfection, cells were incubated with 10 ng/ml IL-1␤ for two days, and apoptotic cells were counted. We noted that IB1 had a small but reproducible "survival" effect on cells in control conditions lowering the apoptotic rate by 0.5-1% (data not shown). Furthermore, overexpression of IB1 reduced IL-1␤induced apoptosis by 80% (Fig. 2). We observed similar protection with the INS-1 cell line activated by a combination of IL-1␤/TNF-␣/IFN-␥ (data not shown). IB1 also decreased apoptosis promoted by ⌬MEKK1 (Fig. 2). Similar protection was also observed with the JBD only of IB1/JIP-1 (data not shown). 2 IB1 Decreased JNK-mediated Activation of c-Jun, ATF2, and Elk1-The effects of JNK on gene transcription are accomplished via the phosphorylation of the trans-acting domains of different transcription factors including c-Jun, ATF2, and Elk1, an event resulting in an increased transcriptional activity. To demonstrate activation of MAP kinases by IL-1␤ in the insulin-secreting cell line ␤TC-3, we first performed in vitro kinase assays in whole cell lysates using as substrate GST-Jun (a substrate for JNK only), GST-ATF2 (phosphorylated by JNK and p38), and GST-Elk1 (phosphorylated by JNK and extracellular signal-regulated kinase 1/2). Kinase activity toward all three fusion proteins was detected in control cells (0Ј, Fig. 3A). IL-1␤ weakly enhanced phosphorylation of c-Jun, whereas ATF2 was strongly activated after 1 h of treatment, and Elk1 activation appeared to be more persistent. To specifically determine whether IL-1␤ activated JNK, we performed a solid phase assay using GST-Jun as substrate. We observed a strong and persistent activation of JNK after exposure to IL-1␤, peaking at 1-3 h and declining at 16 h (Fig. 3B).
To study the regulation of JNK signaling by IB1 in ␤TC-3 cells, we used a heterologous GAL4 reporter system in which the GAL4 DNA binding domain was linked to the transactivation domains of c-Jun, ATF2, or Elk1. As shown in Fig. 4A, IL-1␤ induced a weak activation of ATF2 and Elk1 in ␤TC-3 cells that could be blocked by overexpression of IB1. To more potently activate JNK, we transfected cells with ⌬MEKK1. This lead to a strong activation of Elk1 (Ͼ50-fold, Fig. 4B). Surprisingly, ⌬MEKK1 activated c-Jun and ATF2 only modestly (6 -8-fold). These low levels of c-Jun and ATF2 activation appeared specific to the ␤ cell lines used, as control experiments employing HeLa and Swiss 3T3 cells gave a similar high level of activation of the three transcription factors (about 50 -100-fold). 3 However, co-transfection with vectors encoding IB1 prevented activation of the three transcription factors, an effect consistent with the one observed with the JBD alone of IB1 (2) (Fig. 4B).  Cells were then processed as in A with normalization performed using antibodies against ␤-tubulin. The ratio of IB1/tub in control condition is set to 100. Note that only a combination of the three cytokines induce significant apoptosis in this cell line. n ϭ 3, only the decrease in presence of all three cytokines is statistically significant (p Ͻ 0.05). C, quantification of IB1 decrease in ␤TC-3 cells under different stress conditions. Control (CTRL) and IL-1␤-treated cells were processed as in A. Other treatments: 0.5 mM glucose, cells were incubated at 0.5 mM glucose for two days; UV, cells were exposed to 20 J/m 2 UV light (Stratalinker, Stratagene) and were processed 1 h later; ⌬MEKK1, cells were co-transfected with pcDNA3.1 or ⌬MEKK1 and the pEGFP-FLAG vector for normalization for 24 h. Blots were then exposed to either anti-IB1 and antitubuline antibodies (CTRL, IL-1␤, and 0.5 mM glucose samples) or to anti-IB1 and anti-FLAG antibodies (⌬MEKK1). Following densitometric scanning of the blots, the ratio of IB1/tub or IB1/FLAG was calculated (normalized IB1 content (%)). The ratio of IB1/tub in control condition is set to 100. n ϭ 3-5 for each experimental conditions, p values relative to CTRL are below 0.05.

