|Home | About | Journals | Submit | Contact Us | Français|
Phosphatidylinositol 4,5-bisphosphate (PIP2) controls a surprisingly large number of processes in cells. Thus, many investigators have suggested that there might be different pools of PIP2 on the inner leaflet of the plasma membrane. If a significant fraction of PIP2 is bound electrostatically to unstructured clusters of basic residues on membrane proteins, the PIP2 diffusion constant, D, should be reduced. We microinjected micelles of Bodipy TMR-PIP2 into cells, and we measured D on the inner leaflet of fibroblasts and epithelial cells by using fluorescence correlation spectroscopy. The average ± SD value from all cell types was D = 0.8 ± 0.2 μm2/s (n = 218; 25°C). This is threefold lower than the D in blebs formed on Rat1 cells, D = 2.5 ± 0.8 μm2/s (n = 26). It is also significantly lower than the D in the outer leaflet or in giant unilamellar vesicles and the diffusion coefficient for other lipids on the inner leaflet of these cell membranes. The simplest interpretation is that approximately two thirds of the PIP2 on inner leaflet of these plasma membranes is bound reversibly.
Phosphatidylinositol 4,5 bisphosphate (PIP2) directly regulates many processes at the inner leaflet of the plasma membrane, such as the activation of ion channels, endocytosis, and exocytosis (Di Paolo and De Camilli, 2006 ). It is also the source of three second messengers: how does one lipid do so much? Many investigators have considered the possibility there are different pools of PIP2 in the plasma membrane (Hinchliffe et al., 1998 ; Aikawa and Martin, 2003 ; Janmey and Lindberg, 2004 ). Little is known, however, about how these separate pools come about and how they are regulated. One hypothesis is that membrane-bound clusters of basic residues on peripheral (e.g., myristoylated alanine rich C kinase substrate [MARCKS], K-Ras, and gravin) and integral (e.g., epidermal growth factor receptor [EGFR] and polymeric immunoglobulin receptor [pIgR]) proteins concentrate electrostatically (sequester) a significant fraction of the PIP2 in the membrane; when the local concentration of Ca2+ rises and calcium/calmodulin (Ca/CaM) binds to these basic clusters, the sequestered PIP2 is released, increasing its local free concentration (McLaughlin and Murray, 2005 ).
Theoretical calculations and experimental measurements on model systems support this hypothesis. Specifically, electrostatic theory predicts a membrane-bound cluster of basic residues produces a local positive potential that concentrates multivalent acidic lipids such as PIP2 (Wang et al., 2004 ; McLaughlin and Murray, 2005 ). Fluorescence resonance energy transfer, electron paramagnetic resonance, and phospholipase C (PLC) activity measurements confirm that basic clusters on purified proteins or peptides do indeed laterally concentrate PIP2 in phospholipid vesicles (Murray et al., 2002 ; Gambhir et al., 2004 ). These calculations and model studies also suggest that if the number of membrane-bound basic clusters is equal to or greater than the number of PIP2 on the inner leaflet of a plasma membrane, a significant fraction of PIP2 will be sequestered. Testing this prediction requires direct experiments on living cells, however, because the exact concentration of these basic clusters is not known with certainty for any cell type.
Studies of microvesicles released from human erythrocytes under different conditions (Hagelberg and Allan, 1990 ) provided evidence that ~50% of the PIP2 is bound to cytoskeletal proteins (e.g., Band III and glycophorin) that are depleted in the microvesicles. Specifically, the microvesicles and native membranes contained identical mole fractions of most lipids, but not polyphosphoinositides (e.g., PIP2). The higher mole fraction of PIP2 in the native membranes suggests ~50% of their PIP2 is not free to diffuse into the nascent microvesicles.
One independent way to estimate the fraction of free PIP2 in the plasma membrane is to measure directly the diffusion coefficient, D, of PIP2 on the inner leaflet in a living cell. A simple analysis (see Supplemental Material) reveals that if R is the ratio of reversibly bound or sequestered (S) to free (C) PIP2, the diffusion constant will be scaled by a factor 1/(R + 1).
