If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Lipid Signaling, National Center for Global Health and Medicine (NCGM), Shinjuku-ku, Tokyo, JapanDepartment of Molecular Therapy, National Institute of Neuroscience, National Center of Neurology and Psychiatry (NCNP), Kodaira, Tokyo, Japan
Department of Lipid Signaling, National Center for Global Health and Medicine (NCGM), Shinjuku-ku, Tokyo, JapanDepartment of Lipid Medical Science, Graduate School of Medicine, The University of Tokyo, Bunkyo-ku, Tokyo, Japan
The diversity of glycerophospholipid species in cellular membranes is immense and affects various biological functions. Glycerol-3-phosphate acyltransferases (GPATs) and lysophospholipid acyltransferases (LPLATs), in concert with phospholipase A1/2s enzymes, contribute to this diversity via selective esterification of fatty acyl chains at the sn-1 or sn-2 positions of membrane phospholipids. These enzymes are conserved across all kingdoms, and in mammals four GPATs of the 1-acylglycerol-3-phosphate O-acyltransferase (AGPAT) family and at least 14 LPLATs, either of the AGPAT or the membrane-bound O-acyltransferase (MBOAT) families, have been identified. Here we provide an overview of the biochemical and biological activities of these mammalian enzymes, including their predicted structures, involvements in human diseases, and essential physiological roles as revealed by gene-deficient mice. Recently, the nomenclature used to refer to these enzymes has generated some confusion due to the use of multiple names to refer to the same enzyme and instances of the same name being used to refer to completely different enzymes. Thus, this review proposes a more uniform LPLAT enzyme nomenclature, as well as providing an update of recent advances made in the study of LPLATs, continuing from our JBC mini review in 2009.
Diversity of cellular membrane glycerophospholipids
Biological membranes of mammalian cells are comprised mostly of proteins and lipids. Glycerophospholipids (phospholipids), along with sphingolipids and cholesterol, are the major lipid components. The basic structure of membrane phospholipids, with a glycerol backbone, polar headgroup, and two hydrophobic chains, provides a template for a diverse array of chemical species to support their many biological functions. These functions include formation of lipid bilayers that form hydrophobic barriers in cellular membranes, which encapsulate and compartmentalize cells as well as provide domains of molecular interactions (
). To perform these diverse functions, a wide variety of phospholipid species exists, and their compositions in cellular membranes vary among organelles, cell types, and tissues to impart the membranes with essential properties for their biological functions (
Chemical diversity of phospholipids is introduced largely by the choices of headgroups at the stereospecifically numbered (sn) -3 position and two fatty acids and linkages at the sn-1 and sn-2 positions of their glycerol-based structures (
) (Fig. 1A). For headgroups, the simplest phospholipid class, phosphatidic acid (PA), has just a phosphate. During de novo phospholipid synthesis, PA is a precursor molecule for the production of other phospholipids with more complex headgroups: phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS), phosphatidylinositol (PI), phosphatidylglycerol (PG), and cardiolipin (CL) (
). Chemical diversity is also generated from the choices of fatty chains introduced at sn-1 and sn-2. Carbon chain lengths and double-bond numbers vary, usually between 12 and 24 carbons and between 0 and 6 double bonds, and the linkages of the chains to the glycerol backbone also vary. Acyl linkages are most common at both sn-1 and sn-2; however, alkyl and alkenyl linkages also occur at sn-1 to form plasmanyl- and plasmenyl-phospholipids, respectively (
The choices of headgroups, fatty acids, and linkages are all essential features that affect phospholipid function. For the fatty acids, carbon chain lengths and double-bond numbers/positions are important attributes, which may affect biophysical and biological properties of phospholipid layers (
). The diversity of fatty chain sets is determined in two distinct steps of de novo phospholipid biosynthesis and phospholipid remodeling, as described in detail in Phospholipid biosynthesis section.
