Intersectin activates Ras but stimulates transcription through an independent pathway involving JNK.

Intersectin (ITSN) is a molecular scaffold involved in regulating endocytosis and mitogenic signaling. We previously demonstrated that ITSN transformed rodent fibroblasts, accelerated hormone-induced maturation of Xenopus oocytes, and activated the Elk-1 transcription factor through an MEK- and Erk-independent mechanism. We now demonstrate that ITSN complexes with the Ras guanine nucleotide exchange factor Sos1 leading to increased RasGTP levels. Using fluorescence resonant energy transfer analysis, we demonstrate that ITSN complexes with Ras in living cells leading to Ras activation on intracellular vesicles. These vesicles contain epidermal growth factor receptor but are distinct from transferrin-positive vesicles. However, Ras is not required for ITSN stimulation of transcription. Rather, we demonstrate that ITSN signals through JNK to activate Elk-1. Although ITSN activation of Elk-1 was Ras-independent, ITSN cooperates with Ras to synergistically activate JNK. These findings indicate that ITSN activates multiple intracellular signaling pathways and suggest that this adaptor protein may coordinately regulate the activity of these pathways in vivo.

Preliminary characterization of the pathway(s) involved in Elk-1 activation by ITSN revealed that this activity was independent of MEK1/2 and Erk MAPK (9). This result was surprising given the proposed involvement of ITSN in activation of Ras, a potent activator of the Raf-MEK-Erk pathway (10,11). However, a recent study demonstrated that Ras activates distinct signaling pathways at different endomembrane compartments (12). Thus, we were interested in determining the contribution of Ras to ITSN signaling. The results presented herein demonstrate for the first time that ITSN stimulates RasGTP levels and forms a complex with Ras in vivo on a previously uncharacterized vesicular compartment. However, ITSN stimulation of Elk-1 does not require Ras function. In addition, ITSN signaling does not require the kinase activity of the epidermal growth factor receptor although EGFR and ITSN cooperate to synergistically activate Elk-1 and are co-localized on vesicles. Rather, ITSN stimulates Elk-1 through a JNK-dependent pathway. Inhibition of JNK attenuates Elk-1 activation by ITSN. Conversely, ITSN overexpression activates JNK. These data further define the biochemical pathways through which ITSN functions and support the model that ITSN is a pivotal component in integrating numerous cellular processes, including activation of GTPase cascades, endocytosis, and mitogenesis.

MATERIALS AND METHODS
Cells and Reagents-293T and NIH/3T3 cells were maintained as previously described (13). Growth factors were purchased from the following: recombinant human platelet-derived growth factor BB, recombinant human hepatocyte growth factor, and recombinant human basic fibroblast growth factor were from R & D Systems, Inc.; human recombinant epidermal growth factor was from Upstate Biotechnology; and bovine insulin was from Invitrogen. SB203580 and SP600125 (also known as Jnk inhibitor II) were purchased from Calbiochem. For each 10 M of JNK inhibitor added to the media, 0.1% Me 2 SO was also added (e.g. 20 M SP600125/0.2% Me 2 SO). All the inhibitors were resuspended in Me 2 SO as carrier. PY20 antibody was purchased from BD Transduction Laboratories. FLAG monoclonal antibody was purchased from Sigma. Phosphospecific JNK antibody was purchased from New England Biolabs. Monoclonal anti-hemagglutinin antibody was purchased from Covance. Polyclonal GST antibody conjugated to horseradish peroxidase was purchased from Santa Cruz Biotechnology. Polyclonal antibodies to Sos-1 were purchased from Santa Cruz Biotechnology and Upstate Biotechnology. * 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  Constructs-The Xenopus laevis intersectin full-length, EH, coiledcoil, and SH3 expression constructs in pCGN-Hyg were previously described (9). Mouse intersectin was isolated by PCR from reversetranscribed total RNA from NIH/3T3 