Many cell types in animals and plants are polarized, which means that the cell is subdivided into functionally and structurally distinct compartments. Epithelial cells, for example, possess an apical side facing a lumen or the outside environment and a basolateral side facing adjacent epithelial cells and the basement membrane. Neurons possess distinct axonal and dendritic compartments with specific functions in sending and receiving signals. Migrating cells form a leading edge that actively engages in pathfinding and cell-substrate attachment, and a trailing edge where such attachments are abandoned. In all of these cases, both the plasma membrane and the cytocortex directly underneath the plasma membrane show differences in their molecular composition and structural organization. In this chapter we will focus on a specific type of membrane lipids, the phosphoinositides, because in polarized cells they show a polarized distribution in the plasma membrane. They furthermore influence the molecular organization of the cytocortex by recruiting specific protein binding partners which are involved in the regulation of the cytoskeleton and in signal transduction cascades that control polarity, growth and cell migration.
The functional compartmentalization of a polarized cell can be nicely illustrated in the case of a typical epithelial cell, for instance a cell of the small intestine (Figure 1). The major function of this cell type is the uptake of nutrients from the food that passes through the lumen of the gut. The apical plasma membrane of the gut epithelial cell is in direct contact with the food in the gut lumen and possesses numerous transporters and ion channels for glucose, amino acids, lipids and other important substances that have to be extracted from the food. To facilitate the uptake of substances from the food, the apical surface of the gut epithelial cells is enormously enlarged by the formation of microvilli, finger-like protrusions of the plasma membrane that are formed by a specialized arrangement of actin filaments directly underneath the membrane in the apical cytocortex. On the other hand, the basolateral plasma membrane of the gut epithelial cells is engaged in the formation of various types of intercellular junctions (Figure 1). The most apical of these junctions is the tight junction, also called ZO (zonula occludens), which forms a diffusion barrier between the apical and the basolateral plasma membrane domains and seals the intercellular space to prevent leakage of small molecules and water between the epithelial cells. Basally to the tight junction the ZA (zonula adherens) is formed, an adhesion belt composed of cadherins and catenins, which is closely linked to the cortical actin cytoskeleton. From this example it is obvious that the protein composition of the plasma membrane differs between the apical and the basolateral plasma membrane domain, and that the different composition of the membrane is reflected in a different organization of the cytocortex. In the following paragraphs we will discuss how the polarized localization of phosphoinositide membrane lipids links the polarity of the plasma membrane to the organization of the cytocortex.
Phosphoinositides as components of biological membranes
The core of the plasma membrane is formed by a bilayer of phospholipids. One of the major classes of phospholipids in biological membranes are glycerophospholipids, consisting of a glycerol moiety esterified with two fatty acid chains that protrude into the centre of the lipid bilayer and a polar head group that faces the cytosol or the extracellular space and is linked to the glycerol moiety via a phosphate group.
The polar head group of a class of glycerophospholipids, called phosphoinositides, is the cyclic alcohol inositol that can be phosphorylated at the hydroxy groups 3, 4 and 5 in seven possible combinations (Figure 2). Phosphoinositides are exclusively localized on the cytosolic face of biological membranes and show a strong specificity with respect to their distribution in different membrane compartments. PtdIns3P is enriched in endosomal membranes, whereas PtdIns4P is enriched in membranes of the Golgi apparatus and in the plasma membrane. In this chapter, however, we will focus on PtdIns(4,5)P2 and PtdIns(3,4,5)P3, two phosphoinositides that are commonly present in the cytosolic face of the plasma membrane and are important regulators of cell polarity. PtdIns(4,5)P2 was first identified as a substrate for PLC (phospholipase C). Upon stimulation of many GPCRs (G-protein-coupled receptors), activated PLC selectively binds to PtdIns(4,5)P2 and cleaves it into DAG (diacylglycerol) and Ins(1,4,5)P3 [1⇓–3], which both serve as second messengers for various intracellular signalling cascades.