Phospholipids can be produced by several metabolic routes, and a key reaction that utilizes lysophospholipids and acyl-CoAs as substrates to produce phospholipids is catalyzed by a class of enzymes called lysophospholipid acyltransferases (LPLATs) (Fig. 1B). Chemical diversity is endowed to phospholipid molecules in two distinct phases of their biosynthesis, the Kennedy pathway (de novo pathway) (
In the Kennedy pathway, where phospholipids are newly synthesized, glycerol-3-phosphate (G3P) acyltransferases (GPATs) use G3P and acyl-CoA as substrates to produce a key intermediate in phospholipid synthesis, lysophosphatidic acid (LPA) (Figs. 1B and 2). This LPA is converted to phosphatidic acid (PA) through the introduction of a second fatty acid in a LPLAT-catalyzed reaction. This type of LPLAT reaction that produces PA may also be referred to as an LPA acyltransferase (LPAAT) reaction (
). The PA they produce may be further converted to the other classes of phospholipids via two main routes. In one route, PA is dephosphorylated by PA phosphatases (PAPs; also known as lipins) to produce diacylglycerol (DAG). DAG may be next metabolized to PC and PE, which may be further converted to PS by PS synthases (
). In the other route, PA is changed to cytidine diphosphate (CDP)-DAG by CDP-DAG synthase and further metabolized to produce PI, PG, and CL. Through these reactions, phospholipids having a variety of chain sets and all types of polar head groups are newly biosynthesized during the Kennedy pathway (Fig. 2).
Cellular phospholipids synthesized in the Kennedy pathway are further subjected to fatty chain remodeling reactions in the pathway known as the Lands cycle. In this cycle, fatty acids of phospholipids are selectively replaced by the concerted actions of phospholipase (PLA)1/2s and LPLATs (
), and these imported phospholipids may undergo Lands cycle remodeling. Multiple LPLATs with distinct substrate specificities are active in the Lands cycle, and the levels of these enzymes and the availability of their substrates are major determinants of the compositional diversity of phospholipid species in cells (
). Based on their primary structures, these LPLATs are divided into two families, the 1-acylglycerol-3-phosphate O-acyltransferase (AGPAT) and the membrane bound O-acyltransferase (MBOAT) families (Fig. 3D, see also Table 1) (
). The MBOAT family also contains DGAT1, sterol O-acyltransferase (SOAT) 1 and 2 (also called acyl-CoA:cholesterol acyltransferase 1 and 2), and several protein acyltransferases such as ghrelin O-acyltransferase (GOAT), porcupine O-acyltransferase (PORCN), hedgehog acyltransferase (HHAT), and HHAT-like (
). Thus, different members of the two families, AGPAT and MBOAT, possess primary activities to acylate not just lysophospholipids but also other protein and lipid substrates. The diversity of acylating activities among these enzymes is immense, and in some cases the specific primary biological activities of the enzymes in terms of specific substrates utilized and products generated are still not clear. As discussed in Nomenclature section, this also applies to some of the 14 enzymes functionally classified as LPLATs.
Because LPLATs are membrane-associated enzymes and utilize lipid substrates, their purification and identification had long been elusive after the discoveries of the Kennedy pathway and the Lands cycle in 1950s. However, several LPLATs were identified that functioned in either the Kennedy pathway or the Lands cycle in the 1990s and the 2000s, respectively; 40 to 50 years after the first discoveries of both pathways. Following completion of the human genome project in 2003, multiple additional putative LPLATs were identified in the DNA databases, and consequently their characterizations as LPLATs were performed in rapid succession, especially around 2004 to 2009 (
). As a result, currently a total of 14 LPLATs from the AGPAT (ten LPLATs) and MBOAT (four LPLATs) families are now known, and they are each conserved between human and mouse. In the course of these studies, many enzymes were assigned different names by different groups, and also registered in genomic databases with incorrect information of their substrate specificities as LPLATs, resulting in multiple names being assigned to each LPLAT enzyme. In several instances, the same name was assigned to completely different enzymes, which has generated some confusion in the literature and research field in referring to specific LPLATs and GPATs (Fig. 3A, Tables 1 and 2). For example, the name “lyso-PC acyltransferase (LPCAT)4” sometimes indicates either of two enzymes, MBOAT2 (Gene ID: 129642) or lyso-PE acyltransferase (LPEAT)2 (Gene ID: 254531). Another example is “AGPAT8,” which may indicate two enzymes, lyso-CL acyltransferase (LCLAT)1 (Gene ID: 253558) and GPAT3 (Gene ID: 84803). Similarly, LPCAT1 (Gene ID: 79888) and GPAT3 have each been referred to as both “AGPAT9” and “AGPAT10.”
