Rac-dependent Anti-apoptotic Signaling by the Insulin Receptor Cytoplasmic Domain*

Mutations in the cytoplasmic domain of the insulin receptor that block the ability of the receptor to stimulate glucose uptake do not block the receptor’s ability to inhibit apoptosis (Boehm, J. E., Chaika, O. V., and Lewis, R. E. (1998) J. Biol. Chem. 273, 7169–7176). To characterize this survival pathway we used a chimeric receptor (CSF1R/IR) consisting of the ligand-binding domain of the colony-stimulating factor-1 receptor spliced to the cytoplasmic domain of the insulin receptor and a mutated version of the chimeric receptor containing a 12-amino acid deletion of the juxtamembrane domain (CSF1R/IRΔ960). In addition to the inhibition of apoptosis, activation of either the CSF1R/IR or the CSF1R/IRΔ960 rapidly induced membrane ruffling in Rat1 fibroblasts. The small GTPase Rac mediates membrane ruffling. Activated and dominant-inhibitory mutants of Rac and other small GTPases were expressed in Rat1 fibroblasts to examine a potential link between the intracellular pathways that induce membrane ruffling and promote cell survival. The anti-apoptotic action of the CSF1R/IRΔ960 was reversed by dominant-inhibitory RacN17, but not by RasN17 or Cdc42N17. Activated RacV12, but not RasD12 or Cdc42V12, promoted cell survival in the absence of insulin. These data implicate Rac as a mediator of an unique anti-apoptotic signaling pathway activated by the insulin receptor cytoplasmic domain.

The cellular receptor for insulin controls diverse intracellular pathways that regulate metabolism, growth, and development. One conventional approach for identifying the molecular components that constitute these different pathways has been the selective mutagenesis of the receptor's cytoplasmic domain in an effort to discretely dissociate one biological action of insulin from another (1-7). These observations have led to the suggestion that different regions of the insulin receptor cytoplasmic domain play distinct roles in modulating the biological effects of insulin.
The insulin and IGF I 1 receptors are potent inhibitors of apoptosis (6,8,9). Recent studies with a CSF1R/IR suggest that the insulin receptor cytoplasmic domain activates intracellular signaling pathways that are independent of its ability to phosphorylate IRS proteins and Shc (5,6). The mutated CSF1R/IR⌬960 lacks the ability to phosphorylate the receptor substrates IRS-1 and Shc. Coincident with the loss of receptor substrate phosphorylation, the ⌬960 deletion also blocks the activation of ERK MAP kinases, the activation of PI 3-kinase associated with IRS-1, and the stimulation of glucose transport. However, the ⌬960 deletion does not prevent the receptor from inhibiting apoptosis (6). Previous studies of anti-apoptotic signaling by growth factor receptors have identified PI 3-kinase and the serine/threonine kinase Akt as part of the mechanism that promotes cell survival (10 -13). The insulin and IGF I receptors activate PI 3-kinase and, subsequently, Akt through their ability to phosphorylate IRS proteins at sites that facilitate interaction of these receptor substrates with the noncatalytic subunit of PI 3-kinase, p85 (14 -16). These observations suggested that alternative pathways could contribute to the insulin receptor's anti-apoptotic activity.
To identify additional mediators of anti-apoptotic signaling activated by the insulin receptor cytoplasmic domain, other biological effects of the CSF1R/IR⌬960 were investigated. We observed that the CSF1R/IR⌬960 rapidly induced membrane ruffling in a manner dependent on the small GTPase Rac. A contribution of Rac to anti-apoptotic signaling by the chimeric receptor was indicated by the observation that dominant-negative Rac N17 blocked survival signals from the CSF1R/IR⌬960. Furthermore, activated Rac V12 , but not activated Ras D12 or Cdc42 V12 , was sufficient to promote survival in the absence of growth factor stimulation. These results implicate Rac as a mediator of an additional anti-apoptotic signaling pathway activated by the insulin receptor cytoplasmic domain in a manner independent of IRS phosphorylation. transfected with the CSF1R/IR or the CSF1R/IR⌬960 with Lipo-fectAMINE (Life Technologies, Inc.). Stable lines expressing each chimeric receptor were isolated as described previously (5) by selection with G418 followed by fluorescence-activated cell sorting using an antibody to the ligand binding domain of the human CSF-1 receptor. For the expression of activated and dominant-negative forms of Rac, Ras, and Cdc42, Rat1 mycER cells were infected at a multiplicity of infection of at least 5 with ecotropic retroviruses encoding each Myc-tagged cDNA and selected with G418 (400 g/ml). Analysis of Akt-Cells expressing the CSF1R/IR or the CSF1R/ IR⌬960 (5 ϫ 10 5 /35-mm dish) were treated with or without 10 nM CSF-1 or 100 nM insulin for 10 min, lysed, and the phosphorylation state of Akt was determined by Western blot with PhosphoPlus Akt (Ser 473 ) antibodies (New England Biolabs). Akt kinase activity was determined as described previously (19) by the phosphorylation of histone H2B in immunoprecipitates of Akt from treated and untreated cells.