Specifically, combining the diffusion equation (Fick's second law) with a linear adsorption isotherm (Henry's law) produces a diffusion equation with an apparent diffusion constant, D:
where S is the concentration of sequestered PIP2 (reversibly bound or immobilized), C is the concentration of PIP2 free to diffuse, and Dfree is the diffusion coefficient of free PIP2. If the total concentration of PIP2 T = S + C, then C = T/(R + 1).
If the erythrocyte result can be extrapolated to other cells, such as fibroblasts, a significant fraction of PIP2 in these plasma membranes is also bound reversibly to proteins. Equation 1 then predicts that PIP2 will diffuse less rapidly in the inner leaflet of a fibroblast plasma membrane than in a “bleb” formed on that membrane. It also predicts that PIP2 will diffuse less rapidly in the inner than in the outer leaflet of the plasma membrane and less rapidly on the inner leaflet than either neutral or monovalent lipids that are not bound to proteins.
Experiments on giant unilamellar vesicles (GUVs) showed monomers of fluorescent PIP2 can be incorporated into the outer leaflet of preformed vesicles by exposing them to micelles of PIP2; fluorescence correlation spectroscopy (FCS) measurements showed PIP2 diffuses with a D characteristic of other lipids in the GUVs (Golebiewska et al., 2006 ). Here, we report FCS measurements of the D of fluorescent PIP2 in plasma membranes after microinjecting fluorescent PIP2 micelles into fibroblasts (Rat1, Cos1, REF52, and NIH3T3) and epithelial cells (human embryonic kidney [HEK]293 and Fisher rat thyroid [FRT]).
Our most important result is that Bodipy TMR-PIP2 in the inner leaflet of plasma membranes diffuses significantly more slowly than the same lipid in blebs, in the outer leaflet of plasma membranes, or in GUVs. These observations are consistent with the Hagelberg and Allan (1990) erythrocyte microvesicle partitioning results. Hence, both the partitioning and diffusion experiments suggest a significant fraction of the PIP2 on the inner leaflet is bound reversibly.
The Supplemental Material includes details of the experimental protocols. In brief, we performed confocal imaging and FCS measurements using an LSM 510 Meta/Confocor 2 apparatus (Carl Zeiss, Jena, Germany) with standard configurations. FCS has been reviewed previously (Elson, 2001 ; Rigler and Elson, 2001 ; Marguet et al., 2006 ; Haustein and Schwille, 2007 ). We microinjected the micelles using an InjectMan NI2 with FemtoJet pump from Eppendorf, typically mounted on an Axiovert 200M microscope (Carl Zeiss) fitted with a long working distance 40× phase contrast objective.
We obtained similar results with three different types of Bodipy TMR-phosphatidylinositol 4,5-biphosphate [PI(4,5)P2], by using three different types of micelles, as discussed in Supplemental Material. One major advantage of FCS over fluorescence recovery after photobleaching (FRAP) is that the fluorescent PIP2 we introduce into the membrane does not perturb significantly the endogenous level of PIP2 in the plasma membrane. Specifically, the illuminated (~0.2-μm radius) portion of the plasma membrane contains ~105 lipids and ~103 PIP2, and our technique incorporated 1–100 fluorescent PIP2 into this area.
We made most of our measurements on Rat1 fibroblasts or HEK293 epithelial cells, but we also report data from FCS measurements on REF52, Cos1, NIH3T3 fibroblasts, and FRT epithelial cells. The diffusion coefficient of PIP2 on the inner leaflet did not differ significantly between the cell types or lines, and we pooled all data. All measurements were made at 25°C.
