Differential Regulation of Phospholipase C-β2 Activity and Membrane Interaction by Gαq, Gβ1γ2, and Rac2*

We combined fluorescence recovery after photobleaching (FRAP) beam-size analysis with biochemical assays to investigate the mechanisms of membrane recruitment and activation of phospholipase C-β2 (PLCβ2) by G protein αq and βγ dimers. We show that activation by αq and βγ differ from activation by Rac2 and from each other. Stimulation by αq enhanced the plasma membrane association of PLCβ2, but not of PLCβ2Δ, which lacks the αq-interacting region. Although αq resembled Rac2 in increasing the contribution of exchange to the FRAP of PLCβ2 and in enhancing its membrane association, the latter effect was weaker than with Rac2. Moreover, the membrane recruitment of PLCβ2 by αq occurred by enhancing PLCβ2 association with fast-diffusing (lipid-like) membrane components, whereas stimulation by Rac2 led to interactions with slow diffusing membrane sites. On the other hand, activation by βγ shifted the FRAP of PLCβ2 and PLCβ2Δ to pure lateral diffusion 3- to 5-fold faster than lipids, suggesting surfing-like diffusion along the membrane. We propose that these different modes of PLCβ2 membrane recruitment may accommodate contrasting functional needs to hydrolyze phosphatidylinositol 4,5-bisphosphate (PtdInsP2) in localized versus dispersed populations. PLCβ2 activation by Rac2, which leads to slow lateral diffusion and much faster exchange, recruits PLCβ2 to act locally on PtdInsP2 at specific domains. Activation by αq leads to lipid-like diffusion of PLCβ2 accompanied by exchange, enabling the sampling of larger, yet limited, areas prior to dissociation. Finally, activation by βγ recruits PLCβ2 to the membrane by transient interactions, leading to fast “surfing” diffusion along the membrane, sampling large regions for dispersed PtdInsP2 populations.

Activation of PLC␤ 2 by ␣ q and related ␣ subunits requires the C-terminal region of the enzyme; mutants with deletions in this region (e.g. PLC␤ 2 ⌬, that lacks the Phe 819 -Glu 1166 segment) are resistant to stimulation by ␣ q , but undergo activation by ␤␥ and Rac/Cdc42 (7,(13)(14)(15). Recent results show that ␤␥ and Rac/Cdc42 activate PLC␤ 2 by interacting, at least in part, with different regions of the effector enzyme. Thus, although the pleckstrin homology (PH) domain of PLC␤ 2 is dispensable for activation of the enzyme by ␤␥ dimers (6), this domain also interacts with ␤␥, suggesting that the latter binds to at least two sites on the enzyme (16). In contrast, direct interaction of the PH domain with activated Rho GTPases is both necessary and sufficient for their stimulatory function (6,17,18). Functional evidence for a connection between PLC␤ 2 and Rho GTPases in cells is provided by the chemoattractant receptor system, which activates Rac/Cdc42 and PLC␤ 2 (19 -22).
The mechanisms by which heterotrimeric G proteins and Rho GTPases regulate PLC␤ isozymes are only partially understood (16,23,24), especially in live cells. Because both PLC␤ substrate(s) and stimulators are membrane-bound (16,23,24), recruitment of PLC␤ from the cytoplasm to the membrane is clearly required for effective enzyme activation. This view is consistent with the loss of PLC␤ activation following mutations that interfere with the membrane association of ␣ q or ␤␥ (25,26). PtdInsP 2 , the substrate of PLC␤ enzymes, is located at the internal plasma membrane leaflet in both dispersed and localized populations (27)(28)(29)(30)(31)(32)(33) and diffuses in the cytoplasmic leaflet of cell membranes with a lateral diffusion coefficient (D) of 0.5-1 m 2 /s (32,34). The PH domains of several proteins (e.g. PLC␦ 1 and the Dictyostelium discoideum CRAC protein) were found to undergo dynamic membrane-cytoplasm exchange along with lateral diffusion over short distances (35,36). Interestingly, we found a similar behavior for PLC␤ 2 (7), although this protein is likely to interact with the membrane through several distinct sites; these include the catalytic triosephosphate isomerase barrel, the C-terminal region, and possibly the EF hands motif and the C2 domain, as well as its PH domain, which, unlike that of PLC␦, appears to be unable to bind phosphoinositides (16,18).
We have formerly studied the interactions of green fluorescent protein (GFP)-tagged PLC␤ 2 (PLC␤ 2 -GFP) and PLC␤ 2 ⌬-GFP with the plasma membrane in live cells (7). Using FRAP beam-size analysis (37), which discriminates between recovery by lateral diffusion and exchange, we demonstrated that activation by Rac2 enhances the association of PLC␤ 2 and PLC␤ 2 ⌬ with the plasma membrane via binding to slow diffusing membrane proteins. However, the effects of the G protein subunits ␣ q and ␤␥ on the membrane interactions of PLC␤ 2 and their potential relevance to PLC␤ 2 stimulation were not explored. Here, we show that ␣ q and ␤␥ recruit PLC␤ 2 to the plasma membrane by distinct mechanisms, which differ from the mechanism employed by Rac2 and from each other. Each stimulator leads to a specific ratio between the rates of exchange and lateral diffusion characterizing the interaction of PLC␤ 2 with the membrane, and this in turn allows the enzyme to act preferentially on localized or dispersed PtdInsP 2 populations.

EXPERIMENTAL PROCEDURES
Materials and Plasmids-Murine anti-GFP antibodies were from Roche Applied Science, and peroxidase-conjugated goat anti-mouse IgG was from Sigma. The cDNAs of wild-type (wt) and G12V mutant human Rac2 (Rac2(wt) and Rac2(G12V), respectively), mouse ␣ q (wt) and ␣ q (R183C), human ␤ 1 , bovine ␥ 2 (wt) and ␥ 2 (C68S), and human PLC␤ 2 were ligated into pcDNA3.1(ϩ) or pcDNA3.1(Ϫ) (Invitrogen). Bovine and human ␥ 2 have identical amino acid sequences, and mouse ␣ q differs from human ␣ q in but one residue (S141A). There are several indications that this residue has no role in ␣ q function: (i) In mouse ␣ q , Ser 141 is in a long loop connecting helices ␣E and ␣F of the helical domain (PDB accession codes 2BCJ and 2RGN); it does not reside in any of the three switch domains or in regions known to interact with PLC␤ isozymes and is not among the residues that contact the bound guanine nucleotide and G␤␥ (38,39). (ii) Secondary structure prediction algorithms do not predict a structural difference between mouse and human ␣ q ␣E-loop-␣F region (40, results not shown). (iii) Mouse ␣ q has been successfully reconstituted with many signaling proteins from several other species, including human, for functional and structural analyses (e.g. [41][42][43]. The ␣ q (wt) and ␣ q (R183C) cDNAs were a gift from B. R. Conklin and H. R. Bourne (University of California, San Francisco, CA). The cDNAs of PLC␤ 2 and the deletion mutant PLC␤ 2 ⌬, which lacks a C-terminal region necessary for stimulation by ␣ q (Phe 819 -Glu 1166 ), were inserted in-frame with the cDNA of GFP into the EcoRI/SalI site of pEGFP-N1 (Clontech) to generate the plasmids encoding PLC␤ 2 -GFP and PLC␤ 2 ⌬-GFP (7). Plasmids (pcDNA3.1) encoding the N-terminal half (residues 1-155) of the venus fluorescent protein fused to the N terminus of human ␥ 2 (venus 1-155-␥ 2 ) or venus residues 156 -239 fused to the N terminus of human ␤ 1 (venus 156 -239-␤ 1 ) were described earlier (44) and donated by N. A. Lambert (Medical College of Georgia, Augusta, GA). A vector encoding the Salmonella typhimurium SigD protein cloned in pEGFP-N1 lacking the EGFP-encoding sequences (45), with SigD starting at residue 28 to increase stability, was a gift from B. Brett Finlay (University of British Columbia, Vancouver, Canada).
