Structure and Regulation of Phospholipase Cβ and ε at the Membrane

Abstract

Phospholipase C (PLC) β and ε enzymes hydrolyze phosphatidylinositol (PI) lipids in response to direct interactions with heterotrimeric G protein subunits and small GTPases activated downstream of G protein-coupled receptors (GPCRs) and receptor tyrosine kinases (RTKs). PI hydrolysis generates second messengers that increase the intracellular Ca²⁺ concentration and activate protein kinase C (PKC), thereby regulating numerous physiological processes. PLCβ and PLCε share a highly conserved core required for lipase activity but use different strategies and structural elements to autoinhibit basal activity, bind membranes, and engage G protein activators. In this review, we discuss recent structural insights into these enzymes and the implications for how they engage membranes alone or in complex with their G protein regulators.

Introduction

Phosphatidylinositol (PI) lipids make up only ~1% of the total phospholipid population in cellular membranes, but have an outsized impact on cellular physiology and signaling. Phosphatidylinositol-4,5-bisphosphate (PIP₂) is the most abundant of the PI species, making up ~45% of the total PI pool, and is located primarily in the inner leaflet of the plasma membrane (1). There, it contributes to diverse processes, including signal transduction, ion channel regulation, exocytosis and endocytosis, and regulation of the actin cytoskeleton (2,3). PIP₂ also serves as a substrate for lipid kinases and phosphatases, allowing interconversion of PIP₂ to other PIP species, including phosphatidylinositol-4-phosphate (PI4P) and phosphatidylinositol-3,4,5-triphosphate (PIP₃) (3,4).

The phospholipase C (PLC) family is among the best characterized PIP₂-catabolizing enzymes. In response to diverse cellular signals, they generate the second messengers inositol-1,4,5-triphosphate (IP₃) and diacylglycerol (DAG) (Figure 1). IP₃ diffuses through the cytoplasm and binds receptors on the endoplasmic reticulum (or the sarcoplasmic reticulum in cardiomyocytes), stimulating the release of Ca²⁺ from intracellular stores. DAG remains in the membrane and, together with the increased concentration of intracellular Ca²⁺, activates typical and atypical protein kinase C (PKC) enzymes. Whereas PIP₂ is their primary substrate at the plasma membrane, PLCs can also hydrolyze other PI species, including PI4P and PI from internal membranes, which in turn promote DAG- and PKC-dependent signaling pathways (5–11).

Figure 1. Actions of PLC at the plasma membrane.

PLCs cleave PIP₂ into IP₃ and DAG in response to diverse cellular signals. IP₃ binds IP₃ receptors to increase intracellular Ca²⁺. DAG, together with increasing [Ca²⁺], activates typical and atypical protein kinase C (PKC) enzymes, thereby regulating pathways involved in membrane remodeling, transcription, and cell growth and survival (5, 7).

Both PLCβ and PLCε subfamilies are activated downstream of G protein-coupled receptors (GPCRs) and receptor tyrosine kinases (RTKs) via direct interactions with different families of G proteins (12). Although they have a highly conserved core, unique domains that flank the core and insertions within it confer subfamily-specific regulation through mechanisms including allostery and modulation of membrane association (Figure 2) (13–16). In this review, we highlight recent advances in structural and biophysical studies that provide mechanistic insights into how PLCβ and ε are autoregulated and activated by membrane-associated G proteins and the membrane itself. We also describe outstanding questions in the field, in particular the role of PIP₂ in regulating membrane association and spatial distribution of PLCs, and the hurdles that need to be overcome to understand the fully activated states of these critical signaling enzymes.

Figure 2. Domain architecture and structure of PLCβ and PLCε.

(A) The PH domain, EF hands, TIM barrel, and C2 domain form the conserved core found in almost all PLC enzymes and which is important for lipase activity (14). PLCβ and PLCε feature a highly diverse regulatory loop in in the TIM barrel known as the X–Y linker that is involved in interfacial activation. The PLCβ proximal (prox.) and distal C-terminal domains (CTDs) are connected by an unconserved linker (CTD link.), and are required for subfamily-specific regulation, as are the PLCε CDC25 domain, C2-RA1 linker (purple), and RA domains. The Y-box insertion in PLCε (orange) is required for its activation by Rho GTPases. Domains are shown to scale. (B) The crystal structure of human PLCβ (PDB ID 3OHM (34) and 4GNK (35)) and (C) rat PLCε EF3-RA1 (PDB ID 6PMP (111), colored as in (A). The catalytic Ca²⁺ in the active sites are shown as black spheres, and disordered loops shown as dashed lines. The ends of the disordered and unconserved PLCβ CTD linker are denoted by blue asterisks. The observed N- and C-termini of each crystal structure are labelled N and C.

PLCβ membrane binding domains are not what they seem

The PLCβ subfamily plays prominent role in heterotrimeric G protein signaling by responding to the activation of G_q/11− and G_i-coupled receptors, including those for angiotensin, histamine, and endothelin (15,17–19). The distinguishing feature of these enzymes is a ~400 residue C-terminal extension immediately following the C2 domain of the conserved core (Figure 2), which plays roles in autoinhibition, membrane association, and regulation by the heterotrimeric G protein subunit Gα_q (15). Structural studies subdivided this region into two functional units called the proximal and distal C-terminal domains (CTDs). The four PLCβ isoforms (PLCβ1–4, with two splice variants of PLCβ1 (20)) differ most within this extension, which is thought to account for differences in their basal activity, degree of intrinsic membrane association, and sensitivity to activation by Gα_q (12,14,15).