Direct physical measurements (Hirai et al., 1996 ; Flanagan et al., 1997 ) and our FCS data (not shown) indicate both native PIP2 and long-chain Bodipy TMR-PIP2 [see Figure 2 of Gambhir et al. (2004) for structure] form micelles, rather than bilayer vesicles, in solution; this is because of the disparity in the cross sectional areas of the PIP2 head group and hydrocarbon tails. Earlier work showed adding Bodipy TMR-PIP2 micelles (as well as mixed micelles containing this lipid: see Supplemental Material methods) to preformed GUVS produces spontaneous transfer of fluorescent PIP2 into the exposed bilayer leaflet, where they have a diffusion constant characteristic of other phospholipids (Fernandes et al., 2006 ; Golebiewska et al., 2006 ). Thus, we simply microinjected micelles containing Bodipy TMR-PIP2 into cells (Figure 1A), where they diffused rapidly through the cytoplasm and incorporated PIP2 monomers into the plasma membrane (Figure 1B). We then used the confocal microscope to scan from the bottom to the top of the cell and record the Bodipy TMR-PIP2 fluorescence. Figure 1C shows a typical result for a Rat1 fibroblast; the fluorescence peaks correspond to the locations of the plasma membrane at the bottom and top of the cell. The width at half maximum is similar to that observed in GUVs with fluorescent PIP2, where the fluorescence is confined to a bilayer (Golebiewska et al., 2006 ). The resolution is not sufficient to distinguish between PIP2 incorporated into the plasma membrane bilayer and PIP2 adsorbed to cytoskeletal elements. We consider this issue in the Supplemental Material, where we also discuss experiments that suggest why the micelles deliver PIP2 primarily to plasma membrane versus internal membranes.
The FCS measurements were made on the upper portion of the plasma membrane (Figure 1B). The instrument recorded fluctuations in fluorescence for 10-s intervals 10 times, and then it calculated the autocorrelation function from the experimental results. The micelles we microinjected in these experiments were formed from a mixture of Bodipy TMR-PIP2 and lysophosphatidylcholine (lysoPC) for technical reasons discussed in the Supplemental Material. Because the injections cause some cells to form large blebs (Sheetz et al., 2006 ; Charras et al., 2006 ; Baumgart et al., 2007 ), we were able to compare the D of PIP2 on blebs, which presumably lack much of the cytoskeleton and intrinsic membrane proteins associated with the cytoskeleton, with the D of PIP2 in native Rat1 plasma membranes.
Figure 2 shows autocorrelation functions computed from the signals recorded from a large bleb (green) on a Rat1 cell and from the plasma membrane (red) of a Rat1 cell that lacked blebs; both cells were in the same dish, and they were injected at about the same time. The FCS measurements determine the correlation times of the fluorescent PIP2 diffusing in the inner leaflet of the plasma membrane. (Equations S1 and S3 in Supplemental Material show the correlation time is approximately the midpoint of the curves in Figure 2. If the confocal volume contains one diffusing fluorescent PIP2, this corresponds to the residence time, or average diffusion time required for the lipid to exit the illuminated region, radius r = 0.19 μm). We calculate the diffusion coefficient, D, from the Einstein relation, r2 = 4Dt, where r is the radius of the confocal volume (calibrated daily by measuring the correlation time of rhodamine, which has a known diffusion coefficient) and t is the correlation time (Rigler and Elson, 2001 ). For the data in Figure 2, we calculated the correlation time for Bodipy TMR-PIP2 in the Rat1 bleb as 3.2 ms and in the plasma membrane as 10.2 ms; the diffusion constants corresponding to these correlation times are D = 2.8 and 0.9 μm2/s, respectively. Thus, these two measurements indicate PIP2 diffuses threefold faster in the bleb than in the normal membrane.
The average ± SD D value for Bodipy TMR-PIP2 in an unperturbed inner leaflet of the plasma membrane of Rat1 cells was 0.9 ± 0.2 μm2/s (n = 40), similar to the average value we measured in all experiments with the four fibroblast and two epithelial cell lines, 0.8 ± 0.2 μm2/s (n = 218); the experiments used three different types of micelles containing three different chemical forms of Bodipy TMR-PIP2 (see Supplemental Material). The average D in the Rat1 blebs, 2.5 ± 0.8 μm2/s (n = 26), was 2.8-fold higher than D in the intact Rat1 plasma membrane. Figure 3 is a bar graph summarizing these results. The simplest interpretation of the results (Eq. 1) is that two thirds of the PIP2 in a Rat1 plasma membrane is not free to diffuse, possibly because it is bound reversibly to cytoskeletal proteins that are present at lower concentration in the blebs than in the unperturbed membrane.