cells using the primers 5ЈmInt (5Ј-GGATCCATGGCTCAGTTTCCCACACCT) and 3ЈmInt (5Ј-GGATC-CGGATGGACAACATATGATTCA). Products were subcloned into pCRII (Invitrogen) and subjected to DNA sequence analysis to confirm wild type sequence. The full-length cDNA was digested with BamHI to release the fragment from pCRII and subcloned into pCGN-Hyg as a BamHI fragment. All clones possess an amino-terminal hemagglutinin epitope tag. YFP-tagged mouse ITSN was created by subcloning the BamHI fragment from pCGN-ITSN (mouse) into pEYFP (Clontech) digested with BglII. The pGEX expression constructs were previously described (1) with the exception of SH3B, which was constructed using Vent polymerase (New England Biolabs) to amplify the SH3B from pCGN ITSN SH3 with the primers 5Ј-CGC GGA TCC GTA GAA GGC CTT and 5Ј-TTC CTT TTT TCG CCG GCG TTA ACC GGA TAT AAG. The fragment was subcloned into pCRII T/A vector (Invitrogen), sequenced, then digested with BamHI/NotI (New England Biolabs) and subcloned into pGEX 4T-2. The Ras dominant-negative expression constructs were generated by subcloning a BamHI fragment from pCMV FLAG-based constructs (kindly provided by Dr. Lawrence Quilliam) into pEF6 V5/His digested with BamHI (Invitrogen). The JNK dominant-negative construct (Jnk(APF)) was kindly provided by Dr. Roger Davis.
Reporter Assays-293T cells were transfected with the GAL4-Elk-1 and 5xGAL4-luciferase reporter plasmids as previously described along with the indicated expression constructs, and reporter activation was measured as previously described (13). Following serum starvation overnight, cells were either left unstimulated or stimulated for 4 -6 h with growth factor. Cells were washed with warm PBS then harvested in 1ϫ luciferase lysis buffer (Analytical Luminescence). Luciferase activity was then measured in a Dynex 96-well microtiter plate luminometer, and the relative light units (RLUs) per microgram of protein were determined. For each experiment, relative activation was normalized to the vector control determined by dividing the RLU/g for each experimental point by the RLU/g for unstimulated vector-transfected samples.
Ras Activation Assays-Determination of the RasGTP levels was performed essentially as described previously (14). Briefly, a GST fusion protein of the Ras binding domain of Raf, GST-Raf-RBD, was used to selectively purify RasGTP from cell lysates prepared as follows: 3-5 ϫ 10 6 293T cells per 100-mm dish were transfected with 0.5 g of pCGN-H-Ras wild type along with the various ITSN expression plasmids using the calcium phosphate precipitation method. After allowing precipitates to incubate on cells for several hours, the media were replaced with fresh media. Following an additional incubation, the cells were transferred to serum-free media overnight. On the following day, cells were incubated in the absence or presence of EGF (100 ng/ml) for 5 min, washed once with PBS, and then lysed in 500 l of GTPase lysis buffer (25 mM HEPES, pH 7.5, 10% glycerol, 150 mM NaCl, 1% Triton X-100, 0.25% sodium deoxycholate, 25 mM MgCl 2 , 1 mM EDTA, 25 mM NaF, 1 mM vanadate, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 10 g/ml aprotinin). Lysates were incubated at 4°C for 10 -15 min with gentle mixing and then clarified by centrifugation for 10 min. Equivalent amounts of GST-Raf-RBD bound to GSH beads were then added to 250 l of each lysate and incubated at 4°C for 30 -60 min with gentle mixing. Beads were then spun down and washed 4ϫ with GTPase lysis buffer, resuspended in 50 l of 4ϫ NuPAGE sample buffer supplemented with 5% ␤-mercaptoethanol and then heated to 70°C for 10 min. Ten microliters of each sample was then fractionated on a 4 -12% NuPAGE gel along with equivalent amounts of lysate from each sample as a control for total Ras levels. Gels were transferred to Immobilon-P membranes and then analyzed by Western blot with anti-HA antibodies. To determine the level of Ras activation, the signal for Ras bound to the GST-Raf-RBD beads was divided by the signal for total Ras in each sample and then normalized to the vector control, unstimulated sample.