By the action of a number of specific kinases and phosphatases, phosphoinositides are dynamically modified or converted into other phosphoinositides by phosphorylation and dephosphorylation reactions. PtdIns(4,5)P2 is synthesized from PtdIns in two steps. First the PI4K (phosphoinositide 4-kinase) converts PtdIns into PtdIns4P. Subsequently, the plasma-membrane-associated PI5K (phosphoinositide 5-kinase) phosphorylates the hydroxy group at position 5 of the inositol ring to produce PtdIns(4,5)P2 . Only a small fraction of PtdIns(4,5)P2 gets phosphorylated by class I PI3Ks (phosphoinositide 3-kinases) to produce PtdIns(3,4,5)P3 . This catalytic step is highly regulated and is triggered by activation of class I PI3Ks via the stimulation of a number of cell-surface receptors, including growth factor receptors, e.g. by insulin, growth hormone, nerve growth factor and epidermal growth factor . Class I PI3Ks are antagonized by the PTEN (phosphatase and tensin homologue) phosphatase, which removes the phosphate from position 3 of the inositol ring . Although several phosphoinositides are substrates for class I PI3Ks and PTEN, the phosphorylation of PtdIns(4,5)P2 at position 3 by PI3K to generate PtdIns(3,4,5)P3 and the corresponding dephosphorylation of PtdIns(3,4,5)P3 by PTEN to generate PtdIns(4,5)P2 are the most relevant modifications of phosphoinositides in the context of this chapter.
Recruitment of cytosolic proteins to biological membranes by direct binding to phosphoinositide head groups
Owing to their localization in the cytosolic leaflet of biological membranes and because their production and turnover can be spatially and temporally regulated, phosphoinositides are ideally suited to recruit cytosolic proteins to specific membrane compartments. This is important for the activation of many enzymes, including kinases and phosphatases, and for the proper assembly of the cytoskeleton in the cytocortex. The interaction of cytosolic proteins with phosphoinositides is mediated by selective lipid-binding domains that are able to discriminate between the different phosphoinositides. Well-known examples for such domains are the PH (pleckstrin homology), PX (phagocyte oxidase homology) and FYVE domains (see Table 1). A direct interaction of either one of these complex folded domains or of only a stretch of positively charged (basic) amino acids within a cytosolic protein with phosphoinositides has been described for a large number of proteins. In a proteomic approach, Catimel et al.  pulled down 388 proteins from mammalian cell extracts with PtdIns(3,5)P2- and PtdIns(4,5)P2-coated beads . Even in the absence of a defined phosphoinositide-binding domain or a contiguous stretch of basic amino acids, some proteins are capable of directly binding to phosphoinositides in the plasma membrane. Several basic amino acids which might be scattered over a longer distance within the linear protein sequence are sufficient for binding to PtdIns(4,5)P2 and PtdIns(3,4,5)P3 . By three-dimensional protein folding, these residues come together to form a positively charged pocket for the acidic head groups of the phosphoinositides.
Proteins mediating downstream effects of phosphoinositides
Local accumulation of PtdIns(4,5)P2 and PtdIns(3,4,5)P3 in specific regions of the plasma membrane results in the recruitment of certain proteins containing the corresponding lipid-binding domains. For one of these proteins, PDK1 (phosphoinositide-dependent kinase 1), a function as a master activator of various intracellular signal transduction pathways has been demonstrated . PDK1 is a constitutively active kinase that directly binds to PtdIns(3,4,5)P3 and PtdIns(3,4)P2 via its PH domain. Among the phosphorylation targets of PDK1 are classical, as well as atypical, PKC (protein kinase C; aPKC is atypical PKC) isoforms [11,12] and the kinase Akt [also known as PKB (protein kinase B)] . Like PDK1, Akt contains a PH domain specific for PtdIns(3,4,5)P3 and PtdIns(3,4)P2 [13,14] which is responsible for targeting Akt to the plasma membrane in the vicinity of its activating kinase PDK1. In addition, PtdIns(3,4,5)P3 induces a conformational change in Akt that allows its phosphorylation by PDK1 . Thus the activation of Akt by PDK1 is regulated at the level of substrate recruitment to the plasma membrane and PtdIns(3,4,5)P3-dependent changes in substrate conformation rather than by activation of the upstream kinase PDK1 .