The first crystal structures of PLCβ were reported over a decade ago (21,22), and revealed that the PLCβ2 core domains (PH-C2) form a relatively compact, globular structure (Figure 2B). The TIM barrel domain houses the active site and features a solvent exposed hydrophobic ridge adjacent to the active site. This ridge, first observed in PLCδ, is thought to facilitate substrate binding by inserting into the membrane leaflet (23–26). These and subsequent structures of PLCβ3 also provided insight into how lipase activity is autoinhibited by a loop called the X–Y linker within the TIM barrel (Figures 2B, 3A) (21,27). The C-terminus of the X–Y linker interacts with residues near the active site to physically block substrate binding (21). Disruption of these interactions by the membrane is discussed in greater detail in the following section.

Figure 3. Regulation of PLCβ involves conformational dynamics of its N- and C-terminal domains.

(A) In its basal state, PLCβ is autoinhibited by the X–Y linker (hot pink) and Hα2' (cyan). The PH domain is flexibly connected to the core, potentially through a partial unfolding of EF hands 1/2 (EF1/2). The proximal and distal CTDs are connected by the unconserved CTD linker (dashed gray lines), allowing the distal CTD to adopt multiple orientations with respect to the core and the membrane: 1) it can interact with the TIM barrel to regulate partitioning of the enzyme between membrane-bound and cytosolic populations, 2) it could be fully dissociated, or 3) it could associate with negatively charge lipids in the membrane. Domains are colored as in Fig. 2, with dashed lines representing disordered regions. The black equilibrium arrows illustrate conformational dynamics between the PH domain, EF1/2, and the distal CTD with the core and membrane. (B) Activated Gα_q (navy blue) binds the proximal CTD, displacing Hα2' to allosterically activate the lipase. The palmitoylated N-terminal helix of Gα_q interacts with a hydrophobic patch on the distal CTD. Interactions between Gα_q, PLCβ, and the membrane facilitate displacement of the X–Y linker and optimize the orientation of the PLCβ active site for catalysis. (C) Gβγ (lavender) is membrane localized via its prenylated Gγ subunit, and it is speculated to bind the PLCβ PH domain when it is in an extended conformation. Gβγ may also interact with other domains within PLCβ, such as the TIM barrel (76). While it is not expected to directly interact with the CTDs, its binding changes the dynamics of the distal CTD (36), suggesting long-range changes occur upon activation. Interactions between Gβγ and the membrane with PLCβ displace the X–Y linker and potentially Hα2'. (D) Rac GTPases (green) bind to the PH domain of PLCβ2 and PLCβ3, resulting in membrane translocation. Interactions between Rac, PLCβ, and the membrane likely facilitate displacement of the X–Y linker and potentially Hα2' as well.

PLCβ requires a PH domain for catalytic activity, but the domain does not bind lipids with high affinity or specificity (15,27,28). Although this domain is in close proximity to the catalytic TIM barrel in all published crystal structures, there are surprisingly few interactions between the two. This is also true of the first two EF hands (EF1/2), and contrasts with the extensive interactions formed among EF hands 3/4 (EF3/4), the TIM barrel, and the C2 domain, which are required for activity in all PLC enzymes (14,21,29). Recent studies have now revealed that the PLC core structure is more dynamic in solution than implied by its crystal structures. Cross-linking and small angle X-ray scattering (SAXS) studies of the PLCβ core domains indicate that the PH domain, and likely EF1/2, do not stably associate with the core (Figure 3A) (30–33). The physiological importance of this conformational heterogeneity is currently under investigation, but has been proposed to be an important aspect of Gβγ-dependent activation (30).

Structural studies of the PLCβ3 holoenzyme and its complex with Gα_q revealed complex roles for various elements within the CTD. The proximal CTD, corresponding to the first ~30 residues of the C-terminal extension, contains the primary binding site for the heterotrimeric G protein subunit Gα_q as well as an autoinhibitory Hα2’ helix (29,34). Under basal conditions, the Gα_q binding site is disordered, and the Hα2’ helix binds to a cleft between the TIM barrel and C2 domains (Figure 3A)(29,34). When Gαq binds to PLCβ, Hα2’ is displaced from the core, allosterically activating the enzyme. The proximal CTD is followed by a flexible and unconserved CTD linker that allows the distal CTD domain to adopt multiple conformations with respect to the rest of the protein (Figures 2A,3B). This mobility turns out to be key for full activation by Gα_q (35,36).

The distal CTD folds into a ~350 residue coiled-coil domain that serves as the primary membrane binding determinant of PLCβ, and as such is required for maximum activity (35,37–39). It is the least conserved domain across PLCβ isoforms (27% identity), and this sequence variation is presumed to underly differences in the intrinsic membrane association of the PLCβ isoforms: PLCβ1 and 4 are constitutively associated with the plasma membrane, whereas PLCβ2 and 3 are primarily cytosolic in resting cells (15,35,40). The differences in membrane association may also reflect interactions formed between the distal CTD and the PLCβ core in solution, as observed in both the crystal structure and the 3D electron microscopy (EM) reconstruction of full-length PLCβ3 in complex with Gα_q. In these structures, the distal CTD was observed to interact with residues in the hydrophobic ridge of the TIM barrel. These interdomain contacts are expected to alter the relative partitioning of PLCβ enzymes between membrane and cytosolic fractions, and contributing to autoinhibition (35,36).

The distal CTD has also been proposed to house a PIP₂ binding site, which would aid in spatial localization of the enzyme to regions enriched in PIP₂ (40,41). Although the specific residues required for PIP₂ binding are not known, atomic force microscopy (AFM) studies of PLCβ3 on compressed lipid monolayers containing PIP₂ support this hypothesis. In this system, PLCβ3 preferentially localized to regions of the monolayer surface likely enriched in PIP₂, whereas removal of the distal CTD resulted in nonspecific adsorption to the surface. Thus, the distal CTD may also increase activity by localizing the enzyme to regions with a high local PIP₂ concentration (42,43).