FRET Analysis-COS-1 cells were co-transfected with YFP-tagged mouse ITSN and either CFP-H-Ras or Raf1RBD-CFP and untagged H-Ras. Transfected cells were imaged alive the next day with a Zeiss 510 laser scanning confocal microscope. The distributions of YFP-ITSN and CFP-H-Ras or Raf1RBD-CFP were imaged with conventional CFP and YFP settings. FRET was analyzed as sensitized acceptor emission by excitation at 458 nm with an argon laser and detection of emissions, limited by a 560-nm bandpass filter. Acquired images were processed with Adobe Photoshop 6.0 to enhance contrast. FRET was quantified by reversal of donor quenching after acceptor photobleaching at 514 nm (15). From each cell two to three individual vesicles (for a total of 25 vesicles) that co-expressed ITSN and H-Ras were selected, and CFP fluorescence intensity (Icfp) was measured prior to and following acceptor photobleaching at 514 nm. FRET efficiency was calculated as follows: (Icfp postbleach Ϫ Icfp prebleach )/Icfp postbleach .
EGF and Transferrin Internalization Assays-Cells to be examined by fluorescence microscopy were plated at 2 ϫ 10 5 per plate into 35-mm dishes containing a glass coverslip-covered 15-mm cutout (MatTek) and transfected with YFP-ITSN the next day. Fluorescent loading of endosomes was accomplished by incubating cells with 5 g/ml Texas Redconjugated transferrin (Molecular Probes) for 30 min at 37°C followed by removal of the unincorporated probe. Fluorescent labeling of activated EGFRs was accomplished by incubating cells with 40 ng/ml Texas Red-conjugated EGF (Molecular Probes) for 20 min at 37°C. Dual-color digital laser scanning confocal microscope was used to continuously monitor ligand internalization by individual cells.
JNK Assays-HEK 293T cells were transiently co-transfected with the indicated expression constructs and a FLAG-epitope-tagged JNK expression construct. Following an overnight incubation in serum-free media, cell lysates were harvested and Western blots were performed to assess the expression of FLAG-JNK. Lysates were normalized so that equal amounts of FLAG-JNK were immunoprecipitated with a FLAG monoclonal antibody. Precipitates were washed three times with icecold PLC-LB (50 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1 mM EGTA, 1.5 mM MgCl 2 , 100 mM NaF) supplemented with 1 mM vanadate, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 10 g/ml aprotinin then resuspended in 25 l of 4ϫ NuPAGE sample buffer supplemented with 5% ␤-mercaptoethanol. After heating to 70°C for 10 min, equivalent amounts of samples were fractionated on duplicate 4 -12% NuPAGE gels (Invitrogen), transferred to Immobilon-P membranes, and probed with antibodies to the FLAG epitope to determine total levels of JNK or antibodies to activated, dually phosphorylated JNK (New England Biolabs).
GST Pull-down Assays-Bacterially expressed GST fusion proteins were prepared essentially as described (16) except that bacterial pellets were lysed in PLC-LB. Ten micrograms of the individual GST-SH3 domains bound to Sepharose beads was combined with 250 g of lysate from cells stimulated with EGF for varying lengths of time and then incubated at 4°C for 2 h with gentle mixing. The beads were then pelleted at 20,800 ϫ g for 2 min at 4°C, washed five times with 500 l of PLC-LB with inhibitors, resuspended in 25 l of 4ϫ NuPAGE loading buffer (Invitrogen) supplemented with 5% ␤-mercaptoethanol, boiled at 100°C for 5 min, fractionated on NuPAGE gels, and then transferred to Immobilon-P filters. The filters were then blocked in 5% nonfat dry milk, 1ϫ Tris-buffered saline, 0.05% Tween 20 (Sigma), pH 7.4 (TBS-T), overnight at 4°C. The top half of the filters was then probed with antibodies to Sos1 (2 g/ml) in PBS containing 3% nonfat dry milk overnight at 4°C. The lower half of the blots were probed with ␣GSThorseradish peroxidase TBST containing 3% nonfat dry milk for 1 h at room temperature. Filters were washed five times for a total of 50 min. Signals were developed with SuperSignal chemiluminescence reagent (Pierce).

