TCR-Induced Activation of Ras Proceeds at the Plasma Membrane and Requires Palmitoylation of N-Ras

T cells are versatile cells that react to varying environmental challenges with sometimes strikingly opposed biological outcomes. For example, during positive/negative selection immature T cells either survive or succumb to apoptosis depending on self-Ag recognition strength (1). Similarly, T cells that become engaged by APCs can either buildup a full immunological response or enter a hyporesponsive state known as T cell anergy, depending on intensity, duration, and other characteristics of the trigger. These and other cues conveyed by cognate Ags are translated into intracellular signals by the TCR.

Within seconds of activation, the TCR addresses a large array of signal transduction pathways that ultimately induce changes in gene expression that control cell proliferation or differentiation and cause the shaping of T cell effector functions. Among the many signaling mediators engaged, the Ras/Erk pathway is arguably one of the most important signal transducers downstream of the TCR. Recent genetic and biochemical studies have shown that diacylglycerol (DAG) liberated by TCR-activated PLCγ stimulates Ras-GTP loading via the direct binding and concomitant recruitment of the guanine nucleotide exchange factor (GEF) RasGRP1 (2, 3), whose expression is largely restricted to leukocytes and neuronal cells. In addition to the direct engagement, DAG feeds into RasGRP1 by triggering PKC-dependent phosphorylation of RasGRP1 (4). These and other studies have thus disclosed a pathway initiated by PLCγ-dependent Ca²⁺ and DAG generation, leading to RasGRP1 activation via both PKC dependent and independent means as a major pathway of Ras activation downstream of the TCR (2, 5). In addition to RasGRP1, however, T cells express other GEFs, like Sos1, Sos2, or RasGRF2, and full-blown activation of Ras is likely to involve a complex interplay of various GEFs (6, 7).

Another level of complexity is posed by the subcellular distribution of Ras. K-Ras, H-Ras, and N-Ras (collectively Ras) possess a conserved, yet distinct, C-terminal motif that targets them for posttranslational modifications at the endoplasmic reticulum and Golgi apparatus (collectively endomembranes). These include farnesylation of all Ras variants and additional palmitoylation at cysteine residues in H-Ras and N-Ras (8, 9). The latter modification is reversible and determines both membrane association strength and localization of Ras. Although farnesylated and unpalmitoylated N/H-Ras exhibits loose and reversible binding to membranes and is largely confined to endomembranes (8, 10–12), palmitoylation traps N/H-Ras on membranes, and tags Ras proteins for exocytic transport and residency at the plasma membrane (PM) (8, 13). Given the many different localities and the nodal role of Ras in TCR signaling, it is crucial to know the subcellular sites of Ras activation to understand whether spatial segregation of Ras activity plays a role in TCR signaling. In support of this notion, imaging studies have documented TCR-driven Ras-GTP formation at the PM and at the Golgi of T cells (5, 14). In particular, Ras activation in endomembranes reportedly proceeds via the PLC/DAG/RasGRP1 pathway (5). These findings support a model by which Ras activation at different subcellular locations enables T cells to diversify the signaling output of the TCR. In contrast, active PLCγ and DAG, both essential upstream activators of RasGRP1, localize exclusively to the PM of activated T cells (15, 16), arguing against RasGRP1-dependent Ras activation in endomembranes. Another aspect important to consider is that all reported Ras-GTP imaging data have been recorded in T cells engineered to overexpress Ras. Because Ras overexpression can distort finely tuned processes, such as Ras trafficking, posttranslational processing or the activation process itself (9), it is difficult to judge whether the results from Ras overexpression studies reflect the true behavior of endogenous Ras.

These considerations evidence that the unambiguous identification of the subcellular sites of Ras activation and signaling in T cells can only be accomplished by the visualization of native Ras-GTP in the absence of overexpression. In the current study, we have imaged endogenous Ras-GTP formation in life T cells using novel affinity probes with increased fluorescence and avidity to Ras-GTP. Because the segregation of N/H-Ras to PM and endomembranes is largely dictated by the palmitoylation status, we have also investigated the role of palmitoylation in Ras activation downstream of the TCR.