Quantification of Apoptosis-Apoptosis in cells stably expressing activated or dominant-negative forms of Ras, Rac 1, or Cdc42 was determined by time-lapse video microscopy (8) and by determining the number of apoptotic cells on a specific grid at 0, 10, and 20 h posttreatment. Fields of at least 300 cells were used to quantify apoptosis.
Apoptosis in cells microinjected with Rac N17 was determined by co-injection with 20 ng/l pEFGFP. Cells were treated with 100 nM tamoxifen, 10 nM CSF-1, or 1 M insulin as indicated, and cells expressing GFP were observed every 2 h for a total of 20 h. The amount of apoptosis among injected cells was quantified. Cell death caused by microinjection was determined by the injection of pEFGFP alone and subtracted from the amount of death induced by each experimental condition. Microinjected cells expressing GFP were evaluated within the microscopic field for each condition. Each condition was repeated at least five times.

RESULTS AND DISCUSSION
Previous studies have shown that mutations in the CSF1R/IR can dissociate its ability to phosphorylate important intracellular substrates from certain receptor-stimulated biological effects (5,6). A 12-amino acid deletion (⌬960) blocks the ability of the CSF1R/IR to phosphorylate IRS-1 and Shc, activate the MAP kinases ERK1 and ERK2, activate PI 3-kinase associated with IRS-1, and stimulate glucose transport (5,6). Despite the uncoupling of these intracellular effectors from the receptor, the CSF1R/IR⌬960 still retains its ability to stimulate adipocyte differentiation (5) and protect cells against apoptosis (6). However, inhibition of apoptosis and induction of differentiation are assayed over much longer periods of time (hours to days) than the amount of time (minutes) required to detect biological events that are inhibited by the ⌬960 mutation. This fact prevented us from excluding the possibility that small amounts of IRS or Shc phosphorylation could result from prolonged activation of the CSF1R/IR⌬960 to inhibit apoptosis or stimulate differentiation. We, therefore, looked for biological effects of insulin that could be mimicked by activation of the CSF1R/IR and occurred over the same rapid time course as the phosphorylation of insulin receptor substrates and the stimulation of glucose uptake.
Insulin and insulin-like growth factors rapidly induce membrane ruffling (lamellipodia) in cells (21,22). To investigate the effect of the ⌬960 deletion on the ability of the chimeric receptor to stimulate membrane ruffling, Rat1mycER fibroblasts stably expressing the CSF1R/IR, the CSF1R/IR⌬960, or transfected with the expression vector alone were treated with or without 10 nM CSF-1 or 100 nM insulin for 2 min. The cells were fixed immediately, and polymerized actin was detected with TRITC-labeled phalloidin. Fluorescence microscopy revealed FIG. 1. Membrane ruffling in cells expressing the CSF1R/IR or the CSF1R/IR⌬960. Rat1mycER cells stably transfected with the expression vector alone, or expressing the CSF1R/IR or the CSF1R/ IR⌬960 were incubated in serum-free media for 48 h, treated with or without 100 nM insulin or 10 nM CSF-1 for 2 min, and fixed immediately with 2.5% paraformaldehyde. Actin microfilaments were detected with TRITC-phalloidin. Lamellipodia are indicated by the white arrows.

