PI4P and PI(4,5)P2 Are Essential But Independent Lipid Determinants of Membrane Identity

See allHide authors and affiliations

Science  10 Aug 2012:
Vol. 337, Issue 6095, pp. 727-730
DOI: 10.1126/science.1222483

Phosphoinositide Contributions

To study the roles of phosphoinositides in the plasma membrane of mammalian cells, Hammond et al. (p. 727, published online 21 June; see the Perspective by Fairn and Grinstein) engineered phosphatase molecules that could be targeted to the membrane on demand, where they would alter the concentrations of the phospholipids phosphatidylinositol (4,5)-bisphosphate [PI(4,5)P2] and phosphatidylinositol 4-phosphate (PI4P). PI4P was thought to provide a major source for the synthesis of PI(4,5)P2, but depletion of PI4P did not have much affect on synthesis of PI(4,5)P2. Instead, PI4P appears to help to establish the negative charge at the membrane and thus promote electrostatic interactions with positively charged amino acids in membrane-associated proteins and influencing function of ion channels.


The quantitatively minor phospholipid phosphatidylinositol (4,5)-bisphosphate [PI(4,5)P2] fulfills many cellular functions in the plasma membrane (PM), whereas its synthetic precursor, phosphatidylinositol 4-phosphate (PI4P), has no assigned PM roles apart from PI(4,5)P2 synthesis. We used a combination of pharmacological and chemical genetic approaches to probe the function of PM PI4P, most of which was not required for the synthesis or functions of PI(4,5)P2. However, depletion of both lipids was required to prevent PM targeting of proteins that interact with acidic lipids or activation of the transient receptor potential vanilloid 1 cation channel. Therefore, PI4P contributes to the pool of polyanionic lipids that define plasma membrane identity and to some functions previously attributed specifically to PI(4,5)P2, which may be fulfilled by a more general polyanionic lipid requirement.

The quantitatively minor phospholipid, phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2], is found on the inner surface of the plasma membrane (PM), where it acts as a molecular gatekeeper of both cell signaling and molecular traffic (13). Its major route of synthesis (Fig. 1A) is by phosphorylation of phosphatidylinositol (PI) by PI 4-kinases (PI4K or PI4K2), making phosphatidylinositol 4-phosphate (PI4P), which is then phosphorylated at the 5-position by PI4P 5-kinase (PIP5K). PI4P is generated in many cellular membranes, particularly in the Golgi apparatus, where it is crucial for function (4). Direct evidence for the presence of PI4P in the PM was scarce (5, 6), and the tacit assumption has been that it resides there solely for PI(4,5)P2 synthesis.

Fig. 1

Independent depletion of PM PI4P and PI(4,5)P2. (A) Synthesis of PI(4,5)P2, and effects of inhibitors and activators. DAG, diacylglcerol. (B) Effect of LY294002, PAO, and ionomycin on PI4P and PI(4,5)P2 measured by mass spectrometry (open bars) or staining (filled bars; means ± SEM, n = 3 to 4). (C) Generation of PJ, a fusion of sac and INPP5E phosphatase domains with FKBP, and its rapamycin-induced recruitment to a PM-targeted FRB domain (Lyn11-FRB). CFP, cyan fluorescent protein; mCherry, a red fluorescent protein. (D) Effect of PJ, PJ-Sac (with inactivated INPP5E domain), or INPP5E (lacking the sac domain) on PI4P and PI(4,5)P2 staining intensity after PM recruitment for 2 min with 1 μM rapamycin. Histograms are means ± SEM (n = 4 to 5); gray peaks are the frequency of occurrence of cells with the indicated staining intensity for mock-transfected cells. (E) Effect of PJ constructs on PM recruitment of PI4P/PI(4,5)P2-binding green fluorescent protein (GFP)-PH-Osh2x2 (example images) and the PI(4,5)P2-selective PH-PLCδ1 and Tubbyc domains (means ± SEM of 10 to 18 cells).

Inhibitors of PI4K activity such as LY294002 and phenylarsine oxide (PAO) cause depletion of cellular PI4P, with only minor effects on the total amount of PI(4,5)P2 (5, 7, 8). We confirmed this in COS-7 (African Green monkey fibroblast) cells using either specific immunocytochemical probes (5) or mass spectrometry (9) (Fig. 1B). As a positive control, activation of phospholipase C (PLC) with ionomycin (10) caused depletion of both lipids. Although mass spectrometry cannot distinguish regio-isomers, PI4P and PI(4,5)P2 are the predominant isomers in mammalian cells (11).