We also measured the diffusion coefficient of fluorescent PIP2 incorporated from LysoPC/Bodipy TMR-PIP2 micelles into palmitoyl-oleoyl phosphatidylcholine (POPC)/palmitoyl-oleoyl phosphatidylserine (POPS) (9:1) GUVs using the same FCS technique and temperature (25°C); Figure 3 shows D = 3.7 ± 0.8 μm2/s (average ± SD; n = 6) under these conditions. (As expected, this is slightly faster than the D measured at 21°C: D = 3.3 ± 0.8 μm2/s [n = 23] using same FCS technique; Golebiewska et al., 2006 ). Thus, the D of PIP2 in blebs formed on the plasma membranes of Rat1 cells is comparable with the value observed on pure phospholipid membranes, where the PIP2 diffuses freely (Figure 3). The lower D in the blebs (2.5 vs. 3.7 μm2/s) could be due to presence of cholesterol, which is known to reduce the D of lipids in a fluid membrane (Almeida and Vaz, 1995 ; Korlach et al., 1999 ; Filippov et al., 2003 ), although the mechanism by which this occurs is controversial (Almeida et al., 2005 ). We confirmed with FCS measurements that incorporating physiological mole fractions of cholesterol can decrease D in a POPC/POPS GUV to the value of D measured in blebs (data not shown).
We measured D in the outer leaflets of Rat1, REF52, and HEK293 cells by adding Bodipy-TMR PIP2 micelles to the solution bathing the cells; Figure 3 shows D = 2 ± 1.3 μm2/s (average ± SD; n = 65). Our data agree with FRAP measurements of fluorescent PIP2 diffusion on the outer leaflet of N1E-115 neuroblastoma cells (van Rheenen and Jalink, 2002 ). The 2.2-fold disparity in PIP2 diffusion on the inner and outer leaflets of plasma membranes again suggests that much of the PIP2 on the inner leaflet is not free to diffuse, presumably because it is bound or sequestered in some manner.
One possible explanation for why PIP2 diffuses significantly more slowly on the inner leaflet than either the outer leaflet or blebs is its reversible electrostatic association with basic clusters on proteins. As discussed below, many transmembrane proteins contain clusters with four or more basic residues on their cytoplasmic juxtamembrane regions (Nilsson et al., 2005 ), and many peripheral membrane proteins use clusters of basic residues to bind to the plasma membrane. Our experiments, however, do not exclude other possibilities. For example, PIP2 and other lipids on the inner leaflet could have more tortuous diffusion pathways than lipids on the outer leaflet or in blebs. We tested this possibility by measuring the diffusion of the lipid DiD on the inner leaflet of REF52 cells. DiD (also known as DiI-C18) has a single positive charge, so it is unlikely to interact electrostatically with basic clusters on proteins. The charge is delocalized, and there is good experimental evidence this molecule rapidly flip-flops across bilayers (Melikyan et al., 1996 ). Thus, even if we add it to the outside of the cell, we expect most of the DiD to be located on the inner leaflet: both the negative transmembrane potential and the negative electrostatic surface potential of the inner leaflet (produced by acidic phospholipids) favor this distribution. As expected, D for DiD was the same, within experimental error, whether the lipid was added to the inside (3 ± 1 μm2/s, average ± SD; n = 11) or outside (3 ± 1 μm2/s; n = 12) of cells. These values of D agree with those reported for FRAP measurements of DiD diffusion (1–4 μm2/s) in cell membranes (Fulbright et al., 1997 ; Weisswange et al., 2005 ). The observation that DiD diffuses more rapidly than PIP2 support the interpretation that the D of PIP2 on the inner leaflet is lower because approximately two thirds of the polyphosphoinositide is bound reversibly. As discussed in the Supplemental Material, however, the measurements are difficult, particularly when DiD is added to the bathing solution. Thus, we repeated the experiments on the plasma membrane inner leaflets with rhodamine-labeled phosphoethanolamine (PE).