For studies of inositol phosphate formation, COS-7 cells (1.5 ϫ 10 5 /well) were seeded into 12-well plates. After 24 h, the cells were incubated with fresh medium (1 ml, 1 h), and cotransfected with vector encoding a GFP-tagged PLC␤ 2 derivative (250 ng) together with 750 ng of vectors encoding the various activating proteins. Transfection was done with Lipofectamine TM 2000 (Invitrogen) according to the manufacturer's instructions. At 48 h post-transfection, the cells were taken for experiments on inositol phosphate formation. For experiments on subcellular fractionation of PLC␤ 2 , COS-7 cells (2.5 ϫ 10 6 /10-cm dish) were co-transfected as above with PLC␤ 2 -GFP or PLC␤ 2 ⌬-GFP (7 g of DNA) together with 7 g of vectors encoding one of the PLC␤ 2 stimulators or empty vector.
FRAP-FRAP studies (47,48) were conducted as described (7). The experiments were performed 24 -26 h post-transfection on COS-7 cells transfected with PLC␤ 2 -GFP derivatives as described above. All experiments were conducted at 22°C, in Hanks' balanced salt solution supplemented with 20 mM HEPES, pH 7.2. The monitoring argon ion laser beam (488 nm and 1.2 microwatts) was focused through the microscope (Zeiss Universal, Carl Zeiss MicroImaging) to a Gaussian spot with a radius ϭ 0.85 Ϯ 0.02 m (63ϫ/1.4 numerical aperture (NA) oil-immersion objective) or 1.36 Ϯ 0.04 m (40ϫ/0.75 NA objective). Experiments were conducted with each beam size (beam-size analysis; described previously (7,37)). The ratio between the illuminated areas ( 2 (40ϫ)/ 2 (63ϫ)) was 2.56 (n ϭ 39). After a brief measurement at the monitoring intensity, a 5-milliwatt pulse (4 -6 ms or 10 -20 ms for the 63ϫ and 40ϫ objectives, respectively) bleached 50 -70% of the fluorescence in the spot. Fluorescence recovery was followed by the monitoring beam. The apparent characteristic fluorescence recovery time () and the mobile fraction (R f ) were derived from the FRAP curves by nonlinear regression analysis, fitting to a lateral diffusion process with a single value (49).
Statistical Analysis of FRAP Data-The significance of differences between values measured with the same laser beam size was evaluated by Student's t test. To compare ratio measurements ((40ϫ)/(63ϫ) and 2 (40ϫ)/ 2 (63ϫ)), we employed bootstrap analysis, which is preferable for comparison between ratio values (50). The (40ϫ) and (63ϫ) values were resampled with replacement using Excel, and average values from each group of resampled data ((40ϫ) Boot and (63ϫ) Boot ) were derived. For each beam size, 1000 averaged samples were generated, followed by calculation of the bootstrap ratio dividing (40ϫ) Boot by (63ϫ) Boot . To evaluate whether the ratios thus obtained differ significantly from the beam-size ratio calculated by the same method ( 2 (40ϫ) Boot / 2 (63ϫ) Boot ), the set of the bootstrap ratios was divided by the set of beam area bootstrap ratios, and the p value was derived from the spread of the resulting histogram at ϳ1.
Radiolabeling of Inositol Phospholipids and Analysis of Inositol Phosphate Formation-Twenty-four hours after transfection, COS-7 cells were washed once with 0.5 ml/well of buffer A (10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , 140 mM NaCl, 2.7 mM KCl, pH 7.4), followed by addition of 0.4 ml/well of Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 10 Ci/ml myo-[2-3 H]inositol (Amersham Biosciences) and 10 mM LiCl. After incubation for 20 h, the cells were washed once with 0.4 ml/well of buffer A and lysed in 0.2 ml/well 10 mM ice-cold formic acid (51). Samples were incubated on ice for 30 min, neutralized with 0.3 ml/well of 10 mM NH 4 OH, and centrifuged for 5 min at 15,000 ϫ g. The total inositol phosphates in the supernatants were separated on columns of Dowex 1 ϫ 8 -200 ion exchange resin (Sigma), as described (51,52), and the radioactivity was quantified by liquid scintillation counting.
Subcellular Fractionation-COS-7 cells grown in 10-cm dishes were transiently transfected as described above by vectors encoding wt or mutant GFP-PLC␤ 2 and various activators. After 20 h, the cells were serum-starved (24 h) to avoid seruminduced activation, scraped into 8 ml of buffer A, and centrifuged at 300 ϫ g for 5 min. The cells were lysed in 70 l of ice-cold buffer B (20 mM Tris/HCl, pH 7.5, 2 mM EDTA, 2 g/ml soybean trypsin inhibitor, 3 mM benzamidine, 0.1 mM phenylmethylsulfonyl fluoride, 1 M pepstatin, 1 M leupeptin, and 1 g/ml aprotinin) by freezing in liquid nitrogen and thawing followed by homogenization, forcing the suspension ten times through a 0.40-ϫ 20-mm syringe needle. After removal of unbroken cells and nuclei (300 ϫ g, 10 min, 4°C), particulate (P) and soluble (S) fractions were separated by centrifugation at 100,000 ϫ g for 60 min at 4°C. The P fraction was washed once with 100 l of buffer B and resuspended in 25 l of buffer B. To enable direct comparison of the relative distribution of PLC␤ 2 -GFP (wt or mutant) between the two fractions, equal proportions (identical percentages by volume, v/v) of pairs of the P and S fractions were analyzed by SDS-PAGE and immunoblotting using anti-GFP antibodies followed by peroxidase goat antimouse and ECL. The bands were quantified by densitometry using EZQuant-Gel 2.2 (EZQuant Ltd.).