To more selectively and acutely manipulate the abundance of PM inositol lipids, we turned to the rapamycin-inducible dimerization of FKBP (FK506 binding protein 12) and FRB (fragment of mTOR that binds rapamycin) domains, which can be used to recruit enzymes to the PM (12, 13) (Fig. 1C). We generated an enzymatic chimera of inositol polyphosphate-5-phosphatase E (INPP5E), which converts PI(4,5)P2 to PI4P (12), and the S. cerevisiae sac1 phosphatase, which dephosphorylates PI4P (14). We named this fusion protein Pseudojanin (PJ), in reference to its similarity to Synaptojanin (15). PJ recruited to the PM for 2 min with rapamycin caused decreased PI4P and PI(4,5)P2 staining (Fig. 1D) and the release of the PI4P and PI(4,5)P2-binding Osh2 tandem pleckstrin homology (PH) domain (PH-Osh2x2) (7, 16) from the PM (Fig. 1E). Conversely, recruitment of only an INPP5E domain had no effect on PH-Osh2x2 (fig. S2), caused small increases in PI4P staining, depleted PI(4,5)P2 staining (Fig. 1D and fig. S1E), and released PM-bound PI(4,5)P2-biosensors such as the PLCδ1 PH (PH-PLCδ1) or Tubby C-terminal (Tubbyc) domains (17) (Fig. 1E and fig. S2).

To deplete PI4P specifically, we inactivated PJ’s INPPE domain by mutation, making a chimera we call PJ-Sac. Recruitment of this enzyme to the PM caused depletion of PM PI4P staining but had no effect on PM PI(4,5)P2 staining (Fig. 1D) or localization of PH-Osh2x2, PH-PLCδ1, or Tubbyc (Fig. 1E and fig. S2). In fact, cells showing the largest degree of PI4P depletion induced by LY294002, PAO, or PJ-Sac had scarcely altered PI(4,5)P2 abundance (fig. S1, C and D). The effects of the chimeras depended on rapamycin-induced membrane recruitment (fig. S1B) and were not observed with PJ-Dead, a chimera with inactivated sac and INPP5E domains (fig. S1B). PJ did not affect Golgi PI4P or endosomal PI3P staining (fig. S3).

These observations demonstrate that most PM PI4P is not required to maintain the steady-state PI(4,5)P2 pool. However, PI4P may still act as a reserve for cellular functions associated with continued consumption, and therefore replenishment, of PM PI(4,5)P2. Such processes include clathrin-mediated endocytosis of transferrin (18), continued generation of the lipid second messengers PI(3,4,5)P3 and PI(3,4)P2, and generation of Ca2+-mobilizing inositol 1,4,5-trisphosphate (IP3). Indeed, PM recruitment of PJ or INPP5E inhibited all of these processes (Fig. 2, A, B, and C; and fig. S4). Depletion of PM PI4P with PJ-Sac, on the other hand, had no effect (Fig. 2 and fig. S4), and thus, PI4P is dispensable for maintaining the functionally relevant PI(4,5)P2 pool.

Fig. 2

Dependence of clathrin-mediated endocytosis, PI3K, and PLC signaling on PM PI(4,5)P2, but not PI4P. (A) Effect of PM-recruited PJ constructs (red) on the uptake of transferrin (AlexaFluor488, 20 μg/ml, green) over 15 min before the removal of surface bound transferrin with a pH 2.5 wash. The bar chart shows the proportion of cells showing uptake of transferrin-associated fluorescence. Data are means ± SEM (n = 4). (B) Effect of PM recruitment of PJ constructs on the PI(3,4)P2/PI(3,4,5)P3 reporter GFP-PH-Akt after stimulation of serum-starved cells for 2 min with insulin-like growth factor 1 (IGF-1). Data are means ± SEM from 15 to 26 cells. (C) Effect of PM-recruited PJ constructs on PLC-mediated Ca2+ signals (monitored with Ca2+ indicator Fluo4-AM) after stimulation of endogenous P2Y-receptors with 100 μM adenosine triphosphate in either calcium-free (100 μM EGTA) or Ca2+-containing (1.8 mM) medium. Data are means ± SEM from 11 to 26 cells.