Rhodamine-PE is a more typical fluorescent lipid: it has a net charge of −1 (vs. −4 for PIP2, and it should not flip-flop rapidly across the plasma membrane bilayer (Melikyan et al., 1996 )). Our FCS measurements produced an average ± SD D value for rhodamine-PE on the inner leaflet of HEK293, Cos1, HeLa, or Rat1 cells of 1.5 ± 0.9 μm2/s (n = 107), 1.9-fold higher than the average D of PIP2, 0.8 ± 0.2 μm2/s. This value agrees with previous measurements of lipid diffusion in cell membranes (Jacobson, 1983 ). In contrast to PIP2, rhodamine-PE did not diffuse significantly faster in the blebs (D = 1.5 ± 0.4; n = 5) than in the unperturbed plasma membranes of Rat1 cells. The rhodamine-PE results argue against the possibility that lipids in general have tortuous diffusion pathways on the plasma membrane inner leaflet, lending support to the hypothesis that reversible association with proteins is responsible for the reduced value of D for PIP2 we observe.
The fluorescent PIP2 we injected into the cell was >90% pure, based on TLC (data not shown). Methods to analyze phosphoinositides have been discussed previously (Rusten and Stenmark, 2006 ). Microinjected Bodipy TMR-PIP2 could, however, be degraded by phosphoinositide-specific PLCs (Rebecchi and Pentyala, 2000 ; Rhee, 2001 ) and by lipid phosphatases. PLCs, for example, could hydrolyze a fraction of fluorescent PIP2 to produce fluorescent diacylglycerol (DAG), which should diffuse more rapidly than PIP2 because it will not be electrostatically sequestered by basic clusters on proteins. If this occurs to a significant degree, our estimate that two thirds of the PIP2 is bound reversibly could be an underestimate. Thus, we did three important control experiments. First, we measured the D of Bodipy TMR-phosphatidylinositol 3,4-bisphosphate [PI(3,4)P2] in the inner leaflet; PLCs prefer PI(4,5)P2 and they do not hydrolyze this lipid rapidly (Serunian et al., 1989 ). Nevertheless, PI(3,4)P2 binds to unstructured basic clusters with the same affinity as PI(4,5)P2, as expected for nonspecific electrostatic interactions (Wang et al., 2002 ). Second, we measured D in cells treated with a lipid phosphatase inhibitor. Third, we reduced the time between microinjection and FCS measurements.
FCS measurements of Bodipy TMR-PI(3,4)P2 microinjected into Cos1 and Rat1 cells produced a D = 0.9 ± 0.2 μm2/s (average ± SD; n = 47), which agrees within experimental error with the value obtained with Bodipy TMR-PI(4,5)P2 (0.8 ± 0.2 μm2/s). This suggests PLCs did not hydrolyze a major fraction of fluorescent PI(4,5)P2 into DAG in our experiments. We examined the effects of phosphatases by exposing NIH3T3 and HEK293 cells to 0.5 μM of the lipid phosphatase inhibitor RV001 (Rosivatz et al., 2006 ) ≥10 min before the FCS measurements (RV001 was a generous gift of Drs. Rudiger Woscholski [Imperial College, London] and Banafshe Larijani [Cancer Research UK, London]). The average D of Bodipy TMR-PI(3,4)P2 in cells exposed to phosphatase inhibitor, 0.7 ± 0.3 μm2/s (n = 12), did not differ significantly from the value obtained without inhibitor.
For most FCS measurements we first microinjected about a dozen cells at a time using a fluorescence microscope with a long working distance objective and attached microinjector, then moved the dish to the adjacent FCS microscope for the diffusion measurements. The time from microinjection to measurement of D was 10–30 min, during which lipid phosphatases (or PLCs) could convert a significant fraction of the fluorescent PIP2 to phosphatidylinositol monophosphate, phosphatidylinositol, or DAG. We did not expect this to be an important problem because we starved the cells for 1–2 h, and we maintained them at 25°C (rather than 37°C) for the injection and diffusion measurements. Nevertheless, we tested our assumption by mounting the microinjector on the FCS confocal microscope and making FCS measurements of D of Bodipy TMR-PIP2 as rapidly as possible (<2 min) after microinjection of a single Rat1 cell. These measurements yielded D = 0.7 ± 0.3 μm2/s (average ± SD; n = 6), a value similar to that obtained using our standard protocol (0.8 ± 0.2 μm2/s; n = 218). We also measured D in cells starved for 4–6 h before microinjection: again, the results, D = 0.9 ± 0.2 μm2/s (n = 34), did not differ significantly from those reported above.