DGKα inhibitor R59949, TRITC-phalloidin and fatty-acid-free/endotoxin-low BSA were purchased from Sigma-Aldrich, Taufkirchen, Germany. Bodipy-TR-C₅-ceramide and Blue-CMAC (Molecular Probes, Eugene, OR) and transfection reagent DMRIE-C were from Invitrogen, Carlsbad, CA. Staphylococcal enterotoxin (SE)E, SEB, and SEA were from Toxin Technology, Sarasota, FL. GST-c-Raf-RBD protein was produced in Escherichia coli (17).

Abs used: Pan-Ras (Ab-4), Merck, Bad Soden, Germany. H-Ras (F235), N-Ras (F155), K-Ras (F234), PKCθ, GFP all from Santa Cruz Biotechnology, Santa Cruz, CA. Phospho-T202/Y204- p44/42 Erk (E10) and anti-HA were from Cell Signaling Technology, Danvers, MA. Anti-Erk1 from BD/Transduction Laboratories, Heidelberg, Germany. Anti-CD3 (UCHT-1) and anti-CD28 were purchased from BD/Pharmingen, Heidelberg, Germany. Anti-CD3 IgG (OKT-3) hybridoma was acquired from the American Type Culture Collection (Manassas, VA) cell bank. Anti-CD3 IgM (C305) and anti-CD28 IgM hybridomas were kindly provided by Arthur Weiss (Howard Hughes Medical Institute, University of California San Francisco, San Francisco, CA) and Ulrich Moebius (Morphosys AG, Heidelberg, Germany), respectively.

Jurkat T cells (German Collection of Microorganisms and Cell Cultures cell bank) and Raji B cells (American Type Culture Collection) were grown in RPMI 1640 (Biowest, Nuaillé, France) supplemented with glutamine and 10% heat-inactivated FCS in a 5% CO₂ atmosphere. For live-cell imaging Jurkat T cells were transfected with DMRIE-C, following the manufacturer’s instructions. For imaging of cotransfected T cells (Fig. 2) and for biochemical experiments cells were electroporated with the Microporator device (Invitrogen, Carlsbad, CA) in the 10 μl (imaging) or 100 μl (biochemical assays) format, respectively, using the provided parameters for Jurkat cells.

Primary human T cells were prepared from healthy donors using the pan T cell isolation kit (Miltenyi Biotech, Bergisch Gladbach, Germany) and maintained in RPMI 1640 medium containing 10% FCS, stable l-glutamine, and 1000 U/ml penicillin/streptomycin. Approval for these studies was obtained from the Ethics Committee of the Medical Faculty at the Otto-von-Guericke University, Magdeburg, Germany. Informed consent was obtained in accordance with the Declaration of Helsinki. For electroporation of cDNA constructs, human peripheral T cells (8 × 10⁶) were washed in PBS containing Ca²⁺/Mg²⁺ and resuspended in 200 μl Opti-MEM (Invitrogen). Thirty micrograms of DNA was added and after 3 min, cells were transfected by electroporation (square-wave pulse, 1000V 0.5 ms, two pulses [pulse interval 5 s]; BioRad X-cell). The cells were then added to prewarmed cell culture medium as described previously and cultured for 24 h before use. GFP-expression was evaluated by flow cytometry (FACScalibur flow cytometer and CellquestPro software [BD Bioscience, Heidelberg, Germany]).

3xHA-N-Ras and 3xHA-H-Ras in pcDNA3 were obtained from the Missouri S&T cDNA Resource Center (http://www.cdna.org). GTPases to be cloned in fusion with mCherry were excised from parental dsRed2-C1 constructs (18) by XhoI/HindIII (K-Ras, K-RasG12V), XhoI/BamHI (M-RasQ71L, TC21Q72L), or BglII/EcoRI restriction (Rap1AG12V) and subcloned into pmCherry C1 (Clontech, Mountain View, CA). pmCherry-N-Ras was subcloned accordingly by BglII/KpnI restriction using HA-N-Ras in pCMV (kind gift of Ian Prior, University of Liverpool, Liverpool, U.K.) as the parental plasmid. All other pmCherry-N-Ras and pmCherry-K-Ras mutants as well as 3xHA-N-RasC181S, 3xHA-H-RasC181/184S, and 3xHA-H-RasC181/184A in pcDNA3 were consequently generated by standard point mutagenesis approaches.