Miscellaneous-Protein concentrations were determined according to Bradford (53) using bovine IgG as standard. SDS-PAGE and immunoblotting were as described before (7); immunoreactive proteins were visualized using the ECL TM Western blotting Detection System (Amersham Biosciences).

PLC␤ 2 Activation by Different Stimulators Can Occur at
Varying Extents of Membrane Recruitment-PLC␤ 2 can be activated by G protein subunits (␣ q , ␤␥) and by Rac with different orders of efficacy and potency (1,2,6). Because the different stimulators interact with distinct regions of the enzyme, they may differ in their ability to recruit PLC␤ 2 to the membrane, as well as in the resulting mode of membrane interactions. We therefore compared, in cells and under identical conditions, the activation of PLC␤ 2 and the PLC␤ 2 ⌬ mutant by the different stimulators ( Fig. 1). Because a major part of the studies involves PLC␤ 2 -GFP chimeric proteins, we validated that the GFP tagging does not interfere with the activation of the enzyme by all the stimulators. To this end, we compared side-by-side (Fig. 1) the ability of ␤ 1 ␥ 2 , ␣ q , and Rac2 (wt or mutated) to stimulate inositol phosphate formation by PLC␤ 2 , PLC␤ 2 -GFP, and PLC␤ 2 ⌬-GFP. Inositol phosphate formation was measured in COS-7 cells co-transfected with vectors encoding one of the above PLC␤ 2 constructs (replaced by empty or GFP-encoding vectors as controls) together with: (i) ␤ 1 ␥ 2 (␤ 1 along with ␥ 2 (wt)); (ii) ␤ 1 ␥ 2 (C68S) (␤ 1 along with the isoprenylation mutant ␥ 2 (C68S)); (iii) ␣ q (wt); (iv) Rac2(wt); or (v) Rac2(G12V). Comparison between the inositol phosphate formation levels in cells transfected with PLC␤ 2 versus PLC␤ 2 -GFP (third and fourth bar groups in Fig. 1) clearly demonstrates that PLC␤ 2 -GFP was activated by all stimulators to an extent similar to that of untagged PLC␤ 2 . These results confirm the earlier demonstration that PLC␤ 2 -GFP is effectively stimulated by activated Rac2 (7), and extend them to show that attachment of GFP to the C terminus of PLC␤ 2 does not affect the regulation of the enzyme by the heterotrimeric G protein subunits. For both PLC␤ 2 and PLC␤ 2 -GFP, the degree of stimulation was Rac2(G12V) Ͼ ␣ q (wt) Ͼ ␤ 1 ␥ 2 . Considering that only a portion of ␣ q (wt) is activated under the conditions used here, this is consistent with the rank order of potencies (which cannot be directly determined in cells) of the three stimulators in in vitro assays: activated ␣ q Ͼ activated Rac2 Ͼ ␤ 1 ␥ 2 (6,14,54). As shown in Fig. 1, stimulation by ␤ 1 ␥ 2 required membrane anchorage of ␥ 2 , as suggested by the lack of PLC␤ 2 activation  FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6 upon replacement of ␥ 2 by the isoprenylation-resistant mutant ␥ 2 (C68S). Rac2(wt) had only a weak effect, in accord with the notion that it should undergo activation to stimulate PLC␤ 2 (7). It should be noted that ␣ q (wt) induced a marked stimulatory response in cells expressing PLC␤ 2 or PLC␤ 2 -GFP, but not in cells expressing PLC␤ 2 ⌬-GFP, in line with the inability of the latter to bind ␣ q . However, PLC␤ 2 ⌬-GFP remained responsive to Rac2(G12V) or ␤ 1 ␥ 2 , in accord with earlier in vitro studies on the untagged form of this mutant (6). Constitutively active ␣ q (R183C) gave a very high stimulation of inositol phosphate formation (up to ϳ60-fold; not shown in Fig. 1 due to the outof-range value); this robust activation was obtained already in mock-or GFP-transfected cells, suggesting that it reflects activation of endogenous PLC␤ isozymes other than PLC␤ 2 , which are present in COS-7 cells (13,55). Constitutively active Rac2(G12V) did not stimulate inositol phosphate formation in the absence of a co-transfected PLC␤ 2 construct (Fig. 1), because it specifically activates PLC␤ 2 , whose endogenous expression level in these cells is very low. PLC␤ 2 must be recruited to the membrane to interact with its substrate, PtdInsP 2 , which is membrane-associated. We therefore investigated the ability of the different stimulators to translocate PLC␤ 2 -GFP and PLC␤ 2 ⌬-GFP to the membrane fraction ( Fig. 2). To this end, COS-7 cells were co-transfected with a vector encoding PLC␤ 2 -GFP or PLC␤ 2 ⌬-GFP together with either empty vector or with vectors encoding various PLC␤ 2 activators, and the relative distribution of PLC␤ 2 -GFP or PLC␤ 2 ⌬-GFP between the cytosolic (S) and membrane (P) fractions was determined by cell fractionation as described under "Experimental Procedures." Fig. 2A shows that singly expressed PLC␤ 2 -GFP was mostly in the cytosol (S fraction). Co-expression with the different PLC␤ 2 activators increased the proportion (%) of PLC␤ 2 -GFP in the particulate fraction to varying degrees ( Fig. 2A), to extents that correlated with their abilities to stimulate the enzymatic activity of PLC␤ 2 (cf. Fig. 1

Phospholipase C-␤ 2 Activity and Membrane Interaction
Because the fractionation experiment specifically detects the localization of PLC␤ 2 -GFP, it is not masked by activation of endogenous PLC␤ enzymes (unlike the stimulation of PLC␤ activity), enabling us to measure the effects of constitutively active ␣ q (R183C) on PLC␤ 2 -GFP membrane association. This mutant had a slightly higher effect than ␣ q (wt), which was hard to detect due to the limited sensitivity of the fractionation assay, but is supported by the more sensitive biophysical FRAP studies of PLC␤ 2 membrane interactions (see Fig. 4). Interestingly, even though both Rac2(G12V) and ␣ q (R183C) are both constitutively active, Rac2(G12V) recruited PLC␤ 2 -GFP to the particulate fraction to a higher extent than ␣ q (R183C). Together with the lower membrane recruitment and lower activation of PLC␤ 2 by ␤ 1 ␥ 2 (Figs. 1 and 2A), this raises the possibility that the extent and/or mode of membrane recruitment mediated by the distinct stimulators are different. This view gains strong support from the FRAP beam-size analysis experiments (Fig. 4), which detect with high sensitivity the membrane association dynamics of PLC␤ 2 -GFP in live cells; it is corroborated by fractionation studies on cells expressing PLC␤ 2 ⌬-GFP (Fig. 2B). Here, ␤ 1 ␥ 2 did not induce a measurable increase in the percentage of PLC␤ 2 ⌬-GFP in the membrane pellet, contrasting with its ability to induce mild, but clear stimulation of PLC␤ 2 ⌬-GFP enzymatic activity (Fig. 1). The ability of ␤ 1 ␥ 2 to activate PLC␤ 2 or PLC␤ 2 ⌬ despite its weak effect on their recruitment to the particulate fraction is likely to reflect transient (as opposed to stable) membrane recruitment, enabling significant dissociation of the enzyme from the membrane during fractionation. The notion that ␤ 1 ␥ 2 elicits transient association of the enzyme with the membrane was validated by the biophysical studies described later (see Figs. 4 and 5). In line with the inability of PLC␤ 2 ⌬ to bind ␣ q , neither ␣ q (wt) nor ␣ q (R183C) affected its membrane association. On the other hand, Rac2(G12V) was highly effective in recruiting PLC␤ 2 ⌬ to the particulate fraction, in accord with the concept that activated Rac/Cdc42 GTPases interact with the N-terminal PH domain also present in the mutant (6,7,17,18).