Our final experiment was a positive control: we tested the effect of maintaining cells at 37°C for 45 min after microinjection. Specifically, we microinjected the cells with Bodipy-TMR PIP2, returned them to a 37°C incubator for 45 min., and then performed FCS measurements at 25°C. This protocol increased D significantly, to 1.2 ± 0.3 μm2/s (average ± SD; n = 50). This result suggests phosphatases or lipases may hydrolyze a significant fraction of the PIP2 at 37°C.
Our main result (Figure 3) is that the diffusion coefficient, D, of fluorescent PIP2 on the inner leaflet of fibroblast plasma membranes is significantly (2- to 3-fold) lower than the D of PIP2 in blebs or in the outer leaflet, and the D value of other lipids (rhodamine-PE and DiD) on the inner leaflet. The simplest interpretation is that approximately two thirds of the PIP2 on the inner leaflet is bound reversibly to cytoskeletal or other membrane proteins. (Note that PIP2 must be bound reversibly on the time scale of the FCS experiment, ~10 ms. The Supplemental Material includes a discussion of why this is reasonable for electrostatic interactions with unstructured basic clusters.)
Our measurements and conclusion agree well with the other reported experiment we know of that addresses this question directly. Hagelberg and Allan (1990) measured the phospholipid composition of microvesicles from erythrocytes and noted the only phospholipids depleted in the microvesicles were “polyphosphoinositides: the microvesicles only contained about half as much of the polyphosphoinositides as the original cells (compared on the basis of total phospholipids content), suggesting that only half of these lipids are free to diffuse.” They suggested PIP2 could be bound to proteins such as glycophorin, which contains a cluster of four basic residues in its juxtamembrane cytoplasmic domain and binds PIP2 (Anderson and Marchesi, 1985 ). Glycophorin and Band III (and presumably other cytoskeletal proteins) are depleted in the microvesicles from erythrocytes.
Our FCS experiments measure the average ± SD D value of PIP2 in the plasma membranes of fibroblasts and epithelial cells. Our estimate of D on the inner leaflet (0.8 μm2/s) from FCS measurements agrees well with an estimate (0.7 μm2/s) from measurements made with an indirect technique over longer distances (Yaradanakul and Hilgemann, 2007 ). Both our estimates disagree with the observation of Cho et al. (2005) that the D of PIP2 in atrial myocytes is orders of magnitude lower; Hilgemann (2007) discusses the Cho et al. (2005) measurements.
As we discuss in the Supplemental Material, our conclusion that two thirds of the PIP2 on the plasma membrane inner leaflet is bound reversibly is consistent with many partitioning measurements of green fluorescent protein (GFP)-pleckstrin homology (PH) domains (from both phospholipase C-δ1 and pleckstrin) between the plasma membrane and the cytoplasm. We conclude the free concentration of PIP2 is C = 10 μM, and the sequestered concentration is S = 20 μM (concentrations calculated by assuming PIP2 is dissolved in the cytoplasm of a hypothetical 10-μm-diameter cell).