In addition to the ambiguity and confusion caused by the overlapping of names, these names incompletely describe biochemical characteristics of the enzymes. Most LPLATs utilize multiple lysophospholipids as substrates (Fig. 4). For instance, LPCAT3 acts not only on lyso-PC (LPC) but also lyso-PE (LPE) and lyso-PS (LPS), but it is unclear that the enzyme has specificity to utilize several different lysophospholipids from the name “LPCAT3” (Fig. 3B, left and Fig. 4). Another issue is that reported substrate selectivities of some LPLATs are inconsistent among different studies, as exemplified by the case of LPCAT1 (Fig. 3B, right and Fig. 4). Consistent use of the unique official gene symbols provided by the National Center for Biotechnology Information (NCBI) would alleviate confusion caused by multiple naming of individual enzymes; however, many of these contain incorrect information of their LPLAT enzymatic activities, especially regarding the lysophospholipid polar head selectivity. Therefore, to avoid confusion in this research field, a revised LPLAT nomenclature is urgently needed.
In this review, we would like to propose a new nomenclature of mammalian LPLATs based on three criteria (Fig. 3C and Table 1). (i) LPLATx (x is a number) is used as a name, which does not include any information on polar head groups of substrates, i.e., type of lysophospholipid. (ii) The numbers assigned in each LPLATx name are essentially in the order of discovery. (iii) The LPLATs for which we propose nomenclature belong to two families, AGPAT and MBOAT. Although there are a few reports of enzymes with potential LPLAT activities that belong to other protein families (
), this review and nomenclature proposal focus only on the enzymes of these two major families. The proposed nomenclature is shown in Table 1 along with enzyme information registered in NCBI. Nomenclature for the currently known GPATs (GPAT1-4, all AGPAT-family enzymes) is generally accurate and uniformly adopted; therefore, no revision is proposed for their current nomenclature, which is shown in Table 2 along with enzyme information registered in NCBI. We will introduce the proposed LPLATx names as our working names in this review, along with official gene symbols in parentheses as needed to avoid ambiguity, beginning with the next section, which describes enzymatic motifs and structures.
Motifs and structures
The experimentally solved structure of a mammalian LPLAT has not been reported yet; however, the X-ray structure of PlsC, an AGPAT-family LPAAT of the bacterium Thermotoga maritima, has been determined (
). The model supports that PlsC introduces an acyl chain at the sn-2 position of PA, and the proper arrangement of hydrophobic tunnels, termed “rulers,” determines the length of the fatty acid to be introduced. In addition, an N-terminus two-helix motif anchors the protein firmly to one leaflet of the membrane (
), and it is hoped the structure of an LPLAT member of the MBOAT family will be determined soon.
Conserved motifs essential for LPLAT activities are reported and are distinct between LPLATs of the AGPAT and MBOAT families (Table 3). LPLATs and GPATs of the AGPAT family have four conserved motifs that are essential for substrate recognition and enzymatic activity: Motif 1 (xHxxxxD), Motif 2 (GxxFxxR), Motif 3 (xxEGxx), and Motif 4 (xxxxPxx) (
). MBOAT family LPLATs also contain four conserved motifs, different from the AGPAT motifs that are critical for LPLAT activities: Motif A (WD), Motif B (WHGxxxGYxxxF), Motif C (YxxxxF), and Motif D (YxxxYFxxH) (
). In our homology models, the four AGPAT motifs surround the putative acyl-CoA-binding pocket. LPLAT8 and LPLAT9 are thought to be monotopic membrane proteins, while LPLATs in the MBOAT family are thought to possess multiple membrane spanning domains.