ITSN Activation of Elk-1 Does Not Require EGFR Kinase
Activity-We previously demonstrated cooperativity between ITSN and EGF in the synergistic activation of Elk-1 and in the transformation of rodent fibroblasts (9). Given this finding coupled with the fact that ITSN overexpression is sufficient to inhibit endocytosis (4,17), it is possible that the ability of ITSN to activate Elk-1 may stem from inhibition of EGFR downregulation thereby leading to enhanced or sustained signaling from the activated receptor at the cell surface. Alternatively, ITSN may stimulate an autocrine loop involving EGFR ligands thereby leading to receptor activation and Elk-1 stimulation. To test whether ITSN signaling was dependent on the EGFR kinase activity, we treated cells with a specific pharmacological inhibitor of EGFR, PD153035 (18) (Fig. 1). Inhibition of EGFR did not attenuate ITSN activation of Elk-1, although there was a complete loss of Elk-1 activation with EGF stimulation alone in the presence of this inhibitor. In addition, inhibition of EGFR abolished the cooperative activation of Elk-1 by ITSN and EGF.
ITSN Complexes with Sos to Activate Ras-Given that the signaling activity of ITSN was independent of the EGFR ki-nase, we sought to determine the signaling pathway(s) activated by ITSN. Previous results revealed that ITSN complexed with the Ras guanine nucleotide exchange factor Sos1, and this interaction was mediated by the SH3 domains of ITSN (10). To determine the specificity of this interaction, we used GST-SH3 fusion proteins of the individual ITSN SH3 domains to purify Sos1 from cell lysates. As illustrated in Fig. 2A, SH3A, SH3C, and SH3E each bound Sos1 with apparent equal affinity. GST, SH3B, and SH3D did not complex with Sos1. Given the cooperativity between EGF and ITSN and the ability of growth factor stimulation to regulate complex formation between Sos and Grb2 through phosphorylation (19,20), we tested whether EGF stimulation of cells altered the ability of Sos1 to bind the SH3 domains of ITSN. As illustrated in Fig. 2A, EGF stimulation did not affect in vitro binding of the SH3 domains to Sos1.
Overexpression of the SH3 region of ITSN inhibited EGF activation of both MAPK and Ras suggesting that ITSN might regulate Ras activation through interaction with Sos (9,11). Based on the localization of ITSN to the plasma membrane (21) and the interaction with Sos, we hypothesized that ITSN expression would enhance Ras activation through recruitment of Sos to the plasma membrane (22,23). Indeed, overexpression of either Xenopus or mouse ITSN was sufficient to increase Ras GTP levels (Fig. 2B). The difference in activity between Xenopus ITSN and mouse ITSN is due to higher expression of mouse versus Xenopus ITSN (data not shown).