Rat RasGRP1 in fusion with mCherry was prepared by replacing EGFP in EGFP-RasGRP1 (kind gift of Isabel Merida, Centro Nacional de Biotecnologia/CSIC and Universidad Autonoma de Madrid, Madrid, Spain) with mCherry from pmCherry C2 via AgeI/BsrGI restriction cloning.

Trimeric EGFP (E3) reporter constructs were generated by adding two EGFP modules to the 5′ end of previously described E1 type reporter versions (18) (Fig. 2A). The dimeric EGFP sequence was produced by limited BsrGI digestion of a trimeric EGFPx3-plasmid (19) (kindly provided by Alison L. Barth, Carnegie Mellon University, Pittsburgh, PA) and inserted into BsrGI-cut E1-R3 or E1-R1 plasmids. The final E3-constructs thus feature exactly the same linkers and overall architecture as the parental EGFPx3 plasmid previously shown to exhibit 3-fold higher fluorescence than EGFP (19).

Live-cell imaging was carried out on a Zeiss LSM 510 axiovert confocal microscope equipped with a thermostated stage chamber (IBIDI, München, Germany). Confocal images (optical slice of ≤1 μm) were acquired using a 63× water immersion objective lens. EGFP and Cherry/Bodipy were excited with the Argon 488 nm and the HeNe 543 nm line in subsequent tracks. Emitted fluorescence was collected with a 505–550 nm bandpass and a 560 nm longpass filter, respectively. T cell preparation, including staining with Bodipy-TR-C₅-ceramide, was performed exactly as described (18). Cells were plated on poly-l-lysine-coated self-made glass-bottom 35-mm dishes and monitored for at least 5 min before stimulation. All images of a series were exported as TIF files and subjected to the same processing routine using Zeiss ZEN 2008 Light Edition software.

To quantify organelle-associated fluorescence, we have programmed an automatic cell image segmentation algorithm that yields masks for four cellular regions: PM, nucleus, Golgi, and cytosol. The Golgi compartment is identified by the fluorescence signal of the Golgi tracker Bodipy-TR-C₅-ceramide. GFP fluorescence is used to distinguish PM and nucleus from the cytosol (Supplemental Fig. 1 for a flowchart of the segmentation procedure). Based on the masks, mean fluorescence intensities were calculated for PM, Golgi, and cytoplasm. The nuclear and immediate perinuclear region were excluded from the analysis. PM/cytoplasma and Golgi/cytoplasma ratios were plotted as a function of time and presented as normalized mean values (± SEM). The segmentation algorithm was programmed in MatLab (MathWorks, Natick, MA).

T cells grown to a density of 10⁶ cells/ml were deprived of serum for 2 h in RPMI 1640 supplemented with 0.2% fatty-acid-free/endotoxin-low BSA and 50 mM HEPES pH 7.5, counted and resuspended at 10⁷ cells/ml in the same solution. Cell suspensions were kept in a warm water bath at 37°C and tubes were carefully flipped every 2–3 min. After appropriate stimulation 1 ml of the cell suspension was transferred to 1.5 ml reaction vials and quickly spun in a table top centrifuge. Medium was aspirated off and the cell pellet was lysed with 1 ml ice-cold lysis buffer (50 mM HEPES pH 7.5, 140 mM NaCl, 5 mM MgCl₂, 1 mM DTT, 1% NP-40, protease inhibitors) supplemented with 25 μg GST-RBD protein and 100 μM GDP. GDP, and GST-RBD were included at this point to quench postlytic GTP-loading and GAP-dependent Ras-bound GTP hydrolysis, respectively. Cell extracts were cleared by centrifugation and GST-RBD/Ras-GTP complexes were collected on Glutathione-Sepharose. Precipitates were washed once with 500 μl lysis buffer and processed for SDS-PAGE analysis.