FRAP Studies Demonstrate Distinct Modes of PLC␤ 2 Membrane Recruitment by the Different Stimulators-
The subcellular fractionation experiments (Fig. 2) provide a measure for the population of PLC␤ 2 molecules that exhibit relatively stable association with the total membrane fraction. To investigate the effects of the various PLC␤ 2 activators on the mode and dynamics of PLC␤ 2 interactions with the plasma membrane of live cells, we expressed PLC␤ 2 -GFP in COS-7 cells and employed FRAP to measure its lateral diffusion and membrane association dynamics in the presence or absence of the various stimulators. Typical FRAP experiments are depicted in Fig. 3; quantitative results on multiple cells using two different sizes of a Gaussian laser beam (FRAP beam-size analysis) are shown in Fig. 4. The beam-size analysis (7, 37) explores the membrane interaction mode of proteins capable of both lateral diffusion in the membrane and of exchange between membrane-associated and cytoplasmic pools. If FRAP occurs exclusively by diffusion, the characteristic fluorescence recovery time is identical to the characteristic diffusion time D , which is proportional to the bleached area ( ϭ D ϭ 2 /4D, where is the Gaussian radius of the beam, and D is the lateral diffusion coefficient) (49). In the current studies, the ratio between the recovery times obtained with the two beam sizes generated using the 40ϫ and 63ϫ objectives, (40ϫ)/(63ϫ), should be 2.56 (the measured ratio between the illuminated areas). When FRAP occurs by exchange, reflects the chemical relaxation time, which is independent of the bleached area; i.e. (40ϫ)/(63ϫ) should equal 1. Intermediate ratios suggest mixed recovery, where the faster process has a higher contribution (7,37).
Because PLC␤ 2 -GFP (and PLC␤ 2 ⌬-GFP) have a significant cytoplasmic fraction (cf. Fig. 2 ϭ 1), suggesting a mixed contribution of the two mechanisms (7,37). The contribution of exchange does not allow an accurate calculation of the lateral diffusion coefficient; yet, an estimate of D can be calculated from (63ϫ) and the beam size with the 63ϫ objective, because the recovery at this smaller beam size contains a higher contribution of lateral diffusion (56). This yields D ϭ 2 /4 D ϭ (3.2 Ϯ 0.2) m 2 /s, faster than D of the lipid probe DiIC 16 (1 m 2 /s) in the same cells (7) or the 0.5-1 m 2 /s value reported for PtdInsP 2 (32,34), in accord with the lack of stable binding of the enzyme to the plasma membrane and the significant contribution of exchange to the FRAP measurements.
Stimulation of PLC␤ 2 -GFP by co-expression with Rac2-(G12V) led to a large and highly significant increase in the FRAP times (; slower recovery rates) of PLC␤ 2 -GFP (Fig. 4A), in line with the marked increase in its membrane-associated fraction ( Fig. 2A). Importantly, this was accompanied by a shift of the FRAP mechanism to recovery by nearly pure exchange ((40ϫ)/(63ϫ) Ϸ 1; Fig. 4B), suggesting that the characteristic diffusion time ( D ) must be at least an order of magnitude slower than for exchange ( ex ), resulting in a negligible contribution of the lateral diffusion to the fluorescence recovery. Thus, in the presence of Rac2(G12V), D of PLC␤ 2 -GFP is at least 2 s, 10-fold slower than the measured (63ϫ) (0.2 s, reflecting exchange), providing an upper limit of D ϭ 0.09 m 2 /s. This D value is much lower than D of lipid probes in the plasma membrane, indicating that the enhanced association of PLC␤ 2 with the plasma membrane following activation by Rac2(G12V) is due to interactions with membrane proteins and/or localized protein/lipid clusters. In addition, because the dissociation rate governs fluorescence recovery due to exchange (57,58), measured in the FRAP experiment under conditions where the recovery is due to exchange ( ex ) reflects the time constant for dissociation from the plasma membrane. This allows the calculation of the distance that a given fraction of the membrane-associated protein will diffuse laterally prior to dissociation into the cytoplasm (travel range); for a fraction comprising 63% of the protein population, this range (r) is given by r ϭ (2 ϫ D ϫ ) 1 ⁄2 (59). For Rac2(G12V)-activated PLC␤ 2 -GFP, this calculation yields r ϭ (2 ϫ 0.09 ϫ 0.2) 1 ⁄2 ϭ 0.19 m, suggesting that recruitment of PLC␤ 2 to the membrane by Rac2(G12V) targets the activated PLC␤ 2 mainly to laterally restricted small ranges, where it acts until it dissociates back to the cytoplasm.
To examine whether PtdInsP 2 hydrolysis following PLC␤ 2 activation is involved in altering the membrane interaction dynamics of PLC␤ 2 -GFP upon activation, we co-transfected COS-7 cells with PLC␤ 2 -GFP (150 ng of DNA) together with a 7-fold excess of a plasmid encoding SigD, a bacterial inositol phosphatase shown to hydrolyze PtdInsP 2 and reduce its cellular level (60,61). FRAP studies conducted 24 h post-transfection showed no change in PLC␤ 2 -GFP FRAP parameters, suggesting that PtdInsP 2 hydrolysis per se does not significantly affect the dynamics of PLC␤ 2 membrane interactions. This notion is further supported by the finding that each PLC␤ 2 activator has a distinct effect on the FRAP kinetics of the stimulated PLC␤ 2 (Figs. 4 and 5; see below), although PtdInsP 2 hydrolysis by PLC␤ 2 is elicited in all cases (Fig. 1).