). The number of transmembrane (TM) domains and total number of embedded helices (EH), including TMs, of the AlphaFold structures modeled in flat lipid bilayer membranes were predicted using PPM 3.0 Web Server (
Recently, a computational machine learning method named AlphaFold has been developed that can predict protein structures with high accuracy, even if no similar experimentally solved structure is available (
). Figure 5, Figure 6, Figure 7, Figure 8 summarize the AlphaFold-generated structural predictions of GPATs and LPLATs, which were predicted with high confidence in most regions of each enzyme (Fig. 5). The predicted structures for all AGPAT-family LPLATs show conserved enzymatic core structures that contain Motifs 1 to 4 (Fig. 5 and Table 3), similar to PlsC (
). In addition, these predicted structures reveal unique regions in each structure, some of which could function as transmembrane helices (Fig. 6). Positioning of the AlphaFold structures in lipid bilayer membranes by computational modeling using PPM 3.0 Web Server (
). In contrast, LPLAT1 to 7 and GPAT3 to 4 are predicted to more strongly anchor to membranes via one or two transmembrane helices, while GPAT1 to 2 are predicted to interact weakly with the membrane surface (Fig. 7 and Table 3). Similar positioning of the AlphaFold structures for the MBOAT-family LPLATs, LPLAT11 to 14, indicated that each of these enzymes may possess 11 transmembrane helices (Fig. 7 and Table 3), comparable to the experimentally determined structures of non-LPLAT MBOAT family members such as human HHAT (12 transmembrane helices) (
As representative LPLATs of the AGPAT and MBOAT families, detailed views of the AGPAT Motifs 1 to 4 in LPLAT9 (LPCAT2) and MBOAT Motifs B-D in LPLAT12 (LPCAT3) are shown in Figure 8A. These motifs are clustered to form a cavity where acyl-CoAs and lysophospholipids might bind, as also shown in the solved structures for PlsC (
) (right panels). Figure 8B, upper panel, shows a manually docked model of LPLAT9 with acetyl-CoA and lyso-platelet-activating factor (lyso-PAF), the natural substrates utilized by LPLAT9 in PAF production. LPLAT9 possesses a hydrophobic cavity that faces the cytosol at one end and the membrane surface at the other, and acetyl-CoA and lyso-PAF could enter the cavity from the cytosol and the membrane, respectively, contacting with each other in the vicinity of the catalytic His146. AGPAT Motifs 1 to 3 constitute the cavity, whereas Motif 4 functions as a backing for the cavity. Besides the motifs, Arg166 and Arg195 could function in recognizing the phosphate groups in lyso-PAF and acetyl-CoA, respectively. Figure 8B, lower panel, shows a manually docked model of LPLAT12 with arachidonoyl-CoA and lyso-PC, the natural substrates utilized by LPLAT12 in arachidonoyl-containing-PC production. LPLAT12 possesses a hydrophobic cavity at the center that penetrates the membrane, and arachidonoyl-CoA and lyso-PC could bind to the cavity from the cytosol and the luminal side of the membrane, respectively, contacting with each other in the vicinity of the catalytic His374. MBOAT Motifs B and C, but not A or D, contribute to the construction of the cavity, which possesses a hydrophobic bulge at the center that might be suitable for accommodating the arachidonoyl chain. Recently, Zhang et al. reported on the structure of the LPLAT12 ortholog in chicken as determined by x-ray crystallography, cryo-EM, and sequence analysis (
). The topology of their model with an ER lumen-facing gate for acyl-CoA substrate was surprising in being oriented opposite of to other determined structures for the MBOAT proteins SOAT1, DGAT1, and HHAT (
). The lumen-/cytosol-facing orientations of their model is also opposite to that presented in our modeled structural prediction, and more analyses are needed to firmly establish the topology and membrane orientation of LPLAT12.
Future experimental studies will be required to validate the predicted structures of LPLATs in both families and reveal their true forms, membrane topologies, and catalytic mechanisms. It is expected that the current advances in protein structural predictions for GPATs and LPLATs may hold potential to offer insights regarding the structural bases for substrate access, specificity, product release, and inhibitors development for all of the enzymes. Combined with experimental validation, those studies may go far to fill in the gaps that still exist in our knowledge of the biochemical activities and biological functions of each enzyme, which are summarized in the following sections for GPATs (Characteristics of GPATs section) and LPLATs (Characteristics of lysophospholipid acyltransferases section).