Mock pulsed or superantigen (SEE, SEB, and SEA) pulsed and Blue-CMAC loaded Raji B cells were incubated for 15 or 30 min at 37°C with E3-R3(A/D), E3-R1(A/D), or E1-R3(A/D) expressing Jurkat or human primary T cells on poly-l-lysine-coated coverslips and fixed with 3.5% paraformaldehyde in PBS for 10 min. Cells were permeabilized with 0.1% Triton-X100 in PBS, blocked with 5% horse serum in PBS, and incubated with the indicated Abs (anti-GFP mAb, anti-CD3 IgG, or anti-PKCθ) and TRITC-phalloidin in combination with Cy3-conjugated secondary Abs (Dianova, Hamburg, Germany). Coverslips were mounted in Mowiol 488 and imaged with a LEICA TCS SP2 laser scanning confocal system (Leica Microsystems, Wetzlar, Germany) using a plan apochromatic oil emerging 63× objective (NA 1.4). For each experiment, a minimum of 30 conjugates were analyzed for polarization of E3-R3(A/D) or E1-R3(A/D) to the immunological synapse (IS) in T cells that featured enrichment of F-actin at the T cell/B cell interface. Figure construction of images was performed in COREL Photopaint.

As a first step we determined which Ras species are activated in response to TCR ligation. In agreement with previous reports (14, 20), cross-linking of the TCR with any of three anti-CD3 Abs caused GTP-loading of N-Ras and K-Ras (Fig. 1A). In contrast, H-Ras was not detectable in Tcell extracts (Fig. 1B) (14). This expression pattern is reminiscent of findings in other lymphoid leukemia cell lines (21) and peripheral T cells (20). Importantly, T cell costimulation via CD28, SLAM, or LFA-1 did not lead to significantly increased Ras-GTP formation as triggered by 5 μg/ml anti-CD3 (data not shown). These experiments evidenced that K-Ras and N-Ras are the relevant Ras isoforms in TCR signaling and that costimulation was dispensable for full-blown activation of Ras.

To visualize Ras-GTP, we used affinity probes derived from the Ras binding domain of the Ras-effector c-Raf (RBD). A fusion-protein of EGFP and RBD (EGFP-RBD or E1-R1, Fig. 2A) reports activation of overexpressed Ras (5, 12, 14). However, E1-R1 is incapable of visualizing endogenous Ras-GTP because probe redistribution must be near quantitative to be visible, which requires imaging cells with probe expression levels in the range of endogenous Ras levels that are arduous to detect. To accomplish visualization of endogenous Ras we have previously oligomerized RBD to generate a multivalent probe (E1-R3) with increased affinity/avidity for Ras-GTP (18). In a complementary approach to raise detection sensitivity we have now augmented probe fluorescence by appending two EGFP modules to E1-R3. The resulting construct E3-R3 features a 3-fold increase in fluorescence (19), which allows to reduce probe concentration and thus achieve higher fractional probe distribution without suffering a loss in fluorescence. Expression of E3-R3 in T cells resulted in a polypeptide of the correct size (Fig. 2B). To assess the usefulness of E3-R3 as a live-cell reporter of Ras-GTP in T cells, we studied its recruitment by Ras proteins. As shown in Fig. 2C constitutively active G12V mutants of N-Ras and K-Ras colocalized with E3-R3. Colocalization was specific because the effector loop point mutation D38A, which reduces affinity to RBD by 72-fold (22), impeded colocalization. The endomembrane resident palmitoylation mutant N-RasG12V/C181S (12) also recruited E3-R3 and the D38A mutation, again, abolished colocalization, showing that E3-R3, specifically, decorated Ras-GTP also in endomembranes. Constitutively, active Rap1A, which possesses 66-fold lower affinity for RBD (23), did not recruit E3-R3. E3-R3, however, cross-reacted with the more closely related GTPases TC21 and M-Ras.