Unlike the robust effect of Rac2(G12V) on the FRAP dynamics of PLC␤ 2 -GFP, co-expression with ␣ q (wt or constitutively active) induced modest, albeit significant, effects on the FRAP parameters of the enzyme. This correlated with the biochemical fractionation experiments ( Fig. 2A), where Rac2(G12V) induced a higher increase in the percentage of PLC␤ 2 -GFP associated with the membrane fraction. Accordingly, ␣ q -(R183C) and ␣ q (wt) modulated the FRAP parameters of PLC␤ 2 toward the same direction as Rac2(G12V), but to a lower extent (Fig. 4). They mildly increased the values of PLC␤ 2 -GFP and shifted its (40ϫ)/(63ϫ) ratio to lower values, albeit still higher than 1. Note that the sensitivity of the FRAP beam-size analysis demonstrates that ␣ q (R183C) has a stronger effect than ␣ q (wt) on the ratio (Fig. 4B), a difference that was too mild to detect by the fractionation studies. Because the exchange rates of PLC␤ 2 -GFP stimulated by ␣ q (wt or mutant) were distinctively faster than after stimulation by active Rac2, the increase in the values was very mild. This is especially valid for (40ϫ), which contains a higher contribution of exchange (due to the larger beam size) and was only weakly affected by ␣ q (wt) and even less by ␣ q (R183C). Therefore, the ratio can detect the stronger effect of ␣ q (R183C) with higher sensitivity. The results depicted in Fig. 4 suggest that, although stimulation by ␣ q (R183C) or ␣ q (wt) enhances the mobility-retarding interactions of PLC␤ 2 -GFP with the plasma membrane, the interactions are weaker than those induced by Rac2(G12V) (7; see Fig.  4), and may involve association with different targets in the membrane. Estimation of D for PLC␤ 2 -GFP co-expressed with ␣ q (R183C) or ␣ q (wt) from the (63ϫ) values (which are very similar for the two ␣ q proteins) yields D ϭ (2.1 Ϯ 0.2) m 2 /s, somewhat higher than lipid probe diffusion, most likely due to the residual contribution of exchange. Because the (40ϫ)/(63ϫ) ratio of PLC␤ 2 -GFP upon co-expression with an ␣ q protein is intermediate between the values expected for recovery by diffusion and exchange, the characteristic exchange time ex should be in the same range as D and can be  Fig. 3, on COS-7 cells transfected with PLC␤ 2 -GFP, and an excess of empty vector (Control) or vectors encoding the indicated proteins as described under "Experimental Procedures." Two beam sizes were generated using a 63ϫ and 40ϫ objectives (see "Experimental Procedures"), and the values were determined with each. The ratio between the areas illuminated by the two beams, 2 (40ϫ)/ 2 (63ϫ), was 2.56 (n ϭ 39). This ratio is expected for FRAP by lateral diffusion, whereas a ratio of 1 is expected for recovery by exchange (37). The R f values were high in all cases (Ն0.93). A, values. Bars are means Ϯ S.E. of 40 -60 measurements, each conducted on a different cell. Comparing values measured with the same beam size, Rac2(G12V) and ␣ q induced significant increases in of PLC␤ 2 -GFP relative to the control (***, p Ͻ 10 Ϫ6 ; **, p Ͻ 0.005; Student's t test). ␤ 1 ␥ 2 had no significant effect on (40ϫ), but reduced (63ϫ) (*, p Ͻ 0.02). B, (40ϫ)/(63ϫ) ratios. The ratio values ( ratios and the beam-size ratio) and their S.E. were calculated from the experimentally measured values ((40ϫ) and (63ϫ) for ratio, 2 (40ϫ) and 2 (63ϫ) for the beam-size ratio) using bootstrap analysis. The bootstrap analysis (see "Experimental Procedures") showed that the ratios of PLC␤ 2 differ significantly from the 2.56 beam-size ratio predicted for FRAP by lateral diffusion in all cases (***, p Ͻ 10 Ϫ6 ; *, p Ͻ 0.02), except for co-expression with ␤ 1 ␥ 2 (p Ͼ 0.3). Comparison of the ratios to 1 (the value expected for FRAP by exchange) using bootstrap analysis shows that in the presence of Rac2(G12V) the ratio of PLC␤ 2 -GFP is not significantly different from 1 (p Ͼ 0.4). estimated from (40ϫ) ( measured with the larger beam size, where the relative contribution of exchange is higher). The (40ϫ) values for ␣ q (R183C) and ␣ q (wt) are 0.12 and 0.15 s, respectively, suggesting that the travel range of PLC␤ 2 following activation by ␣ q is roughly r ϭ (2 ϫ D ϫ ex ) 1 ⁄2 ϭ (2 ϫ 2.1 ϫ 0.135) 1 ⁄2 ϭ 0.75 m, 4-fold larger than after stimulation with Rac2(G12V). Thus, activation by ␣ q leads to a less confined recruitment of PLC␤ 2 to the plasma membrane, because the active enzyme can diffuse laterally a longer distance prior to dissociation to the cytoplasm.

. FRAP beam-size analysis suggests activator-dependent distinct modes of PLC␤ 2 -GFP interactions with the plasma membrane. FRAP experiments were conducted at 22°C as in
Unexpectedly, the effects of ␤ 1 ␥ 2 on the FRAP parameters of PLC␤ 2 -GFP were highly different from those of Rac2(G12V) and ␣ q . Unlike the marked increase in of PLC␤ 2 -GFP induced by the latter two stimulators, activation by ␤ 1 ␥ 2 had only very subtle effects on the values (Fig. 4A). Concomitantly, ␤ 1 ␥ 2 shifted the (40ϫ)/(63ϫ) ratio of PLC␤ 2 in a direction opposite to that mediated by Rac2(G12V) and ␣ q , resulting in a ratio indistinguishable from lateral diffusion (Fig. 4B). This suggests that, following stimulation by ␤ 1 ␥ 2 , the exchange rate of PLC␤ 2 -GFP between the plasma membrane and the cytoplasm becomes much slower than its lateral diffusion rate, indicating enhanced membrane interactions. Yet, these interactions are highly transient. This notion is supported by the biochemical fractionation studies, which showed a markedly weaker recruitment of PLC␤ 2 to the membrane fraction by ␤ 1 ␥ 2 relative to Rac2(G12V) or ␣ q (Fig. 2A). The D value of PLC␤ 2 -GFP coexpressed with ␤ 1 ␥ 2 can be accurately calculated from the FRAP experiments, because in the presence of ␤ 1 ␥ 2 the ratio between the two beam sizes is as expected for lateral diffusion. Interestingly, the D value thus obtained, (3.6 Ϯ 0.2) m 2 /s, is markedly higher than D of lipid probes such as DiIC 16 (1 m 2 /s) or PtdInsP 2 (0.5-1 m 2 /s) (7,32,34). It is much higher than the lateral diffusion of ␤ 1 ␥ 2 at the plasma membrane of COS-7 cells, which we measured by FRAP on cells co-transfected with 850 ng DNA (1:1 ratio) of venus 1-155-␥ 2 and venus 155-239-␤ 1 . The two halves of the venus protein are not fluorescent separately and form a stable fluorescent complex due to bifunctional fluorescence complementation upon association (44,62). These measurements yielded D ϭ (0.21 Ϯ 0.2) m 2 /s (n ϭ 31). This value, which is similar to the 0.23 m 2 /s value reported for ␤ 1 ␥ 2 in HEK293 cells (62), was not altered by co-expressing the venus-tagged ␤ 1 ␥ 2 constructs with PLC␤ 2 (150 ng of plasmid DNA). This suggests that ␤ 1 ␥ 2 -stimulated PLC␤ 2 exhibits a surfing-like diffusion along the plasma membrane, spending some of the time transiently bound to membrane lipids and/or fatty acid-anchored proteins, including the ␤ 1 ␥ 2 complex itself. The latter notion is supported by the failure of ␤ 1 ␥ 2 (C68S), where ␥ 2 cannot undergo isoprenylation, to modulate the FRAP parameters of PLC␤ 2 (Fig. 4). These findings have important implications for the travel range of PLC␤ 2 following stimulation by ␤ 1 ␥ 2 . A lower limit for this travel range can be derived based on the assumption that the exchange time ( ex ) of PLC␤ 2 -GFP co-expressed with ␤ 1 ␥ 2 is at least 10-fold slower than the diffusion time D , which is essentially equal to the measured value (e.g. (63ϫ) ϭ D (63ϫ) ϭ 0.045 s, and thus ex is at least 0.45 s). Therefore, the lower limit of the travel range is r ϭ (2 ϫ D ϫ ex ) 1 ⁄2 ϭ (2 ϫ 3.6 ϫ 0.45) 1 ⁄2 ϭ 1.8 m, ϳ10-fold larger than following stimulation with Rac2(G12V). We conclude that activation of PLC␤ 2 by ␤␥ involves a mechanism that recruits the enzyme to the plasma membrane by inducing interactions that enable it to roam relatively large membrane regions prior to detachment to the cytoplasm, as required for hydrolysis of dispersed PtdInsP 2 populations.