Characteristics of GPATs
GPATs function in the common biosynthetic pathways of de novo phospholipid synthesis and TAG production by esterification of a fatty acid at the sn-1 or sn-2 position of G3P to generate LPA. At present, four GPAT enzymes, GPAT1-4, have been identified (Table 2). All four GPATs belong to the AGPAT family and have AGPAT motifs (
). In this section, we briefly summarize the biochemical and biological characteristics of GPATs.
The mitochondrial enzyme GPAT1 (also called GPAM) is highly expressed in lipogenic tissues such as liver and adipose, where its levels are decreased by fasting and increased by insulin, indicating a metabolic function to regulate fat utilization and storage (
), and GPAT1-knockout (KO) mice had phospholipid alterations that included reduced C16:0 in sn-1 of PC and PE and increased C20:4 in sn-2, indicating that GPAT1 may not only function in TAG production but also influence the fatty acid compositions of phospholipids (
). Among several tissues in mice, GPAT2 mRNA was expressed almost exclusively in testis, where it was detected in germ cells. The expression in the liver was at least 50-fold less, and even lower in other tissues including the brown adipose tissue (BAT), brain, lung, and heart (
). GPAT2 is also highly expressed in tumors and was suggested to belong to a class of “cancer-testis” genes, whose expression is normally low in somatic tissues but may be upregulated in cancers of various origins (
). GPAT3-KO mice had increased energy expenditures on high-fat diet, and females but not males showed decreased adiposity and body weight gain, indicating important roles of GPAT3 in energy and lipid homeostasis (
). GPAT4-KO mice showed increased energy expenditure, reduced TAG accumulation in BAT and WAT, and profound lack of subdermal adipose tissues. There was an overall increase in polyunsaturated and decrease of monounsaturated fatty acid chains in TAGs, DAGs, and phospholipids; suggesting GPAT4 has selectivity to incorporate monounsaturated fatty acid substrates (
). During lactation, GPAT4 is upregulated in mammary gland epithelia and participates in TAG and DAG synthesis required to support milk production. Indeed, GPAT4-KO mice had dramatically decreased fat droplets within mammary epithelial cells and ducts. GPAT4-KO nursing females had greatly reduced TAG and DAGs in their milk and were unable to successfully nurse their young, indicating a critical role of GPAT4 for the production of milk fat (
Recently, calcineurin B homologous protein 1 (CHP1) has been found to be an essential cofactor for GPAT4 and thereby regulate ER glycerolipid synthesis. CHP1 binding was required for GPAT4 activation, and myristoylation of CHP1 was required for the full interaction. Loss of CHP1 in mammalian cells led to severely reduced fatty acid incorporation and storage, but was partially compensated for by upregulation of glyceronephosphate O-acyltransferase (GNPAT) and increased synthesis of peroxisomal ether lipids (
Characteristics of lysophospholipid acyltransferases
In this section, biochemical and biological characteristics of each LPLAT are summarized. The new proposed names for LPLATs are also introduced, which are in the format LPLATx, where x is a number (Table 1). As described in Nomenclature section, these names are unambiguous and do not imply substrate specificities. The official gene name is given in parentheses immediately following the first instances of the new names. Enzymatic characteristics for the LPLATs are summarized in Table 4 and Figure 4. Physiological roles of LPLATs and their disease associations are summarized in Table 5.
Acylglycerophosphate acyltransferase 4 (AGPAT4) is a mitochondrial lysophosphatidic acid acyltransferase that regulates brain phosphatidylcholine, phosphatidylethanolamine, and phosphatidylinositol levels.
Acylglycerophosphate acyltransferase 4 (AGPAT4) is a mitochondrial lysophosphatidic acid acyltransferase that regulates brain phosphatidylcholine, phosphatidylethanolamine, and phosphatidylinositol levels.