FIG. 2. Phosphorylation and activation of Akt in cells express-
ing the CSF1R/IR or CSF1R/IR⌬960. Rat1mycER cells expressing the CSF1R/IR or the CSF1R/IR⌬960 were left untreated or treated with 10 nM CSF-1 or 100 nM insulin for 10 min at 37°C. A, activated Akt was immunoprecipitated from the cell lysates using phospho-specific Akt antibodies, separated by electrophoresis, and blotted to nitrocellulose. Activated Akt was detected by probing the blots with the phosphospecific Akt antibodies. B, the Akt expression in the samples from A was determined by immunoprecipitation and probing a portion of the cell lysate with antibodies to Akt. C, the activation of Akt was determined by examining the phosphorylation of histone H2B by Akt immunoprecipitated from cells expressing the CSF1R/IR or the CSF1R/IR⌬960. Cells were treated with CSF-1 or insulin as indicated. D, the Akt expression in the samples from C was determined by immunoprecipitating and probing a portion of the cell lysate with antibodies to Akt. that CSF-1 and insulin rapidly induced lamellipodia in cells expressing either the CSF1R/IR or CSF1R/IR⌬960 (Fig. 1). However, in control cells only insulin was able to induce membrane ruffling (Fig. 1). The ability of the CSF1R/IR⌬960 to rapidly stimulate membrane ruffling suggested that the induction of this cytoskeletal rearrangement was not dependent upon IRS and Shc phosphorylation. Lamellipodia formation is dependent upon the activation of the small GTPase Rac (23). To investigate the role of Rac in lamellipodia formation by the CSF1R/IR⌬960, dominant-inhibitory Rac N17 was microinjected into Rat1mycER fibroblasts expressing the CSF1R/IR⌬960, and the cells were treated with CSF-1. Expression of Rac N17 blocked the ability of CSF-1 to induce membrane ruffling in microinjected cells, but not in adjacent, uninjected cells (data not shown).
We observed previously that the ⌬960 mutation blocks the ability of the CSF1R/IR to stimulate PI 3-kinase activity associated with tyrosine phosphorylated proteins, including IRS-1 (6). Phosphorylation of the 3Ј position of PI is thought to be a key event in the activation of guanine nucleotide exchange factors for Rac (24,25). Binding of PI(3,4,5)P 3 to the pleckstrin homology domain of Akt is also critical to the activation of the serine/threonine kinase (26 -29). To confirm that the CSF1R/ IR⌬960 was not activating known PI 3-kinase signaling pathways, the ability of cells expressing the CSF1R/IR or the CSF1R/IR⌬960 to activate Akt was examined. Insulin (100 nM) stimulated the phosphorylation of Akt in cells expressing the CSF1R/IR or the CSF1R/IR⌬960, as determined by Western blotting with an antibody specific for Akt phosphorylated on Ser 473 . However, only cells expressing the intact CSF1R/IR were able to stimulate phosphorylation in response to CSF-1 treatment (Fig. 2A). The level of Akt expressed in each treatment group was similar, suggesting that the ⌬960 mutation was responsible for the inability of CSF-1 to stimulate Akt phosphorylation in cells expressing the chimeric receptor (Fig.  2B). Similar results were obtained when the kinase activity of Akt was analyzed. CSF-1 stimulated the kinase activity of Akt immunoprecipitated from cells expressing the intact CSF1R/IR but not from cells expressing the CSF1R/IR⌬960 (Fig. 2C). These data indicate the activation of Akt by the insulin receptor kinase is dependent upon IRS-associated PI 3-kinase activity but that the rapid, Rac-dependent induction of lamellipodia by the CSF1R/IR⌬960 is independent of the activation of PI 3-kinase. Alternatively, a second, distinct PI 3-kinase pathway may also be stimulated independent of IRS proteins by the insulin receptor kinase to activate Rac. PI 3-kinase and Akt are known mediators of growth factorstimulated anti-apoptotic signaling pathways (10 -13). CSF-1 inhibited apoptosis (6) and stimulated Rac-dependent membrane ruffling (Fig. 1), but was unable to activate PI 3-kinase (6) or Akt (Fig. 2) in cells expressing the CSF1R/IR⌬960. These observations raise the possibility that Rac mediates anti-apoptotic signaling by the insulin receptor cytoplasmic domain. To test the role of Rac in anti-apoptotic signaling by insulin, constitutively active or dominant-inhibitory forms of small GTPases were stably expressed in Rat1mycER fibroblasts and tested for their ability to inhibit apoptosis. Rat1mycER cells alone or expressing the different small GTPases were treated with 100 nM tamoxifen to induce apoptosis (18) and with or without 1 M insulin for 20 h. Cells expressing Rac V12 and treated with 100 nM tamoxifen showed a significantly reduced level of apoptosis when compared with control Rat1mycER cells treated with tamoxifen (Fig. 3A). Furthermore, insulin treatment did not enhance the survival of tamoxifen-treated cells expressing Rac V12 . The protective role of Rac V12 in tamoxifentreated Rat1mycER fibroblasts was further demonstrated by time-lapse video microscopy (Fig. 3B). In contrast, the expression of activated Cdc42 V12 showed no ability to protect cells from tamoxifen-induced apoptosis and had no effect on insulin's anti-apoptotic activity, suggesting that Cdc42 is not involved in an insulin-mediated survival pathway (Fig. 3A). Activated Ras D12 also had no protective effect against tamoxifeninduced apoptosis, but instead inhibited the anti-apoptotic activities of insulin. The actions of activated Ras D12 are consistent with the observation that activated Ras has pro-apoptotic activity in Rat1mycER cells through its stimulation of ERK MAP kinases (13). These results suggest that Rac is distinct in its ability to promote cell survival in the presence of an apoptotic stimulus.