FIG. 2. IB1 decreases IL-1␤-and ⌬MEKK1-induced apoptosis.
Cells were transfected with pEGFP-FLAG, and the indicated vectors were transfected for either 16 (⌬MEKK1) or 48 h (IL-1␤). Apoptotic cells were then counted. A minimum of 1000 transfected cells were counted in five separate experiments. p Ͻ 0.01 for IL-1␤ and ⌬MEKK1 conditions relative to CTRL. p Ͻ 0.01 for IL1-␤/IB1 and ⌬MEKK1/IB1 conditions relative to IL-1␤ and ⌬MEKK1, respectively in ␤ cells interfered with the apoptotic response, we used the insulin-secreting INS-1 clone AS7 previously described (6). This cell clone expresses IB1 antisense RNA under the control of a doxycycline inducible promoter leading to a 38% decrease in IB1 content. The cell clone INS-1 CTR1 was used here as a control (6).
In pilot experiments, we observed that INS-1 cells were relatively resistant to the action of IL-1␤, so that a combination of IL-1␤, TNF-␣, and IFN-␥ was then used to promote significant apoptosis (32). At day 1, cells were seeded on plates, and expression of the IB1 antisense RNA was induced by doxycycline (200 ng/ml). 24 h later, IL-1␤, TNF-␣, and IFN-␥ were added, and cells were incubated for a further 48 h. The number of apoptotic cells was then evaluated. INS-1 AS7 and INS-1 CTR1 cells have a basal apoptotic rate of 1.5-2%, which is increased to 17-19% in the presence of IL-1␤/TNF-␣/IFN-␥. Doxycycline had no effect on the control INS-1 CTR1 cells in presence or absence of IL-1␤/TNF-␣/ IFN-␥. However, expression of the IB1 antisense RNA in the INS-1 AS7 cell line more than doubled the apoptotic rate in the presence of IL-1␤ (Fig. 5).

INS-1 AS7 Cells Do Not Release More NO in Presence of Doxycycline and
Cytokines-Nitric oxide production in response to IL-1␤ participates to the apoptotic response of the INS-1 cell line (39). We therefore measured NO production in INS-1 CTR1 and AS7 cells that had been treated or not with IL-1␤/TNF-␣/IFN-␥ and doxycycline. As shown in Fig. 6, the decrease in IB1 content induced by doxycycline in AS7 cells did not lead to an increased NO production in response to the cytokines.

INS-1 AS7 Cells Showed Higher Phosphorylation of c-Jun in
Response to Doxycycline and Cytokines-To determine whether the increased apoptotic rate observed in the INS-1 AS7 was associated with a differential ability of JNK to activate c-Jun, we performed solid phase JNK assays. Cells were treated with IL-1␤/TNF-␣/IFN-␥ in the presence/absence of doxycycline for various times, and JNK was pulled down using GST-Jun. As shown in Fig. 7, JNK from the INS-1 AS7 cells in the presence of doxycycline (200 ng/ml) was more active in phosphorylating c-Jun during the initial phase of the response to IL-1␤/TNF-␣/ IFN-␥ (3.8-fold increase in the presence of doxycycline compared with 2.1-fold in its absence after 1 h). It is not clear from these experiments 1) whether the observed increased c-Jun phosphorylation resulted from an actual increase in JNK activity or 2) whether it resulted from a higher JNK availability in the assay that would be secondary to the lower IB1 content. DISCUSSION We showed herein that stress events decrease IB1 levels in insulin-producing cells. Our results using a cell line expressing IB1 antisense RNA indicate that this single event sensitizes FIG. 3. Persistent activation of JNK in ␤TC-3 cells by IL-1␤. A, top, whole cell lysate kinase assays using GST-Jun, GST-ATF2, and GST-Elk1 as substrate. Whole cell extracts were prepared from ␤TC-3 cells as described under "Materials and Methods." Times of exposure with IL-1␤ are indicated. One half of the reaction was loaded on a polyacrylamide gel, and ␥-33 P-phosphorylation of the substrates (P-GSTs) was subsequently analyzed with an InstantImager Apparatus (Perkin-Elmer). Bottom, the other half of the reaction was analyzed by Western blotting with anti-GST antibodies (␣-GSTs) to show equal loading of the samples. B, solid phase JNK assays. JNK was first purified using GST-Jun, and then kinase assays were performed. ␥-33 P-Phosphorylated substrates were separated on a polyacrylamide gel that was analyzed with an InstantImager Apparatus (Perkin-Elmer). Samples were processed as in A.