ITSN and Ras Interact on Cytoplasmic Vesicles-Previous studies reported localization of endogenous ITSN on clathrincoated pits in COS-7 cells and cytoplasmic vesicles in primary hippocampal neurons (21). YFP-tagged mouse ITSN expressed in COS-1 cells was observed in the cytosol and on distinct cytoplasmic vesicles that were clustered in the paranuclear region adjacent to Golgi but were also evident in the cell periphery (Fig. 3). This pattern was very similar to that observed for endogenous ITSN in hippocampal neurons (21). Although the subcellular distribution of these vesicles and the reported co-localization with clathrin are consistent with endosomes, they did not accumulate Texas Red-conjugated transferrin sug-gesting that they are distinct compartments (Fig. 4). However, these ITSN-positive vesicles do contain EGFR consistent with the notion that these vesicles are sites of active signaling complexes (Fig. 4). To determine if H-Ras was associated with ITSN-bearing vesicles, we co-transfected COS-1 cells with CFP-H-Ras and YFP-ITSN (Fig. 3A). H-Ras was associated with a subset of ITSN-bearing vesicles. To determine if H-Ras and ITSN interacted with each other on this compartment, we measured FRET between the two molecules. A FRET signal could be detected as sensitized emission on vesicles bearing both H-Ras and ITSN (Fig. 3A). No FRET signal was observed when YFP-ITSN was co-expressed with CFP alone (data not shown). FRET was further validated by release of donor quenching upon acceptor photobleaching (24). The FRET efficiency was calculated as 10 Ϯ 1% (mean Ϯ S.E., n ϭ 25 vesicles), within the range reported for intracellular protein- protein interactions (15). Furthermore, immunoprecipitation of ITSN from cell lysates resulted in the co-precipitation of wild type H-Ras (Fig. 3C). Thus, H-Ras and ITSN interact on a subset of ITSN-bearing cytoplasmic vesicles.
To determine if the ITSN-H-Ras association led to Ras activation, we co-expressed the Ras binding domain of Raf-1 (RBD) tagged with CFP and YFP-ITSN along with untagged H-Ras (Fig. 3B). CFP-RBD was associated with a subset of ITSNbearing vesicles indicating that at least a portion of the H-Ras on this compartment was in the GTP-bound state. Consistent with this result, we were able to co-precipitate constitutively activated H-Ras with ITSN (Fig. 4C). In contrast to CFP-H-Ras, we detected no FRET signal between CFP-RBD and YFP-ITSN.
ITSN Activation of Elk-1 Is Independent of Ras-Given that ITSN complexed with Ras and stimulated RasGTP levels, we tested whether ITSN activation of Elk-1 was dependent on Ras. Co-expression of Ras dominant-negatives along with ITSN did not block Elk-1 activation by ITSN (Fig. 5). However, EGF activation of Elk-1 was inhibited by the Ras dominant-negatives as was the cooperative activation of Elk-1 by EGF and ITSN. These results indicated that ITSN signals to Elk-1 activation through a Ras-independent mechanism.

ITSN Activation of Elk-1 Is Differentially Regulated by the JNK and p38 Pathways-The observation that ITSN activated
Elk-1 independent of Ras and the Erk MAPK pathway coupled with the fact that p38 and JNK activate Elk-1 in various cell types (25) suggested that alternative MAPK pathways were involved in ITSN signaling. To test this hypotheses, we treated ITSN-expressing cells with a pharmacological inhibitor of JNK (26). As illustrated in Fig. 6A, ITSN activation of Elk-1 was significantly impaired with JNK inhibition. Our previous studies with truncation mutants of ITSN revealed that the signaling activity of ITSN was localized to the amino-terminal EH domains (9). Treatment of EH-expressing cells with the JNK inhibitor also blocked JNK activation. Given a recent report demonstrating that the SP600125 inhibits kinases other than JNK (27), we tested whether a JNK dominant-negative (JNK-APF) (28) would also block ITSN signaling. As illustrated in Fig. 6B, stimulation of Elk-1 by either wild type ITSN (Xenopus or mouse) or the EH domains was blocked by the JNK-APF. However, expression of JNK-APF did not alter ITSN expression (data not shown). Together, these results demonstrate that ITSN activation of Elk-1 was dependent on JNK activity.
Although ITSN signaling activity was JNK-dependent, contrasting results were obtained with an inhibitor to p38. Pharmacological inhibition of the p38 pathway with SB203580 (29) led to a significant increase in ITSN activation of Elk-1 as well as an increase in the cooperativity between EGF and ITSN in activating Elk-1 (Fig. 6C). Western blot analysis of cell lysates indicated that ITSN was expressed at equivalent levels in cells treated with each inhibitor as compared with vehicle-treated cells (data not shown). These results indicated the ITSN signaling was differentially regulated by p38 and JNK.