To validate the specificity of the effects of the various stimulators on the FRAP parameters of PLC␤ 2 , we conducted analogous FRAP studies to measure their effects on PLC␤ 2 ⌬-GFP (Fig. 5), which does not respond to ␣ q due to the F819-E1166 C-terminal deletion. In line with the loss of the response of this mutant to ␣ q as measured by both activation and recruitment to the membrane (Figs. 1 and 2), neither ␣ q (R183C) nor ␣ q (wt) modulated the values or the (40ϫ)/(63ϫ) ratios of PLC␤ 2 ⌬-GFP (Fig. 5). On the other hand, the effects of ␤ 1 ␥ 2 on the FRAP  Fig. 4, except that PLC␤ 2 ⌬-GFP replaced PLC␤ 2 -GFP. The R f values were high in all cases (Ն95%; not shown). A, values. Bars are means Ϯ S.E. of 40 -60 measurements. Comparison of the values measured for cells co-transfected with the various activators relative to control (cells singly transfected with PLC␤ 2 ⌬-GFP) showed highly significant increase in both (63ϫ) and (40ϫ) for PLC␤ 2 ⌬-GFP coexpressed with Rac2(G12V) (***, p Ͻ 10 Ϫ6 ; Student's t test). ␤ 1 ␥ 2 had a weaker but significant effect on (40ϫ) of PLC␤ 2 ⌬-GFP (**, p Ͻ 0.005). In all other cases, the differences from the values measured with the same beam size in the control cells were not significant (p Ͼ 0.05). B, (40ϫ)/(63ϫ) ratios. All the ratio values and their S.E. were calculated using bootstrap analysis. This analysis (see "Experimental Procedures") showed that co-expression with ␤ 1 ␥ 2 or with Rac2(G12V) shifts the ratio of PLC␤ 2 ⌬-GFP to values essentially similar to the 2.56 ratio expected for FRAP by lateral diffusion (p Ͼ 0.15 for co-expression with ␤ 1 ␥ 2 , and p Ͼ 0.3 for co-expression with constitutively active Rac2). In all other cases, (40ϫ)/(63ϫ) of PLC␤ 2 ⌬ differed significantly from the 2.56 beam-size ratio (**, p Ͻ 0.005; *, p Ͻ 0.02), except for co-expression with ␤ 1 ␥ 2 (p Ͼ 0.3), indicative of a significant contribution of exchange to the FRAP mechanism.
dynamics of PLC␤ 2 ⌬ closely resembled their effects on PLC␤ 2 , inducing only a very minor effect on the values, while increasing the (40ϫ)/(63ϫ) ratio very close to the ratio expected for FRAP by pure lateral diffusion. These findings support the notion that ␤ 1 ␥ 2 expression induces highly transient interactions of PLC␤ 2 and PLC␤ 2 ⌬ with the plasma membrane, in line with its inability to measurably enhance the fraction of PLC␤ 2 ⌬-GFP in the membrane pellet (Fig. 2B). Accordingly, the very fast lateral diffusion of ␤ 1 ␥ 2 -stimulated PLC␤ 2 ⌬ (D ϭ (5.2 Ϯ 0.3) m 2 /s, calculated from the (63ϫ) in Fig. 5A)) suggests surfing-like diffusion (even faster than that of ␤ 1 ␥ 2 -stimulated PLC␤ 2 ), resulting in a high travel range prior to dissociation to the cytoplasm; assuming that ex for PLC␤ 2 ⌬-GFP is at least 10-fold slower than (63ϫ), the lower limit of ␤ 1 ␥ 2 -stimulated PLC␤ 2 ⌬ travel range is r ϭ (2 ϫ D ϫ ex ) 1 ⁄2 ϭ (2 ϫ 5.2 ϫ 0.33) 1 ⁄2 ϭ 1.9 m, similar to ␤ 1 ␥ 2 -stimulated PLC␤ 2 . Finally, activation of PLC␤ 2 ⌬-GFP by Rac2(G12V) markedly increased the values (Fig. 5A), as in the case of PLC␤ 2 -GFP, supporting the notion that activation by Rac2 enhances the membrane association of PLC␤ 2 ⌬, in accord with the fractionation studies (Fig. 2). However, the effect of Rac2(G12V) on the (40ϫ)/ (63ϫ) ratio differed markedly between PLC␤ 2 and PLC␤ 2 ⌬ (cf. Figs. 5B and 4B), shifting the ratio of PLC␤ 2 ⌬ toward recovery by pure lateral diffusion. As shown and discussed by us earlier (37), when recovery occurs by lateral diffusion and exchange, their contribution to the measured FRAP is determined by the relative rates of the two processes, with the faster process prevailing. Thus, the shift of Rac2-stimulated PLC␤ 2 ⌬ to FRAP by lateral diffusion directly demonstrates that its exchange rate is at least 10-fold slower than its lateral diffusion rate. This situation differs from that observed for Rac2-stimulated PLC␤ 2 , where exchange becomes the dominant mechanism, reflecting a much slower lateral diffusion rate for the fulllength, Rac2-stimulated PLC␤ 2 (compare Figs. 5B and 4B). This difference, discussed by us extensively earlier (7), is in line with the suggestion (7) that the C-terminal region missing in PLC␤ 2 ⌬ has a role in the membrane interactions of full-length PLC␤ 2 , mainly with membrane proteins that diffuse slower than lipid probes. In its absence (PLC␤ 2 ⌬), the interactions with slow diffusing membrane proteins become weaker, resulting in a loss of the diffusion-restricting interactions with the above proteins and in faster, lipid-like dif- Fig. 5A). Under these conditions, the diffusion of PLC␤2⌬ becomes fast relative to its exchange rate, as indicated by the diffusion-dominated FRAP mechanism (Fig. 5B). Thus, ex of Rac2-stimulated PLC␤2⌬ should be higher than the measured (63ϫ) by at least 10-fold (i.e. ex Ն 1.4 s). This would increase the travel range of Rac2-stimulated PLC␤ 2 ⌬, with a lower limit estimate of r ϭ (2 ϫ D ϫ ex ) 1 ⁄2 ϭ (2 ϫ 1.3 ϫ 1.4) 1 ⁄2 ϭ 1.9 m. This in turn indicates that, unlike the limited travel range of Rac2(G12V)-activated PLC␤ 2 , the travel range on the plasma membrane of the Rac2-activated PLC␤ 2 ⌬ mutant is 10-fold higher, strongly compromising the localized nature of the membrane recruitment.