If the plasma membrane contains 20 μM PIP2 that is reversibly bound, where is it? One possibility is that it is sequestered through nonspecific electrostatic interactions with clusters of basic residues on proteins (McLaughlin and Murray, 2005 ). The many membrane proteins capable of sequestering PIP2 include ~40 members of the Ras, Rab, Arf, and Rho families of GTPases that have basic clusters and use them to interact with PIP2 (Heo et al., 2006 ). Moreover, the major protein kinase C (PKC) substrate MARCKS has a basic cluster that binds to the membrane and sequesters three PIP2 (Gambhir et al., 2004 ). MARCKS has been estimated to be present at a concentration ~10 μM in neuronal tissue and at somewhat lower concentrations in other cells; so, MARCKS by itself could sequester a significant fraction of the PIP2 (McLaughlin and Murray, 2005 ). Many integral proteins (e.g., EGFR, pIgR, glycophorin, and syntaxin) have basic clusters on their cytoplasmic juxtamembrane domains, as predicted by the “positive inside” rule formulated by von Heinje and colleagues (Nilsson et al., 2005 ). These basic clusters produce a local positive potential at the surface of the membrane that acts as a basin of attraction for the tetravalent PIP2, as discussed in detail previously (McLaughlin et al., 2002 ; Gambhir et al., 2004 ; Wang et al., 2004 ; McLaughlin and Murray, 2005 ). Experiments confirm (Gambhir et al., 2004 ) PIP2 is concentrated ~100-fold in the neighborhood of the basic cluster (i.e., PIP2 interacts with a basic cluster with an electrostatic energy of ~3 kcal/mol). Engelman (2005) made the strong point that we should expect a highly nonuniform distribution of a lipid in the plasma membrane except when its interaction energy with proteins is less than thermal energy (~0.6 kcal/mol). The interaction energy of PIP2 with a basic cluster (>4 basic residues) is much greater than thermal energy. Thus, a tentative picture of PIP2 distribution in the plasma membrane emerges: approximately two thirds should be concentrated 100-fold in local neighborhoods (thickness ~1 nm = Debye length; see McLaughlin and Murray, 2005 ) adjacent to basic clusters on proteins such as K-Ras, EGFR, MARCKS, and syntaxin, whereas the remaining one third is free to diffuse and distributes randomly over the bulk of the membrane (~100-fold larger area). As discussed in the Supplemental Material, many cells have regions where proteins with clusters of basic residues are concentrated, e.g., MARCKS in ruffles and nascent phagosomes; syntaxin in regions where exocytosis occurs.
If a significant fraction of plasma membrane PIP2 is sequestered, could physiological factors, such as a local increase in Ca2+ or activation of PKC, release it? Experiments on MARCKS and peptides corresponding to the basic domains from e.g., MARCKS, EGFR, and gravin show these domains laterally sequester PIP2. Additional experiments show Ca/CaM can bind with high affinity to the basic clusters on MARCKS (Kd = 30 pM) and EGFR (Kd = 10 nM), and rapidly rip them from phospholipid vesicles, releasing the sequestered PIP2 (McLaughlin and Murray, 2005 ). There is ample evidence that cells experience local increases in Ca2+—and consequently Ca/CaM—in response to many stimuli (Clapham, 1995 ; Chandler et al., 2003 ; Hilgemann, 2007 ; Muller et al., 2007 ); can local increases in Ca/CaM produce a local increase in the level of free PIP2? The Supplemental Material includes data from experiments attempted to release some of the sequestered PIP2. For example, we reversed the net negative charge on the cytoplasmic leaflet with amphipathic weak bases. This should cause dissociation of some basic clusters from the membrane: as we hoped, this procedure significantly increased the D value of PIP2.
PKC phosphorylation of MARCKS can also cause desorption of the protein from plasma membrane to cytoplasm and release of sequestered PIP2 in a model system (Murray et al., 2002 ). Importantly, J. Sable, M. Sheetz, and colleagues have evidence that PKC activation increases the level of free PIP2 in fibroblasts (personal communication). Specifically, PKC-induced desorption of MARCKS from the plasma membrane resulted in a twofold decrease in the level of cytoplasmic GFP-PH and a concomitant rise in the membrane bound GFP-PH. The simplest interpretation of this result is that MARCKS binds a significant fraction of the PIP2 in the plasma membrane and that translocation of this protein from membrane to cytoplasm increases the level of free PIP2 in the plasma membrane. This experiment provides the strongest evidence to date that the level of free PIP2 in the plasma membrane of living cells can be modified by protein translocation.
We thank Richard A. Anderson, Michael Edidin, Donald Hilgemann, Bertil Hille, Ken Jacobson, and Suzanne Scarlata for valuable discussions and comments. This work was supported by National Institutes of Health grant GM-24971 (to S.M.).
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E07-12-1208) on February 6, 2008.