Finally, the membrane itself clearly plays a role in the control of PLCβ adsorption and activity. The hydrophobic ridge within the TIM barrel must insert into the membrane and the active site must be able to productively engage the PIP₂ head group. As the primary membrane binding determinant, the distal CTD also makes extensive interactions with the membrane. Membrane binding appears to stabilize its structure, evidenced by a decrease in hydrogen-deuterium exchange (HDX), as measured by mass spectrometry (36). In addition, enzymatic activity itself is highly dependent upon membrane surface pressure, with maximum activity observed at a pressure of ~30 mN/m, similar to that of native cell membranes (43–47). At higher surface pressure, lipase activity decreases, likely due to impaired insertion of the hydrophobic ridge. The concentration and distribution of PIP₂ in the membrane is clearly important for association and activation in vitro, where concentrations upwards of 20% PIP₂ are routinely used to measure activity (43,48,49). The first experiments investigating the role of PIP₂ showed that PLCβ activity increased as the PIP₂ concentration increased from ~18% to 30%. However, very high concentrations of PIP₂ inhibited activity, leading to the hypothesis that PLCβ is highly sensitive to the distribution of substrate in the membrane (46). Other studies found that increasing the PIP₂ concentrations did not increase PLCβ binding to liposomes (50–52), but the variable composition and diversity of model membrane systems used in these experiments make direct comparisons difficult.

The topography of the membrane, potentially in combination with the local PIP₂ concentration, may also be a critical factor in PLCβ association and activation. This is because the closest structural homologs of the distal CTD are Bin/Amphiphysin/RVS (BAR) domains, which sense and/or induce membrane curvature, in some cases via a PIP₂-dependent mechanism (35,53,54). Interestingly, it was recently shown that the PLCβ1 distal CTD induced membrane tubulation and contributed to caveolae formation, suggesting a functional connection to BAR domains (55). This raises the possibility that PLCβ activity may itself by regulated by membrane curvature, as detected or induced by the distal CTD. In support of this idea, it was reported that activity of the PLCβ2 core decreased as a function of negative curvature of its substrate liposomes (47). It is apparent that further studies are needed to systematically investigate the importance of surface pressure and other membrane parameters, including physiologically relevant concentrations of PIP₂ concentration and curvature, in regulating PLCβ function.

A complex role for the X–Y linker in regulating substrate access to the active site

The X–Y linker is a major regulatory element in all PLC enzymes. In PLCβ, it is divided into three regions: an unconserved N-terminal region of ~80 amino acids, followed by 10–15 acidic residues, and then a 10–15 residue C-terminal region which forms a short helix that packs like a lid over the active site and blocks substrate access to the active site (21,56) (Figures 2B, 3A). In the current model of interfacial activation for the enzyme, when the PLCβ TIM barrel interacts with the membrane unfavorable electrostatic interactions occur between the acidic stretch in the linker and the negatively charged membrane, resulting in displacement of the lid helix from the active site (21,56,57). Consistent with this model, mutations that disrupt interactions between the TIM barrel and the lid helix increase activity, as does deletion of the lid helix or the entire X–Y linker (21,27,42,56).

However, other studies suggest the acidic stretch plays a more complex role than electrostatic repulsion. Deletion of the acidic stretch or charge-reversal mutations decreased thermal stability of the PLCβ core, consistent with stabilizing interactions between the acidic stretch and basic regions on the PLCβ surface (58). A crystal structure of a PLCβ3 variant, wherein the N-terminus and acidic stretch of the X–Y linker were removed, revealed that the lid helix became disordered, even though the residues forming this element were present in the construct (56). This supports the idea that interactions between the acidic stretch and PLCβ core are needed to stabilize the interaction between the lid helix and TIM barrel. The importance of the acidic stretch and X–Y linker in regulating PLCβ3 adsorption to model membranes have also been directly investigated using AFM. Deletion of the N-terminus and acidic stretch increased adsorption to PIP₂-containing monolayers to the same extent as deletion of the entire X–Y linker. As these variants differed only by the presence of the lid helix, it illustrates that this element alone is insufficient to regulate membrane association and activity (42).

An unexpected autoregulatory element in the C-terminal extension of PLCβ

The Hα2’ autoinhibitory helix in the proximal CTD docks in a hydrophobic cleft formed by the TIM barrel and C2 domains under basal conditions (Figures 2B, 3A). Disruption of this interface by mutagenesis or deletion of Hα2’ decreased stability of the PLCβ core and increased basal activity ~50-fold, consistent with this element inhibiting activity, at least in part, by stabilizing a less active conformation of the lipase (29). However, other mechanisms of inhibition are also possible. When bound to the core, Hα2’ is positioned such that it protrudes into the projected membrane binding plane of PLCβ, which led to the hypothesis that it could act as a wedge to hinder engagement of the TIM barrel with the membrane. This hypothesis was supported by AFM imaging of compressed lipid monolayers incubated with the PLCβ3 core (PH-C2 domains) or the core plus proximal CTD (Figure 2). Although both variants adsorbed nonspecifically to the monolayer, the removal of the proximal CTD increased adsorption to the surface (43). Whether Hα2’ can be displaced like the X–Y linker via interactions with the membrane promoted by other G protein activators is not known, but as described above, it disengages when Gα_q binds.