FIG. 4. Activation of JNK by IL-1␤ and ⌬MEKK1 in a luciferase reporter assay. A, ␤TC-3 cells were transfected with the 5xGALluciferase reporter vector and constructs expressing the indicated factors. Empty vectors were added when necessary. IL-1␤ (10 ng/ml) was added 3 h following transfections as indicated. After 16 h, cell extracts were recovered and normalized to protein content, and luciferase activities were measured. The activities of the GAL-Elk1, GAL-Jun, and GAL-ATF2 constructs in the absence of IL-1␤and IB1-expressing plasmids are set to 100. n ϭ 3, p Ͻ 0.05 for ATF2/IL-1␤ and Elk1/IL-1␤ conditions relative to ATF2 and Elk1, respectively. p Ͻ 0.05 for ATF2/ IL-1␤/IB1 and Elk1/IL-1␤/IB1 conditions relative to ATF2/IL-1␤ and Elk1/IL-1␤. B, ␤TC-3 cells were transfected with the 5xGAL-luciferase reporter vector and constructs expressing the indicated factors. Empty vectors were added when necessary. After 16 h, cell extracts were processed as described in A. The activities of the GAL-Elk1, GAL-Jun, and GAL-ATF2 constructs in the absence of ⌬MEKK1and IB1-expressing plasmids are set to 100. n ϭ 5-7, p Ͻ 0.01 for all ⌬MEKK1 conditions relative to controls. p Ͻ 0.01 for all ⌬MEKK1/IB1 conditions relative to IL-1␤. RLU, relative light units (arbitrary units). cells to IL-1␤-induced apoptosis without an increase in NO synthesis. Because IL-1␤ is recognized as one of the key mediators of pancreatic ␤ cell apoptosis in type 1 diabetes, we propose that the high amount of IB1 normally present in insulin-producing cells is a critical parameter that preserves cells from cytokine-induced destruction (Fig. 8).
The absence of increased NO production in response to IL-1␤ in the INS-1 AS7 relative to INS-1 CTR1 cells indicates that activation of other pathways is responsible for the higher apoptotic rate of this cell line. This does not preclude NO as a necessary mediator of apoptosis in this cell system. Indeed, INS-1 cells have been shown to be protected from the deleterious effects of IL-␤ by the iNOS inhibitor L-NMMA (39). Rather, these and previous data indicate that NO is not the only player mediating ␤ cell apoptosis and that independent activation of JNK is also required. 2 In this sense, endogenous regulators of JNK may play a specific role in the development of ␤ cell apoptosis.
Our results indicate that IB1 plays an important role in limiting the effects of JNK signaling on the fate of insulinproducing cells. Three lines of evidences support this conclusion. First, decreasing IB1 levels with an antisense RNA sensitizes the INS AS7 cells to IL-1␤-induced apoptosis (Fig. 5). This effect is likely to result from an increased ability of JNK to phosphorylate its substrates including c-Jun (Fig. 7). Second, Because IB1 interacts ϳ100-fold more tightly with JNK than does c-Jun or ATF2 (2), IB1 would in such a situation prevent c-Jun and ATF2, but not Elk1, from interacting with JNK. Therefore, most of the JNK activity would be directed toward Elk1. B, in the absence of IB1, JNK is free to interact with c-Jun, ATF2, and Elk1. The model implies that pancreatic ␤ cells activated by IL-1␤ gradually pass from the situation in "A" to the one in "B" therefore allowing for the recruitment of c-Jun and ATF2. These events would then facilitate apoptosis. overexpression of IB1 protects pancreatic ␤ cell lines against the two proapoptotic stimuli IL-1␤ and ⌬MEKK1 (Fig. 2). This effect correlates with the ability of IB1 to block JNK-mediated activation of c-Jun and ATF2 (Fig. 4, A and B). Third, we found a correlation in INS-1 cells between the decrease in IB1 levels and the proapoptotic potential of the three cytokines IL-1␤, TNF-␣, and IFN-␥ alone or in combination (Fig. 1B).