Activated Akt dissociates from the plasma membrane and phosphorylates numerous targets in the cytosol and in the nucleus, among them key regulators of pivotal cellular processes such as apoptosis/survival, cell-cycle control, glucose metabolism, cell migration and cell proliferation . Thus overabundance of PtdIns(3,4,5)P3 in the plasma membrane, caused either by hyperactivation of the PI3K pathway or by loss-of-function of PTEN, is a frequent cause of cancer and Type II diabetes mediated by Akt signalling.
Phosphoinositides control the polarity of migrating cells
The polarized distribution of PtdIns(4,5)P2 and PtdIns(3,4,5)P3 is an ancient evolutionarily conserved mechanism to control cell polarity and cytoskeleton rearrangements in many migratory cell types. Migrating cells of the slime mould Dictyostelium discoideum accumulate active PI3K at the plasma membrane of the leading edge facing a chemoattractant, whereas PTEN localizes at the side and the rear of the cell [18,19]. It has been proposed that this complementary localization of PI3K and PTEN, which causes a corresponding polarity in the localization of PtdIns(4,5)P2 and PtdIns(3,4,5)P3 in the plasma membrane, serves to translate the very small differences in the extracellular concentration of the chemoattractant into a steeper intracellular gradient of binding sites for regulators of the cortical actin cytoskeleton . Indeed, gain- and loss-of-function of both PI3K and PTEN disrupt polarity and cytoskeleton rearrangements in migrating Dictyostelium.
Essentially the same mechanism appears to operate in human white blood cells, called neutrophils, during migration. PtdIns(3,4,5)P3 is enriched at the leading edge, whereas PtdIns(4,5)P2 is excluded from the leading edge. Furthermore, pseudopod formation, which is crucial for directed movement, depends on PtdIns(3,4,5)P3 accumulation at the leading edge [21,22]. Notably, enrichment of PtdIns(3,4,5)P3 results in a positive-feedback loop leading to increased local accumulation of PtdIns(3,4,5)P3, which depends on PI3K and Rho GTPases.
In 1998 Tamura et al.  reported that, in mouse fibroblasts, loss of PTEN results in increased cell motility, whereas overexpression of PTEN inhibits cell motility. This was confirmed by Liliental et al.  who showed that the activity of Rac1 and Cdc42 (cell division cycle 42) was increased upon loss of PTEN, and that expression of dominant-negative versions of these small GTPases abolished the PTEN loss-of-function phenotype . Together, these data clearly indicate that the polarized distribution of PtdIns(4,5)P2 and PtdIns(3,4,5)P3 in migrating cells leads to local differences in the organization of the actin cytoskeleton which are essential for protrusion formation and directional cell movement.
Phosphoinositides control the polarity of non-migratory cells
PtdIns(3,4,5)P3 has also been implicated in the polarization of hippocampal neurons. One of the earliest markers of the single neurite that is going to become the axon is the accumulation of PI3K and Akt, which directly binds to PtdIns(3,4,5)P3. Overexpression of PTEN or inhibition of PI3K abolishes polarization of hippocampal neurons, pointing to an essential function of PtdIns(3,4,5)P3 in determining the axon .
In a three-dimensional cyst model of polarizing mammalian epithelial cells, PtdIns(4,5)P2 is enriched at the apical plasma membrane, whereas PtdIns(3,4,5)P3 is confined to the basolateral membrane . PTEN localizes to the apical plasma membrane and functions to segregate PtdIns(4,5)P2 from PtdIns(3,4,5)P3. This polarized distribution of phosphoinositides apparently functions as a polarity cue, because PtdIns(4,5)P2 ectopically delivered to the basolateral membrane recruits the apical determinant annexin 2, which in turn leads to mislocalization of Cdc42 to the lateral plasma membrane. Cdc42 in turn is a central regulator of actin polymerization that controls the dynamic rearrangements of the cytoskeleton in polarized cells. Since loss of Cdc42 function results in impaired cyst formation, taken together these findings indicate that the apical recruitment of the annexin 2–Cdc42 complex by PtdIns(4,5)P2 is crucial for cyst formation. Consistent with this model, depletion of PTEN results in the unpolarized distribution of PtdIns(4,5)P2, PtdIns(3,4,5)P3 and annexin 2, which prevents proper lumen formation .