ITSN Stimulates JNK Phosphorylation-Because ITSN activation of Elk-1 was JNK-dependent, we next tested whether ITSN activated JNK. Co-expression of either Xenopus or mouse ITSN with FLAG-tagged JNK led to JNK activation (Fig. 6D). Mouse ITSN was more potent at stimulating JNK due to its higher expression. Additionally, expression of the EH domains was sufficient to stimulate JNK activation consistent with the Elk-1 reporter assays (Fig. 6A). Expression of a constitutively activated allele of Ras, H-RasQ61L, led to slight JNK activation. However, co-expression of ITSN with activated Ras led to a dramatic induction in JNK activation. In contrast, there was little cooperativity in the activation of JNK by co-expression of ITSN with wild type Ras. These results indicated that, although ITSN did not require Ras for activation of Elk-1, ITSN cooperated with Ras-activated pathways to potentiate JNK activation.

DISCUSSION
Previous work from our laboratory demonstrated that the endocytic adaptor protein ITSN regulates mitogenic signaling pathways leading to cell growth and differentiation (9). Although ITSN synergized with EGF as well as other growth factors (e.g. hepatocyte growth factor and basic fibroblast growth factor, data not shown) to stimulate Elk-1, we found that Elk-1 activation by ITSN was independent of EGFR kinase activity. These data indicate that ITSN signaling does not result merely from inhibition of EGFR endocytosis thereby prolonging signaling from the receptor at the cell surface. Furthermore, the inability of PD153035 to affect ITSN signaling suggests that ITSN does not stimulate an autocrine loop leading to Elk-1 activation, at least not one involving EGFR ligands Lysates were probed with antibodies to Ras or to HA-tagged ITSN (middle and bottom panels) to confirm expression levels. (30). Rather, our findings suggest that ITSN directly stimulates biochemical pathways that activate Elk-1 independent of EGFR activity (see below).
Recent studies have demonstrated an interaction between ITSN and the Ras guanine nucleotide exchange factor Sos1 suggesting an involvement in Ras activation (10,11). Indeed, overexpression of the SH3 domains of ITSN blocked EGF activation of both Ras and MAPK (9,11). Our current results demonstrate that ITSN stimulates RasGTP levels and physically complexes with Ras in vivo on intracellular vesicles. This interaction is likely mediated through Sos binding to the SH3 domains of ITSN, particularly SH3A, -C, and -E, although we cannot rule out the potential involvement of additional Ras exchange factors in this process. In contrast, however, our initial studies revealed that ITSN activated Elk-1 in an MEK-and MAPK-independent manner suggesting a lack of Ras involvement in ITSN signaling (9). Indeed several lines of evidence suggest that Ras is not involved in ITSN signaling to Elk-1. First, the EH domains are potent activators of Elk-1 yet do not associate with Sos1. Second, pharmacological inhibitors of MEK1/2, a prominent Ras target, do not block Elk-1 activation by ITSN (9). Finally, ITSN activation of Elk-1 is unaffected by expression of Ras dominant-negative proteins, although these proteins potently inhibited EGF activation of Elk-1 as well as abolished the synergy between EGF and ITSN. Together these data indicate that ITSN signals to Elk-1 through a Ras-independent pathway.