DISCUSSION
The mechanisms regulating the membrane recruitment and activation of PLC␤ isozymes by their activators are not fully understood. Here, we investigated these issues in live cells for PLC␤ 2 activated by several stimulators (Rac2, ␣ q , and ␤ 1 ␥ 2 ). Our findings demonstrate that each activator causes a distinct mode of PLC␤ 2 membrane association, ranging between recruitment to confined regions (activation by Rac2, and to a lesser degree by ␣ q ) and fast, surfing-like diffusion of the enzyme along the cytoplasmic leaflet of the plasma membrane (activation by ␤ 1 ␥ 2 ). The diversity of these mechanisms has important implications for the PtdInsP 2 populations targeted by PLC␤ 2 , because the first mechanism targets the activated enzyme to act on discrete PtdInsP 2 populations localized at or near the recruitment sites, while the second directs the enzyme to act on dispersed PtdInsP 2 populations.
In the current study, we combined FRAP beam-size analysis with biochemical and signaling assays to investigate the mechanisms by which Rac2, ␣ q , and ␤ 1 ␥ 2 mediate membrane recruitment and activation of PLC␤ 2 . Because the FRAP studies and some of the biochemical studies employed GFP-tagged PLC␤ 2 or PLC␤ 2 ⌬, we first validated that the GFP-tagged enzymes are as responsive as their untagged counterparts to the various PLC␤ 2 activators (Fig. 1). Moreover, the specificity of the activation was kept in the GFP-tagged constructs, as shown by the loss of their response to ␤ 1 ␥ 2 upon mutational removal of the ␥ 2 membrane anchor site (C68S) and the loss of ␣ q activation in the PLC␤ 2 ⌬ mutant (Fig. 1).
The fractionation studies demonstrate that PLC␤ 2 -GFP and PLC␤ 2 ⌬-GFP are mainly cytosolic prior to activation (Fig. 2). However, transient association can be overlooked in such studies, which require relatively stable association with the membrane. Indeed, the FRAP beam-size analysis shows that unstimulated PLC␤ 2 and PLC␤ 2 ⌬ do interact transiently with the plasma membrane, as indicated by the mixed contribution of lateral diffusion and exchange to their FRAP kinetics, which are much slower than that of free cytoplasmic GFP (Figs. 3-5; see also Ref. 7). Thus, although mainly cytoplasmic, unstimulated PLC␤ 2 (and PLC␤ 2 ⌬) experiences some mobility-retarding interactions with plasma membrane constituents, which are insufficient for stable association. The D values estimated from (63ϫ) for PLC␤ 2 and PLC␤ 2 ⌬ prior to activation (3.2 and 4.8 m 2 /s, respectively) are ϳ4-fold higher than those of lipids and of PtdInsP 2 (7,32,34), most likely due to contributions of exchange and possibly of surfing-like diffusion along the cytoplasmic face of the membrane, a diffusion mode discussed later in the context of ␤ 1 ␥ 2 -activated PLC␤ 2 . It should be noted that the lack of significant enzymatic activity in the unstimulated enzymes (Fig. 1), despite their transient interactions with the plasma membrane, suggests that such interactions per se are not sufficient to activate PLC␤ 2 , and that transient recruitment to the membrane (as observed following ␤ 1 ␥ 2 stimulation) should be accompanied by an additional event (e.g. a conformational change) to induce activation.
A striking finding of the current studies is that, although all the stimulators enhance the membrane interactions of PLC␤ 2 , they induce these effects by different mechanisms. Expression of constitutively active Rac2(G12V) induced a robust recruitment of PLC␤ 2 -GFP to the membrane fraction ( Fig. 2A), accompanied by a major modulation of its membrane interaction dynamics (significantly longer values and a drastic shift of the (40ϫ)/(63ϫ) ratio to ϳ1, indicative of exchange-dominated FRAP (Fig. 4)). The significantly slower values at both laser beam sizes indicate that both the lateral diffusion and exchange of PLC␤ 2 -GFP are retarded following activation by Rac2(G12V), while the simultaneous shift to recovery dominated by exchange suggests that the lateral diffusion of PLC␤ 2 is inhibited at least 10-fold more than its exchange, resulting in a negligible contribution of lateral diffusion to the FRAP. Such slow diffusion (the upper limit estimate of D is 0.09 m 2 /s; see "Results") is well below the typical D values of lipid probes, which are in the 1 m 2 /s range (7,32,34), indicating that Rac2(G12V) recruits PLC␤ 2 to the plasma membrane by enhancing its interactions with slow diffusing entities such as transmembrane proteins or protein-lipid clusters. This has important implications for the PtdInsP 2 populations targeted by Rac2-stimulated PLC␤ 2 ; calculation of the resulting travel range prior to dissociation of the majority (63%) of the Racstimulated PLC␤ 2 molecules from the membrane (59) yields 0.19 m (see "Results"), indicating that the enzyme is recruited to act on substrate populations localized in distinct limited regions or clusters. Interestingly, the localized nature of PLC␤ 2 recruitment and activation by Rac2(G12V) is disrupted in the PLC␤ 2 ⌬ mutant, which is effectively recruited by activated Rac2 to the membrane fraction (Fig. 2B), but shifts to FRAP by lateral diffusion ( ratio very close to the ratio expected for pure lateral diffusion (Fig. 5)). The D value obtained for Rac2-stimulated PLC␤ 2 ⌬ is 1.3 m 2 /s, similar to the typical values for lipid probe diffusion. These results indicate that the modulation of PLC␤ 2 membrane interactions by activated Rac2 involve not only the enzyme's PH domain, which was shown to interact with activated Rho GTPases (6,17,18), but also the C-terminal region, a major portion of which is missing in PLC␤ 2 ⌬. The C-terminal region appears to contribute to the interactions of full-length PLC␤ 2 with the slow diffusing membrane constituents, and its deletion in PLC␤ 2 ⌬ interferes with these interactions, leading to faster lateral diffusion relative to exchange. Due to the higher D (1.3 m 2 /s) and slower exchange time ( ex Ն 1.4 s), the travel range of Rac2-stimulated PLC␤ 2 ⌬ increases by at least an order of magnitude (lower estimate, 1.9 m).