Autoinhibition by the X–Y linker and Hα2’ are both needed to regulate basal PLCβ activity. Deletion of the X–Y linker in the presence of the proximal CTD increases activity to a submaximal threshold, as does deletion of the proximal CTD when the X–Y linker is retained (56). This dual inhibition strategy likely allows greater contextual control over activity. For example, maximum activity may only occur upon displacement of Hα2’ by binding Gα_q and concomitant displacement of the lid helix when in close proximity to highly negatively charged regions of the membrane (Figure 3A, B)(21,56,57).

Gα_q allosterically activates PLCβ, but needs an assist from the membrane

All PLCβ isoforms are directly activated by members of the Gα_q/11 family (15,17–19). Gα_q/11 must be in its GTP-bound state to bind PLCβ, and maximum activity requires full-length PLCβ (15,17,37,59,60). The crystal structure of Gα_q in complex with a PLCβ3 truncation lacking the distal CTD revealed that the primary binding site for Gα_q is located in the proximal CTD, in the region connecting the C2 domain to Hα2’. This region is disordered under basal conditions, but folds into a helix-turn-helix that binds to the canonical effector binding site of Gα_q (34), and in doing so must displace the autoinhibitory Hα2’ helix from the core (Figure 3A, B)(29). Additional contacts with Gα_q are made by the C2 domain and EF3/4, including the loop connecting EF3 and 4 that contains residues essential for the GTPase activating protein (GAP) activity of PLCβ (34,59,61,62). The structure of full-length PLCβ3 bound to Gα_q revealed an addition contact formed between a hydrophobic patch on the distal CTD and the N-terminal amphipathic helix of Gα_q. Although mutation of this interface did not alter the affinity of their interaction, it was required for maximum activation (35). These interactions may also help increase membrane binding or optimize the orientation of the complex at the membrane, given that the Gα_q N-terminal helix is dually palmitoylated (and thus membrane anchored), and the CTD is the major membrane binding domain of PLCβ (Figure 3B)(35,63).

Notably, in both truncated and full-length Gαq–PLCβ structures, there are no significant conformational changes observed in the PLCβ active site, including the structure of the lid helix in the X–Y linker. Such is consistent with a model wherein Gα_q binding does not promote any conformational change in the active site, but instead activates the lipase by releasing Hα2’ autoinhibition and enhancing membrane binding and displacement of the lid helix via interfacial activation, as discussed above. Palmitoylated Gα_q may also help spatially regulate PLCβ adsorption on the membrane, because Gα_q and PIP₂ are proposed to partition into lipid rafts (63–67). Consistent with this hypothesis, Gα_q was recently shown to be sufficient to restore localization of a PLCβ3 variant lacking the distal CTD to PIP₂-enriched regions of a compressed lipid monolayer (42)

The conundrum of PLCβ activation by the Gβγ heterodimer

The Gβγ heterodimer directly activates PLCβ1, PLCβ2, and PLCβ3, but its binding site and mechanism of activation have remained elusive for over two decades (68–70). Gβγ is released upon stimulation of G_i/o-coupled receptors, which are abundantly expressed in cellular contexts where PLCβ is present (14). For example, Gβγ-dependent activation of PLCβ occurs in response to stimulation of μ- and δ-opioid receptors to promote anti-nociception (71,72). The proximal and distal CTDs of PLCβ are dispensable for activation, meaning Gβγ must bind directly to the surface of one or more core domains (37,73). In addition, PLCβ3 can also be synergistically activated by Gα_q and Gβγ in cells and in vitro, demonstrating that Gα_q and Gβγ bind separate sites on the lipase (74,75), thereby further restricting the surface area available for Gβγ binding.

Efforts to identify the Gβγ binding site on PLCβ have focused on the TIM barrel and PH domains. A putative Gβγ binding site on the TIM barrel was identified using PLCβ-derived peptides that blocked Gβγ-dependent activation by binding to its effector binding surface (76–78). These peptides were derived from a helix in the TIM barrel domain in close proximity to the X–Y linker (21,76). Mutation of this region in the background of full-length PLCβ2 had minimal impact on basal activity and increased Gα_q-dependent activation, but essentially eliminated Gβγ-dependent activation. Thus, this region of the TIM barrel domain is clearly important for activation by heterotrimeric G proteins (76). Deletion of the PH domain also eliminated Gβγ-dependent activation (79), and chimeras in which the PLCβ PH domain was replaced that of PLCδ, which is insensitive to Gβγ, was sufficient to confer Gβγ-dependent activation (80,81). Gβγ also engages other effectors through their PH domains, including GPCR kinase 2 (GRK2) (82,83) and Ras-GAP (84). However, a chimera in which the PH domain of PLCβ2 (strongly activated by Gβγ) exchanged with that of PLCβ1 (weakly activated by Gβγ) failed to increase its sensitivity to Gβγ-dependent activation (85). Finally, the solvent-exposed surface of the PH domain, observed in crystal structures (Figure 2), is unlikely to bind directly to Gβγ, as mutagenesis studies focused on these regions have failed to knock down Gβγ activation (22)(Lyon and Tesmer, unpublished data).

The question of where Gβγ binds to PLCβ was recently revisited using intramolecular crosslinking and functional assays (30). Surprisingly, introduction of a crosslink between the PH domain and EF1/2 decreased Gβγ binding and activation of PLCβ, suggesting that conformational flexibility in this region is needed for activation. A comparison of the PLCβ PH domain structure to that of the Gβγ–GRK2 PH domain revealed that the surface equivalent to the Gβγ-binding site in GRK2 is sequestered by the TIM barrel domain in PLCβ (Figures 2B, 3A,C). Mutation of conserved hydrophobic residues on this surface of the PLCβ PH domain decreased Gβγ binding and activation without altering basal activity (30). A subsequent SAXS study showed that the PLCβ3 PH domain is flexibly connected to the rest of the core in solution (31), providing a structural rationale for the crosslinking studies. Taken together, it is clear that Gβγ binding to and activating PLCβ is dependent on dynamics in solution that are not evident in the existing crystal structures, and that multiple regions of the lipase contribute to this process (Figure 3A,C).

