Ras proteins transmit signals from receptor tyrosine kinases to intracellular signalling networks. Activation of Ras is tightly regulated by guanine nucleotide-exchange factors that catalyse GDP release and allow GTP binding to switch Ras into an active conformation. Ras–GTP then recruits effector proteins to the plasma membrane or Golgi for activation. The canonical pathway for plasma membrane Ras activation since 1993 describes the recruitment of cytosolic Grb2–Sos complexes to activated tyrosine kinase receptors on the plasma membrane1. The focus has, therefore, always been on the importance of protein–protein interactions to colocalize Sos with its substrate, Ras–GDP. On page 706 of this issue, Zhao et al.2 now provide compelling evidence that in fact a lipid–protein interaction with phosphatidic acid is what drives Sos plasma-membrane recruitment. In a complementary study on page 713, Mor et al.3 elegantly demonstrate that the spatial pattern of H- and N-ras activation in T-cells, which occurs predominantly on the Golgi, can be remodelled to include plasma membrane Ras. The core pathway again involves phosphatidic acid generation, but the molecular mechanism and target Ras exchange factor, in this case RasGRP1, are different. The studies prompt intriguing new possibilities for the spatio-temporal control of Ras activation.

Informed by homology with a phosphatidic acid-binding site in p47phox, a NADPH oxidase component, Zhao et al. demonstrate that the PH domain of Sos binds phosphatidic acid with high affinity2. The newly identified phosphatidic acid-binding site does not overlap with that of PtdInsP2, another phospholipid engaged by the Sos PH domain. In a striking set of experiments, the authors show that a carboxy-terminally truncated Sos protein (SosΔC), which is unable to bind Grb2, is recruited normally to the plasma membrane in response to EGF receptor (EGFR) activation and activates Ras2. In contrast, a Sos protein with PH domain mutations that abrogate phosphatidic acid-binding is not recruited to the plasma membrane and does not stimulate Ras GTP-loading. These results clearly demonstrate a primary role for phosphatidic acid, not Grb2–EGFR as the critical plasma membrane anchor of Sos that mediates Ras activation. The source of EGFR-generated phosphatidic acid is the enzyme PLD2. Ectopic expression of PLD2 is sufficient to recruit Sos or SosΔC to the plasma membrane and to activate Ras, whereas knockdown of endogenous PLD2, has no effect on EGFR activation, but completely blocks EGF-stimulated Sos recruitment and Ras activation2. Taken together, these results show unequivocally that PLD2 operates between EGFR and Sos and that phosphatidic acid is both necessary and sufficient for Sos plasma-membrane recruitment (Fig. 1).

Figure 1: A role for PLD2 upstream of Ras activation.
figure 1

(a) The classical model for Sos recruitment to the plasma membrane by activated EGFR, mediated by Grb2 (left), is compared with a revised version that requires recruitment and activation of PLD2 (right). Phosphatidic acid generated by PLD2 binds to the PH domain of Sos, colocalizing Sos with the EGFR and Ras–GDP. The involvement of Grb2 in the recruitment process, or in allowing potential complex formation between PLD2 and Sos, is speculative2. (b) Activation of TCR results in Zap70-mediated phosphorylation of LAT, a scaffold that provides binding sites for the recruitment of PLCγ and Grb2–Sos. Hydrolysis of PtdInsP2 generates a calcium signal through InsP3 that drives CAPRI to the plasma membrane, preventing Ras activation by Sos. The same calcium signal also drives RasGRP1 to the Golgi, where H- and N-ras are activated. When LFA-1 is coactivated, PLD2 and PAP function in concert to increase plasma-membrane DAG levels sufficiently to recruit RasGRP1 to the plasma membrane where Ras is now activated. The spatial relationship and signalling crosstalk between the TCR and LFA-1 that is required to coordinate this activation of Ras on both plasma membrane and Golgi remains to be elucidated3. PC, phosphatidylcholine.

Full size image

The results are intriguing and provoke a series of important questions about how growth factors operate. A great deal of work has elucidated the role of protein–protein interactions in assembling signalling pathways, as exemplified by the multiple phosphotyrosine dependent interactions characterized on tyrosine-kinase growth-factor receptors. However, signal transduction pathways are spatially organized on the plasma membrane, where lipid–lipid-based sorting and lipid–protein-based sorting drives the formation of signalling domains4,5,6. EGFR now provides a nice example, where the recruitment and activation of PLD2 remodels the lipid environment of the receptor by hydrolysing membrane phosphatidylcholine to phosphatidic acid, and so uses a lipid–protein interaction to anchor Sos to the plasma membrane. Two important questions arise from the study: first, why did the initial invertebrate genetic experiments identify Grb2 homologues Sem5 and Drk upstream of Let-60 and D-Ras activation?; second, does the Grb2 interaction between Sos and EGFR ever have a physiological role?