IB1/JIP-1 binds three kinases, MLK3, MKK7, and JNK, which together constitute an ordered three-partite signaling module leading to activation of JNK. We described IB1 as being localized in both the nucleus and cytoplasm of pancreatic ␤ cells (1). Because IB1 is expressed at very high levels in pancreatic ␤ cells compared with most other cell types, the high amount of IB1 in pancreatic ␤ cells may exert a very stringent control on JNK signaling. High levels of IB1 may be expected to have negative effects on the ability of the JNK cascade to transmit signaling (2,3,5), and in ␤ cells, this may be accounted for by two distinct mechanisms. First, high amounts of IB1 may have a "dispersive" effect by favoring the formation of incomplete, nonproductive signaling complexes (i.e. complexes lacking either one of the three kinases that normally transmit signaling) (2,3). This situation occurs in experimental conditions like the transfection studies used in this study and as a result leads to uncoupling of JNK from its upstream activators MKK7 and MLK3. This leads to inefficient JNK activation that translates into low c-Jun, ATF2, or Elk1 phosphorylation (5). Second, IB1 may have a "competitive" effect in the nucleus acting against the binding of c-Jun and ATF2 to JNK. This hypothesis is supported by the observation that IB1/JIP-1 binds ϳ100-fold more tightly to the same domain of JNK than does c-Jun or ATF2 (2). In conditions where most of the JNKs available in a cell would be bound to IB1, c-Jun and ATF2 would be efficiently prevented to interact with and to be activated by JNK. As a consequence, high amounts of IB1 may "route" JNK signaling toward factors, such as Elk1 that do not use a "JNK binding domain" similar to that of c-Jun, ATF2, or IB1 to become substrate of JNK (40,41).
Our data indicate that the near absence of c-Jun and ATF2 activation at late time points (3-6 h of IL-1␤, Fig. 3A) in whole cell lysates occurs despite a robust activation of JNK (Fig. 3B). There is thus dissociation between JNK activation and c-Jun and ATF2 phosphorylation in these cells. Transient transfection experiments employing ⌬MEKK1 indicate that only Elk1 is strongly activated by JNK, in contrast to c-Jun and ATF2 (Fig. 4B). In control experiments performed with HeLa and Swiss 3T3 cells (which express only minute amounts of IB1), activation of the three factors reached similar high levels. 3 Therefore in pancreatic ␤ cells, JNK signaling appears to be "routed" preferentially toward Elk1 activation, and our data indicate that IB1 may play a role in this process (Figs. 4B and  7). The selective activation of Elk1 in detriment to c-Jun and ATF2 is presumably not sufficient to induce apoptosis (42). If the model holds true, the observed IB1 decrease following JNK stimulation is therefore likely to represent an important control step to allow for the apoptotic response to develop (Fig. 8). The mechanisms by which the stability of IB1 is regulated by JNK is not known yet, but we may speculate that it involves a phosphorylation-dependent ubiquitination process, as observed with other JNK targets (34,35).
The critical role of IL-1␤ as mediator of pancreatic ␤ cell death in type 1 diabetes is well recognized (12,14). Our data have demonstrated the essential role that activation of the JNK pathway plays in IL-1␤-induced ␤ cell apoptosis. 2 The intracellular events that transmit IL-1␤ signaling involve the sequential activation of the two kinases MKK7 and JNK (7-9). These kinases are held together in a multiple protein complex by the scaffold protein IB1. Here, we have established that high expression of this scaffold protein in pancreatic ␤ cells limits the output of JNK signaling toward apoptosis induced by IL-1␤. As this protective effect may be attenuated following the observed decrease of IB1 in IL-1␤-treated cells, our results may lead to therapeutic strategies aimed at preserving IB1 function in pancreatic ␤ cells.