Gβγ has not been shown to recruit PLCβ to the membrane, nor does it increase the affinity of the lipase for the membrane surface, as measured using bulk solution methods (50–52,86). However, the membrane must be an integral part of Gβγ-dependent activation, as soluble forms of Gγ fail to stimulate PLCβ (69,70,87,88). The distal CTD is needed for maximum activation, but surprisingly, this may only be due in part to its role in membrane association. Using liposomes as a model membrane, it was shown that Gβγ-dependent activation of PLCβ2 actually increased the dynamics of the distal CTD, as measured using HDX mass spectrometry. Thus, while the distal CTD is not needed for Gβγ binding, activation by Gβγ at the membrane may involve long-range conformational changes within the lipase (36). A study using fluorescence recovery after photobleaching (FRAP) in cells co-transfected with GFP-tagged PLCβ2 and Gβγ suggested that the activated complex interacts transiently, but frequently, with the membrane (28,89). These interactions are likely sufficient to displace the X–Y linker, and potentially Hα2’, via interfacial activation (Figure 3C). Because the PLCβ PH domain must be flexibly connected to the core for activation by Gβγ to occur, the membrane may help select and/or stabilize the conformational state of PLCβ that favors Gβγ binding.

Rac GTPases activate PLCβ2 and PLCβ3

The Rac1, Rac2, and Cdc42 small G proteins are direct activators of PLCβ2, and to a lesser extent, PLCβ3 (90–92). While the link between small GTPase activity and PIP₂ hydrolysis is well established (93,94), the pathways by which these GTPases are stimulated to increase PLCβ activity have remained undefined (12,95). Like Gβγ, these small GTPases bind directly to the PLCβ core, independently of the proximal and distal CTDs (85). The crystal structure of activated Rac2 in complex with the PLCβ2 core showed that the GTPase binds exclusively to the PH domain via its switch regions (Figure 3D)(22,96). The Rac binding surface is located on the solvent-exposed face of the PH domain in crystal structures of PLCβ, on the opposite face of the domain from where Gβγ is now postulated to bind (Figure 3C, D). Mutations in the PLCβ2 PH domain that affect Rac2 binding and/or activation have minimal impact on Gβγ-dependent activation, providing functional evidence that these G proteins most likely engage distinct surfaces of the PH domain (22).

Rac-dependent activation of PLCβ also likely to require the membrane, because the Rac2–PLCβ2 structure did not show any conformational changes within the lipase active site (12,22). Interactions between the prenylated GTPases and the membrane would be expected to promote displacement of the lid helix from the active site via interfacial activation. Interestingly, FRAP experiments with GFP-tagged PLCβ2 and Rac2 showed a different membrane interaction strategy, as compared to PLCβ2 and Gβγ. Whereas Gβγ facilitated rapid, transient interactions of PLCβ2 with the membrane, Rac2 slowed the diffusion of lipase on the membrane surface (89,92). These differences could reflect the different conformational states of the lipase recognized or stabilized by Gβγ versus Rac2, or different properties in how the G proteins themselves engage with and diffuse across the membrane.

Phospholipase Cε: partitioning to diverse membranes in response to diverse receptor inputs

PLCε is the largest member of the PLC superfamily and is expressed as two splice variants in humans that differ by ~225 residues at their N-termini. The PLCε1a variant corresponds to the larger variant, which is broadly expressed across tissues and often the only PLCε variant expressed (16,97). PLCε participates in many processes, including inflammation and cancer, although its roles are best characterized with respect to calcium-induced calcium release (CICR) and hypertrophy in the cardiovascular system (16,98). The ability of PLCε to participate in these diverse cellular processes is due to its complex domain architecture that not only integrates inputs from numerous signal transduction pathways, but also initiates multiple downstream signals (Figure 4). In addition to the canonical PLC core, PLCε contains an N-terminal CDC25 domain that is a guanine nucleotide exchange factor for the Rap1A GTPase (99,100) and two C-terminal Ras association domains (RA1 and RA2) that are involved in subcellular localization and activation by Ras and Rap1A GTPases, respectively (Figures 2, 4)(16,101–104). In addition to the autoinhibitory X–Y linker, the PLCε TIM barrel also contains a second insertion, termed the Y-box, that is required for activation by RhoA GTPases (Figures 2C, 4)(105).

Figure 4. PLCε is regulated by diverse inputs.

(A) In its basal state, PLCε autoinhibited by its X–Y (hot pink) and C2-RA1 (purple) linkers. The PH, EF1/2, and RA2 domains are flexibly connected to the core, as indicated by dashed lines. The CDC25 domain may also be flexibly connected. The α_X–Y helix is near the active site and the disordered Y-box (orange). Domains are colored as in Fig. 2A. (B) Rap1A (pink) is activated downstream of G_s receptors and binds the RA2 domain and likely the core, inducing long-range conformational changes. Membrane association relieves autoinhibition by the X–Y linker, and RA1 binding to mAKAP may displace the C2-RA1 linker. Finally, the CDC25 domain activates Rap1A, generating a feed forward-activation loop. (C) Ras isoforms (coral) activated by RTKs bind to the RA2 domain and may also engage other regions of the core. The Ras–PLCε complex hydrolyzes PIP₂ at the plasma membrane. (D) RhoA is activated downstream of G_12/13 receptors and binds the PLCε core. The Y-box is not required for RhoA binding but is needed for activation and PIP₂ hydrolysis at the plasma membrane. (E) Gβγ subunits (lavender) from Gi-coupled receptors require the CDC25 and RA2 domains for activation, though it is not known whether each domain binds one Gβγ molecule, or if the two domains form a single Gβγ binding site, as shown here. Gβγ localizes the complex to the plasma or perinuclear membrane for PIP₂ or PI4P hydrolysis, respectively.