The 3×RBD polypeptides are potent scavengers of Ras-GTP and exhibit cytotoxic effects (18). In Jurkat cells, E3-R3 expression resulted in up to 40% cell death (data not shown). To deal with probe toxicity, we have previously introduced well characterized attenuating point mutations into RBD (18, 24). Accordingly, incorporation of the tandem mutation R59A/N64D into E3-R3 resulted in a nontoxic probe, E3-R3(A/D) (Fig. 2A), that showed the same robust, effector site dependent recruitment by N-RasG12V as E3-R3 (Fig. 2D; data not shown). Colocalization was contingent on the high avidity provided by RBD oligomerization, because the monomeric version E3-R1(A/D), which possesses 100-fold lower affinity for Ras-GTP than RBD (24), did not redistribute (Fig. 2D). E3-RG3, a probe composed of triple RalGDS-RA, a domain of comparable affinity for Ras-GTP as R1(A/D) [K_Ds of 1 μM and 3.8 μM, respectively (24)] was also recruited by active Ras. In sum, E3-R3 and its attenuated derivative E3-R3(A/D) were specific, high-affinity/avidity probes for Ras-GTP, which featured 3-fold higher fluorescence than the E1-R3 precursor probes.

To visualize endogenous Ras activation, we expressed E3-R3 in T cells and used a vital dye to label the Golgi, the major site of residency of N-Ras besides the PM (8). TCR activation with either of three anti-CD3 Abs alone or in combination with anti-CD28 Abs caused rapid translocation of E3-R3 to the PM but not to the Golgi (Fig. 3A; data not shown). The same response was observed with E3-R3(A/D) (Fig. 3C; data not shown). To quantify PM and Golgi associated E3-R3 fluorescence, masks identifying PM and Golgi were created by an automated segmentation routine (bottom lane in Fig. 3A; see Supplemental Fig.1 for a further description of the segmentation process). E3-R3 association with the PM was still apparent 10 min poststimulation (Fig. 3A, 3B) even though biochemically assayed Ras-GTP accumulation had largely vanished by that time (Fig. 1). Differences in the experimental protocols are likely to account for this variation in activation time courses. For example, immobilization on poly-lysine, an assumedly inert matrix used to fix T cells to microscopy slides, prolonged Ras activation in biochemical assays (Supplemental Fig. 2). Moreover, RBD-probes facilitate Ras-GTP formation by protecting Ras against GAP action (18, 25), likely resulting in extended activation kinetics. To subject the specificity of the probes to further scrutiny, we transfected E3-R3(A/D) along with its red-fluorescent monomeric counterpart C3-R1(A/D). The latter is not recruited by Ras-GTP (18, 24) (Fig. 2D) and thus served as internal control of signal specificity. TCR ligation caused translocation of E3-R3(A/D) to the PM, whereas C3-R1(A/D) did not reallocate in the same cell (Fig. 3C), indicating that PM illumination by E3-R3(A/D) truly reflected Ras activation rather than the incidental association of RBD-probes with the PM. Because TC21 and M-Ras are negligibly or not at all activated by the TCR, respectively (26, 27), a contribution of these GTPases to probe redistribution could be excluded. We conclude that activation of endogenous K-Ras and N-Ras by the TCR occurs at the PM.

PLCγ/DAG-dependent engagement of RasGRP1 is a major pathway of Ras activation downstream of the TCR (2, 5). Diacylglycerolkinases (DGKs) convert DAG to phosphatidic acid and thus negatively regulate this pathway, a scenario well documented for DGKα and DGKζ (28, 29). To avoid missing sites of low-intensity Ras-GTP formation, we manipulated the PLC/DGK/RasGRP1 pathway to rev up Ras activation. Inhibition of DGKα with R59949 markedly increased TCR-dependent accumulation of Ras-GTP (Fig. 4A), which occurred only at the PM (Fig. 4B). As a second approach to force Ras-GTP formation, we overexpressed RasGRP1, a measure known to enhance TCR-dependent Ras activation (2). As before, TCR ligation triggered E3-R3 accumulation only at the PM (Fig. 4C), confirming that TCR-driven Ras activation takes place at the PM.