The C-terminal region of PLC␤ 2 is essential for activation by ␣ q (7,(13)(14)(15), in line with the failure of ␣ q or ␣ q (R183C) to activate PLC␤ 2 ⌬ and to modulate its membrane interactions (Figs. 1, 2B, and 5). This contrasts with the ability of activated Rac2 and ␤ 1 ␥ 2 to activate PLC␤ 2 ⌬, suggesting that the recruitment and activation mechanisms of PLC␤ 2 by ␣ q may be different. Indeed, although ␣ q (R183C) and ␣ q enhanced the recruitment of PLC␤ 2 -GFP to the membrane fraction ( Fig. 2A), this effect was weaker than that mediated by activated Rac2, and the ability of the ␣ q proteins to modulate the membrane interaction dynamics of PLC␤ 2 was much milder (Fig. 4). Thus, the effect of ␣ q on prolonging the values was evident but much weaker, and the shift toward a ratio of 1 (recovery by exchange) was partial, indicating that, although stimulation by ␣ q increases the contribution of exchange relative to diffusion, the latter still has a contribution to the fluorescence recovery. This suggests that the retardation of the lateral diffusion of PLC␤ 2 following stimulation by ␣ q is less than that induced by Rac2(G12V); indeed, the D value estimated for ␣ q -stimulated PLC␤ 2 (which may contain some contribution of exchange) is 2.1 m 2 /s, close to but somewhat higher than D of lipid probes. This indicates that ␣ q -stimulated PLC␤ 2 molecules interact with membrane constituents different from those targeted by Rac2 activation, possibly including lipids (e.g. via the PH and/or C2 domains) and/or lipid-anchored proteins. Direct association with ␣ q , which can interact with the membrane via its single fatty-acyl residue, may also contribute to these interactions. Nevertheless, the association of ␣ q -stimulated PLC␤ 2 with membrane constituents (including ␣ q itself) must be dynamic, as indicated by the contribution of exchange to the FRAP kinetics (Fig. 4) and by the lower D value (0.47 m 2 /s) measured for a G␣ subunit (␣ oA ) (63). Based on the D and ex values, the travel range of ␣ q -stimulated PLC␤ 2 is estimated to be 0.75 m, suggesting that PLC␤ 2 recruited by ␣ q stimulation is less constrained than the Rac2stimulated enzyme, roaming larger (albeit still limited) membrane regions.
␤ 1 ␥ 2 dimers can interact with the PH domain of PLC␤ 2 (16) but also appear to interact with other portions of the enzyme (6). In line with the latter, distinct interactions, the modulation of the membrane interactions of PLC␤ 2 by ␤ 1 ␥ 2 are very different from those mediated by Rac2 or ␣ q . Unlike activation by the latter two, activation by ␤ 1 ␥ 2 induced highly transient association with the membrane fraction, as evidenced by the fractionation experiments (Fig. 2). Yet, despite this highly transient nature, the FRAP of ␤ 1 ␥ 2 -activated PLC␤ 2 (or PLC␤ 2 ⌬) is strongly dominated by fast lateral diffusion (Figs. 4 and 5), characterized by D values 3-to 5-fold faster than lipids or lipidanchored proteins such as G protein ␣ or ␤␥ subunits (7,32,34,62,63). Combining the highly transient interactions and the very fast diffusion along the plasma membrane, we propose that stimulation by ␤ 1 ␥ 2 induces surfing-like diffusion of PLC␤ 2 enzymes along the cytoplasmic face of the plasma membrane. Thus, when a ␤ 1 ␥ 2 -stimulated PLC␤ 2 molecule dissociates from the membrane, it diffuses a short distance and quickly re-associates with another membrane component, remaining in the juxtamembrane vicinity and diffusing along the membrane. Juxtamembrane diffusion was recently found for paxillin and vinculin above focal adhesions (58). This unique mechanism enables diffusion of ␤ 1 ␥ 2 -stimulated PLC␤ 2 along the membrane at rates much faster than those of lipids or lipidated proteins, including the ␤ 1 ␥ 2 dimers, which diffuse with D ϭ 0.2 m 2 /s (62). The diffusion of ␤ 1 ␥ 2 -stimulated PLC␤ 2 (or PLC␤ 2 ⌬) is also much faster than that of PtdInsP 2 , which has D values of 0.5-1 m 2 /s (32,34), enabling very fast spatial dispersal of the activated enzyme, not limited by the lateral diffusion of either lipidated protein targets or the substrate. This mechanism is likely relevant under physiological conditions, because PLC␤s are typically much less abundant (at least 100-fold) in cells than ␤␥ dimers (64,65), enabling sequential interaction of a single PLC␤ molecule with multiple ␤␥ dimers by dissociation and fast re-association. An important consequence of this fast diffusion is a dramatic increase in the travel range of ␤ 1 ␥ 2activated PLC␤ 2 (or PLC␤ 2 ⌬) to at least 1.9 m. This implies Phospholipase C-␤ 2 Activity and Membrane Interaction FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6 that stimulation by ␤ 1 ␥ 2 recruits PLC␤ 2 to act on dispersed PtdInsP 2 populations.
The divergent mechanisms described above may have evolved for stimulation of PLC␤ 2 (and possibly other PLC␤ isozymes) to accomplish different tasks of cellular regulation. PtdInsP 2 has long been known to occur in cells not only in dispersed populations, but also to form gradients and locally enriched regions (27)(28)(29)(30)(31)(32)(33). We propose that activation by Rac/ Cdc42 recruits PLC␤ 2 to hydrolyze PtdInsP 2 in discrete, spatially restricted zones. On the other hand, activation by ␤ 1 ␥ 2 dimers results in signals that rapidly propagate along the plasma membrane, because the activated PLC␤ 2 can roam large areas along the membrane, hydrolyzing PtdInsP 2 with low spatial resolution. Activation by ␣ q yields an intermediate situation, which may fit conditions where relatively shallow gradients of PtdInsP 2 hydrolysis are beneficial.