LIPIDS: Illustrating human PLD

Abstract

Phospholipase D (PLD) regulates various cellular functions through the signaling lipid phosphatidic acid (PA). New crystal structures of human PLD1 and PLD2 reveal catalytic mechanism, inhibitor binding, and regulation, enabling future structure-based inhibitor design and functional studies of PLD.

Mammalian PLD hydrolyzes the abundant membrane lipid phosphatidylcholine (PC) to generate the signaling lipid PA and choline ^1,2. PLD and PA are involved in the regulation of multiple key signaling pathways that control a variety of cellular functions such as cell proliferation and death, vesicle trafficking, cytoskeletal reorganization, and autophagy ^2–4. PLD dysfunction has been implicated in many diseases including cancer, vascular disease, virus infection, neurodegeneration, and inflammation ^2,4,5 (Fig. 1a). There are two classical PLD isoforms in mammals: PLD1 and PLD2, which are 50% identical in sequence, and have distinct regulatory and functional properties. Both isoforms contain tandem Pleckstrin homology (PH) and Phox (PX) domains that mediate membrane association, and a catalytic domain including two HKD motifs/domians and a conserved C-terminus that are essential for activity ¹ (Fig. 1b).

Fig. 1.

Structures and functions of mammalian PLD1 and PLD2

(a) Enzymatic activity and functions of mammalian PLD. (b) Diagram of domains of human PLD1 and PLD2 proteins. Regions containing polybasic residues marked with light blue form a binding pocket for PIP2. (c) A model of PLD1 regulation by phosphatidylinositol 4,5-bisphosphate (PIP2) (sticks-balls) and small GTPase RhoA (yellow cartoons) based on the PLD1 crystal structure. The structure of PLD1 catalytic domain is displayed as surface representation consisting of HKD1 (yellow), HKD2 (pink), and C-ter domains (green). A PA molecule (stick-balls) is docked in the catalytic site formed by HKD1 and HKD2. (d) Cartoon representation of the structure of an inhibitor-bound PLD2 comprised of three domains: HKD1 (yellow), HKD2 (red), and C-ter (green). A PLD inhibitor ML299 (stick-balls) is bound in the catalytic site.

PLD has been extensively studied using biochemical and cell biological techniques as well as using animal models. A key missing piece in mammalian PLD research has been a high-resolution structure. Structures would allow rational drug design and helps to better understand the regulation and isoform-specific functions of PLDs. The structure of a bacterial PLD has been crystallized ⁶, but bacterial PLDs are constitutively active, contain only one HKD motif, and share limited homology with mammalian PLDs. Thus, it remained unclear how mammalian PLD recognizes and hydrolyzes its substrate, how it is regulated by lipids and proteins, and why the C-terminus is required for activity. After more than twenty years of attempts, in this issue, Airola and colleagues report the structure of human PLD 1 (hPLD1) ⁷ while Chodaparambil and colleagues report the apo structure of hPLD2 as well as inhibitor-bound structures of hPLD1 and hPLD2 ⁸ (Fig. 1c).

To solve the structures, both groups expressed hPLD1 and hPLD2 in insect cells without the N-terminal PX-PH domains, and deleted the hPLD1-specific unstructured loop between the two HKD motifs. Removal of these sequences improved expression and crystallization of the proteins. The structures show that hPLDs have two lobes, each of which contains an HKD motif. Dimerization of the two lobes forms a funnel-shaped hydrophobic cavity leading to the active site, supporting the notion that hPLDs utilize a ping-pong catalytic mechanism that involves a covalent phosphatidyl-enzyme intermediate, similar to that proposed for bacterial PLD ⁶. Compared to bacterial PLD, hPLDs have a deeper catalytic pocket that confers specificity to PC. In contrast, the shallower and wider catalytic pocket of bacterial PLD allows it to interact with other lipids of different sizes such as phosphatidylserine and phosphatidylethanolamine. The new structures also reveal a membrane-facing polybasic patch, and show that the C-terminus of hPLDs forms a hydrophobic lip that interacts with the tail of PA/PC to properly position them in the catalytic pocket for catalysis, explaining why the C-terminus is critical for PLD activity.

Both studies complemented structure determination by biochemical and biophysical analyses that resulted in some unique observations. Focusing on the regulation of hPLD1 ⁷, Airola and colleagues demonstrated that the polybasic pocket binds to phosphatidylinositol 4,5-bisphosphate, which is required for membrane association and activity of hPLD1 (Fig. 1c). They also proposed a mechanism for the activation of PLD1 by RhoA, supporting the RhoA-binding sites previously identified by random mutagenesis ⁹. Chodaparambil and colleagues were interested in developing novel PLD inhibitors ⁸, and described the binding modes of both dual-active and isoform-selective PLD inhibitors at the catalytic pocket (Fig. 1d). Based on the inhibitor-bound hPLD2 structure, this group was able to design a new PLD inhibitor #5 that showed 8-fold increased potency as compared to the original inhibitor #4.

Although the structures reported here represent a major progress in PLD research, many critical questions still remain. For example, the current structures do not contain the N-terminal PX-PH domains, which are critical to the binding of PLD to proteins and lipids including protein kinase C (PKC) ^1,10. Many proteins and lipids activate or inhibit PLD through direct binding ^1,2,4. The most studied PLD activators, PKC, the Arf and Rho GTPases, synergistically activate PLD1 ^1,9. Full-length hPLD structure in complex with these activators would greatly help to understand the regulatory mechanism or membrane interaction of PLD. In addition, the current hPLD structures have likely captured a partially activated state, as computational studies showed that the catalytic site can only accommodate the catalytic product PA and the glycerol backbone but not the choline head group of PC. To gain insight into substrate recognition and activation, it would be critical to determine the structure of substrate-bound PLD. For example, a catalytically inactive mutant PLD, such as those with mutations in one of the two HKD motifs ¹, may prevent hydrolysis of PC during crystallization. Finally, the new inhibitor #5 derived from the hPLD2 structure has improved potency but loses specificity to PLD1. Additional studies combining mutagenesis, organic synthesis, enzymology, and structural biology will be required to develop the next generation of PLD inhibitors.

In summary, the long sought-after structures of the two human PLD isoforms described in this issue are of extreme importance to the field and are herald more structure-function and inhibitor development studies into this highly important family of lipid modifying enzymes.