Unlike PLCβ, the membrane binding regions of PLCε are poorly characterized. The residues tentatively assigned as the PLCε PH domain were identified through a bioinformatics approach (106), and while this region appears to contribute to membrane association, it has not been shown to bind any lipid with high affinity or selectivity (31). Its PH domain and EF1/2 are dispensable for activity, and therefore it is tempting to speculate the PLCε PH domain may function similarly to that of the PLCδ PH domain, which binds PIP₂ and anchors the enzyme at the membrane for processive catalysis (107). Unlike PLCδ, the PLCε C2 domain lacks conserved residues that would allow Ca²⁺ and/or lipid binding, and therefore it likely functions as a scaffolding domain, as it does in PLCβ (21,34). The ~800 residues at the N-terminus of PLCε, which are largely uncharacterized, could also contribute to membrane association through an unknown mechanism. Thus, the membrane binding surface of PLCε may be comprised of multiple basic regions distant in the primary sequence.

The large size of PLCε (220 kDa), together with its domain complexity and the presence of large insertions in the first ~1,200 residues have made it challenging to isolate pure and functional protein in sufficient quantities for in vitro structural and biochemical characterization (Figure 2A). Thus, nearly all studies using purified protein have relied on N-terminal truncations of PLCε that lack the CDC25 and/or the PH domain and EF1/2, all of which retain basal lipase activity (31,105,106,108,109). SAXS and negative stain single particle electron microscopy (EM) reconstructions of a subset of these variants confirmed their overall structure is globular and relatively compact, with the PH domain able to adopt multiple conformations with respect to the core (31). The flexibility of the PH domain is likely due to a large insertion between EF1/2 and EF3/4 and/or through partial unfolding of EF1/2, which is poorly conserved in PLCε relative to other PLCs (106). At the C-terminus of the enzyme, its RA domains contribute differently to basal function. The RA2 domain is flexibly connected to the rest of the enzyme, as nuclear magnetic resonance (NMR) studies of the RA domains showed that they did not interact with one another in solution (110). It was also shown that the RA2 domain does not contribute to the stability of the PLCε core, and has only a minor role in basal activity, consistent with this domain being flexibly connected (111). In contrast, deletion of the RA1 domain significantly decreased stability and activity, suggesting contacts between RA1 and the core domains are needed for structural integrity and normal activity (111).

A recent breakthrough was the 2.7 Å crystal structure of a catalytically active fragment of the enzyme spanning from EF3 through the RA1 regulatory domain (EF3-RA1, Figure 2C) (111). The structure confirmed the PLCε EF3-C2 domains share a similar architecture with those of PLCβ (21,34,35), PLCδ (23), and PLCγ (112), but also revealed some important differences unique to this subfamily (Figure 2C). The C2-RA1 linker binds a hydrophobic cleft between the TIM barrel and C2 domains, altering the conformation of the TIM barrel-C2 loop. Mutations within this linker, or of residues that comprise the binding site on the core, increased activity in vitro and in cells, consistent with an autoinhibitory function for this element. This interaction site is reminiscent of the autoinhibitory contact formed between the PLCβ Hα2’ helix and the catalytic core (29), suggesting that interactions between regulatory domains and the TIM barrel-C2 interface is a conserved strategy among PLCs. The RA1 domain interacts with EF3, the TIM barrel, and the C2 domain, forming an extensive interface necessary for full stability of the enzyme. Thus, the RA1 domain is integrated into the PLCε core. The Y-box within the TIM barrel was disordered, leaving little insight into how this element contributes to activation by RhoA. Imaging the electrostatic surface of the EF3-RA1 structure revealed the only strongly positive region surrounds the active site of the TIM barrel, where basic residues would be expected to increase interactions with negatively charged lipid headgroups and coordinate substrate PIs. Thus, if there is a PLCε membrane binding domain, it is outside the EF3-RA1 core.

Surprisingly, large portions of the PLCε autoinhibitory X–Y linker were well ordered, including a highly conserved amphipathic helix, α_X–Y. In the primary sequence of PLCε, this helix is located at the start of the X–Y linker, almost immediately N-terminal to its acidic stretch, suggesting it could potentially be linked to interfacial activation. While likely disordered in solution, the α_X–Y helix is stabilized in the structure through a crystal contact with an adjacent molecule of PLCε. Deletion of α_X–Y altered basal activity in vitro and in cells, but the molecular basis for this remains unclear (111). Given the amphipathic nature of the element, it is possible that α_X–Y modulates activity by promoting interactions with the membrane, and/or interacting with other regions of PLCε not present in the existing crystal structure. In addition, the PLCε structure lacks an obvious explanation for how the X–Y linker might block the active site, indicating that autoinhibition by the linker is likely conferred via a different mechanism than that used by PLCβ.

The structure also confirms functional studies indicating that the PH domain and EF1/2 are dispensable for the structure and activity of the PLCε core, as they were removed to facilitate crystallization. Superimposing the structure of the PLCε EF3-C2 domains with those of PLCβ PH-C2 provides a reference for the location of where these domains may be within the context of larger fragments of PLCε. The PH domain and EF1/2 would be positioned to extend away from the core, on the opposite side of the molecule with respect to the RA1 domain, consistent with SAXS studies (Figure 2) (31).