The reverse approach pursued by Gassama-Diagne et al.  underlines the importance of the differential localization of phosphatidylinositol phosphates for the determination of the apical versus the basolateral plasma membrane domains . The authors produced ectopic patches of basolateral plasma membrane by introduction of PtdIns(3,4,5)P3 into the apical plasma membrane. These patches formed protrusions that contained membrane proteins normally only present in the basolateral membrane and excluded apical proteins. Inhibition of PI3K by a specific inhibitor decreased the size of the lateral plasma membrane and overall cell height. Blocking endocytosis by expression of a dominant-negative dynamin abolished the formation of basolateral protrusions in the apical membrane domain, indicating that PtdIns(3,4,5)P3 present in the apical plasma membrane is capable of redirecting endocytosed vesicles containing basolateral membrane proteins to ectopic sites. Taken together, these data convincingly show that the ectopic insertion of PtdIns(3,4,5)P3 into the apical plasma membrane confers basolateral properties on this patch of membrane.
An important question in this context is how plasma membrane polarity, cytoskeleton organization and endocytic trafficking are connected to each other. There is now increasing evidence that this link may be provided by the small Rho GTPases Cdc42 and Rac. Cdc42 and Rac were shown to associate with and activate PI3K [28⇓–30]. Vice versa, PtdIns(3,4,5)P3 directly activates two GEFs (guanine-nucleotide-exchange factors), which promote Rac activity [31,32]. Furthermore, the activity of one of these GEFs is inhibited by PtdIns(4,5)P2 . Thus recruitment and activation of Rac by PtdIns(3,4,5)P3 directly and via Rac-specific GEFs appears to provide an efficient mechanism to locally modulate the actin cytoskeleton in migrating cells and epithelial cells. The situation is not as clear for Cdc42, since it can be recruited by both PtdIns(4,5)P2 and PtdIns(3,4,5)P3 [26,27]. Thus the existence of additional regulatory mechanisms has been postulated that either influence Cdc42 activity or function downstream of Cdc42 in the control of cell polarity .
Interactions between phosphoinositides and PAR (PARtitioning defective) proteins
One important link between Cdc42 and the regulation of cell polarity is provided by the PAR-3–PAR-6–aPKC complex. This protein complex, named the PAR complex below, is a key regulator of cell polarity in many different cell types throughout the animal kingdom [34,35]. Cdc42 binds directly to PAR-6, a core component of the PAR complex [36⇓–38]. The biochemical interaction between Cdc42 and the PAR complex is essential for Rac activation downstream of Cdc42, leading to rearrangements of the cytoskeleton followed by neurite growth and axon specification . In mammalian cultured neurons, localization of PAR-3 at the tip of the axon depends on PtdIns(3,4,5)P3 and PI3K . Overexpression of PAR-3 or PAR-6 resulted in impaired axon determination, leaving cells with two or more axon-like neurites. In Caenorhabditis elegans, the localization of the PAR complex is dependent on Cdc42 activity, either due to the direct interaction of Cdc42 with PAR-6 or indirectly via the impact of Cdc42 on the cytoskeleton [40,41]. Thus the PtdIns(3,4,5)P3–Cdc42 connection appears to be an ancient mechanism to restrict the localization and/or the activity of the PAR complex.
Recent work has revealed a direct interaction between one of the PAR proteins, PAR-3, called Baz (Bazooka) in the fruit fly, and phosphoinositides. We have shown that Baz directly binds to PtdIns(4,5)P2 and PtdIns(3,4,5)P3 via a stretch of basic amino acid residues at its C-terminal region  and this finding was confirmed for mammalian PAR-3 . For many proteins containing polybasic lipid-binding domains similar to the one in Baz it was shown that the depletion of only PtdIns(4,5)P2 or PtdIns(3,4,5)P3 does not affect their cortical localization . Only if both phosphoinositides are depleted do these proteins accumulate in the cytoplasm. The promiscuity in phosphoinositide binding allows recruitment of Baz to the cortex, even in unpolarized cells, prior to the formation of cell–cell junctions and the establishment of cell polarity. This is consistent with Baz being one of the earliest regulators of cell polarity, which is required for triggering the segregation of PtdIns(4,5)P2 and PtdIns(3,4,5)P3 into different plasma membrane domains upon full polarization of the cell.