Our studies indicate that ITSN signals through JNK to stimulate Elk-1. Pharmacological inhibition of JNK or expression of a JNK dominant-negative greatly attenuated activation of Elk-1 by ITSN. Conversely, expression of ITSN stimulated JNK activation. Although ITSN did not depend on Ras for activation of Elk-1, ITSN cooperated with constitutively activated Ras, but not wild type Ras, to potentiate JNK activation. We interpret this result to mean that, although ITSN activates Ras on vesicles, this pool of Ras is not capable of coupling to JNK (Fig. 7). Rather, we speculate that ITSN cooperates with a different pool of Ras. Consistent with this notion, Ras activation of JNK is spatially restricted to certain endomembrane compartments (12). Given the ability of ITSN and EGFR to cooperate in activating Elk-1, our results suggest that ITSN cooperates with a signal from Ras at the plasma membrane, or possibly another endomembrane compartment. Furthermore, the lack of Erk activation with ITSN overexpression suggests that this pool of ITSN-activated Ras is not capable of coupling to the MEK-Erk pathway. These observations suggest that ITSN is either inhibiting Ras signaling or altering the ability of Ras to select effectors that lead to activation of Erk or JNK. Given that ITSN promotes Ras activation, we favor the later possibility. However, formal proof that ITSN-activated Ras

FIG. 5. Intersectin activation of Elk-1 is independent of Ras.
HEK 293T cells were co-transfected with the indicated expression constructs along with the Elk reporter plasmids as in Fig. 1. Following stimulation with EGF (100 ng/ml) for 5 h, cells were lysed and then analyzed for luciferase activity. The data are expressed as the mean Ϯ S.E. from five independent experiments (three for mITSN data) performed in duplicate. couples to distinct biochemical pathways awaits experimental support.
Our findings also reveal a dynamic regulation of ITSN by MAPK pathways. Whereas JNK is an important mediator of ITSN function, p38 negatively regulates ITSN signaling and Erk appears to play no role in ITSN signaling. Other groups have reported opposing actions of p38 and JNK in Ras signaling and transformation (31,32) consistent with our observations for ITSN regulation. The ability of ITSN to stimulate JNK was not due to nonspecific induction of a stress response due to protein overexpression. Whereas ITSN overexpression led to pronounced JNK activation, Ras overexpression had only a modest effect, although both were expressed quite well in 293T cells. This point is clearly illustrated by the difference in synergy between ITSN and activated Ras versus wild type Ras (Fig. 6D). Furthermore, both JNK and p38 are stress-activated kinases yet have opposing roles in ITSN signaling. We do not currently know the mechanism behind p38 inhibition of ITSN signaling. However, one possibility is that p38 phosphorylates ITSN thereby disrupting ITSN signaling complexes. Further work will be necessary to determine the mechanism by which p38 regulates ITSN.
Given that ITSN activated Ras and that Ras potently activates MEK and Erk, how can we reconcile the fact that MEK and Erk are not involved in ITSN signaling? We have recently demonstrated that, in addition to the plasma membrane, Ras proteins are localized to distinct endomembrane compartments where they stimulate different signaling pathways at each site (12,33). Coupled with our findings, it is intriguing to speculate that ITSN may promote Ras activation at distinct subcellular compartments, particularly intracellular vesicles, and that activation of Ras on these vesicles leads to activation of novel signaling pathways due to spatial restrictions of these pathways. Alternatively, given its scaffolding properties, ITSN may promote the interaction of distinct effectors with Ras thereby leading to the selective activation of non-ERK pathways. Such a model is reminiscent of the regulation of yeast MAPK pathways by scaffolding proteins (34,35). Although not mutually exclusive, these two possibilities provide possible mechanisms by which ITSN may promote the selective activation of signal- for four independent experiments performed in duplicate. C, inhibition of JNK or p38 MAPKs have differential effects on ITSN signaling to Elk-1. Experiments were performed as in A except cells were serum-starved overnight in the presence of either vehicle control (Me 2 SO), JNK inhibitor (SP600125, 20 M), or p38 inhibitor (SB203580, 10 M). The data are expressed as the mean Ϯ S.E. for three independent experiments performed in duplicate. D, ITSN activates JNK and cooperates with constitutively activated Ras to potentiate JNK activation. HEK cells were co-transfected with the indicated expression constructs along with a FLAG-epitope-tagged JNK expression construct. Following overnight incubation in serum-free media, cells were lysed and JNK was immunoprecipitated with an anti-FLAG antibody. Immunoprecipitates were then analyzed for activated JNK using the phosphospecific JNK antibody (pJNK) or total JNK (FLAG-JNK). The results shown are representative of at least four independent experiments. ing pathways. Our FRET analysis supports the latter possibility. We detected significant co-localization of YFP-ITSN with both CFP-H-Ras and CFP-RBD on intracellular vesicles particularly in the paranuclear region (Fig. 3); however, these structures do not appear to correspond to ER, a site at which Ras selectively activates JNK (12). Furthermore, a specific FRET signal between CFP-Ras and YFP-ITSN was present on a subset of these vesicles suggesting that ITSN promotes the activation of Ras at these vesicles. The lack of FRET between YFP-ITSN and CFP-RBD may be due to steric constraints that prevent the interaction of the two fluorescent proteins. Alternatively, the lack of FRET may indicate that ITSN preferentially interacts with Ras-GDP versus Ras-GTP. However, both wild type and constitutively activated Ras co-precipitated with ITSN, supporting the notion that ITSN complexes with activated Ras in vivo (Fig. 3C). Additional experiments will explore these possibilities.