The Rap1A GTPase activates PLCε at the perinuclear membrane

Rap1A is the best characterized regulator of PLCε. In response to stimulation of G_s-coupled GPCRs, including β-adrenergic receptors, Rap1A is activated through a pathway involving adenylyl cyclase and exchange protein activated by cAMP (Epac), a GEF for Rap1A (16,98). Activated Rap1A binds the RA2 domain (101,110), translocating the complex to the perinuclear membrane, where localization is maintained via interactions between RA1 and the mAKAP scaffolding protein (104,113,114). The GEF activity of the CDC25 domain also activates Rap1A, resulting in a feed forward activation loop and sustained PLCε activity at the perinuclear membrane (99,115). There, PI4P is hydrolyzed to produce DAG, stimulating a PKC-dependent pathway that results in upregulation of genes involved in cardiac hypertrophy (Figure 4A, B)(10,11,113,116,117).

Because Rap1A is prenylated, it could help localize and/or increase the affinity of the activated complex for the membrane, thereby promoting interfacial activation via rearrangement of the X–Y linker. However, this is insufficient for full activation, because a PLCε variant artificially targeted to the plasma membrane could be further stimulated by activated GTPases (110). Thus, Rap1A activates PLCε through both membrane-dependent and allosteric mechanisms. As the RA2 domain is flexibly connected to the core, the allosteric component of activation could involve the formation and/or stabilization of long-range interdomain contacts between the Rap1A-bound RA2 domain and structurally uncharacterized N-terminal regions, including the CDC25, PH, and EF1/2 domains (Figure 4B). Several lines of evidence support this hypothesis. The PLCε EF3-RA1 crystal structure revealed the N- and C-termini are close to one another, which would situate the RA2 domain near the core (Figures 2C, 4A, B)(111). Secondly, a recent study reported that the PH domain and EF1/2 are required for Rap1A-dependent activation in vitro, but not for binding the GTPase. This study also used SAXS to show that binding of activated Rap1A to a PLCε variant spanning from the PH domain through the C-terminus (Figure 2), induced conformational changes that resulted in a more compact, less flexible structure (Figure 4B) (118).

Activation of PLCε by the Ras GTPases.

PLCε is also activated downstream of RTKs through direct interactions with Ras GTPases, connecting PLCε activity to inflammation (16,119) and tumor suppression or oncogenesis depending on the tissue and cellular context (97,120–125). Ras-dependent activation results in translocation of the activated complex to the plasma membrane, where PIP₂ is hydrolyzed, initiating signaling via Ca²⁺- and PKC-dependent pathways (102,103). The mechanism of Ras-dependent activation of PLCε is thought to be similar to that of Rap1A, involving both membrane-dependent and allosteric components triggered by the GTPase binding to the RA2 domain (Figure 4C) (101,102,110,115). The crystal structure of active H-Ras bound to the isolated RA2 domain confirmed that only activated Ras engages PLCε, as extensive contacts are made between switch regions of the G protein and RA2 (110). Ras binding may promote and/or stabilize interactions between the Ras-bound RA2 domain and the PLCε core, as well as help to displace the X–Y linker as a consequence of interfacial activation (Figure 4C).

Rho GTPases activate PLCε independently of its RA domains.

PLCε is also directly activated by Rho GTPases, including RhoA, which themselves are activated downstream of G_12/13-coupled receptors (16,105,108,126), and include the thrombin (127,128), sphingosine-1-phosphate (128,129), and lysophosphatidic acid receptors (130). In contrast to Rap1A-dependent activation, stimulation of PLCε by RhoA is reported to be cardioprotective (128), and contributes to inflammatory signaling in astrocytes and other tissues (127,129). Rho GTPases also translocate PLCε to the plasma membrane for PIP₂ hydrolysis, though their binding site on the lipase has not yet been identified (Figure 4D)(16). Domain deletion studies of PLCε suggest that RhoA binds to a surface on the EF3/4-C2 fragment of the enzyme, and though the Y-box insertion in the TIM barrel is required for activation, it is not the binding site (Figure 2, 4D)(105,108). Given the proximity of the Y-box to the X–Y linker and its α_X–Y helix, it is tempting to speculate that interactions between these elements, induced by Rho GTPase binding, will be part of the Rho-dependent activation mechanism (Figure 4D).

Activation of PLCε by the Gβγ heterodimer at multiple membranes.

Gβγ-dependent activation of PLCε is best studied in the context of the cardiovascular system. There, stimulation of endothelin receptors increases lipase activity through a Gβγ-dependent mechanism, resulting in cardiac hypertrophy (10,104,113,131). Other G_i-coupled receptors are also poised to activate PLCε via Gβγ heterodimer, similar to regulation of PLCβ (14,106,109). Although Gβγ-dependent activation of PLCε is expected to require prenylation of the Gγ subunit, it has not been experimentally demonstrated. Intriguingly, Gβγ-dependent activation of PLCε has also been reported to occur at both the plasma membrane and internal membranes (11,14,131). Given that PLCβ and PLCε share ~46% sequence identity and are both directly regulated by Gβγ, it is tempting to speculate that they must use similar binding sites and mechanisms of activation. However, this appears unlikely based on a study that identified the minimal fragment of PLCε that could be activated by Gβγ. Robust activation was only achieved when both the CDC25 and RA2 domains were present (109). Because these domains are distant in the primary sequence of the enzyme, it suggests that either each domain binds a Gβγ molecule, or that the CDC25 and RA2 domains together to form a single Gβγ binding site, as supported by the proximity of the N- and C-termini in the PLCε EF3-RA1 structure (Figure 4E)(111).