One possible explanation for this lack of effect is that a small fraction of the total PM PI(4,5)P2 pool is consumed during endocytosis or signaling. In contrast, activation of PLC by muscarinic M1 (8, 19) or angiotensin II receptors (7) leads to consumption of up to 90% of PI(4,5)P2. We therefore used transient overexpression of M1 receptors in COS-7 cells to investigate resynthesis of PI(4,5)P2 (Fig. 3A). Stimulation of M1-expressing cells led to reduced PI(4,5)P2 and PI4P staining, which returned to prestimulation levels after addition of the M1 receptor antagonist atropine (Fig. 3B). PM-recruited PJ-Sac had no effect on this recovery of PI(4,5)P2 staining, despite sustained depletion of PM PI4P (Fig. 3B and fig. S5). Likewise, PI(4,5)P2 biosensors showed translocation from the PM upon PLC activation, but their return to the PM after atropine addition was unaffected by PJ-Sac recruitment (Fig. 3, C and D, and fig. S6).

Fig. 3

Requirements of PI4K activity, but not PM PI4P, for resynthesis of PI(4,5)P2 after robust PLC activation. (A) In muscarinic M1 receptor–expressing cells, PLC activity is stimulated with carbachol (CCh) and inactivated by atropine. (B) Effect of PAO or PJ-Sac (recruited to the PM with rapamycin) on PI4P and PI(4,5)P2 staining before and after 2 min 1 mM CCh stimulation or after a further 3 min of 10 μM atropine treatment. Histograms are the relative frequency of occurrence of PI4P and PI(4,5)P2 staining intensities expressed as means ± SEM (n = 4). Gray peaks are data from control, unstimulated cells. (C and D) Effect of PAO or PJ-Sac recruited to the PM on PI(4,5)P2 resynthesis assayed with the PI(4,5)P2 biosensors Tubbyc-GFP (C) or PH-PLCδ1-GFP (D) during stimulation with CCh and subsequent inhibition with atropine. Data are means ± SEM of 8 to 26 cells. Images show cells coexpressing the reporters with Lyn11-FRB-CFP and PJ-Sac.

These data indicate that PM PI4P seems to be redundant for synthesis of PI(4,5)P2. Intuitively, such a result seems contradictory, given the known requirements for PI4K in this pathway. Indeed, the PI4K inhibitor PAO prevented resynthesis of PI(4,5)P2 assayed with PI(4,5)P2 staining (Fig. 3B and fig. S5) or the Tubbyc and PH-PLCδ1 reporters (Fig. 3, C and D) (7, 10). These experiments show that, despite a requirement for PI4K, PI(4,5)P2 production continues in the absence of PM PI4P, either because of the efficiency of PIP5K in consuming residual PI4P [i.e., the PI4P used for PI(4,5)P2 synthesis is synthesized ad hoc by PI4Ks], or else PI4P is supplied from other membranes (20). Either way, we conclude that the majority of PM PI4P is not required for PI(4,5)P2 synthesis.

If PM PI(4,5)P2 and its functions are independent of PM PI4P, why do cells maintain substantial quantities of PI4P? Many proteins selectively target the PM through basic amino acid stretches that interact with anionic lipid head groups (3, 21); monovalent lipids such as PI and phosphatidylserine (PS) are present at high concentrations in several membranes (22), whereas an abundance of polyanionic inositol lipids is unique to the PM (13, 22). These polyanionic lipids concentrate around stretches of polybasic residues through nonspecific electrostatic interactions, increasing binding affinity (3). We therefore reasoned that PI4P might contribute to this electrostatic interaction. We screened the localization of short peptide sequences from PM proteins before and after depletion of PI4P and/or PI(4,5)P2 (Fig. 4A and fig. S7). These included amphipathic peptides, such as the myristoylated alanine-rich C-kinase substrate effector domain (MARCKS-ED) and Rit1 guanosine triphosphatase C terminus (Rit1-tail), and lipid-anchored polybasic sequences, such as the C terminus of K-Ras (K-Ras tail) and the N terminus of cortical cytoskeleton-associated protein of ~23 kD (CAP2320). We also assayed the kinase-associated 1 (KA1) domain from microtubule-associated protein–microtubule affinity–regulating kinase 1 (MARK1), which interacts nonspecifically with acidic lipids (23). In all cases, combined removal of PI4P and PI(4,5)P2 caused depletion of the proteins from the PM (Fig. 4A and fig. S7), with little effect when either lipid was depleted alone (13). Proteins that retained a secondary membrane targeting motif, such as prenylated K-Ras tail, were still found in the PM but were no longer enriched there compared with the amounts in other [presumably negatively charged (22)] membranes (Fig. 4A and fig. S8A). These effects were due to nonspecific electrostatic interactions, because no effect was seen on the PS-specific lactadherin C2 domain (22) or the C terminus of H-Ras, which interacts with the membrane solely through its hydrophobic lipid moieties (Fig. 4A and fig. S7). Measuring K-Ras tail’s PM dissociation rate by fluorescence recovery after photobleaching (24) after PI4P and/or PI(4,5)P2 depletion revealed that the two lipids made similar contributions to the protein’s electrostatic interactions with the PM in vivo (fig. S8).