Although our data do not address whether JNK activation is important for the role of ITSN in endocytosis, a number of studies have implicated JNK in protein trafficking. The endocytic adaptor protein endophilin regulates JNK activation through binding the germinal center kinase-like kinase through its SH3 domains (36). Bile acid-induced trafficking of CD95 in rat hepatocytes requires JNK (37). ␤-Arrestin 2, which participates in the desensitization of numerous G protein-coupled receptors, also acts as a scaffold for the assembly of an MAPK module resulting in G protein-coupled receptor-regulated activation of JNK3 (38). Finally, constitutively activated forms of Rac and Rho, which are potent activators of JNK in a number of cell types, inhibit receptor-mediated endocytosis (39). Thus, JNK may function both in activating transcription factors as well as regulating protein trafficking.
Several observations suggest a physiological link between ITSN and JNK. Both ITSN and JNK3 have overlapping patterns of expression, particularly within the hippocampus of the brain (40,41). Targeted disruption of JNK3 revealed an important role for this kinase in excitotoxic neuronal apoptosis following seizures (42). Interestingly, ITSN expression is elevated in the brains of Down syndrome patients, consistent with the localization of ITSN to the Down syndrome region of chromo-some 21 (43). Given the increased incidence of seizure activity in Down syndrome patients (44), elevated ITSN levels may lead to elevated JNK activity resulting in an increase in neuronal apoptosis following seizures thereby contributing to increased neurodegeneration and cognitive impairment associated with this syndrome.
We and others (9,11,45), have suggested that ITSN may serve as a scaffold to assemble a heteromeric complex of proteins that coordinately regulates GTPase cycles within cells. Our current results support this hypothesis. Both ITSN and ITSN-L complex with Sos1 through their SH3 domains, and our results demonstrate that ITSN does indeed activate Ras in vivo. In addition, ITSN-L specifically activates Cdc42 (6, 7). Finally, two lines of evidence suggest a potential role of ITSN in Rac regulation as well. The presence of a Rac guanine nucleotide exchange factor domain in Sos1 (46) suggests that ITSN may alter the Rac exchange activity of Sos1. Furthermore, ITSN inhibits the activity of CdGAP, a Cdc42 and Rac GTPaseactivating protein suggesting that ITSN may stimulate Rac and Cdc42 GTP levels through inhibition of CdGAP activity (47). Taken together, these findings support the notion that ITSN proteins are a focal point for the coordinated regulation of the Ras superfamily of GTPases as well as their effector proteins.
In addition to GTPase regulation, ITSN couples to MAPK signaling pathways and the endocytic machinery. Thus, ITSN represents a nexus for the regulation of multiple biochemical pathways important for normal cellular physiology. Dysregulation of ITSN may therefore contribute to the pathogenesis of diseases that are associated with disruption of these processes such as cancer and neurodegeneration.