How could Baz/PAR-3 promote the segregation of PtdIns(4,5)P2 and PtdIns(3,4,5)P3 in the plasma membrane upon cell polarization? We have previously demonstrated that Baz directly binds to PTEN and recruits it to the apical junctional region in epithelial cells . In Drosophila photoreceptor cells, Baz-dependent targeting of PTEN to the ZA is crucial for PtdIns(3,4,5)P3 accumulation in a specialized apical domain of the plasma membrane . In the polarizing ectodermal epithelium of Drosophila embryos, loss of Baz function abolishes the proper junctional localization of the synaptotagmin-like protein Bitesize, which directly binds to PtdIns(4,5)P2 . The interaction of PAR-3 with PTEN has been confirmed in mammalian MDCK (Madin–Darby canine kidney) cells and was shown to be essential for proper establishment of polarity [47,48]. Together with the finding that depletion of PTEN abolishes polarization of three-dimensional cysts of MDCK cells , recruitment of PTEN by Baz/PAR-3 can now be considered an essential functional step in plasma membrane polarization.
Manipulation of phosphoinositide signalling by bacterial pathogens
A fascinating twist to the function of phosphoinositides in cell polarity is their manipulation by bacterial pathogens in order to facilitate the entry of bacteria into the cell. Salmonella expresses a PtdIns 4-phosphatase that hydrolyses PtdIns(4,5)P2 to PtdIns5P. Loss of PtdIns(4,5)P2 from the apical plasma membrane leads to rearrangement of the cytoskeleton and the loss of the tight junctions . This breakdown of epithelial integrity allows the bacterium to overwhelm the epithelial barrier. Another bacterium, Pseudomonas aeruginosa, penetrates epithelial cells more efficiently through the basolateral plasma membrane than through the apical membrane domain. However, attachment of the pathogen to the apical plasma membrane induces focal accumulation of PI3K and PtdIns(3,4,5)P3, as well as re-localization of basolateral markers, facilitating infection of the cell by P. aeruginosa [50,51].
Phosphoinositides are membrane lipids that are unique with respect to their compartmentalized localization in the cell. In the context of cell polarity, PtdIns(4,5)P2 and PtdIns(3,4,5)P3 show a mutually exclusive localization in the plasma membrane and recruit specific binding partners to the cortex that control the organization of the cytoskeleton and various intracellular signalling pathways. On the other hand, polarity regulators, including the PAR complex, contribute to the polarized localization of phosphoinositides in the plasma membrane by localized recruitment of the lipid phosphatase PTEN. Together, phosphoinositides, PAR proteins and regulators of the cytoskeleton, in particular the small GTPases Rac and Cdc42, form a complex network of interactions that controls the polarized phenotype of many different cell types.
• Phosphoinositides are always localized on the cytosolic face of biological membranes.
• Different phosphoinositides are enriched in specific subcellular membrane compartments.
• Phosphoinositides recruit cortical proteins via specific phosphoinositide-binding domains.
• PtdIns(4,5)P2 and PtdIns(3,4,5)P3 show a polarized and mutually exclusive localization in the plasma membrane of migratory cells, epithelia and other polarized cell types.
• PtdIns(4,5)P2 and PtdIns(3,4,5)P3 are functionally linked to proteins of the PAR complex and to the small GTPases Rac and Cdc42.
• Some bacterial pathogens manipulate the phosphoinositide composition of the plasma membrane to gain entry into the cell.
The work of the authors is supported by the Forschungsförderprogramm of the University Medicine Göttingen (to M.P.K.) and by the Deutsche Forschungsgemeinschaft [grant numbers KR 390/1-1 (to M.P.K.), and grant numbers SFB 523, SFB 590, SPP 1109, SPP 1111, FOR 942, FOR 1756, DFG Research Center Molecular Physiology of the Brain) (to A.W.)].
- © The Authors Journal compilation © 2012 Biochemical Society