Fig. 4

(A) Targeting of proteins with polybasic motifs to the PM by PI4P and PI(4,5)P2. Representative images before and after rapamycin treatment, and dissociation index of cells transfected with the indicated GFP-tagged motifs, Lyn11-FRB-CFP and the indicated PJ construct. Lact-C2, lactadherin C2 domain. Data are means ± SEM for 8 to 19 cells. (B to D) Capsaicin-induced currents in human embryonic kidney (HEK) 293 cells expressing TRPV1 are enabled by the presence of either PI4P or PI(4,5)P2. (B) Specimen showing rapamycin-induced inactivation in the presence of PJ. (C) Time course of the capsaicin-induced current in the presence of PJ, INPP5E, PJ-Sac, or PJ-Dead. (D) Percentage of inhibition at the end of the 90-s coapplication of rapamycin (1 μM) (n = 8). (E to G) Menthol-induced currents in HEK293 cells expressing TRPM8 depend primarily on the presence of PI(4,5)P2. (E) Representative specimen showing the effect of PJ. (F) Time course of the menthol-induced current in the presence of PJ, INPP5E, PJ-Sac, or PJ-Dead. (G) Percentage of inhibition at the end of the 90-s coapplication of rapamycin (1 μM) (n = 7).

PI(4,5)P2 has been proposed to be a molecular switch that restricts the activity of several ion channels to the PM (25), a phenomenon that can be highly specific for PI(4,5)P2 (1, 2628). We wondered whether this is typical for all channels or whether some have a more general polyanionic lipid requirement, which can also be fulfilled by PI4P. For example, the heat and capsaicin-activated transient receptor potential vanilloid 1 (TRPV1) cation channel can be both inhibited and activated by PI(4,5)P2 and possibly PI4P (29). Translocation of PJ-Sac or INPP5E had no effect on prolonged (Fig. 4, B to D) or repetitive (fig. S9) capsaicin activation of TRPV1, but it was inhibited when both PI4P and PI(4,5)P2 were depleted by PJ (Fig. 4, B to D, and fig. S9). Therefore, it appears that either lipid is sufficient for TRPV1 channel activity. However, this does not apply to all lipid-activated cation channels. For example, the menthol-activated transient receptor potential melatastatin 8 (TRPM8) channel is specifically dependent on PI(4,5)P2 (12) and was inhibited by PI(4,5)P2 depletion, but not by removing PI4P with PJ-Sac (Fig. 4, E to G).

Our results reveal an unanticipated role for PI4P at the PM of cells: Most of it is not required to support synthesis of PI(4,5)P2. Rather, PI4P makes an autonomous contribution to the polyanionic lipid pool that defines the inner leaflet of the PM, a function it shares with PI(4,5)P2. We suggest that PI4P fulfills the need of any PM functions that simply require polyvalent anionic lipids. This leaves PI(4,5)P2 free to undergo rapid turnover and regulate its large repertoire of specific effector proteins, which may decrease its effective free concentration, without deleteriously perturbing the unique and defining electrostatic properties of the PM.

Supplementary Materials

Materials and Methods

Figs. S1 to S9

References (3037)

References and Notes

  1. Acknowledgments: We thank M. Lemmon, K. Moravcevic, D. Oliver, D. D. Saur, and L. Stephens for helpful discussions and constructs. G.R.V.H. and R.F.I. were supported by the Wellcome Trust and the Isaac Newton Trust, M.J.F. by the Alexander von Humboldt Foundation and the Isaac Newton Trust, K.E.A. by the UK Biotechnology and Biological Sciences Research Council, A.K. by a Dame Rosemary Murray Scholarship, and T.B. by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), NIH. We thank V. Schram of the NICHD Microscopy and Imaging Core for technical assistance with fluorescence recovery after photobleaching experiments. Constructs used in this work are available from
View Abstract

Navigate This Article