Outstanding Questions and Future Directions

PLCβ and ε are recruited to cellular membranes in response to the activation of diverse receptors, hydrolyzing various PI lipid species to increase intracellular Ca²⁺ and activate PKC, and in the case of PLCε, to activate the Rap1A GTPase. Despite different elaborations to their core domains, these subfamilies share conserved themes of regulation and membrane association. Importantly, in all cases experimentally tested thus far, membrane localization (e.g. achieved through intrinsic association or mutation to install a prenylation site) or G protein binding is insufficient for full activation of the PLC. Maximum activity is achieved only with a full-length PLC enzyme, a lipidated and activated G protein, and the PI substrate presented in the context of a membrane surface. Thus, the membrane and G proteins coordinately regulate the lipase activity of PLCβ and PLCε. However, how this coordination is achieved remains unclear. The membrane could simply function as a surface that brings G proteins and PI substrates into close proximity for PLC activation, or the unique and dynamic physical and chemical properties of the membrane, such as curvature and PI concentration and spatial distribution, aid in recruiting and maintaining the most optimal configuration of the PLC enzyme at the right place and right time.

Current high-resolution crystal structures of PLCβ and PLCε, together with structural studies of the PLCδ and PLCγ subfamilies, confirm the PLC EF3-C2 core domains adopt a highly compact structure. However, solution-based studies now show that the other conserved domains, the PH domain and EF1/2, can be conformationally heterogenous in solution (30,31). The picture is further complicated by presence of additional regulatory domains, such as the PLCβ distal CTD and the PLCε RA2 domain, which had already been shown to be flexibly connected to the EF3-C2 core (35,110). The variety of conformational states that each PLC can adopt likely contributes to subfamily- and isoform-specific differences in basal activity, membrane association, and sensitivity to activation by G proteins. Consequently, it has become increasingly clear that the structure of a fully active PLCβ or PLCε enzyme has not yet been determined. All published PLCβ and PLCε crystal structures, alone or in complex with activating proteins or small molecules, preserve known autoinhibitory interactions. For example, in the PLCβ structures, including those in complex with G proteins, the lid helix of the X–Y linker is bound to the TIM barrel, occluding the active site. In the Gα_q–PLCβ complex structures, the Hα2’ helix remains bound to the core through crystal contacts with adjacent molecules (21,29,34,35). Furthermore, because Gαq was stabilized by GDP·AlF₄⁻ in both Gα_q-PLCβ3 complex structures (132), these models may represent the GAP complex between the proteins, wherein in the EF3-EF4 loop is caught in the process of turning Gα_q signaling off. When Gα_q is in a GTP-bound state, this EF3/4–Gα_q interface may not form and the complex may adopt a distinct conformation more compatible with membrane binding. Because the X–Y linkers in PLCβ and PLCε are displaced, at least in part, through a membrane-dependent process, it may therefore only be possible to capture the structure of an activated lipase on a membrane surface. However, to make such structural studies feasible, it will first be necessary to better understand how the properties of the membrane promote PLC association and activity.

The first studies investigating the membrane properties involved in PLC regulation were carried out over 20 years ago, and although limited by the techniques available at the time, they provided fundamental insights into PLC-membrane interactions, including the importance of negatively charged lipids and surface pressure (44,45,133). These studies relied on purified proteins and/or model membranes with non-physiologically relevant concentrations of PIP₂, or in the case of cell-based studies, on overexpression of fluorescently labelled proteins (40,41,46,51,52). These bulk measurements, while informative, can mask small changes in membrane association and/or residence time on the membrane, particularly if only a small proportion of the protein is needed for signaling. These questions and experimental challenges are now being revisited newer approaches, including single molecule studies with model membranes (134,135) and electron cryotomography, which can provide direct structural insights into protein-membrane complexes (136–139). Importantly, the lipase activity of PLCs itself alters the membrane, decreasing its negative charge by the removal of the IP₃ head group, and altering fluidity through accumulation of DAG. Only by direct characterization of the interplay between PLC conformation and membrane itself will we be able to fully understand how these enzymes promote human health and disease. This knowledge can lead to unique strategies to inhibit these enzymes in specific contexts/locales that will ultimately serve as the basis for future therapeutics.

Highlights.

• Recent structural and biophysical studies of phospholipase C (PLC) β and ε subfamilies have revealed new insights into basal and G protein-dependent regulation.

• The chemical and physical properties of the membrane, including its phosphatidylinositol (PI) composition, concentration, and distribution, regulate PLC association and activity

• The membrane and G proteins are co-regulators of PLCβ and PLCε, as membrane localization or G protein binding alone is insufficient for full activation.

• The conformational heterogeneity of these enzymes contributes to subfamily- and isoform-specific differences in basal activity, membrane association, and sensitivity to G protein-dependent activation.

Abbreviations:

PLC

phospholipase

PIP₂

phosphatidylinositol-4,5-bisphosphate

IP₃

inositol-1,4,5-triphosphate

PIP

phosphatidylinositol phosphate

PI

phosphatidylinositol

DAG

diacylglycerol

PKC

protein kinase C

PH

pleckstrin homology

RA

Ras association

CTD

C-terminal domain

EM

electron microscopy

AFM

atomic force microscopy

SAXS

small angle X-ray scattering

GPCR

G protein-coupled receptor

RTK

receptor tyrosine kinase

GEF

guanine nucleotide exchange factor

GAP

GTPase activating protein