Physiological activation of the TCR proceeds at the IS, an intercellular contact area between APC and T cell. To visualize Ras activation in the context of the IS, E3-R3(A/D) redistribution was imaged upon conjugation of T cells and superantigen pulsed Raji B cells (Fig. 5A). Upon 15 min of conjugate formation Ras-GTP reporter probes redistributed to the IS in over 50% of Jurkat cells or 90% of primary T cells that had undergone efficient conjugation, as judged by the enrichment of F-actin at the T cell/APC interface (Fig. 5A, 5B). At 30 min postconjugation, this number increased to 92% in Jurkat cells and remained high in primary T cells (Fig. 5B). Accumulation of the RBD probes at endomembranes was not observed in either cell background. As opposed to E3-R3(A/D) or E1-R3(A/D), the monovalent probe E3-R1(A/D) did not redistribute at any time point of the conjugation process (data not shown). In parallel experiments, we labeled the TCR (Fig. 5A) or PKCθ (data not shown), two IS constituents that accumulate in the central core of the IS known as central supramolecular activation cluster (30). In 92% of conjugates active Ras localization did not strictly overlap with the TCR or PKCθ label but extended to the area surrounding the central supramolecular activation cluster, indicating that Ras-GTP accumulates preferentially in the peripheral supramolecular activation cluster. This pattern matches well the reported distribution of the two upstream regulatory molecules DAG and RasGRP1 in the IS (15, 28, 31). In conclusion, signals elicited by the TCR at the IS promote focalized Ras activation at the PM.

The absence of visible Ras-GTP formation in endomembranes argued against Ras signaling from that organelle, but it did not formally rule out the existence of TCR-signals that reached endomembranes and activated a small pool of Ras. We used the palmitoylation mutant N-RasC181S, which is largely confined to endomembranes (12) (Fig. 2C), as a signal amplifier to investigate if TCR signals reach that compartment. TCR-ligation lead to undistinguishable activation profiles of endogenous Ras and overexpressed 3xHA-tagged N-Ras, whereas 3xHA-N-RasC181S was not significantly activated (Fig. 6A, 6B). Of note, endogenous and heterologous Ras-GTP levels were simultaneously monitored in the same GST-RBD affinity precipitates from one and the same activated cell extracts. Considering that the transfection efficiency ranged between 30 and 40%, this meant that GTP-loading of endomembrane-located N-RasC181S was negligible in the same cells in which the TCR induced full activation of endogenous Ras, excluding T cell unresponsiveness or TCR malfunction as a cause for the lack of N-RasC181S activation. Experiments with palmitoylation mutants of H-Ras yielded essentially the same results (Supplemental Fig. 3). Taken together with the imaging data, the most straightforward interpretation of these results is that the TCR does not address Ras at endomembranes.

