Unlocking the signaling potential of GPI-anchored proteins through lipolytic cleavage

Borza et al. (2025). Trends Cell Biol DOI: 10.1016/j.tcb.2024.12.010 Citations: 8

Cell-intrinsic lipolytic release of GPI-anchored proteins was long elusive. This opinion article reviews the GPI-specific phospholipases GDE2, GDE3, GDE6 and PGAP6, stating that GDE2 cleaves as a PLD and GDE3 as a PLC, leaving phosphatidic acid or diacylglycerol on the surface.

Parts used

Key findings

  • A GPI anchor links a protein to the outer leaflet of the plasma membrane through a phosphatidylinositol (PI) lipid tail, a variable glycan core and a phosphoethanolamine linker that attaches the protein. GPI anchoring is an irreversible post-translational modification, and about 30 genes are involved in GPI-anchored protein biosynthesis in mammals.
  • GPI-anchored proteins (GPI-APs) lack a transmembrane domain and are directed to lipid rafts by the mostly saturated acyl chains of the PI lipid. GPI-APs are cleaved from the cell surface by exogenous bacterial PI-specific phospholipase C (PI-PLC) and by endogenous GPI-specific phospholipases. The human proteome comprises more than 160 GPI-APs.
  • A GPI-anchored protein is released from the cell surface by lipolytic cleavage of its anchor, by membrane-proximal proteolysis, or by both, depending on the cellular context. Four multi-pass GPI-specific phospholipases cleave GPI anchors in cis: GDE2 (GDPD5) as a GPI-specific phospholipase D (PLD), GDE3 (GDPD2) as a phospholipase C (PLC), PGAP6 (TMEM8A) as a phospholipase A2 (PLA2), and GDE6 (GDPD4).
  • GDE2 (GDPD5), GDE3 (GDPD2) and GDE6 (GDPD4) are six-transmembrane proteins with a globular alpha/beta (TIM) barrel catalytic ectodomain inserted between the last two transmembrane helices and cytosolic N- and C-terminal tails. PGAP6 (TMEM8A) is a seven-transmembrane GPI-specific PLA2 with a large N-terminal ecto-domain that mediates protein binding and with catalytic residues inside its transmembrane helices, so it acts as an intramembrane GPI-PLA2 rather than an interfacial PLA2.
  • GDE2 cleaves GPI anchors as a PLD and leaves GPI-derived phosphatidic acid (PA) exposed on the cell surface. GDE3 cleaves GPI anchors as a PLC and leaves GPI-derived diacylglycerol (DAG) exposed. PA and DAG are not normally present in the outer leaflet of the plasma membrane, and their surface-exposed signaling roles are unexplored.
  • Surface-exposed PA can be hydrolyzed by lipase H (LIPH), a membrane-associated PA-specific phospholipase A1, to the lipid mediator 2-acyl-lysophosphatidic acid (LPA). PA and DAG are conical lipids that induce negative membrane curvature, and the review proposes three non-exclusive roles for GPI-derived PA: substrate for LIPH to produce LPA, remodeling of lipid rafts and vesicle formation, and recruitment of extracellular PA-binding proteins. GDE2 activity induces extracellular vesicle formation in HEK293T and neuronal cells through an unknown mechanism.
  • PGAP6 releases GPI-anchored CRIPTO in two lipolytic steps. PGAP6 first converts GPI-anchored CRIPTO to lysoGPI-anchored CRIPTO, which is spontaneously released into the extracellular milieu, and plasma GPI-PLD (GPLD1) then releases lysophosphatidic acid (LPA) from the soluble lysoGPI form to make CRIPTO lipid-free.
  • Soluble anchorless CRIPTO binds Nodal, a TGFbeta family member, to promote embryonic development, and Pgap6 knockout mice are embryonic lethal with phenotypes similar to Cripto knockout embryos. CRIPTO release is strongly induced by multiple growth factors and by LPA.
  • Human GDE6 does not exist as a functional protein. A primate-specific single base deletion in exon 3 of GDPD4 produces an N-terminal truncation that prevents correct folding and localization, and human GDE6 accumulates in intracellular vesicles, whereas chicken GDE6 is functional at the cell surface.
  • GDE2 and GDE3 differ in GPI-AP substrate preference in reconstituted cell systems. GDE2 releases RECK, GPC2, GPC3, GPC4, GPC6, CD59 and TFPI, whereas uPAR, TIKI, contactins and CD58 are released by GDE3 and resist GDE2. GDE3 is the more promiscuous enzyme, and CNTFRalpha is listed as a GDE3 substrate. Phospholipase assays of this kind are commonly performed under overexpression, so the physiological relevance of release is not established for several listed substrates.
  • GDE2 and GDE3 also hydrolyze soluble glycerophospholipid-like substrates. GDE2 hydrolyzes glycerol 3-phosphocholine (GPCho) to release choline and is resistant to glycerol 3-phosphoinositol (GroPIns) and lysophosphatidylinositol (LPI), whereas GDE3 hydrolyzes GroPIns and LPI and is resistant to GPCho.