The answer to the first question may be that tyrosine-phosphorylated PLD2 binds to the SH2 domains of Grb2 and can be recovered in EGFR–Sos complexes after EGF stimulation7. Grb2 could, therefore, still be a critical mediator of growth factor-receptor signalling, but because of a requirement to recruit PLD2 rather than Sos. The Caenorhabditis elegans and Drosophila melanogaster genomes encode phospholipase D homologues, but these have not been biologically characterized. Alternatively, the superimposition of PLD2 control over Sos may simply have been a recent evolutionary event. With regard to the second question, it remains formally possible that Grb2-mediated and phosphatidic acid-mediated Sos recruitment could be used alternatively during different stages of development, on different time scales, or under different growth states. Grb2 binding could also help stabilize Sos in an active conformation on the plasma membrane, or be involved in the nanoscale tuning of plasma membrane localization by co-recruiting Sos and PLD2 to precisely the same sites. In this context, it is interesting to speculate that co-delivery of PLD2 and Sos has the potential to generate an autonomous signalling domain. Recent work has shown that Ras proteins operate in transient nanoclusters on the plasma membrane8. PLD2 could, therefore, modify the lipid content of a nanocluster enriching it in phosphatidic acid, thus retaining Sos and maintaining Ras in a GTP conformation. Moreover, the key Ras effector, Raf, can also be recruited and stabilized on the plasma membrane through binding to phosphatidic acid9,10. Therefore, the existence of nanoscale domains that have both Ras effectors and activators colocalized by a common lipid-based anchoring mechanism is an intriguing proposition that flows from this work.

In the second study, Mor et al.3 show that PLD2 also operates upstream of a second Ras-GEF, RasGRP1 in T-lymphocytes. T-cell receptor (TCR) crosslinking in Jurkat T cells and in primary mouse T-cells results in rapid activation of H- and N-ras on the Golgi. The signalling cascade is complex11: TCR activation through recruitment and activation of the cytosolic tyrosine kinase Zap70 recruits Src and PLCγ by phosphorylating the scaffold protein LAT. Activated PLCγ hydrolyses plasma membrane PtdInsP2 to generate DAG and InsP3. The latter second messenger, by stimulating calcium release, drives RasGRP1 to the Golgi where it activates H- and N-ras11. Although LAT also recruits Grb2–Sos, the same calcium signal suppresses any Ras activation on the plasma membrane by stimulating recruitment of CAPRI, a Ras-GAP12. This profile of Ras activation is substantially modified if TCR activation is accompanied by activation of the integrin LFA-1. The authors demonstrate that LFA-1 crosslinking stimulates the recruitment of RasGRP1 to the plasma membrane and rapid, coincident activation of H- and N-ras on both the plasma membrane and Golgi, and delayed K-ras activation on the plasma membrane3. The effect is only observed if LFA-1 is activated simultaneously with the TCR and not if LFA-1 is crosslinked in isolation. Although the activation of Ras on the Golgi continues to be mediated through PLCγ, this is not the case on the plasma membrane3. Using a fluorescent marker for DAG, the authors show that plasma membrane DAG levels, which remain low after TCR crosslinking, are substantially elevated after LFA-1 coactivation. Through an elegant set of knockdown and inhibitor experiments, it is firmly established that LFA-1 crosslinking activates PLD2, generating phosphatidic acid, which is converted to DAG by the enzyme phosphatidic-acid phosphatase (PAP). Generating DAG through phosphatidic acid does not lead to coincident calcium release, which presumably is why there is no further recruitment of CAPRI, allowing plasma membrane Ras-GTP levels to rise (Fig. 1). Plasma membrane-recruited RasGRP1 predominantly activates H- and N-ras, prompting an interesting question as to how the nanoscale colocalization of PLD2 and PAP with specific Ras isoforms is achieved in lymphocytes.

The clear identification of PLD2 as a new player upstream of Ras activation, in two quite distinct signalling pathways, is an exciting discovery. At a mechanistic level, it will be important to define in detail how PLD2 is, in turn, regulated. In particular, how LFA-1 and EGFR activation of PLD2 integrates with known PLD2 regulators such as Rho GTPases and Ral A. As a start in this direction, Mor et al.3 show that constitutively active Rac1 (a Rho GTPase) can substitute for LFA-1 costimulation in activating Ras on the plasma membrane. The broader biological significance of both studies is substantial; for example, PLD2 cooperates with Sos and Ras in cell transformation assays2 and increased PLD activity has been reported in human tumours13. Cell-fate decisions in thymocytes are determined by the spatial pattern of Ras activation, positive- and negative-selecting ligands are associated with Golgi and plasma-membrane activation, respectively14. The concept that costimulation of the TCR with other surface receptors can spatially shape patterns of Ras activation3 opens up new avenues to explore the biochemical mechanism underlying this critical immunological process. Exciting times are clearly ahead as previously unappreciated phosphatidic acid takes on new and increasingly important roles, and GTPase network diagrams are reassessed and rewired.