The expression of dominant-inhibitory Cdc42 N17 or Ras N17 in Rat1mycER cells had no effect on tamoxifen-induced apoptosis and did not alter insulin's ability to inhibit tamoxifen-induced apoptosis (Figs. 3, C and D). We were unable to create cell lines stably expressing Rac N17 . As a second approach toward determining whether Rac contributed to survival signaling, plasmids encoding dominant-inhibitory Rac N17 were microinjected into Rat1mycER fibroblasts expressing the CSF1R/IR⌬960 (Fig. 4). Apoptosis in untreated cells was 11.5%, whereas ta-moxifen treatment alone elevated the level of apoptosis to 36.3%. Tamoxifen treatment in combination with insulin or CSF-1 significantly reduced the amount of apoptosis to 18.9% and 20%, respectively. Untreated cells injected with an expression plasmid encoding Rac N17 showed a level of apoptosis comparable with untreated cells or cells treated with tamoxifen plus insulin or CSF-1. Tamoxifen increased apoptosis in cells injected with plasmids encoding Rac N17 to the same extent as control injected cells treated with tamoxifen, which demonstrated that dominant-inhibitory Rac N17 has no protective effect on the cells. However, the anti-apoptotic effects of insulin and CSF-1 were lost in cells injected with Rac N17 (Fig. 4).
The data demonstrate that receptor signaling to Rac is independent of the phosphorylation of IRS-1 and Shc, as a deletion in the juxtamembrane domain prevented the chimeric CSF1R/IR (6) and the intact insulin receptor (1, 30 -32) from interacting with and phosphorylating these important signaling intermediates. Ras (33), Cdc42, (34) and PI 3-kinase (35) are considered to be activators of Rac. However, CSF1R/IR activation of Ras-regulated pathways is inhibited by the ⌬960 mutation (6), and neither dominant-inhibitory Ras N17 nor Cdc42 N17 blocked the ability of the CSF1R/IR⌬960 to induce membrane ruffling (data not shown) or survival (Fig. 3). PI 3-kinase associated with IRS-1 is activated by the CSF1R/IR, but is not activated by the CSF1R/IR⌬960 (6). These previous observations and the data presented here suggest the existence of a novel intracellular signaling pathway used by the insulin receptor kinase to induce membrane ruffling and to transmit intracellular signals promoting cell survival.
The PI 3-kinase-dependent activation of Akt and the subsequent phosphorylation of the pro-apoptotic protein Bad (12), caspase 9 (36), and the forkhead transcription factor FKHRL1 (37) have been identified previously as elements of a growth factor-mediated survival mechanism. The data presented here suggest that growth factors with the ability to activate the small GTPase Rac may stimulate additional pathways to inhibit apoptosis. Rac is an effector of multiple signaling pathways (38), any one of which may contribute to the inhibition of apoptosis. Activated Rac promotes survival in BaF3 cells in a manner that is inhibited by the p38 kinase inhibitor SB203580 (39). Rac can also activate the NADPH burst oxidase (40), which, in turn, can activate the anti-apoptotic transcription factor NF␤ (41,42). This latter possibility suggests an additional survival pathway dependent on new gene transcription. In this regard, it is interesting to note that a novel antiapoptotic signaling pathway was recently described for the IGF I receptor that is sensitive to the RNA polymerase II inhibitor ␣-amanitin (19). The identification of Rac as a transitional element in these biological events should provide a valuable point of reference for the identification and ordering of additional signaling intermediates that regulate cytoskeletal changes and survival by the insulin receptor kinase.

FIG. 4. Apoptosis in cells microinjected with Rac N17 .
Rat1mycER cells expressing the CSF1R/IR⌬960 were left untreated or microinjected with DNA encoding GFP (20 ng/l) or GFP plus Rac N17 (50 ng/l) DNA. Five hours after injection, the fluorescent cells were counted, and the cells were treated with 100 nM tamoxifen, 1 M insulin, or 10 nM CSF-1 as indicated. After incubation at 37°C for 20 h the number of fluorescent cells remaining was determined and the percentage of apoptosis was calculated. The results are presented as the mean Ϯ S.D. of three or more independent experiments. Statistical significance was determined by a Student's t test; a-d, p Ͻ 0.01 in comparison to control cells treated with tamoxifen alone.