Imaging studies have documented that several established upstream regulators of Ras are recruited and engaged at the PM of APC-challenged T cells. Thus, PKCθ, which probably contributes to Ras activation by phosphorylating RasGRP1 (4), and RasGRP1 itself are exclusively associated to the PM of APC-triggered T cells (28, 31–34). DAG, which serves to recruit RasGRP1, and the DAG metabolizing enzyme DGKα reside at the PM and DAG further accumulates at the IS upon conjugation with APCs (15, 35, 36). Finally, active PLCγ (as evaluated by phosphorylation of Tyr783) specifically and exclusively localizes to the IS of activated T cells (16). All these findings strongly indicate that TCR-induced Ras activation, at least as mediated by the PLCγ/DAG/RasGRP1 pathway, is bound to proceed at the PM of activated T cells and, specifically, at the IS of T cells challenged with APCs. Contrary to this line of evidence, visualization of overexpressed Ras activation has documented that Ras-GTP accumulates both at the PM and at endomembranes upon TCR ligation (5, 14). Remarkably, endomembrane-activation of Ras has been proposed to be mediated precisely by the same PLCγ/DAG/RasGRP1 pathway that is presumed and predicted to act at the T cell surface based on the imaging data referred above. The images of endogenous Ras-GTP formation presented in the current manuscript evidence that endogenous Ras activation does proceed only at the PM of stimulated T cells, in accordance with the observation that all required upstream modulators reside at the PM. Our findings have far-reaching implications, because they exclude the endomembrane compartment as a signaling platform for Ras in T cells. This notion further implies that residency of Ras at endomembranes is most likely only for the purpose of posttranslational processing and clearance to the PM, a concept supported by the finding that palmitoylation mutants of Ras, which are locked on the endomembrane compartment owing to the interruption of posttranslational processing, are not engaged by the TCR (Fig. 6A). However, it is certainly impossible to rule out that other signals or extracellular cues may under certain circumstances, and possibly via other mechanisms, lead to the biologically relevant accumulation of Ras-GTP at endomembranes. Indeed, the fact that Golgi-activation of overexpressed Ras is observed in nonhematopoietic cell lines, such as COS fibroblasts (5, 12), which evidentially do not express RasGRP1 (17), indicates that mechanisms other than the DAG/RasGRP1 pathway may be in charge of spatial control of Ras activity. Recently, reallocation of GTP-loaded N/H-Ras to endomembranes following depalmitoylation at the PM has been reported to sustain Ras-GTP accumulation at the Golgi (13). However, magnitude and relevance of this process are not precisely known because all collected experimental evidence so far stems from Ras overexpression studies (11, 13, 37). Moreover, the data presented in this study evidence that this process is unlikely to operate in T cells or, at best, it will not affect a meaningful proportion of the Ras population, because any accumulation of Ras-GTP at T cell endomembranes, irrespective of the underlying mechanism, should have become visible in our imaging experiments.

An alternative explanation for the reported accumulation of Ras-GTP in the Golgi apparatus of T cells overexpressing Ras is that this phenomenon may be an incidental result of overexpression. Ras overexpression might cause overstraining of the posttranslational processing machinery of T cells leading to the aberrant accumulation of incompletely processed Ras proteins in endomembranes, and this pool of mislocalized Ras could subsequently escape normal regulation during TCR signaling. Ras overexpression alone, however, is unlikely to suffice as the cause of aberrant activation at endomembranes because overexpressed endomembrane-resident Ras is not significantly activated in response to TCR ligation (Fig. 6), at least at the 2- to 4-fold overexpression rate achieved in our experiments. We presume therefore that it is the combination of Ras overexpression and RBD-probe expression that may distort normal Ras regulation leading to the aberrant accumulation of Ras-GTP at T cell endomembranes. Indeed, Ras activation as a consequence of RBD-polypeptide expression is not unprecedented (25) and is likely to result from the ability of RBD to protect Ras from GAP action. Thus, several mechanisms for the incidental accumulation of overexpressed Ras-GTP at endomembranes can be envisaged.

In conclusion, our findings document that endogenous Ras activation by the TCR occurs only at the PM, implying that all signals transmitted via the TCR/Ras pathway originate at the T cell surface. Consistent with this conclusion, our data also reveal a previously unnoticed requirement for palmitoylation in N-Ras/H-Ras activation, which most likely reflects the role of palmitate as a PM anchor for Ras. This raises the possibility of pharmacologically blocking Ras palmitoylation as a new means of intervention in immunosuppressive therapies.

We thank Alison L. Barth, Jonathan Lindquist, Stefan Lorkowski, Isabel Merida, Ian Prior, and Jim Stone for the generous provision of reagents. We also thank Frank Böhmer, Ian Prior, and Jim Stone for helpful discussions, Michael Grün and Susanne Köthe for technical help, and Stefan Heinemann for providing access to LSM instrumentation.

Disclosures The authors have no financial conflicts of interest.