  • GDE3 hydrolyzes the lipid mediator lysophosphatidylinositol (LPI), a GPR55 agonist, in vitro and in vivo as a lyso-PI-PLC. GDE3 thereby limits receptor-active LPI and produces the cannabinoid receptor agonist 2-arachidonoylglycerol (2-AG), acting as a switch between GPR55 and cannabinoid receptor signaling.
  • GDE2 promotes neuroblastoma cell differentiation by releasing GPI-anchored glypican-6 (GPC6). In these cells GDE2 promotes neurite outgrowth, suppresses cell motility, confers resistance to neurite retraction by shifting the Rac/RhoA GTPase balance, and upregulates differentiation-associated genes. High GDPD5 expression correlates with favorable clinical outcome in neuroblastoma.
  • In the developing spinal cord, GDE2 releases GPI-anchored RECK, which enhances ADAM10 activity and triggers proteolytic shedding of the Notch ligand Delta. This inactivates Notch signaling in adjacent neuronal progenitors and results in neuronal differentiation, and requires cell-cell contact.
  • GDE2 sustains ADAM10 alpha-secretase cleavage of amyloid precursor protein (APP) in the brain by releasing RECK, which yields soluble APPalpha and prevents accumulation of the neurotoxic APPbeta form. Gde2 knockout mice show progressive neurodegeneration with neuronal loss, synaptic dysfunction, increased amyloid beta deposition and memory deficits, and Gde2 loss increases GPI-anchored RECK levels and suppresses ADAM10 activity.
  • GDE3 cleaves GPI-anchored urokinase plasminogen activator receptor (uPAR, PLAUR) to release soluble uPAR (suPAR) in breast cancer cells. By depleting uPAR from basolateral membrane domains, GDE3 suppresses the transformed phenotype of breast cancer cells in vitro and in vivo.
  • In oligodendrocyte precursor cells (OPCs), GDE3 cleaves GPI-anchored ciliary neurotrophic factor receptor alpha (CNTFRalpha) in a cell-autonomous manner. The soluble CNTFRalpha activates CNTF receptor signaling and prevents excessive OPC proliferation.
  • In human vascular endothelial cells (HUVECs), a large proportion of GDE2 is stored in Weibel-Palade bodies and traffics to the cell surface upon histamine stimulation. There GDE2 releases the GPI-anchored complement inhibitor CD59 and the anticoagulant TFPI.
  • In neuronal cells, GDE2 undergoes constitutive clathrin-dependent endocytosis with C-terminal Leu553 essential for internalization, and returns to the surface along fast Rab4-regulated and slow Rab11-regulated recycling routes. C-terminally truncated GDE2(delta560) strongly prefers the fast Rab4-positive route and is dysfunctional at the cell surface, whereas a further C-terminal tail deletion results in cell-surface retention and gain of GDE2 function.
  • AlphaFold3 models of GDE2 and GDE3 predict a catalytic alpha/beta barrel ectodomain, carrying the catalytic histidine, with a well-defined orientation relative to the transmembrane helices. The predicted GDE3 structure has a more open PI-binding cleft than GDE2, consistent with GDE3 being less selective and accepting lysoPI, and GDE2 and GDE3 differ in electrostatic surface charge at the catalytic site and substrate interaction surface.
  • A model based on AlphaFold3 predictions proposes that the PI acyl chains of a GPI-anchored substrate cannot be accommodated in the predicted GDE2 and GDE3 structures, and that relative movement of transmembrane helices TM2 and TM3 opens a cleft for GPI entry. This brings the active-site nucleophile close to the GPI phosphodiester, hydrolysis releases the anchorless protein and leaves GPI-derived PA (GDE2) or DAG (GDE3) at the cell surface. This mechanism is predicted and has not been confirmed by experimental structures.

Abstract

Glycosylphosphatidylinositol (GPI)-anchored proteins (APs) regulate numerous biological processes through interaction with signaling effectors at the cell surface. As a unique feature, GPI-APs can be released from their anchors by multi-pass GPI-specific phospholipases (types A2, C, and D) to impact signaling networks, phenotype, and cell fate; however, many questions remain outstanding. Here, we discuss and expand our current understanding of the distinct GPI-specific phospholipases, their substrates, effector pathways, and emerging physiological roles, with a focus on the six-transmembrane ecto-phospholipases GDE2 (GDPD5) and GDE3 (GDPD2). We provide structural insight into their AlphaFold-predicted inner workings, revealing how transmembrane (TM) domain plasticity may enable GPI-anchor binding and hydrolysis. Understanding lipolytic cleavage of GPI-APs adds a new dimension to their signaling capabilities and biological functions.